Protein NMR Techniques (Methods in Molecular Biology, v831)

METHODS IN MOLECULAR BIOLOGY™ Series Editor John M. Walker School of Life Sciences University of Hertfordshire Hatfi...

Author: Alexander Shekhtman | David S. Burz

98 downloads 1134 Views 10MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

METHODS

IN

MOLECULAR BIOLOGY™

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

For further volumes: http://www.springer.com/series/7651

Protein NMR Techniques Third Edition Edited by

Alexander Shekhtman and David S. Burz Department of Chemistry, University at Albany, State University of New York, Albany, NY, USA

Editors Alexander Shekhtman Department of Chemistry University at Albany State University of New York 1400 Washington Avenue Albany, NY 12222, USA [email protected]

David S. Burz Department of Chemistry University at Albany State University of New York 1400 Washington Avenue Albany, NY 12222, USA [email protected]

ISSN 1064-3745 e-ISSN 1940-6029 ISBN 978-1-61779-479-7 e-ISBN 978-1-61779-480-3 DOI 10.1007/978-1-61779-480-3 Springer New York Dordrecht Heidelberg London Library of Congress Control Number: 2011943883 © Springer Science+Business Media, LLC 2012 All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Humana Press, c/o Springer Science+Business Media, LLC, 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. Printed on acid-free paper Humana Press is part of Springer Science+Business Media (www.springer.com)

Preface The field of protein NMR spectroscopy has rapidly expanded into new areas of biochemistry, molecular biology, and cell biology research that were impossible to study as recently as 10 years ago. The potential to study macromolecular systems that were once considered too large or too transient or too complex by using NMR spectroscopy is now being realized with the development of innovative technologies. Standard NMR technologies are also getting a facelift in part due to the pervasive nature of high-throughput approaches in biochemical and biomedical research. These advances warrant a new edition of Protein NMR Techniques that includes an authoritative but down-to-earth description of new methodologies. This edition consists of 24 chapters divided into four major categories: NMR sample preparation, solution NMR methodologies, solid-state NMR methodologies, and data processing. The material presented contains enough detail for use not only in specialized NMR laboratories, but in biochemical, molecular, and cell biology research labs that have access to high-field NMR spectrometers. Preparing proteins for NMR spectroscopy can be a time-consuming process that may take longer than data collection and analysis combined. Expression in bacterial cells still remains one of the most popular ways of preparing NMR samples. Chapter 1 discusses new methods for optimizing and increasing the production of isotope-labeled protein in bacteria. However, some proteins are difficult to express in bacteria, in these cases an alternative approach involves using yeast cells. Chapter 2 describes a methodology for producing proteins in yeast that are usually secreted into the growth medium. This technique is proving to be as robust and economic as bacterial production. One drawback to using proteins secreted by yeast is that they may exhibit altered patterns of glycosylation and phosphorylation. To avoid this problem and achieve proper posttranslational modifications, proteins are best produced in insect or mammalian cells. Advances in the use of these cells for producing NMR samples are detailed in Chapters 3 and 4. Cell-free expression of proteins has become a method of choice for high-throughput protein production especially in cases, where the yields from in vivo overexpression are very low. Cell-free systems allow for the selective incorporation of any isotope-labeled amino acid into a target protein with minimal scrambling. Chapters 5 and 6 describe the cell-free production of proteins for solution and solidstate NMR, respectively. Finally, although well-expressed in bacterial cells, some soluble proteins do not fold properly in sufficient quantity to permit analyses of structure, dynamics, and interactions. Chapter 7 presents a methodology for expressing and purifying such proteins in a cost-efficient manner. The chapters on solution NMR methodologies range from the study of individual proteins, large multidomain proteins, protein–ligand and protein–nucleic acid complexes in vitro, to the study of proteins inside living cells. A strategy for studying supramolecular systems, which has become possible due to advances in isotope labeling and NMR pulse sequences, is described in Chapter 8. Chapter 9 presents basic protocols and the latest improvements for measuring relaxation rates and analyzing protein dynamics. Methods to help overcome difficulties in applying solution NMR to the study of membrane proteins are

v

vi

Preface

detailed in Chapter 10. Structurally characterizing multi-domain proteins can be challenging due to the inherent flexibility present in these systems and requires special approaches outlined in Chapter 11. To regulate biological activity, proteins engage in interactions with other macromolecules present in the cell. A description of methods used to prepare protein–RNA, protein–DNA, and protein–ligand complexes suitable for study by using NMR spectroscopy are presented in Chapters 12, 13, and 14. Lastly, Chapter 15 describes in-cell NMR spectroscopy, a relatively new area of NMR research that affords atomic resolution information about isotope-labeled proteins inside living cells. Solid-state NMR spectroscopy presents a complementary approach to studying proteins, especially since the method is not limited by the molecular size constraints that hamper solution NMR. With the availability of high-field NMR spectrometers, solid-state NMR has become a viable technique for acquiring unique information about protein systems that are difficult to characterize by using solution NMR. Chapter 16 reviews the use of magic angle spinning solid-state NMR to study the structure and dynamics of perdeuterated proteins. The preparation and characterization of protein complexes for solid-state NMR and methodologies to analyze the structures and dynamics of protein complexes are presented in Chapter 17. The area of membrane protein expression has seen extensive advances of late, spurred by intense interest in signaling pathways, but impeded by difficulties in preparing samples in sufficient quantity for NMR spectroscopy. Chapter 18 details methods for producing membrane proteins suitable for study by using solid-state NMR. Processing and analyzing NMR data has historically been an extremely laborious part of NMR research, requiring skillful NMR spectroscopists to assign chemical shifts and to determine atomic resolution structures of proteins. With the advent of high-throughput assignment protocols, this task has become largely manageable by a trained graduate student. Nevertheless, there are difficult cases for which there is no substitution for the experienced spectroscopist. For example, characterization of eukaryotic kinases by NMR spectroscopy is complicated by the extensive dynamics and large size exhibited by these proteins. Chapter 19 describes the procedures used to assign backbone resonances for ERK2. The reactivity of solvent-exposed backbone amides varies by a factor of at least a billion-fold because of electrostatic interactions at the protein surface. The use of electrostatic analysis of hydrogen exchange rates to analyze protein flexibility is reviewed in Chapter 20. Chapter 21 presents a strategy for assigning the backbone resonances of small- to medium-sized globular proteins in a few hours by using a highly automated program, BATCH, to acquire, process, and analyze NMR data. A versatile protocol, UNIO, that provides nearly fully automated structure determination is described in Chapter 22. In UNIO, user-intervention is encouraged and facilitated by graphical tools for preparing, analyzing, validating, and presenting the NMR structure. Chapter 23 details the use of the ARIA software, which incorporates both solution and solid-state NMR structural constraints to perform structure calculations. The final chapter, Chapter 24, introduces the software DYNAMICS for analyzing relaxation rates that characterize the overall tumbling and local dynamics of a protein. This book presents a comprehensive description of the latest innovations in the field of protein NMR. It focuses on the importance of biochemistry, molecular biology, and cell biology to NMR spectroscopy while avoiding excessive repetition of existing material, which is readily available through a number of excellent texts and reviews that cover topics

Preface

vii

relevant to studying proteins by using NMR. Rather than reiterating the fundamental principles behind NMR methodologies, we have emphasized the practical aspects of experimental design combined with practical advice and examples. We hope that this book will provide both experienced NMR spectroscopists and biochemists, who are new to the field of NMR, with enough background to successfully apply these techniques to their research. Albany, NY, USA

Alexander Shekhtman David S. Burz

Contents Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v xi

1 A Novel Bacterial Expression Method with Optimized Parameters for Very High Yield Production of Triple-Labeled Proteins . . . . . . . . . . . . . . . . . . . 1 Victoria Murray, Yuefei Huang, Jianglei Chen, Jianjun Wang, and Qianqian Li 2 Isotopic Labeling of Heterologous Proteins in the Yeast Pichia pastoris and Kluyveromyces lactis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Toshihiko Sugiki, Osamu Ichikawa, Mayumi Miyazawa-Onami, Ichio Shimada, and Hideo Takahashi 3 Isotope Labeling in Insect Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Krishna Saxena, Arpana Dutta, Judith Klein-Seetharaman, and Harald Schwalbe 4 Isotope Labeling in Mammalian Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Arpana Dutta, Krishna Saxena, Harald Schwalbe, and Judith Klein-Seetharaman 5 Cell-Free Protein Production for NMR Studies. . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 Mitsuhiro Takeda and Masatsune Kainosho 6 Cell-Free Membrane Protein Expression for Solid-State NMR . . . . . . . . . . . . . . . . 85 Alaa Abdine, Kyu-Ho Park, and Dror E. Warschawski 7 Expression and Purification of Src-family Kinases for Solution NMR Studies . . . . . . 111 Andrea Piserchio, David Cowburn, and Ranajeet Ghose 8 NMR Studies of Large Protein Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 Shiou-Ru Tzeng, Ming-Tao Pai, and Charalampos G. Kalodimos 9 Protein Dynamics by 15N Nuclear Magnetic Relaxation . . . . . . . . . . . . . . . . . . . . . . 141 Fabien Ferrage 10 Bacterial Production and Solution NMR Studies of a Viral Membrane Ion Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 Jolyon K. Claridge and Jason R. Schnell 11 Preparation of the Modular Multi-Domain Protein RPA for Study by NMR Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 Chris A. Brosey, Marie-Eve Chagot, and Walter J. Chazin 12 NMR Studies of Protein–RNA Interactions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 Carla A. Theimer, Nakesha L. Smith, and May Khanna 13 Preparation and Optimization of Protein–DNA Complexes Suitable for Detailed NMR Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 My D. Sam and Robert T. Clubb

ix

x

Contents

14 NMR Studies of Protein–Ligand Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 Michael Goldflam, Teresa Tarragó, Margarida Gairí, and Ernest Giralt 15 In-Cell NMR Spectroscopy in Escherichia coli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 Kirsten E. Robinson, Patrick N. Reardon, and Leonard D. Spicer 16 Deuterated Peptides and Proteins: Structure and Dynamics Studies by MAS Solid-State NMR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 Bernd Reif 17 Solid-State NMR Spectroscopy of Protein Complexes . . . . . . . . . . . . . . . . . . . . . . . 303 Shangjin Sun, Yun Han, Sivakumar Paramasivam, Si Yan, Amanda E. Siglin, John C. Williams, In-Ja L. Byeon, Jinwoo Ahn, Angela M. Gronenborn, and Tatyana Polenova 18 Synthesis, Purification, and Characterization of Single Helix Membrane Peptides and Proteins for NMR Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333 Miki Itaya, Ian C. Brett, and Steven O. Smith 19 Assignment of Backbone Resonances in a Eukaryotic Protein Kinase – ERK2 as a Representative Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359 Andrea Piserchio, Kevin N. Dalby, and Ranajeet Ghose 20 Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility . . . . . . . . . . 369 Griselda Hernández, Janet S. Anderson, and David M. LeMaster 21 Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407 Bernhard Brutscher and Ewen Lescop 22 Comprehensive Automation for NMR Structure Determination of Proteins . . . . . . 429 Paul Guerry and Torsten Herrmann 23 ARIA for Solution and Solid-State NMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453 Benjamin Bardiaux, Thérèse Malliavin, and Michael Nilges 24 Determining Protein Dynamics from 15N Relaxation Data by Using DYNAMICS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485 David Fushman Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

513

Contributors ALAA ABDINE • CNRS and Université Paris Diderot, IBPC, Paris, France JINWOO AHN • Department of Structural Biology, Pittsburgh Center for HIV Protein Interactions, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA JANET S. ANDERSON • Department of Chemistry, Union College, NY, USA BENJAMIN BARDIAUX • NMR-supported Structural Biology, Leibnitz-Institut für Molekulare Pharmakologie (FMP), Berlin, Germany IAN C. BRETT • Department of Biochemistry and Cell Biology, Stony Brook University, Stony Brook, NY, USA CHRIS A. BROSEY • Departments of Biochemistry and Chemistry, Center for Structural Biology, Vanderbilt University, Nashville, TN, USA BERNHARD BRUTSCHER • Institut de Biologie Structurale – Jean-Pierre Ebel, CNRS, CEA, UJF, UMR5075, Grenoble Cedex, France IN-JA L. BYEON • Department of Structural Biology, Pittsburgh Center for HIV Protein Interactions, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA MARIE-EVE CHAGOT • Departments of Biochemistry and Chemistry, Center for Structural Biology, Vanderbilt University, Nashville, TN, USA WALTER J. CHAZIN • Departments of Biochemistry and Chemistry, Center for Structural Biology, Vanderbilt University, Nashville, TN, USA JIANGLEI CHEN • Department of Biochemistry and Molecular Biology, School of Medicine, Wayne State University, Detroit, MI, USA JOLYON K. CLARIDGE • Department of Biochemistry, University of Oxford, Oxford, UK ROBERT T. CLUBB • Department of Chemistry and Biochemistry, University of California, Los Angeles, CA, USA DAVID COWBURN • Departments of Biochemistry and Physiology & Biophysics, Albert Einstein College of Medicine of Yeshiva University, Bronx, NY, USA KEVIN N. DALBY • Division of Medicinal Chemistry, University of Texas, Austin, TX, USA; Graduate Programs, Cellular and Molecular Biology, Pharmacy, Biomedical Engineering and Biochemistry, University of Texas, Austin, TX, USA ARPANA DUTTA • Department of Structural Biology, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA FABIEN FERRAGE • Département de chimie, Ecole normale supérieure et Laboratoire des Biomolécules, CNRS UMR 7203, Paris, Cedex, France DAVID FUSHMAN • Department of Chemistry and Biochemistry and Center for Biomolecular Structure and Organization, University of Maryland, MD, USA MARGARIDA GAIRÍ • Servicios Cientifico Tecnicos, Universitat de Barcelona, Barcelona, Spain RANAJEET GHOSE • Department of Chemistry, The City College of New York, New York, NY, USA; Graduate Center of the City University of New York, New York, NY, USA

xi

xii

Contributors

ERNEST GIRALT • Institute for Research in Biomedicine (IRB Barcelona), Parc Científic de Barcelona, Barcelona, Spain; Departament de Química Orgànica, Universitat de Barcelona, Barcelona, Spain MICHAEL GOLDFLAM • Institute for Research in Biomedicine (IRB Barcelona), Parc Científic de Barcelona, Barcelona, Spain ANGELA M. GRONENBORN • Department of Structural Biology, Pittsburgh Center for HIV Protein Interactions, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA PAUL GUERRY • Centre Européen de RMN à très Hauts Champs, Université de Lyon, Ecole Normale Supérieure de Lyon, CNRS, Université Claude, Villeurbanne, France YUN HAN • Department of Chemistry and Biochemistry, University of Delaware, Newark, DE, USA GRISELDA HERNÁNDEZ • Department of Health and Department of Biomedical Sciences, Wadsworth Center, School of Public Health, University at Albany – SUNY, Albany, NY, USA TORSTEN HERRMANN • Centre Européen de RMN à très Hauts Champs, Université de Lyon, CNRS, Ecole Normale Supérieure de Lyon, Université Claude, Villeurbanne, France YUEFEI HUANG • Department of Biochemistry and Molecular Biology, School of Medicine, Wayne State University, Detroit, MI, USA OSAMU ICHIKAWA • Graduate School of Pharmaceutical Sciences, The University of Tokyo, Tokyo, Japan MIKI ITAYA • Department of Biochemistry and Cell Biology, Stony Brook University, Stony Brook, NY, USA MASATSUNE KAINOSHO • Graduate School of Science, Nagoya University, Nagoya, Japan; Center for Priority Areas, Tokyo Metropolitan University, Hachioji, Japan CHARALAMPOS G. KALODIMOS • Department of Chemistry and Chemical Biology, Rutgers University, Piscataway, NJ, USA MAY KHANNA • Department of Biochemistry and Molecular Biology, Indiana University School of Medicine, Indianapolis, IN, USA JUDITH KLEIN-SEETHARAMAN • Department of Structural Biology, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA DAVID M. LEMASTER • Department of Health and Department of Biomedical Sciences, Wadsworth Center, School of Public Health, University at Albany – SUNY, Albany, NY, USA EWEN LESCOP • Laboratoire de Chimie et Biologie Structurales, Institut de Chimie des Substances Naturelles, Centre de Recherche de Gif, CNRS, Gif-sur-Yvette, France QIANQIAN LI • Department of Biochemistry and Molecular Biology, School of Medicine, Wayne State University, Detroit, MI, USA THÉRÈSE MALLIAVIN • Unité de Bioinformatique Structurale, CNRS URA 2185, Institut Pasteur, Paris, France MAYUMI MIYAZAWA-ONAMI • Japan Biological Informatics Consortium (JBiC), Tokyo, Japan; Biomedicinal Information Research Center (BIRC), National Institute of Advanced Industrial Science and Technology (AIST), Tokyo, Japan

Contributors

xiii

VICTORIA MURRAY • Department of Biochemistry and Molecular Biology, School of Medicine, Wayne State University, Detroit, MI, USA MICHAEL NILGES • Unité de Bioinformatique Structurale, CNRS URA 2185, Institut Pasteur, Paris, France MING-TAO PAI • Department of Chemistry and Chemical Biology, Rutgers University, Piscataway, NJ, USA SIVAKUMAR PARAMASIVAM • Department of Chemistry and Biochemistry, University of Delaware, Newark, DE, USA KYU-HO PARK • CNRS and Université Paris Diderot, IBPC, Paris, France ANDREA PISERCHIO • Department of Chemistry, The City College of New York, New York, NY, USA TATYANA POLENOVA • Department of Chemistry and Biochemistry, University of Delaware, Newark, DE, USA PATRICK N. REARDON • Department of Biochemistry, Duke University NMR Center, Durham, NC, USA BERND REIF • Munich Center for Integrated Protein Science (CIPSM) at Department Chemie, Technische Universität München, Garching, Germany; Leibniz-Institut für Molekulare Pharmakologie (FMP), Berlin, Germany; Helmholtz-Zentrum München (HMGU), German Research Center for Environmental Health, Neuherberg, Germany KIRSTEN E. ROBINSON • Department of Biochemistry, Duke University NMR Center, Durham, NC, USA MY D. SAM • Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA, USA KRISHNA SAXENA • Institute for Organic Chemistry and Chemical Biology, Center for Biomolecular Magnetic Resonance, Johann Wolfgang Goethe-University Frankfurt, Frankfurt am Main, Germany JASON R. SCHNELL • Department of Biochemistry, University of Oxford, Oxford, UK HARALD SCHWALBE • Institute for Organic Chemistry and Chemical Biology, Center for Biomolecular Magnetic Resonance, Johann Wolfgang Goethe-University Frankfurt, Frankfurt am Main, Germany ICHIO SHIMADA • Biomedicinal Information Research Center (BIRC), National Institute of Advanced Industrial Science and Technology (AIST), Tokyo, Japan; Graduate School of Pharmaceutical Sciences, The University of Tokyo, Tokyo, Japan AMANDA E. SIGLIN • Department of Molecular Medicine, Beckman Research Institute of City of Hope, Duarte, CA, USA NAKESHA L. SMITH • Department of Chemistry, University at Albany SUNY, Albany, NY, USA STEVEN O. SMITH • Department of Biochemistry and Cell Biology, Stony Brook University, Stony Brook, NY, USA LEONARD D. SPICER • Department of Biochemistry, Duke University NMR Center, Durham, NC, USA; Department of Radiology, Duke University NMR Center, Durham, NC, USA TOSHIHIKO SUGIKI • Japan Biological Informatics Consortium (JBiC), Tokyo, Japan; Biomedicinal Information Research Center (BIRC), National Institute of Advanced Industrial Science and Technology (AIST), Tokyo, Japan

xiv

Contributors

SHANGJIN SUN • Department of Chemistry and Biochemistry, University of Delaware, Newark, DE, USA HIDEO TAKAHASHI • Department of Supramolecular Biology, Graduate School of Nanobioscience, Yokohama City University, Yokohama, Japan; Biomedicinal Information Research Center (BIRC), National Institute of Advanced Industrial Science and Technology (AIST), Tokyo, Japan MITSUHIRO TAKEDA • Graduate School of Science, Nagoya University, Nagoya, Japan TERESA TARRAGÓ • Institute for Research in Biomedicine (IRB Barcelona), Parc Científic de Barcelona, Barcelona, Spain CARLA A. THEIMER • Department of Chemistry, University at Albany SUNY, Albany, NY, USA SHIOU-RU TZENG • Department of Chemistry and Chemical Biology, Rutgers University, Piscataway, NJ, USA JIANJUN WANG • Department of Biochemistry and Molecular Biology, School of Medicine, Wayne State University, Detroit, MI, USA DROR E. WARSCHAWSKI • CNRS and Université Paris Diderot, IBPC, Paris, France JOHN C. WILLIAMS • Department of Molecular Medicine, Beckman Research Institute of City of Hope, Duarte, CA, USA SI YAN • Department of Chemistry and Biochemistry, University of Delaware, Newark, DE, USA

Chapter 1 A Novel Bacterial Expression Method with Optimized Parameters for Very High Yield Production of Triple-Labeled Proteins Victoria Murray, Yuefei Huang, Jianglei Chen, Jianjun Wang, and Qianqian Li Abstract The Gram-negative bacterium Escherichia coli offer a means for rapid, high-yield, and economical production of recombinant proteins. However, when preparing protein samples for NMR, high-level production of functional isotopically labeled proteins can be quite challenging. This is especially true for the preparation of triple-labeled protein samples in D2O (2H/13C/15N). The large expense and time-consuming nature of triple-labeled protein production for NMR led us to revisit the current bacterial protein expression protocols. Our goal was to develop an efficient bacterial expression method for very high-level production of triple-labeled proteins that could be routinely utilized in every NMR lab without changing expression vectors or requiring fermentation. We developed a novel high cell-density IPTG-induction bacterial expression method that combines tightly controlled traditional IPTG-induction expression with the high cell-density of auto-induction expression. In addition, we optimize several key experimental protocols and parameters to ensure that our new high cell-density bacterial expression method routinely produces 14–25 mg of triple-labeled proteins and 15–35 mg of unlabeled proteins from 50-mL bacterial cell cultures. Key words: High yield protein production, Bacterial expression, Isotopic labeling, NMR

1. Introduction To perform NMR structural studies of proteins, we have to produce proteins that are isotopically labeled with 13C and 15N for small proteins (<20 kDa) and with 13C, 15N, and 2H for larger proteins (>20 kDa). Among the many systems available for heterologous protein production, the Gram-negative bacterium Escherichia coli remains one of the most attractive hosts (1, 2). This is especially true for isotopically labeling proteins since bacterial expression Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_1, © Springer Science+Business Media, LLC 2012

1

2

V. Murray et al.

provides the cheapest way to prepare these proteins for NMR studies (3). Protein expression and purification is a routine practice in many NMR labs, but it is not uncommon to see a drastic reduction in protein yield when isotopically labeling the proteins, especially when D2O must be used. To overcome these difficulties, we have developed a novel bacterial expression method that combines the tightly controlled traditional IPTG-induction bacterial expression with the high celldensity of the auto-induction method (4). To summarize our procedure, we first determine how to make a proper starting culture, followed by double colony selection and finally high cell-density expression. With these optimized protocols and parameters, our new bacterial expression method offers near gram quantity production of triple-labeled proteins from one-liter bacterial cell cultures, without changing expression vectors and without fermentation (5). Thus, every NMR laboratory can easily apply this novel bacterial expression method on a routine basis for the production of a very high yield of triple-labeled proteins for their NMR structural studies of proteins.

2. Materials 2.1. Sample Preparation and SDS-PAGE

1. 4× SDS loading buffer: 200 mM Tris–HCl, pH 6.8, 8% (w/v) sodium dodecyl sulfate (SDS), 0.4% (w/v) bromophenol blue, 40% glycerol. Store at room temperature. 2. Dithiothreitol (DTT): Prepare 1 M solution using sterile water. Filter through a 0.22-μm pore size membrane (syringe filter) and store 1-mL aliquots at −20°C. 3. 30% Acrylamide/bis solution (29:1). 4. 1.5 M Tris–HCl, pH 8.8. 5. 10% SDS. 6. 10% ammonium persulfate. 7. TEMED. 8. M Tris–HCl, pH 6.8. 9. Protein molecular weight markers. 10. 5× SDS running buffer: 0.5% SDS, 125 mM Tris base, 1.25 M glycine. Store at room temperature. Do not adjust pH. 11. Coomassie blue-staining solution: 0.25% (w/v) Coomassie brilliant blue, 45% methanol, 10% acetic acid. Add brilliant blue to methanol and stir for 60 min. Add water and acetic acid and stir for another 30 min. Store at room temperature (see Note 1). 12. Destaining solution: 30% methanol, 10% acetic acid. Store at room temperature.

1

2.2. Protein Expression

A Novel Bacterial Expression Method with Optimized Parameters…

3

1. LB medium (Miller): Dissolve 25 g of powdered LB medium in 1 L of distilled water or D2O. Adjust the pH to 7.4 using NaOH. Autoclave and store at room temperature (see Note 2). Add antibiotics (KAN or AMP) prior to use. 2. Kanamycin monosulfate stock solution (KAN): Dissolve KAN monosulfate to a concentration of 30 mg/mL in distilled water. Syringe filter and store 1 mL aliquots at −20°C. 3. Ampicillin sodium sulfate stock solution (AMP): Dissolve AMP sodium sulfate to a concentration of 50 mg/mL in distilled water. Syringe filter and store 1 mL aliquots at −20°C. 4. LB agar plates (Miller): Dissolve 40 g of LB agar (Miller) in 1 L of distilled water and/or D2O in a 2-L flask. Cover with foil and autoclave. Monitor the temperature as it cools. When the temperature reaches ~50°C, add 1 mL of the KAN or AMP stock solution. Pour ~10 mL into 100 × 10-mm Petri dishes and swirl to coat the plate. Let the LB agar solidify at room temperature. Place plates back into a plastic bag, seal with tape, and store at 4°C (see Note 2). 5. Isopropyl-β-D-thiogalactopyranoside (IPTG): Prepare a 1 M solution using distilled water. Syringe filter and store 1 mL aliquots at −20°C. 6. 100% Glycerol: Autoclave to sterilize. Store at room temperature. 7. 5× M9 Salts (1 L): Dissolve 64 g of Na2HPO4, 15 g of KH2PO4, 5 g of NH4Cl, and 2.5 g of NaCl in distilled water or D2O, adjust the volume to 1 L. Autoclave to sterilize and store at room temperature (see Note 2). Omit NH4Cl from this recipe when 5× M9 salts are used for isotope-labeling. Do not adjust pH. 8. 20% Glucose: Dissolve 20 g glucose in distilled water, adjust volume to 100 mL. Sterilize by filtration and store at 4°C. 9. MgSO4: Prepare a 1-M solution, autoclave, and store at room temperature. 10. CaCl2: Prepare 1-M solution, autoclave, and store at room temperature. 11. M9 minimal medium for traditional IPTG method and double colony selection (100 mL): 78 mL of distilled, sterilized water, 20 mL of 5× M9 salts, 2 mL of 20% glucose, 200 μL of 1 M MgSO4, 10 μL of 1 M CaCl2, and 100 μL of antibiotic. Add the CaCl2 last and immediately swirl the flask to dissolve the cloudy precipitate. Adjust the pH to 7.4 using NaOH. 12. M9 minimal medium for high cell-density IPTG-induction method (100 mL): 75 mL of distilled, sterilized water, 20 mL of 5× M9 salts, 5 mL of 20% glucose, 200 μL of 1 M MgSO4, 10 μL of 1 M CaCl2, and 100 μL of antibiotic. Add the CaCl2 last and immediately swirl the flask to dissolve the cloudy precipitate (see Note 3). Adjust the pH to 7.4 using NaOH (see Note 4).

4

V. Murray et al.

13. M9 for double-labeling (100 mL): 80 mL of distilled, sterilized water, 20 mL of 5× M9 salts without NH4Cl, 100 mg of 15 NH4Cl and 0.2 g of 13C-glucose (for traditional IPTG method) or 1 g of 13C-glucose (for high cell-density IPTGinduction method), 200 μL of 1 M MgSO4, 10 μL of 1 M CaCl2, and 100 μL of antibiotic. Add the CaCl2 last and immediately swirl the flask to dissolve the cloudy precipitate (see Note 3). Adjust the pH to 7.4 using NaOH (see Note 4). Use a filtration unit with a 0.22-μm pore size to sterilize the medium (see Note 5). 14. M9 for triple-labeling (100 mL): 80 mL of 99% D2O, 20 mL of 5× M9 salts without NH4Cl in D2O, 100 mg of 15NH4Cl, and 0.2 g of 13C/2H-glucose (for traditional IPTG-induction method), or 1 g of 13C/2H-glucose (for high cell-density IPTG-induction method), 200 μL of 1 M MgSO4, 10 μL of 1 M CaCl2, and 100 μL of antibiotic. You can also use 13 C-glucose; however, this usually generates ~90% deuterated triple-labeled protein samples. Add the CaCl2 last and immediately swirl the flask to dissolve the cloudy precipitate (see Note 3). Adjust the pH to 7.4 using NaOH (see Note 4). Use a filtration unit with a 0.22-μm pore size to sterilize the medium (see Note 5). 15. 1000× Trace metals (100 mL): Dissolve 811 mg of FeCl3 (50 mM), 222 mg of CaCl2 (20 mM), 125.8 mg of MnCl2 (10 mM), 161.5 mg of ZnSO4 (10 mM), 26 mg of CoCl2 (2 mM), 26.9 mg of CuCl2 (2 mM), 25.9 mg of NiCl2 (2 mM), 41.2 mg of Na2MoO4 (2 mM), 34.6 mg of Na2SeO3 (2 mM), and 12.4 mg of H3BO3 (2 mM) in 60 mM HCl. Autoclave to sterilize. Store at room temperature. 16. BME vitamins (see Note 3). 2.3. Protein Purification

1. Affinity resin: His-Bind Resin. 2. 8× Charge buffer: 400 mM NiSO4. Store at 4°C. 3. 8× Binding buffer: 160 mM Tris–HCl, pH 7.9, 2 M NaCl, 20 mM imidazole. Store in an amber bottle at room temperature. 4. 8× Wash buffer: 160 mM Tris–HCl, pH 7.9, 2 M NaCl, 240 mM imidazole. Store in an amber bottle at room temperature. 5. 4× Elute buffer: 80 mM Tris–HCl, pH 7.9, 1 M NaCl, 4 M imidazole. Store in an amber bottle at room temperature. 6. (NH4)2CO3: Dissolve 474.4 g of (NH4)2CO3 in 3.5 L of water. Once dissolved, adjust the volume to 4 L (final concentration 1.23 M) and store at room temperature. Do not adjust pH. 7. Urea.

1

A Novel Bacterial Expression Method with Optimized Parameters…

5

3. Methods When triple-labeling proteins, bacteria have to be grown in D2O, usually causing a significant reduction in protein yields. We sought to overcome this obstacle. Our strategy mainly focuses on increasing the cell density of bacterial expression without manipulation of the expression vector or use of a fermenter. Unfortunately, bacterial expression at a high cell-density in D2O usually causes several major problems, including (1) plasmid loss, (2) significant reduction in the pH of the growth medium due to cell metabolites, and (3) limited availability of dissolved oxygen. These problems often result in a low or even no protein production with high cell-density bacterial expression. We developed several practical protocols that solved these problems, including (1) preparation of a proper starting culture, (2) double colony selection in D2O, (3) optimization of bacterial expression conditions, and (4) better control of the pH of the medium. We further developed a high cell-density IPTG-induction bacterial expression method that combines the tightly controlled traditional IPTG-induction expression with high cell-density autoinduction expression. Our optimized protocols ensure plasmid stability inside bacterial cells, resulting in routine production of 14–25 mg triple-labeled proteins from a 50-mL bacterial cell culture. Importantly, this novel bacterial expression method uses the same expression vectors as the traditional IPTG-induction method and does not require a fermenter. Thus, every NMR laboratory can easily adopt this novel bacterial expression method to produce large quantities of triple-labeled proteins. 3.1. PAGE Sample Preparation

PAGE samples come either from the cell lysate isolated immediately after bacterial expression or from the column flow through during protein purification. 1. For samples collected directly from bacterial cell culture: Collect 500 μL of cells and place in a 1.7-mL microcentrifuge tube. Spin down at 12,000 × g for 5 min at room temperature using a microcentrifuge, discard the supernatant, and tap out the excess on a paper towel. 2. Add 25 μL of 4× SDS loading buffer and 25 μL of water and resuspend the pellet. Store the samples in a freezer until ready to run a gel. 3. Before running the culture samples on a gel, place the samples on a 90°C heat block for 30 min. Remove from the heat block; add 50 μL of water and vortex for 30 s (see Note 6). 4. For samples collected from flow through during protein purification, mix 60 μL of column flow through with 20 μL of

6

V. Murray et al.

4× SDS loading buffer and mix thoroughly by repeated pipetting up and down. Store the samples in a freezer until ready to run a gel. 5. Before running the flow through samples on a gel, place the samples on a 90°C heat block for 5 min (see Note 6). 6. Before loading samples on a gel, centrifuge at 12,000 × g for 10 min at room temperature using a microcentrifuge to pellet cellular debris (see Note 7). 3.2. SDS-PAGE

1. Depending on the protein size, choose an appropriate acrylamide percentage for the resolving gel (see Note 8). 2. To prepare a 10% SDS-PAGE mini-resolving gel (5 mL), using a mini-gel apparatus: Mix 1.9 mL of water, 1.7 mL of 30% acrylamide/bis solution, 1.3 mL of 1.5 M Tris–HCl, pH 8.8, 50 μL of 10% SDS, 50 μL of 10% ammonium persulfate, and 2 μL of TEMED. Mix well and pour between glass plates set in a loading cassette. Leave about 1.5-cm space on top for the stacking gel. Gently pipet water on top and let the gel set (about 20 min). 3. Pour a stacking gel (2 mL) once the resolving gel has set. Mix 1.4 mL of water, 330 μL of 30% acrylamide/bis solution, 250 μL of 1 M Tris–HCl, pH 6.8, 20 μL of 10% SDS, 20 μL of ammonium persulfate, and 2 μL of TEMED. Pour the water off the top of the resolving gel, remove excess water with filter paper, and pour the stacking gel. Insert a comb containing the appropriate number of lanes and let the stacking gel set (about 20 min). 4. Prepare 1× running buffer by diluting 100 mL of 5× running buffer solution with 400 mL of distilled water. Make sure to mix the solution well. Pour the buffer into the inner and outer chambers of the gel apparatus. For cell culture samples, load 7.5 μL of each sample into the lanes. For column flow through samples, load 20 μL into each lane. Make sure to load 5 μL of molecular weight markers in one lane. 5. Secure the lid on the gel box and plug into a power supply. Run the gel at 88 V for ~2 h. Turn off the power supply when the blue dye front reaches the bottom of the gel. 6. Remove the gel from the glass plates and place it in a small box with 20 mL of Coomassie blue-staining solution. Allow the gel to stain for at least 30 min (see Note 1). Pour out the stain, rinse the gel with water to remove excess stain and then add 30 mL of destaining solution. Place a small piece of paper towel in the box to accelerate the destaining process. This process may take a few hours; however, you can start to detect bands within 30–60 min.

1

3.3. A Proper Starting Culture

A Novel Bacterial Expression Method with Optimized Parameters…

7

A critical consideration for high-level bacterial expression is the preparation of a proper starting culture in a rich medium for scaling up in minimal medium. The general practice in NMR laboratories is to grow an overnight culture using a rich medium, such as LB, at 37°C. We observe that an overnight culture usually reaches saturation by the next morning, and may result in plasmid instability and loss due to several factors including basal leakage of the T7 expression system that expresses the toxic target proteins to the host cells under this overgrowth condition (4, 6, 7). This usually results in a poor yield of target protein. Figure 1 shows a growth curve for E. coli BL21 (DE3) cells carrying the LCAT/pET30a vector, in H2O-based LB, suggesting that the bacteria are in the exponential or log phase of growth between 6 and 7.5 h at 37°C (see Note 9). We placed particular emphasis on double colony selection (Subheading 3.5) to ensure that a high percentage of bacterial cells within this colony contain the DNA expression plasmid. 1. Perform a time course of bacterial growth of a new protein expression vector in rich (LB) medium by measuring the OD600 every 30 min for ~10–12 h in water-based rich medium and ~16–18 h in D2O-based rich medium (see Note 10). The OD600

Fig. 1. Plot of E. coli growth in 5 mL of LB medium over a 10-h period at 37°C starting with a glycerol stock. The bacterial strain, BL-21(DE3), contains a pET30a vector expressing the gene for lecithin:cholesterol acyltransferase (LCAT). Based on this plot, the log phase of the culture is between an OD600 of 1 and 3.5. Note: The growth curve is vector, protein, and bacterial strain dependent. Reproduced from Murray 2010 with permission from Cold Spring Harbor.

8

V. Murray et al.

of the log phase and the time to reach saturation are vector, protein, and bacterial strain-dependent; therefore, we suggest performing this experiment before actually expressing protein for purification and labeling (see Note 11). 2. Once the optimal OD600 of the log phase of the starting culture is determined, this will be the OD600 for all future starting cultures using both traditional IPTG expression and high celldensity expression methods. 3.4. Traditional IPTG Method

This method can be scaled up or down to suit your needs. This following protocol is used to make a 100-mL expression culture. 1. Prepare a 5-mL proper starting culture as described in Subheading 3.3 in a 50-mL conical tube with holes poked in the lid. 2. Prepare 100 mL of M9 minimal medium in a 250-mL flask. Add 1.5 mL of the starting culture and measure the OD600. (To check the OD600 at this point, aliquot 1 mL of the cell culture into a cuvette and measure the OD600.) We suggest a starting OD600 between 0.05 and 0.10 for healthy bacterial cell growth. Place the flask in a 37°C incubator with a shaking speed of 200 rpm. 3. Start to monitor the OD600 after 3 h (to check the OD600 at this point, dilute 100 μL of cell culture into a cuvette containing 900 μL of distilled water and measure the OD600. The final OD600 value is ten times the spectrophotometer reading). Once the OD600 reaches between 0.8 and 1.2 (see Note 12), remove 2 mL of cell culture and place it in a 15-mL culture tube (for a noninduced reference). Induce the remaining culture with 0.5 mM IPTG. Place both the flask and the 15-mL culture tube in a 20°C incubator overnight with a shaking speed of 200 rpm. 4. The following morning, measure the final OD600 of the cell culture. If the protein expression is induced at an OD600 of 1, the OD600 of the bacterial cell culture will be around 2–3, indicating healthy bacterial cell growth. Harvest the cells by spinning down at 10,000 × g for 10 min at 4°C using a benchtop centrifuge. Remove the supernatant and either store the cell pellet at −80°C or use immediately for protein purification. 5. To check protein expression levels, take 500 μL samples of each culture (non-induced and IPTG-induced), follow the sample preparation protocol (Subheading 3.1) and run an SDS-PAGE (Subheading 3.2) to compare non-induced and IPTG-induced samples.

3.5. Double Colony Selection

Since plasmid loss is encountered during bacterial expression in D2O much more frequently than in H2O, we describe a double colony selection procedure for triple-labeling protein in D2O. Based on our

1

A Novel Bacterial Expression Method with Optimized Parameters…

9

Fig. 2. SDS-PAGE of protein expression of apoE(1–215)/pTYB1 in D2O before (a), during (b) and after (c) double colony selection. Arrows indicate the expected protein band (~80 kDa). (a) Shows four different colonies before colony selection. (b) Shows the results of three different colonies selected from the single colony selection (lanes 1–3) and another three colonies selected from the double colony selection (lanes 4–6). The second colony selection was based on Colony 3 (lane 3, b). (c) Shows the results of six colonies from the double colony selection, indicating a high level of protein expression for all six colonies. Reproduced from Sivashanmugam 2009 with permission from Wiley Interscience.

experience, this is a critical protocol that significantly increases the yield of triple-labeled proteins. Figure 2 shows the result of a typical double colony selection of apoE(1–215)/pTYB1 in D2O, demonstrating that high protein expression levels are achieved after double colony selection. This high-level expression of apoE(1–215)/pTYB1 in D2O has been stable for more than 2 years. Usually, we perform double colony selection before we optimize the expression conditions. Thus, the traditional IPTGinduction method is the default method with which to start double colony selection. The procedure, described in Subheading 3.4, can be applied here for double colony selection, except for culture volume and the use of D2O. 1. Prepare LB agar plates with 50% D2O. 2. Perform a bacterial transformation using LB agar plates prepared with D2O. 3. Next afternoon, choose nine colonies to make starting cultures of 5 mL of LB in 50% D2O and 5 μL of antibiotic in 50-mL conical tubes, and prepare a master plate (see Note 13). Punch holes in the lids of the tubes. Eight of the nine colonies will be induced with IPTG and the ninth colony will be used as a negative, non-induced control. 4. When the starting culture is ready, add 50–100 μL to 5 mL of M9 minimal medium in 70% D2O and 5 μL of antibiotic in 50-mL conical tubes. Ensure that the starting OD600 is between 0.05 and 0.10. Place the tube in a 37°C incubator with a shaking speed of 200 rpm.

10

V. Murray et al.

5. When the OD600 of the culture reaches between 0.8 and 1 (see Note 14), add 0.5 mM IPTG to eight of the nine cultures to induce protein expression, clearly marking which culture is serving as a negative control. Place the tubes in a 20°C incubator overnight with a shaking speed of 200 rpm. 6. The following morning, remove 500 μL of cell culture from each tube and place in a microcentrifuge tube. Centrifuge at 12,000 × g for 10 min at room temperature and discard the supernatant. Prepare SDS-PAGE samples and run a gel. 7. Choose the colony expressing the biggest protein band (using the negative control as reference) and prepare 5 mL of LB culture in either 70 or 99% D2O and 5 μL of antibiotic containing the colony from the master plate. Grow at 37°C until the OD600 reaches 0.7–0.9 and spread 150 μL on an LB plate prepared in 50% D2O. Invert the plate and incubate it at 37°C overnight. 8. The following day, repeat steps 3–6 using 70 or 99% D2O for the second round of colony selection (see Note 14). When completed, choose the colony expressing the biggest protein band to make a 5-mL LB culture in 100% D2O and 5 μL of antibiotic. Grow the culture halfway through its log phase. Add 800 μL of culture to 200 μL of 100% sterile glycerol in a 2-mL cryogenic tube with a screw top cap. Pipet up and down to mix thoroughly and flash freeze the tubes by dipping them in liquid nitrogen, store at −80°C. If the protein expression level is extremely high, we suggest that you make at least 5–10 glycerol stocks of this double selected colony for future use. 3.6. High Cell-Density IPTG-Induction Method

This is a hybrid method combining traditional IPTG-induction and auto-induction bacterial expression methods. It takes advantage of tightly controlled IPTG-induction and the high cell-density of the auto-induction bacterial expression. We use rich media, such as LB and 2× YT, to reach a high cell-density before IPTGinduction and then switch the culture medium by gently spinningdown the cells and resuspending them into an equal volume of minimal medium. However, many problems may occur during a high cell-density bacterial expression that can cause a significant reduction in protein yield, such as reduced pH of the expression medium, poor aeration, and/or plasmid loss during expression. We describe the following procedure to avoid these problems, ensuring a high yield of triple-labeled protein. 1. Using the glycerol stock prepared after double colony selection (Subheading 3.5), make a starting culture in 50 mL of LB medium containing 50 μL antibiotic in a 250-mL flask (Dip the pipet tip in the glycerol stock and scratch the surface. Place the tip into the culture medium, pipet up and down a few

1

A Novel Bacterial Expression Method with Optimized Parameters…

11

times and remove the tip. Immediately return the glycerol stock to the −80°C freezer). Place the flask in a 37°C incubator with a shaking speed of 200 rpm and incubate until the OD600 is halfway through the log phase (Subheading 3.3, step 1). DO NOT let the starting culture grow overnight since the saturation of cell growth may cause plasmid loss. 2. Transfer the cells into sterile tubes and spin down the culture at 5,000 × g for 7 min at room temperature. Remove the supernatant and tap the tubes on paper towels to remove as much LB as possible. 3. Gently resuspend the pellets in 50 mL of M9 minimal medium (see Note 4) and transfer the resuspended cells to a 250-mL sterile flask. Place the flask in an incubator that is set at the optimal induction temperature (see Subheading 3.7). Maintain the shaking speed at 200 rpm for efficient aeration and keep the culture at the optimal temperature for 1–1.5 h to allow the cells to adapt to the new medium. An enhancement of approximate 0.5–1 U of OD600 should be observed at the end of this time. For example, if the OD600 is 3 right after medium exchange, you can expect the OD600 to reach 3.5–4 at the end of the 1–1.5 h incubation, indicating healthy cell growth (this increase in OD600 will be slightly less when expressing in D2O). 4. Add the optimal concentration of IPTG (Subheading 3.7) to induce protein expression and keep the culture in the incubator for an optimal period of time (Subheading 3.7) at a shaking speed of 200 rpm. 5. Measure the final OD600 before harvesting cells. You can expect to see a two- to fourfold enhancement at the end of the expression. For example, if your OD600 is 4 after the 1–1.5 h incubation, the final OD600 will be between 8 and 16, indicating healthy bacterial growth. 6. Harvest the cells by spinning down the culture at 10,000 × g for 10 min at 4°C. Remove the supernatant and either store the cell pellet at −80°C, or immediately use for protein purification. 3.7. Optimization of Various Conditions

Another important step for high-level protein production using high cell-density bacterial expression is to optimize the expression conditions such as culture temperature and the induction time. These steps are critical for the initial expression of a protein using the high cell-density expression method. We usually use the traditional IPTG-induction method first to check if a new protein can be expressed by using bacteria. Once the protein expression is confirmed by the traditional IPTG-induction method, we can optimize the high cell-density expression method to produce high-yield isotopically labeled proteins.

12

V. Murray et al.

We usually carry out time courses at different temperatures, such as 15, 20, room temperature, 30 and 37°C. We normally prepare a 5-mL starting culture either in D2O or in water for the time course and closely monitor the following parameters during expression: OD600, pH, and target protein production. The detailed procedure of optimization follows: 1. Prepare 10 mL starting cultures in 50-mL flasks: 10 mL of LB medium in 99% D2O, 10 μL of antibiotic and bacterial cells from a glycerol stock after double colony selection. Incubate at 37°C with a shaking speed of 200 rpm. 2. Once the optimal OD600 has been reached, gently centrifuge the cells at 5,000 × g for 7 min at room temperature and discard the supernatant. Resuspend the cell pellets with 10 mL of M9 minimal medium in 99% D2O in 50-mL flasks and incubate at various temperatures, such as 15, 20, room-temperature, 30° and 37°C, for 1 h. Check the OD600 of the cultures before and after this 1-h cell incubation. 3. If the OD600 of each culture after the 1-h incubation increases by 0.5–1 OD600 units, this indicates that the cells are healthy and growing after the medium exchange. Induce protein expression by adding 0.5 mM IPTG. Be sure to choose one culture without IPTG-induction to serve as the negative control. Return the tubes to appropriate shakers at different temperatures. We found that 0.5 mM IPTG usually gives a reasonable protein production, thus we always used this IPTG concentration as our starting point. However, an independent optimization of IPTG concentration can be carried out and is discussed in step 6 of this section. 4. For cultures growing below 25°C, let them grow overnight (~14–16 h). The following morning, collect 500 μL of cell culture samples every 2 h (typically collect between 16 and 28 h and one more sample the next morning). For the cultures growing above 25°C, start to collect samples every 2 h after induction for at least 8 h and one the next morning. At each time point, check the OD600 and the pH of the cell culture (see Note 4). Be sure to keep the collected cell pellets at −20°C. 5. When all of the samples have been collected, prepare samples for SDS-PAGE analysis. A comparison with the negative control (noninduced culture) allows you to determine which temperature and induction time give you the best protein yield (see Note 15). 6. Repeat the above procedure to optimize the IPTG concentration using the optimized temperature and incubation time. We usually test IPTG concentrations of 0.1, 0.25, 0.5, 0.75, and 1 mM. 7. Once the optimal conditions have been determined, you can now perform high cell-density bacterial expression on a larger

1

A Novel Bacterial Expression Method with Optimized Parameters…

13

scale. Be sure to use a large flask for better aeration. We usually use a 250-mL flask for a 50-mL cell culture and a 500-mL flask for a 100-mL cell culture. If you want to grow a 200-mL cell culture, we suggest dividing the culture into 2 × 100 mL cultures in two 500-mL flasks. Figure 3 shows a typical time course experiment performed during optimization of experimental conditions. Table 1 shows the expression parameters for the time course during our optimization of the expression of human apolipoprotein A-I (apoAI). It is clear that at the maximum OD600 the bacterial expression produces the highest yield of triple-labeled apoAI, as confirmed by a Western blot (Fig. 3).

Fig. 3. Left panel: An SDS-PAGE showing auto-induction time course of triple-labeled human apoAI expression in D2O at room temperature. Lanes 1–7 = 24, 28, 32, 36, 40, 44, and 54 h, respectively. Right panel is a Western blot of the same time course using anti-apoAI monoclonal antibody. Reproduced from Sivashanmugam 2009 with permission from Wiley Interscience.

Table 1 Parameters of the time course of human apolipoprotein A–I expression Time

24 h

28 h

32 h

36 h

40 h

44 h

54 h

OD600

2.5

3.9

7.2

9.1

8.4

8

8.1

pH

6.6

6.5

6.3

6

6

6

6.1

Protein yield

−

+

++

+++

++

++

−

14

V. Murray et al.

3.8. Protein Purification

Protein purification depends on the fusion tag that is used, as different tags serve different purposes. In our laboratory, we generally use histidine tags. In this section, we describe a typical protein purification procedure using a His-Bind Resin column. 1. Prepare 1× dilutions of all buffers. Recheck the pH to ensure that they are still 7.9. 2. Centrifuge the cell culture at 10,000 × g for 10 min at 4°C. Remove the supernatant and resuspend the pellets in 20 mL of 1× binding buffer. If the protein is in inclusion bodies and can be refolded readily during dialysis, you can resuspend the pellet in 20 mL of 1× binding buffer containing 6 M urea. 3. Lyse the cells by using either sonication or a French press. Centrifuge the lysate at 16,000 × g for 20 min at 4°C. Collect the supernatant and store it on ice. 4. Add 10 mL of 1× binding buffer and repeat step 2 at least twice, and then combine all the supernatants. Depending on the protein, you may need to add additional binding buffer and repeat step 2 3–5 times to completely extract the protein from the cells. 5. Equilibrate the affinity column with 50 mL of 1× charge buffer (see Note 16). Remove the charge buffer and equilibrate the column with 50 mL of 1× binding buffer. The column should be a light blue color after equilibration. 6. Load the column with the clear lysate from step 3. The flow rate should be ~1 mL/minute (see Note 17). Collect the flow through and remove a 60-μL sample for SDS-PAGE. 7. Wash the column with 200 mL of 1× binding buffer, followed by an additional 100 mL of 1× wash buffer. The flow rate of wash buffer should also be ~1 mL/min. Elute the column with 100 mL of 1× elution buffer. Collect the last drop of elution for SDS-PAGE to make sure that all of the protein has been eluted from the column. 8. Perform SDS-PAGE analysis with all of the collected samples to assess the purification. 9. Place the eluted protein into a dialysis bag and dialyze extensively against water containing 20 mM (NH4)2CO3 to remove imidazole, salts and possibly urea. After dialysis, freeze the protein sample with liquid nitrogen and lyophilize to obtain pure triple-labeled protein powder. Run a gel to assess the purity of the protein powder.

3.9. Conclusion

With the high cell-density IPTG-induction bacterial expression method and the practical protocols described above, we routinely produce 14–25 mg of triple-labeled proteins and 15–35 mg of unlabeled proteins from a 50-mL cell culture for all the proteins we tested. Table 2 lists the final yields of unlabeled and triplelabeled proteins obtained using high cell-density bacterial expression

1

A Novel Bacterial Expression Method with Optimized Parameters…

15

Table 2 Final yields of unlabeled and triple labeled proteins: high cell-density vs. traditional IPTG method Protein

High cell densityb (mg)

IPTGb (mg)

M.W. (Cal) (Da)

M.W. (MS) (Da)

%Dc

Triple-labeled RAP(1–210) RAP(91–323) ApoE(1–183)a Mouse apoAI(1–216) Human apoAI

20 ± 3 25 ± 3 18 ± 4 15 ± 2 14 ± 1

0.5 0.8 2 0.8 0.6

33,801 36,633 22,866 28,014 32,814

33,525 ± 195 36,376 ± 200 22,686 ± 116 27,732 ± 125 32,401 ± 150

~92 ~93 ~89 ~90 ~88

Unlabeled Human apoAI Human apoE

34 ± 1 17 ± 2

1 0.2

M.W. molecular weight Reproduced from Sivashanmugam 2009 with permission from Wiley Interscience a ApoE(1–183) was expressed in 40% D2O, the rest are expressed in 99.7% D2O b High cell density (50-mL culture volume): high cell-density expression methods, including auto-induction and high cell-density IPTG-induction; IPTG: the optimized traditional IPTG-induced expression. We repeated the expressions at least three times for all proteins, the yield shown is the average ± standard deviation c Estimated percentage of deuteration, assuming 100% 13C and 15N-labeling. For apoE(1–183), the %D is the estimated percentage of deuteration based on 40% D2O. For the other four proteins, the %D is the estimated percentage of deuteration based on 99.7% D2O

compared with the yields obtained by using the traditional IPTGinduction method in a 50-mL cell culture, the results suggest a 5–100-fold enhancement in protein yield. In addition, the protocols described produce a consistent high-level of triple-labeled protein, which is always reproducible. Table 2 also gives mass spectroscopic data for the triple-labeled protein, indicating the efficiency of deuteration for triple-labeled protein using auto-induction expressions. Overall, the deuteration efficiency is around 90% if we assume the 13C and 15N-labeling are 100%. This is because we used 99.7% D2O and nondeuterated 13C-glycerol or 13C-glucose in high celldensity expressions. This result is comparable to the deuteration efficiency of the traditional IPTG-induction expression method using single labeled 13C-glucose.

4. Notes 1. The stain can be reused multiple times. When using fresh stain, you only need to stain gels for 15–30 min. Pour the used stain into a separate container. When reusing stain, you may need to stain gels longer.

16

V. Murray et al.

2. When using D2O to replace water in LB agar, broth or M9 salts, solutions CANNOT be autoclaved to sterilize. When making LB broth, follow the directions as stated, but instead of autoclaving, use a filtration unit with a 0.22-μm pore size to sterilize. When making LB agar plates, use a microwave to bring the solution just to a boil (to dissolve agar), add antibiotics once the agar temperature cools to ~50°C, and pour into plates. When making 5× M9 salts, follow the directions as stated, however omit NH4Cl for isotopic-labeling. Once again, do not autoclave; sterilize using a filtration unit with a 0.22-μm pore size. 3. For the auto-induction method, Studier suggests using vitamins and trace metals (4). We found vitamins and trace metals help promote healthy bacterial growth when using our high cell-density IPTG-induction bacterial expression. We purchased the trace metals and BME vitamins stock solution from Sigma. The trace metals used in our laboratory is based on Studier’s recipe provided in the supplement materials of his elegant paper on the auto-induction bacterial expression method (4). For our optimized high cell-density IPTGinduction minimal medium, we added 0.25× vitamins and 0.25× trace metals (see Table 1 in ref. 5). 4. While monitoring OD600 during the optimization time course (Subheading 3.7), the pH should be monitored as well. As the cell density increases, the pH of the culture lowers due to the release of cell metabolites. If the pH becomes too low (pH < 6), it will affect bacterial cell health and protein production. If the pH drops below 6 during the time course, we increase the pH of the M9 minimal medium to 8 using NaOH to allow for a larger buffering capacity of the culture medium. 5. M9 minimal medium containing isotopes (2H, 13C, or 15N) cannot be autoclaved. It must be sterilized using a filtration unit containing a 0.22-μm pore size. 6. If the protein contains cysteine residue(s), add 10–20 mM DTT after adding water. Vortex the sample well and let it sit for 30–60 min at room temperature. 7. The release of DNA can cause the sample to become quite viscous, making it hard to load on the gel. If you notice this, you can simply sonicate your sample at a low wattage (4–6 W) for 5–10 s. Afterward, spin down the sample at 12,000 × g for 2 min at room temperature. Also, when loading the gel, remove the sample from the top portion of the supernatant to avoid the pelleted cellular debris at the bottom. 8. To determine which percentage gel to use, follow the guidelines found in Table A8-8 in Sambrook and Russell (8). Also, if you are working with proteins that weigh less than 15 kDa,

1

A Novel Bacterial Expression Method with Optimized Parameters…

17

12% Tricine gels or gradient gels are highly recommended for better resolution in the molecular weight range of 5–15 kDa. 9. Bacteria display a four-phase pattern of cell growth in liquid media. First, there is an initial lag phase when bacteria are adapting to the growth conditions; at this point, an increase in OD600 will not be seen. Second, bacteria enter their exponential or log phase at which point the bacterial cells start dividing (doubling in number). The OD600 during this log phase climbs steadily. Third, bacteria enter the stationary phase during which the rate of cell growth significantly slows due to a decrease in available nutrients and an accumulation of toxins. The OD600 will level off during this phase. Finally, if fresh medium is not made available and toxins are not removed, bacteria will enter the death phase and a noticeable drop in OD600 will be observed. The key is to utilize a starting culture during the exponential or log phase of their growth curve (see Fig. 1). 10. When performing a time course using rich media, such as LB or 2× YT in D2O, bacteria grow much more slowly than in water. 11. Based on our experience, we suggest the appropriate OD600 range for the starting culture is between 3 and 5 in LB medium and between 5 and 7 in 2× YT medium. However, the OD600 of the log phase is vector, protein, and bacterial strain-dependent. Thus, the best way to determine the middle point of the log phase is to perform a time course of bacterial growth for each new protein expression vector. 12. Based on our experience, for healthy bacterial cell growth, the culture is expected to reach an OD600 of 0.8–1.2 within 4–6 h for bacterial expressions in water and within 6–9 h for bacterial expressions in D2O if the starting OD600 is between 0.05 and 0.10. 13. When performing colony selection, it is helpful to make a “master plate.” Take a KAN or AMP plate and make a 9-square grid under the agar plate. Label boxes 1–9. When you are ready to inoculate the medium, take a sterilized tip, gently touch the selected colony, and then gently touch the agar of the master plate in the corresponding box. Go back to the original plate, retouch the same colony, and drop it into a tube. Repeat this procedure for all selected colonies. Once finished, put the lid back on the plate, invert and incubate at 37°C for about 8 h. Colonies should be about 1–2-mm in diameter. Cut a long strip of parafilm, wrap the edges of the plate, and store inverted at 4°C. You can use this plate to regrow the colonies for future cell cultures. However, the plate is only good for about 2 weeks, so be sure to also make glycerol stocks. 14. If the bacteria do not grow well in 70% D2O, then they must be trained to adapt to D2O medium. For this purpose, pick a colony off a D2O plate and start a 5-mL bacterial culture of LB

18

V. Murray et al.

medium in 25% D2O. Once the OD600 of the culture reaches 1 at 37°C, transfer 100 μL of the cell culture into 5 mL of LB medium in 50% D2O. The starting OD600 of this new culture is about 0.1. Let the cell culture grow at 37°C until the OD600 reaches 1 and transfer 100 μL of the cell culture into 5 mL of LB medium in 75% D2O and let the culture to grow at 37°C until the OD600 reaches 2–3. Use this cell culture as your starting culture. 15. A Western blot of this time course will further allow an unambiguous determination of the time point that produces the best protein yield. 16. Typically, we prepare columns containing 5 mL of affinity resin for a 50-mL cell culture since 2.5 mL of resin can bind about 20 mg of protein. This can be scaled up or down to suit your needs based on the expected protein yield. 17. If the flow rate is too slow, we have found that using a 1.5-μm syringe filter to remove cellular debris from the clear lysate prevents columns from becoming clogged and running slow. References 1. Swartz, J.R. (2001) Advances in Escherichia coli production of therapeutic proteins. Curr. Opin. Biotechnol. 12, 195–201. 2. Hewitt, L., and McDonnell, J.M. (2004) Screening and optimizing protein production in E. coli methods. Methods Mol. Biol. 278, 1–16. 3. McIntosh, L.P. and Dahlquist, F.W. (1990) Biosynthetic incorporation of 15N and 13C for assignment and interpretation of nuclear magnetic resonance spectra of proteins. Q Rev. Biophys. 23, 1–38. 4. Studier, F.W. (2005) Protein production by auto-induction in high density shaking cultures. Protein Expr. Purif. 41, 207–234. 5. Sivashanmugam, A., Murray, V., Cui, C., Zhang, Y., Wang, J., and Li, Q. (2009) Practical

protocols for production of very high yields of recombinant proteins using Escherichia coli. Protein Sci. 18, 936–948. 6. Chen, H.C., Hwang, C.F., and Mou, D.G. (1992) High-density Escherichia coli cultivation process for hyperexpression of recombinant porcine growth hormone. Enzyme Microb. Technol. 14, 321–326. 7. Baneyx, F. (1999) Recombinant protein expression in Escherichia coli. Curr. Opin. Biotechnol. 10, 411–421. 8. Sambrook, J. and Russell, D. (2001) Molecular Cloning: A Laboratory Manual (3rd ed.). Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York (ISBN 978087969577-4).

Chapter 2 Isotopic Labeling of Heterologous Proteins in the Yeast Pichia pastoris and Kluyveromyces lactis Toshihiko Sugiki, Osamu Ichikawa, Mayumi Miyazawa-Onami, Ichio Shimada, and Hideo Takahashi Abstract Several protein expression systems are available for the preparation of stable isotope-labeled recombinant proteins for NMR studies. Yeast expression systems have several advantages over prokaryotic systems, such as the widely used Escherichia coli expression system. Protein expression using the methylotrophic yeast Pichia pastoris is commonly employed for the preparation of isotope-labeled proteins. Recently, the hemiascomycete yeast Kluyveromyces lactis expression system was reported as being useful for preparing proteins for NMR studies. Since each yeast expression system has different features, their applications have increased in number. In this chapter, we describe procedures for the efficient production of uniformly isotope-labeled proteins using the P. pastoris and the K. lactis yeast expression systems. Key words: Yeast expression systems, Pichia pastoris, Kluyveromyces lactis, Stable isotope labeling, Fed-batch fermentation, NMR

1. Introduction The most widely used method for the expression of isotopically labeled heterologous recombinant proteins is the Escherichia coli expression system because of easy handling, rapid and high-density cell growth, high levels of protein production, and relatively low costs for isotope-labeling. However, in many cases, expression of structurally and functionally intact eukaryotic proteins by E. coli and other prokaryotic cells is fundamentally difficult due to (1) the lack of intracellular organelles, (2) a limited number of molecular chaperones, and (3) the absence of posttranslational modification mechanisms (1–6). Yeast, however, combines several advantages of

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_2, © Springer Science+Business Media, LLC 2012

19

20

T. Sugiki et al.

both eukaryotic and prokaryotic expression systems (1) processing, folding, complex disulfide-bond network formation, and posttranslational modification (e.g., glycosylation) of proteins are possible, (2) yeast exhibit rapid growth rates and grow to high-density, and (3) methods for molecular genetic manipulation are well established and are simple to perform (1–6). Here, we introduce the Pichia pastoris and Kluyveromyces lactis yeast expression systems for the stable isotopic labeling of heterologous proteins. 1.1. The P. pastoris Expression System

Host cells and expression vectors are available from Invitrogen. Based upon Invitrogen instruction manuals (7), previous reports (8–15), and our experience, we describe here optimized cell culture procedures for the production of isotope-labeled heterologous proteins by P. pastoris. P. pastoris is capable of utilizing methanol as both a carbon source and to induce protein expression. Methanol is oxidized by alcohol oxidases, AOX1 and AOX2, in yeast cell peroxisomes (though the majority of alcohol oxidase activity is attributed to AOX1). The AOX1 promoter tightly regulates expression of the AOX1 gene, and methanol induces AOX1 promoter activity. Thus, the AOX1 promoter is used to drive the expression of a target protein by replacing the AOX1 gene with a cDNA encoding the desired heterologous protein (1–6). Invitrogen supplies several vector series that are commonly used for protein expression in P. pastoris, including pPIC9K, pPIC3.5K, and pPICZ. Using these vectors, a cDNA cassette containing the target gene under the control of the AOX1 promoter is inserted into the genome of P. pastoris by using homologous recombination along with a gene coding for resistance to a drug, such as Zeocin™ or G418, for subsequent selection of transformed cells (7). Target proteins are expressed either intracellularly or secreted into the medium by P. pastoris. For secretion of target proteins, vectors pPICZα and pPIC9K are recommended (7). The pPICZα and pPIC9K series contain a gene that encodes a prepro signal sequence [such as the Saccharomyces cerevisiae α-mating factor (α-MF)] between the AOX1 promoter and the target gene (7). Although target proteins may contain native secretion signals, in many cases, the α-MF sequence is generally used as the sole secretion signal. Since P. pastoris secretes a low amount of its native proteins, a secreted target protein comprises the vast majority of the total proteins in the culture medium, simplifying purification of the target protein (1–7). A different set of vectors lacking secretion signal sequences are required if the target protein is to be expressed intracellularly. For cytosolic and nonglycosylated proteins, vectors pPICZ or pPIC3.5K are recommended (7). In this chapter, we describe methods and provide technical advice for producing uniformly labeled [U-13C, 15N] and [U-2H, 15N] target proteins in P. pastoris (8–14). The procedures described are

2

Isotopic Labeling of Heterologous Proteins in the Yeast…

21

optimized for secretory expression of isotope-labeled DDR2 from P. pastoris strain X-33 (Mut+ phenotype transformants) using the pPICZα vector system (15). 1.2. The K. lactis Expression System

Host cells and expression vectors are available from New England Biolabs. Based upon instruction manuals (16), previous reports, and our experience, we describe here optimized cell culture procedures for the production of isotope-labeled heterologous proteins by K. lactis. Heterologous recombinant proteins expressed by K. lactis can be sequestered intracellularly or secreted into the medium. As with P. pastoris, K. lactis secretes very low levels of native proteins, therefore secreted target proteins comprise the vast majority of the total protein present in the culture medium, which simplifies purification of the target protein (16–18). In K. lactis, the LAC4 promoter drives expression of the target protein gene. Using the pKLAC vector, a cDNA cassette containing the target gene under the control of the LAC4 promoter is inserted into the genome of K. lactis by using homologous recombination (16–18). The transcriptional activity of the LAC4 promoter is induced by galactose, which K. lactis also utilizes as a carbon source for cell growth. Therefore, expression of the target protein is constitutively induced by cultivation in a medium containing galactose (16–18). The pKLAC vectors impart a fungal acetamidase gene (amdS) to the transformants. Since only transformants that express amdS can utilize acetamide as a sole nitrogen source, amdS is used for selection of the desired transformants, thereby obviating the need for expensive antibiotics, such as Zeocin™ and rendering this auxotrophic selection method more cost-effective (16–18). In addition, auxotrophic acetamide selection enriches populations of transformants in which multiple tandem copies of target cDNA are integrated into the yeast genome (16–18). Thus, the K. lactis protein expression system is simple, cost-effective, easily scaled-up, and highly reproducible. We recently established a cost-effective isotope-labeling method utilizing the hemiascomycete yeast K. lactis (6, 19). In the most commonly employed K. lactis expression system, 20 g/L galactose is required as a carbon source for cell growth and as an inducer of target protein expression (16). However, while a 20 g/L carbon source is acceptable for uniform 15N-labeling, it is economically infeasible for uniform 13C-labeling. Sugiki et al. reported that by combining K. lactis strain GG799, which is characterized by weak glucose suppression for the LAC4 promoter (16), with a fed-batch culture method, larger amounts of protein can be expressed using a smaller amount of glucose, thus reducing costs to a level comparable to isotope-labeling by E. coli systems (6, 19). In this chapter, we describe methods and provide technical advice for producing uniformly labeled [U-13C, 15N] and [U-2H, 15N] target proteins in K. lactis.

22

T. Sugiki et al.

2. Materials The recipes for the culture media described here were derived mainly from the manuals for the EasySelect™ Pichia Expression Kit (Invitrogen), the Pichia Fermentation Process Guidelines (Invitrogen), and the K. lactis Protein Expression Kit (New England Biolabs). Autoclave sterilization is performed at 121°C for 15 min. 2.1. Uniform 13C, 15 N-Labeling Using P. pastoris

1. 40% unlabeled D-glucose stock solution: Dissolve 200 g of D-glucose in 1 L of water. Sterilize by aseptic filtration and store at room temperature. The shelf life of this solution is approximately 1 year. 2. 100 mg/mL Stock solution of Zeocin™ (Invitrogen). 3. Yeast extract peptone dextrose sorbitol (YPDS) agar plates: Dissolve 20 g of bacto peptone, 10 g of yeast extract, 182.2 g of sorbitol, and 20 g of bacto agar in 950 mL of water. Sterilize by autoclaving and cool to 50–60°C. Aseptically add 1 mL of 100 mg/mL Zeocin™ and 50 mL of 40% glucose stock solution, gently mix well, and dispense into sterile disporsable Petri dishes. Store at 4°C in the dark. The shelf life of these agar plates is 1–2 weeks. 4. 10× Yeast nitrogen base (YNB): Dissolve 34 g of YNB without amino acids and ammonium sulfate in 1 L of water (see Note 1). Warm it to 40–50°C to dissolve completely if needed. Sterilize by aseptic filtration and store at 4°C. The shelf life of this solution is approximately 1 year. 5. 1 M Potassium phosphate: Mix 132 mL of 1 M K2HPO4 and 868 mL of 1 M KH2PO4. Adjust the pH to 6.0 ± 0.1 using KOH or phosphoric acid. Sterilize by autoclaving and store at room temperature. The shelf life of this solution is greater than 1 year. 6. 500× Biotin: Dissolve 20 mg of biotin in 100 mL of water and warm it to 40–50°C to dissolve completely. Sterilize by aseptic filtration and store at 4°C. The shelf life of this solution is approximately 1 year. 7. 10% (w/v) Glycerol: Dissolve 50 mL of glycerol in 450 mL of water. Sterilize by aseptic filtration and store at room temperature. The shelf life of this solution is over 1 year. 8. 200× PTM1 trace salts: Mix together the following ingredients and dissolve to a final volume of 1 L in water: 6.0 g of CuSO 4⋅5H2O, 0.08 g of NaI, 3.0 g of MgSO4⋅H2O, 0.2 g of Na2MoO4⋅2H2O, 0.02 g of boric acid, 0.5 g of CoCl2, 20.0 g of ZnCl2, 65.0 g of FeSO4⋅7H2O, 0.2 g of biotin, 5.0 mL of H2SO4. Warm to 40–50°C to dissolve completely if needed. Sterilize by aseptic filtration and store at 4°C or room temperature. The shelf life of this solution is approximately 3 months at 4°C.

2

Isotopic Labeling of Heterologous Proteins in the Yeast…

23

9. Buffered glycerol-complex (BMGY) medium: Dissolve 20 g bacto peptone and 10 g yeast extract in 700 mL of water. Sterilize by autoclaving and cool it to room temperature. Aseptically add 100 mL of sterile 1 M potassium phosphate, 100 mL of sterile 10× YNB, 2 mL of sterile 500× biotin, and 100 mL of sterile 10% (w/v) glycerol solutions. If needed, 5 mL of 200× PTM1 trace salts solution can be added to this medium. Store at 4°C. The shelf life of this medium is approximately 3 months. 10. 10% 15N-ammonium chloride: Dissolve 6 g of 15N-ammonium chloride (98 atom% 15N) in 60 mL of water (see Note 2). Sterilize by aseptic filtration. Prepare this solution just before use. 11.

15

N-BM medium: To prepare 1.2 L of uniformly 15N-labeled buffered minimal (15N-BM) medium, autoclave 900 mL of water in a 2-L media bottle and allow the water to cool to room temperature. Aseptically add 120 mL of 10× YNB, 2.4 mL of 500× biotin, 120 mL of 1 M potassium phosphate, and 60 mL of 10% 15N-ammonium chloride to the autoclaved water. If needed, 6 mL of 200× PTM1 trace salts solution can be added to 15N-BM medium.

12. 5% D-[13C6]glucose: Dissolve 1 g of D-[13C6]glucose (99 atom% 13 C) in 20 mL of 15N-BM medium and sterilize by aseptic filtration. Prepare this solution just before use. 13. Antifoaming agent: (see Note 3). 14.

13

C-methanol (99 atom% 13C).

15. 50% (w/v) ethanol. Ethanol is just necessary for immersing air outlet port of the fermentation system. 2.2. Uniform 2H, 15 N-Labeling Using P. pastoris

1. YPDS agar plates (Subheading 2.1). 2. BMGY medium (Subheading 2.1). 3. 10× YNB(2H2O): Prepare this solution in a manner similar to 10× YNB solution (Subheading 2.1), but use 2H2O (99.8 atom % 2H) instead of water (H2O). Sterilize by aseptic filtration and store at room temperature. The shelf life of this solution is approximately 2–3 months. 4. 1 M potassium phosphate(2H2O): Prepare 150 mL of this solution in a manner similar to 1 M potassium phosphate solution (Subheading 2.1), but use 2H2O instead of water. Sterilize by aseptic filtration. Prepare this solution just before use. 5. 500× biotin(2H2O): Prepare this solution in a manner similar to 500× biotin solution (Subheading 2.1), but use 2H2O instead of water. Sterilize by aseptic filtration and store at 4°C. The shelf life of this solution is approximately 2–3 months. 6. 200× PTM1 trace salts(2H2O): Prepare this solution in a manner similar to 200× PTM1 trace salts solution (Subheading 2.1), but use 2H2O instead of water. Sterilize by aseptic filtration and store at 4°C.

24

T. Sugiki et al.

7. 10% 15N-ammonium chloride(2H2O): Dissolve 7 g of 15 N-ammonium chloride in 70 mL of 2H2O (see Note 2) and sterilize by aseptic filtration. Prepare this solution just before use. 8. 10% unlabeled D-glucose(2H2O): Dissolve 1 g of D-glucose in 10 mL of 2H2O and sterilize by aseptic filtration. Prepare this solution just before use. 9. 15N-BMD (90% 2H2O) medium: To prepare 10 mL of uniformly 15 N-labeled buffered minimal medium containing unlabeled glucose prepared in 90% 2H2O, mix 6 mL of 2H2O and 1 mL of H2O, and sterilize by aseptic filtration. Aseptically add 1 mL of 10× YNB(2H2O), 0.02 mL of 500× biotin(2H2O), 1 mL of 1 M potassium phosphate(2H2O), 0.5 mL of 10% unlabeled D-glucose(2H2O) (0.5% final concentration), 0.5 mL of 10% 15 N-ammonium chloride(2H2O), and 5–10 μL of antifoaming agents. If needed, 0.05 mL of 200× PTM1 trace salts(2H2O) can be added to this medium. Prepare this medium just before use. 10.

15

N-BM(2H2O) medium: To prepare 1.3 L of uniformly N-labeled buffered minimal medium in 100% 2H2O, sterilize 956 mL of fresh 2H2O by aseptic filtration. Aseptically add 130 mL of 10× YNB(2H2O), 3 mL of 500× biotin(2H2O), 130 mL of 1 M potassium phosphate(2H2O), and 66 mL of 10% 15N-ammonium chloride(2H2O) into the sterile 2H2O. If needed, 7 mL of 200× PTM1 trace salts(2H2O) can be added to this medium. Prepare this medium just before use. 15

11.

15

N-BMD(2H2O) medium: To prepare 5 mL of uniformly N-labeled buffered minimal medium containing unlabeled glucose in 100% 2H2O, aseptically add 0.25 mL of 10% unlabeled D-glucose(2H2O) to 4.75 mL of 15N-BM(2H2O) in a 50-mL Corning tube. Prepare this medium just before use. 15

12. 10% D-[2H7]glucose(2H2O): Dissolve 1.5 g of D-[2H7]glucose (98 atom% 2H) in 15 mL of 2H2O and sterilize by aseptic filtration. Prepare this solution just before use. 13. 2H, 15N-BMD(2H2O) medium: To prepare 300 mL of uniformly 15 N-labeled buffered minimal medium containing D-[2H7]glucose in 100% 2H2O, transfer 285 mL of 15N-BM(2H2O) into a sterilize, well-dried 1-L baffled flask. Aseptically add 15 mL of 10% D-[2H7]glucose(2H2O) and 0.15–0.30 mL of antifoaming agents. Prepare this medium just before use. 14. Antifoaming agent (Subheading 2.1). 15. [2H4]methanol (99.5 atom% 2H). 16. 50% (w/v) ethanol. Ethanol is just necessary for immersing air outlet port of the fermentation system. 2.3. Uniform 13C, 15 N-Labeling Using K. lactis

1. Yeast peptone (YP): Dissolve 20 g of bacto peptone and 10 g of yeast extract in 950 mL of water. Sterilize by autoclaving and store at room temperature or 4°C. The shelf life of this solution is approximately 1 year.

2

Isotopic Labeling of Heterologous Proteins in the Yeast…

25

2. 40% unlabeled D-glucose stock solution (Subheading 2.1). 3. YPD medium: Aseptically add 50 mL of sterile 40% D-glucose stock solution to the autoclaved 950 mL YP solution. The shelf life of this solution is approximately 1 month. 4. 1 M potassium phosphate (Subheading 2.1). 5. YCB-acetamide agar plates: Dissolve 5.85 g of yeast carbon base (YCB) medium powder and 10 g of bacto agar in 470 mL of water. Add 25 mL of 1 M potassium phosphate solution. Autoclave and cool it to 50–60°C. Aseptically add 5 mL of 100× acetamide stock solution (supplied by New England Biolabs), gently mix well, and dispense into sterile disporsable Petri dishes. Store at 4°C. The shelf life of these plates is approximately 3 months. 6. 10× YNB (Subheading 2.1). 7. 500× biotin (Subheading 2.1). 8. 200× PTM1 trace salts (Subheading 2.1). 9. 10% 15N-ammonium chloride: Dissolve 7.5 g of 15N-ammonium chloride in 75 mL of water (see Note 2) and sterilize by aseptic filtration. Prepare this solution just before use. 10. 4% D-[13C6]glucose: Dissolve 2 g of D-[13C6]glucose in 50 mL of water and sterilize it by aseptic filtration. Prepare this solution just before use. 11.

13

C, 15N-BMD medium: To prepare 0.5 L of 15N-BM medium containing 0.4% D-[13C6]glucose, autoclave 325 mL of water containing 0.25–0.50 mL of antifoaming agents in a 2-L fermentation vessel and allow it to cool to room temperature. Aseptically add 50 mL of 10× YNB, 1 mL of 500× biotin, 50 mL of 1 M potassium phosphate, 25 mL of 10% 15N-ammonium chloride, and 50 mL of 4% D-[13C6]glucose to the autoclaved water for a final volume of 500 mL. If needed, 2.5 mL of 200× PTM1 trace salts solution can be added to this medium.

12. 6% D-[13C6]glucose: Dissolve 6 g of D-[13C6]glucose in 100 mL of water and sterilize by aseptic filtration. Prepare this solution just before use. 13. 13C, 15N-BMD feeding medium: To prepare 1 L of 15N-BM medium containing 0.6% D-[13C6]glucose feeding medium, autoclave 650 mL of water containing 0.5–1.0 mL of antifoaming agents in a 1-L media bottle and allow it to cool to room temperature. Aseptically add 100 mL of 10× YNB, 2 mL of 500× biotin, 100 mL of 1 M potassium phosphate, 50 mL of 10% 15 N-ammonium chloride and 100 mL of 6% D-[13C6]glucose to the autoclaved water for a total volume of 1 L. If needed, 5 mL of 200× PTM1 trace salts solution can be added to this medium. 14. Antifoaming agent (Subheading 2.1). 15. 50% (w/v) ethanol. Ethanol is just necessary for immersing air outlet port of the fermentation system.

26

T. Sugiki et al.

2.4. Uniform 2H, 15 N-Labeling Using K. lactis

1. YPD medium (Subheading 2.3). 2. YCB-acetamide agar plates (Subheading 2.3): If needed, prepare the plates with 2H2O (see Note 4). 3. 10× YNB(2H2O) (Subheading 2.2). 4. 500× biotin(2H2O) (Subheading 2.2). 5. 1 M potassium phosphate(2H2O) (Subheading 2.2). 6. 200× PTM1 trace salts(2H2O) (Subheading 2.2). 7. 10% 15N-ammonium chloride(2H2O): Dissolve 7.5 g of 15 N-ammonium chloride in 75 mL of 2H2O (see Note 2) and sterilize by aseptic filtration. Prepare this solution just before use. 8. 4% D-[2H7]glucose(2H2O): Dissolve 2 g of D-[2H7]glucose in 50 mL of 2H2O and sterilize by aseptic filtration. Prepare this solution just before use. 9. 2H, 15N-BMD(2H2O): To prepare 0.5 L of uniformly 15 N-labeled buffered minimal medium containing D-[2H7]glucose in 2H2O, sterilize 325 mL of 2H2O by aseptic filtration. Aseptically add 50 mL of 10× YNB(2H2O), 1 mL of 500× biotin(2H2O), 50 mL of 1 M potassium phosphate(2H2O), 25 mL of 10% 15N-ammonium chloride(2H2O), 50 mL of 4% D-[2H7]glucose(2H2O), and 0.25–0.50 mL of antifoaming agent to the sterilized 2H2O for a final volume of 500 mL. If needed, 2.5 mL of 200× PTM1 trace salts(2H2O) can be added to this medium. 10. 6% D-[2H7]glucose(2H2O): Dissolve 6 g of D-[2H7]glucose in 100 mL of 2H2O and sterilize by aseptic filtration. Prepare this solution just before use. 11. 2H, 15N-BMD(2H2O) feeding medium: Sterilize 650 mL of 2 H2O by aseptic filtration. Aseptically add 100 mL of 10× YNB(2H2O), 2 mL of 500× biotin(2H2O), 100 mL of 1 M potassium phosphate(2H2O), 50 mL of 10% 15N-ammonium chloride(2H2O), 100 mL of 6% D-[2H7]glucose(2H2O), and 0.5–1.0 mL of antifoaming agent to the sterilized 2H2O for a final volume of 1 L. If needed, 5 mL of 200× PTM1 trace salts(2H2O) can be added to this medium. 12. Antifoaming agent (Subheading 2.1). 13. 50% (w/v) ethanol. 13. 50% (w/v) ethanol. Ethanol is just necessary for immersing air outlet port of the fermentation system.

3. Methods 3.1. Uniform 13C, 15 N-Labeling Using P. pastoris

1. Prepare necessary media. (a) Using aluminum foil, loosely cover the open lid of an empty 1-L baffled flask and autoclave it.

2

Isotopic Labeling of Heterologous Proteins in the Yeast…

27

(b) Prepare 1.2 L of fresh, sterile 15N-BM medium in a 2-L medium bottle and 20 mL of 5% D-[13C6]glucose. (c) Aseptically combine 180 mL of fresh 15N-BM medium, 20 mL of 5% D-[13C6]glucose and 0.1–0.2 mL of antifoaming agents in the sterile 1-L baffled flask to make 13 C, 15N-BMD medium. (d) Using aluminum foil, loosely mask the open lid of the empty 2-L fermentation vessel. Clamp all of the lines that could come into contact with the growth medium and autoclave the vessel. (e) Aseptically combine 1 L of fresh 15N-BM medium, 0.5–1.0 mL of antifoaming agent and 5 mL of 13C-methanol into the sterile empty 2-L fermentation vessel, to make 13 C, 15N-BMM medium. 2. Using a sterile loop, scoop a small aliquot of P. pastoris transformants from a frozen glycerol stock, and streak it onto a YPDS agar plate containing Zeocin™. Incubate the plate for 24–48 h at 30°C. 3. To produce the primary culture, inoculate a fresh, single colony of the P. pastoris transformants into 5 mL of BMGY medium (in a 50-mL Corning tube) and shake at 200–250 rpm at 30°C for 18–24 h. 4. Pellet the primary culture cells by centrifugation at 3,000 × g for 5 min at 20°C and discard the supernatant. Gently resuspend the pellet with 30 mL of fresh, sterile 13C, 15N-BMD medium, and pour the resuspended cells into the remaining 13 C, 15N-BMD medium in the baffled flask for a final volume of 200 mL (see Note 5). Shake the flask at > 200 rpm at 30°C until the cell density reaches an OD600 of 4–6. 5. Pellet the cells in sterile tubes by centrifuging at 2,000 × g for 10 min at 20°C and discard the supernatant. Gently resuspend the pellet with 30 mL of fresh, sterile 13C, 15N-BMM medium, and pour the resuspended cells into the remaining fresh, sterile 13 C, 15N-BMM medium in the 2-L fermentation vessel to reach an OD600 of approximately 1 (see Notes 6–8). 6. Assemble the fermentation system. Aseptically attach the pH and dissolved O2 sensors to the fermentation vessel and connect the probes to their respective controllers. Connect the outlet port of the air-feeding tube to the air flow inlet port of the vessel. Aseptically attach a Liebig condenser to the outlet port of the vessel, and immerse the air-exhaust port of the Liebig condenser under 1 L of 50 % (w/v) ethanol solution. Insert the temperature probe into the vessel and connect it to a temperature controller, such as a circulating water bath. 7. Agitate the culture medium at 300–800 rpm with 0.1–0.3 L/ min of feeding air at 30°C (see Notes 9 and 10). Add

28

T. Sugiki et al. 13

C-methanol (0.5 % (w/v) of the total volume) to the culture medium every 24 h during this induction phase (20) (see Note 11). The total amount of 13C-methanol required for sufficient protein expression in a cost-effective manner is approximately 10–20 mL per 1 L culture medium (see Note 12). 8. Pellet the cells by centrifuging at 6,000 × g for 20 min at 4°C. If the target protein is secreted into the culture medium, filter the supernatant by passing it through a < 0.45-μm pore size membrane and collect the filtrate. If needed, add appropriate protease inhibitors to the filtrate to prevent proteolytic degradation of secreted target proteins. If the target protein is expressed in the cytoplasm of the host cells, discard the supernatant after centrifugation and retain the cell pellet. 3.2. Uniform 2H, 15 N-Labeling Using P. pastoris

Deuterium isotope-labeling of target proteins is one of the most important techniques for protein NMR studies, especially for analyses of large molecular weight (MW > 25 K) proteins and in crosssaturation experiments to identify intermolecular-binding sites (6, 21–24). The P. pastoris expression system can be used to overexpress deuterium-labeled heterologous proteins. For the efficient production of deuterated target proteins, cells should be adapted to grow in deuterated broth medium. Adaptation is achieved by multistage subculturing in which the deuterium concentration is raised in a stepwise fashion. For instance, subculturing of cells is performed using semideuterated (25–95% 2H2O) medium, with subsequent culturing in fully deuterated medium (14) (see Note 4). In the case of uniform 2H-labeling, using a 2 H-labeled carbon source (such as [2H4]methanol) only during the induction phase is insufficient for producing fully deuterated target proteins (a considerable amount of protons remain on methyl and Lys δ/ε groups) (14). To achieve the nearly-complete deuteration level required for cross-saturation experiments, a deuterium-labeled carbon source, such as D-[2H7]glucose should also be used during the cell growth phase (prior to the induction phase) (7, 14). In this section, we describe a cultivation procedure for overexpressing perdeuterated heterologous proteins in P. pastoris. Using this procedure, Ichikawa and coworkers successfully prepared a discoidin domain of DDR2 that was more than 95% deuterated, and clearly identified the collagen-binding site of DDR2 through transferred cross-saturation experiments (7). 1. Prepare necessary media: (a) Autoclave a 1-L baffled flask and a 2-L fermentation vessel, and dry them in an oven at 60–80°C (for at least 48 h) to completely remove residual H2O. (b) Prepare 1.3 L of 15N-BM(2H2O) medium in a 2-L media bottle, 10 mL of 10% unlabeled D-glucose(2H2O), and 15 mL of 10% D-[2H7]glucose(2H2O) solutions.

2

Isotopic Labeling of Heterologous Proteins in the Yeast…

29

(c) Prepare 5 mL each of 15N-BMD(90% 2H2O), 15N-BMD (2H2O), and 2H, 15N-BMD(2H2O) media in 50-mL Corning tubes. (d) Prepare 300 mL of 2H, 15N-BMD(2H2O) medium in a sterilize, well-dried 1-L baffled flask. (e) Prepare 2H, 15N-BMM(2H2O) medium by mixing 1 L of 15 N-BM(2H2O) medium, 0.5–1.0 mL of antifoaming agent and 5 mL of [2H4]methanol in the sterilize, well-dried 2-L fermentation vessel. 2. Using a sterile loop, scoop a small aliquot of P. pastoris transformants from a frozen glycerol stock and streak it onto a YPDS agar plate containing Zeocin™. Incubate the plate for 24–48 h at 30°C. 3. To produce the primary culture, inoculate 5 mL of BMGY medium in a sterile 50-mL Corning tube with a fresh single colony of the P. pastoris transformant from the YPDS agar plate and shake at 200–250 rpm at 30°C for 18–24 h. 4. Add 0.1 mL of the primary culture to 5 mL of fresh, sterile 15 N-BMD(90% 2H2O) medium in a sterile 50-mL Corning tube. Shake the culture tube at 200–300 rpm at 30°C until the cell density reaches an OD600 of approximately 3–5. 5. Add 0.1 mL of the step 4 culture to 5 mL of fresh, sterile 15 N-BMD(2H2O) medium (in a sterile 50-mL Corning tube). Shake the culture tube at 200–300 rpm at 30°C until the cell density reaches an OD600 of 3–5. 6. Add 0.1 mL of the step 5 culture to 5 mL of fresh, sterile 2H, 15 N-BMD(2H2O) medium in a sterile 50-mL Corning tube. Shake the culture tube at 200–300 rpm at 30°C until the cell density reaches an OD600 of 3–5. 7. Pellet the cells by centrifuging at 3,000 × g for 5 min at 20°C and discard the supernatant. Gently resuspend the pellet with 30 mL of fresh, sterile 2H, 15N-BMD(2H2O) medium and pour the resuspended cells into the remaining 2H, 15N-BMD(2H2O) medium in the 1-L baffled flask to a final volume of 300 mL. Shake the flask at > 200 rpm at 30°C until the cell density reaches an OD600 of 3–4. 8. Pellet the cells in sterile tubes by centrifuging at 2,000 × g for 10 min at 20°C and discard the supernatant. Gently resuspend the pellet with 30 mL of fresh, sterile 2H, 15N-BMM(2H2O) medium and pour the resuspended cells into the remaining sterile 2H, 15N-BMM(2H2O) medium in the 2-L fermentation vessel. The starting OD600 of the culture in the fermenter vessel should be approximately 1 (see Notes 6–8). 9. Assemble the fermentation system Subheading 3.1, step 6 (see Note 13).

as

described

in

30

T. Sugiki et al.

10. Agitate the culture medium at 300–800 rpm with 0.1–0.3 L/min of feeding air at 30°C (see Notes 9 and 10). Add [2H4] methanol (0.5 % (w/v) of the total volume) to the culture medium every 24 h during this induction phase (see Notes 11 and 12). 11. Collect the target proteins as described in Subheading 3.1, step 8. 3.3. Uniform 13C, 15 N-Labeling Using K. lactis

1. Prepare 0.5 L of sterile 13C, 15N-BMD and 1.0 L of 13C, 15N-BMD feeding medium in a 2-L fermentation vessel and 1-L media bottle, respectively. 2. Using a sterile loop, scoop a small aliquot of K. lactis transformants from a frozen glycerol stock and streak it onto a YCBacetamide agar plate. Incubate the plate for 24–48 h at 30°C. 3. To produce the primary culture, inoculate a fresh, single colony of K. lactis transformants into 5 mL of YPD medium (in a 50-mL Corning tube) and shake at 200–250 rpm at 30°C for 48 h to obtain a saturated biomass (OD600 > 20–30). 4. Assemble the fermentation system as described in Subheading 3.1, step 6. Connect the media bottle containing 1 L of fresh 13C, 15N-BMD feeding medium to the inlet port of the vessel with an appropriate length of feeding tube, and attach a peristaltic pump at the midpoint of the feeding tube (Fig. 1). 5. Pellet the primary culture cells by centrifuging at 3,000 × g for 5 min at 20°C and discard the supernatant. Gently resuspend the pellet with 30 mL of fresh, sterile 13C, 15N-BMD medium from the 2-L fermentation vessel, and pour the resuspended cells into the remaining 13C, 15N-BMD medium in the fermentation vessel. Agitate the culture medium at 600–800 rpm with 0.1–0.3 L/min feeding air at 30°C (see Notes 9, 10, and 12). 6. During the fermentation, continuously feed fresh, sterile 13C, 15 N-BMD feeding medium into the fermentation vessel at a constant flow rate of 8.3 mL/h using the peristaltic pump (see Note 14). 7. Pellet the cells by centrifuging at 10,000 × g for 15 min at 4°C. If the target protein is secreted into the culture medium, filter the supernatant by passing it through a < 0.45-μm pore size membrane and collect the filtrate. If needed, add appropriate protease inhibitors to the collected filtrate to prevent proteolytic degradation of the target protein. If the target protein is expressed in the cytoplasm of the host cells, discard the supernatant after centrifugation, and retain the cell pellet.

3.4. Uniform 2H, 15 N-Labeling Using K. lactis

In our experience, perdeuteration (approximately 90% deuteration estimated by MS analysis) of maltose-binding protein (MBP) is successfully achieved using K. lactis with fed-batch fermentation.

2

Isotopic Labeling of Heterologous Proteins in the Yeast…

31

Fig. 1. Overview of the fed-batch fermentation system. (a) Schematic diagram of fed-batch fermentation. (b) Photograph of the assembled fed-batch fermentation device. a: fresh media which is continuously fed into the fermentation vessel, b: peristaltic pump, c: fermentation vessel, d: air flow inlet, and e: air exhaust port.

Furthermore, insertion of one or two subcultivation steps following the primary culture improves cell growth and expression of MBP with a highly efficient (> 92%) level of deuteration (Miyazawa-Onami and coworkers, unpublished data). 1. Prepare necessary media: (a) Autoclave an empty 1-L media bottle and a 2-L fermentation vessel, and dry them in an oven at 60–80°C (for at least 48 h) to completely remove residual H2O. (b) Prepare 0.5 L of fresh, sterile 2H, 15N-BMD(2H2O) and 1.0 L of 2H, 15N-BMD(2H2O) feeding medium in the sterile 2-L fermentation vessel and 1-L media bottle, respectively. 2. Using a sterile loop, scoop a small aliquot of K. lactis transformants from a frozen glycerol stock and streak it onto a YCBacetamide agar plate. Incubate the plate for 24–48 h at 30°C.

32

T. Sugiki et al.

3. To produce the primary culture, inoculate 5 mL of YPD medium in a 50-mL Corning tube using a fresh single colony of the K. lactis transformants from the YCB agar plate and shake at 200–250 rpm at 30°C for 48 h to obtain a saturated biomass (OD600 > 20–30). 4. Assemble the fermentation system as described in Subheading 3.1, step 6 (see Note 13). Connect the bottle containing 1 L of fresh 2H, 15N-BMD(2H2O) feeding medium to the inlet port of the vessel with an appropriate length of feeding tube, and attach a peristaltic pump at the midpoint of the feeding tube (Fig. 1). 5. Pellet the cells by centrifuging at 3,000 × g for 5 min at 20°C and discard the supernatant. Gently resuspend the pellet with 30 mL of fresh, sterile 2H, 15N-BMD(2H2O) medium from the 2-L fermentation vessel, and pour the resuspended cells into the remaining 2H, 15N-BMD(2H2O) medium in the fermentation vessel. Agitate the culture medium at 600–800 rpm with 0.1–0.3 L/min feeding air at 30°C (see Notes 9, 10, and 12). 6. During the fermentation, continuously feed fresh, 2H, 15 N-BMD(2H2O) feeding medium from the 1-L media bottle into the fermentation vessel at a constant flow rate of 8.3 mL/h using the peristaltic pump (see Note 14). 7. Collect the target proteins as described in Subheading 3.3, step 7.

4. Notes 1. Detailed information about the composition of YNB without amino acids and ammonium sulfate can be found in the Becton Dickinson Difco Yeast Media Recipes, available at http://www. bd.com/ds/technicalCenter/inserts/Yeast_Media.pdf. 2. Ammonium sulfate or ammonium chloride is used as a nitrogen source by the yeast. In many cases, 5–10 g/L of ammonium sulfate or ammonium chloride are used in the culture of yeast. In our experience, a sufficient amount of isotope-labeled protein could be expressed while keeping costs to a minimum by using 2 g/L of 15N-ammonium chloride (7). 3. Foaming of the culture medium caused by agitation and aeration will severely affect the yield of target protein due to a reduction in the level of protein expression and partial denaturation of the expressed proteins (25). Foaming should be controlled

2

Isotopic Labeling of Heterologous Proteins in the Yeast…

33

by adding antifoaming agents to the culture medium. We use 0.05–0.10% (w/v) of “Antifoam 204” (supplied from Sigma). 4. For many organisms, 2H2O-containing medium negatively affect physiological processes. In many cases, stagnation of cell growth occurs in 75–100% 2H2O, necessitating adaptation to the deuterated medium by subcultivating the cells several times prior to the induction of protein expression (2). Furthermore, the selection of transformants using deuterated agar plates may be effective in generating special transformants that are more adaptable to deuterated medium. 5. Although glycerol is widely used as a carbon source of P. pastoris during growth phase, in the case of 13C-labeling, 13C-glycerol can be replaced by 13C-glucose to reduce labeling costs (9, 26, 27). 6. An OD600 = 1 is equivalent to 5 × 107 P. pastoris cells (2). 7. Nonmethanolic carbon sources, especially glucose, should be completely eliminated (or consumed) prior to the addition of methanol to achieve full induction efficiency of the AOX1 promoter (2, 7, 20). However, the use of mixtures of carbon sources during the induction phase to improve cell growth and the production of target proteins has been reported (28–32). 8. The optimal starting point of the induction phase should be determined by performing a time course study of cell growth in a “cold-run” prior to beginning a “hot-run,” since the optimal point for starting induction varies according to the type of target protein and culture conditions. 9. The level of oxygen supplied to the fermentation medium should be kept at 0.1–0.3 vvm (volume of oxygen (liters) per volume of fermentation culture (liters) per minute) during the cultivation. This level can be easily achieved using any glass fermenter and mixing the broth medium via an impeller in an air feeding condition. Maintaining an adequate concentration of dissolved O2 during cultivation is crucially important for sufficient cell growth and protein expression (2, 20, 33). 10. Proteolytic degradation of heterologous proteins is one of the drawbacks of the secretory expression strategy. In our experience, this often occurs in a culture grown in a poor-nutrient medium such as the one that may be found under isotope labeling conditions. To prevent extensive proteolytic degradation of target proteins, the following measures can be taken (1) optimizing the pH range of the culture medium (34); (2) cultivating at low-temperature (35–38); (3) adding extracts of algae, such as BioExpress. (26); (4) adding protease inhibitors (2, 27, 39); however, some types of protease inhibitors are toxic; and (5) utilizing protease-deficient strains (34, 40). Several

34

T. Sugiki et al.

protease-deficient strains of P. pastoris and K. lactis are commercially available. 11. In the P. pastoris expression system, the methanol concentration during induction directly affects cell growth and protein production since an excess of methanol (> 1–2% (w/v)) and an excessive rate of feeding is toxic to cells (5). Monitoring the level of dissolved O2 in the culture medium is a convenient method for assessing the timing for feeding and the appropriate dose of methanol because sharp “spike” signals are detected by the O2 electrode when the carbon sources are completely consumed and/or when the concentration of methanol reaches a toxic level (20, 33). 12. The typical induction time is between 48 and 96 h. However, in the case of a deuterium-labeling culture, an induction period longer than 96 h is typically required. The optimal induction period and the amount of methanol should be determined by following the time course of target protein expression level using SDS-PAGE. 13. To achieve a higher level of deuteration, a glass tube filled with dried granular calcium chloride should be connected in front of the air inlet of the fermentation vessel to remove H2O from the supplied air. 14. The optimal culture conditions (composition of media, pH, temperature, etc.) vary according to the type of host and target proteins. In our experience, the most critical factors for obtaining sufficient protein expression using K. lactis are the concentration of the carbon source and the concentration of dissolved oxygen. Especially, in cases involving the use of minimal media for isotope labeling, we recommend fermentation cultivation rather than flask culture. Furthermore, addition of isotopically labeled extracts of algae, such as Celtone, BioExpress (both supplied by Cambridge Isotope Laboratories, Inc.) or C.H.L. (Chlorella Industry Co., Ltd.) significantly improves cell growth and protein expression (41).

Acknowledgments This work was financially supported in part by the Ministry of Economy, Trade and Industry (METI) and the New Energy and Industrial Technology Development Organization (NEDO).

2

Isotopic Labeling of Heterologous Proteins in the Yeast…

35

References 1. Cregg, J. M., Vedvick, T. S., and Raschke, W. C. (1993) Recent advances in the expression of foreign genes in Pichia pastoris. Biotechnology 11, 905–910. 2. Pickford, A. R., and O’Leary, J. M. (2008) Isotopic labeling of recombinant proteins from the methylotrophic yeast Pichia pastoris, in Methods in Molecular Biology (Downing, A. K., Ed.), vol. 278, pp. 17–33, Humana Press, Totowa, NJ. 3. Lin Cereghino, G. P., Cereghino, J. L., Ilgen, C., and Cregg, J. M. (2002) Production of recombinant proteins in fermenter cultures of the yeast Pichia pastoris. Curr. Opin. Biotechnol. 13, 329–332. 4. Daley, R., and Hearn, M. T. (2005) Expression of heterologous proteins in Pichia pastoris : a useful experimental tool in protein engineering and production. J. Mol. Recognit. 18, 119–138. 5. Cos, O., Ramόn, R., Montesinos, J. L., and Valero F. (2006) Operational strategies, monitoring and control of heterologous protein production in the methylotrophic yeast Pichia pastoris under different promoters: A review. Microbial Cell Factories 5, 17–36. 6. Takahashi, H., and Shimada, I. (2010) Production of isotopically labeled heterologous proteins in non-E. coli prokaryotic cells. J. Biomol. NMR 46, 3–10. 7. Invitrogen Corporation. EasySelect™ Pichia Expression Kit: A manual of methods for expression of recombinant proteins using pPICZ and pPICZα in Pichia pastoris, Version I. Available at http://tools.invitrogen.com/content/sfs/manuals/easyselect_man.pdf. 8. Laroche, Y., Strome, V., De Meutter, J., Messens, J., and Lauwereys, M. (1994) Highlevel secretion and very efficient isotopic labeling of tick anticoagulant peptide (TAP) expressed in the methylotrophic yeast, Pichia pastoris. Bio/Technol 12, 1119–1124. 9. Denton, H., Smith, M., Husi, H., Uhrin, D., Barlow, P. N., Batt, C. A. , and Sawyer, L. (1998) Isotopically labeled bovine β-lactoglobulin for NMR studies expressed in Pichia pastoris. Protein Expr. Purif. 14, 97–103. 10. Wood, M. J., and Komives, E. A. (1999) Production of large quantities of isotopically labeled protein in Pichia pastoris by fermentation.. J. Biomol. NMR 13, 149–159. 11. Mine, S., Ueda, T., Hashimoto, Y., Tanaka, Y., and Imoto, T. (1999) High-level expression of uniformly 15N-labeled hen lysozyme in Pichia pastoris and identification of the site in hen

lysozyme where phosphate ion binds using NMR measurements. FEBS Lett. 448, 33–37. 12. van den Burg, H. A., de Wit, P. J., and Vervoort, J. (2001) Efficient 13C/15N double labeling of the avirulence protein AVR4 in a methanolutilizing strain (Mut+) of Pichia pastoris. J. Biomol. NMR 20, 251–261. 13. Rodriguez, E., and Krishna, N. R. (2001) An economical method for 13C/15N isotopic labeling of proteins expressed in Pichia pastoris. J. Biochem. 130, 19–22. 14. Morgan, W. D., Kragt, A., and Feeney, J. (2000) Expression of deuterium-isotope-labelled protein in the yeast Pichia pastoris for NMR study. J. Biomol. NMR 17, 337–347. 15. Ichikawa, M., Osawa, M., Nishida, N., Goshima, N., Nomura, N., and Shimada, I. (2007) Structural basis of the collagen-binding mode of discoidin domain receptor 2, EMBO J. 26, 4168–4176. 16. New England Biolabs, Inc. K. lactis Protein Expression Kit: Instruction manual. Available at http:/ /www.neb.com/nebecomm/ManualFiles/ manualE1000.pdf. 17. Colussi, P. A., and Taron, C. H. (2005) Kluyveromyces lactis LAC4 promoter variants that lack function in bacteria but retain full function in K. lactis, Appl. Environ. Microbiol. 71, 7092–7098. 18. Read, J. D., Colussi, P. A., Ganatra, M. B., and Taron, C. H. (2007) Acetamide selection of Kluyveromyces lactis cells transformed with an integrative vector leads to high-frequency formation of multicopy strains. Appl. Environ. Microbiol. 73, 5088–5096. 19. Sugiki, T., Shimada, I., and Takahashi, H. (2008) Stable isotope labeling of protein by Kluyveromyces lactis for NMR study. J. Biomol. NMR 42, 159–162. 20. Invitrogen Corporation. Pichia Fermentation Process Guidelines. Available at http://toolsja. invitrogen.com/content/sfs/manuals/pichiaferm_prot.pdf. 21. Takahashi, H., Nakanishi, T., Kami, K., Arata, Y., and Shimada, I. (2000) A novel NMR method for determining the interface of large protein-protein complexes. Nat. Struct. Biol. 7, 220–223. 22. Nakanishi, T., Miyazawa, M., Sakakura, M., Terasawa, H., Takahashi, H., and Shimada, I. (2002) Determination of the interface of a large protein complex by transferred cross-saturation measurements. J. Mol. Biol. 318, 245–249. 23. Takahashi, H., Miyazawa, M., Ina, Y., Fukunishi, Y., Mizukoshi, Y., Nakamura, H., and Shimada,

36

T. Sugiki et al.

I. (2006) Utilization of methyl proton resonances in cross-saturation measurement for determining the interfaces of large protein-protein complexes. J. Biomol. NMR 34, 167–177. 24. Shimada, I., Ueda, T., Matsumoto, M., Sakakura, M., Osawa, M., Takeuchi, K., Nishida, N., and Takahashi, H. (2008) Cross-saturation and transferred cross-saturation experiments. Prog. Nucl. Magn. Reson. Spectrosc. 54, 123–140. 25. Koch, V., Rüffer, H.-M., Schügerl, K., Innertsberger, E., Menzel, H., and Weis, J. (1995) Effect of antifoam agents on the medium and microbial cell properties and process performance in small and large reactors. Process Biochem. 30, 435–446. 26. Macauley-Patrick, S., Fazenda, M. L., McNeil, B., and Harvey, L. M. (2005) Heterologous protein production using the Pichia pastoris expression system. Yeast 22, 249–270. 27. Shapiro, R. I., Wen, D., Levesque, M., Hronowski, X., Gill, A., Garber, E. A., Galdes, A., Strauch, K. L., and Taylor, F. R. (2003) Expression of sonic hedgehog-Fc fusion protein in Pichia pastoris. Identification and control of post-translational, chemical, and proteolytic modifications. Protein Expr. Purif. 29, 272–283. 28. Files, D., Ogawa, M., Scaman, C. H., and Baldwin, S. A. (2001) A Pichia pastoris fermentation process for producing high-levels of recombinant human cystatin-C. Enzyme Microbial Technol. 29, 335–340. 29. Zhang, W., Hywood Potter K. J., Plantz, B. A., Schlegel, V. L., Smith, L. A., and Meagher, M. M. (2003) Pichia pastoris fermentation with mixed-feeds of glycerol and methanol: growth kinetics and production improvement. J. Ind. Microbiol. Biotechnol. 30, 210–215. 30. Xie, J., Zhang, L., Ye, Q., Zhou, Q., Xin, L., Du, P., and Gan, R. (2003) Angiostatin production in cultivation of recombinant Pichia pastoris fed with mixed carbon sources. Biotechnol. Lett. 25, 173–177. 31. McGrew, J. T., Leiske, D., Dell, B., Klinke, R., Krasts, D., Wee, S. F., Abbott, N., Armitage, R., and Harrington, K. (1997) Expression of trimeric CD40 ligand in Pichia pastoris: use of a rapid method to detect high-level expressing transformants. Gene 187, 193–200. 32. d’Anjou, M. C., and Daugulis, A. J. (2001) A rational approach to improving productivity in

Pichia pastoris fermentation. Biotechnol. Bioeng 72, 1–11. 33. Cai, M., Huang, Y., Sakaguchi, K., Clore, G. M., Gronenborn, A. M., and Craigie, R. (1998) An efficient and cost-effective isotope labeling protocol for proteins expressed in Escherichia coli. J. Biomol. NMR 11, 97–102. 34. Zhang, Y., Liu, R., and Wu, X. (2007) The proteolytic systems and heterologous proteins degradation in the methylotrophic yeast Pichia pastoris. Annal. Microbiol. 57, 553–560. 35. Li, Z., Xiong, F., Lin, Q., d’Anjou, M., Daugulis, A. J., Yang, D. S., and Hew, C. L. (2001) Lowtemperature increases the yield of biological active herring antifreeze protein in Pichia pastoris. Protein Expr. Purif. 21, 438–445. 36. Jahic, M., Gustavsson, M., Jansen, A.-K., Martinelle, M., and Enfors, S.-O. (2003) Analysis and control of proteolysis of a fusion protein in Pichia pastoris fed-batch processes. J. Biotechnol. 102, 45–53. 37. Jahic, M., Wallberg, F., Bollok, M., Garcia, P., and Enfors, S.-O. (2003) Temperature limited fed-batch technique for control of proteolysis in Pichia pastoris bioreactor cultures. Microbial Cell Factories 2, 6–17. 38. Surribas, A., Stahn, R., Montesinos, J. L., Enfors, S.-O., Valero, F., and Jahic, M. (2007) Production of Rhizopus oryzae lipase from Pichia pastoris using alternative operational strategies. J. Biotechnol. 130, 291–299. 39. Shi, X., Karkut T., Chamankhah, M., AltingMees, M., Hemmingsen, S. M., and Hegedus, D. (2003) Optimal conditions for the expression of a single-chain antibody (scFv) gene in Pichia pastoris. Protein Expr. Purif. 28, 321–330. 40. Yao, X. Q., Zhao, H. L., Xue, C., Zhang, W., Xiong, X. H., Wang, Z. W., Li, X. Y., and Liu, Z. M. (2009) Degradation of HSAAX15(R13K) when expressed in Pichia pastoris can be reduced via the disruption of YPS1 gene in this yeast. J. Biotechnol. 139, 131–136. 41. Madduri, K., Badger, M., Li, Z.-S., Xu, X., Thornburgh, S., Evans, S., and Dhadialla, T. S. (2009) Development of stable isotope and selenomethionine labeling methods for proteins expressed in Pseudomonas fluorescens. Protein Expr. Purif. 65, 57–65.

Chapter 3 Isotope Labeling in Insect Cells Krishna Saxena, Arpana Dutta, Judith Klein-Seetharaman, and Harald Schwalbe Abstract Recent years have seen remarkable progress in applying nuclear magnetic resonance (NMR) spectroscopy to proteins that have traditionally been difficult to study due to issues with folding, posttranslational modification, and expression levels or combinations thereof. In particular, insect cells have proved useful in allowing large quantities of isotope-labeled, functional proteins to be obtained and purified to homogeneity, allowing study of their structures and dynamics by using NMR. Here, we provide protocols that have proven successful in such endeavors. Key words: Isotope labeling, Baculovirus, Nuclear magnetic resonance, Recombinant protein expression, Insect cells

1. Introduction: Baculovirus-Insect Cell Expression System

Isotope labeling of proteins represents an important and often required tool for the application of nuclear magnetic resonance (NMR) spectroscopy to investigate the structure and dynamics of proteins. So far, the great majority of isotope-labeled proteins have been expressed in Escherichia coli (1) because of the ease of cloning and expressing proteins at low cost. When possible, protein production should be performed in a prokaryotic system (i.e., E. coli), since this strategy is the most cost-effective and allows the most flexibility. However, human or complex proteins often cannot be expressed in E. coli in an active, correctly folded, posttranslationally modified form (glycosylated, phosphorylated, etc.). The capacity of E. coli for protein folding and forming disulfide bonds is not

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_3, © Springer Science+Business Media, LLC 2012

37

38

K. Saxena et al.

sufficient for many recombinant proteins, although there are a number of new developments to overcome these limitations: 1. Decreasing the temperature of the cell culture (2, 3) 2. Coexpressing molecular chaperones (4) 3. Fusing highly soluble tags (gst, mbp, trxa, nusa, sumo, etc.) to the target proteins (5) 4. Overexpressing (6, 7) the target protein or using an engineered E. coli strain capable of forming disulfide bonds in the cytoplasm (e.g., Shuffle, New England Biolabs) 5. Refolding in vitro However, low-expression yield and solubility problems of the target protein in E. coli often force the change from a bacterial recombinant protein expression system to a eukaryotic host. Numerous eukaryotic-based expression systems are currently available in protein expression laboratories. For NMR use, only three expression hosts are generally considered: yeast (Pichia pastoris (8), Hansenula polymorpha or Kluyveromyces lactis), baculovirusmediated insect cells, and mammalian cells. More recently, the generation of baculoviruses harboring mammalian promoters (BacMams) have extended the use of baculovirus-mediated expression systems (BvE) to the development of gene delivery (9) and expression vectors in mammalian cells (10). BacMams cannot replicate in mammalian cells, which renders them a much safer alternative to conventional virus vectors (11). Yeast offers a powerful, simple system for expressing recombinant proteins. Besides the capability of performing posttranslational modification to the recombinant protein, the main advantage of these expression hosts is the feasibility of isotope labeling in simple minimal-defined media. Therefore, the costs of 15N, 13C, or 2 H uniform isotope incorporation are negligible in comparison to the costs of other eukaryotic cell media. Higher eukaryotes need well-defined expression media supplemented with expensive, labeled amino acids. But there are also disadvantages to recombinant protein expression in yeast. Since they provide N- and O-linked high-mannose-type glycans that could be immunogenic in humans (12, 13), the production of glycosylated proteins in yeast is not optimal, especially if glycosylation is required for the biological functionality of the target protein. Additionally, yeast cannot perform tyrosine O-sulfation (14), and proteins whose native forms are nonglycosylated may be hyperglycosylated when expressed in yeast (15). A recent study reported the successful reengineering of the glycosylation pathways in P. pastoris to allow the expression of recombinant proteins with human-type glycans (16). This may allow future improvements in the expression of glycosylated proteins in yeast. However, glycosylation is not the only challenge for successful expression of proteins in yeast. Certain proteins are

3

Isotope Labeling in Insect Cells

39

expressed at low levels even when glycosylation is not necessarily an issue (17–19). While the reasons are not fully clear, low-expression yields have been attributed to defects in the ER folding machinery (20–22). Even for proteins in which high-expression yields can be obtained, much of the protein may be misfolded (23). In order to detect which part of the folding machinery may be responsible, protein disulfide isomerase (PDI), an enzyme that catalyzes disulfide exchange in the ER, was overexpressed and found to enhance protein yields (24, 25). On the other hand, human adenosine A2A receptor levels were not found to increase with an increase in PDI expression (26). Thus, the process of disulfide bond formation in yeast remains uncertain. Finally, the case is most uncertain for membrane proteins, where there is a lack of knowledge on folding mechanisms and also on the way chaperones and the translocon participate in the folding pathway. Therefore, it is currently recommended to use higher eukaryotic hosts with advanced cell machinery systems for the production of recombinant proteins that have to be glycosylated, disulfide bonded, and/or membrane inserted for functional activity. Advanced expression systems with higher cell machineries for posttranslational modifications are offered by the baculovirusmediated expression system in insect cells (BvE) and mammalian expression methods. The BvE is one of the most efficient and popular systems among the eukaryotic hosts to use for expressing recombinant proteins. Therefore, its application is widespread in industrial as well as in academic environments for structural and functional studies of diverse proteins. However, for the production of therapeutic recombinant proteins, mammalian expression systems (mainly Chinese hamster ovary cells (CHO) and Human embryonic kidney (HEK) 293 cells) are required. Since CHO and HEK293 cells have been extensively characterized and have the ability for human-like glycosylation, the production of recombinant therapeutic monoclonal antibodies and Fc fusion proteins in this host is safe. The rate of production of therapeutic proteins, the largest class of new products being developed by the biopharmaceutical industry, has increased significantly in recent years (27). Mammalian expression systems have conventionally been considered to be too weak and inefficient for protein expression. However, recent advances have significantly improved the expression levels of these systems. This chapter and the following one, therefore, attempt to provide an overview of some of the recent developments in expression strategies for baculovirus-mediated insect cell and mammalian expression systems in view of NMR investigations. Since NMR requires isotope-labeled protein, the focus of this article is directed toward strategies to incorporate stable isotopes (15N, 13C) into the target protein in insect and mammalian cells (see Chapter 4). A major bottleneck of uniform labeling in higher eukaryotes is the high cost of complex medium

40

K. Saxena et al.

with labeled amino acids. Another limitation of these hosts is that they cannot survive in deuterium oxide (D2O)-containing media, so cost-effective generation of perdeuterated proteins is not available, either for insect or for mammalian cell systems. It is beyond the scope of both chapters to discuss features of the BvE or mammalian expression systems in detail. Comprehensive guides and detailed methodologies for the construction and analysis of recombinant baculovirus for insect cell expression, maintenance of insect cells in culture, and analysis of recombinant protein expression can be found elsewhere (28, 29), including in Baculovirus Manuals from Invitrogen, Pharmingen, Novagen, and others. However, given the number of different BvE strategies, we begin this chapter by explaining the principle of BvE and subsequently give a short survey of BvE options. 1.1. Principle of Baculovirus-Mediated Recombinant Protein Expression

The insect cell baculovirus-mediated expression system (BvE) is a powerful platform to rapidly produce high levels of recombinant proteins (see ref. 28 for an excellent review). Unlike bacterial expression hosts, the baculovirus system relies on a eukaryotic expression system and thus offers protein modification, sorting, and transportation machineries similar to those found in higher eukaryotic organisms. Baculoviruses are insect viruses that predominantly infect insect larvae of the order Lepidoptera (butterflies and moths) (30). A baculovirus expression vector is a recombinant baculovirus that has been genetically modified to contain a foreign gene of interest, which can be expressed in insect cells under the control of a baculovirus gene promoter. The BvE uses a helper-independent virus that can be propagated to high titers in insect cells adapted for growth in suspension cultures, enabling the production of large amounts of protein with relative ease (31). Finally, baculoviruses are noninfectious to vertebrates and their promoters have been shown to be inactive in mammalian cells (32). The most commonly used baculovirus for recombinant protein expression is Autographa californica, a multicapsid nucleopolyhedrovirus (AcMNPV) (33). AcMNPV is a large (130 kb), lytic, double-stranded DNA virus and can accommodate large segments of foreign DNA for the expression of recombinant protein (34). The BvE is based on the infection of cultured insect cells by a recombinant virus vector in which the target DNA (or multiple genes) is integrated under the control of the strong viral polyhedron promoter (28). The polyhedrin gene (polh) is necessary for the formation of polyhedral or occlusion bodies in the cell nucleus, but is nonessential for viral replication in insect cells. Polyhedra are large particles that appear in the nuclei of AcMNPV-infected insect cells. The first recombinant baculoviruses were generated by replacing the viral polyhedrin gene with a foreign gene of interest through homologous recombination (33). Homologous exchange between the flanking sequences common to both DNA molecules facilitates

3

Isotope Labeling in Insect Cells

41

the insertion of the gene of interest into the viral genome at the polh locus, resulting in the production of a recombinant virus genome and allowing the powerful polyhedron promoter to drive protein expression of the foreign gene. Since the efficiency of homologous recombination is quite low, identification, isolation, and selection of recombinant virus were traditionally achieved by labor-intensive, technically demanding plaque assays. Due to the deletion of the polyhedrin gene, the recombinant plaque has a more clearly distinct morphology than the parental virus containing the polh gene. Subsequently, additional rounds of plaque screening are required to separate the desired recombinant virus from the parental wild-type virus. However, discriminating between polyhedron-positive and -negative plaques and isolating recombinant virus turned out to be a serious problem for many investigators who used the BvE for the production of recombinant proteins. Nowadays, these technical issues and the time-consuming plaque purification processes are eliminated. In the next section, some of the key developments in the BvE are presented and discussed. 1.2. Commercially Available Baculovirus Expression Systems

Generally, the baculovirus genome is considered too large to insert a foreign gene directly. In most applications of the BvE, the gene of interest is therefore cloned into a transfer vector, which contains sequences that flank the polyhedron gene in the baculovirus genome. The virus genome and the transfer vector are cotransfected into the insect cells and the gene of interest inserts into the virus genome via homologous recombination (see Fig. 1) under the control of the strong late viral polyhedrin promoter (35). Since a mixture of recombinant and original parental virus is produced after the initial replication, time-consuming plaque purification and isolation are required before protein expression can proceed. BacVector (Merck Biosciences), Baculo-Gold and pBacPAK (BD Biosciences), and Bac-N-Blue (Invitrogen) are commercially available BvEs that use homologous recombination to integrate foreign genes into the virus genome. New developments in generating recombinant virus by using site-specific transpositions (Bac-to-Bac or BaculoDirect, Invitrogen) or progress in recombination methodology with an engineered baculovirus containing a lethal mutation in an essential gene (open reading frame (ORF1629), flashBAC from Oxford Expression Technologies, or BacMagic from EMD Chemicals, Novagen) have facilitated the use of BvE for a larger user community (28). Generally, these improvement strategies can be classified into transfer plasmid modifications and parental baculovirus genome modification (28). The flashBAC and BacMagic are the most promising BvEs so far, since the efficiency of recombination in both systems is 100%. Therefore, these BvEs overcome the requirement of timeconsuming plaque assays and protein expression can be started directly after one or two rounds of virus amplification. This technology

42

K. Saxena et al.

Fig. 1. Construction of baculovirus recombinants with Novagen® BacMagic™ system. This expression system is based on a modified baculovirus genome containing a bacterial artificial chromosome (BAC) at the polyhedrin locus and a partial deletion of the essential ORF1629 viral gene. The BacMagic DNA is mixed with a transfer vector, containing a foreign gene at the polh locus and the complete ORF1629, to generate the recombinant virus via homologous recombination in insect cells. Picture modified from the Novagen manual. © EMD Chemicals Inc., an Affiliate of Merck KGaA. Darmstadt, Germany. BacMagic™ and Novagen® are trademarks of Merck KGaA.

reduces the production of recombinant virus to a one-step procedure, fully amenable to high-throughput and automated production systems. Moreover, this approach is compatible with all baculovirus transfer vectors based on homologous recombination in insect cells at the polyhedrin locus, including several multigene coexpression plasmids. The technology of the flashBAC and BacMagic is driven by a modified bacmid, in which the baculovirus genome AcMNPV with a portion of the essential viral gene (ORF1629) deleted. In addition, a bacterial artificial chromosome (BAC) replaces the polyhedrin-coding region. This combination prevents nonrecombinant virus from replicating in insect cells, yet allows the viral DNA to be propagated as circular DNA in bacteria. This circular viral DNA is then isolated and purified from bacterial cells (flashBAC or BacMagic DNA provided in the kits). Homologous recombination with a compatible expression plasmid (containing the gene of interest

3

Isotope Labeling in Insect Cells

43

flanked by the lef2 and ORF1629 recombination sites) restores the function of the viral ORF1629 allowing the virus DNA to replicate and replaces the BAC sequence with the target coding sequence under the control of the polyhedrin promoter (Fig. 1). Since only recombinant viruses with a restored ORF1629 can replicate, this results in a unique recombinant virus population. This population can then be used directly to infect a larger insect cell culture (50–200 mL) to produce a high-titer working stock. 1.3. Insect Cell Lines

The main insect cell lines used for cotransfections and baculovirus amplification are Spodoptera frugiperda Sf9 or Sf21 (derivatives of the fall armyworm). Trichoplusia ni BTI 5B1-4 (36) (High Five™) cells are generally used for the production of secreted recombinant proteins and not for virus production because of the increased possibility of generating virus mutants (37). Due to the high-mannose and paucimannose types of glycosylation that are obtained in insect cells, no therapeutic protein is currently produced using this system as this would compromise in vivo bioactivity and potentially induce allergenic reactions. Engineering insect cells with glycosyltransferases allows the production of proteins with mammaliantype sugars (38).

1.4. Expression of Labeled Recombinant Protein in the Baculovirus Expression System

Due to the high costs of incorporating stable isotopes into insect cells, it is recommended that the recombinant protein expression is optimized using the chosen BvE before starting with labeled fermentation. There are only a few studies reporting the incorporation of stable isotopes into proteins expressed by baculovirusmediated insect cell expression (39–47). In contrast to the initial trials of amino acid-type selective labeling of proteins in user-defined insect cell media, nowadays there are commercial media (BioExpress-2000, Cambridge Isotope Laboratories, CIL) available for the different labeling applications. Expression of uniformly 13C- 15N-labeled Abelson Kinase domain (13C–15N BioExpress-2000) was the first example of backbone NMR resonance assignments of a recombinant protein expressed using the BvE (48). So far, most uniform labeling protein work in insect cells is only performed by industrial research groups due to the extraordinary costs of the required media. It is not possible to cultivate insect cells in minimal medium, since this host requires essential amino acids for its growth. Reports of selective amino acid isotope labeling in BvE are more frequently cited in the literature, since this approach is easy and fast and does not require expensive medium for labeling. Even in the absence of a backbone assignment of the target protein, structural information can be deduced from the selective labeling approach based on an existing X-ray structure of the protein. From a practical aspect, it should be considered that there are essential and nonessential amino acids (alanine, cysteine, glutamic acid, glutamine, aspartic acid, and asparagine) in insect cells whose content in the medium

44

K. Saxena et al. NH4

Glutamin

Lysine

Aspartate

Asparagine

Glutamate

Tyrosine

Alanine

Phenylalanine

Valine

Leucine

Isoleucine

Glycine

Serine

Fig. 2. A schematic presentation of amino acid metabolism in E. coli and Sf9 with respect to 15N: The black arrows symbolize pathways present in both expression hosts. Pathways that only exist in E. coli are shown in grey. The strength of the arrows reflects the intensity of the conversion. Picture modified from Bruggert et al. (42).

depends on the specific provider of the medium. Before starting with site-specific amino acid labeling, the unlabeled quantity of the desired amino acid in the medium should be checked to calculate the required amount of the amino acid to be labeled. In Fig. 2, a schematic presentation of amino acid metabolism is shown for E. coli and insect cells (42). Interestingly, selective labeling of amino acids in insect cells is more effective than in bacteria, since the amino acid pathways in insect cells do not harbor as many aminotransferases as in prokaryotes, which leads to cross-labeling problems.

2. Materials 2.1. Cotransfection of Insect Cells

1. 35-mm2 tissue culture dishes. 2. Sf9 or Sf21 insect cells. 3. SF900 II insect cell culture medium (serum-free, antibioticfree; Invitrogen). 4. Insect GeneJuice® transfection reagent (Novagen) (see Note 1). 5. BacMagic DNA: 100 ng (5 μL) per cotransfection (20 ng/μL). 6. Sterile baculovirus transfer vector DNA containing the gene under investigation (500 ng per cotransfection). 7. Plastic box to house dishes in the incubator. 8. Sterile pipettes, bijoux (sterile tubes).

3

2.2. Amplification of Recombinant Virus

Isotope Labeling in Insect Cells

45

1. Recombinant virus seed stock (Subheading 3.1). 2. Sf 9 insect cells. 3. SF900 II insect cell culture medium (Subheading 2.1). 4. Inverted phase-contrast microscope.

2.3. Analysis of Recombinant Protein Expression

1. 35-mm2 tissue culture dishes. 2. Sf 9 or High Five insect cells. 3. SF900 II insect cell culture medium (Subheading 2.1). 4. Recombinant virus stock (Subheading 3.2). 5. Phosphate-buffered saline (PBS; Invitrogen): pH 6.2, sterilize by autoclaving.

2.4. Production of Isotopically Labeled Protein

1. Sf 9 insect cells. 2. SF900 II insect cell culture medium (Subheading 2.1). 3. BioExpress-2000-U (CIL): Unlabeled insect cell culture medium. 4. BioExpress-2000-CN (CIL): 15N-, 13C-labeled insect cell culture medium. 5. Recombinant virus stock (Subheading 3.2). 6. Protease inhibitor (Complete™, Roche). 7. PBS (Subheading 2.3).

3. Methods For a general overview of the BvEs and cloning, expression, analysis, and purification of recombinant proteins in insect cells, please refer to ref. 29. The following procedures describe the cloning of a recombinant baculovirus expression vector, production (Subheading 3.1) and amplification of recombinant baculovirus (Subheading 3.2), analysis of recombinant protein expression (Subheading 3.3), and finally production of isotope-labeled protein (Subheading 3.4) expressed in insect cells. All of the procedures are based on the flashBAC or BacMagic systems (no plaque purification required) and must be carried out using sterile technique. Since the flashBAC system is compatible with all transfer vectors designed for homologous recombination in insect cells at the polh l (BacPAK technology), the target gene can be cloned into many suitable transfer vectors (see Note 2). Moreover, this BvE is compatible with traditional (T4 DNA ligase) and elegant recombinatorial cloning techniques, such as Creator or In-Fusion (BD Biosciences) (see Note 3). While Gateway (Invitrogen) is one of the most popular recombinatorial cloning systems, it is limited to

46

K. Saxena et al.

specifically engineered expression vectors with specifically engineered expression vectors with λ recombination sites. These lead to incorporation of additional amino acids into the protein of interest. A new development to avoid multiple cloning of target genes into host-specific expression vectors is triple host transfer vector, pTriEx, from Novagen. Due to the parallel existence of three promoters in this vector series, recombinant protein expression is enabled in vertebrates, insect cells, and bacteria. 3.1. Cotransfection of Insect Cells

For efficient transfection, high-quality DNA of the transfer plasmid is prepared using commercially available plasmid DNA purification kits (Qiagen, Novagen) (see Note 4). Additionally, it is recommended to use fresh, rapidly proliferating cells (see Note 5) for transfection experiments and to have positive and negative transfection controls (see Note 6). 1. For each cotransfection, prepare one 35-mm2 plate. Seed the dishes with insect cells at least 1 h before use. Use 1 × 106 cells/ dish for Sf9 cells and 1.5 × 106 cells/dish for Sf21 cells in 2 mL of SF900 II insect cell culture medium. Allow the cells to attach by incubating at 28°C for 20 min. 2. During the 1-h incubation period, prepare a DNA–liposome complex cotransfection mix of BacMagic DNA and Insect GeneJuice® transfection reagent for each transfection. Assemble the following components, in the order listed, in a sterile tube (bijoux) (see Note 7): 1 mL serum-free, antibiotic-free SF900 II insect cell culture medium 5 μL Insect GeneJuice 5 μL BacMagic DNA (100 ng total) 5 μL Transfer vector DNA (500 ng total) 1.015 mL Total volume 3. Incubate at room temperature for 15–30 min to allow the DNA–liposome complexes to form. 4. Remove the culture medium from the 35-mm2 dishes of cells using a sterile pipette, ensuring that the cell monolayer is not disrupted (see Note 8). 5. Immediately after the medium has been removed from the cells, add the 1 mL of the DNA–liposome complex dropwise to the center of each dish (see Note 9) and incubate in a plastic sandwich box at 28°C for a minimum of 5 h or overnight. 6. After the incubation period, add another 1 mL of SF900 II insect cell culture medium to each dish and continue the incubation for 5 days in total.

3

Isotope Labeling in Insect Cells

47

7. Following the 5-day incubation period (see Note 10), harvest the medium containing the recombinant virus into a sterile bijoux and store in the dark at 4°C. This is the seed stock of recombinant baculovirus (see Note 11). Due to the limited size of the stock, the next step is to amplify the virus recombination sites. 3.2. Amplification of Recombinant Virus

Amplification of the recombinant virus (produced in Subheading 3.1) is necessary before proceeding with recombinant protein expression. The following provides a protocol (adapted from the Novagen and flashBAC Manual) for amplification and preparation of high-titer recombinant virus (passage 1 stock) in cells grown in suspension culture. 1. Observe the health and viability of cells under an inverted phase-contrast microscope (see Note 12). 2. Prepare a 100–200 mL culture of Sf9 cells at an appropriate cell density in serum-free SF900 II insect cell culture medium (e.g., 2 × 106 Sf9 cells/mL in log-phase growth; high aeration is recommended (see Note 13)). Cells should be infected at a multiplicity of infection (MOI) of <1 pfu/cell. 3. Add 0.5 mL of recombinant virus seed stock to the cell culture. Incubate with shaking until the cells are well-infected (usually, 4–5 days) (see Note 14). 4. When the cells appear to be well-infected with virus, harvest the cell culture medium by centrifugation at 1,500 × g for 15 min at 4°C. Remove the supernatant aseptically. Store the supernatant (recombinant virus stock) in the dark at 4°C (see Note 15).

3.3. Analysis of Recombinant Protein Expression

Before proceeding with use of the virus in subsequent protein expression experiments, a plaque assay to determine the accurate titer is strongly recommended by the manufacturer of the flashBAC and BacMagic technology (see Note 16). This allows the calculation of MOI and ensures cross-referencing and reproducibility in the following experiments. However, for analysis of protein expression, it is not absolutely required to know the titer of the recombinant virus (see Note 17). Pilot expression analysis can also be performed by infecting cells with different amounts of P1 virus and monitoring the expression of the recombinant protein. According to the protein expression literature, it is recommended that exponentially growing cells should be infected at a high MOI to ensure that all cells are infected simultaneously and that the culture is synchronous. However, in our lab, various P1 viruses are directly added (0.5–10%) to the insect cells for determining the best ratio for optimal recombinant protein expression without knowing the exact titer of the different viruses. The best time to

48

K. Saxena et al.

harvest the recombinant protein can be examined by taking samples at different time points after infection (see Note 18). It is recommended to also prepare noninfected or mock-infected cells as a control for host cell proteins and a positive control of a successfully tested recombinant virus. Recombinant protein expression can then be evaluated by SDS-PAGE or Western blot analysis. Additionally, protein expression should be compared between Sf9 and High Five cell lines, especially if the protein of interest is secreted (see Note 19). The following protocol is adapted from the Novagen and flashBAC Manual. 1. Seed 35-mm2 dishes (one per virus to be tested plus a positive and negative control) with 1 × 106 of S9 or High Five cells per dish and allow the cells to attach by incubating at 28°C for 1 h. 2. Remove the medium, add 200 μL of the recombinant virus stock to be tested, and incubate at room temperature for an additional hour. 3. Remove the virus inoculum, add 1.5 mL of SF900 II insect cell culture medium, and incubate at 28°C for 48 h. 4. Use a sterile pipette tip to scrape the cells from the dishes into suspension and to transfer them into sterile Eppendorf tubes. 5. Centrifuge the cells at 1,500 × g for 10 min at room temperature. Remove the supernatant and discard (see Note 20). 6. Resuspend the pellet (protein of interest in the cytosol) in 80 μL of PBS. 7. Analyze the extent of protein overexpression by using SDSPAGE and Western blotting if a suitable antibody is available for the recombinant protein (see Note 21 and Fig. 3).

Fig. 3. Protein overexpression and purification using insect cells. SDS PAGE (left ), Western Blot (right ): Protein samples loaded are BC (before Ni-NTA chromatography), FT (Flowthrough), M (Molecular weight marker), W (Wash), E (Elution).

3

3.4. Production of Isotopically Labeled Protein

Isotope Labeling in Insect Cells

49

While fermenters and bioreactors are now routinely used to produce recombinant proteins in insect cells, good yields can be achieved in shake cultures at a fraction of the cost. Large, disposable shake flasks can be obtained that hold up to 1.5 L of insect cell culture on a shaking platform in a warm room or incubator. The most suitable culture protocol for isotope labeling of a recombinant protein in insect cells involves an initial growth phase for the insect cells in an unlabeled growth/expression medium and a subsequent expression phase, where cells are centrifuged and resuspended into the labeling medium prior to infection with the recombinant baculovirus (see Note 22). This centrifugation step allows much higher expression levels of the labeled protein compared to one with growth and expression in labeling medium without the change in medium. Recently, isotope-labeled expression media for insect cells have become commercially available (BioExpress-2000, CIL), but the source and the concentration of the amino acids in these media are not published. However, for site-specific labeling of amino acids in proteins expressed in insect cells, adding the isotopelabeled amino acid of interest to a standard insect cell medium before infection (see Note 23) also works well and the cost for this labeling strategy is comparable to that when using E. coli. The following protocol is adapted from the Application Note: Efficient uniform isotope labeling of proteins expressed in Baculovirus-infected insect cells using BioExpress® 2000 (Insect cell) medium by André Strauss, Gabriele Fendrich, and Wolfgang Jahnke. Prior to performing isotope labeling of a protein, optimize the culture and BV-infection conditions in unlabeled medium (e.g., BioExpress 2000-U or SF900 II for expression of the protein). 1. Cultivate several 100 mL cultures of Sf9 cells, adapted to growth in serum-free SF900 II medium, in 500-mL Erlenmeyer flasks for 3 days at 27°C with shaking at 90 rpm. 2. Prepare the uniform isotope-labeling medium (see Note 24). 3. When the final cell density of the culture has reached 1.5 × 106 cells/mL (~3 days), sterile centrifuge the cells at 400 × g for 20 min at 20°C. 4. Resuspend the pelleted cells in 100-mL portions in labeled BioExpress® 2000-CN medium and transfer them to fresh 500-mL Erlenmeyer flasks. 5. Add the recombinant virus stock at a titer of 0.5–2 × 108 pfu/mL to an MOI = 1–2, according to optimized conditions. 6. Grow the 100 mL cultures of baculovirus-infected Sf9 cells for 3 days post infection in labeled BioExpress® 2000-CN medium at 27°C, with shaking at 90 rpm (see Note 25). 7. Harvest the cells expressing the labeled recombinant protein by centrifuging at 400 × g for 20 min at 20°C (see Note 26); resuspend the pelleted cells in 20 mL of PBS containing

50

K. Saxena et al.

protease inhibitor mix (see Note 27), followed by a second centrifugation in 50-mL plastic tubes at 400 × g for 20 min at 20°C. Store the pelleted cells at −80°C. 8. Isolate and purify the recombinant protein according to protocols generated for the unlabeled protein.

4. Notes 1. Lipofectin® (Invitrogen), FuGENE 6 (Roche), Tfx-20™ (Promega), and CELLFECTIN® (Invitrogen) have also been successfully tested. 2. Several suitable vectors from different providers are available and are summarized: http://www.expressiontechnologies. com/flashBAC/vectors.asp. 3. Ligation protocols are described in the manuals of the specific cloning technology providers. 4. Plasmid DNA purification protocols are provided by the manufacturers of the purification kits. The DNA must be sterile and must be of a quality suitable for transfection into cells. 5. Healthy cells look bright, round, and refractile, and many should be in the process of dividing into daughter cells. 6. A control transfer vector is supplied with the BacMagic kit and can be used to make recombinant virus; the lacZ-positive infected cells can be stained using X-gal. Add 1 mL of appropriate insect cell culture medium containing 15 μL of 2% (w/v) X-gal in N,N-dimethylformamide and incubate at 28°C. After 5 h, the cells and culture medium appear blue in color, confirming the production of recombinant virus. 7. Plasticware used to prepare the transfection mixture has to be made from polystyrene and not from polypropylene, since the complexes bind to polypropylene. 8. When removing liquid from a dish of cells, tip the dish at a 30–60° angle so that the liquid pools toward one side of the dish. It is important not to allow the cell monolayer to dry out at this point. 9. Adding the mixture dropwise should not disturb the cell monolayer if it is done slowly and gently. 10. Cell monolayers in which recombinant virus has been produced appear very different from mock-transfected control cells under the inverted microscope. Control cells form a confluent monolayer while virus-infected cells do not form a confluent monolayer and appear grainy with enlarged nuclei.

3

Isotope Labeling in Insect Cells

51

11. The expected titer of this initial viral seed stock is generally about 1 × 107 pfu/mL. In contrast to virus-free cells, virusinfected cells appear grainy with enlarged nuclei and do not form a confluent monolayer. 12. It is important that the cells are healthy and in the log phase of growth to ensure that virus replication occurs efficiently to generate high-titer stocks of virus for subsequent use in expression studies suitable for transfection into cells. 13. Virus-infected cells have an increased need for oxygen, and therefore the contents of the flasks should be shaken at quite high speeds to maximize aeration. The surface area-to-volume ratio should also be as large as possible for maximum gas exchange. Do not overfill the flasks. 14. Under a phase-contrast inverted microscope, cells infected with virus appear grainy when compared to healthy cells. The infected cells become uniformly rounded and enlarged, with distinct enlarged nuclei. 15. The virus stock can be stored in the dark at 4°C for 6–12 months, although the titer begins to drop after 3–4 months. Titer the virus before use and reamplify if necessary. The addition of 2–5% serum when using serum-free medium can be helpful in avoiding a drop in titer. Virus may be frozen at −80°C for longer periods of time. Avoid multiple freeze–thaw cycles. 16. Protocols for plaque assays can be found in the manuals of Novagen and Oxford Expression Technologies. 17. After cotransfecting and harvesting the seed stock of virus for further amplification, it is possible to harvest the remaining cells from the dish and prepare these for SDS-PAGE/Western blotting. This gives a quick check for gene expression. 18. The most commonly used times are at 24, 48, 72, and 96 h post infection (hpi). Some proteins may be very stable and accumulate to high levels by 96 hpi, and others may start to degrade and thus need to be harvested much earlier. 19. High Five cell lines (Invitrogen) often increase the yield of secreted recombinant proteins. 20. For secreted protein expression, the supernatant has to be analyzed. 21. Standard molecular biology procedures like SDS-PAGE and Western-blotting are precisely described in Molecular Cloning: A Laboratory Manual) (49). 22. A similar cost-reducing labeling strategy in E. coli was also described for the production of isotope-labeled protein by Marley and coworkers (50).

52

K. Saxena et al.

23. 300–600 mg of 15N-Phe or 15N-Leu (per liter culture medium) was immediately added to the insect cells after being infected with virus. The medium used for the amino acid-selective labeling was 1LSF-900 II serum-free medium from Invitrogen. 24. Medium can be stored, and filter-sterilized for several months at 4°C in the dark without loss of capacity for protein expression. Warm to 28°C before using. 25. The harvesting time of the infected cells depends on the stability of the target protein. 26. For secreted protein expression, the supernatant has to be further purified. 27. Protease inhibitor mix: Roche complete (one tablet dissolved in 200 mL of PBS) References 1. Goto, N. K., and Kay, L. E. (2000) New developments in isotope labeling strategies for protein solution NMR spectroscopy. Curr. Opin. Struct. Biol. 10, 585–592. 2. Schein, C. H. (1991) Optimizing protein folding to the native state in bacteria. Curr Opin. Biotechnol. 2, 746–750. 3. Qing, G., Ma, L. C., Khorchid, A., Swapna, G. V., Mal, T. K., Takayama, M. M., Xia, B., Phadtare, S., Ke, H., Acton, T., Montelione, G. T., Ikura, M., and Inouye, M. (2004) Coldshock induced high-yield protein production in Escherichia coli. Nat. Biotechnol. 22, 877–882. 4. Young, J. C., Agashe, V. R., Siegers, K., and Hartl, F. U. (2004) Pathways of chaperonemediated protein folding in the cytosol. Nat. Rev. Mol. Cell. Biol. 5, 781–791. 5. Esposito, D., and Chatterjee, D. K. (2006) Enhancement of soluble protein expression through the use of fusion tags. Curr. Opin. Biotechnol. 17, 353–358. 6. Andersen, C. L., Matthey-Dupraz, A., Missiakas, D., and Raina, S. (1997) A new Escherichia coli gene, dsbG, encodes a periplasmic protein involved in disulphide bond formation, required for recycling DsbA/DsbB and DsbC redox proteins. Mol. Microbiol. 26, 121–132. 7. Bardwell, J. C. (1994) Building bridges: disulphide bond formation in the cell. Mol. Microbiol. 14, 199–205. 8. Pickford, A. R., and O’Leary, J. M. (2004) Isotopic labeling of recombinant proteins from the methylotrophic yeast Pichia pastoris. Methods Mol. Biol. 278, 17–33.

9. Hu, Y. C. (2008) Baculoviral vectors for gene delivery: a review. Curr. Gene Ther. 8, 54–65. 10. Kost, T. A., Condreay, J. P., Ames, R. S., Rees, S., and Romanos, M. A. (2007) Implementation of BacMam virus gene delivery technology in a drug discovery setting. Drug Discov. Today 12, 396–403. 11. Hitchman, R. B., Possee, R. D., and King, L. A. (2009) Baculovirus expression systems for recombinant protein production in insect cells. Recent Pat. Biotechnol. 3, 46–54. 12. Lam, J. S., Huang, H., and Levitz, S. M. (2007) Effect of differential N-linked and O-linked mannosylation on recognition of fungal antigens by dendritic cells. PLoS One 2, e1009. 13. Dasgupta, S., Navarrete, A. M., Bayry, J., Delignat, S., Wootla, B., Andre, S., Christophe, O., Nascimbeni, M., Jacquemin, M., MartinezPomares, L., Geijtenbeek, T. B., Moris, A., Saint-Remy, J. M., Kazatchkine, M. D., Kaveri, S. V., and Lacroix-Desmazes, S. (2007) A role for exposed mannosylations in presentation of human therapeutic self-proteins to CD4+ T lymphocytes. Proc. Natl. Acad. Sci. U.S.A 104, 8965–8970. 14. Moore, K. L. (2003) The biology and enzymology of protein tyrosine O-sulfation. J. Biol. Chem. 278, 24243–24246. 15. Daly, R., and Hearn, M. T. (2005) Expression of heterologous proteins in Pichia pastoris: a useful experimental tool in protein engineering and production. J. Mol. Recognit. 18, 119–138. 16. Hamilton, S. R., and Gerngross, T. U. (2007) Glycosylation engineering in yeast: the advent

3

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

of fully humanized yeast. Curr. Opin. Biotechnol. 18, 387–392. Chisholm, V., Chen, C. Y., Simpson, N. J., and Hitzeman, R. A. (1990) Molecular and genetic approach to enhancing protein secretion. Methods Enzymol. 185, 471–482. Dorner, A. J., and Kaufman, R. J. (1990) Analysis of synthesis, processing, and secretion of proteins expressed in mammalian cells. Methods Enzymol. 185, 577–596. Moir, D. T. (1989) Yeast mutants with increased secretion efficiency. Biotechnology 13, 215–231. Biemans, R., Thines, D., Rutgers, T., De Wilde, M., and Cabezon, T. (1991) The large surface protein of hepatitis B virus is retained in the yeast endoplasmic reticulum and provokes its unique enlargement. DNA Cell. Biol. 10, 191–200. Gennaro, D. E., Hoffstein, S. T., Marks, G., Ramos, L., Oka, M. S., Reff, M. E., Hart, T. K., and Bugelski, P. J. (1991) Quantitative immunocytochemical staining for recombinant tissue-type plasminogen activator in transfected Chinese hamster ovary cells. Proc. Soc. Exp. Biol. Med. 198, 591–598. Shuster, J. R. (1991) Gene expression in yeast: protein secretion. Curr. Opin. Biotechnol. 2, 685–690. Mollaaghababa, R., Davidson, F. F., Kaiser, C., and Khorana, H. G. (1996) Structure and function in rhodopsin: expression of functional mammalian opsin in Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. U.S.A 93, 11482–11486. Robinson, A. S., Hines, V., and Wittrup, K. D. (1994) Protein disulfide isomerase overexpression increases secretion of foreign proteins in Saccharomyces cerevisiae. Biotechnology (N Y) 12, 381–384. Shusta, E. V., Raines, R. T., Pluckthun, A., and Wittrup, K. D. (1998) Increasing the secretory capacity of Saccharomyces cerevisiae for production of single-chain antibody fragments. Nat. Biotechnol. 16, 773–777. Butz, J. A., Niebauer, R. T., and Robinson, A. S. (2003) Co-expression of molecular chaperones does not improve the heterologous expression of mammalian G-protein coupled receptor expression in yeast. Biotechnol. Bioeng. 84, 292–304. Durocher, Y., and Butler, M. (2009) Expression systems for therapeutic glycoprotein production. Curr. Opin. Biotechnol. 20, 700–707. Jarvis, D. L. (2009) Baculovirus-insect cell expression systems. Methods Enzymol. 463, 191–222.

Isotope Labeling in Insect Cells

53

29. O´Reilly, D. R., Miller, L., Luckow, V.A. (1992) Baculovirus Expression Vectors - a Laboratory Manual. WH Freeman, New York. 30. van Regenmortel, M. H., Mayo, M. A., Fauquet, C. M., and Maniloff, J. (2000) Virus nomenclature: consensus versus chaos. Arch. Virol. 145, 2227–2232. 31. Hunt, I. (2005) From gene to protein: a review of new and enabling technologies for multiparallel protein expression. Protein Expr. Purif. 40, 1–22. 32. Carbonell, L. F., Klowden, M. J., and Miller, L. K. (1985) Baculovirus-mediated expression of bacterial genes in dipteran and mammalian cells. J. Virol. 56, 153–160. 33. Smith, G. E., Summers, M. D., and Fraser, M. J. (1983) Production of human beta interferon in insect cells infected with a baculovirus expression vector. Molecular Cell. Biol. 3, 2156–2165. 34. Ayres, M. D., Howard, S. C., Kuzio, J., LopezFerber, M., and Possee, R. D. (1994) The complete DNA sequence of Autographa californica nuclear polyhedrosis virus. Virology 202, 586–605. 35. Kitts, P. A., and Possee, R. D. (1993) A method for producing recombinant baculovirus expression vectors at high frequency. Biotechniques 14, 810–817. 36. Kost, T. A., Condreay, J. P., and Jarvis, D. L. (2005) Baculovirus as versatile vectors for protein expression in insect and mammalian cells. Nat. Biotechnol. 23, 567–575. 37. Friesen, P. D., and Nissen, M. S. (1990) Gene organization and transcription of TED, a lepidopteran retrotransposon integrated within the baculovirus genome. Mol. Cell. Biol. 10, 3067–3077. 38. Harrison, R. L., and Jarvis, D. L. (2006) Protein N-glycosylation in the baculovirusinsect cell expression system and engineering of insect cells to produce “mammalianized” recombinant glycoproteins. Adv. Virus Res. 68, 159–191. 39. DeLange, F., Klaassen, C. H., Wallace-Williams, S. E., Bovee-Geurts, P. H., Liu, X. M., DeGrip, W. J., and Rothschild, K. J. (1998) Tyrosine structural changes detected during the photoactivation of rhodopsin. J. Biol. Chem. 273, 23735–23739. 40. Creemers, A. F., Klaassen, C. H., Bovee-Geurts, P. H., Kelle, R., Kragl, U., Raap, J., de Grip, W. J., Lugtenburg, J., and de Groot, H. J. (1999) Solid state 15N NMR evidence for a complex Schiff base counterion in the visual G-proteincoupled receptor rhodopsin. Biochemistry 38, 7195–7199.

54

K. Saxena et al.

41. Bellizzi, J. J., Widom, J., Kemp, C. W., and Clardy, J. (1999) Producing selenomethioninelabeled proteins with a baculovirus expression vector system. Structure 7, R263-267. 42. Bruggert, M., Rehm, T., Shanker, S., Georgescu, J., and Holak, T. A. (2003) A novel medium for expression of proteins selectively labeled with 15N-amino acids in Spodoptera frugiperda (Sf9) insect cells. J. Biomol. NMR 25, 335–348. 43. Strauss, A., Bitsch, F., Cutting, B., Fendrich, G., Graff, P., Liebetanz, J., Zurini, M., and Jahnke, W. (2003) Amino-acid-type selective isotope labeling of proteins expressed in Baculovirus-infected insect cells useful for NMR studies. J. Biomol. NMR 26, 367–372. 44. Strauss, A., Bitsch, F., Fendrich, G., Graff, P., Knecht, R., Meyhack, B., and Jahnke, W. (2005) Efficient uniform isotope labeling of Abl kinase expressed in Baculovirus-infected insect cells. J. Biomol. NMR 31, 343–349. 45. Betz, M., Vogtherr, M., Schieborr, U., Elshorst, B., Grimme, S., Pescatore, B., Langer, T., Saxena, K., and Schwalbe, H. (2008) Chemical Biology of Kinases Studied by NMR Spectroscopy, In Chemical Biology (Prof. Dr. Stuart L. Schreiber, P. D. T. M. K. P. D. G., Ed.), pp 852–890.

46. Stockman, B. J., Kothe, M., Kohls, D., Weibley, L., Connolly, B. J., Sheils, A. L., Cao, Q., Cheng, A. C., Yang, L., Kamath, A. V., Ding, Y. H., and Charlton, M. E. (2009) Identification of allosteric PIF-pocket ligands for PDK1 using NMR-based fragment screening and 1H-15N TROSY experiments. Chem. Biol. Drug Des. 73, 179–188. 47. Jahnke, W., Grotzfeld, R. M., Pelle, X., Strauss, A., Fendrich, G., Cowan-Jacob, S. W., Cotesta, S., Fabbro, D., Furet, P., Mestan, J., and Marzinzik, A. L. (2010) Binding or bending: distinction of allosteric Abl kinase agonists from antagonists by an NMR-based conformational assay. J. Am. Chem. Soc. 132, 7043–7048. 48. Vajpai, N., Strauss, A., Fendrich, G., CowanJacob, S. W., Manley, P. W., Jahnke, W., and Grzesiek, S. (2008) Backbone NMR resonance assignment of the Abelson kinase domain in complex with imatinib. Biomol. NMR Assign. 2, 41–42. 49. Maniatis, T. (1982) Molecular cloning : a laboratory manual / T. Maniatis, E.F. Fritsch, J. Sambrook, Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 50. Marley, J., Lu, M., and Bracken, C. (2001) A method for efficient isotopic labeling of recombinant proteins. J. Biomol. NMR 20, 71–75.

Chapter 4 Isotope Labeling in Mammalian Cells Arpana Dutta, Krishna Saxena, Harald Schwalbe, and Judith Klein-Seetharaman Abstract Isotope labeling of proteins represents an important and often required tool for the application of nuclear magnetic resonance (NMR) spectroscopy to investigate the structure and dynamics of proteins. Mammalian expression systems have conventionally been considered to be too weak and inefficient for protein expression. However, recent advances have significantly improved the expression levels of these systems. Here, we provide an overview of some of the recent developments in expression strategies for mammalian expression systems in view of NMR investigations. Key words: Isotope labeling, Nuclear magnetic resonance, Recombinant protein expression, Human embryonic kidney cells

1. Introduction: Mammalian Cell Expression System 1.1. Principle of Mammalian Cell-Mediated Protein Expression

Mammalian cell expression systems are being increasingly used to express proteins for structural biology. This is evident from the increase in the number of structures available in the Protein Data Bank (PDB) (1) of proteins purified from such sources. The main reason for switching to these higher eukaryotic expression systems is the ability to produce biologically active cell surface receptors and secreted glycoproteins. The functionality of these proteins is linked to the requirement for posttranslational modifications, such as glycosylation and disulfide bond formation, which can often only be satisfied by using mammalian systems and not other eukaryotic systems, such as yeast and insect cells. Protein expression in mammalian cells involves transfection with a plasmid carrying the gene of interest under the control of a mammalian promoter. The mammalian gene can be introduced

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_4, © Springer Science+Business Media, LLC 2012

55

56

A. Dutta et al.

into cells either by way of transient transfection or by developing a stable cell line. The former can be achieved with both adherent and suspension cells using polyethyleneimine (PEI). The only advantage of this method over stable cell line preparation is that milligram quantities of protein can be prepared in a few days time compared to the months required by the latter method (2–4). However, there are several disadvantages associated with the transient transfection approach (1) Not every protein, in particular membrane proteins, give high yields when transiently transfected; (2) Very large quantities of DNA need to be prepared; (3) The process needs to be repeated every time protein is needed, adding labor and cost. Stable cell line creation overcomes these limitations, although 2–6 months are required to establish a high-level expression system. The effort is often justified, however, by the advantages of a stable mammalian cell line over a transient transfect (1) The yield is typically very high; (2) The process becomes fast and robust once the cell line is created for a particular protein; (3) Stable cell lines are usually created using calcium phosphate transfection, an efficient and inexpensive method; (4) Cell lines can be established in cellular backgrounds that are specifically tailored to the needs of the protein being expressed. For example, heterogenous glycosolyation is a problem in NMR studies which typically require highly homogenous material. To overcome this problem, cell lines can be used that are deficient in complex heterogenous glycosylation capabilities (5); (5) The ability to create inducible cell lines is an additional advantage. In cases where constitutive expression of a particular protein is toxic to cells, placing the gene under an inducible promoter can allow expression of otherwise toxic proteins; (6) Finally, stable cells can be easily scaled up by transferring them to suspension cultures in spinner flasks or bioreactors. Episomal vectors developed from viruses, such as Epstein–Barr virus (EBV) (6), bovine papilloma virus (BPV) (7), BK virus (BKV) (8), and Simian virus 40 (SV40) (9) are mainly used for transfection. A list of commonly used vectors can be found in (10). The advantages of using episomal vectors over integrating genes into the DNA of the host cell are numerous (1) Episomal expression will result in gene expression independent of the regulatory mechanisms of the host cell and the position of integration in the host genome, leading to higher expression levels since these factors may unfavorably influence expression. Random integration may also disrupt the characteristics of the gene of interest (11); (2) There is no interruption in host cell gene expression, which is often the case with integrative vectors, the latter leading to transformation of the host cell and undesirable effects on protein production; (3) Episomal vectors can exist in multiple copies in a cell, thus leading to amplification of the gene of interest. The main criteria for selecting a suitable episomal vector are (1) high copy number in E. coli for large-scale DNA production; (2) a strong mammalian promoter

4

Isotope Labeling in Mammalian Cells

57

for high expression levels of the gene of interest in mammalian cells; and (3) small size so that genes of different lengths can be easily cloned into the vector. Thus, the vector typically consists of a strong mammalian promoter to drive expression of the mammalian gene of interest, a viral origin of replication activated by viral early genes, which is required to carry out its replication thereby maintaining the vector in the host system, and a eukaryotic selection marker, usually an antibiotic resistance gene, such as neomycin, to select for transfected cells. The human cytomegalovirus promoter has been shown to be very powerful in both HEK293 and Chinese Hamster Ovary (CHO) cells and is also active in most mammalian cells (12). 1.2. Mammalian Cell Lines

Mammalian cell lines used for transfection with virus-based vectors carrying the gene of interest usually already carry vectors encoding viral early genes corresponding to the origin of replication used by the new vector. Some of these early genes are SV40 T antigen, EBNA-1, E1, and E2 which bind to SV40, EBV, and BPV ori, respectively. The early genes act in trans to initiate replication by binding to the ori and they also act as enhancers, thereby increasing the copy number of viral vector and the expression levels of the protein of interest (13). Since very high copy numbers of the viral genes can lead to host cell death, regulation of viral replication can be imposed by transfecting mammalian cells with a different vector carrying the early genes (13). Mammalian cell lines that are widely used for protein expression are CHO and Human Embryonic Kidney 293 (HEK293) cells. Both are suitable for use in adherent and suspension cultures. The advantage of using CHO cells is the availability of various auxotrophs that can be used as selection markers for transfection. An example of such an auxotroph is dihydrofolate reductase (DHFR) deficient cells which are triple auxotrophs for hypoxanthine, glycine, and thymidine (14). Transfection of foreign genes along with DHFR genes in these cells allows for the selection of clones in a medium devoid of the above nutrients. Another advantage of this system is that it helps amplify the foreign gene when DHFR deficient cells are grown in the presence of methotrexate, which blocks DHFR activity. This causes the transfected cells to deal with low DHFR activity by amplifying the copy number of DHFR, thus amplifying the copy number of the transfected gene of interest. The disadvantages of the CHO cell line is that these cells do not carry all of the sugar transferring enzymes (15) that are present in human cell lines. This may lead to the production of functionally nonrelevant proteins. Further, some of the posttranslational modifications of human proteins are not appropriately carried out in these cells (16). It has also been seen that expression levels are usually lower in CHO cells than in HEK293 cells (12, 17). Therefore, the human cell line, HEK293, is used for proteins that require functionally sensitive posttranslational modifications.

58

A. Dutta et al.

There are two mutant cell lines of HEK293, HEK293E, expressing EBNA-1, and HEK293T, expressing SV40 large T antigen, that increase the copy number of plasmids with EBV and SV40 origins of replication, respectively, thereby increasing expression levels of the protein of interest (13). 1.3. Expression and Labeling of Recombinant Proteins in Mammalian Cells

Uniform isotope labeling of proteins expressed in mammalian cells is still under development (18). This is because mammalian cells, like insect cells, are unable to grow in the type of minimal media that are used for bacteria where glucose or glycerol and ammonium chloride are the only sources of carbon and nitrogen, respectively. Higher eukaryotes, like the organisms they represent, require certain essential amino acids, without which they are unable to grow. Thus, the use of simple sources for 13C and 15N isotope labeling of proteins in bacteria is not possible for mammalian cells. A number of recent approaches, however, have been developed to address this complication. The first uniformly isotope-labeled protein purified from mammalian cells was obtained by Hansen et al. (19) followed by (20, 21). The method involved purification of a mixture of isotope labeled amino acids from an acid hydrolysate of algae or bacteria grown in 15NH4Cl and 13C glucose or 13CO2 to prepare uniformly labeled urokinase in a Sp2/0 mouse myeloma cell line. Different purification protocols of the labeled mixture were tested but only acid hydrolysis, which removes bacterial and algal products that are toxic to mammalian cells, proved to work. The mixture, however, needed to be supplemented with commercially available amino acids that would degrade during hydrolysis, particularly glutamine and cysteine. A commercially available form of 15 N-cysteine was used. Because of the high cost involved in purchasing 15 N-glutamate, 15N-, and 15N/13C-glutamine were synthesized from labeled glutamate (available commercially), 15NH4Cl, and ATP. Dialyzed serum was used to prevent dilution of isotope labeled amino acids with unlabeled amino acids from the serum. Such purified amino acid mixtures are commercially available today. One such commercially available mixture from Martek Biosciences Corp. was tested to support expression in CHO cells (20). The growth medium was optimized in terms of its concentration, removal of amino acids, such as aspartic acid and asparagine, and addition of amino acids, such as arginine, cysteine, and glutamic acid after isolating and purifying from the mixture so that the highest level of protein expression was obtained. However, there are serious concerns regarding the toxic effects of feeding these mixtures to mammalian cells, leading to extensive methods of purification and optimization, and making it a time and cost inefficient method. HEK293 cells, the most commonly used system of expression for preparing isotope-labeled samples, have, to date, not been grown successfully in the presence of these mixtures (Klein-Seetharaman, unpublished results). An uniformly labeled 15N TGFb1 sample was prepared from CHO cells by

4

Isotope Labeling in Mammalian Cells

59

growing them in Minimum Essential Medium (MEM, Gibco) with dialyzed serum, uniformly 15N labeled choline and uniformly 15 N-labeled amino acids (except tryptophan which was labeled only in the backbone nitrogen) (22), demonstrating the utility of CHO systems. The total cost involved was $1,000/L. Therefore, this method could prove to be expensive if the protein to be labeled does not have high expression levels, as is the case with most membrane proteins. Mammalian growth media containing a mixture of certain isotope-labeled amino acids are commercially available and have been recently used for labeling the G-protein coupled receptor, bovine rhodopsin, using 15N labeled medium in which GKLQSTVW amino acids were 15N labeled (23). If all the amino acids in the mixture do not cover the complete sequence of the protein to be labeled, then this method may not lead to complete labeling, as was the case with rhodopsin where 50% of amino acids were labeled (23). Apart from the high costs involved, an additional problem is the inability to perform perdeuteration because of the sensitivity of mammalian cells to deuterium oxide. Therefore, the most widely used method to obtain structural information on mammalian proteins by NMR has been by conducting amino acid type selective (AATS) isotope labeling. AATS labels a protein with specific isotope-labeled amino acids rather than uniform labels and yields structural information selectively for the amino acid(s) used in labeling. This has been successfully done with numerous proteins (24–26). Extensive AATS labeling has been carried out with rhodopsin, which was labeled by using 15N-isotopes of tryptophan (26), lysine (25), histidine (27) and 13C-isotopes of tryptophan (27), histidine (27) and glutamate (28), all through expression in HEK293 cells. 15N-lysine, 13C-glycine/serine double labeling has also been shown for rhodopsin (25). For example, multiple combinations of 15N- and 13C-isotope-labeled rhodopsin samples were used to assign 15N-tryptophan resonances in the NMR spectrum of rhodopsin, shown in Fig. 1 (29). Since the protocol for such a labeling method has been optimized and works very well at least for rhodopsin, we are providing a step-by-step protocol below. The first step is the creation of an inducible stable cell line of rhodopsin in HEK293 cells. Scaling up the expression level is accomplished by growing these cells in a suspension culture, thereby producing the milligram quantities required for NMR studies. The yield of rhodopsin from such a transient transfection is on the order of 50 mg/15 cm2 plate. After establishing the stable cell line with the opsin (apo form of rhodopsin) gene, the next step is to transfer these cells to suspension culture and grow them in isotope-labeled medium containing unlabeled amino acids along with the labeled amino acid(s) of interest. Up to 10 mg of rhodopsin can be obtained from such suspension cultures when amino acid supplements are provided (30), but typically 2 mg of rhodopsin are obtained from

60

A. Dutta et al.

Fig. 1. Dephased (red line) and nondephased (blue line) 15N detected 13C/15N CP REDOR spectra of selectively labeled rhodopsin in DOPC lipid bilayers at 220 K. (a) a,e-15N-Trp; (b) 13C¢-Leu/a,e-15N-Trp; (c) 13U-Thr/a,e-15N-Trp; (d) 13C¢-Cys/a,e15 N-Trp; (e) 13C¢-Pro/a,e-15N-Trp; (f) 13C¢-Gly/a,e-15N-Trp. Figure reproduced with permission from (29).

4

Isotope Labeling in Mammalian Cells

61

isotope labeled medium in which amino acid mixtures cannot be supplemented, to avoid dilution of the isotope-labeled amino acids provided in the medium.

2. Materials 2.1. Expression of Isotope-Labeled Recombinant Proteins

1. Plasmid containing the gene of interest (see Note 1). 2. HEK293 cells. 3. Dulbecco’s modified Eagle medium (DMEM F-12). 4. Blasticidin (500 mg/mL stock solution prepared in DMEM F-12). 5. Geneticin, G418 (100 mg/mL stock solution prepared in DMEM F-12). 6. 0.05% Trypsin–EDTA (Gibco). 7. Penicillin–streptomycin (PS): 100 U/mL of each. 8. Fetal bovine serum (FBS). 9. Phosphate buffer saline (PBS): Autoclaved. 10. Tetracycline: 200 mg/mL (100× stock). 11. Sodium butyrate: 500 mM (100× stock), filtered. 12. Complete medium: DMEM F-12, 10% FBS, 1% PS. 13. Induction medium: Complete medium containing 2 mg/mL tetracycline and 5 mM sodium butyrate. 14. Selection medium: Complete medium containing 5 mg/mL blasticidin and 1, 2, or 3 mg/mL of G418. 15. Cryo medium: Complete medium containing 10% DMSO. 16. 2.5 M CaCl2. 17. BES: 50 mM N,N-bis(2-hydroxyethyl)-2-aminoethanesulfonic acid, pH 7.2, 250 mM NaCl, 1.5 mM Na2HPO4, 1 M NaOH used to adjust pH. 18. 15- and 10-cm2 Tissue culture plates. 19. 6- and 24-Well cell culture plates. 20. Sterile forceps. 21. Cloning rings. 22. Vacuum grease. 23. Dodecyl maltoside.

2.2. Specific Isotope Labeling of Proteins

Suspension DMEM medium: 1. Table 1 gives the individual components of suspension DMEM in their final concentrations based on the composition of commercially available DMEM. 100× stock solutions are prepared

mg/L

84

42

105

105

146

30

66

95

16

94

Essential amino acid

Arginine HCl

Histidine HCl H2O

Isoleucine

Leucine

Lysine HCl

Methionine

Phenylalanine

Threonine

Tryptophan

Valine

42 104

Tyrosine 2Na+ 2H2O

None

30

584

Serine

Proline

Glycine

Glutamine

None

63

Cystine 2HCl Glutamate

None

None

None

mg/L

Aspartate

Asparagine

Alanine

Nonessential amino acid

Table 1 Amino Acid Composition of suspension DMEM

pantothenate

0.4 4

Thiamine HCl

4

Pyridoxal HCl Riboflavin

4

7.2

4

4

4

mg/L

Niacinamide

i-Inositol

Folic Acid

Choline Chloride

D-Ca

Vitamins

Glucose

Phenol Red Na+

NaH2PO4 H2O

NaCl

KCl

MgSO4

Fe(NO3)3 9H2O

CaCl2

Inorganic salts and other

4,500

15

125

6,400

400

97.7

0.1

50

mg/L

62 A. Dutta et al.

4

Isotope Labeling in Mammalian Cells

63

for most components (see Note 2). Each component is dissolved in distilled water. 2. 0.05% trypsin–EDTA (Subheading 2.1). 3. FBS (Subheading 2.1). 4. PS (Subheading 2.1). 5. Dialyzed FBS: Heat inactivated, dialyzed against PBS and then filtered. 6. 10% Pluronic F-68 (100× stock), filtered. 7. 5 mg/mL Heparin (100× stock), filtered. 8. 15-cm2 tissue culture plates. 9. 2-L spinner flasks. 10. 100× tetracycline (Subheading 2.1). 11. 100× sodium butyrate (Subheading 2.1). 12. 20% (w/v) glucose, sterile filtered. 13. 8% (w/v) NaHCO3. 14. Complete medium (Subheading 2.1). 15. Selection medium (Subheading 2.1).

3. Methods 3.1. Expression of Isotope-Labeled Recombinant Proteins

Following is a protocol for generating an inducible stable cell line of HEK293 (30–32). 1. Wake up HEK293 cells from cryostocks and maintain them in 15 cm2 plates in 25 mL of complete medium containing 5 mg/mL blasticidin. 2. Split the cells into a 10-cm2 dish at 80% confluency. 3. Split the cells 1:10 or 1:8 (~1–2 million cells per plate) into complete medium containing 5 mg/mL blasticidin the day before transfection. 4. On the day of transfection, the cells should be 30–40% confluent. To transfect the cells, prepare the following cocktail, adding components in the order given (volumes are for one 10-cm2 plate): In a falcon tube, mix 30 mg of plasmid DNA, 50 mL of 2.5 M CaCl2, 500 mL of BES, and sterile water to 1 mL. Incubate the mixture for exactly 1 min and gently add it to the cells. After 1 h, verify the efficiency of transfection by the presence of calcium phosphate precipitate which appears as fine sand particles between cells. 5. Incubate the plates at 35°C and 3% CO2 for 19 h.

64

A. Dutta et al.

6. After 19 h, wash the plates twice with complete medium containing 5 mg/mL blasticidin and add fresh complete medium containing 5 mg/mL blasticidin. 7. Incubate the cells at 37°C and 5% CO2 overnight. 8. Split the cells 1:10 (0.5–1 million cells/plate) into nine plates, three plates each for selection with G418 (1, 2, and 3 mg/mL). 9. Replace the medium after 20 h with selection medium. 10. Replace the selection medium every 2–3 days until colonies of workable size are formed. 11. On the day of picking the clones: (a) Circle the colonies that are to be picked such that the circles are big enough to place the cloning rings on its perimeter. (b) Aspirate the medium and look for more colonies on the empty plate by eye. Wash with 8 mL of PBS. (c) Using sterile forceps, dip one end (base) of each cloning ring in vacuum grease and place it over each clone (see Note 3). Make sure there is no grease on the inner walls of the rings. Make sure that the rings form a well around the clones with a leak proof seal at the bottom. (d) Add 40 mL of 0.05% trypsin–EDTA to each well and incubate for 1 min. Then, add 80 mL of fresh selection medium and gently pipette up and down. Transfer the cells (100 mL) to a 24-well plate containing 1 mL of selection medium per well. (e) Repeat for all the clones. 12. Replace the selection medium every 2–3 days until the cells are confluent. Make a note of medium changes and transfers to 6-well plates in a chart (see step 13 for details). 13. As the cells become confluent, transfer them to two 6-well plates at different cell densities. For the transfer: (a) Pipette 3 mL of selection medium into each well of a 6-well plate. (b) Aspirate the medium from the cells (24-well plate) and add 300 mL of 0.05% trypsin–EDTA to each well and incubate for 1 min. (c) Add 1,200 mL (use a 1-mL pipette and add 600 + 600 mL) of selection medium and gently pipette up down. (d) Remove 300 mL and add it to one well and place the remaining (1,200 mL) in another well. The well with higher cell density will be induced in the future to screen the clones while the one with lower density will be maintained and used for making glycerol stocks.

4

Isotope Labeling in Mammalian Cells

65

14. Change the selection medium on the cells every 2–3 days. 15. As soon as the cells in the higher density well reach confluence, induce with induction medium. 16. Harvest the cells 48-h post induction. (a) After harvesting, solubilize the cells in a detergent suitable for the membrane protein of interest (e.g., 1% dodecyl maltoside). (b) Centrifuge the solubilized cells at 126,000 × g for 20 min at room temperature and collect the supernatant. 17. Check for levels of expression by performing a Western dot blot on serial dilutions of the supernatant in PBS, containing 1% of the detergent used for solubilization, versus protein samples of known concentrations (see Note 4). 18. When the cells from the low density well reach confluence, split 1:5 into two 10-cm2 cell culture dishes. Add 0.5 mL of 0.05% trypsin–EDTA to a 6-well plate, and then add 1.5 mL of selection medium. Add 1 mL of this cell suspension to each of the two 10-cm2 plates. 19. After the cells reach confluence on the 10-cm2 plate, prepare three 1-mL glycerol stocks from each plate. To prepare glycerol stocks: (a) Wash 70–90% confluent plates twice with PBS and trypsinize with 1 mL of 0.05% trypsin–EDTA for 1 min. (b) Add 10 mL of complete medium (no selection) and collect the cells in a 15-mL falcon tube. (c) Centrifuge the cells at 600 × g for 10 min at 4°C. (d) Aspirate the medium and gently resuspend the cells in 3 mL of cryo medium. (e) Transfer the cells to cryo vials in 1-mL aliquots and place the tubes in cryo boxes. Incubate at −20°C for 1 h, and then −80°C overnight. (f) On the following day, move the tubes to liquid nitrogen storage tanks. 3.2. Specific Isotope Labeling of Proteins

After estimating the yield from above (Subheading 3.1) and identifying the highest yield clone, the following procedure is used to prepare a protein sample, in suspension culture, in which particular amino acids are specifically isotope labeled (30–32). 1. Grow the highest yielding stable cell line clone in complete medium. 2. Split the cells 1:5 into 10-cm2 dishes after ~3 days using 15 mL of selection medium.

66

A. Dutta et al.

3. Split the cells until a sufficient number of plates are obtained for setting up spinner flasks. For reference, 1 mg of protein per liter of suspension culture is usually obtained for rhodopsin and 3–4 confluent 15-cm2 plates are used to inoculate a 500-mL flask (see Notes 5 and 6). 4. Prepare labeled suspension DMEM medium from 100× stocks of the individual components (Table 1, Subheading 2.2), except for the isotope labeled amino acid(s). Supplement with 10% dialyzed FBS and 1% PS or obtain isotope-labeled medium from a commercial source. (a) Add glucose, NaCl, glutamine, isotope-labeled amino acid(s) as solids. (b) Lower the glutamine concentration to half when starting the suspension culture. (c) Add Pluronic F-68 and heparin to 1× concentration. (d) Add isotope-labeled amino acid(s) as solids. 5. Inoculate the suspension cultures and grow at 37°C, spinning at 47 rpm for 6 days. To inoculate, add 2 mL of 0.05% trypsin– EDTA to each plate and extract the cells with 8 mL of unlabeled suspension medium. Collect cells from each plate (10-mL volume) and centrifuge at 1,000 × g for 10 min at 4°C. Aspirate the medium and add 10 mL of labeled suspension DMEM medium to each plate and resuspend the cells. Add the cell suspension to the suspension culture in the spinner flask. 6. After 6 days (see Note 7), supplement the growth medium with 6 mL of 20% (w/v) glucose and 4 mL of 8% (w/v) NaHCO3 and induce expression with 5 mM sodium butyrate and 2 mg/mL of tetracycline. 7. After the 6th day, feed the cells with 6 mL of 20% glucose every 24 h. 8. Grow the cells for 2 more days after the 6th day (total 8 days in suspension). 9. Harvest at the end of 8 days.

4. Conclusions A variety of expression systems have been developed that can be used to express a protein of choice in isotope-labeled form, each system having advantages and disadvantages. These systems include bacterial, yeast, insect, and mammalian cells. The choice of the system largely depends on the type of the protein to be expressed. Other factors are the costs involved, the need for modifications and

4

Isotope Labeling in Mammalian Cells

67

yield of protein required. Bacterial expression systems are most commonly used for soluble protein expression due to ease of use and cost-effectiveness in cloning and expression. However, this system is not the system of choice for proteins that require posttranslational modification for their activity. In such cases, eukaryotic expression systems are required. Yeast being both eukaryotic and a microorganism is a simple system to use, allowing relatively inexpensive complete isotope labeling of proteins for NMR studies. Although it can carry out some posttranslational modifications, these are often not sufficient and deficiency in folding machinery can be a problem for some proteins, especially membrane proteins. For these reasons, higher eukaryotic systems, such as insect cells have been in use where posttranslational modifications are very similar to those found in mammalian cells. The greatest advantage of insect cell expression is the very large amounts of proteins that can be expressed, in most cases more than that in mammalian cells. However, disadvantages of expensive isotope-labeled media, incomplete labeling and intolerability to perdeuteration that hamper mammalian cell expression are also prevalent in insect cells. Hence, efforts are ongoing towards further developing mammalian expression systems since these are the only systems in which the full complement of folding and posttranslational modification machineries are available, optimally supporting the activity of the expressed protein.

5. Notes 1. A tetracycline (tet)-regulated mammalian expression vector should be used. 2. All solutions were prepared as 100× concentrated stock solutions, except glucose, NaCl, glutamine, and the isotopelabeled amino acid(s), which were added as solids. Typically, 500 mL of stock solutions were prepared and filtered to maintain sterility. Sterile stock solutions were kept at 4°C and were stable for several weeks. 3. Cloning rings, vacuum grease, and forceps should be autoclaved. 4. PVDF membranes should not be used. 5. Cells should be counted before adding them to spinner flasks so that appropriate number of cells (in the range between 60 and 90 million cells/500 mL) can be transferred. 6. A 2-L spinner flask is ideal for a 500-mL suspension culture. 7. The color of the medium should turn yellow after 6 days.

68

A. Dutta et al.

References 1. Nettleship, J. E., Assenberg, R., Diprose, J. M., Rahman-Huq, N., and Owens, R. J. (2010) Recent advances in the production of proteins in insect and mammalian cells for structural biology. J. Struct. Biol. 172, 55–65. 2. Aricescu, A. R., Lu, W., and Jones, E. Y. (2006) A time- and cost-efficient system for highlevel protein production in mammalian cells. Acta. Crystallogr. D Biol. Crystallogr. 62, 1243–1250. 3. Lee, J. E., Fusco, M. L., and Ollmann Saphire, E. (2009) An efficient platform for screening expression and crystallization of glycoproteins produced in human cells. Nat. Protoc. 4, 592–604. 4. Nettleship, J. E., Rahman-Huq, N., and Owens, R. J. (2009) The production of glycoproteins by transient expression in mammalian cells. Methods Mol. Biol. 498, 245–263. 5. Reeves, P. J., Callewaert, N., Contreras, R., and Khorana, H. G. (2002) Structure and function in rhodopsin: high-level expression of rhodopsin with restricted and homogeneous N-glycosylation by a tetracycline-inducible N-acetylglucosaminyltransferase I-negative HEK293S stable mammalian cell line. Proc. Natl. Acad. Sci. U.S.A. 99, 13419–13424. 6. Sclimenti, C. R., and Calos, M. P. (1998) EpsteinBarr virus vectors for gene expression and transfer. Curr. Opin. Biotechnol. 9, 476–479. 7. Sambrook, J., Rodgers, L., White, J., and Gething, M. J. (1985) Lines of BPV-transformed murine cells that constitutively express influenza virus hemagglutinin. EMBO J. 4, 91–103. 8. Grossi, M. P., Caputo, A., Rimessi, P., Chiccoli, L., Balboni, P. G., and Barbanti-Brodano, G. (1988) New BK virus episomal vector for complementary DNA expression in human cells. Arch. Virol. 102, 275–283. 9. Piechaczek, C., Fetzer, C., Baiker, A., Bode, J., and Lipps, H. J. (1999) A vector based on the SV40 origin of replication and chromosomal S/MARs replicates episomally in CHO cells. Nucleic Acids Res. 27, 426–428. 10. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual, CSH Laboratory Press, Cold Spring Harbor, NY. 11. Doerfler, W., Schubbert, R., Heller, H., Kammer, C., Hilger-Eversheim, K., Knoblauch, M., and Remus, R. (1997) Integration of foreign DNA and its consequences in mammalian systems. Trends Biotechnol. 15, 297–301. 12. Xia, W., Bringmann, P., McClary, J., Jones, P. P., Manzana, W., Zhu, Y., Wang, S., Liu, Y., Harvey, S., Madlansacay, M. R., McLean, K.,

Rosser, M. P., MacRobbie, J., Olsen, C. L., and Cobb, R. R. (2006) High levels of protein expression using different mammalian CMV promoters in several cell lines. Protein Expr. Purif. 45, 115–124. 13. Van Craenenbroeck, K., Vanhoenacker, P., and Haegeman, G. (2000) Episomal vectors for gene expression in mammalian cells. Eur. J. Biochem. 267, 5665–5678. 14. Urlaub, G., and Chasin, L. A. (1980) Isolation of Chinese hamster cell mutants deficient in dihydrofolate reductase activity. Proc. Natl. Acad. Sci. U.S.A. 77, 4216–4220. 15. Grabenhorst, E., Schlenke, P., Pohl, S., Nimtz, M., and Conradt, H. S. (1999) Genetic engineering of recombinant glycoproteins and the glycosylation pathway in mammalian host cells. Glycoconj. J. 16, 81–97. 16. Suttie, J. W. (1986) Report of Workshop on expression of vitamin K-dependent proteins in bacterial and mammalian cells. Madison, Wisconsin, USA, April 1986, Thromb. Res. 44, 129–134. 17. Schlaeger, E. J., Kitas, E. A., and Dorn, A. (2003) SEAP expression in transiently transfected mammalian cells grown in serum-free suspension culture. Cytotechnology 42, 47–55. 18. Takahashi, H., and Shimada, I. (2010) Production of isotopically labeled heterologous proteins in non-E. coli prokaryotic and eukaryotic cells. J. Biomol. NMR 46, 3–10. 19. Hansen, A. P., Petros, A. M., Mazar, A. P., Pederson, T. M., Rueter, A., and Fesik, S. W. (1992) A practical method for uniform isotopic labeling of recombinant proteins in mammalian cells. Biochemistry 31, 12713–12718. 20. Lustbader, J. W., Birken, S., Pollak, S., Pound, A., Chait, B. T., Mirza, U. A., Ramnarain, S., Canfield, R. E., and Brown, J. M. (1996) Expression of human chorionic gonadotropin uniformly labeled with NMR isotopes in Chinese hamster ovary cells: an advance toward rapid determination of glycoprotein structures. J. Biomol. NMR 7, 295–304. 21. Shindo, K., Masuda, K., Takahashi, H., Arata, Y., and Shimada, I. (2000) Backbone 1H, 13C, and 15N resonance assignments of the anti-dansyl antibody Fv fragment. J. Biomol. NMR 17, 357–358. 22. Archer, S. J., Bax, A., Roberts, A. B., Sporn, M. B., Ogawa, Y., Piez, K. A., Weatherbee, J. A., Tsang, M. L., Lucas, R., Zheng, B. L., and et al. (1993) Transforming growth factor beta 1: NMR signal assignments of the recombinant protein expressed and isotopically enriched using Chinese hamster ovary cells. Biochemistry 32, 1152–1163.

4 23. Werner, K., Richter, C., Klein-Seetharaman, J., and Schwalbe, H. (2008) Isotope labeling of mammalian GPCRs in HEK293 cells and characterization of the C-terminus of bovine rhodopsin by high resolution liquid NMR spectroscopy. J. Biomol. NMR 40, 49–53. 24. Arata, Y., Kato, K., Takahashi, H., and Shimada, I. (1994) Nuclear magnetic resonance study of antibodies: a multinuclear approach. Methods Enzymol. 239, 440–464. 25. Klein-Seetharaman, J., Reeves, P. J., Loewen, M. C., Getmanova, E. V., Chung, J., Schwalbe, H., Wright, P. E., and Khorana, H. G. (2002) Solution NMR spectroscopy of (alpha -15N) lysine-labeled rhodopsin: The single peak observed in both conventional and TROSYtype HSQC spectra is ascribed to Lys-339 in the carboxyl-terminal peptide sequence. Proc. Natl. Acad. Sci. U.S.A. 99, 3452–3457. 26. Klein-Seetharaman, J., Yanamala, N. V., Javeed, F., Reeves, P. J., Getmanova, E. V., Loewen, M. C., Schwalbe, H., and Khorana, H. G. (2004) Differential dynamics in the G protein-coupled receptor rhodopsin revealed by solution NMR. Proc. Natl. Acad. Sci. U.S.A. 101, 3409–3413. 27. Patel, A. B., Crocker, E., Reeves, P. J., Getmanova, E. V., Eilers, M., Khorana, H. G., and Smith, S. O. (2005) Changes in interhelical hydrogen bonding upon rhodopsin activation. J. Mol. Biol. 347, 803–812.

Isotope Labeling in Mammalian Cells

69

28. Han, M., and Smith, S. O. (1995) Highresolution structural studies of the retinal– Glu113 interaction in rhodopsin. Biophys. Chem. 56, 23–29. 29. Werner, K., Lehner, I., Dhiman, H. K., Richter, C., Glaubitz, C., Schwalbe, H., KleinSeetharaman, J., and Khorana, H. G. (2007) Combined solid state and solution NMR studies of alpha,epsilon-15N labeled bovine rhodopsin. J. Biomol. NMR 37, 303–312. 30. Reeves, P. J., Kim, J. M., and Khorana, H. G. (2002) Structure and function in rhodopsin: a tetracycline-inducible system in stable mammalian cell lines for high-level expression of opsin mutants. Proc. Natl. Acad. Sci. U.S.A. 99, 13413–13418. 31. Reeves, P. J., Callewaert, N., Contreras, R., and Khorana, H. G. (2002) Structure and function in rhodopsin: high-level expression of rhodopsin with restricted and homogeneous N-glycosylation by a tetracycline-inducible N-acetylglucosaminyltransferase I-negative HEK293S stable mammalian cell line 452. Proc. Natl. Acad. Sci. U.S.A. 99, 13419–13424. 32. Reeves, P. J., Thurmond, R. L., and Khorana, H. G. (1996) Structure and function in rhodopsin: high level expression of a synthetic bovine opsin gene and its mutants in stable mammalian cell lines. Proc. Natl. Acad. Sci. U.S.A. 93, 11487–11492.

Chapter 5 Cell-Free Protein Production for NMR Studies Mitsuhiro Takeda and Masatsune Kainosho Abstract The cell-free expression system using an Escherichia coli extract is a practical method for producing isotope-labeled proteins. The advantage of the cell-free system over cellular expression is that any isotopelabeled amino acid can be incorporated into the target protein with minimal scrambling, thus providing opportunities for advanced isotope labeling of proteins. We have modified the standard protocol for E. coli cell-free expression to cope with two problems specific to NMR sample preparation. First, endogenous amino acids present in the E. coli S30 extract lead to dilution of the added isotope. To minimize the content of the remaining amino acids, a gel filtration step is included in the preparation of the E. coli extract. Second, proteins produced by the cell-free system are not necessarily homogeneous due to incomplete processing of the N-terminal formyl-methionine residue, which complicates NMR spectra. Therefore, the protein of interest is engineered to contain a cleavable N-terminal histidine-tag, which generates a homogeneous protein after the digestion of the tag. Here, we describe the protocol for modified E. coli cell-free expression. Key words: Cell-free synthesis, S30 extract from E. coli, NMR, Stable isotope labeling

1. Introduction The NMR study of a protein starts by expressing the isotope-labeled protein. It is now widely accepted that the cell-free expression system is one of the most practical methods available for producing isotope-labeled proteins, along with the cellular expression system (1, 2). The cell-free system employs a cell extract from an organism such as Escherichia coli (3–7), an extract from wheat germ (8) or a mixture of recombinant proteins (9). As a consequence of its reduced amino acid metabolic activity in vitro, isotope-labeled amino acids can be efficiently incorporated into a target protein with minimal scrambling (10), which generates new opportunities for advanced selective isotope labeling of proteins, such as stereoarray isotope-labeled proteins (11–13). Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_5, © Springer Science+Business Media, LLC 2012

71

72

M. Takeda and M. Kainosho

We utilize a cell-free system, in which the transcription and translation systems are reconstituted in a vessel by combining an E. coli extract and several components. The E. coli cell-free system has a long history and the level of expression has been improving due to extensive investigations. The E. coli cell-free protein expression system is generally regarded as a practical method for producing isotope-labeled proteins, along with the cellular expression system. E. coli extract is now commercially available. In many of the E. coli cell-free systems, transcription is mediated by T7 RNA polymerase, and thus any plasmid encoding a DNA sequence under the control of the T7 promoter, such as the pET vectors (Novagen), can be used. Therefore, it is now relatively easy to test for expression of proteins of interest by using the E. coli cell-free system. We encountered two problems in the standard protocol of the E. coli cell-free procedure in terms of NMR sample preparation, and thus addressed the problems (14). First, the endogenous amino acids present in the E. coli S30 extract are incorporated into the target protein, which leads to isotope dilution. When (2H, 13C, 15 N)-calmodulin (CaM) protein was produced by using an E. coli S30 extract prepared by the standard protocol, the labeling efficiency was only ~90%. Therefore, to minimize the residual amino acids, we introduced a gel filtration step into the protocol for preparing the E. coli S30 extract. As a result, the labeling efficiency increased to 96% without loss of activity in the extract (14). Second, the formyl group of the N-terminal methionine in proteins produced by the cell-free system is not processed completely. The presence of the formyl moiety causes large, chemical shift changes in nearby residues, giving rise to doubled peaks, which correspond to formylated and deformylated forms. For example, the NMR signals from CaM, which has a flexible N terminus, are affected by the heterogeneity over an extensive region (14). Larger spectral complications may arise from heterogeneity of the peptide chain in proteins with a structured N terminus. We solved this problem by engineering the target protein with a cleavable N-terminal tag. After cleaving the tag, the protein was homogeneous, and the (1H, 15N) heteronuclear single quantum correlation (HSQC) spectrum of this CaM preparation lacked detectable extra peaks (14). To engineer proteins with an N-terminal cleavable tag, a template vector encoding an N-terminal tag followed by the multicloning site is useful. We constructed a family of vectors by modifying a pIVEX2.3d vector (Roche) to encode the N-terminal tag. In these vectors, the amino acid sequences are identical, but the DNA sequence in the region of the N-terminal tag is different. Introducing silent mutations into the DNA sequence in the N-terminal region can lead to different expression levels of the target protein (14). Therefore, plasmids with a higher expression level can be isolated when the target DNA sequence is introduced into the modified

5

Cell-Free Protein Production for NMR Studies

73

pIVEX vectors. In the case of CaM, for example, improvements of the protein yields relative to original sequence were observed for two of the ten pIVEX vectors. In the large-scale reaction performed with the best sequence and 51.1 mg of amino acid mixture (1.7 mg/mL), the amount of CaM synthesized was 5.2 mg (10 wt%). This was twice the yield from the construct prior to the silent mutagenesis (14).

2. Materials 2.1. Preparation of E. coli S30 Extract

1. RNase-free Milli-Q PF Plus water (Millipore) (see Note 1). 2. LB medium: Dissolve 20 g of Luria Broth powder in 1 L of water. 3. Incomplete rich medium: Dissolve 5.6 g of KH2PO4, 28.9 g of K2HPO4, 1 g of Bacto yeast extract, and 1.5 mg of thiamine hydrochloride in 1 L of water, and autoclave the solution. Cool to room temperature and add 50 mL of 40% (w/v) D-glucose and 10 mL of 100 mM Mg(OAc)2. 4. BL21 Star (DE3) cells (Invitrogen) (see Note 2). 5. French press (see Note 3). 6. S30 buffer: Combine 10 mL of 1 M Tris–OAc, pH 8.2, 10 mL of 1.4 M Mg(OAc) 2, and 10 mL of 6 M KOAc in a final volume of 1 L of RNase-free water, atuoclave, and add 1 mL of 1 M DTT. 7. Gel filtration apparatus: Pack Sephadex G25 resin into a 2.5 × 20-cm chromatography column and set the column vertically at 4°C. After the resin settles, attach a funnel to the top of the column and equilibrate with 500 mL of S30 buffer (see Note 4). 8. Dialysis tubing (molecular weight cutoff: 6–8 kDa). 9. Polyethylene glycol (PEG)-8000: Avg. MW 8,000 Da. 10. 2-mercaptoethanol. 11. Diethyl pyrocarbonate (DEPC). 12. Centrifugation tube: Wash the tubes with DEPC water and autoclave to completely remove RNase. 13. 40% (w/v) D-glucose. 14. 100 mM Mg(OAc)2. 15. 1 M Tris–OAc, pH 8.2. 16. 1.4 M Mg(OAc)2. 17. 6 M KOAc. 18. 1 M DTT.

74

M. Takeda and M. Kainosho

2.2. Preparation of T7 RNA Polymerase

1. M9 medium supplemented with tryptone peptone: Combine 1 g of Tryptone peptone (Bectone), 5 g of NaCl, 1 g of NH4Cl, 3 g of KH2PO4, and 6 g of Na2HPO4, and add water to 1 L. After autoclaving, add 10 mL of sterile-filtered 40% (w/v) glucose and 1 mL of sterile 1 M MgSO4. 2. Sonication buffer A: 50 mM Tris–HCl, pH 8.1, 20 mM NaCl, 2 mM EDTA, and 1 mM DTT. 3. Sonication buffer B: 50 mM Tris–HCl, pH 8.1, 20 mM NaCl, 2 mM EDTA, 1 mM DTT, 1.5 mg/mL lysozyme, 20 μg/mL PMSF, 10 μg/mL bacitracin, and 0.1 mM benzamidine. 4. Dialysis buffer: 20 mM sodium phosphate–NaOH, pH 7.7, 1 mM EDTA, 1 mM DTT, and 5% glycerol. 5. SP-Sepharose Fast Flow (GE Healthcare): This resin is packed into a column (3 × 30 cm). 6. Elution buffer: 20 mM sodium phosphate–NaOH, pH 7.7, 1 mM EDTA, 1 mM DTT, 5% glycerol, 200 mM NaCl, and 20 μg/mL PMSF. 7. Isopropyl-beta-D-thiogalactopyranoside (IPTG). 8. 0.8% (w/v) sodium deoxycholate. 9. 2 M Ammonium sulfate. 10. 10% (w/v) Polyethleneimine–HCl, pH 7.0. 11. Saturated ammonium sulfate: Adjust pH to 7.0 using 1 M Tris. 12. E. coli strain BL21 (DE3) transformed with a plasmid encoding T7 RNA polymerase. 13. LB medium (Subheading 2.1). 14. 40% D-Glucose (Subheading 2.1). 15. 1 M MgSO4: Sterilize by autoclaving.

2.3. Cell-Free Synthesis Reaction

1. LM mixture: To prepare the low-molecular-weight mixture, combine 22 mL of 2 M HEPES–KOH, pH 7.5, 33.4 mL of 6 M K(OAc), 210 mg of DTT, 530 mg of ATP, 338 mg of CTP, 335 mg of GTP, 310 mg of UTP, 172 mg of cAMP, 28 mg of folinic acid, and 140 mg of tRNA, and, if needed, 64 mL 50% (w/v) PEG-8000 (see Note 5), and then add RNase-free water up to 200 mL (see Note 6). 2. Dialysis membrane for cell-free reaction: A size 8 cellulose membrane (Viskase Sales Corporation) ensures a reproducible expression level. Cut the membrane to an appropriate size. First, soak the membrane in RNase-free water and then warm it to 70–80°C. After it has cooled down, wash the membrane with RNase-free water. Second, soak the membrane in water containing ~50 mM of NaHCO3, warm it in a microwave to remove glycerol, and then wash it with RNase-free water. Third, soak the membrane in water containing ~1 mM EDTA, warm it in a microwave to further remove glycerol, and then

5

Cell-Free Protein Production for NMR Studies

75

wash it with RNase-free water three times. Finally, soak the membrane in water and autoclave it. 3. 0.645 M Creatine phosphate (see Note 7). 4. 10 mg/mL Creatine kinase (see Note 7). 5. RNase inhibitor (Porcine liver). 6. 1.4 M NH4OAc or (15N) NH4OAc (see Note 8). 7. T7 RNA polymerase at ~11 mg/mL (see Note 9). 8. Template DNA: A plasmid containing the sequence encoding the target protein at ~1 mg/mL (see Note 10). 9. 2× Loading buffer: 2 mL of 0.5 M Tris–HCl, pH 6.8, 4 mL of 10% sodium dodecyl sulfate (SDS), 1.2 mL of 2-mercaptoethanol, 2 mL of glycerol, 1–2 mg of bromophenol blue, and 0.8 mL of water. 10. RNase-free water (Subheading 2.1). 11. 0.5 M Mg(OAc)2. 12. S30 extract (Subheading 3.1). 13. Reaction solution: Mix 112 μL of RNase-free water, 9.8 μL of 1.4 M NH4OAc, 15 μL of 0.5 M Mg(OAc)2, 20 μL of SAIL amino acids, 40 μL of 0.645 M creatine phosphate, 125 μL of LM mixture, 10 μL of 1 mg/mL template DNA, 4.5 μL of 11 mg/mL T7 RNA polymerase, 1.25 μL of 40 U/mL RNase inhibitor, 12.5 μL of 10 mg/mL creatine kinase, and 150 μL of S30 extract for 0.5 mL total volume. 14. Dialysis solution: Mix 1160.8 μL of RNase-free water, 39.2 μL of 1.4 M NH4OAc, 60 μL of 0.5 M Mg(OAc)2, 80 μL of SAIL amino acids, 160 μL of 0.645 M creatine phosphate, and 500 μL of LM mixture for 2 mL total volume. 15. Phosphate buffer: 10 mM potassium phosphate, pH 7.0. Adjust pH using NaOH or HCl. 16. 6 M KOAc (Subheading 2.1). 17. 50% (w/v) PEG-8000. 18. 10% SDS. 19. 0.5 M Tris–HCl, pH 6.8. 20. 2-Mercaptoethanol.

3. Methods The E. coli S30 extract preparation protocol essentially follows previously reported procedures (3, 7, 15), but contains a gel filtration step to minimize the endogenous amino acids. Typically, preparation of the S30 extract takes 2–3 days for one person. We strongly recommend saving an aliquot of the extract at each step to monitor

76

M. Takeda and M. Kainosho

the activity throughout the purification. T7 RNA polymerase is also prepared following established methods (16–18). The preparation of T7 RNA polymerase takes 1 week for one person. The cell-free expression strategy for new proteins is shown in Fig. 1 . As the first step, we recommend starting with E. coli cellular expression of the target protein with uniform 15N labeling.

Fig. 1. Flowchart of the cell-free strategy (reproduced from ref. 14 with permission from Elsevier Science).

5

Cell-Free Protein Production for NMR Studies

77

It is less costly and more efficient to produce proteins from E. coli cells than by using the cell-free method, and cellular expression is better suited for optimizing the purification protocol and the NMR buffer conditions. In general, a protein that exhibits poor cellular expression is also likely to show poor cell-free expression, unless the protein produced is toxic to E. coli cells. Next, several parameters are optimized to maximize the expression level in the cell-free reaction based on small-scale reactions under different conditions. There are various parameters to consider in the optimization process, including the concentration of magnesium, incubation temperature, incubation time, and concentration of PEG. If necessary, the target DNA sequence can be transferred to the line of pIVEX vectors with different silent mutations, as described in the introduction. When an acceptable expression level is attained, a large quantity of the 15N-labeled protein is produced and its (1H, 15N)-HSQC spectrum is compared with that of the protein produced by cellular expression. This comparison helps to detect any possible difference between in vivo and in vitro expression. Special attention should be paid to the incomplete deformylation of the N terminus of proteins in the cell-free reaction. The presence of formylated and deformylated species complicates NMR spectra. To overcome this problem, the use of an engineered N-terminal cleavable tag is helpful, as shown in Fig. 2. 3.1. Preparation of E. coli S30 Extract

1. Inoculate 10 mL of LB medium with the E. coli stock cells (BL21 star (DE3) strain) and incubate the preculture overnight at 37°C with shaking. 2. Add the preculture to 1 L of incomplete rich medium prewarmed to 37°C, and incubate the culture at the same temperature with shaking to an optical density at 600 nm (OD600) of 0.7 (see Note 11). 3. Centrifuge the cells at 5,000 × g for 10 min at 4°C and then wash them three times with 200 mL of ice-cold S30 buffer supplemented with 0.05% 2-mercaptoethanol (see Note 12). Store the cells as a pellet at −80°C. 4. Gently resuspend the cell pellet in 200 mL of ice-cold S30 buffer supplemented with 0.05% 2-mercaptoethanol. Centrifuge at 5,000 × g for 10 min at 4°C and weigh the pellet. Resuspend the pellet in 1.27 mL of S30 buffer per gram of cells. 5. Disrupt the cells with a French Press at 20,000 psi (1,400 kg cm−2) (see Note 13). Immediately after the disruption, add 30 μL of 1 M DTT and centrifuge the lysate at 30,000 × g for 30 min at 4°C in DEPC-treated/autoclaved centrifuge tubes. Carefully collect ~1.4 mL of supernatant per gram of E. coli without mixing with the precipitate.

M. Takeda and M. Kainosho

a

105

110

Q3E2

Q3E2

T5 Q3E2

Q3E2

T70 D2 D2

115

T70 R74

R74 I9 I9

Q3 Q3 K77 K77

D78 D78

M71 M71

A73 A73

K13 K13

120

15N (p.p.m.)

T5

125 A1

L4

L4

M0

130

11

10

9

8

6

7

1H (p.p.m.)

b

105 G113 G33 G59

G40

N137D2

N137D2

Q135E2

Q135E2

T62 G96

N53D2

T5 G98 G25

G134

T29 T44

G61

G132 V55 G23 Q8E2 Q41E2 T146

D95

T117

T26 D58

T110

N53D2

N111D2

Q143E2 N42D2 S17

Q3E2 Q49E2

110

Q41E2 Q8E2 N111D2 Q143E2 N42D2 Q3E2 Q49E2

D22 M145 T79 N60D2

T70 F19 Q135 N60D2 E127 N97D2 R90 E54 M109 Y99 N97D2 M72 D93 D131 D129 F92 S81 D20 R74 E139 D2E67 R106 Q49 N60 T34 N53 I125 M76 I52 Q143 Y138 A103 V91 Q3 M36 Q41 K75 F65 F89 L69 R126 E87 E47 H107 E119 A128 R37 E84 S38V108 L112 M51 F16 E11 D122 M124 V142 E140 E104 N97 E123 Q8 L32 E7 I9 E83D50 M144 L39 E6 E120 E45 F12 D24 L18 L48 E114 K77 E14 K30 D133 A46 A10 E31 D118 L105 V121 D56 R86 D78 M71 V35 A102 I85 A88 L4 F68 A73 E82 A15 N111 K13 D80 A1 K21 S101 I63 K115 K148 V136 F141 L116 K94 N42

D64 I100

115

T28

120

125

A147

I130

I27

130

N137 A57

11

15N (p.p.m.)

78

10

9

8

7

6

Fig. 2. (1H, 15N) HSQC experiments of (U-13C, 15N) CaM. (a) CaM synthesized by the E. coli cell-free system. (b) CaM synthesized with the N-terminal tag following its removal by thrombin digestion. The extra peaks in (a) and the peaks in (b) are labeled with their assignments (reproduced from ref. 14 with permission from Elsevier Science).

5

Cell-Free Protein Production for NMR Studies

79

6. Transfer the collected supernatant to fresh DEPC-treated/ autoclaved tubes. Centrifuge at 30,000 × g for 30 min at 4°C and collect ~1.0 mL of the supernatant per gram of E. coli and place it into a 50-mL tube. 7. Incubate the 50-mL tube at 37°C for 80 min with shaking (see Note 14). 8. Dialyze the solution for 45 min against 2 L of S30 buffer at 4°C using dialysis tubing with a MWCO of 6–8,000 kDa. Repeat the dialysis, and then centrifuge the solution at 15,000 × g for 10 min at 4°C. 9. Apply the supernatant to a gel-filtration column preequilibrated at 4°C with S30 buffer. Load the supernatant onto the column. Maintain a continuous flow of S30 buffer in the column. When the sample first begins to exit the column, as judged from its color and turbidity (see Note 15), collect, in bulk, 1.4 times the total volume applied to the column. 10. Dialyze the collected bulk solution at 4°C for 40–50 min against 700 mL of an equal weight mixture of PEG-8000 and S30 buffer (see Note 16). Adjust the dialysis time so as to concentrate the extract to 0.86 times the initial volume. Dialyze the extract against 2 L of S30 buffer at 4°C for 60 min to remove PEG-8000. 11. Transfer the dialyzed S30 extract to 1.5-mL tubes, freeze it in liquid nitrogen, and store it at −80°C (see Note 17). 12. Assess the extent to which the endogenous amino acids in the extract are eliminated by preparing a protein by using an 2H-, 13 C-, and 15N-labeled amino acid mixture, e.g., SAIL amino acids set. The remaining endogenous amino acids give rise to proton signals in 13C-filtered 1H NMR experiments, as shown in Fig. 3. 3.2. Preparation of T7 RNA Polymerase

1. Using a glycerol stock of E. coli strain BL21 (DE3) transformed with a plasmid encoding T7 RNA polymerase, inoculate 10 mL of LB medium containing the appropriate antibiotic in a 50-mL tube and then incubate with shaking at 37°C overnight. 2. Inoculate 1 L of M9 medium supplemented with tryptone peptone, containing the appropriate antibiotic, with the 10 mL culture from step 1 into a 2-L flask that has been prewarmed to 37°C. Incubate the cells with shaking at 37°C until the culture reaches an OD600 of 0.5. 3. Induce the expression of T7 RNA polymerase by adding IPTG to a final concentration of 0.5 mM. Incubate the cells with shaking for 8 h at 37°C and then centrifuge the culture at 5,000 × g for 10 min at 4°C. Store the cell pellet at −80°C.

80

M. Takeda and M. Kainosho

Fig. 3. 13C-filtered 1H-NMR spectra of CaM. (a) 15N-labeled CaM synthesized with the conventional extract. (b) 2H-, 13C-, 15N-labeled CaM synthesized with the conventional S30 extract. (c) 2H-, 13C-, 15N-labeled CaM synthesized with the improved, dialyzed S30 extract. All of the spectra are adjusted for the intensities of the amide region. Since the peaks of the protons attached to 13C are filtered, only the protons attached to 12C give rise to resonances in the aliphatic region (reproduced from ref. 14 with permission from Elsevier Science).

4. Suspend the frozen cell pellet in 72 mL of sonication buffer A. Add 18 mL of sonication buffer B and stir the mixture for 15 min at 4°C (see Note 18). Add 7.5 mL of 0.8% sodium deoxycholate and stir the mixture for 15 min at 4°C. Disrupt the cells with an ultrasonic generator (five rounds of 1-min sonication at 5-min intervals). 5. Add 15 mL of 2 M ammonium sulfate and bring the volume of the lysate to 150 mL by adding sonication buffer A. Then, gradually add 15 mL of 10% polyethyleneimine and stir the mixture on ice for 20 min. Centrifuge at 39,000 × g for 15 min at 4°C.

5

Cell-Free Protein Production for NMR Studies

81

6. Gradually add 0.82 volume of saturated ammonium sulfate to the supernatant. Stir on ice for 15 min and centrifuge the mixture at 12,000 × g for 15 min at 4°C. 7. Resuspend the precipitate in 45 mL of dialysis buffer containing 100 mM NaCl and 20 μg/mL PMSF. Dialyze the suspension against 2 L of dialysis buffer containing 100 mM NaCl and 20 μg/mL PMSF for 3 h using a membrane with a molecular weight cutoff of 6,000–8,000. Repeat the dialysis for 3 h and overnight. 8. Centrifuge the dialysate at 12,000 × g for 10 min at 4°C. Dilute the dialysate with one volume of dialysis buffer containing 20 μg/mL of PMSF to produce conductivity equal to that of the dialysis buffer plus 50 mM NaCl. Load the dialysate on an SP-Sepharose FF column (3 × 30 cm), equilibrated with dialysis buffer containing 50 mM NaCl and 20 μg/mL of PMSF, and run the 2-column volumes of dialysis buffer containing 50 mM NaCl and 20 μg/mL of PMSF at a flow rate of 1 mL/min to elute impurities. If necessary, to further remove impurities remaining in the column, load up to 500 mL of dialysis buffer containing 50 mM NaCl and 20 μg/mL PMSF at a flow rate of 2 mL/min (see Note 19). 9. Elute the T7 RNA polymerase with elution buffer and dialyze it against dialysis buffer containing 100 mM NaCl and 20 μg/mL of PMSF at 4°C overnight. 10. Repeat the dialysis for 6 h (see Note 20). Finally, dialyze the T7 RNA polymerase against dialysis buffer containing 100 mM NaCl and 50% glycerol at 4°C overnight. Centrifuge at 18,000 × g for 10 min at 4°C, collect the supernatant, and determine the concentration of the T7 RNA polymerase by measuring the optical density at 280 nm (ε280 = 1.4 × 105 M−1 cm−1) (see Note 21). Transfer the protein to 1.5-mL tubes and store it at −20°C. 3.3. Cell-Free Synthesis Reaction

1. Wear sanitary gloves (see Note 22) and thaw the LM mixture and the S30 extract (see Note 23). 2. Prepare the reaction and dialysis solutions (Subheading 2.3). In the case of a small-scale reaction for evaluating expression levels, the typical volumes of the reaction and dialysis solutions are 0.5 and 2.0 mL, respectively. For large production quantities, each volume is scaled up while maintaining the volume ratio of the reaction solution to the dialysis solution (see Note 24). 3. Transfer the dialysis solution to an RNase-free vessel. Tie one end of the dialysis tube firmly. Transfer the reaction solution to the dialysis tube, and tie off the other end. Fold the tubing 4–6 times and place it into the vessel such that the dialysis tube is completely submerged in the dialysis solution.

82

M. Takeda and M. Kainosho

4. Incubate the vessel with shaking for 4–8 h at 37°C (see Note 25). 5. Retrieve the reaction and dialysis solutions. If the expressed protein has a molecular weight smaller than molecular weight cutoff of the membrane, then it will be found in both the reaction and dialysis solutions. 6. To evaluate the expression level, centrifuge 50-μL aliquots of the reaction and dialysis solutions at 5,000 × g for 1 min and 4°C for. Mix the supernatant with an equal volume of 2× loading buffer, incubate for 5 min at 95°C, and load 10 μL on a polyacrylamide gel. Dissolve the precipitate in 50 μL of phosphate buffer (pH 7.0), mix it with an equal volume of 2× loading buffer, and load 10 μL on a polyacrylamide gel.

4. Notes 1. Water, treated with DEPC and then autoclaved, can also be used as RNase-free water (19). However, handle it carefully, since DEPC is a carcinogen. 2. The expression level of BL21 star is 1.2–1.4 times higher than that of the A19 strain. 3. The disruption of the E. coli cells can be accomplished by using equipment other than a French press. 4. The uniformity of the filled resin affects the resolution of gel filtration. The resin is used only once for each preparation. 5. While the use of PEG increases the expression level in some cases, it hampers SDS-PAGE analysis. Therefore, PEG should be removed by ethanol precipitation when the sample is subjected to SDS-PAGE analysis. 6. The prepared LM mixture can be frozen at −20°C for 1 month or more. 7. The creatine phosphate and creatine kinase should be prepared just prior to use. 8. If 15N labeling of the side-chain amide groups in asparagine and glutamine residues is intended, use (15N)-NH4OAc. 9. While we prepare the T7 RNA polymerase ourselves, commercially available T7 RNA polymerase can also be used. 10. The cell-free system uses a large amount of the plasmid, and thus a high-copy plasmid saves both cost and time. 11. The growth rate of the cells in the culture correlates with the activity of the resulting extract (20). We strongly recommend monitoring the growth rate. If the grow rate is slow, then it

5

Cell-Free Protein Production for NMR Studies

83

might be better to restart the culture. On the other hand, overgrowth, i.e., OD600 larger than 0.9, leads to reduced activity of the cells. 12. Do not allow the suspension to foam. 13. The cell disruption is a critical step that affects the activity of the extract. 14. This runoff reaction was performed with a preincubation mixture (14). The expression level does not change, even without the supplement (21). 15. The fraction to be collected looks yellow. 16. Before use, stir the PEG-S30 buffer at 4°C with a large magnetic stir bar to avoid PEG deposition. 17. The presence of DTT is necessary for preserving the activity over time. The activity of an S30 extract prepared without DTT decreases by 10–20% after 3 months of storage at −80°C. 18. The viscosity of the solution increases because of the lysozyme. 19. Wash until the optical density at 280 nm becomes constant. 20. The protein may precipitate during the dialysis. 21. If necessary, concentrate the solution up to about 10 mg/mL with an ultrafiltration device. 22. When setting up the cell-free reaction, use sanitary gloves to prevent contamination with RNases. 23. The frozen S30 extract should be thawed gently on ice. 24. Tryptophan and tyrosine residues are unlikely to be soluble. If needed, warm the amino acid solutions to 60°C. Do not vigorously shake the solution after the template DNA has been added. 25. PEG is assumed to play a role as a stabilizer of the system. For example, in the case of CaM, the ratio of protein yields with and without PEG was 1.6. However, even a small amount of PEG can disturb the SDS-PAGE quantitation of the production level of a target protein. If the solution contains PEG, then precipitate the supernatant with ethanol. Empirically, however, it is difficult to completely remove PEG even by ethanol precipitation. Therefore, optimizing the PEG concentration is somewhat difficult in small-scale cell-free reactions.

Acknowledgments This work was supported by a grant from the Targeted Protein Research Program (MEXT) to M. Kainosho and a Grant-in-Aid for Young Scientists (B) (21770110) to M. Takeda.

84

M. Takeda and M. Kainosho

References 1. Clemens, M.J., and Prujin, G.J. (1999) Protein synthesis in Eukaryotic Cell-free systems. Oxford University Press, New York, pp. 129–165. 2. Kramer, G., Kudlicki, W., and Hardesty, B. (1999) Cell-free Coupled TranscriptionTranslation systems from Escherichia coli. Oxford University Press, New York, pp. 129–165. 3. Zubay, G. (1973) In vitro synthesis of protein in microbial systems. Ann. Rev. Gen. 7, 267–287. 4. Spirin, A.S., Barano, V.I., Ryabova, L.A., Ovodov, S.Y., and Alakhov, Y.B. (1988) A continuous cell-free translation system capable of producing polypeptides in high yield. Science 242, 1162–1164. 5. Kim, D.M., Kigawa, T., Choi, C.Y., and Yokoyama, S. (1996) A highly efficient cell-free protein synthesis system from Escherichia coli. Eur. J. Biochem. 239, 881–886. 6. Kim, D.M., and Swartz, J.R. (2000) Prolonging cell-free protein synthesis by selective reagent additions. Biotechnol. Prog. 16, 385–390. 7. Kigawa, T., Yabuki, T., Yoshida, Y., Tsutsui, M., Ito, Y., Shibata, T., and Yokoyama, S. (1999) Cell-free production and stable-isotope labeling of milligram quantities of proteins. FEBS Lett. 442,15-19. 8. Madin, K., Sawasaki, T., Ogasawara, T., and Endo, Y. (2000) A highly efficient and robust cell-free protein synthesis system prepared from wheat embryos: plants apparently contain a suicide system directed at ribosomes. Proc. Natl. Acad. Sci. U.S.A. 97, 559–564. 9. Shimizu, Y., Inoue, A., Tomari, Y., Suzuki, T., Yokogawa, T., Nishikawa, K., and Ueda, T. (2001) Cell-free translation reconstituted with purified components. Nat. Biotechnol. 19, 751–755. 10. Kigawa, T., Muto, Y., and Yokoyama, S. (1995) Cell-free synthesis and amino acid-selective stable isotope labeling of proteins for NMR analysis. J. Biomol. NMR 6, 129–134. 11. Kainosho, M., Torizawa, T., Iwashita, Y., Terauchi, T., Ono. A,M., and Güntert, P. (2006) Optimal isotope labelling for NMR

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

protein structure determinations. Nature 440, 52–57. Takeda, M., Ikeya, T., Güntert, P., and Kainosho, M. (2007) Automated structure determination of proteins with the SAIL-FLYA NMR method. Nature Protocol 2, 2896–2902. Kainosho, M., and Güntert, P. (2010) SAILStereo-array isotope labeling. Q. Rev. Biophys. 7, 1–54. Torizawa, T., Shimizu, M., Taoka, M., Miyano, H., and Kainosho, M. (2004) Efficient production of isotopically labeled proteins by cell-free synthesis: A practical protocol. J. Biomol. NMR 30, 311–325. Pratt, J.M. (1984) Transcription and Translation: A Practical Approach, IRL Press, New York, pp. 179–209. Davanloo, P., Rosenberg, A.H., Dunn, J.J., and Studier, F.W. (1984) Cloning and expression of the gene for bacteriophage T7 RNA polymerase. Proc. Natl. Acad. Sci. U.S.A. 81, 2035–2039. Zawadzki, V., and Gross, H.J. (1991) Rapid and simple purification of T7 RNA polymerase. Nucl. Acid Res. 19,1948. Grodberg, J., and Dunn, J.J. (1988) ompT encodes the Escherichia coli outer membrane protease that cleaves T7 RNA polymerase during purification. J. Bacteriol. 170, 1245–1253. Huang, Y.H., Leblanc, P., Apostolou, V., Stewart, B., and Moreland, R.B. (1995) Comparison of Milli-Q PF Plus water to DEPC-treated water in the preparation and analysis of RNA. Biotechniques 19, 656–661. Zawada, J., and Swartz, J.R. (2006) Effects of growth rate on cell extract performance in cellfree protein synthesis. Biotechnol. Bioeng. 94, 618–624. Liu, D.V., Zawada J.F., and Swartz, J.R. (2005) Streamlining Escherichia coli S30 extract preparation for economical cellfree protein synthesis. Biotechnol. Prog. 21, 460–465.

Chapter 6 Cell-Free Membrane Protein Expression for Solid-State NMR Alaa Abdine, Kyu-Ho Park, and Dror E. Warschawski Abstract Although cell-free expression is a relative newcomer to the biochemical toolbox, it has already been reviewed extensively, even in the more specialized cases such as membrane protein expression, nanolipoprotein particles, and applications to crystallography and nuclear magnetic resonance (NMR). Solid-state NMR is also a newcomer to the structural biology toolbox, with its own specificities in terms of sample preparation. Cell-free expression and solid-state NMR are a promising combination that has already proven useful for the structural study of membrane proteins in their native environment, the hydrated lipid bilayer. We describe below several protocols for preparing MscL, a mechanosensitive membrane channel, using cell-free expression destined for a solid-state NMR study. These protocols are flexible and can easily be applied to other membrane proteins, with minor adjustments. Key words: In vitro synthesis, Integral membrane proteins, Membrane protein reconstitution, Liposomes, Solid-state NMR

1. Introduction Cell-free expression is one of the major new developments in structural biology, both for nuclear magnetic resonance (NMR) and X-ray crystallography, because it allows for overproduction of proteins, both wild-type and mutant, that are often produced with prohibitively low yields using classical biosynthetic methods. Proteins that are expressed cell-free end up in the reaction container and hence in water where membrane proteins precipitate, often irreversibly. Cell-free expression was first thought to be impractical for membrane protein expression because the system did not include a biological apparatus for targeting the protein to the membrane (1). Providing the medium with detergent was also considered risky since detergent could perturb protein expression by interfering with the ribosome, polymerase or any other essential

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_6, © Springer Science+Business Media, LLC 2012

85

86

A. Abdine et al.

component of the cell-free expression system. These obstacles were overcome in 2004, when several groups, using optimized cell lysates, developed protocols for membrane protein expression in vitro, in the presence of detergents (2–4). For X-ray crystallography and solution-state NMR, the major techniques used in structural biology today, membrane protein structure determination is still a challenge: first, because the aforementioned necessity of obtaining large quantity of functional proteins is aggravated, in vivo, by the limited membrane surface available (5) and second, because membrane proteins are hydrophobic and have to be manipulated in detergents at all times when they are extracted from their native environment, the lipid bilayer (6). Detergents or other surfactants are used for membrane solubilization and during protein purification (which often requires an affinity tag, such as a polyhistidine stretch). They can interfere with protein crystallization or make aggregates that are too large for solution-state NMR. Last but not the least, detergents can interfere with protein folding or function, and the protein structure or dynamics determined in a detergent environment is not necessarily representative of the native structure or dynamics (7, 8). Nevertheless, since 1985, over 200 membrane protein structures have been determined by using X-ray crystallography (9) and, since 1997, about 30 structures by using solution-state NMR (10). In this context, membrane protein cell-free expression has been developed and reviewed, especially for isotope labeling strategies, another issue for NMR, where in vitro synthesis offers a very efficient and versatile alternative that greatly minimizes amino acid scrambling (11, 12). Solid-state NMR is an alternative for membrane protein structure determination inside the hydrated lipid bilayer, where the protein is correctly folded and stable for a long time. Sample preparation for solid-state NMR is therefore different than for solutionstate NMR. If the protein is expressed cell-free in the presence of detergent, it needs to be purified, renatured and reconstituted in a membrane bilayer. Compared to a protein expressed in a cell, the solubilization step is avoided and the purification step is greatly simplified and accelerated, reducing the chances of spurious proteolysis. Reconstitution of the protein in a lipid membrane is the price to pay to obtain a sample where the membrane protein is almost certainly in its native state, and where the protein function can be checked and monitored (13, 14). All these steps are quite flexible, and they are described below for the expression of the mechanosensitive channel MscL. This protocol is general for membrane proteins, although details such as the nature or the concentration of detergent can vary from one protein to the next. Cell-free expression in presence of liposomes or nanodiscs is an alternative sample preparation approach that has been developed more recently and that has yet to be proven general for membrane

6

Cell-Free Membrane Protein Expression for Solid-State NMR

87

proteins (15–17). When possible, it presents many advantages for solid-state NMR studies. First and foremost, the protein is already in its final state, saving time and avoiding many biochemical steps where the protein may be partially lost or inactivated. Second, the protein is purified simply by centrifugation, alleviating the necessity of adding an affinity tag. In addition, no detergent is used that can interfere and that needs to be removed. Importantly, the protein is in its native environment at all times, where it is well folded, functional and stable (18). Finally, this approach is quite flexible, allowing for a variety of labeling strategies (19). In our hands, cell-free expression of MscL in the presence of liposomes has proven efficient (3, 14, 18, 19) and we describe it below, as it is the best method so far for providing a solid-state NMR sample. The protocols presented here make use of commercial kits for cell-free expression, which we found advantageous for its convenience, reliability, for saving the time and manpower to make the lysate, and also for managing the stocks. Cell lysates can also be prepared, following published protocols from various cell types such as bacteria, wheat germ, or others (11, 20). We describe below an optimized procedure for expressing the membrane protein MscL, using the commercial Roche/5Prime continuous exchange vessel and kits, with an Escherichia coli extract. Since, for each sample preparation protocol, it is necessary to check for protein integrity, we also describe several tests that should be performed to assess the quality of the sample.

2. Materials 2.1. Cell-Free Expression of the Mechanosensitive Channel MscL in Detergent Micelles

All solutions are prepared with autoclaved nuclease-free Milli-Q water (see Note 1). 1. Thermoregulated shaker for the cell-free reaction vessel. 2. Nuclease-free 50-mL tubes. 3. Nuclease-free 1.5-mL tubes. 4. Nuclease-free glass pipettes. 5. Nuclease-free pipette tips (0–10, 10–200, 200–1,000 μL, and 1–5 mL), autoclaved at 121°C for 20 min. 6. pIVEX-2.3-mscL plasmid (3). Encodes the E. coli MscL C-terminally fused to a His6 tag under the control of the T7 promoter. Dissolve in pure nuclease-free water and store aliquoted at −20°C, at 0.5 μg/μL. 7. RTS 9000 cell-free expression kit (Roche/5Prime). Contains lyophilized E. coli lysate, reaction mix, feeding mix, reconstitution buffer, continuous exchange reaction vessel, and a syringe (see Fig. 1, Notes 2–5). Store at −80°C.

88

A. Abdine et al.

Fig. 1. Roche/5Prime RTS 9000 reaction vessel. The reaction compartment (10 mL) and the feeding compartment (100 mL) are each accessible through two screws. The reaction compartment contains the cell lysate, plasmid DNA, detergent or preformed liposomes and is where the coupled transcription/translation reaction takes place. The separate feeding compartment provides additional ions, energy substrates, nucleotides and amino acids, through a semipermeable membrane (MW cutoff 10 kDa). Simultaneously, by-products that may inhibit the reaction are diluted through the same membrane into the feeding compartment. This continuous exchange allows cell-free expression to last for up to 24 h.

8. Unlabeled amino acids (powder), store at −20°C. 9.

13

C/15N labeled amino acids (powder), store at −20°C.

10. Dithiothreitol (DTT): Prepare 40 mM stock, store at −20°C. 11. 20% (w/v) Triton X-100: Prepare 200 mL, stir until homogeneous, store at 4°C for up to 6 months. 12. Unlabeled amino acid solutions: Prepare 168 mM solutions of each amino acid except for leucine (140 mM) and the labeled amino acids, Ile and Thr, with the appropriate solution (see Note 6). The amount of amino acid to be incorporated is calculated in Subheading 3.1 and is indicated in Table 1. Transfer the appropriate volume of each unlabeled amino acid to a 1.5-mL vial, and sonicate until the solution looks clear. Transfer the individual solutions to a 50-mL tube. Add Tyr and Leu last to avoid precipitation; prepare fresh before using and keep on ice. 13. Labeled amino acid solutions, Ile and Thr: Dissolve 23 mg of Ile and 3.9 mg of Thr in 1.0 and 0.20 mL of reconstitution buffer, respectively (~168 mM each), into 1.5-mL tubes; prepare fresh before using and keep on ice (see Subheading 3.1, Table 1 and Note 7). 14. FPLC purification system. 15. 1 M NaOH: To adjust pH, store at room temperature. 16. 100 mM NiSO4: Store at room temperature. 17. 20% Ethanol: Store at room temperature for up to a month. 18. 4-(2-Aminoethyl) benzene sulfonyl fluoride hydrochloride (AEBSF): 100 mM stock solution in water, store at 4°C for up to 6 months.

6

Cell-Free Membrane Protein Expression for Solid-State NMR

89

Table 1 Preparation of amino acid solutions for MscL cell-free expression in a volume V = 110 mL (see Subheading 3.1 and Fig. 1) Amino acid

MWi (g/mol)

n

Amount (mg)

Volume (mL)

Ala (A)

89

15

15

0.98

Arg (R)

174

6

11

0.39

Asn (N)

132

6

Asp (D)

133

7

10

0.46

Cys (C)

121

0

0

0.10

Gln (Q)

146

4

6.4

0.26

Glu (E)

147

8

13

0.52

Gly (G)

75

13

11

0.85

His (H)

155

7

12

0.46

Ile (I)

131

16

23

1.0

Leu (L)

131

13

19

1.0

Lys (K)

146

9

14

0.59

Met (M)

149

5

Phe (F)

165

10

Pro (P)

115

7

8.9

0.46

Ser (S)

105

4

4.6

0.26

Thr (T)

119

3

3.9

0.20

Trp (W)

204

0

0

0.10

Tyr (Y)

181

1

2.0

0.10

Val (V)

117

13

8.7

8.2 18

17

0.39

0.33 0.65

0.85

n is the number of each amino acid type, of molecular weight MWi, in the protein sequence. The weight of each amino acid is (n × MWi × 1.1 × 10−2), expressed in mg. Since each amino acid is solubilized at 168 mM, except for leucine, at 140 mM, the corresponding volume of each amino acid solution is (n × 11)/168, except for leucine where it is (n × 11)/140, expressed in mL

19. Chelating column (5 mL): Stored in 20% ethanol at 4°C. 20. FPLC buffer A1: 50 mM NaH2PO4–NaOH (7.1 g/L), pH 8, 300 mM NaCl (17.5 g/L), 10 mM imidazole (680 mg/L), 4.0% Triton X-100 (40 g/L), store at 4°C. 21. FPLC buffer A2: 50 mM NaH2PO4–NaOH (7.1 g/L), pH 8, 300 mM NaCl (17.5 g/L), 10 mM imidazole (680 mg/L), 0.2% Triton X-100 (2 g/L), store at 4°C.

90

A. Abdine et al.

22. FPLC buffer B: 50 mM NaH2PO4–NaOH (7.1 g/L), pH 8, 300 mM NaCl (17.5 g/L), 500 mM imidazole (34 g/L), 0.2% Triton X-100 (2 g/L), store at 4°C. 23. FPLC sample loading buffer: 50 mM NaH2PO4–NaOH (7.1 g/L), pH 8, 300 mM NaCl (17.5 g/L), 10 mM imidazole (680 mg/L), 1% Triton X-100 (10 g/L), store at 4°C. 24. 10× Dialysis buffer: 0.1 M HEPES–KOH, pH 7.5, 1 M KCl, and 2% (w/v) Triton X-100, store at 4°C for up to 6 months. 25. 1× Dialysis buffer: Prepare 2 L by diluting 200 mL of 10× buffer in 1,750 mL of water, adjusting the pH to 7.5 using 4 M KOH, and adding water to 2 L final volume. Store at 4°C for up to 6 months. 26. 2 dialysis cassettes with a 10 kDa MW cutoff. 27. Methanol. 28. DOPC: 1,2-dioleoyl-sn-glycero-3-phosphocholine powder, store at −20°C. 29. 4 M KOH: To adjust pH, store at room temperature. 30. 0.8% Triton X-100: Prepare 10 mL, dissolve and stir until homogeneous, store at 4°C for up to 6 months. 31. Wet polystyrene beads: Wash 5 g of 300–1,200 μm polystyrene beads with 25 mL of pure methanol and then wash four times with 25 mL of Milli-Q water. The beads are kept in water at 4°C and are stable for months, provided a weekly renewal of Milli-Q water. 32. HEPES–KCl buffer: 10 mM HEPES–KOH, pH 7.5, 100 mM KCl, store at 4°C for up to 6 months. 33. HEPES solutions for solubilizing Tyr, Trp, and Phe (Subheading 2.1, see Notes 6 and 7): 60 mM HEPES, pH 13 for Tyr, pH 1 for Trp, and pH 7.5 for Phe. Adjust the pH using KOH or HCl. Keep on ice. 2.2. Cell-Free Expression of the Mechanosensitive Channel MscL in Liposomes

All solutions are prepared with autoclaved nuclease-free Milli-Q water (see Note 1). 1. Thermoregulated shaker for the cell-free reaction vessel. 2. Mini-Extruder with two 1-mL gastight microsyringes and 0.1-μm cutoff polycarbonate membranes. 3. Nuclease-free 50-mL tubes. 4. Nuclease-free 1.5-mL tubes. 5. Nuclease-free glass pipettes. 6. Nuclease-free pipette tips (0–10, 10–200, 200–1000 µL and 1–5 mL). Autoclave at 121°C for 20 min.

6

Cell-Free Membrane Protein Expression for Solid-State NMR

91

7. pIVEX-2.3-mscL plasmid: For protein expression in liposomes, an affinity tag such as a polyhistidine stretch is preferable but not mandatory. 8. RTS 9000 cell-free expression kit (Roche/5Prime). Contains lyophilized E. coli lysate, reaction mix, feeding mix, reconstitution buffer, a continuous exchange reaction vessel, and a syringe (see Fig. 1, Notes 2–5), store at −80°C. 9. Unlabeled amino acids (powder), store at −20°C. 10.

13

C/15N labeled amino acids (powder), store at −20°C.

11. DTT (dithiothreitol): Prepare 40 mM stock, store at −20°C. 12. 4 M KOH: To adjust pH, store at room temperature. 13. AEBSF (Subheading 2.1). 14. DOPC. 15. Chloroform. 16. HEPES–KCl buffer (Subheading 2.1). 17. DOPC liposome preparation: Prepare a chloroform solution of 10 mg/mL of DOPC (100 mg of DOPC in 10 mL of dry chloroform). Dry the lipids and then resuspend with RTS reconstitution buffer to obtain a 20 mg/mL aqueous solution. Sonicate 2.5 mL of the solution for 5 min, and then extrude it 13 times with a 0.1-μm filter, to obtain a liposome solution containing 40–50 mg of lipids, and store at 4°C. This is the amount of lipids that is expected in the final sample. DOPC can be replaced by other lipids or lipid mixtures, at a higher or lower concentration. 18. HEPES solutions for solubilizing Tyr, Trp, and Phe (Subheading 2.1). 19. Unlabeled amino acid solutions: Prepare 168 mM solutions of each amino acid except for leucine (140 mM) and the labeled amino acids (item 20), with the appropriate solution (see Note 6). The amount of amino acid to be incorporated is calculated in Subheading 3.1 and is indicated in Table 1. The appropriate volume of each unlabeled amino acid is transferred to a 1.5-mL vial, and sonicated until the solution looks clear. The vials are transferred into a 50-mL tube. Tyr and Leu are added last, to avoid precipitation; prepare fresh before using and keep on ice. 20. Labeled amino acid solutions, Arg, Ile, Pro, Met, and Phe: Weigh 11 mg of Arg, 23 mg of Ile, and 8.9 mg of Pro and dissolve in 0.39, 1.0, and 0.46 mL of reconstitution buffer, respectively (~168 mM). Weigh 8.2 mg of Met and dissolve in 0.33 mL of reconstitution buffer plus 50 μL of 40 mM DTT. Weigh 18 mg of Phe and dissolve in 0.65 mL of the appropriate HEPES solution (Subheading 2.1), and sonicate for 1 min. Prepare each amino acid solution fresh before using and keep on ice (see Subheading 3.1, Table 1 and Note 7).

92

A. Abdine et al.

2.3. Sample Characterization

All solutions described should be prepared using Milli-Q water and stored at room temperature unless stated otherwise. 1. FA diluted solution: 37% formaldehyde in water, store at 4°C. 2. Dimethyl sulfoxide (DMSO). 3. 1 M NaOH: To adjust pH. 4. 1 M HCl: To adjust pH. 5. 1 M Tris base, pH 8. 6. SDS running buffer: 100 mM Tris base, 100 mM HEPES, 0.1% SDS, adjust to pH 8 using NaOH or HCl. 7. Staining buffer: 0.5% Coomassie Brilliant Blue R-250 dye, 50% ethanol and 10% acetic acid. 8. Destaining buffer: 20% ethanol and 10% acetic acid. 9. 2× SDS loading buffer: 0.2 M Tris base, pH 8, 8% SDS, 40% glycerol, 0.4% bromophenol blue, 0.4 M DTT, adjust pH 8 using 1 M NaOH or 1 M HCl. 10. 1× SDS loading buffer: dilute 1 mL of 2× SDS loading buffer with 1 mL of water. 11. DSS solution: 50 mM disuccinimidylsuberate in DMSO. 12. HEPES–KCl buffer (see Subheading 2.1). 13. Polyvinylidene fluoride (PVDF) membranes. 14. TBS-Tween buffer 1: 50 mM Tris base, pH 7.4, 150 mM NaCl, 0.05% Tween-20. 15. Monoclonal anti-histidineperoxidase-conjugate antibody solution (for the dot blot): Dilute 2,000× from the stock solution with TBS-Tween buffer 1, prepare fresh. 16. Diaminobenzidine solution: Dissolve one tablet of diaminobenzidine and one tablet of urea hydrogen peroxide in 5 mL of water, prepare fresh. 17. Western-blotting detection kit. 18. Nonfat dry milk (powder). 19. Chemiluminescence reagents (H2O2 and luminol), store at 4°C. 20. Towbin buffer: 25 mM Tris base, 192 mM glycine, 20% isopropanol, and 0.1% SDS. Store without isopropanol at 4°C, add isopropanol just before use. 21. Ponceau S solution: 0.1% Ponceau S, 5% acetic acid. 22. TBS-Tween buffer 2: TBS-Tween buffer 1 with 5% (w/v) nonfat dry milk, freshly made. 23. Mouse anti-histidine antibody solution (for the Western-blot): Dilute 5,000× from the stock solution with TBS-Tween buffer 1, prepare fresh. 24. Anti-mouseperoxidase-coupled antibody solution (for the Western-blot): Dilute 10,000× from the stock solution with TBS-tween buffer 1, prepare fresh.

6

Cell-Free Membrane Protein Expression for Solid-State NMR

93

25. Micro BCA Protein Assay kit (Pierce). 26. Nuclease-free 1.5-mL tubes. 27. Bovine serum albumine (BSA): 1 mg/mL stock solution. 28. Solubilization buffer: 0.2% Triton X-100, solubilized in HEPES–KCl buffer, store at 4°C for up to 6 months. 29. Acetone, stored at −20°C. 30. Nitric acid solution: 1%, store at room temperature in a safety cabinet. Handle under the hood. 31. Perchloric acid solution: 70%, store at room temperature in a safety cabinet. Handle under the hood. 32. Ammonium molybdate solution: Dissolve 2.5 g in 100 mL of water, store at 4°C. 33. Ascorbic acid solution: Dissolve 10 g in 100 mL of water, prepare fresh, and do not keep for more than 1 week at 4°C. 34. Phosphate stock solution: Dissolve 10 mg of KH2PO4 in 100 mL of water and store at 4°C. 35. Nuclease-free 3-mL ultracentrifugation tubes. 36. Sucrose. 37. Sucrose solutions: 10, 20, and 45% (w/v) sucrose, solubilized in HEPES–KCl buffer, store at 4°C. 38. Nitrocellulose membrane (for Western blotting). 39. X-ray film (to develop the proteoliposomes must be mixed with sucrose Western blot). 40. Film developer. 2.4. NMR Sample Preparation

Four millimeter diameter rotors for high-resolution magic-angle spinning solid-state NMR (see Fig. 2).

Fig. 2. Four millimeter diameter rotors for high-resolution magic-angle spinning solidstate NMR. The sample (approximately 50 μL) is contained in the center of the zirconium rotor, thanks to the Teflon insert bottom and top, and is kept tightly in place by the top screw and the Kel-F cap.

94

A. Abdine et al.

3. Methods The methods outlined below describe the following: (1) calculation of the amount of incorporated amino acids, (2) cell-free expression of the mechanosensitive channel MscL in detergent micelles, purification of protein/detergent micelles, protein reconstitution into liposomes, (3) cell-free expression directly into liposomes, (4) sample characterization, and (5) NMR sample preparation. 3.1. Cell-Free Expression of the Mechanosensitive Channel MscL in Detergent Micelles

The method below describes the production of an MscL sample for solid-state NMR, using a cell-free protein expression system in detergent micelles. Although any labeling scheme could be performed, the sample described below was labeled on isoleucines and threonines, using 13C/15N-labeled amino acids, and was subsequently used in solid-state NMR experiments (14).

3.1.1. Day 1: Cell-Free Protein Expression in Detergent Micelles

1. Calculate the amount of amino acids to be incorporated during protein expression. The concentration of each amino acid in the protein cell-free expression system (V = 110 mL, see Fig. 1) is usually ~1 mM. For a given protein, each amino acid concentration depends on its occurrence, n, in the protein sequence. In our case, we have obtained good results with a final concentration of each amino acid equal to n/10, expressed in mM. The corresponding weight of each amino acid, of molecular weight MWi, is therefore (n × V × MWi/10), expressed in mg. The corresponding volume of each amino acid solution of concentration Ci, where Ci is 168 or 140 μM (Subheading 2.1), is (n × V/10)/Ci, expressed in mL. The weights and final volumes calculated for this protocol are summarized in Table 1. 2. Thaw the DTT, RTS reconstitution buffer, and MscL plasmid at room temperature. 3. Thaw the other components of the RTS kit (E. coli lysate, reaction mix, and feeding mix) on ice. 4. Prepare the unlabeled amino acid solutions (see Subheading 2.1). 5. Prepare the labeled amino acid solutions (see Subheading 2.1). 6. Add the labeled amino acid solution to the unlabeled amino acid solution. 7. Add 3 mL of 40 mM DTT, sonicate (80 W for 5 min or until clear), and store on ice. 8. Reconstitute the lyophilized E. coli lysate in 5.2 mL of reconstitution buffer. Shake the bottle gently (see Note 8). 9. Reconstitute the lyophilized reaction mix in 2.2 mL of reconstitution buffer. Shake the bottle gently (see Note 8).

6

Cell-Free Membrane Protein Expression for Solid-State NMR

95

10. Reconstitute the lyophilized feeding mix in 80 mL of reconstitution buffer. Shake the bottle gently (see Note 8). 11. Prepare the feeding solution by adding 26 mL of the reconstituted amino acid solutions (mixture of unlabeled and labeled amino acid solutions) and 3 mL of 40 mM DTT to the feeding mix (80 mL). 12. Prepare the reaction solution by adding the reconstituted reaction mix (2.2 mL), 2.7 mL of the reconstituted amino acid solution and 0.3 mL of 40 mM DTT to the reconstituted E. coli lysate (5.2 mL). Remove a 20 μL aliquot of this solution for later comparison on gel electrophoresis. 13. Add 300 μL of the MscL plasmid (~150 μg) to the reaction solution. 14. Add 200 μL of 20% Triton X-100 solution to the reaction solution. 15. Add 2.2 mL of 20% Triton X-100 solution to the feeding solution. 16. Open both screws of the reaction compartment and fill the reaction compartment with the reaction solution using a nuclease-free pipette (see Fig. 1). Remove air bubbles by tapping the vessel lightly. 17. Open both screws of the feeding compartment and fill the feeding compartment with the feeding solution, using the provided syringe (see Fig. 1). Remove air bubbles by tapping the vessel lightly. 18. Insert the reaction vessel into a thermoregulated shaker. 19. Set the shaking speed to 800 rpm. 20. Set the temperature to 30°C. 21. Incubate the reaction for 22 h. 3.1.2. Day 2: Purification of Detergent/Protein Micelles

1. After 22 h, stop the cell-free protein expression. Remove a 20 μL aliquot of the reaction run-on solution for gel electrophoresis. 2. Remove the reaction run-on mix using a pipette, add 100 μL of the AEBSF solution to prevent protease digestion, and stir at room temperature for 15 min (see Note 9). 3. Dilute the sample to a final volume of 50 mL using FPLC buffer A1 and centrifuge at 10,000 × g for 15 min at 4°C. 4. Connect the chelating column to the FPLC system (see Note 10). Pass 4–5 column volumes (CV) of distilled water through the column, to wash away the ethanol solution in which it has been kept. 5. Degas the FPLC buffers for 10 min in an ultrasonic bath. Wash the pump inlets with the degassed solutions. Wash the system with FPLC buffer A1 until the UV and conductivity baselines are stable.

96

A. Abdine et al.

6. At the end of the sample centrifugation (step 3), dilute the supernatant (reaction run-on mix) with FPLC sample loading buffer to a final volume of 100 mL (final dilution ~10×). 7. Filter this solution through a 0.45-μm filter. 8. Resuspend the pellet with 5 mL of FPLC buffer A1 and remove a 20 μL aliquot for gel electrophoresis. 9. Introduce half (50 mL) of the diluted reaction run-on mix into the column (see Note 11). Collect the flow-through in the first 50-mL tube. 10. Wash off the nonspecifically bound material by raising the concentration of imidazole to 100 mM, by mixing 80% of FPLC buffer A1 with 20% of FPLC buffer B (6 CV total). The concentration of detergent remains high (around 3%). Collect the flow-through in the second 50-mL tube. 11. To elute the protein, switch FPLC buffer A1 to buffer A2, to reduce the detergent concentration to 0.2%. Collect the flowthrough in the second 50-mL tube. 12. Increase the imidazole concentration to 200 mM by mixing 60% of solution A2 with 40% of solution B (2 CV total), and collect eluant fractions of 2 mL each (see Note 12). 13. Increase the imidazole concentration to 500 mM by switching to 100% FPLC buffer B (6 CV total) and collect eluant fractions of 2 mL each. 14. Wash the column with FPLC buffer A1 (10 CV or until the UV and conductivity baselines are stable), then repeat steps 9–13 with the remaining 50 mL (second half of the diluted reaction run-on mix, see Note 11). 15. At the end of the FPLC protein purification run (see Note 10), the potentially interesting fractions are separated into approximately 40 tubes, whereas the flow-though is contained in two 50-mL tubes. Remove a 20 μL aliquot from each of these 42 tubes, for gel electrophoresis. 16. Determine which fractions contain the histidine-tagged protein by performing a dot blot (see Subheading 3.3.4). 17. Wash the two dialysis cassettes in 1× dialysis buffer. 18. For each half of the solution, pool the fractions containing the protein into a dialysis cassette. 19. Dialyze against 1 L of 1× dialysis buffer, overnight at 4°C. 3.1.3. Day 3: Sample Characterization and Reconstitution into Liposomes

1. Dialyze a second time against 1 L of 1× dialysis buffer for 2 h at 4°C. Remove a 20 μL aliquot for gel electrophoresis. 2. Assess the cell-free expression level, FPLC purification efficiency, and oligomeric state by SDS-PAGE (see Subheadings 3.3.1 and 3.3.3).

6

Cell-Free Membrane Protein Expression for Solid-State NMR

97

3. After SDS-PAGE, wash the gel in water and perform a Westernblot transfer (see Subheading 3.3.5). 4. Estimate the protein concentration. Since MscL does not contain any tryptophan it has to be quantified by using the BCA method (see Subheading 3.3.6 and Note 13). 5. BCA quantification of our MscL preparation indicated that each half of the solution contains approximately 4.8 mg of protein in a 10 mL volume that will be reconstituted in 19.2 mg of DOPC liposomes (protein–lipid ratio of 1:4). DOPC can be replaced by other lipids or lipid mixtures, at a higher or lower concentration (21). Lipids can also be replaced by other mixtures to reconstitute the protein into bicelles or nanodiscs (22, 23). 6. Split the solutions into four fractions of 5 mL, each containing 2.4 mg of protein and 10 mg of Triton X-100. To each fraction, add 9.6 mg of DOPC solubilized in 2.5 mL of the 0.8% Triton X-100 solution, such that the final lipid–detergent ratio is 1:3. Stir slowly for 15 min at room temperature. 7. To each fraction of 7.5 mL add 300 mg of wet polystyrene beads (detergent–bead ratio is 1:10) and put the four tubes on a rotator, to remove the detergent by adsorption onto the beads. Incubate overnight at 4°C. For alternative methods (21, 24), see Note 14. 3.1.4. Day 4: Preparation of the Sample for Solid-State NMR

1. Add another 300 mg of wet polystyrene beads to each tube. Incubate for another 2 h, at 4°C. 2. After 2 h, set the tubes vertically for 5 min and discard the pellet and the beads. 3. Divide the supernatant into four tubes and centrifuge at 100,000 × g for 30 min at 4°C. Remove a 20 μL aliquot of one supernatant for gel electrophoresis, discard the rest and resuspend the pellets with 1 mL of HEPES–KCl buffer for each pellet. Split the new resuspended pellets into two tubes and centrifuge at 100,000 × g for 30 min at 4°C. Repeat one more time so that the entire sample fits into a single tube. Remove a final 20 μL aliquot of the supernatant for gel electrophoresis. Dry the final pellet under argon and store it at −20°C. 4. Assess the reconstitution by SDS-PAGE (see Subheading 3.3.1), and by proteoliposome density characterization on a sucrose gradient (see Subheading 3.3.9). Also, estimate the lipid and water content (see Subheadings 3.3.8 and 3.3.10, and Note 15). 5. The average quantity of protein obtained with the RTS 9000 kit is 10 mg in 10 mL of reaction mix, which is generally sufficient to prepare two or three NMR samples. Transfer ~30 mg of pelleted sample to the NMR rotor (see Subheading 3.4). 6. Store the remaining pellet at −20°C.

98

A. Abdine et al.

3.2. Cell-Free Expression of the Mechanosensitive Channel MscL in Liposomes

The method below describes the production of an MscL sample for solid-state NMR, using a cell-free protein expression system directly into lipid vesicles. Although any labeling scheme could be performed, the sample described below was labeled on arginines, isoleucines, methionines, phenylalanines and prolines, using 13 C/15N-labeled amino acids, and was subsequently used in solidstate NMR experiments (19). Lipids used were synthetic DOPC, but other lipids or complex mixtures can also be used, such as asolectin (18) or nanodiscs (17).

3.2.1. Day 1: Cell-Free Protein Expression

1. Thaw the DTT solution, RTS reconstitution buffer and MscL plasmid at room temperature. 2. Thaw the other components of the RTS kit (E. coli lysate, reaction mix, and feeding mix) on ice. 3. Prepare the DOPC liposomes (Subheading 2.2). 4. Prepare the HEPES solutions (Subheading 2.1). 5. Prepare the unlabeled amino acid solutions (Subheading 2.1). 6. Prepare the labeled amino acid solutions (Subheading 2.1). 7. Add the labeled amino acid solutions to the unlabeled amino acid solutions. 8. Add 3 mL of 40 mM DTT, sonicate (80 W for 5 min or until clear), and store on ice. 9. Reconstitute the lyophilized E. coli lysate in 2.7 mL of reconstitution buffer. Shake the bottle gently (see Note 8). 10. Add the liposome solution (2.5 mL) to the E. coli lysate once the latter is completely dissolved. 11. Reconstitute the lyophilized reaction mix in 2.2 mL of reconstitution buffer. Shake the bottle gently (see Note 8). 12. Reconstitute the lyophilized feeding mix in 80 mL of reconstitution buffer. Shake the bottle gently (see Note 8). 13. Prepare the feeding solution by adding 26 mL of the reconstituted amino acid solution (mixture of unlabeled and labeled amino acid solutions) and 3 mL of 40 mM DTT to the feeding mix solution (80 mL). 14. Prepare the reaction solution by adding the reconstituted reaction mix (2.2 mL), 2.7 mL of the reconstituted amino acid solution and 0.3 mL of 40 mM DTT to the reconstituted E. coli lysate with the liposomes (5.2 mL). Take a 20 μL aliquot of this reaction solution for later comparison on gel electrophoresis. 15. Add 300 μL of the MscL plasmid (~150 μg) to the reaction solution. 16. Open both screws of the reaction compartment and fill the reaction compartment with the reaction solution using a nuclease-free pipette (see Fig. 1). Remove any air bubbles by tapping the vessel lightly.

6

Cell-Free Membrane Protein Expression for Solid-State NMR

99

17. Open both screws of the feeding compartment and fill the feeding compartment with the feeding solution, using the provided syringe (see Fig. 1). Remove any air bubbles by tapping the vessel lightly. 18. Insert the reaction vessel into a thermoregulated shaker. 19. Set the shaking speed to 800 rpm. 20. Set the temperature to 30°C. 21. Incubate the reaction for 22 h. 3.2.2. Day 2: Preparation of the Sample for Solid-State NMR

1. After 22 h, stop the cell-free protein expression. Remove a 20 μL aliquot of the reaction run-on solution for gel electrophoresis. 2. Remove the reaction run-on mix using a pipette, add 100 μL of the AEBSF solution to prevent protease digestion, and stir at room temperature for 15 min (see Note 9). 3. Split the sample into six tubes and centrifuge at 100,000 × g for 30 min at 4°C. Remove a 20 μL aliquot of one supernatant for gel electrophoresis, discard the rest and resuspend each pellet with 1 mL of HEPES–KCl buffer. Split the resuspended pellets into three tubes and centrifuge at 100,000 × g for 30 min at 4°C. Repeat one more time so that the entire sample fits into a single tube. Remove a final 20 μL aliquot of the supernatant for gel electrophoresis. Dry the final pellet under argon and store it at −20°C. 4. Assess cell-free expression efficiency and oligomeric state by SDS-PAGE (see Subheadings 3.3.2 and 3.3.3). At this point, the expressed protein is generally almost pure. 5. If the protein was expressed with a polyhistidine tag, after SDSPAGE, wash the gel in water and perform a Western-blot transfer (see Subheading 3.3.5). 6. Estimate the protein and lipid concentration using the BCA and Rouser methods respectively (see Subheadings 3.3.7 and 3.3.8, and Note 13). 7. Characterize the proteoliposome density on a sucrose gradient (see Subheading 3.3.9), and estimate the water content (see Subheading 3.3.10 and Note 15). 8. The average quantity of protein obtained with the RTS 9000 kit is 10 mg in 10 mL of reaction mix, which is generally sufficient to prepare two or three NMR samples. About 30 mg of pelleted sample is transferred into the NMR rotor (see Subheading 3.4). 9. Store the remaining pellet at −20°C.

100

A. Abdine et al.

3.3. Sample Characterization 3.3.1. Electrophoresis of Protein Expressed in Detergent Micelles

1. Perform SDS-PAGE on the following samples: (a) Aliquots from the reaction mix before and after the cell-free expression reaction, to check for expression of the target protein. (b) Aliquots from the centrifugation pellet and supernatant, to check whether the protein was totally solubilized in detergent micelles or if there are some aggregates. (c) Aliquots from the purification flow-through and the purified fractions. (d) Analyze the incorporation of the protein by comparing aliquots of the protein-detergent micelles with the proteoliposomes, after the reconstitution step. (e) Aliquots of the supernatant of all the wash steps. (f) Also check the integrity of the purified protein after dialysis. 2. Dilute all SDS PAGE samples 1:1 with 2× SDS loading buffer. 3. Heat the samples to 90°C for 5 min. 4. Cool the samples and load on a gel. 5. Run the gel using SDS running buffer. 6. Wash the gel in water and stain/destain the gel using staining/ destaining buffers, respectively.

3.3.2. Electrophoresis of Protein Expressed in Liposomes

1. Perform SDS-PAGE on the following samples: (a) Aliquots from the reaction mix before and after the cellfree expression reaction, to check for expression of the target protein. (b) Also check the centrifugation supernatants. 2. Follow steps 2–6 in Subheading 3.3.1.

3.3.3. Electrophoresis for Oligomeric State Characterization

If the protein is an oligomer (like MscL), assess its integrity by cross-linking the protein with different free primary amine targeting reagents, like formaldehyde (FA) or disuccinimidylsuberate (DSS) (3, 25). NB: the protein buffer must not contain primary amines (such as in Tris). 1. Mix 5 μg of purified MscL, either in detergent or liposomes, in 10 μL of HEPES–KCl buffer and 0.54 μL of the FA diluted solution. 2. Mix 5 μg of purified MscL, either in detergent or liposomes, in 10 μL of HEPES–KCl buffer and 0.2 μL of 50 mM DSS. 3. Incubate both samples, without shaking, for 30 min at room temperature.

6

Cell-Free Membrane Protein Expression for Solid-State NMR

101

Fig. 3. Coomassie blue stained SDS-PAGE of MscL: (1) purified in detergent, (2) crosslinked with formaldehyde, (3) cross-linked with disuccinimidylsuberate. Formaldehyde generates five protein bands on the gel, corresponding to various oligomeric forms, from monomers to pentamers, while disuccinimidylsuberate generates mostly pentamers, confirming the pentameric nature of Escherichia coli MscL (3, 25).

4. Stop the cross-linking reactions by adding 2 μL of 1 M Tris base, pH 8, to each sample. Gently mix and incubate for 10 min at room temperature. 5. Add 3 μL of 2× SDS loading buffer to each sample and mix. 6. Heat the FA sample to 60°C and the DSS sample to 90°C, for 5 min in a water bath. 7. Cool the samples and load on a gel. 8. After migration, wash the gel in water and stain/destain the gel using staining/destaining buffers, respectively (see Fig. 3) or perform a Western-blot transfer (see Subheading 3.3.5). 3.3.4. Dot Blot

1. Deposit 2 μL of each fraction of purified protein directly onto a PVDF membrane. 2. Allow the protein solution to air-dry. 3. Prepare the TBS-Tween buffers 1 and 2. 4. Prepare the monocolonal antibody solution. 5. When the protein spots are completely dry, incubate the membrane in the TBS-Tween buffer 2 for 30 min. 6. Wash the membrane twice for 5 min with the TBS-Tween buffer 1.

102

A. Abdine et al.

7. Incubate the membrane with the monoclonal antibody solution for 1 h. 8. Wash the membrane twice for 5 min with the TBS-Tween buffer 1. 9. Prepare the diaminobenzidine solution. 10. Dry the membrane on a tissue. 11. Incubate the membrane with the diaminobenzidine solution for a few seconds, until the spots are colored. 12. Dry the membrane on a tissue. The color remains on the spots containing the histidine-tagged protein, while it vanishes on the other spots. 3.3.5. Semidry Western-Blot

1. Prepare the Towbin buffer. 2. Prepare the transfer sandwich by superimposing two filter papers, the gel, the nitrocellulose membrane and another two filter papers in the Towbin buffer. 3. Place the transfer sandwich on the anode plate and clip the cathode plate on top. 4. Clamp the potential at 25 V and blot for 20 min. 5. Prepare TBS-Tween buffers 1 and 2, mouse anti-histidine antibody, and anti-mouse peroxidase-coupled antibody solutions. 6. Assess the sample transfer and relative concentration by staining the membrane with Ponceau S solution. This is a reversible stain that can be removed by washing in TBS-Tween buffer 1. 7. After protein transfer and destaining, saturate the nitrocellulose membrane by incubating in TBS-Tween buffer 2 for 30 min. 8. Rinse the membrane quickly with TBS-Tween buffer 1. 9. Incubate for 45 min at room temperature with the mouse antihistidine antibodies solution. 10. Wash the membrane for 4 × 5 min with TBS-Tween buffer 1. 11. Incubate for 30 min at room temperature with the anti-mouse peroxidase-coupled antibody solution. 12. Wash the membrane for 5 × 5 min with TBS-Tween buffer 1. 13. Incubate the membrane for 1 min with 1 mL of each of the chemiluminescence reagents (H2O2 and luminol). 14. In a light tight box, expose X-ray films to the membrane for varying times, usually between 30 s and 2 min, and develop the films in the photo developer.

6 3.3.6. Protein Quantification in Detergents

Cell-Free Membrane Protein Expression for Solid-State NMR

103

Protein concentration is measured using the BCA method (26, 27), using the test tube procedure of the Micro BCA Protein Assay Kit: 1. Prepare eight BSA standards by diluting the stock solution with the solubilization buffer (the final BSA concentration should fall between 0.5 and 20 μg/mL). Make three replicates of each dilution. Introduce 500 μL of each into 1.5-mL tubes (24 tubes in total) and store at room temperature. 2. Dilute the unknown protein sample with the solubilization buffer to three different concentrations that are expected to be between 1 and 20 μg/mL. Make three replicates of each dilution. Transfer 500 μL of each sample into 1.5-mL tubes (nine tubes in total). 3. Add 500 μL of Micro BCA Working Reagent to each of the 1.5-mL tubes (24 BSA standards and nine unknown proteins) and incubate for 60 min at 60°C. Cool to room temperature. 4. Measure the absorbance at 562 nm. 5. Plot a standard curve based on the absorbance of the BSA samples. 6. Deduce the protein concentration of each unknown protein sample.

3.3.7. Protein Quantification in Proteoliposomes

Lipids need to be removed from the sample by precipitating the proteins using cold acetone, followed by a centrifugation to separate the protein pellet from the supernatant containing the lipids: 1. Prepare the 24 BSA standard tubes (Subheading 3.3.6, step 1). 2. Dilute the unknown protein sample with the solubilization buffer to three different concentrations that are expected to be between 1 and 20 μg/mL. Make three replicates of each dilution. Transfer 500 μL of each sample into 1.5-mL tubes (nine tubes in total). 3. Add 1 mL of cold acetone to each tube of unknown protein. Vortex the tubes and incubate for 60 min at −20°C. Centrifuge at 10,000 × g for 10 min at room temperature. Discard the supernatant. Incubate the tubes for 30 min at room temperature to allow for acetone evaporation. Add 500 μL of the solubilization buffer, and vortex the tubes again. 4. Add 500 μL of Micro BCA Working Reagent to each of the 1.5-mL tubes (24 BSA standards and nine unknown proteins) and incubate for 60 min at 60°C. Cool to room temperature. 5. Measure the absorbance at 562 nm. 6. Plot a standard curve based on the absorbance of the BSA samples. 7. Deduce the protein concentration of each unknown protein sample.

104

A. Abdine et al.

3.3.8. Lipid Quantification in Proteoliposomes

The phospholipid content in the proteoliposome sample is assessed by the Rouser method, which measures the phosphate concentration in the sample (28). Full eye, face, and skin protection is required for this method. 1. Set a heating block at 180°C. 2. Wash the glass tubes in nitric acid solution before use and dry them in an oven. 3. Prepare five phosphate standard samples by diluting the phosphate stock solution into five samples containing 1–5 μg of KH2PO4 per tube. Classically, 5 μg of KH2PO4 give an absorbance of 0.9 at 800 nm. Store at 4°C. 4. Prepare the ascorbic acid solution (Subheading 2.3). 5. Collect three proteoliposomes volumes containing approximately 1–5 μg of lipids. Transfer the samples into clean glass tubes. 6. Add 0.65 mL of the perchloric acid solution and place the tubes in the heated block at 180°C for 30 min of digestion, until the yellow color has disappeared. 7. Add 0.65 mL of the perchloric acid solution per phosphate standard tube (digestion is not necessary). 8. Put all the tubes on ice, and set the heated block at 100°C (alternatively, a boiling water bath can be used). 9. Once cooled, add to the tubes the following: 3.3 mL of water, 0.5 mL of the ammonium molybdate solution, and 0.5 mL of the ascorbic acid solution. Agitate on a vortex after each addition. 10. Put the tubes in the heated block at 100°C for 5 min. 11. Put the tubes on ice for 5 min. 12. Read the absorbance of the cooled samples at 800 nm. 13. Plot a standard curve based on the absorbance of the phosphate standard samples. The exact concentration in the phospholipid samples is calculated based on the standard curve.

3.3.9. Proteoliposome Density Characterization

A discontinuous sucrose flotation gradient analysis (21, 29) can be performed to check that the membrane protein is correctly reconstituted into proteoliposomes. The proteoliposomes can be layered at the bottom (see below) or at the top (see Note 16) of the sucrose layers. 1. After MscL reconstitution, add 120 mg of sucrose to 0.2 mL of the proteoliposome suspension (containing approximately 96 μg of protein) and mix gently by pipetting, until the sucrose has completely dissolved. The final volume is approximately 0.265 mL and the final sucrose concentration is 45%. Adjust the final volume to 0.4 mL with the buffered 45% sucrose solution.

6

Cell-Free Membrane Protein Expression for Solid-State NMR

105

2. Transfer the resulting suspension to a 3-mL ultracentrifuge tube, and keep it vertical in an appropriate rack for the following steps. 3. Carefully deposit 0.7 mL of the buffered 20% sucrose solution, with minimal flow along the tube wall to avoid layers mixing. 4. Repeat with 0.7 mL of the buffered 10% sucrose solution. 5. Ultracentrifuge at 100,000 × g for 1.5 h in a swinging rotor at 18°C. 6. At the end of the centrifugation, carefully remove and transfer the tube into the rack, keeping it vertical. Carefully collect six fractions of 0.3 mL, from top to bottom, into 1.5-mL tubes, and adjust the final volume to 1 mL with water. 7. Wash the bottom of the 3 mL tube with 100 μL of 1× SDS loading buffer and remove a 15 μL aliquot for gel electrophoresis. 8. Centrifuge the 1.5-mL tubes at 16,000 × g for 15 min at room temperature, to pellet the proteoliposomes and remove the sucrose. 9. Resuspend each pellet in 100 μL of 1× SDS loading buffer and take a 15 μL aliquot of each for gel electrophoresis. 10. Analyze the aliquots by gel electrophoresis. For an expected ratio of protein to lipid of 1:4 (w/w), proteoliposomes appear at the interface between 20% and 10% sucrose. Protein-free liposomes lie at the top of the sucrose gradient, while lipid-free protein aggregates lie at the bottom of the sucrose gradient (see Note 17). 3.3.10. Water Quantification in the Sample

3.4. NMR Sample Preparation

Water content is estimated by weighing 10 mg of sample prior to and after 16 h in vacuum. 1. Weigh the empty rotor, cap, insert bottom, top, and screw (see Fig. 2). 2. Introduce the insert bottom into the rotor. 3. Introduce a couple of mg of the pelleted sample into the rotor on the tip of a spatula. Place the rotor into a 1.5-mL tube and centrifuge it at 10,000 × g for 1 min at room temperature. Repeat this until approximately 30 mg of sample has been introduced into the rotor. 4. Place the insert top into the rotor, without the top screw. Clean the upper part the rotor with a precision wiper before capping tightly. At this point, the rotor containing sample should be stored at −20°C until the NMR experiment is performed. 5. Introduce the rotor into the magic-angle spinning NMR probe and into the magnet of the NMR spectrometer. Spin the rotor at ~10 kHz for 15 min.

106

A. Abdine et al.

6. Extract the rotor and open it. Clean the upper part of the rotor again, in case a drop of water has come out. Place the top screw, tighten it, and then cap the rotor tightly. Weigh the full rotor to deduce the final sample mass. A typical sample consists of 3 mg proteins, 12 mg lipids and 15 mg water.

4. Notes 1. All solutions require Milli-Q water, but only solutions used inside the cell-free expression vessel need to be autoclaved and nuclease-free, to make sure that no ribonucleases are present that would damage the RNAs in the lysate. 2. Continuous-exchange cell-free protein expression with an RTS commercial kit from Roche/5Prime is advantageous for its convenience, reliability, for saving time and manpower in preparing the lysate, and for managing the stocks. The thermoregulated shaker, on the contrary, is not necessarily from Roche/5Prime. 3. Cell-free reactions should be performed on a small scale (RTS 100 or analog kits) for optimization studies, including detergents or lipids. Protein yield should be evaluated in a cell-free kit equipped with a continuous-exchange system (e.g., RTS 500), where the yield is usually higher. Once the protocol is optimized, the proteins can then be expressed on larger scales (e.g., RTS 9000). 4. Cell-free kits were stored at −80°C rather than −20°C, which increases the lifespan of the kits to over a year. Freeze–thaw cycles should be avoided. 5. Circular DNA was used for cell-free expression of MscL. Linear PCR templates can also be used, but they require larger amounts of DNA. DNA quality is important for cell-free expression. A plasmid purification procedure with a good yield (typically 100 μg of plasmid DNA per standard MIDI-prep), including an anion-exchange step, was found to be necessary. DNA should be dissolved in either nuclease-free water (our case) or Tris solution (typically 10 mM, at a pH between 8 and 8.5), but not in a buffer containing EDTA, as it would change the free magnesium ion concentration and reduce the protein expression yield. 6. Some amino acids are difficult to solubilize. Sonication or heating at 60°C can improve solubility. Met (M) and Cys (C) have to be supplemented with DTT (4 and 8 mM respectively). Tyr (Y), Trp (W), and Phe (F) have to be solubilized in 60 mM HEPES in nuclease-free water, adjusting the pH to 13, 1, or

6

Cell-Free Membrane Protein Expression for Solid-State NMR

107

7.5 respectively, using KOH or HCl. However, Trp (W), Asp (D), Asn (N), Cys (C), or Tyr (Y) will not dissolve completely and have to be used as suspensions. Alternatively, the Roche/5Prime RTS Amino acid Sampler provides appropriate stock solutions of each individual unlabeled amino acid. 7. Commercial mixtures of labeled amino-acids are also available (30). All labeled amino-acids can be specifically incorporated into the expressed proteins, except for Gln (Q) and Glu (E) when using the Roche/5Prime buffer, which contains large quantities of unlabeled glutamate. Some scrambling is observed with Ser (S), Asp (D), Asn (N), Gln (Q), and Glu (E). Some metabolic degradation can occur with Arg (R), Cys (C), Trp (W), Met (M), Asp (D), and Glu (E) during prolonged incubation. In such a case, an increased amount of amino acids can enhance the protein production. 8. Too vigorous mixing should be avoided while preparing the different lysates, as it may denature proteins or ribosomes in the extracts. 9. AEBSF is added after the expression is complete to prevent protease degradation of the target product. Proteases inhibitors can also be added to the reaction chamber, but they should be tested on a small scale beforehand. 10. If the chelating column is not precharged with Ni ions, before connecting it to the FPLC system, open it at both ends and wash it with 2 CV of distilled water, 1 CV of 100 mM EDTA, pH 8, 3 CV of distilled water, 2 CV of 0.1 M NiSO4 solution, and 5 CV of distilled water. Once the column is charged, an additional washing step with 5 CV of 20% ethanol is necessary before storage at 4°C. At the end of the protein purification run, wash the column with water (5 CV), 0.5 M NaOH (5 CV), water (5 CV) and store in 20% ethanol. 11. The total reaction run-on is expected to contain approximately 10 mg of protein but also detergent and impurities, which may bind to the column as well. Since the binding saturation of the column surface is on the order of 10 mg/mL, the reaction run-on is split in two so as to always remain below the saturation limit. 12. For MscL, the fractions containing the protein are eluted between 200 and 500 mM of imidazole, but this is highly protein dependent. Depending on the histidine-tag accessibility, the protein may elute at lower or higher imidazole concentrations. 13. Protein concentration is usually estimated using UV fluorescence of aromatic amino acids. If the protein is not sufficiently fluorescent, quantification can be performed using the commercial Micro BCA bicinchonic acid protein assay kit (Pierce). Care should be taken if lipids are present in the sample.

108

A. Abdine et al.

14. Instead of using polystyrene beads, detergent extraction can be performed using cyclodextrin inclusion compounds, as described by DeGrip et al. (24), or by dialysis (21). 15. Proteoliposomes can be dehydrated and rehydrated, but care should be taken, since the process may affect protein activity. 16. The discontinuous sucrose flotation gradient analysis to check that the membrane protein is correctly reconstituted can be performed with the proteoliposomes layered either at the top or at the bottom of the sucrose layers, or both (21, 29). If at the top, the proteoliposomes must be mixed with sucrose for a final sucrose concentration of 10%. The rest of the protocol is the same: After layering 0.7 mL of 45% sucrose, then 0.7 mL of 20% sucrose and then 0.7 mL of the proteoliposomes in 10% sucrose, the proteoliposomes usually appear at the interface between 20 and 10% sucrose. 17. If the amount of lipid-free protein aggregates is not negligible, the discontinuous sucrose flotation gradient can also be used as a purification technique by layering the entire reconstituted proteoliposome suspension.

Acknowledgments This work was supported by fellowships from the Ministère de l’Enseignement Supérieur et de la Recherche and the Fondation pour la Recherche Médicale (to A.A.), by the CNRS (UMR 7099 and 8619), the ANR (ANR-06-JCJC0014), the Univ Paris Diderot and the Université Paris-Sud 11. We thank Alexandre Ghazi, Catherine Berrier, Emmanuelle Billon-Denis, and Michiel A. Verhoeven for helping us optimize the protocols presented here. References 1. Lyford, L. K., and Rosenberg, R. L. (1999) Cell-free expression and functional reconstitution of Homo-oligomeric α7 Nicotinic Acetylcholine Receptors into Planar Lipid Bilayers. J. Biol. Chem. 274, 25675–25681. 2. Elbaz, Y., Steiner-Mordoch, S., Danieli, T., and Schuldiner, S. (2004) In vitro synthesis of fully functional EmrE, a multidrug transporter, and study of its oligomeric state. Proc. Natl. Acad. Sci. U.S.A. 101, 1519–1524. 3. Berrier, C., Park, K. H., Abes, S., Bibonne, A., Betton, J. M., and Ghazi, A. (2004) Cell-free synthesis of a functional ion channel in the absence of a membrane and in the presence of detergent. Biochemistry 43, 12585–12591.

4. Klammt, C., Lohr, F., Schafer, B., Haase, W., Dötsch, V., Ruterjans, H., Glaubitz, C., and Bernhard, F. (2004) High level cell-free expression and specific labeling of integral membrane proteins. Eur. J. Biochem. 271, 568–580. 5. Wagner, S., Bader, M. L., Drew, D., and de Gier, J. W. (2006) Rationalizing membrane protein overexpression. Trends Biotechnol. 24, 364–371. 6. Eshaghi, S. (2009) High-throughput expression and detergent screening of integral membrane proteins. Methods Mol. Biol. 498, 265–271. 7. Tate, C. G. (2010) Practical considerations of membrane protein instability during purification and crystallisation. Methods Mol. Biol. 601, 187–203.

6

Cell-Free Membrane Protein Expression for Solid-State NMR

8. Breyton, C., Pucci, B., and Popot, J.-L. (2010) Amphipols and fluorinated surfactants: Two alternatives to detergents for studying membrane proteins in vitro. Methods Mol. Biol. 601, 219–245. 9. White, S. H. (2010) Membrane Proteins of Known Structure. University of California at Irvine. http://blanco.biomol.uci.edu/ Membrane_Proteins_xtal.html Accessed 25 July 2011. 10. Warschawski, D. E. (2010) Membrane Proteins of Known Structure Determined by NMR. Drorlist. http://www.drorlist.com/nmr/MPNMR.html Accessed 25 July 2011. 11. Schneider, B., Junge, F., Shirokov, V. A., Durst, F., Schwarz, D., Dötsch, V., and Bernhard, F. (2010) Membrane protein expression in cellfree systems. Methods Mol. Biol. 601, 165–186. 12. Sobhanifar, S., Reckel, S., Junge, F., Schwarz, D., Kai, L., Karbyshev, M., Löhr, F., Bernhard, F., and Dötsch, V. (2010) Cell-free expression and stable isotope labelling strategies for membrane proteins. J. Biomol. NMR 46, 33–43. 13. Lehner, I., Basting, D., Meyer, B., Haase, W., Manolikas, T., Kaiser, C., Karas, M., and Glaubitz, C. (2008) The key residue for substrate transport (Glu14) in the EmrE dimer is asymmetric. J. Biol. Chem. 283, 3281–3288. 14. Abdine, A., Verhoeven, M. A., Park, K.-H., Ghazi, A., Guittet, E., Berrier, C., Van Heijenoort, C., and Warschawski, D. E. (2010) Structural study of the membrane protein MscL using cell-free expression and solid-state NMR. J. Magn. Reson. 204, 155–159. 15. Kalmbach, R., Chizhov, I., Schumacher, M. C., Friedrich, T., Bamberg, E., and Engelhard, M. (2007) Functional cell-free synthesis of a seven helix membrane protein: in situ insertion of bacteriorhodopsin into liposomes. J. Mol. Biol. 371, 639–648. 16. Marques, B., Liguori, L., Paclet, M. H., VillegasMendéz, A., Rothe, R., Morel, F., Lenormand, J.-L. (2007) Liposome-mediated cellular delivery of active gp91(phox). PLoS One 2, e856. 17. Katzen, F., Fletcher, J. E., Yang, J. P., Kang, D., Peterson, T. C., Cappuccio, J. A., Blanchette, C. D., Sulchek, T., Chromy, B. A., Hoeprich, P. D., Coleman, M. A., and Kudlicki, W. (2008) Insertion of membrane proteins into discoidal membranes using a cell-free protein expression approach. J. Proteome Res. 7, 3535–3542. 18. Berrier, C., Guilvout, I., Bayan, N., Park, K.-H., Mesneau, A., Chami, M., Pugsley, A. P., and Ghazi, A. (2011) Coupled cell-free synthesis and lipid vesicle insertion of a functional oligomeric channel MscL - MscL does not need the insertase YidC for insertion in vitro. Biochim. Biophys. Acta 1808, 41–46.

109

19. Abdine, A., Verhoeven, M. A., and Warschawski, D. E. (2011) Cell-free expression and labeling strategies for a new decade in solid-state NMR. New Biotechnol. 28, 272–276. 20. He, M. (2008) Cell-free protein synthesis: applications in proteomics and biotechnology. New Biotechnol. 25, 126–132. 21. Rigaud, J.-L., and Lévy, D. (2003) Reconstitution of Membrane Proteins into Liposomes. Methods Enzymol. 372, 65–86. 22. Triba, M. N., Zoonens, M., Popot, J.-L., Devaux, P. F., and Warschawski, D. E. (2006) Reconstitution and alignment by a magnetic field of a β-barrel membrane protein in bicelles. Eur. Biophys. J. 35, 268–275. 23. Leitz, A. J., Bayburt, T. H., Barnakov, A. N., Springer, B. A., and Sligar, S. G. (2006) Functional reconstitution of β2-adrenergic receptors utilizing self-assembling Nanodisc technology. BioTechniques 40, 601–612. 24. De Grip, W. J., Van Oostrum, J., and BoveeGeurts, P. H. M. (1998) Selective detergentextraction from mixed detergent/lipid/protein micelles, using cyclodextrin inclusion compounds: a novel generic approach for the preparation of proteoliposomes. Biochem. J. 330, 667–674. 25. Sukharev, S. I., Schroeder, M. J., and McCaslin, D. R. (1999) Stoichiometry of the Large Conductance Bacterial Mechanosensitive Channel of E. coli. A Biochemical Study. J. Membrane Biol. 171, 183–193. 26. Smith, P. K., Krohn, R. I., Hermanson, G. T., Mallia, A. K., Gartner, F. H., Provenzano, M. D., Fujimoto, E. K., Goeke, N. M., Olson, B. J., and Klenk, D. C. (1985) Measurement of protein using bicinchoninic acid. Anal. Biochem. 150, 76–85. 27. Wiechelman, K. J., Braun, R. D., and Fitzpatrick, J. D. (1988) Investigation of the bicinchoninic acid protein assay: identification of the groups responsible for color formation. Anal. Biochem. 175, 231–237. 28. Rouser, G., Fkeischer, S., and Yamamoto, A. (1970) Two dimensional thin layer chromatographic separation of polar lipids and determination of phospholipids by phosphorus analysis of spots. Lipids 5, 494–496. 29. Laird, D. M., Eble, K. S., and Cunningham, C. C. (1986) Reconstitution of mitochondrial F0.F1-ATPase with phosphatidylcholine using the nonionic detergent, octylglucoside. J. Biol. Chem. 261, 14844–14850. 30. Etezady-Esfarjani, T., Hiller, S., Villalba, C., and Wüthrich, K. (2007) Cell-free protein synthesis of perdeuterated proteins for NMR studies. J. Biomol. NMR 39, 229–238.

Chapter 7 Expression and Purification of Src-family Kinases for Solution NMR Studies Andrea Piserchio, David Cowburn, and Ranajeet Ghose Abstract NMR analyses of the structure, dynamics, and interactions of the Src family kinases (SFKs) have been hindered by the limited ability to obtain sufficient amounts of properly folded, soluble protein from bacterial expression systems, to allow these studies to be performed in an economically viable manner. In this chapter, we detail our attempts to overcome these difficulties using the catalytic domain (SrcCD) of c-Src, the prototypical SFK, as an illustrative example. We describe in detail two general methods to express and purify SrcCD from Escherichia coli expression systems in both fully active wild-type and kinase-deficient mutant forms, allowing the efficient and cost-effective labeling by NMR-active isotopes for solution NMR studies. Key words: Protein tyrosine kinases, Src-family kinases, Escherichia coli expression systems

1. Introduction Protein kinases in higher eukaryotes transfer the γ-phosphate of ATP to the hydroxyl groups of specific serine, threonine (serine/ threonine kinases) or tyrosine (tyrosine kinases) residues in substrate proteins. This covalent modification of the Ser/Thr/ Tyr–OH moiety is a fundamental mechanism of cellular regulation and intracellular signal transduction (1–3) and has a significant role in almost every aspect of cell growth, differentiation, maturation, motility, and regulated cell death. In vivo, the phosphorylation levels of kinase targets are tightly controlled by the opposing actions of kinases and their corresponding phosphatases (4, 5). A loss of this control due to mutations, changes in expression levels, alteration of protein–protein interactions including localization, and/or catalytic activity has been implicated in a variety of diseases

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_7, © Springer Science+Business Media, LLC 2012

111

112

A. Piserchio et al.

including a multitude of human cancers (6–9), making protein kinases significant and established targets for anticancer therapies (10–13). Eukaryotic protein kinases contain a highly conserved catalytic domain (CD) (14, 15) that forms a dual-lobed structure (Fig. 1) with a smaller, mainly β-sheet N-terminal lobe (N-lobe), and a larger, mainly α-helical C-terminal lobe (C-lobe). The catalytic activity, while wholly contained within the CD, is often regulated by additional modular domains (16) or insertions within the CD itself (17). While crystallographic analyses of both serine/threonine (18) and tyrosine kinases (19) have provided valuable insights into the regulation of their catalytic activity, several questions remain. The current hypothesis is that most eukaryotic protein kinases function as simple two-state switches transitioning between well-defined “active” and “inactive” conformations based on the phosphorylation state of a small number of regulatory residues. Mounting evidence suggests that this may be an oversimplification and the boundary between fully active and fully inactive states may be not that well defined. Further, the simple phosphorylation-dependent switch scenario is complicated by mounting experimental evidence that hints towards the role played by longrange allosteric dynamic pathways in the regulation of kinase activity, where active-like conformations may be attained through remote interactions (20). Structural details on the nature of recognition of downstream substrates and of regulating kinases and phosphatases are still poorly defined for many of the subfamilies of

Fig. 1. (a) Domain arrangement in inactive full-length c-Src (PDB ID: 2SRC). The residues comprising the linker between the SH2 and the catalytic domain (CD), which engages the SH3 domain, is shown in ball-and-stick representation, as are the residues that comprise the C-terminal tail. The regulatory tyrosine residues in the activation loop (Y416, unphosphorylated in inactive c-Src) and the C-terminal tail (Y527, phosphorylated in inactive c-Src) are also shown in ball-and-stick representation. The ATP-analog, ANP (phosphoaminophosphonic acid-adenylate), bound to the ATP-binding site, is also shown in ball-and-stick representation. (b) Expanded view of SrcCD with the N-lobe and C-lobes indicated. Important regulatory elements that include the activation loop, Gly-rich loop, the HRD (that precedes the catalytic loop) and DFG motifs are also indicated. Y416 is shown in a ball-and-stick representation and the αC helix is indicated by a dashed ellipse. The ligand ANP is also shown bound to the ATP-binding pocket.

7

Expression and Purification of Src-family Kinases for Solution NMR Studies

113

the human kinome. This is not unexpected, given that some of these interactions may be weak or transient, and the resultant complexes may not be amenable to crystallographic analyses. Given this background, it would seem that solution NMR spectroscopy is tailor-made to resolve these unanswered issues about structure, dynamics, and interactions involving protein kinases. Targeting eukaryotic protein kinases by using solution NMR methodologies is especially important because of considerable interest in the pharmaceutical industry in designing small molecule inhibitors of enzymatic activity (21). In particular, the rational design of inhibitors targeting remote sites exerting allosteric control over catalytic activity in protein kinases (type III inhibitors) holds promise as a strategy for inhibiting kinase signaling in a highly selective fashion. These allosteric sites, which are more likely to be unique to specific protein kinases or kinase subfamilies, could be identified by NMR studies performed on a representative subgroup of the human kinome. These studies are expected to assume greater importance as the role of yet uncharacterized protein kinases in cancer becomes clearer by using a combination of RNAi technologies in tumor cell lines and genetic knockdown studies (22–25). Unfortunately, there have been only a handful of NMR studies reported on protein kinases. These include the receptor tyrosine kinase Eph (26), c-Abl (27), c-Src (28), ERK2 (29) protein kinase A (PKA) (30, 31), and Csk (32). A large number of these targets, including PKA and Eph, for which detailed NMR analyses are available, may be expressed and purified with relative ease and are, to a large extent, well behaved in solution, in contrast to a large majority of the human kinome. NMR studies on a broader class of protein kinases have been hindered by difficulties in bacterial expression and purification of sufficient quantities of soluble, properly folded protein for economically viable labeling with NMRactive isotopes. Indeed, the only comprehensive NMR study on a nonreceptor tyrosine kinase examined c-Abl expressed using the baculovirus Sf9 insect cell expression system (27). This system allows limited labeling options and is generally not an economically viable option in a noncommercial setting. Thus, to make these studies viable for a broader set of representative members of the kinome, the optimization of protocols for protein production in bacterial expression systems is required. The problems encountered in NMR studies of eukaryotic protein kinases are exemplified by the Src-family of nonreceptor tyrosine kinases (SFKs) (Fig. 1). c-Src is the prototypical representative of this nine member family (c-Src, Blk, Fgr, Fyn, Hck, Lck, Lyn, Yes, and Yrk) (16, 33, 34). SFKs have been found to be involved in a multitude of human cancers including those of the breast, colon, gastrointestinal tract, lung and in leukemias, lymphomas, and myelomas (35). In particular, the catalytic activity of

114

A. Piserchio et al.

c-Src is highly elevated in a majority of human colon cancers (36). SFKs in general and c-Src in particular are therefore key targets for the design of drugs to treat a wide range of human cancers (37). The catalytic activity of c-Src, and indeed of other SFKs, depends on the phosphorylation state of two conserved tyrosine residues (chicken c-Src numbering used throughout): Tyr416 (activation) and Tyr527 (suppression). Phosphorylation occurs at Tyr416 through an intermolecular mechanism, while Tyr527 is phosphorylated by another family of nonreceptor tyrosine kinases that is comprised of two members: C-terminal Src kinase (Csk) and Cskhomologous kinase (Chk) (38, 39). Several researchers have utilized a variety of systems, including insect cells (40), yeast (Schizosaccharomyces pombe) (41), and platelets (42) to express c-Src. Weijland et al. (41) used a S. pombe based strategy to express the catalytic domain of c-Src by coexpressing the catalytic domain of the tyrosine phosphatase PTP-PEST to reduce catalytic activity and the resultant cytotoxicity. As an extension of this strategy, Seeliger et al. (43) have reported the Escherichia coli expression of the catalytic domains of c-Src and c-Abl by coexpressing the downregulating YopH phosphatase (43). However, multiple issues, including poor sample homogeneity, multiple phosphorylation states (44), sample aggregation or degradation are often encountered when using these strategies. Escherichia coli, the system of choice for production of material amenable to analysis by solution NMR spectroscopy, is particularly affected by these drawbacks (28). The expression and purification of the catalytic domain of c-Src (SrcCD) from E. coli for NMR studies was hindered by several major complications that include the following: 1. Cytotoxicity resulting in poor cell growth after induction. 2. Protein misfolding leading to the accumulation of large amounts of insoluble material in inclusion bodies resulting in vanishing low yields of properly folded soluble protein. Attempts to refold these misfolded aggregates using chaperones in vitro have also not yielded promising results (Piserchio et al. unpubl. results). 3. The production of heterogeneously phosphorylated protein. Resolution of these issues is absolutely necessary for NMR, with its ability to provide both structural and dynamic information over a wide range of timescales, to play a role in studies of these key signaling molecules for drug discovery (45). Here, we describe our efforts toward alleviating these issues by optimal choice of constructs, the use of solubility-enhancing tags, coexpression with chaperones, and carefully optimizing overexpression conditions (28). These methods should provide a useful guide for the bacterial expression and purification of similar protein kinases for NMR studies.

7

Expression and Purification of Src-family Kinases for Solution NMR Studies

2. Expression of the Catalytic Domain of c-Src in Escherichia coli

115

The first strategy that we utilized for the bacterial expression of c-Src involved (1) the addition of an N-terminal MBP fusion tag to enhance both solubility and folding, (2) the coexpression of GroEL and GroES chaperones to further facilitate the accumulation of properly folded protein, an idea originally proposed by Cole (46), and (3) a decrease in the growth temperature after induction to 15°C, thus reducing the rapid accumulation of large quantities of proteins and macromolecular crowding effects that tend to facilitate aggregation. MBP is widely used to enhance the production of fused proteins through a mechanism that is far from clear. Several possible mechanisms have been proposed including the formation of large vesicular aggregates with the misfolded or unfolded proteins contained inside (47) or a general chaperoning effect (48). Our initial attempts to produce the catalytic domain of the so-called “kinasedead mutant” bearing a Lys → Met mutation (SrcCDK295M) (49) in M9 minimal medium showed that in fact, just the introduction of a N-terminal MBP fusion, while not eliminating the presence of protein in the insoluble inclusion bodies, drastically increased the amount of soluble material. However, a majority of the protein produced in this fashion found in solution consisted of soluble aggregates, presumably formed by misfolded material. Observation of similar phenomena has added credence to the speculation that MBP fusions can form micellar-like systems, formed by unfolded material in the core, and by the soluble MBP domain on the surface (47). Though a relatively small amount of properly folded SrcCDK295M was obtained by these preparations, it allowed us to collect the first 15N-HSQC spectra of SrcCDK295M. However, the quantity of soluble, properly folded SrcCD obtained using this procedure was not sufficient for detailed NMR analyses. This was because a large number of experiments, collected over long periods of time, are required for the resonance assignment process, necessitating multiple NMR samples produced using expensive labeling supplies. Nevertheless, the observation that material suitable for NMR experiments could be obtained from an E. coli expression system was indeed encouraging. The largest improvement in the yield of properly folded SrcCDK295M came when the N-terminal MBP fusion system was transformed into E. coli cells that overexpress GroEL and GroES chaperones. Cole and coworkers have shown that overexpression of these bacterial chaperones allow the purification of small amounts of SrcCDK295M from E. coli (49). We found that a more substantial improvement, also achievable in the minimal medium used for NMR sample preparation, can be obtained by integrating the MBP fusion and the GroELS systems. Furthermore, the same approach

116

A. Piserchio et al.

Fig. 2. Production and purification of SrcCDK295M. (a) SDS-PAGE gel (20%) showing different steps in the purification process: lane 1 – GroEL and GroES (expressed alone), lane 2 – cell lysate supernatant, lane 3 – cell lysate pellet, lane 4 – metal affinity column flow through, lane 5 – metal affinity column eluate, lane 6 – product after thrombin cleavage (MBP – higher molecular weight and SrcCDK295M – lower molecular weight), lane 7 – SrcCDK295M after Q-column purification. (b) Effects of controlled phosphorylation and dephosphorylation of SrcCDK295M on the Q column elution profiles. Dashed line: sample preincubated with catalytic amounts of commercial wild-type c-Src and Mg+2/ATP. Dark solid line: sample preincubated with alkaline phosphatase (nonspecific phosphatase). Light solid line: sample preincubated with Lyp phosphatase (specific for phosphorylated Y416). The identity of pSrcCDK295M was confirmed by immunoblots using anti-pTyr antibodies and MS/MS, following trypsin digestion.

could also be used to grow wild-type SrcCD (as opposed to SrcCDK295M), albeit with lower yields, with minimal alteration of the overall protocol. The introduction of the chaperones, however, did not completely eliminate the formation of soluble aggregates mentioned above. However, we noticed that, while these undesired aggregates bound to the amylose columns that are routinely employed to purify MBP fusion proteins, they were unable to bind Co+2 affinity columns despite the presence of an N-terminal His6tag, most likely because the His6-tag was occluded in the misfolded species. This fortuitous scenario allowed us to use metal affinity chromatography to efficiently separate the soluble aggregates from properly folded SrcCD. When a target protein is expressed as an MBP fusion, a critical step prior to NMR analyses is the removal of the bulky MBP tag from the fusion. SrcCD can be efficiently separated from the MBP tag by inserting a thrombin cleavage site between the two components of the fusion construct, although some care has to be taken to avoid protein degradation in the presence of thrombin. We found that resins that bind the N-terminal end of the cleaved fusion (metal-chelating for His6 or amylose for the MBP tag) cannot be efficiently used to isolate SrcCD, given the tendency of the kinase to bind nonspecifically to these columns. We then found that separation of MBP and SrcCD can be carried out using ion exchange chromatography (Fig. 2). The catalytically compromised SrcCDK295M could be efficiently separated from cleaved MBP in this fashion. Interestingly, during purification we found that two forms of SrcCDK295M (one selectively phosphorylated at the activating

7

Expression and Purification of Src-family Kinases for Solution NMR Studies

117

Tyr416 and the other not; confirmed by Western blot against anti-pTyr antibodies and MS/MS analysis) (28) were obtained as two distinct peaks in the elution chromatogram (Fig. 2). This last observation was surprising to some extent, because Q columns do not usually have sufficient resolution to separate identical proteins with unit changes in charge states. Crystallographic and computational analyses (e.g., refs. 50, 51) have revealed that activating phosphorylation alters the overall conformation of the catalytic domain, thus the protein surface that interacts with the column is also likely altered. This feature, we believe, allows the separation of the phosphorylated and nonphosphorylated species using ion-exchange chromatography. We provide below a step-by-step protocol for the purification of SrcCD as an MBP fusion in Subheading 4.1. The MBP/GroELS based protocol, though complex in the nature of its purification scheme, results in high yields of SrcCD purified to homogeneity (up to 700 nmol/L of culture). However, the biggest disadvantage lies in the fact that the catalytic domain of c-Src can only be purified either as a kinase-deficient mutant (SrcCDK295M) or in wild-type form in complex with a high-affinity inhibitor (see Subheading 4.1.2), using this strategy. This prevents access to the full range of catalytically relevant conformations and motional modes, in spite of having the ability to purify both constructs (mutant and wild-type) with the activating Tyr416 in the unphosphorylated and selectively phosphorylated forms. Thus, we developed an alternative strategy to purify wild-type SrcCD that displays full catalytic activity in solution. Our experiences seem to indicate that proper folding appears to be the main hindrance to SrcCD production in bacterial expression systems. In eukaryotes, the folding of several protein kinases including c-Src, occurs with the assistance of the Hsp90 chaperone machinery (52–54). This process is further assisted by an essential kinasespecific cochaperone Cdc37/p50, that can independently bind both the kinase target and the Hsp90 chaperone, thus approximating the kinase and the chaperone machinery, and stabilizing the transient ternary complex (55, 56). Indeed, ternary complexes of Hsp90, v-Src (a transforming viral form of c-Src that contains a tail truncation and lacks Tyr527) and Cdc37 (57) have been identified. We found that coexpressing the Cdc37 cochaperone (the plasmid encoding Cdc37 was a kind gift from Dr. Avrom Caplan, CCNY) with wild-type SrcCD leads to a properly folded soluble kinase domain even in the absence of the N-terminal MBP fusion protein (Fig. 3). Consequently, the His6-tagged kinase can be purified easily using metal affinity chromatography followed by size exclusion chromatography. Thrombin cleavage and ion-exchange chromatography, steps that can both lead to substantial losses in material, become unnecessary. Although the yields are somewhat lower, the methodology is less time consuming, and the wild-type kinase domain can be isolated without the use of an inhibitor, allowing

118

A. Piserchio et al.

Fig. 3. (a) SDS-PAGE gel (20%) showing the various steps in the purification of wild-type SrcCD (His6-tagged): lane 1 – cell lysate supernatant, lane 2 – cell lysate pellet, lane 3 – metal affinity column flow-through (unbound material), lane 4 – metal affinity column eluate, lane 5 – cSrc after purification using gel filtration. The positions of Cdc37, MBP-Lyp and SrcCD in lanes 3–5, respectively, are highlighted by boxes. Samples in lanes 4 and 5 were concentrated by a factor of 10. (b) Size exclusion column profile of the eluate from the metal affinity column purification. (c) Immunoblots of phosphorylated wild-type SrcCD (preincubated with 2 mM ATP and 10 mM Mg+2) with anti-pTyr antibody. All immunoblots were visualized using the Bio-Rad ChemiDoc system. Lane 1 – phosphorylated SrcCD, lane 2 – phosphorylated SrcCD after incubation with alkaline phosphatase.

access to the full complement of structural and dynamics changes that accompany activation. The detailed protocols used are described at length in Subheading 4.2.

3. Materials Most of the materials utilized in the experiments described below are routinely used in NMR laboratories for protein expression and purification. 3.1. Expression Vectors and Competent Cells

1. H-MBP-3C vector (58): a kind gift from Dr. Kaushik Dutta, New York Structural Biology Center (see Note 1). 2. pACYCDuet vector: for simultaneous expression of two proteins (see Note 2): SrcCD and Cdc37 should be cloned into this vector following the protocol described below. 3. 0.1-cm gap sterile electroporation cuvette for bacteria. 4. Electroporator (see Note 3). 5. pREP4-groELS (see Note 4).

electrocompetent

BL21

(DE3)

cells

6. BL21 (DE3) T1 competent cells. 3.2. Media for Bacterial Growth

1. LB (Luria Bertani) Miller medium: Dissolve 20 g of LB powder in 1 L of water, and sterilized by autoclaving. Store at room temperature.

7

Expression and Purification of Src-family Kinases for Solution NMR Studies

119

2. M9 salts: 47.9 mM anhydrous Na2HPO4 (6.8 g/L), 22.0 mM anhydrous KH2PO4 (3 g/L), and 8.6 mM NaCl (0.5 g/L), sterilized by autoclaving. Store at room temperature. 3. Ampicillin stock (1,000×): 100 mg/mL Ampicillin in water. Store at 4°C. 4. Kanamycin stock (1,000×): 50 mg/mL Kanamycin in water. Store at 4°C. 5. Chloramphenicol stock (1,000×): 35 mg/mL Chloramphenicol in Isopropanol. Store at 4°C. 6. LB plates: Mix 40 g/L of LB Miller agar and autoclave the suspension, let the material cool to ~55°C and add the desired antibiotic (100 mg/L Ampicillin, or 50 mg/L Kanamycin, or 35 mg/L Chloramphenicol) or combination of antibiotics (100 mg/L Ampicillin and 50 mg/L Kanamycin, or 100 mg/L Ampicillin and 35 mg/L Chloramphenicol). 7. 1M Isopropyl β-D-1-thiogalactopyranoside (IPTG): Prepare in water and store at 4°C. 8. Metal Solution I (1,000×): 1 mM MnCl2·4H2O, 0.2 mM CoCl2·6H2O, 2 mM ZnCl2, 1 mM CuSO4·5H2O, 50 mM H3BO4, and 0.001 mM Na2MoO4. The compounds are added in the order listed above, dissolving each compound completely before adding the following one. 9. Metal Solution II (1,000×): 1 mM FeSO4, filter-sterilize and store at 4°C. 10. MgSO4 stock (500×): 1M MgSO4, autoclave and store at room temperature. 11. CaCl2 stock (1,000×): 0.1M CaCl2, autoclave and store at room temperature. 12. Vitamin solution (100×): Prepare 10 mL aliquots in 15-mL centrifuge tubes, and store at −20°C (see Note 5). 13.

15

NH4Cl: The material is weighed and carefully dissolved in the minimum amount of preparation medium (40 mL for 0.5 g of NH4Cl), sterile-filtered, and added immediately to the medium.

14. [13C6]-Glucose: The material is weighed and carefully dissolved in the minimum amount of preparation medium (40 mL for 2.5 g of glucose), sterile-filtered, and added immediately to the medium. 15. [U-15N,13C,2H] CELTONE® powder (see Note 6). The material is weighed and carefully dissolved in the minimum amount of preparation medium, sterile-filtered, and added immediately to the medium. Store at room temperature. 16. Anhydrous dextrose.

120

A. Piserchio et al.

17.

14

N ammonium chloride.

18. 0.22-μm Filters. 3.3. Protein Purification

1. Human α-thrombin, store at −80°C. 2. Empty 20-mL bed volume chromatography columns. 3. Preequilibrated cobalt resin: Resuspend the cobalt-based beads, (see Note 7) used for metal-affinity purification of His6-tagged proteins, by gently inverting the bottle. Typically 4 mL of bed volume is used for every 0.5 L of culture (2 mL of slurry provides 1 mL of bed volume). Once resuspended, pipette the slurry into a 20 mL empty column until all the methanol is washed away and the resin settles. Gently wash the column with three bed volumes of deionized H2O followed by three bed volumes of lysis buffer. Resuspend the resin in 1–2 mL of lysis buffer (by gently inverting the closed column) and add to the cell lysate. 4. Lysis buffer: 20 mM Tris–HCl, pH 7.8, 0.5M NaCl, 1% Triton X-100, 0.1% β-mercaptoethanol. The buffer is prepared without β-mercaptoethanol and Triton-X, and stored at room temperature with these two chemicals added immediately before usage. 5. Q columns (5 mL): for ion-exchange chromatography (see Note 8). 6. Ion-exchange chromatography buffer A: 20 mM Tris–HCl, pH 7.4, 10% glycerol. Store at room temperature. 7. Ion-exchange chromatography buffer B: 20 mM Tris–HCl, pH 7.4, 1M NaCl, 10% glycerol. Store at room temperature. 8. Dialysis buffer (before size-exclusion chromatography): 20 mM Tris, pH 7.5 (titrated with 12M HCl), 150 mM NaCl, 1 mM EDTA, and 1 mM DTT (Dithiothreitol), prepared immediately before use. 9. Size exclusion chromatography buffer: 50 mM sodium phosphate, pH 6.5, 150 mM NaCl, 1 mM DTT. 4.82 g of NaH2PO4, and 2.14 g of Na2HPO4 are added to 1 L of water, the pH is adjusted if necessary using 12M HCl or 1M NaOH, The DTT is added immediately before use and the rest stored at room temperature. 10. Gel filtration column (see Note 9): Preequilibrate the column with 1.5 volumes of size exclusion chromatography buffer. The typical flow rate is 0.5 mL/min. 11. AP23464: 10 mM in DMSO. The high-affinity kinase inhibitor (59) is a kind gift from Dr. David Dalgarno, Ariad Pharmaceuticals. Dissolve 4.7 mg in 1 mL of dimethyl sulofoxide. 12. Imidazole buffer: 20 mM Tris, 0.5M NaCl, 1M Imidazole. Dissolve 1.2 g Tris, 14.61 g NaCl, and 34.04 g of Imidazole in ~ 450 mL H2O. Adjust the pH to 7.8 using 12M HCl, bring the solution to 0.5 L volume using a graduated cylinder, and

7

Expression and Purification of Src-family Kinases for Solution NMR Studies

121

readjust the pH to 7.8 using 12M HCl or 1M NaOH. Cover the bottle with aluminum foil and store at room temperature. 13. Cleavage buffer: 20 mM Tris–HCl, pH 7.8, 150 mM NaCl. Add β-mercaptoethanol to a final concentration of 4 mM just before the cleavage reaction. 14. 1M sodium hydroxide: dissolve 40 g of NaOH in ~ 950 mL of water and bring to 1 L volume using a graduated cylinder. 15. EDTA solution: 20 mM Tris, 0.5M EDTA. Dissolve 1.2 g of Tris and 73 g of ethylenediaminetetraacetic acid in 450 mL of water, adjust the pH to 8.0 using 1M NaOH, bring to volume with a graduated cylinder, and adjust the pH to 8.0 using 12M HCl or 1M NaOH. 16. AEBSF: 41.7 mM solution, 1 mL aliquots. Dissolve 10 mg of 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride in 1 mL of water and store at −20°C. 17. French press. 18. 12M HCl.

4. Methods 4.1. Expression and Purification of the Catalytic Domain of c-Src as an N-terminal MBP Fusion 4.1.1. Overexpression of SrcCD

The detailed protocol provided below is for the bacterial expression and purification of the catalytic domain of the kinase deficient construct (SrcCDK295M) or the wild-type domain in the presence of a high-affinity inhibitor.

1. Thaw a tube containing the pREP4-groELS electrocompetent BL21 (DE3) cells, on ice (see Note 4). 2. Add ~20 ng of H-MBP-3C vector, encoding SrcCD, to the electrocompetent cells. Transfer the contents into a prechilled 0.1 cm gap electroporation cuvette and incubate for 2 min on ice. 3. Transform the cells with a 6 ms pulse at 1,730 V, and add 960 μL of LB medium, prewarmed to 37°C. Transfer into a sterile 15-mL conical centrifuge tube and shake at 250 rpm for 1 h in an incubator set to 37°C. 4. Plate ~100 μL of transformed cells on a prewarmed (37°C) LB plate containing both Kanamycin and Ampicillin (50 and 100 mg/L, respectively). Incubate overnight at 37°C. 5. Transfer the plate to 4°C for storage. 6. Autoclave 0.5 L of M9 salts in a 2-L flask, transfer 20 mL of this solution into a 50-mL conical tube. From the stock solutions, add 20 μL of 0.1M CaCl2 (final concentration: 0.1 mM),

122

A. Piserchio et al.

40 μL of 1M MgSO4 (final concentration: 2.0 mM), 20 μL of Metal Solutions I and II, 200 μL of Vitamins solution, 20 μL each of Ampicillin and Kanamycin (final concentration: 100 and 50 mg/L, respectively). Finally, add 0.1 g (5 g/L) of anhydrous dextrose and 20 mg (1 g/L) of ammonium chloride. Dissolve by agitation or vortexing, and filter the M9 medium into a preautoclaved 125-mL flask using a sterile 50-mL syringe and a sterile 0.22-μm filter. 7. Scoop a few colonies from the transformed plate using a sterile loop, transfer into the M9 medium and incubate overnight in a shaker (250 rpm) at 37°C. 8. Using a sterile culture pipette, transfer ~20 mL of the M9 salt solution into a sterile 50-mL centrifuge tube. Use this solution to dissolve 2.5 g of dextrose and 0.5 g of ammonium chloride (final concentration: 5 and 1 g/L, respectively). Then, add 5 mL of vitamin solution, dissolve by shaking or vortexing, and filter the entire contents into the original 2-L flask. Add 500 μL of 0.1M CaCl2 1 mL of 1M MgCl2 stock solutions (final concentration: 0.1 and 2.0 mM, respectively), 500 μL each of Metal Solutions I and II, 500 μL of Ampicillin and Kanamycin (final concentration: 100 and 50 mg/L, respectively) to the flask. 9. Add the 20 mL overnight culture directly into the 0.5 L of M9 medium. Incubate at 250 rpm at 37°C. 10. After 1 h, check the OD600 every 30 min; when the OD600 reaches ~0.8, add 100 μL of 1M IPTG (final concentration: 200 μM, see Note 10). Keep the flask at 37°C for about 30 min (see Note 11). 11. Transfer the flask to a second shaker with the temperature set to 15°C, and incubate at 250 rpm overnight (see Note 12). 12. Centrifuge the cells at 4,900 g for 20 min at 4°C. Store the cell pellet at −80°C. 13. For 15N,13C enrichment, replace 14NH4Cl and dextrose in steps 6 and 8 described above, with 15NH4Cl and [U-13C]-glucose, respectively. 14. Under normal circumstances, for uniform 15N,13C,2H enrichment, grow the cells in M9 medium in which H2O is replaced by D2O. However, in the case of SrcCD, we found that the introduction of 2H2O dramatically reduces the yields. Therefore, we follow the approach developed by Fiaux et al. (60), where a pool of 15N,13C,2H-labeled amino acids were used to supplement M9 medium prepared in H2O. In this case: In step 6 above, replace 14NH4Cl with 15NH4Cl; in step 8, above, replace 14 NH4Cl with 15NH4Cl, and use 0.5 g of dextrose (1 g/L instead of 5 g/L) and add 0.75–1.0 g of 15N,13C,2H-labeled algal lysate medium (e.g., CELTONE® 1.5–2.0 g/L).

7

Expression and Purification of Src-family Kinases for Solution NMR Studies

4.1.2. Protocol for the Purification of SrcCD from the MBP Fusion Protein

123

1. Lyse cells using a French Press (1,100 psi), followed by centrifugation at 16,000 g for 20 min at 4°C. Transfer the supernatant to a centrifuge tube. 2. Check the pH of the supernatant (see Note 13), if it is below 7.5, adjust by adding small amounts of solid tris-hydroxymethyl-aminomethane (Tris). 3. Transfer preequilibrated cobalt-resin into the centrifuge tube containing the lysate supernatant, and incubate for about 1 h under rotation or slight agitation at 4°C. 4. Pour the lysate and beads into an empty column. Elute the lysate and wash the column with four bed volumes of lysis buffer and two bed volumes of lysis buffer containing 3 mM imidazole (3 μL Imidazole buffer/mL of lysis buffer). 5. Elute the protein with four bed volumes of the lysis buffer containing 300 mM imidazole (combine three volumes of Imidazole buffer with seven volumes of lysis buffer). 6. Add 20 μL of EDTA solution/mL of eluate (final concentration: 10 mM; see Note 14) to the eluate prior to dialysis and then dialyze against cleavage buffer at 4°C. 7. Determine the concentration of the fusion protein by measuring the OD280 using 119,100 M-1-cm-1 as the extinction coefficient. 8. Add 5 units of α-thrombin/mg of fusion protein to the solution, immediately add β-mercaptoethanol to a final concentration of 4 mM. Cleave overnight at 4°C. 9. Check for completion of the cleavage reaction by running an SDS–PAGE gel. Add more thrombin if necessary. When the cleavage is complete, add solid DTT and the protease inhibitor, AEBSF (4-(2-aminoethyl) benzenesulfonyl fluoride) to final concentrations of 5 mM and 100 μM, respectively. Add AP23464, a high-affinity kinase inhibitor, to the cleavage buffer to a final concentration of 100 μM for wild-type SrcCD. 10. Dilute the completed reaction fivefold into ion exchange chromatography buffer A, and load the resulting sample onto a Q column (ion-exchange) at a flow-rate of 4 mL/min (the volume at this stage is likely to be large, possibly more than 100 mL; see Note 15). 11. Elute the cleaved kinase by running a salt gradient from 0 to 300 mM NaCl, using ion exchange chromatography buffer B, in 120 min using a flow-rate of 1 mL/min. Unphosphorylated SrcCDK295M appears at around 15.4 mS/cm and SrcCDK295M phosphorylated at Tyr416 appears at around 16.9 mS/cm. Wildtype SrcCD bound with the kinase inhibitor, AP23464, appears at around 20.4 mS/cm. MBP elutes at around 8–10 mS/cm, multiple peaks may appear due to nonspecific phosphorylation (see Note 16).

124

A. Piserchio et al.

4.2. Expression and Purification of the Fully Active, Wild-Type Catalytic Domain of c-Src

We provide below a step-by-step protocol for the expression and purification of wild-type SrcCD without the use of an N-terminal MBP tag. The SrcCD obtained using this protocol is fully active and does not require a kinase inhibitor to allow purification to homogeneity. To implement this protocol, first SrcCD and Cdc37 have to be cloned into a pACYCDuet vector (Chloramphenicol resistant) using standard procedures. SrcCD should be inserted in the first cloning site using the EcoRI and SalI restriction enzymes (resulting in a N-terminal His6-tag for SrcCD, see Note 17), and Cdc37 in the second site using XhoI and NdeI (see Note 18). SrcCD expression was also tested with and without the concomitant coexpression of a third protein, the tyrosine phosphatase, Lyp (61) (a kind gift from Dr. Ronald Seidel, Albert Einstein College of Medicine). Lyp selectively dephosphorylates the activating Tyr residue in SFKs (specifically Tyr394 in Lck), and was obtained in a H-MBP-3C Ampicillin resistant vector (see Note 1). 1. If using electrocompetent cells, transfer ~20 ng of pACYCDuet vector into the Eppendorf tube containing BL21(DE3) T1 electrocompetent cells. Transfer the contents of the Eppendorf tube into a prechilled 0.1 cm gap electroporation cuvette and incubate for 2 min on ice (see Note 19). If also overexpressing the phosphatase, then transfer both vectors (pACYC and H-MBP-3C, 20 ng each) into the cuvette at the same time (see Note 20). 2. Follow steps 4–14 of Subheading 4.1.1 but use Chloramphenicol (35 mg/L) and Ampicillin only if overexpressing the phosphatase. This applies to step 4 (Chloramphenicol, or Ampicillin and Chloramphenicol double resistant plates) and step 6 (Chloramphenicol, or Ampicillin and Chloramphenicol added to the M9 medium). 3. Purify the protein following steps 1–6 of Subheading 4.1.2. Note that the final purified protein retains the His6-tag. If the tag needs to be removed, a thrombin cleavage site could be inserted at step 1 in the SrcCD forward primer (described in Note 10). Then steps 7–9 from Subheading 4.1.2. would also apply. Note that, in our construct, the MBP-phosphatase fusion protein has an N-terminal His6 tag and coelutes with the kinase. 4. After elution, dialyze the protein(s) against dialysis buffer. 5. Concentrate the protein(s) down to 2 mL and inject it onto a size exclusion column equilibrated with size exclusion chromatography buffer. Note that the phosphatase elutes in the void (presumably due to some level of self-association) while the kinase elutes consistently at the volume expected for its molecular weight, and are both readily isolated.

7

Expression and Purification of Src-family Kinases for Solution NMR Studies

125

Both methods of expression (with and without coexpression with Lyp) lead to the isolation of active, wild-type SrcCD (confirmed by analysis of kinetics, see Note 21). However, a higher final OD600 and overall yield of purified protein was observed when the kinase was coexpressed with Lyp phosphatase (~190 nmol/L of culture).

5. Conclusions We have described how careful optimization of overexpression protocols allows the expression and purification of the catalytic domain of c-Src both in mutant, as well as active wild-type forms using E. coli expression systems. In particular, we developed two related strategies, the first employs both an N-terminal MBP fusion and coexpression with the GroEL and GroES chaperones, while the second, by taking advantage of the higher specificity of the Cdc37 cochaperone for the nascent c-Src chain, does not require the production of SrcCD as an MBP fusion protein. Both methods present advantages and disadvantages. In particular, the first method leads to higher yields, but it is more efficient in producing the catalytically compromised (K295M) mutant compared to the wild-type protein. The wild-type protein can only be purified in the presence of a high-affinity kinase inhibitor when using this method. The second method produces lower yields; however it also less labor intensive and more importantly, allows the production of the catalytic domain in active, wild-type form. Given the higher yields, the first method is probably more useful when SrcCD is required in large amounts for studies such as those to determine specific protein–protein interactions that do not require the intact catalytic properties of the kinase. However, given that the MBPdependent protocol also produces a considerable amount of soluble misfolded aggregates that have not been observed using the Cdc37-dependent protocol, we believe that the latter may be better suited for “in-cell” NMR studies in E. coli (62–64). Obviously, the latter protocol is required when wild-type, fully active SrcCD needs to be produced. While this manuscript was in production, Campos-Olivas et. al. (65) utilized some of the procedures described herein, to produce sufficient amounts of SrcCD, and obtain backbone resonance assignments in complex with the kinase inhibitor imatinib.

126

A. Piserchio et al.

6. Notes 1. The H-MBP-3C vector contains a thrombin cleavage site and features a maltose binding protein (MBP) tag to enhance solubility and a His6 sequence to facilitate protein purification using a metal affinity chromatography. The protease 3C site described in the original publication (58) has been replaced by a thrombin site in the present case. 2. The pACYCDuet vector was purchased from Novagen (EMD Biosciences). 3. We used an Eppendorf 2510 electroporator. 4. We observed that the transformation of the pREP4-groESL vector was often inefficient, and therefore, we avoided the simultaneous double transformation of pREP4-groESL and H-MBP-3C vector encoding SrcCD, by preparing a stock of electrocompetent cells pretransformed with the former. Electrocompetent cells were preferred over chemically competent cells, due to their ease of preparation and high efficiency of transformation. These cells were prepared by streaking out commercial BL21 (DE3) competent cells on an LB plate and incubating the plate overnight at 37°C. From this, electrocompetent cells were prepared using standard protocols. 5. We used Kao and Michaylchuk vitamin solution (100×) purchased from Sigma. 6. CELTONE® is a rich growth medium for bacteria derived from algal extracts. It may be purchased from Cambridge Isotopes Laboratories. 7. We used TALON beads purchased from Clontech. 8. We used HiTrap Q columns from GE Healthcare Biosciences Corp. 9. We used a HiPrep 16/60 Sephacryl S-100 column from GE Healthcare Biosciences Corp. All protein purification was carried out using an Akta Explorer 100 from GE Healthcare Biosciences Corp. 10. Lower IPTG concentrations are recommended at low temperature to avoid toxicity. 11. This helps to increase the final yield by boosting mRNA production, while most protein is synthesized during the following lower temperature growth. In the first hour after induction, bacteria mainly activate mRNA transcription, while actual translation is still minimal.

7

Expression and Purification of Src-family Kinases for Solution NMR Studies

127

12. All bacterial growth postinduction (after the initial half an hour at 37°C, see Note 11) was carried out at a low temperature (15°C). It is well known that low temperatures decrease the degradation rates of unstable proteins; in the case of SrcCD, we observed that low temperature growth is essential for the production of soluble, properly folded SrcCD. We speculate that SrcCD folding is a slow process that is in competition with the formation of aggregates both soluble and insoluble. In our protocol both the N-terminal MBP fusion and the GroEL/ GroES chaperones (or the Cdc37 cochaperone) provide assistance in stabilizing the nascent chain and protect it from aggregation. Lower temperatures likely assist this process by reducing the rates of protein synthesis and aggregation. 13. We noticed that after long induction times at low temperature, E. coli cells release significant amounts of acid molecules into the lysis buffer. This is, in some instances, sufficient to overcome the buffering power of the solution leading to a pH below 7.0, compromising efficient binding of the His6 tag to the metal affinity column. 14. EDTA is introduced to chelate Co2+ cations. Metal leaks from the resin can be caused by mechanical stress and by high Imidazole concentrations. If not chelated by EDTA, Co2+ can be difficult to eliminate by dialysis, since it can bind the His6-tag as well as the metal binding pockets in SrcCD if left in the sample. Co2+ forms a reddish precipitate once a substantial amount of reducing agent has been added (interfering with the remaining purification process), while the Co2+ remaining in solution, being paramagnetic, degrades the quality of the NMR spectra. 15. We normally do not load more than 15 mg of fusion protein into the ion exchange column at a given time. 16. Unlike the catalytically compromised mutant, wild-type SrcCD does not elute once loaded on the Q column. Presumably the wild-type protein is less stable under conditions of low ionic strength used for sample loading, and precipitates on the column. We found however, that this construct could also be successfully loaded and eluted using the protocol described above by adding a high-affinity kinase inhibitor (59) after cleaving the fusion protein. 17. For SrcCD (human), we used the following primers: Forward: GGTGGTGGAATTCGTCCAAGCCGCAGA CTCAGGG. Reverse: GGTGGTGGTCGACCTAGAGGTTCTCCCC GGGCTGGTA.

128

A. Piserchio et al.

18. For Cdc37 we used the following primers: Forward: GGTGGTGCATATGGTGGACTACAGCG TGTGGGAC. Reverse: GGTGGTGCTCGAGTCACACACTGACATC CTTCTCATCGCC. 19. The electrocompetent cells were homemade and derived from the commercial BL21 (DE3) T1 expression system (chemical competent). We expect that transformation directly into the commercial electrocompetent cells would produce equivalent results. 20. In contrast to what was observed for the pREP4-GroESL vectors, cotransformation of both the H-MBP-3C Lyp vector and the pACYCDuet hosting SrcCD and Cdc37 does not present any difficulties, and can be performed following standard protocols with chemically competent BL21 (DE3) cells. 21. The kcat for the phosphorylation of a substrate peptide polyE4Y by SrcCD was measured using a photometric assay based on the recycling of ADP and the concomitant consumption of NADH catalyzed by L-lactate-dehydrogenase (LDH) and pyruvate kinase (PK). PK regenerates ATP from ADP (a byproduct of the SrcCD phosphorylation process) by converting phosphoenolpyruvate into pyruvate. LDH converts then the pyruvate into lactate by oxidizing NADH. Velocities are obtained following the time-dependent decrease of the NADH fluorescence. This method yielded a kcat of 240 ± 10/min, a value comparable to that obtained previously (41).

Acknowledgments The following grants from the National Institutes of Health, GM084278 (to R. G.), GM047021 (to D. C.) and RR03060 (toward partial support of the NMR facilities at CCNY), are acknowledged. The authors are members of the New York Structural Biology Center, a STAR center supported by the New York State Office for Science, Technology and Academic Research. The authors thank Dr. Avrom Caplan (CCNY), Dr. Ronald Siedel (Albert Einstein College of Medicine), and Dr. Kaushik Dutta (NYSBC) for gifts of materials, and members of the Ghose and Cowburn labs for useful discussions.

7

Expression and Purification of Src-family Kinases for Solution NMR Studies

129

References 1. Hunter, T. (2000) Signaling – 2000 and beyond. Cell 100, 113–127. 2. Johnson, L. N. (2009) The regulation of protein phosphorylation. Biochem. Soc. Trans. 37, 627–641. 3. Tarrant, M. K., and Cole, P. A. (2009) The chemical biology of protein phosphorylation. Annu. Rev. Biochem. 78, 797–825. 4. Ostman, A., and Bohmer, F. D. (2001) Regulation of receptor tyrosine kinase signaling by protein tyrosine phosphatases. Trends Cell Biol. 11, 258–266. 5. Shi, Y. (2009) Serine/threonine phosphatases: mechanism through structure. Cell 139, 468–484. 6. Kolibaba, K. S., and Druker, B. J. (1997) Protein tyrosine kinases and cancer. Biochim. Biophys. Acta 1333, F217–248. 7. Blume-Jensen, P., and Hunter, T. (2001) Oncogenic kinase signalling. Nature 411, 355–365. 8. Chong, P. K., Lee, H., Kong, J. W., Loh, M. C., Wong, C. H., and Lim, Y. P. (2008) Phosphoproteomics, oncogenic signaling and cancer research. Proteomics 8, 4370–4382. 9. Pawson, T., and Kofler, M. (2009) Kinome signaling through regulated protein–protein interactions in normal and cancer cells. Curr. Opin. Cell Biol. 21, 147–153. 10. Johnson, L. N. (2009) Protein kinase inhibitors: contributions from structure to clinical compounds. Q. Rev. Biophys. 42, 1–40. 11. Noble, M. E., Endicott, J. A., and Johnson, L. N. (2004) Protein kinase inhibitors: insights into drug design from structure. Science 303, 1800–1805. 12. Nichols, G. L. (2003) Tyrosine kinase inhibitors as cancer therapy. Cancer Invest. 21, 758–771. 13. Shawver, L. K., Slamon, D., and Ullrich, A. (2002) Smart drugs: tyrosine kinase inhibitors in cancer therapy. Cancer Cell 1, 117–123. 14. Nolen, B., Taylor, S., and Ghosh, G. (2004) Regulation of protein kinases; controlling activity through activation segment conformation. Mol. Cell 15, 661–675. 15. Kornev, A. P., and Taylor, S. S. (2010) Defining the conserved internal architecture of a protein kinase. Biochim. Biophys. Acta 1804, 440–444. 16. Parsons, S. J., and Parsons, J. T. (2004) Src family kinases, key regulators of signal transduction. Oncogene 23, 7906–7909.

17. Roux, P. P., and Blenis, J. (2004) ERK and p38 MAPK-activated protein kinases: a family of protein kinases with diverse biological functions. Microbiol. Mol. Biol. Rev. 68, 320–344. 18. Chen, Z., Gibson, T. B., Robinson, F., Silvestro, L., Pearson, G., Xu, B., Wright, A., Vanderbilt, C., and Cobb, M. H. (2001) MAP kinases. Chem. Rev. 101, 2449–2476. 19. Hubbard, S. R., and Till, J. H. (2000) Protein tyrosine kinase structure and function. Annu. Rev. Biochem. 69, 373–398. 20. Shi, Z., Resing, K. A., and Ahn, N. G. (2006) Networks for the allosteric control of protein kinases. Curr. Opin. Struct. Biol. 16, 686–692. 21. Fedorov, O., Muller, S., and Knapp, S. (2010) The (un)targeted cancer kinome. Nature Chem. Biol. 6, 166–169. 22. Grueneberg, D. A., Degot, S., Pearlberg, J., Li, W., Davies, J. E., Baldwin, A., Endege, W., Doench, J., Sawyer, J., Hu, Y., Boyce, F., Xian, J., Munger, K., and Harlow, E. (2008) Kinase requirements in human cells: I. Comparing kinase requirements across various cell types. Proc. Natl. Acad. Sci. U.S.A. 105, 16472–16477. 23. Baldwin, A., Li, W., Grace, M., Pearlberg, J., Harlow, E., Munger, K., and Grueneberg, D. A. (2008) Kinase requirements in human cells: II. Genetic interaction screens identify kinase requirements following HPV16 E7 expression in cancer cells. Proc. Natl. Acad. Sci. U.S.A. 105, 16478–16483. 24. Bommi-Reddy, A., Almeciga, I., Sawyer, J., Geisen, C., Li, W., Harlow, E., Kaelin, W. G., Jr., and Grueneberg, D. A. (2008) Kinase requirements in human cells: III. Altered kinase requirements in VHL−/− cancer cells detected in a pilot synthetic lethal screen. Proc. Natl. Acad. Sci. U.S.A. 105, 16484–16489. 25. Grueneberg, D. A., Li, W., Davies, J. E., Sawyer, J., Pearlberg, J., and Harlow, E. (2008) Kinase requirements in human cells: IV. Differential kinase requirements in cervical and renal human tumor cell lines. Proc. Natl. Acad. Sci. U.S.A. 105, 16490–16495. 26. Wiesner, S., Wybenga-Groot, L. E., Warner, N., Lin, H., Pawson, T., Forman-Kay, J. D., and Sicheri, F. (2006) A change in conformational dynamics underlies the activation of Eph receptor tyrosine kinases. EMBO J. 25, 4686–4696. 27. Vajpai, N., Strauss, A., Fendrich, G., CowanJacob, S. W., Manley, P. W., Grzesiek, S., and Jahnke, W. (2008) Solution conformations and

130

28.

29.

30.

31.

32.

33.

34. 35.

36.

37.

38.

39.

A. Piserchio et al. dynamics of Abl kinase-inhibitor complexes determined by NMR substantiate the different binding modes of imatinib/nilotinib and dasatinib. J. Biol. Chem. 283, 18292–18302. Piserchio, A., Ghose, R., and Cowburn, D. (2009) Optimized bacterial expression and purification of the c-Src catalytic domain for solution NMR studies. J. Biomol. NMR 44, 87–93. Piserchio, A., Warthaka, M., Devkota, A. K., Kaoud, T. S., Lee, S., Abramczyk, O., Ren, P., Dalby, K. N., and Ghose R. (2011) Solution NMR insights into docking interactions involving inactive ERK2. Biochemistry 50, 3660–3672. Masterson, L. R., Mascioni, A., Traaseth, N. J., Taylor, S. S., and Veglia, G. (2008) Allosteric cooperativity in protein kinase A. Proc. Natl. Acad. Sci. U.S.A. 105, 506–511. Masterson, L. R., Cheng, C., Yu, T., Tonelli, M., Kornev, A., Taylor, S. S., and Veglia, G. (2010) Dynamics connect substrate recognition to catalysis in protein kinase A. Nature Chem. Biol. 6, 821–828. Liu, D., Xu, R., and Cowburn, D. (2009) Segmental isotopic labeling of proteins for nuclear magnetic resonance. Meth. Enzymol. 462, 151–175. Boggon, T. J., and Eck, M. J. (2004) Structure and regulation of Src family kinases. Oncogene 23, 7918–7927. Martin, G. S. (2001) The hunting of the Src. Nature Rev. Mol. Cell Biol. 2, 467–475. Alvarez, R. H., Kantarjian, H. M., and Cortes, J. E. (2006) The role of Src in solid and hematologic malignancies: development of newgeneration Src inhibitors. Cancer 107, 1918–1929. Russello, S. V., and Shore, S. K. (2003) Src in human carcinogenesis. Front. Biosci. 8, s1068–1073. Trevino, J. G., Summy, J. M., and Gallick, G. E. (2006) Src inhibitors as potential therapeutic agents for human cancers. Mini Rev. Med. Chem. 6, 681–687. Grgurevich, S., Mikhael, A., and McVicar, D. W. (1999) The Csk homologous kinase, Chk, binds tyrosine phosphorylated paxillin in human blastic T cells. Biochem. Biophys. Res. Commun. 256, 668–675. Zrihan-Licht, S., Deng, B., Yarden, Y., McShan, G., Keydar, I., and Avraham, H. (1998) Csk homologous kinase, a novel signaling molecule, directly associates with the activated ErbB-2 receptor in breast cancer cells and inhibits their proliferation. J. Biol. Chem. 273, 4065–4072.

40. Sicheri, F., Moarefi, I., and Kuriyan, J. (1997) Crystal structure of the Src family tyrosine kinase Hck. Nature 385, 602–609. 41. Weijland, A., Neubauer, G., Courtneidge, S. A., Mann, M., Wierenga, R. K., and SupertiFurga, G. (1996) The purification and characterization of the catalytic domain of Src expressed in Schizosaccharomyces pombe. Comparison of unphosphorylated and tyrosine phosphorylated species. Eur. J. Biochem. 240, 756–764. 42. Feder, D., and Bishop, J. M. (1990) Purification and enzymatic characterization of pp60c-src from human platelets. J. Biol. Chem. 265, 8205–8211. 43. Seeliger, M. A., Young, M., Henderson, M. N., Pellicena, P., King, D. S., Falick, A. M., and Kuriyan, J. (2005) High yield bacterial expression of active c-Abl and c-Src tyrosine kinases. Protein Sci. 14, 3135–3139. 44. Stover, D. R., Liebetanz, J., and Lydon, N. B. (1994) Cdc2-mediated modulation of pp60csrc activity. J. Biol. Chem. 269, 26885–26889. 45. Pellecchia, M., Bertini, I., Cowburn, D., Dalvit, C., Giralt, E., Jahnke, W., James, T. L., Homans, S. W., Kessler, H., Luchinat, C., Meyer, B., Oschkinat, H., Peng, J., Schwalbe, H., and Siegal, G. (2008) Perspectives on NMR in drug discovery: a technique comes of age. Nature Rev. Drug Discov. 7, 738–745. 46. Cole, P. A. (1996) Chaperone-assisted protein expression. Structure 4, 239–242. 47. Nomine, Y., Ristriani, T., Laurent, C., Lefevre, J. F., Weiss, E., and Trave, G. (2001) Formation of soluble inclusion bodies by HPV E6 oncoprotein fused to maltose-binding protein. Protein Expr. Purif. 23, 22–32. 48. Kapust, R. B., and Waugh, D. S. (1999) Escherichia coli maltose-binding protein is uncommonly effective at promoting the solubility of polypeptides to which it is fused. Protein Sci. 8, 1668–1674. 49. Wang, D., Huang, X. Y., and Cole, P. A. (2001) Molecular determinants for Csk-catalyzed tyrosine phosphorylation of the Src tail. Biochemistry 40, 2004–2010. 50. Ozkirimli, E., and Post, C. B. (2006) Src kinase activation: a switched electrostatic network. Protein Sci. 15, 1051–1062. 51. Cowan-Jacob, S. W. (2006) Structural biology of protein tyrosine kinases. Cell Mol. Life Sci. 63, 2608–2625. 52. Welch, W. J., and Feramisco, J. R. (1982) Purification of the major mammalian heat shock proteins. J. Biol. Chem. 257, 14949–14959.

7

Expression and Purification of Src-family Kinases for Solution NMR Studies

53. Hahn, J. S. (2009) The Hsp90 chaperone machinery: from structure to drug development. BMB Rep. 42, 623–630. 54. Zuehlke, A., and Johnson, J. L. (2010) Hsp90 and co-chaperones twist the functions of diverse client proteins. Biopolymers 93, 211–217. 55. Caplan, A. J., Mandal, A. K., and Theodoraki, M. A. (2007) Molecular chaperones and protein kinase quality control. Trends Cell Biol. 17, 87–92. 56. Karnitz, L. M., and Felts, S. J. (2007) Cdc37 regulation of the kinome: when to hold ‘em and when to fold ‘em. Science STKE 2007, pe22. 57. Whitelaw, M. L., Hutchison, K., and Perdew, G. H. (1991) A 50-kDa cytosolic protein complexed with the 90-kDa heat shock protein (hsp90) is the same protein complexed with pp60v-Src hsp90 in cells transformed by the Rous sarcoma virus. J. Biol. Chem. 266, 16436–16440. 58. Alexandrov, A., Dutta, K., and Pascal, S. M. (2001) MBP fusion protein with a viral protease cleavage site: one-step cleavage/purification of insoluble proteins. Biotechniques 30, 1194–1198. 59. Dalgarno, D., Stehle, T., Narula, S., Schelling, P., van Schravendijk, M. R., Adams, S., Andrade, L., Keats, J., Ram, M., Jin, L., Grossman, T., MacNeil, I., Metcalf, C., 3rd, Shakespeare, W., Wang, Y., Keenan, T., Sundaramoorthi, R., Bohacek, R., Weigele, M., and Sawyer, T. (2006) Structural basis of Src tyrosine kinase inhibition with a new class of potent and

60.

61.

62.

63.

64.

65.

131

selective trisubstituted purine-based compounds. Chem. Biol. Drug Des. 67, 46–57. Fiaux, J., Bertelsen, E. B., Horwich, A. L., and Wuthrich, K. (2004) Uniform and residue-specific 15N-labeling of proteins on a highly deuterated background. J. Biomol. NMR 29, 289–297. Wu, J., Katrekar, A., Honigberg, L. A., Smith, A. M., Conn, M. T., Tang, J., Jeffery, D., Mortara, K., Sampang, J., Williams, S. R., Buggy, J., and Clark, J. M. (2006) Identification of substrates of human protein-tyrosine phosphatase PTPN22. J. Biol. Chem. 281, 11002–11010. Burz, D. S., Dutta, K., Cowburn, D., and Shekhtman, A. (2006) In-cell NMR for protein–protein interactions (STINT-NMR). Nature Protoc. 1, 146–152. Selenko, P., and Wagner, G. (2006) NMR mapping of protein interactions in living cells. Nature Meth. 3, 80–81. Sakakibara, D., Sasaki, A., Ikeya, T., Hamatsu, J., Hanashima, T., Mishima, M., Yoshimasu, M., Hayashi, N., Mikawa, T., Walchli, M., Smith, B. O., Shirakawa, M., Guntert, P., and Ito, Y. (2009) Protein structure determination in living cells by in-cell NMR spectroscopy. Nature 458, 102–105. Campos-Olivas, R., Marenchino, M., Scapozza, L. and Gervasio, F. L. (2011) Backbone assignment of the tyrosine kinase Src catalytic domain in complex with imatinib. Biomol. NMR Asgn. 5, 221–224.

Chapter 8 NMR Studies of Large Protein Systems Shiou-Ru Tzeng, Ming-Tao Pai, and Charalampos G. Kalodimos Abstract Over the recent years, there has been increased interest in applying NMR spectroscopy for the characterization of proteins and protein complexes of large molecular weight. The combination of multidimensional NMR, novel pulse sequences allowing for the selection of slowly relaxing coherence pathways, and the development of a range of labeling techniques has enabled high-resolution NMR analyses of supramolecular systems of even megadalton size. Here, we describe how NMR can be used to obtain structural information in large systems by using as an example the recent structure determination of SecA ATPase (204 kDa) in complex with a signal peptide. Key words: Paramagnetic resonance enhancement, Macromolecular complex, Molecular machinery, Stable isotope labeling

1. Introduction Application of NMR spectroscopy to supramolecular systems (>200 kDa) has been revolutionized by specific labeling of methyl groups (1). The labeling protocol, pioneered by the group of Lewis Kay, is very simple and robust (2, 3). The approach exploits some very favorable properties of methyl groups in proteins: (a) they occur frequently in the hydrophobic cores of proteins and at the interfaces of biomolecular complexes, and are thus excellent reporters of structure and dynamics; (b) the three protons of the methyl group all contribute to the intensity of the same signal, and therefore methyl probes are significantly more sensitive than other candidates; and (c) methyl groups are intrinsically optimized for use in TROSY

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_8, © Springer Science+Business Media, LLC 2012

133

134

S.-R. Tzeng et al.

spectroscopy and the simple 1H-13C HMQC experiment can be used to select for pathways with favorable relaxation properties (4). Currently, the methyl groups of five different amino acids can be labeled in a highly specific and scramble-free manner: Ala, Ile (δ1), Leu, Met, and Val. These five residues are highly abundant, typically accounting for 35–45% of the total number of residues in a protein and are distributed throughout the protein, thus providing almost complete coverage of the protein space. The methyl-labeling approach combined with methyl-TROSY currently provides the method of choice for NMR characterization of large protein systems. Although this approach has proven to be very robust for recording spectra of large proteins with high sensitivity and resolution, a major hurdle in obtaining site-specific information remains the difficulty in obtaining assignments. While the traditional approach of assigning the backbone and subsequently linking the methyl side chains to the backbone has worked efficiently for smaller proteins, it is not applicable to larger systems. The only approach currently is to “disassemble” the supramolecular system. For higher order oligomeric systems, such as the proteasome (5), this means preparing the subunit in its monomeric form and for large, single-chain proteins, such as the SecA (6), preparing isolated domains or fragments. In principle, determining solution structures of supramolecular protein–ligand complexes by NMR should be feasible, provided that the crystal structures of the free partners are previously known. Because usually only methyl groups can be robustly and unambiguously detected for supramolecular systems, in cases where complex interactions are mediated by hydrophobic contacts involving methyl-bearing residues it is likely that intermolecular NOEs can be detected, thereby enabling the reliable docking of the complex. Unfavorable motions commonly observed at protein interfaces, however, may result in line broadening and render NOE detection unfeasible. Although the NOE has served as the gold standard for protein structure determination by NMR, the old, but recently resurrected, paramagnetic relaxation enhancement (PRE) (7, 8) technique holds great promise for obtaining both structural and dynamic information in supramolecular protein complexes (6). By combining transferred NOESY, line broadening, and PRE experiments, the structure of the 204 kDa SecA ATPase in complex with a secretory signal peptide was recently determined (6). Using this system as an example, we describe strategies to (a) obtain samples optimally labeled for methyl detection, (b) assign the methyl resonances of the large protein system, and (c) obtain intermolecular distance restraints for the structure determination of large protein–ligand complexes.

8

NMR Studies of Large Protein Systems

135

2. Materials 1. Frozen, transformed Escherichia coli BL21(DE3) cells to overexpress protein of interest. 2. M9 medium: 6 g/L Na2HPO4, 3 g/L KH2PO4, pH 7.0–7.4., 0.5 g/L NaCl, 1.0 g/L NH4Cl or 15NH4Cl. Autoclave, let the medium cool down, and then add 0.1 mL/L of 1M CaCl2, 1 mL/L of 1M MgSO4, and 2 g/L D-[2H,12C]-glucose or D-[2H,13C]-glucose. 3. 1M CaCl2 stock: Dissolve 11.0 g of CaCl2 in 100 mL of D2O; filter sterilize. 4. 1M MgSO4 stock: Dissolve 12.04 g of MgSO4 in 100 mL of D2O; filter sterilize. 5. D2O (Cambridge Isotope Laboratories, CIL). 6. BIOEXPRESS (CIL). 7. ISOGRO (Isotec). 8.

13

CH3-2H-alanine.

9. α-Ketobutyrate (CIL or Isotec). 10. α-Ketoisovalerate (CIL or Isotec). 11.

13

CH3-methionine (CIL).

12. IPTG. 13. AMICON stir cell (Millipore). 14. L-broth: 10 g tryptone/L, 5 g yeast extract/L, 5 g NaCl/L, adjusted to pH 7.4 with NaOH; autoclave. 15. (2,2,5,5-Tetramethyl-2,5-dihydro-1H-pyrrol-3-yl)methyl methanesulfonothioate (MTSL) (Toronto Research Chemicals Inc.): Dissolve in acetonitirile. 16. Ascorbic acid.

3. Methods 3.1. Protein Labeling for Methyl Detection

1. Pick a freshly transformed colony (see Note 1) of BL21(DE3) cells and inoculate a 1–2 mL culture of L-broth in D2O containing 0.1% of glucose at 37°C until cells reach an OD600 of ~0.7–0.8. 2. Centrifuge the cells at 1,200 × g for 20 min at room temperature and resuspend them in 5–10 mL of sterile M9 medium in D2O (M9/D2O) in a sterile flask to a starting OD600 of ~0.05. Incubate the culture in a shaking incubator (220–250 rpm) at 37°C until it reaches an OD600 of ~0.6.

136

S.-R. Tzeng et al.

3. Centrifuge as in step 2, and resuspend the cells in 50–100 mL of M9/D2O prepared with either D-[2H,13C]glucose (see Note 2) or D-[2H, 12C]glucose (see Note 3) and 15NH4Cl containing 2% of BIOEXPRESS (CIL) or ISOGRO (Sigma) (see Note 4). The starting OD600 should be ~0.1. Incubate the culture in a shaking incubator (220–250 rpm) at 37°C until the OD600 is ~0.6. 4. Centrifuge as in step 2, resuspend the cells in 1 L of M9/D2O, and grow until the OD600 is ~0.25. 5. At this point, amino acid precursors for methyl labeling can be added. (a) Add precursors (see Note 5) or amino acids 30–60 min prior to IPTG induction; add the following quantities (final concentration): 100 mg/L of 13CH3-2Hα-alanine (see Note 6) for Ala labeling; 45–50 mg/L of α-ketobutyrate for isoleucine labeling; 85–100 mg/L of α-ketoisovalerate for leucine and valine labeling (see Note 7); 250 mg/L of [13CH3]-methionine for Met labeling. The methyl groups of all five residues can be labeled in one sample in a scramblefree manner (see Note 8) (Fig. 1a). (b) Continue incubating the culture for approximately 1 h. The OD600 should reach a value of ~0.3–0.4. 6. Add IPTG to 0.5 mM to induce protein overexpression. 7. Continue postinduction growth for 6–8 h (see Note 9). 8. Harvest the cells by centrifugation at 5,000 × g for 15 min at 4°C. 9. Freeze the wet cell pack at −80°C. a

b 12

20

Met

16 18

Leu, Val 20

(p.p.m.)

21 Ala

13C

13C

(p.p.m.)

14

Ile

22 23 24

22 25 24 26 26 27 2.0

1.6

1.2 0.8 1H (p.p.m.)

0.4

0.0

1.2 1.0

0.8 0.6

0.4

0.2

0.0

Fig. 1. (a) 1H-13C HMQC of [U-2H,12C], Ala-, Leu-, Met-, Val-, Ile-δ1-[13CH3] labeled protein. The methyl groups of all five amino acids can be labeled with no scrambling. (b) 1H-13C HMQC of the same protein as in (a, black) but prepared using 10% BIOEXPRESS (green). The Leu methyl groups are completely suppressed, whereas the Val methyl groups are only minimally affected.

8

3.2. NMR Assignment

NMR Studies of Large Protein Systems

137

To assign a large protein, such as SecA (204 kDa; 901 residues per subunit), a domain-parsing strategy is followed. 1. Isolate and characterize, by using NMR, virtually all domains of the full-length protein and a number of fragments comprising contiguous domains (see Note 10). The size of the isolated domains and fragments should be such that backbone and side-chain assignment is feasible using standard approaches. 2. Prepare Ala, Ile, Met, Leu, and Val methyl-labeled samples for the full-length protein and the domains thereof (see Note 11). Record methyl-TROSY for all of the samples, overlay, and compare the spectra of the individual domains against the spectra of the longer fragments and full-length protein. If good resonance correspondence among domains, fragments, and the full-length protein can be demonstrated, then assignment is in principle transferable. 3. Record standard triple-resonance NMR experiments for isolated domains and obtain backbone assignments. Record the threedimensional spectra required for the assignment of the methyl groups (see Note 12). 4. Transfer the methyl assignments obtained for the isolated domains to the larger fragments and finally to the full-length protein by visually inspecting the methyl-TROSY spectra. Only the assignment of the obvious and well-dispersed resonances can be safely transferred this way. 5. Record 13C HMQC-NOESY-HMQC spectra (see Note 13) for the methyl-labeled samples. Use the NOE patterns to confirm and extend the assignment transfer from the domains to the full-length protein. If a crystal structure is available, it can be used to determine the distances between the methyl groups and assist with the assignment. 6. Prepare site-directed mutations to assign ambiguous resonances and further extend and confirm the assignments (see Note 14).

3.3. Paramagnetic Relaxation Enhancement Measurements

1. To prepare nitroxide spin label (MTSL)-derivatized ligand via cysteine-specific modification of engineered ligand derivatives containing single-solvent-accessible cysteine residues at sites of interest (see Note 15), add MTSL from a concentrated stock in acetonitrile to the ligand solution (free from any reducing agent) at a tenfold molar excess over the ligand and allow the reaction to proceed at 4°C for ~12 h. If available, confirm the completion of the reaction by mass spectrometry. 2. Remove excess MTSL by extensive dialysis using an Amicon stirred cell. 3. Determine PRE-derived distances from 1H-13C HMQC spectra by measuring peak intensities before (paramagnetic) and after

138

S.-R. Tzeng et al.

(diamagnetic) reduction of the nitroxide spin label by the addition of 5 mM ascorbic acid (see Note 16). 4. Convert PRE values to distances by using a modified SolomonBloembergen equation for transverse relaxation (7). 5. Incorporate distance (intermolecular) restraints into the structure calculation protocol of the complex. 6. Restrain resonances strongly affected by the presence of the spin label in the ligand (Ipara /Idia < 0.15) and whose resonances broaden beyond detection in the paramagnetic spectrum with only an upper bound distance estimated from the noise of the spectrum plus 4 Å. 7. Restrain resonances that appear in the paramagnetic spectra (Ipara /Idia < 0.85) as the calculated distance with ±4 Å upper/ lower bounds. 3.4. Structure Determination

1. Determine the interface between the ligand and the large protein using differential line broadening (9). The residues affected by complex formation can be used as ambiguous restraints. 2. If the ligand is a flexible peptide, use transferred NOESY (10) to determine the structure of the peptide in the complex. Determine the structure of the complex by using a CNS-based software, such as HADDOCK (11) or Xplor-NIH (12). Use the crystal structure of the large protein to define the starting conformation, and both unambiguous and ambiguous restraints obtained from NOE, PRE, line broadening, and chemical shift perturbation experiments.

4. Notes 1. Freshly transformed colonies always give better protein yield. 2. [U-2H,13C]-glucose should be used when uniform 13C labeling is desired or when all side-chain carbons of the methyl-bearing residues are to be 13C labeled for magnetization transfer from methyls to the backbone. In this case, the uniformly 13C-labeled ketoacid precursor must be used. 3. [U-2H,12C]-glucose should be used to produce an NMR sample in which all carbons are 12C labeled, except the methyl carbons of interest. 1H-13C HMQC spectra of such samples are recorded without the use of the constant time version and typically provide the best resolution. Such a sample can also be used for relaxation experiments (1). 4. Up to ~2.5% of a rich labeling medium can be used to increase the protein yield with no effect on the specific labeling of the methyl groups.

8

NMR Studies of Large Protein Systems

139

5. Precursors can be purchased in protonated form and dissolved in D2O for exchange to take place: at pH 12.5 (45°C), 2–3 h for α-ketoisovalerate, and at pH 10.5 (45°C), 12–14 h for α-ketobutyrate; the pH values are optimized for exchange and prevent the generation of dimers through condensation of two ketoacid molecules. 6.

CH3-2Hα-alanine can be prepared by using the tryptophan synthase enzyme to catalyze the proton-to-deuterium exchange of the α hydrogen, as described by Matthews and coworkers (13). 13

7. Incorporation of 13CH3/12CD3 isotope labels into the isopropyl moieties of Val and Leu residues should be used for very large proteins since the inter-methyl dipolar relaxation is significantly reduced. The methyl-TROSY spectra show significant gains in resolution with practically no losses in sensitivity despite the twofold dilution of the NMR-active methyls in such samples. Precursors have also become available that allow any of the methyl isotopomers (13CHD2, 13CH2D, and 13CH3) to be incorporated into the protein (1). The different isotopomers can be used for relaxation experiments. 8. In this case, addition of ~2% of a rich labeling medium (e.g., BIOEXPRESS or ISOGRO) is required to suppress scrambling associated with the addition of the alanine amino acid. Interestingly, further increase of the rich labeling medium (~10%) suppresses completely the methyl labeling of Leu while having a minimal effect on the methyl labeling of Val (Fig. 1b). Since the methyl groups of these two residues often overlap, this labeling scheme can be used to differentiate between the two. 9. Critical step: Excessively prolonged growth after induction should be avoided to prevent generation of methyl groups with undesired isotopomers. 10. The design of domains and fragments thereof that would retain their fold and are soluble in isolation can be quite tricky. In this respect, the availability of a crystal structure can be of tremendous help. 11. When the methyl residues of all five residues are labeled in the same sample of a very large protein, the signal may be significantly compromised due to enhanced inter-methyl relaxation. The preparation of multiple samples each containing a single amino acid labeled may be desirable in such a case. 12. An arsenal of pulse sequences are available for methyl assignment (14). 13. The highly deuterated background suppresses spin diffusion, and thus the mixing time for the NOESY experiments can be set as high as 500 ms allowing for NOEs to be observed between methyl groups as far as ~8 Å.

140

S.-R. Tzeng et al.

14. Amino acids should be typically substituted by an isosteric amino acid to prevent significant changes in the local environment and protein packing, which could introduce significant chemical shift effects. 15. Nonreactive Cys residues can be judged by the Elman’s test. Sites for MTSL incorporation should be selected so that they cause no or minimal effect on protein structure. This can be assessed by NMR. 16. PRE rates should typically be measured using several MTSLderivatized ligands, each containing a single MTSL at a different site. Because PRE rates provide long-range distance information, in the absence of available NOE data a large number of PREs are required to properly determine the structure of a protein– ligand complex. The complex between SecA and the signal peptide was determined using 160 PRE-derived intermolecular restraints. References 1. Ruschak, A. M., and Kay, L. E. (2010) Methyl groups as probes of supra-molecular structure, dynamics and function. J. Biomol. NMR 46, 75–87. 2. Goto, N., Gardner, K., Mueller, G., Willis, R., and Kay, L. (1999) A robust and cost-effective method for the production of Val, Leu, Ile (delta 1) methyl-protonated 15N-, 13C-, 2H-labeled proteins. J. Biomol. NMR 13, 369–374. 3. Tugarinov, V., Kanelis, V., and Kay, L. E. (2006) Isotope labeling strategies for the study of highmolecular-weight proteins by solution NMR spectroscopy. Nat. Protoc. 1, 749–754. 4. Tugarinov, V., Hwang, P., Ollerenshaw, J., and Kay, L. (2003) Cross-correlated relaxation enhanced 1H-13C NMR spectroscopy of methyl groups in very high molecular weight proteins and protein complexes. J. Am. Chem. Soc. 125, 10420–10428. 5. Sprangers, R., and Kay, L. E. (2007) Quantitative dynamics and binding studies of the 20S proteasome by NMR. Nature 445, 618–622. 6. Gelis, I., Bonvin, A., Keramisanou, D., Koukaki, M., Gouridis, G., Karamanou, S., Economou, A., and Kalodimos, C. G. (2007) Structural basis for signal-sequence recognition by the translocase motor SecA as determined by NMR. Cell 131, 756–769. 7. Battiste, J., and Wagner, G. (2000) Utilization of site-directed spin labeling and high-resolution heteronuclear nuclear magnetic resonance for global fold determination of large proteins with

limited nuclear overhauser effect data. Biochemistry 39, 5355–5365. 8. Tang, C., Schwieters, C., and Clore, G. (2007) Open-to-closed transition in apo maltose-binding protein observed by paramagnetic NMR. Nature 449, 1078–1082. 9. Takeuchi, K., and Wagner, G. (2006) NMR studies of protein interactions. Curr. Opin. Struct. Biol. 16, 109–117. 10. Post, C. (2003) Exchange-transferred NOE spectroscopy and bound ligand structure determination. Curr. Opin. Struct. Biol. 13, 581–588. 11. de Vries, S. J., van Dijk, M., and Bonvin, A. M. (2010) The HADDOCK web server for data-driven biomolecular docking. Nat. Protoc. 5, 883–897. 12. Schwieters, C. D., Kuszewski, J. J., Tjandra, N., and Clore, G. M. (2003) The Xplor-NIH NMR molecular structure determination package. J. Magn. Reson. 160, 65–73. 13. Isaacson, R., Simpson, P., Liu, M., Cota, E., Zhang, X., Freemont, P., and Matthews, S. (2007) A new labeling method for methyl transverse relaxation-optimized spectroscopy NMR spectra of alanine residues. J. Am. Chem. Soc. 129, 15428–15429. 14. Tugarinov, V., and Kay, L. (2003) Ile, Leu, and Val methyl assignments of the 723-residue malate synthase G using a new labeling strategy and novel NMR methods. J. Am. Chem. Soc. 125, 13868–13878.

Chapter 9 Protein Dynamics by 15N Nuclear Magnetic Relaxation Fabien Ferrage Abstract Nitrogen-15 relaxation is the most ubiquitous source of information about protein (backbone) dynamics used by NMR spectroscopists. It provides the general characteristics of hydrodynamics as well as internal motions on subnanosecond, micro- and millisecond timescales of a biomolecule. Here, we present a full protocol to perform and analyze a series of experiments to measure the 15N longitudinal relaxation rate, the 15N transverse relaxation rate under an echo train or a single echo, the 15N–1H dipolar cross-relaxation rate, as well as the longitudinal and transverse cross-relaxation rates due to the cross-correlation of the nitrogen-15 chemical shift anisotropy and the dipolar coupling with the adjacent proton. These rates can be employed to carry out model-free analyses and can be used to quantify accurately the contribution of chemical exchange to transverse relaxation. Key words: Nuclear magnetic resonance, Protein dynamics, Relaxation rates, Nitrogen-15, Longitudinal relaxation, Transverse relaxation, Cross-relaxation, Cross-correlated relaxation, Chemical exchange

1. Introduction Nuclear magnetic resonance is a fantastic tool to investigate the dynamics of biomolecules and, in particular, proteins. Among the numerous techniques available, which provide access to internal motions over a wide range of timescales, measurements of backbone nitrogen-15 relaxation rates have proven to be by far the most popular method to sample and quantify protein dynamics (1–3). Nitrogen-15 relaxation rates can be analyzed to determine the hydrodynamic properties of proteins (4), internal motions faster than overall motions (subnanosecond) as well as slower motions on microsecond–millisecond timescales that give rise to chemical exchange-induced relaxation. Fast subnanosecond motions are closely related to the local atomic density of a protein, and they

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_9, © Springer Science+Business Media, LLC 2012

141

142

F. Ferrage

contribute to conformational entropy (5, 6) and are a good indication of the local malleability of the protein structure. Hydrodynamic properties can be used for structural refinement of a single-domain (7) or tight multidomain protein or protein complex (8), interdomain motions may also be analyzed (9). Microsecond and millisecond motions and chemical reactions are essential to protein function. In most cases, the kinetics of these events can be determined by nitrogen-15 relaxation methods (10, 11), while, in favorable cases, both thermodynamics and structural changes between exchanging states can be characterized (12). The canonical set of nitrogen-15 relaxation experiments comprises the measurements of longitudinal, R1, and transverse, R2, auto-relaxation rates, as well as the 15N–{1H} nuclear Overhauser effect, from which the dipolar cross-relaxation rate between the nitrogen-15 nucleus and its neighboring proton, sNH, can be extracted (13). In this review, I present basic protocols to set up and analyze these experiments, including the latest developments. In addition, we provide protocols for two additional experiments for measuring longitudinal and transverse cross-relaxation rates due to the cross-correlation of the nitrogen-15 chemical shift anisotropy (CSA) and the dipole–dipole (DD) coupling with its attached proton. These rates can be analyzed alongside the canonical set in order to evaluate the contribution of chemical exchange effects to transverse relaxation (14, 15). 1.1. Theory

We provide here the minimal theoretical background that is necessary to define the various terms that are used in this review. For a detailed understanding of relaxation theory, the reader should refer to some of the many reviews that have been published on the subject (16–19). Relaxation is the irreversible process through which a spin system evolves toward a steady state. Different elements of the density operator evolve with different auto-relaxation rates, and they convert into one another with well-defined cross-relaxation rates. The molecular processes underlying relaxation are the fluctuations of high-amplitude orientation-dependent interactions, such as dipole–dipole couplings and the anisotropy of the chemical shift. The amplitude of these interactions is known so that one can extract the characteristics of motions from relaxation rates. Relaxation rates do not depend directly on the correlation function of interaction Hamiltonians but on their Fourier transform: the spectral density function J(w). Here, we describe a protocol to measure the following relaxation rates for 15N nuclei in a 15N–1H pair. The longitudinal auto-relaxation rate, R1: R1 = D (6(1 + a 2 ) J (w N ) + 2 J (w H - w N ) + 12 J (w H + w N )). (1) The transverse auto-relaxation rate, R2:

9

Protein Dynamics by 15N Nuclear Magnetic Relaxation

143

R2 = D (4(1 + a 2 ) J (0) + 3(1 + a 2 ) J (w N ) + J (w H - w N ) + 6 J (w H ) + 6 J (w H + w N )).

(2)

The 1H–15N dipolar cross-relaxation rate, sNH: s NH = D (-2 J (w H - w N ) + 12 J (w H + w N )).

(3)

The CSA/DD transverse cross-correlated cross-relaxation rate, dtN: (4)

dtN = D a(8 J (0) + 6 J (w N )).

The CSA/DD longitudinal cross-correlated cross-relaxation rate, dlN: dlN = 12D aJ (w N ),

(5) 3 rNH

2

2 2 2 with D = 1 æ m0 ö g H g N and a = - 2 B0 (s / / - s^ ) 6 20 çè 4p ÷ø rNH 3g H

; gH and gN

are the gyromagnetic ratios of the proton and nitrogen-15 nuclei, respectively; wH and wN are the Larmor angular frequencies of the proton and nitrogen-15, respectively (see Note 1); rNH is the internuclear distance; m0 is the permeability of free space; ħ is the Planck constant divided by 2p; B0 is the static magnetic field and s// and s ^ are the axial and perpendicular components of the anisotropic chemical shift tensor of the nitrogen-15 nucleus (which we consider to be axially symmetric). The 1H–15N dipolar cross-relaxation rate is measured indirectly from nuclear Overhauser effects. In this case, signal intensities are measured under two conditions: at the steady state under effective proton saturation Iss and at equilibrium Ieq. The ratio of these intensities is: (6) I ss / I eq = 1 + (g H s NH ) / (g N R1 ) In addition, we measure the transverse relaxation rate R2 under two different conditions, the more typical CPMG (Carr-PurcellMeiboom-Gill) train of echoes and by a single echo combined with continuous 1H composite pulse decoupling. The later rate includes, in most cases, the full contribution of chemical exchange to transverse relaxation, Rex, which is a probe of chemical reactions and motions on micro- to millisecond timescales (20, 21), a particularly relevant range for biological processes. In the case of fast exchange between two sites A and B (with a timescale tex and a difference of chemical shift between the two sites Dw), we have: Rex = p A pB Dw 2 t ex ,

(7)

where pA and pB are the populations of the two exchanging sites A and B respectively, pA + pB = 1. The focus of this review is not to describe how to determine the various parameters of the exchange process, which is well

144

F. Ferrage

presented in the following reviews (10, 22). Nevertheless, we describe a protocol to identify the presence of chemical exchange.

2. Materials 1. Appropriately labeled protein, preferably (2H, 15N) or (15N) [while studies of (2H, 15N, 13C) and (15N, 13C) labeled proteins are also possible] sample at ~1 mM concentration (see Note 2). 2. NMR spectrometer with a magnetic field strength of ³11.7 T (see Note 3). 3. nmrPipe: software to process, display and analyze 2D spectra (http://spin.niddk.nih.gov/NMRPipe/). 4. Curvefit: software to fit the relaxation decays and build-up curves (http://biochemistry.hs.columbia.edu/labs/palmer/ software/curvefit.html). 5. Grace: software to display the results of the Curvefit analysis (http://plasma-gate.weizmann.ac.il/Grace/).

3. Methods 3.1. Preliminary Set-up

A few preliminary procedures should be followed before starting the series of experiments. First, when all experiments are run on the same spectrometer, one should ensure the temperature calibration is very recent, in any case, it is safer to calibrate the temperature prior to recording all experiments. When data are collected by using two or more spectrometers, the exact match of temperatures should be verified directly on the same standard methanol sample on each spectrometer. It is more important to run all experiments at the exact same temperature than to have a small error on the nominal temperature of the experiments. 1. When running experiments on the first spectrometer, introduce a sample of perdeuterated methanol into the spectrometer. Set the temperature at the desired value according to the latest calibration. Let the sample equilibrate for 5 min. Match and tune the probe, and shim the magnet. 2. Run a simple 1D proton experiment with a small angle excitation (typically with a 1 ms 1H pulse at high power). Measure the difference in chemical shifts between the hydroxyl and methyl protons, Dd. The temperature is (23): T = -16.7467 Dd 2 - 52.5130 Dd + 419.1381.

(8)

9

Protein Dynamics by 15N Nuclear Magnetic Relaxation

145

3. Change the nominal temperature on the spectrometer and repeat step 2 until the desired temperature is reached. Write down the value of Dd. 4. When running experiments on another spectrometer, directly match the value of Dd. Second, pulses should be calibrated for each set of measurements collected on each spectrometer. The typical values obtained by using a standard sample should be used as a guess to start calibration but not directly to run experiments. 1. Introduce the protein sample in the spectrometer, let it equilibrate for a few minutes. Shim, tune, and match each channel that is going to be used. Set the carrier on the water signal and run a 1D proton spectrum with a short (1 ms) excitation at high power (see Note 4). Phase and use this spectrum as a reference. 2. Set the pulse duration to the expected value for a 360° flip angle. Collect a spectrum. Optimize the value of the pulse duration to get a null signal. Write down the duration of the corresponding 90° pulse. 3. Prepare a 1D version of a 15N–1H Heteronuclear Single Quantum Coherence (HSQC) experiment, preferably with a heteronuclear gradient echo. Set the calibrated duration for proton pulses and the expected value for 15N pulses. Make sure the carrier on the 15N channel is set at 117 ppm. Set the number of scans to 16. Run a 1D reference spectrum. If the signalto-noise is very high, reduce the number of scans. If it is low, increase accordingly. 4. Now, set the duration of the first 90° pulse on 15N to the expected duration for a 180° pulse. Run the experiment. Optimize to get a null signal. Write down the corresponding value for the calibrated 90° pulse. 5. If the sample is carbon-13 labeled, follow steps 3 and 4 with 13 C instead of 15N. The 13C carrier should be set at 35 ppm and the proton signal should be null between 2 and 3 ppm when the flip angle is 180°. 6. Alternatively, or, when using a carbon-13 and perdeuterated sample, run steps 3 and 4 with a 1D version of an HNCO experiment, where the first 13C pulse is adjusted to 180°. The carrier on the 13C channel should be set to 174 ppm (see Note 5). Third, the carrier, spectral width, and number of time points should be optimized in the 15N dimension (this should have been done during the assignment experiments). In particular, cross-correlated relaxation experiments on small proteins can have a long duration because of long phase cycles,

146

F. Ferrage

but are not signal-to-noise limited. Such optimization is an important time saver. 1. Set up a 2D HSQC experiment with a wide spectral width (larger than 30 ppm) so that only the arginine NeHe signals are folded. 2. Process the spectrum and adjust the spectral width and carrier. Folded peaks should not overlap with nonfolded peaks, and it is advised to keep about 1 ppm on each side of the spectrum with no peak. This process may require a few iterations. In the end, a spectral width between 16 and 24 ppm should be obtained for most small- or medium-size proteins. 3. This last step should be carried each time one runs experiments at a new B0 field. Record a final spectrum with the optimal carrier and spectral width as well as a large number of points in the indirect dimension. Process it with the lowest number of points that provides good resolution of all peaks of interest. Use this value for all experiments. Be careful not to underestimate the number of points, it may save 10% of the time but you may also have to repeat the full series of experiments. 3.2. Auto-Relaxation Rates Measurements 3.2.1. Longitudinal Relaxation Rate

The sequences presented in Fig. 1a, b should be employed for this experiment. There is no fully satisfactory combination of water-flip back schemes and suppression of CSA/DD cross-correlated crossrelaxation pathways during the relaxation delay. One solution is to saturate the water resonance in each scan and use a long recovery delay between scans. To saturate the water resonance, drop the water-flip back pulse at the end of the first INEPT in Fig. 1a and use a strong gradient G2. In that case, the two shaped pulses in Fig. 1b can be substituted either by composite pulse decoupling (see Fig. 1c) or a series of proton 180° pulses (see Fig. 1d). Here, we use a scheme that can be used when no amide resonance is lying too close to the water resonance. It is one of the standard sequences available on Bruker spectrometers. 1. Set pulse durations and amplitudes according to the spectrometer specifications, the calibration and the description of the pulse sequence in Fig. 1a, b. Set delays according to the pulse sequence. Calibrate proton shape pulses according to the spectrometer-specific protocol. 2. Set the recovery delay between scans to a large value, usually 2 s; 3 s can be used if spectrometer time is not a problem. 3. For a well-behaved protein, start with eight scans (16–32 for low-concentration samples). Set the relaxation delay to a low value (we use 20 ms). Run the first 1D experiment. Phase and save this spectrum. 4. Set the relaxation delay to a value close to the expected average T1. At 300 K, this is 500 ms for a globular 100-residue protein and 700 ms for a 150-residue protein. Acquire a new spectrum.

9

Protein Dynamics by 15N Nuclear Magnetic Relaxation

147

Fig. 1. Pulse sequences used for measuring 15N auto-relaxation rates. (a) General scheme. (b) Relaxation sequence for measuring the longitudinal relaxation rate R1; (c) relaxation sequence for measuring the transverse relaxation rate under a single echo R2echo; (d) relaxation sequence for the measurement of the transverse relaxation rate under a CPMG echo train R2CPMG. All narrow (filled) and (wide ) open rectangles represent 90° and 180° pulses, respectively. Pulse phases are along the x-axis of the rotating frame unless otherwise mentioned. Proton composite pulse decoupling during the delay t1 was performed with a GARP scheme and a radio frequency field (rf) amplitude of 1 kHz. The 13C channel pulse was a 500 ms smoothed CHIRP pulse (37), with a sweep of 60 kHz on a 600 MHz spectrometer; the carrier at the center of the pulse was 110 ppm. Composite-pulse decoupling during acquisition was performed on the 15N channel with a GARP scheme (38) and an rf amplitude of 1,090 Hz. The delay ta is 2.56 ms; the delay tb can be adjusted around 5 ms. The phase cycles were: f1 = {x, −x }; f3 = {x, x, −x, −x }; f4 = {x, x, −x, −x }; f5 = {−y, −y, y, y }; facq = {x, −x, −x, x }. When relaxation block (b) is used, f2 = { y, y, y, y, −y, −y, −y, −y } and facq = {x, −x, −x, x, −x, x, x, −x }. The amplitude profile of the pulsed field gradient was a sine bell shape. Their durations and peak amplitudes over the x, y, and z orientations (when triple axis gradients are available) were, respectively: G1; 600 ms, 9.5 G/cm, 9.5 G/cm, 0; G2; 1 ms, 0, 0, 30 G/cm; G3; 600 ms, 15 G/cm, –15 G/cm, 0; G4; 1 ms, 0, 0, 40 G/cm; G5; 1 ms, 0, 0, 8.1 G/cm. Coherence selection was achieved by inverting the amplitude of the gradient G4 and phase f1. (b) The carrier is placed at 8.2 ppm during the relaxation block; gray bell-shaped pulses are 1.6 ms Q3 Gaussian cascade pulses at 600 MHz (39) (see Note 9). (c) WALTZ-16 decoupling should be used for 1H decoupling during the relaxation block (40). See text for how to choose the relaxation delays. (d) Gray rectangles are 180° pulses, depending on the spectrometer and probe, they should be either at high power or less, but should not be longer than 100 ms. t should be set to 500 ms.

148

F. Ferrage

5. Compare the intensities and repeat with adjusted delays until the ratio of intensities is about 0.3. We call this delay tmax + 20 ms. 6. The best sampling of the decay is achieved with an even decay of intensities between each time point. If n different relaxation delays are to be acquired, the intensity of the j-th experiment should be: I j = I 1 (1 - 0.7( j - 1) / (n - 1)).

(9)

We expect a mono-exponential decay with an average rate R1av and we know that:

(

)

exp -R1av t max = 0.3 Û R1av = - ln 0.3 / t max

(

)

(10)

and exp -R1avt j = I j .

(11)

t j = (t max / ln0.3)ln(1 - 0.7( j - 1) / (n - 1)).

(12)

So that

7. A typical number of relaxation delays is 8. This number can be reduced to 6, even 4 in cases where each spectrum takes a very long time to acquire. Record all experiments in an interleaved manner, changing the relaxation time once each full 1D spectrum is acquired. This requires a small modification of standard pulse sequence programs. When using standard pulse sequences, the sequence of delays should be: t1, tn, t2, tn−1, t3, tn−2, etc., so that the decay of intensities would not appear mono-exponential if the quality of the spectra was to decay, because of a bubble for instance, which would otherwise be reflected by a higher, erroneous rate. Similarly, at least t1 and, if possible t3 should be repeated to identify a possible decay in the quality of spectra. If this decay is larger than the error bars, the dataset should be discarded. 8. Before starting the series of experiments, set the number of “dummy scans” in the first experiment to a very high value (at least 256, preferably 512). This leaves enough time for the temperature control system of the spectrometer to reach equilibrium. 9. Import the spectra into nmrPipe (24). Process them with typical parameters, the window function should be a sine-bell function, the power should be set to 2 and the shift parameter between 0.5 and 0.35 in each dimension (the lower the value, the more resolution). Truncation artifacts in the 2D spectrum should be avoided at all cost, as they may contaminate the intensity of neighboring peaks. 10. Peak pick the first spectrum, or export a peak list from the assignment software, save the peak list as peakX_0.tab, where X is the number of the spectrum. To extract the intensity of the peaks with the best possible accuracy, use the nonlinear line

9

Protein Dynamics by 15N Nuclear Magnetic Relaxation

149

shape analysis tool nlinLS, which is provided with nmrPipe. First, measure the full width, in points, of the peaks in each dimension: WH and WN. Then, run the following line: nlinLS -in peakX_0.tab -out peakX_1.tab -data spectrumX.ft2 -w WH/6 WN/6 -delta X_AXIS WH/10 Y_AXIS WN/10. This supposes that the proton dimension is the X dimension of the spectrum. The option w defines the region, where the peak is fitted (the current command specifies a square of dimensions WH and WN); the option delta adds a constraint on the following parameters, here the position of each peak. 11. It is likely that nlinLS will initially provide a series of error messages, most often indicating bad convergence of the fit. In spite of these error messages, the output file will be of better quality than the input file and should be used as an input for a second round of analysis with nlinLS. The process is iterative and may require up to five iterations for convergence and a proper fit with no error message. If the width of a peak is diverging from typical values in the output file, an additional constraint can be added to the peak width parameters, WX and WY. Peaks may also be added or excluded from clusters to improve convergence. Use the final, error-free list obtained from the first spectrum as the first input for all other fits. 12. Run this iterative procedure for each spectrum. 13. Collect the intensities and peak numbers with the getCols tool, which is provided with nmrPipe. nlinLS provides both peak intensities and volumes. Only the intensities should be used. Although the precision should be better than the spectral noise, this turns out to be a good estimate of the error in the intensities. Use this value as the error for all the peaks of each spectrum. 14. Using a short script or a spreadsheet editor, prepare a table with peak numbers in the first column, then the intensities of peaks in the series of spectra in columns 2, 4, 6, etc., and the respective error in columns 3, 5, 7, etc. 15. Run a short script to create a series of Curvefit input files with names: N.in, where N is the peak number (see Note 6). 16. Run the command: batch_curve in. The fit for each peak number appears (see Fig. 2). It is very important to check every single fit and identify errors that may come from improper intensity fits or other potential errors. Write down the peaks and spectra that lead to apparently wrong intensities and inspect the spectra for potential errors. There may be no error. 3.2.2. Transverse Relaxation Rates with a Single Echo

The sequence presented in Fig. 1b should be employed for this experiment. Only relaxation delays that lead to full cycles of the proton composite pulse decoupling (CPD) should be used, which modifies slightly the protocol from the measurement of longitudinal relaxation rates.

150

F. Ferrage

Fig. 2. Example of a longitudinal relaxation decay curve as shown by Grace during the Curvefit procedure.

1. Most parameters are identical to those used in longitudinal relaxation measurements. To cover the entire amide region the carrier for the proton CPD is set to 8.2 ppm and the amplitude to about 2 × x Hz; where x is the proton Larmor frequency of the spectrometer in MHz. Set the CPD scheme to WALTZ-16 (see Note 7). The half relaxation delay T should be set to multiples of 96 × tcpd, where tcpd is the duration of a 90° pulse at the decoupling power. 2. Contrary to other sequences, proton CPD with a carrier on the amides prevents a proper control of the polarization of water. Set the radiofrequency amplitude to zero for the water flip back pulse following the first INEPT and use a long recovery delay (at least 2 s) in between scans. 3. Follow steps 2–7 of the protocol for longitudinal relaxation (Subheading 3.2.1) with the following changes. Start with the shortest relaxation delay, i.e., zero. The fact that the total relaxation delay has to be a multiple of 192 × tcpd, severely limits the choice of possible delays, use all accessible delays between zero and tmax. 4. To obtain the transverse relaxation rate under a single echo R2echo, the analyses of spectra are exactly the same as that described in steps 8–15 of the preceding protocol (Subheading 3.2.1).

9 3.2.3. Transverse Relaxation Rates with a CPMG Scheme

Protein Dynamics by 15N Nuclear Magnetic Relaxation

151

For the sake of consistency, this protocol is described here. However, this is the most challenging experiment as heating from high radiofrequency fields may lead to bubble formation and sample degradation. It is strongly advised to run this experiment as the last one of the series. 1. Most parameters are identical to those used in longitudinal relaxation measurements. Set the power level for 15N 180° pulses during the CPMG echo train. This experiment is one of the most demanding that can be run on a high-resolution probe. The maximum power (i.e., shortest nitrogen-15 pulses) that can be employed during a CPMG echo train is usually provided to users for each probe on each spectrometer. If no value is recommended, ask the person in charge of the spectrometer maintenance. If you are in charge of the spectrometer maintenance, ask the manufacturer. Often, particularly on cryogenic probes, these pulses should not be applied at full power. For proper accuracy of the experiment, 180° pulses should be no longer than 100 ms. 2. In addition to Fig. 1d, include a temperature-compensation loop at the beginning of the sequence. At the end of the recycling delay, include a series of far off-resonance (200 kHz works fine) 15N 180° pulses at the amplitude of the CPMG echo train. Do not include 1H 180° pulses as these would alter the 1 H longitudinal polarization. The duration of this train should be such that the total number of 15N 180° pulses is constant whatever the relaxation delay is. An additional delay, equal to the relaxation delay, should also be added so that the total duration for the recovery of proton polarization is constant. 3. Follow steps 2–7 of the protocol for longitudinal relaxation (Subheading 3.2.1). Zero can be used as the shortest relaxation delay. 4. Use a long recovery delay (preferably 2 s) and a large number of dummy scans (512) to let the temperature control system reach equilibrium before the first experiment. 5. Follow steps 9–16 of the R1 analysis (Subheading 3.2.1) to obtain transverse relaxation rates under a CPMG train, R2CPMG.

3.3. 15N–{ 1H} Nuclear Overhauser Effect Measurements

The sequence presented in Fig. 3 should be used for this experiment. It displays a series of improvements introduced recently (25–27). The two experiments (under effective proton saturation and at equilibrium) have to be run in an interleaved manner. 1. Set up the saturation scheme (Fig. 3b). The saturation element is symmetric, 1H pulses have a 180° flip angle, the carrier is placed in the amide region (8.2 ppm), the interpulse delay tNOE should be a multiple of 1/JNH, where JNH is the 1H–15N one-bond scalar coupling constant. In globular proteins at low pH, tNOE = 22 ms

152

F. Ferrage

Fig. 3. Pulse sequence used for recording steady-state 15N–{1H} nuclear Overhauser effects. For each measurement, reference and steady-state experiments have to be recorded in an interleaved manner. In the reference experiment, one should run the part of the pulse program displayed in box (a). At the end of the recovery delay TNOE = 10 s (or more), the proton carrier is placed on resonance with the water signal and a very selective water-flip back pulse is applied (3 ms sinc shaped or longer). To record steady-state experiments, the boxed sequence in (a) is substituted by the scheme shown in (b) for the effective saturation of amide proton resonances. After an optional delay T¢NOE = 2 s for stable detection of the lock signal, the proton carrier is placed in the center of the amide region (at 8.2 ppm) as shown by the arrow labeled by N. The motif [delay tNOE/2 – 180° pulse – delay tNOE/2] is repeated nNOE times. The interpulse delay, tNOE, is typically 22 ms (11 ms may also be used, see text). The rf amplitude for the pulses should be 7.5 kHz at a 500 MHz Larmor frequency and 9 kHz at 600 MHz. A gradient G1 is applied at the end of the last tNOE/2 delay to suppress all transverse components of the proton polarization. The carrier was moved on-resonance with the water signal as indicated by the W arrow. The number of cycles, nNOE, was set so that the total duration for effective saturation was 4 s. All narrow (filled ) and wide (open) rectangles represent 90° and 180° pulses, respectively. Pulse phases are along the x-axis of the rotating frame unless otherwise mentioned. Proton composite pulse decoupling during the delay t1 was performed with a GARP scheme and an rf amplitude of 1 kHz. Composite-pulse decoupling during acquisition was performed on the 15N channel with a GARP scheme (38) and an rf amplitude of 1,090 Hz. The delay ta is 2.56 ms. The phase cycles were: f1 = {y, –y}; f2 = {x, x, –x, –x}; f3 = {x, x, –x, –x }; f4 = {−y, –y, y, y }; facq = {x, –x, –x, x }. The amplitude profile of the pulsed field gradient was a sine bell shape. The durations and peak amplitudes over the x, y, and z orientations were, respectively: G1; 600 ms, 15 G/cm, 15 G/cm, 0; G2; 1 ms, 0, 0, 25 G/cm; G3; 1 ms, 0, 0, 40 G/cm; G4; 1 ms, 0, 0, 8.1 G/cm. Coherence selection was achieved by inverting the amplitude of the gradient G3 and phase f4.

is a safe value, but tNOE = 11 ms should be used in disordered proteins or on high pH samples, where proton exchange with the solvent is faster. The amplitude of these radiofrequency pulses should be 1.5 × X kHz, where X is the 1H Larmor frequency in MHz divided by 100 (e.g., 9 kHz on a 600 MHz spectrometer). 2. Set up the duration of the saturation scheme. The total duration of the saturation is linked to the 15N R1 rates, but not to the 1H relaxation rates, since the present scheme ensures an effective saturation of protons after each saturation element. If the longitudinal relaxation experiments have been fully analyzed, identify the lowest value of all 15N R1 rates. Set the value of nNOE so that the total duration of the saturation scheme is 4/R1. If only an estimate of the average longitudinal relaxation rates R1av is available from a comparison of 1D spectra (steps 3–5 of the protocol for the measurement of R1 rates; Subheading 3.2.1),

9

Protein Dynamics by 15N Nuclear Magnetic Relaxation

153

set the total duration of saturation to at least 6/R1av and preferably 8/R1av, if time permits. 3. Set up the reference experiment (Fig. 3a). The soft water-flip back pulse should be long (at least a 3 ms sinc-shaped pulse, and longer for spectrometers with a Larmor frequency under 700 MHz) as it should not touch amide resonances closest to the water resonance. The total duration of the recovery delay, TNOE, should be long. In this case, the spin system should be at equilibrium, including the protons. The use of a water-flip back scheme permits the use of significantly shorter recovery delays. In general, these delays should be no shorter than 10 s. It is strongly advisable to spend up to an hour to compare the intensity of the first 1D spectrum (obtained with t1 = 0) for various values of TNOE and identify for which value of TNOE the signal has reached its maximum value. Since the signal-to-noise ratio will be too low to properly evaluate the difference between 97% and 100% of the signal, add at least 2 s to the value of TNOE where the signal appears to have reached its maximum value. 4. The total number of scans should be at least 24 and up to 64. The experiment should last between 12 and 24 h. If longer acquisition times are necessary, reduce the number of scans for each experiment and run a series of experiments no longer than 24 h. The number of dummy scans should not be greater than 64. 5. Follow steps 9–14 of the protocol for longitudinal relaxation (Subheading 3.2.1) to obtain intensities. 6. Compute the NOE ratio (Eq. 6). Although it may not be necessary for the following analysis, derive the dipolar crossrelaxation rate from the NOE ratio using Eq. 6. 3.4. Measurements of CSA/DD CrossCorrelated CrossRelaxation Rates

These experiments are less commonly run in spite of their great utility. The set up turns out to be straightforward after more typical relaxation experiments have been run. 1. Set pulse durations and amplitudes according to the spectrometer specifications, the calibration and the description of the pulse sequence, which is shown in Fig. 4. Set delays according to the pulse sequence. Calibrate proton shape pulses according to the spectrometer-specific protocol. Most of this part has already been set up in the preceding experiments. 2. For both experiments, the minimum number of scans imposed by the phase cycle is higher, leading to very long experimental times when high resolution data needs to be collected (large number of t1 points). On the other hand, shorter recycle delays can be used, typically between 1 and 1.5 s with little loss of sensitivity and no loss of accuracy. 3. These experiments use symmetrical reconversion (15, 28). This means that all four relaxation pathways in a two-operator space are detected. These four experiments have to be run in an

154

F. Ferrage

Fig. 4. Pulses sequences for measuring CSA/DD cross-correlated cross-relaxation rates. (a) General scheme; (b) relaxation block for measuring the transverse cross-relaxation rate; (c) relaxation block for measuring the longitudinal cross-relaxation rate. Only specific elements are detailed here. The delay d is equal to the shortest value of t1 so that the first effective value of t1 is zero. The phase cycles were: f1 = { y, –y }; f2 = {x, x, x, x, –x, –x, –x, –x}; f3 = {x, x, x, x, x, x, x, x, –x, –x, –x, –x, –x, –x, –x, –x}; f4 = {x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, –x, –x, –x, –x, –x, –x, –x, –x, –x, –x, –x, –x, –x, –x, –x, –x}; f5 = {x}; f6 = {x, x, –x, –x}; f¢6 = {x, x, y, y}; facq = {x, –x, –x, x, –x, x, x, –x, –x, x, x, –x, x, –x, –x, x, –x, x, x, –x, x, –x, –x, x, x, –x, –x, x, –x, x, x, –x}. Gradient durations and peak amplitudes over the x, y, and z orientations were, respectively: G1; 600 ms, 15 G/ cm, 15 G/cm, 0; G2; 600 ms, 6.5 G/cm, 0, 0; G3; 2 ms, 0, 0, 40 G/cm; G4; 1 ms, –9.5 G/cm, 9.5 G/cm, 0 G/cm; G5; 600 ms, 3.5 G/cm, 0 G/cm, 14.5 G/cm; G6; 1 ms, –35 G/cm, –35 G/cm, –35 G/cm; G7; 1 ms, 35 G/cm, 35 G/cm, 35 G/cm. Frequency sign discrimination was performed using States-TPPI.

9

Protein Dynamics by 15N Nuclear Magnetic Relaxation

155

interleaved manner (see Note 8). Remember to set the number of points in the indirect dimension to the proper value (i.e., eight times the number of complex time points). 3.4.1. Transverse Cross-Correlated Cross-Relaxation Rates

1. Set up the relaxation delay(s). The limiting factor for sensitivity is the intensity in the cross-relaxation experiments (pathways II and III in Fig. 4a). This intensity is maximum when the relaxation delay is equal to the auto-relaxation time. The optimal relaxation delay typically lies between 30 and 80 ms. Explore this interval in 20 ms steps. The intensity versus time curve is usually flat around the maximum. When two relaxation delays appear to give the same maximum intensity, first check a relaxation delay at the midpoint between the two. If the intensity is better, keep this delay. If the intensity is the same, choose the shortest delay so the intensity in the auto-relaxation experiments will be better. 2. This experiment can be run with a single relaxation delay. If time permits, repeat the optimal relaxation delay and run another delay with sufficient intensity on the cross-relaxation experiments. 3. Follow steps 9–14 of the longitudinal relaxation protocol (Subheading 3.2.1) to obtain the intensities of the four interleaved experiments. The following quantity should be computed: S (4T ) = I II (4T )I III (4T ) / I I (4T )I IV (4T ).

(

)

S (4T ) = tanh dNt 4T .

(13) (14)

4. If only one relaxation delay is recorded, simply invert Eq. 14. If several time points are recorded, compile the values of S and the corresponding errors in a text file and follow steps 15 and 16 of the longitudinal relaxation protocol (Subheading 3.2.1). The header of each N.in file should set the hyperbolic tangent mode for the fit to evaluate dtN . 3.4.2. Longitudinal Cross-Correlated Cross-Relaxation Rates

1. Set up the relaxation delay(s). The constraints are similar to those for the transverse cross-relaxation rate (Subheading 3.4.1). A difference is that the cross-relaxation experiments are now number I and IV because of the conversion between longitudinal polarization and two-spin order in the middle of the crossrelaxation delay. The optimal delay should be found between 80 and 200 ms. However, particularly in nondeuterated proteins, there is a strong advantage to using shorter delays. If not, one has to calculate a correction for the rates that takes into account the effects of proton–proton cross-relaxation (15). In that case, at least two well-separated time points should be recorded. 2. This experiment can be run with a single relaxation delay if it is short and/or the protein is deuterated. To make sure that the correction is small, it is advisable to run two experiments with different delays.

156

F. Ferrage

3. Follow steps 3–4 for the transverse cross-relaxation protocol (Subheading 3.4.1 ). Figure 5 shows the build-up of the symmetrical reconversion observable for longitudinal crosscorrelated cross-relaxation. These data were recorded with a small deuterated protein, thus allowing the use of long crossrelaxation delays. 4. If two or more relaxation delays are recorded on a nondeuterated sample, evaluate the correction factor, D, for each time point, and compute the ratio: 2

(15)

I IV / I I = C exp(-D 4T ) / (1 + 32d1N DT 3 / 3.

From these2 ratios, fit the value of D. In most cases, evaluate the factor 32d1N DT 3 / 3 , it should be much smaller than 1 so that D can be obtained from a simple exponential fit. 5. Compute for each time point: dlN T = atanh

(II

I IV

)(

)

/ I II I III / 1 + 2D 2T 2 / 3 .

(16)

Proceed to a linear fit to obtain dlN . 3.5. Analysis and Interpretation of 15N Relaxation Rates

The analysis of this dataset closely follows the one presented by Kroenke et al. (14). The use of more accurate measurements of nuclear Overhauser effects and longitudinal CSA/DD cross-correlated

Fig. 5. Example of a longitudinal cross-correlated cross-relaxation build-up curve as shown by Grace during the Curvefit procedure.

9

Protein Dynamics by 15N Nuclear Magnetic Relaxation

157

cross-relaxation rates as well as the comparison of transverse relaxation rates measured under a CPMG echo train and a single echo are the main improvements. 3.5.1. Identification of Chemical Exchange

1. Compute the transverse relaxation rate expected in the absence of exchange R20, for each backbone 15N nucleus: R20 = (R1 - 1.25s NH )dtN / dlN - 1.08s NH .

(17)

2. Derive the exchange contribution to the transverse relaxation rate measured with a CPMG sequence: Rexecho = R2echo - R20 .

(18)

3. Derive the exchange contribution to the transverse relaxation rate measured with a single echo: RexCPMG = R2CPMG - R20 .

(19)

4. Figure 6 shows, as an example, a plot of RexCPMG for Calbindin D9k at 296 K. All but one residue shows contributions of exchange to transverse relaxation rates during CPMG that are significantly larger than 1/s. This confirms that motions of the backbone of Calbindin D9k are very limited on timescales in the 10 ms/ms range. Only L6 and T45 show small but significant contributions of chemical exchange to transverse relaxation. RexCPMG should never be negative, so the presence of a few values around −0.7/s (with an expected error smaller than 0.2/s) indicates that the accuracy of the method is not as good as its precision. In this particular case, the use of a carbon-13 labeled sample makes the derivation of R20 less accurate since contributions of 15 N–13C dipolar interactions to 15N auto-relaxation also have to be taken into account and predicted. 5. The comparison of RexCPMG and Rexecho provides insights on the order of magnitude of the timescales of the exchange processes. If a nonzero value of Rexecho is determined while RexCPMG is zero, the exchange process is slower than 1 ms. If both values are identical, the exchange process is faster than 1 ms. If RexCPMG is smaller than Rexecho but significantly larger than zero, the timescale of the exchange process is close to 1 ms or two exchange processes take place, one faster and one slower than 1 ms. Note that RexCPMG should never be larger than Rexecho. 3.5.2. A Note on the Model-Free Analysis of 15N Relaxation Rates

The ensemble of rates can be used as the input for a model-free analysis (29). A series of software are available: modelfree (30); fast modelfree (31); tensor2 (32); and dynamics (33) are good examples. I do not describe here the use of such software. The only point that discussed is the choice of the ensemble of relaxation rates that should be employed. The above-mentioned softwares are designed to use the longitudinal relaxation rate R1, the transverse relaxation

158

F. Ferrage

Fig. 6. Contribution of chemical exchange processes to transverse relaxation in a single echo experiment RexCPMG in Calbindin D9k at 296 K.

rate R2CPMG, and the NOE ratio as inputs. In most cases, these rates should be used. One of the most difficult tasks of the analysis is the detection of an exchange contribution to transverse relaxation rates R2CPMG. When a nonzero contribution RexCPMG is identified, the rates are at best fit with a simpler model for fast motions. When a protein shows significant contributions of chemical exchange to the relaxation of many 15N nuclei, this can be detrimental to the overall quality of the analysis of hydrodynamic properties and fast local motions. The analysis described herein provides exchangefree transverse relaxation rates R20. If, and only if, the ensemble of RexCPMG rates shows a very flat baseline around zero (see for instance ref. 15), the R20 rates can be employed instead of the R2CPMG rates in the model-free analysis.

4. Notes 1. Note that since the gyromagnetic ratio of 15N is negative, these two frequencies have opposite signs. Keeping this in mind helps understand the analysis of relaxation rates. 2. Particular care should be given to the sample conditions for relaxation studies. The stability of the sample is very important since data are collected as series of experiments in several days (or possibly more) and will be used together for the analysis.

9

Protein Dynamics by 15N Nuclear Magnetic Relaxation

159

We advise against running such a series of experiments on samples with very short (2 days or less) lifetimes unless high reproducibility can be achieved in sample preparation and several samples can be used to collect the ensemble of data discussed here. In order to ensure the quality of the data and a safe interpretation of hydrodynamic properties and chemical exchange, it is advised to verify the concentration dependence of relaxation rates which can be affected by aggregation and oligomerization. The transverse relaxation rate measured with a single echo is very sensitive to such processes and can be used to find the optimal concentration (34). One of the common problems that can arise during the weeklong data collection is the appearance of bubbles, which can degrade the field homogeneity and decrease the signal intensity, especially when a Shigemi tube is being used (as bubbles will be trapped in the detection volume). It is advisable that samples should be degassed before recording long relaxation experiments. 3. The series of experiments described herein is most easily adapted for small and middle-size proteins (less than 200 amino acids, monomeric). In most cases, the data are easily recorded on any spectrometer (B0 ³ 11.7 T) when a concentration around ~1 mM can be used. When the samples are concentration limited, the use of spectrometers equipped with cryogenic probes should be preferred, particularly at low fields (B0 £ 14.1 T). When using cryogenic probes, optimal homogeneity of the magnetic field B0 should be reached in order to ensure proper suppression of the water signal. Probes equipped with triple-axis gradients enhance the water suppression (35). When only z-axis gradients are available, the sign of gradients should be chosen such that the phase of the water transverse polarization builds up (36). For instance, two pulsed field gradients separated in the pulse sequence by a single 180° pulse on the proton channel should have opposite signs in the manner of bipolar gradients. 4. Using a small angle to start helps reduce the effects of radiation damping. It is also practical, since the signal is directly proportional to the flip angle (modulo 360°), one can use the intensity in this first experiment as a meter for the deviations of following pulses from 360°. For example, if the intensity in the first attempt at a 360° pulse is positive and about twice as large as that of the 1 ms pulse experiment, it is likely that a pulse 2 ms shorter will be an exact 360° pulse. 5. If, as described here, only adiabatic pulses are used on the 13C channel, pulse calibration on this channel is optional, since the quality of the inversion by the adiabatic pulse will not be affected by a small deviation of the rf amplitude from its nominal value. 6. Sample script for an exponential decay:

160

F. Ferrage

9

Protein Dynamics by 15N Nuclear Magnetic Relaxation

161

The input file is a text file made of 25 columns; the first one is the peak/residue number while columns 2–25 are the intensity and error for the 12 longitudinal relaxation experiments. 7. WALTZ-16 is not as good a decoupling scheme as GARP. However, its basis element is much shorter. Therefore, when very low amplitude decoupling is performed, or when short durations are favored (like in the present case) WALTZ-16 should be used instead of GARP. 8. Since the four experiments for each relaxation delay are run in an interleaved manner, there is no need to record different relaxation delays in series. If doubts arise about the stability of the sample, run one set for a given value of the relaxation delay. Additional datasets for other relaxation delays (or repeats of the same) can be acquired after the full series of experiments. 9. When working on a spectrometer with a different Larmor frequency, all pulse durations should be scaled with the inverse of the field (e.g., 1.6 ms at 600 MHz corresponds to 1.2 ms at 800 MHz). There are a few exceptions: high-power, 15N CPMG, and 15N decoupling radiofrequency amplitudes usually do not scale up with B0.

Acknowledgments I am grateful to Geoffrey Bodenhausen, David Cowburn, Ranajeet Ghose, Arthur G. Palmer, and Philippe Pelupessy for their many contributions to my training, from hands-on practice to many insightful discussions. I thank Mikael Akke for the sample of Calbindin D9k and Kaushik Dutta for carefully reading this manuscript. References 1. Mittermaier, A., and Kay, L. E. (2006) Review – New tools provide new insights in NMR studies of protein dynamics. Science 312, 224–228. 2. Palmer, A. G. (2004) NMR characterization of the dynamics of biomacromolecules. Chem. Rev. 104, 3623–3640. 3. Massi, F., Wang, C. Y., and Palmer, A. G. (2006) Solution NMR and computer simulation studies of active site loop motion in triosephosphate isomerase. Biochemistry 45, 10787–10794. 4. Tjandra, N., Feller, S. E., Pastor, R. W., and Bax, A. (1995) Rotational Diffusion Anisotropy of Human Ubiquitin from N-15 NMR Relaxation. J. Am. Chem. Soc. 117, 12562–12566. 5. Akke, M., Brüschweiler, R., and Palmer III, A. G. (1993) NMR Order Parameters and Free Energy:

An Analytical Approach and Its Application to Cooperative Ca2+ Binding by Calbindin Dgk. J. Am. Chem. Soc. 115, 9832–9833. 6. Frederick, K. K., Marlow, M. S., Valentine, K. G., and Wand, A. J. (2007) Conformational entropy in molecular recognition by proteins. Nature 448, 325–329. 7. Kuszewski, J., Gronenborn, A. M., and Clore, G. M. (1999) Improving the packing and accuracy of NMR structures with a pseudopotential for the radius of gyration. J. Am. Chem. Soc. 121, 2337–2338. 8. Ryabov, Y., and Fushman, D. (2007) Structural Assembly of Multidomain Proteins and Protein Complexes Guided by the Overall Rotational Diffusion Tensor. J. Am. Chem. Soc. 129, 7894–7902.

162

F. Ferrage

9. Ryabov, Y. E., and Fushman, D. (2007) A model of interdomain mobility in a multidomain protein. J. Am. Chem. Soc. 129, 3315–3327. 10. Palmer, A. G., and Massi, F. (2006) Characterization of the dynamics of biomacromolecules using rotating-frame spin relaxation NMR spectroscopy. Chem. Rev. 106, 1700–1719. 11. Palmer, A. G. (2004) NMR characterization of the dynamics of biomacromolecules. Chem. Rev. 104, 3623–3640. 12. Vallurupalli, P., Hansen, D. F., and Kay, L. E. (2008) Structures of invisible, excited protein states by relaxation dispersion NMR spectroscopy. Proc. Natl. Acad. Sci. U.S.A. 105, 11766–11771. 13. Kay, L. E., Torchia, D. A., and Bax, A. (1989) Backbone Dynamics of Proteins as Studied by N-15 Inverse Detected Heteronuclear NMRSpectroscopy – Application to Staphylococcal Nuclease. Biochemistry 28, 8972–8979. 14. Kroenke, C. D., Loria, J. P., Lee, L. K., Rance, M., and Palmer III, A. G. (1998) Longitudinal and Transverse H-1-N-15 Dipolar N-15 Chemical Shift Anisotropy Relaxation Interference: Unambiguous Determination of Rotational Diffusion Tensors and Chemical Exchange Effects in Biological Macromolecules. J. Am. Chem. Soc. 120, 7905–7915. 15. Pelupessy, P., Ferrage, F., and Bodenhausen, G. (2007) Accurate Measurement of Longitudinal Cross-Relaxation Rates in Nuclear Magnetic Resonance. J. Chem. Phys. 126, 134508. 16. Korzhnev, D. M., Billeter, M., Arseniev, A. S., and Orekhov, V. Y. (2001) NMR Studies of Brownian Tumbling and Internal Motions in Proteins. Prog. Nucl. Magn. Reson. Spectrosc. 38, 197–266. 17. Luginbuhl, P., and Wuthrich, K. (2002) Semiclassical nuclear spin relaxation theory revisited for use with biological macromolecules. Prog. Nucl. Magn. Reson. Spectrosc. 40, 199–247. 18. Nicholas, M. P., Eryilmaz, E., Ferrage, F., Cowburn, D., and Ghose, R. (2010) Nuclear spin relaxation in isotropic and anisotropic media, Prog. Nucl. Magn. Reson. Spectrosc. 57, 111–158. 19. Cavanagh, J., Fairbrother, W. J., Palmer III, A. G., Rance, M., and Skelton, N. J. (2006) Protein NMR Spectroscopy: Principles and practice, Academic Press, San Diego. 20. Wang, L. C., Pang, Y. X., Holder, T., Brender, J. R., Kurochkin, A. V., and Zuiderweg, E. R. P. (2001) Functional dynamics in the active site of the ribonuclease binase. Proc. Natl. Acad. Sci. U.S.A. 98, 7684–7689. 21. Wang, C. Y., and Palmer, A. G. (2003) Solution NMR methods for quantitative identification of

chemical exchange in N-15-labeled proteins. Magn. Reson. Chem. 41, 866–876. 22. Kempf, J. G., and Loria, J. P. (2004) Measurement of Intermediate Exchange Phenomena. Meth. Mol. Biol. 278, 185–231. 23. Findeisen, M., Brand, T., and Berger, S. (2007) A H-1-NMR thermometer suitable for cryoprobes. Magn. Reson. Chem. 45, 175–178. 24. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J., and Bax, A. (1995) NMRPipe: a Multidimensional Spectral Processing System Based on UNIX Pipes. J. Biomol. NMR 6, 277–293. 25. Ferrage, F., Piserchio, A., Cowburn, D., and Ghose, R. (2008) On the measurement of N-15-{H-1} nuclear Overhauser effects. J. Magn. Reson. 192, 302–313. 26. Ferrage, F., Cowburn, D., and Ghose, R. (2009) Accurate Sampling of High-Frequency Motions in Proteins by Steady-State 15N-{1H} Nuclear Overhauser Effect Measurements in the Presence of Cross-Correlated Relaxation. J. Am. Chem. Soc. 131, 6048–6049. 27. Ferrage, F., Reichel, A., Battacharya, S., Cowburn, D., and Ghose, R. (2010) On the measurement of N-15-{H-1} nuclear Overhauser effects. 2. Effects of the saturation scheme and water signal suppression. J. Magn. Reson. 207, 294–303. 28. Pelupessy, P., Espallargas, G. M., and Bodenhausen, G. (2003) Symmetrical reconversion: measuring cross-correlation rates with enhanced accuracy. J. Magn. Reson. 161, 258–264. 29. Lipari, G., and Szabo, A. (1982) Model-Free Approach to the Interpretation of Nuclear Magnetic Resonance Relaxation In Macromolecules 1. Theory and Range of Validity. J. Am. Chem. Soc. 104, 4546–4559. 30. Mandel, A. M., Akke, M., and Palmer III, A. G. (1995) Backbone Dynamics of Escherichia coli Ribonuclease HI : Correlations with Structure and Function in an Active Enzyme. J. Mol. Biol. 246, 144–163. 31. Cole, R., and Loria, J. P. (2003) FAST-Modelfree: A program for rapid automated analysis of solution NMR spin-relaxation data. J. Biomol. NMR 26, 203–13. 32. Dosset, P., Hus, J. C., Blackledge, M., and Marion, D. (2000) Efficient analysis of macromolecular rotational diffusion from heteronuclear relaxation data. J. Biomol. NMR 16, 23–28. 33. Fushman, D., Cahill, S., and Cowburn, D. (1997) The main chain dynamics of the dynamin pleckstrin homology (PH) domain in solution: analysis of 15N relaxation with

9

Protein Dynamics by 15N Nuclear Magnetic Relaxation

monomer/dimer equilibration. J. Mol. Biol. 266, 173–194. 34. Butterwick, J. A., Loria, J. P., Astrof, N. S., Kroenke, C. D., Cole, R., Rance, M., and Palmer, A. G. (2004) Multiple time scale backbone dynamics of homologous thermophilic and mesophilic ribonuclease HI enzymes. J. Mol. Biol. 339, 855–871. 35. Sarkar, R., Moskau, D., Ferrage, F., Vasos, P. R., and Bodenhausen, G. (2008) Single or triple gradients? J. Magn. Reson. 193, 110–118. 36. Muhandiram, D. R., Yamazaki, T., Sykes, B. D., and Kay, L. E. (1995) Measurement of 2H T1ro Relaxation Times in Uniformly 13 C-Labeled and Fractionally 2H-Labeled Proteins in Solution. J. Am. Chem. Soc. 117, 11536–11544.

163

37. Bohlen, J. M., and Bodenhausen, G. (1993) Experimental Aspects of Chirp NMRSpectroscopy. J. Magn. Reson. A 102, 293–301. 38. Shaka, A. J., Barker, P. B., and Freeman, R. (1985) Computer-Optimized Decoupling Scheme for Wideband Applications and LowLevel Operation. J. Magn. Reson. 64, 547–552. 39. Emsley, L., and Bodenhausen, G. (1990) Gaussian Pulse Cascades-New Analytical Functions for Rectangular Selective Inversion and In-phase Excitation in NMR. Chem. Phys. Lett. 165, 469–476. 40. Shaka, A. J., Keeler, J., Frenkiel, T., and Freeman, R. (1983) An Improved Sequence for Broad Band Decoupling – WALTZ-16. J. Magn. Reson. 52, 335–38.

Chapter 10 Bacterial Production and Solution NMR Studies of a Viral Membrane Ion Channel Jolyon K. Claridge and Jason R. Schnell Abstract Advances in solution nuclear magnetic resonance (NMR) methodology that enable studies of very large proteins have also paved the way for studies of membrane proteins that behave like large proteins due to the added weight of surfactants. Solution NMR has been used to determine the high-resolution structures of several small, membrane proteins dissolved in detergent micelles and small bicelles. However, the usual difficulties with membrane proteins in producing, purifying, and stabilizing the proteins away from native membranes remain, requiring intensive screening efforts. Low levels of heterologous expression can be the most detrimental aspect to studying membrane proteins. This is exacerbated for NMR studies because of the costs of isotopically enriched media. Thus, solution NMR studies have tended to focus on relatively small, membrane proteins that can be expressed into inclusion bodies and refolded. Here, we describe the methods used to produce, purify, and refold the proton channel M2 into detergent micelles, and the procedures used to determine chemical shift assignments and the atomic level structure of the closed form of the homotetrameric channel. Key words: Membrane proteins, Ion channel, NMR, M2

1. Introduction Several difficulties arise in applying solution nuclear magnetic resonance (NMR) to the study of membrane proteins. Foremost among the difficulties are obtaining sufficient quantities of pure and natively folded membrane proteins at a reasonable cost, adverse effects of increased relaxation rates from complexes of protein and detergent micelles or detergent and lipid bicelles, limited lifetime of natively folded membrane proteins outside of lipid bilayers, and cost and availability of deuterated detergents. The requirement that the native structure be stable for days or months remains a general problem

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_10, © Springer Science+Business Media, LLC 2012

165

166

J.K. Claridge and J.R. Schnell

for structural studies of membrane proteins and largely relies on trial and error to overcome. Access to relatively inexpensive deuterated detergents facilitates collection of information-rich methyl-based experiments, and in some cases it is necessary also to prevent rapid spin diffusion in backbone amide-based NOESY experiments. Increases in transverse relaxation rates with increasing protein size often mean that structure determinations to high resolution by solution NMR becomes very difficult for proteins with rotational correlation times of less than ~20 ns, unless extraordinary efforts are made. In general, this limit corresponds to protein sizes of ~35 kDa for water-soluble proteins at 30°C. However, studies of membrane proteins become challenging at much lower sizes because of the additional mass of the detergent micelles, which might double or triple the effective tumbling times. The introduction of TROSY-based pulse sequences (1, 2) and methyl-based approaches (3) that have been used extensively for much larger water-soluble proteins greatly benefit studies of small- and mediumsized membrane proteins (4, 5) as well. A common difficulty in studying membrane proteins by solution NMR (or, indeed, by any other method) arises from the inability to produce the proteins in suitable quantities. For NMR, the high unit cost of culture media enriched in the rare isotopes needed for multidimensional NMR studies prohibits the large-scale cultures (>10 L) common for crystallographic studies. Although isotopic labeling in eukaryotic systems (6, 7) or via cell-free methods (8) have been reported, the predominant method by which isotopically labeled proteins are produced remains by expression in bacteria. Using such methods, both the number and size of membrane proteins that have been solved by solution NMR have grown rapidly in the past few years (9, 10). Several of these studies expressed the proteins into bacterial membranes (e.g., refs. 3–5, 11); however, a larger trend has been to use solution NMR to study relatively small-membrane proteins that can be refolded from bacterial inclusion bodies. Membrane proteins containing a small number of transmembrane helices, which are more likely to fold in vitro, appear to be abundant in genomes (12, 13). Expression into inclusion bodies may happen spontaneously or by fusing the target protein to one that is efficiently “packaged” into inclusion bodies. One such fusion protein is “trpLE” (14), which has been used to express and purify several hydrophobic peptides: Vpu from HIV (15), transmembrane regions of GPCRs (16), caveolins (17), and several others (18–20) (see Note 1). TrpLE contains the leader sequence of the trp operon of Escherichia coli fused to a sequence of 97 residues found near the C-terminus of the anthranilate synthase gene (21, 22). Fusion to trpLE directs the protein to inclusion bodies, with the advantages of reduced

10

Bacterial Production and Solution NMR Studies of a Viral…

167

toxicity and protease resistance (23). In the trpLE construct, cysteines have been mutated to alanines to prevent disulfide bond formation, and methionines have been mutated to leucine to allow for the introduction of a single CNBr cleavage site at the fusion junction (14). In this chapter, we outline the approach used to produce, purify, and determine the structure of the M2 proton channel from influenza A in detergent micelles (24). M2 is one of the smallest ion channels known, being assembled from a homotetramer of a single-pass membrane protein. However, it is surprisingly sophisticated, being pH gated and highly proton selective. Past circulating strains of M2 are the target of a class of antiviral drugs, the adamantanes, and understanding drug resistance may help in developing drugs that are again effective. We describe how M2 is expressed at high yield into bacterial inclusion bodies as a trpLE fusion, chemically cleaved, purified, and refolded into detergent micelles to provide high-resolution NMR spectra of the native homotetramer (24) (Fig. 1).

Fig. 1. An overview of the homotetrameric structure of the high pH, closed M2 proton channel. Alpha helices are shown as cylinders. An idealized schematic of the DHPC micelle is shown as a single layer of detergent molecules with the hydrocarbon chains coating the transmembrane domain. The positioning of the micelle is based on the absence of cross peaks in amide backbone strips at the resonance frequency of water in an 15N-edited NOESY (24).

168

J.K. Claridge and J.R. Schnell

2. Materials 2.1. Expression of the TrpLE-M2 Fusion Protein

2.2. Fusion Protein Purification, Cleavage, and Preparation of Pure M2

●

Expression strain of E. coli: BL21(DE3) pLysS cells (Novagen).

●

LB agar plates: 10 mg/mL tryptone, 5 mg/mL yeast extract, 10 mg/mL NaCl, and 16 mg/mL agar. Adjust pH to 7.0, if necessary, with NaOH or HCl and autoclave. Add 100 mM kanamycin and 100 mM chloramphenicol just before solidification begins, and pour into plates.

●

LB medium: 10 mg/mL tryptone, 5 mg/mL yeast extract, and 10 mg/mL NaCl. Adjust pH to 7.0, if necessary, with NaOH or HCl and autoclave.

●

M9 medium: 42 mM Na2HPO4, 22 mM KH2PO4, 20 mM NH4Cl, 10 mM NaCl, 22 mM glucose, 2 mM MgSO4, 0.1 mM CaCl2, and 1× MEM vitamin solution (10 mL of a 100× stock purchased from Sigma–Aldrich), 100 mM kanamycin, and 100 mM chloramphenicol. The salts (Na2HPO4, KH2PO4, NH4Cl, and NaCl) are first dissolved into 990 mL of distilled and 0.22-mm filtered water and autoclaved, followed by addition of the remaining components. For production of 15N-labeled protein, uniformly 15N-labeled ammonium chloride is substituted. For 13C-labeled protein, 11 mM of uniformly 13C-labeled glucose is substituted. For partial deuteration, the M9 solutes are dissolved in 99.8% 2H2O and sterile filtered through 0.2-mm filter flasks, rather than autoclaved.

●

1 M DTT: Prepare stock solution in distilled, filtered water.

●

10% SDS: Prepare by adding 1 g of SDS to 10 mL of distilled, filtered water.

●

NuPAGE Novex 12% Bis–Tris Gel (Invitrogen).

●

4× NuPAGE LDS Sample Buffer (Invitrogen).

●

NuPAGE® MES SDS Running Buffer (Invitrogen).

●

1 M IPTG: Prepare stock solution in distilled, filtered water.

●

8 M Urea: Prepare stock solution in distilled, filtered water.

●

Lysis buffer: 50 mM Tris–HCl, pH 8.0, 200 mM NaCl.

●

Guanidine buffer: 20 mM Tris–HCl, pH 8.0, 200 mM NaCl, 6 M guanidine HCl, and 15 mM imidazole.

●

Elution buffer: 20 mM Tris–HCl, pH 7.0, 200 mM NaCl, 6 M guanidine HCl, and 400 mM imidazole.

●

Reverse phase C4 column for HPLC was purchased from GraceVydac (214TP C4, 300 Å silica, 5 mm beads, 2.1 × 150 mm).

●

Hexafluoroisopropanol: ³98% purity.

●

Formic acid: ³98% purity.

●

Syringe filter: 0.2 mm polytetrafluoroethylene.

2.3. Reconstitution of M2 into Detergent Micelles

2.4. NMR Chemical Shift Assignments and Restraint Measurements

2.5. Structure Calculation and Measurement of Tryptophan Gate Dynamics

10

Bacterial Production and Solution NMR Studies of a Viral…

169

●

HPLC buffer A: 5% isopropanol, 95% water, 0.1% trifluoroacetic acid.

●

HPLC buffer B: 57% isopropanol, 38% acetonitrile, 5% water, 0.1% trifluoroacetic acid.

●

Ni-NTA agarose beads.

●

Cyanogen bromide: Solid, ³98.5 % purity.

●

Dialysis tubing: 10 kDa MWCO, 22-mm internal diameter.

●

Dialysis cassette with 3.5 kDa MWCO.

●

Reconstitution buffer: 50 mM sodium phosphate, pH 7.5, 6 M guanidine HCl, 0.3 M dihexanoyl-sn-glycerol-3-phosphocholine (DHPC) detergent, 30 mM sodium glutamate. Adjust pH with NaOH.

●

NMR buffer: 50 mM sodium phosphate, pH 7.5, 30 mM sodium glutamate. Adjust pH with NaOH.

●

Deuterium oxide: 98% purity.

●

Rimantadine: 0.3 M stock prepared in 80 mM DHPC.

●

Mini dialysis cups for protein refolding: 3.5 kDa MWCO, 10–100 mL capacity (Thermo Scientific).

●

Centrifugal concentrator: 5 kDa MWCO.

●

All NMR experiments are conducted on a 14.1 T spectrometer equipped with a cryogenic probe.

●

Processing and preliminary analysis of data, peak fitting, and extraction of intensities for determining scalar bond couplings are performed using NMRPipe (25).

●

Resonance assignments and quantitation of NOE cross-peak intensities are performed using CARA (26).

●

Prediction of backbone dihedral angles from chemical shifts is performed using TALOS (27).

●

Cylindrically shaped polyacrylamide gel: A solution containing 4.5% acrylamide concentration with an acrylamide/bisacrylamide molar ratio of 40 is cast in a 6-mm cylindrical vessel.

●

A gel press kit (New Era Enterprises, Inc.) is used to push the cylindrical gel (6 mm in diameter) into an open-ended 4.2-mm inner diameter NMR tube.

●

Structure calculations were performed using the program XPLOR-NIH (28).

●

The program PALES (29) was used to evaluate the fit of residual dipolar couplings (RDCs) to structures.

●

Fitting of relaxation dispersion data was accomplished using the CPMGfit software (Dr. Art Palmer).

170

J.K. Claridge and J.R. Schnell

3. Methods The following protocols describe in detail our process for expressing, purifying, reconstituting the M2 proton channel into detergent micelles, assigning chemical shifts, and performing structural analyses. However, our lab has used the same protocol for several membrane proteins containing one or two transmembrane domains. A generalized flowchart of the procedure is provided in Fig. 2. The gene of interest was inserted between the HindIII (5¢) and BamHI (3¢) sites in the pMMHb plasmid (kindly provided by Stephen Blacklow, Brigham and Women’s Hospital, Boston), which contains a gene conferring kanamycin resistance. Cyanogen bromide is used to cleave the expressed construct on the C-terminal side of a unique methionine, resulting in release of the N-terminal (His)9-tagged trpLE, and the C-terminal M2 construct with no extra residues. Purified peptide of different isotopic composition is gently refolded from denaturing conditions into buffered detergent solution suitable for solution NMR experiments. Samples are typically stable for up to 3 weeks. Resonance assignments are obtained

Fig. 2. A generalized flowchart showing the production of tetrameric M2 samples from trpLE fusions expressed in Escherichia coli.

10

Bacterial Production and Solution NMR Studies of a Viral…

171

from a combination of triple-resonance NMR experiments, an 15 N-separated NOESY, and a 13C-separated NOESY collected on 15 N, 13C, 2H-labeled, 15N-labeled, and 15N, 13C-labeled peptide, respectively. Structure calculations are performed by an iterative process that incorporates local experimental restraints first, followed by longrange restraints. RDCs are incorporated last with a low-temperature refinement to avoid local minima that arise from the degeneracy of RDC magnitudes. 3.1. Expression of the TrpLE-M2 Fusion Protein

1. Transform the plasmid into E. coli BL21(DE3) pLysS cells, plate on LB agar containing kanamycin and chloramphenicol, and incubate overnight at 37°C. 2. Inoculate cultures of 200–500 mL Luria–Bertani (LB) media with individual colonies of freshly transformed bacteria and grow overnight at 37°C with moderate shaking (150 rpm). 3. Centrifuge the cultures at 2,000 × g for 25 min at 4°C and resuspend into 40 mL of chilled M9 medium. Add the resuspended cells to the large-scale M9 cultures (2–4 L) such that the OD600 is 0.2–0.3 relative to water. Grow each liter of culture in a 2.5-L baffled flask at 37°C with moderate shaking (150 rpm). 4. When the OD600 reaches 0.6–0.7, induce expression of the trpLE-M2 fusion by adding IPTG from a stock of 1 M IPTG to a final concentration of 1 mM. Grow overnight. The final OD600 is typically between 1.2 and 1.4 (see Note 2). 5. Analyze protein expression levels by using SDS-PAGE (12% Bis–Tris gel). Spin down cell quantities equivalent to 250 mL of OD 600 = 0.6 cell culture at 5,000 × g for 5 min at room temperature and redissolve the pellet into 40 mL of 8 M urea, 20 mL of 4× LDS sample buffer, 5 mL of 1 M DTT, and 5 mL of 10% SDS. Load 20 mL of sample per lane and run the gel using MES SDS running buffer.

3.2. Fusion Protein Purification, Cleavage, and Preparation of Pure M2

1. Harvest the cells by centrifugation at 5,000 × g for 30 min at 4°C. Resuspend the cell pellets in lysis buffer using a Dounce homogenizer, and sonicate on ice for a total of 3 min with a 20% duty cycle. Spin down the inclusion bodies and cell debris at 10,000 × g for 25 min at 4°C. Solubilize the pellets in 40 mL of lysis buffer and spin down at 10,000 × g for 25 min at 4°C to purify away additional water-soluble contaminants (see Note 3). 2. Dissolve the water-insoluble matter in 50 mL (per 1 L culture) of guanidine buffer using a Dounce homogenizer. Pellet undissolved matter, which includes nucleic acids, by centrifugation at 100,000 × g for 1.5 h at 4°C. Add the supernatant to 2 mL of Ni-NTA agarose beads (see Note 4) preequilibrated with guanidine buffer. After a 1-h incubation at 4°C with gentle rotation, pour the slurry into a gravity column and wash with 80 mL of guanidine buffer. Elute bound trpLE-M2 fusion

172

J.K. Claridge and J.R. Schnell

protein from the column by adding elution buffer in three stages of 5 mL each. 3. Dialyze the elution containing the trpLE-M2 fusion in 10 kDa molecular weight cutoff (MWCO) dialysis tubing against 4 L of H2O, with several exchanges over 4 h. The trpLE-M2 fusion precipitates as white flakes. Centrifuge in a swinging-bucket rotor at 1,500 × g for 30 min at 4°C. 4. Chemically cleave the M2 peptide from trpLE by dissolving the pellet into 5 mL of 70% formic acid containing 1 g of CNBr (see Note 5). Cover the reaction vessel with aluminum foil and allow the reaction to proceed for 2 h under a low-pressure stream of nitrogen gas. Load the sample via syringe into a dialysis cassette with a 3.5 kDa MWCO and dialyze against 4 L of water for 1 h, snap frozen in liquid nitrogen, and then lyophilize. 5. In preparation for reverse-phase HPLC, dissolve the lyophilized sample in 1.5 mL of hexafluoroisopropanol, which clarifies after 15 min at 40°C. Subsequently, add 0.5 mL of formic acid and 2 × 1 mL of water. Draw this solution into a 5-mL syringe, degas using a vacuum hose, and load onto a C4 reverse-phase column through a syringe filter. Collect fractions over a linear gradient from 0 to 100% buffer B (100–0% buffer A). Cleaved trpLE elutes first (~55% buffer B), followed by the uncut fusion (~65% buffer B), and the cleaved M2 peptide (~75% buffer B). Cleavage efficiencies for the trpLE-M2 fusion are routinely between 70 and 80%. Lyophilize the pooled fractions. For quantitation and aliquoting of peptide, dissolve the dried samples in 50% acetonitrile containing 0.1% TFA, quantitate by absorbance at 280 nm based on amino acid composition, and relyophilize (see Note 6). 3.3. Reconstitution of M2 into Detergent Micelles

1. Dissolve purified peptide (1.2 mg) into reconstitution buffer (see Note 7) at a concentration of 250 mM (final volume is 960 mL), split into three 3.5 kDa MWCO dialysis cups, and dialyze for 12 h against 2 L of NMR buffer with slow stirring and one buffer change at 10 h. 2. Concentrate the sample to ~0.7 mM monomer using a centrifugal concentrator with a 5 kDa MWCO. 3. Add rimantadine to 10 mM. Rimantadine binds at four equivalent sites near the gate on the lipid-facing side of the channel and stabilizes the closed conformation of the pore. Because of the poor water solubility of rimantadine, it is added from a 0.3 M stock that contains 80 mM DHPC (Subheading 2.3).

3.4. NMR Chemical Shift Assignments and Restraint Measurements

4. Add deuterium oxide for the magnetic field lock to a concentration of 5% in three steps. 1. To achieve nearly complete sequence-specific backbone chemical shift assignment of 1HN, 15N, 13C¢, 13Ca, and 13Cb, use

10

Bacterial Production and Solution NMR Studies of a Viral…

173

TROSY versions of the HNCA, HNCACB, and HNCO experiments (30, 31) on fully 15N-, 13C-, and 85% 2H-labeled protein (see Note 8). Process spectra in NMRPipe (25), and assign resonances and quantitate NOE cross-peak intensities in CARA (26). 2. Once backbone chemical shifts are known, use TALOS (27) to predict regions of secondary structure. For M2, this indicates that the transmembrane domain and the C-terminal juxtamembrane region are alpha-helical. This is confirmed by the characteristic local NOE patterns, and is used to aid assignment of most 1Ha and 1Hb intraresidue and sequential NOEs in a 3D 15 N-edited NOESY spectrum (110 ms; see Note 9 and Fig. 3). 3. Collect a methyl-based 3D 13C-edited NOESY (150 ms mixing time) on fully 15N-, 13C-labeled protein; this is particularly helpful in the methyl-rich transmembrane segment to confirm backbone proton assignments, extend the side-chain proton assignments, and identify intermolecular contacts (Fig. 4; see Note 8). Obtain stereospecific assignments of gamma methyls of valine and delta methyls of leucine from 10% 13C-labeled protein by recording a constant time 1H–13C HSQC with 28-ms carbon evolution, which allows discrimination between coupled and uncoupled methyls based on the sign of the cross peak (32). 4. Complete assignment of side-chain proton resonances is facilitated by determining a large number of c1 and c2 rotamers.

Fig. 3. Projection of the 1H,1H plane of 15N-separated NOESY spectra on an M2 sample containing (a) fully protonated or (b) fully deuterated DHPC detergent showing the loss of information due to spin diffusion when the hydrocarbon chain of DHPC is protonated (see Note 9). NOE mixing times were 90 and 110 ms, respectively.

174

J.K. Claridge and J.R. Schnell

Fig. 4. Final, lowest-energy structure of the high pH M2 channel showing the position of methyl groups throughout the transmembrane domain, including the helix–helix interfaces. Helices are shown as ribbons and methyl protons as filled circles. Structural elements in adjacent helices are shaded darker or lighter.

Determine the c1 of isoleucines, threonine, and valines from methyl-based measurements of the 3-bond scalar couplings 3 JNCg and 3JC¢C g (33, 34), and the c2 of leucine and isoleucine from 3JCaCd (35, 36). Determine the c1 of long-chain aliphatic (arginine, leucine, and lysine) and aromatic (histidine, phenylalanine, tryptophan, and tyrosine) side chains from 3JNCg and 3JC¢Cg values measured in an 1H–15N constant-time TROSY experiments on 15N-, 13C-, and 85% 2H-labeled protein (37, 38). 5. NOEs that cannot be explained by intramonomer distances based on the local secondary structure are identified as intermonomer NOEs. Carry out the assigning of intermonomer distance restraints and structure calculations iteratively until all NOE cross peaks in the NOESY spectra are self-consistent (see Subheading 3.5). First, identify protein-drug NOEs in the 15 N-edited and 13C-edited NOESYs described in step 3, and subsequently confirm them by acquiring an 15N-edited NOESY (500-ms mixing time) on a sample containing uniformly 15N- and 2 H-labeled protein, protonated rimantadine, and perdeuterated DHPC (see Note 10). 6. Weakly align the ion channel relative to the magnetic field by using a strained gel (39–41). Soak the protein and detergent solution into a cylindrically shaped polyacrylamide gel, initially 6 mm in diameter and 9 mm in length; squeeze this into the

10

Bacterial Production and Solution NMR Studies of a Viral…

175

4.2-mm inner diameter of an open-ended NMR tube. Obtain the RDCs by subtracting the scalar coupling (J), measured on an unaligned sample, from the measured coupling (J + RDC) of the aligned sample. In both cases, obtain couplings by interleaving a regular gradient-enhanced HSQC and a gradient-selected TROSY (42), both acquired with 80 ms of 15N evolution. 3.5. Structure Calculation and Measurement of Tryptophan Gate Dynamics

1. Begin structure calculations from a random coil structure generated in XPLOR-NIH that incorporates intramonomer NOEs, backbone dihedral angle restraints derived from chemical shifts, and side-chain c1 and c2 restraints. Enforce the intramonomer NOE restraints by using flat-well harmonic potentials, with the force constant fixed at 50 kcal-mol-1-Å-2. For fixed side-chain c1 and c2 angles, apply flat-well (±30°) harmonic potentials with a force constant of 30 kcal-mol-1-rad-2. During the simulation, ramp the van der Waals, improper angle, and bond angle force constants to 4.0 kcal-mol-1-Å-2, 1.0 kcal-mol-1-degree-2 and 1.0 kcal-mol-1-degree-2, respectively. Calculate a total of 20 monomer structures with a standard high-temperature simulated annealing protocol in which the bath temperature is cooled from 1,000 to 200 K. 2. To obtain an initial set of tetramer structures, replicate and uniquely translate the lowest-energy monomer structure four times. Perform another high-temperature simulated annealing run (1,000 to 200 K) using all previous restraints plus intermonomer NOEs, which are applied in a fourfold manner consistent with the C4 symmetry implied by a single set of resonances. Calculate a total of 100 tetramer structures. 3. Independently cross validate the 100 tetramer structures by 1 H–15N RDCs using singular value decomposition (SVD) as implemented in PALES. Assess the goodness of fit by the Pearson correlation coefficient (r) and quality factor (Q) (43). Select the 15 structures with the best agreement with RDCs (r ~ 0.91 and Q ~ 0.25). 4. Set the approximate initial values of the magnitude (Da) and rhombicity (Rh) of the alignment tensor to the average values of Da = 14.0 Hz and Rh = 0.20 that were obtained from the best SVD fits. Due to flexibility in the tetrameric assembly on the timescale of RDCs (up to microseconds (44)), the rhombicity is nonzero when the monomers are fit to a single alignment tensor. Thus, the monomers are restrained against independent alignment tensors during refinement. During this final refinement, cool the bath from 200 to 20 K. Fix the force constants for NOE and dihedral restraints at 100 kcal-mol-1-Å-2 and 40 kcal-mol-1-rad-2, respectively. In addition, ramp a weak database-derived “Rama” potential function from 0.02 to 0.20

176

J.K. Claridge and J.R. Schnell

(dimensionless force constant) for the general treatment of sidechain rotamers. The RDC restraint force constant was ramped from 0.010 to 0.125 kcal-mol-1-Hz-2, thereby supplementing, but not supplanting, the NOE restraints. Generate 10 RDCrefined structures for each of the 15 structures validated by RDCs, and add the structure with the lowest total energy to the final ensemble. Choose the structure with heavy atom conformation closest to the mean to represent the ensemble. 5. Measure the timescale of chemical shift exchange of the Trp41 side chain using an 15N relaxation dispersion CPMG experiment (45). Because the single tryptophan He1 of M2(18–60) is downfield from the backbone amides, measure relaxation dispersion to high precision using a 1D experiment with many scans (>1,000). 6. Fit the dependence of 15N relaxation due to chemical exchange on the frequency of refocusing (1/tcp) of chemical shift evolution to a two-site exchange model given by Rex ∝ 1 – (2tex/tcp) tanh (tcp/2tex), where Rex is the contribution to transverse relaxation due to chemical shift exchange and tex is the correlation time of the process that is generating the chemical shift exchange (46). Analyze the relaxation-compensated CPMG experiment (45) using the program CPMGfit.

4. Notes 1. Single-pass membrane proteins approaching 100 amino acids are robustly expressed into bacterial inclusion bodies with N-terminal trpLE fusions. We have observed that the expression yields of longer proteins or those containing additional TM helices can be increased by the addition of a second, C-terminal trpLE with an intervening methionine for cleavage. 2. Optimal temperature and IPTG conditions should be explored for each construct. In some cases, lower IPTG and lower temperature (100 mM IPTG and 18°C) or higher IPTG and higher temperature (1 mM IPTG and 42°C) result in higher expression levels. In some cases, better expression can be obtained using low volumes (200 mL) in 2.5-L baffled flasks and shaking at very high speeds (300 rpm) to increase aeration. 3. Beta-mercaptoethanol at 0.05% can be added to all cellprocessing buffers if cysteines are present in the target protein. 4. For maximum yields, it may be necessary to use fresh Ni-NTA resin for each preparation because of degradation that occurs in 6 M guanidine.

10

Bacterial Production and Solution NMR Studies of a Viral…

177

5. Use of CNBr for cleavage of the fusion protein requires a unique methionine at the fusion site. However, the side chain of the residue following the methionine can affect cleavage efficiencies. The hydroxyls of serine and threonine residues impair cleavage efficiency and should be avoided. In contrast, we have observed particularly high cleavage efficiencies when a glycine follows methionine. In addition, lower yields are observed for reactions with low concentrations of the fusion protein. Formylation of tryptophan and lysine side chains can occur when the reaction is allowed to proceed for longer than 3 h. CNBr is very toxic and corrosive, requiring a chemical hood and all safety precautions necessary to avoid direct contact. 6. A light, fluffy consistency of the lyophilized peptide prior to reconstitution correlated with best-quality samples for NMR. 7. The presence of glutamate is necessary to avoid nonspecific aggregation of M2 above ~200 mM (47). It is important to include glutamate in the 2H2O that is added to NMR samples to prevent localized aggregation. 8. Deuteration of the protein and the use of TROSY is critical for observing many of the cross peaks in the transmembrane domain because of the slow, rotational tumbling time of the tetrameric protein (~20 kDa) plus bound detergent. 9. The 3D 13C-edited NOESY spectrum was collected in the presence of perdeuterated detergent to prevent obfuscation of the aliphatic region. Collection of the 15N-edited NOESY spectrum also required deuterated detergent to prevent loss of magnetization due to rapid spin diffusion. Subsequent tests established that deuteration of only the acyl chains was sufficient to suppress rapid spin diffusion. The cost of perdeuterated DHPC (D35) is approximately tenfold more than that of DHPC deuterated only at the acyl chains (D22). 10. To ensure that essentially all nonexchangeable protein protons were replaced with deuterium, cells were grown in 99.9% 2H2O and perdeuterated glucose (Cambridge Isotope Laboratories).

Acknowledgments James J. Chou and the National Institutes of Health, USA (NIH) are acknowledged for supporting development of the approaches described. J.R.S. was supported by a Ruth Kirschstein Fellowship from the NIH. Matthew E. Call is acknowledged for many useful modifications and additions to the trpLE fusion expression and purification protocol.

178

J.K. Claridge and J.R. Schnell

References 1. LeMaster, D. M. (1994) Isotope labeling in solution protein assignment and structural analysis. Prog. Nucl. Magn. Reson. Spectrosc. 26, 371–419. 2. Pervushin, K., Riek, R., Wider, G., and Wuthrich, K. (1997) Attenuated T2 relaxation by mutual cancellation of dipole-dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution. Proc. Natl. Acad. Sci. U.S.A 94, 12366–12371. 3. Ruschak, A. M., and Kay, L. E. (2010) Methyl groups as probes of supra-molecular structure, dynamics and function. J. Biomol. NMR 46, 75–87. 4. Gautier, A., Mott, H. R., Bostock, M. J., Kirkpatrick, J. P., and Nietlispach, D. (2010) Structure determination of the seven-helix transmembrane receptor sensory rhodopsin II by solution NMR spectroscopy. Nat. Struct. Mol. Biol. 17, 768–774. 5. Imai, S., Osawa, M., Takeuchi, K., and Shimada, I. (2010) Structural basis underlying the dual gate properties of KcsA. Proc. Natl. Acad. Sci. U.S.A. 107, 6216–6221. 6. Bruggert, M., Rehm, T., Shanker, S., Georgescu, J., and Holak, T. A. (2003) A novel medium for expression of proteins selectively labeled with 15 N-amino acids in Spodoptera frugiperda (Sf9) insect cells. J. Biomol. NMR 25, 335–348. 7. Strauss, A., Bitsch, F., Cutting, B., Fendrich, G., Graff, P., Liebetanz, J., Zurini, M., and Jahnke, W. (2003) Amino-acid-type selective isotope labeling of proteins expressed in Baculovirusinfected insect cells useful for NMR studies. J. Biomol. NMR 26, 367–372. 8. Makino, S., Goren, M. A., Fox, B. G., and Markley, J. L. Cell-free protein synthesis technology in NMR high-throughput structure determination. Methods Mol. Biol. 607, 127–147. 9. Kim, H. J., Howell, S. C., Van Horn, W. D., Jeon, Y. H., and Sanders, C. R. (2009) Recent Advances in the Application of Solution NMR Spectroscopy to Multi-Span Integral Membrane Proteins. Prog. Nucl. Magn. Reson. Spectrosc. 55, 335–360. 10. Warchawski, D. (2010) Membrane proteins of known structure determined by NMR. http:// www.drorlist.com/nmr/MPNMR.html. 11. Chill, J. H., Louis, J. M., Miller, C., and Bax, A. (2006) NMR study of the tetrameric KcsA potassium channel in detergent micelles. Protein Sci. 15, 684–698. 12. Liu, Y., Engelman, D. M., and Gerstein, M. (2002) Genomic analysis of membrane protein families: abundance and conserved motifs. Genome Biol. 3, research0054.

13. Marsden, R. L., Lee, D., Maibaum, M., Yeats, C., and Orengo, C. A. (2006) Comprehensive genome analysis of 203 genomes provides structural genomics with new insights into protein family space. Nucleic Acids Res. 34, 1066–1080. 14. Staley, J. P., and Kim, P. S. (1994) Formation of a native-like subdomain in a partially folded intermediate of bovine pancreatic trypsin inhibitor. Protein Sci. 3, 1822–1832. 15. Ma, C., Marassi, F. M., Jones, D. H., Straus, S. K., Bour, S., Strebel, K., Schubert, U., OblattMontal, M., Montal, M., and Opella, S. J. (2002) Expression, purification, and activities of full-length and truncated versions of the integral membrane protein Vpu from HIV-1. Protein Sci. 11, 546–557. 16. Zheng, H., Zhao, J., Wang, S., Lin, C. M., Chen, T., Jones, D. H., Ma, C., Opella, S., and Xie, X. Q. (2005) Biosynthesis and purification of a hydrophobic peptide from transmembrane domains of G-protein-coupled CB2 receptor. J. Pept. Res. 65, 450–458. 17. Diefenderfer, C., Lee, J., Mlyanarski, S., Guo, Y., and Glover, K. J. (2009) Reliable expression and purification of highly insoluble transmembrane domains. Anal. Biochem. 384, 274–278. 18. Call, M. E., Schnell, J. R., Xu, C., Lutz, R. A., Chou, J. J., and Wucherpfennig, K. W. (2006) The structure of the zetazeta transmembrane dimer reveals features essential for its assembly with the T cell receptor. Cell 127, 355–368. 19. Chong, Y. H., Ball, J. M., Issel, C. J., Montelaro, R. C., and Rushlow, K. E. (1991) Analysis of equine humoral immune responses to the transmembrane envelope glycoprotein (gp45) of equine infectious anemia virus. J. Virol. 65, 1013–1018. 20. Smith, J. G., Mothes, W., Blacklow, S. C., and Cunningham, J. M. (2004) The mature avian leukosis virus subgroup A envelope glycoprotein is metastable, and refolding induced by the synergistic effects of receptor binding and low pH is coupled to infection. J. Virol. 78, 1403–1410. 21. Bertrand, K., Squires, C., and Yanofsky, C. (1976) Transcription termination in vivo in the leader region of the tryptophan operon of Escherichia coli. J. Mol. Biol. 103, 319–337. 22. Miozzari, G. F., and Yanofsky, C. (1978) Translation of the leader region of the Escherichia coli tryptophan operon. J. Bacteriol. 133, 1457–1466. 23. Kleid, D. G., Yansura, D., Small, B., Dowbenko, D., Moore, D. M., Grubman, M. J., McKercher, P. D., Morgan, D. O., Robertson, B. H., and Bachrach, H. L. (1981) Cloned viral protein vaccine for foot-and-mouth disease: responses in cattle and swine. Science 214, 1125–1129.

10

Bacterial Production and Solution NMR Studies of a Viral…

24. Schnell, J. R., and Chou, J. J. (2008) Structure and mechanism of the M2 proton channel of influenza A virus. Nature 451, 591–595. 25. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J., and Bax, A. (1995) NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293. 26. Keller, R. (2004) The Computer Aided Resonance Assignment Tutorial, First Edition ed., CantinaVerlag, Goldau, Switzerland. http://cara.nmrsoftware.org/downloads/3-85600-112-3.pdf. 27. Cornilescu, G., Delaglio, F., and Bax, A. (1999) Protein backbone angle restraints from searching a database for chemical shift and sequence homology. J. Biomol. NMR 13, 289–302. 28. Schwieters, C. D., Kuszewski, J., Tjandra, N., and Clore, G. M. (2002) The Xplor-NIH NMR molecular structure determination package. J. Magn. Reson. 160, 66–74. 29. Zweckstetter, M., Bax, A. (2000) Prediction of sterically induced alignment in a dilute liquid crystalline phase: aid to protein structure determination by NMR. J. Am. Chem. Soc. 122, 3791–3792. 30. Salzmann, M., Wider, G., Pervushin, K., and Wuthrich, K. (1999) Improved sensitivity and coherence selection for [N-15,H-1]-TROSY elements in triple resonance experiments. J. Biomol. NMR 15, 181–184. 31. Kay, L. E., Ikura, M., Tschudin, R., and Bax, A. (1990) Three-dimensional triple resonance NMR spectroscopy of isotopically enriched proteins. J. Magn. Reson. 89, 496–514. 32. Neri, D., Szyperski, T., Otting, G., Senn, H., and Wuthrich, K. (1989) Stereospecific nuclear magnetic resonance assignments of the methyl groups of valine and leucine in the DNAbinding domain of the 434 repressor by biosynthetically directed fractional 13C labeling. Biochemistry 28, 7510–7516. 33. Grzesiek, S., Vuister, G. W., and Bax, A. (1993) A simple and sensitive experiment for measurement of JCC couplings between backbone carbonyl and methyl carbons in isotopically enriched proteins. J. Biomol. NMR 3, 487–493. 34. Vuister, G. W., Wang, A. C., and Bax, A. (1993) Measurement of three-bond nitrogen-carbon J couplings in proteins uniformly enriched in 15N and 13C. J. Am. Chem. Soc. 115, 5334–5335. 35. Bax, A., Vuister, G. W., Grzesiek, S., Delaglio, F., Wang, A. C., Tschudin, R., and Zhu, G. (1994) Measurement of homo- and heteronuclear J couplings from quantitative J correlation. Methods Enzymol. 239, 79–105. 36. MacKenzie, K. R., Prestegard, J. H., and Engelman, D. M. (1996) Leucine side-chain rotamers in a glycophorin A transmembrane

179

peptide as revealed by three-bond carbon-carbon couplings and 13C chemical shifts. J. Biomol. NMR 7, 256–260. 37. Hu, J.-S., Grzesiek, S., and Bax, A. (1997) Chi1 angle information from a simple two-dimensional NMR experiment which identifies trans 3JNCg couplings in isotopically enriched proteins. J. Biomol. NMR 9, 323–328. 38. Hu, J.-S., Grzesiek, S., and Bax, A. (1997) Two-dimensional NMR methods for determining c1 angles of aromatic residues in proteins from three-bond JC¢Cg and JNCgcouplings. J. Am. Chem. Soc. 119, 1803–1804. 39. Tycko, R., Blanco, F. J., and Ishii, Y. (2000) Alignment of biopolymers in strained gels: A new way to create detectable dipole-dipole couplings in high-resolution biomolecular NMR. J. Am. Chem. Soc. 122, 9340–9341. 40. Sass, H. J., Musco, G., Stahl, S. J., Wingfield, P. T., and Grzesiek, S. (2000) Solution NMR of proteins within polyacrylamide gels: diffusional properties and residual alignment by mechanical stress or embedding of oriented purple membranes. J. Biomol. NMR 18, 303–309. 41. Chou, J. J., Gaemers, S., Howder, B., Louis, J. M., Bax, A. (2001) A simple apparatus for generating stretched polyacrylamide gels, yielding uniform alignment of proteins and detergent micelles. J. Biomol. NMR 21, 377–382. 42. Weigelt, J. (1998) Single Scan, Sensitivity- and Gradient-Enhanced TROSY for Multidimensional NMR Experiments. J. Am. Chem. Soc. 120, 10778–10779. 43. Cornilescu, G., Marquardt, J. L., Ottiger, M., and Bax, A. (1998) Validation of protein structure from anisotropic carbonyl chemical shifts in a dilute liquid crystalline phase. J. Am. Chem. Soc. 120, 6836–6837. 44. Lakomek, N. A., Lange, O. F., Walter, K. F., Fares, C., Egger, D., Lunkenheimer, P., Meiler, J., Grubmuller, H., Becker, S., de Groot, B. L., and Griesinger, C. (2008) Residual dipolar couplings as a tool to study molecular recognition of ubiquitin. Biochem. Soc. Trans. 36, 1433–1437. 45. Loria, J. P., Rance, M., and Palmer, A. G. (1999) A relaxation-compensated Carr-PurcellMeiboom-Gill sequence for characterizing chemical exchange by NMR spectroscopy. J. Am. Chem. Soc. 121, 2331–2332. 46. Allerhand, A., and Thiele, E. (1966) Analysis of Carr-Purcell Spin-Echo NMR Experiments on Multiple-Spin Systems. II. The Effect of Chemical Exchange. J. Chem. Phys. 45, 902–916. 47. Golovanov, A. P., Hautbergue, G. M., Wilson, S. A., and Lian, L. Y. (2004) A simple method for improving protein solubility and long-term stability. J. Am. Chem. Soc. 126, 8933–8939.

Chapter 11 Preparation of the Modular Multi-Domain Protein RPA for Study by NMR Spectroscopy Chris A. Brosey, Marie-Eve Chagot, and Walter J. Chazin Abstract The integrity and propagation of the genome depend upon the fidelity of DNA processing events, such as replication, damage recognition, and repair. Requisite to the numerous biochemical tasks required for DNA processing is the generation and manipulation of single-stranded DNA (ssDNA). As the primary eukaryotic ssDNA-binding protein, Replication Protein A (RPA) protects ssDNA templates from stray nuclease cleavage and untimely reannealment. More importantly, RPA also serves as a platform for organizing access to ssDNA for readout of the genetic code, recognition of aberrations in DNA, and processing by enzymes. We have proposed that RPA’s ability to adapt to such a broad spectrum of multiprotein machinery arises in part from its modular organization and interdomain flexibility. While requisite for function, RPA’s modular flexibility has presented many challenges to providing a detailed characterization of the dynamic architecture of the full-length protein. To enable the study of RPA’s interdomain dynamics and responses to ssDNA binding by biophysical methods including NMR spectroscopy, we have successfully produced recombinant full-length RPA in milligram quantities at natural abundance and enriched with NMR-active isotopes. Key words: Replication Protein A, DNA processing, Protein modularity, Isotopic labeling, Recombinant expression, Protein purification, NMR spectroscopy

1. Introduction As the primary eukaryotic single-stranded DNA (ssDNA)-binding protein, Replication Protein A (RPA) prevents reannealment of unwound DNA strands, controls access to DNA templates, and serves as a scaffold for the assembly and disassembly of DNA processing machinery (1, 2). A heterotrimer, RPA’s three subunits (RPA70, RPA32, and RPA14) contain seven structured domains interconnected by flexible linkers. Three of these domains form

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_11, © Springer Science+Business Media, LLC 2012

181

182

C.A. Brosey et al.

the trimeric core of the protein (70C, 32D, 14), from which emanate the flexibly linked N-terminal domains of RPA70 (70N, 70A, 70B), as well as the disordered N-terminal and structured C-terminal domains of RPA32 (32N and 32C, respectively). Binding of ssDNA is facilitated by domains 70A, 70B, 70C, and 32D, which together occupy an occluded site size of 30 nucleotides (1). Interactions with other DNA-processing proteins are primarily mediated by domains 70N and 32C, and the principal DNA-binding domains 70A and 70B (1, 2). As a universal participant in DNA processing, RPA must interact with a wide array of structurally unique multiprotein complexes. The flexible, modular organization of the protein is thought to be critical for enabling such structural adaptability (3). Although high-resolution X-ray or NMR structures of all individual RPA domains are available (4–8), the dynamic interdomain organization of full-length RPA and the accompanying structural alterations imposed by DNA processing have not been extensively characterized. The full-length protein’s intrinsic flexibility poses several challenges to study by X-ray diffraction; and at 116 kDa, RPA falls outside the size limit of conventional NMR methods (30–40 kDa). Application of advanced NMR approaches, however, namely, deuterium labeling and TROSY- or CRINEPT-based techniques, has allowed this size limitation to be extended to proteins in excess of 100 kDa (9–12). This, combined with the discrete distribution of molecular mass among RPA domains (50 kDa for the trimer core and 10–14 kDa for the remaining domains), makes feasible characterization of the full-length protein by NMR (13). Here, we describe the production of full-length RPA by recombinant expression in Escherichia coli and subsequent purification of the protein by a series of FPLC steps. The protocols provided include those required for preparation of 2H-, 15N-enriched RPA for study by NMR spectroscopy.

2. Materials 2.1. Cell Transformation

1. RPA pET15b plasmid (see Note 1). 2. BL21(DE3) pLyS competent cells: 100-μL aliquots stored at −80°C. 3. LB medium plates: 10 g/L tryptone, 10 g/L NaCl, 5 g/L yeast extract, 15 g/L agar dissolved in Milli-Q water (filtered to a resistance of 18.3 MΩ-cm) and autoclaved at 121°C for 15 min. Add antibiotic stocks (ampicillin and chloramphenicol) at 1:1,000 dilution when the medium has cooled to 50–60°C (14) (see Note 2).

11

Preparation of the Modular Multi-Domain Protein RPA…

183

4. 1,000× Ampicillin stock: 100 mg/mL in Milli-Q water, sterilize by filtration at 0.2 μm (see Note 3). 5. 1,000× Chloramphenicol stock: 34 mg/mL in ethanol, sterilize by filtration at 0.2 μm (see Note 3). 6. SOC recovery medium: 20 g/L tryptone, 0.5 g/L NaCl, 5 g/L yeast extract, 2.5 mM KCl, 5 mM MgCl2, 5 mM MgSO4, 20 mM glucose dissolved in Milli-Q water and autoclaved at 121°C for 15 min (14). 2.2. Cell Expression Testing

1. RPA pET15b BL21(DE3) pLysS LB plate (Subheading 2.1). 2. Sterile 10-mL test culture tubes. 3. LB medium: 10 g/L tryptone, 10 g/L NaCl, 5 g/L yeast extract dissolved in Milli-Q water and autoclaved at 121°C for 15 min. 4. 1,000× Antibiotic stocks (Subheading 2.1). 5. 1 M IPTG: Sterilize by filtration at 0.2 μm and store at −20°C. 6. 2× SDS loading buffer: 100 mM Tris–HCl, pH 6.8, 4% (w/v) SDS (electrophoresis grade), 0.2% (w/v) bromophenol blue, 20% (w/v) glycerol, 200 mM β-mercaptoethanol (βME, added fresh) (14). 7. 8 M urea. 8. Precast 4–12% Bis-Tris SDS-PAGE gel (Invitrogen). 9. 1× MES SDS running buffer (Invitrogen). 10. 1× Prestained molecular weight standards. 11. SimplyBlue SafeStain.

2.3. Preparation of Culture Media

1. LB medium (Subheading 2.2).

2.3.1. LB Medium

3. 500-mL Erlenmeyer flask with baffles.

2. 1,000× Antibiotic stocks (Subheading 2.1). 4. Six 2.8-L Fernbach flasks with baffles.

2.3.2. Minimal Medium

1. Milli-Q water (900 mL/L medium). 2. 10× M9 salts: 5 g/L NaCl, 30 g/L KH2PO4, 60 g/L Na2HPO4 dissolved in Milli-Q water, adjusted to pH 7.4 with 10 M NaOH, and autoclaved at 121°C for 15 min. 3. 1 M MgSO4: Sterilize filter at 0.2 μm and store at room temperature. 4. 1 M CaCl2: Sterilize filter at 0.2 μm and store at room temperature.

184

C.A. Brosey et al.

5. 20% (w/v) Glucose: Sterilize filter at 0.2 μm and store at room temperature. 6. 1 M Thiamine hydrochloride: Sterilize filter at 0.2 μm and store at room temperature. 7. 1,000× Antibiotic stocks (Subheading 2.1). 8.

15

NH4Cl.

9. 500-mL Erlenmeyer flask with baffles. 10. Six 2.8-L Fernbach flasks with baffles. 2.3.3. Deuterated Minimal Medium

1. 99% D2O. 2. Dry components: 0.5 g/L NaCl, 3 g/L KH2PO4, 6 g/L Na2HPO4, 0.24 g/L-MgSO4, 11.1 mg/L-CaCl2, 2 g/L-glucose, 0.337 g/L thiamine hydrochloride, 0.5 g/L 15NH4Cl, and 0.1 g/L ampicillin. 3. Six sterile vacuum filtration systems with 1-L storage containers. 4. Six sterile 2.8-L Fernbach flasks with baffles.

2.4. Starter Cultures

1. RPA pET15B BL21(DE3) pLysS LB plate (Subheading 2.1).

2.4.1. LB Medium

2. 250 mL LB starter culture (Subheading 2.3.1).

2.4.2. Minimal Medium and Deuterated Minimal Medium

1. RPA pET15B BL21(DE3) pLysS LB plate (Subheading 2.1). 2. Sterile 10-mL test culture tubes. 3. LB medium (Subheading 2.2). 4. 1,000× Antibiotic stocks (Subheading 2.1). 5. 250 mL Minimal medium starter culture (Subheading 2.3.2).

2.5. Large-Scale Cell Culture and Overexpression

1. Starter culture (Subheading 2.4). 2. 6 L Sterile media in Fernbach flasks (Subheading 2.3). 3. 1 M IPTG (Subheading 2.2). 4. Bleach or 1% Terg-a-Zyme solution for decontamination of spent media.

2.6. RPA Purification 2.6.1. Cell Lysis

1. Lysis buffer: Dissolve two complete EDTA-free protease inhibitor cocktail tablets (Roche) in 80 mL of Ni-NTA buffer A (Subheading 2.6.2) in a 150-mL glass beaker on ice immediately prior to use. 2. 100-mL Glass homogenizer. 3. Sonic dismembrator. 4. 25-mm diameter, 0.45-μm syringe filter.

2.6.2. Ni-NTA Chromatography

1. Refrigerated Äkta FPLC purification system and accessories. 2. Ni-NTA buffer A: 20 mM HEPES, pH 7.5, 500 mM NaCl, 5 mM βME, 10 μM ZnCl2, 10 mM imidazole; adjusted to

11

Preparation of the Modular Multi-Domain Protein RPA…

185

target pH with concentrated HCl, filtered at 0.45 μm under vacuum, and stored at 4°C (see Notes 4–7). 3. Ni-NTA buffer B: 20 mM HEPES, pH 7.5, 500 mM NaCl, 5 mM βME, 10 μM ZnCl2, 300 mM imidazole; adjusted to target pH with concentrated HCl, filtered at 0.45 μm under vacuum, and stored at 4°C (see Notes 4–7). 4. 25 mL Ni-NTA pre-packed FPLC column (Sigma–Aldrich). 2.6.3. Desalting Exchange and Heparin Chromatography

1. Centrifugal concentrators (15 mL, 30 kDa MWCO). 2. Refrigerated Äkta FPLC purification system and accessories. 3. Heparin buffer A: 20 mM HEPES, pH 7.5, 50 mM NaCl, 5 mM βME, 10 μM ZnCl2, 10% glycerol; adjusted to target pH with concentrated HCl, filtered at 0.45 μm under vacuum, and stored at 4°C (see Notes 6 and 7). 4. Heparin buffer B: 20 mM HEPES, pH 7.5, 1 M NaCl, 5 mM βME, 10 μM ZnCl2, 10% glycerol; adjusted to target pH with concentrated HCl, filtered at 0.45 μm under vacuum, and stored at 4°C (see Notes 6 and 7). 5. HiPrep 26/10 Desalting column (GE Healthcare). 6. HiTrap 5-mL Heparin HP column (GE Healthcare).

2.6.4. Superdex 200 Gel Filtration Chromatography

1. Centrifugal concentrators (15 mL, 30 kDa MWCO). 2. 0.22-μm centrifugal spin filters. 3. Refrigerated Äkta FPLC purification system and accessories. 4. Gel filtration buffer: 20 mM HEPES, pH 7.5, 100 mM NaCl, 5 mM βME, 10 μM ZnCl2, 200 mM arginine; adjusted to target pH with concentrated HCl, filtered at 0.45 μm under vacuum, and stored at 4°C (see Notes 6 and 7). 5. Superdex 200 HR 10/30 column (GE Healthcare).

2.7. Preparation of Samples for NMR

1. Centrifugal concentrators (15 mL, 30 kDa MWCO). 2. 5- and 4-mm NMR tubes. 3. 99% D2O.

3. Methods This section describes a protocol for the production of full-length RPA in E. coli and its subsequent purification, including preparation of 2H-,15N-enriched protein for study by NMR. As a rule, robust expression of full-length RPA in E. coli is challenging as RPA is a relatively large protein (>100 kDa) and its ssDNA binding properties are toxic to bacterial cells. Average yields of RPA overexpressed from the pET15b vector range from 17 mg for 6 L of

186

C.A. Brosey et al.

rich LB culture to 4–8 mg of 15N-enriched RPA for 6 L of minimal medium culture (1–2 NMR samples at approximately 130 μM concentration and 260-μL volume). Working with RPA cultures grown in deuterated minimal medium requires patience and careful monitoring, as a deuterated environment is particularly stressful to the bacterial metabolism. Consequently, deuterated cultures take much longer to reach their target induction densities and usually result in a diminished yield of the recombinant protein. Due to the high cost of D2O and time investment required for growth and expression, we found it beneficial to pilot a small-scale test culture (100 mL) to develop an expected timeline for the growth and to ensure that all reagents were functioning as expected. Subsequent purification of this culture allowed us to determine that the overall yield of RPA had not suffered significantly from production in a deuterated environment. In the expression protocol below, we describe the full 6 L production run; however, when embarking upon deuterium labeling for the first time, we highly recommend starting with the smaller pilot culture. As RPA is a trimeric, DNA-binding protein, the purification protocol below is designed to ensure samples free of contaminating ssDNA, as well as uniform stoichiometry among all three RPA subunits. Expression of RPA from the pET15b vector results in an excess of the RPA70 subunit, which can be successfully separated from the intact heterotrimer with heparin and gel filtration chromatography. The heparin purification step also selects for RPA free from ssDNA contamination. 3.1. Cell Transformation

Ensuring robust antibiotic selection of the RPA vector on solid medium is vital to enabling the success of subsequent liquid cultures (see Note 2). 1. Thaw 100 μL of BL21(DE3) pLyS competent cells on ice, gently combine with 100 ng of RPA pET15b vector, and incubate for 30 min on ice. 2. Heat shock the cells at 42°C for 45 s and incubate on ice for 2 min. Add 900 μL of sterile SOC recovery medium. 3. Incubate the cells for 1 h at 37°C and 200–230 rpm, then centrifuge the cells at 16,100 × g for 1 min at room temperature (see Note 8). Remove 900 μL of the clarified SOC medium and gently resuspend the cells in the remaining medium prior to plating on an LB medium plate. 4. Incubate the plates overnight at 37°C.

3.2. Cell Expression Testing

Expression testing allows confirmation of the expression capability of the transformed bacterial colonies prior to scaling up protein production. The testing also allows for selection of colonies with the most robust expression.

11

Preparation of the Modular Multi-Domain Protein RPA…

187

1. Prepare five LB test cultures as follows: Transfer 5 mL of LB medium to a 10-mL sterile culture tube and add 1:1,000 dilutions of ampicillin and chlormaphenicol antibiotic stocks. Inoculate each culture with a colony selected from the center of a freshly transformed RPA pET15b LB plate (see Note 9) and incubate the cultures at 37°C, 200–230 rpm, until they reach an A600 of 0.5–0.6 (approximately 3–4 h). 2. Transfer 250 μL of each LB test culture into an Eppendorf tube as a preinduction sample. Centrifuge the sample at 16,100 × g for 1 min, decant the supernatant, add 7 μL each of 2× SDSPAGE loading buffer and 8 M urea, and vortex to mix. 3. Add IPTG to a final concentration of 1 mM to the remainder of the test culture to induce expression and continue to incubate with shaking at room temperature for 3 h. Collect a final 250 μL postinduction sample and process as in step 2. 4. Boil pre- and postinduction SDS-PAGE samples for 5–10 min to denature the lysates and load 2–4 μL from each sample into a precast 4–12% Bis–Tris SDS-PAGE gel preloaded into an electrophoresis cell filled with 1× MES SDS running buffer. Reserve one lane for 5 μL of 1× prestained molecular weight standards. Run the gel at 200 V. 5. Remove the gel from the electrophoresis cell and place in a loosely capped container filled with Milli-Q water (see Note 10). Fix the gel by heating on a high setting in a microwave oven for 1 min, followed by 1 min of cooling. Exchange the water and repeat. Remove the final rinse and stain with SimplyBlue SafeStain for 20 min. Remove the stain and refill the container with deionized water to destain the gel. 6. The relative proportion of RPA32 and RPA14 subunits is too low to observe on the gel. RPA70 should be just distinguishable at the appropriate molecular weight in lanes containing postinduction samples. Select colonies exhibiting the most abundant RPA production for subsequent largescale expression. 3.3. Preparation of Culture Media 3.3.1. LB Medium

This medium serves for the production of unlabeled RPA. Preparation should include a 250-mL starter culture to accommodate a 6-L large-scale culture. 1. Dissolve LB components (Subheading 2.2) in Milli-Q water in a 500-mL baffled Erlenmeyer flask (starter culture) or 2.8-L baffled Fernbach flasks (large-scale culture), autoclave at 121°C for 15 min, and cool to 50–60°C. 2. Add ampicillin and chloramphenicol at 1:1,000 dilution immediately prior to inoculation.

188

C.A. Brosey et al.

3.3.2. Minimal Medium

This medium serves for the production of 15N-enriched RPA. Preparation should include a 250 mL starter culture to accommodate a 6 L large-scale culture. A 250 mL minimal medium culture also serves as an adaptation culture for production of deuterated protein. 1. Dilute 10× M9 salts in Milli-Q water to 1× in a 500-mL baffled Erlenmeyer flask (starter culture) or 2.8-L baffled Fernbach flasks (large-scale culture), autoclave at 121°C for 15 min, and cool to 50–60°C. 2. Add the following components immediately prior to inoculation: 0.5 g/L of 15NH4Cl (see Note 11), 2 mL/L of 1 M MgSO4, 100 μL/L of 1 M CaCl2, 10 mL/L of 20% glucose, 1 mL/L of 1 M thiamine hydrochloride, and antibiotic stocks at 1:1,000 dilution.

3.3.3. Deuterated Minimal Medium

This medium serves for six 1 L large-scale production cultures and is prepared immediately prior to inoculation after the success of the 250 mL minimal medium adaptation culture has been ascertained. 1. Dry autoclave six 2.8-L Fernbach flasks with baffles and allow to dry thoroughly overnight. 2. Dissolve all dry components in 6 L of 99% D2O (see Notes 12 and 13) and immediately sterilize the medium in 1-L batches by using sterile vacuum filtration systems (i.e., 1 L/unit). This apparatus filters the medium directly into a sterile 1-L bottle. Chloramphenicol is not included in the medium at this stage to ease the metabolic burden on the cells. 3. Carefully transfer each 1 L of sterile deuterated minimal medium to a dry, sterile Fernbach flask (see Note 14).

3.4. Starter Cultures 3.4.1. LB Medium

1. Inoculate a 250 mL LB starter culture (Subheading 3.3.1) directly with an RPA pET15b colony selected from the test expression. 2. Grow the culture overnight at 37°C, 200–230 rpm. The culture should be cloudy in the morning.

3.4.2. Minimal Medium and Deuterated Minimal Medium

1. Inoculate a 4 mL LB starter culture (4 mL LB + 1:1,000 ampicillin/chloramphenicol in a 10-mL culture tube) with an RPA pET15b colony selected from the test expression. Grow for 3–4 h at 37°C, 200–230 rpm, or until cloudy. 2. Inoculate the 250 mL minimal medium starter culture (Subheading 3.3.2) with the 4 mL LB starter culture and shake overnight at 37°C. The culture should be cloudy in the morning.

11

3.5. Large-Scale Cell Culture and Overexpression

Preparation of the Modular Multi-Domain Protein RPA…

189

1. Prepare six 1 L cultures of rich LB medium, minimal medium, or deuterated minimal medium as described above (Subheading 3.3) and inoculate each with 30 mL (40 mL for deuterated minimal medium) of the corresponding overnight starter culture. 2. Grow the cultures at 37°C, 200–230 rpm, until an A600 of 0.6–0.7 is reached (see Note 15). 3. Allow the cultures to equilibrate for half an hour with agitation at 18°C (or room temperature for deuterated minimal medium) prior to induction. Collect a preinduction SDS-PAGE sample as described above (Subheading 3.2) and induce the cells with 1 mM IPTG. Allow cells to express overnight (approximately 16–18 h). 4. The A600 at the end of the expression period should be 1.8–2.0 for LB medium cultures and 0.9–1.0 for standard and deuterated minimal media cultures. Collect postinduction SDS-PAGE samples from the cultures as described above (Subheading 3.2). Harvest the cultures by centrifuging at 10,000 × g for 20 min at 4°C. 5. Decant the supernatant and reserve the spent deuterated media for recycling (15). Spent LB or minimal media may be decontaminated by the addition of bleach or a 1% Terg-a-zyme solution for 30 min, and then discarded. If purification does not follow immediately, transfer the pellets to sterile 50-mL conical tubes and freeze at −80°C. Run pre- and postinduction SDSPAGE samples as in Subheading 3.2 to confirm the presence of RPA expression.

3.6. RPA Purification

3.6.1. Cell Lysis

Purification of RPA involves three primary steps: Ni-NTA affinity, heparin, and size-exclusion chromatography. For best results, the protocol should be completed over the course of 2 days, where Ni-NTA and heparin chromatography steps are accomplished the first day and the final gel filtration step is carried out on the second day. If necessary, the Ni-NTA and heparin steps may be divided into two separate days and protein fractions from each purification kept at 4°C overnight. Ideally, though, the time from cell lysis to the final gel filtration exchange should be kept to a minimum. 1. If cells have been frozen at −80°C, thaw the pellets by submerging the 50-mL conical tubes in cool water. Meanwhile, prepare and chill the lysis buffer and pre-chill the 100-mL glass homogenizer on ice (see Note 16). 2. Transfer all 6 L of RPA cell pellets into the 100-mL homogenizer, rinse the 50-mL conical tubes with ice-cold lysis buffer, and add the rinse and any remaining lysis buffer to the homogenizer.

190

C.A. Brosey et al.

3. Homogenize the lysate until smooth (10–15 strokes). 4. Return the lysate to the 150-mL glass beaker and pack this into a 2-L plastic beaker filled with an ice-water bath (see Note 17). Ensure that there is sufficient ice to securely brace the beaker and prevent floating. 5. Sonicate the lysate with a macrotip set at 60% power for 5.0 min of total process time (pulsing 5.0 s on and 5.0 s off). Pause the sonicator half-way through this cycle to replenish the ice-water bath and to check the temperature of the lysate (see Note 16). The lysate should become translucent and less viscous as the sonicator disrupts the cellular material. If the lysate viscosity remains unchanged after the cycle is complete, repeat the cycle once more, monitoring the ice-water bath and lysate temperature. 6. Clarify the lysate by centrifuging at 48,000 × g for 20 min at 4°C. Ensure that both centrifuge and rotor are pre-chilled to at least 4°C. 7. Decant and filter the clarified supernatant through a 0.45-μm membrane. Store on ice for immediate loading onto the FPLC Ni-NTA column. 3.6.2. Ni-NTA Chromatography

Steps 2–4 are implemented as a pre-programmed Äkta FPLC method. 1. Equilibrate the prepacked 25 mL Ni-NTA column with three column volumes (3 CVs) each of filtered Milli-Q water and Ni-NTA buffer A (see Note 18). 2. Load the filtered lysate onto the equilibrated Ni-NTA column at 1.0–1.5 mL/min. 3. Wash unbound lysate from the column with 4 CVs of Ni-NTA buffer A at 2.5 mL/min. 4. Elute RPA with a 4 CV gradient (0–100% Ni-NTA buffer B, 10–300 mM imidazole), collecting 6-mL fractions at 2.5 mL/min. 5. Assess the presence of RPA from the A280 chromatogram trace and SDS-PAGE of relevant fractions (sampling 5 μL of each fraction). Pool fractions containing all three RPA subunits for further processing (typically, a 60-mL pool).

3.6.3. Desalting Exchange and Heparin Chromatography

The charged DNA-binding clefts of RPA render it sensitive to the absence of ambient salt. Effective binding of RPA to the heparin matrix, however, requires a low salt content in the loading buffer. Direct dialysis into the loading buffer (heparin buffer A) usually provokes extensive precipitation of RPA. Buffer exchange by desalting, however, allows for the rapid and successful transfer of RPA into the loading buffer with minimal aggregation. Once the series of FPLC desalting runs are complete, it is imperative to load the exchanged protein directly onto the heparin column to restore a stabilizing ionic environment. As before, loading, washing, and

11

Preparation of the Modular Multi-Domain Protein RPA…

191

elution are implemented automatically using pre-programmed Äkta FPLC methods. 1. Pre-rinse two centrifugal concentrators (15 mL, 30 kDa MWCO) with Milli-Q water by centrifuging at 3,700 × g for 10 min at 4°C. Concentrate the Ni-NTA RPA pool to ~30 mL and store on ice for desalting into heparin buffer A (see Notes 19 and 20). 2. Equilibrate the HiPrep 26/10 Desalting column with 2 CVs of filtered Milli-Q water and 1.5 CVs of heparin buffer A. Equilibrate the HiTrap 5 mL Heparin HP column with 5 CVs of filtered Milli-Q water and 3 CVs of heparin buffer A (see Note 21). 3. Filter and load 10 mL of the Ni-NTA RPA concentrate onto the desalting column at 2.0 mL/min, collecting 4-mL fractions. The protein should elute within the first four fractions (16 mL) of the run. Re-equilibrate the column and repeat the run twice for the remaining 20 mL of RPA Ni-NTA concentrate. Store fractions on ice until all runs are complete. 4. Combine all three desalting pools (48 mL), filter at 0.45 μm, and load directly onto the equilibrated heparin column at 1.0 mL/min. 5. Wash out unbound sample with 3 CVs of heparin buffer A at 2.5 mL/min. 6. Elute RPA with a 20 CV gradient (0–100% heparin buffer B, 50 mM to 1 M NaCl), collecting 4-mL fractions at 2.5 mL/min. 7. Assess the presence of RPA from the A280 chromatogram trace and SDS-PAGE of relevant fractions (sampling 5 μL of each fraction). The elution should include two major peaks: the first corresponding to RPA70 exclusively and the second to trimeric RPA. Fractions containing all three RPA subunits are pooled for further final purification (typically, a 40-mL pool). 3.6.4. Superdex 200 Gel Filtration Chromatography

Gel filtration provides a final polishing step for the purification and ensures the removal of any trace RPA70 or low-molecular-weight contaminants. As before, loading and elution are implemented automatically using a pre-programmed Äkta FPLC method. 1. Equilibrate the Superdex 200 HR 10/30 column with 1.5 CVs of filtered Milli-Q water and 1.5 CVs of gel filtration buffer. 2. Pre-rinse two centrifugal concentrators (15 mL, 30 kDa MWCO) with Milli-Q by centrifuging at 3,700 × g for 10 min at 4°C. Concentrate the heparin RPA pool to ~300–500 μL (see Note 19). 3. Filter the concentrate by using a 0.22-μm centrifugal spin filter in a refrigerated (4°C) centrifuge and load onto the Superdex 200 HR 10/30 column at 0.3 mL/min, collecting 0.5-mL fractions for 1.5 CVs.

192

C.A. Brosey et al.

4. As before, assess the presence of RPA from the A280 trace and SDSPAGE of relevant fractions (sampling 5 μL of each fraction). 5. Before the final RPA fractions are pooled, acquire final A280 and A260 measurements by UV-Vis spectrophotometry to ensure that the selected fractions are DNA-free, as determined by A260/A280 ratios of 0.64 or less. 3.7. Preparation of Samples for NMR

1. Pre-rinse two centrifugal concentrators (15 mL, 30 kDa MWCO) with Milli-Q water by centrifuging at 3,700 × g for 10 min at 4°C. Concentrate the S200 RPA pool to ~500– 600 μL (see Note 19), monitoring the protein concentration by UV-Vis spectrophotometry (the gel filtration buffer may serve as a blank). 2. The target NMR concentration for RPA is ~100–130 μM (10–15 mg/mL) with a minimum sample volume of 260 μΛ (using a 4-mm-diameter NMR tube). Once the target concentration is reached, centrifuge 260–300 μL of the concentrate at 16,100 × g for 5 min at 4°C to remove any stray precipitation. 3. Load the protein into a standard 4-mm diameter NMR tube, which may be fitted with an adaptor to fit a 5-mm spinner or slipped into a 5-mm diameter tube containing 120 μL D2O without the need for the adaptor.

4. Notes 1. The tricistronic RPA pET15b vector was a gift from the lab of Alexey Bochkarev. 6×-His tags with thrombin cleavage sites precede the RPA70 and RPA14 subunits. The order of subunit open reading frames (ORFs) is as follows: RPA70, RPA14, RPA32. 2. Using freshly prepared ampicillin stock in the LB medium is important for ensuring robust RPA transformation. Even though the choice of the BL21(DE3) pLysS cell line is designed to circumvent leaky expression, even small amounts of noninduced RPA can potentially result in resistant cells with less than robust expression. 3. Antibiotic stocks may be aliquoted and stored at −20°C for future use. For long-term storage of ampicillin stocks, storage at −80°C is recommended. 4. The 70C domain of RPA contains a zinc-binding motif, for which ZnCl2 is included in the purification buffers. 5. The most effective imidazole concentrations in the Ni-NTA buffers will depend on how recently the Ni-NTA resin has been charged. For freshly charged resin, the imidazole concentration

11

Preparation of the Modular Multi-Domain Protein RPA…

193

for Ni-NTA buffer A is often raised to 30 mM for the first few purifications to compensate for the higher nonspecific affinity of the resin. 6. βME is added fresh to each buffer immediately prior to use. Buffers can be prepared without βME and stored at 4°C if their use is anticipated to last beyond 1–2 days. 7. Preparation of 1 L of each buffer should provide more than enough for the entire purification. 8. As the transformation efficiency of the RPA pET15b vector is low and the double-antibiotic selection with the BL21(DE3) pLysS strain is quite stringent, plating the entire transformation culture is recommended. 9. When transferring a selected colony to the test expression culture, be sure to leave a portion behind for future inoculation of the large-scale cultures. If the colony is too small to divide in this manner, the plate may be left at room temperature for half a day to allow the colony to regrow. 10. An empty gel tip box will suffice. The level of water should be enough to immerse the gel. 11. The 15NH4Cl is measured out and added directly to the sterile medium as a powder. This ensures that the labeled material is not wasted should a step prior to the inoculation fails. 12. As mentioned at the beginning of Subheading 3, the volume of deuterated minimal medium may be adjusted for small-scale testing. 13. Mixing of the deuterated minimal medium dry components should take place in a clean, dry container and be carried out as efficiently as possible to prevent exchange with ambient water vapor. If a container large enough to accommodate 6-L volume is not available, substitution of two 3-L containers with subsequent exchange and mixing between the two batches can be used to ensure the homogeneity of the medium across all cultures. 14. Performing this sterile transfer in the presence of a Bunsen burner flame is advised. 15. In our experience, growth in rich LB medium requires 3–4 h to reach the target A600 while growth in minimal medium requires 8–10 h. For deuterated minimal medium, this timeline extends to 1.5–2 days. To ensure that the induction occurred during daylight hours, cultures were switched to agitation at 20°C overnight after a full day of growth at 37°C and then returned to 37°C the next morning. The target A600 was reached during the afternoon of the second day. 16. RPA is susceptible to proteolytic cleavage, particularly at the unstructured 60–70 amino-acid linker that connects the 70N and 70A domains. Throughout the purification, it is essential

194

C.A. Brosey et al.

that all buffers are kept ice cold and that heating from other steps of the lysis (sonication, centrifugation) is kept to a minimum. 17. The ice-water bath serves as a heat sink during sonication of the lysate. 18. Initializing the FPLC system (pump washing, cleaning superloops, setting up fraction collectors), as well as equilibration of the Ni-NTA column, should occur prior to or concurrently with cell lysis to ensure that the clarified, filtered lysate can be loaded directly into the system as soon as it is available. 19. Centrifugal concentrators are usually spun in 10–15-min increments and carefully mixed with each addition of the Ni-NTA pool to prevent buildup and aggregation of RPA at the base of the concentrator. 20. The desalting resolution of the HiPrep 26/10 Desalting column (GE Healthcare) is 10 mL; that is, the column can effectively exchange 10 mL of injected sample into the target buffer without contamination from the original buffer. To avoid desalting too concentrated a volume of RPA and triggering aggregation, the Ni-NTA pool is processed in three sequential 10-mL batches. 21. As with the Ni-NTA step, initializing the FPLC system and equilibrating the desalting and heparin columns should occur prior to or concurrently with concentrating the RPA Ni-NTA pool to allow for immediate loading once the target volume is reached.

Acknowledgments The authors would like to thank Dr. Dalyir Pretto and Susan Meyn. This work was supported by the National Institutes of Health operating grant R01 GM65484 and graduate training grant T32 GM08320. References 1. Wold, M. S. (1997) Replication protein A: A heterotrimeric, single-stranded DNA-binding protein required for eukaryotic DNA metabolism. Annu. Rev. Biochem. 66, 61–92. 2. Fanning, E., Klimovich, V., and Nager, A. R. (2006) A dynamic model for replication protein A (RPA) function in DNA processing pathways. Nuc. Acids Res. 34, 4216–4137. 3. Stauffer, M. E., and Chazin, W. J. (2004) Structural mechanisms of DNA replication,

repair, and recombination. J. Biol. Chem. 279, 30915–30918. 4. Bochkarev, A., Pfuetzner, R. A., Edwards, A. M., and Frappier, L. (1997) Structure of the singlestranded-DNA-binding domain of replication protein A bound to DNA. Nature 385, 176–181. 5. Jacobs, D. M., Lipton, A. S., Isern, N. G., Daughdrill, G. W., Lowry, D. F., Gomes, X., and Wold, M. S. (1999) Human replication

11

6.

7.

8.

9.

Preparation of the Modular Multi-Domain Protein RPA…

protein A: Global fold of the N-terminal RPA70 domain reveals a basic cleft and flexible C-terminal linker. J. Biomol. NMR 14, 321–331. Bochkarev, A., Bochkareva, E., Frappier, L., and Edwards, A. M. (1999) The crystal structure of the complex of replication protein A subunits RPA32 and RPA14 reveals a mechanism for single-stranded DNA binding. EMBO J. 18, 4498–4504. Mer, G., Bochkarev, A., Gupta, R., Bochkareva, E., Frappier, L., Ingles, C. J., Edwards, A. M., and Chazin, W. J. (2000) Structural basis for the recognition of DNA repair proteins UNG2, XPA, and RAD52 by replication factor A. Cell 103, 449–456. Bochkareva, E., Korolev, S., Lees-Miller, S. P., and Bochkarev, A. (2002) Structure of the RPA trimerization core and its role in the multistep DNA-binding mechanism of RPA. EMBO J. 21, 1855–1863. Pervushin, K., Riek, R., Wider, G., and Wuthrich, K. (1997) Attenuated T2 relaxation by mutual cancellation of dipole-dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution. Proc. Natl. Acad. Sci. U.S.A. 94, 12366–12371.

195

10. Riek, R., Wider, G., Pervushin, K., and Wuthrich, K. (1999) Polarization transfer by cross-correlated relaxation in solution NMR with very large molecules. Proc. Natl. Acad. Sci. U.S.A. 96, 4918–4923. 11. Riek, R., Pervushin, K., and Wuthrich, K. (2000) TROSY and CRINEPT: NMR with large molecular and supramolecular structures in solution. Trends Biochem. Sci. 25, 462–468. 12. Tugarinov, V., Hwang, P. M., and Kay, L. E. (2004) Nuclear magnetic resonance spectroscopy of high-molecular-weight proteins. Annu. Rev. Biochem. 73, 107–146. 13. Brosey, C. A., Chagot, M. E., Ehrhardt, M., Pretto, D. I., Weiner, B. E., and Chazin, W. J. (2009) NMR analysis of the architecture and functional remodeling of a modular multidomain protein, RPA. J. Am. Chem. Soc. 131, 6346–6347. 14. Sambrook J and Russell D. (2001) Molecular cloning: A laboratory manual, vol. 3, 3rd ed. Cold Spring Harbor Laboratory Press, New York. 15. Li, M. X., Corson, D. C., and Sykes, B. D. (2002) Structure determination by NMR isotope labeling. Meth. Mol. Biol. 173, 255–265.

Chapter 12 NMR Studies of Protein–RNA Interactions Carla A. Theimer, Nakesha L. Smith, and May Khanna Abstract This chapter describes the preparation of NMR quantities of RNA purified to single-nucleotide resolution for protein–RNA interaction studies. The protocol is easily modified to make nucleotide-specific isotopically labeled RNAs or uniformly labeled RNA fragments for ligation to generate segmentally labeled RNAs. Key words: In vitro transcription, Single-nucleotide resolution, RNA synthesis, Protein–RNA interactions, RNA purification, Isotopic labeling

1. Introduction Understanding RNA–protein interactions and the structures of RNA–protein complexes is an important avenue of research. Ribonucleoprotein (RNP) complexes and RNA–protein interactions have been found to be central to many biological processes, ranging from genomic stability through telomeric maintenance (1–3), and alternative splicing of RNA by the spliceosome (4), to protein synthesis by the ribosome (5–8). Additionally, exciting new discoveries regarding small, noncoding RNAs, including snoRNAs, snRNAs, and RNA interference, continue to unfold. Many of these RNAs are associated with proteins at various stages of maturation, from trafficking to processing to their final destination, and any defect in these pathways can lead to serious diseases (9). Thus, dissecting the structure of RNA–protein interactions is crucial for deciphering the roles of these RNP complexes as well as identifying potential targets for pharmaceutical intervention. Protocols are well-established and documented for the generation of sufficient quantities of 13C-,15N-isotopically labeled proteins

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_12, © Springer Science+Business Media, LLC 2012

197

198

C.A. Theimer et al.

from bacterial sources for NMR studies; the purification of such proteins uses removable protein tags (e.g., His6 and GST) and various permutations of affinity, ion exchange, and size-exclusion chromatography. In addition, methodologies for generating 13C-,15Nisotopically labeled amino acid-specific (10) or segmentally labeled proteins are also available (11, 12). Similar strategies for segmental and nucleotide-specific labeling are particularly important for the analyses of RNAs by using NMR spectroscopy (13, 14), given the extremely poor dispersion of RNA chemical shifts. Selectively labeled RNAs are particularly useful in combination with NMR experiments that utilize filtering and editing schemes to identify intramolecular NOEs (between labeled and unlabeled nucleotides in a single-RNA sequence for unambiguous assignment of overlapped chemical shifts) and intermolecular NOEs (between RNAs or in RNA– protein complexes) (15). However, since there is no commercially available source for chemically synthesized 13C-,15N-isotopically labeled RNA, these studies rely on the ability to synthesize large quantities of RNA in vitro in the laboratory. The following sections outline start-to-finish detailed protocols for the synthesis and purification of milligram quantities of RNA using in vitro transcription by T7 RNA polymerase, suitable for use in NMR studies (Fig. 1a).

Fig. 1. RNA synthesis and purification. (a) An RNA synthesis and purification flowchart, including the general steps from initial template design to NMR data collection. (b) DNA template design for RNA transcription. The RNA product is presented on the annealed DNA template/T7 Top promoter duplex in lower case letters. The DNA sequences of the T7 RNA polymerase promoter sequence and DNA template for transcription are presented in upper case letters. The two nucleotides at the 5¢ end of the template, which should have 2¢-OMe substitutions, are indicated by asterisks. (c) A representative 20% (19:1) acrylamide:bisacrylamide gel showing the observed bands from a typical analytical transcription experiment compared to the DNA template.

12

NMR Studies of Protein–RNA Interactions

199

2. Materials 2.1. DNA Template Design for RNA Synthesis

1. Template DNA: Complementary DNA template for RNA transcription, 0.1–1.0 mM in water, store at −20°C, Fig. 1b (see Note 1). 2. T7 Top DNA: Purified T7 RNA polymerase promoter DNA (coding strand, 5¢-TAATACGACTCACTATA-3¢), 1 mM in water, store at −20°C.

2.2. RNA Synthesis

1. 50 mM Annealed DNA template: T7 Top DNA stock solution in water, store at −20°C. 2. 10× Transcription buffer: 400 mM Tris–HCl, pH 8.0, 10 mM spermidine, 0.1% (w/v) Triton X-100; store at room temperature. 3. 100 mM dithiothreitol (DTT): Store in 1–2-mL aliquots at −20°C. 4. 100 mM ATP, CTP, GTP, and UTP: Dissolve nucleotides in water and adjust to pH 8.0 using 1M NaOH; store in 1–2-mL aliquots at −20°C (see Note 2). 5. 1M MgCl2: Store at room temperature. 6. T7 RNA polymerase/ribonuclease (RNAse) inhibitor mixture: T7 RNA polymerase in solution, dialyzed into 30 mM HEPES (adjusted to pH 7.5 with 1M NaOH), 0.1M potassium glutamate, 0.25 mM EDTA, 0.05% Tween-20, 1 mM DTT, 200 mM NaCl. Following dialysis, RNAse inhibitor (see Note 3) and glycerol (50% (w/v) final concentration) are added. Store in 1-mL aliquots at −80°C. 7. Gel running buffer (1× TBE): 90 mM Tris base, 90 mM boric acid, and 2 mM EDTA (using 0.5M EDTA, pH 8.0, stock solution). Typically made as a 5× or 10× concentrated stock and diluted as needed; store at room temperature. 8. Denaturing acrylamide gel solution: 20% acrylamide/bisacrylamide (19:1), 1× TBE, and 7.8 M urea; store at 4°C (see Note 4). 9. 10% (w/v) Ammonium persulfate solution (APS): Store at 4°C. 10. Denaturing gel loading buffer: 80% formamide, 10 mM EDTA (using 0.5M EDTA, pH 8.0, stock solution), 0.025% (w/v) xylene cyanol, and 0.025% (w/v) bromophenol blue; store at room temperature. 11. Toluidine blue stain: 0.25% (w/v) in water; prepare as needed, and store at room temperature. 12. 100% Ethanol; store at −20°C. 13. 0.5M EDTA stock solution: 0.5M EDTA, adjust to pH 8.0 using solid NaOH pellets. 14. N,N,N´,N´-tetramethyl-ethane-1,2-diamine (TEMED).

200

C.A. Theimer et al.

2.3. RNA Purification

1. Denaturing gel loading buffer, denaturing acrylamide gel solution, APS, and gel running solutions are identical to those described in Subheading 2.2. 2. Polyester-backed, silica-based, fluorescent thin-layer chromatography (TLC) plates. 3. Low-salt buffer: 10 mM monosodium phosphate/disodium phosphate buffer, pH 7.6 (adjusted with 1M NaOH or 1M HCl if the pH is off by more than 0.2 pH units), 1 mM EDTA, and 200 mM KCl; store at room temperature (see Note 5). 4. High-salt buffer: 10 mM monosodium phosphate/disodium phosphate buffer, pH 7.6 (adjusted with 1M NaOH or 1M HCl if the pH is off by more than 0.2 pH units), 1 mM EDTA, and 1.5M KCl; store at room temperature (see Note 5). 5. HiTrap-Q anion exchange column (5 mL prepacked). 6. 20% (w/v) Ethanol. 7. Amicon-pressurized stirred cell and 1,000 and/or 3,000 molecular weight cutoff (MWCO) membranes. 8. NMR buffer for preliminary RNA experiments or for RNA– protein interaction studies (typically, 10–20 mM sodium phosphate buffer, pH 6.0–7.0, 0–200 mM KCl). 9. 70% Ethanol. 10. 100% Ethanol. 11. Electroelution chamber.

2.4. Preliminary RNA Analysis by NMR

1. 99.99% D2O: Store in sealed ampoules at room temperature. 2. 4M potassium chloride: Store at room temperature. 3. Acid and base solutions for pH optimization: Usually, 0.1 and 1M hydrochloric acid and 0.1 and 1M potassium (or sodium) hydroxide; store at room temperature. 4. Sodium azide. 5. Standard and Shigemi NMR tubes.

2.5. RNA–Protein Interactions by NMR

1. Polyethyleneimine (PEI): 5% (w/v) solution at pH 7.9 (MW 50–100 K). 2. Solid NaCl. 3. Amicon-pressurized stirred cell and 1,000 and/or 3,000 MWCO membranes. 4. NMR buffer similar to that used for RNA experiments (see Subheading 2.3). 5. Shigemi NMR tubes, D2O matched. 6. Solid ammonium sulfate.

12

NMR Studies of Protein–RNA Interactions

201

3. Methods 3.1. DNA Template Design for RNA Synthesis

Commonly, a partially double-stranded DNA template is used for in vitro transcription of large quantities of RNA, since only the promoter region of the template DNA must be double stranded for T7 RNA polymerase to bind and initiate transcription (Fig. 1b) (16). Standard PCR can be used to make the DNA template completely double stranded if it improves transcription efficiency and the overall yield for a particular template, although we do not often find this to be necessary. The biggest issue for synthesizing RNA by using in vitro transcription is RNA product heterogeneity due to in vitro transcription artifacts (Fig. 1c). It has been demonstrated that, in vitro, T7 RNA polymerase can produce off-target products as a result of 5¢-heterogeneity, 3¢-heterogeneity, and RNAtemplated RNA addition (16–19). T7 RNA polymerase only synthesizes RNA products that contain at least one 5¢-G nucleotide. Increased transcription efficiency is observed with two or three 5¢-G nucleotides, but multiple G nucleotides at the 5¢-end of the RNA has been shown to cause 5¢-heterogeneity (additional G nucleotides) in the RNA product (18). To strike a balance between transcription efficiency and artifact generation, we prefer, when possible, to start sequences with no more than two G nucleotides in a row. 3¢-heterogeneity appears to be a result of the runoff transcription mechanism and often results in the nontemplated addition of one or two nucleotides (N + 1 and, less often, N + 2 products) (16). This problem can be diminished or overcome completely by having the DNA template synthesized with 2¢-methoxyl groups on the two terminal 5¢ nucleotides (20, 21), and we strongly recommend that template DNAs be synthesized with this modification; Fig. 1b. Typically, we see RNA-templated RNA addition products that vary in size from ~10–20 nucleotides longer than the expected transcribed product. Since this phenomenon is sequence dependent, some sequences display no extraneous long products and some sequences make large amounts of long products (17, 19). A study has been performed on DNA sequence mutations that reduce or abolish this behavior (22); when sequence alteration is not feasible, there are transcription conditions that can help to reduce this problem (19). In addition, strategies to eliminate 5¢- and 3¢-heterogeneity in RNA samples using ribozyme-based cleavages are also performed, as described in detail elsewhere (13). Finally, mass spectrometry is an excellent technique to check that the final RNA product is the correct length and has the correct sequence composition. The DNA template is complementary to the sequence of the RNA and the top strand of the T7 promoter sequence (Fig. 1b). Both DNA sequences for RNA transcription can either be ordered

202

C.A. Theimer et al.

from an oligonucleotide synthesis company typically at a 250-nmol to 1-mmol scale, unpurified, or synthesized on site if you have access to a DNA synthesizer. The chemically synthesized DNA should be desalted by the company and arrive as a lyophilized pellet. 1. Dissolve the pellet in 100–300 mL of water. Depending on the quality of the DNA synthesis, it may be necessary to purify the DNA template using denaturing polyacrylamide gel electrophoresis (PAGE), although we do not always do so. Use a UV/ VIS spectrophotometer and the molar extinction coefficient of the specific DNA sequence to calculate the concentration of the DNA template (see Note 6) and adjust to ~0.2–1 mM. A dilution between 1:100 and 1:1,000 should be sufficient for spectrometry. 2. Prepare 1 mL of 50 mM annealed stock DNA for transcription, which is enough for 50 mL of transcription at a final template concentration of 1 mM. Pipette 50 mL of 1 mM T7 Top DNA (total 50 nmol) into a sterile labeled Eppendorf tube. Add 50 nmol of the template DNA for RNA transcription (based on the calculated concentration) and water to make a total volume of 1 mL. 3. Vortex the tube containing the DNAs and heat it at 95°C for 5 min. 4. Allow the annealed DNA template to cool to room temperature. Use the DNA immediately or store for later use at 4 or −20°C. 3.2. RNA Synthesis

The transcription reaction is essentially the same for making selectively or uniformly 13C-,15N-isotopically labeled RNA as it is for unlabeled RNA. All nucleotides are prepared and stored separately so that one or more unlabeled nucleotides can be replaced with 13 C-,15N-isotopically labeled nucleotides for making selectively labeled RNA samples or all four unlabeled nucleotides can be replaced with 13C-,15N-isotopically labeled nucleotides for uniform labeling. The primary differences are as follows: (1) since labeled nucleotides are prohibitively expensive, we typically run labeled transcriptions at a concentration of 2 mM for each nucleotide and (2) we often explore additional transcription optimization conditions (additional transcription components), such as adding inorganic pyrophosphatase (IPP) or polyethylene glycol (avg. MW 8,000), for improved transcription efficiency (23). Before performing large-scale transcription reactions (10– 50 mL), it is necessary to determine the optimum transcription conditions for every new DNA template (Fig. 1c). The conditions that need to be considered include NTP concentration (2–4 mM),

12

NMR Studies of Protein–RNA Interactions

203

magnesium chloride concentration (15–50 mM), annealed DNA template concentration (0.5–1 mM), and T7 RNA polymerase/ RNAse inhibitor mixture concentration. We typically use a concentration of 4 mM for each NTP when working with unlabeled nucleotides, and ~2 mL of T7 RNA polymerase/RNAse inhibitor mixture per 100 mL test reaction. Test reactions are performed in 1.5-mL Eppendorf tubes (see Note 7). 1. Thaw the T7 RNA polymerase/RNAse inhibitor mixture at −20°C, and then keep on ice; handle very gently (see Note 8). 2. Thaw all other frozen components on ice (except for the T7 RNA polymerase/RNAse inhibitor mixture), and vortex them briefly to ensure homogeneity (see Note 9). 3. For a single-DNA template, run between 8 and 16 different test reaction conditions, each reaction consisting of a total of 100 mL. Add 10 mL of 10× transcription buffer (1× final concentration) and 2.5 mL of 100 mM DTT (2.5 mM final concentration) to each of the labeled test reaction tubes. 4. Based on the exact conditions to be tested, add 1–2 mL of 50 mM annealed DNA template (0.5–1.0 mM final concentration), 2–4 mL of each 100 mM NTP stock (2.0–4.0 mM final concentration), and 1.5–5.0 mL of 1M MgCl2 (10–50 mM final concentration) to each tube. 5. Add water to bring the volume to 100 mL, leaving space for the T7 RNA polymerase/RNAse inhibitor mixture to be added later. Vortex the samples and centrifuge in a standard tabletop at 14,000 × g for 3 min at room temperature to collect the solution droplets. 6. Add 1–5 mL of T7 RNA polymerase/RNAse inhibitor mixture to the samples, pipetting the solution up and down gently to mix in the polymerase and to prevent denaturation of the protein. Always add the T7 RNA polymerase/RNAse inhibitor mixture last. 7. Incubate the samples at 37°C for 2–4 h. 8. When the test transcription reactions are complete, centrifuge in a standard tabletop at 14,000 × g for 3 minutes to pellet precipitated magnesium pyrophosphate, and transfer 10 mL of each reaction into clean labeled 1.5-mL Eppendorf tubes and add 10 mL of denaturing gel loading buffer to each tube. A standard sample is also made containing 5 mL of the annealed DNA template, 5 mL of water, and 10 mL of denaturing gel loading buffer. The samples are heated at ~95°C for 5 min immediately prior to loading on the denaturing gel. 9. These instructions assume the use of a vertical electrophoresis system (for example: ASU-250, C.B.S. Scientific) and 17-cm-length gels, although any vertical gel apparatus with a reasonable gel

204

C.A. Theimer et al.

size can be used. Rinse the glass plates and spacers with water, followed by ethanol, and wipe dry immediately prior to use. Although very small gels can be made using standard protein SDS-gel apparatus, we do not advise this as the resolution on such gels is generally too low for single-nucleotide resolution and thus not satisfactory for this purpose. 10. Prepare a 0.75-mm-thick 20% gel by mixing 50 mL of denaturing acrylamide gel solution with 500 mL of 10% APS. To this solution, mix in 50 mL of TEMED and immediately pour the gel. These gels are a single layer and the comb (10, 14, or 20 well) is inserted immediately after the entire gel is poured. The gel takes approximately 30 min to polymerize. The percentage of acrylamide:bisacrylamide solution used depends on the size of the RNA product that is run on the gel. We typically use 20% for RNA transcripts up to 40–50 nucleotides long, 15% for RNA transcripts between 45 and 70 nucleotides, and 10–12% for longer RNA transcripts. 11. When the gel is polymerized, remove the comb and carefully rinse the wells with water by using a 30-mL syringe equipped with a 22-gauge needle. Place the gel in the apparatus and fill the buffer chambers with gel running buffer. The gel should be pre-run at 150 V (or 15 W) for 15–20 min and the wells rinsed again with gel running buffer prior to loading the samples. 12. Once the samples are loaded, run the gel for 2–3 h at 150 V or until the bromophenol blue dye front (dark blue) is within ~3 cm of the bottom of the gel. 13. Stain the gel with toluidine blue stain on an orbital shaker for 15 min. Destain the gel with water (multiple exchanges) until you observe good contrast between the dark blue nucleic acid bands and the background of the gel. The optimal transcription conditions are chosen based on the test conditions that produce the highest intensity of the RNA product band. 14. Once the optimal solution conditions have been identified, the large-scale transcription reaction can be performed. Generally, we perform a 30 mL transcription reaction in a sterile, disposable, blue-capped 50-mL centrifuge tube. We find that this is a large enough transcription volume to generate a reasonable quantity of RNA for NMR samples (150–400 nmol of RNA) for most RNA sequences, although (rarely) some RNA sequences transcribe much better or much worse than this. The large-scale reaction is a direct scale-up of the small-scale reaction. The only differences are the volumes of reagents used and that it is usually not necessary to centrifuge the solution after vortex mixing. The T7 RNA polymerase/RNAse inhibitor mixture is still the last ingredient added before incubation and the tube is swirled gently after adding the polymerase.

12

NMR Studies of Protein–RNA Interactions

205

15. Incubate the reactions for 4–8 h at 37°C. The reactions can be incubated overnight, although this can be risky if there is any possibility of RNAse contamination or if the RNA sequence is prone to undesirable side reactions, like RNA-primed RNA addition by T7 RNA polymerase, as described above (see Note 10). 16. Centrifuge the reaction solution at 4,000 × g for 5 min at 4°C and decant the supernatant, which contains the RNA product, into a sterile 250-mL centrifuge bottle. There is a large pellet from precipitated inorganic pyrophosphate; this does not contain any RNA and can be discarded. 17. Add 1/10th the volume of 0.5M EDTA, pH 8.0 (1 mL of EDTA for every 10 mL of transcription), swirl to mix, and then add 2.5–3 times the volume of cold 100% ethanol (~75–90 mL of ethanol for a 30 mL transcription) to the transcription solution. 18. Place the centrifuge bottle containing the RNA product at −20°C overnight to precipitate (see Note 11). 3.3. RNA Purification

Denaturing PAGE is our primary means of purifying large quantities of RNA to single-nucleotide resolution for NMR. Although we have not experimented recently with currently available preparative-scale HPLC columns, in the past we found that HPLC purification would not purify RNAs larger than ~30 nucleotides to single-nucleotide resolution on a preparative scale. In addition, while we frequently purify small quantities of RNA using native PAGE, the native gels must be run in the cold room (4°C), typically take a long time to run, and rarely yield single-nucleotide resolution when the single-band products are checked on analytical denaturing gels for purity. 1. Centrifuge the bottle at 14,000 × g for 45 min at 4°C. 2. Very gently, decant the supernatant, as soon as the centrifuge stops running (see Note 12). There should be a visible white pellet on the wall of the bottle. The size of the pellet is not a direct reflection of product yield, since the majority of the size of the pellet is due to salt precipitation. 3. If the pellet is very large, put 20–50 mL of cold 70% (w/v) ethanol in the bottle very gently to wash away excess salt. Centrifuge with the pellet located against the outer wall at 14,000 × g for 30 min at 4°C, and again decant the supernatant immediately (see Note 13). 4. Dry the pellet by placing the capless centrifuge bottle underneath the hood, angled so that the residual ethanol is not lying directly on top of the pellet. Evaporating off all residual ethanol typically takes about 2 h if the supernatant was properly poured off without disturbing the pellet.

206

C.A. Theimer et al.

5. When the pellet has completely dried, put 2–3 mL of water in the bottle directly over the pellet and allow it to sit for 10 min. Pipette up and down gently to resuspend and transfer to a sterile, disposable 15-mL centrifuge tube. Rinse the bottle with an additional 1–2 mL of water to gather up any remaining sample and transfer to the centrifuge tube. 6. Add denaturing gel loading buffer to the sample. Use 1:1 RNA solution:gel loading buffer; although if the RNA solution volume is too large, you can use 2:1 RNA solution:gel loading buffer and it loads and runs fine on the purification gels. 7. These instructions assume the use of gel electrophoresis equipment for large, sequencing gels, 20 cm (w) × 42 cm (l) (for example: DDH-400-20, C.B.S. Scientific), spacers, and 3-well preparative scale (3-mm thick) combs. Rinse the glass plates, combs, and spacers with water, followed by ethanol, and wipe dry immediately prior to use. 8. Pour the number of gels needed to obtain the appropriate purification level (see Note 14). Preparative-scale gels typically take ~400 mL of acrylamide solution to fill each gel. Place 400 mL of denaturing acrylamide gel solution into a beaker, add 4 mL of 10% APS solution, and stir briefly to mix the APS in evenly. Add 400 mL of TEMED and stir the solution again briefly. In general, a 1/100 ratio of APS and 1/1,000 ratio of TEMED to gel solution are used for polymerization. The gel must be poured immediately at this ratio of APS and TEMED. The gels take 2 h to completely polymerize and cool, although they can sit for longer or even overnight. 9. Remove the combs and clean the wells by using a wash bottle of water. It is important to also carefully clean the back plate of the gel, which is in contact with the ceramic heat exchange plate, to prevent poor contact and smiling of the gel. Set up the gels in the sequencing apparatus and fill the upper and lower buffer chambers with gel running buffer. The gels should be pre-electrophoresed for 20–30 min at 20 W per gel, and the wells thoroughly rinsed with 1× TBE prior to loading the RNA samples. 10. Heat the RNA samples at ~95°C for 5 min immediately prior to loading on the purification gels. Load the gels with a standard 1,000 mL pipettor, since the tip easily fits down into the 3-mm wells. 11. The size of the RNA dictates how long the gels should be run and the wattage. Typically, we run these gels overnight (~18 h) at 20 W per gel for smaller RNAs (15–25 nucleotides), 25 W per gel for medium-sized RNAs (25–40 nucleotides), and 30 W per gel for larger RNAs (over 40 nucleotides). The gels are monitored based on the location of the dyes and the previous observation from test transcription gels, of where the RNA runs compared to the xylene cyanol and bromophenol blue dyes (see Note 15).

12

NMR Studies of Protein–RNA Interactions

207

12. Remove the gel from the apparatus, carefully remove one glass plate, and cover the gel with plastic wrap. Flip the gel over (plastic wrap on the bottom) onto a fluorescent TLC plate and gently remove the top glass plate. When a handheld shortwavelength UV light (254 nm) is shone onto the gel, the plate glows green and the RNA band is visible as a grey to black shadow on the fluorescent green background (UV shadowing). Carefully cut the product bands out of the gel with a clean razor blade and transfer them to a sterile, disposable 50-mL centrifuge tube and discard the rest of the gel. Repeat for all of the gels (exposure to UV light should be minimized to avoid UV damage of RNA). 13. Elute the RNA product by using a 4-trap Elutrap electroelution chamber (Whatman), with 1× TBE as the running buffer. Gel slices should be cut into small pieces but not crushed, and placed into the gel holding chamber between the BT1 and BT2 membranes (see Note 16). 14. Run the elution at 150 V and collect the RNA sample from the trap at 2–3-h intervals, typically for 9 h. For longer RNAs, we frequently reduce the voltage to 50 V and run overnight to ensure complete elution of the product from the gel slices. The progress of the elution can be tracked by calculating the amount of RNA in the eluant at each time point, based on UV absorbance at 260 nm. Time points should be stored at −20°C until all of the RNA has been collected, and the elution is finished. 15. Thaw and pool all of the eluted RNA fractions. Load this material onto an HiTrap-Q anion-exchange column hooked up to a peristaltic pump (flow rate 3–5 mL/min) and pre-equilibrated with low-salt buffer. Wash the column with ten-column volumes (50 mL) or more of low-salt buffer. Elute the RNA with high-salt buffer and collect 4-mL fractions. The purified RNA is typically found entirely in the second and third fractions. The column must always be thoroughly rinsed with low- and high-salt buffers between samples and immediately stored in 20% ethanol when not in use to prevent contamination. 16. Pool the two RNA containing fractions (8 mL) in a 50-mL centrifuge tube, add 24 mL of cold 100% ethanol, and store the RNA overnight at −20°C to precipitate. 17. Remove the centrifuge tube directly from the freezer, make a weight-matched balance tube, and centrifuge at 15,000 × g for 45 min at 4°C. Very gently, decant the supernatant as soon as the centrifuge stops running. At this point, it is possible to again perform a 70% ethanol wash as described above (step 3). 18. Dry the pellet by placing the capless centrifuge tube underneath the hood, angled so that the residual ethanol is not lying directly on top of the pellet. Evaporating off all residual ethanol typically takes about 2 h.

208

C.A. Theimer et al.

19. Dissolve the RNA sample in 50 mL of water or NMR buffer, if you already know what the required buffer should be, based on the protein to be investigated. For new RNAs under investigation, desalt the RNA (step 21) into water for pH and salt titrations in the NMR to determine optimal conditions for obtaining a single-RNA conformation in solution. 20. Anneal the RNA by placing the 50-mL tube at ~95°C for 5 min and then slow cool to room temperature on the benchtop (see Note 17). 21. When cool, load the solution into an Amicon-stirred cell (MWCO 3,000 or 1,000 membrane, depending on the size of your RNA), pressurize the chamber with 55 psi of nitrogen or argon, and stir. Concentrate the RNA down to ~1–2 mL and then add fresh buffer or water up to the 50-mL line. Perform at least three washes to remove excess salt. Concentrate the RNA down to ~0.5–1 mL (~0.2–2 mM) and transfer to a sterile 1.5-mL Eppendorf tube. The RNA is ready for NMR studies and should be stored at −20°C when not in use. 3.4. Preliminary RNA Analysis by NMR

Once an initial unlabeled RNA sample is made, a basic set of NMR experiments is run, including 1D 1H NMR experiments collected in 95% H2O/5% D2O (5–10°C) for pH and salt titrations, NOESYs collected in 95% H2O/5% D2O (5–10°C) and D2O (20–30°C), and a TOCSY collected in D2O (20–30°C), to assess the properties of the RNA in solution at NMR concentrations. Optimal salt and pH conditions are assessed by the number, intensity, and line widths of the detectable imino proton resonances compared to the number of imino protons which are expected to be protected from rapid exchange due to hydrogen bonding in Watson-Crick and noncanonical base pairs (Fig. 2). Generally, we find that the optimal pH for RNA samples falls between 6.0 and 7.0 and the optimal salt conditions vary from 0 to 200 mM monovalent salt, depending on the ability of the sequence to form alternative conformations (see Note 18). The optimal solution conditions are typically dictated by the protein of interest, but it is important to be aware of the expected behavior of the RNA under the appropriate solution conditions. Degradation products and alternative conformations (including unwanted dimerization) can be identified both from additional imino proton resonances in the 1D and 2D spectra collected in 95% H2O/5% D2O (Fig. 2). In addition, the 1D 1H spectrum helps identify any small-molecule contaminants in the RNA sample. The most common contaminants in RNA samples (and their causes) are as follows: acrylamide and/or urea (insufficient volume of low-salt wash of the anion-exchange column before eluting the RNA product), EDTA and/or ethanol (too few buffer exchanges during desalting and concentration), and ethanol and/or glycerol (insufficient soaking and rinsing of the Amicon membrane prior to installing in the

12

NMR Studies of Protein–RNA Interactions

209

Fig. 2. 1D and 2D imino proton NMR spectra of RNA. The secondary structure of a hairpin RNA and the imino proton region of 1D and 2D NOESY 1H spectra (500 MHz) of the RNA in 95% H2O/5% D2O at 10°C demonstrate the presence of sharp imino proton resonances. A clear sequential walk through the stem base pairs is indicated for the expected hairpin stem. The smaller imino proton resonances and NOE peaks are due to an alternative conformation in solution.

stirred cell). EDTA can be eliminated from the high- and low-salt anion-exchange buffers if it is a recurring contaminant in the final samples. Ethanol contamination can easily be removed by freezing the sample in liquid nitrogen, lyophilizing the sample to dryness, and resuspending in 95% H2O/5% D2O (repeated as necessary). Glycerol typically requires exhaustive dialysis or buffer exchanges in the Amicon-stirred cell to be completely removed. 1. For titrations, the RNA sample was previously concentrated into water during the last steps of RNA purification (Subheading 3.3, steps 19–21). Add 5% (w/v) 99.99% D2O to the RNA sample (usually, ~0.1–0.3 mM RNA for solution condition optimization) and adjust to the starting pH (~pH 5.5) using acid and base solutions (see Note 19). 2. Transfer the sample to a standard NMR tube, insert the NMR tube into the spectrophotometer, and allow the sample to equilibrate to 10°C for 5 min before the instrument is set up for data collection. 3. Collect a baseline 1D imino proton spectrum (using 1,1 echo water suppression) for a low-salt, low-pH (~5.5–6.0), RNA sample at 10°C; usually, 128 scans are sufficient (see Note 20).

210

C.A. Theimer et al.

4. After each spectrum is collected, transfer the RNA sample into a 1.5-mL Eppendorf tube, adjust the pH by 0.5 pH units using the appropriate acid and base solutions, return the sample to the NMR tube, and allow it to equilibrate to 10°C for 5 min in the spectrophotometer before the setup for data collection. Due to increased RNA degradation under basic conditions, we typically do not take RNA samples above pH 8.0 unless absolutely necessary. 5. After the NMR spectra are collected and compared, identify an optimal pH based on the number of expected and observed peaks, intensities of the peaks, line widths, and pH at which there is the least peak overlap in terms of chemical shift. The optimal pH for most RNA samples typically lies between pH 6.0 and 7.0. 6. For salt titrations, use a new RNA sample that is already adjusted to the correct pH. The same procedure is followed for salt titrations. Typically, use either 25- or 50-mM increments for the monovalent salt concentration, aliquoted from a 4 M KCl stock, and choose the best conditions again based on the number of expected and observed peaks, intensities of the peaks, line widths, and monovalent salt concentration at which there is the least peak overlap in terms of chemical shift. If the optimal conditions for the RNA are consistent with the conditions needed for protein stability and protein–RNA interactions, all of the RNA from the NMR samples can then be pooled with the stock RNA and exchanged into the appropriate NMR buffer at this point as described in RNA purification (steps 19–21). 3.5. RNA–Protein Interactions by NMR

This section focuses on preparing the protein for investigating RNA– protein interactions. Protein purifications are well-established and isotopic labeling of proteins has been reviewed extensively (12, 24, 25). Other than the standard issues and concerns for preparing NMR quality protein samples, there are two issues that also need to be considered. First, RNAses can bind to or copurify with the protein of interest and may potentially degrade the RNA during complex formation. Second, high-affinity cellular RNA can copurify with the protein. Usually, several different purification steps, such as affinity tag purification, followed by ion-exchange and size-exclusion chromatography, are sufficient to ensure complete removal of RNAses (26). If it is also necessary to remove RNA bound to protein, this can be done through treatment with 0.1–0.5% PEI (27) to ensure removal of all RNA bound to the protein of interest. Uniformly 15N-labeled protein is typically obtained by overexpression from E. coli in M9 minimal medium and purified using affinity columns for either His6- or GST-tagged proteins. This step is often followed by size-exclusion chromatography. Depending on the purity of the protein, as judged by Coomassie staining of SDS

12

NMR Studies of Protein–RNA Interactions

211

polyacrylamide gels, it may be necessary to include a third-column purification step, such as anion-exchange chromatography. Following these purification steps, there should be no RNAse contamination left in the purified protein. However, there may still be nucleic acid contamination, which can be detected by running the purified protein on an agarose or acrylamide gel and staining with ethidium bromide. If nucleic acid contamination is detected, a PEI precipitation following the first affinity column can ensure complete removal. 1. Pool the appropriate fractions that contain the protein of interest from the affinity column and add solid NaCl to a final concentration of 1M. 2. Place all samples at 4°C, add PEI to a final concentration of 0.5% (w/v), and incubate the solution for 1 h. 3. Centrifuge the solution at 15,000 × g for 35 min at 4°C to remove the precipitated PEI–nucleic acid complexes. 4. Recover the protein from the supernatant through precipitation by adding ammonium sulfate to a final concentration of 75% (see Note 21). 5. Centrifuge the protein solution at 10,000 × g for 20 min at 4°C, decant the supernatant, and recover the pelleted protein by dissolving in low-salt buffer (Subheading 2.3), if the next column is an anion-exchange column, or in the final NMR buffer (Subheading 2.3), if the next column is a size-exclusion column. The final buffer should be identical to the buffer chosen for optimal RNA stability (see Subheading 3.4). 6. A final column purification step, such as an ion-exchange column, can be inserted at this step, if the previous purification steps do not yield protein of sufficient purity. 7. Pool all the fractions that contain the protein of interest and concentrate the protein to 0.2–1 mM concentration using either Amicon-pressurized stir cells or disposable Amicon ultra centrifugal filters (Ultracel). 8. Monitor complex formation between the RNA and protein by acquiring a 1H–15N heteronuclear single-quantum correlation (15N-HSQC) spectrum of the 15N-labeled protein (isotopic labeling of proteins has been extensively reviewed elsewhere (11)). 9. First, collect the 15N-HSQC spectrum of the free protein. The 2D 15N-HSQC should be acquired from 6 to 13 ppm in the hydrogen dimension and 95–140 ppm in the nitrogen dimension, which is typically referred to as the fingerprint region of proteins (see Note 22). 10. Prepare the protein sample in a Shigemi NMR tube using 300 mL of a 0.2 mM protein sample in 5% D2O to obtain a good quality spectrum and define the effect of buffers on the overall fold of the protein.

212

C.A. Theimer et al.

11. Following the initial test of protein folding, measure the chemical shift perturbation in the 15N-HSQC of the amide protons from the free protein upon titration with unlabeled RNA. The protein should be between 0.1 and 0.5 mM and the unlabeled RNA has to be concentrated to obtain different ratios for the titration ranging from 0.25 to 2:1 of RNA:protein without requiring the need for large volumes of RNA solution to be added to the NMR sample (see Note 23). 12. Since each 15N peak on the protein represents a distinct amide peak of the protein, this titration can yield information on the position of the RNA binding site at the protein interface as well as the binding affinity, based on the NMR timescale. 13. The reverse titration (protein into the RNA sample) can also be performed. First, using the same purification protocol, obtain a concentrated, unlabeled protein (1–2 mM) that can be titrated into a 13C-,15N-labeled RNA sample, which is purified as described above for unlabeled RNA (see Subheading 3.3). 14. Obtain a 13C-HSQC of RNA, which typically exhibits distinct peaks for the carbons attached to the aromatic and H1¢ protons. Binding to the protein should cause perturbations of these peaks, and thus yields information of the binding site on the RNA (28) (see Note 24). Once a few sites on the protein or RNA have been identified using the chemical shift perturbation method, other more targeted experiments narrow down the binding region (for a review of different experiments and how they can be applied to solve the RNA–protein complex, see ref. 26). Ultimately, measuring NOEs between the RNA and the protein provides the most important details for the structural calculations of the RNA– protein complex. Filtered and edited NOESY experiments (outlined in the introduction based on ref. 15) have been instrumental in identifying NOEs for RNA structure determination and greatly simplify the identification of NOEs between the RNA and protein components of the RNA–protein complex.

4. Notes 1. All water for RNA studies must be distilled deionized 18.0 MW-cm quality from a purification system that is properly maintained (cartridges replaced as suggested by manufacturer). Large bottles of autoclaved water are kept on hand in the laboratory for solution making, and all solutions that come into contact with RNA are sterile filtered using Nalgene sterile filter units with a 0.2-mm pore size prior to storage. Unless otherwise noted, all solutions should be handled in this manner and “water” in the

12

NMR Studies of Protein–RNA Interactions

213

context of this protocol indicates autoclaved water of this quality which has been sterile filtered prior to use. As long as the water quality is maintained, we have not found it necessary to treat water with diethylpyrocarbonate (DEPC) in order to prevent RNAse contamination and degradation of RNA samples. 2. It is important that the pH of the nucleotides is adjusted using sodium hydroxide and not Tris base. High Tris base concentrations appear to inhibit transcription by T7 RNA polymerase. 3. For the final solution of T7 RNA polymerase/RNAse inhibitor mixture, a tube (10,000 U) of SUPERase·In RNase Inhibitor (Ambion/Applied Biosystems) is mixed with the enzyme prior to the addition of glycerol to a final concentration of 50% (w/v), prior to making aliquots and storing at −80°C. RNAse inhibitor is not essential, but is added as a precaution against any residual RNAse from the T7 RNA polymerase enzyme purification. The final concentration of T7 RNA polymerase is rarely measured; we use test transcriptions to identify the amount of each preparation to use for optimal RNA transcription efficiency. 4. It is important when making this solution that it is not diluted to volume until all of the urea has dissolved (very-high-quality ultrapure urea should be used) and that it is not heated. When using stored acrylamide–urea solutions, the solution must be examined to ensure that urea has not precipitated out of solution. It is for this reason that we store these solutions in transparent brown glass bottles. In addition, since acrylamide is a cumulative neurotoxin, it is important that gloves are worn at all times even when it is polymerized. We prefer to buy premixed acrylamide to reduce handling. 5. The low- and high-salt buffers for anion exchange can use different buffer salts and sodium chloride, rather than potassium chloride. We use phosphate buffer and potassium chloride for this column because this is basically the last step of RNA purification and they are the usual buffer and salt compounds that we use for RNA NMR studies. 6. If the extinction coefficient for a DNA is not provided in the documentation from the company, it can be calculated (for DNA or RNA) using the extinction coefficients for individual NMPs (A = 15.4 M-1-cm-1, C = 7.4 M-1-cm-1, G = 11.8 M-1-cm-1, T = 9.6 M-1-cm-1, U = 9.9 M-1-cm-1) or a good approximation is to set the extinction coefficient at 10 M-1-cm-1 multiplied by the number of nucleotides. It is not advisable to use the general assumption that an A260 of 1 corresponds to 50 mg/mL as this is not accurate for short DNAs or short RNAs. 7. All disposable plasticware (pipette tips and Eppendorf tubes) that do not come presterilized and certified RNAse and DNAse free should be autoclaved prior to use and kept in sealed

214

C.A. Theimer et al.

containers to prevent contamination. We also cover the openings of glassware with aluminum foil and autoclave all glassware in the laboratory routinely after washing and rinsing with distilled deionized water. 8. Although it is possible to use purchased T7 RNA polymerase for in vitro transcription, the amounts necessary for large-scale synthesis of RNA for NMR tend to be prohibitively expensive. Therefore, most laboratories, ours included, purify their own supply of T7 RNA polymerase for transcription. The plasmid for His6-tagged T7 RNA polymerase is readily available from any number of academic sources. In general, we do not quantitate the T7 RNA polymerase; we simply use analytical transcription reactions to determine the appropriate amount of T7 RNA polymerase to be used for optimal transcription efficiency. 9. These solutions do not freeze in a homogeneous fashion, nor do they thaw that way. Before pipetting out of a thawed tube of solution, you must make sure that the solution is completely thawed, then vortex mix the tube briefly to make sure that it is well-mixed, and centrifuge the solution briefly to push all the droplets to the bottom of the tube. 10. If the reaction is working, it will start to get cloudy 1–2 h after you put the reaction into the water bath. The reaction gets cloudy because, as nucleotide triphosphates are linked together to form RNA (which has only one phosphate between each nucleotide), the other two phosphate groups are released as pyrophosphate (PPi). The pyrophosphate interacts with the magnesium in solution to form an insoluble magnesium pyrophosphate complex. The more precipitation you observe, the more nucleotide triphosphates have been converted into RNA. 11. If you use 2.5–3× the volume of 100% ethanol, your final solution will be 70–75% ethanol. It is not recommended to leave a sample in ethanol for more than a few days. The longer the sample sits, the more salt precipitates out of solution. The greater the amount of salt in the precipitate, the bigger effect it has on the separation efficiency and running of the purification gels. Ideal timing for RNA ethanol precipitation is either overnight at −20°C or for a few hours at −80°C. 12. It is important to pour off the supernatant as soon as the rotor stops spinning, before the pellet detaches from the wall of the bottle. 13. This step removes excess salt and helps the purification gels to run with better resolution. It is critical that you do not resuspend the pellet or rinse it off the wall with rough handling or it is very difficult to recover without significant loss of sample. 14. The number of gels needed to purify your RNA needs to be determined before you run your entire RNA sample on the gels. Each purification gel has three wells that can be loaded with

12

NMR Studies of Protein–RNA Interactions

215

different amounts of your dissolved RNA: loading buffer mixture. Load three different amounts of RNA in the three wells and run one gel to see which amount gives you a good-sized RNA band. Generally, loading amounts of the dissolved RNA solution that correspond to ~2, 4, and 6 mL of the original transcription give you a good idea what the proper amount for optimal purification might be. The RNA band should be somewhere between the width of a pencil and the width of your finger. Much less than this is hard to find on the gels and any more than that eliminates single-nucleotide resolution. The optimal loading amount is used for subsequent gels for this specific RNA. 15. To obtain the full separation capacity of the 42-cm-long gel, it is important to run the RNA to the bottom 1/3 of the gel or as close to the bottom as possible. The location of the RNA of interest on the gel with respect to the dyes should be noted at the test-preparative-gel stage. When purifying RNAs that run very close to the dye front (an ~27-nucleotide RNA runs on top of the xylene cyanol dye on a 20% 19:1 denaturing acrylamide gel), it is necessary to omit the overlapping dye from the samples completely and run a lane on one of the gels that contains only dye as a marker. 16. The MWCO for the BT1 membrane is ~5,000. To prevent loss of smaller product RNAs (~30 nucleotides or less), the BT1 membrane should be replaced with thoroughly rinsed 1,000 MWCO dialysis tubing for these samples. 18-mm wide dialysis tubing fits exactly in the trap chambers and only needs to be cut at the open ends to fit the length of the BT1 membrane. It is also possible to make a second trap by placing an additional BT1 membrane after the next insert in the trap. It is absolutely critical that elution traps are not overloaded with gel slices, which significantly reduces elution efficiency. 17. The conditions for annealing should be based on the behavior of the RNA in question. RNAs that are intended to form unimolecular structures should be annealed under dilute conditions (50 mL) while RNAs that are intended to form duplexes (dimers) should be annealed in more concentrated solutions (2–5 mL). In addition, snap cooling on ice can be used to affect the conformational exchange in solution. The timing of adding monovalent salts (before annealing or immediately after snap cooling) can be used to bias the sample toward forming a single conformation as needed. The appropriate annealing conditions require trial and error for each RNA sample. 18. We typically include 0.2% (w/v) sodium azide in RNA samples to prevent bacterial contamination. Usually, the sodium azide is included in the NMR buffer solution used for the final stages of RNA concentration and desalting (RNA purification: steps 19–21), after the optimal NMR solution conditions have been identified. While divalent salts, such as MgCl2, do stabilize

216

C.A. Theimer et al.

RNA structures and are often needed for larger RNA structures to form, completely small RNAs typically fold without magnesium, and the NMR samples last longer (degrade less quickly) if magnesium is excluded from the NMR buffer, whenever possible. 19. It is very important that, when the pH of RNA samples in water is being adjusted, you are careful not to overshoot the desired pH. Alternately adding acid and base to reach the appropriate pH results unnecessarily increases the salt concentration, which also affects the intensity of the imino proton resonance peaks. 20. It is important to ensure that the sweep width is wide enough not to miss any unusual imino proton chemical shifts (protonated cytosine residues, for example). Therefore, we routinely collect out to 18 ppm initially, although almost all imino proton peaks are found between ~9 and 15 ppm. 21. PEI can be very difficult to remove from a protein preparation, as it cannot be dialyzed out effectively. Ammonium sulfate precipitation is the standard method of separating the protein of interest from the PEI, with the protein precipitating and the PEI remaining in the supernatant solution. The precise percentage of ammonium sulfate necessary is protein dependent and the optimal percentage to be used should be tested on small samples of the protein to ensure maximal protein recovery. 22. Examining the fingerprint region of the free protein spectrum: If the peaks are well-dispersed with minimal overlap in the center region of the spectrum, this is an indication that the protein is well-folded. Each peak in this spectrum represents a unique amide peak from the protein. If the protein is not well-folded, it is still worth testing the titration with RNA, as it is possible that the RNA may induce conformational change of the protein upon binding. 23. It is best to start with a lower end of the range of protein concentrations since formation of RNA:RNA multimers can be an issue at high concentrations. This titration also gives some idea of the stoichiometry of RNA:protein complex. If a large amount of RNA (high RNA:protein ratio) must be added to observe chemical shift perturbations, this may be an indication that measuring NOEs between the RNA and protein is difficult. 24. One must keep in mind that it is possible to get shifted resonances (on the protein or the RNA) not only in the binding site, but also in other regions due to allosteric modulation or, more likely, changes in the pH or monovalent salt concentration of the solution. This is why it is particularly important to try to have both the RNA and protein prepared such that their final solution conditions are identical.

12

NMR Studies of Protein–RNA Interactions

217

References 1. Autexier, C., and Triki, I. (1999) Tetrahymena telomerase ribonucleoprotein RNA–protein interactions. Nucl. Acids Res. 27, 2227–2234. 2. Bachand, F., Triki, I., and Autexier, C. (2001) Human telomerase RNA–protein interactions. Nucl. Acids Res. 29, 3385–3393. 3. Greider, C. W., and Blackburn, E. H. (1987) The telomere terminal transferase of Tetrahymena is a ribonucleoprotein enzyme with two kinds of primer specificity. Cell 51, 887–898. 4. Staley, J. P., and Woolford, J. L., Jr. (2009) Assembly of ribosomes and spliceosomes: complex ribonucleoprotein machines. Curr. Opin. Cell Biol. 21, 109–118. 5. Ban, N., Nissen, P., Hansen, J., Moore, P. B., and Steitz, T. A. (2000) The complete atomic structure of the large ribosomal subunit at 2.4 A resolution. Science 289, 905–920. 6. Cech, T. R. (2000) Structural biology. The ribosome is a ribozyme, Science 289, 878–879. 7. Schluenzen, F., Tocilj, A., Zarivach, R., Harms, J., Gluehmann, M., Janell, D., Bashan, A., Bartels, H., Agmon, I., Franceschi, F., and Yonath, A. (2000) Structure of functionally activated small ribosomal subunit at 3.3 angstroms resolution. Cell 102, 615–623. 8. Wimberly, B. T., Brodersen, D. E., Clemons, W. M., Jr., Morgan-Warren, R. J., Carter, A. P., Vonrhein, C., Hartsch, T., and Ramakrishnan, V. (2000) Structure of the 30S ribosomal subunit. Nature 407, 327–339. 9. Ule, J. (2008) Ribonucleoprotein complexes in neurologic diseases. Current Opinion in Neurobiology 18, 516–523. 10. Whittaker, J. W. (2007) Selective isotopic labeling of recombinant proteins using amino acid auxotroph strains. Methods. Mol. Biol. 389, 175–188. 11. Cowburn, D., Shekhtman, A., Xu, R., Ottesen, J. J., and Muir, T. W. (2004) Segmental isotopic labeling for structural biological applications of NMR. Methods Mol. Biol. 278, 47–56. 12. Liu, D., Xu, R., and Cowburn, D. (2009) Segmental isotopic labeling of proteins for nuclear magnetic resonance. Methods Enzymol. 462, 151–175. 13. Lu, K., Miyazaki, Y., and Summers, M. F. (2010) Isotope labeling strategies for NMR studies of RNA. J. Biomol. NMR 46, 113–125. 14. Nelissen, F. H., van Gammeren, A. J., Tessari, M., Girard, F. C., Heus, H. A., and Wijmenga, S. S.

(2008) Multiple segmental and selective isotope labeling of large RNA for NMR structural studies. Nucl. Acids Res. 36, e89. 15. Peterson, R. D., Theimer, C. A., Wu, H., and Feigon, J. (2004) New applications of 2D filtered/edited NOESY for assignment and structure elucidation of RNA and RNA–protein complexes. J. Biomol. NMR 28, 59–67. 16. Milligan, J. F., Groebe, D. R., Witherell, G. W., and Uhlenbeck, O. C. (1987) Oligoribonucleotide synthesis using T7 RNA polymerase and synthetic DNA templates. Nucl. Acids Res. 15, 8783–8798. 17. Cazenave, C., and Uhlenbeck, O. C. (1994) RNA template-directed RNA synthesis by T7 RNA polymerase. Proc. Natl. Acad. Sci. U.S.A. 91, 6972–6976. 18. Pleiss, J. A., Derrick, M. L., and Uhlenbeck, O. C. (1998) T7 RNA polymerase produces 5¢ end heterogeneity during in vitro transcription from certain templates. RNA 4, 1313–1317. 19. Triana-Alonso, F. J., Dabrowski, M., Wadzack, J., and Nierhaus, K. H. (1995) Self-coded 3¢-extension of run-off transcripts produces aberrant products during in vitro transcription with T7 RNA polymerase. J. Biol. Chem. 270, 6298–6307. 20. Kao, C., Rudisser, S., and Zheng, M. (2001) A simple and efficient method to transcribe RNAs with reduced 3¢ heterogeneity. Methods 23, 201–205. 21. Kao, C., Zheng, M., and Rudisser, S. (1999) A simple and efficient method to reduce nontemplated nucleotide addition at the 3 terminus of RNAs transcribed by T7 RNA polymerase. RNA 5, 1268–1272. 22. Nacheva, G. A., and Berzal-Herranz, A. (2003) Preventing nondesired RNA-primed RNA extension catalyzed by T7 RNA polymerase. Eur. J. Biochem./FEBS 270, 1458–1465. 23. Cunningham, P. R., and Ofengand, J. (1990) Use of inorganic pyrophosphatase to improve the yield of in vitro transcription reactions catalyzed by T7 RNA polymerase. BioTechniques 9, 713–714. 24. Gardner, K. H., and Kay, L. E. (1998) The use of 2H, 13C, 15N multidimensional NMR to study the structure and dynamics of proteins. Ann. Rev. Biophys. Biomol. Struct. 27, 357–406. 25. Marley, J., Lu, M., and Bracken, C. (2001) A method for efficient isotopic labeling of recombinant proteins. J. Biomol. NMR 20, 71–75.

218

C.A. Theimer et al.

26. Wu, H., Finger, L. D., and Feigon, J. (2005) Structure determination of protein/RNA complexes by NMR. Methods Enzymol. 394, 525–545. 27. Marenchino, M., Armbruster, D. W., and Hennig, M. (2009) Rapid and efficient purification of RNA-binding proteins: application to

HIV-1 Rev. Protein Expression and Purification 63, 112–119. 28. Khanna, M., Wu, H., Johansson, C., CaizerguesFerrer, M., and Feigon, J. (2006) Structural study of the H/ACA snoRNP components Nop10p and the 3¢ hairpin of U65 snoRNA. RNA 12, 40–52.

Chapter 13 Preparation and Optimization of Protein–DNA Complexes Suitable for Detailed NMR Studies My D. Sam and Robert T. Clubb Abstract This chapter describes the methods to form and optimize samples of protein–DNA complexes that are suitable for detailed structure and dynamics studies by NMR spectroscopy. Key words: Protein–DNA complex, NMR, Structure, Intermolecular NOEs

1. Introduction Interactions between proteins and DNA molecules play an essential role in a wide range of important biological processes, including gene expression and genomic replication, recombination, and repair. Of particular interest are site-specific DNA-binding transcription factors, which regulate gene expression by recognizing specific nucleotide sequences. These proteins locate, and bind to, the correct site within the genome despite the presence of a vast number of competitor sites with similar geometries and electrostatic surfaces. Over the past several decades, X-ray crystallography has been used extensively to determine a large number of high-resolution structures of protein–DNA complexes (1, 2). This work has provided a wealth of detailed stereochemical information about binding site recognition, which typically is achieved through complementary hydrogen-bonding and van der Waals interactions that can be maximized by protein folding and/or DNA distortions (3, 4). However, crystallography provides only a static view of a protein– DNA complex, and thus little insight into the conformational dynamics that underpin macromolecular recognition (5).

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_13, © Springer Science+Business Media, LLC 2012

219

220

M.D. Sam and R.T. Clubb

NMR spectroscopy is a powerful tool that can be used to investigate protein–DNA recognition in the solution state. When the spectra obtained are of good quality, NMR can be used to elucidate high-resolution structures and atomic-level conformational dynamics (6, 7). NMR can also be applied to investigate other key aspects of recognition, including the basis of nonspecific binding (8), hydration lifetimes (9), on/off rates of binding, and the process by which a protein locates its binding site (10). Even when the quality of the NMR spectra are poor due to resonance line broadening, structural models of a protein–DNA complex can be generated using chemical shift mapping techniques. In recent years, the size and complexity of protein–DNA complexes amenable for NMR studies have increased due to several methodological advances, such as selective isotopic labeling, residual dipolar coupling measurements (11), paramagnetic relaxation enhancement methods (12), and transverse relaxation-optimized experiments, that exploit ultrahigh magnetic field strengths (13, 14). However, one of the largest obstacles to successfully studying a protein–DNA complex by NMR is the preparation of sufficiently stable, concentrated, and homogeneous samples of the complex. For a protein–DNA complex to be suitable for detailed NMR studies, it typically must satisfy several criteria. The components and nature of the interaction should be well-defined. In particular, biochemical experiments should have been performed to clearly delineate the specific nucleotide sequence recognized by the protein, as well as the stoichiometry and affinity of the resulting complex. The same molecular weight limitations that hinder NMR studies of other macromolecules apply, so the final size of the complex is also an important consideration. Ideally, the complex should have a dissociation constant (Kd) in the submicromolar range, thus making it more likely that it will be in the slow-exchange regime on the chemical shift timescale. However, weaker affinity complexes that are in fast exchange have also been successfully studied. To optimize the production and spectral qualities of a complex, large quantities of DNA and isotopically labeled protein are also needed to enable different preparative procedures and conditions to be tested. Pilot studies typically make use of purified 15N-enriched protein and a range of DNA species that differ in their length and sequence. In our experience, the greatest chance of success occurs when the protein is soluble in its DNA-free state and its 1H–15N HSQC spectrum is well-resolved. However, DNA is highly soluble and as a result the aggregation behavior of the protein may be greatly reduced upon complex formation. To create a stable complex suitable for NMR studies, the DNA molecule must have the appropriate nucleotide sequence and length to form productive contacts with the protein. For sequencespecific DNA binding proteins, this information can be obtained from previously reported biochemical studies, which should define

13

Preparation and Optimization of Protein–DNA…

221

the specificity, affinity, and stoichiometry of the complex. If the protein is known to bind to several DNA sites, then an alignment of their nucleotide sequences may reveal conserved positions essential for binding. This knowledge is helpful later in optimizing the spectra of the protein–DNA complex, since it identifies nucleotides within the DNA molecule that can presumably be altered without affecting stability. To reduce spectral overlap, the minimal DNA sequence with good binding affinity for the protein should be used to make the complex. If the structure of the protein in the DNA-free state is known, a model of it docked to B-form DNA should be constructed. This may help to determine the minimum DNA length that can be used, and whether additional nucleotides outside of the known binding site are required to form nonspecific stabilizing contacts. In practice, DNA molecules studied in our group almost always contain a G:C base pair (bp) at each end, which limits fraying by increasing the melting temperature. Minor changes in the DNA and protein sequences can dramatically affect the NMR spectrum of a protein–DNA complex and are therefore parameters that can be optimized. A common mistake is to choose a DNA fragment that is too long with unnecessary base pairs at either its 3¢ or 5¢ end. This can be problematic as longer DNA fragments can contain weaker, “cryptic” binding sites for the protein that become occupied at the high protein concentrations present in the NMR sample (typically > 0.5 mM). For example, a protein that forms numerous interactions with an A-T sequence located at the center of the primary site might also bind to a secondary A-T dinucleotide sequence present elsewhere in a longer DNA fragment. If this occurs, the multiple binding modes of the protein cause resonance line broadening. Modeling studies and a comparison of the DNA-binding sites can be used to identify potential “cryptic” sites, which can then be eliminated by altering the nucleotide sequence of the DNA molecule. A nucleotide sequence comparison can also identify dsDNA molecules that have NMR spectra that can more readily be assigned. For example, it may be preferable to maximize the number of thymine bases in the sequence, as its methyl groups are good anchor points in the assignment process. The length and sequence of the protein can also be adjusted to improve spectral quality. Typically, this involves deleting unstructured amino acids at the polypeptide termini to reduce spectral overlap. However, even subtle single amino acid changes can have a dramatic impact on spectral quality. For example, in our studies of an ARID-DNA complex, a single phenylalanine-to-leucine mutation was found to dramatically reduce line broadening, salvaging a protein–DNA complex that was originally ill suited for structural analysis by NMR (15). This biochemical approach is not a general method, but may prove useful in the spectral optimization of other protein complexes that suffer from interfacial line broadening caused by dynamic changes in proximal aromatic rings.

222

M.D. Sam and R.T. Clubb

Fig. 1. Flowchart showing the procedures used to form and optimize protein–DNA complexes for NMR studies.

In this chapter, we outline the approaches we typically use to form protein–DNA complexes suitable for NMR studies. The overall procedure for this protocol is outlined in Fig. 1.

2. Materials The exact reagents used to form protein–DNA complexes suitable for high-resolution solution-state NMR studies vary depending upon the specific system that is being studied. In this section, the materials used to produce the Integrase(Int)-DNA complex are described (16). 2.1. Binding Affinity Measurements

1. 2× binding buffer: 40 mM Tris–HCl, pH 7.5, 40 mM NaCl, 40 mM KCl, 10% (w/v) glycerol, 2 mM EDTA, and 2 mM DTT. 2.

32

P-labeled DNA: Labeled at its 5¢-termini with T4 polynucleotide kinase and g-32P-ATP.

3. PhosphorImager (Molecular Dynamics Inc.): To quantify radioactivity in the gels. 4. 1× TBE: For 1 L of 10× TBE, dissolve 108 g of Tris base, 55 g of boric acid, and 7.4 g of disodium EDTA salt. Dilute tenfold to get 1× TBE. 5. Bovine serum albumin (BSA): Stock concentration 1 mg/mL in H2O. 6. Poly dI/dC: Stock concentration 0.5 mg/mL.

13

Preparation and Optimization of Protein–DNA…

223

7. Protein stock solution: 20 mM (4× of the highest protein concentration used in the binding assay). For protein–DNA complexes with nanomolar affinity, a typical titration range is as follows (nM): 5,000, 1,000, 200, 100, 50, 25, 5, 1, 0.5, and 0. Protein should be dissolved in a solution that is most stable for that particular protein. This protein buffer is diluted in the binding assay and replaced with binding buffer solution. 8. 6× DNA loading dyes: 0.03% (w/v) xylene cyanol FF and 0.03% (w/v) bromphenol blue in 10 mM Tris, pH 8, and 30% (w/v) glycerol. 9. 1 M Tris, pH 8: Dissolve 121.1 g of Trizma (MW = 121.1 g/mol) in 800 mL of H2O, titrate with concentrated HCl until pH 8 is achieved, and then bring final volume to 1 L. 2.2. Preparation of Purified SingleStranded DNA

1. DNA oligonucleotides: 1 mmol scale synthesis (Integrated DNA Technologies, IDT) (see Note 1). 2. 17% acrylamide–urea gel: For 400 mL total volume, mix 168.2 g of urea, 170 mL of 40% acrylamide (37.5:1), 40 mL of 10× TBE (Subheading 2.1), 60 mL of H2O, 1 mL of 10% (w/v) APS, and 100 mL of TEMED. 3. DNA loading buffer: 7 M urea, 50 mM Tris, pH 8.0 (using 1 M Tris, pH 8; Subheading 2.1), 5 mM EDTA (using 0.5 M EDTA, pH 8), and 10% (w/v) glycerol. 4. 0.5 M EDTA, pH 8: Add 186.1 g of EDTA disodium salt and ~20 g of NaOH pellets to 700 mL of H2O (EDTA dissolves as it approaches pH 8), and then bring the solution to 1 L once pH 8 is achieved. 5. 6x DNA loading dyes (see Subheading 2.1). 6. 1× TBE (see Subheading 2.1). 7. FLEX TLC plates. 8. Electroelution chamber. 9. Dialysis buffer: 50 mM Tris–HCl, pH 7.5, 200 mM NaCl, and 2 mM EDTA (using 0.5 M EDTA, pH 8).

2.3. Preparation of Duplex DNA for NMR Studies

1. Annealing buffer: Same as dialysis buffer (Subheading 2.2).

2.4. Preparation of the Protein–DNA Complex

1. High-salt protein buffer: 50 mM Hepes, pH 7.0 (using 1 M Hepes, pH 7), 500 mM NaCl, and 1 mM DTT.

2. D2O.

2. 1 M Hepes, pH 7: Stock solution is adjusted to pH with NaOH. 3. High-salt DNA buffer: 50 mM Tris, pH 7.5 (using 1 M Tris, pH 7.5, see Subheading 2.1, except adjust pH to 7.5), 500 mM NaCl, and 0.1 mM EDTA.

224

M.D. Sam and R.T. Clubb

4. Low-salt buffer: 25 mM Hepes, pH 7.0 (using 1 M Hepes, pH 7), 15 mM NaCl, 2 mM DTT, 7% D2O, and 0.01% NaN3. 5. Centricon YM-3 Bioseparations).

centrifugal

filter

device

(Amicon

6. ~50 mM protein solution. 7. ~50 mM dsDNA solution.

3. Methods Our lab has solved the structures of six protein–DNA complexes. Four have been determined by NMR spectroscopy and two have been determined by X-ray crystallography (17–22). Below, we describe the procedures we generally use to form complexes between a sequence-specific binding protein and a duplex-DNA molecule. Unless otherwise stated, the procedures described below are used to produce samples of the complex between the Int protein and its cognate DNA site (17–22). Four procedures are presented: (1) an electrophoretic mobility shift assay (EMSA) for affinity and specificity measurements, (2) methods to purify singlestranded DNA (ssDNA), (3) methods to prepare duplex DNA (dsDNA), and (4) the procedures used to assess the spectral quality of a protein–DNA complex to determine if additional NMR studies are warranted. 3.1. Binding Affinity Measurements

Biochemical assays should be available to rapidly estimate the affinity of wild-type and mutant proteins for different DNA molecules. A variety of methods can be employed to measure binding, such as the EMSA, isothermal titration calorimetry (ITC), fluorescence anisotropy, surface plasmon resonance (SPR; e.g., Biacore), and fluorescence quenching (if the protein contains an appropriately positioned tryptophan). However, we favor the EMSA because it is robust and simple to perform (23). This procedure has been described in detail previously (24) and is outlined below. 1. Mix the following components: 12 mL of 2× binding buffer, 3 mL of 1 mg/mL BSA, 2 mL of 0.5 mg/mL Poly dI/dC (competitor DNA) for a total of 17 mL. 2. Add to this mixture the protein stock solution and the appropriate amount of H2O to achieve a final volume of 23 mL (see Subheading 2.1 and Note 2). 3. Incubate on ice for 20 min. 4. Add 1 mL of 32P-labeled DNA probe (~4,000 cpm/mL). 5. Incubate on ice for 20 min.

13

Preparation and Optimization of Protein–DNA…

225

6. Prepare an 8% polyacrylamide gel and pre-electrophorese the gel by running it for 30–60 min at 10 V/cm in 1× TBE at ambient temperature or 4°C (see Note 3). 7. Load the reaction mixtures onto an 8% polyacrylamide/TBE gel at 4°C (see Note 4). Load 6× DNA loading dyes in a separate lane as a reference to track the migration of the free DNA. The gel run time should be optimized for each specific system so as to resolve the species of interest and to be as short as possible. A typical gel run time is ~1.5 h at constant voltage (10 V/cm), but the voltage should be reduced if the gel becomes warm during electrophoresis. 8. Quantify the amount of free and bound DNA in each lane by using a PhosphorImager system or the equivalent. 9. Determine the dissociation constant by fitting to the following equation: θ = [L ]/ ([L ]+ K d ), where q, [L], and Kd are the fraction of DNA bound, the total protein concentration in the reaction, and the dissociation constant, respectively. q is equal to the counts present in the shifted band divided by the total counts for the DNA (free plus shifted bands). 3.2. Preparation of Purified SingleStranded DNA

In this section, we discuss how to purify large quantities of commercially available ssDNA oligonucleotides for NMR studies. We initially purchase the ssDNA in a crude form at a cost of ~$2 per nucleotide for 1 mmole of material. ssDNA with lengths less than 20 base pairs is purified on 20% acrylamide–urea gels and with lengths longer than 20 bp is purified on 17% gels. The ssDNA is then eluted from the gel and dialyzed into native buffer for further use. The procedure we use generally yields ssDNA that is >98% pure for molecules up to 40 bp in length. For ssDNA shorter than 15 bp, more conventional approaches are sufficient to produce ssDNA suitable for NMR studies (see Note 5). 1. Prepare one 17% acrylamide/urea gel with a single lane (see Note 6). 2. Dissolve the DNA oligonucleotide (1 mmol or 3–6 mg) in 2 mL of DNA loading buffer. 3. Load 20 mL of 6× DNA loading dyes on the right and left edge of the gel within the single lane. The migration of the xylene cyanol FF and bromophenol blue dyes along the gel gives an estimate of how far the DNA has migrated (see Note 7). 4. Run the gel in 1× TBE. It typically takes ~10–13 h for ssDNA (depending on the length) to migrate ~3/4 of the entire gel length. For a single gel running at a constant 50 W, the voltage is ~600–700 V (65–80 mA) (see Note 8). 5. Transfer the gel onto saran wrap by removing the top gel plate and placing a layer of saran wrap directly on top of the gel.

226

M.D. Sam and R.T. Clubb

Invert the gel onto FLEX TLC plates with the saran wrap on top of the TLC plates. Remove the remaining gel plate and in a darkroom, use a handheld UV lamp (254 nm) to locate the ssDNA. Quickly excise the DNA using a razor blade. Cut the excised gel containing the desired ssDNA into ½-in. pieces to ensure efficient DNA electroelution. The purpose of the FLEX TLC plate is to enhance the DNA signals during exposure to UV lights. 6. Assemble an Elutrap™ electroelution device (Whatman). Membranes for trapping DNA are BT1 (14 nucleotide cutoff) and BT2 (cellulose acetate membrane). Alternatively, BT1 can be replaced with a low-molecular-weight cutoff dialysis membrane. 7. Elute the DNA in 1× TBE. Run the Elutrap at 150 V for 8–10 h (remove eluted DNA from the Elutrap two to three times during this period). 8. Dialyze extensively with dialysis buffer to remove all denaturing reagents from the ssDNA. 9. Determine the DNA concentration from a UV absorbance reading at 260 nm (A260). The extinction coefficient may be calculated online using a program provided by IDT (http:// www.idtdna.com/analyzer/Applications/OligoAnalyzer/). 3.3. Preparation of Duplex DNA for NMR Studies

Complementary purified ssDNA molecules are annealed to produce the appropriate DNA duplex. NMR and/or chromatographic approaches are then used to verify duplex formation prior to forming the protein–DNA complex. The steps used to generate dsDNA are outlined below. 1. Dissolve complementary ssDNA molecules in annealing buffer to a final concentration of ~100 mM. The samples should be free of urea and EDTA. 2. Heat the sample to ~95–100°C for 10 min and slowly cool to ambient temperature in the heat block. A water bath may be used in place of a heating block for larger annealing volumes. 3. Perform NMR experiments to ensure that the sample has properly annealed and that no excess ssDNA is present. Add an appropriate amount of D2O to maintain the field lock, and acquire an 1H 2D TOCSY spectrum (mixing time ~40 ms). Acquire spectra of the ssDNA components of the duplex as well. Compare the three spectra; signals from the ssDNA spectra appearing in the duplex spectrum indicate ssDNA excess. Typically, the H5-H6 cross peaks of the cytosine bases are used to discriminate between the single-stranded and doublestranded forms of DNA (see Note 9).

3.4. Preparation of the Protein–DNA Complex

Our group follows a conservative approach when forming protein– DNA complexes to minimize sample loss due to precipitation (17–22) (see Note 10). In this procedure, dilute concentrations of

13

Preparation and Optimization of Protein–DNA…

227

the components are mixed in the presence of high salt, followed by concentration and removal of the salt to form the final NMR sample. Typically, the starting concentration of the dsDNA is ~50–100 mM and it is dissolved in at least 150 mM NaCl, near physiological pH (see Note 11). Pilot studies should be performed prior to embarking on a large-scale production of the complex as we have found that the order of addition of the protein and DNA components can have a dramatic impact on the results obtained. To test for precipitation of the sample upon component mixing, slowly titrate the protein into a ~20 mL sample of the DNA (the final volume of the fully formed complex should be ~40 mL). The concentration of the components is as described above and enough protein should be added to generate a 1:1 complex. The reverse titration in which dsDNA is titrated into a sample of the protein should also be performed. In both cases, carefully observe if any precipitation occurs during the mixing procedure and, more importantly, if any precipitation remains after the titration is complete. A light microscope can provide an easy way to assay for the presence of precipitation. As already mentioned, the order of addition can be critical. For example, in our studies of the Excisionase(Xis)-DNA complex, the addition of dsDNA to a solution of Xis resulted in irreversible precipitation, whereas the reverse titration, titrating Xis into a solution of dsDNA, yielded a soluble complex after mixing (20). To obtain an NMR sample once the protocol for complex formation has been optimized, we typically mix the DNA and protein components such that the final volume of the complex is ~10 mL (~25 mM of the complex). The salt and unwanted buffer components are then removed by dialysis or using a centrifugal filter unit. This step is important for complex stability as the presence of salt tends to destabilize the protein–DNA complex by shielding electrostatic interactions. Generally, to obtain a sample with good NMR spectral properties, several variants of the complex are studied that differ in their pH, complex concentration, and ionic strength. Initial screening is typically performed using ~200–500 mM samples of the complex, which are then concentrated further to construct the final sample once the best conditions are discovered (see Note 12). 1. Prepare ~10 mL of 50 mM Int protein solution dissolved in high-salt protein buffer. Prepare ~10 mL of 50 mM dsDNA solution dissolved in high-salt DNA buffer (see Note 11). 2. Prepare the final NMR sample by slowly titrating 10 mL of 50 mM Int into 10 mL of 50 mM dsDNA. Add ~0.5 mL of protein, mix, and repeat until the titration is complete. The specific order of addition and the conditions used should have been optimized as described above.

228

M.D. Sam and R.T. Clubb

3. Exchange the final sample into low-salt buffer using a protein concentrator (see Note 12). Concentrate the sample to a volume suitable for NMR experiments. Monitor the total amount of complex at each step of this process by measuring the A260 (see Note 13). 3.5. Assessing the Quality of the Protein– DNA Complex

1. To assess the quality of the DNA spectrum, record 1D 1H spectra using a 1331 pulse adjusted to maximally excite the imino protons (25). 2. Compare the spectra of the complex and the isolated dsDNA molecule to assess protein binding. 3. If sufficient material is present, use a similar excitation scheme to record a 2D 1H NOESY spectrum, which reveals whether the appropriate imino–imino cross peaks are present. Ascertain the quality of the protein spectrum by comparing the 1H–15N HSQC spectra of the free and DNA-bound forms of the protein. The spectrum of the complex should be well-resolved, exhibit uniform signal intensities, and differ substantially from the NMR spectrum of the DNA-free protein (Fig. 2). In favorable cases, it may be possible to observe intermolecular NOEs using protein–DNA complexes containing only 15 N-labeled protein. Depending on the structure of the complex, intermolecular NOE cross peaks are sometimes observed in the 2D 1H NOESY spectrum between the imino protons of

Fig. 2. The 1H–15N HSQC spectrum of the Int-DNA complex, which exhibits good dispersion and uniform line widths. Data for this complex were suitable to determine a highresolution NMR structure (17–22).

13

Preparation and Optimization of Protein–DNA…

229

the DNA and protein protons that resonate upfield of 1.2 ppm. This is because the most upfield resonances in the 1H spectrum of DNA are the thymine H5 methyl groups (1.2 and 1.6 ppm). Intermolecular NOEs between the protein amide and DNA imino protons can sometimes also be seen in the 3D 15N-edited NOESY spectrum of the 15N-labeled complex. However, a full assessment of whether a complex is suitable for structure determination by NMR requires the acquisition of the appropriate 2D and 3D edited and filtered NOESY experiments using samples in which the protein is labeled with 13C and 15N.

4. Notes 1. A variety of companies sell pure or partially purified singlestranded oligonucleotides that are synthesized using phosphoramidite chemistry (e.g., IDT, Invitrogen, Sigma, and Applied Biosystems). To save money, we typically obtain standard, desalted, unpurified oligonucleotides, which are then further purified in-house. Olignonucleotides greater than 60 nt in length can be produced via enzymatic reactions, which may also be applied to produce DNA molecules enriched with 13C and 15N (26, 27). 2. In this procedure, binding isotherms are generated by varying the protein concentration (0, 0.5, 1, 5, 25, 50, 100, 200, 1,000, 5,000 nM protein). 3. Alternatively, Tris–acetate–EDTA (TAE) may be used for the gel casting mixture and running buffer. 4. A 5–15% gel could be used depending on the respective sizes of the individual components and formed complex. 5. A classical and more straightforward method to purify ssDNA from a crude synthesis is to use an HPLC reverse-phase column (C4 to C18; porous hydrocarbon silica gel) with 0.1 M triethylammonium acetate (TEAA; mobile phase A) and ace-tonitrile (mobile phase B). A complete purification protocol and a column selection guide are described in Current Protocols Nucleic Acid Chemistry by Andrus et al. (28). A Mono Q column (GE, Mono Q HR5/5) on an FPLC can be used to purify ssDNA under denaturing conditions. Buffers for this purification scheme are as follows: (a) Buffer A: 50 mM Tris, pH 7.5, 6 M urea, and (b) Buffer B: 50 mM Tris, pH 7.5, 6 M urea, 1.5 M NaCl. No more than 1.5 mmol of ssDNA should be loaded onto this column for good separation. A 25 mL gradient (flow rate = 1 mL/min) from 10 to 30% buffer B should be sufficient to ensure good separation of ssDNA from impurities. It should be noted that all buffers and resuspended DNA

230

M.D. Sam and R.T. Clubb

should be filtered through a 0.2-mm filter before applying to the column. 6. If the DNA is longer than ~20 nucleotides, a 20% acrylamide gel should be used instead. 7. On a 10% gel, xylene cyanol migrates equivalent to a 55-nucleotide ssDNA molecule and bromophenol blue migrates equivalent to a 12-nucleotide ssDNA molecule. On a 20% gel, xylene cyanol and bromophenol blue migrate equivalent to 28- and 8-nucleotide ssDNA molecules, respectively. 8. Power settings for a single 20% acrylamide/7 M urea gel should be set to 30 W and maximum voltage and current. For a single gel running at a constant 30 W, the voltage is ~600–650 V and the current is 45–50 mA. 9. An alternative approach to ensure that solutions of dsDNA do not contain excess ssDNA is to provide a 10% excess of one of the DNA strands in the annealing reaction. After annealing, a Mono Q column is then used to separate the dsDNA molecule from excess ssDNA under native conditions. The following buffers can be used: Buffer A: 50 mM Tris, pH 7.5, and 1 mM EDTA; Buffer B: 50 mM Tris, pH 7.5, 1 mM EDTA, and 1.5 M NaCl. For good separation of the two DNA species, no more than 2 mg of DNA should be loaded onto a 1 mL Mono Q column. ssDNA typically elutes around 25% buffer B (0.375 M NaCl) and dsDNA typically elutes at ~35% buffer B (0.525 M NaCl). A 25 mL gradient from 10 to 45% NaCl running at 1 mL/min works well for oligos between 10 and 30 bp. 10. A variety of approaches can be used to successfully make protein–DNA complexes for NMR studies. A strategy commonly used in the literature is to titrate a ~1 mM 15N-labeled sample of the protein with a concentrated stock solution of the DNA molecule. 1H–15N HSQC spectra are recorded at various points during the titration until the desired stoichiometry is reached. Although the titration method is simple, mixing concentrated protein and DNA samples can cause the resulting complex to precipitate. 11. These conditions prevent a shift in the equilibrium from dsDNA to ssDNA that occurs at low-salt concentrations. A similar protein concentration is used when forming the complex; however, the salt and pH in the protein solution can vary and are chosen to maximize the stability and solubility of the free protein. 12. Frequently, the components of the complex are precious and therefore methods that minimize sample loss during concentration are desired. One trick to concentrate small-volume samples (< 2 mL) is to partially evaporate the complex. In this procedure, a weak flow of nitrogen gas is blown over the

13

Preparation and Optimization of Protein–DNA…

231

sample while it rests in the NMR tube. This can be accomplished by passing the gas through drawn-out pipette that is in turn inserted into the NMR tube. The buffer components are also concentrated during this process, and therefore the initial buffer conditions must be chosen accordingly. 13. A rough estimate of the concentration of the complex can be obtained by measuring its A260 immediately after mixing the components. The amount of the material present at this point in the preparation procedure is known. Therefore, it is possible to estimate the extinction coefficient of the complex by determining the optical absorbance of the complex at 260 and 280 nm. Once estimated, the extinction coefficient enables the concentration of the complex to be readily determined as it is concentrated to a volume suitable for NMR. It also enables the yield of the concentration procedure to be determined as the total amount of complex before and after concentrating can be determined.

Acknowledgments We thank Dr. Evgeny Fadeev for making Fig. 2. This work was supported by a grant from the National Institutes of Health to R.T.C. (R01 AI52217). References 1. Pabo, C. O., and Sauer, R. T. (1992) Transcription Factors-Structural Families and Principles of DNA Recognition. Annu. Rev. Biochem. 61, 1053–95. 2. Garvie, C. W., and Wolberger, C. (2001) Recognition of specific DNA sequences. Mol. Cell. 8, 937–946. 3. Nadassy, K., Wodak, S. J., and Janin, J. (1999) Structural features of protein-nucleic acid recognition sites. Biochemistry 38, 1999–2017. 4. Jen-Jacobson, L. (1997) Protein-DNA recognition complexes: conservation of structure and binding energy in the transition state. Biopolymers 44, 153–180. 5. Boehr, D. D., Nussinov, R., and Wright, P. E. (2009) The role of dynamic conformational ensembles in biomolecular recognition. Nat. Chem. Biol. 5, 789–796. 6. Billeter, M., Qian, Y. Q., Otting, G., Muller, M., Gehring, W., and Wuthrich, K. (1993) Determination of the nuclear magnetic resonance

7.

8.

9.

10.

solution structure of an Antennapedia homeodomain-DNA complex. J. Mol. Biol. 234, 1084–1093. Omichinski, J. G., Clore, G. M., Schaad, O., Felsenfeld, G., Trainor, C., Appella, E., Stahl, S. J., and Gronenborn, A. M. (1993) NMR structure of a specific DNA complex of Zn-containing DNA binding domain of GATA-1. Science 261, 438–446. Kalodimos, C. G., Biris, N., Bonvin, A. M., Levandoski, M. M., Guennuegues, M., Boelens, R., and Kaptein, R. (2004) Structure and flexibility adaptation in nonspecific and specific protein-DNA complexes. Science 305, 386–389. Qian, Y. Q., Otting, G., and Wuthrich, K. (1993) NMR detection of hydration water in the intermolecular interface of a protein-DNA complex. J. Am. Chem. Soc. 115, 1189–1190. Iwahara, J., and Clore, G. M. (2006) Direct observation of enhanced translocation of a

232

11.

12.

13.

14.

15.

16.

17.

18.

M.D. Sam and R.T. Clubb homeodomain between DNA cognate sites by NMR exchange spectroscopy. J. Am. Chem. Soc. 128, 404–405. Tjandra, N., and Bax, A. (1997) Direct measurement of distances and angles in biomolecules by NMR in a dilute liquid crystalline medium. Science 278, 1111–1114. Clore, G. M., and Iwahara, J. (2009) Theory, practice, and applications of paramagnetic relaxation enhancement for the characterization of transient low-population states of biological macromolecules and their complexes. Chem. Rev. 109, 4108–4139. Pervushin, K., Riek, R., Wider, G., and Wuthrich, K. (1997) Attenuated T2 relaxation by mutual cancellation of dipole-dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution. Proc. Natl. Acad. Sci. USA 94, 12366–12371. Cavanagh, J., Fairbrother, W. J., Palmer, A. G. I., Rance, M., and Skelton, N. J. (2006) Protein NMR Spectroscopy: Principles & Practice (2nd ed.), Academic Press, San Diego. Iwahara, J., Wojciak, J. M., and Clubb, R. T. (2001) Improved NMR spectra of a proteinDNA complex through rational mutagenesis and the application of a sensitivity optimized isotope-filtered NOESY experiment. J. Biomol. NMR 19, 231–241. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning, A laboratory manual, 2nd ed., Cold Spring Harbor Laboratory Press. Abbani, M. A., Papagiannis, C. V., Sam, M. D., Cascio, D., Johnson, R. C., and Clubb, R. T. (2007) Structure of the cooperative Xis-DNA complex reveals a micronucleoprotein filament that regulates phage lambda intasome assembly. Proc. Natl. Acad. Sci. USA 104, 2109–2114. Iwahara, J., Iwahara, M., Daughdrill, G. W., Ford, J., and Clubb, R. T. (2002) The structure of the Dead ringer-DNA complex reveals how AT-rich interaction domains (ARIDs) recognize DNA. EMBO J. 21, 1197–1209.

19. Fadeev, E. A., Sam, M. D., and Clubb, R. T. (2009) NMR structure of the amino-terminal domain of the lambda integrase protein in complex with DNA: immobilization of a flexible tail facilitates beta-sheet recognition of the major groove. J. Mol. Biol. 388, 682–690. 20. Sam, M. D., Cascio, D., Johnson, R. C., and Clubb, R. T. (2004) Crystal structure of the excisionase-DNA complex from bacteriophage lambda. J. Mol. Biol. 338, 229–240. 21. Wojciak, J. M., Connolly, K. M., and Clubb, R. T. (1999) NMR structure of the Tn916 integrase-DNA complex. Nature Struct. Biol. 6, 366–373. 22. Wojciak, J. M., Iwahara, J., and Clubb, R. T. (2001) The Mu repressor-DNA complex contains an immobilized “wing” within the minor groove. Nature Struct. Biol. 8, 84–90. 23. Buratowski, S., and Chodosh, L. A. (2001) Mobility shift DNA-binding assay using gel electrophoresis. Curr. Protoc. Mol. Biol., Chapter 12, Unit 12 2. 24. Taylor, J. D., Ackroyd, A. J., and Halford, S. E. (1994) The gel shift assay for the analysis of DNA-protein interactions, in DNA-protein interactions, principles and protocols (Kneale, G. G., Ed.), Humana Press, Totowa, NJ. 25. Hore, P. J. (1983) A new method for water suppression in the proton NMR spectra of aqueous solutions. J. Magn. Reson. 54, 539–542. 26. Louis, J. M., Martin, R. G., Clore, G. M., and Gronenborn, A. M. (1998) Preparation of uniformly isotope-labeled DNA oligonucleotides for NMR spectroscopy. J. Biol. Chem. 273, 2374–2378. 27. Xiong, A. S., Yao, Q. H., Peng, R. H., Duan, H., Li, X., Fan, H. Q., Cheng, Z. M., and Li, Y. (2006) PCR-based accurate synthesis of long DNA sequences. Nat Protoc. 1, 791–797. 28. Andrus, A., and Kuimelis, R. G. (2001) Analysis and purification of synthetic nucleic acids using HPLC. Curr. Protoc. Nucleic Acid Chem., Chapter 10, Unit 10 5.

Chapter 14 NMR Studies of Protein–Ligand Interactions Michael Goldflam, Teresa Tarragó, Margarida Gairí, and Ernest Giralt Abstract Nuclear magnetic resonance (NMR) has evolved into a powerful tool for characterizing protein–ligand interactions in solution under near physiological conditions. It is now frequently harnessed to assess the affinity and specificity of interactions; to identify binding epitopes on proteins and ligands; and to characterize the structural rearrangements induced by binding. The first section of this chapter provides a general overview of the NMR study of protein–ligand interactions. The section is divided according to two main categories of experiments: those based on observing protein signals and those based on observing ligand signals. The next section explains two case studies performed in the authors’ laboratory. The first of these deals with the interaction between vascular endothelial growth factor and a peptidic ligand, and includes a detailed protocol of chemical shift perturbation experiments. The second one reports on the interaction between prolyl oligopeptidase and a small molecule as monitored by ligand saturation transfer difference (STD), and illustrates how NMR can be used to confirm binding and to identify the binding epitope of a ligand. Key words: Protein–ligand interactions, Chemical shift perturbation, Saturation transfer difference, NMR, Vascular endothelial growth factor, Prolyl oligopeptidase, Protein observed experiments, Ligand observed experiments

1. Introduction Protein–ligand interactions are integral to diverse biological processes. They include the interaction of proteins with signaling molecules, such as neurotransmitters and hormones, or cofactors, as well as antigen recognition and enzyme–substrate interactions. In all of these processes, correct biological functioning of the protein requires that it specifically recognize a ligand at a particular binding area on its surface.

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_14, © Springer Science+Business Media, LLC 2012

233

234

M. Goldflam et al.

Deep knowledge of these processes and their underlying mechanisms is necessary not only for understanding these events at the molecular level, but also for being able to selectively modulate these interactions to provoke a desired biological response. This can be done by modifying natural compounds or by developing completely new compounds. Both cases offer a nearly unlimited pool of small organic molecules, peptides, carbohydrates, or mixtures thereof. Whatever the potential of these compounds to interact with a given protein, recognition itself is steered by the structural orientation of the protein’s functional groups. Thus, elucidation of these interactions greatly facilitates selection of appropriate functional groups in an appropriate framework. Protein–ligand interactions can be studied with several tools, nearly all of which can provide information on binding strength and specificity. This information can be complemented with data acquired by isothermal titration calorimetry (ITC), mass spectrometry (MS), surface plasmon resonance (SPR), and nuclear magnetic resonance (NMR). ITC records the change in temperature of a protein solution upon titration with a ligand solution in an isolated chamber (1). It enables determination of thermodynamic parameters, including the free energy (ΔG), enthalpy (ΔH), and entropy (ΔS) of the interaction, and the change in heat capacity (ΔCp). In mass spectrometry, various techniques are used to ionize compounds or complexes and subsequently analyze their mass-tocharge ratios. Recent developments in MS have facilitated the study of protein–ligand interactions, allowing the detection and characterization of individual conformational states of protein complexes (2). Owing to the high sensitivity of MS, only minute amounts of sample are needed. The study of hydrogen–deuterium exchange of protein backbone amide hydrogens can give information on the binding epitope of a ligand; however, this is most amenable to higher affinity ligands. Finally, MS is one of the few methods that enable study of complexes in gas phase. Comparison between binding energies in gas phase and in solution may advance understanding of the forces behind protein–ligand interactions and, more precisely, help establish the role of solvation in molecular recognition at protein surfaces (3). SPR probes the interaction between an analyte in solution and a biomolecular recognition element immobilized on a sensor surface (4). It enables direct determination of the binding kinetics parameters kon and koff, from which thermodynamic parameters can be quantified. If the protein is the immobilized binding partner, then only small amounts are necessary. The main drawback of SPR is that it requires immobilization of one of the binding partners, which may influence the protein–ligand interaction. NMR has evolved into a powerful tool for obtaining massive amounts of data on inter- and intramolecular processes. Use of NMR

14

NMR Studies of Protein–Ligand Interactions

235

to detect protein–ligand interactions is widely documented in the literature (5–7). An advantage of NMR over other techniques is that it provides access to a broad set of experiments that have been optimized for various objectives: determination of affinity and specificity; identification of binding epitopes on the protein and on the ligand; characterization of structural rearrangements induced by binding; and turnover of substrates by enzymes. Furthermore, since the experiments are performed in solution, physiological or near physiological conditions are possible. Another advantage of NMR is that it is not limited to high-affinity systems: it can be applied to study very weak interactions (i.e., mM range), for which other techniques are often unsuitable (8). Moreover, for low-affinity systems, NMR offers a relatively low incidence of false positive and false negatives compared to other analytical approaches. The main limitation of NMR is its low sensitivity. Also, compared to other techniques, NMR experiments are intrinsically low-throughput. Nevertheless, improved automation, and development of high sensitivity probes (e.g., cryoprobes), new pulse sequences, efficient isotopic labeling techniques, and more powerful magnets, have all contributed significantly to minimize these limitations. NMR experiments for protein–ligand interactions fall into two main categories: either studying them from the perspective of the protein or from the perspective of the ligand. In the following section, both approaches are overviewed and some typical experiments from each group are analyzed. 1.1. Protein Observed Experiments

Although in some cases monodimensional (1D) 1H-NMR experiments have been used to characterize the protein–ligand binding by following the 1H chemical shifts of specific residues in the protein, most experiments on protein observation entail bidimensional (2D) NMR. Conventional 1D-1H spectra typically cannot resolve the individual proton signals of the protein. This limitation can be overcome by distributing the information along two dimensions and by employing heteronuclear spectroscopy (i.e., studying magnetically active nuclei other than protons that are present in proteins, such as 15N and 13C). Since the natural abundance of 15N and 13 C (0.37% and 1.1%, respectively) is too low for NMR experiments, the protein to be studied must be isotopically labeled, usually, through expression in E. coli. Several efficient labeling schemes are available, and choosing the right one can greatly simplify NMR studies of proteins. The most widely used labeling method is uniform labeling with 15 N. For proteins expressed recombinantly in E. coli, 15N (in the form of an ammonium salt) is added to the expression media. This simple modification provides near quantitative isotopic labeling of the protein: all backbone amides as well the nitrogen containing side chains are labeled with this magnetically active nucleus. Heteronuclear 1H-15N correlation NMR experiments that allow

236

M. Goldflam et al.

direct observation of J-coupled 1H to 15N nuclei generate spectra containing at least one signal for each amino acid, except proline. Additional signals arise from amides in the side chains. When signal assignment is available, this strategy enables mapping of changes in the protein’s backbone amides that are induced by binding of a ligand, and if the 3D structure of the protein is known, then the regions directly involved in the binding process can be easily identified. Another common labeling scheme for studying protein–ligand interactions is amino acid specific labeling, in which the desired amino acid – or a suitable precursor – is added prelabeled to the expression media and an auxotrophic bacterial strain is used. This is advantageous for large proteins (i.e., >40 kDa), for which it provides far simpler spectra than those obtained with uniform labeling of the backbone. In this context, a type of amino acid is selected which is well distributed throughout the protein sequence and which can serve as a probe for changes induced by the ligand–protein interaction. The authors of this review recently used this strategy to map changes induced by ligand binding to POP, an 80-kDa protein, using a 15N-indole selective labeling scheme of the 12 Trp residues in the enzyme (9). Selective labeling of one type of amino acid can also be attractive for small proteins. According to the Hot Spot theory proposed by Bogan et al., specific amino acids are concentrated in regions of the protein that contribute to interactions with other proteins or ligands (10). Therefore, one of these amino acids may serve as a site-specific probe. Even if the assignment is incomplete, this scheme can identify ligands that bind to a zone of interest. For methyl-bearing side chains, 13C labeling provides a very sensitive probe. Due to its mobility and the presence of three degenerated protons, the methyl group generates a high intensity signal in 2D heteronuclear 1H-13C correlation NMR experiments, while being extremely sensitive to environmental changes. The advantages and applications of using selectively 13C-labeled methyl groups in the NMR study of large biomolecules have been reviewed by Tugarinov (11). The key experiment for the study of protein–ligand interactions in 15N-labeled target proteins is the 1H-detected 2D-[15N, 1 H]-HSQC experiment (12, 13). For uniformly 15N-labeled samples, at least one signal per amino acid is observed. The basic experiment comprises four main blocks (Fig. 1a). Block A comprises an INEPT module (14), whose purpose is to transfer nuclear spin polarization between J-coupled nuclei – in this case, from the more sensitive one, 1H, to the less sensitive one, 15 N. Since the scalar coupling constant is adjusted to the 1JHN value (ca. 90–95 Hz), only magnetization of amide protons is transferred to the adjacent 15N nucleus. During block B, a 15N frequency labeling is achieved by incrementing the variable delay t1, which leads to generation of the indirect dimension of the 2D spectrum

14

NMR Studies of Protein–Ligand Interactions

237

Fig. 1. (a) Basic pulse sequence for the 1H-15N HSQC experiment. The narrow and wide bars depict 90° and 180° pulses, respectively. The delays (t ; equal to 1/[41JHN]) allow magnetization evolution to be transferred between coupled nuclei. 15N magnetization evolves for t1 and 1H magnetization is directly detected during t2. Double Fourier transformation along t1 and t2 generates a 2D correlation spectrum with frequencies F1 and F2, respectively, as shown in (b) Every signal in the spectrum corresponds to one NH group in the protein and gives information on the chemical shifts of an amide nitrogen (F1) and an amide proton (F2) that are directly coupled through the coupling constant 1JHN.

(F1 frequency). Block C comprises a reverse INEPT module. Nuclear spin polarization is again transferred – this time, from 15N to 1H. This enables data acquisition in block D, in which 1H magnetization is directly detected during t2, which corresponds to the F2 frequency in the 2D spectrum (see Fig. 1b). Both excitation and direct detection of 1H, the nucleus with the higher gyromagnetic ratio, provide a highly sensitive NMR experiment. Since labile (e.g., amide) protons are observed in the experiment, protein NMR must be performed in H2O, rather than in D2O. Proton concentration in H2O (ca. 100 M) is usually several orders of magnitude higher than that of the protein (mM range), which implies a wide dynamic range. Thus, the H2O signal must be strongly attenuated in order to observe the protein protons at a sufficient signal-to-noise ratio in the NMR spectrum. Water suppression is thereby a critical requisite that must be experimentally optimized. Currently, most schemes that provide good water elimination (15) are based on using pulsed field gradients and proton selective pulses that enable manipulation of the H2O magnetization independently of that of the protein. Protein observed experiments for studying protein–ligand interactions are very simple: the chemical shifts of the protein signals change upon binding of the ligand. The resulting chemical shift perturbation (CSP) provides the basis for detecting binding. Moreover, if signal assignment is available, the exact location of the interaction on the protein surface can be mapped. Although this approach was pioneered by several authors, including Gerhard Wagner, it is strongly associated with Stephen Fesik and his colleagues at Abbott Laboratories, who coined the term SAR by NMR (16) to describe the use of CSP for establishing structure– activity relationships (SAR) in drug discovery.

238

M. Goldflam et al.

Interaction of the protein with a ligand affects not only the local magnetic environment of the backbone amides, but also the protein’s dynamics. In principle, NMR is well suited for studying protein dynamics, although this approach is still in its infancy for protein–ligand interactions. Smrcka et al. (17) studied the intensities of 15N-Trp labeled G-protein βγ subunits in the presence and absence of ligands to gain insight into these subunits ability to interact with diverse molecular partners. They concluded that the wide range of signal intensities corresponding to different Trp residues is related to differences in local mobility, which is the underlying mechanism behind their molecular promiscuity. The experiments done in the presence of a ligand supported this idea, since the intensities of residues close to the ligand decreased upon binding (17). In addition to binding affinity, binding kinetics are also decisive in CSP experiments. However, since the theory behind this is already covered in the literature (7, 18), only a qualitative description of the phenomenon and its impact on CSP are provided here. Depending on the system being studied, the kinetic constants of the binding event can be much faster or much slower than the difference between the chemical shifts of the bound and free states. This leads to a range of behaviors in CSP experiments, whereby increasing amounts of a ligand are titrated into a protein sample and ligand-induced chemical shift changes are subsequently detected. In the fast exchange regime (see Note 1) the exchange between the bound and free form is faster than the difference in chemical shifts. Only one set of protein signals is visible, and their positions typically shift according to the ratio between the bound and free species. Therefore, the chemical shifts move from the free form of the protein to the position of the bound state, which is reached once the protein sample has been completely saturated with ligand. If the same amount of ligand is used in each titration step, then the chemical shifts will change asymptotically. This can be fit to a mathematical model and used to calculate the affinity (KD) of the interaction. In the slow exchange regime the situation is reversed: exchange is slower than the difference in chemical shifts between the bound and free states. Therefore, the bound and free states give separate signals. In the course of the titration experiment, the signal of the free protein declines while a new signal appears at the position of the bound state, which increases in intensity until becoming the only observable signal at the saturation point. Fast and slow exchange regimes are not isolated extremes: they are linked by the intermediate regime, whereby the rate of exchange between the bound and free states is comparable to the difference in chemical shifts between these two states. Consequently, the behavior is more complicated, as it entails a mixture of signal shifts, decreasing signals, and newly appearing signals. This results in very

14

NMR Studies of Protein–Ligand Interactions

239

broad signals and non-Lorentzian line shapes, which makes analysis very difficult (19). The equilibrium dissociation constant KD can be used for quantification of exchange regimes. If a diffusion controlled on rate with kon ~ 108 M-1s-1 is assumed, then koff can be estimated: ligands with KD < 1–10 nM and koff ~ 0.1–1/s will be in the slow regime; ligands with KD > 10 μM and koff > 103/s will be in the fast regime; and ligands with values in between these will fall in the intermediate regime. However, these values are only valid if the association is indeed diffusion-controlled. CSP experiments are not always easy to interpret, chiefly due to the difficulty in distinguishing between short- and long-distance effects. Short-distance effects are perturbations resulting from the interaction of residues with the ligand. They delineate the binding zone of the ligand. Long-distance effects are perturbations caused by structural rearrangements of the protein under ligand binding. Although detection of long-distance effects may be of interest, they can give misleading information if the ligand interaction zone is unknown. Long distance effects markedly complicate the study of very flexible systems and weak ligands. To overcome this problem, Fesik et al. performed CSP studies of FKBP and of Bcl-XL using closely related ligands (20). Although in both cases all ligands caused massive perturbations, the differences among the perturbations of these related ligands enabled identification of the binding site and the crude orientation of the ligands. More recently, Krishnamoorthy et al. addressed this issue by proposing a new way to analyze NMR CSP data in detail (21). 1.2. Ligand Observed Experiments

All ligand observation experiments are based on the difference in NMR parameters between the bound and free states of the ligand. The changes in nuclear Overhauser effects (NOEs) when ligands bind to receptor proteins are especially interesting (22). Ligands with molecular weight lower than 1,000 U exhibit short correlation times (τc) and show only weak positive NOEs, very small negative NOEs, or no NOEs at all, depending on the magnetic field strength and the molecular weight. Proteins, due to their size, show large τc, large negative NOEs, and highly efficient spin diffusion. Upon binding, the ligand forms a high molecular weight complex with the protein; consequently its properties change, especially its NOE behavior, with the appearance of strong negative NOEs, usually called transferred NOEs (trNOEs). The difference in the properties between the bound and free ligand is as large as the difference in molecular weight between the ligand and the protein–ligand complex. Since these differences have a direct impact on the observable NMR parameters of the ligand, several experiments can be used to detect and characterize the binding event. Most ligand observed methods are based on one of the following: assessment of changes in conventional NMR parameters

240

M. Goldflam et al.

of the ligand (e.g., line widths, chemical shits, relaxation properties, and diffusion); or observation of intermolecular proton magnetization transfer from the protein to the free ligand (via the bound ligand), to distinguish between binding and nonbinding ligand molecules. There are myriad ligand observed experiments currently available, some of which are briefly introduced in the following section. One of the first reported applications from the first category above entailed using 1H NMR to observe the binding-induced chemical shift changes in certain signals of a ligand upon its interaction with a protein. However, because changes in chemical shifts are small compared to line width changes, experiments based on relaxation rate effects have been more extensively used. One such experiment is the Carr–Purcell–Meiboom–Gill (CMPG) filtered 1 H spectrum (23), in which, an R2 relaxation filter comprising of a train of conveniently spaced 180° pulses is applied prior to data acquisition. Provided that the ligand remains bound to the protein long enough to adopt its relaxation behavior, this method, when adjusted properly, removes signals from the quickly relaxing protons of the bound ligand as well as those of the protein. This procedure is also useful since the degree of the attenuation in ligand signals can be used to rank the affinity of various ligands. As mentioned above, the sign and size of the NOEs of a small molecule (i.e., ligand) will change when binding to a receptor protein. Transient NOE experiments (see Note 2) such as 2D NOESY (22) can be performed to observe transferred-NOEs to determine conformations of ligands bound to proteins (24). During the mixing time of the NOESY, the NOEs build up to a maximum value, and the difference in build-up rate among transferred-NOES and NOEs from the free ligand is the key point for ligand-binding detection: for binding ligands, trNOE rates range from 50 to 100 ms, whereas for nonbinding ligands, larger values (200–1,000 ms) are typical. However, this experiment is less sensitive than other experiments (e.g., STD). Nevertheless, its value lies in enabling structure determination of the bound conformation of the ligand in the complex, when intramolecular trNOEs are detected. The intermolecular trNOEs between a ligand and a protein can be used to establish the orientation of the bound ligand in the protein’s binding pockets. The transfer NOE effect can be considered as a precursor to experiments in the second category described above, which are currently very popular. Responses of magnetization transfer experiments are based on exchange-averaged parameters and are affected by many experimental parameters. Saturation transfer difference (STD) and Water-Ligand Observed via Gradient Spectroscopy (WaterLOGSY) are among the most important of these experiments. Case study 2 is based on the use of STD NMR; therefore, this experiment is described in more detail.

14

NMR Studies of Protein–Ligand Interactions

241

STD was introduced in 1999 by Bernd Meyer in two seminal papers (25, 26). He described the experiment in studying the interaction of wheat germ agglutinin with saccharides, and reported its potential use for analyzing mixtures of putative ligands. Several other STD experiments have since been reported, using a broad range of targets, including transmembrane receptors on whole cells. An ideal sample for an STD experiment comprises a medium to high molecular weight protein plus a low molecular weight ligand in a deuterated buffer. The ligand is in high excess over the protein, to which it binds in fast exchange; these are common conditions for low affinity ligands. A conventional 1D 1H NMR spectrum of this sample will show a combination of broad peaks, corresponding to the protein protons and narrow peaks, corresponding to the ligands protons. The low concentration of the protein and its short T2 will generate low intensity signals distributed along the entire spectrum as follows: δ 10–6 ppm (amide and aromatic protons); δ 6–4 ppm (α-protons); and δ 4 to −1 ppm (aliphatic protons). Ligand signals vary strongly by ligand structure, but usually appear at δ > 0.8. This leaves a high-field spectral region occupied exclusively by protein signals. STD is based on difference spectroscopy, so two sets of experiments, the on-resonance and the off-resonance, are acquired. In the on-resonance experiment a frequency-selective pulse is repetitively applied to the sample in the aforementioned range in which only protein signals are present (e.g., at −1 ppm) to saturate these protons, which are chiefly methyl groups of aliphatic side chains. The magnetic saturation will transfer to protons located in proximity, and then spread over the entire protein due to fast spindiffusion and fast cross-relaxation mechanisms. This process is observed in the 1H-spectrum as a nearly complete disappearance of protein signals. If a ligand present in the sample then binds to the protein, it will form part of this high molecular weight system, and consequently, will receive part of that magnetic saturation. Interestingly, the degree of saturation received by each ligand proton is not equal, but rather depends on their proximity to the protein. Therefore, this property can be used to determine the binding epitope of the ligand. The on-resonance experiment must be compared to a reference (i.e., the off-resonance experiment), in which no magnetic saturation of the protein is performed, and therefore, no change in signal intensities is observed. Subtracting the on-resonance spectrum from the off-resonance spectrum provides the STD spectrum, in which distinguishing whether or not a ligand binds to the protein is easy, since only the signals of the binding molecule are visible. To minimize the appearance of artifacts in the resulting difference spectrum, the on- and off-resonance experiments must be completely comparable. Because of this, in the off-resonance

242

M. Goldflam et al.

Fig. 2. Basic pulse sequence for an STD experiment. Selective saturation via a train of N selective 90° pulses separated by a delay δ is performed during block A. Protein signals are suppressed during an R2 relaxation filtering delay (block B). After a module for water suppression (block C), the 1H signal is detected during the FID (block D).

experiment a train of frequency-selective pulses is applied to a spectral region lacking ligand and protein signals (e.g., 40 ppm). Furthermore, both experiments are acquired in an interleaving manner to reduce the impact of equipment instabilities. Figure 2 is a schematic of the STD pulse scheme. The core of the program comprises three main blocks. During block A, a train of selective pulses is applied for a total saturation time (tSat). During the on-resonance experiment, the selective pulses are applied only to the protein signals at high-field, whereas in the off-resonance experiment the pulses are applied to a region far off-resonance from the protein and ligand signals. During block B, after a hard 90° pulse, a spin lock (R2 relaxation filter) is applied to remove protein background signals. Block C comprises a water suppression module for samples containing a significant amount of H2O (i.e., >20%). This block can be omitted when working in D2O or other organic solvents. The widespread use of STD NMR stems from its many attractive features. First, STD is amenable to high molecular weight therapeutic targets. In fact, the larger the target, the more favorable the conditions for the experiment. Magnetic saturation is easily achieved for larger targets, which promotes saturation transfer from the protein to the ligand better than for smaller targets. STD works well for large receptors (>30 kDa). For masses lower than 10 kDa, special attention must be paid because the R2 relaxation rate may be insufficient for the intramolecular spreading of the saturation and the intermolecular transfer to the ligand. These cases may demand longer saturation pulse trains, or either addition of viscosity enhancing reagents or use of lower temperatures to slow molecular tumbling. Secondly, STD experiments can be run with a minute amount of protein, only low micromolar concentrations are usually chosen. This is true because the ligand is present in molar excess (generally, 100-fold) over the protein. Assuming fast exchange, one molecule of protein can bind to a multitude of ligand molecules during the total saturation time tSat (usually 1–3 s). Due to the small R1 relaxation values for the free ligand state, free ligand molecules conserve

14

NMR Studies of Protein–Ligand Interactions

243

the magnetic saturation received by the protein, which leads to a buildup of saturated ligands in the sample imprinted with the information of the binding event. This signal amplification is what makes STD more sensitive than other techniques. Thirdly, the STD experiment is easy to implement. Optimization of the on-resonance frequency for each protein is important, such that only protein signals are selectively irradiated. Although the ligand-to-protein ratio, saturation time, and length of the R2 relaxation filter can all be optimized, in most cases, STD signals will already be observed with standard (default) parameters. Unfavorable kinetics of the ligand exchange may be improved by changing the temperature at which the experiment is performed (27). Fourth, the binding epitope of a ligand (the specific portions of the ligand surface critical for molecular recognition) can be estimated from STD experiments (28) by exploiting the fact that STD signal intensities (ISTD) are not equal for the different protons in the ligand. The usual interpretation is that the larger the STD response, the closer the contact between the protein and the ligand. However, the magnitude of the STD signals depends not only on the proximity to the receptor, but also on the longitudinal relaxation times (T1) of the free ligand; thus, the STD response depends both on intermolecular cross-relaxation with the saturated receptor protons and on autorelaxation. STD effects at long saturation times may be misinterpreted for protons in molecules having significantly different T1 values. To determine the binding epitope without the bias of different relaxation times (T1), the STD experiment must be performed at different saturation time (tSat) values (29). Experimental data are fitted to the STD build-up curves for each proton (having a different T1) to obtain the slope of the monoexponential equation (STDmax) and the saturation rate constant (ksat): STD Ampl. = STDmax × [1 − exp(−kSat × t Sat )]

(1)

whereby STD Ampl. = e × h = e ×

I STD I0

(2)

STDAmpl. corresponds to the STD amplification factor (28) and is a correction for total ligand concentration to the STD effect; ISTD is the intensity of an individual proton in the STD spectrum, and I0, intensity of the same proton in the reference spectrum; e is the ligand excess; h is the fraction of ISTD from I0; STDmax is the maximal STD intensity achievable with long saturation times and corresponds to the STD intensity in the absence of T1 bias. Finally, STD experiments can be used to determine dissociation constants if a titration curve is recorded with varying ligand concentrations at the same saturation time (30). To do this, the

244

M. Goldflam et al.

STD-amplification factors are first determined as described above, and then plotted against the ligand concentration. For one-site binding models, the curve can be fitted to the following equation: STDmax × [L] STD Ampl. = (3) K D + [L] whereby [L] is the concentration of the ligand; and KD is the affinity constant of the ligand (relative to the protein). The range in KD has been estimated to be from 10−8 to 10−3 M, assuming a diffusion-limited on-rate constant (ca. 108/s/M). The intrinsic sensitivity of the STD experiment is limited by the efficiency of the signal amplification and the magnetization transfer. The signal amplification depends on the kinetics of the binding process, especially on the off-rate. For KD < 10−8 M, small off-rates cause a low turnover of ligands into saturated ligands; the binding is so tight that saturation transfer from the bound to the free ligand molecules is very inefficient. Additionally, when binding is very weak, the population of the ligand–protein complex is so low that it leads to either weak STD signals or no signals at all. Despite being extremely utile and versatile, STD suffers from certain limitations. Among these is that the large excess of ligand relative to the protein may promote nonspecific binding (7) once the specific binding site has been saturated. Another limitation is that protein saturation is suboptimal in the case of low proton density, local proton deficiency, or molecular motion which compromises the intramolecular 1H–1H dipole interaction network. In such cases, WaterLOGSY may be a more effective experiment than STD (31). The main difference between STD and WaterLOGSY is the way in which the system receives magnetic saturation. Whereas STD NMR uses direct saturation of the protein, WaterLOGSY applies indirect saturation of the protein, namely, by selective saturation of the bulk water protons (H2O). Therefore, the transfer magnetization flows from water to protein to ligand. Technically, there are several options to achieve the selective bulk water saturation. Dalvit et al. use the selective inversion of the water resonance via the e-PHOGSY scheme (32). The transfer of magnetization from the water to the protein-bound ligand occurs via labile receptor protons (NH and OH protein protons) situated in the ligand-binding site as well as via remote labile protons in the protein, through spin diffusion. Additionally, direct proton– proton cross-relaxation between the bound ligand and long-lived water molecules within the binding pocket is an effective pathway in the magnetization transfer process. Differential cross-relaxation properties of binding and nonbinding molecules with water allow distinguishing between binding and nonbinding ligands. Whereas binding molecules interact with the proton spins of inverted water

14

NMR Studies of Protein–Ligand Interactions

245

via dipolar interactions, which lead to negative cross-relaxation rates, nonbinding molecules yield positive cross-relaxation rates. The result is that signals of nonbinding molecules show opposite sign to, and are usually weaker than, the resonances of binding ligands. 1.3. Ligand Versus Protein Observed Experiments

Ligand-based and protein-based approaches have distinct advantages and disadvantages. The former yield information about the strength of the interaction, the binding epitope, and the conformation of the ligand. They can be used to simultaneously screen several compounds for their ability to bind to a protein of interest. Ligandbased experiments have simple requirements. First, they do not require isotopically labeled protein. Secondly, there is no upper limit for protein size, but the difference between protein and ligand has to be substantial enough to result in differential relaxation behaviors. Contrariwise, protein observation experiments are currently only feasible for proteins weighing ca. 40 kDa or less. Moreover, they demand considerable amounts of isotopically labeled protein, which must be stable at high concentrations for long periods of time. When signal assignment is available, proteinbased experiments may provide more information. Most importantly, they can be used to identify one or several binding sites of the ligand on the protein and indicate zones of structural rearrangement. Furthermore, in these experiments, formation of ligand aggregates cannot be misinterpreted as an interaction. However, the data in protein observation experiments are easiest to interpret when the binding site has been saturated. In the case of low affinity ligands, this point may be beyond the limit of solubility. In conclusion, there are cases for which one approach is better suited than the other. Nevertheless, the full power of NMR to characterize protein–ligand interactions can only be exploited if both approaches are combined. Several examples from academic and industrial drug discovery projects are testament to the great success of a combined approach (7, 8, 33, 34).

2. Materials 2.1. Protein-Based Study on the Binding of VEGF to the Peptidic Ligand P-7i

1. Purified samples of uniformly 15N-labeled vascular endothelial growth factor (VEGF): 160 μ L at 100 μ M in 25 mM phosphate buffer, pH 7.0 (see Note 3), 50 mM NaCl, 90% H2O, 10% D2O in a 3-mm NMR tube (see Note 4). VEGF is obtained by recombinant expression as previously described (35). 2. Peptide P-7i: Prepared by using standard solid phase peptide synthesis (3). 3. Bruker Digital Avance 600 MHz spectrometer equipped with a cryoprobe (see Note 5).

246

M. Goldflam et al.

4. Data processing and analysis programs: TopSpin (36), Cara (37), and Origin (38), and the results are visualized using the program MOE (39). 2.2. Ligand-Based Study on Binding of POP to the Ligand Baicalin

1. Prolyl oligopeptidase (POP):160 μL of 100 μM POP in 20 mM phosphate buffer, pH 7.0 (see Note 3) in 100% D2O in a 3-mm NMR tube (see Note 4). 2. Baicalin: 160 μL of 500 μM baicalin in 20 mM phosphate buffer, pH 7.0 in 100% D2O in a 3-mm NMR tube. 3. 160 μL of POP (10 μM) and baicalin (500 μM) in 20 mM phosphate buffer, pH 7.0 in 100% D2O in a 3-mm NMR tube. 4. 160 μL of POP (20 μM) and baicalin (180 μM) in 20 mM phosphate buffer, pH 7.0 in 100% D2O in a 3-mm NMR tube. 5. Bruker Digital Avance 600 MHz NMR spectrometer equipped with a cryoprobe (see Note 5). 6. Data processing and analysis programs, TopSpin and Origin.

3. Methods 3.1. Protein-Based Study on the Binding of VEGF to the Peptidic Ligand P-7i

The following case study describes NMR monitoring of the interaction between VEGF and the peptidic ligand P-7i, presented here as a representative example of a CSP experiment (3). The 23-kDa VEGF11–109 construct used is a truncated version of VEGF121 which exhibits excellent solubility and stability. Moreover, it is readily labeled with 15N and is a symmetric homodimer, making it highly suited to protein-based NMR experiments. Additionally, the signal assignment for this construct is available (35). P-7i is a 19 amino acid-long analog of v107 that was discovered by phage display (40). It differs from v107 by a single mutation: the Ile-7 is D rather than L, which translates to a reduced affinity for VEGF (252 μM) compared to the wild type (1.0 μM). Although P-7i is relatively large (MW = ca. 2.3 kDa) for NMR studies of protein–ligand binding, it was selected as an example because it exhibits fast and intermediate exchange behavior and is amenable to CSP studies.

3.1.1. Sample Preparation

1. Starting from a stock solution of P-7i in water, prepare six aliquots (two at each of three concentrations) with a total amount of ligand corresponding to 50, 100, or 200 μM in a volume of 160 μL (19, 37, or 75 μg, respectively). Freeze the aliquots in 1.5-mL Eppendorf tubes and lyophilize (Fig. 3).

14 3.1.2. Recording of CSP NMR Spectra

NMR Studies of Protein–Ligand Interactions

247

1. Equilibrate the VEGF sample inside the NMR spectrometer for 15 min at 318 K before starting the NMR spectra acquisition. This temperature is chosen because the available signal assignment was performed at this temperature (35). Execute standard NMR procedures: tune the probe, shim the magnetic field, calibrate the length of the 90° pulses for 1H and 15N, and optimize the water suppression. 2. Record a 1H-15N HSQC spectrum using the FAST-HSQC experiment (41), which uses pulsed field gradients and a WATERGATE (42) module to efficiently suppress the water signal. For this, 2,048 × 256 complex points with a total of 8 transients per increment are used. The total experiment time is ca. 40 min. 3. Once the spectrum is obtained, remove the sample from the spectrometer and transfer it to the first Eppendorf tube containing the lyophilized ligand (see Fig. 3). Add 0.5% of DMSO-d6 to ensure ligand solubility (see Note 6), vortex the sample and centrifuge (5,000 RCF, 1 min, RT), and transfer all of the liquid to the previously used NMR tube. Introduce the resulting sample, containing the protein and the ligand at the first titration concentration, into the spectrometer. Equilibrate the sample at 318 K for 15 min, re-shim the magnetic field and record a new spectrum using the same conditions described in step 2. 4. Repeat step 3 until spectra are recorded for each titration point. By dissolving the lyophilized ligand in the sample incrementally, the concentration of P-7i increases stepwise over the course of the titration from 0 to 50 μM, 100, 200 , 300, 500, and finally, 700 μM (Fig. 3).

Fig. 3. Stepwise addition of P-7i to VEGF: overview of lyophilized ligand aliquots, the respective concentration increments and the total concentration over the course of titration. After the reference experiment on the sample containing only protein, the ligand concentration was increased stepwise by transferring the protein sample to Eppendorf tubes with the denoted amounts of lyophilized P-7i prior to acquisition of the next NMR spectrum. A total of seven spectra were acquired by repeating this procedure: one reference spectrum, plus one spectrum at each of the six titration points.

248

M. Goldflam et al.

3.1.3. Data Analysis

1. To process the HSQC spectra, increase the number of points in the indirect dimension (F1) from 256 to 512 by linear prediction and then zero fill to 1,024 points to yield a 2,048 × 1,024 matrix. Adjust the phase correction manually and apply a squared sine weighting function in both dimensions. Process all the spectra acquired in the titration experiment identically using Topspin 2.0. Figure 4 shows the seven superimposed spectra acquired from the titration of VEGF with P-7i. 2. Determination of CSP requires peak picking and subsequent assignment of the peaks to the corresponding residues. Use the program Cara to do this for the first and last spectra of the titration (see Note 7). 3. Extract the relevant data for mapping the binding site by calculating the distance between the position of the reference

Fig. 4. (a) Seven superimposed 1H-15N HSQC spectra of a 100-μM sample of uniformly-15N labeled VEGF11–109 at 318 K titrated with 0–700 μM of P-7i (600 MHz with cryoprobe). (b) Zoom of Lys48 shifts (reproduced from ref. 3 with kind permission from Wiley).

14

NMR Studies of Protein–Ligand Interactions

249

peak in the spectrum of the protein without ligand and the peak position in the spectrum of the highest ligand concentration. Calculate the distance between two peaks as the difference between the average chemical shift Δd NH for each peak computed from proton and nitrogen chemical shits (d H and d N, respectively), according to the following formula (see Note 8): ⎛ 2 ⎛ Δδ N ⎞ 2 ⎞ Δδ NH = ⎜ Δδ H +⎜ ⎝ 5 ⎠⎟ ⎟⎠ ⎝

(4)

The majority of the peaks exhibit fast exchange behavior; however, residues 17, 21, 26, 64, and 104 exhibit intermediate exchange. Signal broadening and a rapid decrease in signal intensity are observed for these residues; however, despite the broad signals, assignment is still possible. Residue 21, located in the ligand-binding zone, exhibits slow exchange behavior. The signal disappears completely after the second titration point. CSP is not feasible for this behavior (see Note 9). To identify the binding site, consider as significant only changes greater than the sum of the mean shift and the standard derivation. Figure 5 shows the calculated changes for each residue.

Fig. 5. Histogram of the P-7i induced CSP of the backbone amides Δd for every residue of VEGF11–109 observed in the 1H-15N HSQC experiment. The histogram was calculated for the shifts between the reference spectrum (VEGF alone) and the spectrum of the sample with the highest ligand concentration (VEGF plus 700 μM of P-7i). The lower (dashed ) horizontal line represents the mean shift, and the upper (dotted ) horizontal line, the cutoff for significant changes (mean shift plus one standard derivation).

250

M. Goldflam et al.

Fig. 6. Surface representation of the homodimer VEGF11–109 (PDB:2VPF). Residues encoded in red exhibit significant CSP, thereby indicating the binding zone for the ligand P-7i (residues encoded in black show no observable signals).

The residues with significant changes (10 of 83) are mapped to the 3D-structure of VEGF and depicted in red (Fig. 6) using the software MOE. 4. Calculate the binding affinity by plotting ΔdNH against the ligand concentration, which requires that ΔdNH is calculated for each of the seven titration points. Although a KD can be calculated for each residue, this exercise is only performed for some of the residues with a strong shift: those that exhibit the best signal-to-noise ratios and are free from signal overlap. Thus, a KD is calculated for residues 25, 48, 50, and 66, assuming a model of two independent identical binding sites (3):

[L ]− [L] [P ] K + [L ]+ 2 [P ]− (K + [L ]) + 4 [P ](K − [L ]+ [P ]) =F× 2 [P ]

Δδ NH = F ×

0

0

2

D

0

0

D

0

0

D

0

(5)

0

0

whereby [P0] is the total protein concentration; [L0], the total ligand concentration; [L], the concentration of unbound ligand in solution; F, a scaling factor; and KD, the affinity to be calculated. The calculated average KD for the VEGF – P-7i system is 252 μM (Fig. 7). The fact that the R2 values are all greater than 0.99 validates this model of the system. 3.2. Ligand-Based Study on Binding of POP to the Ligand Baicalin

The following case study was chosen to give a step-by-step explanation of an STD experiment. The example is based on work performed in the authors’ laboratory to characterize the interaction between the protease POP and the flavonoid baicalin (43).

14

NMR Studies of Protein–Ligand Interactions

251

Fig. 7. (a) Relative amount of bound ligand plotted against the total ligand concentration and fitted to a model of two independent identical binding sites. Analysis is based on Lys48 peak displacement in the spectra of the VEGF-(P-7i) complex. (b) Results of fitting for the four selected residues (reproduced from ref. 3 with kind permission from Wiley).

These experiments were designed to confirm baicalin as a ligand of POP and to obtain structure–activity information on POP–baicalin binding, especially on the influence of the baicalin sugar moiety. This entailed recording of a saturation buildup curve. Based on its molecular weight (80 kDa), POP is not appropriate for simple protein-based experiments; however, it is well suited for ligand-based experiments. Likewise, baicalin (MW 446 Da), as a relatively small to medium-sized ligand, is ideal for ligand-based experiments. Concerning its affinity properties, baicalin is a weak ligand of POP, having an IC50 value of 12 μM (43) (see Note 10). 3.2.1. Optimization of STD Parameters and Confirmation of Baicalin as a POP Ligand

1. Equilibrate the sample containing only POP in the NMR spectrometer to 308 K (see Note 11) for 15 min. Follow the standard procedure: tune the probe, shim the magnetic field, calibrate the length of the 90° pulse, and optimize the water suppression (see Note 12). 2. To optimize the protein saturation, record a 1H-spectrum of POP to identify promising signals in the aliphatic region. The aliphatic region of POP shows a signal at ca. 0.9 ppm that decreases in intensity in the up-field direction without occurrence of any new maxima. Therefore, the closer to this value the protein is irradiated, the greater the saturation (see Note 13). 3. To determine if protein irradiation is affecting any of the ligand’s signals, acquire several STD spectra of a sample of baicalin alone, using values of 0, −1, and −2 ppm for the on-resonance frequency. No STD signals are observed in any case. This negative control experiment confirms that no direct irradiation of baicalin or baicalin-aggregates occurs.

252

M. Goldflam et al.

For the subsequent STD experiments, select an on-resonance frequency of 0 ppm and an off-resonance frequency of 80 ppm (see Note 14). 4. Optimize the total saturation time with a sample containing POP (10 μM) and baicalin (500 μM) (see Note 15). Acquire STD spectra with different tSat values (from 1 to 4 s) using the same total experimental time. Superimpose all on-resonance spectra and select the spectrum with the best signal-to-noise ratio as the optimum one (in this case, 2 s). 5. Optimize the spin lock filter by testing different mixing times (from 20 to 70 ms). Under optimal conditions the protein signals are completely suppressed, thereby reducing the background noise in the STD experiment, whereas ligand signals are not affected. In this case, this is achieved with a spin lock length of 30 ms. 6. Acquire the final STD spectrum with the aforementioned optimized conditions and 2 k scans. Process the NMR data by multiplication with an exponential line-broadening function of 0.5 Hz prior to Fourier transformation. The resulting spectrum exhibits clear STD signals (Fig. 8) for baicalin, thereby confirming that it binds to POP.

Fig. 8. (a) 1H STD spectrum of baicalin (500 μM) in the presence of POP (10 μM) recorded at 600 MHz and 308 K. The protein signals were suppressed by applying a spin lock filter. (b) 1H-reference spectrum of baicalin. *Signals arising from sample impurities (reproduced from ref. 43 with kind permission from Elsevier).

14

NMR Studies of Protein–Ligand Interactions

253

Fig. 9. (a) Saturation transfer difference amplification for individual protons at different saturation times. The data were acquired with a ninefold excess of baicalin over POP (20 μM). (b) Build-up curves obtained for individual protons in baicalin. The key to the nomenclature for the individual protons is provided in Fig. 10 (adjusted from ref. 43 with kind permission from Elsevier).

3.2.2. Identifying the Binding Epitope on Baicalin

A sample containing 20 μM POP and 180 μM baicalin is used to identify the binding epitope (see Note 16). Experiments are performed at 308 K. 1. Obtain data for the saturation buildup curve using the previously optimized parameters (on-resonance irradiation: 0 ppm; 1 k scans; 308 K). Record a total of four experiments, using tSat values of 0.5, 1.0, 1.5, and 2.0 s. 2. For each proton, determine the values for STDAmpl. for the different saturation times (Fig. 9). Superimpose the STD and offresonance spectra for each saturation time, measure the difference between the STD and off-resonance signal for a specific proton (as a percentage), and multiply this value by the ligand excess (Eq. 2), which was equal to 9 (see Note 17). 3. Determine the binding epitope by fitting the values for STDAmpl. against the saturation times for each proton, according to Eq. 1. Calculate the initial slope v0 by multiplying STDmax by kSat. The saturation build-up curves of the protons of baicalin shows different initial slopes. The values were adjusted by setting the proton with the highest initial slope to 100% (Fig. 10). These data indicate that the protons in the γ-chromenone and the phenyl ring are in close contact with the protein, whereas those in the saccharide moiety contribute less to binding.

254

M. Goldflam et al.

Fig. 10. The STD data were fitted to a monoexponential equation, from which the STDmax and the saturation rate constant (ksat) were obtained. The initial slope directly correlates to the proximity of the corresponding proton to the protein and is the product of STDmax and ksat. The relative STDs were calculated by setting the proton with the greatest STD effect to 100% (adjusted from ref. 43 with kind permission from Elsevier).

4. Notes 1. Exchange between the free (L) and bound (PL) states of a ligand is considered in this context, assuming that the binding follows a bimolecular association reaction with second-order kon kinetics: P + L PL ; K = [P][L] = koff . D koff [PL] kon 2. In transient NOE experiments, a nonequilibrium state is generated via high-frequency pulses, and in a subsequent mixing period, returns to equilibrium by relaxation. 3. A buffer of the desired pH is created by mixing 25 mM solutions of NaH2PO4 and Na2HPO4 in deuterated water until the final pH is reached. 4. The most widely used NMR tubes are 3- and 5-mm standard tubes and 5-mm Shigemi tubes, which require sample volumes of 160, 600–700, and 300 μL, respectively. Voehler et al. analyzed the influence of tube type on the sensitivity of HSQC experiments (44). In general, 5-mm tubes are recommended for cases of abundant, poorly soluble protein; 3-mm standard or 5-mm Shigemi tubes when the sample is limited; and 3-mm standard tubes for high salt concentrations. Moreover, for titration experiments 3-mm tubes are easier to manipulate than

14

NMR Studies of Protein–Ligand Interactions

255

Shigemi tubes. Higher protein concentrations are generally preferable, as they enable shorter acquisition times. One exception to this is the case of a low affinity, poorly soluble ligand, for which a major excess of ligand must be employed to saturate the binding site, which in turn is only possible at lower protein concentrations. 5. In a cryoprobe the coil and the preamplifier are cooled to decrease thermal noise. The signal-to-noise ratio can be increased by three- to fourfold by using a cryoprobe instead of a conventional probe operating at room temperature. The authors have obtained a signal-to-noise ratio of ca. 7,000:1 for 1 H using a cryoprobe and a standard sample of 0.1% ethylbenzene in CDCl3. 6. Ligands soluble at high concentrations (ca. 100 mM) in a vehicle such as DMSO-d6 can usually be added directly to the sample, without resulting in any significant dilution; however, if this does not apply, then lyophilized aliquots of ligand can be used, as in Case Study 1. Vehicle may be added simply to ensure the solubility of the ligand; however, then a control experiment is necessary to evaluate any changes induced by it. If lyophilized aliquots of ligand are used, then a vehicle can also be added to the reference experiment (i.e., protein alone) to eliminate any vehicle-induced changes during the course of the titration. Finally, data quality may be improved if the ligand concentration in the sample is controlled. This is easily achieved by adding a reference compound at a known concentration to the sample, such as trimethylsilyl propionate-d4 (TSP), and then comparing the signal intensities of the ligand and of the reference in a 1H-spectrum. 7. The authors found the freeware program Cara (available at http://www.nmr.ch) to be utile and straightforward; nevertheless, other assignment tools (e.g., NMRview and Sparky) are equally suitable. 8. This formula is one of the most commonly used ones for calculating the distance between two peaks in the twodimensional plane. Other approaches and their impact on CSP mapping have been reviewed by Schumann et al. (45). 9. For the wild-type peptide v107, the binding site could not be mapped by observing CSP. This is because the binding kinetics are in the slow exchange regime. All residues that are directly involved in ligand binding or are in close proximity do not shift, but simply decrease in intensity as a new signal appears. In this case an alternative mapping of the binding site for slow exchanging ligands was feasible based on the changes of signal intensities induced by the binding of the ligand. 10. As mentioned in Subheading 1.2, STD NMR enables calculation of KD. However, since the IC50 of baicalin for POP had

256

M. Goldflam et al.

previously been determined, this calculation was not performed. Information on the calculation of binding strengths can be found in references (5) and (30). 11. Temperature is another parameter that can be optimized in an STD experiment, as it affects the binding kinetics of the system, the efficiency of protein saturation, and the relaxation rates. In STD, signal intensities can be increased by changing the temperature (27); however, once temperature has been optimized, all other parameters must also be adjusted. 12. Before performing the STD experiments, a 1D 1H spectrum with a water suppression module was employed to optimize water suppression. The residual H2O signal was suppressed with an excitation sculpting block (46), which uses a double gradient spin-echo to defocus the H2O resonance. Squared pulses of 2 ms length were used as 180° pulses. Optimized parameters for water suppression in this experiment were then used for the STD experiments. 13. This sample can also be used to verify if the saturation is well spread over the whole protein. This requires that an STD spectrum without spin lock filter is acquired. A decrease in intensity of all protein signals indicates a good distribution, which is the case for most proteins. 14. Some proteins exhibit several up-field shifted signals (ca. 1 to -1 ppm) that differ in intensities. In these cases, the onresonance frequency can be optimized by recording a set of STD spectra of a sample containing only protein, without the spinlock filter but using different saturation frequencies. The saturation frequency that provides the strongest decrease between on- and off-resonance spectra is deemed the most favorable (upon confirmation that no direct saturation of ligand signals occurs). Commonly used values for the on-resonance irradiation lie between 0 and −1 ppm, and for the off-resonance, either 40 or 80 ppm. 15. The range of ligand excess in STD NMR can be very wide. A good starting point is to use a 50-fold molar excess of ligand at a protein concentration of 10 μM. However, the optimal ratio between protein and ligand depends on the system’s kinetics. The faster the exchange, the better the signal can be amplified which makes higher protein-to-ligand ratios (100-fold and higher) useful and will ultimately yield stronger STD effects. Independent of the choice of the ratio, the ligand concentration should always be in a range in which aggregation can be excluded. 16. To map a binding epitope, lower ligand-to-protein ratios are typically used. The objective is not to reach the maximal STD amplification, but rather to cover the range of STDAmpl. values

14

NMR Studies of Protein–Ligand Interactions

257

in order to calculate the build-up curve. A smaller excess of ligand leads to more pronounced differences between ISTD and I0 for the different saturations times than does a larger excess. 17. In this example the STD amplification factor STDAmpl. was used, rather than η, the fraction of ISTD from I0. The difference between the two is that STDAmpl. accounts for the ligand excess (Eq. 2). Since in this example the ligand excess is equal for all saturation times, then it turns into a scaling factor between STDAmpl. and η, which is not required for calculating the binding epitope. STDAmpl. is more commonly used because it enables calculation of KD. References 1. Wiseman, T., Williston, S., Brandts, J. F., and Lin, L. N. (1989) Rapid measurement of binding constants and heats of binding using a new titration calorimeter. Anal. Biochem. 179, 131–137. 2. Baldwin, M. A. (2005) Mass spectrometers for the analysis of biomolecules. Methods Enzymol. 402, 3–48. 3. Dyachenko, A., Goldflam, M., Vilaseca, M., and Giralt, E. (2010) Molecular recognition at protein surface in solution and gas phase: Five VEGF peptidic ligands show inverse affinity when studied by NMR and CID-MS. Biopolymers 94, 689–700. 4. Englebienne, P., Hoonacker, A. V., and Verhas, M. (2003) Surface plasmon resonance: principles, methods and applications in biomedical sciences. Spectroscopy 17, 255–273. 5. Fielding, L. (2003) NMR methods for the determination of protein-ligand dissociation constants. Curr. Top. Med. Chem. 3, 39–53. 6. Carlomagno, T. (2005) Ligand-target interactions: what can we learn from NMR? Annu. Rev. Biophys. Biomo.l Struct. 34, 245–266. 7. Lepre, C. A., Moore, J. M., and Peng, J. W. (2004) Theory and applications of NMR-based screening in pharmaceutical research. Chem. Rev. 104, 3641–3676. 8. Dalvit, C. (2009) NMR methods in fragment screening: theory and a comparison with other biophysical techniques. Drug Discov. Today 14, 1051–1057. 9. Tarrago, T., Claasen, B., Kichik, N., RodriguezMias, R. A., Gairi, M., and Giralt, E. (2009) A cost-effective labeling strategy for the NMR study of large proteins: selective 15 N-labeling of the tryptophan side chains of prolyl oligopeptidase. Chembiochem. 10, 2736–2739. 10. Bogan, A. A., and Thorn, K. S. (1998) Anatomy of hot spots in protein interfaces. J. Mol. Biol. 280, 1–9.

11. Tugarinov, V., and Kay, L. E. (2005) Methyl groups as probes of structure and dynamics in NMR studies of high-molecular-weight proteins. Chembiochem. 6, 1567–1577. 12. Bodenhausen, G., and Ruben, D. J. (1980) Natural abundance nitrogen-15 NMR by enhanced heteronuclear spectroscopy. Chemical Physics Letters 69, 185–189. 13. Kay, L., Keifer, P., and Saarinen, T. (1992) Pure absorption gradient enhanced heteronuclear single quantum correlation spectroscopy with improved sensitivity. J. Am. Chem. Soc. 114, 10663–10665. 14. Morris, G. A., and Freeman, R. (1979) Enhancement of nuclear magnetic resonance signals by polarization transfer. J. Am. Chem. Soc. 101, 760–762. 15. Gang, Z., and Price, W. S. Solvent signal suppression in NMR. Prog. Nucl. Magn. Reson. Spectrosc. 56, 267–288. 16. Shuker, S. B., Hajduk, P. J., Meadows, R. P., and Fesik, S. W. (1996) Discovering HighAffinity Ligands for Proteins: SAR by NMR. Science 274, 1531–1534. 17. Smrcka, A. V., Kichik, N., Tarrago, T., Burroughs, M., Park, M. S., Itoga, N. K., Stern, H. A., Willardson, B. M., and Giralt, E. (2010) NMR analysis of G-protein betagamma subunit complexes reveals a dynamic G(alpha)Gbetagamma subunit interface and multiple protein recognition modes. Proc. Natl. Acad. Sci. USA 107, 639–644. 18. Pellecchia, M. (2005) Solution nuclear magnetic resonance spectroscopy techniques for probing intermolecular interactions. Chem. Biol. 12, 961–971. 19. Reibarkh, M., Malia, T. J., and Wagner, G. (2006) NMR distinction of single- and multiple-mode binding of small-molecule protein ligands. J. Am. Chem. Soc. 128, 2160–2161.

258

M. Goldflam et al.

20. Medek, A., Hajduk, P., Mack, J., and Fesik, S. (2000) The Use of Differential Chemical Shifts for Determining the Binding Site Location and Orientation of Protein-Bound Ligands. J. Am. Chem. Soc. 122, 1241–1242. 21. Krishnamoorthy, J., Yu, V. C., and Mok, Y. K. (2010) Auto-FACE: an NMR based binding site mapping program for fast chemical exchange protein-ligand systems. PLoS One 5, e8943. 22. Neuhaus, D., and Williamson, M. P. (2000) The Nuclear Overhauser Effect in Structural and Conformational Analysis, 2nd Edition ed., Wiley, New York. 23. Meiboom, S., and Gill, D. (1958) Modified Spin-Echo Method for Measuring Nuclear Relaxation Times, Review of Scientific Instruments 29, 688–691. 24. Ni, F., and Scheraga, H. A. (1994) Use of the Transferred Nuclear Overhauser Effect To Determine the Conformations of Ligands Bound to Proteins. Accts. Chem. Res. 27, 257–264. 25. Moriz, M., and Bernd, M. (1999) Characterization of Ligand Binding by Saturation Transfer Difference NMR Spectroscopy. Angew. Chem. Int. Ed. Engl. 38, 1784–1788. 26. Klein, J., Meinecke, R., Mayer, M., and Meyer, B. (1999) Detecting Binding Affinity to Immobilized Receptor Proteins in Compound Libraries by HR-MAS STD NMR. J. Am. Chem. Soc. 121, 5336–5337. 27. Groves, P., Kover, K. E., Andre, S., Bandorowicz-Pikula, J., Batta, G., Bruix, M., Buchet, R., Canales, A., Canada, F. J., Gabius, H. J., Laurents, D. V., Naranjo, J. R., Palczewska, M., Pikula, S., Rial, E., StrzeleckaKiliszek, A., and Jimenez-Barbero, J. (2007) Temperature dependence of ligand-protein complex formation as reflected by saturation transfer difference NMR experiments. Magn. Reson. Chem. 45, 745–748. 28. Mayer, M., and Meyer, B. (2001) Group epitope mapping by saturation transfer difference NMR to identify segments of a ligand in direct contact with a protein receptor. J. Am. Chem. Soc. 123, 6108–6117. 29. Mayer, M., and James, T. L. (2004) NMRbased characterization of phenothiazines as a RNA binding scaffold. J. Am. Chem. Soc. 126, 4453–4460. 30. Meyer, B., and Peters, T. (2003) NMR spectroscopy techniques for screening and identifying ligand binding to protein receptors. Angew. Chem. Int. Ed. Engl. 42, 864–890. 31. Dalvit, C., Pevarello, P., Tato, M., Veronesi, M., Vulpetti, A., and Sundstrom, M. (2000) Identification of compounds with binding

32.

33.

34.

35.

36.

37.

38. 39. 40.

41.

42.

affinity to proteins via magnetization transfer from bulk water. J. Biomol. NMR 18, 65–68. Dalvit, C., Fogliatto, G., Stewart, A., Veronesi, M., and Stockman, B. (2001) WaterLOGSY as a method for primary NMR screening: Practical aspects and range of applicability. J. Biomol. NMR 21, 349–359. Pellecchia, M., Sem, D. S., and Wuthrich, K. (2002) NMR in drug discovery. Nat. Rev. Drug Discov. 1, 211–219. Oltersdorf, T., Elmore, S. W., Shoemaker, A. R., Armstrong, R. C., Augeri, D. J., Belli, B. A., Bruncko, M., Deckwerth, T. L., Dinges, J., Hajduk, P. J., Joseph, M. K., Kitada, S., Korsmeyer, S. J., Kunzer, A. R., Letai, A., Li, C., Mitten, M. J., Nettesheim, D. G., Ng, S., Nimmer, P. M., O’Connor, J. M., Oleksijew, A., Petros, A. M., Reed, J. C., Shen, W., Tahir, S. K., Thompson, C. B., Tomaselli, K. J., Wang, B., Wendt, M. D., Zhang, H., Fesik, S. W., and Rosenberg, S. H. (2005) An inhibitor of Bcl-2 family proteins induces regression of solid tumours. Nature 435, 677–681. Fairbrother, W. J., Champe, M. A., Christinger, H. W., Keyt, B. A., and Starovasnik, M. A. (1997) 1H, 13C, and 15N backbone assignment and secondary structure of the receptorbinding domain of vascular endothelial growth factor. Protein Sci. 6, 2250–2260. Bruker Corporation, (2007) Topspin 2.0, http://www.bruker-biospin.com/nmr_software.html. Keller, R. (2004) The Computer Aided Resonance Assignment Tutorial, 1st edition ed., CANTINA Verlag. Origin Corporation, (2007) Origin 8.0, http://www.originlab.com/. Chemical computing group, (2009) http:// www.chemcomp.com/index.htm. Pan, B., Li, B., Russell, S. J., Tom, J. Y., Cochran, A. G., and Fairbrother, W. J. (2002) Solution structure of a phage-derived peptide antagonist in complex with vascular endothelial growth factor. J. Mol. Biol. 316, 769–787. Mori, S., Abeygunawardana, C., Johnson, M. O., and van Zijl, P. C. (1995) Improved sensitivity of HSQC spectra of exchanging protons at short interscan delays using a new fast HSQC (FHSQC) detection scheme that avoids water saturation. J. Magn. Reson. B 108, 94–98. Piotto, M., Saudek, V., and Sklenář, V. (1992) Gradient-tailored excitation for single-quantum NMR spectroscopy of aqueous solutions. J. Biomol. NMR 2, 661–665.

14

NMR Studies of Protein–Ligand Interactions

43. Tarrago, T., Kichik, N., Claasen, B., Prades, R., Teixido, M., and Giralt, E. (2008) Baicalin, a prodrug able to reach the CNS, is a prolyl oligopeptidase inhibitor. Bioorg. Med. Chem. 16, 7516–7524. 44. Voehler, M. W., Collier, G., Young, J. K., Stone, M. P., and Germann, M. W. (2006) Performance of cryogenic probes as a function of ionic strength and sample tube geometry. J. Magn. Reson. 183, 102–109.

259

45. Schumann, F. H., Riepl, H., Maurer, T., Gronwald, W., Neidig, K. P., and Kalbitzer, H. R. (2007) Combined chemical shift changes and amino acid specific chemical shift mapping of protein-protein interactions. J. Biomol. NMR 39, 275–289. 46. Hwang, T. L., and Shaka, A. J. (1995) Water suppression that works. Excitation sculpting using arbitrary waveforms and pulsed field gradients. J. Magn. Reson. 112, 275–279.

Chapter 15 In-Cell NMR Spectroscopy in Escherichia coli Kirsten E. Robinson, Patrick N. Reardon, and Leonard D. Spicer Abstract A living cell is a complex system that contains many biological macromolecules and small molecules necessary for survival, in a relatively small volume. It is within this crowded and complex cellular environment that proteins function making in-cell studies of protein structure and binding interactions an exciting and important area of study. Nuclear magnetic resonance (NMR) spectroscopy is a particularly attractive method for in-cell studies of proteins since it provides atomic-level data noninvasively in solution. In addition, NMR has recently undergone significant advances in instrumentation to increase sensitivity and in methods development to reduce data acquisition times for multidimensional experiments. Thus, NMR spectroscopy lends itself to studying proteins within a living cell, and recently “in-cell NMR” studies have been reported from several laboratories. To date, this technique has been successfully applied in Escherichia coli (E. coli), Xenopus laevis (X. laevis) oocytes, and HeLa host cells. Demonstrated applications include protein assignment as well as de novo 3D protein structure determination. The most common use, however, is to probe binding interactions and structural modifications directly from proton nitrogen correlation spectra. E. coli is the most extensively used cell type thus far and this chapter is largely confined to reviewing recent literature and describing methods and detailed protocols for in-cell NMR studies in this bacterial cell. Key words: In-cell NMR spectroscopy, Protein NMR spectroscopy, Escherichia coli, Fast NMR spectroscopy

1. Introduction 1.1. In-Cell NMR Spectroscopy: The Technique and Information It Reveals

Within a cell, there are many different proteins, other biological macromolecules, and small molecules that must function properly for a cell to survive. The simplest organisms are estimated to utilize a few hundred types of small molecules and genomically encode up to 1,000 different proteins (1, 2), with the human genome estimated to encode 10–100-fold more (3). The large number and diversity of small molecules and biological macromolecules form a cellular environment, where these molecules function, that is very crowded

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_15, © Springer Science+Business Media, LLC 2012

261

262

K.E. Robinson et al.

and complex (4). Understanding the influences that this packed environment has on individual protein structure, stability, and behavior as well as on protein complexes is one of the challenges of contemporary biomedical science (1, 5, 6). The knowledge that can be gained from complete molecular- and atomic-level observations in a live cellular environment is invaluable in understanding detailed protein mechanisms of action as well as in the areas of drug development and protein engineering and numerous other research frontiers. In-cell nuclear magnetic resonance (NMR) spectroscopy is currently the most comprehensive technique for examining proteins at the atomic level within a living cell (4). The ultimate goal is molecular structure determination. Historically, high-resolution protein structures have been determined only on purified proteins. The fact that NMR is a noninvasive solution technique suggests that NMR is uniquely suited to examining full atomic-level macromolecular structures in a representative environment, although not necessarily a natural environment (6). The strength of in-cell NMR spectroscopy not only lies in the determination of de novo 3-D structure, but also in the ability to observe structural changes of biological macromolecules in a native environment. Structural changes can be observed directly by monitoring changes that occur in a protein 2-D 1H–15N (or 1H–13C) heteronuclear single-quantum coherence (HSQC) fingerprint spectrum, since every protein generates a unique 1H–15N HSQC that is dependent upon the secondary, tertiary, and quaternary structure as well as the chemical environment in which the protein resides. Alterations in the structure of the target protein or its chemical environment give rise to changes in the 2-D fingerprint. This enables studies of molecular interactions within a living cell, such as mapping the binding interface between two proteins (7). 1.2. In-Cell NMR Spectroscopy: Current Achievements and Challenges

In-cell NMR spectroscopy within E. coli has been used to determine the 3-D backbone assignment of GB1 (8), determine the 3-D de novo structure of the gene product TTHA1718 (9, 10), as well as to study protein–protein interactions (7), protein–DNA interactions (11), binding events (12), and identify potential new drugs (13). These studies have all been performed using 15N and/or 13C isotopic labeling of proteins under 20 kDa. Both uniform labeling and selective methyl-group labeling strategies have been applied (14, 15). Labeling schemes typically use isotope 15N enrichment for 2-D fingerprint data collection and also incorporate uniform 13C labeling when collecting 3-D data for assignment and structure determination. The 1H–13C HSQC fingerprint can also be useful, but is often complicated by significant background signals that arise from metabolites within the cell incorporating the 13C isotope (15). However, it has been demonstrated that utilizing specific 13C methylgroup labeling can be beneficial for observing larger proteins that

15

In-Cell NMR Spectroscopy in Escherichia coli

263

have attenuated 1H–15N HSQC spectra since methyl groups most often have independent rotational motion and give better line shape and intensity (15–17). An alternative specific labeling strategy is the incorporation of 19F as an NMR probe, which can enhance sensitivity (18), but typically does not provide sufficient data for structural characterization. In-cell NMR spectroscopy enables protein studies within a living cellular environment with some limitations. Most success has been achieved for proteins under 12 kDa; the largest protein observed in-cell with clear spectral features is calmodulin at 16.8 kDa. Calmodulin was observed using specific labeling techniques to incorporate either 13C-labeled methyl groups or 15N-labeled lysines (15, 19). Protein that aggregates or interacts with the cellular membrane, large cellular complexes, or DNA may have significantly longer rotational correlation times due to slower molecular motions and/or may be involved in an intermediate exchange regime. This can lead to attenuated NMR signals due to line broadening, thus making the protein quite difficult to detect and often “invisible” by NMR within a living cell. This was found to be the case with the MetJ repressor protein (11), which binds nonselectively to DNA. Care must be taken to demonstrate the source of signal attenuation by incorporating appropriate controls, as well as alternate analytical methods and when possible rescue experiments (11). To characterize proteins by NMR within a living E. coli cell, it is necessary to isotopically label the protein with 15N and/or 13C. When the protein is expressed within the cell type of interest (see Note 1), it is essential to keep in mind that the isotopic labels are incorporated throughout the cell, into the protein of interest and into the native proteins and metabolites. The presence of isotopes in the native biomolecules and metabolites results in background signals: signals that are not associated with the protein of interest (Fig. 1 ). Serber et al. used rifampicin to suppress production of bacterial protein; however, the detected background signal remained (19). The background signals that are being detected are, therefore, thought to belong to small metabolites, such as amino acids, that are incorporating the stable isotope labels during growth (19). Growing the cells to log phase in unlabeled medium and then switching to isotopically labeled medium have also been studied (19). While this approach did not reduce the intensity of the observable background signal, this method can be used for cost and time savings. An E. coli in-cell preparation for NMR is composed typically of a 20% cell slurry. In this slurry, the cells settle over time in an NMR tube without some form of mixing. It takes a few hours for a 20% cell slurry to start layering at the bottom of the tube and this may or may not be confined within the active volume of the probe coil (see Note 2). This cell settling problem is seemingly easily overcome by packing the cells close together, and there are in-cell NMR

264

K.E. Robinson et al. 110.0

115.0

N (ppm)

120.0

125.0

130.0

135.0

9.50

9.00

8.50

8.00

7.50

7.00

NH (ppm)

Fig. 1. 1H–15N HSQC spectrum from E. coli cells grown in isotopically labeled medium without protein overexpression. The signals detected here are referred to as the background signal for in-cell NMR spectroscopy.

reports using a denser sample. Careful studies of the effect of cell density on spectral quality, however, demonstrate that a 20–30% cell slurry yields the best signal (19) and that higher cell densities result in line broadening. The source of this spectral degradation has not been thoroughly studied, but contributing factors are likely bulk magnetic susceptibility and inhomogeneity of the sample. Another strategy to prevent cell settling that has been employed is to encapsulate E. coli cells in alginate microcapsules (20). This method involves preparing a mixture of warm alginate with E. coli cells and then using an electric current to force the mixture through a needle into a calcium bath to cause polymerization. This technique, however, has not been commonly used. Other options to prevent cell settling beyond the active volume of the coil may include using Shigemi tubes or susceptibility plugs, but these have not been explored. As mentioned above, another important concern when performing in-cell NMR experiments is that cells remain viable during the course of the NMR experiment. Cells that die during the experiment can lyse and release their contents into the supernatant, thus leading to the protein of interest being found in the medium and not in the concentrated cellular environment. This leads to spurious protein signals. The NMR tube is a harsh environment, limited in nutrients and oxygen, which causes eventual cell death. It is necessary to determine the viable cell count in the NMR sample before and after performing the NMR experiment. Cell viability

15

In-Cell NMR Spectroscopy in Escherichia coli

265

must be determined by a quantitative assay rather than a qualitative assay as having only a small percentage of viable cells is sufficient for a positive result by a qualitative assay. The quantitative assay we use is a serial dilution plating experiment which is described in Subheading 3.2 (see Note 3). A related problem that must be avoided and diligently tested for is leakage of the protein of interest out of intact cells. It is always necessary after performing an in-cell NMR experiment to harvest the cells from the sample and examine the supernatant for the presence of the protein of interest to ensure that the spectral signals recorded do not originate partially or fully from outside the living cells. Investigators should be warned that even healthy cells frequently release protein into their surroundings and stressed cells expressing nonnative proteins are often more prone to this release. There is at least one reported strategy for overcoming NMR detection of protein outside the cell (8). In our own studies, we found that a small amount of the protein GB1, a domain of a protein from Group G Streptococcus, can escape from E. coli cells during NMR experiments however. By including IgG antibody in the cell slurry medium, the extracellular GB1 is efficiently scavenged by the large IgG molecule eliminating this GB1 signal from the supernatant. Protein leakage from E. coli is also reported to be correlated with the total fraction of overexpressed protein in the cell (21). In general, careful controls must be established to ensure that the supernatant gives no detectable NMR signal for the protein of interest, thus assuring that the detected protein is located within living cells. These considerations place significant time limitations on an in-cell NMR experiment. Cells settle in a few hours, unless settling is attenuated (20). Cell viability along with cell wall permeability affect the intracellular versus extracellular protein concentration, thus restricting data acquisition time. If protein leakage is observed, effective extracellular protein scavengers can be employed to extend the available data collection time (8). Even though these limitations are highly variable, they usually do not affect 2-D HSQC experiments since the data can be collected within the time frame of a few minutes. Therefore, 2-D HSQC has been the most common in-cell NMR experiment performed to date. HSQC spectra can be very useful when the in vitro assignments for the protein of interest are already known. The fingerprints obtained from the in-cell NMR experiment can often be compared to the in vitro spectrum to obtain the assignments if the structure does not change. So far, most in-cell NMR HSQC spectra have shown only slight differences when compared with the in vitro spectrum, so it is straightforward to transfer the assignments. This also indicates that the molecular crowding and corresponding limited protein-accessible free volume characteristic of the intracellular milieu are not significant determinants of the folded state.

266

K.E. Robinson et al.

With the HSQC spectrum assigned, it is straightforward to study how specific changes within the cellular environment affect the protein structure. Through examination of 2-D 1H–15N HSQC experiments, it is possible to study how drugs bind to the protein of interest within the cellular environment (12), and to identify potential new drugs (13). It is also possible to study protein–protein interactions (7) and to examine the effect that a posttranslational modification on one of the binding partners may have on the binding interface (22). Even the effect of nonselective protein–DNA interactions on the NMR spectrum has been studied by in-cell NMR spectroscopy, leading to new insights into repressor activity in transcription regulation (11). In-cell NMR spectra have also shown that some, although not all, intrinsically disordered proteins may gain structure within the cellular environment (23, 24). These studies, utilizing 2-D 1H–15N HSQC experiments, illustrate the diversity and strength of in-cell NMR spectroscopy to expand the knowledge of protein function within a living cell. While 2-D HSQC experiments are quite useful, if the assignments have not been previously determined in vitro or if the HSQC fingerprint is dramatically altered in-cell, then it becomes necessary to collect additional multidimensional data. As mentioned above, there are time limitations that can significantly affect the ability to collect multidimensional NMR data. The requirement for short acquisition times is important when performing 3-D in-cell NMR experiments; however, the type of experiment implemented should not compromise sensitivity or resolution of the detected signal unless specific, well-dispersed spectral features are targeted. These time limitations generally make it necessary to utilize fast NMR methodologies and ultrasensitive probes that are continuing to be developed and improved. There are several ways to collect fast NMR data (25–28); however, to date, only sparse sampling techniques have been applied to in-cell NMR spectroscopy. Most fast NMR methodologies require two aspects to be considered: the sampling pattern used and the processing method applied. The sampling patterns used for fast NMR are designed to reduce the number of points sampled to significantly decrease the time it takes to collect adequate data for a specific experiment when compared to the standard Cartesian sampling pattern. Sparse sampling patterns used for multidimensional NMR include a radial sampling pattern (28, 29), concentric ring sampling (30), and random sampling (31). Radial sampling is a special case of concentric ring sampling, where the same number of points is taken for each ring. Random sampling involves collecting data points that are distributed randomly. These patterns are different from the standard Cartesian sampling pattern as Cartesian sampling distributes the data points equally on a grid while these other sampling patterns either are not on a grid at all or, if positioned to be on a grid, only partially fill it.

15

In-Cell NMR Spectroscopy in Escherichia coli

267

The other aspect of performing fast NMR experiments is the ability to properly process the data. There are several processing methods available (31–38). The sampling pattern chosen can lead to artifacts being incorporated in the spectrum upon processing. It is well-known that radial sampling introduces artifacts. As a result, the field is moving away from radial sampling and toward concentric ring sampling. In any of the methods used, it is important to appreciate the potential for introducing artifacts in the processed data collected from fast acquisition experiments. Programs are available to aid with processing the data (see Note 4). Currently, there are two studies that have been performed using these fast NMR techniques. The first study used projection reconstruction NMR (PR-NMR) with radial sampling and the hybrid back-projection/lower value (HBLV) reconstruction algorithm to walk the backbone of GB1 (8). The backbone assignment of GB1 was accomplished by collecting PR-NMR versions of the 3-D HNCA, HNCO, and HA(CA)NH experiments. It should be noted that the HA(CA)NH experiment works only for very small proteins. The second study used random sampling followed by maximum entropy processing (9). The 3-D structure of TTHA1718 was determined using several 3-D heteronuclear NMR experiments and distance restraints obtained from NOE data. Fast NMR data collection is continuously growing and evolving. 1.3. In-Cell NMR Spectroscopy: Looking Ahead

The studies cited above illustrate that in-cell NMR utilizing E. coli enables the study of proteins within a physiologically relevant environment. However, the technique requires that the protein of interest is expressed to intracellular concentrations sufficient for NMR detection that are greater than the concentrations of most cellular proteins. Thus, overexpression generates a labeled protein in the cellular environment but is not strictly in vivo. Ideally, the protein concentration should be tightly controlled to understand what effects a high concentration may cause within a living cell as well as potentially lower the concentration while remaining within the detection limits of NMR. It has been demonstrated that protein expression levels for in-cell NMR can be controlled with tunable promoters, such as the arabinose promoter (7). External delivery of isotopically labeled protein can provide another strategy to control concentrations of labeled proteins in cells; however, to date, this has only been reported using Xenopus laevis oocytes or HeLa cells (39, 40) (Fig. 2). The overexpression protocol is a useful strategy because the protein of interest is never exposed to a noncellular environment; however, E. coli lack much of the complexity found within eukaryotic organisms as they are not capable of performing most posttranslational modifications and do not have organelles. Therefore, it is also of interest to develop complementary methods to overexpress proteins within a eukaryotic organism, such as yeast or insect cells, so that the protein of interest is always exposed to

268

K.E. Robinson et al. Target protein Cell-penetrating peptide Injection needle

Overexpression

Delivery

X. laevis Oocytes

E. coli

HeLa

Fig. 2. Methods for in-cell NMR spectroscopy. E. coli cells are used for overexpressing protein within the cell of interest. Proteins are difficult to overexpress in eukaroptic cells; therefore, currently, two protocols are used. For X. laevis oocytes, physical injection has been used while for HeLa cells cell-penetrating peptides have been used.

a natural eukaryotic environment, but this has yet to be achieved. This chapter discusses detailed planning and implementation of incell NMR spectroscopy studies in live E. coli using overexpression methods to introduce stable isotope labels into the target protein.

2. Materials 2.1. Protein Expression

1. E. coli BL21 DE3 cells containing the expression plasmid (see Note 1). 2. Luria-Broth (LB) Medium: 10 g NaCl, 10 g Bacto-Tryptone, 5 g yeast extract in 1 L dIH2O. Autoclave to sterilize and store at 25°C. 3. LB agar selection plates: LB medium with 20 g Bacto-Agar in 1 L dIH2O. Autoclave to sterilize. Add the desired antibiotic after the medium has cooled to 50°C. Store plates at 4°C. 4. 5× M9 Salts: 64 g Na2HPO4, 15 g KH2PO4, 2.5 g NaCl, 5.0 g NH4Cl in 1 L dIH2O. Autoclave to sterilize. To incorporate 15 N, use 15NH4Cl; final pH is 7.2. Store at 25°C. 5. 1,000× trace metal mix: 2 mM H3BO3, 2 mM CuSO4, 2 mM CoCl2, 10 mM MnCl2, 2 mM NiSO4, 2 mM (NH4)6Mo7O24. Filter sterilize and store at 25°C. 6. M9 minimal medium: 1× M9 salts, 2 mM MgSO4, 1 mM FeCl3, 25 mM ZnSO4, 100 mM CaCl2, 1× trace metal mix, 0.0005% thiamine, 0.3% (w/v) glucose (see Note 5). To incorporate 13 C, use 0.3% (w/v) 13C glucose.

15

In-Cell NMR Spectroscopy in Escherichia coli

269

7. Inducers (see Table 1). 8. Antibiotics (see Table 2). 9. 99.9% D2O. 1. Luria–Broth medium (Subheading 2.1).

2.2. Cell Viability Assay

2. LB agar selection plates (Subheading 2.1). 3. Antibiotics (Table 2). 1. 10× phosphate buffer: 42.3 g NaH2PO4, 27.46 g Na2HPO4 in 1 L dIH2O. Use NaOH or HCL to adjust pH to 7. Autoclave and store at 25°C.

2.3. Freeze–Thaw Lysis

2. 10× NaCl: 8.766 g NaCl in 50 mL dIH2O. Filter sterilize and store at 25°C. 3. 1,000× Phenylmethanesulfonylfluoride (PMSF) stock solution: 0.871 g PMSF in 100% ethanol. Store at 25°C. 4. 1,000× Deoxyribonuclease (DNase) stock solution: 20 mg/mL DNase (from lyophilized powder) in 10 mM Tris–HCl, pH 7.5, 50 mM NaCl, 10 mM MgCl2, 1 mM dithiothreitol (DTT), and 50% (w/v) glycerol. Store at −20°C. 5. Dry ice. 6. 100% Ethanol.

Table 1 Induction Agent Composition Inducer a

a

[Stock solution] [Working]

Storage

Isopropyl b-D-1-thioga- 1 M in H2O lactopyranoside (IPTG)

1 mM

Arabinose

0.02–0.2% (w/v) 25°C

20% (w/v)

−20°C

All inducers are filter sterilized

Table 2 Selection Agent Composition

a

Antibiotic a

[Stock solution]

[Working]

Ampicillin

100 mg/mL in H2O

100 mg/mL

Kanamycin

25 mg/mL in H2O

25 mg/mL

Tetracycline

12.5 mg/mL in 100% Ethanol

12.5 mg/mL

Streptomycin

50 mg/mL in H2O

50 mg/mL

All antibiotics are filter sterilized and stored at −20°C

270

K.E. Robinson et al.

7. 2× Loading dye: 40% (w/v) glycerol, 125 mM Tris–HCl, pH 6.8, 100 mM DTT, 2% (w/v) sodium dodecyl sulfate (SDS), 0.025% (w/v) bromophenol blue. 8. Freeze–thaw lysis buffer: 1× phosphate buffer, 1× NaCl, 1× PMSF, 1× DNase, and 1.5 mg/mL lysozyme (from lyophilized powder).

3. Methods 3.1. Protein Expression for Preparing an In-Cell NMR Sample

The typical strategy for studying a protein using in-cell NMR spectroscopy entails incorporating 15N uniform isotope labeling and collecting a 2-D 1H–15N HSQC. Uniform 13C labeling for the 1 H–13C HSQC experiment is rarely done as the background signal detected is quite high. However, selective labeling schemes to incorporate 13C isotope into methyl groups can be beneficial, particularly for larger proteins, and significantly reduce the background signal detected as very few metabolites incorporate the 13C label introduced by this strategy. An alternative to labeling methyl groups with 13C isotope is to use 19F trifluoromethyl groups, but this procedure has limitations both in protein production and the spectral information produced. If the target protein yields a successful in-cell HSQC spectrum, then higher dimensional experiments and/or binding studies can be designed and implemented. Binding studies require tight regulation over the delivery and concentration of the type of binding partner to be studied. For protein–protein interaction studies, it is desirable to be able to detect both proteins individually and to have independent control over expressing or delivering both proteins to the same cell (see Note 6). The experimental design depends on the goal of the experiment; however, all in-cell NMR experiments have a similar basic protocol as described below (Fig. 3). 1. Using a frozen glycerol stock or single LB selection plate colony, inoculate 5 mL of LB medium with the appropriate antibiotic (Table 2). Incubate the culture for 12–16 h at 37°C in a shaking incubator. The container used for all cultures should be at least three to four times the volume of the medium volume to ensure proper aeration. We find it convenient to start this growth the afternoon before expression. 2. Start two 50 mL cultures in LB medium containing the appropriate antibiotic (Table 2) by inoculating from the overnight culture to a starting OD600 » 0.05 (see Note 7). Two cultures are grown to have a sample that can be used for setting NMR shims and other parameters on the spectrometer without losing valuable data collection time.

15

In-Cell NMR Spectroscopy in Escherichia coli

Centrifuge Switch media

Centrifuge 20% Slurry

al

ign

tS tec

De

No Grow

271

Induce

Sig

na

l

Lyse cells

Fig. 3. Schematic of the standard in-cell NMR experiment.

3. Place the cultures in a 37°C shaking incubator and grow until the cells reach log phase (OD600 » 0.4–0.6). Remove two 1-mL samples from each culture and harvest the cells by spinning at 1,000 × g for 15 min at 25°C. These are the “before induction” SDS-PAGE samples. Two samples are harvested so that one sample can be lysed and the other sample can be left intact to identify artifacts that may be caused by the lysis protocol. 4. Harvest the remainder of the cells at 1,000 × g for 15 min at 4°C. 5. Resuspend the cell pellets in an equivalent volume of isotopically labeled M9 minimal medium containing the appropriate antibiotic (Table 2). 6. Allow the cells to recover for ~10 min, then add the appropriate inducing agent (Table 1), and allow the cells to grow for a predetermined optimum expression time. Remove two 1-mL samples for the SDS-PAGE gel analysis “after induction” time point as was done in step 3. 7. Two hours before the induction is complete, gently harvest one 50-mL culture by spinning at 1,000 × g for 15 min at 4°C. Pour the spent medium into a beaker and determine the volume of the cell pellet by comparing it to a known volume of water. Resuspend the cell pellet to a 20% slurry using spent medium containing 10% D2O. Place the cell slurry into a 5-mm NMR tube and set up the NMR instrument using this sample (tune, shim, and calibrate rf pulses). At the end of the optimal induction time, take the second culture, collect two 1-mL samples for gel analysis, and then gently harvest the remainder of the cells as described in step 4. Resuspend the cell pellet as before, place in a clean 5-mm NMR tube, and collect the desired data. Before and after running the NMR experiment, perform the serial dilution assay (Subheading 3.2) to assess cell viability.

272

K.E. Robinson et al.

(a) If protein signals are detected during the NMR experiment, gently harvest the cells from the sample and collect an HSQC on the supernatant. This is done to ensure that the observed protein is found within the cells and does not originate from the medium. (b) If no protein signal is detected, lyse the cells by freeze– thaw lysis (Subheading 3.3, see Note 8) using a buffer volume that is equivalent to the original NMR sample volume, and the cleared lysate can be examined by NMR. This is done to determine if the protein concentration achieved during expression is sufficient to detect by NMR spectroscopy. 3.2. Cell Viability Assay (41) (see Note 3)

1. Using a marker, draw partitions on the outside of an LB agar selective plate to section it as shown in Fig. 4. 2. Perform two to three serial 1:100 dilutions, followed by three serial 1:10 dilutions into fresh LB medium (Fig. 4). To reduce error, adjust each dilution to a final volume of 1 mL. Vortex each dilution prior to preparing the next dilution. This is done to ensure that cells are evenly suspended. 3. Plate 10 mL drops of each dilution on the previously marked selective plate, one drop per section (Fig. 4). Vortex the sample in between each drop. Allow drops to thoroughly dry on the plate and incubate at 37°C for 12–16 h. 4. Count the number of colonies per section. To calculate cell viability, identify the section(s) containing 3–30 single colonies, count the number of colonies for each 10 mL drop (these are typically called colony-forming units (CFU)), and divide by the volume (V) plated (in mL) times the dilution factor (D) (10−x): (CFU/(V × D)). A healthy bacterial sample with an OD600 of ~1 should give ~109 CFU/mL. It is important to keep in mind that cell viability altered by just one dilution indicates that 90% of the viable cells have died off.

11 4 11

11

4

10 6

6 10 8 8

102 104 106 107 108 109 Serial Dilutions

NMR Tube

Fig. 4. Serial dilution assay schematic.

9

8 9 9

Selective Plate

10

15

3.3. Freeze–Thaw Lysis Protocol (see Note 8)

In-Cell NMR Spectroscopy in Escherichia coli

273

1. Prepare the freeze–thaw lysis buffer immediately before use (see Note 9). 2. Resuspend one of each of the two samples gathered from each point of the protein induction (i.e., before and after induction) in 50 mL of lysis buffer. 3. Place the samples in a dry ice/ethanol bath for 5 min. Completely thaw at room temperature. Repeat the freeze–thaw cycle two more times. Be careful when thawing as some proteins are temperature sensitive and may be aggregated by high or low temperatures (see Note 7). 4. Spin the samples at 16,000 × g for 10 min at 25°C. Transfer the supernatant to a fresh Eppendorf tube; this is the “cleared lysate” sample. The pellet left behind is the “insoluble” sample. 5. Mix 50 mL of cleared lysate with 50 mL of 2× loading dye. Resuspend the insoluble pellet using 25 mL of 2× loading dye and 25 mL of dIH2O. Boil the samples for 10 min. 6. Resuspend the cell pellets that were not used in step 2 in 25 mL of 2× loading dye and 25 mL of dIH2O. Boil the samples for 10 min. 7. Run an SDS-PAGE gel to analyze expression levels and determine the solubility of the protein (42). Solubility is determined by what sample the target protein appears in. If the target protein appears in the cleared lysate sample, then the protein is soluble. If it appears in the insoluble pellet, then it is insoluble. 8. This method can also be applied to an in-cell NMR sample if there was no protein signal detected during the experiment. The volume of lysis buffer should be equal to the volume of the NMR sample. Follow the freeze–thaw protocol through step 4, then place the cleared lysate into an NMR tube, and examine it for a protein signal.

4. Notes 1. Currently, E. coli BL21 (DE3), BL21 (DE3) Gold, BL21 (DE3) Rosetta, and JM109 (DE3) have been used for performing in-cell NMR experiments. There are many other expression strains of E. coli that serve specific purposes. We recommend examining the available expression strains and selecting a strain in which overexpression gives the highest yield of soluble protein of interest. To do this, we typically lyse induced cells, separate the soluble and insoluble portions, and examine by SDS-PAGE analysis.

274

K.E. Robinson et al.

2. We recommend examining the cell slurry at the end of the experiment to ensure that the cell slurry remains within the coil volume in the probe. If the cells have all settled out of the coil volume, then it can be assumed that the detected signal arises from protein that has leaked out of the cell, making the experiment no longer an in-cell NMR experiment. 3. There are several quantitative cell viability assays that can be used, including cell counting, serial dilution, colorimetric assays, etc. The important aspect of the quantitative assay is that it has the ability to detect a range of viability allowing the determination of the percentage of viable cells at the end of the experiment. 4. New software for FT processing is free and available upon request from Dr. Pei Zhou or Dr. Brian Coggins. Their e-mail addresses can be found at this Web site: http://zhoulab.biochem. duke.edu/. The software for maximum entropy processing can be obtained (free for not-for-profit organizations) by going to Dr. Jeffrey Hochs’ Web site: http://structbio.uchc.edu/HochLab_files/ Hoch_Lab/Software.html. 5. The glucose, thiamine, MgSO4, FeCl3, ZnSO4, and CaCl2 stocks should all be filter sterilized before use. 6. Protein–protein interaction studies within a living cell can be very informative. They generally require that both proteins can be independently detected within a living cell. If both proteins are expressed using the same promoter, it is very difficult to selectively label one protein. Therefore, the protein expression should be designed to use two different promoter systems (e.g., one protein is expressed using the T7 promoter system and the other protein is expressed using the arabinose system, as was reported by Burz et al. (7)). Since this requires two separate vectors, it is necessary to use two different antibiotics for selection purposes. This means that both antibiotics should be present in all cultures to ensure that neither vector is lost from the cell. If the target proteins have well-dispersed spectra, it may be possible to coexpress the proteins using a single vector (such as pCDFDuet), thus negating the requirement for multiple antibiotics. 7. The timing used for growing E. coli cells can vary. The procedure discussed here has been used successfully on several small proteins (8, 14, 15, 19); however, a different timing was used for the first 3-D de novo structure (10). Timing of protein expression is dependent on the expressed protein. 8. This lysis protocol is a gentle method that can be used to determine if the protein is located in the soluble or insoluble portion of the E. coli cell. For in-cell NMR spectroscopy, it is necessary

15

In-Cell NMR Spectroscopy in Escherichia coli

275

that the protein is located in the soluble portion of the cell. Proteins that are located in the insoluble portion of the cell are not detected by solution NMR spectroscopy. There are other methods of cell lysis for small volumes that utilize detergents (e.g., Bugbuster Protein Extraction Reagent (Pierce)); however, we have found that the detergent lysis methods often cause proteins to appear in the insoluble portion of the cell artificially. There are also physical methods that can lyse small volumes, such as the French Press and sonication. 9. Be aware that using lysozyme in the lysis buffer causes lysozyme to appear on the SDS-PAGE gel. Lysozyme is a 14.7-kDa protein, so if the protein of interest runs in that range the use of lysozyme is not recommended. The lysis protocol works without lysozyme present, although it is less efficient.

Acknowledgments The authors wish to thank Dr. Ronald A. Venters and Dr. Brian E. Coggins for their useful discussions and comments on this manuscript. References 1. Dobson, C. M. (2004) Chemical space and biology. Nature 432, 824–828. 2. Goto, S., Okuno, Y., Hattori, M., Nishioka, T., and Kanehisa, M. (2002) LIGAND: database of chemical compounds and reactions in biological pathways. Nucl. Acids Res. 30, 402–404. 3. Lander, E. S., et al. (2001) Initial sequencing and analysis of the human genome. Nature 409, 860–921. 4. Goodsell, D. S. (1991) Inside a living cell. Trends Biochem. Sci. 16, 203–206. 5. Ellis, R. J., and Minton, A. P. (2003) Cell biology: join the crowd. Nature 425, 27–28. 6. Hall, D., and Minton, A. P. (2003) Macromolecular crowding: qualitative and semiquantitative successes, quantitative challenges. Biochim. Biophys. Acta. 1649, 127–139. 7. Burz, D. S., Dutta, K., Cowburn, D., and Shekhtman, A. (2006) Mapping structural interactions using in-cell NMR spectroscopy (STINT-NMR). Nat. Methods 3, 91–93. 8. Reardon, P. N., and Spicer, L. D. (2005) Multidimensional NMR spectroscopy for protein characterization and assignment inside cells. J. Am. Chem. Soc. 127, 10848–10849.

9. Sakakibara, D., Sasaki, A., Ikeya, T., Hamatsu, J., Hanashima, T., Mishima, M., Yoshimasu, M., Hayashi, N., Mikawa, T., Walchli, M., Smith, B. O., Shirakawa, M., Guntert, P., and Ito, Y. (2009) Protein structure determination in living cells by in-cell NMR spectroscopy. Nature 458, 102–105. 10. Ikeya, T., Sasaki, A., Sakakibara, D., Shigemitsu, Y., Hamatsu, J., Hanashima, T., Mishima, M., Yoshimasu, M., Hayashi, N., Mikawa, T., Nietlispach, D., Walchli, M., Smith, B. O., Shirakawa, M., Guntert, P., and Ito, Y. (2010) NMR protein structure determination in living E. coli cells using nonlinear sampling. Nat. Protoc. 5, 1051–1060. 11. Augustus, A. M., Reardon, P. N., and Spicer, L. D. (2009) MetJ repressor interactions with DNA probed by in-cell NMR. Proc. Natl. Acad. Sci. USA 106, 5065–5069. 12. Hubbard, J. A., MacLachlan, L. K., King, G. W., Jones, J. J., and Fosberry, A. P. (2003) Nuclear magnetic resonance spectroscopy reveals the functional state of the signalling protein CheY in vivo in Escherichia coli. Mol. Microbiol. 49, 1191–1200.

276

K.E. Robinson et al.

13. Xie, J., Thapa, R., Reverdatto, S., Burz, D. S., and Shekhtman, A. (2009) Screening of small molecule interactor library by using in-cell NMR spectroscopy (SMILI-NMR). J. Med. Chem. 52, 3516–3522. 14. Serber, Z., Keatinge-Clay, A. T., Ledwidge, R., Kelly, A. E., Miller, S. M., and Dotsch, V. (2001) High-resolution macromolecular NMR spectroscopy inside living cells. J. Am. Chem. Soc. 123, 2446–2447. 15. Serber, Z., Straub, W., Corsini, L., Nomura, A. M., Shimba, N., Craik, C. S., Ortiz de Montellano, P., and Dotsch, V. (2004) Methyl groups as probes for proteins and complexes in in-cell NMR experiments. J. Am. Chem. Soc. 126, 7119–7125. 16. Tugarinov, V., and Kay, L. E. (2003) Ile, Leu, and Val methyl assignments of the 723-residue malate synthase G using a new labeling strategy and novel NMR methods. J. Am. Chem. Soc. 125, 13868–13878. 17. Goto, N. K., Gardner, K. H., Mueller, G. A., Willis, R. C., and Kay, L. E. (1999) A robust and cost-effective method for the production of Val, Leu, Ile (delta 1) methyl-protonated 15N-, 13 C-, 2H-labeled proteins. J. Biomol. NMR 13, 369–374. 18. Li, C., Wang, G. F., Wang, Y., Creager-Allen, R., Lutz, E. A., Scronce, H., Slade, K. M., Ruf, R. A., Mehl, R. A., and Pielak, G. J. (2010) Protein (19)F NMR in Escherichia coli. J. Am. Chem. Soc. 132, 321–327. 19. Serber, Z., Ledwidge, R., Miller, S. M., and Dotsch, V. (2001) Evaluation of parameters critical to observing proteins inside living Escherichia coli by in-cell NMR spectroscopy. J. Am. Chem. Soc. 123, 8895–8901. 20. Li, C., Charlton, L. M., Lakkavaram, A., Seagle, C., Wang, G., Young, G. B., Macdonald, J. M., and Pielak, G. J. (2008) Differential dynamical effects of macromolecular crowding on an intrinsically disordered protein and a globular protein: implications for in-cell NMR spectroscopy. J. Am. Chem. Soc. 130, 6310–6311. 21. Barnes, C. O., and Pielak, G. J. (2010) In-cell protein NMR and protein leakage, Proteins 79, 347–351. 22. Burz, D. S., and Shekhtman, A. (2008) In-cell biochemistry using NMR spectroscopy. PLoS One 3, e2571. 23. Dedmon, M. M., Patel, C. N., Young, G. B., and Pielak, G. J. (2002) FlgM gains structure in living cells. Proc. Natl. Acad. Sci. USA 99, 12681–12684. 24. McNulty, B. C., Young, G. B., and Pielak, G. J. (2006) Macromolecular crowding in the Escherichia coli periplasm maintains alpha-synuclein disorder. J. Mol. Biol. 355, 893–897.

25. Mandelshtam, V. A., Taylor, H. S., and Shaka, A. J. (1998) Application of the filter diagonalization method to one- and two-dimensional NMR spectra. J. Magn. Reson. 133, 304–312. 26. Kupce, E., and Freeman, R. (2003) Fast multidimensional NMR of proteins. J. Biomol. NMR 25, 349–354. 27. Schanda, P., Kupce, E., and Brutscher, B. (2005) SOFAST-HMQC experiments for recording two-dimensional heteronuclear correlation spectra of proteins within a few seconds. J. Biomol. NMR 33, 199–211. 28. Kupce, E., and Freeman, R. (2003) Projectionreconstruction of three-dimensional NMR spectra. J. Am. Chem. Soc. 125, 13958–13959. 29. Coggins, B. E., Venters, R. A., and Zhou, P. (2004) Generalized reconstruction of n-D NMR spectra from multiple projections: application to the 5-D HACACONH spectrum of protein G B1 domain. J. Am. Chem. Soc. 126, 1000–1001. 30. Coggins, B. E., and Zhou, P. (2007) Sampling of the NMR time domain along concentric rings. J. Magn. Reson. 184, 207–221. 31. Barna, J. C. J., Laue, E. D., Mayger, M. R., Skilling, J., and Worrall, S. J. P. (1987) Exponential Sampling, an alternative method for sampling in two-dimensional NMR experiments. J. Magn. Reson. 73, 69–77. 32. Hiller, S., Fiorito, F., Wuthrich, K., and Wider, G. (2005) Automated projection spectroscopy (APSY). Proc. Natl. Acad. Sci. USA 102, 10876–10881. 33. Eghbalnia, H. R., Bahrami, A., Tonelli, M., Hallenga, K., and Markley, J. L. (2005) Highresolution iterative frequency identification for NMR as a general strategy for multidimensional data collection. J. Am. Chem. Soc. 127, 12528–12536. 34. Kupce, E., and Freeman, R. (2004) Projectionreconstruction technique for speeding up multidimensional NMR spectroscopy. J. Am. Chem. Soc. 126, 6429–6440. 35. Venters, R. A., Coggins, B. E., Kojetin, D., Cavanagh, J., and Zhou, P. (2005) (4,2)D Projection – reconstruction experiments for protein backbone assignment: application to human carbonic anhydrase II and calbindin D(28 K). J. Am. Chem. Soc. 127, 8785–8795. 36. Coggins, B. E., Venters, R. A., and Zhou, P. (2005) Filtered backprojection for the reconstruction of a high-resolution (4,2)D CH3-NH NOESY spectrum on a 29 kDa protein. J. Am. Chem. Soc. 127, 11562–11563. 37. Kupce, E., and Freeman, R. (2003) Reconstruction of the three-dimensional NMR spectrum of a protein from a set of plane projections. J. Biomol. NMR 27, 383–387.

15

In-Cell NMR Spectroscopy in Escherichia coli

38. Coggins, B. E., and Zhou, P. (2006) Polar Fourier transforms of radially sampled NMR data. J. Magn. Reson. 182, 84–95. 39. Selenko, P., Serber, Z., Gadea, B., Ruderman, J., and Wagner, G. (2006) Quantitative NMR analysis of the protein G B1 domain in Xenopus laevis egg extracts and intact oocytes. Proc. Natl. Acad. Sci. USA 103, 11904–11909. 40. Inomata, K., Ohno, A., Tochio, H., Isogai, S., Tenno, T., Nakase, I., Takeuchi, T., Futaki, S., Ito, Y., Hiroaki, H., and Shirakawa, M. (2009)

277

High-resolution multi-dimensional NMR spectroscopy of proteins in human cells. Nature 458, 106–109. 41. Beckman, J. S., and Siedow, J. N. (1985) Bactericidal agents generated by the peroxidasecatalyzed oxidation of para-hydroquinones. J. Biol. Chem. 260, 14604–14609. 42. Sambrook, J., and Russell, D. W., (Eds.) (2001) Molecular Cloning: A Laboratory Manual, Vol. 3, Cold Spring Harbor Laboratory Press, Cold Spring Harbor.

Chapter 16 Deuterated Peptides and Proteins: Structure and Dynamics Studies by MAS Solid-State NMR Bernd Reif Abstract Perdeuteration and back substitution of exchangeable protons in microcrystalline proteins, in combination with recrystallization from D2O-containing buffers, significantly reduce 1H, 1H dipolar interactions. This way, amide proton line widths on the order of 20 Hz are obtained. Aliphatic protons are accessible either via specifically protonated precursors or by using low amounts of H2O in the bacterial growth medium. The labeling scheme enables characterization of structure and dynamics in the solid-state without dipolar truncation artifacts. Key words: Magic angle spinning solid-state NMR, Perdeuteration, 2H labeling, Microcrystalline proteins, 15N relaxation, Order parameters, Protein dynamics

1. Introduction Magic angle spinning (MAS) solid-state nuclear magnetic resonance (NMR) spectroscopy has rapidly progressed over the past 10 years. Whereas samples with one or two NMR active nuclei were investigated in the past, uniformly labeled samples are now the focus of investigations. This development was made possible with the advent of microcrystalline proteins (1, 2). Clearly, those samples hold the potential to characterize a multitude of interactions at the same time using only one sample. On the other hand, the problem of dipolar truncation (3), i.e., the suppression of weak interactions in the presence of strong couplings, needs to be addressed to derive a structure in the end. Dipolar truncation can be circumvented by preparing samples that are magnetically dilute in the carbon spin system. This can be achieved by growing bacteria that overexpress the protein of interest in a medium that contains (1,3)-13C-glucose

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_16, © Springer Science+Business Media, LLC 2012

279

280

B. Reif

or (2)-13C-glucose (4). Progress in hardware (5–7) and sample preparation (1, 8–11) resulted in the structural characterization of these crystalline proteins (12–17). Due to the fact that the line width in the solid state is independent of molecular tumbling, very large crystalline protein complexes can be investigated (18). This was shown for the 143-kDa tryptophane synthase (19), as well as for the 480-kDa ferretin (20). Experiments are not limited to precipitated/crystalline proteins, but can also be carried out in solution, where the viscosity is large enough that the tumbling correlation time exceeds the MAS rotor period (21). In addition to investigations involving crystalline/soluble proteins, NMR experiments are performed on noncrystalline, uniformly isotopically enriched samples, like membrane proteins and amyloid fibrils. In this context, a GPCR-bound ligand (22), a toxin binding to the nicotinic acetylcholine receptor (23), and a potassium channel KcsA with an interacting toxin (24) were characterized by using NMR spectroscopy. Furthermore, the uniformly isotopically labeled membrane proteins phospholamban (25), OmpG (26), EmrE (27), sensory rhodopsin (28), proteorhodopsin (29, 30), the ABC transporter ArtJ (31), and DsbB (32, 33) were investigated, and amyloid fibrils formed from the Alzheimer’s disease ß-amyloid peptide (34–37), transthyretin (38, 39), the WW domain (40), Het-s (41, 42), IAPP (43–45), α-synuclein (46, 47), and Ure2p (48) were assigned and structurally characterized. Furthermore, the microtubule-binding protein CAP-Gly (49, 50) was characterized using MAS solid-state NMR and, large protein complexes, such as the small heat-shock protein αB-crystallin, which are not amenable to solution-state NMR methods, were successfully studied by solid-state NMR (21, 51). Recently, a number of high-quality reviews on MAS solid-state NMR methodology on biomolecules have been published ( 52– 57). The focus of this chapter is, therefore, on NMR spectroscopic investigations using perdeuterated proteins. Perdeuteration significantly simplifies spectroscopy by eliminating most of the strong homo- and heteronuclear interactions that are inherent to “normal” solid-state NMR samples. In the past, MAS solid-state NMR experiments were restricted to the observation of heteronuclear spins. CRAMPS (58–60) type 1 H, 1H homonuclear decoupling approaches, such as WHH-4- (61), BR-24- (62), MSHOT- (63), and Lee-Goldburg-derived sequences, such as FSLG (64), PMLG (65), w-PMLG (66), DUMBO (67), or symmetry-based methods (68), did not succeed in reducing the unscaled proton line widths to values below 150–300 Hz (i.e., 0.25–0.5 ppm at 600 MHz). Given the small chemical shift dispersion for protons of the same kind (e.g., HN, Hα or methyl protons) with a chemical shift range of ca. 3 ppm (for each of them), the line width achieved is not sufficient to resolve individual resonances of larger molecules. Due to their high gyromagnetic

16

Deuterated Peptides and Proteins: Structure and Dynamics Studies…

281

ratio, protons should be the nucleus with the highest sensitivity. In the solid state, however, 1H, 1H dipolar couplings make proton detection ineffective as they induce significant line broadening. Application of ultrafast MAS (60–70 kHz) (69) represents an alternative to proton homonuclear decoupling. At the moment, however, with ultrafast spinning, the resolution achieved is comparable to the resolution that is obtained at moderate spinning frequencies using an advanced CRAMPS technique. Alternatively, suppression of strong 1H, 1H dipolar couplings can be chemically achieved by perdeuterating the sample. In this approach, all nonexchangeable proton sites are occupied by deuterium atoms which have a gyromagnetic ratio that is a factor of 6.5 smaller than the proton gyromagnetic ratio. Correspondingly, all 1H, 2H dipolar couplings are reduced. In addition, the interaction Hamiltonian becomes heteronuclear which makes manipulation of the interaction by MAS or RF pulses straightforward. Exchangeable sites are subsequently back substituted with protons. This strategy was pioneered in solution-state NMR (70–72). In the solid state, deuteration was first applied to small molecules (73–75) and then later extended to peptides (76–78) and proteins (79–82). The achieved dilution of the proton bath is illustrated in Fig. 1.

2. Correlation Spectroscopy To record proton-detected experiments, solvent suppression becomes a major issue. Water suppression can be achieved using pulsed field gradients (79), cross-polarization (CP) periods as spin locks to purge unwanted solvent magnetization (82), or a combination of both (84). In experiments that are carried out with proteins that are recrystallized from buffers containing 100% H2O, the 1H line width of most of the resonances is typically on the order of 150–250 Hz and 80–150 Hz in the absence and presence of homonuclear decoupling, respectively (85). We showed recently that ultrahigh-resolution 1H spectra are obtained in MAS solidstate NMR if the respective perdeuterated protein is recrystallized from a buffer containing 90% D2O (86) (Fig. 2). The resulting 1H line width is on the order of 17–35 Hz for MAS spinning frequencies in the range of 8–24 kHz. This approach enables 1H-detected 2D 1H, 15N correlation spectroscopy without the need for homoand heteronuclear dipolar decoupling. Similarly, high-resolution spectra can be recorded for methyl protons in perdeuterated peptides and proteins (Fig. 3a) (87). The bacteria that overexpress the SH3 domain are grown in a medium containing glucose that is only ~97% enriched in deuterium. The likelihood that a proton gets incorporated into a methyl group is, therefore, on the order of 10%. The canonical line widths in the 1H

282

B. Reif

Fig. 1. Proton density in the α-spectrin SH3 domain upon deuteration. (a) Protonated sample. (b) Sample recrystallized from 100% H2O: Only exchangeable protons of the protein and protons of refined hydration water are displayed. (c) Sample recrystallized from 10% H2O and 90% D2O. On average, every molecule contains only 21 protons, assuming that there are 53 hydration water molecules as found in the X-ray structure (PDB: 1U06) (83). (d) Upon deuteration, 1H, 1H dipolar interactions are strongly attenuated due to the chemical dilution of the proton spins in the sample.

and 13C dimensions are 20–25 Hz and 5–8 Hz, respectively, at a MAS rotation frequency of 22 kHz. An increase in sensitivity can be achieved by making use of precursors that allow selective labeling of methyl groups in aliphatic side chains, like pyruvate (88) or α-ketoisovalerate (89). This kind of labeling strategy was pioneered by Lewis Kay and coworkers for solution-state NMR applications (90). In the solid state, care should be taken to preferentially incorporate CHD2 isotopomers, as the dipolar couplings among methyl protons in the CH3 group can induce severe line broadening.

16

Deuterated Peptides and Proteins: Structure and Dynamics Studies…

a

G51

N Chemical shift [ppm]

110

15

283

Y13

V53 V9

R21

V58 V23

A56

115

G28 F52

120

M25 V44

K27

L10 S36

125 We41 L31

K18 I30

Y15

E45 E17 E22 L61

V46 L34

Q16

D62

L12

130

T24

L8

W41

K26

K60 A11

A55

L33

10.0

9.5

9.0

8.5 1H

Y57

D14

K59 T32 R49 K39

K43

N35

D40

Q50

S19

8.0

7.5

7.0

6.5

Chemical shift [ppm]

b

1H

Line Width [Hz]

30

20 G51 G28 We41 A56 L61

10

0 0

20

40

60

80

100

120

Rotor Period [•s]

Fig. 2. (a) 1H-detected 1H, 15N correlation recorded with a perdeuterated α-spectrin SH3 sample that was recrystallized from a buffer containing 90% D2O. (b) Amide proton line widths as a function of MAS rotation frequencies for selected residues. Reproduced by permission of Wiley from Chevelkov et al. (86).

A prerequisite to achieve high-resolution spectra for long-lasting multidimensional experiments is an internal 2H lock that can be employed to decouple 2H, 13C scalar couplings, which induce a significant broadening of the resonances in 13C evolution periods (88). In principle, ultrafast MAS probes (69) might allow the use of higher proton concentrations while maintaining high 1H resolution as the residual dipolar interactions are more efficiently suppressed. However, ultrafast MAS probes impose challenges in shimming, resulting in intrinsic line widths on the order of 25–30 Hz. Problems arise from the small dimensions of the sample, in particular at high magnetic fields. This might change when susceptibility matched

B. Reif

a I30δ

12

13C

Chemical Shift (ppm)

14 A55

16

M1 V53γ1

A56

18

I30γ2 V44γ1 V58γ1 V23γ1 V44γ2 V23-γ2

A11

20

V9γ1

M25 T24

T4 V53γ2

22

L12δ1 L34δ1 L8δ1 V58γ2 L8δ2 L61δ1 T32 L33δ2 V9γ2 L10δ2 L12δ2 L33δ1

T37

24 26

L34δ2 L10δ1

L31

28 0.5 1.0 1.5 H Chemical Shift (ppm)

2.0

2.5

0.0

1

b

P54δ

S36β A55

K27

S19β

G28

8

T24β T32β

T37α

A56

P54α

G51

4

G28

5

K39 K60 Y15

10

P20α T32α

T37β

A11

SQ Chemical Shift (ppm)

3

S36β

2Hα

DQ Chemical Shift (ppm)

6

2 α H

284

V23 V53 I30

T24α

12

6

V44 F52

70

13Cα

V58

60 50 Chemical Shift (ppm)

Fig. 3. (a) 1H, 13C correlation recorded for a [U – 2H, 13C, 15N]-labeled sample of the α-spectrin SH3 domain. The experiment makes use of the residual protonation of the precursors employed during protein biosynthesis. Spectra in gray and black are recorded using INEPT and CP, respectively, for magnetization transfer. Cross peaks highlighted with circles are only visible in the INEPT version of the experiment. Reproduced by permission of Elsevier from Agarwal et al. (87). (b) Cα spectral region of the 13C-detected 2H-DQ, 13C correlation experiment applied to the SH3 domain. Dα resonances are as narrow as 16 Hz (A56) in the 2 H DQ dimension. Reproduced by permission of ACS from Agarwal et al. (93).

Deuterated Peptides and Proteins: Structure and Dynamics Studies…

285

wire and isolation become available. Then, higher concentrations of exchangeable protons might be achieved without compromising resolution (91, 92). On the other hand, 2H can be used as an additional nucleus or chemical shift dimension to disperse overlapping resonances. In the solid state, high-resolution 2H, 13C correlation experiments are possible, since overall tumbling is absent in immobilized crystalline systems (93). Figure 3b shows a 13C-detected 2H-DQ, 13C correlation recorded for a perdeuterated α-spectrin SH3 sample. The 2H dimension is realized by evolving double-quantum (DQ) coherences, since 2H-DQ are independent from deviations of the spinning axis from the magic angle, and insensitive to the motional effects that would interfere with MAS. In addition, 2H-DQ allow a doubling of the effective resolution since they evolve twice as fast as singlequantum (SQ) coherences. The efficiency of the 2H, 13C magnetization transfer is strongly coupled to the RF field strength on the 2H channel (>80 kHz). The use of optimum control (OC) in the design of better magnetization transfer sequences might alleviate this problem in the future (94). In solution-state NMR, exchangeable hydroxyl protons are difficult to assign due to their rapid exchange with the solvent. In the solid state, magnetization can be transferred approximately 50× faster using dipolar couplings. This way, many exchangeable hydroxyl protons are, thus, accessible. Figure 4 shows a 13C-detected 1 H, 13C correlation spectrum recorded for the microcrystalline α-spectrin SH3 domain (95). All three threonine hydroxyl protons

T24-Cβ

S19-Cβ

T32-Cβ

5 1H Chemical Shift (ppm)

16

S36-Cβ

H2 O T37-Cβ

6

T37-Cα T37-OH 1J

T24-HN

7

8

HN

T32-Cα

T32-OH

T37-HN

T24-OH

9

L33-HN N38-HN

10 74

72

70 13C

68

66

S36-HN

64

62

Chemical Shift (ppm)

Fig. 4. Threonine spectral region of a 13C-detected 1H, 13C correlation recorded for a [U – 2H, 13 C, 15N]-labeled sample of the α-spectrin SH3 domain. The spectrum was recorded at 5°C. Peaks are split into doublets in the 1H indirect dimension due to evolution of the 1JNH scalar coupling. Reproduced by permission of ACS from Agarwal et al. (95).

286

B. Reif

are readily assigned by correlations to the Cβ and Cα carbon chemical shifts. In EXSY-type experiments, the exchange characteristics of the respective hydroxyl proton can be probed. Dipolar recoupling experiments allow the proton to be localized within a hydrogen bond. From the dephasing behavior in REDOR-type experiments, the distance between the hydroxyl proton and a carbon atom in the donor as well as acceptor group can be deduced. We expect that these experiments have an impact on the understanding of enzymes that are involved in proton transfer reactions. Figure 5 shows the effect of miscalibration of the spinning angle on the experimental 15N and 2H spectra. The 15N spectrum in Fig. 5a is split into a doublet due to the one-bond scalar coupling between the nitrogen and the amide proton. Even small deviations of the spinning angle from the magic angle ( 0.02° for 15N) result in a noticeable deterioration of the spectral quality. The situation is worse for 2H-SQ spectroscopy. Because of the large deuterium quadrupolar interaction, 2H resonances are sensitive to a miscalibration of the spinning angle as small as 0.005° (Fig. 5b). We expect that in the future Hall devices, which allow a direct adjustment of the spinning angle without the need for an external reference sample, will greatly improve the quality of solid-state spectra (96). In addition to using specifically protonated precursors of amino acid biosynthesis, doping the sample with Cu-EDTA results in a reduction of the 1H T1 (up to a factor of 15), and thus the recycle delay of the experiment (99, 100). Therefore, the paramagnetic complex leaves the proton line width unaffected. Doping is straightforward in perdeuterated samples for which no highpower proton decoupling is required in a direct or indirect evolution period. Typical duty cycles are on the order of 30% (100-ms acquisition time, 300-ms recycle delay). However, care has to be taken if protonated samples are employed. The use of probes that largely exclude the electric field from the active volume of the sample seems mandatory (5, 6). The high resolution that is achievable in the solid state of these highly deuterated proteins enables solution state-type, scalar coupling-based correlation experiments, e.g., HNCO, HNCA, HNCACB, HNCACO, and HNCOCA, which yield reliable backbone resonance assignments (18, 101). Note that only by taking into account the HN proton chemical shift is an unambiguous backbone assignment of uniformly isotopically enriched proteins in the solid state possible. Even with 15N line widths on the order of 10 Hz (which is the typical line width of the resonances in perdeuterated SH3 recrystallized from 90% D2O), many 15N chemical shifts overlap (Fig. 2a). Finally, deuteration of nonexchangeable sites in combination with back exchange of amide protons allows the determination of HN–HN long-range distances (76, 80, 81), detection of dynamic water molecules (81, 83), and characterization of side-chain dynamics using deuterium (102, 103).

16

Deuterated Peptides and Proteins: Structure and Dynamics Studies…

a

287

b ΔνFWHM

βRL = βMA

ΔνFWHM

= 10 Hz

2H-DQ

= 10 Hz

βRL =

54.71° , βMA

βRL = 54.79°

24 kHz 10 kHz

ΔνFWHM

βRL = 54.715°

= 10 Hz

ΔνFWHM = 16 Hz 24 kHz

βRL = βMA

2H-SQ

10 kHz

βRL = 54.73° βRL = 54.6°

βRL = 54.75° 24 kHz

βRL = 54.71°

10 kHz

250

200

150 15

100

50

N Frequency [Hz]

0

–50

60

30

0

–30

–60

2H Chemical Shift (Hz)

Fig. 5. Effect of the MAS spinning axis on the 15N (a) and 2H (b) line width. (a) The simulation of the 15N spectrum (15N–1H spin pair) assumes an 15N chemical shift anisotropy of 100 ppm and a dipolar and scalar coupling to a directly bonded proton of 10 kHz and −95 Hz, respectively. The external magnetic field strength was set to 14.1 T, corresponding to a proton Larmor frequency of 600 MHz. Simulations are shown for a MAS rotation frequency of 10 kHz (dashed) and 24 kHz (solid). Mis-setting the spinning angle from the “magic angle” ( b MA = b RL = arctan 2 ≈ 54.73561) reintroduces the sum and difference anisotropy for the upfield and downfield components, respectively. Reproduced by permission of ACS from Chevelkov et al. (97). (b) Simulation of the 2H DQ (top) and DQ (bottom) spectrum. The quadrupolar coupling was assumed to be 100 kHz, setting h = 0.1. In the simulation, the Euler angle b RL, which describes the angle between the principal axes of the rotor fixed frame and the laboratory coordinate system, was varied as indicated. All simulations were carried out using the program SIMPSON (98). Reproduced by permission of ACS from Agarwal et al. (93).

( )

3. Characterization of Dynamics in the Solid State

In solution-state NMR, the relaxation properties of an NMR observable are largely determined by the overall tumbling of molecule in the solvent. Local structural fluctuations, which are often of greater interest than the characterization of the overall correlation time of a molecule, are therefore difficult to access. The situation is different when microcrystalline proteins are considered. In MAS solid-state NMR experiments, overall tumbling is absent and relaxation is mostly driven by local structural fluctuations. Therefore, the 15N-T1 relaxation time of an amide nitrogen in the protein backbone can fluctuate by several orders of magnitude (104–107). Figure 6 shows the 15N-T1 relaxation times that were

288

B. Reif

a

b 6 5 4

15

3

3

15

2

2

10

10

7 6 5 4

7 6 5 4 3

15N

– T1 [sec]

6 5 4

3

N N

(500 MHz) (600 MHz) 1 1

2

2

1

1

15N

T1 (600 MHz) 15N T (900 MHz) 1

7 6 5

10

20

30 40 Residue Number

50

7 6 5 4

10

60

20

40 30 Residue Number

50

60

Fig. 6. 15N T1 relaxation times of the α-spectrin SH3 domain recorded in the solid state (a) and in solution (b). Reproduced by permission of ACS from Chevelkov et al. (108).

obtained for a perdeuterated sample of the α-spectrin SH3 domain that was recrystallized from a buffer that contained 90% D2O (108). The results obtained in the solid state (Fig. 6a) are contrasted with the 15N-T1 relaxation times that are obtained in solution-state NMR experiments (Fig. 6b) recorded for the same protein. Clearly, the dynamic range is increased in the solid state. The use of deuterated samples ensures that 1H, 1H spin diffusion does not perturb the experimental rates. Spin diffusion would result in an averaging of the experimental rate. A comparative analysis of 15N relaxation times and order parameters in the solid state and solution shows that both techniques can be combined to allow for a more reliable quantification of motional processes (109, 110). The measured relaxation rate R1(15N) is related to the size of the N-H dipolar coupling, d, the chemical shift anisotropy, c, and the spectral density function, J(ω), according to (111, 112): d2 ⎡ J 0 (w H − wN ) + 3 J 1 (wN ) + 6 J 2 (w H + wN )⎤⎦ (1) 10 ⎣ 2 + c 2 J 1 (wN ) 15

R1 (15 N ) =

with 2

⎛ γ γ h⎞ 2 d = ⎜ H N3 ⎟ ≡ ω HN ⎝ 2π r ⎠ 2

NH

(

)

2

c = ⎡⎣ γ N B0 σ|| − σ ⊥ ⎤⎦ ≡ ωN2 ·Δσ 2 2

(2)

16

Deuterated Peptides and Proteins: Structure and Dynamics Studies…

289

rNH refers to the 1H–15N bond length, wH and wN represent the 1H and the 15N Larmor frequencies, respectively, and γH and γN are the gyromagnetic magnetic ratios of 1H and 15N, respectively. The frequency of the 1H, 15N dipole–dipole interaction is denoted as wHN. The 15N chemical shift anisotropy (CSA) can be assumed to be axially symmetric. The 15N–1H bond is tilted by approximately 20° with respect to the principal axis of the 15N CSA tensor (113). Typical values for the anisotropy of the 15N chemical shift are Δs = s|| − s = 170 ± 8 ppm or sz = 106 ± 6 ppm (9, 113–116). In the absence of motion, an effective N–H bond length of rNH = ()1/3 = 1.015 Å is assumed (117). Until recently, transverse relaxation properties of nuclear spins in the solid state were not generally accessible. An exception is represented by chemical exchange phenomena, which have a direct impact on the spectral line shape or the powder pattern. In protonated samples, the decay of transverse magnetization cannot be easily related to the motional properties of the molecule, since the magnetization decay can arise from insufficient 1H decoupling or other experimental issues. Given the high resolution that is achievable in highly deuterated proteins, proton and nitrogen spectra can be recorded without decoupling in the direct or indirect dimension (97). The resulting spectrum yields a doublet in either dimension for every amide moiety. In the solid state, two effects contribute to differences in the intensities of each of the doublets: first, the multiplet intensites are affected by a coherent effect which is MAS frequency dependent; second, static, coherent effects are welldocumented and have been exploited by many solid-state NMR groups (118–123). In brief, the chemical shift w PASN of a particular amide nitrogen of a given crystallite is determined by two contributions that are due to the 1H, 15N dipolar interaction and the 15N chemical shielding in the nitrogen principal axis system: Dipol wNPAS (b ) = wNCSA (b ) + wNH (b )

⎡ 3cos 2 b − 1 ⎤ 3cos 2 b − 1 = ⎢s Niso + dN N + D 2N z H z z HN ⎥ 2 2 ⎣ ⎦ 3cos 2 b − 1 = s isoN z + (dN + 2DHN H z )N z 2 with

(3)

⎛m ⎞g g DHN = ⎜ 0 ⎟ H 3 N , ⎝ 4π ⎠ rNH

in which s iso and dΝ describe the isotropic and anisotropic chemical shift of the nitrogen spin, respectively. b refers to the angle of the principal axis of the dipolar/shielding tensor with respect to the external magnetic field. For simplicity, it is assumed that the dipolar and chemical shielding tensors are collinear and that the 15N shielding tensor is axially symmetric. DHN represents the size of the 1H, 15N

290

B. Reif

dipolar interaction, which is dependent on the magnetic permeability m0, the gyromagnetic ratios, γH and γN, of the proton and the nitrogen nuclei, respectively, Planck’s constant, ħ, and the N–H bond length rNH. Given the fact that the proton spin state can adopt the spin quantum number ±1/2, one obtains ⎡⎛ 1 ⎞ ⎛1 ⎞⎤ dN + DHN H z = dN + DHN ⎢⎜ + H z ⎟ − ⎜ − H z ⎟ ⎥ ⎝ ⎠ ⎝ ⎠⎦ 2 2 ⎣ = d N + DHN ⎡⎣H a ⎤⎦ − DHN ⎡⎣H b ⎤⎦

(4)

⎧⎪ d N + DHN ⎡⎣H a ⎤⎦ ; upfield component =⎨ b ⎪⎩dN − DHN ⎡⎣H ⎤⎦ ; downfield component As a consequence, the b spin state of the multiplet experiences effectively only the difference anisotropy dN−DHN, whereas the a spin state experiences the sum anisotropy dN + DHN. The intensity of the spinning sideband resonances associated with the sum anisotropy is, therefore, distributed over a larger spectral region and the respective central band intensity is decreased. The opposite applies for the resonance associated with the difference tensor. As the 15N CSA and 1H–15N dipolar interactions are purely inhomogeneous as per Maricq and Waugh (124), no contribution of the coherent static effect to the 15N-Hα/Hβ multiplet line width is expected. This explanation is supported by numerical simulations in which a two-site exchange process is explicitly included in the calculation of the powder average of a 1H, 15N two-spin system (125). Static interference effects are not observable in solution-state NMR, since they are averaged to zero because of the tumbling of the molecule in solution. However, a second-order dynamic interference effect, which is based on dipolar and CSA relaxation interference, still influences the spectra. This effect is the physical basis of TROSY (126) and cross-correlated relaxation experiments (127–129). The size of the 15N CSA, 15N–1H dipole cross-correlated relaxation rate can be expressed as (127) hCSA/DD = 2ad {4 J (0) + 3 J (wN )}P2 (cos q ),

(5)

where a = −4π/3Β0(σ|| − σ )rHN3/(hγH) and d = γH2γN2h2/ (80π2rHN6). rHN refers to the H–N bond length and P2 refers to the second-order Legendre polynomial ½(3cos2q–1) with q corresponding to the angle between the principal axis of the N–H dipolar vector and the 15N CSA shielding tensor. For an isotropic motional model without internal motion, the spectral density function J(w) is given as J(w) = tC/(1 + w 2tC2). In addition to the size and relative orientation of the CSA and dipolar tensors, the cross-correlated relaxation rate is, therefore, directly proportional to the molecular correlation time tC. An exact quantification of h is typically performed

16

Deuterated Peptides and Proteins: Structure and Dynamics Studies…

291

in experiments in which the magnetization of 15N-Hα and 15N-Hβ spin states is allowed to relax for a constant time, Δ (130). Keeping the MAS rotation frequency constant, variation of the effective sample temperature allows a direct probe of CSA-dipole cross-correlated relaxation effects on the differential 15N-Hα/Hβ line width (Fig. 7). Clearly, the anisotropy of the intensities associated with the multiplet components of the Hα/Hβ spin states becomes larger at lower temperature for both L61 and D62, indicating slower motional correlation times at lower temperatures. This behavior is expected. Nevertheless, it is surprising to see that backbone motions have large enough amplitudes and correlation times to produce this effect. A significant effect can only be expected for motional correlation times that are on the order of or larger than the inverse of the 15N Larmor frequency, i.e., several ns (see Eq. 3). In the solid state, order parameters are directly accessible by measuring the dipolar interaction between two spins, e.g., 13C/15N

10 °C

4°C

17 °C

13.0 Hz

13.0 Hz

Effective Temperature

24 °C

10.2 Hz

11.2 Hz

Experimental FWHM

L61 17.9 Hz

128

126

17.7 Hz

128

17.0 Hz

126

12.5 Hz

13.8 Hz

128

16.5 Hz

126

128

13.8 Hz

126

12.6 Hz

Experimental FWHM

D62 ~30 Hz

130

128

130

128

130

128

21.4 Hz

130

128

15

N Chemical Shift [ppm]

Fig. 7. 15N columns extracted from a 2D 1H, 15N correlation experiment that was recorded without 1H decoupling in the 15N evolution period. The MAS rotation frequency was kept constant at 13 kHz. Contributions from a static correlation between the 15N CSA tensor and the 1H, 15N dipole as a possible source of the effect should, therefore, be constant and independent of a change in temperature. The anisotropy of the multiplet intensities increases at lower temperatures indicating that the correlation time implied by dynamics is decreased. The effects are, in particular, pronounced for D62 for which the upfield component cannot be detected when the sample temperature is adjusted to 4°C. Reproduced by permission of ACS from Chevelkov et al. (97).

292

B. Reif

and its directly bonded 1H (9, 131–133). Again, the dipolar interaction can be determined more reliably if perdeuterated proteins are employed since residual 1H, 1H dipolar interactions can be neglected as a possible source of error (134). Dipolar couplings, extracted from CPPI-type experiments (135–137), are almost unaffected by the radio frequency inhomogeneities of the probe, and thus enable a more accurate determination of the absolute value of the coupling. A comparison of solid-state and solution-state order parameters allows identification of slow motional processes that are normally not easily observed in solution (110). To analyze motion quantitatively, the spectral density functions in Eqs. 1 and 5 need to be expressed explicitly. This is not trivial, since the exact form of the spectral density function depends on the underlying motional model (111). In the framework of extended model-free formalism (138, 139), the spectral density functions Jm(w) are expressed as Lorentzian functions that depend on two correlation times, ts and tF, and two order parameters, SS and SF, referring to slow and fast motional processes, respectively:

(

J (w ) = 1 − S F2

) 1 + tw t F

2

2 F

(

+ S F2 1 − SS2

) 1 + tw t S

2

2 S

.

(6)

To find the best fit in the framework of an extended model-free analysis, all experimental results are combined (15R1 measured at an external field of 14.1 T and 21.1 T, corresponding to an 1H Larmor frequency of 600 MHz and 900 MHz; 1H–15N dipole, 15N CSA cross-correlated relaxation rate ηDD/CSA, and 1H, 15N dipolar couplings). In total, the data contain four experimental observables which are just enough to yield a determined system. The best fit corresponds to the minimum root mean square deviation χ between experimental and theoretical rates which is defined as ⎧ ⎡ 1 ⎪ χ = ⎨∑ ⎢ expt R1,theo − R1,expt i i R i ⎢ 1, i ⎩⎪ ⎣ 2

(

2 2⎫ ⎤ ⎡ 1 theo expt ⎤ ⎪ (7) ⎥ + ⎢ expt η − η ⎥ ⎬. ⎦ ⎪⎭ ⎥⎦ ⎣ η

)

(

)

Superscripts theo and expt denote the theoretical and experimental values for the 15N longitudinal relaxation rate R1 and the 1H–15N dipole, 15N CSA cross-correlated relaxation rate ηDD/ CSA . In the grid search, the order parameter of fast motion SF2 was calculated according to SF = S/SS while the parameter times tS, tF, and SS2 were allowed to float freely. Figure 8 shows RMSD contour plots for residue Q16 of the α-spectrin SH3 domain as a function of tS and SS2. tF was set to 22 ps, which corresponds to its optimal value obtained in the course of the grid search. In fitting the curves in Fig. 8a, only 15 N-T1 are included; (B), (C), and (D) contain crosscorrelated relaxation data (hDD/CSA) as well. We find that the minimum for the fit of the motional correlation time tS is more restricted if h DD/CSA is taken into account. Inclusion of an additional 15N-T1 relaxation time measured at a different external field strength increases the

16

Deuterated Peptides and Proteins: Structure and Dynamics Studies…

a

b

S S2

S S2

τS (ns)

τS (ns)

c

293

d

SS2

SS2

τS (ns)

τS (ns)

Fig. 8. Rms difference plots between experimental and theoretical data as a function of SS2 and τS for the residue Q16 in α-spectrin SH3. For the best fit, we obtain τF = 22 ps, and SF2 = 0.819. Data included in the fit are: (a) 15N T1 measured at 14.1 T and 21.1 T. (b) 15N T1 measured at 14.1 T and hDD/CSA. (c) 15N T1 measured at 21.1 T and hDD/CSA. (d) 15N T1 measured at 14.1 T, 15N T1 measured at 21.1 T and h DD/CSA. Reproduced by permission of Springer from Chevelkov et al. (140).

steepness of the minimum, but leaves the best fit for τS and SS2 approximately unaltered. This is in agreement with previous findings (106, 109). We expect that this kind of analysis will become more and more important in the future to characterize the dynamics of membrane proteins and amyloid fibrils. However, for soluble/crystalline proteins, solid-state NMR might also be the method of choice to quantitate dynamic processes. Overall tumbling, which is the major source of relaxation in solution, is absent in the solid state. Local motional processes are directly reflected in the respective relaxation rates, and the quantitative characterization of dynamics should thus be more accurate. In addition to backbone dynamics, information on side-chain dynamics is obtained from analysis of the 2H Pake tensor. In the past, specific deuterium labeling was used to investigate the dynamics of various crystalline and amorphous solids, like liquid crystals (141), polymers (3, 142, 143), biomembranes (144, 145), membrane proteins (146–148), and enzymes (149). If the increment in the

294

B. Reif 2

H dimension of a multidimensional experiment is chosen to be small enough, the resulting spinning sideband manifold can be employed to extract the anisotropy and asymmetry parameters for the 2H quadrupolar tensor in uniformly perdeuterated proteins (102, 150). The (scaled) anisotropy and asymmetry yield the order parameter and give direct information on the implicated motional model, respectively. The quadrupolar interaction, which dominates the deuterium spectral line shape, is very sensitive to molecular motion over a large kinetic window (3, 151). This method should apply to motional processes that are faster compared to the size of the quadrupolar interaction (ca. 165 kHz of a sp3 hybridized carbon (152)). Intermediate motions (10−4 to 10−7 s) result in line shape distortions due to anisotropic 2H T2 relaxation (142, 153, 154). In crystalline proteins, this scenario is difficult to observe due to low signal-to-noise ratios. Faster processes can, in principle, be analyzed by measuring 2H T1 relaxation times. The anisotropy of the spinlattice relaxation time T1 was used in the past to study fast molecular motion (10−8 to 10−12 s) (151). Uniformly deuterated spin systems, similar to the case of 15N–2H T1 relaxation times, also suffer from (2H, 2H) spin diffusion. The measured 2H R1 rates are generally averaged because of cross talk among the 2H spins (155). An alternative route to assess fast side-chain dynamics involves the incorporation of selective 13C labels into the methyl groups, making use of selectively isotopically enriched amino acid precursors. This can be achieved by employing α-ketoisovalerate ((12CD3) (13CHD2)-CD-CO-COO−) in protein biosynthesis, which yields efficient labeling of one methyl group (−CD2H) in valine and leucine residues (89, 90). Interestingly, the resulting side-chain 13C-T1 relaxation times match those found in solution, demonstrating that motional processes in the solid state and in solution are highly similar (89, 155). This similarity opens the door for future characterization of biomolecular dynamics in which MAS solid-state NMR might play a major role given the fact that relaxation parameters are independent of molecular tumbling. Thus, local structural fluctuations are accessible with much higher precision in the solid-state compared to solution-state NMR experiments. References 1. Pauli, J., Van Rossum, B.-J., Förster, H., De Groot, H. J. M., and Oschkinat, H. (2000) Sample Optimization and Identification of Signal Patterns of Amino Acid Side Chains in 2D-RFDR Spectra of the α-Spectrin SH3 Domain. J. Magn. Reson. 143, 411–416. 2. McDermott, A., Polenova, T., Böckmann, A., Zilm, K. W., Paulsen, E. K., Martin, R. W., and Montelione, G. T. (2000) Partial Assignments for uniformly (13C,15N)-enriched BPTI in the solid state. J. Biomol. NMR 16, 209–219.

3. Schmidt-Rohr, K., and Spiess, H. W. (1994) Multidimensional Solid-State NMR and Polymers, Academic Press, London. 4. LeMaster, D. M., and Kushlan, D. M. (1996) Dynamical Mapping of E. coli Thioredoxin via 13 C NMR Relaxation Analysis. J. Am. Chem. Soc. 118, 9255–9264. 5. Stringer, J. A., Bronnimann, C. E., Mullen, C. G., Zhou, D. H. H., Stellfox, S. A., Li, Y., Williams, E. H., and Rienstra, C. M. (2005) Reduction of RF-induced sample heating with

16

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

Deuterated Peptides and Proteins: Structure and Dynamics Studies…

a scroll coil resonator structure for solid-state NMR probes. J. Magn. Reson. 173, 40–48. Doty, F. D., Kulkarni, J., Turner, C., Entzminger, G., and Bielecki, A. (2006) Using a cross-coil to reduce RF heating by an order of magnitude in triple-resonance multinuclear MAS at high fields. J. Magn. Reson. 182, 239–253. Dillmann, B., Elbayed, K., Zeiger, H., Weingertner, M.-C., Plotto, M., and Engelke, F. (2007) A novel low-E field coil to minimize heating of biological samples in solid-state multinuclear NMR experiments. J. Magn. Reson. 187, 10–18. Martin, R. W., and Zilm, K. W. (2003) Preparation of protein nanocrystals and their characterization by solid state NMR. J. Magn. Reson. 165, 162–174. Franks, W. T., Zhou, D. H., Wylie, B. J., Money, B. G., Graesser, D. T., Frericks, H. L., Gurmukh, S., and Rienstra, C. M. (2005) Magic-Angle Spinning Solid-State NMR Spectroscopy of the beta 1 Immunoglobulin Binding Domain of Protein G (GB1): 15 N and 13 C Chemical Shift Assignments and Conformational Analysis. J. Am. Chem. Soc. 127, 12291–12305. Lorch, M., Lehner, I., Siarheyeva, A., Basting, D., Pfleger, N., Manolikas, T., and Glaubitz, C. (2005) NMR and fluorescence spectroscopy approaches to secondary and primary active multidrug efflux pumps. Biochem. Soc. Trans. 33, 873–877. Marulanda, D., Tasayco, M. L., Cataldi, M., Arriaran, V., and Polenova, T. (2005) Resonance Assignments and Secondary Structure Analysis of E. coli Thioredoxin by Magic Angle Spinning Solid-State NMR Spectroscopy. J. Phys. Chem. B 109, 18135–18145. Nomura, K., Takegoshi, K., Terao, T., Uchida, K., and Kainosho, M. (1999) Determination of the Complete Structure of a Uniformly Labeled Molecule by Rotational Resonance Solid-State NMR in the Tilted Rotating Frame. J. Am. Chem. Soc. 121, 4064–4065. Rienstra, C. M., Tucker-Kellogg, L., Jaroniec, C. P., Hohwy, M., Reif, B., McMahon, M. T., Tidor, B., Lozano-Pérez, T., and Griffin, R. G. (2002) De Novo Determination of Peptide Structure with Solid-State MAS NMR Spectroscopy. Proc. Natl. Acad. Sci. USA 99, 10260–10265. Castellani, F., van Rossum, B.-J., Diehl, A., Schubert, M., Rehbein, K., and Oschkinat, H. (2002) Structure of a protein determined by solid-state magic-angle spinning NMR. Nature 420, 98–102. Zech, S. G., Wand, A. J., and McDermott, A. E. (2005) Protein Structure Determination

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

295

by High-Resolution Solid-State NMR Spectroscopy: Application to Microcrystalline Ubiquitin. J. Am. Chem. Soc. 127, 8618–8626. Loquet, A., Bardiaux, B., Gardiennet, C., Blanchet, C., Baldus, M., Nilges, M., Malliavin, T., and Bockmann, A. (2008) 3D structure determination of the Crh protein from highly ambiguous solid-state NMR restraints. J. Am. Chem. Soc. 130, 3579–3589. Franks, W. T., Wylie, B. J., Frericks Schmidt, H. L., Nieuwkoop, A. J., Mayrhofer, R.-M., Shah, G. J., Graesser, D. T., and Rienstra, C. M. (2008) Dipole tensor-based atomic-resolution structure determination of a nanocrystalline protein by solid-state NMR. Proc. Natl Acad. Sci. USA 105, 4621–4626. Linser, R., Fink, U., and Reif, B. (2010) Narrow carbonyl resonances in proton-diluted proteins facilitate NMR assignments in the solid-state. J. Biomol. NMR 47, 1–6. Tian, Y., Chen, L., Niks, D., Kaiser, J. M., Lai, J., Rienstra, C. M., Dunn, M. F., and Mueller, L. J. (2009) J-Based 3D sidechain correlation in solid-state proteins. Phys. Chem. Chem. Phys. 11, 7078–7086. Turano, P., Lalli, D., Felli, I. C., Theil, E. C., and Bertini, I. (2010) NMR reveals pathway for ferric mineral precursors to the central cavity of ferritin. Proc. Natl Acad. Sci. USA 107, 545–550. Mainz, A., Jehle, S., van Rossum, B. J., Oschkinat, H., and Reif, B. (2009) Large Protein Complexes with Extreme Rotational Correlation Times Investigated in Solution by Magic-AngleSpinning NMR Spectroscopy. J. Am. Chem. Soc. 131, 15968–15969. Luca, S., White, J. F., Sohal, A. K., Filippov, D. V., van Boom, J. H., R., G., and Baldus, M. (2003) The conformation of neurotensin bound to its G protein-coupled receptor. Proc. Natl. Acad. Sci. USA 100, 10706–10711. Krabben, L., van Rossum, B. J., Castellani, F., Bocharov, E., Schulga, A. A., Arseniev, A. S., Weise, C., Hucho, F., and Oschkinat, H. (2004) Towards structure determination of neurotoxin II bound to nicotinic acetylcholine receptor: a solid-state NMR approach. FEBS Lett. 564, 319–324. Lange, A., Giller, K., Hornig, S., MartinEauclaire, M. F., Pongs, O., Becker, S., and Baldus, M. (2006) Toxin-induced conformational changes in a potassium channel revealed by solid-state NMR. Nature 440, 959–962. Andronesi, O. C., Becker, S., Seidel, K., Heise, H., Young, H. S., and Baldus, M. (2005) Determination of membrane protein structure and dynamics by magic-angle-spinning

296

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

B. Reif solid-state NMR spectroscopy. J. Am. Chem. Soc. 127, 12965–12974. Hiller, M., Krabben, L., Vinothkumar, K. R., Castellani, F., Van Rossum, B., Kühlbrandt, W., and Oschkinat, H. (2005) Solid-State Magic-Angle Spinning NMR of OuterMembrane Protein G from Escherichia coli. ChemBioChem. 6, 1679–1684. Agarwal, V., Fink, U., Schuldiner, S., and Reif, B. (2007) MAS Solid-State NMR Studies on the Multidrug Transporer EmrE. BBA- Biomembranes 1768, 3036–3043. Etzkorn, M., Martell, S., Andronesi, O. C., Seidel, K., Engelhard, M., and Baldus, M. (2007) Secondary structure, dynamics, and topology of a seven-helix receptor in native membranes, studied by solid-state NMR spectroscopy. Angew. Chem. Int. Edt. 46, 459–462 Shi, L., Lake, E. M. R., Ahmed, M. A. M., Brown, L. S., and Ladizhansky, V. (2009) Solidstate NMR study of proteorhodopsin in the lipid environment: Secondary structure and dynamics. Biochim. Biophys. Acta 1788, 2563–2574. Pfleger, N., Woerner, A. C., Yang, J., Shastri, S., Hellmich, U. A., Aslimovska, L., Maier, M. S. M., and Glaubitz, C. (2009) Solid-state NMR and functional studies on proteorhodopsin. Biochim. Biophys. Acta 1787, 697–705. Lange, V., Becker-Baldus, J., Kunert, B., van Rossum, B.-J., Casagrande, F., Engel, A., Roske, Y., Scheffel, F. M., Schneider, E., and Oschkinat, H. (2010) A MAS NMR Study of the Bacterial ABC Transporter ArtMP. ChemBioChem. 11, 547–555. Li, Y., Berthold, D. A., Frericks, H. L., Gennis, R. B., and Rienstra, C. M. (2007) Partial C-13 and N-15 chemical-shift assignments of the disulfide-bond-forming enzyme DsbB by 3D magic-angle spinning NMR spectroscopy. Chembiochem 8, 434–442. Li, Y., Berthold, D. A., Gennis, R. B., and Rienstra, C. M. (2008) Chemical shift assignment of the transmembrane helices of DsbB, a 20-kDa integral membrane enzyme, by 3D magic-angle spinning NMR spectroscopy. Protein Science 17, 199–204. Petkova, A. T., Ishii, Y., Balbach, J. J., Antzutkin, O. N., Leapman, R. D., Delaglio, F., and Tycko, R. (2002) A structural model for Alzheimer’s β-amyloid fibrils based on experimental constraints from solid state NMR. Proc. Natl. Acad. Sci. USA 99, 16742–16747. Petkova, A. T., Leapman, R. D., Guo, Z. H., Yau, W. M., Mattson, M. P., and Tycko, R. (2005) Self-propagating, molecular-level polymorphism in Alzheimer’s beta-amyloid fibrils. Science 307, 262–265.

36. Tycko, R. (2006) Molecular structure of amyloid fibrils: insights from solid-state NMR. Quart. Rev. Biophys. 39, 1–55. 37. Paravastu, A. K., Qahwash, I., Leapman, R. D., Meredith, S. C., and Tycko, R. (2009) Seeded growth of beta-amyloid fibrils from Alzheimer’s brain-derived fibrils produces a distinct fibril structure. Proc. Natl Acad. Sci. USA 106, 7443–7448. 38. Jaroniec, C. P., MacPhee, C. E., Astrof, N. S., Dobson, C. M., and Griffin, R. G. (2002) Molecular conformation of a peptide fragment of transthyretin in an amyloid fibril. Proc. Natl. Acad. Sci. USA 99, 16748–16753. 39. Jaroniec, C. P., MacPhee, C. E., Bajaj, V. S., McMahon, M. T., Dobson, C. M., and Griffin, R. G. (2004) High-resolution molecular structure of a peptide in an amyloid fibril determined by magic angle spinning NMR spectroscopy. Proc. Natl. Acad. Sci. USA 101, 711–716. 40. Ferguson, N., Becker, J., Tidow, H., Tremmel, S., Sharpe, T. D., Krause, G., Flinders, J., Petrovich, M., Berriman, J., Oschkinat, H., and Fersht, A. R. (2006) General structural motifs of amyloid protofilaments. Proc. Natl. Acad. Sci. USA 103, 16248–16253. 41. Ritter, C., Maddelein, M.-L., Siemer, A. B., Lührs, T., Ernst, M., Meier, B. H., Saupe, S. J., and Riek, R. (2005) Correlation of structural elements and infectivity of the HET-s prion. Nature 435, 844–848. 42. Wasmer, C., Lange, A., Van Melckebeke, H., Siemer, A. B., Riek, R., and Meier, B. H. (2008) Amyloid fibrils of the HET-s(218–289) prion form a beta solenoid with a triangular hydrophobic core. Science 319, 1523–1526. 43. Luca, S., Yau, W.-M., Leapman, R. D., and Tycko, R. (2007) Peptide Conformation and Supramolecular Organization in Amylin Fibrils: Constraints from Solid-State NMR. Biochemistry 46, 13505–13522. 44. Madine, J., Jack, E., Stockley, P. G., Radford, S. E., Serpell, L. C., and Middleton, D. A. (2008) Structural Insights into the Polymorphism of Amyloid-Like Fibrils Formed by Region 20–29 of Amylin Revealed by Solid-State NMR and X-ray Fiber Diffraction. J. Am. Chem. Soc. 130, 14990–15001. 45. Nielsen, J. T., Bjerring, M., Jeppesen, M. D., Pedersen, R. O., Pedersen, J. M., Hein, K. L., Vosegaard, T., Skrydstrup, T., Otzen, D. E., and Nielsen, N. C. (2009) Unique Identification of Supramolecular Structures in Amyloid Fibrils by Solid-State NMR Spectroscopy. Angew. Chem. Int. Edt. 48, 2118 –2121. 46. Heise, H., Hoyer, W., Becker, S., Andronesi, O. C., Riedel, D., and Baldus, M. (2005) Molecularlevel secondary structure, polymorphism, and

16

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

Deuterated Peptides and Proteins: Structure and Dynamics Studies…

dynamics of full-length alpha-synuclein fibrils studied by solid-state NMR. Proc. Natl. Acad. Sci. USA 102, 15871–15876. Kloepper, K. D., Zhou, D. H., Li, Y., Winter, K. A., George, J. M., and Rienstra, C. M. (2007) Temperature-dependent sensitivity enhancement of solid-state NMR spectra of alpha-synuclein fibrils. J. Biomol. NMR 39, 197–211. Loquet, A., Luc, Gardiennet, C., Sourigues, Y., Wasmer, C., Habenstein, B., Schutz, A., Meier, B. H., Melki, R., and Bockmann, A. (2009) Prion Fibrils of Ure2p Assembled under Physiological Conditions Contain Highly Ordered, Natively Folded Modules. J. Mol. Biol. 394, 108–118. Sun, S., Siglin, A., Williams, J. C., and Polenova, T. (2009) Solid-State and Solution NMR Studies of the CAP-Gly Domain of Mammalian Dynactin and Its Interaction with Microtubules. J. Am. Chem. Soc. 131, 10113–10126. Ahmed, S., Sun, S., Siglin, A. E., Polenova, T., and Williams, J. C. (2010) DiseaseAssociated Mutations in the p150(Glued) Subunit Destabilize the CAP-gly Domain. Biochemistry 49, 5083–5085. Jehle, S., van Rossum, B.-J., Stout, J. R., Noguchi, S. M., Falber, K., Rehbein, K., Oschkinat, H., Klevit, R. E., and Rajagopal, P. (2009) alpha B-Crystallin: A Hybrid SolidState/Solution-State NMR Investigation Reveals Structural Aspects of the Heterogeneous Oligomer. J. Mol. Biol. 385, 1481–1497. Baldus, M. (2002) Correlation experiments for assignment and structure elucidation of immobilized polypeptides under magic angle spinning. Prog. NMR Spect. 41, 1–47. Baldus, M. (2006) Molecular interactions investigated by multi-dimensional solid-state NMR. Curr. Opin. Struct. Biol 16, 618–623. Brown, S. P. (2007) Probing proton–proton proximities in the solid state. Prog. NMR Spect. 50, 199–251 Böckmann, A. (2008) 3D protein structures by solid-state NMR: ready for high resolution. Angew. Chem. Int. Edt. 47, 6110–6113. Wylie, B. J., and Rienstra, C. M. (2008) Multidimensional solid state NMR of anisotropic interactions in peptides and proteins. J. Chem. Phys. 128, 052207. McDermott, A. (2009) Structure and Dynamics of Membrane Proteins by Magic Angle Spinning Solid-State NMR. Ann. Rev. Biophys. 38, 385–403. Schnabel, B., Haubenreisser, U., Scheler, G., and Müller, R. (1976) in 19th Congress Ampere pp 441, Heidelberg.

297

59. Gerstein, B. C., Chow, C., Pembleton, R. G., and Wilson, R. C. (1977) Utility of Pulse Nuclear Magnetic Resonance in Studying Protons in Coals. J. Phys. Chem. 81, 565–570. 60. Burum, D. P. (1990) Combined Rotation and Multiple Pulse Spectroscopy (CRAMPS). Concepts in Magn. Reson. 2, 213–227. 61. Waugh, J. S., Huber, L. M., and Haeberlen, U. (1968) Approach to High-Resolution NMR in Solids. Phys. Rev. Lett. 20, 180. 62. Burum, D. P., and Rhim, W. K. (1979) Analysis of multiple pulse NMR in solids .3. J. Chem. Phys. 71, 944–956. 63. Hohwy, M., Bower, P. V., Jakobsen, H. J., and Nielsen, N. C. (1997) A high-order and broadband CRAMPS experiment using z-rotational decoupling Chem. Phys. Lett. 273, 297–303. 64. Bielecki, A., Kolbert, A. C., and Levitt, M. H. (1989) Frequency-switched pulse sequences – Homonuclear decoupling and dilute spin NMR in solids. Chem. Phys. Lett. 155, 341–346. 65. Vinogradov, E., Madhu, P. K., and Vega, S. (1999) High-resolution proton solid-state NMR spectroscopy by phase-modulated Lee-Goldburg experiment. Chem. Phys. Lett. 314, 443–450. 66. Vinogradov, E., Madhu, P. K., and Vega, S. (2002) Proton spectroscopy in solid state nuclear magnetic resonance with windowed phase modulated Lee–Goldburg decoupling sequences. Chem. Phys. Lett. 354, 193–202. 67. Lesage, A., Sakellariou, D., Hediger, S., Elena, B., Charmont, P., Steuernagel, S., and Emsley, L. (2003) Experimental aspects of proton NMR spectroscopy in solids using phase-modulated homonuclear dipolar decoupling. J. Magn. Reson. 163, 105–113. 68. Madhu, P. K., Zhao, X., and Levitt, M. H. (2001) High-resolution H-1 NMR in the solid state using symmetry-based pulse sequences. Chem. Phys. Lett. 346, 142–148. 69. Samoson, A., Tuherm, T., Past, J., Reinhold, A., Anupold, T., and Heinmaa, N. (2005) New horizons for magic-angle spinning NMR. Top. Curr. Chem. 246, 15–31. 70. LeMaster, D. M., and Richards, F. M. (1988) NMR Sequential Assignment of Escherichia Coli Thioredoxin Utilizing Random Fractional Deuteration. Biochemistry 27, 142–150. 71. LeMaster, D. M. (1989) Deuteration in protein proton magnetic resonance. Methods Enzymol. 177, 23–43. 72. Kay, L. E., and Gardner, K. H. (1997) Solution NMR spectroscopy beyond 25 kDa. Curr. Op. Struct. Biol. 7, 722–731.

298

B. Reif

73. McDermott, A. E., Creuzet, F. J., Kolbert, A. C., and Griffin, R. G. (1992) High-Resolution Magic-Angle-Spinning NMR Spectra of Protons in Deuterated Solids. J. Magn. Reson. 98, 408–413. 74. Zheng, L., Fishbein, K. W., Griffin, R. G., and Herzfeld, J. (1993) Two-Dimensional Solid-State 1H NMR and Proton Exchange. J. Am. Chem. Soc. 115, 6254–6261. 75. Zorin, V. E., Brown, S. P., and Hodgkinson, P. (2006) Origins of linewidth in 1H magicangle spinning NMR. J. Chem. Phys. 125, 144508. 76. Reif, B., Jaroniec, C. P., Rienstra, C. M., Hohwy, M., and Griffin, R. G. (2001) 1H-1H MAS Correlation Spectroscopy and Distance Measurements in a Deuterated Peptide. J. Magn. Reson. 151, 320–327. 77. Reif, B., and Griffin, R. G. (2003) 1H detected 1 H,15N Correlation Spectroscopy in Rotating Solids. J. Magn. Reson. 160, 78–83. 78. Zhou, D. H., Graesser, D. T., Franks, W. T., and Rienstra, C. M. (2006) Sensitivity and resolution in proton solid-state NMR at intermediate deuteration levels: Quantitative linewidth analysis and applications to correlation spectroscopy. J. Magn. Reson. 178, 297–307. 79. Chevelkov, V., van Rossum, B. J., Castellani, F., Rehbein, K., Diehl, A., Hohwy, M., Steuernagel, S., Engelke, F., Oschkinat, H., and Reif, B. (2003) 1H detection in MAS solid state NMR spectroscopy employing pulsed field gradients for residual solvent suppression. J. Am. Chem. Soc. 125, 7788–7789. 80. Reif, B., van Rossum, B. J., Castellani, F., Rehbein, K., Diehl, A., and Oschkinat, H. (2003) Determination of 1H 1H distances in a uniformly 2H,15N labeled SH3 domain by MAS solid state NMR spectroscopy. J. Am. Chem. Soc. 125, 1488–1489. 81. Paulson, E. K., Morcombe, C. R., Gaponenko, V., Dancheck, B., Byrd, R. A., and Zilm, K. W. (2003) High-Sensitivity Observation of Dipolar Exchange and NOEs between Exchangeable Protons in Proteins by 3D Solid-State NMR Spectroscopy. J. Am. Chem. Soc. 125, 14222–14223. 82. Paulson, E. K., Morcombe, C. R., Gaponenko, V., Dancheck, B., Byrd, R. A., and Zilm, K. W. (2003) Sensitive High Resolution Inverse Detection NMR Spectroscopy of Proteins in the Solid State. J. Am. Chem. Soc. 125, 15831–15836. 83. Chevelkov, V., Faelber, K., Diehl, A., Heinemann, U., Oschkinat, H., and Reif, B. (2005) Detection of dynamic water molecules in a microcrystalline sample of the SH3

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

94.

domain of alpha-spectrin by MAS solidstate NMR. J. Biomol. NMR 31, 295–310. Zhou, D. H., and Rienstra, C. M. (2008) High-Performance Solvent Suppression for Proton-Detected Solid-State NMR. J. Magn. Reson. 192, 167–172. Morcombe, C. R., Paulson, E. K., Gaponenko, V., Byrd, R. A., and Zilm, K. W. (2005) H-1N-15 correlation spectroscopy of nanocrystalline proteins. J. Biomol. NMR 31, 217–230. Chevelkov, V., Rehbein, K., Diehl, A., and Reif, B. (2006) Ultra-high resolution in proton solid-state NMR at high levels of deuteration. Angew. Chem. Int. Ed. 45, 3878–3881. Agarwal, V., and Reif, B. (2008) Residual Methyl Protonation in Perdeuterated Proteins for Multidimensional Correlation Experiments in MAS solid-state NMR Spectroscopy. J. Magn. Reson. 194, 16–24. Agarwal, V., Diehl, A., Skrynnikov, N., and Reif, B. (2006) High Resolution 1H Detected 1 H,13C Correlation Spectra in MAS Solid-State NMR using Deuterated Proteins with Selective 1 H,2H Isotopic Labeling of Methyl Groups. J. Am. Chem. Soc. 128, 12620–12621. Agarwal, V., Xue, Y., Reif, B., and Skrynnikov, N. R. (2008) Protein side-chain dynamics as observed by solution- and solid-state NMR: a similarity revealed J. Am. Chem. Soc. 130, 16611–16621. Goto, N., and Kay, L. E. (2000) New developments in isotope strategies for protein solution NMR spectroscopy. Curr. Opin. Cell Biol. 10, 585–592. Zhou, D. H., Shah, G., Cormos, M., Mullen, C., Sandoz, D., and Rienstra, C. M. (2007) Proton-detected solid-state NMR Spectroscopy of fully protonated proteins at 40 kHz magic-angle spinning. J. Am. Chem. Soc. 129, 11791–11801. Zhou, D. H., Shea, J. J., Nieuwkoop, A. J., Franks, W. T., Wylie, B. J., Mullen, C., Sandoz, D., and Rienstra, C. M. (2007) SolidState Protein-Structure Determination with Proton-Detected Triple-Resonance 3D Magic-Angle-Spinning NMR Spectroscopy. Angew. Chemie Int. Edt. 46, 8380–8383. Agarwal, V., Faelber, K., Schmieder, P., and Reif, B. (2009) High-Resolution DoubleQuantum Deuterium Magic Angle Spinning Solid-State NMR Spectroscopy of Perdeuterated Proteins. J. Am. Chem. Soc. 131, 2–3. Tosner, Z., Vosegaard, T., Kehlet, C., Khaneja, N., Glaser, S. J., and Nielsen, N. C. (2009) Optimal control in NMR spectroscopy: Numerical implementation in SIMPSON. J. Magn. Reson. 197, 120–134.

16

Deuterated Peptides and Proteins: Structure and Dynamics Studies…

95. Agarwal, V., Linser, R., Fink, U., Faelber, K., and Reif, B. (2010) Identification of Hydroxyl Protons, Determination of their Exchange Dynamics, and Characterization of Hydrogen Bonding by MAS solid-state NMR Spectroscopy in a Microcrystalline Protein. J. Am. Chem. Soc. 132, 3187–3195. 96. Mamone, S., Dorsch, A., Johannessen, O. G., Naik, M. V., Madhu, P. K., and Levitt, M. H. (2008) A Hall effect angle detector for solidstate NMR. J. Magn. Reson. 190, 135–141. 97. Chevelkov, V., Faelber, K., Schrey, A., Rehbein, K., Diehl, A., and Reif, B. (2007) Differential Line Broadening in MAS solidstate NMR due to Dynamic Interference. J. Am. Chem. Soc. 129, 10195–10200. 98. Bak, M., Rasmussen, J. T., and Nielsen, N. C. (2000) SIMPSON: A General Simulation Program for Solid-State NMR Spectroscopy. J. Magn. Reson. 147, 296–330. 99. Wickramasinghe, N. P., Kotecha, M., Samoson, A., Past, J., and Ishii, Y. (2007) Sensitivity enhancement in C-13 solid-state NMR of protein microcrystals by use of paramagnetic metal ions for optimizing H-1 T-1 relaxation. J. Magn. Reson. 184, 350–356. 100. Linser, R., Chevelkov, V., Diehl, A., and Reif, B. (2007) Sensitivity Enhancement Using Paramagnetic Relaxation in MAS Solid State NMR of Perdeuterated Proteins. J. Magn. Reson. 189, 209–216. 101. Linser, R., Fink, U., and Reif, B. (2008) Proton-detected Scalar Coupling based Assignment Strategies in MAS Solid-State NMR Spectroscopy applied to Perdeuterated Proteins J. Magn. Reson. 193, 89–93. 102. Hologne, M., Faelber, K., Diehl, A., and Reif, B. (2005) Characterization of Dynamics of Perdeuterated Proteins by MAS Solid-State NMR. J. Am. Chem. Soc. 127, 11208–11209. 103. Hologne, M., Chen, Z., and Reif, B. (2006) Characterization of dynamic processes using deuterium in uniformly 2H,13C,15N enriched peptides by MAS solid-state NMR. J. Magn. Reson. 179, 20–28. 104. Cole, H. B. R., and Torchia, D. A. (1991) An NMR-study of the Backbone Dynamics of Staphylococcal Nuclease in the Crystalline State. Chem. Phys. 158, 271–281. 105. Giraud, N., Böckmann, A., Lesage, A., Penin, F., Blackledge, M., and Emsley, L. (2004) SiteSpecific Backbone Dynamics from a Crystalline Protein by Solid-State NMR Spectroscopy. J. Am. Chem. Soc. 126, 11422–11423. 106. Giraud, N., Blackledge, M., Goldman, M., Böckmann, A., Lesage, A., Penin, F., and Emsley, L. (2005) Quantitative Analysis of

107.

108.

109.

110.

111.

112.

113.

114.

115.

116.

299

Backbone Dynamics in a Crystalline Protein from Nitrogen-15 Spin-Lattice Relaxation. J. Am. Chem. Soc. 127, 18190–18201. Giraud, N., Blackledge, M., Böckmann, A., and Emsley, L. (2007) The influence of nitrogen-15 proton-driven spin diffusion on the measurement of nitrogen-15 longitudinal relaxation times. J. Magn. Reson. 184, 51–61. Chevelkov, V., Diehl, A., and Reif, B. (2008) Measurement of 15 N-T1 Relaxation Rates in a Perdeuterated Protein by MAS Solid-State NMR Spectroscopy. J. Chem. Phys. 128, 052316. Chevelkov, V., Zhuravleva, A. V., Xue, Y., Reif, B., and Skrynnikov, N. R. (2007) Combined Analysis of 15 N Relaxation Data from Solidand Solution-State NMR Spectroscopy. J. Am. Chem. Soc. 129, 12594–12595. Chevelkov, V., Xue, Y., Linser, R., Skrynnikov, N. R., and Reif, B. (2010) Comparison of Solid-State Dipolar Couplings and Solution Relaxation Data Provides Insight into Protein Backbone Dynamics. J. Am. Chem. Soc. 132, 5015–5017. Torchia, D. A., and Szabo, A. (1982) SpinLattice Relaxation in Solids. J. Magn. Reson. 49, 107–121. Cavanagh, J., Fairbrother, W. J., Palmer, A. G., and Skelton, N. J. (1996) Protein NMR Spectroscopy: Principles and Practice, Academic Press, San Diego. Chekmenev, E. Y., Zhang, Q., Waddell, K. W., Mashuta, M. S., and Wittebort, R. J. (2004) 15 N Chemical Shielding in Glycyl Tripeptides: Measurement by Solid-State NMR and Correlation with X-ray Structure. J. Am. Chem. Soc. 126, 379–384. Wylie, B. J., Franks, W. T., and Rienstra, C. M. (2006) Determinations of N-15 chemical shift anisotropy magnitudes in a uniformly N-15, C-13-labeled microcrystalline protein by three-dimensional magic-angle spinning nuclear magnetic resonance spectroscopy. J. Phys. Chem. B 110, 10926–10936. Hall, J. B., and Fushman, D. (2006) Variability of the N-15 chemical shielding tensors in the B3 domain of protein G from N-15 relaxation measurements at several fields. Implications for backbone order parameters. J. Am. Chem. Soc. 128, 7855–7870. Wylie, B. J., Sperling, L. J., Frericks, H. L., Shah, G. J., Franks, W. T., and Rienstra, C. M. (2007) Chemical-shift anisotropy measurements of amide and carbonyl resonances in a microcrystalline protein with slow magicangle spinning NMR spectroscopy. J. Am. Chem. Soc. 129, 5318–5319.

300

B. Reif

117. Yao, L., Vögeli, B., Ying, J., and Bax, A. (2008) NMR Determination of Amide N-H Equilibrium Bond Length from Concerted Dipolar Coupling Measurements. J. Am. Chem. Soc. 130, 16518–16520. 118. Zilm, K. W., and Grant, D. M. (1981) Carbon-13 Dipolar Spectroscopy of Small Organic Molecules in Argon Matrices. J. Am. Chem. Soc. 103, 2913–2922. 119. Harris, R. K., Packer, K. J., and Thayer, A. M. (1985) Slow Magic-Angle Rotation 13 C NMR Studies of Solid Phosphonium Iodides. The Interplay of Dipolar, Shielding and Indirect Coupling Tensors. J. Magn. Reson. 62, 284–297. 120. Griffey, D., and Redfield, A. (1987) Protondeteceted heteronuclear edited and correlated nuclear-magnetic-resonance and nuclear Overhauser effect in solution. Quart. Rev. Biophys. 19, 51–82. 121. Wu, G., Sun, B., Wasylishen, R. E., and Griffin, R. G. (1997) Spinning Sidebands in Slow-Magic-Angle-Spinning NMR Spectra Arising from Tightly J-Coupled Spin Pairs. J. Magn. Reson. 124, 366–371 122. Duma, L., Hediger, S., Lesage, A., Sakellariou, D., and Emsley, L. (2003) Carbon-13 lineshapes in solid-state NMR of labeled compounds. Effects of coherent CSA-dipolar cross-correlation. J. Magn. Reson. 162, 90–101. 123. Igumenova, T. I., and McDermott, A. E. (2003) Improvement of resolution in solid state NMR spectra with J-decoupling: an analysis of lineshape contributions in uniformly 13 C-enriched amino acids and proteins. J. Magn. Reson. 164, 270–285. 124. Maricq, M. M., and Waugh, J. S. (1979) NMR in rotating solids. J. Chem. Phys. 70, 3300–3316. 125. Skrynnikov, N. R. (2007) Asymmetric doublets in MAS NMR: coherent and incoherent mechanisms. Magn. Reson. Chem. 45, S161–S173. 126. Pervushin, K., Riek, R., Wider, G., and Wüthrich, K. (1997) Attenuated T2 relaxation by mutual cancellation of dipole-dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution. Proc. Natl. Acad. Sci. USA 94, 12366–12371. 127. Tjandra, N., Szabo, A., and Bax, A. (1996) Protein Backbone Dynamics and 15 N Chemical Shift Anisotropy from Quantitative Measurement of Relaxation Interference Effects. J. Am. Chem. Soc. 118, 6986–6991. 128. Reif, B., Hennig, M., and Griesinger, C. (1997) Direct Measurement of Angles Between Bond Vectors in High-Resolution NMR. Science 276, 1230–1233.

129. Reif, B., Diener, A., Hennig, M., Maurer, M., and Griesinger, C. (2000) Cross Correlated Relaxation for the Measurement of Angles between Tensorial Interactions. J. Magn. Reson. 143, 45–68. 130. Chevelkov, V., Diehl, A., and Reif, B. (2007) Quantitative Measurement of Differential 15 N-Hα/β T2 Relaxation Times in a Perdeuterated Protein by MAS Solid-State NMR Spectroscopy. Magn. Reson. Chem. 45, S156–S160. 131. Lorieau, J. L., and McDermott, A. E. (2006) Order parameters based on (CH)-C-13-H-1, (CH2)-C-13-H-1 and (CH3)-C-13-H-1 heteronuclear dipolar powder patterns: a comparison of MAS-based solid-state NMR sequences. Magn. Reson. Chem. 44, 334–347. 132. Lorieau, J. L., and McDermott, A. E. (2006) Conformational Flexibility of a Microcrystalline Globular Protein: Order Parameters by SolidState NMR Spectroscopy. J. Am. Chem. Soc. 128, 11505–11512. 133. Lorieau, J. L., Day, L. A., and McDermott, A. E. (2008) Conformational dynamics of an intact virus: Order parameters for the coat protein of Pf1 bacteriophage. Proc. Natl Acad. Sci. USA 105, 10366–10371. 134. Chevelkov, V., Fink, U., and Reif, B. (2009) Accurate Determination of Order Parameters from 1 H,15N Dipolar Couplings in MAS solid-state NMR experiments. J. Am. Chem. Soc. 131, 14018–14022. 135. Wu, X. L., and Zilm, K. W. (1993) CrossPolarization with High-Speed Magic-Angle Spinning. J. Magn. Reson. A 104, 154–165. 136. Dvinskikh, S. V., Zimmermann, H., Maliniak, A., and Sandstrom, D. (2003) Heteronuclear dipolar recoupling in liquid crystals and solids by PISEMA-type pulse sequences. J. Magn. Reson. 164, 165–170. 137. Dvinskikh, S. V., Zimmermann, H., Maliniak, A., and Sandström, D. (2005) Heteronuclear dipolar recoupling in solid-state nuclear magnetic resonance by amplitude-, phase-, and frequency-modulated Lee–Goldburg crosspolarization. J. Chem. Phys. 122, 044512. 138. Lipari, G., and Szabo, A. (1982) Model-Free Approach to the Interpretation of Nuclear Magnetic Resonance Relaxation in Macromolecules. 1. Theory and Range of Validity. J. Am. Chem. Soc. 104, 4546–4559. 139. Clore, G. M., Szabo, A., Bax, A., Kay, L. E., Driscoll, P. C., and Gronenborn, A. M. (1990) Deviations from the Simple 2-Parameter Model-Free Approach to the Interpretation of N-15 Nuclear Magnetic Relaxation of Proteins. J. Am. Chem. Soc. 112, 4989–4991.

16

Deuterated Peptides and Proteins: Structure and Dynamics Studies…

140. Chevelkov, V., Fink, U., and Reif, B. (2009) Analysis of the Dynamics of Backbone Motion in the Solid-State. J. Biomol. NMR 45, 197–206. 141. Sandström, D., and Zimmermann, H. (2000) Correlation of deuterium quadrupolar couplings and carbon-13 chemical shifts in ordered media by multiple-quantum NMR. J. Phys. Chem. B 104, 1490–1493. 142. Spiess, H. (1985) Deuteron NMR – a new tool for studying chain mobility and orientation in polymers. Adv. Polym. Sci. 66, 23–58. 143. Hirschinger, J., Miura, H., Gardner, K. H., and English, A. D. (1990) Segmental dynamics in the crystalline phase of Nylon 66 : Solid State 2 H NMR. Macromolecules 23, 2153–2169. 144. Seelig, J. (1977) Deuterium magnetic resonance: Theory and application to lipid membranes. Q. Rev. Biophys. 10, 353–418. 145. Davis, J. H. (1983) The description of membrane lipid conformation, order and dynamics by 2 H NMR. Biochim. Biophys. Acta 737, 117–171. 146. Copié, V., McDermott, A. E., Beshah, K., Williams, J. C., Spijker-Assink, M., Gebhard, R., Lugtenburg, J., Herzfeld, J., and Griffin, R. G. (1994) Deuterium Solid-State Nuclear Magnetic Resonance Studies of Methyl Group Dynamics in Bacteriorhodopsin and Retinal Model Compounds: Evidence for a 6-s-Trans Chromophore in the Protein. Biochemistry. 33, 3280–3286. 147. Williamson, P. T. F., Watts, J. A., Addona, G. H., Miller, K. W., and Watts, A. (2001) Dynamics and orientation of N+ (CD3) (3)-bromoacetylcholine bound to its binding

148.

149.

150.

151.

152.

153.

154.

155.

301

site on the nicotinic acetylcholine receptor. Proc. Natl. Acad. Sci. USA 98, 2346–2351. Howard, K. P., Liu, W., Crocker, E., Nanda, V., Lear, J., Degrado, W. F., and Smith, S. O. (2005) Rotational orientation of monomers within a designed homo-oligomer transmembrane helical bundle. Protein Sci. 14, 1019–1024. Williams, J. C., and McDermott, A. E. (1995) Dynamics of the flexible loop of triosephosphate isomerase – the loop motion is not ligand-gated Biochemistry 34, 8309–8319. Hologne, M., Chevelkov, V., and Reif, B. (2006) Deuteration of Peptides and Proteins in MAS Solid-State NMR. Prog. NMR Spect. 48, 211–232. Hoatson, G. L., and Vold, R. L. (1994) 2 H NMR Spectroscopy of Solids and Liquid Crystals. NMR Basic Principles and Progress 32, 3–61. Emsley, J. W. (2002) Solid-State NMR Spectroscopy- Principles and Applications, Duer, M.J. edt., Blackwell Science, Oxford. Wittebort, R. J., Olejniczak, E. T., and Griffin, R. G. (1987) Analysis of deuterium nuclear magnetic resonance line shapes in anisotropic media. J. Chem. Phys. 86, 5411–5420. Hologne, M., and Hirschinger, J. (2004) Molecular Dynamics as Studied by StaticPowder and MAS 2 H NMR. Solid State NMR 26, 1–10. Reif, B., Xue, Y., Agarwal, V., Pavlova, M. S., Hologne, M., Diehl, A., Ryabov, Y. E., and Skrynnikov, N. R. (2006) Protein Side-Chain Dynamics Observed by Solution- and Solidstate NMR: Comparative Analysis of Methyl 2 H Relaxation Data. J. Am. Chem. Soc. 128, 12354–12355.

Chapter 17 Solid-State NMR Spectroscopy of Protein Complexes Shangjin Sun, Yun Han, Sivakumar Paramasivam, Si Yan, Amanda E. Siglin, John C. Williams, In-Ja L. Byeon, Jinwoo Ahn, Angela M. Gronenborn, and Tatyana Polenova Abstract Protein–protein interactions are vital for many biological processes. These interactions often result in the formation of protein assemblies that are large in size, insoluble, and difficult to crystallize, and therefore are challenging to study by structure biology techniques, such as single crystal X-ray diffraction and solution NMR spectroscopy. Solid-state NMR (SSNMR) spectroscopy is emerging as a promising technique for studies of such protein assemblies because it is not limited by molecular size, solubility, or lack of longrange order. In the past several years, we have applied magic angle spinning SSNMR-based methods to study several protein complexes. In this chapter, we discuss the general SSNMR methodologies employed for structural and dynamics analyses of protein complexes with specific examples from our work on thioredoxin reassemblies, HIV-1 capsid protein assemblies, and microtubule-associated protein assemblies. We present protocols for sample preparation and characterization, pulse sequences, SSNMR spectra collection, and data analysis. Key words: SSNMR, Magic angle spinning, Protein complexes

1. Introduction Protein–protein interactions are involved in many important biological processes such as signal transduction (1), cellular transport (2), viral infection (3, 4), and immune response (5). These interactions often result in large protein complexes that are insoluble and difficult to crystallize. Because of the insolubility and inherent lack of long-range order in protein assembles, the mature structural techniques that yield atomic-level information, such as solution NMR spectroscopy and X-ray crystallography, cannot be applied to studies of such protein complexes. Solid-state NMR (SSNMR)

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_17, © Springer Science+Business Media, LLC 2012

303

304

S. Sun et al.

spectroscopy has emerged as one of the very few techniques that can yield atomic level structural information for these types of systems. Recently, several studies have been reported on SSNMR applications for analysis of protein assemblies, such as bacteriophage viruses (6), oligomeric membrane peptides and proteins (7–10), amyloid fibrils (11–17), HIV-1 capsid protein assemblies (18), microtubule-associated protein assemblies (19), as well as assemblies of soluble proteins (20–22). The major strength of SSNMR spectroscopy is that there is no intrinsic limitation on molecular size or solubility, and long-range order is not required. In large systems where the resonance lines are narrow but spectral congestion presents a challenge, sparse (23), differential (20), and selective isotopic labeling (24, 25) enables simplification of SSNMR spectra and hence detailed atomic-resolution information can be attained (21, 22, 26, 27). Furthermore, with SSNMR spectroscopy residue-specific dynamics can be probed for protein complexes on multiple timescales ranging from picoseconds to many seconds (28), which fosters a deeper understanding of their biological function. Resonance assignment (or chemical shift assignment) is a prerequisite for extracting site-specific structural and dynamics information in proteins by NMR spectroscopy, including SSNMR (29). NMR experiments for resonance assignments generate two types of information. The first is correlations between atoms within the same residue, which allow for amino acid type identification. The second is correlations between atoms belonging to neighboring residues, which allow for establishing sequential connectivities. With the intraresidue and sequential correlations and from the known primary sequence of a protein, site-specific resonance assignments are extracted. In SSNMR spectroscopy, either throughspace (dipolar) or through-bond (scalar) correlation spectroscopy can be employed for assignments under the magic angle spinning (MAS) conditions (Fig. 1 illustrates the orientation of the sample rotor with respect to the static magnetic field). MAS (30) frequencies of 8–20 kHz are usually employed for multidimensional correlation spectroscopy. In the multidimensional MAS NMR correlation experiments, the typical building blocks for constructing the pulse sequences are (1) cross polarization (CP) for 1H-15N or 1H-13C polarization transfer, (2) double cross polarization (DCP) (31) and its band-selective version (SPECIFIC-CP) (32) for 15N–13C polarization transfer, (3) PDSD (33), DARR (34), DREAM (35), RFDR (36), SPC5 (37), and several other sequences for 13C–13C magnetization transfer through proton-driven spin diffusion or its rotary-assisted variant or by direct 13C–13C dipolar recoupling, (4) TOBSY (38), CTUC-COSY (39–41), and several other sequences for 13C–13C magnetization transfer through scalar couplings. For reviews of the homonuclear and heteronuclear dipolar recoupling methods see refs. 42, 43, and 44, respectively. Figure 2 shows the

17

Solid-State NMR Spectroscopy of Protein Complexes

305

Fig. 1. Magic angle spinning: (a) the sample rotor is spun at a 54.7° angle (magic angle) with respect to the static magnetic field; (b) a Varian 3.2-mm thick wall rotor is loaded into a Varian T3 probe. The stator holds the rotor at the magic angle and allows the bearing/drive air flow to spin the rotor at desired frequency.

typical 2D and 3D MAS NMR experiments for NMR assignments based on these building blocks, and the corresponding 2D and 3D spectra for thioredoxin reassembly, CAP-Gly/MT reassembly, and HIV-1 CA assembly are presented in Fig. 3. Differential (20), selective (24, 25), and sparse (23) isotopic labeling enables simplification of NMR spectra as well as the distinction between intra- and intermolecular correlations. Nonuniform labeling is commonly employed in the structural and dynamics analysis of large protein assemblies by SSNMR spectroscopy. Differential labeling with paramagnetic tags for gaining long-range intermolecular constraints in protein interfaces is another emerging area (45–47). There is quite extensive literature on applications of these various labeling schemes to solid-state protein NMR spectroscopy (20, 21, 48, 49). In this chapter, we discuss one of the possible labeling schemes, namely the differential labeling of two interacting proteins where one molecule is enriched in 15N, and the second molecule in 13C, 15N. This approach enables detailed structural analysis of the 13C, 15 N-labeled protein and at the same time extraction of the intermolecular interface information by a suitable dipolar dephasing technique. We employed this labeling protocol in the 1-73(U-13C,15N)/74-108(U-15N) thioredoxin reassembly (Subheading 3.1), and developed a set of 2D MAS NMR experiments, which allow for simultaneous identification of the residues constituting the intermolecular interface and resonance assignment of the binding partners (27). These experiments, REDOR–PAINCP, REDOR–PDSD, REDOR–HETCOR, and HETCOR–REDOR are presented below. Pulse sequences for

306

S. Sun et al.

Fig. 2. Pulse sequences for resonance assignments of proteins in MAS solid-state NMR: (a) 2D 13C–13C DARR; (b) 2D dipolar-based NCA/NCO with SPECIFIC-CP for heteronuclear 15N–13C polarization transfer; (c) 3D dipolar-based NCACX or NCOCX with SPECIFIC-CP and DARR mixing periods for 15N–13C and 13C–13C polarization transfers, respectively; (d) 3D dipolar-based NCACB with SPECIFIC-CP and DREAM mixing periods for 15N–13C and 13C–13C polarization transfers, respectively. Filled and open rectangles represent p/2 and p pulses, respectively, unless specified otherwise.

17

Solid-State NMR Spectroscopy of Protein Complexes

307

Fig. 3. Representative solid-state NMR spectra for resonance assignments of protein complexes. (a) 2D spectra of the 1–73(U-13C,15N)/74–108(U-15N) thioredoxin reassembly demonstrating the examples of intraresidue and sequential backbone and side chain assignments; (a1) 13C–13C DARR; (a2) NCO and (a3) NCA. All spectra are recorded at 14.1 T with the MAS frequency of 10 kHz. Reproduced from ref. 22 with permission from John Wiley and Sons. (b) Overlay of 2D DARR spectra of CAP-Gly/MT (black) and CAP-Gly alone (green). The spectra of free CAP-Gly and of CAP-Gly/MT complex are acquired at 21.1 T and MAS frequency of 14 kHz. (b2) and (b3) are expansions around selected aliphatic regions (Ca–Cb or Ca–Cg correlations) to demonstrate chemical shift perturbations of CAP-Gly upon binding to microtubules. Reproduced from ref. 19 with permission from the American Chemical Society. (c) Sequential backbone connectivity for the sequence stretch A105-L111 in HIV-1 CA assemblies of conical morphology based on 3D NCOCX, NCACX, and NCACB experiments at 14.1 T and MAS frequency of 10 kHz. The residue names are shown on top of the spectra at their 15N chemical shift plane. Negative cross-peaks resulting from two-bond N–Cb correlations in the NCACB spectra are displayed in green. Reproduced from ref. 18 with permission from the American Chemical Society.

these experiments are shown in Fig. 4 and representative spectra acquired by these experiments are shown in Fig. 5. There are a number of software packages for multidimensional data processing and analysis, such as NMRPipe/NMRDraw (50),

308

S. Sun et al.

Fig. 4. Pulse sequences for interface studies by solid-state NMR. (a) 15N–13C REDOR– PAINCP; ( b ) 15N– 15N PDSD–REDOR; ( c ) 1 H– 15N HETCOR–REDOR; ( d ) 1H– 113C REDOR–HETCOR. Filled and open rectangles represent p and p/2 pulses, respectively, unless specified otherwise. XY-8 phase cycle is used in the rotor-synchronous REDOR-p pulse train.

RNMRTK (51), NMRView (52), Sparky (53), ccpNMR (54), ANSIG (55), and, SIFT (56). The choice of a particular software package is somewhat judicial as many of these programs offer similar capabilities. In our laboratory, we typically employ NMRPipe for multidimensional NMR data processing and Sparky for spectral analysis. In multidimensional processing, the choice of the processing parameters is determined by the specifics of the experiment, and in some cases, it is beneficial to process the SSNMR spectra in two or more different ways, tailored for either sensitivity or resolution enhancement. The window functions and other processing functions are applied as necessary. For example, a common processing

17

Solid-State NMR Spectroscopy of Protein Complexes

309

Fig. 5. 2D spectra for studies of intermolecular interfaces in 1–73(U-13C,15N)/74–108(U-15 N) thioredoxin reassembly: (a) REDOR–PAINCP, (b) REDOR–HETCOR, (c) HETCOR–REDOR, and (d) PDSD–REDOR. All spectra are acquired at 14.1 T with a MAS frequency of 10 kHz. Reproduced from ref. 27 with permission from the American Chemical Society.

sequence may include (in one or all dimensions, as needed): 90° or 60° shifted sine bell/sine square apodization followed by a Lorentzianto-Gaussian transformation (for sensitivity or resolution enhancement, respectively); forward linear predication in the indirect dimension(s), zero filling, phase correction, polynomial, or multipoint baseline correction. Depending on a particular experiment, maximum entropy reconstruction (57, 58) and/or nonuniform sampling algorithms (56, 59) may be beneficial. Numerical simulations of SSNMR spectra are an integral part of most of the data analysis protocols. Numerical simulations can be employed for any part of the SSNMR investigation, from pulse sequence design to interpretation of anisotropic lineshapes to

310

S. Sun et al.

quantitative calculations of specific spectra. In the past decade, several powerful software packages have been developed for numerical simulations of SSNMR experiments, including ANTIOPE (60), GAMMA (61), BlochLib (62), SIMPSON (63), and SPINEVOLUTION (64). In addition to these multipurpose simulation packages, researchers in the field often use custom-coded programs, for example, under Mathematica and Matlab environments. In our workonproteinassemblies,weutilizeSIMPSON,SPINEVOLUTION, as well as home-written Mathematica- and Fortran-based programs. Our laboratories have been working on the development of MAS SSNMR spectroscopy for investigation of protein complexes. In this chapter, we present experimental protocols for sample preparation techniques, resonance assignments by MAS NMR spectroscopy, structure analysis, and dynamics studies of protein complexes based on our work on three classes of protein complexes: thioredoxin reassembly, HIV-1 capsid protein assembly, and microtubule/CAP-Gly assembly (18, 19, 22, 27, 28). Figure 6 illustrates representative morphologies of HIV-1 CA assemblies, microtubules (MT), and CAP-Gly/MT assemblies before and after MAS.

2. Materials 2.1. Preparation of Thioredoxin Reassemblies for Solid-State NMR Studies

1. M9 Minimal Medium (1 L): Add 200 mL of 5× M9 salts, 2 mL of 1 M MgSO4, 0.1 mL of 1 M CaCl2, 20 mL of 20% glucose, 10 mL of 100 mg/mL of NH4Cl, and 1 mL of 50 mg/mL ampicillin in 767 mL of water (see Note 1). 2. M9 salts (5×): Dissolve 64 g of Na2HPO4·7H2O, 15 g of KH2PO4, and 2.5 g of NaCl in 500 mL of water. Adjust volume to 1 L with water. Divide the solution into aliquots of 200 mL. Sterilize by autoclaving for 20 min. 3. NH4Cl (100 mg/mL): Dissolve 1 g of NH4Cl into 10 mL of water. Sterilize by filtration. 4.

15

NH4Cl (100 mg/mL): Dissolve 1 g of 15NH4Cl into 10 mL of water. Sterilize by filtration.

5. Ampicillin (50 mg/mL): Dissolve 0.5 g into 10 mL of water. Sterilize by filtration. 6. Size exclusion chromatography buffer: 20 mM sodium phosphate, pH 7.0, 3 mM EDTA. Adjust pH with 1 M HCl or 1 M NaOH. 7. Anion exchange chromatography buffer: 20 mM sodium phosphate, pH 7.0, 500 mM KCl, 3 mM EDTA. Adjust pH with 1 M HCl or 1 M NaOH. 8. HiLoad Superdex 75 column (see Note 2). 9. DEAE-cellulose resin.

17

Solid-State NMR Spectroscopy of Protein Complexes

311

Fig. 6. Morphology of HIV-1CA assemblies, microtubules and CAP-Gly/MT characterized by confocal and TEM microscopy. (a) Confocal images of HIV-1CA assemblies before and after magic angle spinning of the sample; (b) TEM images of MT and MT/CAP-Gly assemblies before and after magic angle spinning of the sample.

10. Citraconylation buffer: 500 mM potassium phosphate, pH 8.5. Adjust pH with 1 M HCl or 1 M KOH. 11. Citraconic anhydride. 12. 5 M NaOH. 13. Citraconylated thioredoxin purification buffer (size exclusion): 0.5% NH4HCO3, pH 7.9. Adjust pH with 10–35% ammonium hydroxide or 1 M HCl.

312

S. Sun et al.

14. Desalting column: PD-10 disposable column packed with Sephadex G-25 medium resin. 15. Trypsin: Dissolve sequencing grade modified trypsin lyophilized powder in 50 mM acetic acid to 100 mg/mL. 16. 50% Acetic acid. 17. Sephadex G-25 and Sephadex G-50 resins (see Note 3). 18. Denaturing buffer: 10 mM potassium phosphate, pH 7.4, 7.6 M urea, pH is adjusted by titrating with 1 M HCl or 1 M KOH. 19. Refolding buffer: 100 mM potassium phosphate, pH 5.7. Adjust pH with 1 M HCl or 1 M KOH. 20. Amicon stirred cell, microcon, membrane with a 3,000 Da molecular weight cut off. 21. Precipitation buffer: 35% PEG-4,000 in 10 mM NaCH 3COO, 1 mM NaN3, pH 3.5. Adjust pH with 1 M HCl or 1 M NaOH. 22. 10–35% Ammonium hydroxide. 23. 1 M HCl. 24. 1 M KOH. 25. 1 M NaOH. 26. 1 M MgSO4. 27. 1 M CaCl2. 28. 20% (w/v) D-glucose. 29. 20% (w/v) U-13C6 D-Glucose. 30. 4 mm Bruker HRMAS or Varian 3.2 mm thick wall NMR sample rotor. 31. E. coli BL21 (DE3). 2.2. Preparation of HIV-1 CA Assemblies

1. cDNA encoding gag polyprotein, pr55gag: Obtained from the NIH AIDS Research and Reference Reagent Program (88). 2. pET21 vector (EMD). 3. Basal Vitamins Eagle medium. 4. Modified M9 growth medium: Prepared by supplementing the 1 L of standard M9 medium (Subheading 2.1) with 10.0 mL of Basal Vitamins Eagle medium (65). 5. Growth medium for selective labeling: Prepared by adding a 13 C, 15N isotopically labeled amino acid and the other 19 unlabeled amino acids to the cultures at 100 mg/L in M9 medium. 6. Anion exchange chromatography buffer: 25 mM sodium phosphate, pH 7.0, 1 mM DTT, 0.02% NaN3. Adjust pH with 1 M HCl or 1 M NaOH.

17

Solid-State NMR Spectroscopy of Protein Complexes

313

7. Cation exchange chromatography buffer A: 25 mM sodium phosphate, pH 5.8, 1 mM DTT, 0.02% NaN3. Adjust pH with 1 M HCl or 1 M NaOH. 8. Cation exchange chromatography buffer B: 25 mM sodium phosphate, pH 5.8, 1 M NaCl, 1 mM DTT, 0.02% NaN3. Adjust pH with 1 M HCl or 1 M NaOH. 9. Size exclusion chromatography buffer: 25 mM sodium phosphate, pH 6.5, 100 mM NaCl, 1 mM DTT, 0.02% NaN3. Adjust pH with 1 M HCl or 1 M NaOH. 10. Anion exchange chromatography column: HiTrap Q HP (GE healthcare). 11. Cation exchange chromatography column: HiTrap SP HP (GE healthcare). 12. Size exclusion chromatography column: HiLoad Superdex 200 (GE healthcare). 13. CA dialysis buffer: 25 mM sodium phosphate pH 5.5. Adjust pH with 1 M HCl or 1 M NaOH. 14. PEG-20,000 solution: 17.5% in H2O (e.g. dissolve 1.75 g PEG-20,000 to 8.25 mL water). 15. 10 mM EDTA-Cu(II): Dissolve EDTA-Cu(II) in 90% D2O/10%H2O. 16. CA tubular morphology incubation buffer: 50 mM Tris HCl buffer, pH 8.0, 1 M NaCl. Adjust pH with 1 M HCl or 1 M NaOH. 17. 1 M HCl. 18. 1 M NaOH. 19. E. coli Rosetta 2 (DE3). 20. 4 mm Bruker HRMAS or Varian 3.2 mm thick wall NMR sample rotor. 21. Isopropyl b-D-1-thiogalactopyranoside (IPTG) stock: 200 mM solution. 2.3. Transmission Electron Microscopy of HIV-1 CA Protein Assemblies

1. TEM staining solution: ammonium molybdate (5% w/v) in water, filtered with a 0.2-mm syringe filter. 2. Nonsterile 72-well mini trays with lids. 3. 55 mm diameter qualitative circle filter paper. 4. 60 mm × 15 mm-Petri dish. 5. Transmission electron microscope: Zeiss CEM 902, operating at 80 kV. 6. 400 mesh, Formval/carbon-coated copper grids, stabilized with evaporated carbon films.

314

S. Sun et al.

2.4. Confocal Microscopy of HIV-1 CA Protein Assemblies

1. One-well chambered cover glasses. 2. Staining solution: 0.5% (w/v) Nile Blue A in water, filtered with a 0.2-mm syringe filter. 3. Laser scanning microscope: Zeiss LSM 510 NLO (25 mW HeNe laser; 543 nm) equipped with a Zeiss 40× (NA 1.3) oil immersion objective lens.

2.5. Cryo-SEM Microscopy of HIV-1 CA Protein Assemblies

1. EM PACT high-pressure freezer (Leica). 2. Gold carrier plates. 3. Copper hats. 4. Gold (for deposition). 5. Liquid nitrogen.

2.6. Preparation of CAP-Gly/Microtubule Complexes

1. Modified M9 growth medium (Subheading 2.2). 2. IPTG stock (Subheading 2.2). 3. Buffers for Ni affinity chromatography: 20 mM Tris, pH 7.5, containing 10 mM, 50 mM, 200 mM. Adjust pH with 1 M HCl or 1 M NaOH. 4. Anion exchange buffer A: 20 mM Tris, pH 7.5, 1 mM DTT. Adjust pH with 1 M HCl or 1 M NaOH. 5. Anion exchange buffer B: 20 mM Tris, pH 7.5, 1 M NaCl, 1 mM DTT. Adjust pH with 1 M HCl or 1 M NaOH. 6. Microtubule polymerization buffer: 25 mM sodium phosphate, pH 6.0, 25 mM NaCl, 0.4 mM DTT. Adjust pH with 1 M HCl or 1 M NaOH. 7. Ni affinity chromatography column: HisTrap (GE healthcare). 8. Anion exchange chromatography column: HiTrap FF-Q (GE healthcare). 9. Lypholized bovine tubulin powder: Stored at 4°C. Generally, fresh tubulin solution is used for assays. Excess tubulin solution is quick frozen by liquid nitrogen and stored at −80°C. 10. Paclitaxel (Taxol): 3 mM paclitaxel dissolved in dimethyl sulfoxide (DMSO). Store at −20°C. 11. GTP. 12. Luria-Bertani (LB) Liquid Medium (1 L): Dissolve 10 g bactotryptone, 5 g bacto-yeast and 10 g NaCl into 950 mL of tap water. Adjust pH to 7.0 with NaOH. Adjust volume to 1 L with tap water. Sterilize by autoclaving for 20 min at 15 lb/sq.in on the liquid cycle. Let it cool to ca. 40°C to add antibiotics. 13. E. coli BL21 (DE3). 14. 1 M HCl. 15. 1 M NaOH.

17

Solid-State NMR Spectroscopy of Protein Complexes

315

3. Methods 3.1. Preparation of Thioredoxin Reassemblies for Solid-State NMR Studies

For a detailed description of the overexpression system and the purification protocol for E. coli thioredoxin, see refs. 66 and 67. For a description of proteolytic cleavage of thioredoxin at Arg-73 by trypsin digestion see ref. 68. The salient steps pertaining to the preparation of the SSNMR samples of differentially enriched thioredoxin reassembly are outlined below. 1. Prepare differentially enriched thioredoxin reassemblies by overexpressing two batches of thioredoxin in E. coli BL21(DE3) separately. Use M9 minimal medium containing 15NH4Cl and U-13C6 glucose for expression of U-13C,15N thioredoxin and use M9 minimal medium containing 15NH4Cl and natural abundance glucose for 15N thioredoxin (69). 2. Purify each batch of thioredoxin by loading the crude cell extract onto a size exclusion (Superdex 75) column. Elute the protein using size exclusion chromatography buffer. 3. Apply the eluant to an anion exchange (DEAE-cellulose) column. Elute the protein. 4. Validate the purity of thioredoxin by using SDS-PAGE and measure its concentration by UV absorbance (extinction coefficient e 280 = 14,100/M/cm) (66). 5. Cleave each protein batch at the Arg-73 site by trypsin digestion. First, dialyze thioredoxin against citraconylation buffer and concentrate the protein in an Amicon stirred cell (molecular weight cut off: 3,000 Da) to 0.3 mM. Second, block the lysine side chain amine groups by adding 25 mL of citraconic anhydride at 20 min intervals (total amount of citraconic anhydride is 50 mL for every 1 mmol of thioredoxin) and allow the reaction to continue for 2 h after the final addition. Add 5 M NaOH to maintain the pH at 8.5. Third, remove excess reagent by running the reaction mixture through a desalting column. Fourth, add trypsin to citraconylated thioredoxin (trypsin to thioredoxin ratio is 1:100 w/w) and allow the enzyme digestion to continue for 6 h at 37°C. Finally, lyophilize the mixture and incubate in 50% acetic acid for 1 h to remove citraconyl groups on lysine side chains. 6. Separate the two peptide fragments, thioredoxin (1–73) and thioredoxin (74–108), on a size exclusion chromatography (Sephadex G-50) using 50% acetic acid as elution buffer. 7. Purify each fragment separately by size exclusion chromatography (Sephadex G-25) using citraconylated thioredoxin purification elution buffer. 8. Validate the purity of each fragment by SDS-PAGE and measure the concentration of each fragment by UV absorbance.

316

S. Sun et al.

Extinction coefficients for the N fragment (1–73) and the C fragment (74–108) are e 280 = 14,100 M-1-cm-1 and e 215 = 39,700 M-1-cm-1, respectively. 9. Reconstitute thioredoxin by mixing each 13C,15N-enriched fragment with its complementary 15N-enriched counterpart. This step results in two reassembled thioredoxin samples: 1–73(U-13C,15N)/74–108(U-15N), and 1–73(U-15N)/74–108(U13 C,15N). Mix equimolar amounts of N and C fragments at low concentration (~50 mM) in denaturing buffer, then dialyze against refolding buffer. 10. Concentrate the reconstituted thioredoxin to 70 mg/mL using an Amicon stirred cell and microcon (molecular weight cut off: 3,000 Da). Gradually add precipitation buffer (10 mL every 10 min or longer) into 0.5 mL of concentrated thioredoxin solution until no further protein precipitation is observed. Quantify the extent of precipitation by measuring residual absorbance at 280 nm. 11. Centrifuge the hydrated thioredoxin/PEG precipitate at 14,000 × g for 15 min at 4°C and transfer the pellet into 4 mm Bruker HRMAS NMR sample rotor or Varian 3.2-mm thick wall NMR sample rotor. Seal the samples with the upper spacer and the top spinner (see Note 4). 3.2. Preparation of HIV-1 CA Assemblies

The cDNA encoding gag polyprotein, pr55gag, was obtained from the NIH AIDS Research and Reference Reagent Program (88). The DNA sequence coding for CA (gag residues 133–363) was amplified and subcloned into pET21 vector using the NdeI and XhoI sites (70). The primers used for PCR amplification are 5¢GAT ATA CAT ATG CCT ATA GTG CAG AAC ATC CAG GGG-3¢, and 5¢-GTG GTG CTC GAG TCA TCA CAA AAC TCT TGC CTT ATG GCC GGG-3¢, respectively. Restriction sites are underlined. 1. Express U-13C,15N isotopically labeled CA protein in E. coli Rosetta 2 (DE3) in modified M9 medium using 15NH4Cl and U-13C6 glucose as the sole nitrogen and carbon sources. Induce the protein expression with 0.4 mM IPTG at 23°C for 16 h. 2. Express selectively labeled CA protein in E coli Rosetta 2 (DE3) in M9 growth medium for selective labeling prepared by adding a 13C, 15N isotopically labeled amino acid and the other 19 unlabeled amino acids to the cultures at 100 mg/L when 0.4 mM IPTG is added to induce protein expression (see Note 5). 3. Purify the CA protein by anion exchange chromatography using anion exchange chromatography buffer. Use a flow rate of 2 mL/min and collect the flow through (nonbinding part). 4. Purify the CA protein produced in step 3 by cation exchange chromatography using a gradient formed by cation exchange

17

Solid-State NMR Spectroscopy of Protein Complexes

317

chromatography buffer A and B. The flow rate is 2 mL/min. CA containing fraction was eluted at conductivity of ca. 10 ms/cm. 5. Remove aggregates by size exclusion chromatography using size exclusion buffer. The flow rate is 2 mL/min. 6. Validate the purity of CA protein by SDS-PAGE and measure the concentration of proteins by UV absorbance (extinction coefficient e 280 = 33,585 M-1-cm-1). 7. Dialyze the CA protein against CA dialysis buffer. To prepare CA assemblies containing mixed labels, mix two solutions containing CA protein, each isotopically labeled with a different desired amino acid, in a 1:1 ratio, followed by the assembly step to produce CA assemblies of one of the three morphologies: conical, spherical, or tubular, as described below. 8. Lyophilize the purified CA protein (see Note 6). Prepare CA protein assemblies of conical morphology by adding PEG-20,000 solution to the lyophilized protein to a final protein concentration of 32 mg/mL if needed for experiments, add EDTA-Cu(II) (see Note 7). Incubate the mixture for 1 h at 37°C. Recover the assembled material as the pellet after centrifugation at 18,800 × g for 5 min at room temperature. Pack 15 mg of the precipitate into a 3.2-mm Varian NMR sample rotor and seal the sample with an upper spacer and a top spinner. 9. Prepare CA assemblies of spherical morphology by mixing a 32-mg/mL CA solution, prepared in CA dialysis buffer, with PEG-20,000 buffer (1:1 volume ratio). Incubate the resulting mixture on ice for 30 min and dilute it fourfold. Dry the solution containing the spherical assemblies with N2 gas to remove any excess water. Pack 12 mg of the dried sample into a 3.2mm Varian NMR sample rotor and seal the sample using an upper spacer and a top spinner (see Note 8). 10. Prepare CA assemblies of tubular morphology by incubating a 32-mg/mL CA solution prepared in tubular morphology incubation buffer at 37°C for 1 h. 3.3. Transmission Electron Microscopy of HIV-1 CA Protein Assemblies and CAP-Gly/Microtubule Protein Assemblies

The morphologies of the HIV-1 CA assemblies are analyzed using a Zeiss CEM 902 transmission electron microscope operating at 80 kV. Samples are stained with TEM staining solution, deposited onto 400 mesh, Formval/carbon-coated copper grids, and dried for 40 min. Follow the protocol below to prepare the TEM grids: 1. Transfer 5 mL of CA assemblies slurry and 5 mL of TEM staining solution to two separate wells of a mini tray. 2. Place the copper grid on the CA assemblies slurry first with the shiny side down, incubate for 1 min. 3. Use the edge of a filter paper to remove excess solution. 4. Place the grid on top of the staining solution and incubate for 30 s with the same side facing down.

318

S. Sun et al.

5. Use the edge of a filter paper to remove excess solution. 6. Place the grid on the CA assemblies slurry drop again for 30 s with the same side facing down. 7. Use the edge of a filter paper to remove excess solution. 8. Place the grid on the staining solution drop again for 30 s with the same side facing down. 9. Use the edge of a filter paper to remove excess solution. 10. Place one piece of filter paper in the Petri dish. Then place the copper grid on the filter paper, which is already in the Petri dish. 11. Place the Petri dish under a lamp to dry the copper grid for 40 min. 12. Place the dried copper grid on the TEM sample holder and then acquire the images. 3.4. Confocal Microscopy of HIV-1 CA Protein Assemblies

The morphologies of the HIV-1 CA assemblies are analyzed in solution using confocal microscopy. The stain, Nile Blue A is excited under a 543-nm laser line and emits fluorescence in the hydrophobic environment (protein assemblies). For confocal imaging of the CA assemblies, follow the steps below. 1. Place 1 mL of protein assemblies slurry on a cover glass (see Note 9). 2. Add 5 mL of staining solution to the protein assemblies slurry (see Note 10). 3. Acquire the images under a 543-nm laser line of a 25-mW HeNe laser scanning microscope using a Zeiss 40× (NA 1.3) oil immersion objective lens. Turn on the transmitted light channel when appropriate.

3.5. Cryo-SEM Microscopy of HIV-1 CA Protein Assemblies

The morphologies of the HIV-1 CA assemblies are analyzed using Cryo-SEM microscopy on a cold stage using a high-pressure freezer. The cryo-fixed specimens are cryo-fractured under vacuum to reveal internal structure. For cryo-SEM imaging of the CA assemblies, follow the steps below (see Note 11). 1. Place 1 mL of protein assemblies slurry on the gold carrier plates. Carefully cover the gold carrier plate with copper hat. 2. Transfer the gold plate set to the Leica EM PACT highpressure freezer and freeze the sample at 2,000 bar at dT/ dt > 10,000°C/s. 3. Transfer the frozen sample to the precooled sample preparation chamber (−125°C) with liquid nitrogen. 4. Fracture the copper cover with the knife in the preparation chamber.

17

Solid-State NMR Spectroscopy of Protein Complexes

319

5. Deposit 10 nm of gold on the freshly fractured surface and lower the temperature to −125°C. 6. Transfer the sample to the cryostage for observation and increase the temperature to −90°C for 5–7 min to remove surface water. 7. Acquire images at −125°C and 1.0 kV at a working distance of approximately 4–5 mm. 3.6. Preparation of CAP-Gly/Microtubule Complexes

CAP-Gly domain of the p150Glued subunit of mammalian dynactin encompassing residues 19–107 was subcloned into the pET28bHis6-SMT3 vector (71) using the BamHI and XhoI restriction sites. Successful subcloning was confirmed by DNA sequencing. (SMT-His6)-CAP-Gly containing plasmid was transformed into E. coli BL21(DE3) cells. For subsequent production of isotopically enriched protein, purification and assembly steps, follow the protocols below. 1. Overexpress U-13C,15N (SMT-His6)-CAP-Gly in M9 medium or modified M9 medium (see Note 12) containing 15NH4Cl and U-13C6 glucose. 2. Purify the tagged protein by Ni affinity chromatography using the buffers for Ni affinity chromatography. Remove nonbinding and nonspecifically bound protein impurities by washing the Ni-affinity column with buffers containing 10 and 50 mM imidazole, respectively. Elute tagged CAP-Gly by using the buffer containing 200 mM imidazole. 3. Overexpress His6-ULP1 protease (71) in LB medium (see Note 13). Purify the enzyme by Ni affinity chromatography following the same procedure described in step 2. Divide the His6-ULP1 protease expressed in 1 L of LB medium into Eppendorf tubes (0.5 mL/tube) and store at −80°C without assaying enzyme activity. 4. Mix 24 mL of (SMT-His6)-CAP-Gly expressed in 250 mL of M9 medium or modified M9 medium with 1–1.5 mL of His6ULP1 protease to cleave the SMT-His6 tag from CAP-Gly (see Note 13). Incubate the mixture at 4°C overnight. 5. Dilute the mixture (step 4) to 40 mL with Ni affinity chromatography buffer containing 10 mM imidazole and load the diluted mixture onto a 5-mL HisTrap (Ni affinity) column. Elute the CAP-Gly(19–107) with 10 mM imidazole buffer (CAPGly(19–107) does not bind to the column). Elute His6-SMT3 tag and His6-ULP-1 with 200 mM imidazole buffer. 6. Purify the CAP-Gly containing fractions by anion exchange chromatography to remove the residual protein and nucleic acid impurities using a gradient formed by anion exchange buffer A and B. The flow rate is set as 1 mL/min. The gradient

320

S. Sun et al.

is set as 0–40% buffer B in 100 mL elution (or 100 min). The protein is eluted when buffer B is 15–20% in the gradient. 7. Validate the purity of CAP-Gly(19–107) by SDS-PAGE. 8. Dialyze U-13C,15N CAP-Gly(19–107) against microtubule polymerization buffer. Concentrate CAP-Gly(19–107) to 7.3 mg/mL. Measure the concentration of CAP-Gly by UV absorbance (e 280 = 8,250/M/cm). 9. Dissolve bovine tubulin (lyophylized powder) in the microtubule polymerization buffer. Add GTP and paclitaxel to the tubulin solution (final concentration is 30 mM for tubulin, 1 mM for GTP, and 15 mM for paclitaxel). Incubate the mixture at 37°C for 40–45 min. 10. Mix a 1.5-mL solution of 7.3 mg/mL U-13C,15N CAP-Gly(19–107) with 3.165 mL of 23 mM paclitaxel-stabilized microtubules. Centrifuge the resulting complex at 80,000 × g for 40 min at 4°C. Pack 14.2 mg of hydrated gel-like pellets into a 3.2-mm Varian NMR sample rotor and seal the sample using an upper spacer and a top spinner (see Note 14). 3.7. Transmission Electron Microscopy of CAP-Gly/ Microtubule Protein Assemblies

The morphologies of the CAP-Gly/microtubule protein assemblies are analyzed using a Zeiss CEM 902 transmission electron microscope operating at 80 kV. Samples are stained with 5% (w/v) ammonium molybdate, deposited onto 400 mesh, Formval/ carbon-coated copper grids, and dried for 40 min. 1. Polymerize microtubules from bovine tubulin in vitro as described in Subheading 3.6. 2. Express and purify natural abundance CAP-Gly according to the procedure described in Subheading 3.6. 3. Gently mix 10 mM CAP-Gly and 10 mM microtubules to prepare CAP-Gly/microtubule assembly (see Note 15). 4. Follow steps 2–12 in Subheading 3.3.

3.8. Solid-State NMR Spectroscopy for Resonance Assignments

The typical SSNMR experimental parameters for 2D and 3D resonance assignment experiments conducted at 14.1 T on a Varian InfinityPlus instrument equipped with 3.2 mm triple tuned T3 probe are detailed below. 1. For the MAS frequency of 10 kHz, set the radio frequency (rf) field strengths to 95 kHz (1H), 50 kHz (13C), and 50 kHz (15N) for hard pulses. For 1H–13C CP or 1H–15N CP, contact times are 0.85 and 1.1 ms, respectively; 1H radio frequency field is 50 kHz, and 13C or 15N radio frequency field is ca. 40 kHz in the center of a linear or tangential ramp. Use TPPM decoupling (72); the decoupling field strengths range between 80 and 100 kHz in different experiments. Recycle delays in all

17

Solid-State NMR Spectroscopy of Protein Complexes

321

experiments are temperature dependent, and for temperatures between 0 and −30°C are typically set to 2 s. 2. For selective magnetization transfers from 15N to 13Ca (NCA) or to 13C¢ (NCO), match the 15N and 13C radio frequencies according to wN ± wC = nwr (32). For example, at 14.1 T and when the MAS frequency is 10 kHz, the typical rf field strengths are wC = 25 kHz (constant amplitude) and wN = 15 kHz (at the center of a tangential ramp). The mixing time is 6–7 ms. 3. For the DARR and PDSD sequences employed for the 13C–13C correlation spectroscopy either as stand alone experiments or as part of 3D NCACX experiment, tune the mixing time to observe cross peaks within the desired range of distances (see Note 16). Typical mixing times for one-bond correlations are 10 and 50 ms at 14.1 and 21.1 T, respectively. 4. Use the DREAM sequence in the 2D/3D NCACB experiment to establish predominantly one-bond Ca–Cb correlations following the 15N–13Ca transfer by SPECIFIC-CP. The doublequantum matching condition for the DREAM sequence is nwr = (wrf2 + W12)1/2 + (wrf2 + W22)1/2 (35). The typical mixing time in the DREAM step is ca. 2 ms (see Note 17). The one-bond Ca-Cb correlations result in negative cross peaks. Under these conditions, a number of two-bond correlations (e.g., Ca-Cg correlations for Thr residues) will appear in the spectra, and these are positive-intensity cross peaks. 3.9. Solid-State NMR Experiments for Probing Protein Backbone Dynamics

The pulse sequences used for backbone dynamics experiments are shown in Fig. 7 and described below. 1.

15

N longitudinal relaxation rates (T1) provide information about motions on the pico- to nanosecond time scales. Insert a p/2 − t − p/2 block into the NCA experiment before the DCP block, by which magnetization is transferred from 15N to 13 Ca. Monitor the decay of 15N magnetization by tracing the peak intensities in a series of 2D NCA spectra (28). Generate relaxation curves from the spectra acquired with a series of delays; for each residue, plot the cross peak intensities as a function of the delay time, and fit the experimental points to a single-exponential function I = I0exp(−R1t) to extract the residuespecific relaxation rates R1 (see Note 18).

2. For qualitative detection of submillisecond motions in backbone amide protons, insert a t/2 − p − t/2 echo before the 1H–15N CP block. Proton magnetization is rapidly dephased during the two t/2 delays if it is in the rigid environment of the solid protein. Only the amide protons with high mobility on the submillisecond time scale can survive the relatively long delay (e.g., 400 ms), and the backbone nitrogen atoms bonded to these protons detected (see Note 18).

322

S. Sun et al.

Fig. 7. Pulse sequences for dynamics studies of proteins and protein assemblies by MAS solid-state NMR. (a) 3D 1H T2¢ filtered NCA experiment; (b) 3D NCA-based 15N T1 relaxation experiment; (c) 3D DIPSHIFT-NCA experiment with R1817 block employed for dipolar recoupling; (d) 3D ROCSA-NCA experiment. Filled and open rectangles represent p/2 and p pulses, respectively, unless specified otherwise.

3.

15

N chemical shift anisotropy (CSA) is also sensitive to protein dynamics. The CSA is reduced in the presence of motions occurring at frequencies faster than the magnitude of the CSA interaction. Therefore, the ratio of the anisotropy in the presence of dynamics to the static-limit anisotropy as well as the asymmetry parameter of the dynamically averaged CSA tensor are a probe of the amplitude and geometry of the motions. 15N CSA at 14.1 T is ca. 10 kHz, and therefore, it is sensitive to the motions occurring on the time scales faster than 100 ms. Measure the 15N CSA tensors site-specifically by introducing a ROCSA CSA recoupling block (73) before the 15N chemical shift evolution period in the NCA experiment in a 3D ROCSANCA (74) experiment (see Note 19). Representative 15N CSA lineshapes are illustrated for reassembled thioredoxin in Fig. 8.

4. 1H–15N dipolar couplings are also sensitive to motions on the time scales of less than 100 ms. Record residue-specific 1H–15N dipolar lineshapes in a 3D DIPSHIFT-NCA experiment by introducing a DIPSHIFT period (75) in the basic NCA sequence. A number of dipolar recoupling sequences can be employed (the original DIPSHIFT (75), TMREV (76), LGCP (77–79)). In our work, we employ an RN-type recoupling block, R1817 (43), for the recoupling of the 1H–15N dipolar coupling and at the same time the suppression of the 1H–1H homonuclear dipolar interactions. During the R1817 recoupling period, the 15N chemical shift is refocused by a spin echo (see Note 20). Representative 1H–15N dipolar lineshapes are illustrated for reassembled thioredoxin in Fig. 8. 5. Extract the CSA and dipolar tensor parameters from the 3D ROCSA-NCA and 3D DIPSHIFT-NCA experiments, by numerical simulations of the experimental lineshapes in

17

Solid-State NMR Spectroscopy of Protein Complexes

323

Fig. 8. Dynamics information extracted from 3D-ROCSA and 3D-DIPSHIFT experiments in 1–73(U-13C,15N)/74–108(U-15N) reassembled thioredoxin. (a) Representative 15N CSA lineshapes; the fit values are: G21, ds = 34 ± 2 ppm, h = 1.0 ± 0.25; R73, ds = 75 ± 5 ppm, h = 0.16 ± 0.11; V25, ds = 90 ± 4 ppm, h = 0.20 ± 0.10; T8, ds = 97 ± 4 ppm, h = 0.20 ± 0.08. (b) Chemical shift anisotropy ds plotted as a function of the residue number; (c) Representative 15N–1H dipolar lineshapes; (d) Dipolar order parameters <S> plotted as a function of the residue number. Reproduced from ref. 28 with permission from the American Chemical Society.

SIMPSON (63) or SPINEVOLUTION (64) to find the best fit to the experimental results. 3.10. Solid-State NMR Experiments for Structural Analysis of Protein Interfaces

1. The REDOR block (80) is incorporated into a family of NMR pulse sequences to differentiate between 15N nuclei in two distinct environments. REDOR reintroduces the dipolar coupling between 15N and 13C (or between 13C and 1H) nuclei, which would otherwise be suppressed by MAS. Therefore, in the differentially enriched 1–73(U-13C,15N)/74–108(U-15N) thioredoxin reassembly the 15N or 1H magnetization in the U-13C,15N-enriched fragment can be selectively dephased during the REDOR period by the reintroduced 15N–13C or 1H–13C heteronuclear dipolar coupling, respectively, while the 15N or 1H magnetization in the U-15N fragment is retained allowing for subsequent polarization transfer through the interface or within the 15N fragment resulting in either intermolecular or intramolecular isotopically edited correlations, depending on the desired information (see Note 21).

324

S. Sun et al.

2. In the REDOR–PAINCP sequence, the 13C–15N REDOR period is introduced after the initial 1H–15N CP. The residual unwanted 13C transverse magnetization excited by REDOR is removed by the 15N Z-filter, after which only the 15N magnetization of the 15N-enriched fragment is retained. After the 15N t1 chemical shift evolution period, the 15N magnetization is transferred to 13C using the heteronuclear 15N–13C PAINCP (11, 76) step followed by the detection of 13C chemical shift evolution in the t2 period. Since only one of the two fragments (U-15N,13C) is 13C enriched, the 15N–13C cross peaks represent exclusively intermolecular through interface correlations (see Note 22). 3. In the PDSD–REDOR sequence, a 15N proton-driven spin diffusion (PDSD) mixing period is introduced after the initial 1 H–15N CP step followed by an 15N t1 chemical shift evolution period to establish sequential 15N–15N correlations. The subsequent 13C–15N REDOR dephasing period removes the 15N signals arising from the U-13C,15N-enriched fragment, and therefore, in the final spectrum only sequential 15N–15N correlations from the U-15N labeled fragment can be detected, resulting in considerable spectral simplification due to isotopic editing (see Note 23). 4. In the HETCOR–REDOR sequence, the initial part is the FSLG-based 1H–15N HETCOR experiment (81) employing a flat 1H–15N CP with a short contact time to establish one-bond 1 H–15N correlations between the amide proton and nitrogen atoms in the entire protein. The 15N magnetization arising from the U-13C,15N-enriched fragment is dephased in the subsequent REDOR period, which is introduced after the FSLG-CP part of the sequence. In the final spectrum only 1H–15N correlations from nuclei in the U-15N labeled fragment are detected, resulting in considerable spectral simplification due to isotopic editing (see Note 24). 5. In the REDOR–HETCOR sequence, the 13C-1H REDOR filter is employed for the dephasing of the 1H magnetization from the (U-15N-13C) enriched fragment. Under the experimental conditions, 1H magnetization dephasing is also observed in the part of the (U-15N) enriched fragment constituting the intermolecular interface. Following the t1 evolution under FSLG, the 1H magnetization is transferred to 15N by a flat CP with a short contact time. The 15N signal is detected during the t2 period. The final spectrum contains the 1H–15N correlations arising solely from the residues of the U-15N-enriched fragment, while cross peaks that would be due to the residues constituting the intermolecular interface being either absent or displaying reduced intensity because of their full or partial 13C/1HN REDOR dephasing. A combination of HETCOR/REDOR and REDOR/ HETCOR experiments therefore yields information on the

17

Solid-State NMR Spectroscopy of Protein Complexes

325

intramolecular 1H–15N correlations in the (U-15N) enriched fragment as well as on the 1H–15N correlations of the residues composing the intermolecular interface (see Note 25).

4. Notes 1. Distilled water should be used instead of Millipore pure water for preparing M9 medium. 2. To improve the efficiency of purification, we used chromatography columns purchased from GE healthcare (unless otherwise indicated), which is also the producer of the AKTA FPLC system used in our lab. Columns or resins from other vendors may also work but the procedure details (e.g., buffer conditions) will be different and need to be optimized. 3. To achieve adequate resolution, the size exclusion column packed with Sephadex G-50 should be long enough (>160 cm for separating two fragments and >140 cm for repurification). 4. To obtain SSNMR spectra with narrow lines, it is critical to preserve conformational homogeneity during the preparation of SSNMR samples. Controlled precipitation is a general protocol that allows for generating conformationally homogeneous SSNMR samples of proteins and protein assemblies that are intrinsically soluble. For controlled precipitation, hanging drop screening is performed first to identify the suitable conditions. In our work, we typically employ polyethylene glycol (PEG) of various molecular weights as the precipitant because in our experience, precipitation conditions for virtually any well-behaved protein or protein complex can be successfully established, leading to high-quality samples. In order to pack more protein sample into the MAS rotor, supernatant should be removed from the protein/PEG pellet generated by controlled precipitation. To preserve conformational homogeneity, the pellet should be kept hydrated. 5. Protein expression is induced only for 2 h 30 min to limit shuffling between labeled and unlabeled amino acids. 6. To prepare CA assembly of conical morphology, the final concentration should be 32 mg/mL. Lyophilization of CA enables the direct preparation of CA solution in PEG-20,000 at an initial concentration of 32 mg/mL. 7. Introducing 10 mM Cu(II)-EDTA in the precipitant allows the Cu(II)-EDTA complex to diffuse into the CA sample and to enhance proton longitudinal relaxation, thus permiting shorter recycle delays in the NMR experiments under very fast-MAS conditions (MAS frequencies of 40 kHz or greater) (82, 83).

326

S. Sun et al.

8. The spherical assemblies are not stable in solution, but are stable and retain their morphology for many weeks when dried under N2 gas. 9. When observing conical assemblies with confocal microscopy, an excess of staining solution is desirable. 10. When confocal microscopy is employed for imaging the tubular CA assemblies, the high salt content quenches some of the fluorescence, and a higher receiver gain in the fluorescence channel is needed to get good quality images. 11. The humidity in the ambient environment is critical for the cryo-SEM experiment. Ice tends to accumulate on the surface and cover the details of the structures when humidity is high. 12. The protocol of Marley et al. employs modified M9 medium (89). The timings and the conditions of the individual steps have to be optimized for a specific protein. For CAP-Gly, E. coli cells are grown in LB medium until O.D. at 600 nm reaches 0.8. Cells are pelleted, washed with M9 medium without a nitrogen or carbon source, and then transferred to the M9 growth medium whose volume is a quarter of that of the LB medium culture. After 1 h of recovery, expression of SMT-His6CAP-Gly is induced by addition of IPTG to 0.8 mM. After another 4 h, cells are harvested for protein purification. 13. The His-ULP1 expression system (in E. coli) is a gift from Weill Cornell medical college. ULP1 is a cysteine protease (71). For efficient cleavage by the His6-ULP1 enzyme, DTT is added to the His6-ULP1, SMT-His6-CAP mixture. The final concentration of DTT is 1–5 mM. 14. The CAP-Gly/microtubule ratio is optimized by a co-sedimentation assay (19). 15. Microtubules are fragile protein assemblies and their morphology may be altered during various biochemical manipulations. To prevent shearing of microtubules upon pipetting, 100 mL pipette tips are cut at the sharp end. Prior to and after MAS experiments, microtubule morphologies have to be examined by TEM to ensure that the microtubules remain intact. 16. DARR and PDSD mixing times are strongly magnetic field dependent, and polarization transfer at higher fields is slower. The mixing time is determined experimentally for a specific magnetic field strength. For example, the mixing times for onebond correlations at 14.1 T are ca. 2–10 ms; at 17.6 T ca. 10 ms, and at 21.1 T ca. 50 ms. 17. DREAM is a double-quantum homonuclear recoupling sequence, and the magnetization generated by DREAM is of opposite phase of the original polarization. Therefore for NCACB experiment, the N-Ca SPECIFIC-CP should be carefully optimized to

17

Solid-State NMR Spectroscopy of Protein Complexes

327

avoid two-bond N-Cb magnetization transfer. Cb magnetization generated by nonselective SPECIFIC-CP would cancel signals generated by DREAM. 18. The pulse lengths and power levels are similar to those in experiments for resonance assignments, with additional 15N T1 or T2 filter delays introduced in the corresponding experiments. 19. In our experiments, a C2 21 POST block ( 84 ) is used with (a, b) = (0.0329, 0.467) and one rotor period (100 ms) increment per t1 point. During ROCSA, a 10-ms 13C p pulse with XY-8 phase cycling scheme (85) is introduced in the middle of every rotor period on the 13C channel, and 110 kHz CW decoupling is employed on the 1H channel. 20. In this study, we used the R1817 = {18070180−70}9 element (43) with a 10 kHz MAS spinning frequency. 21. During the 13C–15N Rotational Echo Double Resonance (REDOR) dephasing, 100 kHz 1H TPPM decoupling is employed, and the XY-8 phasing scheme (86) is applied to minimize the resonance offset of the rotor-synchronized p-pulse train. The 13 C and 15N radio frequency field strengths are both 50 kHz. Generally, the REDOR dephasing time needs to be optimized, and under our experimental conditions the dephasing time longer than 6 ms ensured complete suppression of 15N signals from the U-15N, 13C labeled fragment. 22. During the Proton-Assisted Insensitive Nuclei Cross Polarization (PAINCP) transfers (11), the radio frequency field strengths on the 13C and 15N channels are 45 kHz, while the field strength on the 1H channel is optimized for each experiment and is 57–63 kHz. In the 1H–15N heteronuclear correlation experiments, a flat CP with a short contact time of 170 ms is used. 23. The N–N PDSD mixing time utilized in the REDOR–PDSD experiments is 4 s. Under these conditions, almost all of the cross-peaks are from the sequential Ni–Ni–1 correlations, and the cross peak intensities are 10–30% of the corresponding diagonal signals. The cross peaks from Ni–Ni–2 correlations are too weak to be detected. 24. The 1H–1H homonuclear dipolar couplings are suppressed by Frequency-Switched Lee-Goldburg scheme (FSLG) (81) which in the PMLG variant (87) can be implemented by ramping the phase of the proton radio frequency while keeping the proton carrier frequency unchanged. 25. 1HN–13C dipolar interaction is ten times stronger than the 15 N–13C coupling when the internuclear distances are identical. Therefore, the dephasing effect of 13C–1HN REDOR is generally stronger than that of 13C–15N REDOR. Therefore, the 13C–1HN REDOR, in addition to dephasing the HN signals from the

328

S. Sun et al.

U-15N,13C labeled thioredoxin fragment, will give rise to partial dephasing of HN signals belonging to the singly 15N labeled fragment and lining the intermolecular interface of reassembled thioredoxin. Under our experimental conditions where 3.2 ms 1H–13C REDOR dephasing is employed, HN signals from the 15N,13C labeled fragment are eliminated completely, and 1HN signals corresponding to the residues at the interface also disappear. The signals belonging to residues away from the interface are not affected. This experiment allows identification of amino acid residues that constitute the intermolecular interface.

Acknowledgments The projects discussed here are supported by the National Institutes of General Medical Sciences (NIH Grants P50GM082251 and R01GM085306) and the National Center for Research Resources (NIH Grants P20RR017716-07 and P20RR015588). The authors thank Maria Luisa Tasayco, Dabeiba Marulanda, Jun Yang, Marcela Cataldi, Vilma Arriaran for their contributions to the preparation of thioredoxin reassemblies and/or solid-state NMR studies of these reassemblies. References 1. Yool A. J. (2007) Aquaporins: Multiple roles in the central nervous system. Neuroscientist 13, 470–485. 2. Vale R. D. (2003) The molecular motor toolbox for intracellular transport. Cell 112, 467–480. 3. Grunewald K. & Cyrklaff M. (2006) Structure of complex viruses and virus-infected cells by electron cryo tomography. Curr. Opin. Microbiol. 9, 437–442. 4. Klein K. C., Reed J. C., & Lingappa J. R. (2007) Intracellular destinies: Degradation, targeting, assembly, and endocytosis of HIV gag. AIDS Rev. 9, 150–161. 5. Uysal H., et al. (2010) Antibodies to citrullinated proteins: molecular interactions and arthritogenicity. Immunol. Rev. 233, 9–33. 6. Goldbourt A., Gross B. J., Day L. A., & McDermott A. E. (2007) Filamentous phage studied by magic-angle spinning NMR: Resonance assignment and secondary structure of the coat protein in Pf1. J. Am. Chem. Soc. 129, 2338–2344. 7. Hong M. (2007) Structure, topology, and dynamics of membrane peptides and proteins

8.

9.

10.

11.

12.

from solid-state NMR Spectroscopy. J. Phys. Chem. B 111, 10340–10351. Lange A., et al. (2006) Toxin-induced conformational changes in a potassium channel revealed by solid-state NMR. Nature 440, 959–962. Porcelli F., Buck-Koehntop B. A., Thennarasu S., Ramamoorthy A., & Veglia G. (2006) Structures of the dimeric and monomeric variants of magainin antimicrobial peptides (MSI-78 and MSI-594) in micelles and bilayers, determined by NMR spectroscopy. Biochemistry 45, 5793–5799. Zheng Z., Yang R., Bodner M. L., & Weliky D. P. (2006) Conformational flexibility and strand arrangements of the membrane-associated HIV fusion peptide trimer probed by solidstate NMR spectroscopy. Biochemistry 45, 12960–12975. Lewandowski J. R., De Paepe G., & Griffin R. G. (2007) Proton assisted insensitive nuclei cross polarization. J. Am. Chem. Soc. 129, 728–729. Chimon S. & Ishii Y. (2005) Capturing intermediate structures of Alzheimer’s beta-amyloid,

17

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

Solid-State NMR Spectroscopy of Protein Complexes

A beta(1–40), by solid-state NMR spectroscopy. J. Am. Chem. Soc. 127, 13472–13473. Jaroniec C. P., et al. (2004) High-resolution molecular structure of a peptide in an amyloid fibril determined by magic angle spinning NMR spectroscopy. Proc. Natl. Acad. Sci. USA 101, 711–716. Petkova A. T., et al. (2004) Solid state NMR reveals a pH-dependent antiparallel beta-sheet registry in fibrils formed by a beta-amyloid peptide. J. Mol. Biol. 335, 247–260. Shewmaker F., Wickner R. B., & Tycko R. (2006) Amyloid of the prion domain of Sup35p has an in-register parallel beta-sheet structure. Proc. Natl. Acad. Sci. USA 103, 19754–19759. Siemer A. B., et al. (2006) Observation of highly flexible residues in amyloid fibrils of the HET-s prion. J. Am. Chem. Soc. 128, 13224–13228. Tycko R. (2006) Molecular structure of amyloid fibrils: insights from solid-state NMR. Q. Rev. Biophys. 39, 1–55. Han Y., et al. (2010) Solid-State NMR Studies of HIV-1 Capsid Protein Assemblies. J. Am. Chem. Soc. 132, 1976–1987. Sun S. J., Siglin A., Williams J. C., & Polenova T. (2009) Solid-State and Solution NMR Studies of the CAP-Gly Domain of Mammalian Dynactin and Its Interaction with Microtubules. J. Am. Chem. Soc. 131, 10113–10126. Etzkorn M., Bockmann A., Lange A., & Baldus M. (2004) Probing molecular interfaces using 2D magic-angle-spinning NMR on protein mixtures with different uniform labeling. J. Am. Chem. Soc. 126, 14746–14751. Marulanda D., et al. (2004) Magic angle spinning solid-state NMR spectroscopy for structural studies of protein interfaces. Resonance assignments of differentially enriched Escherichia coli thioredoxin reassembled by fragment complementation. J. Am. Chem. Soc. 126, 16608–16620. Yang J., et al. (2007) Magic angle spinning NMR spectroscopy of thioredoxin reassemblies. Magn. Reson. Chem. 45, S73-S83. Castellani F., et al. (2002) Structure of a protein determined by solid-state magic-angle-spinning NMR spectroscopy. Nature 420, 98–102. Hong M. & Jakes K. (1999) Selective and extensive 13 C labeling of a membrane protein for solid-state NMR investigations. J. Biomol. NMR 14, 71–74. Muchmore D. C., McIntosh L. P., Russell C. B., Anderson D. E., & Dahlquist F. W. (1989) Expression and nitrogen-15 labeling of proteins for proton and nitrogen-15 nuclear magnetic resonance. Methods Enzymol. 177, 44–73.

329

26. Marulanda D., Tasayco M. L., Cataldi M., Arriaran V., & Polenova T. (2005) Resonance assignments and secondary structure analysis of E. coli thioredoxin by magic angle spinning solid-state NMR spectroscopy. J. Phys. Chem. B 109, 18135–18145. 27. Yang J., Tasayco M. L., & Polenova T. (2008) Magic angle spinning NMR experiments for structural studies of differentially enriched protein interfaces and protein assemblies. J. Am. Chem. Soc. 130, 5798–5807. 28. Yang J., Tasayco M. L., & Polenova T. (2009) Dynamics of Reassembled Thioredoxin Studied by Magic Angle Spinning NMR: Snapshots from Different Time Scales. J. Am. Chem. Soc. 131, 13690–13702. 29. Franks W., Kloepper K., Wylie B., & Rienstra C. (2007) Four-dimensional heteronuclear correlation experiments for chemical shift assignment of solid proteins. J. Biomol. NMR 39, 107–131. 30. Andrew E. R., Bradbury A., & Eades R. G. (1958) Nuclear Magnetic Resonance Spectra from a Crystal Rotated at High Speed. Nature 182, 1659–1659. 31. Schaefer J., McKay R. A., & Stejskal E. O. (1979) Double-cross-polarization NMR of solids. J. Magn. Reson. 34, 443–447. 32. Baldus M., Petkova A. T., Herzfeld J., & Griffin R. G. (1998) Cross polarization in the tilted frame: assignment and spectral simplification in heteronuclear spin systems. Mol. Phys. 95, 1197–1207. 33. Szeverenyi N. M., Sullivan M. J., & Maciel G. E. (1982) Observation of spin exchange by two-dimensional fourier transform 13C cross polarization-magic-angle spinning. J. Magn. Reson. 47, 462–475. 34. Takegoshi K., Nakamura S., & Terao T. (2001) 13C-1H dipolar-assisted rotational resonance in magic-angle spinning NMR. Chem. Phys. Lett. 344, 631–637. 35. Verel R., Baldus M., Ernst M., & Meier B. H. (1998) A homonuclear spin-pair filter for solid-state NMR based on adiabatic-passage techniques. Chem. Phys. Lett. 287, 421–428. 36. Bennett A. E., et al. (1998) Homonuclear radio frequency-driven recoupling in rotating solids. J. Chem. Phys. 108, 9463–9479. 37. Hohwy M., Rienstra C. M., Jaroniec C. P., & Griffin R. G. (1999) Fivefold symmetric homonuclear dipolar recoupling in rotating solids: Application to double quantum spectroscopy. J. Chem. Phys. 110, 7983–7992. 38. Ernst M., Detken A., Bockmann A., & Meier B. H. (2003) NMR spectra of a microcrystalline

330

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

S. Sun et al. protein at 30 kHz MAS. J. Am. Chem. Soc. 125, 15807–15810. Chen L. L., et al. (2007) J-based 2D homonuclear and heteronuclear correlation in solid-state proteins. Magn. Reson. Chem. 45, S84-S92. Chen L., et al. (2006) Constant-Time Through-Bond 13C Correlation Spectroscopy for Assigning Protein Resonances with SolidState NMR Spectroscopy. J. Am. Chem. Soc. 128, 9992–9993. Chen L., et al. (2007) Backbone assignments in solid-state proteins using J-based 3D Heteronuclear correlation spectroscopy. J. Am. Chem. Soc. 129, 10650–10651. Griffin R. G. (1998) Dipolar recoupling in MAS spectra of biological solids. Nat. Struct. Biol. 5 Suppl, 508–512. Zhao X., Edén M., & Levitt M. H. (2001) Recoupling of heteronuclear dipolar interactions in solid-state NMR using symmetry-based pulse sequences. Chem. Phys. Lett. 342, 353–361. Ladizhansky V. (2009) Homonuclear dipolar recoupling techniques for structure determination in uniformly 13C-labeled proteins. Solid State Nucl. Magn. Reson. 36, 119–128. Balayssac S. p., Bertini I., Lelli M., Luchinat C., & Maletta M. (2007) Paramagnetic Ions Provide Structural Restraints in Solid-State NMR of Proteins. J. Am. Chem. Soc. 129, 2218–2219. Nadaud P. S., Helmus J. J., Kall S. L., & Jaroniec C. P. (2009) Paramagnetic Ions Enable Tuning of Nuclear Relaxation Rates and Provide LongRange Structural Restraints in Solid-State NMR of Proteins. J. Am. Chem. Soc. 131, 8108–8120. Xu X., et al. (2009) Intermolecular dynamics studied by paramagnetic tagging. J. Biomol. NMR 43, 247–254. Lian L.-Y. & Middleton D. A. (2001) Labelling approaches for protein structural studies by solution-state and solid-state NMR. Prog. Nucl. Magn. Reson. Spectrosc. 39, 171–190. Schubert M., Manolikas T., Rogowski M., & Meier B. H. (2006) Solid-state NMR spectroscopy of 10% 13C labeled ubiquitin: spectral simplification and stereospecific assignment of isopropyl groups. J. Biomol. NMR 35, 167–173. Delaglio F., et al. (1995) NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293. Mobli M., Maciejewski M. W., Gryk M. R., & Hoch J. C. (2007) Automatic maximum entropy spectral reconstruction in NMR. J. Biomol. NMR 39, 133–139. Johnson B. A. & Blevins R. A. (1994) NMR View: A computer program for the visualization and

53. 54.

55.

56.

57.

58.

59.

60.

61.

62. 63.

64.

65.

66.

67.

analysis of NMR data. J. Biomol. NMR 4, 603–614. Goddard T. D. & Kneller D. G. Sparky 3 (University of California, San Francisco). Vranken W. F., et al. (2005) The CCPN data model for NMR spectroscopy: development of a software pipeline. Proteins 59, 687–696. Kraulis P. J. (1989) ANSIG: A program for the assignment of protein 1H 2D NMR spectra by interactive computer graphics. J. Magn. Reson. 84, 627–633. Matsuki Y., Eddy M. T., & Herzfeld J. (2009) Spectroscopy by Integration of Frequency and Time Domain Information for Fast Acquisition of High-Resolution Dark Spectra. J. Am. Chem. Soc. 131, 4648–4656. Laue E. D., Skilling J., Staunton J., Sibisi S., & Brereton R. G. (1985) Maximum Entropy Method in Nuclear Magnetic Resonance Spectroscopy. J. Magn. Reson. 62, 437–452. Hoch J. C., Stern A. S., Donoho D. L., & Johnstone I. M. (1990) Maximum-Entropy Reconstruction of Complex (Phase-Sensitive) Spectra. J. Magn. Reson. 86, 236–246. Barna J. C. J., Laue E. D., Mayger M. R., Skilling J., & Worrall S. J. P. (1987) Exponential Sampling, an Alternative Method for Sampling in Two-Dimensional NMR Experiments. J. Magn. Reson. 73, 69–77. de Bouregas F. S. & Waugh J. S. (1992) ANTIOPE, a program for computer experiments on spin dynamics. J. Magn. Reson. 96, 280–289. Smith S. A., Levante T. O., Meier B. H., & Ernst R. R. (1994) Computer Simulations in Magnetic Resonance. An Object-Oriented Programming Approach. J. Magn. Reson., Ser A 106, 75–105. Blanton W. B. (2003) BlochLib: a fast NMR C++ tool kit. J. Magn. Reson. 162, 269–283. Bak M., Rasmussen J. T., & Nielsen N. C. (2000) SIMPSON: A General Simulation Program for Solid-State NMR Spectroscopy. J. Magn. Reson. 147, 296–330. Veshtort M. & Griffin R. G. (2006) SPINEVOLUTION: A powerful tool for the simulation of solid and liquid state NMR experiments. J. Magn. Reson. 178, 248–282. Erickson-Viitanen S., et al. (1989) Cleavage of HIV-1 gag polyprotein synthesized in vitro: sequential cleavage by the viral protease. AIDS Res. Hum. Retroviruses 5, 577–591. Langsetmo K., Fuchs J., & Woodward C. (1989) Escherichia coli Thioredoxin Folds into 2 Compact Forms of Different Stability to Urea Denaturation. Biochemistry 28, 3211–3220. Tasayco M. L. & Chao K. (1995) NMR study of the reconstitution of the beta-sheet of

17

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

Solid-State NMR Spectroscopy of Protein Complexes

thioredoxin by fragment complementation. Proteins 22, 41–44. Slaby I. & Holmgren A. (1975) Reconstitution of Escherichia coli thioredoxin from complementing peptide fragments obtained by cleavage at methionine-37 or arginine-73. J. Biol. Chem. 250, 1340–1347. Sambrook J., Fritsch E. F., & Sambrook J. (1989) Molecular cloning: a laboratory manual (Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.) 2nd Ed. Byeon I. J., et al. (2009) Structural convergence between Cryo-EM and NMR reveals intersubunit interactions critical for HIV-1 capsid function. Cell 139, 780–790. Mossessova E. & Lima C. D. (2000) Ulp1SUMO Crystal Structure and Genetic Analysis Reveal Conserved Interactions and a Regulatory Element Essential for Cell Growth in Yeast. Mol. Cell 5, 865–876. Bennett A. E., Rienstra C. M., Auger M., Lakshmi K. V., & Griffin R. G. (1995) Heteronuclear Decoupling in Rotating Solids. J. Chem. Phys. 103, 6951–6958. Chan J. C. C. & Tycko R. (2003) Recoupling of chemical shift anisotropies in solid-state NMR under high-speed magic-angle spinning and in uniformly 13C-labeled systems. J. Chem. Phys. 118, 8378–8389. Wylie B. J., Franks W. T., & Rienstra C. M. (2006) Determinations of 15N Chemical Shift Anisotropy Magnitudes in a Uniformly 15 N,13C-Labeled Microcrystalline Protein by Three-Dimensional Magic-Angle Spinning Nuclear Magnetic Resonance Spectroscopy. J. Phys. Chem. B 110, 10926–10936. Munowitz M., Aue W. P., & Griffin R. G. (1982) Two-dimensional separation of dipolar and scaled isotropic chemical shift interactions in magic angle NMR spectra. J. Chem. Phys. 77, 1686–1689. Hohwy M., Jaroniec C. P., Reif B., Rienstra C. M., & Griffin R. G. (2000) Local structure and relaxation in solid-state NMR: Accurate measurement of amide N-H bond lengths and H-N-H bond angles. J. Am. Chem. Soc. 122, 3218–3219. Hong M., Yao X., Jakes K., & Huster D. (2002) Investigation of Molecular Motions by LeeGoldburg Cross-Polarization NMR Spectroscopy. J. Phys. Chem. B 106, 7355–7364.

331

78. van Rossum B. J., de Groot C. P., Ladizhansky V., Vega S., & de Groot H. J. M. (2000) A method for measuring heteronuclear (1H-13C) distances in high speed MAS NMR. J. Am. Chem. Soc. 122, 3465–3472. 79. Lorieau J. L. & McDermott A. E. (2006) Conformational flexibility of a microcrystalline globular protein: order parameters by solidstate NMR spectroscopy. J. Am. Chem. Soc. 128, 11505–11512. 80. Gullion T. & Schaefer J. (1989) Rotationalecho double-resonance NMR. J. Magn. Reson. 81, 196–200. 81. Bielecki A., Kolbert A. C., & Levitt M. H. (1989) Frequency-switched pulse sequences: Homonuclear decoupling and dilute spin NMR in solids. Chem. Phys. Lett. 155, 341–346. 82. Wickramasinghe N. P., Kotecha M., Samoson A., Past J., & Ishii Y. (2007) Sensitivity enhancement in (13)C solid-state NMR of protein microcrystals by use of paramagnetic metal ions for optimizing (1)H T(1) relaxation. J. Magn. Reson. 184, 350–356. 83. Wickramasinghe N. P., et al. (2009) Nanomolescale protein solid-state NMR by breaking intrinsic 1HT1 boundaries. Nat. Methods 6, 215–218. 84. Carravetta M., Edén M., Zhao X., Brinkmann A., & Levitt M. H. (2000) Symmetry principles for the design of radiofrequency pulse sequences in the nuclear magnetic resonance of rotating solids. Chem. Phys. Lett. 321, 205–215. 85. Holl S. M., McKay R. A., Gullion T., & Schaefer J. (1990) Rotational-echo triple-resonance NMR. J. Magn. Reson. 89, 620–626. 86. Gullion T., Baker D. B., & Conradi M. S. (1990) New, compensated Carr-Purcell sequences. J. Magn. Reson. 89, 479–484. 87. Vinogradov E., Madhu P. K., & Vega S. (1999) High-resolution proton solid-state NMR spectroscopy by phase-modulated Lee-Goldburg experiment. Chem. Phys. Lett. 314, 443–450. 88. Erickson-Viitanen S., Manfredi J., Viitanen P., Tribe D. E., Tritch R., Hutchison C. A., 3rd, Loeb D. D., & Swanstrom R. (1989) Cleavage of HIV-1 gag polyprotein synthesized in vitro: sequential cleavage by the viral protease. AIDS Res. Hum. Retroviruses 5, 577–591. 89. Marley J., Lu M., & Bracken C. (2001) A method for efficient isotopic labeling of recombinant proteins. J. Biomol. NMR 20, 71–75.

Chapter 18 Synthesis, Purification, and Characterization of Single Helix Membrane Peptides and Proteins for NMR Spectroscopy Miki Itaya, Ian C. Brett, and Steven O. Smith Abstract Membrane proteins function as receptors, channels, transporters, and enzymes. These proteins are generally difficult to express and purify in a functional form due to the hydrophobic nature of their membrane spanning sequences. Studies on membrane proteins with a single membrane spanning helix have been particularly challenging. Single-pass membrane proteins will often form dimers or higher order oligomers in cell membranes as a result of sequence motifs that mediate specific transmembrane helix interactions. Understanding the structural basis for helix association provides insights into how these proteins function. Nevertheless, nonspecific association or aggregation of hydrophobic membrane spanning sequences can occur when isolated transmembrane domains are reconstituted into membrane bilayers or solubilized into detergent micelles for structural studies by solid-state or solution NMR spectroscopy. Here, we outline the methods used to synthesize, purify, and characterize single transmembrane segments for structural studies. Two synthetic strategies are discussed. The first strategy is to express hydrophobic peptides as protein chimera attached to the maltose binding protein. The second strategy is by direct chemical synthesis. Purification is carried out by several complementary chromatography methods. The peptides are solubilized in detergent for solution NMR studies or reconstituted into model membranes for solid-state NMR studies. We describe the methods used to characterize the reconstitution of these systems prior to NMR structural studies to establish if there is nonspecific aggregation. Key words: Membrane protein, NMR spectroscopy, Gp55-P, Epo receptor, Transmembrane

1. Introduction Membrane proteins containing single membrane-spanning helices are involved in a broad range of cellular functions, such as signal transduction, cellular mobility, and apoptosis. Despite the importance of these proteins, relatively few structures have been reported. For protein crystallography, the single transmembrane (TM) helix presents a problem because it must remain embedded in a membrane environment. The strategy for structural studies has often Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_18, © Springer Science+Business Media, LLC 2012

333

334

M. Itaya et al.

been to separately crystallize the extracellular and intracellular domains of these proteins and assume that the TM sequence forms a passive tether. In contrast, nuclear magnetic resonance (NMR) spectroscopy has emerged as an effective tool for probing the membrane-spanning elements of membrane proteins (1–3). In addition, it is increasingly recognized that single TM helices can mediate biological function by associating in specific orientations. Solid-state NMR spectroscopy is well suited for structural studies of TM proteins reconstituted into lipid bilayers (4–8) and has been used to study helix–helix interfaces in single TM helix dimers (9–11) as well as to determine their structure and orientation with respect to the bilayer normal (12–14). Solution NMR spectroscopy has been used to solve the structures of several membrane proteins (15–23), and the structures of single TM helix dimers and higher order oligomers have been deposited in the Protein Data Bank (24). We discuss the methods that have been developed for the synthesis, purification, and characterization of isolated TM helices using two different membrane proteins. The first membrane protein is the murine erythropoietin (Epo) receptor (P14753). The Epo receptor belongs to the cytokine receptor family and is responsible for the production and development of red blood cells. The second membrane protein is gp55-P, a viral membrane protein from the murine spleen focus-forming virus (SFFV) that interacts with and activates the murine Epo receptor (25–27). Both proteins are thought to exist as TM-mediated homodimers (28–30). For each protein, constructs of varying lengths are produced, ranging from peptides containing only the TM domain to constructs that include the TM domain and portions of the intracellular or extracellular domains. For the Epo receptor, the longest peptide sequence (EpoR218–368) used includes the TM domain (residues 226–248) and two intracellular regions (the switch region and Box 1/2 regions) that are important for activity. The switch region (residues 249–256) and Box 1/2 region (residues 257–265, 303–312) are conserved across the cytokine receptors. The orientation of the switch region controls receptor activity (31), while residues in Box 1 are involved in binding the JAK2 kinase (32). 1.1. Synthesis of Peptides and Proteins with Single TM Helices

One of the limiting factors of membrane protein structure determination is the increased difficulty of membrane protein production and purification. Many different production and purification schemes have been described for structural studies (22, 33–37). For short membrane protein sequences, two particularly useful methods are solid-phase peptide synthesis (38–40) and expression in E. coli (41–43).

1.1.1. Solid-Phase Peptide Synthesis

The synthesis of membrane peptides by conventional solid-phase methods can be challenging due to their hydrophobic nature.

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

335

During synthesis, the peptides tend to aggregate on the resin and consequently the efficiency of the reactions needed to sequentially add amino acids is reduced. Some of the concerns involving hydrophobic peptide synthesis have previously been discussed (40, 44). We list below several tips for improving the yield of the desired peptide. 1. Incorporate charged amino acids at the C terminus. Typically, single membrane spanning sequences are terminated by a series of basic amino acids. Because chemical synthesis starts with the C terminus and progresses toward the N terminus, the inclusion of these amino acids in the synthesis helps to extend the growing peptide from the surface of the resin. 2. Use a low density of reactive sites on the resin. The density of peptides being synthesized on the solid resin support is determined by the density of the first amino acid coupled to the resin. A lower density of peptide helps to prevent aggregation. For hydrophobic peptide synthesis, 0.2 mmol of reactive sites/ gram of the resin (or less) are typically used. 3. Use strong amino acid activating reagents. Peptide aggregation lowers the efficiency of the reactions for elongating hydrophobic peptides (44). The efficiency of these reactions can be improved by using a combination of HOAt and HATU in place of the more commonly used reagents O-benzotriazoleN , N , N ¢ , N ¢ -tetramethyl-uronium-hexafluoro-phosphate (HBTU) and hydroxybenzotriazole (HOBt) or dicyclohexylcarbodiimide (DCC) and HOBt (45–47). 4. Incorporation of pseudoproline. Amino acids such as serine, threonine, and cysteine are difficult to incorporate into hydrophobic peptides during the elongation step. Insertion of pseudoprolines as temporary side chain protection helps to reduce peptide self-association and β-sheet formation during synthesis (48). The native amino acid is regenerated by acid treatment and ring-opening upon cleavage from the resin. 1.1.2. Protein Overexpression

Many recombinant protein expression methods have been developed, but the basic differences are in the properties of the vector that govern the purification method (affinity tag) and the mechanism of separation of the affinity tag from the protein of interest (49). Common affinity tags are the hexahistidine (His), glutathione-S-transferase (GST), and Flag tags. Incubating the cellular lysate with a modified resin that interacts with the tag allows purification of the protein of interest. These tags, while useful for expression and purification of soluble proteins, have had mixed success when used with hydrophobic proteins. More recent approaches focus on combining a traditional affinity tag with another tag that promotes solubility of the otherwise hydrophobic fusion protein. For example, maltose binding protein (MBP) is

336

M. Itaya et al.

thought to act as a chaperone or to recruit chaperones to promote the solubility of the hydrophobic fusion peptide (50, 51). Initial attempts at membrane protein expression and purification used fusion proteins with a simple affinity tag, such as a His tag, but neglected a mechanism for removal of the tag. This approach is undesirable because solubility and success of expression are variable (43, 52). Also, residual His tags can drive protein oligomerization (53). Proteolytic separation of the protein of interest from the affinity/solubility tag can be accomplished using a protease, such as thrombin, factor Xa, enterokinase, or the tobacco etch virus (TEV) protease. A less common method of chemical proteolysis makes use of cyanogen bromide, which cleaves the peptide backbone at methionine residues. Recently, a ligation-independent cloning (LIC) vector encoding a His tag with a TEV protease site (54) has been adapted to include MBP for the expression of otherwise insoluble or TM domain containing constructs (21). A major difference between this and other purification methods is that the protein of interest remains soluble and does not get shuttled into inclusion bodies. One drawback is that there are three non-native residues (SNA) at the N terminus of the protein after TEV protease cleavage. We have successfully used this vector to express and purify single TM containing constructs of varying lengths. 1.2. Purification of TM Peptides and Proteins 1.2.1. Peptide Purification

Reverse-phase HPLC is a widely used technique for purifying hydrophobic peptides (55). The columns used are generally made by attaching hydrocarbon chains of different lengths to a solid support. The chain lengths (e.g., C4, C8, C18) characterize the type of column. Columns with short hydrocarbon chain lengths generally exhibit weaker interactions with the long hydrophobic peptide being purified. The consequence is that the resolution is poorer, but the yield of peptide is higher. Typically, the yield of pure peptide per crude peptide weight is ~10–20% with a C4 column (40). The contaminants for expressed and chemically synthesized peptides are different. For chemically synthesized peptides, the major protein contaminants are peptides that are one to two amino acids shorter than the target sequence. In these peptides, the coupling reaction failed at one or more steps. For expressed peptides, the contaminants are generally other protein products with larger differences in molecular weight than the target peptide. Different hydrophobic peptides often require different solvent conditions to achieve optimal separation. However, a water–acetonitrile gradient is often a good starting point (40). The gradient starts with a low concentration of acetonitrile in water, and the acetonitrile concentration is increased in a linear fashion during the purification (see Note 1). We present, in Subheading 3.2.1, a general strategy for HPLC purification of a crude peptide mixture from solid phase synthesis by using reverse phase chromatography.

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

1.2.2. Protein Purification

337

Once the fusion protein has been expressed and the TM domaincontaining construct has been separated from the His-MBP tag, the protein of interest must be separated from this mix. Two different methods can be employed to accomplish this: organic extraction (Subheading 3.2.3) and aqueous (Subheading 3.2.4) purification. Constructs that have a high percentage of hydrophobic residues can be solubilized from a dried mixture after TEV protease cleavage using organic solvents. This method depends on two factors: a high hydrophobic content of the protein of interest and low solubility in organic solvents of the other contaminants (His-TEV, His-MBP, His-MBP-fusion protein). A drawback of this method is that the denaturation of the protein of interest is likely. In contrast, the aqueous method leaves the protein of interest in its natively folded state and uses the His tag present on all contaminants to remove them from solution. To follow the purification and cleavage reactions, we use SDS-PAGE (Fig. 1) and mass spectrometry (Fig. 2). Sample conditions require that detergent which is present in excess when purifying membrane proteins; therefore, most of the preceding steps are done at relatively high detergent concentration. Because the signals from the detergent alkyl groups will be overwhelming in some NMR experiments, it would be preferable to have the sample in a deuterated detergent. It is impractical to conduct the purification in deuterated detergents because of the expense, but it is possible to exchange the purified sample into a deuterated detergent. Detergent exchange can lead to longer protein relaxation times. However, this is generally not appreciable.

Fig. 1. Analysis of protein expression, cleavage, and purification using 15% SDS-PAGE. The ~48-kDa fusion protein (HisMBP EpoR218–268) is the major protein in the lysate. The eluted protein is nearly pure and, when exposed to TEV protease for 22 h, almost 100% cleavage occurs, resulting in a 5.9-kDa band corresponding to EpoR218–268 (see boxed area). NEB MWM NEB broad range molecular weight markers.

M. Itaya et al.

a

b M

His-MBP EpoR

66 kDa

His-MBP His-TEV

27 kDa

Lane 1

1000

1

20 kDa

TEV Intensity (AU)

338

14 kDa

800 His-MBP 600 400 200 0 20

c

40

60

80

100 120

(103)

m/z M

1

2

d

300

Lane 2

27 kDa 20 kDa 14 kDa

EpoR(218-368)

Intensity (AU)

66 kDa 200 EpoR

+2 100

0 10

20

30

40

50 (103)

m/z

Fig. 2. Characterization of a typical aqueous purification by 15% SDS-PAGE and mass spectrometry. (a) SDS-PAGE gel from a typical purification. Lane 1 contains the TEV cleavage mixture ~24 h after the addition of the TEV protease. The components in the sample are His-TEV protease (27 kDa), His-MBP (43 kDa), and EpoR218–368 (17.4 kDa). (b) MALDI-TOF mass spectrum (large MW window) of the TEV cleavage mixture showing peaks for the His-TEV protease and His-MBP. (c) SDSPAGE gel of the same sample mixture from (a) after the aqueous purification described in the text. Lanes 1 and 2 contain 1 and 5 μL of sample, respectively. (d) MALDI-TOF mass spectrum (small MW window) of the EpoR218–368 peptide after aqueous purification. The 17.4-kDa band is clearly visible without contaminants from His-MBP or His-TEV protease. The mass chromatograms shown here were obtained using a Bruker AutoFlex II MALDI-TOF–TOF mass spectrometer.

Detergent exchange can be accomplished by binding the protein to an ion exchange (IEX) column and extensively washing in a buffer with the deuterated detergent (56). 1.3. Characterization of NMR Samples 1.3.1. Mass Spectrometry

Mass spectrometry (MS) (Subheading 3.3) is used to verify the molecular weight and purity of the final purified sample. Figure 2 presents a combination of SDS-PAGE gels and mass spectra to show the usefulness of mass spectrometry for assaying the purity of expressed hydrophobic TM peptides after the removal of His-MBP and His-TEV by aqueous extraction. Mass spectrometry is also routinely used for assessing the purity of peptides produced by solid phase synthesis and organic extraction.

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

339

1.3.2. Size Exclusion Chromatography

One of the pitfalls in the purification of membrane proteins is aggregation. While centrifugation can remove large aggregates from solution, microaggregrates and oligomers can remain and are undesirable for solution NMR spectroscopy. Changing the type or the amount of detergent used for solubilization can disrupt microaggregates. Size exclusion chromatography (SEC) can optimize the peptide solubilization by providing a picture of the aggregation state. Figure 3 shows that adding more detergent disperses microaggregates of the EpoR218–268 peptide.

1.3.3. Polarized Attenuated Total Reflection Fourier Transform Infrared Spectroscopy

Attenuated total reflection (ATR)-Fourier transform infrared (FTIR) spectroscopy is used to characterize the global secondary structure and TM orientation of the reconstituted peptide (57). The amide I vibrational bands observed between 1,600 and 1,700cm-1 are sensitive to secondary structure. Helical secondary structure yields an amide I vibration between 1,650 and 1,660cm-1, while aggregated strand and β-sheet structure yields amide I vibrations below ~1,645cm-1. For membrane spanning helices, the dichroic ratio of the amide I vibration can be used to calculate the helix orientation relative to the membrane normal (57), and consequently provides a rapid assay to characterize whether the peptide is properly inserted into the membrane bilayer (58). The purified bands from sucrose gradients containing peptide inserted with a TM orientation generally yield dichroic ratios for the helical amide I band at ~1,655cm-1 of greater than ~3. Bands from sucrose gradients containing aggregated peptide exhibit dichroic

a

b 80

EpoR(218-268) aggregates

10

A 220 (mAU)

A 220 (mAU)

15

EpoR(218-268) dimer

60 40 20

5

0

0 8

10 12 14 16 18 20 22 Elution Volume (mL)

8

10 12 14 16 18 20 22 Elution Volume (mL)

Fig. 3. Fast protein liquid chromatography. (a) FPLC analysis of the EpoR218–268 peptide solubilized in DPC at 20× CMC shows that the peptide elutes in the first fractions from the gel filtration column, indicating the presence of molecular aggregates. (b) FPLC analysis of the same sample as in (a) after the addition of DPC to 100× CMC and sonication for 1 min shows that the peptide elutes at the molecular weight corresponding to a dimer. The FPLC separations were performed with a Superdex 200 10/300 GL column calibrated with known standards. The buffer for the samples and standards contained 100 mM sodium phosphate and 150 mM NaCl at pH 7.0, and the injected sample volume was 0.1 mL. SDS-PAGE confirmed the identity of the elution fractions (data not shown).

340

M. Itaya et al. 0.2

Absorbance

0.15

0.1

0.05

0 1800

1760

1720

1680

Wavenumber

1640

1600

(cm–1)

Fig. 4. Polarized ATR-FTIR spectroscopy. Polarized ATR-FTIR spectra were obtained of the gp55-P peptide reconstituted into DMPC bilayers by detergent dialysis using IR light polarized parallel (solid line) and perpendicular (dotted line) relative to the bilayer normal. Only the region between 1,600cm-1 and 1,800cm-1 is shown. The amide I vibration is observed at 1,654cm-1, indicating that the reconstituted peptide has an α-helical conformation. The intense vibration at 1,735cm-1 is due to the C=O stretching vibration of the lipid acyl chains. The dichroic ratio (A║/A┴) of the amide I band of 3.3 corresponds to a helix orientation of ~20°. The FTIR spectrum shown here was obtained using a Bruker IFS 66 V/S spectrometer.

ratios of less than ~3. Figure 4 presents the amide I region from the polarized FTIR spectrum of gp55-P reconstituted into dimyristoylphosphatidylcholine (DMPC) bilayers. There is a single amide I band observed at a frequency of 1,654cm-1, characteristic of helical secondary structure. The dichroic ratio of the 1,654cm-1 vibration is 3.3, which corresponds to a tilt of the helix axis of ~20° relative to the membrane normal. Together, the frequency and dichroic ratio of the amide I vibration indicate that the gp55-P peptide is properly reconstituted in a homogeneous α-helical conformation. 1.3.4. Circular Dichroism Spectroscopy

Circular dichroism (CD) spectroscopy is a widely used technique to deduce protein secondary structure (59). CD can inform us about the helical content of the sample, confirming that the sample is folded or refolded properly. FTIR and CD spectroscopy provide complementary information on secondary structure. CD can easily distinguish random coil from α-helical and β-sheet secondary structure. Light scattering from membrane vesicles often leads to a red shift and damping of the CD bands making it more difficult to distinguish the signature bands for α-helix and β-sheet. In contrast, α-helix and β-sheet are easily distinguished by FTIR spectroscopy, but they are more difficult to distinguish from random coil. Figure 5 presents the CD spectrum of the pure EpoR218–268 peptide in dodecylphosphocholine (DPC) at a detergent concentration corresponding to 1.4 times its CMC. It exhibits negative CD absorption bands at approximately 208 and 222 nm characteristic of α-helical secondary structure. The peptide sequence includes

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

341

30

Ellipticity (millideg)

20

10

0 200

220

240

260

280

–10 Wavelength (nm)

Fig. 5. Circular dichroism spectroscopy. The CD spectrum is shown of pure EpoR218–268 peptide solubilized in DPC at 1.4× CMC. The spectrum was taken at room temperature in 10 mM sodium phosphate buffer at pH 7.0. Minima are observed at approximately 208 and 222 nm, which are characteristic of α-helical secondary structure and in agreement with the expected global secondary structure of the EpoR218–268 sequence. The CD spectrum shown here was obtained using an Olis RSM 1000 CD spectrophotometer.

the TM domain and two intracellular regions (the switch region and Box 1) and we expect to observe global α-helical structure for a well-reconstituted sample. 1.4. NMR Spectroscopy 1.4.1. Solid-State 2H NMR Spectroscopy

Deuterium NMR spectroscopy is well suited for probing dynamic processes in membrane proteins (60–62). Deuterium NMR has been widely used to look at lipid dynamics, but much less so to investigate side chain dynamics in membrane proteins because of sensitivity and selectivity issues (63, 64). However, by combining magic angle spinning (MAS) with specifically deuterated membrane proteins and peptides, the sensitivity issue can be resolved (61). We use deuterium MAS NMR to assess homo-oligomerization of the reconstituted membrane peptides prior to undertaking structural studies with more advanced NMR methods. There are several advantages of deuterium MAS NMR. The first is that it is a simple experiment, using a single pulse at relatively low MAS frequencies. The spinning side bands in slow MAS spectra reveal the envelope of the static 2H lineshape. Second, the measurements are carried out in the liquid crystalline phase (above the lipid phase transition temperature). Third, the experiments are comparative in nature. Figure 6 presents deuterium MAS NMR spectra of gp55-P selectively labeled at four consecutive leucines (Leu396–399) in the middle of the TM region. Comparison of the deuterium MAS side band patterns shows that Leu399 has the most restricted motion and consequently is likely to be oriented toward the TM dimer interface. Leu397 and Leu398 exhibit the narrowest deuterium side band patterns. The narrow lineshapes correspond to

342

M. Itaya et al. Facing lipids

Facing dimer interface

Leu397

Leu396

397

L

398

396

L

399

L

L

399

L

Leu398

Leu399

397

L

398

L

396

L gp55-p –20 –10

0

10

20

–20 –10

Frequency (kHz)

0

10

20

Frequency (kHz) Restricted motion of side chain

Mobile side chain RRPPWFTTLISTIMGSLIILLLLLILLIWTLYS

Fig. 6. Deuterium MAS NMR spectroscopy. Gp55-P has five consecutive leucines in the middle of the TM region that allow us to map out the dimer interface by using deuterium MAS NMR (9). Four different gp55-P TM peptides were chemically synthesized, each with one of four sequential leucines methyl deuterium labeled. By examining the deuterium spectrum of each peptide, we can infer which leucines are in the helix dimer interface and which ones face the lipid side chains. Comparison of lineshapes of deuterated leucines clearly show that Leu396 and Leu399 are most likely in the dimer interface and Leu397 and Leu398 oriented away from the dimer interface.

increased mobility relative to Leu399. These leucines are likely facing the surrounding lipids. The modulation of leucine deuterium lineshape around one turn of the gp55-P TM helix argues that the helix is dimerizing in membrane bilayers and not forming nonspecific aggregates. Together, the results from polarized IR and deuterium NMR presented in Figs. 4 and 6 indicate that the gp55-P TM peptide can be reconstituted as a helical membrane-spanning dimer, which is now suitable for high-resolution solid-state NMR structural studies. 1.4.2. Solution-State NMR Spectroscopy

After production of a pure sample, preliminary scouting experiments must be performed to determine sample conditions that promote both long-term sample stability and sample homogeneity. Several factors that must be considered are buffer type and concentration, detergent type and concentration, pH, and temperature for data collection. These variables are covered well elsewhere (2, 65). Solubility/stability assessments may be carried out with unlabeled protein. However, the proton-nitrogen heteronuclear single quantum correlation (15N-HSQC) experiment should be used to assess the resolution and the sensitivity of particular sample conditions; this requires 15N-labeled protein. These experiments can be conducted at a relatively low protein concentration (~0.1 mM),

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

343

ppm 112 113 114 115 116 117 15

N 118 119 120 121 122 123 124 125 8.8

8.6

8.4

8.2

7.8

8.0

7.6

7.4

7.2

ppm

1

H

Fig. 7. Solution NMR spectroscopy. The 1H-15N HSQC spectrum of the EpoR220–248 peptide is shown. The peptide was prepared by the aqueous extraction method and solubilized in d38-DPC (100× CMC). The spectrum was collected with 32 scans at a temperature of 313 K. About 30 N-H peaks are visible, which are expected from the 31 amino acid construct, along with the pair of N-H resonances from the single Asn residue side chain.

but a general rule is that most expected peaks should be visible after an 8-scan experiment; otherwise the time required for threedimensional NMR experiments will make these experiments impractical. Once optimal conditions for data collection have been identified, proteins with different labeling schemes can be produced for a variety of solution studies. Figure 7 presents the 15 N-HSQC spectrum of the TM domain of the Epo receptor (EpoR220–248). There are 31 residues in the 15N-labeled peptide, and the observation of ~30 N-H peaks of roughly the same intensity indicates that the sample is suitable for three-dimensional NMR measurements. Here, we describe methods to produce, purify, and characterize peptides and proteins that are either specifically or fully labeled with 13 C and 15N for both solid-state and solution NMR spectroscopy.

2. Materials The experimental protocols described below require access to equipment typically found in a laboratory conducting recombinant protein expression and instrumentation used in protein characterization, which is often located in laboratories involved in biophysics. Equipment needed for protein expression includes shaker incubators, centrifuges, a French press, bath sonicator, and a setup

344

M. Itaya et al.

for gel electrophoresis. Instrumentation needed for protein characterization includes spectrometers for circular dichroism (CD), FTIR, and NMR spectroscopy. 2.1. Protein Overexpression

1. His-MBP TEV expression vector (21, 66). 2. Escherichia coli BL21 DE3 cells. 3. Luria-Bertani (LB) broth: 10 g tryptone, 10 g NaCl, 5 g yeast extract per liter of water. Adjust to pH 7.0 with 5N NaOH. Add ampicillin to a final concentration of 100 μg/mL. 4. M9 salts: 2.5 g of NaCl, 15 g of KH2PO4, 5 g of NH4Cl, 64 g of Na2HPO4·7H2O, per 1 L sterile deionized water. 15NH4Cl can be substituted in the same amount as unlabeled NH4Cl to express uniformly labeled protein. 5. M9 medium: 750 mL of sterile deionized water, 200 mL of autoclaved M9 salts, 20 mL of 20% glucose, 2 mL of 1 M MgSO4, 0.1 mL of 1 M CaCl2. Ampicillin is added to a final concentration of 100 μg/mL. 6. Labeled M9 medium: Same as M9 medium except 15NH4Cl and U13C-glucose are substituted in the same amounts as unlabeled materials to express uniformly labeled protein. Ampicillin is added to a final concentration of 100 μg/mL. 7. Isopropyl β-D-1-thiogalactopyranoside (IPTG). 8. 20% (w/v) glucose: Filter sterilize. 9. 1 M MgSO4. 10. 1 M CaCl2. 11. Binding buffer: 20 mM Tris–HCl, pH 8.0 (1.95 g/L Tris–HCl and 0.92 g/L Tris-Base, adjust pH with 1N NaOH or 1N HCl), 500 mM NaCl, 5 mM imidazole.

2.2. Purification of TM Peptides and Proteins

1. C4 Reversed-phase semipreparatory HPLC column: 200 Å C4 5 μm (Higgins Analytical).

2.2.1. Peptide Purification

2. 2,2,2 Trifluoroethanol, 99% (TFE). 3. Trifluoroacetic acid (TFA). 4. Solvent A: Water containing 0.1% TFA. 5. Solvent B: Acetonitrile containing 0.1% TFA (see Note 2). 6. Liquid nitrogen.

2.2.2. Protein Purification, Organic and Aqueous Extractions

1. Binding buffer (Subheading 2.1). 2. Wash buffer: 20 mM Tris–HCl, pH 8.0 (1.95 g/L Tris–HCl and 0.92 g/L Tris-Base, adjust pH with 1N NaOH or 1N HCl), 500 mM NaCl, 20 mM imidazole. 3. Elution buffer: 20 mM Tris–HCl, pH 8.0 (1.95 g/L Tris–HCl and 0.92 g/L Tris-Base, adjust pH with 1N NaOH or 1N HCl), 500 mM NaCl, 500 mM imidazole.

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

345

4. n-Octyl-β-D-glucopyranoside (β-OG): Critical micelle concentration (CMC) = 0.25 mM. 5. Ni+/NTA resin. 6. n-Dodecyl-β-D-maltopyranoside (DDM): CMC = 0.015 mM. 7. His-tagged TEV (His-TEV) protease (Invitrogen) or plasmids for in-house expression available. 8. Trichloroacetic acid. 9. MilliQ water (18.2 mW water). 10. Methanol:chloroform mixture: 90% methanol:10% chloroform. 11. 0.2-μm PTFE syringe filters. 12. Dialysis membrane 1 kDa MWCO. 13. Amicon spin concentrator (3 kDa MWCO). 2.2.3. Detergent Exchange

1. Detergent exchange dialysis buffer: 20 mM Tris–HCl, pH 8.5 (0.884 g/L Tris–HCl and 1.744 g/L Tris-Base). 2. Detergent exchange start buffer: 20 mM Tris–HCl, pH 8.5 (0.884 g/L Tris–HCl and 1.744 g/L Tris-Base), 3.08 mM d38-DPC (2× CMC of 1.54 mM). 3. Detergent exchange elution buffer: 20 mM Tris–HCl, pH 8.5 (0.884 g/L Tris–HCl and 1.744 g/L Tris-Base), 3.08 mM d38-DPC, 1 M NaCl. 4. IEX column: 1 mL Q-Sepharose FF anion exchange column (GE Healthcare). 5. Amicon spin concentrator (3 kDa MWCO).

2.3. Mass Spectroscopy

1. Saturated sinapinic acid matrix solution: Dissolve ~20 mg/mL in 50% acetonitrile and 0.1% TFA. Spin down excess matrix by centrifugation at 15,000 × g for 10 s at room temperature and use only the supernatant (saturated sinapinic acid solution). 2. Washing buffer: 10 mM ammonium phosphate, monobasic dissolved in 0.1% TFA. 3. Recrystallization buffer: 60% ethanol, 30% acetone, and 10% water with 0.1% TFA.

2.4. NMR Sample Preparation

1. β-OG (Subheading 2.2.2).

2.4.1. Reconstitution into Membrane Lipids

3. TFE.

2. DMPC (Avanti Polar Lipids). 4. Liquid nitrogen. 5. Phosphate dialysis buffer (1×): 10 mM sodium phosphate, 50 mM NaCl, pH 7.0 (prepared using 100 mM sodium phosphate dibasic, 500 mM NaCl and titrate with 100 mM sodium phosphate monobasic, 500 mM NaCl to adjust the pH. This concentration corresponds to a 10× stock solution).

346

M. Itaya et al.

6. MES rehydration buffer (for a peptide sequence that contains cysteine): 5 mM 2-(N-morpholino)ethanesulfonic acid (MES), 50 mM NaCl, 5 mM dithiothreitol. Adjust the pH to 6.2 with 10N NaOH. 7. MES dialysis buffer (for a peptide sequence that contains cysteine): 5 mM MES, 50 mM NaCl. Adjust the pH to 6.2 with 10N NaOH. 8. Sucrose. 9. Deuterium depleted water (Cambridge Isotope Laboratories). 2.4.2. Solubilization in Deteregnt Micelles

1. NMR buffer: 10 mM sodium phosphate, pH 7.0 (prepared by mixing 5.77 mL of 1 M Na2HPO4 and 4.23 mL of 1 M NaH2PO4 with 990 mL of milliQ water). 2. Argon or nitrogen gas. 3. Deuterium oxide (D2O). 4. 4,4-Dimethyl-4-silapentane-1-sulfonic acid (DSS): 0.1–0.25 mM. 5. Sodium azide (NaN3). 6. Bath sonicator.

3. Methods 3.1. Protein Overexpression

1. Inoculate 25 mL of LB broth with a single colony of E. coli BL21 (DE3) transformed with the His-MBP-TEV expression vector containing the TM clone, isolated from a plate or taken from a frozen glycerol stock solution (see Note 3). Grow overnight at 37°C. 2. Pellet the cells by centrifuging at 6,000 × g for 20 min at 4°C, wash twice with M9 medium, and inoculate 1 L of labeled M9 medium with the resuspended cell pellet (see Note 4). Grow at 37°C at 200 rpm until the optical absorbance at 600 nm (A600) reaches 0.5–0.8, then reduce the temperature to 23°C and induce by adding IPTG to a final concentration of 0.4 mM. Continue incubating for 12–16 h, then pellet cells by centrifuging at 4,000 × g for 30 min at 4°C. 3. Resuspend the cell pellet in 10 mL of binding buffer. Freeze at −20°C until ready for extraction.

3.2. Purification of TM Peptides and Proteins 3.2.1. Peptide Purification

1. Dissolve the chemically synthesized crude peptide in TFE. If it does not promptly dissolve, use a minimum amount of TFA and dilute immediately with TFE (40). 2. Equilibrate the C4 reversed phase column with 5% acetonitrile containing 0.1% TFA (95% solvent A + 5% solvent B) (see Note 5).

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

347

3. Inject the sample as prepared in step 1. 4. Elute the peptide using a linear gradient that varies the acetonitrile concentration from the starting concentration to 95% solvent B over 45 min at a flow rate of 2.5 mL/min, then clean the column for 10 min with 95% solvent B. 5. Monitor the elution by the optical absorbance at 220 nm (peptide backbone) and 260–280 nm (aromatic amino acids). Collect fractions across the major peaks. Assess the purity by using analytical HPLC and MALDI-TOF mass spectrometry (see Subheading 3.3). 6. Measure the concentration of pure peptide by absorbance or by amino acid analysis. 7. Freeze with liquid nitrogen and lyophilize (see Note 6), then store at −20°C. 3.2.2. Protein Purification

1. Lyse the cells using one of several methods (e.g., French press, cell homogenizer). 2. Clarify the lysate by centrifuging at 25,000 × g for 25 min at 4°C. 3. Transfer the supernatant to a tube and add β-OG to ~2× its CMC (CMC = 0.25 mM), nutate ~5 min at room temperature to dissolve. 4. Apply the supernatant to a Ni+/NTA column (12.5 mL bed volume) previously equilibrated with binding buffer, nutate (in column) for 2–4 h at 4°C to bind the fusion protein, and then allow the lysate to flow through. 5. Wash with 14 column volumes of wash buffer, collecting all fractions for SDS-PAGE analysis. Figure 1 shows that most of the contaminating protein is removed in the flow-through and the first wash. 6. Elute the column with 3 column volumes of elution buffer, collecting 1 mL fractions. Measure the optical absorbance at 280 nm (A280), combine the fractions that have protein, measure the A280 of that mixture (for a rough determination of the protein concentration). Add DDM to 10× its CMC, and dissolve by sonication or nutation. 7. Add an appropriate amount of His-TEV protease (see Note 7), nutate for 16–24 h at 23°C to cleave. Monitor the cleavage by the loss of the band corresponding to the His-MBP-fusion protein and the appearance of two lower molecular weight bands corresponding to His-MBP and the TM peptide on SDS-PAGE (see Fig. 1, lane 10). At this point, further purification proceeds using either the organic extraction or the aqueous extraction method.

348

M. Itaya et al.

3.2.3. Organic Extraction

1. Precipitate the protein in the TEV protease mixture (Subheading 3.2.2) by adding trichloroacetic acid to a final concentration of 6% (w/v). 2. Centrifuge at 9,000 × g for 20 min at 4°C to collect the precipitate. Decant the supernatant, wash with milliQ water, centrifuge again, decant the supernatant; repeat once. Lyophilize to remove water. 3. Nutate the dried pellet for 2 h at room temperature with 10 mL of 90% methanol/10% chloroform. 4. Filter the supernatant through a 0.2-μm syringe filter (PTFE) to remove particulate matter. Continue to Subheading 3.4 for reconstitution of organic solvent purified samples. 5. Assess the purity of the protein by SDS-PAGE analysis and mass spectrometry as described in Fig. 2.

3.2.4. Aqueous Extraction

1. Dialyze the sample overnight against binding buffer containing DDM at 1× its CMC to remove the imidazole and glycerol from the Ni+/NTA purification and TEV cleavage, respectively (1 kDa MW cutoff is appropriate). 2. Nutate the dialyzed sample in a column with a 20-mL bed volume of Ni+/NTA (pre-equilibrated with binding buffer plus 1× CMC of DDM) for 2–4 h at 4°C. 3. Collect the flow-through and wash the beads with 2 column volumes of binding buffer containing DDM at 1× its CMC. Combine the flow-through with the first column volume in the wash step. Elute the column with 6 column volumes of elution buffer. 4. Re-equilibrate the Ni+/NTA with binding buffer. Perform a second incubation using the combined flow-through and washes from above. Again, combine the flow-through and first wash fractions. 5. Dialyze overnight against the final NMR buffer, or if detergent exchange will occur, against the detergent exchange dialysis buffer of choice (see Subheading 3.2.5, step 1). 6. If NMR will be performed directly on this sample, concentrate the result of step 5 using an Amicon spin column (3 kDa MWCO), and verify the purity and molecular weight by SDSPAGE and mass spectrometry (see Fig. 2).

3.2.5. Detergent Exchange

1. Dialyze the aqueous purification sample against detergent exchange dialysis buffer without deuterated detergent (see Note 8). 2. Concentrate the sample to 1–3 mL using an Amicon Ultra spin concentrator. 3. Centrifuge at 20,000 × g for 15 min at 4°C to remove particulate matter or aggregated protein, save the supernatant.

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

349

4. Load the sample on an IEX column pre-equilibrated with detergent exchange start buffer containing deuterated detergent. 5. Wash with 10 column volumes of detergent exchange start buffer, collect 1 mL fractions. 6. Elute with 10 column volumes of detergent exchange elution buffer, collect 1 mL fractions. 7. Check the fractions by SDS-PAGE, combine the fractions containing protein. 3.3. Characterization of Purified Protein

Mass spectrometry is used to verify the molecular weight and purity of the final purified sample. 1. Mix the sample and saturated sinapinic acid matrix solution in a volume ratio of 1:5 to 1:10 (see Note 9). Apply 1–2 μL of solution onto the target plate for MALDI-TOF MS and let it dry at room temperature (67). 2. Wash the target with 5–10 μL of washing buffer. 3. Remove the washing buffer after a few seconds with a pipet and let the liquid evaporate. 4. Apply 0.5–1 μL of recrystallization buffer on the washed spot and let it dry.

3.4. NMR Sample Preparation 3.4.1. Reconstitution into Membrane Lipids

1. Co-solubilize β-OG and DMPC in TFE. The amount of detergent is determined by the final detergent concentration (5% w/v) when the sample is rehydrated in step 5. In general, a 1:50 peptide-to-lipid molar ratio is used (see Note 10). Freeze the solution with liquid nitrogen and remove under vacuum; a small amount of water can be added if a powder cannot be obtained with TFE alone. 2. Dissolve the dry mixture of β-OG and DMPC in a minimum amount of water (~0.5 mL). 3. Dissolve the appropriate amount of peptide in 1 mL of TFE and incubate for 6 h at 37°C (for solid-state NMR experiments, we typically use ~2 μmol peptide). 4. Add β-OG /DMPC drop-wise to the peptide solution while stirring. Then add water drop-wise while stirring the sample until bubbles form. Bubbling can be achieved when the ratio of organic solution to water is between 1:2 and 1:4 (19). After the titration is complete, freeze and lyophilize the sample (see Note 11). 5. Rehydrate the sample with 4 mL of phosphate dialysis buffer for 6 h while stirring slowly at 37°C. If the peptide sequence contains cysteine, use MES rehydration buffer to rehydrate the sample. 6. Dialyze the rehydrated sample against 2 L of phosphate dialysis buffer for 48 h at 37°C. If the peptide sequence contains

350

M. Itaya et al.

cysteine, use MES dialysis buffer. Change the dialysis buffer every 5–12 h. As the sample dialyzes and detergent concentration decreases, the sample will become cloudy. 7. Save 100 μL (~0.3 mM peptide) of the dialyzed sample for FTIR analysis (see Note 12, Subheading 1.3.3 and Fig. 4). Layer the membrane vesicles containing peptides on the surface of the germanium plate. 8. Purify the membrane vesicles containing reconstituted membrane peptides by sucrose gradient ultracentrifugation. Make sucrose gradients (10–40% w/v) using either a gradient maker or by careful layering of different densities (1.4 mL for each layer in 5% increments for a 10–40% gradient for ~10 mL total volume) of sucrose in an ultracentrifuge tube appropriate for a swinging bucket ultracentrifuge rotor. Clear or transparent tubes are preferred. Load the sample from step 6 on the top of the gradient and ultracentrifuge at 150,000 × g for 8–12 h at 15°C. A peptide oriented in a transmembrane fashion can be found in the upper band and the aggregates collect in lower bands or pellet. Collect the upper band and dialyze (repeat step 6) to remove sucrose. 9. For deuterium NMR measurements, centrifuge the dialyzed sample at 228,556 × g for 1 h at 4°C (see Note 13). Discard the supernatant and save the pellet. 10. Freeze and lyophilize the pellet. 11. Rehydrate the pellet with 50% (w/w) deuterium depleted water. 12. Incubate overnight at 37°C. 13. Pack the rehydrated pellet in a 4-mm MAS rotor. 3.4.2. Solubilization in Detergent Micelles

Following the organic extraction procedure (Subheading 3.2.3): 1. Blow down the organic solvent with a fine stream of argon or nitrogen gas to ~2 mL, add the detergent of choice (deuterated or otherwise) to the concentration desired in the final sample volume (~300 μL for Shigemi tubes, 600 μL for regular NMR tubes). 2. Add water drop-wise while stirring the sample; the sample may become cloudy. Stop adding water when the detergent bubbles do not immediately pop after sample agitation (19). 3. Freeze the sample with liquid nitrogen and lyophilize in a low temperature (< −70°C), low pressure (~20 mTor) lyophilizer. The low temperature and pressure are needed for the water– organic solvent mixtures. 4. Dissolve the dried mixture in water (~2 mL), refreeze, and lyophilize to drive off residual organic solvent.

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

351

5. Dissolve in an appropriate volume of water or buffer; add D2O to the final desired concentration. If desired, also add DSS (for spectral referencing) and NaN3 (antimicrobial) to a final concentration of 0.1–0.25 mM and 0.05% w/v, respectively. Following the (Subheading 3.2.4):

aqueous

extraction

procedure

1. Dialyze the sample that results from the detergent exchange (Subheading 3.2.5) into the buffer in which NMR experiments will be run (without the detergent). 2. Concentrate the sample to roughly 500 μL (~300 μL for a Shigemi tube) using Amicon spin concentrators, then add detergent (deuterated or otherwise) to the final desired concentration and sonicate in a bath sonicator until the detergent is dissolved (~60 s). 3. Add D2O for the solvent lock, DSS for spectral referencing, and NaN3 as an antimicrobial, if desired.

4. Notes 1. Using water/formic acid/isopropanol as the mobile phase can dramatically improve resolution (47, 68, 69). However, formic acid can lead to peptide formylation as well as shorten the lifetime of an HPLC column. 2. Adding 1-propanol or 2-propanol to solvent B is often used to increase the resolution. 3. This protocol assumes that the desired His-MBP-TEV expression construct has been prepared as described and transformed into the E. coli expression strain of interest (here, BL21 (DE3) cells). For convenience, we prepare 50% glycerol stocks of our transformants and freeze at −80°C by inoculating 5 mL of LB broth with a single isolated colony from an LB agar plate, growing overnight at 37°C, adding sterile glycerol to 50% (w/v) and then freezing. Scraping the surface of the frozen glycerol stock with a sterile pipet tip is all that is needed to remove a small amount for inoculation of the LB broth. 4. Isotopic labeling in E. coli. There are several protocols available for isotopic labeling that allow incorporation of specifically labeled or deuterated amino acids (70, 71). The protocol described here is for full 13C and 15N incorporation. Other protocols can be substituted at this point. 5. A higher starting concentration of acetonitrile (e.g., 30% solvent B) is used for more hydrophobic peptides.

352

M. Itaya et al.

6. The reverse phase HPLC fractions containing the peptide of interest often have sufficient water to yield a fluffy powder by lyophilization. If the peptide elutes in a high concentration of acetonitrile and dries as a film under vacuum, then a sublimating solvent, such as cyclohexane, can be added to dissolve the film and to repeat the lyophilization procedure. Alternatively, a small amount of water can be added directly to the HPLC elution fraction. 7. TEV protease cleavage. For TEV protease that is expressed and purified in-house, the typical yield is ~0.35 mg/mL (the concentration of eluate off of the column, total yield is 18 mg/L of culture). TEV protease is stored in 50% glycerol, 5 mM DTT, 1 mM EDTA at −20°C until use. Cleavage reactions are set up in a 1:1 v:v ratio consisting of 1 part (2 mg/mL) expressed MBP fusion protein to 1 part (0.175 mg/mL) TEV solution. Regardless of the source of the TEV protease, it is essential to determine experimentally the cleavage efficiency of a particular construct. Some fusions will cleave quickly and efficiently with little TEV protease, some constructs require more TEV protease or require more time. Finally, for some constructs proteolysis may not proceed to completion even after 24 h. 8. IEX/detergent exchange. The efficiency of detergent exchange using an IEX column depends upon strong binding of the protein to the IEX column. Binding is determined by the protein charge, which in turn is determined by the pI of the protein and the pH of the buffer chosen. For instance, EpoR218–268 has an estimated pI of 5.78, so a basic buffer is used for anion exchange chromatography. Tris buffer at a pH of 8.5 yields an overall protein charge of −1.8. This charge allows complete binding of the EpoR218–268 protein to the IEX column and efficient detergent exchange. 9. Sample preparation for mass spectrometry. Sinapinic acid is used in this method and is generally suitable for peptides and proteins larger than 3 kDa. However, other MALDI-TOF matrices are available. Steps 2–4 in this section may be omitted if the results are acceptable. 10. Selection of the optimum protein-to-lipid ratio. There are competing factors that must be considered when selecting the protein-to-lipid ratio for NMR studies on membrane-reconstituted peptides. The sensitivity of the NMR measurement increases as the protein-to-lipid ratio increases. However, increasing peptide concentrations leads to nonspecific aggregation. In a series of control studies using the transmembrane domain of glycophorin A, it was found that the peptide began to nonspecifically aggregate above protein-to-lipid ratios of

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

353

~1:50 (57). A typical ratio of cellular membrane has a peptide:lipid molar ratio of 1:60 (72). 11. Organic solvents for peptide – membrane reconstitution. TFE is often our first choice of an organic solvent for solubilization of hydrophobic membrane peptides, lipids, and detergent. A second choice is hexafluoroisopropanol (HFIP). Both solvents are co-soluble with water and allow good co-mixing with detergent and lipids. The strategy for the reconstitution is to monomerize the lipid in TFE and then to add mixed detergent–lipid micelles in a minimum amount of water to induce the formation of mixed micelles containing peptide. In this procedure, when the mixed detergent–lipid micelles containing peptide are frozen and lyophilized to remove the organic solvent, the peptide has not had the opportunity to nonspecifically aggregate. 12. Step 8 in this section may be omitted if the dichroic ratio of short helical TM peptides (~30 amino acids) is ³3, which represents proper reconstitution with a helical tilt angle of £25° from the bilayer normal. For longer TM peptides, a lower dichroic ratio may be observed as a result of the peptide sequence that is not embedded in the membrane and can adopt other secondary structures and orientations. 13. Water content in solid-state NMR samples. For NMR measurements other than deuterium, the reconstituted membranes containing peptide after step 6 or 8 are pelleted in an ultracentrifuge (e.g., SW60Ti Beckman Coulter rotor, 407,506 × g, 24 h) and then loaded into an NMR rotor as a wet paste. The sample is then typically spun in an MAS rotor at 3–4 kHz for 30 min to further pellet the membranes and remove excess water. This step helps balance the rotor for high-speed MAS experiments. The level of hydration can be measured based on the intensity of the water 1H resonances relative to those of the lipid and peptide. The hydration levels after this procedure are typically in the range of 80–100% (w/w) water. At this level of hydration, lipid phase transition temperatures are not changed.

Acknowledgments This work was supported by NIH-NSF instrumentation grants (S10 RR13889 and DBI-9977553), a grant from the NIH to S.O.S (GM-46732). We gratefully acknowledge the W.M. Keck Foundation for support of the NMR facilities in the Center of Structural Biology at Stony Brook.

354

M. Itaya et al.

References 1. Opella, S. J., and Marassi, F. M. (2004) Structure determination of membrane proteins by NMR spectroscopy. Chem. Rev. 104, 3587–3606. 2. Sanders, C. R., and Sönnichsen, F. (2006) Solution NMR of membrane proteins: Practice and challenges. Magn. Reson. Chem. 44, S24–S40. 3. McDermott, A. (2009) Structure and dynamics of membrane proteins by magic angle spinning solid-state NMR. Ann. Rev. Biophys. 38, 385–403. 4. Ketchem, R. R., Hu, W., and Cross, T. A. (1993) High-resolution conformation of gramicidin A in a lipid bilayer by solid-state NMR. Science 261, 1457–1460. 5. Andronesi, O. C., Becker, S., Seidel, K., Heise, H., Young, H. S., and Baldus, M. (2005) Determination of membrane protein structure and dynamics by magic-angle-spinning solidstate NMR spectroscopy. J. Am. Chem. Soc. 127, 12965–12974. 6. Cady, S. D., Schmidt-Rohr, K., Wang, J., Soto, C. S., DeGrado, W. F., and Hong, M. (2010) Structure of the amantadine binding site of influenza M2 proton channels in lipid bilayers. Nature 463, 689–692. 7. Chu, S. D., Coey, A. T., and Lorigan, G. A. (2010) Solid-state 2H and 15N NMR studies of side-chain and backbone dynamics of phospholamban in lipid bilayers: Investigation of the N27A mutation. Biochim. Biophys. ActaBiomembr. 1798, 210–215. 8. Etzkorn, M., Martell, S., Andronesi, O. C., Seidel, K., Engelhard, M., and Baldus, M. (2007) Secondary structure, dynamics, and topology of a seven-helix receptor in native membranes, studied by solid-state NMR spectroscopy. Angew. Chem. Int. Ed. Engl. 46, 459–462. 9. Liu, W., Crocker, E., Constantinescu, S. N., and Smith, S. O. (2005) Helix packing and orientation in the transmembrane dimer of gp55P of the spleen focus forming virus. Biophys. J. 89, 1194–1202. 10. Smith, S. O., Song, D., Shekar, S., Groesbeek, M., Ziliox, M., and Aimoto, S. (2001) Structure of the transmembrane dimer interface of glycophorin A in membrane bilayers. Biochemistry 40, 6553–6558. 11. Smith, S. O., Smith, C. S., and Bormann, B. J. (1996) Strong hydrogen bonding interactions involving a buried glutamic acid in the transmembrane sequence of the neu/erbB-2 receptor. Nat. Struct. Biol. 3, 252–258. 12. Ramamoorthy, A., and Opella, S. J. (1995) Two-dimensional chemical shift/heteronuclear

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

dipolar coupling spectra obtained with polarization inversion spin exchange at the magic angle and magic-angle sample spinning (PISEMAMAS). Solid State Nucl. Magn. Reson. 4, 387–392. Wang, J. F., Kim, S., Kovacs, F., and Cross, T. A. (2001) Structure of the transmembrane region of the M2 protein H+ channel. Protein Sci. 10, 2241–2250. Kim, M. J., Park, S. H., Opella, S. J., Marsilje, T. H., Michellys, P. Y., Seidel, H. M., and Tian, S. S. (2007) NMR structural studies of interactions of a small, nonpeptidyl Tpo mimic with the thrombopoietin receptor extracellular juxtamembrane and transmembrane domains. J. Biol. Chem. 282, 14253–14261. Bocharov, E. V., Pustovalova, Y. E., Pavlov, K. V., Volynsky, P. E., Goncharuk, M. V., Ermolyuk, Y. S., Karpunin, D. V., Schulga, A. A., Kirpichnikov, M. P., Efremov, R. G., et al. (2007) Unique dimeric structure of BNip3 transmembrane domain suggests membrane permeabilization as a cell death trigger. J. Biol. Chem. 282, 16256–16266. Bocharov, E. V., Mineev, K. S., Volynsky, P. E., Ermolyuk, Y. S., Tkach, E. N., Sobol, A. G., Chupin, V. V., Kirpichnikov, M. P., Efremov, R. G., and Arseniev, A. S. (2008) Spatial structure of the dimeric transmembrane domain of the growth factor receptor ErbB2 presumably corresponding to the receptor active state. J. Biol. Chem. 283, 6950–6956. Roosild, T. P., Greenwald, J., Vega, M., Castronovo, S., Riek, R., and Choe, S. (2005) NMR structure of Mistic, a membrane-integrating protein for membrane protein expression. Science 307, 1317–1321. MacKenzie, K. R., Prestegard, J. H., and Engelman, D. M. (1997) A transmembrane helix dimer: Structure and implications. Science 276, 131–133. Sulistijo, E. S., and MacKenzie, K. R. (2009) Structural basis for dimerization of the BNIP3 transmembrane domain. Biochemistry 48, 5106–5120. Oxenoid, K., and Chou, J. J. (2005) The structure of phospholamban pentamer reveals a channel-like architecture in membranes. Proc. Natl. Acad. Sci. USA 102, 10870–10875. Hu, J., Qin, H., Li, C., Sharma, M., Cross, T. A., and Gao, F. P. (2007) Structural biology of transmembrane domains: Efficient production and characterization of transmembrane peptides by NMR. Protein Sci. 16, 2153–2165. Tamm, L. K., and Liang, B. Y. (2006) NMR of membrane proteins in solution. Prog. Nucl. Magn. Reson. Spectrosc. 48, 201–210.

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

23. Mineev, K. S., Bocharov, E. V., Pustovalova, Y. E., Bocharova, O. V., Chupin, V. V., and Arseniev, A. S. (2010) Spatial structure of the transmembrane domain heterodimer of ErbB1 and ErbB2 receptor tyrosine kinases. J. Mol. Biol. 400, 231–243. 24. White, S. H. (2009) Biophysical dissection of membrane proteins. Nature 459, 344–346. 25. Ruscetti, S. K., Janesch, N. J., Chakraborti, A., Sawyer, S. T., and Hankins, W. D. (1990) Friend spleen focus-forming virus induces factor independence in an erythropoietin-dependent erythroleukemia cell line. J. Virol. 64, 1057–1062. 26. Hoatlin, M. E., Kozak, S. L., Lilly, F., Chakraborti, A., Kozak, C. A., and Kabat, D. (1990) Activation of erythropoietin receptors by Friend viral gp55 and by erythropoietin and down-modulation by the murine Fv-2r resistance gene. Proc. Natl. Acad. Sci. USA 87, 9985–9989. 27. Li, J. P., D’Andrea, A. D., Lodish, H. F., and Baltimore, D. (1990) Activation of cell growth by binding of Friend spleen focus-forming virus gp55 glycoprotein to the erythropoietin receptor. Nature 343, 762–764. 28. Constantinescu, S. N., Keren, T., Socolovsky, M., Nam, H. S., Henis, Y. I., and Lodish, H. F. (2001) Ligand-independent oligomerization of cell-surface erythropoietin receptor is mediated by the transmembrane domain. Proc. Natl. Acad. Sci. USA 98, 4379–4384. 29. Gurezka, R., Laage, R., Brosig, B., and Langosch, D. (1999) A heptad motif of leucine residues found in membrane proteins can drive self-assembly of artificial transmembrane segments. J. Biol. Chem. 274, 9265–9270. 30. Constantinescu, S. N., Keren, T., Russ, W. P., Ubarretxena-Belandia, I., Malka, Y., Kubatzky, K. F., Engelman, D. M., Lodish, H. F., and Henis, Y. I. (2003) The erythropoietin receptor transmembrane domain mediates complex formation with viral anemic and polycythemic gp55 proteins. J. Biol. Chem. 278, 43755–43763. 31. Constantinescu, S. N., Huang, L. J. S., Nam, H. S., and Lodish, H. F. (2001) The erythropoietin receptor cytosolic juxtamembrane domain contains an essential, precisely oriented, hydrophobic motif. Mol. Cell 7, 377–385. 32. Witthuhn, B. A., Quelle, F. W., Silvennoinen, O., Yi, T. L., Tang, B., Miura, O., and Ihle, J. N. (1993) Jak2 associates with the erythropoietin receptor and is tyrosine-phosphorylated and activated following stimulation with erythropoietin. Cell 74, 227–236. 33. Dawson, P. E., and Kent, S. B. H. (2000) Synthesis of native proteins by chemical ligation. Annu. Rev. Biochem. 69, 923–960.

355

34. Loll, P. J. (2003) Membrane protein structural biology: the high throughput challenge. J. Struct. Biol. 142, 144–153. 35. Wang, D. N., Safferling, M., Lemieux, M. J., Griffith, H., Chen, Y., and Li, X. D. (2003) Practical aspects of overexpressing bacterial secondary membrane transporters for structural studies. Biochim. Biophys. Acta-Biomembr. 1610, 23–36. 36. Klammt, C., Löhr, F., Schäfer, B., Haase, W., Dötsch, V., Rüterjans, H., Glaubitz, C., and Bernhard, F. (2004) High level cell-free expression and specific labeling of integral membrane proteins. Eur. J. Biochem. 271, 568–580. 37. Laage, R., and Langosch, D. (2001) Strategies for prokaryotic expression of eukaryotic membrane proteins. Traffic 2, 99–104. 38. Bormann, B. J., Knowles, W. J., and Marchesi, V. T. (1989) Synthetic peptides mimic the assembly of transmembrane glycoproteins. J. Biol. Chem. 264, 4033–4037. 39. Kochendoerfer, G. G., Salom, D., Lear, J. D., Wilk-Orescan, R., Kent, S. B. H., and DeGrado, W. F. (1999) Total chemical synthesis of the integral membrane protein influenza A virus M2: Role of its C-terminal domain in tetramer. Biochemistry 38, 11905–11913. 40. Fisher, L. E., and Engelman, D. M. (2001) High-yield synthesis and purification of an α-helical transmembrane domain. Anal. Biochem. 293, 102–108. 41. Tian, C. L., Karra, M. D., Ellis, C. D., Jacob, J., Oxenoid, K., Sonnichsen, F., and Sanders, C. R. (2005) Membrane protein preparation for TROSY NMR screening. Meth. Enzymol. 394, 321–334. 42. Page, R. C., Moore, J. D., Nguyen, H. B., Sharma, M., Chase, R., Gao, F. P., Mobley, C. K., Sanders, C. R., Ma, L., Sonnichsen, F. D., et al. (2006) Comprehensive evaluation of solution nuclear magnetic resonance spectroscopy sample preparation for helical integral membrane proteins. J. Struct. Funct. Genomics 7, 51–64. 43. Qin, H. J., Hu, J., Hua, Y. Z., Challa, S. V., Cross, T. A., and Gao, F. P. (2008) Construction of a series of vectors for high throughput cloning and expression screening of membrane proteins from Mycobacterium tuberculosis. BMC Biotechnol. 8, 51–59. 44. Kent, S. B. H. (1988) Chemical synthesis of peptides and proteins. Annu. Rev. Biochem. 57, 957–989. 45. Carpino, L. A. (1993) 1-Hydroxy-7Azabenzotriazole – an efficient peptide coupling additive. J. Am. Chem. Soc. 115, 4397–4398. 46. Glover, K. J., Martini, P. M., Vold, R. R., and Komives, E. A. (1999) Preparation of insoluble transmembrane peptides: Glycophorin-A, prion

356

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

M. Itaya et al. (110–137), and FGFR (368–397). Anal. Biochem. 272, 270–274. Heukeshoven, J., and Dernick, R. (1982) Reversed-phase high-performance liquid-chromatography of virus proteins and other large hydrophobic proteins in formic-acid containing solvents. J. Chromatogr. 252, 241–254. Mutter, M., Nefzi, A., Sato, T., Sun, X., Wahl, F., and Wohr, T. (1995) Pseudo-prolines (PsiPro) for accessing inaccessible peptides. Pept. Res. 8, 145–153. Arnau, J., Lauritzen, C., Petersen, G. E., and Pedersen, J. (2006) Current strategies for the use of affinity tags and tag removal for the purification of recombinant proteins. Protein Expr. Purif. 48, 1–13. Kapust, R. B., and Waugh, D. S. (1999) Escherichia coli maltose-binding protein is uncommonly effective at promoting the solubility of polypeptides to which it is fused. Protein Sci. 8, 1668–1674. Nallamsetty, S., and Waugh, D. S. (2006) Solubility-enhancing proteins MBP and NusA play a passive role in the folding of their fusion partners. Protein Expr. Purif. 45, 175–182. Korepanova, A., Gao, F. P., Hua, Y. Z., Qin, H. J., Nakamoto, R. K., and Cross, T. A. (2005) Cloning and expression of multiple integral membrane proteins from Mycobacterium tuberculosis in Escherichia coli. Protein Sci. 14, 148–158. Amor-Mahjoub, M., Suppini, J. P., GomezVrielyunck, N., and Ladjimi, M. (2006) The effect of the hexahistidine-tag in the oligomerization of HSC70 constructs. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 844, 328–334. Stols, L., Gu, M. Y., Dieckman, L., Raffen, R., Collart, F. R., and Donnelly, M. I. (2002) A new vector for high-throughput, ligation-independent cloning encoding a tobacco etch virus protease cleavage site. Protein Expr. Purif. 25, 8–15. Lew, S., and London, E. (1997) Simple procedure for reversed-phase high-performance liquid chromatographic purification of long hydrophobic peptides that form transmembrane helices. Anal. Biochem. 251, 113–116. Fleming, K. G., Ackerman, A. L., and Engelman, D. M. (1997) The effect of point mutations on the free energy of transmembrane alpha-helix dimerization. J. Mol. Biol. 272, 266–275. Smith, S. O., Eilers, M., Song, D., Crocker, E., Ying, W. W., Groesbeek, M., Metz, G., Ziliox, M., and Aimoto, S. (2002) Implications of threonine hydrogen bonding in the glycophorin

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

69.

A transmembrane helix dimer. Biophys. J. 82, 2476–2486. Tamm, L. K., and Tatulian, S. A. (1997) Infrared spectroscopy of proteins and peptides in lipid bilayers. Q. Rev. Biophys. 30, 365–429. Johnson, W. C. (1999) Analyzing protein circular dichroism spectra for accurate secondary structures. Proteins 35, 307–312. Siminovitch, D. J. (1998) Solid-state NMR studies of proteins: the view from static 2H NMR experiments. Biochem. Cell Biol. 76, 411–422. Ying, W. W., Irvine, S. E., Beekman, R. A., Siminovitch, D. J., and Smith, S. O. (2000) Deuterium NMR reveals helix packing interactions in phospholamban. J. Am. Chem. Soc. 122, 11125–11128. Sharpe, S., Barber, K. R., Grant, C. W. M., Goodyear, D., and Morrow, M. R. (2002) Organization of model helical peptides in lipid bilayers: Insight into the behavior of singlespan protein transmembrane domains. Biophys. J. 83, 345–358. Siminovitch, D. J., Ruocco, M. J., Olejniczak, E. T., Das Gupta, S. K., and Griffin, R. G. (1988) Anisotropic 2H-nuclear magnetic resonance spin-lattice relaxation in cerebroside- and phospholipid-cholesterol bilayer membranes. Biophys. J. 54, 373–381. Bloom, M., and Smith, I. C. P. (1985) Manifestations of lipid-protein interactions in deuterium NMR, in Progress in Protein-Lipid Interactions (Watts, A. & De Pont, J. J. H. H. M., Eds.) pp 61–88, Elsevier, Amsterdam. Krueger-Koplin, R. D., Sorgen, P. L., KruegerKoplin, S. T., Rivera-Torres, A. O., Cahill, S. M., Hicks, D. B., Grinius, L., Krulwich, T. A., and Girvin, M. E. (2004) An evaluation of detergents for NMR structural studies of membrane proteins. J. Biomol. NMR 28, 43–57. Nallamsetty, S., and Waugh, D. S. (2007) A generic protocol for the expression and purification of recombinant proteins in Escherichia coli using a combinatorial His6-maltose binding protein fusion tag. Nat. Protoc. 2, 383–391. Karas, M., Bachmann, D., Bahr, U., and Hillenkamp, F. (1987) Matrix-assisted ultraviolet laser desorption of non-volatile compounds. Int. J. Mass Spectrom. Ion Process. 78, 53–68. Bollhagen, R., Schmiedberger, M., and Grell, E. (1995) High-Performance LiquidChromatographic Purification of Extremely Hydrophobic Peptides – Transmembrane Segments. J. Chromatogr. A 711, 181–186. Sato, T., Kawakami, T., Akaji, K., Konishi, H., Mochizuki, K., Fujiwara, T., Akutsu, H., and

18

Synthesis, Purification, and Characterization of Single Helix Membrane¼

Aimoto, S. (2002) Synthesis of a membrane protein with two transmembrane regions. J. Pept. Sci. 8, 172–180. 70. Gardner, K. H., and Kay, L. E. (1998) The use of 2H, 13C, 15N multidimensional NMR to study the structure and dynamics of proteins. Annu. Rev. Biophys. Biomol. Struct. 27, 357–406.

357

71. Goto, N. K., and Kay, L. E. (2000) New developments in isotope labeling strategies for protein solution NMR spectroscopy. Curr. Opin. Struct. Biol. 10, 585–592. 72. Gennis, R. B. (1989) Membrane Dynamics and Protein-Lipid Interactions, in Biomembranes pp 166–198, Springer-Verlag, New York.

Chapter 19 Assignment of Backbone Resonances in a Eukaryotic Protein Kinase – ERK2 as a Representative Example Andrea Piserchio, Kevin N. Dalby, and Ranajeet Ghose Abstract A first step toward the analysis of the structure, dynamics, and interactions of proteins by NMR is obtaining an acceptable level of resonance assignments. This process is nontrivial in most eukaryotic kinases given their size and suboptimal behavior in solution. Using inactive ERK2 as a representative example, we describe the procedures we utilized to achieve a significant degree of completeness of backbone resonance assignment. Key words: MAP kinase, ERK2, TROSY, Backbone resonance assignment, Selective labeling, Spinlabeled ATP

1. Introduction ERK2 is a member of the extracellular signal-regulated kinase (ERK) subfamily of the mitogen-activated protein kinases (MAPKs). ERKs are upregulated in response to the activation of cell surface receptors mediated by extracellular cues, such as hormones, cytokines, and growth factors (1–3). ERKs play a central role in growth factor-related apoptosis in colorectal cancer (4), making the ERK signaling pathway a key target for cancer therapy (5, 6). The activation of ERKs (ERK1 and ERK2) occurs downstream of the Ras/ Raf pathway upon dual phosphorylation of the conserved 183ThrX-Tyr185 motif by MAP/ERK kinase kinase (MEKK) (7). While the bacterial expression and purification of ERK2, at least in its inactive state (on which we focus here), are more straightforward than some other eukaryotic kinases, e.g., c-Src (see Chapter 7, this volume, (8)), complications in NMR characterization due to the large size and extensive dynamics remain a general trend in most eukaryotic kinases. A necessary step before NMR

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_19, © Springer Science+Business Media, LLC 2012

359

360

A. Piserchio et al.

studies of structure, dynamics, and interactions of these important signaling molecules can be undertaken is to obtain a sufficient number of assignments for backbone resonances. However, standard methodologies (9, 10) that are successfully applied to smaller or more well-behaved systems tend to fail for these important signaling molecules. This would explain why only a few detailed NMR studies on eukaryotic kinases are available in the literature (11–15). Here, we provide a description of the procedures that we applied to obtain backbone resonance assignments for inactive ERK2. These strategies illustrate how similar protocols can be utilized for other protein kinases. It is to be noted, that generation of homogeneous samples of active, dual-phosphorylated (on Thr183 and Tyr185) ERK2 for NMR studies is nontrivial and studies on the active, dual-phosphorylated species will be described elsewhere.

2. Assignment of Backbone Resonances for Full-Length Inactive ERK2

2.1. Standard Triple-Resonance Experiments

The size of ERK2 (42 kDa), its tendency for nonspecific aggregation at concentrations above approximately 200 mM, and dynamics on the slow to intermediate timescale leading to line-broadening effects make the assignment of backbone resonances quite difficult. Given these issues, the resonance assignment procedure for fulllength inactive ERK2 is described below in some detail. For inactive ERK2 (referred to hereon forward as ERK2), TROSYbased experiments (10, 16, 17) consistently displayed significantly narrower line widths, even at fields as low as 600 MHz, and were therefore preferred over their non-TROSY counterparts. However, for samples prepared in D2O-based media, the slow back exchange of several well-protected amide groups resulted in reduced sensitivity in backbone-directed NMR experiments, complicating their analysis. This problem is evident when an 15N-, 1H-TROSY spectrum of perdeuterated ERK2 (prepared in a D2O-based medium) was compared to that of a sample prepared from cells grown in a H2O-based medium supplemented with uniformly 2H-, 15N-labeled amino acids. The latter spectrum displayed additional sets of resonances not visible in the former. Incomplete back exchange complicates resonance assignment both by restricting the number of detectable resonances, and limiting the degree of correlations obtained (complicating the so-called backbone walk) for unambiguous correspondences between the resonances that are observed. This complication adds to the problem of the overall quality of the triple-resonance experiments being poor presumably because of the aforementioned aggregation phenomenon and dynamics. A TROSYHNCO experiment collected at 800 MHz exhibits around 65–70% of the expected peaks. Clearly, this represents the upper limit for

19

Assignment of Backbone Resonances in a Eukaryotic Protein Kinase – ERK2...

361

NMR assignment (for protein prepared in a D2O-based medium, where there is incomplete back exchange of amide protons), given that HNCO is by far the most sensitive triple-resonance experiment. In an HNCACB dataset collected at the same field and requiring a week of acquisition time, Cb (i − 1) peaks were identified for approximately 43% of the expected resonances, if only the resonances also appearing in the HNCO experiment were considered (30% of the expected resonances considering the entire protein). Spectral overlap only partially justifies such an incomplete peak count. An HN(COCA)CB experiment collected at 600 MHz shows 53% of the spin systems visible in the HNCO (38% of the overall expected resonances). Fortunately, the Ca-based experiments were, by comparison, far more complete; the HN(CO)CA (also collected at 600 MHz), for example, includes 90% of the Ca(i − 1) peaks expected from the HNCO-detected spin systems. By combining this with an HNCA (collected at 800 MHz) experiment, most of the expected Ca(i), Ca(i − 1) patterns matching the observed HNCO peaks could be successfully recognized. In addition, roughly 70% (relative to the HNCO-observed resonances) of the intraresidue HN(CA)CO peaks were also found. However, CO- or Ca-based experiments are limited in scope since they allow the identification of only a limited number of amino acid types unlike the Cb-based experiments. Clearly, these statistics indicate that the extent of the NMR assignment achievable by using conventional approaches is fairly limited. Therefore, we relied heavily on an approach that takes into consideration the structural and biochemical features of ERK2. A similar approach has been employed by Langer and coworkers (18) for assignment of the catalytic subunit of protein kinase A (PKA). 2.2. Use of Predicted Chemical Shifts

Several crystal structures of ERK2 can be found in the PDB, in both the inactive (19) and dual-phosphorylated (on T183 and Y185) active (20) forms. A simple way to take advantage of the available structural data is to utilize them to predict the protein NMR chemical shifts. This can be done with reasonable accuracy for 13C resonances, as long as the crystal structure reflects the structure in solution. Toward this purpose, we used the software Sparta (21), freely available from the Bax group. For well-ordered regions, like sheets and helices, these predictions are expected to be more accurate than for loops and regions of noncanonical secondary structure. Discrepancies may also be introduced in highly structured areas by the presence (or by the lack of) ligands or by intermolecular interactions either in solution (aggregation) or in crystallo (crystal-packing forces). The latter scenario is especially true in flexible and highly dynamic molecules (22, 23) as the protein kinases are known to be. It should be noted that in most of the crystal structures of ERK2 available in the PDB it is in complex with a variety of ligands. In addition, the measured chemical shift values

362

A. Piserchio et al.

are also affected by several sources of experimental error, including those resulting from low digital resolution (especially for 13Cb), spectral overlap, poor signal-to-noise ratio, and artifacts introduced due to pulse-sequence imperfections and nonidealities. Therefore, we used a relatively large cutoff (2.3 ppm) for differences between measured and predicted chemical shifts when evaluating a potential match for a 13C resonance for a particular position along the protein sequence. Due to this large uncertainty and the large number of potential matches along the polypeptide chain, any comparison done at the level of individual residues leads to several ambiguous matches and is not particularly informative. If however, the comparison is done using stretches of three (or more) resonance peaks sequentially linked together, the method becomes much more useful. In particular, we found that links typically of four, and sometimes three, residues belonging to well-structured regions were sufficient to assign these resonances. We also found that, when analyzing areas of less well-defined secondary structure, this approach still remains useful when combined with the traditional analysis based on average chemical shift values expected for a given residue type as obtained from the Biological Magnetic Resonance Data Bank (http://www.bmrb.wisc.edu). Often, a link comprising four residues can be assigned to a protein loop if just three of the four residues in a given link correlate favorably to the corresponding predicted chemical shift values, provided that the chemical shifts observed for all four of them are compatible with the expected average database values of residues comprising the sequence. 2.3. Use of Structural Information

Clearly, the assignments obtained using predicted chemical shifts should not be considered reliable until confirmed using more conventional experimental spectroscopy-based approaches. An obvious way to accomplish is to take advantage of the known three-dimensional structure of ERK2, and use the internuclear distances available from them for comparison with cross peaks between amide protons that appear in a three-dimensional 15N-edited NOESY-TROSY experiment. The verification of the existence (or the absence) of specific cross peaks predicted from the crystal structure is an effective way to validate assignments, especially for b-strands and loops. In the case of b-sheets, the expected (and observed) NOEs are mainly long range, allowing confirmation of sequentially nonproximal stretches of residues that comprise individual strands of a b-sheet that have been independently assigned. The reproduction of proper patterns of internuclear distances from incorrectly assigned resonances is highly unlikely. In case of loops, generally only few specific residues would be expected to be wellstructured and generate amide–amide NOEs, so again the observed NOE pattern can be used to confirm a tentative assignment. For helices, however, the expected NOEs are mostly short range and only involve amino acids in the particular helical segment, so they

19

Assignment of Backbone Resonances in a Eukaryotic Protein Kinase – ERK2...

363

can be used principally to distinguish a helical motif from a nonhelical one. Interamide NOEs in helices can nevertheless be used as an aid to, or an alternative for, triple-resonance experiments in order to sequentially link successive amino acid spin systems. Since the early days of protein NMR spectroscopy when heteronuclear labeling was not commonplace, “walking” the sequential NH–NH NOEs represented a simple path to assign resonances corresponding to helical fragments (24). Furthermore, in samples of low proton density (as in the present case), spin diffusion can be utilized to generate excellent medium-range connectivities (i, i + 2; i, i + 3, etc.) within helical stretches. For example, an 15N-edited NOESYTROSY experiment with a long mixing time (400 ms) effectively generates a TOCSY-like pattern among the NH resonances in a tight turn (Fig. 1). This process is greatly simplified by perdeuteration that reduces magnetization transfer to other regions of the protein. This is particularly useful in regions of spectral crowding, when resonance overlaps prevent the unambiguous identification of sequential NOEs at several positions. 2.4. Use of Selective Labeling Strategies

Selective amino acid labeling represents another route to aid in the linking of neighboring spin systems and to fill gaps in assignments when the information content of the triple-resonance experiments, especially in the Cb region, is poor. Usually, selective labeling (25) is performed by supplementing the M9 medium with unlabeled (14 N) ammonium chloride, 1H-12C glucose, or similar nutrients (sometimes, LB is used directly (26)), a particular 15N-labeled (15N, 12 C, 1H) amino acid, and sometimes an unlabeled (14N, 12C, 1H) pool of the remaining amino acids. Then, a simple 15N, 1H HSQC experiment should highlight the amide resonances belonging to the residue type selected. This method can also be more rigorously applied using E. coli strains that are auxotrophic for the specific amino acid to be labeled (27). Unfortunately, this labeling approach did not perform well when applied to ERK2. Independently of the specific amino acid tested, the resulting HSQC spectrum lacked discernable peaks. We attributed this problem to extensive line broadening resulting from efficient 1H–1H relaxation in the absence of deuteration. Reducing the overall 1H density by growing the bacteria in D2O did not significantly improve the quality of the spectra suggesting that this was the result of the contribution of the local dipolar interactions between the amide and alpha protons of the selectively labeled amino acids. This phenomenon leads to an increase in the contribution of the 1H homonuclear R1 to the relaxation rate of the antiphase term between the amide 15N and 1 H nuclei, and results in a broadening of the resonances in an 15N, 1 H-HSQC experiment. We then decided to alter our selective labeling approach and use amino acids selectively 13C-labeled only at the carbonyl position (14N, 12C, 13CO,1H) in a uniformly 15 N-labeled, deuterated background. As shown by Takeuchi and

364

A. Piserchio et al.

Fig. 1. (a) Structure of ERK2 with the N- and C-terminal lobes colored light and dark grey, respectively. The MAP kinase insert and the C-terminal extension are colored black. Side chains for the regulatory T183 and Y185 residues are shown and labeled. Side chains for the tight turn encompassing residues T92-M96 are shown on the structure and expanded on the right panel. (b) Strips taken from an 15N-edited NOESY-TROSY spectrum collected with a 400-ms mixing time at 800 MHz on a uniformly 2H-, 13C-, 15N-labeled inactive ERK2 sample in a buffer containing 150 mM NaCl, 2 mM DTT, 10 mM MgCl2, 2 mM ADP, 50 mM phosphate, pH 6.8, 10% 2H2O. Shown here is the effect of spin diffusion generating long-range connections among the amides of the segment comprising residues T92-M96. The lines highlight the total correlation-like (as in a TOCSY experiment, where transfer occurs through scalar rather than dipolar couplings) effect of the magnetization transfer. The source (first label) and target (last label) amide 1HN nuclei for the cross peaks are labeled. Only a single label is used for the diagonal peaks.

coworkers (28), this 15N-labeled, deuterated background can be achieved by adding 15NH4Cl, 12C-2H, glucose, and a pool of 15N, 12 Cl, 2H amino-acids (CELTONE base powder, Cambridge Isotope Laboratories) to the growth medium and by replacing H2O with

19

Assignment of Backbone Resonances in a Eukaryotic Protein Kinase – ERK2...

365

Fig. 2. 13CO, 1H planes for TROSY-based HNCO spectra for representative examples (Leu, Ala) of residue-selective 13 CO-labeled samples of ERK2 in a uniformly 15N-labeled, perdeuterated background. Also shown in the extreme left panel is the corresponding plane from uniformly 13C-, 2H-, 15N-labeled ERK2. The labels correspond to the residue that contributes the 13CO nucleus (i.e., the i − 1 residue).

D2O. HNCO experiments would then be expected to show peaks at a position corresponding to the selectively labeled carbonyl and the nitrogen of the following residue. While the alpha position of the labeled residues (i − 1) is still protonated, the resonance detected corresponds to the amide (1HN) for the ith residue that carries a deuteron at the Ca position. A further advantage of this approach is the higher resolution offered by the 3D HNCO compared to the extensive resonance overlap seen in 2D HSQC experiments used with the 15N-selective labeling approach. Another piece of information provided by this labeling scheme is the disappearance of the resonances corresponding to the 14N-labeled amino acids (for those selectively 13C labeled at the carbonyl position) that can be monitored using 2D TROSY experiments. We successfully utilized this strategy for Gly, Ala, Leu, Val, and Ile residues in ERK2 (representative examples are shown in Fig. 2). 2.5. Use of SpinLabeled ATP Analogs

Like all protein and indeed nonprotein kinases, ERK2 binds ATP and ADP. However, the chemical shift perturbations induced by binding of these molecules (or corresponding slowly hydrolyzed analogs) are not limited to the ATP-binding pocket; therefore, the shifts of unknown resonances can be difficult to correlate to a specific portion of the structure simply by monitoring chemical shift perturbations. It has already been shown for PKA that spin-labeled ATP (sl-ATP) molecules can be successfully employed to highlight those residues within a certain distance from the nucleotide binding pocket (18). The sl-ATP we employed, sl-N3-ATP (a kind gift from Dr. Pia Vogel, SMU), carries a stable nitroxide spin label as part of a 2,2,5,5 tetramethyl 3-pyrroline scaffold attached to the 3¢ (70–80%) or 2¢ (20–30%) positions of the ribose moiety (29). A crystal structure of ERK2 bound to this specific ligand does not exist; therefore,

366

A. Piserchio et al.

Fig. 3. Paramagnetic relaxation enhancement (PRE) monitored using TROSY-based HNCO experiments. Partial and complete quenching for L114 and K115, respectively, induced by substoichiometric amount of sl-ATP (1:0.25 ratio) are illustrated. Both residues are a distance of ~12 Å from the label.

we relied on the structure of ATP-bound ERK2 (PDB: 1GOL) to estimate distances from the spin label. We estimated that spin-labeled ATP in a one-fourth (or half) substoichiometric amount is capable of significantly quenching the HNCO peaks corresponding to residues within 20–25 Å of the 3¢ ribose position (a representative example is shown in Fig. 3). Given the substoichiometric amounts of sl-ATP used and the low affinity (30) of ATP for inactive ERK2 (KD > ~700 mM), the conformational changes induced by simple ATP binding are expected to be negligible. This approach helped extend the assignments in the area at the interface between the Nand C-lobes of ERK2, a critical region that was difficult to assign by other means. Using these strategies, we have unambiguously assigned ~90% (~65% of all nonproline resonances) of the resonances seen/ resolved to date at 800 MHz. The largest unassigned continuous stretch corresponds to the catalytic segment that can be expected to be in conformational exchange, a phenomenon that would lead to line-broadening effects. We are investigating alternative strategies to obtain assignments for this region, including experiments that allow better visualization of exchange-broadened lines (31).

3. Conclusions We focused on the problem of the NMR backbone assignment of eukaryotic kinases using ERK2 as an example. NMR studies of this class of proteins is hindered by a number of problems, namely, incomplete amide protons back exchange, aggregation/ oligomerization at high protein concentration, and internal dynamics. We have shown here that the careful analysis of otherwise

19

Assignment of Backbone Resonances in a Eukaryotic Protein Kinase – ERK2...

367

well-established NMR experiments that can be normally found in most standard pulse sequence libraries can lead to an acceptable level of resonance assignment. However, this process requires multiple sample conditions (different ligands, various selectively labeled samples, spin-labeled ATP analogs, etc.) and available structural information. In general, we have found that the resonance assignment of the sites of protein–protein interactions (where known, as in ERK2) is significantly less challenging than the highly dynamic regions around the catalytic site. This process of resonance assignment is certainly time and resource consuming, but obtaining these assignments is clearly worthwhile given their utility in investigating protein-protein interactions involving these key signaling molecules, especially in the large number of cases where crystallographic information about the interaction interfaces is not available (32).

Acknowledgments This research has been supported by the following grants from the National Institutes of Health: GM084278 (to RG), GM059802 (to KND), and 5G12 RR03060 (toward partial support of the NMR facilities at The City College of New York). RG is a member of the New York Structural Biology Center, NYSTAR facility. KND is a recipient of a grant from the Welch Foundation (F-1390). The authors thank Dr. Pia Vogel (SMU) for the kind gift of spinlabeled ATP. References 1. Murphy, L. O., and Blenis, J. (2006) MAPK signal specificity: the right place at the right time. Trends Biochem. Sci. 31, 268–275. 2. Chen, Z., Gibson, T. B., Robinson, F., Silvestro, L., Pearson, G., Xu, B., Wright, A., Vanderbilt, C., and Cobb, M. H. (2001) MAP kinases. Chem. Rev. 101, 2449–2476. 3. Pearson, G., Robinson, F., Beers Gibson, T., Xu, B. E., Karandikar, M., Berman, K., and Cobb, M. H. (2001) Mitogen-activated protein (MAP) kinase pathways: regulation and physiological functions. Endocrine Rev. 22, 153–183. 4. Fang, J. Y., and Richardson, B. C. (2005) The MAPK signalling pathways and colorectal cancer. The Lancet Oncol. 6, 322–327. 5. Kohno, M., and Pouyssegur, J. (2006) Targeting the ERK signaling pathway in cancer therapy. Annal. Med. 38, 200–211. 6. Kohno, M., and Pouyssegur, J. (2003) Pharmacological inhibitors of the ERK signaling

7.

8.

9.

10.

pathway: application as anticancer drugs. Prog. Cell Cyc. Res. 5, 219–224. Roux, P. P., and Blenis, J. (2004) ERK and p38 MAPK-activated protein kinases: a family of protein kinases with diverse biological functions. Microbiol. Mol. Biol. Rev. 68, 320–344. Piserchio, A., Dalby, K. N., and Ghose, R. (2012) Expression and Purification of Srcfamily Kinases for Solution NMR Studies. Meth. Mol. Biol. 831, 111–132. Sattler, M., Schleucher, J., and Griesinger, C. (1999) Heteronuclear multidimensional NMR experiments for the structure determination of proteins in solution employing pulsed field gradients. Prog. NMR Spectr. 34, 93–158. Salzmann, M., Pervushin, K., Wider, G., Senn, H., and Wuthrich, K. (1998) TROSY in triple-resonance experiments: new perspectives for sequential NMR assignment of large proteins. Proc. Natl. Acad. Sci. USA 95, 13585–13590.

368

A. Piserchio et al.

11. Masterson, L. R., Mascioni, A., Traaseth, N. J., Taylor, S. S., and Veglia, G. (2008) Allosteric cooperativity in protein kinase A. Proc. Natl. Acad. Sci. USA 105, 506–511. 12. Masterson, L. R., Cheng, C., Yu, T., Tonelli, M., Kornev, A., Taylor, S. S., and Veglia, G. (2010) Dynamics connect substrate recognition to catalysis in protein kinase A. Nature Chem. Biol. 6, 821–828. 13. Wiesner, S., Wybenga-Groot, L. E., Warner, N., Lin, H., Pawson, T., Forman-Kay, J. D., and Sicheri, F. (2006) A change in conformational dynamics underlies the activation of Eph receptor tyrosine kinases. EMBO J. 25, 4686–4696. 14. Vajpai, N., Strauss, A., Fendrich, G., CowanJacob, S. W., Manley, P. W., Grzesiek, S., and Jahnke, W. (2008) Solution conformations and dynamics of ABL kinase-inhibitor complexes determined by NMR substantiate the different binding modes of imatinib/nilotinib and dasatinib. J. Biol. Chem. 283, 18292–18302. 15. Vogtherr, M., Saxena, K., Hoelder, S., Grimme, S., Betz, M., Schieborr, U., Pescatore, B., Robin, M., Delarbre, L., Langer, T., Wendt, K. U., and Schwalbe, H. (2006) NMR characterization of kinase p38 dynamics in free and ligand-bound forms. Angew. Chem. Intl. Ed. Engl. 45, 993–997. 16. Riek, R., Pervushin, K., and Wuthrich, K. (2000) TROSY and CRINEPT: NMR with large molecular and supramolecular structures in solution. Trends Biochem. Sci. 25, 462–468. 17. Pervushin, K. (2000) Impact of transverse relaxation optimized spectroscopy (TROSY) on NMR as a technique in structural biology. Q. Rev. Biophys. 33, 161–197. 18. Langer, T., Vogtherr, M., Elshorst, B., Betz, M., Schieborr, U., Saxena, K., and Schwalbe, H. (2004) NMR backbone assignment of a protein kinase catalytic domain by a combination of several approaches: application to the catalytic subunit of cAMP-dependent protein kinase. ChemBioChem 5, 1508–1516. 19. Zhang, F., Strand, A., Robbins, D., Cobb, M. H., and Goldsmith, E. J. (1994) Atomic structure of the MAP kinase ERK2 at 2.3 Å resolution. Nature 367, 704–711. 20. Canagarajah, B. J., Khokhlatchev, A., Cobb, M. H., and Goldsmith, E. J. (1997) Activation mechanism of the MAP kinase ERK2 by dual phosphorylation. Cell 90, 859–869.

21. Shen, Y., and Bax, A. (2007) Protein backbone chemical shifts predicted from searching a database for torsion angle and sequence homology. J. Biomol. NMR 38, 289–302. 22. Fushman, D., Xu, R., and Cowburn, D. (1999) Direct determination of changes of interdomain orientation on ligation: use of the orientational dependence of 15 N NMR relaxation in Abl SH(32). Biochemistry 38, 10225–10230. 23. Piserchio, A., Nair, P. A., Shuman, S., and Ghose, R. (2010) Solution NMR studies of Chlorella virus DNA ligase-adenylate. J. Mol. Biol. 395, 291–308. 24. Wüthrich, K. (1986) NMR of proteins and nucleic acids, John Wiley and Sons, New York. 25. Muchmore, D. C., McIntosh, L. P., Russell, C. B., Anderson, D. E., and Dahlquist, F. W. (1989) Expression and nitrogen-15 labeling of proteins for proton and nitrogen-15 nuclear magnetic resonance. Meth. Ezymnol. 177, 44–73. 26. Englander, J., Cohen, L., Arshava, B., Estephan, R., Becker, J. M., and Naider, F. (2006) Selective labeling of a membrane peptide with 15 N-amino acids using cells grown in rich medium. Biopolymers 84, 508–518. 27. LeMaster, D. M., and Cronan, J. E., Jr. (1982) Biosynthetic production of 13 C-labeled amino acids with site-specific enrichment. J. Biol. Chem. 257, 1224–1230. 28. Takeuchi, K., Ng, E., Malia, T. J., and Wagner, G. (2007) 1-13 C amino acid selective labeling in a 2H15N background for NMR studies of large proteins. J. Biomol. NMR 38, 89–98. 29. Vogel-Claude, P., Schafer, G., and Trommer, W. E. (1988) Synthesis of a photoaffinity-spinlabeled derivative of ATP and its first application to F1-ATPase. FEBS Lett. 227, 107–109. 30. Prowse, C. N., and Lew, J. (2001) Mechanism of activation of ERK2 by dual phosphorylation. J. Biol. Chem. 276, 99–103. 31. Li, Y., and Palmer, A. G., III. (2010) Narrowing of protein NMR spectral lines broadened by chemical exchange. J. Am. Chem. Soc. 132, 8856–8857. 32. Piserchio, A., Warthaka, M., Devkota, A. K., Kaoud, T. S., Lee, S., Abramczyk, O., Ren, P., Dalby, K. N., and Ghose R. (2011) Solution NMR insights into docking interactions involving inactive ERK2. Biochemistry, 50, 3660–3672.

Chapter 20 Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility Griselda Hernández, Janet S. Anderson, and David M. LeMaster Abstract Electrostatic interactions at the protein–aqueous interface modulate the reactivity of solvent-exposed backbone amides by a factor of at least a billion fold. The brief (~10 ps) lifetime of the peptide anion formed during the hydroxide-catalyzed exchange reaction helps enable the experimental rates to be robustly predictable by continuum dielectric methods. Since this ability to predict the structural dependence of exchange reactivity also applies to the protein amide hydrogens that are only rarely exposed to the bulk solvent phase, electrostatic analysis of the experimental exchange rates provides an effective assessment of whether a given model ensemble is consistent with the properly weighted Boltzmann conformational distribution of the protein native state. Key words: Hydrogen exchange, Protein flexibility, Electrostatics, Conformational distribution, Dielectric shielding, Poisson–Boltzmann, Protein ensemble

1. Introduction Both the flexibility and the conformational dynamics of proteins are generally thought to play critical roles in biological function. Accurate experimental and computational characterization of these properties for any given protein remains challenging. In the equilibrium distribution of the protein native state, every energetically feasible conformation has a nonzero probability. As a result, the quantitative analysis of protein flexibility is synonymous with determining the proper Boltzmann-weighting of this conformational distribution. To effectively compare between experimental measurements and computational modeling on the conformational distribution of the protein native state, several conditions should be met. The computational modeling must be sufficiently detailed so that a Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_20, © Springer Science+Business Media, LLC 2012

369

370

G. Hernández et al.

quantitative structure-based prediction of the observed experimental data can be made. Conversely, given a set of protein conformations against which to test, the experimental method needs to be predictable on the basis of that distribution. If instead, interpretation of the experimental measurements depends not only upon the conformational distribution but also upon the rate of interchange between those conformations, then the computational modeling approach must encompass the complete conformational dynamics of the system. Although a full dynamical analysis is appealing in principle, in practice, the ability to use experimental data to distinguish the degree to which a given dynamical simulation is consistent with physical reality is often significantly decreased. The kinetics of the hydrogen exchange reaction for the amides along the protein backbone have long been interpreted as providing a passive monitor of what fraction of time a given amide hydrogen is directly exposed to the solvent phase. In reality, the reactivity to exchange for a solvent-exposed amide is acutely sensitive to its electrostatic environment. Poisson–Boltzmann continuum dielectric methods offer usefully accurate predictions of those electrostatic environments. Owing to the highly transient peptide anion charge state, the kinetics of hydroxide-catalyzed amide hydrogen exchange provide a “snapshot” of the Boltzmann conformational distribution which is nearly independent from the dynamics of interchange between protein conformations. Not only is hydrogen exchange analysis of the well-exposed backbone amides acutely sensitive to the detailed conformations of the highly populated states, these data also reflect both the frequency and structural detail of the exchange-competent states that arise from rare conformational transitions of the structurally buried amides. As a result, electrostatic analysis of amide hydrogen exchange provides a robust experimental basis upon which to assess the consistency of any given model ensemble with the properly weighted Boltzmann conformational distribution.

2. Steric Interpretation of Protein Hydrogen Exchange 2.1. Hydrogen Exchange as a Measure of Solvent Accessibility

Before the first protein X-ray structure was reported, LinderstrømLang and colleagues (1) described the so-called EX2 analysis of hydrogen exchange from structurally buried backbone amides, as summarized in the following kinetic scheme: kop

kch

closed open → exchanged kcl

If the rate of the closing reaction is rapid compared to the open state chemical exchange step (i.e., kcl > > kch), a preequilibrium of the open and closed conformational states is established and the

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

371

overall exchange rate constant kex equals (kop/kcl) kch, in which kop/kcl is the equilibrium constant for the conformational opening transition. Since that time, the conventional steric interpretation of hydrogen exchange has identified the rate constant kch with the corresponding kinetics of exchange in model peptides under analogous sample conditions (2, 3). The ratio of the observed exchange rate to that of the model peptide defines a protection factor that is assumed to specify the fraction of the population in the open-state and thus a residue-specific free energy [i.e., ΔG = −RT ln(kex/kpep)] for the conformational transition that gives rise to exchange (3). Central to the peptide normalization analysis is the assumption that the residual tertiary structure in the exchange-competent open state does not influence the kinetics of exchange. When applied to the slowest exchanging amides, the peptide normalization analysis has been shown to yield reasonable predictions of global thermodynamic stability for a number of proteins (4), as might be expected when hydrogen exchange is occurring from a conformationally disordered unfolded state. However, under physiological conditions, the amides of well folded proteins that exchange via a global unfolding transition generally constitute only a small fraction of the peptide backbone. A survey of 20 wellstudied proteins identified less than 10% of all backbone amides as such “core” amides (5). The desire to extend the residue-specific stability interpretation of experimental hydrogen exchange data has stimulated the development of a number of structure-based analysis algorithms (6–12) to predict the population of solvent-exposed conformations for each backbone amide. As illustrated by the COREX conformational sampling algorithm (6), the reported success in predicting protein hydrogen exchange rates has been invoked to validate the application of this algorithm to a broad range of questions including the structural propagation of ligand binding effects (13), analysis of the localized energetics of allosteric coupling pathways (14), partitioning of protein structures into high, medium and low thermodynamic stability environments (15) and characterizing the determinants of fold specificity (16) as well as structurally interpreting protein cold denaturation (17), the framework model for folding (18) and pathological protein misfolding transitions (19). Yet when the protection factor predictions given in the initial COREX manuscript (6) were directly compared against the corresponding experimental hydrogen exchange values, no net correlation was observed (20, 21). Although this lack of predictive capability suggests a limited utility for the specific conformational sampling algorithm used, it does not provide an unambiguous test for the validity of the peptide normalization analysis. Since independent experimental data characterizing the properties of the transient, partially ordered conformations that give rise to hydrogen exchange are rarely available, the predicted residue-specific conformational free energies can not generally be directly verified or refuted.

372

G. Hernández et al. 2.0

E53 S25

log kex (s–1)

1.0 K2

0.0

D35

K3

K46

D36

D14 I12

D21

–1.0

S47

K29 K51 V38*

5.0

6.0

7.0

8.0

9.0

10.0

11.0

12.0

pH

Fig. 1. Magnetization transfer-based hydrogen exchange rate measurements on the solvent-exposed amides of P. furiosus A2K rubredoxin. CLEANEX-PM [71, 72] measurements were carried out at 25°C. Dashed lines with slope of 1.0 were drawn for the pH dependent data of each solvent-exposed amide, indicating a simple hydroxide ion dependence on the exchange rates over most of the pH range. The exchange rate value for Val 38, marked with an asterisk, is derived by extrapolation from measurements at 52°C. Reprinted from ref. 22 with permission from the American Chemical Society.

On the contrary, the physical plausibility of the peptide normalization analysis can be straightforwardly examined. Specifically, is it chemically reasonable to assume that exposure of an amide hydrogen to the bulk solvent phase is sufficient to establish exchange kinetics equivalent to those of the corresponding model peptide? This assumption can be directly examined by consideration of the exchange behavior for protein backbone amide hydrogens that are well-exposed to solvent in the high resolution X-ray structure so that no conformational transition is required for the exchange reaction. The well exposed Val 38 amide hydrogen of rubredoxin from Pyrococcus furiosus (22) exchanges at a rate that is nearly 107-fold slower than that of the corresponding Trp-Val model peptide (23). Conversely, His 38 in the active site of the a domain of the human protein disulfide isomerase exchanges at a rate 400-fold faster than the corresponding model peptide value (24). Application of the standard protection factor analysis to these two static solventexposed amides yields a 13 kcal/mol range of apparent conformational stabilities. This range is at least as large as the maximal global stability of any protein predicted from hydrogen exchange measurements, which has been independently verified by either calorimetric or spectroscopic methods (4, 25, 26). Figure 1 illustrates the exchange rates for all of the backbone amide hydrogens of Pyrococcus furiosus rubredoxin that are exposed to solvent in the 1.1 Å resolution X-ray structure (27). At each pH, exchange rates can be accurately quantified over the range of 0.2– 70 s−1 (28). The exchange rates of the individual static solvent accessible (29) amides increase directly with the hydroxide ion

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

373

concentration up to pH 10.77. At the highest pH values, the slope for the amides of Lys 29 and Lys 46 decreases, presumably reflecting the increasingly negative protein charge arising from the neutralization of these amino acid side chains. These data yield hydroxide-catalyzed rate constants kOH− at 25°C ranging from 109.68 M−1 s−1 for Lys 2 to 101.55 M−1 s−1 for Ile 12. The exchange of Val 38 at pH 11.85 and 25°C is too slow to be observed under these conditions by the magnetization transfer method used. Taking advantage that this protein is stable to 94°C at pH 11.6 (30), exchange measurements on the pH 11.85 sample were carried out at elevated temperature and extrapolated to a rate constant of 100.67 M−1 s−1 at 25°C, a billion-fold slower than that of Lys 2. 2.2. Comparing Solvent Accessibility from Molecular Simulations to that Inferred from the Steric Interpretation of Hydrogen Exchange

The physical plausibility of the peptide normalization analysis of hydrogen exchange can also be tested by comparison with molecular simulation studies that strive to predict the correct Boltzmannweighted native state conformational distribution of a protein. Ubiquitin has served as the primary model system for such detailed studies of the conformational ensemble. To facilitate such comparisons, our recent measurements for ubiquitin provide the first reported data set describing the hydroxide-catalyzed exchange rate constants kOH− for every backbone amide of a protein under near physiological solution conditions (20). One such model conformational ensemble has been deposited in the Protein Data Bank by Vendruscolo and colleagues (PDB code 2NR2 (31)) for which molecular simulations of ubiquitin were restrained to match both experimental NOE and NMR relaxation data. More recently, de Groot and colleagues (32) also deposited a molecular dynamics simulation of ubiquitin (PDB code 2K39) that was initially restrained to match the same set of NOE restraints. This set of conformations was then used to iteratively select subsets of conformations that were consistent with experimental residual dipolar coupling data. The 51 amide hydrogens of ubiquitin that become exposed to solvent in at least one of the 144 structures of the 2NR2 ensemble or one of the 116 structures of the 2K39 ensemble were compared to the accessibility predictions derived from protection factor analysis of the amide hydrogen exchange (33) (Fig. 2). When the experimental exchange rate constants for these 51 residues were normalized against the model peptide values to obtain an estimate of the population of exchange-competent conformations for each residue, the fraction of solvent-exposed conformations varies by more than a factor of 107. This variation corresponds to a range in excess of 10 kcal/mol for the apparent residue-specific conformational stabilities ΔGHX. Given the number of conformations included, the molecular dynamics-derived ensembles can only sample fractional accessibilities over a range of ~102. For this set of 51 amides, the fractional accessibility predictions from the ensembles,

G. Hernández et al. β1

β2

α

β3

β4

β5

0 log solvent accessibility

374

–1 –2 –3 –4 –5 –6 –7 –8

0

10

20

30 40 50 residue number

60

70

Fig. 2. The fraction of conformations in which the backbone amide hydrogen is predicted to be exposed to solvent for each residue of ubiquitin. Estimations based on protection factor analysis [3, 23] of hydrogen exchange measurements [20], normalized to model peptide values, are indicated (filled circle). Illustrated as well is the fraction of conformations in the 2NR2 (filled triangle) and 2K39 (filled inverted triangle) NMR-restrained ensembles for which the solvent accessibility of the amide hydrogen is greater than 0.5 Å2. The position of the secondary structure elements of ubiquitin are indicated along the top of the figure. Reprinted from ref. 33 with permission from Elsevier Limited.

as compared to the peptide normalization-based estimates, differ by up to a factor of 105 (ΔΔG ~ 7 kcal/mol). Indeed, 21 of these amides yield ΔGHX values that differ from the molecular simulationderived ensemble predictions by at least half that much (ΔΔG ~ 3.5 kcal/mol). Within the degree to which these two NMR-restrained molecular simulations faithfully model the Boltzmann conformational distribution of ubiquitin, the conventional interpretation of the hydrogen exchange data severely underestimates the flexibility of this protein. As discussed in the following section, the physical basis for this systematic error in flexibility predictions derived from the conventional hydrogen exchange analysis is straightforward. Although occasional exceptions arise due to specific local electrostatic interactions (24), most solvent-exposed amides that lie along the surface of a partially or fully folded protein will have lower acidities than the corresponding model peptides, due to the presence of the low dielectric volume of the adjacent protein interior. When the peptide normalization analysis is applied to protein hydrogen exchange data, these depressed ionization equilibria are misinterpreted as a lower fraction of solvent-accessible conformations. The direct implication of the electrostatic and conformational contributions to hydrogen exchange kinetics is that normalization against the model peptide exchange rates can only be expected to provide useful conformational equilibria data when the exchange-competent state exhibits both solvation and conformational sampling behavior similar to that of the model peptide (34, 35).

20

3. Kinetics and Electrostatics of Hydrogen Exchange 3.1. Implications of Amides Being Weak Normal Eigen Acids

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

375

Hydroxide-catalyzed amide hydrogen exchange is a straightforward acid-base reaction. A number of researchers (36–41) have pointed out that electrostatic interactions modulate the kinetics of amide hydrogen exchange. Nevertheless, many of the earlier reported electrostatic effects have appeared to be relatively modest when compared with the 107 to 108-fold decrease in exchange rates, which is commonly observed for the most slowly exchanging amides of moderately stable proteins. Consistent with such an assessment, it has been argued that titrating the formal charges of the side chains modulates the observed hydrogen exchange rates much more strongly via an indirect effect on protein stability than they do via a direct electrostatic interaction (42). As first predicted by Eigen (43), amides have been experimentally demonstrated (44, 45) to act as normal Eigen acids such that the reaction rate with hydroxide ions is attenuated from the diffusion limit by the fraction of forward-reacting encounters Ki/ (Ki + 1), where Ki is the equilibrium constant for the transfer of a proton from the amide to an hydroxide ion. Therefore, the thermodynamic acidity of an amide directly predicts its kinetic acidity as monitored by the hydrogen exchange reaction. Most all protein backbone amides have appreciably lower thermodynamic acidities than that of water. As a result, nearly every collision with a neutral water molecule will quench the peptide anion charge state. This low acidity implies that near neutral pH, most backbone amides will be in the peptide anion state at a fractional population of less than one part in 1010. The key advantage in predicting the ionization behavior of the backbone amide, as compared to predicting ionization of protein side chains, stems from the short lifetime of the peptide anion (22, 46, 47). In contrast to the μs–ms lifetimes for the charge states of the ionizable side chains near neutral pH, the range of protein conformational responses to the peptide anion charge state is strongly limited by its brief lifetime. Although a direct measurement of how rapidly the peptide anion is quenched by a neutral water molecule has not been reported, NMR relaxation studies indicate that the residence lifetime of an hydroxide ion in water is ~ 5 ps (48), and lifetimes near 10 ps have often been observed for photoactivated strong acids and bases (49, 50). Given that the dominant phase of the Debye dielectric relaxation profile for water has a time constant of 8 ps at 25°C (51), it has been argued that the dynamics of water reorientation are limiting in these fast proton transfer reactions (49, 50). By analogy, the lifetime of the peptide anion intermediate is likewise anticipated to be ~10 ps (22, 46, 47).

376

G. Hernández et al.

3.2. Electronic Polarizability in the Dielectric Shielding of the Peptide Anion

As long discussed in electron transfer theory (52), dielectric shielding is frequency dependent. The lifetime of a transient charge state determines the range of conformational motions that can give rise to effective dielectric shielding since conformational transitions that are slower than the charge state lifetime cannot adjust rapidly enough to stabilize that state. As a result, the dielectric shielding of the hydrogen exchange reaction that arises from the protein molecule is expected to be dominated by electronic polarizability (47). Owing to the highly transient peptide anion charge state, the kinetics of amide hydrogen exchange provides a “snapshot” of the Boltzmann conformational distribution that is nearly independent from the dynamics of interchange between protein conformations. The electrostatic free energy of a generalized-Born ion of charge Q and radius R is given by the formula (53): ΔG elec = −(1 / e int − 1 / e ext )Q 2 / 2R When such a low dielectric ion (eint) is embedded in a high dielectric solvent (eext), its electrostatic free energy is essentially inversely proportional to the value of the internal dielectric (eint). As discussed in Subheading 5.1, Poisson–Boltzmann continuum dielectric calculations on the static solvent-accessible amides from a set of four globular proteins have demonstrated that this inverse proportionality is well preserved for these more complex geometries (47). As a result, the slope of the correlation between the experimental and predicted peptide acidities provides a sensitive monitor of the optimal effective internal dielectric value, which was found to equal 3 for these same four globular proteins (22, 47). An internal dielectric value of 2.0 is commonly used to model the electronic polarizability of the protein interior. This value is derived from refractive index measurements on typical organic liquids that monitor the dielectric response at optical frequencies (~1015 s−1). However, noting that the density within the protein interior is 30–40% higher than that of analogous small molecule liquids (54, 55), Krishtalik and colleagues (56) have argued that the average contribution of electronic polarizability implies a dielectric shielding value of at least 2.5 for protein molecules. On the slower time scale of ~10−13 s, nuclei respond to an altered electric field by adjusting bond lengths and angles as well as the corresponding vibrational frequencies. Although estimates vary, the nuclear relaxation response may account for as little as 5% of the total polarizability in the high frequency range (57). These studies provide strong support for the interpretation that our experimentally derived determination of an effective internal dielectric value of 3 for the protein hydrogen exchange reaction indicates that the dielectric shielding of the peptide anion is dominated by electronic polarizability.

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility O

R N

N H

OH–

O

O

H

O

H

R

OH–

N

N R

O

H

R

N

N R

377

O

R

DV Fig. 3. Electrostatic free energy of peptide ionization. The reaction of hydroxide ion forms a peptide anion at one or another site along the protein backbone. The differential electrostatic free energy for these two species is given by the product of the charge and the difference in electrostatic potential for the two sites ΔeV which, in turn, is proportional to the ΔpK for these two amide nitrogens. Reprinted from ref. 22 with permission from the American Chemical Society.

Not only do the continuum dielectric predictions of experimental protein hydrogen exchange data provide a direct estimation of the effective internal dielectric value, the quality of the individual residue predictions provides an upper bound on how much the spatial distribution of that dielectric shielding can deviate from uniformity. The internal effective dielectric value primarily represents the volume polarizability of electronic shielding as averaged over the length scale of the electrostatic interactions for the ionizing peptide. The electrostatic potential around the amide nitrogen is sensitive to a large set of significant nonbonded interactions that range in length from van der Waals contact out to 14 Å or more (22, 58). The assumption of a uniform internal effective dielectric value proves to be a considerably more robust approximation in hydrogen exchange analysis than for the analogous protein side chain pK predictions. It is well known that applying continuum dielectric methods to protein side chain ionizations generally does not yield predictions that are consistent with a well-determined uniform internal dielectric value (59–64). Protein conformational reorganization through reorientation of the various mobile charged side chains are believed to provide the primary contribution to the comparatively high dielectric shielding observed for the ionizable side chains (65), although larger scale structural motion can play a critical role, particularly for buried side chains (66). If chemical induction effects do not differentially alter the intrinsic acidity of the protein amides, the difference in electrostatic free energy for each such pair of amide anions will correspond to the free energy of transferring an amide hydrogen from one site to the other, which is proportional to the ΔpK between those two amide nitrogens (Fig. 3) (22). The free energy of this proton transfer is necessarily equivalent to the difference in free energy of protonating each amide anion site, since in both cases the identical neutral backbone protein structure is generated. After setting all of the side chain and C-terminal carboxyls as well as the side chain and N-terminal amines to the desired charge

378

G. Hernández et al.

state, the amide proton for each of the exchanging residues is removed one at a time and the electrostatic free energy for each species is calculated. One significant complication in the prediction of protein side chain pK values is largely circumvented in the case of hydrogen exchange. Since many of the ionizable side chains of a protein have similar pK values, whether a given side chain is in the neutral or charged state will alter the ionization behavior of these other side chains so that the final population distribution of charge states must generally be determined in an iterative fashion. Although the ionization of individual backbone amides is obviously sensitive to the charge distribution of the ionizable side chains, given peptide anion concentrations of ~1 in 1010, the thermodynamics of the side chain ionizations are insensitive to the backbone ionizations. Most of the relevant hydrogen exchange data can be obtained in the range of pH 7 to pH 10. Besides histidines, relatively few side chains undergo ionization over this pH range in most small proteins. In the case of side chain pK titrations in this pH range, the pH dependence of the hydrogen exchange kinetics for nearby amides can provide an effective means of determining the hydroxide-catalyzed rate constants in both the neutral and charged state of a given side chain (47). When the protein conformational transition to the exchangecompetent state becomes rate-limiting for the hydrogen exchange reaction (i.e., the EX1 condition), the kinetic acidity of an amide is necessarily less than its thermodynamic acidity. The formally analogous condition holds for the ionization of most carbon-bound hydrogens. For example, the reaction rates of nitroalkanes with hydroxide are more than 1010-fold slower than that predicted for a normal Eigen acid (67). In this case, the charge delocalization that provides resonance stabilization of the anion progresses more slowly than does proton transfer (68), and heavy-atom intramolecular reorganization is generally the rate-limiting process (69). Establishing a clear physicochemical basis for interpreting protein hydrogen exchange provides a means to characterize both the structure and the population of protein heavy-atom reorganization processes that facilitate solvent access for the structurally buried amides.

4. Hydrogen Exchange Techniques 4.1. Magnetization Transfer Methods

Rapid hydrogen exchange can be monitored by magnetization transfer techniques in which the water resonance is selectively excited. The NMR experiment then monitors the transfer of this magnetization to the amide resonances. A particularly robust implementation of magnetization transfer-based hydrogen exchange monitoring is that of CLEANEX-PM (70, 71), in which

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

379

most NOE/ROE and TOCSY-derived contributions to the observed resonances are efficiently suppressed. We (72) have introduced a modification to this experiment that compensates for the effects arising from transverse relaxation, which limit the accuracy of the deduced exchange rates. Elimination of NOE and ROE cross-relaxation effects in the CLEANEX-PM sequence is based on their mutual cancelation in the slow tumbling limit (73). Concerns over the applicability of the CLEANEX-PM sequence to more mobile peptide groups recently prompted Skrynnikov and colleagues to develop a modified (HACACO)NH sequence to detect solvent exchange via the 15N amide resonance (74). Although the lower power level used in their SOLEXSY sequence yielded reduced sample heating in the high ionic strength solutions, the exchange rates obtained for the backbone amides of the denatured drkN SH3 domain were virtually indistinguishable from their CLEANEX-PM results. In the present application, many of the potential complications to obtaining accurate exchange rates were directly suppressed by the use of perdeuterated protein samples which further served to enhance the sensitivity of the CLEANEX-PM measurements by reducing 1H transverse relaxation effects. When the hydrogen exchange data of rubredoxin, summarized in Fig. 1, was combined with that from three other model proteins collected under analogous conditions, a total of 46 residues are found for which a linear hydroxide-dependent rate constant can be reliably fitted at two or more pH values with an overall uncertainty of 0.053 in the log kOH− rate constants, at least an order of magnitude more accurate than the current ability to predict these data by continuum dielectric methods (20). 4.2. Solvent Exchange by 1H Exchange-In Protocol

In the case of ubiquitin, exchange kinetics for the amides that do not exhibit exchange in the CLEANEX-PM experiments have been analyzed using an 1H exchange-in protocol (20). By preexchange of the amide hydrogen positions with deuterium and then dissolution of the protein sample in a 1H2O-containing buffer, one can circumvent the complications from the isotope dependence of solvent, buffer and protein side chain ionizations that plague quantitative interpretation of exchange rates measured using the conventional 2H exchange-in protocol. Furthermore, no correction is needed for the significant differences in protein stability that can result from comparing measurements in normal and heavy water buffer solutions (75). When compared to the magnetization transfer-based hydrogen exchange measurements (47), the 1H exchange-in protocol suffers mainly from the differential effect in breakage of an N–D or an N–H amide bond. Measurements on poly d,l-alanine indicate a 0.08 shift in the log rate constant for this isotope effect in the hydroxide-catalyzed exchange reaction (76). An additional benefit

380

G. Hernández et al.

of the 1H exchange-in protocol is that the final buffer conditions correspond to those used to produce the earlier reported magnetization transfer-based measurements (47). As a result, CLEANEX-PM experiments (71, 72) can be carried out on the 1H exchange-in sample so as to provide a precise calibration of the relative pH values between the two sets of measurements. Based on extrapolation from unfolding measurements in guanidinium chloride, transition to the EX1 kinetic condition, in which protein unfolding limits the hydrogen exchange rate, does not apply to the most slowly exchanging amides of ubiquitin in normal buffer conditions for pH values less than 9.5 at 25°C (77). The average rmsd fit to the [1−exp(−kext)] dependence on the amide 1H peak intensities in the ubiquitin 1H exchange-in experiments was 1.2%. Only the amides of Thr 22 and Leu 50 provided robust rate constants in both sets of experimental measurements, yielding log kOH− values from the CLEANEX-PM and 1H exchangein experiments of 3.66 and 3.62 for Thr 22 as well as 3.79 and 3.67 for Leu 50, respectively.

5. Continuum Dielectric Analysis for Exchange of Static Solvent Accessible Protein Amides 5.1. Electrostatic Parameter Set Dependence of Peptide Acidity Predictions

Hydroxide-catalyzed exchange rate constants were determined for those amides of rubredoxin, FK506-binding protein (FKBP12), ubiquitin and chymotrypsin inhibitor 2 (CI2) that are solventaccessible in the high-resolution X-ray structures (22, 47). The acidity of these amides were calculated using the Poisson–Boltzmann finite difference algorithm DelPhi (78) as a function of the nonpolarizable electrostatic parameter set, the internal dielectric value and the charge distribution of the peptide anion. As illustrated in Fig. 4, the best performance was obtained using the CHARMM22 electrostatic atomic partial charge and radius parameters (79) (these parameters are preserved in the current CHARMM27 force field), an ab initio-derived peptide anion charge distribution (47), and an internal dielectric value of 3. These parameters yielded an rmsd value of 7 for the 56 amide exchange rate constants ranging from 100.67 to 109.0 M−1 s−1. The optimal internal dielectric value was obtained via its (1/eint) scaling effect on the differences in electrostatic potential for the various peptide anions predicted by the Poisson–Boltzmann calculations and linear correlation against the experimental hydrogen exchange rates. The OPLS-AA electrostatic parameter set (80) yielded comparably robust predictions, as might be expected from its strong similarity to the CHARMM atomic charge and radii set. By contrast, the nonpolarizable AMBER parm99 (81) and AMBER ff03 (82) parameter sets performed more poorly. As illustrated in Fig. 5, the parm99 electrostatic parameters from the AMBER force field do

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

381

Fig. 4. Dependence of amide acidity predictions on the atomic charge distribution of the peptide anion. Protein amide pK values predicted using CHARMM22 atomic charge and radius parameters [79] and an internal dielectric constant of three at 25°C, with the excess anion charge density distributed throughout the peptide unit as predicted from B3LYP DFT calculations [47]. Reprinted from ref. 47 with permission from the American Chemical Society.

not reliably predict the experimental hydrogen exchange data. Many of the outlying predicted values arise from peptide groups that have neighboring charged side chains. In contrast to the CHARMM22 and OPLS-AA parameter sets considered above, the parm99 set does not assume that the charges on the O, C, N, and H of the peptide group are common to every residue type. For the lysine and arginine side chains, 11.5% of the formal charge of the side chain is assumed to reside on the backbone carbonyl group, and an additional 7.1% of the formal charge resides on the backbone nitrogen and amide hydrogen. Similarly, for the aspartate and glutamate residues, the parm99 set assumes that, relative to the atomic charges of the neutral amino acid types, the backbone carbonyl group bears 7.5% of the side chain formal negative charge, while together the backbone nitrogen and amide hydrogen bear an additional 7.9%. The partial charges of the AMBER force field are derived by fitting to the distribution of the ab initio-derived electrostatic potentials surrounding each atom (83). However, in the present context, the projection of these charges onto each nucleus operationally corresponds to a modeling of chemical induction effects. The ability to predict amide exchange in both proteins and peptides much more accurately with the electrostatic parameter

382

G. Hernández et al.

Fig. 5. Correlation of hydroxide-catalyzed hydrogen exchange rate constants with Poisson–Boltzmann-derived pK values, using AMBER parm99 electrostatic parameters [81] and an internal dielectric constant of 3 at 25°C. The open symbols represent amides for which either of the two neighboring side chains is ionized. Only amides surrounded by neutral side chains were used in the scaling of the predicted pK values. Reprinted from ref. 47 with permission from the American Chemical Society.

sets that do not incorporate large shifts in the atomic charges of the backbone atoms, indicates that the magnitude of charge migration within the individual amino acids that is modeled into the AMBER parm99 electrostatic parameter set appears to be well beyond what might be needed to rationalize local sequence-dependent variations. These considerations apply even more markedly to calculations using the AMBER ff03 electrostatic parameters (47). 5.2. Atomic Charge Distribution in the Peptide Anion

Our initial rubredoxin hydrogen exchange predictions assumed that the excess negative charge of the peptide anion resides exclusively on the nitrogen, following the earlier results from continuum dielectric modeling of hydrogen exchange in simple peptides by McCammmon and colleagues (41). That assumption conflicts with the long-standing tradition of representing the product formed by deprotonation of an amide as an imidate anion. However, in contrast to predictions from early valence bond theory studies, there

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

383

Fig. 6. Dependence of amide acidity predictions on the atomic charge distribution of the peptide anion. Protein amide pK values predicted using CHARMM22 electrostatic parameters and an internal dielectric constant of 3 at 25°C, with the excess anion charge density localized to the carbonyl oxygen. Reprinted from ref. 47 with permission from the American Chemical Society.

are no experimental results or high level quantum mechanical calculations to support the interpretation that a dominant fraction of the excess negative charge for a secondary alkyl amide anion shifts to the oxygen atom (21). Recently, we reported (47) B3LYP (84) density functional theory calculations at the aug-cc-pVTZ basis set level on the neutral and anionic states of N-methylacetamide. These calculations predicted an electron charge distribution for the peptide anion that assigns a threefold higher excess charge density for the nitrogen atom than for the oxygen atom. The DFT-derived peptide anion charge distribution provided somewhat better protein hydrogen exchange predictions than did an assignment of the excess charge to the nitrogen (47). By contrast, assignment of the excess negative charge to the oxygen, so as to generate the imidate form, yielded markedly poorer predictions of hydrogen exchange for the four model proteins (Fig. 6) than those obtained from either nitrogen-centered or ab initio-derived peptide anion charge distributions (47).

384

G. Hernández et al.

5.3. Dominant Acidic Conformer Analysis

Although these continuum dielectric calculations were based on high resolution X-ray structures, a discrete set of adjustments was applied to several side chain types during the calculation of the intraresidue peptide acidity. The most significant case involved aspartate side chains in which the χ1 side chain torsion angle is gauche to the backbone nitrogen. This orientation places the negatively charged carboxylate near the intraresidue amide, thus strongly suppressing its predicted ionization. The sterically unhindered rotation of an Asp carboxylate to a trans rotamer can enhance the acidity of the intraresidue amide by 5 pH units or more (22, 35, 47). A second type of systematic modulation in the predicted electrostatic free energy of the peptide anions as a function of the residue side chain conformation was applied when the χ1 side chain torsion angle is near +60°. In this rotamer, the Cγ is gauche to both the main chain nitrogen and carbonyl carbon and is often in van der Waals contact with the amide hydrogen. Unhindered rotation to another c1 rotamer tends to increase the solvation of the amide anion with a resultant increase the peptide acidity of that residue. Regarding the physical validity of these ad hoc side chain rotations for identifying energetically favorable conformations near the X-ray coordinates that have enhanced peptide acidities, it should be noted that the model conformation (molecule 92) within the independently generated 2NR2 ubiquitin ensemble (further discussed below) that most accurately predicts the experimental hydrogen exchange has undergone the Asp and gauche+ side chain rotamer transitions that were identified by this earlier published side chain reorientation protocol (33). The assumption of limited protein conformational reorganization during the lifetime of the peptide anion surely can not apply generally to the side chain hydroxyl hydrogens, since the analogous reorientation of the hydrogens on water molecules gives rise to the dominant dielectric shielding of that phase. Particularly for side chain hydroxyl hydrogens that are not involved in an intramolecular hydrogen bond, continuum dielectric calculations based on a fixed orientation are potentially misleading. This is most notably the case when amide acidity is estimated with an intraresidue serine or threonine hydroxyl in either a gauche+ or gauche− c1 rotamer. Given that the exchange rates are similar for serine- and threoninecontaining model peptides, as compared to the alanine reference (23), the side chain hydroxyl does not generally serve as a catalyst for peptide hydrogen exchange. Consistent with that observation, the peptide acidity analyses for serine and threonine residues with a gauche c1 rotamer assume that the dielectric shielding of the side chain hydroxyl is equal to that of the equivalent volume of water. In such cases, the serine side chain is computationally truncated to alanine, and threonine is tranformed into α-aminobutyrate.

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

6. Ensemble Averaging in Prediction of Ubiquitin Hydrogen Exchange 6.1. Population Averaging of Conformer Acidities

385

All model ensembles that are justified by their collective ability to predict experimental measurements necessarily invoke the assumption that they represent an accurate Boltzmann sampling of conformational space. Yet not all experimental approaches offer comparable sensitivity to variations in the model distribution. In contrast to experimental techniques that are equally sensitive to every protein conformation and thus are generally dominated by the most populated states, hydrogen exchange reactivity is highly sensitive to conformation. The fact that structurally buried amides are effectively unreactive to hydrogen exchange forms the basis for the widespread application of this experimental technique for monitoring rare conformational states. On the one hand, the highly exposed amides exhibit exchange rates that are quite sensitive to the well-populated protein conformations. On the other hand, this sensitivity to conformation applies to the rarely exposed amides as well so that hydrogen exchange measurements for these sites provide a powerful experimental monitor of both the population and conformation of the transient exchange-competent state. Although population averaging of the conformer acidities (ΣKi) is more formally correct (85), population averaging of the conformer pKi values (ΣpKi or, equivalently, averaging of the conformer electrostatic potential values) has occasionally been used to estimate the effect of conformer sampling in the prediction of protein side chain ionization. Karplus and colleagues (86) have concluded that, in assessment of the ionization midpoint for each titrating residue, averaging over the Ki values or averaging over the log Ki values usually has little effect on the predicted pK values. The issue of Ki vs. pKi averaging is markedly different in the estimation of hydroxide-catalyzed hydrogen exchange rates near neutral pH where, in most cases, less than 1 out of every 1010 molecules will have a given amide in the ionized state. As a result, whenever there is a substantial range in conformer acidities, the most acidic conformers can make the dominant contribution to the observed hydrogen exchange rate, even if they constitute only a modest fraction of the overall conformer population. Molecular simulation techniques have been increasingly employed to predict the Boltzmann-weighted conformational distribution of the protein native state. In principle, under the assumption of ergodicity, an unconstrained constant temperature molecular dynamics simulation can provide the Boltzmann conformational distribution. In practice, given the roughness of protein energy landscapes, even simulations extending for hundreds of nanoseconds will generally suffer from incomplete conformational sampling. Furthermore, force field parameterizations are only approximate. As a result, the predicted conformational distribution can drift away from the physical values.

386

G. Hernández et al.

6.2. Consistency of Ubiquitin Model Ensembles with the Native State Conformational Distribution

Concerns arising from approximate force fields and incomplete sampling have been approached by incorporating experimentally derived restraints into molecular dynamics simulations, as applied to ubiquitin. The MUMO algorithm of Vendruscolo and colleagues (31) introduced NOE-derived distance bound restraints, averaged over subsets of protein conformations, as a mechanism for maintaining the predicted molecular dynamics ensemble distribution to within the neighborhood of the experimentally determined structure. In parallel, order parameters S2, derived from backbone 15N and side chain 13C methyl NMR relaxation measurements, were incorporated into the restrained molecular simulation where they enforce enhanced conformational sampling. The resultant set of 144 protein conformations (PDB code 2NR2 (31)) serves as a model for the random sampling of the native state Boltzmann distribution of ubiquitin. In generating an alternate model ensemble (PDB code 2K39), de Groot and colleagues (32) applied the CONCOORD algorithm (87) using the same 2,727 NOE constraints from the 1D3Z solution structure analysis (88) to generate 1,000 model conformations of ubiquitin. In the EROS (ensemble refinement with orientational restraints) protocol a subset of 400 conformations were initially selected as most consistent with the residual dipolar coupling (RDC) data. An iterative process of simulated annealing followed by reselection against the RDC data was then applied until the initial set of 1,000 conformations was winnowed down to a final set of 116 conformations. When ensemble averaging of hydrogen exchange reactivity was applied to the NOE, S2-restrained 2NR2 ubiquitin ensemble (31), the hydroxide-catalyzed exchange rates for nearly all of the highly exposed amide hydrogens (solvent-accessible in >50% of conformations) were quite accurately predicted (black circles in Fig. 7) (20, 33). For 16 of these highly exposed amides (Gly 47 and Asp 52 discussed below), the 2NR2 ensemble predicted the 105-fold range in experimental rates, yielding an rmsd of 0.51 and a correlation coefficient r = 0.94 for the log kOH− values. This correlation is markedly better than that obtained using a single crystallographically derived ubiquitin structure (47). Most strikingly, for the backbone amides that are exposed to solvent above 0.5 Å2 in more than one but less than half of the models in the NMR relaxation-restrained ensemble, with the exception of Lys 48, the amide pKa predictions are nearly as accurate (rmsd for log kOH− of 0.69) as those for the more highly exposed sites (Fig. 7). Despite being structurally buried by most conventional criteria, the exchange rate constants for these 12 residues, spanning nearly a million-fold range, are predictable to within a factor of 5. The underestimation of the hydrogen exchange rates for residues in which only one model conformation has an amide hydrogen

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

387

10 G47 K48

8

6

log kOH– (PB)

20

4

D52

2

0

–2 –2

0

2 4 6 log kOH– (M–1 s–1)

8

10

Fig. 7. Hydroxide-catalyzed rate constants predicted from the NMR relaxation-restrained 2NR2 ensemble of ubiquitin. For residues in which the amide hydrogen is exposed to solvent by more than 0.5 Å2 in at least one ensemble model, conformer acidities were predicted for all solvent-exposed amides. Each residue is distinguished according to whether the amide hydrogen is exposed to solvent by more than 0.5 Å2 in at least 50% of the models (filled circle) or exposed in only a single model (filled square). The other transiently exposed amides are denoted as (filled diamond ). Residues Gly 47, Lys 48, and Asp 52 are denoted with open symbols. Reprinted from ref. 20 with permission from the American Chemical Society.

accessibility above 0.5 Å2 (filled square in Fig. 7) is consistent with that expected from the statistics of undersampling. As a result of the ΣKi averaging of conformer acidities discussed above, a sufficient number of models must be sampled not only to establish the relative fraction of solvent-exposed conformations. In addition, the solvent-exposed conformations must be sufficiently sampled so as to approximate the distribution of conformer acidities therein. Indeed, many of the residues in the 2NR2 ensemble exhibit a range of conformer acidity values (103–106) comparable to those of the four residues in the conformationally disordered C-terminus, despite remaining well-ordered as indicated by the Cα rmsd of 0.56 Å for residues 1–72 in this ensemble (Fig. 8). Regarding the overestimated peptide acidities for Gly 47 and Lys 48 predicted from the 2NR2 ensemble, a more marked overestimation is also obtained from the 2K39 ensemble discussed below that spans the segment Ile 44 to Lys 48, which constitutes a major portion of the recognition site for enzymes involved in formation of Lys 48-linked poly ubiquitylation signals for proteasomal targeting. In the 144 structures of the 2NR2 ensemble, not a single Asp 52 side chain is predicted to occupy a trans rotamer and all

388

G. Hernández et al. 14 16

pKPB

18 20 22 24 26 28

0

10

20

30

40 residue

50

60

70

Fig. 8. The range of conformer acidities for amide hydrogens exposed to solvent by at least 0.5 Å2 in the NMR relaxationrestrained 2NR2 ensemble of ubiquitin. Residues for which the amide hydrogen is solvent-exposed in less than 50% of the ensemble models are indicated in gray, while those that are solvent-accessible in more than 50% of the models are marked in black. The peptide acidities are placed on an absolute scale based on their normal Eigen acid behavior, the diffusion-limited rate for hydroxide-catalyzed exchange of 2 × 1010 M−1 s−1, and the pK of 15.7 for water at 25°C. These properties imply an exchange rate constant of 1.0 M−1 s−1 for an amide with a pK value of 26.0 [47]. Reprinted from ref. 20 with permission from the American Chemical Society.

except 8 of these side chain conformations have the carboxylate bound in a salt bridge to the Lys 27 side chain so that no sampling of the higher acidity conformations are included. Although the NOE-restrained, RDC-selected 2K39 ensemble yields predictions for the highly exposed amides that are nearly as robust as those from the 2NR2 ensemble (Fig. 9), the 2K39 ensemble provides substantially overestimated exchange rates for a number of the more weakly exposed amide sites (20, 33). All eight residues for which only a single ensemble model has an amide hydrogen accessibility above 0.5 Å2 have predicted exchange rates that exceed the experimental results. As indicated in the discussion above, it is highly unlikely that all of these overestimates arise as the result of undersampling. 6.3. Comparison Against the Set of Known UbiquitinProtein Complexes

Of particular significance is the pattern of exchange rates seen for the proteasome targeting interaction site around residue Lys 48. In explicit contrast to the 2NR2 ubiquitin ensemble of Vendruscolo and colleagues (31), de Groot and colleagues (32) contended that their 2K39 ensemble spans a conformational space that includes the ubiquitin structures found in all of the available X-ray studies of ubiquitin-protein complexes (41 complexed-ubiquitin molecules +5 X-ray structures of uncomplexed ubiquitin). These authors further claimed that conformations of ubiquitin found in these

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

389

10 G47 K48

8 F45

log kOH– (PB)

20

6

I44

4

D52

2

0 0

2

4 6 log kOH– (M–1 s–1)

8

10

Fig. 9. Hydroxide-catalyzed rate constants predicted from the NMR residual dipolar couplingrestrained 2K39 ensemble of ubiquitin. For residues in which the amide hydrogen is exposed to solvent by more than 0.5 Å2 in at least one ensemble model, conformer acidities were predicted for all solvent-exposed amides. Each residue is distinguished according to whether the amide hydrogen is exposed to solvent by more than 0.5 Å2 in at least 50% of the models (filled circle) or exposed in only a single model (filled square). The other transiently exposed amides are denoted as (filled diamond ). Asp 52 and the residues involved in the primary interaction site for proteasome targeting are individually identified. Reprinted from ref. 20 with permission from the American Chemical Society.

protein complexes are well represented in their 2K39 conformational ensemble of free ubiquitin, providing what has been widely regarded to be a compelling demonstration of the conformational selection mechanism of protein-protein recognition (89–91). However, the log kOH− values for Ile 44, Phe 45, Gly 47, and Lys 48 predicted from the 2K39 ensemble exceed the experimental results by 2.4, 3.2, 1.7, and 3.1, respectively. An indication that anomalous sampling statistics do not explain these discrepancies is that for six of the seven ensemble models in which the Lys 48 amide hydrogen is exposed to solvent by more than 0.5 Å2 each predict a conformer hydrogen exchange rate that is more than 4,000-fold above the experimentally observed value. As a result, even after normalization to the 116 models in the ensemble, the predicted amide pKa value is significantly above the experimental value. Furthermore, the experimental S2 order parameter values of 0.838, 0.872, 0.840, 0.821, and 0.843 for the N–H bond vectors of residues Ile 44 to Lys 48 (92) indicate that any substantial internal motion in this segment must be very weakly populated on the ps-ns timescale.

390

G. Hernández et al.

This discontinuity between the claims for conformational sampling in the 2K39 ensemble study and the prediction of hydrogen exchange reactivity from this ensemble prompted a reexamination of the conformational distribution of the 2K39 ensemble. One line of evidence that these authors (32) provided for indicating that the NOE-restrained, RDC-selected 2K39 ensemble spans the conformational space of the ubiquitin-protein complexes is that each of the X-ray structures is within a backbone rmsd of 0.8 Å from at least 1 of the 116 members of the ensemble (for the Cα atoms of residues 1–70). Yet when the analogous calculation was carried out on the 2NR2 ensemble, a maximum backbone rmsd value of 0.7 Å was obtained between each of the 46 X-ray structures and the nearest member of the 2NR2 ensemble (33). More significantly, for 36 of the 46 ubiquitin X-ray structures, each of the 116 backbone conformations in the 2K39 ensemble is farther from that X-ray structure than is the 1D3Z solution structure model (88) from which the 2K39 ensemble was initiated. In comparison, for 41 of the 46 ubiquitin X-ray structures, the 2NR2 ensemble contains a backbone conformation that is closer to the X-ray structure than is any member of the 2K39 ensemble. Overall, the 2K39 ensemble clearly represents a drifting away from the conformational space spanned by the crystal structures of ubiquitinprotein complexes. 6.4. Hydrogen Exchange Analysis as a Monitor for Completeness of Ensemble Sampling

Ideally, these two ubiquitin ensembles represent ~102 random samplings of the Boltzmann conformational distribution so that amides, which become exposed to solvent at less than a 1% frequency, will generally be unrepresented in these peptide acidity predictions. As indicated in Fig. 2, for nearly every case in which an amide hydrogen is exposed to solvent in at least one conformation from either the 2NR2 or 2K39 ensembles, the experimental exchange rate is less than what would be predicted for a model peptide having the same fraction of solvent-exposed conformations (only for Thr 9 is the apparent solvent accessibility estimated from peptide normalization significantly above that from both of the ensembles). The log exchange rate constants for most model peptides are >8 (23). Hence, one may anticipate that for a proper 1% Boltzmann sampling of the conformational distribution nearly all backbone amides having log kOH− values >6 should have solvent-accessible conformations within that 1% sampling. Indeed, each of the 23 ubiquitin amides that have experimental log kOH− values >6 are exposed to solvent in at least one conformation in both the 2NR2 and 2K39 ensembles (20). On the contrary, there are some amides of ubiquitin, which are exposed to solvent in these two ensembles, that have predicted and observed exchange log rate constants that are significantly less than 6, reflecting the fact that their exchangecompetent conformations have strongly depressed exchange reactivities. Nevertheless, a number of backbone amides that are solvent

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

391

inaccessible in the X-ray structures of ubiquitin have conformers within these two model ensembles with amide acidities that are similar to those of simple peptides (Fig. 8) (20). For such a residue, if its log kOH− value is below 6, then its amide hydrogen could be expected to remain solvent inaccessible in most Boltzmann samplings at a 1% level. There are 12 amides in ubiquitin that have log exchange rate constants between 5 and 6. For 8 of these 12 residues, the amide hydrogen is solvent inaccessible in every conformation of the 2NR2 ensemble, despite the fact that all 23 amides with log kOH− values >6 are solvent accessible. By contrast, only 3 of the 12 residues with log exchange rate constants between 5 and 6 are solventinaccessible in every 2K39 conformation, consistent with an overly expanded sampling of conformational space for that ensemble. This analysis supports the expectation that the fraction of solventaccessible conformations as a function of the log kOH− values can provide a useful monitor of the degree of completeness with which a given ensemble has sampled the energetically accessible conformational space.

7. Continuum Dielectric Analysis of Hydrogen Exchange in Model Peptides 7.1. Backbone Conformation Dependence of Side Chain Correction Factors for Hydrogen Exchange

Predicting the experimental side-chain-dependent differential exchange rates for conformationally unstructured peptides presents a rather stringent challenge. Although early studies argued that the differences in exchange rates among various simple model peptides arise from chemical induction effects (44, 93), more recent electrostatic calculations have indicated that amide hydrogen exchange rates are strongly dependent upon the relative orientation of the adjacent peptide groups (41, 94). Experimentally measured hydroxide-catalyzed amide exchange rates for conformationally unstructured alanine peptides are essentially unaffected by the intraresidue substitution of a methionine side chain, while substitution of a phenylalanine, tyrosine, or tryptophan side chain decreases the exchange rate by approximately twofold (23). A significantly larger (~fivefold) attenuation of the exchange rates results from substituting any of the branched side chains from valine, leucine, or isoleucine. A similar pattern is obtained when these side chain substitutions are introduced into the residue preceding the site of amide exchange, although the magnitude of the variations in exchange rates is approximately fourfold smaller. As a result of these fairly small effects, a compelling correlation between predicted and observed side-chain-dependent hydrogen exchange rate differences for conformationally unstructured peptides requires substantially more accurate predictions than have yet been demonstrated in protein studies.

392

G. Hernández et al.

Applying continuum dielectric methods with an internal dielectric value of 4, Fogolari et al. (41) reported a substantial dependence of peptide acidity on the conformation of the backbone. Using a 70 to 30% weighted average from a pair of calculations with an extended backbone and an α-helical backbone conformation, respectively, they found that the log correction factors for the side chains preceding the exchanging amide could be predicted with an rmsd of 0.17 for the set of experimental hydroxide-catalyzed exchange rate constant values. For the larger range of differential log exchange rates that arise from altering the intraresidue side chain, those authors obtained an appreciably worse fit with an rmsd of 0.38, comparable to the range of experimental log rate values. Avbelj and Baldwin (94) presented a Poisson–Boltzmann analysis of the steric contributions to hydrogen exchange that demonstrates the dependence of the predicted side chain correction factors on the assumed backbone conformation. For the various side chain types, these authors predicted differences in electrostatic free energies, relative to the alanine reference, that were roughly twice as large when the residue bearing the exchanging amide was placed in a polyproline II conformation as compared to when it was placed in an extended conformation. In both of these earlier studies of the side-chain-dependent exchange rates, the problem was simplified by comparing between residue types assuming the same backbone conformational distribution. As a result, insight into the impact of the different side chain types upon the conformational distribution of the backbone is lost. As compared to the conformational complexity of the protein native state, it might be anticipated that accurate prediction of the Boltzmann-weighted conformational distribution for simple model peptides should be relatively straightforward. In fact, even the basic question of the relative fraction of extended vs. α conformational populations in model peptides remains an actively debated issue (95–97). Current implementations of classical molecular dynamics simulations (98, 99), as well as density functional theory-based modeling (100), continue to yield disparate predictions for the backbone conformational distributions of simple model peptides. 7.2. Dependence of Peptide Acidity on Backbone Conformation

Poisson–Boltzmann analysis was carried out utilizing the Protein Coil Library of Rose and colleagues (101) as a model for the Boltzmann-weighted distribution of the unstructured state. In this structural library, protein segments lying outside of regular secondary structures were identified from high resolution X-ray analysis. Generally implicit in the application of these coil libraries is the assumption that the other forms of long range interactions that are present in the protein crystal structure do not systematically shift the average conformational distribution of the individual residue types away from that of conformationally disordered polypeptides.

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility 100

Coil library entries

20

393

N-Acetyl-Ala-Ala-N-Methylamide *

80 60 40 20 0

–3.0

–2.0

–1.0 0 ΔpK (NMA)

1.0

2.0

Fig. 10. Peptide acidities of Ala–Ala conformers. The electrostatic potential was calculated for the central peptide anion (asterisks) formed from blocked peptides derived from the 679 Ala–Ala segments found in the Protein Coil Library [101], utilizing the CHARMM22 atomic charge and radius parameters [79] and an internal dielectric value of 3. The N-methylacetamide anion was used as reference for the electrostatic potential calculations. Reprinted from ref. 34 with permission from Elsevier Limited.

The 17,422 protein segments in this coil library, taken from X-ray structures having at least 1.6 Å resolution and R values of 0.25 or better, were divided into dipeptide segments. From the atomic coordinates of these segments, model N-acetyl-[X-Ala]-Nmethylamides and N-acetyl-[Ala-Y]-N-methylamides were constructed and then used to predict the experimental side chain correction factors for hydrogen exchange (23), while the other N-acetyl-[X–Y]-N-methylamides were subsequently used to assess whether anticipated deviations from additivity for these correction factors (95, 102) are observed. When the CHARMM22 electrostatic parameters were applied to the coil library-derived Ala-Ala peptides with the internal dielectric set to 3, the conformer acidities were found to span a range of 6 pH units (Fig. 10). A similar range of conformer acidities was predicted for the other coil library peptides considered here. Hence, within the range of conformations observed in native protein structures, the local backbone conformation of the adjacent peptide groups is predicted to give rise to a million-fold range in hydroxide-catalyzed amide hydrogen exchange rates. As illustrated in Fig. 11, nearly all of Ala-Ala peptides predicted to be highly acidic have the N-terminal residue in either a β (center at f » −130°, y » 125°) or a polyproline II (center at f » −80°, y » 145°) conformation and the C-terminal residue in the α conformation. The reverse pattern holds true for the least acidic peptide conformers. It is the residue with a conformation near the α-helix basin of the Ramachandran map that dominates this behavior. When the C-terminal residue is an α conformation, the positive end of this peptide dipole points toward the ionizing nitrogen,

G. Hernández et al. 180 150 120 90 60 30 psi

394

0 –30 –60 –90

–120 –150 –180 –180 –150 –120

–90 phi

–60

–30

0

Fig. 11. The backbone conformational distribution of the most acidic and least acidic N-acetyl-Ala-Ala-N-methylamide conformers in the Protein Coil Library. The (f,ψ) torsion angle values for the 50 most acidic peptides are plotted in gray, while the values for the 30 least acidic peptides are plotted in black. The N-terminal residues are denoted by circles and the C-terminal residues by triangles. Dotted lines are used to correlate the N- and C-terminal residue backbone torsion angles for peptides that do not bridge between the extended and α conformational regions. None of the most acidic peptides have positive f torsion angles. Reprinted from ref. 34 with permission from Elsevier Limited.

thus stabilizing the anionic intermediate. Similarly, the negative end of that peptide dipole points toward the ionizing nitrogen when the N-terminal residue is in an α conformation. The subset of acidic Ala-Ala conformers with an extended N-terminal residue and a C-terminal residue in an α conformation is predicted to account for over 60% of the total hydrogen exchange, while constituting only 12% of the population. By contrast, over half of the Ala-Ala peptides in the coil library have either both residues in the extended conformation or both in the α conformation. However, when combined, these two sets of conformers are predicted to account for only 11% of the total hydrogen exchange reaction.

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

395

15 16

pKPB (Ala-Val)

17 18 19 20 21 22 23 23

2 22

21

20

19

18

17

16

15

pKPB (Ala-Ala)

Fig. 12. Relative contributions of side chain and main chain interactions determining the peptide acidities of N-acetyl-[Ala-Val]-N-methylamide conformers. For each of the N-acetyl-[Ala-Val]-N-methylamide conformers, the methyl groups were truncated to form an [Ala-Ala] conformer, and the electrostatic free energy was calculated. The corresponding pairs of amide pK values are denoted by their χ1 side chain rotamer (g− as diamonds, g+ as squares, and t as circles). Reprinted from ref. 35 with permission from Elsevier Limited.

7.3. Hydrogen Exchange for Nonpolar Side Chains and Nonadditivity of Correction Factors

The degree to which the backbone geometry contributes to the predicted conformer acidities can be assessed by analysis of the Protein Coil Library [Ala-Val] peptide conformations with the valine side chain truncated to alanine. For each of the three c1 rotamers of the valine side chain, the acidity of the central amide is closely correlated with that of the [Ala-Ala] peptide in the same backbone geometry (Fig. 12). However, conformers with the gauche− side chain rotamer, in which both methyl groups are oriented gauche to the backbone nitrogen, are predicted to have appreciably lower amide acidities (on average ~0.7 pH units). Throughout the range of amide acidities, differences in the side chain c1 torsion angle give rise to variations in pK values spanning ~1 pH unit. For each of the nonpolar N-acetyl-[X-Ala]-N-methylamides and N-acetyl-[Ala-Y]-N-methylamides, the predicted acidities of the central amide in the individual conformers of each peptide span nearly a million-fold range. Nevertheless, population averaging of the conformer reactivities predicts the standard side-chain-dependent hydrogen exchange correction factors for model peptides to within a factor of 30% (100.11) with a correlation coefficient r = 0.91 (Fig. 13).

G. Hernández et al. 0.2 YA FA WA

0

MA

LA

D log kOH– (PB)

396

IA

–0.2

AA VA

AY

AM AF

–0.4

AL AW

–0.6

AV AI

–0.8 –0.8

–0.6

–0.4

–0.2

0

0.2

D log kOH– (exp)

Fig. 13. Predicted and observed nonpolar side-chain-dependent differences in the hydroxide-catalyzed log rate constants for model peptides. Poisson–Boltzmann electrostatic free energies were calculated for the N-acetyl-[X-Ala]-N-methylamide conformers and N-acetyl-[Ala-Y]-N-methylamide conformers derived from the Protein Coil Library [101]. Hydrogen exchange rate constants were predicted from the ensemble averaging of the conformer exchange reactivities and were then compared to the standard experimental side-chain-dependent hydrogen exchange correction factors [23]. Reprinted from ref. 35 with permission from Elsevier Limited.

Hydrogen exchange rate predictions for unstructured peptides are standardly derived by adding the side-chain-dependent exchange rate correction factors for the side chain preceding and the side chain following the peptide group undergoing exchange (23). This assumption of additivity for the individual side chain correction factors has its conceptual justification in the isolated residue hypothesis of Flory, developed in his classic analysis of the statistical mechanics of random coil polymers (103). In that paradigm, the conformational distribution of each residue is assumed to be independent of the conformational distribution of any other residue in the chain. However, direct evidence for violation of the isolated residue hypothesis (103) was first reported by Penkett et al. (104) who observed that the backbone ϕ torsion angle of a given residue, as assessed on the basis of the 3JHNα NMR scalar coupling constant, is dependent on whether the preceding residue has a β-branched or an aromatic side chain. The presence of a bulky nonpolar side chain on the preceding residue was found to increase the population of conformers having extended main chain torsion angles. Studies of coil libraries drawn from high resolution protein X-ray structures

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

397

have found nearest-neighbor effects in the backbone torsion angle preferences (105, 106). If the conformer distributions from the Protein Coil Library (101) perfectly conformed to the isolated residue hypothesis, then the differential hydrogen exchange rates predicted for each N-acetyl-[X–Y]-N-methylamide might be expected to be precisely determined by the sum of the Δlog kOH− values for the corresponding N-acetyl-[X-Ala]-N-methylamide and N-acetyl-[Ala-Y]-Nmethylamide. However, the predictions assuming additivity in the calculated nonpolar side-chain-dependent hydrogen exchange correction factors deviate from the N-acetyl-[X–Y]-N-methylamide calculations with an rmsd of 0.14 and a correlation coefficient r = 0.78. This correlation is substantially worse than that for the predictions of the experimental nonpolar side-chain-dependent hydrogen exchange correction factors (Fig. 13), despite the fact that the calculations of the exchange reactivities for the N-acetyl[X-Ala]-N-methylamides and N-acetyl-[Ala-Y]-N-methylamides have several additional sources of uncertainty. 7.4. Evidence for Dielectric Shielding from Conformational Reorganization in Carboxamide Side Chains

It should be noted that for the nonpolar side chains illustrated in Fig. 13, an assumed internal dielectric value of three provides an optimal correlation between the experimental and predicted peptide acidities. As expected, the slope of this correlation exhibits the anticipated inverse dependence on that dielectric value. Hence, even in the case of the highly mobile nonpolar peptides, the conformational contribution to dielectric shielding of amide ionization appears to be severely limited. However, evidence for a modest contribution from conformational reorganization was observed for the Asn and Gln side chains. Poisson–Boltzmann calculations on the coordinates of the Asn and Gln residues in Protein Coil Library distribution yielded amide reactivity predictions that were appreciably less than the experimentally determined values (open symbols in Fig. 14). In contrast to unhindered rotamer transitions around the sp3–sp3 bonds that generally occur in the timeframe of hundreds of picoseconds to nanoseconds (107, 108), the sp3–sp2 hybridization of the carboxamide side chain results in more rapid dihedral angle transitions due to the lower intrinsic torsional potential barrier. Quantum mechanical analysis indicates a barrier of only 0.15 kcal/mol for acetamide (109). As a result, within each c1 rotamer state of Asn, extensive sampling of the c2 torsion angle can potentially occur during the peptide anion lifetime. Given the large dipole of the side chain carboxamide group, such a bond rotation can substantially alter the degree of stabilization provided for the peptide anion. Calculations were conducted to estimate the magnitude of the shielding effect from conformational reorganization of the Asn side chain by assuming rapid averaging around the c2 torsion angle. Upon deprotonation of the peptide unit, the statistical weighting

398

G. Hernández et al. 0.6 0.4 AN

D log kOH– (PB)

0.2

NA

0

QA AQ

–0.2 –0.4 –0.6 –0.8 –0.8

–0.6

–0.4

–0.2

0

0.2

0.4

0.6

D log kOH– (exp)

Fig. 14. The effect of conformational reorganization within the individual side chain rotamer states for Asn (χ1) and Gln (χ1 and χ2). Electrostatic free energies were calculated for the [Asn-Ala], [Ala-Asn], [Gln-Ala], and [Ala-Gln] methylamide conformers derived from the Protein Coil Library [101]. Hydrogen exchange rate constants were predicted from the ensemble averaging of these conformer exchange reactivities with (filled circle) and without (open circle) allowance for conformational reorganization of the peptide anions within each side chain rotamer state. The other data are displayed as given in Fig. 13. Reprinted from ref. 35 with permission from Elsevier Limited.

of this distribution will shift according to the conformer-dependent strength of the electrostatic interaction between the carboxamide group and the peptide anion. The differences in energy among all of the conformer-dependent electrostatic interactions within a given c1 rotamer state were used to assign a Boltzmann factor weighting to each conformer within that c1 rotamer state. For the Asn peptides, this correction for conformational reorganization brings the predicted peptide acidity fully in line with the experimental results (Fig. 14). A smaller correction is predicted for the Gln side chain, reflecting the larger average distance between the backbone nitrogen and the side chain carboxamide. 7.5. Deviations in Dielectric Continuum Modeling of Hydrogen Exchange Arising from Chemical Induction and Aspartate Side Chain Interactions

The Ser, Thr, Cys and His+ intraresidue side chains all accelerate hydroxide-catalyzed peptide hydrogen exchange (23), consistent with an electron-withdrawing effect from the substituent. However, the challenges facing an adequate modeling of the electrostatic potential for these side chains complicate the deconvolution of an additional contribution to peptide acidity arising from chemical induction (35). With the exception of the Asp residue, the other charged side chains yield correction factors for model peptide hydrogen

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

399

16

18 pKPB (Ala-Asp)

20

20

22

24 24

22

20 18 pKPB (Ala-Ala)

16

Fig. 15. The relative contributions of side chain and main chain interactions in determining the peptide acidities of N-acetyl-[Ala-Asp]-N-methylamide conformers. For each N-acetyl[Ala-Asp]-N- methylamide, the carboxylate group was truncated to form an [Ala-Ala] conformer, and the electrostatic free energy was then calculated. The corresponding pairs of amide pK values are denoted by their c1 side chain rotamer (g− as diamonds, g+ as squares, and t as circles). Reprinted from ref. 35 with permission from Elsevier Limited.

exchange that are robustly predictable from the Protein Coil Library (21). However in marked contrast to each of the other residue types, the hydrogen exchange prediction derived from the calculated peptide acidities for the Asp conformers is significantly lower than the experimental result, even after the elevated ionic strength of the hydrogen exchange measurements are taken into account. These deviations strongly suggest that the present continuum dielectric calculations overestimate the effective strength of the electrostatic interaction between the peptide anion and the negatively charged Asp side chain. Insight into these interactions can be gained from consideration of the distribution of conformer acidities as a function of side chain rotamer state. When the [Ala-Asp] peptide conformers and the [Ala-Ala] peptide conformers generated by truncation of the carboxylate were compared (Fig. 15), both gauche c1 rotamer states yielded side chain conformation-dependent variations in peptide acidities that were substantially larger than those obtained from the analogous predictions for the [Ala-Val] peptide conformers (Fig. 12). This effect is particularly marked for the g+ rotamer state in which a substantial proportion of conformers have a carboxylate oxygen positioned close to the peptide nitrogen. By contrast, the [Ala-Asp] peptide conformers in the trans c1 rotamer state yield peptide conformer acidities

400

G. Hernández et al.

that tightly correlate with the electrostatic interactions of the backbone, although the predictions are uniformly shifted to lower acidities, due to the long range interaction between the peptide anion and the trans carboxylate. The strongly enhanced peptide acidities predicted from rotating to the trans c1 rotamer reflect the effects previously noted in our analysis of hydrogen exchange in rubredoxin (22) and other model proteins (47).

8. Future Directions Future studies will provide insight into the degree to which the residual inaccuracies in predictions of hydrogen exchange for either proteins or model peptides reflect errors in the modeling of the Boltzmann conformational distribution or rather reflect inadequacies in the electrostatic modeling used to analyze those conformations. Both of these avenues for improved predictive capability will require pursuit. However, as recently observed by Senn and Thiel (110), despite many years of intense research effort, there are as yet no generally established polarizable biomolecular force fields. Fortunately, the present studies further demonstrate that, in the context of dielectric shielding without substantial conformational reorganization, the classic paradigm of uniform volume polarizability is strikingly robust. As such, continued insights into modeling of the Boltzmann conformational distribution from hydrogen exchange analysis can be anticipated on the basis of the continuum dielectric representation. References 1. Berger, A., and Linderstrøm-Lang, K. (1957) Deuterium exchange of poly-DL-alanine in aqueous solution. Arch. Biochem. Biophys. 69, 106–118. 2. Hvidt, A., and Nielsen, S.O. (1966) Hydrogen exchange in proteins. Advances in Protein Chem. 21, 287–386. 3. Bai, Y.W., Milne, J.S., Mayne, L., and Englander, S.W. (1994) Protein stability parameters measured by hydrogen exchange. Proteins: Struct., Funct., Genet. 20, 4–14. 4. Huyghues-Despointes, B.M.P., Scholtz, J.M., and Pace, C.N. (1999) Protein conformational stabilities can be determined from hydrogen exchange rates. Nat. Struct. Biol. 6, 910–912. 5. Li, R., and Woodward, C. (1999) The hydrogen exchange core and protein folding. Prot. Sci. 8, 1571–1591. 6. Hilser, V.J., and Freire, E. (1996) Structurebased calculations of the equilibrium folding pathway of proteins. Correlation with hydrogen

exchange protection factors. J. Mol. Biol. 262, 756–772. 7. Wallqvist, A., Smythers, G.W., and Covell, D.G. (1997) Identification of cooperative folding units in a set of native proteins. Prot. Sci. 6, 1627–1642. 8. Bahar, I., Wallqvist, A., Covell, D.G., and Jernigan, R.L. (1998) Correlation between native-state hydrogen exchange and cooperative residue fluctuations from a simple model. Biochemistry 37, 1067–1075. 9. Sheinerman, F.B., and Brooks, C.L. (1998) Molecular picture of folding of a small a/ß protein. Proc. Natl. Acad. Sci USA 95, 1562–1567. 10. Garcia, A.E., and Hummer, G. (1999) Conformational dynamics of cytochrome c: Correlation to hydrogen exchange. Prot. Struct. Funct. Genet. 36, 175–191. 11. Dixon, R.D.S., Chen, Y., Ding, F., Khare, S.D., Prutzman, K.C., Schaller, M.D.,

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

Campbell, S.L., and Dokholyan, N.V. (2004) New insights into FAK signaling and localization based on detection of a FAT domain folding intermediate. Structure 12, 2161–2171. 12. Livesay, D.R., Dallakyan, S., Wood, G.G., and Jacobs, D.J. (2004) A flexible approach for understanding protein stability. FEBS Lett. 576, 468–476. 13. Freire, E. (1999) The propagation of binding interactions to remote sites in proteins: Analysis of the binding of the monoclonal antibody D1.3 to lysozyme. Proc. Natl. Acad. Sci. USA 96, 10118–10122. 14. Pan, H., Lee, J.C., and Hilser, V.J. (2000) Binding sites in Escherichia coli dihydrofolate reductase communicate by modulating the conformational ensemble. Proc. Natl. Acad. Sci USA 97, 12020–12025. 15. Wrabl, J.O., Larson, S.A., and Hilser, V.J. (2001) Thermodynamic propensities of amino acids in the native state ensemble: Implications for fold recognition. Prot. Sci. 10, 1032–1045. 16. Wrabl, J.O., Larson, S.A., and Hilser, V.J. (2002) Thermodynamic environments in proteins: Fundamental determinants of fold specificity. Prot. Sci. 11, 1945–1957. 17. Babu, C.R., Hilser, V.J., and Wand, A.J. (2004) Direct access to the cooperative substructure of proteins and the protein ensemble via cold denaturation. Nat. Struct. Mol. Biol. 11, 352–357. 18. Wang, S.W., Gu, J., Larson, S.A., Whitten, S.T., and Hilser, V.J. (2008) Denatured-state energy landscapes of a protein structural database reveal the energetic determinants of a framework model for folding. J. Mol. Biol. 381, 1184–1201. 19. Cremades, N., Sancho, J., and Freire, E. (2006) The native-state ensemble of proteins provides clues for folding, misfolding and function. Trends Biochem. Sci. 31, 494–496. 20. LeMaster, D.M., Anderson, J.S., and Hernández, G. (2009) Peptide conformer acidity analysis of protein flexibility monitored by hydrogen exchange. Biochemistry 48, 9256–9265. 21. Anderson, J.S., Hernandez, G., and LeMaster, D.M. (2010) Conformational Electrostatics in the Stabilization of the Peptide Anion. Curr. Org. Chem. 14, 162–180. 22. Anderson, J.S., Hernández, G., and LeMaster, D.M. (2008) A billion-fold range in acidity for the solvent-exposed amides of Pyrococcus furiosus rubredoxin. Biochemistry 47, 6178–6188. 23. Bai, Y.W., Milne, J.S., Mayne, L., and Englander, S.W. (1993) Primary structure effects on peptide group hydrogen-exchange. Proteins: Struct., Funct., Genet. 17, 75–86.

401

24. Hernández, G., Anderson, J.S., and LeMaster, D.M. (2008) Electrostatic stabilization and general base catalysis in the active site of the human protein disulfide isomerase a domain monitored by hydrogen exchange. ChemBioChem 9, 768–778. 25. Radford, S.E., Buck, M., Topping, K.D., Dobson, C.M., and Evans, P.A. (1992) Hydrogen exchange in native and denatured states of hen egg-white lysozyme. Proteins 14, 237–248. 26. Hollien, J., and Marqusee, S. (2002) Comparison of the folding processes of T. thermophilus and E. coli Ribonucleases H. J. Mol. Biol. 316, 327–340. 27. Bau, R., Rees, D.C., Kurtz-Jr., D.M., Scott, R.A., Huang, H.S., Adams, M.W.W., and Eidsness, M.K. (1998) Crystal-structure of rubredoxin from Pyrococcus furiosus at 0.95 Angstrom resolution, and the structures of N-terminal methionine and formylmethionine variants of Pf Rd. Contributions of N-terminal interactions to thermostability. J. Biol. Inorg. Chem. 3, 484–493. 28. LeMaster, D.M., Tang, J., Paredes, D.I., and Hernández, G. (2005) Enhanced thermal stability achieved without increased conformational rigidity at physiological temperatures: Spatial propagation of differential flexibility in rubredoxin hybrids. Proteins 61, 608–616. 29. Lee, B., and Richards, F.M. (1971) The Interpretation of Protein Structures: Estimation of Static Accessibility. J. Mol. Biol. 55, 379–400. 30. Hiller, R., Zhou, Z.H., Adams, M.W.W., and Englander, S.W. (1997) Stability and dynamics in a hyperthermophilic protein with melting temperature close to 200 degrees C. Proc. Natl. Acad. Sci. USA 94, 11329–11332. 31. Richter, B., Gsponer, J., Varnai, P., Salvatella, X., and Vendruscolo, M. (2007) The MUMO (minimal under-restraining minimal overrestraining) method for the determination of native state ensembles of proteins. J. Biomol. NMR 37, 117–135. 32. Lange, O.F., Lakomek, N.A., Fares, C., Schroder, G.F., Walter, K.F.A., Becker, S., Meiler, J., Grubmuller, H., Griesinger, C., and deGroot, B.L. (2008) Recognition dynamics up to microseconds revealed from an RDC-derived ubiquitin ensemble in solution. Science 320, 1471–1475. 33. Hernández, G., Anderson, J.S., and LeMaster, D.M. (2010) Assessing the native state conformational distribution of ubiquitin by peptide acidity. Biophys. Chem., doi: 10.1016/j. bpc.2010.10.006. 34. Anderson, J.S., Hernández, G., and LeMaster, D.M. (2009) Backbone conformational dependence of peptide acidity. Biophys. Chem. 141, 124–130.

402

G. Hernández et al.

35. Anderson, J.S., Hernandez, G., and LeMaster, D.M. (2010) Sidechain conformational dependence of hydrogen exchange in model peptides. Biophys. Chem. 151, 61–70. 36. Kim, P.S., and Baldwin, R.L. (1982) Influence of charge on the rate of amide proton exchange. Biochemistry 21, 1–5. 37. Tüchsen, E., and Woodward, C. (1985) Hydrogen kinetics of peptide amide protons at the bovine pancreatic trypsin inhibitor protein-solvent interface. J. Mol. Biol. 185, 405–419. 38. Delepierre, M., Dobson, C.M., Karplus, M., Poulsen, F.M., States, D.J., and Wedin, R.E. (1987) Electrostatic effects and hydrogen exchange behavior in proteins. The pH dependence of exchange rates in lysozyme. J. Mol. Biol. 197, 111–122. 39. Dempsey, C.E. (1995) Hydrogen bond stabilities in the isolated alamethicin helix: pHdependent amide exchange measurements in methanol. J. Am. Chem. Soc. 117, 7526–7534. 40. Forsyth, W.R., and Robertson, A.D. (1996) Intramolecular electrostatic interactions accelerate hydrogen exchange in diketopiperazine relative to 2-piperidone. J. Am. Chem. Soc. 118, 2694–2698. 41. Fogolari, F., Esposito, G., Viglino, P., Briggs, J.M., and McCammon, J.A. (1998) pKa shift effects on backbone amide base-catalyzed hydrogen exchange rates in peptides. J. Am. Chem. Soc. 120, 3735–3738. 42. Matthew, J.B., and Richards, F.M. (1983) The pH dependence of hydrogen exchange in proteins. J. Biol. Chem. 258, 3039–3044. 43. Eigen, M. (1964) Proton transfer, acid-base catalysis, and enzymatic hydrolysis. (I) Elementary processes. Angew. Chem. Int. Ed. 3, 1–19. 44. Molday, R.S., and Kallen, R.G. (1972) Substituent effects on amide hydrogen exchange rates in aqueous solution. J. Am. Chem. Soc. 94, 6739–6745. 45. Wang, W.H., and Cheng, C.C. (1994) General base catalyzed proton exchange in amides. Bull. Chem. Soc. Jpn. 67, 1054–1057. 46. LeMaster, D.M., Anderson, J.S., and Hernández, G. (2007) Spatial distribution of dielectric shielding in the interior of Pyrococcus furiosus rubredoxin as sampled in the subnanosecond timeframe by hydrogen exchange. Biophys. Chem. 129, 43–48. 47. Hernández, G., Anderson, J.S., and LeMaster, D.M. (2009) Polarization and polarizability assessed by protein amide acidity. Biochemistry 48, 6482–6494.

48. Luz, Z., and Meiboom, S. (1964) The activation energies of proton transfer reactions in water. J. Am. Chem. Soc. 86, 4768–4769. 49. Tolbert, L.M., and Solntsev, K.M. (2002) Excited-state proton transfer: From constrained systems to “Super” photoacids to superfast proton transfer. Acc. Chem. Res. 35, 19–27. 50. Leiderman, P., Genosar, L., and Huppert, D. (2005) Excited-state proton transfer: Indication of three steps in the dissociation and recombination process. J. Phys. Chem A 109, 5965–5977. 51. Ellison, W.J., Lamkaouchi, K., and Moreau, J.M. (1996) Water: A dielectric reference. J. Molec. Liquids 68, 171–279. 52. Marcus, R.A. (1964) Chemical and electrochemical electron-transfer theory. Annu. Rev. Phys. Chem. 15, 155–196. 53. Schaefer, M., and Karplus, M. (1996) A comprehensive analytical treatment of continuum electrostatics. J. Phys. Chem. 100, 1578–1599. 54. Richards, F.M. (1974) The interpretation of protein structures: Total volume, group volume distributions and packing density. J. Mol. Biol. 82, 1–14. 55. Tsai, J., Taylor, R., Chothia, C., and Gerstein, M. (1999) The packing density in proteins: Standard radii and volumes. J. Mol. Biol. 290, 253–266. 56. Mertz, E.L., and Krishtalik, L.I. (2000) Low dielectric response in enzyme active site. Proc. Natl. Acad. Sci. USA 97, 2081–2086. 57. Hawranek, J.P., Wrzeszcz, W., Muszyñski, A.S., and Pajdowska, M. (2002) Infrared dispersion of liquid triethylamine. J. Non-Crystal. Solids 305, 62–70. 58. LeMaster, D.M., Anderson, J.S., and Hernández, G. (2006) Role of native-state structure in rubredoxin native-state hydrogen exchange. Biochemistry 45, 9956–9963. 59. Antosiewicz, J., McCammon, J.A., and Gilson, M.K. (1994) Prediction of pH dependent properties of proteins. J. Mol. Biol. 238, 415–436. 60. Antosiewicz, J., McCammon, J.A., and Gilson, M.K. (1996) The determinants of pKas in proteins. Biochemistry 35, 7819–7833. 61. Demchuk, E., and Wade, R.C. (1996) Improving the continuum dielectric approach to calculating pKa’s of ionizable groups in proteins. J. Phys. Chem. 100, 17373–17387. 62. Georgescu, R.E., Alexov, E.G., and Gunner, M.R. (2002) Combining conformational flexibility and continuum electrostatics for

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

calculating pKas in proteins. Biophys. J. 83, 1731–1748. 63. Wisz, M.S., and Hellinga, H.W. (2003) An empirical model for electrostatic interactions in proteins incorporating multiple geometrydependent dielectric constants. Proteins 51, 360–377. 64. Song, Y., Mao, J., and Gunner, M.R. (2009) MCCE2: Improving protein pKa calculations with extensive side chain rotamer sampling. J. Comput. Chem. 30, 2231–2247. 65. Simonson, T., and Perahia, D. (1995) Internal and interfacial dielectric properties of cytochrome c from molecular dynamics in aqueous solution. Proc. Natl. Acad. Sci. USA 92, 1082–1086. 66. Harms, M.J., Schlessman, J.L., Chimenti, M.S., Sue, G.R., Damjanovic, A., and GarcíaMoreno, B.E. (2008) A buried lysine that titrates with a normal pKa: Role of conformational flexibility at the protein-water interface as a determinant of pKa values, Prot. Sci. 17, 833–845. 67. Bordwell, F.G., W. J. Boyle, J., and Yee, K.C. (1970) Equilibrium and kinetic acidities of nitroalkanes and their relationship to transition state structures. J. Am. Chem. Soc. 92, 5926–5932. 68. Bernasconi, C.F. (1987) Intrinsic barriers of reactions and the principle of nonperfect synchronization. Acc. Chem. Res. 20, 301–308. 69. Costentin, C., and Saveant, J.M. (2004) Why are proton transfers at carbon slow? Selfexchange reactions. J. Am. Chem. Soc. 126, 14787–14795. 70. Hwang, T.L., Mori, S., Shaka, A.J., and vanZijl, P.C.M. (1997) Application of Phase-Modulated CLEAN Chemical EXchange Spectroscopy (CLEANEX-PM) to detect water-protein proton exchange and intermolecular NOEs. J. Am. Chem. Soc. 119, 6203–6204. 71. Hwang, T.L., vanZijl, P.C.M., and Mori, S. (1998) Accurate quantitation of water-amide proton exchange rates using the phase-modulated CLEAN chemical EXchange (CLEANEX-PM) approach with a fast-HSQC (FHSQC) detection scheme. J. Biomol. NMR 11, 221–226. 72. Hernández, G., and LeMaster, D.M. (2003) Relaxation compensation in chemical exchange measurements for the quantitation of amide hydrogen exchange in larger proteins. Magn. Reson. Chem. 41, 699–702. 73. Griesinger, C., and Ernst, R.R. (1987) Frequency offset effects and their elimination in NMR rotating-frame cross-relaxation spectroscopy. J. Magn. Reson. 75, 261–271.

403

74. Chevelkov, V., Xue, Y., Rao, D.K., FormanKay, J.D., and Skrynnikov, N.R. (2010) N-15(H/D)-SOLEXSY experiment for accurate measurement of amide solvent exchange rates: application to denatured drkN SH3. J. Biomolec. NMR 46, 227–244. 75. Makhatadze, G.I., Clore, G.M., and Gronenborn, A.M. (1995) Solvent isotope effect and protein stability. Nat. Struct. Biol. 2, 852–855. 76. Connelly, G.P., Bai, Y.W., Jeng, M.F., and Englander, S.W. (1993) Isotope effects in peptide group hydrogen-exchange. Proteins: Struct., Funct., Genet. 17, 87–92. 77. Sivaraman, T., Arrington, C.B., and Robertson, A.D. (2001) Kinetics of unfolding and folding from amide hydrogen exchange in native ubiquitin. Nat. Struct. Biol. 8, 331–333. 78. Rocchia, W., Sridharan, S., Nicholls, A., Alexov, E., Chiabrera, A., and Honig, B. (2002) Rapid grid-based construction of the molecular surface and the use of induced surface charge to calculate reaction field energies: Applications to the molecular systems and geometric objects. J. Comput. Chem. 23, 128–137. 79. MacKerell.Jr., A.D., Bashford, D., Bellott, M., Dunbrack.Jr., R.L., Evanseck, J.D., Field, M.J., Fischer, S., Gao, J., Guo, H., Ha, S., Joseph-McCarthy, D., Kuchnir, L., Kuczera, K., Lau, F.T.K., Mattos, C., Michnick, S., Ngo, T., Nguyen, D.T., Prodhom, B., ReiherIII, W.E., Roux, B., Schlenkrich, M., Smith, J.C., Stote, R., Straub, J., Watanabe, M., Wiorkiewicz-Kuczera, J., Yin, D., and Karplus, M. (1998) All-atom empirical potential for molecular modeling and dynamics studies of proteins. J. Phys. Chem. B 102, 3586–3616. 80. Jorgensen, W.L., Maxwell, D.S., and TiradoRives, J. (1996) Development and testing of the OPLS all-atom force field on conformational energetics and properties of organic liquids. J. Am. Chem. Soc. 118, 11225–11236. 81. Cheatham, T.E., Cieplak, P., and Kollman, P.A. (1999) A modified version of the Cornell et al. force field with improved sugar pucker phases and helical repeat. J. Biomolec. Struct. Dynamics 16, 845–862. 82. Duan, Y., Wu, C., Chowdhury, S., Lee, M.C., Xiong, G.M., Zhang, W., Yang, R., Cieplak, P., Lou, R., Lee, T., Caldwell, J., Wang, J.M., and Kollman, P. (2003) A point-charge force field for molecular mechanics simulations of proteins based on condensed-phase quantum mechanical calculations. J. Comp. Chem. 24, 1999–2012. 83. Bayly, C.I., Cieplak, P., Cornell, W.D., and Kollman, P.A. (1993) A well behaved electrostatic potential-based method using charge

404

G. Hernández et al.

restraints for deriving atomic charges: The RESP model. J. Phys. Chem. 97, 10269–10280. 84. Becke, A.D. (1993) Density-functional thermochemistry III. The role of exact exchange. J. Chem. Phys. 98, 5648–5652. 85. You, T.J., and Bashford, D. (1995) Conformation and hydrogen ion titration of proteins: A continuum electrostatic model with conformational flexibility. Biophys. J. 69, 1721–1733. 86. vanVlijmen, H.W.T., Schaefer, M., and Karplus, M. (1998) Improving the accuracy of protein pKa calculations: Conformational averaging versus the average structure. Proteins 33, 145–158. 87. deGroot, B.L., vanAalten, D.M.F., Scheek, R.M., Amadei, A., Vriend, G., and Berendsen, H.J.C. (1997) Prediction of protein conformational freedom from distance constraints. Proteins 29, 240–251. 88. Cornilescu, G., Marquardt, J.L., Ottiger, M., and Bax, A. (1998) Validation of Protein Structure from Anisotropic Carbonyl Chemical Shifts in a Dilute Liquid Crystalline Phase. J. Am. Chem. Soc. 120, 6836–6837. 89. Boehr, D.D., Nussinov, R., and Wright, P.E. (2009) The role of dynamic conformational ensembles in biomolecular recognition. Nature Chem. Biol. 5, 789–796. 90. Mittermaier, A.K., and Kay, L.E. (2009) Observing biological dynamics at atomic resolution using NMR. Trends Biochem. Sci. 34, 601–611. 91. Dikic, I., Wakatsuki, S., and Walters, K.J. (2009) Ubiquitin-binding domains - from structures to functions. Nature Rev. Molec. Cell Biol. 10, 659–671. 92. Tjandra, N., Feller, S.E., Pastor, R.W., and Bax, A. (1995) Rotational diffusion anisotropy of human ubiquitin from 15 N NMR relaxation. J. Am. Chem. Soc. 117, 12562–12566. 93. Sheinblatt, M. (1970) Determination of an acidity scale for peptide hydrogens from nuclear magnetic resonance kinetic studies. J. Am. Chem. Soc. 92, 2505–2509. 94. Avbelj, F., and Baldwin, R.L. (2004) Origin of the neighboring residue effect on peptide backbone conformation. Proc. Natl. Acad. Sci. USA 101, 10967–10972. 95. Makowska, J., Rodziewicz-Motowid³o, S., Bagiñska, K., Vila, J.A., Liwo, A., Chmurzyñski, L., and Scheraga, H.A. (2006) Polyproline II conformation is one of many local conformational states and is not an overall conformation of unfolded peptides and proteins. Proc. Natl. Acad. Sci. USA 103, 1744–1749.

96. Chen, K., Liu, Z., Zhou, C., Bracken, W.C., and Kallenbach, N.R. (2007) Spin relaxation enhancement confirms dominance of extended conformations in short alanine peptides. Angew. Chem. Int. Ed. 46, 9036–9039. 97. Graf, J., Nguyen, P.H., Stock, G., and Schwalbe, H. (2007) Structure and dynamics of the homologous series of alanine peptides: A joint molecular dynamics/NMR study. J. Am. Chem. Soc. 129, 1179–1189. 98. Wickstrom, L., Okur, A., and Simmerling, C. (2009) Evaluating the Performance of the ff99SB Force Field Based on NMR Scalar Coupling Data. Biophys. J. 97, 853–856. 99. Pizzanelli, S., Forte, C., Monti, S., Zandomeneghi, G., Hagarman, A., Measey, T.J., and Schweitzer-Stenner, R. (2010) Conformations of Phenylalanine in the Tripeptides AFA and GFG Probed by Combining MD Simulations with NMR, FTIR, Polarized Raman, and VCD Spectroscopy. J. Phys. Chem B 114, 3965–3978. 100. Tsai, M., Xu, Y.J., and Dannenberg, J.J. (2009) Ramachandran Revisited. DFT Energy Surfaces of Diastereomeric Trialanine Peptides in the Gas Phase and Aqueous Solution. J. Phys. Chem B 113, 309–318. 101. Fitzkee, N.C., Fleming, P.J., and Rose, G.D. (2005) The Protein Coil Library: A structural database of nonhelix, nonstrand fragments derived from the PDB. Prot. Struct. Funct. Bioinform. 58, 852–854. 102. Avbelj, F., and Baldwin, R.L. (2006) Limited validity of group additivity for the folding energetics of the peptide group. Prot. Struct. Funct. Bioinform. 63, 283–289. 103. Flory, P.J. Statistical mechanics of chain molecules, (Wiley Interscience, New York, 1969). 104. Penkett, C.J., Redfield, C., Dodd, I., Hubbard, J., McBay, D.L., Mossakowska, D.E., Smith, R.A.G., Dobson, C.M., and Smith, L.J. (1997) NMR analysis of main-chain conformational preferences in an unfolded fibronectin-binding protein. J. Mol. Biol. 274, 152–159. 105. Keskin, O., Yuret, D., Gursoy, A., Turkay, M., and Erman, B. (2004) Relationships between amino acid sequences and backbone torsion angle preferences. Prot. Struct. Funct. Bioinform. 55, 992–998. 106. Jha, A.K., Colubri, A., Zaman, M.H., Koide, S., Sosnick, T.R., and Freed, K.F. (2005) Helix, sheet, and polyproline II frequencies and strong nearest neighbor effects in a restricted coil library. Biochemistry 44, 9691–9702.

20

Electrostatics of Hydrogen Exchange for Analyzing Protein Flexibility

107. LeMaster, D.M. (1999) NMR relaxation order parameter analysis of the dynamics of protein sidechains. J. Am. Chem. Soc. 121, 1726–1742. 108. Skrynnikov, N.R., Millet, O., and Kay, L.E. (2002) Deuterium spin probes of side-chain dynamics in proteins. 2. Spectral density mapping and identification of nanosecond

405

time-scale side-chain motions. J. Am. Chem. Soc. 124, 6449–6460. 109. Darley, M.G., and Popelier, P.L.A. (2008) Role of short-range electrostatics in torsional potentials. J. Phys. Chem A 112, 12954–12965. 110. Senn, H.M., and Thiel, W. (2009) QM/MM methods for biomolecular systems. Angew. Chem. Int. Ed. 48, 1198–1229.

Chapter 21 Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy Bernhard Brutscher and Ewen Lescop Abstract Probing protein structure, dynamics, and interaction surfaces by NMR requires initial backbone resonance assignment. The protocol for this step has been progressively developed in the last 15 years to provide robust assignments. However, even in the case of favorable conditions (high field magnets and cryogenically cooled probes, small globular proteins, high sample concentration), the assignment step generally takes several days of data collection and analysis, thus precluding studies of unstable proteins and limiting highthroughput applications. Recently, we have introduced the BATCH strategy for fast protein backbone resonance assignment. BATCH benefits from the combination of several tools (BEST/ASCOM/TargetedSampling/COBRA/HADAMAC) for time-optimized and highly automated NMR data acquisition, processing, and analysis. In this chapter, we discuss the individual steps of the BATCH method and describe its practical implementation to obtain the backbone resonance assignment of small globular proteins in a few hours of time. Key words: Protein, Fast NMR, Resonance assignment, Chemical shift, Amino acid type discrimination, Algorithm

1. Introduction Protein NMR studies usually start by assigning the backbone resonances required for the subsequent measurement of residue-specific parameters that are related to protein structure and dynamics. In many cases, the assignment of the 15N HSQC spectrum is sufficient, although 13CO, 13Ca, and 13Cb chemical shifts are also desirable for secondary structure assessment and for structure determination. Over the years, a robust strategy has been developed to assign backbone resonances of uniformly 15N/13C labeled proteins. In this strategy, a series of pairs of triple resonance experiments are collected. These experiments are usually performed to record three-dimensional (3D) datasets correlating the 1H, 15N chemical shifts of one amino acid and the 13CO, 13Ca, and/or 13Cb chemical

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_21, © Springer Science+Business Media, LLC 2012

407

408

B. Brutscher and E. Lescop

shifts of the same amino acid (intraresidual correlation) or of the previous amino acid (sequential correlation) in the peptide sequence. The analyses of these 3D spectra provide the connectivity information for 1H/15N frequency pairs corresponding to neighboring amino acids, as well as the 13C chemical shifts that contain some information on the amino acid type of the corresponding residue. The combination of sequential connectivity and amino acid type information, in addition to the known protein sequence, is virtually sufficient to obtain complete backbone assignment and is easily obtained using one of the numerous dedicated computer software packages available for this purpose. This standard strategy is very robust for a wide range of protein samples (MW, concentration). However, for a favorable experimental setup providing high sensitivity (related to protein size, concentration, and NMR hardware), this strategy is far from optimal in terms of overall experimental time. In such a situation, the acquisition time needed for many 3D triple resonance experiments is essentially limited by the timeconsuming sampling of the incremented time domains needed to achieve sufficient resolution along all spectral dimensions. In addition, amino acid type discrimination based on 13C chemical shifts only is highly ambiguous: while glycine, alanine, serine, and threonine residues can be reasonably well differentiated, other amino acid types cannot be unambiguously identified, and a probabilistic treatment is required for the assignment algorithm. Bearing in mind the limitations of the commonly used assignment protocols, we and others have developed techniques to improve data collection and analysis for fast and highly automated backbone resonance assignment (1–3). In this chapter, we focus on the BATCH strategy (3) recently developed in our laboratories. The BATCH strategy has been designed to collect a minimal dataset sufficient to achieve complete backbone resonance assignment. BATCH benefits from a suite of individual spectroscopic and computational tools, each devoted to the time optimization of a specific task. In BATCH, sequential connectivity and amino acid type information are obtained from separate types of experiments, a situation that contrasts with the conventional strategy. The BATCH strategy is particularly efficient and robust for fast assignment of small-to-medium sized 15N/13C labeled proteins.

2. Description of the Various Tools Implemented in BATCH

In this section, we briefly describe the set of methods implemented in BATCH allowing for significantly faster data collection and analysis. We emphasize the main characteristics of the introduced techniques and describe their benefits in the context of the BATCH strategy.

21

Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy

409

For more details, the interested reader is advised to refer to the original publications of the BATCH protocol (3) and of the respective tools (4–8). 2.1. The BEST Principle for Fast Pulsing Multidimensional NMR

Sequential resonance assignment is based on a set of 3D triple resonance (H–N–C) experiments. One crucial parameter governing the overall experimental time is the interscan (or recycling) delay that is required for magnetization recovery between successive repetitions of the pulse sequence. In order to maximize the signal-to-noise ratio per unit of time (sensitivity), this delay is usually set to ~1–1.5 s, accounting for the average 1H T1 in proteins. It has been demonstrated that in experiments that excite and detect amide protons, amide 1H T1 can be significantly reduced by leaving all nonamide protons unperturbed throughout the pulse sequence (9–11). This so-called longitudinal relaxation enhancement effect is mainly due to 1H–1H dipolar interaction-mediated polarization (energy) transfer from the excited amide 1H to other nearby 1H in thermal equilibrium and chemical exchange between labile amide protons and water protons. This observation has led to the development of BEST-type HNC correlation experiments used for backbone assignment (6, 7). BEST pulse sequences differ from conventional pulse sequences by the extensive use of shaped pulses for amide 1H band selective excitation, inversion, and refocusing (PC9 (12), EBURP2 (13) and REBURP (13) shapes), as well as pairs of broadband inversion pulses (BIP (14)) that achieve minimal perturbation of aliphatic and water 1H magnetizations. Composite broadband 1H decoupling sequences, such as WALTZ or DIPSI are incompatible with the BEST effect. Therefore another difference in BEST pulse sequences is the use of a simple 1H inversion (BIP) pulse for refocusing 1H–15N or 1H–13C coupling evolution. This slightly reduces the overall sensitivity of BEST triple-resonance experiments for fast relaxing systems such as high molecular weight proteins. In BEST experiments, maximal sensitivity is obtained for interscan delays of ~200–400 ms. This yields a reduction in the overall experimental time by a factor of ~3 without compromising spectral resolution and sensitivity.

2.2. ASCOM 15N Spectral Width for Rapid Sampling of the 15N Dimension

3D triple resonance experiments can be viewed as a repetitive recording of 15N HSQC spectra with an amplitude modulation of the 1H–15N correlation peaks according to the 13C chemical shift evolution during an incremented time delay. Therefore, once the 1 H–15N chemical shift pairs are known from a single 15N HSQC spectrum, this information can be exploited to optimize the 15N spectral width for subsequent triple-resonance experiments. For a spectral width that has been deliberately chosen to be smaller than the actual chemical shift range, the position of all peaks in the spectrum can be accurately predicted from the well-known aliasing property of complex Fourier transformation. The ASCOM tool (8)

410

B. Brutscher and E. Lescop

has been developed for optimal spectral compression leading to the smallest possible 15N spectral width without creating any additional peak overlap. In practice, the 15N HSQC spectrum is subjected to peak-picking. The resulting peak list is then used to simulate and analyze the 15N HSQC spectrum with various 15N spectral widths. The smallest 15N spectral width value providing no new peak overlap is selected. The optimized 15N spectral width can then be used in subsequent 3D triple resonance experiments to reach the same 15 N spectral resolution with a reduced number of increments (reduced experimental time). 2.3. COBRA for Automated Extraction of Sequential Connectivities

In the conventional approach, peak picking of the 3D Fouriertransformed (FT) matrices is usually performed after data collection, and 13C chemical shifts are compared between spectra to identify sequential connectivities. To facilitate this assignment step, we introduced the COBRA algorithm (5) that directly extracts sequential correlation information from the raw data. During the COBRA procedure, the 3D time domain data are Fourier transformed along the 1H and 15N dimensions, and the 13C time domain signal is extracted for each of the n individual 15N HSQC cross peaks (residues). Then, for a pair of intraresidue and sequential experiments, a correlation (COBRA) map is computed to yield a n × n matrix with elements given by the (weighted) correlation coefficients LC(i,j) of the two time domain signals, Si(t) and Ij(t), extracted from the sequential and intraresidual experiments at the 1 H/15N positions of cross peaks i and j, respectively. The following equation is used for LC(i,j): LC (i , j ) = corr (Si , I j ) * e

⎛ f ⎞ −⎜ ⎝ fcut ⎟⎠

4

where f refers to the angular phase of the complex correlation coefficient (corr) of the two traces Si(t) and Ij(t). corr is calculated as cov(Si , I j ) / var(Si )var(I j ) , with cov and var being the covariance and variance of the two signals, respectively. The LC coefficient is a real number and ranges from 0 (for uncorrelated signals, i.e., weak “probability” for sequential connectivity between 1 H/15N cross peaks j i) to 1 (for perfect correlation, i.e., possible “physical” connectivity). The phase weighted function depends on a user adjustable fcut phase cutoff parameter and allows better frequency discrimination (resolution) for close but distinct frequencies (corr close to 1 and f close to 0°) while leaving the coefficient unchanged for identical frequency composition (f = 0°). A small value of fcut leads to improved frequency discrimination. However, when taking noise into account, too small a fcut value may lead to vanishing COBRA element values for traces containing peaks of identical frequency but low signal intensity. Overall, the informational content of the COBRA map is limited by the experiment

21

Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy

411

with the lowest signal-to-noise ratio. As an intuitive but extreme illustration, the COBRA map calculated on two H–N–C experiments with one of them containing only noise is completely uninformative. In the current version of COBRA, the phase cutoff parameter is set by the user and has to be manually optimized for the signal-to-noise ratio of the two experiments. An attempt for automated optimization of this adjustable parameter is also included in the current version of BATCH and will be presented elsewhere. When several pairs of experiments are available, all individual maps (corresponding to different 13C nuclei) are combined together, resulting in a final COBRA map representing the sequential connectivity information contained in the whole set of NMR data. COBRA performs best with traces containing a minimal number of signals. Therefore, intraresidue experiments (e.g., iHNCA) are preferred over their bidirectional counterparts (e.g., HNCA), and the CA CB transfer in HN(CO)CACB and iHNCACB is adjusted such to select only correlations with CB and not with CA frequencies (except for glycine residues). The extraction of sequential connectivities from triple resonance experiments using the COBRA approach is very fast (a few seconds of time) and is used in the BATCH strategy to significantly speed up NMR data analysis. 2.4. Targeted Sampling Approach for Rapid Sampling of the 13C Dimension

In contrast to FT processing, the COBRA processing tool is compatible with regular but incomplete sampling of the indirect 13C dimensions (5). Typically, the 13C time domain is sampled over two time regions (targeted sampling): t1 = 0 to 3 ms (standard STD region) and t1 = 25 to 28 ms (time-shifted TS region). This second time window is chosen such that the signal attenuation due to evolution under 13Ca–13Cb scalar coupling is negligible: (cos (pJ CCt 1 ) ≅ −1). A COBRA map is calculated separately for each time window and subsequently combined. The additional sampling of long t1 values in the TS time window improves frequency discrimination in the final COBRA map compared with the STD window-only based COBRA map. The targeted sampling approach combined with COBRA analysis provides a high level of frequency discrimination while requiring only a small number of t1 increments in the 13C dimension. In practice, the targeted sampling scheme is applied to the most sensitive pair of experiments (H–N–CA).

2.5. The HADAMAC Experiment for Amino Acid Typing

Amino acid type information for individual residues (15N HSQC cross peaks) is required to assign fragments of sequentially connected 1H–15N frequency pairs to a particular location on the protein sequence. In the BATCH strategy, we use the recently introduced HADAMAC experiment (4) for amino acid-type discrimination. In this experiment, the 15N HSQC cross peaks are edited along an additional “amino acid-type” dimension according

412

B. Brutscher and E. Lescop

to the amino acid type of the preceding residue in the protein sequence. The HADAMAC experiment results in six 1H–15N planes corresponding to the following amino acid groups: (1) Gly, (2) Ser, (3) Thr, (4) Ala/Val/Ile(AVI), (5) Asn/Asp(Asx), and (6) Cys and Aromatic residues (Cys-Arom), and all other residues (Rest). In plane (6) cross peaks corresponding to the Cys-Arom and Rest groups have opposite signs, allowing their separation. The HADAMAC pulse sequence, based on a CBCACONH transfer experiment, exploits the differences between the 13Cb chemical shift range and spin topology (numbers of 1H attached to 13Cb and 13C a atoms; numbers and types (carbonyl, aromatic, or aliphatic) of 13Cg atoms to achieve this selection. Four spin manipulations (filters) are independently applied to change the relative signs of signals originating from different amino acid groups. The filters are applied using a Hadamard encoding scheme to ensure maximal sensitivity in a short overall experimental time. Recently, we have introduced an improved version of the HADAMAC experiment that yields slightly improved sensitivity (HADAMAC-2 (3)), especially for larger proteins. The principal advantage of HADAMAC with respect to conventional 13C chemical shift based methods is that a high level of amino acid type discrimination is achieved from a simple visual (or automated) inspection of the six HADAMAC 1 H–15N planes. 2.6. 13C Chemical Shift Extraction

To extract the 13C chemical shift information present in the available triple resonance experiments, an efficient algorithm has been developed (3). Briefly, 1D 13C time domain data are extracted at the 1H/15N position of each 15N HSQC cross peak from a given 3D experiment. These traces are then subjected to Fourier transformation, resulting in complex valued S(w) traces. Analogous to COBRA4 ⎛ f (ω) ⎞ −⎜ f ⎟

processing, the traces are weighted as F (ω) = ℜ(S (ω))* e ⎝ cut ⎠ where f(w) corresponds to the angular phase of S(w). The phase weighting transformation greatly improves frequency resolution. Using the same fcut as for COBRA processing is recommended. In the case of delayed acquisition (TS region defined as t1 = t10 + k*Dt1), zero and first order phase corrections are required and are calculated as f0 = t10/Dt1*p and f1 = t10/Dt1*2p. An additional signal inversion (180° zero-order phase correction) is applied to account for sign inversion due to 13C–13C scalar coupling evolution. When the STD and TS regions are collected for the H–N–CA pair, the traces obtained from the two regions are multiplied point-by-point, resulting in a single trace. In principle, a given 13C nucleus gives rise to peaks in sequential and intraresidue experiments at the 1 H/15N positions of sequential residues. Prior to chemical shift extraction, the two corresponding traces are also multiplied. One 13 C chemical shift is extracted for every trace (residue) as the frequency of maximum (absolute) amplitude.

21

Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy

413

3. Experimental Setup 3.1. Protein Sample

Protein backbone resonance assignment using the BATCH strategy requires a 15N/13C labeled protein sample, typically in the concentration range of 100 mM to a few mM. Since high-level deuteration of aliphatic protons is incompatible with the BATCH strategy, this approach is best suited for proteins below ~20 kDa. Aliphatic protons are required as a relaxation source for amide protons in BESTtype triple-resonance experiments, and as starting magnetization for the amino acid-type edited HADAMAC experiment. A file containing the sequence of the protein in NMRView format is also needed.

3.2. NMR Hardware and Pulse Sequences

The time efficiency of the assignment relies on the spectrometer sensitivity. Therefore high magnetic fields equipped with cryogenic probes are preferred. A 600 MHz spectrometer may represent a good trade-off to limit the loss of sensitivity due to the CSAinduced increase in 13CO transverse relaxation at high field strengths. In the following sections, we assume that the spectrometer (Bruker or Agilent) is equipped with Topspin 2.1 (or newer) or VNMRJ (installed with the last version of BioPack) software. We also assume that pulses on the 15N and 13C channels have been correctly calibrated and that 1H amplifiers have been linearized. This is particularly important for automated calibration of shaped pulses in BEST-type sequences. For the BATCH strategy, several NMR pulse sequences should be available on the spectrometer. They include the BEST versions for 15N HSQC and the following triple resonance experiments: HN(CO)CA, iHNCA, HN(CO)CB, iHNCB, HNCO, and iHNCO as well as HADAMAC. The iHNCA (iHNCB, and iHNCO) only retains the intraresidual H–N–C coherence transfers as described elsewhere (15). The HN(CO)CB experiment is identical to HN(CO)CACB with the exception of the longer Ca→Cb transfer delay for performing a full CA→CB transfer. Therefore, in HN(CO)CB (and iHNCB) only one cross peak is present for each 1 H/15N frequency pair. In principle, the sensitivity enhanced versions of these pulse sequences should be used. However, for fast relaxing systems, shorter INEPT-based versions can be used to obtain higher signal to noise ratios. The HADAMAC experiment should be collected first on a protein sample with known resonance assignments for the assignment of each plane to the corresponding amino acid group. For VNMRJ users, all pulse sequences are provided within the latest updated Biopack pulse sequence library. For Bruker users, most experiments are directly available for Topspin 2.1 and later versions. Pulse sequences are also available upon request from the authors of this chapter.

414

B. Brutscher and E. Lescop

3.3. Software

The BATCH strategy requires the ASCOM tool for 15N spectral optimization, which consists of a simple Perl script that can be run on the spectrometer. The software is available at the Web site http:// www.icsn.cnrs-gif.fr/download/nmr. For Agilent spectrometers equipped with the most recent versions of BioPack, the macro BestSW allows for automated peak picking of 2D 15N HSQC spectra and provides the ASCOM optimized 15N spectral width. The BATCH software platform is written in Tcl language and is embedded in the NMRView software (http://www.onemoonscientific. com/nmrview) (16) as a new functionality to make use of the variety of NMRView native functions. BATCH was validated for the Aqua (for MaxOS) and C (Linux system) versions of the NMRView software and is also available for the platform independent Java version. The BATCH software package can be downloaded from the Web site http://www.icsn.cnrs-gif.fr/download/nmr. The downloaded package includes instructions for installation, a manual, as well as a tutorial. The NMRPipe processing software (http://www.nmrscience. com/nmrpipe.html) (17) is also required to execute scripts generated by BATCH. The final assignment can be carried out using the native algorithm embedded in the BATCH software. Alternatively, the BATCH software provides a convenient interface for the Mars ( http://www.mpibpc.mpg.de/groups/zweckstetter/_links/ software_mars.htm) (18) or SmartNotebook (http://www.bionmr. ualberta.ca/bds/software/snb/) (19) software packages that may also be installed. For the following description, we assume that the user has a minimal proficiency with NMRView and NMRPipe software.

4. NMR Data Collection In the following sections, specific commands for Topspin and VNMRJ are given in small caps and italics, respectively, for example: PULSECAL and pulsecal. 4.1. Experimental Setup and Pulse Calibration

After sample injection and temperature regulation, the probe is tuned and shimmed. The 90° 1H pulse is calibrated, typically by determining the 360° pulse duration measured on the on-resonance water frequency or by using the Bruker PULSECAL tool. A water suppressed 1H spectrum is collected as a first evaluation of spectral quality, and for 1H chemical shift calibration in case of the presence of an internal reference (e.g., DSS, TSP).

4.2. The 15N HSQC Experiment

In a new folder, load the 15N HSQC experiment (RPAR BHSQC, best_Nhsqc). Amide 1HN band selective pulses are adjusted to cover the 1HN spectral width (typically 4 ppm centered at 8.5 ppm). Care is taken, however, to avoid water saturation for an efficient BEST effect and solvent suppression. Correct pulse calibration for 1H and

21

Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy

415

15

N channels are set (GETPROSOL 1H 9 1.5; pw = 9). For Bruker, the getprosol command automatically generates the power level for all shaped pulses present in the loaded pulse sequence. For VNMRJ, the power levels are automatically adjusted “on the fly”using the Pbox tool. The 15N HSQC experiment is run using the following parameters by default: recycling delay (d1) of 250 ms, 15N spectral width of 35 ppm centered at 118 ppm, 2 scans, 16 dummy scans, and 80–100 increments in the 15N dimension. For cryogenic probe safety, it is advised to reduce the acquisition time (AQ, at) to less than 100 ms and the power level of the 15N decoupling sequence applied during acquisition. The 15N HSQC is acquired within ~1–2 min. After double FT processing, the spectrum is visually inspected in terms of averaged signal-to-noise ratio, intensity heterogeneity, and spectral dispersion to assess the amenability of sequential resonance assignment through the BATCH approach. Possible aliasing of backbone 1H/15N cross peaks is detected and the experiment is possibly run again with increased 15N spectral width. This step is important to extract exact 15N frequencies before ASCOM optimization. Care should be taken for side chain H–N moieties (arginine, asparagine, and glutamine residues). The automated optimal 15N spectral width is obtained by ASCOM either “on the fly” (BestSW) or after peak picking and analysis (Bruker), and should be memorized for the following experiments. The extent of spectral compression by ASCOM software should be carefully set. In the course of COBRA processing and for increasing the signal-to-noise ratio, the residue specific 13C time domain is obtained by the integration of 3D matrices along the 1H and 15N dimensions (for the typical 1H/15N line widths) around the 1H/15N chemical shifts position. To avoid cross talk between 13C time domains corresponding to different residues, the cross peaks should be as separate as possible in the 1H/15N planes. Therefore, the cutoff parameters (RH and RN), defining whether two 1H/15N cross peaks overlap or not in ASCOM software, are set to the averaged 1 H and 15N line widths to limit partial peak overlapping. 4.3. The HADAMAC Experiment

In a new folder the HADAMAC pulse sequence (RPAR HADAMAC, hadamac) is loaded, preferably the HADAMAC-2 version for increased sensitivity. Set the correct pulse power values (GETPROSOL 1H 9 1.5; pw = 9), the number of repetitions to 2, the relaxation delay d1 to 0.8–1 s, the 15N spectral width (ASCOM-optimized or not) and the maximum number of 15N increments as allowed by the constant-time period inserted in the pulse sequence for increased resolution. For VNMRJ, set had_flg to (1, 2, …, 8), phase = 1,2 and array=“(had_flg, phase).” Depending on the chosen 15N spectral width, the HADAMAC experiment lasts from ~30 min to 1 h. This time can be used for the initial setup of BATCH, inspection of the 15N HSQC spectrum (Subheading 5.2), and the setup of triple resonance experiments.

416

B. Brutscher and E. Lescop

4.4. The Triple Resonance Experiments

In a new folder, the sequential BEST HN(CO)CA pulse sequence is loaded (RPAR BHNCOCA, best_hncocaP). Set the correct pulse power values (vide supra), the number of repetitions to 2, the relaxation delay (d1) to 250 ms, the ASCOM-optimized 15N spectral width with the maximum number of increments as allowed by the constant-time period inserted in the pulse sequence, and a typical 13C spectral window (20 ppm) centered at 56 ppm. The initial evolution delay under 13C chemical shifts is set to 0 (d0 = 0, d1 = 0), and 10 complex points are recorded to sample the 0…3 ms time region (for a 600 MHz 1H field strength). Repeat the procedure for the BEST iHNCA pulse sequence. This pair of experiments represents the standard (STD) H–N–CA pair. For correct COBRA processing, the 13C time domain of the sequential and intraresidual experiments should be collected in an exactly identical manner, including the pulse sequence element for 13C chemical shift evolution, sampling points, and the 13C carrier frequency. The phase weighted correlation coefficient embedded in COBRA is able to discriminate 13C signals that are 180° phase shifted (corresponding to peaks of opposite sign in frequency domain). Such sign inversion occurs, for example, when the first t1 time value is set to half dwell in the 15N dimension in the case of aliased cross peaks. For correct COBRA processing, the first t1 time value should be set to 0 ms. Using the same procedure, the time shifted (TS) H–N–CA pair of experiments is prepared in two additional folders. This pair is exactly identical to the previous STD H–N–CA pair except that the initial evolution delay under 13C chemical shifts is set to 25 ms (d0 = 25 m, d1 = 25 m) instead of 0 ms. Alternatively, the TS time region can be stored in the same files as the STD H–N–CA and can also be analyzed by the BATCH software. The same procedure is then applied to prepare the H–N–CB pair, consisting of the BEST HN(CO)CB and iHNCB pulse sequences and the H–N–CO pair, consisting of the BEST HNCO and iHNCO pulse sequences. For the 13Cb dimension, set the spectral window to 60 ppm centered at 46 ppm and collect 30 complex points to sample the t = 0…3 ms time region (for a 600 MHz 1H frequency). For the carbonyl 13C dimension, set the spectral window to 10 ppm centered at 176 ppm and collect 10 complex increments. Owing to the intrinsic low sensitivity of the iHNCB and the iHNCO experiments, the number of transients may be set to 4. Launch all experiments in a row. The time required for acquisition of the triple resonance data can be used to process and analyze the HADAMAC experiment (Subheading 5.2). After completion of the first pair of triple-resonance experiments (STD H–N–CA), they can be processed (Subheading 5.2), analyzed using the COBRA algorithm (Subheading 6.2), and a first attempt at resonance assignment can be done (Subheading 7). Every newly collected pair can be immediately processed and

21

Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy

417

analyzed with the BATCH software. Data are collected in an iterative way until complete assignment is achieved. In case resonance assignment is not complete upon the collection of the entire series of experiments, the spectroscopist has to identify the reasons and eventually collect additional data for increased signal-to-noise ratio. For example, less than 10–20% assigned resonances likely suggest low signal-to-noise ratio in at least one of the datasets that may be collected again and added up to the previously collected datasets.

5. Setup and Raw Data Processing in BATCH

5.1. General Parameters

Here we assume that NMRView, including the BATCH module, has been successfully installed on a computer (either on the spectrometer or a remote personal computer), and that the collected raw data are accessible on the same computer. We further assume that all other software packages, mentioned in Subheading 3.3, have been successfully installed and that the main BATCH window is open (Analysis > BATCH, Fig. 1a). This window consists of several sections for parameter settings and specific actions required for data processing and analysis. It provides a convenient interface to the external NMRPipe, MARS, and SmartNoteBook software packages. We illustrate the BATCH backbone assignment strategy on NMR data collected on the 77-residue Hyl1 protein (20). The data include the 15N HSQC, the HADAMAC and the pair of H–N–CA experiments with 13C dimensions sampled on the STD (0…3 ms) and TS (25…28 ms) time regions. The experimental data are also provided in the BATCH software package as a tutorial and knowledge training. In the following sections, the button clicking action is indicated as underlined command (ex: Load Setup). A few paths have to be set in the PATH Window (Fig. 1b): the location of the directory containing BATCH scripts (see installation procedure), the base directory for NMR data containing one folder per experiment and the sequence file. An additional path can be set for already existing 2D peak lists in the xpk (NMRView) format. Define the names of the folders containing the collected experiments. Additional parameter definitions will be automatically set in the course of preprocessing. The last section of this window allows the setting of a signal-free region of the 15N HSQC and of the integration box. The signal-free region is defined by a 1H/15N cross peak position and is used to estimate the noise in H–N–C spectra for the automated phase cutoff setting algorithm during COBRA processing. Owing to the extensive folding in 15N dimension in H–N–C experiments, it is advised to locate the cross peak in a signal-free region

418

B. Brutscher and E. Lescop

Fig. 1. (a) Main graphical interface of the BATCH software. (b) Window for paths definition and peak integration. (c) Window for rapid preprocessing of NMR data.

of the 1H projection of the 15N HSQC. The integration box refers to the spectral integration of the 3D H–N–C spectra around each 1 H/15N cross peak for 13C time domain extraction (by default 0.015 ppm and 0.1 ppm in 1H and 15N dimensions, respectively).

21

Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy

419

All parameter settings can be saved (Save Setup) in the current directory for future usage (Load Setup). If available, a 2D peak list can be loaded from the main window (ReadXPK), and the FT processed 15N HSQC spectrum can be automatically loaded in a new window (LoadHSQC). 5.2. Preprocessing of the Raw Data

The 15N HSQC, the HADAMAC and the triple resonance experiments require different preprocessing steps: the 15N HSQC spectrum is processed using conventional double FT, the HADAMAC experiment is subjected to double FT followed by Hadamard decoding, and triple resonance experiments are Fourier transformed along the 1H and 15N dimensions, while leaving the 13C dimension in the time domain for COBRA processing (files stored in the cobra folder). An additional Fourier transformation is also applied along the 13C dimension of triple resonance experiments (files stored in ft folder and converted to the NMRview format) for visualization (Load NvFile). Preprocessing is performed with NMRPipe in two steps: data conversion to NMRPipe format (fid. com script file) and data manipulation (nmrproc.com script file). Since for a given experimental setup (spectrometer and pulse sequences), the same NMRPipe processing script files can be used with minor modifications, the BATCH scripts directory contains one nmrproc.com-type file for each experiment. These files are modified and validated for the specific experimental setup only once and can be trustfully used for following protein studies. We put emphasis on the adequate processing of the HADAMAC experiment for correct subspectra assignment. The preprocessing step, facilitated by the “Process Window” (Fig. 1c), consists of going through each collected and previously defined experiment. Basically, for each recorded dataset, an nmrDraw window is opened (fid.com) for the preparation of the conversion script. The label of the 1H and 15N dimensions should be consistent over all experiments and should be set to “HN” and “N15” respectively. After adequate parameter adjustment, the conversion is executed. In the next step, the processing file is opened (nmrproc.com). This file may be chosen from different locations: default script library, folder of the current experiment or from the STD H–N–CA experiment (in the case of TS H–N–CA). The nmrproc.com script is then adjusted for the specific protein (mainly 1H phase correction), saved in the correct subdirectory and executed by NMRPipe (Save & Execute). The individual spectra can be inspected in NMRDraw (Check in NMRDraw) for optimal processing. This initial setting is carried out without linear prediction (LP) in the 15N dimensions to save time. An additional button (Process All with LP) allows reprocessing of all datasets with additional LP. During the preprocessing step, the paths to all newly generated files are automatically set (see Path window).

420

B. Brutscher and E. Lescop

6. 2D PeakPicking, HADAMAC Analysis and COBRA Calculation 6.1. 2D Peak List Generation and Extraction of Amino Acid Type

One unique (2D) peak list is required in BATCH, containing the protein’s 1H/15N chemical shifts. Although such a peak list can be obtained directly from the 15N HSQC through the 2D peak-picking procedure embedded in NMRView, the better spectral dispersion in the pseudo-3D HADAMAC spectra makes this experiment well suited to discriminate partially overlapping peaks. The HADAMAC subspectra are loaded (Show HADAMAC spectra) and inspected for spectral quality. A dedicated algorithm for the HADAMAC-based 2D 1H/15N peak list generation is called (AutoPeakPick), and the resulting peak list is inspected and manually adjusted (Fig. 2b). The peak list (defined as the Current 2D peak list in Fig. 1a) is then visualized on the 15N HSQC spectrum (LoadHSQC). Possible folded peaks in the 15N dimensions are identified and unfolded. More details about this procedure (specific to NMRView) are provided in the BATCH manual. The 2D peak list should contain exact (nonfolded) 15N frequencies for the subsequent extraction of 13C time domain at the correct 1H/15N chemical shifts position in triple resonance experiments. The individual box sizes, used for the extraction of amino acid type information, should be manually adjusted for partially overlapping cross peaks. The cross-peak-dependent amino acid type information is stored into the Comment section of the NMRView peak list window (Transfer HADAMAC information to Current Peaklist).

Fig. 2. (a) 15N HSQC spectrum of Hyl1. (b) Overlay of the HADAMAC subspectra of Hyl1. The amino acid groups are colored as follows: AVI (black), Gly (red), Thr (yellow), Ser (Blue), Asx (Magenta), Cys-Arom (Green), and Rest (Cyan). Of note, the cross peaks corresponding to the Cys-Arom and Rest groups are present in the same plane but with opposite signs. The result from the automated peak picking is shown as boxes with (random) numbering.

21

Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy

421

The algorithm for amino acid type extraction is fairly conservative, and for the difficult cases of overlapping peaks (as listed in NMRView Console), the assignment should be checked manually. The HADAMAC-derived information provides a very strong constraint for the backbone assignment and should be carefully validated before pursuing with the COBRA analysis. 6.2. COBRA Analysis

The COBRA maps are calculated from the main BATCH window (Fig. 1a). Each pair of H–N–C experiments (i.e., one COBRA map) corresponds to one line allowing the setting of three parameters: the first point (usually 1), the number of points in the 13C time domain (−1 to use all available data) to be used for COBRA calculation and the phase cutoff parameter fcut. In case the STD and TS sampled regions are stored in the same file for the H–N–CA pair, an additional list of 13C evolution delays can be defined on the same dataset. Only selected pairs of experiments are used for COBRA calculation (Calculate COBRA maps) and the individual and final maps are displayed (Show COBRA maps). At the same time the contour level for the visualization is automatically set to have on average ~1–2 visible COBRA elements per column and row (Fig. 3a), and can be further adjusted using classical NMRView tools. The contour level will be used later for binning. Spectral noise present in the individual triple resonance experiments propagates through the COBRA calculation. Noise in experimental 13C traces translates into overall reduction of COBRA elements values. In addition, noise leads to the deviation of the angular phase f value of the correlation coefficient corr when compared with noisefree 13C traces. This deviation is larger for low “averaged” signalto-noise ratio and has to be taken into account for setting the phase cutoff parameter fcut. Default initial cutoff phase values have to be set to low values (e.g., 15°). In absence of assignment, the cross peak numbering is random and real (physical) connectivities are not apparent as a (shifted) diagonal. In addition, for nonoptimized phase settings, several nonphysical connectivities (reconstruction artifacts) may appear as additional nonzero values in the COBRA map and/or correct connectivites may be absent. For these reasons, the informational content of a given map is difficult to assess visually. To facilitate the COBRA processing, a new peak list can be generated (Order Peaklist) that differs from the current one by the cross peak numbering (ordering). The new order of the peaks is guided by the unambiguous connectivities extracted from the current COBRA map to create unambiguously defined fragments of different lengths. An unambiguous connectivity between cross peaks i and j is obtained if the M(i,j) element is larger than the cutoff value (defined in the main BATCH window) and if no other values higher than the cutoff value exist along the ith column and the jth row. This intensity cutoff value is automatically set during processing

422

B. Brutscher and E. Lescop

Fig. 3. COBRA maps calculated from the H–N–CA experiments collected on Hyl1 with different phase cutoff values for the STD (fSTD) and TS (fTS) region and different peak ordering. (a) (fSTD = 15°; fTS = 15°) and original (random) peak ordering. (b) (fSTD = 15°; fTS = 15°) and peak ordering obtained after unambiguous fragment identification. (c) (fSTD = 15°; fTS = 45°) and same peak order as in (b). (d) (fSTD = 15°; fTS = 45°) and peaks ordered according the final assignment.

and is also used for contour level definition. Optionally, it can also be adjusted manually. The same cutoff value is used for binning in the assignment step (vide infra). After peak ordering and calculation of the new COBRA maps (Calculate COBRA maps), the identified fragments are ordered from the longest one to the shortest one as illustrated in Fig. 3b-c. This ordering step is recommended once a reasonably good COBRA map is obtained (see rule 1 below).

21

Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy

423

Table 1 Strategy for setting cutoff phases in COBRA algorithm Rule

Observation

1

A “good” final COBRA map is obtained if the majority (> ~80–90%) of columns and rows contains on average only ~1 element with nonzero intensity

2A

The COBRA product map contains several (a majority of) columns/ rows with only zero element values. This situation corresponds to experiments of low SNR

A subset of COBRA maps shows low SNR

The COBRA product map contains a majority of columns/rows with two or more nonzero values. This situation corresponds to experiments of high SNR but weak frequency discrimination

This observation applies only to a subset of COBRA maps This observation applies to all COBRA maps

2B

3A

3B

Origin

Action Try the assignment (Subheading 7)

All COBRA maps show low SNR

Increase by 15° the phase cutoff parameter for the corresponding pair Increase by 15° the phase cutoff parameters for all pairs Decrease by 15° the phase cutoff parameter for the corresponding pairs Decrease by 15° the phase cutoff parameter for all pairs

For phase cutoff adjustment, an ensemble of observation rules and corresponding actions for phase optimization are gathered together in Table 1. One example of strategy for setting phases in COBRA processing is described for the case of the Hyl1 protein. The phase cutoff parameter is initially set to 15° for both pairs (STD and TS H–N–CA) as illustrated in Fig. 3a. This map contains (visually) a high number of row/columns with only ~1 element with a value larger than the intensity cutoff. After peak ordering (Fig. 3b) several fragments of unambiguously connected cross peaks are clearly apparent. Nevertheless, several columns at the extremity of these fragments contain only zero values. By applying rule 2A, we increase the phase parameter of the TS H–N–CA pair to 45° (Fig. 3c). This map follows now rule 1 and the first round of assignment is carried out (Subheading 7). The COBRA map obtained after assignment based on the available data and ordering of the cross peaks is shown in Fig. 3d. Additional considerations are listed below: (a) Sensitive pairs of experiments (such as STD H–N–CA and H–N–CB) are processed with lower phase cutoff values compared to less sensitive datasets (TS H–N–CA and H–N–CO). (b) If a 1H/15N peak gives rise to no signal at all in COBRA maps, it may correspond to a side chain H–N moiety (and remove it from the list) or to a residue with low sensitivity due to line-broadening for example (such a peak will need to be treated manually).

424

B. Brutscher and E. Lescop

(c) If the right part of the COBRA product map (after ordering) contains many columns with more than 2 possible connectivities, this suggests either a lack of 13C frequency discrimination due to close or overlapping 13C chemical shifts for different residues in the protein or to partially overlapping 1H/15N. In the former case, the HADAMAC information may be sufficient to alleviate the ambiguities in the course of the assignment. The latter case can be identified from the inspection of the 15N HSQC spectrum and the integration box around 1 H/15N frequencies can be reduced (see Path window). (d) The HNCO/iHNCO pair of experiments should be used carefully during the course of assignment using COBRA for glycine residues. The iHNCO experiment contains one Ca selective pulse designed to optimize the Ca -> CO transfer while limiting Ca -> Cb transfer. As a consequence, signals corresponding to glycine residues may be severely attenuated in the iHNCO experiment. This translates into missing COBRA connectivities for glycine residues in the final COBRA maps (after combination of all available COBRA maps). (e) Overall, less than ~5 COBRA calculations should in principle be sufficient to obtain a COBRA map that reflects the information content of the triple resonance experiments. Otherwise, the suitability of the BATCH method for the particular protein under investigation should be questioned.

7. Backbone Assignment Based on the previous steps, a 2D 1H–15N peak list containing the cross-peak-dependent HADAMAC amino acid group and the corresponding COBRA sequential connectivity map are available. Three alternative options with different degrees of automation have been developed to facilitate sequential resonance assignment. They include the embedded BATCH algorithm (3), the interface to MARS (18) and SmartNotebook (19) software packages. The COBRA product map (Subheading 6.2) contains element values between 0 and 1. Before application of any of the three methods, this map is first binned to 0 and 1 values according to the intensity cutoff parameter defined in the main BATCH window. The cutoff parameter automatically proposed after COBRA map calculation often represents an excellent starting value. However, it can be manually adjusted from the (visual) inspection of the COBRA maps. Based on the newly calculated matrix, fragments of unambiguously connected cross peaks are built and fed to the assignment algorithm together with the additional (ambiguous) sequential connectivities and the HADAMAC information.

21

Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy

425

An automated assignment algorithm is integrated in the BATCH software that is based on a Best-First algorithm. This algorithm quickly converges to a solution when fragments of sufficient lengths can be unambiguously located onto the protein sequence. Additional cross peaks are also assigned based on the HADAMAC information and the compatible COBRA connectivities to allow the extension of already assigned fragments. The BATCH assignment mode is accessible by BATCH assign. Additional information can be found in separate windows for the further manual assignment. MARS embeds an assignment protocol widely exploring the assignment space and can also be chosen for assignment (Start MARS Calculation). SmartNotebook (Launch SNB) is another assignment module available for NMRView that significantly helps the manual assignment by proposing allowed protein locations for a given chain. For each assignment mode, cross peak assignments are stored directly in the current peak list. An additional tool (Create Ordered Peaklist From Assigned Peaks) is also provided for reordering the cross peaks according to the amino acid sequence. The new COBRA map calculated using the reordered peak list allows the rapid analysis of still unassigned cross peaks. Fully automated methods are of great help to quickly assign a large part of the protein. However, the completion of the assignment may require human intervention in particular for the identification of missing or overlapping 1H/15N cross peaks. As an advantage of the BATCH strategy, the triple resonance experiments can also be fully Fourier-transformed and analyzed using the conventional methods. The default processing scripts contain the additional 13C dimension Fourier transformation step. The resulting 3D matrices can be directly loaded in NMRView (Load NvFile in the Path Window) for manual inspection and identification of overlapping peaks for example.

8. 1H, 15N, and 13C Chemical Shift Extraction

After completion of the BATCH assignment only 1H and 15N resonances are assigned. This contrasts with the conventional strategy where 13C chemical shifts are at the core of the assignment step. In the BATCH strategy, 13C chemical shifts can also be easily extracted from the H–N–C experiments. The lower part of the main BATCH graphical interface (Fig. 1a) allows parameters to be set for the processing method (Subheading 2.6) including the first t1 value (used for first order phase correction), possible spectral reverse operation and 13C calibration. One pseudo spectrum is calculated for every nucleus Ca, Cb, and CO (Process spectra) that contains the 13C traces extracted from the intraresidual (left half of the spectrum)

426

B. Brutscher and E. Lescop

Fig. 4. Pseudo 2D spectrum processed as described in Subheading 2.6. 13C frequency traces from the intraresidual (left half ) and sequential (right half ) H–N–CA experiments are shown for each 1H–15N cross peak (ordered according to the final assignment). The two halves of the spectrum show similar patterns due to the presence of traces containing signals at the same 13C frequency in the two experiments. For the sequential experiment, cross peak numbers are incremented by n (total number of cross peaks).

and the sequential (right half) experiments for every 15N HSQC cross peak (Fig. 4). This calculation is based on the same phase cutoff parameters as defined for COBRA map calculation. Residuespecific 13C chemical shifts are automatically extracted for the assigned cross peaks from the pseudo-spectra (Get Shifts) and are eventually unfolded to account for possible aliasing in the 13C dimension (Unfold?). This operation is based on the expected chemical shifts for known amino acids. 1H, 15N, and available 13C chemical shifts data are stored in the NMRView assignment table and can be directly analyzed in terms of secondary structure prediction by CSI (21) within NMR view.

9. Conclusions In this chapter, we describe the BATCH protocol for fast and highly automated sequential protein backbone resonance assignment. In the most favorable situations, the entire assignment step, from sample injection to complete analysis may take only a few hours of time. However, the tools forming the basis of the BATCH

21

Fast Protein Backbone NMR Resonance Assignment Using the BATCH Strategy

427

strategy have intrinsic limitations. The COBRA and HADAMAC methods assume a minimal number of 15N HSQC cross peak overlaps for efficient extraction of single-residue 13C traces and amino acid type discrimination. In addition, the minimal experimental times reported herein for the HADAMAC and 3D triple resonance experiments are only sufficient for slowly relaxing systems, e.g., small proteins. Altogether, these restrictions make BATCH optimal for well-folded small-to-medium size (diamagnetic) proteins that yield well-dispersed NMR spectra while retaining high sensitivity for triple resonance experiments. In our hands, proteins up to 10–15 kDa may be assigned using the BATCH strategy. Of equal importance, the quality of the 15N HSQC in terms of spectral dispersion and assessment of line-broadening serves as a critical test for the possible outcome of BATCH. Future developments are expected to extend the applicability of the BATCH method to higher molecular weight proteins, as well as to intrinsically disordered proteins characterized by low chemical shift dispersion.

Acknowledgments Special thanks to Rodolfo Rasia and Jérôme Boisbouvier (IBS, Grenoble) for allowing us to use their NMR data on Hyl1 to illustrate this book chapter. References 1. Jaravine, V. A., Zhuravleva, A. V., Permi, P., Ibraghimov, I., and Orekhov, V. Y. (2008) Hyperdimensional NMR spectroscopy with nonlinear sampling. J. Am. Chem. Soc. 130, 3927–3936. 2. Hiller, S., Fiorito, F., Wuthrich, K., and Wider, G. (2005) Automated projection spectroscopy (APSY). Proc. Natl. Acad. Sci. USA 102, 10876–10881. 3. Lescop, E., and Brutscher, B. (2009) Highly automated protein backbone resonance assignment within a few hours: the “BATCH” strategy and software package. J. Biomol. NMR 44, 43–57. 4. Lescop, E., Rasia, R., and Brutscher, B. (2008) Hadamard amino-acid-type edited NMR experiment for fast protein resonance assignment. J. Am. Chem. Soc. 130, 5014–5015. 5. Lescop, E., and Brutscher, B. (2007) Hyperdimensional protein NMR spectroscopy in peptide-sequence space. J. Am. Chem. Soc. 129, 11916–11917. 6. Lescop, E., Schanda, P., and Brutscher, B. (2007) A set of BEST triple-resonance experiments for

7.

8.

9.

10.

11.

time-optimized protein resonance assignment. J. Magn. Reson. 187, 163–169. Schanda, P., Van Melckebeke, H., and Brutscher, B. (2006) Speeding up three-dimensional protein NMR experiments to a few minutes. J. Am. Chem. Soc. 128, 9042–9043. Lescop, E., Schanda, P., Rasia, R., and Brutscher, B. (2007) Automated spectral compression for fast multidimensional NMR and increased time resolution in real-time NMR spectroscopy. J. Am. Chem. Soc. 129, 2756–2757. Deschamps, M., and Campbell, I. D. (2006) Cooling overall spin temperature: protein NMR experiments optimized for longitudinal relaxation effects. J. Magn. Reson. 178, 206–211. Pervushin, K., Vogeli, B., and Eletsky, A. (2002) Longitudinal (1)H relaxation optimization in TROSY NMR spectroscopy. J. Am. Chem. Soc. 124, 12898–12902. Diercks, T., Daniels, M., and Kaptein, R. (2005) Extended flip-back schemes for sensitivity enhancement in multidimensional HSQC-type

428

12.

13.

14.

15.

16.

17.

B. Brutscher and E. Lescop out-and-back experiments. J. Biomol. NMR 33, 243–259. Kupce, E., and Freeman, R. (1993) Polychromatic Selective Pulses. J. Magn. Reson. 102A, 122–126. Geen, H., and Freeman, R. (1991) Bandselective radiofrequency pulses. J. Magn. Reson. 93, 93–141. Smith, M. A., Hu, H., and Shaka, A. J. (2001) Improved Broadband Inversion Performance for NMR in Liquids. J. Magn. Reson. 151, 269–283. Brutscher, B. (2002) Intraresidue HNCA and COHNCA experiments for protein backbone resonance assignment. J. Magn. Reson. 156, 155–159. Blevins, R. A., and Johnson, B. A. (1994) NMRView: a computer program for the visualization and analysis of NMR data. J. Biomol. NMR 4, 603–614. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J., and Bax, A. (1995)

18.

19.

20.

21.

NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293. Jung, Y. S., and Zweckstetter, M. (2004) Mars – robust automatic backbone assignment of proteins. J. Biomol. NMR 30, 11–23. Slupsky, C. M., Boyko, R. F., Booth, V. K., and Sykes, B. D. (2003) Smartnotebook: a semiautomated approach to protein sequential NMR resonance assignments. J. Biomol. NMR 27, 313–321. Rasia, R. M., Mateos, J., Bologna, N. G., Burdisso, P., Imbert, L., Palatnik, J. F., and Boisbouvier, J. (2010) Structure and RNA interactions of the plant MicroRNA processing-associated protein HYL1. Biochemistry 49, 8237–8239. Wishart, D. S., and Sykes, B. D. (1994) The 13 C chemical-shift index: a simple method for the identification of protein secondary structure using 13 C chemical-shift data. J. Biomol. NMR 4, 171–180.

Chapter 22 Comprehensive Automation for NMR Structure Determination of Proteins Paul Guerry and Torsten Herrmann Abstract This chapter gives an overview of automated protein structure determination by nuclear magnetic resonance (NMR) with the UNIO protocol that enables high to full automation of all NMR data analysis steps involved. Four established algorithms, namely, the MATCH algorithm for sequence-specific resonance assignment, the ASCAN algorithm for side-chain resonance assignment, the CANDID algorithm for NOE assignment, and the ATNOS algorithm for signal identification in NMR spectra, are assembled into three principal UNIO NMR data analysis components (MATCH, ATNOS/ASCAN, and ATNOS/CANDID) that are accessed thanks to a particularly intuitive and flexible, yet powerful graphical user interface (GUI). UNIO is designed to work independently or in association with other NMR software. The principal data analysis components for sequence-specific backbone, side-chain and NOE assignment may be run separately or out of sequence. User-intervention at individual stages is encouraged and facilitated by graphical tools included for the preparation, analysis, validation, and subsequent presentation of the NMR structure. Key words: Protein structure, NMR structure determination, Resonance assignment, NOE assignment, Automated NMR structure determination, MATCH, ASCAN, ATNOS, CANDID, UNIO protocol

1. Introduction Little more than 10 years ago, protein NMR structure determination projects were framed in terms of months if not years of laborious, interactive work that required the expertise of a well-trained NMR spectroscopist. Nowadays, owing to stunning advances in NMR experiments, instrumentation and computational data analysis, a relatively propitious protein candidate may be solved in a few weeks (1, 2). More importantly, even newcomers to the NMR field are increasingly able to pursue a small- to medium-sized protein NMR structure determination, with a minimum of supervised training, by following standard protocols for NMR data acquisition and

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_22, © Springer Science+Business Media, LLC 2012

429

430

P. Guerry and T. Herrmann

Fig. 1. Scheme of the stepwise standard protocol for protein structure determination by NMR.

data interpretation. Despite these significant advances that promote NMR spectroscopy as a universal tool for the broader structural biology community, the motivation remains high and ongoing to describe and establish a general and robust NMR structure determination protocol in terms of man-hours, not man-weeks. The commonly exploited protocol for NMR structure determination (Fig. 1) includes the preparation of the protein sample, the acquisition of multidimensional NMR experiments, NMR data processing, signal identification (peak picking), sequence-specific resonance assignment, NOE assignment as the primary source of conformational restraints and structure calculation followed by structure refinement and structure validation (3). This well established stepwise protocol has successfully been applied to thousands of de novo structure determinations but results in lengthy NMR data analysis requiring massive human manpower and expertise on top of the time-consuming data acquisition process. In recent years, solution NMR has attained a level of development where much interest is focused on replacing laborious and time-consuming manual NMR data analysis with computationaltheoretical approaches (1, 2, 4–8). This interest in streamlining the process of NMR structure determination, to achieve a level of sophistication presently used by X-ray crystallography, has been further enhanced by the demands for high-throughput NMR studies on proteins in Structural Proteomics Initiatives (1). Currently, various computational expert systems for solution NMR structure determination are available aiming either at supporting the interactive

22

Comprehensive Automation for NMR Structure Determination of Proteins

431

spectral analysis by visualization tools and systematic bookkeeping of the collected spectral data (computer-aided approach) (9–16) or at providing semi- or full automation for specific parts of an NMR structure determination (7). Most progress has been achieved for the final part of NOE assignment and structure calculation (17–22). Although expert systems for NMR data analysis are commonly exploited for the collection of conformational NOE-derived distance restraints, most of the proposed approaches operate on listings of peak positions and volumes (peak lists) rather than on the raw NMR spectra, and their performance critically depends on careful preprocessing of the input data with the notable caveat that this laborious task is still mainly performed interactively and hence subjectively (23). In common practice, several rounds of NOE assignment and structure calculation with steadily refined NOE peak lists are required to obtain an accurate and precise threedimensional protein structure (23). The preceding step of sequencespecific resonance assignment has also been subjected to the development of data analysis models (24–48). Here, most of the proposed algorithms target the assignment of the polypeptide backbone atoms, where again, extensively preprocessed input data are typically required to lead to satisfactory results. Despite many promising attempts, manual or semi-automated approaches still prevail, and the critically important chemical shift assignment of the amino acid side-chain atoms represents a major bottleneck for efficient NMR data analysis and structure determination. Initially, expert systems were expected to be most suitable for the first step of NMR signal identification (peak picking) providing efficient, objective and reliable handling of the large sets of NMR signals comprising thousands of resonance frequencies (6, 27, 38, 49–55). In practice however it has turned out that robust peak picking is limited to spectral regions with scarcity of signal overlap and artifacts, with manual reinspection of the results being generally advised. Nonetheless, and the impressive progress of research on the subject notwithstanding, it could be argued that the major stumbling block for automated NMR structure determination has proven to be experimental in that a large number of NMR spectra, i.e., high data redundancy, is generally required for robust performance of the individual automated approaches. Indeed, it is often the case that an interactive data analysis approach, on fewer NMR spectra, is more attractive and more time-efficient once the tedious demands of the proposed automated approaches (i.e., labor-intensive data preprocessing and collection of a comprehensive, highly redundant set of NMR spectra) are taken into consideration. In this chapter we describe the UNIO protocol for highly to fully automated protein NMR structure determination that performs all NMR data analysis tasks, i.e., sequence-specific backbone and side-chain resonance assignment, and NOE assignment, reliably and efficiently. UNIO associates previously published algorithms

432

P. Guerry and T. Herrmann

within a single computational framework and emphasizes ease of use through intuitive and powerful graphical interfaces and utilities, making it attractive to aficionados and casual users alike. Most importantly, UNIO is undemanding in terms of experimental input, both in the number of NMR spectra and the subsequent data preprocessing required for proper performance. The standard UNIO protocol requires the acquisition of only six NMR spectra (three APSY and three NOESY spectra), and the flexibility of UNIO means that the setup for any part of the step-wise protocol for NMR structure determination can be tailored according to the particular problem at hand. Time is therefore gained during data analysis without increasing the experimental load. UNIO has been designed for immediate and general applicability to problems in structural biology. The underlying ethos is fundamentally pragmatic, and the algorithms and protocols described in this chapter follow the path of optimal robustness and efficiency. In this context, expert algorithms solve those problems that are tedious and time-consuming whereas the user remains in control of the overall process and if necessary, performs tasks where human judgment and intuition (currently) know no numerical equal. Last but not least, the simplicity and ease of use of the software ensure that an entire NMR structure determination can be completed within a few man-hours of processing the experimental data.

2. Specificities of the UNIO Program

The simplicity and ease of use of the UNIO suite ensure that a structure calculation is up and running within only a few manhours of installation, making UNIO attractive to expert and casual users alike. The following list provides an overview of computer requirements, compatible input file formats, and molecular dynamics programs that can be used in combination with UNIO. 1. Display resolution of 1,024 × 768 pixel or higher. True color display (16-bit or 32-bit depth). Computer with either Linux kernel 2.4 or above, or Mac OSX operating system 10.5 or higher with Intel processors. A minimum of 100 megabytes of disk space is required. 2. UNIO software application suite for automated protein NMR structure determination. UNIO is free-of-charge for academic use at http://www.unio-nmr.eu. 3. CNS (56), XPLOR-NIH (57), or CYANA(58) software package for NMR structure calculation by simulated annealing. 4. Input file for the amino acid sequence in any of the following formats: BioMagResBank (59), XEASY (11), FASTA, ANSIG (9), NMRVIEW (10), SPARKY (14), CYANA, CNS, XPLOR-NIH.

22

Comprehensive Automation for NMR Structure Determination of Proteins

433

5. For MATCH, input peak lists containing information about the frequency coordinates of the NMR signals in APSY file format. 6. For ATNOS/ASCAN and ATNOS/CANDID: input chemical shift list in any of the following formats: BioMagResBank, XEASY, NMRVIEW, SPARKY, CYANA, CNS, XPLOR-NIH; 3D 13C or 15N-resolved [1H, 1H]-NOESY spectra in either BRUKER or XEASY file format.

3. Standard UNIO Protocol for Protein NMR Structure Determination

The UNIO protocol for substantially automated protein NMR structure determination comprises sequence-specific backbone and side-chain assignment followed by NOE assignment and NMR structure calculation (Fig. 2). 1. Sequence-specific backbone resonance assignment with MATCH. The standard experimental input data for MATCH (47) consists of a set of three APSY datasets: 4D APSY-HACANH, 5D APSY-HACACONH, and 5D APSY-CBCACONH (60, 61). The MATCH algorithm yields polypeptide backbone resonance assignments for the 1HN, 1Ha, 15N, 13Ca, 13Cb, and 13C¢ atoms. 2. Side-chain resonance assignment with ATNOS/ASCAN. The standard experimental input data for ATNOS/ASCAN (48, 54) comprises the previously obtained sequence-specific backbone resonance assignments and a set of three NOESY spectra: 3D [1H,1H]-NOESY-15N-HSQC and two 3D [1H,1H]NOESY-13C-HSQC with the 13C carrier frequency in the aliphatic and the aromatic spectral regions. The ATNOS/ASCAN approach yields meaningful side-chain resonance assignments, i.e., resonance frequencies of atoms involved in many NOEs. 3. Automated NOESY assignment and NMR structure calculation. The standard experimental input data for ATNOS/ CANDID (19, 54) consists of the previously determined backbone and side-chain resonance assignments, and the three aforementioned 3D NOESY spectra. ATNOS/CANDID in combination with a simulated annealing program yields listings of assigned NOESY peaks and the 3D protein structure. In the following sections, the individual UNIO data analysis components are presented in detail. Automated NMR signal identification is described in Subheading 4, automated backbone assignment in Subheading 5, automated side-chain assignment in Subheading 6, and automated NOESY assignment in Subheading 7. The entire UNIO protocol presented here has been successfully applied to more

434

P. Guerry and T. Herrmann

Fig. 2. Schematic outline of the UNIO protocol for highly automatic NMR protein structure determination. The three principal UNIO modules are in bold font inside solid boxes along with the standard input data and output, in dashed and shaded rectangles, respectively. The input data common to all three modules is shown in a thick dashed box at the top. Cyclic symbols denote reevaluation of the experimental input data at the start of each iteration guided by the output of the previous cycle (new resonance assignments for ATNOS/ASCAN, intermediate protein structures for ATNOS/CANDID). In case, the UNIO validation criteria are not met, the required interactive refinement is facilitated by UNIO reports.

than a dozen de novo NMR structure determination projects. Individual NMR data analysis components of UNIO such as the ATNOS/CANDID (19, 54) approach for combined automated signal identification and NOE assignment, or the CANDID module

22

Comprehensive Automation for NMR Structure Determination of Proteins

435

alone, have already evolved into standard processing tools routinely applied by the biology-oriented NMR community and have contributed to several hundreds of protein NMR structure determinations with hitherto unknown protein folds. Note that although the three principal NMR data analysis components of UNIO are designed to work one after the other, one of UNIO’s strengths is its flexibility; the different modules may be launched separately, out of sequence and with different experimental input data from those listed above: MATCH also supports peak lists from conventional triple-resonance experiments as input; the ATNOS/ASCAN approach can also be used with TOCSY input datasets; the ATNOS/CANDID approach can be supplemented by additional conformational restraints such as residual dipolar couplings, pseudo contact shifts, torsion angle restraints, hydrogen and disulfide bond restraints. These optional conformational restraints are not used for any of the NMR data analysis tasks performed by UNIO, but are directly passed on to the structure calculation algorithm.

4. Automated NMR Signal Identification

4.1. Overview of the ATNOS Algorithm

In the UNIO protocol, automated NOESY peak picking and NOE signal identification in 2D homonuclear and heteronuclearresolved 3D [1H, 1H]-NOESY spectra is performed with the ATNOS algorithm (54) in association with either ASCAN (48) automated side-chain assignment (see Subheading 6) or CANDID (19) automated NOE assignment and NMR structure calculation (see Subheading 7). The main elements of ATNOS for NOESY spectral analysis are local baseline correction and evaluation of local noise level amplitudes, automated determination of spectrum-specific threshold parameters, the use of symmetry relations, and the inclusion of chemical shift information and the intermediate protein structures to distinguish between NOE cross peaks and artifacts. 1. Input data for ATNOS. The input data consists of the amino acid sequence of the protein, resonance frequencies of the assigned atoms, and 2D or 3D NOESY spectra. 2. Determination of local baseline and local noise level. These techniques are based on those previously introduced by the FLATT (62) and AUTOPSY (53) algorithms. 3. Generation of a comprehensive set of NMR signals. Highly permissive criteria are applied that only require an initial minimal signal-to-noise ratio and a local minimum.

436

P. Guerry and T. Herrmann

4. Identification of “covalent NMR signals”. Assignment of NMR signals to atom pairs with covalent structure-imposed upper distance limits are based on compatibility with the input chemical shifts: the fixed bond lengths, bond angles, and chiralities of the covalent polypeptide structure impose NOE-observable upper limits on certain intraresidual and sequential 1H–1H distances. These conformation-independent upper distance limits can be calculated analytically for all atom pairs that are separated by one or two dihedral angles. A covalent NMR signal is defined such that in its initial list of chemical-shift based assignments there is at least one assignment possibility that corresponds to a hydrogen pair with maximal upper distance limit smaller than 5 Å. The set of identified covalent NMR signals in a given NOESY spectrum can then be used to derive spectrumspecific threshold parameters for minimal signal-to-noise ratio and adaptation of the input chemical shift. 5. Determination of spectrum-specific threshold parameters. Threshold values for minimally required signal-to-noise ratio and peak volume are determined using the previously identified covalent NMR signals as a reference. 6. Adaptation of input chemical shifts for each individual NOESY spectrum. 7. Peak validation. The first validation filter is based on peak classification, compatibility with adapted chemical shifts, networkanchoring, and symmetry considerations (ATNOS/ASCAN and ATNOS/CANDID cycles 1, 2, …). 8. Peak validation. The second validation filter is based on compatibility with the intermediate protein structure (ATNOS/ CANDID cycles 2, 3, …). The ATNOS approach for automated NOESY signal identification differs from most conventional automated peak picking programs by incorporating chemical shift information (ATNOS in combination with ASCAN or CANDID) and intermediate protein structures (ATNOS in combination with CANDID) into the process of NMR signal identification. Most of the routinely used algorithms for automated NOE assignments operate on intermediate listings of NOE cross peak positions and volumes, such as NOAH (17), ARIA (20, 63), AUTOSTRUCTURE (22), KNOWNOE (18), CANDID (19), and PASD (21, 64). In practice, the use of intermediate listings of NOE cross peak positions entails that the automated NOE routines are applied in several rounds with successively refined NOE peak lists as input data. This is clearly a conceptual limitation of the present practice of automated NOE assignment and results in time-consuming, laborious editing of the input data for most automated NOE assignment programs to obtain an accurate and precise 3D protein structure. The listings of

22

Comprehensive Automation for NMR Structure Determination of Proteins

437

NOE cross peak positions can also be obtained by automated peak picking methods. However, even sophisticated pattern recognition methods easily fail for all but ideal artifact-free, well-separated NMR signals. Under realistic, experimental conditions, difficulties in automated NMR signal identification arise mainly from signal overlap and spectral distortions due to artifacts. Sophisticated algorithms have been introduced at the outset of a spectral analysis, but in practice their use in spectral regions of strong peak overlap and weak noisy peaks is limited, and manual reinspection of the resulting listings of NOE cross peaks is generally advised. Therefore, in present practice, NOESY peak picking is still dominantly performed with interactive graphic computer programs. Automated and interactive NOE cross peak identification must be able to clearly distinguish between both real and artifact signals, with the signal-to-noise ratio as the primary filter. Because of the inverse 6th power-relationship between NOE cross peak intensity and interatomic distance between the pair of hydrogen atoms giving rise to a NOE cross peak, a significant fraction of the most informative long-range NOE signals in a NOESY spectrum may have signal-to-noise ratios only slightly above the average noise level, which emphasizes the importance of working with powerful and sophisticated signal filtering procedures. A weakness of many automated peak picking routines is introduced by the underlying recognition technique focusing only on limited regions in close proximity to a local extremum, without taking into account mutually inclusive peak patterns in a given NOESY spectrum, or across several NOESY spectra. More recently, algorithms for automated peak-picking perform better when the spectral data is supplemented with additional information, such as chemical shift lists of atoms that are correlated with as yet unidentified signals in the spectra to be analyzed, or information on expected peak patterns derived from the magnetization pathways in the NMR experiments used. These “constrained” peak picking algorithms mimic the modus operandi of an experienced spectroscopist, who analyzes new signals in the context of previously assigned resonances. When used in the context of NOESY assignment (Subheading 7), the ATNOS approach performs multiple cycles of NOE peak identification in concert with automated NOE assignment with the CANDID algorithm followed by protein structure calculation by simulated annealing using either CNS (56) or XPLOR-NIH (65) or CYANA (58). At the outset of a de novo structure calculation (ATNOS/CANDID cycle 1), ATNOS NOE peak validation is primarily guided by the input chemical shifts. In the second and subsequent cycles of automated ATNOS/CANDID NOESY analysis, intermediate protein structures are used as an additional guide for the interpretation of the NOESY spectra. Since the precision and accuracy of the intermediate protein structures tend to improve from cycle to cycle, the structure-based criteria for ATNOS NOE

438

P. Guerry and T. Herrmann

identification are loosened to facilitate identification of weaker signals. By incorporating the analysis of raw NMR data into the process of automated protein structure determination, ATNOS enables direct feedback between the protein structure, the NOE assignments and the experimental NOESY spectra. Thereby the list of verified NOE peaks is updated between subsequent cycles of combined NOE assignment and structure calculation based on the intermediate protein structures. Notably, within this scheme of using chemical shifts and intermediate 3D structures, ATNOS achieves more extensive and reliable NOE cross peak identification than routines that rely exclusively on the information content of the NOESY spectrum without further guidance by already available chemical and structural information. The combination of ATNOS with an automated NOE assignment routine avoids the iterative refinement of static peak lists common to most other popular NOE assignment programs.

5. Automated Backbone Assignment

5.1. Overview of the MATCH Algorithm

In the UNIO protocol, automated sequence-specific polypeptide backbone NMR assignment is performed with the MATCH algorithm (47). MATCH employs local optimization for tracing partial sequence-specific assignments within a global, population-based search environment, where the simultaneous application of local and global optimization heuristics guarantees high efficiency and robustness. MATCH thus makes combined use of the two predominant concepts in use for automated assignment of proteins (see Subheading 5.2). The MATCH algorithm is founded on two main building blocks: initialization and optimization. Novel concepts in MATCH are dynamic transition and inherent mutation that enable automatic adaptation to the variable quality of experimental input data. The concept of dynamic transition is incorporated in all major building blocks of the MATCH algorithm, where it enables switching between local and global optimization heuristics at any time during the assignment process. Inherent mutation restricts the intrinsically required randomness of the evolutionary algorithm to those regions of the conformation space that are compatible with the experimental input data. 1. Input data for MATCH. The input consists of the amino acid sequence of the protein, a statistical analysis of chemical shift values of proteins contained in the BioMagResBank, and the experimental NMR data in form of the frequency coordinates of the NMR correlation signals.

22

Comprehensive Automation for NMR Structure Determination of Proteins

439

2. Generation of generic spin-systems. The experimental input peak lists are consolidated and transformed into a single set of generic spin systems containing all available intra- and interresidual chemical shifts for a given spin system. 3. Buffer of candidate fragments. A graph exploration routine identifies all possible sequential connectivities between generic spin-systems up to a user-specified maximal fragment length. 4. Calibration of control parameters. All MATCH control parameters used in the optimization routine are automatically adapted to the degree of ambiguity contained in the experimental input data. 5. Genesis. The generation of an initial population of arbitrarily generated sequence-specific resonance assignments (set of individuals) represents the start of the optimization routine. 6. Assignment optimization. Local optimization is applied by repositioning candidate fragments. A candidate fragment is relocated if and only if its sequence-specific assignment score increases for the newly proposed sequence position. 7. Assignment management. Temporary and permanent sequencespecific assignments are performed for a given fragment based on the sequence-specific scoring function and the presence in the population of the same fragment at the same sequence position. 8. Cross-over. This key module of the evolutionary MATCH algorithm identifies the most promising individuals that are used subsequently for the generation of a new population. 9. Intervention. Control parameters are adapted based on the progress of the optimization process. Return to step 6. 10. Elite buffer. The optimization for a given individual is completed either when all generic spin-systems are permanently assigned, or when the total sequence-specific score of all individuals is equal. The final result is then stored and the optimization restarts with step 5 until a predetermined number of elite individuals have been generated. MATCH initialization (steps 1–4) is needed to load all the necessary input data, to consolidate the experimental NMR data, to generate an initial set of measured graphs, and to calibrate intrinsic MATCH control parameters. The result of the initialization process represents the input for the first cycle of optimization (steps 5–10). Each MATCH optimization cycle starts with the creation of an initial population of individuals. This is followed by multiple evolutionary cycles, each consisting of local optimization and a global “cross-over,” where new individuals are created and low-scoring individuals are eliminated. Within each evolutionary cycle, the configuration space is reduced whenever possible, and, if

440

P. Guerry and T. Herrmann

necessary, the threshold for the assignment of a generic spin-system to a specific-sequence position is decreased. The result of each round of optimization, which typically includes a large number of evolutionary cycles, is stored as a new population of “elite individuals.” Subsequent optimization rounds always start from a newly created population. 5.2. Memetic Algorithm

In general, algorithms for solving the resonance assignment problem employ either local or global optimization. Local optimization algorithms refine a preliminary solution by screening the adjacent configuration space in search of information on the best candidate solution. They can work in a highly deterministic fashion, following a concrete optimization strategy. The benefit of local optimization is high efficiency resulting from the assumption that the underlying data do not contain information that is incompatible with the rationale used by the algorithm. This efficiency is predictably gained at the expense of robustness. Global optimization algorithms, on the contrary, solve combinatorial problems by optimizing all problem parameters independently and at once. They are usually implemented in a population-based fashion such that multiple candidate solutions located in different regions of the configuration space are optimized simultaneously. A certain degree of randomness may be involved, analogous to mutation in biological evolution, e.g., genetic algorithms, the deteriorating influence of misleading experimental input data is thus muted, and the risk of getting trapped in local minima is greatly reduced. Overall, a population-based global optimization approach has high robustness but low efficiency due to the fact that numerous candidate solutions have to be managed concurrently. A memetic algorithm is the logical attempt to merge both approaches, since it contains a local optimization routine embedded in an evolutionary, global optimization algorithm. The evolutionary algorithm is meant to explore the overall problem space, while the local search heuristic refines discrete areas of this space. With MATCH, the advantages of both approaches are exploited to the fullest extent as local optimization efficiently traces partial solutions inside a population-based (genetic) environment that preserves robustness.

5.3. APSY-NMR Input Data

The NMR method APSY (Automated Projection Spectroscopy) (61) enables the automatic generation of high-dimensional heteronuclear-resolved correlation peak lists from the analysis of a suitably selected group of experimental 2D projections of the higher-dimensional experiment. The use of high dimensions enables a significant reduction of the number of spectra needed for the resonance assignment. A further important merit of APSY spectroscopy is the determination of highly precise correlation peak chemical shifts (66), which is a key asset for fully automated

22

Comprehensive Automation for NMR Structure Determination of Proteins

441

sequence-specific resonance assignment. MATCH has been optimized for high efficiency and reliability of automatic backbone NMR assignment of proteins when using input from APSY-NMR experiments. Note that MATCH can also be used with conventional triple-resonance data.

6. Automated Side-Chain Assignment

6.1. Overview of the ATNOS/ASCAN Approach

In the UNIO protocol, automated sequence-specific NMR assignment of amino acid side-chain atoms is performed according to the ATNOS/ASCAN approach (48, 54). ATNOS/ASCAN operates on the 3D heteronuclear-resolved [1H,1H]-NOESY datasets that are subsequently used to collect the input of NOE-distance constraints for the structure calculation. ATNOS/ASCAN makes use of the chemical shift lists for the previously assigned backbone atoms, and the knowledge of the covalent polypeptide structure. To make inevitable imperfections of experimental input NMR data tractable, the chemical shifts of the previously assigned backbone and Cb atoms are used to guide both the peak-picking of the NOESY spectra and the search for new side-chain resonance assignments. The ATNOS/ASCAN procedure assigns new resonances based on a comparison of the NMR signals expected from the chemical structure (“covalent peaks”) with the experimentally observed NOESY peak patterns. The ATNOS/ASCAN approach differs from most previous procedures for automated resonance assignment of backbone and/or side-chain atoms in that it operates on raw NMR data rather than on interactively generated peak lists, which often require extensive preprocessing to lead to satisfactory results. The underlying techniques of the ATNOS/ASCAN approach are a procedure for generating expected peak positions, and a corresponding set of acceptance criteria for assignments based on the NMR experiments used. Expected patterns of NOESY cross peaks involving unassigned resonances are generated using the list of previously assigned resonances, and tentative chemical shift values taken from the Biological Magnetic Resonance Data Bank (BMRB) statistics for globular proteins. 1. Input data for ATNOS/ASCAN. The input consists of the amino acid sequence of the protein, a statistical analysis of chemical shift values of proteins contained in the BioMagResBank, and the experimental NMR data comprising chemical shift lists of the 1HN, 15N, 13Ca, 13Cb, and possibly 1Ha atoms, and one or several 3D 13C- or 15N-resolved [1H,1H]-NOESY spectra. Optional input data can be provided. 2. Generation of expected 3D NOESY peak patterns. The unassigned atoms are correlated with the previously assigned

442

P. Guerry and T. Herrmann

atoms via the chemical structure to generate the set of expected peaks. 3. Automated signal identification in the NOESY spectra. Chemical-shift guided peak picking with ATNOS is performed in which each cycle yields an updated set of identified signals. 4. Determination of experimental peak patterns. The set of observed peaks is generated based on the knowledge of the magnetization pathway employed by the NMR experiment. 5. Mapping of expected peak onto observed peak pattern. A set of potential resonance frequencies for each unassigned atom is generated by best fit of the expected onto to the observed peak pattern. 6. Acceptance criteria for resonance assignment. First, the agreement between expected and observed peaks is assessed using an iteration-dependent threshold value. Second, the agreement between the remaining potential resonance frequencies and predictions based on chemical shift statistics is assessed. Third, a new resonance assignment is stored if only one potential resonance frequency is retained. Return to step 1. Starting from the backbone chemical shift lists, ATNOS/ ASCAN generates expected NOE-signal patterns that are subsequently compared with those observed in the experimental NOESY spectra. In each cycle of the iterative ATNOS/ASCAN protocol, the information on expected signal patterns is updated based on the new assignments obtained in the preceding cycle. Experimental peak patterns to be compared with these predicted peak patterns are also updated after each iteration cycle, making use of the new assignments obtained in the previous cycles for chemical shiftguided ATNOS signal identification. The next step consists in an evaluation of the closeness of fit between expected peak patterns and the experimental data, which in turn generates the input for a set of acceptance criteria for new resonance assignments. The ATNOS/ASCAN protocol for automated side-chain resonance assignment is composed of two assignment phases. During the first assignment phase, only 1H atoms for which the resonance frequency of the covalently bound heavy atom was present in the input are assigned. Thus, the first phase completes the assignment of the 1Ha and 1Hb atoms before the second phase aims at assigning the more peripheral amino acid side-chain atoms. The entire assignment procedure makes use of fixed as well as iteration-dependent parameters. These control parameters are chosen in such a way that initially atoms can only be assigned if the agreement between experimental peak positions and chemical shift values of previously assigned atoms is extremely good. Moreover, the initial peak pattern is composed exclusively of NMR signals with a high signalto-noise ratio. In later iterations the control parameters are loosened to allow the assignment of atoms for which the agreement is not so

22

Comprehensive Automation for NMR Structure Determination of Proteins

443

good. NMR signals with reduced intensity due to line broadening, solvent suppression, or inefficient magnetization transfer can at this stage be used for obtaining new resonance assignments. 6.2. Resonance Assignments Obtained by the ATNOS/ASCAN Approach

7. Automated NOESY Assignment and NMR Structure Calculation

7.1. Overview of the ATNOS/CANDID Approach

As is the case for interactive side-chain assignment, the ATNOS/ ASCAN procedure performs better on interior, buried residues than on extensively solvent-exposed residues. In general, for both approaches, the completeness of the side chain assignments correlates inversely with the degree of solvent-accessibility. This is readily rationalized if one considers that a much larger number of NOEs is generally observed for interior atoms than for atoms at or near the protein surface. ATNOS/ASCAN thus primarily assigns sidechain atoms that are involved in numerous inter-residue NOEs. Note that it is a special advantage of the [1H,1H]-NOESYbased ATNOS/ASCAN approach that the same datasets are used for the amino acid side chain assignments and for the collection of NOE upper distance constraints. Since adjustments of polypeptide backbone chemical shifts have already been made when preparing the input for ATNOS/ASCAN, this eliminates the need for further chemical shift adjustments between datasets recorded with different experimental conditions, which is an intrinsically laborious procedure that may introduce unnecessary ambiguity into the following step of NOESY assignment and structure calculation.

In the UNIO protocol, automated NOESY spectral analysis follows the ATNOS/CANDID approach (19, 54) that proceeds, as all commonly used NOE assignment algorithms, in iterative cycles, each consisting of exhaustive NOE signal identification and, in part, ambiguous NOE assignments followed by a structure calculation. But in contrast to many other NOE assignment approaches that operate on listings of peak positions and chemical shifts invariant in all NOE assignment cycles, the combined use of ATNOS NOE signal identification and CANDID NOE assignment waives the common requirement for multiple rounds of manual peak list preparation and refinement, and leads to a dramatic increase in the efficiency and reproducibility of the NOESY spectral analysis. Each cycle of the iteratively performed NOESY spectral analysis consists of automated NOESY peak picking with ATNOS, use of the resulting lists of peak positions and peak intensities as input for CANDID automated NOE assignment, and use of a set of NOE distance restraints from CANDID as input for the structure calculation. Between subsequent ATNOS/CANDID cycles, information is transferred exclusively through the intermediate 3D structures, in that the protein molecular structure obtained in a given cycle is

444

P. Guerry and T. Herrmann

used to guide NOE signal identification and NOE assignment in the following cycle. The three main techniques that form the basis of the automated CANDID NOE assignment algorithm are ambiguous distance restraints (Subheading 7.2), network-anchored assignment (Subheading 7.3) and constraint combination (Subheading 7.4). The latter two concepts make the ATNOS/ CANDID approach robust with respect to the inevitable imperfections of NMR spectra. Network-anchored assignment and constraint combination ensure that the correct protein fold is already obtained after the first ATNOS/CANDID cycle. 1. Input data for ATNOS/CANDID. The input consists of the amino acid sequence of the protein, the resonance frequencies of the previously assigned atoms, and one or several 2D or 3D NOESY spectra. Optional conformational restraints can be provided. 2. Automated ATNOS signal identification. The ATNOS algorithm yields a listing of NOE cross peak positions and volumes. 3. Generation of initial assignment possibilities. For each NOESY cross peak, one or multiple assignments are determined based on chemical shift fitting within a user-defined tolerance range. 4. Ranking and elimination of initial assignment possibilities. Only those initial assignment possibilities that contribute more than an iteration-dependent threshold to the overall peak volume are retained. Thereby, the contribution of each initial assignment possibility to a given peak volume is calculated as function of the closeness of the chemical shift fit, the compatibility with the covalent polypeptide structure, the networkanchored score (see Subheading 7.3) and the compatibility with the intermediate protein 3D structures (in ATNOS/ CANDID cycles 2, 3, …). 5. Calibration of NOE upper distance restraints. Upper unambiguous or ambiguous distance bounds (see Subheading 7.2) are derived from the NOESY peak intensities. 6. Elimination of spurious NOESY cross peaks. Only those NOE cross peaks that have at least one assignment possibility with a network-anchored score above an iteration-dependent threshold and are compatible with the intermediate 3D protein structure of the preceding cycle are retained (ATNOS/CANDID cycles 2,3, …). 7. Constraint combination. In the first ATNOS/CANDID cycle, unrelated long-range distance restraints are randomly combined into new virtual distance restraints (see Subheading 7.4). 8. Structure calculation. A 3D protein structure is calculated using torsion angle dynamics. The UNIO-ATNOS/CANDID approach interfaces with either CNS or XPLOR-NIH or CYANA. Return to step 1.

22

Comprehensive Automation for NMR Structure Determination of Proteins

445

After ATNOS NOE signal identification, each CANDID cycle starts with the generation for each NOESY cross peak of an initial chemical shift-based assignment list, i.e., hydrogen atom pairs, within the user-defined tolerance range, that contribute to the peak are identified from the fit of chemical shifts. Subsequently, for each cross peak these initial chemical shift-based assignments are weighted with respect to several criteria, and initial assignments with low overall scores are discarded. For each cross peak, the retained assignments are interpreted in the form of an upper distance limit derived from the NOE cross peak volume. Thereby, a conventional distance restraint is obtained for cross peaks with a single retained assignment, and otherwise an ambiguous distance restraint (see Subheading 7.2) is generated that embodies several assignments. In addition, all NOE cross peaks with a poor score are temporarily discarded. In order to reduce deleterious effects on the resulting structure from erroneous distance restraints that may pass the preceding filter step, long-range distance restraints are combined into new virtual distance restraints in ATNOS/CANDID cycle 1 (see Subheading 7.4). The standard ATNOS/CANDID protocol consists of seven cycles. The second and subsequent cycles differ from the first cycle by the use of additional selection criteria for NOE assignment that are based on assessments relative to the intermediate protein 3D structure from the preceding cycle. Since the precision of the protein 3D structure model normally improves with each subsequent cycle, the criteria for accepting cross peaks and NOE assignments are successively tightened in more advanced stages of the calculation. For proper performance and structure validation the following two input criteria must be fulfilled prior to starting the ATNOS/ CANDID procedure: (1) The input chemical shift list must contain more than 90% of the nonlabile and backbone amide 1H chemical shifts. If 3D heteronuclear-resolved NOESY are used, more than 90% of the 15N and/or 13C chemical shifts must be available. (2) ATNOS must validate NOE signals for at least 85% of all pairwise combinations of protons for which sequence-specific NMR assignments are available, and which have covalent structure-imposed upper distance limits shorter than 5 Å. This condition requires high quality of the NOESY spectra and accurate calibration of the input chemical shifts to the NOESY spectra. A low percentage of validated NOE cross peaks typically results when the signal-to-noise ratio is too poor for automated spectral analysis, or the input chemical shifts are not well-calibrated to the NOESY spectra. In this situation, the input data need to be critically reevaluated before attempting a new automated NOESY interpretation. In particular, the adaptation of the chemical shifts to the NOESY spectra needs to be improved. The following three criteria have to be met for validation of the resulting 3D structure. (1) The average final target function value

446

P. Guerry and T. Herrmann

from the first ATNOS/CANDID cycle should be below 250 Å2, and the corresponding value for the last ATNOS/CANDID cycle should be below 10 Å2, with more than 80% of all picked NOESY cross peaks assigned and less than 20% of the peaks with exclusively long-range assignments eliminated by the filtering step applied in CANDID. (2) The average backbone RMSD to the mean coordinates for the structured parts of the polypeptide chain should be below 3 Å for the bundle of conformers used to represent the protein structure from the first ATNOS/CANDID cycle. (3) The RMSD drift between the mean atom coordinates after the first and the last ATNOS/CANDID cycles calculated for the backbone heavy atoms of the structured part of the polypeptide chain should be smaller than 3 Å. The three output criteria emphasize the crucial importance of getting the correct protein fold already after ATNOS/CANDID cycle 1. For reliable automated NOESY analysis, the initial 3D structure obtained should be reasonably compatible with the input data and show a defined fold of the protein. Structural changes between the first and subsequent ATNOS/ CANDID cycles should only occur within the conformation space determined by the initial bundle of conformers obtained after ATNOS/CANDID cycle 1. 7.2. Ambiguous Distance Restraints

The high NOE assignment ambiguity at the outset of a protein structure determination can be resolved by temporarily ignoring cross peaks with too many (typically, more than two) assignment possibilities and instead generating distance restraints for all assignment possibilities of the remaining cross peaks. However, such a procedure requires highly accurate chemical shift values and NOE cross peak positions to be present in the input data and is hardly achievable under realistic, experimental conditions. A more elegant way for handling the initial chemical shift-based assignment ambiguity is given by the concept of ambiguous distance restraints (63, 67). When using ambiguous distance restraints, each individual NOE cross peak is treated as the superposition of n degenerate signals arising from each of its multiple initial chemical shift-based assignments, using relative weights proportional to the inverse sixth power of the corresponding interatomic distance. A NOE cross peak uniquely assigned to a pair of hydrogen atoms, α and β , gives rise to an upper distance limit b for the corresponding distance dαβ ≤ b . A NOESY cross peak with two or more assignment possibilities (n ≥ 2 ) is then interpreted as an ambiguous distance restraint with an effective, d eff , or r −6 -summed distance d

eff

⎛ n ⎞ = ⎜ ∑ di−6 ⎟ ⎝ i =1 ⎠

−

1 6

The sum runs over all distances di = d(a , b )i corresponding to the given chemical shift-based assignment possibility between the

22

Comprehensive Automation for NMR Structure Determination of Proteins

447

two hydrogen atoms, a and b . In this way, information from NOE cross peaks with multiple initial assignment possibilities can be used for the structure calculation, and although inclusion of erroneous assignments for a given cross peak results in a loss of information, it will not lead to inconsistencies as long as one or several correct assignments are among the initial assignments. This is due to the fact that the effective distance d eff is always shorter than any of the individual distances di = d(a , b )i. 7.3. NetworkAnchored NOE Assignment

The concept of ambiguous distance restraint is quite efficient for improving and completing the NOESY assignment once a correct preliminary polypeptide 3D fold is available, e.g., based on a limited set of interactively assigned NOESY cross peaks. However, obtaining a correct initial protein fold at the outset of a de novo structure determination often proves to be difficult, because structure-based filters used for the detection and elimination of erroneous cross peaks in the input data and for the discrimination between multiple initial chemical shift-based cross peak assignments are not yet operational. To achieve reliable and robust automated NOE assignment for de novo protein NMR structure determination, the NOE assignment process cannot solely rely on chemical shift agreement between resonance frequencies of assigned atoms and frequency coordinates of the NMR signals, and the subsequent use of ambiguous distance restraints. Indeed, techniques to remove artifacts prior to any knowledge of a structure model must also be included. One powerful concept for robust automated NOE assignment is network-anchored assignment (19). Network-anchoring imitates the modus operandi of an experienced spectroscopist who typically decides on the assignment of an individual NOE cross peak on the basis of the set of already assigned NOE cross peaks. Networkanchored assignment exploits the observation that the correctly assigned restraints form a self-consistent subset in any network of distance restraints that is sufficiently dense for the determination of a protein 3D structure. Network-anchoring thereby evaluates the self-consistency of the NOE assignments independently of any knowledge about the 3D protein structure, and in this way compensates for the absence of 3D structural information at the outset of a de novo structure determination. The requirement that each NOE assignment must be embedded in the network of all other assignments makes network-anchoring a sensitive approach for detecting erroneous restraints that might artificially constrain unstructured parts of the protein. Such restraints might not lead to systematic constraint violation during the structure calculation, and therefore might also escape 3D structure-based filtering methods. The concept of network-anchored assignment has proved efficient and reliable in searching for the correct fold especially in the initial phase of de novo NMR structure determinations.

448

P. Guerry and T. Herrmann

7.4. Constraint Combination

In the practice of NMR structure determination with biological macromolecules, the presence of spurious distance restraints is hardly avoidable in the input for a structure calculation at the outset of the NOESY analysis, i.e., before a 3D structure is available to filter artifacts. A key technique to weaken structural distortions caused by erroneous distance restraints is constraint combination (19). Constraint combination generates virtual distance restraints with combined assignments from different, in general unrelated (medium- and long-range) NOE cross peaks. Constrain combination is thus an extension of the concept introduced by ambiguous distance restraints. The basic property of an ambiguous distance restraint is that the restraint will be satisfied by the correct protein structure provided that at least one of the assignments is correct. Combined restraints therefore have a correspondingly lower probability of being erroneous than individual ones. Constraint combination, thus, significantly reduces the impact of artifacts on the resulting 3D protein structure, at the expense, however, of a temporary loss of information.

References 1. Billeter, M., Wagner, G., and Wüthrich, K. (2008) Solution NMR structure determination of proteins revisited. J. Biomol. NMR 42, 155–158. 2. Williamson, M. P., and Craven, C. J. (2009) Automated protein structure calculation from NMR data. J. Biomol. NMR 43, 131–143. 3. Wüthrich, K. (1986) NMR of Proteins and Nucleic Acids. Wiley, New York. 4. Altieri, A. S., and Byrd, R. A. (2004) Automation of NMR structure determination of proteins. Curr. Opin. Struct. Biol. 14, 547–553. 5. Baran, M. C., Huang, Y. J., Moseley, H. N. B., and Montelione, G. T. (2004) Automated analysis of protein NMR assignments and structures. Chem. Rev. 104, 3541–3555. 6. Huang, Y. P. J., Moseley, H. N. B., Baran, M. C., Arrowsmith, C., Powers, R., Tejero, R., et al. (2005) An integrated platform for automated analysis of protein NMR structures. Methods Enzymol. 394, 111–141. 7. Gronwald, W., and Kalbitzer, H. R. (2004) Automated structure determination of proteins by NMR spectroscopy. Prog. Nucl. Magn. Reson. Spectrosc. 44, 33–96. 8. Güntert, P. (2009) Automated structure determination from NMR spectra. Eur. Biophys. J. 38, 129–143. 9. Kraulis, P. J. (1989) Ansig - a Program for the Assignment of Protein H-1 2d-Nmr Spectra by Interactive Computer-Graphics. J. Magn. Reson. 84, 627–633.

10. Johnson, B. A., and Blevins, R. A. (1994) Nmr View - a Computer-Program for the Visualization and Analysis of Nmr Data. J. Biomol. NMR 4, 603–614. 11. Bartels, C., Xia, T. H., Billeter, M., Güntert, P., and Wüthrich, K. (1995) The Program Xeasy for Computer-Supported Nmr Spectral-Analysis of Biological Macromolecules. J. Biomol. NMR 6, 1–10. 12. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J., and Bax, A. (1995) Nmrpipe - a Multidimensional Spectral Processing System Based on Unix Pipes. J. Biomol. NMR 6, 277–293. 13. Neidig, K. P., Geyer, M., Gorler, A., Antz, C., Saffrich, R., Beneicke, W., et al. (1995) Aurelia, a Program for Computer-Aided Analysis of Multidimensional Nmr-Spectra. J. Biomol. NMR 6, 255–270. 14. Goddard, T. D., and Kneller, D. G. (2001) SPARKY 3. University of Californai, San Francisco. 15. Keller, R. L. J. (2004) Optimizing the process of nuclear magnetic resonance spectrum analysis and computer aided resonance assignment. Ph.D. thesis. Diss. ETH Nr. 15947. ETH Zurich, Zurich, Switzerland. 16. Kobayashi, N., Iwahara, J., Koshiba, S., Tomizawa, T., Tochio, N., Güntert, P., et al. (2007) KUJIRA, a package of integrated modules for systematic and interactive analysis of NMR data directed to high-throughput

22

Comprehensive Automation for NMR Structure Determination of Proteins

NMR structure studies. J. Biomol. NMR 39, 31–52. 17. Mumenthaler, C., Güntert, P., Braun, W., and Wüthrich, K. (1997) Automated combined assignment of NOESY spectra and threedimensional protein structure determination. J. Biomol. NMR 10, 351–362. 18. Gronwald, W., Moussa, S., Elsner, R., Jung, A., Ganslmeier, B., Trenner, J., et al. (2002) Automated assignment of NOESY NMR spectra using a knowledge based method (KNOWNOE). J. Biomol. NMR 23, 271–287. 19. Herrmann, T., Güntert, P., and Wüthrich, K. (2002) Protein NMR structure determination with automated NOE assignment using the new software CANDID and the torsion angle dynamics algorithm DYANA. J. Mol. Biol. 319, 209–227. 20. Linge, J. P., Habeck, M., Rieping, W., and Nilges, M. (2003) ARIA: automated NOE assignment and NMR structure calculation. Bioinformatics 19, 315–316. 21. Kuszewski, J., Schwieters, C. D., Garrett, D. S., Byrd, R. A., Tjandra, N., and Clore, G. M. (2004) Completely automated, highly errortolerant macromolecular structure determination from multidimensional nuclear overhauser enhancement spectra and chemical shift assignments. J. Am. Chem. Soc. 126, 6258–6273. 22. Huang, Y. J., Tejero, R., Powers, R., and Montelione, G. T. (2006) A topology-constrained distance network algorithm for protein structure determination from NOESY data. Methods Enzymol 62, 587–603. 23. Güntert, P. (2003) Automated NMR protein structure calculation. Prog. Nucl. Magn. Reson. Spectrosc. 43, 105–125. 24. Bernstein, R., Cieslar, C., Ross, A., Oschkinat, H., Freund, J., and Holak, T. A. (1993) ComputerAssisted Assignment of Multidimensional NmrSpectra of Proteins - Application to 3d Noesy-Hmqc and Tocsy-Hmqc Spectra. J. Biomol. NMR 3, 245–251. 25. Olson, J. B., and Markley, J. L. (1994) Evaluation of an Algorithm for the Automated Sequential Assignment of Protein Backbone Resonances - a Demonstration of the Connectivity Tracing Assignment Tools (Contrast) Software Package. J. Biomol. NMR 4, 385–410. 26. Lukin, J. A., Gove, A. P., Talukdar, S. N., and Ho, C. (1997) Automated probabilistic method for assigning backbone resonances of (C-13,N15)-labeled proteins. J. Biomol. NMR 9, 151–166. 27. Bartels, C., Güntert, P., Billeter, M., and Wüthrich, K. (1997) GARANT - A general algorithm for resonance assignment of

449

multidimensional nuclear magnetic resonance spectra. J. Comput. Chem. 18, 139–149. 28. Choy, W. Y., Sanctuary, B. C., and Zhu, G. (1997) Using neural network predicted secondary structure information in automatic protein NMR assignment. J. Chem. Inf. Comput. Sci. 37, 1086–1094. 29. Buchler, N. E. G., Zuiderweg, E. R. P., Wang, H., and Goldstein, R. A. (1997) Protein NMR assignments using mean-field simulated annealing. Biophys. J. 72, Wp447–Wp447. 30. Croft, D., Kemmink, J., Neidig, K. P., and Oschkinat, H. (1997) Tools for the automated assignment of high-resolution three-dimensional protein NMR spectra based on pattern recognition techniques. J. Biomol. NMR 10, 207–219. 31. Zimmerman, D. E., Kulikowski, C. A., Huang, Y. P., Feng, W. Q., Tashiro, M., Shimotakahara, S., et al. (1997) Automated analysis of protein NMR assignments using methods from artificial intelligence. J. Mol. Biol. 269, 592–610. 32. Gronwald, W., Willard, L., Jellard, T., Boyko, R. E., Rajarathnam, K., Wishart, D. S., et al. (1998) CAMRA: Chemical shift based computer aided protein NMR assignments. J. Biomol. NMR 12, 395–405. 33. Leutner, M., Gschwind, R. M., Liermann, J., Schwarz, C., Gemmecker, G., and Kessler, H. (1998) Automated backbone assignment of labeled proteins using the threshold accepting algorithm. J. Biomol. NMR 11, 31–43. 34. Moseley, H. N. B., Monleon, D., and Montelione, G. T. (2001) Automatic determination of protein backbone resonance assignments from triple resonance nuclear magnetic resonance data. Nuc. Magn. Reson. Biol. Macromol. 339, 91–108. 35. Coggins, B. E., and Zhou, P. (2003) PACES: Protein sequential assignment by computerassisted exhaustive search. J. Biomol. NMR 26, 93–111. 36. Malmodin, D., Papavoine, C. H. M., and Billeter, M. (2003) Fully automated sequencespecific resonance assignments of heteronuclear protein spectra. J. Biomol. NMR 27, 69–79. 37. Hitchens, T. K., Lukin, J. A., Zhan, Y. P., McCallum, S. A., and Rule, G. S. (2003) MONTE: An automated Monte Carlo based approach to nuclear magnetic resonance assignment of proteins. J. Biomol. NMR 25, 1–9. 38. Moseley, H. N. B., Riaz, N., Aramini, J. M., Szyperski, T., and Montelione, G. T. (2004) A generalized approach to automated NMR peak list editing: application to reduced dimensionality triple resonance spectra. J. Magn. Reson. 170, 263–277.

450

P. Guerry and T. Herrmann

39. Eghbalnia, H. R., Bahrami, A., Wang, L. Y., Assadi, A., and Markley, J. L. (2005) Probabilistic identification of spin systems and their assignments including coil-helix inference as output (PISTACHIO). J. Biomol. NMR 32, 219–233. 40. Lin, H. N., Wu, K. P., Chang, J. M., Sung, T. Y., and Hsu, W. L. (2005) GANA - a genetic algorithm for NMR backbone resonance assignment. Nucleic Acids Res. 33, 4593–4601. 41. Masse, J. E., Keller, R., and Pervushin, K. (2006) SideLink: Automated side-chain assignment of biopolymers from NMR data by relative-hypothesis-prioritization-based simulated logic. J. Magn. Reson. 181, 45–67. 42. Masse, J. E., and Keller, R. (2005) AutoLink: Automated sequential resonance assignment of biopolymers from NMR data by relativehypothesis-prioritization-based simulated logic. J. Magn. Reson. 174, 133–151. 43. Wang, J. Y., Wang, T. Z., Zuiderweg, E. R. P., and Crippen, G. M. (2005) CASA: An efficient automated assignment of protein mainchain NMR data using an ordered tree search algorithm. J. Biomol. NMR 33, 261–279. 44. Kamisetty, H., Bailey-Kellogg, C., and Pandurangan, G. (2006) An efficient randomized algorithm for contact-based NMR backbone resonance assignment. Bioinformatics 22, 172–180. 45. Vitek, O., Bailey-Kellogg, C., Craig, B., and Vitek, J. (2006) Inferential backbone assignment for sparse data. J. Biomol. NMR 35, 187–208. 46. Wu, K. P., Chang, J. M., Chen, J. B., Chang, C. F., Wu, W. J., Huang, T. H., et al. (2006) RIBRA - An error-tolerant algorithm for the NMR backbone assignment problem. J. Comput. Biol. 13, 229–244. 47. Volk, J., Herrmann, T., and Wüthrich, K. (2008) Automated sequence-specific protein NMR assignment using the memetic algorithm MATCH. J. Biomol. NMR 41, 127–138. 48. Fiorito, F., Herrmann, T., Damberger, F. F., and Wüthrich, K. (2008) Automated amino acid side-chain NMR assignment of proteins using C-13- and N-15-resolved 3D [H-1,H1]-NOESY. J. Biomol. NMR 42, 23–33. 49. Neidig, K. P., Saffrich, R., Lorenz, M., and Kalbitzer, H. R. (1990) Cluster-Analysis and Multiplet Pattern-Recognition in 2-Dimensional Nmr-Spectra. J. Magn. Reson. 89, 543–552. 50. Garrett, D. S., Powers, R., Gronenborn, A. M., and Clore, G. M. (1991) A Common-Sense Approach to Peak Picking in 2-Dimensional, 3-Dimensional, and 4-Dimensional Spectra

Using Automatic Computer-Analysis of Contour Diagrams. J. Magn. Reson. 95, 214–220. 51. Carrara, E. A., Pagliari, F., and Nicolini, C. (1993) Neural Networks for the Peak-Picking of Nuclear-Magnetic-Resonance Spectra. Neural Networks 6, 1023–1032. 52. Antz, C., Neidig, K. P., and Kalbitzer, H. R. (1995) A General Bayesian Method for an Automated Signal Class Recognition in 2d Nmr-Spectra Combined with a Multivariate Discriminant-Analysis. J. Biomol. NMR 5, 287–296. 53. Koradi, R., Billeter, M., Engeli, M., Güntert, P., and Wüthrich, K. (1998) Automated peak picking and peak integration in macromolecular NMR spectra using AUTOPSY. J. Magn. Reson. 135, 288–297. 54. Herrmann, T., Güntert, P., and Wüthrich, K. (2002) Protein NMR structure determination with automated NOE-identification in the NOESY spectra using the new software ATNOS. J. Biomol. NMR 24, 171–189. 55. Dancea, F., and Gunther, U. (2005) Automated protein NMR structure determination using wavelet de-noised NOESY spectra. J. Biomol. NMR 33, 139–152. 56. Brünger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., et al. (1998) Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Cryst. D 54, 905–921. 57. Schwieters, C. D., Kuszewski, J. J., Tjandra, N., and Clore, G. M. (2003) The Xplor-NIH NMR molecular structure determination package. J. Magn. Reson. 160, 65–73. 58. Güntert, P., Mumenthaler, C., and Wüthrich, K. (1997) Torsion angle dynamics for NMR structure calculation with the new program DYANA. J. Mol. Biol. 273, 283–298. 59. Doreleijers, J. F., Mading, S., Maziuk, D., Sojourner, K., Yin, L., Zhu, J., et al. (2003) BioMagResBank database with sets of experimental NMR constraints corresponding to the structures of over 1400 biomolecules deposited in the Protein Data Bank. J. Biomol. NMR 26, 139–146. 60. Hiller, S., Wider, G., and Wüthrich, K. (2008) APSY-NMR with proteins: practical aspects and backbone assignment. J. Biomol. NMR 42, 179–195. 61. Hiller, S., Fiorito, F., Wüthrich, K., and Wider, G. (2005) Automated projection spectroscopy (APSY). Proc. Natl. Acad. Sci. USA 102, 10876–10881. 62. Güntert, P., and Wüthrich, K. (1992) Flatt - a New Procedure for High-Quality Base-Line

22

Comprehensive Automation for NMR Structure Determination of Proteins

Correction of Multidimensional Nmr-Spectra. J. Magn. Reson. 96, 403–407. 63. Nilges, M. (1995) Calculation of Protein Structures with Ambiguous Distance Restraints Automated Assignment of Ambiguous Noe Crosspeaks and Disulfide Connectivities. J. Mol. Biol. 245, 645–660. 64. Kuszewski, J. J., Thottungal, R. A., Clore, G. M., and Schwieters, C. D. (2008) Automated errortolerant macromolecular structure determination from multidimensional nuclear Overhauser enhancement spectra and chemical shift assignments: improved robustness and performance

451

of the PASD algorithm. J. Biomol. NMR 41, 221–239. 65. Schwieters, C. D., Kuszewski, J. J., and Clore, G. M. (2006) Using Xplor-NIH for NMR molecular structure determination. Prog. Nucl. Magn. Reson. Spectrosc. 48, 47–62. 66. Fiorito, F., Hiller, S., Wider, G., and Wüthrich, K. (2006) Automated resonance assignment of proteins: 6D APSY-NMR. J. Biomol. NMR 35, 27–37. 67. Nilges, M. (1993) A Calculation Strategy for the Structure Determination of Symmetrical Dimers by H-1-Nmr. Proteins 17, 297–309.

Chapter 23 ARIA for Solution and Solid-State NMR Benjamin Bardiaux, Thérèse Malliavin, and Michael Nilges Abstract In solution or solid-state, determining the three-dimensional structure of biomolecules by Nuclear Magnetic Resonance (NMR) normally requires the collection of distance information. The interpretation of the spectra containing this distance information is a critical step in an NMR structure determination. In this chapter, we present the Ambiguous Restraints for Iterative Assignment (ARIA) program for automated cross-peak assignment and determination of macromolecular structure from solution and solid-state NMR experiments. While the program was initially designed for the assignment of nuclear Overhauser effect (NOE) resonances, it has been extended to the interpretation of magic-angle spinning (MAS) solidstate NMR data. This chapter first details the concepts and procedures carried out by the program. Then, we describe both the general strategy for structure determination with ARIA 2.3 and practical aspects of the technique. ARIA 2.3 includes all recent developments. such as an extended integration of the Collaborative Computing Project for the NMR community (CCPN), the incorporation of the log-harmonic distance restraint potential and an automated treatment of symmetric oligomers. Key words: Ambiguous distance restraint, Structure calculation, Automated assignment, MAS, Solid-state NMR, CCPN, NOE, ARIA, PDSD, CHHC

1. Introduction Nuclear Magnetic Resonance (NMR) is widely used in the field of structural biology. Most structure determinations by NMR rely on the measurement of distances and angles between nuclei, the distances playing a crucial role in the fold determination. In solution, these distances are measured by nuclear Overhauser effect spectroscopy (NOESY) (1). The intensity of the nuclear Overhauser effect (NOE), produced by the magnetization transfer through the dipolar coupling between the observed spins, is related to the distance between the two interacting spins. The qualitative estimate of distances from NOE intensities is then translated into interatomic

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_23, © Springer Science+Business Media, LLC 2012

453

454

B. Bardiaux et al.

restraints and the structure is calculated from these restraints. Structure determination from NOEs thus requires the assignment of the NOE cross-peaks to pairs of magnetically interacting spins. However, this assignment cannot generally be obtained without the knowledge of the structure. In fact, unambiguously assigning NOE cross-peaks is sometimes very difficult due to inadequate spectral resolution, chemical shift degeneracy and potentially overlapped cross-peaks. The introduction of the concept of Ambiguous Distances Restraints (ADR) (2) was a breakthrough in the treatment of degenerate NOE assignments, since it actually derives distance information from ambiguously assigned cross-peaks. The intricate relationship between the structure determination and the NOE assignment led to the development of an iterative automatic procedure to simultaneously calculate the structure and assign the NOEs. In this procedure, structure calculation from ADRs and cross-peak assignment are performed alternatively by comparing the tentative ambiguous assignment to an ensemble of molecular conformations determined on the basis of ADRs. The implementation of this iterative strategy is largely automated in the software ARIA (Ambiguous Restraints for Iterative Assignment) (3–6), which is described in more detail. ARIA is an open source software, widely disseminated in the biological NMR community. The use of Magic-Angle Spinning (MAS) with solid-state NMR (ssNMR) spectroscopy was applied to the structure determination of proteins in microcrystalline or fibrilar form. Long-range structural restraints can also be obtained from proton-driven spin diffusion experiments or proton-mediated rare-spin detected correlation experiments. However, cross-peak assignment is complicated by the larger band widths that induce substantial ambiguities in resonance assignments. The first de novo determination of a protein structure from MAS ssNMR marked an important step in the field (7). Shortly after, it was demonstrated that automated methods for cross-peak assignment, such as ARIA or CYANA (8), could be successfully applied to carbon–carbon or proton–proton correlation NMR experiments in the solid-state (9–12). ARIA now incorporates routines for ssNMR structure determination by using various solid-state NMR spectra. In addition, the in-depth integration of the Collaborative Computing Project for the NMR community in ARIA streamlines the structure determination process by NMR by facilitating import and export of data. Other recent improvements to ARIA include: (i) implementation of the network anchoring approach (8, 13) adapted to the ARIA philosophy, (ii) automated treatment of symmetric oligomers (13) and (iii) the availability of the logharmonic potential and the Bayesian estimation of optimal restraint weight (14).

23

ARIA for Solution and Solid-State NMR

455

2. Materials 2.1. ARIA Software Package

The following software packages are required to use ARIA. 1. ARIA software package. ARIA (6) is written in the programming language Python (15). The current version is 2.3. Instructions on how to install ARIA can be found in the ARIA installation archive, which should be downloaded from http:// aria.pasteur.fr. ARIA can be installed on computers operating under Linux, Windows or Mac OS X. 2. CNS software. To enable specific features used by ARIA, it is necessary to compile the CNS program (16) with libraries provided within the ARIA package. 3. Optional: CCPNmr Analysis software package (version 2 or later) (17). ARIA uses the CCPN data model to read input data and to store all results in a general format. 4. Optional: Access to a computer cluster for distributed calculation.

2.2. Input Data

The minimal set of data required by ARIA consists of (see Note 1): 1. Definition of the molecular system. 2. List(s) of chemical shift assignments of 1H (for 2D-NOESY) and 13C/15N if necessary for 3D-NOESY or for MAS solidstate NMR spectra (see Note 2). 3. One or more lists of cross-peaks with chemical shift positions in each dimension and peak volumes/intensities. Individual peaks can be either fully assigned, partially assigned or completely unassigned. A list of cross-peaks generally corresponds to the peaks picked in a particular spectrum. It is recommended that similar experiments performed with different mixing times are entered as separated lists. ARIA also integrates various data types for additional experimental information. All restraints must be in CNS “tbl” format (see Note 1). 1. Hydrogen bonds: The distance between hydrogen donor and acceptor as well as the distance between acceptor and hydrogen. 2. Dihedral angles: Dihedral angle restraints incorporated using a flat-bottom harmonic-wall potential. 3. J-couplings: Calculated J-couplings are directly refined against observed J-couplings. 4. Residual dipolar couplings: Residual dipolar coupling (RDC) data as restraints. 5. Distance restraints: Preformatted distance restraints, e.g., from manual assignments.

456

B. Bardiaux et al.

6. Preliminary structure or structure ensemble:. PDB formatted file(s) from a previous calculation or models (see Note 1). 7. CCPN Project. CCPN project containing the same data as listed above but directly imported into ARIA without format conversion. 2.3. Software for Structural Quality Checks

Additional software required to analyze the quality of the final structure ensembles: 1. PROCHECK (18). 2. WHAT IF (or WHAT_CHECK) (19). 3. ProSa II (or ProSa 2003) (20). 4. MolProbity suite (21).

3. Methods The general workflow of the ARIA methodology is presented in Fig. 1. After an initial chemical-shift based cross-peak assignment and a calibration step, ambiguous distance restraints are derived from the cross-peaks (NOEs or C-C correlations). From these restraints, an ensemble of conformers is calculated. On the basis of these structures, noise peaks are detected with a violation analysis, and unlikely assignment possibilities are discarded. This process is iterated several times (nine by default) with optimized parameters for each iteration. Each step of the protocol is described in detail in the sections below. 3.1. Preparation Phase

Before cross-peak assignment and structure calculation, the following steps are automatically performed by ARIA. First, the data are checked and filtered for errors and inconsistencies. The program then creates the molecular topology of the system.

3.1.1. Data Filtering

When checking the chemical shift assignments for consistency, ARIA considers three possible situations: 1. A unique assignment consisting of a single atom and a single chemical shift. 2. A degenerate chemical shift assignment, where one group of equivalent atoms is assigned to exactly one chemical shift. 3. An assignment of the two substituents of a prochiral group, which can have one or two chemical shifts. In the latter case, floating chirality assignment (22) is used in the resulting restraints (cf. Subheading 3.3.5). Peaks that lack frequency information or with incorrect/missing peak sizes are removed (see Note 3).

23

ltering Molecular topology creation

ARIA for Solution and Solid-State NMR

a

457

b

Initial cross-peak assignment

ARIA Iterative protocol c

Chemical shits assignments

Cross-peak lists

Calibration d

Structure ensemble

e

Noise peaks removal f

Partial Assignment Additional restraints (dihedrals, RDC, distance)

ARIA (GUI)

Distance restraints

g

Restraints Merging

Structure ensemble

Violation Analysis Molecular ntion

Structure Calculation

nition

h

Floating chirality assignment j

Generation of report Quality analysis

Cross-peak assignments

Restraint violation list

i

nement in explicit solvent

PDB structure ensemble

Structure quality statistics

Fig. 1. Description of the ARIA protocol workflow. Rounded rectangles indicate steps performed by ARIA, folded rectangles correspond to user provided input-data and trapezoids represent results.

3.1.2. Molecular Topology Creation

From the definition of the molecular system provided as input data, ARIA creates a molecular topology file (MTF) with the program CNS (16). Name, chemical type, charge and mass of each atom as well as the covalent connectivity are defined in the MTF. An extended conformation of the molecule is then generated by CNS and the coordinates are stored in a PDB file (cf. Subheading 3.8). The molecular topology is created automatically for standard biopolymers. If applicable, topological features can be easily defined by the user through the graphical interface (cf. Subheading 3.7).

3.2. Initial Cross-Peak Assignment

For every cross-peak, ARIA uses the chemical shift lists from the sequential resonance assignment to derive possible assignments. As illustrated in Fig. 2, the peak position is defined by its frequency coordinates (c1, c2) in each dimension of the spectrum. To account for the limited precision in chemical shift measurements, for the uncertainty of the cross-peak coordinates and for systematic experimental errors, chemical shift tolerances (d1, d2) are applied around the peak position. The tolerances should be chosen to be sufficiently

3.2.1. Chemical-Shift Based Assignment

B. Bardiaux et al. dimension 2

458

c1−δ1

c1

c1+δ1

c2+δ2

pz py

c2

px

c2−δ2

pa pb pc pd

dimension 1

Fig. 2. Illustration of the assignment of a cross-peak. c1,c2 denote the peak coordinates in frequency space. The assignment frequency window is indicated by the solid black square, defined from the chemical shifts tolerances d1 and d2. The coordinates of the (hypothetical) correct assignment are represented by the gray dashed lines (pb, py). Multiple resonances within the tolerance window (pa, pb, pc, pd in dimension 1 and px, py, pz in the other dimension) give rise to 12 assignment possibilities.

large to obtain frequency windows that can compensate for all sources of inconsistencies between the list of resonance assignments and the cross-peak lists. Then, for each peak dimension, all protons (or 13C/15N spins for MAS ssNMR) whose chemical shifts fall in the peak frequency windows are collected (see Note 4). In the case of 3D or 4D heteronuclear spectra, the hetero atom attached to the proton must also match the corresponding chemical shift window. The list of all assignment possibilities (or contributions) for a cross-peak is generated from the combination of the resonances assignment (Fig. 2). The sizes of the frequency windows play an important role in the initial cross-peak assignment step (see Note 5). In addition, the completeness of the chemical shift assignments influences the accuracy of the initial assignment (see Note 6). For symmetric oligomers, since symmetric nuclei will have the same chemical shifts, ARIA will collect possible assignments for all monomers. To simplify the treatment of the resulting highly ambiguous assignments (see Note 7), ARIA considers only one dimension (of the two corresponding to the through-space correlation) as ambiguous in terms of chain assignment. Later on, the corresponding symmetric restraints will be automatically generated by ARIA prior to structure calculation. ARIA also takes into account information about the intramolecular or intermolecular nature of the experiment (if applicable and specified by the user) by excluding the nonvisible contributions. 3.2.2. Structural Rules for Symmetric Oligomers

ARIA can use information about the secondary structure organization of the system under investigation to remove unlikely assignments. ARIA uses simple rules (23) to assign some cross-peaks as intermonomer before the structure calculation, using the predicted

23

ARIA for Solution and Solid-State NMR

459

secondary structure elements (see Note 8). If two symmetric secondary structure elements are facing each other in the interface, cross-peaks observed within the same element between residues separated by more than five residues in sequence cannot arise from intramolecular contacts and are thus unambiguously classified as intermolecular. 3.2.3. Network Anchoring

ARIA implements a network anchoring approach (8) to reduce the number of possibilities of cross-peak assignments prior to structure calculation. The approach is based on the ranking of each assignment, calculated using the information about the assignments of neighboring nuclei in 3D space, and is efficient because true assignments form a self-consistent subset of the network of all possible assignments (see ref. 8, 13 for details). The behavior of network anchoring is controlled by a set of user-defined parameters: 1. “High network-anchoring (NA) score per residue threshold” high (N res ). min 2. “Minimal NA score per residue threshold” (N res ). min 3. “Minimal NA score per residues threshold” (N atom ).

A peak is conserved if one of the following rules is verified: high S res ≥ N res

(1)

min min S res ≥ N res and S atom ≥ N atom

(2)

where Sres and Satom are respectively the residue-wise and atom-wise network anchoring score. Even though the network anchoring approach does not directly rely on 3D structure information, it is still possible to use it after the first ARIA iteration. 3.3. Iterative Structure Calculation 3.3.1. Ambiguous Distance Restraints

The most important idea that underlies the ARIA methodology is the concept of Ambiguous Distance Restraints (ADR) (2). In the framework of the ADR, each NOESY cross-peak is treated as the superposition of the signals from each of its multiple assignments possibilities: the NOE intensity depends on the sum of the inverse sixth power of all the individual proton–proton distances that contribute to the signal. An effective distance D is thus derived as: ⎛ Nc ⎞ D = ⎜ ∑ dc−6 ⎟ ⎝ c =1 ⎠

−

1 6

(3)

where c runs through all Nc assignment possibilities and dc is the interatomic distance between the two protons corresponding to the c-th contribution. During structure calculation, in a similar fashion as for unambiguous distance constraints, the distance D in the molecular coordinates is restrained through the distance target energy function (cf. Subheading 3.3.5).

460

B. Bardiaux et al.

3.3.2. Distance Calibration

The simplest model to derive distances from NOE signal intensity is the Isolated Spin Pair Approximation (ISPA), which considers only the observed spin pair, neglecting spin diffusion through third nuclei. For short mixing times, ISPA provides a good approximation to relate an NOE volume (Vij) to the distance dij of two interacting spins i and j: Vij = Cdij−6

(4)

The scale factor C (also named calibration factor) cannot be measured directly since it depends on the system under investigation and on the experimental setup. The calibration factor is estimated for all NOEs from the ratio of the average of the experimental volume, Vexp, to the average of the theoretical volume: C=

∑V

exp

i

∑ dˆ

(5)

−6

i

i

where dˆi is the average effective distance for NOE i in the conformer ensemble. In the case of multiple assignment possibilities, dˆi is calculated according to equation of ADR Eq. 3. Finally, the calibrated distance is obtained by: d = (C −1V exp )

−

1 6

(6)

In the case of NOE between two groups of magnetically equivalent spins (e.g., methyl groups and aromatic rings), averaging effects are taken into account by expanding Eq. 4 (see Note 9). Magnetization can also be transferred from one spin to another not only directly but also by spin diffusion, i.e., indirectly via other spins in the vicinity. For longer mixing times, the spin-diffusion phenomenon must be considered in the estimation of the distance. When applying ISPA the resulting interproton distances are therefore mostly underestimated. ARIA employs relaxation matrix theory to account for indirect magnetization transfer. In this formalism, cross-peak volumes at mixing time tm can be calculated given the volumes at tm = 0 and the matrix of auto- and cross-relaxation rates, R (24): Vij (t m ) = CVij (0)(exp(−Rt m ))ij

(7)

The resulting NOE back-calculated volumes, which take into account the bias induced by spin-diffusion, are then converted into corrected target distances d: V exp ⎞ ⎛ d = dˆ ⎜C −1 ⎟ V th ⎠ ⎝

−

1 6

(8)

23

ARIA for Solution and Solid-State NMR

461

where dˆ is the average effective distance, and V exp and V th are the experimental and theoretical NOE volumes, respectively. When using spin-diffusion corrected distances, the distance bounds calculated from the theoretical volume may also be of use for the structure calculation (25). In ARIA 2.3, the spin-diffusion correction is performed by the python core of ARIA and not by CNS routines. It is also important to note that every spectrum is independently calibrated. Still, these models are approximate and it is common practice to restrain the distance to an interval to account for uncertainties in the distances (see Note 10). This interval is thus defined by lower and upper distance bounds, L and U: L = d − Δ,U = d + Δ where Δ = 0.125d 2 3.3.3. Violation Analysis and Noise Peak Removal

(9)

To identify incorrect assignments and noise peaks, the calibrated restraints are treated with a violation analysis, following the structural consistency hypothesis (3, 26): incorrectly assigned peaks or noise peaks are not consistent with the 3D structure determined with all experimental data. To assess whether a particular restraint follows the “general trends” imposed on the structures by the entire data set, the obtained distance bounds are compared to the corresponding distances found in the conformer ensemble. A restraint is considered as violated if the distance found in the structure lies outside the bounds by more than a user-defined violation tolerance, t. To identify systematically violated restraints, each conformer in the ensemble is analyzed. The fraction, f i , of conformers violating restraint i is calculated according to: fi =

1 S ∑ max(Θ(Li − t − di(k) ), Θ(di(k) −U i − t )) S k =1

(10)

where Li and Ui denote the lower and upper bounds of the i-th restraint, di(k) designates the distance found in the k-th conformer; Q is the Heaviside step function and S is the total number of conformers analyzed. A restraint is classified as violated if f i exceeds a user-defined violation threshold (50% by default). The corresponding cross-peak is thus removed from the list of active peaks for the next iteration. During the course of the protocol, the violation tolerance, is reduced from iteration to iteration to ensure that most of the inconsistent peaks are removed. 3.3.4. Partial Assignment

The assignment of cross-peaks is made in an indirect fashion by progressively eliminating unlikely assignment possibilities. Due to the r −6 dependence, assignments with large distances contribute only little to the NOE intensity. Thus, for a particular cross-peak, each assignment possibility is weighted by its normalized partial volume, wc , calculated as follows: wc ∝ dc−6

(11)

462

B. Bardiaux et al. Nc

∑w

c

c =1

=1

(12)

where dc is the average distance of the contribution c in the structure ensemble and Nc, the number of contributions for the crosspeak. To reduce the number of assignment possibilities, only the m largest contributions satisfying the following condition are kept: m

∑w

c

≥p

(13)

1

where p designates a user-defined ambiguity cut-off. This cut-off is set to 1.0 in the first iteration and progressively reduced to 0.8 so that for most peaks unambiguous assignments can be derived in the last iteration. The quality of NMR structure ensembles might also be improved by excluding peaks that involve a large number of contributions. This function is controlled by the parameter max_n, which defines the maximum number of assignment possibilities (4). Symmetric peaks or duplicate peaks from different experiments lead to equivalent restraints (restraints involving the same set of atoms). To avoid overrepresentation of certain distance data, nonviolated restraints with equivalent atom content are detected. The restraint with the smallest distance is kept, while the others are discarded for the rest of the protocol. For every iteration, the file noe_restraints.merged lists restraints discarded by the merging procedure. 3.3.5. Calculation of Structure Ensemble

On the basis of the merged restraints list, a new structure ensemble is calculated with the program CNS (16) through a molecular dynamics simulated annealing (MDSA) protocol. ARIA provides two forms of molecular dynamics : in Cartesian or torsion angle space. Torsion angle molecular dynamics (TAD) (27) reduces the calculation time and allows for higher MDSA temperatures, while generally increasing the convergence radius. The molecular structures obtained with TAD also provide better local geometries. The MDSA protocol used in ARIA is divided into two phases : an initial high temperature search phase, and a cooling phase where the temperature slowly decreases. The second part of the cooling stage is performed in Cartesian coordinates. The length of the cooling stages determines the slope of the bath temperature cooling function. It has been shown that this parameter plays an important role in the convergence properties of the ARIA calculation for highly ambiguous data (28). The MDSA protocols implemented in ARIA (3) are optimized for the application of ambiguous distance restraints and for the violation analysis method. The minimization protocols are based primarily on separate scaling of different energy terms with relatively low force constants. Any other structural

23

ARIA for Solution and Solid-State NMR

463

Table 1 Important protocol parameters, their location in the GUI, and defaults values (if applicable) Parameter

GUI item

Default value

Project environment Project name File root Working directory Temporary directory

Project

Data specification Frequency window (proton) Frequency window (hetero) Trust assignments Use only assigned Symmetry CNS topology file CNS parameter file

Data Spectra Spectra Spectra Spectra Symmetry Molecular system Molecular system

0.02 0.5 No No None topallhdg5.3.pro parallhdg5.3.pro

Protocol parameters Number of structures (n_structures) Violation tolerance (t) Violation threshold Ambiguity cutoff (wc) Maximum nb. of contributions (max_n) Number of lowest energy structures (S) Solvent for refinement

Protocol Iterations

20

Iterations Iterations Iterations Iterations

1000.0–0.1 0.5 1.0–0.8 20

Iterations

7

Water refinement

Water

Structure calculation Local CNS executable Command to start remote calculation High temperature steps Cooling 1 steps Cooling 2 steps Log-Harmonic potential

Structure Generation CNS Job Manager

1

CNS Dynamics CNS Dynamics CNS Dynamics CNS Annealing Parameters

10,000 5,000 4,000 No

restraints available are also used during the structure calculation (hydrogen bond restraints, dihedral angles and RDCs). The number of calculated conformers is an important parameter of the structure calculation protocol. Among all calculated conformers, only the n-lowest energy ones (usually n = 30%) will be used in the next ARIA iteration to recalibrate and reassign NOEs. For every iteration, the number of structures is a user defined parameter (see Table 1).

464

B. Bardiaux et al.

3.3.6. Restraint Energy Function

The aim of the MDSA protocol is to find a global energy minimum of an objective function that incorporates experimental data and physical energy. The latter is quantified by using a molecular dynamics force field. Experimental data are integrated in the form of conformational restraints entering the objective function via an energy potential. For distance restraints, ARIA employs an flatbottom harmonic-wall potential with zero-energy between the distance bounds and linear asymptotes (3). This potential allows for large distance violations as may occur in an automated assignment procedure. Nevertheless, it is still difficult to correctly evaluate the bounds and the relative weight to apply to the data. Recently, we have introduced an new error-tolerant potential where lower and upper bounds are replaced by a bounds-free log-harmonic potential (14). This potential derives from a Bayesian analysis showing that NOEs and the derived distances ideally follow the log-normal distribution (29, 30). In ARIA, we also retain another important feature of this Bayesian approach: automatic determination of the optimal weight for the experimental data (31). The log-harmonic potential is applied during the second cooling stage of the MDSA and during water refinement. The weight for the distance restraints, wdata , is iteratively evaluated as: wdata =

n χ (X ) 2

(14)

where n is the number of restraints, and: ⎡d ⎤ χ 2 (X ) = ∑ log 2 ⎢ i ⎥ ˆ i ⎣⎢ di ⎦⎥

(15)

where, for each restraint i, dˆi is the effective distance Eq. 3 calculated from the current structure, and di is the target distance of the restraint. This approach was shown to generally improve the accuracy as well as the quality of the structures calculated from assigned restraints (14). Our initial experience in using ARIA with real (noisy and ambiguous) data indicates that the log-harmonic restraint potential is preferable. 3.3.7. Symmetric Oligomers

The symmetry of the system is maintained during the calculation by adding a symmetry target function to the objective energy function (32). This target function contains terms that ensure the symmetry relation between the monomers and keep them in the vicinity of each other (Packing, see Note 11).

3.3.8. Floating Chirality Assignment

The treatment of unassigned prochiral groups is realized with a floating chirality assignment approach (22). The two substituents of a prochiral center (methylene protons or methyl protons of isopropyl groups) are often difficult to assign stereo-specifically, in terms of chemical shifts. In each proton dimension, a resonance

23

ARIA for Solution and Solid-State NMR

465

matching one of the chemical shifts may potentially involve either of the two prochiral substituents. In ARIA, the two assignment alternatives are tested during the course of the structure calculation and the most energetically favorable possibility is used. The result is written for each conformer in a file with a .float extension. The simplified force field parameters for nonbonded contacts applied to structure calculations in vacuo often produce structures that contain artifacts (unrealistic side-chain packing and unsatisfied hydrogen bond donors or acceptors). Therefore, the final structures of the last ARIA iteration are automatically refined in a shell of explicit solvent (water or DMSO molecules). This refinement consists in a short MD with a complete force field, which includes coulombic and Lennard-Jones potentials. The covalent parameters used in the refinement (33) are consistent with the force field used for structure calculation and validation, thus avoiding systematic differences that could influence validation results. It has been shown that the refinement in solution significantly improves the quality of the structure (33–35).

3.4. Solvent Refinement

3.5. Results Export and Generation of Output Files 3.5.1. Export to CCPN

At the end of the ARIA protocol, assigned peak lists, restraint lists, along with violations, and final structure ensembles (last iteration and solvent refined) are automatically exported into a CCPN project (see Fig. 3). Data exchange, further analysis of results, and management of ARIA runs are then facilitated through the use of the CCPN program suite (cf. Subheading 3.11).

ARIA Cross-peak assignments Distance restraints

nition

Violations

Distance restraints Hydrogen bond restraints

EXPORT

Chemical shift assignments

IMPORT

Cross-peaks lists

Final structure ensemble

Dihedral angle restraints RDC restraints Initial structure ensemble

CCPN project

CCPN Analysis

Fig. 3. Communication interface between ARIA and CCPN for import of input data and export of results.

466

B. Bardiaux et al.

3.5.2. Report Files

For every iteration, ARIA creates the following report files: 1. report summarizes analyses of the restraint lists and the structure ensemble (number of restraints applied, violations, ensemble precision). 2. noe_restraints.unambig, noe_restraints.ambig tabulates information about unambiguous and ambiguous restraints, respectively. For each restraint, the reference crosspeak, restraint bounds and the average distance found in the ensemble are provided. The result of violation analysis is also given here (see Note 12). 3. noe_restraints.violations lists all violated restraints. 4. noe_restraints.assignments lists the tentative assignments corresponding to every restraint. The nature of the assignment(s) is also given (fully, partially or unassigned crosspeaks). 5. noe_restraints.xml, noe_restraints.pickle stores the complete list of cross-peak based distance restraints in XML format and Python binary format. The latter is required for further assignment analysis in the ARIA GUI (cf. Subheading 3.10.2).

3.5.3. Quality Checks

To evaluate the structural quality of both the final set of structures and the solvent-refined ensemble, ARIA makes use of the programs WHAT IF (19), PROCHECK (18), ProSa (20) and MolProbity (21). Separate report files are generated for every program, named quality_checks.*, and are stored in the directories of the respective ensembles (last iteration and solvent-refined). Overall quality scores are tabulated in the file quality_checks, whereas WHAT-IF score profiles along the molecular sequence are generated in both textual and graphical forms (cf. Subheading 3.10.3).

3.5.4. CNS Analyses

CNS scripts calculate restraint energies, ensemble RMSDs, an optimal superposition of the final structure ensemble (with automated determination of flexible and rigid regions), and an unminimized average structure. Analyses of restraints from complementary experimental data are also given. Results are stored in the directory analysis/. In the following sections, we detail the typical procedure to be followed by a user to perform an ARIA calculation. In a structure determination project, the general procedure consists of repeated ARIA runs using revised results from a previous calculation as input data (Fig. 4).

3.6. Conversion of Input Data

Since most NMR software packages use proprietary formats for data storage, the interconversion step required to transfer data with other applications such as ARIA can lead to a loss of information.

23

ARIA for Solution and Solid-State NMR

Initial stage

Series of ARIA runs

Preparation of input data Parameters and project setup

Setup of a new run

467

Adjustment of frequency windows

Revision of input data

ARIA Automated cross-peak assignment and structure calculation

Examination of quality checks for nal structure ensemble

Completion of cross-peak assignments Removal of potential noise peaks

les

Analysis of violations and proposed assignments

Final result

Fig. 4. A series of ARIA runs in a typical structure determination project, with several cycles of structure calculations and cross-peak assignments punctuated by manual inspection and correction of experimental input data.

To facilitate data validation and integration, ARIA uses a data format based on the extensible markup language (XML) (36) to describe molecular systems, chemical shifts, and cross-peaks lists. If input data are intended to be read from a previously created CCPN project, the conversion step described here is no longer required (see Fig. 3). Input data will be read directly and internally converted from the CCPN data model into ARIA at run-time. It is otherwise necessary to convert input data to ARIA XML format before starting the ARIA program per se. This step is simplified by the internal conversion routine provided by ARIA. To use this routine, one must prepare a simple XML conversion file. 1. Conversion template. A preformatted conversion file can be auto-generated by typing the following command in a terminal: aria2 --convert -t conversion.xml An empty conversion file template, “conversion.xml” is then created and must be completed. 2. Editing the conversion file. In addition to formats and filenames of the raw data (sequence, chemical shift lists and spectra) (see Note 13), the user has to specify the mapping between nuclei and frequency dimensions. If the molecular system is a symmetric multimer, it is mandatory to specify the molecular chains involved (segment id or segid). For the cross-peak lists, the user needs to indicate the chains involved and the level of chain-wise

468

B. Bardiaux et al.

ambiguity. Possibilities are intramolecular, intermolecular, or unknown. (cf. Subheading 3.7). For solid-state NMR experiments, a parameter has to be filled in by the user to designate the type of experiment and transfer (see Note 14). 3. Conversion step. Then, invoke the command aria2 --convert conversion.xml to start the data conversion. Converted data will be written in ARIA XML format; a project file, which has to be completed by the user, will be generated as well. 3.7. Specification of ARIA Project Parameters

1. Project creation. All program parameters and locations of the input data are stored in single project file (in XML format). To conveniently change or review the project settings, ARIA provides a Graphical User Interface (GUI) (Fig. 5). Entering the following command will start the GUI and load the project definition from project.xml (see Note 15) aria2 --gui project.xml

Fig. 5. Graphical User Interface of ARIA 2.3 for project management, where data and protocol settings can be modified graphically.

23

ARIA for Solution and Solid-State NMR

469

Important program and protocol parameters are listed in Table 1. Default settings are provided for the rest of the parameters. 2. General settings. Mandatory parameters are related to the general infrastructure of the project, e.g., the name of the project, the directory where an ARIA run will be stored (Working directory) or the prefix (File root) used by ARIA throughout the project for naming PDB files. 3. Sequence definition. It is necessary to provide here the definition of the molecular system. A project file created during the conversion step will already display the location of the XML file of the molecular sequence. Otherwise, the “Browse” button assists in locating the sequence definition XML file. If the sequence has to be read from a CCPN project, the user should first locate the CCPN project in the CCPN data model panel in the GUI. Then, the “CCPN” format has to be chosen for the sequence, and hitting the “Select” button will open a popup window displaying available molecular systems contained in the CCPN project. This procedure is common to all steps where import of data from a CCPN project is available. 4. Adding input data. Spectra and additional experimental data can be added by clicking the “Add” button in the GUI menu. When adding a spectrum, it is necessary to provide both the location of the cross-peak list and the corresponding chemical shift list. Additional experimental data can be supplied in the form of CNS “tbl” formatted files or from a CCPN Project. In the latter case, supplementary options are offered when the distance restraints list is added. For instance, distance restraints can be selected to enter the iterative protocol, where they will be recalibrated and filtered like restraints derived from the internal ARIA cross-peak assignments procedure. Otherwise, they will be kept untouched by the program during the entire protocol. 5. Adjusting data parameters. For each spectrum the default frequency window sizes should be adjusted. When a user wants to apply spin-diffusion correction, the necessary parameters need to be entered (molecule correlation time, spectrometer frequency and mixing time). The nature of the cross-peak in terms of possible chain assignment should also be specified here in the case of symmetric oligomers. This option intends to make better use of possible information arising from filtered/ separated experiments recorded on asymmetrically labeled samples. Finally, for solid-state NMR spectra, we recommend specifying lower and upper distance bounds that will be applied to the cross-peak derived restraints (see Note 10). If applicable, its is furthermore possible to pick an appropriate labeling scheme (see Note 14). In addition, parameters relative to dihedral

470

B. Bardiaux et al.

angle restraints (see Note 8), RDC (see Note 16) and J-couplings (see Note 17) should be defined. 6. Symmetry. ARIA can treat oligomers with C2, C3, C5 or D2 symmetry (see Note 11). 7. Specifying topology patches. By default, ARIA supports the following cases: Disulfide bridges (unambiguous or ambiguous) (2), Histidine protonation states, cis-proline and tetrahedral coordination of Zinc ions. In the case of nonstandard residues or other chemical compounds, manual intervention of the user is required (see Note 18). 8. Iteration parameters. The mode of restraint calibration has to be specified : ratio of average (default), spin-diffusion correction or fixed bounds (see Note 10). For every iteration, default values are provided for protocol parameters (Table 1) and the network-anchoring thresholds (see Note 19). 9. Job Manager. Distributing structure calculations to multiple processors speeds up the ARIA protocol. ARIA provides support for several job submission modes (see Note 20). The appropriate command should be entered and the correct path to the remote CNS program executable should be specified. 10. Structure calculation parameters. The remaining parameters are related to the molecular dynamics simulated annealing, and in particular the number of steps, restraint force constants and potential shape (flat-bottom-harmonic-wall and log-harmonic). 3.8. Project Setup

At this point, the project must be set up with the following command. aria2 --setup project.xml The project is then validated and ARIA creates the directory tree for the project (directory run1). As shown in Fig. 6, the results of the successive iterations are stored in structures/, each iteration having its own subdirectory, e.g., structures/it0/. Experimental data files are copied into their respective directory in data/ (see Note 21). Report files for the cross-peak filtering procedure are stored in data/spectra/. All data, protocols, parameters, and topology files used by CNS reside in the cns/ subdirectory.

3.9. Starting an ARIA Run

It is now possible to launch the ARIA calculation, using the following command: aria2 project.xml ARIA will then automatically perform all the steps listed in Subheadings 3.1–3.5. The main ARIA job will be executed on the local machine where it has been started. According to the job manager settings of the project, the structure calculations will be

23

ARIA for Solution and Solid-State NMR le

begin sequence

nition (XML)

templates

Template structures (PDB)

spectra

Cross-peaks and chemical shifts lists (XML) de bonds (TBL)

ssbonds data

hbonds

Location where ARIA stores input data

jcouplings rdcs

Distance restraints for hydrogen bonds (TBL) Restraints for J-Couplings (TBL) Residual Dipolar couplings restraints (TBL)

dihedrals

Dihedral angle restraints (TBL)

distances

User provided distances restraints (TBL) rst ARIA iteration (iteration 0)

it0 it1 run1

structures

...

ARIA run directory

les for each iteration are stored here

it8

analysis

Various analysis results (performed by CNS) les (PostScript)

graphics le to visualize restraints

molmol

Last iteration

les

cns ne protocols cns

data

Files used for structure calculation

471

W

ned structures and quality-checks analysis

Simulated-annealing protocols (CNS) Input data for simulated annealing nition

toppar

T

begin

Template structure for simulated-annealing

Fig. 6. Illustration of the directory tree of an ARIA project and details about the content. Final results can be found in the directories marked in gray.

successively launched on the local processor (default behavior) or dispatched to a computer cluster (see Note 20). 3.10. Checking the Results

3.10.1. Convergence

In the next paragraphs, we list the points of interests when inspecting the calculation results, along with some guidance on how to correct input data and adapt the protocol parameters. The level of convergence indicates how well the protocol managed to find a well-defined structure and a consistent set of assignments. Convergence can be estimated with two indicators: 1. The average (and variation) of the total energy of the structure ensemble 2. The conformational variance of the structure ensemble (or precision) expressed as a RMSD. A low average energy (see Note 22) and a high precision (RMSD < 1.5 Å) generally mean that convergence has been reached. Other situations may stem for unadapted protocol settings or incomplete or low quality data. The average energy can be found in structures/it8/analysis/energy.disp and the precision in the report file or in structures/it8/analysis/rmsdave.disp.

472

B. Bardiaux et al.

3.10.2. Automated Assignments

1. The report files listed in Subheading 3.5 provide analyses on all restraints and particularly which restraints have been classified as violated. Restraints showing consistent violation greater than 0.1 Å should be inspected manually. Restraints with large upper-bound violations (³5 Å) in the majority of the conformers (³85%) usually result from incorrect assignments. Restraints detected as such should not be used in a later ARIA run and the corresponding cross-peak removed from its respective spectrum. Other assignments should be considered as “reliable” in a subsequent run. 2. Analyzing text files for violations and assignments can be a tedious task. ARIA also provides ways to investigate this in a graphical manner (37). Postscript files describing the restraints, based on the RMS of violations are generated automatically during a run. These values are displayed at the residue level, in the form of a profile along the protein sequence, or as a contact map for the RMS of violations per residue pair (Fig. 7a). The contact map displays the sum of the RMS of violations per residue pair. In the profile, the sum of the RMS of violations per residues is plotted along the protein sequence. In addition, the program provides an interactive tool to browse assignments at the residue level (Peak map). A peak-map can be viewed for all iterations in the ARIA GUI (Fig. 8). Clicking on a contact

Fig. 7. Per-residue quality plots. (a) Contact map displaying the sums of RMS deviations and a profile of the RMS deviations. (b) WHATIF score profiles along the sequence. The RMS deviations are plotted on a color scale (figure adapted from ref. 25).

23

ARIA for Solution and Solid-State NMR

473

Fig. 8. Interactive peak map. Right panel of the ARIA 2.3 GUI showing the interactive peak map at iteration 8 of an ARIA run. Each pixel of the map located between residues i and j is clickable and opens an assignment report, which contains the list of peaks that exist between residues i and j, along with their contributions (figure adapted from ref. 25).

between residues i and j opens a pop-up window that shows a list of ARIA restraints involving atoms from both residues, where restraints are labeled. Such graphical representations can be useful to detect regions of the structure where violations are concentrated, indicating where restraints and assignments should be more thoroughly investigated. 3. Finally, the resulting restraints and assignments that are exported to a CCPN project can be later investigated with the CCPNmr Analysis software. As illustrated on Fig. 9, CCPNmr Analysis offers utilities to inspect restraints through a customizable user interface. Moreover, a user will be able to examine the proposed resonance assignments directly in a spectral display window at the positions in frequency space where the peaks were picked. 3.10.3. Quality Indices

The quality of structure ensembles as determined by independent structure validation is widely acknowledged as a good indicator of the performance of the structure calculation protocol and of the reliability of the structure. The application of NMR restraints for structure calculation may induce distortions in the geometry of the molecular structure. For this purpose, ARIA applies four major programs (PROCHECK (18), WHAT IF (19), ProSa (20) and MolProbity (21)) that aim at detecting outliers and abnormalities

474

B. Bardiaux et al.

Fig. 9. Screenshot of CCPNmr Analysis windows showing the result of an ARIA run.

in macromolecular structure by comparing several characteristic geometric properties to a database of small molecules and/or highresolution X-ray structures. The summary of all global quality indices is given in the quality_checks file. For thorough reviews of tools to evaluate the quality of NMR structures, we suggest consulting the following references (38, 39). We would like to stress here that despite the apparent lower resolution of solid-state NMR data, a great deal of attention should still be given to the inspection of such quality checks. The following scores should be investigated further (see Note 23). 1. Procheck Ramachandran percentage. For typical NMR structures deposited in the PDB, 80% of the dihedral angles lie within the preferred region of the Ramachandran plot. For high-resolution NMR structures, a higher percentage is expected (90%).

23

ARIA for Solution and Solid-State NMR

475

2. WHAT-IF Z-scores. WHAT-IF results are presented in the form of overall Z-scores. In general, structures with Z-scores between −2 and +2 are considered to be within a normal range and are thus good structures, while structures with Z-scores lower than −2 should be inspected further. Useful indicators of good quality are “Backbone conformation” and “Packing quality”. The “bump-score” also reports the number of van der Waals violations per 100 residues. 3. WHAT-IF profiles. Recently, some studies have stressed that global structural indicators are not sufficient to detect errors in structures and suggested examining parameters on a per-residue basis (40, 41). Such profiles for the WHAT-IF scores are produced by ARIA in the form of a PostScript file (Fig. 7b). Thus, poor quality regions can be precisely identified (see Note 24). 4. Molprobity clashscore. This reports the number of overlaps >0.4 Å per thousand atoms. For typical NMR structures deposited in the PDB, this score is generally high (>10). From our experience, the application of the log-harmonic potential along with automated weight estimation significantly improves this situation. 3.11. Preparing a New Run

To use the result of an ARIA run to further improve the structure, it may be necessary to correct the input data. At this stage, we recommend preparing a new ARIA project for better bookkeeping. CCPNmr Analysis also offers a utility to manage the input and output of successive ARIA runs (Fig. 9). The same CCPN project can be used in multiple ARIA runs.

3.11.1. Correction of Input Data

1. Peaks identified as erroneous (noise peaks) should be deleted from the input data. 2. Automated assignments may be added in the initial cross-peak assignment and incorrect assignments removed. 3. To improve convergence, reliable assignments can be used either as distance restraints or set individually as reliable in the input XML file.

3.11.2. Adjusting Parameters

In the new project file, protocol parameters may also be changed according to the result of a previous calculation. We list here the most important parameters that ought to be adapted. 1. The number of dynamic steps required for convergence is determined by the system size and the level of ambiguity or incompleteness of the input data. Default values work well for systems up to about 100 residues studied with NOESY. However, for larger systems (e.g., symmetric oligomers) or when MAS solid-state NMR data are used, it might become necessary to increase the number of steps in the cooling stage

476

B. Bardiaux et al.

of the simulated annealing protocol. On the one hand, the computation time to calculate a structure will increase with the length of the dynamics. On the other hand, a slow-cooling strategy substantially increases the probability of success of the minimization protocol (see Note 25). 2. In case of poor convergence, one should also check frequency window sizes. Narrow windows affect the completeness of a cross-peak assignment. It may therefore be judicious to slightly increase the individual window size (e.g., by 10%). Conversely, when the final set of restraints is still largely ambiguous, it is reasonable to reduce the window sizes. 3. Achieving convergence may also be hampered by a tight violation tolerance. If a large number of restraints are rejected, the data may be become too sparse. Also, if an initial ensemble of template structures (from a previous calculation for instance) is specified, the default tolerance must be reduced for the first iteration (e.g., 5 Å).

4. Notes 1. Data can be read from common NMR formats or via the CCPN program suite. Compliant formats are the following : Ansig (42), NMRDraw (43), NMRView (44), Pipp (45), Pronto (46), Sparky (T. D. Goddard and D. G. Kneller, University of California), XEasy (47), Diana (48), and NMRStar (49). PDB files with CNS(16), IUPAC(50), or DYANA (51) atom name nomenclatures can be read by ARIA. Restraints files should follow the CNS/XPLOR syntax and nomenclature. Mismatch in segment id (segid) between the restraints and the molecular definition is often a source of errors. ARIA internally follows the IUPAC (50) recommendations for the atom name nomenclature. Most common naming problems are the following: ●

The C-terminal carboxyl group is named O¢ and O″. O″ contains two apostrophes (ASCII 39), not a quotation mark (ASCII 34). The PDB uses O and OXT or OT1 and OT2 instead.

●

The N-terminus consists of H1, H2, and H3 (not HT1, HT2 and HT3).

●

The protein backbone amide proton is called H (instead of HN).

●

The glycine alpha protons are HA2 and HA3.

●

Pseudoatoms (52) are not supported, r–6-averaging is applied to equivalent groups.

23

ARIA for Solution and Solid-State NMR

477

2. ARIA supports CHHC/NHHC (53) and 2D/3D 13C-13C correlation spectra, i.e., PDSD (54, 55), DARR (56), and PAR (57). 3. We always use absolute values of peak sizes (volume or intensity). 4. For C/NHHC experiments, cross-peak assignment is performed on the basis of 13C/15N chemical shifts, but later transformed in proton–proton distance restraints. 5. Windows that are too narrow induce potentially incomplete assignments, while large window sizes lead to highly ambiguous initial assignments, which are often the source of severe convergence issues during the ARIA protocol. Therefore, window size must be chosen carefully; the ideal situation is reached when the windows size is sufficiently large to contain the correct assignments, but without unduly increasing the number of assignment possibilities. Typical window size values for NOESY spectra are 0.02 and 0.04 ppm for the direct and indirect proton dimensions, respectively, and 0.5 ppm for the heteronuclear dimensions. The maximum number of assignment possibilities (max_n) also affects the quality of the initial assignment, since some peaks that could correctly be assigned are rejected due to an excessively large number of assignment possibilities. Fossi et al. have developed a strategy, based on a precalculation analysis, for choosing optimal values for d and and max_n for a particular data set (58). The size of the windows is directly linked to the line-width of the spectra. Thus, for MAS solid-state NMR experiments, line broadening would require larger assignment windows. From the literature, typical values for proton-driven spin diffusion experiments or protonmediated rare-spin correlation experiments are in the range of 0.25–0.6 ppm. 6. Atoms with missing resonance assignments will not be assigned to any cross-peak. In this case, automatically generated assignments are almost certainly wrong. From our experience, to achieve reasonable convergence, the completeness of a chemical shift list should not be less than 90%. 7. In addition to the standard ambiguity arising from chemical shift degeneracy, symmetry degeneracy leads to a larger number of assignment possibilities. 8. Different methods can be used to estimate secondary structures. For instance, CSI (59), TALOS (60) or DANGLE (61) predict likely values of phi/psi main-chain dihedral angles from a list of chemical shift assignments. Such predictions can be incorporated as dihedral angle restraints using an harmonic square-well potential.

478

B. Bardiaux et al.

9. The theoretical cross-peak volume is then calculated as an r −6 average over all pairwise contributions: VIJ = CnI n J dˆIJ−6 where dˆIJ−6 =

1 N IN J

∑d

I×J

−6 ij

(16)

and where I and J denote two groups of spins having nI and nJ members, respectively. Introduction of the effective distance dˆIJ retains the functional form of Eq. 4. Equation 16 relies on a discrete slow jump model where spins I and J jump between NI and NJ equilibrium sites, respectively (24). 10. For solid-state NMR data, approximation is more severe. Because of additional effects that influence the relation between peak intensity and the actual distance (dipolar truncation, partial mobility, transfer efficiency), the calibration routine implemented in ARIA may not be adapted to correctly model the cross-peak signals. However, the use of fixed distance bounds has been shown to be sufficient in numerous solid-state NMR studies. In fact, the calibration is less important since the essential feature of the ambiguous distance restraint remains valid: if at least one of the assignment possibilities is smaller than the upper limit, the restraint is satisfied. Bounds can be estimated, for instance, from buildup curves. We recommend consulting the following references for details (7, 9, 55, 62, 63). 11. The packing restraint intends to compensate for lack of unambiguous intermonomer restraints in early ARIA iterations. If convergence is achieved and a sufficient number of meaningful intermonomer cross-peaks have been assigned, we advise not to use this restraint. 12. Restraints discarded by the merging procedure are excluded from the list. 13. To use the CCPNmr FormatConverter (17) for data conversion with file formats not natively supported by ARIA, it is necessary to use the following command aria2 --convert_ccpn conversion.xml 14. If solid-state NMR experiments are performed on site-directed 13 C-enriched samples (7, 64), it is necessary to specify the appropriate labeling scheme, i.e., [1,3-13C]-glycerol and [2-13C]-glycerol. ARIA automatically removes assignment options that are not permitted by the labeling pattern, as first described in the SOLARIA program (9). Alternatively, CCPNmr Analysis provides routines to create ambiguous distance restraints respecting the labeling patterns. Such restraints can be then imported into ARIA.

23

ARIA for Solution and Solid-State NMR

479

15. A user can also choose the “New” item in the GUI menu “Project” to create a new project. As an alternative, the following command aria2 --project_template project.xml will create a new project file. 16. Residual dipolar coupling data can be incorporated as restraints following two alternative approaches: direct (SANI) or indirect (VEAN). For SANI, the user has to specify the rhombicity and magnitude of the alignment tensor (65). Several methods exist to predict these parameters, from the distribution of the RDC values (66) or from the shape of the molecule (67). VEAN uses intervector projection angle restraints which must be generated with a separate program (68). 17. The correlation between a three-bond measured J-coupling and the corresponding dihedral angle is modeled by the Karplus curve. Default values for the parameters of the Karplus curve are given for 3J(HNHa). 18. An MTF can be specified in the project file. Changes must be also made to the CNS topology, linkage, and parameter files. Definitions of the additional residues or compounds must be added to the ARIA dictionary (files atomnames.xml and iupac.xml). A detailed explanation is given on the ARIA Web site. 19. We recommend the use of the network-anchoring only for the first 3 iterations. Too stringent thresholds or an application of network-anchoring during more ARIA iterations may bias the assignment process toward an incorrect structure (13). 20. Jobs can be submitted via ssh commands or with the following batch queuing systems: PBS (69), SGE (70) or Condor (71). Alternatively, CCPN users can submit their ARIA calculation to the CCPNGrid portal server at http://www.webapps. ccpn.ac.uk/ccpngrid/. 21. Only local copies of data files are used for structure calculation. Changes in the original files will thus become active only in the next project setup. 22. For systems of about 100 residues, well converged ensembles show average energies of the order of 1,000 kcal/mol. Normal energy variation is about 10%, the total average energy scaling is approximately linear with the system size. 23. Others methods are available to estimate the credibility of the structures, notably by scrutinizing the information content of the data (72). For instance, the completeness (73) of a restraint set provides insight into the local reliability of each structure. The completeness is the ratio between the number of observed restraints and the number of expected restraints. We recommend the method AQUA (73) to perform such analysis.

480

B. Bardiaux et al.

Moreover, several Web servers exist where a user can submit structures for quality checking and validation, e.g., PSVS (74) and Cing (75). 24. Comparing such quality profiles can be very helpful to detect reliable solutions when multiple conformations are obtained (13). 25. A recent study on the effect of the cooling rate of the simulatedannealing with highly ambiguous data reported an increased efficiency of slower cooling, e.g., 100,000 (equivalent Cartesian) steps (28). The same order of value was successfully used to determine the structure of the SH3 domain (9), Crh (10), and aB crystallin dimer from MAS solid-state NMR data (76). Note that ARIA divides the number of steps for the torsion angle phase by the value of the parameter TAD time-steps factor to allow a larger time-step (default factor value is 9).

Acknowledgments This work was supported by the EU grants SPINE (QLG2CT-2002-00988) and ExtendNMR (LSHG-CT- 2005–018988). The Ministère de l’Enseignement Supérieur (ACI IMPBio, project ICMD-RMN) and Institut Pasteur are also acknowledged for financial support. The authors would like to thank Wolfgang Rieping, Michael Habeck, Aymeric Bernard, and the CCPN team for their active participation in the development of ARIA, as well as Anja Böckmann and Barth-Jan van Rossum for fruitful collaborations on solid-state NMR. Benjamin Bardiaux thanks Hartmut Oschkinat for support. References 1. Wuthrich, K. (1986) NMR of Proteins and Nucleic Acids, Wiley-Interscience New York. 2. Nilges, M. (1995) Calculation of protein structures with ambiguous distance restraints. Automated assignment of ambiguous NOE crosspeaks and disulphide connectivities. J. Mol. Biol. 245, 645–660. 3. Nilges, M. and O’Donoghue, S. I. (1998) Ambiguous NOEs and automated NOESY assignment. Prog. NMR Spec. 32, 107–139. 4. Linge, J. P., O’Donoghue, S. I., and Nilges, M. (2001) Automated assignment of ambiguous nuclear overhauser effects with ARIA. Methods Enzymol. 339, 71–90. 5. Linge, J. P., Habeck, M., Rieping, W., and Nilges, M. (2003) ARIA: automated NOE assignment and NMR structure calculation. Bioinformatics 19, 315–316.

6. Rieping, W., Habeck, M., Bardiaux, B., Bernard, A., Malliavin, T., and Nilges, M. (2007) ARIA2: automated NOE assignment and data integration in NMR structure calculation. Bioinformatics 23, 381–382. 7. Castellani, F., van Rossum, B., Diehl, A., Schubert, M., Rehbein, K., and Oschkinat, H. (2002) Structure of a protein determined by solid-state magic-angle-spinning NMR spectroscopy. Nature 420, 98–102. 8. Herrmann, T., Güntert, P., and Wüthrich, K. (2002) Protein NMR structure determination with automated NOE assignment using the new software CANDID and the torsion angle dynamics algorithm DYANA. J. Mol. Biol. 319, 209–227. 9. Fossi, M., Castellani, F., Nilges, M., Oschkinat, H., and van Rossum, B. (2005) SOLARIA: a

23

10.

11.

12.

13.

14.

15. 16.

17.

18.

19.

20.

21.

protocol for automated cross-peak assignment and structure calculation for solid-state magicangle spinning NMR spectroscopy. Angew. Chem. Int. Ed. Engl. 44, 6151–6154. Loquet, A., Bardiaux, B., Gardiennet, C., Blanchet, C., Baldus, M., Nilges, M., Malliavin, T., and Böckmann, A. (2008) 3D Structure Determination of the Crh Protein from Highly Ambiguous Solid-State NMR Restraints. J. Am. Chem. Soc. 130, 3579–3589. Manolikas, T., Herrmann, T., and Meier, B. (2008) Protein structure determination from (13)C spin-diffusion solid-state NMR spectroscopy. J. Am. Chem. Soc. 130, 3959–3966. Wasmer, C., Lange, A., Melckebeke, H. V., Siemer, A., Riek, R., and Meier, B. (2008) Amyloid fibrils of the HET-s(218–289) prion form a beta solenoid with a triangular hydrophobic core. Science 319, 1523–1526. Bardiaux, B., Bernard, A., Rieping, W., Habeck, M., Malliavin, T. E., and Nilges, M. (2009) Influence of different assignment conditions on the determination of symmetric homodimeric structures with ARIA. Proteins 75, 569–585. Nilges, M., Bernard, A., Bardiaux, B., Malliavin, T., Habeck, M., and Rieping, W. (2008) Accurate NMR structures through minimisation of an extended hybrid energy. Structure 16, 1305–1312. van Rossum, G., http://www.python.org/. Brünger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J.-S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Crystallography and NMR system (CNS): A new software suite for macromolecular structure determination. Acta Cryst. sect. D 54, 905–921. Vranken, W. F., Boucher, W., Stevens, T. J., Fogh, R. H., Pajon, A., Llinas, M., Ulrich, E. L., Markley, J. L., Ionides, J., and Laue, E. D. (2005) The CCPN data model for NMR spectroscopy: development of a software pipeline. Proteins 59, 687–696. Laskowski, R. A., MacArthur, M. W., Moss, D. S., and Thornton, J. M. (1993) PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Cryst. 26, 283–291. Vriend, G. (1990) WHAT IF: a molecular modeling and drug design program. J. Mol. Graph. 8, 52–56. Sippl, M. J. (1993) Recognition of errors in three-dimensional structures of proteins. Proteins Struct. Funct. Genet. 17, 355–362. Davis, I. W., Leaver-Fay, A., Chen, V. B., Block, J. N., Kapral, G. J., Wang, X., Murray, L. W.,

ARIA for Solution and Solid-State NMR

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

481

Arendall, W. B., Snoeyink, J., Richardson, J. S., and Richardson, D. C. (2007) MolProbity: allatom contacts and structure validation for proteins and nucleic acids. Nucleic Acids Res. 35, W375–383. Folmer, R. H., Hilbers, C. W., Konings, R. N., and Nilges, M. (1997) Floating stereospecific assignment revisited: application to an 18 kDa protein and comparison with J-coupling data. J. Biomol. NMR 9, 245–258. Duggan, B., Legge, G., Dyson, H., and Wright, P. (2001) SANE (Structure Assisted NOE Evaluation): an automated model-based approach for NOE assignment. J. Biomol. NMR 19, 321–329. Görler, A. and Kalbitzer, H. R. (1997) Relax, a flexible program for the back calculation of NOESY spectra based on complete relaxation matrix formalism. J. Magn. Reson. 124, 177–188. Linge, J., Habeck, M., Rieping, W., and Nilges, M. (2004) Correction of spin diffusion during iterative automated NOE assignment. J. Magn. Reson. 167, 334–342. Mumenthaler, C. and Braun, W. (1995) Automated assignment of simulated and experimental NOESY spectra of proteins by feedback filtering and self-correcting distance geometry. J. Mol. Biol. 254, 465–480. Stein, E. G., Rice, L. M., and Brünger, A. T. (1997) Torsion-angle molecular dynamics as a new efficient tool for NMR structure calculation. J. Magn. Reson. 124, 154–164. Fossi, M., Oschkinat, F., Nilges, M., and Ball, L. (2005) Quantitative study of the effects of chemical shift tolerances and rates of SA cooling on structure calculation from automatically assigned NOE data. J. Magn. Reson. 175, 92–102. Rieping, W., Habeck, M., and Nilges, M. (2005) Modeling errors in NOE data with a log-normal distribution improves the quality of NMR structures. J. Am. Chem. Soc. 127, 16026–16027. Rieping, W., Habeck, M., and Nilges, M. (2005) Inferential Structure Determination. Science 309, 303–306. Habeck, M., Rieping, W., and Nilges, M. (2006) Weighting of experimental evidence in macromolecular structure determination. Proc. Natl. Acad. Sci. USA 103, 1756–1761. Nilges, M. (1993) A calculation strategy for the structure determination of symmetric dimers by 1 H NMR. Proteins 17, 297–309. Linge, J. P., Williams, M. A., Spronk, C. A., Bonvin, A. M., and Nilges, M. (2003) Refinement of protein structures in explicit solvent. Proteins Struct. Funct. Genet. 20, 496–506.

482

B. Bardiaux et al.

34. Linge, J. P. and Nilges, M. (1999) Influence of non-bonded parameters on the quality of NMR structures: a new force-field for NMR structure calculation. J. Biomol. NMR 13, 51–59. 35. Nederveen, A., Doreleijers, J., Vranken, W., Miller, Z., Spronk, C., Nabuurs, S., Guntert, P., Livny, M., Markley, J., Nilges, M., Ulrich, E., Kaptein, R., and Bonvin, A. M. (2005) RECOORD: a REcalculated COORdinates Database of 500+ proteins from the PDB using restraints from the BioMagResBank. Proteins 59, 662–672. 36. The World Wide Web Consortium (2008), Extensible Markup Language (XML) 1.0 (Fifth Edition), http://www.w3.org/TR/xml/. 37. Bardiaux, B., Bernard, A., Rieping, W., Habeck, M., Malliavin, T., and Nilges, M. (2008) Graphical analysis of NMR structural quality and interactive contact map of NOE assignments in ARIA. BMC Struct. Biol. 8, 30–34. 38. Spronk, C. A. E. M., Nabuurs, S. B., Krieger, E., Vriend, G., and Vuister, G.W. (2004) Validation of protein structures derived by NMR spectroscopy. Progress in Nuclear Magnetic Resonance Spectroscopy 45, 315–337. 39. Saccenti, E. and Rosato, A. (2008) The war of tools: how can NMR spectroscopists detect errors in their structures? J. Biomol. NMR 40, 251–261. 40. Nabuurs, S., Krieger, E., Spronk, C., Nederveen, A., Vriend, G., and Vuister, G. (2005) Definition of a new information-based per-residue quality parameter. J. Biomol. NMR 33, 123–134. 41. Nabuurs, S., Spronk, C., Vuister, G., and Vriend, G. (2006) Traditional biomolecular structure determination by NMR spectroscopy allows for major errors. PLoS Comput. Biol. 2, e9. 42. Kraulis, P., Domaille, P. J., Campbell-Burk, S. L., van Aken, T., and Laue, E. D. (1994) Solution structure and dynamics of ras p21. GDP determined by heteronuclear three- and four-dimensional NMR spectroscopy. Biochemistry 33, 3515–3531. 43. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J., and Bax, A. (1995) NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293. 44. Johnson, B. A. and Blevins, R. A. (1994) NMRView: A computer program for the visualization and analysis of NMR data. J. Biomol. NMR 4, 603–614. 45. Garrett, D., Powers, R., Gronenborn, A., and Clore, G. (1991) A common sense approach to peak picking two-, three- and four-dimensional spectra using automatic computer analysis of

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

contour diagrams., J. Magn. Reson. 95, 214–220. Kjær, M., Andersen, K. V., and Poulsen, F. M. (1994) Automated and semiautomated analysis of homo- and heteronuclear multidimensional nuclear magnetic resonance spectra of proteins: the program PRONTO. Methods Enzymol. 239, 288–308. Bartels, C., Xia, T.-H., Billeter, M., Güntert, P., and Wüthrich, K. (1995) The program XEASY for computer-supported NMR spectral analysis of biological macromolecules. J. Biomol. NMR 5, 1–10. Güntert, P., Braun, W., and Wüthrich, K. (1991) Efficient computation of three-dimensional protein structures in solution from nuclear magnetic resonance data using the program DIANA and the supporting programs CALIBA, HABAS and GLOMSA. J. Mol. Biol. 217, 517–530. Hall, S. R. and Cook, A. P. F. (1995) STAR dictionary definition language: Initial specification. J. Chem. Inf. Comput. Sci. 35, 819–825. Markley, J. L., Bax, A., Arata, Y., Hilbers, C. W., Kaptein, R., Sykes, B. D., Wright, P. E., and Wüthrich, K. (1998) Recommendations for the presentation of NMR structures of proteins and nucleic acids. J. Mol. Biol. 280, 933–952. Güntert, P., Mumenthaler, C., and Wütrich, K. (1997) Torsion Angle Dynamics for NMR Strucutre Calculation with the New Program DYANA. J. Mol. Biol. 273, 283–298. Wüthrich, K., Billeter, M., and Braun, W. (1983) Pseudo-structures for the 20 common amino acids for use in studies of protein conformations by measurements of intramolecular proton-proton distance constraints with nuclear magnetic resonance. J Mol Biol 169, 949–961. Lange, A., Luca, S., and Baldus, M. (2002) Structural constraints from proton-mediated rare-spin correlation spectroscopy in rotating solids. J. Am. Chem. Soc. 124, 9704–9705. Szeverenyi, N., Sullivan, M., and Maciel, G. (1982) Observation of spin exchange by twodimensional fourier transform 13 C cross polarization-magic-angle spinning. J. Magn. Reson. 47, 462–475. Castellani, F., van Rossum, B., Diehl, A., Rehbein, K., and Oschkinat, H. (2003) Determination of solid-state NMR structures of proteins by means of three-dimensional 15 N-13 C-13 C dipolar correlation spectroscopy and chemical shift analysis. Biochemistry 42, 11476–11483. Takegoshi, K., Nakamura, S., and Terao, T. (2003) 13 C-1 H dipolar-driven 13 C-13 C recoupling without 13 C rf irradiation in nuclear

23

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

magnetic resonance of rotating solids. J. Chem. Phys. 118, 2325–2341. Lewandowski, J. R., Paëpe, G. D., Eddy, M. T., and Griffin, R. G. (2009) (15)N-(15)N proton assisted recoupling in magic angle spinning NMR. J. Am. Chem. Soc. 131, 5769–5776. Fossi, M., Linge, J., Labudde, D., Leitner, D., Nilges, M., and Oschkinat, H. (2005) Influence of chemical shift tolerances on NMR structure calculations using ARIA protocols for assigning NOE data. J. Biomol. NMR 31, 21–34. Wishart, D. S. and Sykes, B. D. (1994) The 13 C chemical-shift index: a simple method for the identification of protein secondary structure using 13 C chemical-shift data. J. Biomol. NMR 4, 171–180. Cornilescu, G., Delaglio, F., and Bax, A. (1999) Protein backbone angle restraints from searching a database for chemical shift and sequence homology. J. Biomol. NMR 13, 289–302. Cheung, M.-S., Maguire, M. L., Stevens, T. J., and Broadhurst, R. W. (2010) DANGLE: A Bayesian inferential method for predicting protein backbone dihedral angles and secondary structure. J. Magn. Reson. 202, 223–33. Loquet, A., Gardiennet, C., and Böckmann, A. (2010) Protein 3D structure determination by high-resolution solid-state NMR. Comptes. Rendus - Chimie 13, 423–430. Gardiennet, C., Loquet, A., Etzkorn, M., Heise, H., Baldus, M., and Böckmann, A. (2008) Structural constraints for the Crh protein from solid-state NMR experiments. J. Biomol. NMR. 40, 239–250. LeMaster, D. M. and Kushlan, D. M. (1996) Dynamical mapping of E. coli thioredoxin via 13 C NMR relaxation analysis. J. Am. Chem. Soc. 118, 9255–9264. Tjandra, N., Garrett, D. S., Gronenborn, A. M., Bax, A., and Clore, G. M. (1997) Defining long range order in NMR structure determination from the dependence of heteronuclear relaxation times on rotational diffusion anisotropy. Nature Struct. Biol. 4, 443–449. Clore, G., Gronenborn, A., and Bax, A. (1998) A robust method for determining the magnitude of the fully asymmetric alignment tensor of oriented macromolecules in the absence of structural information. J. Magn. Reson. 133, 216–221.

ARIA for Solution and Solid-State NMR

483

67. Zweckstetter, M. and Bax, A. (2000) Prediction of sterically induced alignment in a dilute liquid crystalline phase: Aid to protein structure determination by NMR. J. Am. Chem. Soc. 122, 3791–3792. 68. Meiler, J., Blomberg, N., Nilges, M., and Griesinger, C. (2000) A new approach for applying residual dipolar couplings as restraints in structure calculations. J. Biomol. NMR 16, 245–252. 69. Jones, J. P. (2002) PBS: portable batch system, Beowulf cluster computing with Linux, MIT Press, Cambridge, MA, USA, 369–390. 70. Gentzsch, W. (2001) Sun Grid Engine: Towards creating a compute power grid, CCGRID ’01: Proceedings of the 1st International Symposium on Cluster Computing and the Grid, IEEE Computer Society, Washington, DC, USA, 35. 71. Thain, D., Tannenbaum, T., and Livny, M. (2005) Distributed computing in practice: the Condor experience. Concurr. Comput.: Pract. Exper. 17, 323–356. 72. Nabuurs, S., Spronk, C., Krieger, E., Maassen, H., Vriend, G., and Vuister, G. (2003) Quantitative evaluation of experimental NMR restraints. J. Am. Chem. Soc. 125, 12026–12034. 73. Doreleijers, J. F., Raves, M. L., Rullmann, T., and Kaptein, R. (1999) Completeness of NOEs in protein structure: a statistical analysis of NMR data. J. Biomol. NMR 14, 123–132. 74. Bhattacharya, A., Tejero, R., and Montelione, G. T. (2007) Evaluating protein structures determined by structural genomics consortia. Proteins 66, 778–795. 75. Doreleijers, J. F., Vranken, W. F., Schulte, C., Lin, J., Wedell, J. R., Penkett, C. J., Vuister, G. W., Vriend, G., Markley, J. L., and Ulrich, E. L. (2009) The NMR restraints grid at BMRB for 5,266 protein and nucleic acid PDB entries. J. Biomol. NMR 45, 389–396. 76. Jehle, S., Rajagopal, P., Bardiaux, B., Markovic, S., Kühne, R., Stout, J. R., Higman, V. A., Klevit, R. E., van Rossum, B.-J., and Oschkinat, H. (2010) Solid-state NMR and SAXS studies provide a structural basis for the activation of alphaB-crystallin oligomers. Nat. Struct. Mol. Biol. 17, 1037–1042.

Chapter 24 Determining Protein Dynamics from 15N Relaxation Data by Using DYNAMICS David Fushman Abstract Motions are essential for protein function, and knowledge of protein dynamics is a key to our understanding the mechanisms underlying protein folding and stability, ligand recognition, allostery, and catalysis. In the last two decades, NMR relaxation measurements have become a powerful tool for characterizing backbone and side chain dynamics in complex biological macromolecules such as proteins and nucleic acids. Accurate analysis of the experimental data in terms of motional parameters is an essential prerequisite for developing physical models of motions to paint an adequate picture of protein dynamics. Here, I describe in detail how to use the software package DYNAMICS that was developed for accurate characterization of the overall tumbling and local dynamics in a protein from nuclear spin-relaxation rates measured by NMR. Step-bystep instructions are provided and illustrated through an analysis of 15N relaxation data for protein G. Key words: Relaxation, Protein dynamics, Order parameter, Spectral density, Dipolar coupling, Chemical shift anisotropy, CSA, Overall tumbling, Rotational diffusion tensor, Monomer–dimer equilibrium

1. Introduction Proteins are molecular nanomachines. Understanding of how they work requires detailed knowledge of not only their three-dimensional structure but also of various motions that take place in a protein and the roles they play in protein’s folding and stability, ligand recognition, allostery, and catalysis. NMR is perhaps the most powerful analytical tool in structural biology, because it is capable of providing site-specific information on the structure, dynamics, and electronic environment of essentially any nucleus in a molecule, even as complex as a protein or nucleic acid. Moreover, solution NMR methods allow studies of molecules in their native milieu, the reporter groups used do not cause any structural perturbations, and the applied magnetic fields are still so weak

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3_24, © Springer Science+Business Media, LLC 2012

485

486

D. Fushman

compared to the thermal energy that they do not affect molecular structure or dynamics. Recent decades witnessed a burst in NMR studies of protein motions (reviewed in ref. 1). A popular reporter group for protein dynamics studies has been the N–H group, for several reasons, most importantly: 1H–15N is an isolated spin-pair (to a good approximation), conveniently located in the backbone and abundant in proteins which, combined with the oftentimes good spread of NMR signals in the 1H–15N correlation spectra, allows almost complete coverage of a protein sequence (except prolines), and 15N enrichment is relatively easy and affordable. There is growing interest in developing and extending the NMR methodology (e.g., see ref. 2, 3) to understand motions in other groups in a protein, both in the backbone (CO, C–Ca, CaHa) and in the side chains (e.g., methyl groups). The underlying concept of NMR being a sensitive tool for accessing equilibrium protein dynamics is that nuclear spin relaxation is caused by modulation (by the overall and internal motions) of the magnetic field sensed by a nucleus under observation. This field is a result of magnetic interactions (dipolar and scalar) with the magnetic moments of surrounding nuclei and of perturbations in the electronic environment of the nucleus, resulting in the shielding effect. All of these mechanisms contribute to nuclear-spin relaxation, albeit to a different extent. Modulation of dipolar coupling can be caused by reorientation of the internuclear vector as well as variation in its length. For a bonded pair of atoms, bond vibrations are usually very fast (on the NMR time scale) and therefore do not contribute to nuclear-spin relaxation rates directly (except for an altered effective bond length). Likewise, the effect of motions on the chemical shift (shielding) tensor could be envisioned as a reorientation of the tensor, as well as a modulation of its principal components. The former contributes in a similar way as reorientation of the dipolar coupling while the latter could lead to the so-called conformational (or chemical) exchange contributions to relaxation. Note that unlike the dipolar coupling, the shielding tensor generally is not axially symmetric; however, a fully anisotropic tensor can always be presented as a sum of two axially symmetric tensors, and for each of them the above statement applies. I do not describe in this chapter how relaxation rates are measured; this information can be found in various sources, e.g., see ref. 4. Instead, here I focus on how these rates can be analyzed. 1.1. The Underlying Equations

The experimental spin relaxation parameters (longitudinal and transverse relaxation rates, R1 and R2, and the steady-state heteronuclear NOE) are directly related to power spectral densities, J(w), which are Fourier transforms of the corresponding correlation functions describing reorientations of the internuclear vector of interest. In the case of the backbone amide 15N nucleus, the major

24

Determining Protein Dynamics from 15N Relaxation Data…

487

sources of 15N spin relaxation are modulations by motions of (1) the dipole–dipole interaction (=dipolar coupling) of the nuclear magnetic moment of 15N with that of the directly bonded 1H, and (2) the anisotropy of the 15N chemical shift tensor (CSA). The standard equations read as follows: R1 = 3(d 2 + c 2 ) J (w N ) + d 2 [ J (w H − w N ) + 6 J (w H + w N )]

(1)

R2 = 12 (d 2 + c 2 )[4 J (0) + 3 J (w N )] + 12 d 2 [ J (w H − w N )

(2)

+ 6 J (w H ) + 6 J (w H + w N )] + Rex NOE = 1 − g H / g N d 2 ⎡⎣6 J (w H + w N ) − J (w H − w N )⎤⎦ / R1

(3)

Here d = − (mo/(4p))gHgNh/(4prHN3) is the strength of the 15 N–1H dipolar coupling, c = −wN·CSA/3, wH and wN are the resonance frequencies of 1H and 15N, respectively, and Rex is the conformational exchange contribution (if any) to measured R2. These equations assume that the effects of reorientational motion on the 1 H–15N dipolar interaction and on the 15N CSA can be described by the same autocorrelation function. Corrections to the above equations that account for noncollinearity of these two interactions are discussed in (5). Equations 1–3 provide the basis for extracting information on protein dynamics from NMR relaxation measurements. Given the experimental data, the primary objectives here are (1) to determine the spectral densities J(w) and, most importantly, (2) to translate them into an adequate physical picture of protein dynamics. As it is generally impossible to determine all parameters of complex motions from a limited set of measurements, the latter objective requires adequate theoretical models of motion that can be obtained from comparison with molecular dynamics simulations (e.g., see ref. 6–8). Nevertheless, accurate analysis of experimental data (objective (1)) is an essential prerequisite for such a comparison. Extracting the spectral densities directly from Eqs. 1–3 is problematic because this system of equations in underdetermined: the number of unknowns (d, c, J(w)’s, and possibly Rex) exceeds the number of available experimental data (e.g., see ref. 9). A widely accepted way to circumvent this problem, the so-called model-free approach (10, 11), is based on a rather simple parameterization of the spectral density function by approximating the correlation function describing local dynamics as monoexponential ⎛ −t ⎞ C loc (t ) = S 2 + (1 − S 2 )exp ⎜ . ⎝ t loc ⎟⎠

(4)

In this parameterization, tloc has the meaning of the correlation time of the bond’s motion, and the angular amplitude of bond reorientations is characterized by the so-called squared order

488

D. Fushman

parameter, S 2, a dimensionless measure of the amplitude on a scale from 0 to 1: S 2 = 0 for unrestricted bond motions, while S 2 = 1 when this motion is completely restricted. Assuming that the strength of the dipolar coupling (d) and the 15N CSA term (c) are known, this leaves only two fitting parameters (three if Rex is present), S 2 and tloc, to be determined for each residue, since the overall tumbling of the molecule is described by a small number of global parameters (see below). It was found however, that analysis of 15N relaxation data in proteins sometimes requires a more complex, dual-exponential parameterization (the so-called extended model-free model (12)): ⎛ −t ⎞ ⎛ −t ⎞ 2 2 C loc (t ) = S 2 + (S fast − S 2 )exp ⎜ + (1 − S fast )exp ⎜ . (5) ⎟ ⎝ t slow ⎠ ⎝ t fast ⎟⎠ This correlation function represents a superposition of two independent motions, “fast” and “slow,” characterized by the corresponding order parameters (Sfast2 and Sslow2) and correlation times (tfast and tslow) and occurring on entirely separated time scales: tfast < < tslow. Note that for consistency with Eq. 4, here I introduced the generalized order parameter S2 that represents the total amplitude of the combined motion: S2 = Sslow2 Sfast2. In this chapter, I refer to the modelfree characteristics of local motion (S2, tloc or Sfast2, tfast, S 2, tslow) as well as Rex as microdynamic parameters. Various parameterizations of the correlation function Cloc(t) are referred to as models of local motion. If there is no correlation between the local dynamics and the overall rotational diffusion of a molecule, as assumed in the modelfree approach, the total correlation function that determines J(w) and hence the rates of 15N relaxation Eqs. 1–3 can be written in the following form: C (t ) = C ovrl (t )C loc (t ),

(6)

where Covrl(t) is the autocorrelation function describing the overall tumbling of a rigid molecule characterized by the (generally anisotropic) rotational diffusion tensor D (13, 14). In the simplest case of isotropic overall tumbling with a correlation time tc, 1 C ovrl (t ) = e −t /t c . In the case of rotational anisotropy, the expres5 sions for Covrl(t) are more complex and depend on the bond’s orientation with respect to the diffusion tensor frame (e.g., see ref. 15). The corresponding equations can be found, for example, in (13, 14, 16). Recall that Fourier transforms of C(t) give the power spectral densities J(w) in Eqs. 1–3. For example, in the case of isotropic rotation diffusion, combining Eqs. 4 and 6 gives J (w ) = S 2 j (w , t c ) + (1 − S 2 ) j (w , t e ), where j (w , t ) =

2 t and 1 / t e = 1 / t c + 1 / t loc. 5 1 + (wt )2

(7)

24

Determining Protein Dynamics from 15N Relaxation Data…

489

Numerous studies in the last two decades (by NMR relaxation, as well as MD simulations, (e.g., see ref. 6, 7)) revealed quite restricted backbone motions in well-ordered regions (secondary structure) in proteins, with the amplitudes of S2 ~ 0.87, whereas significantly lower S2 (reflecting greater amplitudes) are often observed in the flexible unstructured regions, such as loops and termini. Likewise, the associated time scales (tloc or tfast) are in the 1–100 ps range for protein-core elements and slower, up to several nanoseconds (tslow) in the flexible parts, possibly reflecting concerted motion of several residues or segments. It should be pointed out that the information about protein motions is limited by the time window imposed by the overall tumbling. Being unrestricted, the overall tumbling eventually averages both the dipolar- the CSA-related energies to zero such that there is essentially nothing left for slower motions to modulate, except for the principal values of the dipolar or chemical shift tensors. The latter modulation manifests itself in the so-called chemical or conformational exchange processes, which provide access to motions slower than tc and have been studied quite extensively recently (e.g., see ref. 17). In this chapter, I describe the use of computer program DYNAMICS (18, 19) designed to extract parameters characterizing protein motions from NMR-measured spin-relaxation parameters. For convenience, kplot to indicate the names of the parameters in DYNAMICS. Courier font is used to indicate Matlab commands and screen output messages; the lines containing Matlab commands throughout this chapter begin with the Matlab prompt (>>).

2. Description of DYNAMICS Software 2.1. Highlights of the Program DYNAMICS

DYNAMICS is a computer program for model-free analysis of spin relaxation data. The current version of the program (version 3.0) includes the following features: ●

Overall tumbling. All possible models of the overall rotational diffusion are allowed: isotropic, axially symmetric, and fully anisotropic. The overall rotational diffusion tensor can be an input variable, but can also be determined simultaneously with the model-free analysis of the relaxation data.

●

Multiple-field data. Simultaneous or separate analysis of experimental data from measurements at multiple magnetic fields. Data sets at various fields do not have to be complete.

●

Chemical shift anisotropy. The CSA can be treated as uniform (fixed) or site-specific, the program also allows determining site-specific CSA values simultaneously with data analysis.

490

D. Fushman ●

2.2. The Overall Organization of the Program

Monomer–dimer equilibrium. The program allows data analysis when the molecule of interest exists in a fast dynamic equilibrium between monomeric and dimer states.

The overall organization of DYNAMICS is depicted in the flowchart in Fig. 1. The program is written in Matlab (The MathWorks, Inc); the current version of the program is compatible with Matlab versions 6.5 and newer. It is assumed here that the user is familiar with very basic Matlab commands that allow loading and saving data and navigation to the desired folder/directory.

Fig. 1. Flowchart of the program DYNAMICS.

24

Determining Protein Dynamics from 15N Relaxation Data… 4

j(ω,τc), τc=10 ns j(ω,τc), τc=5 ns j(ω,τe), τc=10 ns, τloc=100 ps

3 j(ω,τ), a.u.

491

0.2

2 0.1

1

0.0 100

900

0 1

10

100

900

ω/2π, MHz

Fig. 2. Relative contributions to the power spectral density J (w) from the overall tumbling and local motion. Shown as a function of frequency w are j ( w,tc) (see Eq. 7) for tc = 5 ns (green) and 10 ns (blue), and j (w,te) for tloc = 100 ps and tc =10 ns (red ). The factors S 2 and (1−S 2) are not included.

2.3. Treating the Overall Rotational Diffusion of a Molecule

As shown in Fig. 2, the contribution to the spectral density function from the overall tumbling is quite substantial (if not dominant) and often overshadows that from local motions. As our main goal here is to characterize internal motions, accurate treatment of the overall tumbling is absolutely critical for accurate analysis of the local dynamics in a protein (20). Thus, the first and foremost step in relaxation data analysis is to determine, and “subtract,” the contribution from the overall tumbling. Significant attention in the past was paid to developing tools for accurate analysis of the overall rotational diffusion (16, 21–28). In principle, the overall rotational diffusion can also be characterized simultaneously with the analysis of local dynamics, and in fact, DYNAMICS includes a mechanism for doing this (see Subheadings 3.2 and 3.3). However, beyond the simplest case of isotropic tumbling, this determination becomes less straightforward and can require significant effort, as multiple parameters need to be optimized manually,. Therefore, if protein atom coordinates are available, the most straightforward and reliable way to characterize the overall rotational diffusion is directly from relaxation data and separately from (and prior to) the analysis of local motions. The underlying reason for this is based on the fact that, for welldefined structural regions in a protein, the “reduced” relaxation rates R1¢ and R2¢ are both proportional (to a good approximation) to the squared order parameter. (The “reduction” is achieved by subtracting from Eqs. 1–2 the contributions from the high-frequency components, J(wH) and J(wH ± wN), of the spectral density function, e.g., see ref. 27). Thus the R2¢/R1¢ ratio is S2-independent, and the determination of the overall motion can be de-convoluted

492

D. Fushman

(hence performed separately) from the analysis of local protein dynamics (22, 26, 27, 29). Moreover, as discussed in (26, 27, 29, 30), the R2¢/R1¢ ratio is independent of site-specific variations in the actual values of d and c, Eqs. 1–2, and therefore depends solely on the structure of a protein (i.e., orientations of the NH bonds with respect to the diffusion tensor axes) and on the diffusion tensor itself (22). This concept is implemented in the computer program RotDif (26) available online from our Web site: http://www.gandalf. umd.edu/FushmanLab/. The use of this program is illustrated in (26–28); therefore, we do not describe these steps in detail here. Briefly, the RotDif program uses NH-vector coordinates and relaxation rates, R1, R2, as well as NOE (if available) as input parameters, and outputs the principal components of the diffusion tensor D (Dx, Dy, Dz) along with the orientation (given by the Euler angles a, b, and g) of the principal axes of the tensor with respect to the protein coordinate frame. The output also includes the overall rotational correlation time tc (=TAUc) and the anisotropy of the tensor (Dz/Dx, Dz/Dy), to be directly entered as input to DYNAMICS. For the B3 domain of protein G (GB3), used here as an example, RotDif analysis of 15N relaxation data measured at 14.1 T (600 MHz 1H frequency) yielded the following characteristics of the (axially symmetric) diffusion tensor: D|| ≡ Dz = 6.05 ± 0.44 10−7 s−1, D^ ≡ Dx = Dy =4.45 ± 0.15 10−7 s−1, a = 90° ± 8°, b = 70° ± 10°, which give tc = 3.34 ± 0.14 ns and the anisotropy D||/D^ = 1.36 ± 0.09. The results at other fields are very similar (19). 2.4. Selection of the Appropriate Model for Local Motion

At the heart of the DYNAMICS program is the model selection algorithm, which, based on how a particular model of local motion fits experimental data, selects the most appropriate model. It is similar to the approach described in (31) and is based on the Occam’s razor principle, in that the simplest model that fits the data is considered sufficient. All models of local motion used in DYNAMICS are listed in Table 1. The model selection process starts with the simplest model, LS_00, and first determines if it is acceptable, i.e., the following two criteria are satisfied: (1) the model yields physically reasonable values of the microdynamic parameters (in this case, 0 £ S2 £1, but more generally for all models: 0 £ S2, Sfast2 £ 1; tloc, tfast >5 ps; 100 ps < tslow < tc, and Rex ³ sR2) and (2) it provides a reasonable fit to the experimental data, i.e., passes the goodness-of-fit test (32). If this model is acceptable, the program proceeds to the next-level-complexity models (in this case, LS_tl and LS_ex) and applies the same acceptance rules as above. If any of these models are acceptable and yield lower residuals of fit (c2) than the lower-complexity model (in this case, LS_00), the program uses the F-statistics test to determine if this improvement in the fit is genuine and reflects a better-fit model or merely

24

Determining Protein Dynamics from 15N Relaxation Data…

493

Table 1 Microdynamic parameters for the various models of local motion used in DYNAMICS Modela

S 2 or Sslow2b

tloc or tslowb

Sfast2

tfast

Rex

#expc

Npard

Indexe

LS_00

Vf

0

1

N/A

0

1

1

0

LS_tl

V

V

1

N/A

0

1

2

1

LS_ex

V

0

1

N/A

V

1

2

2

LS_tx

V

V

1

N/A

V

1

3

3

CL_00

V

0

V

V

0

2

3

4

CL_tl

V

V

V

V

0

2

4

5

CL_ex

V

0

V

V

V

2

4

6

CL_tx

V

V

V

V

V

2

5

7

Matlab name

S2 g

TAUloc or TAUsl b

S2f

TAUf

Rex

a

The name of the corresponding model of local motion as used in DYNAMICS Naming convention used in DYNAMICS: the corresponding motional parameters in the monoexponential model are S2 and tloc, whereas in the double-exponential (“extended”) model these parameters are called Sslow2 and tslow c The number of exponentials in the corresponding correlation function of local motion see Eqs. 4 and 5 d The total number of fitting parameters in a given model e A numerical index of the model in DYNAMICS, plotted in the output graphs (see Figs. 3 and 4) f “V ” indicates that the corresponding parameter is present in a given model and is fitted (not fixed); N/A = not applied g The name of the corresponding Matlab variable in DYNAMICS output. Note that in the case of “extended” models the reported S2 value is in fact S2 = Sfast2 × Sslow2 b

reflects the greater number of fitting parameters (32). This is possible because the models being tested are nested: each higher complexity model retains the same parameters as the lower complexity model and introduces an additional parameter. If neither model provides a better fit than LS_00, the latter model is accepted, and the program moves to the next residue. If not, the program proceeds to higher complexity models and so on. The program keeps increasing the level of complexity until the number of fitting parameters (Npar) reaches the number of experimental data (Ndat) for a given residue, hence the number of degrees of freedom (df = Ndat -Npar) becomes 0. Note that when df = 0, the F-statistics test does not work. In this case we implement a simple rule: if c2 < 0.01, the model is accepted. This is somewhat arbitrary, and therefore selection of a model with df = 0 (e.g., LS_tx or CL_00 for a set of R1, R2, NOE measured at a single field) should be taken with some caution.

494

D. Fushman

It could happen that none of the models of local motion for a given residue pass the goodness-of-fit test because of a poor fit or underestimated experimental errors, resulting in c2 values higher than the acceptance level. In this case, if at least one model yielded a physically meaningful set of microdynamic parameters, the lowest-c2/df model that satisfies the latter criterion is selected, and the residue will be marked as belonging to the NOMOD category. If none of the models yield a physically meaningful solution, the residue is marked as program-excluded residue (EXCL category), and DYNAMICS will output a message: no model found, at all. 2.5. Running DYNAMICS 2.5.1. Getting Started

All the scripts and functions of DYNAMICS package come in a single compressed file. When you uncompress it (using one of the standard programs), it will by default put all the content of the package in a folder called dynamics. I recommend that you run all the analysis from the directory containing your relaxation data, which is separate from the dynamics directory: this will prevent you from “littering” the latter with output files that DYNAMICS creates automatically (see Subheading 3.5). For this, you will need to add the dynamics directory to your Matlab path, for example by using the following command: > > path(path,’c:/MyMatlab/dynamics’) (here I assumed that all DYNAMICS scripts are located in the folder c:/MyMatlab/dynamics on your computer).

2.5.2. Before You Run the Program

Navigate to your data directory and load all required input parameters (Table 2) into the Matlab workspace (use Matlab function load for this). Make sure that all of the parameters are in the proper format and units as specified in Table 2. The auxiliary program pdb2nh (see Subheading 3.6.1) will help you retrieve NH-vector coordinates from the protein coordinate file.

2.5.3. Run-Time Dialog

To start the program type the following command in the Matlab Command window: > > dynamics If you added the dynamics directory to the Matlab path, you can type this command directly from your data directory (recommended). At the start, the program performs preliminary analysis of the input data and outputs on the screen various estimates of the overall rotational correlation time and the statistics of the distribution of the R2/R1 ratios. Here is an example of such output for 15N relaxation data at 14.1 Tesla (1H frequency = 600.13 MHz) for GB3 (19). We use these data throughout this chapter. If relaxation data at more than one field are included, the analysis and the output will be done for each field separately.

Yes

Yes

Yes

Only for anisotropic diffusion models (kovrl = 1 or −1) Only for fixed site-specific CSA (kcsa = −1). If kcsa = 0, the program will ask you to input CSA manually

Array Nresb x 3: [Residue# R1c sR1c]

Array Nres x 3: [Residue# R2c sR2c]

Array Nres x 3: [Residue# NOE sNOE] Array Nres x 4: [Residue# x y zd] Array Nres x 3: [Residue# CSAe sCSAe] Scalar or vector

Scalars

H frequency, in MHz, could be several (Nfreq) frequencies, if data at multiple fields

R1 data for Nres residues at freq (i)

R2 data for Nres residues at freq (i)

NOE data for Nres residues at freq (i)

NH-vectors (normalized) for Nres residues

CSA values for Nres residues

Overall rotational correlation time, tc, in ns

Ratios of the principal values of the diffusion tensor (Dz/Dx, Dy/Dx)

r11 (if single field) or several arrays r11, r12, etc., generally: r1i where i = 1,2,…,Nfreq

r21 (if single field) or several arrays r21, r22, etc., generally: r2i where i = 1,2,…,Nfreq

r31 (if single field) or several arrays r31, r32, etc., generally: r3i where i = 1,2,…,Nfreq

vNH

csa

TAUc

Dz2Dx, Dy2Dx

Only for anisotropic diffusion models (kovrl = 1 or −1). The program will ask you to input them manually (continued)

Yes. If TAUc is missing in the workspace, the program will ask you to input it manually

Yes

Vector of length Nfreqa

1

freq

Required?

Data format/structure

The meaning

Parameter or data array name (case sensitive)

Table 2 Input parameters for DYNAMICS

Scalars or vectors

Scalars

=0 for isotropic (default) =1 for axially symmetric =−1 for fully anisotropic =2 for monomer–dimer equilibrium =0 for fixed uniform CSA (default) =−1 for fixed site-specific CSA =1 to fit site-specific CSA =0 =1 (default) =0 open new a figure (default) =−1 plot to the same figure otherwise figure # = kfig e.g., for Matlab version 7.01, ML_ver = 7.0 Vector

Euler angles {a,b,g}, in degrees, that define the orientation of the diffusion tensor axes with respect to the protein coordinate frame

Molar concentration of the protein (Ct) and the dimer’s dissociation constant (Kd), both in mM

Flag indicating various rotational diffusion models

Flag for selecting various CSA options

Flag to suppress (0) or allow (1) visual output in a form of data plots

Flag controlling figure numbers for plotting the results

Matlab version

List of residues you want to exclude from the analysis

alpha, beta, gamma

Ct, Kd

kovrl

kcsa

kplot

kfig

ML_ver

Exclude

NO

Set manually or let the program determine

Only to output to a specific figure#, otherwise set to 0 by default

Only for suppressing plot, otherwise set to 1 by default

Only for nonuniform CSA model, set to 0 by default

Only for nonisotropic motion, otherwise set to 0 by default

Only for monomer–dimer equilibrium model (kovrl = 2). The program will ask you to input them manually

Only for anisotropic diffusion models (kovrl = 1 or −1). The program will ask you to input them manually

Required?

b

Nfreq = number of frequencies in the freq list Nres = number of residues in the list. If data for some residues are unavailable, do not include these residues in the list or use NaN (see footnote c) c The values of relaxation rates R1, R2 and their experimental errors, sR1, sR2, should be in 1/s, the values of NOE and the experimental error, sNOE, are dimensionless. If for a given residue the relaxation parameter (R1, R2, or NOE) is not available, input NaN (“non-assigned-number”) in the corresponding position (second column) in the array d x,y,z should be coordinates of a unit vector in the direction of the NH bond (can be obtained by running an auxiliary program pdb2nh, see Subheading 3.6.1) e CSA values and their errors (sCSA) should be in ppm

a

Data format/structure

The meaning

Parameter or data array name (case sensitive)

Table 2 (continued)

24

Determining Protein Dynamics from 15N Relaxation Data…

497

- - - - - - - 600.13 MHz - - - - - - - - TOTAL: MEAN = 2.2131 SD(MEAN) = 0.12582 TAU = 3.3506 L&S: MEAN = 2.1806 SD(MEAN) = 0.062933 TAU(MEAN) = 3.2995 MEAN(TAU) = 3.2981 SD_TAU = 0.098668 TAUmc = 3.3001 SD_TAUmc = 0.11508 resid. with the R2/R1 within MEAN +/− SD: 37 resid. with the R2/R1 above MEAN + SD: 9 resid. with the R2/R1 below MEAN - SD: 5 The purpose of this analysis is to estimate the overall rotational 6R2 1 correlation time (as t c = − 7 , (33)) and to count, in 2wN R1 the spirit of (34), how many residues have the R2/R1 ratio within one standard deviation (SD) from the mean R2/R1 value. These residues are expected to fit into the “standard” Lipari & Szabo model (10, 11). Residues with the R2/R1 ratio more than one standard deviation below the mean value could require the Rex term, see Eq. 2, while those residues that have R2/R1 more than one standard deviation above the mean value might need the “extended” model-free model (12). At the start the program plots the experimental data (R2, R1, NOE, and residue-specific CSAs, if applicable) as a function of residue number (see Figs. 3 and 4). This output can be suppressed by setting kplot to 0 (or any number other than 1). The run-time dialog that follows is shown step-by-step below. Note that many questions that appear on the screen have a default answer (indicated in the square brackets): this answer will be assumed if you press ENTER, and if the question was about a parameter involved in computations, the program will output a message confirming that the corresponding value was assumed. Input a CSA value [−160] ==> This line appears if kcsa is set to 0 (default), i.e., a fixed uniform CSA value will be used. Input the desired value (only numeric input) or simply press ENTER: in this case CSA = −160 ppm will be assumed. Note that if kcsa was set to −1, a list of fixed (site-specific) CSA values must exist in the workspace; otherwise, the program will output an error message and exit. If you did not define TAUc value(s), the program will ask you the following: Input TAUc value(s) (in ns) ==> Here you can input a single value (e.g., 3.3) or a list of values, e.g., [3.28 3.3 3.32]. If you selected the isotropic rotational diffusion model (i.e., kovrl was set to 0 on undefined), the program will proceed to actual model-free analysis and model selection on a residue by residue

498

D. Fushman

Fig. 3. Output of DYNAMICS analysis of backbone motions in GB3 from 15 N relaxation data at 600 MHz. (a) Input data; (b) the results of analysis assuming isotropic overall tumbling with TAUc = 3.33 ns; and (c) the results of analysis assuming anisotropic (axially symmetric) overall tumbling with TAUc = 3.33 ns and other diffusion tensor characteristics presented in Subheading 2.3. A uniform 15N CSA value of −174.2 ppm was assumed throughout the protein. The circles on the “model” plot in b indicate residues that fall into the NOMOD category.

basis (see below). However, if the anisotropic diffusion model was selected (i.e., kovrl was set to −1 or 1), additional input requests will appear. In case kovrl was set to 1, you will see the following messages: <<<<>>>>> Input the Dz/Dx ratio (Dz/Dx < =0 -stop) [1]==> Enter the actual value of the ratio (in the case of axial symmetry, it is the same as D||/D^). Input a range of BETA values [0:10:90]==> Enter a range of b values, if you want to screen different orientations of the diffusion tensor, or just a single value (for example, the output of RotDif analysis). Input the ALPHA angle [0]==>

24

Determining Protein Dynamics from 15N Relaxation Data…

499

Fig. 4. Output of DYNAMICS analysis of backbone motions in GB3 from 15N relaxation data at five magnetic fields. (a) Input data. (b–c) the results of analysis assuming (axially symmetric) anisotropic overall tumbling (b) with a uniform (fixed) CSA = −174.2 ppm (as in Fig. 3) and (c) site-specific CSAs obtained simultaneously with the microdynamic parameters from fitting these relaxation data. The circles on the “model” plots indicate residues that fall into the NOMOD category.

Enter the value of angle a, in degrees. Input a starting BETA angle [0]==> If you entered a single b value when answering the BETAquestion above, reenter it here. Otherwise enter a single value of b that you want to start with. Note that the b value you enter in this line will be used first, even if you entered a different value or a range of values above. You will be then given the option to proceed with the above-entered b values. If kovrl was set to −1, the dialog will be similar, except that you will see the following message: <<<<<<<< anisotropic model selected >>>>>>>>

500

D. Fushman

And in addition to the questions listed above you will be asked to enter Dz/Dy and the angle g: Input Dy/Dx ratio [1]==> Input the GAMMA angle [0]==> If you selected the monomer–dimer equilibrium model (18) (i.e., kovrl = 2), you will be asked to input the total protein concentration and the dissociation constant (both in mM) prior to starting the analysis: <<<<<<<< monomer-dimer equilibrium >>>>>>>> Input protein concentration (in mM) ==> Input the dissociation constant, Kd (in mM) ==> The program will then compute and output the [monomer]/ [dimer] molar ratio and proceed to the data analysis (as described above for the isotropic tumbling option). After all required parameters (depending on the overall tumbling model) have been entered, the program will start model-free analysis. This analysis is performed on a per-residue basis, and for each residue the program outputs the results in the following format (these data are taken from the GB3 analysis): res# 48 LS_tx -model, S2 = 0.74325 TAUloc = 0.015475 Rex = 0.10926 chi2 = 5.904e-006 res# 49 LS_00 -model, S2 = 0.78293 TAUloc = 0 Rex = 0 chi2 = 0.61294 res# 50 CL_00 -model, S2 = 0.7772 TAUsl = 1.9757 Rex = 0 S2f = 0.85552 TAUf = 0 chi2 = 4.6997e-009 res# 52 LS_tl -model, S2 = 0.8388 TAUloc = 0.0064478 Rex = 0 chi2 = 2.6929e-005 res# 55 LS_tl -model, S2 = 0.80301 TAUloc = 0.016452 Rex = 0 chi2 = 0.48655 The models of local motion and the corresponding parameters are defined in Table 1; chi2 represents the residuals of fit (c2) for a given residue. In addition to numeric output, DYNAMICS visualizes/plots some of the results of the latest run on the screen (see examples in Figs. 3 and 4): the relevant microdynamic parameters (e.g., S2, tloc, Rex) and the selected local motion model (represented by its Index, see Table 1). As mentioned above, the plot option can be suppressed by the user by setting kplot = 0. If CSA was among the fitting parameters (i.e., kcsa = 1), the output also shows the resulting CSA values. After completing a run through all nonexcluded residues, the program outputs a summary of the results (see Subheading 2.5.3) and either continues the calculations for all other TAUc and/or b

24

Determining Protein Dynamics from 15N Relaxation Data…

501

values (if there is more than one value for each of these parameters), or stops and waits for user’s input. If the isotropic tumbling model was selected (kovrl = 0 or 2), the message on the screen will read as follows: Input TAUc (TAUc < = [0] - break)==> Entering a positive number will trigger another round of calculations with this TAUc value, whereas zero or a negative number will be interpreted as the signal to proceed to exit or error analysis. Note that the latest positive TAUc value will be taken as the final/ accepted value and used for error analysis. If the TAUc value that you want to accept is not the latest one, you need to reenter the desired value, let the program run through all residues again (this is quite fast anyway), and only after that enter 0 or a negative TAUc to exit or proceed to error analysis. In the case of anisotropic tumbling (kovrl = −1 or 1), the message on the screen reads as follows: Satisfied? (1-yes(calc.err), [0]-cont.(betarange), 2-man.input, -1-stop/exit)==> Enter 1 here to proceed to error analysis, 0 to continue computations with other β values (if more than one b value was entered above), 2 if you want to return to manual input of the diffusion tensor parameters (see above), and −1 to exit the program. If you choose to exit the program, it will automatically remove unnecessary (run-time) variables from the workspace and finish. If you choose to proceed to error analysis, the program will ask you to select the method of error estimation: Choose MC simulation of exper.data(0) or fitted params(1 or 2(vis = on)) ==> Selecting option 0 will generate synthetic experimental data (assuming normally distributed noise with the standard deviation sR1, sR2, or sNOE), and for each set of generated data will perform the fit using the same model of local motion as selected for the real data. By default, 500 runs will be performed for each residue, and the standard deviation will be displayed and included in the ERR array and in the final report RESERR. If you select option 1, the program will determine experimental errors using the constant c2-boundaries method (32), which assumes that the residuals of fit are distributed according to a c2 distribution, and therefore a deviation of the fitted parameters from the optimal value by one standard deviation would result in a specific increase in c2 that depends on the number of fitting parameters (e.g., Dc2 =1, 2.3 or 3.53 for Npar = 1, 2 or 3, respectively). Thus the program determines the confidence boundaries for the fitted parameters by generating their values randomly and keeping only those values that

502

D. Fushman

led to Dc2 below the corresponding threshold. By default, the simulation runs until 500 generated points fall into the defined Dc2 region. This method is usually faster, except for those rare cases when the errors in fitting parameters are extremely small. You can visualize the confidence regions for selected parameters if you select option 2. As the error estimation proceeds, the program will output on the screen the results (standard deviation, SD) for every residue and, after it is finished, will also update the results plot with error bars. If you choose options 1 or 2 in the isotropic tumbling mode, the program will present you with an option to vary TAUc together with the other parameters such that the estimated errors reflect the possible uncertainty in tc as well. However, since tc is not a fitting parameter, this option should be used for evaluation purposes only. After the program run is finished, you can save the results that you want to keep by using Matlab’s save command, for example: > > save results.mat ANISO

RESERR

NOMOD

EXCL

TAUCHI

This will save RESERR, NOMOD, and other parameters listed in that command line to a Matlab file results.mat (which stores data in a binary format). If you want to save your results in ascii format (to be easily opened by a text editor or any spreadsheet program), type: > > save results.dat RESERR –ascii Type help save to see other saving options that Matlab provides. See also the description of DYNAMICS’s automatic saving feature in Subheading 3.5. 2.5.4. Understanding the Output

DYNAMICS outputs the results of the analysis on the screen, both in numerical format and as plots, and stores them in several output parameters/arrays, summarized in Table 3. As discussed above, during each run the program outputs the results of analysis for each residue. In addition to the obvious issues, such as the model and the actual values of the microdynamic parameters, the user should also pay attention to the residuals of fit (chi2 = c2), which are represented by the last number in each row (or the one before last if CSA is also a fitting parameter). The chi2 information is important, because it tells you how well the data fit the model. Ideally, a good fit would give chi2 values of about 1 per degree of freedom. Thus, chi2 numbers in the range of single digits (» df) or lower indicate a reasonable fit, whereas much higher chi2 values indicate a potential problem with data analysis for a particular residue: either the “best”-fit model is not ideal for that residue or perhaps the experimental errors are underestimated (hence elevated chi2).

24

Determining Protein Dynamics from 15N Relaxation Data…

503

Table 3 Output parameters created by program DYNAMICS Parameter

Meaning and data format

NOMOD

List of residues in the NOMOD category (see Subheading 2.4), i.e., residues for which none of the models of local motion passed the goodness-of-fit test (c 2 too high), although at least one model provides a physically meaningful set of microdynamic parameters

EXCL

List of residues that have been excluded by the program because none of the tested models of local motion are able to provide a physically meaningful set of microdynamic parameters (see Subheading 2.4)

RES

The results of fit in the following format (array Nres x 10): [Residue# tc S2 tloc Rex Sfast2 tfast model-Index c 2 CSA]

ERR

The results of error analysis in the following format (array Nres x 7): [Residue# dtc dS 2 d tloc dRex dSfast2 dtfast]

RESERR

Combined results of fit (RES) and error analysis (ERR) in the following format (array Nres x 16): [Residue# tc dtc S2 d S2 tloc d tloc Rex d Rex Sfast2 dSfast2 tfast dtfast model-Index c 2 CSA]

TAUCHI

Record of all evaluations performed during the current DYNAMICS run in the isotropic tumbling mode or monomer–dimer equilibrium (empty if anisotropic tumbling). Each line is a summary statistics for all residues, in the following format: tc, c 2(mod), Nnomod, c 2(nomod), runs, df, c 2(total), c 2(total)/df, Nexcl

ANISO

Record of all evaluations performed during the current DYNAMICS run in the anisotropic mode (empty matrix in the isotropic or monomer-dimer equilibrium). Each line is a summary of statistics for all residues, in the following format (for axially symmetric model): tc, Dz/Dx,b,a, c 2(mod),Nnomod,c 2(nomod), runs, df, c 2(total), c 2(total)/df, NexclIn the case of fully anisotropic tumbling model, the format is tc, Dz/Dx, Dy/Dx b,a,g, c 2(mod),Nnomod,c 2(nomod), runs, df, c 2(total), c2(total)/df, Nexcl

runs

This parameter counts how many times the selected model switches between model-free and “extended” model-free models in adjacent residues along the protein sequence

df

Total number of degrees of freedom, df = Ndat – Npar

chi2

Residuals of fit, c 2 χ2 =

⎡⎛ R exp − R calc ⎞ 2 ⎛ R exp − R calc ⎞ 2 ⎛ NOEexp − NOEcalc ⎞ 2 ⎤ 1 2 2 ⎢⎜ 1 ∑ ⎟⎠ ⎥ ⎟⎠ + ⎜⎝ ⎟⎠ + ⎜⎝ s R s R sNOE freq ⎢ ⎝ ⎥⎦ 1 2 ⎣

In addition to results for each residue, after each run through all residues is completed, the program outputs a summary of the results of the current run, which looks like this: NOMODEL 2 res.: 36 45 chi2 = 52.8946 TAUc = 3.33 MODEL chi2 = 107.0356 runs = 14 df = 46 TOTAL Chi2 = 159.9302 Chi2/df = 3.4767

504

D. Fushman

The first line here is optional and appears only if there is at least one residue that falls into the NOMOD category (see Subheading 2.4): it lists those residues and their total chi2. The second line reports the current tc value, the total chi2 value for “MODEL” residues, i.e., those for which proper model selection was obtained, and a summary of other statistics of the results (see Table 3). The last entry in this line is the total chi2 divided by the total number of degrees of freedom. The program also summarizes similar statistics for all iterations (various TAUc values) performed so far in a form shown below for the isotropic tumbling model: TAUc chi2mod nomod chi2nomod runs df total_chi2 totalCHI2/df nexcl 3.15 87.4701 5 72.5883 0 39 160.0584 4.1041 0 3.33 107.0356 2 52.8946 14 46 159.9302 3.4767 0 < −− 3.45 100.6942 5 157.2367 22 42 257.9309 6.1412 0 Here mod and nomod refer to MODEL and NOMOD categories, and nexcl is the number of residues in the EXCL category, i.e., for which no model could be found; the rest of the parameters are defined in Table 3. These data are also stored in array TAUCHI. The horizontal arrow on the right indicates the entry with the lowest total c2/df. For anisotropic diffusion model the summary table is slightly different (because of the additional overall-tumbling-related parameters) and looks like this (also saved in array ANISO, see Table 3): TAUc Dz/Dx beta alpha - > chi2mod nomod chi2nomod runs df total_chi2 totalCHI2/df nexcl 3.33 1.36 65 90 - >76.8789 0 0 6 49 76.8789 1.569 0 3.33 1.36 70 90 ->78.0965 0 0 4 50 78.0965 1.5619 0 For fully anisotropic diffusion model, the summary table also includes Dy/Dx and angle gamma. 2.6. Practical Examples

The examples of DYNAMICS graphics outputs showing the input 15N relaxation data for GB3 at 14.1 T (600 MHz) and the resulting microdynamic parameters and models of local motion are shown in Fig. 3. The analysis was performed assuming isotropic or anisotropic overall tumbling of the protein. Differences in the microdynamic parameters and selected models of local motion illustrate the need to use an adequate model for the overall tumbling (anisotropic rotational diffusion in the case of GB3 (35)). The analysis of GB3 data measured at five fields (9.4, 11.7, 14.1, 16.4, and 18.8 T) assuming a uniform (fixed) 15N CSA or including site-specific 15N CSA as a fitting parameter is illustrated in Fig. 4.

24

Determining Protein Dynamics from 15N Relaxation Data…

505

3. Miscellaneous Issues 3.1. General Notes on Using DYNAMICS

1. Make sure the input list of frequencies freq contains all the pertinent 1H frequencies (magnetic fields) for the data that you want to analyze. It is critical that the order in which the frequencies are listed in the freq-list is coordinated with the second “index” in the relaxation parameters names. For example, if freq = [500, 600], then r11, r21, and r31 should be R1, R2, and NOE data at 500 MHz, respectively, while r12, r22, and r32 should be R1, R2, and NOE data at 600 MHz. 2. Small Rex values (<1 s−1) could be artifacts of model selection (for example, due to an inadequate overall tumbling model, as illustrated in Fig. 3) or a result of elevated 15N CSA, rather than real exchange contributions. Relaxation data at more than one magnetic field are required (analyzed separately or together) to verify the consistency of the model selection as well as that the resulting Rex term has the expected field dependence (∝ Bo2). 3. To fit CSA values, relaxation data at more than one magnetic field are required. 4. When conformational exchange (Rex) is present, the CSA values obtained from the analysis could be biased, since both the Rex and the CSA contributions to relaxation rates have a similar field dependence (∝ Bo2). 5. Errors in the derived CSA values, as well as the associated errors in microdynamic parameters from such analysis (kcsa = 1), can currently be calculated only using the “experimental data simulation” option. 6. In the output RES matrix, the CSA values are included as column #10, following the c2 values. In the output RESERR matrix the CSA values and errors in CSA are included as the last two columns (#16, 17). 7. Even if you input a single TAUc value for the isotropic or monomer–dimer equilibrium model, the program will run calculations for this TAUc twice. This is done to compare the results of at least two runs and, because DYNAMICS is fast, does not take too much time.

3.2. Strategy for Determining the Overall Rotational Correlation Time When the Structure is Unknown or RotDif Results are Unreliable

What can one do if protein coordinates are not available? One can always use the isotropic overall tumbling model in DYNAMICS, which does not require knowledge of protein structure. However, one should be aware of the fact that the microdynamic parameters derived from such analysis could be biased by the possible oversimplification of the tumbling model used, particularly if the protein shape turns out to be far from spherical.

506

D. Fushman

While RotDif provides an efficient and reliable way to determine the overall rotational diffusion tensor, it is also possible to estimate the tensor (or at least some of its characteristics) simultaneously with DYNAMICS analysis. This feature was built into DYNAMICS in the old days when RotDif was unavailable, and still remains. Here is the recipe that I recommend if you want to determine tc from 15N relaxation data at a single field. 1. Use only residues from structurally well-defined regions (“core”) of the protein. Exclude any residues that are in highly flexible/unstructured parts (e.g., loops and termini) of your protein: they would likely require the “extended”-model-free models. You can do this based on the structure of the protein, or using only residues for which the R2/R1 ratio falls within one standard deviation from the mean (the estimates are provided at the very beginning of each DYNAMICS run). You also might want to exclude residues with noticeably high R2 values but “average” R1 values. It might be helpful to plot the R2/R1 ratio for such an analysis. The rationale for these exclusions is that at a single field there is no reliable F-statistics test for models (CL_00, LS_tx) with df = 0, so you try to minimize these contributions. 2. Start with the estimate of TAUc that DYNAMICS provides at the very beginning of the run. 3. Run DYNAMICS once at this TAUc, and analyze the output results for various residues. Identify residues that give very high chi2, compared to the rest of residues. You will want to exclude these residues during the next run. The rationale is that these residues will dominate the total chi2, thus biasing the optimization based on the residuals of fit. For example, if the average chi2 level is in the single digits, it might be prudent to exclude residues that gave chi2 above 30–50. 4. With the new set of residues, run DYNAMICS while varying TAUc. The goal is to find TAUc that minimizes the total chi2/ df. This optimization could be tricky, because when the input TAUc deviates from the actual tc, the model selection procedure could compensate by choosing a different model. For example as TAUc increases, there is a tendency toward the CL_00 model, for which chi2 could be very low (recall that for that model df = 0 when using data at a single field, thus the F-statistics test does not work in this case, and acceptance of this model is somewhat arbitrary). The same happens when TAUc is lower than the actual tc: the model selection will compensate by choosing LS_ex or LS_tx models (recall that for the latter df = 0, hence the F-statistics does not work). This would also result in a decrease in the total number of degrees of freedom. Therefore the name of the game here is to find a

Determining Protein Dynamics from 15N Relaxation Data…

24

b 10

Number of residues

Rex-models CL-model

5 0 4.5

Rex-models CL-model

0 4 2

NOMOD

0 4.5

45

40

4.0

4.0 40

3.5

df

3.0 35

2.5

χ2/df

3.5

χ2/df

10

35

3.0

df

Number of residues

a

20

507

2.5 2.0

2.0

30

30

1.5

1.5 1.0

1.0 3.25

3.30

3.35

3.40

TAUc, ns

3.45

3.15 3.20 3.25 3.30 3.35 3.40 3.45

TAUc, ns

Fig. 5. Illustration of the use of DYNAMICS to optimize tc simultaneously with model-free analysis of 15N relaxation data for GB3. (a) The analysis assumed an anisotropic tumbling model with the diffusion tensor’s anisotropy and orientation as specified in Subheading 2.3. (b) The analysis assumed isotropic tumbling model. The 15N CSA was set to −174.2 ppm (fixed) in both cases. Only secondary structure residues were included; in addition, residues D36 and Y45 were excluded from this analysis because of high c 2 values. Solid line in both panels shows c 2/df, while the dashed line depicts the total number of degrees of freedom, df. The vertical bars in the top plots show the number of residues with Rex (blue) (LS_ex and LS_tx models) or with extended model (CL_00) (red) selected for each value of TAUc, as well as the number of NOMOD residues (black). There were no NOMOD residues in the analysis in a. The RotDif analysis gives the tc value of 3.34 ± 0.14 ns (19) (or 3.37 ± 0.20 ns for these selected residues); a very similar value (3.36 ns) calculated from the mean R2/R1 value was reported at the beginning of the DYNAMICS run.

TAUc value (or range) in the “middle” region, where the total chi2 is at or close to its minimum and at the same time the total df is at or close to its maximum. There might not be a clear single TAUc value, but one should remember there is always an uncertainty in tc even from RotDif analysis. Examples of such analyses are shown in Fig. 5. In addition, to help deal with an artificial selection of the CL_00 model, DYNAMICS also uses another parameter, called runs (Table 3). The rationale behind it is to avoid an artificial situation when a single residue shows large-amplitude motions (this is what the extended model is usually needed for) while its neighbors do not. To avoid this, runs counts how many times the selected model switches between “LS” and “CL” models in adjacent residues. Naturally, one wants to minimize the number of such switches, because intuitively one would expect that largeamplitude fluctuations involve several adjacent residues in the polypeptide chain, and not just a single one.

508

D. Fushman

One should also bear in mind that the total chi2/df (as well as chi2) is generally not a smooth function because changes in the selected model of local motion for individual residues would result in abrupt changes in the residuals of fit and df. 5. After the TAUc value has been optimized, use this value to run DYNAMICS again, this time for all residues in the protein. You might need to do several iterations of such an analysis. 3.3. Strategy for Determining the Overall Rotational Diffusion Tensor when RotDif Results are Unreliable

The strategy here is similar to that described in the previous section, except that (1) protein atom coordinates are required and (2) you will need to vary several parameters as the same time. For example, for the axially symmetric model, you might want to vary Dz/Dx and the angles a and b. While doable, this optimization is not straightforward, can require significant effort, and can result in a local rather than a global minimizer. Thus, I would recommend using it only when no other option is available.

3.4. Cleaning the Workspace

DYNAMICS generates and keeps most of the necessary intermediate variables in the current Matlab workspace (computer memory). They are automatically removed from the workspace when the program terminates successfully. If the program run was terminated prematurely either by the user (e.g., via CTRL/C) or in case of a run-time error, these variables will remain in the workspace. This might cause an interruption in the normal program execution when you start it next time during the same Matlab session. To ensure uninterrupted program flow, it is recommended to remove the remaining intermediate variables from the computer memory before restarting DYNAMICS. Removal of only intermediate variables can be achieved by issuing the following command: > > dynclean

3.5. Automatic Saving of the Results

To prevent accidental loss of the computed data, the results (RES, NOMOD, EXCL, TAUCHI, ANISO) are automatically saved to a Matlab file after completion (and acceptance) of the model-free analysis and again after error analysis (the same parameters as above plus ERR and RESERR). To reduce the chance of overwriting this file when you run DYNAMICS again, the name of the file contains the current date followed by a random number from 0 to 99, e.g., dyn16jan2011_92.mat.

3.6. Auxiliary Programs

The DYNAMICS package includes several programs designed to help the user prepare data for running the calculations. Some of these programs are briefly described below. Their use and the actual command lines are not detailed here: the reader can find all relevant information in the header of each program using any text editor (e.g., Matlab editor).

24

Determining Protein Dynamics from 15N Relaxation Data…

509

3.6.1. pdb2nh

This program extracts coordinates of backbone NH vectors from a given protein atom coordinates file and normalizes these vectors (to be used as input for DYNAMICS). In case hydrogens cannot be found (e.g., crystal structure), the program builds amide hydrogens from coordinates of the heavy atoms C¢, O, N, Ca in the corresponding peptide plane using conventional rules.

3.6.2. Reldata, Reldatae

Given all pertinent parameters of the overall and local dynamics, as well as the orientation of the NH vector (if necessary), the reldata program computes 15N relaxation rates: R1, R2, and NOE. The input options also include the ability to add random noise to the data. The program reldatae performs the same task as reldata, but in addition also computes the longitudinal (hz) and transverse (hxy) cross-correlation rates between the 1H–15N dipolar interaction and 15 N CSA, e.g., see ref. 19, 36.

3.6.3. conv2temp

This program allows conversion between tc values at different temperatures, by taking into account the temperature dependence of water viscosity, see e.g., ref. 37.

3.6.4. Demo Scripts

The package includes several demo scripts, designed to help the user learn how to run DYNAMICS: demo_iso.m

Isotropic overall tumbling

demo_ani.m

Anisotropic overall tumbling

demomdeq.m

Monomer–dimer equilibrium

demo_csa_iso.m

CSA fit + isotropic overall tumbling

demo_csa_ax.m

CSA fit + axially symmetric anisotropic overall tumbling

demo_csa_ani.m

CSA fit + fully anisotropic overall tumbling

All “demo_csa” scripts use data at five magnetic fields and include CSA as a fitting parameter. Note that these scripts can be modified to use site-specific CSA values as an external fixed parameter (kcsa = −1) rather than as an adjustable parameter. All you need is to open this file with any text editor and uncomment (remove %) the line kcsa = −1. Each of these scripts starts by generating synthetic sets of relaxation data using reldata.m, and then runs DYNAMICS using these data as input. Additional three text files, demo_iso.txt, demo_ani. txt, and demomdeq.txt, contain copies of the screen outputs and the dialog, to illustrate the main steps in data analysis using DYNAMICS.

510

D. Fushman

Acknowledgments The development of DYNAMICS program was supported by NIH grant GM 065334. My work on this chapter has led to several modifications of the program, which hopefully made it user-friendlier, and I would like to thank the editors, Alex Shekhtman and David Burz, for being so patient with me during this process. References 1. Palmer, A. G., 3 rd. (2004) NMR characterization of the dynamics of biomacromolecules. Chem. Rev. 104, 3623–3640. 2. Sheppard, D., Sprangers, R., and Tugarinov, V. (2010) Experimental approaches for NMR studies of side-chain dynamics in high-molecular-weight proteins. Prog. Nucl. Magn. Reson. Spectrosc. 56, 1–45. 3. Godoy-Ruiz, R., Guo, C., and Tugarinov, V. (2010) Alanine methyl groups as NMR probes of molecular structure and dynamics in highmolecular-weight proteins. J. Am. Chem. Soc. 132, 18340–18350. 4. Cavanagh, J., Fairbrother, W. J., III, A. J. P., and Skelton, N. J. (1996) Protein NMR Spectroscopy, Academic Press, San Diego. 5. Fushman, D., and Cowburn, D. (1999) The effect of noncollinearity of 15 N-1 H dipolar and 15 N CSA tensors and rotational anisotropy on 15 N relaxation rates, CSA/DD cross correlation, and TROSY. J. Biomol. NMR 13, 139–147. 6. Fushman, D., Ohlenschlager, O., and Rüterjans, H. (1994) Determination of the backbone mobility of ribonuclease T1 and its 2’GMP complex using molecular dynamics simulations and NMR relaxation data. J. Biomol. Struct. Dyn. 11, 1377–1402. 7. Pfeiffer, S., Fushman, D., and Cowburn, D. (2001) Simulated and NMR derived backbone dynamics of a protein with significant flexibility: A comparison of spectral densities for the < beta > ARK PH domain. J. Am. Chem. Soc. 123, 3021–3036. 8. Maragakis, P., Lindorff-Larsen, K., Eastwood, M. P., Dror, R. O., Klepeis, J. L., Arkin, I. T., Jensen, M. O., Xu, H., Trbovic, N., Friesner, R. A., Palmer, A. G., and Shaw, D. E. (2008) Microsecond molecular dynamics simulation shows effect of slow loop dynamics on backbone amide order parameters of proteins. J. Phys. Chem. B 112, 6155–6158. 9. Fushman, D., and Cowburn, D. (2001) Nuclear magnetic resonance relaxation in

10.

11.

12.

13.

14.

15.

16.

17.

18.

determination of residue-specific 15 N chemical shift tensors in proteins in solution: protein dynamics, structure, and applications of transverse relaxation optimized spectroscopy, in Methods in Enzymology (James, T., Schmitz, U., and Doetsch, V., Eds.), 339, 109–126. Lipari, G., and Szabo, A. (1982) Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 2. J. Am. Chem. Soc. 104, 4559–4570. Lipari, G., and Szabo, A. (1982) Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 1. Theory and range of validity, J. Am. Chem. Soc. 104, 4546–4559. Clore, G. M., Szabo, A., Bax, A., Kay, L. E., Driscoll, P. C., and Gronenborn, A. M. (1990) Deviations from the simple two-parameter model-free approach to the interpretation of nitrogen-15 nuclear magnetic relaxation of proteins. J. Am. Chem. Soc 112, 4989–4936. Woessner, D. (1962) Nuclear spin relaxaion in ellipsoids undergoing rotational brownian motion. J.Chem.Phys. 37, 647–654. Favro, D. L. (1960) Theory of the Rotational Brownian Motion of a Free Rigid Body. Phys. Rev. 119, 53–62. Ryabov, Y. E., and Fushman, D. (2007) A Model of Interdomain Mobility in a Multidomain Protein. J. Am. Chem. Soc. 129, 3315–3327. Tjandra, N., Feller, S. E., Pastor, R. W., and Bax, A. (1995) Rotational diffusion anisotropy of human ubiquitin from 15 N NMR relaxation. J. Am. Chem. Soc. 117, 12562–12566. Palmer, A. G., 3 rd, Grey, M. J., and Wang, C. (2005) Solution NMR spin relaxation methods for characterizing chemical exchange in highmolecular-weight systems. Methods Enzymol. 394, 430–465. Fushman, D., Cahill, S., and Cowburn, D. (1997) The main chain dynamics of the dynamin pleckstrin homology (PH) domain in solution: Analysis of 15 N relaxation with monomer/dimer equilibration. J. Mol. Biol. 266, 173–194.

24

Determining Protein Dynamics from 15N Relaxation Data…

19. Hall, J. B., and Fushman, D. (2006) Variability of the 15 N Chemical Shielding Tensors in the B3 Domain of Protein G from 15 N Relaxation Measurements at Several Fields. Implications for Backbone Order Parameters. J.Am.Chem. Soc. 128, 7855–7870. 20. Fushman, D., and Cowburn, D. (1998) Studying protein dynamics with NMR relaxation, in Structure, Motion, Interaction and Expression of Biological Macromolecules (Sarma, R., and Sarma, M., Eds.), pp 63–77, Adenine Press, Albany, NY. 21. Blackledge, M., Cordier, F., Dosset, P., and Marion, D. (1998) Precision and uncertainty in the characterization of anisotropic rotational diffusion by 15 N relaxation. J.Am.Chem.Soc. 120, 4538–4539. 22. Fushman, D., Xu, R., and Cowburn, D. (1999) Direct determination of changes of interdomain orientation on ligation: use of the orientational dependence of 15 N NMR relaxation in Abl SH(32). Biochemistry 38, 10225–10230. 23. Fushman, D., Ghose, R., and Cowburn, D. (2000) The effect of finite sampling on the determination of orientational properties: A theoretical treatment with application to interatomic vectors in proteins. J. Am. Chem. Soc. 122, 10640–10649. 24. Dosset, P., Hus, J. C., Blackledge, M., and Marion, D. (2000) Efficient analysis of macromolecular rotational diffusion from heteronuclear relaxation data. J. Biomol. NMR 16, 23–28. 25. Ghose, R., Fushman, D., and Cowburn, D. (2001) Determination of the Rotational Diffusion Tensor of Macromolecules in Solution from NMR Relaxation Data with a Combination of Exact and Approximate Methods - Application to the Determination of Interdomain Orientation in Multidomain Proteins. J. Magn. Reson. 149, 214–217. 26. Walker, O., Varadan, R., and Fushman, D. (2004) Efficient and accurate determination of the overall rotational diffusion tensor of a molecule from 15 N relaxation data using computer program ROTDIF. J. Magn. Reson. 168, 336–345. 27. Fushman, D., Varadan, R., Assfalg, M., and Walker, O. (2004) Determining domain orientation in macromolecules by using spin-relaxation and residual dipolar coupling measurements. Prog. NMR Spectros. 44, 189–214. 28. Hall, J. B., Walker, O., and Fushman, D. (2004) Characterization of the overall rotational

29.

30.

31.

32.

33.

34.

35.

36.

37.

511

diffusion of a protein from 15 N relaxation measurements and hydrodynamic calculations, in Protein NMR techniques (Methods in Molecular Biology) (A.K.Downing, Ed.), pp 139–160, Humana Press Inc. Fushman, D., and Cowburn, D. (2002) Characterization of Inter-Domain Orientations in Solution Using the NMR Relaxation Approach, in Protein NMR for the Millenium (Biological Magnetic Resonance Vol 20) (N. R. Krishna, L. B., Ed.), pp 53–78, Kluwer. Fushman, D. (2002) Determination of protein dynamics using 15 N relaxation measurements, in BioNMR in drug research (O.Zerbe, Ed.), pp 283–308, Wiley-VCH. Mandel, A. M., Akke, M., and Palmer, A. G. I. (1995) Backbone dynamics of E. coli Ribonuclease HI: correlations with structure and function in an active enzyme. J. Mol. Biol. 246, 144–163. Press, W. H., Teukolsky, S. A., Vetterling, W. T., and Flannery, B. P. (1992) Numerical Recipes in C, Cambridge University Press, NY. Fushman, D., Weisemann, R., Thüring, H., and Rüterjans, H. (1994) Backbone dynamics of ribonuclease T1 and its complex with 2’GMP studied by two-dimensional heteronuclear NMR spectroscopy. J. Biomol. NMR 4, 61–78. Kay, L. E., Torchia, D. A., and Bax, A. (1989) Backbone dynamics of proteins as studies by N15 inverse detected heteronuclear NMR spectroscopy: application to staphylococcal nuclease. Biochemistry 28, 8972–8979. Hall, J. B., and Fushman, D. (2003) Characterization of the overall and local dynamics of a protein with intermediate rotational anisotropy: Differentiating between conformational exchange and anisotropic diffusion in the B3 domain of protein G. J. Biomol. NMR 27, 261–275. Hall, J. B., and Fushman, D. (2003) Direct measurement of the transverse and longitudinal 15 N chemical shift anisotropy-dipolar crosscorrelation rate constants using 1 H-coupled HSQC spectra. Mag. Res. in Chemistry 41, 837–842. Ryabov, Y. E., Geraghty, C., Varshney, A., and Fushman, D. (2006) An efficient computational method for predicting rotational diffusion tensors of globular proteins using an ellipsoid representation. J. Am. Chem. Soc. 128, 15432–15444.

INDEX A Abelson Kinase domain (13C–15N labeled)......................... 43 Acetamidase gene (amdS), 21 ADR. See Ambiguous distance restraints Affinity/solubility tags Flag............................................................................ 335 glutathione-S-transferase (GST) ............................... 335 hexahistidine (His) .................................................... 335 maltose binding protein (MBP) ........................ 335–336 trpLE-tag .................................................................. 336 AMBER force field ................................................. 380–381 Ambiguous distance restraints (ADR) ................... 444–448, 454, 456, 459, 460, 462, 478 Ambiguous Restraints for Iterative Assignment (ARIA) .......................................... 436, 453–480 Amino acid specific labeling ............................................ 236 Amino acid type selective (AATS) isotope labeling .......... 59 Amplification of recombinant virus ............................. 45, 47 APSY. See Automated projection spectroscopy ARIA. See Ambiguous Restraints for Iterative Assignment Automated backbone assignment .................... 433, 438–441 Automated projection spectroscopy (APSY ) 5D APSY-CBCACONH ......................................... 433 5D APSY-HACACONH ......................................... 433 4D APSY-HACANH ............................................... 433 Auto relaxation rates ........................................ 142, 146–151

B Backbone resonance assignment ...............286, 408, 413, 433 BacMagic....................................................41–42, 44–47, 50 BacMams........................................................................... 38 Bacmid............................................................................... 42 Bacterial over-expression apoE.... .................................................................... 9, 15 calmodulin (CaM) ........................ 72–73, 78, 80, 83, 263 MscL.................................. 86–91, 94–101, 104, 106, 107 replication protein A (RPA) .............................. 185, 189 rhodopsin..................................................................... 59 SrcCD-MBP fusion .................................. 115–118, 127 trpLE-M2 ......................................................... 168, 171 Bacterial transformation ......................................................9 Baculovirus-mediated expression (BvE) system ............................................ 38–41, 43, 45

Baculovirus-mediated insect cells .......................... 38, 39, 43 BATCH protocol (strategy for resonance assignments) ASCOM............................................ 409–410, 414–416 BEST... ...............................................409, 413, 414, 416 COBRA .............................410–412, 415–417, 419–427 HADAMAC .............. 411–413, 415–417, 419–425, 427 targeted-sampling ...................................................... 411 B3 domain of protein G (GB3) .............................. 492, 494, 498–500, 504, 507 Boltzmann conformational distribution ................. 370, 374, 376, 385, 390, 400

C Calibration factor (C) ...................................................... 460 CCPN. See Collaborative Computing Project for the NMR Cdc37 (kinase-specific chaperone) .......................... 117–118 13 C-detected 1H, 13C correlation spectrum.............. 285–286 13 C-detected 2H-DQ, 13C correlation ..................... 284, 285 Cell-free membrane protein expression in detergent micelles ........... 87–90, 94–97, 100, 350–351 in liposomes ............................................................... 100 of MscL in liposomes ................................ 90–91, 98–99 Cell free synthesis/expression ........ 71, 74–77, 81–82, 85–91, 94–100, 106 Cell viability assay ........................................... 269, 272, 274 CELTONE® ............................................120, 123, 126, 364 Charge distribution atomic.. .............................................................. 381–383 electron ...................................................................... 383 CHARMM ..............................................380, 381, 383, 393 Chemical exchange fast exchange regime .................................................. 238 intermediate exchange regime ................................... 263 slow exchange regime ................................ 220, 238, 255 Chemical shift anisotropy (CSA) ....................142, 287–289, 322, 323, 413, 487–489, 497–500, 502–505, 509 Chemical shift perturbation (CSP) ........................ 138, 212, 216, 237–239, 246–250, 255, 307, 365 Chemical shifts .......................... 72, 142, 143, 169, 172–175, 198, 210, 220, 235, 238, 324, 361–362, 407–410, 412, 416, 425–426, 435–447, 455–458, 464–465, 467, 469, 477, 486 CHHC...... ...................................................................... 477

Alexander Shekhtman and David S. Burz (eds.), Protein NMR Techniques, Methods in Molecular Biology, vol. 831, DOI 10.1007/978-1-61779-480-3, © Springer Science+Business Media, LLC 2012

513

PROTEIN NMR TECHNIQUES 514 Index Chromatography affinity, Ni-NTA........................................................ 189 desalting..................................................... 185, 190–191 heparin....................................................... 185, 189–191 ion exchange (IEC) ............................116, 118, 120, 124 reverse phase ............................... 168, 172, 229, 336, 352 size-exclusion (SEC) .................................118, 120, 121, 125, 189, 198, 210, 310, 313, 315, 317, 339 Circular dichroism (CD) spectroscopy .................... 340–341 CLEAN chemical EXchange-phase modulator (CLEANEX-PM) ......................... 372, 378–380 CMC. See Critical micelle concentration Collaborative Computing Project for the NMR (CCPN) .................................454–456, 465, 467, 469, 473–476, 478, 479 Combined rotation and multiple pulse spectroscopy (CRAMPS) technique ................................... 281 Confocal microscopy HIV-1 CA assemblies ........ 305, 307, 310–313, 316–318 Conformation distribution ..............................369, 370, 373, 374, 376, 385–388, 390, 392, 394, 396, 400 Conformer acidities .................. 385, 387–389, 393, 395, 399 Constraint combination........................................... 444, 448 COREX algorithm .......................................................... 371 Correlation spectroscopy ..........................281–286, 304, 321 Co-transfection of insect cells ......................... 43, 44, 46–47 CP. See Cross polarization 1 H–15N CP block ..................................................... 320, 321 Critical micelle concentration (CMC) ................... 339–341, 343, 345, 347–348 Cross polarization (CP) ................................... 281, 304, 327 Cryo-SEM HIV-1 CA assemblies ....................................... 318–319 Crystallography and NMR system (CNS) software.................................................. 138, 455 CSA. See Chemical shift anisotropy CSA/DD cross-correlated cross-relaxation rates ..... 153–157 15 N CSA shielding tensor ................................................ 290 CSP. See Chemical shift perturbation CYANA............................................432–433, 437, 444, 454

D DARR sequence ............................... 304, 306, 307, 321, 326 DD coupling. See Dipole-dipole coupling DelPhi....... ...................................................................... 380 Detergent micelles reconstitution of M2 into .................................. 169, 172 Detergents (membrane proteins) d38-DPC ........................................................... 343, 345 dihexanoyl-sn-glycerol-3-phosphocholine (DHPC) .........................167, 169, 172–174, 177 dimyristoylphosphatidylcholine (DMPC) ........................................ 340, 345, 349 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) ...................................60, 90, 91, 97, 98

n-dodecyl-β-D-maltopyranoside (DDM) .......................................... 345, 347, 348 dodecylphosphocholine (DPC) ..........339–341, 343, 345 n-octyl-β-D-glucopyranoside (β-OG) ...................... 345 Deuterated target proteins ................................................. 28 Dielectric shielding...........................376–378, 384, 397–398 4,4-Dimethyl-4-silapentane-1-sulfonic acid (DSS) ........................ 92, 100–101, 346, 351, 414 Dimyristoylphosphatidylcholine (DMPC) bilayer ............................................ 340, 345, 349 Dipolar coupling 1 H, 1H dipolar coupling ..................................... 281, 487 N–H dipolar coupling, d .................................... 288–289 Dipole-dipole (DD) coupling.......................................... 142 Discoidin domain of DDR2 .............................................. 28 Dissociation constant ....................... 225, 239, 243, 496, 500 DNA processing ...................................................... 181, 182 DNA template...................... 75, 83, 181, 198, 199, 201–203 DOPC liposome preparation ................................ 91, 97, 98 Double colony selection ................................2, 3, 5, 7–10, 12 Double cross polarization (DCP) block .......................... 321 1D proton experiment ..................................................... 144 DREAM sequence .......................................... 321, 326–327 3D ROCSA-NCA experiment ............................... 322–323 DYNAMICS protocol ............................................ 485–509

E Eigen acid... ..................................................... 375, 378, 388 Electronic polarizability........................................... 376–378 Electrophoretic mobility shift assay (EMSA).................. 224 Episomal vectors.......................................................... 56–57 Eukaryotic protein kinase-2 (ERK2)....................... 359–367 Expression system bacteria, E. coli BL21(DE3) ........................... 7, 74, 77, 79, 135, 168, 171, 268, 273, 312, 314, 315, 319, 344, 346, 351 BL21(DE3) pLysS .......................168, 171, 183, 184 Rosetta 2 (DE3) .......................................... 313, 316 baculovirus Autographa californica ............................................. 40 multicapsid nucleopolyhedrovirus (AcMNPV) ............................................... 40, 42 insect cells Spodoptera frugiperda Sf9/Sf21 ......................... 43–49 Trichoplusia ni BTI 5B1-4 (High Five™) cells .................................................................. 43 mammalian cells human embryonic kidney 293 (HEK293) cells .........................39, 57–59, 61, 63 optimization .............................................................. 125 yeast, Kluyveromyces lactis (K. lactis) acetamidase gene (amdS) ....................................... 21 Lac4 promoter........................................................ 21 yeast, Pichia pastoris (P. pastoris) AOX1 promoter/gene............................................. 20

PROTEIN NMR TECHNIQUES 515 Index F FAST-heteronuclear single quantum correlation (FAST-HSQC) ............................................. 247 Fast NMR methodologies concentric ring sampling.................................... 266, 267 hybrid back projection/lower value (HBLV) reconstruction algorithm................................ 267 projection reconstruction NMR (PR-NMR) ............ 267 radial sampling pattern .............................................. 266 random sampling ............................................... 266, 267 sparse sampling pattern ............................................. 266 Fed-batch fermentation ............................................... 30, 31 FlashBAC... ..........................................41–42, 45, 47, 48, 50 Floating chirality assignment .................................. 464–465 FSLG-based 1H–15N HETCOR experiment.................. 324

G Gp55-P, viral membrane protein from murine spleen focus-forming virus ........................ 334, 340–342 GroEL pREP4-groELS electrocompetent BL21 (DE3) .................................................... 119, 121 GroES.......................................................115–117, 125, 127 Growth culture media deuterated minimal medium....... 184, 186, 188, 189, 193 Luria Bertani (LB) medium ...................... 119, 171, 314 minimal medium (M9) ........................ 3, 8, 9, 11, 12, 16, 115, 183–184, 188, 210, 268, 271, 310, 315 Gyromagnetic ratio................... 143, 158, 280–281, 289–290

H Hansenula polymorpha ........................................................ 38 1 H–13C-HMQC ...................................... 133–134, 136–138 1 H–13C-HSQC................................................ 173, 262, 270 Hemiascomycete yeast ....................................................... 21 Heteronuclear spectroscopy ............................................. 235 High cell-density expression.......................2, 5, 8, 10–16, 19 High cell-density induction ......................2–5, 10–11, 14–16 1 H–15N-HSQC ................................... 77, 78, 220, 228, 230, 236, 237, 247–249, 262–264, 266, 270, 343, 363 1 H 15N NOE ........................................................... 151, 174 Human apolipoprotein A-I (apoAI)............................ 13, 15 Hydrogen exchange electrostatics of .................................................. 369–400 kinetics of .................................................. 370, 374–378

I In-cell NMR ........................................... 125–126, 261–275 Inducible stable cell line of HEK293........................... 59, 63 INEPT.............................................. 150, 236, 237, 284, 413 Integral membrane proteins .............................................. 85 Intermolecular NOEs ...............................134, 198, 228–229 Intramolecular NOEs .............................................. 198, 240 In vitro synthesis.........................................86, 198, 201, 214

In vitro transcription ....................................... 198, 201, 214 Isolated Spin Pair Approximation (ISPA) ............... 460, 486 Isopropyl-β-D-thiogalactopyranoside (IPTG) induction ...........................2–5, 9–12, 14–16, 136 Isotope labeling double 13 C 15N labeling ......................... 1, 15, 20–32, 39, 43, 58–60, 78, 80, 88, 91, 94, 98, 173, 174, 197–198, 202, 212, 229, 235, 262, 263, 305, 312, 316, 319, 327–328, 343, 360, 407–408, 413 single 13 C labeling .....................1, 21, 38, 59, 138, 145, 168, 236, 262, 294, 363 2 H labeling ............................................... 28, 38, 172 methyl labeling ............................133, 134, 136–139, 236, 262–263, 270, 282, 294 15 N labeling .................................. 21, 52, 59, 76, 144, 168, 210, 236, 263, 324, 342, 363, 419, 485 perdeuteration ........................................................ 67 triple 2 H 13C 15N labeling................. 1, 16, 38, 72, 79, 120, 123, 144, 169–170, 172–175, 284, 363–365 ISPA. See Isolated Spin Pair Approximation

J J-couplings................................ 235–236, 455, 469–470, 479

L Ligand observed experiments .................................. 239–245 Ligation independent cloning (LIC) vector .................... 336 Liposomes, reconstitution into .............................. 94, 96–97 Longitudinal cross-correlated cross-relaxation rates ............................................... 143, 155–156 Longitudinal relaxation rate (T1) .....................146–149, 152, 157–158, 292, 321

M Magic angle spinning (MAS) ....................93, 105, 279–294, 304–307, 309–311, 320–323, 325–327, 341, 342, 454, 455, 458, 475, 477, 480 rotor..... ...............................................280, 325, 350, 353 Magnetization transfer .............................372, 378–379, 460 Maltose-binding protein (MBP) fusion protein.............. 115–118, 123–125, 337, 347, 352 MARS....... ...................................................... 417, 424, 425 MAS. See Magic angle spinning Mass spectrometry (MS) MALDI-TOF mass spectrometry ............. 338, 347, 349 Matlab....... ................310, 489, 490, 493, 494, 496, 502, 508 MBP. See Maltose-binding protein Membrane proteins .....................................39, 56, 59, 65, 67, 85–108, 165–167, 170, 176, 280, 293–294, 333–334, 336, 337, 339, 341 reconstitution ..................................................... 104, 108

PROTEIN NMR TECHNIQUES 516 Index Membrane solubilization hexafluoroisopropanol (HFIP) .................................. 353 Methylotrophic yeast ......................................................... 19 Microcrystalline proteins ..........................279, 285, 287, 454 Microdynamic parameters order parameters (S2) ................................................. 488 Sfast................................................... 2, 488, 492, 493, 503 TAUC4.. ................................. 92, 495, 497, 498, 500–504 τfast4........................................................ 88, 492, 493, 503 τloc4....... ................................. 87–489, 492, 493, 500, 503 τslow4...... ................................................................ 88, 492 Mitogen-activated protein (MAP) kinase ............... 359, 364 Model-free analysis .................................157–158, 292, 489, 497–498, 500, 507, 508 Models of local motion....................................488, 492–494, 500, 503, 504, 508 Molecular dynamics simulated annealing (MDSA) protocol .................................. 462, 464 MolProbity ...............................................456, 466, 473–475 Monomer-dimer equilibrium ................................. 490, 496, 500, 503, 505, 509 M2 proton channel .................................................. 167, 170 MS. See Mass spectrometry Multiple field data ........................................................... 489 Murine erythropoietin receptor (Epo receptor) ............... 334

N NCA experiment 3D DIPSHIFT-NCA experiment .................... 322–323 Network-anchored assignment/network anchoring (NA) .....................436, 444, 447, 454, 459, 470, 479 NMR assignment of/spectra of SecA..... ..................................................... 134, 137, 140 STP prolyl oligopeptidase (POP)-baicalin (flavonoid)...................................... 246, 250–256 vascular endothelial growth factor (VEGF)-P-7i (peptidic ligand) ............. 245–251 NMRPipe................................ 144, 148, 149, 169, 172–173, 307–308, 414, 417, 419 NMR sample preparation dsDNA-Int........................................................ 224, 227 solid-state NMR.................................. 72, 93, 94, 97, 99, 105–106, 116, 310–312 NMRView.. ............................. 255, 307–308, 413, 414, 417, 419–421, 425, 426, 432, 433, 476 Nuclear Overhauser effect (NOE) assignments ............. 137, 169, 172–174, 198, 362–363, 430, 431, 433–438, 443–447, 454, 459, 466 Nuclear Overhauser effect spectroscopy (NOESY ) 13 C-edited NOESY ................................... 173, 174, 177 13 C HMQC-NOESY-HMQC ................................. 137 13 C-separated NOESY ...................................... 170–171 3D [1H,1H]-NOESY-13C-HSQC............................. 433

3D [1H,1H]-NOESY-15N-HSQC ............................ 433 2D proton NOESY ................................................... 209 15 N[1H 1H] nuclear Overhauser effect.............. 433, 435, 441, 443 15 N-separated NOESY .............................. 170–171, 173 relaxation-compensated CPMG NOESY experiment ..................................................... 176 transferred NOESY ............................134, 138, 239, 240

O Oligomeric state characterization ............................ 100–101 Overall tumbling .....................................285, 287, 293, 488, 489, 491, 498, 499, 504, 505, 509

P 2

H Pake tensor ................................................................. 293 Paramagnetic relaxation enhancement (PRE) ..................... 134, 137–138, 140, 220, 366 PDSD sequence. See Proton-driven spin diffusion sequence Plasmids pACYCDuet ......................................118, 124, 126, 128 pET15b .............................. 182, 183, 185–188, 192, 193 pET vectors ...............7, 72, 182–188, 192, 193, 312, 316 pIVEX2 ....................................................................... 72 pKLAC........................................................................ 21 pMMHb.................................................................... 170 pPIC3.5K .................................................................... 20 pPIC9K ....................................................................... 20 pPICZ. .................................................................. 20, 21 pPICZα ..................................................................20, 21 pTriEx. ........................................................................ 46 pTYB1...........................................................................9 Poisson–Boltzmann .........................................369, 370, 376, 380, 382, 392, 396, 397 Polarized attenuated total reflection (ATR) Fourier transform infrared (FTIR) spectroscopy ........................................... 339–340 Polyhedra... ........................................................................ 40 Polyhedrin gene (polh)/promoter ................................. 40–41 PRE. See Paramagnetic relaxation enhancement PROCHECK ................................................. 456, 466, 473 ProSa......... ...................................................... 456, 466, 473 Protein dynamics ............................. 141–161, 238, 322, 485–509 flexibility ............................................................ 369–400 modularity ......................................................... 181–194 observed experiments................................. 235–239, 245 purification/preparation of CAP-Gly/microtubule complexes ....... 314, 319–320 detergent/protein micelles ....................... 94–96, 100 duplex DNA ........................................ 223, 224, 226 HIV-1 CA assemblies................................. 305, 307, 310–313, 316–318 replication protein A (RPA) ........................ 181–194 RNA ............................................................ 197–216

PROTEIN NMR TECHNIQUES 517 Index SrcCD ......................................... 115–118, 121–125 SrcCD from MBP fusion ........................... 115–117, 123–124, 127 ssDNA ......................................................... 225–226 thioredoxin reassemblies ..............305, 307, 309–312, 315–316 T7 RNA polymerase.................................. 74, 79–81 trpLE-M2 ................................................... 168, 171 tyrosine kinases .................................................. 111–114 Protein–DNA interactions ...................................... 262, 266 Protein–protein interactions ................................... 112, 125, 262, 266, 270, 274, 303, 367 Protein–RNA interactions ....................................... 197–216 Proteolytic separation of affinity/solubility tags cyanogen bromide...................................................... 336 enterokinase ............................................................... 336 factor Xa .................................................................... 336 thrombin .................................................................... 336 tobacco etch virus (TEV) protease ............................ 336 Proton-driven spin diffusion (PDSD) sequence .............. 321 Pseudoproline .................................................................. 335

Q Quantification of lipids in proteoliposomes ....................................... 104 of proteins in detergent.............................................. 103 of proteins in proteoliposomes ........................... 103–104 of proteoliposome density .................................. 104–105

R RDC. See Residual dipolar coupling Recombinant protein expression........................... 38, 40–41, 43–48, 335, 343 REDOR block ................................................................ 323 REDOR-HETCOR ........................305, 308, 309, 324–325 REDOR-PAINCP...................................305, 308, 309, 324 REDOR-PDSD .............................. 305, 308, 309, 324, 327 Relaxation 13 C relaxation ............................................................. 137 cross-correlated relaxation .........................142, 143, 145, 146, 153–156, 290–292 2 H relaxation .............................................................. 294 15 N CSA,15N-1H dipole cross-correlated relaxation rate ........................................ 290–291 15 N relaxation ..................................................... 141–161 Residual dipolar coupling (RDC) ............169, 171, 174–176, 373, 386, 388–390, 435, 455, 463, 470, 479 Restraint energy function flat-bottom harmonic-wall potential ......... 455, 464, 470 log-harmonic potential ............... 454, 463, 464, 470, 475 Reverse phase HPLC ...............................172, 229, 336, 352 RNA synthesis ..........................................198, 199, 201–205 RNA-templated RNA addition....................................... 201 RN-type recoupling block ............................................... 322

ROCSA CSA recoupling block ...................................... 322 Rotamer..... .............................. 173–174, 175–176, 384, 387, 395, 397–400 Rotational diffusion tensor ...................................... 506, 508 RotDif....... ...............................................492, 498, 505–508 Rubredoxin (Pyrococcus furiosus) ....................................... 372

S Saturation transfer difference (STD) amplification factor .................................... 243–244, 257 binding epitope .......................... 234, 235, 241, 243, 245, 253–254, 256 off-resonance .....................................241–242, 252, 253, 256, 327 on-resonance..............................152, 241–243, 251–253, 256, 414 Scalar coupling ( J) 2 H 13C scalar coupling............................................... 283 S30 cell extract (E. coli) preparation....................................................... 73, 77–79 SDS-PAGE protein expressed in liposomes................................... 100 protein expressed in micelles...................................... 100 Secretion of target proteins α-MF sequence ........................................................... 20 Saccharomyces cerevisiae α-mating factor (α-MF)......... 20 Selective labeling ............................................................. 236 Shigemi tube ....................................159, 254–255, 350, 351 Single nucleotide resolution ............................ 204, 205, 215 Solid-state NMR....................... 85–108, 279–275, 279–294, 303–328, 334, 342, 349, 353, 452–480 Solid state peptide synthesis ............................ 245, 334–335 Solvent accessibility ........................ 137, 370–374, 376, 380–384, 390, 391, 443 exchange ............................................................ 379–380 refinement ................................................................. 465 Spectral density ....................................................... 292, 487 Spectral density function, J(w).........................142, 290, 292, 486–488, 491 Spin label ATP-sl-N3-ATP ........................................................ 365 nitroxide spin-label .................................... 137–138, 365 2,2,5,5 tetramethyl 3-pyrroline scaffold............. 365–366 Src-family kinases.................................................... 111–128 Stable isotope labeling ..................................20, 43, 263, 268 Structure, determination/calculation of CPMGfit ........................................................... 169, 176 Pearson correlation coefficient ................................... 175 TALOS ..................................................... 169, 173, 477 Structure ensemble ........... 456, 462–463, 465, 466, 471, 473 Subculturing adapting to deuterated medium ................................... 33 Symmetry target function ................................................ 464

PROTEIN NMR TECHNIQUES 518 Index T TEM. See Transmission electron microscopy TOCSY..... ............................... 208, 226, 363, 364, 379, 435 TPPM decoupling ................................................... 320, 327 Transfection .................................... 46, 50, 51, 55–57, 59, 63 Transmembrane (TM) helix protein................................ 334 Transmission electron microscopy (TEM) CAP-Gly/microtubule assemblies ............. 317–318, 320 HIV-1 CA assemblies ....................................... 317–318 Transverse cross-correlated cross-relaxation rates .... 143, 155 Transverse relaxation optimized spectroscopy (TROSY ) methyl-TROSY ......................................... 134, 137, 139 15 N-edited NOESY-TROSY ............................. 362–364 15 N relaxation dispersion CPMG TROSY experiment ..................................................... 176 TROSY-HNCO ....................................................... 360 Transverse relaxation rate (T2) Carr-Purcell-Meiboom-Gill (CPMG) echo train ................ 143, 147, 150–151, 157, 158 single echo .......................... 143, 147, 149–150, 157–159 Triple resonance NMR experiments HA(CA)NH.............................................................. 267 HNCA .......................................172–173, 267, 286, 361 HNCACB ..........................................172–173, 286, 361 HNCACO ................................................................ 286 HNCO .............................. 145, 172–173, 267, 360–361, 365, 366, 413, 416, 424 HNCOCA ................................................................ 286 iHNCA ..................................................... 411, 413, 416 iHNCB.............................................................. 413, 416 iHNCO ..................................................... 413, 416, 424

T7 RNA polymerase preparation of .................................................. 74, 79–81 TROSY. See Transverse relaxation optimized spectroscopy

U Ubiquitin... ..............................................373, 374, 379, 380, 384–391 UNIO protocol ASCAN algorithm ............................ 433–436, 441–443 ATNOS algorithm ............................................ 435–438 CANDID algorithm ......................... 433–437, 443–446 MATCH algorithm............................433, 435, 438–441

V Violation analysis .....................................456, 461, 462, 466 Violation tolerance .................................................. 461, 463

W WATERGATE ............................................................... 247 Water-Ligand Observed via Gradient Spectroscopy (WaterLOGSY ) .................................... 240, 244 Water suppression ...................................159, 209, 242, 247, 251, 256, 281 WHAT IF. ....................................... 456, 466, 472, 473, 475

X Xplor-NIH ........................138, 169, 175, 432, 433, 437, 444

Z Zeocin.............................................................. 20–22, 27, 29