Peptide and Protein Design for Biopharmaceutical Applications

Peptide and Protein Design for Biopharmaceutical Applications

Peptide and Protein Design for Biopharmaceutical Applications © 2009 John Wiley & Sons, Ltd. ISBN: 978-0-470-31961-1

Edited by Knud J. Jensen

Peptide and Protein Design for Biopharmaceutical Applications Editor Knud J. Jensen Faculty of Life Sciences, University of Copenhagen, Denmark

A John Wiley and Sons, Ltd, Publication

This edition first published 2009 Ó 2009 John Wiley & Sons Ltd Registered office John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, United Kingdom For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com. The right of the author to be identified as the author of this work has been asserted in accordance with the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books. Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought. The publisher and the author make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of fitness for a particular purpose. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for every situation. In view of ongoing research, equipment modifications, changes in governmental regulations, and the constant flow of information relating to the use of experimental reagents, equipment, and devices, the reader is urged to review and evaluate the information provided in the package insert or instructions for each chemical, piece of equipment, reagent, or device for, among other things, any changes in the instructions or indication of usage and for added warnings and precautions. The fact that an organization or Website is referred to in this work as a citation and/or a potential source of further information does not mean that the author or the publisher endorses the information the organization or Website may provide or recommendations it may make. Further, readers should be aware that Internet Websites listed in this work may have changed or disappeared between when this work was written and when it is read. No warranty may be created or extended by any promotional statements for this work. Neither the publisher nor the author shall be liable for any damages arising herefrom. The cover figure showing Cilengitide was kindly provided by Horst Kessler and Oliver Demmer. Library of Congress Cataloging-in-Publication Data Peptide and protein design for biopharmaceutical applications / editor, Knud J. Jensen. p. ; cm. Includes bibliographical references and index. ISBN 978-0-470-31961-1 1. Peptide drugs—Design. 2. Protein drugs—Design. 3. Peptides—Design. 4. Protein engineering. I. Jensen, Knud J. [DNLM: 1. Peptides—chemistry. 2. Proteins—chemistry. 3. Drug Design. QU 68 D4575 2009] RS431.P38D47 2009 6150 .19—dc22 2009019177 A catalogue record for this book is available from the British Library. ISBN: 978-0-470-31961-1 (H/B) Set in 10.5/13pt Sabon by Integra Software Services Pvt. Ltd, Pondicherry, India Printed and bound in Great Britain by TJ International Ltd., Padstow, Cornwall

Contents

List of Contributors Preface

ix xi

1 Introduction Knud J. Jensen

1

2 Computational Approaches in Peptide and Protein Design: An Overview Gregory V. Nikiforovich and Garland R. Marshall 2.1 Introduction 2.2 Basics and Tools 2.2.1 The Importance of Computational Approaches 2.2.2 Tools and Procedures: Force Fields and Sampling 2.3 Computational Study of Cyclopentapeptide Inhibitors of CXCR4 2.3.1 The 3D Pharmacophore Model for FC131 2.3.2 A 3D Model of the TM Region of CXCR4 2.3.3 Docking of FC131 to CXCR4 Acknowledgements References 3 Aspects of Peptidomimetics Veronique Maes and Dirk Tourwe´ 3.1 Introduction 3.2 Modified Peptides 3.3 Pseudopeptides

5

5 6 6 9 31 32 36 39 42 42 49

49 51 65

vi

CONTENTS

3.4 Secondary Structure Mimics (Excluding Turn Mimics) 3.4.1 b-strand Mimetics 3.4.2 Helix Mimetics 3.5 Examples of Peptidomimetics 3.6 Conclusion References 4 Design of Cyclic Peptides Oliver Demmer, Andreas O. Frank and Horst Kessler 4.1 Introduction 4.1.1 Pharmaceutical Research Today 4.1.2 General Advantages of Cyclic Peptide Structures 4.1.3 Examples of Cyclic Peptides of Medicinal Interest 4.1.4 General Considerations 4.2 Peptide Cyclization 4.2.1 Possibilities of Peptide Cyclization 4.2.2 Synthesis of Cyclic Peptides 4.2.3 Chemical Modifications of Cyclic Peptides 4.2.4 Concluding Remarks 4.3 Conformation and Dynamics of Cyclic Peptides 4.3.1 Reductions in Conformational Space 4.3.2 Conformational Arrangements in Cyclic Structures 4.3.3 Flexibility of Cyclized Scaffolds 4.3.4 Experimental Structure Characterization 4.4 Concepts in the Rational Design of Cyclic Peptides 4.4.1 The Influence of Amino Acid Composition 4.4.2 The Dunitz–Waser Concept 4.4.3 The Spatial Screening Technique 4.4.4 General Strategy for Finding Active Hits 4.5 Examples of Cyclic Peptides as Drug Candidates 4.5.1 Cilengitide as Integrin Inhibitor 4.5.2 CXCR4 Antagonists 4.5.3 Sandostatin and the Veber–Hirschmann Peptide as Examples of Rational Design 4.6 Conclusion References

75 75 87 92 104 105 133

133 133 134 135 137 138 138 139 141 146 146 146 148 151 152 154 154 155 156 157 159 159 163 164 166 166

CONTENTS

5 Carbohydrates in Peptide and Protein Design Knud J. Jensen and Jesper Brask 5.1 Introduction 5.2 Configurational and Conformational Properties of Carbohydrates 5.3 Carbohydrates in Peptidomimetics 5.4 Glycopeptides 5.5 Carbohydrates as Scaffolds in the Design of Nonpeptide Peptidomimetics 5.6 Sugar Amino Acids 5.7 Cyclodextrin–Peptide Conjugates 5.8 Carboproteins: Protein Models on Carbohydrate Templates 5.9 Conclusion References 6 De Novo Design of Proteins Knud J. Jensen 6.1 Introduction 6.2 Secondary Structure Elements 6.2.1 The a-helix 6.2.2 The b-sheet 6.2.3 Loops, Turns and Templates 6.3 Assembling a Specified Tertiary Structure from Secondary Structural Elements 6.3.1 Computational Methods 6.3.2 Coiled Coils 6.3.3 a-helical Bundles 6.3.4 Fluorous Interactions 6.3.5 Additional Topics 6.4 Proteins on Templates 6.5 Foldamers 6.6 Biopharmaceutical Applications of De Novo Design 6.6.1 a-helical Structures in Biopharmaceutical Applications 6.6.2 Foldamers in Biopharmaceutical Applications References

vii

177

177 178 181 183 185 187 193 198 199 200 207

207 208 208 214 214 215 215 216 220 225 228 229 234 236 236 238 238

viii

CONTENTS

7 Design of Insulin Variants for Improved Treatment of Diabetes Thomas Hoeg-Jensen 7.1 Introduction 7.2 Diabetes Management and the Need for Insulin Engineering 7.3 Insulin Structure 7.4 Prolonged-acting Insulin Solids 7.5 Prolonged-acting Insulin Solutions 7.6 Fast-acting Insulins 7.7 Glucose-sensitive Insulin Preparations 7.8 Alternative Insulin Delivery 7.9 Insulin Mimetics 7.10 Pushing the Limits of Insulin Engineering 7.11 Conclusion References Index

249

249 251 256 258 259 265 267 271 272 273 274 275 287

List of Contributors

Jesper Brask, Novozymes A/S, 6Bs.98, Krogshøjvej 36, DK-2880 Bagsværd, Denmark Oliver Demmer, Institute for Advanced Study at the Department of Chemistry, Technische Universita¨t Mu¨nchen, Lichtenbergstraße 4, D-85747 Garching, Germany Andreas O. Frank, Institute for Advanced Study at the Department of Chemistry, Technische Universita¨t Mu¨nchen, Lichtenbergstraße 4, D-85747 Garching, Germany Thomas Hoeg-Jensen, Novo Nordisk A/S, DK-2760 Maaloev, Denmark Knud J. Jensen, Faculty of Life Sciences, University of Copenhagen, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Copenhagen, Denmark Horst Kessler, Institute for Advanced Study at the Department of Chemistry, Technische Universita¨t Mu¨nchen, Lichtenbergstraße 4, D-85747 Garching, Germany Veronique Maes, Organic Chemistry Department, Vrije Universiteit Brussel, Pleinlaan 2, B-1050 Brussels, Belgium Garland R. Marshall, Center for Computational Biology, Department of Biochemistry and Molecular Biophysics, Washington University, 700 South Euclid Ave., St. Louis, MO 63130, USA

x

LIST OF CONTRIBUTORS

Gregory V. Nikiforovich, MolLife Design LLC, 751 Aramis Drive, St. Louis, MO 63141, USA Dirk Tourwe´, Organic Chemistry Department, Vrije Universiteit Brussel, Pleinlaan 2, B-1050 Brussels, Belgium

Preface

Ever since the discovery of the therapeutic value of insulin, at the beginning of the twentieth century, peptides have been successfully applied as drugs. In the fifties and sixties, with the advent of new methodologies for isolation, identification and total synthesis, the range of peptide hormones was further expanded and additional peptide drugs were launched. Since then, a generally upward trend has been seen, despite occasional statements to the contrary. Improvements in techniques have allowed the identification of many naturally occurring peptides, which have provided the starting point for the design of peptide drug candidates. In addition, de novo design has emerged as a new approach for the invention of peptide drug candidates. Designed peptides and small proteins have become ubiquitous tools for biochemical and biophysical studies. The latter studies have had a significant impact on the design of peptides as potential drug candidates, and are thus to some extent covered in the present volume. This book comprehensively presents central topics in the design of peptides and proteins, especially those with the goal of biopharmaceutical applications. It starts with an outline of computational methods, then moves on to cyclic peptides, which are often important in the development of peptide drug candidates; it provides an overview of peptidomimetics, carbohydrates in the design of peptides and proteins, de novo design of proteins, and finally, as a key example, the design of new insulin variants. This book is aimed at peptide scientists in academia and in industry, as well as at graduate students entering the field. I wish to express my sincere gratitude to Gregory Nikiforovich, Garland Marshall, Horst Kessler, Oliver Demmer, Andreas O. Frank, Dirk Tourwe´, Veronique Maes, Thomas Hoeg-Jensen and my

xii

PREFACE

former PhD student Jesper Brask for their excellent contributions. I am pleased to acknowledge Paul Deards and Richard Davies from Wiley for their support and guidance during the process. It is a privilege to be able to acknowledge Laura Quartara for her advice throughout this project.

180

beta alpha R

pass ψ

0

alpha R

–180 –180

0 φ

180 –180

0 φ

180

2(b)

2(a)

Plate 1 (a) Sampled conformational distribution of Ac-Ala-OMe obtained with the QM/MM approach. Lines show 99.8%, 99.5%, 98%, 95% and 90% (from purple to pink) levels of the experimental distribution of the (f, c) points for alanine residues (adapted from [1], Figure 6). (b) Ramachandran map of free energy for Ac-Ala-OMe in water, obtained with the AMOEBA polarizable force field. Lines show energy levels of 3.2, 2.8, 2.4, 2.0, 1.6, 1.2, 0.8 and 0.4 kcal/mol, from red to dashed orange (map courtesy of Prof. Jay Ponder) (see Figure 2.2)

K282 D262 K38

Y255

Arg4 Nal5

E288 Y255 Nal5

D-Tyr2

V197

E288 Arg4

H113 Y116 Arg3

Y116 H113

D-Tyr2

K110

A

B

Plate 2 Diagram of the two possible complexes of CXCR4 with FC131, A and B. FC131 is coloured green; FC131 backbone is shown as a pentagon. The TM helices are represented by purple ribbons, and only the side chains of CXCR4 contact residues mentioned in the text are displayed (see Figure 2.6)

Peptide and Protein Design for Biopharmaceutical Applications © 2009 John Wiley & Sons, Ltd. ISBN: 978-0-470-31961-1

Edited by Knud J. Jensen

Plate 3 Conformation of Cilengitide after a 100 ps restrained MD simulation [138] (see Figure 4.14)

Plate 4 Left: derivatization of polyphemusin II into a cyclic pentapeptide with excellent antagonistic activity on CXCR4. Optimization of the residues marked in red led from the initial peptide to its shortened analogue (middle). The amino acids marked in blue and an additional glycine were subjected to the spatial screening. Right: cartoon of the backbone structure with disulfide bonds of the polyphemusin II analogue tachyplesin I, showing the regions of interest in colours corresponding to the scheme on the left (see Figure 4.16)

B1

B1 T

R

c(phenol) > mM

Plate 5 Insulin T- and R-folds. The monomer structures were extracted from crystal structures 4IN and 2CTI. Insulin A-chain red and B-chain blue (see Figure 7.5)

Plate 6 The 0377 3D structure (1UZ9) showing hexamer–hexamer interactions. The zoomed box shows the cross-hexamer binding of the B29Ne-lithocholyl residue in a pocket in the opposing hexamer, lined by among others ProB28 and TyrB26 (H-bond in green). A phenyl–phenyl stacking interaction is clear between B1s from opposing hexamers. The residual lithocholyl residues that seemingly protrude into open space are in fact bound to neighbouring hexamers, which were removed for clarity (see Figure 7.9)

Plate 7 3D structure (1ZEG) of the dimer of AspB28 human insulin, illustrating the engineered destabilization of the dimer interface by charge repulsion between the engineered AspB28 (represented with spheres) with the native residues B21Glu and A4Glu. Natively, B28 is proline. Copyright 2005 (see Figure 7.10)

1 Introduction Knud J. Jensen

The aim of this book is to provide a comprehensive introduction to the concepts and methods behind the design of peptides and small proteins. The individual chapters are written by experts in each field. We have striven to coordinate the chapters to create coherence in the book. Inevitably, there is some constructive overlap between the topics in the chapters, and the chapters refer to one another. Chapter 2, ‘Computational Approaches in Peptide and Protein Design: An Overview’, by Gregory V. Nikiforovich and Garland R. Marshall, provides a comprehensive overview of computational approaches to the modelling of peptides and proteins. This chapter surveys computational methods and principles as well as some of the available software. The authors illustrate their points with specific examples, such as the design of cyclopentapeptides as inhibitors of CXCR4. Here they describe the conformational study of both the cyclopentapeptides, as a 3D pharmacophore model for FC131, and of the G-protein-coupled receptor CXCR4, where they build on a 3D model of the transmembrane region of CXCR4. They then discuss docking of the peptide FC131 to CXCR4. Chapter 3, ‘Aspects of Peptidomimetics’, by Veronique Maes and Dirk Tourwe´, provides an overview of a hierarchical approach to peptidomimetic design. This includes the role of cyclic peptides in the development of peptidomomimetics, referring to Chapter 4. Maes and Tourwe´ then Peptide and Protein Design for Biopharmaceutical Applications © 2009 John Wiley & Sons, Ltd. ISBN: 978-0-470-31961-1

Edited by Knud J. Jensen

2

INTRODUCTION

describe the concept of retroinverso structures and deliver an extensive overview of backbone modifications, before discussing side-chain constraints. They describe peptoids and secondary structure mimetics – a topic that is taken up again in Chapter 6 – as well as topomimetics. An important aspect of Chapter 3 is the examples given of modifications of peptide hormones, especially of somatostatin, as peptidomimetics. The chapter also covers a range of protease inhibitors. Chapter 4, ‘Design of Cyclic Peptides’, by Oliver Demmer, Andreas O. Frank and Horst Kessler, provides a comprehensive overview of its topic. It starts with naturally-occurring cyclic peptides (cyclosporin A, for example) and moves on to different ways of cyclizing peptides. Then some backbone modifications are discussed, a theme covered in Chapter 3, as well as other modifications of cyclic peptides. A central part of the chapter is a description of the conformation and dynamics of cyclic peptides, especially the reduction in conformational space. The authors describe turn structures in cyclic peptides and concepts in the rational design of cyclic peptides, leading to the outline of a general strategy for finding active hits. The text exemplifies this with the development of the peptide drug candidate Cilengitide as an integrin inhibitor and CXCR4 antagonist. Chapter 5, ‘Carbohydrates in Peptide and Protein Design’, by Jesper Brask and the editor, describes how carbohydrates are used to introduce new structural and conformational features to peptides and proteins. The topics in this chapter include sugar amino acids, cyclodextrins and carbohydrates as templates in the design of peptides and proteins. Chapter 6, ‘De Novo Design of Proteins’, by the editor, gives an overview of concepts in the design of proteins from general principles, rather than through a redesign of natural structures. The focus is on structural aspects, especially rules for the design of secondary structural elements such as a-helical peptides, and the assembly of these into tertiary structures. Some de novo turn motifs, used to connect the secondary structural elements, are also included. The chapter features an introduction to foldamers, especially b- and g-peptides. It ends with examples of biopharmaceutical applications of de novo design. Chapter 7, ‘Design of Insulin Variants for Improved Treatment of Diabetes’, by Thomas Hoeg-Jensen, provides a comprehensive overview of the classical therapeutic peptide hormone insulin. The focus is on insulin as a modern biopharmaceutical drug and the development of new insulin variants with modulated therapeutic profiles, e.g. prolonged-acting vs. fast-acting insulins, either by modifications in the

INTRODUCTION

3

51 AA structure or by appending moieties. Novel glucose-sensitive insulins and insulin mimetics are also covered. As mentioned above, there is some constructive overlap between chapters. We have striven to make the index a powerful tool in accessing topics across chapters.

2 Computational Approaches in Peptide and Protein Design: An Overview Gregory V. Nikiforovich and Garland R. Marshall

2.1

INTRODUCTION

Over the last decade, new examples of applications of computational methods to the design of peptides and proteins appeared in the literature literally every week, if not every day. Among the factors that contributed to this growth of computational design studies are the rapidly evolving capacity of computer systems, the availability of software packages for molecular modelling – both commercial and freeware – and ease of access to multiple databases, such as the Cambridge Structural Database (CSD), the Protein Data Bank (PDB) and Swiss-Prot/TrEMBL. Even more important, perhaps, is that nowadays there is almost universal agreement that computational approaches, along with experimental methods, are indispensable components of the general pipeline of drug design. Fifteen years ago, doubts about the potential practicality of computational methods were still widespread among the drug design community (see our earlier reviews on the subject [3,4]). Today, computational approaches cover the wide field ranging from suggesting plausible 3D structures of short oligopeptides in solution, Peptide and Protein Design for Biopharmaceutical Applications © 2009 John Wiley & Sons, Ltd. ISBN: 978-0-470-31961-1

Edited by Knud J. Jensen

6

COMPUTATIONAL APPROACHES: AN OVERVIEW

through determining the peptide sequences most suitable for performing certain biological functions, to de novo predictions of large proteins interacting with one another. In fact, many of the most recent examples are described and reviewed in later chapters of this book. Therefore, we do not feel a real need to review the current state of computational approaches in detail. Instead, this chapter focuses only on the most general problems of computational design of proteins and especially peptides, as well as on the basic techniques required for the practical implementation of design methods. Our main goal is to introduce the basic elements of the computational approaches to those interested in peptide and protein design, as well as to share some of our thoughts and reflections on the current state of the field. In the first part of the chapter, we discuss computational tools and procedures connected with conformational flexibility of peptides and proteins; in our opinion, these problems can be satisfactorily resolved only by applying computational approaches. The second part contains a recent example of the application of these tools to the specific task of studying possible complexes between the CXCR4 G-protein-coupled receptor (GPCR) and its cyclopentapeptide inhibitors [6,7].

2.2 2.2.1

BASICS AND TOOLS The Importance of Computational Approaches

From a very general point of view, most of the problems of modern peptide and protein drug design fall into two main areas: structure-based and target-based design. In the first case, one starts from a parent peptide (‘ligand’) with no details of the structural information on its specific target (‘receptor’). In the second case, structural information on the receptor, at least on the receptor site that binds the ligand, is available at varying resolution. In both cases, the goal is to suggest compounds that would exhibit certain biological qualities (affinity, activity, etc.) as well as or better than the parent ligand. In structure-based design, the most essential requirement is to determine the 3D arrangement of the functional groups of ligand comprising the so-called ‘3D pharmacophore’, which is responsible for ensuring correct interaction between the ligand and the receptor. In target-based design, one also aims to determine the correct binding mode of each ligand within the binding site of the common receptor. However, in the frame of structure-based design, the available experimental methods of structural determination often fail to determine possible 3D pharmacophores characteristic for the interaction of the peptide ligand

BASICS AND TOOLS

7

with its specific receptor. The apparent reason is the inherent conformational flexibility of peptides. Most peptides exist under physiological conditions as a mixture of more or less well-defined, interconverting conformers. The interconversion rate is such that, for instance, NMR spectroscopy with characteristic resolution times of 105–103 seconds does not distinguish separate conformers of linear peptides in the kilodalton range in solution (with the exception of cis/trans peptide bond isomers). The same is true for CD, IR and ESR spectroscopy. In the absence of one highly predominant conformer, the 3D peptide structure deduced from physicochemical measurements (e.g. from NMR parameters such as NOEs, vicinal coupling constants, etc.) reflects the average over the ensemble of conformers present in solution and, in this sense, could not be related to any of the ‘real’ peptide conformers at all. On the other hand, the conformation of peptide ligands corresponding to the 3D pharmacophore, i.e. to the ligand conformation in the complex with receptor, may not necessarily be the one with the highest statistical weight in solution, since some other conformers may acquire the highest statistical weight in the peptide–receptor complex, being compensated by much more favourable interaction in the complex with the receptor. At the same time, X-ray crystallography produces only individual ‘snapshots’ of peptides, each representing a single 3D structure stabilized by the crystalline lattice from among the set of possible conformers existing in solution. For the highly flexible enkephalin molecule, for instance, X-ray crystallography obtained snapshots of the four drastically different 3D structures ranging from fully extended to various types of b-reversals (see review [8]). On the other hand, computational methods, being applied to various analogues of the same peptide that differ by values of affinity (or activity) toward a specific receptor, may model all 3D structures feasible for the parent peptide and its analogues from the energetic and/or sterical point of view. Then one may compare sets of those structures to one another and select those among the biologically active analogues in which the important functional groups are arranged in space similarly. These structures may be regarded as reasonable candidates for 3D pharmacophores, which in turn may be stabilized by introducing constraints through chemical synthesis. The structure-based design employing this approach has been successful in developing novel cyclic analogues of linear peptides, many of which are biologically active (such as analogues of opioid peptides, angiotensin, a-melanotropin, etc.; see earlier review [3]). Historically, target-based design came after structure-based design, since detailed information on the receptor molecules only became readily available in recent decades. This progress was made largely due to rapid

8

COMPUTATIONAL APPROACHES: AN OVERVIEW

development of technology for co-crystallization of ligand–receptor complexes (especially enzymes and their substrates/inhibitors), as well as advances in X-ray crystallography of proteins. Seemingly, experimental determination of ligand–receptor complexes by X-ray crystallography abolishes the need for computational approaches. Indeed, detailed information on both the recognition motif (3D pharmacophore) and the binding mode of the ligand within the receptor is readily available from the X-ray structure of the complex. In reality, however, there are several important limitations that still require emphasis on computational approaches in target-based design. First, many biologically active peptides (and over 30% of drugs in clinical use [9]) act through interactions with GPCRs, which are integral membrane proteins that include seven transmembrane helical stretches (TM helices) connected by loops that form the intracellular (IC) and extracellular (EC) domains, together with the fragments containing the N- and C-termini. Being membrane proteins, GPCRs are extremely difficult to express and to extract from the membrane in quantity, and have resisted chemical synthesis. Accordingly, X-ray structures are known presently only for four GPCRs, namely the photoreceptor rhodopsin, the b2- and b1-adrenergic receptors and the A2A adenosine receptor [10,11, 114, 115]. Therefore, the only way to address ligand–receptor interactions involving other GPCRs (the largest human gene family) for target-based design is to apply computational approaches to model (either by homology or by de novo approaches) the 3D structure of GPCR. Second, while many ligand-receiving sites in receptors feature more or less well-defined 3D ‘pockets’ inside the protein globule (as, for example, in enzymes), other recognition sites are formed by flexible loops protruding away from the bulk of the protein (as, for example, in antibodies or GPCRs). In the latter case, one faces the same problem as in determining the 3D structure of a flexible peptide: namely, even if the X-ray structure of the loops in the receptor–ligand complex is resolved, the X-ray snapshot may capture only a single specific conformation out of several that may be more characteristic for the given complex and more representative of a functional complex. For instance, the conformations of the IC loops connecting TM helices in the five X-ray structures of rhodopsin published so far drastically differ from one another. Again, determining the set of plausible conformations of the loops in question requires use of computational modelling. Third, depending on the shape and rigidity of the receiving site of the receptor, the binding modes of the ligand also may be nonunique. That may be especially important for small ligands with medium affinity toward the

BASICS AND TOOLS

9

receptor (e.g. in micromolar range). In this case, again, the X-ray snapshot of the receptor cocrystallized with the ligand may not represent the optimal binding mode, nor the ensemble of binding conformations, for the ligand. In these cases, the binding modes have to be refined by computational sampling of various spatial positions of the ligand within the receiving site of the receptor to sample orientations in which the ligand binds the receptor more tightly. The roles of entropy and enthalpy of complex formation are not independent and simple comparisons of affinity (DG) without dissection into its components, DH and DS, can be misleading [12]. These considerations illustrate the point that in peptide and protein design, whether structure-based or target-based, there are certain problems for which we require computational modelling. At the same time, one should not overestimate the precision of current computational results. For typical applications, computational approaches generate plausible suggestions regarding structural aspects of the functional recognition of peptides and proteins. In all cases, these suggestions have to be independently validated either by direct experimental structural measurements or by confirmation of predictions through biological experiments.

2.2.2

Tools and Procedures: Force Fields and Sampling

General protocols of any computational approach to peptide and protein design regularly include two key aspects: the generation of possible molecular conformations and relative orientations of interacting molecules (sampling) and the evaluation of the plausibility of the generated conformations or orientations in terms of their relative energies (scoring). The more thorough the sampling protocol and the more accurate the scoring function, the more reliable the predictions. Ultimately, the best results may be obtained when all possible states of a system in question (conformations and relative orientations) are sampled and the energy for each of the states is calculated employing high-level quantum calculations. However, this best-case scenario is seldom applicable in peptide and protein design, simply because of the system size, which overwhelms available computer resources. Even with current rapid expansion of computer capacity, it is unrealistic to expect adequate quantum chemical calculations for, say, a linear octapeptide in water – the system featuring thousands of possible conformations of the peptide and millions of configurations of the solvent – within the next decade.

10

COMPUTATIONAL APPROACHES: AN OVERVIEW

2.2.2.1 Force fields Atom–atom force fields currently in use: validation and applicability In the so-called Born–Oppenheimer approximation, interactions within molecular systems are limited to atom–atom interactions that allow the estimation of energies for various states of peptides and proteins, with an accuracy sufficient for many practical applications. In this approximation (molecular mechanics, MM), peptide molecules are considered systems of points (atoms) in space, which interact with each other by different types of forces. The forces can be divided into two main classes, namely those between atoms that are bonded and those that are nonbonded in the valence structure. Usually, the forces between nonbonded atoms include at least two terms: van der Waals and electrostatic forces. A special term describing hydrogen bonding between corresponding atoms is often included, as well as an additional dihedral angle (‘torsional’) term. The forces between bonded atoms include bond stretching, valence angle bending and improper dihedral angle forces. Summarily, molecular energy calculated with a typical atom–atom force field in MM may be expressed as follows: VðrÞ ¼

X bonds

X k ð  0 Þ 2 þ k ½cosðn þ Þ þ 1 torsions angles " # X qi qj Aij Cij þ þ 12  6 rij r ij r ij nonbond pairs

kb ðb  b0 Þ 2 þ

X

ð2:1Þ

Generally, all forces depend on the distance between interacting atoms, and on the parameters selected for each term. There is no one single atom–atom force field universally adapted as standard for calculating energies in peptides and proteins. Several force fields are currently used for this purpose, differing mostly in the sets of parameters. The parameters are usually selected to fit the experimental data on crystal packing of amides and amino acids, such as in AMBER (Assisted Model Building with Energy Refinement) [13,14] and ECEPP (Empirical Conformational Energy Program for Peptides) [15–17], or on properties of organic liquids, such as in OPLS (Optimized Potentials for Liquid Simulations) [18,19] and GROMOS (GROningen MOlecular Simulation) [20], as well as to fit the results of quantum chemistry calculations, such as in AMBER, OPLS and CHARMM (Chemistry at HARvard Molecular Mechanics) [21]. Historical development of atom–atom force fields was strongly impacted by the availability of computer resources. The oldest

BASICS AND TOOLS

11

force fields, such as ECEPP, employ rigid valence geometry and therefore do not include the first two terms of the potential expression, which results in significant reduction of the computer time needed for energy calculations. All of the other force fields include flexibility of valence geometry. Options are to consider all hydrogens separately (slower calculations) or consider aliphatic hydrogens united with their bonded carbon atoms (united-atom assumption; speedier calculations). The force fields with flexible valence geometry are utilized also for other classes of biochemical compounds, such as nucleic acids, carbohydrates, etc. Specifically, force fields for peptides and proteins have been reviewed many times, emphasizing different aspects of their applications; the reader is referred to the recent excellent reviews by Ponder and Case [22] and Mackerell [23]. It is difficult to determine which force field is preferable for conformational calculations involving peptides and proteins. On the one hand, obviously, the force fields with flexible valence geometry and those calibrated to fit the results of quantum chemistry calculations are more likely to yield an accurate value of energy for a given state of a peptide system including solvent. But on the other hand, computational approaches are especially valuable for problems involving conformational flexibility of peptide chains and orientations of ligands within the binding site of the receptor. In both cases, estimations of energy for a large number of possible states of the system are required in order to select the most plausible states as fast as possible with as few as possible false positives and false negatives, providing some justification for the more computationally efficient approximations. In this regard, a convenient test utilizes reconstruction of the Ramachandran map for the simplest element of the peptide chain, the acetyl-N-Methyl-L-alanine (Ac-Ala-OMe), by various force fields. The Ramachandran map is a function of the two torsional angles, f(rotation around the bond NH-Ca) and c(Ca-CO), adjacent to the a-carbon that maps the potential surface of peptide backbone conformations. The Ramachandran map can be roughly divided into four quadrants: the upperleft, corresponding to (f < 0, c > 0); the lower-left (f < 0, c < 0); the upper-right (f > 0, c > 0); and the lower-right (f > 0, c < 0). The upper-left quadrant contains the (f, c) points corresponding to an extended structure, such as a b-strand (f  140, c  140), and the lower-left and upper-right quadrants contain points corresponding to the right- (f  60, c  60) and left-handed (f  60, c  60) a-helices, respectively. The upper-left quadrant also contains the (f, c) points (75, 140) and (80, 80), corresponding to conformations PII (polyproline II) and C7eq (the

12

COMPUTATIONAL APPROACHES: AN OVERVIEW

inverted g-turn). The lower-right quadrant contains the (80, 80) point that corresponds to the C7ax conformation (the g-turn). At this rough ‘quadrant’ approximation, all quadrants except the lower-right are considered sterically allowed for the L-amino acid residues. All four quadrants are allowed for Gly residues but only two of them, namely those at the left side of the plot, are allowed for L-Pro residues. The Ramachandran maps for L- and D-amino acid residues are symmetrical with respect to rotation by 180 around an orthogonal axis at the centre of the map. Earlier calculations of the Ramachandran map for Ac-Ala-OMe using CHARMM and AMBER, but not those using ECEPP, showed that conformation C7ax possessed relative energies close to those of C7eq, the conformation with the lowest energy [24]. Moreover, the CHARMM and AMBER maps showed fairly large regions of energetically-allowed conformations in the lower-right quadrant, which was contradictive to data on the X-ray structures of amino acid residues in proteins available at the time (1989). It was argued that the calculations were performed without proper account for solvent and, in fact, modelled the Ac-Ala-OMe in the gas phase, for which no experimental data exists. On the other hand, high-level quantum calculations for Ac-Ala-OMe also found that conformation C7ax possesses energy close to C7eq and lower than conformations corresponding to the right- or left-handed a-helical structures (e.g. [25]). Recently, the Ramachandran map for Ac-Ala-OMe was extensively sampled by molecular dynamics simulations employing several force fields with flexible valence geometry [1]. Simulations included interactions with water molecules described explicitly. Figure 2.1(a)(d) depicts the Ramachandran maps obtained with the AMBER, CHARM22, GROMOS and OPLS-AA force fields, respectively. Figure 2.1(e) depicts the distribution of the (f, c) positions for each of 97 368 residues derived from 500 high-resolution X-ray structures of proteins [2]; to avoid bias, only the data for residues not involved in regular secondary structures such as b-strands or a-helices were included in the map in Figure 2.1(e). Assuming that the distribution of the X-ray data on residues in proteins is a good approximation of the general distribution of plausible (f, c) values for peptides and proteins, one can utilize this distribution to evaluate the relative validity of various force fields for computational studies of peptides and proteins. Experimental data presented in Figure 2.1(e) suggest that most plausible conformations of the L-amino acid residues in proteins are concentrated into three regions. The largest and most populated region is located in the upper-left quadrant of the Ramachandran map, encompassing conformations corresponding to the b-strand and PII (the ‘beta’ region).

BASICS AND TOOLS

13

Figure 2.1 (a)(d) Sampled conformational distributions of Ac-Ala-Me obtained with the AMBER, CHARM22, GROMOS and OPLS-AA force fields, respectively (adapted from [1], Figures 1, 3, 4 and 5, with permission from John Wiley & Sons, Inc). (e) The (f, c) data points for the amino acid residues derived from the X-ray structures of 500 proteins (adapted from [2], Figure 7, with permission from John Wiley & Sons, Inc). (f) The Ramachandran map calculated with the ECEPP force field showing equipotential levels of relative energy (adapted from [5], Figure 9, with permission from Wiley-VCH)

14

COMPUTATIONAL APPROACHES: AN OVERVIEW

It also includes conformation C7eq, located into the ‘pass’ zone (that with c values around 60 – 30) extended toward the second main region of plausible conformations centred around (90, 0). This region spreads through both left quadrants and contains conformations close to the right-handed a-helix (the ‘alpha R’ region). The third region is the smallest one, located in the upper-right quadrant, and contains conformations close to the left-handed a-helix (the ‘alpha L’ region). Only a few conformations were experimentally found in the lower-right quadrant, which includes conformation C7ax. Interestingly, experimental distributions in the regions corresponding to the right- and left-handed a-helices show a somewhat diagonal shape. Sampling of the Ramachandran map with the AMBER and CHARMM22 force fields (Figure 2.1(a,b)) resulted in two main regions located in the left half of the map. For AMBER, population of the alpha R region was significantly higher than that of the beta region, and only a few conformations were found in the alpha L region (after performing additional molecular dynamics simulations; see [1] for details). These results are not consistent with the experimental data in Figure 2.1(e); one may expect that energy estimations involving the AMBER force field will lead to large numbers of false positives in the alpha R region when sampling conformational possibilities of peptide systems. The results obtained with CHARMM22 (Figure 2.1(b)) showed a high population of the beta region; however, the second equally populated region is shifted down to the lower-left quadrant and possesses only a small overlap with the experimentally determined alpha R region. Much more consistent with experimental data were the Ramachandran maps obtained with the GROMOS and OPLS-AA (‘AA’ stands for ‘all atoms’, meaning ‘including all hydrogens’) force fields (Figure 2.1(c,d)). In both cases, the sampled beta and alpha R regions were located close to the experimentally determined ones, and populations of the beta regions clearly exceeded populations of the alpha R regions. Sampling also found conformations belonging to the pass zone between two regions, as well as conformations in the upper-right quadrant. These latter conformations were more frequent in the GROMOS map, but more consistent with the experimentally observed alpha L region in the OPLS-AA map. One may expect that energy estimations based on the GROMOS or OPLS-AA force fields will lead to more adequate sampling of peptide systems with less false positives or negatives. In fact, both force fields are currently more frequently used in conformational calculations of peptides and proteins than either AMBER or CHARMM.

BASICS AND TOOLS

15

At the same time, the Ramachandran map in Figure 2.1(f), calculated with the much older ECEPP force field using rigid valence geometry without accounting for solvent, whether implicit or explicit, shows at least as good consistency with experimental data in Figure 2.1(e) as the GROMOS and OPLS-AA maps. Indeed, the allowed areas of this map cover all three regions, determined experimentally with approximately the same relative populations. The zone region was also populated; in fact, conformation C7eq had the lowest relative energy. Even the diagonal shape of the experimental distribution was preserved in the alpha R region (and, though just slightly, in the alpha L region) of Figure 2.1(f). The main difference from the experimental data was in the extension of the alpha R region in the ECEPP map toward (f, c) values around (60 – 30, 150 – 30); these are virtually unpopulated in Figure 2.1(e). One reason for the good consistency of the ECEPP map with the experimental distribution was that the parameters of the ECEPP force field were calibrated specifically to reproduce the X-ray data on crystal packing of amino acids. This made the ECEPP force field limited in applications to molecules other than peptides and proteins, which precluded its acceptance by commercially available modelling packages, such as SYBYL, INSIGHT or MacroModel. Nevertheless, comparison of maps in Figure 2.1(e) and (f) clearly suggests that the ECEPP force field, though less sophisticated than GROMOS or OPLS, may be successfully used in the sampling of possible conformational states of peptide and protein systems. At the same time, practical applications of the ECEPP force field require significantly less computer resources, with the additional advantage that it is able to perform sampling in dihedral angle space, which is much less complex than Cartesian coordinate space (see Section 2.2.2.3). Further developments of molecular force fields The MM force fields outlined above are routinely used in computational design of peptides and proteins. However, as was concluded in a recent review, ‘Currently, force fields are not perfect (even the all-atom ones). It is possible to obtain different results with different force fields. Therefore, improving force fields (both the all-atom and reduced ones, and the water potential) is a priority.’ [26]. Further development of force fields has occurred recently, offering several possible improvements. One approach is to combine quantum mechanics with molecular mechanics (QM/MM), whereby selected parts of the molecular system under study, for instance the active sites of enzymes, are treated by QM approximations, and the larger surrounding molecular areas are

16

COMPUTATIONAL APPROACHES: AN OVERVIEW

represented with MM [27,28]. This approach was applied to Ac-AlaOMe and showed excellent consistency with the experimental data in Figure 2.1(e) [1]. Specifically, the molecule of Ac-Ala-OMe was treated with the fast approximate QM method SCCDFTB [29], whereas interactions between Ac-Ala-OMe and the explicit water molecules were calculated either with SCCDFTB (electrostatic interactions) or with force fields with flexible valence geometry (such as AMBER). Interactions between the water molecules were calculated with the same force fields. Figure 2.2(a) presents the results of sampling overlapped with contours corresponding (from purple to pink) to 99.8%, 99.5%, 98%, 95% and 90% of the levels of the experimental distribution of the (f, c) points for the alanine residues from the experimental distribution in Figure 2.1(e). Distribution of sampled conformations in Figure 2.2(a) is remarkably close to that in Figure 2.1(e). All main regions of the experimental distributions have comparable relative populations, including the pass zone. Also, the diagonal shapes of distributions in the alpha R and alpha L regions are reproduced, whereas only a few conformations appear in the region of the (f, c) values around (60 – 30, 150 – 30). One slight discrepancy is the relatively low population of the narrow zone around c 100. 180

beta alpha R

pass ψ

0

alpha R

–180 –180 2(a)

0 φ

180 –180

0 φ

180

2(b)

Figure 2.2 (a) Sampled conformational distribution of Ac-Ala-OMe obtained with the QM/MM approach. Lines show 99.8%, 99.5%, 98%, 95% and 90% (from purple to pink) levels of the experimental distribution of the (f, c) points for alanine residues (adapted from [1], Figure 6). (b) Ramachandran map of free energy for Ac-Ala-OMe in water, obtained with the AMOEBA polarizable force field. Lines show energy levels of 3.2, 2.8, 2.4, 2.0, 1.6, 1.2, 0.8 and 0.4 kcal/mol, from red to dashed orange (map courtesy of Prof. Jay Ponder). Adapted with permission from John Wiley & Sons, Inc (see colour Plate 1)

BASICS AND TOOLS

17

Despite this comparison, which distinguishes the QM/MM force field as preferable for energy estimations in peptides and proteins, practical applications of this approach to systems larger than the alanine dipeptide are rather rare due to its computational demands, though several studies have employed it [28]. On the other hand, the QM/MM approach has already been incorporated into some molecular modelling packages, such as a recent version of AMBER [30]. Another significant improvement to force fields emerged from a longstanding problem to do with the correct accounting for electrostatic interactions. In the Born–Oppenheimer approximation, the short-ranged repulsive and attractive nonbonded forces acting between atoms can be assigned to atomic centres to provide an estimation of the nonbonded energy. But in reality electrostatic potentials are delocalized over the entire molecular volume, and representing electrostatic energy simply by sum over interactions between point changes localized on the same atomic centres – as happens with all the force fields discussed above – is the source of errors in electrostatic energy. Also, electrostatic potentials for a given molecule may vary from one conformation to another, since different spatial arrangements of the charges on the atomic centres induce variations of those changes through dipole–dipole (and, more generally, multipole–multipole) mechanisms. Additionally, electrostatic potentials around polar molecules also depend on the influence of a solvent, especially a polar solvent, such as water. In other words, delocalized electrostatic potentials depend on the conformation of the molecule in its particular electrostatic environment, which in turn depends on electrostatic potentials. Accordingly, the main problem in dealing with this obstacle is to develop procedures that efficiently converge to a stable distribution of delocalized components of electrostatic potentials for each conformation. Several types of such procedures were suggested, leading to the new generation of the so-called ‘polarizable’ force fields. Delocalization of electrostatic potentials was achieved by employing various schemes, including the creation of additional artificial atomic centres (e.g. ‘lone pairs’) and the decomposition of the charges into sophisticated systems of multipoles. The effect of additional polarization from water was also included in some procedures, as were various models for water–water interactions (see [22]). The polarizable force fields aiming at possible applications for peptides and proteins were suggested by several groups (e.g. [22,31]) and showed improvements in energy estimations of local minima on the Ramachandran map of Ac-Ala-OMe. The typical benchmark for validation of polarizable force fields was the ability to

18

COMPUTATIONAL APPROACHES: AN OVERVIEW

reproduce the results of high-level quantum calculations as closely as possible. Direct comparison with experimental data was recently performed by simulating the Ramachandran map of Ac-Ala-OMe in water with the AMBER polarizable force field [32] and by calculating the same freeenergy map in water with the polarizable AMOEBA (Atomic Multipole Optimized Energetics for Biomolecular Applications) force field (Ponder, personal communication). While the former study produced the Ramachandran map with heavy populated b-region and alpha R region, but not the pass or the alpha L regions (see [32], Figure 4), the AMOEBA map appears very consistent with the experimental distribution of the (f, c) points (see Figure 2.2(b)). Keeping in mind that calculation of the AMOEBA map required computation only eight times larger than that for the same calculation with the nonpolarizable AMBER force field (Ponder, personal communication), prospects for the application of the AMOEBA force field to peptides and proteins look encouraging. However, polarizable force fields still have to be validated for their ability to reproduce thermodynamics in a variety of examples to ensure their wide applicability to peptide and protein design. In summary, as comparison with the experimental data shows, developments of recent novel approaches to molecular force fields should lead to a much wider use of the QM/MM and polarizable force fields in peptide and protein design. Surprisingly, however, the comparison also shows that the relatively simple ECEPP force field, which was developed more than two decades ago, reproduced the distribution of experimental (f, c) points on a Ramachandran map with accuracy comparable to that of the QM/MM and polarizable force field approaches, and surpassing that of the more sophisticated GROMOS and OPLS force fields.

2.2.2.2 Scoring functions The other current trend in describing interactions in peptides and especially in proteins suggests not making the interactions more complex but, on the contrary, simplifying them. In such a system, each amino acid residue, and some functional atomic groups, is reduced to one of several interaction centres. (In fact, combining the aliphatic hydrogens together with the bonded carbon atom into a united centre such as CHn, which is a common feature of many atom–atom force fields, is already an example of such simplification.) Potential functions for interactions between the centres may be deduced by averaging atom–atom interactions between

BASICS AND TOOLS

19

the corresponding functional groups, or from statistics of close contacts between different types of amino acid residues in the known X-ray structures of proteins (the ‘knowledge-based potentials’), or by applying other models. By drastically reducing the number of interaction centres in the large molecule, one gains the opportunity to estimate the energy of the interactions for many possible conformational states in a reasonable amount of computer time. This may be especially valuable for complex problems such as ab initio protein folding. For instance, reducing the all-atom protein chain to the system of two centres of interaction per residue, one on each peptide group and one on each side chain, allowed folding of a 75-residue protein in 4 hours on a single processor [26], which was remarkably fast. However, this is paid for in terms of the accuracy of the estimation of energy, which effectively becomes a ‘scoring function’ that distinguishes between conformational states with high scores and low scores. If the highest score is significantly higher than others (as may be expected, for instance, for the native fold of a protein), a simplified system of interactions may produce clear predictions. Otherwise, one must expect predictions of an ensemble of folds with comparable score levels, the most plausible of which must then be selected by comparison with experiment. Various types of scoring function were developed specifically to find the optimal orientations of the ligand within the binding site of the receptor, the main goal of the virtual screening. Several recent reviews addressed the application of scoring functions in studies of ligand– receptor binding that involve peptides and proteins [33–36]. Usually, scoring functions are loosely divided into three main categories: force field-based, empirical and knowledge-based scoring functions. Force field-based scoring functions utilize the routine MM force fields such as AMBER, OPLS, ECEPP and so on to estimate the energy of atom–atom interactions between ligand and receptor. They may also include terms estimating the desolvation of ligand and receptor in order to increase the chances of finding the correct (native) orientation of the ligand from among many decoys [37]. These scoring functions yield the most precise estimations of potential and free energies of ligand–receptor binding, but obviously they are the most expensive from a computational point of view. Also, the rugged energy landscapes of molecular interactions require additional adjustments during sampling, as discussed below. Empirical scoring functions are derived primarily from experimental data on binding energies measured for a limited number of

20

COMPUTATIONAL APPROACHES: AN OVERVIEW

ligand-binding complexes (the training sets). Components of these functions may include terms roughly related to free energy of binding, systems of hydrogen bonding and hydrophobic/hydrophilic interactions between ligand and receptor, values of solvent-accessible atomic surfaces and so on. All terms are usually combined into one linear equation, and coefficients for each term are calibrated to fit the data on the training set. A typical empirical scoring function is the one pioneered by Bo¨hm more than a decade ago [38,39]. Empirical scoring functions are convenient for the virtual screening of a large number of ligand compounds that are structurally similar to the relatively small number of compounds in the training set, but dependence on the given training set remains a limitation. Knowledge-based scoring functions use structural data on protein– ligand complexes instead of the experimental data on binding energies. The data are analysed in detail to derive simple distancedependent potential functions for interactions between specific atomic groups of ligands and receptors by weighting the probability of observed experimental distances between the atomic groups as energies of interaction according to an inverse Boltzmann relationship. In this respect, the knowledge-based scoring functions are, in fact, statistical atom–atom potentials, which rely on the ‘training sets’ of the available high-resolution data on protein–ligand complexes. For instance, the prototypic knowledge-based scoring function, a potential of mean force initially suggested by analysing the 697 protein–ligand complexes available in 1999 [40], was recently updated with the data on 7152 protein–ligand complexes available in the PDB in 2005 [41]. The three types of scoring function are often combined in practical tasks for peptide and protein design, especially in virtual screening, into a united procedure of consensus screening, where ligand orientations with the high scores predicted by different scoring functions are pooled to compensate for errors due to biases in each single function (e.g. [34,42]). Many existing scoring functions are incorporated into available molecular modelling packages and are widely used in drug design [33,43,44]. However, the search for an optimal scoring function related to protein–ligand complexes is far from complete. Examples of the recently described scoring functions are new statistically-derived atom–atom potentials [45], knowledge-based potentials accounting for protein flexibility [46,47], a scoring function based on properties of residues at the ligand–receptor interface [48] and several others (e.g. [49–51]).

BASICS AND TOOLS

21

2.2.2.3 Sampling Sampling of peptide conformational space The most thermodynamically stable conformational state of a molecular system is, obviously, the one with the lowest free energy (the global minimum-energy conformer). However, energy minimizations performed using MM find only values of potential energy; to calculate values of free energy, one needs to know the entire multidimensional potential surface, or at least all local low-energy conformers, in order to estimate the partition function and calculate the entropic contribution to the free energy. In this regard, sampling of peptide conformational space is, in fact, mapping of the energy surface of the peptide in search of the local energy minima. Several strategies have been developed to map the potential energy surface and to find local minima. Stochastic methods such as Monte Carlo can be employed not only to find local minima, but also to estimate local fluctuations inside a given minimum. Molecular dynamics can also be used to explore the potential energy surface, often with some enhancements such as simulated annealing or replica exchange to help overcome energy barriers between minima, i.e. kinetic barriers related to conformational dynamics. Systematic, or grid, search samples conformations in a regular fashion in the space of parameters (usually by dihedral torsional angles) that are incremented. In a sense, the same approach is represented by various build-up procedures, whose aim is to explore the conformational states of the entire peptide molecule by systematically combining results of conformational samplings for fragments of this peptide. These different methods of conformational sampling are briefly discussed below. Examples of systematic sampling Systematic search consists of systematic generation of all possible conformations at the selected torsional grid, in order to determine the set of sterically allowed ones. Therefore, systematic search is the most exhaustive way of sampling the conformational space, but due to its combinatorial nature the number of conformations considered increases enormously with an increase of torsional space dimensionality. Also, since the energetics of the molecular system are very sensitive to interatomic distances, a conformation generated at, say, the 10 increment may be sterically disallowed, but very close to a minimum, so relaxation of the structure by allowing a torsional angle to vary by 1 or less may find the missing minimum. However, the computational cost for that systematic change would be prohibitive. Indeed, using a 10 grid for, say, a seven-rotatable bond problem, the entire number of conformations explored will be 736 (2.65  1030), whereas at

22

COMPUTATIONAL APPROACHES: AN OVERVIEW

a 1 grid it will be 7360 (1.72  10304) conformations. Even using rather sophisticated computational algorithms to minimize this problem [52], only a few short peptides, such as the tripeptide TRH and its analogues, have been investigated in this manner (see [3]). In many cases, it seems more practical to consider those grid points with relatively low energies as starting places for consequent minimization, allowing convergance to the nearest local minimum. Build-up procedures are somewhat similar to systematic search in considering all possible combinations of local energy minima found in the previous systematic search of smaller peptide fragments (e.g. combinations of all local minima found in the Ramachandran maps for the acetyl-Nmethylamides of amino acid residues; see above). To examine, for instance, an octapeptide, one can start by performing energy minimization for all possible combinations of the starting backbone conformations for each residue in the C-terminal pentapeptide 4–8. The number of starting lowenergy backbone conformations for a separate residue usually varies from 5 to 10 depending on the nature of the residue, so the number of possible combinations would be in the range of 55–105. The subsequent steps of the build-up procedure can involve the molecule ‘growing’ to the fragment 3–8, then to the fragment 2–8 and finally to the entire molecule. The backbone conformers selected for further consideration at each step would be those satisfying the criteria of DE ¼ E  Emin. The starting points for the new residues attached to the conformers selected at the previous steps would again be the backbone conformations for each separate residue. At each step, the spatial arrangement of the side chains might be optimized before energy minimization by a special algorithm, which includes several cycles of a stepwise grid search for energy minimum in the space of side-chain dihedral angles (see also the section ‘Sampling of Side-chain Rotamers’, below). A build-up procedure can equally well start with a central peptide fragment, or involve overlapping peptide fragments. Owing to the computational feasibility of this procedure, it has been applied for many peptides: opioid peptides, angiotensin, CCK-related peptides, a-melanotropins, GnRH agonists and antagonists, and others (see our earlier review, [3]). The main weak point of any build-up procedure is the necessity for a priori estimates of DE value at each step of the procedure. A too-high value of DE will result in an unmanageable number of conformers to be processed; on the other hand, lowering of this value might cause significant losses in the list of low-energy conformers. The DE value also depends on the force field used for energy calculations; for instance, for the ECEPP force field, a value of DE of 1 kcal/mol per residue has been recommended as practical [3].

BASICS AND TOOLS

23

Examples of stochastic sampling Molecular dynamics (MD) is a process for simulating the evolution of the movement of a molecule in time, where the motions of each atom are governed by Newton’s equations, including the atom positions, masses and velocities. As the position of one atom changes with respect to the others, the forces which it experiences also change, as do those of its neighbours. The forces on any particular atom are calculated using the appropriate force field. The velocities are introduced by the average kinetic energy of the system, i.e. by the temperature value. The time step chosen for calculations should be smaller than the period of fastest local motion, in order to ensure that each atom moves in sufficiently small increments to leave the positions of surrounding atoms unchanged. The typical time increment in molecular dynamics of peptides is on the order of 1015 seconds, which reflects the need to adequately represent atomic vibrations of similar timescale. For MD simulations of molecules in solvent, sufficient solvent molecules must be included to adequately represent all classes of solvent–solute interactions. This requires several hundred solvent molecules for even small solutes, so the longest MD trajectories simulated for peptides are usually of the order of not longer than 100 ns (e.g. [53]). Because of the short time steps of molecular dynamics, events requiring longer times such as diffusion are difficult to simulate at the molecular level of detail. In this case, Brownian dynamics are used, and the particles (consisting of many atoms) move under the Langevin equations, which govern diffusion, rather than Newton’s equations of motion. Electrostatic forces are derived from the relative positions of the charged particles in the simulation by the Poisson–Boltzmann equation, which describes dielectric behaviour in a nonhomogeneous system. Monte Carlo sampling is based on statistical mechanics and generates sufficient different configurations of a system by computer simulation to allow the desired structural, statistical and thermodynamic properties to be calculated as a weighted average of these properties over the set of configurations. Monte Carlo simulations are successfully performed by sampling only a limited set of the energetically feasible conformations, say 106 out of 10100 theoretical possibilities. One could sample all states, calculate the energy of each and then average their contributions according to the Boltzmann distribution of energies. Instead, Monte Carlo sampling focuses only on energetically feasible answers from among the ensemble of possibilities. The term ‘Monte Carlo’ comes from the random selection of the parameter (for example, coordinate or torsion angle) that determines the next configuration (conformation). The energy of the new state is

24

COMPUTATIONAL APPROACHES: AN OVERVIEW

compared with that of the old state. If it is the same or lower, the new configuration is kept and becomes the basis for calculation of the next configuration. If the new configuration has a higher energy than the previous one then it is either kept or discarded depending on the energetic difference between the two states (DE) and a random number, x, chosen between 0 and 1. If exp(DE/kT)  x then the new configuration is accepted. In other words, there is a unitary probability of accepting a move which results in an energy decrease, and an exponential probability based on the Boltzmann factor of accepting a move with a higher energy (the so-called Metropolis technique [54]). This procedure generates ensembles of configurations weighted in accord with the canonical Boltzmann distribution, and the average thermodynamic properties of the system can be calculated simply by averaging the thermodynamic properties associated with each configuration. At the same time, non-Boltzmann Monte Carlo sampling may be instrumental in accessing high-energy states that are infrequently reached in the Boltzmann simulations. For instance, energetics associated with a conformational transition from conformer A to conformer B would involve one or more transition states, which are of higher energy and would not be frequently sampled in an unconstrained simulation. In this case, one can perform a series of individual simulations with specifically selected constraints to focus on discrete regions of the reaction coordinate of the transition from A to B. This procedure is called umbrella sampling [55], and the effects of the different constraining potentials for each individual simulation can be mathematically removed to generate a potential of mean force for the transition coordinate that estimates the activation energy. One aspect shared by Monte Carlo methods and molecular dynamics is the ability to cross barriers between local minima. In the case of Monte Carlo, barrier crossing occurs by random change and acceptance of higher energy states is a function of temperature. In the case of molecular dynamics, the ability of crossing the barriers depends mostly on the initial velocities of atoms, or, in effect, on the average temperature. Because it is difficult to simulate systems (especially with explicit solvent) for long enough to allow conformational transitions, there will always be a concern that sampling of the potential surface was insufficient. One approach to this problem is to do multiple runs from different starting configurations of the system. One can examine convergence of the ensemble averaged properties of each run to determine if adequate sampling has occurred. Obviously, the ability of the system to make transitions over activation energy barriers depends on the temperature. In order to

BASICS AND TOOLS

25

increase the efficiency of sampling of conformational space, one can elevate the temperature for a short period and then resume the simulation at the desired temperature, allowing time for reequilibration. This technique, called simulated annealing, is useful for overcoming activation energy barriers between local minima. Another approach generates several ensembles (in the Monte Carlo approach) or several trajectories (in molecular dynamics) in parallel at different temperatures with periodic exchange between ensembles or trajectories (replicas) [56]. This method, called replica exchange, rather efficiently avoids situations where Monte Carlo or molecular dynamics runs are trapped in local minima. Replica exchange was applied to flexible peptides, such as enkephalin [56], but also for simulation of protein folding in the case of the small model proteins [57]. Generally, both the Monte Carlo and the molecular dynamics approaches have been widely used in many computational studies of peptides (see, for example, our earlier reviews [3,4]) and proteins (e.g. a recent review [26]). As an example, one can mention the Folding@home approach; the application of molecular dynamics sampling to small proteins [58,59]. This tool specifically aims to reproduce the dynamical process of protein folding from any extended 3D structure to the native fold. Current conformational resources and molecular dynamics algorithms are not sufficient to generate trajectories long enough to observe this kind of folding transition. However, even in very short trajectories there is a very small chance to generate a native-like structure as a result of stochastic crossing of the corresponding barriers. If a very large number of short molecular dynamics trajectories are generated independently and their results are pooled, the number of observed near-native structures may be large enough to be interpreted as an ensemble of structures in the nearest vicinity of the native one. The Folding@home method facilitates running hundreds of thousands of independent short trajectories of the same protein on a worldwide distributed grid of computers and then pools the results. In this way, it was possible to fold, for instance, a 36-residue protein from the villin headpiece with a root mean square (RMS) distance between atoms of the ˚ [60]. It is noteworthy, calculated and experimental structures of 1.7 A however, that the combination of molecular dynamics trajectories collected by Folding@home can be regarded as data on conformational sampling rather than a ‘real’ trajectory of protein folding. Sampling of ligand orientations within the receptor In fact, almost all methods of sampling the conformational state of peptides, such as systematic search, Monte Carlo and molecular dynamics are also applicable to the sampling of possible orientations (poses, binding modes) of a small

26

COMPUTATIONAL APPROACHES: AN OVERVIEW

ligand within the binding site of the receptor. In this case, however, parameters for sampling may be not only the dihedral angles of the Cartesian atom coordinates of the ligand, but also ‘global’ parameters determining orientation of the ligand with respect to the receptor as a whole. On the other hand, there are sampling methods summarily known as genetic algorithms. Unlike Monte Carlo or molecular dynamics simulations, this approach starts with an initial population of different conformations and/ or orientations (configurations) of the ligand. Each configuration is defined by a set of parameters, both global and ligand-conformational ones, which are treated as sets of genes in a chromosome, i.e. each may experience ‘mutations’, ‘crossovers’ and ‘migrations’ by analogy with genetic processes. Acceptance or rejection of the next population of configurations is governed by values of the corresponding scoring functions. The process of sampling stops when the value of the scoring function has converged, usually after several hundreds of steps of ‘genetic perturbations’, yielding an ensemble of plausible configurations (poses) of the ligand in the binding site. Genetic algorithms are available in a variety of molecular modelling programs, such as the ligand-receptor docking program AutoDock, which still is the single most widely employed docking program [34]. Sampling of protein conformational space The problem of sampling conformational space for proteins has, in fact, two very different limitations. On the one hand, the number of parameters (dimensions) defining conformational space in proteins is much larger than in peptides, which makes any of the sampling protocols outlined above both more complicated and more time consuming. But on the other hand, most proteins, unlike peptides, exist in a single native well-stabilized 3D structure of main interest in most practical applications of protein design. Historically, procedures for sampling conformational space in proteins were developed mostly in attempts to understand possible mechanisms of protein folding, a very general problem of molecular biology. However, as noted recently by Dill [61], protein folding is, in fact, three somewhat different problems. The first is the computational problem of predicting the stable 3D structure of the protein from its amino acid sequence (protein-fold prediction), and the other two relate to the thermodynamics of folding (how the native structure results from interatomic forces acting on the amino acid sequence) and to its kinetics (how the protein achieves the native fold from non-native ones). The first problem relates to equilibrium and is path-independent and therefore much closer to solution than the latter two. It is also the most relevant to the sampling problem in computational protein design. Some approaches used for protein design are briefly outlined below.

BASICS AND TOOLS

27

Comparative or homology modelling has proven to be the most successful method of determining the all-atom 3D models of proteins [62]. This approach takes advantage of the fact that the 3D structures of proteins belonging to the same evolutionary family are more conserved than their amino acid sequences. If it is therefore possible to establish a high level of homology between two sequences: one of the protein to be modelled (the target) and the other of the homologous protein with a known 3D structure (the template); there is a high likelihood that they possess very similar 3D structures. Accordingly, the first step of comparative modelling is the search for template sequences of known 3D structure in the PDB that can be aligned with the target sequence to reveal high levels of sequence homology (at least 30–40%). Then the target sequence is aligned with the selected 3D templates and a rough 3D structure of the target is built by introducing constraints (such as residue–residue distances, disulfide bridges, etc.) derived from the template structures. This draft structure is then transformed into the all-atom model and further refined using MM techniques. This general pathway of comparative modelling was implemented into various computational procedures and modelling packages, the most well known being, perhaps, the Modeller package [63,64]. The quality of predictions for the 3D target structure obtained by comparative modelling depends on several factors, but primarily on the quality of the target– template sequence alignment. Since different methods of alignment may produce different results, alignment to multiple homologous sequences is usually suggested. Also, the target–template sequence alignment often contains gaps, or insertions, in the overlapped sequences. A gap in the target sequence aligned to the 3D structure of the template can be eliminated by applying distance constraints, but an insertion (a loop in most cases) should be modelled separately [65]. There are also some specifically designed criteria determining the quality of the final 3D target models suggested by comparative modelling. Generally, comparative modelling is able to generate models with accuracy sufficient for further use in protein design (atom–atom root mean square deviation (RMSD) ˚ ) [62]. However, such results are only possible when an values of 2–3 A adequately homologous 3D template exists. Threading (or fold recognition) procedures have been developed to suggest possible 3D structures for proteins whose sequences do not possess strong homology to any sequences of known 3D structures [66]. The basis of threading was the observation that many proteins share similar 3D folds despite having relatively low sequence homology [67]. More recent studies confirmed this observation and revealed a limited

28

COMPUTATIONAL APPROACHES: AN OVERVIEW

number of significantly different folds in the PDB [68,69]. Therefore, threading was developed to suggest plausible folds for given sequences when direct sequence alignment with those in the PDB resulted in homology of less than 30% (see also [70]). Threading was not expected to yield an accurate 3D model of the target protein, but rather a rough model (such as Ca-trace) similar to a chosen 3D template. A suitable template may be determined by various procedures, such as by establishing distant evolutional relations between the sequence of the target protein and the sequences available in the PDB [71]; by building templates out of fragments possessing high sequence homology to some parts of the target sequence only [72]; by involving statistical predictions of the regular elements in the sequence, such as fragments of a-helices and b-strands [73]; or by combining these procedures with one another. To evaluate the quality of the templates with the suggested 3D models for the targets, threading procedures use various sequence-structure fitness functions. Threading is also often used as the first step in building much more detailed 3D structures of proteins. After an approximate 3D model of the protein chain has been obtained from threading, it is refined through calculation of an appropriate scoring function (e.g. the statisticallyderived residue–residue potentials) and then in all-atom approximation, adding backbone atoms and sampling different rotamers of the side chains, as exemplified in one of the recent studies [74]. Recently, several servers accessible on the Internet offered automated threading of protein sequences, employing various threading techniques or their combinations (e.g. [75]). The best predictions of 3D structures of proteins based on automated servers are in the range of about 2–6 A˚ differences in the atom–atom distance RMS values calculated between the predicted and experimental structures (see assessment [76]). There have also been recent attempts to derive 3D structures for sequences that proved difficult for threading by developing specific neural network-based procedures for machine learning [70]. De novo predictions of the protein structure became successful in the last several years, mostly due to the development of fragment-based assembly, the recently christened Rosetta algorithm [77]. The basic idea was initiated after an analysis of six-residue fragments of proteins in the PDB showed rather discrete clusters of conformers. Based on the limitation in the statistics of the observed conformer vocabulary, combinatorial assembly of hexamer fragments of known proteins present in the targeted sequence was suggested as an approach to generate nativelike 3D structures for the target. Modern-day Rosetta starts with selection of the sets of the backbone conformations represented in the

BASICS AND TOOLS

29

PDB for each short fragment (usually, three to nine residues long) of the target sequence. (In this regard, the Rosetta algorithm is not an ab initio physics-based approach, but is based on heuristics.) The backbone conformations of each fragment are then selected by Monte Carlo sampling, starting with the most represented, and are inserted into the entire sequence, which is initially put into the extended 3D structure. In parallel, fragments of the regular structures, such as a-helices and b-strands, which are predicted by statistical methods, are also inserted into the structure. An ensemble of various structures is generated by including various backbone conformations for each fragment segment, and by various perturbations made in the dihedral angles of the peptide backbone of the entire chain and in the space of the Cartesian coordinates. The generated rough structures are ranged by calculation of specifically developed scoring function, and the most probable are refined by an all-atom approximation with insertion of various rotamers of the side chains and by performing fine Monte Carlo sampling in the space of dihedral angles, followed by energy minimization with MM (Monte Carlo plus minimization procedure [78]). The loop fragments, which may not be represented in the PDB, have to be modelled separately and usually with less accuracy [79]. Over the last decade, the Rosetta approach has repeatedly shown good performance in the community-wide critical assessment of structure predictions (CASP), where researchers are asked to predict 3D structures of proteins with known sequences with experimentally determined 3D structures withheld until after the predictions are made [80–84]. The Rosetta algorithm has allowed numerous applications, including the design of novel proteins with a desired 3D structure [85]; design of protein sequences with new folds [86]; design of protein sequences aimed at avoiding specific 3D folds (‘negative design’) [87]; as well as applications for protein–protein docking [88–90]. Though it would be premature to evaluate the overall success of the Rosetta approach, it can be concluded that Rosetta provides a powerful tool for the computational design of peptides and proteins. Sampling of side-chain rotamers Many procedures for sampling the conformational space of peptides and proteins, as well as for sampling the orientations of the ligands, are developed primarily considering backbone conformers. At the same time, all-atom models should include lowenergy combinations of the side-chain rotamers, which must pack with one another and with the backbone without steric clashes. While the resulting 3D shape of the entire structure is not significantly affected by selecting a particular side-chain rotamer, the number of combinations of side-chain

30

COMPUTATIONAL APPROACHES: AN OVERVIEW

rotamers in a peptide chain may be roughly estimated as 10N, where N is the number of residues in the sequence. Therefore, efficient sampling of possible rotamers of side chains in peptides and proteins remains an important problem, which is approached in several ways as briefly discussed below. The theorem of the dead-end elimination proposes an energy-based equation and an algorithm to eliminate any rotamers inconsistent with the global energy minimum [91]. The theorem uses an assumption that the sum energy of the interactions of a given residue in a given side-chain rotamer consists of the energy of its interaction with the backbone and with all other side chains. If, for a specific residue, it is possible to determine the rotamer for which any combination of rotamers of all other residues would be associated with an energy value larger than the energy values associated with any other rotamer of the same residue, the former rotamer cannot be a part of the structure corresponding to the global energy minimum; this dead-end rotamer can be discarded from further consideration. Then another rotamer in another residue may be determined as a dead-end one and eliminated, and so on, until only one combination of side-chain rotamers remains. This procedure may be appropriate for producing a unique lowest-energy combination of the rotamers; obviously, it depends on the accuracy of the energy estimations, i.e. on the force field used. Though algorithms based on the dead-end elimination have been used in several practical applications [92,93], most sampling procedures employ selection of the rotamer combinations guided by libraries of discrete side-chain conformations (see e.g. [94] and references therein). Such libraries are obtained from statistical analysis of the PDB, and can either be backbone-independent or backbone-dependent. The latter differentiate distributions of rotamers that depend on backbone conformations, and the former ignore such dependence. Selection of side-chain rotamers based on libraries can be performed by simple procedures such as Monte Carlo sampling [95,96] or by more sophisticated algorithms that estimate the energies of interactions between side chains of different rotamers to determine the most energetically feasible combinations of rotamers (e.g. the SCWRL procedure [94,97]). Usually, estimation of energies involves simple steric energy combined with probabilities of rotamer populations for each residue and each backbone conformation. The library-based SCWRL procedure is computationally fast and produces several possible combinations of rotamers that are close to those combinations observed in the X-ray structures of proteins (see [94,97]). The procedure and its modifications have been used in many studies of protein conformational space (e.g. the recent studies [74,88,90,98]).

CYCLOPENTAPEPTIDE INHIBITORS OF CXCR4

31

Both these methods of sampling the rotamers of side chains were, in fact, developed specifically for computational studies of proteins. In applying the dead-end elimination theorem, one expects to find one single combination of rotamers corresponding to the single energy-minimal conformation of the backbone. The library-based approaches are also biased by using experimental distribution of rotamers in tightly packed proteins. For peptides and protein fragments, where several different backbone conformations can exist with comparable energies, a different approach to sampling of side-chain rotamers, which involves the sequential rotation algorithm, was developed [99]. This algorithm utilizes a stepwise grid search and consists of several steps. First, the side-chain dihedral angle q1 chosen from the qi angles (i ¼ 1 . . . n for all side chains), which possess the initial values of q01, is rotated with a chosen grid step, normally 30, from 150 to 180. All other angles are fixed in their q0i values. Rotation results in the energy profile where some angle value qmin1 corresponds to local energy minimum Emin(q1). Then the q1 angle is fixed in the qmin1 value, and the procedure is repeated for each qi angle to qn; at the end of this run all q01 values became equal to qmin1. The second run starts again from q1, and so on, until all qmini do not change any more, which means that the optimal values of qi angles are achieved. The algorithm has been extensively used to optimize the starting (prior to energy minimization) and final (after energy minimization) values of the dihedral angles of side chains, wi (i.e. qi ¼ wi), and has been validated by successful design of many biologically active analogues of peptides [3]. As an additional benefit, the algorithm produces the energy profiles along the wi angles in the final point of energy minimization, revealing a ‘slice’ of the multidimensional energy surface for each given wi. The algorithm is a path-dependent one, since its results may depend on the choice of the initial q0i values; however, this is easily compensated for by changing the order of the initial q0i angles.

2.3

COMPUTATIONAL STUDY OF CYCLOPENTAPEPTIDE INHIBITORS OF CXCR4

The second part of this chapter illustrates applications of the computational approaches as exemplified by our recent studies of cyclopentapeptide inhibitors of CXCR4, a GPCR involved in the process of entry of T-tropic HIV-1 strains into T-cells [6,7]. Accordingly, the main emphasis is not the final results, but rather an overview of corresponding computational protocols, accompanied by some caveats and comments based on practical experience.

32

COMPUTATIONAL APPROACHES: AN OVERVIEW

CXCR4 belongs to the chemokine family of GPCRs, and its natural ligand is the 67-residue peptide CXCL12, also known as SDF-1 [100,101]. Like all GPCRs, it includes seven helical transmembrane stretches (TM helices) as well as non-TM parts, namely the N- and C-terminal fragments and the extra- and intracellular loops connecting the TM helices. A large number of CXCR4 antagonists have been reported in the literature, one of the most potent being the cyclic pentapeptide FC131 (c(Gly1-D-Tyr2-Arg3-Arg4-Nal5), Nal is 2-naphthylalanine) , possessing affinity to the receptor in the nanomolar range [102]. This peptide represents a natural starting point for further design of potent and orally active peptidomimetic CXCR4 inhibitors with high specificity. Obviously, it would be of the utmost importance for such a design to know the 3D structure of the complex of FC131 and CXCR4. However, as for most other human GPCRs, the experimental 3D structure of the CXCR4 receptor remains unknown. The 3D structure of the peptide FC131 itself was determined by NMR studies [103] in solution, but not while complexed with CXCR4. Therefore, plausible 3D models for the complex of FC131 and CXCR4 were suggested by computational approaches. Specifically, two main problems needed to be addressed: (i) What were the possible ‘pharmacophoric’ conformations of FC131 (that is, feasible conformations while complexed with CXCR4)? (ii) What possible binding modes (orientations) for this (these) conformation(s) of FC131 exist in complex with CXCR4?

2.3.1

The 3D Pharmacophore Model for FC131

The 3D pharmacophore model for FC131 [6] was developed employing a typical ligand-based approach. SAR studies on FC131 analogues have shown that the presence of the four side chains (Tyr2, Arg3, Arg4, Nal5) results in compounds with highest affinity to CXCR4 [103,104]. However, substitution studies have shown that Arg3, alone among these four residues, can be replaced by Ala without a dramatic loss in affinity [103,104], i.e. Arg in the Xaa3 position is not an absolute prerequisite for binding. The stereochemistry of the chiral residues in FC131 also plays a very important role: of the 16 possible stereoisomers of FC131, only three, namely FC131 itself, 3 3 5 D-Arg -FC131 and D-Arg -D-Nal -FC131, possess nanomolar affinities to CXCR4 [102]. One could expect significant differences in conformational possibilities for the different stereoisomers of FC131 involved in the SAR studies. In turn, one could assume that geometric comparison of these conformations of stereoisomers could select the limited set of conformers

CYCLOPENTAPEPTIDE INHIBITORS OF CXCR4

33

similar in all analogues with high affinity, but dissimilar to all conformations available to stereoisomers with low affinity. This selected set of common active conformers would be the most plausible candidate for the 3D pharmacophoric model of FC131. Conformational sampling for several cyclopentapeptide compounds with high and low affinities to CXCR4 required several steps. First, all possible templates for backbone conformations were generated systematically by combining all local minima on the Ramachandran map characteristic for each chiral residue in a given cyclopentapeptide. This approach was justified by the analysis of experimentally determined cyclopentapeptide structures [105], showing that those compounds almost exclusively adopted combinations of (f, c) dihedral angles that were close to allowed regions for linear peptides in the Ramachandran map. Backbone conformations that allowed closing of the pentapeptide ring were selected as starting points for further energy minimization. Energy minimization was performed employing the ECEPP force field, and all redundant conformations (those conformers with similar values of all f and c angles) were removed. The choice of the ECEPP force field at this step avoided low-energy conformers with unrealistic (f, c) values (such as those in the lower-right quadrant of the Ramachandran map for L-amino acid residues), which might have occurred in sampling of cyclopentapeptides employing other force fields with flexible valence geometry [106]. The important side chains in FC131 and its analogues (Tyr2, Arg3, Arg4, Nal5) are of a size comparable with the cyclic backbone, so sampling of the side-chain rotamers should be rather exhaustive. Starting conformations for the side-chain rotamers were generated systematically for each backbone by including all the relevant side-chain rotamers (i.e. 60, 60 and 180) for most of the wi dihedral angles. Then each starting conformation was subjected to energy minimization that was, in turn, performed in two sequential steps. Since the number of generated starting ring conformations varied between 105and 106 from analogue to analogue, energy minimization was first performed with the ECEPP/2 force field, which employs rigid valence geometry that drastically reduces the computational time required for each minimization run. This in-house-developed program was used for all calculations employing the ECEPP force field. The calculations were performed in vacuo with a dielectric constant (e) of 80.0. There are considerable uncertainties involved in mimicking the heterogeneous transmembrane (TM) protein environment of the CXCR4 receptor, and this treatment was chosen to dampen the strong electrostatic interactions between charged groups, thereby allowing exploration of a wider set of low-energy conformations. After selection of low-energy conformations by a rather high

34

COMPUTATIONAL APPROACHES: AN OVERVIEW

cutoff (10 kcal/mol), the selected conformations were again subjected to energy minimization using the OPLS-AA force field as implemented in the MacroModel package for molecular modelling. Finally, the sets of lowenergy conformations for FC131 and all analogues were selected by another energy cutoff value (3 kcal/mol), the numbers of conformations in the sets for different analogues varying from about 50 to 250. In total, energetic optimizations were performed for 15 cyclopentapeptide compounds. When the sets of low-energy conformers were determined, they could be compared to locate possible 3D pharmacophoric models. Such structural comparison was performed by overlapping of the atomic centres defining the most functionally important groups in the molecules. Therefore, it was crucial to determine which atomic centres would best represent the important functional groups. In our case, 10 atomic centres were selected to represent a ‘minimalistic’ pharmacophore model in accordance with the SAR studies (see Figure 2.3). A more inclusive model included additional centres in the Arg3 side chain. Overlapping of the selected atomic centres in the low-energy conformers of the stereoisomers with high and low affinity toward CXCR4 allowed elucidation of two conformations present in the sets of low-energy conformations for all analogues with high or medium affinity, which at the same time were not present in the sets of low-energy conformations for all analogues with low affinity (Figure 2.4). An RMSD ˚ was used to differentiate similar and non-similar cutoff value of 1 A structures. The two selected conformations differed only in the side-chain rotamer of Nal5; both were regarded as plausible 3D models for the conformation of FC131 when bound with CXCR4.

Gly1 O

Nal5

H H N H

N H

NH O H N

O

H

HN

D-Tyr2

OH

O

N O

H N

H2N NH Arg4

NH2 HN Arg3

Figure 2.3 Structure of the cyclopentapeptide CXCR4 antagonist FC131. The ten atoms defining the minimalistic pharmacophore are shown as eight solid circles (C atoms) and two open squares (O and H). Courtesy of Professor Jay Ponder

CYCLOPENTAPEPTIDE INHIBITORS OF CXCR4

35

Figure 2.4 Superimposition of the two conformations of FC131, together representing a plausible 3D pharmacophoric model for binding to CXCR4

2.3.1.1 Caveats and comments It is noteworthy that the final results of the above computational protocol significantly depended on several empirical (user-specified) parameters that were applied at different steps of the procedure, such as selection of the functionally important atomic centres, the values of energy cutoffs for selection of low-energy conformations, or RMSD cutoffs for establishing similarity criteria between conformations. The choice of such parameters is always somewhat arbitrary, and in general varies from one specific application to another. Some other uncertainties, mainly those from the inevitable errors in energy estimation (inadequate force field, simplifications in description of electrostatic interactions and/or the molecular environment) are present in any computational procedure, so it is essential to validate the resulting hypotheses either by comparison with experimental data (when available) or by additional independent calculations. In the case of the 3D pharmacophoric model for FC131, one aspect of validation was convergence of the entire computational procedure to the hypothetical structures in Figure 2.4 – it might happen that no choice of the energy and RMSD cutoffs will yield a clear distinction between analogues with high and low affinity toward CXCR4. Another aspect was additional calculations performed for new analogues of FC131 with different conformational possibilities that were not used in deducing the 3D pharmacophore. Such calculations showed that the hypothetical models were still present in the sets of low-energy conformations for the new analogues with high affinity to CXCR4 and at the same time were absent in the sets of low-energy conformations for all new analogues with low affinity. In other words, self-consistency is a requirement for model building that must be maintained when confronted with new data.

36

2.3.2

COMPUTATIONAL APPROACHES: AN OVERVIEW

A 3D Model of the TM Region of CXCR4

Building a 3D model of the TM region of CXCR4 [7] was the next step in modelling the complex. Generally, computational procedures used in this step were the same as those developed and described previously [107,108]. CXCR4 belongs to the so-called rhodopsin-like group of GPCRs, so it was logical to assume that the TM regions in the CXCR4 sequence could be found by pairwise-sequence alignment of human CXCR4 with bovine rhodopsin. The alignment was performed using the ClustalW program (http:// www.ch.embnet.org/software/ClustalW.html), as shown in Figure 2.5. Assignment of the first and last residue in the TM helices of rhodopsin was generally based on the (f, c) torsions (20  (f, c)  100) of the X-ray structure (PDB entry 1F88 [10]; monomer A). Minor deviations from these limitations were accepted for residues that were obviously part of a helix, i.e. the Pro171–Leu172 fragment in the TM helix 4 (TMH4) of rhodopsin. Proline was not assigned as a TM residue in rhodopsin when it was located at the termini of the TM helices. Also, TMH1 was shortened by two and one residues on the extra- and intracellular sides, respectively, and

Figure 2.5 Final alignment of the transmembrane regions of CXCR4 with the rhodopsin sequence. Identical residues are shown with black background. Numbering is based on the rhodopsin sequence

CYCLOPENTAPEPTIDE INHIBITORS OF CXCR4

37

the intracellular part of TMH4 was extended by one residue. This resulted in the following assignment of the TM helices for CXCR4: TMH1: K38–T51– G64 (the first, middle and last residue, respectively); TMH2: T73–L86– A100; TMH3: F107–S122–I138; TMH4: K149–V160–F172; TMH5: V196–L208–I221; TMH6: Q233–F248–F264; and TMH7: K282–F292– Y302 (Figure 2.5). Then the TM helical fragments of the CXCR4 receptor were assembled in a helical bundle based on the following procedure: (i) determination of the conformation of each individual helix by sampling of the side-chain rotamers by a sequential rotation algorithm and energy minimization involving all dihedral angles; (ii) superimposition of the obtained helix conformations over the X-ray structure of Rh (Ca atoms only) according to the sequence alignment; (iii) packing of the seven helices into the energetically best arrangement while keeping the dihedral angles of the helical backbone fixed in the values obtained for individual helices (step (i)). Energy minimization for each individual TM helix of CXCR4 started from the backbone conformation (f, c and o dihedral angles) of the corresponding rhodopsin helix (PDB entry 1F88A; monomer A). The f and c angles were allowed to rotate with the limitation 60 – 40, which to some extent mimics limitations on intrahelical mobility of TM helices immobilized in the membrane. For the same reason, the o angle in Pro residues was limited to a value of 180 – 30. Side-chain torsions were optimized before and after energy minimization by a sequential rotation algorithm. Packing of the seven TM helices consisted of minimization of the sum of all intra- and interhelical interatomic energies in their multidimensional parameter space. These included the 6  7 ¼ 42 ‘global’ orientation parameters (related to movement of the individual helices as rigid bodies, namely translations along the coordinate axes X, Y, Z, and rotations around these axes, Tx, Ty, Tz) and the ‘local’ parameters (the dihedral angles of the side chains for all helices, but not for the backbone, which remained fixed with values obtained for the individual helices). Side-chain torsions were optimized prior to each energy minimization step by the sequential rotation algorithm. The coordinate system for the global parameters was selected as follows: the long axial X-coordinate axis for each TM helix was directed from the first to the last Ca atom; the Y axis was perpendicular to X and went through the Ca atom of the ‘middle’ residue of each helix; and the Z axis was built perpendicular to X and Y to maintain the right-handed coordinate system. The RMSD values (all Ca atoms) of the CXCR4 TM helical bundle relative to the X-ray structure of rhodopsin before and after ˚ , respectively. helix packing were 1.56 and 2.35 A

38

COMPUTATIONAL APPROACHES: AN OVERVIEW

All energy calculations were performed with the ECEPP force field using an in-house-developed program. Employing the ECEPP force field provided at least two significant advantages compared to employment of the force fields with flexible valence geometry. First, energy minimization occurred in the space of the dihedral angles (about 700 dihedral angles in the TM region of CXCR4) and not in the space of the Cartesian atomic coordinates (about 6000 atomic coordinates), which ensured much faster computations. Second, practical experience showed that the process of energy minimization over the large number of Cartesian atomic coordinates using a force field with flexible valence geometry almost always converged to a local energy minimum closest to the starting point, due to accumulation of small deviations of the starting Cartesian atomic coordinates. As a result, the final 3D structure of a TM bundle was very close to the initial TM bundle of rhodopsin. In other words, modelling of the TM bundles for different GPCRs by homology to rhodopsin using the force fields with flexible valence geometry yielded 3D structures virtually the same as the X-ray structures of rhodopsin. These results could hardly be consistent with the real 3D structures for the amino acid sequences of different GPCRs.

2.3.2.1 Caveats and comments Since there were no energy estimations for different conformations of the helix bundle, many uncertainties discussed above, such as selection of the values for the energy cutoffs, were not applicable in this case. However, the packing protocol we have developed [108] assumes that the backbone conformations of the TM helices do not change during packing and only the side chains undergo mutual adjustments (the ‘hard core’ and ‘soft shell’ assumption [109]). This assumption allows computation runs to be accelerated significantly, since energy minimization can be performed only in the space of the dihedral angles for the side chains, and not for those of the backbone, reducing the dimension of the search space from about 700 dihedral angles to about 300. Also, there is no need to calculate the energy of interatomic interactions within the peptide backbones of the TM helices, since the energy cannot change during minimization. The assumption is based on several experimental findings. For instance, the ˚ [110], average diameter of the TM helices in TM proteins is about 10 A which suggests that the backbone elements of the helices are not involved in direct interaction. Also, the (f, c) values in the X-ray structures of the TM helices rarely exceed the limits of 60 – 40 used in energy

CYCLOPENTAPEPTIDE INHIBITORS OF CXCR4

39

calculations for individual TM helices (see e.g. [10,111]). Nevertheless, independent experimental validation of the computational results obtained using this assumption would be highly desirable. However, in the absence of direct structural data on CXCR4, the only available experimental data are those from site-directed mutagenesis. Comparison with those data would be more meaningful for results obtained after computational docking of FC131 to CXCR4.

2.3.3

Docking of FC131 to CXCR4

Docking of FC131 to CXCR4 [7] was achieved in several sequential steps. In fact, docking was performed not only for FC131, but also for 10 analogues of FC131 with high or medium affinity to CXCR4 in order to elucidate orientations (binding modes) common for all potent cyclopentapeptides. The previously deduced 3D pharmacophore model (see Figure 2.4) was docked to the obtained 3D models of the TM region of CXCR4 using the AutoDock program, which assumed a rigid receptor (CXCR4) with flexibility (rotation of bonds) of the ligand (cyclopentapeptides). Sampling of possible orientations of the ligand employed a grid map of 1273 points with a spacing of 0.375 A˚, which covered the whole TM helical bundle of CXCR4, with the exception of a few residues on the intracellular side. The Lamarckian genetic algorithm (LGA), which uses a combination of a genetic algorithm and a local search, was used as the sampling method, and the standard scoring function implemented in AutoDock was used to evaluate ligand orientations. LGA docking selected 100 orientations with the most preferable values of the scoring function for each of the 11 compounds. The orientations obtained for each compound were then divided into clusters using an RMS cutoff of 1.0 A˚ (all atoms, no translation). Only clusters found near or within the extracellular pocket formed by the seven TM helices of CXCR4, as judged by visual inspection, were considered further, i.e. clusters representing binding to the lipid-oriented transmembrane faces of the receptor were discarded. For each of the remaining clusters, the orientation with the lowest value of the scoring function was extracted as a representative. Based on the 10 atoms defining the 3D pharmacophore (see Figure 2.3), an in-house program was used to compare orientations within the TM helices of CXCR4 for different compounds with the less ˚ (no translation). This resulted in the stringent RMS cutoff of 3.0 A identification of two binding modes (A and B) that were common for

40

COMPUTATIONAL APPROACHES: AN OVERVIEW

all 11 compounds. The obtained binding modes were then optimized by repacking the side chains in the FC131–CXCR4 complex. The optimization (repacking) procedure was essentially the same as for packing of the seven TM helices of CXCR4 (see above), with the exception that the ligand was introduced as the eighth component in the bundle, giving 6  8 ¼ 48 ‘global’ parameters. The ‘local’ parameters were the dihedral angles of the side chains for all helices, plus some of the dihedral angles of the side chains for the ligands. The optimized binding modes of FC131 within CXCR4 are illustrated in Figure 2.6. K282 D262 K38

Y255

Arg4 Nal5

E288 Y255 D-Tyr

Nal5

2

V197

E288 Arg4

H113 Y116 Arg3

Y116

2

D-Tyr

H113

K110

A

B

Figure 2.6 Diagram of the two possible complexes of CXCR4 with FC131, A and B. FC131 is coloured green; FC131 backbone is shown as a pentagon. The TM helices are represented by purple ribbons, and only the side chains of CXCR4 contact residues mentioned in the text are displayed (see colour Plate 2)

Computational docking procedures did not elucidate a single most probable binding mode of FC131 in the complex with CXCR4, but instead suggested two plausible options, which differed in the system of interactions between the side chains of the ligand and the receptor. Binding mode A was characterized by hydrogen bonding between the OH group of D-Tyr2 in FC131 and the e-amino group of K38 in CXCR4, and by a salt bridge formed by the guanidinium group of Arg4 and E288. The naphthyl ring of Nal5 in FC131 was surrounded by aromatic residues H113, Y116 and Y255. In binding mode B, Arg4 and E288 formed a salt bridge in the same way as for binding mode A; in addition, the guanidinium group of Arg4 was in contact with the phenyl ring of Y116, which might represent a cation-p interaction. The naphthyl ring of Nal5 was mainly in contact with Y255 and I284. The available experimental data of site-directed mutagenesis for CXCR4 were very limited (briefly reviewed in [7]) and were not obtained by direct evaluation of the affinity of the CXCR4 mutants to FC131. Still, there were clear indications that Y255 and D262 seemed to be important for HIV

CYCLOPENTAPEPTIDE INHIBITORS OF CXCR4

41

co-receptor activity of CXCR4, as well as for SDF-1a signaling, and possibly also for SDF-1a binding. Y255 and D262 also appeared to be involved in interaction with HIV, and some studies suggested that K38 and K110 also have some importance in this respect. These experimental data were in general consistent with the binding modes proposed by computational procedures. For instance, K38 was involved in the important interaction with FC131 in binding mode A, and D262 was identified as a contact residue for the ligand in binding mode B. Also, E288 and Y255 were involved in interactions with the ligand in both of the candidate binding modes, suggesting that the competing interaction of FC131 and other cyclopentapeptide compounds with these residues could be the reason for the inhibition of HIV entry by this class of compound. Therefore, final selection between the two binding modes depicted in Figure 2.6 was difficult to achieve by direct comparison of both models with available data on site-directed mutagenesis for CXCR4. However, these models provide a basis for rational design of the CXCR4 mutants with specifically selected modifications that will distinguish between the two binding modes. For instance, mutations of K38 introducing oppositely charged residues, such as K38E, will interrupt the H-bonds and/or salt bridges between the receptor and the ligand in binding mode A only. On the other hand, studies of FC131 binding to the E288A or E288K mutants will confirm that E288 is the key residue for interaction with the crucial Arg4 of the ligand in both binding modes. In summary, the computational procedures narrowed possible ‘binding space’ for the cyclopentapeptide compounds in the complex with CXCR4 down to two plausible binding modes. Both of these predict the involvement of E288 as the anchor point for this ligand class. This prediction was consistent with existing data on site-directed mutagenesis of CXCR4, showing E288 as an important residue for HIV co-receptor activity and CXCR4–SDF-1a interaction. The results should serve as a guide for the design of CXCR4 mutants, in order to provide further experimental data on binding of cyclopentapeptides to CXCR4, which in turn will facilitate the search for CXCR4 antagonists with improved druglike properties.

2.3.3.1 Caveats and comments While computational docking of FC131 to CXCR4 was successful in producing independent results consistent with available experimental data, it tacitly used an important assumption. Namely, it was assumed that FC131 and its analogues bind to the pocket formed by the TM helices of CXCR4

42

COMPUTATIONAL APPROACHES: AN OVERVIEW

without interfering with the extracellular loops connecting the helices or with the N-terminal fragment. Since the extracellular loops are flexible, it is quite possible that in the most ‘open’ conformation they will create an opening wide enough for a cyclopentapeptide molecule to pass through without significant sterical clashes between the cyclopentapeptide and the loops. In fact, our computational modelling of binding of other peptides to their corresponding GPCRs that accounted for the flexibility of the extracellular loops (such as that of angiotensin II to the AT receptor type 2 [112]) confirmed this suggestion. However, one would need to perform similar studies for the complex of FC131 and CXCR4 to be sure that this assumption remains valid; such modelling could use the computational techniques for restoring the large loops in GPCRs developed earlier [113].

ACKNOWLEDGEMENTS The authors are grateful to Prof. Jay Ponder for sharing the unpublished data regarding the AMOEBA polarizable force field and for general discussion of the subject.

REFERENCES [1] H. Hu, M. Elstner and J. Hermans, Comparison of a QM/MM force field and molecular mechanics force fields in simulations of alanine and glycine ‘dipeptides’ (Ace-Ala-Nme and Ace-Gly-Nme) in water in relation to the problem of modeling the unfolded peptide backbone in solution, Proteins, 50, 451–463 (2003). [2] S. C. Lovell, I. W. Davis, W. B. Arendall, 3rd, P. I. de Bakker, J. M. Word, M. G. Prisant, J. S. Richardson and D. C. Richardson, Structure validation by Ca geometry: ’, c and Cb deviation, Proteins, 50, 437–450 (2003). [3] G. V. Nikiforovich, Computational molecular modeling in peptide design, Int. J. Peptide Protein Res., 44, 513–531 (1994). [4] G. R. Marshall, D. D. Beusen and G. V. Nikiforovich, Peptide conformation: Stability and dynamics, in Peptides. Synthesis, structures, and applications, B. Gutte (Ed), Academic Press, San Diego, 1995. [5] H. A. Scheraga, Predicting three-dimensional structures of oligopeptides, in Reviews in computational chemistry, K. B. Lipkowitz and D. B. Boyd (Eds), VCH Publishers, New York, 1992. [6] J. Va˚benø, G. V. Nikiforovich and G. R. Marshall, A minimalistic 3D pharmacophore model for cyclopentapeptide CXCR4 antagonists, Biopolymers, 84, 459–471 (2006). [7] J. Va˚benø, G. V. Nikiforovich and G. R. Marshall, Insight into the binding mode for cyclopentapeptide antagonists of the CXCR4 receptor, Chem. Biol. & Drug Design, 67, 346–354 (2006).

REFERENCES

43

[8] J. R. Deschamps, C. George and J. L. Flippen-Anderson, Structural studies of opioid peptides: A review of recent progress in x-ray diffraction studies, Biopolymers, 40, 121–139 (1996). [9] J. Drews, Drug discovery: A historical perspective, Science, 287, 1960–1964 (2000). [10] K. Palczewski, T. Kumasaka, T. Hori, C. A. Behnke, H. Motoshima, B. A. Fox, I. LeTrong, D. C. Teller, T. Okada, R. E. Stenkamp, M. Yamamoto and M. Miyano, Crystal structure of rhodopsin: A G protein-coupled receptor, Science, 289, 739–745 (2000). [11] S. G. F. Rasmussen, H. Choi, J,, D. M. Rosenbaum, T. S. Kobilka, F. S. Thian, P. C. Edwards, M. Burghammer, V. R. P. Rathala, R. Sanishvili, R. F. Fischetti, G. F. Shertler, W. I. Weiss and B. K. Kobilka, Crystal structure of the human b2 adrenergic G-protein-coupled receptor, Nature, 450, 383–387 (2007). [12] V. Lafont, A. A. Armstrong, H. Ohtaka, Y. Kiso, L. Mario Amzel and E. Freire, Compensating enthalpic and entropic changes hinder binding affinity optimization, Chem. Biol. & Drug Design, 69, 413–422 (2007). [13] D. A. Pearlman, D. A. Case, J. W. Caldwell, W. S. Ross, T. E. Cheatham III, S. Debolt, D. Ferguson, G. Seibel and P. Kollman, AMBER, a package of computer programs for applying molecular mechanics, normal mode analysis and free energy calculations to simulate the structural and energetic properties of molecules, Comp. Phys. Comm., 91, 1–141 (1995). [14] D. A. Case, T. E. Cheatham III, T. Darden, H. Gohlke, R. Luo, K. M. Mertz, A. Onufriev, C. Simmerling, B. Wang and R. J. Woods, The AMBER biomolecular simulation programs, J. Comp. Chem., 26, 1668–1688 (2005). [15] L. G. Dunfield, A. W. Burgess and H. A. Scheraga, Energy parameters in polypeptides. 8. Empirical potential energy algorithm for the conformational analysis of large molecules, J. Phys. Chem., 82, 2609–2616 (1978). [16] G. Nemethy, M. S. Pottle and H. A. Scheraga, Energy parameters in polypeptides. 9. Updating of geometrical parameters, nonbonded interactions, and hydrogen bond interactions for the naturally occuring amino acids, J. Phys Chem, 87, 1883–1887 (1983). [17] G. Nemethy, K. D. Gibson, K. A. Palmer, C. N. Yoon, G. Paterlini, A. Zagari, S. Rumsey and H. A. Scheraga, Energy parameters in polypeptides. 10. Improved geometrical parameters and nonbonded interactions for use in the ECEPP/3 algorithm, with application to proline–containing peptides, J. Phys. Chem., 96, 6472–6484 (1992). [18] W. L. Jorgensen and J. Tirado-Rives, The OPLS potential functions for proteins. Energy minimizations for crystals of cyclic peptides and crambin, J. Am. Chem. Soc., 110, 1657–1666 (1988). [19] W. L. Jorgensen and J. Tirado-Rives, Potential energy functions for atomic-level simulations of water and organic and biomolecular systems, PNAS, 102, 6665– 6670 (2005). [20] W. R. P. Scott, P. H. Hunenberg, I. G. Tironi, A. E. Mark, S. R. Billeter, J. Fennen, A. E. Torda, T. Huber, P. Kruger and W. F. van Gunsteren, The GROMOS biomolecular simulation program package, J. Phys Chem, 103, 3596–3607 (1999). [21] B. R. Brooks, R. E. Bruccoleri, B. D. Olafson, D. J. States, S. Swaminathan and M. Karplus, CHARMM: A program for macromolecular energy, minimization and dynamics calculations, J. Comp. Chem., 4, 187–217 (1983). [22] J. W. Ponder and D. A. Case, Force fields for protein simulations, Adv. Protein Chem., 66, 27–85 (2003). [23] A. D. Mackerell, Empirical force fields for biological macromolecules, J. Comp. Chem., 25, 1584–1604 (2004).

44

COMPUTATIONAL APPROACHES: AN OVERVIEW

[24] I. K. Roterman, M. H. Lambert, K. D. Gibson and H. A. Scheraga, A comparison of the CHARMM, AMBER and ECEPP potentials for peptides. II. ’–c maps for N-acetyl alanine N0 -methyl amide: Comparisons, contrasts and simple experimental tests, J. Biomol. Struct. & Dynamics, 7, 421–453 (1989). [25] A. D. Mackerell, M. Feig and B. R. Brooks, Extending the treatment of backbone energetics in protein force fields: Limitations of gas-phase quantum mechanics in reproducing protein conformational distrubutions in molecular dynamics simulations, J. Comp. Chem., 25, 1400–1415 (2004). [26] H. A. Scheraga, M. Khalili and A. Liwo, Protein-folding dynamics: Overview of molecular simulation techniques, Ann. Rev. Phys. Chem., 58, 57–83 (2007). [27] R. A. Friesner, Combined quantum and molecular mechanics (QM/MM), Drug Discovery Today: Technologies, 1, 253–260 (2004). [28] R. A. Friesner and V. Guallar, Ab initio quantum chemical and mixed quantum mechanics/molecular mechanics (QM/MM) methods for studying enzymatic catalysis, Ann. Rev. Phys. Chem., 56, 389–427 (2005). [29] M. Elstner, T. Frauenheim, E. Kaxiras, G. Seifert and S. Suhai, A self-consistent charge density-functional based tight-binding scheme for large biomolecules, Phys. Status Solidi, B217, 357–376 (2000). [30] G. M. Seabra, R. C. Walker, M. Elstner, D. A. Case and A. E. Roitberg, Implementation of the SCC-DFTB method for hybrid QM/MM simulations within the amber molecular dynamics package, J. Phys Chem, A111, 5655–5664 (2007). [31] S. Patel and B. R. Brooks, Fluctuating change force fields: Recent developments and applications from small molecules to macromolecular biological systems, Molecular Simulations, 32, 231–249 (2006). [32] Z.-X. Wang, W. Zhang, C. Wu, H. Lei, P. Cieplak and Y. Duan, Strike a balance: Optimization of backbone torsion parameters of amber polarizable force field for simulations of proteins and peptides, J. Comp. Chem., 27, 781–790 (2006). [33] P. Ferrara, H. Gohlke, D. J. Price, G. Klebe and C. L. Brooks, 3rd, Assessing scoring functions for protein–ligand interactions, J. Med. Chem., 47, 3032–3047 (2004). [34] S. F. Sousa, P. A. Fernandes and M. J. Ramos, Protein–ligand docking: Current status and future challenges, Proteins, 65, 15–26 (2006). [35] R. Rajamani and A. C. Good, Ranking poses in structure-based lead discovery and optimization: Current trends in scoring function development, Curr. Opin. Drug Discov. & Dev., 10, 308–315 (2007). [36] A. N. Jain, Scoring functions for protein–ligand docking, Curr. Prot. & Pept. Sci., 7, 407–420 (2006). [37] J. Fernandez-Recio, R. Abagyan and M. Totrov, Improving CAPRI predictions: Optimized desolvation for rigid-body docking, Proteins, 60, 308–313 (2005). [38] H. J. Bo¨hm, The development of a simple empirical scoring function to estimate the binding constant for a protein–ligand complex of known three–dimensional structure, J. Comp.-Aid. Mol. Design, 8, 243–256 (1994). [39] H. J. Bo¨hm, Prediction of binding constants of protein ligands: A fast method for the prioritization of hits obtained from de novo design or 3D database search programs, J. Comp.-Aid. Mol. Design, 12, 309–323 (1998). [40] I. Muegge and Y. C. Martin, A general and fast scoring function for protein–ligand interactions: A simplified potential approach, J. Med. Chem., 42, 791–804 (1999). [41] I. Muegge, PMF scoring revisited, J. Med. Chem., 49, 5895–5902 (2006).

REFERENCES

45

[42] R. Teramoto and H. Fukunishi, Supervised consensus scoring for docking and virtual screening, J. Chem. Inf. & Model., 47, 526–534 (2007). [43] R. Wang, Y. Lu and S. Wang, Comparative evaluation of 11 scoring functions for molecular docking, J. Med. Chem., 46, 2287–2303 (2003). [44] E. Perola, W. P. Walters and P. S. Charifson, A detailed comparison of current docking and scoring methods on systems of pharmaceutical relevance, Proteins, 56, 235–249 (2004). [45] W. T. Mooij and M. L. Verdonk, General and targeted statistical potentials for protein–ligand interactions, Proteins, 61, 272–287 (2005). [46] D. Tobi and I. Bahar, Optimal design of protein docking potentials: Efficiency and limitations, Proteins, 62, 970–981 (2006). [47] C. Y. Yang, R. Wang and S. Wang, M-score: A knowledge-based potential scoring function accounting for protein atom mobility, J. Med. Chem., 49, 5903–5911 (2006). [48] J. Bernauer, J. Aze, J. Janin and A. Poupon, A new protein–protein docking scoring function based on interface residue properties, Bioinformatics, 23, 555–562 (2007). [49] W. Muller and H. Sticht, A protein-specifically adapted scoring function for the reranking of docking solutions, Proteins, 67, 98–111 (2007). [50] F. Fogolari, L. Pieri, A. Dovier, L. Bortolussi, G. Giugliarelli, A. Corazza, G. Esposito and P. Viglino, Scoring predictive models using a reduced representation of proteins: Model and energy definition, BMC Struct. Biol., 7, 15 (2007). [51] P. Ferrara, A. Curioni, E. Vangrevelinghe, T. Meyer, T. Mordasini, W. Andreoni, P. Acklin and E. Jacoby, New scoring functions for virtual screening from molecular dynamics simulations with a quantum-refined force-field (QRFF-MD). Application to cyclin-dependent kinase 2, J. Chem. Inf. & Model., 46, 254–263 (2006). [52] R. A. Dammkoehler, S. F. Karasek, E. F. Shands and G. R. Marshall, Constrained search of conformational hyperspace, J. Comp.-Aid. Mol. Design, 3, 3–21 (1989). [53] X. Zhang, G. V. Nikiforovich and G. R. Marshall, Conformational templates for rational drug design: Flexibility of cyclo(D-Pro1-Ala2-Ala3-Ala4-Ala5) in DMSO solution, J. Med. Chem., 50, 2921–2925 (2007). [54] N. Metropolis, R.A.E., M. N. Rosenbluth, A. H. Teller and E. Teller, Equation of state calculations by fast computing machines, J. Chem. Phys., 21, 1087–1092 (1953). [55] G. M. Torrie and J. P. Valleau, Nonphysical sampling distributions in Monte Carlo free-energy estimation – umbrella sampling, J. Comp. Chem., 23, 187–199 (1977). [56] Y. Sugita and Y. Okamoto, Replica-exchange molecular dynamics method for protein folding, Chem. Phys. Lett., 314, 141–151 (1999). [57] Y. M. Rhee and V. S. Pande, Multiplexed-replica exchange molecular dynamics method for protein folding simulation, Biophys. J., 84, 775–786 (2003). [58] M. R. Shirts and V. S. Pande, Screensavers of the world unite!, Science, 290, 1903–1904 (2001). [59] B. Zagrovic, E. J. Sorin and V. Pande, Beta-hairpin folding simulations in atomistic detail using an implicit solvent model, J. Mol. Biol., 313, 151–169 (2001). [60] B. Zagrovic, C. D. Snow, M. R. Shirts and V. S. Pande, Simulation of folding of a small alpha-helical protein in atomistic detail using worldwide-distributed computing, J. Mol. Biol., 323, 927–937 (2002). [61] K. Dill, S. B. Ozkan, T. R. Weikl, J. D. Chodera and V. A. Voelz, The protein folding problem: When it will be solved?, Curr. Opin. Struct Biol., 17, 342–346 (2007).

46

COMPUTATIONAL APPROACHES: AN OVERVIEW

[62] J. Moult, A decade of CASP: Progress, bottlenecks and prognosis in protein structure prediction, Curr. Opin. Struct Biol., 15, 285–289 (2005). [63] A. Sali and T. L. Blundell, Comparative protein modelling by satisfaction of spatial restraints, J. Mol. Biol., 234, 779–815 (1993). [64] A. Fiser and A. Sali, Modeller: Generation and refinement of homology–based protein structure models, Methods in Enzymology, 374, 461–491 (2003). [65] A. Fiser and A. Sali, Modloop: Automated modeling of loops in protein structures, Bioinformatics, 19, 2500–2501 (2003). [66] A. Godzik, Fold recognition methods, Methods of Biochemical Analysis, 44, 525–546 (2003). [67] R. T. Miller, D. T. Jones and J. M. Thornton, Protein fold recognition by sequence threading: Tools and assessment techniques, FASEB Journal, 10, 171– 178 (1996). [68] R. Day, D. A. Beck, R. S. Armen and V. Daggett, A consensus view of fold space: Combining SCOP, CATH, and the DALI domain dictionary, Protein Sci., 12, 2150–2160 (2003). [69] Y. Zhang and J. Skolnick, The protein structure prediction problem could be solved using the current PDB library, PNAS, 102, 1029–1034 (2005). [70] R. Casadio, P. Fariselli, P. L. Martelli and G. Tasco, Thinking the impossible: How to solve the protein folding problem with and without homologous structures and more, Methods in Molecular Biology, 350, 305–320 (2007). [71] H. Zhou and Y. Zhou, Fold recognition by combining sequence profiles derived from evolution and from depth–dependent structural alignment of fragments, Proteins, 58, 321–328 (2005). [72] Y. Zhang, A. K. Arakaki and J. Skolnick, TASSER: An automated method for the prediction of protein tertiary structures in CASP6, Proteins, 61 Suppl. 7, 91–98 (2005). [73] Y. An and R. A. Friesner, A novel fold recognition method using composite predicted secondary structures, Proteins, 48, 352–366 (2002). [74] S. Wu, J. Skolnick and Y. Zhang, Ab initio modeling of small proteins by iterative tasser simulations, BMC Biology, 5, 17 (2007). [75] S. Wu and Y. Zhang, Lomets: A local meta-threading-server for protein structure prediction, Nucleic Acids Res., 35, 3375–3382 (2007). [76] L. Rychlewski and D. Fischer, Livebench-8: The large-scale, continuous assessment of automated protein structure prediction, Protein Sci., 14, 240–245 (2005). [77] C. A. Rohl, C. E. M. Strauss, K. M. S. Misura and D. Baker, Protein structure prediction using Rosetta, Methods in Enzymology, 383, 66–93 (2004). [78] Z. Li and H. A. Scheraga, Monte Carlo-minimization approach to the multipleminima problem in protein folding, PNAS, 84, 6611–6615 (1987). [79] C. A. Rohl, C. E. Strauss, D. Chivian and D. Baker, Modeling structurally variable regions in homologous proteins with Rosetta, Proteins, 55, 656–677 (2004). [80] C. Bystroff and D. Baker, Blind predictions of local protein structure in CASP2 targets using the i-sites library, Proteins, Suppl. 1, 167–171 (1997). [81] K. T. Simons, R. Bonneau, I. Ruczinski and D. Baker, Ab initio protein structure prediction of CASP III targets using Rosetta, Proteins, Suppl. 3, 171–176 (1999). [82] R. Bonneau, J. Tsai, I. Ruczinski, D. Chivian, C. Rohl, C. E. Strauss and D. Baker, Rosetta in CASP4: Progress in ab initio protein structure prediction, Proteins, Suppl. 5, 119–126 (2001).

REFERENCES

47

[83] P. Bradley, D. Chivian, J. Meiler, K. M. Misura, C. A. Rohl, W. R. Schief, W. J. Wedemeyer, O. Schueler-Furman, P. Murphy, J. Schonbrun, C. E. Strauss and D. Baker, Rosetta predictions in CASP5: Successes, failures, and prospects for complete automation, Proteins, 53 Suppl. 6, 457–468 (2003). [84] P. Bradley, L. Malmstrom, B. Qian, J. Schonbrun, D. Chivian, D. E. Kim, J. Meiler, K. M. Misura and D. Baker, Free modeling with Rosetta in CASP6, Proteins, 61 Suppl. 7, 128–134 (2005). [85] R. Bonneau, C. E. Strauss, C. A. Rohl, D. Chivian, P. Bradley, L. Malmstrom, T. Robertson and D. Baker, De novo prediction of three–dimensional structures for major protein families, J. Mol. Biol., 322, 65–78 (2002). [86] B. Kuhlman, G. Dantas, G. C. Ireton, G. Varani, B. L. Stoddard and D. Baker, Design of a novel globular protein fold with atomic-level accuracy, Science, 302, 1364–1368 (2003). [87] G. L. Butterfoss and B. Kuhlman, Computer–based design of novel protein structures, Ann. Rev. Biophys. & Biomol. Struct., 35, 49–65 (2006). [88] J. Meiler and D. Baker, Rosettaligand: Protein–small molecule docking with full side-chain flexibility, Proteins, 65, 538–548 (2006). [89] O. Schueler-Furman, C. Wang and D. Baker, Progress in protein–protein docking: Atomic resolution predictions in the CAPRI experiment using Rosettadock with an improved treatment of side-chain flexibility, Proteins, 60, 187–194 (2005). [90] J. J. Gray, S. Moughon, C. Wang, O. Schueler-Furman, B. Kuhlman, C. A. Rohl and D. Baker, Protein–protein docking with simultaneous optimization of rigid-body displacement and side-chain conformations, J. Mol. Biol., 331, 281–299 (2003). [91] J. Desmet, M. De Maeyer, B. Hazes and I. Lasters, The dead-end elimination theorem and its use in protein side-chain positioning, Nature, 356, 539–542 (1992). [92] A. R. Leach, Ligand docking to proteins with discrete side–chain flexibility, J. Mol. Biol., 235, 345–356 (1994). [93] M. De Maeyer, J. Desmet and I. Lasters, The dead-end elimination theorem: Mathematical aspects, implementation, optimizations, evaluation, and performance, Methods in Molecular Biology, 143, 265–304 (2000). [94] A. A. Canutescu, A. A. Shelenkov and R. L. Dunbrack, Jr., A graph-theory algorithm for rapid protein side-chain prediction, Protein Sci., 12, 2001–2014 (2003). [95] B. Kuhlman and D. Baker, Native protein sequences are close to optimal for their structures, PNAS, 97, 10383–10388 (2000). [96] C. Wang, O. Schueler-Furman and D. Baker, Improved side-chain modeling for protein–protein docking, Protein Sci., 14, 1328–1339 (2005). [97] M. J. Bower, F. E. Cohen and R. L. Dunbrack, Jr., Prediction of protein side-chain rotamers from a backbone-dependent rotamer library: A new homology modeling tool, J. Mol. Biol., 267, 1268–1282 (1997). [98] R. L. Dunbrack, Jr., Comparative modeling of CASP3 targets using psi–blast and SCWRL, Proteins, Suppl. 3, 81–87 (1999). [99] G. V. Nikiforovich, V. J. Hruby, O. Prakash and C. A. Gehrig, Topographical requirements for delta-selective opioid peptides, Biopolymers, 31, 941–955 (1991). [100] C. C. Bleul, M. Farzan, H. Choe, C. Parolin, I. Clark-Lewis, J. Sodroski and T. A. Springer, The lymphocyte chemoattractant SDF-1 is a ligand for lestr/fusin and blocks HIV-1 entry, Nature, 382, 829–833 (1996). [101] E. Oberlin, A. Amara, F. Bachelerie, C. Bessia, J. L. Virelizier, F. ArenzanaSeisdedos, O. Schwartz, J. M. Heard, I. Clark-Lewis, D. F. Legler, M. Loetscher, M. Baggiolini and B. Moser, The CXC chemokine SDF-1 is the ligand for lestr/

48

[102]

[103]

[104]

[105]

[106]

[107]

[108]

[109]

[110] [111]

[112]

[113]

[114]

[115]

COMPUTATIONAL APPROACHES: AN OVERVIEW fusin and prevents infection by t-cell-line-adapted HIV-1, Nature, 382, 833–835 (1996). N. Fujii, S. Oishi, K. Hiramatsu, T. Araki, S. Ueda, H. Tamamura, A. Otaka, S. Kusano, S. Terakubo, H. Nakashima, J. A. Broach, J. O. Trent, Z. X. Wang and S. C. Peiper, Molecular-size reduction of a potent CXCR4–chemokine antagonist using orthogonal combination of conformation– and sequence–based libraries, Angew. Chem. Int. Ed., 42, 3251–3253 (2003). H. Tamamura, T. Araki, S. Ueda, Z. Wang, S. Oishi, A. Esaka, J. O. Trent, H. Nakashima, N. Yamamoto, S. C. Peiper, A. Otaka and N. Fujii, Identification of novel low molecular weight CXCR4 antagonists by structural tuning of cyclic tetrapeptide scaffolds, J. Med. Chem., 48, 3280–3289 (2005). H. Tamamura, A. Esaka, T. Ogawa, T. Araki, S. Ueda, Z. Wang, J. O. Trent, H. Tsutsumi, H. Masuno, H. Nakashima, N. Yamamoto, S. C. Peiper, A. Otaka and N. Fujii, Structure–activity relationship studies on CXCR4 antagonists having cyclic pentapeptide scaffolds, Org. & Biomol. Chem., 3, 4392–4394 (2005). J. H. Viles, J. B. Mitchell, S. L. Gough, P. M. Doyle, C. J. Harris, P. J. Sadler and J. M. Thornton, Multiple solution conformations of the integrin-binding cyclic pentapeptide cyclo(-Ser-D-Leu-Asp-Val-Pro-). Analysis of the (’, c) space available to cyclic pentapeptides, Eur. J. Biochem., 242, 352–362 (1996). G. V. Nikiforovich, K. E. Ko¨ve´r, W. J. Zhang and G. R. Marshall, Cyclopentapeptides as flexible conformational templates, J. Am. Chem. Soc., 122, 3262–3273 (2000). G. V. Nikiforovich, B. Mihalik, K. J. Catt and G. R. Marshall, Molecular mechanisms of constitutive activity: Mutations at position 111 of the angiotensin AT1 receptor, J. Pept. Res., 66, 236–248 (2005). G. V. Nikiforovich and G. R. Marshall, 3D model for TM region of the AT-1 receptor in complex with angiotensin II independently validated by site-directed mutagenesis data, Biochem. Biophys. Res. Commun., 286, 1204–1211 (2001). G. V. Nikiforovich, S. Galaktionov, J. Balodis and G. R. Marshall, Novel approach to computer modeling of seven-helical transmembrane proteins: Current progress in the test case of bacteriorhodopsin, Acta Biochim. Pol., 48, 53–64 (2001). J. U. Bowie, Helix packing in membrane proteins, J. Mol. Biol., 272, 780–789 (1997). J. Deisenhofer, O. Epp, I. Sinning and H. Michel, Crystallographic refinement at ˚ resolution and refined model of the photosynthetic reaction centre from 2.3 A Rhodopseudomonas viridis, J. Mol. Biol., 246, 429–457 (1995). C. Skold, G. V. Nikiforovich and A. Karlen, Modeling binding modes of angiotensin II and pseudopeptide analogues to the AT2 receptor, J. Mol. Model. Graph., 26, 991–1003 (2008). G. V. Nikiforovich and G. R. Marshall, Modeling flexible loops in the dark-adapted and activated states of rhodopsin, a prototypical G–protein coupled receptor, Biophys. J., 89, 3780–3789 (2005). T. Warne, M. J. Serrano-Vega, J. G. Baker, R. Moukhametzianov, P. C. Edwards, R. Henderson, A. G. Leslie, C. S. Tate and G. F. Schertler, Structure of a b1-adrenergic G-protein-coupled receptor, Nature, 454, 486–491 (2008). V. P. Jaakola, M. T. Griffith, M. A. Hanson, V. Cherezov, E. Y. Chien, J. R. Lane, A. P. Ijzerman and R. C. Stevens, The 2.6 angstrom crystal structure of a human A2A adenosine receptor bound to an antagonist, Science, 322, 1211–1217 (2008).

3 Aspects of Peptidomimetics Veronique Maes and Dirk Tourwe´

3.1

INTRODUCTION

Despite the fact that an increasing number of peptides are marketed as drugs, their rapid proteolytic cleavage, poor oral and tissue absorption, and rapid excretion remain major limitations. Therefore, considerable efforts have been made to develop methods to overcome these limitations. Many years of intense research efforts have led to a hierarchical approach to gradually transform a peptide into derivatives with less peptide character, and finally into a peptide mimetic, also termed a peptidomimetic [1,2] (Figure 3.1). A peptidomimetic should possess good receptor affinity and selectivity, the required agonist or antagonist character, stability against proteolytic degradation, and good bioavailability and biodistribution properties. In the first step in this process, SAR data of a bioactive peptide are generated using methods that maintain the peptide character of the derivative, leading to modified peptides [3]. These include N- and C-terminal truncations, chirality changes, amino acid substitutions, N-alkylations and so on. Following structural information, information about the topographic requirements for receptor interaction is obtained by a conformational constraining strategy. A global constraint of the peptide backbone is obtained by cyclization, whereas local constraints are obtained by using designed amino acids or secondary structure mimetics.

Peptide and Protein Design for Biopharmaceutical Applications © 2009 John Wiley & Sons, Ltd. ISBN: 978-0-470-31961-1

Edited by Knud J. Jensen

Figure 3.1

Pseudopeptide: amide bond modification Cα substitution backbone extensions

3D pharmacophore model

Peptidomimetic

hit optimization

random screening of compound collections

A hierarchical approach to peptidomimetic design

Modified peptide: conformational constraints: cyclic peptides constrained amino acids

Modified peptide: N- and C-terminal truncations, chirality changes, Ala scans amino acid substitutions

Bioactive peptide

MODIFIED PEPTIDES

51

A further step includes peptide bond modifications, Ca substitutions and chain extensions, resulting in derivatives with partial peptide character, or pseudopeptides. This combined knowledge, supplemented by conformational studies using NMR spectroscopy, X-ray diffraction, CD and molecular modelling, is used to generate a 3D pharmacophore model, which then allows the de novo design of a peptidomimetic. Peptide pharmacophoric groups can be displayed with a defined 3D orientation on a variety of scaffolds, many of which are small heterocycles, including ‘privileged templates’. Various definitions of peptidomimetics have been given in the literature [4]. Some investigators use the term ‘peptide mimetic’ to refer to pseudopeptides and even modified peptides. Moore uses a strict definition of a peptide mimetic as a molecule that no longer contains peptide bonds [5]. A definition given by V. Hruby describes a peptidomimetic as a designed compound whose pharmacophoric structural elements mimic a peptide’s binding element in 3D space, mimicing the binding and agonist or antagonist activity of a natural peptide ligand [2]. In parallel to this rational design, many compounds acting at peptide receptors as agonists or antagonists have been discovered through random screening strategies, followed by classical medicinal chemistry optimization. Although some of these do not bind to the peptide receptor in the same way as the native peptide does [6], most authors refer to them as peptidomimetics. Peptidomimetics have been classified as type I, II or III [6]. Type I mimetics are backbone-modified peptides. Type II mimetics are small nonpeptide molecules that bind to peptide receptors, and were mostly discovered by screening. Those designed mimetics possessing novel templates on which the pharmacophoric groups are displayed to generate a topographical mimetic belong to type III [6]. Oligomeric compounds such as oligoureas and oligocarbamates have also been termed peptidomimetics. Other examples include b- and g-peptides, which adopt well-defined structures such as helices or sheets, and are therefore called ‘foldamers’. In this chapter, some of the aspects of the chemistry and applications of modified peptides, pseudopeptides and peptidomimetics will be discussed.

3.2

MODIFIED PEPTIDES

When peptide chemists are faced with a new bioactive peptide sequence, the traditional approach to obtain structure-activity data generally involves the use of N- and C-terminal truncations,

52

ASPECTS OF PEPTIDOMIMETICS

and a D-amino acid or alanine scan to determine the importance of chirality and of the side chain of each constituent amino acid. This allows the identification of the groups that are important for binding to the receptor and for signal transduction. In order to optimize the steric and electronic properties of the peptide, and to limit its metabolic breakdown, extensive use is made of modified or noncoded a-amino acids. Successful examples of extensively modified peptides include the GnRH antagonist Ganirelix and the Bradykinin B2 antagonist Icatibant (Figure 3.2).

pGlu - His - Trp - Ser - Tyr - Gly - Leu - Arg - Pro - Gly - NH2

GnRH

Ac - D-2Nal - D-4Cpa - D-3Pal - Ser - Tyr - D-Har(Et)2 - Leu - Har(Et)2 - Pro - D-Ala - NH2 Ganirelix HN

HN

HN

O

N

O N H

H N

H N

N H

O

OH

O N H

O

O H N

N H

O

HN

N

O H N

O

N

N H

O

Cl

Bradykinin

N

NH2

O

O

OH

H - Arg - Pro - Pro - Gly - Phe - Ser - Pro - Phe - Arg - OH

H - D-Arg - Arg - Pro - Hyp - Thi - Ser - D-Tic - Oic - Arg - OH Icatibant (HOE 140) HN

H

S

O

H

N

O H2N

NH

OH

NH

H N

NH2

HN

NH2

O N

O

O

H N

N H

O

N OH

N

O

N H

O

O OH NH HN

Figure 3.2

NH2

Structure of modified GnRH and Bradykinin analogues

MODIFIED PEPTIDES

53

A wide range of such modified amino acids are now commercially available, which facilitates the application of this strategy to optimize the pharmacological profile of bioactive peptides. The consideration that the interaction of a peptide with its biological receptor would mainly involve the side chains led to the interesting concept that a reversal of the peptide bond from CO-NH to NH-CO in a peptide would increase the metabolic stability, while maintaining the receptor affinity [7]. The topographical similarity of a peptide structure with its retroenantiomeric structure was first recognized for cyclic peptides [4] (Figure 3.3). In the cycloretroenantiomer, the sequence of the peptide was reversed, and the configuration of each residue was inverted. Although differences in sidechain topology were noted [8], the concept of cycloretroenantiomerization has been applied successfully to many peptide sequences [9]. In a recent application, a library of retroenantiomeric cyclic pentapeptides was used to develop a CXCR4 antagonist that is more potent than the parent peptide [10] (Figure 3.3). Also, the cyclic osteogenic growth peptide C-terminal peptide OGP(10–14) and its retroenantiomer were shown to be equipotent [11]. R1 O

O

NH

R4

NH

R1 R2

HN HN

NH cycloretroenantiomers

O

R2 HN R4

NH

HN

R3

O

O O O O

R3 OH

OH

O

H N

O N H

cycloretroenantiomers

HN

NH O

HN NH

NH H2N

HN

O

HN

NH

NH

NH2

NH2

H2N

N H

O

NH

O O O O O

N H

HN

NH

CXCR4 antagonist O

OH

H N

O HN

NH

HN

OH

O NH

O

O

O

HN NH

HN

O

O cyclic OGP(10–14)

Figure 3.3

NH

O N H

O

cyclic retroinverso OGP(10–14)

Cyclic peptides and their cycloretroenantiomers

NH

54

ASPECTS OF PEPTIDOMIMETICS

In linear peptides, the retroinverso principle has been exploited to design modified peptides. Such a retroinverso peptide has a similar side-chain topology as the parent peptide, but the end groups are noncomplimentary, which can result in inactive compounds (Figure 3.4). This problem can be solved by incorporating a gem-diaminoalkyl residue at the amino terminus and a C2-substituted malonyl residue at the C-terminus [7]. The incorporation of these diaminoalkyl and malonyl residues within the peptide sequence results in partially modified retroinverso (PMRI) peptides. The intervening amino acids have an inverted configuration, while the end groups are identical to those in the parent peptide. Extensive reviews of the synthetic methods [4,9] and biological applications [7,9] have been published. A critical review [12] claims that the retroinverso strategy has been especially successful in immunological applications, where the partial retroinversion is more promising since crucial hydrogen bonding through the peptide bonds can be maintained, while still obtaining a metabolic stabilization of the crucial peptide bonds. As an alternative to the configurationally unstable malonyl residue, the [NH-CH(CF3)]Gly modification has been developed for application in partially modified retro peptides [13–15] (Figure 3.4). In this unit, a carbonyl CO is replaced by a CH(CF3). The resulting NH-CH(CF3) group closely resembles the NH-CO bond, since the C-CF3 bond is isopolar with the C¼O, the NH has a low basicity and the o backbone angle is close to 120°. It also, however, has some of the

R2

O H2N

N H

R1

O

R4

O

H N

OH

N H

R3

O

peptide R2

O H N

HO R1

N H O

R2

R4

O H N R3

NH2

H2N R1

O

R2

H2 N R1

N H

N H

gem-diaminoalkyl

R4

O

O

R3

O

Bn H N

OH

N H

2-alkyl-malonyl

N H

O

R4

O H N R3

OH O

O

end-group modified retroinverso peptide

retroinverso peptide(all-D peptide) O

H N

MeO O

R1

O

CF3 N H

O N H

OBn O

PMRI Ψ[NHCH(CF3)]Gly peptide

partial retroinverso peptide(PMRI)

Figure 3.4 The principle of partially modified retroinverso (PMRI) peptides

MODIFIED PEPTIDES

55

properties of the tetrahedral intermediate involved in the proteasemediated hydrolysis of a peptide bond [15]. The stereoelectronic proporties of the CF3 group lead to a turn-like conformation, similar to the one observed in malonyl-based retro peptides [14]. The CH(CF3)-NH substitution for an amide CO-NH is an example of an amide bond replacement resulting in metabolically stabilized pseudopeptides. Isosteric replacements of the amide function are discussed below. Further metabolic stabilizations can be achieved by modifications of the peptide backbone. R

H

replace with alkyl groups

N

R′

Me

O

R

Me

R

H

N H

N H O

N H

alkylate

R′

unsaturate

O Z-didehydro amino acid

N H

O E-didehydro amino acid

O

Figure 3.5 Examples of peptide backbone modifications

The N-methylation of the amide function imparts metabolic stability and improves membrane permeability. It also influences the peptide conformation, due to its steric hindrance and the elimination of a hydrogen bond donor. It restricts the conformational space of the amino acid largely to extended conformations [16,17]. Methylation of the a-carbon also results in an increase in metabolic stability, and has a strong effect on the conformation of the peptide. a,a-disubstituted amino acids (or Ca-tetrasubstituted a-amino acids) are very effective in inducing secondary structure, in particular helical structure. A prominent example is the achiral Aib (a-aminoisobutyric acid) residue, which is a strong helix inducer [18]. However, other a,a-disubstituted amino acid analogues such as the carbocyclic Acnc series have also been used for their helix-inducing properties. In contrast, Ca-ethylated residues, or achiral residues with two identical side chains longer than a methyl group, have been shown to induce fully-extended conformations [19]. In small peptides, Ca-methylated amino acids such as a-MePro induce turn structures [20]. Because of the steric hindrance of the a,a-disubstituted amino acids, their synthesis requires appropriate coupling conditions [21]. In a,b-didehydro-a-amino acids (DAaa), the chirality at the a-carbon is removed, while the orientation of the side chain is fixed in the E- or Z-geometric isomers (Figure 3.5). These unsaturated amino acids are components of many naturally occurring linear and cyclic peptides of microbial, plant and animal origin. Didehydropeptides have an increased

56

ASPECTS OF PEPTIDOMIMETICS

resistance to proteolytic degradation. The incorporation of dehydro amino acids in a peptide has a pronounced rigidifying effect on the peptide backbone conformation. Small peptides, containing a single DPhe residue, generally adopt a b-turn structure, while helices are formed in larger peptides [22]. This tendency is less pronounced for the E- than for the Z-configuration of the double bond [23]. In contrast, DAla-containing peptides adopt extended conformations. Several methods for the synthesis of the E- and Z-stereoisomers have been developed, and methods for the incorporation of these unsaturated amino acids into bioactive peptides such as enkephalin, angiotensin, bradykinin and dermorphin have been reviewed [24]. Initially, efforts were concentrated on the development of strategies to induce specific backbone conformations of a peptide, such as a a-helix, b-sheet, or turn structures, by controlling the F and C angles. In this context, cyclic peptides have played a major role in providing a global constraint of the peptide backbone, which will be reviewed in Chapter 4. The design of templates that mimic the secondary structures of a peptide is a field of enormous interest. A full discussion of turn mimetics is beyond the scope of this chapter, while helix and sheets mimics will be described in Section 3.4. However, it has become increasingly apparent that the side-chain dihedral angles w, in conjunction with the backbone angles, are critical for molecular recognition and for the transduction process. The importance of the sidechain orientation on a stable backbone template led Hruby to propose the concept of design in w-space, or topographical design [25]. The side chain of an amino acid can adopt three low-energy staggered conformations: gauche(), gauche(þ) and anti (also termed trans) (Figure 3.6) [25–27]. (Note that for (S)- and (R)-enantiomers of the amino acids, the orientation of the side chain in the gauche() and gauche(þ) is different [26,28].) The orientation of the side chain relative to the peptide backbone differs greatly between the three conformers. In the gauche() conformation the side chain points to the N-terminus of the peptide, in the anti conformation it points to the C-terminus, whereas in the gauche(þ) conformation it is oriented perpendicular to the peptide backbone.

+60°

H –HN –60°

R

CO–

H H gauche(–)

–HN

R

H H gauche(+) H

180°

H CO–

–HN

CO–

H H anti

R

Figure 3.6 Newman projections of the three low-energy staggered conformations in L-amino acids

MODIFIED PEPTIDES

57

The energy difference and barrier between these conformations is low, so that generally all three conformations are accessible. However, during the interaction of the peptide with its receptor, each of the critical side chains will adopt one of these low-energy conformations, thereby creating a unique pharmacophore. The ability to create topographical surfaces with specific stereoelectronic functions is a powerful tool in the design of peptidomimetics. This concept highly depends on the design and synthesis of side chain-constrained amino acids. The a,b-didehydro amino acid geometrical isomers of Phe (DZPhe and DEPhe) analogues have a fixed geometry of the side chain, which corresponds to a gauche() and an anti orientation, respectively [28,29]. The planar nature of the backbone, however, introduces considerable deviations from the original geometry. The w1 torsion angle can be restricted by van der Waals forces, which can be introduced by substituents at the b-carbon. Alternatively, the rotation around the Ca-Cb (and Cb-Cg, etc.) bond can be restricted by incorporation of the side chain into various ring structures. Evidently, any modification of the amino acid structure with the aim of constraining its w angles will have some impact on the backbone dihedrals as well. The first topographical probes which were introduced by V. Hruby were b-methylphenylalanine and b-methyltyrosine [25] (Figure 3.7). Later, b-methylTrp [25,30,31], b-Me(2’)Nal [32,33] and b-MeCha [34] were also reported.

H

H H H2N

COOH

(2S,3R)-βMePhe

H H2N

H

COOH

(2S,3S)-βMePhe

H H2N

COOH

β-iPrPhe

X H H H2N

H

COOH

TMP(X = H)/TMT(X = OH)

Figure 3.7

H H2N

COOH Dip

H H2N

NH

COOH

β-PhTrp

Examples of b-substituted amino acids for the control of w-space

58

ASPECTS OF PEPTIDOMIMETICS

A single b-methyl substitution does not constrain significantly the (w1,w2) space of an aromatic amino acid. Generally, there is a small energy preference of the gauche() rotamer for the erythro isomer (2S,3S) and of the anti rotamer for the threo isomer (2S,3R), which can be rationalized on the basis of steric interactions in the different rotamers (Figure 3.8).

H H H2N

H CH3 COOH

–HN R

(2S, 3S)-βMePhe H H H2N

CO–

CH3 H gauche(–)

COOH

(2S, 3R)-βMePhe

–HN R

CO–

–HN H3C

H H gauche(+)

CH3 CH3

CH3

R –HN H

H

H H gauche(–)

Figure 3.8

CO–

–HN

CH3 H gauche(+)

H3C

–HN H

R H anti

R CO–

CO-

CO– R H anti

Rotamers b-MePhe

Evidently, when the steric bulk of the b-substituent is increased, as in the b-isopropyl derivatives, the stereochemical bias is more pronounced [25,27]. A variety of other b-alkyl- and b-phenyl-substituted Phe, Tyr and Trp analogues have been reported [31,35–37]. However, except for the Dip residue [29], their use in bioactive peptides remains to be demonstrated. A very strong effect is obtained when the b-methyl substitution in Phe or Tyr is accompanied by methylation of the aromatic ring at the 20 ,60 -positions, leading to trimethylPhe (TMP) or trimethylTyr (TMT) [38] (Figure 3.7). The incorporation of (2S,3S)TMT in the opioid sequences of DPDPE, deltorphin-I and TMT-Tic dipeptides, as well as in a bicyclic oxytocin analogue, provided insight into the topographical requirements for receptor recognition by these ligands [38–40]. New synthetic methods for the efficient preparation of this interesting class of b-substituted amino acids continue to be developed [36,41]. The fluorenylglycine (Fgl) and 1-indanylglycine (Ing) structures (Figure 3.9) do not only constrain w1, but also w2. The Ing isomers can be use as conformational probes which constrain the w1 angle depending on the backbone conformation [28,29]. For a backbone with extended conformation, the (2S,3S)-Ing prefers the anti, and the (2S,3R)-Ing the gauche() conformation, with w2 þ60° and 60°, respectively. In a

MODIFIED PEPTIDES

59

H HN H

H H2N

COOH Fgl

CO

H H

COOH

(2S,3S)-Ing

CO–

HN –

HN

χ2 H H2N

H CO–

gauche(–)

H

H

gauche(+)

anti

H N χ2 H H2N

COOH

H H2N

COOH

H H2N

COOH

Figure 3.9 w1 and w2 constraints

b-turn conformation, the preferences are gauche() and gauche(þ), respectively. The w2 angle can also be efficiently restricted by the interaction between the phenyl group in o-substituted aromatic amino acids and the b-hydrogens (Figure 3.9). Their incorporation into the melanotropin analogue MT-II revealed interesting effects on receptor selectivity [42]. Many applications of these topographically constrained b-substituted amino acids in bioactive peptides such as the opioid peptides [25,34,43], glucagon [25], somatostatin [25,30,33,44], MSH [25], CCK [25], substance P [28,29,45], DPPIV inhibitors [46] and so on have been reported. Very interesting effects on affinity, selectivity, agonist versus antagonist character, stability and duration of action were observed. Whereas b-alkyl substitution provides a conformational bias at w1, tethering Ca to Cb through an alkylidene bridge of variable length in an amino acid results in the formation of a ring with concomitant fixation of the side chain, depending on the relative stereochemistry of Ca to Cb. For Phe this results in the 1-amino-2-phenylcycloalkanecarboxylic acid (cnPhe, n ¼ number of carbons in the ring) series. Tethering Ca to Cg results in 2,4-methano analogues (Figure 3.10). R

( )n

H H2N

R

R

R

COOH

H2N

COOH

n( ) H2N

COOH

H2N

COOH

n = 1–4

Figure 3.10 1-aminocycloalkanecarboxylic acids and 2,4-methano analogues

60

ASPECTS OF PEPTIDOMIMETICS

The synthesis and application of 1-aminocycloalkanecarboxylic acids has been reviewed up to 2000 [47], with an update up to 2007 [48], and those of the enantiomerically pure Phe analogues – cnPhe (n ¼ 3–6) – were also reviewed recently [49]. Asymmetric synthesis methods and efficient resolution on a cellulose-derived stationary phase have been reported. Next to cn analogues of the aromatic amino acids Phe, Tyr, Trp and His [26], various other substituted cycloalkane analogues have been prepared [50], such as: R¼OH (n ¼ 3, 4) [51–55], R¼CH3 (n ¼ 1, 2, 4) [54,56,57], R¼iPr (n ¼ 2, 3) [57,58], R¼PhCH2CH2, (10 )Nal, p-ClPh (n ¼ 1) [59]. Studies on model dipeptides RCO-Pro-cnPheNHR0 revealed the influence of the side-chain orientation on the conformation of the backbone modulating the b-turn preferences. Whereas the cyclopropane analogues (ACCs or r-amino acids) have been used in peptide chemistry for many years [60] and were shown to result in an increased metabolic stability [28,29], the application potential of the other analogues remains largely to be explored [49]. Enkephalin analogues containing a c5Ser or c6Ser residue at the 2-position were shown to be more potent than the native peptide, with subnanomolar affinities for the d-receptors [53,61]. In contrast to the 1-aminocycloalkane carboxylic acid analogues discussed above, little has been reported on the synthesis and use of the 2,4-methano analogues (Figure 3.10). Racemic methano analogues of Arg, Lys, Orn, Thr and Val were incorporated into the tuftsin sequence. Some analogues were considerably more active than the parent peptide, and showed high resistance to enzymatic degradation [62]. Asymmetric syntheses of cis- and trans-2,4-methanoVal and Leu were reported, as well as syntheses of the achiral 2,4-methanoPro, and of some of its chiral homologues [56,57,63]. In a similar fashion, the b-carbon of an amino acid can be tethered to the a-nitrogen, leading to 3-substituted cyclic iminoacids, which can be considered chimeras of amino acids and azetidine, proline or pipecolic acid (Figure 3.11). Similarly, linking Cg to the a-nitrogen leads to 4-substituted proline or pipecolic acid analogues, in which there is an additional constraint of w2. 3-substituted azetidine-2-carbocyclic analogues bearing side chains of Phe, 1-Nal, Leu, Val, Glu, Lys, Arg and Nle have been reported, but applications in bioactive peptides are lacking [50,64,65]. Various methods for the preparation of enantiomerically pure 3-substituted prolines as amino acid–proline chimeras or prolino amino acids have been reported, leading to chimeras of Asp [66], Arg [67–69], Lys, Glu, Gln, homoSer [69], Phe, homoPhe [70] and Tyr [69,71,72]. These amino

MODIFIED PEPTIDES

61 R

R + H2N

HN COOH

R′

H2N

COOH 3-R-Aze

COOH

N H

COOH

3-R-Pro

N COOH H 3-R-Pip

R′

R′ +

R

( )n N H

COOH

R

( )n N H

COOH

N H 4-R-Pro

COOH

N COOH H 4-R-Pip

Figure 3.11 Aze, Pro and Pip chimeras

acids were incorporated into the sequences of the melanotropin analogue MT-II [73], CCK, Angiotensin II, opioids [72], SP [74] and an IL-1 receptor antagonist [71]. A 4-substituted prolino–Arg analogue was prepared from Hyp, and incorporated into atrial natriuretic peptide analogue [75]. The chemistry of substituted pipecolic acid was recently reviewed [76]. Several amino acid–pipecolic acid chimeras have become available [76–78]. Considering that the side chain of aromatic amino acids can adopt three low-energy conformations around the Ca-Cb bond, V. Hruby introduced the 1,2,3,4-tetrahydroisoquinoline-3-carboxylic acid or Tic residue to limit the conformations to gauche() or gauche(þ), excluding the anti. According to the same reasoning, linking the aromatic side chain to the alfa carbon, resulting in 1-amino-tetralin-1-carboxylic acid (Atc), limits w1 to the gauche() and anti conformations, while the Aba residue limits w1 to the gauche(þ) and anti conformations (Figure 3.12). These three types of constrained amino acid therefore have complementary side-chain constraints (but different main-chain preferences), which makes them very useful for studying the conformational preferences during receptor interaction [79–81]. The use of the commonly available L- and D-Tic in many bioactive peptides has resulted in remarkable effects on affinity and selectivity and shifts in agonist/antagonist character. V. Hruby showed that the replacement of D-Phe1 by D-Tic1 in the cyclic somatostatin analogue CTP resulted in a selective m-opioid agonist [25,82]. One of the most successful applications in peptide design was the development by Schiller of the TIPP family (Tyr-Tic-Phe-Phe) of opioid peptides, resulting in very potent d-opioid

62

ASPECTS OF PEPTIDOMIMETICS X H

H –HN

CO–

–HN

H X

H X

H H gauche(+)

H

gauche(–) X

anti

X

X

X H2N

H

CO–NH

–HN

H

CO–NH–

COOH

HN

link

N NH2 COOH

COOH

O

Tic

COOH

NH2

Atc

R′

R

Aba

Figure 3.12 Complementary side-chain constraints by Tic-, Atc- and Aba-type amino acid analogues

antagonists [83,84]. In combination with the 20 ,60 -dimethyltyrosine (Dmt) residue, the tetrapeptide could be reduced to dipeptide analogues with remarkable potency. The Tic residue was key in the development of bradykinin antagonists [85]. Some recent applications include the melanocortin tetrapeptide Tic-D-Phe-Arg-Trp-NH2, where the Tic residue induces significant hMC4 selectivity [86]. It is also present in tetra- and pentapeptide HCV protease inhibitors [87]. Analogues of Tic were also prepared from Tyr (7-HO-Tic), mTyr (6-HO-Tic), Trp (Tcc or Tpi) and His (Spi), as were the benzo[f]-, [g]- and [h]Tic isomers [88]. (Figure 3.13)

OH HO N HN

N H 7-HO-Tic

COOH

N H 6-HO-Tic

COOH

N H

COOH

benzo[g]Tic

Figure 3.13

HN

N H

COOH

Tcc

Tic analogues

N H Spi

( )n N H

COOH

COOH

n = 1: Tic, n = 2: Sic, n = 3: Hic, n = 4: Nic, n = 5: Xic

MODIFIED PEPTIDES

63

When the benzo-substituted Tic isomers were used to replace the Tic residue in TIPP, activity dropped considerably [89]. The 7-HO-Tic residue was used in opioid peptides [72,81], and is present in the potent and selective k-opioid antagonist JDTic [90] and in macrocyclic inhibitors of HCV protease inhibitors [87], whereas 6-HO-Tic was used to probe the topological requirements for substrate recognition by protein tyrosine kinases [80]. Homologues of the Tic residue were prepared, and when incorporated into the CCK2 sequence, an interesting dependence on the ring size was observed. The nine-membered analogue Nic was equipotent to the parent Phe-containing compound [91]. Introduction of the Tcc residue in the bombesin sequence resulted in potent antagonists [92]. R R

NH

H2N

COOH

R = H: Atc R = OH: Hat

H2N

COOH

H2N

COOH

H2N

COOH

H2N

COOH

R = H: Aic R = OH: Hai

Figure 3.14 Tetralin- and indane-constrained amino acids

The 2-aminotetralin-2-carboxylic acid (Atc) and its 6-hydroxy analogue (Hat) have been used in opioid peptides (Figure 3.14) [72,93–96], tyrosine kinase substrates [80], antigenic peptides [79], melanocortins [97] and so on. A 6-phosphono-Atc analogue was used in Grb SH2 domain-directed tripeptides [98]. Conformational studies have indicated that the (R)- and (S)-tetralin amino acids are significantly more helixpromoting than Aib [99]. Recently, an asymmetric synthesis of this constrained residue has been published [100]. A solid-phase synthesis of the equivalent tetrahydrocarbazole-constrained Trp analogue was developed, which could be upscaled to a multigram scale [101,102]. This scaffold was used in the development of GnRH antagonists [101], neuromedin B partial agonists [103] and NK1/NK2 ligands [104]. The related achiral (when R¼H) indane-constrained amino acid (Aic) and its chiral hydroxyl-substituted analogue (R¼OH, Hai) have also frequently been used to probe the influence of the orientation of aromatic amino acids [72,79,80,94,96,105]. Replacing Tyr11 with Hai in the

64

ASPECTS OF PEPTIDOMIMETICS

NT(8–13) sequence resulted in a partial inverse agonist that was selective for the hNTR2 versus the hNTR1 receptor [106]. The 1-substituted isomers of Atc and Aic, which can be considered as constrained phenylglycine analogues, have also been used as rigid templates in modified peptides [97]. The azepine-constrained Phe (Aba, Figure 3.12) was originally developed by Flynn [107] and de Laszlo [108] for the design of ACE and renin inhibitors, ultimately leading to highly potent dual ACE/NEP inhibitors [109,110]. Later, this heterocycle was also used in farnesyl transferase inhibitors [111] (Figure 3.15).

COOH

NC

NH

O N

N H SH

N

O N

dual ACE/NEP inhibitor

N N H

Cl O

farnesyl transferase inhibitor

R′

N N H

R O Indolo-azepine analogue: Aia

Figure 3.15 Azepine-constrained peptide analogues

Several synthetic methods leading to this type of constrained Phe or Tyr analogue have been developed [107,108,111–114]. When this type of fixation was used as a constrained Phe3 residue in the opioid peptide dermorphin, it was able to shift the selectivity from the m- to the d-receptor [81]. In the N-terminal dermorphin tetrapeptide, the replacement of both Tyr1 and Phe3 by the corresponding Hba and Aba residues resulted in a highly constrained analogue that was shown to have potent in vivo analgesic effects when administered intrathecally as well as intravenously [114]. Further applications of this constraint include the melanocortin analogue MT-II [115], bradykinin [116] and opioid dipeptide analogues [114,117–119]. A related spiro-Aba analogue was shown to effectively induce a b-turn [120], and when incorporated into the HOE140 sequence resulted in a potent bradykinin B2 antagonist [116]. The equivalent indole-constrained analogue (Aia, Figure 3.15) can easily be prepared [121], and was used to prepare selective somatostatin peptide mimetics [122] (see Figure 3.47). A large number of other constrained amino acids have been described in the literature. A comprehensive discussion would be beyond the scope of this chapter. However, the general principles which underly the design of many of these analogues have been discussed here in order to make it

PSEUDOPEPTIDES

65

clear that these concepts have a proven value in peptide design, and their application constitutes an interesting step toward the transformation of a peptide into a peptide mimetic.

3.3

PSEUDOPEPTIDES

The observation that the metabolic breakdown of peptides occurs through enzyme-catalysed hydrolysis of the peptide bonds has been a primary motivation for the replacement of the scissile amide bonds by nonhydrolysable isosteres (Figure 3.16). Such isosteres were at the basis of the development of transition state analogue enzyme inhibitors, which led to a breakthrough in the development of inhibitors for several therapeutically important enzymes. Moreover, peptide chemists realized that it was important to investigate whether the backbone had a functional role or just served to orient and align the side chains [123]. In the latter case, the transposition of the side-chain groups from the a-carbon to the a-nitrogen, resulting in N-peptoids, seems a logical step. Other backbone modifications include the replacement of the a-carbon by, for instance, a nitrogen, leading to achiral azapeptides. In addition, the insertion of extra atoms into the backbone chain by using b-homo amino acids, g-amino acids or even longer-chain amino acids has been explored. replace α-carbon: α-aza-amino acid modify amide bond: pseudo peptide bond move side chain to N: N-peptoid H N O insert atom(s): β3-homo amino acid,

O N H

H N O

insert atom(s): β2-homo amino acid, vinylogous amino acid, γ-amino acid, urea analogue

Figure 3.16 Examples of strategies used in pseudopeptide design

Many amide bond replacements have been developed over the years. Some representative examples are given in Figure 3.17. Modifications of the amide function include a change at the nitrogen, at the carbonyl, or at both positions. The change of the amide bond is indicated by the symbol C, as an indication of a pseudo-amide function, while the replacing function is

66

ASPECTS OF PEPTIDOMIMETICS

indicated between brackets. Theoretical studies have compared the steric and electronic equivalence of different isosteres with the amide bond [124,125]. It is clear that the different types of modification will influence the physicochemical properties of the resulting pseudopeptide, such as the polarity, flexibility, hydrogen bond donating or accepting character. Therefore, it was soon realized that as well as preventing the enzymatic cleavage and influencing the membrane permeability, peptide bond replacements were also giving important information on the role of the amide function in receptor interaction by influencing the affinity and selectivity. Moreover, in many cases they also influenced the agonist or antagonist character of the native peptide, and provided a method to design enzyme inhibitors based on the concept of transition state analogues. R1

O H N

N H

R1

R1

O H N

N H

N H

R2

N H

Ψ(CH2NH2)

N H

R2

O

R1

N H

R1

O

N H

Ψ(CX = CY)

N H

R2

R1

O

R2

X

Ψ( )

N H

Y

O

N H

R2

O

N H

N H

R2

Ψ(cyclopropane)

Ψ(CH2-CH3)

X = H,F,CH3,CF3; Y = H,CH3 R1

R2

R1

N H

N H OH

N H

O

N H

R2 Ψ(CH2-OH)

R

R1

O O

OH

N H

N H

2

N H

R2 Ψ(CH2-SH)

N H

R2

OH

Ψ(CHOH-CH2OH) R1

O S

O

OH

N H

Ψ(CHOH-CH3)

Ψ(CHOH-CH2-NH), HEA isostere R1

R1

O

H N

N H

N H

O

R1

O

S R2 Ψ(CH2-SO)

N H

O

O S

N H

R2

O N H

Ψ(CH2-SO2)

Figure 3.17 Examples of amide bond modifications

A comprehensive discussion would be beyond the scope of this chapter, but the potential of this type of pseudopeptide modification will be illustrated with some representative examples. One of the most popular isosteres has been the reduced amide isostere: C(CH2NH). Synthetically, it is one of the easiest modifications of the peptide backbone, mostly obtained by reductive amination of the growing peptide with Boc- or Fmoc-protected aminoaldehydes, both in solution and on solid phase [126,127]. Recent alternatives such as Mitsunobu

PSEUDOPEPTIDES

67

alkylations have extended the range of available methods [128]. Therefore, a ‘reduction scan’ of each peptide bond can be performed in order to probe the backbone function, as was demonstrated for bombesin [129], CCK [130], somatostatin [131], dynorphin A [127,132] and neurotensin [133,134]. The resulting pseudopeptides have an improved metabolic stability [127,132,134–137] but also an improved chemical stability, as was demonstrated for the d-opioid antagonist TIPP, where cleavage of an N-terminal dipeptide was prevented [83]. This method also provides information about the involvement of the NH or CO groups in the stabilization of the conformation of the peptide by forming intramolecular hydrogen bonds. The reduction of a carbonyl group that is involved in the stabilization of a turn structure will result in a large drop in receptor affinity [131]. Despite the fact that it is generally expected that the replacement of the amide bond by a CH2NH will increase the flexibility of the peptide, it was demonstrated that the nitrogen which is protonated at physiological pH can be involved in a strong hydrogen bond, and stabilize turn structures [138,139], and thereby has a stong rigidifying effect on the backbone [140]. In many cases, this method resulted in the transformation of an agonist into an antagonist. In the case of bombesin, the change from agonist to antagonist by reduction of the Leu–Met as well as of the Ala–Val amide bond was correlated with the involvement of both amide functions in an intramolecular hydrogen bond, stabilizing a hydrogen-bonded b-turn in the bioactive conformation [141] (Figure 3.18).

H

H antagonist

Val

O

Ala Met H2N O

H N

N H

O

O

N H

H

H H O HN

H N Leu

His O

H antagonist

Figure 3.18 Amide bond reductions resulting in bombesin antagonists

68

ASPECTS OF PEPTIDOMIMETICS

Further antagonists obtained by reduction of an amide bond are collected in Table 3.1. In other examples, however, amide bond reduction maintained the agonist character of the parent peptides. This was the case for somatostatin [130], CCK [130] neurotensin [133,134], bradykinin [135], dynorphin [127,142] and xenin 6 [143].

Table 3.1 Antagonist peptides obtained by amide bond reduction Peptide analogue Gastrin [144] SP(6–11) [145] b-CM [146]

Sequence Boc-Trp-Leu C(CH2NH) Asp-Phe-NH2 D-Nal-Phe-Phe-Gly-LeuC(CH2NH)Phe-NH2 Tyr-ProC(CH2NH)Phe-Pro-Gly-OH

Ile-Glu-Pro-Dpr-Tyr-Arg-Leu-ArgΨ(CH2NH)Tyr-NH2 NPY(28–36) dimer [147] Nociceptin [148] BN [129] BN(6–14) [141] Neuromedin C [149]

Ile-Glu-Pro-Dpr-Tyr-Arg-Leu-ArgΨ(CH2NH)Tyr-NH2 [Phe1C(CH2NH)Gly2]nociceptin (1-13)NH2 [Leu14C(CH2NH)Leu15]BN [D-Phe6, Leu13C(CH2NH)Cpa14 ]BN(6-14) [Leu14C(CH2NH)Leu15]NMC

Reduced amide analogues of endomorphin 1 and 2 were shown to be partial agonists with significant in vivo antinociceptive action [150]. The discovery by Szelke in 1982 that the reduction of a Leu–Leu amide bond in an octapeptide sequence of equine angiotensinogen increased the inhibitory potency for renin 10 000-fold came at the start of the successful use of amide bond isosteres in the design of transition-state analogue enzyme inhibitors [151]. The tetrahedral geometry of this isostere resembles the transition state for peptide bond hydolysis, as was shown for the binding of a C(CH2NH) inhibitor to Rhizopus chinensis [152]. This isostere was later used in the design of inhibitors of HIV protease [153], farnesyl transferase [154] and b-secretase [155]. However, it was never as successful as other isosteres such as the hydroxyethyl isostere or the hydroxyethylamine isostere, which will be discussed below. Whereas many of the peptide bond modifications shown in Figure 3.17 result in an increased flexibility of the main chain due to the removal of the partial double bond character of the amide function, its replacement by an alkene – C(CH¼CH) – perfectly maintains the trans geometry. In order to mimic the cis amide bond geometry, Z-alkene isosteres have been prepared, mainly to mimic cis-Xxx-Pro amide bonds. However, the alkene function does not reproduce the polar

PSEUDOPEPTIDES

69

character of the peptide bond, its hydrogen bond donor and acceptor functions, or its steric bulk. Therefore, substituted alkenes have been prepared (Figure 3.19). The fluoroalkene isostere restores the polarity of the original amide function, and several stereoselective synthetic approaches to this isostere have been developed [156].

O H N

X

H N N H

O

O

H N O

Y

O H N

H N

NH H N

O O

Figure 3.19

H N O

X Y H N

O O

Alkene isosteres as trans and cis amide bond mimics

The (E)- and (Z)-fluoroalkene and alkene isosteres were used to probe the structural requirements for dipeptide recognition by the peptide transporter PEPT1. A preference for the trans amide equivalents was found, but contrary to the expectation, no increased affinity for the fluoroalkene versus the alkene isostere was observed [157]. Fluoroalkene tripeptide analogues were proposed as ground-state analogue inhibitors for thermolysin: they bound to thermolysin about one order of magnitude more thightly than the substrates [158]. A CF3-substituted (E)-alkene isostere was proposed to provide an improved mimicry of the electrostatic potential surface of the amide bond, as well as of its dipole moment [159]. Efficient synthetic methods for the asymmetric synthesis of this dipeptide isostere have been reported [160,161]. Methyl-substituted alkenes were shown to induce b-turn conformations in a peptide [159,162]. In Gramicidins S, the incorporation of the LeuC[(E)-C(CF3)¼CH]D-Phe isostere resulted in a better agreement with the secondary structure of the native Gramicidin S, than was that of the LeuC[(E)C(CH3)¼CH]D-Phe isostere [160]. In the same cyclic peptide, the D-PheC[(E)-C(CH3)¼CH]Ala dipeptide isostere was shown to be an effective replacement for D-Phe-Pro in a type II0 b-turn conformation [163]. In contrast to the C(CH2NH) analogues, the synthesis of the alkene isosteres requires a considerable synthetic effort. This limits the

70

ASPECTS OF PEPTIDOMIMETICS

widespread application of this type of isostere. Synthetic methods for the C(CH¼CH) isostere have been reviewed [164], and new ones are still being developed [165,166]. A recent report on a solid-phase procedure to assemble the alkene isostere using aziridine building blocks might improve this situation [166,167]. Despite the differences in polar and steric character, alkene replacements have resulted in active and stable pseudopeptides. The Leu13C[(E)-CH¼CH]Leu14 bombesin(6–14) analogue turned out to be an antagonist, like its C(CH2NH) analogue [168]. Recently, C(CH¼CH)-containing dipeptide mimics have been converted into cyclopropane amide bond isosteres (CPDIs), which have a more flexible backbone. These CPDIs should have a greater resistance to oxidative metabolism that the corresponding alkene isosteres [169]. Other rigid replacements mimicking an amide bond use aromatic or heteroaromatic rings. The 1,2,4-oxadiazole and 1,2,4-triazole rings mimic the trans amide bond geometry [170,171], while the o-substituted benzene (o-AMPA), the tetrazole and the pyrole rings mimic the cis amide bond geometry [172–174]. The 1,2,3-triazole motif has recently become an attractive isostere because of its easy synthesis through Cu(I)-catalysed Huisgen 1,3-dipolar cycloaddition chemistry [175,176] (Figure 3.20). O R1 H N

N H

R1

O

R2

O

N H

N H

R2 N H NH O O

O R2

R1

O N

N H

O

N H

N

HN

N H o-AMPA

O R2

N H

N N N

1,2,4-triazole

N

tetrazole

N H

N H N

N N

O

O N

N H

N H

N

N H

1,2,4-oxadiazole R1

R1

N H

R2

1,2,3-triazole

N H

N

1,2-pyrole

Figure 3.20 (Hetero)aromatic ring replacements for the amide bond

N H

PSEUDOPEPTIDES

71

The amide bond replacement with a hydroxy-substituted tetrahedral carbon (Figure 3.17) has been particularly successful as transition state analogues in the design of enzyme inhibitors. Many pseudopeptides containing the C(CHOHCH2NH) hydroxy-ethylamine (HEA) isostere replacement turned out to be potent enzyme inhibitors of aspartic proteases such as renin and HIV protease (Figure 3.21). Further medicinal chemistry optimization has resulted in the peptidomimetic HIV protease inhibitors that are now on the market, such as Saquinavir (HEA isostere), Lopinavir (HE isostere) and Darunavir (HEA isostere) [177,178] (Figure 3.21).

H N

O O

tBu

H N N

N H

O

N

N H

H

OH

O

O O

O

H N OH

N

NH

O

H H 2N Saquinavir: Ψ(CHOH-CH2-N) isostere

H O

NH2

OH H N

O H

H

Lopinavir: Ψ(CHOH-CH2) isostere

N

S

O

O

O

O Darunavir: Ψ(CHOH-CH2-N) isostere

Figure 3.21 HIV protease inhibitors

In contrast, the pseudopeptidic renin inhibitors did not reach the market; more efficient ones such as the ‘kirens’ were developed based on a vicinal diol motif [179]. Potent b-secretase inhibitors were developed based on the C(CHOH-CH2) and C(CHOH-CH2-NH) motifs as transition state analogues [180,181] (Figure 3.22). H3C

I

S

O

H N

Boc-NH O

OH

CH3

O

H N

O N H

H N

N O

OH

O

H N F

F

Figure 3.22 inhibitors

C(CHOH-CH2)- and C(CHOH-CH2-NH)-containing b-secretase

72

ASPECTS OF PEPTIDOMIMETICS

These examples have demonstrated the power of amide bond modifications to design novel stable pseudopeptidic agonists and antagonists for peptide receptors and powerful inhibitors of peptidases. Another successful backbone modification is the replacement of the a-carbon by the achiral nitrogen, resulting in azapeptides (Figure 3.23). R2

O H N R1

N H

R2

O H N O

N R1

N H

O

Figure 3.23 Azapeptide modification

Due to the formation of a rigid urea structure, the rotation around the Ca-C(O) bond (C angle) is greatly reduced in azapeptides. Electronic repulsions of the lone pairs of the two adjacent nitrogens also restrict the F angle. Both effects contribute to a reduction of the flexibility of the backbone. Aza-amino acids were shown to favour the formation of turn structures [182]. A method for aza-amino acid scanning of a peptide using Fmoc–solid phase chemistry has been described, and the synthetic strategies were discussed [183]. A large number of aza-analogues of biologically active peptides have been prepared, including angiotensin II, oxytocin, enkephalin, GnRH and somatostatin. (see cited refs in [183]). Moreover, aza-amino acids were shown to be effective building blocks for obtaining inhibitors of serine or cysteine proteinases: Z-Arg-Leu-Arg-[a-aza-Gly]-IleVal-OMe is the most potent and selective inhibitor of cathepsin B described [184]. An original approach to develop metabolically stable pseudopeptides consists of shifting the side chain of an a-amino acid to the a-nitrogen [185]. This results in N-alkylated glycine oligomers, termed ‘peptoids’ [186] (Figure 3.24). If the original peptide sequence is maintained, the Ca to N shift results in an increased distance between the side chain and the carbonyl group. This relative distance can be restored by using the reverse peptide sequence. The peptoid backbone is achiral and exhibits a larger flexibility than that of a peptide. Both the cis and trans conformations of the backbone amide can be significantly populated [187]. Nevertheless, it was shown that peptoids containing a-chiral aromatic or aliphatic side chains adopt remarkably stable helical structures, which can mimic the antibacterial activity of magainin [188,189]. Their synthesis can be achieved either by using preassembled monomeric N-substituted glycines, or by

PSEUDOPEPTIDES R2

O

H N

73

N H

R1

BocHN N H

R3

O

N

O

H N

R3

O

N

N

N

N

O

O

N

R2

N

H N

O N H

OBn

O

C

peptide

R1

H N

hydrazinoazapeptoid N H

O

O C

peptoid

O N

OH

R O N R1 C

R2

O

n extended peptoid

N

N R3

O peptoid

N

Figure 3.24 Peptoids and modifications

the so-called submonomer approach, which consists of coupling bromoacetic acid to the growing peptoid sequence, followed by substitution using primary amines with substituents corresponding to the side chains of the proteinogenic amino acids [17,186]. The resulting N-alkylated glycines are indicated as Naaas, for example Nphe, Nleu and so on. However, many other functionalities can easily be introduced in the side chains, including unprotected heterocyclic structures [190], and even a method for clicking various groups to azidoalkyl-containing oligomers by azide–alkyne cycloaddition reaction has been developed [191]. The full peptide sequence can be transformed into a peptoid structure, but single residues can also be modified to the N-peptoid, allowing an N-peptoid scan [192–194]. This has resulted in modified peptide structures with an increased affinity, stability or selectivity compared to the original peptide, as was illustrated for ligands for the opioid, melanocortin, vanilloid, somatostatin and andrenergic receptors (for a list see [195] and [186]). The partial peptoid [Nphe1, Arg14,Lys15] nociceptin-NH2 is a potent and selective nociceptin antagonist [196]. Peptoid and retropeptoid analogues of amylin(20–29) were shown to be b-sheet breakers with limited ability to inhibit amyloid formation [197]. Various modifications to the original method have been developed, leading to extended peptoids [198] or hydrazinoazapeptoids [199,200].

74

ASPECTS OF PEPTIDOMIMETICS

Several modifications which result in an extension of the peptide backbone have been developed (Figure 3.25). These modifications have been applied as single substitutions in a peptide sequence, or as multiple substitutions leading to oligomeric pseudopeptides [201]. A most prominent example is the use of homologated amino acids. The use of b2- and b3-homo amino acids results in an increased metabolic stability, and the formation of turn and helix structures. A discussion of the conformational properties of b-peptides is beyond the scope of this chapter, but has been reviewed extensively elsewhere [202]. Also, vinylogous peptides and the related saturated g-peptides exhibit stong folding properties [202,203]. Even peptides with longer alkyl chains such as d-peptides show a stong structuration [204]. Pseudopeptides containing a urea modification have been shown to behave as foldamers [205]. Oligourea analogues of neurotensin and Leu-enkephalin analogue have been prepared [206]. A single urea-substituted Leu-enkephalin analogue was shown to be equipotent to the natural peptide [207]. Oligocarbamate ligands with nM affinity for human thrombin were discovered trough screening of a library [208]. R1 N H

O H N R1

O

β3-homo-amino acid

R1 N H O

γ-peptide

R1

R2

N H

N H

H N

O

H N

N H

N H

R2

R2

O

β2-homo-amino acid O

H N

R1

O

H N

N H

R2

O

R1

H N

N H

R2

O

H N

O

H N

N H

vinylogous peptide R1

O

H N 2

R O ureidopeptide

N H

O H N

O N H

O

R2

N H

carbamate

Figure 3.25 Examples of backbone-extended pseudopeptides

Although this discussion is far from comprehensive, and a wide variety of other backbone modifications for the construction of metabolically stable, potent and selective pseudopeptides have been developed, the examples presented here convincingly demonstrate the power of the strategy. Moreover, these modifications provide important information

SECONDARY STRUCTURE MIMICS (EXCLUDING TURN MIMICS)

75

about the functional groups and their orientation, which are essential for interaction with the receptors or enzymes, allowing further reduction of the peptide character of the molecules, ultimately resulting in peptidomimetics.

3.4

SECONDARY STRUCTURE MIMICS (EXCLUDING TURN MIMICS)

Within proteins and peptides, secondary structure elements such as a-helices, b-turns and b-strands determine the 3D orientation of the side chains. These secondary structure elements are recognition motifs for the interaction with other proteins, receptors and enzymes. Short peptide sequences corresponding to the helical or b-sheet regions of proteins generally have only low populations of conformations corresponding to an ahelix or b-strand. Given the general flexibility of a peptide, it is important to stabilize, or fix, the conformation which interacts with the receptor protein. The minimization of the entropy penalty for adopting the bioactive conformation generally results in a higher affinity. Mimicking the functional helix or sheet region of a protein by appropriate mimetics allows the preparation of protein mimetics with a much reduced size and better synthetic accessibility [209]. In order to induce secondary structure in peptides, amino acids with known propensities to adopt various secondary structures are often used, as explained in the previous sections on for instance the propensity of a,a-disubstituted amino acids to induce helices. A second approach is to use ‘nucleators’ of secondary structures. Such nucleators are responsible for inducing for instance a b-sheet or a-helical structure in the peptide that is appended to them. The design and synthesis of structures that mimic the secondary structure of a peptide or protein is a very active field of research. Given the importance of turn structure for receptor recognition, extensive research efforts have been and are still being devoted to the development of turn mimics.

3.4.1

b-strand Mimetics

In an extended b-strand, the peptide backbone is in a fully extended conformation; the side chains of contiguous residues are presented on alternating sides of the strand. The amide NH and CO groups are exposed

76

ASPECTS OF PEPTIDOMIMETICS

for hydrogen bonding to the receptor, or to another – parallel or antiparallel – b-strand. In the latter case, b-sheets are formed. The b-strand is an important structural element that is preferentially recognized by proteases and MHC proteins [210,211]. This b-strand conformational selection of proteases explains the resistance of folded/structured regions of proteins to proteolytic degradation. Therefore, the fixation of an extended b-strand ligand conformation should produce high-affinity inhibitors even for proteases of unknown structure [211]. Additionally, b-strand mimetics are interesting as inhibitors of b-sheet aggregation, a phenomenon which is associated with a number of neurodegenerative diseases such as Alzheimer’s, Huntington’s and Parkinson’s diseases and prion potein diseases [212]. Comprehensive reviews on b-strand mimetics up to 2004 have been published [210,213]. More recently, de Vega reviewed examples up to 2007 [212].

3.4.1.1 b-sheet nucleating templates In proteins, antiparallel b-sheet formation is quite often nucleated by tight turns, resulting in b-hairpin structures. The chain reversal is achieved by a two-residue nucleus in a b-turn conformation that is stabilized by a 4 ! 1 hydrogen bond. Several templates for ensuring polypeptide chain reversal have been developed, and their capability for inducing b-sheet formation of the attached peptide chains has been demonstrated. Figure 3.26 shows a representative set of such templates. In all cases, the templates provide two reactive functional handles which can be used to attach N- and C-terminal polypeptide chain segments. The dimensions of the template are chosen such that the pendant antiparallel and parallel peptide chains can be brought into hydrogenbonding distance, which then facilitates the formation of artificial b-sheets. For this application, no effort is made to include the original side chains of the turn unit, which differentiates them from the b-turn mimics. The dibenzofuran scaffold with n ¼ 2 was more efficient in stabilizing a b-sheet structure of the appended peptide sequences than the shorter analogue with n ¼ 1, provided it was flanked by hydrophobic amino acids to form the necessary hydrophobic cluster [214] (Figure 3.26). Also, the 2,30 substituted biphenyl required the hydrophobic clustering to induce the formation of a b-hairpin [215]. This scaffold was used successfully in the design of folded peptides that have antiangiogenic activity [216] or which inhibit the dimerization of HIV-1 protease [217]. The

SECONDARY STRUCTURE MIMICS (EXCLUDING TURN MIMICS)

( )n

O

O

O

N H

( )n O

O

( )n

H N

N H

( )n dibenzofuran

77

N H 2,2'-substituted biphenyl

2, 3'-substituted biphenyl

diphenylacetylene

O O

O O N N H

N

N

HN type VI β-turn mimic

H

N

HN

N

HN

X

O

O

X

O

O O

O H O N

Pro-D-Pro

N

X = CH2: Nip-D-Nip X = O: di-oxaNip

diketopiperazine

Figure 3.26 Examples of aromatic-based and proline-based b-sheet nucleating templates

2,20 -substituted biphenyls were developed by Feigel as turn mimics in cyclic peptides [218]. The analogue with n ¼ 2 was recently shown to be capable of adopting an intramolecularly hydrogen-bonded turn conformation, in contrast to the shorter-chain (n < 2) analogues [219]. The stability of the hairpin conformation induced by the 2-amino-20 -carboxydiphenylacetylene scaffold was shown to depend on the relative configuration of the appended peptide strands [220,221]. Amino acids conjugated to the indolizidinone type VI b-turn mimic were shown to form interstrand hydrogen bonds, with a strong dependence on the sequence [222]. The bis-proline-derived diketopiperazine scaffold induces an extended conformation in a cyclic peptide containing the NPNA motif, which elicited antisera in mice against P. falciparum sporozoites [223]. The heterochiral dipeptide sequence L-Pro-D-Pro and the di-b-amino acid segment (R)-nipecotic-(S)-nipecotic acid (Nip-Nip) were more recently shown by Marshall and Gelman, respectively, to be very efficient turn inducers [224,225]. The dioxanipecotic acid (X¼O) segment was shown to lead to even more stable turn motifs than the dinipecotic acid (X¼CH2) segment [226].

78

ASPECTS OF PEPTIDOMIMETICS H N

H N

N

H N

N

O

H

N

O

N

O

O

N

O

O HN O O

N H O

N H R

N O HH O N N R H O

H N

H N HO N O

O R

O

H H

N

O

N

O

H

H N O

O

N OH N H

R

X = H, NO2 X

N

HN X

O

R

O

R

N

N

H

H

O

O O N H

O R'

N O R'

O CDHA-Gly linker

NH

N

H N

O

H

N H

O

D-Pro-DA DME linker

Figure 3.27 Further examples of b-sheet nucleating templates

In the bipyridine-based scaffolds (Figure 3.27), b-sheet formation is induced by addition of Cu(II) ions, which induces the cisoid conformation and brings the attached peptide strands into the required arrangement for the formation of interstrand hydrogen bonds [227]. In the endo-norbonene b-amino acid or dicarboxylic acid, the scaffold constrains the functional groups to an eclipsed conformation. The b-amino acid scaffold allows the attachement of antiparallel peptide strands. The NMR data are consistent with the formation of a b-sheet structure, but with a hydrogen bonding pattern that is different from the traditional parallel b-sheet [228]. On the other hand, the diacid norbornene scaffold can be used to attach parallel peptide strands. Because of the desymmetrization method used, the first amino acids attached to the norbornene scaffold needs to be proline. The resulting pseudopeptides using this norbornene diacid were shown to form intrastrand hydrogen-bonded b-turns rather than interstrand hydrogenbonded sheets. The built-in U-architecture of the norbornene ring system was shown to be required for the formation of the observed conformations. Using another desymmetrization method, resulting in the saturation of double bond in the norbornene, Bolm later demonstrated that when proline is replaced by phenylalanine, interstrand hydrogen-bonded structures are formed [229]. The benzo-substituted norbornane scaffold was proposed as a protein core mimetic (PCM). The aromatic ring in the PCM

SECONDARY STRUCTURE MIMICS (EXCLUDING TURN MIMICS)

79

scaffolds was shown to induce T-shaped aromatic–aromatic interactions that determine the conformation of the amino acids, which are attached in a parallel or antiparallel fashion [230]. Whereas oligomers of the b-amino acid trans-2-aminocyclohexane carboxylic acid adopt helical conformations, oligomers formed by alternating trans-1,2-diaminocyclohexane and cyclohexane trans-1,2-dicarboxylic acid have an extended conformation that is complementary to an extended tripeptide strand [231]. Gelman showed that the cis-cyclohexanedicarboxylic acid-Gly (CHDA) linker enabled parallel b-sheet formation in water between strands attached via their N-termini, but the linker was not a dominant driving force [232]. This linker complements the D-Pro-Dadme (1,1-dimethyl-1,2-diamino-ethane) linker, which allows parallel b-sheet formation between peptide strands that are attached via their C-termini [233]. The 2,8-diaminoepindolidione scaffold (Figure 3.28), developed by Kemp [234], was one of the first reported artificial b-sheet inducers. The peptide strands are coupled to this tetracyclic template by way of a turninducing D-Ala-Pro dipeptide. The appended peptide then forms a parallel artificial b-sheet by hydrogen bonding to the NH and CO of the template. When the peptide strand is attached to the Pro residue through a urea function as a chain-reversing element, an antiparallel b-sheet is formed. Ph N

O

O H N

N O

H

O

H N

O

H N

O

H N

N

H

O

HN H N

O

O

NH

O

O

O N

H N N H

O Ph

Figure 3.28 formation

N H NH

O

N H

N

O

HN

O

N

O

O

O

O H N

O N H

N

N

O

H O

H N

N O

N H

2,8-diaminoepindolidione scaffold and antiparallel or parallel sheet

A related 3,6-diaminoquinolone scaffold was developed, which, in combination with the D-Ala-Pro turn unit, induced a stable b-strand in a tripeptide sequence that was derived from ICAM-1 [235] (Figure 3.29). A methoxypyrrole amino acid (MOPAS) dimer, in combination with the

80

ASPECTS OF PEPTIDOMIMETICS

Pro-Gly turn unit, was shown to induce a b-strand conformation in the attached tetrapeptide sequence [236].

CH3

CH3

COOH

O O H3C

O

H N

N H

H N

N H

O

N

O

H

O

NH

Figure 3.29

O

H

O O

H N

Boc

H O

H

OMe

N

N H

O

O

N

N

N

O

HN

H N

H N

H N

N

O

HN

O

N H

O

O

3,6-diaminoquinolone and MOPAS sheet inducers

Using the same design principles, Nowick developed 5-aminoanisic acid amide as a b-sheet nucleator [237] (Figure 3.30). In combination with a urea scaffold, this b-strand mimic induced the formation of an antiparallel b-sheet. Further modifications include the use of the hydrazide and an oxalamide derivative of this b-strand mimic. When the oxalamide strand mimic Hao and a tripeptide were coupled to the urea scaffold, an intramolecular antiparallel sheet was formed that dimerized in chloroform to a well-defined b-sheet dimer.

NC O

O H N

H2N

N CH3 N

R N H

Ph

O H N O

O

O

N H N

CH3 H2N

O

NH2 Ph

H3C O

HO O

H N

N H

H N

N

O

O

O

Ph

CH3

N H

O

N

O

O

Hao

O

H N

NH2

R

1

R2 N H

O

O N

O

H

H N

O

R

3

N H

CH3

NC

Figure 3.30 Nowick artificial b-sheets

H N

N H

O

R

O N

O

CH3

N H

R′

CH3 O

CH3

H N

N H O

O

NC

CH3 O

CH3

N H

H N O

R'

O N H

CH3

O N H

CH3

SECONDARY STRUCTURE MIMICS (EXCLUDING TURN MIMICS)

81

The use of Hao in combination with a d-linked ornithine as a turn mimic provided models for intramolecular sheet formation, which were also shown to form intermolecular antiparallel sheets by dimerization. Since one edge of the sheet is blocked for further hydrogenbond formation, aggregation does not occur in these systems [237] (Figure 3.31).

Val

O N H O CH3

Phe

O

H N

N

O

N H

H N

Leu

O

H

O

N H

O

H

O

N

N H

O

H N

R1

O CH3

H3N

N

R3

O

H N

N

O

H

O

R2

H

O

H N

O

H

R6

O

H

O

O

N

N H

R7

O

H N

N R4

H

O

O

N

N H

R5

O

H N

H N H

O

N

NH3 O

CH3

Figure 3.31

Hao-induced b-sheets

Moreover, a combination of the Hao unit with two d-linked ornithines provided cyclic peptides that form an intramolecular b-sheet in water, independently of the sequence. These ‘cyclic modular b-sheets’ can be linked to form structures containing more than one b-sheet domain. Tuning of the lower strand allows the highest folding to be obtained and a b-strand structure of the upper pentapeptide strand to be formed [238]. When the 5-aminoanisic acid hydrazide was connected through an (S)-2-aminoadipic acid turn unit to a second strand of two N-terminally linked peptides, the resulting folded structure dimerized by parallel b-sheet formation [239] (Figure 3.32).

O

Leu H N

O

N

Me2N

Ile

H N

N

O

H

O

Leu

H

O

H

O

H

Val

O

H

N O

O

H N

N

N H

N O

O

N H

Bu

O

CH3

Figure 3.32

Hao-induced sheet formation in N-terminally linked peptide strands

82

ASPECTS OF PEPTIDOMIMETICS

It is interesting to note that Fmoc-Hao was shown to inhibit the dimerization of HIV protease [217].

3.4.1.2 Macrocyclic b-strand mimics b-hairpins are widely-occurring secondary structural elements in proteins, consisting of two adjacent strands of antiparallel b-sheet and a connecting loop. Cross-strand hydrogen bonding usually stabilizes b-hairpin structures. The shortest common loop involves two residues, in which case the loop and the two adjacent residues constitute a b-turn [240]. Based on the work of Marshall [224], showing that the D-Pro-L-Pro sequence (Figure 3.26) is a strong reverse turn inducer, the Robinson group has developed a strategy for the design of b-hairpin mimetic cyclic peptides by using this sequence to connect the N- and C-termini of the hairpin part of the native protein [223,241] (Figure 3.33). This sequence is a strong turn inducer and was shown to be able to maintain the b-hairpin conformation in many attached loops. This strategy allowed the preparation of b-hairpin mimetics as trypsin inhibitors [242], CXCR4 inhibitors [243], antimicrobial peptides [244], peptides binding to a human antibody [245] or inhibitors of the Tat–TAR interaction of bovine immunodeficiency virus [246]. Interestingly, the b-hairpin scaffold can be used to preorganize the side chains in a geometry similar to that seen in a helical peptide (see Section 3.4.2) [247,248]. Biaryl amino acid templates were shown to be less effective in promoting b-hairpin conformations than the D-Pro-L-Pro sequence, although the resulting cyclic analogues of protegrin 1 retained significant antimicrobial activity [249].

β-strand

β-strand

protein β-strand

β-strand

template

O N

O H

O N

NH

N O

Figure 3.33 (PEM)

Template-bound hairpin mimetic or b-hairpin protein epitope mimetics

SECONDARY STRUCTURE MIMICS (EXCLUDING TURN MIMICS)

83

In another approach, the peptide backbone is preorganized in an extended form by linking a side chain to the main chain, or a side chain to a side chain, thus forming a macrocycle [250]. The macrocycle prevents intramolecular hydrogen bond formation, and results in a b-strand conformation of the peptide backbone that is effectively recognized by proteases [251,252]. Moreover, such macrocycles are also selective toward other proteases, making them resistant to degradation. Naturally occurring macrocyclic enzyme inhibitors containing the amino acid isodityrosine as a linking unit are inhibitors of the metalloproteases angiotensin converting enzyme (ACE) or aminopeptidase N (APN) [250] (Figure 3.34). This has inspired the use of Tyr as a tool for forming macrocycles, resulting for example in powerful inhibitors of TNF-a converting enzyme (TACE) [253], HIV protease [254] and hepatitis C virus (HCV) NS3 protease, as shown in Figure 3.34 [87].

HO

OH

O

O

O

H N

N H

N H

O

O

O

H2N O

O

O

H N

OH

OH

N H

O

HOHN

N H

O

O

H N

O CH3

N H

O

H2N O

K-13, ACE OH

H N O

O N H

OH

N H

HIV protease

H N O

TACE

O

O

O

OF4949-IV APN

O H N

N H

H N

N O

O

O

H N

CH3

O

O

N H

N

CH3

O

O HCV NS3 protease

Figure 3.34 Macrocyclic protease inhibitors

Ring-closing metathesis (RCM) is also a powerful method for obtaining various macrocycles (Figure 3.35). Examples shown in Figure 3.35 include inhibitors of plasmepsin [255], b-secretase (BACE) [256,257] and HCV NS3 protease [258,259]. These examples convincingly demonstrate that macrocyclization of peptides is an effective strategy for obtaining b-strand mimetic protease inhibitors.

84

ASPECTS OF PEPTIDOMIMETICS

R

O N H

OH

O

O

OH

H N

N

H N

HNBoc O

OH

COOH

O

O

HCV NS3 protease plasmepsin CH3 O NH O

H N

O

S O

H N

OH

H N

HN O

O

O

N

O N H

O

Bn HN

N H

O BACE-1

Figure 3.35

H N O

BACE-1

Macrocyclic protease inhibitors obtained by RCM reactions

3.4.1.3 Backbone-modified strand mimetics A different approach to mimicking b-strands is to replace all or part of the peptide backbone with small carbo- or heterocyclic structures that have the appropriate geometry and are able to display the functional groups in approximately the same orientation as in the native b-strand (Figure 3.36). Pioneering work has been done by Smith and Hirschmann [260], who developed polypyrrolinone b-strand mimetics. This scaffold was used for the preparation of inhibitors of HIV protease [261], matrix metalloproteases [262] and a hybrid peptide ligand for the MHC protein HLA-DR1 [263]. The heterochiral pyrrolinone stereoisomer was shown to adopt a turn conformation, and provided a somatostatin mimetic with mM potency [264]. The @-Tides framework was proposed by the Bartlett group [265]. In linear oligomers, this unit was shown to adopt a b-strand conformation and to dimerize. It also induces b-hairpin folding in attached peptide squences [266]. Peptidomimetics incorporating this @-Tides unit have been shown to be potent ligands to a PDZ protein-interaction domain, with high resistance to protease-mediated degradation [267]. Recently, the aza-@-Tides unit has been developed. It allows the ready incorporation of a side chain, starting from an amino acid precursor [268]. A Leu-Gly-Gly b-strand tripeptide mimetic, based on a pyridine scaffold (R1 ¼ iPr, R2, R3 ¼ H), was recently reported [269].

SECONDARY STRUCTURE MIMICS (EXCLUDING TURN MIMICS) R2

O

H N

H

N H

R1 H

O H N R3 H

O

R4 N H

R2

O

H

H N Ac

O H N

H N

OMe

R1 H

O

85

R3 H

N

β-strand conformation

H

O R R1

O

O

R4

N

2

H O

R3

N

O

H

N

O

H

OH

H N

HN NH2 HN

O

O

O

polypyrrolinone β-strand mimetic HIV protease inhibitor O H N

O

H N

N

H N

N

R1 H

R2

O

H N

R3 H

H

N

R1 H

O

O

H N

N

O

R3 H

O

dihydropyridinone oligomer: @-Tides

dihydropyrazinone unit: aza-@-unit

Figure 3.36 Backbone-modified b-strand mimics

A variety of nonoligomeric scaffolds have been used in the design of enzyme inhibitors that bind in an extended conformation [210]. These include aromatic or heteroaromatic scaffolds such as the ones shown in Figure 3.37, but also many saturated heterocycles, including lactams (Figure 3.38). HS HS

O

H N

H2 N

N H

O

CH3

S

H N H2N

COOH

O

FTase

O

FTase

COOH

O N H

ICE

O

O

N O

COOH

N H

O

O

O

S

H Cl

N H

N H

N O

N H

CF3 O

HLE

Figure 3.37 Inhibitors of farnesyl transferase (FTE) [270,271], interleukin-1b convertase (ICE) [272,273] and human leucocyte elastase (HLE) [274] based on aromatic or heteroaromatic scaffolds

86

ASPECTS OF PEPTIDOMIMETICS

N

N H

HOOC

COOH

O

S O

O

ACE inhibitors

N

H N H

NH

H

N H

O

O

H

O

N O

N H

NH2

thrombin inhibitor

X H2N

N

O

( )n

O

O

S

HS N S

N

N H O

COOH

X = CH X=N NH

thrombin inhibitor

ACE/NEP inhibitor

HN NH2

Figure 3.38 Inhibitors of ACE, thrombin, ICE and HLE based on saturated lactam scaffolds

Saturated lactams such as those found in the ACE (n ¼ 1  5) or thrombin inhibitors (Figure 3.38) belong to the class of so-called Freidinger lactams [275], which were frequently claimed to be b-turn mimetics [276]. However, it was shown that when bound to the enzyme, such inhibitors adopted an extended conformation [210]. Interestingly, another group of compounds claimed to be b-turn mimetics, the azabicyclo-alkanone lactams [277], have also been used successfully to prepare enzyme inhibitors that bind in an extended form to the enzyme, as exemplified by the thrombin and ACE/NEP inhibitors [278,279] (Figure 3.38).

3.4.1.4 Topomimetics The Hamilton and the Mayo groups proposed substituted calixarenes as topomimetics of a b-sheet as well as of an a-helix [280–282]. The molecular dimensions of a calixarene correspond approximately to

SECONDARY STRUCTURE MIMICS (EXCLUDING TURN MIMICS)

87

those of a segment of a b-sheet about three residues long on each strand, but also to about two turns of an a-helix. Moreover, the calixarene scaffold allows the attachement of various substituents, which allow creation of a hydrophilic and a hydrophobic surface, corresponding to those found in amphipatic sheets or helices. Topomimetics of the antiangiogenic peptide anginex, which adopts an antiparallel b-sheet, were developed (Figure 3.39), as were those of SC4, an a-helix-forming bactericidal peptide, which displayed high potency [280,281]. Calixarenes functionalized with four cyclic peptides containing the Gly-Asp-Gly-Tyr sequence were shown to have nM potency in inhibiting PDGF (platelet-derived growth factor)-induced PDGFR autophosphorylaton [282]. The analogue with Gly-Asp-Gly-Asp sequence was able to block cytochrome c from binding to cytochrome c peroxidase [283]. R'

R'

H N NMe2

R' = H, R =

K

K

K

R'

R'

O K

R O R

O

O

R

R

R' = tBu, R =

O R

H N

NH2 NH

Figure 3.39 Topomimetics of a b-sheet

3.4.2

Helix Mimetics

Helices are the predominant protein secondary structure. They are important recognition motifs for protein–protein and protein–nucleic acid interactions. Short peptide sequences, corresponding to these recognition motifs, usually do not adopt a helical conformation. Several strategies have been developed to stabilize short peptides in a-helical conformations, which include the use of unnatural amino acids such as the a,a-dialkyl amino acids, and the introduction of noncovalent and covalent side-chain constraints [284–286]. Also, foldamers, consisting of oligomers of b, g, d-amino acids, vinylogous g-peptides or peptoids were shown to form various kinds of helices [188,287,288]. Several helix nucleating scaffolds and nonpeptidic scaffolds and modified peptide backbones that mimic the recognition properties of a-helices have been shown to lead to compounds with important biological properties. These approaches have been reviewed recently [212,224,282,286,289,290].

88

ASPECTS OF PEPTIDOMIMETICS

3.4.2.1 Helix nucleators As for the b-sheet nucleators, templates have been designed in which the orientation of the NH or CO groups is fixed in a rigid structure in order to initiate helix formation [291] (Figure 3.40). The Kemp group was the first to propose a helix scaffold that positions three carbonyl groups in an orientation that allows them to form hydrogen bonds with the amide NH protons of the attached peptide, thereby inducing a distorted a-helix [292]. Satterthwaith introduced the concept of a hydrogen-bond surrogate (HBS) in his template, which mimics one turn of an a-helix, and uses a hydrazone-ethylene bridge as a covalent replacement of the hydrogen bond [293]. A solid-phase procedure to assemble this helix nucleator has been developed [294] and was applied to the development of a restricted peptide derived from the human papillomavirus that was recognized by sera from women having cervical carcinoma [295]. This concept was recently further developed by using an allyl group as the HBS. This helix-nucleator can be very conveniently prepared by metathesis reaction of a bis-allyl precursor [296–298] and induces a-helix formation in short peptides, as confirmed by X-ray crystallography [299–301]. A constrained Bak BH3 peptide sequence was shown to bind to the Bcl-xL protein and to be resistant to trypsin cleavage [302]. An HBS-based artificial helix inhibited the gp41-mediated HIV-1 fusion [303]. O OH O O

H O

N

H

N

N

O

HN

Ri+2 HN

O

O

Ri+1

HN CH3

s

O

R

CH3

Ri

Satterthwait HBS template

Kemp template

O

O

N HN

N

N

s

R

O

O

N H

O HN N

R

O

N

Ri+3 N

N

H N

O

H N

O

O

N

O

Ri+4

O

N

O O

N H R

O

O H

R R

H

H3C

HN

O

N

O O

COOH N H

CH3

H3C

O

N

R

HN H O

Cbz-HN

H N

H3C N

NH-Boc

OH

CH3

O

Bartlett template

R allyl-based HBS template

Bartlett template

Figure 3.40

N H

CH3

Helix-inducing templates

SECONDARY STRUCTURE MIMICS (EXCLUDING TURN MIMICS)

89

The Bartlett group described the hexahydroindol-4-one diacid and the diamino-substituted diazabicyclo[4,3,0]nonene amidinium salt as helix nucleators [304,305]. In the latter, the three NH hydrogen-bond donors are positioned in an ananalogous arrangement to the three carbonyl group hydrogen-bond acceptors of the N-terminal diacid template and therefore could mimic the hydrogen-bonding pattern of the first turn at the C-terminus of an a-helix. This remains to be demonstrated, however.

3.4.2.2 Helix surface mimetics A number of approaches to develop nonpeptide a-helix mimetics have been described. Side chains along one face of a helix are often responsible for critical interactions with the target proteins. 1,1,6Trisubstituted indanes reported by Horwell and coworkers are used to represent amino acids at the i and i þ 1 positions of an a-helix (Figure 3.41). The designed mimetics of dipeptides (Phe-Phe and TrpPhe) using this scaffold show micromolar affinity similar to the original dipeptides, in an attempt to mimic tachykinins and other neuropeptide targets [306]. However, the indanes can mimic only two amino acids, and therefore are not suitable for covering more residues in a helix. Jacoby suggested 2,6,30 ,5,-tetrasubstituted biphenyl analogues to mimic one helical turn [307]. Hamilton and coworkers have reported scaffolds which can represent two helical turns. Trifunctionalized 3,20 ,200 -terphenyl derivatives hold side chains located at the i, i þ 3 or i þ 4, and i þ 7 positions [224] in two helical turns. When fitted with the appropriate side chains, these helix surface mimics were found to effectively disrupt the interactions between calmodulin and smooth muscle light-chain kinase [308], between Bcl-xL and Bak proteins [309,310], and between the p53 protein and MDM2 [311], and also to inhibit the self-assembly of a six-helix bundle gp41 core [312], demonstrating the potential use of a-helix mimetics as therapeutics. In order to be able to exploit additional interactions by residues adjacent to those forming the hydrophobic face of the helix, the Hamilton group designed the 4,7-diphenyl-substituted indane scaffold, which can display the i, i þ 3, i þ 4 and i þ 7 side chains, thereby mimicking two faces of a helix [313]. A tris-pyridylamide scaffold was also designed to display side chains found at the i, i þ 4 and i þ 7 positions, in order to inhibit the formation of a Bak BH3/Bcl-xL complex [314]. Recently, Ahn and coworkers

90

ASPECTS OF PEPTIDOMIMETICS

developed a tris-benzamide scaffold with substituents that mimic the i, i þ 4 and i þ 7 positions of an ideal a-helix. This scaffold is easy to synthesize and possesses a slightly higher conformational flexibility than the corresponding tris-pyridyl amide, due to the lack of a hydrogen bond between the amide and the nitrogen on the pyridine ring. According to the authors, this results in a less flattened structure, which gives a better topological correspondence with the a-helix [315]. The terphenyl scaffold has recently been replaced by a terephtalamide scaffold, which has an increased water solubility and membrane permeability [316]. Antagonists of the Bcl-xL/Bak protein interaction with mM potency were obtained.

COOH

CH3

COOH

O

O Ri

i

O Ri

Ac

O

O

HN

H

N

CH3

O

N

Ri

O

R

i+3 i+4

Ri + 3

R i + 1/ i + 4

Ri + 1

N

O

Ri + 4

Ri

i

O

H

H

N

O N

R

i+7

i+7

O

Ri + 7 N

O

COOH

COOH

O

N

Ri

Ri

Ri

N N

Ri + 4

H N

R

i+4

O

N

R

N

N

R

O

i+3

Ri + 4 N

N

Ri + 7 N

pyridazin

N

N

R

N Ac

pyrrole

Figure 3.41

Ri + 7

i+7

Ri + 7 N

Ac enaminone

N R

HN

Ri

O i+7

CH3 terephtalamide

Ri + 3

Ri + 3

CH3

Ri

O

N

O

tris-benzamide

N

N Ri + 7

O

Ri + 7

Bn

Ri

Ri

HN i+3

N

O Ri + 7 O

N

H

Ri + 7

HN H tris-pyridylamide

H O

N

O O

Hamilton terphenyls

HOOC

O

H

N

Horwell indane

Ri + 4

Ri + 4

Bn dipiperidinobenzene

N

terpyridyl

Helix surface mimetics

Considering that six-membered ring hydrogen bonds can impart structural rigidity, Hamilton recently designed an enaminone helix mimetic. This scaffold can be considered as a simplified terphenyl scaffold in which the central phenyl ring has been replaced by a more polar ring [317]. As alternatives for the terphenyl scaffold, with better

SECONDARY STRUCTURE MIMICS (EXCLUDING TURN MIMICS)

91

synthetic accessibility and an amphiphilic structure with a hydrophobic surface for recognition along one side and a polar edge for solubility along the other side, Rebek designed pyridazine- and pyrrole-based scaffolds [318,319]. These scaffolds can be assembled in a modular way and should allow the targeting of a range of protein–protein interactions. A similar modular assembly was used for the preparation of 1,4-dipiperidino benzenes [320]. An X-ray crystal structure of the mimetic revealed a good similarity between the side chain orientations and that in an a-helix. In a recent study, Marshall computationally evaluated various helix mimetic scaffolds and found that the terphenyl scaffold is not rigid but can adopt 16 conformations with almost equal energy. Various other a-helix mimetics, such as the terpyridyl scaffold (Figure 3.41), that should be more effective than the terphenyl scaffold were proposed, based on theoretical calculations [224,321]. Trisubstituted imidazoles, which can be prepared by van Leusen multicomponent reactions, were designed as alternatives to the Hamilton terphenyl mimics (Figure 3.42). A good correspondence with the i, i þ 3 and i þ 7 positions in an a-helix was demonstrated by molecular modelling [322]. Selective inhibitors of the Bcl-w/Bak-BH3 interaction with mM potency were prepared. Ri

Ri + 3

H

N Ri + 3 N

H O

N ( )n

imidazoles

N

O

O

H

NH

HN NH COOBn O

H

H O polyether

Ri + 8

N

N

i+3

N

H N

O

O Ri+ 7

N

Ri + 4

H

i O

O

O

H

NH

N N H

H

Ri + 7

Ri

H O

N

( )n

O

RO

Ri

O

i=4

O

bis-imidazole

triazine

Figure 3.42 Helix surface mimetics

In addition, ladder-like polyethers were developed by Hirama and coworkers to mimic i, i þ 4 and i þ 8 positions in a helix [323,324]. A different approach was followed by Todd and collaborators, who proposed a bis-imidazole scaffold in which the amino acids attached to the imidazoles mimic the residues in the i and i þ 4 positions of the helix (Figure 3.42). The compound was shown to block HCV-E2 binding to CD81 in mM concentration [325].

92

ASPECTS OF PEPTIDOMIMETICS H - Ala - Gly - Cys - Lys - Asn - Phe - Phe - Trp HO - Cys - Ser - Thr - Phe - Thr - Lys

somatostatin-14

OH O

O

H3C N H

H3C N O

H N

H N H2N O

O O

O N H

HN

H N

O S

H N NH2

O

HO

O

O

O

HN

H N

N H

OH

MK-678,Seglitide

Figure 3.43

N H

H N

N H

S

OH

O

N H

NH2

Octreotide

Somatostatin-14 and reduced-size cyclic analogues

A 13 amino acid peptide binds to the oestrogen receptor in a helical conformation. A head-on view of this coactivator peptide shows three key leucine residues whose Ca carbons are rougly positioned at the corners of a triagle. A triazene scaffold was found to best mimic this arrangement, and provided a coactivation binding inhibitor with mM potency [326] (Figure 3.42). As mentioned in Section 3.4.1.2, cyclic peptides with b-hairpin scaffold Pro-D-Pro (Figure 3.33) were shown to mimic helical conformations of the p53 peptide (see Figure 3.53) and of a Rev helical peptide [247,248]. Various nonpeptide helix mimetics have been discovered, mainly by high-througput screening. A selection is shown in Figure 3.54, as ihibitors of the p53/HDM2 interaction.

3.5

EXAMPLES OF PEPTIDOMIMETICS

The strategies that were discussed in the previous sections have been applied to a variety of bioactive peptides, and have resulted in a large number of modified peptides and peptidomimetics. Moreover, intense screening efforts, followed by lead optimizations, have revealed a wealth of new nonpeptide lead structures for peptide receptors [327]. For many years, it was thought that the peptidomimetic approach would mainly result in inhibitors of enzymes and antagonists of peptide receptors. However, it is now clear that peptide mimicry by nonpeptides has also

EXAMPLES OF PEPTIDOMIMETICS

93

resulted in numerous agonists. The question ‘Can peptides be mimicked?’ [328] has certainly been given a positive answer in recent years. The subject of nonpeptidic ligands for G-protein-coupled receptors and for disrupting protein–protein interactions has been the topic of recent comprehensive reviews [212,224,282,289,327,329]. Some selected examples which illustrate the power of the strategies discussed in the previous sections will be discussed here. One of the best examples of the hierarchical approach to peptidomimetic design is the case of somatostatin (or somatotropin release-inhibiting factor, SRIF), where almost all of the previously discussed strategies have been applied [330]. SRIF is a cyclic tetradecapeptide for which five subtype receptors have been identified. Systematic studies resulted in the development of the cyclic octapeptide octreotide (SMS 201–995) by Sandoz and of the cyclic hexapeptide MK-678 (Seglitide) by Merck [330,331] (Figure 3.43).

N

O

HN

NH

O

O

O

O

O O

O

NH

O O

O

NH

O O O

NH2

N

NH2

NH2

HO NH

HO

HO

NH

N

HO

N

BnO

O NH2

BnO

HN

NH2

Figure 3.44 Glucose-, catechol and iminosugar-based somatostatin mimetics

A large number of analogues of these cyclic peptides have been prepared, including a retroinverso cyclohexapeptide with 25% of the potency of somatostatin [332] and N-peptoid analogues [333], resulting in a spectrum of affinities and selectivities for the five subtype receptors. All these studies indicated that the Phe-D-Trp-Lys-Thr tetrapeptide fragment was the essential pharmacophore. Conformational studies have shown that this fragment was maintained in a type II0 b-turn conformation by the remaining part of the cyclic

94

ASPECTS OF PEPTIDOMIMETICS

peptides. This 3D arrangement of the D-Trp and Lys residues has been the basis for the design of a number of scaffolds that mimic this arrangement of the critical side chains (Figures 3.44–3.50). The design of the somatostatin mimetic, based on the b-D-glucose scaffold by Hirschmann and coworkers, was a landmark in this field [334] (Figure 3.44). The structure–activity relationships in this series correlates with the one found for the cyclic peptides. Unexpectedly, these glucose-based mimetics also showed affinity for the NK1 receptor [335,336]. Replacement of the 2-benzyl substituent by an imidazol-4ylmethyl group improved affinity, and further incorporation of a pyridyl3-ylmethyl substituent at position 4 resulted in a potent analogue that was also selective for the sstr4 receptor, and was no longer recognized by the NK1 receptor. Iminosugar-based scaffolds provided low-affinity analogues [337,338]. More recently, catechol was used as a minimal scaffold to display three of the somatostatin pharmacophores, resulting in mM affinity for sstr2 and sstr4 [339]. The compounds shown in Figure 3.45 all contain scaffolds that were designed as mimics for the b-turn conformation of the cyclic peptides, and display the important Trp and Lys pharmacophores and one or two additional aromatic groups [340–344]. Whereas most of these mimetics showed mM affinities, the cyclic thioether analogue was shown to be a sstr5-selective ligand with nM potency [343].

NH NH2

NH2

O H N

N

N N

N N

O

NH

O

O

O HN

O

O HN

O

NH2

O

NH

NH

N

S

H3C N

O HN N

O

N

O O O

NH2

NH2

Figure 3.45 Molecular scaffolds in somatostatin peptide mimetics

EXAMPLES OF PEPTIDOMIMETICS

95

A 3D similarity search of the Merck compound collection using the Tyr-D-Trp-Lys motif, and biological testing of only 75 compounds, yielded compound L-264,930 with 100 nM affinity for the human sstr2 [345,346] (Figure 3.46). Further screening of combinatorial libraries, using the privileged scaffold concept [347], led to highly active and selective agonists for each of the five sstrs [348] (Figure 3.46).

NH

NH

O

NH2

HN

NH2

HN O

N

O

O

HN

N

N H

N

COOtBu

NH2

HN O

N

N L-264,930 lead

HN

HN O

L-054,522 sstr2

O

L-054,522 sstr1

F H N

O NH2

HN HN

HN O O

F O

COOtBu

O

HN L-796,778 sstr3

H N

HN NO2

Figure 3.46

L-803,087 sstr4

NH2

HN N H

HN NH2 O

COOMe NH2

O

L-817,818 sstr5

Subtype-selective somatostatin agonists

The introduction of the cyclic constrained Trp analogue Aia (see Figure 3.15) in an analogue based on the L-054,522 structure recently led to a potent ligand for all receptor subtypes, except for sstr1, and further modifications of the constrained dipeptide analogue resulted in a potent sstr5 agonist (0.6 nM) [122,349] (Figure 3.47). A screening programme based on a scaffold containing two aromatic groups and a basic group identified NNC 26-9100 as an sstr4-selective agonist [350]. Similarly, screening of a 700-membered library of 3-thio-1,2,4-triazoles with an indole and an alkylamine substituent identified selective agonists for sstr2 and sstr3 [351]. The 4,1-benzoxazepine analogue was reported in the patent literature as a potent sstr5-selective agonist [331,352]. Small modifications of the Merck compounds resulted in the conversion of agonists into antagonists. A change in the N-terminal substituent gave sstr2-selective antagonists [353] (Figure 3.48). A chemogenomics approach, using the H1 antagonist astemizole as the lead structure, led to

96

ASPECTS OF PEPTIDOMIMETICS

the development of the first nonpeptidic sstr5 antagonist [354]. The screening of an imidazolyl library led to a tetrahydro-b-carboline analogue antagonist that was selective for the sstr3 receptor subtype [355].

NH

NH

O

O N

N H O

NH2

N

NH

O

N H

N HN

N O

NH 2

O

N H

O sstr 2-5

sstr5

Ph Br

F

O

N

Cl N

HN

H N

H N

N

Cl

Cl

S

NH

O O

H2N

NNC-26-9100 sstr4

sstr5

H 2N

NH N N

N

S N

O

sstr2/3

Figure 3.47 Somatostatin mimetic agonists

Seebach used the propensity of b2/b3-dipeptide and g-dipeptide combinations to adopt turn structures and to develop derivatives with nanomolar affinity and selectivity for some of the hsstrs [356–358] (Figure 3.49). The observation that some of the glucose-based somatotatin mimetics (Figure 3.50) also showed affinity for the NK1 receptor [335,336] led the Hirschman group to develop a cyclic hexapeptide NK1 antagonist, based on the somatostatin scaffold [359], which is entirely different from the approach that led to nonpeptide NK1 antagonists [327]. A similar cyclic b-turn scaffold is also present in an NK2 antagonist [360].

EXAMPLES OF PEPTIDOMIMETICS

97

NH

NH O

O

H N N H

N

NH2 O

O

S

O O

H N

N H

NH 2

O

N

O

O

O

sstr2

sstr2

N

H N

O

HN NH

N H

O

NH2

N H2 N

sstr3

N O

S O

Et

O

Et

sstr5

Figure 3.48 Somatostatin mimetic antagonists

H N

H2N

NH

O

O H N

Ph

HN

O

O

HN

O

H N

H N

NH2

NH2

H O

NH

H N

O OH

OH

H N

O

NH

O

HN

O

H N NH2 O

Figure 3.49 Somatostatin mimetics based on b- or g-homo amino acids

The structural similarity of the somatostatin pharmacophore with that of urotensin (U-II) stimulated the application of the established somatostatin SAR to U-II. U-II is a cyclic undecapeptide with potent vasoactive properties (Figure 3.51). The endocyclic residues were shown to be critical for biological activity.

98

ASPECTS OF PEPTIDOMIMETICS

H N

O Bn

N H

NH

O

N

Bn Bn

H N

O

O

O

O

O O

O O

O

H N

N H

NHAc

HN

NH

O O

F

N H

HN

H N

N H

O

O NK-2 antagonist

NK-1antagonist

Figure 3.50 Use of the somatostatin template to design an NK1 antagonist

H - Glu - Thr - Pro - Asp - Cys - Phe - Trp HO - Val - Cys - Tyr - Lys

H - Asp - Pen - Phe - Trp

H - Asp - Pen - Phe - D -Trp

HO - Val - Cys - Tyr - Lys

HO - Val - Cys - Tyr - Orn

urotensinII

Urantide

P5U

O O

O H3C N H

HN

O

N

NH

O

O H3C

H N

O

O N H

N H

HN

H N

N

NH Cl

AC-7954

NH2

NH2

O

S6716 O

O Cl

N

H N

N H

N O

O

NH Cl N

N Boc

Figure 3.51

Urotensin peptide analogues and mimetics

The cyclic hexapeptide Ac-c(Cys-Phe-Trp-Lys-2-Nal-Cys)-NH2 was identified as a potent U-II agonist. Further conformational constraint by substituting Cys with Pen resulted in the highly potent octapeptide agonist (P5U) Asp-c(Pen-Phe-Trp-Lys-Tyr-Cys)-Val-OH [361], and in the antagonist urantide Asp-c(Pen-Phe-D-Trp-Orn-Tyr-Cys)-Val-OH [362], which were shown to adopt a turn conformation. Therefore, the straightforward use of the somatostatin strategy resulted in the development of a cystine-free cyclic hexapeptide agonist for the U-II receptor [363].

EXAMPLES OF PEPTIDOMIMETICS

99

The U-II pharmacophore was used by several companies to screen libraries, resulting in various nonpeptide U-II agonists and antagonists, which were recently reviewed by Blakeney and Carotenuto [327,364]. One of the first reported examples used the three-point pharmacophore model, composed of the Trp8, Lys9 and Tyr10 side chains, to screen the Aventis compound collection [365]. Compound S6716 was identified as a potent U-II receptor antagonist. Some more recent examples are included in Figure 3.51. The first reported nonpeptide agonist was the isochromanone AC-7954. Its structure served as lead for the development of the more potent 4-phenylbenzamide [366]. The Johnson & Johnson group idendified a phenylpiperidine-benzoxazinone lead from a library screening. It was optimized to yield a low-nanomolar antagonist against both rat and human U-II receptors [367]. Many of the concepts that were discussed in the previous sections have been applied to the design of inhibitors of the interaction between the tumour-suppression gene p53 and the human or murine double minute-2 (HDM2 or MDM2) oncogene [368–370]. The p53 protein regulates cell proliferation by induction of growth arrest or apoptosis in response to DNA damage or stress stimuli. The HDM2 protein binds to the transactivation domain of p53 and downregulates its activity. The disruption of this protein–protein interaction has been intensively studied as an approach for cancer therapy, because it would allow the upregulation of the p53 response. Binding to the MDM2 protein occurs through the N-terminal transactivation domain of p53. A crystal structure of the complex between MDM2 and a 15-residue sequence of p53 revealed a narrow hydrophobic cleft in the MDM2 protein to which three hydrophobic side chains of the p53 sequence, Phe19, Trp23 and Leu26, make direct contact. These three residues are aligned along one face of the amphipatic a-helix that is adopted by the p53 N-terminal domain. This discovery stimulated research toward peptide mimetics that display three hydrophobic groups in an orientation that mimics their presentation in the a-helix. A team at Novartis screened peptide libraries, obtained via phage display, and identified a 12-mer peptide with a greater inhibition than that of the wild-type p53 sequence [371]. Truncation of this peptide sequence revealed an octapeptide sequence Ac-Phe-Met-Asp-Tyr-Trp-Glu-Gly-Leu-NH2 as the minimal sequence retaining micromolar affinity for HDM2. Exploiting the propensity of a,a-disubstituted amino acids to induce helical conformations, the Asp and Gly residues were replaced by a-aminoisobutyric acid (Aib) and 1-aminocyclopropanecarboxylic acid (Ac3c) residues, respectively, resulting in an increased affinity. Subsequent replacement of the Tyr

100

ASPECTS OF PEPTIDOMIMETICS

residue by phosphonometylphenylalanine (Pmp) to make an electrostatic contact to the E-amine of the MDM2 Lys94, and of Trp with 6-chloro-Trp, resulted in a peptide with low nanomolar affinity (Figure 3.52). A recent crystallographic analysis of the co-complex of this peptide and MDM2 confirms that the 6-chloro substituent of the Trp residue fills a hydrophobic pocket, but also indicates the absence of a salt bridge involving the Pmp residue, which is exposed to the solvent [372]. b-peptides, composed of b-homoaminoacids, were shown to adopt helical structures. Seebach showed that b-peptides composed of b3-homo aminoacids adopt a 14-helix, while Gellman demonstrated that the use of the five-membered ring constrained trans-2-aminocyclopentanecarboxylic acid (ACPC) or trans-3-aminopyrrolidine-4-carboxylic acid (APC) leads to the formation of a 12-helix. Based on this concept, b-peptides were designed to display the key Phe, Trp and Phe side chains along one side of the helix structures. The Gellman group developed a 12-helical b-peptide inhibitor, which was shown however to have a low potency [369]. In contrast, 14-helical b-peptide inhibitors with low mM affinity were obtained by the Shepartz group [373,374] (Figure 3.52). Starting from the wild-type p53 N-terminal sequence, Verdine stabilized the a-helix (SAH) conformation by using a hydrocarbon stapling strategy. In this strategy, side chains of the i and i þ 7 residues are connected by an all-hydrocarbon linker generated by olefin metathesis [375]. SAH-p53 sequences were obtained with high binding affinity for HDM2 and good cellular uptake, with the ability to increase p53 levels in these cells. The retroinverso and N-peptoid concepts have also been applied to the design of analogues of the p53 peptide sequence. A retroinverso allD peptide was obtained with a potency comparable to that of the natural peptide sequence. This shows that, despite the fact that the nitrogen and carbonyl groups in the amide bonds were reversed, that the interresidue hydrogen bonds point in opposite directions, that the chirality was inverted and that D-peptides preferentially adopt lefthanded helices, these retroinverso peptides can mimic a natural righthanded helix [376]. The Apella group designed N-peptoids with chiral substituents to induce a helical conformation, and with the Phe, Trp and Leu side chains. These peptoids did not however interact with HDM2. In contrast, achiral residues increased the binding affinity, and introduction of the 6-Cl-substituent on the indole increased affinity only slightly. The importance of the nitro substituents in the peptoid suggests that interactions outside the hydrophobic cleft of the MDM2 protein can be important [377].

EXAMPLES OF PEPTIDOMIMETICS

101

O HO P

O

O

H N

N H

O

HO

S

O

H N

Pmp

OH

H3C

O

O

H N

N H

O

O

H N

N H

NH2

N H

O

O

Ac3c

Aib HN (6-Cl)Trp Novartispeptide

Cl

O O S

HN

O

O N H

H N

N H

N H

COOH

N

O

O

O

O

NH2

O

N H

NH

N H

O S O O

HOOC

N H

Gellman 12-helical β-peptide

H2N

COOH O

H2N

O

O

O N H

N H

H2N

HN

COOH

O

N H

O

N H

N H

N H

O

O

O N H

N H

OH

Shepartz 14-helical β-peptide

COOH O

H N

H2N

N H

O

O

H N

H N

N H OH

O

HN

HOOC O

HO

N H

O

O

H N

N H

O

O

H N

N H

O

O

H N

N

HO CONH2

p53 15-mer peptide

NH2

COOH

O

HN

HOOC

HO H2N

N H

O

H N

O

H N

N H

O

O

H N

N H

O OH

N H

O

O

H N

N H

O

O

H N

N H

O

O

H N

N O

O CONH2

retroinverso all-D peptide OH O P

O2N

O2N

O N

HN

NH2

OH

O N

N

O

O N

N

O

O

O

P O OH

N

N

O

HO

HN

O2N

O N

N

NH2

O H2N

S O O

Cl N-peptoid

Figure 3.52

CONH2 O

H N

HO

CONH2

O2N

COOH

N H

O

O

CONH2

O

CONH2

O

H N

Inhibitors of the p53/HDM2 or MDM2 interaction

N H

102

ASPECTS OF PEPTIDOMIMETICS

Robinson noted that the distance between the Ca atoms of the Phe and Trp residues in the p53 helix is close to that between the Ca atoms of residues i and i þ 1 in a b-strand. Therefore, the three key residues were

NH O HOOC

H N

O N H

O

Glu

O

N

N H

O

NH O

Glu

H N

N H

O

O

O N H

N

H N O

HN

Cl

Figure 3.53

Robinson hairpin helix mimetic

incorporated into a designed cyclic hairpin mimetic (see Section 3.4.1.2) using the D-Pro-L-Pro turn-inducing template [248] (Figure 3.53). Lead optimization, involving the use of the 6-ClTrp residue, resulted in a 1000-fold affinity increase. The X-ray structure of the hairpin mimetic– HDM2 complex confirmed the binding of the Phe1, (6-Cl)Trp3 and Leu4 side chains in the first strand into the hydrophobic cleft of the HDM2 protein, but also indicated the presence of additional interactions on the side of the cleft, involving the Trp6, Phe8 and Asp5 residues of the second b-strand [241,248,378]. Various nonpeptide mimetics have been shown to be able disrupt this protein–protein interaction, and thereby qualify as proteomimetics [368,369]. The Hamilton group used their concept that a terphenyl scaffold can mimic one face of a a-helical peptide, and showed that the terphenyl with hydrophobic substituents intended to mimic the three key p53 binding groups was able to disrupt the p53/HDM2 complex [311] (Figure 3.54).

EXAMPLES OF PEPTIDOMIMETICS COOH O

103

O

H N

O

H N

N

O

N

OH

O

O

O

HN

N

NO2

O

H3C N N

COOH O Cl

O

O

H N

NH

N

O

N Cl

N O N

Cl

N

I

COOH

O

CH3

N

I O

NH2

O Cl Cl

nutlin-3

Cl

CH3

Cl O

N

N Cl

CH3

F O

NH

HO O N

NH

OH

NH OH

O N H

Cl

N

N H

Cl

S

N

O

MI-63

O

H2N CN

NH

NH

OH N

H2N

O

NH

NH

F NSC66811

Figure 3.54

COOH

Peptidomimetic inhibitors of the p53–HDM2 interaction

A variey of small-molecule inhibitors were discovered as a result of screening programmes, some of them including preceding virtual screening (Figure 3.54). Among the first-reported small-molecule antagonists were the tryptophyl-piperazine and the indole-substituted tryptophans [379,380]. In the latter, the 2-phenoxy substituent was intended to target the p53 Phe19 binding site, and the substituted Trp to target the Trp23 site. The indole N1 substituent was intended to target the Leu26 binding site. A series of imidazoline compounds named nutlins were reported by Hoffmann-La Roche [381]. They showed activity in the 100–300 nM

104

ASPECTS OF PEPTIDOMIMETICS

range. The binding mode to HDM2 was determined by X-ray crystallography. The halophenyl rings were shown to occupy the Trp23 and Leu26 pockets. The piperazine substituent is exposed to the solvent and serves as a solubilizing moiety. The 1,4-benzodiazepine-2,5-dione scaffold as p53-HDM2 antagonist was discovered by library screening [382]. Optimization of the lead provided submicromolar inhibitors. Crystallographic analysis showed that the three substituted phenyl rings of the benzodiazepine bind to the same pockets as the Phe, Trp and Leu side chains of the peptide inhibitors [382]. Further studies established the importance of the stereochemistry and led to the removal of the carboxylic acid moiety, the introduction of an amine functional group in the ortho position of the benzylic ring to form an additional hydrogen bond with Val93 of the protein, and the introduction of a 1-methyl-4propylpiperazine solubilizing group. The resulting benzodiazepinone analogue was shown to be active in cell-based assays [383]. A structurebased design came next, starting from the spirooxindole core of the natural alkaloid spirotryprostatin A as a Trp replacement, followed by the introduction of substituents to provide the required hydrophobic interactions with MDM2, resulting in the spirooxindole inhibitor MI63 with low mM potency in cellular assays [384]. Further optimization yielded a compound with 3 nM affinity for MDM2 and high selectivity [385]. Recently, molecular modelling and virtual database screening strategies identified the novel scaffolds based on a polysubstituted pyridine, a bis-indolylmethyl or the 8-quinolinol derivative NSC66811. All of these were docked into the MDM2 binding cleft and superimposed on to the p53 peptide, showing good overlap with the three key p53 residues [386,387]. A structure-based library design identified the cyanobenzyland fluorobenzyl-substituted benzoylamino-benzoic acid as an inhibitor with mM potency [388].

3.6

CONCLUSION

While the design of nonpeptide mimetics from the structure of a bioactive peptide is still a formidable challenge, spectacular successes have been made over the last few years. A large variety of techniques have been elaborated, as described in this chapter, allowing rapid progress to be made toward selective, stable and bioavailable modified peptides or peptide mimetics. The contribution of high-throughput screening of combinatorial libraries has also identified a large number of leads with

REFERENCES

105

novel scaffolds, and has allowed the identification of a number of privileged scaffolds which have proven to be highly successful for peptidomimetic design.

REFERENCES [1] G. R. Marshall,A hierarchical approach to peptidomimetic design, Tetrahedron, 49, 3547–3558 (1993). [2] V. J. Hruby and P. M. Balse, Conformational and topographical considerations in designing agonist peptidomimetics from peptide leads, Curr.Med.Chem., 7, 945–970 (2000). [3] C. Adessi and C. Soto, Converting a peptide into a drug: Strategies to improve stability and bioavailability, Curr.Med.Chem., 9, 963–978 (2002). [4] M. D. Fletcher and M. M. Campbell, Partially modified retro-inverso peptides: development, synthesis, and conformational behavior, Chem.Rev., 98, 763–795 (1998). [5] G. J. Moore, Designing peptide mimetics, Trends Pharmacol.Sci., 15, 124–129 (1994). [6] A. S. Ripka and D. H. Rich, Peptidomimetic design, Curr.Opin.Chem.Biol., 2, 441–452 (1998). [7] M. Chorev, The partial retro-inverso modification: A road traveled together, Biopolymers, 80, 67–84 (2005). [8] Freidinger, R., Peptides and their retro enantiomers are topologically nonidentical, personal communication. [9] L. Scheibler and M. Chorev, Synthesis of Retro-Inverso-Peptides., In Methods of Organic Chemistry (Houben-Weyl). Synthesis of Peptides and Peptidomimetics., 4th edn., 2002, 528–551. [10] H. Tamamura, M. Mizumoto, K. Hiramatsu, S. Kusano, S. Terakubo, N. Yamamoto, J. O. Trent, Z. Wang, S. C. Peiper, H. Nakashima, A. Otaka and N. Fujii, Topochemical exploration of potent compounds using retro-enantiomer libraries of cyclic pentapeptides, Org.Biomol.Chem., 2, 1255–1257 (2004). [11] Y.-C. Chen, A. Muhlrad, A. Shteyer, M. Vidson, I. Bab and M. Chorev, Bioactive pseudopeptidic analogues and cyclostereoisomers of osteogenic growth peptide C-terminal pentapeptide, OGP(10–14), J.Med.Chem., 45, 1624–1632 (2002). [12] P. M. Fischer, The design, synthesis and application of stereochemical and directional peptide isomers: A critical review, Curr.Prot.Pept.Sci., 4, 339–356 (2003). [13] M. Molteni, C. Pesenti, M. Sani, A. Volonterio and M. Zanda, Fluorinated peptidomimetics: synthesis, conformational and biological features, J.Fluorine Chem., 125, 1735–1743 (2004). [14] A. Volonterio, S. Bellosta, F. Bravin, M. C. Bellucci, L. Bruche´, G. Colombo, L. Malpezzi, S. Mazzini, S. V. Meille, M. Meli, C. Ramirez de Arellano and M. Zanda, Synthesis, structure and conformation of partially-modified retro- and retro-inverso c[NHCH(CF3)]Gly peptides, Chem.Eur.J., 9, 4510–4522 (2003). [15] M. Zanda, Trifluoromethyl group: an effective xenobiotic function for peptide backbone modification, New J.Chem., 28, 1401–1411 (2004).

106

ASPECTS OF PEPTIDOMIMETICS

[16] S. Sagan, P. Karoyan, O. Lequin, G. Chassaing and S. Lavielle, N- and Ca-methylation in biologically active peptides: Synthesis, structural and functional aspects, Curr.Med.Chem., 11, 2533–2554 (2004). [17] C. Gilon, M. A. Dechantsreiter, F. Burkhart, A. Friedler and H. Kessler, Synthesis of N-Alkylated Peptides., In Methods of Organic Chemistry (Houben-Weyl). Synthesis of Peptides and Peptidomimetics., 4th edn., 2002, 215–271. [18] C. Toniolo, F. Formaggio, B. Kaptein and Q. B. Broxterman, You are sitting on a gold mine!, Synlett., 9, 1295–1310 (2006). [19] M. Tanaka, Design and synthesis of chiral alpha, alpha-disubstituted amino acids and conformational study of their oligopeptides, Chem.Pharm.Bull., 55, 349–358 (2007). [20] M. G. Hinds, J. H. Welsh, D. M. Brennand, J. Fisher, M. J. Glennie, N. G. J. Richards, D. L. Turner and J. A. Robinson, Synthesis, conformational properties, and antibody recognition of peptides containing beta-turn mimetics based on alpha-alkylproline derivatives, J.Med.Chem., 34, 1777–1789 (1991). [21] F. Formaggio, Q. B. Broxterman and C. Toniolo, Synthesis of Peptides Based on Ca-Tetrasubstituted a-Amino Acids., In Methods of Organic Chemistry (HoubenWeyl). Synthesis of Peptides and Peptidomimetics, 4th edn., 2002, 292–310. [22] P. Mathur, S. Ramakumar and V. S. Chauhan, Peptide design using a,b-dehydro amino acids: from b-turns to helical hairpins, Biopolymers (Pept.Sci), 76, 150–161 (2004). [23] M. A. Broda, E. M. Ciszak, A. E. Koziol, G. Pietrzynski and B. Rzeszotarska, Conformational investigation of a,b–dehydropeptides. XVI. b-turn tendency in AcPro-DXaa-NHMe:crystallographic and theoretical studies, J.Pept.Sci., 12, 538–549 (2006). [24] R. M. Joshi and V. S. Chauhan, Synthesis of Peptides Based on a,b-Didehydro-a-Amino Acids., In Methods of Organic Chemistry (Houben-Weyl). Synthesis of Peptides and Peptidomimetics, 4th edn., 2002, 636–662. [25] V. J. Hruby, G. Li, C. Haskell-Luevano and M. Shenderovich, Design of peptides, proteins, and peptidomimetics in Chi Space, Biopolymers., 43, 219–266 (1997). [26] S. E. Gibson, N. Guillo and M. J. Tozer, Towards control of w-space: Conformationally constrained analogues of Phe, Tyr, Trp, and His, Tetrahedron, 55, 585–615 (1999). [27] S. M. Cowell, Y. S. Lee, J. P. Cain and V. J. Hruby, Exploring Ramachandran and chi space: confomationally constrained amino acids and peptides in the design of bioactive polypeptide ligands, Curr.Med.Chem., 11, 2785–2798 (2004). [28] H. Josien, O. Convert, J.-P. Berlose, S. Sagan, A. Brunissen, S. Lavielle and G. Chassaing, Topographic analysis of the S7 binding subsite of the tachykinin neurokinin-1 receptor, Biopolymers, 39, 133–147 (1996). [29] H. Josien, S. Lavielle, A. Brunissen, M. Saffroy, Y. Torrens, J.-C. Beaujouan, J. Glowinski and G. Chassaing, Design and synthesis of side-chain conformationally restricted phenylalanines and their use for structure–activity studies on tachykinin NK-1 receptor, J.Med.Chem., 37, 1601 (1994). [30] S. B. Moore, M. Grant, Y. Rew, E. Bosa, M. Fabbri, U. Kumar and M. Goodman, Synthesis and biologic activity of conformationally constrained analogues of L-363,301, J.Peptide Res., 66, 404–422 (2005). [31] C. Nemes, L. Jeannin, J. Sapi, M. Laronze, H. Seghir, F. Auge´ and J.-Y. Laronze, A convenient synthesis of conformationally constrained b-substituted tryptophans, Tetrahedron, 56, 5479–5492 (2000).

REFERENCES

107

[32] J. Affaro–Lopez, W. Yuan, B. C. Phan, J. Kamath, Q. Lou, K. S. Lam and V. J. Hruby, Discovery of a novel series of potent and selective substrate-based inhibitors of p60c-src protein tyrosine kinase: conformational and topographical constraints in peptide design, J.Med.Chem., 41, 2252–2260 (1998). [33] J. Erchegyi, B. Penke, L. Simon, S. Michaelson, S. Wenger, B. Waser, R. Cescato, J.-C. Schaer, J. C. Reubi and J. Rivier, Novel sst4-selective somatostatin (SRIF) agonists. 2. Analogues with b-methyl-3-(2-naphtyl)alanine substitutions at position 8, J.Med.Chem., 46, 5587–5596 (2003). [34] G. To´th, E. Ioja, C. To¨mboly, S. Ballet, D. Tourwe´, A. Pe´ter, T. Martinek, N. N. Chung, P. W. Schiller, S. Benyhe and A. Borsodi, b-methyl substitution of cyclohexylalanine in Dmt-Tic-Cha-Phe peptides results in highly potent d opioid antagonists, J.Med.Chem., 50, 328–333 (2007). [35] J. Lin, S. Liao and V. J. Hruby, Syntheses of optically pure, conformationally constrained, and highly hydrophobic unusual amino acids: 2-amino-3, 3-diarylpropionic acids, J.Peptide Res., 65, 105–112 (2005). [36] M. Zhang, A. Porte, G. Diamantidis, K. Sogi, D. Kubrak, L. Resnick, S. C. Mayer, Z. Wang, A. F. Kreft and B. L. Harrison, Asymmetric synthesis of novel a-amino acids with b-branched side chains, Bioorg.Med.Chem.Lett., 17, 2401–2403 (2007). [37] J. Zhou, M. C. Ye, Z. Z. Huang and Y. Tang, Controllable enantioselective Friedel– Crafts reaction between indoles and alkylidene malonates catalyzed by pseudo-C-3symmetric trisoxazoline copper(II) complexes, J.Org.Chem., 69, 1309–1320 (2004). [38] X. Qian, K.E. Ko¨ver, M. D. Shenderovich, B.-S. Lou, A. Misicka, T. Zalewska, R. Horva´th, P. Davis, E. J. Bilsky, F. Porreca, H. I. Yamamura and V. J. Hruby, Newly discovered stereochemical requirements in the side-chain conformation of d opioid agonists for recognizing opioid d receptors, J.Med.Chem., 37, 1746–1757 (1994). [39] S. B. Lian, M. D. Shenderovich, Z. G. Zhang, L. Maletinska, J. Slaninova and V. J. Hruby, Substitution of the side–chain–constrained amino acids beta-methyl20 ,60 -dimethyl-40 -methoxytyrosine in position 2 of a bicyclic oxytocin analogue provides unique insights into the bioactive topography of oxytocin antagonists, J.Am.Chem.Soc., 120, 7393–7394 (1998). [40] S. B. Liao, J. Lin, M. D. Shenderovich, Y. L. Han, K. Hasohata, P. Davis, W. Qiu, F. Porreca, H. I. Yamamura and V. J. Hruby, The stereochemical requirements of the novel delta-opioid selective dipeptide antagonist TMT-TIC, Bioorg.Med.Chem.Lett., 7, 3049–3052 (1997). [41] G. J. Roff, R. C. Lloyd and N. J. Turner, A versatile chemo-enzymatic route to enantiomerically pure beta-branched alpha-amino acids, J.Am.Chem.Soc., 126, 4098–4099 (2004). [42] W. Wang, M. Cai, C. Xiong, J. Zhang, D. Trivedi and V. J. Hruby, Design and synthesis of novel w2-constrained phenylalanine, naphtylalanine, and tryptophan analogues and their use in biologically active melantropin peptides, Tetrahedron, 58, 7365–7374 (2002). [43] D. Tourwe´, E. Mannekens, T. N. T. Diem, P. Verheyden, H. Jaspers, G. To´th, A. Pe´ter, I. Kerte´sz, G. To¨ro¨k, N. N. Chung and P. W. Schiller, Side chain methyl substitution in the d-opioid receptor antagonist TIPP has an important effect on the activity profile, J.Med.Chem., 41, 5167–5176 (1998). [44] J. Erchegyi, B. Waser, J.–C. Schaer, R. Cescato, J. F. Brazeau, J. Rivier and J. C. Reubi, Novel sst4-selective somatostatin (SRIF) agonists. 3. Analogues amenable to radiolabeling, J.Med.Chem., 46, 5597–5605 (2003).

108

ASPECTS OF PEPTIDOMIMETICS

[45] D. M. Birney, D. C. Cole, C. E. Crosson, B. F. Kahl, B. W. Neff, T. W. Reid, K. Ren and R. D. Walkup, Use of b-methylphenylalanine (bMeF) residues to probe the nature of the interaction of substance P with its receptor: effects of bMeF-containing substance P analogues on rabbit iris smooth muscle contraction, J.Med.Chem., 38, 2478–2482 (1995). [46] J. Xu, L. Wei, R. Mathvink, J. He, Y.-J. Park, H. He, B. Leiting, K. A. Lyons, F. Marsilio, R. A. Patel, J. K. Wu, N. A. Thornberry and A. E. Weber, Discovery of potent and selective phenylalanine based dipeptidyl peptidase IV inhibitors, Bioorg.Med.Chem.Lett., 15, 2533–2536 (2005). [47] C. Cativiela and M. D. Dı`az-de-Villegas, Stereoselective synthesis of quaternary a-amino acids. Part 2: Cyclic compounds, Tetrahedron – Asymmetry, 11, 645–732 (2000). [48] P. Maity and B. Konig, Enantio- and diastereoselective syntheses of cyclic C-alphatetrasubstituted alpha–amino acids and their use to induce stable conformations in short peptides, Biopolymers, 90, 8–27 (2008). [49] M. Lasa and C. Cativiela, Synthesis of enantiomerically pure 1-amino-2-phenylcycloalkanecarboxylic acids (c(n)Phe), Synlett, 16, 2517–2533 (2006). [50] K. H. Park and M. J. Kurth, Cyclic amino acid derivatives, Tetrahedron, 58, 8629–8659 (2002). [51] U. Meyer, E. Breitling, P. Bisel and A. W. Frahm, Asymmetric Strecker synthesis of enatiopure 2,4-ethanothreonines, Tetrahedron – Asymmetry, 15, 2029–2037 (2004). [52] K. Namba, M. Kawasaki, I. Takada, S. Iwama, M. Izumida, T. Shinada and Y. Ohfune, Novel and efficient transformation of a-amino nitrile to a-imino and a-amide nitriles in asymmetric Strecker synthesis, Tetrahedron Lett., 42, 3733–3736 (2001). [53] Y. Ohfune and T. Shinada, Enantio– and diastereoselective construction of a,a-disubstituted a-amino acids for the synthesis of biologically active compounds, Eur.J.Org.Chem., 5127–5143 (2005). [54] J. Wede, F.-J. Volk and A. W. Frahm, Carbocyclic a-amino acids: asymmetric Strecker synthesis of a series of 2-alkylated 1-aminocyclopentanecarboxylic acids, Tetrahedron – Asymmetry, 11, 3231–3252 (2000). [55] A. Avenoza, J. I. Barriobero, C. Cativiela, M. A. Fernandez-Recio, J. M. Peregrina and F. Rodriguez, New synthesis of all four 1-amino-2-hydroxycyclohexanecarboxylic acids, Tetrahedron, 57, 2745–2755 (2001). [56] F.-J. Volk, M. Wagner and A. W. Frahm, Cyclobutane amino acids (CBAAs): asymmetric Strecker synthesis of enantiopure cis- and trans-2,4-methanovalines, Tetrahedron – Asymmetry, 14, 497–502 (2003). [57] M. Truong, F. Lecornue´ and A. Fadel, First synthesis of (1S,2S)- and (1R,2R)-1amino-2-isopropylcyclobutanecarboxylic acids by asymmetric Strecker reaction from 2-substituted cyclobutanones, Tetrahedron – Asymmetry, 14, 1063–1072 (2003). [58] P. Bisel, E. Breitling, M. Schlauch, F.-J. Volk and A. W. Frahm, Asymmetric synthesis and chromatographic resolution of all four stereomers of the a,b-propanoleucine, Pharmazie, 56, 770–772 (2001). [59] R. P. Wurz and A. B. Charette, An expedient and practical method for the synthesis of a diverse series of cyclopropane a-amino acids and amines, J.Org.Chem., 69, 1262– 1269 (2004).

REFERENCES

109

[60] C. H. Stammer, Cyclopropane amino-acids (2,3-methanoamino and 3,4-methanoamino acids), Tetrahedron,46, 2231–2254 (1990). [61] M. Horikawa, Y. Shigeri, N. Yumoto, S. Yoshikawa, T. Nakajima and Y. Ohfune, Syntheses of potent Leu-enkephalin analogues possessing b-hydroxy-a,a-disubstituted-a-amino acid and their characterization to opioid receptors, Bioorg.Med.Chem.Lett., 8, 2027–2032 (1998). [62] E. Gershonov, R. Granoth, E. Tzehoval, Y. Gaoni and M. Fridkin, 1-aminocyclobutanecarboxylic acid derivatives as novel structural elements in bioactive peptides: application to tuftsin analogues, J.Med.Chem., 39, 4833–4843 (1996). [63] O. O. Grygorenko, O. S. Artamonov, G. V. Palamarchuk, R. I. Zubatyuk, O. V. Shishkin and I. V. Komarov, Stereoselective synthesis of 2,4-methanoproline homologues, Tetrahedron – Asymmetry, 17, 252–258 (2006). [64] F. Couty, G. Evano and N. Rabasso, Synthesis of enantiopure azetidine 2-carboxylic acids and their incorporation into peptides, Tetrahedron – Asymmetry, 14, 2407– 2412 (2003). [65] Z. Sajjadi and W. D. Lubell, Amino acid–azetidine chimeras: synthesis of enantiopure 3-substituted azetidine-2-carboxylic acids, J.Peptide Res., 65, 298–310 (2005). [66] I. B. Parr, S. K. Boehlein, A. B. Dribben, S. M. Schuster and N. G. J. Richards, Mapping the Aspartic Acid Binding Site of Escherichia coli Asparagine Synthetase B Using Substrate Analogues, J.Med.Chem., 39, 2367–2378 (1996). [67] A. Mamai, N. E. Hughes, A. Wurthmann and J. S. Madalengoitia, Synthesis of conformationally constrained arginine and ornithine analogues based on the 3-substituted pyrrolidine framework, J.Org.Chem., 66, 6483–6486 (2001). [68] N. Pellegrini, M. Schmitt, S. Guery and J. J. Bourguignon, New strategies towards proline derivatives as conformationally constrained arginine analogues, Tetrahedron Lett., 43, 3243–3246 (2002). [69] J. Quancard, A. Labonne, Y. Jacquot, G. Chassaing, S. Lavielle and P. Karoyan, Asymmetric synthesis of 3-substituted proline chimeras bearing polar side chains of proteinogenic amino acids, J.Org.Chem., 69, 7940–7948 (2004). [70] J. Quancard, H. Magellan, S. Lavielle, G. Chassaing and P. Karoyan, Amino-zincene-enolate cyclisation: a short access to (2S,3R)- and (2S,3S)-3-benzylprolines (3-benzylpyrrolidine-2-carboxylic acids), Tetrahedron Lett., 45, 2185–2187 (2004). [71] G. A. Flynn, A. L. Akeson, R. Dharanipragada, M. J. Genin, J. A. Malikayil, R. Pottorf, J. S. Sabol, H. Schreuder, R. Tomlinson, P. Waid, R. Barrett, J. Jacobs and S. Yanofsky, Application of conformationally restricted peptidomimetics to modeling the bound conformation of peptide antagonists with the IL-1 receptor, Lett.Pept.Sci., 5, 93–100 (1998). [72] H. I. Mosberg and H. B. Kroona, Incorporation of a novel conformationally restricted tyrosine analogue into a cyclic, d opioid receptor selective tetrapeptide (JOM-13) enhances d receptor binding affinity and selectivity, J.Med.Chem., 35, 4498–4500 (1992). [73] M. Cai, C. Cai, A. V. Mayorov, C. Xiong, C. M. Cabello, V. A. Soloshonok, J. R. Swift, D. Trivedi and V. J. Hruby, Biological and conformational study of beta-substituted prolines in MT-II template: steric effects leading to human MC5 receptor selectivity, J.Peptide Res., 63, 116–131 (2004). [74] S. Sagan, J. Quancard, O. Lequin, P. Karoyan, G. Chassaing and S. Lavielle, Conformational analysis of the C-terminal Gly-Leu-Met-NH2 tripeptide of substance P bound to the NK-1 receptor, Chem.Biol., 12, 555–565 (2005).

110

ASPECTS OF PEPTIDOMIMETICS

[75] K. Sugase, M. Horikawan, M. Sugiyama and M. Ishiguro, Restriction of a peptide turn conformation and conformational analysis of guanidino group using arginineproline fused amino acids: application to mini atrial natriuretic peptide on binding to the receptor, J.Med.Chem., 47, 489–492 (2004). [76] C. Kadouri-Puchot and S. Comesse, Recent advances in asymmetric synthesis of pipecolic acid and derivatives – Review article, Amino Acids, 29, 101–130 (2005). [77] I. S. Kim, J. S. Oh, O. P. Zee and Y. H. Jung, Synthesis and conformational analysis of 3-hydroxypipecolic acid analogues via CSI-mediated stereoselective amination, Tetrahedron, 63, 2622–2633 (2007). [78] L. Le Corre and H. Dhimane, Synthesis of 5-substituted pipecolic acid derivatives as new conformationally constrained ornithine and arginine analogues, Tetrahedron Lett., 46, 7495–7497 (2005). [79] J. R. Casimir, K. Iterbeke, W. Van den Nest, M. C. Trescol-Biemont, H. Dumortier, S. Muller, D. Gerlier, C. Rabourdin-Combe, D. Tourwe´ and J. Paris, Conformational restriction of the Tyr(53) side-chain in the decapeptide HEL[52–61]: effects on binding to MHC-II I-A(k) molecule and TCR recognition, J.Peptide Res., 56, 398–408 (2000). [80] P. Ruzza, L. Cesaro, D. Tourwe´, A. Calderan, B. Biondi, V. Maes, I. Menegazzo, A. Osler, C. Rubini, A. Guiotto, L. A. Pinna, G. Borin and A. Donella-Deana, Spatial conformation and topography of the tyrosine aromatic ring in substrate recognition by protein tyrosine kinases, J.Med.Chem., 49, 1916–1924 (2006). [81] D. Tourwe´, K. Verschueren, A. Frycia, P. Davis, F. Porreca, V. J. Hruby, G. Toth, H. Jaspers, P. Verheyden and G. VanBinst, Conformational restriction of Tyr and Phe side chains in opioid peptides: Information about preferred and bioactive side-chain topology, Biopolymers, 38, 1–12 (1996). [82] W. M. Kazmierski, H. I. Yamamura and V. J. Hruby, Topographic design of peptide neurotransmitters and hormones on stable backbone templates – relation of conformation and dynamics to bioactivity, J.Am.Chem.Soc., 113, 2275–2283 (1991). [83] P. W. Schiller, G. Weltrowska, T. M. D. Nguyen, B. C. Wilkes, N. N. Chung and C. Lemieux, TIPP[c] – A highly potent and stable pseudopeptide delta-opioid receptor antagonist with extraordinary delta-selectivity, J.Med.Chem., 36, 3182–3187 (1993). [84] P. W. Schiller, G. Weltrowska, I. Berezowska, T. M. D. Nguyen, B. C. Wilkes, C. Lemieux and N. N. Chung, The TIPP opioid peptide family: Development of delta antagonists, delta agonists, and mixed mu agonist/delta antagonists, Biopolymers, 51, 411–425 (1999). [85] D. J. Kyle, J. A. Martin, S. G. Farmer and R. M. Burch, Design and conformationalanalysis of several highly potent bradykinin receptor antagonists, J.Med.Chem., 34, 1230–1233 (1991). [86] Z. X. Ye, T. MacNeil, D. H. Weinberg, R. N. Kalyani, R. Tang, A. M. Strack, B. A. Murphy, R. T. Mosley, D. E. MacIntyre, L. H. T. Van der Ploeg, A. A. Patchett, M. J. Wyvratt and R. R. Nargund, Structure–activity relationship of linear tetrapeptides Tic-DPhe-Arg-Trp-NH2 at the human melanocortin-4 receptor and effects on feeding behaviors in rat, Peptides, 26, 2017–2025 (2005). [87] K. X. Chen, F. G. Njoroge, J. Pichardo, A. Prongay, N. Butkiewicz, N. Yao, V. Madison and V. Girijavallabhan, Potent 7-hydroxy-1,2,3,4-tetrahydroisoquinoline3-carboxylic acid-based macrocyclic inhibitors of hepatitis C virus NS3 protease, J.Med.Chem., 49, 567–574 (2006).

REFERENCES

111

[88] C. G. Wang and H. I. Mosberg, Synthesis of a novel series of topographically constrained amino-acids – benzo-1,2,3,4-tetrahydroisoquinoline-3-carboxylic acids, Tetrahedron Lett., 36, 3623–3626 (1995). [89] S. Van Cauwenberghe, J. Martins, M. Gosselin, L. Hodzic, J. Butterworth and D. Tourwe´, Replacement of the Tic residue by benzo-1,2,3,4-tetrahydroisoquinoline3-carboxylic acids in TIPP peptides, In: Chorev, M. and Sawyer, T. K., Peptides. Peptide Revolution: Genomics, Proteomics & Therapeutics. Proceedings of the 18th American Peptide Symposium, 665–666, 2003. [90] J. B. Thomas, R. N. Atkinson, N. A. Vinson, J. L. Catanzaro, C. L. Perretta, S. E. Fix, S. W. Mascarella, R. B. Rothman, H. Xu, C. M. Dersch, B. E. Cantrell, D. M. Zimmerman and F. I. Carroll, Identification of (3R)-7-hydroxy-N-((1S)-1{[(3R,4R)-4-(3-hydroxyphenyl)-3,4-dimethyl-1-piperidinyl]methyl}-2-methylpropyl)1,2,3,4-tetrahydro-3-isoquinolinecarboxamide as a novel potent and selective opioid kappa receptor antagonist, J.Med.Chem., 46, 3127–3137 (2003). [91] S. E. Gibson, N. Guillo, J. O. Jones, I. M. Buck, S. B. Kalindjian, S. Roberts and M. J. Tozer, CCK2 receptor antagonists containing the conformationally constrained phenylalanine derivatives, including the new amino acid Xic, Eur.J.Med.Chem., 37, 379–389 (2002). [92] R. Z. Cai, S. Radulovic, J. Pinski, A. Nagy, T. W. Redding, D. B. Olsen and A. V. Schally, Pseudononapeptide bombesin antagonists containing C-terminal Trp or Tpi, Peptides, 13, 267–271 (1992). [93] Z. Darula, K. E. Kover, K. Monory, A. Borsodi, E. Mako, A. Ronai, D. Tourwe´, A. Peter and G. Toth, Deltorphin II analogues with 6-hydroxy-2-aminotetralin-2carboxylic acid in position 1, J.Med.Chem., 43, 1359–1366 (2000). [94] S. Salvadori, S. D. Bryant, C. Bianchi, G. Balboni, V. Scaranari, M. Attila and L. H. Lazarus, Phe(3)-substituted analogues of deltorphin-C – spatial conformation and topography of the aromatic ring in peptide recognition by delta-opioid receptors, J.Med.Chem., 36, 3748–3756 (1993). [95] G. Toth, Z. Darula, A. Peter, F. Fulop, D. Tourwe´, H. Jaspers, P. Verheyden, Z. Bocskey, Z. Toth and A. Borsodi, Conformationally constrained deltorphin analogues with 2-aminotetralin-2-carboxylic acid in position 3, J.Med.Chem., 40, 990–995 (1997). [96] B. S. Vig, M. Q. Zheng, T. F. Murray and J. V. Aldrich, Effects of the substitution of Phe(4) in the opioid peptide [D–Ala(8)]dynorphin A-(1–11)NH2, J.Med.Chem., 46, 4002–4008 (2003). [97] X. J. Chu, D. Bartkovitz, W. Danho, J. Swistok, A. W. H. Cheung, G. Kurylko, K. Rowan, M. Yeon, L. Franco, L. D. Qi, L. Chen and K. Yagaloff, Discovery of 1-amino-4-phenylcyclohexane-1-carboxylic acid and its influence on agonist selectivity between human melanocortin-4 and -1 receptors in linear pentapeptides, Bioorg.Med.Chem.Lett., 15, 4910–4914 (2005). [98] S. Oishi, R. G. Karki, S. U. Kang, X. Z. Wang, K. M. Worthy, L. K. Bindu, M. C. Nicklaus, R. J. Fisher and T. R. Burke, Design and synthesis of conformationally constrained Grb2 SH2 domain binding peptides employing alpha-methylphenylalanyl based phosphotyrosyl mimetics, J.Med.Chem., 48, 764–772 (2005). [99] D. Obrecht, M. Altorfer, U. Bohdal, J. Daly, W. Huber, A. Labhardt, C. Lehmann, K. Muller, R. Ruffieux, P. Schonholzer, C. Spiegler and C. Zumbrunn, Design and synthesis of novel nonpolar host peptides for the determination of the 3(10)- and alpha-helix compatibilities of alpha-amino acid building blocks: An assessment of alpha,alpha-disubstituted glycines, Biopolymers, 42, 575–626 (1997).

112

ASPECTS OF PEPTIDOMIMETICS

[100] T. Satoh, M. Hirano, A. Kuroiwa and Y. Kaneko, A synthesis of optically active alpha-quaternary alpha-amino acids and esters by assembling three components, ketones, (R)-chloromethyl p-tolyl sulfoxide and sodium azide, via sulfinyloxiranes, Tetrahedron, 62, 9268–9279 (2006). [101] M. Koppitz, G. Reinhardt and A. van Lingen, Solid-phase synthesis of substituted 3-amino-30 -carboxy-tetrahydrocarbazoles, Tetrahedron Lett., 46, 911–914 (2005). [102] S. Pruhs, C. Dinter, T. Blume, A. Schutz, M. Harre and H. Neh, Upscaling the solidphase synthesis of a tetrahydrocarbazole in chemical development, Org.Process Res.Dev., 10, 441–445 (2006). [103] S. J. Shuttleworth, M. E. Lizarzaburu, A. Chai and P. Coward, Identification and optimization of novel partial agonists of Neuromedin B receptor using parallel synthesis, Bioorg.Med.Chem.Lett., 14, 3037–3042 (2004). [104] R. Millet, J. F. Goossens, K. Bertrand-Caumont, P. Chavatte, R. Houssin and J. P. Henichart, Synthesis and biological evaluation of conformationally restricted derivatives of tryptophan as NK1/NK2 ligands, Lett.Pept.Sci., 6, 221–233 (1999). [105] P. W. Schiller, G. Weltrowska, T. M. D. Nguyen, B. C. Wilkes, N. N. Chung and C. Lemieux, Conformationally restricted deltorphin analogues, J.Med.Chem., 35, 3956–3961 (1992). [106] D. Tourwe´, K. Iterbeke, G. To¨ro¨k, G. Laus, F. Fulop, A. Pe´ter, F. Richard and P. Kitabgi, Pro10–Tyr11 substitutions provide potent or selective NT(8–13) analogues, In: Benedetti, E. and Pedone, C., Peptides. Proceedings of the 27th European Peptide Symposium, 304–305, Napoli, Italy, 2002. [107] G. A. Flynn, E. L. Giroux and R. C. Dage, An acyl-iminium ion cyclization route to a novel conformationally restricted dipeptide mimic – applications to angiotensinconverting enzyme-inhibition, J.Am.Chem.Soc., 109, 7914–7915 (1987). [108] S. E. Delaszlo, B. L. Bush, J. J. Doyle, W. J. Greenlee, D. G. Hangauer, T. A. Halgren, R. J. Lynch, T. W. Schorn and P. K. S. Siegl, Synthesis and use of 3-amino-4-phenyl-2-piperidones and 4-amino-2-benzazepin-3-ones as conformationally restricted phenylalanine isosteres in renin inhibitors, J.Med.Chem., 35, 833–846 (1992). [109] G. A. Flynn, D. W. Beight, S. Mehdi, J. R. Koehl, E. L. Giroux, J. F. French, P. W. Hake and R. C. Dage, Application of a conformationally restricted Phe–Leu dipeptide mimetic to the design of a combined inhibitor of angiotensin I-converting enzyme and neutral endopeptidase-24.11, J.Med.Chem., 36, 2420–2423 (1993). [110] A. M. Warshawsky, G. A. Flynn, J. R. Koehl, S. Mehdi and R. J. Vaz, The synthesis of aminobenzazepinones as anti-phenylalanine dipeptide mimics and their use in NEP inhibition, Bioorg.Med.Chem.Lett., 6, 957–962 (1996). [111] T. Le Diguarher, J. C. Ortuno, D. Shanks, N. Guilbaud, A. Pierre, E. Raimbaud, J. L. Fauchere, J. A. Hickman, G. C. Tucker and P. J. Casara, Synthesis of N,N0 -disubstituted 3-aminobenzo[c] and [d]azepin-2-ones as potent and specific farnesyl transferase inhibitors, Bioorg.Med.Chem.Lett., 14, 767–771 (2004). [112] J. R. Casimir, D. Tourwe´, K. Iterbeke, G. Guichard and J. P. Briand, Efficient synthesis of (S)-4-phthalimido-1,3,4,5-tetrahydro-8-(2,6-dichlorobenzyloxy)-3-oxo2H- 2-benzazepin-2-acetic acid (Pht-Hba(2,6-Cl-2-Bn)-Gly-OH), J.Org.Chem., 65, 6487–6492 (2000). [113] K. Van Rompaey, I. Van den Eynde, N. De Kimpe and D. Tourwe´, A versatile synthesis of 2-substituted 4-amino-1,2,4,5-tetrahydro-2-benzazepine-3-ones, Tetrahedron, 59, 4421–4432 (2003).

REFERENCES

113

[114] S. Ballet, A. Frycia, J. Piron, N. N. Chung, P. W. Schiller, P. Kosson, A. W. Lipkowski and D. Tourwe´, Synthesis and biological evaluation of constrained analogues of the opioid peptide H-Tyr-D-Ala-Phe-Gly-NH2 using the 4-amino-2-benzazepin-3-one scaffold, J.Peptide Res., 66, 222–230 (2005). [115] S. Ballet, A. V. Mayorov, M. Y. Cai, D. Tymecka, K. B. Chandler, E. S. Palmer, K. Van Rompaey, A. Misicka, D. Tourwe´ and V. J. Hruby, Novel selective human melanocortin-3 receptor ligands: Use of the 4-amino-1,2,4,5-tetrahydro-2-benzazepin-3-one (Aba) scaffold, Bioorg.Med.Chem.Lett., 17, 2492–2498 (2007). [116] S. Ballet, R. De Wachter, K. Van Rompaey, C. Tomboly, D. Feytens, G. Toth, L. Quartara, P. Cucchi, S. Meini and D. Tourwe´, Bradykinin analogues containing 4-amino-2-benzazepin-3-one the scaffold at the C-terminus, J.Pept.Sci., 13, 164–170 (2007). [117] I. Van den Eynde, G. Laus, P. W. Schiller, P. Kosson, N. N. Chung, A. W. Lipkowski and D. Tourwe´, A new structural motif for mu-opioid antagonists, J.Med.Chem., 48, 3644–3648 (2005). [118] S. Ballet, S. Salvadori, C. Trapella, S. D. Bryant, Y. Jinsmaa, L. H. Lazarus, L. Negri, E. Giannini, R. Lattanzi, D. Tourwe´ and G. Balboni, New 20 ,60 -dimethyl-L-tyrosine (Dmt) opioid peptidomimetics based on the Aba–Gly scaffold. Development of unique mu-opioid receptor ligands, J.Med.Chem., 49, 3990–3993 (2006). [119] P. Ruzza, A. Calderan, A. Donella–Deana, B. Biondi, L. Cesaro, A. Osler, S. Elardo, A. Guiotto, L. A. Pinna and G. Borin, Conformational constraints of tyrosine in protein tyrosine kinase substrates: Information about preferred bioactive side–chain orientation, Biopolymers, 71, 478–488 (2003). [120] K. Van Rompaey, S. Ballet, C. Tomboly, R. De Wachter, K. Vanommeslaeghe, M. Biesemans, R. Willem and D. Tourwe´, Synthesis and evaluation of the beta-turn properties of 4-amino-1,2,4,5-tetrahydro-2-benzazepin-3-ones and of their spirocyclic derivative, Eur.J.Org.Chem., 2899–2911 (2006). [121] K. Pulka, D. Feytens, I. Van den Eynde, R. De Wachter, P. Kosson, A. Misicka, A. Lipkowski, N. N. Chung, P. W. Schiller and D. Tourwe´, Synthesis of 4-amino-3oxo-tetrahydroazepino[3,4-b]indoles: new conformationally constrained Trp analogues, Tetrahedron, 63, 1459–1466 (2007). [122] D. Feytens, R. Cescato, J. C. Reubi and D. Tourwe´, New sst(4/5)-selective somatostatin peptidomimetics based on a constrained tryptophan scaffold, J.Med.Chem., 50, 3397–3401 (2007). [123] A. F. Spatola, Peptide Backbone Modifications: a Structure–Activity Analysis of Peptides Containing Amide Bond Surrogates. Conformational Constraints, and Related Backbone Replacements., In Chemistry and Biochemistry of Amino Acids, Peptides, and Proteins, Volume 7 edn., 1983, ch. 5, 267–357. [124] S. Gupta and J. W. Payne, Evaluation of the conformational propensities of peptide isosteres as a basis for selecting bioactive pseudopeptides, J.Peptide Res., 58, 546–561 (2001). [125] P. Gillespie, J. Cicariello and G. L. Olson, Conformational analysis of dipeptide mimetics, Biopolymers., 43, 191–217 (1997). [126] D. H. Coy, S. J. Hocart and Y. Sasaki, Solid-phase reductive alkylation techniques in analogue peptide-bond and side-chain modification, Tetrahedron, 44, 835–841 (1988). [127] J. P. Meyer, P. Davis, K. B. Lee, F. Porreca, H. I. Yamamura and V. J. Hruby, Synthesis using a Fmoc-based strategy and biological–activities of some reduced

114

[128]

[129]

[130]

[131]

[132]

[133]

[134]

[135]

[136]

[137]

[138]

[139]

[140]

ASPECTS OF PEPTIDOMIMETICS peptide-bond pseudopeptide analogues of dynorphin A(1), J.Med.Chem., 38, 3462–3468 (1995). N. P. Boyarskaya, D. I. Prokhorov, Y. G. Kirillova, E. N. Zvonkova and V. I. Shvets, Synthesis of protected pseudopeptides from dicarboxylic amino acids by Mitsunobu condensation, Tetrahedron Lett., 46, 7359–7362 (2005). D. H. Coy, P. Heinzerian, N. Y. Jiang, Y. Sasaki, J. Taylor, J. P. Moreau, W. T. Wolfrey, J. D. Gardner and R. T. Jensen, Probing peptide backbone function in bombesin – a reduced peptide-bond analogue with potent and specific receptor antagonist activity, J.Biol.Chem., 263, 5056–5060 (1988). M. Rodriguez, M. F. Lignon, M. C. Galas, P. Fulcrand, C. Mendre, A. Aumelas, J. Laur and J. Martinez, Synthesis and biological–activities of pseudopeptide analogues of the C-terminal heptapeptide of cholecystokinin – on the importance of the peptide-bonds, J.Med.Chem., 30, 1366–1373 (1987). Y. Sasaki, W. A. Murphy, M. L. Heiman, V. A. Lance and D. H. Coy, Solid-phase synthesis and biological properties of c-[CH2NH] pseudopeptide analogues of a highly potent somatostatin octapeptide, J.Med.Chem., 30, 1162–1166 (1987). A. Ambo, T. Adachi and Y. Sasaki, Synthesis and opioid activities of [D-Leu(8)]Dynorphin(1–8) analogues containing a reduced peptide–bond, c(CH2NH), Chem.Pharm.Bull., 43, 1547–1550 (1995). J. Couder, D. Tourwe´, G. VanBinst, J. Schuurkens and J. E. Leysen, Synthesis and biological-activities of c(CH2NH) pseudopeptide analogues of the C-terminal hexapeptide of neurotensin, Int.J.Pept.Protein Res., 41, 181–184 (1993). D. Lugrin, F. Vecchini, S. Doulut, M.Rodriguez, J. Martinez and P. Kitabgi, Reduced peptide-bond pseudopeptide analogues of neurotensin – binding and biological-activities and in vitro metabolic stability, Eur.J.Pharmacol., 205, 191–198 (1991). G. Drapeau, N. E. Rhaleb, S. Dion, D. Jukic and D. Regoli, [Phe8-c(CH2NH)Arg9]Bradykinin, a B2-receptor selective agonist which is not broken down by either kininase-I or kininase-II, Eur.J.Pharmacol., 155, 193–195 (1988). R. Bergmann, M. Scheunemann, C. Heichert, P. Mading, H. Wittrisch, M. Kretzschmar, H. Rodig, D. Tourwe´, K. Iterbeke, K. Chavatte, D. Zips, J. C. Reubi and B. Johannsen, Biodistribution and catabolism of F-18-labeled neurotensin (8–13) analogues, Nucl.Med.Biol., 29, 61–72 (2002). P. Blauenstein, E. G. Garayoa, D. Ruegg, A. Blanc, D. Tourwe´, A. Beck-Sickinger and P. A. Schubiger, Improving the tumor uptake of Tc-99m-labeled neuropeptides using stabilized peptide analogues, Cancer Biother.Radio., 19, 181–188 (2004). L. Elmasdouri, A. Aubry, C. Sakarellos, E. J. Gomez, M. T. Cung and M. Marraud, A beta-turn-like conformation in reduced peptides – crystal-structures of 2 dipeptide analogues with Pro-Gly-c[CH2NH] sequence, Int. J. Peptide Protein Res., 31, 420–428 (1988). C. Toniolo, G. Valle, M. Crisma, J. S. Kaltenbronn, J. T. Repine, G. Van Binst, M. Elseviers and D. Tourwe´, c[CH2NH] Backbone-modified peptides: first unequivocal observation of a C7 structure in a linear peptide, Peptide Research, 2, 332– 337 (1989). A. Geyer, G. Muller and H. Kessler, Conformational-analysis of a cyclic RGD peptide–containing a c[CH2NH] bond – A positional shift in backbone structure caused by a single dipeptide mimetic, J.Am.Chem.Soc., 116, 7735–7743 (1994).

REFERENCES

115

[141] D. Coy, L. H. Wang, N. Y. Jiang and R. Jensen, Short chain bombesin pseudopeptides with potent bombesin receptor antagonist activity in rat and guinea-pig pancreatic acinar-cells, Eur.J.Pharmacol., 190, 31–38 (1990). [142] M. Hiramatsu, K. Inoue, A. Ambo, Y. Sasaki and T. Kameyama, Long-lasting antinociceptive effects of a novel dynorphin analogue, Tyr-D-Ala-Phe-Leu-Arg c (CH2NH) Arg-NH2, in mice, Brit.J.Pharmacol., 132, 1948–1956 (2001). [143] G. E. Feurle, H. E. Meyer and G. Hamscher, Metabolism and potency of xenin and of its reduced hexapseudopeptide c fragment in the dog, Life Sci., 74, 697–707 (2003). [144] J. Martinez, J. P. Bali, M. Rodriguez, B. Castro, R. Magous, J. Laur and M. F. Lignon, Synthesis and biological-activities of some pseudo-peptide analogues of tetragastrin – the importance of the peptide backbone, J.Med.Chem., 28, 1874–1879 (1985). [145] S. Zacharia, W. J. Rossowski, N. Y. Jiang, P. Hrbas, A. Ertan and D. H. Coy, New reduced peptide-bond substance-P agonists and antagonists – effects on smoothmuscle contraction, Eur.J.Pharmacol., 203, 353–357 (1991). [146] N. G. J. Delaet, P. M. F. Verheyden, D. Tourwe´, G. VanBinst, P. Davis and T. F. Burks, Morphiceptin and beta-casomorphin-5 analogues containing a reduced peptide-bond – selective mu-receptor agonists and a novel mu antagonist, H-TyrPro-c(CH2–NH) Phe-Pro-Gly-OH, Biopolymers, 32, 957–969 (1992). [147] A. Balasubramaniam, V. C. Dhawan, D. E. Mullins, W. T. Chance, S. Sheriff, M. Guzzi, M. Prabhakaran and E. M. Parker, Highly selective and potent neuropeptide Y (NPY) Y1 receptor antagonists based on [Pro(30), Tyr(32), Leu(34)]NPY (28–36)-NH2 (BW1911U90), J.Med.Chem., 44, 1479–1482 (2001). [148] G. Calo’, R. Guerrini, R. Bigoni, A. Rizzi, C. Bianchi, D. Regoli and S. Salvadori, Structure-activity study of the nociceptin(1–13)-NH2 N-terminal tetrapeptide and discovery of a nociceptin receptor antagonist, J.Med.Chem., 41, 3360–3366 (1998). [149] J. R. Wisner, B. G. Xue, D. H. Coy and I. G. Renner, (D-Ala1, Leu9-c-CH2NHLeu10)Neuromedin-C antagonizes neuromedin C-stimulated amylase release by acini isolated from the rat pancreas, Pancreas, 5, 408–414 (1990). [150] Q. Y. Zhao, Q. Chen, D. J. Yang, Y. Feng, Y. Long, P. Wang and R. Wang, Endomorphin 1[c] and endomorphin 2[c], endomorphins analogues containing a reduced (CH2NH) amide bond between Tyr(1) and Pro(2), display partial agonist potency but significant antinociception, Life Sci., 77, 1155–1165 (2005). [151] M. Szelke, B. Leckie, A. Hallett, D. M. Jones, J. Sueiras, B. Atrash and A. F. Lever, Potent new inhibitors of human renin, Nature, 299, 555–557 (1982). [152] K. Suguna, E. A. Padlan, C. W. Smith, W. D. Carlson and D. R. Davies, Binding of a reduced peptide inhibitor to the aspartic proteinase from rhizopus-chinensis – implications for a mechanism of action, P.Natl.Acad.Sci.USA, 84, 7009–7013 (1987). [153] D. Skrbec and D. Romeo, Inhibition of Candida albicans secreted aspartic protease by a novel series of peptidomimetics, also active on the HIV-1 protease, Biochem.Bioph.Res.Co., 297, 1350–1353 (2002). [154] J. S. Wai, D. L. Bamberger, T. E. Fisher, S. L. Graham, R. L. Smith, J. B. Gibbs, S. D. Mosser, A. I. Oliff, D. L. Pompliano, E. Rands and N. E. Kohl, Synthesis and biological activity of Ras Farnesyl Protein Transferase inhibitors. Tetrapeptide analogues with amino methyl and carbon linkages, Bioorg.Med.Chem., 2, 939–947 (1994).

116

ASPECTS OF PEPTIDOMIMETICS

[155] C. A. Coburn, S. J. Stachel, K. G. Jones, T. G. Steele, D. M. Rush, J. DiMuzio, B. L. Pietrak, M. T. Lai, Q. Huang, J. Lineberger, L. X. Jin, S. Munshi, M. K. Holloway, A. Espeseth, A. Simon, D. Hazuda, S. L. Graham and J. P. Vacca, BACE-1 inhibition by a series of c[CH2NH] reduced amide isosteres, Bioorg.Med.Chem.Lett., 16, 3635–3638 (2006). [156] T. Narumi, K. Tomita, E. Inokuchi, K. Kobayashi, S. Oishi, H. Ohno and N. Fujii, Diastereoselective synthesis of highly functionalized fluoroalkene dipeptide isosteres and its application to Fmoc-based solid-phase synthesis of a cyclic pentapeptide mimetic, Tetrahedron, 64, 4332–4346 (2008). [157] A. Niida, K. Tomita, M. Mizumoto, H. Tanigaki, T. Terada, S. Oishi, A. Otaka, K. Inui and N. Fujii, Unequivocal synthesis of (Z)-alkene and (E)-fluoroalkene dipeptide isosteres to probe structural requirements of the peptide transporter PEPT1, Org.Lett., 8, 613–616 (2006). [158] P. A. Bartlett and A. Otake, Fluoroalkenes as peptide isosteres – ground-state analogue inhibitors of thermolysin, J.Org.Chem., 60, 3107–3111 (1995). [159] P. Wipf, T. C. Henninger and S. J. Geib, Methyl- and (trifluoromethyl)alkene peptide isosteres: Synthesis and evaluation of their potential as beta-turn promoters and peptide mimetics, J.Org.Chem., 63, 6088–6089 (1998). [160] J. B. Xiao, B. Weisblum and P. Wipf, Electrostatic versus steric effects in peptidomimicry: Synthesis and secondary structure analysis of gramicidin S analogues with (E)-alkene peptide isosteres, J.Am.Chem.Soc., 127, 5742–5743 (2005). [161] E. Inokuchi, T. Narumi, A. Niida, K. Kobayashi, K. Tomita, S. Oishi, H. Ohno and N. Fujii, Efficient synthesis of trifluoromethyl and related trisubstituted alkene dipeptide isosteres by palladium-catalyzed carbonylation of amino acid derived allylic carbonates, J.Org.Chem., 73, 3942–3945 (2008). [162] R. R. Gardner, G. B. Liang and S. H. Gellman, beta-turn and beta-hairpin mimicry with tetrasubstituted alkenes, J.Am.Chem.Soc., 121, 1806–1816 (1999). [163] J. B. Xiao, B. Weisblum and P. Wipf, Trisubstituted (E)-alkene dipeptide isosteres as beta-turn promoters in the gramicidin S cyclodecapeptide scaffold, Org.Lett., 8, 4731–4734 (2006). [164] D. Tourwe´, K. Iterbeke and E. Mannekens, Synthesis of Peptides Containing Carbaamide Bond Replacements, In Houben-Weyl. Methods of Organic Chemistry. Synthesis of Peptides and Peptidomimetics., 4th edn., 2003, vol. E22c, ch. 10.5, 324–372. [165] N. G. Bandur, K. Harms and U. Koert, First stereoselective synthesis of a Tyr-Tyr E-alkene isostere, Synlett., 99–102 (2007). [166] P. Wipf, J. B. Xiao and S. J. Geib, Imine additions of internal alkynes for the synthesis of trisubstituted (E)-alkene and cyclopropane peptide isosteres, Adv.Synth.Catal., 347, 1605–1613 (2005). [167] C. A. Olsen, H. Franzyk and J. W. Jaroszewski, Aziridines in parallel- and solidphase synthesis, Eur.J.Org.Chem., 1717–1724 (2007). [168] M. Wada, R. Doi, R. Hosotani, S. Higashide, T. Ibuka, H. Habashita, K. Nakai, N. Fujii and M. Imamura, Effect of a new bombesin receptor antagonist, (E)-alkene bombesin isostere, on amylase release from rat pancreatic acini, Pancreas, 10, 301–305 (1995). [169] P. Wipf and J. B. Xiao, Convergent approach to (E)-alkene and cyclopropane peptide isosteres, Org.Lett., 7, 103–106 (2005).

REFERENCES

117

[170] Z. Jakopin, R. Roskar and M. S. Dolenc, Synthesis of 3,5-disubstituted 1,2,4oxadiazoles as peptidomimetic building blocks, Tetrahedron Lett., 48, 1465– 1468 (2007). [171] J. Cesar and M. Sollner, Use of 3,5-disubstituted 1,2,4-triazoles for the synthesis of peptidomimetics, Synthetic Commun., 30, 4147–4158 (2000). [172] D. D. Beusen, J. Zabrocki, U. Slomczynska, R. D. Head, J. L. F. Kao and G. R. Marshall, Conformational Mimicry – Synthesis and Solution Conformation of A Cyclic Somatostatin Hexapeptide Containing A Tetrazole Cis Amide Bond Surrogate, Biopolymers, 36, 181–200 (1995). [173] V. Brecx, P. Verheyden and D. Tourwe´, Alpha-Bn-o-AMPA as a cis peptide bond mimic in somatostatin analogues, Lett.Pept.Sci., 5, 67–70 (1998). [174] A. D. Abell, Heterocyclic-based peptidomimetics, Lett.Pept.Sci., 8, 267–272 (2002). [175] J. K. Pokorski, L. M. M. Jenkins, H. Q. Feng, S. R. Durell, Y. W. Bai and D. H. Appella, Introduction of a triazole amino acid into a peptoid oligomer induces turn formation in aqueous solution, Org.Lett., 9, 2381–2383 (2007). [176] V. D. Bock, D. Speijer, H. Hiemstra and J. H. van Maarseveen, 1,2,3-triazoles as peptide bond isosteres: synthesis and biological evaluation of cyclotetrapeptide mimics, Org.Biomol.Chem., 5, 971–975 (2007). [177] A. K. Ghosh, P. R. Sridhar, N. Kumaragurubaran, Y. Koh, I. T. Weber and H. Mitsuya, Bis-tetrahydrofuran: a privileged ligand for darunavir and a new generation of HIV protease inhibitors that combat drug resistance, ChemMedChem, 1, 939 ff. (2006). [178] A. Brik and C. H. Wong, HIV-1 protease: mechanism and drug discovery, Org.Biomol.Chem., 1, 5–14 (2003). [179] B. B. Scott, G. M. McGeehan and R. K. Harrison, Development of inhibitors of the aspartyl protease renin for the treatment of hypertension, Curr.Protein Pept.Sc., 7, 241–254 (2006). [180] A. K. Ghosh, G. Bilcer, C. Harwood, R. Kawahama, D. Shin, K. A. Hussain, L. Hong, J. A. Loy, C. Nguyen, G. Koelsch, J. Ermolieff and J. Tang, Structurebased design: Potent inhibitors of human brain memapsin 2 (beta–secretase), J.Med.Chem., 44, 2865–2868 (2001). [181] M. C. Maillard, R. K. Hom, T. E. Benson, J. B. Moon, S. Mamo, M. Bienkowski, A. G. Tomasselli, D. D. Woods, D. B. Prince, D. J. Paddock, T. L. Emmons, J. A. Tucker, M. S. Dappen, L. Brogley, E. D. Thorsett, N. Jewett, S. Sinha and V. John, Design, synthesis, and crystal structure of hydroxyethyl secondary aminebased peptidomimetic inhibitors of human beta-secretase, J.Med.Chem., 50, 776–781 (2007). [182] H. J. Lee, H. M. Park and K. B. Lee, The beta-turn scaffold of tripeptide containing an azaphenylalanine residue, Biophys.Chem., 125, 117–126 (2007). [183] D. Boeglin and W. D. Lubell, Aza-amino acid scanning of secondary structure suited for solid-phase peptide synthesis with Fmoc chemistry and aza-amino acids with heteroatomic side chains, J.Comb.Chem., 7, 864–878 (2005). [184] E. Wieczerzak, S. Rodziewicz–Motowidlo, E. Jankowska, A. Gieldon and J. Ciarkowski, An enormously active and selective azapeptide inhibitor of cathespin B, J.Pept.Sci., 13, 536–543 (2007). [185] R. J. Simon, R. S. Kania, R. N. Zuckermann, V. D. Huebner, D. A. Jewell, S. Banville, S. Ng, L. Wang, S. Rosenberg, C. K. Marlowe, D. C. Spellmeyer,

118

[186]

[187]

[188] [189]

[190]

[191]

[192]

[193]

[194]

[195]

[196]

[197]

[198] [199]

ASPECTS OF PEPTIDOMIMETICS R. Y. Tan, A. D. Frankel, D. V. Santi, F. E. Cohen and P. A. Bartlett, Peptoids – A modular approach to drug discovery, P.Natl.Acad.Sci.USA, 89, 9367–9371 (1992). J. A. Patch, K. Kirshenbaum, S. L. Seurynck, R. N. Zuckermann and A. E. Barron, Versatile Oligo(N-Substituted) Glycines: The Many Roles of Peptoids in Drug Discovery, In: Pseudo-Peptides in Drug Discovery, 2004, ch. 1, 1–31. Q. Sui, D. Borchardt and D. L. Rabenstein, Kinetics and equilibria of cis/trans isomerization of backbone amide bonds in peptoids, J.Am.Chem.Soc., 129, 12042–12048 (2007). J. A. Patch and A. E. Barron, Helical peptoid mimics of magainin-2 amide, J.Am.Chem.Soc., 125, 12092–12093 (2003). C. W. Wu, K. Kirshenbaum, T. J. Sanborn, J. A. Patch, K. Huang, K. A. Dill, R. N. Zuckermann and A. E. Barron, Structural and spectroscopic studies of peptoid oligomers with alpha-chiral aliphatic side chains, J.Am.Chem.Soc., 125, 13525– 13530 (2003). T. S. Burkoth, A. T. Fafarman, D. H. Charych, M. D. Connolly and R. N. Zuckermann, Incorporation of unprotected heterocyclic side chains into peptoid oligomers via solid-phase submonomer synthesis, J.Am.Chem.Soc., 125, 8841– 8845 (2003). J. M. Holub, H. J. Jang and K. Kirshenbaum, Clickity-click: highly functionalized peptoid oligomers generated by sequential conjugation reactions on solid-phase support, Org.Biomol.Chem., 4, 1497–1502 (2006). J. R. Holder, R. M. Bauzo, Z. M. Xiang, J. Scott and C. Haskell-Luevano, Design and pharmacology of peptoids and peptide–peptoid hybrids based on the melanocortin agonists core tetrapeptide sequence, Bioorg.Med.Chem.Lett., 13, 4505– 4509 (2003). R. Ruijtenbeek, J. A. W. Kruijtzer, W. van de Wiel, M. J. E. Fischer, M. Fluck, F. A. M. Redegeld, R. M. J. Liskamp and F. P. Nijkamp, Peptoid–peptide hybrids that bind Syk SH2 domains involved in signal transduction, ChemBioChem, 2, 171–179 (2001). J. A. W. Kruijtzer, W. A. J. Nijenhuis, N. Wanders, W. H. Gispen, R. M. J. Liskamp and R. A. H. Adan, Peptoid–peptide hybrids as potent novel melanocortin receptor ligands, J.Med.Chem., 48, 4224–4230 (2005). C. Gilon, M. A. Dechantsreiter, F. Burkhart, A. Friedler and H. Kessler, Synthesis of N–Alkylated Peptides, In Houben-Weyl. Methods of Organic Chemistry. Synthesis of Peptides and Peptidomimetics., 4th edn., 2003, vol. E22c, ch. 10.1, 262. G. Calo, A. Rizzi, D. Rizzi, R. Bigoni, R. Guerrini, G. Marzola, M. Marti, J. McDonald, M. Morari, D. G. Lambert, S. Salvadori and D. Regoli, [Nphe(1),Arg(14),Lys(15)]nociceptin–NH2, a novel potent and selective antagonist of the nociceptin/orphanin FQ receptor, Brit.J.Pharmacol., 136, 303–311 (2002). R. C. Elgersma, G.E. Mulder, J. A. W. Kruijtzer, G. Posthuma, D. T. S. Rijkers and R. M. J. Liskamp, Transformation of the amyloidogenic peptide amylin(20–29) into its corresponding peptoid and retropeptoid: Access to both an amyloid inhibitor and template for self-assembled supramolecular tapes, Bioorg.Med.Chem.Lett., 17, 1837–1842 (2007). D. J. Combs and R. S. Lokey, Extended peptoids: a new class of oligomers based on aromatic building blocks, Tetrahedron Lett., 48, 2679–2682 (2007). O. Busnel, L. R. Bi, H. Dali, A. Cheguillaume, S. Chevance, A. Bondon, S. Muller and M. Baudy-Floc’h, Solid-phase synthesis of ‘mixed’ peptidomimetics using

REFERENCES

[200]

[201] [202]

[203]

[204]

[205]

[206]

[207]

[208] [209] [210] [211] [212]

[213] [214]

[215]

[216]

119

Fmoc-protected aza-/beta(3)-amino acids and alpha-amino acids, J.Org.Chem., 70, 10701–10708 (2005). A. Cheguillaume, F. Lehardy, K. Bouget, M. Baudy-Floc’h and P. Le Grel, Submonomer solution synthesis of hydrazinoazapeptoids, a new class of pseudopeptides, J.Org.Chem., 64, 2924–2927 (1999). J. A. Patch and A. E. Barron, Mimicry of bioactive peptides via non-natural, sequencespecific peptidomimetic oligomers, Curr.Opin.Chem.Biol., 6, 872–877 (2002). D. Seebach, A. K. Beck and D. J. Bierbaum, The world of beta- and gamma-peptides comprised of homologated proteinogenic amino acids and other components, Chem.Biodivers., 1, 1111–1239 (2004). C. Grison, S. Geneve, S. Claudel, P. Coutrot and M. Marraud, Catalytic hydrogenation of vinylogous peptides: a route towards gamma-peptide foldamers, Tetrahedron Lett., 44, 2297–2300 (2003). C. Baldauf, R. Gunther and H. J. Hofmann, Delta–peptides and delta–amino acids as tools for peptide structure design – A theoretical study, J.Org.Chem., 69, 6214– 6220 (2004). A. Violette, M.C. Verlant-Petit, V. Semetey, C. Hemmerlin, R. Casimir, R. Graff, M. Marraud, J. P. Briand, D. Rognan and G. Guichard, N,N0 -linked oligoureas as foldamers: Chain length requirements for helix formation in protic solvent investigated by circular dichroism, NMR spectroscopy, and molecular dynamics, J.Am.Chem.Soc., 127, 2156–2164 (2005). A. Boeijen, J. van Ameijde and R. M. J. Liskamp, Solid-phase synthesis of oligourea peptidomimetics employing the Fmoc protection strategy, J.Org.Chem., 66, 8454– 8462 (2001). A. Wiszniewska, D. Kunce, N. N. Chung, P. W. Schiller and J. Izdebski, p-Nitrophenoxycarbonyl derivatives of Boc-protected diaminoalkanes in the synthesis of enkephalin peptidomimetics, J.Pept.Sci., 11, 579–583 (2005). C. Y. Cho, C. W. Liu, D. E. Wemmer and P. G. Schultz, Cyclic and linear oligocarbamate ligands for human thrombin, Bioorg.Med.Chem., 7, 1171–1179 (1999). J. Venkatraman, S. C. Shankaramma and P. Balaram, Design of folded peptides, Chem.Rev., 101, 3131–3152 (2001). W. A. Loughlin, J. D. A. Tyndall, M. P. Glenn and D. P. Fairlie, Beta-strand mimetics, Chem.Rev., 104, 6085–6117 (2004). J. D. A. Tyndall, T. Nall and D. P. Fairlie, Proteases universally recognize beta strands in their active sites, Chem.Rev., 105, 973–999 (2005). M. J. Pe´rez de Vega, M. Martin-Martinez and R. Gonzalez-Mun˜iz, Modulation of protein–protein interactions by stabilizing/mimicking protein secondary structure elements, Curr.Top.Med.Chem., 7, 33–62 (2007). T. Moriuchi and T. Hirao, Highly ordered structures of peptides by using molecular scaffolds, Chem.Soc.Rev., 33, 294–301 (2004). K. Y. Tsang, H. Diaz, N. Graciani and J. W. Kelly, Hydrophobic cluster formation is necessary for dibenzofuran-based amino acids to function as b-sheet nucleators, J.Am.Chem.Soc., 116, 3988–4005 (1994). C. L. Nesloney and J. W. Kelly, A 2,30 -substituted biphenyl-based amino acid facilitates the formation of a monomeric beta-hairpin-like structure in aqueous solution at elevated temperature, J.Am.Chem.Soc., 118, 5836–5845 (1996). K. H. Mayo, R. P. M. Dings, C. Flader, I. Nesmelova, B. Hargittai, D. W. J. van der Schaft, L. I. van Eijk, D. Walek, J. Haseman, T. R. Hoye and A. W. Griffioen,

120

[217]

[218]

[219]

[220]

[221]

[222] [223] [224]

[225] [226]

[227]

[228] [229]

[230]

[231] [232]

[233]

ASPECTS OF PEPTIDOMIMETICS Design of a partial peptide mimetic of anginex with antiangiogenic and anticancer activity, J.Biol.Chem., 278, 45746–45752 (2003). M. C. Song, S. Rajesh, Y. Hayashi and Y. Kiso, Design and synthesis of new inhibitors of HIV-1 protease dimerization with conformationally constrained templates, Bioorg.Med.Chem.Lett., 11, 2465–2468 (2001). V. Brandmeier, W. H. B. Sauer and M. Feigel, Antiparallel beta-sheet conformation in cyclopeptides containing a pseudo-amino acid with a biphenyl moiety, Helv.Chim.Acta, 77, 70–85 (1994). E. Perissutti, F. Frecentese, A. Lavecchia, F. Fiorino, B. Severino, F. De Angelis, V. Santagada and G. Caliendo, Design and synthesis of potential beta-sheet nucleators via Suzuki coupling reaction, Tetrahedron, 63, 12779–12785 (2007). M. Erdelyi, V. Langer, A. Karlen and A. Gogoll, Insight into beta-hairpin stability: a structural and thermodynamic study of diastereomeric beta-hairpin mimetics, New J.Chem., 26, 834–843 (2002). D. S. Kemp and Z. Q. Li, 2-amino-20 -carboxydiphenylacetylenes as beta-turn mimetics – Synthesis and conformational properties, Tetrahedron Lett., 36, 4175–4178 (1995). K. Kim and J. P. Germanas, Peptides constrained to type VI beta-turns. 2. Antiparallel beta-ladder formation, J.Org.Chem., 62, 2853–2860 (1997). J. A. Robinson, The design, synthesis and conformation of some new beta-hairpin mimetics: Novel reagents for drug and vaccine discovery, Synlett., 429–441 (2000). Y. Che, B. R. Brooks and G. R. Marshall, Development of small molecules designed to modulate protein–protein interactions., J.Comput.Aid.Mol.Des., 20, 109–130 (2006). B. R. Huck, J. D. Fisk and S. H. Gellman, Promotion of sheet formation in alphapeptide strands by a beta-peptide reverse turn, Org.Lett., 2, 2607–2610 (2000). B. H. Baek, M. R. Lee, K. Y. Kim, U. I. Cho, D. W. Boo and I. Shin, Di-oxanipecotic acids as more stable turn motifs than di-nipecotic acids, Tetrahedron Lett., 44, 3447–3450 (2003). J. P. Schneider and J. W. Kelly, Synthesis and efficacy of square-planar coppercomplexes designed to nucleate beta-sheet structure, J.Am.Chem.Soc., 117, 2533– 2546 (1995). M. North, Incorporation of conformationally constrained beta-amino acids into peptides, J.Pept.Sci., 6, 301–313 (2000). C. P. R. Hackenberger, I. Schiffers, J. Runsink and C. Bolm, General synthesis of unsymmetrical norbornane scaffolds as inducers for hydrogen bond interactions in peptides, J.Org.Chem., 69, 739–743 (2004). J. Fotins and D. B. Smithrud, Creation and investigation of protein-core mimetics with parallel and antiparallel aligned amino acids, J.Org.Chem., 70, 4452–4459 (2005). K. Ryan, L. J. Gershell and W. C. Still, A synthetic receptor motif designed for extended peptide conformations, Tetrahedron, 56, 3309–3318 (2000). F. Freire, J. D. Fisk, A. J. Peoples, M. Ivancic, I. A. Guzei and S. H. Gellman, Diacid linkers that promote parallel b-sheet secondary structure in water, J.Am.Chem.Soc., 130, 7839–7841(2008). J. D. Fisk, M. A. Schmitt and S. H. Gellman, Thermodynamic analysis of autonomous parallel beta-sheet formation in water, J.Am.Chem.Soc., 128, 7148–7149 (2006).

REFERENCES

121

[234] D. S. Kemp and B. R. Bowen, Synthesis of peptide-functionalized diacylaminoepindolidiones as template for b-sheet formation, Tetrahedron Lett., 29, 5077–5080 (1988). [235] W. F. Michne and J. D. Schroeder, Design and synthesis of a beta-strand inducer – Application to ICAM-1/LFA-1 mediated cellular adhesion, Int. J. Peptide Protein Res., 47, 2–8 (1996). [236] C. Bonauer, M. Zabel and B. Konig, Synthesis and peptide binding properties of methoxypyrrole amino acids (MOPAS), Org.Lett., 6, 1349–1352 (2004). [237] J. S. Nowick, What I have learned by using chemical model systems to study biomolecular structure and interactions, Org.Biomol.Chem., 4, 3869–3885 (2006). [238] R. J. Woods, J. O. Brower, E. Castellanos, M. Hashemzadeh, O. Khakshoor, W. A. Russu and J. S. Nowick, Cyclic modular beta-sheets, J.Am.Chem.Soc., 129, 2548–2558 (2007). [239] S. Levin and J. S. Nowick, An artificial beta-sheet that dimerizes through parallel b-sheet interactions, J.Am.Chem.Soc., 129, 13043–13048 (2007). [240] L. Belvisi, C. Gennari, A. Mielgo, D. Potenza and C. Scolastico, Conformational preferences of peptides containing reverse-turn mimetic bicyclic lactams: Inverse gamma-turns versus type-II 0 beta-turns – Insights into beta-hairpin stability, Eur.J.Org.Chem., 389–400 (1999). [241] J. A. Robinson, b-Hairpin peptidomimetics: design, structures and biological activities, Accounts Chem.Res., ASAP, (2008). [242] A. Descours, K. Moehle, A. Renard and J. A. Robinson, A new family of beta-hairpin mimetics based on a trypsin inhibitor from sunflower seeds, ChemBioChem, 3, 318– 323 (2002). [243] S. J. DeMarco, H. Henze, A. Lederer, K. Moehle, R. Mukherjee, B. Romagnoli, J. A. Robinson, F. Brianza, F. O. Gombert, S. Lociuro, C. Ludin, J. W. Vrijbloed, J. Zumbrunn, J. P. Obrecht, D. Obrecht, V. Brondani, F. Hamy and T. Klimkait, Discovery of novel, highly potent and selective beta-hairpin mimetic CXCR4 inhibitors with excellent anti-HIV activity and pharmacokinetic profiles, Bioorg.Med.Chem., 14, 8396–8404 (2006). [244] J. A. Robinson, S. C. Shankaramma, P. Jettera, U. Kienzl, R. A. Schwendener, J. W. Vrijbloed and D. Obrecht, Properties and structure–activity studies of cyclic beta-hairpin peptidomimetics based on the cationic antimicrobial peptide protegrin I, Bioorg.Med.Chem., 13, 2055–2064 (2005). [245] R. L. A. Dias, R. Fasan, K. Moehle, A. Renard, D. Obrecht and J. A. Robinson, Protein ligand design: From phage display to synthetic protein epitope mimetics in human antibody Fc-binding peptidomimetics, J.Am.Chem.Soc., 128, 2726–2732 (2006). [246] Z. Athanassiou, K. Patora, R. L. A. Dias, K. Moehle, J. A. Robinson and G. Varani, Structure-guided peptidomimetic design leads to nanomolar beta-hairpin inhibitors of the Tat-TAR interaction of bovine immunodeficiency Virus, Biochemistry, 46, 741–751 (2007). [247] K. Moehle, Z. Athanassiou, K. Patora, A. Davidson, G. Varani and J. A. Robinson, Design of beta-hairpin peptidomimetics that inhibit binding of alpha-helical HIV-1 rev protein to the rev response element RNA, Angew.Chem.Int.Edit., 46, 9101– 9104 (2007).

122

ASPECTS OF PEPTIDOMIMETICS

[248] R. Fasan, R. L. A. Dias, K. Moehle, O. Zerbe, J. W. Vrijbloed, D. Obrecht and J. A. Robinson, Using a beta-hairpin to mimic an alpha-helix: Cyclic peptidomimetic inhibitors of the p53-HDM2 protein–protein interaction, Angew.Chem.Int.Edit., 43, 2109–2112 (2004). [249] N. Srinivas, K. Moehle, K. bou-Hadeed, D. Obrecht and J. A. Robinson, Biaryl amino acid templates in place of D-Pro-L-Pro in cyclic b-hairpin cationic antimicrobial peptidomimetics, Org.Biomol.Chem., 5, 3100–3105 (2007). [250] J. D. A. Tyndall and D. P. Fairlie, Macrocycles mimic the extended peptide conformation recognized by aspartic, serine, cysteine and metallo proteases, Curr.Med.Chem., 8, 893–907 (2001). [251] R. C. Reid, M. J. Kelso, M. J. Scanlon and D. P. Fairlie, Conformationally constrained macrocycles that mimic tripeptide beta-strands in water and aprotic solvents, J.Am.Chem.Soc., 124, 5673–5683 (2002). [252] M. P. Glenn, L. K. Pattenden, R. C. Reid, D. P. Tyssen, J. D. A. Tyndall, C. J. Birch and D. P. Fairlie, beta-strand mimicking macrocyclic amino acids: Templates for protease inhibitors with antiviral activity, J.Med.Chem., 45, 371–381 (2002). [253] C. B. Xue, M. E. Voss, D. J. Nelson, J. J. W. Duan, R. J. Cherney, I. C. Jacobson, X. H. He, J. Roderick, L. H. Chen, R. L. Corbett, L. Wang, D. T. Meyer, K. Kennedy, W. F. DeGrado, K. D. Hardman, C. A. Teleha, B. D. Jaffee, R. Q. Liu, R. A. Copeland, M. B. Covington, D. D. Christ, J. M. Trzaskos, R. C. Newton, R. L. Magolda, R. R. Wexler and C. P. Decicco, Design, synthesis, and structure–activity relationships of macrocyclic hydroxamic acids that inhibit tumor necrosis factor alpha release in vitro and in vivo, J.Med.Chem., 44, 2636– 2660 (2001). [254] G. Abbenante, D. A. Bergman, R. I. Brinkworth, D. R. March, R. C. Reid, P. A. Hunt, I. W. James, R. J. Dancer, B. Garnham, M. L. Stoermer and D. P. Fairlie, Structure-activity relationships for macrocyclic peptidomimetic inhibitors of HIV-1 protease, Bioorg.Med.Chem.Lett., 6, 2531–2536 (1996). [255] K. Ersmark, M. Nervall, H. Gutierrez-de-Teran, E. Hamelink, L. K. Janka, J. C. Clemente, B. M. Dunn, A. Gogoll, B. Samuelsson, J. Aqvist and A. Hallberg, Macrocyclic inhibitors of the malarial aspartic proteases plasmepsin I, II, and IV, Bioorgan.Med.Chem., 14, 2197–2208 (2006). [256] S. J. Stachel, C. A. Coburn, S. Sankaranarayanan, E. A. Price, B. L. Pietrak, Q. Huang, J. Lineberger, A. S. Espeseth, L. X. Jin, J. Ellis, M. K. Holloway, S. Munshi, T. Allison, D. Hazuda, A. J. Simon, S. L. Graham and J. P. Vacca, Macrocyclic inhibitors of beta-secretase: Functional activity in an animal model, J.Med.Chem., 49, 6147–6150 (2006). [257] A. K. Ghosh, T. Devasamudram, L. Hong, C. DeZutter, X. M. Xu, V. Weerasena, G. Koelsch, G. Bilcer and J. Tang, Structure–based design of cycloamide-urethanederived novel inhibitors of human brain memapsin 2 (beta–secretase), Bioorg.Med.Chem.Lett., 15, 15–20 (2005). [258] Y. S. Tsantrizos, J. M. Ferland, A. McClory, M. Poirier, V. Farina, N. K. Yee, X. J. Wang, N. Haddad, X. D. Wei, J. H. Xu and L. Zhang, Olefin ring-closing metathesis as a powerful tool in drug discovery and development – potent macrocyclic inhibitors of the hepatitis C virus NS3 protease, J.Organomet.Chem., 691, 5163–5171 (2006). [259] N. Goudreau, C. Brochu, D. R. Cameron, J. S. Duceppe, A. M. Faucher, J. M. Ferland, C. Grand-Maitre, M. Poirier, B. Simoneau and Y. S. Tsantrizos,

REFERENCES

[260]

[261]

[262]

[263]

[264]

[265]

[266] [267] [268]

[269] [270]

[271]

[272]

[273]

123

Potent inhibitors of the hepatitis C virus NS3 protease: Design and synthesis of macrocyclic substrate-based beta-strand mimics, J.Org.Chem., 69, 6185–6201 (2004). A. B. Smith III, M. C. Guzman, P. A. Sprengeler, T. P. Keenan, R. C. Holcomb, J. L. Wood, P. J. Carroll and R. Hirschmann, De novo design, synthesis and X-ray crystal structures of pyrrolinone-based b-strand peptidomimetics, J.Am.Chem.Soc., 116, 9947–9962 (1994). A. B. Smith, R. Hirschmann, A. Pasternak, R. Akaishi, M. C. Guzman, D. R. Jones, T. P. Keenan, P. A. Sprengeler, P. L. Darke, E. A. Emini, M. K. Holloway and W. A. Schleif, Design and synthesis of peptidomimetic inhibitors of HIV-1 protease and renin – Evidence for improved transport, J.Med.Chem., 37, 215–218 (1994). A. B. Smith, T. Nittoli, P. A. Sprengeler, J. J. W. Duan, R. Q. Liu and R. F. Hirschmann, Design, synthesis, and evaluation of a pyrrolinone-based matrix metalloprotease inhibitor, Org.Lett., 2, 3809–3812 (2000). K. H. Lee, G. L. Olson, D. R. Bolin, A. B. Benowitz, P. A. Sprengeler, A. B. Smith, R. F. Hirschmann and D. C. Wiley, The crystal structure of a pyrrolinone–peptide hybrid ligand bound to the human class II MHC protein HLA–DR1, J.Am.Chem.Soc., 122, 8370–8375 (2000). A. B. Smith, A. K. Charnley, E. F. Mesaros, O. Kikuchi, W. Y. Wang, A. Benowitz, C. L. Chu, J. J. Feng, K. H. Chen, A. Lin, F. C. Cheng, L. Taylor and R. Hirschmann, Design, synthesis, and binding affinities of pyrrolinone–based somatostatin mimetics, Org.Lett., 7, 399–402 (2005). S. T. Philips, M. Rezac, U. Abel, M. Kossenjans and P. A. Bartlett, ‘@-tides’: The 1,2-dihydro-3(6H)-pyridinone unit as a b-strand mimic, J.Am.Chem.Soc., 124, 58–59 (2002). S. T. Phillips, L. K. Blasdel and P. A. Bartlett, @-tides as reporters for molecular associations, J.Am.Chem.Soc., 127, 4193–4198 (2005). M. C. Hammond, B. Z. Harris, W. A. Lim and P. A. Bartlett, beta strand peptidomimetics as potent PDZ domain ligands, Chem.Biol., 13, 1247–1251 (2006). M. C. Hammond and P. A. Bartlett, Synthesis of amino acid-derived cyclic acyl amidines for use in beta-strand peptidomimetics, J.Org.Chem., 72, 3104–3107 (2007). D. Blomberg, K. Brickmann and J. Kihlberg, Synthesis of a beta-strand mimetic based on a pyridine scaffold, Tetrahedron, 62, 10937–10944 (2006). Y. M. Qian, M. A. Blaskovich, M. Saleem, C. M. Seong, S. P. Wathen, A. D. Hamilton and S. M. Sebti, Design and structural requirements of potent peptidomimetic inhibitors of P21(Ras) farnesyltransferase, J.Biol.Chem., 269, 12410– 12413 (1994). Y. M. Qian, A. Vogt, S. M. Sebti and A. D. Hamilton, Design and synthesis of nonpeptide Ras CAAX mimetics as potent farnesyltransferase inhibitors, J.Med.Chem., 39, 217–223 (1996). G. Semple, D. M. Ashworth, G. R. Baker, A. R. Batt, A. J. Baxter, D. W. M. Benzies, L. H. Elliot, D. M. Evans, R. J. Franklin, P. Hudson, P. D. Jenkins, G. R. Pitt, D. P. Rooker, A. Sheppard, M. SZELKE, S. Yamamoto and Y. Isomura, Pyridonebased peptidomimetic inhibitors of interleukin-1 beta-converting enzyme (ICE), Bioorg.Med.Chem.Lett., 7, 1337–1342 (1997). J. M. C. Golec, M. D. Mullican, M. A. Murcko, K. P. Wilson, D. P. Kay, S. D. Jones, R. Murdoch, G. W. Bemis, S. A. Raybuck, Y. P. Luong and

124

[274]

[275]

[276]

[277]

[278]

[279]

[280]

[281]

[282] [283]

[284]

[285]

ASPECTS OF PEPTIDOMIMETICS D. J. Livingston, Structure-based design of non-peptidic pyridone aldehydes as inhibitors of interleukin-1 beta converting enzyme, Bioorg.Med.Chem.Lett., 7, 2181–2186 (1997). F. J. Brown, D. W. Andisik, P. R. Bernstein, C. B. Bryant, C. Ceccarelli, J. R. Damewood, S. A. Woolson, P. D. Edwards, R. A. Earley, S. Feeney, R. C. Green, B. Gomes, B. J. Kosmider, R. D. Krell, A. Shaw, G. B. Steelman, R. N. Thomas, E. P. Vacek, C. A. Veale, P. A. Tuthill, P. Warner, J. C. Williams and D. J. Wolanin, Design of orally-active, nonpeptidic inhibitors of human-leukocyte elastase, J.Med.Chem., 37, 1259–1261 (1994). R. M. Freidinger, D. F. Veber, D. S. Perlow, J. R. Brooks and R. Saperstein, Bioactive conformation of luteinizing-hormone-releasing hormone – evidence from a conformationally constrained analogue, Science, 210, 656–658 (1980). A. Perdih and D. Kikelj, The application of Freidinger lactams and their analogues in the design of conformationally constrained peptidomimetics, Curr.Med.Chem., 13, 1525–1556 (2006). J. Cluzeau and W. D. Lubell, Design, synthesis, and application of azabicyclo[X.Y.O]alkanone amino acids as constrained dipeptide surrogates and peptide mimics, Biopolymers, 80, 98–150 (2005). P. D. Boatman, C. O. Ogbu, M. Eguchi, H. O. Kim, H. Nakanishi, B. L. Cao, J. P. Shea and M. Kahn, Secondary structure peptide mimetics: Design, synthesis, and evaluation of beta-strand mimetic thrombin inhibitors, J.Med.Chem., 42, 1367–1375 (1999). W. A. Slusarchyk, J. A. Robl, P. C. Taunk, M. M. Asaad, J. E. Bird, J. Dimarco and Y. Pan, Dual metalloprotease inhibitors. 5. Utilization of bicyclic azepinonethiazolidines and azepinonetetrahydrothiazines in constrained peptidomimetics of mercaptoacyl dipeptides, Bioorg.Med.Chem.Lett., 5, 753–758 (1995). X. M. Chen, R. P. M. Dings, I. Nesmelova, S. Debbert, J. R. Haseman, J. Maxwell, T. R. Hoye and K. H. Mayo, Topomimetics of amphipathic beta-sheet and helixforming bactericidal peptides neutralize lipopolysaccharide endotoxins, J.Med.Chem., 49, 7754–7765 (2006). R. P. M. Dings, X. M. Chen, D. M. E. I. Hellebrekers, L. I. van Eijk, Y. Zhang, T. R. Hoye, A. W. Griffioen and K. H. Mayo, Design of nonpeptidic topomimetics of antiangiogenic proteins with antitumor activities, J.Natl.Cancer Inst., 98, 932–936 (2006). H. Yin and A. D. Hamilton, Strategies for targeting protein–protein interactions with synthetic agents, Angew.Chem.Int.Edit., 44, 4130–4163 (2005). Y. Wei, G. L. McLendon, A. D. Hamilton, M. A. Case, C. B. Purring, Q. Lin, H. S. Park, C. S. Lee and T. N. Yu, Disruption of protein–protein interactions: design of a synthetic receptor that blocks the binding of cytochrome c to cytochrome c peroxidase, Chem.Commun., 1580–1581 (2001). F. Z. Zhang, O. Sadovski, S. J. Xin and G. A. Woolley, Stabilization of folded peptide and protein structures via distance matching with a long, rigid cross-linker, J.Am.Chem.Soc., 129, 14154 ff. (2007). A. K. Boal, I. Guryanov, A. Moretto, M. Crisma, E. L. Lanni, C. Toniolo, R. H. Grubbs and D.J. O’Leary, Facile and E-selective intramolecular ring-closing metathesis reactions in 3(10)-helical peptides: A 3D structural study, J.Am.Chem.Soc., 129, 6986 ff. (2007).

REFERENCES

125

[286] J. Garner and M. M. Harding, Design and synthesis of alpha-helical peptides and mimetics, Org.Biomol.Chem., 5, 3577–3585 (2007). [287] J.-S. Park, H.-S. Lee, J. R. Lai, B. M. Kim and S. H. Gellman, Accommodation of a-substituted residues in the b-peptide 12-helix: expanding the range of substitution patterns available to a foldamer scaffold, J.Am.Chem.Soc., 125, 8539–8545 (2003). [288] C. Baldauf, R. Gunther and H. J. Hofmann, Control of helix formation in vinylogous gamma-peptides by (E)- and (Z)-double bonds: A way to ion channels and monomolecular nanotubes, J.Org.Chem., 70, 5351–5361 (2005). [289] L. Zhao and J. Chmielewski, Inhibiting protein–protein interactions using designed molecules, Curr.Opin.Struct.Biol., 15, 31–34 (2005). [290] J. M. Davis, L. K. Tsou and A. D. Hamilton, Synthetic non-peptide mimetics of alpha-helices, Chem.Soc.Rev., 36, 326–334 (2007). [291] J. P. Schneider and J. W. Kelly, Templates that induce alpha-helical, beta-sheet, and loop conformations, Chem.Rev., 95, 2169–2187 (1995). [292] D. S. Kemp and T. P. Curran, (2S,5S,8S,11S)-1-Acetyl-1,4-diaza-3-keto-5-carboxy10-thia-tricyclo-[2.8.04,8]-tridecane, 1 the preferred conformation of 1 (1 ¼ a-TempOH) and its peptide conjugates a-Temp-L-(Ala)n-OR (n ¼ 1 to 4) and a-Temp-L-Ala-LPhe-L-Lys(E-Boc)-L-Lys(E-Boc)-NHMe studies of templates for a-helix formation, Tetrahedron Lett., 29, 4935–4938 (1988). [293] T. Arrhenius and A. C. Satterthwait, The substitution of an amide–amide backbone hydrogen bond in an a-helical peptide with a covalent hydrogen bond mimic, In: Peptides: Chemistry, structure and biology, Proceedings of the 11th American Chemical Symposium, 870–872, 1990. [294] E. Cabezas and A. C. Satterthwait, The hydrogen bond mimic approach: Solidphase synthesis of a peptide stabilized as an alpha–helix with a hydrazone link, J.Am.Chem.Soc., 121, 3862–3875 (1999). [295] J. C. Calvo, K. C. Choconta, D. Diaz, O. Orozco, M. M. Bravo, F. Espejo, L. M. Salazar, F. Guzman and M. E. Patarroyo, An alpha helix conformationally restricted peptide is recognized by cervical carcinoma patients’ sera, J.Med.Chem., 46, 5389–5394 (2003). [296] R. N. Chapman, G. Dimartino and P. S. Arora, A highly stable short alpha-helix constrained by a main-chain hydrogen-bond surrogate, J.Am.Chem.Soc., 126, 12252–12253 (2004). [297] R. N. Chapman and P. S. Arora, Optimized synthesis of hydrogen-bond surrogate helices: Surprising effects of microwave heating on the activity of grubbs catalysts, Org.Lett., 8, 5825–5828 (2006). [298] G. Dimartino, D. Y. Wang, R. N. Chapman and P. S. Arora, Solid-phase synthesis of hydrogen-bond surrogate-derived alpha-helices, Org.Lett., 7, 2389–2392 (2005). [299] D. Y. Wang, K. Chen, G. Dimartino and P. S. Arora, Nucleation and stability of hydrogen-bond surrogate-based alpha-helices, Org.Biomol.Chem., 4, 4074–4081 (2006). [300] D. Wang, K. Chen, J. L. Kulp and P. S. Arora, Evaluation of biologically relevant short alpha-helices stabilized by a main-chain hydrogen-bond surrogate, J.Am.Chem.Soc., 128, 9248–9256 (2006). [301] J. Liu, D. Wang, Q. Zheng, M. Lu and P. S. Arora, Atomic structure of a short alpha-helix stabilized by a main chain hydrogen–bond surrogate, J.Am.Chem.Soc., 130, 4334–4337 (2008).

126

ASPECTS OF PEPTIDOMIMETICS

[302] D. Y. Wang, W. Liao and P. S. Arora, Binding properties of artificial a helices derived from a hydrogen-bond surrogate: Application to Bcl-xL, Angew.Chem.Int.Edit., 44, 6525–6529 (2005). [303] D. Wang, M. Lu and P. S. Arora, Inhibition of HIV-1 fusion by hydrogen-bondsurrogate-based alpha helices, Angew.Chem.Int.Edit., 47, 1879–1882 (2008). [304] C. A. Hutton and P. A. Bartlett, Preparation of diazabicyclo[4.3.0]nonene-based peptidomimetics, J.Org.Chem., 72, 6865–6872 (2007). [305] R. E. Austin, R. A. Maplestone, A. M. Sefler, K. Liu, W. N. Hruzewicz, C. W. Liu, H. S. Cho, D. E. Wemmer and P. A. Bartlett, Template for stabilization of a peptide alpha-helix: Synthesis and evaluation of conformational effects by circular dichroism and NMR, J.Am.Chem.Soc., 119, 6461–6472 (1997). [306] D. C. Horwell, W. Howson, G. S. Ratcliffe and H. M. G. Willems, The design of dipeptide helical mimetics: The synthesis, tachykinin receptor affinity and conformational analysis of 1,1,6-trisubstituted indanes, Bioorgan.Med.Chem., 4, 33–42 (1996). [307] E. Jacoby, Biphenyls as potential mimetics of protein alpha-helix, Bioorg.Med.Chem.Lett., 12, 891–893 (2002). [308] B. P. Orner, J. T. Ernst and A. D. Hamilton, Toward proteomimetics: Terphenyl derivatives as structural and functional mimics of extended regions of an alphahelix, J.Am.Chem.Soc., 123, 5382–5383 (2001). [309] H. Yin, G. Lee, K. A. Sedey, O. Kutzki, H. S. Park, B. P. Orner, J. T. Ernst, H. G. Wang, S. M. Sebti and A. D. Hamilton, Terphenyl-based bak BH3 a-helical proteomimetics as low-molecular-weight antagonists of Bcl-xL, J.Am.Chem.Soc., 127, 10191–10196 (2005). [310] O. Kutzki, H. S. Park, J. T. Ernst, B. P. Orner, H. Yin and A. D. Hamilton, Development of a potent Bcl-xL antagonist based on alpha-helix mimicry, J.Am.Chem.Soc., 124, 11838–11839 (2002). [311] H. Yin, G. I. Lee, H. S. Park, G. A. Payne, J. M. Rodriguez, S. M. Sebti and A. D. Hamilton, Terphenyl-based helical mimetics that disrupt the p53/HDM2 interaction, Angew.Chem.Int.Edit., 44, 2704–2707 (2005). [312] J. T. Ernst, O. Kutzki, A. K. Debnath, S. Jiang, H. Lu and A. D. Hamilton, Design of a protein surface antagonist based on alpha-helix mimicry: Inhibition of gp41 assembly and viral fusion, Angew.Chem.Int.Edit., 41, 278 ff. (2001). [313] I. C. Kim and A. D. Hamilton, Diphenylindane-based proteomimetics reproduce the projection of the i, i þ 3, i þ 4, and i þ 7 residues on an a-helix, Org.Lett., 8, 1751–1754 (2006). [314] J. T. Ernst, J. Becerril, H. S. Park, H. Yin and A. D. Hamilton, Design and application of an alpha-helix-mimetic scaffold based on an oligoamide-foldamer strategy: Antagonism of the bak BH3/Bcl-xL complex, Angew.Chem.Int.Edit., 42, 535 ff. (2003). [315] J. M. Ahn and S. Y. Han, Facile synthesis of benzamides to mimic an alpha-helix, Tetrahedron Lett., 48, 3543–3547 (2007). [316] H. Yin and A. D. Hamilton, Terephthalamide derivatives as mimetics of the helical region of Bak peptide target Bcl-xL protein, Bioorg.Med.Chem.Lett., 14, 1375– 1379 (2004). [317] J. M. Rodriguez and A. D. Hamilton, Intramolecular hydrogen bonding allows simple enaminones to structurally mimic the i, i þ 4, and i þ 7 residues of an a-helix, Tetrahedron Lett., 47, 7443–7446 (2006).

REFERENCES

127

[318] A. Volonterio, L. Moisan and J. Julius, Synthesis of pyridazine-based scaffolds as alpha-helix mimetics, Org.Lett., 9, 3733–3736 (2007). [319] L. Moisan, S. Odermatt, N. Gombosuren, A. Carella and J. Rebek, Synthesis of an oxazole-pyrrole-piperazine scaffold as an alpha-helix mimetic, Eur. J. Org. Chem., 10, 1673–1676 (2008). [320] P. Maity and B. Konig, Synthesis and structure of 1,4-dipiperazino benzenes: Chiral terphenyl-type peptide helix mimetics, Org.Lett., 10, 1473–1476 (2008). [321] Y. Che, B. R. Brooks and G. R. Marshall, Protein recognition motifs: Design of peptidomimetics of helix surfaces, Biopolymers, 86, 288–297 (2007). [322] W. Antuch, S. Menon, Q. Z. Chen, Y. C. Lu, S. Sakamuri, B. Beck, V. SchauerVukasinovic, S. Agarwal, S. Hess and A. Domling, Design and modular parallel synthesis of a MCR derived alpha-helix mimetic protein–protein interaction inhibitor scaffold, Bioorg.Med.Chem.Lett., 16, 1740–1743 (2006). [323] H. Oguri, A. Oomura, S. Tanabe and M. Hirama, Design and synthesis of a transfused polycyclic ether skeleton as an a–helix mimetic scaffold, Tetrahedron Lett., 46, 2179–2183 (2005). [324] H. Oguri, S. Tanabe, A. Oomura, M. Umetsu and M. Hirama, Synthesis and evaluation of a-helix mimetics based on a trans-fused polycyclic ether: sequenceselective binding to aspartate pairs in a-helical peptides, Tetrahedron Lett., 47, 5801–5805 (2006). [325] S. E. Van Compernolle, A. V. Wiznycia, J. R. Rush, M. Dhanasekaran, P. W. Baures and S. C. Todd, Small molecule inhibition of hepatitis C virus E2 binding to CD81, Virology, 314, 371–380 (2003). [326] A. L. Rodriguez, A. Tamrazi, M. L. Collins and J. A. Katzenellenbogen, Design, synthesis, and in vitro biological evaluation of small molecule inhibitors of estrogen receptor a coactivator binding, J.Med.Chem., 47, 600–611 (2004). [327] J. S. Blakeney, R. C. Reid, G. T. Le and D. P. Fairlie, Nonpeptidic ligands for peptideactivated G protein-coupled receptors, Chem.Rev., 107, 2960–3041 (2007). [328] N. R. A. Beeley, Can peptides be mimicked?, Drug Discov.Today, 5, 354–363 (2000). [329] R. M. Jones, P. D. Boatman, G. Semple, Y.-J. Shin and S. Y. Tamura, Clinically validated peptides as template for de novo peptidomimetic drug design at G-proteincoupled receptors, Curr.Opin.Pharmacol., 3, 530–543 (2003). [330] A. Janecka, M. Zubrzycka and T. Janecki, Somatostatin analogues, J. Peptide Res., 58, 91–107 (2001). [331] L. Yang, Non-peptide Somatostatin Receptor Ligands, In: Annu. Rep. Med. Chem., 1999, ch. 21, 209–218. [332] R. M. Freidinger, Design and synthesis of novel bioactive peptides and peptidomimeties, J.Med.Chem., 46, 5553–5566 (2003). [333] R. H. Mattern, S. B. Moore, T. A. Tran, J. K. Rueter and M. Goodman, Synthesis, biological activities and conformational studies of somatostatin analogues, Tetrahedron, 56, 9819–9831 (2000). [334] V. Prasad, E. T. Birzin, C. T. McVaugh, R. D. van Rijn, S. P. Rohrer, G. Chicchi, D. J. Underwood, E. R. Thornton, A. B. Smith and R. Hirschmann, Effects of heterocyclic aromatic substituents on binding affinities at two distinct sites of somatostatin receptors. Correlation with the electrostatic potential of the substituents, J.Med.Chem., 46, 1858–1869 (2003). [335] R. Hirschmann, K. C. Nicolaou, S. Pietranico, J. Salvino, E. M. Leahy, P. A. Sprengeler, G. Furst, A. B. Smith, C. D. Strader, M. A. Cascieri,

128

[336]

[337] [338]

[339]

[340] [341]

[342]

[343]

[344]

[345]

[346]

ASPECTS OF PEPTIDOMIMETICS M. R. Candelore, C. Donaldson, W. Vale and L. Maechler, Nonpeptidal peptidomimetics with a beta-D-glucose scaffolding – A partial somatostatin agonist bearing a close structural relationship to a potent, selective substance-P antagonist, J.Am.Chem.Soc., 114, 9217–9218 (1992). R. Hirschmann, J. Hynes, M.A. Cichy-Knight, R. D. van Rijn, P. A. Sprengeler, P. G. Spoors, W. C. Shakespeare, S. Pietranico-Cole, J. Barbosa, J. Liu, W. Q. Yao, S. Rohrer and A. B. Smith, Modulation of receptor and receptor subtype affinities using diastereomeric and enantiomeric monosaccharide scaffolds as a means to structural and biological diversity. A new route to ether synthesis, J.Med.Chem., 41, 1382–1391 (1998). S. G. Gouin and P. V. Murphy, Synthesis of somatostatin mimetics based on the 1-deoxymannojirimycin scaffold, J.Org.Chem., 70, 8527–8532 (2005). D. Damour, M. Barreau, J. C. Blanchard, M. C. Burgevin, A. Doble, F. Herman, G. Pantel, E. JamesSurcouf, M. Vuilhorgne, S. Mignani, L. Poitout, Y. LeMerrer and J. C. Depezay, Design, synthesis and binding affinities of novel non-peptide mimics of somatostatin/sandostatinÒ, Bioorg.Med.Chem.Lett., 6, 1667–1672 (1996). B. P. Mowery, V. Prasad, C. S. Kenesky, A. R. Angeles, L. L. Taylor, J. J. Feng, W. L. Chen, A. Lin, F. C. Cheng, A. B. Smith and R. Hirschmann, Catechol: A minimal scaffold for non-peptide peptidomimetics of the i þ 1 and i þ 2 positions of the beta-turn of somatostatin, Org.Lett., 8, 4397–4400 (2006). C. Papageorgiou and X. Borer, A non-peptide ligand for the somatostatin receptor having a benzodiazepinone structure, Bioorg.Med.Chem.Lett., 6, 267–272 (1996). D. Damour, F. Herman, R. Labaudiniere, G. Pantel, M. Vuilhorgne and S. Mignani, Synthesis of novel proline and gamma-lactam derivatives as non-peptide mimics of Somatostatin/SandostatinÒ, Tetrahedron, 55, 10135–10154 (1999). D. Damour, M. Barreau, J. C. Blanchard, M. C. Burgevin, A. Doble, G. Pantel, R. Labaudiniere and S. Mignani, Synthesis and binding affinities of novel spirocyclic lactam peptidomimetics of somatostatin, Chem.Lett., 943–944 (1998). A. J. Souers, A. A. Virgilio, A. Rosenquist, W. Fenuik and J. A. Ellman, Identification of a potent heterocyclic ligand to somatostatin receptor subtype 5 by the synthesis and screening of beta-turn mimetic libraries, J.Am.Chem.Soc., 121, 1817–1825 (1999). D. Chianelli, Y. C. Kim, D. Lvovskiy and T. R. Webb, Application of a novel design paradigm to generate general nonpeptide combinatorial scaffolds mimicking beta turns: Synthesis of Ligands for somatostatin receptors, Bioorg.Med.Chem., 11, 5059–5068 (2003). S. P. Rohrer, E. T. Birzin, R. T. Mosley, S. C. Berk, S. M. Hutchins, D. M. Shen, Y. S. Xiong, E. C. Hayes, R. M. Parmar, F. Foor, S. W. Mitra, S. J. Degrado, M. Shu, J. M. Klopp, S. J. Cai, A. Blake, W. W. S. Chan, A. Pasternak, L. H. Yang, A. A. Patchett, R. G. Smith, K. T. Chapman and J. M. Schaeffer, Rapid identification of subtype-selective agonists of the somatostatin receptor through combinatorial chemistry, Science, 282, 737–740 (1998). L. H. Yang, S. C. Berk, S. P. Rohrer, R. T. Mosley, L. Q. Guo, D. J. Underwood, B. H. Arison, E. T. Birzin, E. C. Hayes, S. W. Mitra, R. M. Parmar, K. Cheng, T. J. Wu, B. S. Butler, F. Foor, A. Pasternak, Y. P. Pan, M. Silva, R. M. Freidinger, R. G. Smith, K. Chapman, J. M. Schaeffer and A. A. Patchett, Synthesis and biological activities of potent peptidomimetics selective for somatostatin receptor subtype 2, P.Natl.Acad.Sci.USA, 95, 10836–10841 (1998).

REFERENCES

129

[347] A. Pasternak, Y. Pan, R. T. Mosley, S. P. Rohrer, E. T. Birzin, J. M. Schaeffer, A. A. Patchett and L. Yang, Diverse peptidyl privileged structures as potent somatostatin receptor ligands, Lett. Drug Des. Discov., 1, 121–125 (2004). [348] S. P. Rohrer and J. M. Schaeffer, Identification and characterization of subtype selective somatostatin receptor agonists, J.Physiol.-Paris, 94, 211–215 (2000). [349] D. Feytens, M. De Vlaeminck, R. Cescato, D. Tourwe, J. C. Reubi, Highly potent 4-amino-indola[2,3-c]azepin-3-one-containig Somatostatin mimetics with a range of sst receptor selectivities, J. Med. Chem, 52, 95–104 (2009). [350] M. Ankersen, M. Crider, S. Q. Liu, B. Ho, H. S. Andersen and C. Stidsen, Discovery of a novel non-peptide somatostatin agonist with SST4 selectivity, J.Am.Chem.Soc., 120, 1368–1373 (1998). [351] M.O. Contour-Galcera, A. Sidhu, P. Plas and P. Roubert, 3-Thio-1,2,4-triazoles, novel somatostatin sst(2)/sst(5) agonists, Bioorg.Med.Chem.Lett., 15, 3555–3559 (2005). [352] H. Mabuchi, N. Suzuki and T. Miki, WO98/47882, 1998. [353] B. A. Hay, B. M. Cole, F. DiCapua, G. W. Kirk, M. C. Murray, R. A. Nardone, D. J. Pelletier, A. P. Ricketts, A. S. Robertson and T. W. Siegel, Small molecule somatostatin receptor subtype-2 antagonists, Bioorg.Med.Chem.Lett., 11, 2731– 2734 (2001). [354] R. E. Martin, L. G. Green, W. Guba, N. Kratochwil and A. Christ, Discovery of the first nonpeptidic, small-molecule, highly selective somatostatin receptor subtype 5 antagonists: A chemogenomics approach, J.Med.Chem., 50, 6291–6294 (2007). [355] L. Poitout, P. Roubert, M.O. Contour-Galera, C. Moinet, J. Lannoy, J. Pommier, P. Plas, D. Bigg and C. Thurieau, Identification of potent non-peptide somatostatin antagonists with sst(3) selectivity, J.Med.Chem., 44, 2990–3000 (2001). [356] D. Seebach, L. Schaeffer, M. Brenner and D. Hoyer, Design and synthesis of gamma-dipeptide derivatives with submicromolar affinities for human somatostatin receptors, Angew.Chem.Int.Edit., 42, 776–778 (2003). [357] D. Seebach, M. Rueping, P. I. Arvidsson, T. Kimmerlin, P. Micuch, C. Noti, D. Langenegger and D. Hoyer, Linear, peptidase-resistant beta(2)/beta(3)-di- and alpha/beta(3)-tetrapeptide derivatives with nanomolar affinities to a human somatostatin receptor –Helv.Chim.Acta, 84, 3503–3510 (2001). [358] K. Gademann, T. Kimmerlin, D. Hoyer and D. Seebach, Peptide folding induces high and selective affinity of a linear and small beta–peptide to the human somatostatin receptor 4, J.Med.Chem., 44, 2460–2468 (2001). [359] R. Hirschmann, W. Q. Yao, M. A. Cascieri, C. D. Strader, L. Maechler, M. A. CichyKnight, J. Hynes, R. D. vanRijn, P. A. Sprengeler and A. B. Smith, Synthesis of potent cyclic hexapeptide NK-1 antagonists. Use of a minilibrary in transforming a peptidal somatostatin receptor ligand into an NK-1 receptor ligand via a polyvalent peptidomimetic, J.Med.Chem., 39, 2441–2448 (1996). [360] D. Giannotti, E. Perrotta, C. Di Bugno, R. Nannicini, N. J. S. Harmat, A. Giolitti, R. Patacchini, A. R. Renzetti, L. Rotondaro, S. Giuliani, M. Altamura and C. A. Maggi, Discovery of potent cyclic pseudopeptide human tachykinin NK-2 receptor antagonists, J.Med.Chem., 43, 4041–4044 (2000). [361] P. Grieco, A. Carotenuto, P. Campiglia, E. Zampelli, R. Patacchini, C. A. Maggi, E. Novellino and P. Rovero, A new, potent urotensin II receptor peptide agonist containing a pen residue at the disulfide bridge, J.Med.Chem., 45, 4391–4394 (2002).

130

ASPECTS OF PEPTIDOMIMETICS

[362] P. Grieco, A. Carotenuto, P. Campiglia, L. Marinelli, T. Lama, R. Patacchini, P. Santicioli, C. A. Maggi, P. Rovero and E. Novellino, Urotensin-II receptor ligands. From agonist to antagonist activity, J.Med.Chem., 48, 7290–7297 (2005). [363] S. Foister, L. L. Taylor, J. J. Feng, W. L. Chen, A. Lin, F. C. Cheng, A. B. Smith and R. Hirschmann, Design and synthesis of potent cystine-free cyclic hexapeptide agonists at the human urotensin receptor, Org.Lett., 8, 1799–1802 (2006). [364] A. Carotenuto, P. Griecio, P. Rovero and E. Novellino, Urotensin-II receptor antagonists, Curr.Med.Chem., 13, 267–275 (2006). [365] S. Flohr, M. Kurz, E. Kostenis, A. Brkovich, A. Fournier and T. Klabunde, Identification of nonpeptidic urotensin II receptor antagonists by virtual screening based on a pharmacophore model derived from structure–activity relationships and nuclear magnetic resonance studies on urotensin II, J.Med.Chem., 45, 1799–1805 (2002). [366] F. Lehmann, A. Pettersen, E. A. Currier, V. Sherbukhin, R. Olsson, U. Hacksell and K. Luthman, Novel potent and efficacious nonpeptidic urotensin II receptor agonists, J.Med.Chem., 49, 2232–2240 (2006). [367] D. K. Luci, S. Ghosh, C. E. Smith, J. Qi, Y. Wang, B. Haertlein, T. J. Parry, J. Li, H. R. Almond, L. K. Minor, B. P. Damiano, W. A. Kinney, B. E. Maryano and E. C. Lawson, Phenylpiperidine-benzoxazinones as urotensin-II receptor antagonists: Synthesis, SAR, and in vivo assessment, Bioorg.Med.Chem.Lett., 17, 6489–6492 (2007). [368] P. M. Fischer, Peptide, peptidomimetic, and small-molecule antagonists of the p53-HDM2 protein–protein interaction, Int. J. Pept. Res. Ther., 12, 3–19 (2006). [369] J. K. Murray and S. H. Gellman, Targeting protein–protein interactions: Lessons from p53/MDM2, Biopolymers, 88, 657–686 (2007). [370] A. S. Dudkina and C. W. Lindsley, Small molecule protein–protein inhibitors for the p53-MDM2 interaction, Curr. Top. Med. Chem., 7, 952–960 (2007). [371] C. Garcia-Echeverria, P. Chene, M. J. J. Blommers and P. Furet, Discovery of potent antagonists of the interaction between human double minute 2 and tumor suppressor p53, J. Med. Chem., 43, 3205–3208 (2000). [372] K. Sakurai, C. Schubert and D. Kahne, Crystallographic analysis of an 8-mer p53 peptide analogue complexed with MDM2, J. Am. Chem. Soc., 128, 11000–11001 (2006). [373] J. A. Kritzer, O. M. Stephens, D. A. Guarracino, S. K. Reznik and A. Schepartz, beta-Peptides as inhibitors of protein–protein interactions, Bioorgan. Med. Chem., 13, 11–16 (2005). [374] J. A. Kritzer, M. E. Hodsdon and A. Schepartz, Solution structure of a beta-peptide ligand for hDM2, J. Am. Chem. Soc., 127, 4118–4119 (2005). [375] F. Bernal, A. F. Tyler, S. J. Korsmeyer, L. D. Walensky and G. L. Verdine, Reactivation of the p53 tumor suppressor pathway by a stapled p53 peptide, J. Am. Chem. Soc., 129, 2456 (2007). [376] K. Sakurai, H. S. Chung and D. Kahne, Use of a retroinverso p53 peptide as an inhibitor of MDM2, J. Am. Chem. Soc., 126, 16288–16289 (2004). [377] T. Hara, S. R. Durell, M. C. Myers and D. H. Appella, Probing the structural requirements of peptoids that inhibit HDM2–p53 interactions, J. Am. Chem. Soc., 128, 1995–2004 (2006).

REFERENCES

131

[378] R. Fasan, R. L. A. Dias, K. Moehle, O. Zerbe, D. Obrecht, P. R. E. Mittl, M. G. Grutter and J. A. Robinson, Structure–activity studies in a family of betahairpin protein epitope mimetic inhibitors of the p53-HDM2 protein–protein interaction, ChemBioChem, 7, 515–526 (2006). [379] Luke, R. W. A., Jewsbury, P. J. and Cotton, R., WO 2000015657, 2000. [380] R. M. Zhang, T. Mayhood, P. Lipari, Y. L. Wang, J. Durkin, R. Syto, J. Gesell, C. McNemar and W. Windsor, Fluorescence polarization assay and inhibitor design for MDM2/p53 interaction, Anal.Biochem., 331, 138–146 (2004). [381] L. T. Vassilev, B. T. Vu, B. Graves, D. Carvajal, F. Podlaski, Z. Filipovic, N. Kong, U. Kammlott, C. Lukacs, C. Klein, N. Fotouhi and E. A. Liu, In vivo activation of the p53 pathway by small-molecule antagonists of MDM2, Science, 303, 844–848 (2004). [382] M. D. Cummings, C. Schubert, D. J. Parks, R. R. Calvo, L. V. LaFrance, J. Lattanze, K. L. Milkiewicz and T. B. Lu, Substituted 1,4-benzodiazepine-2,5-diones as alpha-helix mimetic antagonists of the HDM2-p53 protein–protein interaction, Chemical Biology & Drug Design, 67, 201–205 (2006). [383] J. J. Marugan, K. Leonard, P. Raboisson, J. M. Gushue, R. Calvo, H. K. Koblish, J. Lattanze, S. Y. Zhao, M. D. Cummings, M. R. Player, C. Schubert, A. C. Maroney and T. B. Lu, Enantiomerically pure 1,4-benzodiazepine-2,5-diones as Hdm2 antagonists, Bioorg.Med.Chem.Lett., 16, 3115–3120 (2006). [384] K. Ding, Y. Lu, Z. Nikolovska-Coleska, S. Qiu, Y. S. Ding, W. Gao, J. Stuckey, K. Krajewski, P. P. Roller, Y. Tomita, D. A. Parrish, J. R. Deschamps and S. M. Wang, Structure-based design of potent non-peptide MDM2 inhibitors, J.Am.Chem.Soc., 127, 10130–10131 (2005). [385] K. Ding, Y. P. Lu, Z. Nikolovska-Coleska, G. P. Wang, S. Qiu, S. Shangary, W. Gao, D. G. Qin, J. Stuckey, K. Krajewski, P. P. Roller and S. M. Wang, Structure-based design of spiro–oxindoles as potent, specific small-molecule inhibitors of the MDM2–p53 interaction, J.Med.Chem., 49, 3432–3435 (2006). [386] A. L. Bowman, Z. Nikolovska-Coleska, H. Z. Zhong, S. M. Wang and H. A. Carlson, Small molecule inhibitors of the MDM2–p53 interaction discovered by ensemblebased receptor models, J. Am. Chem. Soc., 129, 12809–12814 (2007). [387] Y. P. Lu, Z. Nikolovska-Coleska, X. L. Fang, W. Gao, S. Shangary, S. Qiu, D. G. Qin and S. M. Wang, Discovery of a nanomolar inhibitor of the human murine double minute 2 (MDM2)–p53 interaction through an integrated, virtual database screening strategy, J.Med.Chem., 49, 3759–3762 (2006). [388] F. Lu, S. W. Chi, D. H. Kim, K. H. Han, I. D. Kuntz and R. K. Guy, Proteomimetic libraries: Design, synthesis, and evaluation of p53–MDM2 interaction inhibitors, J.Comb.Chem., 8, 315–325 (2006).

4 Design of Cyclic Peptides Oliver Demmer, Andreas O. Frank and Horst Kessler

4.1 4.1.1

INTRODUCTION Pharmaceutical Research Today

Preclinical drug development has changed its paradigms several times over the years. Since the focus shifted more and more toward rational strategies, concepts for computational drug design were developed. However, it became obvious that such approaches had great difficulties due to mutual conformational adaptation in the docking process (flexible keys and locks) and unpredictable water binding in the molecular interphase [1]. Thus, in most cases, high-throughput screening (HTS) assays, which are almost exclusively focused on small organic molecules, serve as the available tool for discovering new drug molecules. Due to the high number of ‘false positives’ and absent hits for several target receptors, the limitations of HTS are obvious even when huge libraries are tested for biological activity. In conclusion, chemical biology has progressed to explore the ‘chemical space’, but it remains difficult to realize a substantial part of the ‘biological space’, i.e. to achieve sufficient but also directed diversity in practice [2,3]. Due to the problems of HTS and common rational design approaches, the interest of medicinal chemists turned back to peptides as potential drug molecules. Although having a high biochemical potential (as Peptide and Protein Design for Biopharmaceutical Applications © 2009 John Wiley & Sons, Ltd. ISBN: 978-0-470-31961-1

Edited by Knud J. Jensen

134

DESIGN OF CYCLIC PEPTIDES

described in detail below), unmodified peptide sequences are poor drugs because they are usually cleaved by enzymes in vivo within a short time, thus losing all biological activity. To overcome this obstacle, various concepts for peptide modifications have been suggested. Among the various approaches, cyclization of a linear peptide strand seems to be the most promising way to achieve peptidic or peptidomimetic molecules that display druglike properties. When the cyclized peptide matches the bioactive conformation – the structure required to bind strongly to a receptor and activate or inhibit it – superactivity and often receptorsubtype selectivity is found [4].

4.1.2

General Advantages of Cyclic Peptide Structures

In general, use of peptides in pharmaceutical applications provides many advantages compared to small organic molecules. They usually have low to zero pharmacotoxicity (even though peptidic toxins are known) because their enzymatic degradation results in biogenic, nontoxic molecules which are either excreted or further used in metabolism. However, linear peptides are cleaved in vivo within minutes, before they develop their pharmaceutical function, making them unattractive drug candidates (the term ‘peptide’ originates from the Greek word pEpto& (peptos), which means ‘digestible’). Cyclization of peptides leads to a considerable increase of stability in serum because exopeptidases cannot cleave the peptide at its (nonexistent) ends. Cyclic peptides, especially those with a small ring size, are also protected against endopeptidases. Due to the constrained peptide backbone, which blocks the adaptation of the usually required extended conformation, an enzymatic cleavage is no longer possible. In contrast to the above-mentioned benefits, it should be pointed out that peptide cyclization is usually accompanied by a loss of activity in its initial step. However, when the peptidic cycle is constructed in such a way that it matches the required conformation for a given target receptor, activity and selectivity can be crucially enhanced compared to the linear precursor. In this context, small cyclic peptides (five to six amino acids) are excellent, well-investigated scaffolds for the design of highly active and selective drug candidates [5–15]. In contrast to frequent statements in the literature, small peptide cycles are in most cases very stable against enzymatic degradation and sometimes an adequate bioavailability is achieved. The latter aspects make cyclic peptides even more interesting for medicinal research.

INTRODUCTION

135

Finally, restriction of cyclic peptides often allows a reliable determination of 3D structures in solution as linear peptides are conformationally inhomogeneous due to their intrinsic flexibility. Accordingly, identification of pharmacophoric groups is possible in many cases that may further serve as starting points for an actually rational design of first ‘hits’.

4.1.3

Examples of Cyclic Peptides of Medicinal Interest

Cyclic peptides have been used in medicine as active ingredients of natural extracts for thousands of years. In modern medicine, their application was rediscovered in the middle of the last century, when the cyclic decapeptide antibiotic Gramicidin S was isolated together with linear gramicidins from the bacterium Bacillus brevis [16–18]. At almost the same time, an assortment of polymixins was extracted from Bacillus polymyxa [19–21]. Both antibiotics are administered as mixtures of naturally occurring peptides and are used in anti-infective therapy. Since the late 1950s, the cyclic peptides polymixin B and E have been used as antibiotics under the name Colistin. Despite their toxicity, these branched cyclic decapeptides recently had a kind of revival because multidrug-resistant gram-negative bacterial strains arose which could only be treated by a combination of Colistin and Rifampicin. Among the plethora of cyclic peptides that are of medicinal interest, some representative examples have found important applications in disease treatment. To show the high pharmacological potential of cyclic peptides, some natural products are briefly presented below. 4.1.3.1 Cyclosporin A as a prominent drug compound One of the best-known cyclic peptides in pharmaceutical use is the immunosuppressive agent cyclosporin A (Sandimmun [22–25]; see Figure 4.1), which is one of the best-selling drugs worldwide over recent decades. Its turnover of about 1 billion USD p.a. is about one third of the whole market for pharmacological peptides (excluding insulin) [26]. Cyclosporin A is an undecapeptide with high lipophilicity, and seven out of its eleven amide bonds are N-methylated. Depending on the environment in which the peptide is located, differing conformations are observed. Therefore cyclosporin has a different structure when

136

DESIGN OF CYCLIC PEPTIDES

Figure 4.1 The cyclic peptide cyclosporin A. This natural product is one of the bestselling drug compounds in the peptidic agent market

bound to its receptor cyclophilin [23], compared to when the isolated molecule is crystallized or placed in solution [24,25]. Cyclosporin A was discovered in 1972 and has been used since 1980 to suppress immune responses when transplants are grafted, and to treat autoimmuneinduced diseases like rheumatoid arthritis.

4.1.3.2 Further medium-sized and larger natural products Other important natural cyclopeptides in the pharmaceutical market are vasopressin, oxytocin [27], vancomycin [28] and insulin [29], as well as the recently-developed Integrilin [30] (see Section 5.1). Many of these peptides were originally discovered as hormones and often served as starting point for drug design (see e.g. somatostatin [31], enkephalin [32–34] and melanotan [35]). Cyclization of these peptides (except vancomycin) is achieved via disulfide bonds between two cysteine residues. In general, disulfide bridging is the most common way to reduce the conformational space of peptides and proteins in nature. However, evolution was never – or at least rarely – faced with having to ‘construct’ a cyclic peptide that encompasses all important druglike properties, such as protection against serum enzymes and in particular human

INTRODUCTION

137

digestive enzymes (stability), and adequate uptake from gut into blood (bioavailability). Larger cyclic peptides have also been explored as potential drugs. A particular group of them is dubbed the cyclotides, which are small proteins with remarkable stability toward temperature, pH and enzymatic degradation [36–38]. This stability is attributed to conformational restriction via strong interactions between helices, as in compound A of Figure 4.2, or several disulfide bridges, as present in cyclotide B and peptide C.

Figure 4.2 Large cyclic peptides are found in plants, bacteria and mammals. The peptide backbone of bacteriocin (A; 70 AA) is stabilized by its five alpha helices, while cyclotide Kalata B1 (B; 29 AA) and the immune defensin RTD-1 (C; 18 AA) are constrained by disulfide bonds [36, 37]

4.1.4

General Considerations

Regarding peptides as drug candidates that could be put on the market, activity, selectivity, and stability (especially in serum) are the most important prerequisites. However, even when fulfilling these conditions, most peptidic drugs must be injected, because they are usually not bioavailable. In order to design compounds that could be orally administered, resistance against the highly aggressive enzymes in the

138

DESIGN OF CYCLIC PEPTIDES

brush border membrane of the gut must be achieved. Furthermore, paracellular (via tight junctions) or transcellular uptake is required. For some compounds, for instance di- or tripeptides, active transporters exist (e.g. the peptide transporter PEPT1 [39]), which could hypothetically be exploited as a natural ‘transmission machinery’. So far, only cyclic peptides with multiple N-methylations such as cyclosporin A are orally available to a great extent. Therefore, it might be possible that cyclization of peptide templates in combination with alterations like N-methylation will allow control of biological activity, selectivity, stability and bioavailability. In the following sections, the widespread effects of cyclization on peptides with respect to druglike properties are discussed. After treating the basic concepts underlying peptide cyclization and the synthetic pathways for generating such compounds (Section 4.2), conformational and dynamic features of cyclic peptides are illustrated (Section 4.3). Moreover, strategies for the (semi)rational design of cyclic and modified peptide drug candidates are presented (Section 4.4). Finally, three cases of successful design of cyclic peptides with high medicinal potential are given (Section 4.5).

4.2 4.2.1

PEPTIDE CYCLIZATION Possibilities of Peptide Cyclization

Biologically active, linear peptide sequences are mostly derived from parts of proteins or by screening of synthetic libraries prepared on solid phase. In addition, epitope mapping, sequence comparison between different binding proteins, structural consideration, phage display or pepscan methods provide promising peptidic templates [40,41]. Alternatively, natural products that show the desired pharmacological effects can serve as a starting point for further optimization. Usually templates are found which exhibit a reasonable to good binding affinity but often lack the pharmacological properties like specificity and stability. There are different ways to cyclize peptides. The linear peptide strand can be cyclized not only from head to tail, connecting the Cand N-termini, but also by linking to amine and carboxylic functions in amino acid side chains, giving side-chain-to-head or side-chain-to-tail connections. Side-chain-to-side-chain bridges have also been described (Figure 4.3) [42].

PEPTIDE CYCLIZATION

139

Figure 4.3 Different modes of cyclization for peptides. Left: conventional cyclization. Right: backbone cyclization with all possible connections to backbone amides; instead of the amide, a Ca atom can be analogously connected

Another possibility for linking a linear peptide strand is the socalled backbone cyclization. This method restrains the linear peptide strand by inserting a bridge from a backbone amide nitrogen or a Ca carbon atom to any other position [43]. Such connections have the advantage that the peptide chain termini remain free, which is important for peptide hormones that are often found as N-terminal primary amides, for example. However, it requires more synthetic effort than conventional peptide cyclization, as unnatural amino acids are incorporated. Generally speaking, the actual problem is to find restraints that fix the peptide into a conformation that is still recognized by the receptor (matched case) [4]. In most cases the constraints cluster a molecule in a family of conformations that does not contain the bioactive conformation (mismatched case). This becomes plainly evident when comparing the multitude of possible conformational families that represent mismatched cases with the sparse amount of matched cases. This is analogous to a jigsaw puzzle, where of the many similar pieces only one fits exactly in the designated place.

4.2.2

Synthesis of Cyclic Peptides

In order to obtain cyclic peptides, sophisticated synthetic strategies have been developed. The most common is head-to-tail cyclization of linear precursors. The synthetic yields vary strongly with ring size,

140

DESIGN OF CYCLIC PEPTIDES

specific structure, and even differing sequences that lead to the same cyclic peptide. For more details, the reader is referred to the recent literature [42]. One of the most efficient methods of cyclizing a linear peptide strand from head to tail is to use a combination of diphenylphosphoryl azide (DPPA) and NaHCO3 in dimethylformamide (DMF). DPPA is a mild cyclization reagent that helps to suppress racemization. On one hand, NaHCO3 is a weak base which helps to reduce racemization of the amino acid building blocks; on the other, it is insoluble in DMF. Therefore the reaction occurs only at the surface of the bicarbonate (solid base method) and the low concentration of deprotonated ammonium groups prevents the formation of cyclic dimers and higher multimers of the peptide strand. Another useful cyclization reagent especially suited for secondary amines is 2-(7-aza-1H-benzotriazole1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HATU), as the C-terminus is more strongly activated and racemization rates are still low. Increasing yields are found when the conformation of the linear precursor allows the N- and C-terminal groups of the peptide chain to come close to each other. This is the case in sequences containing both D- and L-residues, proline, or other N-alkylated amino acids, since preformation of turns is induced, which reduces the distance between the termini. For example, linear pentapeptides exclusively built of L-amino acids are difficult to cyclize. However, when an N- or C-terminal D-amino acid is present, cyclization occurs smoothly [42]. Also, N-alkylated amino acids can favour cyclization, e.g. a linear tripeptide containing solely L-amino acids is hardly convertible into a cyclic tripeptide since it usually forms cyclic dimers, the hexapeptides. On the other hand, there is no problem in the cyclization of triproline or trisarcosine. Apart from cyclizing linear peptide strands in solution, modern orthogonal protecting group strategies allow ring closure on solid support with standard coupling reagents. For example, the 4-(N-[1(4,4-dimethyl-2,6-dioxocyclohexylidene)-3-methylbutyl]amino)benzyl (Dmab) protection group for acid functionalities is cleaved under mild conditions and can be combined with the Fmoc strategy to synthesize cyclic, resin-bound peptides. However, one should be aware that the concentration of the growing peptide chain on the resin is relatively high. Hence, usually higher amounts of unwanted cyclic dimers or even higher oligomers are formed compared to cyclization in solution where low concentrations are present (principle of dilution).

PEPTIDE CYCLIZATION

4.2.3

141

Chemical Modifications of Cyclic Peptides

There are numerous possibilities for chemical alterations of a cyclic peptidic precursor molecule. Modifications retaining the peptide chain include variation of amino acid stereochemistry and of the peptide strand orientation. Other alterations are the introduction of small groups (e.g. N-methylation of the peptide bond) or replacement of complete amino acid building blocks by nonproteinogenic ones. Although all these modifications have been applied to linear peptides, they have quite a different impact on cyclic molecules. When incorporated in peptidic cycles, the entire backbone strand is influenced via its global connection, unlike a linear compound, where just a local area is affected. An overview is given in the following sections.

4.2.3.1 Peptide strand arrangements The simplest modification, which is single substitution of natural L-amino acids by D-enantiomers, usually has drastic effects on conformation and biological activity. In linear peptides, a D-amino acid scan may provide information on possible turn structures important for biological activity. However, such a modification is hardly detectable (e.g. by NMR spectra) as it only leads to a shift in the thermodynamical equilibrium of a very large number of conformations. In cyclic peptides, the influence of a change in stereochemistry on the conformation is much more pronounced, because cyclization already results in a dramatic reduction in the number of distinct conformations. Complete substitution of all-L-amino acids by D-enantiomers leads to inverso peptides, which lose their biological activity in most cases [44,45]. A few exceptions are known where such peptides interact with ‘achiral receptors’. In retro peptides, the direction of the peptide bond and therefore the primary sequence is reversed [46]. Similarly to inverso peptides, compounds constructed in this way also lose their biological activity, as the orientation of the amino acid side chains is different from the parent peptide. To yield a comparable orientation of the side chains, the retroinverso concept was developed [47–51]. Here, the backbone chirality of the amino acid building blocks of a retro peptide is changed (see Figure 4.4). Neglecting the fact that – in relation to the template structure – the backbone conformation could be influenced by this kind of chemical alteration, retroinverso peptides exhibit their side chains in

142

DESIGN OF CYCLIC PEPTIDES

Figure 4.4 Modifications of the peptide backbone by change of the orientation of the amide bonds (indicated by the arrows) and/or chirality

the same orientation as the original peptide. The advantage of this approach is that the high content of D-amino acids makes retroinverso peptides more stable against enzymatic degradation. However, if hydrogen bridging between amide bonds in the peptide and certain groups in the target is essential for interaction, retroinverso peptides lose their activity, since the positions of hydrogen bond donors and acceptors are altered. For smaller cyclic peptides, complete retroinverso peptides can easily be prepared. In contrast, partial retroinverso peptides are more difficult to synthesize as they contain unusual diamino or dicarboxylic building blocks.

4.2.3.2 Peptide bond modifications Linear or cyclic peptides containing only standard peptide bonds are called homodetic, while in heterodetic peptides different functional groups are additionally used to connect amino acids. Such modifications include esters (depsipeptides), thioamides, reduced amide bonds, ethers, thioethers, sulfoxides, carbon double bonds, heterocycles and many others. A few examples of peptide bond mimetics are shown in Figure 4.5. The consequences of changes in steric and electronic properties are manifold. Beside distortions of local geometry, and a modified mobility of the artificial peptide bond, the global conformation of the peptide cycle is altered compared to the linear precursor. It is almost impossible to investigate the participation of a distinct peptide bond in biological activity by introducing a mimetic in cyclic peptides. Conclusions can only be drawn when conformational changes upon an alteration are

PEPTIDE CYCLIZATION

143

Figure 4.5 A small number of chemical modifications of the peptide bond that are more or less suitable mimetics. The general structure of peptoids and azapeptides is given on the right side in the framed section

completely explored [52]. A seemingly simple substitution of an amide bond by a sulfonamide illustrates the dramatic effects on the overall conformation, as it has a preferred orientation that is not planar but has a bond angle of 90° relative to the original peptide bond. This effect has recently been used in peptidomimetics to switch selectivity between two integrin receptor subtypes [53]. Although a small chemical modification, N-methylation of amide bonds can have a strong influence due to incorporation of ‘steric hindrance’, loss of hydrogen bond donor ability and changes in the partial charge pattern [42,54,55]. Thus, N-methylation usually modifies the properties of cyclic peptides drastically. Steric effects become increasingly important the more N-methyl groups are introduced. In particular, exposure of NH groups interacting with transporters, receptors or other biomolecules is altered, finally resulting in strong biological effects like changed selectivity [56]. It is also observed that peptide bond cleaving enzymes (proteases) cannot work properly, leading to an enhanced metabolic stability. After problems concerning synthesis of multiple N-methylated peptides were solved [57], this modification was applied to small cyclic penta- and hexapeptides to achieve oral availability of biologically active peptides [58], as for example in cyclosporin (see Section 4.1.3.1). Other types of modification that indirectly affect the peptide bond are azapeptides [59] or peptoids [60], which are introduced in linear

144

DESIGN OF CYCLIC PEPTIDES

and cyclic peptides (see framed section in Figure 4.5). In an azapeptide, the Ca-proton moiety of a peptide is substituted by nitrogen, whereas in peptoids the side chains of the amino acids are shifted to the amide nitrogen. In both mimetics, the chiral information of the former Ca is lost, which may lead to global conformational changes in cyclic peptides. Peptide bond modifications are also present in natural products: oxazoles and thiazoles have been found in non-ribosomally synthesized peptides. It is obvious that such drastic changes have a strong impact on activity and conformation. When incorporated in peptidic cycles, the entire backbone strand is influenced, as opposed to a linear compound, where just a local area is affected.

4.2.3.3 Amino acid side-chain alterations The most commonly observed and utilized side-chain alteration is cysteine bridging in cyclic peptides and proteins, which is easily formed by oxidation of two thiol functions of the amino acid side chains. Disulfide bonds help to stabilize the tertiary and quaternary structures of proteins and also play an important role in constraining peptide conformations. However, the resulting cycles usually retain considerable flexibility. In contrast to homodetic head-to-tail cyclized peptides, even small peptides that contain a side-chain linkage are conformationally inhomogeneous. For example, head-to-tail cyclized tetrapeptides exhibit a strongly reduced conformational space, whereas a disulfide-bridged tetrapeptide of the general structure CXYC (C: Cys; X,Y: any amino acid) allows a multitude of conformations [61]. Although it is claimed in the literature [62] that such Cys bridges stabilize b-turns, a simple comparison of published structures of b-turns [63] shows that the distance of the b-carbons of the cysteine residues is too large to form a b-turn when they are bridged by a disulfide bond. Hence, standard b-turns cannot be formed in such structures. Another drawback of disulfide bonds is their easy reductive cleavage under many physiological conditions, e.g. in the cytosol. A further obvious way to cyclize linear peptide strands is linkage via amide bonds between carboxylic and amine functionalities. Such connections are also found in many natural and unnatural amino acids. In principle, all kinds of feasible chemical modifications could be introduced for cyclization. However, much higher synthetic efforts are necessary because not only the functionalities themselves have to be built, but

PEPTIDE CYCLIZATION

145

an appropriate protection group strategy for the remaining amino acid building blocks must be developed. For special cases, like the incorporation of N-alkylated side chains, standard protocols have been developed to introduce nonproteinogenic functionalities to amino acids [64]. Further modifications are also feasible as the protection group strategy was optimized to be orthogonal and compatible to standard Fmoc solidphase peptide synthesis (SPPS).

4.2.3.4 Modifications in natural peptides In nature, e.g. in bacteria, fungi, lower animals and plants, an enormous number of homo- and heterodetic cyclic peptides are found, including most of the aforementioned alterations [65–67]. Two prominent examples of (poly)cyclic peptides are given in Figure 4.6. On the left side, vancomycin is shown, which was a last-resort antibiotic until resistant strains arose in the late 1980s. On the right side, phalloidin is depicted, which belongs together with the amatoxin family to the poisonous ingredients of the death cap mushroom (Amanita phalloides). In these examples, typical modifications like (thio)ethers and adjoined sugars, in combination with nonproteinogenic amino acids, form cyclic substances that disguise the peptidic origin of the natural products at first glance.

Figure 4.6 Natural products containing complex cyclic peptide structures. On the left side is the antibiotic vancomycine and on the right side phalloidin, one of the poisonous substances composing the death cap mushroom

146

4.2.4

DESIGN OF CYCLIC PEPTIDES

Concluding Remarks

It is beyond the scope of this chapter to give a complete description of all feasible cyclization methods and peptide modifications, and their influences on the conformation of cyclic peptides. In our opinion, the most promising way to turn an interesting (e.g. active) linear peptide into a cyclic molecule with druglike properties is head-to-tail cyclization, followed by the introduction of D-amino acids and, subsequently, multiple N-methylation. Afterwards, side-chain modifications like the exchange of natural for nonproteinogenic amino acids or specific modifications of distinct functional groups may further improve a peptidic drug candidate, thus, at best, finally resulting in a ‘lead compound’.

4.3

CONFORMATION AND DYNAMICS OF CYCLIC PEPTIDES

Cyclization of peptides leads to compounds displaying features that are completely different from those of the linear precursors. In particular, conformation and dynamics are strongly altered. The differing characteristics of cyclized peptides tremendously bias the biological profile in relation to the linear template. Since the effects of dynamics are not yet fully explored, the main focus of this section is on the conformational influences of cyclization on druglike properties.

4.3.1

Reductions in Conformational Space

Linear peptides are flexible molecules that exhibit a plethora of different conformations in solution. Considering a minimum energy barrier of kT between two conformations, a tetrapeptide with six rotatable bonds in the peptide chain can already adopt thousands of different 3D arrangements. However, a receptor recognizes and binds to only a few of these conformations [68]. Binding to a receptor requires mutual adaption of both the conformation of the receptor and that of the ligand. If constraints can be introduced in the flexible ligand that fix it in the bioactive conformation or in a family of conformations close to it, the loss of entropy upon binding is reduced and/or the negative-binding energy increased. Both effects lead to

CONFORMATION AND DYNAMICS OF CYCLIC PEPTIDES

147

an overall stronger binding (higher potency) that can even surpass the affinity of a natural binder. As explained above, cyclization of a linear peptide chain induces a preference in backbone conformations (often called increasing ‘rigidity’). This effect is frequently accompanied by the formation of preferred sidechain conformations not present in linear peptides. While the spatial arrangement of the side chains is generally the most important factor in the 3D molecular recognition process, the peptide backbone primarily serves as a template for their orientation. Unique structures enable selectivity between receptor subtypes, as these have often evolved to recognize distinct conformations of the same binding motif [69]. Receptor-subtype selectivity is important in preventing drugs from binding to a whole family of similar receptors instead of a single target, thereby causing unwanted side effects. However, one has to remember that terms like ‘rigid’ and ‘the bioactive conformation’ refer not to a single fixed 3D structure but to an assembly of closely related conformers in (fast) exchange with one another [70–72]. In general, one can say that the smaller the size of the cyclic peptides, the higher the conformational restraints and the less the observable flexibility. A graphical representation of the effects of conformational restriction is schematically shown in Figure 4.7. The more the

Figure 4.7 Reduction of the multidimensional conformational space into two dimensions: three different receptors, A, B and C, require different ligand conformations. Different chemical modifications (cyclizations) restrain peptide conformations to a, b, c or d. The effect of the restrictions to activity and selectivity is illustrated

148

DESIGN OF CYCLIC PEPTIDES

conformational space of ligands and receptors (depicted as circles) overlaps, the stronger the binding (higher activity) and the better the selectivity for a receptor subtype. Molecule a has all desired characteristics while b, c, and d lack selectivity, activity or both due to unfavourable restraints.

4.3.2

Conformational Arrangements in Cyclic Structures

4.3.2.1 Peptide bonds in cis and trans orientation The partial double-bond character of the peptide bond results in cis–trans conformers, which are separated by a rotational barrier of about 60–80 kg/mol [73]. This corresponds to a lifetime of each conformation of about 0.1 seconds at room temperature, yielding separate signals for cis and trans forms in NMR spectra of peptides. Within the cis–trans equilibrium, the higher allylic strain [74,75] in the cis conformer results in a higher preference for the trans form of all secondary peptide bonds (-CONR- instead of -CONH-; see Figure 4.8). Therefore, peptides containing proline or N-alkylated amino acids have a higher propensity of cis peptide bonds about the N-alkylated amide bond because both substituents at the amide nitrogen (the alkyl group and the Ca) are of similar size (see Figure 4.8) [42,76–81].

Figure 4.8 cis/trans isomers and allylic strain in normal and alkylated amide bonds

CONFORMATION AND DYNAMICS OF CYCLIC PEPTIDES

149

Usually the cis conformer is not observed in the NMR spectrum of linear peptides. However, it has been found that the population of cis conformers can reach levels of 0.1–1.0% in linear peptides [76]. Also, homodetic headto-tail cyclized peptides containing five, six or more amino acids always prefer all-trans peptide bonds like linear peptides. Only in rare cases is a population of more than 5% cis of a secondary peptide bond found [78]. By contrast, cyclic tetrapeptides usually contain one or more cis peptide bond, and in even smaller cyclic tripeptides [82,83] and dipeptides (diketopiperazines), trans conformations are rare or impossible. 4.3.2.2 Turn structures and their relevance for biological activity All structural elements are defined by the dihedral angles in the peptide backbone. The o angle defines the cis (0°) and trans (180°) arrangement about the peptide bond; j and c specify the other backbone angles from N to Ca or Ca to CO, respectively; and the w angle defines the orientation of side chains. The Ramachandran plot exhibits the potential energy landscape as function of the j and c angles. In small cyclic peptides the preferred Ramachandran angles are j ¼ 120° and c ¼ 60°, which typically occur in turn structures or helical motifs [84]. A c angle of 120° is energetically preferred but results in an extended conformation, which is prevented by the cyclization in small rings. There is a very strong tendency for the preference of j ¼ 120° because the allylic strain is minimized due to the smallest residue at the a-carbon being syn-periplanar oriented with respect to the largest residue at the nitrogen, the CO group. In addition, the syn orientation of the oppositely charged dipoles of C ¼ O and Ca-H bonds leads to electronic stabilization (Figure 4.9).

Figure 4.9 Preferred j and c angles induced by the allylic strain in the peptide bond

150

DESIGN OF CYCLIC PEPTIDES

The resulting turn conformations in constrained cyclic peptides can mimic loop structures of proteins, which are often found as recognition elements in protein–protein interactions. These loops are exposed at the surface of the protein or are used in peptidic ligands as elements which dock into the binding pocket of a target [85–87]. In these loop regions, b- and g-turns are convex structures with a compact and strongly defined folding of the peptide backbone that are recognized by concave surfaces of receptors. Cyclic penta- and hexapeptides frequently incorporate b- and/or g-turn-like structures [87–97] and consist of sufficient amino acid building blocks for incorporation of the required recognition motif. Hence, small- and medium-sized cyclic peptides may serve as scaffolds for mimicking turn structures of pharmacologically active parts of a protein [98]. Having reached this level of affinity, the selectivity and activity (agonistic or antagonistic) can be further enhanced by structural optimization [4,87,99]. In general, this knowledge enables prediction of a rigid backbone conformation and the 3D presentation of the pharmacophoric groups on the side chains, thus allowing rational conformationally based design.

4.3.2.3 Hydrogen-bond networks Hydrogen-bond networks are another typical trait encountered in cyclic peptides [100–102]. Hydrogen bridges typically occur in turn structures (Figure 4.10) between NH of the last amino acid and CO of the first amino acid within a turn (this holds for a-peptides). In bpeptides the formation of hydrogen bonds is more diverse [103]. However, contrary to the general opinion that these hydrogen bonds force the aforementioned structures in cyclic peptides, their

Figure 4.10 Denotation of the rotational angels of a peptide and the relative positions of amino acids in b- (middle) and g- (right) turns

CONFORMATION AND DYNAMICS OF CYCLIC PEPTIDES

151

contribution to the formation of a specific conformation is small [101,104]. Turns in small peptides are mainly determined by steric effects [104]. The stability of the hydrogen bonds is also very sensitive to the surrounding medium. Interaction of peptide bonds with a polar medium can easily break internal hydrogen bonds, while in nonpolar media, like membranes or lipophilic organic solvents, hydrogen bonds can actually impose constraints on the conformational freedom of cyclized peptides.

4.3.3

Flexibility of Cyclized Scaffolds

Compared to linear template structures, cyclic peptides differ not only in structural characteristics but also in displaying a modified dynamical behaviour. Since, in particular, backbone flexibility seems to play an important role for cellular uptake (bioavailability), the dynamics of cyclic peptides and the differences in relation to the linear precursor are treated below. 4.3.3.1 Backbone dynamics Linear peptides intrinsically have several degrees of freedom. As has already been explained, even small compounds can adopt thousands of different conformations. These structures, however, cannot directly be observed, by NMR spectroscopy for example, because they are in fast exchange. Only in cases where stable secondary structural elements in slow exchange are formed, more than one conformation is detectable via NMR. To drastically reduce fast dynamics, head-to-tail cyclization is a particularly suitable method. The main effect of ring closure is visible when analysing the dynamical behaviour of the peptide backbone. Here, most of the degrees of freedom are frozen, resulting in a more or less rigid conformation. If there are more structural arrangements that are separated by a high-energy barrier, several signal sets will appear in the NMR spectrum. At this point, however, it should be explicitly noted that there are still dynamics left; in particular, synchronous dihedral bond flips about both adjacent single bonds flanking a peptide bond (f and c) are found. These motions are usually fast on the NMR time scale, thus only one signal set is observed. This example clearly demonstrates that it is not mandatory for conformational preference to be accompanied by complete rigidity of the peptide

152

DESIGN OF CYCLIC PEPTIDES

backbone. Cyclization always strongly decreases flexibility but does not eliminate dynamics at all.

4.3.3.2 Side-chain flexibility Due to fast thermodynamic equilibrium motions in linear peptides, no information about side-chain orientations is normally obtainable via NMR spectroscopy. In cyclized structures, not only the backbone flexibility but also the dynamics of amino acid side chains is strongly reduced. This is particularly true for the common three rotamers around the Ca-Cb bond. Often, preferred conformations about w1 are observed, which are easily identified and assigned by NMR spectroscopy [105]. For example, measurement of NMR coupling constants reveals whether side-chain rotations are still present (an averaged constant is yielded) or if one orientation is preferred, thus indicating that dynamic processes have more or less vanished. Vice versa, preferred side chains are a good indicator for a preferred backbone as well.

4.3.4

Experimental Structure Characterization

Knowledge of the 3D conformation of cyclic peptides is essential in order to completely understand biological features like activity, selectivity, stability and bioavailability. Hence, in the following we are dealing with the methods present for structure elucidation and errors frequently observed in the literature.

4.3.4.1 Methods for studying conformational features Usually there are three different ways to obtain 3D peptide structures at high resolution. In solid form, X-ray diffraction of single crystals is most commonly used, but solid-state NMR spectroscopy is also frequently applied. For structure elucidation in solution, NMR spectroscopy is used in combination with modelling techniques such as distance geometry (DG) and/or molecular dynamics (MD) calculations. Compared to their overall volume, small- and medium-sized compounds display a huge surface that is accessible for surrounding molecules. Consequently, intermolecular interactions clearly contribute to both internal dynamics

CONFORMATION AND DYNAMICS OF CYCLIC PEPTIDES

153

and conformations in a (cyclic) peptide. Depending on the techniques used for structure determination, the resulting conformational models often differ. In particular, the structures in crystal and solution may be different [106]. Since the solution structure is usually closer to the receptor-bound conformation than the X-ray structure, we clearly prefer the application of liquid-state NMR for structure elucidation. The methods for yielding NMR structures of high quality are well advanced and have recently been reviewed in the literature [107].

4.3.4.2 Reliability of structural models Sometimes 3D structures in solution are presented on the basis of insufficient data material. For example, small chemical-shift changes of certain NH protons in temperature gradient measurements are interpreted as evidence for the presence of a b-turn. However, such a result only indicates shielding from solvent and not necessarily an internal hydrogen bond, which is found in said b-turns. For a precise investigation, a careful analysis by special NMR experiments is essential, which can directly identify hydrogen bridges via J-coupling from 13CO to the 15N involved in a hydrogen bond [108]. As a drawback, this requires expensive isotopic labelling of the cyclic peptide. In addition, there are many other sources of error. NMR structures of cyclic peptides are often calculated via methods originally developed for protein structure determination. Since spin-diffusion processes have only minimal influence on the NOESY/ROESY cross-peak volumes of small- and medium-sized cyclic peptides, distance information should be used as distinct values (upper and lower bounds) and not divided into distance classes. In addition, it is strongly recommended that NMR coupling constants are included for the structure determination, because they provide important information (e.g. dynamics) not available from NOE cross-peak volumes. Furthermore, NMR peptide structures derived from DG calculations are often refined by MD simulations without incorporation of explicit solvent molecules, thus resulting in strongly distorted backbone conformations. This is especially true for cyclic peptides because of the distinct and before-mentioned volume: surface ratio. Finally, 3D models of cyclic peptides determined by X-ray crystallography often display artificial backbone arrangements, since forces resulting from crystal packing usually dominate the conformation.

154

DESIGN OF CYCLIC PEPTIDES

To conclude, the purpose of mentioning these facts is to warn the reader against believing noncritically all published conformations of (cyclic) peptides.

4.4 4.4.1

CONCEPTS IN THE RATIONAL DESIGN OF CYCLIC PEPTIDES The Influence of Amino Acid Composition

The introduction of special amino acids and the use of building blocks with varying chiralities can strongly bias the conformation of cyclic peptides. Since this knowledge contributes evidently to the concepts applied in rational design, a brief overview of the most prominent examples is given below.

4.4.1.1 Glycine and proline as building blocks of special impact Conformational investigations of cyclic peptides by NMR spectroscopy show that their structure is strongly influenced by the amino acids and their chiralities. Glycine, which has no side chain, is usually found in flexible regions and also in proteins when other side chains sterically interfere (see e.g. the GXXXG motif in helix–helix interactions of proteins) [109,110]. Proline is the only natural proteinogenic amino acid with a secondary amino group. Hence, glycine and proline have a strong impact on the conformational preference in cyclic peptides. Glycine is the sterically least hindered amino acid. Therefore, it easily substitutes L- as well as D-amino acids and usually occupies the i þ 1 position in a type II0 b-turn if the remaining amino acids are in L-configuration. Proline, on the other hand, is the conformationally most restrained amino acid and thus induces stronger structural preferences to the remaining peptide chain.

4.4.1.2 The influence of changes in amino acid chiralities The incorporation of amino acids with different chiralities offers an efficient way of actively influencing the side-chain orientation by controlled induction or (re)positioning of turns. Due to steric hindrance

CONCEPTS IN THE RATIONAL DESIGN OF CYCLIC PEPTIDES

155

imposed by the b-carbon [74,111], there is a strong impact on the backbone a-carbon. For cyclic pentapeptides with DLLLL configuration the three amino acids in the ‘south direction’ (see e.g. Figure 4.12) usually adopt a g-turn in equilibrium with a structure that is formed by flipping the amide bond on the lower-left side about its two adjacent j and c angles [95,98]. The same procedure can also be applied to hexapeptides, using cyclic structures containing one D- and five L-amino acids. If we assume that the D-amino acid (e.g. Gly) is in the i þ 1 position of the bII0 turn, the other amino acids opposite this b-turn should adopt another b-turn. This second turn is usually flexible (experimentally, an equilibrium between bI and bII was observed) [96,111]. In addition, the hexapeptides also allow further g-turns in both sides of the extended b-strand, which leads to a bent hexapeptide conformation [56,112]. This knowledge can be used to design a desired conformation of a bioactive sequence, as has been done for tendamistat [111].

4.4.2

The Dunitz–Waser Concept

The structurally most intensely studied cyclic peptides are hexapeptides. Quite early a basic structure containing two internal b-turns was proposed, and later this conformation was confirmed in many crystal and solution structures. This structural preference is even more the case if a 0 D-amino acid is present in the i þ 1 position of a bII -turn. To understand the dynamics and conformational possibilities, it is useful to simplify the view on cyclic peptides to the essential and most prominent features by substituting all peptide bonds with C ¼ C double bonds. A cyclic peptide then corresponds to a nonconjugated cyclic polyene. The conformation of the latter can be reduced to saturated cyclic hydrocarbons using the principle of Dunitz and Waser [113]. A trans double bond can be substituted by a long single bond, and a cis-carbon double bond by a ‘phantom methylene group’ (see Figure 4.11). For example, a cyclic pentapeptide with all-trans peptide bonds is reduced by this procedure to a cyclopentane (Figure 4.12). The flexibility of cyclopentane via pseudorotation with small energetic barriers between the conformers is well known. Hence high flexibility in cyclic all-trans pentapeptides is frequently present. Preferred conformations are usually found, but ‘rigidity’ is not expected [83,114,115]. On the other hand, a cyclic all-trans hexapeptide is transferred into a cyclohexane, and a ‘rigid’ chair (including bIturns) or a flexible boat conformation (including bII-turns) can be expected.

156

DESIGN OF CYCLIC PEPTIDES

Figure 4.11 Simplification of cyclic peptide structures via conversion of peptide bonds from carbon–carbon double bonds cyclopolyenes (Figure 4.12) and applying the Dunitz–Waser concept (see text)

Figure 4.12 The spatial screening method applied to hexa- and pentapeptides in order to find the optimal side-chain orientation. The preference of D-configured amino acids (indicated with lower-case letters and black dots) to occupy the i þ 1 position of the bII0 -turn is used to fix each of them in the upper-left corner of the drawn peptides. A bioactive sequence, e.g. -ABCDE- in the cyclic pentapeptide scaffold cyclo(-D-Ala-Ala4-) and -ABCDEF- in the corresponding hexapeptide with one D-Ala, is then projected in all possible spatial arrangements. The typical turns in the templates are depicted

4.4.3

The Spatial Screening Technique

Having found an active linear peptide sequence, it is normally hard to predict which spatial orientation the pharmacophoric groups might adopt when the peptide is cyclized. In order to find the optimal arrangement, the so-called spatial screening procedure can be applied [12,116–121]. In this procedure, a systematic search for templates of all-alanine peptides which strongly prefer one conformation on the NMR timescale is performed. For this effect, all possible penta- and hexapeptides containing D-Ala and L-Ala were synthesized. Subsequently, the conformations in solution were determined via

CONCEPTS IN THE RATIONAL DESIGN OF CYCLIC PEPTIDES

157

NMR spectroscopy and the most homogeneous ones were selected. For example, this technique revealed that the pentapeptide cyclo(-D-AlaAla4-) displays a very stable conformation containing a bII0 - and a g-turn (see Figure 4.12). Based on this knowledge, a bioactive sequence, for example ABCDE, can now be head-to-tail cyclized, where each of the amino acids is used once in D-configuration. In this way, one obtains five new peptides that have the same constitution (connectivity) but differ in their stereochemistry. Hence, the pharmacophoric side chains are presented in five different conformations (see Figure 4.12). The synthesized peptides have different biological activity, and the bioactive conformation is best represented in the most active peptide (‘spatial screening’). The same procedure can be applied when working with hexapeptides or even larger cyclic peptides.

4.4.4

General Strategy for Finding Active Hits

As pointed out above, cyclic peptides can exhibit superior properties over their linear analogues, such as higher activity, better enzymatic stability and bioavailability, and receptor subtype selectivity. The remaining question is how to design cyclic peptides in which these properties are optimized. In this context, the following procedure is suggested: Step 1: Identification of the target recognition motif. The de novo approach starts with the identification of molecules that bind to a suitable target protein. Such molecules can be small peptidic ligands or large biomacromolecules. To identify the binding sequence, several different techniques can be used, as mentioned in Section 4.2.1. Step 2: Elucidation of the pharmacophoric groups. Having found an attractive binding sequence, an alanine scan [122] is usually performed, in which each amino acid is systematically substituted by alanine. When an essential residue is exchanged, a tremendous drop of biological activity can be observed. For nonessential amino acids, the binding affinity is affected less drastically, which means that these residues can be optimized and/or removed. Step 3: Obtaining initial information of structural arrangements. When the important amino acids are identified, an initial hint about turn structures is

158

DESIGN OF CYCLIC PEPTIDES

gained by replacement of the original amino acids with their corresponding D-enantiomers (D-scan). As was explained in Sections 4.2.3.1 and 4.3.2.2, D-residues within all-L peptides induce turn structures that are essential in biological recognition processes. Double D-scans have also been proposed [123] for this purpose. Step 4: Cyclization of the linear precursor. The most crucial step in the following is the design of cyclic peptides that are still biologically active. Different methods of cyclization can be tested; however, head-to-tail linkage seems to have the most advantages regarding reduced conformational spaces and druglike properties (see Section 4.2.1). Step 5: Spatial screening to explore the bioactive conformation. Since in most cases the bioactive sequence is not matched when having cyclized an active linear precursor, spatial screening (see Section 4.4.3) can be applied. By using different cyclic template peptides, the optimal conformation of the backbone and of the important side chains can be identified, as was described above. Step 6: Optimization of the first ‘hit’ by modifications. Cyclic peptide structures obtained in the described way can be further optimized with respect to their activity, receptor subtype selectivity or bioavailability. To match the bioactive conformation of a specific receptor subtype even better, one can optimize the functional groups in the side chains but also restrict the structure further to mimic the chemical requirements for binding better than the precursor (see Section 4.2.4). In particular, N-methylation of peptide bonds can beneficially contribute. As will be shown in Section 4.5.1, peptides with a high pharmaceutical potential can already be obtained at this point. Step 7: Generating peptidomimetics. If compounds have been designed that lack an important druglike property, further modifications can be performed. Sometimes, it is useful to develop peptidomimetics which share the spatial arrangement of the pharmacophores with the optimized cyclic peptidic structure. In particular, when stability or selectivity of the cyclic peptides is not adequately present, this approach is an efficient tool to finally gain a new drug candidate. In Figure 4.13, a flow chart of the above-described peptide drug design concept is given. The central point is cyclization of peptidic precursors; however, preparation and final optimization steps are also essential for success in this highly interesting field of drug research.

EXAMPLES OF CYCLIC PEPTIDES AS DRUG CANDIDATES

159

Figure 4.13 Flowchart showing approaches from a chosen protein target to peptidic and nonpeptidic drugs

4.5

EXAMPLES OF CYCLIC PEPTIDES AS DRUG CANDIDATES

The following examples will show how small cyclic peptides have been derived from linear or larger cyclic structures and successfully implemented into biomedical research. The illustrated peptides are ligands for transmembrane proteins like integrins or G-protein-coupled receptors (GPCRs). At present, 40% of all drugs target GPCRs, but so far only four mammalian GPCRs (rhodopsin, b1 and b2 adrenergic receptor and the human A2A adenosine receptor) have been structurally determined [124–127]. In spite of recent successes in homology modelling of GPCRs, ligand-oriented design or screening is still the usual way to develop new drugs with GPCRs as target. Therefore, as cyclic peptides allow a controlled modulation of their spatial structure, they are ideally suited to explore the unknown conformational space of GPCRs.

4.5.1

Cilengitide as Integrin Inhibitor

Integrins are a class of heterodimeric receptors of one a and one b subunit found on the surface of most cell lines that facilitate attachment to other

160

DESIGN OF CYCLIC PEPTIDES

cells and the extracellular matrix (ECM) [128]. Moreover, their bidirectional signalling pathways can affect cell growth, differentiation, division, migration and apoptosis. Apart from these functions, integrins are responsible for the binding of certain viruses to the cell and take part in blood coagulation. These characteristics also make them targets for pathological processes like inflammation, cardiovascular disorders, thrombosis, restenosis, vascular homeostasis, osteoporosis, cancer invasion, metastasis and tumour angiogenesis [69,129–132]. The challenge when targeting integrins is to develop selective drugs that discriminate between the receptor subtypes and thereby cause fewer side effects. Starting in the early 1980s, the first steps in finding the binding domain of the nonselective fibronectin were to take fragments of different lengths to determine whether the binding motif was near the N- or C-terminus. The minimal recognition sequence for some intergrins was found to be the RGD motif [133], which is also present in snake venoms (so-called disintegrins). The influence on specificity of different integrin receptor subtypes by the flanking amino acids has been considered, but we investigated the role of the conformation using the spatial screening procedure (see above; Figure 4.12). For that purpose, the sequence RGD was elongated by a N-terminal Val and an C-terminal Phe. This pentapeptide (and additionally the hexapeptide, which contains an additional Gly residue) was introduced in five (or six) different head-to-tail cyclic peptides in which each amino acid is once taken in the D-configuration [9,12,134]. Of the resulting five cyclic pentapeptides, the one which contained a Dphenylalanine turned out to have 1000 times higher activity for the aVb3 receptor subtype but 10 times reduced activity for the aIIbb3 integrin, (the platelet receptor), always compared to the linear reference peptide GRGDSPK. Hence, superactivity and selectivity were obtained in this procedure [12,116,135]. A detailed study of the conformation showed that there is a well-defined bII0 -turn around D-Phe-Val and a lowerpopulated g-turn around Gly, which is in fast equilibrium with a nonhydrogen-bonded amide bond between Arg and Gly. This peptide bond can flip between two different orientations [95,108]. In return, some of the hexapeptides showed selectivity for the aIIbb3 receptor. It turned out that ligands active for aVb3 have a stronger kink at the central glycine, resulting in a shorter distance between the arginine and aspartic functional groups than in the aIIbb3-specific peptides [136]. To explore the potential participation of the peptide bonds in the binding to the receptor, all 16 stereoisomers of the aVb3-selective

EXAMPLES OF CYCLIC PEPTIDES AS DRUG CANDIDATES

161

parent peptide cyclo(-RGDfV-) and their retro sequences were prepared and investigated [137]. One of the retroinverso analogue cyclo(-VfdGr-) exhibited significantly reduced activity, although the NMR–MD-based structure showed an almost identical side-chain orientation, whereas the direction of the peptide bond was reversed. This finding clearly gives evidence that at least one peptide bond participates in the binding to the receptor. Other chemical modifications of the peptide bond such as thioamides or reduced amide bonds often do not give satisfying conclusions, as these modifications usually also result in conformational changes. Hence, the observed change in biological activity cannot be unequivocally interpreted by the substitution alone. To further rigidify the conformation, different b-turn mimetics were incorporated at the b-turn of the peptide, replacing D-Phe and Val. Different classes of turn mimetics were chosen, including dipeptides, peptide analogues and spiro compounds, as well as structures with planar aromatic or carbohydrate cores. It turned out that several of the b-turn mimetics do not substitute a b-turn but are only allowed in the conformational space of b-turns [139–144]. It is beyond the scope of this chapter to give details about the different conformations and their activities for the integrin receptor subtypes [136,140,141]. Many different modifications have been investigated, but the result of an N-methyl scan yielded cyclo(-RGDfNMeVal-), which was chosen by the Merck company as a drug candidate (Cilengitide) [138]. This peptide has entered clinical phase III for the treatment of glioblastoma multiforme, metastatic prostate cancer and lymphoma (Figure 4.14).

Figure 4.14 Conformation of Cilengitide after a MD simulation [138] (see colour Plate 3)

162

DESIGN OF CYCLIC PEPTIDES

Another cyclic peptide specific for aIIbb3 is already successfully sold by Schering-Plough under the name Integrilin as prevention for myocardial infarction (turnover: 325 million USD in 2004). The heptapeptide cyclo-S,S-[Mpa-Har-Gly-Asp-Trp-Pro-Cys]-NH2 behind the marketing name is called eptifibatide and contains two-non-natural building blocks 3mercaptopropanoic acid (Mpa) and homoarginine (Har). (Figure 4.15).

Figure 4.15 Eptifibatide, an aIIbb3 antagonist marketed under the name Integrilin, and two peptides with more constrained disulfide bonds and better affinities. All IC50 values are given for the inhibition of binding of integrin aIIbb3 to fibronectin

Eptifibatide is an analogue of the highly specific aIIbb3 disintegrin barbourin, which originally contained a lysine instead of the arginine in the RGD motif. Here, cyclic disulfide-bridged templates were successfully chosen to optimize affinity and bioavailability [145–148]. Along with other modifications like N-methylation and reduction of the ring size, the flexible disulfide bridge was rigidified by incorporation of penicillamine (Pen) instead of cysteine. This led to the peptide Ac-cyclo-S,S-[Cys-(NMe)Arg-Gly-Asp-Pen]-NH2 and an increase in affinity over 100-fold better than eptifibatide. Subsequent derivatization in the direction of semipeptides led to the replacement of the disulfide tether with 2-mercaptobenzoyl/2-mercaptoaniline (Mba/Man), constraining the w torsion angle to 0°, which led to significant enhancement of affinity and potency [149].

EXAMPLES OF CYCLIC PEPTIDES AS DRUG CANDIDATES

4.5.2

163

CXCR4 Antagonists

Chemokine receptors belong to the family of GPCRs and allow cells, viruses and bacteria to recognize certain chemoattractants or -repellents and induce migration to or from their source. As the name suggests, their corresponding ligands belong to the subfamily of cytokines (chemotactic cytokines). In the human body this signalling system is initially involved in stem-cell migration during embryogenesis and later used by the immune system to guide cells [150,151]. However, this normally vital and benign system has been hijacked by various diseases like HIV, rheumatoid arthritis (RA), and cancer metastasis. As peptidic antagonists of CXCR4 have already shown anti-HIV [152], anti-RA [153,154] and anti-metastatic [155,156] activity, they are a valuable target for medicinal chemistry. In contrast to the RGD motif, size reduction did not originate from the natural ligand CXCL12 itself but from biologically active peptides found by screening of natural extracts. In the late 1980s, four peptides with 17–18 amino acids and constrained by two cysteine bridges were found in horseshoe crabs [157,158]. One of these peptides was optimized after additional biological studies showed antiviral activity [159, 160]. Due to these effects the structure was investigated in this early stage [161]. By converting two existing phenylalanines into tyrosines and introducing an additional positive charge by substitution of a valine with lysine, the affinity was improved by three orders of magnitude (Figure 4.16) [162]. Subsequent modifications included the shortening of the peptide to a monocyclic strand with 14 amino acids by replacing the central cysteine bridge and the attached loop region with a bII0 -inducing dipeptide unit consisting of D-Lys and L-Pro [163]. After isosteric replacement of Trp3 with L-3-(2-naphthyl)alanine (Nal) [164], substitution of Lys12 with citrulline (Cit) [165] and removal of the C-terminal amide, a better affinity and a reduced cytotoxicity to the precursor peptides were observed. A consecutive Ala-scan [166] showed that four potentially important amino acids were in close proximity to one another because of the remaining cysteine bridge, and partly out of the macrocycle, and therefore more flexible. These were, together with an additional glycine, subjected to a scan of cyclic pentapeptides with two orthogonal libraries consisting of a sequence- and a conformation-based approach following the spatial screening method, resulting in a small lead compound with equal affinity in the low nanomolar range and high biostability [167]. Further modifications have been introduced, including retro enantiomer libraries, amino acid substitutions and attempts to create

164

DESIGN OF CYCLIC PEPTIDES

Figure 4.16 Left: derivatization of polyphemusin II into a cyclic pentapeptide with excellent antagonistic activity on CXCR4. Optimization of the residues marked in red led from the initial peptide to its shortened analogue (middle). The amino acids marked in blue and an additional glycine were subjected to the spatial screening. Right: cartoon of the backbone structure with disulfide bonds of the polyphemusin II analogue tachyplesin I, showing the regions of interest in colours corresponding to the scheme on the left (see colour Plate 4)

tetrapeptidic scaffolds. However, the most promising has been N-methylation of the arginine (D-configured, in contrast to the starting compound) next to the D-tyrosine, with further enhancement of affinity [168].

4.5.3

Sandostatin and the Veber–Hirschmann Peptide as Examples of Rational Design

Somatostatins are a family of cyclopeptides that exhibit a broad inhibitory effect on the secretion of hormones like insulin, glucagon and growth hormones. Therefore, their derivatives are used in the treatment of acromegaly and endocrine tumours [169]. A remarkable example of the (semi)rational design of a drug candidate is the chemical modification of the metabolically unstable peptide

EXAMPLES OF CYCLIC PEPTIDES AS DRUG CANDIDATES

165

hormone somatostatin (t½ ¼ 1–2 minutes). In vivo, the resulting cyclic derivative Sandostatin [31] (see Figure 4.17) is much more stable (t½ ¼ 70–110 minutes). The enhanced stability is achieved by shifting the disulfide bridge closer to the active part of the sequence, by introducing two D-amino acids (the D-Trp8 in the b-turn, forming the recognition region, and the N-terminal D-Phe) and by reduction of the C-terminal carboxyl group of Thr into the corresponding alcohol [169,170]. The cyclic Veber–Hirschmann hexapeptide cyclo(-PFwKTF-) is a highly active somatostatin agonist [171]. Compared to Sandostatin, it is an even smaller somatostatin derivative. However, as the hexapeptide is cyclized from head-to-tail, the two cysteine residues of Sandostatin are replaced by a dipeptide unit forming a second b-turn. This consists of a phenylalanine also present in somatostatin and a proline which favours b-turns when placed in the i þ 1 position, as is the case here.

Figure 4.17 The natural peptide hormone somatostatin (Somatotropin Releasing Factor, SRIF-14; upper side), the modified octreotide (Sandostatin; right) and the reduced derivative, called Veber–Hirschmann peptide left [171]. The atoms marked in dark grey are modified in comparison to somatostatin

166

4.6

DESIGN OF CYCLIC PEPTIDES

CONCLUSION

Cyclization of linear peptide strands improves their pharmacological properties and makes them more druglike by reducing the accessible conformational space. In particular, biostability is enhanced, as enzymes that cleave linear peptides within minutes do have significantly less points of vantage for digestion. Additionally, cyclic peptides exhibit better biological activity if their structures are fixed closer to the receptorbound conformation. The reduction in conformational flexibility also causes receptor-subtype specificity and helps to suppress unwanted side effects caused by promiscuous binding of linear peptides to whole receptor families. Apart from improving pharmacological properties, cyclization enables rational drug design of peptides. Certain cyclic peptidic backbones have a strongly preferred or even homogeneous 3D arrangement and can therefore be used as scaffolds to present amino acid side chains with their pharmacophoric groups in a defined structure. The spatial screening approach makes use of this rigidity to find the optimal presentation of side chains by retaining the active peptide sequence but varying its conformation. The success of this simple procedure results from the fact that recognition motifs in protein–protein (receptor–ligand) interactions are often found in exposed loop regions, which are efficiently mimicked by cyclic penta- and hexapeptides. Additionally, knowledge of the 3D structure of the active sequence enables (semi)rational design of peptidomimetics. Alternatively, the cyclic peptides obtained by spatial screening can be further optimized to match the required pharmacological properties. Improvement of pharmacophoric properties, in combination with rational design approaches, makes cyclic peptides interesting targets in medicinal research; moreover, they have the potential to become successful drugs like Cyclosporin A, Sandostatin or Integrilin.

REFERENCES [1] H.-J. Bo¨hm, In: H.-J. Bo¨hm and G. Schneider, Protein–Ligand Interactions, 2005, 3–20. [2] P. A. Bartlett and M. Entzeroth, Exploiting Chemical Diversity in Drug Discovery, RSC Publishing, Cambridge, 2006. [3] R. Flaumenhaft, In: J. B. Taylor and D. J. Triggle, Chemical Biology in Comprehensive Medicinal Chemistry, Elsevier, 2007, 129–149.

REFERENCES

167

[4] H. Kessler, Peptide Conformations. 19. Conformation and biological-activity of cyclic-peptides, Angewandte Chemie – International Edition, 21, 512–523 (1982). [5] V. J. Hruby, Conformational restrictions of biologically-active peptides via aminoacid side-chain groups, Life Sciences, 31, 189–199 (1982). [6] A. Meyer, J. Auernheimer, A. Modlinger and H. Kessler, Targeting RGO recognizing integrins: drug development, biomaterial research, tumor imaging and targeting, Current Pharmaceutical Design, 12, 2723–2747. [7] H. Weisshoff, T. Wieprecht, P. Henklein, C. Antz and C. Mugge, Cyclic cholecystokinin-analog pentapeptide cyclo (Asp-Trp-Met-Asp-Phe) – an unexpected solution conformation, Biochemical and Biophysical Research Communications, 213, 506– 512 (1995). [8] G. V. Nikiforovich, K. E. Kover, W. J. Zhang and G. R. Marshall, Cyclopentapeptides as flexible conformational templates, Journal of the American Chemical Society, 122, 3262–3273 (2000). [9] R. Haubner et al., Structural and functional aspects of RGD-containing cyclic pentapeptides as highly potent and selective integrin alpha(v)beta(3) antagonists, Journal of the American Chemical Society, 118, 7461–7472 (1996). [10] A. M. Rodriguez et al., Characteristics of Ramachandran maps of L-alanine diamides as computed by various molecular mechanics, semiempirical and ab initio MO methods. A search for primary standard of peptide conformational stability, Journal of Molecular Structure: THEOCHEM, 455, 275–301 (1998). [11] R. Schwyzer and U. Ludescher, Homodetic cyclic polypeptides. XIV. Conformation of the cyclic hexapeptides cyclo-glycyl-L-prolylglycylglycyl-L-prolylglycyl by proton magnetic resonance studies and parallel to the cyclodecapeptide gramicidin S, Helvetica Chimica Acta 52, 2033–2040 (1969). [12] M. Aumailley et al., Arg-Gly-Asp constrained within cyclic pentapeptides – strong and selective inhibitors of cell-adhesion to vitronectin and laminin fragment-P1, FEBS Letters, 291, 50–54 (1991). [13] J. Rizo and L. M. Gierasch, Constrained peptides: Models of bioactive peptides and protein substructures, Annual Review of Biochemistry, 61, 387–418 (1992). [14] C. Ramakrishnan, P. K. C. Paul and K. Ramnarayan, Cyclic-peptides – Small and big and their conformational aspects, Journal of Biosciences, 8, 239–251 (1985). [15] V. J. Hruby, F. Alobeidi and W. Kazmierski, Emerging approaches in the molecular design of receptor-selective peptide ligands – Conformational, topographical and dynamic considerations, Biochemical Journal, 268, 249–262 (1990). [16] G. F. Gause, M. G. Brazhnikova and N. P. Lissovskaya, Preparation and properties of crystalline gramicidine C, Comptes Rendus De L Academie Des Sciences De L Urss, 43, 217–219 (1944). [17] G. F. Gause and M. G. Brazhnikova, Gramicidin sand – Its use in the treatment of infected wounds, Nature, 154, 703 (1944). [18] G. F. Gause and M. G. Brazhnikova, Gramicin S – Origin and mode of action, Lancet, 2, 715–716 (1944). [19] J. N. Porter, R. Broschard, G. Krupka, P. Little and J. S. Zellat, Isolation and production of polymyxin, Annals of the New York Academy of Sciences, 51, 857–865 (1949). [20] R. G. Shepherd et al., Chemical studies on polymyxin. 1. Isolation and preliminary purification, Journal of the American Chemical Society, 70, 3771–3774 (1948).

168

DESIGN OF CYCLIC PEPTIDES

[21] P. G. Stansly and M. E. Schlosser, Studies on polymyxin – isolation and identification of bacillus-polymyxa and differentiation of polymyxin from certain known antibiotics, Journal of Bacteriology, 54, 549–556 (1947). [22] A. Ruegger et al., Cyclosporin-a, a peptide metabolite from trichoderma-polysporum (link ex pers) rifai, with a remarkable immunosuppressive activity, Helvetica Chimica Acta, 59, 1075–1092 (1976). [23] S. W. Fesik et al., NMR-studies of [U-13C] cyclosporin A bound to cyclophilin – Bound conformation and portions of cyclosporine in binding, Biochemistry, 30, 6574–6583 (1991). [24] J. Klages, C. Neubauer, M. Coles, H. Kessler and B. Luy, Structure refinement of cyclosporin a in chloroform by using RDCs measured in a stretched PDMS-gel, Chembiochem, 6, 1672–1678 (2005). [25] H. R. Loosli et al., Peptide conformations. Part 31. The conformation of cyclosporine A in the crystal and in solution, Helvetica Chimica Acta, 68, 682–704 (1985). [26] A. Sehgal, Peptides 2006 – New Applications in Discovery, Manufacturing, and Therapeutics, D&MD Publications, Westborough, 2006. [27] E. Emanuele, M. Arra and S. Pesenti, Vasopressin and oxytocin as neurohormonal mediators of MDMA (ecstasy) sociosexual behavioural effects, Medical Hypotheses, 67, 1250–1251 (2006). [28] J. F. Mohr and B. E. Murray, Point: Vancomycin is not obsolete for the treatment of infection caused by methicillin-resistant Staphylococcus aureus, Clinical Infectious Diseases, 44, 1536–1542 (2007). [29] S. Bellary and A. H. Barnett, Inhaled insulin: new technology, new possibilities, International Journal of Clinical Practice, 60, 728–734 (2006). [30] M. P. Curran and G. M. Keating, Eptifibatide: A review of its use in patients with acute coronary syndromes and/or undergoing percutaneous coronary intervention, Drugs, 65, 2009–2035 (2005). [31] W. Bauer et al., Sms 201–995 – A very potent and selective octapeptide analog of somatostatin with prolonged action, Life Sciences, 31, 1133–1140 (1982). [32] Y. S. Lee et al., Development of novel enkephalin analogues that have enhanced opioid activities at both mu and delta opioid receptors, Journal of Medicinal Chemistry, 50, 5528–5532 (2007). [33] P. W. Schiller, In: C. G. Wermuth, N. Koga, H. Ko¨nig and B. W. Metcalf, Medicinal Chemistry for the 21st Century, Blackwell Scientific Publications, London, 1992, 215–232. [34] I. Berezowska, N. N. Chung, C. Lemieux, B. C. Wilkes and P. W. Schiller, Dicarba analogues of the cyclic enkephalin peptides H-Tyr-c[D-Cys-Gly-Phe-D(or L)Cys]NH2 retain high opioid activity, Journal of Medicinal Chemistry, 50, 1414–1417 (2007). [35] M. E. Hadley and C. Haskell-Luevano, The proopiomelanocortin system, Cutaneous Neuroimmunomodulation: The Proopiomelanocortin System, 885, 1–21 (1999). [36] D. J. Craik, S. Simonsen and N. L. Daly, The cyclotides: novel macrocyclic peptides as scaffolds in drug design, Current Opinion in Drug Discovery & Development, 5, 251–260 (2002). [37] D. J. Craik, Chemistry – Seamless proteins tie up their loose ends, Science, 311, 1563–1564 (2006).

REFERENCES

169

[38] R. H.Kimura, A. T. Tran and J. A. Camarero, Biosynthesis of the cyclotide Kalata B1 by using protein splicing, Angewandte Chemie – International Edition, 45, 973–976 (2006). [39] I. Rubio-Aliaga and H. Daniel, Mammalian peptide transporters as targets for drug delivery, Trends in Pharmacological Sciences, 23, 434–440 (2002). [40] N. J. P. Dubree et al., Selective alpha 4 beta 7 integrin antagonists and their potential as antiinflammatory agents, Journal of Medicinal Chemistry, 45, 3451–3457 (2002). [41] R. H. Li, R. H. Hoess, J. S. Bennett and W. F. DeGrado, Use of phage display to probe the evolution of binding specificity and affinity in integrins, Protein Engineering, 16, 65–72 (2003). [42] C. Gilon, C. Mang, E. Lohof, A. Friedler and H. Kessler, In: M. Goodman, A. Felix, L. Moroder and C. Toniolo, Houben-Weyl: Methods of Organic Chemistry, Thieme Verlag, Stuttgart, 2003, 461–542. [43] C. Gilon, D. Halle, M. Chorev, Z. Selinger and G. Byk, Backbone cyclization – A new method for conferring conformational constraint on peptides, Biopolymers, 31, 745– 750 (1991). [44] R. C. D. Milton, S. C. F. Milton and S. B. H. Kent, Total chemical synthesis of a Denzyme – The enantiomers of HIV-1 protease show demonstration of reciprocal chiral substrate-specificity, Science, 256, 1445–1448 (1992). [45] M. Chorev and M. Goodman, Recent developments in retro peptides and proteins – An ongoing topochemical exploration, Trends in Biotechnology, 13, 438–445 (1995). [46] S. Lande, Retrobradykinin – peptide with reverse sequence of bradykinin, Journal of Organic Chemistry, 27, 4558–4562 (1962). [47] J. Rudinger, In: E.J. Arie¨ns, Drug Design, Academic Press, 1971, 319–649. [48] H. Gerlach, J. A. Owtschinnikow and V. Prelog, Cycloenantiomerie þ cyclodiastereomerie. 2. Mitteilung uber cycloenantiomere cyclo-hexaalanyle þ ein cycloenantiomeres cyclo-diglycyl-tetraalanyl, Helvetica Chimica Acta, 47, 2294–2301 (1964). [49] H. Gerlach, G. Haas and V. Prelog, Uber zwei cycloisomere triglycyl-Tri-L-alanyle, Helvetica Chimica Acta, 49, 603–607 (1966). [50] V. Prelog and H. Gerlach, Cycloenantiomerie and cyclodiastereomerie. 1. Mitteilung, Helvetica Chimica Acta, 47, 2288–2294 (1964). [51] M. M. Shemyakin, Y. A. Ovchinnikov and V. T. Ivanov, Topochemical investigations on peptide systems, Angewandte Chemie – International Edition, 8, 492–499 (1969). [52] H. Kessler et al., Alanine containing analogs of cyclo(-D-Pro-Phe-Thr-Lys(Z)-TrpPhe-) – Conformationally controlled structure–activity-relationships, Tetrahedron, 44, 745–759 (1988). [53] D. Heckmann et al., Probing integrin selectivity: Rational design of highly active and selective ligands for the alpha 5 beta 1 and alpha v beta 3 integrin receptor, Angewandte Chemie – International Edition, 46, 3571–3574 (2007). [54] D. Ron et al., N-methylated analogs of Ac[Nle28,31]Cck(26–33) – Synthesis, activity, and receptor selectivity, Journal of Medicinal Chemistry, 35, 2806–2811 (1992). [55] C. Gilon, M. A. Dechantsreiter, F. Burkhart, A. Friedler and H. Kessler, In: M. Goodmann, A. Felix, L. Moroder and C. Toniolo, Houben-Weyl: Methods of Organic Chemistry, Thieme Verlag, Stuttgart, New York, 2003, 215–271.

170

DESIGN OF CYCLIC PEPTIDES

[56] J. Chatterjee et al., Multiple N-methylation by a designed approach enhances receptor selectivity, Journal of Medicinal Chemistry, 50, 5878–5881 (2007). [57] E. Biron, J. Chatterjee and H. Kessler, Optimized selective N-methylation of peptides on solid support, Journal of Peptide Science, 12, 213–219 (2006). [58] E. Biron et al., Improved oral bioavailability of peptides via N-methylation: Somatostatin analogs, Angewandte Chemie – International Edition, 47, 2595– 2599 (2008). [59] J. Gante, H. Kessler, and C. Gibson, In: M. Goodmann, A. Felix, L. Moroder and C. Toniolo, Houben-Weyl: Methods of Organic Chemistry, Thieme Verlag, Stuttgart, 2003, 311–323. [60] H. Kessler, Peptoids – A new approach to the development of pharmaceuticals, Angewandte Chemie – International Edition, 32, 543–544 (1993). [61] H. Weisshoff et al., Novel disulfide-constrained pentapeptides as models for beta-via turns in proteins, FEBS Letters, 372, 203–209 (1995). [62] A. Ravi and P. Balaram, Cyclic peptide disulfides – consecutive beta-turn conformation of a synthetic model peptide corresponding to the active-site of thioredoxin, Biochimica Et Biophysica Acta, 745, 301–309 (1983). [63] M. Hata and G. R. Marshall, Do benzodiazepines mimic reverse-turn structures?, Journal of Computer-aided Molecular Design, 20, 321–331 (2006). [64] O. Demmer, I. Dijkgraaf, M. Schottelius, H. J. Wester and H. Kessler, Introduction of functional groups into peptides via N-alkylation, Organic Letters, 10, 2015–2018 (2008). [65] J. S. Davis, In: J. Jones, A Specialist Periodical Report: Amino Acids, Peptides and Proteins, The Royal Society of Chemistry, Cambridge, 1968–1998. [66] D. J. Faulkner, Marine natural-products, Natural Product Reports, 9, 323–364 (1992). [67] Y. A. Ovchinnikov and V. T. Ivanov, In: H. Neurth and R. L. Hill, The Cyclic Peptides: Structure, Conformation and Function, Academic Press, New York, 1982, 307. [68] J. W. Payne, N. J. Marshall, B. M. Grail and S. Gupta, Conformer profiles and biological activities of peptides, Current Organic Chemistry, 6, 1221–1246 (2002). [69] T. Arndt, U. Arndt, U. Reuning and H. Kessler, In: G. F. Weber, Cancer Therapy: Molecular Targets in Tumor–Host Interactions, Horizon Bioscience, Norfolk, 2005, 93–141. [70] V. J. Hruby, Designing peptide receptor agonists and antagonists, Nature Reviews Drug Discovery, 1, 847–858 (2002). [71] L. H. M. Janssen, Conformational flexibility and receptor interaction, Bioorganic & Medicinal Chemistry, 6, 785–788 (1998). [72] K. D. Kopple et al., Conformational mobility in cyclic oligopeptides, Biopolymers, 33, 1093–1099 (1993). [73] H. Kessler, Detection of intramolecular mobility by NMR spectroscopy. 13. Detection of hindered rotation and inversion by NMR spectroscopy, Angewandte Chemie – International Edition, 9, 219–235 (1970). [74] R. Haubner, D. Finsinger and H. Kessler, Stereoisomeric peptide libraries and peptidomimetics for designing selective inhibitors of the alpha(v)beta(3) integrin for a new cancer therapy, Angewandte Chemie – International Edition, 36, 1375–1389 (1997). [75] R. W. Hoffmann, Flexible molecules with defined shape – Conformational design, Angewandte Chemie – International Edition, 31, 1124–1134 (1992).

REFERENCES

171

[76] G. Fischer, Chemical aspects of peptide bond isomerisation, Chemical Society Reviews, 29, 119–127 (2000). [77] L. M. Gierasch, C. M. Deber, V. Madison, C. H. Niu and E. R. Blout, Cyclicpeptides – Conformations of (X-L-Pro-Y)2 cyclic hexapeptides – Preferred betaturn conformers and implications for beta-turns in proteins, Biochemistry, 20, 4730–4738 (1981). [78] H. Kessler, U. Anders and M. Schudok, An unexpected cis peptide-bond in the minor conformation of a cyclic hexapeptide containing only secondary amide bonds, Journal of the American Chemical Society, 112, 5908–5916 (1990). [79] J. L. Kofron et al., Lithium–chloride perturbation of cis-trans peptide-bond equilibria – Effect on conformational equilibria in cyclosporine A and on timedependent inhibition of cyclophilin, Journal of the American Chemical Society, 114, 2670–2675 (1992). [80] G. Scherer, M. L. Kramer, M. Schutkowski, U. Reimer and G. Fischer, Barriers to rotation of secondary amide peptide bonds, Journal of the American Chemical Society, 120, 5568–5574 (1998). [81] C. Schiene-Fischer and G. Fischer, Direct measurement indicates a slow cis/trans isomerization at the secondary amide peptide bond of glycylglycine, Journal of the American Chemical Society, 123, 6227–6231 (2001). [82] H. Kessler, In: R. H. Sarma, Stereodynamics of Molecular Systems, Pergamon Press, Oxford, New York, Frankfurt, Paris, 1979, 187–196. [83] H. Kessler, P. Kramer and G. Krack, Peptide conformation. 9. Conformational studies on cyclotripeptides containing N-benzyl-glycine, Israel Journal of Chemistry, 20, 188–195 (1980). [84] G. N. Ramachandran, C. Ramakrishnan and V. Sasisekharan, Stereochemistry of polypeptide chain configurations, Journal of Molecular Biology, 7, 95–99 (1963). [85] M. Chorev, Probing the bimolecular interface between RGD-containing ligands and the alpha(v)beta(3) integrin receptor – An important therapeutic target. Part 2, Chimica Oggi, 20, 81–85 (2002). [86] V. Saudek, R. A. Atkinson and J. T. Pelton, 3-dimensional structure of echistatin, the smallest active RGD protein, Biochemistry, 30, 7369–7372 (1991). [87] G. D. Rose, L. M. Gierasch and J. A. Smith, Turns in peptides and proteins, Advances in Protein Chemistry, 37, 1–109 (1985). [88] H. Balaram, K. Uma and P. Balaram, Synthetic peptide models for protein secondary structures. b-sheet formation in acyclic cystine peptides, International Journal of Peptide and Protein Research, 35, 495–500 (1990). [89] C. Bisang et al., Synthesis, conformational properties, and immunogenicity of a cyclic template-bound peptide mimetic containing an NPNA motif from the circumsporozoite protein of plasmodium falciparum, Journal of the American Chemical Society, 120, 7439–7449 (1998). [90] B. Diblasio, A. Lombardi, X. Yang, C. Pedone and V. Pavone, Beta-alanyl-betaalanine in cyclic beta-turned peptides, Biopolymers, 31, 1181–1188 (1991). [91] R. M. Freidinger, Design and synthesis of novel bioactive peptides and peptidomimeties, Journal of Medicinal Chemistry, 46, 5553–5566 (2003). [92] A. Giannis, Peptidomimetics for receptor ligands discovery, development, and medical perspectives, Angewandte Chemie – International Edition, 32, 1244–1267 (1993).

172

DESIGN OF CYCLIC PEPTIDES

[93] E. G. Hutchinson and J. M. Thornton, A revised set of potentials for beta-turn formation in proteins, Protein Science: A Publication of the Protein Society, 3, 2207–2216 (1994). [94] O. Kitagawa et al., Structural-analysis of beta-turn mimics containing a substituted 6-aminocaproic acid linker, Journal of the American Chemical Society, 117, 5169– 5178 (1995). [95] D. F. Mierke, M. Kurz and H. Kessler, Peptide flexibility and calculations of an ensemble of molecules, Journal of the American Chemical Society, 116, 1042–1049 (1994). [96] R. Schwyzer, A. Tunkyi, J. P. Carrion, B. Gorup and H. Nolting, Verdoppelungserscheinungen beim Ringschluss von Peptiden. V. Relative Bedeutung der sterischen Hinclerung and der Assoziation } uber Wasserstoff-Bra¨cken bei Tripeptiden. Spektroskopische Versuche zur Konformationsbestimmung. 12. Mitteilung } uber homodet cyclische Polypeptide Helvetica Chimica Acta, 47, 441–464 (1964) [97] H. Weisshoff et al., Mimicry of beta II 0 -turns of proteins in cyclic pentapeptides with one and without D-amino acids, European Journal of Biochemistry, 259, 776– 788 (1999). [98] J. Chatterjee, D. Mierke and H. Kessler, N-methylated cyclic pentaalanine peptides as template structures, Journal of the American Chemical Society, 128, 15164– 15172 (2006). [99] J. A. Smith and L. G. Pease, Reverse turns in peptides and proteins, CRC Critical Reviews in Biochemistry, 8, 315–399 (1980). [100] C. A. Bewley, H. He, D. H. Williams and D. J. Faulkner, Aciculitins A–C: cytotoxic and antifungal cyclic peptides from the lithistid sponge Aciculites orientalis, Journal of the American Chemical Society, 118, 4314–4321 (1996). [101] K. D. Kopple, On the location of beta-turns, Biopolymers, 20, 1913–1920 (1981). [102] C. Weber et al., The NMR structure of cyclosporine A bound to cyclophilin in aqueous-solution, Biochemistry, 30, 6563–6574 (1991). [103] D. Seebach, A. K. Beck and D. J. Bierbaum, The world of beta- and gamma-peptides comprised of homologated proteinogenic amino acids and other components, Chemistry & Biodiversity, 1, 1111–1239 (2004). [104] J. P. Snyder, Probable unimportance of intramolecular hydrogen-bonds for determining the secondary structure of cyclic hexapeptides – Roseotoxin-B, Journal of the American Chemical Society, 106, 2393–2400 (1984). [105] H. Kessler, C. Griesinger and K. Wagner, Peptide conformations. 42. Conformation of side-chains in peptides using heteronuclear coupling-constants obtained by twodimensional NMR-spectroscopy, Journal of the American Chemical Society, 109, 6927–6933 (1987). [106] H. Kessler et al., Does a molecule have the same conformation in the crystalline state and in solution – Comparison of NMR results for the solid-state and solution with those of the X-ray structural determination, Angewandte Chemie – International Edition, 20, 1053–1055 (1981). [107] B. Luy, A. O. Frank and H. Kessler, In: R. Mannhold, Drug Properties: Measurement and Computation, 2008, 207–254. [108] M. Heller et al., The conformation of cyclo(-D-Pro-Ala(4)-) as a model for cyclic pentapeptides of the DL4 type, Journal of the American Chemical Society, 128, 13806–13814 (2006). [109] W. Liu et al., Structural role of glycine in amyloid fibrils formed from transmembrane alpha-helices, Biochemistry, 44, 3591–3597 (2005).

REFERENCES

173

[110] A. Senes, D. E. Engel and W. F. DeGrado, Folding of helical membrane proteins: the role of polar, GxxxG-like and proline motifs, Current Opinion in Structural Biology, 14, 465–479 (2004). [111] H. Matter and H. Kessler, Structures, dynamics, and biological-activities of 15 cyclic hexapeptide analogs of the alpha–amylase inhibitor tendamistat (HOE467) in solution, Journal of the American Chemical Society, 117, 3347–3359 (1995). [112] Y. B. He, Z. W. Huang, K. Raynor, T. Reisine and M. Goodman, Syntheses and conformations of somatostatin-related cyclic hexapeptides incorporating specific alpha-methylated and beta-methylated residues, Journal of the American Chemical Society, 115, 8066–8072 (1993). [113] J. D. Dunitz and J. Waser, Geometric constraints in 6-membered and 8-membered rings, Journal of the American Chemical Society, 94, 5645–5650 (1972). [114] Y. A. Bara et al., Synthesis of cyclic pentapeptides containing L-phenylalanine and glycine, Chemische Berichte-Recueil, 111, 1029–1044 (1978). [115] J. Dale and K. Titlestad, Conformational processes in simple cyclic peptides, Acta Chemica Scandinavica Series B – Organic Chemistry and Biochemistry, B 29, 353–361 (1975). [116] M. Gurrath, G. Mu¨ller, H. Kessler, M. Aumailley and R. Timpl, Conformation activity studies of rationally designed potent antiadhesive RGD peptides, European Journal of Biochemistry, 210, 911–921 (1992). [117] H. Kessler and B. Kutscher, Synthesis and NMR investigation of cyclic pentapeptide analogs of thymopentin, Tetrahedron Letters, 26, 177–180 (1985). [118] H. Kessler et al., Design, synthesis, and conformation of superacitve thymopoietinanalogs, Pept Struct Funct Proc Am Pept Symp 9th, 83–92 (1985). [119] H. Kessler, B. Kutscher and A. Klein, Peptide conformations. 39. NMR-studies of the conformation of cyclic pentapeptide analogs of thymopoietin, Liebigs Annalen der Chemie, 893–913 (1986). [120] H. Kessler and B. Kutscher, Peptide conformations. 40. Cyclic hexapeptide analogs of thymopoietin – Synthesis and conformational-analysis, Liebigs Annalen der Chemie, 914–931 (1986). [121] H. Kessler, R. Gratias, G. Hessler, M. Gurrath and G. Muller, Conformation of cyclic peptides. Principle concepts and the design of selectivity and superactivity in bioactive sequences by ‘spatial screening’, Pure and Applied Chemistry, 68, 1201– 1205 (1996). [122] V. Duvigneaud, G. S. Denning, W. Y. Chan and S. Drabarek, Synthesis and pharmacological study of 4-decarboxamido-oxytocin (4-alpha-aminobutyric acidoxytocin) and 5-decarboxamido-oxytocin (5-alanine-oxytocin), Journal of Biological Chemistry, 239, 472–478 (1964). [123] S. Rothemund et al., Structure effects of double D-amino-acid replacements – A nuclear-magnetic-resonance and circular-dichroism study using amphipathic model helices, Biochemistry, 34, 12954–12962 (1995). [124] K. Palczewski et al., Crystal structure of rhodopsin: A G protein-coupled receptor, Science, 289, 739–745 (2000). [125] S. G. F. Rasmussen et al., Crystal structure of the human beta(2) adrenergic G-protein-coupled receptor, Nature, 450, 383–387 (2007). [126] T. Warne et al., structure of a b (1)-adrenergic G-protein-coupled receptor, Nature, 454, 486–491 (2008).

174

DESIGN OF CYCLIC PEPTIDES

[127] V. Jaakola et al., The 2.6 A° Crystal Structure of a Human A2A Adenosine Receptor Bound to an Antagonist, science, 322, 1211–1217 (2008). [128] J. A. Elbe and K. Ku¨hn, Integrin–Ligand Interaction, Springer, Heidelberg, 1997. [129] K. J. Clemetson and J. M. Clemetson, Integrins and cardiovascular disease, Cellular and Molecular Life Sciences, 54, 502–513 (1998). [130] H. Jin and J. Varner, Integrins: roles in cancer development and as treatment targets, British Journal of Cancer, 90, 561–565 (2004). [131] A. J. Rojas and A. R. Ahmed, Adhesion receptors in health and disease, Critical Reviews in Oral Biology and Medicine, 10, 337–358 (1999). [132] H. Yusuf-Makagiansar, M. E. Anderson, T. V. Yakovleva, J. S. Murray and T. J. Siahaan, Inhibition of LFA-1/ICAM-1 and VLA-4/VCAM-1 as a therapeutic approach to inflammation and autoimmune disease, Medicinal Research Reviews, 22, 146–167 (2002). [133] M. D. Pierschbacher and E. Ruoslahti, Cell attachment activity of fibronectin can be duplicated by small synthetic fragments of the molecule, Nature, 309, 30–33 (1984). [134] L. Tranqui et al., Differential structural requirements for fibrinogen binding to platelets and to endothelial cells, The Journal of Cell Biology, 108, 2519–2527 (1989). [135] M. Pfaff, et al., Selective recognition of cyclic RGD peptides of NMR defined conformation by alpha IIb beta 3, alpha V beta 3, and alpha 5 beta 1 integrins, The Journal of Biological Chemistry, 269, 20233–20238 (1994). [136] G. Mu¨ller, M. Gurrath and H. Kessler, Pharmacophore refinement of gpIIb/IIIa antagonists based on comparative studies of antiadhesive cyclic and acyclic RGD peptides, Journal of Computer-Aided Molecular Design, 8, 709–730 (1994). [137] J. Wermuth, S. L. Goodman, A. Jonczyk and H. Kessler, Stereoisomerism and biological activity of the selective and superactive alpha(v)beta(3) integrin inhibitor cyclo(-RGDfV-) and its retro-inverso peptide, Journal of the American Chemical Society, 119, 1328–1335 (1997). [138] M. A. Dechantsreiter et al., N-methylated cyclic RGD peptides as highly active and selective alpha(v)beta(3) integrin antagonists, Journal of Medicinal Chemistry, 42, 3033–3040 (1999). [139] D. Finsinger, In: Institut fu¨r Organische Chemie und Biochemie, Technische Universita¨t Mu¨nchen, Mu¨nchen, 1997, 161. [140] F. Schumann, A. Mu¨ller, M. Koksch, G. Mu¨ller and N. Sewald, Are beta-amino acids gamma-turn mimetics? Exploring a new design principle for bioactive cyclopeptides, Journal of the American Chemical Society, 122, 12009–12010 (2000). [141] D. Zimmermann, et al., Integrin alpha(5)beta(1) ligands: Biological evaluation and conformational analysis, ChemBioChem, 6, 272–276 (2005). [142] S. A. W. Gruner, E. Locardi, E. Lohof and H. Kessler, Carbohydrate-based mimetics in drug design: sugar amino acids and carbohydrate scaffolds, Chemical Reviews, 102, 491–514 (2002). [143] E. von Roedern and H. Kessler, A sugar amino-acid as a novel peptidomimetic, Angewandte Chemie – International Edition, 33, 687–689 (1994). [144] E. von Roedern, E. Lohof, G. Hessler, M. Hoffmann and H. Kessler, Synthesis and conformational analysis of linear and cyclic peptides containing sugar amino acids, Journal of the American Chemical Society, 118, 10156–10167 (1996). [145] S. Cheng et al., Design and synthesis of novel cyclic RGD-containing peptides as highly potent and selective integrin alpha(IIb)beta(3) antagonists, Journal of Medicinal Chemistry, 37, 1–8 (1994).

REFERENCES

175

[146] K. D. Kopple et al., Conformations of Arg-Gly-Asp containing heterodetic cyclicpeptides – Solution and crystal studies, Journal of the American Chemical Society, 114, 9615–9623 (1992). [147] R. M. Scarborough and D. D. Gretler, Platelet glycoprotein IIb–IIIa antagonists as prototypical integrin blockers: Novel parenteral and potential oral antithrombotic agents, Journal of Medicinal Chemistry, 43, 3453–3473 (2000). [148] J. Samanen et al., Development of a small RGD peptide fibrinogen receptor antagonist with potent antiaggregatory activity in vitro, Journal of Medicinal Chemistry, 34, 3114–3125 (1991). [149] F. E. Ali et al., Conformationally constrained peptides and semipeptides derived from RGD as potent inhibitors of the platelet fibrinogen receptor and plateletaggregation, Journal of Medicinal Chemistry, 37, 769–780 (1994). [150] M. Burger et al., Small peptide inhibitors of the CXCR4 chemokine receptor (CD184) antagonize the activation, migration, and antiapoptotic responses of CXCL12 in chronic lymphocytic leukemia B cells, Blood, 106, 1824–1830 (2005). [151] C. Murdoch, CXCR4: Chemokine receptor extraordinaire, Immunological Reviews, 177, 175–184 (2000). [152] K. Kanbara et al., Biological and genetic characterization of a human immunodeficiency virus strain resistant to CXCR4 antagonist T134, AIDS Research and Human Retroviruses, 17, 615–622 (2001). [153] T. Nanki et al., Stromal cell-derived factor-1-CXC chemokine receptor 4 interactions play a central role in CD4(þ) T cell accumulation in rheumatoid arthritis synovium, Journal of Immunology ,165, 6590–6598 (2000). [154] H. Tamamura et al., Identification of a CXCR4 antagonist, a T140 analog, as an anti-rheumatoid arthritis agent, FEBS Letters, 569, 99–104 (2004). [155] M. Takenaga et al., A single treatment with microcapsules containing a CXCR4 antagonist suppresses pulmonary metastasis of murine melanoma, Biochemical and Biophysical Research Communications, 320, 226–232 (2004). [156] H. Tamamura et al., T140 analogs as CXCR4 antagonists identified as anti– metastatic agents in the treatment of breast cancer, FEBS Letters, 550, 79–83 (2003). [157] T. Nakamura et al., Tachyplesin, a class of antimicrobial peptide from the hemocytes of the horseshoe-crab (tachypleus-tridentatus) – Isolation and chemical-structure, Journal of Biological Chemistry, 263, 16709–16713 (1988). [158] T. Miyata, et al., Antimicrobial peptides, isolated from horseshoe-crab hemocytes, tachyplesin-Il, and polyphemusin-I and polyphemusin-II – Chemical structures and biological-activity, Journal of Biochemistry, 106, 663–668 (1989). [159] M. Morimoto et al., Inhibitory effect of tachyplesin-I on the proliferation of humanimmunodeficiency-virus in vitro, Chemotherapy, 37, 206–211 (1991). [160] T. Murakami, M. Niwa, F. Tokunaga, T. Miyata and S. Iwanaga, Direct virus inactivation of tachyplesin-I and its isopeptides from horseshoe-crab hemocytes, Chemotherapy, 37, 327–334 (1991). [161] K. Kawano et al., Antimicrobial peptide, tachyplesin-I, isolated from hemocytes of the horseshoe-crab (tachypleus-tridentatus) – NMR determination of the beta-sheet structure, Journal of Biological Chemistry, 265, 15365–15367 (1990).

176

DESIGN OF CYCLIC PEPTIDES

[162] M. Masuda et al., A novel anti-HIV synthetic peptide, T-22 ([Tyr5,12, Lys7]Polyphemusin-II), Biochemical and Biophysical Research Communications, 189, 845–850 (1992). [163] H. Tamamura et al., Downsizing of an HIV-cell fusion inhibitor, T22 ([Tyr(5,12), Lys(7)]-polyphemusin II), with the maintenance of anti-HIV activity and solution structure, Bioorganic & Medicinal Chemistry, 6, 473–479 (1998). [164] H. Tamamura et al., A low-molecular-weight inhibitor against the chemokine receptor CXCR4: A strong anti-HIV peptide T140, Biochemical and Biophysical Research Communications, 253, 877–882 (1998). [165] H. Tamamura et al., Effective lowly cytotoxic analogs of an HIV-cell fusion inhibitor, T22 ([Tyr(5,12), Lys(7)]-polyphemusin II), Bioorganic & Medicinal Chemistry, 6, 231–238 (1998). [166] H. Tamamura et al., Pharmacophore identification of a specific CXCR4 inhibitor, T140, leads to development of effective anti-HIV agents with very high selectivity indexes, Bioorganic & Medicinal Chemistry Letters, 10, 2633–2637 (2000). [167] N. Fujii et al., Molecular-size reduction of a potent CXCR4-chemokine antagonist using orthogonal combination of conformation- and sequence-based libraries, Angewandte Chemie – International Edition, 42, 3251–3253 (2003). [168] S. Ueda et al., Structure-activity relationships of cyclic peptide-based chemokine receptor CXCR4 antagonists: Disclosing the importance of side-chain and backbone functionalities, Journal of Medicinal Chemistry, 50, 192–198 (2007). [169] G. Weckbecker et al., Opportunities in somatostatin research: Biological, chemical and therapeutic aspects, Nature Reviews Drug Discovery, 2, 999–1017 (2003). [170] C. E. DiLiberti, The best targets for biogenerics, BioPharm International, 19, 50–52, 54, 56, 58, 60, 62, 64 (2006). [171] D. F. Veber, et al., A potent cyclic hexapeptide analog of somatostatin, Nature, 292, 55–58 (1981).

5 Carbohydrates in Peptide and Protein Design1 Knud J. Jensen and Jesper Brask

5.1

INTRODUCTION

Monosaccharides and amino acids are fundamental building blocks in the assembly of nature’s polymers. They have different structural aspects and, to a significant extent, different functional groups. Oligomerization gives rise to oligosaccharides and peptides, respectively. Carbohydrates and peptides can be found conjoined in nature, e.g. in glycopeptides. It is testimony to the genius of Emil Fischer that he is the founding father of the chemistry of both peptides and carbohydrates, two of nature’s fundamental building blocks in heteroatom-linked biopolymers. Since then, carbohydrate chemistry has progressed through the pioneering work of Claude S. Hudson, Raymond Lemieux, Hans Paulsen and many others, while peptide chemistry has been developed through the work of Vincent du Vigneaud, R. Bruce Merrifield and many more. The topic of ‘carbohydrates in peptide and protein design’ brings these two fields of bioorganic chemistry together again. The aim of this review is the radical redesign of peptide structures, using carbohydrates – particularly monosaccharides and cyclic 1 This chapter is a revised version of K. J. Jensen and J. Brask, Carbohydrates in peptide and protein design, Biopolymers (Pept. Sci.), 80, 747–761 (2005).

Peptide and Protein Design for Biopharmaceutical Applications © 2009 John Wiley & Sons, Ltd. ISBN: 978-0-470-31961-1

Edited by Knud J. Jensen

178

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

oligosaccharides – to produce novel peptides, peptidomimetics and de novo-designed proteins. These hybrid molecules, chimeras, have properties arising largely from the combination of structural characteristics of carbohydrates with the functional group diversity of peptides. This field includes synthetic glycopeptides, sugar (carbohydrate) amino acids as peptidomimetics (see also Chapter 3), carbohydrate scaffolds for nonpeptidal peptidomimetics of cyclic peptides (see also Chapter 4), cyclodextrin-functionalized peptides and carboproteins, i.e. carbohydrate-based protein mimetics (see also Chapter 6). These successful applications demonstrate the general utility of carbohydrates in peptide and protein architecture. In this chapter we will attempt to provide an overview of novel hybrid molecules or chimeric compounds. We will cover carbohydrates in peptide design, but not the reverse, i.e. not peptides in carbohydrate design such as are found in the replacement of glycosidic linkages with amides. There are exhaustive reviews on synthetic glycopeptides [1]; in this chapter we only show a few applications of glycopeptides. But why bother with ‘carbohydrates in peptide and protein design’? In a much-quoted review on peptidomimetics from 1993, Giannis and Kolter anticipated: ‘. . . the potential of the carbohydrate skeleton in the design of nonpeptide ligands. Carbohydrates offer advantages of structural diversity and the facile derivatization with a multitude of functional groups’ [2]. Very briefly, ‘carbohydrates in peptide and protein design’ promises to combine the structural properties of carbohydrates with the functional group diversity of peptides for the design and synthesis of new peptidomimetics and proteinomimetics, as well as for functionalized cyclodextrins.

5.2

CONFIGURATIONAL AND CONFORMATIONAL PROPERTIES OF CARBOHYDRATES

There is no sharp and unequivocal definition of what counts as a carbohydrate. In fact, the very name ‘carbohydrate’ is slightly misleading as it suggests the sole constituents to be carbon and water. While the formula Cx(H2O)x covers many carbohydrates, numerous others do not fit with this, e.g. glucosamine (2-amino-2-deoxy-D-glucose). Carbohydrates, as a wide group of compounds, can be described by Wittgenstein’s concept of ‘family resemblance’, i.e. they are related not by unifying, underlying

PROPERTIES OF CARBOHYDRATES

179

criteria but by a set of interconnecting resemblances among the members of the family [3]. The most prominent carbohydrates are the two trioses (glyceraldehyde), four tetraoses, eight pentoses and sixteen hexoses (for a general overview of carbohydrate chemistry, see [4]). They are distributed in enantiomeric (mirror image) pairs, in which the highest-numbered (furthest from the aldehyde/ketone) asymmetric centre defines whether the pair is of the D- or L-form. Pentoses and hexoses can exist as five- and six-membered cyclized forms, i.e. as furanoses (f) and pyranoses (p), respectively. Substituents on these rings can normally either be equatorially or axially oriented. Aldoses contain an aldehyde functionality, and ketoses (uloses) incorporate a keto functionality, most often at C-2, as in fructose. While the aldose open-chain form contains an aldehyde moiety, the ring-closed form contains a hemiacetal or acetal moiety. The hemiacetal hydroxyl at C-1 (anomeric centre in aldoses) can be in an a- or b-orientation. A prominent nomenclature example is a-D-Glcp: the ‘p’ indicates a sixmembered pyranoid ring, while ‘a’ indicates an axial orientation for the anomeric hydroxyl [5]. There are three different received ways to depict cyclic monosaccharides as pyranoses. Shown in Figure 5.1 is a-D-Glcp in (a) the Haworth representation, (b) the six-membered puckered ring (e.g. chair-like) and (c,d) the representation in which the ring is coplanar with the paper. While the two former are used very widely in carbohydrate chemistry, the coplanar representation is used more in e.g. natural product chemistry. Although they contain the same information, the two former more readily visualize the 3D structure of carbohydrates. At a first approximation, the conformational properties of pyranoses can be exemplified by cyclohexane [4]. Cyclohexane exists in two major OH OH OH

O OH OH

A

HO HO

OH

HO

OH O

HO

HO OH B

Figure 5.1

HO

O

OH

OH HO HO

O

OH C

D

OH

Representations of a-D-Glcp

conformational classes: a rigid chair form and a flexible form. The chair form is free from torsional strain and van der Waals interaction, while the flexible forms suffer from both and thus are energetically less favoured. The group of flexible forms includes boat and skew forms, each having

180

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

four carbons coplanar-aligned, as well as other forms. Large groups attached to the cyclohexane ring prefer the equatorial orientation. However, the 1,3-diaxial interaction between bulky groups in cyclohexane rings can disfavour these conformations. Additional effects, such as hydrogen bonding in 1,3-diols, can make the prediction of conformational properties complicated. Substitution of a methylene moiety with an oxygen atom, i.e. applying this conformational analysis to pyranoses, does not change the general conformational properties. However, the heterocycle has non-equivalent bond lengths and features new stereoelectronic effects. Most prominent among these effects is the anomeric effect, i.e. the preferred axial orientation of an anomeric substituent (most often oxygen). Related effects are the reverse anomeric effect – positively charged anomeric substituents prefer equatorial orientation – and the exoanomeric effect related to the lone pair on the exocyclic oxygen. Other bulky substituents, i.e. other than at C-1, on pyranoid rings still prefer equatorial orientations. In b-D-glucopyranose, all secondary hydroxyls, including 1-OH, are equatorially oriented. Hexopyranoses can have two chair conformations, 4C1 and 1C4, depending on whether the 4- or the 1-substituent is above the plane of the ring (Figure 5.2). Additional important conformational parameters are given by the j, c glycosidic angles. Furanoses, which have a fivemembered heterocyclic ring, prefer puckered forms in order to avoid eclipsed cis-adjacent hydrogen atoms. The envelope (E) and twist (T) conformations are preferred, in which there are quasi-equatorial and quasi-axial orientations of (hydroxyl) substituents. R1 N H

O

H N φ O

R2

N ψ H

R3

4

O

5

5

O 2

3 4

C1

1

O 1

4

3 1

C4

O

O

2

ϕ, ψ angles

Figure 5.2 Left: torsional angles defining the peptide backbone consisting of a-amino acids. Not shown is the o-angle, which determines the amide bond conformation as trans (180°) or cis (0°) [6]. Right: the two chair conformations in pyranoses (4C1 and 1C4), and glycosidic angles [4]

Oligosaccharides can show a staggering structural complexity, as they can be linear, branched or cyclic. Numerous disaccharides and many trisaccharides are easily available. Saccharides can be 1,2-, 1,3-, 1,4-, 1,6- but also 1,1-linked, as in trehalose. Combined with the possibility of

CARBOHYDRATES IN PEPTIDOMIMETICS

181

a- and b-orientations of glycosidic bonds, as well as furanoid and pyranoid ring forms, this generates a high degree of structural diversity for a biopolymer. Functional groups in natural carbohydrates include not only hemiacetal and acetal moieties, primary and secondary alcohols, but also amino groups, e.g. in glucosamine and amino-glycosides, and carboxylic acid groups, e.g. in glucuronic acid. However, these are rare compared to the hydroxyls. Additional modifications of oligosaccharides, such as acetyl and carbamoyl groups in lipochitin oligosaccharide nodulation factors, can give other functionalities to oligosaccharides. The glycosidic linkage can contain O-, S-, N- and C-atoms. Man-made, abiotic carbohydrates can, of course, contain any imaginable chemical modification. It is often stressed that oligosaccharides posses an unsurpassed positional diversity, based on their hydroxyls, due to the possibility of positional isomers (e.g. 1,2- vs. 1,3-linkages), equatorial vs. axial orientation, and the a- vs. b-orientation of the glycosidic linkage (for a clear presentation of this concept, see [7]). However, this takes a purely ‘digital’ view of the information content, based on hydroxyl functional groups, and one should contrast this with the high functional group diversity found in amino acid side chains. The functional group diversity in most common oligosaccharides is similar to that found in the side chains of Ser/Thr, Asn/Gln and Asp/Glu, while the side chains of e.g. Arg (guanidinium), Cys (thiol), His (imidazole), Tyr (phenol) and Trp (indole) do not have an equivalent in common oligosaccharides.

5.3

CARBOHYDRATES IN PEPTIDOMIMETICS

For an extensive overview of peptidomimetics, see Chapter 3. Here, only a few topics comparing conformational aspects of carbohydrates and peptides are presented. While oligosaccharides are secondary gene products, proteins are primary products assembled from a-amino acids. However, a large variety of nonproteinogenic, nonribosomal amino acids and peptides are also found; they are biosynthesized by enzymes and thus are also secondary gene products or elements of these. Examples include b- and g-amino acids as well as a-alkylated amino acids such as a-aminoisobutyric acid (Aib). Furthermore, post-translational modification of ribosomal peptides and proteins, e.g. glycosylation, are also frequently-occurring. Thus, the structural diversity of natural peptides extends far beyond oligomers of proteinogenic a-amino acids. Two of

182

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

the main problems with peptides as drug candidates are their poor bioavailability and their lability to proteolysis. Cyclization is an approach to overcoming the latter, as it can improve metabolic stability (see Chapter 4). The conformation of the peptide backbone can be described by three torsional angles: f (phi), which at 180° implies that the two carbonyls are trans; c (psi), which at 180° implies that the two amides are trans; and o (omega), which at 180° indicates that the amide bond has a trans orientation [6]. Since the amide bonds of the peptide units are relatively rigid groups – with restricted rotation around the C(O)-N amide bond – that are linked into a chain by covalent bonds at the Ca atoms, the only degrees of freedom they normally have are rotations around two bonds (angle of rotation around N-Ca is f and angle of rotation around Ca-C(O) is c) (Figure 5.2). However, most combinations of f and c angles are not allowed because of steric collisions between the side chains and the main chains; Ramachandran plots give combinations of experimental f and c angles in a diagram. Incorporation of carbohydrate structures into peptides, i.e. the use of carbohydrates in peptide design, generates new nonpeptide peptidomimetic structures. Structural elements in peptides include a-helices, parallel and antiparallel b-sheets, other types of helices and a-, b- and g-turns. Of these structures, carbohydrates can readily be envisioned as mimetics of turn motifs, based on their pseudocyclic structure with (pseudo)axial and (pseudo)equatorial orientation of substituents. Turns (see also Chapters 3 and 4) in peptides consisting of a-amino acids are classified by the number of residues in the regular structure: b-turns, four amino acids; g-turns, three amino acids [8]. Turns are stabilized, or constrained, by a hydrogen bond across them, holding the two ends together. The first residue is defined as i; in idealized b-turns, the hydrogen bond is between the carbonyl of the i residue and the NH of the i þ 3 residue, forming a 10-membered ring [9]. In g-turns the hydrogen bond is between the carbonyl of the i residue and the NH of the i þ 2, forming a 7-membered ring. Classical types of b-turns include I, I0 , II, II0 b-turns, characterized by different angles. b-turns are generally located at the surfaces of proteins, are hydrophilic with potentially reactive functional groups, and are often recognition sites [8]. Thus, a b-turn mimetic that often represents an appealing drug target would require control of four separate asymmetric centres [8]. This chapter focuses on structural aspects of carbohydrates in peptide design. In most cases the carbohydrate will function as a scaffold, i.e. as a

GLYCOPEPTIDES

183

relatively rigid structure to which amino acids, amino acid mimetics, sidechain mimetics, peptides and so on can be attached. Thus, the carbohydrate scaffold can display substituents to control distance and geometry between them. In this context, Hirschmann described nonpeptidal peptidomimetics with novel scaffolding [10,11]. In 1977, Walter had already suggested that backbone amide elements might not be required for receptor interaction and that a cyclic scaffold could be used [12]. Scaffolding was used and proven in the 1980s, when Hirschmann described an approach to the design of peptidomimetics, wherein the entire amide backbone of a b-turn was replaced by novel scaffoldings devoid of amide bonds or isosteric replacements (described in more detail in Section 3.5) [10,11]. In summary, carbohydrates as scaffolds in peptide design can utilize a range of structural features, including furanoid vs. pyranoid ring forms, axial vs. equatorial orientation of hydroxyls, relatively rigid pyranoside ring forms, regioisomeric hydroxyls, introduction of amino and carboxyl groups, and monosaccharide vs. oligosaccharide (e.g. cyclodextrin) structures. Carbohydrates are thus candidates as mimetics for e.g. turns or cyclic peptide structures and as templates for the control of the distancegeometry of attached moieties.

5.4

GLYCOPEPTIDES

The chemistry of glycopeptide synthesis has been covered in extensive reviews [1,13], and only a few general considerations shall be outlined here. Glycopeptide synthesis was pioneered by H. Paulsen, H. Kunz, M. Meldal, T. Norberg and others [13]. In the buildingblock approach, the amino acid is glycosylated in solution and then incorporated into the peptide, e.g. by solid-phase synthesis. O-glycosidic linkages are generally acid-labile, but whereas they are often stable enough to withstand treatment with trifluoroacetic acid (TFA) at room temperature, HF will cleave most O-glycosidic bonds, while N-glycosidic linkages from N-acetyl-glycosamines are generally stable to HF. This makes solid-phase glycopeptide synthesis by a 9-fluorenylmethyloxycarbonyl (Fmoc) strategy the preferred approach; upon completion of the synthesis, the final glycopeptide can be released by TFA-containing ‘cocktails’ at room temperature. In the glycosylation of the amino acid, it is most conveniently Na-Fmoc protected, while the carboxylic acid can be masked as the

184

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

pentafluorophenyl (Pfp) ester. This serves a dual function as it is also an ‘activated ester’ very suitable for the following coupling/acylation step [14,15]. Other N- and C-protecting groups have been used with success, as have derivatives with a free carboxylate [1,13]. Redesign of naturally-occurring glycopeptides such as vancomycin will not be covered here. Introduction of glycosylation in a peptide also offers the prospect of protection against proteolysis and improved aqueous solubility. A leading pharmaceutical example is the work to improve the poor solubility of the potent, selective, conformationally-constrained tachykinin NK2 receptor antagonist, MEN 10627, cyclo(Met-Asp-Trp-Phe-DapLeu)cyclo(2b-5b) [16,17]. To overcome this problem, Quartara and coworkers designed, among others, the bicyclic glycopeptide MEN 11420, which incorporates an amide linked N-acetyl-D-glucosamine moiety (Figure 5.3). Indeed, the glycopeptide MEN 11420 proved to be a potent and selective tachykinin NK2 receptor antagonist [17]. Interestingly, the N-acetyl-D-glucosamine moiety did not produce major changes in the affinity profile of the antagonist as compared to the analogous nonglycosylated cyclopeptide MEN 10627; however, MEN 11420 showed significantly improved in vivo potency and duration of action. Thus, Quartara and coworkers demonstrated the concept of glycosylation of peptides as a means to improve ADME properties of peptides. In a different approach, glycosyl-enkephalin conjugates have been prepared to modulate the properties of enkephalin [18,19]. Otvos and

Figure 5.3 The Menarini glycopeptide MEN 11420 [16,17]

CARBOHYDRATES AS SCAFFOLDS

185

coworkers have described a glycopeptide antagonist of the leptin receptor, where the carbohydrate moiety was expected to enhance penetration of the blood–brain barrier [20].

5.5

CARBOHYDRATES AS SCAFFOLDS IN THE DESIGN OF NONPEPTIDE PEPTIDOMIMETICS

In the early 1990s, Hirschmann, Smith, Nicolaou and coworkers published a series of groundbreaking papers in which they described the use of b-D-glucopyranosides as scaffolds in the design and synthesis of nonpeptidal peptidomimetics of the hormone somatostatin, which led to the discovery of a partial somatostatin agonist [10,11]. This was one of the very first successful examples of de novo-designed nonpeptidal peptidomimetics on a novel scaffold. Somatostatins are a family of disulfide-bridged peptide hormones that are found in many vertebrates and mammals, including humans (for a monograph, see [21]; see also [22,23]). The tetradecapeptide somatostatin-14 (SST-14; also referred to as somatotropin release-inhibiting factor (SRIF)-14) – found in the central nervous system, pancreas, stomach, small intestine and thyroid – inhibits the release of a number of hormones, including growth hormone, glucagon and insulin [24–27]. The diverse biological actions attributed to members of the somatostatin family have been correlated to interactions with one or more of five receptor subtypes, termed SSTR1–5 [28,29]. After this work by Hirschmann, Smith, Nicolaou and coworkers these subtypes have been cloned, and hypotheses have been developed about their respective functions [30]. The development of potent somatostatin variants that show selective binding to each particular subtype is of interest in drug discovery research. Some cyclic hexapeptides, e.g. cyclo(Phe-D-Trp-Lys-Thr-Phe-Pro), had been shown to be potent somatostatin agonists [31]. Here, the dipeptide moiety Phe-Pro induces the side chains of the four other amino acids, which make up the b-turn, to assume conformations which permit binding and agonism. In addition, the Phe-Pro moiety provides a favourable hydrophobic interaction with the SRIF receptor. Hirschmann, Smith, Nicolaou and coworkers designed a SRIF mimetic, starting from the relatively inflexible cyclic hexapeptide cyclo(Phe-D-Trp-Lys-Thr-Phe-Pro), L-363,301, on a b-D-glucopyranoside scaffold; b-D-glucopyranoside was chosen because of its well-defined

186

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

conformation and equatorial orientations of the O-1–4 [11]. Their initial target was a glucoside derivative, which carried an O-1 indole side chain (Trp mimetic), an O-6 alkyl amine side chain (Lys mimetic) and 2,3,4-triO-benzyl (Bn), in which the O-4 Bn could mimic Phe-Pro and the O-2 Bn could mimic the other Phe. Similarly, they also prepared the 3-deoxy analogue, which does not carry the ‘superfluous’ 3-O Bn, as well as a whole range of other permutations. Computer modelling (MM2) showed that a local minimum overlapped well with the bioactive conformation of the cyclic hexapeptide. Considerable synthetic effort was required to access these carbohydrate-based peptidomimetic, as depicted in Figure 5.4. O BnO BnO

NH2 O BnO

O NSO2Ph

Figure 5.4 Hirschmann’s SRIF mimetic on a b-D-Glcp template; Bn, benzyl [11]

These peptidomimetics were tested in vitro in radioligand displacement assays on membranes from subclones of the AtT-20 cell line using different radiolabelled somatostatin analogues, as well as in other tests. The compounds bound to the SRIF receptor in a dose-dependent manner and with an affinity in the mM range, qualifying as SRIF agonists. The results validated the claim that hydrogen-bonding to backbone amides is not required for either binding or activation of the SRIF receptor. However, only three of the four side chains in the b-turn of SRIF are required for binding. The 2-O Bn moiety mimics Phe7 and the 4-O Bn moiety mimics Phe-Pro of the cyclic hexapeptide. The authors find that the substituents mimic the interactions that were intended, but the indole ring may contribute very little to binding, unlike the Trp residues in the peptides. Hirschmann and Smith have suggested that b-D-glucose, and more generally monosaccharides, could be privileged platforms for the design of therapeutically important agents [32]. Inspired by the work of Hirschmann et al., Murphy and coworkers have prepared somatostatin mimetics with the iminosugar 1-deoxynojirimycin as a central scaffold [33]. 1-deoxynojirimycin is a nitrogencontaining pyranoside and its secondary amine allows hydrogen binding or chemical modifications.

SUGAR AMINO ACIDS

5.6

187

SUGAR AMINO ACIDS

A sugar (carbohydrate) amino acid (SAA) is a compound with immediate linkages of both amino and carboxy functionalities to a carbohydrate frame [34]. SAAs can be found widely distributed in nature, with sialic acid being the most prominent example (Figure 5.5). Sialic acid is the name for both the family of N- and O-acyl derivatives of neuraminic acid that are found peripherally on glycoproteins and for the most common member of this group, Neu5Ac.

Figure 5.5 The naturally-occurring sugar (carbohydrate) amino acid (SAA) neuraminic acid

Synthetic SAAs can be prepared from commercially available monosaccharides. The amino functionality can be introduced as an azide, cyanide or nitromethane equivalent, followed by subsequent reduction. Likewise, the carboxylic acid moiety can be introduced directly with CO2, by reduction of a cyanide, by oxidation of an olefin or via selective oxidation of a primary alcohol. The latter method was used by Heyns and Paulsen in 1955 for the first synthesis of an SAA, glucosaminuronic acid, by oxidation of glucosamine [35]. Since then, SAAs have been synthesized by Kessler [34], Fleet, Le Merrer, Chakraborty [36,37], Dondoni [38] and many others, and include b-, g- and d-SAAs, as well as both furanoside and pyranoside ring forms (Figure 5.6). Also, sugar mimetics in which the Carbohydrate H2N

O H2N

n

O

n COOH

H2N

COOH n

n O R

O R R COOH n

H2N

COOH n

n

H2N

COOH n

n

Figure 5.6 Schematical presentation of five classes of SAA: a-amino acids, epoxides, oxetanes, furanosides and pyranosides. Inspired by Ref. 39

188

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

ring oxygen has been replaced by nitrogen have gained much interest as inhibitors of glucosidases. Many of these, natural as well as synthetic, can be considered SAAs. Thus, SAAs representing a-, b-, g- and d-amino acids have been prepared, while others are better viewed as dipeptide isosteres or mimetics (for a review of SAA structures, see [39]; see also [34]). SAAs most often contain 3-, 4-, 5- and 6-membered rings; in this chapter we will focus on 5- and 6-membered rings, i.e. furanosides and pyranosides. SAAs are thus conformationally restricted and are used as constrained amino acids, turn mimetics, or as mimetics of cyclic peptides (Figure 5.7). Containing amino and carboxylic acid functional groups, SAAs are particularly well suited as peptidomimetic scaffolds. OH O

HO HO

COOH

HOOC HO HO

NH2

O NH2

O

NH2 OCH3 δ-SAA (also β-anomer)

HO

β-SAA

HOOC HO HO

γ-SAA

HOOC HO HO

O NH2

HO

δ-SAA, Gum (also epimer)

O OH

OH

HO

HO HO

HO

OH HO

O HN HN

O

β-turn mimetic

OH

O

O

O HN

COOH

γ-turn mimetic

OH

HO

OCH3

linear dipeptide isostere

O

NH flexible β-turn mimetic dipeptide isostere

Figure 5.7 Examples of some SAAs and their structural use. Above are SAAs in 4C1 conformation, below are their respective structural roles. For example, the b-SAA can mimic a g-turn when incorporated into a peptide, whereas the g-SAA can mimic a b-turn [34]. Gum is a C-glycosyl glucuronic acid

Kessler, Overhand and coworkers have reported a detailed conformational study of cyclic tetra- and pentamers containing one or two furanoid e-sugar amino acids [40]. For example, a tetramer containing three a-amino acids and a furanoid e-SAA corresponds to a pentapeptide of a-amino acids, except that it has one additional carbon atom due to the SAA (Figure 5.8). As the SAA contains a furan ring, the conformational analysis relies on concepts from nucleoside conformations. In a tetramer containing two a-amino acids and two copies of this SAA, the two SAAs adopted different conformations, one of which has a b-turn-like 9-member internal hydrogen-bond ring structure. This is in contrast to

SUGAR AMINO ACIDS

189

the 10-member hydrogen-bonded ring formed in b-turns and to dipeptide isosteric pyranoid d-SAAs serving as b-turn mimetics. The 9-member hydrogen-bond ring structure formed by the furanoid SAA resembles the retro version of a type VIa b-turn, as the peptide backbone has been reversed, but it also has some similarity to type I and III turns. HO

OH

O N O

H

Figure 5.8 Furanoid e-SAA which can adopt a 9-member hydrogen-bond ring structure

Hirschmann and coworkers prepared an inhibitor of mammalian ribonucleotide reductase (mRR) on an SAA pyranoside scaffold, mimicking a b-turn of the sequence Ac-Phe-Thr-Leu-Asp-Ala-Asp-Phe-OH (Figure 5.9) [41]. The Leu3 and Asp4 functionalities were attached to the 4- and 2-hydroxyls of L-glucose, respectively. Conversion to the tetrahydropyran scaffold then permitted introduction of the methyl group corresponding to Ala5. Finally, Ac-Phe- was attached to the 6-position and -Asp-Phe-OH was connected to the 1-position. The functionalized SAA was found to inhibit mRR with a Ki of 400–500 mM, versus a Ki of 15–20 mM for the heptapeptide sequence. For a combinatorial solid-phase approach, Sofia et al. have prepared SAAs to be used as scaffolds in the assembly of pharmacophore mapping libraries [42]. Chemical diversity was achieved by functionalization of the C-6 carboxylic acid, the amine and the O-4 hydroxy group. The solid-phase strategy relied on anchoring the Fmoc-protected SAA to an amino acidderivatized trityl resin. The hydroxy and amino functionalities could then be manipulated, after which acid treatment would cleave the compound from the solid phase. In another combinatorial approach, Fleet and

Figure 5.9

Hirschmann’s mRR inhibitor [41].

190

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

coworkers prepared a 99-membered library of derivatized furanose SAAs containing three orthogonal points of diversification: amine, hydroxy and carboxylic acid [43]. The library was prepared by parallel synthesis in solution using scavenger resins. Despite some progress, synthesis of libraries of oligo- and polysaccharides in solution or on solid phase can still be challenging. With SAAs, biopolymers can be assembled using well-established peptide chemistry to mimic nature’s sugar structures, as originally envisioned by Nicolaou in his outline for carbopeptoids [44]. Further, hybrids of amino acids, carbohydrates and SAAs can be created to give promising drug candidates with resistance to glycosidases as well as to many proteases. SAAs have also been assembled to mimic oligonucleotides. Kessler and Fleet have prepared a range of cyclic SAA macrolactams [45–48], while Gregar and Gervay-Hague have studied linear oligomers derived from amidelinked neuraminic acid analogues [49]. SAAs can be used in the design of foldamers (for foldamer design, see Sections 6.5 and 6.6.2). The research groups of van Boom and Kessler have both explored the potential of linear and cyclic oligomers, consisting of both SAAs and a-amino acids. Often prepared by solid-phase methodologies, the conformations of these structures were studied by NMR and CD spectroscopy, and modelled by MD calculations. In 2002, Kessler and coworkers reported b-Ala and GABA for these studies, as these amino acids are unsubstituted and hence known not to induce any secondary structure. With linear hexameric structures with every second residue an SAA, a helical conformation could be detected for the b-Ala case in acetonitrile, whereas the GABA structures apparently did not form a stable conformation [50]. SAAs are often incorporated into peptide sequences as turn mimetics. In 2001, van Boom and coworkers described the use of the sequence Ac-Lys-Lys-Tyr-Thr-Val-Ser-Ile(SAA)-Lys-Lys-Ile-Thr-Val-Ser-Ile-OH for the first noncyclic b-hairpin structure in an SAA-containing peptide (Figure 5.10) [51]. The SAA building block was easily incorporated into the polypeptide by means H Ac-Lys-Lys-Tyr--Thr-Val-Ser-Ile N N-terminus O HO-Ile-Ser-Val-Thr-Ile-Lys-Lys C-terminus

OH O

Figure 5.10 The van Boom b-hairpin peptide [51]

SUGAR AMINO ACIDS

191

of Fmoc chemistry. While unfolded in water, the sequence was shown to adopt a b-hairpin structure upon addition of methanol. Thus the SAA construction kit [34], including capabilities for linear, flexible b-turns, rigid b-turns and g-turns, has been utilized in these peptide sequences (some are illustrated in Figure 5.7). Once the conformational influence of SAAs in peptides is known, the information can be utilized for the construction of biologically-active sequences, with SAA-induced turns or constrains in general. In line with this, Kessler and coworkers have prepared a selection of somatostatin analogues, originally based on the sequence cyclo[(SAA)-Phe-D-Trp-LysThr-] [52]. Inhibition of growth-hormone release has been demonstrated, as have antiproliferative and apoptotic activity. The sequences were assembled on solid phase with cyclization in solution using standard methods. Likewise, a series of Leu-enkephalin sequences were prepared as H-Tyr-(SAA)-Phe-Leu-OMe, with the SAA replacing Gly-Gly of the natural sequence [53,54]. Such sequences prepared by Chakraborty et al. have shown analgesic activity similar to that of Leu-enkephalin methyl ester [55,56]. Similar, the group of Toth has used SAAs to form glycosylated enkephalins via amide-bond linkages to the C-terminal [57]. A 1-azido glucuronic acid was anchored to a trityl resin through the carboxylic acid, after which the amine functionality was formed by onresin reduction with propane-1,3-dithiol. Subsequent acylation with the SAA would generate a disaccharide analogue, and with Fmocbased peptide synthesis the Leu- and Met-enkephalins could be assembled. One glycosylated Leu-enkephalin was found to inhibit electrically-stimulated muscle contractions much more effectively than Leu-enkephalin amide, with a strong selectivity toward the d-receptor. Integrins are cell-surface receptors that play a role in the interaction with the extracellular matrix and in signal transduction. Kessler and coworkers prepared integrin ligands based on the RGD-motif cyclo(Arg-Gly-Asp-D-Phe-Val), with fully benzylated glucosyluronic acid methylamine (Gum; Figure 5.7) substituting D-Phe-Val (for more details, see Chapter 4) [58]. Both a- and b-anomers showed high avb3activity, but the latter showed reduced selectivity, with high activity against the aIIbb3 receptor as well. This effect is explained by larger flexibility in the structure based on b-Gum. In another approach to improving pharmacokinetic properties, Val was substituted by glycosylated amino acids, such as Lys-SAA conjugates. In this case, the resulting free amino functionality of the SAA could be used for 18F radio-labelling with 4-nitrophenyl 2-[18F]fluoropropionate, to enable positron emission

192

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

Figure 5.11

Kessler’s 18F-labelled integrin ligand for PET studies [59,60]

tomography (PET) studies of avb3 integrins (Figure 5.11) [59,60]. Also, the group of van Boom reported the preparation of cyclic RGD peptides containing either one or two furanoside SAAs [61]. Solid-phase synthesis on Kaiser’s oxime resin using Na-Boc-protected building blocks gave the cyclic, SAA-containing peptide. Finally, Stick and coworkers have prepared some a,a-dialkylated amino acids derived from protected hexo-furanosides and -pyranosides using the Corey–Link reaction (Figure 5.12) [62]. Coupling of a,a-dialkylated azido acids with the corresponding amino acid methyl esters proved somewhat difficult; however, one hexapeptide was prepared [63]. These a-alkylated SAAs are analogues of Aib and appear to cause similar difficulties to Aib residues in peptide couplings. O O

X O Y

X O O

O O X

O

O

O

BnO

O

O

O O BnO

O Y

Y

OCH3

X O Y

OCH3

Figure 5.12 Examples of a-alkylated SAAs, X, Y: N3, CO2Me; N3, CO2H; NH2, CO2Me [63]

CYCLODEXTRIN–PEPTIDE CONJUGATES

5.7

193

CYCLODEXTRIN–PEPTIDE CONJUGATES

Cyclodextrins are cyclic oligosaccharides assembled from a-1,4-linked Glcp units, which are obtained by partial enzymatic hydrolysis of starch (for a general review, see [64]). The ring size is indicated by the prefix, hence a-, b- and g-cyclodextrins comprise six, seven and eight glucose units, respectively. They are crystalline nonhygroscopic compounds with torus-like macrocycles of a-1,4-Glcp in the 4C1 conformation, which imparts them with interesting properties [64]. Larger structures are also known but are more rare. The cyclic structure with a-1,4-linkages in combination with the relatively rigid monomer provides cyclodextrins with a central hydrophobic cavity and a hydrophilic exterior. The inside is lined by the H-3 and H-5 of the glucose monomers and the glycosidic oxygens, which together provide the hydrophobic environment. The diameters of the tubular cavities are 0.57, 0.78 and 0.95 nm, for a-, b- and g-cyclodextrins, respectively, while the outer diameters are 1.37, 1.53 and 1.69 nm, and the height for all is 0.78 nm. The ring has the shape of a conical cylinder or lamp shade. The cavities of cyclodextrins can thus act as hosts for hydrophobic small organic molecules, e.g. substituted benzenes, but also for peptides and the aromatic side chains of peptides. They can thus form inclusion complexes. The secondary hydroxyls, 2- and 3-OH, are above the tube arrangement, while the primary hydroxyls, 6-OH, are below. Substituents on the rim can provide cyclodextrins with some catalytic activity. The driving force for the inclusion of a guest, i.e. complex formation, is substitution of water molecules in the hydrophobic cavity for a hydrophobic molecule. A fair number of substituted cyclodextrins are commercially available and a plethora of chemicallyor enzymatically-modified cyclodextrins have been reported (for a review of nonpeptide caps on cyclodextrin, see [65]). This chapter focuses on designs and molecules in which carbohydrates and peptides are covalently linked. Several different types of covalent peptide–cyclodextrin conjugate have been reported. Among the first were publications by Parrot-Lopez et al. on the preparation of cyclodextrins singly substituted with amino acids [66] or peptides [67], by Vecchio and ˚ kerfeldt and DeGrado in 1994, by Hanessian coworkers in 1992, by A and coworkers in 1995 and by Moroder and coworkers and Stoddart in separate publications in 1996, all of which will be discussed below. Cyclodextrins have been used extensively in the de novo design of potential biomimetic catalysts. However, in general these designs do not incorporate peptides as such, with a few exceptions [68]. In this context it is

194

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

interesting that a Pd(II) conjugate with a thiol-functionalized cyclodextrin showed biomimetic peptidase activity at Xxx-Pro sites [69]. Here we should also mention that noncovalent interactions between carbohydrates and peptides have been studied, e.g. binding of b-cyclodextrin to the Alzheimer Ab(1–40) peptide at two sites [70]. Another noncovalent interaction is the insertion of functionalized cyclodextrins into the cavity of bacterial pore-forming toxin, staphylococcal a-hemolysin, which modulates the charge-selectivity of the transmembrane pore [71]. In an early work, Galons et al. prepared mono-(6-amino-6-deoxy-2,3di-O-methyl)-hexakis (2,3,6-tri-O-methyl)-b-cyclodextrin, in which a single 6-OH in b-cyclodextrin had been substituted for an amino group [67]. This derivative was obtained by introduction of a single 6-azido group, followed by methylation of the 2,3-hydroxyls and finally reduction of the azide to give the amine. This monoamine derivative of b-cyclodextrin was N-acylated with Na-Boc-protected amino acids or Leu-enkephalin, which were subsequently deprotected (Figure 5.13) [72]. H-Tyr-Gly-Gly-Phe-Leu NH O O

OCH3 O

O O O

O

O

6

Figure 5.13 Leu-enkephalin-functionalized b-cyclodextrin [72]

In 1992, Di Blasio et al. reported a b-cyclodextrin derivative with cyclo-(His-Pro), a diketopiperazine (DKP), covalently attached through the side-chain imidazole to C-6 [73]. X-ray studies revealed that 6-deoxy6-cyclo(His-Leu)-b-cyclodextrin was in a ‘sleeping swan’-like fold, with the hydrophobic side chain of Leu nested inside the cavity of the macrocycle [73], i.e. with self-inclusion of the grafted molecule. Self-inclusion (self-complexation) has also been observed in a tyrosinyl–cyclodextrin conjugate, where the tyrosinyl moiety occupied the cyclodextrin cavity [74]. In an attempt to improve the bioavailability of the highly d-opiod receptor-selective disulfide-linked cyclic peptide [p-I-Phe4]DPDPE, a lead compound for the design of d-opiod analgesics, Hruby and coworkers acylated per-O-methyl-derivatized mono-6-amino-6-deoxyb-cyclodextrin with [p-I-Phe4]DPDPE [75]. The conjugate, per-O-methyl

CYCLODEXTRIN–PEPTIDE CONJUGATES

195

mono-6-([p-I-Phe4]DPDPE-NH)-6-deoxy-b-cyclodextrin, proved less potent and selective in the rat brain-binding assay. However, it showed antinociceptive properties (painful stimulus) in the mouse tail flick test. The authors speculated that the loss of potency could be due to the loss of a negative charge in the C-terminus or due to steric interference. In 1995, Hanessian and coworkers described a range of methods to covalently modify a-, b- and g-cyclodextrins with amino and alkenyl functionalities [76]. They reported mainly monofunctionalized derivatives, but also some di-, tri- and tetrafunctionalized ones. In 1996, Stoddart and coworkers reported the synthesis of highly substituted, symmetrical amino acid derivatives of b-cyclodextrin [77]. The amino groups of hepta-(6-amino-6-deoxy) b-cyclodextrin, obtained through the hepta-azide, were per-acylated with Boc-Phe-OH; the Boc groups were then removed with TFA. H-Cys-OH was anchored through the side chain as the thioether by displacement of halides in hepta-iodo or -bromo cyclodextrin derivatives; this gave a poly-zwitterionic compound with improved solubility compared to the parent cyclodextrin. The corresponding heptakis (2,3-di-O-methyl) derivative was also prepared. X-ray crystal structures showed a loss of C7 symmetry, possibly due to circumferal intramolecular hydrogen bonds. ˚ kerfeldt and DeGrado reported an adaptation of Mutter’s In 1994, A TASP (template-assembled synthetic protein; see Chapter 6) concept, in which they used a b-cyclodextrin derivative instead of a cyclic peptide as a template for the anchoring of peptides [78]. They prepared the 2,3permethyl ether of b-cyclodextrin, which was 6-O alkylated with tertbutyl bromoacetate to generate the hepta-ester, which upon treatment with TFA-CH2Cl2 gave the hepta-carboxylic acid derivative. HBTUmediated coupling of b-sheet heptapeptide amide H-Trp-Ser-Leu-SerLeu-Ser-Leu-NH2, through the Na-amine, gave the hepta-functionalized peptide–cyclodextrin conjugate. The novel conjugate was not subjected to biophysical characterization. The synthetic work also highlighted the problems with obtaining uniform, homogenous products of functionalized cyclodextrins. In 1998, Moroder and coworkers reported an extensive study of b-cyclodextrin as a carrier of peptide hormones by mono- or multivalent display [79]. The C-terminal tetrapeptide amide sequence (H-Trp-NleAsp-Phe-NH2) of the gastrointestinal hormone gastrin is the shortest sequence capable of exhibiting all the biological properties, though with reduced potency. Moroder and coworkers functionalized mono(6-deoxy-6-succinoylamino-)-b-cyclodextrin with a 4-carbon succinoyl spacer, to which they anchored the tetrapeptide through the Na-amine,

196

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN O

O

Peptide

Peptide NH O

NH O

OH

HO HO

O

O

O HO

O HO

HO

O 7

HO 6 O

Figure 5.14 Mono- and multifunctionalized b-cyclodextrin. For monofunctionalized cyclodextrin, peptide: H-Trp-Nle-Asp-Phe-NH2 or H-Ala-Tyr-Gly-Trp-NleAsp-Phe-NH2. For heptafunctionalized cyclodextrin, peptide: H-Trp-Nle-Asp-PheNH2 [79]

forming an amide (Figure 5.14). Next, they prepared the analogous heptapeptide derivative with an N-terminal extended sequence. Finally, they anchored seven copies of the tetrapeptide sequence to hepta-(6deoxy-6-succinoylamino-)-b-cyclodextrin. After coupling of the protected peptides, the peptides were deprotected with TFA. The monoconjugated tetra- and heptapeptides largely retained the (GPCR) binding and signal transduction efficacy; the additional tripeptide in the heptapeptide sequence appeared to mainly function as a spacer. In contrast, hereto, oligomeric representation in the heptafunctionalized derivative significantly impaired the ligand receptor recognition process, as the heptaconjugate showed, per peptide moiety, a significantly reduced binding affinity. The failure of multivalent display could be due to steric interferences of the peptide moieties, as well as collapse of the chains on the template. They found indications for beneficial interactions of the b-cyclodextrin moiety with the receptor surface. The nonspecific binding events make b-cyclodextrins promising as candidates for conjugation. These studies were preceded by work in the same group on the amidation of mono-(6-deoxy-6-succinoylamino-)-b-cyclodextrin with the protease inhibitor H-Leu-Leu-Nle-H [80] They have also reported the crystal structure of human b-tryptase as a 1 : 2 complex with b-cyclodextrin difuntionalized with short peptide inhibitors [81]. This proved the concept of cyclodextrins functioning as scaffolds to control the distance-geometry in the display of ligands for interaction with multimeric protein complexes. The ability of cyclodextrins to bind hydrophobic guest molecules in their central cavity has been utilized in the design of novel peptide-based catalysts as potential enzyme mimetics. In 2000, Ueno and coworkers reported the synthesis of four 19-mer peptide-b-cyclodextrin hybrids [82].

CYCLODEXTRIN–PEPTIDE CONJUGATES

197

They designed a-helical peptides, which were stabilized by two Glu-Arg salt bridges on one side of the a-helix, while a b-cyclodextrin (anchored through a 6-amino-6-deoxy group to a Glu side chain) was placed on the other, as well as a Glu and a His. They prepared two of these hybrids with different structural permutations, as well as two constructs lagging the catalytic Glu. They studied the hydrolysis of the D- and L-enantiomers of Boc-Phe-OpNP (pNP: 4-nitrophenyl) and observed a modest catalytic activity by these trifunctional constructs. The authors further described the synthesis of three a-helical 17-mer peptides functionalized with a g-cyclodextrin and one or two pyrene units [83]. Two pyrene units can be included in the cavity of a g-cyclodextrin and these peptides were studied for their ability to dimerize. Similarly, the Ueno group has also prepared four a-helical peptides carrying a b-cyclodextrin and a dansyl (Dns) moiety as a model system for molecular sensing [84]. Binding of guest compounds displaced the Dns moiety from the b-cyclodextrin cavity, an event which could be followed by fluorescence spectroscopy and circular dichroism spectroscopy. In another design, a-helical peptides carrying b-cyclodextrin and two naphthalene moieties were used to study binding of guests, as displacement of one of the naphthalene groups from the cyclodextrin cavity could be followed by excimer fluorescence [85]. This design could potentially be used to detect molecules in aqueous solutions. These principles have also been used in the design of two a-helical peptides carrying an a-cyclodextrin as well as a pyrene and a nitrobenzene moiety as molecule-responsive fluorescence sensors [86]. Finally, several groups have used anchoring of peptide chains through the modified O-6 of cyclodextrin scaffolds as a means to improve the pharmacological properties of peptides [75,87] and as an aid in drug targeting [88]. Cyclodextrins are fascinating molecules with very interesting properties due to their central cavity, with its ability to bind guest molecules, and the hydroxyls on the rims, which can be functionalized. Cyclodextrins are likely to find numerous new applications, and new methods for their derivatization will improve their utility. However, at present they have a number of limitations: it is difficult to access cyclodextrins other than the a-D-glucose-based a-, b- and g-cyclodextrins; it is nontrivial to modify the central cavity; and the cyclic nature of these molecules has made it difficult to differentiate between the different 6-OHs, which has limited their use as structural elements in molecular architecture with peptides. However, a procedure for the regioselective modification of benzylated cyclodextrins has been reported [89].

198

5.8

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

CARBOPROTEINS: PROTEIN MODELS ON CARBOHYDRATE TEMPLATES

‘Carbopeptides’ and ‘carboproteins’ (for a review, see [90]) are peptide and protein chimeras assembled on a carbohydrate template, inspired by Mutter’s concept of TASPs [91]. Monosaccharides were utilized as templates due to their polyfunctionality, the relative rigidity of their pyranose ring forms and their ease of access to epimers. Furthermore, a vast literature describes the regiospecific manipulation of their functional groups. Using the primary and secondary hydroxyls of mono- or disaccharides should provide flexible control of the directionality and distances between anchoring points for peptide chains in the de novo design of protein models. Carboproteins have been designed to mimic a-helix bundles [92]. A convergent modular synthesis strategy was adopted in which peptide aldehydes were coupled to the carbohydrate template by oxime ligation [93–97]. This provided carboproteins on Galp, Glcp and Altp templates (Figure 5.15) [95,96]. Biophysical studies with CD and H-D exchange NMR spectroscopy clearly indicated formation of a-helix bundles.

P e p t i d e

P e p t i d e

NH

HN P e p t i d e

N

N

O O

O

OO

N O

O

O

R′O O

R′O

R′O

OR

NH

O O

R′O

Altp

O

O

HN

P e p t i d e

OR′

O–N R′O R′O

OR

O R′O

OR

Glcp

Figure 5.15 Left: a carboprotein prepared by chemoselective oxime ligation on an aminooxy acetyl functionalized Galp template. R: OCH3 or C6H4-NHCOCH2SH; peptide: Ac-EALEKALKEALAKLG- or Ac-YEELLKKLEELLKKA-. Right: Glcp and Altp templates; R: peptide with C-terminal linker [95–97]

CONCLUSION

199

Surprisingly, a carboprotein assembled on an Altp template showed a higher content of a-helix than similar carboproteins prepared on Galp and Glcp [96]. This indicated that the template has a controlling effect on the a-helix bundle structure. However, small-angle X-ray scattering (SAXS) indicated that a 3 þ 1-helix structure had formed, instead of the expected 4-helix bundle [94]. Carboproteins are described in greater detail in Chapter 6.

5.9

CONCLUSION

The hydrophilicity of carbohydrates has been employed in the design of neoglycopeptides with improved physical and biochemical properties compared to the unmodified peptides. The great potential of carbohydrates as scaffolds in peptide redesign is based on their structural features, including furanoid vs. pyranoid ring forms, axial vs. equatorial orientation of hydroxyls, relatively rigid pyranoside ring forms, regioisomeric hydroxyls, introduction of amino and carboxyl groups, and monosaccharide vs. oligosaccharide (e.g. cyclodextrin) structures. This potential has been converted into successful applications of carbohydrates in the de novo design of glycopeptides, peptidomimetics, proteinmimetics and cyclodextrin–peptide conjugates. Carbohydrate derivatives, for example SAAs, have proven especially valuable in the design of turn mimetics, as furanosides and pyranosides can display substituents with an equatorial orientation. Monosaccharides have also been used quite successfully as central, organizing scaffolds in mimetics of cyclic peptides and for template-assembled proteinmimetics. For the most part, two properties of cyclodextrins have been utilized: the well-defined distances between different 6-OHs and the hydrophobicity of the cavity. Still, only a limited selection of carbohydrates has been used so far; for example, the axial vs. equatorial orientation of hydroxyls has not been fully exploited for display of ligands. The configurational and conformational properties of carbohydrates have to be balanced with appropriate spacers from the carbohydrate ring substituents. Too-long spacers are likely to abolish the chiral information in the carbohydrate moiety. There are only a few examples of a direct effect from the stereochemical nature of a carbohydrate template, one being Altp vs. Glcp carboproteins [96].

200

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

REFERENCES [1] A. M. Jansson, P. M. St Hilaire and M. Meldal, Synthesis of glycopeptides, In: M. Goodman, A. Felix, L. Moroder and C. Toniolo, Synthesis of Peptides and Peptidomimetics, Houben–Weyl Methods of Organic Chemistry, E22b, Thieme, Stuttgart, 2004, 235–332. [2] A. Giannis and T. Kolter, Peptidomimetics for receptor ligand – discovery, development, and medical perspectives, Angew. Chem. Int. Ed. Engl., 32, 1244–1267 (1993). [3] L. Wittgenstein, Philosophische Untersuchungen, Suhrkamp Verlag, Frankfurt, 1977. [4] P. Collins, R. Ferrier, Monosaccharides, Wiley, Chicester, UK, 1995. [5] IUPAC nomenclature for carbohydrates, Carbohydr. Res., 297, 1–92, 1997. [6] C. Branden and J. Tooze, Introduction to Protein Structure, 2nd. Ed., Garland Publishing, New York, 1999. [7] R. Laine, The information-storing potential of the sugar code, In: H.-J. Gabius and S. Gabius, Glycosciences, Chapman & Hall, Weinheim, 1997, 1–14. [8] M. Kahn and M. Eguchi, Synthesis of peptides incorporating b-turn inducers and mimetics, In: M. Goodman, A. Felix, L. Moroder and C. Toniolo, Synthesis of Peptides and Peptidomimetics, Houben-Weyl Methods of Organic Chemistry, E22c, Thieme, Stuttgart, 2004, 695–740. [9] C. M. Venkatachalam, Stereochemical criteria for polypeptides and proteins. V. Conformation of a system of 3 linked peptide units, Biopolymers, 6, 1425–1436 (1968). [10] R. Hirschmann, K. C. Nicolaou, S. Pietranico, J. Salvino, E. M. Leahy, P. A. Sprengeler, G. Furst, A. B. Smith, C. D. Strader, M. A. Cascieri, M. R. Candelore, C. Donaldson, W. Vale and L. Maechler, Nonpeptidal peptidomimetics with a b-D-glucose scaffolding. A partial somatostatin agonist bearing a close structural relationship to a potent, selective substance P antagonist, J. Am. Chem. Soc., 114, 9217–9218 (1992). [11] (a) K. C. Nicoloau, J. M. Salvino, K. Raynor, S. Pietranico, T. Reisine, R. M. Freidinger and R. Hirschmann, Design and synthesis of a peptidomimetic employing b-D-glucose for scaffolding, In: J. E. Rivier and G. R. Marshall, Peptides (Chemistry, Structure and Biology), Proc. 11th American Peptide Symposium, ESCOM, Leiden, 1990, 881–884. (b) R. Hirschmann, K. C. Nicolaou, S. Pietranico, E. M. Leahy, J. Salvino, B. Arison, M. A. Cichy, P. G. Spoors, W. C. Shakespear, P. A. Sprengeler, P. Hamley, A. B. Smith, T. Reisine, K. Raynor, L. Maechler, C. Donaldson, W. Vale, R. M. Freidinger, M. R. Cascieri and C. D. Strader, De novo design and synthesis of somatostatin non-peptide peptidomimetics utilizing b-D-glucose as a novel scaffolding, J. Am. Chem. Soc., 115, 12550–12568 (1993). [12] R. Walter, Identification of sites in oxytocin involved in uterine receptor recognition and activation, Federation Proceedings, 36, 1872–1878 (1977). [13] H. Paulsen, Advances in selective chemical synthesis of complex oligosaccharides, Angew. Chem. Int. Ed. Engl., 21, 155–224 (1982). [14] M. Meldal and K. J. Jensen, Pentafluorophenyl esters for the temporary protection of the a-carboxy group in solid phase glycopeptide synthesis, Chem. Comm., 483–485 (1990).

REFERENCES

201

[15] K. J. Jensen, M. Meldal and K. Bock, Glycosylation of phenols: Preparation of 1,2-cis and 1,2-trans glycosylated tyrosine derivatives to be used in solid-phase glycopeptide synthesis, J. Chem. Soc. Perkin Trans. 1, 2119–2129 (1993). [16] D. Pinzani, A. M. Papini, M. E. Vallecchi, M. Chelli, M. Ginanneschi, G. Rapi, L. Quartara, R. Patacchini, C. A. Maggi and F. M. Arcamone, Glycosyl derivatives of NK2 tachykinin receptor antagonists, Bioorg. Med. Chem. Lett., 6, 367–372 (1996). [17] R.-M. Catalioto, M. Criscuoli, P. Cucchi, A. Giachetti, D. Giannotti, S. Giuliani, A. Lecci, A. Lippi, R. Patacchini, L. Quartara, A. R. Renzetti, M. Tramontana, F. Arcamone and C. A. Maggi, MEN 11420 (Nepadutant), a novel glycosylated bicyclic peptide tachykinin NK2 receptor antagonist, Brit. J. Pharmacology, 123, 81–91 (1998). [18] L. Varga, S. Horvat, C. Lemieux and P. W. Schiller, Synthesis and biological activity of some glucose–enkephalin conjugates, Int. J. Peptide Protein Res., 30, 371–378 (1987). [19] M. Skuric, J. Horvat, S. Horvat, N. N. Chung and P. W. Schiller, Acetylated glucopyranosyl esters of enkephalins, Int. J. Peptide Protein Res., 43, 402–409 (1994). [20] L. Otvos, M. Terrasi, S. Cascio, M. Cassone, G. Abbadessa, F. De Pascali, L. Scolaro, D. Knappe, M. Stawikowski, P. Cudic, J. D. Wade, R. Hoffmann and E. Surmacz, Development of a pharmacologically improved peptide agonist of the leptin receptor, Biochim. Biophys. Acta., 1783, 1745–1754 (2008). [21] M. T. McQuillan, Somatostatin, vol. 2, Eden, Quebec, 1980. [22] C. H. S. McIntosh, Gastrointestinal somatostatin: Distribution, secretion and physiological significance, Life Sci., 37, 2043–2058 (1985). [23] C.-J. Su, J. W. White, W. H. Li, C.-C. Luo, M. L. Frazier, G. F. Saunders and L. Chan, Structure and evolution of somatostatin genes, Mol. Endocrinol, 2, 209–216 (1988). [24] P. Brazeau, W. Vale, R. Burgus, J. Rivier and R. Guillemin, Inhibition of growth hormone secretion in the rat by synthetic somatostatin, Science, 179, 77–79 (1973). [25] D. Koerker, W. Ruch, E. Chideckel, J. Palmer, C. Goodner, J. Ensinck and C. Gale, Somatostatin: Hypothalamic inhibitor of the endocrine pancreas, Science, 184, 482–484 (1974). [26] J. Polak, A. Pearse, L. Grimelius, S. R. Bloom and A. Arimura, Growth-hormone release-inhibiting hormone in gastrointestinal and pancreatic D-cells, Lancet, 1, 1220–1222 (1975). [27] S. J. Konturek, J. Tasler, W. Obtulowicz, D. H. Coy and A. V. Schally, Effect of growth hormone-release inhibiting hormone on hormones stimulating exocrine pancreatic-secretion, J. Clin. Invest., 58, 1–6 (1976). [28] Y. Yamada, S. R. Post, K. Wong, H. Tager, G. I. Bell and S. Seino, Cloning and functional characterization of a family of human and mouse somatostatin receptors expressed in brain, gastrointestinal trcat, and kidney, PNAS, 89, 251–255 (1992). [29] K. Raynor, A. O’Carroll, H. Kong, K. Yasuda, L. C. Mahan, G. I. Bell and T. Reisine, Characterization of cloned somatostatin receptors SSTR4 and SSTR5, Mol. Pharmacol., 44, 385–392 (1993). [30] K. Raynor, W. A. Murphy, D. H. Coy, J. E. Taylor, J.-P. Moreau, K. Yasuda, M. I. Bell and T. Reisine, Cloned somatostatin receptors – Identification of subtype-selective peptides and demonstration of high-affinity binding of linear peptides, Mol. Pharmacol., 43, 838–844 (1993).

202

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

[31] D. F. Veber, R. M. Freidinger, D. S. Perlow, W. J. Paleveda, F. W. Holly, R. G. Strachan, R. F. Nutt, B. H. Arison, C. Homnick, W. C. Randall, M. S. Glitzer, R. Saperstein and R. Hirschmann, A potent cyclic hexapeptide analog of somatostatin, Nature, 292, 55–57 (1981). [32] R. Hirschmann, L. Ducry and A. B. Smith, Development of an efficient, region- and stereoselective route to libraries based on the b-D-glucose scaffold, J. Org. Chem., 65, 8307–8316 (2000). [33] V. Chanault, J. Lalot and P. V. Murphy, Synthesis of somatostatin mimetics based on 1-deoxy nojirimycin, ChemMedChem, 3, 1071–1076 (2008). [34] S. A. W. Gruner, E. Locardi, E. Lohof and H. Kessler, Carbohydrate-based mimetics in drug design: Sugar amino acids and carbohydrate scaffolds, Chem. Rev., 102, 491–514 (2002). [35] K. Heyns and H. Paulsen, Synthese der D–Glucosaminuronsa¨ure (2-amino-2-desoxyD-glucuronsa ¨ ure) und einige ihrer Derivate, Chem. Ber., 88, 188–195 (1955). [36] T. K. Chakraborty, S. Ghosh and S. Jayaprakash, Sugar amino acids and their uses in designing bioactive molecules, Curr. Med. Chem., 9, 421–435 (2002). [37] T. K. Chakraborty, S. Jayaprakash and S. Ghosh, Sugar amino acids – Novel peptidomimetics and their potential in combinatorial synthesis, Comb. Chem. High Throughput Screen., 5, 373–387 (2002). [38] A. Dondoni and A. Marra, Methods for anomeric carbon-linked and fused sugar amino acid synthesis: The gateway to artificial glycopeptides, Chem. Rev., 100, 4395–4421 (2000). [39] M. D. P. Risseuw, M. Overhand, G. W. J. Fleet and M. I. Simone, A compendium of sugar amino acids (SAA): Scaffolds, peptide- and glycomimetics, Tetrahedron Asymmetry 18, 2001–2010 (2007). [40] R. M. van Well, L. Marinelli, C. Altona, K. Erkelens, G. Siegal, M. van Raaij, A. L. Llamas–Saiz, H. Kessler, E. Novellino, A. Lavecchia, J. H. van Boom and M. Overhand, Conformational analysis of furanoid e-sugar amino acids containing cyclic peptides by NMR spectroscopy, molecular dynamics simulation, and X-ray crystallography: Evidence for a novel turn structure, J. Am. Chem. Soc., 125, 10822–10829 (2003). [41] A. B. Smith, S. Sasho, B. A. Barwis, P. Sprengeler, J. Barbosa, R. Hirschmann and B. S. Cooperman, Design and synthesis of a tetrahydropyran-based inhibitor of mammalian ribonucleotide reductase, Bioorg. Med. Chem. Lett., 8, 3133–3136 (1998). [42] M. J. Sofia, R. Hunter, T. Y. Chan, A. Vaughan, R. Dulina, H. Wang and D. Gange, Carbohydrate-based small-molecule scaffolds for the construction of universal pharmocophore mapping libraries, J. Org. Chem., 63, 2802–2803 (1998). [43] A. A. Edwards, O. Ichihara, S. Murfin, R. Wilkes, M. Whittaker, D. J. Watkin and G. W. J Fleet, Tetrahydrofuran-based amino acids as library scaffolds, J. Comb. Chem., 6, 230–238 (2004). [44] K. C. Nicolaou, H. Flo¨rke, M. G. Egan, T. Barth and V. A. Estevez, Carbanucleotides and carbopeptoids – New carbohydrate oligomers, Tetrahedron Lett., 36, 1775–1778 (1995). [45] E. Locardi, M. Sto¨ckle, S. Gruner and H. Kessler, Cyclic homooligomers from sugar amino acids: Synthesis, conformational analysis, and significance, J. Am. Chem. Soc., 123, 8189–8196 (2001). [46] B. A. Mayes, R. J. E. Stetz, W. G. Ansell and G. W. J. Fleet, Cyclo(6-amino-6-deoxyD-galactonic acid)4,: a new class of carbopeptoid-cyclodextrins, Tetrahedron Lett., 45, 153–156 (2004).

REFERENCES

203

[47] B. A. Mayes, L. Simon, D. J. Watkin, C. W. G. Ansell and G. W. J. Fleet, Huge (14-, 21-, 28-, 35-, 56- and 70-membered ring) macrocyclic lactams – A novel family of carbopeptoid-cyclodextrins, Tetrahedron Lett., 45, 157–162 (2004). [48] B. A. Mayes, A. R. Cowley, C. W. G. Ansell and G. W. J. Fleet, Mixed sugar-nylon 14-, 28-, and 42-membered ring mocrocyclic lactams, Tetrahedron Lett., 45, 163–166 (2004). [49] T. Q. Gregar and J. Gervay-Hague, Synthesis of oligomers derived from amide-linked neuraminic acid analogues, J. Org. Chem., 69, 1001–1009 (2004). [50] S. A. W. Gruner, V. Truffault, G. Voll, E. Locardi, M. Sto¨ckle and H. Kessler, Design, synthesis, and NMR structure of linear and cyclic oligomers containing novel furanoid sugar amino acids, Chem. Eur. J., 8, 4366–4376 (2002). [51] B. Aguilera, G. Siegal, H. S. Overkleeft, N. J. Meeuwenoord, F. P. J. T. Rutjes, J. C. M. van Hest, H. E. Schoemaker, G. A. van der Marel, J. H. van Boom and M. Overhand, Synthesis and structural analysis of a b-hairpin peptide containing a sugar amino acid, Eur. J. Org. Chem., 1541–1547 (2001). [52] S. A. W. Gruner, G. Ke´ri, R. Schwab, A. Venetianer and H. Kessler, Sugar amino acid containing somatostatin analogues that induce apoptosis in both drug-sensitive and multidrug-resistant tumor cells, Org. Lett., 3, 3723–3725 (2001). [53] E. G. von Roedern, E. Lohof, G. Hessler, M. Hoffmann and H. Kessler, Synthesis and conformational analysis of linear and cyclic peptides containing sugar amino acids, J. Am. Chem. Soc., 118, 10156–10167 (1996). [54] E. G. von Roedern and H. Kessler, A sugar amino acid as a novel peptidomimetic, Angew. Chem. Int. Ed. Engl., 33, 687–689 (1994). [55] T. K. Chakraborty, S. Jayaprakash, P. V. Diwan, R. Nagaraj, S. R. B. Jampani and A. C. Kunwar, Folded conformation in peptides containing furanoid sugar amino acids, J. Am. Chem. Soc., 120, 12962–12963 (1998). [56] T. K. Chakraborty, S. Ghosh, S. Jayaprakash, J. A. R. P. Sharma, V. Ravikanth, P. V. Diwan, R. Nagaraj and A. C. Kunwar, Synthesis and conformational studies by peptidomimetics containing furanoid sugar amino acids and a sugar diacid, J. Org. Chem., 65, 6441–6457 (2000). [57] B. Drouillat, B. Kellam, G. Dekany, M. S. Starr and I. Toth, Solid phase synthesis of C-terminal carbohydrate modified enkephalins, Bioorg. Med. Chem. Lett., 7, 2247–2250 (1997). [58] E. Lohof, E. Planker, C. Mang, F. Burkhart, M. A. Dechantsreiter, R. Haubner, H.-J. Wester, M. Schwaiger, G. Ho¨lzemann, S. L. Goodman and H. Kessler, Carbohydrate derivatives for use in drug design: Cyclic av-selective RGD-peptides, Angew. Chem. Int. Ed., 39, 2761–2764 (2000). [59] R. Haubner, H.-J. Wester, W. A. Weber, C. Mang, S. I. Ziegler, S. L. Goodman, R. Senekowitsch-Schmidtke, H. Kessler and M. Schwaiger, Noninvasive imaging of avb3 integrin expression using F18-labeled RGD-containing glycopeptide and positron emission tomography, Cancer Res., 61, 1781–1785 (2001). [60] R. Haubner, B. Kuhnast, C. Mang, W. A. Weber, H. Kessler, H.-J. Wester and M. Schwaiger, F18-Galacto-RGD: Synthesis, radiolabeling, metabolic stability, and radiation dose estimates, Bioconj. Chem., 15, 61–69 (2004). [61] R. M. van Well, H. S. Overkleeft, G. A. van der Marel, D. Bruss, G. Thibault, P. G. de Groot, J. H. van Boom and M. Overhand, Solid-phase synthesis of cyclic RGD-furanoid sugar amino acids peptides as integrin inhibitors, Bioorg. Med. Chem. Lett., 13, 331–334 (2003).

204

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

[62] A. Scaffidi, B. W. Skelton, R. V. Stick and A. H. White, The synthesis of carbohydrate a-amino acids utilizing the Corey–Link reaction, Aust. J. Chem., 57, 723–732 (2004). [63] A. Scaffidi, B. W. Skelton, R. V. Stick and A. H. White, The synthesis of some oligopeptides from novel carbohydrate a-amino acids, Aust. J. Chem., 57, 733–740 (2004). [64] J. Szejtli, Introduction and general overview of cyclodextrin Chemistry, Chem. Rev., 98, 1743–1753 (1998). [65] E. Engeldinger, D. Armspach and D. Matt, Capped cyclodextrins, Chem. Rev., 103, 4147–4173 (2003). [66] (a) H. Parrot-Lopez, H. Galons, A. W. Coleman, F. Djedani, N. Keller and B. Perly, Intramolecular host–guest complexes of D-mono-6-phenylalanylamino-6-deoxy and L-mono-6-phenylalanyl-amino-6-deoxy-cyclomalto-heptaoses, Tetrahedron Asymm., 1, 367–370 (1990); (b) H. Parrot–Lopez, F. Djedaini, B. Perly, A. W. Coleman, H. Galons and M. Miocque, An approach to vectorization of pharmacologically active molecules – The covalent binding of Leu-enkephalin to a modified b-cyclodextrin, Tetrahedron Lett., 31, 1999–2002 (1990); (c) H. ParrotLopez, H. Galons, S. Dupas, M. Miocque and G. Tsoucaris, Regioselective synthesis of 2,6-dimethyl-b-cyclodextrin coupled with R-phenylalanine and S-phenylalanine, Bull. Soc. Chim. Fr., 127, 568–571 (1990). [67] H. Galons, H. Parrot-Lopez, S. Dupas, B. Perly, A. W. Coleman, G. Tsoucaris and M. Miocque, Synthesis and structural study of amino acids and a peptide bound to permethyl b-cyclodextrin, In: D. Duchene, Minutes of the Fifth International Symposium on Cyclodextrins, Editions de Sante´, Paris, 1990, 134–137. [68] R. Breslow and S. Dong, Biomimetic reactions catalyzed by cyclodextrins and their derivatives, Chem. Rev., 98, 1997–2011 (1998). [69] N. M. Milovic, J. D. Badjic and N. M. Kostic, Conjugate of palladium(II) complex and b-cyclodextrin acts as a biomimetic peptidase, J. Am. Chem. Soc., 126, 696–697 (2004). [70] J. Danielsson, J. Jarvet, P. Damberg and A. Gra¨slund, Two-site binding of b-cyclodextrin to the Alzheimer Ab(1–40) peptide measured with combined PFG–NMR diffusion and induced chemical shifts, Biochemistry, 43, 6261–6269 (2004). [71] L.-Q Gu, M. D. Serra, J. B. Vincent, G. Vigh, S. Cheley, O. Braha and H. Bayley, Reversal of charge selectivity in transmembrane protein pores by using noncovalent molecular adapters, PNAS, 97, 3959–3964 (2000). [72] F. Djedaı¨ni-Pilard, J. De´salos and B. Perly, Synthesis of a new molecular carrier: N-(Leu-enkephalin)yl 6-amido-6-deoxy-cyclomaltoheptaose, Tetrahedron Lett., 34, 2457–60 (1993). [73] B. Di Blasio, V. Pavone, F. Nastri, C. Isernia, M. Saviano, C. Pedone, V. Cucinotta, G. Impellizzeri, E. Rizzarelli and G. Vecchio, Conformation for a b-cyclodextrin monosubstituted with a cyclic dipeptide, PNAS, 89, 7218–7221 (1992). [74] P. Berthault, D. Duchesne, H. Desvaux and B. Gilquin, A self-included cyclomaltoheptaose derivative studied by NMR-spectroscopy and molecular modeling, Carbohydr. Res., 276, 267–287 (1995). [75] M.K. Hristova-Kazmierski, P. Horan, P. Davis, H. I. Yamamura, T. Kramer, R. Horvath, W. M. Kazmierski, F. Porreca and V. J. Hruby, A new approach to enhance bioavailabiliy og biologically-active peptides – Conjugation of a d-opioid agonist to b-cyclodextrin, Bioorg. Med. Chem. Lett., 3, 831–834 (1993).

REFERENCES

205

[76] S. Hanessian, A. Benalil and C. Laferriere, The synthesis of functionalized cyclodextrins as scaffolds and templates for molecular diversity, catalysis, and inclusion phenomena, J. Org. Chem., 60, 4786–4797 (1995). [77] P. R. Ashton, R. Ko¨niger, J. F. Stoddart, D. Alker and V. D. Harding, Amino acid derivatives of b-cyclodextrin, J. Org. Chem., 61, 903–908 (1996). [78] K.S. A˚kerfeldt and W. F. DeGrado, Synthesis and per-functionalization of heptakis(6-O-carboxymethyl-2,3-di-O-methyl)cyclomaltoheptaose, Tetrahedron Lett., 35, 4489–4492 (1994). [79] N. Schaschke, S. Fiori, E. Weyher, C. Escrieut, D. Fourmy, G. Mu¨ller and L. Moroder, Cyclodextrin as carrier of peptide hormones. Conformational and biological properties of b-cyclodextrin/gastrin constructs, J. Am. Chem. Soc., 120, 7030–7038 (1998). [80] N. Schaschke, H.-J. Musiol, I. Asfalg-Machleidt, W. Machleidt, S. Rudolph-Bo¨hner and L. Moroder, Cyclodextrins as templates for the presentation of protease inhibitors, FEBS Lett., 391, 297–301 (1996). [81] N. Schaschke, G. Matschiner, F. Zettl, U. Marquardt, A. Bergner, W. Bode, C. P. Sommerhoff and L. Moroder, Bivalent inhibition of human b-tryptase, Chem. & Biol., 8, 313–327 (2001). [82] (a) H. Tsutsumi, K. Hamasaki, H. Mihara and A. Ueno, Rate enhancement and enantioselectivity in ester hydrolysis catalyzed by cyclodextrin-peptide hybrids, J. Chem. Soc. Perkin Trans. 2, 1813–1818 (2000); (b) H. Tsutsumi, K. Hamasaki, H. Mihara and A. Ueno, Cyclodextrin-peptide hybrid as a hydrolytic catalyst, Bioorg. Med. Chem. Lett., 10, 741–743 (2000). [83] M. A. Hossain, S. Matsumura, T. Kanai, K. Hamasaki, H. Mihara and A. Ueno, Association of a-helix peptides that have g-cyclodextrin and pyrene units in their side chain, and induction of dissociation of the association dimer by external stimulant molecules, J. Chem. Soc. Perkin Trans. 2, 1527–1533 (2000). [84] S. Matsumura, S. Sakamoto, A. Ueno and H. Mihara, Construction of a-helix peptides with b-cyclodextrin and dansyl units and their conformational and molecular sensing properties, Chem. Eur. J., 6, 1781–1788 (2000). [85] D. Yana, T. Shimizu, K. Hamasaki, H. Mihara and A. Ueno, Double naphthalenetagged cyclodextrin-peptide capable of exhibiting guest-induced naphthalene excimer fluorescence, Macromol. Rapid Comm., 23, 11–15 (2002). [86] M. A. Hossain, K. Takahashi, H. Mihara and A. Ueno, Molecule-responsive fluorescent Sensors of a-helix peptides bearing a-cyclodextrin, pyrene and nitrobenzene units in their side chains, J. Incl. Phenom. Macro., 43, 271–277 (2002). [87] N. Schaschke, H.-J. Musiol, I. Assfalg–Machleidt, W. Machleidt and L. Moroder, Oligopresentation of protease inhibitors with b-cyclodextrin as template, Bioorg. Med. Chem. Lett., 7, 2507–2512 (1997). [88] C. Pe´an, C. Cre´minon, A. Wijkhuisen, J. Grassi, P. Guenot, P. Je´han, J.-P. Dalbiez, B. Perly and F. Djedaı¨ni-Pilard, Synthesis and characterization of peptidyl-cyclodextrins dedicated to drug targeting, J. Chem. Soc. Perkin Trans. 2, 853–863 (2000). [89] A. J. Pearce and P. Sinay, Diisobutylaluminum-promoted regioselective de-O-benzylation of perbenzylated cyclodextrins: A powerful new strategy for the preparation of selectively modified cyclodextrisn, Angew. Chem. Int. Ed., 39, 3610–3612 (2000). [90] K. J. Jensen and J. Brask, Carbohydrates as templates for control of distance-geometry in de novo designed proteins, Cell. Mol. Life Sci., 59, 859–869 (2002).

206

CARBOHYDRATES IN PEPTIDE AND PROTEIN DESIGN

[91] M. Mutter and S. Vuilleumier, A chemical approach to protein design – Templateassembled synthetic proteins (TASP), Angew. Chem. Int. Ed. Engl., 28, 535–554 (1989). [92] J. Brask and K. J. Jensen, Carbopeptides: Chemoselective ligation of peptide aldehydes to an aminooxy-functionalized D-galactose template, J. Pept. Sci., 6, 290–299 (2000). [93] K. Rose, Facile synthesis of homogenous artificial proteins, J. Am. Chem. Soc., 116, 30–33 (1994). [94] R. Høiberg-Nielsen, A. P. Tofteng, K.K. Sørensen, P. W. Thulstrup, K. J. Jensen and L. Arleth, 3- vs. 4-helix formation in a de novo designed protein in solution revealed by small-angle X-ray scattering: Role of peptide sequence and template, ChemBioChem, 9, 2663–2672 (2008). [95] J. Brask and K. J. Jensen, Carboproteins: A 4-a-helix bundle protein model assembled on a D-galactopyranoside template, Bioorg. Med. Chem. Lett., 11, 697–700 (2001). [96] J. Brask, J. M. Dideriksen, J. Nielsen and K. J. Jensen, Monosaccharide templates for de novo designed 4-a-helix bundle proteins: Template effects in carboproteins, Org. Biomol. Chem., 1, 2247–2252 (2003). [97] J. Brask, H. Wackerbarth, K. J. Jensen, J. Zhang, I. Chorkendorff and J. Ulstrup, Monolayer assemblies of a de novo designed 4-a-helix bundle carboprotein and its sulfur anchor fragment on Au(111)-surfaces addressed by voltammetry and in situ scanning tunneling microscopy, J. Am. Chem. Soc., 125, 94–104 (2003).

6 De Novo Design of Proteins Knud J. Jensen

6.1

INTRODUCTION

The prediction of a protein’s 3D structure from its sequence alone remains a difficult problem. However, the inverse question, i.e. finding or designing a peptide sequence that will fold to a specific 3D structure, is more approachable. Although there might be numerous possible structures that can adopt this specified fold, only one solution, i.e. sequence, is needed. The focus of this chapter is on de novo design, i.e. the radical design of peptide sequences from general principles, rather than the use or redesign of a naturally occuring sequence. The biopharmaceutical relevance of de novo design has also emerged in the application of some de novo design principles in the drug discovery process. It is a continuing challenge for bioorganic chemistry and structural biology to design novel molecules that mimic the 3D structure and function of proteins. Protein de novo design offers the ultimate test of our understanding of the factors governing protein structure, folding and stability [1–3]. Not only can complex interactions in natural proteins be studied in greater detail using smaller de novo-designed systems, but the approach also offers the prospect of access to tailor-made proteins. DeGrado has suggested that there are different levels in protein design [2]. In its most radical form it is the design of a protein which folds into a welldefined 3D structure, with a sequence that is not directly derived from any

Peptide and Protein Design for Biopharmaceutical Applications © 2009 John Wiley & Sons, Ltd. ISBN: 978-0-470-31961-1

Edited by Knud J. Jensen

208

DE NOVO DESIGN OF PROTEINS

natural protein. In addition, there are a variety of strategies and methods for the redesign or reengineering of naturally occurring peptides and proteins. The ultimate goal of de novo protein design is the design from first principles of proteins which fold in a specified way and possibly with defined properties, e.g. enzymatic, without relying on a natural protein as a paradigm, i.e. without starting from a natural sequence. Thus the aim is to specify a certain 3D fold (a tertiary structure) constituted from specific secondary structure elements and appropriate loops, turns and templates. It is not redecoration of a natural protein but the use of general knowledge about protein folding and packing to construct new proteins. However, obviously there is no opposition here between ‘natural’ and ‘manmade’ and the same rules govern the folding of all molecules regardless of their heritage. De novo design has often taken the form of – in DeGrado’s words – a ‘minimalist’ approach, which involves design of minimalist sequences that are simpler than their natural counterparts but retain sufficient complexity for folding and function. The alternative, structure-based strategy – which is not de novo design in the strict sense – begins with an experimentally determined 3D structure of a protein. This has been useful for the design of proteins with enhanced stability or novel functions. However, the designed proteins have often behaved as molten globules, i.e. with non-native states, where most of the a-helices and b-sheets have formed but the tertiary structure is loose, showing dynamically averaging conformations, containing poorly packed hydrophobic cores behaving more like a liquid. Bryson and DeGrado have cautioned that ‘. . . between the conception of a designed protein and its realization lies the molten globule – an energy well of surprising depth and breadth that must be overcome en route to the final goal’ [4]. This chapter will describe de novo design of secondary structural elements and tertiary structures, with an emphasis on general applicability. The chapter is closely connected with Chapter 2 by Nikiforovich and Marshall on computational methods and Chapter 3 by Maes and Tourwe´ on peptidomimetics.

6.2 6.2.1

SECONDARY STRUCTURE ELEMENTS The a-helix

The most common secondary structural element is the a-helix (Figures 6.1 and 6.2). In the a-helix, the carbonyl of residue i is hydrogen

SECONDARY STRUCTURE ELEMENTS

209 g

g c

g

c

c

g

d

d

d

c

d

f f

f f a b

a

a

a b

e

e

b

b

e

e

A g c

e

g

e

c

d

d

a a

b b f

f f a b

a

d

g

e

b

f

d c

c

g

e

B

E E

K K c

g

L d L

L L d

K K g

E E

c

S S f

f S S a b

K K

e

E E

Y L

L Y

a e

b

K K

E E

C

E E

K K

E E

L

g c d

L

L Y

Y L

L L

e

K K b

a

S S f

f S S a b

K K

e

E E

d g

K K

c

E E

D

Figure 6.1 General helical wheel presentations of dimeric coiled coils: (a) parallel, (b) antiparallel. Helical wheel presentation of the sequence hypothetical YKELESKLKELESK depicted as a parallel, dimeric coiled coil with interhelical salt bridges: (c) parallel, (d) antiparallel. Created with DrawCoil 1.0

210

DE NOVO DESIGN OF PROTEINS

bonded to the amide NH of the i þ 4 residue. The a-helix is defined as 3.613-P, with a 3.6 pitch (number of residues per turn) and 13 atoms in the rings formed by hydrogen bonds between backbone CO and NH in i and i þ 4 residues, respectively. P indicates a right-handed helicity [5]. ˚ . Supercoiled a-helices (coiled The rise in the helix per residue is 1.5 A coils) have 3.5 residues per turn. The intrachain hydrogen bonds stabilize the helix, such that the carbonyls are pointing downward in the direction of the C-terminus, while the NHs point upward in the direction of the N-terminus. The helix forms a polyamide ‘cylinder’ with a strong dipole moment (a macrodipole), with the positive end at the N-terminus. There is a handedness to the helix, i.e. the screw sense of the helix. Looking down the helical axis from the N-terminal, generally, there is a clockwise sense in a right-handed helix. Another but less common helical secondary structure is the 310 helix. With three residues per turn, it has ten atoms in each repetitive hydrogenbonding structure, as the carbonyl of residue i is hydrogen bonded to the amide NH of the i þ 3 residue [6]. The propensity of amino acids to form a helix in a peptide segment is an important feature. Helix propensities accumulate over a peptide segment, with Ala being the most helix-stabilizing and Gly the least (it ‘breaks’ helices). The ability or potential of amino acids to form helices can to a large extent be understood in terms of conformational entropy and, maybe to a lesser extent, of solvent-accessibility of the side chain and main chain in the random coil vs. helical states [2,7]. Asn is nearly isosteric with Leu; however, Asn is able to H-bond in nonhelical conformations as well. Single a-helices can be stabilized by a number of methods (Figure 6.2) [4]: (i) Introduction of multiple Ala units. (ii) Addition of salt bridges between side chains of residues one a-helical turn apart [8–12]. (iii) Addition or engineering of new linkages connecting side chains one or two a-helical turns apart [13]. For example, Schultz and coworkers used disulfide formation between i and i þ 7 positions, bridging two turns, by introduction of S-2-amino-6mercaptohexanoic acid residues (i.e. (CH2)4SH side chain) ¨ sapay and Taylor introduced three lactam bridges in a [14]. O peptide with three heptad repeats at positions i and i þ 4 and observed a preference for Lysi, Aspi þ 4 combinations [15]. Verdine and coworkers have focused on so-called ‘hydrocarbon-

SECONDARY STRUCTURE ELEMENTS

(iv) (v)

(vi)

(vii) (viii)

211

stapling’ as a means to stabilize a-helical peptides to target protein–protein interactions, including the enforced helicity in a BAD BH3 that interacts with antiapoptotic BCL-XL [16,17]. The ‘stapling’ was achieved by incorporation of an a,a-disubstituted unnatural amino acid [18,19], an alkenyl Ala analogue with a terminal double bond, placed in i and i þ 4 or i þ 7 positions. The ‘staple’ was formed by ruthenium-catalysed olefin metathesis, providing the cyclic peptide. Addition of nonpeptide templates to initiate helix formation [20]. Introduction of charged residues near the amino or carboxyl termini to electrostatically stabilize the helix [12,21–23] due to the macrodipole, i.e. negatively-charged amino acid residues in the first turn (N-terminus) and positively-charged ones in the last turn (C-terminus) [24]. Introduction of ‘caps’ at the N- or C-termini to hydrogen bond to the helix ends, thus stabilizing the helix [24–30]. In an a-helix, the NH protons point toward the N-terminal while the CO carbonyls point toward the C-terminal end of the helix. This leaves ‘unsatisfied’ hydrogen-bond donors and acceptors at the two termini. At the N-terminus of the helix, the side chains of Asn, Ser, Asp or Thr residues can hydrogen bond to exposed amide protons: for example, from the Od to a backbone NH of residues i þ 3 (N3) or, less often, i þ 2 (N2) [24,25,31]. The residue preceding the N-cap is often apolar and can form a hydrophobic interaction with the side chain of the fourth helical residue, forming a hydrophobic stable [32]. Also, the so-called ‘SXXE box’ (a sequence of SerXxxXxxGlu), which provides hydrogen bonding between the N-cap amide and the side chain of Glu or Gln at an i þ 3 position [31]. C-terminal caps include Gly in the aL conformation (as in a left-handed a-helix, which is not allowed for L-amino acids; see Ramachandran plot), or a Pro residue two residues from the C-terminus of the helix [33,34]. Interestingly, Pro in position N1 (first residue after the ‘cap’) can be a helix-initiator. Addition of hydrogen bonds between side chains one turn apart to stabilize the helix [10,12]. Introduction of a hydrogen-bond surrogate by replacement of one of the main-chain intramolecular hydrogen bonds with a covalent linkage (for review, see [35]).

212

DE NOVO DESIGN OF PROTEINS N-terminal Idealized α-helix

C-terminal

Side-chain electrostatic interactions COO

N-cap Asn, Ser, Asp, Thr

OOC

Covalent ‘stable’ Side-chain macrodipole electrostatic interactions δ+

NH3

C-cap

H2N

NH2

Side-chain macrodipole electrostatic interactions δ−

Helix macrodipole due to alignment of amide bonds

Figure 6.2 Schematic presentation of some methods for stabilization of an a-helix. Redrawn after [4]

Often, a-helical peptides exhibit a hydrophilic and a hydrophobic side. These amphipatic a-helical peptide sequences have a heptad (seven residue) repeat pattern, designated abcdefg, each encompassing two turns in the helix. The study of these structures builds on pioneering work by Hodges and coworkers [36,37], who devised the prototypical sequence LeuaGlubAlacLeudGlueGlyfLysg to mimic the two-stranded coiled-coil conformation of tropomyosin. These and related structures are often referred to as Leu zippers because of the (interdigitating) Leu hydrophobic core. The a-helical structures that have been aimed at in de novo protein design have primarily been coiled coils and 3-, 4-, 5- and 6-helix bundles. In the coiled coil, the peptide helices are not covalently linked, while in helix bundles the peptide helices are typically linked by loops or turns [38]. The a-helices, typically two but up to seven, are wound around one another. Coiled coils show a ‘knobs-into-holes’ packing, first suggested by Crick [39], in which a hydrophobic side chain of one a-helix can contact four side chains from the second helix. Coiled coils are sometimes also referred to as super-secondary structures.

SECONDARY STRUCTURE ELEMENTS

213

In most 4-helix bundles, the a-helices pack against one another as ‘ridges-into-grooves’ [40]. The side chains in an a-helix can be viewed as being arranged in helical rows at the surface of the helix, thus forming ridges and grooves. The ridges and grooves are formed by residues whose separation in the sequence is usually four and sometimes three. The geometry of ridges and grooves depends not only on the helix but also on the actual sequence, i.e. its side chains. Thus, when a-helices pack against one another, the ridges from one helix are fitted into the groove of another. As a result, the helices that pack are inclined by an angle (often 50° or 20°) relative to one another. In artificial protein-like structures a nonlinear topology can be achieved with templates. The peptides can be packed either parallel or antiparallel and with varying helix crossing angles (Figure 6.3). A 4-helix bundle can also have a ‘bisecting U’ topology (Figure 6.4) [2].

N

N N

N

Helix 2’

C

Helix 1’

N

C C

Helix 2

Helix 1

C N

Helix1’

C

N

C

Helix2’

C

Helix 2

Helix 1

N

C

Figure 6.3 Schematic topology presentation of 4-a-helix bundles, parallel and antiparallel, respectively, by self-assembly of helix1–loop–helix2 sequences. Here righthanded turn loops are depicted, but left-handed turn loops are also possible (seen from the N-terminal helix, a left-handed bundle has the second helix in the bundle to the left, while a right-handed bundle has the second helix in the bundle to the right) [41]. When the ‘bisecting U’ topology is included, this gives rise to a total of six 4-helix bundle topologies

1’ Helix

N

2’ Helix

C

N

C

Figure 6.4 Schematic topology presentation of the ‘bisecting U’ topology of helix1– loop–helix2 sequences [2]

214

6.2.2

DE NOVO DESIGN OF PROTEINS

The b-sheet

The b-sheet is the second most common secondary structure in proteins, consisting of an assembly of b-strands connected by at least three interstrand hydrogen bonds. These interstrand hydrogen bonds occur between the carbonyl and amide NHs of every other residue in the backbone. The b-strands can be connected in either a parallel or an antiparallel manner, with different geometries; the parallel orientation is less stable due to nonplanarity. b-Sheets are composed of b-strands in an extended conformation with a slight twist and with interchain hydrogen bonds. The antiparallel arrangement provides a stronger interstrand stability than the corresponding parallel. For the parallel arrangement, it is rare to find fewer than five interacting strands in a motif. b-Sheets can participate in several different structural motifs, including b-hairpin, Greek key motif, b-a-b motif and a/b protein folds, such as in a TIM barrel with eight ahelices and eight parallel b-strands, where the latter form a b-barrel. a-Helical structures have been pursued in de novo design to a significantly larger extent than b-sheet-containing structures. Helix requirements are well understood and the hydrogen bonds in helices are intrachain, and thus the sequences are less likely to form insoluble aggregates, unless they also have a propensity for b-sheet formation. However, some b-sheet-containing assemblies are described in Section 6.3.5.

6.2.3

Loops, Turns and Templates

In the design of proteins, it is necessary to consider how the secondary structure elements are connected. In proteins with linear sequences, turns and loops connect the secondary structural elements, whereas some artificial proteins have a central scaffold to which the peptide strands are fixed (template-assembled synthetic proteins, TASPs). Turns and turn mimetics are briefly introduced in Chapter 3 by Maes and Tourwe´. Thus, the discussion here will be limited to a few notes. DeGrado and coworkers have been successful in engineering turns in their de novo-designed proteins, notably the GPRRG sequence, which also directs antiparallel dimerization by charge repulsion. Helix nucleators are also described by Maes and Tourwe´ in Chapter 3. b-Sheets can be stabilized by incorporation of artificial structures, especially b-turn mimics (for a review, see [42]). b-Hairpins (in which antiparallel strands are connected with a reverse turn) can be assembled with b-turn mimics. Here the reverse turns induce b-sheet structure by

ASSEMBLING A SPECIFIED TERTIARY STRUCTURE

215

aligning the peptide sequences, and b-turn mimics have been used to nucleate b-sheet folding in b-hairpins. Succesful b-turn structures have included Aib-Xxx (Xxx: D-Ala, D-Val, D-Pro or Gly) [43,44]; some form turn conformations in aqueous solutions [45]. Other turn mimics have included Gellman’s D-Pro-Gly [46].

6.3

6.3.1

ASSEMBLING A SPECIFIED TERTIARY STRUCTURE FROM SECONDARY STRUCTURAL ELEMENTS Computational Methods

Chapter 2, written by Nikiforovich and Marshall, provides an excellent overview of computational methods for peptide folding. In the present section, just a few specialized approaches are summarized, as the focus of this chapter is the general rules of de novo design and their applicability. For additional reviews on computational methods for design of proteins, see [47,48]. The Rosetta program, developed by Baker and coworkers, aims at prediction and design of protein structures, primarily for the de novo prediction of structures for structural genomics [49,50]. However, the program RosettaDesign, which iterates between sequence optimization and structure prediction to identify low-energy sequences for specified protein backbones, has been used for de novo design of a novel globular protein fold (vide infra) [51]. Importantly, this program is accessible to the scientific community. An important aspect of de novo protein design has been hydrophobic side-chain repacking algorithms, using the program ROC [52], for example. Algorithms which have successfully been applied to the de novo design of proteins have utilized sequence search methods such as simulated annealing, genetic algorithms and dead-end elimination (elimination of rotamers that cannot be members of the minimum energy sequence). Other promising methods have involved statistical or probabilistic design. Also, principles of negative design have proven important in distinguishing native and non-native structures, i.e. energetically separating the folded state from competing structures. The program MetalSearch was used to build a His3Cys into a protein domain to bind Zn2þ [53]. The computational design algorithm SCADS (statistical computationally assisted design strategy) [54] was used in the

216

DE NOVO DESIGN OF PROTEINS

design of a 4-helix bundle which binds a hemelike cofactor, as described below. Dahiyat and Mayo have reported a design algorithm which takes the coordinates of a specific backbone fold – from a crystal structure, for example – as input and then predicts as the output a peptide sequence that will stabilize this fold ([55]; for a commentary, see [56]). They include three categories of interactions: (i) side chains that are exposed to solvent; (ii) side chains that are buried in the protein interior; and (iii) residues with an interfacial position. They include a dead-end elimination theorem in their program. The use of this program was demonstrated by the radical redesign of a zinc finger, which has a bba fold. They then searched for a sequence that would stabilize this fold in the absence of metal ions. They identified a 28-mer sequence, which they synthesized and which folded according to NMR studies essentially as predicted.

6.3.2

Coiled Coils

Maybe the most simple tertiary structure – or structural motif – in proteins is the coiled coil [39], in which two to seven a-helices are coiled together like the strands of a rope [57,58]. Most common are the 2- and 3-helix coiled coils, i.e. dimers and trimers. The a-helices are arranged either parallel or antiparallel; usually in a left-handed super coil [2]. However, some de novo-designed coiled coils [59] have adopted right-handed coiled coils. The helices are not covalently linked by loops or turns. The amino acid sequences in coiled coils usually contain heptad repeats, which provide an amphipathic pattern in which the hydrophobic residues are typically leucine (primarily at d positions), which can form a Leu zipper. They assemble into coiled coils by ‘hydrophobic collapse’, in which burial of the hydrophobic side chains in water is the major driving force. The GCN4 leucine zipper (PDB 1zik), which is a parallel, lefthanded dimer, is a classic example of a coiled coil. In pioneering work, Hodges [60–62] and coworkers studied idealized two-stranded coiled coils, which contain heptad repeats with positions designated as a–g, where hydrophobic residues a and d are directed toward the interior of the structure: LeuaGlubAlacLeudGlueGlyfLysg and multimers thereof, where Glu and Lys at positions e and g, respectively, provide opposite charges which can further stabilize the structure by salt-bridge formation when the helices pack together (Figure 6.1). Both antiparallel homodimers and parallel heterodimers have been

ASSEMBLING A SPECIFIED TERTIARY STRUCTURE

217

designed and synthesized through modifications of the charges on residues e and g. In an important early paper, Harbury et al. mutated the buried a and d positions of the Leu zipper sequence GCN4 [63]. The mutants were designated p-IL, p-II and p-LI, depending on whether they had Ile or Leu in positions a and d, respectively (e.g. p-LI had Leu in all a positions and Ile in all d positions). p-IL, p-II and p-LI appeared to be dimeric, trimeric and tetrameric, respectively. Thus, the differences conserved the volume of the hydrophobic side chain and it was the packing interactions which determined the switch between dimer, trimer and tetramer. One conclusion from their study was that b-branched residues in d positions disfavour dimers, while b-branched residues in a positions disfavour tetramers, and b-branched residues in a and d positions should favour trimer formation. The p-LI tetramer had four parallel a-helices organized in a left-handed superhelix forming a cylinder that was 27 A˚ wide and 48 A˚ long. The cylinder had a cavity in the middle of each Leu and Ile layer, which formed a continuous central channel. The radius of the channel varied from 1.0 to 1.3 A˚, which was too small for water, which would have required a 1.4 A˚ radius. A determining factor in the side-chain packing was the occurrence of three different knobsinto-holes arrangements. As expected, more surface was buried in the tetramers (1640 A˚2 per helix) than in the dimer (900 A˚2). Residues at the b, c, e and g positions are more buried in the tetramer structure than in the dimer; thus residues at these positions can influence the oligomerization state. DeGrado and coworkers summarized that peptides with Leu in position a and d form dimers, trimers, tetramers, pentamers or hexamers, depending on the hydrophobicity and steric properties of the residues e and g. Thus, while hydrophobicity can drive the formation of secondary structure and self-assembly, the oligomeric state and specific topology depend on side-chain packing, hydrogen bonding and ‘negative design’ to selectively destabilize alternative folds [4]. O’Neil and DeGrado designed a 29-mer peptide (Ac-EWEALEKKLAALE-X-KLOALEKKLEALEHG) to study the helix propensity of amino acids [64]. Interestingly, when the peptide was crystallized it revealed an antiparallel triple-helix structure, rather than the expected dimeric (i.e. two-helix) structure [65]. They demonstrated that the peptide in solution was in a concentration-dependent monomer–dimer–trimer equilibrium and that the self-assembly was noncooperative [66].

218

DE NOVO DESIGN OF PROTEINS

The association of peptides into trimeric coiled coil seems in general to be determined by hydrophobic residues in positions a and d of the heptad repeat, apart from specific effects such as hydrogen-bonding effects [38]. The a and d residues in the core pack as ‘knobs-into-holes’ [39], where the ‘hole’ is in between two residues. Glu, Arg and Lys are often found in e and g positions. Also, hydrophobic interactions between side chains of b and e are more important for stabilizing a trimer than a dimeric assembly. While two-stranded structures only have substantially apolar residues at positions a and d (with the remaining being highly polar), three-stranded coils show decreasing polarity in other residues in moving away from the central axis (and toward the solvent-exposed surface). Although the substitution of a hydrophobic residue in position a or d by a hydrophilic residue is likely to destabilize the protein, the resulting buried polar interaction may favour one aggregational state over another (i.e. by negative design) and thus contribute to a more native-like behaviour. In coiled coils with three parallel strands, all the a residues form one layer, while the d residues form another; the two types of layers are not geometrically equivalent. In the antiparallel structure there are alternating layers of either two a and one d or one a and two d. While the parallel orientation provides three identical helix–helix interfaces with interactions between residues e and g, the antiparallel structures have geometrically distinct interfaces, providing g–g, g–e and e–e interactions (Figure 6.5). Interestingly, the coil-Ser protein from DeGrado and coworkers, which has an all-Leu core with Leu in positions a and d, forms an antiparallel trimer [64,65]. The extensively studied ‘coil-Ser’ sequence by DeGrado and coworkers incorporated Trp-2 and His-28 to facilitate NMR and UV spectroscopy. It had Leu in positions a and d (except Trp-2) and Glu and Lys at positions e and g. It formed a trimer both in solution, as observed by ultracentrifugation, and in the crystal, as studied by X-ray crystallography. All helices had eight helical turns and the trimer formed a hydrophobic core with eight hydrophobic layers. The crystal structure revealed an antiparallel triple-stranded a-helical bundle 44 A˚ in length and 18 A˚ in diameter. Helix I and II were ‘up’ while helix III was ‘down’ (up-up-down topology; like an ‘N’). The crossing angle between all three helix pairs was about þ20° and the three helices wrapped around the superhelical axis to form one sixth of a turn of a left-handed super coil. The distance between the axes of the helices was in the range 11.1–12.6 A˚. In solution, it exhibited a monomer–dimer–trimer equilibrium.

ASSEMBLING A SPECIFIED TERTIARY STRUCTURE

219

K

K K E b

L

G f

L e

G

c E E

K E

E g

E c

L

d

K

L

L

a

g

L

d L L

L

L L a

E K G

K G f a b

e L L

E K

L

L L

L

L

e

d

b E K

E g K

f

c E

G

G K

E

A

K

K

K K E b

L

G f

L e

G

c E E

K E E c

E g

L

d

K

L

L

a

g

L

d L L

L

L L d

E K G

K G f a b E K

e L L

L

L L

g

a

c E E

L e L

B

G K E

b K E

f K K G

Figure 6.5 Helical wheel presentation of an amphipathic sequence in a triple-helical arrangement (a1B, GELaEELLKKLKELLKG). The parallel topology (a) has three identical e–g interactions, wheras the antiparallel topology (b) has three different interfaces, e–e, e–g and g–g. Created with DrawCoil 1.0

220

DE NOVO DESIGN OF PROTEINS

There are only relatively few examples of parallel triple helical structures. However, the de novo sequence CoilVaLd, b-branched and with Val in position a forms a parallel trimeric coiled coil (VEALEKKVAALESK-VQALEKKVEALEHG, 28AA, helix–helix design, VaLd) [67,68]. In solution, a monomer–trimer equilibrium was observed.

6.3.3

a-helical Bundles

Among the most frequently used structural motifs for de novo design is the 4-a-helix bundle. Its folding is driven by the hydrophobic collapse of amphiphilic helices and is guided by the position and nature of loop regions, by electrostatic effects and by shape complementarity in sidechain packing (Figure 6.6). Several artificial, de novo-designed 3-, 4- and 5-helix structures have been reported (for a review, see [69]). They rely on a-helical, amphipathic peptide sequences either (i) by linear connection in a helix–loop–helix pattern, or (ii) by assembled on a template to provide artificial TASP structures. Here we focus on 3- vs. 4-helix structures. It is important to restrict the side-chain mobility of the hydrophobic core, i.e. to avoid a dynamic or fluid character, which can be achieved by incorporation of b-branched and aromatic residues in the core [70]. DeGrado and coworkers have since the mid 1980s been on a long quest toward de novo design of proteins, especially 4-helix bundles, which has culminated in the design of proteins and metalloproteins with native-like properties [2,3]. They started out with single-helix sequences and then moved on to longer helix1–loop–helix2 sequences (see Table 6.1). It has been concluded that helices of four or more turns pack preferably in elongated bundles, while shorter helices can pack into a number of other geometries [4,71]. Also, it has been concluded that conformational specificity requires a correct balance of hydrophobic and hydrophilic interactions, as too few hydrophobic residues leads to inadequate stability, while too many leads to highly stable but dynamic structures [4,72].

6.3.3.1 3-helix bundles DeGrado and coworkers have concluded that the way helices interact is less regular in a 3-helix bundle than in a three-stranded coiled coil [38]. Starting from coil-Ser, DeGrado and coworkers developed the uniquely folded 73residue 3-helix bundle proteins a3C and a3D [73,74]. The

ASSEMBLING A SPECIFIED TERTIARY STRUCTURE

221

K K f

E E L K g

E c

c

b

L e A L

K

E

E

L d L

E K

g K

L a Y

d L L

E

A L e L Y

E b

a f K K

K K f a b

Y

L e LA

E E

L d L

Y a L

L L d

L

K g

E

c

b f K

E

E E

K

A e L

K

c

g K

L

E E

K

A

K K

E E K

a

d

L

L

L L

L

E

A L

A

L d

K

L

e L

Y g

c

E

b

L

K E

c

K g

K E

E

f

E

e

b

Y a

K f

f K a b

e

E

Y

E

L

L

L

L A

A L

L

L d

a

E E K

g K

L

K E E

B

c

g

K

L

Y L e

K

d

b

c f

K

E E

K

Figure 6.6 Schematic helical wheel presentation of a 4-helix bundle: (a) parallel, (b) antiparallel. Created with DrawCoil 1.0

by formation of five disulfide bridges

helix–loop–helix

a2D

1

helix–loop–helix

Ac-ELLEKL- HCys(SR)-K- HCys(SR)-LEELLKK-NH2, R: Acm, SCH3, H

5-helix bundle1

a2C

R MKQLEDK VEELLSK NYHLENE VARLKKL VGER g abcdefg abcdefg abcdefg abcdefg abcdef

GCN4 Leu-zipper motif

GCN4-p1

Ac-GEVEELEKKFKELWK-GPRRG-EIEELHKKFHELIKGNH2

Ac-GEVEELLKKFKELWK-GPRRG-EIEELFKKFKELIKGNH2

Ac-GELEELLKKLKELLK-GPRRG-ELEELLKKLKELLKGNH2

Ac-GKLEELLKKLLEELKG-OH Ac-GELEELLKKLKELLKG-OH

a1A a1B

helix–loop–helix

E VEALEKK VAALESK CQALEKK VEALEHG

CoilVaLd V16C

a2B

E VEALEKK VAALESK VQALEKK VEALEHG

CoilVaLd

G LKALEEK LKALEEK LKALEEK LKALEEK G

TRI

triple helix

Ac- E WEALEKK LAALESK LQALEKK LEALEHG -NH2

Coil-Ser

Sequence K LEALEGK LEALEGK LEALEGK LEALEGK LEALEG g abcdefg abcdefg abcdefg abcdefg abcdef

Topology, comments

Coiled coil

Name

DeGrado

DeGrado

DeGrado

Chmielewski [88]

DeGrado, Eisenberg

DeGrado [80] DeGrado

Pecoraro

DeGrado

Pecoraro (1998)

DeGrado [64]

Hodges [36]

Reference

Table 6.1 List of some a-helical sequences used in the de novo design of coiled coils and helix bundles; small letters indicate heptad repeat pattern

ASSEMBLING A SPECIFIED TERTIARY STRUCTURE

223

computer program ROC was utilized in this design. The means to achieving the design were: (i) shortening helix sequence length (to 20 residues, which is the typical helical length in 3-helix bundles and thus shorter than in coiled coils); (ii) incorporating loops with N-cap motifs; and (iii) incorporating charges to electrostatically favour desired helix–helix pairings and disfavour others. Interfacial Glu, Arg and Lys side chains were important. A counterclockwise topology was observed by NMR. However, NMR relaxation studies revealed a higher degree of fluidity in the core of a3D than that observed in natural proteins of similar size [75]. Furthermore, a helix-binding 3-helix structure was developed by DeGrado, Lombardi and coworkers [76]. They also developed a 3-helix structure in which the helices were linked by disulfide bridges [77]. Pecoraro and coworkers have developed a series of triple-helical structures, self-assembled from Cys-containing homotrimers, which together bind Hg(II) [78,79]. The starting point provided the TRI sequence, which does not contain Cys and which forms a trimer. They also mutated CoilVaLd (see above) into CoilVaLd–V16C. While TRI can be both parallel and antiparallel, CoilVaLd was designed to be parallel.

6.3.3.2 4-helix bundles The purposely simplistic a1A and a1B sequences (Table 6.1) by Ho and DeGrado were reported to form tetrameric assemblies as 4-helix bundles according to size-exclusion chromatography (and concentration-dependent CD) [80]. Starting from these single-helix peptides, which selfassemble weakly to tetramers, Ho and DeGrado designed helix–loop– helix sequences that dimerize to 4-helix bundle assemblies. These constructs, a1B-Pro-a1B and a1B-Pro-Arg-Arg-a1B, form relatively stable dimers in solution. The latter contained Pro-Arg-Arg as a loop, which was designed to favour antiparallel dimerization by repulsion of electrostatic charges. This loop was eventually developed further to the useful sequence GPRRG. In helix–loop–helix sequences that were designed to dimerize to form 4-helix bundles, DeGrado and coworkers had previously observed a dynamic character of the de novo-designed protein a2B. a2B had only Leu, Glu and Lys in the sequence and an all-Leu core; its dynamic character was most likely due to rapid motions of side chains in the bundle and the formation of multiple, interconverting topologies [2,3]. They designed several 35 AA helix–loop–helix sequences (N-terminus acetylated, C-terminus amidated) intended to dimerize. The GPRRG loop causes

224

DE NOVO DESIGN OF PROTEINS

charge repulsion if the dimerization is parallel. a2B was stable but appeared to have adopted a molten globule-like conformation. Starting from a2B, DeGrado and coworkers first designed a2C, in which half of the Leu residues were substituted with aromatic and b-branched apolar side chains [81]. However, only after substituting three of the interfacial apolar residues for polar ones, providing the sequence a2D, did dimerization of this sequence lead to a native-like protein structure [82]. Actually, these mutations were originally intended to introduce a metal-ion binding site into a2C. However, dimeric a2D also showed native-like character even in the absence of metal ions, i.e. a2D had characteristics of folded proteins. Remarkably, a2D had a ‘bisecting U’ folding topology, with diagonal crossover loops and two parallel and two antiparallel helices [83]. Dimeric a2D was thermodynamically less stable than a2B, which indicated that the native-like character of a2D was obtained by negative design, which destabilized alternatively-folded states. In negative design, a structural feature is built into the sequence that will disfavour certain folds, rather than directly favouring a topology (as for example the introduction of a salt bridge might). Schafmeister, Stroud and coworkers have expressed a de novodesigned protein of 108 AA which folds as 4-helix bundles, as determined by X-ray crystallography [84]. The protein, termed DHP1, was based on the amphipathic, a-helical 24 AA sequence PD1 [85]. Betz and DeGrado, and then Marsh and coworkers, have concluded that antiparallel and parallel bundles differ in that the former are more conformationally stable toward changes in the hydrophobic core, as subtle changes in sequence of the latter or in buffer conditions can readily change the oligomerization state between dimer, trimer and tetramer [86,87]. Baltzer and coworkers have in a series of papers described the design and utilization of a 42-residue helix–loop–helix peptide designed to dimerize [88–90]. For example, they reported the site-selective chemical acylation of Lys and Orn residues. They were able to incorporate an amino acid carrying a triaza-cyclononane that allowed binding of Zn(II), which showed some transesterification activity.

6.3.3.3 a-helical metalloproteins Once some of the first rules for de novo design of protein-like structures with a-helices were established, work began to incorporate metal-ion binding sites into de novo-designed proteins. The efforts by Pecoraro and coworkers to this end have already been discussed above. Haehnel

ASSEMBLING A SPECIFIED TERTIARY STRUCTURE

225

and coworkers’ use of a TASP strategy for design of metalloproteins is described below (Section 6.4). The position of the ligands that will coordinate the metal ion is crucial; however, the positioning of additional functional groups close to the coordination site may also be important. Coordination of Zn2þ typically requires three His imidazole ligands or a Cys2His2 binding site. Pecoraro and coworkers have focused on Cys-containing sequences that bind mercury. Proteins have been engineered which fold to provide a heme binding site, thus creating artificial heme proteins [80,91,92]. DeGrado and coworkers designed a 4-helix bundle (four copies of the peptide in Table 6.2, entry 1) which bound two DPP-FE heme-like cofactor by complexation to His moieties [93]. Also, iron-sulfur (Fe4S4) clusters have been engineered into proteins [94–96]. Table 6.2 List of some a-helical sequences (helix1–turn–helix2 topology) used in the de novo design of metal-ion binding helix bundles [101,102] Entry

Name

1

Sequence SLEEALQEAQQTAQEAQQALQKGQQAFQKFQKYG

2

DF1

DYLRELLKLELQAIKQYREALEYV KLPVLAKILEDEEKHIEWLETING

3

DF2

MDYLRELYKLEQQAMKLYREASEYV GD PVLAKILEDEEKHIEWLETING

4

DF2t

MDYLRELYKLEQQAMKLYREASE KARN PEKKSVLAKILEDEEKHIEWLETING

Metal-ion binding sites have been incorporated into the de novo protein a4. A His3 binding site was engineered into a4 [97,98], as was a Cys2His2 binding site [99]. Lombardi, DeGrado and coworkers have designed a series of di-iron (III) proteins formed from antiparallel dimers of helix1–turn–helix2-type sequences (Table 6.2, entries 2–4) [100,101]. De novo-designed metalloproteins based on b-sheet secondary structures were achieved in a rubredoxin mimic, which bound Fe(II/III) in a Cys4 site formed by dimerization [102].

6.3.4

Fluorous Interactions

Proteins containing fluorous side chains were first used in 19F-NMR studies [103]. However, in a visionary paper, Marsh described the

226

DE NOVO DESIGN OF PROTEINS

potential of using fluorous interactions in the design of proteins [104]. The van der Waals radius of fluorine is 1.35 A˚, only 0.15 A˚ larger than hydrogen; however, the C-F bond is significantly longer (1.4 A˚) than the C-H bond (1.0 A˚) and fluorine is better considered an isostere of oxygen. Obviously, a C-F bond has the opposite polarity of a corresponding C-H bond and introduction of fluorine induces polarization of C-H bonds in the proximity and has an effect on the pKa several bonds away. Fluorocarbon molecules appear much more hydrophobic than the analogous hydrocarbon molecules. Interestingly, although fluorocarbons are hydrophobic, perfluorinated molecules tend also to be poorly soluble in hydrocarbon solvents. Hexane, water and perfluorohexane are mutually immiscible and separate into three layers after mixing. Perfluorocarbons are better described as fluorophilic, rather than lipophilic or hydrophobic. However, the cohesive dispersion forces between hydrocarbon molecules are larger than those between fluorocarbon molecules, as hydrocarbons are more polarizable, which excludes fluorocarbons from hydrocarbons, and the ‘fluorous effect’ is thus at least to some extent caused by this exclusion from hydrocarbons. This fluorous effect is responsible for the nonstick property of Teflon, as it interacts with neither hydrophilic nor lipophilic compounds. The use of fluorous interactions in the de novo design of proteins has focused on the interior of proteins, especially fluorinated substitutes for Leu and Val. Ideally, fluorous–fluorous interactions in proteins would be orthogonal to other interactions. Tirrel and coworkers have prepared a fluorinated coiled-coil ensemble [105]. In this Leu zipper, Leu in the d positions of the heptad was substituted by trifluoroleucine (Tfl, racemic mixture). The sequence was expressed in vivo, but not with 100% incorporation. It formed a stable dimer, but at high-mM concentrations it associated to higher-order species. It proved highly resistant to denaturation. Kumar and coworkers [106–108] designed a 30-residue coiled-coil sequence with Lys at e positions and Glu at g positions, which due to unfavourable interhelical electrostatic in the antiparallel arrangements favoured the parallel alignment. A single Asn was placed in the core such that it could only hydrogen bond in the parallel arrangement, and the sequence was N-terminally extended with Cys-Gly-Gly to allow interstrand disulfide formation. In addition to Hb with an all-Leu core, Kumar and coworkers also prepared Fb with hexafluoro-leucine (hFLeu) in the core. In a key experiment, the disulfide-linked heterodimer HbFb was allowed to ‘self-sort’ under redox conditions. Homodimerization into

ASSEMBLING A SPECIFIED TERTIARY STRUCTURE

227

HbHb and FbFb was strongly favoured over heterodimerization, which indicated a potential for orthogonality between fluorous and hydrophobic interactions in protein design. Marsh and coworkers designed a 27-residue peptide intended to form a 4-helix bundle-like structure with an antiparallel topology (Ac-GN ADELYKE LEDLQER LRKLRKK LRSG NH2, a4-H) [86,109]. Stepwise substitution of Leu for hFLeu provided a4-F2, a4-F4 and a4-F6, in which 2, 4 and 6 layers, respectively, of the hydrophobic core were repacked with hFLeu. CD spectroscopy revealed for all fluoroproteins an extensively a-helical secondary structure. At high concentrations (up to 500 mM) all structures occurred predominantly as the tetramers, while at lower concentrations monomer–tetramer equilibria were observed. Apparently, the bigger hFLeu side chains could be accommodated in the core. The stability of the protein structures increased with increasing numbers of fluorinated side chains in the core. In addition, the core of a4-F6 appeared to be less conformationally dynamic than that of a4-H. Interestingly, a4-F6 proved more stable to proteolysis than a4-H. Surprisingly, TFE did not preferentially dissociate a4-F6. In contrast to the observations by Kumar and coworkers, Marsh and coworkers reported that they did observe indications for interactions between a4-F6 and a4-H, thus did not observe self-segregation in this protein-like assembly. They speculated that in the Kumar peptide, not only fluorous interactions but also steric effects from hFLeu, which is larger than Leu, played a role. Koksch and coworkers [110] used as a model system a 41-mer, RLEEL REKLE SLRKK LACLK YELRK LEYEL KKLEY ELSSL E, which could form an a-helical coiled coil as an antiparallel homodimer. In this sequence, positions a and d had alkyl side chains, while e and g had charged residues capable of forming interhelical salt bridges. Koksch and coworkers studied the effect of mutating Lys8 and Leu9 with fluorinated analogues. They employed non-natural building blocks with increasing degrees of fluorination (S)-ethylglycine (EGly), (S)-4,4difluoroethylglycine (DfeGly), (S)-4,4,4-trifluoroethylglycine (TfeGly), (S)-4,4-difluoropropylglycine (DfpGly). Assuming that the trifluoromethyl moiety has approximately the size of an isopropyl group, Leu would have space-filling properties as (S)-4,4,4-trifluoroethylglycine (TfeGly). They evaluated the effects of introduction of fluorinated analogues by two different methods: first thermodynamic stability and second the ability to ‘self-replicate’ by the reaction of two peptide segments through native chemical ligation to yield the full sequence. Substitutions

228

DE NOVO DESIGN OF PROTEINS

in the hydrophobic core destabilized the a-helical coiled coil. They also found that electrostatic consequences of alkyl-fluorination can have a stronger effect on hydrophobic interactions in a protein than the increase in molecular volume. They observed self-sorting behaviour of their a-helical coiled coil, i.e. that fluorine in a native peptide/protein environment prefers to interact with fluorine.

6.3.5

Additional Topics

Above, selected topics in the wide-ranging field of de novo design of proteins have been described. However, a few additional approaches need to be briefly mentioned. Garcia-Echeverria [111] described the unequivocal determination by fluorescence spectroscopy of the formation of a parallel Leu-zipper homodimer. He synthesized two related peptides, only differing in the presence of a pyrene label on one. The pyrene-labelled Leu-zipper peptide shows excimer fluorescence (max. at 480 nm) when the polypeptide selfassociates to form a parallel homodimer, while monomer fluorescence (max. at 380, 400 nm) would be observed if the peptide adopted a singlestranded conformation. Redesign of a protein is in general not covered in this chapter on de novo design. However, in the redesign of ROP, a natural 4-helix bundle, Regan and coworkers concluded that amino acids with small hydrophobic side chains (Ala) in position a and amino acids with a larger hydrophobic side chain in position d give a stable protein with native-like character [112]. Interestingly, an a-helical sequence containing Cys residues designed to form a 4-helix struxture by disulfide formation actually formed a disulfide-linked 5-helix assembly Table 6.1 [88]. There are only a few examples of the use of b-sheet structures in de novo design. One is the 50-residue BetaCore, designed and chemically synthesized by Carulla, Woodward and Barany [113]. The sequence was originally derived from bovine pancreatic trypsin inhibitor (BPTI) and the design is thus not strictly de novo, by the definition used in this chapter. BetaCore consists of two units of b-sheet structure, each cyclized by a disulfide bridge and further stabilized by an intrachain oxime linkage. BetaCore forms a water-soluble, four-stranded, antiparallel b-sheet protein at pH 3 and lower temperatures. Struthers, Cheng and Imperiali iteratively developed a 23-mer peptide which folded into a bba motif similar to a zinc finger but

PROTEINS ON TEMPLATES

229

without metal-ion complexation [114]. The folded peptide incorporated a type II0 b-turn to assist in the formation of the b-hairpin structure. Baker and coworkers used their program, RosettaDesign, to design a 93-mer a/b protein, Top7, with a novel sequence and topology [51]. It was characterized by a range of biophysical techniques, which revealed that it was monomeric and showed native-like behaviour. Structure determination by X-ray crystallography revealed that it adopted the intended fold. This important study thus combined computer design using a program that is now generally accessible with full structural characterization of the de novo protein.

6.4

PROTEINS ON TEMPLATES

Helix 4 Helix 3

Helix 2

Helix 1

This section describes steps toward a class of radically de novo-designed proteins that are assembled on templates or scaffolds, which gives them an artificial topology for the connection of secondary structural elements (Figure 6.7). Mutter and coworkers suggested the term ‘template-assembled synthetic proteins’ (TASPs) to describe structures in which peptide strands are anchored to regioselectively addressable templates [115–117]. The secondary structural elements – so far mainly a-helices have been used – are anchored to the template or scaffold to preorganize them, typically in a parallel or antiparallel orientation. Underlying this, there have been two basic but opposite assumptions: either that the template to some extent controls or directs the distance

Figure 6.7 Schematic representation of the TASP concept, here depicted with four helices (peptide sequences in Table 6.3)

230

DE NOVO DESIGN OF PROTEINS

geometry of the peptide-attachment points, or that the template merely controls the oligomeric state of the assembly (overall topology), for example that four and five helical strands on the template would give 4-helix vs. 5-helix bundles, respectively. While this template strategy for protein design has led to very fascinating science, it is notable that no X-ray crystal or NMR structures of TASPs have been reported so far, which has limited the development of this field. The lack of NMR structures could in part be due to the often degenerated structures (i.e. multiple copies of the same peptide sequence) in these proteins. However, very recently, small-angle Xray scattering (SAXS) has provided novel information on the topology of so-called carboproteins, vide infra. Although these are ‘low-resolution’ structures that do not provide atomic resolution, they do provide key information on the overall fold of the proteins, and hence their topology. Table 6.3 Some amphiphilic peptide sequences used in template-based proteins Entry 1 2 3 4 5 6 7 8 1

Sequence

Reference

LEALEKALKEALAKLG LKALKEAFEKAMAELG EELLKKLEELLKKG -CEKLLKELKELLEKG-NH2 CGGGEELLKKXEELLKKG1 GEELLKKLEELLKKGGGC Ac-EALEKALKEALAKLGG-H Ac-YEELLKKLEELLKKAG-H

Mutter [119] Mutter Sherman [131] Sherman [130] Sherman [130] Sherman Carboprotein [140] Carboprotein [145]

X: Leu, Ile, Nle or Val; anchoring through Cys

The underlying concept in TASPs is the preorganization of peptide strands on a molecular scaffold, to reduce the entropy of the construct’s unfolded state, thus to reduce the entropic cost in going from an unassembled conformation to a constrained protein structure [118]. The templates originally utilized by Mutter and coworkers [118– 120] were short, linear peptides inspired by gramicidin S with two antiparallel b-sheet segments, which were eventually cyclized to provide cyclo[-Pro-Gly-Lys-Ala-Lys-Pro-Gly-Lys-Ala-Lys-] and cyclo[-DPro-Gly-Lys-Ala-Lys-D-Pro-Gly-Lys-Ala-Lys-], and their derivatives. Turn mimetics have also been used in these types of template. The

PROTEINS ON TEMPLATES

231

four Lys residues were protected by a set of almost orthogonal Fmoc-, Dde-, Aloc- and Boc-protecting groups, which rendered the template regioselectively addressable. However, functionalization of templates with groups that will allow chemoselective anchoring of peptide strands is clearly preferable to direct amide-bond formation. The chemoselective reactions used have included formation of oximes, hydrazones, thiazolidines, disulfides and thioesters (reaction between thiol(ate) and bromoacetyl moiety). Non-peptide templates, which provide different properties and geometries, have also been used in protein designs. They include porphyrin derivatives [121,122], metal ions (by complexation) [123–125], a cyclohexane derivative (Kemp’s triacid) [126], substituted phenyl rings [127], calix[4]arenes [128], cavitands [129,130] and monosaccharides (vide infra). However, the significance of the template geometry has been debated, as indicated above. Does the template merely tie the peptide strands together, or can it with the proper choice of geometry between ‘anchoring points’ be used to affect the folding? Based on 4-a-helix bundle TASP structures assembled on a series of aromatic and cyclized aromatic templates, Fairlie and coworkers have concluded that, at least with a sufficiently long linker between the peptide strands and the template, the template geometry is of less importance [128]. However, although these templates had different sizes, the peptide anchoring points were all in the same plane and had the same directionality within the same template. In a series of papers, Haehnel and coworkers have applied the TASP approach with cyclic peptide templates to the construction of artificial proteins [131–134]. One key achievement was the construction of an ensemble of 96 potential Cu(II)-binding de novo proteins. The design was based on a 4-helix bundle design with antiparallel strands derived from the backbone structure of the ROP protein. The de novo proteins contained one Cys and two His residues. It was implemented as a combinatorial synthesis on a cellulose membrane (SPOT [135] synthesis). The cellulose membrane-bound proteins were studied by UV-Vis spectroscopy, through which about one third of the 96 proteins indicated copper complexation. However, the copper–protein complexes decomposed over time. Some of the TASP proteins were resynthesized in solution for more extensive studies by 1D NMR, CD and EPR spectroscopy, and resonance Raman spectroscopy. Carboproteins are protein models assembled on a central carbohydrate template, which holds the peptide secondary structural element strands

232

DE NOVO DESIGN OF PROTEINS

together ([136]; for a review, see [137]). So far, mostly monosaccharides in the 4C1 pyranoside conformation have been used. The relatively rigid 4 C1 conformation of most pyranosides enforces well-defined axial or equatorial orientations of the hydroxyls. Monosaccharides differ in the orientation of their hydroxyls, and when the hydroxyls are used as attachment points for the peptides, the choice of carbohydrate template could be expected to determine or influence the distance geometry of the peptide strands. We have developed an efficient strategy for the synthesis of carboproteins, in which amphiphilic C-terminal peptide aldehydes were ligated by oxime bond formation to tetra-aminooxyacetylfunctionalized monosaccharide templates (Figure 6.8) [138–141]. A sequence used in some carboproteins was Ac-YaEbEcLdLeKfKgLa EbEcLdLeKfKgAaG-H (with the heptad repeat assignment in superscript), which was N-acetylated and coupled as the C-terminal aldehyde. It thus allows for four turns in an a-helix, which should be sufficient for helix bundle formation, and includes three Leu and one Tyr (N-terminal) and one Ala (C-terminal) in a and d positions. The sequence is related to a1B and a1A (Table 6.1) by DeGrado and coworkers (for a review, see [142]), as well as a sequence by Sherman and coworkers used to design 4- and 3-helix bundles [143]. This sequence was used for the synthesis of 8.1 kDa a-helical bundles by oxime ligation of tetra-aminooxyacetyl-functionalized D-galacto-, D-gluco- and D-altropyranoside templates [144]. CD spectroscopy indicated that the choice of template has an effect on the overall structure of the carboprotein, as the altro-based carboprotein was found to be more a-helical than the corresponding galacto- and gluco-carboproteins. On the other hand, no influence on stability was detected in these experiments, as the three carboproteins gave similar free energy of foldings (DGFH2O) and melting points in chemical and thermal denaturation experiments. However, CD spectroscopy only provides information on secondary structural elements. To obtain the crucial information on the overall topology, these carboproteins were studied by SAXS, which clearly indicated that the topology was not that of the expected 4-helix bundle but rather of a 3-helix bundle with the fourth peptide chain extending opposite to the helix bundle [145]. The overall topology was independent of the template used, which clearly indicated that the topology was determined by the peptide sequence. However, the template did have a noticeable impact on the solution structure. This was in particular evident when comparing the 2  2helix carboprotein dimer (DTT template; Figure 6.8) with the 4-helix

PROTEINS ON TEMPLATES

233

RO OR

OR RO O

OR O

RO RO

RO RO

O RO

OCH3

RO

RO

OCH3

OCH3

R = peptides attached though linker regions

P e p t i d e

P e p t i d e P e p t i d e

HN

NH

OO

O

O O O

O

P e p t i d e

NH

HN

N

N O

NH

O O

O O

HN

P e p t i d e

P e p t i d e

OCH3

O

O

O S S

D-Galp template

cyclo-DTT template

Figure 6.8 Schematic presentation of carboproteins

carboprotein monomers, as the 2  2-helix structure adopted a more compact conformation. Furthermore, the clear conformational differences observed between the two 4-helix (3 þ 1) carboproteins based on D-altropyranoside and D-galactopyranoside supports the notion that the folding is affected by the template and that subtle variations in template distance-geometry design may be exploited to templatecontrol the solution fold. In an application to nanobioscience, a hemi-4-helix bundle designed to dimerize was synthesized by attaching two copies of an a-helical peptide to a cyclo-DTT template [146]. This hemi-4-helix bundle formed a SAM on Au(111) with 2  2-helix bundle formation, providing ˚ . Both parallel and antiparallel 2  24-helix bundle diameters of 23–27 A helix bundles were believed to be formed, with the latter providing better electron tunneling due to compensation of the dipole moment of the two halves.

234

6.5

DE NOVO DESIGN OF PROTEINS

FOLDAMERS

Foldamers are non-natural oligomers that adopt a secondary structure by stabilization through noncovalent bonds (Figure 6.9). These artificial molecules are designed to mimic proteins and other biopolymers by folding into well-defined conformations, including helices and b-sheets (for a review, see [147]). Probably the most prominent members of biomimetic foldamer oligomers are the b-peptides, which are assembled from homologated amino acids, i.e. a CH2 is inserted between carbonyl and amino moieties. Other building blocks include cyclic structures, for example cyclopentanes. Foldamers that mimic proteins can be considered peptidomimetics. For an overview of peptidomimetics and other aspects of foldamers, see Chapter 3 by Maes and Tourwe´. 3

2

R

N H

β

1

O

α

β3-peptide

Figure 6.9 Example of a foldamer unit

Foldamers are peptidomimetics and so far primarily short sequences have been assembled, thus foldamers will only be mentioned briefly here (for a review, see [148]). However, a few applications to the construction of protein mimics have been reported. Foldamers of b3-amino acids have their side chain on C3 relative to the carbonyl, while b2-amino acids have it at C2. As for peptides assembled from a-amino acids, the amide bonds are rigid. Foldamers assembled from b-amino acids can be stabilized by hydrogen bonds, forming either a 14-member ring between the NH at residue i and the carbonyl at i þ 2, which forms a 14-helix that can be either left- or right-handed [149]. b3-amino acids derived from proteinogenic L-amino acids have a high propensity to form left-handed 14-helices [150]. The 14-helix has approximately three residues per turn, i.e. a three-residue repeat, giving an approximately triangle-like helical wheel presentation. In a b-peptide 14-helix, the C3 atoms of residues i and i þ 3 ˚ compared to 6.3 A ˚ in an a-helix) and side chains projecting are close (4.8 A from these positions are nearly parallel to one another [151]. Also, the

FOLDAMERS

235

14-helix has a slightly wider radius and a shorter rise than a comparable a-helix. The macrodipole in a 14-helix has partial negative and positive charges on the N-terminus and the C-terminus, respectively. Notably, compared to an a-helix, carbonyl groups in a 14-helix point in the opposite direction. An interesting feature is the greater conformational stability observed for short b-peptides compared to a-peptides. Gellman and coworkers have reported that sequences of cyclopentanecontaining amino acids trans-2-amino-cyclopentanecarboxylic acid (ACPC) form a 12-helix with a 12-membered ring between a carbonyl at position i and an NH at position i þ 3 [152]. The helix repeat is approximately 2.5 and it has the same polarity as an a-helix. 10/12helices with intertwined 10- and 12-membered hydrogen bonds have also been observed [153,154]. In an application of b-peptides to materials science, Gellman and coworkers have shown that helical b-decapeptides, assembled from trans-2-aminocyclohexane carboxylic acids, can serve as mesogens for lyotropic liquid crystal phase formation in water [155]. Difficulties in assembling longer sequences of b-peptides by solid phase synthesis have limited the application of foldamers to protein design [149]. However, microwave heating during couplings may offer improvements in the assembly [156]. In one of only a few applications of foldamers to protein design through construction of a tertiary structure, Schepartz and coworkers have reported that b-peptides can assemble into octameric bundles, which would be protein-like but nonproteinaceous [157–160]. Initially, their b3-dodecapeptide constructs exhibit alternating cationic (homoornithine) and anionic side chains (homo-aspartic acid) on one helical face, while a second face exhibits b-homoleucine residues. The third face exhibits homo-glutamic acid, homo-phenylalanine or homo-4-iodo-phenylalanine residues. Eight copies of the b-peptides assemble into an octamer with low affinity. Starting from the above, Petersson and Schepartz designed and synthesized an antiparallel helix–loop–helix structure, containing homo-ornithine, homo-glutamic acid, homo-aspartic acid and homo-4-iodo-phenylalanine in the helices [161]. With a loop of four homo-glycine (b-alanine) residues, this gave a 28-mer b-peptide, Z28, which was synthesized using microwave heating. CD spectroscopy, analytical ultracentrifugation and size-exclusion chromatography coupled with light scattering indicated a monomer–tetramer equilibrium, predominantly tetrameric in the concentration range 9–90 mM.

236

6.6

DE NOVO DESIGN OF PROTEINS

BIOPHARMACEUTICAL APPLICATIONS OF DE NOVO DESIGN

This section will highlight a few applications of the principles of protein de novo design to the design of potential peptide drug candidates. We foresee a large potential for the application of these methods in peptide drug design.

6.6.1

a-helical Structures in Biopharmaceutical Applications

Stroud and coworkers have used a de novo-designed 4-helix bundle protein, DHP-1, as a platform for the development of an IL-4 antagonist [162]. The goal was to develop an IL-4 antagonist that competes for the receptor IL-4Ra. IL-4 is a member of the short-chain helical cytokine family; it is a 4-helix bundle with an up-up-down-down topology. Stroud and coworkers practiced ‘side-chain transplantation’ (grafting) on to DHP-1, based on the coordinates from the crystal structure of the IL4IL-4Ra complex. The crossing angle (O, interhelical angle, i.e. the angle between the helix axes when projected on to their plane of contact [163]) between helices was as important as the overlap (correct placement) of Ca positions in helices. Stroud and coworkers designed and expressed a 108 AA 4-helix bundle, which is one of the largest, most stable and functional de novo-designed proteins, adressing only the high-affinity receptor binding site, not the putative second receptor binding site. Although relatively stable proteins were obtained, the binding to IL-4Ra was significantly lower than for native IL-4, maybe because of non-optimal display of side chains. One of the most significant new peptide drugs of recent years is Enfuvirtide (Fuzeon, T-20), developed by Trimeris in collaboration with Roche [164]. While most – if not all other – peptide drugs are potent peptide hormones, Enfuvirtide appears to interfere with the transient formation of a 6-helix structure during membrane fusion. Enfuvirtide, a 36-mer peptide derived from the natural gp41 HR2 sequence, is the first HIV fusion inhibitor. The first synthetic peptide in this series to show inhibition of HIV membrane fusion was DP-107, a 38-mer peptide corresponding to gp160 residues 558–595 (N-terminal end of gp41) [165,166]. Another important early fusion inhibitor was C34 [167]. Also, Enfuvirtide, originally designated DP-178, is not a de novo peptide

BIOPHARMACEUTICAL APPLICATIONS OF DE NOVO DESIGN

237

sequence, but is derived from the natural gp41 sequence (residues 643–678 of HIV-1LAI gp160; C-terminal region of the gp41 ectodomain). However, it is relevant in this context, as it is an a-helical sequence with a heptad repeat sequence, which interacts with coiled-coil trimer formation, thus blocking the formation of the native 6-helix bundle structure. It is produced on a ton scale by chemical synthesis [168,169]. A follow-up compound, T-1249, as 39-mer peptide, was developed but eventually discontinued. Numerous peptides have been designed as potential membrane fusion inhibitors [170]. From a de novo-design perspective, it is interesting that some of the designs for inhibiting 6-helix formation centred around reengineering of the heptad repeat structure. Kim and coworkers developed hybrid structures for HIV membrane fusion inhibition, which combine a helical segment from the N-terminus (which by itself has a tendency to aggregate) and an unrelated sequence, forming a trimeric coiled coil and solubilizing the construct. The first of these was IQN17, a 46-mer peptide, with a C-terminal 17-mer derived from HIV-1 gp41 NHR region (N), while the N-terminal segment was derived from the sequence of a designed trimeric coiled coil, GCN4-pIQI0 [171]. This de novo, chimeric structure formed highly stable trimeric helices, showed good solubility and inhibited membrane fusion (in nM (IC50) range). In a second generation of compounds, the binding region (N) was permutated, and a modified isoleucine zipper, instead of the GCN4-pIQI0 , was studied [172]. Kim and coworkers also tried the opposite strategy to interfere with trimer-of-hairpin formation, i.e. instead of a single peptide, they created a 5-helix bundle structure, designed to bind one peptide segment in the C-peptide region of gp41 [173]. Three N-peptide segments (derived from NHR, HR1) and two C-peptide segments (derived from CHR, C-terminal heptad repeat, HR2) were joined (N-C-N-C-N) by short loops containing Gly and Ser. The 5-helix protein displayed potent (nanomolar) inhibitory activity against diverse HIV-1 variants. Incidentally, they also constructed a related 6-helix structure, which proved very stable. Truncated analogues of DP-178, conformationally restrained by sidechain-to-side-chain cyclization with a linker, have also been studied [174]. These short peptides have a high degree of a-helicity due to covalent links between i and i þ 7 and show good binding. In an example of grafting, i.e. transplantation of side chains from a peptide to a protein scaffold, Sia and Kim transferred the binding epitope of the peptide C34 on to variants of the GCN4 Leu zipper [175]. Nineteen

238

DE NOVO DESIGN OF PROTEINS

amino acids were transferred; however, the first C34-GCN4 hybrid was hydrophobic and exhibited a low solubility. A disulfide-linked heterodimer, C34coil, overcame this problem and in addition had a high a-helical content. Apparently, GCN4 is well suited to acting as a scaffold protein due to its general tolerance for substitutions on the solventexposed surface.

6.6.2

Foldamers in Biopharmaceutical Applications

Antimicrobial peptides have been engineered from cationic b-peptides. Natural antimicrobial peptides from a-amino acids are in general amphilic, with cationic and hydrophobic sides [176]. Seebach et al. and DeGrado et al. have reported 14-helix antimicrobial b-peptides, using triads of hydrophobic-cationic-hydrophobic b-amino acids [177–179]. Gellman and coworkers reported a cationic, antimicrobial b-peptide with low hemolytic activity [180]. This amphiphilic 12-helix, assembled from cyclic b-amino acids, had 2,5 residues per turn and thus 5-residue repeat, creating pentad repeats of cationic-hydrophobic-cationic-hydophobichydrophobic. Other antimicrobial b-peptides have been found to be hemolytic. However, the oligomers with the appropriate length, amphiphilicity and balance between hydrophilicity and hydrophobicity could be selective antimicrobial agents [151]. In addition, a b-peptide was a low-affinity inhibitor of HIV membrane fusion [181]. A 14-helical b-decapeptide was reported to have antifungal activity against Candida albicans [182].

REFERENCES [1] L. Baltzer, H. Nilsson and J. Nilsson, De novo design of proteins – what are the rules?, Chem. Rev., 101, 3153–3163 (2001). [2] W. F. DeGrado, C. M. Summa, V. Pavone, F. Nastri, and A. Lombardi, De novo design and structural characterization of proteins and metalloproteins, Annu. Rev. Biochem., 68, 779–819 (1999). [3] R. B. Hill, D. P. Raleigh, A. Lombardi and W. F. DeGrado, De novo design of helical bundles as models for understanding protein folding and function, Acc. Chem. Res., 33, 745–754 (2000). [4] J. W. Bryson, S. F. Betz, H. S. Lu, D. J. Suich, H. X. Zhou, K.T. O’Neil and W. F. DeGrado, Protein design: A hierarchic approach, Science, 270, 935–941 (1995).

REFERENCES

239

[5] C. Branden and J. Tooze, Introduction to Protein Structure, 2nd Ed., Garland, New York, 1999. [6] C. Toniolo and E. Benedetti, The polypeptide 310 helix, Trends Biochem. Sci., 16, 350–353 (1991). [7] M. Petukhov, V. Munoz, N. Yumoto, S. Yoshokawa and L. Serrano, Position dependence of non-polar amino acid intrinsic helical propensities, J. Mol. Biol., 278, 279–289 (1998). [8] S. Marqusee, V. H. Robbins and R. L. Baldwin, Unusually stable helix formation in short alanine-based peptides, PNAS, 86, 5286–5290 (1992). [9] P. C. Lyu, M. I. Liff, L. A. Marky and N. R. Kallenbach, Side-chain contributions to the stability of a-helical structure in peptides, Science, 250, 669–673 (1990). [10] J. M. Scholtz, H. Qian, V. H. Robbins and R. L. Baldwin, The energetics of ion-pair and hydrogen-bonding interactions in a helical peptide, Biochemistry, 32, 9668–9676 (1993). [11] S. H. Park, W. Shalongo and E. Stellwagen, Residue helix parameters obtained from dichromic analysis of peptides of defined sequence, Biochemistry, 32, 7048–7053 (1993). [12] B. M. P. Huyghens-Despointe, J. M. Scholtz and R. L. Baldwin, Effect of a single aspartate on helix stability at different positions in a neutral alanine-based peptide, Protein Sci., 2, 1604–1611 (1993). [13] A. M. Felix, E. P. Heimer, C. T. Wang, T. J. Lambros, A. Fournier, T. F. Mowles, S. Maines, R. M. Campbell, B. B. Wegrzynski, V. Toome, D. Fry and V. S. Madison, Synthesis, biological activity and configurational analysis of cyclic GRF analogues, Int. J. Pept. Protein Res., 32, 441–454 (1988). [14] D. Y. Jackson, D. S. King, J. Chmielewski, S. Singh and P. G. Schultz, General approach to the synthesis of short a-helical peptides, J. Am. Chem. Soc., 113, 9391–9392 (1991). ¨ sapay and J. W. Taylor, Multicyclic polypeptide model compounds. 2. [15] G. O Synthesis and conformational properties of a highly a-helical uncosapeptide constrained by three side-chain to side-chain lactam bridges, J. Am. Chem. Soc., 114, 6966–6973 (1992). [16] L. D. Walensky, A. L. Kung, I. Escher, T. J. Malia, S. Barbuto, R. D. Wright, G. Wagner, G. L. Verdine and S. J. Korsmeyer, Activation of apoptosis in vivo by a hydrocarbon-stapled BH3 helix, Science, 305, 1466–1470 (2004). [17] L. D. Walensky, K. Pitter, J. Morash, K. J. Oh, S. Barbuto, J. Fischer, E. Smith, G. L. Verdine and S. J. Korsmeyer, A stapled BID BH3 helix directly binds and activates BAX, Molecular Cell, 24, 199–210 (2006). [18] H. E. Blackwell and R. H. Grubbs, Highly efficient synthesis of covalently crosslinked peptide helices by ring-closing metathesis, Angew. Chem. Int. Ed. Engl., 37, 3281–3284 (1998). [19] C. Schafmeister, J. Po and G . L. Verdine, An all-hydrocarbon cross-linking system for enhancing the helicity and metabolic stability of peptides, J. Am. Chem. Soc., 122, 5891–5892 (2000). [20] D. S. Kemp, J. G. Boyd and C. C. Muendel, The helical S-constant for alanine in water derived from template–nucleated helices, Nature, 352, 451–454 (1991). [21] V. Munoz and L. Serrano, Elucidating the folding problem of helical peptides using empirical parameters. 2. Helix macrodipole effects and rational modification of the helical content of natural peptides, J. Mol. Biol., 245, 275–296 (1995).

240

DE NOVO DESIGN OF PROTEINS

[22] H. Nicholson, W. J. Becktel and B. W. Matthews, Enhanced protein thermostability from designed mutations that interact with a–helix dipoles, Nature, 336, 651–656 (1988). [23] H. Nicholson, D. E. Anderson, S. Dao-Pin and B. W. Matthews, Analysis of the interaction between charged side-chains and the a-helix dipole using designed thermostable mutants of phage T4-lysozyme, Biochemistry, 30, 9816–9828 (1991). [24] J. S. Richardson and D. C. Richardson, Amino acid preferences for specific locations at the ends of a-helices, Science, 240, 1648–1652 (1988). [25] L. G. Presta and G. D. Rose, Helix signals in proteins, Science, 240, 1632–1641 (1988). [26] D. F. Stickle, L. G. Presta, K. A. Dill and G. D. Rose, Hydrogen-bonding in globular proteins J. Mol. Biol., 226, 143 (1992). [27] E. T. Harper and G. D. Rose, Helix stop signals in proteins and peptides – The capping box, Biochemistry, 32, 7605–7609 (1993). [28] R. Aurora, R. Srinivasan and G. D. Rose, Rules for a-helix termination by glycine, Science, 264, 1126–1130 (1994). [29] P. Lyu, H. X. Zhou, N. Jelveh, D. E. Wemmer and N. R. Kallenbach, Positiondependent stabilizing effects in a-helices – N-terminal capping in synthetic model peptides, J. Am. Chem. Soc., 114, 6560–6562 (1992). [30] R. Aurora and G. D. Rose, Helix capping, Protein Science, 7, 21–38 (1998). [31] E. N. Baker and R. E. Hubbard, Hydrogen-bonding in globular proteins, Prog. Biophys. Mol. Biol., 44, 97–179 (1984). [32] V. Munoz, F. J. Blanco and L. Serrano, The hydrophobic stable motif and a role for loop-residues in a-helix stability and protein-folding, Nat. Struct. Biol., 2, 380–385 (1995). [33] J. Prieto and L. Serrono, C-capping and helix stability: The Pro C-capping motif, J. Mol. Biol., 274, 276–88 (1997). [34] K. Gunasekaran, H. A. Nagarajaram, C. Ramakrishnan and P. Balaram, Stereochemical punctuation marks in protein structurs: Glycine and proline containing helix stop signals J. Mol. Biol., 275, 917–932 (1998). [35] L. K. Henchey, A. L. Jochim and P. S. Arora, Contemporary strategies for the stabilization of peptides in the a-helical conformation, Curr. Opin. Chem. Biol., 12, 692–697 (2008). [36] R. S. Hodges, A. K. Saund, P. C. S. Chong, S. A. St.Pierre and R. E. Reid, Synthetic model for 2-stranded a-helical coiled-coils – Design, synthesis, and characterization of an 86-residue analogue of tropomyosin, J. Biol. Chem., 256, 1214–1224 (1981). [37] N. E. Zhou, C. M. Kay and R. S. Hodges, Synthetic model proteins: – Positional effects of interchain hydrophobic interactions on stability of 2-stranded a-helical coiled-coils, J. Biol. Chem., 267, 2664–2670 (1992). [38] J. P. Schneider, A. Lombardi and W. F. DeGrado, Analysis and design of threestranded coiled coils and three-helix bundles, Folding & Design, 3, R29–R40 (1998). [39] F. H. C. Crick, The packing of a–helices: Simple coiled-coils, Acta Cryst., 6, 689–697 (1953). [40] C. Chothia, M. Levitt and D. Richardson, J. Mol. Biol. 145, 215–250 (1981). [41] S. R. Presnell and F. E. Cohen, Topological distribution of four-a-helix bundles, PNAS, 86, 6592–6596 (1989).

REFERENCES

241

[42] O. Khakshoor and J. S. Nowick, Artificial b-sheets: Chemical models of b-sheets, Curr. Opin. Chem. Biol., 12, 722–729 (2008). [43] S. Aravinda, N. Shamala, R. Rajkishore, H. N. Gopi and P. Balaram, A crystalline b-hairpin peptide nucleated by a type I0 Aib-D-Ala b-turn: Evidence for cross-strand aromatic interactions, Angew. Chem. Int. Ed., 41, 3863–3865 (2002). [44] R. Rai, S. Raghothama, R. Sridharan and P. Balaram, Tuning the b-turn segment in designed peptide b-hairpins: construction of a stable type I0 b-turn nucleus and hairpin–helix transition, Biopolymers, 88, 350–361 (2007). [45] L. R. Masterson, M. A. Etienne, F. Porcelli, G. Barany, R. P. Hammer and G. Veglia, Nonstereogenic a-aminoisobuturyl-glycyl-dipeptidyl unit nucleates type I0 b-turn in linear peptides in aqueous solution, Biopolymers, 88, 746–753 (2007). [46] F. A. Syud, J. F. Espinosa and S. H. Gellman, NMR-based quantification of b-sheet populations inaqueous solution through use of reference peptides for the folded and unfolded states, J. Am. Chem. Soc., 121, 11577–11578 (1999). [47] S. Kang and J. G. Saven, Computational protein design: structure, function and combinatorial diversity, Curr. Opin. Chem. Biol., 11, 329–334 (2007). [48] S. Park, X. Yang and J. G. Saven, Advances in computational protein design, Curr. Opin. Struct. Biol., 14, 487–494 (2004). [49] K. T. Simons, C. Strauss and D. Baker, Prospects for ab initio protein structural genomics, J. Mol. Biol., 306, 1191–1199 (2000). [50] D. Baker and A. Sali, Protein structure prediction and structural genomics, Science, 294, 93–96 (2001). [51] B. Kuhlman, G. Dantas, G. C. Ireton, G. Varani, B. L. Stoddard and D. Baker, Design of a novel globular protein fold with atomic-level accuracy, Science, 302, 1364–1368 (2003). [52] J. R. Desjarlais and T. M. Handel, De novo design of the hydrophobic cores of proteins, Protein Sci., 4, 2006–2018 (1995). [53] N. D. Clarke and S. M. Yuan, Metal search – A computer program that helps design tetrahedral metal-binding sites, Proteins, 23, 256–263 (1995). [54] H. Kono and J. G. Saven, Statistical theory for protein combinatorial libraries. Packing interactions, backbone flexibility, and the sequence variability of a mainchain structure, J. Mol. Biol., 306, 607–628 (2001). [55] B. I. Dahiyat and S. L. Mayo, De novo protein design: Fully automated sequence selection, Science, 278, 82–87 (1997). [56] W. F. DeGrado, Proteins from scratch, Science, 278, 80–81 (1997). [57] J. Liu, Q. Zheng, Y. Deng, C. S. Cheng, N. R. Kallenbach and M. Lu, A seven-helix coiled coil, PNAS, 103, 15457–15462 (2006). [58] J. M. Mason and K. M. Arndt, Coiled coil domains: Stability, specificity, and biological implications, Chembiochem, 5, 170–176 (2004). [59] P. B. Harbury, J. J. Plecs, B. Tidor, T. Alber and P. S. Kim, High-resolution protein design with backbone freedom, Science, 282, 1462–1467 (1998). [60] S. Y. M. Lau, A. K. Taneja and R. S. Hodges, Synthesis of a model protein of defined secondary and quarternary structure – Effect of the chain-length on the stabilization and formation of 2-stranded a-helical coiled-coils, J. Biol. Chem., 259, 3253–3261 (1984). [61] O. D. Monera, N. E. Zhou, C. M. Kay and R. S. Hodges, Comparison of antiparallel and parallel 2-stranded a-helical coiled-coils. Design, synthesis and characterization, J. Biol. Chem., 268, 19218–19227 (1993).

242

DE NOVO DESIGN OF PROTEINS

[62] O. D. Monera, C. M. Kay and R. S. Hodges, Electrostatic interactions control the parallel and antiparallel orientation of a-helical chains in 2-stranded a-helical coiled coils, Biochemistry, 33, 3862–3871 (1994). [63] P. B. Harbury, T. Zhang, P. S. Kim and T. Alber, A switch between two-, three-, and four-standed coiled coils in GCN4 leucine zipper mutants, Science, 262, 1401–1407 (1993). [64] K. T. O’Neil and W. F. DeGrado, A thermodynamic scale for the helix-forming tendencies of the commonly occuring amino acids, Science, 250, 646–651 (1990). [65] B. Lovejoy, S. Choe, D. Cascio, D. K. McRorie, W. F. DeGrado and D. Eisenberg, Crystal structure of a synthetic triple-stranded a-helical bundle, Science, 259, 1288–1293 (1993). [66] S. Betz, R. Fairman, K. O’Neil, J. Lear and W. F. DeGrado, Design of 2-stranded and 3-stranded coiled-coil peptides, Philos. Trans. R. Society London Ser. B., 348, 81–88 (1995). [67] J. A. Boice, G. R. Dieckmann, W. F. DeGrado and R. Fairman, Thermodynamic analysis of a designed three-stranded coiled coil, Biochemistry, 35, 14480–14485 (1996). [68] N. L. Ogihara, M. S. Weiss, W. F. DeGrado and D. Eisenberg, The crystal structure of the designed trimeric coiled coil coil-VaLd: Implications for engineering crystals and supramolecular assemblies, Protein Sci., 6, 80–88 (1997). [69] W. F. DeGrado, C. M. Summa, V. Pavone, F. Nastri and A. Lombardi, De novo design and structural characterization of proteins and metalloproteins, Annu. Rev. Biochem., 68, 779–819 (1999). [70] L. Brive, G. T. Dolphin and L. Baltzer, Structure and function of an aromatic ensemble that restricts the dynamics of the hydrophobic core of a designed helix– loop–helix dimer, J. Am. Chem. Soc., 119, 8598–8607 (1997). [71] A. G. Murzin and A. V. Finkelstein, General architecture of the a-helical globule, J. Mol. Biol., 204, 749–769 (1988). [72] B. Honig and A.-S. Yang, Free energy balance in protein-folding, Adv. Protein Chem., 46, 27–58 (1995); K. Yue and K. Dill, PNAS, 89, 4163–4167 (1992). [73] J. W. Bryson, J. R. Desjarlais, T. M. Handel and W. F. DeGrado, From coiled coils to small globular proteins: Design of a native-like three-helix bundle, Protein Sci., 7, 1404–1414 (1998). [74] S. T. R. Walsh, H. Cheng, J. W. Bryson, H. Roder and W. F. DeGrado, Solution strcuture and dynamics of a de novo designed three-helix bundle protein, PNAS, 96, 5486–5491 (1999). [75] S. T. R. Walsh, A. L. Lee, W. F. DeGrado and A. J. Wand, Dynamics of a de novo designed three-helix bundle protein studied by N-15, C-13, and H-2 NMR relaxation methods, Biochemistry, 40, 9560–9569 (2001). [76] A. Lombardi, J. W. Bryson, G. Ghirlanda and W. F. DeGrado, Design of a synthetic receptor for the calmodulin-binding domain of calcineurin, J. Am. Chem. Soc., 119, 12378–12379 (1997). [77] G. Ghirlanda, J. D. Lear, A. Lombardi and W. F. DeGrado, From synthetic coiled coils to functional proteins: Automated design of a receptor for the calmodulinbinding domain of calcineurin, J. Mol. Biol., 281, 379–381 (1998). [78] G. R. Dieckmann, D. K. McRorie, J. D. Lear, K. A. Sharp, W. F. DeGrado and V. L. Pecoraro, The role of protonation and metal chelation preferences in defining the properties mercury-binding coiled-coils, J. Mol. Biol., 280, 897–912 (1998).

REFERENCES

243

[79] D. Ghosh and V. L. Pecoraro, Understanding metalloprotein folding using a de novo design strategy, Inorg. Chem., 43, 7902–7915 (2004). [80] S. P. Ho and W. F. DeGrado, Design of a 4-helix bundle protein – Synthesis of peptides which self-associate into a helical protein, J. Am. Chem. Soc., 109, 6751–6758 (1987). [81] D. P. Raleigh and W. F. DeGrado, A de novo designed protein shows a thermally induced transition from a native to a molten globule like state, J. Am. Chem. Soc., 114, 10079–10081 (1992). [82] D. P. Raleigh, S. F. Betz and W. F. DeGrado, A de novo designed protein mimics the native-state of natural proteins, J. Am. Chem. Soc., 117, 7558–7559 (1995). [83] R. B. Hill, W. F. DeGrado, Solution Structure of a2D, a native-like de novo designed protein, J. Am. Chem. Soc., 120, 1138–1145 (1998). [84] C. E. Schafmeister, S. L. LaPorta, L. J. W. Miercke and R. M. Stroud, A designed four helix bundle protein with native-like structure, Nat. Struct. Biol., 4, 1039–1046 (1997). [85] C. E. Schafmeister, L. J. W. Miercke and R. M. Stroud, Structure at 2.5 A˚ngstro¨m of a designed peptide that maintains solubility of membrane proteins, Science, 262, 734–738 (1993). [86] K.-H. Lee, H.-Y. Lee, M. S. Slutsky, J. T. Anderson and E. N. G. Marsh, Fluorous effect in proteins: De novo design and characterization of a four-ahelix bundle protein containing hexafluoroleucine, Biochemistry, 2004, 43, 16277–16284. [87] S. F. Betz and W. F. DeGrado, Controlling topology and native-like behavior of de novo-designed peptides: Design and characterization of antiparallel four-stranded coiled coils, Biochemistry, 35, 6955–6962 (1996). [88] L. K. Andersson, M. Caspersson and L. Baltzer, Control of lysine reactivity in four-helix bundle proteins by site-selective pKa depression: Expanding the versatility of proteins by postsynthetic functionalization, Chem. Eur. J., 8, 3687–3697 (2002). [89] K. Enander, G. T. Dolphin, B. Liedberg, I. Lundstro¨m and L. Baltzer, A versatile polypeptide platfom for integrated recognition and reporting: Affinity arrays for protein–ligand interaction analysis, Chem. Eur. J., 10, 2375–2385 (2004). [90] P. Rossi, P. Tecilla, L. Baltzer and P. Scrimin, De novo metallonucleases based on helix–loop–helix motifs, Chem. Eur. J., 10, 4163–4170 (2004). [91] C. T. Choma, J. D. Lear, M. J. Nelson, P. L. Dutton, D. E. Robertson and W. F. DeGrado, Design of a heme-binding 4-helix bundle, J. Am. Chem. Soc., 116, 856–865 (1994). [92] D. E. Robertson, R. S. Farid, C. C. Moser, J. L. Urbauer, S. E. Mulholland, R. Pidikiti, J. D. Lear, A. J. Wand, W. F. DeGrado and P. L. Dutton, Design and synthsis of multi-heme proteins, Nature, 368, 425–432 (1994). [93] F. V. Cochran, S. P. Wu, W. Wang, V. Nanda, J. G. Saven, M. J. Therien and W. F. DeGrado, Computational de novo design and characterization of a four-helix bundle protein that selectively binds a nonbiological cofactor, J. Am. Chem. Soc., 127, 1346–1347 (2005). [94] B. R. Gibney, S. E. Mulholland, F. Rabanal and P. L. Dutton, Ferredoxin and ferredoxin-heme maquettes, PNAS, 93, 15041–15046 (1996). [95] L. Regan and W. F. DeGrado, Characterization of a protein designed form 1st principles, Science, 241, 976–978 (1988).

244

DE NOVO DESIGN OF PROTEINS

[96] M. P. Scott and J. Biggins, Introduction of a [4Fe-4S (S-Cys)4)(þ1, þ2)iron-sulfur center into a four-a-helix protein using design parameters from the domian of the F-x cluster in the Photosystem I reaction center, Protein Sci., 6, 340–346 (1997). [97] T. M. Handel and W. F. DeGrado, De novo design of a Zn2þ binding protein, J. Am. Chem. Soc., 112, 6710–6711 (1990). [98] T. M. Handel, S. A. Williams and W. F. DeGrado, Metal-ion dependent modulation of the dynamics of a designed protein, Science, 261, 879–885 (1993). [99] L. Regan and N. D. Clarke, A tetrahedral Zinc(II)-binding site introduced into a designed protein, Biochemistry, 29, 10878–83 (1990). [100] S. J. Lahr, D. E. Engel, S. E. Stayrock, O. Maglio, B. North, S. Geremia, A. Lombardi and W. F. DeGrado, Analysis and design of turns in a-helical hairpins, J. Mol. Biol., 346, 1441–1454 (2005). [101] F. Wade, S. E. Stayrock and W. F. DeGrado, The structure of a designed diiron(III) protein: Implications for cofactor stabilization and catalysis, Angew. Chem. Int. Ed., 45, 4951–4954 (2006). [102] V. Nanda, M. M. Rosenblatt, A. Osyczka, H. Kono, Z. Getahun, P. L. Dutton, J. G. Saven and W. F. DeGrado, De novo design of a redox-active minimal rubredoxin mimic, J. Am. Chem. Soc., 127, 5804–5805 (2005). [103] J. T. Gerig, Fluorine NMR of proteins, Prog. NMR Spectrosc., 26, 293–370 (1994). [104] E. N. G. Marsh, Towards the nonstick egg: Designing fluorous proteins, Chem. & Biol., 7, R153–R157 (2000). [105] Y. Tang, G. Ghirlanda, W. A. Petka, T. Nakajima, W. F. DeGrado and D. A. Tirrell, Fluorinated coiled-coil proteins prepared in vivo display enhanced thermal and chemical stability, Angew. Chem. Int. Ed., 40, 1494–1496 (2001). [106] X. Xing, A. Fichera and K. Kumar, A novel synthesis of enantiomerically pure 5,5,5,50 ,50 ,50 -hexafluoroleucine, Org. Lett., 3, 1285–1286 (2001). [107] B. Bilgicer, X. Xing and K. Kumar, Programmed self-sorting of coiled coils with leucine and hexafluoroleucine cores, J. Am. Chem. Soc., 123, 11815–11816 (2001). [108] N. C. Yoder and K. Kumar, Fluorinated amino acids in protein design and engineering, Chem. Soc. Rev., 31, 335–341 (2002). [109] L. M. Gottler, R. de la Salud-Bera and E. N. G. Marsh, The fluorous effect in proteins: Properties of a4F6, a 4-a-helix bundle protein with a fluorocarbon core, Biochemistry, 47, 4484–4490 (2008). [110] C. Ja¨ckel, M. Salwiczek and B. Koksch, Fluorine in a native protein environment – How the spatial demand and polarity of fluoroalkyl groups affect protein folding, Angew. Chem. Int. Ed., 45, 4198–4203 (2006). [111] C. Garcia-Echeverria, Unequivocal determination by fluorescence spectroscopy of the formation of a parallel leucine-zipper homodimer, J.Am. Chem. Soc., 116, 6031–6032 (1994). [112] M. Munson, R. O’Brien, J. M. Sturtevant and L. Regan, Redesigning the hydrophobic core of a 4-helix-bundle protein, Protein Sci., 3, 2015–2022 (1994). [113] N. Carula, C. Woodward and G. Barany, BetaCore, a designed water soluble fourstranded antiparallel b-sheet protein, Protein Sci., 11, 1539–1551 (2002). [114] M. D. Struthers, R. P. Cheng and B. Imperiali, Design of a monomeric 23-residue polypeptide with defined tertiary structure, Science, 271, 342–345 (1996). [115] G. Tuchscherer and M. Mutter, In: M. Goodman, A. Felix, L. Moroder and C. Toniolo, Houben-Weyl, E22d, Thieme, Stuttgart, 7–67.

REFERENCES

245

[116] M. Mutter and G. Tuchscherer, Non-native architectures in protein design and mimicry, Cell. Mol. Life Sci., 53, 851–863 (1997). [117] M. Mutter and S. Vuilleumier, A chemical appraoch to protein design – Templateassembled synthetic proteins (TASP), Angew. Chem. Int. Ed. Engl., 28, 535–554 (1989). [118] M. Mutter, G. G. Tuchscherer, C. Miller, K.-H. Altmann, R. I. Carey, D. F. Wyss, A. M. Labhardt and J. E. Rivier, Template-assembled synthetic proteins with fourhelix-bundle topology. Total chemical synthesis and conformational studies, J. Am. Chem. Soc., 114, 1463–1470 (1992). [119] M. Mutter, P. Dumy, P. Garrouste, C. Lehmann, M. Mathieu, C. Peggion, S. Peluso, A. Razaname and G. Tuchscherer, Template assembled synthetic proteins (TASP) as functional mimetics of proteins, Angew. Chem. Int. Ed. Engl., 35, 1482–1485 (1996). [120] S. Peluso, P. Dumy, C. Nkubana, Y. Yokokawa and M. Mutter, Solid–phase strategies for the assembly of template-based protein mimetics, J. Org. Chem., 64, 7114–7120 (1999). [121] T. Sasaki and E. T. Kaiser, Helichrome – Synthesis and enzymatic-activity of a designed hemeprotein, J. Am. Chem. Soc., 111, 380–381 (1989). ˚ kerfeldt, R. M. Kim, D. Camac, J. T. Groves, J. D. Lear and W. F. DeGrado, [122] K. S. A Tetraphilin – A 4-helix proton channel built on a tetraphenylporphyrin framework, J. Am. Chem. Soc., 114, 9656–9657 (1992). [123] M. R. Ghadiri, C. Soares and C. Choi, A convergent approach to protein design – Metal ion-assisted spontaneous self-assembly of a polypeptide into a triple-helix bundle protein, J. Am. Chem. Soc., 114, 825–831 (1992). [124] M. R. Ghadiri, C. Soares and C. Choi, Design of an artificial 4-helix bundle metalloprotein via a novel ruthenium(II)-assisted self-assembly process, J. Am. Chem. Soc., 114, 4000–4002 (1992). [125] M. W. Mutz, M. A. Case, J. F. Wishart, M. R. Ghadiri and G. L. McLendon, De novo design of protein function: Predictable structure–function relationships in synthetic redox proteins, J. Am. Chem. Soc., 121, 858–859 (1999). [126] M. Goodman, Y. Feng, G. Melacini and J. P. Taulane, A template-induced incipient collagen-like triple-helical structure, J. Am. Chem. Soc., 118, 5156–5157 (1996). [127] A. K. Wong, M. P. Jacobsen, D. J. Winzor and D. P. Fairlie, Template assembled synthetic proteins (TASPs). Are template size, shape, and directionality important in formation of four-helix bundles?, J. Am. Chem. Soc., 120, 3836–3841 (1998). [128] Y. Hamuro, M. C. Calama, H. S. Park and A. D. Hamilton, A calixarene with four peptide loops: An antibody mimic for recognition of proteins surfaces, Angew. Chem. Int. Ed. Engl., 36, 2680–2683 (1997). [129] A. S. Causton and J. C. Sherman, Design of proteins using rigid organic macrocycles as scaffolds, Bioorg. Med. Chem., 7, 23–27 (1999). [130] A. R. Mezo and J. C. Sherman, Cavitands are effective templates for inducing stability and nativelike structure in de novo four-helix bundles, J. Am. Chem. Soc., 121, 8983–8994 (1999). [131] H. K. Rau and W. Haehnel, Design, synthesis, and properties of a novel cytochrome b model, J. Am. Chem. Soc., 120, 468–476 (1998). [132] H. K. Rau, N. DeJonge and W. Haehnel, Modular synthesis of de novo designed metalloproteins for light-induced elecetron-transfer, PNAS, 95, 11526–11531 (1998).

246

DE NOVO DESIGN OF PROTEINS

[133] H. K. Rau, N. DeJonge and W. Haehnel, Combinatorial synthesis of four-helix bundle hemoproteins for tuning of cofactor properties, Angew. Chem. Int. Ed., 39, 250–253 (2000). [134] R. Schnepf, P. Ho¨rth, E. Bill, K. Wieghardt, P. Hildebrandt and W. Haehnel, De novo design and characterization of copper centers in synthetic four-helix-bundle proteins, J. Am. Chem. Soc., 123, 2186–2195 (2001). [135] R. Frank, SPOT synthesis – An easy technique for the positionally addressable, parallel chemical synthesis on a membrane, Tetrahedron, 48, 9217–9232 (1992). [136] K. J. Jensen and G. Barany, Carbopeptides: Carbohydrates as templates for de novo design of proteins, J. Peptide Res., 56, 3–11 (2000). [137] K. J. Jensen and J. Brask, Carbohydrates as templates for control of distancegeometry in de novo designed proteins, Cell. Mol. Life Sci., 59, 859–869 (2002). [138] J. Brask and K. J. Jensen, Carbopeptides: Chemoselective ligation of peptide aldehydes to an aminooxy-functionalized D-galactose template, J. Peptide Sci., 6, 290–299 (2000). [139] J. Brask and K. J. Jensen, Carboproteins: A 4-a-helix bundle protein model assembled on a D-galactopyranoside template, Bioorg. Med. Chem. Lett., 11, 697–700 (2001). [140] J. Brask, H. Wackerbarth, K. J. Jensen, J. Zhang, J. U. Nielsen, J. E. T. Andersen and J. Ulstrup, Monolayers of a de novo designed 4-h-helix bundle carboprotein and partial structures on Au(111)-electrodes, Bioelectrochem., 56, 27–32 (2002). [141] J. Brask, H. Wackerbarth, K. J. Jensen, J. Zhang, I. Chorkendorff and J. Ulstrup, Monolayer assemblies of a de novo designed 4-h-helix bundle carboprotein and its sulfur anchor fragment on Au(111)-surfaces addressed by voltametry and in situ scanning tunneling microscopy, J. Am. Chem. Soc., 125, 94–104 (2003). [142] W. F. DeGrado, Z. R. Wassermann and J. D. Lear, Protein design, a minimalistic approach, Science, 243, 622–628 (1989). [143] A. S. Causton and J. C. Sherman, A comparison of three- and four-helix bundle TASP molecules, J. Peptide Sci., 8, 275–282 (2002). [144] J. Brask, J. M. Dideriksen, J. Nielsen and K. J. Jensen, Monosaccharide templates for de novo designed 4-a-helixbundle proteins: Template effects in carboproteins, Org. Biomol. Chem., 1, 2247–2252 (2003). [145] R. Høiberg-Nielsen, A. P. T. Shelton, K. K. Sørensen, M. Roessle, D I. Svergun, P. W. Thusltrup, K. J. Jensen and L. Arleth, 3- instead of 4-helix foramtion in de novo designed protein in solution revealed by small-angle X-ray scattering, ChemBioChem, 9, 2663–2672 (2008). [146] H. Wackerbarth, A. P. Tofteng, K. J. Jensen, I. Chorkendorff and J. Ulstrup, Hierarchical self-assembly of designed 2  2-helix bundle proteins on Au(111) surfaces, Langmuir, 22, 6661–6667 (2006). [147] D. Seebach, A. K. Beck and D. J. Bierbaum, The world of b- and g-peptides comprised of homlogated proteinogenic amino acids and other components, Chemistry & Biodiversity, 1, 1111–1239 (2004). [148] R. P. Cheng, Beyond de novo protein design – De novo design of non-natural folded oligomers, Curr. Opin. Struct. Biol., 14, 512–520 (2004). [149] P. I. Arvidsson, M. Rueping and D. Seebach, Design, machine syntehsis, and NMRsolution structure of a b-heptapeptide forming a salt-bridge stabilized 314-helix in methanol and in water, Chem. Commun., 649–650 (2001).

REFERENCES

247

[150] C. M. Goodman, S. Choi, S. Shandler and W. F. DeGrado, Foldamers as versatile frameworks for the design and evolution of function, Nature Chem. Biol., 3, 252–262 (2007). [151] R. Cheng, S. Gellman and W. F. DeGrado, b-peptides: From structure to function, Chem. Rev., 101, 3219–3232 (2001). [152] H. S. Lee, F. A. Syud, X. Wang and S. H. Gellman, Diversity in short b-peptide 12helices: High-resolution structural analysis in aqueous solution of a hexamer containing sulfonylated pyrrolidine residues, J. Am. Chem. Soc., 123, 7721–7722 (2001). [153] S. A. W. Gruner, V. Truffault, G. Voll, E. Locardi, M. Sto¨ckle and H. Kessler, Design, synthesis, and NMR structure of linear and cyclic oligomers containing novel furanoid sugar amino acids, Chem. Eur. J., 8, 4366–4376 (2002). [154] M. Rueping, J. V. Schreiber, G. Lelais, B. Jaun and D. Seebach, Mixed b2/b3hexapeptides and b2/b3-nonapeptides folding to (P)–helices with alternating twelveand ten-membered hydrogen-bonded rings, Helv. Chim. Acta, 85, 2577–2593 (2002). [155] W. C. Pomerantz, N. L. Abbott and S. H. Gellman, Lyotropic liquid crystals from designed helical b-peptides, J. Am. Chem. Soc., 128, 8730–8731 (2006). [156] J. K. Murray and S. H. Gellman, Application of microwave irradiation to the syntehsis of 14-helical b-peptides, Org. Lett., 7, 1517–1520 (2005). [157] J. X. Qiu, E. J. Petersson, E. E. Matthews and A. Schepartz, Toward b-amino acid proteins: A cooperatively folded b-peptide quaternary structure, J. Am. Chem. Soc., 128, 11338–11339 (2006). [158] D. S. Daniels, E. J. Petersson, J. X. Qiu and A. Schepartz, High-resolution structure of a b-peptide bundle, J. Am. Chem. Soc., 129, 1532–1533 (2007). [159] E. J. Petersson, C. J. Craig, D. S. Daniels, J. X. Qiu and A. Schepartz, Biophysical characterization of b-peptide bundle, comparison to natural proteins, J. Am. Chem. Soc., 129, 5344–5345 (2007). [160] J. L. Goodman, E. J. Petersson, D. S. Daniels, J. X. Qiu and A. Schepartz, Biophysical and structural characterization of a robust octameric b-peptide bundle, J. Am. Chem. Soc., 129, 14746–14751 (2007). [161] E. J. Petersson and A. Schepartz, Toward b-amino acid proteins: design, synthesis, and characterization of a fifteen kilodalton b-peptide tetramer, J. Am. Chem. Soc., 130, 821–823 (2008). [162] S. L. LaPorte, C. M. Forsyth, B. C. Cunningham, L. J. Mierke, D. Akhavan and R. M. Stroud, De novo design of an IL-4 antagonist and its structure at 1.9 A˚, PNAS, 102, 1889–1894 (2005). [163] C. Chothia, M. Levitt and D. Richardson, Helix to helix packing in proteins, J. Mol. Biol., 145, 215–250 (1981). [164] T. Matthews, M. Salgo, M. Greenberg, J. Chung, R. DeMasi and D. Bolognesi, Enfuvirtide: The first therapy to inhibit the entry of HIV-1 into host CD4 lymphocytes, Nature Rev. Drug Discovery, 3, 215–225 (2004). [165] C. Wild, T. Oas, C. McDanal, D. Bolognesi and T. Matthews, A synthetic peptide inhibitor of human immunodeficiency virus replication: Correlation between solution structure and viral inhibition, PNAS, 89, 10537–10541 (1992). [166] C. Wild, J. W. Dubay, T. Greenwell, T. Baird, T. G. Oas, C. McDanal, E. Hunter and T. Matthews, Propensity for a leucine zipper-like domain of human immunodeficiency virus type 1 gp41 to form oligomers correlates with a role in virus-

248

[167] [168] [169]

[170]

[171]

[172] [173] [174]

[175] [176] [177]

[178] [179] [180] [181] [182]

DE NOVO DESIGN OF PROTEINS induced fusion rather than assembly of the glycoprotein complex, PNAS, 91, 12676–12680 (1994). S. Liu, S. Wu and S. Jiang, HIV entry inhibitors targeting gp41: From polypeptides to small-molecyle compounds, Curr. Pharm. Design, 13, 143–162 (2007). B. L. Bray, Large-scale manufacture of peptide therapeutics by chemical synthesis, Nature Rev. Drug Discovery, 2, 587–593 (2003). S. E. Schneider, B. L. Bray, C. J. Mader, P. E. Friedrich, M. W. Anderson, T. S. Taylor, N. Boshernitzan, T. E. Niemi, B. C. Fulcher, S. R. Whight, J. M. White, R. J. Greene, L. E. Stoltenberg and M. Lichty, Development of HIV fusion inhibitors, J. Pept. Sci., 11, 744–753 (2005). S. W. Liu, S. G. Wu and S. B. Jiang, HIV entry inhibitors targeting gp41: From polypeptides to small-molecule compounds, Curr. Pharm. Design, 13, 143–162 (2007). D. M. Eckert, V. N. Malashkevich, L. H. Hong, P. A. Carr and P. S. Kim, Inhibiting HIV-1 entry: Discovery of D-peptide inhibitors that target the gp41 coiled-coil pocket, Cell, 99, 103–115 (1999). D. M. Eckert and P. S. Kim, Design of potent inhibitors of HIV-1 entry from the gp41 N-peptide region PNAS, 98, 11187–11192 (2001). M. J. Root, M. S. Kay and P. S. Kim, Protein design of an HIV-1 entry inhibitor, Science, 291, 884–888 (2001). J. K. Judice, J. Y. K. Tom, W. Huang, T. Wrin, J. Vennari, C. J. Petropoulis and R. S. McDowell, Inhibition of HIV type 1 infectivity by constrained a-helical peptides: Inportance for the viral fusion mechanism, PNAS, 94, 13426–13430 (1997). S. K. Sia and P. S. Kim, Protein grafting of an HIV-1-inhibiting epitope, PNAS, 100, 9756–9761 (2003). A. Tossi, L. Sandri and A. Giangaspero, Amphipathic, a-helical antimicrobial peptides, Biopolymers, 55, 4–30 (2000). P. I. Arvidsson, J. Frackenpohl, N. S. Ryder, B. Liechty, F. Petersen, H. Zimmerman, G. P. Camenisch, R. Woessner and D. Seebach, On the antimicrobial and hemolytic activities of amphiphilic b-peptides, ChemBioChem, 2, 771–773 (2001). Y. Hamuro, J. P. Schneider and W. F. DeGrado, De novo design of antibacterial b-peptides, J. Am. Chem. Soc., 121, 12200–122001 (1999). D. Liu and W. F. DeGrado, De novo design, synthesis, and characterization of antimicrobial b-peptides, J. Am. Chem. Soc., 123, 7553–7559 (2001). E. A. Porter, X. Wang, H.-S. Lee, B. Weisblum and S. H. Gellman, Antibiotics – Non-haemolytic b-amino acid oliogomers, Nature, 404, 565–565 (2000). O. M. Stephens, S. Kim, B. D. Welch, M. E. Hodsdon, M. S. Kay and A. Schepartz, J. Am. Chem. Soc., 127, 13126–13127 (2005). A. J. Karlsson, W. C. Pomerantz, B. Weisblum, S. H. Gellman and S. P. Palecek, Antifungal activity from 14-helical b-peptides, J. Am. Chem. Soc., 128, 12630–12631 (2006).

7 Design of Insulin Variants for Improved Treatment of Diabetes Thomas Hoeg-Jensen

7.1

INTRODUCTION

Prior to the discovery of insulin in Toronto around 1920 [1], diabetes was a terminal disease, a death sentence. The miraculous recovery of diabetes patients following treatment with insulin from animal pancreatic extracts prompted fast establishment of industrial production of insulin, and diabetes changed from a terminal to a manageable disorder. However, although patients could now survive diabetes treatment for many years, it gradually became clear that living with diabetes often resulted in long-term complications such as damage to eyes, kidneys, heart and the cardiovascular system. Diabetes is today the most common course of blindness, and the life expectancy for diabetes patients is lower compared to the average population. Large-scale clinical studies have shown that intensive insulin treatment can limit the risk of long-term complications from diabetes [2,3]. Intensive insulin care involves monitoring of blood glucose several times a day and adjustments of the timing and dosing of insulin administrations (Figure 7.1) [4] in order to approach the tightly regulated glucose levels of a healthy Peptide and Protein Design for Biopharmaceutical Applications © 2009 John Wiley & Sons, Ltd. ISBN: 978-0-470-31961-1

Edited by Knud J. Jensen

250

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

person. Basal insulin levels should be present in the body 24 hours a day, and extra supplies of insulin must be given for handling of the glucose ingested with each meal. However, constant achievement of normal glucose levels is near impossible for diabetes patients due to the many factors involved in blood glucose fluctuations, such as meals of diverse character, exercise, sleep, infections and endogenous glucose production by the liver [5]. Furthermore, intensive insulin treatment includes the risk of having the blood sugar dropping too low, which is acutely dangerous. The ideal therapeutic window for insulin dose is very narrow. Blood glucose levels should ideally stay within the range 4–6 millimols per litre.

Insulin (μU/ml)

100 80 60 40 20 0

0800

1200

2000 1600 Clock Time (hours)

2400

0400

Figure 7.1 Average daily insulin secretion profile in healthy individuals. Insulin drug delivery targets this pattern in pursuit of optimal glucose control, which lowers the risk of long-term complications from diabetes. A selection of fast- and long-acting insulins is required in order to approach this goal. Reproduced by permission of the American Society for Clinical Investigation. Copyright 1988

In order to help the timing and dosing of the insulin drug, scientists have engineered insulin toward achieving slower, faster and more predictable profiles than observed when using native insulin as drug. The current chapter will review and evaluate the known approaches for modulating insulin drug properties, with focus on structural engineering of the insulin molecule targeted toward modulation of insulin pharmacology. Native insulin structure–activity relationships, insulin chemistry and insulin biology have been the subjects of recent reviews [6–8], and these topics will be included here only as necessary for following the drug engineering discussion. Searching for insulin in the public database Medline/PubMed in January 2008 resulted in identification of 229 051 scientific papers. Insulin is clearly one of the best-studied peptide hormones and drugs,

DIABETES MANAGEMENT AND INSULIN ENGINEERING

251

and a review of all aspects of insulin literature is not intended here. Even within the given scope, the cited papers are only a selection. The text is intended for nonspecialists of insulin and diabetes, so terms from the field of diabetology and other technical terms are explained as they first appear.

7.2

DIABETES MANAGEMENT AND THE NEED FOR INSULIN ENGINEERING

In a fasted, healthy person, blood glucose is tightly regulated near 5 millimols per litre (millimolar, mM). Human cells uptake and metabolize glucose as part of the body’s energy balance, and the cellular glucose uptake is regulated by insulin via binding and activation of the insulin receptor present on cellular surfaces. Basal levels of insulin are required 24 hours a day. The pancreatic b-cells continuously excrete insulin in the needed amounts, and sensing mechanisms in the b-cells ensure that insulin is excreted in levels necessary to maintain the glucose level in the normal range of 4–6 mM [9]. Following a meal, blood glucose can rise to around 10 mM, but this rise is countered by excretion of extra insulin (bolus insulin) from the b-cells in order to increase glucose uptake and bring the glucose level back to normal. The insulin secretion is stimulated not just by glucose, but also by incretine hormones like glucagon-like peptide 1 (GLP-1), which is secreted from gut L-cells in response to a meal [10]. Insulin is produced in the b-cells as inactive single-chain proinsulin [11], which is subsequently processed to active two-chain insulin and stored in the b-cells as zinc(II) insulin hexamer granules. In case of exercise or fasting, where blood glucose levels may drop below normal, the pancreatic a-cells will excrete glucagon [12], which in turn stimulates the liver to produce glucose from stored glycogen, thereby maintaining the normal glucose level. There are two main types of diabetes, type 1 and type 2, both diagnosed by raised blood glucose levels. Type 1 diabetes often appears at a young age and is also known as juvenile-onset diabetes. Type 1 is an autoimmune disorder. The insulin-producing b-cells are killed by the immune system, for reasons that are not fully understood. Genetic disposition is well-documented and infection or shock may play a role [13]. The progression of type 1 diabetes is quite fast, and blood glucose levels at the time of diagnosis are often very high, > 30 mM, which results in excessive excretion of glucose in the urine, weight loss and craving thirst, and these

252

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

symptoms usually lead to fast diagnosis. Daily insulin treatments of type 1 patients are crucial. Type 1 is also known as insulin-dependent diabetes mellitus, IDDM. Type 2 diabetes is a metabolic disorder affecting blood glucose, among other things. b-cells and insulin production are still (partially) intact in type 2, but the insulin sensitivity in the peripheral tissue is lower than normal [14]. The low insulin sensitivity leads to moderately raised blood glucose levels, typically 10–15 mM. There are often no clear symptoms for years of type 2 progression, but the raised glucose levels nevertheless give rise to increased levels of protein glycations, i.e. covalent reactions between protein amino groups and the aldehyde function of glucose (Figure 7.2). Glycations proceed via Amadori rearrangement to form advanced glycation end-products (AGEs) [15–17], which include protein crosslinks and other plaque, all of which result in sedimentations in blood vessels and organ damage. Type 2 diabetes is often only diagnosed after years of progression, when late complications like heart or kidney problems appear.

O

OH

OH

HO

proteins

HO

OH

OH Glucose hemiacetal

OH

OH OH

Glucose

N

O

HO

HO

OH

protein

protein

HO

HO HO

OH

Amadori rearrangement

HO

O OH

OH Imine

Fructosamine

COOH protein

H N

N H

carboxymethyl amines

+

protein

N

N

N

protein

imidazolium cross-links

protein

N H

OH N

fluorescent pyrimidines

Figure 7.2 Protein glycation reactions and a few of the known end-products

DIABETES MANAGEMENT AND INSULIN ENGINEERING

253

Excessive body weight and unhealthy lifestyle are risk factors for development of type 2 diabetes. The prevalence of type 2 is dramatically increasing, especially in developing countries, due to adoption of a modern lifestyle with too much unhealthy food and a lack of exercise [18]. Type 2 is the more common form of diabetes, as 90–95 % of diabetes patients are type 2. The WHO has estimated that there are 170 million diabetes patients today, and this number will increase to 370 million by 2030, mainly due to increase in type 2 patients. The prevalence of type 1 diabetes is also increasing, but less dramatically than type 2. If diagnosed early, type 2 diabetes can often be treated by lifestyle adjustments. Various oral drugs are also available for type 2 treatment, such as insulin sensitizers (insulin sensitivity boosters) or b-cell provokers (insulin secretors) [19]. However, the extra workload on b-cells in type 2 patients often leads to b-cell failure after some years. In fact, early insulin treatment of type 2 patients is recommended in order to improve glucose control and spare the b-cells [20,21]. Type 2 diabetes is also known as non-insulin-dependent diabetes mellitus, NIDDM, but this term is something of a misnomer, since type 2 diabetes patients typically enter insulin treatment sooner or later. The needles required for insulin injections are a common fear in regard to starting insulin therapy. It is unfortunately not well recognized that use of modern needles is virtually painless [22]. Many adults remember their childhood vaccination injections as painful, but insulin injections use much smaller volumes, with much narrower needles, and insulin is available in handy pens and pumps. Injected insulin doses are placed under the skin (subcutaneously), not directly into the bloodstream. The clinical measure for an insulin dose is the ‘unit’, which equals 6 nmol (for most insulins) [23]. As a rule of thumb, a diabetes patient will need a daily number of insulin units roughly equal to their body weight in kilograms. Due to low insulin sensitivity, some type 2 patients use much more insulin. The standard drug concentration of insulin is 600 micromolar, so 1 unit equals 10 microlitres. The daily required distribution between basal and meal (bolus) insulin supplies is typically in the range of 30:70 to 50:50. Apart from the lowering of blood glucose, insulin has other biological activities. Insulin is a weak growth factor (mitogen) and shares some sequence similarity with insulin-like growth factor, IGF, of which there are two known variants [6,24]. A common side effect of insulin-based diabetes treatment is weight gain, particularly in type 2 diabetes. It is unclear what causes the weight gain, but the mitogenic effects of insulin may play a role, as well as effects on appetite due to rapidly dropping

254

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

blood glucose levels during treatment of high blood sugars. With regard to weight, the new glucagon-like peptide-1 incretin drugs (exenatide, alias Byetta, from Eli Lilly, and liraglutide, alias Victoza, from Novo Nordisk) appear very promising for type 2 treatment, since they not only help glucose control, but also affect weight loss [25]. As mentioned, tight glucose control is crucial for good diabetes treatment. Home blood glucose monitors were not widely available until around 1990, but blood glucose can now be measured with these handy devices [26]. They do however require fresh blood, and a finger prick is the most common source. The finger prick is in fact more painful than insulin injections. However, with the modern blood glucose monitors, only about 0.5 microlitres of blood is required, so the procedure is quite gentle. Unfortunately, many diabetes patients rarely test their blood sugar, despite this being the proven way to achieve good control. The pain, the hassle and perhaps the unpleasant facing of many non-ideal values are some of the reasons for neglect of blood glucose monitoring. The average blood glucose level over the last two to three months can be evaluated by measuring glycated haemoglobin (HbA1c) [27], a procedure which is typically carried out at the diabetes clinic. In a healthy person, HbA1c is in the range of 4–6% relative to total haemoglobin. Poorly controlled diabetes can lead to HbA1c values above 10%. HbA1c values below 7 or 6.5% have been recommended as targets for diabetes management [28]. With such low values, the risk of long-term complications has been shown to be significantly reduced in both type 1 and type 2 diabetes [2,3]. Accordingly, the main goal of diabetes management is to achieve blood glucose values near normal, ideally near 4–6 mM in the fasted state, and not much higher than 8–10 mM after a meal. However, too much insulin at the wrong time will lead to low blood sugar (hypoglycaemia), which can be acutely dangerous. Blood glucose in the range of 2–3 mM can make the patient feel tired, hungry and/or confused, and levels below 1–2 mM can lead to coma or death. The brain neurons are highly dependent on glucose and can die after short periods with too low glucose. Insulin treatment is therefore a narrow balance between too high and too low blood glucose (hyperglycaemia and hypoglycaemia) [29,30]. It is very difficult to reach normal values without also nearing the dangerous hypoglycaemic state at times. For this reason, patients and caretakers dose insulin cautiously, which results in overall higher levels than ideal. Some patients use the same timing and dosing of insulin shots every single day, but motivated patients, who are more careful with home blood glucose monitoring, can self-adjust their doses to the individual pattern of any given day, and this will generally provide better results.

DIABETES MANAGEMENT AND INSULIN ENGINEERING

255

In order to meet the discussed blood glucose goals, insulin must be supplied as a combination of basal and bolus (meal) administrations. Injected native insulin is cleared from the body within a few hours, so supplying basal insulin as drug over 24 hours is a problem. Pharmaceutical engineering of insulin for ensuring basal supply has come a long way, as will be discussed below, but the ideal basal insulin is yet to be invented. For meals, the injected insulin should take action rather quickly, as passage of meal components from the stomach to the blood occurs within minutes to a few hours. In order to ensure stability during storage, insulin drugs are administered as Zn(II) hexamers. The insulin hexamer solution is injected under the skin (subcutaneously), but the dissociation of insulin hexamer to dimers and monomers and their diffusion to the bloodstream takes some hours (Figure 7.3). For this reason, regular insulin must be taken 0.5–1 hour prior to a meal [31,32], and this timing can be difficult to handle under some circumstances, for instance at a restaurant. In order to allow insulin to be taken with a meal (not half an hour before), fast-acting insulins have been engineered, as will be discussed below.

Zn2+ Zn2+ Zn2+ Zn2+

K = 103

K = 105

Zn2+ Zn2+

Zn2+ Zn2+

crystals or soluble self-assemblies > 500 kDa

hexamers

dimers

monomers

> 10 hours

5 hours

2 hours

1 hour

Absorption rates t½

Figure 7.3 Insulin oligomer states with stability constants K and approximate absorption rates t½

Other engineered insulins which will be discussed below include insulins for improved uptake in alternative administrations (non-injection administrations), such as lung administration (pulmonary), and attempts at nasal or oral administration.

256

7.3

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

INSULIN STRUCTURE

Human insulin, 5808 Da, consists of 51 amino acid residues in two chains, with one intrachain disulfide and two interchain disulfides (Figure 7.4). The insulin structure is folded as three a-helixes and a b-strand, as has been documented by X-ray crystallography [33], as well as NMR [34,35] and other techniques (Figure 7.5). S H

S

G I V E Q C C T S I C S L Y Q L E N Y C N OH S

S S

S H

F V N Q H L C G S H L V E A L Y L V C G E R G F F Y T P K T OH

Figure 7.4 Human insulin sequence and disulfide pairings

B1

B1 T

R

c(phenol) > mM

Figure 7.5 Insulin T- and R-folds. The monomer structures were extracted from crystal structures 4IN and 2CTI. (see colour Plate 5)

There have been some suggestions that insulin undergoes a conformational change upon binding to its receptor [36]. Despite intense efforts, it has not yet been possible to crystallize and elucidate the structure of insulin in complex with its receptor [37–40]. Mutation studies like Ala-scan and so on have given some information about the binding epitopes of insulin [41,42]. The critical parts of the insulin structure with regard to receptor affinity, and the parts that can be modified without compromising the biological activity, are therefore quite well recognized. It has been suggested that insulin contacts its receptor via two separate epitopes [43,44]. Upon the extracellular binding of insulin to the insulin receptor, the intracellular domain of the insulin receptor is phosphorylated and a signal transduction occurs inside the cell, leading to

INSULIN STRUCTURE

257

translocation of glucose transporter proteins to the cell surface (GLUT4), resulting in glucose uptake and metabolism [45]. Native insulin forms homodimers at concentrations above approximately 10 micromolar [46]. In the pancreas, as well as in most commercial preparations, three insulin dimers are assembled to form a hexamer structure via HisB10-complexation with two Zn(II) ions [47]. Insulin has a tendency toward fibrillation [48], particularly if agitated, but the insulin hexamer provides some protection against fibrillation. Hexamer formulation is therefore preferred in commercial preparations in order to achieve sufficient transport and storage stability. Commercial insulin stability is typically two years at refrigeration and one month at room temperature. In order to preserve a sterile commercial insulin preparation [49], phenol and related substances are added at noticeably smelly concentrations (> 10 mM). Serendipitously, the phenol excipients also improve the hexamer insulin stability [50], and phenol above 1 mM induces a conformational shift in the N-terminal of the insulin B-chain from random coil to a-helix. The two types of folding of insulin are known as the T- and R-conformations, respectively (Figure 7.5) [51]. Insulin sequence is quite well conserved throughout the animal kingdom [52]. Until around 1980, all commercial insulins were extracted from animal pancreas, initially from cow (bovine insulin) and later mainly from pigs (porcine insulin). Human use of bovine insulin tends to cause immunogenic problems [53]. The difference between human insulin and pig insulin is only one residue, ThrB30 and AlaB30, respectively [54]. In the early 1980s, commercial human insulin was first introduced commercially by Novo. The transformation from pig to human insulin was performed by enzymatic exchange of AlaB30 with Thr using transpeptidation at industrial scale [55–57]. The transpeptidation utilizes the B29 lysine residue (the only lysine in the sequence) as a semisynthetic coupling point for transpeptidation with trypsin (which is reactive with Arg and Lys) or achromobactor lyticus protease (ALP), which is Lysspecific [58]. Starting from the 1980s, insulin production by extraction of animal pancreas has been gradually replaced by recombinant methods [59]. Virtually limitless amounts of insulin can now be produced by fermentation in high yields with no risk of contaminations with animal pathogens. Twochain insulin can be produced from refined single-chain precursors expressed in yeast or other microorganisms [60], and transformed in a process similar to the mentioned pig-to-human insulin transformation [61].

258

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

Importantly, the advent of recombinant access to insulin also made possible relatively easy production of insulin mutants, which, along with chemical modification and formulation, is the toolset for pharmaceutical insulin engineering.

7.4

PROLONGED-ACTING INSULIN SOLIDS

In the early days, basal supply of insulin over 24 hours was attempted via multiple daily injections, but this approach was inefficient and annoying, especially during sleep. Ironically, the early supplies of insulin from animal sources were quite impure, and high-molecular-weight protein impurities helped retain insulin at the site of injection. This helped achieve prolonged insulin action, but the impurities also gave other problems such as allergies and immunogenic reactions. Today, commercial insulin preparations are highly pure [62,63], and several novel mechanisms for prolongation of the action to basal coverage have been engineered. In the 1930s to 1940s, Nordisk Gentofte introduced prolonged-acting insulin via insulin co-crystallization with protamine [64,65]. The product name is NPH (Neutral Protamine Hagedorn) or isophane. Protamine is a basic (cationic) protein isolated from fish sperm. Insulin is an anionic peptide with overall net charge of 4 at physiological pH. Insulin– protamine co-crystals dissolve slowly upon injection in human subcutis, and thereby provide basal insulin supply for approximately 12 hours. NPH crystals can be mixed with regular insulin, and such mixtures can provide both basal and bolus coverage in one injection. Injection of a mixture of NPH and regular insulin before breakfast and dinner is a widespread insulin regiment still used today. Several NPH:insulin mix ratios are commercially available, with 30:70 as the most common mix. The described morning-plus-evening regimen is easy to adhere to, providing good patient compliance, but it lacks proper coverage for lunch and snacks. NPH-mix insulins were recently upgraded to include engineered fast-acting insulins as the bolus component, thus allowing injection of the mix with the meal, not half an hour before the meal [66]. NPH insulin is still widely used today, but it suffers from several problems. Since NPH is crystalline, it is supplied as a suspension, which must be carefully resuspended before each injection. Even with the most careful resuspension (gentle turning and rolling of the vial), the heterogeneity of the preparation can result in large differences in insulin dosings between each injection. NPH dosing can be compared to serving ice cubes

PROLONGED-ACTING INSULIN SOLUTIONS

259

from a can. The rate of absorption from the compact crystalline depot can also lead to large variations between doses [67]. Furthermore, the appearance of basal insulin in the blood from NPH formulation is not as flat as desired. NPH gives a pronounced peak after 4–6 hours [68,69], and this can lead to dangerously low blood sugar, for instance at night following the evening injection [70]. In the 1930s, Hoechst introduced Surfen insulin, which was based on acidic co-formulation of 1,3-bis(4-amino-2-methyl-6-quinolyl) urea and insulin, which co-precipitated upon injection. The urea additive has been implicated in allergic reactions to Surfen insulin [71,72], and the product is no longer available. In the 1940s to 1950s, Novo introduced porcine and bovine insulin crystallized with excess Zn(II) [73–75], known as the Lente insulins. These preparation provide long basal coverage, but they cannot be mixed with regular insulin, because the excess Zn(II) precipitates and delays the effect of the regular insulin. Like NPH, Lente insulins are crystalline and must be resuspended before injection, and this leads to the problems with tricky reproducibility of day-to-day dosings. Lente insulin is no longer on the market.

7.5

PROLONGED-ACTING INSULIN SOLUTIONS

Because of the dosing problems with insulin suspensions, soluble prolonged-acting insulins have been desired by diabetes clinicians for many years. In the 1980s, scientists at Novo and Hoechst initiated projects with insulin variants of altered isoelectric points [76–80]. The isoelectric point (pI) of native insulin is 5.4, and native insulin is therefore net uncharged and relatively insoluble at pH 5.4. By introducing positively-charged residues or blocking or removing negative charges from native insulin, scientists altered the pI of insulin toward neutral pH. Thereby, insulin variants were created that were soluble at slight acidic pH, but precipitated at physiological pH 7.5. Accordingly, such insulins could be dosed as solutions, but precipitated in the tissue upon injection and neutralization to form prolonged-acting depots. Since the dosed solutions of such pI-altered insulins are acidic, some patients experience pain or sting upon injections, but patients have different sensitivities to this issue. GlyA21 ArgB27 ThrB30-amide human insulin (NovoSol) was clinically investigated by Novo as a pI-altered basal insulin [81,82], but development was stopped because of low bioavailability and local irritations.

260

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

Hoechst (now part of Sanofi-Aventis) successfully developed HOE901, alias insulin glargine, alias GlyA21 ArgB31 ArgB32 human insulin, as a pI-shifted prolonged-acting insulin [83,84]. The trademark is Lantus. The glargine preparation provides flat and long basal coverage. Once-daily dosing is most commonly used, but some patients require twice-daily dosing for full coverage of 24 hours. Since the glargine preparation is acidic (pH 4), glargine cannot be mixed with regular insulin [85], because regular insulin tends to precipitate at pH 4. Furthermore, the AsnA21 residue of regular insulin is unstable toward deamidation at acidic pH, so the mix would not be stable during storage. In glargine, the instability of AsnA21 is fixed by mutation of the A21 residue to glycine. GlyA21 cannot deamidate, but preserves the biological activity of insulin. The A21Gly mutation itself was originally invented at Novo [82]. Glargine displays moderately increased affinity toward the IGF receptor (growth factor) compared to native insulin, and it has been discussed whether this observation has clinical implications such as cancer risk or increased retinopathy (eye disease) [86,87]. In June 2009, the journal of the European Association for Study of Diabetes, Diabetologia, published the online versions of four papers which suggests a possible link between insulin glargine and cancer. It was speculated that the increased affinity of glargine for the IGF-1 receptor was involved. Reference: http://www.diabetologia-journal.org/cancer.html (accessed June 30 2009). Recently, Eli Lilly has published design and preclinical studies of novel pI-shifted insulins based on ArgA0 mutations, among other things. Such insulins display normal or decreased IGF receptor affinity compared to native insulin [88], but these insulins appear not to have been further developed. In the 1980s, proinsulin and related sequences [89] were investigated by Eli Lilly as prolonged-acting soluble insulins, but the profiles were not as long as desired. In the 1990s, Co(III) insulin hexamer was investigated by Novo Nordisk [90], but the increased stability of this hexamer did not provide sufficiently prolonged action [91]. Various polymer formulations (encapsulations) [92–94] of insulin have been pursued for prolonged action, but none have reached the market. Stability of insulin in the polymer environments is a problem for this approach, as the insulin hexamer structure is difficult to maintain in the polymer formulation. A fundamental problem with polymer formulations is the burst effect, i.e. release of too much drug in the

PROLONGED-ACTING INSULIN SOLUTIONS

261

initial stages upon polymer injection [95]. Burst effect can be tolerable for some drugs, but for insulin this phenomenon can easily lead to dangerously low blood sugar. In the mid-1990s, Eli Lilly and Novo Nordisk separately reported on the development of fatty-acid-acylated insulins as soluble prolonged-acting drugs [96–100]. The epsilon-amino group of LysB29, Ne, was used as a handle for insulin derivatization. Insulin has three amino groups, but Ne is significantly more basic than the two Na-terminals [101,102]. Ne is hence more nucleophilic than Na above pH values where Ne is deprotonated, and Ne can therefore be selectively acylated at pH > 10 (Figure 7.6). pKa 8.6

NH2

NH2 S

S

I V EQCC T S I C S L YQ L E N Y CN

Ph

S

O

pKa 11.2

S

OH S

S

H2N

V N Q H L C G S H L V E A L Y L V CG E R G F F Y T P N

OH O

O

pKa 6.8

acylation reagents, pH > 10 O R

NH2 S

NH

S

I V E Q C C T S I C S L Y Q L E N Y C N OH

Ph O H2N

S

S

S

S

V N Q H L C G S H L V E A L Y L V CG E R G F F Y T P N O

OH O

Figure 7.6 Lysine-Ne selective acylation at pH > 10. With kind permission from Springer Science and Business Media

The Eli Lilly compound was B29Ne-palmityl human insulin (hexadecanoyl, C16) and the Novo Nordisk compound was B29Ne-myristyl desB30 human insulin (tetradecanoyl, C14). The Eli Lilly compound W99-S-32 was not fully developed, apparently because of too low potency in humans. The low potency may be related to the high hydrophobicity of the compound, which in turn could lead to high unspecific clearance. The Novo Nordisk compound NN304, alias insulin detemir, alias Levemir, was also found to be of relatively low

262

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

potency in humans (15–30 %) [103], but this problem was corrected by adjusting the dose and redefining the unit for this particular insulin. Levemir was fully developed and market-introduced in 2005. Levemir provides a flat and reproducible profile with coverage for up to 24 hours. Once-daily dosing is the most common regime [104], but some patients need twice-daily dosing for full basal coverage. The mechanism of protraction is based on binding to albumin [105–108], an endogenous protein of 66 kDa, which is present in both subcutis and the circulation. Increased oligomerization of acylated insulin, and binding of acylated insulin dimer to two albumins have been suggested to participate in the prolonged drug residence in subcutis [109,110]. Detemir–albumin binding in the circulation advantageously provides an insulin-buffering effect, which seemingly protects against insulin spikes, which could otherwise lead to hypoglycaemia [111,112]. Detemir is based on desB30 human insulin, because acylated desB30 insulin gave stronger binding to albumin compared to full-chain insulin [96]. Levemir cannot be mixed with regular insulin, because the fast-acting insulin would be partially retained in the Levemir depot (blunting of individual profiles by formation of mixed hexamers). Initiation of insulin therapy typically leads to weight gain, especially in type 2 diabetes. However, Levemir turns out be weight-neutral, i.e. patients retain their weight after initiation of Levemir therapy [113]. It is not fully understood why this is the case [114], but the increased hydrophobicity of the myristyl insulin relative to other insulins or the albumin-binding property could be partly involved. Alternatively, since Levemir provides a more predictable profile relative to other basal insulins, the generally more stable blood sugar may lead to less influence on appetite and thus less snacking by patients. Apart from fatty acids, other albumin-binding ligands have been investigated in pursuit of protraction of soluble insulin. Fatty diacids have been used, and these can provide very long insulin effect due to increased albumin binding [109,115], but in some cases these compounds seem to suffer from low potency [115]. Bile acid derivatives of insulin bind to albumin, but these insulins are in some cases longer acting than should be expected from the albumin binding alone. Studies by size-exclusion chromatography (SEC), among other methods, have revealed that bile acid insulins can form high-molecularweight self-assemblies (multihexamers) at the site of injection [116,117]. As long as phenol is present (in the vial), the insulin is in R-fold, and as shown by SEC, the R-fold does not give rise to the high-molecular-weight

PROLONGED-ACTING INSULIN SOLUTIONS

263

Figure 7.7 Glucose infusion rate profiles obtained in euglycaemic glucose clamp experiments in pigs after s.c. injection. Eight pigs received NN344 432 nmol per animal, glucose level 4.3 – 0.3 mmol mM. Seven pigs received NPH insulin 148 nmol per animal, glucose level 4.4 – 0.3 mmol mM. Five pigs received NPH insulin 216 nmol per animal, glucose level 4.4 – 0.1 mM. With kind permission from Springer Science and Business Media

self-assemblies. R-fold keeps insulin in a storage stable form, which does not self-assemble above dihexamer (dodecamer). Upon injection in the subcutis, the small amphiphilic phenol additive quickly diffuses away from the injection site. Insulin now adopts the T-fold. Some bile acid derivatives of insulin when in T-form transform into a soluble highmolecular-weight self-assembled state, as shown by SEC [116]. Such soluble multihexamer complexes can provide prolonged insulin action for more than 24 hours (Figure 7.7). NN344, alias B29Ne-lithocholyl-gglutamyl desB30 human insulin, is currently in clinical development by Novo Nordisk. A related structure, NN377, alias B29Ne-lithocholyl desB30 human insulin (same structure as NN344 except for the g-glutamyl linker), has been structurally characterized by X-ray crystallography (Figures 7.8 and 7.9) [118]. The crystal structure shows how the lithocholyl residues ‘hook’ different insulin hexamers together via binding to pockets in neighbouring hexamers. The cross-hexamer binding includes hydrophopic packing against a proline residue and a hydrogen bond between hydroxy groups of the bile acid and of TyrB26. A related interaction may well exist for NN344 and other bile acid insulins in solution and this could explain the formation of soluble high-molecular-weight selfassemblies (multihexamers) and thus the prolonged action. An attractive (but expensive) alternative to engineered basal insulins is the programmed delivery of insulin in basal and bolus patterns by using mechanical insulin pumps. Fast-acting insulins are preferred in pumps; see below for further details.

264

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES OH H N

O

O

OH

O

B29Nε-desB30 human insulin

B29Nε-desB30 human insulin O

OH

Figure 7.8 Structures of NN344 B29Ne-lithocholyl-g-glutamyl desB30 human insulin and 0377 B29Ne-lithocholyl desB30 human insulin

Figure 7.9 The 0377 3D structure (1UZ9) showing hexamer–hexamer interactions. The zoomed box shows the cross-hexamer binding of the B29Ne-lithocholyl residue in a pocket in the opposing hexamer, lined by among others ProB28 and TyrB26. A phenyl– phenyl stacking interaction is clear between B1s from opposing hexamers. The residual lithocholyl residues that seemingly protrude into open space are in fact bound to neighbouring hexamers, which were removed for clarity (see colour Plate 6)

FAST-ACTING INSULINS

7.6

265

FAST-ACTING INSULINS

For reasons of storage stability, commercial insulins are administered as Zn(II) hexamers [119]. The high stability of the native insulin hexamers, like pig and human insulin hexamers, results in relatively slow dissociation and absorption of insulin upon injection in the subcutaneous tissue. Native insulins must therefore be dosed ½–1 hour before a meal in order to match the increase in blood glucose after a meal [31,32]. This timing can be difficult to get right. Scientists have worked on this problem by engineering insulins which form less stable hexamers, and thus provide faster absorption from the subcutaneous tissue. The first results were published by Novo in the mid-1980s [120]. By examining the insulin dimer and hexamer interfaces in insulin crystal structures and inserting charge repulsions or steric repulsions at strategic positions (Figure 7.10), it was shown to be possible to design insulins that form less stable oligomers. The design process excluded the residues putatively involved in insulin–insulin receptor contacts, in order to seek preservation of the biological activity of insulin. Clinical investigations confirmed faster onset of action with the oligomer-destabilized insulins [121,122]. The AspB10 analogue, alias X10, was one such mutant displaying weaker oligomer stability, and this analogue was brought to clinical trials. Unfortunately, animal toxicology studies of AspB10 human insulin showed formation of tumours [123], and the development of X10 was stopped. The toxicology of AspB10 has been suggested to reside with its increased affinity for the IGF receptor [124–126]. Meanwhile, Eli Lilly had started development of their version of hexamer-destabilized fast-acting insulin. Lilly went through clinical development faster, and therefore became the first to reach the market with their variant, LysB28 ProB29 human insulin, alias insulin lyspro, alias Humalog, in 1996. The insulin lyspro sequence was designed by considering the C-terminal part of the B-chain of IGF-1, inspired by the observation that IGF-1 does not form stable hexamers [127]. Novo Nordisk managed to find an alternative to AspB10, namely AspB28 human insulin, alias X14, alias insulin aspart, alias NovoRapid, alias Novolog, which passed the clinical trials and reached the market in 1999. Although insulin lyspro and insulin aspart are often titillated monomeric insulins, they are in fact Zn(II) hexamers in the storage vials (crucial for storage-stability reasons), but the hexamers are obviously destabilized compared to regular insulin.

266

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

Figure 7.10 3D structure (1ZEG) of the dimer of AspB28 human insulin, illustrating the engineered destabilization of the dimer interface by charge repulsion between the engineered AspB28 (represented with spheres) with the native residues B21Glu and A4Glu. Natively, B28 is proline. Copyright 2005 (see colour Plate 7)

Recently, Sanofi-Aventis has introduced their version of fast-acting insulin, LysB3 GluB29 human insulin, alias glulisine, alias Apidra [128]. In contrast to the other two fast-acting insulins, glulisine is not administered as a weakened hexamer, but is formulated without zinc(II), i.e as insulin dimer/ monomer. The usual storage-stability problems with nonhexamer insulins have been overcome for glulisine by formulating the drug with detergent. Notably, glulisine has been described as activating the downstream cascade from the insulin receptor differently than other insulins (different insulin receptor substrate activations, IRS-1 and IRS-2) [129]. Some protection of b-cell function is speculated in this context, but any clinical outcome of this altered biochemistry is not yet apparent. Furthermore, the alleged IRS-1/IRS-2 selectivity has since been questioned [130].

GLUCOSE-SENSITIVE INSULIN PREPARATIONS

267

Various insulin derivatives, such as glycosylated insulins [131,132], have been investigated in search of fast-acting insulin, but none appear to have reached clinical trials. Apart from their use as bolus insulins, the fast-acting insulins are also recommended for use in insulin pumps (continuous subcutaneous insulin infusion, CSII) [133]. An insulin pump can slowly and steadily supply insulin in a programmed basal pattern, accompanied by manual entering of bolus doses into the pump computer, as needed to cover meals and snacks. Modern insulin pumps are roughly the size of a mobile phone. The insulin is delivered to the body via a tube and a catheter inserted on the stomach or hip. The catheter must be replaced every three days or inflammation becomes a problem. A common annoyance with insulin pump therapy is precipitation of insulin in the tubing, which can lead to failed delivery. Insulin aspart (NovoRapid) is the most soluble of the commercially available fast-acting insulins, because of the extra negative charge from the Asp residue. Insulin aspart has therefore been recommended as the pump insulin of choice [134].

7.7

GLUCOSE-SENSITIVE INSULIN PREPARATIONS

Blood glucose concentration is influenced by many factors, and diabetes patients often experience unexpected and unpredictable fluctuations in blood glucose. This problem makes it difficult to always predict the right dosing and timing of insulin administrations. Against this background, scientists have devised glucose-dependent insulin release systems in pursuit of autonomous regulation of insulin in synchronization with blood glucose. Various polymers with glucose-binding motifs have been prepared, which can bind or entrap insulin or insulin derivatives in the subcutaneous tissue and release the drug in a glucose-dependent fashion [135–141]. The glucosebinding motifs in these polymers have been large proteins like concanavalin A, or small-molecule carbohydrate binders such as aryl boronic acids. Although the research area of glucose-controlled insulin release has been active since the late 1970s, none of the systems have apparently matured to reach clinical trials. Problems with the polymer approaches include, among other things: 1. Larger volume of drug when an external polymer is needed. A large volume can lead to unmanageable vials and pens and to unacceptable pain at the injection site.

268

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

2. Instability of insulin in the polymer environment (nonhexamer insulin). 3. Burst effect (too much insulin released in the early stages). 4. Time lag of glucose transport from blood to subcutis [142], and of insulin transport from subcutis to blood. The time lag of glucose from blood to the subcutis is only about 10 minutes, but the lag of released insulin from subcutis to blood is approximately 1 hour. 5. The insulin bolus release from healthy pancreas is amplified 6–10 times over the basal level relative to any given rise in blood glucose. Unless a strong amplification can be build into the artificial release systems, not nearly enough insulin will be released to cover steep rises in blood glucose after a meal. Still, soft glucose-dependent adjustments of basal insulin release could be better than no adjustments. A recent effort in the field of glucose-controlled insulin release used high-molecular-weight insulin self-assemblies based on reversible boronic acid–polyol crosslinks between insulin hexamers (Figures 7.11 and 7.12) [143]. The advantage of this approach is that no external polymer is needed, as insulin reversibly self-assembles (polymerizes) with itself. Insulin can be released from the soluble boronate–polyol insulin multihexamers by treatment with carbohydrates (Figure 7.13).

insulin:

polyol:

boronate:

carbohydrate:

Figure 7.11 Illustration of carbohydrate-controlled self-assembly of boronate– polyol insulin. Reprinted with permission from the American Chemical Society. Copyright 2005

A long-standing goal in diabetes treatment has been the artificial -cell or closed-loop CSII, i.e. a mechanical insulin pump controlled by a continuous glucose sensor [144,145]. Insulin pumps are today flexible and handy and can be adjusted on the fly. They are however quite

GLUCOSE-SENSITIVE INSULIN PREPARATIONS

4

HN OH

5

HN Ac2O

Boc2O

OH

HO

O

O

O

O H2N HO

269

HO

MeOH-water

OOCMe

OH

MeCOO

pyridine OH

HO

OOCMe

MeCOO

HO

MeCOO

HO 94 %

100 %

O

O

O CF3COO–

H3N+ TFA

6

OOCMe MeCOO

OOCMe

MeCOO

93 % O O B

O

OH

O N CF3COO–

O

O

O

O 2N

O

O

8

HN

O B

MeCOO

OOCMe

N H NO2

OOCMe

O 9 HN OOCMe

MeCOO

OOCMe

MeCOO

MeCOO

82 %

N O

O

O

O B

O

O 1. DesB30 insulin, H O/acetonitrile

O N H

O

10

OH HO

B29Nε desB30 human insulin

O

B

N H

2. NaOH

O HN

HN

DCM NO2

MeCOO

63 %

O

DCC, HOSu

OH

O

DIEA, DCM MeCOO

TFA

OOCMe

MeCOO

78 %

H3N+

7

HN

MeCOO

MeCOO

O

O

N H

HATU, DIEA, DCM

OOCMe

MeCOO

O

Boc-Glu( Bu)

MeCOO MeCOO

OOCMe OOCMe

NO2

OH

HO HO

OH

HO

98 % MeCOO

Figure 7.12 Synthetic route to boronate–polyol insulin. Reprinted with permission from the American Chemical Society. Copyright 2005

expensive. But the main problem with the artificial pancreas concept lies with the reliability of available continuous glucose sensors. Despite many years of development, current sensors are not of sufficient precision and reliability for it to be safe to let them control an insulin pump. Continuous glucose sensors based on small electrodes inserted in the skin are today marketed by Medtronic, Dexcom and soon Abbott. The recognition element used in the electrodes and handheld glucose monitors are based on enzymes like glucose oxidase [146], which is very specific for glucose. However, it is an inherent problem with these invasive sensors that the tissue around the foreign object soon begins the healing process, where the local glucose concentration is no longer representative for the body.

270

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES OD 220 nm

(a)

Ferritin 500 kDa Insulin hexamer 36 kDa

Insulin monomer 6 kDa void

(b) Insulin 1, no phenol in SEC eluent

(c) Insulin 1, 4 mM phenol in SEC eluent

(d) Insulin 1, 4 mM phenol + 50 mM sorbitol

4

8

12

16 mins

Figure 7.13 Size-exclusion chromatographic analysis showing high-molecularweight soluble self-assembly of boronate–polyol insulin in presence of phenol (c), and the carbohydrate-controlled release of insulin from the self-assembly (d). Reprinted with permission from the American Chemical Society. Coyright 2005

ALTERNATIVE INSULIN DELIVERY

271

Inflammation will rather quickly initiate at the site of foreign-object insertion [147], and the current sensors must only be used for three days. They must furthermore be calibrated against two to three daily finger pricks. Many scientists have attempted to develop non-invasive blood glucose monitoring by using infrared spectroscopy, among other things, but so far with limited success [148]. Glucose is difficult to detect and distinguish from other substances in the human tissue. Semi-invasive methods like glucose monitoring using contact lenses or optic fibres are under investigation [149,150].

7.8

ALTERNATIVE INSULIN DELIVERY

Injection is not the most convenient form of drug administration and many people think that insulin injections are painful, although this is not true with modern needles, as discussed previously. Against this background, scientists have investigated alternative administration forms including various engineerings of the insulin molecule for better stability of insulin in proteolytic environments or for better uptake through biological barriers. Oral delivery of native insulin leads to very low bioavailability due to degradation in the gut by acid and by proteases, and due to the generally very low gut absorption of large molecules [151]. Insulin PEGylation (derivatization of insulin with functionalized polyethylenglycol) has been claimed to provide better oral uptake of insulin, and Nobex/Biocon currently conduct oral clinical trials of insulin derivatized by short-chain alkyl-PEGylation [152]. Despite improvements over native insulin, bioavailability is still very low. Furthermore, the influence on insulin absorption of different stomach content from day to day will make predictable and reliable dosing of oral insulin a significant problem. Pfizer launched pulmonary insulin, Exubera [153], in 2006, based on an insulin powder inhaler. The bioavailability of Exubera is low (10%), but this issue could be adjusted via dosing, and glucose control comparable to that of injectable insulin can be reached with pulmonary insulin [154]. Notably, Exubera is based on human insulin, which does not provide basal coverage, so basal insulins must be added via injections.

272

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

The penetration of Exubera into clinics was very slow, and Pfizer dramatically withdrew the product after only one year on the market (in October 2007), with multi-billion dollar losses. Nevertheless, other companies are proceeding toward advanced clinical stages with various devices for pulmonary insulin delivery [155]. Some clinicians hesitate to use pulmonary insulin due to fear of adverse long-term effects. Indeed, the lung is a sensitive and vital organ. Higher levels of induced insulin antibodies have been reported with pulmonary insulin compared to injected insulin [153], but no dramatic clinical outcomes have yet turned up. Notably, tobacco smokers uptake pulmonary insulin better than nonsmokers [153]. Other attempts at alternative insulin delivery include buccal (via the mouth mucous) [156], nasal (via the nose) and transdermal (across the skin) delivery, but these approaches have typically included various absorption enhancers (detergents), which open the mucous tissue. The enhancer approach is therefore problematic for chronic use. A buccal insulin formulation, Oralin, has been marketed by Generex in Ecuador since 2005 [157], but the drug is so far only approved in a few South American countries.

7.9

INSULIN MIMETICS

As mentioned above, oral delivery of insulin proceeds with very low bioavailability. However, if a small-molecule insulin mimetic could be identified, it could provide a route to oral diabetes therapy [158,159]. There are many claims of discovery of insulin mimetics present in the literature, but the term mimetic is often misused. A true insulin mimetic should bind and activate (i.e. phosphorylate) the insulin receptor in the same manner as insulin. There are other ways to decrease blood sugar than via activation of the insulin receptor, but such routes may lead to adverse effects. For example, vanadates are known to lower blood sugar, not via binding to the insulin receptor, but via inhibition of phosphatases downstream of the insulin receptor [160]. There are many other crucial phosphatases in the body and vanadates are not sufficiently specific for the insulin pathway to be of clinical use in diabetes. Merck published a high-profile paper in 1999 on a set of ‘insulin mimetic’ asterriquinones [161–164], but the mechanism of action was unclear, and the compounds have apparently not entered clinical trials.

PUSHING THE LIMITS OF INSULIN ENGINEERING

273

Notably, the central Merck compound was already known for several other biological activities [165,166], but these details were left out of the original Merck paper. In 2002, Novo Nordisk and DGI Biotech reported on insulin mimetic peptides identified from screening of phage display peptide libraries (peptides expressed on the surface of viral phages) [167]. Two different motifs for each of the alleged receptor-binding epitopes were identified and subsequently linked, to provide fully-active insulin agonists (and antagonists), although of quite long sequences [168]. Testing in vivo showed a correlation between length of peptide sequence and bioavailability.

7.10 PUSHING THE LIMITS OF INSULIN ENGINEERING The currently marketed insulins are produced by recombinant methods, in some cases supplemented by chemical transformations. Recombinant methods are traditionally limited to the 20 genetically-encoded amino acids. Recently, novel recombinant methods have been developed in which stop codons are used as codes for the introduction of unnatural amino acids in recombinant sequences [169]. However, these methods have apparently not yet been applied to insulin or matured to industrial scale. Various chemical methods are known for the addition or exchange of amino acids with unnatural residues in the terminals of native insulin [7], but these methods can only be used in a very limited number of positions. Full chemical synthesis of insulin has been described by a couple of methods, but the chemistry is difficult and low-yielding [170]. The relatively long sequences of insulin near the limits of modern peptide synthesis, but the worst problem is the required pairing of three disulfide bridges. Recent efforts to improve synthetic insulin disulfide pairing include orthogonal protections and directed pairings of each disulfide bridge [171,172], as well as synthetic use of a solubilizing extension accompanied by artificial C-peptide to direct the correct folding and disulfide formation [173], followed by enzymatic processing to twochain insulin (Figure 7.14) [174]. However, the yields of these methods are still low and neither seems ready for industrial-scale production.

274

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES SH

SH

G I V E QC C T S I C S L Y Q L E N Y C N SH SH

H

E EEK

pH 9.5

SH

F V NQH L CG S H L V E A L Y L V CG E RG F F Y T P K

S

S

EWK

S

G I V E QC C T S I C S L Y Q L E N Y C N S S

H

ALP

S

F V NQH L CG S H L V E A L Y L V CG E RG F F Y T P K

S

H

OH

S

S

E EEK

EWK

S

G I V E QC C T S I C S L Y Q L E N Y C N

H

OH

SH

OH

S S

F V NQH L CG S H L V E A L Y L V CG E RG F F Y T P K

OH

Figure 7.14 Folding of synthetic insulin via solubilizing extension EEEK and folddirecting C-peptide EWK. Both auxiliaries are removed by ALP-treatment (enzymatic cleavage C-terminally of K)

7.11 CONCLUSION After 50 years of virtually no change in the clinically-available assortment of insulin variants, the last 10 years have witnessed the commercial introduction of five engineered modern insulins, which have been quickly adopted in diabetes practice. The modern insulins have helped the management of diabetes, but there is still room for improvement. Novo Nordisk has recently announced insulin NN5401 as a novel insulin undergoing clinical trials. A number of possibilities for insulin engineering present themselves, for example prolonged-acting pulmonary insulin [175], orally deliverable insulin [151] or, as discussed above, glucose-sensitive insulin displaying autonomous adjustment to glucose fluctuations. Eventually, insulin treatment may even become obsolete, for instance if pancreas transplantations become successful in full scale [176]. Despite recent success with the Edmonton protocol [177,178], transplanted human pancreas is at best only sustainable for some years. Also, the number of donors required for treatment of all diabetes patients is far from available. Discouragingly, pancreas transplantations must

REFERENCES

275

furthermore be accompanied by chronic treatment with immunosuppressive drugs – not a good trade for insulin. Transplantation of animal pancreas to humans is used in a Mexican clinic [179], but animal organs will likely never be accepted in Europe or the USA [180]. Other future scenarios could include diabetes treatment via engineered insulin-secreting cells [181] or transformation of stem cells to b-cells [182]. However, for type 1 diabetes, the autoimmune destruction of b-cells would also have to be overcome. For type 2 diabetes, low insulin sensitivity in the peripheral tissue is more of a problem than lack of b-cells, so cell therapy may not be viable for type 2. Notably, the numerous principles invented for insulin delivery modulation could be better exploited as general tools for peptide drug engineering. Peptide activity prolongation by subcutaneous precipitation via modulation of the peptide pI point should be a generally applicable principle, but this method does not seem to have been exploited outside of insulin drugs. On the other hand, prolongation of peptide bioactivity by fatty acid acylation has been successfully adopted to other peptides, such as GLP-1 in the form of Liraglutide, among others [183,184]. Only time will tell where insulin and general peptide and protein engineering will take us. Stay tuned.

REFERENCES [1] F. Banting, C. Best, J. Collip, W. Campbell and A. Fletcher, Pancreatic extracts in the treatment of diabetes mellitus, Can. Med. Assoc. J., 12, 141–146 (1922). [2] The Diabetes Control and Complications Trial Research Group, The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus, N. Engl. J. Med., 329, 977–986 (1993). [3] THE UKPDS Study Group, Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33), Lancet, 352, 837–853 (1998). [4] K. S. Polonsky, B. D. Given and C. E. Van, Twenty-four-hour profiles and pulsatile patterns of insulin secretion in normal and obese subjects, J. Clin. Inv., 81, 442–448 (1988). [5] P. Dandona, D. Hooke and J. Bell, Exercise and insulin absorption from subcutaneous tissue, Br. Med. J., 1, 479–480 (1978). [6] P. De Meyts and J. Whittaker, Structural biology of insulin and IGF1 receptors: Implications for drug design, Nature Rev. Drug Disc., 1, 769–783 (2002). [7] J. P. Mayer, F. Zhang and R. D. DiMarchi, Insulin structure and function, Pept. Sci., 5, 687–713 (2007).

276

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

[8] S. Heller, P. Kozlovski and P. Kurtzhals, Insulin’s 85th anniversary – An enduring medical miracle, Diab. Res. Clin. Pract., 78, 149–158 (2007). [9] J. C. Henquin, Triggering and amplifying pathways of regulation of insulin secretion by glucose, Diabetes, 49, 1751–1760 (2000). [10] J. J. Holst, Glucagon-like peptide-1: From extract to agent. The Claude Bernard Lecture, 2005, Diabetologia, 49, 253–260 (2006). [11] D. F. Steiner, Proinsulin and the biosynthesis of insulin, N. Engl. J. Med., 280, 1106–1113 (1969). [12] A. Luyckx, The role of glucagon in hyperglycemia. A review, Diabetes Metab., 1, 201–208 (1975). [13] H. Peng and W. Hagopian, Environmental factors in the development of Type 1 diabetes, Rev. Endocr. Metab. Disord., 7, 149–162 (2006). [14] Z. T. Bloomgarden, Insulin resistance: Current concepts, Clin. Ther., 20, 216–231 (1998). [15] N. Ahmed and P. J. Thornalley, Advanced glycation endproducts: What is their relevance to diabetic complications?, Diab. Obes. Metab., 9, 233–245 (2007). [16] M. Brownlee, Biochemistry and molecular cell biology of diabetic complications, Nature, 414, 813–820 (2001). [17] P. Ulrich and A. Cerami, Protein glycation, diabetes, and aging, Rec. Progr. Horm. Res., 56, 1–22 (2001). [18] R. Sicree and J. Shaw, Type 2 diabetes: An epidemic or not, and why it is happening, Diab. Metab. Syndr. Clin. Res. Rev., 1, 75–81 (2007). [19] P. J. Guillausseau, Influence of oral antidiabetic drugs compliance on metabolic control in type 2 diabetes. A survey in general practice, Diabetes Metab., 29, 79–81 (2003). [20] S. Heinzl, Early insulin therapy in diabetis mellitus lowers the cardiovascular risk, Med. Monats. Pharm., 30, 160–161 (2007). [21] A. Liebl, The when and how of insulin initiation in type 2 diabetes, Curr. Med. Res. Opin., 23, 129–132 (2007). [22] L. Arendt-Nielsen, H. Egekvist and P. Bjerring, Pain following controlled cutaneous insertion of needles with different diameters, Somatos. Mot. Res., 23, 37–43 (2006). [23] J. C. Scheller and J. A. Galloway, The development of the insulin unit, Am. J. Pharm. Sci. Supp. Publ. Health, 147, 29–32 (1975). [24] A. Bayes-Genis, C. A. Conover and R. S. Schwartz, The insulin-like growth factor axis – A review of atherosclerosis and restenosis, Circ. Res., 86, 125–130 (2000). [25] E. Bonora, Antidiabetic medications in overweight/obese patients with type 2 diabetes: Drawbacks of current drugs and potential advantages of incretin-based treatment on body weight, Int. J. Clin. Pract., 61, 19–28 (2007). [26] Portable blood glucose monitors, Health Devices, 29, 201–232 (2000). [27] M. Procopiou, HbA1c: Review and recent developments, Rev. Med. Suis., 2, 1473–1476 (2006). [28] J. H. Simmons, K. K. Mcfann, A. C. Brown, A. Rewers, E. Cruz and G. J. Klingensmith, Achieving pediatric ADA HbA1c goals: Insulin pump therapy vs injection therapy, Diabetes, 55, A413 (2006). [29] H. S. Seltzer, Drug-induced hypoglycemia. A review of 1418 cases, Endo. Metab. Clin. N. Am., 18, 163–183 (1989). [30] G. H. Murata, W. C. Duckworth, R. M. Hoffman, C. S. Wendel, M. J. Mohler and J. H. Shah, Hypoglycemia in type 2 diabetes: A critical review, Biomed. Pharm., 58, 551–559 (2004).

REFERENCES

277

[31] E. W. Kraegen, D. J. Chisholm and M. E. McNamara, Timing of insulin delivery with meals, Horm. Metab. Res, 13, 365–367 (1981). [32] G. D. Dimitriadis and J. E. Gerich, Importance of timing of preprandial subcutaneous insulin administration in the management of diabetes mellitus, Diabetes Care, 6, 374–377 (1983). [33] T. Blundell, G. Dodson, D. Hodgkin and D. Mercola, Insulin: The structure in the crystal and its reflection in chemistry and biology, Adv. Prot. Chem., 26, 279–402 (1972). [34] S. Ludvigsen, M. Roy, H. Thøgersen and N. C. Kaarsholm, High-resolution structure of an engineered biologically potent insulin monomer, B16 Tyr—>His, as determined by nuclear magnetic resonance spectroscopy, Biochemistry, 33, 7998–8006 (1994). [35] H. B. Olsen, S. Ludvigsen and N. C. Kaarsholm, Solution structure of an engineered insulin monomer at neutral pH, Biochemistry, 35, 8836–8845 (1996). [36] H. Q. Xin, S. E. Shoelson, M. Kochoyan and M. A. Weiss, Receptor binding redefined by a structural switch in a mutant human insulin, Nature, 354, 238–241 (1991). [37] N. M. McKern, M. C. Lawrence, V. A. Streltsov, M. Z. Lou, T. E. Adams, G. O. Lovrecz, T. C. Elleman, K. M. Richards, J. D. Bentley, P. A. Pilling, P. A. Hoyne, K. A. Cartledge, T. M. Pham, J. L. Lewis, S. E. Sankovich, V. Stoichevska, S. E. Da, C. P. Robinson, M. J. Frenkel, L. G. Sparrow, R. T. Fernley, V. C. Epa and C. W. Ward, Structure of the insulin receptor ectodomain reveals a foldedover conformation, Nature, 443, 218–221 (2006). [38] C. W. Ward, M. C. Lawrence, V. A. Streltsov, T. E. Adams and N. M. McKern, The insulin and EGF receptor structures: new insights into ligand-induced receptor activation, Trends Biochem. Sci., 32, 129–137 (2007). [39] R. Z. Luo, D. R. Beniac, A. Fernandes, C. C. Yip and F. P. Ottensmeyer, Quaternary structure of the insulin–insulin receptor complex, Science (New York, N.Y.), 285, 1077–1080 (1999). [40] S. R. Hubbard, L. Wei, L. Ellis and W. A. Hendrickson, Crystal structure of the tyrosine kinase domain of the human insulin receptor, Nature, 372, 746–754 (1994). [41] C. Kristensen, T. Kjeldsen, F. C. Wiberg, L. Scha¨ffer, M. Hach, S. Havelund, J. Bass, D. F. Steiner and A. S. Andersen, Alanine scanning mutagenesis of insulin, J. Biol. Chem., 272, 12978–12983 (1997). [42] R. A. Pullen, D. G. Lindsay, S. P. Wood, I. J. Tickle, T. L. Blundell, A. Wollmer, G. Krail, D. Brandenburg, H. Zahn, J. Gliemann and S. Gammeltoft, Receptor–binding region of insulin, Nature, 259, 369–373 (1976). [43] L. Scha¨ffer, A model for insulin binding to the insulin receptor, Eur. J. Biochem., 221, 1127–1132 (1994). [44] P. DeMeyts, A. R. Bainco and J. Roth, Site–site interactions among insulin receptors. Characterization of the negative cooperativity, J. Biol. Chem., 251, 1877–1888 (1976). [45] S. Huang and M. P. Czech, The GLUT4 glucose transporter, Cell Metab., 5, 237–252 (2007). [46] I. M. A. Nooren and J. M. Thornton, Structural characterisation and functional significance of transient protein–protein interactions, J. Mol. Biol., 325, 991–1018 (2003). [47] M. J. Adams, T. L. Blundell, E. J. Dodson, G. G. Dodson, M. Vijayan, E. N. Baker, M. M. Harding, D. C. Hodgkin, B. Rimmer and S. Sheat, Structure of rhombohedral 2 zinc insulin crystals, Nature, 224, 491–495 (1969).

278

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

[48] J. Brange, L. Andersen, E. D. Laursen, G. Meyn and E. Rasmussen, Toward understanding insulin fibrillation, J. Pharm. Sci., 86, 517–525 (1997). [49] R. N. Longfield, L. P. Smith and J. N. Longfield, Multiple-dose vials: Persistence of bacterial contaminants and infection control implications, Inf. Cont., 6, 194–199 (1985). [50] U. Derewenda, Z. Derewenda, E. J. Dodson, G. G. Dodson, C. D. Reynolds, G. D. Smith, C. Sparks and D. Swenson, Phenol stabilizes more helix in a new symmetrical zinc insulin hexamer, Nature, 338, 594–596 (1989). [51] M. Roy, M. L. Brader, R. W. Lee, N. C. Kaarsholm, J. F. Hansen and M. F. Dunn, Spectroscopic signatures of the T to R conformational transition in the insulin hexamer, J. Biol. Chem., 264, 19081–19085 (1989). [52] J. M. Conlon, Evolution of the insulin molecule: Insights into structure–activity and phylogenetic relationships, Peptides, 22, 1183–1193 (2001). [53] S. E. Fineberg, J. A. Galloway, N. S. Fineberg and J. Goldman, Effects of species of origin, purification levels, and formulation on insulin immunogenicity, Diabetes, 37, 592–599 (1983). [54] M. A. Ruttenberg, Human insulin: facile synthesis by modification of porcine insulin, Science, 177, 623–626 (1972). [55] K. Morihara, T. Oka and H. Tsuzuki, Semi-synthesis of human insulin by trypsincatalyzed replacement of Ala-B30 by Thr in porcine insulin, Nature, 280, 412–413 (1979). [56] K. Morihara, T. Oka, H. Tsuzuki, Y. Tochino and T. Kanaya, Achromobacter protease I-catalyzed conversion of porcine insulin into human insulin, Biochem. Biophys. Res. Commun., 92, 396–402 (1980). [57] J. Markussen, U. Damgaard, M. Pingel, L. Snel, A. R. Sørensen and E. Sørensen, Human insulin (novo): chemistry and characteristics, Diabetes Care, 6 Suppl. 1, 4–8 (1983). [58] T. Masaki, M. Tanabe, K. Nakamura and M. Soejima, Studies on a new proteolytic enzyme from A chromobacter lyticus M497-1. I. Purification and some enzymatic properties, Biochim. Biophys. Acta, 660, 44–50 (1981). [59] I. S. Johnson, Human insulin from recombinant DNA technology, Science, 219, 632–637 (1983). [60] T. Kjeldsen, J. Brandt, A. S. Andersen, M. Egel-Mitani, M. Hach, A. F. Pettersson and K. Vad, A removable spacer peptide in an alpha-factor-leader/insulin precursor fusion protein improves processing and concomitant yield of the insulin precursor in Saccharomyces cerevisiae, Gene, 170, 107–112 (1996). [61] J. Markussen, U. Damgaard, I. Diers, N. Fiil, M. T. Hansen, P. Larsen, F. Norris, K. Norris, O. Schou et al., Biosynthesis of human insulin in yeast via single-chain precursors, Peptides 1986; 19th Eur. Symp. 189–194 (1987). [62] J. Schlichtkrull, J. Brange, Aa. H. Christiansen, O. Hallund, L. G. Heding, ˚ . Vølund, K. H. Jørgensen, S. Munkgaard Rasmussen, E. Sørensen and A Monocomponent insulin and its clinical implications., Horm. Metab. Res. Suppl. Series, 5, 134–143 (1974). [63] J. Schlichtkrull, Purity and antigenicity of insulin preparations, Acta Paediatr. Scand. Suppl., 270, 37–40 (1978). [64] P. Felig, Landmark perspective: Protamine insulin. Hagedorn’s pioneering contribution to drug delivery in the management of diabetes, J. Am. Med. Ass., 251, 393–396 (1984).

REFERENCES

279

[65] H. C. Hagedorn, B. N. Jensen, N. B. Krarup and I. Wodstrup, Protamine insulinate, J. Am. Med. Ass., 106, 177–180 (1936). [66] P. R. Raskin, P. A. Hollander, A. Lewin, R. A. Gabbay, B. Bode and A. J. Garber, Basal insulin or premix analogue therapy in type 2 diabetes patients, Eur. J. Inter. Med., 18, 56–62 (2007). [67] J. A. Galloway, C. T. Spradlin, R. L. Nelson, S. M. Wentworth, J. A. Davidson and J. L. Swarner, Factors influencing the absorption, serum insulin concentration, and blood glucose responses after injections of regular insulin and various insulin mixtures, Diabetes Care, 4, 366–376 (1981). [68] J. R. Woodworth, D. C. Howey and R. R. Bowsher, Establishment of time–action profiles for regular and NPH insulin using pharmacodynamic modeling, Diabetes Care, 17, 64–68 (1994). [69] P. O. Olsson, H. Arnqvist and S. H. Von, Free insulin profiles in insulin-dependent diabetics treated with one or two insulin injections per day, Acta Med. Scand., 220, 133–141 (1986). [70] A. Sola, E. Larger, J. M’Bemba, F. Elgrably and G. Slama, New insulins: Interests and convenients, Reanimation, 15, 454–460 (2006). [71] M. Kasa and F. Borda, Surfen allergy causing insulin depot S hypersensitivity, Orv. Hetil., 131, 965–6, 969 (1990). [72] H. Hasche, M. Haslbeck, W. Bachmann and H. Mehnert, Patterns of allergic skin reactions during insulin treatment [Author’s translation], Deut. Med. Woch., 106, 1451–1456 (1981). [73] E. J. Cohn, J. D. Ferry, J. J. Livingood and M. H. Blanchard, Studies on the physical chemistry of insulin. I. The solubility and dielectric properties of insulin and its crystallisation with radioactive zinc, Science, 90, 183–185 (1939). [74] E. J. Cohn, J. D. Ferry, J. J. Livingood and M. H. Blanchard, Studies on the physical chemistry of insulin. II. Crystallisation of radioactive zinc insulin containing two or more zinc atoms, J. Am. Chem. Soc., 63, 17–21 (1941). [75] K. Hallas–Møller, K. Petersen and J. Schlichtkrull, Crystalline and amorphous insulin–zinc compounds with prolonged action, Science, 116, 394–399 (1952). [76] U. Grau, Insulin arginine: A new retard-principle based on a natural proinsulinderived processing intermediate, Diabates Res. Clin. Pract., S204 (1985). [77] J. Markussen, P. Hougaard, U. Ribel, A. R. Sørensen and E. Sørensen, Soluble, prolonged-acting insulin derivatives. I. Degree of protraction and crystallizability of insulins substituted in the termini of the B-chain, Prot. Eng., 1, 205–213 (1987). [78] J. Markussen, I. Diers, A. Engesgaard, M. T. Hansen, P. Hougaard, L. Langkjaer, K. Norris, U. Ribel, A.R. Sørensen, E. Sørensen and H. O. Voigt, Soluble, prolongedacting insulin derivatives. II. Degree of protraction and crystallizability of insulins substituted in positions A17, B8, B13, B27 and B30, Prot. Eng., 1, 215–223 (1987). [79] J. Markussen, I. Diers, P. Hougaard, L. Langkjaer, K. Norris, L. Snel, A. R. Sørensen, E. Sørensen and H. O. Voigt, Soluble, prolonged-acting insulin derivatives. III. Degree of protraction, crystallizability and chemical stability of insulins substituted in positions A21, B13, B23, B27 and B30, Prot. Eng., 2, 157–166 (1988). [80] H. P. Neubauer and U. Grau, Insulin-arginine-2: Evaluation of a novel principle of retarded biological activity, Diab. Res. Clin. Prac., S404 (1985). [81] S. Jorgensen, A. Vaag, L. Langkjaer, P. Hougaard and J. Markussen, NovoSol Basal: Pharmacokinetics of a novel soluble long acting insulin analogue, Br. Med. J., 299, 415–419 (1989).

280

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

[82] A˚. Vølund, U. Ribel, P. L. Dam, J. Markussen, L. Langkjaer, J. Dolben and D. R. Owens, Application of the euglycaemic clamp technique to bioassay of insulin analogues, Biologicals: J. Inter. Assoc. Biol. Stand., 20, 135–142 (1992). [83] J. Buse, Insulin glargine (HOE901), Diabetes Care, 23, 576–578 (2000). [84] P. Home, Insulin glargine: The first clinically useful extended-acting insulin in half a century?, Exp. Opin. Invest. Drugs, 8, 307–314 (1999). [85] G. B. Bolli and D. R. Owens, Insulin glargine, Lancet, 356, 443–445 (2000). [86] M. Bahr, T. Kolter, G. Seipke and J. Eckel, Growth promoting and metabolic activity of the human insulin analogue GlyA21,ArgB31, ArgB32insulin (HOE 901) in muscle cells, Eur. J. Pharm., 320, 259–265 (1997). [87] P. Kurtzhals, L. Schaffer, A. Sørensen, C. Kristensen, I. Jonassen, C. Schmid and T. Trub, Correlations of receptor binding and metabolic and mitogenic potencies of insulin analogs designed for clinical use, Diabetes, 49, 999–1005 (2000). [88] W. D. Kohn, R. Micanovic, S. L. Myers, A. M. Vick, S. D. Kahl, L. Zhang, B. A. Strifler, S. Li, J. Shang, J. M. Beals, J. P. Mayer and R. D. DiMarchi, pI-shifted insulin analogs with extended in vivo time action and favorable receptor selectivity, Peptides, 28, 935–948 (2007). [89] W. F. Heath, R. M. Belagaje, G. S. Brooke, R. E. Chance, J. A. Hoffmann, H. B. Long, S. G. Reams, C. Roundtree, W. N. Shaw, L. J. Slieker, K. L. Sundell and R. D. DiMarchi, (A-C-B) human proinsulin, a novel insulin agonist and intermediate in the synthesis of biosynthetic human insulin, J. Biol. Chem., 267, 419–425 (1992). [90] P. Kurtzhals, B. Kiehr and A.R. Sørensen, The cobalt(III)–insulin hexamer is a prolonged-acting insulin prodrug, J. Pharm. Sci., 84, 1164–1168 (1995). [91] P. Kurtzhals and U. Ribel, Action profile of cobalt(III)–insulin. A novel principle of protraction of potential use for basal insulin delivery, Diabetes, 44, 1381–1385 (1995). [92] T. M. Allen and P. R. Cullis, Drug delivery systems: Entering the mainstream, Science, 303, 1818–1822 (2004). [93] S. Huille, N. Bryson, A. Constancis, L. Mohadjer, R. Meyrueix and G. Soula, Controlled release of insulin from nanoparticles of amphiphilic block copolyamino acid, J. Control. Rel., 64, 319–321 (2000). [94] T. Laugel, O. Crambes, S. Huille, L. Mohadjer, P. Vivet, C. Bloy, M. Rousseau, R. Meyrueix and G. Soula, In vivo profile of BasulinTM, a new long-acting insulin formulation with polyamino acid nanoparticles, J. Control. Rel., 64, 276–278 (2000). [95] S. Y. Lin, L. T. Ho and H. L. Chiou, Microencapsulation and controlled release of insulin from polylactic acid microcapsules, Biomat. Med. Dev. Art. Org, 13, 187–201 (1985). [96] P. Kurtzhals, S. Havelund, I. Jonassen, B. Kiehr, U. D. Larsen, U. Ribel and J. Markussen, Albumin binding of insulins acylated with fatty acids: characterization of the ligand–protein interaction and correlation between binding affinity and timing of the insulin effect in vivo, Biochem. J., 312 (Pt 3), 725–731 (1995). [97] P. Kurtzhals, S. Havelund, I. Jonassen, B. Kiehr, U. Ribel and J. Markussen, Albumin binding and time action of acylated insulins in various species, J. Pharm. Sci., 85, 304–308 (1996). [98] P. Kurtzhals, S. Havelund, I. Jonassen and J. Markussen, Effect of fatty acids and selected drugs on the albumin binding of a long-acting, acylated insulin analogue, J. Pharm. Sci., 86, 1365–1368 (1997).

REFERENCES

281

[99] S. R. Myers, F.E. Yakubu–Madus, W. T. Johnson, J. E. Baker, T. S. Cusick, V. K. Williams, F. C. Tinsley, A. Kriauciunas, J. Manetta and V. J. Chen, Acylation of human insulin with palmitic acid extends the time action of human insulin in diabetic dogs, Diabetes, 46, 637–642 (1997). [100] J. Radziuk, S. Pye, B. Bradley, J. Braaten, L. Vignati, P. Roach, R. Bowsher, R. Dimarchi and R. Chance, Basal activity profiles of NPH and [Nepsilon-palmitoyl Lys (B29)] human insulins in subjects with IDDM, Diabetologia, 41, 116–120 (1998). [101] J. H. Bradbury and L. R. Brown, Nuclear magnetic resonance spectroscopic studies of the amino groups of insulin, Eur. J. Biochem., 76, 573–582 (1977). [102] V. K. Naithani and H. G. Gattner, Preparation and properties of citraconylinsulins, Hoppe-Seyler Zeit. Phys. Chem., 363, 1443–1448 (1982). [103] G. A. Brunner, G. Sendlhofer, A. Wutte, M. Ellmerer, B. Sogaard, A. Siebenhofer, S. Hirschberger, G. J. Krejs and T. R. Pieber, Pharmacokinetic and pharmacodynamic properties of long-acting insulin analogue NN304 in comparison to NPH insulin in humans, Exper. Clin. Endocrin. Diab., 108, 100–105 (2000). [104] T. Heise and T. R. Pieber, Towards peakless, reproducible and long-acting insulins. An assessment of the basal analogues based on isoglycaemic clamp studies, Diab. Obes. Metab., 9, 648–659 (2007). [105] M. K. Dea, M. Hamilton-Wessler, M. Ader, D. Moore, L. Schaffer, M. Loftager, A. Volund and R. N. Bergman, Albumin binding of acylated insulin (NN304) does not deter action to stimulate glucose uptake, Diabetes, 51, 762–769 (2002). [106] M. S. Dennis, M. Zhang, Y. G. Meng, M. Kadkhodayan, D. Kirchhofer, D. Combs and L. A. Damico, Albumin binding as a general strategy for improving the pharmacokinetics of proteins, J. Biol. Chem., 277, 35035–35043 (2002). [107] M. Hamilton-Wessler, M. Ader, M. Dea, D. Moore, P. N. Jorgensen, J. Markussen and R. N. Bergman, Mechanism of protracted metabolic effects of fatty acid acylated insulin, NN304, in dogs: Retention of NN304 by albumin, Diabetologia, 42, 1254–1263 (1999). [108] T. Kjeldsen, P. Kurtzhals, A. F. Pettersson, L. Drube, S. Havelund, I. Jonassen and J. Markussen, Localization of albumin-binding site(s) for NN304, Diabetologia, 40, 1377 (1997). [109] S. Havelund, A. Plum, U. Ribel, I. Jonassen, A. Vølund, J. Markussen and P. Kurtzhals, The mechanism of protraction of insulin detemir, a long-acting, acylated analog of human insulin, Pharm Res, 21, 1498–1504 (2004). [110] J. L. Whittingham, S. Havelund and I. Jonassen, Crystal structure of a prolongedacting insulin with albumin-binding properties, Biochemistry, 36, 2826–2831 (1997). [111] T. Heise, L. Nosek, B. B. Ronn, L. Endahl, L. Heinemann, C. Kapitza and E. Draeger, Lower within-subject variability of insulin detemir in comparison to NPH insulin and insulin glargine in people with type 1 diabetes, Diabetes, 53, 1614–1620 (2004). [112] O. Klein, Albumin-bound basal insulin analogues (insulin detemir and NN344): comparable time–action profiles but less variability than insulin glargine in type 2 diabetes, Diab. Obes. Metab., 9, 290–299 (2007). [113] P. Home, P. Bartley, D. Russell-Jones, H. Hanaire–Broutin, J. E. Heeg, P. Abrams, M. Landin-Olsson, B. Hylleberg, H. Lang, E. Draeger and t.E.t.A.o.D.I.E.S.a.S.S.S.G.-Study, Insulin detemir offers improved glycemic control compared with NPH insulin in people with type 1 diabetes: A randomized clinical trial, Diabetes Care, 27, 1081–1087 (2004).

282

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

[114] K. Hermansen and M. Davies, Does insulin detemir have a role in reducing risk of insulin-associated weight gain?, Diab. Obes. Metab., 9, 209–217 (2007). [115] M. Ellmerer, M. Hamilton–Wessler, S. P. Kim, M. K. Dea, E. Kirkman, A. Perianayagam, J. Markussen and R. N. Bergman, Mechanism of action in dogs of slow-acting insulin analog O346, J. Clin. Endo. Metab., 88, 2256–2262 (2003). [116] I. Jonassen, S. Havelund, U. Ribel, A. Plum, M. Loftager, T. Hoeg-Jensen, A. Volund and J. Markussen, Biochemical and physiological properties of a novel series of long-acting insulin analogs obtained by acylation with cholic acid derivatives, Pharm. Res., 23, 49–55 (2006). [117] S. Lee, K. Kim, T. S. Kumar, J. Lee, S. K. Kim, D. Y. Lee, Y. K. Lee and Y. Byun, Synthesis and biological properties of insulin – Deoxycholic acid chemical conjugates, Bioconjugate Chem., 16, 615–620 (2005). [118] J. L. Whittingham, I. Jonassen, S. Havelund, S. M. Roberts, E. J. Dodson, C. S. Verma, A. J. Wilkinson and G. G. Dodson, Crystallographic and solution studies of N-lithocholyl insulin: A new generation of prolonged-acting human insulins, Biochemistry, 43, 5987–5995 (2004). [119] J. Brange and L. Langkjaer, Chemical stability of insulin. 3. Influence of excipients, formulation, and pH, Acta Pharm. Nord., 4, 149–158 (1992). [120] J. Brange, U. Ribel, J. F. Hansen, G. Dodson, M. T. Hansen, S. Havelund, S. G. Melberg, F. Norris, K. Norris and L. Snel, Monomeric insulins obtained by protein engineering and their medical implications, Nature, 333, 679–682 (1988). [121] J. Brange, D. R. Owens, S. Kang and A˚. Vølund, Monomeric insulins and their experimental and clinical implications, Diabetes Care, 13, 923–954 (1990). [122] S. Kang, J. Brange, A. Burch, A˚. Vølund and D. R. Owens, Subcutaneous insulin absorption explained by insulin’s physicochemical properties. Evidence from absorption studies of soluble human insulin and insulin analogues in humans, Diabetes Care, 14, 942–948 (1991). [123] K. Drejer, The bioactivity of insulin analogues from in vitro receptor binding to in vivo glucose uptake, Diabetes Metab. Rev., 8, 259–285 (1992). [124] G. Milazzo, L. Sciacca, V. Papa, I. D. Goldfine and R. Vigneri, ASPB10 insulin induction of increased mitogenic responses and phenotypic changes in human breast epithelial cells: evidence for enhanced interactions with the insulin-like growth factor-I receptor, Mol. Carcin., 18, 19–25 (1997). [125] F. G. Hamel, G. L. Siford, J. Fawcett, R. E. Chance, B. H. Frank and W. C. Duckworth, Differences in the cellular processing of AspB10 human insulin compared with human insulin and LysB28ProB29 human insulin, Metab. Clin. Exp., 48, 611–617 (1999). [126] K. Drejer, V. Kruse, U. D. Larsen, P. Hougaard, S. Bjørn and S. Gammeltoft, Receptor binding and tyrosine kinase activation by insulin analogues with extreme affinities studied in human hepatoma HepG2 cells, Diabetes, 40, 1488–1495 (1991). [127] R. D. DiMarchi, R. E. Chance, H. B. Long, J. E. Shields and L. J. Slieker, Preparation of an insulin with improved pharmacokinetics relative to human insulin through consideration of structural homology with insulin-like growth factor I, Horm. Res, 41 Suppl. 2, 93–96 (1994). [128] R. H. Becker, Insulin glulisine complementing basal insulins: A review of structure and activity, Diab. Tech. Ther., 9, 109–121 (2007). [129] I. Rakatzi, G. Seipke and E. Eckel, [LysB3, GluB29] insulin: a novel insulin analog with enhanced beta-cell protective action, Biochem. Biophys. Res. Commun., 310, 852–859 (2003).

REFERENCES

283

[130] A. M. Hennige, R. Lehmann, C. Weigert, K. Moeschel, M. Schauble, E. Metzinger, R. Lammers and H. U. Haring, Insulin glulisine: Insulin receptor signaling characteristics in vivo, Diabetes, 54, 361–366 (2005). [131] J. Halstrøm, U. Ribel, K. Kovacs, L. Scha¨ffer and A. R. Sørensen, Glycation at B1 makes insulin fast acting, Diabetologia, 34 Suppl. 1, A193, abstract 745 (1991). [132] M. Baudys, T. Uchio, D. Mix, D. Wilson and S. W. Kim, Physical stabilization of insulin by glycosylation, J. Pharm. Sci., 84, 28–33 (1995). [133] K. M. Stewart, M. F. Wilson and J. M. Rider, Insulin Delivery Devices, J. Pharm. Pract., 17, 20–28 (2004). [134] C. Poulsen, L. Langkjaer and G. Worsoe, Precipitation of insulin aspart and insulin glulisine products used for continuous subcutaneous insulin infusion, Diab. Tech. Ther., 9, 26–35 (2007). [135] M. Brownlee and A. Cerami, Glycosylated insulin complexed to concanavalin A. Biochemical basis for a closed-loop insulin delivery system, Diabetes, 32, 499–504 (1983). [136] I. Hisamitsu, K. Kataoka, T. Okano and Y. Sakurai, Glucose-responsive gel from phenylborate polymer and poly(vinyl alcohol) – Prompt response at physiological pH through the interaction of borate with amine group in the gel, Pharm. Res., 14, 289–293 (1997). [137] S. W. Kim, C. M. Pai, K. Makino, L. A. Seminoff, D. L. Holmberg, J. M. Gleeson, D. E. Wilson and E. J. Mack, Self-regulated glycosylated insulin delivery, J. Control. Rel., 11, 193–201 (1990). [138] A. Matsumoto, S. Ikeda, A. Harada and K. Kataoka, Glucose-responsive polymer bearing a novel phenylborate derivative as a glucose-sensing moiety operating at physiological pH conditions, Biomacromol., 4, 1410–1416 (2003). [139] D. Shiino, K. Kataoka, Y. Koyama, M. Yokoyama, T. Okano and Y. Sakurai, Controlled-release of insulin from boronic acid gel under physiological conditions, J. Control. Rel., 28, 317–318 (1994). [140] D. Shiino, Y. Murata, K. Kataoka, Y. Koyama, M. Yokoyama, T. Okano and Y. Sakurai, Preparation and characterization of a glucose-responsive insulin-releasing polymer device, Biomaterials, 15, 121–128 (1994). [141] D. Shiino, Y. Murata, A. Kubo, Y. J. Kim, K. Kataoka, Y. Koyama, A. Kikuchi, M. Yokoyama, Y. Sakurai and T. Okano, Amine-containing phenylboronic acid gel for glucose-responsive insulin release under physiological pH, J. Control. Rel., 37, 269–276 (1995). [142] P. J. Stout, J. R. Racchini, M. E. Hilgers and S. E. Noujaim, Continuous glucose monitoring: Key challenges to replacing episodic SMBG, Diab. Res. Clin. Pract., 74, S97–S100 (2006). [143] T. Hoeg–Jensen, S. Havelund, P. K. Nielsen and J. Markussen, Reversible insulin self-assembly under carbohydrate control, J. Am. Chem. Soc., 127, 6158–6159 (2005). [144] C. M. Peterson, L. B. Jovanovic and M. Brownlee, Closing the loop: Practical and theoretical, Diabetes Care, 3, 318–321 (1980). [145] E. Renard, G. Costalat, H. Chevassus and J. Bringer, Artificial beta-cell: Clinical experience toward an implantable closed-loop insulin delivery system, Diab. Metab., 32, 497–502 (2006). [146] E. S. Wilkins, Towards implantable glucose sensors: a review, J. Biomed. Eng., 11, 354–361 (1989).

284

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

[147] P. H. Kvist, T. Iburg, M. Bielecki, M. Gerstenberg, T. Buch-Rasmussen, E. Hasselager and H. E. Jensen, Biocompatibility of electrochemical glucose sensors implanted in the subcutis of pigs, Diab. Tech. Ther., 8, 463–475 (2006). [148] A. Tura, A. Maran and G. Pacini, Non–invasive glucose monitoring: Assessment of technologies and devices according to quantitative criteria, Diab. Res. Clin. Prac., 77, 16–40 (2007). [149] R. Badugu, J. R. Lakowicz and C. D. Geddes, A glucose-sensing contact lens: From bench top to patient, Curr. Opin. Biotech., 16, 100–107 (2005). [150] P. Thoniyot, F. E. Cappuccio, S. Gamsey, D. B. Cordes, R. A. Wessling and B. Singaram, Continuous glucose sensing with fluorescent thin-film hydrogels. 2. Fiber optic sensor fabrication and in vitro testing, Diab. Tech. Ther., 8, 279–287 (2006). [151] M. Morishita and N. A. Peppas, Is the oral route possible for peptide and protein drug delivery?, Drug Disc. Today, 11, 905–910 (2006). [152] S. Clement, P. Dandona, J. G. Still and G. Kosutic, Oral modified insulin (HIM2) in patients with type 1 diabetes mellitus: Results from a phase I/II clinical trial, Metab. Clin. Exp., 53, 54–58 (2004). [153] C. Dunn and M. P. Curran, Inhaled human insulin (Exubera ) – A review of its use in adult patients with diabetes mellitus, Drugs, 66, 1013–1032 (2006). [154] D. R. Owens, J. Grimley and P. Kirkpatrick, Inhaled human insulin, Nat. Rev. Drug Disc., 5, 371–372 (2006). [155] W. T. Cefalu, The new diabetes inhalers: New tools for the clinician, Curr. Diabetes Rep., 7, 165–167 (2007). [156] S. Rossi, G. Sandri and C. Caramella, Buccal delivery systems for peptides: Recent advances, Am. J. Drug Del., 3, 215–225 (2005). [157] M. Brackbill and R. Schuller, Oralin: Buccally absorbed insulin, Adv. Pharm., 3, 5–13 (2005). [158] M. Schlein, S. Ludvigsen, H. B. Olsen, A. S. Andersen, G. M. Danielsen and N. C. Kaarsholm, Properties of small molecules affecting insulin receptor function, Biochemistry, 40, 13520–13528 (2001). [159] C. Pender, I. D. Goldfine, V. P. Manchem, J. L. Evans, W. R. Spevak, S. Shi, S. Rao, S. Bajjalieh, B. A. Maddux and J. F. Youngren, Regulation of insulin receptor function by a small molecule insulin receptor activator, J. Biol. Chem., 277, 43565–43571 (2002). [160] A. K. Srivastava and M. Z. Mehdi, Insulino-mimetic and anti-diabetic effects of vanadium compounds, Diabetic Med., 22, 2–13 (2005). [161] B. Zhang, G. Salituro, D. Szalkowski, Z. H. Li, Y. Zhang, I. Royo, D. Vilella, M. T. Diez, F. Pelaez, C. Ruby, R. L. Kendall, X. Z. Mao, P. Griffin, J. Calaycay, J. R. Zierath, J. V. Heck, R. G. Smith and D. E. Moller, Discovery of a small molecule insulin mimetic with antidiabetic activity in mice, Science, 284, 974–977 (1999). [162] S. A. Qureshi, V. Ding, Z. Li, D. Szalkowski, D.E. Biazzo-Ashnault, D. Xie, R. Saperstein, E. Brady, S. Huskey, X. Shen, K. Liu, L. Xu, G. M. Salituro, J. V. Heck, D. E. Moller, A. B. Jones and B. B. Zhang, Activation of insulin signal transduction pathway and anti-diabetic activity of small molecule insulin receptor activators, J. Biol. Chem., 275, 36590–36595 (2000). [163] M. G. Roper, W. j. Qian, B. B. Zhang, R. N. Kulkarni, C. R. Kahn and R. T. Kennedy, Effect of the insulin mimetic L-783,281 on intracellular [Ca2þ] and insulin secretion from pancreatic b-cells, Diabetes, 51, S43–S49 (2002).

REFERENCES

285

[164] H. B. Wood, Jr., R. Black, G. Salituro, D. Szalkowski, Z. Li, Y. Zhang, D. E. Moller, B. Zhang and A. B. Jones, The basal SAR of a novel insulin receptor activator, Bioorg. Med. Chem. Lett., 10, 1189–1192 (2000). [165] K. Ono, H. Nakane, S. Shimizu and S. Koshimura, Inhibition of HIV-reverse transcriptase activity by asterriquinone and its analogues, Biochem. Biophys. Res. Commun., 174, 56–62 (1991). [166] U. Mocek, L. Schultz, T. Buchan, C. Baek, L. Fretto, J. Nzerem, L. Sehl and U. Sinha, Isolation and structure elucidation of five new asterriquinones from Aspergillus, Humicola and Botryotrichum species, J. Antibiot., 49, 854–859 (1996). [167] R. C. Pillutla, K. C. Hsiao, J. R. Beasley, J. Brandt, S. Ostergaard, P. H. Hansen, J. C. Spetzler, G. M. Danielsen, A. S. Andersen, R. E. Brissette, M. Lennick, P. W. Fletcher, A. J. Blume, L. Schaffer and N. I. Goldstein, Peptides identify the critical hotspots involved in the biological activation of the insulin receptor, J. Biol. Chem., 277, 22590–22594 (2002). [168] L. Schaffer, R. E. Brissette, J. C. Spetzler, R. C. Pillutla, S. Ostergaard, M. Lennick, J. Brandt, P. W. Fletcher, G. M. Danielsen, K. Hsiao, A. S. Andersen, O. Dedova, U. Ribel, T. Hoeg-Jensen, P. H. Hansen, A. J. Blume, J. Markussen and N. I. Goldstein, Assembly of high-affinity insulin receptor agonists and antagonists from peptide building blocks, Proc Nat. Aca. Sci. US, 100, 4435–4439 (2003). [169] W. Liu, A. Brock, S. Chen, S. Chen and P. G. Schultz, Genetic incorporation of unnatural amino acids into proteins in mammalian cells, Nature Meth., 4, 239–244 (2007). [170] P. G. Katsoyannis, Synthesis of Insulin, J. Am. Chem. Soc., 154, 1509–1514 (1966). [171] K. Akaji, K. Fujino, T. Tatsumi and Y. Kiso, Total synthesis of human insulin by regioselective disulfide formation using the silyl chloride sulfoxide method, J. Am. Chem. Soc., 115, 11384–11392 (1993). [172] R. A. D. Bathgate, F. Lin, N. F. Hanson, L. Otvos, A. Guidolin, C. Giannakis, S. Bastiras, S. L. Layfield, T. Ferraro, S. Ma, C. X. Zhao, A. L. Gundlach, C. S. Samuel, G. W. Tregear and J. D. Wade, Relaxin-3: Improved synthesis strategy and demonstration of its high-affinity interaction with the relaxin receptor LGR7 both in vitro and in vivo, Biochemistry, 45, 1043–1053 (2006). [173] T. Kjeldsen, S. Ludvigsen, I. Diers, P. Balschmidt, A. R. Sorensen and N. C. Kaarsholm, Engineering-enhanced protein secretory expression in yeast with application to insulin, J. Biol. Chem., 277, 18245–18248 (2002). [174] A. P. Tofteng, K. J. Jensen, L. Scha¨ffer, T. Hoeg-Jensen, Total synthesis of insulin via biomimetic folding of single-chain precursors, ChemBioChem, 9, 2663–2672 (2008). [175] Insulin inhalation – Pfizer/Nektar therapeutics: HMR 4006, inhaled PEG-insulin – Nektar, PEGylated insulin, Nektar, Drugs R D, 5, 166–170 (2004). [176] A. Pileggi, R. D. Molano, C. Ricordi, E. Zahr, J. Collins, R. Valdes and L. Inverardi, Reversal of diabetes by pancreatic islet transplantation into a subcutaneous, neovascularized device, Transplantation, 81, 1318–1324 (2006). [177] B. Hatipoglu, E. Benedetti and J. Oberholzer, Islet transplantation: Current status and future directions, Curr. Diab. Rep., 5, 311–316 (2005). [178] A. M. J. Shapiro, J. R. T. Lakey, E. A. Ryan, G. S. Korbutt, E. Toth, G. L. Warnock, N. M. Kneteman and R. V. Rajotte, Islet transplantation in seven patients with type 1 diabetes mellitus using a glucocorticoid-free immunosuppressive regimen, N. Engl. J. Med., 343, 230–238 (2000).

286

INSULIN VARIANTS FOR IMPROVED TREATMENT OF DIABETES

[179] G. Valdes, L. M. Dorantes, G. N. Garibay, E. Bracho-Blanchet, A. J. Mendez, P. Davila, R. B. Elliott, L. Teran and D. J. G. White, Xenotransplantation of porcine neonatal islets of Langerhans and Sertoli cells: A 4-year study, Eur. J. Endo., 153, 419–427 (2005). [180] M. Sykes and E. Cozzi, Xenotransplantation of pig islets into Mexican children: Were the fundamental ethical requirements to proceed with such a study really met?, Eur. J. Endo., 154, 921–922 (2006). [181] A. T. Cheung, B. Dayanandan, J. T. Lewis, G. S. Korbutt, R. V. Rajotte, M. Bryer-Ash, M. O. Boylan, M. M. Wolfe and T. J. Kieffer, Glucose-dependent insulin release from genetically engineered K cells, Science, 290, 1959–1962 (2000). [182] O. D. Madsen, Stem cells and diabetes treatment, APMIS, 113, 858–875 (2005). [183] K. Raun, V. P. von and L. B. Knudsen, Liraglutide, a once-daily human glucagonlike peptide-1 analog, minimizes food intake in severely obese minipigs, Obesity, 15, 1710–1716 (2007). [184] T. Vilsboll, M. Zdravkovic, T. Le-Thi, T. Krarup, O. Schmitz, J. P. Courreges, R. Verhoeven, I. Buganova and S. Madsbad, Liraglutide, a long-acting human glucagon-like peptide-1 analog, given as monotherapy significantly improves glycemic control and lowers body weight without risk of hypoglycemia in patients with type 2 diabetes, Diabetes Care, 30, 1608–1610 (2007).

Index

Note: Page numbers in italics refer to figures; page numbers in bold refer to tables. All Greek prefixes are ignored in the index order; e.g. for ‘‘a-helix’’, search under ‘‘helix’’. Acylated insulins 261–2 Alanine 156–7 scan (ala scan) 52, 163, 256 Albumin, as binder for insulin derivatives 262 b-alkyl amino acids 57, 57–9, 74 Allylic strain 148, 148, 149 AMBER (assisted model building with energy refinement) 10, 12, 13, 14, 17 Amide bond modification alkylation 55, 148, 148–9 N-peptoids 72–3, 73, 100, 143–4 reduction 66–70, 68, 69 replacements 55, 65–6, 66, 141–4, 143 reversals 53–5 see also Isosteres (pseudo-amide bonds); Methylation Amino acids b-alkyl substituted 57, 57–9, 74 Ca-substituted 55, 72 chimeras 60–1, 61 fluorinated 225–8 influence on cyclopeptide conformation 154 modified, commercial use 52–3 stereochemistry, effect on peptides 141, 156–7

sugar (SAAs) 187, 187–192 Tic-, Atc- and Aba-analogues 61–4, 62 unsaturated (dehydro-) 55–6, 57 see also Side chains, Amino acid 4-amino-tetrahydro-2-benzazepin-3-one (Aba) 62, 64 4-amino-tetrahydro-indolo[2, 3-c] azepin-3-one (Aia) 64, 64 2-amino-tetralin-2-carboxylic acid (Atc) 61, 63, 63 5-aminoanisic acid amide 80, 80 1-aminocycloalkane carboxylic acids (Acnc) 55, 59, 59–60 trans-2-aminocyclohexane carboxylic acid 79, 235 trans-2-aminocyclopentane carboxylic acid (ACPC) 100, 235 Aminoisobutyric acid (Aib) 55, 192 Aminopeptidase N (APN) 83, 83 AMOEBA (atomic multipole optimized energetics for biomolecular applications) 16, 18 Amylin, peptoid analogues 73 Anginex 87 Angiotensin 42 Angiotensin converting enzyme (ACE) 83 Antibiotics 135, 145, 238

Peptide and Protein Design for Biopharmaceutical Applications © 2009 John Wiley & Sons, Ltd. ISBN: 978-0-470-31961-1

Edited by Knud J. Jensen

288

INDEX

Aspart insulin (NovoRapid, Novolog, AspB28, X14) 265, 266, 267 Atom-atom force fields, see Force fields, atom-atom Atrial natriuretic peptide 61 Autodock (modelling program) 26, 39 Azapeptides 72, 72, 143, 143–4 Azepine-constrained amino acids 64, 64 BACE-1 (b-secretase inhibitor) 83, 84 Backbone modifications 55, 55–6, 65, 84–6 see also Amide bond modification; Cyclization; Pseudopeptides Basal insulin 255, 258–64 Bcl-xL (TM protein) 88, 89, 211 Tris-benzamide scaffold 90, 90 1, 4-benzodiazepine-2, 5-dione 104 BetaCore 228 Bile acid insulins 262–3 Binding sites conformation modes 8–9, 39–41, 146–8, 147 metal 224–5, 225 RGD recognition sequence motif 159, 160–1 see also Epitopes, Insulin Bioavailability 137–8 Biosynthesis and natural peptide modification 181 Biphenyls 76–7, 77, 89 Bipyridine-based scaffolds 78, 78 Blood glucose, see Glucose, blood Bolus insulin 251, 255, 265–7 Bombesin antagonist activity 63, 67, 67 Born–Oppenheimer approximation 10, 17 Boronic acids, in polyol insulin self-assembly 268, 268, 270 Bradykinin 52, 52, 64 Build-up procedures 22 Burst effect (drug release) 260, 268 Ca-substitution, amino acid/peptide 55, 72 Calixarenes, as topomimetics 86–7, 87 Cancer 99, 160, 161, 163, 260 Caps, N- and C–terminal 211 Carbocyclic Acnc series analogues 55 Carbohydrates 3D structural conformation 179, 179–81, 180

definition and nomenclature 178–9, 193 as peptidomimetic scaffolds 182–3, 185–6, 231–3, 233 Carboproteins 198–9, 231–3, 233 Cathepsin B, inhibitor of 72 b-cells, pancreatic 251, 253, 275 CHARMM (chemistry at Harvard molecular mechanics) 10, 12, 13, 14 Chemokine receptors 32, 163 Chimeras 60–1, 61, 178, 198–9 Cholecystokinin (CCK) 63, 68 Cilengitide (peptide drug candidate) 159, 61 Cis/trans conformers 148, 148–9 ClustalW program 36 Coiled coils 209, 212, 216–20, 219, 222 Comparative (homology) modelling 27, 36, 36–8 Computational chemistry comparison with alternative tools 6–9 current status 5–6, 9, 215–16 force field options 10–18, 33 sampling 21–31 scoring 18–20, 39 Conformation of carbohydrates 179, 179–81, 180 effect on in vivo stability 134, 137 flexibility, backbone 7, 55–6, 72, 75, 182 cyclization constraints 147, 152 and receptor recognition 19, 75, 139, 146–8, 147 reliability of models, cyclic peptides 153–4 sampling problems 21–2, 26 side chain topography 56, 56–7 transition states 24 Conjugates, cylodextrin-peptide 193–7 Corey–Link reaction 192, 192 CXCL12 (natural ligand, SDF-1) 32, 41 CXCR4 (a G-protein-coupled receptor) 31–2 antagonists design 163–4, 164 docking 39–42, 40 modelling, TM region 36–9

INDEX Cyclic peptides enantiomers 53, 53 hairpin mimetic 102, 102 medicinal (designed) 159–65, 185–6 natural (bioactive examples) 135–7, 145, 145 structural features hydrogen bond networks 150–1 peptide bond cis/trans orientation 148, 148–9 turn structures 149–50 as TASP templates 231 Cyclization design strategies 154–9 choice of amino acids 154–5 comparison with cyclopolyenes (Dunitz-Waser concept) 155–6, 156 spatial screening 156, 156–7, 160 effects on peptide properties 134, 146–8, 152 synthetic pathways 138–9, 139 backbone linkages 144–5 head-to-tail 139–40 side-chain (disulfide) bonds 144 Cyclodextrins 193–7 Cyclosporin A (immunosuppressant) 135–6, 136, 138 Cyclotides 137, 137 Cytokines 163, 236 Dead-end elimination, of side-chain rotamers 30, 216 Deltorphin analogues 58 Dermorphins (opioid peptides) 64 DesB30 human insulins 261–3, 264 Detemir (Levemir, acylated insulin) 261–2 Diabetes current and future treatments 274–5 history of treatment research 249–51, 258–9 Type 1 (IDDM) 251–2 Type 2 (NIDDM) 252–4 2, 8-diaminoepindolidione (b-sheet scaffold) 79, 79 3, 6-diaminoquinolone (b-sheet scaffold) 79, 80 Didehydroamino acids, E/Z stereoisomers 55–6, 57 Diketopiperazine (DKP) 194 2’, 6’-dimethyltyrosine (Dmt) 62

289

Diphenylphosphoryl azide (DPPA, cyclization reagent) 140 Dipiperidinobenzene 90, 91 Disulfide bridges 273 in cyclopeptides 136, 137, 144, 164 in a-helix assemblies 210, 222, 226, 228 Docking 39–42 DPDPE, in d-opioid analgesic design 194–5 Dunitz–Waser concept 155–6, 156 Dynorphin 67, 68 ECEPP (empirical conformational energy program for peptides) 10–11, 13, 22, 33 comparison with other force fields 15, 18, 38 Electrostatic interactions, in computational chemistry 17, 33 Enfuvirtide (T-20, Fuzeon, DP-178) 236–7 Enkephalin, glycosylation 184–5, 191 Enzymatic (proteolytic) degradation 134 Enzyme mimetics 196–7 Epitopes, insulin 256, 273 Eptifibatide (Integrilin) 136, 162–3, 162 Exenatide (Byetta, GLP-1 incretin drug) 254 Farnesyl transferase (FTE) 64, 85 FC131 (cyclic pentapeptide) 32–5, 34, 35 Fibronectin 160 Flexibility, see Conformation; loops Fluorenylglycine (Fgl) 58, 59 Fluorinated artificial proteins 225–8 Foldamers 74, 234, 234–5 Folding@home (sampling application) 25 Fold recognition (threading) 27–8 Force fields, atom-atom 9–18, 19, 22, 33–4, 38 Galactose (Galp) templates 198, 198–9, 232, 233, 233 Ganirelix 52, 52 Gastrin, b-cyclodextrin as carrier 195–6 GCN4 (scaffold protein) 237–8 Genetic algorithm sampling 26, 39–41

290

INDEX

Glargine (Lantus; insulin mimetic) 260 Glioblastoma 161 Glucagon 59, 164, 185, 251 Glucose, blood monitoring methods 254, 269, 271 natural hormonal control 251 raised level, consequences 252 reactive insulin release systems 267–9, 270 Glulisine (Apidra, GluB29 human insulin) 266 Glycation, protein 252, 252 Glycine 154, 210, 211 Glycopeptides 183–5, 184 GnRH (gonadotrophin-releasing hormone) 52, 52 Gp41 (glycoprotein) 236–7 GPCRs (G-protein-coupled receptors) 8, 159 Grafting 237–8 Gramicidin S 69, 135, 230 GROMOS (Groningen molecular simulation) 10, 12, 13, 14 b-hairpins assembly, using turn mimics 214–15, 228–9 induction 76–7, 82, 82, 190, 190–1 Hao (b-sheet inducing template) 80–2, 81 HATU (cyclization reagent) 140 HbA1c (glycated haemoglobin) 254 HCV (hepatitis C virus) protease, inhibitors of 62, 83, 83, 84, 91 HDM2 (human double minute-2) protein 99–104 a-helix 310 variant 210 314 variant 234–235, 238 bundles 222 3-helix 220, 223 4-helix 213, 213, 223–4 computational modelling 37–9 mimicry by carboproteins 198–99 template assembly (TASP) 229, 229–33, 236 mimetics 87–92 nucleating templates 88, 88–9, 211 stabilization 210–11, 212 structures 208, 210, 234–35 Heptad repeat 216, 237

HIV membrane fusion inhibitors 236–37 CXCR4/FC131 interaction 40–1 protease inhibitors 71, 71 Homology (comparative) modelling 27, 36, 36–8 Human leucocyte elastase (HLE) 85 Hydrazinoazapeptoids 73, 73 Hydrogen bond surrogates (HBS) 88, 88, 211 Icatibant 52, 52 IGF (insulin-like growth factor) 253 Imidazole scaffold 91, 91 Iminosugars, as somatostatin mimetic scaffold 186 Indane-constrained amino acids (Aic and Hai) 63, 63–4 Indanes 89, 90 Indole-constrained amino acids (Aia) 64, 64 INSIGHT (molecular modelling package) 15 Insulin administration 253, 255, 267, 271–72 artificial synthesis 273, 274 commercial preparation and sources 257–58 daily secretion, healthy 250, 251 fast-acting, as bolus dose 265–67 glucose-sensitive 267–69, 270 long-acting, for basal coverage 258–64, 263 peptidomimetics 272–3 pumps 263, 267, 268–9, 271 structure (human) 256, 256–7 Integrilin (eptifibatide) 136, 162–3, 162 Integrin inhibitors 159–62, 162 SAA ligands, RGD motif 191–92, 192 Interleukin-1b convertase (ICE) 85 Isoelectric point-shifted insulins 259–60 Isosteres (pseudo-amide bonds) alkene 68–70, 69 effect on peptide properties 65–6, 142–44 (hetero)aromatic ring 70, 70 hydroxy-ethylamine (HEA) 71, 71 reduced amide 66–8, 68 see also Amide bond modification

INDEX Lactams, as enzyme inhibitor scaffolds 85–6, 86 Lente insulins 259 Leu zippers 212, 216–17, 226 Ligand orientation, computational sampling 25–6, 39–41 Liraglutide (Victoza, GLP-1 incretin drug) 254 Loops cyclopeptide turns as mimics of 149–50 extracellular, flexibility of 42 as helix bundle connectors 220, 222 as receptor recognition sites 8 Lyspro insulin (Humalog) 265 Macrocycles, as enzyme inhibitors 83, 83, 84 Macrodipole 210, 211 MacroModel (molecular modelling package) 15, 34 MDM2 (murine double minute-2) protein 99–104 Melanotan 136 A-melanotropin (melanocortin) 7, 22, 62 MT-II analogue 59, 61, 64 Menarini glycopeptide (MEN 11420) 184, 184 Metalloprotein design 224–5, 225, 231 2, 4-methanoamino acid analogues 60 Methoxypyrrole amino acid (MOPAS) 79–80, 80 Methylation at amide bond 55, 143, 160 in aromatic amino acids 58 N-methylation 55, 138, 143, 164 Modeller computational package 27 Modified peptides backbone modifications 55–6 cyclic retroenantiomers 53, 53 definition 49 linear retroinversion 54, 54–5 side chain constraints 56–65 use of non-coded amino acids 52, 52–3 Molecular dynamics (MD) 12, 23 Folding@home application 25 Molecular mechanics (MM) 10, 15

291

Molecular modelling computational software packages 15, 17, 27, 34 scoring functions, in drug design 20 Molecular sensing 197 Monte Carlo simulations 23–4 mRR (mammalian ribonucleotide reductase) inhibitor 189, 189 Negative design 29, 215, 218, 224 Neurodegenerative diseases 76 Neuromedin 63, 68 Neuropeptide Y (NPY) 68, 89 Neurotensin 67, 68, 74 NMR spectroscopy 152–3 Nociceptin 68, 73 Norbornene (b-sheet scaffold) 78 Nowick artificial b-sheets 80, 80 NPH insulin (Neutral Protamine Hagedorn, isophane) 258–9 Nucleating templates helix 88, 88–9, 211 b-sheet 76–82, 77, 78, 214–215 Octreotide (SMS 201–995) 92, 93, 165 OGP(10–14) (cyclic osteogenic growth peptide) 53, 53 Oligomeric state coiled coils 217–20 insulin 255, 257, 265 Oligosaccharides 180–1, 193 OPLS (optimised potentials for liquid simulations) 10, 12, 13, 14, 34 Oral delivery, insulin 271, 272 1, 2, 4-oxadiazole 70, 70 Oxalamides 80 Oxime ligation 198, 198 Oxytocin analogues 58 P53 (tumour suppression gene) protein 99–104 Pancreas artificial 268–69, 271 cell types and activity 251 transplants 274–75 PEGylation, insulin 271 Pentapeptides, cyclic, structure determination 32–5 Peptide bond modification, see Amide bond modification b-peptides 74, 100, 234, 238

292

INDEX

Peptidomimetics (peptide mimetics) compared with peptides 49–51, 50 definitions 51 design procedure 157–9, 159 examples of design applications CXCR4 antagonists 163–64, 164 inhibitors of p53-H/MDM2 protein interaction 99–104, 101 integrin inhibitors 159–62, 162 somatostatin analogues 93–7, 165, 165, 185–86, 191 see also Cyclization; Modified peptides Peptoids 72–3, 73, 100, 143–44 Phalloidin 145, 145 Pharmacophore 3D model, design of scaffold mimics 93–4 determination of 3D arrangement 6–8, 32–5, 135 Plasmepsin 83, 84 Platelet-derived growth factor (PDGF) 87 PMRI (partially modified retroinverso) peptides 54, 54 Polarizable force fields 17–18 Polyethers, as helix mimetic scaffolds 91 Polymer formulations, insulin 260 Polymixins 135 Polypyrrolinone b-strand scaffold 84, 85 Positron emission tomography (PET) 191–92 Preferred conformation backbone 143, 147, 149 cyclic structures 148, 156 side chain rotamers 152 sugars 180 Privileged templates/scaffold concept 95, 186 Proline 154, 211 Prolino amino acid (chimera) 60–1 Protease inhibitors 62, 63, 71, 83, 196 Protein design, de novo approach 207–8, 224–5 Pseudopeptides amide bond isosteres ( pseudo-peptide bonds) 65–72, 66 azapeptides 72, 72 extensions, backbone 74, 74 N-peptoids 72–3, 73

Pulmonary (lung) insulin administration 271–2 Pumps, insulin 263, 267, 268–9, 271 Pyridazine scaffold 91 tris-pyridylamide scaffold 89–90 Pyrrole scaffold 91 QM/MM (quantum/molecular mechanics) modelling 15–17, 16 Ramachandran maps 11–15, 13, 33, 149 Receptor-subtype selectivity 147–8, 160 Renin, inhibitors of 64, 68, 71 Replica exchange simulation 25 Retroenantiomers, cyclic 53, 53 Retro-inverso structures 54–5, 93, 100, 141–2, 142 Rheumatoid arthritis (RA) 136, 163 Rigidity 147 see also Conformation Ring-closing metathesis (RCM) 83, 84 ROP protein 228, 231 Rosetta algorithm 28–9, 215, 229 Rotamers, side-chain 29–31, 33, 56–9, 58, 61 constraint by cyclization 152 Salt bridges 209, 210, 216, 227 Sampling methods (computational chemistry) 215 comparative (homology) modelling 27, 36, 36–8 genetic algorithms 26, 39–41 Rosetta algorithm 28–9 side-chain rotamers 29–31 dead-end elimination 30, 216 discrete conformation libraries 30 sequential rotation algorithm 31, 37 stochastic 23–5 systematic 21–2 threading (fold recognition) 27–8 Sandostatin 165, 164–5 Scaffolds carbohydrate 182–183, 185–6, 231–3, 233 helix 88–92 for secondary structure elements (TASP) 229, 229–33

INDEX for somatostatin mimetics 94–8, 185–6 b-sheet stabilization 76–82 b-strand 84–6 see also Templates Scoring functions 18–20, 39 Screening high throughput (HTS) assay 92, 104, 133 library 74, 95, 96, 99 spatial 156, 156–7, 160 see also Scoring functions SDF-1 (CXCL12, natural ligand) 32, 41 Secondary structure mimetics connecting elements 214–15 helices 87 helix nucleators 88, 88–9 helix surface mimetics 89–92, 90, 91 b-strands and sheets 75–6 backbone-modified b-strands 84–6 macrocyclic b-strand mimics 82–4 topomimetics 86–7, 87 b-sheet nucleating templates 76–82, 77, 78, 214–15 b-secretase inhibitors 71, 71 Seglitide (MK-678, cyclic hexapeptide) 92, 93 Self-assembly, of insulin multihexamers 263, 264, 270 Sequential rotation algorithm 31, 37 b-sheets in BetaCore (designed protein) 228 nucleating templates 76–82, 77, 78, 214–15 Sialic acid (Neu5Ac) 187, 187 Side chains, amino acid 29–31, 56–65 for helix stabilization 210–11 tethering 59, 59–60 see also Rotamers, Side-chain Simulated annealing 24–5 Small-angle X-ray scattering (SAXS) 230, 232 Somatostatins (SSTs) designed mimetics 93–7, 165, 164–5, 185–6, 191 SRIF-14 (natural peptide hormone) 92, 93, 185 Spirooxindole 104 Stapling 210–211 Stochastic sampling 23–5

293

b-strand mimetics 75–6, 82–6 Structure-activity relationships (SAR) studies 32, 51–2 Sugar amino acids (SAAs) 187, 187–92 Sulfonamide, substitution of amide bond 143 Super-secondary structures, see Coiled coils Surfen insulin 259 SYBYL (molecular modelling package) 15 Synthesis Fmoc (solid phase) strategy 183–4, 189 of oligomers/polymers, including SAAs 190–2 of SAAs 187 Systematic (grid) search sampling 21–2 TASP (template-assembled synthetic protein) concept for artificial proteins 214, 229, 229–33 for cyclodextrin-peptide conjugates 195, 198 Templates nucleating helix 88, 88–9, 211 b-sheet 76–82, 77, 78, 214–15 privileged 95, 186 see also Scaffolds Terephtalamide scaffold 90 Terphenyl scaffold 89–90, 91, 102 Tertiary structures 215–29 coiled coils 209, 212, 216–20, 219, 222 fluoroproteins 225–28 a-helix bundles 213, 213, 220–25, 222 metalloproteins 224–25, 225 using foldamers 235 Tetrahydroisoquinoline-3-carboxylic acid (Tic) in aromatic amino acid side chains 61–3, 62 incorporation of TMT 58 Tetrahydro-b-carboline 96 Tetrazole rings 70 Threading (fold recognition) 27–8 Thrombin 74, 86 @Tides b-strand scaffold units 84, 85 TIPP family (opioid peptides) 61–2 use of Tic residues in 63

294

INDEX

TNF-a converting enzyme (TACE) 83 Topographical (w-space) design 56–65 Topomimetics 86–7, 87 Toxicity, of peptide drugs 134 Transmembrane (TM) region modelling 36–9 Transplantation, pancreas 274–275 Triazine scaffolds 91, 92 Triazole rings 70, 70, 95 Trimeric coiled coils 218–20, 219, 237 Turns carbohydrate mimetics 182–83, 188, 188–89, 190–91 in cyclic peptides 149–51 peptide, structural types 182 see also b-sheets, Nucleating templates Tyrosine kinases 63

Umbrella sampling 24 Urantide 98, 98 Urotensin (U-II) 97–9 Vancomycin 136, 145, 145 Vasopressin 136 Veber–Hirschmann peptide 164, 165 X-ray crystallography data, for evaluating force fields 13 limitations 7–9, 153 Zinc(II) insulin hexamers 255, 265

12,

251, 255,