Cell-Free Protein Expression [1st ed.] 9781587061233, 1587061236, 2007040344

The role of cell-free rabbit reticulocyte expression systems in functional proteomics / Michele Arduengo, Elaine Schenbo

210 20 5MB

English Pages 240 Year 2007

Report DMCA / Copyright

DOWNLOAD PDF FILE

Recommend Papers

Cell-Free Protein Expression [1st ed.]
 9781587061233, 1587061236, 2007040344

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

BiotEChnology intElligEnCE unit

KudliCKi • KAtzEn • BEnnEtt

Biu

Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett

Cell-Free Protein Expression

Cell-Free Protein Expression

Biotechnology Intelligence Unit

Cell-Free Protein Expression Wieslaw A. Kudlicki, Ph.D. Federico Katzen, Ph.D. Robert P. Bennett, Ph.D. Invitrogen Corporation Carlsbad, California, U.S.A.

Landes Bioscience Austin, Texas U.S.A.

Cell-Free Protein Expression Biotechnology Intelligence Unit Landes Bioscience Copyright ©2007 Landes Bioscience All rights reserved. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Printed in the U.S.A. Please address all inquiries to the Publisher: Landes Bioscience, 1002 West Avenue, 2nd Floor, Austin, Texas 78701, U.S.A. Phone: 512/ 637 6050; Fax: 512/ 637 6079 www.landesbioscience.com ISBN: 978-1-58706-123-3 While the authors, editors and publisher believe that drug selection and dosage and the speciications and usage of equipment and devices, as set forth in this book, are in accord with current recommendations and practice at the time of publication, they make no warranty, expressed or implied, with respect to material described in this book. In view of the ongoing research, equipment development, changes in governmental regulations and the rapid accumulation of information relating to the biomedical sciences, the reader is urged to carefully review and evaluate the information provided herein.

Library of Congress Cataloging-in-Publication Data Cell-free protein expression / [edited by] Wieslaw A. Kudlicki, Federico Katzen, Robert P. Bennett. p. ; cm. -- (Biotechnology intelligence unit) Includes bibliographical references and index. ISBN-13: 978-1-58706-123-3 (alk. paper) ISBN-10: 1-58706-123-6 (alk. paper) 1. Proteins--Synthesis. 2. Genetic translation. I. Kudlicki, Wieslaw A. II. Katzen, Federico. III. Bennett, Robert P., Ph. D. IV. Series: Biotechnology intelligence unit (Unnumbered) [DNLM: 1. Cell-Free System--physiology. 2. Protein Biosynthesis. 3. Gene Expression Regulation. 4. Proteins--chemical synthesis. QU 350 C3935 2007] QP551.C4452 2007 572'.645--dc22 2007040344

About the Editors...

WIESLAW A. KUDLICKI, Ph.D., is a Research Fellow and a Group Leader for Cloning and Protein Expression, at Invitrogen Corporation. Wieslaw graduated with honors from the University of M. Curie Sklodowska, Poland in 1974. He earned his doctorate degree in 1981 and achieved habilitation in 1989. Ater completing his postdoctoral work in the Department of Chemistry and Biochemistry at the University of Texas at Austin, he returned to Poland, continuing his research as a scientist at his alma mater. In the early 1990s Dr. Kudlicki returned to the University of Texas at Austin where he worked to develop a highly eicient cell-free coupled transcription system from E. coli. Before joining Invitrogen to lead a Cloning and Protein expression team, Wieslaw held a Senior Scientist position at Ambion, Inc. in Austin Texas where he led the development of a variety of RNA and protein Expression technologies. Dr. Kudlicki has been the recipient of many honors and awards and has authored many papers in high quality, peer reviewed journals.

About the Editors...

FEDERICO KATZEN, Ph.D., is a Scientist at Invitrogen Corporation where he has been involved in the development of a variety of products ranging from cell-free protein expression to DNA recombination. Before joining Invitrogen, Federico held a postdoctoral research position at Harvard Medical School in Boston, where he uncovered the molecular basis of the transport of reduced thiol groups in proteins across the cytoplasmic membrane in bacteria. Federico earned his doctorate from the University of Buenos Aires in Argentina studying the biochemical and genetic aspects of polysaccharide biosynthesis in bacteria.

About the Editors...

ROBERT P. BENNETT, Ph.D., is Vice-President of Research and Development for the Molecular Biology Business at Invitrogen Corporation, a biotechnology company that develops technologies and services for basic and applied research. Rob has over 12 years of experience at Invitrogen in the successful product development and commercialization of molecular biology platforms. In particular, Rob led the development of a variety of cloning and protein expression technologies in prokaryotic and eukaryotic systems. Before joining Invitrogen, Rob earned his doctorate from Penn State Medical Center studying retroviral replication and particle assembly.

CONTENTS Preface.......................................................................................................xvii 1. he Role of Cell-Free Rabbit Reticulocyte Expression Systems in Functional Proteomics .............................................................................1 Michele Arduengo, Elaine Schenborn and Robin Hurst Membrane Topology ................................................................................................2 Glycosylation ..............................................................................................................3 Lipid Modiication and Acetylation of Proteins ................................................4 Reconstituting ER-Associated Protein Degradation ........................................5 In Vitro Viral Assembly ...........................................................................................6 Protein Microarray Technology .............................................................................6 Protein Interaction with Other Molecules..........................................................8 Display Technologies ................................................................................................9 Screening ...................................................................................................................11 2. Advances in Insect-Based Cell-Free Protein Expression ............................19 Uritza von Groll, Stefan Kubick, Helmut Merk, Wolfgang Stiege and Frank Schäfer Materials and Methods ......................................................................................... 20 Protein Expression ................................................................................................. 20 Protein Analysis ...................................................................................................... 20 Luciferase Activity Assay ...................................................................................... 20 Strep Tag Pull-Down Assay ................................................................................. 23 Protein Deglycosylation ........................................................................................ 23 Results and Discussion.......................................................................................... 23 Soluble Expression of Human Proteins............................................................. 24 Protein Function .....................................................................................................25 Posttranslationally Modiied Proteins .............................................................. 26 3. Advantages and Applications of the Batch-Formatted E. coli Cell-Free Expression System ......................................................................................31 Julia E. Fletcher, Federico Katzen, Shiranthi Keppetipola, Ashley Getbehead and Wieslaw A. Kudlicki Introduction .............................................................................................................31 High-Yield Formats ................................................................................................32 References..................................................................................................................39 4. Energetics in Escherichia coli-Based Batch Cell-Free Systems ...................42 Kalavathy Sitaraman and Deb K. Chatterjee History ...................................................................................................................... 42 Cell Extracts ............................................................................................................ 43 Components of E. coli-Based Cell-Free System .............................................. 43 Energy Sources for Cell-Free Protein Expression Systems............................ 43 Phosphoenolpyruvate ........................................................................................... 44 Pyruvate ................................................................................................................... 44 Glucose-6-Phosphate............................................................................................. 46

3-Phosphoglycerate ............................................................................................... 46 Glucose...................................................................................................................... 47 Other Phosphate Containing Secondary Energy Sources ............................ 47 Secondary Energy Source from Citric Acid Cycle .......................................... 47 Large-Scale Protein Production in Batch Reaction........................................ 49 5. Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression ......................................................................................53 Aaron R. Goerke and James R. Swartz 6. he PURE System: A Minimal Cell-Free Translation System...................76 Bei-Wen Ying, Yoshihiro Shimizu and Takuya Ueda Construction of the PURE System for Protein Generation ........................ 77 Reconstitution of the PURE System for Protein Maturation Studies....... 79 7. Cell-Free Expression Approaches for the Production and Characterization of Membrane Proteins ................................................................................84 Daniel Schwarz, Christian Klammt, Alexander Koglin, Florian Durst, Frank Löhr, Volker Dötsch and Frank Bernhard Why is Cell-Free Expression Exceptionally Suitable for the High-Level Production of Membrane Proteins?............................85 Cell-Free Produced Precipitates: A Fast and Reliable Approach for the Production of High Amounts of Pure Membrane Protein Samples.................................................................................................. 87 Cell-Free Expression into Detergent Micelles: A Completely New and Unique Mode of Membrane Protein Production............................... 88 Cell-Free Expression of Membrane Proteins into Preformed Artiicial Liposomes: A Challenging Perspective of Future Developments ...........91 Membrane Protein Targets Suitable for Cell-Free Expression .................... 92 hroughput Approaches and Optimization Strategies for the Production of MPs ..................................................................................... 93 Synergies of Cell-Free Expression and the Structural Analysis of Membrane Proteins by NMR Techniques .............................................. 94 8. Cell-Free Expression for Protein NMR....................................................101 A.J. Shaka A Bird’s Eye View of NMR Structure Determination................................. 101 Practical Limits for NMR Structure Determination .................................. 106 Cell-Free Protein Expression ............................................................................. 108 Future Prospects ................................................................................................... 114 9. Cell-Free Synthesis of Membrane Proteins for X-Ray Crystallography ..117 Julia E. Fletcher, Federico Katzen, Wieslaw Kudlicki, Yen-Ju Chen, Andrew Chen, Samatha Lieu and Geof rey Chang Results ..................................................................................................................... 118 Concluding Remarks ........................................................................................... 120 Materials and Methods ....................................................................................... 121

10. Bacterial Cell-Free Expression Systems for High-hroughput Protein Production ...................................................................................124 T.V.S. Murthy, Leonardo Brizuela and Joshua LaBaer Introduction to High-hroughput Protein Production .............................. 124 Cell-Free Systems for High-hroughput Protein Expression .................... 125 Bacterial Cell-Free System for High-hroughput Protein Production ... 126 Advances in Bacterial Cell-Free System .......................................................... 127 Cell-Free Protein Expression in Our Laboratory ......................................... 129 Limitations and Potential Solutions ................................................................ 129 11. Cell-Free Protein Synthesis for Protein Microarrays ...............................134 Gregory A. Michaud, Michael Salcius, Rebecca Martone, Diane Buhr, John C. Duarte, Jennifer E. McCague, Xiangdong Liu, Michael Samuels, Christine Stalder, James Ball, Alex Tikhonov, Shiranthi Keppetipola, Wieslaw Kudlicki, James Meegan and Barry I. Schweitzer Protein Content ....................................................................................................134 Yersinia Pestis Protein Arrays ............................................................................ 136 12. Cell-Free Protein Expression Screening and Protein Immobilization Using Protein Microarrays........................................................................145 Matthew A. Coleman, Paul Hoeprich, Peter Beernink and Julio A. Camarero Fluorescence-Based Protein Microarray Expression Screens ..................... 146 Array-Based Site-Speciic and Traceless Immobilization of Cell-Free Expressed Proteins .......................................................................................... 148 Materials and Methods ....................................................................................... 150 13. Cell-Free Protein Expression Labeling with Fluorophores......................155 Jerzy Olejnik Future Perspectives .............................................................................................. 162 14. Cell-Free Synthesis of Deined Protein Conjugates by Site-Directed Cotranslational Labeling ..........................................................................166 Michael Gerrits, Jan Strey, Iris Claußnitzer, Uritza von Groll, Frank Schäfer, Martina Rimmele and Wolfgang Stiege Protein Conjugates .............................................................................................. 166 Cell-Free Systems as a Tool for Protein Conjugate Production ................ 167 Incorporation Eiciencies of Diferent Unnatural Amino Acids ............. 169 Depletion of RF1 from a Cell-Free System Results in Eicient Incorporation of Biocytin ............................................................................. 169 Applications of Site-Speciically and Stoichiometrically Deined Protein Conjugates...........................................................................................174

15.

C-Terminal Labeling of Proteins Using Fluorescently Conjugated Puromycin Derivatives..............................................................................181 Ichiro Tabuchi he Principle of C-Terminal Fluorescence Labeling Using Puromycin Derivatives ................................................................................... 181 Advantages of C-Terminal Fluor-Puro Labeling Technology ................... 182 Successful Fluor-Puro Labeling Applications ................................................ 183 Fluor-Puro Conjugates: Labeling Eiciency and Choice of Fluorophore .................................................................................................. 183 he Choice of he Translation System ............................................................ 185 Concentration of Fluor-Puro............................................................................. 185 Advanced Applications for Fluor-Puro Labeling .......................................... 185

16. Translation Engineering and Synthetic Biology ......................................190 David A. Roth, Liza S.Z. Larsen and G. Wesley Hatield Translation Engineering™—Codon Context and Translational “Pausing”.......................................................................... 190 he CODA Translational Engineering Toolbox .......................................... 192 Summary and Applications................................................................................ 196 17. Accelerated Protein Evolution Using Ribosome Display.........................198 Julie Douthwaite, Lutz Jermutus, Ronald Jackson Introduction to in Vitro Display Technology................................................ 198 he Ribosome Display Process .......................................................................... 199 Cell-Free Translation for Ribosome Display .................................................200 In Vitro Evolution by Ribosome Display ........................................................200 Evolution of High Ainity Antibodies ........................................................... 203 Evolution of Improved Proteins ........................................................................205 18. Application of in Vitro Virus (IVV) Technique for High-hroughput Analysis of Protein-Protein Interactions..................................................208 Etsuko Miyamoto-Sato and Hiroshi Yanagawa Introduction ..........................................................................................................208 IVV: Genotype-Phenotype Assignment Molecules Formed via Puromycin...................................................................................................209 Applications of IVV as in Vitro Display Technology .................................. 210 Protein Complexes Analysis; a Co-Translation Technique ........................ 211 Reliability of IVV Data; the Problem of False Positives and Negatives ................................................................................................... 211 In Silico Analysis of IVV Selection for AP-1 (Fos/Jun) Complex............ 214 Outlook ...................................................................................................................215 Index .........................................................................................................219

EDITORS Wieslaw A. Kudlicki Invitrogen Corporation Carlsbad, California, U.S.A. Email: [email protected] Chapters 3, 9, 11

Federico Katzen Invitrogen Corporation Carlsbad, California, U.S.A. Email: [email protected] Chapters 3, 9

Robert P. Bennett Invitrogen Corporation Carlsbad, California, U.S.A. Email: [email protected]

CONTRIBUTORS Michele Arduengo Research and Development Promega Corporation Madison, Wisconsin, U.S.A.

Leonardo Brizuela Harvard Institute of Proteomics Cambridge, Massachusetts, U.S.A. Chapter 10

Chapter 1

James Ball Protein Array Center Branford, Connecticut, U.S.A.

Diane Buhr Protein Array Center Branford, Connecticut, U.S.A. Chapter 11

Chapter 11

Peter Beernink Children’s Hospital Oakland Research Institute Oakland, California, U.S.A. Chapter 12

Julio A. Camarero Chemistry and Material Life Sciences Directorate Lawrence Livermore National Laboratory Livermore, California, U.S.A. Email: [email protected] Chapter 12

Frank Bernhard Centre for Biomolecular Magnetic Resonance Institute for Biophysical Chemistry University of Frankfurt/Main Frankfurt/Main, Germany Email: [email protected] Chapter 7

Geofrey Chang he Scripps Research Institute La Jolla, California, U.S.A. Email: [email protected] Chapter 9

Deb K. Chatterjee Protein Expression Laboratory Research Technology Program SAIC-Frederick, Inc. NCI-Frederick Frederick, Maryland, U.S.A. Email: [email protected] Chapter 4

Andrew Chen he Scripps Research Institute La Jolla, California, U.S.A.

John C. Duarte Protein Array Center Branford, Connecticut, U.S.A. Chapter 11

Florian Durst Centre for Biomolecular Magnetic Resonance Institute for Biophysical Chemistry University of Frankfurt/Main Frankfurt/Main, Germany Chapter 7

Chapter 9

Yen-Ju Chen he Scripps Research Institute La Jolla, California, U.S.A.

Julia E. Fletcher Invitrogen Corporation Carlsbad, California, U.S.A. Chapter 3,9

Chapter 9

Iris Claußnitzer RiNA Netzwerk RNA Technologien GmbH Berlin, Germany

Michael Gerrits RiNA Netzwerk RNA Technologien GmbH Berlin, Germany Chapter 14

Chapter 14

Matthew A. Coleman Chemistry and Material Life Sciences Directorate Lawrence Livermore National Laboratory Livermore, California, U.S.A. Email: [email protected] Chapter 12

Volker Dötsch Centre for Biomolecular Magnetic Resonance Institute for Biophysical Chemistry University of Frankfurt/Main Frankfurt/Main, Germany Chapter 7

Julie Douthwaite Cambridge Antibody Technology Granta Park Cambridge, U.K. Email: julie.douthwaite @cambridgeantibody.com Chapter 17

Ashley Getbehead Invitrogen Corporation Carlsbad, California, U.S.A. Chapter 3

Aaron R. Goerke Department of Chemical Engineering Stanford University Stanford, California, U.S.A. Chapter 5

G. Wesley Hatield CODA Genomics, Inc. Laguna Hills, California, U.S.A. and University of California Irvine, California, U.S.A. Email: [email protected] Chapter 16

Paul Hoeprich Chemistry and Material Life Sciences Directorate Lawrence Livermore National Laboratory Livermore, California, U.S.A. Chapter 12

Robin Hurst Research and Development Promega Corporation Madison, Wisconsin, U.S.A. Email: [email protected]

Stefan Kubick Qiagen GmbH Hilden, Germany Chapter 2

Joshua LaBaer Harvard University Cambridge, Massachusetts, U.S.A. Chapter 10

Ronald Jackson Cambridge Antibody Technology Cambridge, U.K.

Liza S. Z. Larsen he Institute for Genomics and Bioinformatics he Donald Bren School of Information and Computer Sciences University of California Irvine, California, U.S.A.

Chapter 17

Chapter 16

Lutz Jermutus Cambridge Antibody Technology Cambridge, U.K.

Samatha Lieu he Scripps Research Institute La Jolla, California, U.S.A

Chapter 17

Chapter 9

Shiranthi Keppetipola Invitrogen Corporation Carlsbad, California, U.S.A.

Xiangdong Liu Protein Array Center Branford, Connecticut, U.S.A.

Chapter 3,11

Chapter 7

Christian Klammt Centre for Biomolecular Magnetic Resonance Institute for Biophysical Chemistry University of Frankfurt/Main Frankfurt/Main, Germany

Frank Löhr Centre for Biomolecular Magnetic Resonance Institute for Biophysical Chemistry University of Frankfurt/Main Frankfurt/Main, Germany

Chapter 7

Chapter 7

Alexander Koglin Centre for Biomolecular Magnetic Resonance Institute for Biophysical Chemistry University of Frankfurt/Main Frankfurt/Main, Germany

Rebecca Martone Protein Array Center Branford, Connecticut, U.S.A.

Chapter 1

Chapter 7

Chapter 11

Jennifer E. McCague Protein Array Center Branford, Connecticut, U.S.A. Chapter 11

James Meegan Protein Array Center Branford, Connecticut, U.S.A.

Michael Salcius Protein Array Center Branford, Connecticut, U.S.A.

Chapter 11

Chapter 11

Helmut Merk Qiagen GmbH Hilden, Germany

Michael Samuels Protein Array Center Branford, Connecticut, U.S.A.

Chapter 2

Chapter 11

Gregory A. Michaud Invitrogen Corporation Branford, Connecticut, U.S.A. Email: [email protected]

Frank Schäfer Qiagen GmbH Hilden, Germany Email: [email protected]

Chapter 11

Chapters 2 and 14

Etsuko Miyamoto-Sato Department of Biosciences and Informatics Keio University Yokohama, Japan

Elaine Schenborn Research and Development Promega Corporation Madison, Wisconsin, U.S.A. Chapter 1

Chapter 18

T.V.S. Murthy Harvard University Cambridge, Massachusetts, U.S.A. Email: [email protected] Chapter 10

Daniel Schwarz Centre for Biomolecular Magnetic Resonance Institute for Biophysical Chemistry University of Frankfurt/Main Frankfurt/Main, Germany Chapter 7

Jerzy Olejnik Bio-Comm, LLC Boston, Massachusetts, U.S.A. Email: [email protected]

Barry I. Schweitzer Protein Array Center Branford, Connecticut, U.S.A.

Chapter 13

Chapter 11

Martina Rimmele RiNA Netzwerk RNA Technologien GmbH Berlin, Germany

A. J. Shaka Chemistry Department University of California Irvine, California, U.S.A. Email: [email protected]

Chapter 14

Chapter 8

David A. Roth CODA Genomics, Inc. Laguna Hills, California, U.S.A. Email: [email protected] Chapter 16

Yoshihiro Shimizu Department of Medical Genome Sciences Graduate School of Frontier Sciences he University of Tokyo Kashiwa, Chiba, Japan Chapter 6

Kalavathy Sitaraman Protein Expression Laboratory Research Technology Program SAIC-Frederick, Inc. NCI-Frederick Frederick, Maryland, U.S.A.

Alex Tikhonov Protein Array Center Branford, Connecticut, U.S.A. Chapter 11

Christine Stalder Protein Array Center Branford, Connecticut, U.S.A.

Takuya Ueda Department of Medical Genome Sciences Graduate School of Frontier Sciences he University of Tokyo Kashiwa, Chiba, Japan Email: [email protected]

Chapter 11

Chapter 6

Wolfgang Stiege RiNA Netzwerk RNA Technologien GmbH Berlin, Germany Email: [email protected]

Uritza von Groll Qiagen GmbH Hilden, Germany Email: [email protected]

Chapter 4

Chapters 2 and 14

Chapters 2 and 14

Jan Strey RiNA Netzwerk RNA Technologien GmbH Berlin, Germany Chapter 14

Hiroshi Yanagawa Department of Biosciences and Informatics Keio University Yokohama, Japan Email: [email protected] Chapter 18

James R. Swartz Department of Chemical Engineering Stanford University Stanford, California, U.S.A. Email: [email protected] Chapter 5

Bei-Wen Ying Department of Medical Genome Sciences Graduate School of Frontier Sciences he University of Tokyo Kashiwa, Chiba, Japan Chapter 6

Ichiro Tabuchi Tokyo Evolution Research Center Tokyo, Japan Email: [email protected] Chapter 15

PREFACE Cell-Free Protein Expression: Old Challenges and New Opportunities Since its inception in the 1950s, cell-free protein synthesis has had a tremendous impact on basic life sciences. he use of cell-free systems was key to the understanding the molecular mechanisms underlying one of the most complicated processes found in nature: protein translation. Fity years ago, it would have been unimaginable that the same principle that would enable Marshall Nirenberg and Heinrich Matthaei to crack the genetic code would enable the synthesis of milligram quantities of proteins in a microfuge tube. It took a few decades before cell-free protein expression became broadly accepted in protein science. Aggressive cutting-edge research and stiff commerical competition drove the development of a variety of systems with increased productivity, improved protein quality and relatively low production costs. Incidentally, the increasing popularity and capabilities of this technology, largely driven by the explosion of genomic information, has generated an expanding market illed with a variety of protein synthesis kits and related products. he implications of these advances in cell-free protein on the advances on cell-free protein synthesis go well beyond the mere in vitro reaction. he technology has generated myriad applications that have enabled advances in ields as diverse as systems biology, structural biology, and drug discovery. Many of these applications are described and expanded upon in this book. We have divided this volume into six main sections. In the irst section, many of the most popular sources of cell-free lysates are introduced. he second section focuses on extraordinary advances made in the Escherichia coli-based systems that enabled reconstituting the entire translational process, incorporating post-translational modiications, increasing yield, and producing functional membrane proteins. hese extend the usefulness of cell-free systems into structural biology applications described in the third section and highcontent platforms like protein microarrays discussed in the fourth section. Finally, we include two sections covering the use of cell-free protein expression technologies in the rational design and directed evolution of proteins. Last but not least, we are indebted to all the authors who contributed to this book. A common theme which arises from their work is that as powerful as cell-free protein expression systems are today, the technologies and applications are at the beginning of a rapid evolutionary process, making what used to be impossible, possible. It's a good time to be a protein scientist. Wieslaw A. Kudlicki Federico Katzen Robert P. Bennett

Chapter 1

he Role of Cell-Free Rabbit Reticulocyte Expression Systems in Functional Proteomics Michele Arduengo, Elaine Schenborn and Robin Hurst*

Abstract

I

n the broad area of functional proteomics, that is the global characterization of proteins and their function, cell-free rabbit reticulocyte lysate (RRL) has been used extensively to elucidate the mechanisms of mammalian translation, cotranslational modiications, post-translational modiications and translocation of proteins. More recently, RRL has been used as the workhorse for manufacturing the proteins engaged in interaction, selection and protein evolution studies from DNA or mRNA libraries either in microarray, display or in vitro expression cloning (IVEC) technologies. his chapter highlights recent functional proteomics applications that use cell-free mammalian RRL.

Abbreviations

CMM, canine microsomal membrane; ER, endoplasmic reticulum; ERAD, endoplasmic reticulum-associated degradation; GPI, glycosylphosphatidylinositol; IVC, in vitro compartmentalization; IVEC, in vitro expression cloning; PCR, polymerase chain reaction; PTT, protein truncation test; RRL, rabbit reticulocyte lysate; RT-PCR, reverse transcription-polymerase chain reaction; SP cells, semipermeabilized cells. HighWire Press is a registered trademark of Stanford University. TnT is a registered trademark of Promega Corporation.

Introduction

In late 1950s and early 1960s researchers irst demonstrated that radioactive amino acids could be incorporated into hemoglobin in cell-free rabbit reticulocyte lysate (RRL).1,2 Since then RRL has been used to elucidate the highly complex events that encompass translation, from initiation to termination.3-5 RRL has also proven useful in understanding cotranslational folding of nascent polypeptide chains, protein targeting and post-translational folding. During the 1970s, researchers showed that RRL could be manipulated for exogenously directed mRNA protein synthesis, so that only a single protein of interest was synthesized.6 In the 1990s, the development of coupled transcription/translation, in which RRL is supplemented with T7, T3, or SP6 RNA polymerases further simpliied the expression of protein targets.7,8 DNA-directed protein synthesis in RRL has some advantages over mRNA-primed RRL including the elimination of mRNA handling, and it usually achieves higher levels of protein synthesis. An advantage of cell-free RRL over other cell-free systems (E. coli or wheat germ extracts) is that the mammalian environment more *Corresponding Author: Robin Hurst—Research and Development, Promega Corporation, 2800 Woods Hollow Rd., Madison, WI, U.S.A. 53711. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

2

Cell-Free Protein Expression

closely mimics human cells. Cell-free RRL is generated from lysed reticulocytes isolated from phenylhydrazine-treated rabbits. he lysed reticulocytes are treated with microccocal nuclease to remove endogenous mRNA. he RRL is optimized and supplemented with components that give optimal translation when priming with mRNA and, in the case of a coupled system, optimal translation/transcription for priming with DNA that contains the appropriate RNA phage polymerase promoter sequences. Cell-free RRL has the same advantages as all cell-free systems over cell-based expression systems: substantial time-savings (two hours versus 24–48 hours for protein expression), the ability to adapt to high-throughput formats, increased tolerance to additives and less sensitivity to toxic or proteolytic proteins. he current use of cell-free RRL is substantial as illustrated by a HighWire Press® search covering January 2000 to April 2006 that yielded more than 3,000 articles containing the phrase “rabbit reticulocyte.” his chapter will focus on recent applications that use RRL for cell-free functional proteomics.

Membrane Topology Synthesis of Membrane Proteins

Membrane protein topology is oten described based on the predicted amino acid sequence and algorithms that estimate hydrophobicity and probable secondary structure of a stretch of amino acids.9 However, diferent algorithms or diferent stringencies applied to the same algorithm can predict diferent topologies, and many algorithms fail to account for cotranslational processing events or the efects of post-translational modiications on protein topology.9 Translation in a cell-free system containing microsomes or semipermeabilized cells can provide empirical data about membrane protein topology. Using a prepared lysate supplemented with endoplasmic reticulum (ER)-derived microsomal membranes such as canine pancreatic microsomal membranes (CMMs)10-12 or digitonin-permeabilized cells (semipermeabilized cells),13 membrane proteins can be successfully synthesized, translocated and modiied in vitro. Semipermeabilized (SP) cells have some advantages over ER-derived microsomes. SP cells are more likely to contain the necessary components for the correct folding and modiication of proteins normally expressed by the cells, and they have a spatially intact ER and Golgi system that better approximates the cellular environment.13 Additionally, SP cells can be more eicient at specialized modiications such as adding glycosylphosphatidylinositol (GPI) anchors.14 Using SP cells and RRL without dithiothreitol allows disulide bond formation and eicient folding of proteins such as MHC class I heavy chain.15 Proteins can be associated with the membrane in a variety of ways. Integral membrane proteins span both lealets of the phospholipid bilayer with one or more alpha helical transmembrane domains consisting of approximately 20 hydrophobic amino acids. Peripheral membrane proteins can be associated with a single membrane lealet either by means of a fatty acid modiication, such as prenylation, myristoylation, or a GPI anchor, or by association of a predominantly hydrophobic stretch of amino acids. Cell-free translation systems such as RRL supplemented with microsomes or SP cells can provide information about how a particular protein associates with a membrane. For instance, treating isolated microsomes with sodium carbonate dissociates peripheral but not integral membrane proteins from microsomes or semipermeabilized cells.16 Such treatment has been used to show that the glycoprotein GP4 produced by the equine arteritis virus is an integral membrane protein, while the GP3 protein, produced by the same virus, is membrane-anchored.17

Protease Protection Assays

Protease protection assays are oten used to help determine membrane protein topology and orientation. Water-soluble proteases cannot freely cross the lipid bilayer of microsomes, so segments of proteins that are in the lumen of the microsomes will not be subject to protease digestion unless the microsomes are irst permeabilized. Assuming that the lumen of microsomes represents the

he Role of Cell-Free Rabbit Reticulocyte Expression Systems in Functional Proteomics

3

lumen of the ER, the segments of proteins in the lumen of the microsomes become extracellular once a protein is inserted into the plasma membrane. Several studies have used protease protection assays to determine the topology of membrane proteins, including determining which of two proposed topologies is correct for cytochrome b5.18 In this study, failure of carboxypeptidase Y to remove C-terminal labeled methionines of cytochrome b5 suggests that the C-terminus is inside the lumen of the ER and inaccessible to the protease.18 Oten, the protease assay is performed on proteins translated in the presence of microsomes or SP cells. he microsomes are incubated with the protease with or without concomitant detergent solubilization, and protease fragments are compared between the solubilized or unsolubilized samples. Using a proteinase K protection assay of proteins translated in RRL supplemented with CMMs, Umigai et al19 showed that the M2 domain of the K+ channel Kir 2.1 is oriented with the C-terminus toward the cytoplasm. A similar assay has been used to explore the efect of pathogenic mutations on the prion (PrP) protein.14 In addition to determining topology of a particular protein, researchers can use protease assays to dissect the process of ER binding and translocation.20

Tagging Membrane Proteins to Determine Orientation

Another strategy for determining membrane topology involves tagging a protein with an enzyme or epitope. Tagging is oten used in conjunction with other studies such as glycosylation or protease assays to give a more complete picture of membrane topology. In one study, the C-terminal end of each of several deletion mutants of Presenilin I was tagged with E. coli leader peptidase (LPase). Anti-LPase antibody was able to immunoprecipitate in vitro translated protein from some but not all of the deletion mutants based on the location of the C-terminus of the protein (either cytosolic or luminal).21 C-terminal and N-terminal glycosylation tags have also been used in experiments to investigate the topology of vitamin K epoxide reductase.22

Glycosylation N-Linked Glycosylation of Membrane and Secreted Proteins

Glycosylation studies in RRL supplemented with CMMs or SP cells can provide information about membrane topology. Portions of proteins translocated into the lumen of microsomes or SP cells are exposed to enzymes responsible for core N-linked glycosylation. N-linked glycosylation acceptor sites can be inserted into the protein, at the N- or C-terminus, for example. Any sugar residues that are added should be removed by the actions of glycosidases, such as endoglycosidase H, when microsomes containing the proteins are treated with detergents to allow the glycosidase access to the luminal portion of the protein. Such a strategy was used to determine the membrane topology of vitamin K epoxide reductase.22 Zhang and Ling used sensitivity to peptide N-glycosidase (PNGaseF) to determine whether an 18 kDa protease-protected fragment from mouse P-glycoprotein is glycosylated.23 Additionally, carrying out translation in RRL supplemented with microsomes in the presence or absence of tunicamycin (a glycosylation inhibitor) can allow comparison of glycosylated and unglycosylated forms of proteins produced in vitro.24 he membrane topology of polytopic proteins (proteins that span the membrane multiple times) can be especially diicult to predict. he Presenilin-1 protein is a polytopic protein predicted by hydrophobicity analysis to span the membrane from six to eight times. To determine membrane topology for Presenilin-1, a series of C-terminal deletions was made to remove predicted transmembrane regions of the protein. he truncated proteins were translated in vitro in the presence of microsomes. Endoglycosidase H sensitivity of protein from solubilized microsomes changed as deletions of the transmembrane domains altered the orientation of the protein in the membrane.21 Combined with protease protection assays and epitope tag labeling at the C-terminus, these glycosylation results supported a seven-transmembrane domain structure with an additional membrane-embedded domain for Presenilin-1.

4

Cell-Free Protein Expression

O-linked Glycosylation of Cytosolic and Nuclear Proteins

Secretory and membrane proteins are not the only proteins in the cell that are glycosylated. Many nuclear and cytosolic proteins are modiied by O-linked glycosylation (O-GlcNAcylation).25 RRL contains the enzymes and substrates necessary for the O-GlcNAcylation of proteins,26 and addition of microsomes or SP cells provides the environment necessary for correct membraneprotein folding. To assess whether the insulin-responsive glucose transporter GLUT4 undergoes O-GlcNAcylation, GLUT4 cDNA was transcribed and translated in RRL supplemented with CMMs.25 RRL was able to successfully modify GLUT4 protein.25

Lipid Modiication and Acetylation of Proteins Glycosylphosphatidyl Inositol (GPI) Anchors

Some proteins are anchored to the cell membrane by means of a glycosylphosphatidyl inositol (GPI) modiication at the C-terminus.27 Unlike proteins incorporated into the membrane by a transmembrane domain, proteins that are GPI-anchored are reversibly associated with the lipid bilayer. GPI anchoring does not appear to be necessary for cell survival, but it is necessary for development.28 Proteins that are modiied by the addition of a GPI anchor contain two signal sequences, one at the N-terminus that directs protein synthesis to the ER and a second at the C-terminus that directs the addition of the GPI-anchor by a transamidase activity.29 Small nucleophilic compounds like hydrazine can substitute for GPI, providing a means to assess whether GPI modiication has occurred by comparing the molecular weight of proteins translated in the presence or absence of hydrazine.30 Human placental alkaline phosphatase (PLAP) is a GPI-anchored protein.24 GPI-anchored mini-PLAP has been generated by numerous groups using nuclease-treated RRL supplemented with microsomal membranes from CHO, F9, EL4 or K562 cells,24,28,29,31-33 demonstrating that GPI modiication can be reconstituted in a cell-free system.

Prenylation

Some proteins are modiied by prenylation, the attachment of one or more isoprenoid groups, such as the 15-carbon farnesyl group or the 20-carbon geranylgeranyl group, to a cysteine residue. Prenylation can mediate membrane association of some proteins, particularly the Ras-like GTPases, and protein-protein interactions (e.g., nuclear lamins).34 Prenylated proteins can be produced and detected in RRL supplemented with the labeled isoprenoid precursor mevalonic acid ater the translation reaction is complete; additionally proteins synthesized in RRL can be modiied using photoactivatable analogs of isoprenoids.35-37 Gel-based assays to detect changes in protein migration as a result of prenylation are also used, but these are indirect assays and are usually performed along with a labeling experiment. Most prenylation assays require autoradiography of the labeled lysate, which can take weeks or months. Benetka and colleagues have developed an in vitro prenylation assay using N-terminal GST-tagged proteins and detection of 3H-labeled precursors using a TLC linear analyzer.38 he incorporation of the GST tag allows the labeled protein of interest to be separated from free radioactive label and other proteins in the RRL, and using the TLC scanner to detect the incorporated lipid molecule signiicantly reduces the time required to obtain results.

N-Myristoylation and Palmitoylation

Many proteins in eukaryotic cells are subject to N-myristoylation, the addition of a 14carbon fatty acid to the N-terminus.39 In addition to being prenylated, the alpha subunits of some G-proteins, including pp60src and p21ras, are myristoylated. Other G-protein alpha subunits, including some of the Gs and Gq subunits, are not myristoylated, but are instead modiied by the attachment of a 16-carbon palmitic acid (palmitoylation), and others are both myristoylated and palmitoylated. Some studies suggest that N-myristolyation and palmitoylation contribute to the membrane association of these proteins.40-42 N-myristoylation can occur cotranslationally, immediately ater the removal of the N-terminal methionine from a protein or even post-translationally as with the protein BID (BCL-2 interacting domain), a substrate of caspase-8.39,43 Upon cleavage

he Role of Cell-Free Rabbit Reticulocyte Expression Systems in Functional Proteomics

5

by caspase-8, BID reveals an N-myristoylation site. RRL contains the components necessary to complete N-myristoylation, and has even been used to myristoylate tumor necrosis factor, a normally nonmyristoylated protein, when it is modiied to contain the N-myristolyation motif of other myristoylated proteins.44,45 he Arabidopsis SOS3 protein involved in plant salt tolerance has also been synthesized and myristoylated in an RRL system.46

N-Acetylation

A majority (70–85%) of the proteins found in the cytoplasm or nucleus of eukaryotes may be modiied by N-acetylation.47,48 Examples of N-acetylated proteins include ovalbumin, actin and cytochrome c. Acetylation is catalyzed by N-acetyltransferases cotranslationally ater the initiator methionine has been cleaved.48 Many proteins are also acetylated post-translationally at internal sites by acetyltransferase enzymes diferent from those involved in cotranslational modiication.48 Acetylation of modiied tumor necrosis factor (TNF) has been demonstrated in RRL,47 and acetylation in RRL can be inhibited by the use of S-acetonyl-CoA, an analog of acetyl-CoA.49

Reconstituting ER-Associated Protein Degradation

Conditions such as environmental stress, viral infection and the absence of required partner proteins can result in the accumulation of aberrantly folded proteins in the rough endoplasmic reticulum (RER). he RER has a “quality control” system that targets these misfolded proteins for degradation. his process, ER-Associated Degradation (ERAD), requires ATP and is distinct from the lysosomal degradation pathway in cells.50 RRL in conjunction with CMMs or SP cells has been used to reconstitute ERAD activity.15, 51 RRL has several advantages over intact-cell systems for such study. he protein of interest will be the only protein labeled in the RRL, making its degradation easy to follow.51 Second, a variety of microsomal membranes can be used with RRL to reconstitute the activity.15,51 Because hemin inhibits the proteasome activity of ERAD, degradation studies are best performed in RRL that does not contain exogenous hemin. Researchers have reported that RRL that works well for studies of degradation is usually poor for translation.51 Additionally, since ERAD is ATP-dependent, the RRL will need to be supplemented with ATP and an ATP regeneration system,51 and some authors report that excess unlabeled methionine seems to aid in reconstituting ERAD activity.52 RRL-based protein degradation systems have been used to investigate the synthesis, stability and degradation of a variety of wild type and mutant proteins. In one such study, tyrosinase ERAD was reconstituted in a commercially available RRL system supplemented with an ATP regeneration system.53 Wild type and mutant tyrosinase associated with albinism were translated in the presence of SP melanocytes; the SP cells were isolated and then resuspended in RRL with the ATP regeneration system. Both proteins were degraded, although the mutant protein degraded at a higher rate than the wild type protein. RRL-based and rat cytosol-based degradation systems have also been used to investigate the degradation of Apoprotein B (apoB).54 Aliquots of the transcription/translation reaction of HA-tagged apoB were incubated in the presence of an ATP regeneration system in fresh RRL or rat hepatocyte cytosol with or without proteasome inhibitors. Inhibition of apoB degradation was more obvious in the rat hepatocyte cytosol, presumably because RRL contains factors that interfere with the proteasome inhibitors. To assess the role of the chaperone protein, hsp90, in the degradation of apoB48, geldanamycin (GA), an antibiotic that competes for the ATP binding site on hsp90, was added to pelleted CMMs before the degradation assay. here was no signiicant decrease in the amount of apoB48 in the presence of GA, indicating that GA did inhibit degradation and hsp90 was required for apoB48 degradation.54 RRL has been used to reconstitute the degradation of α1-antitrypsin Z [(α1AT)Z].52 Individuals who are homozygous recessive for a mutation resulting in a Glu342 to Lys substitution have increased susceptibility to liver disease.52 he amino acid substitution disrupts proper folding of (α1AT)Z, and individuals susceptible to the liver disease are not able to degrade the misfolded protein eiciently. Mutant and wild type (α1AT)Z degradation were examined using an RRL degradation assay system. he mutant (α1AT)Z was degraded eiciently. Mutant protein produced in the presence

6

Cell-Free Protein Expression

of salt-washed or puromycin-treated, salt-washed microsomes was also degraded, indicating that the full complement of RER proteins was not required for degradation.52 Several mechanisms have been suggested to control the targeting of proteins accumulated in the ER for degradation, including regulating the trimming of N-linked oligosaccharide chains. Oligosaccharide side chains can be modiied by mannosidase I in the ER. Inhibiting this activity seems to stabilize misfolded proteins.15 Wild type MHC class I heavy chain and a mutant heavy chain that lacks the N-linked glycosylation site but that can assemble into functional MHC class I molecules were translated in RRL in the presence of SP HT1080 cells.15 he wild type protein was degraded more quickly than the mutant, indicating that glycosylation is important for ERAD.

In Vitro Viral Assembly

he ability to reconstitute cotranslational assembly events and protein interactions using cell-free RRL systems can be extended to the study of in vitro mammalian viral protein assembly, viral protein interactions with other cellular components, and viral protein efects on translation. Early studies of viral protein assembly demonstrated that capsid proteins expressed in RRL were capable of self-assembling in vitro. For example, adenovirus type 2 iber protein synthesized in RRL formed trimers without requiring additional viral proteins or components.55 From this relatively well-deined, single-protein model, more complex viral protein interactions and assembly studies evolved. Human papillomavirus-like particles have been assembled in vitro from L1 capsid protein expression in RRL.56 hese particles also mimicked endogenous virus in its conformational epitope exposure, and antibodies generated against the in vitro-assembled L1 particles were efective in recognizing similar epitopes in patient samples. In addition to human papillomavirus, human hepatitis C (HCV) core viral capsid precursor structures were also generated de novo in reticulocyte lysates.57 Cell-free systems have allowed detailed examination of HCV core capsid assembly processes and properties, whereas mammalian culture systems have been limited by low viral titers. In vitro expression of Gag precursor proteins in RRL has allowed detailed examination of the more complex pathway for retroviral assembly, which includes not only protein interactions, but also plasma membrane interactions and budding. Viral capsid structures of immunodeiciency virus type 1 (HIV-1) have been assembled from the Gag precursor protein p55gag expressed in RRL, and these particles resembled immature HIV-1 viral structures.58 Additional processing of proteins in viral capsids can include prenylation and glycosylation. Myristoylation of HIV-1 viral particles59 and glycosylation of woodchuck hepatitis virus capsid proteins60 have been investigated using cell-free RRL systems. hese results illustrate the importance of cell-free protein expression in delineating processes involved in viral particle assembly pathways.

Protein Microarray Technology

In the ield of functional proteomics, protein microarrays are illing a niche for miniaturization coupled with high-throughput assay capability.61,62 Protein microarray concepts are patterned ater DNA microarrays, but immobilization of diverse types of proteins in a manner that preserves conformation and functionality is a complex and challenging problem to solve. Continuing advances in microarray surface chemistries coupled with improvements in protein production capabilities, sensitive detection methods, and instrumentation are accelerating the pace of development for protein microarrays. Microarray formats include printing proteins at high density on a glass slides or other solid surfaces, using miniaturized reaction chambers adapted for slides, and assaying samples in multiwell plates. Currently two general types of protein microarray applications are being pursued: antibody- or peptide-based arrays and functional protein arrays. Antibody- or peptide-based arrays bind proteins of interest in given samples, such as serum, and can provide information about the amount and speciicity for binding of such proteins. his is referred to as protein proiling. he search for disease biomarkers for diagnostic purposes and drug screening capabilities drives many of the innovations for these types of arrays. Another strategy is to use protein microarray formats

he Role of Cell-Free Rabbit Reticulocyte Expression Systems in Functional Proteomics

7

Figure 1. Schematic of multiwell format protein array experiment. Tagged proteins are generated through coupled translation and then bind to coated multiwell surface. Printed with permission from Promega Corporation.

to interrogate and carefully perturb protein functions such as protein-protein, protein-nucleic acid or protein-small molecule interactions, as well as protein enzyme activities. Cell-free protein expression systems, such as RRL, are well suited to supply the functional proteins required for these types of protein microarrays. Cell-free protein expression in RRL is versatile and allows incorporation of speciic moieties into the protein sequence for immobilization or detection strategies, as well as post-translational modiications, in an automatable manner. Early proof-of-principle for use of RRL in combination with a multiwell array-type format was demonstrated by He and Taussig in 2001,63,64 and named “PISA” (Protein In Situ Array). Proteins with double (His)6 tags were expressed directly from DNA templates with RRL in each well, and the expressed proteins were immobilized by the (His)6 tag to nickel-coated surfaces. Expression and immobilization of functional proteins were demonstrated by enzymatic activity of a cloned (His)6 tagged-luciferase and a tagged-single chain antibody fusion that bound its corresponding antigen, progesterone. Microwells ofer an advantage of maintaining aqueous reaction conditions that are compatible with native protein conformation, compared to more harsh conditions present on a spotted glass slide. Expression of tagged protein expressed in cell-free extracts eliminates laborious protein puriication schemes (Fig. 1).

8

Cell-Free Protein Expression

Oleinikov et al65 describe a variation on protein arrays based on expression of protein from RRL with concomitant immobilization. his strategy exploits features of electronic semiconductor microchips for protein binding and detection. A second variation of protein arrays takes advantage of the stability of immobilized DNA in microarrays, protein expression and spatially deined protein capture.66 his approach, named NAPPA (nucleic acid programmable protein array) involves generating protein in situ with a coupled transcription/translation RRL layered onto slides printed with a mixture of biotinylated DNA, avidin and polyclonal GST antibody. Target proteins are expressed from template DNA encoding a C-terminal GST fusion tag and then the fusion proteins are spatially bound to the GST antibody. he target proteins are detected using a monoclonal antibody to GST.

Protein Interaction with Other Molecules

Cell-free RRL has been used to examine numerous protein-protein,70 protein-DNA,71,72 and protein-RNA interactions73-75 via immunoprecitation67,74-76 and tagged protein pull-down.68,70,72 Conirmation of protein-protein interactions in vitro oten includes expression of target proteins in RRL and pull-down with a tagged protein, such as GST-tagged protein, which is bound to a solid support. hese types of GST-pull-down experiments have been used to conirm speciic interactions of protein involved in signal transduction,77,78 transcription regulation,79-81 ion channels82 and splicesosomes.83 Reconstitution studies can be performed in which expressed proteins form a complex in cell-free RRL and give a measurable biochemical response that mimics an in vivo response. One interesting recent example is the demonstration that p43, a telomerase accessory protein, can afect the in vitro nucleotide addition activity and processivity of the conserved core consisting of the protein telomerase reverse transcriptase (TERT) and the telomerase RNA subunit. he resulting reconstituted ternary complex [TERT•RNA•p43] was identiied and examined by immunoprecipitation in coupled transcription/translation in RRL.76 Another interesting example of protein-RNA interaction in RRL involved the depletion of the internal ribosome entry site (IRES)-interacting protein of the RRL by the immobilized foot-and-mouth virus (FMDV)-IRES.84 he efect of the depleted lysate was assessed by translation eiciency of transcripts that were either capped or had FMDV IRES in the sense or antisense orientation. his procedure should be useful for analysis of protein-RNA interactions and their role in IRES-dependent translation. Cell-free translations in RRL can also provide information about protein-protein interactions. Translation in RRL with microsomal membranes and immunoprecipitations have helped to elucidate the mechanism of activation of the endogenous p21-activated kinase 2 (PAK-2) by Nef proteins that are encoded by human immunodeiciency virus (HIV) and simian immunodeiciency virus (SIV) in vitro.67 he cell-free system allows further investigation of the molecular mechanism of activation because the timing and order of component additions can be easily controlled. Protein interactions that involve ribosome-associated chaperones (cotranslational folding and targeting) and post-translational interactions have also been explored using RRL.68-70 During cotranslational folding the nascent chains interact with chaperones such as the 70-kD heat shock protein cognate (Hsc-70) and nascent polypeptide-associated complex (NAC)68,85 and chaperonins such as the Tailless complex polypeptide 1 (TCP1) ring complex (TRiC).61,87 Recently the interaction of the TRiC with the nascent polypeptide chain as it emerges from the ribosome was demonstrated using photoreactive Nε-(5-azido-2-nitrobenzoyl)-Lys-tRNAlys along with translation of truncated actin in the RRL. Post-translation interactions of chaperones have been investigated using mutagenesis and immunoprecipitation from RRL ater the expression of protein kinases. hese studies showed that phosphorylation of Ser12 of the Hsp-90 cochaperone Cdc37 is critical for its interaction with eukaryotic protein kinases and Hsp-90.70 Historically, protein-DNA interactions have been identiied via mobility shit assays71,72 in which the DNA binding activity of proteins expressed in RRL are visualized by a shit of molecular weight on native polyacrylamide gels. Human biliverdin reductase (hBVR) is a serine/threonine kinase that catalyzes the reduction of biliverdin to bilirubin in response to oxidative stress. Using

he Role of Cell-Free Rabbit Reticulocyte Expression Systems in Functional Proteomics

9

hBVR and hBVR mutants that were translated in RRL and analyzed using a mobility shit assay, hBVR was found to bind to speciic DNA sequences.71

Display Technologies

Cell-free display technologies, such as ribosome display,86-91 mRNA display92-94 or in vitro virus (IVV),95 and in vitro compartmentalization (IVC)96 are powerful technologies that can be used to identify protein-target molecule interactions and for directed evolution of proteins for desired improvements. hese technologies rely on coupled transcription/translation or translation using RRL or other sources of lysates. Cell-free display technologies have advantages over cell-based display technologies such as phage display,97 and cell surface display on bacteria98 or yeast.99 he cell-based display systems have limited library diversity because of transfection ineiciencies, the inability to specify incorporation of nonnatural amino acids via amber suppressor tRNAs and bias against cytotoxic proteins. Display technologies rely on coupling genotype (mRNA) to phenotype (protein) to retrieve the genetic information along with protein function.

Eukaryotic Ribosome Display

Eukaryotic ribosome display is an entirely cell-free technology that screens and selects functional proteins and peptides from large libraries. For ribosome display, the link between genotype and phenotype is accomplished by an mRNA-ribosome-protein (PRM) complex that is stable under controlled conditions. he eukaryotic method of ribosome display using RRL for coupled transcription/translation has been used to display single-chain antibodies to form an antibody-ribosome-mRNA (ARM) complex.87,88,91 he function of the single-chain antibody is evaluated by its binding properties to an immobilized antigen. he function of other non-antibody proteins can be evaluated by using a diferent immobilized target, such as a partner protein, ligand or substrate, to capture the relevant PRM complexes. he mRNA that is complexed with the protein can then be ampliied by reverse transcription polymerase chain reaction (RT-PCR) and recovered as DNA. If screening and selection is the goal, then proofreading DNA polymerases are necessary; however, if evolution or diversiication of the DNA sequence pool is required, then a nonproofreading polymerase such as Taq DNA polymerase is used. he major distinguishing feature between eukaryotic ribosome display and prokaryotic ribosome display86,90 is that in eukaryotic ribosome display, the RT-PCR is carried out on the intact PRM complexes rather than on mRNA that has been released from PRM complex. Eukaryotic ribosome display has been used to select the enzyme, sialyltransferase II, from a cDNA library in a 96-well plate coated with the substrate, ganglioside GM3.100 Coupled transcription/translation in an RRL expression system from the cDNA library resulted in an enzyme-speciic protein-ribosome-mRNA (PRIME) complex. A recently described modiication of eukaryotic ribosome display incorporates Qβ RNA-dependent RNA polymerase into the display and selection process.101 his allows a continuous in vitro evolution (Fig. 2). he cell-free RRL is used in the coupled transcription/translation mode to generate mRNA and then protein. he Qβ RNA-dependent RNA polymerase mutates the generated mRNA and thus the simultaneous display of the protein generated from the original mRNA. he ribosome ternary complexes display the synthesized proteins/single chain antibodies and are selected against immobilized antigens. For the selection process, the displayed wild type and mutants are competing for the target. he recovery of the mRNA is the same as in ARM display. Recently, improvements have been developed for eukaryotic ribosome display that allow 20.8-fold more eicient selection102 than current methods, making ribosome display a readily accessible technique for all researchers.

mRNA Display or in Vitro Virus

A diferent approach to the selection and identiication of functional proteins is mRNA display, also called in vitro virus,95 a technique that uses the cell-free RRL translation system to link a peptide or protein covalently to its encoding mRNA.92-94 he mRNA has a puromycin-tagged DNA linker ligated or photo-crosslinked to the 3´-terminus.103 he ribosome will stall when

10

Cell-Free Protein Expression

Figure 2. The proposed continuous in vitro evolution (CIVE) cycle. The coupled transcription/translation system with inclusion of Qβ replicase. This produces a process in which, in the reaction mix, mRNAs are transcribed from the DNA template by T7 polymerase, replicated and mutated by the Qβ replicase, translated and displayed on the surface of the ribosomes. In selecting against a target, the displayed wild type and mutant are competing for the target. Reprinted from: Irving RA et al, J Immunol Meth 248:31-45; ©2001 with permission from Elsevier.101

it reaches the mRNA-DNA junction, and puromycin enters the ribosomal A-site. he nascent peptide is coupled to the puromycin by the peptidyl-transferase. The complex of peptidepuromycin-DNA linker-mRNA is dissociated from the ribosome and can then be used for selection and identiication of functional protein (Fig. 3). mRNA display or IVV in combination with in vitro selection are powerful tools for evolving and discovering new functional proteins. mRNA display of a random peptide library has been used to determine the epitope-like consensus motifs that deine the determinants for binding of the anti-c-Myc antibody.104 Recently, a method has been described for mRNA display using a unidirectional nested deletion library.105 he method identiied high-ainity, epitope-like peptides for an anti-polyhistidine monoclonal antibody and should be useful for determining minimal binding domains and novel protein-protein interactions. Also, mRNA display can be used to select mRNA templates capable of eiciently incorporating the nonnatural amino acid, biotinyl-lysine, a lysine derivatized at the epsilon amino via amide linkage to biotin.106 To generate the mRNA-peptide fusions, the template library and the tRNA pools are added to RRL for translation. he mRNA-peptide fusions contain a mixture of peptides, some of which contain the biotinyl-lysine. hose that contain biotinyl-lysine can be puriied by binding to streptavidin-agarose.

In Vitro Compartmentalization

In vitro compartmentalization (IVC) links genotype and phenotype by compartmentalization into discrete water-in-oil emulsions.96 Until recently, IVC has mainly been used in conjunction

he Role of Cell-Free Rabbit Reticulocyte Expression Systems in Functional Proteomics

11

Figure 3. Protein:RNA fusion. Covalent RNA:protein complexes can be generated by ligation of a DNA:puromycin linker to the in vitro transcribed mRNA. The ribosome stalls at the RNA:DNA junction. Puromycin then binds to the ribosomal A site. The nascent polypeptide is thereby transferred to puromycin. The resulting covalently linked complex can be used for selection experiments. Reprinted from: Schaffitzel C et al, J Immunol Meth 231:119-135; ©1999 with permission from Elsevier.90

with prokaryotic coupled transcription and translation (E. coli S30 extract) for selection of peptide ligands107-109 and directed evolution of Taq DNA polymerase,110 bacterial phosphotriesterase111 and DNA methyltransferase.112 However, IVC has been used in conjunction with coupled transcription/translation in RRL to select active restriction enzymes from a randomized large (109–1010 molecules) mutant Fok I library.103 Use of RRL in this way is made possible by a new inert emulsion formulation that is compatible with coupled transcription/translation,114 allowing an expanded range of vertebrate protein targets that may be diicult to express as soluble and functional in S30 (bacterial) or wheat germ lysates.

Screening In Vitro Expression Cloning (IVEC)

IVEC115,116 is another approach that uses RRL for coupled transcription/translation to identify genes and elucidate protein interactions with other molecules. Using this method cDNAs or small plasmid pools (50–100 clones) are expressed and then assayed for a speciic function. Plasmids from the positive pools are further deconvoluted and rescreened (Fig. 4). he process is repeated until a single positive plasmid has been identiied. IVEC is only successful if the speciic function assayed can be distinguished from the endogenous activity in cell-free RRL. IVEC has been successfully used to identify protein substrates,117-120 protein-protein interactions,121,122 enzymatic activity,123-125 protein-DNA interactions,126 and phospholipid-protein interactions.136

12

Cell-Free Protein Expression

Figure 4. The strategy of in vitro expression cloning. An unamplified cDNA expression library is plated at a density of approximately 100 clones per bacterial plate. Pooled plasmid DNA is obtained by scraping colonies from each plate and performing a small-scale plasmid purification. Each plasmid pool is then transcribed and translated in vitro with a commercially available system, such as the TNT® Coupled Transcription/Translation Systems from Promega Corporation. The resulting protein pool is then assayed for the presence of an activity. In the illustrated experiment, a radioactive amino acid is included in the translation system to specifically label the pool of proteins. Incubation of a pool with a modifying enzyme (lanes labeled +) such as a protease or kinase can result in a change in mobility of a substrate (bands marked with asterisk). Pool 1 contains a protein whose mobility is reduced following treatment with a kinase; Pool 2 contains a protein that is degraded following treatment with an extract containing an activated proteolytic system; Pool 3 contains a protein that is specifically cleaved following treatment with a protease, decreasing its apparent molecular mass. Once a pool containing a candidate activity is identified, the original cDNA pool is subdivided and retested until the single cDNA encoding the protein of interest is isolated. Reprinted with permission from: King RW et al, Science 277:973; ©1997 AAAS.115

he Role of Cell-Free Rabbit Reticulocyte Expression Systems in Functional Proteomics

13

Protein Truncation Test

he protein truncation test (PTT) is a mutation detection technique that speciically identiies pathogenic premature termination codons and has the advantage of not detecting polymorphisms. Using extracted RNA, the coding region is screened for truncated mutations. he RNA is subjected to RT-PCR such that the cDNA product contains a T7 promoter. he cDNA is subjected to coupled transcription/translation in cell-free RRL, and the translated products are analyzed on gels to identify the truncated proteins. he PTT has been applied to screening for many clinical conditions including hereditary breast and ovarian cancer (BRCA I and BRCA II),128 colorectal cancer (APC),129 Duchenne Muscular Dystrophy (DMD),130 and neuroibromatosis type I (NFI).131

Other Screening

An approach to screening that uses cell-free RRL and site-directed mutagenesis to identify elements critical for protein N-myristoylation132 has recently been described. Sequential verticalscanning mutagenesis in the N-terminal region of tumor necrosis factor (TNF), followed by cotranslation N-myristoylation in RRL revealed the major sequence requirements for protein N-myristoylation. RRL has also been used to functionally screen a randomly mutagenized phage library.133 In this example, critical amino acids in the protein C1 were identiied that were responsible for binding RNA. Other amino acids were identiied that were important for protein oligomerization. Protein C1 is a member of the heterogeneous ribonucleoproteins that bind nascent RNA transcripts. Screens that use cell-free RRL have also been developed to identify small-molecule inhibitors of translation.134 Numerous reports have implicated alternative translation initiation controls occurring in cancer cells,135-139 and small-molecule inhibitors may provide tools for determining the molecular mechanism of this alternative regulation. A high-throughput screen was designed to identify small-molecule inhibitors of eukaryotic translation as well as inhibitors that interact at the mRNA-ribosome level to inhibit gene-speciic translation. his multiplex in vitro translation was used to screen over 900,000 distinct compounds identiied novel inhibitors.134 A secondary high-throughput eukaryotic translation screen to discover broad spectrum antibacterial compounds has also been used to assess the biochemical selectivity of the compounds for prokaryotic translation.140 Another screen using RRL is a PCR-based rapid detection screen for pyrazinamide (PZA)-resistant Mycobacterium tuberculosis.141 Ater ampliication of the pncA gene and coupled transcription/translation of the PCR product, the activity of the enzyme pyrazinamidase is measured by the conversion of PZA to pyrazinoic acid. Other PCR-based coupled transcription/translation methods for rapid phenotypic screening have also been developed, including screening of the thymidine kinase gene for monitoring acyclovir resistant herpes simplex virus and varicella-zoster virus.142

Conclusions

Cell-free RRL plays an important role in functional proteomics, whether the protein is destined for the ER, modiication, degradation, or forms a complex with DNA, RNA and other proteins. RRL provides a mammalian environment for elucidating quasi-cellular mechanisms and can be easily manipulated by depleting or adding protein, tRNA or membranes to provide the desired environment so that the function of a protein or proteins can be studied. Additionally, cell-free RRL is proving to be a useful tool for high-throughput protein synthesis in protein microarrays and other screening situations.

References

1. Allen EH, Schweet RS. Synthesis of hemoglobin in a cell-free system. J Biol Chem 1962; 237:760-7. 2. Schweet R, Lamfrom H, Allen E. he Synthesis of Hemoglobin in a Cell-Free System. Proc Natl Acad Sci USA 1958; 44:1029-35. 3. Hershey JWB, Matthews MB. he pathway and mechanism of initiation of protein synthesis. In: Sonenberg N, Hershey JWB, Matthews MB, eds. Translational Control of Gene Expression. Cold Spring Harbor: Cold Spring Harbor Laboratory Press, 2000:33-88.

14

Cell-Free Protein Expression

4. Merrick WC. Cap-dependent and cap-independent translation in eukaryotic systems. Gene 2004; 332:1-11. 5. Browne GJ, Proud CG. Regulation of peptide-chain elongation in mammalian cells. Eur J Biochem 2002; 269:5360-8. 6. Pelham HR, Jackson RJ. An eicient mRNA-dependent translation system from reticulocyte lysates. Eur J Biochem 1976; 67:247-56. 7. Craig D, Howell MT, Gibbs CI et al. Plasmid cDNA-directed protein synthesis in a coupled eukaryotic in vitro transcription-translation system. Nucl Acids Res 1992; 20:4987-95. 8. homspon D, van Oosbree T, Beckler G et al. he TnT® Lysate Systems: One-step transcription/translation in rabbit reticulocyte lysate. Promega Notes 1992; 35:1-4. 9. Ott CM, Lingappa VR. Integral membrane protein biosynthesis: why topology is hard to predict. J Cell Sci 2002; 115:2003-9. 10. Bocco JL, Panzetta GM, Flury A et al. Processing of SP1 precursor in a cell-free system from poly(A+) mRNA of human placenta. Mol Biol Rep 1988; 13:45-51. 11. MacDonald MR, McCourt DW, Krause JE. Posttranslational processing of alpha-, beta-, and gamma- preprotachykinins. Cell-free translation and early posttranslational processing events. J Biol Chem 1988; 263:15176-83. 12. Ray RB, Lagging LM, Meyer K et al. Transcriptional regulation of cellular and viral promoters by the hepatitis C virus core protein. Virus Res 1995; 37:209-20. 13. Wilson R, Allen AJ, Oliver J et al. he translocation, folding, assembly and redox-dependent degradation of secretory and membrane proteins in semi-permeabilized mammalian cells. Biochem J 1995; 307 (Pt 3):679-87. 14. Stewart RS, Harris DA. Most pathogenic mutations do not alter the membrane topology of the Prion protein. J Biol Chem 2001; 276:2212-20. 15. Wilson CM, Farmery MR, Bulleid NJ. Pivotal role of calnexin and mannose trimming in regulating the endoplasmic reticulum-associated degradation of major histocompatibility complex class I heavy chain. J Biol Chem 2000; 275:21224-32. 16. Howell KE, Palade GE. Hepatic Golgi fractions resolved into membrane and content subfractions. J Cell Biol 1982; 92:822-32. 17. Wieringa R, de Vries AAF, Raamsman MJB et al. Characterization of two new structural glycoproteins, GP(3) and GP(4), of equine arteritis virus. J Virol 2002; 76:10829-40. 18. Vergères G, Ramsden J, Waskell L. he carboxyl terminus of the membrane-binding domain of cytochrome b(5) spans the bilayer of the endoplasmic reticulum. J Biol Chem 1995; 270:3414-22. 19. Umigai N, Sato Y, Mizutani A et al. Topogenesis of two transmembrane type K+ channels, Kir 2.1 and KcsA. J Biol Chem 2003; 278:40373-84. 20. Nicchitta CV, Blobel G. Nascent secretory chain binding and translocation are distinct processes: differentiation by chemical alkylation. J Cell Biol 1989; 108:789-95. 21. Nakai T, Yamasaki A, Sakaguchi M et al. Membrane topology of Alzheimer’s disease-related Presenilin 1. Evidence for the existence of a molecular species with a seven membrane-spanning and one membrane-embedded structure. J Biol Chem 1999; 274:23647-58. 22. Tie J-K, Nicchitta C, von Heijne G et al. Membrane topology mapping of vitamin K epoxide reductase by in vitro translation/cotranslocation. J Biol Chem 2005; 280:16410-6. 23. Zhang JT, Ling V. Study of membrane orientation and glycosylated extracellular loops of mouse P-glycoprotein by in vitro translation. J Biol Chem 1991; 266:18224-32. 24. Bailey CA, Gerber L, Howard AD et al. Processing at the carboxyl terminus of nascent placental alkaline phosphatase in a cell-free system: evidence for speciic cleavage of a signal peptide. Proc Natl Acad Sci USA 1989; 86:22. 25. Buse MG, Robinson KA, Marshall BA et al. Enhanced O-GlcNAc protein modiication is associated with insulin resistance in GLUT1-overexpressing muscles. AJP—Endocrinology and Metabolism 2002; 283:241-50. 26. Ren JM, Marshall BA, Gulve EA et al. Evidence from transgenic mice that glucose transport is ratelimiting for glycogen deposition and glycolysis in skeletal muscle. J Biol Chem 1993; 268:16113-5. 27. Ferguson MA. he structure biosynthesis and functions of glycosylphosphatidyl inositol anchors, and the contribution of Typanosome research. J Cell Sci 1999; 112:2799-809. 28. Ohishi K, Inoue N, Maeda Y et al. Gaa1p and Gpi8p Are Components of a Glycosylphosphatidylinositol (GPI) Transamidase hat Mediates Attachment of GPI to Proteins. Mol Biol Cell 2000; 11:1523-33. 29. Dalley JA, Bulleid NJ. he endoplasmic reticulum (ER) translocon can diferentiate between hydrophobic sequences allowing signals for glycosylphosphatidylinositol anchor addition to be fully translocated into the ER lumen. J Biol Chem 2003; 278:51749-57. 30. Maxwell SE, Ramalingam S, Gerber LD et al. An active carbonyl formed during glycosylphosphatidylinositol addition to a protein is evidence of catalysis by a transamidase. J Biol Chem 1995; 270:19576-82.

he Role of Cell-Free Rabbit Reticulocyte Expression Systems in Functional Proteomics

15

31. Chen R, Walter EI, Parker G et al. Mammalian glycophosphatidylinositol anchor transfer to proteins and posttransfer deacylation. Proc Natl Acad Sci USA 1998; 95:9512-7. 32. Vidugiriene J, Vainauskas S, Johnson AE et al. Endoplasmic reticulum proteins involved in glycosylphosphatidylinositol-anchor attachment: Photocrosslinking studies in a cell-free system. Eur J Biochem 2001; 268:2290-300. 33. Spurway TD, Dalley JA, High S et al. Early Events in Glycosylphosphatidylinositol Anchor Addition. Substrate proteins associate with the transamidase subunit Gpi8p. J Biol Chem 2001; 276:15975-82. 34. Quellhorst JGJ, Allen CM, Wessling-Resnick M. Modiication of Rab5 with a photoactivatable analog of geranylgeranyl diphosphate. J Biol Chem 2001; 276:40727-33. 35. Sanford J, Codina J, Birnbaumer L. Gamma-subunits of G proteins, but not their alpha- or beta-subunits, are polyisoprenylated. Studies on post-translational modiications using in vitro translation with rabbit reticulocyte lysates. J Biol Chem 1991; 266:9570-9. 36. Newman CM, Giannakouros T, Hancock JF et al. Post-translational processing of Schizosaccharomyces pombe YPT proteins. J Biol Chem 1992; 267:11329-36. 37. Khosravi-Far R, Lutz RJ, Cox AD et al. Isoprenoid Modiication of Rab Proteins Terminating in CC or CXC Motifs. Proc Natl Acad Sci USA 1991; 88:6264-8. 38. Benetka W, Koranda M, Maurer-Stroh S et al. Farnesylation or geranylgeranylation? Eicient assays for testing protein prenylation in vitro and in vivo. BMC Biochemistry 2006; 7. 39. Gordon JI, Duronio RJ, Rudnick DA et al. Protein N- myristoylation. J Biol Chem 1991; 266:8647-50. 40. Jones TLZ, Simonds WF, Merendino JJ et al. Myristoylation of an inhibitory GTP-binding protein {alpha} subunit is essential for its membrane attachment. Proc Natl Acad Sci USA 1990; 87:568-72. 41. Mumby SM, Heukeroth RO, Gordon JI et al. G-protein {alpha}-subunit expression, myristoylation, and membrane association in COS cells. Proc Natl Acad Sci USA 1990; 87:728-32. 42. Mumby SM, Kleuss C, Gilman AG. Receptor regulation of G-protein palmitoylation. Proc Natl Acad Sci USA 1994; 91:2800-4. 43. Farazi TA, Waksman G, Gordon JI. he biology and enzymology of protein N-myristoylation. J Biol Chem 2001; 276:39501-4. 44. Utsumi T, Kuranami J, Tou E et al. In vitro synthesis of an N-myristoylated fusion protein that binds to the liposomal surface. Arch Biochem Biophys 1996; 326:179-84. 45. Utsumi T, Tou E, Takemura D et al. Met-Gly-Cys motif from G-protein alpha subunit cannot direct palmitoylation when fused to heterologous protein. Arch Biochem Biophys 1998; 349:216-24. 46. Ishitani M, Liu J, Halter U et al. SOS3 function in plant salt tolerance requires N-myristoylation and calcium binding. Plant Cell 2000; 12:1667-78. 47. Utsumi T, Sato M, Nakano K et al. Amino acid residue penultimate to the amino-terminal Gly residue strongly afects two cotranslational protein modiications, N-myristoylation and N-acetylation. J Biol Chem 2001; 276:10505-13. 48. Polevoda B, Sherman F. Nalpha -terminal acetylation of eukaryotic proteins. J Biol Chem 2000; 275:36479-82. 49. Rubenstein P, Smith P, Deuchler J et al. NH2-terminal acetylation of Dictyostelium discoideum actin in a cell-free protein-synthesizing system. J Biol Chem 1981; 256:8149-55. 50. Kostova Z, Wolf DH. For whom the bell tolls: protein quality control of the endoplasmic reticulum and the ubiquitin-proteasome connection. EMBO J 2003; 22:2306-17. 51. Carlson E, Bays N, David L et al. Reticulocyte lysate as a model system to study endoplasmic reticulum membrane protein degradation. Vol 301. Totowa, New Jersey USA: Humana Press 2005. 52. Teckman JH, Gilmore R, Perlmutter DH. Role of ubiquitin in proteasomal degradation of mutant alpha 1-antitrypsin Z in the endoplasmic reticulum. Amer J Physiol Gastrointestinal Liver Physiol 2000; 278:39-48. 53. Svedine S, Wang T, Halaban R et al. Carbohydrates act as sorting determinants in ER-associated degradation of tyrosinase. J Cell Sci 2004; 117:2937-49. 54. Gusarova V, Caplan AJ, Brodsky JL et al. Apoprotein B degradation is promoted by the molecular chaperones hsp90 and hsp70. J Biol Chem 2001; 276:24891-900. 55. Novelli A, Boulanger PA. Assembly of adenovirus type 2 iber synthesized in cell-free translation system. J Biol Chem 1991; 266:9299-303. 56. Iyengar S, Shah KV, Kotlof KL et al. Self-assembly of in vitro-translated human papillomavirus type 16 L1 capsid protein into virus-like particles and antigenic reactivity of the protein. Clin Diagn Lab Immunol 1996; 3:733-9. 57. Klein KC, Polyak SJ, Lingappa JR. Unique features of hepatitis C virus capsid formation revealed by de novo cell-free assembly. J Virol 2004; 78:9257-69. 58. Spearman P, Ratner L. Human Immunodeiciency Virus Type 1 capsid formation in reticulocyte lysates. J Virol 1996; 70:8187-94.

16

Cell-Free Protein Expression

59. Lee YM, Liu B, Yu XF. Formation of virus assembly intermediate complexes in the cytoplasm by wild-type and assembly-defective mutant Human Immunodeiciency Virus Type 1 and their association with membranes. J Virol 1999; 73:5654-62. 60. Schildgen O, Roggendorf M, Lu M. Identiication of a glycosylation site in the woodchuck hepatitis virus preS2 protein and its role in protein traicking. J Gen Virol 2004; 85:787-93. 61. Katzen F, Chang G, Kudlicki W. he past, present and future of cell-free protein synthesis. Trends Biotechnol 2005; 23:150-6. 62. Bertone P, Snyder M. Advances in functional protein microarray technolog y. FEBS J 2005; 272:5400-11. 63. He MT, M. Single step generation of protein arrays from DNA by cell-free expression and in situ immobilisation (PISA method). Nucl Acids Res 2001; 29:e73. 64. He MT, M. DiscernArray technology: a cell-free method for the generation of protein arrays from PCR DNA. J Immunol Meth 2003; 254:265-70. 65. Oleinikov AV, Gray MD, Zhao J et al. Self-assembling protein arrays using electronic semiconductor microchips and in vitro translation. J Proteome Res 2003; 2:313-9. 66. Ramachandran N, Hainsworth E, Bhullar B et al. Self-Assembling protein microarrays. Science 2004; 305:86-90. 67. Raney A, Kuo LS, Baugh LL et al. Reconstitution and molecular analysis of an active Human Immunodeiciency Virus Type 1 Nef/p21-activated kinase 2 complex. J Virol 2005; 79:12732-41. 68. Beatrix B, Sakai H, Wiedmann M. he alpha and beta subunit of the nascent polypeptide-associated complex have distinct functions. J Biol Chem 2000; 275:37838-45. 69. McCallum CD, Do H, Johnson AE et al. he interaction of the chaperonin Tailless Complex Polypeptide 1 (TCP1) Ring Complex (TRiC) with ribosome-bound nascent chains examined using photo-cross-linking. J Cell Biol 2000; 149:591-602. 70. Shao J, Prince T, Hartson SD et al. Phosphorylation of serine 13 is required for the proper function of the Hsp90 cochaperone, Cdc37. J Biol Chem 2003; 278:38117-20. 71. Ahmad Z, Salim M, Maines MD. Human biliverdin reductase is a leucine zipper-like DNA-binding protein and functions in transcriptional activation of heme oxygenase-1 by oxidative stress. J Biol Chem 2002; 277:9226-32. 72. Niyaz Y, Frenz I, Petersen G et al. Transcriptional stimulation by the DNA binding protein Hap46/ BAG-1M involves hsp70/hsc70 molecular chaperones. Nucl Acids Res 2003; 31:2209-16. 73. Bryan TM, Goodrich KJ, Cech TR. Telomerase RNA bound by protein motifs speciic to telomerase reverse transcriptase. Mol Cell 2000; 6:493-9. 74. Bryan TM, Goodrich KJ, Cech TR. A mutant of Tetrahymena telomerase reverse transcriptase with increased processivity. J Biol Chem 2000; 275:24199-207. 75. Collins K, Gandhi L. he reverse transcriptase component of the Tetrahymena telomerase ribonucleoprotein complex. Proc Natl Acad Sci USA 1998; 95:8485-90. 76. Aigner S, Cech TR. he Euplotes telomerase subunit p43 stimulates enzymatic activity and processivity in vitro. RNA 2004; 10:1108-18. 77. Lamark T, Perander M, Outzen H et al. Interaction codes within the family of mammalian Phox and Bem1p domain-containing proteins. J Biol Chem 2003; 278:34568-81. 78. Olsten MEK, Canton DA, Zhang C et al. he Pleckstrin Homology domain of CK2 interacting protein-1 is required for interactions and recruitment of protein kinase CK2 to the plasma membrane. J Biol Chem 2004; 279:42114-27. 79. Choi CY, Kim YH, Kim Y-O et al. Phosphorylation by the DHIPK2 protein kinase modulates the corepressor activity of groucho. J Biol Chem 2005; 280:21427-36. 80. Eisbacher M, Holmes ML, Newton A et al. Protein-protein interaction between Fli-1 and GATA-1 mediates synergistic expression of megakaryocyte-speciic genes through cooperative DNA binding. Mol Cell Biol 2003; 23:3427-41. 81. Poulin G, Lebel M, Chamberland M et al. Specific protein-protein interaction between basic helix-loop-helix transcription factors and homeoproteins of the Pitx family. Mol Cell Biol 2000; 20:4826-37. 82. Santoro B, Wainger BJ, Siegelbaum SA. Regulation of HCN channel surface expression by a novel C-terminal protein-protein interaction. J Neurosci 2004; 24:10750-62. 83. Ingelinger D, Gothel SF, Marahiel MA et al. Two protein-protein interaction sites on the spliceosomeassociated human cyclophilin CypH. Nucl Acids Res 2003; 31:4791-6. 84. Stassinopoulos IA, Belsham GJ. A novel protein-RNA binding assay: functional interactions of the foot-and-mouth disease virus internal ribosome entry site with cellular proteins. RNA 2001; 7:114-22. 85. Hartl FU, Hayer-Hartl M. Molecular chaperones in the cytosol: from nascent chain to folded protein. Science 2002; 295:1852-8.

he Role of Cell-Free Rabbit Reticulocyte Expression Systems in Functional Proteomics

17

86. Hanes J, Pluckthun A. In vitro selection and evolution of functional proteins by using ribosome display. Proc Natl Acad Sci USA 1997; 94:4937-42. 87. He M, Menges M, Groves MA et al. Selection of a human anti-progesterone antibody fragment from a transgenic mouse library by ARM ribosome display. J Immunol Meth 1999; 231:105-17. 88. He M, Taussig MJ. Antibody-ribosome-mRNA (ARM) complexes as eicient selection particles for in vitro display and evolution of antibody combining sites. Nucl Acids Res 1997; 25:5132-4. 89. Mattheakis LC, Bhatt RR, Dower WJ. An in vitro polysome display system for identifying ligands from very large peptide libraries. Proc Natl Acad Sci USA 1994; 91:9022-6. 90. Schaitzel C, Hanes J, Jermutus L et al. Ribosome display: an in vitro method for selection and evolution of antibodies from libraries. J Immunol Meth 1999; 231:119-35. 91. He M, Taussig MJ. Ribosome display: Cell-free protein display technology. Brief Funct Genomic Proteomic 2002; 1:204-12. 92. Keefe AD, Szostak JW. Functional proteins from a random-sequence library. Nature 2001; 410:715-8. 93. Roberts RW, Szostak JW. RNA-peptide fusions for the in vitro selection of peptides and proteins. Proc Natl Acad Sci USA 1997; 94:12297-302. 94. Wilson DS, Keefe AD, Szostak JW. he use of mRNA display to select high-ainity protein-binding peptides. Proc Natl Acad Sci USA 2001; 98:3750-5. 95. Nemoto N, Miyamoto-Sato E, Husimi Y et al. In vitro virus: bonding of mRNA bearing puromycin at the 3'-terminal end to the C-terminal end of its encoded protein on the ribosome in vitro. FEBS Lett 1997; 414:405-8. 96. Tawik DS, Griiths AD. Man-made cell-like compartments for molecular evolution. Nat Biotechnol 1998; 16:652-6. 97. Smith GP. Filamentous fusion phage: novel expression vectors that display cloned antigens on the virion surface. Science 1985; 228:1315-7. 98. Georgiou G, Stathopoulos C, Daugherty PS et al. Display of heterologous proteins on the surface of microorganisms: from the screening of combinatorial libraries to live recombinant vaccines. Nat Biotechnol 1997; 15:29-34. 99. Boder ET, Wittrup KD. Yeast surface display for screening combinatorial polypeptide libraries. Nat Biotechnol 1997; 15:553-7. 100. Bieberich E, Kapitonov D, Tencomnao T et al. Protein-ribosome-mRNA display: ainity isolation of enzyme-ribosome-mRNA complexes and cDNA cloning in a single-tube reaction. Anal Biochem 2000; 287:294-8. 101. Irving RA, Coia G, Roberts A et al. Ribosome display and ainity maturation: from antibodies to single V-domains and steps towards cancer therapeutics. J Immunol Meth 2001; 248:31-45. 102. Douthwaite JA, Groves MA, Dufner P et al. An improved method for an eicient and easily accessible eukaryotic ribosome display technology. Protein Eng Des Sel 2006; 19:85-90. 103. Kurz M, Gu K, Lohse PA. Psoralen photo-crosslinked mRNA-puromycin conjugates: a novel template for the rapid and facile preparation of mRNA-protein fusions. Nucl Acids Res 2000; 28:e83. 104. Baggio R, Burgstaller P, Hale SP et al. Identiication of epitope-like consensus motifs using mRNA display. J Mol Recognit 2002; 15:126-34. 105. Ja WW, Olsen BN, Roberts RW. Epitope mapping using mRNA display and a unidirectional nested deletion library. Protein Eng Des Sel 2005; 18:309-19. 106. Frankel A, Roberts RW. In vitro selection for sense codon suppression. RNA 2003; 9:780-6. 107. Doi N, Yanagawa H. STABLE: protein-DNA fusion system for screening of combinatorial protein libraries in vitro. FEBS Lett 1999; 457:227-30. 108. Sepp A, Tawik DS, Griiths AD. Microbead display by in vitro compartmentalisation: selection for binding using low cytometry. FEBS Lett 2002; 532:455-8. 109. Yonezawa M, Doi N, Kawahashi Y et al. DNA display for in vitro selection of diverse peptide libraries. Nucl Acids Res 2003; 31:e118. 110. Ghadessy FJ, Ong JL, Holliger P. Directed evolution of polymerase function by compartmentalized self-replication. Proc Natl Acad Sci USA 2001; 98:4552-7. 111. Griiths AD, Tawik DS. Directed evolution of an extremely fast phosphotriesterase by in vitro compartmentalization. EMBO J 2003; 22:24-35. 112. Cohen HM, Tawik DS, Griiths AD. Altering the sequence speciicity of Hae III methyltransferase by directed evolution using in vitro compartmentalization. Protein Eng Des Sel 2004; 17:3-11. 113. Doi N, Kumadaki S, Oishi Y et al. In vitro selection of restriction endonucleases by in vitro compartmentalization. Nucl Acids Res 2004; 32:e95. 114. Ghadessy FJ, Holliger P. A novel emulsion mixture for in vitro compartmentalization of transcription and translation in the rabbit reticulocyte system. Protein Eng Des Sel 2004; 17:201-4. 115. King RW, Lustig KD, Stukenberg TP et al. Functional genomics: expression cloning in the test tube. Science 1997; 277:973-4.

18

Cell-Free Protein Expression

116. Lustig KD, Stukenberg PT, McGarry TJ et al. Small pool expression screening: identiication of genes involved in cell cycle control, apoptosis, and early development. Meth Enzymol 1997; 283:83-99. 117. Gocke CB, Yu H, Kang J. Systematic identiication and analysis of mammalian small ubiquitin-like modiier substrates. J Biol Chem 2005; 280:5004-12. 118. Katou S, Yoshioka H, Kawakita K et al. Involvement of PPS3 Phosphorylated by Elicitor-Responsive Mitogen-Activated Protein Kinases in the Regulation of Plant Cell Death. Plant Physiol 2005; 139:1914-26. 119. Kothakota S, Azuma T, Reinhard C et al. Caspase-3-generated fragment of gelsolin: efector of morphological change in apoptosis. Science 1997; 278:294-8. 120. Potts PR, Yu H. Human MMS21/NSE2 is a SUMO ligase required for DNA repair. Mol Biol Cell 2005; 25:7021-32. 121. Marignani PA, Kanai F, Carpenter CL. LKB1 associates with Brg1 and is necessary for Brg1-induced growth arrest. J Biol Chem 2001; 276:32415-8. 122. Pridgeon JW, Geetha T, Wooten MW. A method to identify p62's UBA domain interacting proteins. Biol Proced Online 2003; 5:228-37. 123. Haushalter KA, Todd Stukenberg MW, Kirschner MW et al. Identiication of a new uracil-DNA glycosylase family by expression cloning using synthetic inhibitors. Curr Biol 1999; 9:174-85. 124. McGarry TJ, Kirschner MW. Geminin, an inhibitor of DNA replication, is degraded during mitosis. Cell 1998; 93:1043-53. 125. Stukenberg PT, Lustig KD, McGarry TJ et al. Systematic identiication of mitotic phosphoproteins. Curr Biol 1997; 7:338-48. 126. Mead PE, Zhou Y, Lustig KD et al. Cloning of Mix-related homeodomain proteins using fast retrieval of gel shit activities, (FROGS), a technique for the isolation of DNA-binding proteins. Proc Natl Acad Sci USA 1998; 95:11251-6. 127. Rao VR, Corradetti MN, Chen J et al. Expression cloning of protein targets for 3-phosphorylated phosphoinositides. J Biol Chem 1999; 274:37893-900. 128. Geisler JP, Hatterman-Zogg MA, Rathe JA et al. Ovarian cancer BRCA1 mutation detection: Protein truncation test (PTT) outperforms single strand conformation polymorphism analysis (SSCP). Hum Mutat 2001; 18:337-44. 129. Traverso G, Diehl F, Hurst R et al. Multicolor in vitro translation. Nat Biotechnol 2003; 21:1093-7. 130. Whittock NV, Roberts RG, Mathew CG et al. Dystrophin point mutation screening using a multiplexed protein truncation test. Genet Test 1997; 1:115-23. 131. Rasmussen SA, Friedman JM. NF1 Gene and Neuroibromatosis 1. Am J Epidemiol 2000; 151:33-40. 132. Utsumi T, Nakano K, Funakoshi T et al. Vertical-scanning mutagenesis of amino acids in a model Nmyristoylation motif reveals the major amino-terminal sequence requirements for protein N-myristoylation. Eur J Biochem 2004; 271:863-74. 133. Wan L, Kim JK, Pollard VW et al. Mutational deinition of RNA-binding and protein-protein interaction domains of heterogeneous nuclear RNP C1. J Biol Chem 2001; 276:7681-8. 134. Novac O, Guenier AS, Pelletier J. Inhibitors of protein synthesis identiied by a high throughput multiplexed translation screen. Nucl Acids Res 2004; 32:902-15. 135. Graf JR, Zimmer SG. Translational control and metastatic progression: enhanced activity of the mRNA cap-binding protein eIF-4E selectively enhances translation of metastasis-related mRNAs. Clin Exp Metastasis 2003; 20:265-73. 136. Hershey JWB, Miyamoto S, eds. Translational control and cancer. Plainview, NY: Cold Spring Harbor Laboratory Press; 2002. Sonenberg N, Hershey JWB, Matthews MB, eds. Translational control of gene expression. 137. Meric F, Hunt KK. Translation Initiation in Cancer: A Novel Target for herapy. Mol Cancer her 2002; 1:971-9. 138. Ruggero D, Pandoli PP. Does the ribosome translate cancer? Nat Rev Can 2003; 3:179-92. 139. Watkins SJ, Norbury CJ. Translation initiation and its deregulation during tumorigenesis. Br J Cancer 2002; 86:1023-7. 140. Pratt SD, David CA, Black-Schaefer C et al. A strategy for discovery of novel broad-spectrum antibacterials using a high-throughput Streptococcus pneumoniae transcription/translation screen. J Biomol Screen 2004; 9:3-11. 141. Suzuki Y, Suzuki A, Tamaru A et al. Rapid detection of pyrazinamide-resistant Mycobacterium tuberculosis by a PCR-based in vitro system. J Clin Microbiol 2002; 40:501-7. 142. Suzutani T, Saijo M, Nagamine M et al. Rapid phenotypic characterization method for Herpes Simplex Virus and Varicella-Zoster Virus thymidine kinases to screen for acyclovir-resistant viral infection. J Clin Microbiol 2000; 38:1839-44.

Chapter 2

Advances in Insect-Based Cell-Free Protein Expression Uritza von Groll,* Stefan Kubick, Helmut Merk, Wolfgang Stiege and Frank Schäfer*

Abstract

C

ell-free protein expression systems are becoming more widely used as they allow fast synthesis of recombinant proteins and an easy manipulation of reaction conditions. Here, we report on recent advances of a eukaryotic system based on extracts derived from Spodoptera rugiperda insect cells. We demonstrate the use of PCR products as DNA templates for protein production and how this application speeds up in vitro protein expression and potentially cell-based production processes. We also show the expression of a variety of functionally active eukaryotic proteins, including kinases and membrane proteins. Finally, we compare the eiciency of this improved insect-based in vitro protein expression system to traditionally used eukaryotic cell-free extracts, such as rabbit reticulocyte lysate.

Introduction

Cell-free protein expression is a widely used technology for fast production of recombinant proteins. In vitro translation systems provide powerful tools to generate proteins of highly diverse origin suitable for a wide range of downstream applications. In particular, these systems allow the expression of proteins toxic to living cells. Cell-free extracts contain the cellular folding machinery and thus produce functionally active proteins. Furthermore, cell free systems enable the addition of cofactors important for maximum activity. In vitro translation systems can be manipulated to generate proteins carrying labeled amino acids for structural analyses1-3 and other unnatural modiications.4-6 he ease and speed of cell-free expression makes this technology an attractive tool for expression screening of constructs in a small scale. he obtained results allow predictions on protein expression in vivo,7,8 e.g., E. coli-based in vitro translation systems can be scaled up to produce milligram amounts of proteins for structural studies.1,2,9 While many proteins can be eiciently produced in prokaryotic cell-free extracts, a broad range of eukaryotic proteins require an eukaryotic environment to fold properly. Very oten, additional posttranslational modiications such as phosphorylation, glycosylation and signal peptide cleavage are required to display full functional activity. Widely used eukaryotic systems include the rabbit reticulocyte lysate (RRL) and wheat germ (WG) extracts. Both systems are suitable to produce correctly folded eukaryotic proteins but have signiicant shortfalls. For expression of integral membrane proteins, the RRL requires the addition of a heterologous membrane fraction. Furthermore, quality of RRL’s is subject to variation as it is based on cells collected from living animals. he WG system can be scaled up9 but production of posttranslational modiications is limited. *Corresponding Authors: Dr. Frank Schäfer, Dr. Uritza von Groll—QIAGEN GmbH, Qiagen Strasse 1, 40724 Hilden, Germany. Email: [email protected]; Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

20

Cell-Free Protein Expression

Recently, a new eukaryotic system for cell free expression based on insect cells has been described.10 Spodoptera rugiperda (SF) cells are grown under deined conditions in well controlled fermenters. Subsequently, the cells are lysed mechanically and the obtained extract retains intact subcellular membraneous structures derived from the endoplasmatic reticulum (ER). hese vesicular structures are an essential prerequisite for many post-translational modiications and the expression of functionally active membrane proteins. he system is a linked transcription-translation procedure and may be performed in two modes. In the high-throughput mode, an aliqout of the transcription step is directly pipetted into the extract to start the translation, thus enabling multi-parallel and automated processing of various reactions. he high-yield mode is designed to maximize protein yield which is achieved by an intermediate gel iltration step to clean up the mRNA prior to addition to the cell-free extract. For production of proteins in a larger scale the reaction volume can be linearly scaled up into the milliliter range. his system has been developed further for commercialization. Recent advances and applications are described here and are partly compared to the eukaryotic RRL system. We show the use of various types of templates including linear templates generated by PCR. hese templates are suitable for screening expression constructs. Soluble expression of eukaryotic kinases, transcription factors and membrane proteins which can hardly be expressed solubly and functionally in E. coli is demonstrated. Furthermore, the formation of post-translational modiications and their analysis is shown.

Materials and Methods Construction of DNA Templates

Proteins were expressed using either PCR products or plasmid DNA as template. cDNA’s coding for proteins to be expressed were ampliied by PCR and cloned into an expression vector. Alternatively, PCR fragments were used as linear expression constructs generated by PCR using the EasyXpress Linear Template Kit PLUS (QIAGEN) and following the instructions given in the kit’s manual. Table 1 summarizes the proteins and constructs used here.

Protein Expression

Cell-free protein expression was performed using the following kits according to the manufacturer’s instructions: EasyXpress Insect Kit II (high-yield protocol, QIAGEN) for insect-based cell-free protein expression, TNT T7 Quick Coupled Transcription/Translation System (Promega) for rabbit reticulocyte lysate protein expression and EasyXpress Protein Synthesis Mini Kit (QIAGEN) for E. coli-based cell-free expression. Protein expression in E. coli in vivo was performed in strain BL21 (DE3) pLysS grown in LB medium plus appropriate antibiotics at 37°C and protein synthesis was induced at OD600nm 1.0 for 4 hours using 1 mM IPTG.

Protein Analysis

Proteins were analysed by Western blotting and chemiluminescence detection using Penta-His Antibody (QIAGEN, for 6xHis-tagged proteins), Strep-tag Antibody (QIAGEN, for Strep II-tagged proteins), anti-pS473-AKT1 (Upstate, for detection of AKT1 kinase serine-phosphorylated at position 473) and an anti-mouse IgG secondary antibody conjugated to horseraddish peroxidase (Dianova) at dilutions recommended by the manufacturers.

Luciferase Activity Assay

Luciferase activity was determined using the Luciferase Assay System (Promega) according to the manufacturer’s recommendations.

Thymoma viral proto-oncogene 1, protein kinase B (h)

Interleukin-1 receptor-associated kinase 4 (h)

Mitogen-activated protein kinase kinase 3 (h)

cAMP-dependent protein kinase, beta catalytic subunit (h)

Cannabinoid receptor 1 (r)

DnaJ homolog subfamily C member 1 (h)

2-oxoglutarate/malate carrier protein (h)

Transient receptor ootential vanilloid 4 (cation channel, h)

Tumor necrosis factor alpha (h)

Transcription factor II A (precursor of alpha- and beta-subunits, h)

AKT1

IRAK4

MKK3

PKACβ

CNR1

Mtj1

OGCP

TrpV 4

TNFα

TFIIAαβ

Protein†

Table 1. Proteins and expression constructs

pET11a

pIX3.0

pET11a

pIX3.0

Plasmid DNA Construct E. coli E. coli in Vivo Cell-Free

pET11a

pIX2.0

pIX2.0

pTriEx-1

pIX4.0

pIX3.0

pIX4.0

pIX4.0

Insect II Cell-Free

continued on next page

N-His,C-His N-Strep, C-Strep N-His/C-Strep, N-Strep/C-His

N-His, C-His N-Strep, C-Strep N-His/C-Strep, N-Strep/C-His

No tag

N-His, C-His N-Strep, C-Strep, N-His/C-Strep, N-Strep/C-His

PCR Construct Insect II Cell-Free*

Advances in Insect-Based Cell-Free Protein Expression 21

Luciferase (f)

Erythropoietin (h)

Alpha-1-acid glycoprotein (h)

Glycoprotein 67 (bv)

TFIIAγ

Luc

EPO

ORM1

gp67

pIX4.0

pIX4.0

pIX4.0

pIX4.0

pIX3.0

Insect II Cell-Free

origin of proteins: bv, Baculovirus; f, Firefly; h, human; r, rat * generated using the EasyXpress Linear Template Kit PLUS (N/C-Strep: protein with a N- or C-terminal Strep II tag, respectively)



pIX3.0

Transcription factor II A (gamma subunit, h)

Protein† pIX3.0

Plasmid DNA Construct E. coli E. coli in Vivo Cell-Free

Table 1. Continued PCR Construct Insect II Cell-Free*

22 Cell-Free Protein Expression

Advances in Insect-Based Cell-Free Protein Expression

23

Strep Tag Pull-Down Assay

TFIIAγ was expressed with a N-terminal Strep tag and a C-terminal 6xHis tag (Strep-TFIIAγ-His) and TFIIAαβ with a N-terminal 6xHis (His-TFIIAαβ) tag in two separate 50 μl EasyXpress Insect II reactions. Strep-TFIIAγ-His was mixed with 50 μl Strep-Tactin Magnetic Agarose Bead suspension (QIAGEN) and processed as to the product manual for puriication of Strep-tagged proteins except that the elution step was not performed. Instead, the beads carrying immobilized Strep-TFIIAγ-His were subsequently incubated with the reaction mixture of the TFIIAαβ expression and treated as before. Finally, the TFIIAαβ/γ protein complex was eluted using 2.5 mM Biotin and analysed by Western blot detection with Penta-His Antibody. As a positive control, Strep-TFIIAγ-His was expressed and puriied protein eluted and analysed by Western blotting and Penta-His detection. As a negative control, TFIIAαβ was expressed and the reaction mixture directly incubated with Strep-Tactin beads. An aliquot of the elution fraction was analysed to check for unspeciic binding of His-TFIIAαβ to the Strep-Tactin matrix.

Protein Deglycosylation

Protein glycosylation was analysed by enzymatic deglycosylation and SDS-PAGE or by tunicamycin-mediated inhibition of glycosylation. PNGase F and EndoH deglycosidases were from New England Biolabs and used according to the manufacturer’s recommendations. Tunicamycin was from Sigma Aldrich and applied prior to the protein synthesis reaction in a inal concentration of 10 μg/ml reaction volume to inhibit protein glycosylation.

Results and Discussion Construct Screening

he design of an expression template has signiicant inluence on the eiciency of heterologous protein production. One reason may be mRNA secondary structure, leading to ineicient initiation of translation, improper folding of a polypeptide, unsuitable codon usage and others. Tag sequences and positions of these tags may also inluence the yield and solubility of the expressed protein.11 Furthermore, it is frequently not possible to heterologously express a protein of full length while a truncation may lead to success.12 hese observations point out that evaluation of many constructs is required to successfully express a heterologous protein and that parallel in vitro screening of several templates which have been constructed following diferent strategies speeds up protein production projects signiicantly. E. coli based cell-free expression based on linear templates generated by PCR is a fast and convenient methodology to determine the optimal expression construct.7,13,14 his approach was further investigated to determine PCR products which are suitable templates for the expression in insect cell-based in vitro systems. Two diferent tags, a 6xHis and a Strep-tag II, were introduced in diferent positions and combinations to investigate positional efects and the inluence of the amino acid composition. Figure 1A shows that both factors have a signiicant efect on expression rate and solubility of the produced proteins: A 6xHis tag located at the C-terminus gave lower yields compared to the tag positioned at the N-terminus of the test proteins shown. Similar efects can be observed for the Strep tag and in the case of TFIIAαβ a double-tagged construct (6xHisTFIIAαβStrep) resulted in higher yields of expressed protein than the N-terminally 6xHis-tagged protein. Soluble protein expression in E. coli was signiicantly afected by the chosen tag and the tag position.11 he underlying mechanism has not been thoroughly investigated but may be explainable with the inluence of introduced sequences on formation or avoidance of mRNA secondary structures thus afecting the eiciency of translation initiation. hese results demonstrate that construct screening is an important initial step in recombinant protein production projects. Insect cell expression, e.g., in Baculovirus/Sf-systems, can now be speeded up by using cell-free screening approaches. Furthermore, using PCR products as templates saves time by eliminating several rounds of transfection, virus construction and cell culture optimization. he best construct determined in the in vitro screening can subsequently be used for eicient baculovirus construction and cloning into a vector suitable for upscaling of protein production both in vitro and in vivo.

24

Cell-Free Protein Expression

Figure 1. Expression screening using PCR product-based template constructs. A) PCR products were generated by two-step PCR using all six possible tag combinations provided by the adapter primers in the EasyXpress Linear Template Kit PLUS. B) 2.5 μl aliquots from EasyXpress Insect II and 12.5 μl from RRL reactions were loaded onto a single SDS gel for Western blotting. Blots were probed using a mixture of Penta-His and Strep-tag antibodies. M: marker (6xHis protein ladder, size of marker bands indicated in kDa).

PCR product-based expression constructs giving the highest yield (see Fig. 1A, 6xHis-TFIIAαβ-Strep, Strep-TNFα, Strep-MKK3) were used as an expression template and cell-free expression eiciency was compared to a rabbit reticulocyte lysate system repeatedly reported to result in highly soluble and correctly folded protein preparations.15-18 Both systems gave >90% soluble protein whereas expression rates in the insect cell-based system were considerably higher. he amount of protein synthesized in the insect system exceeded the yield of protein obtained in the RRL cell-free system by a factor of 10-15 (Fig. 1B).

Soluble Expression of Human Proteins

Heterologous expression of human proteins was further investigated with proteins that were reported to be sparingly soluble when expressed in E. coli.19 Prokaryotic expression of TFIIA subunits αβ and γ and the human kinase MKK3 was investigated in vivo and in vitro and in the insect based cell-free system. Both subunits of the TFIIA complex were almost insoluble when expressed separately in E. coli in vivo and in vitro (Fig. 2, let and middle let panel). A functional and soluble complex could only be formed when the recombinant proteins were denatured and refolded together.19 Similarly, MKK3 was only partially soluble when expressed in E. coli-based systems. he same constructs were then used as templates in the EasyXpress Insect II extract. Figure 2 (middle right panel) shows that without further optimization on the template level more than 90% of the expressed proteins is soluble. hese results indicate that the Spodoptera rugiperda-derived

Advances in Insect-Based Cell-Free Protein Expression

25

Figure 2. Soluble expression of human proteins. Western blot analysis of human 6xHis-tagged proteins MKK3, TFIIAαβ and TFIIAγ expressed in EasyXpress E. coli- and Insect II-based cell-free extracts or in 5 ml cultures of E. coli BL21 (DE3) pLysS. 2.5 μl of cell-free extracts and 0.5% of BL21 crude (T, total) or cleared (S, soluble) lysates were loaded on a SDS gel and the blot probed with Penta-His antibody. Western blot analysis of human 6xHis-tagged integral membrane proteins CNR1, Mtj1, TrpV4 and OGCP expressed in EasyXpress Insect II extracts. 2.5 μl of cell-free extracts were loaded for SDS-PAGE and immuno detection. TrpV4 was expressed in the presence of 14C-Leucin and 8 μl aliquots separated by SDS-PAGE and analysed in a phosphoimager. The soluble fraction (S) was obtained by centrifugation at 15,000xg for 10 minutes and saving the supernatant. The construct for TrpV4 expression was a kind gift of Dr. Christian Harteneck, Institute of Pharmacology, Freie Universität Berlin, Germany and the construct for Mtj1 expression was from Prof. Richard Zimmermann, Medizinische Biochemie & Molekularbiologie, Universität des Saarlandes, Homburg, Germany.

extracts are more readily capable of expressing complex proteins such as heterodimeric TFIIA and homodimeric MKK3 in a soluble and therefore presumably functional form where E. coli-based systems fail or require optimization. To further investigate whether this also holds true for membrane proteins, another class of pharmacologically relevant proteins where soluble and functional expression frequently turns out to be diicult, diferent types of integral human membrane proteins were expressed and analysed. CNR1 (cannabinoid receptor 1) represents a GPCR (G-protein coupled receptor) located in the plasma membrane, Mtj1 is located in the endoplasmatic reticulum membrane,20 OGCP (2-oxoglutarate/malate carrier protein) is a transporter of the inner mitochondrial membrane and TrpV 4 represents a cation channel. All of these proteins could be expressed and could not be sedimented by centrifugation at 15,000xg for 10 minutes (Fig. 2, right panel) indicating that the proteins are properly folded.

Protein Function

To investigate the functionality of proteins expressed in the Insect II cell-free system irely luciferase enzymatic activity and protein-protein interaction of subunits αβ and γ of the human transcription factor IIA were analysed (Fig. 3). Firely luciferase activity is frequently used as a reporter system to determine activity of cell-free extracts and as model proteins when studying cellular processes such as refolding.10,21 Full activity of Luciferase seems to be dependent of the eukaryotic folding machinery as refolding of the denatured protein requires the Hsp90 chaperone.21 he luminescent protein was cloned into the pIX4.0 vector and expressed as an C-terminally 6xHis-tagged form in the Insect II extract (Fig. 3A, insert). Up to 40 μg active luciferase per ml reaction volume can be expressed as determined by a microtiter-based activity assay (Fig. 3A). he ability of the system to generate proteins capable of forming a protein complex was analysed in a pull-down assay with TFIIA subunits αβ and γ. Both subunits are expressed soluble in the Insect II extract (Fig. 2). For the pull-down assay, TFIIAγ was synthesized in a double-tagged form carrying a N-terminal Strep tag II and a C-terminal 6xHis tag. TFIIAαβ was N-terminally 6xHis-tagged. he Strep tag on the γ subunit was used to pull out the protein (complex) using

26

Cell-Free Protein Expression

Figure 3. Analyses of catalytic and binding activities. A) Luciferase assay; firefly luciferase was expressed in a Insect II 50 μl reaction from a vector pIX4.0 construct as a C-terminally 6xHis-tagged protein and analysed for luciferase activity in an activity assay using the crude extract (Insect II). A control expression (NTC, no template control) and activity assay was performed without DNA template added to the transcription/translation reaction. Measured activities were converted into μg luciferase protein expressed per reaction volume using a standard curve. The insert shows the expressed protein and a no template control (NTC) in a Western blot analysis using Penta-His antibody for detection of 6xHis-tagged protein. B) Pull-down assay; TFIIA subunits αβ (N-terminally 6xHis-tagged) and γ (N-terminally Strep-, C-terminally 6xHis-tagged) were expressed in 50 μl Insect II reactions and were successively incubated with equilibrated Strep-Tactin Magnetic Agarose beads. Beads were washed using wash buffer between the binding steps to preclude unspecific binding. Proteins bound specifically to the beads were eluted with 2.5 mM Biotin, 10 μl loaded on a SDS-PAGE and analysed by Western blotting using Penta-His antibody (lane 3). As controls, reactions with TFIIAγ (lane 1) and αβ (lane 2) expression alone were incubated, washed, eluted and analysed. M: marker (6xHis protein ladder, size of marker bands indicated in kDa).

Strep-Tactin magnetic agarose beads, the 6xHis tag allows the detection of the eluted protein complex. Strep-TFIIAγ-6xHis was bound to Strep-Tactin beads and could be eluted and detected when expressed individually (Fig. 3B, lane 1). As expected, 6xHis-TFIIAαβ failed to bind to Strep-Tactin beads (lane 2). To test complex formation between the αβ and γ subunits, extracts from γ and αβ expression assays were incubated with the beads, washed and eluted with Biotin. Both subunits can be detected (lane 3) indicating that they formed a stable heterodimeric TFIIA complex. his result was also achieved when both subunits were coexpressed in a single reaction (data not shown).

Posttranslationally Modiied Proteins

Many proteins require posttranslational modiications to display functional activity or even solubility. Two of the most common modiications, glycosylation and phosphorylation have been analysed. hree glycoproteins were expressed in the Insect II extract: ORM1 (alpha-1-acid glycoprotein), one of the most abundant glycoproteins in human plasma, EPO (erythropoietin), a hormone which stimulates the formation of erythrocytes and the baculoviral glycoprotein gp67. For these proteins, Western blotting and autoradiography show the multi-band pattern typical for glycoproteins (Fig. 4A). To analyse whether this band pattern is originated from glycosylation,

Advances in Insect-Based Cell-Free Protein Expression

27

Figure 4. Posttranslational modifications: analysis of glycosylation and phosphorylation. A) 6xHis-EPO and 6xHis-ORM1 were expressed in 50 μl Insect II reactions, samples treated according to the recommendations of the EndoH manufacturer and splitted into 2 aliquots. One aliquot was loaded directly on a SDS-PAGE (lanes marked -), the other one was incubated with EndoH before loading onto the gel (lanes marked +). Western blotting was performed using Penta-His antibody. Gp67 was expressed in two separate 50 μl Insect II reactions in the presence of 14C-Leucin whereas one reaction (marked +) was supplemented with 10 μg/ml tunicamycin. 8 μl aliquots of the reactions were separated by SDS-PAGE and analysed in a phosphoimager. B) Strep-tagged PKACβ and IRAK4 and 6xHis-tagged AKT1 kinases were expressed in 50 μl cell-free reactions and aliquots loaded onto a SDS-PAGE and analysed by Western blotting using anti-Strep and Penta-His antibodies, respectively (upper panels: Insect II reactions, 2.5 μl loaded per lane; lower panels: RRL reactions, 12.5 μl loaded per lane). Anti-pS 473 monoclonal antibody was used to probe for phosphorylation of AKT1 (upper and lower right panels). Sizes of expressed proteins are indicated in kDa on the left.

the proteins were incubated with EndoH, a endoglycosidase enzyme capable of cleaving N-linked carbohydrates, or tunicamycin, an inhibitor of protein N-glycosylation,22 was added prior to the translation reaction. Tunicamycin led to formation of a smaller gp67 protein compared to the gp67 from an untreated synthesis reaction (Fig. 4A, right panel) indicating that the slower migration was due to the formation of protein N-glycosidic linkages. Endo H treatment reduced the apparent

28

Cell-Free Protein Expression

molecular weight of both the EPO and ORM1 glycoprotein indicating that the blood proteins are eiciently glysosylated in the insect cell-based extracts (Fig. 4A, let panel). Comparable results were achieved using the peptide N-glycosidase F (PNGaseF, data not shown). Conirmation of the importance of glycosylation for the blood proteins was obtained when solubility analyses were performed. A residual, unglycosylated part of the protein expressed was insoluble and could be sedimented whereas the glycosylated forms of ORM1 and EPO remained in the supernatant ater centrifugation at 15,000 × g for 10 minutes (data not shown). Glycosylation processes take place in the ER and the Golgi of eukaryotic cells. he import of protein into the ER is followed by the cleavage of their signal peptide. he results of the experiments shown in Figure 4A demonstrate that the insect cell-derived extracts are capable of such a signal peptide cleavage. Cell-free extracts derived from eukaryotic cells have been described to have protein phosphorylation activity.23,24 We analysed the expression of three human kinases, AKT1, IRAK4 and PKACβ, in the Insect II extracts and in parallel in the rabbit reticulocyte lysate. Expression of the three kinases was low in E. coli (data not shown). In contrast, eicient protein expression was observed in the Insect II extract (Fig 4B, upper let and middle panels). he kinases could also be expressed in RRL but at a rate approximately one order of magnitude lower (Fig. 4B, lower let and middle panels). AKT1 is a well characterized serine/threonine kinase which plays an important role in balancing cell survival and apoptosis and is involved in several signaling pathways connected to these tasks.25 AKT1 has two phosphorylation sites, one within the kinase domain at T308, one on the hydrophobic tail on S473 and is regarded as active if phosphorylated at this serine residue.26 An antibody directed against pS473 has been described which was shown to recognize active AKT kinases solely.27,28 his antibody was used to probe AKT1 expressed in Insect II and RR lysates. Figure 4B (upper and lower right panels) shows phosphorylated AKT1 at S473 in insect-derived extracts. A faint band was also visible for the RRL-expressed protein when loading the ive-fold sample volume compared to the insect lysate. his experiment demonstrates that the insect-derived cell-free extract is able to produce phosphorylated proteins such as human kinases. Further experiments will be performed to conirm the catalytic activity of AKT1 and other human kinases.

Conclusion

Protein research projects are turning strongly towards pharmacologically relevant proteins such as human and eukaryotic proteins in general and membrane proteins in particular. his is demonstrated by the fact that the Protein Structure Initiative II funded by the National Institue of General Medical Sciences (NIGMS), includes three new structural genomics consortia which concentrate speciically on eukaryotic and membrane proteins. Several of such proteins demand complex folding machineries and posttranslational modiications as well as further processing to display maximum functional activity. herefore, eukaryotic expression systems are expected to become more widely used in the near future as they fulill the listed needs and due to the fact that some proteins cannot be produced eiciently in E. coli-based systems. However, protein production in eukaryotic cells also has its drawbacks such as diicult transfection technologies, slow cell growth and high costs. Here, cell-free expression represents an attractive alternative for synthesis of recombinant proteins as it is a convenient, fast and easy to handle procedure, requiring only minimal lab equipment and making knowledge in cell culture techniques obsolete. We have shown recently10 and in the work presented here that insect cell-based in vitro synthesis of proteins allows expression of soluble and active human proteins including membrane proteins such as GPCR’s. he formation of posttranslational modiications in particular expands the possibilities of eukaryotic cell-free systems. he use of PCR products as linear DNA templates enables multiparallel expression construct generation and evaluation of protein expression within one working day. his reduces costs signiicantly and speeds up insect cell based protein production projects as a single or few in vitro optimized expression constructs can be chosen for upscaling expression in Spodoptera frugiperda cell culture, the presently most widely used eukaryotic protein production system.29 Ongoing Experiments will demonstrate the correlation of protein quality produced in vitro versus in vivo in insect cells which had previously been shown for E. coli.7,8

Advances in Insect-Based Cell-Free Protein Expression

29

Furthermore, the increased expression rates and the functionality of the proteins produced in the Insect II system make it the ideal system to perform protein activity studies such as enzymatic screens and membrane protein assays.

References

1. Kigawa T, Yamaguchi-Nunokawa E, Kodama K et al. Selenomethionine incorporation into a protein by cell-free synthesis. J Struct Functl Gen 2001; 2:29-35. 2. Klammt C, Löhr F, Schäfer B et al. High level cell-free expression and speciic labeling of integral membrane proteins. Eur J Biochem 2004; 271:568-580. 3. Kainosho M, Torizawa T, Iwashita Y et al. Optimal isotope labelling for NMR protein structure determinations. Nature 2006; 440:52-57. 4. Budisa N. Expression of ‘tailor-made’ proteins via incorporation of synthetic amino acids by using cell-free protein synthesis. In: Swartz JR, ed. Cell-free protein expression, 1st ed. Berlin: Springer Verlag 2003; 89-98. 5. Agafonov DE, Huang Y, Grote M et al. Eicient suppression of the amber codon in E. coli in vitro translation system. FEBS Lett 2005; 579:2156-2160. 6. Gerrits M, Strey J, Claußnitzer I et al. Cell-free synthesis of deined protein conjugates by site-directed cotranslational labeling. In: Kudlicki W, ed. Cell-free protein expression, 1st ed. Austin: RG Landes Co 2007. 7. Lamla T, Hoerer S, Bauer MMT et al. Screening for soluble expression constructs using cell-free protein synthesis. Int J Biol Macromol 2006; in press, available on-line: doi:10.1016/j.ijbiomac 2006. 8. Murthy TVS, Wu W, Qiu QQ et al. Bacterial cell-free system for high-throughput protein expression and a comparative analysis of Escherichia coli cell-free and whole cell expression systems. Prot Expr Purif 2004; 36:217-225. 9. Vinarov DA, Lytle BL, Peterson FC et al. Cell-free protein production and labeling protocol for NMR-based structural proteomics. Nature Meth 2004; 1(2):1-5. 10. Kubick S, Schacherl J, Fleischer-Notter H et al. In vitro translation in an insect-based cell-free system. In: Swartz JR, ed. Cell-free protein expression, 1st ed. Berlin: Springer Verlag 2003; 209-217. 11. Zacharias A, Schäfer F, Müller S et al. Recombinant-protein solubility screening using the EasyXpress in vitro translation system. QIAGEN, Qiagen News 2004; e6. 12. Cornvik T, Dahlroth S-L, Magnusdottir A et al. Colony iltration blot: a new screening method for soluble protein expression in Escherichia coli. Nature Meth 2005; 2:507-509. 13. Nemetz C. Generation of linear expression elements by PCR. In: Swartz JR, ed. Cell-free protein expression, 1st ed. Berlin: Springer Verlag 2003; 3-7. 14. Merk H, Meschkat, D, Stiege W. Expression-PCR: from gene pools to puriied proteins within 1 day. In: Swartz JR, ed. Cell-free protein expression, 1st ed. Berlin: Springer Verlag 2003; 209-217. 15. Blagosklonny MV, Toretsky J, Bohen S et al. Mutant conformation of p53 translated in vitro or in vivo requires functional HSP90. Proc Natl Acad Sci 1996; 93:8379-8383. 16. Murthy MS, Pande SV. A stress-regulated protein, GRP58, a member of thioredoxin superfamily, is a carnitine palmitoyltransferase isoenzyme. Biochem J 1994; 304:31-34. 17. Fessing MY, Belkov VM, Krynetski EY et al. Molecular cloning and functional characterization of the cDNA encoding the murine thiopurine S-methyltransferase (TPMT). FEBS Lett 1998; 424:143-145. 18. Shin D-Y, Ishibashi T, Choi TS et al. A novel human ERK phosphatase regulates H-ras and v-raf signal transduction. Oncogene 1997; 14:2633-2639. 19. Sun X, Ma D, Sheldon M et al. Reconstitution of human TFIIA activity from recombinant polypeptides: a role in TFIID-mediated transcription. Genes Dev 1994; 8:2336-2348. 20. Dudek J, Volkmer J, Bies C et al. A novel type of cochaperone mediates transmembrane recruitment of DnaK-like chaperones to ribosomes. EMBO J 2002; 21(12):2958-2967. 21. Schneider C, Sepp-Lorenzino L, Nimmesgern E et al. Pharmacologic shiting of a balance between protein refolding and degradation mediated by Hsp90. Proc Natl Acad Sci 1996; 93:14536-14541. 22. Heifetz A, Keenan RW, Elbein AD. Mechanism of action of tunicamycin on the UDP-GlcNAc: dolichyl-phosphate GlcNAc-1-phosphate transferase. Biochem 1979; 18:2186-2192. 23. Prochownik EV, VanAntwerp ME. Diferential patterns of DNA binding by myc and max proteins. Proc Natl Acad Sci 1993; 90:960-964. 24. Joshi B, Cai A-L, Keiper BD et al. Phosphorylation of eukaryotic protein synthesis initiation factor 4E at Ser-209. J Biol Chem 1995; 270:14597-14603. 25. Bellacosa A, Testa JR, Moore R et al. A portrait of AKT kinases. Cancer Biol her 2004; 3:268-275. 26. Au CS, Wagner A, Chong T et al. Insulin regulates hepatic apolipoprotein B production independent of the mass or activity of AKT/PKBalpha. Metabolism 2004; 53:228-235.

30

Cell-Free Protein Expression

27. Chan TO, Rittenhouse SE, Tsichlis PN. AKT/PKB and other D3 phosphoinositide-regulated kinases: kinase activation by phosphoinositide-dependent phosphorylation. Ann Rev Biochem 1999; 68:965-1014. 28. Sun M, Wang G, Paciga JE et al. AKT1/PKBα kinase is frequently elevated in human cancers and its constitutive activation is required for oncogenic transformation in NIH3T3 cells. Am J Pathol 2001; 159:431-437. 29. Bahia D, Cheung R, Buchs M et al. Optimisation of cell growth in deep-well blocks: development of a high-throughput insect cell expression screen. Prot Expr Purif 2005; 39:61-70.

Chapter 3

Advantages and Applications of the Batch-Formatted E. coli Cell-Free Expression System Julia E. Fletcher, Federico Katzen, Shiranthi Keppetipola, Ashley Getbehead and Wieslaw A. Kudlicki*

Abstract

F

rom the irst procedure standardized in the early 1960s to its most recent modiications, cell-free expression using Escherichia coli lysates has evolved into a robust method for protein expression capable of producing protein in milligram quantities. Free from the constraints of a cell, the components of the E. coli-based cell-free system can be modiied easily to customize its use for studying all aspects of protein expression. In this chapter we will discuss the latest advances in E. coli-based cell-free protein synthesis focusing on the batch method and its use in high-throughput and structural applications.

Introduction

One of the most eicient cell-free systems for protein synthesis utilizes extracts prepared from Escherichia coli (E. coli). In contrast to eukaryotic systems, protein translation is less complex in E. coli because its translational apparatus has fewer components than the eukaryotic counterparts and translation initiation is less tightly regulated.1,2 As in some eukaryotic-derived systems, translational eiciency is enhanced in the E. coli lysate by coupling transcription and translation so ribosomes start translating the mRNA while it is still being transcribed. Coupled transcription/translation makes it possible to start the protein synthesis reaction with a DNA template instead of exogenously added mRNA. Being an open system, the cell-free synthesis reaction provides an environment that is easily controlled and manipulated. Cell-free systems are also tolerant to a number of toxic proteins unlike whole cells where over-expression of such proteins results in cell death. Standardized procedures for the preparation of E. coli-based cell-free systems were reported in 1963 with the publication of the irst method for cell-free protein synthesis3 followed by the slightly modiied and more popularly referenced methods of Zubay and Pratt.4,5 his early work supplied the foundation for the development of a number of cell-free systems from both prokaryotic and eukaryotic organisms (for further details see additional reviews6-8). While eukaryotic-based cell-free lysates are recognized for their ability to post-translationally modify synthesized proteins, the E. coli-based cell-free system remains the most productive and cost eicient means to synthesize protein in milligram quantities. he expanded capabilities of the E. coli-based cell-free system have evolved from a succession of technical advances. his chapter will *Corresponding Author: Wieslaw A. Kudlicki—Invitrogen Corporation, 1610 Faraday Avenue, Carlsbad, California 92008, U.S.A. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

32

Cell-Free Protein Expression

examine the most recent improvements to E. coli-based cell-free expression system. In particular, the batch format will be discussed as most of the advantages of cell-free synthesis can be realized only when this format provides acceptable and consistent product yields.

High-Yield Formats Membrane Exchange Systems

he translational activity of an E. coli-based cell-free system in a simple batch reaction has a limited life span (Fig. 1A). he reason for loss in activity is attributed to the depletion of reaction substrates by the coupled transcription/translation machinery (i.e., NTPs, amino acids, primary energy sources) and the accumulation of inhibitory by-products (i.e., nucleosides monophosphates, inorganic phosphates, polypeptides).9 Solutions for prolonging the reaction irst came in the form of a matrix-immobilized solid-phase translation system where proteins could be synthesized from template polynucleotides immobilized to a resin, which could be washed and recharged with new substrates.10,11 In an adaptation of this idea, substrates and byproducts can be exchanged through a semipermeable barrier as demonstrated by the continuous-low cell-free (CFCF) (Fig. 1B) and the continuous-exchange cell-free (CECF) (Fig. 1C) methods.12-14 In CFCF vessels substrates are continuously supplied to the reaction, and the reaction by-products are continuously removed via a semi-permeable membrane. he disadvantage of this approach is the need for a pump to provide continuous low across the membrane. In the CECF reaction exchange of components occurs passively through the membrane. he original CECF system in E. coli has been improved by (1) including exogenous RNA polymerase like T 7,15 (2) modifying the energy system16 and (3) modifying the E. coli extract preparation.16,17 he widespread use and popularity of the E. coli CECF system has increased since the introduction of the Rapid Translation System (RTS) by Roche Diagnostics. he RTS/CECF system provides instruments and dialysis reactors for continuous cell-free transcription and translation in a customized reaction. However, the format can be cumbersome for experiments that necessitate automation or miniaturization. he requirement of

Figure 1. Diagram comparing cell-free reaction formats. Formats are classified according to how the reaction is fed: A) Batch, B) Continuous-Flow cell-free, C) Continuous exchange cell-free, D) Thin layer, E) Bilayer and F) Batch/feed.

Advantages and Applications of the Batch-Formatted E. coli Cell-Free Expression System

33

a membrane and/or instrumentation for extending translational activity make continuous systems generally less applicable to high-throughput processes and have led several groups to reinvestigate new ways to improve the batch reaction.

Longer-Lived Batch Systems

he batch format ofers several advantages over the CECF system. First, the batch reaction chamber is freely accessible (as opposed to the inner chamber of the CECF system). Next, the batch format does not require complex instrumentation. Finally, batch reaction volumes can be scaled (up and down) more easily making this method highly amenable for automated/high-throughput expression screening. Several laboratories have found ways to improve the activity and/or longevity of the E. coli-based cell-free batch reaction by optimizing the reaction components. Swartz’s group, for example, has done a number of studies examining alternative energy sources, conditions and extract preparation for a more eicient, cost efective batch reaction. Some of the improvements include (1) the use of phosphoenolpyruvate (PEP) as a secondary energy source,18,19 (2) the development of a cytoplasmic-like environment for protein translation,20,21 (3) the reduction in time and cost of preparing the E. coli extract22,23 and (4) the use of glucose as an energy source and replacement of nucleoside triphosphates (NTPs) with monophosphates (NMPs).24 In other areas, Kim and coworkers found they could counteract the chelation of magnesium caused by inorganic phosphates by repeated magnesium repletion.25 Altering growth conditions during preparation of the S30 E. coli lysate also has counteracted phosphate accumulation (i.e., toxic by-product).26 A more comprehensive approach to controlling reaction components is the complete reconstitution of the essential elements and enzymes of the E. coli translational machinery, which is found in the PURE system translation system.27,28 All components in this system are puriied separately through the use of histidine tags and can be added individually or eliminated depending on the application. However, reconstitution of the translation machinery by this method is labor intensive and reaction scale-up becomes cost prohibitive for synthesizing preparative quantities of proteins.

New Conigurations

Other alternatives to optimize the coniguration of the translation machinery have also been investigated. Swartz and coworkers found that by using a “thin ilm” approach, protein yields could be maintained during scale-up in the batch mode by increasing the volume to surface ratio for gas transfer and by providing a hydrophobic surface for improved protein folding (Fig. 1D).29 Another example is the bilayer method (irst developed with wheat germ but easily applied to E. coli).30,31 In the bilayer method (Fig. 1E) a dense liquid reaction phase is overlaid by a less dense substrate phase allowing exchange to occur through passive difusion without a membrane. he main beneit is a slower reaction time for improved protein folding. A disadvantage to both the thin-ilm and bilayer formats is the need to maintain surface uniformity, which limits the ability to alter reaction substrates and kinetics. Previous studies have shown that repeated addition of certain substrates during the time course of the reaction could signiicantly improve protein yields.18,21,32 In our own lab, we have modiied this approach and developed a new system (ExpresswayTM Maxi/NMR) in which substrates can be supplied to an E. coli base reaction by supplemental “feeds” (Fig. 1F). By restoring depleted energy components and diluting spent byproducts, the feeding approach signiicantly improves the protein yield (see next section). A summary of the key beneits and limitations of each method is presented in Table 1. he following sections will highlight some applications where batch-formatted E. coli cell-free expression has been used to further enhance protein expression and analysis. hese areas include: luorescent labeling of proteins, automation, development of protein arrays, membrane protein crystallization and protein synthesis for NMR analysis.

34

Cell-Free Protein Expression

Table 1 E. coli cell-free synthesis formats and applications

Configuration Batch

CFCF

CECF

Bilayer

Thin Film

PURE

Batch/Feed

Description

Applications, Advantages and Disadvantages

• Synthesizes ug range of proteins in 2 hours • Protein engineering and • No dilution of reaction byproducts functional assays • Scale-up volume not limited to reactor size • Short-lived reactions • Automation • Synthesizes mg range of proteins • X-ray crystallography, NMR in 24 hours and protein engineering and functional assays • Reaction byproducts removed • Long-lived reactions • Scale-up volume limited by reactor size • Not useful for automation • Ten times more labeled amino acids • Less cost effective for protein needed than batch/feed reaction engineering • Synthesizes mg range of proteins • X-ray crystallography, NMR, in 24 hours protein engineering and • Reaction byproducts removed functional assays • Scale-up volume limited by reactor size • Long-lived reactions • Ten times more labeled amino acids • Not useful for automation needed than batch/feed reaction • Less cost effective for protein engineering • Synthesizes ug range of proteins in • Protein engineering and 24 hours functional assays • Phase integrity must be maintained • Short-lived reactions • Not useful for X-ray crystallography and NMR • Not useful for automation • Synthesizes ug range of proteins in 6 hours • Protein engineering and functional assays • Scale-up volume limited by reactor size • Short-lived reactions • Reaction byproducts not diluted • Not useful for X-ray crystallography and NMR • Phase integrity must be maintained • Less useful for automation • Synthesizes ug range of protein in hours • Protein engineering and functional assays • Native protein expression • Native purification of proteins • No dilution of reaction byproducts • Short-lived reactions • Highly specific labeling without • Protein engineering background • Not useful • Scale-up cost prohibitive X-ray crystallography and NMR • Synthesizes mg range of proteins in • X-ray crystallography, NMR, 6 hours protein engineering, functional • Reaction byproducts diluted assays and automation • Scale-up volume not limited to reactor size • Long-lived reactions • Ten times less labeled amino acids needed • Automation than CFCF and CECF reactions • More cost effective for NMR and protein engineering

Advantages and Applications of the Batch-Formatted E. coli Cell-Free Expression System

35

High-hroughput Cell-Free Expression of Fluorescently Labeled Proteins

he completion of the Human Genome Project has led to the establishment of annotated banks of sequence-veriied open reading frames (ORFs) as essential tools in the functional characterization of genetic information for proteomic and functional genomics studies. he systematic analysis of protein function requires the rapid expression of the corresponding ORFs. he expression of these large collections using the traditional in vivo systems is labor intensive. Batch-formatted cell-free protein expression ofers a streamlined platform for high-throughput expression screening.33 A way to further expedite sample processing is to use a sensitive luorescent protein visualization method that overcomes the limitations of the current staining or immuno-based detection procedures. We have combined several technologies to accelerate the analysis of these large clone collections. For the cloning part, we have employed the Gateway recombination system34 that makes use of lambda phage site-speciic recombination proteins to shuttle genes among diferent genetic backbones. hus, we chose one of the Gateway-Based available ORFeomes, (Invitrogen’s human UltimateTM ORF collection)35 as the source of the genetic material. A subset of this collection was transferred simultaneously via recombination into a destination vector that encodes a Cys-Cys-Pro-Gly-Cys-Cys tag. he resulting plasmids were then used to program multi-well cell-free protein expression reactions. One of the key advantages of this method is that protein

Figure 2. Real-time synthesis and visualization of N-terminal and C-terminal Lumio™-tagged Kinases. Three human kinases were expressed and labeled simultaneously using batchformatted Expressway™ reactions (BKG, background signal). Lumio™ Green was added directly to the reaction. Products were followed by fluorometry as they emerged from the ribosome (A) or by in-gel visualization under a laser scanner. B) Schematic principle of discretional c-terminal labeling. C) Example of discretional c-terminal labeling. D) Upper panel: in-gel visualization of the products under laser scanner. Lower panel: autoradiograph exposure of the same gel. Symbols (“+” and “–“) indicate the presence or absence of suppressor tRNA in the reaction. Abbreviation: supp, suppressors.

36

Cell-Free Protein Expression

expression levels can be monitored directly by in-gel luorescence in real-time since the tag covalently binds to the detection reagent (Lumio),36 which becomes luorescent upon binding to the target (Fig. 2A). Alternatively, the labeled proteins can be loaded on polyacrylamide gels and the proteins can be visualized by exposing the gel to UV light or scanned by laser densitometry (Fig. 2B). he system works both with N-terminal or C-terminal fusions. In the latter case, an amber suppressor tRNA is added to the reactions (Fig. 2C). he synthesis and detection of full-length luorescently labeled proteins for rapid expression screening or synthesis of corresponding untagged variants is possible from the same construct (Fig. 2D).

Automation Robotic operating systems ofered by Tecan, Zymark, Beckman and PerkinElmer, for example, provide multiple conigurations for the automation of high-throughput experiments. hese systems are indispensable for screening hundreds of thousands of samples and play a critical role in the pharmaceutical industry’s search for drug targets. he Gateway® cloning format and batch-based systems described above are easily adapted to robotics. Starting from either 96 well or 384 well plates, Gateway® recombination reactions can be automated to produce template DNA for protein synthesis in batch-formatted reactions (see worklow schematic, Fig. 3). For a complete description of the application please see http://www.beckman.com/resourcecenter/labresources/automatedsolutions/t3_articles_1q2005/invitorgen_orf_pres_032005.asp Also, scripts for TECAN or BECKMAN lab automation systems are available at http://www.invitrogen.com/Expressway

Figure 3. Schematic of a workflow for the automation of ExpresswayTM Maxi reactions.

Advantages and Applications of the Batch-Formatted E. coli Cell-Free Expression System

37

Bacterial Protein Arrays

Protein micro and macro arrays are emerging as powerful tools for analyzing large numbers of proteins for their binding capabilities and functional interactions.37-41 Usually, cell-based protein expression, is used as the primary method to generate the protein samples that will be used for the arrays.42-44 As the number of proteins increases, the in vivo expression approach becomes more laborious and less practical. he lexibility of the batch-formatted cell-free synthesis systems overcomes this shortcoming. First, it has been demonstrated that the batch-formatted reactions can be downscaled to nanoliter volumes making more amenable the screening of large libraries (genes or compounds).45 Remarkably, this level of miniaturization has not been achieved with any other expression system. he second approach is the use of the cell-free systems to express individual tagged proteins from PCR templates directly on a tag-binding surface; the products are immobilized on a surface as they are synthesized, bypassing the need to purify the proteins.46 he third technology originates from a nucleic acid programmable protein array (NAPPA), where the targets are expressed from immobilized cDNAs and captured in situ by means of a tag-binding molecule that is printed on the surface next to the DNA template.47 he reaction components are overlaid onto the arrays where targets and baits are expressed simultaneously. his approach eliminates the need to express and purify proteins separately and produces proteins at the time of the assay, reducing the concerns about protein stability during storage. Although this technology has been described for eukaryotic-based systems, it can be easily adapted to a prokaryotic batch-formatted cell-free expression system for those instances where post-translational modiications are not critical. An example of this are the bacterial protein arrays, which are being used mostly for the identiication of diagnostic markers of infectious diseases.48 One speciic example is the generation of Yersinia pestis (Y. pestis) protein arrays using proteins expressed in the batch/feed reaction format (please see Michaud’s chapter in this book for more details). In this case, cell-free expression signiicantly reduced the lengthy preparation time of samples previously expressed in vivo.

Membrane Protein Expression

Membrane proteins account for nearly a third of the genes encoded by most fully sequenced genomes, including the human genome. Because approximately 50% of current drug targets are membrane proteins, the importance and relevance for this class of proteins in drug discovery is tremendous. Currently, research and development with membrane proteins is impeded by a lack of availability to this class of proteins. Over-expression of membrane proteins in vivo oten results in cell toxicity and protein aggregation, obstacles that can be overcome by cell-free protein expression systems. A variety of groups have been successful in synthesizing up to milligram quantities of membrane proteins per milliliter of reaction using cell-free expression technologies. Strategies ranging from the use of non-ionic detergents, lipids, to the solubilization of inclusion bodies and the use of solubility tags have been described.49-55 he PURE system (see above) has also been adapted for membrane protein expression by the inclusion of inverted vesicles containing the translocation machinery allowing membrane proteins to be synthesized and correctly folded.56 In another approach a synthetic biological system has been developed for expressing membrane proteins in vesicle bioreactors consisting of the E. coli cell-free machinery encapsulated in large unilamellar vesicles.57,58 Finally, it is important to mention that the irst crystal structure of a membrane protein synthesized in vitro has recently been described.59 Basically, a modiied version of the batch/feed system (see above) was employed, where dimyristoylphosphatidylcholine liposomes were supplemented to the reaction. hese results emphasize another advantage of the batch-formatted cell-free coniguration since this strategy would have been more diicult to apply in a CECF reactor as liposomal particles may stick to the permeable membrane. More details of this method are presented in Chang’s chapter in this book. For further details on the expression of membrane proteins in a cell-free context please see Bernhard’s chapter in this book.

38

Cell-Free Protein Expression

NMR Analysis

Nuclear magnetic resonance (NMR) spectroscopy is another method for resolving protein structure at the atomic level. NMR has the distinct advantage of providing not only structural information but also a more dynamic view of the movement of individual sites in a protein. he major limitation of this technique is inherent to the lack of sensitivity dictated by the size of the instruments needed for measurement. Limits in the spatial resolution have placed constraints on the size of proteins, which can be analyzed (generally ≤ 30kDa). As with X-ray crystallography, NMR analysis requires milligram quantities of protein. Recent advances in E. coli cell-free expression have made it an excellent tool for NMR analysis (for a recent review see ref. 60). Advantages include but are not limited to enhanced sample stability, reduced amino acid scrambling, minimal sample puriication strategies, lower amounts of heavy amino acids needed, higher perdeuteration eiciency, the possibility to introduce diferent heavy isotopes at predetermined positions and the use of multiplexed reactions. Recently, a high-throughput NMR protocol was developed using the batch/feed system where enough uniformly 15N-labeled protein was expressed in less than 5 hours to get a high resolution 15N-1H HSQC spectra without the need of purifying the sample (Fig. 4).61 For further details on NMR applications derived from the use of cell-free expression systems please see Shaka’s chapter in this book.

he Future

he advances in E. coli-based cell-free technology have progressed due to an increased need for new tools to understand protein structure and function in the post-genomic era. he future of E. coli-based cell-free expression lies in the next generation of improvements facilitating expression of diicult-to-express proteins and examining their function. Already, E. coli-based cell-free expression has been used to make improvements in the ield of protein evolution. his area, which aims at improving enzyme technology and protein: protein interactions, is based on the strategy of physically linking the nucleic acids with its protein product allowing multiple rounds of ampliication and selection. In contrast to traditional whole cell phage display, the cell-free methods of ribosome display, mRNA display and in vitro compartmentalization have improved the dynamic range of this technology while reducing the time and cost of

Figure 4. Protein expression and labeling in the cell-free reaction provides protein samples sufficient for NMR analysis without the need for purification. A) 800 MHz 15 N-1H HSQC of the purified protein SUMO-1 B) 800 MHz 15 N-1H HSQC of crude SUMO-1 reaction mixture. A full description of this technique can be found in Keppetipola et al.61 Reprinted with permission from Keppetipola et al.61 Copyright 2006 American Chemical Society.

Advantages and Applications of the Batch-Formatted E. coli Cell-Free Expression System

39

experimental preparation (for more details see reviews in refs. 62,63). Again, the batch-formatted cell-free reaction is the only coniguration compatible with this technology. In other areas, E. coli-based cell-free expression has been implemented as a great tool for synthesizing proteins with unnatural amino acids as researchers ind new applications for manipulating the genetic code.64 Eliminating issues of permeability found in vivo, the direct incorporation of nonnatural amino acids using the cell-free technology enables a full range of protein engineering to analyze protein function by luorescent tagging, PEGylation and glycosylation to name a few.65 Finally, the improvements, which reduce the cost of generating the cell-free reaction components, are now making it a viable option for protein bioproduction. he full potential of the E. coli-based cell-free expression system will be realized only with time. For now, the batch-formatted system and applications described here are an excellent indication of things to come.

Acknowledgements

We are very grateful to Professor A.J. Shaka and his group for generating the NMR spectra from the cell-free expression reactions. We also would like to thank Dr. Jon Chesnut and his group for their help and guidance with the Gateway® automation experiments and protocol development.

References

1. Kozak M. Regulation of translation via mRNA structure in prokaryotes and eukaryotes. Gene 2005; 361:13-37. 2. Kozak M. Initiation of translation in prokaryotes and eukaryotes. Gene 1999; 234(2):187-208. 3. Nirenberg MW. Cell-free protein synthesis directed by messenger RNA. Methods Enzymol 1963; 6:17-23. 4. Zubay G. In vitro synthesis of protein in microbial systems. Annu Rev Genet 1973; 7:267-287. 5. Pratt JM. Coupled transcription-translation in prokaryotic cell-free system. In: Hames BD, Higgins, S.J., eds. Transcription and Translation: A Practical Approach. Oxford, UK: IRL Press; 1984:179-209. 6. Katzen F, Chang G, Kudlicki W. he past, present and future of cell-free protein synthesis. Trends Biotechnol 2005; 23(3):150-156. 7. Spirin AS. Cell-free translation systems. New York: Springer 2002. 8. Swartz JR. Cell-free protein expression. Berlin: Springer 2003. 9. Kim DM, Choi CY. A semicontinuous prokaryotic coupled transcription/translation system using a dialysis membrane. Biotechnol Prog 1996; 12(5):645-649. 10. Belitsyna NV, Girshovich AS, Spirin AS. [Translation of a polynucleotide matrix on a solid carrier]. Dokl Akad Nauk SSSR 1973; 210(1):224-227. 11. Belitsina NV, Spirin AS. Translation of matrix-bound polyuridylic acid by Escherichia coli ribosomes (solid-phase translation system). Methods Enzymol 1979; 60:745-760. 12. Spirin AS, Baranov VI, Ryabova LA et al. A continuous cell-free translation system capable of producing polypeptides in high yield. Science 1988; 242(4882):1162-1164. 13. Baranov VI, Spirin AS. Gene expression in cell-free system on preparative scale. Methods Enzymol 1993; 217:123-142. 14. Alakhov YB, Baranov VI, Ovodov SY et al. Method of preparing polypeptides in a cell-free translation system. US Patent 5478730. 1995(991757). 15. Baranov VI, Morozov I, Ortlepp SA et al. Gene expression in a cell-free system on the preparative scale. Gene 1989; 84(2):463-466. 16. Kigawa T, Yabuki T, Yoshida Y et al. Cell-free production and stable-isotope labeling of milligram quantities of proteins. FEBS Lett 1999; 442(1):15-19. 17. Kigawa T, Yabuki T, Matsuda N et al. Preparation of Escherichia coli cell extract for highly productive cell-free protein expression. J Struct Funct Genomics 2004; 5(1-2):63-68. 18. Kim DM, Swartz JR. Prolonging cell-free protein synthesis with a novel ATP regeneration system. Biotechnol Bioeng 1999; 66(3):180-188. 19. Kim DM, Swartz JR. Regeneration of adenosine triphosphate from glycolytic intermediates for cell-free protein synthesis. Biotechnol Bioeng 2001; 74(4):309-316. 20. Jewett MC, Swartz JR. Mimicking the Escherichia coli cytoplasmic environment activates long-lived and eicient cell-free protein synthesis. Biotechnol Bioeng 2004; 86(1):19-26. 21. Jewett MC, Swartz JR. Substrate replenishment extends protein synthesis with an in vitro translation system designed to mimic the cytoplasm. Biotechnol Bioeng 2004; 87(4):465-472.

40

Cell-Free Protein Expression

22. Liu DV, Zawada JF, Swartz JR. Streamlining Escherichia coli S30 extract preparation for economical cell-free protein synthesis. Biotechnol Prog 2005; 21(2):460-465. 23. Zawada J, Swartz J. Efects of growth rate on cell extract performance in cell-free protein synthesis. Biotechnol Bioeng 2006; 94(4):618-624. 24. Calhoun KA, Swartz JR. An economical method for cell-free protein synthesis using glucose and nucleoside monophosphates. Biotechnol Prog 2005; 21(4):1146-1153. 25. Kim TW, Kim DM, Choi CY. Rapid production of milligram quantities of proteins in a batch cell-free protein synthesis system. J Biotechnol 2006. 26. Kang SH, Oh TJ, Kim RG et al. An eicient cell-free protein synthesis system using periplasmic phosphatase-removed S30 extract. J Microbiol Methods 2000; 43(2):91-96. 27. Shimizu Y, Inoue A, Tomari Y et al. Cell-free translation reconstituted with puriied components. Nat Biotechnol 2001; 19(8):751-755. 28. Shimizu Y, Kanamori T, Ueda T. Protein synthesis by pure translation systems. Methods 2005; 36(3):299-304. 29. Voloshin AM, Swartz JR. Eicient and scalable method for scaling up cell free protein synthesis in batch mode. Biotechnol Bioeng 2005; 91(4):516-521. 30. Sawasaki T, Hasegawa Y, Tsuchimochi M et al. A bilayer cell-free protein synthesis system for high-throughput screening of gene products. FEBS Lett 2002; 514(1):102-105. 31. Endo Y, Sawasaki T. High-throughput, genome-scale protein production method based on the wheat germ cell-free expression system. Biotechnol Adv 2003; 21(8):695-713. 32. Jewett MC, Swartz JR. Rapid expression and puriication of 100 nmol quantities of active protein using cell-free protein synthesis. Biotechnol Prog 2004; 20(1):102-109. 33. Murthy TV, Wu W, Qiu QQ et al. Bacterial cell-free system for high-throughput protein expression and a comparative analysis of Escherichia coli cell-free and whole cell expression systems. Protein Expr Purif 2004; 36(2):217-225. 34. Hartley JL, Temple GF, Brasch MA. DNA cloning using in vitro site-speciic recombination. Genome Res 2000; 10(11):1788-1795. 35. Liang F, Matrubutham U, Parvizi B et al. ORFDB: an information resource linking scientiic content to a high-quality Open Reading Frame (ORF) collection. Nucleic Acids Res 2004; 32(Database issue): D595-599. 36. Feldman G, Bogoev R, Shevirov J et al. Detection of tetracysteine-tagged proteins using a biarsenical luorescein derivative through dry microplate array gel electrophoresis. Electrophoresis 2004; 25(15):2447-2451. 37. Zhu H, Snyder M. Protein chip technology. Curr Opin Chem Biol 2003; 7(1):55-63. 38. Smith MG, Jona G, Ptacek J et al. Global analysis of protein function using protein microarrays. Mech Ageing Dev 2005; 126(1):171-175. 39. Predki PF, Mattoon D, Bangham R et al. Protein microarrays: a new tool for proiling antibody cross-reactivity. Hum Antibodies. 2005;14(1-2):7-15. 40. Ptacek J, Devgan G, Michaud G et al. Global analysis of protein phosphorylation in yeast. Nature 2005; 438(7068):679-684. 41. Michaud GA, Samuels ML, Schweitzer B. Functional protein arrays to facilitate drug discovery and development. IDrugs 2006; 9(4):266-272. 42. Martzen MR, McCraith SM, Spinelli SL et al. A biochemical genomics approach for identifying genes by the activity of their products. Science 1999; 286(5442):1153-1155. 43. Bussow K, Cahill D, Nietfeld W et al. A method for global protein expression and antibody screening on high-density ilters of an arrayed cDNA library. Nucleic Acids Res 1998; 26(21):5007-5008. 44. Uetz P, Giot L, Cagney G et al. A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae. Nature 2000; 403(6770):623-627. 45. Angenendt P, Nyarsik L, Szalarski W et al. Cell-free protein expression and functional assay in nanowell chip format. Anal Chem 2004; 76(7):1844-1849. 46. He M, Taussig MJ. DiscernArray technology: a cell-free method for the generation of protein arrays from PCR DNA. J Immunol Methods 2003; 274(1-2):265-270. 47. Ramachandran N, Hainsworth E, Bhullar B et al. Self-assembling protein microarrays. Science 2004; 305(5680):86-90. 48. Steller S, Angenendt P, Cahill DJ et al. Bacterial protein microarrays for identiication of new potential diagnostic markers for Neisseria meningitidis infections. Proteomics 2005; 5(8):2048-2055. 49. Elbaz Y, Steiner-Mordoch S, Danieli T et al. In vitro synthesis of fully functional EmrE, a multidrug transporter and study of its oligomeric state. Proc Natl Acad Sci USA 2004; 101(6):1519-1524. 50. Klammt C, Lohr F, Schafer B et al. High level cell-free expression and speciic labeling of integral membrane proteins. Eur J Biochem 2004; 271(3):568-580.

Advantages and Applications of the Batch-Formatted E. coli Cell-Free Expression System

41

51. Berrier C, Park KH, Abes S et al. Cell-free synthesis of a functional ion channel in the absence of a membrane and in the presence of detergent. Biochemistry 2004; 43(39):12585-12591. 52. Schwarz D, Klammt C, Koglin A et al. Preparative scale cell-free expression systems: New tools for the large scale preparation of integral membrane proteins for functional and structural studies. Methods 2006. 53. Klammt C, Schwarz D, Lohr F et al. Cell-free expression as an emerging technique for the large scale production of integral membrane protein. Febs J 2006; 273(18):4141-4153. 54. Ishihara G, Goto M, Saeki M et al. Expression of G protein coupled receptors in a cell-free translational system using detergents and thioredoxin-fusion vectors. Protein Expr Purif 2005; 41(1):27-37. 55. Klammt C, Schwarz D, Fendler K et al. Evaluation of detergents for the soluble expression of alpha-helical and beta-barrel-type integral membrane proteins by a preparative scale individual cell-free expression system. Febs J 2005; 272(23):6024-6038. 56. Kuruma Y, Nishiyama K, Shimizu Y et al. Development of a minimal cell-free translation system for the synthesis of presecretory and integral membrane proteins. Biotechnol Prog 2005; 21(4):1243-1251. 57. Noireaux V, Libchaber A. A vesicle bioreactor as a step toward an artiicial cell assembly. Proc Natl Acad Sci USA 2004; 101(51):17669-17674. 58. Noireaux V, Bar-Ziv R, Godefroy J et al. Toward an artiicial cell based on gene expression in vesicles. Phys Biol 2005; 2(3):P1-8. 59. Pornillos O, Chen YJ, Chen AP et al. X-ray structure of the EmrE multidrug transporter in complex with a substrate. Science 2005; 310(5756):1950-1953. 60. Ozawa K, Dixon NE, Otting G. Cell-free synthesis of 15N-labeled proteins for NMR studies. IUBMB Life 2005; 57(9):615-622. 61. Keppetipola S, Kudlicki W, Nguyen BD et al. From gene to HSQC in under ive hours: high-throughput NMR proteomics. J Am Chem Soc 2006; 128(14):4508-4509. 62. He M, Taussig MJ. Ribosome display: cell-free protein display technology. Brief Funct Genomic Proteomic 2002; 1(2):204-212. 63. Rothe A, Hosse RJ, Power BE. In vitro display technologies reveal novel biopharmaceutics. FASEB J 2006; 20(10):1599-1610. 64. Shimizu Y, Kuruma Y, Ying BW et al. Cell-free translation systems for protein engineering. Febs J 2006; 273(18):4133-4140. 65. Wang L, Xie J, Schultz PG. Expanding the genetic code. Annu Rev Biophys Biomol Struct 2006; 35:225-249.

Chapter 4

Energetics in Escherichia coli-Based Batch Cell-Free Systems Kalavathy Sitaraman and Deb K. Chatterjee*

Abstract

T

raditionally, cell-free transcription/translation systems have been used mainly for analytical purposes to characterize gene products. Recently, this technology has become a powerful alternative to cell based methods to produce proteins. It exploits the availability of highly active cellular machinery in S-30 extracts to direct the protein synthesis in the presence of an energy source. Protein synthesis requires 4 to 5 ATP/GTP molecules per peptide bond formation. hus, the eiciency of protein synthesis is governed by the availability of constant supply of energy. Recent advances and improvements made in this ield directly focus on the use of suitable alternative energy sources to prolong the duration of the synthesis thereby increasing the yield.

Introduction

In this post genomic era cell-free translation systems have become the focus of renewed interest for addressing myriads of problems facing protein research. Clear understanding of the role of genes in an organism is only possible if the functions of all its proteins at the molecular level have been deciphered. Acquiring this knowledge will depend, in part, on the cloning, expression, production and puriication of these proteins in a streamlined high throughput manner. Recent advances made in cell-free protein synthesis have widened its usefulness to various applications in molecular biology and biotechnology. Being open systems, cell-free protein expression is amenable to manipulations and modiications to inluence protein folding, stabilization of proteins, disulide bond formation, incorporation of unnatural amino acids and protein synthesis in general. As cellular survival and regulation are no longer required, cell-free protein synthesis has advantages in producing proteins that are otherwise detrimental to cell survival. In addition, protein can be synthesized by using plasmid or linear DNA templates obtained by polymerase chain reaction (PCR) or by other methods containing the gene of interest. Previously, the technology had low acceptance because of poor productivity. Fortunately, various improvements have been made in last few years to produce highly active cellular extract, and more stable energy sources making the system highly competitive with cell-based protein production.

History

As early as in 1950s scientists have shown that protein synthesis continues in cell lysates. Later in the early 1960s Nirenberg and Matthaei1 showed how mRNAs are translated into polypeptides in bacterial systems using polyribonucleotides as templates. he system was called ‘cell-free protein synthesis system’. he conventional in vitro translation system was ‘uncoupled’ as opposed to *Corresponding Author: Deb K. Chatterjee—Protein Expression Laboratory, Research Technology Program, SAIC-Frederick, Inc., NCI-Frederick, Frederick, MD 21702, U.S.A. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

Energetics in Escherichia Coli-Based Batch Cell-Free Systems

43

‘coupled transcription-translation systems’ programmed with DNA as template. In this format, translation continues while mRNA is still being synthesized. Coupled systems are in general simpler, more user friendly and eicient.

Cell Extracts

Any organism could potentially be used for the preparation of cell-free extracts.2 he most well known systems are those derived from Escherichia coli (prokaryotic), rabbit reticulocyte lysates and wheat germ extracts (eukaryotic). hese extracts contain most of the cellular cytoplasmic components essential for protein synthesis. Typically, cells are lysed and the components of transcription/translation are harvested by centrifugation. he choice of the system to be used depends on the synthesis of protein of interest and their downstream application. Both prokaryotic and eukaryotic systems are commercially available for cell-free protein synthesis. E. coli system is more eicient and is capable of generating milligram per ml reaction volume of both prokaryotic and eukaryotic proteins in a batch mode. Eukaryotic systems, even though they are less productive, are more suitable for post translational modiications of eukaryotic proteins.

In Vitro vs in Vivo

In vitro systems have a number of advantages compared to in vivo cell based expression of target proteins: 1. Expression of proteins from circular or linear DNA templates in test tubes without need for transformation, cell growth followed by cell lysis. 2. By virtue of its nature, the system is free from cellular regulation of gene expression and so it can produce even toxic and unstable proteins. 3. Various kinds of unnatural amino acids can be eiciently incorporated into proteins at speciic sites to study protein function. 4. Adaptability to automation and miniaturization suitable for applications such as in vitro evolution and protein micro arrays. Despite all its promising aspects, in vitro protein synthesis has its own limitations, in part due to the short reaction life-span resulting in poor yield compared to in vivo protein synthesis. A constant supply of ATP is one of the key factors for eicient synthesis of proteins in cell-free systems. Over the last few years many groups including ours are channeling their eforts to improve the performance of cell-free systems. In this chapter, we review the bioenergetics involved in bacterial cell-free system.

Components of E. coli-Based Cell-Free System • • • • • • • • • • •

S-30 E. coli extract (without any endogenous mRNA) tRNAs All 20 amino acids ATP, GTP, CTP and UTP NTP regenerating system (energy source) Folinic acid, c-AMP RNA polymerase, preferably T7 RNA polymerase Bufers with magnesium, potassium and ammonium ions Sulhydryl compounds like DTT or β-ME RNAse and protease inhibitors (optional) Plasmid or PCR products as template

Energy Sources for Cell-Free Protein Expression Systems

Recent improvements in cell-free protein expression based on E. coli S-30 extracts include: preparation of the cell lysate in a more concentrated form, removal of endogenous mRNAs and amino acids during the processing of the extract, addition of diferent energy sources and enzymes that generate the ATP necessary to power the process of aminoacetylation and protein synthesis.

44

Cell-Free Protein Expression

he very essence of cell-free protein synthesis depends on the availability of energy in the form of nucleoside triphosphates (NTPs). he maintenance of adequate levels of adenosine triphosphate (ATP) and guanosine triphosphate (GTP) is critical for the eiciency of protein synthesis. ATP is necessary for the activation of amino acid substrates and GTP is required for the function of the ribosomes and accessory factors.

Phosphoenolpyruvate

he most commonly used secondary energy source (Fig. 1) to generate ATP in E. coli cell-free system has been phosphoenolpyruvate (PEP).3 In this reaction one molecule of ATP is generated in the conversion of one molecule of phosphoenolpyruvate to pyruvate, catalyzed by pyruvate kinase. However, the reaction is very short lived because, among other reasons, of the degradation of PEP. hus, only small amount of protein is synthesized. he system is revived and more protein can be synthesized when more PEP and amino acids are supplemented again indicating that the catalytic activity is intact.4,5 Other high energy compounds like acetyl phosphate6 and creatine phosphate7 have been used for regeneration of ATP in the presence of enzymes acetate kinase and creatine kinase, respectively (see Fig. 1). However, similar to PEP, both of these compounds are found to be unstable in E. coli extracts due to the enzymatic degradation by phosphatase activity.5

Pyruvate

Swartz and coworkers have reported several improvements with promising outcome using pyruvate as the energy source (Fig. 2A). his system is dependent on pyruvate oxidase which is not present in E. coli extracts and thus, needs to be added exogenously. In this system, stable ATP concentration was maintained for over an hour. When a mixture of pyruvate and amino acids are added every hour, ATP concentration was maintained for a prolonged time and protein synthesis continued for seven hours yielding 700 ug/ml of chloramphenicol acetyl transferase (CAT).4 To avoid the use of costly exogenous pyruvate oxidase the scheme as detailed in Figure 2 was used to regenerate ATP from ADP using pyruvate and endogenous enzymes in the presence of NAD and CoA.9 In this scheme, pyruvate is converted to acetyl CoA, then acetyl phosphate, and inally to ATP and acetate in sequential steps. he yield of protein synthesis by this method is about 70% compared to that of the reaction using PEP. Pyruvate, though more stable than PEP in E. coli extract, can still be non-productively utilized by phosphoenol pyruvate synthetase (PPS) present in the cell extract. his enzyme catalyzes the

Figure 1. Conventional energy source for cell-free synthesis using Phosphoenol pyruvate.

Energetics in Escherichia Coli-Based Batch Cell-Free Systems

45

Figure 2. Pyruvate as energy source using (A) an exogenous enzyme, pyruvate oxidase and (B) an endogenous enzyme pyruvate dehydrogenase.

conversion of pyruvate to PEP at the expense of ATP which would otherwise be used for protein synthesis. hus, in a modiied system, called the PANOx system10 oxalate was used to inhibit PPS minimizing the unproductive use of ATP.10,11 his modiied system indeed improved protein synthesis considerably. In another system developed by Jewett and Swartz the reaction environment was altered in such a way as to mimic the E. coli cytoplasmic environment,12 called the ‘Cytomin System’. he reactions devoid of inluences from bufers and polyethylene glycol are supplemented with spermidine and putrescine. Spermidine and putrescine have been shown to improve the idelity of translation.13 Under these conditions, with pyruvate as the energy source, CAT was synthesized at 700 μg/ml over a period of 6 hrs, a yield similar to one achieved with PANOx system.

46

Cell-Free Protein Expression

Glucose-6-Phosphate

It was thought that glucose-6-phosphate would produce more ATP during its oxidation to pyruvate (three molecules of ATP per molecule of glucose-6-phosphate compared to one per molecule of PEP). When glucose-6-phosphate was used in the presence of NAD and CoA, the protein synthesis continued for 2 hours ater an initial slow start and resulted in protein synthesis, however, only the 30% more of CAT9 protein was synthesized compared to PEP as the energy source. his result indicated that glucose-6-phosphate can be used as secondary energy source. However, the amount of protein synthesized from did not correlate with the projected ATP synthesis from glucose-6-phosphate suggesting that this substrate was also used unproductively in E. coli extract. Because glucose-6-phosphate, PEP and pyruvate appear to be unstable in E. coli extract, we postulated that endogenous phosphatases are capable of attacking these substrates because the initial concentrations are perhaps higher than the theoretical Km for the phosphatases. We reasoned that if PEP could be made in situ from another glycolytic pathway intermediate, at any given time the concentration of PEP would be too low for phosphatases to use it as a substrate. he net result would be sustained synthesis of ATP. We demonstrated this hypothesis by using 3-phosphoglycerate (3-PGA) and showed that it is indeed a better secondary energy source in a cell-free protein synthesis system using E . coli. extracts.14

3-Phosphoglycerate

In cells, 3-PGA is converted to PEP by two sequential enzymatic reactions catalyzed by phosphoglycerate mutase and enolase, respectively (Fig. 3). We observed in a real-time study of expression of GFP that protein synthesis continued for up to three hours when 3-PGA was used as the secondary energy source. Under identical reaction condition, the rate of protein synthesis slowed down considerably ater 45 min when 3-PGA was replaced by PEP. We believe that 3-PGA is more stable in E. coli extracts than glucose-6-phosphate or its intermediates prior to 3-PGA. We were able to achieve close to 1 mg/ml of GFP which is almost 2 fold compared to PEP as energy source (Fig. 4). Recent experiments conirmed that there is sustained level of ATP synthesis over long period of time when 3-PGA is used as a secondary energy source.15

Figure 3. Biochemical pathway using 3-phosphoglycerate to generate ATP.

Energetics in Escherichia Coli-Based Batch Cell-Free Systems

47

Figure 4. Comparison of GFP synthesis with PEP and 3-PGA in real-time. Expression levels were measured at five minutes interval. Reactions were performed in triplicate for each substrate. Adapted from: Sitaraman K et al. J Biotechnol 110:257-263; ©2004, with permission from Elsevier.14

In brief, any of the glycolytic pathway intermediates can be used as secondary energy source to fuel the cell free protein synthesis since all catalytic enzymes needed for ATP regeneration are active in the E. coli S-30 cell extracts.

Glucose

Recently it has been shown by Swartz’s group that even glucose can serve as an eicient energy source in E. coli cell extracts.16 In this system, NTPs were also replaced by monophosphates as the nucleotide source. It seems that within a short span of twenty minutes monophosphates are converted to NTPs. Under stable pH conditions and in the presence of 10 mM phosphate, CAT yield of about 500 ug/ml was achieved in 3 hr batch mode.

Other Phosphate Containing Secondary Energy Sources

As stated above, in E. coli cell-free systems, both acetyl phosphate (AP) and creatine phosphate (CP) can be used as secondary energy sources (Fig. 5). he use of CP and AP to generate ATP requires the enzymatic activities of creatine kinase (provided exogenously) and acetate kinase respectively. hese conventional energy sources are quite expensive and as indicated earlier these two compounds are relatively unstable due to enzymatic degradation. his greatly reduces the duration and eiciency of protein synthesis. It is worthwhile to note that eukaryotic cell-free systems like wheat germ and rabbit reticulocyte lysates also use energy generating system consisting of creatine phosphate and creatine kinase.

Secondary Energy Source from Citric Acid Cycle

he citric acid cycle ( TCA) produces at least one molecule of GTP (substrate level) during the conversion of succinyl-CoA to succinate in addition to ATP production through oxidative

48

Cell-Free Protein Expression

Figure 5. Secondary energy sources: (A) acetyl phosphate and (B) creatine phosphate.

phosphorylation. We tried many of the intermediates involved in this pathway namely, oxaloacetate, citrate, α-ketoglutarate, succinate and malate in the presence of necessary cofactors in the cell-free protein system. Most of the enzymes needed to catalyze the reactions were found to be active in our E. coli cell extracts and were able to support protein synthesis in cell-free system to a considerable extent using these substrates. he results obtained with one of the substrates, citrate, is shown in Figure 6. Table 1 shows the relative level of GFP synthesis using the TCA cycle intermediates as secondary energy sources compared to 3-PGA. Although our E. coli extracts may contain membrane particles, oxidative phosphorylation may not be operative in this case (unpublished) and the sole energy may be derived from substrate level phosphorylation. It is to be noted that the above substrates from TCA cycle further enhanced the protein synthesis when used in conjunction with 3-PGA (Fig. 6). Use of TCA cycle intermediates as energy source would further reduce the cost of cell-free protein synthesis as the costs of these substrates are only a fraction of other energy sources.

Table 1. Relative synthesis of GFP using various substrates Substrate

Amount of Synthesis

3-PGA Oxaloacetate Citrate α-Ketoglutarate Succinate Malate

+++++++ ++++ +++++ +++++ +++ +++++

Energetics in Escherichia Coli-Based Batch Cell-Free Systems

49

Figure 6. Citrate as the energy source. Three different proteins were expressed with citrate in the presence of CoA and NAD as cofactors and compared against 3-PGA and also in combination with 3-PGA. The reactions were performed in a 50μl volume in a thermomixer. 1 μl was run on a 4-20% gel and coomassie stained.

Large-Scale Protein Production in Batch Reaction

Large scale production of proteins is necessary for structural analysis and other biological applications. he simple, static batch version done in a microfuge tube or large conical tubes is suicient enough to produce a considerable amount of desired protein. he challenges of generating proteins in a preparative scale was addressed by the development of continuous (CECF) and semi-continuous exchange (CFCF) low cell-free systems by Spirin17 and Swartz18 respectively. Despite a slight improvement in yield these methods are rather costly, time consuming and not suitable for high throughput applications. A simple and user friendly dialysis system called Dispolyzer has also been used instead.19 his device is quite simple consisting of a dialyzer in a sterile 15 ml conical tube containing 3.5 ml to 7 ml of dialysis solution containing all necessary ingredients. he dialyzer containing about 100-700 ul of standard in vitro reaction was immersed in a 15 ml conical tube containing 1x dialysis solution without any template or enzyme. Using PEP as the energy source 11-20-fold increase of luciferase synthesis ater 4 and 20 hours respectively has been reported.

50

Cell-Free Protein Expression

Ater evaluating several methods described in the literature such as dialysis, bilayer20 and others., we have started using very simple devices readily available in all molecular biology, cell culture laboratories bypassing all costly instrumentations. he following materials are needed: 1. A 14 ml polypropylene round bottom FALCON 2059 tube for up to one ml of cell-free reaction volume (unpublished). 2. A 24 well four ml wide round bottom culture plates for semi high-throughput application (unpublished). Each well contains one ml of cell-free reaction. he reaction conditions are the same as in 50 ul reaction but the components are proportionately increased to the desired volume. he standard reaction mixture consisted of the following components: 57 mM HEPES-KOH, pH 7.6, 230 mM K-glutamate, 80 mM ammonium acetate, 2 mM each of ATP and GTP, 0.85 mM each of CTP and UTP, 40 mM 3-phosphoglycerate, 1.7 mM DTT, 12 mM Mg-acetate, 0.17 mg/ml E. coli t-RNA mixture, 34 μg/ml folinic acid, 30-50 μg/ml T7 RNA polymerase, 1 mM of each amino acid, 10-20 mg/ml extract, 0.65 mM cAMP and various amounts of DNA template. he tubes or plates are incubated at 30º for 6 hours in a shaking incubator. We made use of this method to successfully express milligram quantities of GFP, CAT and K-Ras (Fig. 7). In addition, we also produced other proteins such as RPA, YopB and YopD by this method (unpublished). Membrane proteins encode about 30% of the genome. It is extremely diicult to produce and purify the recombinant membrane proteins. In vivo expression of these membrane proteins are characterized by high cell toxicity, aggregation and oten results in very poor yield. here are some reports of expressions of G-protein coupled receptors by cell free systems.21 We were also able to

Figure 7. Cell-free expression of GFP, CAT and K-ras in a batch mode using 3-PGA as energy source. Reactions were set up in a microfuge tube using a thermomixer (TM), in a falcon tube (INC) and by dialysis (DIA) using a dispolyzer. The reactions were incubated at 30º C for 6 hours. 1 μl from each reaction was run on a 4-20% Tris-glycine gel and stained with Coomassie blue. CON, control reaction with no DNA.

Energetics in Escherichia Coli-Based Batch Cell-Free Systems

51

Figure 8. Expression of two orphan GPCRs in cell-free synthesis reactions using 3-PGA as the energy source. Reactions were done in duplicate. One μl samples were run on a 4-12% bis-tris gel and stained with coomassie blue. The arrow indicates the position of the proteins.

express milligram quantities of several orphan GPCRs in large scale cell-free system using 3-PGA as the energy source in the presence of lipid mixture. he synthesized proteins were easily detected by simple Coomassie stain (Fig. 8). Preliminary solubilization studies indicated that about 25-30% of the expressed GPCR proteins can be solubilized, properly folded, in the presence of appropriate detergents following high speed centrifugation (manuscript in preparation).

Conclusion

Cell-free protein synthesis has come a long way over the last few decades. It ofers a simple lexible alternative to cell based methods. It is now possible to express milligram quantities of proteins in a matter of few hours. he achievement was mainly due to the strides made in understanding the bioenergetics and some modiications of the reaction. he batch mode is now applicable to large scale high throughput formats. he possibility of using any of the energy sources derived from glycolytic or TCA pathway in a cost efective manner should enable researchers to express proteins for structure determination, functional analysis, antibody production, high throughput screening and molecular evolution.

52

Cell-Free Protein Expression

Acknowledgements

his project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under contract N01-CO-12400. he content of this publication does not necessarily relect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.

References

1. Nirenberg MW, Mattaei JH. he dependence of cell-free protein synthesis in E. coli upon naturally occurring or synthetic polyribonucleotides. Proc Natl Acad Sci USA 1961; 47:1588-1602. 2. Jackson AM, Boutell J, Cooley N et al. Cell-free protein synthesis for proteomics. Brief Funct Genomic Proteomic 2004; 2(4):308-319. 3. Zubay G. In vitro synthesis of protein in microbial systems. Annu Rev Genet 1973; 7:267-287. 4. Kim DM, Swartz JR. Prolonging cell-free protein synthesis with a novel ATP regeneration system. Biotechnol Bioeng 1999; 66:180-188. 5. Kim DM, Swartz JR. Prolonging cell-free protein synthesis by selective reagent additions. Biotechnol Prog 2000a; 16:385-390. 6. Ryabova LA, Vinokurov LM, Shekhovtsova EA et al. Acetyl phosphate as an energy source for bacterial cell-free translation systems. Anal Biochem 1995; 226:184-186. 7. Kigawa T, Yabuki T, Yoshida Y et al. Cell-free production and stable isotope labeling of milligram quantities of proteins. FEBS Lett 1999; 442:15-19. 8. Muller YA, Schulz GE. Strucure of the thiamine and lavin dependent enzyme pyruvate oxidase. Science 1993; 259(5097):965-7. 9. Kim DM, Swartz JR. Regeneration of adenosine triphosphate from glycolytic intermediates for cell-free protein synthesis. Biotechnol Bioeng 2001; 74:309-316. 10. Kim DM, Swartz JR. Oxalate improves protein synthesis by enhancing ATP supply in a cell-free system derived from Escherichia coli. Biotechnol Bioeng 2000b; 22:1537-1542. 11. Narindraorasak S, Bridger WA. Probes of the structure of phosphoenolpyruvate synthetase: efects of a transition state analogue on enzyme conformation. Can J Biochem 1978; 56:816-819. 12. Jewett MC, Swartz JR. Substrate replenishment extends protein synthesis with an in vitro translation system designed to mimic the cytoplasm. Biotechnol Bioeng. 2004; 87:465-472. 13. Jelene PC, Kurland CG. Nucleoside triphosphate regeneration decreases the frequency of translation errors. PNAS 1979; 76(7):3174-3178. 14. Sitaraman K, Esposito D, Klarmann G et al. A novel cell-free protein synthesis system. J Biotechnol 2004; 110:257-263. 15. Kuem JW, Kim TW, Park CG et al. Oxalate enhances protein synthesis in cell-free synthesis system utilizing 3-phosphoglycerate as energy source. J Biosci Bioeng 2006; 101(2):162-165. 16. Calhoun KA, Swartz JR. An economical method for cell-free protein synthesis using glucose and nucleosidemono-phosphates. Biotechnol Prog 2005; 21(4):1146-1153. 17. Spirin AS, Barnov VI, Ryabova LA et al. A continuous cell-free translation system capable of producing polypeptides in high yield. Science 1998; 242:1162-1164. 18. Kim DM, Choi CY. A semicontinous prokaryotic coupled transcription/translation system using a dialysis membrane. Biotechnol Prog 1996; 12:645-649. 19. Julie D, Dave T, Gregory SB. Large scale dialysis reaction using E. coli S30 extract systems. Promega Notes 1996; 56:14. 20. Tatsuya S, Yoshinori H, Masteru T et al. A bilayer cell-free protein synthesis system for high-throughput screening of gene products. FEBS Letter 2002; 514:102-105. 21. Goshi I, Mie G, Mihoro S et al. Expression of G protein coupled receptors in a cell-free translational system using detergents and thioredoxin-fusion vectors. Protein Exp Pur 2005; 41:27-37.

Chapter 5

Disulfide Bond Formation in Bacteria-Based Cell-Free Protein Expression Aaron R. Goerke and James R. Swartz*

Abstract

C

ell-free protein synthesis systems have many advantages over conventional in vivo (cellular) expression. For example, they ofer the potential for higher productivity, parallel production and simpliied puriication. Moreover, the openness of the cell-free system allows control of the reaction environment to promote folding of disulide bonded proteins in a rapid and economically feasible format. hese advantages make in vitro protein expression systems particularly well suited for the production of patient-speciic therapeutic vaccines for diseases such as cancer, for vaccines to protect against threats from natural and man-made biological agents and for pharmaceutical proteins that are diicult to produce in living cells. Yet the promotion of the cell-free expression system from a useful laboratory tool to a commercially viable technology has been slow. his is at least partially due to the inability to eiciently fold complex proteins, especially those containing disulide bonds. In this chapter, we describe controlled disulide/sulhydryl redox chemistry and improved disulide isomerization in a cell-free system. We then also describe genetic engineering that in combination results in cell-free systems producing greater complex protein yields than were previously thought possible. To achieve elevated protein yields, cell extracts were generated that maintain amino acid stability and decrease the amount of nonspeciic chemical treatment required to eliminate cytoplasmic oxidoreductase activity. In combination, these modiications substantially lower substrate costs since glucose can be used to fuel the cell-free system. We then report the successful scale-up of a B-cell lymphoma fusion vaccine to the 30 mL scale. In considering all of these advances, we believe that our in vitro protein synthesis technology has achieved a signiicant milestone and is ready for widespread research and commercial implementation.

Introduction Mammalian Protein herapeutics and Vaccines

Since the irst approval of rDNA insulin in 1982, the number of recombinant proteins used for therapeutic applications has increased dramatically. Most of these applications involve proteins and antibodies with multiple disulide bonds. For example, recombinant granulocyte-colony stimulating factor (G-CSF) and granulocyte macrophage-colony stimulating factor (GM-CSF) have emerged as hematopoietic growth factors with proven eicacy in treating patients with chemotherapy-induced neutropenia.1 Also, the interferons (IFNs) are a class of proteins that serve in the body’s natural defensive responses to such foreign components as microbes, tumors and *Corresponding Author: James R. Swartz—Department of Chemical Engineering, Stanford University, Stanford, California 94305, U.S.A. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

54

Cell-Free Protein Expression

antigens.2 Interferon alpha has been approved for the treatment of hairy-cell leukemia, condyloma acuminatum, Kaposi’s sarcoma in the acquired immunodeiciency syndrome and Type C viral hepatitis. On the biomedical front, central venous access devices are part of current medical therapy but oten become blocked by a blood clot. Tissue plasminogen activator (tPA) and urokinase (UK) have been efective enzymatic treatments to restore blood low in the human body.3 Yet, the clinical importance and high cost of these therapeutic agents highlights the need to establish more rapid, reliable and cost-efective production protocols. his becomes even more pressing with the global emergence of biogeneric protein pharmaceuticals. A streamlined process would both accelerate protein pharmaceutical development and provide for simpliied, highly productive manufacturing procedures. Such advances are now possible with cell-free protein synthesis (CFPS). If the vaccines to deter a natural or man-made pandemic could also be produced via cell-free technology, the response time and production cost would be signiicantly reduced. Moreover, if the vaccine were produced using linear DNA templates, the process could go from patient specimen to vaccine product very rapidly and without requiring living cells. An application for this technology could be NonHodgkin’s lymphoma (NHL). NHL is a heterogeneous group of lymphoproliferative malignancies with difering patterns of behavior and responses to treatment.4 Over 300,000 Americans sufer from NHL and 58,000 new cases will be diagnosed in 2006.5 Using standard treatment, the 5 year survival rate is only 50 to 60% and although 30% of patients with NHL can be completely cured, relapse is common within 2 years following aggressive radiation, chemotherapy and immunotherapy treatments. he limited success of radiation and chemotherapy treatments has suggested that every malignancy is unique and requires individualized biological treatment for increased eicacy. Cell-free technology is the appropriate production methodology for the rapid, parallel and economical manufacture of time sensitive and patient-speciic treatments that are diicult to produce in traditional bacterial fermentations. In this chapter we will describe improved cell-free technology capable of serving such applications.

Comparing Cellular and Cell-Free Production of Disulide-Bonded Proteins

he majority of therapeutic proteins have been produced in either mammalian cell-culture systems, with Chinese hamster ovary (CHO) cells representing the most common system, or in Escherichia coli.6,7 A variety of alternative expression systems have also been developed and evaluated.8-10 It is not obvious which of these systems will ultimately be the most useful. To minimize the overall production cost and to gain the advantages of using a well characterized organism,11 individuals have repeatedly attempted to produce complex proteins in E. coli. Yet, success has been limited due to an inability to precisely reproduce the folding environment in the endoplasmic reticulum where the natural folding occurs. Insoluble aggregates of misfolded protein oten result. Although these can commonly be isolated from the cytoplasm and refolded in vitro,12 the additional refolding step is expensive and time consuming, especially on the industrial scale. Clearly, it would be advantageous to develop a process that does not require refolding, yet still takes advantage of the low cost and simplicity of a bacterial expression system. Another production approach is to export the recombinant protein to the periplasm of E. coli for maturation.13 However, folding eiciency, the rate of translocation of the nascent polypeptide into the periplasm and cell autolysis oten limit recombinant protein yields.14 hus, several groups have been seeking to develop novel signal sequences and production protocols that would allow more eicient secretory production of recombinant proteins.15 Signiicant yields of active proteins that require disulide bonds can be achieved in the cytoplasm of E. coli combining various measures including co-expressing molecular chaperones, reducing protein synthesis rate,16 using lower temperature culture conditions,17 using highly soluble polypeptides as fusion partners18 and deleting the genes encoding the cytoplasmic reductases. For example, the elimination of thioredoxin reductase (trxB) allows accumulation of active alkaline phosphate in the cytoplasm.19 Although deletion of both trxB and glutathione reductase (gor) genes leads to a strain that grows extremely slowly, normal growth is restored by a mutation in the peroxiredoxin gene, ahpC.20 his is the strain (named FÅ113) currently marketed for the formation of disulide bonds in the E. coli cytoplasm.19,21,22 Complex therapeutic proteins such as functional single-chain

Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression

55

variable fragments (scFv) and Fab antibody fragments have been produced in the cytoplasm of FÅ113 when folding is aided by co-expression of DsbC and/or the chaperone, Skp.22-24 Similarly, a number of diferent cellular production methods have been used to produce the whole immunoglobulin (Ig) portion of the NHL vaccine.25,26 Two on-going clinical trials use a full-length Ig fused to keyhole limpet hemocyanin (KLH), a highly immunogenic protein, which is co-administered with GM-CSF. he idiotype speciic vaccine proteins are either produced in mammalian cells (Genitope Incorporated) or by insect cells (Favrille Incorporated), but the associated time and labor costs impose signiicant barriers for the production of patient-speciic vaccines. Using an E. coli-based cell-free protein expression system can overcome the limitations of current cellular production systems. Advantages of a cell-free system include but are not limited to: Rapid production of a single protein product, high-throughput production and control of the reaction environment to direct metabolism and improve protein folding. While plasmid-encoded expression of the tumor-speciic Ig variable region genes is currently required for our patient-speciic application, it is possible to express patient-speciic vaccines from linear DNA templates ampliied by PCR.27 his provides additional time savings when compared to common recombinant DNA production methods. Because the cell-free reaction mixture is not enclosed by a cell-membrane, the puriication is also simpliied relative to other E. coli processes. he product can be puriied, characterized and ready to use as a therapeutic or vaccine in a matter of days. Use of a simple prokaryotic organism, however, creates unique challenges when producing complex proteins that require disulide bond formation or other post-translational modiications such as glycosylation. To address the latter issue, signiicant work is being done to incorporate unnatural amino acids into proteins, which could then serve as sites for post-translational glycosylation in a prokaryotic system.28-31 Another issue to be considered for cell-free production is the presence of the formylated methionine at the N-terminus. Since most recombinant proteins for pharmaceutical purposes are of human origin and are mostly secreted proteins lacking the N-terminal methionine, the retention of this methionine ater cell-free expression can be a problem when a protein is used for a human therapeutic. Overexpression of methionine aminopeptidase can solve this problem in some cases.32 his is an on-going area of work for cell-free protein expression.

Expression of Disulide Bonded Proteins in a Commercially Viable Cell-Free System

CFPS was irst used in biochemical studies such as the deciphering of the genetic code in the 1960s.33,34 hen and for the next few decades cell-free synthesis was not a reliable or eicient process for the production of recombinant proteins due to high reagent costs, low protein yields and diicult scale-up. Improved batch cell-free reactions have since been developed that use phosphoenolpyruvate (PEP),35 creatine phosphate (CP),36 or acetyl phosphate (AcP)37 to directly phosphorylate adenosine diphosphate (ADP) to adenosine triphosphate (ATP). However, typical protein yields suggest ineicient use of these substrates. Most likely, the major cause of the low yields is instability of the energy sources as they are degraded by nonspeciic phosphatases present in the cell extract.38 Not only does this reduce the productivity of the energy source, but it also generates high concentrations of inorganic phosphate which are inhibitory to protein synthesis at concentrations above 30 mM.39 Reduction of phosphatase activity has been attempted with some success. Kim and Choi suggest that addition of phosphate to the growth media for cell extract preparation reduces phosphate levels in an E. coli system.40 A wheat-germ cell-free system has been improved through immunoprecipitation of phosphatases.41 Another method for increasing the supply of the energy source (but not the eiciency) is to use diferent reactor conigurations, such as continuous-low or semi-continuous reactors.35,36 hese designs allow continued supply of the secondary energy source, while providing a means for removal of inhibitory byproducts such as inorganic phosphate. Alternatively, the PURE cell-free system, which uses all puriied components to catalyze protein synthesis, has no phosphatase activity.42 Overall, the use of compounds with high-energy phosphate bonds is a simple, yet expensive way to fuel cell-free protein synthesis.

56

Cell-Free Protein Expression

A large amount of work has been devoted to reducing the cost and increasing the productivity of cell-free reactions. Speciically, eforts have been focused on the discovery of a low cost energy supply that does not create inhibitory levels of inorganic phosphate in the reaction. A signiicant breakthrough was the demonstration that metabolic pathways, not just simple one-step phosphorylation reactions, could be harnessed to supply energy for protein biosynthesis.39 It was then discovered that oxidative phosphorylation could be activated by using a cell-free reaction composition that more closely resembles the E. coli cytoplasm.43 Here the nonphosphorylated energy source, pyruvate, resulted in protein yields similar to the traditional systems, except with a much less expensive energy source. Additionally, energy-generating pathways such as glycolysis have been shown to be active in cell-free reactions.43,44 Additional development showed that glucose, the preferred low cost substrate for bacterial fermentations, could be used if the pH of the system was stabilized.45 Other cost reductions include replacing expensive nucleoside triphosphates (NTPs) with less expensive nucleoside monophosphates (NMPs) and eliminating cofactors from the cell-free reaction.46 With these advances, cell-free protein synthesis is becoming competitive with traditional fermentation in cost and productivity. Systems biotechnology has also been key to the improvement of CFPS consistency and eiciency. his has led to the development of many diferent cell strains used to generate cell extract.47-49 For example, cell-free production limitations have been addressed by identifying enzymatic activities responsible for amino acid instabilities. he genes encoding arginine decarboxylase (speA), tryptophanase (tnaA), serine deaminases (sdaA and sdaB) and glutamate-cysteine ligase (gshA) were removed from the E. coli strain used to make extract thereby allowing complete amino acid stabilization and improving protein yields.49 Nonetheless, several laboratories around the world have demonstrated efective folding for hundreds of proteins using CFPS, many of these examples are relatively simple proteins with few or no disulide bonds. Most cell-free systems fall short in delivering the simplicity, robustness and cost-efectiveness needed for CFPS to meet expectations for producing complex protein products quickly and inexpensively. hus, a more robust cell-free platform is necessary. In addition, we need the ability to easily and eiciently increase the volume of protein synthesis reactions. he reaction scales are limited to, in most cases, tens of microliters per batch. In some cases milliliter scale reactions are realized but the reactor complexity makes them unsuitable for wide adoption and for industrial applications.35,50-53 In this chapter we summarize advances that have enabled us to better understand and control the central metabolic processes that occur in cell-free reactions producing disulide bonded proteins. In addition, we describe methods to produce active disulide bonded proteins in which an oxidized glutathione bufer is used to adjust the redox potential and DsbC is added to catalyze rearrangement of incorrectly formed disulide bonds. Protocols incorporating iodoacetamide (IAM) pretreatment of the cell-free extract54 aided in achieving these results; however, there was no control over which cysteine residues were derivatized by IAM. he inactivation of thioredoxin reductase (TrxB) and glutathione reductase (Gor) is necessary, but colateral modiication of other proteins such as DsbA and DsbC may be detrimental to protein folding. A new extract was engineered that does not contain glutathione reductase (Gor) and requires 20-fold less IAM to stabilize oxidized glutathione. his provides an extract that produces proteins requiring disulide bonds without fully inactivating key metabolic enzymes such as DsbC and glyceraldehyde 3-phosphate dehydrogenase (G-3PDH). hese platforms are now more practical at the manufacturing scale because they can use glucose or glutamate as their primary energy source.46 Following this success, we used a new scale-up technique to increase protein production from a 15 μL to a 30 mL batch reaction volume.55 hese advances provide an economically feasible cell-free system for the production of secreted mammalian proteins suitable for human therapeutics and vaccines.

Batch Mode Production in a Bacterial Cell-Free System

he pioneering work of the Spirin laboratory,36 expanded upon by Kim and Choi and Kigawa et al35,50 demonstrated the potential of the cell-free approach. However, the continuous system has

Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression

57

proven to be rather cumbersome and expensive. While the semicontinuous system ofers high yields via a simpliied format, the excess volume of reagents adds signiicant expense, particularly with respect to nucleotides and energy. More recently the focus has been on batch and fed-batch modes. In batch mode, the use of alternative energy sources has been investigated to reduce costs while maintaining high cell-free yields. Most of these alternative compounds require multi-step enzymatic reactions to generate ATP.44,56 Such reaction systems have advantages over traditional energy sources in cost, eiciency and stability. A reliable and eicient system that activates multi-step enzymatic phosphorylation pathways is referred to as PANOx-SP (PEP, Amino Acids, NAD, Oxalic Acid, Spermidine and Putrescine). his is a second generation PANOx system51 and uses both PEP and its reaction product, pyruvate, as energy sources. Normally the reaction mixture contains the following components from Sigma-Aldrich (St. Louis, MO), except where noted: 16 to 20 mM magnesium glutamate, 10 mM ammonium glutamate, 175 mM potassium glutamate, 1.2 mM ATP, 0.85 mM each of GTP, UTP and CTP, 34 μg/mL folinic acid, 170.6 μg/mL E. coli tRNAs (Roche Molecular Biochemicals, Indianapolis, IN), 2 mM each of twenty amino acids, 33 mM phosphenol pyruvate (Roche Molecular Biochemicals, Indianapolis, IN), 1.5 mM spermidine, 1.0 mM putrescine, 13.3 μg/mL template plasmid, 100 μg/mL T7 RNA polymerase (prepared as described by Jewett et al)47, 10 μM L-[U-14C]-Leucine (Amersham Pharmacia, Uppsala, Sweden), 0.33 mM nicotinamide adenine dinucleotide (NAD), 0.26 mM Coenzyme A (CoA), 2.7 mM sodium oxalate and 0.24 volumes of E. coli S30 extract (prepared as described by Zawada and Swartz57 and Liu et al58). he PANOx-SP reaction composition is listed in Table 1. he genes for all proteins expressed in our cell-free systems were cloned into the expression plasmid pK7 under the control of the standard T7 promoter. he PANOx-SP system increases the eiciency of ATP generation through the addition of the cofactors, NAD and CoA, to activate an additional ATP-generating reaction.43 In this way, up to 1.5 moles of ATP can be obtained from each mole of PEP instead of just 1 mole-ATP/mole-PEP and the duration of energy supply is signiicantly extended. One of the advances from the original PANOx system includes replacing polyethylene glycol (PEG), a high molecular weight compound used for nucleic acid stability, with the natural polyamines putrescine and spermidine. his provides a more natural chemical composition. he use of oxalic acid in this system is beneicial because it inhibits the activity of PEP synthase,59 an enzyme that wastes ATP by converting pyruvate to PEP at a cost of two high energy phosphate bonds per PEP molecule. he ability to activate multi-step reactions through addition of cofactors suggested the possibility of activating entire energy-generating pathways in cell-free reactions. For a long time, the metabolic activities of the crude cell extract were not well characterized. However, with advanced understanding of this class of cell-free reactions, it was possible to activate complex metabolic pathways for generating nucleoside triphosphates. he basic hypothesis was that replicating intracellular conditions would activate complex intracellular functions. his concept led to the development of the Cytomim cell-free system (see Table 1).43 he most notable changes include replacing PEG with putrescine and spermidine and replacing PEP with a nonphosphorylated energy source, pyruvate, thus dramatically lowering costs and avoiding phosphate accumulation during the course of the reaction. Together with optimization of component concentrations (6 to 12 mM magnesium glutamate, 10 mM ammonium glutamate, 130 mM potassium glutamate), these changes resulted in high model protein yields. Interestingly, the protein yields were well above what was expected through ATP generation by conversion of pyruvate to acetate alone. Consequently, another energy-generating pathway, most likely oxidative phosphorylation, is responsible for the additional energy. In fact, the addition of oxidative phosphorylation inhibitors signiicantly reduces protein yields from pyruvate.60 It has been discovered that proper folding of a recombinant protein with disulide bonds requires a relatively oxidizing redox potential.54,61,62 However the S30 cell extract prepared from E. coli is normally in a reducing state relative to disulide bonds. In order to maximize proper protein folding, modiication of the sulhydryl redox potential of the cell-free reaction mixture has been performed by combining two measures. First, 0.5 to 1 mM iodoacetamide (IAM) is used to pretreat

58

Cell-Free Protein Expression

the S30 cell extract (room temperature for 30 minutes) thereby inactivating the free sulhydryl groups required in the active sites of the cytoplasmic oxidoreductases. Second, a glutathione bufer composed of oxidized (GSSG, 4 mM) and reduced glutathione (GSH, 1 mM) is used to create an oxidizing environment. In this way a stabilized oxidizing environment is created (see Table 1). Although an oxidizing environment can support eicient disulide bond formation, it cannot guarantee correct disulide linkages. Additional enzymes must be incorporated into the cell-free reaction to ensure correct disulide bond formation. Dependence on chaperone function has been highlighted by Kudlicki et al.63 A wheat germ-derived cell-free protein synthesis system produced 13 μg/mL active scFv by removing the reductant DTT from the reaction and supplementing with a eukaryotic protein disulide isomerase (PDI).64 An E. coli cell-free system was also used to produce 8 μg/mL of an active scFv with supplemental PDI.65 his success may be due to the relatively small number (2 total) and the stability of the disulide bonds,54 which would most likely form during the early period of protein synthesis before the system becomes reduced.66 Various investigators have also observed that the addition of molecular chaperones such as DnaJ/DnaK and GroEL/GroES

Table 1. Final concentrations of components in the PANOx-SP and Cytomim cell-free systems. Concentrations include components added to the cell extract during its preparation Component

PANOx-SP System

Cytomim System

Magnesium acetate (mM)† 3.3 Magnesium glutamate (mM) 16 to 20 6 to 12 Ammonium glutamate (mM) 10 Potassium acetate (mM)† 14.4 Potassium glutamate (mM) 175 130 Nucleoside triphosphates (NTPs, mM) 1.2 ATP; 0.85 GTP, UTP, & CTP Tris (pH 8.2, mM)† 2.4 Folinic acid (μg/mL) 34 E. coli tRNA mixture (μg/mL) 170 20 amino acid mixture (mM of each AA) 2* Phosphoenolpyruvate (mM) 33 --Sodium pyruvate (mM) --33 Nicotinamide adenine diucleotide (NAD, mM) 0.33 Coenzyme-A (CoA, mM) 0.26 Sodium oxalate (mM) 2.7 4 Putrescine (mM) 1 Spermidine (mM) 1.5 L-[U-14C]-leucine (μM) 5 to 10 Plasmid (μg/mL) 13.3 T7 polymerase (mg/mL) 0.10 S30 cell extract (% volume) 24 Iodoacetamide (IAM, mM) 0.05 to 1‡ Oxidized glutathione (GSSG, mM) 4 Reduced glutathione (GSH, mM) 1 Disulfide isomerase (DsbC, μg/mL) 75 to 100 Sterile water Remainder to final volume *Due to cysteine instability in selected extracts (i.e. A19, NMR1, NMR2, KC1) an additional 2 mM of cysteine was included. †Includes carryover from cell extract. ‡Cell extract is incubated with IAM for 30 minutes at room temperature prior to being added to the cell-free reaction.

Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression

59

could increase soluble protein yields in cell-free systems.65-70 However, the extent of the favorable efect is dependent on the speciic protein being expressed.71 We have concentrated speciically on the E. coli periplasmic chaperone, Skp and on the Dsb system of E. coli proteins that have evolved to catalyze the formation and isomerization of disulide bonds. he periplasmic chaperone, Skp, has been seen to increase the solubility and resulting activity of a number of protein products.61,71,72 Overproduction of the DsbABCD proteins have increased periplasmic production of recombinant mammalian proteins with multiple disulide bonds in E. coli.73,74 DsbA typically serves as an oxidase to catalyze disulide formation75,76 and is reoxidized by the membrane protein, DsbB.77,78 In order to catalyze disulide exchange, E. coli uses DsbC as a disulide isomerase.79,80 DsbC is maintained in its active reduced state by another membrane protein, DsbD.81 Although derived from a bacterium, E. coli DsbC is still efective in catalyzing disulide isomerization for mammalian proteins thereby increasing their soluble and active yields. herefore 100 ± 50 μg/mL of DsbC, prepared as per Kim and Swartz, is included within our cell-free systems when producing disulide bonded proteins (see Table 1).54 he disulide bonded proteins summarized in Figure 1 were expressed under the conditions outlined in Table 1, respectively. he batch reactions were incubated at 30 or 37oC for 3 to 6 hours. he total protein was determined by the amount of L-[U-14C]-Leucine incorporated into the inal product and the soluble yield was measured ater the reaction mixture was centrifuged at 15,000 x g for 15 minutes. Figure 1 indicates early yields of disulide containing proteins with iodoacetamide pretreatment (0.5 to 1 mM) of the extract, addition of glutathione bufer (4 mM GSSG and 1 mM GSH) and addition of DsbC (75 to 100 μg/mL) with or without Skp addition (300 μg/mL). Using a simple batch reaction, high soluble yields of the protease domain of murine urokinase (UK), murine GM-CSF (mGM-CSF), novel B-cell lymphoma fusion proteins and a variant tPA (vtPA) containing nine disulide bonds were achieved. he soluble protein yields using these cell-free systems are greater than previously reported.54,64,65 However, larger proteins with more disulide bonds were consistently produced at lower concentrations, highlighting the need for additional process development. It was encouraging that many of the proteins listed in Figure 1 were produced with specific activities comparable to commercial standards and were largely the desired product (see Fig. 2). he latter was determined by applying ive microliters of reaction mixture to a 10% Bis-Tris SDS-PAGE gel and analyzing the resulting protein pattern by autoradiography. he enzymatic activity of urokinase and vtPA was quantiied ater centrifuging samples at 4oC, 15,000 x g for 15 minutes. For urokinase, the microtiter plate contained 10 μL of supernatant, 80 μL of assay bufer (50 mM Tris-HCl, 38 mM NaCl, pH 8.8) and 10 μL of substrate solution (2 mM Chromozym U; Roche Molecular Biochemicals, Indianapolis, IN). For the cell-free synthesized vtPA, 100 μL of assay reagent mixture bufer (nine parts Tris bufer and one part Chromozym t-PA solution: Tris bufer = 100 mM Tris-HCl pH 8.5 and 0.15% (w/v) Tween 80 and 4 mM Chromozym t-PA, Roche Molecular Biomaterials, in redistilled water) was added and the sample mixed. he mixtures were incubated at 37oC for 10 minutes. he rate of change of absorbance at 405 nM was measured in a microplate reader (SpectraMax 190, Molecular Devices). he concentration of the sample was calculated by comparison with standards as described by Kim and Swartz and Yin and Swartz.54,61 he properly folded murine (38C13) B-Cell lymphoma Id scFv concentration was determined using a radioimmunoassay,71 although the biological activity of the murine GM-CSF and murine GM-CSF portion of the cell-free expressed fusion proteins were assayed using a murine GM-CSF-dependent cell line, NFS-60 (see Fig. 2).82 he soluble cell-free expressed fusion proteins and the standard murine GM-CSF (E. coli derived murine GM-CSF from R&D Systems) were serially diluted in triplicate. Equal volumes of RPMI media (Invitrogen, NY) with 10% FCS and the log phase NFS-60 cell culture were added to the serially diluted cell-free products to test their ability to stimulate cell proliferation relative to a commercial standard (R&D Systems, Minneapolis). Volumes were adjusted to add 5000 NFS-60 cells to each well in lat-bottom 96-well tissue culture

60

Cell-Free Protein Expression

Figure 1. Production of disulfide bonded proteins by cell-free protein synthesis. UK: Urokinase protease54; vtPA: Variant of human tissue-type PA61; mGM-CSF: Murine granulocyte-macrophage colony stimulating factor62; scFv: Murine 38C13 single chain variable fragment71; mGM-scFv: Murine GM-CSF and scFv B-cell lymphoma fusion vaccine71; mGM-CSF-Im9-scFv: B-cell lymphoma fusion vaccine with the linker Im9 bacterial immunity protein. The original S30 cell extract prepared from E. coli K12 (strain A19) and the original amino acid stabilized E. coli strain, NMR2 (see Table 2) were used in these cell-free reactions.

plates (Falcon Microtest 96). Ater incubation at 37oC and 5% CO2 for 16-20 hours, 50 μL of [3H]-hymidine (Amersham) was added to each well at a inal concentration of 6.7 μCi/mL and proliferation was monitored by the incorporation of [3H]-hymidine. Following 8-10 hours of incubation at 37oC and 5% CO2, the cells were harvested onto glass iber ilter mats and washed. he [3H]-hymidine incorporation was measured using a Wallach 1450 Microbeta scintillation counter (Perkin Elmer Life Sciences). Standard murine GM-CSF, cell-free synthesized GM-CSF and fusion proteins all stimulated NFS-60 proliferation (Fig. 2) while an irrelevant product produced in the cell-free system did not (data not shown). he ED50, the concentration of protein stimulating one-half the maximum incorporation of [3H]-hymidine into NFS-60 cells, was assessed for all mGM-CSF containing protein products. he assay results shown in Figure 2 suggest that a signiicant amount of active disulide bonded proteins was produced. he speciic activities of cell-free produced mGM-CSF and fusion proteins were higher than the mGM-CSF standard. his suggests that cell-free protein synthesis produces active disulide bonded proteins with activities comparable to those produced commercially. It was concluded that our bacteria-based cell-free protein synthesis system can produce active complex proteins containing 2 to 9 disulide bonds. his is signiicant since it has been stated that most therapeutic proteins, speciically fusion proteins, cannot be produced in E. coli and must be produced in mammalian cells.83,84

Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression

Protein Urokinase vtPA murine GM-CSF murine scFv* murine GM-CSF-scFv murine GM-CSF-Im9-scFv

Specific Activity, ED50† (pM)

Active Protein Concentration (μg/mL)

----9.1 ± 1.8 --12.1 ± 3.1 25 ± 5

40x 60x 344+ 43 ± 1.4‡ 139+/ 12 ± 2.3‡ 146+/3.8 ± 1.2‡

61

*Murine scFv is the B-cell tumor (38C13) murine single chain variable fragment in the light chain-heavy chain gene sequence. †ED50, concentration of GM-CSF or the fusion protein which stimulates one-half the maximum incorporation of [3H]-Thymidine into NFS-60 cells. The ED50 of the commercial mGM-CSF standard is ~21.0 ± 0.8 pM. xActive protein yield was quantified by colorimetric based assays. ‡Active protein yield is the quantity of scFv recognized by the anti-idiotype antibody, S1C5, and quantified by a radioimmunoassay. +Active protein concentration was quantified by the cell proliferation assay.

Figure 2. Full-length therapeutic and fusion protein activity from cell-free protein synthesis reactions. Lane 1: Murine GM-CSF; Lane 2: 38C13 single chain variable fragment (scFv); Lane 3: B-cell lymphoma fusion vaccine murine GM-CSF-scFv; Lane 4: B-cell lymphoma fusion vaccine murine GM-CSF-Im9-scFv. (DS = number of disulfide bonds)

Amino Acid Stabilization by Genetic Modiication of the E. coli Genome

Recent advances in cell-free protein synthesis have enabled over 500 μg/mL of model protein (chloramphenicol acetyl transferase) production and commonly over 100 μg/mL of complex proteins (see Fig. 1) to be produced in simple batch reactions.47 hese increases in yields can be partially attributed to alleviation of limitations in energy and amino acid supply. Limitations in amino acids were irst identiied by Kim and Choi who determined that higher protein synthesis results were possible when twice the concentration of amino acids was used.35 his was conirmed by Kim and Swartz who measured the time course of amino acid depletion.39 his led to the increase in the standard concentration of amino acids within our cell-free system from 0.5 mM to 2.0 mM.39 In addition it was shown that batch feeding of amino acids could prolong protein synthesis and increase protein yields.42,44,85 hese results suggest that one of the critical factors that limited prolonged protein production in cell-free protein synthesis was the stability of amino acids. We have made signiicant advances in maintaining a stable amino acid supply by using mutated source strains to generate cell extract. his stabilization approach avoids the production of enzymes that deplete these critical substrates. E. coli strain NMR1 was constructed as described by Michel-Reydellet et al.48 his strain served as the control since it did not contain any genetic deletions for amino acid stabilization. It incorporated an active methionine synthesis pathway and the endA deletion to enhance DNA

62

Cell-Free Protein Expression

Table 2. Bacterial strains engineered to achieve amino acid stability and minimize glutathione reductases Strains

Genotype

A19 NMR1 NMR2 KC1 KC6 KGK9 KGK10

Rna-9gdhA2his-95arelA1spoT1metB1 A19ΔendAmet+ NMR1ΔtonAΔspeAΔtnaA NMR2ΔsdaAΔsdaB KC1ΔgshA KC6Δgor KC6ΔgortrxB-HA

Reference 89 48 48 48 49 103 103

a

The A19 strain in our laboratory has reverted to histidine prototrophy.

template stabilization. Similarly, as shown in Table 2, strain NMR2 was constructed per Bessette et al86 to carry unmarked deletions for the tonA gene, to minimize bacteriophage susceptibility; for the speA gene (encoding arginine decarboxylase) to stabilize arginine; and for the tnaA gene (encoding tryptophanase) to stabilize tryptophan.48 In bacteria, arginine can be used as an anabolic precursor for a pathway leading to the formation of putrescine and then to spermidine.87 he cause of arginine depletion in cell-free reactions was hypothesized to be its utilization to form these natural polyamines. he irst enzyme in this biosynthetic pathway, arginine decarboxylase (encoded by the speA gene), is inhibited by both spermidine and putrescine, although not at the levels supplied within the cell-free systems.88,89 Tryptophan catabolism in E. coli is mediated by tryptophanase, the product of the tnaA gene, whose regulation has been thoroughly described.90 his enzyme was deleted since it has been shown to degrade tryptophan as well as cysteine.91 Literature also revealed that during growth on a complex medium, E. coli consumes serine more rapidly than any other amino acid through the activity of two serine deaminases, encoded by the sdA and sdaB genes. hese serine deaminases convert serine to pyruvate and ammonia.87 Previous work has shown that the disruption of the sdaA and sdaB genes completely removed serine deaminase activity from E. coli,92 thereby suggesting that these genes were suitable targets for deletion in the cell strain used to prepare extract. he sdA and sdaB mutations were each performed with an inactivation method using the phage λ Red recombinase system.93 his resulted in a mutant strain of genotype A19ΔspeAΔtnaAΔsdaAΔsdaBΔtonAΔendAmet+, named KC1 (see Table 2). Upon completion of the genetic engineering, cell extract was prepared as described by Jewett et al.47 Strains NMR1 and NMR2 were either grown on complex 2YT media or on deined media with glucose as the carbon source.94 Strain KC1 is sensitive to serine because it contains gene deletions for the serine deaminases.95 herefore, KC1 was grown either on deined medium that does not contain serine94 or a complex 2YT medium supplemented with glucose and phosphate.43 Regardless of the medium, all strains grew similarly at 37oC with a growth rate of 1.1-1.2 hr–1. he reactions described in this section used the PANOx-SP system detailed in Table 1.43 he pK7CAT plasmid that contains the bacterial chloramphenicol acetyl transferase (cat) gene between the T7 promoter and a T7 terminator47 was used to evaluate extract performance. In addition, two other proteins were expressed in the cell-free system: he protease portion of murine urokinase (UK) and the cytokine, murine GM-CSF. As described previously, the latter two proteins require the formation of disulide bonds for activity. herefore, these proteins were expressed with the oxidative modiications to the reaction to allow formation of disulide bonds.54,61 he cell extract was pretreated for 30 minutes with 1 mM iodoacetamide (IAM) at room temperature. In addition, 5 mM glutathione bufer was added to the cell-free reaction at a 4:1 ratio of oxidized: Reduced glutathione. Finally, 75 or 100 μg/mL E. coli DsbC was added to the cell-free reaction to encourage disulide bond isomerization. Although cell extract from strain KC1 improved cell-free protein synthesis yields for bacterial proteins,48 results

Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression

63

Table 3. Summary of enzymatic activities responsible for amino acid depletion during cell-free reactions Amino Acid

Enzyme (gene)

Reaction

Arginine

Arginine decarboxylase (speA)

Arginine → agmatine + CO2

Tryptophan

Tryptophanase (tnaA) Serine deaminase (sdaA, sdaB) Glutamate-cysteine ligase (gshA)

Serine Cysteine

L-tryptophan + H20 → indole + pyruvate + NH3 L-serine → pyruvate + NH3 L-cysteine + H20 → pyruvate + NH3 + H2S L-serine → pyruvate + NH3 L-cysteine + L-glutamate + ATP → γ -glutamylcysteine + ADP + Pi

for active disulide bonded proteins, speciically urokinase and mGM-CSF, remained similar to the yields shown in Figure 1 (data not shown). Cysteine depletion has been a commonly reported problem in many cell-free reactions.40,48 Unfortunately, determining the enzymes responsible for its disappearance has been elusive. Many enzymes degrade cysteine, such as tryptophanase (encoded by tnaA) and cysteine desulhydrase (encoded by yfhO); however, deletions of these genes from our cell extract did not stabilize cysteine.48 Kim and Choi suggested that the glutathione synthesis pathway, initiated by the glutamate-cysteine ligase enzyme (gene gshA) was responsible.40 Following HPLC analysis (Dionex Amino Acid Analysis System), Calhoun and Swartz conirmed that the majority of cysteine was being nonproductively used by the cell-free reaction to synthesize glutathione.49 hus, the gshA gene was deleted from the KC1 strain, resulting in the mutant strain, KC6 (see Table 2). he depletion reactions for the four amino acids being stabilized are summarized in Table 3. he NMR1 and KC6 derived cell extracts were used in cell-free protein synthesis reactions producing CAT. Amino acid concentrations in the two strains were compared using HPLC analysis (Dionex Amino Acid Analysis System).49 As shown in Figure 3, the four amino acids were stabilized during a three-hour cell-free reaction using extract generated from the KC6 source strain. Ater stabilizing the four unstable amino acids, concentrations of the remaining amino acids were measured in reactions with extract from strain KC6. he concentrations of all amino acids remained above 1 mM during the course of a three-hour batch reaction (data not shown).49 While CAT was a convenient protein to quickly evaluate the new extract’s ability to supply cysteine to the cell-free reaction, it is a cytoplasmic bacterial protein and therefore does not require disulide bonds for activity. Since cysteine depletion was reported to be a major limitation for cell-free reactions and was not previously stabilized by the creation of the KC1 cell strain,48 it was necessary to test the production of proteins that require disulide bonds for activity. Using similar reaction conditions (described in Table 1), we found that the yields of total and soluble urokinase were 200 ± 30 and 170 ± 10 μg/mL (see Fig. 4) when using extract from the modiied cell strain KC6. his is an 80% improvement over the amount of active protein produced with extract from the control strain, NMR1. he yields of total and soluble mGM-CSF were 690 ± 80 and 550 ± 90 μg/mL, respectively (Fig. 4). Previous reports show that all of the soluble product is active as assessed by the previously described cell proliferation assay.96 his level of murine GM-CSF production is a 75% improvement over that produced with extract from control strain NMR1 (in Fig. 4) and a 37% improvement over the yields with the NMR2 extract shown in Figure 1. While the KC6 cell extract enabled improved urokinase and murine GM-CSF yields, it was not clear if such a trend would continue for a number of diferent disulide bonded proteins.

64

Cell-Free Protein Expression

Figure 3. Concentrations of four amino acids during cell-free protein synthesis reactions with PANOx-SP conditions using extract from strain KC6 (dashed line, triangle) and control strain NMR1 (solid line, squares). Results are the average of n = 3 cell-free reactions. (A) Tryptophan (B) Arginine (C) Serine (D) Cysteine.

Listed in Table 4 are the yields of various therapeutic proteins and vaccine candidates of human and murine origin. he cell extract preparation and PANOx-SP cell-free reaction conditions are described in our previous work.97 Total, soluble and active protein concentration were quantiied following a four-hour 30oC incubation by liquid scintillation counting, cell proliferation assays,62 immunoprecipitation97 and colorimetric based assays as appropriate.54,61

Figure 4. Protein synthesis yields of proteins requiring disulfide bonds in cell-free reactions using control strain NMR1 (white) versus cell extract source strain KC6 (black). The protease domain of mammalian urokinase and mGM-CSF were expressed in cell-free reactions. Results are the average of n = 9 PANOx-SP cell-free reactions.

65

Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression

Table 4. Cell-free protein synthesis of commercially relevant disulfide bonded proteins. All proteins were expressed with the PANOx-SP cell-free system using KC6 extract. Protein activity was measured using cell proliferation,62 immunoprecipitation,97 and colorimetric based methods.49,61,62 Results are the average of n = 6 cell-free reactions and n = 3 cell extracts49,61,97 Protein Concentration with KC6 + 1 mM IAM (n = 6 reactions, μg/mL) Protein*

Total

mGM-CSF 493 ± 22 hGM-CSF 488 ± 31 hG-CSF 539 ± 16 hIFNα2b 484 ± 44 mvlvh† ‡ 451 ± 41 hvlvh‡ 162 ± 15 Im9-hvlvh‡ 302 ± 41 mGM-Im9-mvlvh†‡ 719 ± 49 mGM-Im9-hvlvh‡ 137 ± 6 vtPA 257 ± 49 Urokinase 431 ± 73

Soluble 478 ± 31 466 ± 22 410 ± 14 440 ± 10 306 ± 22 87 ± 23 251 ± 63 476 ± 27 94 ± 11 174 ± 18 183 ± 24

Active 1

478 4661 4101 1712 2832 722 2012 4761/1252 941/612 >603 ~403

Average Protein Concentration (n = 3 extracts, μg/mL) Soluble 536 ± 73 575 ± 74 453 ± 76 497 ± 117 239 ± 94 93 ± 31 290 ± 59 378 ± 142 169 ± 84 151 ± 11 167 ± 48

1 Cell proliferation activity assay. 2Immunoprecipitation activity assay. 3Colorimetric activity assay. *The mvlvh is a 38C13 murine single chain variable fragment in the light chain-heavy chain sequence. The hvlvh is a human B-cell single chain variable fragment in the light chain-heavy chain sequence with a (GGGGS)3 linker. †The E. coli chaperone, Skp (300 μg/mL), had a positive affect on protein folding. ‡Contains a chloramphenicol acetyl transferase leader sequence.

An immunoprecipitation assay was developed to assess correct folding of human Interferon alpha 2b (hIFNα2b), single chain variable fragments and the C-terminus of the fusion proteins. Twenty microliters of Protein G Sepharose (Amersham Biosciences) resin in PBS was mixed with 100 μg of the relevant monoclonal antibody for saturation of the bead surface. he negative control antibodies were mouse IgG1 (1C1) or IgG2a (Chemicon Inter., Temecula, CA). he full-length murine (S1C5) and human (7D11/C6) anti-idiotype antibodies were obtained from the Ronald Levy lab at Stanford University.98-100 he anti-human interferon alpha antibody (mouse IgG1) was purchased from Assay Designs, Inc. (Ann Arbor, Michigan). hese suspensions were rotated for 48 hours at 4oC and subsequently washed 3 times with PBS. Approximately 20 μL of resin was mixed with 10-30 μL of the resultant cell-free reaction mixture ater expression of the single chain variable fragment or hIFNα2b with [U-14C]-Leucine incorporation and the suspension was rotated for 24 hours at 4oC. he experimental procedure allowed all active protein to bind and negative controls using irrelevant proteins indicated that all binding was speciic. he resin was pelleted and resuspended in 200 μL NDET (1% (v/v) IGEPAL CA-630 from Sigma-Aldrich, 0.4% (w/v) deoxycholate, 66 mM EDTA, 10 mM Tris-HCl pH 7.4) and 0.3% SDS. he slurry was then layered over a 1 mL sucrose pad (50:50 mix of 60% sucrose & NDET + 0.3% SDS), centrifuged and washed with 1 mL of NDET + SDS and water. he supernatant was removed and the resin resuspend in sample bufer and boiled for 2 minutes. he supernatant was loaded onto a NuPAGE 10% Bis-Tris NuPAGE gel (Invitrogen, CA) for quantiication of correctly folded protein. Gels were stained with Simply Blue SafeStain (Invitrogen, CA) and dried with a gel dryer, Model 583 (Bio-Rad, CA) before being exposed to a Kodak scientiic imaging ilm (Rochester, NY). Quantiication was performed using an AlphaImager® Imaging System and AlphaEase®FC Analysis Sotware (Alpha Innotech Corporation). he concentration of correctly folded protein

66

Cell-Free Protein Expression

was determined by comparison to a standard curve prepared with the corresponding cell-free product of known concentration. As listed in Table 4, 410 μg/mL of active hG-CSF was produced. his was signiicant because the commercial product will soon be of patent and a generic G-CSF could save the healthcare system thousands of dollars per patient. Also new B-cell lymphoma fusion vaccines were synthesized in yields of 94 to 476 μg/mL. hese yields are 2 to 3 times greater than those previously reported for similar complex fusion proteins.71 Murine scFv (mvlvh) production and active yields were also increased 3 and 7-fold relative to yields previously reported.71 he large increase in active yields is likely due to the cell-free system optimization, the improved KC6 cell extract and the speciicity of the immunoprecipitation assay. Interestingly, the yield of a human lymphoma scFv (hvlvh) was consistently lower than its murine counterpart. However the amount of protein produced is still above or equal to yields for similar proteins that were previously reported.71 As listed in Table 4, this increase in yields was achieved for all the proteins using three KC6 cell extracts that had been prepared diferentially and generic PANOx-SP cell-free reaction conditions.97

Elimination of Glutathione Reductase in the E. coli Cell-Extract

In the previous section, iodoacetamide (IAM) was added to the cell-free extract to derivatize the active site cysteines of thioredoxin reductase (TrxB) and glutathione reductase (Gor), thereby inactivating those enzymes and stabilizing an oxidized disulide bond environment.54 his has proven to be an important component for the cell-free production of proteins that require disulide bonds. However, there is no control over which cysteine residues in the cell extract are derivatized by IAM. he inactivation of TrxB and Gor is advantageous, but chemical modiication of other proteins such as DsbA and DsbC is potentially detrimental to protein folding. For example IAM pretreatment causes roughly a 15% reduction in protein yields when PEP is used as an energy source.54 Although initial attempts to delete either glutathione reductase or thioredoxin reductase from the cell strain used to make the extract had little efect on the rate of GSSG reduction,54 we predicted that such modiications would be beneicial to our improved KC6 cell strain. However, it was noted that deletion of both gor and trxB results in the mutational conversion of the enzyme AhpC from a peroxiredoxin to a disulide reductase. he ahpC mutation promotes more rapid growth but also stimulates disulide bond reduction.20 In this section we show advances toward producing a cell-free extract with stabilized oxidized glutathione and with the ability to produce a wide variety of important therapeutic proteins that require disulide bonds, without inactivating key metabolic enzymes such as DsbC and glyceraldehye 3-phosphate dehydrogenase (G-3PDH). he Datsenko and Wanner93 method was used to make two chromosomal changes to the cell strain KC6.49 his resulted in a cell-free extract devoid of Gor activity and from which Trx protein could be removed (see Table 2). he irst mutation deleted gor, eliminating the direct enzymatic reduction of GSSG, creating the mutant cell strain KGK9 (see Table 2). GSSG reduction activity was smaller but was still present in the KGK9 extract. he reduction activity could be eliminated by pretreating the extract with a lower IAM concentration than was required for KC6 extract. We reasoned that continued reduction of GSSG in gor deleted strains was due to the activity of the thioredoxin reductase mediated system. However as stated previously, the thioredoxin reductase gene cannot be deleted from the gor deleted background without a compensatory mutation in ahpC. In order to disable the thioredoxin system, a puriication tag was added to the C-terminus of trxB in the chromosome. his strain grows normally and does not promote the ahpC mutation because the thioredixon mediated reduction pathway is still intact. Yet, the hemagglutinin (HA) puriication tag allows TrxB to be removed from the extract before being used in CFPS reactions. his resulted in the mutant cell strain KGK10 (see Table 2). he extracts from strains KC6 and KGK10 were evaluated during chloramphenicol acetyl-transferase (CAT) production in cell-free reactions within the cytomim cell-free system (see Table 1). Cell-free reactions were incubated at 37oC for 3 hours. Reactions were conducted with KC6 and KGK10 extracts with no pretreatment with KGK10 with TrxB removal (KGK10–TrxB) and ater 1 mM IAM pretreatment of each extract. he rate of reduction of GSSG was monitored by

Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression

67

Figure 5. The increase in total sulfhydryl concentration over time due to reduction of GSSG in cell-free protein synthesis reactions. Error bars are +/– one standard deviation for n = 3. KC6 and KGK10 were not pretreated with 1 mM IAM (KC6 and KGK10), the TrxB was removed from KGK10 (KGK10 – TrxB) and 1 mM IAM cell extract pretreatment was conducted on KGK10 (KGK10 + IAM).102

measuring the formation of free sulhydryl (–SH) groups by the DTNB reaction.101 As shown in Figure 5, the KGK10 strain catalyzes a much slower rate of GSSG reduction than KC6, lowering the rate from 44 to 14 μM/min. his was not surprising because KGK10 lacks gor, but a comparable decrease was not observed in another Δgor strain.54 As seen in Figure 5, the removal of TrxB from KGK10 extract did not result in any further stabilization of the GSSG. herefore, it appears that TrxB does not directly, or indirectly through another enzyme, reduce GSSG in CFPS reactions. he ahpC gene was sequenced and was found not to be mutated in KGK10, so none of the known cytoplasmic reduction pathways in E. coli are responsible for the reduction of GSSG in cell-free reactions using KGK10 extract with TrxB removed. It was hypothesized that the reduction is due to unknown enzyme(s) in the extract, because control reactions without extract have a slight rate of oxidation of thiol groups. Furthermore, the unknown enzyme(s) must utilize cysteine in their active site because IAM pretreatment completely stabilizes KGK10 extracts (see Fig. 5). Since the removal of TrxB did not further reduce the rate of GSSG and required a costly puriication procedure (anti-HA resin retails for greater than $500/mL), GSSG stabilization by IAM pretreatment of KGK10 extracts was evaluated.102 he results shown in Figure 5 suggest that the deletion of the gor gene would signiicantly lower the amount of IAM required for pretreatment. KGK10 reactions with 10 μM or less IAM pretreatment have GSSG reduction rates between 20 and 30 μM/min regardless of pH. Near physiological pH, 50 μM IAM pretreatment is suicient to eliminate the reduction of GSSG. Concentrations greater than 50 μM IAM led to a slow rate of oxidation of sulhydryls, similar to what is observed in reactions with no extract. To test the productivity of these extracts, PANOx-SP cell-free reactions were conducted at 30oC for 4 hours in the oxidized environment described by Goerke and Swartz.97 he beneits of the KGK10 extract were clear when producing the complex disulide-containing proteins listed in Table 5. With the lower IAM concentration, pretreated KGK10 extract supported high level protein expression. Active yields of hG-CSF, human scFv (hvlvh) and human fusion vaccines, were improved 61 to 131% above those achieved using the KC6 cell-free platform. Many of the other proteins listed in Table 5 displayed modest gains while a few displayed equivalent yields to that seen in the KC6 cell-free platform. However there was a reproducible and statisti-

68

Cell-Free Protein Expression

Table 5. Product yield using KGK10 extracts with 50 μM IAM pretreatment versus using KC6 extract with 1 mM IAM pretreatment. All proteins were expressed with the PANOx-SP cell-free system. Protein activity was measured using cell proliferation assays,62 immunoprecipitation,97 and colorimetric based methods.49,61,62 Results are the average of n = 6 reactions49,61,97 Protein Concentration with KGK10 + 50 μM IAM (n = 6 reactions, μg/mL) Protein* mGM-CSF hGM-CSF hG-CSF hINFα2b mvlvh†‡ hvlvh‡ Im9-hvlvh‡ mGM-Im9-mvlvh†‡ mGM-Im9-hvlvh‡ tPA Urokinase

Total 592 ± 22 640 ± 10 755 ± 40 228 ± 7 410 ± 17 196 ± 12 412 ± 37 659 ± 36 348 ± 19 208 ± 10 540 ± 26

Soluble 541 ± 47 610 ± 10 713 ± 36 200 ± 8 248 ± 13 116 ± 5 397 ± 44 396 ± 5 328 ± 12 206 ± 40 323 ± 28

Percent Change KGK10 versusKC6

Active 1

541 6101 7131 782 2312 962 2982 3961/1572 3281/1442 -->533

Soluble Protein Active Protein +1 +6 +58 –60 +4 +24 +37 +5 +94 +36 +93

+13 +31 +74 –55 –19 +61 +57 –1.0 +131 --+47

1

Cell proliferation activity assay. 2Immunoprecipitation activity assay. 3Colorimetric activity assay. *The mvlvh is a 38C13 murine single chain variable fragment in the light chain-heavy chain orientation. The hvlvh is a patient specific B-cell human single chain variable fragment in the light chain-heavy chain orientation. †The E. coli chaperone, Skp (300 μg/mL), had a positive affect on protein folding. ‡ Contains a chloramphenicol acetyl transferase leader sequence.

cally signiicant reduction in the amount of soluble and active hIFNα2B produced. Interestingly, a more oxidizing redox potential was optimal for all proteins listed in Table 5, accept hIFNα2b. Stabilization of the oxidized cell-free environment highlights this fact since 55% less hIFNα2b was produced relative to the yield from KC6 cell extract treated with 1 mM IAM.

Using Glucose as the Energy Source for Disulide Bond Containing Proteins

In addition to oxidative phosphorylation, the glycolytic pathway has been shown to generate energy for cell-free protein synthesis.45 Glucose is the preferred carbon and energy source for many organisms and is one of the least expensive and most desirable commercial substrates for industrial biotechnology applications. In order for cell-free protein synthesis to efectively complete with conventional cellular approaches to protein production, it would be highly advantageous to develop a system where glucose can be used as the energy source. To use glucose for cell-free protein synthesis, the cytomim system was adapted slightly (see Table 1). For example, the conversion of glucose to acetate and lactate results in a decline in pH, so the use of a bufer or other appropriate pH control is necessary. In addition, without the use of a phosphorylated energy source, there is too little inorganic phosphate present in the reaction since phosphate is necessary for the initial steps in the glycolytic pathways. When 90 mM Bis-Tris bufer and 10 mM phosphate are added to cell-free reaction with glucose as the energy source, signiicant model protein yields result (~550 μg/mL of chloramphenicol acetyl transferase (CAT)).45 To determine if the decreased concentration of IAM required for the KGK10 extract would preserve sulhydryl requiring activities such as that of glyceraldehyde 3-phosphate dehydrogenase (which is required for glucose utilization), reactions were conducted for the production

Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression

69

Figure 6. Cell-free production of murine GM-CSF in reactions fueled with glucose. KC6 and KGK10 extracts were pretreated with the indicated concentration of IAM. The total (black bars) and active (white bars) yields are indicated. The data are an average of n = 6 experiments, with error bars of +/– one standard deviation.

of mGM-CSF. Figure 6 shows that the KGK10 extract both with and without 50 μM IAM pretreatment allows glucose to be used as an energy source when producing proteins that require disulide bonds. More than a 2-fold increase in soluble mGM-CSF was produced with KGK10 rather than KC6 in the glucose fueled cell-free system when IAM was not used to pretreat the cell extract. In comparison to glucose fueled KC6 extract treated with 1 mM IAM, a 3-fold increase in mGM-CSF resulted using the KGK10 cell extract treated with 50 μM IAM. When comparing the PEP fueled cell-free system and the glucose cell-free system, Knapp et al produced approximately 50% as much mGM-CSF.103 his is signiicant since PEP is the most expensive cell-free reaction component. he elimination of PEP paves the way for further cost reduction in this cell-free production platform.

Scale-Up of Cell-Free Batch Systems

Despite all the recent success, a signiicant barrier remained: he inability to easily and eiciently increase the volume of protein synthesis reactions. In this section we describe principles used to scale-up the cell-free technology. In addition, we discuss the incorporation of these principles into practical methods that can be easily combined with technologies used for conventional bacterial protein expression. In theory this would then allow us to scale-up the cell-free system to any size. he scale-up methodology was veriied by expressing the customized B-cell lymphoma vaccine containing four disulide bonds. Several types of reactors have been developed in an efort to increase the scale of the cell-free reactions. As described previously, the main types of reactors are batch, continuous, semi-continuous and hollow-iber. Even though each of these conigurations approaches the problem of scale-up diferently, so far all have been unable to surpass the reaction scale of a few milliliters because of inherent ineiciencies and complexity of design. As discussed previously, batch reactions work

70

Cell-Free Protein Expression

well at the analytical scale and are capable of expressing a variety of properly folded bacterial and mammalian proteins. What makes batch reactors attractive is the ease of operation, eicient use of substrates and simple reactor design. While it was clear that our cell-free system was capable of producing soluble protein therapeutics and vaccine candidates at the 15 to 100 μL scale,51 our vision was to rapidly produce enough puriied product for in vivo animal studies during drug development. his would require approximately 2 to 5 mg of each vaccine product. Since yields of the mGM-CSF-scFv fusion protein decrease signiicantly when produced at the 0.5 mL scale in a test tube (see Fig. 7), a new approach was needed. herefore a scale-up technology was developed on the presumption that low yields were caused by oxygen transfer limitations, resulting in reduced energy supply and poor protein expression. he “thin-ilm” approach was developed by Voloshin and Swartz.55 his system provided the simplicity and rapid scalability that would be ideal for the synthesis of patient speciic vaccines. One can envision stacks of large trays in an incubator, each producing vaccine for a diferent patient. It was found that this new method of scale-up avoided any signiicant reduction in yields. his was true for both the Cytomim and PANOx-SP cell-free systems. he resulting production kinetics for the mGM-CSF-scFv vaccine at the 0.5 mL scale are shown in Figure 7. hese data illustrate that the thin-ilm approach preserved the fusion protein synthesis kinetics as reaction volume increased. his is due to the increase in surface to volume ratio, enabling good gas transfer and eicient energy regeneration and also to providing more hydrophobic surface area. To produce fusion protein vaccines for tumor challenge studies, the cell-free reactions were scaled up to 30 mL. A 5 mL reaction mixture was placed as a large drop in each of six standard sterile Fisherbrand petri dishes (100 x 15 mm) and incubated at 30°C for 4 hours. Ater 4 hours, the soluble protein was separated by centrifugation at 15,000 x g for 15 min. he soluble fraction was loaded on a 5 mL Ni-NTA column (Qiagen), which was equilibrated with a 10 mM imidazole, 300 mM NaCl and 50 mM phosphate bufer (pH 8.0). he column was then washed with 30 mL of 25 mM imidazole in the same bufer and eluted with 250 mM imidazole. he puriied products were then concentrated with Amicon Ultra-15 Centrifugal Filter Units (5,000 MWCO) and dialyzed with 7,000 MWCO Slide-A-Lyzer (PIERCE) units against phosphate bufered saline (PBS). Lastly, the vaccine was formulated to a inal concentration of 0.1% Tween-20 (Sigma) and sterilized by iltration

Figure 7. Murine GM-CSF-scFv fusion vaccine candidate accumulation kinetics in the Cytomim and PANOx-SP cell-free systems. 0.5 μL reaction conducted with the thin film format produced significantly higher yields than test tube reactions. 55

Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression

71

through a 0.2 μM ilter (Nalgene). his scale-up methodology allowed us to quickly provide a number of biologically active vaccine candidates for murine tumor challenge studies.104

Conclusions

Using typical cellular production techniques with mammalian cells, production, puriication and characterization takes 30 days on average, while the manufacturing process takes 10 weeks per batch.83 Each of these steps in manufacturing must be optimized for the particular biologic compound produced and there are no fundamental techniques that can be used for all products as is true for the manufacture of chemical compounds. Optimizing each technique is expensive and extremely labor-intensive, which is why large biotechnology companies keep such techniques conidential. his complicates issues for prospective generic biologic manufacturers and commonly results in additional process development requirements. One solution to this dilemma is to use cell-free protein synthesis platforms. We have engineered generic cell-free platforms and here shown that cell-free protein synthesis can be used for expression of active therapeutic proteins and patient speciic vaccines. We initially demonstrated that complex synthetic mammalian proteins can be expressed and folded correctly in cell-free protein synthesis by modifying the reaction environment to promote oxidative folding. We have taken advantage of this lexibility and have developed a dramatically improved cell-free system. First, four amino acids were identiied that were depleted during the cell-free protein synthesis reaction: Arginine, tryptophan, cysteine and serine. Next, the speciic enzymatic activities responsible for the amino acid instabilities were determined. hese enzymatic activities include arginine decarboxylase, tryptophanase, serine deaminase and glutamate-cysteine ligase. Finally, the genes encoding those enzymes (speA, tnaA, sdaA, sdaB and gshA, respectively) were then deleted from the E. coli strain used to make the cell extract for the cell-free reactions. By combining ive genetic modiications, a cell extract has been produced that maintains all amino acid concentrations over 1 mM in a 3-hour batch, cell-free reaction. his eliminates one of the major limitations of cell-free reactions and allows a number of active disulide bonded proteins to be produced at acceptable levels. Extensive development was conducted to avoid using IAM pretreatment for cell-free reactions by disabling the cytoplasmic reduction systems in the cell extract. he glutathione reductase mediated system was disabled by deleting the gor gene from the KC6 strain. he thioredoxin reductase mediated system was to be minimized by adding a hemagglutinin puriication tag to the trxB gene in the chromosome and by then removing TrxB from the extract. It was anticipated that ainity removal of the tagged TrxB would result in a cell-free extract devoid of all reduction systems, but this was not the case. However the concentration of IAM required to stabilize GSSG in KGK10 extract was reduced 20-fold. For many proteins this decease in IAM concentration resulted in increased protein yields. he resulting cell-free platform now uses glucose as an energy source. hese advances enabled 195 μg/mL of active murine GM-CSF to be synthesized in a glucose fueled cell-free system. Although these yields are somewhat lower than those obtained when using PEP, the substantially lower substrate costs suggest that the KGK10-based system is much more commercially viable. Additionally we have shown that the thin-ilm scale-up approach can be successfully applied to scale-up vaccine synthesis to rapidly and eiciently produce active vaccines in the amounts required for clinical studies. hese advancements have a broad impact on the utility of the cell-free system and the ield of patient speciic medicine. his work demonstrates the advantages of cell-free protein synthesis: Short expression time, ability to easily multiplex production and simpler downstream processing. All of the advantages of the cell-free expression technology can now be eiciently utilized in various research applications. Commercial applications such as pharmaceutical protein production of targets that are diicult to express and fold in living cells or that require patient speciic approaches now come within reach as well as other critical applications such as rapid response for bio-agent defense using protein vaccines. hese applications are further aided by the short developmental production cycles that this cell-free protein synthesis technology ofers.

72

Cell-Free Protein Expression

Acknowledgements

A.R. Goerke is funded by a Doctoral Fellowship from Merck & Company, Inc. Special thanks to the Merck Biologics and Engineering Department for additional funding.

References

1. Sylvester R. Clinical applications of colony-stimulating factors: A historical perspective. Am J Health-Syst Pharm 2002; 59:S6-S12. 2. Baron S, Tyring S, Fleischmann W, Jr et al. The interferons. Mechanisms of action and clinical applications. JAMA 1991; 266:1375-1383. 3. Ponec D, Irwin D, Haire W et al. Recombinant Tissue Plasminogen Activator (Alteplase) for Restoration of Flow in Occluded Central Venous Access Devices: A Double-Blind Placebo-Controlled Trial—he Cardiovascular hrombolytic to Open Occluded Lines (COOL) Eicacy Trial. J Vasc Interv Radiol 2001; 12:951-955. 4. Armitage J. Treatment of nonHodgkin’s lymphoma. N Engl J Med 1993; 328:1023-1030. 5. www.lymphoma.org 6. Chu L, Robinson D. Industrial choices for protein production by large-scale cell culture. Curr Opin Biotechnol 2001; 12:80-187. 7. Swartz J. Advances in Escherichia coli production of therapeutic proteins. Curr Opin Biotechnol 2001; 12:95-201. 8. Anderson C, Straus J, Dudock B. Preparation of a cell-free protein-synthesizing system from wheat germ. Methods Enzymol 1983; 101:635-644. 9. Andersen D, Krummen L. Recombinant protein expression for therapeutic applications. Curr Opin Biotechnol 2002; 13:117-123. 10. Balkow K, Hunt T, Jackson R. Control of protein synthesis in reticulocyte lysates: the efect of nucleotide triphosphates on formation of the translational repressor. Biochem Biophys Res Commun 1975; 67:366-375. 11. Blattner F, Plunkett G, 3rd Bloch C et al. he complete genome sequence of Escherichia coli K-12. Science 1997; 277:1453-1474. 12. Chang J, Swartz J. In Protein Folding: In vivo and in vitro (ed. Cleland J.L.) (ACS, 1993). 13. Qiu J, Swartz J, Georgiou G. Expression of active human tissue-type plasminogen activator in Escherichia coli. Appl Environ Microbiol 1998; 64:4891-4896. 14. Ortenberg R, Beckwith J. Functions of thiol-disulide oxidoreductases in E. coli: redox myths, realities, and practicalities. Antioxid Redox Signal 2003; 5:403-411. 15. Jeong K, Lee S. Secretory production of human leptin in Escherichia coli. Biotechnol Bioeng 2000; 67:398-407. 16. Bowden G, Georgiou G. Folding and aggregation of beta-lactamase in the periplasmic space of Escherichia coli. J Biol Chem 1990; 265:16760-16766. 17. Schein C. Production of soluble recombinant proteins in bacteria. Biotechnology 1989; 4:1141-1149. 18. Murby M, Uhlen M, Stahl S. Upstream strategies to minimize proteolytic degradation upon recombinant production in Escherichia coli. Protein Expr Purif 1996; 7:129-136. 19. Prinz W, Aslund F, Holmgren A et al. he role of the thioredoxin and glutaredoxin pathways in reducing protein disulide bonds in the Escherichia coli cytoplasm. J Biol Chem 1997; 272:15661-15667. 20. Ritz D, Lim J, Reynolds C et al. Conversion of a peroxiredoxin into a disulide reductase by a triplet repeat expansion. Science 2001; 294:158-160. 21. Bessette P, Aslund F, Beckwith J et al. Eicient folding of proteins with multiple disulide bonds in the Escherichia coli cytoplasm. Proc Natl Acad Sci USA 1999; 96:13703-13708. 22. Jurado P, Ritz D, Beckwith J et al. Production of functional single-chain Fv antibodies in the cytoplasm of Escherichia coli. J Mol Biol 2002; 320:1-10. 23. Levy R, Weiss R, Chen G et al. Production of correctly folded Fab antibody fragment in the cytoplasm of Escherichia coli trxB gor mutants via the coexpression of molecular chaperones. Protein Expr Purif 2001; 23:338-347. 24. Venturi M, Seifert C, Hunte C. High level production of functional antibody Fab fragments in an oxidizing bacterial cytoplasm. J Mol Biol 2002; 315:1-8. 25. McCormick A, Kumagai M, Hanley K et al. Rapid production of speciic vaccines for lymphoma by expression of the tumor-derived single-chain Fv epitopes in tobacco plants. Proc Natl Acad Sci USA 1999; 96:703-708. 26. McCormick A, Reinl S, Cameron T et al. Individualized human scFv vaccines produced in plants: humoral anti-idiotype responses in vaccinated mice conirm relevance to the tumor Ig. J Immunol Methods 2003; 278:95-104.

Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression

73

27. Michel-Reydellet N, Woodrow K, Swartz J. Increasing PCR fragment stability and protein yields in a cell-free system with genetically modiied Escherichia coli extracts. J Mol Microbiol Biotechnol 2005; 9:26-34. 28. Kigawa T, Yamaguchi-Nunokawa E, Kodama K et al. Selenomethionine incorporation into a protein by cell-free synthesis. J Struct Funct Genomics 2002; 2:29-35. 29. Noren C, Anthony-Cahill S, Griith M et al. A general method for site-speciic incorporation of unnatural amino acids into proteins. Science 1989; 244:182-188. 30. Hirao I, Ohtsuki T, Fujiwara T et al. An unnatural base pair for incorporating amino acid analogs into proteins. Nat Biotechnol 2002; 20:177-182. 31. Zhang Z, Gildersleeve J, Yang Y et al. A new strategy for the synthesis of glycoproteins. Science 2004; 303:371-373. 32. Liao Y, Jeng J, Wang C et al. Removal of N-terminal methionine from recombinant proteins by engineered E. coli methionine aminopeptidase. Protein Science 2004; 13:1802-1810. 33. Nirenberg M, Matthaei J. he dependence of cell-free protein synthesis in E. coli upon naturally occurring or synthetic polyribonucleotides. Proc Natl Acad Sci USA 1961; 47:1588-1602. 34. Nirenberg M, Matthaei J, Jones O et al. Approximation of genetic code via cell-free protein synthesis directed by template RNA. Fed Proc 1963; 22:55-61. 35. Kim D, Choi C. A semicontinuous prokaryotic coupled transcription/translation system using a dialysis membrane. Biotechnol Prog 1996; 12:645-649. 36. Spirin A, Baranov V, Ryabova L et al. A continuous cell-free translation system capable of producing polypeptides in high yield. Science 1988; 242:1162-1164. 37. Ryabova L, Vinokurov L, Shekhovtsova E et al. Acetyl phosphate as an energy source for bacterial cell-free translation systems. Anal Biochem 1995; 226:184-186. 38. Shen X-C, Yao S-L, Terada S et al. Protein productivity of cell-free translation was improved by removing phosphatase from wheat germ extract with immunoprecipitation. Biochemical Engineering Journal 1998; 2:23-28. 39. Kim D, Swartz J. Regeneration of adenosine triphosphate from glycolytic intermediates for cell-free protein synthesis. Biotechnol Bioeng 2001; 74:309-316. 40. Kim R, Choi C. Expression-independent consumption of substrates in cell-free expression system from Escherichia coli. J Biotechnol 2000; 84:27-32. 41. Shen X, Yao S, Terada S et al. Protein productivity of cell-free translation was improved by removing phosphatase from wheat germ extract with immunoprecipitation. Biochem Eng J 1998; 2:23-28. 42. Shimizu Y, Inoue A, Tomari Y et al. Cell-free translation reconstituted with puriied components. Nat Biotechnol 2001; 19:751-755. 43. Jewett M, Swartz J. Mimicking the Escherichia coli cytoplasmic environment activates long-lived and eicient cell-free protein synthesis. Biotechnol Bioeng 2004; 86:19-26. 44. Kim D, Swartz J. Prolonging cell-free protein synthesis with a novel ATP regeneration system. Biotechnol Bioeng 1999; 66:180-188. 45. Calhoun K, Swartz J. Energizing cell-free protein synthesis with glucose metabolism. Biotechnol Bioeng 2005; 90:606-613. 46. Calhoun K, Swartz J. An economical method for cell-free protein synthesis using glucose and nucleoside monophosphates. Biotechnol Prog 2005; 21:1146-1153. 47. Jewett M, Voloshin A, Swartz J. in Gene cloning and expression technologies (eds. Weiner, M. & Lu, Q.) 391-411 (Eaton Publishing, Westborough, MA, 2002). 48. Michel-Reydellet N, Calhoun K, Swartz J. Amino acid stabilization for cell-free protein synthesis by modiication of the Escherichia coli genome. Metab Eng 2004; 6:197-203. 49. Calhoun K, Swartz J. Total amino acid stabilization during cell-free protein synthesis reactions. J Biotechnol 2006. 50. Kigawa T, Yabuki T, Yoshida Y et al. Cell-free production and stable-isotope labeling of milligram quantities of proteins. FEBS Lett 1999; 442:15-19. 51. Jewett M, Swartz J. Rapid expression and puriication of 100 nmol quantities of active protein using cell-free protein synthesis. Biotechnol Prog 2004; 20:102-109. 52. Chekulayeva M, Kurnasov O, Shirokov V et al. Continuous-exchange cell-free protein-synthesizing system: synthesis of HIV-1 antigen Nef. Biochem Biophys Res Commun 2001; 280:914-917. 53. Lamla T, Stiege W, Erdmann V. An improved protein bioreactor: eicient product isolation during in vitro protein biosynthesis via ainity tag. Mol Cell Proteomics 2002; 1:466-471. 54. Kim D, Swartz J. Eicient production of a bioactive, multiple disulide-bonded protein using modiied extracts of Escherichia coli. Biotechnol Bioeng 2004; 85:122-129. 55. Voloshin A, Swartz J. Eicient and scalable method for scaling up cell free protein synthesis in batch mode. Biotechnol Bioeng 2005; 91:516-521.

74

Cell-Free Protein Expression

56. Sitaraman K, Esposito D, Klarmann G et al. A novel cell-free protein synthesis system. J Biotechnol 2004; 110:257-263. 57. Zawada J, Swartz J. Maintaining rapid growth in moderate-density Escherichia coli fermentations. Biotechnol Bioeng 2005; 89:407-415. 58. Liu D, Zawada J, Swartz J. Streamlining Escherichia coli S30 extract preparation for economical cell-free protein synthesis. Biotechnol Prog 2005; 21:460-465. 59. Kim D, Swartz J. Oxalate improves protein synthesis by enhancing ATP supply in a cell-free system derived from Escherichia coli. Biotechnol Lett 2000; 22:1537-1542. 60. Jewett M. in Department of Chemical Engineering 240 (Stanford University, Stanford, 2005). 61. Yin G, Swartz J. Enhancing multiple disulide bonded protein folding in a cell-free system. Biotechnol Bioeng 2004; 86:188-195. 62. Yang J, Kanter G, Voloshin A et al. Expression of Active Murine Granulocyte-Macrophage ColonyStimulating Factor in an Escherichia coli. Cell-Free System. Biotechnol Prog 2004; 20:1689-1696. 63. Kudlicki W, Odom O, Kramer G et al. Chaperone-dependent folding and activation of ribosome-bound nascent rhodanese. Analysis by luorescence. J Mol Biol 1994; 244:319-331. 64. Kawasaki T, Gouda M, Sawasaki T et al. Eicient synthesis of a disulide-containing protein through a batch cell-free system from wheat germ. Eur J Biochem 2003; 270:4780-4786. 65. Ryabova L, Desplancq D, Spirin A et al. Functional antibody production using cell-free translation: efects of protein disulide isomerase and chaperones. Nat Biotechnol 1997; 15:79-84. 66. Kim D, Fernholz E, Swartz J. Cell-free expression of proteins containing multiple disulide bonds (ed. Swartz J.R.) (Springer-Verlag Berlin Heidelberg, New York, 2003). 67. Tsalkova T, Zardeneta G, Kudlicki W et al. GroEL and GroES increase the speciic enzymatic activity of newly-synthesized rhodanese if present during in vitro transcription/translation. Biochemistry 1993; 32:3377-3380. 68. Jiang X, Ookubo Y, Fujii I et al. Expression of Fab fragment of catalytic antibody 6D9 in an Escherichia coli in vitro coupled transcription/translation system. FEBS Lett 2002; 514:290-294. 69. Merk H, Stiege W, Tsumoto K et al. Cell-free expression of two single-chain monoclonal antibodies against lysozyme: efect of domain arrangement on the expression. J Biochem (Tokyo) 1999; 125:328-333. 70. Buchner J. (Roche Diagnostics Operations, Inc. (Indianapolis, IN), United States Patent, 2002). 71. Yang J, Kanter G, Voloshin A et al. Rapid expression of vaccine proteins for B-cell lymphoma in a cell-free system. Biotechnol Bioeng 2005; 89:503-511. 72. Bothmann H, Pluckthun A. Selection for a periplasmic factor improving phage display and functional periplasmic expression. Nat Biotechnol 1998; 16:376-380. 73. Joly J, Leung W, Swartz J. Overexpression of Escherichia coli oxidoreductases increases recombinant insulin-like growth factor-I accumulation. Proc Natl Acad Sci USA 1998; 95:2773-2777. 74. Kurokawa Y, Yanagi H, Yura T. Overproduction of bacterial protein disulide isomerase (DsbC) and its modulator (DsbD) markedly enhances periplasmic production of human nerve growth factor in Escherichia coli. J Biol Chem 2001; 276:14393-14399. 75. Bardwell J, McGovern K, Beckwith J. Identiication of a protein required for disulide bond formation in vivo. Cell 1991; 67:581-589. 76. Kamitani S, Akiyama Y, Ito K. Identiication and characterization of an Escherichia coli gene require for the formation of correctly foled alkaline phsphatase, a periplasmic enzyme. EMBO J 1992; 11:57-62. 77. Bardwell J, Lee J, Jander G et al. A pathway for disulide bond formation in vivo. Proc Natl Acad Sci USA 1993; 90:1038-1042. 78. Guilhot C, Jander G, Martin N et al. Evidence that the pathway of disulide bond formation in Escherichia coli involves interactions between the cysteines of DsbB and DsbA. Proc Natl Acad Sci USA 1995; 92:9895-9899. 79. Zapun A, Missiakas D, Raina S et al. Structural and functional characterization of DsbC, a protein involved in disulide bond formation in Escherichia coli. Biochemistry 1995; 34:5075-5089. 80. Sone M, Akiyama Y, Ito K. Diferential in vivo roles played by DsbA and DsbC in the formation of protein disulide bonds. J Biol Chem 1997; 272:10349-10352. 81. Missiakas D, Schwanger F, Raina S. Identiication and characterization of a new disulide isomerase-like protein (DsbD) in Escherichia coli. EMBO J 1995; 14:3415-3424. 82. Miyajima A, Otsu K, Schreurs J et al. Expression of murine and human granulocyte-macrophage colony-stimulating factors in S. cerevisiae: mutagenesis of the potential glycosylation sites. EMBO J 1986; 5:1193-1197. 83. Molowa D. he State of Biologic Manufacturing, Industry Analysis 2001. 84. Fersko R, Latei N. Biologics move of patent but new paradigms are unlikely to emerge in the absence of public or private economic incentives. Biolaw and Business 2004.

Disulide Bond Formation in Bacteria-Based Cell-Free Protein Expression

75

85. Jewett M, Swartz J. Substrate replenishment extends protein synthesis with an in vitro translation system designed to mimic the cytoplasm. Biotechnol Bioeng 2004; 87:465-472. 86. Bessette P, Qiu J, Bardwell J et al. Efect of sequences of the active-site dipeptides of DsbA and DsbC on in vivo folding of multidisulide proteins in Escherichia coli. J Bacteriol 2001; 183:980-988. 87. McFall E, Newman E. In Escherichia coli and Salmonella: Cellular and molecular biology (eds. Neidhardt F.C. et al) 358-379 (ASM Press, Washington DC, 1996). 88. Shaibe E, Metzer E, Halpern Y. Metabolic pathway for the utilization of L-arginine, L-ornithine, agmatine, and putrescine as nitrogen sources in Escherichia coli K-12. J Bacteriol 1985; 163:933-937. 89. Wu W, Morris D. Biosynthetic arginine decarboxylase from Escherichia coli. Puriication and properties. J Biol Chem 1973; 248:1687-1695. 90. Yanofsky C. Transcription attenuation: Once viewed as a novel regulatory strategy. J Bacteriol 2000; 182:1-8. 91. Newton W, Snell E. Catalytic Properties of Tryptophanase, a Multifunctional Pyridoxal Phosphate Enzyme. Proc Natl Acad Sci USA 1964; 51:382-389. 92. Su H, Newman E. A Novel L-Serine Deaminase Activity in Escherichia-Coli K-12. J Bacteriol 1991; 173:2473-2480. 93. Datsenko K, Wanner B. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci USA 2000; 97:6640-6645. 94. Zawada J, Richter B, Huang E et al. High-density, deined media culture for the production of Escherichia coli cell extracts (ed. B, S.) (ACS Press, Washington, DC, 2003). 95. Cosloy S, McFall E. L-Serine-sensitive mutants of Escherichia coli K-12. J Bacteriol 1970; 103:840-841. 96. Yang J, Kanter G, Voloshin A et al. Expression of Active Murine Granulocyte-Macrophage ColonyStimulating Factor in an Escherichia coli. Cell-Free System. Biotechnol Prog 2004; 20:1689-1696. 97. Goerke A, Swartz J. Development of Cell-Free Protein Synthesis Platforms for Production of Disulide Bonded Proteins. Being Submitted 2006; TBD. 98. Meeker T, Lowder J, Cleary M et al. Emergence of idiotype variants during treatment of B-cell lymphoma with anti-idiotype antibodies. N Engl J Med 1985; 312:1658-1665. 99. Cleary M, Meeker T, Levy S et al. Clustering of extensive somatic mutations in the variable region of an immunoglobulin heavy chain gene from a human B cell lymphoma. Cell 1986; 44:97-106. 100. Levy S, Mendel E, Kon S et al. Mutational hot spots in Ig V region genes of human follicular lymphomas. J Exp Med 1988; 168:475-489. 101. Jandik P, Cheng J, Evrovski J et al. Simultaneous analysis of homocysteine and methionine in plasma. J Chromatogr B Biomed Sci Appl 2001; 759:145-151. 102. Knapp K, Swartz J. Evidence for an Additional Disulide Reduction Pathway in Escherichia coli. To be Submitted 2006; TBD:TBD. 103. Knapp K, Goerke A, Swartz J. Cell-free Synthesis of Proteins that Require Disulide Bonds Using Glucose as an Energy Source. Biotechnol Bioeng. 2006; TBD:Accepted. 104. Kanter G, Yang J, Voloshin A et al. Cell-free production of scFv fusion proteins: An efective and eicient approach for custom lymphoma vaccines. TBD 2006; In preparation.

Chapter 6

he PURE System: A Minimal Cell-Free Translation System Bei-Wen Ying, Yoshihiro Shimizu and Takuya Ueda*

Abstract

A

reconstituted protein synthesis system, comprising a minimal set of components for the production of active proteins, is described. he system consists of 32 puriied soluble factors from Escherichia coli together with other biochemical components, and is designated as the PURE (Protein synthesis Using Recombinant Elements) system. he PURE system can be conveniently used for both the production of functional target proteins and for investigation of protein translation and maturation processes. he synthesis of various proteins, including model proteins, single-chain antibodies, multiple-domain proteins, etc., is examined. Moreover, biochemical applications for protein folding, particularly chaperone-mediated processes are further discussed.

Introduction

In the post-genome era, a simple and rapid system for the expression of any gene or ORF is essential. To meet this demand, the development and application of many in vitro protein biosynthesis systems has been attempted. These studies have led to the widespread use of cell-free translation systems as a powerful tool in the ield of protein science.1,2 Cell-free translation systems are typically based on crude-cell extracts comprising all of the necessary components, such as the E. coli S30 fraction,3-5 wheat germ extract,6,7 and rabbit reticulocyte lysate.8 Using these systems, screening techniques with great potential have been developed, such as ribosome display9,10 and mRNA display.11,12 The application of cell-free translation systems to structural proteomics is also becoming more efective, considering the number of protein sequences with unknown functions that are continuously being discovered in various genome sequence projects. However, the rapid depletion of the energy source and the degradation of both proteins and nucleic acids are serious problems in batch- based systems, leading to a bottleneck for further development. To overcome these problems, we have developed an optimized cell-free translation system, reconstructed from a minimum set of puriied factors and enzymes required for translation, and designated it as the PURE (Protein synthesis Using Recombinant Elements) system (Fig. 1). The PURE approach resolves problems related to protein generation and maturation, and is therefore considered to be a suitable system for various types of research. One of the central concerns in biotechnological applications of cell-free translation is correct protein folding. During both co and post-translational folding, molecular chaperones, such as heat shock proteins, block pathways to aggregation, and catalysts, such as prolyl cis-trans isomerases and protein disulide isomerases, catalyze individual folding steps. Further studies on protein folding and maturation are becoming crucial for cell-free translation. Meanwhile, proper folding in cell-free translation systems results in the eicient production of functional proteins in vitro, which may *Corresponding Author: Takuya Ueda—Department of Medical Genome Sciences, Graduate School of Frontier Sciences, The University of Tokyo, FSB-401, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8562, Japan. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

he PURE System

77

Figure 1. Construction of the PURE system. EF, elongation factor; IF, initiation factor; RF, release factor; RRF, ribosomal recycling factor.

then be rapidly used in therapeutics and/or crystallography. herefore, in vitro translation, when carried out under optimized conditions for protein folding, is an extremely powerful tool for the resolution of diverse scientiic questions and for the development of new and more eicient methods in protein science.

Construction of the PURE System for Protein Generation Composition of the PURE System

he translation process is comprised of three steps: initiation, elongation, and termination. In E. coli the translation factors responsible for these steps comprise three initiation factors (IF1, IF2, and IF3), three elongation factors (EF-G, EF-Tu, and EF-Ts), and three release factors (RF1, RF2, and RF3), as well as RRF for the recycling of ribosomes. In addition, gene transcription requires a corresponding RNA polymerase. hus, 32 essential elements, IF1, IF2, IF3, EF-G, EF-Tu, EF-Ts, RF1, RF3, RRF, 20 aminoacyl-tRNA synthetases (ARSs), methionyl-tRNA transformylase (MTF), T7 RNA polymerase, and ribosomes, were puriied individually to establish a minimal in vitro translation system, termed the PURE system. The system also includes other essential components: 46 tRNAs, NTPs, creatine phosphate, 10-formyl-5,6,7,8-tetrahydrofolic acid, 20 amino acids, creatine kinase, myokinase, nucleoside-diphosphate kinase, and pyrophosphatase (see Table 1 for detailed information). he protein factors (IF1, IF2, IF3, EF-G, EF-Tu, EF-Ts, RF1, RF3, RRF, 20 ARSs, MTF, and T7 RNA polymerase) were puriied from E. coli cells over-expressing these factors. he genes were cloned as His-tagged fusions for easy and efective puriication by Ni2+ column chroamtography. All of the puriied factors and enzymes were veriied to be biologically active, in spite of the fused His-tag. Using these puriied components and an energy recycling system, an active transcription/ translation-coupled PURE system was successfully constructed.

Productivity of the PURE System

he PURE system was capable of synthesizing full-length polypeptides of dihydrofolate reductase (DHFR), lysozyme, green luorescent protein (GFP), and bacteriophage T7 gene 10

78

Cell-Free Protein Expression

Table 1. Composition of the PURE system Concentrations of All Components in the PURE System E. coli tRNA mix Magnesium acetate Hepes-KOH (pH 7.6) Potassium acetate Potassium glutamate Spermidine Creatine phosphate 20 Amino acids Ribosome T7 RNA polymerase FD DTT GTP UTP CTP ATP

10A 280/ml 9 mM 50 mM 18.7 mM 157 mM 2 mM 20 mM 0.1 mM each 1 μM 10 μg/ml 10 μg/ml 1 mM 2 mM 1 mM 1 mM 2 mM

EF-TS EF-Tu EF-G RF-1 RRF RF-3 IF-1 IF-2 IF-3 PPiase CK Myokinase MTF NDK LysRS PheRS HisRS

0.66 μM 0.92 μM 0.26 μM 0.25 μM 0.5 μM 0.17 μM 10 μg/ml 40 μg/ml 1.5 μg/ml 0.1 μg/ml 4 μg/ml 3 μg/ml 20 μg/ml 1.08 μg/ml 6.4 μg/ml 16.52 μg/ml 0.8 μg/ml

MetRS AlaRS lleRS AspRS FluRS GlyRS ProRS ThrRS ArgRS CysRS GlnRS AsnRS TrpRS LeuRS SerRS ValRS TyrRS

2.08 μg/ml 68.79 μg/ml 39.53 μg/ml 7.97 μg/ml 112.63 μg/ml 9.6 μg/ml 10.24 μg/ml 6.29 μg/ml 2 μg/ml 1.23 μg/ml 3.79 μg/ml 22 μg/ml 1.05 μg/ml 4.02 μg/ml 1.87 μg/ml 1.81 μg/ml 0.61 μg/ml

CK, creatine kinase; DTT, dithiothreitol; FD, 10-formyl-5,6,7,8-tetrahydrofolic acid; MTF, methionyl-tRNA transformylase; NDK, nucleoside diphosphate kinase; PPiase, pyrophosphatase.

product, with a productivity of more than 100 μg/ml/hr reaction. The biological activities of these proteins were evaluated, to prove that they were all in their active conformations. For example, the speciic activity of the synthesized DHFR was similar to that of over-expressed DHRF obtained from cells.13 hus, the PURE system contained all of the components essential for translation, and these components were both necessary and suicient for producing active proteins. hese results also showed that the PURE system could productively generate active proteins derived from a diverse range of other species. In this post-genome era, there is a clear need for the high-throughput synthesis of target proteins on a library scale. he production of 18 target proteins synthesized using the PURE system, including 16 genes newly cloned from the E. coli genome, was examined. As shown in Figure 2, all of them were produced in high amounts compared to the model protein DHFR. In addition, most of the proteins had relatively high solubility (data not shown), indicating the potential of the PURE system for the production of genome-wide proteins for proteomic research.

Advantages of the PURE System

Since the PURE system is constructed using puriied elements, there are few RNases or proteases in the system. he stability of both mRNAs and proteins in this system is greatly increased compared to cell-extract systems. he PURE system can produce target proteins directly from PCR-ampliied linear DNA fragments, without the need for cloning. hus, it is considered to be an excellent tool for in vitro screening techniques, such as ribosome display,14,15 mRNA display, and other selection approaches.16 he omission or addition of factors is both feasible and easy to perform in the PURE system, ofering researchers a highly controllable system. It provides considerable advantages for the study of translation mechanisms and protein maturation processes, such as translation initiation and/or termination mechanisms,17,18 trans-translation processes,19,20 and chaperone-mediated protein-folding pathways,21,22 as well as application development, such as the incorporation of unnatural amino acids.13

he PURE System

79

Figure 2. Productivity of protein synthesis. Target proteins were synthesized using the PURE system at 37˚C for 2 hr. The productivity represents the total amount of the synthesized product in each 20-μl reaction. Enolase, 2-phosphoglycerate dehydratase; DCEA, glutamate decarboxylase; GatD, galactitol-1-phosphate 5-dehydrogenase; TDH, threonine-3-dehydrogenase; MetF, 5,10-methylenetetrahydrofolate reductase; MetK, S-adenosylmethionine synthetase; GatY, D-tagatose-1,6-bisphosphate aldolase; DapA, dihyodipicolinate synthetase; TPIS, triosephosphate isomerase; ThiD, phosphomethylpyrimidine kinase; Upp, uracil phosphoribosyltransferase; RpoA, DNA-directed RNA polymerase alpha; MBP5, maltose-binding protein (double mutant); MDH, malate dehydrogenase; G6PDH, glucose-6-phosphate dehydrogenase; LDH, D-lactate dehydrogenase (ldha) DLD, D-lactate dehydrogenase, DHFR, dihydrofolate reductase.

he biggest advantage of the PURE system is that the target protein is expressed and puriied in its native form and without any tags, within just a few hours. Since most components of the PURE system exist in His-tagged forms, they can be eliminated by passage through a Ni2+ column. Ribosomes can be removed by ultrailtration using a membrane with a molecular weight cutof of 100 kDa. Highly puriied DHFR was obtained expeditiously, as described previously.13 Such rapid and simple puriication procedures for preparing proteins will enable high-throughput protein preparation and accelerate structural determination of proteins by NMR and X-ray analyses.

Reconstitution of the PURE System for Protein Maturation Studies Functional Production of Single-Chain Antibodies

Compared to conventional in vivo expression techniques, cell-free translation technology has been utilized extensively,1 due to its convenience and low cost. However, cell-free systems, as well as in vivo expression systems, occasionally result in aggregation during protein generation. Nevertheless, relatively high product solubility is observed in the PURE system as compared to the S30 extract system. As shown in Figure 3, two single-chain antibodies, anti-BSA scFv and HyHEL10 scFv, were successfully synthesized using either the PURE system or E. coli S30 extracts. he relative amounts of soluble products were signiicantly higher in the PURE system than that in the S30 extract system; whereas the total amount of product was higher with the S30 extract system. Although it has been generally considered that intrinsic chaperones in the cell extracts could assist protein folding, the results shown here indicated that the crowded and condensed environment of the extract system caused aggregation and hindered the folding

80

Cell-Free Protein Expression

Figure 3. Detection of single-chain antibodies synthesized in vitro. Anti-BSA scFv and HyHEL10 scFv were synthesized using the PURE system and the E. coli S30 extract system. The pellet (P) and the supernatant (S) of the translation (37˚C, 1 hr) products were detected by western blotting, as described previously.22 M, molecular weight (3 bands stand for 49.1 KDa, 34.8 KDa and 28.9 KDa, respectively).

process. We hypothesized that the selective or optional addition of folding helpers to the PURE system would enhance functional protein production. A subset of nascent polypeptides requires corresponding helper proteins,23 such as molecular chaperones and/or isomerases, etc., to produce active protein conformations in cells. he requirements for the representative prokaryotic chaperones, DnaK-DnaJ-GrpE (KJE), trigger factor (TF), GroEL and GroES (GroEL/ES), and HSP104 (ClpB), were examined in the PURE system to develop a production system for biologically active single-chain antibodies and to clarify folding mechanisms. he chaperones were added, either simultaneously or separately, to the chaperone-free PURE system, resulting in a combined translation-folding system. he scFv antibodies were translated in the presence or absence of these chaperones at 37˚C for 1 hr, followed by centrifugation at 14,000 rpm for 10 min. he relative solubility was calculated from [35S]-methionine incorporation into the translation products. he results showed that a single addition of either KJE or TF greatly increased the soluble amounts of both scFv antibodies; whereas the addition of ClpB alone hardly suppressed the aggregation of either of the newly synthesized scFvs (Fig. 4). he addition of multiple chaperones, for example, TF and KJE, increased the amount of soluble scFv antibody more than either chaperone alone (Fig. 4). In conclusion, the eicient generation of active protein was improved by the addition of most chaperones, but the simultaneous addition of all types of chaperones appeared to be unnecessary. In particular, in the case of scFv antibodies, supplementation with KJE and TF signiicantly increased the amount of soluble protein (Fig. 4) and biologically active antibody.22

Precise Investigation of Protein-Folding Mechanisms

In addition to the productive generation of polypeptides, the PURE system is appropriate for elucidating protein-maturation pathways. Since the PURE system is a reconstituted cell-free translation system containing no intrinsic chaperones or enzymes, the simple addition of exogenous factors leads to precise and conclusive veriication of the efectiveness and function of these factors during the folding process.

81

he PURE System

Figure 4. Solubility of the single-chain antibodies generated using the chaperone-supplemented PURE system. Anti-BSA scFv and HyHEL10 scFv were each translated at 37˚C for 1 hr, in the presence or absence of individual or multiple chaperones. The relative amounts of soluble protein are indicated. KJE, DnaK-DnaJ- GrpE system; GroEL/ES, GroEL- GroES system; TF, trigger factor; ClpB, heat-shock protein 104.

In order to illustrate the correlation between individual chaperone systems and their substrate proteins in cells, the candidate substrate proteins in Figure 2 were synthesized using the PURE system in the presence or absence of various chaperone systems. he inluence of the major chaperone systems, DnaK and GroEL, on the solubility of these newly synthesized proteins was evaluated. he results showed that the addition of these chaperones increased the proportion of soluble protein products and that each chaperone system was individually responsible for particular substrate proteins (Table 2). For example, the single addition of GroEL/ES was suicient for maximal Table 2. Chaperone dependency of protein folding

Substrates

DnaK System

Chaperone Dependency GroEL/ES

Both Systems

DCEA GatD THD MetF MetK GatY DapA TPIS ThiD MBP5 GFP Rhodanese DHFR

+ +/+ + ++ ++ + +/+ + + -

+ +/++ +++ + +++ +/+/+/+ + -

+ +/+ +++ +++ ++ +++ +/+ + + + -

The protein substrates were synthesized in the presence of DnaK-DnaJ-GrpE (DnaK system) and/or the GroEL-GroES system (GroEL/ES). Protein solubility resulting from the chaperone-supplemented systems was compared to that from the chaperone-free PURE system. +++, ++, +, and - refer to high, medium, low, and no dependency, respectively. GFP, green fluorescent protein.

82

Cell-Free Protein Expression

solubility of the model proteins, MetK and DapA, but was not suicient for the substrate MetF. In the latter case, it was necessary to additionally supplement the assays with the DnaK system in order to prevent the aggregation of newly synthesized MetF. Co and/or post-translational folding processes were studied using this translation-folding-c oupled system. In contrast to test-tube experiments, ex vivo methods, such as pull-down assays,24 can reveal chaperone-executed events inside cells. However, these approaches are inadequate for distinguishing co and post-translational chaperone-dependent folding. With the PURE system, simply arresting in vitro translation with antibiotics or other chemicals, such as EDTA, can be used to determine whether chaperones recognize their substrates during or ater translation. Using this strategy, the co and/or post-translational actions of chaperones on newly synthesized polypeptides were evaluated precisely using the PURE system, leading to a new understanding of chaperone function.22 In addition, the elimination of release factors from the PURE system resulted in the efective preparation of the translation complex, mRNA-ribosome-polypeptide, which has been employed to detect the co-translational association of chaperones.21 he recent observation of the possible co-translational involvement of GroEL in the folding of its stringent substrate MetK was due the ability to add and omit components of this reconstituted translation system.21 In addition to the translation/folding system, a combined translation and membrane integration/translocation system was constructed, using inverted membrane vesicles (INVs) and other related puriied proteins, to produce soluble pre secretory and integral membrane proteins. he eicient membrane translocation of the pre secretory protein, pOmpA, and the integration of the inner-membrane protein MtlA into INVs were achieved.25

Concluding Remarks

he eicient and rapid expression of functional proteins is required for the latest post-genomic research and emerging high-throughput screening strategies. A genuine cell-free gene-expression system, known as the PURE system, was successfully constructed in vitro with puriied soluble protein factors. he productivity and activity of the PURE system was demonstrated. The absence of all endogenous chaperones provides an ideal condition for identifying protein substrates that strictly require chaperones (chaperone-dependency) and for elucidating the individual steps of protein folding that are controlled by chaperones. he highly controllable translation/folding and translation/secretion systems were reconstituted and used to determine the precise role of each chaperone in co and/or post-translational folding processes. Global analysis and the creation of a protein databank using this system are expected to greatly advance proteomic research. In addition, the PURE system is a powerful tool for the emerging ields of constructive biology and synthetic biology.

References

1. Jermutus L, Ryabova LA, Pluckthun A. Recent advances in producing and selecting functional proteins by using cell-free translation. Curr Opin Biotechnol 1998; 9:534-548. 2. Swartz JR. Advances in Escherichia coli production of therapeutic proteins. Curr Opin Biotechnol 2001; 12:195-201. 3. Ryabova LA, Morozov I, Spirin AS. Continuous-low cell-free translation, transcription-translation and replication-translation systems. Methods Mol Biol 1998; 77:179-193. 4. Kudlicki W, Kramer G, Hardesty B. High eiciency cell-free synthesis of proteins: reinement of the coupled transcription/translation system. Anal Biochem 1992; 206:389-393. 5. Spirin AS, Baranov VI, Ryabova LA et al. A continuous cell-free translation system capable of producing polypeptides in high yield. Science 1988; 242:1162-1164. 6. Endo Y, Sawasaki T. Advances in genome-wide protein expression using the wheat germ cell-free system. Methods Mol Biol 2005; 310:145-167. 7. Kawasaki T, Gouda MD, Sawasaki T et al. Eicient synthesis of a disulide-containing protein through a batch cell-free system from wheat germ. Eur J Biochem 2003; 270:4780-4786. 8. Beckler GS, hompson D, Van Oosbree T. In vitro translation using rabbit reticulocyte lysate. Methods Mol Biol 1995; 37:215-232. 9. Schaitzel C, Hanes J, Jermutus L et al. Ribosome display: an in vitro method for selection and evolution of antibodies from libraries. J Immunol Methods 1999; 231:119-135.

he PURE System

83

10. Hanes J, Pluckthun A. In vitro selection and evolution of functional proteins by using ribosome display. Proc Natl Acad Sci USA 1997; 94:4937-4942. 11. Nemoto N, Miyamoto-Sato E, Husimi Y et al. In vitro virus: bonding of mRNA bearing puromycin at the 3’-terminal end to the C-terminal end of its encoded protein on the ribosome in vitro. FEBS Lett 1997; 414:405-408. 12. Lipovsek D, Pluckthun A. In-vitro protein evolution by ribosome display and mRNA display. J Immunol Methods 2004; 290:51-67. 13. Shimizu Y, Inoue A, Tomari, Y et al. Cell-free translation reconstituted with puriied components. Nat Biotechnol 2001; 19:751-755. 14. Villemagne D, Jackson R, Douthwaite JA. Highly eicient ribosome display selection by use of puriied components for in vitro translation. J Immunol Methods 2006; 313:140-148. 15. Ogawa A, Sando S, Aoyama Y. Termination-free prokaryotic protein translation by using anticodonadjusted E. coli tRNASer as uniied suppressors of the UAA/UGA/UAG stop codons. Read-through ribosome display of full-length DHFR with translated UTR as a buried spacer arm. Chembiochem 2006; 7:249-252. 16. Ying BW, Suzuki T, Shimizu Y et al. A novel screening system for self-mRNA targeting proteins. J Biochem (Tokyo) 2003; 133:485-491. 17. Umekage S, Ueda T. Spermidine inhibits transient and stable ribosome subunit dissociation. FEBS Lett 2006; 580:1222-1226. 18. Udagawa T, Shimizu Y, Ueda T. Evidence for the translation initiation of leaderless mRNAs by the intact 70 S ribosome without its dissociation into subunits in eubacteria. J Biol Chem 2004; 279:8539-8546. 19. Shimizu Y, Ueda T. SmpB triggers GTP hydrolysis of EF-Tu on ribosome by compensating for the lack of codon-anticodon interaction during trans-translation initiation. J Biol Chem 2006; in press. 20. Shimizu Y, Ueda T. he role of SmpB protein in trans-translation. FEBS Lett 2002; 514:74-77. 21. Ying BW, Taguchi H, Kondo M et al. Co translational involvement of the chaperonin GroEL in the folding of newly translated polypeptides. J Biol Chem 2005; 280:12035-12040. 22. Ying BW, Taguchi H, Ueda H et al. Chaperone-assisted folding of a single-chain antibody in a reconstituted translation system. Biochem Biophys Res Commun 2004; 320:1359-1364. 23. Ellis RJ, Hartl FU. Principles of protein folding in the cellular environment. Curr Opin Struct Biol 1999; 9:102-110. 24. Ewalt KL, Hendrick JP, Houry WA et al. In vivo observation of polypeptide lux through the bacterial chaperonin system. Cell 1997; 90:491-500. 25. Kuruma Y, Nishiyama K, Shimizu Y et al. Development of a minimal cell-free translation system for the synthesis of presecretory and integral membrane proteins. Biotechnol Prog 2005; 21:1243-1251.

Chapter 7

Cell-Free Expression Approaches for the Production and Characterization of Membrane Proteins Daniel Schwarz, Christian Klammt, Alexander Koglin, Florian Durst, Frank Löhr, Volker Dötsch and Frank Bernhard*

Abstract

M

embrane proteins still represent one of the major challenges of structural biology. he preparation of high quality samples of functionally folded protein is usually the major bottleneck that restricts further approaches. Cell-free expression has emerged in recent times as a promising tool for the fast production of membrane proteins at high levels and data from a variety of diferent systems are rapidly accumulating. In this chapter we review the current knowledge on the cell-free expression of membrane proteins in preparative scales. We summarize data on the functional and structural evaluation of cell-free produced membrane proteins as well as on the improvement of sample quality. Already established techniques like the cell-free expression of membrane proteins into detergent micelles will be presented in detail. In addition, we will comment on current developments like the involvement of liposomes and throughput applications. Potential synergies of cell-free expression and structural techniques will further be discussed with focus on structural approaches of cell-free produced transporters by NMR spectroscopy.

Introduction

Membrane proteins (MPs) comprise 20-40% of cellular proteomes but their molecular analysis lacks far behind to that of cytosolic soluble proteins. Main reasons for the strongly biased availability of structural data are generally tremendous diiculties in the preparative scale synthesis and preparation of MPs. heir strong hydrophobic nature, the tight association with lipids and localisation in membranes and their requirement for specialized targeting pathways involving complex signal and translocon systems prevented so far in most approaches an eicient sample production. MPs belong therefore to the most diicult proteins to be studied. he optimization of production protocols in conventional expression systems based on bacterial, yeast or higher eukaryotic cells is extremely time consuming and can take even years. Moreover, expression rates obtained with the distinct strategies difer considerably even upon employment of closely related MPs. Conventional approaches for the in vivo production of new MP targets are therefore always highly critical and nonpredictable.1-3 Signiicant optimizations of reaction design and protocols helped to develop the cell-free (CF) expression technique to a powerful systems and the continuous exchange cell-free expression *Corresponding Author: Frank Bernhard—Centre for Biomolecular Magnetic Resonance, University of Frankfurt/Main, Institute for Biophysical Chemistry, Max-von-Laue-Str. 9, D-60438 Frankfurt/Main, Germany. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

Cell-Free Expression Approaches for the Production and Characterization of Membrane Proteins

85

(CECF) mode enables now the production of mg amounts of protein in less than 12 hrs.4-6 In addition, the productivity of CF batch systems is continuously being improved by the modiication of protocols and this less complicated reaction design might become attractive for the preparative scale protein production in the next future.7-11 CF expression techniques can be established in most biochemical labs within a few days. he preparative scale CF expression of proteins is based either on extracts of E. coli or of wheat germ embryos. he extract quality is crucial for the productivity of the system and various common lab strains like BL21 derivatives or RNAse deicient strains like A19 or D10 can be used as source for bacterial extracts.12-14 A variety of protocols for the preparation of E. coli extracts have been published.13-18 An important step is the elimination of endogenous mRNA by either a “run-of ” translation procedure or by high salt incubation at elevated temperatures.15-17 In addition, all cellular amino acids are removed by extensive dialysis. hese procedures ensure background elimination in subsequent expression approaches as well as in labelling applications. CF expression techniques have emerged in recent times to one of the most promising tools for the production of diicult proteins. Even strong toxins could be synthesized in considerable amounts.19 In particular, the fast synthesis of high amounts of MPs provides a fascinating perspective. Successful reports on the CF production of MPs in preparative scales include already a comprehensive variety of prokaryotic and eukaryotic α-helical and β-barrel type proteins.16,20-26 Detailed individual reaction protocols as well as complete commercial systems that can result in the high-level expression of MPs are available.12,16,20-24 CF expression as a tool of primary choice for the fast generation of preparative amounts of high quality MP samples is therefore strongly supported by continuously accumulating data.

Why Is Cell-Free Expression Exceptionally Suitable for the High-Level Production of Membrane Proteins?

In conventional cellular expression a variety of intrinsic characteristics upon synthesis of MPs have to be considered. In contrast to soluble proteins that are released from the ribosome into the cytoplasm, MPs usually have to be synthesized into the two-dimensional space of cellular membranes like the plasma membrane in prokaryotes or the system of the endoplasmic reticulum in eukaryotes. he initial targeting of the nascent polypeptide chains of MPs into those membranes usually depends on the sophisticated interplay between signal sequences and their cognate recognition systems. In eukaryotes, the posttranslational modiication of MPs is mostly essential for the directed transportation through the inner membrane system to inal destinations in the plasma membrane or in membranes of organelles. Frequent modiications are speciic glycosylation patterns, phosphorylation, extensive disulide bridge formation and the attachment of lipid chains that act as membrane anchors. Speciic glycosylation patterns are in particular important as sorting and traicking signals. Overexpression of heterologous MPs in foreign host backgrounds oten causes ineicient targeting and results in nonnative or incomplete glycosylation patterns due to incompatibilities of the recombinant MP with the host speciic modiication and transportation systems. Mis-targeting will be the consequence most likely followed by degradation of the proteins that might become trapped somewhere in the inner membrane system. Even if a recombinant MP inally becomes inserted into the cytoplasmic membrane, the lack of glycosylation can cause high instability and a dramatically reduced lifetime. In the rare cases, where recombinant MPs succeed to become eiciently inserted into the plasma membrane, subsequent toxic efects due to pore or channel forming activities, simple overcrowding of the limited membrane space, or by overloading central translocon systems may afect the viability of the host cells. A special expression strategy commonly employed for the production of β-barrel-type MPs is their production as insoluble inclusion bodies in E. coli. However, eicient refolding procedures could be established only for a small number of proteins and most MPs do not even form inclusion bodies. As a result of these speciic problems, the successful overproduction of MPs in living cell systems is still the exception and structural data of MPs are limited to only few prominent examples.27

86

Cell-Free Protein Expression

he exceptional high potential of CF systems for the ield of membrane research is primarily based on the elimination of these general problems during MP production. Toxic efects caused by the recombinant MPs do not play any roles. In addition, native membranes are almost completely removed in the expression system and they can be replaced by artiicial hydrophobic compartments like detergent micelles. Successful overexpression needs therefore not longer to rely on the presence of eicient translocon systems and of compatible targeting signals. It is further important to note that proteolytic activities in CF systems are largely reduced. Cells are harvested for extract preparation in mid-log growth phase when proteases are generally low abundant. Most membrane associated proteases or proteasome complexes are eliminated ater extract preparation and residual proteolysis due to soluble proteases can easily be controlled by the addition of broad range protease inhibitors. In addition, a variety of further unique and valuable intrinsic properties of CF systems contribute to their success in MP production. (1) he speediness of CF MP production is exceptional and mg amounts of puriied protein can generally be obtained in less than 24 hrs (Fig. 1). his dramatically increases the screening throughput of expression parameters like concentration and supply of individual reaction compounds. (2) he CF expression system does not require special equipments and can be operated in standard temperature controlled shakers in virtually most biochemical labs without the need for expensive set-up investments. Comprehensive preparation protocols for the preparation of CF extracts and for the isolation and puriication of T7-RNA polymerase are available.13,15,17 Eicient individual reaction protocols have furthermore been published in detail and could be established within a few weeks.12,16 In addition, no organisms with potential biohazards are involved. (3) he low volumes of CF reactions are convenient to handle and a 1 ml reaction can already represent the preparative scale set-up. No large fermenters have to be operated and many problems associated with the long term incubation of larger cell volumes like maintenance of sterile conditions or cell viability are eliminated. In addition, optimized analytical reactions in the microlitre scale can be scaled up in a linear ratio to preparative scale reactions covering a few millilitres. (4) Template DNA remains stable during CF expression and is not subject of degradation or mutagenesis as frequently observed in permanent expressing cell lines. Tight repression and eicient induction mechanisms as well as stringent selection conditions are furthermore not required. Background expression due to leaky promoters is not a problem in CF systems as protein synthesis can simply be started by addition of the template DNA. (5) CF systems are predestined for the production of multimeric MPs, either by coexpression of diferent templates or by addition of already puriied subunits to a newly synthesized protein. Plasmid incompatibility, eicient replication and vector partition are not longer of relevance. In addition, the stochiometry of the MP multimers during synthesis can be modulated by variation of the supplied template concentrations. (6) CF systems provide considerably extended options for expression screens. he reduced complexity and the easy accessibility make it to an ideal system for systematic functional expression studies. Physiologies of host cells, growth requirements or strain variations need not to be considered. (7) CF expression of MPs is a fast and straight forward strategy (Fig. 1). A number of commonly employed and highly critical puriication steps are eliminated. Cells need not to be disrupted and changes in pH or contacts with lysosomal enzymes are avoided. A very critical step in conventional puriication protocols is the extraction of the overproduced MPs out of cellular membranes. It oten results in large losses of yields or even in complete MP denaturation. his problem is solved by using CF expression systems and the synthesized MPs in the RM can be directly loaded on chromatography columns for subsequent puriication. (8) No large fusion proteins have to be employed in order to achieve overproduction or solubilization of recombinant MPs in bacterial CF systems. While many MPs might become synthesized already without any modiication, the addition of small N-terminal tags like the 12 amino acid T7-tag might be beneicial for an optimized expression or for analysis by immunodetection. he addition of a poly(His)10-tag might further be helpful for the fast puriication of the MP.

Cell-Free Expression Approaches for the Production and Characterization of Membrane Proteins

87

Figure 1. Comparison of conventional cellular expression versus cell-free expression.

Cell-Free Produced Precipitates: A Fast and Reliable Approach for the Production of High Amounts of Pure Membrane Protein Samples

In CF systems, the mechanism of MP synthesis is basically identical to that of soluble proteins and both protein types are released directly into the reaction mixture already during translation. Certainly, as only residual hydrophobic environments may be present in standard CF reactions, the majority of the synthesized MPs deinitely will precipitate. In the irst instance, this expression mode might resemble the inclusion body formation in conventional cellular expression systems. However, there are several important diferences that have to be considered. Inclusion bodies contain completely unfolded protein that oten can only be solubilized in presence of strong denaturants like guandinium hydrochloride. he proteins might then become refolded by elaborated protocols involving usually slow bufer exchange in excessive dilutions. In contrast, CF produced MP precipitates already solubilize fast and eiciently upon suspension in bufers containing relatively

88

Cell-Free Protein Expression

Figure 2. Molecular structure of detergents suitable for cell-free expression. A, Brij-58; B, digitonin; C, LMPG.

mild detergents like n-dodecyl-β-D-maltoside (DDM) or 1-myristoyl-2-hydroxy-sn-glycero-3[phospho-rac-(1-glycerol)] (LMPG) (Fig. 2).16 LMPG was found to be exceptionally suitable for the solubilization of a wide variety of CF produced MP precipitates. Prewashing of the MP precipitate with speciic detergents or modulation of the incubation temperature upon solubilization can result in samples of high purity within a few hours ater the CF expression reaction has been inished.16 his fast and relatively simple procedure already has the potential to yield functionally folded MPs as demonstrated with the multidrug transporter EmrE.16,21 However, the functional resolubilization out of CF generated precipitate is clearly not a general mechanism and has to be analysed for each new MP target. It was not possible to solubilize the bacterial nucleoside transporter Tsx out of CF produced precipitates in a functional form and we made similar observations with a variety of precipitates formed of eukaryotic G-protein coupled receptors (unpublished data).28 In our opinion, the solubilization of CF produced MP precipitates still bears a considerable potential by various modiications of the procedure. Attempts analysing the refolding of MP precipitates might also be interesting but are not documented so far. It should inally be highlighted that inclusion body formation of MPs in cellular expression systems occurs only in few cases that oten deserve special vector design and expression conditions and it is oten subject of trial and error. he formation of precipitates upon CF expression however is always the clear consequence if not enough hydrophobic compartments are present. In combination with the general suppression of proteolysis in CF systems as discussed above, it is therefore a reliable expression mode for the fast generation of MP samples suitable for a variety of analysis like antibody production. Although functional folding of the MP might not be obtained instantly, it deinitely represents an interesting approach especially if eicient refolding protocols are available or about to be established.

Cell-Free Expression into Detergent Micelles: A Completely New and Unique Mode of Membrane Protein Production

Besides the special case of inclusion body formation, in conventional in vivo systems the production of soluble MPs into membranes is usually anticipated. In CF systems, almost all natural membranes are removed, but the open nature of the reaction allows the addition of artiicial compounds like detergents. In this context it is important to realize that CF systems not only principally ofer the production of MPs that otherwise cannot be obtained; they moreover allow the synthesis of MPs in a completely new expression mode (Fig. 3). Supplied detergents are able to interact with MPs already during translation and thus prevent any inter- or intramolecular aggregation that is followed by precipitation. his mode of MP synthesis directly into detergent

Cell-Free Expression Approaches for the Production and Characterization of Membrane Proteins

89

Figure 3. Cell-free production of membrane proteins in presence of detergents. Schematic illustration of a new expression mode.

micelles is unique for CF systems and not possible with any other of the conventional in vivo expression systems. What kinds of detergents are suitable for this new expression mode and in which concentrations they should be supplied? hese important questions have been addressed by systematic evaluations of comprehensive varieties of detergent types for their suitability to become included into CF reactions.23,28 Fortunately, only few detergents like n-octyl-β-glucopyranoside (β-OG), dodecyl-phosphocholine (DPC) or sodium cholate appeared to be harmful for the CF system and abolished any protein synthesis already at low concentrations that hardly exceeded the critical micellar concentrations (CMC). Solubilization and inactivation of proteins essential for the transcription/translation process in the CF extract most likely account for this efect. Many other detergents comprising the bi-chain-phosphocholines 1, 2-dioctanoyl-sn-glycero-3-phosphocholine (diC8PC) and 1, 2-diheptanoyl-sn-glycero-3-phosphocholine (DHPC), the alkyl-glucosides n-dodecyl-β-D-maltoside (DDM) and n-decyl-β-D-maltoside (DM), the alkyl-ether polyoxyethylene-sorbitan-monolaurate 20 (Tween 20) and the polyethylene-glycol derivatives polyethylene-glycol P-1, 1, 3, 3-tetramethyl-butylphenyl-ether (Triton X-100) and polyethylene-glycol 400 dodecyl-ether (hesit) are tolerated by CF expression systems in concentrations in between 2 times CMC and 15 times CMC.20,21,23,28 Most of these detergents are efective in the solubilization of some MPs like the α-helical transporter EmrE or the β-barrel type nucleoside transporter. However, they are obviously not useful for a more general application and they failed to solubilize CF synthesized larger α-helical eukaryotic MPs like members of the family of G-protein coupled receptors (GPCRs).28 One group of detergents deserves more special consideration. he family of polyoxyethylene-alkyl-ethers appears to be exceptionally suitable for the cotranslational solubilization of a wide range of structurally diferent protein.23,28 Besides bacterial α-helical and β-barrel type MPs, also larger eukaryotic MPs like GPCRs, aquaporins or ion transporters can be produced in the

90

Cell-Free Protein Expression

soluble CF expression mode in preparative amounts. hese detergents show a bipartite structure composed of an alkyl- and a polyoxyethylene moiety (Fig. 2). he particular structural details of the individual polyoxyethylene derivatives are important prerequisites that determine their properties in the solubilization of MPs. Detergents with extended polyoxyethylene chains containing >20 residues like polyoxyethylene-(23)-lauryl-ether (Brij-35), polyoxyethylene-(20)-cetyl-ether (Brij-58), polyoxyethylene-(20)-stearyl-ether (Brij-78) and polyoxyethylene-(20)-oleyl-ether (Brij-98) are highly eicient in MP solubilization, while in presence of short chain derivatives like polyoxyethylene-(10)-cetyl-ether (Brij-56) or polyoxyethylene-(10)-oleyl-ether (Brij-97) predominantly MP precipitates will be formed. he length of the alkyl moiety can have an additional remarkable efect. Larger chain length with C >12 were found to be much more efective in the solubilization of the vasopressin receptor, a member of the GPCR family.28 Final yields of up to 6 mg soluble receptor per one ml of CF reaction could be obtained in presence of Brij-58 (C16) or Brij-78 (C18) in contrast to only 1 mg of soluble receptor with Brij-35 (C12). Higher lexibility of the alkyl chain by exchange of the stearyl moiety in Brij-78 against an oleyl moiety in Brij-98 (C18-1) again reduced the eiciency of vasopressin receptor solubilization for 50%. A similar efect was observed for the soluble expression of the β-barrel type nucleoside transporter Tsx, while in that case also the shorter chain length in Brij-35 was suitable for high solubilization eiciencies. A diferent result was obtained upon solubilization of the multidrug transporter EmrE. In that case the short chain detergent Brij-35 and Brij-98 containing the nonsaturated oleyl moiety have been most efective for the CF production of soluble protein. Besides detergent type and structure, its inal concentration in the CF reaction is an important parameter. It is evident that any detergent will sooner or later become inhibitory to the transcription/translation mechanism above certain critical concentrations. he inal detergent concentration has therefore oten to be a compromise between eicient MP solubilization and least toxicity to the CF system. Some general guidelines about the critical threshold concentrations of the most commonly used detergents have been published.20,23,28 Detergent speciic characteristics like CMC and the resulting micellar concentration (Cmic) have to be considered. As prerequisite for the complete solubilization of a CF synthesized MP, the inal Cmic of a detergent should be at least equimolar to the MP concentration at the end of the CF reaction. Excessive MP might form undesired precipitates or heterogeneous micelles that could prevent further structural approaches. Certainly, both concentrations are diicult to determine and only vague estimates are generally available. In principle, the over-titration of the MP by supplying a molar excess of detergent micelles might be a reasonable strategy. Quantitative solubilization could be achieved and in addition, the homogeneity of the sample would be improved as multiple MP insertions into micelles could be avoided. Additional empty micelles that might be formed should not cause any problems. Unfortunately, many detergents are tolerated at only relatively low concentrations by the CF system and the amount to be added is therefore quite restricted. he limits for the detergents DM, DDM, digitonin, diC8PC, DHPC, Tween or Triton X-100 hardly exceed 10 × CMC. he maximal resulting Cmic can only be roughly estimated. In case of digitonin, we found concentration limits for CF reactions in between 3 mM and 4 mM inal concentration. he aggregation number, i.e., the number of detergent molecules forming a micelle, is not known for digitonin but it is generally assumed to be in between 50 and 100.29 In that case, we would approach a Cmic for digitonin somewhere between 30-100 μM. Upon CF expression of the vasopressin receptor, we can obtain levels of up to 6 mg per ml reaction, corresponding to approx. 100-120 μM. his calculation demonstrates that, at high level CF expression, the maximal possible concentration of many detergents might not be suicient for the complete or homogeneous solubilization of MPs. However, exceptions are several detergents from the Brij family. Brij-58, Brij-78 and Brij-98 are highly tolerated by the CF system in concentrations even exceeding 200 × CMC.28 CF expression of MPs in presence of increasing concentrations of these Brij derivatives shows a typical saturation efect. he amount of soluble produced Tsx transporter continuously increased and remained constant above a concentration level of approx. 50 × CMC. In that speciic case, this concentration most likely resembles now an equimolar ratio between micelles and protein. Further increases

Cell-Free Expression Approaches for the Production and Characterization of Membrane Proteins

91

even above 200 × CMC did not further change the amount of soluble produced Tsx and only empty micelles might be further formed. his “plateau” screening might be a reasonable approach to deine the optimal detergent concentration in speciic MP expression experiments. However, with some detergents it should be realized that due to low tolerances, the “plateau” expression level will not be reached. In summary, a variety of detergents can be used for the CF expression of a particular MP in the soluble mode. he eiciencies might be diferent, but the tolerated detergent concentrations are usually high enough to ensure the production of mg amounts of MP. However, what will be the impact of diferent detergent types on the quality of the synthesized MP? Structural efects upon MP insertion into micelles are still a largely mystic phenomenon. Diferent models are being discussed and MPs loosely swimming within spherical micelles as well as tightly packed protein/detergent complexes are being proposed. Whatever will be the case, one should be aware that any selected detergent can have a dramatic efect on the conformation and on the functional folding of a MP. Conductance measurements of the CF produced nucleoside transporter Tsx in planar membranes indicated a clear correlation between determined activity and the mode of expression.28 Tsx has to be synthesized in the soluble mode in presence of Triton X-100 in order to show highest pore forming activity. he activity of Tsx produced in presence of Brij derivatives was signiicantly lower and the protein was completely inactive when expressed as precipitate and solubilized in LMPG. We observed similar efects with the CF produced human endothelin B receptor whose functional folding is dependent on speciic expression conditions.30 In addition to the high-level MP production, the analysis of sample quality and functional folding of the synthesized MP should therefore be included in initial CF expression screens whenever possible.

Cell-Free Expression of Membrane Proteins into Preformed Artiicial Liposomes: A Challenging Perspective of Future Developments

MPs inserted into detergent micelles are known to be suitable for functional and structural analysis. However, lipid bilayers are the natural environment of MPs and the close association with distinct lipids oten causes subtle changes of protein conformations and modulates activities. In addition, certain functions like substrate transport or channel forming activities can only be analysed in lipids. Cellular membranes containing recombinant MPs are usually not suitable environments for further analyses as their compositions are nonpredictable and highly variable. In contrast, liposomes of deined lipid composition and containing only the desired MPs are generally required. Conventional preparation protocols for MPs include therefore the extraction of the recombinant MPs out of the membranes of the expression host by detergents and the subsequent reconstitution back into deined liposomes. As already discussed, this indispensable detour that usually has to be implemented into MP sample preparations harbours highly critical steps oten resulting in protein unfolding. In CF systems, the addition of lipids as preformed liposomes in analogy to the expression mode in presence of detergents is feasible. Supplementation of CF systems with lipids is not as critical as with detergents and they are tolerated at even high inal concentrations.16 However, a major problem will be that MP insertion into preformed liposomes is not as easy as the insertion into micelles and might require speciic signal or translocation systems. he eicient insertion of synthesized MPs into liposomes at preparative scales will thus be the major challenge. he transporter EmrE was synthesized in presence of 1, 2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) liposomes and probably inserted spontaneously.24 However, the lack of quantitative characterizations of the proteoliposomes and the involvement of detergents in subsequent puriication steps still leave uncertainties about the eiciencies of EmrE reconstitution. he production of MPs in mixtures of detergents and lipids might be a further perspective for the modulation of CF expression protocols. Increased yields of functionally folded mechanosensitive channel MscL were obtained in CF reactions with mixed micelles of Triton X-100 and E. coli lipids or in presence of liposomes prepared from E. coli lipid mixtures.20 Final yield of MscL protein could be increased to approx. 3.6 mg per ml RM by adding 18 μg/ml E. coli lipids. hese irst data are promising and further

92

Cell-Free Protein Expression

technical improvements might help to establish another new CF expression based mode for the eicient production of MPs directly into liposomes of desired compositions. he CF expression of MPs into detergent/lipid bicelles or mixed detergent micelles might ofer additional potentials for membrane protein analysis.

Membrane Protein Targets Suitable for Cell-Free Expression

What kind of MPs can be CF expressed? his question is certainly diicult to answer and no system works for anything. However, the principal elimination of the central barriers in MP production as discussed in previous chapters clearly recommend CF expression for a general and wide range of applications. he CF production of MPs in preparative scales and their subsequent functional analysis, has been approached in recent times by few individual labs. However, it can already be emphasized that the currently obtained variety of successfully synthesized MPs comprises many major MP families. Prokaryotic as well as eukaryotic MPs of α-helical or β-barrel type structure can be produced in functional forms.16,20-23 CF produced MPs include multidrug transporter, amino acid exporter, nucleoside transporter, bacterial glutamate transporter homologues, mechanosensitive channels, light harvesting proteins, aquaporins, GPCRs and ion channels, while the list is continuously growing. he functional folding has been demonstrated for a variety of CF produced MPs by transport assays, ligand binding or speciic absorption. Substrate speciicity and determined dissociation constants of CF produced EmrE matched nicely to those of in vivo expressed protein.21 he transport of speciic substrates could be monitored with CF expressed and reconstituted protein.16,21 Conductance measurements with the reconstituted CF produced mechanosensitive channel MscL by patch clamp assays showed a comparable activity with 8.3 ± 1.8 open channels per patch to the in E. coli cells overproduced channel with 9.3 ± 5.5 open channels per patch. MscL appeared to form a functional pentameric state already in micelles of the detergent Triton X-100.20 he CF produced 6.1 kDa α apoprotein of the light harvesting complex (LH1) from Rhodospirillum rubrum could be functionally refolded in a bufer system containing 0.5 to 2.0% TX-100. CD-spectra and the speciic absorbance at 820 nm of a structural complex with the β-subunit of the LH1 complex veriied its functional folding.22 he β-barrel type nucleoside transporter Tsx was expressed in CECF set-ups in levels of up to 4 mg per ml reaction mixture (RM) as soluble protein in presence of various detergents like Brij-35, Brij-78 and Triton X-100. he substrate speciic transport activity was veriied in black lipid membrane assays with the reconstituted Tsx.28 he native folding of many MPs like transporters, channels, porins and others cannot be assessed by functional assays in detergent micelles. In addition, for most MPs, in vitro assays haven’t been established yet or the function is even not known. However, an approach for the relatively rapid evaluation for the adoption of a deined conformation ofers the recording of spectra by high resolution nuclear magnetic resonance (NMR) spectroscopy. he bacterial α-helical tellurite transporter TehA (36 kDa, 10 transmembrane segments (TMSs)) and the cysteine exporter Yik (22 kDa, 6 TMSs) showed reasonable dispersion in NMR spectra, giving strong evidence for the presence of homogenously folded proteins.16,31 he yields in CECF systems for both proteins reached levels of up to 2.7 mg per ml RM. he family of GPCRs comprises many proteins of high pharmaceutical interest. Unfortunately, GPCRs are generally of low natural abundance and overexpression in E. coli or other conventional systems is in most cases very ineicient.32-34 he human β2 adrenergic receptor (β2AR), the human M2 muscarinic acetylcholine receptor (M2) and the rat neurotensin receptor (NTR) could be synthesized in batch CF systems in amounts of several 100 μg.23 N-terminal fusions to thioredoxin had to be constructed in order to receive high expression levels. However, the relatively large fusion partner could be replaced by small tags like the 12 amino acid T7-tag used for the CF expression of the porcine vasopressin type 2 receptor or the human endothelin B receptor.28,30 CF expression levels for the vasopressin receptor are with up to 6 mg per ml RM among the highest so far reported and amino acid incorporation eiciencies of 10% can be obtained. he quality of CF produced GPCR samples is excellent and not distinguishable from protein obtained out of

Cell-Free Expression Approaches for the Production and Characterization of Membrane Proteins

93

in vivo systems.30,35 Radioligand binding-assays with reconstituted CF produced β2AR protein revealed ligand speciic binding constants of 5.5 nM that are comparable to the KD of 3.9 nM of β2AR isolated from Sf9 insect cells.23 Similar ligand speciic ainities in the nanomolar range were obtained with CF produced ETB receptor.30 Size and number of TMSs seem not to play a primary role for the CF express-ability of a MP. Small multidrug transporter like EmrE or SugE with 12-14 kDa and 4 TMS as well as GPCRs with 7 TMSs and even large ion channels containing 12 TMSs could be produced in mg amounts. One feature extracted from the currently available data is a relatively strong bias of the expressed MPs to those that are almost completely embedded into membranes. A broader application to only anchored MPs or to those having extended loops or domains still has to be demonstrated. What yields of recombinant MP can be obtained by CF expression? When comparing the productivities of CF systems, several issues have to be considered. In CECF systems, the amount of synthesized protein is usually given as mg per ml of RM. he volume of the feeding mixture (FM) is not taken into account but it certainly contributes to cost calculations. Protein synthesis depends furthermore strongly on the ratio between RM and FM which usually is in between 1:10 and 1:20. Yields of recombinant protein can easily be manipulated by varying the ratio of the two compartments or by refreshing the complete FM during the reaction. he speciic reaction conditions should therefore be considered when the eiciencies of diferent CF reactions are compared. A more general basis for the evaluation of productivities would be the incorporation eiciency of the supplied amino acids into the synthesized target protein. In a CECF set-up of 1 ml RM and with a total of 17 ml FM, an average production of 2 mg of recombinant MP can be obtained. Top values obtained so far are even 6 mg in the case of porcine vasopressin type 2 receptor.28 Incorporation eiciencies of supplied amino acids can thus be routinely in the range of 3% and could go up to 10%.

hroughput Approaches and Optimization Strategies for the Production of MPs

Traditional throughput expression approaches usually employ only few variable screening parameters like diferent vectors and/or host environments. he tightly controlled and highly restricted transport of compounds through the plasma membrane of cellular expression hosts, the rapid conversion of supplied substances by metabolic activities and the sensitive physiology of living cells restrict extensive variations of expression protocols. he expression of a high number of target proteins is therefore analysed at nearly identical conditions. Typically, some 20% of the initial targets might be inally suitable for further analysis while the majority of the proteins and probably still including many of the most interesting targets, remain at too low synthesis rates. With CF expression systems, the design of throughput strategies can now be completely diferent. he strength and most powerful property of CF systems is the option to individualize expression protocols according to the requirements of the target proteins. he open nature allows a much higher variety of screening parameters and many potential problems can be directly addressed by addition of speciic compounds. Some key compounds, additives and critical steps of CF reactions with a potentially high impact on the expression of MPs are listed in Table 1. CF systems enable therefore the possibility of excessive expression condition screens for the production of important MP targets. his strategy is a classical application for automatic processes implementing robotic devices. he throughput components that will be screened in such processes are varieties of types and concentrations of reaction compounds in addition to a high number of protein targets. A prototype platform for the automatic throughput CF expression screening is shown in Figure 4. he reactions are set up in microplate formats by high performance nanoscale pipetting roboters. Incubation at various temperatures as well as the subsequent centrifugation of the RM, the puriication of the synthesized recombinant proteins and its quantiication is operated by the automatic platform in an integrated process. Optimal reaction condition parameters are evaluated by special programs that provide basic conditions for the next levels of screening experiments, thus continuously increasing the expression rates of the selected target MPs.

94

Cell-Free Protein Expression

Table 1. Optimization parameters for high level cell-free production of membrane proteins Screening Parameter

Potential Problem

Expression mode (precipitate, soluble mode) Template design (tags, regulatory regions, circular/linear DNA) Reaction temperature Detergent type Detergent concentration Lipids Ion concentrations (Mg2+, K+) tRNA composition/concentration Amino acid composition/concentration Oxidizing/reducing conditions Protease inhibitors RNAse inhibitors Energy regeneration system Polyethylene glycol Glycerol Target specific additives (cofactors, inhibitors, chaperones.)

no/low expression, inactive proteins no/low expression inactive protein, low expression precipitate, inactive protein, low expression residual precipitate, low expression inactive proteins, precipitate low expression premature termination, low expression low expression inactive protein proteolysis, inactive protein low expression low expression low expression low expression instable/inactive protein, low expression

his approach appears to be highly suitable for the fast determination of ideal CF expression conditions. In an attempt to express a comprehensive subset of the E. coli inner membrane proteome, we could already synthesize a considerable variety of MPs in preparative scale amounts ater intensive optimization of the reaction conditions. Some examples are the putative signalling protein AmpE, the camphor resistance protein CrcB, the fumarate reductase anchor FrdD, the proposed substrate locator of the aspartate/glutamate transporter GltK, the prepilin leader peptidase GspO, the hydrogenase-4 component HyfE and the so far not functionally assigned MPs YbaN, YchQ and YijD (Fig. 5). he successfully expressed MPs are involved in transport, elux, signalling, biogenesis and metabolism and they contain up to 15 TMSs.

Synergies of Cell-Free Expression and the Structural Analysis of Membrane Proteins by NMR Techniques

One of today’s major challenges in structural biology are MPs. Restrictions for their structural analysis are caused by the highly problematic sample preparation but also by technical limitations of high-resolution methods. Additional problems for the structural analysis of MPs by NMR spectroscopy are caused by efects like the rotation-correlation time, which directly depends on the size of the analysed protein, resulting in massive line broadening in NMR spectra with increasing protein size. Partial or complete overlap with adjacent signals and even vanishing of signals down to the noise level can be the consequences. Folded MPs require liposomes or micelles as environment that oten further increases the size of the rotating protein complex. In addition, typically strongly biased amino acid compositions and fold similarities of TMSs result in similar chemical shits in protein back bone selective NMR spectra, like the commonly employed 15N-1H-HSQC-correlation spectra for detection of the back bone amides. he α-helical fold of most MPs further reduces the signal dispersion of backbone amides in HSQC spectra, causing additional spectral overlaps. A key advantage of CF sample preparation of MPs for structural analysis by NMR is the complete control over the amino acid pool in the RM, consequently followed by the eicient incorporation of any kind of amino acid based labels into the target protein. Generation of stable

Cell-Free Expression Approaches for the Production and Characterization of Membrane Proteins

95

Figure 4. Throughput design for the optimization of cell-free membrane protein production. An integrated process for pipetting, incubation, separation, purification, quantification and parameter selection of CF MP expression conditions is shown. The expression screening is demonstrated with an example of Mg 2+/K+ ion concentration optimization.

isotope enriched proteins is absolutely crucial for the structural and functional analysis by NMR methods. Uniformly labelled MP samples might also be obtained with conventional in vivo expression systems by growing cells in isotope enriched minimal media, at least as long the protein is synthesized in appreciable amounts. However, amino acid type selective labelling or even combinatorial labelling of several diferent amino acid types (Fig. 6) by bacterial in vivo expression

96

Cell-Free Protein Expression

Figure 5. Preparative scale production of selected E. coli membrane proteins after optimization of the cell-free expression conditions. Sample aliquots of CECF RMs are analysed by SDS-PAGE with 16.5% gels. A) 2 μl samples of proteins produced as precipitate detected by Coomassie Blue staining; B) 0.5 μl samples of proteins soluble produced in presence of 1% Brij-78 and detected by immunoblotting with anti- GFP antibodies directed against the C-terminal His-tags. Arrows indicate the corresponding overproduced MPs.

Figure 6. Combinatorial amino acid labelling scheme. Two dimensional NMR spectra, [15N, 1H]-TROSY (upper row) and [15N, 1H]-TROSY-HN(CO) (lower row), were recorded for three specifically labelled protein samples. 15N or 13C labels were implemented as follows: sample 1, 15N: alanine, phenylalanine and isoleucine; 13C: serine and leucine; sample 2, 15N: alanine and phenylalanine; 13C: serine and glycine; sample 3, 15N: alanine and isoleucine; 13 C: leucine and valine. The detected HN(CO)-signals in the lower row identify the marked resonance as an alanine N-terminally preceded by a glycine. The indicated signal is present in all TROSY spectra and thus consequently identified as the amide resonance of an alanine. In the 2D-HN(CO) spectra, a signal is only detected in sample 2, thus indicating that the preceding amino acid residue must be a glycine.

Cell-Free Expression Approaches for the Production and Characterization of Membrane Proteins

97

Figure 7. Paramagnetic relaxation enhancement (PRE) approach for the estimation of inter-helical distances. Representative TROSY spectra of two selectively 15N-labeled mutants (V98C and L193C), containing a single cysteine residue covalently tagged with MTSL (upper row: reduced state-diamagnetic; lower row: oxidized state-paramagnetic). Annotated peaks are either completely bleached out or considerably broadened in the spectra of the oxidized species due to their proximity to the spin-label.

systems is tremendously diicult as multiple auxotrophic strains have to be employed. he required growth of E. coli in minimal media is furthermore associated with reduced growth, low protein yields and isotopic scrambling caused by the bacterial amino acid metabolisms. he production of speciically or combinatorial labelled samples is a perfect application of CF expression systems. While the technique does not improve the quality of NMR spectra in general, it can be used to simplify NMR spectra dramatically and it facilitates the precise structural analysis of even larger MPs by combination of diferent NMR methods. he recently developed combinatorial amino acid labelling scheme applied to the structural analysis of the CF produced bacterial tellurite transporter TehA enabled the rapid speciic chemical shit assignment of amino acid residues with limited or ambiguous sequential assignments.31 Two dimensional NMR spectra, [15N, 1H]-TROSY and [15N, 1H]-TROSY-HN(CO), were recorded for three speciically labelled protein samples as speciied in Figure 6 and superimposed. he labelling scheme was chosen in order to provide the highest possible information about the so far unassigned sequential stretches in TehA. he advantage of this approach is that not only amino acid types but also the corresponding linkages can be identiied, which is illustrated with one example in Figure 6. he presence or absence of cross peaks in the TROSY and 2D HN(CO) spectra in each of the samples identiies pairs of adjacent amino acid types. When such pairs are unique in the protein sequence, residues

98

Cell-Free Protein Expression

Figure 8. Preliminary structural model of TehA. A first model of a CF produced α-helical MP with seven TMSs could derived from 443 distance constraints using the dyana 1.5 simulated annealing protocol in vacuum. This calculated model symbolizes the proposed average topology of TehA in LMPG micelles. N-terminus and C-terminus are oriented on opposite sites of an assumed conformational space of a lipid bi-layer membrane.

are already unambiguously assigned and can serve as new anchor points for further backbone assignments. A remaining challenge for the structure determination of MPs by NMR methods is to obtain reliable and consistent numbers of fold-determining structural restraints. Direct distance depended NOE (Nuclear Overhauser Efect) or angle constraints mainly based on the dispersion of α-carbon and α-proton chemical shits are not eicient for a consistent determination of MP structures.36 Since unambiguous long-range NOEs are very diicult to extract from NH based 3D NOESY spectra, higher dimensional 4D-versions of NN based spectra efectively reduce spectral overlaps of overall α-helical proteins. he observed NOEs from backbone amides are sequentially, helical or caused by structurally important inter-helix contacts. he 4D-NN-NOESY veriies the secondary structural elements and supports the deinition of the global fold. However, additional structure deining constraints for simulated annealing protocols by NMR based structure determinations are necessary. A recently emerging method uses paramagnetic relaxation enhancement (PRE) in order to estimate inter-helical distances (Fig. 7).37 For site directed spin-labeling with the chemical nitroxide (1-oxyl-2, 2, 5, 5-tetramethyl-3 pyrroline-3-methyl)-methane-thiosulfonate (MTSL), deined single site cysteine containing mutants of the MP were generated, while all native cysteins had to be previously replaced by alanine. Amino acid selective 15N-labeling of the protein sample is necessary to simplify the spectra and to reduce spectral overlap. TROSY spectra recorded for both the reduced (diamagnetic) and the oxidized (paramagnetic) states of the MTSL-modiied protein sample deine then pairs of upper and lower limit for an individual amide-amide distance. he CF produced functionally active 24 kDa fragment of the putative tellurite and multi-drug transporter TehA could inally be assigned to 85% of all back bone resonances and a irst preliminary model could be obtained by using the NOESY and PRE based distance constraints (Fig. 8). Based on 120 meaningful distance constraints as pairs of upper and lower limits for amide-amide distances derived from PRE experiments, 205 interhelical or long-range distance constraints based on unambiguous NOEs from 4D-NN-Noesy spectra, 118 mid-range NOEs and 170 angle

Cell-Free Expression Approaches for the Production and Characterization of Membrane Proteins

99

constraints to deine the secondary structure, a irst preliminary structural model of TehA could be calculated (Fig. 8). his irst model already shows the seven expected TMSs while still containing relatively large nonstructured loop regions and it is currently subject of further reinements. However, this irst example already demonstrates the indispensable and highly valuable support of CF expression techniques for the structural determination of MPs.

Conclusions

he immense potential of CF expression systems either for the expression of diicult proteins or for special applications like labelling is still by far not exhausted. he discussed development of eicient liposome based MP expression strategies will only be one of the fascinating perspectives. Co-expression of multisubunit protein complexes, cotranslational incorporation of supplied artiicial cofactors, site-speciic or segmental labelling of proteins for NMR spectroscopy, even posttranslational modiications of MPs by providing corresponding enzyme systems are some further feasible achievements. Alternative sources for CF extracts in addition to E. coli and wheat germs might provide additional interesting properties. he much easier to handle batch systems will become more and more attractive due to improved productivities and optimized protocols and the design of CECF reaction processes will be further streamlined in order to become more user-friendly. he rapidly accumulating information obtained from throughput approaches and systematic reaction condition screens will provide a comprehensive and reliable database of knowledge for the preparative scale CF synthesis of many MP targets. Based on the growing variety of successfully CF produced MPs, it is tempting to speculate that especially fully membrane embedded proteins characterized by multiple TMS topologies, once the most diicult target for expression attempts, might even switch to one of the easiest expression targets in the next future.

Acknowledgements

he work was inancially supported by the DFG and the SFB 628 ‘Functional Membrane Proteomics’.

References

1. Tate C. Overexpression of mammalian integral membrane proteins for structural studies. FEBS Letts 2001; 504:94-98. 2. Bannwarth M, Schulz GE. he expression of outer membrane proteins for crystallization. Biochim Biophys Acta 2003; 1610:37-45. 3. Wang DN, Saferling M, Lemieux MJ et al. Practical aspects of overexpressing bacterial secondary membrane transporters for structural studies. Biochim Biophys Acta 2003; 1610:23-36. 4. Alakov YB, Baranov VI, Ovodov SJ et al. Method of preparing polypeptides in cell-free translation system in: US Patent 5 1995; 478:730. 5. Kim DM, Choi CY. A semicontinuous prokaryotic coupled transcription/translation system using a dialysis membrane. Biotechnol Prog 1996; 12:645-9. 6. Kigawa T, Yabuki T, Yoshida Y et al. Cell-free production and stable-isotope labeling of milligram quantities of proteins. FEBS Lett 1999; 442:15-9. 7. Kim DM, Swartz JR. Prolonging cell-free protein synthesis with a novel ATP regeneration system. Biotechnol Bioeng 1999; 66:180-8. 8. Kim DM, Swartz JR. Prolonging cell-free protein synthesis by selective reagent additions. Biotechnol Prog 2000; 16:385-90. 9. Jewett MC, Swartz JR. Substrate replenishment extends protein synthesis with an in vitro translation system designed to mimic the cytoplasm. Biotechnol Bioeng 2004; 87:465-72. 10. Jewett MC, Swartz JR. Mimicking the Escherichia coli cytoplasmic environment activates long-lived and eicient cell-free protein synthesis. Biotechnol Bioeng 2004; 86:19-26. 11. Michel-Reydellet N, Calhoun K, Swartz J. Amino acid stabilization for cell-free protein synthesis by modiication of the Escherichia coli genome. Metab Eng 2004; 6:197-203. 12. Torizawa T, Shimizu M, Taoka M et al. Eicient production of isotopically labeled proteins by cell-free synthesis: a practical protocol. J Biomol NMR 2004; 30:311-25. 13. Kigawa T, Yabuki T, Matsuda N et al. Preparation of Escherichia coli cell extract for highly productive cell-free protein expression. J Struct Funct Genomics 2004; 5:63-8. 14. Ozawa K, Jergic S, Crowther JA et al. Cell-free protein synthesis in an autoinduction system for NMR studies of protein-protein interactions. J Biomol NMR 2005; 32:235-41.

100

Cell-Free Protein Expression

15. Zubay G. In vitro synthesis of protein in microbial systems. Annu Rev Genet 1973; 7:267-87. 16. Klammt C, Lohr F, Schafer B et al. High level cell-free expression and speciic labeling of integral membrane proteins. Eur J Biochem 2004; 271:568-80. 17. Liu DV, Zawada JF, Swartz JR. Streamlining Escherichia coli S30 extract preparation for economical cell-free protein synthesis. Biotechnol Prog 2005; 21:460-5. 18. Yamane T, Ikeda Y, Nagasaka T et al. Enhanced cell-free protein synthesis using a S30 extract from Escherichia coli grown rapidly at 42 degrees C in an amino acid enriched medium. Biotechnol Prog 2005; 21:608-13. 19. Martemyanov KA, Shirokov VA, Kurnasov OV et al. Cell-free production of biologically active polypeptides: Application to the synthesis of antibacterial peptide cecropin. Protein Expr Purif 2001; 21:456-461. 20. Berrier C, Park KH, Abes S et al. Cell-free synthesis of a functional ion channel in the absence of a membrane and in the presence of detergent. Biochemistry 2004; 43:12585-91. 21. Elbaz Y, Steiner-Mordoch S, Danieli T et al. In vitro synthesis of fully functional EmrE, a multidrug transporter and study of its oligomeric state. Proc Natl Acad Sci USA 2004; 101:1519-24. 22. Shimada Y, Wang ZY, Mochizuki Y et al. Functional expression and characterization of a bacterial light-harvesting membrane protein in Escherichia coli and cell-free synthesis systems. Biosci Biotechnol Biochem 2004; 68:1942-8. 23. Ishihara G, Goto M, Saeki M et al. Expression of G-protein coupled receptors in a cell-free translational system using detergents and thioredoxin-fusion vectors. Protein Expr Purif 2005; 41:27-37. 24. Pornillos O, Chen YJ, Chen AP et al. X-ray structure of the EmrE multidrug transporter in complex with a substrate. Science 2005; 310:1950-3. 25. Klammt C, Schwarz D, Löhr F et al. Cell-free expression as an emerging tool for the large scale production of integral membrane proteins. FEBS J 2006; 273:4141-53. 26. Schwarz D, Klammt C, Koglin A et al. Preparative scale cell-free expression systems: new tools for the large scale preparation of integral membrane proteins for functional and structural studies. Methods 2007; 41:355-69. 27. Raman P, Cherezov V, Caffrey M. The membrane protein data bank. Cell Mol Life Sci 2006; 63:36-51. 28. Klammt C, Schwarz D, Fendler K et al. Evaluation of detergents for the soluble expression of alpha-helical and beta-barrel-type integral membrane proteins by a preparative scale individual cell-free expression system. FEBS J 2005; 272:6024-38. 29. Garavito RM, Ferguson-Miller S. Detergents as tools in membrane biochemistry. J Biol Chem 2001; 276:32403-6. 30. Klammt C, Srivastava A, Eiler N et al. High level cell-free expression of human endothelin B receptor: Transmembrane segment 1 represents a core area for ET-1 binding and dimer formation. FEBS J accepted. 31. Trbovic N, Klammt C, Koglin A et al. Eicient strategy for the rapid backbone assignment of membrane proteins. J Am Chem Soc 2005; 127:13504-5. 32. Stanasila L, Pattus F, Massotte D. Heterologous expression of G-protein-coupled receptors: human opioid receptors under scrutiny. Biochimie 1998; 80:563-71. 33. Massotte D. G-protein-coupled receptor overexpression with the baculovirus-insect cell system: a tool for structural and functional studies. Biochim Biophys Acta 2003; 1610:77-89. 34. Sarramegna V, Talmont F, Demange P et al. Heterologous expression of G-protein-coupled receptors: comparison of expression systems fron the standpoint of large-scale production and puriication. Cell Mol Life Sci 2003; 60:1529-46. 35. Klammt C, Schwarz D, Eiler N et al. Cell-free production of G-protein-coupled receptors for functional and structural studies. J Struct Biol 2007; doi:10.1016/j.jsb.2007.01.006. 36. Wüthrich K. Protein structure determination in solution by NMR spectroscopy. J Biol Chem 1990; 265:22059-62. 37. Liang B, Bushweller JH, Tamm LK. Site-directed parallel spin-labeling and paramagnetic relaxation enhancement in structure determination of membrane proteins by solution NMR spectroscopy. J Am Chem Soc 2006; 128:4389-97.

Chapter 8

Cell-Free Expression for Protein NMR A.J. Shaka*

Abstract

C

ell-free expression systems ofer new possibilities to produce protein samples that are suitable for nuclear magnetic resonance (NMR) spectroscopy. Advantages of the cell-free approach include enhanced sample stability, minimal sample puriication strategies and, especially, the prospect of introducing stable isotopes (13C, 15N, and/or 2H) at predetermined amino acid sites in the primary sequence. While some amino acid-speciic labeling has been attempted and accomplished with more conventional E. coli expression systems, facile conversion of some amino acids to others in living organisms can lead to cross-talk that complicates NMR spectral analysis and makes this approach less attractive. In addition, when perdeuteration of the protein is desired, as is oten the case with 25 kDa or larger proteins, yield from the in vivo systems can be quite low. In addition, the cost can be an issue, as 2H2O, deuterated bufers, and deuterated glucose must be employed in rather large volumes. By contrast, the cell-free systems need only employ deuterated amino acids. Yield is similar to that obtained with 1H amino acids, reaction volumes are small, and speciic amino acids can be deuterated to help with spectral simpliication. For all these reasons, cell-free protein expression is expected to play a signiicant future role in solution structure determination by NMR.

Introduction

Most recombinant proteins for nuclear magnetic resonance (NMR) studies are produced because the three-dimensional solution structure of the protein target is of interest. hese structure determinations are labor- and time-intensive undertakings and newcomers require considerable training from an experienced practitioner before they can achieve success. here are other, more specialized, reasons to want to obtain spectra from a recombinant protein as well, but most oten at least the NMR spectrum, if not the structure of the target itself, must be fairly well characterized before useful information can be obtained (e.g., ligand binding, protein dynamics, etc.,). A brief introduction to the sequence of events that must occur in an NMR solution structure determination is thus important to put cell-free protein expression into context.

A Bird’s Eye View of NMR Structure Determination

NMR is a proven method to investigate the three-dimensional structure of proteins in solution,1-3 but with standard NMR structure determination methods come a number of sample requirements that are generally more stringent than other spectroscopic methods. Without attention to these prerequisites, the structure determination is doomed from the outset. Briely, and in broad brushstrokes, the essential elements we must keep in mind are as follows: (i) there must be enough material to obtain a decent NMR spectrum in a reasonable amount of time, typically around 500 μM concentration in a volume of several hundred microliters, or around 0.25 μmol of protein, so that e.g., 5 mg of a 20 kDa protein is required; (ii) the protein must be pure enough that contaminant *A.J. Shaka—Chemistry Department, University of California, Irvine, California 92697-2025, U.S.A. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

102

Cell-Free Protein Expression

peaks do not confuse the spectrum; (iii) the protein must be stable enough over time, in solution and sometimes at elevated temperatures (>40 °C), that high-resolution multidimensional spectra can be obtained, with acquisition times from several days to several weeks oten required; (iv) the protein must be soluble enough, the lower limit being ca. 50 μM with state-of-the-art 900 MHz spectrometers and the highest sensitivity NMR probes and preampliiers, while a 1 mM solution is required at lower magnetic ield strength; (v) the protein must assume a well-deined structure under the conditions of study, as lexible regions may give signals that are weak or broad and not discernible above the noise; and (vi) it must be possible to enrich the protein with stable isotopes (13C, 15N and sometimes 2H), usually nearly 100% incorporation of these being required.4,5 his last point means that nearly all proteins studied these days are produced using recombinant DNA technology. he standard sequence of events in an NMR structure determination is irst to obtain an assignment. NMR spectral peaks, or groups of such resonance peaks, are tentatively linked to speciic atomic positions in the polypeptide chain. Usually the backbone HN, 15N, 13C’ and 13Cα resonances are assigned irst, followed by the 13Cβ and side chain positions, and usually a uniformly 100% enriched 15N/13C protein sample is employed. To distinguish identical amino acids at diferent positions in the primary sequence, the chemical shit is centrally important. Each NMR peak occurs at a frequency that relects not just the applied magnetic ield, B0, but the local magnetic ield experienced by the nuclear spin in question. his local ield is sensitive to the environment the nucleus “sees” including magnetic ields produced by electrons in chemical bonds. his means that a proton attached to a particular Cα will have an exact resonance frequency that depends on the side chain of the amino acid, and any functional groups such as aromatic rings that may be spatially close in the correctly-folded protein. Nevertheless, when there are many amino acids of the same type, and if the peaks are broadened, as they are for proteins of relative molecular mass greater than 20 kDa, we may replace many of the hydrogen nuclei (protons) with deuterium to try to improve the resolution (see below). One unique aspect of NMR is our ability to display the spectrum as amplitude as a function of one, two, three, or four diferent frequencies; in most kinds of spectroscopy only a graph of amplitude as a function of a single frequency (or, equivalently, energy) can be displayed meaningfully. his should be familiar from IR or UV-VIS spectroscopy. Two-dimensional (2D) NMR spectra display amplitude versus two orthogonal frequencies, which we typically contour to visualize the peaks, like a topographic map. Higher-dimensional spectra can be sliced at certain coordinates, and 2D planes extracted and contoured, although it becomes increasingly diicult to form an accurate mental image of the high-dimensional object itself. As the number of dimensions is increased, overlapping peaks are resolved. For historical reasons, spectra in which all frequency dimensions refer to protons are called homonuclear whereas spectra in which at least one dimension refers to a diferent nuclear spin species (13C or 15N most commonly) are called heteronuclear. For example, in a heteronuclear 2D experiment designed to show a peak whenever an amide HN proton is bonded to a 15N, identical amino acids at diferent positions along the primary sequence may have very close chemical shits in both the proton and 15N dimensions. he contour plot may get very crowded, and we may not be able to even tell how many peaks are present. However, by including a third dimension, in which the carbonyl carbon at position i-1 is included, there is far less chance of overlap: each identical amino acid at say positions i and j could have diferent predecessors i-1 and j-1 or, even if these are also identical by chance, the C’ chemical shits may still be slightly diferent due to diferences in the local environment. By separating out the peaks it is easier to identify the correct number of peaks and which peaks, if any, may be weak or missing. his is shown schematically in Figure 1 for a 1D versus a 2D spectrum. As protein spectra are much more crowded than these, a full suite of at least 3D NMR experiments5 is usually required to tackle the assignment problem, and each experiment may take a day or two to perform.6 Figure 2 shows the detected proton NMR spectrum in one of these experiments applied to a small protein. his 1D spectrum gives us a better idea of the sheer number of peaks that are present, and the challenge that the assignment of all these transitions will be.

Cell-Free Expression for Protein NMR

103

Figure 1. A schematic illustration of the resolving power of multidimensional NMR spectroscopy. A hypothetical stretch of the amide NH region is shown. The 1D spectrum shows some peaks that may overlap (top) because the proton chemical shifts just happen to be degenerate. However, as long as the 15N chemical shifts are not also degenerate, the 2D contour plot will tease apart the peaks and allow them to be counted. As we know the primary structure of any protein being investigated by NMR, we know how many of these peaks to expect, and the 2D spectrum alerts us to weak, missing, or possibly overlapping peaks, as well as the possible presence of foreign contaminants.

Another aspect of multidimensional NMR spectra is that there is considerable freedom concerning which nuclear frequency is used for detection. In a 1D spectrum, the kind of spectrum dictates the kind of nuclear spins that must be detected: for carbon-13 NMR we detect the 13C frequency, which is only ¼ of the 1H frequency. he magnetic moment is proportional to the NMR frequency, so that carbon-13 spins give a factor of ¼ smaller magnetization. As a Faraday detection

104

Cell-Free Protein Expression

Figure 2. A typical 1D proton NMR spectrum of a protein. The trace is actually the first scan in a multidimensional experiment, so that the strong H2O resonance at 4.8 ppm, which would be many thousandfold stronger than any of the peaks shown, can be effectively filtered out of the spectrum. Amide NH protons and aromatic protons give peaks to the left (downfield) whereas Hα and side chain protons give peaks to the right. Matching up each and every peak to a particular magnetic spin flip of nucleus at a certain position on the molecule is a very tough problem.

scheme is used, in which the derivative of the magnetic lux induces a voltage in the receiver coil that ultimately gives the signal we see, another frequency factor enters in: d sin 2πft = 2πf cos 2πft . dt

(1)

he relative signal-to-noise ratio (SNR) thus scales, conservatively, as at least f 2 meaning that the time to obtain equivalent SNR scales at least like f 4. Even in a molecule that is 100% enriched with carbon-13, it is still around 256 times quicker to detect the proton NMR spectrum rather than the carbon-13. In multidimensional NMR we can watch the proton peaks change phase or amplitude while incrementing inter-pulse delays on the 13C spins, or vice-versa. However, the proton peaks will show far better SNR than carbon-13. For this reason, proton detection is the norm when undertaking an assignment. While multidimensional NMR is very powerful, it may also be time consuming. In a 2D experiment, the entire NMR pulse sequence must be repeated in entirety some number N1 times, the only diference being the incrementing of a time delay somewhere in the sequence of pulses, to allow the chemical shit information of the spins not being directly recorded to be obtained indirectly. he digital resolution in the inal spectrum depends on the number of points acquired in each dimension, and the frequency range (or spectral width) in each dimension. For example, the 15N amide chemical shit range is around 40 ppm for most proteins, and the resonance frequency of

Cell-Free Expression for Protein NMR

105

15 N is around 50 MHz in a magnet in which protons resonate at 500 MHz. hus, the frequency range in the 15N dimension must be at least (50 x 106 Hz)(40 x 10-6) = 2 kHz. To obtain 10 Hz digital resolution thus requires around 200 points, or 200 separate executions of the entire NMR pulse sequence. In addition, some pulses usually must be changed in phase or amplitude on different transients and the results co-added to remove unwanted signals. his may require 4 or 8 repetitions, leading to 800 or 1600 experiments. As some time is required between experiments to allow the spin system to return to a reproducible steady state magnetization, each experiment requires a second or two. hus, even a 2D experiment might take from 10 minutes to an hour. A 3D experiment is simply a series of 2D experiments and so the time requirement is multiplicative. he necessity to obtain multiple 3D NMR spectra to obtain the assignment is one bottleneck in solution structure determination. Once the assignment is nearly complete, experiments to identify which nuclei are spatially proximate, usually nuclear Overhauser spectroscopy or NOESY7,8 experiments, are performed. As shown schematically in Figure 3, the presence of a “cross peak” between two assigned resonances is interpreted to mean that the two nuclei are spatially proximate, and the stronger the peak, roughly speaking, the closer the spins should be. Clearly, in the upield region that is boxed, there are a host of peaks that cannot be discerned individually. hus, a stronger peak might actually be two or more degenerate NOEs rather than one. Even if it were not, the information based on intensity analysis of the cross peaks is qualitative at best, and cannot be relied upon to give actual distances in most cases. Furthermore, the absence of a NOESY cross peak does not mean that the spins in question are not proximate, as there are many other factors that can lead to weak or missing peaks in the spectrum. Further constraints on dihedral angles can be obtained if the peak splittings due to scalar couplings can be measured, and some additional information based on the measured chemical shits of backbone atoms can be utilized. Recently, the advent of weakly-aligned liquid crystal media, in which scaled down dipolar couplings (so-called residual dipolar couplings, or RDCs) can be seen, has proven to be a powerful addition to the arsenal of experiments.9 Once enough spectral information has been obtained, a structure calculation can be undertaken. he methods used are constantly being improved, but a standard sequence might run as follows. Mock structures are generated randomly, and numerically optimized, oten by Monte Carlo simulated annealing10 (MCSA) in torsional space: atoms are kept at accepted bond distances and bond angles are randomly changed with a certain maximum step size.11 Ater each move the energy is computed using a potential function optimized for proteins, and including a pseudo-energy ENOE based on whether known contacts from the NMR experiments are preserved. he additional potential is zero and typically lat if two nuclei are within range, quadratically rises as the distance constraint is violated, and inally becomes linear at very large violation distances. he last point is essential to avoid enormous pseudo-forces that can potentially destabilize the numerical calculation when the conformation is well away from the true minimum.12 A good structure is one in which all covalent bonds have correct distances and acceptable geometry, and all nuclei that are supposed to be close to each other, on the basis of the NMR assignment and NOESY experiments, are actually close to each other. An acceptable structure may have a certain number of restraint violations, i.e., nuclei are father apart than the NOESY spectrum indicates. hese violations may indicate an incorrect assignment, or may arise from lexibility or dynamics. No structure is trusted without conirmation, so this whole process is repeated a number of times, to check whether convergence to a similar structure is observed. For regions where a wealth of NMR constraints are known, the family of structures are usually in quite good registration with low root-mean-square deviation (RMSD) of heavy atoms, whereas regions of the structure with fewer constraints show larger deviations. his could mean that the region in question is a lexible loop, or it could just mean that the lack of NMR data allows a number of diferent low-energy conformations, each of which match the sparse data available. he methods to reine and validate NMR structures are continually being improved.13 Nevertheless, the typical quality of NMR structures is not as good as X-ray structures. Furthermore, a few false NOESY positives, which can occur whenever the assignment is erroneous, can lead to very bad structures

106

Cell-Free Protein Expression

Figure 3. A typical 2D NOESY spectrum, in which we can imagine the conventional 1D spectrum to be along the diagonal of the figure (diagonal peaks). The presence of symmetrical cross peaks at positions (δA ,δB) and (δB,δA) indicates that protons with chemical shifts δA and δB are spatially proximate. An example highlighting two methyl groups from different side chains is shown. The protein is the yeast N-terminal domain (1-115 residues) of Tfb1 (a subunit of General Transcription Factor II H, TFHIIH). The spectrum was recorded on a Varian 600 MHz spectrometer at 25°C at pH 6.5 with a 100 ms mixing time in 100% D2O solution. (Spectrum courtesy of Dr. Bao Nguyen, UCI).

even though the apparent clustering of the ensemble looks decent. It is crucial to get the correct assignments in the irst place if we are to have any conidence in the inal 3D structure.

Practical Limits for NMR Structure Determination

Conversely, there is essentially no hope of determining a protein structure by NMR if the spectra are so crowded that resonance assignments cannot be made. When peak overlap is heavy the result is just a featureless lump: it is impossible to ferret out the individual peaks and assign them to atomic positions. his lack of resolution imposes an upper limit to the size of proteins that we can tackle. here are only a handful of examples of proteins 40 kDa or larger that have had their NMR structures solved.14 Increasing the magnetic ield strength B0 will linearly increase the peak

Cell-Free Expression for Protein NMR

107

separations in each dimension of the NMR spectrum and so improve the resolution greatly. But the huge cost of instruments at 900 MHz (21 tesla magnet, ca. $5M) or higher, and the very slow rate of progress in the design and fabrication of high-stability, very homogeneous, superconducting solenoids makes it unlikely that much bigger magnets will solve the resolution problem. Hand in hand with the resolution problem is the sensitivity problem. Very large proteins are oten not very soluble. Using high salt concentrations is common to stabilize the molecules and facilitate their solubility. However, high salt interferes with the radiofrequency penetration of the pulses used to excite the nuclear magnetic spins in the molecule, and contributes background random noise in the NMR receiver coil, thereby lowering the signal-to-noise ratio.15 he twin problems of resolution and sensitivity are a bottleneck for every structure determination by NMR, so that any improvement in either area is a major advance. Another way to efectively improve the resolution is to make the resonance peaks narrower. For backbone assignment, we can do this by deuterating the protein, i.e., replacing every 1H nucleus by a 2H (D) nucleus throughout the entire molecule. he line width, in particular, of the Cα peaks depends predominately on the interaction with its closest luctuating magnetic partner, the directly-bonded Hα spin. Random magnetic luctuations caused by molecular motion in solution lead to slightly diferent integrated environments over time, which yields a distribution of frequencies, or homogeneous line broadening. Equivalently, viewed in the time domain, the tiny precessing magnetic moments gradually get out of step: when the magnetic moments are distributed uniformly around a circle there is no signal let. he decay is exponential with a time constant T2, the transverse relaxation time. Short T2s limit the ability to transfer magnetization from one spin to another and obtain multidimensional spectra. By replacing the carbon-bound hydrogen with deuterium, which has a magnetic moment only about 1/6th that of proton, much narrower peaks are obtained for the important backbone Cα signals, leading to improved resolution. Most recombinant proteins for NMR, whether deuterated or not, have been expressed in the well-characterized bacterial host E. coli as a matter of course. Deuteration in cell-based expression systems can be done by using D2O instead of H2O in growth medium. Levels of around 70% deuterium incorporation are obtained. To get 100% deuterium labeling requires expensive deuterated bufers, as well. he bacteria do not grow as well in D2O as in H2O, so that cells selected to be tolerant to heavy water are used. Even so, doubling times may increase by several hours so that cell growths may require 2-3 days from inoculation to harvest, and yield is oten only 1/3 to 1/10 that of the yield in fully protonated media.16,17 Ater protein expression, the amide ND groups must exchanged in protonated media to NH groups or there would be no proton signal to observe at all! Usually the standard 90%/10% H2O/D2O solvent, the 10% D2O being required for the ield/frequency lock system, is used for the back exchange of the ND groups to mostly NH. As noted above, proton detection is used to obtain best SNR. However, for proteins in the range 30-40 kDa, where deuteration would most likely be used, the D/H back-exchange strategy may fail for some deeply buried sites, or in cases where there are strong hydrogen bonds. Denaturing large proteins by chemical means, to try to expose these sites for exchange, is risky because refolding them correctly again may not be simple. In these cases some alternatives have been put forward, including expressing the protein in a fully deuterated algal lysate medium in 100% H2O. While the method has been demonstrated to have potential, it does lead to partial protonation of Cα sites in a residue-type dependent manner, which lowers SNR.18 he cost of liters of deuterated media, the longer growth time, the decreased protein yield, and the potential diiculty of reintroducing protons into the protein are all negative aspects of perdeuteration of recombinant proteins. Another downside is the loss of most of the side chain information. Side chain packing is one area where de novo computational methods still have signiicant diiculty in getting the correct low-energy solution, and it is one in which the NMR constraints can be exceedingly useful. However, if all carbons on the side chains are deuterated, then it is not possible to obtain any distance constraints between any of the side chains, because there are no 1H nuclei to detect in a NOESY experiment. While NH-NH contacts supply some distance information, it is typically too sparse to deine the tertiary structure with any certainty.

108

Cell-Free Protein Expression

An extremely elegant, although partial solution to this diiculty was introduced by Rosen et al19 and subsequently used to identify global folds of high deuterated proteins.20 he breakthrough here relies on knowledge of the biochemical pathways in E. coli to introduce protonated H3Cgroups into the amino acids alanine, valine, leucine and isoleucine (γ2 only) by using protonated [13C]pyruvate as the sole carbon source. hese four amino acids are among the most ubiquitous in the hydrophobic core of the protein and at protein-protein interfaces in complexes. As clearly demonstrated in reference 20, the addition of Me-Me, and Me-NH distance constraints by selective methyl protonation of a perdeuterated protein greatly improves the structure compared to the NH-NH distance information available from a perdeuterated protein. his groundbreaking work clearly showed how more elaborate patterns of isotopic labeling can help us tackle larger proteins by NMR, and set the stage for further exploration in technology to enable novel, controlled patterns of isotope incorporation into recombinant proteins.

Cell-Free Protein Expression

As we will see, cell-free expression is the next logical step when it comes to precision control of isotopic labeling. However, one barrier to the adoption of cell-free expression of recombinant proteins for NMR is that the scientists involved in structural biology studies may not have the kind of detailed biochemical expertise to be able to formulate their own cell-free expression systems successfully. However, that aspect is now changing rapidly, as commercial suppliers are making cell-free systems available as kits that deliver good results in non-expert hands. Cell-free systems, using E. coli or wheat germ optimized lysates are available as kits from companies such as Invitrogen (Expressway™ NMR), Roche Applied Science (Rapid Translation System, RTS™), QIAGEN (EasyXpress™) and CellFree Sciences (ENDEXT®). he availability of this technology is now changing the landscape in protein expression. Gradually, more and more groups are turning to cell-free protein expression to prepare samples for NMR spectroscopy. While in principle almost any living organism would possess the translational machinery to express proteins, in practice E. coli21-23 or wheat germ23,24 extracts seem to be the most commonly used for NMR applications.25-32 he cell-free expression systems hold considerable interest for NMR spectroscopy mainly because the in vitro system allows considerable lexibility in the incorporation of stable isotopes, allowing the formulation of new NMR strategies for assignment and higher throughput. his lexibility will be a major advantage for cell-free expression in the future. he most common isotopic labeling protocol by conventional E. coli protein expression uses minimal media in combination with either 15N ammonium chloride and/or uniformly 13C-enriched glucose as the raw materials for isotope incorporation. As a consequence, every single nitrogen and/or carbon is labeled throughout the protein (and throughout other molecules present in the bacteria as well). Having every backbone and side-chain atom be NMR-active gives us the maximum available information because every possible peak that can appear will do so; but it also means that the NMR spectrum can become so crowded with peaks that the assignments cannot be made reliably. Residue-selective labeling of proteins using cell-based expression systems33-35 can be done in some cases by supplying the labeled amino acids themselves, but the technology and its implementation is not routine and may fail if transaminase activity scrambles one amino acid into another. In addition, incorporation of the labeled amino acids is oten less than 100% and depends on amino acid type, complicating the NMR spectra and reducing sensitivity.36 Compared with the ease of use and simplicity of in vitro expression, it is clear that these pioneering and in some cases nearly heroic eforts to generate selectively labeled proteins will see less use in the future.

Amino Acid Speciic Labeling by Cell-Free Expression

Figure 4 shows an idealized 2D 15N-1H spectrum that might be observed from a large protein. here are in fact sixteen peaks present in the center spectrum, but they overlap to such an extent that even in this noiseless example it is impossible to discern each individual feature with idelity. However, the four subspectra around the perimeter each show four well resolved peaks, making their assignment routine. Uniform labeling will always lead to the most crowded spectrum, like

Cell-Free Expression for Protein NMR

109

Figure 4. An illustration of the resolving power of a divide-and-conquer strategy to isotopic labeling. The center panel shows a crowed region containing 16 individual peaks of different widths and positions: four peaks from each of four different amino acid types. The four spectra around the perimeter show each subspectrum to be expected when only one amino acid type is labeled. Although the subspectra need not be resolved in the general case, the chance of distinguishing the individual peaks is much higher when there are fewer of them.

the center panel, whereas selective labeling by amino acid type would yield something like each of the four subspectra. Furthermore, we know which amino acids have been labeled in each of the subspectra, and this information is incredibly useful for a computer-aided automated assignment procedure. he ield is moving toward fully automated assignment to try to boost throughput, but with only a single uniformly labeled sample the results so far have been mixed. Hopefully, cell-free expression technology will be able to tip this balance in the future, as assignment is viewed as a necessary evil to determine structure and is tedious even for a human expert. Of the 20 amino acids, 19 have HN protons (proline being the exception) and so the very best resolution in the NH plane would result by preparing 19 diferent protein samples, each one with only a single type of amino acid enriched 100% in 15N. his kind of control is essentially out of reach in cell-based protein expression, but is simple to implement with cell-free systems. he key diference is that the amino acids themselves are supplied directly to the in vitro reaction mixture, so that using labeled or unlabeled amino acids is simply a matter of synthesizing, or purchasing

110

Cell-Free Protein Expression

from companies specializing in stable isotopes (CIL, Isotec, Spectra, Novachem, etc.,). With an extensive collection of labeled amino acids, the kinds of samples that can be prepared by cell-free expression are limited only by our imagination. Unlabeled amino acids are inexpensive, so that the cost of producing a host of diferent protein samples is limited mostly by cost of the other reagents supplied in the kits. he cost of these will most likely drop as demand increases, in much the same way that the cost of 15N and 13C isotopes for conventional protein expression for NMR has come down in price over the last ten years as improved technology and economy of scale have entered into the marketplace. Preparing 19 diferent samples is overkill for most applications, although it has been done simply to demonstrate unequivocally the speciicity, high incorporation, and lack of any scrambling by Ozawa et al for the case of human cyclophilin A, a 17 kDa protein.37 Protein yields of 1.8 mg/mL were obtained in the E. coli cell-free expression system, and all metabolic byproducts that could give rise to spurious NMR peaks in crude extracts were identiied.37 hese byproduct peaks containing 15 N were usually quite small compared to the 15N peaks from the desired protein, although care must be exercised in some cases. For example, the reaction mixture is carried out in a 200 mM glutamate bufer, making incorporation of 15N-glutamic acid impossible unless all the bufer were labeled, an expensive proposition. In addition, [15N]Asp showed new peaks attributable to [15N]Asn when glutamate bufer was used.37 hese extra peaks, and the diiculty with selective Glu labeling both disappear when an acetate bufer is used in lieu of glutamate. Use of D-glutamate rather than L-glutamate in the bufer gave similar protein yield, but was still unsuitable for selective Glu labeling, perhaps because of the presence of a glutamate racemase in the E. coli S30 extract. Figure 5 shows results obtained with the Invitrogen Expressway NMR™ kit on the small ubiquitin-like modiier protein SUMO-1, a 12 kDa protein. Using a commercial mixture of uniformly 15N amino acids (CIL, Andover Mass.) as feed, high yield of [15N]-labeled SUMO-1 results (Fig. 5A), but when [15N]Glu and [15N]Gln are used in acetate bufer, a sparse spectrum results in which the correct positions are labeled (Fig. 5B).32 Original work on the structure of SUMO-1, expressed in E. coli described very poor sample stability: protein had to be prepared anew every few days to continue data acquisition, as degradation occurred in the NMR tube.38 By contrast,

Figure 5. Results on the 12 kDa protein SUMO-1 obtained by cell-free protein synthesis using the Invitrogen Expressway™ NMR kit. A) Uniformly 15N-labeled SUMO-1 2D HSQC spectrum, showing the chemical shifts of the 1H (horizontal axis) and 15N (vertical axis) amide functional groups along the backbone (and some nitrogen-containing side chains). B) Same as (A) except labeled only with glutamic acid and glutamine, using an acetate buffer rather than glutamate buffer for the in vitro reaction. The spectra were obtained at pH 6.0 and 25°C in 90%/10% H2O/D2O on an 800 MHz Varian Inova spectrometer that forms the core of the UCI Biomolecular NMR Facility.

Cell-Free Expression for Protein NMR

111

the recombinant SUMO-1 from the cell-free expression system was stable in the NMR tube for many months, as evidenced by the rock-solid 15N-1H 2D NMR spectrum.32

Optimized Labeling Schedules for Backbone Assignment

Merely labeling the nitrogen nuclei does nothing to assign the carbon backbone, and [13C] amino acids are expensive. hus making the full suite of 19 individually-labeled 15N-protein samples, each of which is also labeled with 13C at other positions would be prohibitively costly. he question that arises naturally, then, is this: “Assuming we have complete control over the distribution of isotopes in the protein, how many diferent samples should we synthesize, and what combinations of 15N/13C labeling should be used?” he answer will depend on (a) the assignment strategy; (b) the size of the protein; (c) the primary sequence and the numbers of each type of amino acid therein; and (d) the (unknown) structure of the protein, as this may inluence the spreading of the peaks in the NMR spectra. As almost all the spectra used for assignment involve detection of those amide HN protons that are bonded to 15N, this plane is the most important one to consider. By surveying the existing database of assigned protein chemical shits at the Biological Magnetic Resonance Bank (www. bmrb.wisc.edu), we can get some idea of how crowded the NH plane is likely to be before we ever prepare the irst sample. Figure 6 shows the peak probability distribution for four example amino acids: glycine, threonine, tyrosine, and alanine. he ellipses show the area of highest probability, of

Figure 6. Probability distributions reflecting the observed positions of 2D 15N-1H peaks from all measured proteins that have been assigned (excluding paramagnetic proteins and/or peaks that differ by more than three standard deviations from the mean) in the BMRB. The contours are drawn at one standard deviation. It is clear that some amino acids are sufficiently different that they are very unlikely to accidentally overlap in the 2D spectrum (e.g., glycine and alanine) whereas other pairs have a significantly greater chance of overlap (e.g. threonine and tyrosine). Making plots such as these for all the amino acids allows an accurate assessment of the resolution likely to be had for any combination of labels, given the size of the protein, the expected line widths, and the number of each kind of amino acid present. This information, in turn, lets us generate labeling schedules that maximize the resolution while minimizing the number of samples that we have to prepare. (Plots courtesy of Mr. Mike Sweredoski).

112

Cell-Free Protein Expression

one standard deviation, where we expect to see the NMR peaks. Clearly it is safe to label all glycine and all alanine residues simultaneously in a protein with 15N because the signals are very unlikely to overlap. On the other hand, labeling threonine and tyrosine simultaneously may lead to some chance of overlap. he actual chance of overlap then will depend on how many of each of these types of amino acids is present in the protein, and perhaps on the kind of secondary structural element in which the residue is found. Secondary structure prediction of soluble proteins by new computer methods based only on the primary sequence (see, for example, www.igb.uci.edu/servers/psss. html), is already pretty good (ca. 80% accuracy)39 and subdividing the distributions according to the presumed secondary structure would lead to a more sophisticated labeling strategy than that suggested by the unimodal distributions shown in Figure 6. It is also feasible to predict disordered regions of the protein,40 in which the chemical shits will probably show less dispersion. However, these regions may also have missing NMR peaks altogether because the peaks may become quite wide and get lost in the baseline noise. In any case, once the probability distributions have been leshed out to the appropriate level of detail, the idea is for us to prepare the minimum number of diferent labeled protein samples that will, in all probability, result in a complete and unambiguous assignment of all the NMR peaks. Note that the 15N assignments alone are insuicient, so that 13C labels must also be introduced in an optimized way. hese labeling schedules41 represent a new area for research and further development, including the eventual complete automation42 of the entire assignment procedure. In addition, using N samples to perform the assignment, and the N + 1th to validate the assignment, provides an attractive way to safeguard against ambiguous or incorrect identiication of peaks.

Other Combinatorial Labeling Schemes

here are several recent examples in the literature that show the utility of combinatorial labeling strategies, in which certain subsets of amino acids are introduced with 15N or 13C enrichment at diferent levels. he scheme, dubbed combinatorial selective labeling (CSL)43 is an advance based on the previous idea of a dual amino acid selective labeling technique.44,45 In this earlier technique, all residues of type I are labeled with 13C, while those of type II are labeled with 15N. If one instance of a (I)II pair exists in the protein, then only a single cross-peak will appear with appreciable intensity in the 2D 15N-1H plane of a 3D HNCO spectrum. Obviously, making all the assignments would require a prohibitively large number of samples to be prepared. In CSL each of 16 amino acid types are labeled either 100% 13C/15N or 50% 15N/14N (and no carbon-13 enrichment), and a number of diferent samples are prepared. he choice of 16 of the 20 amino acids was made due to supply problems with histidine, typtophan and tyrosine, and the inability to label glutamate in the standard Roche RTS 500 system that was employed in the study.43 For each sample, two NMR spectra are obtained. he irst is a conventional 15N-1H 2D spectrum, which will show all the NH pairs in the sample. he second is a 2D 15N-1H plane from a 3D HNCO spectrum, which will show only those NH pairs at position i with a 13C label at the (i-1) position. For a truncated version of green luorescent protein, of 27 kDa size and with a long correlation time of ca. 21 ns, the preparation of ive samples using CSL and the analysis of ten normalized 2D spectra (300 hours of 500 MHz NMR time using a state-of-the-art cryoprobe) allowed 61 of the residues to be unambiguously assigned. While only a partial success, the low 100 μM concentration of each sample, and the limited chemical shit resolution provided at 500 MHz, would have made a conventional uniformly-labeled sample an impossible assignment task. hus, the partial success of CSL still pointed the way for further improvement. Depending on the particular information sought, and the availability of the requisite amino acids, many kinds of custom-labeling schemes could certainly be contemplated.

More Exotic Labeling Schemes

Amino acid type labeling should be considered routine in cell-free expression. However, chemical synthesis of the amino acids themselves can lead to very elaborate labeling patterns in the recombinant protein. Perhaps the best example, and a true tour de force demonstration of synthetic control is the Stereo-array isotopic labeling (SAIL) approach, which has been demonstrated on

Cell-Free Expression for Protein NMR

113

calmodulin (17 kDa) and maltodextrin binding protein (MBP, 41 kDa).46 Only one sample is prepared in SAIL, as described in reference 46, but each amino acid in the sample has been painstakingly synthesized to deliver the maximum available NMR information, particularly the distance constraints available from the NOESY spectrum. Figure 7 shows the chemical modiications that were made. hese amino acids, with speciic patterns of 2H/1H labeling as well as 13C/12C labeling, require multi-step syntheses to produce. Using SAIL, Kainosho et al were able to greatly improve the resolving power of the NMR spectra on calmodulin and MBP, allowing many more distance constraints to be used in the structure calculation. he results obtained compare favorably with the existing X-ray structures for these molecules.46

Figure 7. The SAIL labeling technique. Compared to conventional 13C labeling (indicated with an asterisk) of amino acid side chains (left) in which protons are not modified, in the SAIL technique (right) specific protons are substituted with deuterium, silencing these sites in the NOESY spectrum, narrowing the line widths of the resonances, and removing ambiguity about stereochemistry. The only stumbling block for the widespread adoption of SAIL is the huge cost and limited availability of the stereospecifically deuterated amino acids, each of which requires a nontrivial chemical synthesis.

114

Cell-Free Protein Expression

Speciic Information

Cell-free expression with amino acid type selective labeling can also be used as an adjunct to conventional cell-based expression to “tidy up” some loose assignments by preparing a few samples with very speciic labeling patterns in order to identify certain types, pairs, or triples of amino acids. For example, Shi et al47 showed that in vitro expression of four diferent labeled samples of a 23.5 kDa phosphoserine phosphatase (PSP) protein, each of which contained two doubly-labeled 15 N/13C amino acids and one 15N labeled amino acid could yield 135 type and 14 sequential assignments on this challenging example. Combined with the conventional uniformly-labeled sample, enough information was obtained to manually assign the backbone HN, 15N, 13C’, 13Cα, and 13Cβ signals. Using one automated assignment algorithm called AutoAssign,48 30% more assignments were obtained when the extra information from the speciically-labeled samples was included. While the automated assignment did not yield as much information as the manual assignment, and while some lower sensitivity NMR experiments were still required to get the information required, this work clearly shows the potential value of speciic information obtained by cell-free protein synthesis.

Future Prospects

Predicting the future of science is as tricky as predicting the weather: short-term predictions are pretty reliable but also rather obvious, whereas long-term predictions are both risky and speculative. Nevertheless, I’ll make a few observations of each kind. First, it is clear that cell-free protein expression is going to play an ever-increasing role in NMR structure determinations. he ability to simplify spectra, introduce novel isotope labeling patterns, obtain speciic assignments, and obtain proteins that are diicult to express in vivo due to cytotoxicity or puriication problems ensure that the in vitro approach will gather momentum in future. It seems that the overall success rate in producing suicient amounts of correctly folded protein is substantially higher using cell-free methods. In addition such samples oten are stable long enough for a full NMR structure determination. Another important and less explored aspect is simply to rethink, from the ground up, the entire strategy to make the assignments and obtain the NMR structure, by a collection of carefully thought-out speciically-labeled samples rather than a single diicult uniformly-labeled sample and some auxiliary samples with speciic labeling. Such an up-front approach might allow larger structures to be solved at lower magnetic ield strengths. As there are perhaps ten to iteen times more 500 MHz instruments than 800 MHz or higher instruments in operation currently, and as the cost of the lower-ield instruments is only a fraction of the higher-ield instruments, the higher cost of the custom extracts and labeled amino acids for in vitro expression can be amortized versus the huge diference in cost for acquisition and ongoing maintenance of the magnet, spectrometer and sometimes even a new building in which to house the huge magnet. As economies of scale kick in, NMR structures using speciic labeling by cell-free synthesis will be ever more attractive to determine using lower-ield spectrometers, freeing up the higher-ield instruments for truly state-of-the-art problems, and opening up protein structure determination to less well-versed users with interesting biological problems but less experience with NMR methodology itself. When uniform isotopic labeling was irst introduced a decade and a half ago, skeptics branded it as too costly, too time consuming, and excessive in terms of instrument time. Proton-only NMR was seen to be the cost-efective alternative that was proven, prudent, and conservative. Now there is no protein structure of any size solved by NMR that does not use labeling of some kind. his trend toward more elaborate sample preparation will only increase in the future as the cell-free technology is perfected further, packaged for turn-key use, and shown to speed up structure determination greatly or, perhaps, allow its automation altogether.

Acknowledgements

his work was supported by the UC Discovery Grant Program, and by the National Institutes of Health, GM-77673. he author wishes to warmly acknowledge help from his research group, and from Invitrogen.

Cell-Free Expression for Protein NMR

References

115

1. Wüthrich K. NMR of proteins and nucleic acids (Baker Lecture Series). New York: Wiley-Interscience 1986. 2. Cavanagh J, Fairbrother WJ, Palmer AG et al. Protein NMR spectroscopy: Principles and practice. San Diego: Academic Press 1996. 3. Rule, GS, Hitchens, TK. Fundamentals of protein NMR spectroscopy. Dordrecht: Springer 2006. 4. Bax A, Grzesiek, S. Methodological advances in protein NMR, Acc Chem Res 1993; 26:131-138. 5. Sattler M, Schleucher J, Grisinger C. Heteronuclear multidimensional NMR experiments for the structure determination of proteins in solution employing pulsed ield gradients, Prog NMR Spectr 1999; 34:93-158. 6. Ikura M, Kay LE, Bax A. A novel approach for sequential assignment of 1H, 13C, 15N spectra of larger proteins: Heteronuclear triple-resonance three-dimensional NMR spectroscopy. Application to calmodulin, Biochemistry 1990; 29:4659-4667. 7. Jeener J, Meier BH, Bachmann, P et al. Investigation of exchange processes by two-dimensional NMR spectroscopy, J Chem Phys 1979; 71:4546-4553. 8. Ikura M, Kay LE, Tschudin R et al. hree-dimensional NOESY-HMQC spectroscopy of a C-13 labeled protein, J Magn Reson 1990; 86:204-209. 9. Bax A, Grishaev A. Weak alignment NMR: a hawk-eyed view of biomolecular structure, Curr Opin Struct Biol 2005; 15:563-570. 10. Brunger AT, Adams PD, Rice LM. New applications of simulated annealing in X-ray crystallography and solution NMR, Structure 1997; 5:325-336. 11. Nilges M, Clore GM, Gronenborn AM. Determination of three dimensional structures of proteins from interproton distance data by dynamical simulated annealing from a random array of atoms, FEBS Lett 1988; 239:129-136. 12. Nilges M, Gronenborn AM, Brunger AT et al. Determination of three-dimensional structures of proteins by simulated annealing with interproton distance restreaints: application to crambin, potato carboxypeptidase inhibitor and barley serine proteinase inhibitor 2, Protein Engineering 1988; 2:27-38. 13. Bardiaus B, Malliavin TE, Nilges M et al. Comparison of diferent torsion angle approaches for NMR structure determination, J Biomol NMR 2006; 34:153-166. 14. Kay LE. NMR studies of protein structure and dynamics, J Magn Reson 2005; 173:193-207. 15. Kelley AE, Ou HD, Withers, R et al. Low-conductivity bufers for high-sensitivity NMR measurements, J Am Chem Soc 2002; 124:12013-12019. 16. Venters RA, Huang C-C, Farmer BT II et al. High-level 2H/13C/15N labeling of proteins for NMR studies, J Biomol NMR 1995; 5:339-344. 17. Marley J, Lu M, Bracken C. A method for eicient isotopic labeling of recombinant proteins, J Biomol NMR 2001; 20:71-75. 18. Löhr L, Katsemi, V, Hartleib J et al. A strategy to obtain backbone resonance assignments of deuterate proteins in the presence of incomplete amide 2H/1H back-exchange, J Biomol NMR 2003; 25:291-311. 19. Rosen MK, Gardner KH, Willis RC et al. Selective methyl group protonation of perdeuterated proteins, J Mol Biol 1996; 263:627-636. 20. Gardner KH, Rosen MK, Kay LE. Global folds of highly deuterated, methyl-protonated proteins by multidimensional NMR, Biochemistry 1997; 36:1389:1401. 21. Kudlicki W, Kramer G, Hardesty B. High eiciency cell-free synthesis of proteins: Reinement of the coupled transcription/translation system, Anal. Biochem. 1992; 206:389-393. 22. Katzen F, Chang G, Kudlicki W. he past, present and future of cell-free protein synthesis, Trends Biotechnol 2005; 23:150-156. 23. Nakano H, Kawarasaki Y, Yamane T. Cell-free protein synthesis systems: increasing their performance and applications, Adv Biochem Eng Biotechnol 2004; 90:135-149. 24. Endo Y, Sawasaki T. High-throughput, genome-scale protein production method based on the wheat germ cell-free expression system, Biotechnol Adv 2003; 21:695-713. 25. Vinarov DA, Markley JL. High-throughput automated platform for nuclear magnetic resonance-based structural proteomics, Expert Rev Proteomics 2005; 2:49-55. 26. Ozawa K, Dixon NE, Otting G. Cell-free synthesis of 15N-labeled proteins for NMR studies, IUBMB Life 2005; 57:615-622. 27. Tyler RC, Aceti DJ, Bingham CA et al. Comparison of cell-based and cell-free protocols for producing target proteins from the Arabidopsis thaliana genome for structural studies, Proteins 2005; 15:633-643. 28. Shi J, Pelton JG, Cho HS et al. Protein signal assignments using speciic labeling and cell-free synthesis, J Biomol NMR 2004; 28:235-247.

116

Cell-Free Protein Expression

29. Klammt C, Lohr F, Schafer B et al. High level cell-free expression and speciic labeling of integral membrane proteins, Eur J Biochem 2004; 271:568-580. 30. Klammt C, Schwarz D, Fendler K et al. Evaluation of detergents for the soluble expression of alpha-helical and beta-barrel-type integral membrane proteins by a preparative scale indicidual cell-free expression system, FEBS J 2005; 272:6024-6038. 31. Kainosho M, Torizawa T, Iwashita Y et al. Optimal isotope labelling for NMR protein structure determinations, Nature 2006; 45:2692-3702. 32. Keppetipola S, Kudlicki W, Nguyen BD et al. From gene to HSCQ in under ive hours: high-throughput NMR proteomics, J Am Chem Soc 2006; 128:4508-4509. 33. Muchmore CD, McIntosh LP, Russell CB et al. Expression and nitrogen-15 labeling of proteins for proton and nitrogen-15 nuclear magnetic resonance, Methods Enzymol 1989; 177:44-73. 34. McIntosh LP, Dahlquist FW. Biosynthetic incorporation of 15N and 13C for assignment and interpretation of nuclear magnetic resonance spectra of proteins, Q Rev Biophys 1990; 23:1-38. 35. Strauss A, Bitsch F, Cutting B et al. Amino-acid-type selective isotope labeling of proteins expressed in Baculovirus-infected insect cells useful for NMR studies, J Biomol NMR 2003; 26:367-372. 36. Chen CY, Cheng CH, Chen YC et al. Preparation of amino-acid-type selective isotope labeling of protein expressed in Pichia pastoris, Proteins 2006; 62:279-287. 37. Ozawa K, Headlam MJ, Schaefer PM et al. Optimization of an Escherichia coli system for cell-free synthesis of selectively 15N-labelled proteins for rapid analysis by NMR spectroscopy, Eur J Biochem 2004; 271:4084-4093. 38. Bayer P, Arndt A, Metzger S et al. Structure determination of the small ubiquitin-related modiier SUMO-1, J Mol Biol 1998; 280:275-286. 39. Cheng J, Randall AZ, Sweredoski, MJ et al. SCRATCH: a protein structure and structural feature prediction server, Nuclei Acids Research 2005; 33:W72-W76. 40. Cheng J, Sweredoski MJ, Baldi P. Accurate prediction of protein disordered regions by mining protein structure data, Data Mining and Knowledge Discovery 2005; 11:213-222. 41. Sweredoski MJ, Donovan KJ, Nguyen BD et al. PINOT NOIR: Integration and analysis of protein NMR data, J Biomol NMR 2006 (submitted). 42. Gronwald W, Kalbitzer HR. Automated structure determination of proteins by NMR spectroscopy, Prog NMR Spectrosc 2004; 44:33-96. 43. Parker MJ, Aulton-Jones M, Hounslow AM et al. A combinatorial selective labeling method for the assignment of backbone amide NMR resonances, J Am Chem Soc 2004; 126:5020-5021. 44. Kainosho M, Tsuji T. Assignment of the three methionyl carbonyl carbon resonances in Streptomyces subtilisin inhibitor by a C-13 and N-15 double-labeling technique: A new strategy for structural studies of proteins in solution, Biochemistry 1982; 21:6273-6279. 45. Yabuki T, Kigawa T, Dohmae N et al. Dual amino acid-selective and site-directed stable-isotope labeling of the human c-Ha-Ras protein by cell-free synthesis, J Biomol NMR 1998; 11:295-306. 46. Kainosho M, Torizawa T, Iwashita Y et al. Optimal isotope labelling for NMR protein structure determinations, Nature 2006; 440:52-57. 47. Shi J, Pelton JG, Cho HS et al. Protein signal assignments using speciic labling and cell-free synthesis, J Biomol NMR 2004; 28:235-247. 48. Zimmerman DE, Kulikowski CA, Huang CA et al. Automated analysis of protein NMR assignments using methods from artiicial intelligence, J Mol Biol 1997; 269:592-610.

Chapter 9

Cell-Free Synthesis of Membrane Proteins for X-Ray Crystallography Julia E. Fletcher, Federico Katzen, Wieslaw Kudlicki, Yen-Ju Chen, Andrew Chen, Samatha Lieu and Geofrey Chang*

Abstract

I

n this chapter we present a general protocol for producing small alpha helical membrane proteins in vitro suitable for X-ray crystallography. Increased protein solubility and highly eicient selenomethionine labeling makes this approach a convenient and relatively inexpensive method for producing membrane proteins, notoriously diicult to synthesize, purify and crystallize.

Introduction

Membrane proteins account for nearly a third of the genes encoded by most fully sequenced genomes. However, only a handful of integral membrane protein structures (~100) have been solved to high resolution. his disparity is largely a consequence of several hurdles, which include the requirement for large amounts (~100 mg) of these proteins for biophysical analysis such as X-ray crystallography, electron microscopy, or nuclear s magnetic resonance. his requirement has generally been diicult to meet by conventional in vivo expression techniques. Cell-free protein expression ofers a unique opportunity to circumvent those shortcomings (for a review see ref. 1). First, it allows expression of otherwise toxic proteins. Second, it provides direct access to the reaction milieu to modify the conditions. And inally, modiied amino acids that may be toxic such as selenomethionine (Se-met) or selenocysteine (Se-cys), can be used to eiciently label proteins for Multiwavelength Anomalous Difraction (MAD) phasing.2 Successful attempts to produce milligram quantities of integral membrane proteins using cell-free protein expression have been previously described3-7 but no 3D protein crystals were reported. In these cases the semi-continuous format was employed.8 Briely, the method makes use of a semi-permeable membrane that allows the continuous supply of substrates and removal of inhibitory byproducts from the reaction vessel. Despite the popularity and the eiciency of this method, all the advantages of the cell-free protein expression can be realized only by using the batch format, where access to the reaction compartment is not restricted. We have recently reported for the irst time an X-ray structure of a Se-met substituted integral membrane protein synthesized in vitro.9 Here we present a set of general conditions to synthesize small α-helical membrane proteins suitable for crystallization using a batch-formatted cell-free protein expression approach. *Corresponding Author: Geoffrey Chang—Department of Molecular Biology, The Scripps Research Institute, 10550 North Torrey Pines Road, CB-105, La Jolla, CA 92037, U.S.A. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

118

Cell-Free Protein Expression

Results Efects of Detergents and Lipids on the Cell-Free Protein Reaction

Previous attempts to improve yield, folding and solubility of membrane proteins using cell-free protein expression systems included the use of a variety of detergents (for example see).6 In general higher solubility was always achieved at the expense of the yield. In addition, the propensity of these proteins to form crystals and their ability to difract has never been studied. In our case, ater screening more than 230 detergents and lipids (some of them not commercial), there was no clear candidate that signiicantly enhanced the yield and solubility of integral membrane proteins in a batch-formatted cell-free expression system (not shown). We decided to investigate the efects of liposomes on the cell-free reactions. Unlike detergent micelles, liposomes consist of bi-layer membranes that closely mimic those found in living cells. Bicelles were produced using diferent ratios of 1,2-Dihexanoyl-sn-Glycero-3-Phosphocholine (DHCP) and 1,2-Dimyristoyl-sn-Glycero-3-Phosphocholine (DMPC) (see Materials and Methods). he best efects were observed with bicelles composed of pure DMPC at a concentration of 2 mg/ml in the batch-formatted cell-free expression system. Higher concentration of lipids in the cell-free reaction has a negative efect on cell-free synthesis for reasons that are not yet clear and does not improve the solubility of the protein. We found that this approach works well mainly for small alpha helical membrane proteins and chose two candidates to characterize further in preparative reactions: the multidrug resistance elux transporter EmrE from E. coli3,10 and the gated mechanosensitive ion channel MscL from Mycobacterium tuberculosis.11 We must mention that aspects of the system could be adapted to synthesize integral membrane proteins with higher molecular weights and refer the reader to a growing list of both eukaryotic and prokaryotic membrane protein being synthesized using the E. coli cell-free technique (for further details see Klammt, et al).12

Increased Solubility of α-Helical Membrane Proteins

Although previous eforts to synthesize small alpha-helical membrane proteins in vitro rendered milligram amounts of functional product,4,6,7 the samples were not homogenous enough for crystallization. We decided, thus, to investigate whether the quality of the proteins produced in the presence of liposomes was suitable for X-ray crystallography. he irst attribute we decided to analyze was protein solubility, and we set up cell-free reactions using EmrE and MscL as models. Reactions were performed in the presence or absence of 2 mg/ml DMPC liposomes, and products were trace labeled with [35S] Met. Reactions were processed as indicated in Materials and Methods. Using these conditions we obtained 0.25 mg/ul and 0.20 mg/ul of EmrE and MscL respectively (Fig. 1). Remarkably, the addition of 2 mg/ml DMPC liposomes to the cell-free reaction increased the solubility of EmrE and MscL from 3% to 19% and from 26% to 36% respectively (Fig. 1).

Efficient Labeling of Membrane Proteins with Se-Met

We examined the efect that Se-met had in cell-free protein reactions that already contain 2 mg/ml DMPC liposomes. Se-met labeling has been the method of choice to facilitate protein phasing by multi-wavelength dispersion. Still, it frequently leads to poorly substituted proteins in vivo owing to its cytotoxicity. Although cell-free systems have been proven permissive to Se-met labeling,2 labeling of integral membrane proteins in the presence of liposomes and further crystallization was not previously reported. We therefore set out to perform Se-met substitution in vitro using one of our alpha-helical membrane proteins of choice, EmrE. In vitro reactions were performed using Met or Se-met and 2 mg/ml DMPC liposomes as described in materials and methods. he results showed that the amount and solubility of Se-met-labeled EmrE was indistinguishable from the nonlabeled EmrE (Fig. 2, compare panels A and B). Total yield in both cases was approximately 0.2 mg/ml of puriied protein. he proteins were subjected to MALDI-TOF-MS to determine the relative incorporation of Se-met. An overlay of the spectra exhibited two peaks (m/z = 14299.03 and 14486.58), which

Cell-Free Synthesis of Membrane Proteins for X-Ray Crystallography

119

Figure 1. Effect of DMPC liposomes on the expression and solubility of membrane proteins. EmrE and MscL were expressed and processed as described in Materials and Methods. Numbers below the corresponding lanes correspond to the relative amount of soluble synthesized protein as determined by laser scanning (see Materials and Methods). S, soluble, T, total.

closely match the predicted mass of EmrE(Met) and EmrE(SeMet) (14295.80 and 14483.80 Da, respectively) (Fig. 2C). Considering the signal-to-noise ratio of the analysis, we concluded that the SeMet incorporation eiciency was higher than 99%. Importantly, we emphasize that the low reaction volume used by the batch method compared to the semi-continuous method reduces the amount of reagent necessary and subsequent cost for labeling by ten-fold.

Crystallization of EmrE

Finally, we used the sample obtained above in crystallization trials using similar conditions as those used for crystallizing EmrE produced in living cells. he success in obtaining EmrE crystals from a sample coming from a cell-free reaction suggested that the EmrE produced in vitro in the presence of liposomes should have similar physicochemical properties as that one obtained in vivo. Se-met EmrE-tetraphenyl-phosphonium (TPP) crystals tended to be even more anisotropic and difraction resolution was slightly lower. he weaker difraction is likely due to the presence the Se-met, which is commonly observed. Two crystal forms have been observed using the material derived from the cell free system. One in space group P21 with cell dimensions a = 75.5 Å, b = 44.2 Å, c = 114.1 Å, β = 109.9 Å as originally published9 and the other in C 2 space group with similar cell dimensions a = 75.4 Å, b = 44.3 Å, c = 114.1 Å, β = 109.9 Å. Both unit cells are related and the Se-met positions as derived from anomalous Fourier using the EmrE-TPP model and are nearly superimposable.

120

Cell-Free Protein Expression

Figure 2. In vitro Se-met labeling of EmrE. Se-met-EmrE (A) and Met-EmrE (B) were synthesized and purified as described in Materials and Methods. Aliquots of purification steps were run on 4-12% SDS-PAGE gels and visualized by Coomassie® staining. Lanes are as follows 1, MW markers; 2, whole cell-free reaction; 3, flow through; 4, third washing step; 5, 6, and 7, elutions 1, 2, and 3 respectively. Overlay of Se-met-EmrE and Met-EmrE mass spectrometry spectra. The uniform and dotted lines correspond to Met-EmrE and Se-met-EmrE respectively. Experimental conditions are described in Materials and Methods.

Concluding Remarks

In this chapter we have successfully demonstrated that, with minor modiications, a simple batch-formatted cell-free protein expression vessel is capable of synthesizing high amounts of small alpha-helical membrane proteins suitable for X-ray crystallography. No specialized equipment other than that usually present in an ordinary biochemistry laboratory is required for this procedure. his is particularly useful for those circumstances in which high throughput conditions should be applied in order to ine-tune the optimum conditions for the synthesis, puriication and crystallization of membrane proteins more diicult to produce and crystallize. Not long ago, cell-free protein expression was a tedious approach for the synthesis of nanogram levels of proteins, used only for functional studies. With the arrival of the proteomics era, the ield has experienced a technical resurgence expanding into a multitude of applications also covering structural proteomics. Now, for the irst time, cell-free protein expression has proven useful for the synthesis and crystallization of a membrane protein. Although a small step, this work sets the basis for further research that might lead to the design and development of new cell-free protein expression conditions involving perhaps novel surfactants with a wider range and larger membrane protein targets.

Cell-Free Synthesis of Membrane Proteins for X-Ray Crystallography

121

Figure 3. Crystals of Se-met-EmrE protein synthesized in a cell-free reaction. Average crystal dimensions are 0.1 mm x 0.1 mm x 0.15 mm.

Materials and Methods Plasmids

E. coli EmrE and M. tuberculosis MscL genes were amplified from pET15b-EmrE and peT15b-MscL respectively10,11 and cloned into pEXP5-NT/TOPO (Invitrogen, Carlsbad, CA). Linear DNA fragments used in cell-free protein expression reactions were obtained by PCR-ampliication using the same templates and included the T7-promoter, ribosome binding site, His-tag, thrombin cleavage site, ORF and T7 terminator sequences.

Preparation of Liposomes

Bicelles were prepared using diferent premixed ratios of DMPC/DHPC (Avanti Polar Lipids, Alabaster, AL) following the manufacturer’s protocols. Briely, lipids were resuspended in chloroform and dried under nitrogen in a glass tube. Chloroform was completely evacuated by vacuum for 10 minutes and distilled water was added. he mix was vortexed and sonicated until a transparent solution was achieved.

122

Cell-Free Protein Expression

In Vitro Protein Synthesis

Proteins were expressed using the Expressway™ Milligram Cell-Free E. coli. Expression system (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions except for the following modiications: (i) when indicated DMPC liposomes were added at a concentration of 2 mg/ml, (ii) when indicated 50 ul reactions were trace-labeled with 1 ul of [35S] Met (5 μCi, 1175 Ci/mmol) plus 1.25 mM of non radioactive Met, and (iii) circular or linear DNA template (0.5-1 μg) was employed. he reactions were performed in 1.5-2 ml micro centrifuge tubes in an Eppendorf hermomixer at 37°C with moderate shaking (1000-1400 rpm) for 2-6 hours. Reactions were fed with half volume feed bufer at two intervals over the reaction period. For Se-met labeling of EmrE, reactions were scaled up to 20 ml. In this case, 10-15ug of PCR-generated DNA, 4.5 mM Se-met (EMD Biosciences, San Diego, CA), 2 mg/ml DMPC liposomes (Avanti Polar Lipids, Alabaster, AL) and 1mM tetraphenyl-phosphonium (TPP, Sigma-Aldrich, St. Louis, MO) or tetraphenyl-arsonium (TPA, Sigma-Aldrich, St. Louis, MO) were added to the reaction. For determining solubility, a 1.25-ul aliquot of a [35S] Met trace-labeled reaction and a 1.25-ul aliquot of the cleared supernatant (10 min, 15000 xg) were run on a 4-12% SDS-PAGE and exposed to a Phosphorimager screen. Products were visualized and quantiied by laser scanning (Typhoon 8600 Variable Mode Imager, GE Healthcare, Piscataway, NJ) following the directions of the manufacturer.

Protein Puriication

Puriication of His-tagged proteins was performed using Ni-NTA resin by gravity low with (20 mM Tris pH 7.5, 200 mM NaCl pH 7.5, 0.05% Dodecyl-maltoside (DDM). To ensure complex formation, 1 mM of substrate (TPP or TPA) was added to all bufers used for puriication. he N-terminal His-tag leader sequence was removed by overnight thrombin digestion. he inal puriied construct used for crystallization trials contained Gly-Ser-His-Met in the N-terminus instead of the native Met. Protein identity was veriied to within 10 Da of expected by Matrix-Assisted Laser Desorption/Ionization Time-Of-Flight (MALDI-TOF) mass spectrometry.

Mass Spectroscopy Analysis

MALDI-TOF-MS analysis of intact proteins was performed using an Applied Biosystems Voyager DE STR instrument (Applied Biosystems, Foster City, CA). Analysis was performed with a constant laser intensity setting of 1998 in the linear mode. Samples were mixed 1:1 (v/v) with a saturated solution of Sinapinic acid (Sigma-Aldrich, St. Louis, MO) in 50% acetonitrile 0.1% Triluoroacetic Acid (TFA, Pierce, Rockford, IL). he samples were calibrated both internally and externally against Invitromass-IV and Invitromass-30kDa (Invitrogen, Carlsbad, CA).

Large Scale Se-Met Labeling and X-Ray Data Collection

Se-met EmrE was crystallized according to.9 In brief, proteins were synthesized as indicated above in the presence of TPP. he puriied protein was desalted using 20mM Tris, pH 8, 20 mM NaCl, 0.3% N-nonyl-α-D-glucoside (NG) (Anatrace, Maumee, OH), 1mM TPP and was concentrated to 15 mg/ml for crystallization. Crystallization trials were set-up in sitting drops by combining protein and well solutions at drop ratios of 1:1 to 3:1. he Se-met EmrE-TPP readily crystallized in conditions similar that of the wild-type: 11-14% Polyethelene Glycol (PEG) 2,000 Mono-Methyl-ether (MME), 100 mM Tris-HCl pH 6.4-6.6, 150 mM CaCl2, 0.35% NG.9 Data collection parameters and statistics for the Se-met crystal form were described previously.9

References

1. Katzen F, Chang G, Kudlicki W. he past, present and future of cell-free protein synthesis. Trends Biotechnol 2005; 23(3):150-156. 2. Kigawa T, Yamaguchi-Nunokawa E, Kodama K et al. Selenomethionine incorporation into a protein by cell-free synthesis. J Struct Funct Genomics 2002; 2(1):29-35. 3. Elbaz Y, Steiner-Mordoch S, Danieli T et al. In vitro synthesis of fully functional EmrE, a multidrug transporter, and study of its oligomeric state. Proc Natl Acad Sci USA 2004; 101(6):1519-1524. 4. Klammt C, Lohr F, Schafer B et al. High level cell-free expression and speciic labeling of integral membrane proteins. Eur J Biochem 2004; 271(3):568-580.

Cell-Free Synthesis of Membrane Proteins for X-Ray Crystallography

123

5. Ishihara G, Goto M, Saeki M et al. Expression of G protein coupled receptors in a cell-free translational system using detergents and thioredoxin-fusion vectors. Protein Expr Purif 2005; 41(1):27-37. 6. Klammt C, Schwarz D, Fendler K et al. Evaluation of detergents for the soluble expression of alpha-helical and beta-barrel-type integral membrane proteins by a preparative scale individual cell-free expression system. Febs J. 2005; 272(23):6024-6038. 7. Berrier C, Park KH, Abes S et al. Cell-free synthesis of a functional ion channel in the absence of a membrane and in the presence of detergent. Biochemistry 2004; 43(39):12585-12591. 8. Kim DM, Choi CY. A semicontinuous prokaryotic coupled transcription/translation system using a dialysis membrane. Biotechnol Prog 1996; 12(5):645-649. 9. Pornillos O, Chen YJ, Chen AP et al. X-ray structure of the EmrE multidrug transporter in complex with a substrate. Science 2005; 310(5756):1950-1953. 10. Ma C, Chang G. Structure of the multidrug resistance elux transporter EmrE from Escherichia coli. Proc Natl Acad Sci USA 2004; 101(9):2852-2857. 11. Chang G, Spencer RH, Lee AT et al. Structure of the MscL homolog from Mycobacterium tuberculosis: a gated mechanosensitive ion channel. Science 1998; 282(5397):2220-2226. 12. Klammt C, Schwarz D, Lohr F et al. Cell-free expression as an emerging technique for the large scale production of integral membrane protein. Febs J 2006; 273(18):4141-4153.

Chapter 10

Bacterial Cell-Free Expression Systems for High-hroughput Protein Production T.V.S. Murthy,* Leonardo Brizuela and Joshua LaBaer

Abstract

T

he combination of rapid protein expression systems with high-throughput protein isolation methodologies leads to increased throughput in protein production, improves eiciency and promotes multiplexed high-throughput experimentation. Such pipelines gain wide importance in proteomic applications, which may involve simultaneous puriication of several thousands of proteins/protein variants. hese optimized platforms beneit from simple and eicient worklows and a minimized number of steps. To address the need for soluble protein screening and to obtain microgram levels of proteins for discovery applications, investigators are turning towards processes that utilize bacterial cell-free protein expression systems coupled to high-throughput protein isolation platforms. he operational design used in these pipelines enables rapid proteome scale biological experimentation.

Introduction to High-hroughput Protein Production

Proteins are an important class of biological macromolecules that maintain the structural and functional integrity of the cell. Simultaneous study of the behavior of families of proteins enables us to understand the underlying principle of biological networks and such studies in combination with discovery assays/structural biology aid in the design of therapeutic agents (reviewed in refs. 1,2). Advances in recombinational genome cloning methods enabled production of the recombinant clone resources, an important raw material for protein production.3-9 In recent years, the high-throughput protein isolation platforms have primarily been used in biochemical assays, screening for soluble proteins to be used in structural studies, or to generate tools such as antibodies or protein arrays.8,10-14 he availability of complete or near complete sets of clones representing the open reading frames (ORFs) of various species has increased dramatically in the past ive years.5,9,15 In order to rapidly exploit the clone repositories for protein assay development, simple pipelines need to be developed in the downstream protein production and assay development methods as well. Two general strategies prevail here. Some laboratories utilize large libraries of clones in order to attempt proteomic studies with those proteins that can be easily expressed using simple and well established expression systems, ignoring any proteins that fail in the irst pass attempt (i.e., capturing the low hanging fruit). Other laboratories adopt a more directed approach towards obtaining speciic sets of proteins of interest even if this requires using multiple heterologous protein expression systems until all proteins of interest are obtained. It is expected that both of these approaches will comple*Corresponding Author: T.V.S. Murthy—Harvard Institute of Proteomics 320 Charles St., Cambridge, Massachusetts 02141, U.S.A. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

Bacterial Cell-Free Expression Systems for High-hroughput Protein Production

125

ment each other and drive progress in proteomic discovery applications. here is no universal heterologous protein expression system that succeeds in making all proteins or that produces proteins useful for all applications. Investigators oten must test multiple expression systems to ind one that meets the requirement of a speciic application. his may require testing diferent plasmid vectors, depending on the requirements of the expression system (e.g., bacteria vs. insect cell vs. mammalian cell). In a high-throughput context, the ability to rapidly shuttle the same expression clone repositories into more than one expression system can signiicantly maximize production of proteins, minimize the cost and labor involved in developing the expression clone resource and thereby may subsequently reduce the time involved in discovery eforts. Among the gamut of heterologous protein expression systems, the cell-free systems ofer an attractive alternative for heterologous high-throughput protein expression.

Cell-Free Systems for High-hroughput Protein Expression

High-throughput recombinant protein expression in cells has been attempted in various prokaryotic and eukaryotic expression systems.16-18 Among these systems, in vivo expression in bacteria gained popularity due to its relative simplicity and low cost compared to other systems.16,19-22 Although in vivo expression maintains wide importance for the high throughput expression of proteins, the multiple steps required in this worklow such as transformation, synchronization of cultures, addition of transcription inducing agents, cell lysis and centrifugation/iltration/separation steps, are time consuming and oten require manual intervention thereby reducing throughput. For some applications, simpler expression systems and pipelines are necessary to increase throughput and decrease error rate in proteome scale applications. Cell-free (in vitro) systems ofer a viable alternative and simplify the process of protein production. he three predominantly used in vitro protein expression systems are bacterial, rabbit reticulocyte and wheat germ systems. Other choices such as yeast, insect and mammalian cell-line derived cell-free systems show promise but have not yet proved their ability to express large numbers of proteins reproducibly or in suicient yields.23-25 In all these systems, the crude cell extracts used for protein synthesis primarily contain all the

macromolecular components (70S or 80S ribosomes, tRNAs, aminoacyl-tRNA synthetases, initiation, elongation, termination factors, etc.,) required for translation of exogenous RNA. To ensure eicient translation, the cell extract must be supplemented with optimal amounts of amino acids, energy sources (ATP, GTP), energy regenerating systems and other cofactors such as Mg2+, K+, etc.

Operationally, cell-free protein expression involves addition of template DNA imbued with the appropriate cis-acting transcriptional and translational elements to a cell extract containing the relevant RNA polymerase, ribosomal machinery and necessary substrates (tRNAs, amino acids, nucleotide triphosphates, etc.,) which then transcribe and translate the template to produce protein.26,27 An example of a typical schema for cell-free protein production is shown in Figure 1. he proteins synthesized in the cell-free system display the same accuracy as in in vivo translation.28 Due to their “open” nature, the cell-free systems are amenable to biochemical manipulation in order to probe the mechanistic aspects of protein synthesis or to speciically label newly translated proteins.29 he thorough understanding of the mechanism of protein synthesis led to the development of a cell-free translational system using reconstituted puriied components.30 In a cell-free system it is possible to alter the milieu in which the protein is synthesized and also modulate the biophysical properties of the proteins by extraneous addition of chemical agents during the actual process of protein synthesis.31 he in vitro translation systems are preferred over in vivo systems when rapid expression of proteins or protein variants is needed, when the over-expressed product is toxic to the host cell or when the protein undergoes rapid proteolytic degradation by intracellular proteases.32,33 Although it is possible to prepare cell extracts for protein synthesis from various cell types, in practice, only a few cell types, generally those that exhibit high transcription/translation capacity have been used to generate the cell-free protein expression systems. In general, E. coli BL21/A19 cells for bacterial, erythrocytes for rabbit reticulocyte and wheat embryos for wheat-germ system are used for preparation of stable cell extracts due to their rela-

126

Cell-Free Protein Expression

Figure 1. Making of the bacterial cell-free protein synthesis system. Panels A and B show a schema of the process of making the bacterial (S30) extract and the ingredients required for protein synthesis respectively. Panel C shows the process flow for 96-well protein production.

tively high protein synthetic capacity.29,34,35 he low yield of recombinant proteins from the rabbit reticulocyte system (~5 ng/50 μl reactions) and the need to sacriice rabbits has retracted its use in high-throughput protein production to some extent. he wheat germ and bacterial cell-free systems exhibit a capability to produce sub-milligram/ml levels of proteins for a wide range of applications.29,33,36

Bacterial Cell-Free System for High-hroughput Protein Production

he bacterial cell-free system is one of the best-studied systems and exhibits a capacity to rapidly express proteins with diverse properties from diferent species.37 his system has been reported to produce protein yields ranging from micrograms to milligrams, depending on the volumes of extract used, and to produce active protein of suicient quality to perform structural studies.38,39Eforts from several independent laboratories in the past few years led to design of protocols to generate highly synthetic bacterial extracts capable of producing several hundreds of micrograms of proteins per milliliter reaction volumes38,40 (reviewed in refs. 41-43). he cell-free protein synthesis enables incorporation of unnatural amino acid analogs and allows addition of detergents, chaperones, and appropriate ligands during protein synthesis to aid in proper folding of the proteins and shows promise for obtaining soluble and functional proteins44 (reviewed in refs. 37,45). Protein synthesis in the bacterial cell-free system can be scaled up and prolonged by incorporating a dialysis step during protein synthesis reaction, which simultaneously removes the toxic by products and regenerates/reintroduces the precursors required for protein synthesis.46 In our experience, an increase of at least three fold was observed by introducing dialysis during the protein expression when tested with one-milliliter reaction volumes (Fig. 2). he dialysis step is advantageous to maximize protein production for individual proteins in one-milliliter reaction volumes but we did not observe a signiicant improvement in protein yields with one hundred microliter reaction volumes in a high-throughput setting. Moreover, introducing dialysis in high-throughput protein production is expensive due to the cost of reagents used in cell-free protein production.

Bacterial Cell-Free Expression Systems for High-hroughput Protein Production

127

Figure 2. Effect of dialysis on synthesis of chloramphenicol acetyl transferase (CAT). CAT was expressed in 1 ml cell-free reactions with or without dialysis as indicated. 50 μl of the final reaction was subjected to one step purification and analyzed on a coomassie gel.

Production of proteins with multiple disulide bonds and expressing integral membrane proteins was also reported.47-49 Proteins, which were not expressed in the in vivo bacterial and insect cells, were also successfully expressed in the bacterial cell-free system.50 Moreover, the use of linearized fragment as a template (such as that obtained by PCR) in cell-free systems eliminates the need for cloning/sub cloning.51 Linear templates can be readily generated using a two-step PCR that introduces the necessary sequence elements for transcription/translation (Fig. 3). In cases where plasmids are used as templates, appropriate design of the promoter elements and selection markers enables use of the same expression clones in both the in vitro and in vivo systems, a feature of notable importance in high-throughput settings where high-throughput sub-cloning of genes into multiple expression vectors is tedious and expensive.52

Advances in Bacterial Cell-Free System

he ability of bacterial cell-free system to rapidly produce proteins in small and large-scale methods for discovery and structural proteomics prompted researchers to work towards further improving the eicacy of the cell-free protein synthesis and tailor it to production of speciic classes of proteins.31,53 Proteins that are prone to aggregation were successfully produced in soluble and active form in a modiied bacterial cell-free system which has the capacity to overexpress chaperones and disulide isomerase.54 Production of functionally active G protein coupled receptors in the bacterial cell-free system was also achieved by using the detergents, Brij35 and digitonin.55 A recent advancement towards improvement of protein expression was achieved by construction of

128

Cell-Free Protein Expression

Figure 3. Synthesis of Chloramphenicol acetyl transferase (CAT) and green fluorescent protein (GFP) from linear (PCR) and circular (Plasmid) templates. The linear template is generated by a two round PCR. The first round of PCR amplifies the gene and the second round of PCR uses the first round product as the template to introduce the 6HIS tag and T7 promoter elements. Panel A shows the schema to generate the linear template by two-step PCR. Agarose gel analysis of templates generated by first and second steps of the PCR is shown in panel B. The linear and circular templates were used in 50 μl cell-free batch reactions. 2 μl of the cell-free reaction mix after protein synthesis was analyzed by coomassie gel analysis (Panel C) or western blot (Panel D).

novel bacterial strains lacking the enzymatic activities that destabilize amino acids. In the study, the authors identiied the key enzymes responsible for degrading the amino acids, which lead to reduced production of the recombinant protein in the cell-free systems. Cell extracts prepared from strains lacking these enzymatic activities showed a signiicant increase in the yield of the recombinant protein.56 Enhanced protein synthetic capacity by introducing inorganic polyphosphate in the cell-free systems was also reported.57 Furthermore, use of novel energy systems for cell-free protein expression had a signiicant impact on the protein synthetic capacity.58-60 Despite these improvements, two major limitations of the cell-free systems are the relatively high cost and low yield of protein compared to the in vivo systems. Comparison of similar sets of proteins indicated that the ratio of recombinant protein to total protein was signiicantly lower in

Bacterial Cell-Free Expression Systems for High-hroughput Protein Production

129

cell-free systems as opposed to the cell-based systems.52 While microgram amounts of protein are easily obtained in the cell-free system for high expressers, some low expressers yielded nanogram amounts of protein, which reduced the apparent overall success rate in high-throughput operations. Moreover, due to low proportion of recombinant protein in the total reaction, high purity could not be achieved in single step high-throughput ainity puriication.52 Design of eicient strategies to reduce the cost of protein synthesis in bacterial cell-free systems was reported.61,62

Cell-Free Protein Expression in Our Laboratory

We have adapted the cell-free system for high-throughput protein production and partially automated the process by using liquid handling robotics. In our laboratory, we prepare cell extracts from three liter bacterial cultures which yield about eight milliliters of cell extract (S30 extract). Extract preparation is briely described in Murthy et al52 and is based on the method of Kigawa et al.29 We aliquot the extract into 96-well plates (12 μl per well) using a multi channel pipette, snap freeze in liquid nitrogen and store the plates at -80oC. To initiate protein expression, recombinant expression plasmids are mixed with a master mix of reagents (i.e., T7 polymerase, ribonucleotides, tRNAs, amino acids, etc.,) which is then added to the extract to produce recombinant protein whenever needed. he recombinant proteins are expressed with an ainity tag and subjected to one step ainity puriication as described in Murthy et al.52 Briely, the cell-free protein synthesis reaction containing the recombinant protein was diluted using the respective denaturing or nondenaturing puriication bufer and added to a 96-well Whatman GF/C plate aliquoted with prewashed Ni2+ NTA agarose beads for ainity puriication. he ainity puriication steps included binding of recombinant protein to the agarose beads, washes to remove the other cellular proteins and elution of the recombinant protein. An overall schematic worklow of the steps in cell-free protein production is shown in Figure 4A. he simple operational process increased the throughput and reduced the error rate. Proteins produced in this fashion in a 96-well protein production worklow are shown in Figure 4B. he yield of protein puriied in a 96-well format from 50 μl of cell-free reaction mixture varied from less than 100 ng to greater than 1 μg as determined by visual comparison with a known amount of bovine serum albumin (Fig. 4B).

Limitations and Potential Solutions

Like other heterologous protein expression systems, all the cell-free systems have several limitations. While the cell-free systems exhibit several advantages for certain applications, they are not a universal system for production of all proteins for all applications. his is more conspicuous in high-throughput applications where the proteome consists of several diverse types of proteins. One of the major deterring factors for the use of cell-free systems is the cost of the reaction. Since these reactions do not involve growing cells producing proteins, the reactions need to be supplemented with ingredients to continue protein synthesis. he accumulated cost of several reagents when multiplied across many reactions, as well as the cost of commercial extracts, can place limits on the availability of this approach to many labs. Fortunately, protocols for preparing bacterial cell-free system in an economical manner have been published by several laboratories.38,63 Although the preparation of cell-free protein synthesis system in a reproducible fashion requires some practice, we have found that yields corresponding to those published could be readily achieved and coupled to high-throughput applications. Since many of the reagents were prepared in-house, reaction cost was signiicantly reduced and we were also able to tailor the concentration of the components according to our requirement. he relatively low target protein/total protein ratio, however, remains a major concern for the cell-free systems.52 Despite optimization of the cell-free reaction by diferent groups, the yield of protein could be about a hundred fold less compared to in vivo systems (although an accurate comparison of the systems is diicult). he solubility of proteins expressed either in vivo or in vitro is comparable.52 Synthesis of truncated protein products, which may be due to transcriptional stalling or protein degradation, was also observed in the cell-free system.52,64 However, the truncated products can be eliminated during puriication by expressing the protein with a C-terminal tag.45

130

Cell-Free Protein Expression

Figure 4. A) Process of recombinant protein production using cell-free system. The cell extract after preparation is poured into a reagent reservoir and subsequently transferred into 96-well plates with the aid of liquid handling robots or multi channel pipettes (step 1). The plates are stored at -80 oC for future use. When required, the plates are thawed on ice and the necessary transcription/translation reagents are added along with recombinant plasmids to produce recombinant protein (step 2). The cell-extract is subjected to one-step affinity purification which includes binding of recombinant protein to the affinity matrix, washes to remove unbound proteins and elution of the recombinant protein (step 3). B) Coomassie gel analysis showing one-step 6xHIS affinity purification of thirty-two independent proteins expressed in cell-free system in 50 μl batch reaction. Twenty-six proteins were detected on the coomassie gel. A yield of 0.1 to 1 micrograms and a purity of 50 to 90% were observed. The putative protein bands are indicated by a dot on the left side of the band. The numbers below indicate the expected molecular weights.

Bacterial Cell-Free Expression Systems for High-hroughput Protein Production

131

Conclusion

Although no protein expression methodology is optimal for all circumstances, the bacterial cell-free system remains one of the best systems for rapid high-throughput protein expression. he knowledge gained in the art of extract preparation enables improvement in the yield of protein and tailors the system for speciic applications. he higher throughput, stability of the extracts, capacity to synthesize milligram/ml amounts of proteins and the ability to use the same protein expression vectors in both in vivo and in vitro settings are some of the major advantages of the bacterial cell-free system in high-throughput operations. Ready availability of cell-extracts to research laboratories at a reasonable price will obviate the necessity to use the in vivo system for many routine biochemical studies. However, the cell-based and cell-free systems are not mutually exclusive and parallel use of diferent expression systems should be considered wherever possible to obtain maximum number of proteins for discovery applications.

Reference

1. Bleicher KH, Bohm HJ, Muller K et al. Hit and lead generation: Beyond high-throughput screening. Nat Rev Drug Discov 2003; 5:369-378. 2. Han JD, Dupuy D, Bertin N et al. Efect of sampling on topology predictions of protein-protein interaction networks. Nat Biotechnol 2005; 23:839-844. 3. Dieckman L, Gu M, Stols L et al. High-throughput methods for gene cloning and expression. Protein Expr Pur 2002; 25:1-7. 4. Reboul J, Vaglio P, Rual JF et al. C elegans ORFeome version 1.1: experimental veriication of the genome annotation and resource for proteome-scale protein expression. Nat Genet 2003; 34:35-41. 5. Rual JF, Hirozane-Kishikawa T, Hao T et al. Human ORFeome version 1.1: a platform for reverse proteomics. Genome Res 2004; 10B:2128-2135. 6. Parrish JR, Limjindaporn T, Hines JA et al. High-throughput cloning of Campylobacter jejuni ORFs by in vivo recombination in Escherichia coli. J Proteome Res 2004; 3:582-586. 7. Dricot A et al. Generation of the Brucella melitensis ORFeome version 1.1. Genome Res 2004; 14: 2201-2206. 8. Park J, Hu Y, Murthy TV et al. Building a human kinase repository: bioinformatics, molecular cloning and functional validation. Proc Natl Acad Sci USA 2005; 102:8114-8119. 9. LaBaer J, Qiu JQ, Anumanthan A et al. he Pseudomonas aeruginosa PA01 gene collection. Genome Res 2004; 14:2190-2200. 10. Gilbert M, Edwards TC, Albala JS. Protein expression arrays for proteomics. Methods Mol Biol 2004; 264:15-23. 11. O’Toole N, Grabowski M, Otwinowski Z et al. he structural genomics experimental pipeline: insights from global target lists. Proteins 2004; 56:201-210. 12. Jung JW, Jung SH, Kim HS et al. High-throughput analysis of GST-fusion protein expression and activity dependent protein interactions on GST-fusion protein arrays with a spectral surface plasmon resonance biosensor. Proteomics 2006; 6:1110-1120. 13. Davies DH, Liang, X, Hernandez JE et al. Proiling the humoral immune response to infection by using proteome microarrays: high-throughput vaccine and diagnostic antigen discovery. Proc Natl Acad Sci USA 2005; 102:547-552. 14. Jones RB, Gordus A, Krall JA et al. quantitative protein interaction network for the ErbB receptors using protein microarrays. Nature 2006; 439:168-174. 15. Ecker A, Moon R, Sinden RE et al. Generation of gene targeting constructs for Plasmodium berghei by a PCR-based method amenable to high-throughput applications. Mol Biochem Parasitol 2006; 145:265-268. 16. Braun P, Hu Y, Shen B et al. Proteome-scale puriication of human proteins from bacteria. Proc Natl Acad Sci USA 2002; 99:2654-2659. 17. Chambers SP, Austen DA, Fulghum JR et al. High-throughput screening for soluble recombinant expressed kinases in Escherichia coli and insect cells. Protein Expr Purif 2004; 36:40-47. 18. Holz C, Lang C. High-hroughput Expression in Microplate Format in Saccharomyces cerevisiae. Methods Mol Biol 2004; 267:267-276. 19. Scheich C, Leitner D, Sievert V et al. Fast identiication of folded human protein domains expressed in E. coli suitable for structural analysis. BMC Struct Biol 2004; 4:4. 20. Vincentelli R, Canaan S, Ofant J et al. Automated expression and solubility screening of His tagged proteins in 96-well format. Anal Biochem 2005; 346:77-84.

132

Cell-Free Protein Expression

21. Lin CT, Moore PA, Auberry DL et al. Automated puriication of recombinant proteins: Combining high-throughput with high yield. Protein Expr Purif 2006; 47:16-24. 22. Steen J, Uhlen M, Hober S et al. High-throughput protein puriication using an automated set-up for high-yield ainity chromatography. Protein Expr Purif 2006; 46:173-178. 23. Katzen F, Kudlicki W. Eicient generation of insect-based cell-free translation extracts active in glycosylation and signal sequence processing. J Biotechnol 2006; 125:194-197. 24. Mikami S, Masutani M, Sonenberg N et al. An eicient mammalian cell-free translation system supplemented with translation factors. Protein Expr Purif 2006; 46:348-357. 25. Wang Z. Controlled expression of recombinant genes and preparation of cell-free extracts in yeast. Methods Mol Biol 2006; 313:317-331. 26. Devries JK, Zubay G. DNA-directed peptide synthesis. 3. Repression of beta-galactosidase synthesis and inhibition of repressor by inducer in a cell-free system. Proc Natl Acad Sci USA 1967; 58:1669-1675. 27. Zubay G. 1973. In vitro synthesis of protein in microbial systems. Annual Rev Genet 1973; 7:267-287. 28. Kurland CG. Translational accuracy in vitro. Cell 1982; 28:201-202. 29. Kigawa T, Muto Y, Yokoyama S. Cell-free synthesis and amino acid-selective stable isotope labeling of proteins for NMR analysis. J Biomol NMR 1995; 6:129-134. 30. Shimizu Y, Kanamori T, Ueda T. Protein synthesis by pure translational systems. Methods 2005; 36:299-304. 31. Boyer ME, Wang CW, Swartz JR. Simultaneous expression and maturation of the iron-sulfur protein ferredoxin in a cell-free system. Biotechnol Bioeng 2006; 94:128-138. 32. Golf SA, Goldberg AL. An increased content of protease La, the lone gene product, increases protein degradation and blocks growth in Escherichia coli. J Biol Chem 1987; 262:4508-4515. 33. Jermutus L, Ryabova LA, Pluckthun A. Recent advances in producing and selecting functional proteins by using cell-free system. Curr Opin Biotechnol 1998; 9:534-548. 34. Madin K, Sawasaki T, Ogasawara T et al. A highly eicient and robust cell-free protein synthesis system prepared from wheat embryos: Plants apparently contain a suicide system directed at ribosomes. Proc Natl Acad Sci USA 2000; 97:559-564. 35. Sawasaki T, Ogasawara T, Morishita R et al. A cell-free protein synthesis system for high-throughput proteomics. Proc Natl Acad Sci USA 2002; 99:14652-14657. 36. Endo Y, Sawasaki T. High-throughput genome-scale protein production method based on wheat germ cell-free protein expression system. J Struct Funct Genomics 2004; 5:45-57. 37. Betton JM. Rapid translation system (RTS): A promising alternative for recombinant protein production. Current Protein and Peptide Sci, 2003; 4:73-80. 38. Kigawa T, Yabuki T, Yoshida Y et al. Cell-free production and stable-isotope labeling of milligram quantities of proteins. FEBS Lett 1999; 442:15-19. 39. Swartz JR. Advances in Escherichia coli production of therapeutic proteins. Curr Opin Biotechnol 2001; 12:195-201. 40. Kim D, Swartz J. Prolonging cell-free protein synthesis with a novel ATP regeneration system. Biotechnol Bioengg 1999; 66:180-188. 41. Spirin AS. High-throughput cell-free systems for synthesis of functionally active proteins. Trends Biotechnol 2004; 22:538-545. 42. Voloshin AM, Swartz JR. Eicient and scalable method for scaling up cell free protein synthesis in batch mode. Biotechnol Bioeng 2005; 91:516-521. 43. Katzen F, Chang G, Kudlicki W. he past, present and future of cell-free protein synthesis. he past, present and future of cell-free protein synthesis. Trends Biotechnol 2005; 23:150-156. 44. Hirao I, Ohtsuki T, Fujiwara T et al. Nat Biotechnol 2002; 20:177-182. 45. Busso D, Kim R, Kim SH. Expression of soluble recombinant proteins in a cell-free system using a 96-well format. J Biochem Biophys Methods 2003; 55:233-240. 46. Chekulayeva MN, Kurnasov OV, Shirokov VA et al. Continuous-exchange cell-free protein-synthesis system. Synthesis of HIV 1 antigen Nef. Biochem Biophys Res Commun 2001; 280:914-917. 47. Kim DM, Swartz JR. Eicient production of a bioactive, multiple disulide-bonded protein using modiied extracts of Escherichia coli. Biotechnol Bioengg 2003; 85:122-129. 48. Yin G, Swartz JR. Enhancing multiple disulide-bonded protein folding in a cell-free system. Biotechnol Bioeng 2004; 86:188-195. 49. Klammt C, Lohr F, Schafer B et al. High level cell-free expression and speciic labeling of integral membrane proteins. Eur J Biochem 2004; 271:568-580. 50. Renesto P, Raoult D. From genes to proteins: in vitro expression of Rickettsial proteins. Ann N Y Acad Sci 2003; 990:642-652.

Bacterial Cell-Free Expression Systems for High-hroughput Protein Production

133

51. Lesley SA, Brow MA, Burgess RR. Use of in vitro protein synthesis from polymerase chain reaction-generated templates to study interaction of Escherichia coli transcription factors with core RNA polymerase and for epitope mapping of monoclonal antibodies. J Biol Chem 1991; 266:2632-2638. 52. Murthy TV, Wu W, Qiu QQ et al. Bacterial cell-free system for high-throughput protein expression and a comparative analysis of Escherichia coli cell-free and whole cell expression systems. Protein Expr Purif 2004; 36:217-225. 53. Knapp KG, Swartz JR. Cell-free production of active E. coli thioredoxin reductase and glutathione reductase. FEBS Lett 2004; 559:66-70. 54. Kang SH, Kim DM, Kim HJ et al. Cell-free production of aggregation-prone proteins in soluble and active forms. Biotechnol Prog 2005; 21:1412-1419. 55. Ishihara G, Goto M, Saeki M et al. Expression of G protein coupled receptors in a cell-free translational system using detergents and thioredoxin-fusion vectors. Protein Expr Purif 2005; 41:27-37. 56. Calhoun KA, Swartz JR. Total amino acid stabilization during cell-free protein synthesis reactions. J Biotechnol 2006; 123:193-203. 57. Itoh H, Kawazoe Y, Shiba T. Enhancement of protein synthesis by inorganic polyphosphate in an E. coli cell-free system. J Microbiol Methods 2006; 64:241-249. 58. Sitaraman K, Esposito D, Klarmann G et al. A novel cell-free protein synthesis system. J Biotechnol 2004; 110:257-263. 59. Calhoun KA, Swartz JR. An economical method for cell-free protein synthesis using glucose and nucleoside monophosphates. Biotechnol Prog 2005; 21:1146-1153. 60. Kuem JW, Kim TW, Park CG et al. Oxalate enhances protein synthesis in cell-free synthesis system utilizing 3-phosphoglycerate as energy source. J Biosci Bioeng 2006; 101:162-165. 61. Calhoun KA, Swartz JR. Energizing cell-free protein synthesis with glucose metabolism. Biotechnol Bioeng 2005; 90:606-613. 62. Liu DV, Zawada JF, Swartz JR. Streamlining Escherichia coli S30 extract preparation for economical cell-free protein synthesis. Biotechnol Prog 2005; 21:460-465. 63. Swartz JR, Jewett MC, Woodrow KA. Cell-free protein synthesis with prokaryotic combined transcription-translation. Methods Mol Biol 2004; 267:169-182. 64. Ramachandiran V, Kramer G, Hardesty B. Expression of diferent coding sequences in cell-free bacterial and eukaryotic systems indicates translational pausing on Escherichia coli ribosomes. FEBS Lett 2000; 482:185-188.

Chapter 11

Cell-Free Protein Synthesis for Protein Microarrays Gregory A. Michaud,* Michael Salcius, Rebecca Martone, Diane Buhr, John C. Duarte, Jennifer E. McCague, Xiangdong Liu, Michael Samuels, Christine Stalder, James Ball, Alex Tikhonov, Shiranthi Keppetipola, Wieslaw Kudlicki, James Meegan and Barry I. Schweitzer

Introduction

P

rotein microarray technology has empowered investigators with a tool that ofers the potential to study the function of thousands of proteins in a single experiment. he approach typically involves a process by which proteins are tethered to either a modiied microscope slide or to the bottom of a well in a multi-well plate. Numerous applications have been developed that measure interactions, both covalent and noncovalent, of a probe with target proteins on the array. Almost every type of interaction observed in biomolecular signaling pathways has been reconstituted on protein arrays, including interactions with small molecules, lipids, proteins, DNA, RNA, carbohydrate and enzymes.1,2 One of the most common questions related to adoption of the technology as a complementary platform for research and development is, “How will protein microarray technology bring value to my research?” One answer to this question is that protein microarray technology can greatly accelerate the speed at which new information can be generated. Standard approaches to investigating protein function can take months to years to address the function of just one protein. In contrast, a standard protein microarray experiment can be performed in as little as 5 hours. Because the technology enables such rapid screening, the cost savings in materials and labor can be signiicant. Secondly, screens on protein microarrays can be comprehensive in that an entire proteome can be proiled in a single experiment which is generally not achievable by alternate proteomic technologies. Below, we highlight some recent examples of how protein microarray technology has been applied to further understand the biology of complex systems. In this discussion, we will focus on the application of cell-free protein synthesis and how it has been applied in protein microarray experiments.

Protein Content

he hallmark of protein microarrays is the presentation of peptides or proteins typically in very high density on a solid state support. he source of the proteins is usually determined by the organism of interest that is being studied. Zhu et al used protein microarrays containing almost every protein from the yeast Saccharomycees cerevisiae to investigate proteome wide interactions with proteins, lipids and DNA.3 Synthesized peptide and full-length protein arrays have been used to characterize protein kinases from mammalian and yeast cells by identifying consensus motifs for sites of protein phosphoryation and through the discovery of novel protein kinase substrates.4-6 *Corresponding Author: Gregory A. Michaud—Protein Array Center, Branford, Connecticut 06405, U.S.A. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

Cell-Free Protein Synthesis for Protein Microarrays

135

he ability to produce these types of arrays typically requires extensive support from several groups (Bioinformatics, Manufacturing, Research and Development and Quality Control) within an organization and, therefore, the time and cost commitment for generating high-content protein arrays can be quite high. For this reason, any approach that simpliies the process of protein array manufacturing has the potential to decrease costs and ultimately make the technology more accessible to researchers. Generally, three methods have been used to produce protein content for protein arrays. he irst and perhaps the most widely used method is cell-based protein production. In this approach, peptides and proteins, which are very oten tagged to facilitate protein puriication, are expressed in either prokaryotic or eukaryotic expression systems. he cells then serve as a source to produce a lysate that can be directly applied to the surface of the array or subjected to an enrichment process to purify the protein(s) of interest prior to immobilization. he direct application of lysates to protein arrays has been quite useful for proiling the abundance or state of post-translational modiication between two states, typically normal and disease/and/or treated. By probing these so-called “reverse-phase” protein arrays with phosphospeciic antibodies, new insights into the dysregulation of phosphoprotein signaling have been obtained for several diseases, including but not limited to prostate cancer, ovarian carcinoma and glioma.7-9 he chief disadvantage of reverse-phase protein microarray technology is that the approach is highly dependent on prior knowledge of which proteins to proile and consequently on the availability of speciic antibodies that exhibit little or no cross-reactivity. Similarly, if reverse phase protein arrays are proiled with serum to identify an autoantibody response that, for example, is enhanced in disease relative to a normal state, the identiication of the proteins that produce the elevated signal requires several downstream steps such as fractionation and mass spectrometry.10 In most commercially available protein microarrays, the proteins are puriied prior to immobilization on the array. Several groups have investigated a number of approaches to optimize the production of proteins that are of high purity and activity.11,12 he hexahistidine tag (His6) and glutathione-S-tranferase (GST) are the most widely used tags to facilitate protein puriication due to the simplicity of the protocols and their validation in high throughput processes.13,14 An alternative approach involves a machine-directed synthesis of peptides that can be directly attached to protein arrays. In a recent paper by Anderson et al synthesized biotinylated peptides were coupled to NeutrAvidin (NA) and the biotinylated peptide-NA complexes spotted directly on to a surface activated glass slide.15 hese peptide arrays were used to show very sensitive and speciic detection of antibody-antigen interactions. Peptide arrays have also been very useful for studying the functions of other enzymes such as protein kinases, proteases and ubiqutin ligases.16-18 Some shortcomings of peptide array technology are the fact that the current state of the art of peptide synthesis limits this approach to peptides of typically less than 20 amino acids. hus only very small domains of proteins can be analyzed. Secondly, although motif analysis has been very powerful for explaining receptor-ligand interactions, the technology currently has limited ability to predict biological signiicance due to the nature that many sequences predicted to interact with ligands are in fact not observed.19 A third method for producing content for protein arrays involves cell-free synthesis of proteins. He et al were the irst to develop a process that involves DNA-directed coupled transcription-translation of proteins and subsequent immobilization on the surface (PISA, protein in situ array).20 Ramachandran et al have used a similar approach that involves the capture of GST-tagged proteins by an anti-GST antibody on the surface.21,22 By coexpressing probe proteins with the targets proteins, pairwise interactions among several human DNA replication initiation proteins were mapped.21 he disadvantages of protein in situ arrays for protein-protein interacting proiling are that protein pairs must be coexpressed. As such, proteome mapping studies will be complicated by the fact that N (prey) X N (target) coexpression pairs must be prepared to perform comprehensive studies. Secondly, the anticipated variability in protein production by cell-free synthesis within and across protein in situ arrays is likely to present serious complications for data analysis for applications such as biomarker-based discovery assays that aim to develop group classiiers.23

136

Cell-Free Protein Expression

Cell-free protein synthesis of proteins that are puriied prior to being spotted have also been used to produce functional protein arrays.24,25 Jung et al were able to reconstitute a pathway for trehalose synthesis using RNA-protein fusion (PROfusion) technology.26 Recently, Davies et al described the development of a proteome-scale poxvirus microarray.27 In this report, 185 out of the 273 proteins encoded by the vaccinia genome were expressed from PCR products in an Escherichia coli-based cell-free in vitro transcription/translation system, and the crude reactions containing expressed proteins were printed directly onto microarrays without puriication. he chips were used to determine antibody proiles in serum from vaccinia virus-immunized humans, primates, and mice. Although this study demonstrated the potential of this technology to comprehensively scan humoral immunity from vaccinated or infected humans and animals, it sufered from a number of drawbacks including the lack of quality control (e.g., DNA sequencing or Western blotting) on the cloned genes or expressed proteins, the use of nonpuriied proteins, and the use of a bacterial host to express proteins from a nonbacterial organism. Cell-free synthesis of proteins requires that all the components for transcription and translation be present. Lysates from E. coli, rabbit reticulocytes and wheat germ are most commonly used systems for cell-free protein production. he choice of system for protein production certainly depends on the downstream application. In general, E. coli produces more protein (up to a milligram of protein/ml of reaction) than observed with either wheat germ or rabbit reticulocyte lysates (1-10 ug/ml of reaction).28 For a more in depth review of the advantages and disadvantages of each system for protein production see Katzen et al.28 Below, we describe how cell free synthesis has been applied to the production of a protein array containing thousands of proteins for the bacterial pathogen, Yersinia pestis.

Yersinia pestis Protein Arrays

he plague-causing bacterium Yersinia pestis (Y. pestis) is capable of producing wide-spread illness and terror if employed as a weapon of mass destruction. Although current therapies for this agent possess some eicacy, there remain signiicant issues with availability of resources, toxicity of therapies, and the potential for terrorists to employ or develop agents highly resistant to available therapeutic agents. Current research eforts into improved detection and therapy strategies involve the application of information from the sequencing of the genomes of these agents and the molecular biological detailing of the mechanisms by which they produce human illness. A major challenge in developing efective vaccines or selecting antigens for diagnostic purposes is to identify the particular antigen from the pathogen that will be most efective. his is especially problematic for bacterial pathogens, which usually have large proteomes; obviously, it would not be practical to test thousands of potential antigens for an organism such as Y. pestis. Unfortunately, no algorithm exists that can use protein sequence data to accurately predict which antigens will elicit the most efective immune response.27 To evaluate if cell free in vitro protein production would be suicient to produce Y. pestis protein arrays, we irst compared the quality of proteins obtained using optimized protocols for both E. coli cell-based and cell-free protein production. Several Y. pestis proteins were cloned into an expression vector which produces proteins with a N-terminal fusion to glutathione-S-transferase. To express the proteins in cells, the constructs were transformed into BL21 AI (Invitrogen) cells and induced at 30°C for four hours. To express the proteins in vitro, approximately 1 μg of plasmid DNA was incubated with an E. coli lysate (Expressway™, Invitrogen, Carlsbad, CA) optimized for coupled transcription-translation (Fig. 1) at 30°C for 6 hours. he proteins from each expression condition were puriied in parallel using glutathione-ainity chromatography. We measured the yield and purity of the proteins obtained and did not observe signiicant diferences with the exception that there were slightly elevated levels of a copuriied protein migrating at the molecular weight of GST for some proteins expressed with Expressway™ (Fig. 2). Considering the possibility that some Y. pestis proteins may be diicult to produce in cells

Cell-Free Protein Synthesis for Protein Microarrays

137

Figure 1. Expressway™. As a DNA template driven by a T7 promoter is transcribed, the 5’ end of the mRNA is bound by ribosomes and undergoes translation. A specifically engineered ATP energy renewal system coupled with early ribosome binding for transcript stability results in high protein yields.

and the considerable time saved by using cell-free protein production, we utilized Expressway™ to express 3358 proteins from the pathogen. Of these 3358 proteins, 2727 (81%) proteins were obtained that displayed the expected molecular weight ater puriication. Interestingly, when the distribution of proteins that were of the expected molecular weight were analyzed relative to their predicted size, we observed that the success rate for producing longer proteins decreased signiicantly as the predicted molecular weight increased (Fig. 3). For proteins that were in the 20-80 kilodalton molecular weight range, a 90% success rate (2970/3768) was observed. In contrast, the success rate dropped to 59% (265/451) for the 80-110 kilodalton range and to 36% (43/119) for the 110-140 range (Fig. 3). he puriied proteins were transferred to 384 well plates and arrayed (OmniGrid, Genomic Solutions) on glass slides to produce protein arrays containing 2727 proteins representing approximately 67% of the proteome. An image of a Y. pestis proteome array is shown in Figure 4.

138

Cell-Free Protein Expression

Figure 2. In vitro versus in vivo protein production. Twelve Y. pestis proteins (Y0066, Y0020, Y0021, Y0013, Y0078, Y0082, Y0030, Y0049, y1029, Y0027, Y1046, Y0043) were produced using optimal protocols for in vitro (Expressway™) versus in vivo (BL21 AI) protein production. Purified proteins were electrophoresed in NuPAGE® and stained with Simply Blue SafeStain. Odd lanes (in vivo), even lanes (in vitro), molecular weight standard (ST) and purified glutathione-S-transferse, GST (C).

Figure 3. Y. pestis protein production yield. The percentage of proteins successfully purified at the correct molecular weight are plotted relative to their predicted molecular weight. Four bins of proteins in the molecular weight ranges of 20-50N = 1283, 50-80N = 1687, 80-110N = 265, 110-140N = 43

Cell-Free Protein Synthesis for Protein Microarrays Figure 4. A fluorescent image of a microarray printed with 2727 Y. pestis proteins. Proteins were visualized with a rabbit anti-GST antibody and AlexaFluor647-labeled anti-rabbit antibody.

he image results from staining the array with an anti-GST antibody followed by detection using a secondary antibody conjugated to a luorescent dye. Currently, we are testing the utility of Y. pestis proteome arrays for measuring protein-protein interactions and also have developed and optimized a protocol for proiling the antibody immune response. he type III secretion system of Y. pestis is essential for virulence and forms an injectosome that delivers virulence factors, or Yops for Yersinia outer proteins (i.e., YopH), into the cytosol of mammalian cells.29 his process requires an interaction of YopH with the SycH chaperone. As shown in Figure 5, when SycH was expressed as a GST-fusion, ainity-puriied, biotinylated, and used as a probe, the expected interaction with YopH on the array was observed (Fig. 5B). To further address the quality of proteins produced using an in vitro E. coli transcription-translation system, experiments were carried out in which two proteins were coexpressed in the same Expressway ™ reaction and then immunoprecipitated.30 Two protein-protein interaction pairs from the yeast Saccharomycees cerevisiae were chosen for these experiments. The yeast proteins, Tif34 and Tif35, proteins in the core complex of the translation initiation factor 3 (eIF3) that have been shown to interact using yeast two-hybrid, mass spectroscopy and on protein arrays (Michaud et al, unpublished results).31-33 A histidine-tagged fusion of Tif34 was coexpressed using Expressway ™ with GST-tagged Tif35. Tif34 was purified using nickel-chromatography and copurification of Tif35 was demonstrated using western analysis (Fig. 6B, lane 1). The Ran-interacting protein (Mog1) and the GTP-binding protein

139

140

Cell-Free Protein Expression

Figure 5. Protein-protein interaction probing on Y. pestis array. A) Subarray image from array probed with Alexa-647-labeled streptavidin. B) Subarray image from array probed with biotinylated SycH. Expected interaction with YopH (green box) is observed. A biotinylated antibody (bio-Ab) control for detection that is bound to the detection reagent, streptavidin-AlexaFluor647 conjugate (red box). A color version of this figure is available online at www.eurekah.com.

(Gsp2) are proteins that form a complex that is important for the maintenance of nuclear organization, RNA processing and transport.34 The interaction of these two proteins has been confirmed on protein arrays using protein probes produced in E. coli cells (data not shown). We observed that Mog1 indeed forms a complex with Gsp2 by in vitro coexpression-affinity purification (Fig. 6B, lane 2). Tif35, which is not expected to form a complex with Mog1, does not copurify with Mog1 (Fig. 6B, lane 3). These experiments demonstrate that functional eukaryotic proteins can be expressed using E. coli cell-free protein synthesis. he Y. pestis proteins, F1 and V, either alone or in combination, have been efective vaccines in animal models.35-37 Highly immunogenic Y. pestis proteins that have yet to be characterized represent potential vaccines and/or reagents valuable for diagnosis. Protein arrays containing 149 Y. pestis proteins have been developed by Li et al and proiled with serum from rabbits exposed

Figure 6. Coexpression-affinity purification assay. Four yeast proteins (Tif35, Tif34, Gsp2, and Mog1) were expressed in vitro using Expressway™. Tif35 and Gsp2 are GST-tagged. Tif34 and Mog1 are his-tagged. The following coexpression pairs were tested: lane 1 (Tif34-Tif35), lane 2 (Mog1-Gsp2) and lane 3 (Mog1-Tif35). His-tagged protein complexes were purified by nickel chromatography, electrophoresed in SDS-page gels, western blotted and stained with either the general protein stain, Ponceau S (A) or with an anti-GSTrabbit (B) antibody/anti-rabbit HRP conjugate.

Cell-Free Protein Synthesis for Protein Microarrays

141

to a live Y. pestis vaccine.38 Although the experiments were performed with only a small subset of the Y. pestis proteome, the study identiied several immunodominant antigens such as the outer membrane proteins OmpA and MlpA that have the potential to be protective antigens for prevention of disease. To address the utility of Y. pestis proteome arrays for immune response in proiling humans, Y. pestis proteome arrays were probed with human serum at the standard dilution used to probe arrays of human proteins produced in insect cells (ProtoArray, Invitrogen, Carlsbad, CA).39 In these initial experiments, signiicant signals were observed with most of the Y. pestis protein spots on the array (Fig. 7A,B). his reactivity, which is not observed when proteins are puriied from insect cells (data not shown), is likely due to trace amounts of E. coli proteins that copurify with the Y. pestis proteins. To eliminate this human antibody reactivity towards E. coli proteins, we optimized the serum proiling protocol by performing the assay in the presence of E. coli lysate used in the in vitro expression reaction in an efort to titrate the human antibodies against E. coli proteins. As shown in Figure 7A, a signiicant reduction in background reactivity was observed at all concentrations tested (0.1% to 20%). he performance of the antibody proiling application in the presence of E. coli lysate was also assessed for known auto antibody-antigen pairs using ELISA and protein array qualiied sera. From these studies, we concluded that the addition of E. coli extract during the probing did not afect the reactivity of auto antibodies with protein antigens (data not shown). As a result of this work, a standard operating protocol was deined using 1% E. coli extract, which lowers the nonspeciic signals for the Y. pestis protein spots approximately 12 fold (Fig. 6C/D) while leaving true auto antibody signals virtually unaltered (data not shown).

Conclusions

he application of cell-free protein synthesis to protein microarray technology has several beneits. he irst is the speed at which proteins can be produced. We have expressed and puriied thousands of proteins in less than 8 hours, as compared to approximately 5 days required to produce proteins using a standard cell-based process. As a result, use of cell-free methods for protein production can result in dramatic reductions in cost mainly through savings in labor. A second beneit is the ability to produce proteins that are diicult to express in cells due to toxicity. Lastly, when cell-free protein synthesis is coupled to protein array production, any potential instability of proteins on the surface can be mitigated by producing the proteins directly before experimentation. Cell-based methods for protein production have been the mainstay for many years because several cellular processes ensure that correct protein folding and post-translation modiications occur for protein activity. A question that is still yet to be answered is the robustness of cell-free protein synthesis systems for producing proteins for functional assays on protein arrays. Can cell-free synthesis produce proteins of suicient function, yield and purity? Sawasaki et al demonstrated using wheat germ cell-free protein synthesis that several hundred protein kinases from Arabidopsis are in fact functional in an autophosphorylation assay.40 In addition, several of these protein kinases phosphorylate an expected peptide substrate. Likewise, several protein-protein interaction pairs have also been reconstituted using both wheat germ and rabbit reticulolysate cell-free production.41,21 Cell-free protein synthesis has also been applied for structural analysis of the small multidrug transporter (EmrE) from E. coli.30,42 It seems clear, therefore, that cell-free protein production can certainly generate proteins that retain both structure and function. However, only more experimentation will determine if cell-free protein synthesis will supplant cell-based methods for producing proteins for functional studies on protein arrays.

142

Cell-Free Protein Expression

Figure 7. Auto-antibody profiling on Y. pestis protein arrays. A) Optimization of amount of Expressway™ extract for background minimization. Protein arrays were probed with serum in the presence of E. coli extract at several concentrations (0.1, 0.25, 0.5, 1, 2.5, 5, 10, 20%). The average fluorescence intensity (y-axis), with standard deviations, for the Y. pestis proteins is plotted as a function of extract concentration (x-axis) B) Subarray image from array probed with human serum (1:150). C) Subarray image from array probed with human serum (1:150) in the presence of 1% E. coli extract (Expressway™). IgG antibodies bound to proteins on the array were detected with an anti-IgG antibody AlexaFluor647™ conjugate. In (B) and (C), the features for the Y. pestis proteins are boxed in white. D) The fluorescence intensity of the arrays (B) and (C) were quantified and plotted as a frequency (number of proteins) relative to fluorescence intensity.

Cell-Free Protein Synthesis for Protein Microarrays

References

143

1. Predki PF. Functional protein microarrays: ripe for discovery. Curr Opin Chem Biol 2004; 8(1):8-13. 2. Michaud GA, Samuels ML, Schweitzer B. Functional protein arrays to facilitate drug discovery and development. IDrugs 2006; 9(4):266-272. 3. Zhu H, Klemic JF, Chang S et al. Analysis of yeast protein kinases using protein chips. Nat Genet 2000; 26(3):283-289. 4. Turk BE. Measuring kinase activity: inding needles in a haystack. Nat Methods 2005; 2(4):251-252. 5. Ptacek J, Devgan G, Michaud G et al. Global analysis of protein phosphorylation in yeast. Nature 2005; 438(7068):679-684. 6. Mah AS, Elia AE, Devgan G et al. Substrate speciicity analysis of protein kinase complex Dbf2-Mob1 by peptide library and proteome array screening. BMC Biochem 2005; 6:22. 7. Jiang R, Mircean C, Shmulevich I et al. Pathway alterations during glioma progression revealed by reverse phase protein lysate arrays. Proteomics 2006; 6(10):2964-2971. 8. Sheehan KM, Calvert VS, Kay EW et al. Use of reverse phase protein microarrays and reference standard development for molecular network analysis of metastatic ovarian carcinoma. Mol Cell Proteomics 2005; 4(4):346-355. 9. Grubb RL, Calvert VS, Wulkuhle JD et al. Signal pathway proiling of prostate cancer using reverse phase protein arrays. Proteomics 2003; 3(11):2142-2146. 10. Nam MJ, Madoz-Gurpide J, Wang H et al. Molecular proiling of the immune response in colon cancer using protein microarrays: occurrence of autoantibodies to ubiquitin C-terminal hydrolase L3. Proteomics 2003; 3(11):2108-2115. 11. Waugh DS. Making the most of ainity tags. Trends Biotechnol 2005; 23(6):316-320. 12. Braun P, Hu Y, Shen B et al. Proteome-scale puriication of human proteins from bacteria. Proc Natl Acad Sci USA 2002; 99(5):2654-2659. 13. Drees J, Smith J, Schafer F et al. High-throughput expression and puriication of 6xHis-tagged proteins in a 96-well format. Methods Mol Med 2004; 94:179-190. 14. Scheich C, Sievert V, Bussow K. An automated method for high-throughput protein puriication applied to a comparison of His-tag and GST-tag ainity chromatography. BMC Biotechnol 2003; 3:12. 15. Andresen H, Grotzinger C, Zarse K et al. Functional peptide microarrays for speciic and sensitive antibody diagnostics. Proteomics 2006; 6(5):1376-1384. 16. Bullock AN, Debreczeni J, Amos AL et al. Structure and substrate speciicity of the Pim-1 kinase. J Biol Chem 2005; 280(50):41675-41682. 17. Turk BE, Cantley LC. Using peptide libraries to identify optimal cleavage motifs for proteolytic enzymes. Methods 2004; 32(4):398-405. 18. Tang X, Orlicky S, Liu Q et al. Genome-Wide Surveys for Phosphorylation-Dependent Substrates of SCF Ubiquitin Ligases. Methods Enzymol 2005; 399:433-458. 19. Michaud GA, Salcius M, Zhou F et al. Analyzing antibody speciicity with whole proteome microarrays. Nat Biotechnol 2003; 21(12):1509-1512. 20. He M, Taussig MJ. Single step generation of protein arrays from DNA by cell-free expression and in situ immobilisation (PISA method). Nucleic Acids Res 2001; 29(15):E73-73. 21. Ramachandran N, Hainsworth E, Bhullar B et al. Self-assembling protein microarrays. Science 2004; 305(5680):86-90. 22. Ramachandran N, Hainsworth E, Demirkan G et al. On-chip protein synthesis for making microarrays. Methods Mol Biol 2006; 328:1-14. 23. Wang X, Yu J, Sreekumar A et al. Autoantibody signatures in prostate cancer. N Engl J Med 2005; 353(12):1224-1235. 24. Tabuchi M, Hino M, Shinohara Y et al. Cell-free protein synthesis on a microchip. Proteomics 2002; 2(4):430-435. 25. Weng S, Gu K, Hammond PW et al. Generating addressable protein microarrays with PROfusion covalent mRNA-protein fusion technology. Proteomics 2002; 2(1):48-57. 26. Jung GY, Stephanopoulos G. A functional protein chip for pathway optimization and in vitro metabolic engineering. Science 2004; 304(5669):428-431. 27. Davies DH, Liang X, Hernandez JE et al. Proiling the humoral immune response to infection by using proteome microarrays: high-throughput vaccine and diagnostic antigen discovery. Proc Natl Acad Sci USA 2005; 102(3):547-552. 28. Katzen F, Chang G, Kudlicki W. he past, present and future of cell-free protein synthesis. Trends Biotechnol 2005; 23(3):150-156. 29. Swietnicki W, O’Brien S, Holman K et al. Novel protein-protein interactions of the Yersinia pestis type III secretion system elucidated with a matrix analysis by surface plasmon resonance and mass spectrometry. J Biol Chem 2004; 279(37):38693-38700.

144

Cell-Free Protein Expression

30. Elbaz Y, Steiner-Mordoch S, Danieli T et al. In vitro synthesis of fully functional EmrE, a multidrug transporter, and study of its oligomeric state. Proc Natl Acad Sci USA 2004; 101(6):1519-1524. 31. Ito T, Chiba T, Ozawa R et al. A comprehensive two-hybrid analysis to explore the yeast protein interactome. Proc Natl Acad Sci USA 2001; 98(8):4569-4574. 32. Valasek L, Mathew AA, Shin BS et al. he yeast eIF3 subunits TIF32/a, NIP1/c, and eIF5 make critical connections with the 40S ribosome in vivo. Genes Dev 2003; 17(6):786-799. 33. Gavin AC, Aloy P, Grandi P et al. Proteome survey reveals modularity of the yeast cell machinery. Nature 2006; 440(7084):631-636. 34. Belhumeur P, Lee A, Tam R et al. GSP1 and GSP2, genetic suppressors of the prp20-1 mutant in Saccharomyces cerevisiae: GTP-binding proteins involved in the maintenance of nuclear organization. Mol Cell Biol 1993; 13(4):2152-2161. 35. Anderson GW Jr, Leary SE, Williamson ED et al. Recombinant V antigen protects mice against pneumonic and bubonic plague caused by F1-capsule-positive and -negative strains of Yersinia pestis. Infect Immun 1996; 64(11):4580-4585. 36. Titball RW, Williamson ED. Vaccination against bubonic and pneumonic plague. Vaccine 2001; 19(30):4175-4184. 37. Williamson ED. Plague vaccine research and development. J Appl Microbiol 2001; 91(4):606-608. 38. Li B, Jiang L, Song Q et al. Protein microarray for proiling antibody responses to Yersinia pestis live vaccine. Infect Immun 2005; 73(6):3734-3739. 39. Mattoon D, Michaud G, Merkel J et al. Biomarker discovery using protein microarray technology platforms: antibody-antigen complex proiling. Expert Rev Proteomics 2005; 2(6):879-889. 40. Sawasaki T, Hasegawa Y, Morishita R et al. Genome-scale, biochemical annotation method based on the wheat germ cell-free protein synthesis system. Phytochemistry 2004; 65(11):1549-1555. 41. Kawahashi Y, Doi N, Takashima H et al. In vitro protein microarrays for detecting protein-protein interactions: application of a new method for luorescence labeling of proteins. Proteomics 2003; 3(7):1236-1243. 42. Pornillos O, Chen YJ, Chen AP et al. X-ray structure of the EmrE multidrug transporter in complex with a substrate. Science 2005; 310(5756):1950-1953.

Chapter 12

Cell-Free Protein Expression Screening and Protein Immobilization Using Protein Microarrays Matthew A. Coleman,* Paul Hoeprich, Peter Beernink and Julio A. Camarero*

Abstract

C

urrent, cell-free methodologies allow for robust and rapid protein production and screening. he use of protein arrays attached selectively or nonspeciically to various solid supports is rapidly becoming a common research tool to explore the function and potential relationships of proteins encoded within any genome. Array-based approaches are also ideal for parallel analysis of multiple binary interactions between proteins and other molecules. In addition, engineering novel tagging techniques allows the orientation of proteins of interest and expands the capabilities and use of protein microarrays. Combining cell-free expression with array-based proteomics promises to be a powerful tool for protein biochemistry, molecular diagnostics and therapeutics.

Introduction

Cell-free production of proteins has become a widely accepted means to overcome bottlenecks in protein expression and puriication in support of high-throughput structural proteomics applications.1-4 Cell free protein expression is capable of overcoming those speciic problems associated with obtaining recombinant proteins such as cloning, transfection, cell growth, lysis and subsequent puriication by avoiding multiple steps in the process.4,5 Cell-free protein expression has also proven beneicial for structural determination techniques such as NMR and X-ray crystallography.6-10 Cell-free based methods are also faster than cell-based systems, with the added capability to produce proteins that are diicult to obtain because they are toxic or subject to proteolysis using cell-based expression.11 Cell-free expression can also overcome other expression barriers through the use of additives since it is an “open” system. Additives that have been successfully employed include: chaperonins,12,13 lipids,14 redox factors15-17 and detergents and protease inhibitors14,18 as well as a means for glycosylation.19 One of the quickly expanding uses of cell-free expression-based technologies is the generation of protein microarrays. Array-based methods for protein analysis provide a high-throughput format with which to study protein–protein, protein-DNA and protein-small molecule interactions, which provide important functional information. As well as interaction studies, protein arrays have become an important molecular tool for diagnostic and therapeutic applications.20-28 Combining array-based technologies with cell-free expression is a recent development for highly parallel strategies to analyze protein functions (see Fig. 1). his approach has been used to provide both proteins for spotting5,29-31 as well as means to produce self assembled arrays by transcribing and translating *Corresponding Authors: Matthew A. Coleman and Julio A. Camarero—Lawrence Livermore National Laboratory, Livermore, California U.S.A. Email: [email protected]; [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

146

Cell-Free Protein Expression

Figure 1. Proteomics using cell-free expression linked to protein microarrays. The development of our proteomics project is broken down into five major components, as shown on the left of the scheme, with the net outcome being sufficient yield of soluble protein for subsequent studies. Methodologies using cell-free transcription and translation may enhance the throughput of such projects by circumventing the time consuming steps of cloning, protein production, immobilization and functional characterization.

protein directly from spotted DNA or PCR products.32-35 A disadvantage of these techniques to date, is the lack of selective immobilization and stabilization of the arrayed proteins.

Fluorescence-Based Protein Microarray Expression Screens

Cell-free expression systems are especially well suited for expression screening and for biochemical studies such as enzyme assays or protein interaction analyses, when only microgram quantities of protein (or less) are needed. hese low quantities are not a limitation for multiplexed protein detection and sensitive quantiication.27,36-39 Fluorescence-based protein expression screens have been implemented in several formats, the microplate, dot blot and protein microarray, that utilize luorescent labeling to enable rapid, reproducible detection of proteins with high sensitivity and a wide dynamic range.5,29 Such techniques also facilitate extremely small-scale expression screening either directly using luorescently labeled protein or indirectly by luorescence-based immunoassay. As implemented in a microarray format, the cell-free protein expression arrays provide highly sensitive detection of cell-free expressed proteins.

Cell-Free Protein Expressions Screening and Protein Immobilization Using Protein Microarrays

147

To validate the use of cell-free protein expression screening for compatibility with bacterial expression, we irst examined the correlation of cell-free and bacterially-expressed proteins using C-terminal GFP fusion proteins.40-42 he clones of interest encoded both human and bacterial proteins. he expression data could be grouped into two subsets (Fig. 2), the more highly expressed clones for which the correlation was good (cc = 0.89) and the more poorly (in vivo) expressed clones for which the correlation was weaker (overall cc = 0.69). For the latter clones, the cell-free expression levels were consistently higher than those in vivo. herefore, we concluded that this set comprises proteins that are either cytotoxic or proteolytically sensitive, which underscores the beneits of cell-free expression for certain classes of proteins that are expressed at low levels in cell-based systems. Notably four of the ive proteins were produced at signiicantly higher quantities using the cell free system. hese proteins are DNA repair proteins, which can be diicult to express, suggesting that cell-free expression may be particularly useful for production of diicult classes of proteins. Cell-free expressed proteins were then spotted on microarray format for multiple applications. First, as a tool for rapid expression screening, in which the arrayed proteins were compared for relative expression levels. Control experiments with known amounts of GFP were conducted to bracket the limit of detection for GFP luorescence on a glass slide (Fig. 3, panel A). Next, GFP fusion proteins were in vitro expressed and arrayed directly to demonstrate that spotting was reproducible and therefore useful to quickly identify relative diferences in expression levels (Fig. 3, panel B and C). Such arrays are also useful for protein-speciic detection, as performed by array-based immunoassays to detect GFP fusion proteins (data not shown). hese sorts of experiments can also be easily adapted for automated protein expression proiling studies. Once suicient expression is established based on luorescence detection strategies, these approaches can be used to obtain native or tagged versions of the desired proteins. Overall, in vitro expression and detection methods are relatively rapid since they require no electrophoresis or transfer to membranes. In addition, the detection method were lexible and could use either luorescent fusion proteins or covalently incorporated labels.43,44 Cell-free expression in conjunction with C-terminal tags such as His6 may also have an advantage for detecting exclusively full-length translation products. Since molecules may afect the conformation of the protein, immunological detection of unlabeled proteins may be preferable if the resulting protein will be used for functional assays. labeling would be preferable if no tag or a variety of tags are present on diferent expression clones. Alternatively, immunological detection may be employed using antibodies against ainity tags, or native proteins may be visualized with35 S-Met. hese lexible

Figure 2. Correlation of E. coli-based in vivo and cell-free protein expression levels. Proteins were expressed cell-free using E. coli extracts or in E. coli cultures. Proteins were expressed as GFP fusions, lysed and the soluble fraction was transferred to a 96 well plate and the fluorescence quantified. The correlation coefficient (ml) was 0.89 for the 9 most highly expressed proteins and the overall ml was 0.69.

148

Cell-Free Protein Expression

Figure 3. Microarray-based expression screening. Cell-free expressed GFP and GFP fusion proteins were spotted in quadruplicate on a glass slide (only duplicates are shown) for analysis of yields. One nL of the reaction was spotted in a spot of approximately 100 μm in diameter. A) Known (ug to ng/mL; limit of detection (2 X 10 -13 g) of GFP)amounts of GFP diluted for array quantification and falsely colored red. B) Green, GFP fluorescence; red, Cy3 labeled DNA. Column 1, LcrH; 2, GFP; 3, XRCC1; 4, LcrG; 5, Lysate control; 6, DNA; 7, SFN5; 8, DNA. C) The spot intensity is reproducible and correlates with expression levels observed as detected with a plate reader or by spotting onto a nylon membrane. The error bars indicate the standard deviation among four replicates and the dashed line illustrates background fluorescence. Y axis units are average pixel intensities as measured from the spots in panel B.

approaches for cell-free protein expression enables automated production of many proteins and their subsequent puriication. Coupled with label-free array-based technologies such approaches will become very powerful in the future.45

Array-Based Site-Speciic and Traceless Immobilization of Cell-Free Expressed Proteins

A key element for the rapid and eicient production of protein microarrays using cell-free expression systems is the method of protein immobilization. In order to be successful it has to be able to selectively immobilize the protein of interest from a complex mixture and under diluted conditions, typically around lower than μM concentrations. During the last few years several enzymatic capture approaches have been developed for the covalent and site-speciic immobilization of enzyme-fusion proteins from complex mixtures without the need for puriication and reconcentration steps (see recent reviews in refs. 46, 47). Most of them rely in the expression of the protein of interest fused to an enzyme (typically an esterase or transferase) that is selectively immobilized onto an appropriate ligand-coated surface.48-50 One of the main limitations of these ligand-capture techniques for site-speciic immobilization is that the enzyme remains attached to

Cell-Free Protein Expressions Screening and Protein Immobilization Using Protein Microarrays

149

Figure 4. General scheme for the site-specific and traceless immobilization of proteins through protein trans-splicing. A) Schematic of specific immobilization of proteins on a chemically modified solid support through the use of intein mediated protein trans-splicing. B) Structures for the linkers 1 and 2 that are used for derivation of the surface on the glass slides.

Figure 5. Selective immobilization of MBP-IN from complex mixtures using protein through protein trans-splicing; (A) Soluble cellular fraction of E. coli cells overexpressing MBP-IN. B) MBP-IN expressed in vitro using a cell-free system. Protein concentrations in the cell lysate and IVT crude reaction were estimated by Western Blotting. In both cases, MBP was detected by immunofluorescence on the slide after several washing steps.

the surface ater the immobilization step has taken place. In some cases, the presence of such a large linker could give rise to problems, especially in those applications where the immobilized proteins will be involved in studying protein–protein interactions within complex protein mixtures.51 To address this problem, we have developed a new traceless capture ligand approach for the selective immobilization of proteins to surfaces based on protein trans-splicing process30,52 (Fig. 4A). his process is similar to protein splicing53,54 with the only diference being the intein self-processing domain is split in two fragments (called N-intein and C-intein, respectively).55,56 In our approach, the C-intein fragment is covalently immobilized onto a glass surface through a PEGylated-peptide linker while the N-intein fragment is fused to the C-terminus of the protein to be attached to the surface (Fig. 4B). When both intein fragments interact, they form an active intein domain,

150

Cell-Free Protein Expression

which ligates the protein of interest to the surface at the same time the split intein is spliced out into solution (Fig. 4). he trans-splicing reaction mediated by the split intein results in the direct covalent attachment of the protein through its C-terminus onto a PEGylated surface. A direct covalent attachment between the protein and the surface using poly-ethylene glycol (PEG) linkers eliminates the need for enzymatic and other protein mediators.30,57 hese PEG-based linkers are also very well know to prevent nonspeciic interactions and act as hydrophilic spacers minimizing any detrimental interaction between the attached protein and the solid surface.30,57,58 Key to this approach is the use of the naturally split DnaE intein from Synechocystis sp. PCC6803.59 he C- and N-intein fragments of the DnaE intein are able to self-assemble spontaneously (Kd = 0.1-0.2 μM), not requiring any refolding step.30,60 he DnaE intein-mediated trans-splicing reaction is also very eicient under physiological-like conditions (τ1/2 ≈ 4 h and trans-splicing yields ranging from 85% to almost quantitative).30 Using this strategy, we have successfully speciically immobilized several proteins to chemically modiied SiO2-based substrates from complex mixtures, including cell-free expression reactions as well as soluble cell lysates (Fig. 5). Site-speciic immoblization of proteins using protein trans-splicing is highly speciic and eficient. It allows the use of protein mixtures and eliminates the need for the puriication and/or reconcentration of the proteins prior to the immobilization step. he required minimum protein concentration for eicient immobilization was estimated to be sub-micromolar.30 More importantly, once the protein is immobilized to the surface, both intein fragments are spliced out into solution and easily removed by washing, providing a completely traceless method of attachment. All these features allow this methodology to be easily interfaced with cell-free protein expression systems with rapid access to the high-throughput production of protein chips and other types of biosensing platforms.

Conclusions

We have developed several approaches for producing and screening proteins in a microarray format that will be useful for future assay development. Coupled with high-throughput methods to express and purify proteins, protein arrays have the potential to identify and characterize newly discovered proteins and protein complexes. Production of diicult proteins, such as membrane bound proteins for protein microarrays remains a challenge as does providing quality control for proteins on the array. One area that protein arrays have lagged well behind mRNA expression arrays is the development of statistical tools for the myriad of possible protein array applications. Although, several tools exist (for a complete review see, 61-64 many of these ofer only the most basic image and statistical analysis across replicate spots. Furthermore, the use of antibodies for array-based proteomics relies on the availability of highly sensitive and speciic antibodies that are oten expensive or hard to produce. Array-based approaches are a powerful tool for studying protein, DNA and small molecule interactions.37,65-68 Protein arrays also hold potential for miniaturization and portability and therefore have a multitude of potential applications in basic biological research, enhanced protein functional annotation and veriication, identiication of disease markers and targets and diagnosis of disease. Future plans to fully automate the process through linking of these methodologies could advance the screening and characterization of these proteins and further our understanding of the cellular proteome.

Materials and Methods Cloning and Bacterial Expression

For the construction of C-terminal GFP fusion proteins,69 a set of microbial genes and human cDNAs were ampliied with primers containing restriction site adapters for NdeI and BamHI and a high-idelity polymerase, Pwo DNA polymerase (Roche Diagnostics). he PCR products were digested with NdeI and BamHI restriction enzymes and subcloned into the pET28-derived plasmid for GFP-fusion69 using the same restriction sites. Genes were expressed as C-terminal

Cell-Free Protein Expressions Screening and Protein Immobilization Using Protein Microarrays

151

GFP fusion proteins in 2 mL Escherichia coli cultures, grown in LB media at 37 °C with vigorous shaking until mid-exponential phase (OD600 = 0.6) was reached and expression was induced with 1 mM IPTG. Cultures were grown for an additional 2 h and harvested by centrifugation. Fluorescence of GFP was determined from the soluble fraction of bacterially or IVT-expressed proteins. Detection and quantiication of GFP luorescence was carried out in a 96-well Genios luorescence plate reader (Tecan).

Template Preparation and Cell-Free Reactions Expression Screening

Sequential PCR and IVT reactions were performed in 25 μL volumes in 96-well plates. DNA ampliication was performed using primers speciic to the T7 promotor (5´-GCGCGCGAGAT CTCGATCCCGCGAAATTAATACGAC) and terminator (5´-GCGCGCGTATCCGGAT ATAGTTCCTCCTTTCAG) sequences and Taq DNA polymerase (Roche Diagnostics). PCR conditions consisted of 5 cycles with a 50 °C annealing temperature followed by 20 cycles with a 60 °C annealing temperature, with 2 min extension times for all cycles. Subsequent IVT reactions contained 1 μL of the PCR reaction product and 20 μL of a master mix containing the RTS kit components and 0.13 μL of a -Lys-tRNALys conjugate, FluoroTect GreenLys (Promega). he reactions were incubated at 30 °C for 4 h and analyzed immediately or stored at –20 °C.

pMBP-IN

he gene fragment encoding DnaE-IN with 4 N-terminal extein residues (FAEY) was ampliied by polymerase chain reaction (PCR) using a plasmid containing the DnaE genes from of Synechocystis sp. strain PCC6803 (Ssp) as template.59 he 5´-primer (5´- TG GAA TTC TTT GCG GAA TAT TGC CTC AGT TTT GG-3´) and the 3´-primer (5´- TTT GGA TCC TTA TTT AAT TGT CCC AGC GTC AAG TAA TGG AAA GGG -3´) introduced EcoR I and BamH I restriction sites, respectively. he PCR ampliied DNA was puriied, digested simultaneously with EcoR I and BamH I and then ligated into a EcoR I, BamH I-treated pMAL-C2X plasmid (New England Biolabs).

pTXB1-IC

he gene fragment for the DnaE-IC was prepared by PCR using a plasmid containing the DnaE genes from of Synechocystis sp. strain PCC6803 (Ssp) as template.59 he 5´-primer (5´- A AAA AGG CAT ATG GTT AAA GTT ATC GGT CGT CGT TCC CTC-3´) and 3´-primer (5´- TAA AAT GGC TCT TCG GCA ATT GGC GGC GAT C-3´) introduced Nde I and Sap I restriction sites, respectively. he PCR product was puriied, double-digested with Nde I and Sap I and ligated into an Nde I, Sap I-treated pTXB-1 plasmid (New England Biolabs).

pIVEX-MBP-IN

he DNA fragment encoding the protein MBP-IN was prepared by digesting pMBP-IN with Nde I and BamH I restriction enzymes. he DNA fragment was puriied and inserted between the Nde I and BamH I restriction sites of the pIVEX2.3 plasmid (Roche Diagnostics). Cell-free protein synthesis reactions were performed using Roche RTS 100 HY system (Roche Diagnostics).

Protein Microarray Spotting and Detection

To analyze GFP proteins in an array format, crude IVT-expressed proteins (~1 nL) were spotted in triplicate on CMT-GAPS glass slides (Corning) with a robotic arrayer (Norgren Systems). Arrays contained up to 224 spots of ~200 μm diameter. Spotting controls included IVT lysate alone and Cy-labeled DNA fragments (Molecular Probes). he arrays were dried at 25 °C and stored at 4 °C until use. Fluorescence was quantiied using a ScanArray 5000 (Packard Bioscience) and visualized with false color.

152

Cell-Free Protein Expression

Functionalization of Glass Substrates

Glass slides coated with g-aminopropyl-silane (GAPS IITM; Corning) were treated with 200 μL of a solution of MPS (3-maleimidopropionic acid N-hydroxysuccinimide ester, 2 mM) in 0.1 M Tris·HCl bufer at pH 7.5 for 40 min at room temperature using a hybridization chamber (Schleicher & Schuell, Keene, NH). he glass slides were washed with deionized H2O, MeOH and dried under a N2 stream. he modiied glass slides were immediately treated with 200 μL of a solution of thiol linkers 1 (0.05 mM) and 2 (1.5 mM) in freshly degassed 0.5 mM EDTA, 1 mM TCEP, 50 mM sodium phosphate, 150 mM NaCl bufer at pH 7.0 for 16 hours at room temperature. Ater the glass slides were washed and dried as described above, the S-tBu protecting group on the C-terminal Cys residue was deprotected by treating the glass slides with 50% β-mercaptoethanol in DMF for 2 h at room temperature. he glass slides were washed with deionized H2O, MeOH, dried under a N2 stream and used immediately.

Generation of Protein Microarrays for Protein Immobilization

Protein solutions (0.1 mM-40 mM) in spotting bufer (0.5 mM EDTA, 1 mM TCEP, 50 mM sodium phosphate, 150 mM NaCl bufer at pH 7. 0 containing 10% glycerol) were arrayed in functionalized glass slides using a robotic arrayer (Norgren Systems). Proteins were spotted with a center-to-center spot distance of 250 μm with an average spot size of 100 μm in diameter. Ater spotting, the array was kept in a humidiied chamber at 37 °C for 16 h. he glass substrate was thoroughly washed with PBST (50 mM sodium phosphate, 500 mM NaCl bufer at pH 7.2 containing 0.2% Triton X-100). Immobilized EGFP was imaged using a ScanArray 5000 at 488 nm without further modiication. Immobilized MBP was detected by immunoluorescence at 543 nm using a primary murine anti-MBP antibody and then a secondary goat anti-mouse antibody conjugated to TRITC (tetramethylrhodamine isothiocyanate). he amount of luorescence was quantiied using the QuantArray sotware package (Packard Bioscience, Billerica, MA, USA).

Acknowledgement

We would like to thank Drs. B. Segelke and J. Albala for helpful discussions during the course of this work. We would also like to thank V. Lao and Y. Kwon for technical assistance. Figures 2 and 3 were adapted from Coleman et al 2004. Figures 4 and 5 were adapted from Kwon et al 2006. his work was performed under the auspices of the US Department of Energy by the University of California, Lawrence Livermore National Laboratory under contract No. W-7405-Eng-48.

References

1. Kigawa T, Yabuki T, Yokoyama S. Large-scale protein preparation using the cell-free synthesis. Tanpakushitsu Kakusan Koso 1999; 44(4 Suppl):598-605. 2. Kigawa T et al. Cell-free production and stable-isotope labeling of milligram quantities of proteins. FEBS Lett 1999; 442(1):15-9. 3. Sawasaki T et al. A cell-free protein synthesis system for high-throughput proteomics. Proc Natl Acad Sci USA 2002; 99(23):14652-7. 4. Segelke BW et al. Laboratory scale structural genomics. J Struct Funct Genomics 2004; 5(1-2):147-57. 5. Coleman MA et al. High-throughput, luorescence-based screening for soluble protein expression. J Proteome Res 2004; 3(5):1024-32. 6. Endo Y, Sawasaki T. High-throughput, genome-scale protein production method based on the wheat germ cell-free expression system. Biotechnol Adv 2003; 21(8):695-713. 7. Endo Y, Sawasaki T. Advances in genome-wide protein expression using the wheat germ cell-free system. Methods Mol Biol 2005; 310:145-67. 8. Klammt C et al. High level cell-free expression and speciic labeling of integral membrane proteins. Eur J Biochem 2004; 271(3):568-80. 9. Yokoyama S. Protein expression systems for structural genomics and proteomics. Curr Opin Chem Biol 2003; 7(1):39-43. 10. Keppetipola S et al. From gene to HSQC in under ive hours: high-throughput NMR proteomics. J Am Chem Soc 2006; 128(14):4508-9. 11. Katzen F, Chang G, Kudlicki W. he past, present and future of cell-free protein synthesis. Trends Biotechnol 2005; 23(3):150-6.

Cell-Free Protein Expressions Screening and Protein Immobilization Using Protein Microarrays

153

12. Frydman J, Hartl FU. Principles of chaperone-assisted protein folding: diferences between in vitro and in vivo mechanisms. Science 1996; 272(5267):1497-502. 13. Tsalkova T et al. GroEL and GroES increase the speciic enzymatic activity of newly-synthesized rhodanese if present during in vitro transcription/translation. Biochemistry 1993; 32(13):3377-80. 14. Klammt C et al. Evaluation of detergents for the soluble expression of alpha-helical and beta-barrel-type integral membrane proteins by a preparative scale individual cell-free expression system. Febs J 2005; 272(23):6024-38. 15. Kim DM, Swartz JR. Eicient production of a bioactive, multiple disulide-bonded protein using modiied extracts of Escherichia coli. Biotechnol Bioeng 2004; 85(2):122-9. 16. Jewett MC, Swartz JR. Substrate replenishment extends protein synthesis with an in vitro translation system designed to mimic the cytoplasm. Biotechnol Bioeng 2004; 87(4):465-72. 17. Jewett MC, Swartz JR. Mimicking the Escherichia coli cytoplasmic environment activates long-lived and eicient cell-free protein synthesis. Biotechnol Bioeng 2004; 86(1):19-26. 18. Mori M, Morris SM Jr, Cohen PP. Cell-free translation and thyroxine induction of carbamyl phosphate synthetase I messenger RNA in tadpole liver. Proc Natl Acad Sci USA 1979; 76(7):3179-83. 19. Katzen F, Kudlicki W. Eicient generation of insect-based cell-free translation extracts active in glycosylation and signal sequence processing. J Biotechnol 2006; 125(2):194-7. 20. Cahill DJ. Protein and antibody arrays and their medical applications. J Immunol Methods 2001; 250(1-2):81-91. 21. Kricka LJ et al. Current perspectives in protein array technology. Ann Clin Biochem 2006; 43(Pt 6):457-67. 22. Predki PF et al. Protein microarrays: a new tool for proiling antibody cross-reactivity. Hum Antibodies 2005; 14(1-2):7-15. 23. Poetz O et al. Protein microarrays for antibody proiling: speciicity and ainity determination on a chip. Proteomics 2005; 5(9):2402-11. 24. Zhu H, Snyder M. Protein chip technology. Curr Opin Chem Biol 2003; 7(1):55-63. 25. Templin MF et al. Protein microarray technology. Drug Discov Today 2002; 7(15):815-22. 26. Albala JS. Array-based proteomics: the latest chip challenge. Expert Rev Mol Diagn 2001; 1(2):145-52. 27. Haab BB. Advances in protein microarray technology for protein expression and interaction proiling. Curr Opin Drug Discov Devel 2001; 4(1):116-23. 28. Zhu H, Snyder M. Protein arrays and microarrays. Curr Opin Chem Biol 2001; 5(1):40-5. 29. Coleman MA et al. Identiication of chromatin-related protein interactions using protein microarrays. Proteomics 2003; 3(11):2101-7. 30. Kwon Y, Coleman MA, Camarero JA. Selective immobilization of proteins onto solid supports through split-intein-mediated protein trans-splicing. Angew Chem Int Ed Engl 2006; 45(11):1726-9. 31. Weng S et al. Generating addressable protein microarrays with PROfusion covalent mRNA-protein fusion technology. Proteomics 2002; 2(1):48-57. 32. Tabuchi M et al. Cell-free protein synthesis on a microchip. Proteomics 2002; 2(4):430-5. 33. Angenendt P et al. Generation of high density protein microarrays by cell-free in situ expression of unpuriied PCR products. Mol Cell Proteomics 2006; 5(9):1658-66. 34. Ramachandran N et al. Self-assembling protein microarrays. Science 2004; 305(5680):86-90. 35. He M. Generation of protein in situ arrays by DiscernArray technology. Methods Mol Biol 2004; 264:25-31. 36. Pawlak M et al. Zeptosens’ protein microarrays: a novel high performance microarray platform for low abundance protein analysis. Proteomics 2002; 2(4):383-93. 37. Haab BB. Applications of antibody array platforms. Curr Opin Biotechnol 2006; 17(4):415-21. 38. Haab BB, Lizardi PM. RCA-enhanced protein detection arrays. Methods Mol Biol 2006; 328:15-29. 39. Haab BB. Multiplexed protein analysis using antibody microarrays and label-based detection. Methods Mol Med 2005; 114:183-94. 40. Cabantous S, Waldo GS. In vivo and in vitro protein solubility assays using split GFP. Nat Methods 2006; 3(10):845-54. 41. Cabantous S, Terwilliger TC, Waldo GS. Protein tagging and detection with engineered self-assembling fragments of green luorescent protein. Nat Biotechnol 2005; 23(1):102-7. 42. Waldo GS. Improving protein folding eiciency by directed evolution using the GFP folding reporter. Methods Mol Biol 2003; 230:343-59. 43. Olejnik J et al. N-terminal labeling of proteins using initiator tRNA. Methods 2005; 36(3):252-60. 44. Gite S et al. Ultrasensitive luorescence-based detection of nascent proteins in gels. Anal Biochem 2000; 279(2):218-25. 45. Yu X, Xu D, Cheng Q. Label-free detection methods for protein microarrays. Proteomics 2006; 6(20):5493-503.

154

Cell-Free Protein Expression

46. Camarero JA, Kwon Y, Coleman MA. Chemoselective attachment of biologically active proteins to surfaces by expressed protein ligation and its application for “protein chip” fabrication. J Am Chem Soc 2004; 126(45):14730-1. 47. Woo YH, Camarero JA. Interfacing “sot” and “hard” matter with exquisite chemical control. Curr Nanoscience 2006; 2:93-103. 48. Hodneland CD et al. Selective immobilization of proteins to self-assembled monolayers presenting active site-directed capture ligands. Proc Natl Acad Sci USA 2002; 99(8):5048-52. 49. Yin J et al. Labeling proteins with small molecules by site-speciic posttranslational modiication. J Am Chem Soc 2004; 126(25):7754-5. 50. Sielaf I et al. Protein function microarrays based on self-immobilizing and self-labeling fusion proteins. Chembiochem 2006; 7(1):194-202. 51. Zhu H et al. Analysis of yeast protein kinases using protein chips. Nat Genet 2000; 26(3):283-9. 52. Kwon Y, Coleman MA, Camarero JA. Selective Immobilization of Proteins onto Solid Supports through Split-Intein-Mediated Protein Trans-Splicing. Proceedings of the 19th American Peptide Symposium. In:Blondelle SE, ed. Understanding Biology Using Peptides. New York: Springer, 2005:728-730. 53. Chong S et al. Protein splicing involving the Saccharomyces cerevisiae VMA intein. he steps in the splicing pathway, side reactions leading to protein cleavage and establishment of an in vitro splicing system. J Biol Chem 1996; 271(36):22159-68. 54. Perler FB, Adam E. Protein splicing and its applications. Curr Opin Biotechnol 2000; 11(4):377-83. 55. Lew BM, Mills KV, Paulus H. Protein splicing in vitro with a semisynthetic two-component minimal intein. J Biol Chem 1998; 273(26):15887-90. 56. Perler FB. A natural example of protein trans-splicing. Trends Biochem Sci 1999; 24(6):209-11. 57. Camarero JA et al. Fmoc-based synthesis of peptide alpha-thioesters using an aryl hydrazine support. J Org Chem 2004; 69(12):4145-51. 58. Cheung CL et al. Fabrication of assembled virus nanostructures on templates of chemoselective linkers formed by scanning probe nanolithography. J Am Chem Soc 2003; 125(23):6848-9. 59. Wu H, Hu Z, Liu XQ. Protein trans-splicing by a split intein encoded in a split DnaE gene of Synechocystis sp. PCC6803. Proc Natl Acad Sci USA 1998; 95(16):9226-31. 60. Shi J, Muir TW. Development of a tandem protein trans-splicing system based on native and engineered split inteins. J Am Chem Soc 2005; 127(17):6198-206. 61. Haab BB et al. Immunoassay and antibody microarray analysis of the HUPO Plasma Proteome Project reference specimens: systematic variation between sample types and calibration of mass spectrometry data. Proteomics 2005; 5(13):3278-91. 62. Hamelinck D et al. Optimized normalization for antibody microarrays and application to serum-protein proiling. Mol Cell Proteomics 2005; 4(6):773-84. 63. Steel LF, Haab BB, Hanash SM. Methods of comparative proteomic proiling for disease diagnostics. J Chromatogr B Analyt Technol Biomed Life Sci 2005; 815(1-2):275-84. 64. Jung K et al. A renewed approach to the nonparametric analysis of replicated microarray experiments. Biom J 2006; 48(2):245-54. 65. Phizicky E et al. Protein analysis on a proteomic scale. Nature 2003; 422(6928):208-15. 66. Hall DA et al. Regulation of gene expression by a metabolic enzyme. Science 2004; 306(5695):482-4. 67. Huang J et al. Finding new components of the target of rapamycin (TOR) signaling network through chemical genetics and proteome chips. Proc Natl Acad Sci USA 2004; 101(47):16594-9. 68. Robinson WH. Antigen arrays for antibody proiling. Curr Opin Chem Biol 2006; 10(1):67-72. 69. Waldo GS et al. Rapid protein-folding assay using green luorescent protein. Nat Biotechnol 1999; 17(7):691-5.

Chapter 13

Cell-Free Protein Expression Labeling with Fluorophores Jerzy Olejnik*

Abstract

T

his chapter is devoted to the emerging ield of engineering artiicial luorescent proteins by using tRNA mediated expression labeling. he method of producing luorescent proteins utilizes in vitro (cell-free) or in vivo (cellular) expression and relies on the use of engineered tRNAs misaminoacylated with nonnative or luorescent amino acids. he basic principles and requirements of this method and methods for producing misaminoacylated tRNAs with luorescent labels, are reviewed. he selected applications utilizing luorescent proteins produced using tRNA mediated expression labeling are also discussed and include: protein expression monitoring, cellular imaging, detecting protein-protein interactions, conformational changes and truncation mutations.

Introduction

Protein engineering is widely used in basic research and biotechnology to obtain luorescent proteins with improved or unique properties. Fluorophore-labeled proteins are suitable for a variety of downstream applications including expression screening, detection of gene products, protein-protein and protein-ligand interactions to name just a few. Fluorescent proteins can be used in conjunction with a variety of luorescence techniques including microarrays,1 homogenous assays (luorescence intensity, polarization and FRET),2 SDS-PAGE detection with laser induced luorescence in conjunction with gel scanners,3 capillary electrophoresis,4 or live cell imaging.5 A variety of chemical and biosynthetic strategies have been developed to incorporate luorescent labels into proteins. he simplest method to introduce luorescence properties to any given protein is via chemical labeling. For example, luorescent labels can be attached to an exposed lysine or cysteine side chains on a native target protein or protein engineered using mutagenesis4 using amine or thiol reactive reagents. However, this approach leads to a nonhomogenous labeling6 and in extreme cases to a complete loss of protein activity. Yet another approach relies on construction of fusion proteins with one of the naturally occurring luorescent proteins, such as GFP and its variants.7-9 Intein-mediated or native chemical ligation10-12 allows to construct luorescent proteins prepared from two separately prepared and labeled protein fragments. One novel approach involves incorporation of an epitope tag which contains a tetra-cysteine (CCXXCC) motif and reacts with a speciic luorophore carrying a bi-arsenical moiety leading to a luorescent conjugate.13,14 An alternative approach to site-speciic introduction of luorescent probes relies on the selective oxidation of N-terminal serine or threonine to an aldehyde followed by a speciic reaction with aromatic amines or hydrazine derivatives (hydrazides, or thiosemicarbazide). Ketone moieties can also be introduced during solid phase peptide synthesis and later derivatized with luorescent labels).15 *Jerzy Olejnik—Bio-Comm, LLC. Brookline, Massachusetts U.S.A. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

156

Cell-Free Protein Expression

his chapter is devoted to the emerging ield of engineering artiicial luorescent proteins by using tRNA mediated expression labeling. he method of producing luorescent proteins utilizes in vitro (cell-free) or in vivo (cellular) expression and relies on the use of engineered tRNAs misaminoacylated with nonnative or luorescent amino acids. For site-speciic incorporation this method also requires the use of suppressor or frameshit tRNAs and corresponding codons on the DNA/mRNA template. his approach has opened the door to many novel applications of luorescent proteins in the ields of biophysics, cell biology, diagnostics, drug discovery and proteomics.

he Basics of tRNA Mediated Expression Labeling

In addition to chemical engineering and site-directed mutagenesis, an approach based on tRNA mediated cell-free or cellular expression labeling presents new opportunities for protein engineering. his approach relies on 1) the development of novel tRNAs and their misaminoacylation with amino acids carrying luorescent labels or nonnative amino acids, 2) the recognition of special codons by the tRNAs and 3) the eicient expression of these modiied proteins. he pioneering studies of Nirenberg and Khorana in the early 1960s16,17 provided scientists with the genetic code, the critical information necessary to engineer proteins at the genetic level. In addition, the ability of the translation machinery in cells to accommodate amino acids outside the canonical 20 has been long known.18 tRNA mediated labeling, by exploiting this fact, allows expansion of the genetic code so that custom designed nonnative amino acids can be selectively introduced into proteins.

Misaminoacylated tRNAs

Early studies using cell-free expression labeling were based on the chemical modiication of aminoacyl-tRNAs prepared enzymatically with one of the twenty native amino acids. his approach allows speciic amino acids in the protein to be replaced by their derivatized analogs. For example, it was demonstrated by Johnson and coworkers that a tRNALys enzymatically aminoacylated with lysine could be derivatized with a light-activated cross-linking reagent to form N-ε-(5-azido-2-ni trobenzoyl)-Lys-tRNALys.19-21 Using these misaminocylated tRNAs in a cell-free protein synthesis system enabled investigation of secretory protein translocation across the ER membrane by photo-crosslinking techniques. More recently, efective methods have been developed to incorporate nonnative amino acids at a speciic point in a protein. his approach is based on the use of stop (amber) codons incorporated at speciic positions in the DNA/mRNA coding for proteins and complementary suppressor tRNAs which cause the stop codon to be read-through.22,23 Figure 1 illustrates the basic approach for the incorporation of nonnative or luorescent amino acids at speciic locations in a protein. As seen in this Figure, the most important components for incorporation of luorescent labels into proteins using the in vitro (cell-free) or in vivo (cellular) translation systems are: 1) luorescent/non native amino acid; 2) tRNA, which could be an elongator, initiator, or suppressor tRNA; 3) DNA/mRNA containing codon(s) complementary to the anticodons present on the misaminoacylated tRNA; 4) the translation system capable of incorporating of the luorescent derivatives into the growing polypeptide chain. he aminoacylation of a suppressor tRNA is accomplished either by using methods for chemical aminoacylation of tRNAs or by modifying an aminoacyl-tRNA aminoacylated enzymatically. In the chemical aminoacylation method, the nonnative amino acid of interest is used to acylate the dinucleotide pdCpA and then ligated to a truncated tRNA.24,25 In contrast to enzymatic aminoacylation of tRNAs, which is generally limited to only natural amino acids or in special cases their analogs, it is possible to aminoacylate tRNAs with any nonnative amino acid by chemical aminoacylation. A third method is to misaminoacylate a tRNA using a native aminoacyl synthetase. Until recently, this approach was restricted to a relatively small set of amino acid analogs such as luoro-phenylalanine26 or the proline analog thiaproline.27 Recently, the in vitro evolution of synthetases and tRNAs allowed for the generation of tRNAs and synthetases capable of incorporating nonnative or luorescent amino acids using cellular systems through the expansion of the genetic code.28-30 his work holds great promise for future developments in protein engineering.

Cell-Free Protein Expression Labeling with Fluorophores

157

Figure 1. Cell-free expression labeling using fluorescent amino acids.

he nonnative amino acid is then targeted to the proper position in the nascent protein by choosing a speciic codon out of the possible 64 which corresponds to the anticodon of the misaminoacylated tRNA. For example, for the purpose of labeling a suppressor tRNA which acts to suppress a stop codon (i.e., UAG (amber); UAA (ochre) or UGA (opal)) can be placed at a speciic position in the gene. In addition, it is also possible to use a nonnatural 4-base and 5-base frameshit anticodons and corresponding codons in the DNA/mRNA. Finally, the protein is synthesized in a cell free or cellular translation system to which the misaminoacylated tRNA is introduced. One of the newest methods for tRNA misaminoacylation uses in vitro evolved ribozyme for tRNA aminoacylation.31-33 he evolved ribozyme (Flexizyme) can be programmed for selectivity toward speciic tRNAs. his method therefore has a great potential as a means of highly eicient way of generating aminoacyl-tRNAs that are charged with nonnatural amino acids. Recently, the method has been improved and tRNA speciicity has been programmed into the Flexizymes31 which opens new possibilities for generation of misaminoacylated tRNAs. In this new Flexizyme design, tRNA speciicity have been programmed by appending to its 3’-end a tRNA-speciic sequence (TSS), which is complementary to the acceptor stem of the cognate tRNA. his new ribozyme was able to recognize its cognate tRNA and aminoacylate it with a high degree of speciicity and also discriminate against the noncognate tRNAs. A versatile approach for the large scale synthesis of misaminoacylated tRNAs has been also described.34 his approach relies on the preparation of the dinucleotides aminoacylated with amino acids and protected with protective groups (see Fig. 2). Ater synthesis and puriication these protective groups can be selectively removed while the 2’,-(3’)-O-aminoacyl linkage preserved. In the next stage the free amino group can be reacted with a variety of markers available in the form of active NHS esters under carefully controlled conditions. A variety of misaminoacylated tRNAs (elongators and suppressors) were prepared using this approach34 including tRNAs carrying -FL, 5-FAM, 6-FAM, NBD, Cy,3 PDBA, Biotin and PC-Biotin. hese tRNAs were then prepared and the markers incorporated into nascent proteins during in vitro translation.

158

Cell-Free Protein Expression

Figure 2. General scheme for the preparation of marker-aminoacyl-pdCpA conjugates using protective group replacement.

Sisido et al35 describe eicient scheme for the preparation of misaminocylated 4-base codon tRNAs via the use of Peptide Nucleic Acid (PNA) assisted chemical aminoacylation. he aminoacylation is essentially achieved by hybridizing an amino acid thioester attached to the PNA with the 3’-end of tRNA. he net result is the aminoacylated tRNA with nonnative amino acid as a result of transesteriication reaction. Taira et al36 describes an eicient in vitro selection scheme for the selection of four-base tRNAs capable of highly eicient incorporation of nonnative amino acids, where a puromycin-tRNA conjugate was fused with streptavidin. As a result of the tRNA library screen tRNAs with best suppression activity could be selected.

Frameshit Suppression

To avoid competition between endogenous release factors and suppressor tRNAs, a new strategy based on frameshit suppression mutations has been developed.37-39 his strategy, based on ability of tRNAs with extended anti-codons to suppress frameshit mutations, has two advantages over conventional suppression using the three-base amber, opal and ochre stop codons: 1) competition

Cell-Free Protein Expression Labeling with Fluorophores

159

between tRNA and release factors (RF) is avoided and 2) this approach allows more than one unnatural amino acids to be incorporated into a single protein by using two or more frameshit suppressor tRNAs. In fact, it was shown that not only four- but even ive-base anti-codon tRNAs are able to incorporate unnatural amino acids with eiciency up to 86%.40-43 One recent example of protein engineering in vivo involves the combination of frameshit and nonsense suppressor tRNAs to incorporate simultaneously two and three nonnative amino-acids, including 5-F-Trp, at speciic sites.44 his approach would enable the use of FRET probes or structural probes for structure-function studies.

Fluorophore Labeling at the N-Terminus

A sensitive, non-isotopic, luorescence-based method for the detection of nascent proteins directly in polyacrylamide gels was reported.45 A luorescent reporter group was incorporated at the N-terminus of nascent proteins using an E. coli initiator tRNAfmet misaminoacylated with methionine modiied at the α-amino group. In addition to the normal formyl group, the protein translational machinery accepts -FL, a relatively small luorophore with a high luorescent quantum yield, as an N-terminal modiication. A direct visualization of a nascent protein bands with good sensitivity was achieved either by using a luorescence scanner or conventional UV-transilluminator. his approach eliminates the need for radioactivity and provides rapid detection of the protein bands immediately ater electrophoresis without any downstream processing. Under optimal conditions, the incorporation of -FL using the above approach was found to be low (1-2%) even though it was suicient to detect nascent proteins with good sensitivity using SDS-PAGE and luorescent gel scanner. For this reason a more eicient method has been developed46,47 which utilizes an amber initiator suppressor tRNA chemically aminoacylated with a nonnative-amino acid conjugate which is introduced into an E. coli S30 cell-free translation system. he normal initiator codon (ATG) of the gene was replaced by an amber codon (TAG) in the DNA template. Using this approach, several luorophores and ainity tags have been incorporated at the N-terminus of various proteins such as α-hemolysin, dihydrofolate reductase and irely luciferase. Typically in this system the speciic labeling achieved was 25-50 % while protein yield was around 20% compared to the yield of wild type protein translation. For the proteins studied, the presence of modifying groups at the N-terminus did not afect enzymatic activity. Protein translation has been initiated with diferent labeled amino acids.47 In one case, lysine carrying two diferent labels on both amino groups was successfully incorporated at the N-terminus. he resulting protein had a luorophore -FL group on the ε-amino group and a biotin or PC-biotin48 on α-amino group of the N-terminal lysine.

Fluorophore Labeling at the C-Terminus

Nemoto et al have developed a method for the luorescence labeling of the C-terminal of nascent proteins using luorescein-puromycin conjugates.49 he luorescent analog of puromycin (an antibiotic that inhibits protein synthesis by binding to the A site on the ribosome) forms a bond with the C-terminal amino acid residue of the protein but only when the coding sequence does not contain a termination codon. his method has been used to monitor protein-protein interactions by luorescence polarization.49 More recently, Doi et al50 used this method along with microarrays and luorescence cross-correlation spectroscopy (FCCS) to monitor protein-protein interactions. he reader is also referred to the separate chapter in this book devoted to puromycin-mediated labeling.

Labeling at Speciic Amino Acids

A general method of incorporation of luorophores into nascent proteins can be based on the use of elongator tRNAs (e.g., lysyl-tRNA) mis-aminoacylated with luorophores. For example, an expression labeling luorescent tRNA carrying -FL has been developed.51 his tRNA inserts -FL luorescent labels at random lysine positions in the protein chain using an E. coli tRNAlys aminoacylated with lysine and modiied at the ε-amino group. his particular tRNA is available under the trade name FluoroTectTM-GreenLys from Promega Corp (Madison, WI). Similar to

160

Cell-Free Protein Expression

-FL-Met-tRNAfmet -FL-Lys-tRNALys, enables sensitive detection of cell-free synthesized proteins in a prokaryotic translation system (E. coli S30 extract) as well as eukaryotic translation systems (wheat germ and rabbit reticulocyte cell-free translation extracts)52 and has been used for a variety of applications (see below). In addition to nonspeciic lysine labeling, Lien et al53 have reported site-speciic incorporation of a luorescent tag into nascent proteins using a Cys-tRNAcys. Ater aminoacylation of E. coli Cys-tRNAcys, it was modiied with a SH-reactive luorophore (-FL). his method may prove valuable when used in conjunction with site-directed mutagenesis to create protein containing single-cysteine residues by which one can prepare the nascent protein containing single luorophore at deined position.

Selected Applications of tRNA Mediated Expression Labeling Incorporating Fluorescent Reporters

One attractive use of tRNA mediated expression labeling is the incorporation of reporter groups into molecules that can provide information about protein structure and conformational changes, especially in membrane proteins not easily analyzed by x-ray crystallography. One of the earliest demonstrations of this capability was the work of Johnson and coworkers who were able to label proteins with luorescent amino acid derivatives.54 Cornish et al55 reported site-speciic incorporation of spin-labeled, luorescent and photoactivatable amino acids into T4 lysozyme. Chamberlin and coworkers incorporated three diferent luorescent analogs, 5-hydroxytryptophan, 7-azatryptophan and ε-dansyllysine into β-galactosidase.56 One novel aspect of this work was the use of an E. coli S30 extract derived from a mutant strain producing a faulty release factor (RF1). he highest expression yield obtained was for 5’-OH Trp (7.6 micrograms/ml) using the suppressor tRNAGlyCUA , whereas much lower yields were observed for 7-azatryptophan and ε-dansyl-lysine. As an example of a similar approach using in vivo expression labeling, Turcatti et al57,58 incorporated a luorescent amino acid, 3-N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)-2,3-diaminopropionic acid (NBD-Dpr), at speciic sites in the tachykinin neurokinin-2 receptor, a member of the large family of 7 α-helix G-protein coupled receptors (GPCRs). he use of Xenopus oocytes allowed for functional assays of the receptor. Structural information was obtained by measuring the intermolecular distance between the luorescent amino acids placed at diferent sites and a luorescently labeled heptapeptide antagonist using luorescence resonance energy transfer (FRET). he site-speciic incorporation of a luorescent amino acid, Aladan (Alanyl-6-dimethylamin o-2-acylnaphthalene), into both soluble and membrane proteins have recently been reported.59 Aladan was shown to be exceptionally sensitive to the polarity of its surroundings and could be incorporated site-selectively at buried and exposed sites, in both soluble and membrane proteins. In one example of this capability, Aladan was incorporated at various sites into the Kir2.1 and Shaker potassium channels. hese proteins showed normal activity, indicating that the incorporation of Aladan did not interfere with proper folding. In addition, Aladan was incorporated by solid-phase synthesis at four diferent sites in GB1, a highly thermostable IgG-binding domain that has been extensively characterized. In all cases the luorescence emission wavelength was dependent on the site of incorporation and local microenvironment.

Expression Monitoring/Screening and Cellular Imaging

In addition to custom engineered proteins which normally require site-speciic modiications at particular internal residues, tRNA mediated expression labeling can be used to incorporate luorescent moieties at nonspeciic positions, such as at random lysine residues or only at the N-terminal position of a protein. his approach can be very useful in the ields of gene expression, proteomics and even medical molecular diagnostic. For example, luorescent labels can be used to monitor in vitro or even in vivo expression of proteins, to build arrays of proteins which detect speciic protein-protein or drug-protein interactions and even to detect genetic defects which are expressed on the protein level.

Cell-Free Protein Expression Labeling with Fluorophores

161

he FluoroTect GreenLys tRNA51,52 has been used for high throughput expression screening of various cDNA constructs for proteomic applications.60 Speciically labeled proteins in cellular lysates were detected in one of three formats: a microplate using a luorescence plate reader, a dot-blot using a luorescence scanner or a microarray using a laser scanner. his work established a correlation between the various detection formats and validated the use of protein microarrays for fast expression screening. he same tRNA was also used in a large project devoted to the preparation of 1463 native human and 1343 fusion expression ready clones of human cDNAs encoding large proteins.61 he rapid cell-free luorescent expression labeling was used to verify that the cDNA are expressible and that they produce the desired size gene product on the denaturing SDS-PAGE gel. he -lysyl-tRNAlys has also been used to demonstrate the presence of coupled transcription and translation within the nuclei of mammalian cells.62 he visualization of translation was achieved by incubating permeabilized mammalian cells with lysyl-transfer RNA tagged with (-lysyl-tRNAlys). Although most nascent polypeptides were found in the cytoplasm, some were found in discrete nuclear sites known as transcription “factories”. his coupling is simply explained if nuclear ribosomes translate nascent transcripts as those transcripts emerge from still-engaged RNA polymerases, much as they do in bacteria. Similarly, it has been recently discovered that mammalian sperm cells perform translation of nuclear-encoded proteins via the mitochondrial ribosomes.63 his has been achieved by incubating partially permeabilized bovine sperm cells with FluoroTect GreenLys and examining the cells under confocal microscope. he translation of speciic proteins was then conirmed using Western blot.

Detecting Interactions, Conformational Changes and Proteolysis

A four-base codon/anticodon (CGGG/CCCG) pair was recently used to incorporate luorescent amino acid analogues at speciic positions in streptavidin.64 Streptavidin mutants with nonnatural amino acid were found to bind biotin indicating that they retained their native conformation. When an anthryl group was incorporated at residue 120, its luorescence is markedly decreased upon biotin binding due to suppression of energy transfer from excited tryptophan. In addition, when the modiied amino acids 7-methoxycoumaryl-alanine was incorporated at this position, the luorescence intensity was modulated by biotin binding. More recently, this approach was used to incorporate a position-speciic luorophorequencher pair (β-anthraniloyl-L-alpha,beta-diaminopropionic acid (atnDap) as a luorophore and p-nitrophenylalanine (ntrPhe) as a quencher) into streptavidin.65 Such an approach holds great promise for designing proteins that can “sense” binding of native and nonnative ligands present at very low concentrations.66,67 Another example of the incorporation of two unnatural luorescent groups into single protein was shown by Hecht and coworkers.68 Dihydrofolate reductase (DHFR) was engineered with a fusion peptide at its N-terminus that contains the quencher (Nβ-dabcyl-1,2-diaminopropionic acid) and luorophore (7-azatryptophan) incorporated via an engineered HIV-1 protease cleavage site. In vitro expression involved use of both an amber suppressor tRNA and a four-base anticodon tRNA. he resulting DHFR fusion complex exhibited an increased in luorescence intensity upon treatment with HIV-protease, indicating speciic cleavage of the polypeptide chain carrying the luorescence donor. Kajihara et al69 describe the synthesis of several new luorescent amino acids and their incorporation into proteins by E.coli cell free translation system. he amino acid derivates were based on p-amino-phenylalanine that was carrying labels at its amino group. -FL, 558 and 576 resulted in eicient incorporation into streptavidin using tRNAs with four base codon. -FL and 558 were then incorporated into calmodulin at various positions using two diferent four base codons tRNAs and the calmodulin conformational changes studied using FRET technique.

Detection of Truncation Mutations in Gene Products

Truncations in a protein can occur due to frameshit and point mutations which result in the creation of a premature stop codon in the reading frame of a gene. Such truncated polypeptides

162

Cell-Free Protein Expression

can be detected by translating a speciic region of the DNA (mRNA) corresponding to the target gene in a cell-free translation system in the presence of radioactive labels (e.g., 35S-methionine or 14 C-leucine) and then analyzing SDS-PAGE and autoradiography. Such an approach known as the protein truncation test (PTT) has been successfully used for the analysis of truncating mutations in a variety of cancer-linked genes including BRCA1/BRCA2, ATM, MHS2, MLH1, APC.70-74 However, in conventional PTT, the use of radioactive isotopes presents several problems including detection time (>8 hours), which is critical for high-throughput analysis. For this reason, it would be highly advantageous to replace radioactivity with a more rapid means of detection. To overcome this, a non-isotopic, high-throughput PTT method was developed using cell-free protein synthesis and incorporation of biotin and -FL.51 he incorporation of luorescent moieties into nascent proteins or protein fragments using cell-free expression labeling allowed detection of premature chain truncating mutations that occur in variety of tumor suppressor proteins. Traverso et al75 describes the detection of APC protein truncating mutations using multicolor in vitro translation in conjunction with digital PCR approach. he tRNAs charged with lysine and carrying -FL, -TR, -TMR and 650 were used in coupled transcription and translation reactions and their use resulted in generation of labeled gene products that were visualized on the gel and imaged using laser scanner. No incorporation was observed for the Alexa dye family. FluoroTect GreenLys was also used to analyze the size of the ibulin-5 gene products from 5 patients with cutis laxa.76 he in vitro translation and detection of truncated gene products proved that heterozygous mutations in ibulin-5 gene can indeed cause the disease. he in vitro expression eiciency was also recently studied using cell-free system and FluoroTect GreenLys for apolipoprotein APOA5 polymorphs, that are associated with diferences in plasma trigliceryde levels.77 he authors concluded that the polymorphisms observed have no signiicant inluence on the translation eiciency of APOA5 variants.

Future Perspectives

his review focused mainly on basic aspects and some applications of cell-free expression labeling with luorescent labels. his approach has clearly proven to be a useful tool for studying structure and function, expression and protein-protein interactions. While useful and fast, the cell-free approach has several limitations. One of these limitations is the need for relatively complex preparation of the labeled tRNAs. Another is the limited amount of the target proteins that can be obtained using cell-free systems. he approach based on evolving orthogonal tRNAs and synthetase pairs for in-vivo incorporation of engineered proteins does not sufer from the many limitations of the cell-free systems. A considerable progress has been made in this area28,29 and recently indirect78-80 as well as direct81 in vivo luorescent expression labeling has been achieved using this methodology. It is likely that in the future the cell-free method will still be used for the rapid screening applications while the cellular engineering with luorescent amino acids will be used for large scale expression projects.

References

1. Templin MF, Stoll D, Schrenk M et al. Protein microarray technology. Drug Discov Today 2002; 7(15):815-22. 2. Jager S, Brand L, Eggeling C. New luorescence techniques for high-throughput drug discovery. Curr Pharm Biotechnol 2003; 4(6):463-76. 3. Patton WF. A thousand points of light: the application of luorescence detection technologies to two-dimensional gel electrophoresis and proteomics. Electrophoresis 2000; 21(6):1123-44. 4. Tim RC, Kautz RA, Karger BL. Ultratrace analysis of drugs in biological luids using ainity probe capillary electrophoresis: analysis of dorzolamide with luorescently labeled carbonic anhydrase. Electrophoresis 2000; 21(1):220-6. 5. Giepmans BN, Adams SR, Ellisman MH et al. he luorescent toolbox for assessing protein location and function. Science 2006; 312(5771):217-24. 6. Richards DP, Stathakis C, Polakowski R et al. Labeling efects on the isoelectric point of green luorescent protein. J Chromatogr A 1999; 853(1-2):21-5. 7. Tsien RY, he green luorescent protein. Annu Rev Biochem 1998; 67:509-44.

Cell-Free Protein Expression Labeling with Fluorophores

163

8. Zhang J, Campbell RE, Ting AY et al. Creating new luorescent probes for cell biology. Nat Rev Mol Cell Biol 2002; 3(12):906-18. 9. Gross LA, Baird GS, Hofman RC et al. he structure of the chromophore within DsRed, a red luorescent protein from coral. Proc Natl Acad Sci USA 2000; 97(22):11990-5. 10. Evans Jr TC, Xu MQ. Intein-mediated protein ligation: harnessing nature’s escape artists. Biopolymers 1999; 51(5):333-42. 11. Dawson PE, Kent SB. Synthesis of native proteins by chemical ligation. Annu Rev Biochem 2000; 69:923-60. 12. Muralidharan V, Muir TW. Protein ligation: an enabling technology for the biophysical analysis of proteins. Nat Methods 2006; 3(6):429-38. 13. Griin BA, Adams SR, Jones J et al. Fluorescent labeling of recombinant proteins in living cells with FlAsH. Methods Enzymol 2000; 327:565-78. 14. Adams SR, Campbell RE, Gross LA et al. New biarsenical ligands and tetracysteine motifs for protein labeling in vitro and in vivo: synthesis and biological applications. J Am Chem Soc 2002; 124(21):6063-76. 15. Gaertner HF, Oford RE. Site-speciic attachment of functionalized poly(ethylene glycol) to the amino terminus of proteins. Bioconjug Chem 1996; 7(1):38-44. 16. Nirenberg M, Leder P. RNA Codewords and Protein Synthesis. he Efect of Trinucleotides Upon the Binding of Srna to Ribosomes. Science 1964; 145:1399-407. 17. Khorana HG, Buchi H, Ghosh H et al. Polynucleotide synthesis and the genetic code. Cold Spring Harb Symp Quant Biol 1966; 31:39-49. 18. Richmond MH. he efect of amino acid analogues on growth and protein synthesis in microorganisms. Bacteriol Rev 1962; 26:398-420. 19. Krieg UC, Walter P, Johnson AE. Photocrosslinking of the signal sequence of nascent preprolactin to the 54-kilodalton polypeptide of the signal recognition particle. Proc Natl Acad Sci USA 1986; 83(22):8604-8. 20. Krieg UC, Johnson AE, Walter P. Protein translocation across the endoplasmic reticulum membrane: identiication by photocross-linking of a 39-kD integral membrane glycoprotein as part of a putative translocation tunnel. J Cell Biol 1989; 109(5):2033-43. 21. hrit RN, Andrews DW, Walter P et al. A nascent membrane protein is located adjacent to ER membrane proteins throughout its integration and translation. J Cell Biol 1991; 112(5):809-21. 22. Noren CJ, Anthony-Cahill SJ, Griith MC et al. A general method for site-speciic incorporation of unnatural amino acids into proteins. Science 1989; 244(4901):182-8. 23. Anthony-Cahill SJ, Griith MC, Noren CJ et al. Site-speciic mutagenesis with unnatural amino acids. Trends Biochem Sci 1989; 14(10):400-3. 24. Noren CJ, Anthony-Cahill SJ, Suich DJ et al. In vitro suppression of an amber mutation by a chemically aminoacylated transfer RNA prepared by runof transcription. Nucleic Acids Res 1990; 18(1):83-8. 25. horson JS, Cornish VW, Barrett JE et al. A biosynthetic approach for the incorporation of unnatural amino acids into proteins. Methods Mol Biol 1998; 77:43-73. 26. Furter R. Expansion of the genetic code: site-directed p-luoro-phenylalanine incorporation in Escherichia coli. Protein Sci 1998. 7(2):419-26. 27. Budisa N, Minks C, Medrano FJ et al. Residue-speciic bioincorporation of nonnatural, biologically active amino acids into proteins as possible drug carriers: structure and stability of the per-thiaproline mutant of annexin V. Proc Natl Acad Sci USA 1998; 95(2):455-9. 28. Wang L, Xie J, Schultz PG. Expanding the genetic code. Annu Rev Biophys Biomol Struct 2006; 35:225-49. 29. Link AJ, Tirrell DA. Reassignment of sense codons in vivo. Methods 2005; 36(3):291-8. 30. Tan Z, Blacklow SC, Cornish VW et al. De novo genetic codes and pure translation display. Methods 2005; 36(3):279-90. 31. Ramaswamy K, Saito H, Murakami H et al. Designer ribozymes: programming the tRNA speciicity into lexizyme. J Am Chem Soc 2004; 126(37):11454-5. 32. Murakami H, Ohta A, Goto Y et al. Flexizyme as a versatile tRNA acylation catalyst and the application for translation. Nucleic Acids Symp Ser (Oxf ) 2006; 50:35-6. 33. Bessho Y, Hodgson DR, Suga H. A tRNA aminoacylation system for nonnatural amino acids based on a programmable ribozyme. Nat Biotechnol 2002; 20(7):723-8. 34. Olejnik J, Krzymanska-Olejnik E, Mamaev S et al. Methods for the preparation of chemically misaminoacylated tRNA via protective groups, US Pat Appl # 20030219780. 2003. 35. Sisido M, Ninomiya K, Ohtsuki T et al. Four-base codon/anticodon strategy and non-enzymatic aminoacylation for protein engineering with nonnatural amino acids. Methods 2005; 36(3):270-8.

164

Cell-Free Protein Expression

36. Taira H, Hohsaka T, Sisido M. In vitro selection of tRNAs for eicient four-base decoding to incorporate nonnatural amino acids into proteins in an Escherichia coli cell-free translation system. Nucleic Acids Res 2006; 34(5):1653-62. 37. Ma C, Kudlicki W, Odom OW et al. In vitro protein engineering using synthetic tRNA(Ala) with diferent anticodons. Biochemistry 1993; 32(31):7939-45. 38. Hohsaka TA, Murakami Y, Sisido M. Incorporation of nonnatural amino acids into streptavidin through in vitro frame-shit suppression. J Am Chem Soc 1996; 118(40):9778-9779. 39. Kramer G, Kudlicki W, Hardesty B. In vitro engineering using synthetic tRNAs with altered anticodons including four-nucleotide anticodons. Methods Mol Biol 1998; 77:105-16. 40. Hohsaka T, Ashizuka Y, Taira H et al. Incorporation of nonnatural amino acids into proteins by using various four-base codons in an Escherichia coli in vitro translation system. Biochemistry 2001; 40(37):11060-4. 41. Hohsaka T, Ashizuka Y, Murakami H et al. Five-base codons for incorporation of nonnatural amino acids into proteins. Nucleic Acids Res 2001; 29(17):3646-51. 42. Magliery TJ, Anderson JC, Schultz PG. Expanding the genetic code: selection of eicient suppressors of four-base codons and identiication of “shity” four-base codons with a library approach in Escherichia coli. J Mol Biol 2001; 307(3):755-69. 43. Anderson JC, Magliery TJ, Schultz PG. Exploring the limits of codon and anticodon size. Chem Biol 2002; 9(2):237-44. 44. Rodriguez EA, Lester HA, Dougherty DA. In vivo incorporation of multiple unnatural amino acids through nonsense and frameshit suppression. Proc Natl Acad Sci USA 2006; 103(23):8650-5. 45. Gite S, Mamaev S, Olejnik J et al. Ultrasensitive luorescence-based detection of nascent proteins in gels. Anal Biochem 2000; 279(2):218-25. 46. Mamaev S, Olejnik J, Olejnik EK et al. Cell-free N-terminal protein labeling using initiator suppressor tRNA. Anal Biochem 2004; 326(1):25-32. 47. Olejnik J, Gite S, Mamaev S et al. N-terminal labeling of proteins using initiator tRNA. Methods 2005; 36(3):252-60. 48. Olejnik J, Sonar S, Krzymanska-Olejnik E et al. Rothschild, Photocleavable biotin derivatives: a versatile approach for the isolation of biomolecules. Proc Natl Acad Sci USA 1995; 92(16):7590-4. 49. Nemoto N, Miyamoto-Sato E, Husimi Y et al. In vitro virus: bonding of mRNA bearing puromycin at the 3’-terminal end to the C-terminal end of its encoded protein on the ribosome in vitro. FEBS Lett 1997; 414(2):405-8. 50. Doi N, Takashima H, Kinjo M et al. Novel luorescence labeling and high-throughput assay technologies for in vitro analysis of protein interactions. Genome Res 2002; 12(3):487-92. 51. Gite S, Lim M, Carlson R et al. A high-throughput nonisotopic protein truncation test. Nat Biotechnol 2003; 21(2):194-7. 52. Kobs G, Hurst R, Betz Net al. FluoroTect GreenLys in vitro Translation Labeling System. Promega Notes 2001; 77:23–27. 53. Lien L, Ananda P, Seneviratne K et al. Site-speciic biosynthetic incorporation of a luorescent tag into proteins via Cysteine-tRNA(Cys). Anal Biochem 2002; 307(2):252-7. 54. Johnson AE. Protein translocation across the ER membrane: a luorescent light at the end of the tunnel. Trends Biochem Sci 1993; 18(12):456-8. 55. Cornish VW, Benson DR, Altenbach CA et al. Site-speciic incorporation of biophysical probes into proteins. Proc Natl Acad Sci USA 1994; 91(8):2910-4. 56. Steward LE, Collins CS, Gilmore MA et al. In vitro site-speciic incorporation of luorescent probes into beta-galactosidase. J Amer Chem Soc 1997; 119:6-11. 57. Turcatti G, Nemeth K, Edgerton MD. Probing the structure and function of the tachykinin neurokinin-2 receptor through biosynthetic incorporation of luorescent amino acids at speciic sites. J Biol Chem 1996; 271(33):19991-8. 58. Turcatti G, Nemeth K, Edgerton MD et al. Fluorescent labeling of NK2 receptor at speciic sites in vivo and luorescence energy transfer analysis of NK2 ligand-receptor complexes. Receptors Channels 1997; 5(3-4):201-7. 59. Cohen BE, McAnaney TB, Park ES et al. Probing protein electrostatics with a synthetic luorescent amino acid. Science 2002; 296(5573):1700-3. 60. Coleman MA, Lao VH, Segelke BW et al. High-throughput, luorescence-based screening for soluble protein expression. J Proteome Res 2004; 3(5):1024-32. 61. Nakajima D, Saito K, Yamakawa H et al. Preparation of a Set of Expression-Ready Clones of Mammalian Long cDNAs Encoding Large Proteins by the ORF Trap Cloning Method. DNA Res 2005; 12(4):57-67. 62. Iborra FJ, Jackson DA, Cook PR. Coupled transcription and translation within nuclei of mammalian cells. Science 2001; 293(5532):1139-42.

Cell-Free Protein Expression Labeling with Fluorophores

165

63. Gur Y, Breitbart H. Mammalian sperm translate nuclear-encoded proteins by mitochondrial-type ribosomes 10.1101/gad.367606. Genes Dev 2006; 20(4):411-416. 64. Murakami H, Hohsaka T, Ashizuka Y et al. Site-directed incorporation of luorescent nonnatural amino acids into streptavidin for highly sensitive detection of biotin. Biomacromolecules 2000; 1(1):118-25. 65. Taki M, Hohsaka T, Murakami H et al. A novel luorescent nonnatural amino acid that can be incorporated into a speciic position of streptavidin. Nucleic Acids Res Suppl 2002; (2):203-4. 66. Hohsaka T, Ashizuka Y, Sisido M. Incorporation of two nonnatural amino acids into proteins through extension of the genetic code. Nucleic Acids Symp Ser 1999; (42):79-80. 67. Taki M, Hohsaka T, Murakami H et al. A nonnatural amino acid for eicient incorporation into proteins as a sensitive luorescent probe. FEBS Lett 2001; 507(1):35-8. 68. Anderson RD 3rd, Zhou J Hecht SM. Fluorescence resonance energy transfer between unnatural amino acids in a structurally modiied dihydrofolate reductase. J Am Chem Soc 2002; 124(33):9674-5. 69. Kajihara D, Abe R, Iijima I et al. FRET analysis of protein conformational change through positionspeciic incorporation of luorescent amino acids. Nat Methods 2006; 3(11):923-9. 70. Powell SM, Petersen GM, Krush AJ et al. Molecular diagnosis of familial adenomatous polyposis. N Engl J Med 1993; 329(27):1982-7. 71. Powell SM, Zilz N, Beazer-Barclay Y et al. APC mutations occur early during colorectal tumorigenesis. Nature 1992; 359(6392):35-7. 72. Den Dunnen JT, Van Ommen GJ. The protein truncation test: A review. Hum Mutat 1999; 14(2):95-102. 73. Roest PA, Roberts RG, Sugino S et al. Protein truncation test (PTT) for rapid detection of translation-terminating mutations. Hum Mol Genet 1993; 2(10):1719-21. 74. Roest PA, Roberts RG, van der Tuijn AC et al. Protein truncation test (PTT) to rapidly screen the DMD gene for translation terminating mutations. Neuromuscul Disord 1993; 3(5-6):391-4. 75. Traverso G, Diehl F, Hurst R et al. Multicolor in vitro translation. Nat Biotechnol 2003; 21(9):1093-7. 76. Markova D, Zou Y, Ringpfeil F et al. Genetic heterogeneity of cutis laxa: a heterozygous tandem duplication within the ibulin-5 (FBLN5) gene. Am J Hum Genet 2003; 72(4):998-1004. 77. Talmud PJ, Palmen J, Putt W et al. Humphries, Determination of the functionality of common APOA5 polymorphisms. J Biol Chem 2005; 280(31):28215-20. 78. Beatty KE, Xie F, Wang Q et al. Tirrell, Selective dye-labeling of newly synthesized proteins in bacterial cells. J Am Chem Soc 2005; 127(41):14150-1. 79. Deiters A, Schultz PG. In vivo incorporation of an alkyne into proteins in Escherichia coli. Bioorg Med Chem Lett 2005; 15(5):1521-4. 80. Beatty KE, Liu JC, Xie F et al. Fluorescence visualization of newly synthesized proteins in Mammalian cells. Angew Chem Int Ed Engl 2006; 45(44):7364-7. 81. Wang J, Xie J, Schultz PG. A genetically encoded luorescent amino acid. J Am Chem Soc 2006; 128(27):8738-9.

Chapter 14

Cell-Free Synthesis of Deined Protein Conjugates by Site-Directed Cotranslational Labeling Michael Gerrits, Jan Strey, Iris Claußnitzer, Uritza von Groll, Frank Schäfer, Martina Rimmele and Wolfgang Stiege*

Abstract

P

roteins provided with unique functional groups such as ainity labels or luorescence moieties ofer high potential in many biotechnological or biomedical investigations, e.g., immobilization studies or high throughput screenings. An attractive alternative to known posttranslational methods of protein modiication is the site-directed cotranslational incorporation of unnatural amino acids. Here we point out diferent aspects of this method with regard to the synthesis of protein conjugates bearing diferent functionalities. Moreover we describe a cell-free system enabling tRNA-based synthesis of modiied proteins even with large functional groups in high yields. his specialized cell-free system contains a substantially decreased level of release factor 1 (RF1) and originates from an E. coli strain encoding a tagged RF1 variant. We present the eicient amber suppressor tRNA-mediated incorporation of the unnatural amino acid biocytin (biotinylated lysine) into stoichiometrically deined protein conjugates, containing the biotin label at the desired position. As potential applications we demonstrate the usefulness of the system for the immobilization of biotinylated proteins directly from a cell-free system and for interaction studies. Finally we show the cotranslational incorporation of a large luorescent amino acid using our system.

Protein Conjugates

he ability to incorporate a wide array of functional groups at speciic sites in proteins provides a powerful tool for characterization of protein function and the development of protein tools or therapeutics. In this context protein conjugates play a key role. Protein conjugates are used in biophysical and functional analytics, they are tools for biomedical research, medical diagnostics or are applied as therapeutically active substances.1,2 Traditionally the production of protein conjugates is an important step on the way to tailor proteins for the analysis of biological processes on a molecular level. he conjugation of proteins with tags such as biotin permits the selective binding of proteins to surfaces. In this way microarrays, beads and nano-particles can be equipped with proteins. Crosslinkers permit linking of proteins among themselves or with other molecules, e.g., nucleic acids. A conjugation of proteins with spectroscopic reporter groups is of signiicant importance, permitting the study of protein interactions or the localization of proteins within cellular structures and on protein chips. Protein conjugates become increasingly important with *Corresponding Author: Wolfgang Stiege—RiNA Netzwerk RNA Technologien GmbH, Takustraße 3, 14195 Berlin, Germany. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

Cell-Free Synthesis of Deined Protein Conjugates by Site-Directed Cotranslational Labeling

167

the development of innovative products for therapeutics and diagnostics. A prominent example is the equipment of proteins with polyethylenglycol chains, which improves their solubility and pharmacokinetics.3,4 Conventionally protein conjugates are synthesized by posttranslational modiication of the puriied proteins with functional groups.5 Naturally occurring reactive side chains within the proteins, preferably lysine or cysteine, are treated with commercially available reactive reagents. In certain cases posttranslational modiication is a viable tool for the desired purpose but oten it is not advantageous to use the natural occurring amino acid side chains for labeling, since the label is introduced at multiple sites and multiple protein derivatives arise. Moreover, typical posttranslational labeling methods may interfere with protein function and solubility, maintaining protein activity only to a minor degree.6 A frequently chosen way to produce site-speciically deined protein conjugates comprises the design of protein variants with a singular cysteine residue that is subsequently used for the introduction of the desired modiication.7-9 Typically several variants of the desired protein must be synthesized, puriied and modiied, until a variant with optimal characteristics is found. he described methodology is complex and laborious and usually hampers the management of a larger number of protein sequences in parallel, thereby rendering it hardly compatible with high throughput concepts required by the pharmaceutical industry. here is a need for a faster and easier to handle methodology that allows the production of stoichiometrically and site-speciically deined protein conjugates maintaining the highest possible degree of protein solubility and activity. An alternative and generally applicable strategy should allow the labeling of proteins site-speciically even in the presence of many identical amino acid side chains. he most reliable method to produce deined site-speciic modiications is the combination of tRNAs carrying a functionalized amino acid with a cell-free protein expression system.

Cell-Free Systems as a Tool for Protein Conjugate Production

Cell-free protein biosynthesis represents a valuable tool, that allows an economic and parallel synthesis of a large number of proteins in analytical or semipreparative scale.10,11 An outstanding perspective of cell-free protein biosynthesis is the production of proteins containing unnatural amino acids at deined positions.12-16 An advantage of cell-free systems is the fact that they are open systems. hus, they can easily be supplemented with precharged tRNAs.17,18 A common method to synthesize site-speciically modiied proteins containing unnatural amino acids is based on the use of suppressor tRNAs, which are aminoacylated with a desired unnatural amino acid, subsequently added to a cell-free translation system and which insert the amino acid cotranslationally and site-speciically into the growing peptide chain (Fig. 1). With cell-free systems even amino acids bearing large functional side chains such as biotin or luorescent moieties can be incorporated into proteins. hus, the production of protein conjugates in one step is possible, avoiding the need of posttranslational modiication methods with their caveats. In addition the concentration of precharged tRNA required for the synthesis of adequate amounts of modiied proteins its well to the achievable protein yield in cell-free batch systems. Hence it is highly advantageous to use “tRNA-mediated” incorporation of unnatural amino acids in cell-free protein biosynthesis systems for the production of protein conjugates. One of the irst experiments that provided a basis for engineered tRNA mediated site-directed incorporation of unnatural amino acids was performed in 1957, when the acceptance of a irst unnatural amino acid (selenomethionine) by the protein biosynthesis machinery could be shown.19 he fact that a misacylated tRNA could lead to protein production and therefore that codon recognition occurs independent from the amino acid attached to the tRNA veriied the “adapter hypothesis”.20 his implied that tRNAs could in principle be charged with arbitrary noncognate amino acids which could then be incorporated into proteins. he irst amber suppressor tRNAs were discovered as intergenic suppressors.21 hus nonsense codons, which are normally not decoded by endogenous tRNAs but serve as a signal for translation termination, were shown to have the potential to be recoded by changing the tRNA anticodon.

168

Cell-Free Protein Expression

Figure 1. Scheme of site-specific unnatural amino acid incorporation. The site-directed introduction of a certain unnatural amino acid, e.g., modified with a biotin of fluorescent moiety, into a target protein can be achieved by the use of an amber suppressor tRNA, that has been chemically misacylated with the desired amino acid. A codon in the gene of interest is mutated to TAG (nonsense) by site-directed mutagenesis. During protein synthesis this nonsense codon is only recognized by the chemically aminoacylated suppressor tRNA. Thus the addition of the acylated amber suppressor tRNA to a cell-free translation reaction results in the specific synthesis of the desired modified protein.

Meanwhile natural mechanisms have been elucidated, in which non canonical amino acids are inserted based on nonsense suppressor tRNAs. he cotranslational insertion of selenocysteine in many organisms is directed by a specialized suppressor tRNA, that reads the UGA (opal) codon.22 Another example is the site-directed incorporation of pyrrolysine. In some methyltransferase encoding genes of methanogenic archaea the nonsense codon UAG (amber) serves as a sense codon and is decoded by an amber suppressor tRNA, that inserts pyrrolysine.23,24 A corresponding mechanism is used by the recently developed “orthogonal systems” enabling the in vivo site-directed incorporation of small useful amino acids.25-29 hese systems are based on the coexpression of an engineered aminoacyl tRNA synthetase-variant and an appropriate suppressor tRNA in a suitable host cell. P-acetyl-phenylalanine and p-azido-phenylalanine for instance are amino acids bearing chemical reactivities difering from those of the canonical amino acids, allowing the posttranslational site-directed conjugation of the synthesized proteins with appropriate functionalized reagents.30-34 At present in vivo incorporation is restricted to comparatively small unnatural amino acids with structures resembling those of the canonical amino acids, preventing the synthesis of protein conjugates in one step. he ribosome however is able to accept a broad range of unnatural amino acids, among them large amino acids with structures signiicantly diferent to the canonical amino acids. he biotin-containing amino acid biocytin for example can be site-directed incorporated into proteins ater microinjection of biotinylated amber suppressor tRNAs into Xenopus oocytes.35

Cell-Free Synthesis of Deined Protein Conjugates by Site-Directed Cotranslational Labeling

169

In cell-free systems the site-directed incorporation of unnatural amino acids became possible by the development of chemical aminoacylation, which enables the aminoacylation of an amber suppressor tRNA, that is not recognized by naturally occurring aminoacyl tRNA synthetases.36-43 Even amino acids which cannot be taken up by cells or activated by aminoacyl tRNA synthetases, like acids containing biotin35 or large luorescent side chains can be attached to the tRNA.44 he strategy introduces the unnatural amino acid into the growing peptide chain by a stoichiometric approach, introducing one unnatural amino acid per tRNA and protein. he tRNA is added to the translation system as an aminoacylated molecule and is exhausted during expression of the desired protein. Recharging of the tRNA by canonical amino acids is excluded, since the tRNA does not contain recognition elements for aminoacyl tRNA synthetases present in the respective expression system. In this way, the desired unnatural amino acid is exclusively incorporated at the predicted position of the growing protein chain and a competition of canonical amino acids for this position is excluded. hus site-speciically deined pure derivatives of the desired modiied proteins emerge. In recent years alternative methods for the aminoacylation of tRNAs with the desired amino acids have been developed in addition to chemical aminoacylation. hese methods include the use of peptide nucleic acids (PNAs)45,46 and ribozymes.47,48 he most promising approaches for tRNA-mediated site-directed unnatural amino acid incorporation by precharged tRNAs in cell-free systems involve the use of nonsense suppressor tRNAs40-43 and frameshit suppressor tRNAs.44,46 he unnatural amino acids are preferably incorporated at the amber stop codon (UAG) or at four base codons with the possibility to incorporate two or more unnatural amino acids into a single protein.49 Recently a combination of the amber and the four base strategy has been described.50 A special case is the use of the start codon by modiied initiator tRNAs.51

Incorporation Eiciencies of Different Unnatural Amino Acids

Meanwhile several useful unnatural amino acids have been incorporated into proteins based on the use of precharged tRNAs in cell-free systems. Among them were amino acids for structural investigations,40,52-54 with selective reactive side chains,55-57 Stable isotope labelled amino acids58,59 and luorescent amino acids.44,49,50,55,60-64 Unnatural amino acids are known to be incorporated with diverse eiciencies.55 For example, the small hydrophobic luorescent amino acid naphthylalanine is reported to be very well incorporated in cell-free systems, while other unnatural amino acids are hardly incorporated.44,62 We could demonstrate the same in our standard, i.e., untreated E. coli lysate. An almost complete suppression of termination was achieved by an amber suppressor tRNA (tRNAPheCUA from yeast) charged with 2-naphthylalanine without further manipulation of the system, whereas the suppression eiciency with the bulky amino acid biocytin (biotinylated lysine) was very low in this system (Fig. 2). here are many factors inluencing the ability of a certain amino acid to be incorporated into a protein, including binding of the aminoacylated tRNA to elongation factor Tu (EF-Tu), competition of the tRNA for codon recognition with Release Factor 1 (RF1), binding of the tRNA to the ribosome by cognate codon-anticodon interaction, behavior of the amino acid in the peptidyl transferase center, and channeling of the amino acid within the growing peptide through the ribosome tunnel. Suppression eiciencies of diferent amber suppressor tRNAs have been traced back to structural features of the tRNAs, mainly to the anticodon arm.65-69 One possibility to improve suppression eiciency is to screen for more eicient suppressor tRNAs.64,70 Screening diferent amber suppressor tRNAs in order to increase the eiciency of biocytin incorporation, we could detect considerable diferences in suppression eiciencies even among tRNAs with optimized anticodon arms (will be published elsewhere).

Depletion of RF1 from a Cell-Free System Results in Eicient Incorporation of Biocytin

he major factor regarding the suppression eiciency of an amber suppressor tRNA is its ability to compete with release factor 1 (RF1) for decoding the amber (UAG) codon. Preference for the charged suppressor tRNA by the ribosome results in the synthesis of full-length protein with

170

Cell-Free Protein Expression

Figure 2. Efficiency of amber suppression with different unnatural amino acids. tRNAPheCUA from yeast was chemically acylated with naphthylalanine (I) and biocytin (biotinylated lysine) (II), respectively. The resulting aminoacylated amber suppressor tRNAs were used in cell-free protein biosynthesis reactions with the EasyXpress standard batch system (distributed by QIAGEN GmbH, Hilden, Germany). The gene for the Fatty Acid Binding Protein from bovine liver (FABP) carrying an amber codon at position 88 served as the template. Protein synthesis was performed in the presence of 14C leucine. Following protein synthesis and SDS-PAGE radiolabeled proteins were visualized with a Typhoon 8600 variable mode imager (GE Healthcare Europe GmbH, Munich, Germany). The efficiency of suppression strongly depends on the nature of the unnatural amino acid. With biotinylated lysine release factor 1 (RF1) mediated termination is favored, whereas suppression is predominant with naphthylalanine. w/o = control reaction without amber suppressor tRNA.

the desired unnatural amino acid incorporated, while selection of RF1 results in the synthesis of truncated protein. herefore, a decrease in the RF1 content of the expression system will increase the suppression eiciency of added tRNAs. Diferent approaches have been described to increase suppression eiciency of amber suppressor tRNAs via a manipulation of RF1 activity, e.g., RF1-inhibitory aptamers71 or antibodies72 have been used in in vitro translation systems. Another approach has been the use of E. coli strains with thermosensitive RF1 variants.73 In lysates of these strains the activity of RF1 could be diminished considerably by heat inactivation which led to a signiicant increase in suppression rate. However, overall protein synthesis rates were relatively low in this system. In principle, the “pure system” which is synthetically reconstituted from all single translation factors provides another possibility to lower termination rates by just omitting RF1 alltogether.74 In our laboratory we developed an alternative strategy to produce RF1-depleted lysates, based on genetic engineering of a suitable Escherichia coli strain. he starting point was one of our proven high-producer E. coli strains to guarantee optimal lysate activity. Since the termination factor RF1 is essential for cell growth and a construction of knock out mutants is not possible,75 we created a strategy involving the construction of an E. coli high-producer strain containing the chromosomal sequence of an RF1 variant with a C-terminal tag instead of the wild type RF1 gene

Cell-Free Synthesis of Deined Protein Conjugates by Site-Directed Cotranslational Labeling

171

Figure 3. Genomic organization of the PrfA encoding region in E. coli wild-type (A) and DSM 15756 strain (B). To enable Release Factor 1 (RF1) depletion during lysate preparation the E. coli wild-type locus for RF1 (prfA) was replaced via homologous recombination by a mutant prfA’ gene encoding C-terminally tagged RF1. As selection marker the kanamycin resistance gene (kanR) from transposon Tn903 was used. To maintain expression of hemK in the mutant strain a ribosome binding site (rbs) derived from phage T7 gene10 was introduced upstream of hemK. For production of RF1-depleted lysates affinity chromatography was performed with extracts from strain DSM 15756.

(Fig. 3). Due to the tag an almost complete removal of RF1 using ainity chromatography during lysate production was achieved. Using such an RF1-depleted lysate, we could demonstrate an 8-fold increase of Biotin-incorporation into a model protein (Fig. 4). A slight additional reduction of RF1 activity was observed with a polyclonal RF1-antibody (Fig. 5), but this had only a marginal auxiliary efect on biotin incorporation. A complete switch-of of RF1 activity led to a reduction of protein synthesis (data not shown). hese results suggest, that our “depletion of RF1-strategy” is comparable to an inactivation of RF1 by antibodies in its efects on the incorporation of unnatural amino acids into proteins (Fig. 5). RF1-depleted lysates however are easier to handle, more cost efective and show a better reproducibility. Furthermore, our lysate preparation is associated with removal of biotinylated endogenous E. coli proteins (Fig. 6), which otherwise may interfere with subsequent biotin-based applications.

Figure 4. Effect of RF1 removal on the incorporation of biotinylated lysine. A) During translation elongation amber suppressor tRNA competes with Release Factor 1 (RF1) for the amber codon. Selection of amber suppressor tRNA by the ribosome results in the synthesis of full-length protein ( = suppression product (SP)), whereas selection of RF1 leads to the synthesis of a truncated peptide referred to as termination product (TP). Hence, the removal of RF1 is expected to increase suppression. B) The Fatty Acid Binding Protein from bovine liver (FABP) containing an amber codon at position 88 within the coding sequence was synthesized in a lysate from strain DSM 15756 before and after RF1 removal via affinity chromatography. For radiolabeling of the synthesized proteins 14C leucine was added. Reactions were performed in the absence and presence of an amber suppressor tRNA chemically charged with biotinylated lysine (Biocytin tRNA). Following protein synthesis and SDS-PAGE radiolabeled peptides were visualized (autoradiogram) and quantified (graph) by using a Typhoon 8600 variable mode imager. The synthesis of the suppression product corresponding to biotinylated full-length FABP is clearly enhanced after RF1 removal. More than 80% suppression was obtained in the RF1 deficient lysate.

172

Cell-Free Protein Expression

Figure 5. RF1 activity in standard and depleted E. coli lysates. RF1 activity was defined as the molar ratio of termination to read-through product, the synthesis of which is caused by endogenous misreading tRNAs. Therefore no amber suppressor tRNA was added in this experiment. The gene for the Fatty Acid Binding Protein from bovine liver (FABP) carrying an amber codon at position 88 was expressed in a wild-type and a RF1-depleted lysate, respectively. Reactions (30 μ1) were performed in the presence of radiolabeled leucine (14C) and increasing concentrations of a RF1-specific polyclonal antibody (AB, kindly provided by Prof Sprinzl, University of Bayreuth, Germany). Following protein synthesis and SDS-PAGE protein bands were quantified by using a Typhoon 8600 variable mode imager. In the standard lysate RF1 activity decreases with increasing antibody concentration. The RF1-depleted lysate shows very low RF1 activity, even in the absence of antibody.

Figure 6. Synthesis of a C-terminally biotinylated protein. The gene for E. coli Elongation Factor Tu (EF-Tu) containing an additional amber stop codon (TAG) upstream of its native stop codon (TAA) was expressed in the EasyXpress Protein Synthesis Mini Kit (standard lysate) and EasyXpress Site-Specific Biotin Kit (RF1-depleted lysate), respectively (both kits distributed by QIAGEN GmbH, Hilden, Germany). Reactions were performed in the presence of an amber suppressor tRNA charged with biotinylated lysine. The synthesized proteins were separated by SDS-PAGE and blotted onto a PVDF membrane. Subsequently biotin-containing proteins were detected by using streptavidin-peroxidase and a Typhoon 8600 variable mode imager. The RF1-depleted lysate is characterized by a significantly enhanced level of biotin incorporation. Additionally, the endogenous biotin-containing E. coli protein BCCP (Biotin Carboxyl Carrier Protein) cannot be detected, since it has been removed during lysate preparation together with tagged RF1.

Cell-Free Synthesis of Deined Protein Conjugates by Site-Directed Cotranslational Labeling

173

For most applications not only site-speciically deined, but also stoichiometrically uniform derivatives of the desired protein conjugates with biotin need to be synthesized. he incorporation of the desired unnatural amino acid by a supplemented tRNA is not only in competition to a termination by RF1, but also to a spontaneous hydrolysis of the aminoacylation and to an unwanted read-through due to misreading by endogenous tRNA. It is therefore necessary to harmonise the RF1 content of the lysates, the velocity of protein synthesis and the duration of the reaction. Synthesis of full length protein due to unwanted read-through is negligible as long as some RF1 is present. However, at a complete inhibition of RF1 by antibodies, the synthesis of full length protein due to read-through can reach a signiicant level of its corresponding wild type protein even in the absence of suppressor tRNA. In our RF1-depleted system as well, we could detect some read-through in the absense of suppressor tRNA. For this reason it is highly advisable to arrest the protein biosynthesis reaction before the added tRNA is completely hydrolyzed and becomes unable to compete with endogenous read-through. his is easily achieved by limiting incubation time or amino acid concentration. Using optimized conditions we could demonstrate that the biotinylated tRNA is able to repress read-through by misreading endogenous tRNA completely (Fig. 7). he half-life of the aminoacyl-bond of the tRNA due to spontaneous deacylation has been determined to be between 10 and 30 minutes in our translation system.69 In order to avoid the deprivation of aminoacyl-tRNA due to deacylation, it is necessary to use lysates with fast kinetics of synthesis. We could show for diferent model proteins, that a stoichiometric incorporation of biocytin (one residue per one protein molecule) was provided at a tRNA concentration of 12 μM in a 30 minutes reaction (Fig. 7). Using these conditions up to 7 μM and over 200 μg/ml of site-speciically biotinylated protein was synthesized, i.e., up to 60% of the biotin residues attached to the tRNA were incorporated into protein.

Figure 7. Stoichiometry of site-specific biotin incorporation. Three derivatives of the E. coli EF-Tu gene containing His-Tag encoding sequences at the 3’-end and amber codons at positions 2 (amb2), 184 (amb184) and 396 (amb396), respectively, were used as templates for cell-free protein biosynthesis with the EasyXpress Site-Specific Biotin Kit. Following protein synthesis aliquots from the reactions were incubated with and without streptavidin and subjected to native polyacrylamide gel electrophoresis. Subsequently proteins were analyzed by western blotting using an antibody against His-Tag (Penta His HRP conjugate). Irrespective of the amber codon position synthesized EF-Tu was almost completely shifted by streptavidin indicating a stoichiometric biotin incorporation (i.e., each protein molecule contains one biotin moiety).

174

Cell-Free Protein Expression

Applications of Site-Speciically and Stoichiometrically Deined Protein Conjugates

Biotin moieties in proteins are widely used for the detection and immobilization of protein conjugates, since corresponding biotin-based methods are robust and readily available. In one of our application examples the usefulness of our cell-free system was demonstrated to detect protein protein interactions eiciently. he interaction of elongation factor Tu (EF-Tu) and elongation factor Ts (EF-Ts) was probed using immobilized EF-Tu and biotinylated EF-Ts as the interaction partner (Fig. 8). Diferent sites of EF-Ts were site-speciically biotinylated with the EasyXpress Site-Speciic Biotin Kit (QIAGEN GmbH, Hilden, Germany) to allow its detection in the LiquiChip system (QIAGEN GmbH, Hilden, Germany) by using a Streptavidin-Phycoerythrin conjugate as a reporter. We could show that the detection signal was strongly dependent on the site of biotinylation. Our identiication of the best site (position 47) for biotinylation to achieve a strong signal correlated well with a site in the published structure of the EF-Tu*EF-Ts complex,76 for which a modiication would probably not interfere with complex formation. Accordingly a biotinylation of position 94 protruding into the same direction as the immobilization tag of EF-Tu or a C-terminal biotinylation did not lead to a detectable signal in our experiments. N- or C-terminal labeling positions, which are typically favored for tag addition, did not lead to a satisfying signal strength, instead an internal position was best for prominent signal production. In a second example we present the usefulness of our system to site-speciically biotinylate a wide range of diferent proteins for protein immobilization applications. Various proteins of different size, including prokaryotic translation factors, an aminoacyl tRNA synthetase, single chain antibodies and diferent human proteins, were immobilized on magnetic beads directly from the reaction mixtures (Fig. 9). In our system there is no need for removal of free biotin because we use precharged biotinylated suppressor tRNA as the sole source of biotin, which is almost totally

Figure 8. Optimization of a protein-protein interaction assay via site-specific biotin incorporation. (A) EF-Tu with C-terminal His-Tag was synthesized by cell-free expression using the EasyXpress Linear Template Kit Plus and the EasyXpress Protein Synthesis Mini Kit (QIAGEN GmbH, Hilden, Germany) and subsequently immobilized on LiquiChip Penta-His beads. (B) Different variants of EF-Ts containing biotin at alternative positions (2-B, 47-B, 94-B, 284-B) were generated with the EasyXpress Site-Specific Biotin Kit and subjected to an interaction assay with immobilized EF-Tu. (C) Interaction of immobilized EF-Tu and biotinylated EF-Ts was monitored in QIAGEN’s LiquiChip system using a Streptavidin-Phycoerythrin conjugate. (D) Optimal signal strength is achieved with EF-Ts site-specifically biotinylated at position 47.

Cell-Free Synthesis of Deined Protein Conjugates by Site-Directed Cotranslational Labeling

175

Figure 9. Immobilization of site-specifically biotinylated proteins. A) Various proteins with biotin incorporated at different postions were synthesized using the EasyXpress Site-Specific Biotin Kit (QIAGEN GmbH, Hilden, Germany). For radiolabeling of the proteins reactions were performed in the presence of defined concentrations of 14C leucine. Following protein synthesis and 1:20 dilution of reactions with phosphate buffered saline (PBS) the biotinylated proteins were immobilized on streptavidin-coated magnetic beads (Dynal Biotech ASA, Oslo, Norway). Beads were washed with PBS to remove unspecifically bound proteins. The amounts of immobilized proteins were calculated via liquid scintillation counting. Proteins that were selected for subsequent SDS-PAGE analysis are designated by numbers in squares. B) To analyze homogeneity of the immobilized proteins beads were boiled in Laemmli sample buffer and supernatants subjected to SDS-PAGE. Proteins were visualized by Coomassie staining of the gel. Streptavidin (SA) released from the beads due to boiling is visible at the bottom of the gel. Immobilized site-specifically biotinylated proteins are indicated by small arrows. The upper bands in lanes 4 and 5 correspond to contaminating chaperone DnaK, which was added to the protein synthesis reactions to increase protein solubility. M = Molecular weight marker (protein sizes in kDa are indicated on the left).— = streptavidin-coated magnetic beads treated with lysate lacking specific template DNA and processed in the same manner as the samples.

176

Cell-Free Protein Expression

incorporated into protein. he negligible amount of free biotin due to spontaneous hydrolysis of the amino acyl bond at the tRNA does not perturb following applications. herefore an immobilization of proteins is possible without prior puriication. his allows the immobilization even of proteins that are diicult or impossible to purify. Cyclophilin A, for example, could not be immobilized ater puriication, presumably due to aggregation of the puriied protein. Figure 9 exempliies that proteins of diferent size can be biotinylated in considerable amounts in our system. Up to 150 pmol of protein could be immobilized from a 25 μl synthesis reaction, using only 250 pmol biotinylated tRNA as the source for biotin. We can show in our system that site-speciic labeling of each of two interacting proteins is possible, rendering labeling more lexible for subsequent microarray or phage display applications. As an example two single chain antibodies and their corresponding antigens (Cyclophilin A and Ubiquitin) were expressed in the system and immobilized (Fig. 9). he immobilized antigens could be subsequently detected with high sensitivity by corresponding nonbiotinylated single chain antibodies (data not shown). A template encoding an amber stop codon in an appropriate position can be used for the incorporation of biocytin as well as for other useful unnatural amino acids. In our opinion the site-directed incorporation of luorescent amino acids provides an outstanding beneit allowing the direct and sensitive detection of synthesized proteins, e.g., in single molecule studies and interaction studies including FRET-based investigations. We have charged an amber suppressor tRNA with the luorescent group TMR (Molecular Probes, Eugene, OR, U.S.A.) attached to the ε-aminogroup of lysine. he resulting large luorescent amino acid could be incorporated into proteins within the RF1 depleted system (Fig. 10).

Figure 10. Site-specific incorporation of a fluorescent moiety. Genes for C-terminally His-tagged EF-Ts, N-terminally His-tagged tandem ubiquitin and FABP containing amber codons at positions 47, 2 and 88, respectively, were expressed in the EasyXpress Site-Specific Biotin Kit (QIAGEN GmbH, Hilden, Germany). Instead of biotinylated amber suppressor tRNA supplied with the Kit an amber suppressor tRNA that we charged chemically with the fluorophor -TMR (Invitrogen GmbH, Karlsruhe, Germany) was added to the reactions. All reactions were performed in the presence of radiolabeled leucine (14C). The synthesized proteins were separated by SDS-PAGE and radioactive as well as fluorescent bands visualized by using a Typhoon 8600 variable mode imager. As can be seen from the fluorogram (right panel) -TMR has been incorporated into all model proteins. Termination products (indicated by asterisks) are only visible in the autoradiogram, since they contain no fluorescent moiety.

Cell-Free Synthesis of Deined Protein Conjugates by Site-Directed Cotranslational Labeling

177

Conclusions

We have demonstrated the high yield synthesis of site-speciically and stoichiometrically deined protein conjugates with biotin using precharged tRNAs in an improved cell-free protein biosynthesis system. Since biocytin is diicult to incorporate due to its size, our success advocates the assumption that the system will allow the site-directed incorporation also of other valuable pharmaceutically desired groups or dyes. his could be veriied by the site-directed incorporation of the luorescent TMR. he most important advantage of our cotranslational strategy for the production of sitespeciically deined protein conjugates with biotin is that protein conjugate synthesis is possible in one step, enabling a rapid synthesis of a wide range of biotinylated proteins in parallel. hus the use of posttranslational modiication procedures, which may interfere with solubility and activity of proteins, can be avoided. he usefulness of the system is further advanced by the fact that an immobilization of biotinylated proteins is possible directly from translation reactions without the need for puriication. Additionally the site-directed incorporation of biocytin allows a lexible and gentle handling maintaining the highest possible activity of the biotinylated proteins. Biocytin and luorescent amino acids represent functionalities with exquisite potential for interaction-based applications. We have shown that both modiications can be incorporated deploying the same template encoding an amber stop codon in an appropriate position. In conjunction with Expression-PCR,77 the technology for rapid generation of ready to express templates, our system enables the parallel synthesis and immobilization of a huge number of cotranslationally modiied proteins. In the future this technology will facilitate proteomics-related applications such as protein-protein interaction studies, drug screening, antibody selection, FRET analyses and single molecule detection.

Acknowledgements

We thank Prof. Matthias Sprinzl (University of Bayreuth) for providing the antibody against RF1 and Dr. Zoltán Konthur (MPI for Molecular Genetics, Berlin) for plasmids encoding the two single chain antibodies and the corresponding antigens. his work was supported by the German ministry of research and education (BMBF) and the Senate of Berlin.

References

1. Kapanidis AN, Weiss S. Fluorescent probes and bioconjugation chemistries for single-molecule luorescence analysis of biomolecules. J Chem Phys 2002; 117(24):10953-64. 2. Elvira C, Gallardo A, Roman JS et al. Covalent Polymer-Drug Conjugates. Molecules 2005; 10:114-125. 3. Chapman AP. PEGylated antibodies and antibody fragments for improved therapy: a review. Adv Drug Deliv Rev 2002; 54(4):531-45. 4. Duncan R. Nanomedicines in action. he Pharmaceutical Journal 2004; 273:485-488. 5. Brunner J. New photolabeling and crosslinking methods. Annu Rev Biochem 1993; 62:483-514. 6. Service RF. Unnatural amino acid could prove boon for protein therapeutics. Science 2005; 308(5718):44. 7. Spruijt RB, Wolfs CJ, Verver JW et al. Accessibility and environment probing using cysteine residues introduced along the putative transmembrane domain of the major coat protein of bacteriophage M13. Biochemistry 1996; 35(32):10383-91. 8. Kumita JR, Jain L, Safroneeva E et al. A cysteine-free irely luciferase retains luminescence activity. Biochem Biophys Res Commun 2000; 267(1):394-7. 9. Rosendahl MS, Doherty DH, Smith DJ et al. Site-speciic protein PEGylation: application to cysteine analogs of recombinant human granulocyte colony-stimulating factor. BioProcess International 2005; 3(52-62). 10. Katzen F, Chang G, Kudlicki W. he past, present and future of cell-free protein synthesis. Trends Biotechnol 2005; 23(3):150-6. 11. Swartz J. Developing cell-free biology for industrial applications. J Ind Microbiol Biotechnol 2006; 33(7):476-85. 12. Hohsaka T, Sisido M. Incorporation of nonnatural amino acids into proteins. Curr Opin Chem Biol 2002; 6(6):809-15.

178

Cell-Free Protein Expression

13. Hendrickson TL, de Crecy-Lagard V, Schimmel P. Incorporation of nonnatural amino acids into proteins. Annu Rev Biochem 2004; 73:147-76. 14. Wang L, Schultz PG. Expanding the genetic code. Angew Chem Int Ed Engl 2004; 44(1):34-66. 15. Budisa N. Prolegomena to future experimental eforts on genetic code engineering by expanding its amino acid repertoire. Angew Chem Int Ed Engl 2004; 43(47):6426-63. 16. Magliery TJ. Unnatural protein engineering: producing proteins with unnatural amino acids. Medicinal Chemistry Reviews 2005; 2:303-323. 17. Kurzchalia TV, Wiedmann M, Breter H et al. tRNA-mediated labeling of proteins with biotin. A nonradioactive method for the detection of cell-free translation products. Eur J Biochem 1988; 172(3):663-8. 18. Stiege W, Erdmann VA. he potentials of the in vitro protein biosynthesis system. J Biotech 1995; 41:81-90. 19. Cowie DB, Cohen GN. Biosynthesis by Escherichia coli of active altered proteins containing selenium instead of sulfur. Biochim Biophys Acta 1957; 26(2):252-61. 20. Chapeville F, Lipmann F, Von Ehrenstein G et al. On the role of soluble ribonucleic acid in coding for amino acids. Proc Natl Acad Sci USA 1962; 48:1086-92. 21. Garen A, Garen S, Wilhelm RC. Suppressor genes for nonsense mutations. I. he Su-1, Su-2 and Su-3 genes of Escherichia coli. J Mol Biol 1965; 14(1):167-78. 22. Leinfelder W, Zehelein E, Mandrand-Berthelot MA et al. Gene for a novel tRNA species that accepts L-serine and cotranslationally inserts selenocysteine. Nature 1988; 331(6158):723-5. 23. Srinivasan G, James CM, Krzycki JA. Pyrrolysine encoded by UAG in Archaea: charging of a UAG-decoding specialized tRNA. Science 2002; 296(5572):1459-62. 24. Hao B, Gong W, Ferguson TK et al. A new UAG-encoded residue in the structure of a methanogen methyltransferase. Science 2002; 296(5572):1462-6. 25. Liu DR, Magliery TJ, Pastrnak M et al. Engineering a tRNA and aminoacyl-tRNA synthetase for the site-speciic incorporation of unnatural amino acids into proteins in vivo. Proc Natl Acad Sci USA 1997; 94(19):10092-7. 26. Furter R. Expansion of the genetic code: site-directed p-luoro-phenylalanine incorporation in Escherichia coli. Protein Sci 1998; 7(2):419-26. 27. Ohno S, Yokogawa T, Nishikawa K. Changing the amino acid speciicity of yeast tyrosyl-tRNA synthetase by genetic engineering. J Biochem (Tokyo) 2001; 130(3):417-23. 28. Kowal AK, Kohrer C, RajBhandary UL. Twenty-irst aminoacyl-tRNA synthetase-suppressor tRNA pairs for possible use in site-speciic incorporation of amino acid analogues into proteins in eukaryotes and in eubacteria. Proc Natl Acad Sci USA 2001; 98(5):2268-73. 29. Wang L, Xie J, Schultz PG. Expanding the genetic code. Annu Rev Biophys Biomol Struct 2006; 35:225-49. 30. Behrens C, Nielsen JN, Fan X-J et al. Development of strategies for the site-speciic in vivo incorporation of photoreactive amino acids. Tetrahedron 2000; 56:9443-9. 31. Chin JW, Santoro SW, Martin AB et al. Addition of p-azido-L-phenylalanine to the genetic code of Escherichia coli. J Am Chem Soc 2002; 124(31):9026-7. 32. Deiters A, Cropp TA, Summerer D et al. Site-speciic PEGylation of proteins containing unnatural amino acids. Bioorg Med Chem Lett 2004; 14(23):5743-5. 33. Tsao ML, Tian F, Schultz PG. Selective Staudinger modiication of proteins containing p-azidophenylalanine. Chembiochem 2005; 6(12):2147-9. 34. Ryu Y, Schultz PG. Eicient incorporation of unnatural amino acids into proteins in Escherichia coli. Nat Methods 2006; 3(4):263-5. 35. Gallivan JP, Lester HA, Dougherty DA. Site-speciic incorporation of biotinylated amino acids to identify surface-exposed residues in integral membrane proteins. Chem Biol 1997; 4(10):739-49. 36. Hecht SM, Alford BL, Kuroda Y et al. “Chemical aminoacylation” of tRNA’s. J Biol Chem 1978; 253(13):4517-20. 37. Bruce AG, Atkins JF, Wills N et al. Replacement of anticodon loop nucleotides to produce functional tRNAs: amber suppressors derived from yeast tRNAPhe. Proc Natl Acad Sci USA 1982; 79(23):7127-31. 38. Shih LB, Bayley H. A carbene-yielding amino acid for incorporation into peptide photoainity reagents. Anal Biochem 1985; 144(1):132-41. 39. Baldini G, Martoglio B, Schachenmann A et al. Mischarging Escherichia coli tRNAPhe with L-4’-[3-(triluoromethyl)-3H-diazirin-3-yl]phenylalanine, a photoactivatable analogue of phenylalanine. Biochemistry 1988; 27(20):7951-9. 40. Roesser JR, Xu C, Payne RC et al. Preparation of misacylated aminoacyl-tRNA(Phe)’s useful as probes of the ribosomal acceptor site. Biochemistry 1989; 28(12):5185-95.

Cell-Free Synthesis of Deined Protein Conjugates by Site-Directed Cotranslational Labeling

179

41. Noren CJ, Anthony-Cahill SJ, Griith MC et al. A general method for site-speciic incorporation of unnatural amino acids into proteins. Science 1989; 244(4901):182-8. 42. Ellman J, Mendel D, Anthony-Cahill S et al. Biosynthetic Method for Introducing Unnatural Amino Acids Site-Speciically into Proteins. Meth Enzymol 1991; 202:301-336. 43. Bain JD, Diala ES, Glabe CG et al. Site-speciic incorporation of nonnatural residues during in vitro protein biosynthesis with semisynthetic aminoacyl-tRNAs. Biochemistry 1991; 30(22):5411-21. 44. Hohsaka T, Kajihara D, Ashizuka Y et al. Eicient incorporation of nonnatural amino acids with large aromatic groups into streptavidin in in vitro protein synthesizing systems. J Am Chem Soc 1999; 121:34-40. 45. Ninomiya K, Minohata T, Nishimura M et al. In situ chemical aminoacylation with amino acid thioesters linked to a peptide nucleic acid. J Am Chem Soc 2004; 126(49):15984-9. 46. Sisido M, Ninomiya K, Ohtsuki T et al. Four-base codon/anticodon strategy and non-enzymatic aminoacylation for protein engineering with nonnatural amino acids. Methods 2005; 36(3):270-8. 47. Bessho Y, Hodgson DR, Suga H. A tRNA aminoacylation system for nonnatural amino acids based on a programmable ribozyme. Nat Biotechnol 2002; 20(7):723-8. 48. Kourouklis D, Murakami H, Suga H. Programmable ribozymes for mischarging tRNA with nonnatural amino acids and their applications to translation. Methods 2005; 36(3):239-44. 49. Taki M, Hohsaka H, Taira K et al. Position-Speciic Incorporation of a Fluorophore-Quencher Pair into a Single Streptavidin through Orthogonal Four-Base Codon/Anticodon Pairs. J Am Chem Soc 2002; 124:14568-90. 50. Muranaka N, Hohsaka T, Sisido M. Four-base codon mediated mRNA display to construct peptide libraries that contain multiple nonnatural amino acids. Nucleic Acids Res 2006; 34(1):e7. 51. Gite S, Mamaev S, Olejnik J et al. Ultrasensitive luorescence-based detection of nascent proteins in gels. Anal Biochem 2000; 279(2):218-25. 52. Cornish VW, Mendel D, Schultz PG. Probing Protein-Structure and Function with an Expanded Genetic-Code. Angewandte Chemie-International Edition in English 1995; 34(6):621-633. 53. Dougherty DA. Unnatural amino acids as probes of protein structure and function. Curr Opin Chem Biol 2000; 4(6):645-52. 54. Dedkova LM, Fahmi NE, Golovine SY et al. Enhanced D-amino acid incorporation into protein by modiied ribosomes. J Am Chem Soc 2003; 125(22):6616-7. 55. Cornish VW, Benson DR, Altenbach CA et al. Site-speciic incorporation of biophysical probes into proteins. Proc Natl Acad Sci USA 1994; 91(8):2910-4. 56. Cornish VW, Hahn KM, Schultz PG. Site-speciic protein modiication using a ketone handle. Journal of the American Chemical Society 1996; 118(34):8150-8151. 57. Schmidt RR, Castro-Palomino JC, Retz O. New aspects of glycoside bond formation; 1999. 58. Ellman JA, Volkman BF, Mendel D et al. Site-Speciic Isotopic Labeling of Proteins for Nmr-Studies 1992. 59. Yabuki T, Kigawa T, Dohmae N et al. Dual amino acid-selective and site-directed stable-isotope labeling of the human c-Ha-Ras protein by cell-free synthesis. J Biomol NMR 1998; 11(3):295-306. 60. Turcatti G, Nemeth K, Edgerton MD et al. Probing the structure and function of the tachykinin neurokinin-2 receptor through biosynthetic incorporation of luorescent amino acids at speciic sites. J Biol Chem 1996; 271(33):19991-8. 61. Karginov VA, Mamaev SV, Hecht SM. In vitro suppression as a tool for the investigation of translation initiation. Nucleic Acids Res 1997; 25(19):3912-6. 62. Sisido M, Hohsaka T. Extension of protein functions by the incorporation of nonnatural amino acids. Bull Chem Soc Jpn 1999; 72(7):1409-25. 63. Murakami H, Hohsaka T, Ashizuka Y et al. Site-directed incorporation of luorescent nonnatural amino acids into streptavidin for highly sensitive detection of biotin. Biomacromolecules 2000; 1(1):118-25. 64. Taira H, Hohsaka T, Sisido M. In vitro selection of tRNAs for eicient four-base decoding to incorporate nonnatural amino acids into proteins in an Escherichia coli cell-free translation system. Nucleic Acids Res 2006; 34(5):1653-62. 65. Yarus M. Translational eiciency of transfer RNA’s: uses of an extended anticodon. Science 1982; 218(4573):646-52. 66. Yarus M, Cline SW, Wier P et al. Actions of the anticodon arm in translation on the phenotypes of RNA mutants. J Mol Biol 1986; 192(2):235-55. 67. Yarus M, Cline S, Ratery L et al. he translational eiciency of tRNA is a property of the anticodon arm. J Biol Chem 1986; 261(23):10496-505. 68. Kleina LG, Masson JM, Normanly J et al. Construction of Escherichia coli amber suppressor tRNA genes. II. Synthesis of additional tRNA genes and improvement of suppressor eiciency. J Mol Biol 1990; 213(4):705-17.

180

Cell-Free Protein Expression

69. Gerrits M. Funktion und Eizienz von amber-Suppressor-tRNAs in der zellfreien Proteinbiosynthese, PhD hesis. Berlin: Freie Universität 2001. 70. Cload ST, Liu DR, Froland WA et al. Development of improved tRNAs for in vitro biosynthesis of proteins containing unnatural amino acids. Chem Biol 1996; 3(12):1033-8. 71. Szkaradkiewicz K, Nanninga M, Nesper-Brock M et al. RNA aptamers directed against release factor 1 from hermus thermophilus. FEBS Lett 2002; 514(1):90-5. 72. Agafonov DE, Huang Y, Grote M et al. Eicient suppression of the amber codon in E. coli in vitro translation system. FEBS Lett 2005; 579(10):2156-60. 73. Short GF, 3rd, Golovine SY, Hecht SM. Efects of release factor 1 on in vitro protein translation and the elaboration of proteins containing unnatural amino acids. Biochemistry 1999; 38(27):8808-19. 74. Shimizu Y, Inoue A, Tomari Y et al. Cell-free translation reconstituted with puriied components. Nat Biotechnol 2001; 19(8):751-5. 75. Ryden M, Murphy J, Martin R et al. Mapping and complementation studies of the gene for release factor 1. J Bacteriol 1986; 168(3):1066-9. 76. Kawashima T, Berthet-Colominas C, Wulf M et al. he structure of the Escherichia coli EF-Tu.EF-Ts complex at 2.5 A resolution. Nature 1996; 379(6565):511-8. 77. Merk H, Meschkat D, Stiege W. Expression-PCR: from gene pools to puriied proteins within 1 day. In: Swartz JR, editor. Cell-free protein expression. Berlin, Heidelberg, New York: Springer Verlag 2003; p. 15-23.

Chapter 15

C-Terminal Labeling of Proteins Using Fluorescently Conjugated Puromycin Derivatives Ichiro Tabuchi*

Abstract

A

novel C-terminal labeling technology has been developed using luorescently conjugated puromycin derivatives in cell free extracts. It is an easy to handle, rapid and low cost method for protein research. he labeling process requires only the addition of a labeling reagent to the translation mixture. Puromycin is conjugated with a luorophore that is incorporated into a protein at the C-terminus, thus linking the luorophore to the protein. Recent advances in cell-free protein expression combined with the puromycin analogue C-terminal labeling technology resulted in novel applications, such as single molecule imaging, Kd determination by luorescence cross-correlation spectroscopy (FCCS) among others. C-terminal labeling can be used for the introduction of a luorophore and many other types of molecules such as ainity tags. his technology can be used for library screening, proteomics analysis and biosensors.

Introduction

One of the most crucial steps in luorescence assay systems is obtaining the protein sample labeled with luorescent dye. Novel technologies using the properties of puromycin have been developed.1 he use of luorescent puromycin conjugates (“Fluor-Puro”) (Fig. 1) allows C-terminal speciic luorescent labeling of a protein. Also, protein production by in vitro translation systems has become important because of its rapidity and ease of handling. Recently several new in vitro translation methods have been developed yielding large amounts of protein (in the order of several mg).2,3 Novel technologies have emerged by combining these in vitro translation techniques with the puromycin analogue technique. A display technology for screening (Fig. 2)4,5 has already achieved great results.6,7 Other useful applications include luorescence labeling,8 ainity puriication1,19 and protein chips for proteomics.9 Here I report the present status and future prospects of puromycin analogue technology especially C-terminal luorescence (Fluor-Puro) labeling is reported.

he Principle of C-Terminal Fluorescence Labeling Using Puromycin Derivatives

Puromycin is a well-known antibiotic that inhibits protein synthesis by competitive incorporation for the ribosome A-site with an aminoacyl tRNA10 so that truncated proteins are produced. However, at low concentrations puromycin preferentially binds only to the C-terminus of a *Ichiro Tabuchi—Tokyo Evolution Research Center, 1-1-45-504, Okubo, Shinjuku-ku, Tokyo 169-0072, Japan. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

182

Cell-Free Protein Expression

Figure 1. Structure of Fluor-Puro. The structure of Cy5-dC-puromycin.

full-length protein.8 Puromycin can be modiied with a luorophore (“Fluor-Puro”) (Fig. 1) that is incorporated into a protein at the C-terminal, linking the luorophore to the protein. hereby, translation of an mRNA with a puromycin-luorescein conjugate (Fluor-Puro) renders a luorescein-labeled protein. he labeling eiciency usually ranges between 50% and 95%.8

Advantages of C-Terminal Fluor-Puro Labeling Technology

Where would be the best labeling position for most of the downstream applications? he most common method is random position labeling such as at Lysine or Cystein residues. his method is simple but it has some limitations. For example, the labeling density depends on the number of those residues. Also, if the target residue is part of the catalytic site, then the enzymatic activity may be lost. N-terminal labeling has been also reported.11 However, as the labeling reagent prior elongation may inhibit protein synthesis or interfere with protein folding. herefore the C-terminus represents the best choice for luorescent labeling. he major advantages of the Fluor-Puro technology are: 1. he method is fast and easy. he labeling procedure involves adding only the Fluor-Puro reagent during translation. he protein is cotranslationally labeled. he method is suitable for quantitative analysis because there is just one luorophore per protein molecule. 2. he method has fewer spatial labeling restrictions and fewer potential structural changes compared with GFP fusions. 3. he label is extremely stable; it luoresces even under strong denaturing conditions, such as those used in SDS-PAGE. he major disadvantage of this method is that a molar excess of the Fluor-Puro must be added to the cell-free reaction. Under these conditions some luorophores bind to the protein nonspeciically. Unincorporated Fluor-Puro must be then removed for those downstream applications where the residual free luorophore-puromycin conjugate may disturb the analysis. To overcome this shortcoming a novel method based on the formation of an RNA-protein fusion12 has been developed. Briely, the puromycin is linked to the mRNA via a DNA oligomer increasing the

C-Terminal Labeling of Proteins Using Fluorescently Conjugated Puromycin Derivatives

183

Figure 2. cDNA display (in vitro virus).

local concentration of the reagent. his approach uses one reagent molecule per protein substrate, eliminating the need for molar excess of the luorophore. Once the reaction is completed the linker DNA is then digested by T7 exonuclease. hus, the residual luorophore efect is minimized, the luorescently-labeled protein is disturbed less by the residual luorophore.1

Successful Fluor-Puro Labeling Applications

he irst reported application of Fluor-Puro labeling was the determination of a dissociation constant using luorescence polarization measurement.8 C-terminal luorescence labeling is suitable for the analysis of protein-protein interactions. In addition, by combining this technology with the single-molecule imaging technique, the physicochemical properties of a single molecule can be readily determined. For example, the movement of kinesin along a microtubule has been observed at the single-molecule level.13 In this case the luorophore Cy5-dC-puromycin was employed (Fig. 1). Remarkably, the entire labeling process took only two hours.13 he usefulness of this luorescence-labeling method for analyzing protein-protein or protein-DNA interactions with a DNA microarray or with luorescence cross-correlation spectroscopy (FCCS) in model systems has been reported.14 Although only model system demonstrations have been presented, they showed that C-terminal luorescence labeling is very useful for many applications. FCCS has good compatibility with the determination of interaction parameters with the added advantage that only small samples are needed. Finally, the technology can be used to study premature stop in protein translation.5

Fluor-Puro Conjugates: Labeling Efficiency and Choice of Fluorophore

he antibiotic puromycin mimics the 3’ terminus of an amino-acylated tRNA and causes nonspeciic, premature termination of translation. It interacts with the ribosome through the terminal nucleotides rCrC.15 herefore a variety of tail linkers, like dC, dCdC, rC, rCrC and other unnatural bases have been tested with the Fluor-Puro technology to study a potential increase in

184

Cell-Free Protein Expression

the labeling eiciency.1,19 he results indicated that either dC or dCdC exhibited good labeling eiciency at a relatively low cost. In some speciic cases, like in a DNA-protein interaction study, 2’-O-methyl-C also makes a good “tail,” because it is resistant to DNase activity.1,19 Other groups obtained similar results but without having a luorophore attached to the molecule14 suggesting that dC is usually the best nucleotide replacement. he fact that in our hands, sometimes dCdC exhibited better labeling eiciency suggests that the type of luorophore and the spacer afect the accessibility to the ribosomal A-site.1 Various luorophores are compatible with C-terminal Fluor-Puro labeling technology, including FITC, Cy2, Cy3, Cy5, TAMRA, BODYPY-FL and Alexa Fluor dyes. (Table 1). Labeling eiciency depends on the kind of luorophore attached. Generally bulky luorophores have low labeling eiciencies, however, adding a long lexible spacer reduces this tendency and increases the labeling eiciency. here are some points to consider when choosing a luorophore to use with Fluor-Puro. For example, longer wavelength luorophores are relatively free of biological noise but they tend to be unstable and in some cases, the synthetic eiciency is lower. Also, the stability of the luorophore is important as intense laser light can result in cleavage. For example an unstable luorophore (e.g., Cy5) loses luorescence within a minute under strong laser beam. he synthetic route is also very important when choosing a luorophore. Long wave luorophores are cleaved by strong bases or acids. Water solubility is also important. Solubility is related to nonspeciic adsorption that can lead to noise. Table 1. Fluorophore for Fluor-Puro

Fluorophore

Absorbance (nm)

Emission (nm)

Stability7

Commercially Available Phosphoramidite

Alexa Fluor 3501 Cy22 BODYPY-FL3 Fluorescein4 Alexa Fluor 4301 Yakima Yellow5 Alexa Fluor 5321 Cy32 TAMRA6 Redmond Red5 Cy52 Cy5.52 Cy72 Alexa Fluor 7501

346 489 505 495 434 531 531 546 565 579 646 683 743 749

442 506 513 525 539 549 554 563 580 595 662 707 767 775

++++ ++ ++++ +++ +++ +++ +++ ++ + ++ + + + ++

No No No Yes No Yes No Yes Yes Yes Yes Yes No No

1

The Alexa Fluor series is composed by a new generation set of dyes. They work under a wide wavelength range and are relative stable even under long wavelengths. 2 The Cy- series dyes are well-known fluorophores that work under a wide wavelength range but are not very stable. 3 The BODYPY series also covers a wide range of wavelengths. The fluorophore is relatively stable even under long wavelengths. 4 Fluorescein (or FITC) is a standard, stable and low cost fluorescent reagent. 5 This dye has been recently developed. It is relatively bright and stable. 6 TAMRA is a derivative of rhodamine. 7 Relative stability is indicated by “+” symbols.

C-Terminal Labeling of Proteins Using Fluorescently Conjugated Puromycin Derivatives

185

he Choice of he Translation System

he Fluor-Puro labeling technology has been successfully applied using the three major commercially available cell-free translation systems derived from Escherichia coli, wheat germ (WGE) and rabbit reticulocytes (RRL) (Table 2). Despite the fact that puromycin apparently inhibits transcription, coupled transcription-translation systems can be also used.13 he WGE system is relatively easy to handle and is recommended for the inexperienced user. he RRL system presents some advantages over the other platforms such as low RNase levels and post-translational modiications. However this lysate carries large amounts of globin, which may be problematic in SDS-PAGE and may be a source of luorescence noise. he E.coli S-30 system is a powerful, easy to handle and eiciently incorporates puromycin. However, the standard S-30 system contains undesirable amounts of RNase. he generation of 3’ digested mRNA molecules usually yield truncated proteins, which are also labeled. herefore this systems requires the addition of strong RNase inhibitors such as vanadyl ribonucleoside complexes. As the Fluor-Puro labeling is a cotranslational process, ribosome recycling is required. In theory the system proposed by Spirin16 should be advantageous for Fluor-Puro labeling. In the practice, the use of these kinds of systems did not result in a increased protein yield drastically. It may be caused a limitation on ribosome amount.19 Finally, advanced translation systems2,3,17 may also be used with Fluor-Puro labeling. Feature of these are listed in Table 2.

Concentration of Fluor-Puro

It is not easy to determine the optimum Fluor-Puro concentration that should be added to the translation system. Although high Fluor-Puro concentrations maximize the yield of the labeled products, the level of truncated protein products is also higher. he optimum concentration level is closely related to the length and secondary structure of the mRNA template. As the mRNA length increases, so does the amounts of truncated product that result from premature termination. hereby, Fluor-Puro concentration must be lowered, which usually results in a slightly lower labeling eiciency. he structure of the Fluor-Puro itself also afects the labeling eiciency. herefore, in some cases the use of higher Fluor-Puro concentration may be desirable. he choice of translation system also afects the optimal concentration.

Advanced Applications for Fluor-Puro Labeling FRET System

GFP has been chemically modiied for the purpose of C-terminal luorescence labeling.18 he internal Cysteins were replaced and a C-terminal Cystein was added at the C-terminus for chemical modiication and luorescent labeling. his GFP variant labeled with corresponding luorophores exhibits altered luorescence characteristics due to luorescence resonance energy transfer (FRET) and it can be used as a biosensor for a variety of compounds.18 he use of this modiied GFP as a biosensor in combination with Fluor-Puro labeling does not require further protein engineering or chemical modiications. he protein should be expressed in vitro in the presence of the Fluor-Puro reagent, a process that takes only one hour.19

Increase the Label of Matured Proteins

One way to increase the label of those products where the puromycin label has been incorporated in matured protein in the case gene has stop codon is by the inactivation or removal of the release factors. Proteins expressed under these conditions are released from the translational machinery. hereby, puromycin derivatives does not encounter competition for precise termination. herefore the concentration of the puromycin derivatives can be signiicantly reduced even if the gene has stop codon minimizing unwanted premature termination events.

+ ++ +++

+++ ++++

++ ++++ ++++

RRL E.coli S-30

E.coli reconstitute Advanced WGE Advanced E-coli S30 +++

++

+++

++ +++

++

Puromycin Incorporation2

Many different commercial sources. 2 Relative amounts or performance is indicated by “+” symbols.

1

++ ++++

++

WGE

+++

Productivity2

System

RNase Presence2

High yield

Less RNase, flexible system Easy handling, high yield

Easy handling, low cost Protein glycosylation Easy handling, low cost

Advantages

Table 2. Translation systems compatible with the Fluor-Puro technology

No data available

Recommended as a first choice High globin levels High RNase levels, Recommend for prokaryotic genes Suitable but complicated to operate Recommended

Comments

2

17 3

1

1

1

Source or Reference

186 Cell-Free Protein Expression

187

C-Terminal Labeling of Proteins Using Fluorescently Conjugated Puromycin Derivatives

For example, the release factor 1 (RF1), indispensable for termination at amber codons, can be inactivated by the incorporation of anti-RF1 antibodies. Under these conditions proteins whose genes end in amber codons, could be released from the ribosome only in the presence of a puromycin-derivative reagent. An example of this application using a puromycin analogue has been presented.20 In a more elaborated example, the puromycin analogue technology has been successfully applied using a reconstituted E. coli translation cell-free system (Table 2).17 One of the advantages of using this approach is that the user can decide which factors to include in the cell-free reaction and which ones to leave out. he elimination of the release factors from the in vitro reaction resulted in virtually no labeled truncated products.19 Following a similar strategy, a C-terminal Fluor-Biotin-Puro label could be eiciently incorporated into GFP.19

Selection of Multi Interaction Proteins

Accelerated evolution methods, such as mRNA display, have proven to be extremely reliable for the enrichment and selection of “improved” genes.4,5 Some performance has been reported21,22 but, unfortunately, this method is not suitable for selection of multi interaction proteins. Because at mRNA display, the selected mRNA coding only one protein. In those cases a cell-based strategy (Table 3) is preferred. In Vitro Compartmentalization (IVC)23,24 is a cell based selection system that uses in vitro translation. his system using micro cell made by oil contains translation mixture and translation done in the micro cell. he isolation of a functional peptide using IVC in combination with cell-sorting has been reported.24 A combination of IVC and Fluor-Puro technology (e.g., FRET based biosensor,18 see above) may emerge as a novel selection system for interacting proteins. Also, as IVC is not afected by the presence of stop codons, full length mRNA libraries can be used.

Proteomics Analysis

Full length mRNA libraries can be translated and labeled using Fluor-Puro reagents with diverse emission spectra. hereby, these complex samples can be easily analyzed by a single 2D-PAGE for multi-sample/multi-colored proteins in the same fashion as described for luorescence two-dimensional gel electrophoresis (2D-DIGE).25 his latter method, though somewhat popular, has a major shortcoming as it labels only a fraction of the proteins that are present in a given sample. Commercially available 2D-DIGE dyes come in two formats, the minimal dye and the maximum dye. he minimal dye labels only 1% of the proteins of a given sample, whereas the maximum dye fails to label proteins that lack cysteine residues. Fluor-Puro labeling, in contrast, has no limitations regarding sample complexity or protein composition. As the method results in one luorescent label per protein, accurate results are obtained. Also, since no harsh physicochemical methods, commonly applied for cell lysis, are required, protein unfolding is unlikely and even weak protein-protein interactions can be detected. he use of the reconstituted cell-free protein expression system for Fluor-Puro labeling17 results in extremely limited amounts of truncated products, as the system lacks proteases that are the most common cause of truncated products. Table 3. Genotype-phenotype linkage strategy for in vitro selection

Name

Strategy of Linking Genotype with Phenotype

Method

RNA type Virus type

Same molecule Chemical Bond

Ribozyme cDNA display

Advantage in Selection

Ribozyme single molecule protein Cell type Compartmentalization In vitro compartment multi interaction talization (IVC) proteins Assignment type human handling Biochip Proteomics

Reference 32 5 23 9

188

Cell-Free Protein Expression

Library Screening for Functional Protein

Recently, rationally designed artiicial libraries28,29 were developed. Multi Line Split DNA Synthesis (MLSDS)29 consists of a codon based library that uses a standard DNA synthesizer and DNA synthesis reagents. A split synthesis in codon units was performed with mixtures of bases optimally designed by using a genetic algorithm computer program. MLSDS has many advantages, including system lexibility, high speed, low cost and very high quality. In combinatorial chemistry, the higher the molecular diversity of a given library is, the higher the chances to obtain the target molecule. Fluor-Puro technology emerges as an important tool to generate diversity. Although mRNA display has demonstrated successful results,30,31 it has disadvantages on this respect. For example, it only uses substrate mRNAs that lack stop codons. hereby, full-length libraries are not substrates for this technology. Also, genes that encode proteins with functional domains near the C-terminus cannot be selected. In contrast, Fluor-Puro readily accepts full length libraries as substrates, as by the inactivation or elimination of the release factors from cell-free translation reactions (see above), puromycin labeling can generate nearly all full-length labeled products. Molecular diversity as high as 1014 can be achieved with this method. Fluor-Puro technology can be considered as an alternative strategy for selecting improved molecules as does not require the use of luorescent targets. he products are luorescent by themselves. Also in combination with microanalysis technologies such as LC-MS or SPR MS, very small amounts of labeled proteins are needed for the process.

Conclusion

C-terminal labeling of proteins using fluorescence conjugated puromycin derivatives (Fluor-Puro technology) is novel and very useful for many applications. Recently, advanced in vitro translation systems and luorescence analyzers have been developed. Combining these, fruitful results are promising.

Acknowledgement

I thank Mrs. Retta Hardy and Mrs. Miho Wakabayashi for helpful advice.

References

1. Tabuchi I. Next-generation protein-handling method: puromycin analogue technology. Biochem Biophys Res Commun 2003; 305(1):1-5. 2. Kigawa T, Yabuki T, Yoshida Y et al. Cell-free production and stable-isotope labeling of milligram quantities of proteins. FEBS Lett 1999; 442:15-19. 3. Madin K, Sawasaki T, Ogasawara T et al. A highly eicient and robust cell-free protein synthesis system prepared from wheat embryos: plants apparently contain a suicide system directed at ribosomes. Proc Natl Acad Sci USA 2000; 97:559-564. 4. Roberts RW, Szostak JW. RNA-peptide fusions for the in vitro selection of peptides and proteins. Proc Natl Acad Sci USA 1997; 94:12230-12297. 5. Tabuchi I, Soramoto S, Nemoto N et al. An in vitro DNA virus for in vitro protein evolution. FEBS Lett 2001; 508:309-312. 6. Keefe AD, Szostak JW. Functional proteins from a random sequence library. Nature 2001; 410:715-718. 7. Barrick JE, Takahashi TT, Ren J et al. Large libraries reveal diverse solutions to an RNA recognition problem. Proc Natl Acad Sci USA 2001; 98:12374-12378. 8. Nemoto N, Miyamoto-Sato E, Yanagawa H. Fluorescence labeling of the C-terminus of proteins with a puromycin analogue in cell-free translation systems. FEBS Lett 1999; 462:43-46. 9. Weng S, Gu K, Hammond PW et al. Generating addressableprotein microarrays with PROfusion covalent mRNA-protein fusion technology. Proteomics 2002; 2:48-57. 10. Monro RE, Marcker KA. Ribosome-catalysed reaction of puromycin with a formylmethionine-containing oligonucleotide. J Mol Biol 1967; 25:347-350. 11. Mamaev S, Olejnik J, Olejnik EK et al. Cell-free N-terminal protein labeling using initiator suppressor tRNA. Anal Biochem 2004; 326:25-32.

C-Terminal Labeling of Proteins Using Fluorescently Conjugated Puromycin Derivatives

189

12. Tabuchi I, Soramoto S, Suzuki M et al. An eicient ligation method in the making of an in vitro virus for in vitro protein evolution. Biol Proc Online 2002; 4:49-54. 13. Yamaguchi J, Nemoto N, Sasaki T et al. Rapid functional analysis of protein-protein interactions byluorescent C-terminal labeling and single-molecule imaging. FEBS Lett 2001; 502:79-83. 14. Doi N, Takashima H, Kinjo M et al. Novel luorescence labeling and high-throughput assay technologies for in vitro analysis of protein interactions. Genome Res 2002; 12:487-492. 15. Ban N, Nissen P, Hansen J et al. he Complete Atomic Structure of the Large Ribosomal Subunit at 2.4 Å Resolution. Science 2000; 289:905-919. 16. Spirin AS, Baranov VI, Ryabova LA et al. A continuous cell-free translation system capable of producing polypeptides in high yield. Science 1988; 242:1162-1164. 17. Shimizu Y, Inoue A, Tomari Y et al. Cell-free translation reconstituted with puriied components. Nat Biotechnol 2001; 751-5 18. Suzuki M, Ito Y, Savage HE et al. Protease-sensitive signalling by chemically engineered intramolecular luorescent resonance energy transfer mutants of green luorescent protein. Biochim Biophys Acta 2004; 1679:222-229. 19. Tabuchi I. in preparation to press. 20. Agafonov DE, Rabe KS, Grote M et al. C-terminal modiications of a protein by UAG-encoded incorporation of puromycin during in vitro protein synthesis in the absence of release factor 1. Chembiochem 2006; 7(2):330-6. 21. Miyamoto-Sato E, Ishizaka M, Horisawa K et al. Cell-free cotranslation and selection using in vitro virus for high-throughput analysis of protein-protein interactions and complexes. Genome Res 2005; 15:710-717. 22. Tateyama S, Horisawa K, Takashima H et al. Ainity selection of DNA-binding protein complexes using mRNA display. Nucleic Acids Res 2006; 34 e27. 23. Tawik DS, Griiths AD. Man-made cell-like compartments for molecular evolution. Nature Biotechnol 1998; 652-656. 24. Sepp A, Tawik DS, Griiths AD. Microbead display by in vitro compartmentalization: selection for binding using low cytometry. FEBS Lett 2002; 532:455-458. 25. Unlu M, Morgan ME, Minden JS et al. Diference gel electrophoresis: a single gel method for detecting changes in protein extracts. Electrophoresis 1997; 18:2071-2077. 26. Suzuki M, Nemoto N, Husimi Y. EMBO IRBM Work-shop on Molecular Repertories and Methods of Selection, Gabbio, Italy 1993; p97. 27. Husimi Y, Aita T, Tabuchi I. Correlated lexible molecularcoding and molecular evolvability. J Biol Phys 2002; 28:499-507. 28 Cho G, Keefe AD, Liu R et al. Constructing high complexity synthetic libraries of long ORFs using in vitro selection. J Mol Biol 2000; 297:309-319. 29 Tabuchi I, Soramoto S, Ueno S et al. Multi-line split DNA synthesis: a novel combinatorial method to make high quality peptide libraries. BMC Biotechnol 2004; 4:19. 30 Hammond PW, Alpin J, Rise CE et al. Invitro selection and characterization of Bcl-X(L)-binding proteins from a mix of tissue-speciic mRNA display libraries. J Biol Chem 2001; 276:20898-20906. 31. McPherson M, Yang Y, Hammond PW et al. Drug receptor identiication from multiple tissues using cellular-derived mRNA display libraries. Chem Biol 2002; 9:691-698. 32. Ellington A, Szostak JW. In vitro selection of RNA molecules that bind speciic ligands. Nature 1990; 346:818-822.

Chapter 16

Translation Engineering and Synthetic Biology David A. Roth,* Liza S.Z. Larsen and G. Wesley Hatield

Abstract

W

e have developed a toolbox approach for protein engineering to enhance protein expression yield and functionality by manipulating ribosomal “pause” signals correlated with over-represented codon pairs contained in the open reading frame of genes. CODA’s Translation Engineering algorithms simultaneously optimize each gene design for both organism-speciic codon usage and for organism-speciic codon pair usage. To further implement Translation Engineering, CODA has developed propriety gene sotware, SpeedPlotTM, a design tool that predicts and graphically displays the positions of translational pause sites in genes expressed in native and heterologous biological hosts. For high protein expression levels, SpeedPlot information is used to design Hot-rodTM genes with all predicted pause signals removed for maximal translation elongation rates. For improved protein function, Speedplot information is used to rationally design gene sets using Hot-rod genes as parental templates, from which directed variant genes are constructed to contain speciied combinations of pause signals (Planned Pause Gene SetsTM). Each deined gene set can be tested in parallel for optimal expressors that encode proteins with premium activities. hese Translation Engineering methods are enabled by CODA’s proprietary Computationally Optimized DNA Assembly (CODA) technology to assemble full-length genes by thermodynamic necessity. In combination, these technologies make it possible to deliberately alter the predicted translation kinetics patterns of genes without altering their amino acid sequences. In this chapter, we discuss the application of these tools for the design and assembly of synthetic genes optimized for expression and function in any biological host or cell-free protein synthesis system.

Translation Engineering™—Codon Context and Translational “Pausing”

A variety of techniques exist today to address the numerous issues faced in protein expression. Alternate expression vectors, cell lines, chaperones and expression conditions are just a few of the many “tricks-of-the-trade” that protein scientists utilize to coax the maximum yield of active proteins from host cells and “cell-free” expression systems. Codon usage optimization has been a common tool available to scientists to utilize “silent” mutations for improved protein expression. However, the results of codon usage optimization are inconsistent. his is because simply correcting for codon usage scrambles species-speciic codon context (codon pair bias) efects that control protein translation step-times important for mRNA stability, protein folding and other cotranslational events.1-8 hese inappropriately positioned codon pairs can compromise protein *Corresponding Author: David A. Roth—Senior Vice President, Research and Development, CODA Genomics, Inc., 26061 Merit Circle, Laguna Hills, CA 92653-7015, U.S.A. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

Translation Engineering and Synthetic Biology

191

expression of codon optimized genes and of native genes in heterologous biological hosts or cell-free systems. To address this problem, scientists at CODA Genomics have developed an assortment of molecular and computational tools, collectively known as Translation EngineeringTM. hese tools enable researchers to express optimally-designed synthetic genes for manufacturing large amounts of functional protein. he underlying science of Translation Engineering is providing new insights into protein structure/function relationships and generating hypotheses for rational processes by which proteins can be produced in large quantities with native or even enhanced activities. Translation Engineering is being used for scalable, rational production of proteins for developing protein and antibody-based therapeutics, drug screens, industrial enzymes and other commercial protein products. Living organisms from bacteria to yeast and human all exhibit extremes in their distribution and utilization of distinct codon pairs1-4 and some highly over-represented codon pairs are recognized by ribosomes as translational pause signals.5 Computational scientists at CODA have developed SpeedPlotTM sotware to identify and graphically display the codon pair positions of these predicted translational pause sites along a gene sequence. hey also have developed Pause Conservation MapsTM to compare the positions of these pause signals in the genes of structurally-related proteins of diferent organisms across large evolutionary distances. hese comparisons have shown that, in spite of the facts that codon usage and codon pair usage patterns are unrelated and that there exists little nucleotide sequence identity among evolutionarily diverse organisms, the positions of many predicted translational pause sites are conserved. hese comparisons of structurally related protein families also have revealed that these pause sites are oten correlated with structure domain boundaries. CODA SpeedPlot and Pause Conservation Maps sotware packages are used to design gene sequences for expressing, testing and producing commercially important proteins.

Codon Pairs Control Translation Kinetics

It has been suggested that ribosome pause signals serve as an explicit, universal, evolutionarily conserved signaling mechanism to ensure correctly folded, structurally intact and functional proteins.3 he actual codon pairs used to encode ribosome pauses vary widely from organism-to-organism. hus, moving a gene’s open reading frame (ORF) from a source organism into a heterologous biological host can scramble these signals and result in random pause signals and failed protein expression and function. his can explain why many attempts to improve protein expression and yield with synthetic genes fail. For example, attempts to express a native gene, or even a synthetic codon usage-optimized gene from Pichia pastoris in the heterologous expression host, Saccharomyces cerevisiae, scramble these signals. his results in random redistribution of pause signals that can cause unpredictable levels of gene expression and impaired protein function such as lack of solubility, improper cotranslational folding, aggregation, or other problems.7,8,10 Karlin et al12 carefully considered the earlier work of Irwin et al5 on how codon pair utilization biases inluence translational elongation step times and in a comprehensive study and independent discussion these scientists showed the relative importance of codon usage and di-codon (codon pair) bias for the control of protein expression levels and for protein function. More recently, Buchan et al11 conirmed the early observations of Gutman and Hatield1,3 and of Irwin et al5 and their interpretations about codon pair bias and tRNA incompatibilities in the A and P sites of a translating ribosome. However, because like Gutman and Hatield they also ind little correlation between codon pair bias and codon usage, they preferred to suggest that the di-nucleotide bias observed between adjacent codons is responsible for “translational eiciency.” hus, optimization for codon usage alone is not a reliable method for high level protein expression. Translation Engineering takes the gene design process beyond simple codon usage to control translation kinetics and further ensure high level protein expression, as well as proper protein structure and function.

Why Is Translational Pausing Important?

Pauses appear to have been evolutionarily conserved for a variety of reasons.3,6-8 While biological signiicance of many predicted translational pauses is not completely understood, we do know

192

Cell-Free Protein Expression

that many of these pauses are evolutionarily conserved in structurally-related protein families. We also know that the introduction or removal of predicted translational pauses can have signiicant efects on a protein’s production and/or activity while the amino acid sequence remains unchanged.3,5-8,10 he common occurrence of evolutionarily conserved pauses near protein domain boundaries suggests that many pauses are critical for the successful folding, secretion and activity of a given protein. Scientists at CODA and elsewhere have proposed that silent mutations that afect translational pausing afect cotranslational events associated with human genetic diseases. For example, Kimchi-Sarfaty et al recently reported that translational pausing encoded by synonymous mutations is important for cotranslational events afecting protein folding and function of the human multi-drug resistance gene, MDR1.8 hey demonstrated that single nucleotide polymorphisms (SNPs) that do not change the amino acid sequence of this protein do change the structure and function of the resultant MDR1 protein. hey suggested that these efects of a silent mutation in the human MDR1 gene were caused by the generation of a slowly translated rare codon that afects the timing of cotranslational folding and insertion of the protein into the membrane. We suggest that, in addition to translational slowing by rare codon usage, translational slowing mechanisms encoded by codon context efects also might, at least in part, explain the efects of synonymous SNPs on the structural and functional properties of the MDR1 gene product. It is now clear that the efects of species-speciic codon usage and codon-pair biases on translation kinetics afect protein structures and functions. his knowledge already is being used to design synthetic genes with cotranslational properties that enable high protein expression and provide strategies for the design of gene sequences with preserved or enhanced function in any natural or heterologous host organism or cell-free expression system for which the codon pair utilization statistics have been calculated. Below we describe a variety of computational and biological tools that implement these principles to solve diicult protein expression, structure and function problems.

he CODA Translational Engineering Toolbox Overview

Scientists at CODA Genomics take a rational approach to the analysis and design of gene sequences for optimized protein expression, structure and function. Depending on the nature of the project, multiple steps may be used to achieve the desired protein characteristics in an optimized cell line or cell-free expression system of interest. We begin our analyses with an evaluation of the translational pausing pattern of the native gene in the native organism and the native gene in the new host organism. his information is used to design the best gene, or rationally designed sets of genes, to address the goals of each synthetic gene and Translational Engineering project. he irst step of this Translation Engineering strategy involves the design and synthesis of a Hot-rodTM gene. his is a codon optimized gene in which all pausing signals are removed. his produces a “domesticated” gene sequence with predictably high protein expression levels that serves as the “parental” construct and baseline for subsequent functional optimizations. Comparisons of the translational kinetics of genes for structurally related protein families provide further improvements which may involve introducing pause sites, or combinations of pause sites to the Hot-rod gene to improve protein product properties such as solubility, secretion, structure, or function. hese general methods are used for the design of deined permutated gene sets optimized for high protein expression with rationally altered translational kinetics for the selection of gene products with improved target properties.

Mechanisms and Utilities

A widely held hypothesis for the role of codon context in the mechanism of slowing translation involves compatibilities of adjacent tRNA isoacceptor molecules on the surface of a translating ribosome deined by a speciic codon pair.9 In 1989, Gutman and Hatield1 proposed that the frequency of one codon next to another co-evolved with the structure and abundance of tRNA isoacceptors in order to control translational step-times without imposing additional constraints

Translation Engineering and Synthetic Biology

193

on amino acid sequences or protein structures, Later, Irwin et al5 presented in vivo evidence that a silent mutation can change a rapidly translated codon pair into a slowly translated codon pair and vice versa. Since this irst report that silent mutations can alter translational kinetics, others have provided experimental evidence that silent mutations that afect translation kinetics also afect in vivo protein folding and function. For example, Cortazzo et al7 reported the efects of codon substitutions in the EgFABP1 (Echinococcus granulosus fatty acid binding protein-1) gene that replaced ive codons with their synonymous ones. he altered region corresponded to a turn between two short alpha helices. One of the silent mutations markedly decreased the solubility of the protein when expressed in E. coli. Expression of this protein also caused strong activation of a reporter gene designed to detect mis-folded proteins, suggesting that the turn region has special translation kinetic requirements that ensure proper folding of the protein. Cortazzo et al interpreted their results in terms of the importance of codon context (codon pairs) for in vivo protein folding. In another example, Trinh et al10 showed that a single silent mutation predicted by the work of Hatield and Gutman3 in a recombinant human fusion antibody gene that replaced a single over-represented (slowly translated) codon pair with an under-represented (rapidly translated) codon pair and did not alter either the amino acid sequence or codon usage, resulted in a 30-fold increase in active protein expression in mammalian cells.

Tool 1: SpeedPlotTM

Computational scientists at CODA Genomics have developed a tool named SpeedPlot to display relative translational step-times across each codon pair of the protein coding sequence of a gene. Each SpeedPlot displays a Z-score based on the chi-square statistics (i.e., the relative deviation from random utilization of each successive codon pair), corrected for di-nucleotide and amino acid pair bias,1 for each of the 3,721 (61 by 61) nonterminating codon pairs that can occur in a gene sequence. High Z score values associated with over-represented codon pairs have been shown to be correlated with slowly translated codon pairs (translational pauses; reference 5). Randomly and under-represented codon pairs are translated more rapidly. A SpeedPlot for the gene encoding the capsid protein of the Saccharomyces cerevisiae Ty3 retrovirus-like element is shown in Figure 1.

Figure 1. SpeedPlot. A graphical display of Saccharomyces cerevisiae codon pair usage values for the GAG gene of the S. cerevisiae Ty3 retrovirus-like element. Slower translational step-times are correlated with codon pairs exhibiting high Z- score values. Faster translational step-times are correlated with low Z-score values (a negative sign is arbitrarily added to the Z-score values of under-represented codon pairs; refs. 1, 3). The predicted translational pause sites at codon pair positions 110-111 and 120-121 coincide with Gag protein domain boundaries identified in the inset with arrows.

194

Cell-Free Protein Expression

his SpeedPlot suggests that the irst helix-bundle domain of the Ty3 capsid protein is translated relatively rapidly until it encounters a pause signal represented by an over-represented codon pair at positions 110-111. his initial pause is followed by a short relatively unstructured inter-domain sequence and then a second pause. It is reasonable to assume that these pause signals deine a domain boundary region that allows the N-terminal and C-terminal domains of this protein to sequentially fold without interference from one another.

Tool 2: Hot-rodTM Genes

A common problem that occurs during codon optimization in heterologous expression systems is the creation of “accidental,” or nonplanned random pauses that occur because of diferent codon pair usage statistics in diferent organisms. he SpeedPlots shown in Figure 2 compare codon pair biases observed for the expression of the native yeast Ty3 GAG gene sequence expressed in its host organism, S. cerevisiae (Fig. 2A) and in a heterologous host, E. coli (Fig. 2B). hese SpeedPlots demonstrate a random redistribution of pauses for expression of the native yeast gene in E. coli. he occurrence of “accidental” pause sites, especially those that occur early in the gene (Fig. 2B), correctly predict low expression of this gene in E. coli due to ribosome queuing and steric hindrance of translation initiation.5 Moreover, we have observed that randomly positioned over-represented codon pairs that occur later in the gene may compromise proper protein folding, solubility or activity. he SpeedPlot of the CODA optimized Hot-rod gene, designed to eliminate all over-represented codon pairs to ensure maximal translational rates, is shown in Figure 2C. he data in Figure 3 conirm the SpeedPlot prediction that, unlike the native yeast gene, the CODA-optimized Hot-rod gene expresses protein at high levels in E. coli. For many applications discussed in other chapters of this book, it oten is desirable to express proteins in cell-free systems. However, like protein synthesis in live cells, protein synthesis yields in cell-free systems also are unpredictable. We addressed this obstacle by comparing cell-free protein yield from the yeast Ty3 GAG gene product and an E. coli CODA-optimized Ty3 GAG Hot-rod gene in an E. coli coupled S30 transcription/translation system for linear DNA templates obtained from Promega (Madison, WI). Like the in vivo system, no protein was obtained when the native yeast Ty3 GAG gene was used as a DNA template. However, high level expression was observed in this E. coli cell-free system with a DNA template encoding a CODA-optimized Ty3 GAG gene (Fig. 3). It is assumed that high level expression is observed with the Hot-rod gene for the same reasons it is highly expressed in live cells; that is, the removal of translation pause signals that compromise translational initiation and elongation.

Tool 3: Pause Conservation MapsTM

Many predicted translation pauses appear to be tolerated while others are problematic. In our eforts to distinguish these pauses, we have discovered that many are conserved in structurally related protein families from evolutionarily disperse organisms. We identify these conserved pauses with a newly developed CODA sotware tool called Pause Conservation MapsTM. Many conserved pauses are correlated with structural domain boundaries, secretion signals and other protein structural components. It is reasonable that these conserved pauses are important for protein properties associated with their structure and function. Our Pause Conservation Maps help to isolate these pauses from other, perhaps species-speciic, pauses that are less important for expression in heterologous biological hosts. As additional genomes are sequenced and available in the public domain or obtained through commercial licensing and partnerships, CODA continues to analyze, compare and characterize translational pauses. To date, CODA scientists have computed the codon pair utilization statistics for over 500 organisms. his exclusive ability to predict pause signals in these organisms and to manipulate them individually and in combination, facilitates solutions to complicated challenges to the enhancement of protein expression, structure and function in biological hosts and cell-free systems of choice. his provides an unique opportunity to selectively engineer planned pauses in genes of interest to produce proteins with desired physical and performance characteristics.

Translation Engineering and Synthetic Biology

195

Figure 2. SpeedPlotsTM for native and Hot-rodTM genes in S. cerevesiae and E. coli GAG genes. See legend of Figure 1 for SpeedPlotTM description. A) SpeedPlotTM for the native S. cerevesiae GAG gene expressed in S. cerevesiae. B) SpeedPlotTM for the native S. cerevesiae GAG gene expressed in E. coli. C) SpeedPlotTM for CODA-optimized GAG Hot-rodTM gene expressed in E. coli.

Tool 4: Planned Pause Gene SetsTM

CODA computational scientists also have developed powerful design tools to engineer combinations of conserved codon pair-determined translation pause signals that occur at selected sites in structurally related proteins without altering amino acid sequences. he result is a set of Planned

196

Cell-Free Protein Expression

Figure 3. Coupled in vitro transcription/ translation product of CODA-optimized GAG Hot-rod gene. Coomassie-stained SDS-PAGE gel of reaction products. Lane 1, Mol. Wt. markers. Lane 2, reaction products of CODA-optimized Hot-rod gene. See text for experimental details.

Pause GenesTM containing single pauses and combinations of pauses strategically engineered into the ORF as “silent” nucleotide sequences for a parental “problem protein” to enhance domain folding, secretion and other cotranslational events.

Summary and Applications

While the ability to identify proteins that represent exciting targets for therapeutic intervention has been scaled up in recent years, a current roadblock for biologics-based drug discovery is the ability to obtain signiicant quantities of reliably functioning proteins for economical protein production systems. Here we describe a new Translational Engineering technology for the design and assembly of synthetic genes for large scale bioreactor and fermentation manufacture of functional proteins of immediate importance to pharmaceutical, biotechnology and other protein-dependent industries. We are intrigued by and anticipate the large impact of successful cell-free expression systems that can take advantage of CODA-optimized genes in a variety of cell-free protein expression systems. CODA DNA assembly and Translational Engineering technologies are enabling the design and synthesis of genes for the production of therapeutic proteins, antibodies, protein subunit vaccines, diagnostic products, industrial enzymes and proteins for a variety of research purposes, through ever-improving, high throughput, low cost, bacterial, insect, plant, animal and cell-free expression systems. We anticipate that scientists both in discovery research, in R&D and in production facilities might ind the combination of CODA synthetic genes and cell-free expression systems to be rapid, inexpensive and possibly more amenable to high-throughput technologies and applications than existing organism-derived protein expression systems.

Translation Engineering and Synthetic Biology

References

197

1. Gutman GA, Hatield GW. Nonrandom utilization of codon pairs in Escherichia coli. Proc Natl Acad Sci USA 1989; 86:3699-3703. 2. Hatield GW, Gutman GA. Nonrandom utilization of codon pairs in Escherichia coli, US Patent #5,082,767, 1992. 3. Hatield GW, Gutman GA. Codon Pair Utilization Bias in Bacteria, Yeast and Mammals. In: Hatield DL, Lee BJ, Pirtle RM eds. Transfer RNA in Protein Synthesis, Boca Raton, FL, USA: CRC Press, 1993. 4. Hatield GW, Roth DA. Optimizing Scaleup Yield for Protein Production: Computationally Optimized DNA Assembly (CODA) and Translation EngineeringTM. In: Raafat El-Gewely M ed. Biotechnology Annual Review. Atlanta:Elsevier, B.V., 2007:In Press. 5. Irwin B, Heck JD, Hatield GW. Codon pair utilization biases inluence translational elongation step times. J Biol Chem 1995; 270(39):22801-6. 6. Kittle JD. Radical Changes in the Engineering of Synthetic Genes for Protein Expression. BioPharm International 2006; 12-18. 7. Cortazzo P, Cervenansky C, Marin M et al. Silent mutations afect in vivo protein folding in Escherichia coli. Biochem Biophys Res Commun 2002; 293(1):537-41. 8. Kimchi-Sarfaty C, Oh JM, Kim IW et al. A “silent” polymorphism in the MDR1 gene changes substrate speciicity. Science 2007; 315(5811):525-8. 9. Smith D, Yarus M. tRNA-tRNA interactions within cellular ribosomes. Proc Natl Acad Sci USA 1989; 86:4397. 10. Trinh R, Gurbaxani B, Morrison S. Optimization of codon pair use within the (GGGGS)3 linker sequence results in enhanced protein expression. Molecular Immunology 2004; 40:717-722. 11. Buchan JR, Aucott, LS, Stansield I. tRNA properties help shape codon pair preferences in open reading frames. Nucleic Acids Research 2006; 34:1015-1027. 12. Karlin S, Mrazek J, Campbell AM. Codon usage in diferent gene classes of genes in the Escherichia coli genome. Mol Micro 1998; 29:1341-1355.

Chapter 17

Accelerated Protein Evolution Using Ribosome Display Julie Douthwaite,* Lutz Jermutus, Ronald Jackson

Abstract

R

ibosome display is a powerful selection technology for the identiication of proteins with desired functions and has wide applicability across biologics drug discovery and basic research. Ribosome display is based on polymerase chain reaction and cell-free translation technologies that combine to provide an entirely in vitro process for selection and directed evolution that is rapid and highly efective. Using ribosome display, human monocolonal therapeutic antibodies are eiciently evolved for optimum ainity.

Introduction to in Vitro Display Technology

Display technologies allow genes encoding proteins with desired properties to be identiied from very large populations, and are most commonly exempliied by the identiication of monocolonal antibodies from very large antibody gene libraries.1 his capability has revolutionized the discovery of antibodies for therapeutic use, allowing multiple human antibodies speciic for disease targets to be discovered or optimized in a matter of weeks, compared to the several months oten required for the production of antibodies using mouse immunization and hybridoma technologies. Central to any display technology is the ability to link genotype and phenotype, i.e., to link a gene to its encoded protein. his enables a selection process where many proteins can be simultaneously examined for a desired property, such as binding to a molecule of interest. Selection is a highly eicient way to examine a large number of gene sequences and allows many more variants to be searched than would be feasible using a one-by-one screening approach. Numerous display technologies now exist, the principle of each being the same, but difering mechanistically in how gene-protein coupling is achieved. In phage display,2 the irst display technology to be developed, proteins are displayed on the surface of phage particles. In this method, individual phage contain a gene encoding a single protein variant or library member, such that the protein is produced as a fusion to a phage coat protein and is therefore displayed on the surface of phage. Phage display is a robust and simple technology that has been extremely well proven by the production of many therapeutic antibodies that are now in the clinic. In vitro selection techniques have also been developed,3 such as ribosome display that achieve display without the use of living organisms. his article will focus on the ribosome display process and present examples of its application over recent years.

*Corresponding Author: Julie Douthwaite—Cambridge Antibody Technology, Milstein Building, Granta Park, Cambridge, CB21 6GH, United Kingdom. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

Accelerated Protein Evolution Using Ribosome Display

he Ribosome Display Process

199

Ribosome display is an elegant approach to coupling gene and protein and exploits cell-free translation to produce a selection unit comprising a stalled ribosome noncovalently linking the nascent displayed protein to its encoding mRNA (Fig 1.). A polymerase chain reaction (PCR)-generated DNA library, encoding up to 1014 diferent proteins, is transcribed in vitro. he resulting mRNA library is then translated in a cell-free reaction under speciic conditions to produce an array of stabilized ternary ribosome complexes where both mRNA and protein remain bound to the ribosome. Stop codons that would otherwise signal release of mRNA and the newly translated protein are omitted from the DNA library. Such stalled ribosome complexes are further stabilized by rapid cooling of the cell-free translation reaction and dilution in a high magnesium containing bufer (ribosome movement is restricted by the interaction of magnesium cations with negatively charged phosphate groups of the ribosomal RNA). One research group has further codisplayed an RNA binding protein that binds to its own encoding mRNA to increase ribosome complex stability,4 although this is not routinely required. Ribosome complexes produced in the typical way as described have been shown to be stable for at least 2 weeks.5 A ‘tether’ sequence encoding a relatively nonstructured protein is included downstream within the DNA library to allow the displayed protein to fully exit the ribosome tunnel and fold free of steric hindrance. Stem loop sequences at the 5’ and 3’ ends of the mRNA increase its stability and timing of the in vitro translation reaction is optimised and precisely controlled to achieve the most productive balance between translation of protein, degradation of mRNA and spontaneous ribosome complex dissociation. Stable ribosome complexes can be selected by a binding interaction between the displayed protein and a target of interest, nonbound ribosome complexes that display proteins lacking the desired binding function are washed away. he DNA sequences of genes encoding the desired selected proteins are recovered by reverse transcription of mRNA and PCR to regenerate an enriched and ampliied DNA library for further selection or for sequence analysis and protein characterization. Several recent publications describe detailed protocols for ribosome display.6-8 Table 1 presents highlights from the literature illustrating the variety of applications of ribosome display for protein selection and evolution applications and other uses that exploit the genotype-phenotype linkage of ribosome display.

Figure 1. Schematic representation of the ribosome display process.

200

Cell-Free Protein Expression

Cell-Free Translation for Ribosome Display

he key step in ribosome display in which the gene-protein coupled selection unit is produced is a cell-free translation reaction that should be optimal for the production of the maximal number of stable ribosome complexes and tailored to achieve correct folding of the proteins being displayed. Cell extracts from a number of sources have been used, the E. coli S30 extract being the most common (see Table 1). Protein production in E. coli extracts can be limited, for example by lack of biochemical energy due to rapid ATP breakdown and poor mRNA stability, however these disadvantages have been overcome suiciently well in the ribosome display process to allow the successful selection and evolution of many proteins. Translation times are relatively short (typically less than 10 minutes), especially by comparison to those required for preparative cell-free translation reactions and protocols include many strategies to overcome RNAse activity, including use of RNAse deicient strains of E. coli, inclusion of secondary structure in the ribosome display construct, reaction processing at low temperature and the inclusion of RNAse inhibitors. Detailed information regarding the optimization of cell-free translation for ribosome display has previously been published.9 Cell extracts derived from eukaryotic sources, such as the rabbit reticulocyte lysate (RRL) and the wheat germ extract have also been used for ribosome display. A notable advantage of a eukaryotic translation system is the potential for posttranslational modiication of displayed proteins, for example phosphorylation or glycosylation, although this has yet to be demonstrated functionally. Use of ribosome display was initially limited to a small number of specialist laboratories due to the considerable knowledge and investment required for setting up the technique, especially for preparation of the E. coli S30 extract. Commercial sources of E. coli S30 extract are typically not suitable for translation of proteins requiring formation of disulphide bonds due to the presence of reducing agents in the extract. RRL translation reagents lacking reducing agents are now available commercially and a recent publication describing improved protocols for their use should render ribosome display more widely accessible to nonspecialist laboratories.6 As an alternative to the use of cell extracts for in vitro translation, a highly pure prokaryotic translation system has been reconstituted using puriied 70S ribosomes, recombinantly expressed and puriied proteins and other necessary factors.10 Such a system is an attractive development in cell-free translation for ribosome display due to the near absence of RNAses and proteases that are inherent in cell extracts. It was recently shown that selection outputs are signiicantly higher when a reconstituted translation system is employed instead of an S30 extract and this is most likely due to improved mRNA stability and protein integrity and potentially better energy provision and lack of unknown inhibitiors.11 hese features presumably beneit selections by an increase in the overall functional library size, that is the number of correctly folded, full-length proteins coupled via a ribosome to their full length, encoding mRNA. Reconstituted translation systems (named the PUREsystem) are also commercially available and so represent another signiicant step forward in increasing the accessibility of ribosome display to the wider research community.

In Vitro Evolution by Ribosome Display

he selection of antibodies from libraries is essentially a very efective enrichment or puriication exercise, where each library member is tested for a certain property and undesired members are removed. Selection may also be applied to many variants of an existing protein, e.g., a mutagenesis library comprising many mutants of a monocolonal antibody and in this case the process becomes comparable to Darwinian evolution. Mutations that improve function, e.g., binding ainity, are selected for in the presence of a selection pressure such as reduced concentration of target antigen, while mutations that lower binding ainity or render the protein nonfunctional in other ways are lost. PCR and cell-free translation underlie ribosome display and in combination these technologies allow the processes of library generation, display and selection to be performed without use of living organisms. his provides several beneits over display technologies that do require in vivo steps and makes ribosome display particularly well suited to protein evolution. Larger library

E. coli S30 extract

E. coli S30 extract

E. coli S30 extract

E. coli S30 extract

E. coli S30 extract

E. coli S30 extract

Human scFv population previously enriched by phage display Designed ankyrin repeat protein libraries

Mammalian nogo receptor random mutagenesis library Rodent-human hybrid scFvs

scFv random mutagenesis library

Bcl-xL

continued on next page

E. coli S30 extract

Human scFv random mutagenesis library

Evolution a human monoclonal antibody for therapeutic use and the mapping of amino acid residues of that antibody that are tolerant to change and associated with improved affinity. Demonstrates the broad applicability of ribosome display to inform a directed mutagenesis strategy.24 Evolution of antibody fragments specific for a model antigen, demonstrating the use of ribosome display in combination with phage display for a streamlined isolation-evolution approach.20 Selection and evolution of designed ankyrin repeat proteins for specific, high affinity target binding using maltose binding protein and mitogen-activated protein kinases (JNK2 and p38) as model targets.25 Display of a protein that is aggregation prone under heterologous expression in vivo. Lack of aggregation in ribosome display format and display of mutants enabled the identification of a functional epitope of the receptor.26,27 Guided selection of human antibody fragment from a rodent antibody fragment. Heavy and light chains are separately selected in combination with existing rodent light and heavy chains respectively.28 Evolution of a very high affinity antiprion protein antibody fragment (1 pM; the highest affinity ribosome display optimised antibody to date) by use of a stringent selection strategy to select for improved dissociation constants.29 Display of a model membrane protein for ligand identification. The method also incorporates a specific protein-mRNA interaction (codisplay of an mRNA-binding protein) favours production of polysomes and enhances ribosome complex stability.4

Translation System

Protein Displayed

Application

Table 1. Examples of the application of ribosome display to the selection and/or evolution of a diverse range of proteins using different cell-free translation systems

Accelerated Protein Evolution Using Ribosome Display 201

Rabbit reticulocyte lysate Rabbit reticulocyte lysate Rabbit reticulocyte lysate Wheat germ extract Wheat germ extract Reconstituted E. coli system Reconstituted E. coli system Rabbit reticulocyte lysate

Zinc finger targeted mutagenesis libraries

scFvs and growth hormone

Selection under a variety of scenarios demonstrating the improvements in ribosome display output by use of a purified translation system.11 Demonstrating a 12,000-fold enrichment of an antibody fragment in a single ribosome display selection cycle in a model system.36 Use of an error prone RNA-dependent RNA polymerase for mutagenesis and affinity maturation.13

Model library comprising known ratios of scFv to an irrelevant protein Single domain antibody fragment based on the shark IgNAR antibody

20-mer random peptide library

Selection of peptides with binding specificity for progesterone specific antigen.35

Selection of enzyme mutants based on binding affinity to a substrate analog.34

Human scFv library from a transgenic mouse Dihydrofolate reductase

T4 DNA Ligase

E. coli S30 extract

10-mer random peptide library

Selection based on the enzymatic activity of DNA ligase, demonstrating the principle of selection of DNA modifying enzymes.32 Selection of antiprogesterone antibody fragments from an antibody library.33

E. coli S30 extract

E. coli S30 extract

Model protein with certain folding properties scFv random mutagenesis library

Selection of proteins based on protease resistance and removal of hydrophobic proteins that are likely to be unfolded.22 Selection and directed evolution of antibody fragments using specific selection pressures to improve affinity or stability. Demonstrates ribosome complexes are stable in the selection environment for at least 10 days.5 Selection of peptides based on binding to a monoclonal antibody. Selected peptides contained a consensus sequence similar to the known epitope of the antibody used for selection.30 Selection of zinc-finger DNA binding proteins.31

Translation System

Protein Displayed

Application

Table 1. Continued

202 Cell-Free Protein Expression

Accelerated Protein Evolution Using Ribosome Display

203

sizes are enabled in ribosome display because the libraries are generated by PCR and need not be transformed into E. coli, a step that limits the size of phage display libraries to that of the eiciency of transformation, usually up to 109 independent clones. In theory, a ribosome display library is only limited by the scale of the PCR and cell-free translation reactions, typically library sizes up to 1014 independent clones are used.12 Larger libraries are particularly useful for protein optimization where a mutagenesis library representing many variants of a single protein is used. In this case, it is likely that the majority of variants will be nonfunctional or at least of reduced ainity due to mutation of essential regions. hose mutations, or combinations of mutations, that will give the desired improved properties are very rare in comparison. he convenience of in vitro versus in vivo techniques also adds to the attraction of ribosome display for protein evolution, especially the opportunity to include genetic variation during selection cycles. PCR-based randomization strategies are conveniently incorporated between iterative rounds of selection. Moreover, there exists an inherent potential for evolution due to opportune incorporation of amino acid changes resulting from random mutations introduced by DNA polymerase. As an alternative to PCR-based mutation, genetic variation may also be introduced by replication of the mRNA library by an RNA-dependent RNA polymerase that ampliies mRNA with a high error rate.13

Evolution of High Affinity Antibodies

he power and convenience of ribosome display for directed evolution have already been described; however in addition ribosome display possesses other features that make it particularly well suited to ainity maturation of antibodies and other binding proteins. First, recovery of the genetic material encoding selected proteins does not require disruption of the binding interaction; therefore genes encoding very high ainity binding molecules are eluted very eiciently. Secondly, the cell-free translation reaction for ribosome display is typically biased to produce monovalent display, i.e., one protein displayed per ribosome, by having an excess of mRNA compared to ribosome content. In this way, selection is based on ‘true’ antibody kinetics in contrast to methods such as phage display, with a degree of polyvalency, where selection of lower ainity antibodies can be assisted by an avidity efect, i.e., the simultaneous binding of antigen by multiple copies of a displayed antibody fragment. Optimum antibody binding equilibrium and kinetics can be paramount when designing a novel therapeutic; optimization tends to be required to reduce the antibody dosage, while maintaining its therapeutic window in vivo. Phage display commonly allows the isolation of antibodies from human naïve libraries (i.e., antibody populations derived from human DNA sources not previously enriched by exposure of the donor to antigen) with initial IC50 values (the antibody concentration that results in 50% inhibition) in biological assays in the 0.2—200 nM range.14,15 Such antibodies may typically be ainity matured by the preparation of phage display mutagenesis libraries of individual clones, leading to antibodies with improved ainity.16,17 Similarly, selection of mutagenesis libraries of individual clones by ribosome display has lead to over 1000-fold increases in potency.18,19 Despite the success of both these approaches, the more rapid isolation of a range of lead antibodies with improved ainities without optimizing individual clones would be desirable, especially for the isolation of antibodies for immunoassays and biomarker studies. An alternative to performing separate antibody isolation and optimization phases in series is to combine both technologies by the application of ribosome display to ainity mature populations of antibodies already enriched by phage display. Antibody gene populations in the plasmid form used for phage display are readily converted to the linear DNA form required for ribosome display. he selection process can then be continued using ribosome display employing appropriate stringent selection pressure and capturing the advantages of ribosome display for ainity optimization. Stable and easily propagated phage libraries are irst enriched over two or three rounds of selection for binding to a target of interest. At this stage populations are still very large (107) and capacity for screening is likely to limit the probability of identifying the highest ainity antibodies. However, the power of ribosome display stringent selections coupled with a randomization strategy can be applied to many potential lead antibodies without requiring their individual identiication and

204

Cell-Free Protein Expression

characterization. A recent study demonstrates this evolutionary power of ribosome display for the ainity maturation of phage display populations.20 Parallel comparisons using either phage display or ribosome display for selection of single chain antibody fragment (scFv) populations derived from three rounds of phage display selection were performed. Stringent selections were performed by reducing concentrations of the antigen step-wise to 1 nM. he highest ainity scFv antibody produced by phage display had a dissociation constant (Kd) of 5.8 nM, whereas ribosome display generated higher ainity variants of this antibody with Kd values of 189 pM and 152 pM, without or with the use of error prone mutagenesis respectively. In addition to these variants, other unrelated antibodies of comparable or higher ainity for insulin, were isolated by ribosome display, but not by phage display, indicating that ribosome display can enrich for diferent populations of antibodies. hese ‘new’ antibodies may be variants of antibodies with ainities too low to be recovered once selection stringency was increased. Alternatively, this may relect the ability of ribosome display to focus phage display derived populations toward high ainity antibodies by lack of an ‘avidity efect.’ As a further alternative, diferences in the expression of antibodies between phage display and ribosome display may have altered the functional representation of clones. Following success in this model system, ainity maturation of phage antibody populations using ribosome display has been applied in our laboratory to the isolation of anti-idiotype antibodies (i.e., antibody-speciic antibodies) for use in pharmacokinetic and immunogenicity assays for a therapeutic antibody against interleukin-13 (IL-13). hree rounds of phage display selections from naïve human libraries were performed using the anti-IL-13 antibody, generating a diverse population of scFv fragments enriched for anti-idiotype antibodies. his population was converted to ribosome display format by PCR and subjected to 4 rounds of ribosome display selection until the antigen concentration had been reduced from 250 nM to 50 pM. Figure 2 shows the neutralization of anti-IL-3 antibody mediated inhibition of IL-13 biological activity by three of the anti-idiotype antibodies isolated by ainity maturation of the phage display pools and by one of the anti-idiotype antibodies isolated from the error prone mutagenesis library. he most potent antibody has a Kd for the anti-IL-13 antibody of 39 pM and has been used in the development of an immunogenicity assay for this antibody to support its clinical development. his strategy,

Figure 2. Neutralization of anti-IL-13 antibody activity by anti-idiotype antibodies generated by the evolution of enriched phage display populations by ribosome display (P_009, P_008 and P_027). The figure shows anti-IL-13 antibody-inhibited induction of VCAM-1 by interleukin-13 is reduced by the anti-idiotypes. Dissociation constants for the anti-idiotypes for the anti-IL-13 antibody are shown in brackets. An additional phage-derived antibody was evolved as a separate clone by ribosome display (C3_034) and is included to show the comparative affinities obtained using the two approaches.

Accelerated Protein Evolution Using Ribosome Display

205

harnessing the advantages of both display technologies and streamlining the isolation-optimization process, is equally applicable to other proteins and features being optimized.

Evolution of Improved Proteins

In vitro translation conditions can readily be tailored and this allows the display of correctly folded proteins. For example, correct folding of disulphide bond containing proteins, such as scFvs, in an E.coli S30 extract is favoured by a nonreducing environment and the inclusion of protein disulphide isomerase.9 Furthermore, addition of the trigger factor and DnaK/DnaJ/GrpE to a reconstituted cell-free translation system lacking endogeneous chaperones increased the solubility and function of a serum albumin-binding scFv.21 In contrast, the cell-free translation environment may also be exploited to discover stabilized protein variants by the deliberate use of a less favourable protein-folding environment for translation and display. Ribosome display has been used to evolve variants of therapeutic proteins with improved expression and increased stability while retaining the same biological properties. For the improvement of stability, expression levels and solubility, the selection strategy is based in principle on using conditions that would normally destabilise and aggregate the protein. hese conditions can be readily adapted to ribosome display since the method operates entirely in vitro. Reducing agents, detergents or proteases are easily added to the translation reaction to interfere with early folding events of the nascent polypeptide chain and eliminate unstable or long-lived folding intermediates.5,22 For many of these strategies it is crucial that the destabilising factor is present during the folding process of the protein. For example, many disulide bonds are diicult to reduce once they are formed as they tend to be engulfed within the protein structure, however the addition of the reducing agent dithiotheritol can be included in a cell-free translation reaction at a suicient level such that all disulides remain reduced. In addition, protein-folding chaperones can be de-

Figure 3. Improved stability compared to wild-type (A) of an erythropoietin variant evolved by ribosome display (B). Percentage relative abundance (%RA) of intact monomer, breakdown products and aggregates when stored for 2 weeks at 45°C are illustrated.

206

Cell-Free Protein Expression

pleted from the translation mix by immunoprecipitation or puriied translation systems that lack endogeneous chaperones can be used.21,23 hese strategies can be combined with the inclusion of proteases or hydrophobic interaction chromatography resins, together with high temperature (10 to 40º C), to eliminate misfolded molecules. Only those molecules that have accumulated stabilizing mutations can survive these selection pressures and importantly, this selection strategy remains compatible with a speciic binding step. he result is that molecules that are stabilized but have lost ainity to their binding partner are eliminated from the sequence pool, leading to the evolution of stabilized but still functional molecules. Such a selection strategy has been used for the generation of variants of the therapeutic protein erythropoietin with improved thermostability (Fig. 3), demonstrating the applicability of ribosome display to evolve therapeutic proteins for the development of second-generation biological drugs.

Conclusion

Cell-free translation technology underpins ribosome display and is exploited to achieve in vitro genotype and phenotype coupling and the application of deined selection pressures. Ribosome display continues to be a leading technology for in vitro protein engineering, with the irst ribosome display evolved therapeutic antibody now in clinical trials. Increasingly wider application of ribosome display, both to the selection of binding proteins other than antibodies and to the optimization of properties other than target binding, are also constantly emerging. Current developments in ribosome display technology are now aimed at the selection of proteins containing nonnatural amino acids19 and these developments will continue to challenge cell-free translation.

Acknowledgements

he authors are grateful to Jon Large for the preparation of illustrations and to Siobhan O’Brien and Andrew Buchanan for their experimental data on anti-idiotype evolution and erythropoietin stability respectively.

References

1. Hoogenboom HR. Selecting and screening recombinant antibody libraries. Nat Biotechnol 2005; 23(9):1105-1116. 2. Carmen S, Jermutus L. Concepts in antibody phage display. Brief Funct Genomic Proteomic 2002; 1(2):189-203. 3. Leemhuis H, Stein V, Griiths AD et al. New genotype-phenotype linkages for directed evolution of functional proteins. Curr Opin Struct Biol 2005; 15(4):472-478. 4. Sawata SY, Suyama E, Taira K. A system based on speciic protein-RNA interactions for analysis of target protein-protein interactions in vitro: successful selection of membrane-bound Bak-Bcl-xL proteins in vitro. Protein Eng Des Sel 2004; 17(6):501-508. 5. Jermutus L, Honegger A, Schwesinger F et al. Tailoring in vitro evolution for protein ainity or stability. Proc Natl Acad Sci USA 2001; 98(1):75-80. 6. Douthwaite JA, Groves MA, Dufner P et al. An improved method for an eicient and easily accessible eukaryotic ribosome display technology. Protein Eng Des Sel 2006; 19(2):85-90. 7. Amstutz P, Binz HK, Zahnd C et al. Ribosome display: in vitro selection of protein-protein interactions. In: Celis JE, ed. Cell biology, A laboratory handbook. Vol 1. 3rd ed. Elsevier Academic Press, 2006:497-503. 8. Schaitzel C, Zahnd C, Amstutz P et al. In Vitro Selection and Evolution of Protein-Ligand Interactions by Ribosome Display. In: Golemis E, Adams P, eds. Protein-Protein Interactions A Molecular Cloning Manual. 2nd ed. New York: Cold Spring Harbor Laboratory Press, 2005:517-548. 9. Schaitzel C, Hanes J, Jermutus L et al. Ribosome display: an in vitro method for selection and evolution of antibodies from libraries. J Immunol Methods 1999; 231(1-2):119-135. 10. Shimizu Y, Inoue A, Tomari Y et al. Cell-free translation reconstituted with puriied components. Nat Biotechnol 2001; 19(8):751-755. 11. Villemagne D, Jackson R, Douthwaite JA. Highly eicient ribosome display selection by use of puriied components for in vitro translation. J Immunol Methods 2006; 313(1-2):140-148. 12. Lamla T, Erdmann VA. Searching sequence space for high-ainity binding peptides using ribosome display. J Mol Biol 2003; 329(2):381-388. 13. Kopsidas G, Roberts AS, Coia G et al. In vitro improvement of a shark IgNAR antibody by Qbeta replicase mutation and ribosome display mimics in vivo ainity maturation. Immunol Lett 2006.

Accelerated Protein Evolution Using Ribosome Display

207

14. Edwards BM, Barash SC, Main SH et al. he remarkable lexibility of the human antibody repertoire; isolation of over one thousand diferent antibodies to a single protein, BLyS. J Mol Biol 2003; 334(1):103-118. 15. Hoet RM, Cohen EH, Kent RB et al. Generation of high-ainity human antibodies by combining donor-derived and synthetic complementarity-determining-region diversity. Nat Biotechnol 2005; 23(3):344-348. 16. Schier R, McCall A, Adams GP et al. Isolation of picomolar ainity anti-c-erbB-2 single-chain Fv by molecular evolution of the complementarity determining regions in the center of the antibody binding site. J Mol Biol 1996; 263(4):551-567. 17. Baker KP, Edwards BM, Main SH et al. Generation and characterization of LymphoStat-B, a human monoclonal antibody that antagonizes the bioactivities of B lymphocyte stimulator. Arthritis Rheum 2003; 48(11):3253-3265. 18. Groves MA, Osbourn JK. Applications of ribosome display to antibody drug discovery. Expert Opin Biol her 2005; 5(1):125-135. 19. Zahnd C, Spinelli S, Luginbuhl B et al. Directed in vitro evolution and crystallographic analysis of a peptide-binding single chain antibody fragment (scFv) with low picomolar ainity. J Biol Chem 2004; 279(18):18870-18877. 20. Groves M, Lane S, Douthwaite J et al. Ainity maturation of phage display antibody populations using ribosome display. J Immunol Methods 2006; 313(1-2):129-139. 21. Ying BW, Taguchi H, Ueda H et al. Chaperone-assisted folding of a single-chain antibody in a reconstituted translation system. Biochem Biophys Res Commun 2004; 320(4):1359-1364. 22. Matsuura T, Pluckthun A. Selection based on the folding properties of proteins with ribosome display. FEBS Lett 2003; 539(1-3):24-28. 23. Tan Z, Blacklow SC, Cornish VW et al. De novo genetic codes and pure translation display. Methods 2005; 36(3):279-290. 24. hom G, Cockrot AC, Buchanan AG et al. Probing a protein-protein interaction by in vitro evolution. Proc Natl Acad Sci USA 2006; 103(20):7619-7624. 25. Binz HK, Amstutz P, Kohl A et al. High-ainity binders selected from designed ankyrin repeat protein libraries. Nat Biotechnol 2004; 22(5):575-582. 26. Schimmele B, Grafe N, Pluckthun A. Ribosome display of mammalian receptor domains. Protein Eng Des Sel 2005; 18(6):285-294. 27. Schimmele B, Pluckthun A. Identiication of a functional epitope of the Nogo receptor by a combinatorial approach using ribosome display. J Mol Biol 2005; 352(1):229-241. 28. Osbourn J, Groves M, Vaughan T. From rodent reagents to human therapeutics using antibody guided selection. Methods 2005; 36(1):61-68. 29. Luginbuhl B, Kanyo Z, Jones RM et al. Directed evolution of an anti-prion protein scFv fragment to an ainity of 1 pM and its structural interpretation. J Mol Biol 2006; 363(1):75-97. 30. Mattheakis LC, Bhatt RR, Dower WJ. An in vitro polysome display system for identifying ligands from very large peptide libraries. Proc Natl Acad Sci USA 1994; 91(19):9022-9026. 31. Ihara H, Mie M, Funabashi H et al. In vitro selection of zinc inger DNA-binding proteins through ribosome display. Biochem Biophys Res Commun 2006; 345(3):1149-1154. 32. Takahashi F, Funabashi H, Mie M et al. Activity-based in vitro selection of T4 DNA ligase. Biochem Biophys Res Commun 2005; 336(3):987-993. 33. He M, Menges M, Groves MA et al. Selection of a human anti-progesterone antibody fragment from a transgenic mouse library by ARM ribosome display. J Immunol Methods 1999; 231(1-2):105-117. 34. Takahashi F, Ebihara T, Mie M et al. Ribosome display for selection of active dihydrofolate reductase mutants using immobilized methotrexate on agarose beads. FEBS Lett 2002; 514(1):106-110. 35. Gersuk GM, Corey MJ, Corey E et al. High-ainity peptide ligands to prostate-speciic antigen identiied by polysome selection. Biochem Biophys Res Commun 1997; 232(2):578-582. 36. Ohashi H, Shimizu Y, Ying BW et al. Eicient protein selection based on ribosome display system with puriied components. Biochem Biophys Res Commun 2007; 352(1):270-276.

Chapter 18

Application of in Vitro Virus (IVV) Technique for High-hroughput Analysis of Protein-Protein Interactions Etsuko Miyamoto-Sato and Hiroshi Yanagawa*

Abstract

G

lobal analysis of protein functions and networks has become the focus of considerable attention since the sequencing of the human genome. he development of proteomics has led to an increasing interest in cell-free translation systems because of their rapidity and ease of handling. We have developed the so-called in vitro virus (IVV) as a more stable and eicient tool for evolutionary protein engineering. his system is applicable as an mRNA display technique to analyze protein functions and networks in proteomics. Accordingly, we developed a high-throughput IVV system for analysis of protein-protein interactions employing cell-free co-translation and selection. Here, we overview this system and discuss the advantages of analyzing protein-protein interactions using IVV as a genotype-phenotype assignment molecule in combination with cell-free co-translation. Moreover, we discuss how we address false positives and negatives in this system, and the in silico mass data processing required to derive biologically signiicant interactions from the wealth of raw data.

Introduction

Since the completion of the Human Genome Project, research has focused on proteome analyses of protein networks and pathways to identify mechanisms of biological functions and diseases. he development of proteomics has led to an increasing interest in cell-free translation systems primarily because it’s a rapid technology. Several new in vitro protein production technologies yielding large amounts of protein have been developed.1,2 mRNA display methods, originally developed for evolutionary protein engineering based on in vitro translation systems, were subsequently applied for the analysis of protein-protein interactions. In mRNA display methods such as in vitro virus (IVV)3-5 and RNA-peptide fusions.6,7 he genotype molecule (mRNA) is linked to the phenotype molecule (protein) through puromycin in a cell-free translation system.8,9 mRNA display is a high-throughput technique because it requires the use of a library, however, the current protocol6 is tedious requiring additional work including isolation of mRNA-protein fusions. Our stable and eicient IVV9 technology allows simple selection without any requirement for post-translational work. An additional technological advancement is the elimination of the need to express and purify bait proteins for downstream protein-protein interaction studies. Our totally in vitro cell-free co-translation system provides a simpler solution that is suitable for high-throughput, genome-wide analysis as baits are synthesized within each reaction. Moreover, *Corresponding Author: Hiroshi Yanagawa—Department of Biosciences and Informatics, Faculty of Science and Technology, Keio University, Yokohama 223-8522, Japan. Email: [email protected]

Cell-Free Protein Expression, edited by Wieslaw A. Kudlicki, Federico Katzen and Robert P. Bennett. ©2007 Landes Bioscience.

Technique for High-hroughput Analysis of Protein-Protein Interactions

209

as co-translation of bait and prey proteins is favorable for the formation of multi-protein complexes10,11, this approach ofers a better chance to obtain a more comprehensive data set including both direct and indirect interactions in a single experiment. Here, we provide an overview of this system and discuss the advantages of analyzing protein-protein interactions and multi-protein complexes using IVV. We also discuss solutions to the issue of false positives and negatives in this system and the in silico mass data processing required to make sense of the data.

IVV: Genotype-Phenotype Assignment Molecules Formed via Puromycin

Puromycin (Fig. 1A), an analogue of the 3' end of aminoacyl-tRNA, is transferred non-specifically to growing polypeptide chains, causing premature termination of translation12-19. However, we found that at very low concentration, puromycin is transferred speciically to the carboxyl (C-) terminus of the full-length protein20. he term puromycin technology9 refers to methods, including IVV display, where an mRNA (genotype molecule) and its protein (phenotype molecule) are linked via puromycin (Fig. 1B), using the technique of speciic in vitro C-terminal protein labeling.9,20 We developed IVV as the irst application of puromycin technology for mRNA display in 1997.8 Shortly thereater, RNA-peptide fusions were independently reported for mRNA display.21 he principle of puromycin technology is employed in both methods, each based on the use of puromycin as a physical linkage between mRNA and the full-length protein which it encodes to obtain an assignment molecule. hese methods have already achieved substantial results including improvements and optimizations in ligation eiciencies and process simpliication. Initial ineiciencies in mRNA-puromycin ligation (about 20-30%8,21) was eventually improved to about 70% by alternative ligation methods such as splint ligation22 and photo-linked ligation.23 Further, we have reported a highly eicient single-strand ligation (over 90% eiciency) using a luorescein-conjugated polyethylene glycol puromycin spacer (PEG-Puro spacer; Fig. 2A).9 he eiciency of fusion formation was initially less than 10%8 or 1%,21 but this has been improved to 40%.22,25 In addition, we have obtained

Figure 1. Structure of puromycin and the principle of IVV formation.20 A) Puromycin is an analog of the 3’-terminal of aminoacyl-tRNA. At very low concentrations, puromycin is transferred specifically to the carboxyl (C-) terminus of the full-length protein.23 B) IVV formation on the ribosome.20 Puromycin at the 3’-terminal end of a spacer (Fig. 2A) ligated to mRNA can enter the ribosomal A-site to bind covalently to the C-terminal end of the encoded full-length protein in the ribosomal P-site.

210

Cell-Free Protein Expression

Figure 2. IVV formation: structure of the PEG puromycin spacer and the mRNA template for puromycin-based technologies. A) The structure of PEG puromycin spacer.3 Fluor-PEG Puro spacer [p(dCp)2-T(Fluor)p-PEGp-(dCp)2-puromycin] was synthesized from Puro(Fmoc)-CPG, polyethylene glycol (PEG, average mol. wt 2,000), deoxycytidine phosphoramidite (dC-amidite), and thymidine(fluorescein) phosphoramidite [T(Fluor)-amidite]. Fluor stands for fluorescein. Puro represents puromycin. B) The structure of the mRNA template for IVV formation.9 The mRNA template is comprised of 5’ UTR, T7-tag, ORF, Flag-tag, and 3’ tail. Five mRNA templates have different 5’ UTR sequences: SP6 + Ω29 [SP6 RNA polymerase transcription promoter (SP6) and translation enhancer Ω29 (CAACAACAACAACAAACAACAACAAAATG), which is a part of the tobacco mosaic virus omega (Ω) sequence (23)], SP6 + Ω, T7 + Ω29 [T7 RNA polymerase promoter (T7) andΩ29], T7 + delta-TMV [T7 and delta-TMV sequence]. and T7 + K [T7 and Kozak (K) sequence]. Four mRNA templates have different 3’ tail sequences: XA8 (CTCGAGAAAAAAAA), X’A8 (CATCACAAAAAAAA), X (CTCGAG), and A8 (AAAAAAAA). The standard template has SP6 + Ω29 as a 5’ UTR and XA8 as a 3’ tail. C) Efficiency of IVV formation on the ribosome. Puromycin at the 3’-terminal end of a spacer (Fluor-PEG Puro) ligated to mRNA can enter the ribosomal A site to covalently bind to the C-terminal end of the protein which it encodes.9

more than 70% fusion formation eiciency (Fig. 2) through single-strand ligation using a lexible PEG-Puro spacer (Fig. 2A) and an optimized mRNA template (Fig. 2B) in a cell-free wheat germ translation system or rabbit reticulocyte lysate translation system. We demonstrated that improvements in the stability of IVV formation can be achieved9 by single-strand ligation using the lexible luorescein-conjugated PEG-Puro spacer (Fig. 2A) instead of DNA8,21 or a triethylene glycol (TEG) spacer.22 We also found that the stability of IVV formation was improved by using mRNA having a 5'-untranslated region including SP6 promoter and Ω29 enhancer (a part of tobacco mosaic virus Ω), and an A8 sequence (eight consecutive adenylate residues) without a stop codon at the 3' end9 (Fig. 2B). Optimization of the 5'- and 3'-terminal sequences of mRNA templates ofers the advantage of easy handling of IVV.9

Applications of IVV as in Vitro Display Technology

In vitro display technology such as IVV,8,9 which were originally developed for evolutionary protein engineering is a powerful tool for analyzing protein functions.25 Advantages include greater library diversity (1012-13/ml) compared with phage display.26 In fact, various aptamers have been

Technique for High-hroughput Analysis of Protein-Protein Interactions

211

successfully evolved from random libraries using this approach.27-30 Evolution of antibody mimics by using mRNA display31 or ribosome display32 and identiication of epitope-like consensus motifs,33 screening a streptavidin-binding aptamer with 10,000-fold-increased ainity,34 and good results in ATP binding27 have also been reported. In many cases, desired functional peptides or proteins have been obtained by the application of display technology. mRNA display methods, including IVV, have been successfully applied to characterizing protein-protein interactions,3-7 DNA-protein interactions,35 and drug-protein interactions.36 Although applications to both evolutionary engineering28 and genome analyses6 have been reported, current protocols are either tedious or not amenable to automated or genomic applications.37 he results of large-scale yeast proteome analysis using the two-hybrid method38,39 and the TAP-mass spectrometry (TAP-MS) method40,41 make it clear that simple large-scale protein-protein interaction data sets give an incomplete picture. Integrating many diferent experimental data sets derived using diferent methods yield a better overall view than any single method alone.42,43 herefore, mRNA display is a potentially powerful tool, adaptable to high-throughput in vitro analysis of protein-protein interactions and multi-protein complexes.

Protein Complexes Analysis; a Co-Translation Technique

Our cell-free co-translation technique using IVV provides a totally in vitro method (Fig. 3) for high-throughput, genome-wide analysis with multiple bait proteins by virtue of using in vitro bait cotranslation instead of in vivo bait preparation.3 As already noted, this approach ofers the best opportunity to identify both direct and indirect protein-protein interactions in a single experiment. One of the key issues in cell-free translation is whether the proteins obtained are properly folded and competent to engage in biologically relevant interactions. Co-translational folding/association is an essential characteristic of prokaryotes and eukaryotes, and so is the case in cell-free translation systems as well.10,11 he fact that protein interaction analysis including both direct and indirect interactions was achieved,3 suggests that IVV selection based on cell-free co-translation is an advance over not only mRNA display technology6,7 but also over the two-hybrid method as well.38,39 IVV ofers several clear advantages over the analysis of amino acid sequences by the TAP-MS or other in vivo methods. As a totally in vitro method, IVV is not biased by cytotoxicity or secretion issues. IVV can also detect proteins present in small amounts because it analyzes genome sequences using the ampliication of mRNA tags instead of amino acid sequences.

Reliability of IVV Data; the Problem of False Positives and Negatives

To minimize the appearance of false positives, we used multiple selection rounds, two-step puriication of the IVV selection products, followed by a post-selection using both real-time PCR and pull-down assays3 to minimize experimental non-speciic interactions (false positives). Two-step puriication of the IVV selection (Fig. 3) is based on the TAP method,40 which provides for clean analysis of protein complexes formed by capture from a crude mixture using a tagged bait protein.40 To further decrease false positives and to obtain information about direct and indirect interactions, an in vitvo post-selection was performed. he post-selection is composed of a pull-down assay to conirm the interactions using C-terminal protein labeling3,9,46,47 and a real-time polymerase chain reaction (PCR) assay to conirm the enrichments. he use of two-step puriication of the IVV selection followed by post-selection provides reliable data for protein-protein interaction analysis. here remains the problem of false negatives buried in the library. False negatives could depend on the number of sequences determined, the initial number of genes in the library, and the interacting ainity. We obtained more than 10 reliable interactors using Fos as bait in a single IVV selection,3 whereas the average number of interactors was only three using the two-hybrid method.44 his suggests that not only false positives, but also false negatives might be lower when using IVV selection and post-selection systems. For example, Jun, Jund1, and Junb, which are all members of the Jun family, were all detected by real-time PCR.3 Although Jun and Jund1 were also detected by

212

Cell-Free Protein Expression

Figure 3. IVV selection based on cell-free co-translation.3 The IVV selection procedure3 is composed of transcription (I), cell-free co-translation (II) coupling translation and interaction of tagged bait protein and prey IVV library (up to 1014 members) to form complexes, IVV selection (III) with bait protein (one or two-step affinity purifications), and RT-PCR (IV) to amplify mRNA tags. Selection rounds are repeated until sufficient enrichment is obtained, followed by cloning and sequencing. Sequence data are analyzed automatically with the IVV analysis system (IWAS; Fujitsu Ltd.), and PPI mapping and in silico analysis are done with Genesphere (Fujitsu Ltd.).

sequence analysis, Junb was not detected, presumably because of insuicient enrichment.3 Recently, the intersection of two protein/peptide interaction data sets obtained with yeast two-hybrid and phage display strategies was utilized to extract even more detailed biological information on the binding partners.43 We propose to use puromycin technology9 in conjunction with IVV selection to identify complex interactions with low false positives and microarrays of C-terminally labeled proteins48,49 to identify direct interactions with low false negatives, thereby allowing the eicient extraction of physicochemical interactions and complexes. We consider that our totally in vitro selection using IVV is reliable for the identiication of physicochemical interactions with low false positives and negatives. We believe that the strength of procedures such as IVV selection is in its ability to accurately identify a large number of direct and indirect physicochemical interactions. Although the biological relevance of the identiied interactions must then be addressed using other methods, in vitro interaction data itself is precious and useful as its sensitivity can provide clues to uncover previously unknown associations, pathways and help elucidate mechanisms underlying biological processes or diseases.

Figure 4. PPI map with phenotype analysis. A PPI map including direct (solid black lines) and indirect interactions (dotted black lines) of Fos interactors3 and Jun interactors4 drawn with Genesphere (Fujitsu). Fos and Jun are the bait proteins (pink square). Proteins (genes) interacting with bait protein are indicated by colored squares: yellow squares, interaction-known proteins; blue squares, interaction-unknown proteins; white squares, function-unknown proteins. Several phenotypes (green hexagons with solid green lines) were analyzed with OMIM (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db = OMIM) using Genesphere (Fujitsu). Zellweger syndrome, Psmc5; kidney tumor, Optn, Jund1, Ras; glaucoma, Optn; liver tumor, Optn, Jun; Alzheimer’s disease, Fos, Gfap; epilepsy, Gfap; brain carcinoma, Cspg6; lymphoma, Jun; Niemann-Pick disease, Jun; leukemia, Atf4.

Technique for High-hroughput Analysis of Protein-Protein Interactions 213

214

Cell-Free Protein Expression

In Silico Analysis of IVV Selection for AP-1 (Fos/Jun) Complex

Mapping protein-protein interactions of IVV selection products (Fig. 3) was demonstrated using the selection results obtained with Fos3 and Jun4 as bait, followed by automatic sequence data analysis. In the irst step of such analysis, IWAS (Fujitsu Ltd.), which is capable of dealing with 100 sequences/hr,3 was used to characterize the cDNA region by omitting vector and IVV common sequences and then analyzing the remaining sequences using gene databases such as nt, refseq, nBLAST, and FANTOM2. he result was a gene catalog of the IVV database.3 he gene catalog was then subjected to protein-protein interaction mapping. In silico analyses of phenotype, function and localization was then performed with LocusLink,49 UniGene and OMIM using Genesphere (Fujitsu Ltd.). Genesphere is a sotware capable of searching for connected and related genes and proteins from a comprehensive data set using LocusLink, UniGene, OMIM, and PubMed (Fig. 5). We previously obtained 31 interactors in IVV selection from a mouse brain cDNA library using Fos3 and Jun,4 components of transcription factor AP-1,50 as bait. We identiied 26 unknown interactors (84%) and 5 known interactors (16%). Among our interactors, 22 were direct (71%) and 9 were indirect, including some with weak ainity (29%). he results of phenotype analysis are summarized in Figure 4. In the case of Jun57 as bait, we obtained two novel Jun interactors, Gfap and Cspg6. hese have been identiied in brain-related diseases such as Alzheimer’s disease, brain carcinoma, and epilepsy. In the case of Fos50 as bait, we obtained Optn, which was identiied as a causative gene of glaucoma.51 Optn is an unknown-function gene whose product has been implicated in the tumor necrosis factor signaling pathway and interacts with Huntington and Ras-associated protein.51 his was the irst report of the transcription factor Fos interacting with Optn, presumably via their bZIP domains. Using PubMed, Eef1d, a protein containing an L-ZIP domain, was identiied as a potential proto-oncogene, which is also overexpressed with fos, jun, and myc under certain conditions.52 An in silico analysis of the biological functions of these AP-1 (Fos/Jun) interactors was also performed. Fos functions as a transcription factor, having nucleic acid-binding and transcription-regulatory activities, while Jun is also associated with completely diferent gene functions, such

Figure 5. In silico function analysis. The function analysis was performed with LocusLink in the gene ontologies of MGI (http://www.informatics.jax.org/searches/GO_form.shtml) using Genesphere (Fujitsu). Genesphere is software that searches for connected and related genes and proteins from a comprehensive data set using LocusLink, UniGene, OMIM, and PubMed.

Technique for High-hroughput Analysis of Protein-Protein Interactions

215

as motor-related activities and structural molecules (Fig. 5). he results suggest that Fos and Jun may have biological functions that extend beyond their partnership in the AP-1 complex. Since these interactions include products of disease-related genes, a further study of these novel interactions may provide clues to new pathways or mechanisms of biological functions and diseases.

Outlook

Protein-protein interactions have been recently studied on a large scale by using yeast two-hybrid systems for the human interactome (2,80053 and 3,18654 interactions). However, a probabilistic analysis integrating model organism interactome data predicts nearly 40,000 protein-protein interactions in humans.55 herefore, it appears that existing large-scale protein interaction data sets are nonsaturating and that integrating many diferent experimental data sets should yield a clearer biological view than a single method alone. herefore, mRNA display should become a useful tool in uncovering these interactions. In addition, proteomics with the TAP-MS method40,41 faces the problem of the detection of low levels of proteins in cells and analysis of the amino acid sequence-based data using standard bioinformatics. However, IVV’s sensitivity lends itself to picking up interactions between even low abundance proteins. Coupled with the ability to screen using mRNA tag sequence data, this technique ofers a strong advantage over traditional techniques. Also, since IVV is a totally in vitro technique, cytotoxicity is not a concern. In addition, by working in vitro, very large cDNA libraries can be screened to detect multiple interactors.3,9 Lastly, not only protein-protein,3 but also protein-DNA interactions35 can be analyzed using IVV. Accordingly, the IVV method represents a large technological step forward in comprehensive analysis of biological interactions on a genomic scale.

Acknowledgements

he authors would like to thank T. Yamashita of Fujitsu Ltd. for preparing the igures. hey gratefully acknowledge the contributions of Drs. M. Ishizaka, S. Tateyama, and K. Horisawa of Keio University. Special thanks are also due to K. Masuoka, and N. Hirai of Keio University. he part of this work was supported by grants of the Genome Network Project from the Ministry of Education, Culture, Sports, Science and Technology, Japan.

References

1. Kigawa T, Yabuki T, Yoshida Y et al. Cell-free production and stable-isotope labeling of milligram quantities of proteins. FEBS Lett 1999; 442(1):15-19. 2. Madin K, Sawasaki T, Ogasawara T et al. A highly eicient and robust cell-free protein synthesis system prepared from wheat embryos: plants apparently contain a suicide system directed at ribosomes. Proc Natl Acad Sci USA 2000; 97(2):559-564. 3. Miyamoto-Sato E, Ishizaka M, Horisawa K et al. Cell-free cotranslation and selection using in vitro virus for high-throughput analysis of protein-protein interactions and complexes. Genome Res 2005; 15(5):710-717. 4. Horisawa K, Tateyama S, Ishizaka M et al. In vitro selection of Jun-associated proteins using mRNA display. Nucleic Acids Research 2004; 32(21):e169. 5. Miyamoto-Sato E, Yanagawa H. Toward functional analysis of protein interactome using “in vitro virus”: in silico analyses of Fos/Jun interactors. Journal of Drug Targeting 2006; in press. 6. Hammond PW, Alpin J, Rise CE et al. In vitro selection and characterization of Bcl-X(L)-binding proteins from a mix of tissue-speciic mRNA display libraries. J Biol Chem 2001; 276(24):20898-20906. 7. Shen X, Valencia CA, Szostak L et al. Scanning the human proteome for calmodulin-binding proteins. Proc Natl Acad Sci USA 2005; 102(17):5969-5974. 8. Nemoto N, Miyamoto-Sato E, Husimi Y et al. In vitro virus: bonding of mRNA bearing puromycin at the 3’-terminal end to the C-terminal end of its encoded protein on the ribosome in vitro. FEBS Lett 1997; 414(2):405-408. 9. Miyamoto-Sato E, Takashima H, Fuse S et al. Highly stable and efficient mRNA templates for mRNA-protein fusions and C-terminally labeled proteins. Nucleic Acids Res 2003; 31(15):e78. 10. Jemutus L, Ryabova LA, Plückthun A. Recent advances in producing and selecting functional proteins by using cell-free translation. Recent advances in producing and selecting functional proteins by using cell-free translation. Curr Opin Biotechnol 1998; 9(5):534-548. Review.

216

Cell-Free Protein Expression

11. Fedorov AN, Baldwin TO. Cotranslational protein folding. J Biol Chem 1996; 272(52):32715-32718. Review. 12. Yarmolinsky MB, Haba GL. Proc. Natl Acad Sci USA 1959; 45(12):1721-1729. 13. Allen DW, Zamecnik PC. he efect of puromycin on rabbit reticulocyte ribosomes. Biochim Biophys Acta 1962; 55:865-874. 14. Nathans D. Puromycin inhibition of protein synthesis: incorporation of puromycin into peptide chains. Proc Natl Acad Sci USA 1964; 51:585-592. 15. Traut RR, Monro RE. he puromycin reaction and its relation to protein synthesis. J Mol Biol 1964; 10:63-72. 16. Zamir A, Leder P, Elson D. A ribosome-catalyzed reaction between n-formylmethionyl-trna and puromycin. Proc Natl Acad Sci USA 1966; 56(6):1794-1801. 17. Nathans D, Neidle A. Structural requirements for puromycin inhibition of protein synthesis. Nature 1963; 197:1076-7. 18. Steiner G, Kuechler E, Barta A. Photo-ainity labelling at the peptidyl transferase centre reveals two diferent positions for the A- and P-sites in domain V of 23S rRNA. EMBO J 1988; 7(12):3949-55. 19. Kirillov S, Porse BT, Vester B et al. Movement of the 3'-end of tRNA through the peptidyl transferase centre and its inhibition by antibiotics. FEBS Lett 1997; 406(3):223-233. 20. Miyamoto-Sato E, Nemoto N, Kobayashi K et al. Speciic bonding of puromycin to full-length protein at the C-terminus. Nucleic Acids Res 2000; 28(5):1176-1182. 21. Roberts RW, Szostak JW. RNA-peptide fusions for the in vitro selection of peptides and proteins. Proc Natl Acad Sci USA 1997; 94(23):12297-12302. 22. Liu R, Barrick JE, Szostak JW et al. Optimized synthesis of RNA-protein fusions for in vitro protein selection. Methods Enzymol 2000; 318:268-293. 23. Kurz M, Gu K, Lohse PA. Psoralen photo-crosslinked mRNA-puromycin conjugates: a novel template for the rapid and facile preparation of mRNA-protein fusions. Nucleic Acids Res 2000; 28(18):E83. 24. Barrick JE, Takahashi TT, Balakin A et al. Selection of RNA-binding peptides using mRNA-peptide fusions. Methods 2001; 23(3):287-293. 25. Amstutz P, Forrer P, Zahnd C et al. In vitro display technologies: novel developments and applications. Cur Opin Biotechnol 2001; 12(4):400-405. 26. Smith GP. Filamentous fusion phage: novel expression vectors that display cloned antigens on the virion surface. Science 1985; 228(4705):1315-7. 27. Wilson DS, Keefe AD, Szostak JW. he use of mRNA display to select high-ainity protein-binding peptides. Proc Natl Acad Sci USA 2001; 98(7):3750-3755. 28. Keefe AD, Szostak JW. Functional proteins from a random-sequence library. Nature 2001; 410(6829):715-718. 29. Barrick JE, Takahashi TT, Ren J et al. Large libraries reveal diverse solutions to an RNA recognition problem. Proc Natl Acad Sci USA 2001; 98(22):12374-12378. 30. Cujec TP, Medeiros PF, Hammond P et al. Selection of v-abl tyrosine kinase substrate sequences from randomized peptide and cellular proteomic libraries using mRNA display. Chem Biol 2002; 9(2):253-264. 31. Xu L, Aha P, Gu K et al. Directed evolution of high-ainity antibody mimics using mRNA display. Chem Biol 2002; 9(8):933-942. 32. Hanes J, Schaitzel C, Knappik A et al. Picomolar ainity antibodies from a fully synthetic naive library selected and evolved by ribosome display. Nat Biotechnol 2000; 18(12):1287-1292. 33. Baggio R, Burgstaller P, Hale SP et al. Identiication of epitope-like consensus motifs using mRNA display. J Mol Recognit 2002; 15(3):126-134. 34. Li S, Millward S, Roberts R. In vitro selection of mRNA display libraries containing an unnatural amino acid. J Am Chem Soc 2002; 124(34):9972-9973. 35. Tateyama S, Horisawa K, Takashima H et al. Ainity selection of DNA-binding protein complexes using mRNA display. Nucleic Acids Res 2006; 34:e27 36. McPherson M, Yang Y, Hammond PW et al. Drug receptor identiication from multiple tissues using cellular-derived mRNA display libraries. Chem Biol 2002; 9(6):691-698. 37. Gold L. mRNA display: Diversity matters during in vitro selection. Proc Natl Acad Sci USA 2001; 98(9):4825-4826. 38. Uetz P, Giot L, Cagney G et al. A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae. Nature 2000; 403(6770):623-627. 39. Ito T, Chiba T, Ozawa R et al. A comprehensive two-hybrid analysis to explore the yeast protein interactome. Proc Natl Acad Sci USA 2001; 98(8):4569-4574. 40. Rigaut G, Shevchenko A, Rutz B et al. A generic protein puriication method for protein complex characterization and proteome exploration. Nat Biotechnol 1999; 17(10):1030-1032.

Technique for High-hroughput Analysis of Protein-Protein Interactions

217

41. Gavin AC, Bösche M, Krause R et al. Functional organization of the yeast proteome by systematic analysis of protein complexes. Nature 2002; 415(6868):141-147. 42. Bader GD, Hogue CWV. Analyzing yeast protein-protein interaction data obtained from diferent sources. Nat Biotechnol 2002; 20(10):991-7. 43. Legrain P. Protein domain networking. Nat Biotechnol 2002; 20(2):128-129. Legrain P, Selig L. Genome-wide protein interaction maps using two-hybrid systems. FEBS Lett 2000; 480(1):32-36. Review. 44. Legrain P, Wojcik J, Gauthier JM. Protein-protein interaction maps: a lead towards cellular functions. Trends Genetics 2001; 17(6):346-352. Review. 45. Nemoto N, Miyamoto-Sato E, Yanagawa H. Fluorescence labeling of the C-terminus of proteins with a puromycin analogue in cell-free translation systems. FEBS Lett 1999; 462(1-2):43-46. 46. Doi N, Takashima H, Kinjo M, Sakata K et al. Novel luorescence labeling and high-throughput assay technologies for in vitro analysis of protein interactions. Genome Res 2002; 12(3):487-492. 47. Kawahashi Y, Doi N, Takashima H et al. In vitro protein microarrays for detecting protein-protein interactions: Application of a new method for luorescence labeling of proteins. Proteomics 2003; 3(7):1236-1243. 48. Wheeler DL, Church DM, Edgar R et al. Database resources of the National Center for Biotechnology Information: update. Nucleic Acids Res 2004; 34(Database issue):D173-180. 49. Chinenov Y, Kerppola TK. Close encounters of many kinds: Fos-Jun interactions that mediate transcription regulatory speciicity. Oncogene 2001; 20(19):2438-2452. Review. 50. Rezaie T, Child A, Hitchings R et al. Adult-onset primary open-angle glaucoma caused by mutations in optineurin. Science 2002; 295(5557):1077-1079. 51. Joseph P, Lei YX, Whong WZ et al. Oncogenic potential of mouse translation elongation factor-1δ, a novel cadmium-responsive proto-oncogene. J Biol Chem 2002; 277(8):6131-6136. 52. Rual JF, Venkatesan K, Hao T et al. Towards proteome-scale map of the human protein-protein interaction network. Nature 2005; 437(7062):1173-1178. 53. Stelzl U, Worm U, Lalowski M et al. A human protein-protein interaction network: a resource for annotating the proteome. Cell 2005; 122 (6):957-968. 54. Rhodes DR, Tomlins SA, Varambally S et al. Probabilistic model of the human protein-protein interaction network. Nat Biotechnol 2005; 23(8):951-959.

Index A Ainity 10, 71, 93 Aggregation 37, 50, 76, 79-81, 88, 90 AKT1 20, 21, 27, 28 Antibody 3, 6-10, 20, 23-28, 51, 53, 55, 61, 65, 76, 79-81, 88, 96

B B-barrel 85, 89, 90, 92 B-cell lymphoma 53, 59-61, 66, 69 Batch/feed reaction 34, 37 Bicell 92 Biological activity 59 Biosynthesis 56, 76 Biotin 10, 23, 26 Brij 88, 90-92, 96 Brij-35 90, 92 Brij-58 88, 90 Brij-78 90, 92, 96 Brij-98 90

C Cell-free 1, 2, 4, 6-9, 11, 13, 19-25, 27, 28, 31-35, 37-39, 42-44, 46-51, 53-71, 76, 79, 80, 82, 84, 85, 87-89, 91, 92, 94-96

Cell-free expression 19, 20, 23, 24, 28, 31, 33, 35, 37-39, 50, 53, 55, 71, 84, 85, 87, 88, 91, 92, 94, 96 Cell-free protein synthesis 31, 42, 43, 46, 48, 51, 53-56, 58, 60-68, 71 Cell-free translation 2, 42, 76, 79, 80 Chaperone 5, 8, 25, 54, 55, 58, 59, 65, 68, 76, 78-82, 94 Chaperone-dependency 82 Citric acid cycle (TCA) 47, 48, 51 ClpB 80, 81 Co-translational 1, 2, 5, 6, 8, 82, 89, 99 Construct screening 23 Continuous exchange cell-free (CECF) 32-34, 37, 49, 84, 85, 92, 93, 96, 99 Continuous in vitro evolution (CIVE) 9, 10 Coupled transcription and translation 11 Crystallization 33

D Deglycosylation 23 Detergent 3, 37, 51, 84, 86, 88-92, 94 DHPC (1, 2-diheptanoyl-sn-glycero-3phosphocholine) 89, 90 Diagnostic 6, 32, 37 diC8PC (1,2-dioctanoyl-sn-glycero-3phosphocholine) 89, 90

220

Dihydrofolate reductase (DHFR) 77-79, 81 Directed evolution 9, 11 Dissociation constant 92 Disulide bond formation 2, 42, 53, 55, 58 DMPC (1, 2-dimyristoyl-sn-glycero-3phosphocholine) 91 DnaJ 21, 58, 80, 81 DnaK 58, 80, 81 Dodecyl-phosphocholine (DPC) 89

E EasyXpress 20, 22-25, 170, 172-176 E. coli S30 extract 11, 57, 80, 110, 160, 200-202 EndoH 23, 27 Endoplasmic reticulum-associated degradation (ERAD) 1, 5, 6 Energy sources 32, 33, 42-51, 55-57, 66, 68, 69, 71, 76, 125 Esherichia coli 1, 3, 11, 19-25, 28, 31-34, 37-39, 43-48, 50, 54-62, 65-68, 71, 76-80, 85, 91, 92, 94, 96, 97, 99, 101, 107, 108, 110, 118, 121, 122, 125, 136, 139-142, 147, 149, 159, 160, 166, 169-173, 187, 193-195, 200-203 Eukaryotic ribosome display 9 Evolutionary protein engineering 208, 210 Expression labeling 155-157, 159-162 Expression screen 86, 91, 146 Expressway 36, 110

Cell-Free Expression

F Feeding mixture (FM) 93 Fluor-Puro 181-188 Fluorescein isothiocyanate (FITC) 184 Fluorescence 33, 35, 36, 39, 77, 81, 112, 128, 39, 142, 146-148, 151, 152, 155-157, 159-162, 166-169, 176, 177, 181, 183-185, 187, 188 Fluorescence cross-correlation spectroscopy (FCCS) 159, 181, 183 Fluorescent amino acid 155-157, 160-162, 166, 169, 176, 177 Fluorophore 155, 159-161, 181-184 Fusion protein 8, 59-61, 65, 66, 70, 86, 147, 148, 150, 151, 155

G G-protein coupled receptor (GPCR) 25, 28, 51, 89, 90, 92, 93, 160 Gateway recombination 35 Genome 35, 37, 50, 61, 76, 78, 117, 124, 136, 145, 194, 208, 211, 215 Genotype-phenotype assignment molecule 208 Glycoprotein 2, 3, 22, 26, 28 Glycosylation 3, 4, 6, 19, 23, 26-28, 39, 55, 85, 145, 186, 200 GPI anchor 2, 4 Green luorescent protein (GFP) 46-48, 50, 77, 81, 96, 112, 128, 147, 148, 150, 151, 155, 182, 185, 187 GroEL 58, 80, 81, 82 GroES 58, 80, 81 GrpE 80, 81, 205

221

Index

H Hexahistidine 135 High throughput 42, 49, 51, 120, 125, 135, 161, 166, 167, 196 High-throughput protein expression 125, 131 Hot-rod gene 190, 192, 194, 196 Human b2 adrenergic receptor (b2AR) 92, 93 Human M2 muscarinic acetylcholine receptor (M2) 3, 92

I Immobilization 6-8, 135, 145, 146, 148-150, 152, 166, 174-177 Immunoprecipitation 8, 55, 65, 66, 68, 206 Immunoprecipitation assay 65, 66 Insect cells 19, 20, 23, 24, 28, 55, 93, 125, 127, 141 Intein 149, 150, 155 Interactome 215 In vitro In vitro compartmentalization (IVC) 1, 9-11, 38, 187 In vitro display 210 In vitro display technology 210 In vitro expression 1, 6, 11, 12, 108, 114, 141, 147, 161, 162 In vitro expression cloning (IVEC) 1, 11, 12 In vitro protein expression 19, 53, 125 In vitro selection 10, 158, 187, 198, 212 In vitro translation 13, 19, 42, 76, 77, 82, 125, 157, 162, 170, 181, 187, 188, 199, 200, 205, 208

In vitro virus (IVV) 9, 10, 208, 209-212, 214, 215 Iodoacetamide (IAM) 56-59, 62, 65-69, 71 Isotopic labeling 108, 109, 112, 114

K KD

8 Kinase 8, 12, 13, 19-21, 24, 27, 28, 35, 44, 47, 77-79, 134, 135, 141, 201

L Labeling 3, 4, 33-35, 38, 98, 101, 107-114, 117-120, 122, 146, 147, 155-157, 159-162, 166, 167, 174, 176, 181-185, 187, 188, 209 Library screening 181, 188 Linear template 20, 22, 23, 24, 127, 128, 174 Liposomes 37, 84, 91, 92, 94, 99, 118, 119, 121, 122 Liquid chromatography mass spectroscopy (LC-MS) 188 LMPG (1-myristoyl-2-hydroxy-sn-glycero-33 [phospho-rac-(1-glycerol)]) 88, 91, 98 Luciferase 7, 20, 22, 25, 26, 49, 159 Lysate 1, 2, 4, 6, 8, 9, 11, 19, 20, 24, 25, 28, 31, 33, 42, 43, 47, 76, 107, 108, 135, 136, 141, 148-151, 161, 169-173, 175, 185, 200, 202, 210

M Magnetic bead 174, 175 Mass spectroscopy 122, 139 Maturation 54, 76, 78- 80, 202-204

222

Membrane protein (MP) 2-4, 19, 20, 25, 28, 29, 33, 37, 50, 59, 82, 84-99, 117-120, 126, 141, 160, 201 Membrane topology 2, 3 Micellar concentration (CMIC) 89, 90 Micell 84, 86, 88-92, 94, 98, 118 Microarray 1, 6-8, 13, 134-136, 139, 141, 145-148, 150-152, 155, 159, 161, 166, 176, 183, 212 Molecular chaperone 54, 58, 76, 80 mRNA display 9, 10, 38, 76, 78, 187, 188, 208, 209, 211, 215 mRNA library 1, 187, 199, 203 MTSL (1-oxyl-2,2,5,5-tetramethyl-3pyrroline -3-methyl-methane-thiosulfonate 97, 98 Multidimensional NMR 103, 104 Multi-line split DNA synthesis 188 Myristoylation 2, 4-6, 13

N N-decyl-b-D-maltoside (DM) 89, 90 N-dodecyl-b-D-maltoside 88-90 N-octyl-b-glucopyranoside (b-OG) 89 Naphthylalanine 169, 170 Nuclear magnetic resonance (NMR) 33, 34, 38, 39, 79, 84, 92, 94-99, 101-108, 110-114, 145 NMR assignment 105 NMR spectroscopy 84, 94, 99, 103, 108 Nuclear overhauser efect (NOE) 98 Nucleoside triphosphates 33, 43, 56-58

Cell-Free Expression

P Paramagnetic relaxation enhancement (PRE) 97, 98 Phage display 9, 38, 176, 198, 201, 203, 204, 210, 212 Phosphorylation 8, 19, 26-28, 48, 56, 57, 68, 85, 200 Planned pause gene setsTM 190, 194 Post-translational 1, 2, 7, 8, 20, 37, 55, 76, 82, 135, 185, 208 Potency 203 Prenylation 2, 4, 6 Productivity 42, 53, 55, 56, 67, 77, 79, 82, 85 Protein Protein-DNA interactions 8, 11, 183 Protein-protein interaction 4, 8, 10, 11, 25, 139-141, 155, 159, 162, 174, 177, 183, 187, 208, 209, 211, 214, 215 Protein-RNA interaction 8 Protein array 6-8, 33, 37, 124, 134-137, 139-142, 145, 150 Protein biosynthesis 56, 76, 167, 170, 173, 177 Protein chip 150, 166, 181 Protein complex 11, 23, 25, 26, 94, 99, 140, 150, 209, 211 Protein conjugate 166-168, 171, 174, 177 Protein domain 192, 193 Protein engineering 34, 39, 155, 156, 159, 185, 190, 206, 208, 210 Protein evolution 1, 38, 198, 200, 203

223

Index

Protein expression 2, 6, 7, 8, 19, 20, 23, 28, 31, 33-35, 37, 38, 42, 43, 53, 55, 67, 69, 70, 101, 107-110, 114, 117, 118, 120, 121, 124-129, 131, 145-148, 150, 155, 167, 181, 187, 190-194, 196 Protein folding 4, 33, 42, 55-57, 65, 66, 68, 76, 79, 81, 82, 141, 182, 190, 192-194 Protein function 7, 9, 25, 35, 39, 43, 134, 145, 166, 167, 190, 191, 208, 210 Protein interaction 4, 6, 8, 10, 11, 25, 38, 139-141, 146, 149, 155, 159, 160, 162, 166, 174, 177, 183, 184, 187, 208, 209, 211, 214, 215 Protein labeling 209 Protein microarray 6, 7, 13, 134, 135, 141, 145, 146, 148, 150-152, 161 Protein modiications 166 Protein optimization 203 Protein solubility 81, 117, 118, 167, 175 Protein structure 28, 38, 106, 114, 117, 160, 191, 192, 205 Protein therapeutics 53, 70 Protein truncation test (PTT) 1, 13, 162 Proteome 84, 136 Proteomics 1, 2, 6, 13, 76, 99, 120, 127, 145, 146, 150, 156, 160, 177, 181, 187, 208, 215 Pure system 170 Puromycin 6, 9-11, 158, 159, 181-183, 185-188, 208-210, 212 Puromycin analogue technology 181, 187 Puromycin technology 209

R Rabbit reticulocyte 1, 2, 19, 20, 24, 28, 43, 47, 76, 125, 136, 160, 185, 200, 202, 210 Rabbit reticulocyte lysate (RRL) 1-11, 13, 19, 20, 24, 27, 28, 76, 185, 186, 200, 202, 210 Rapid protein production 145 Rat neurotensin receptor (NTR) 92 Reconstitution 8, 33, 79, 91 Release factor (RF1) 77, 160, 166, 169-173, 176, 177, 187 Ribosome display 9, 38, 78, 198-206

S Screening 6, 9, 11, 13, 19, 20, 23, 24, 33, 35-37, 51, 76, 78, 82, 86, 91, 93-95, 118, 124, 134, 145-148, 150, 151, 155, 160-162, 169, 177, 181, 188, 198, 203, 211 SDS-PAGE 23, 25-27, 59, 96, 120, 122, 155, 159, 161, 162, 170-172, 175, 176, 182, 185, 196 Selection 1, 9-11, 38, 78, 86, 95, 127, 158, 169, 171, 177, 187, 192, 198-206, 208, 211, 212, 214 Selenomethionine 117, 167 Serum proiling 141 Single-chain variable fragments (scFv) 55, 58-61, 66, 67, 70, 79-81, 201, 202, 204, 205 Single chain 7, 9, 60, 61, 65, 68, 174, 176, 177, 204 Single nucleotide polymorphisms (SNPs) 192 Site-directed 13, 156, 160, 166-169, 176, 177 Site-speciic 35, 99, 148-150, 155, 156, 160, 167, 168, 173, 174, 176

224

Cell-Free Expression

Solubility 23, 26, 28, 37, 59, 78-81, 107, 117-119, 122, 129, 167, 175, 177, 184, 191-194, 205 Speciic labeling 34, 101, 108, 114, 159, 176 Spodoptera frugiperda (SF) 19, 20, 24, 28 Stability 5, 8, 37, 38, 53, 57, 58, 61, 62, 78, 101, 107, 110, 131, 137, 184, 190, 199-202, 205, 206, 210 Stoichiometric 169, 173 Strep tag 23, 25 Sulhydryl redox potential 57 Suppression 88, 158, 161, 169-171 Suppressor 9, 35, 36, 156-162, 166-174, 176 Suppressor tRNA 35, 36, 156, 157, 159, 161, 166, 168-174, 176 Synthetic biology 82, 190

Transverse relaxation optimised spectroscopy (TROSY) 96-98 Trigger factor 80, 81, 205 Triton x-100 (polyethylene-glycol P-1, 1, 3, 3-tetramethyl-butylphenyl-ether) 89 tRNA 10, 13, 35, 36, 58, 77, 78, 94, 125, 155-161, 166-174, 176, 183, 191, 192, 209 Tunicamycin 3, 23, 27

T

W

hroughput expression 33, 35, 93, 125, 161 Trans-splicing 149, 150 Transcription factor 20-22, 25, 214 Translational kinetics 192, 193 Translational step-times 192, 193 Translation engineering 190-192 Transmembrane segment (TMS) 93, 99 Transporter 4, 25, 84, 88-94, 97, 98, 118, 141

Wheat germ 1, 11, 19, 33, 43, 47, 58, 76, 85, 99, 108, 125, 126, 136, 141, 160, 185, 200, 202, 210

U Unnatural amino acid 126, 166-170, 173

V Vaccine 53-56, 60, 61, 64, 66, 67, 69-71, 136, 140, 196 Viral assembly 6

X X-ray crystallography 34, 38, 117, 118, 120, 145

Y Yersinia pestis 37, 136

BIOTECHNOLOGY INTELLIGENCE UNIT

INTELLIGENCE UNITS

The chapters in this book, as well as the chapters of all of the five Intelligence Unit series, are available at our website.

KudliCKi • KAtzEn • BEnnEtt

Biotechnology Intelligence Unit Medical Intelligence Unit Molecular Biology Intelligence Unit Neuroscience Intelligence Unit Tissue Engineering Intelligence Unit

Biu

Cell-Free Protein Expression