Peptide Drug Discovery and Development E-Book

• Herausgeber: Wiley-VCH
• Kategorie: Wissenschaft und neue Technologien
• Sprache: Englisch

Seitenzahl: 685

Veröffentlichungsjahr: 2011

Beschreibung

Filling a real knowledge gap, this handbook and ready reference is both modern and forward-looking in its emphasis on the "bench to bedside" translational approach to drug development.
Clearly structured into three major parts, the book stakes out the boundaries of peptide drug development in the preclinical as well as clinical stages. The first part provides a general background and focuses on the characteristic strengths and weaknesses of peptide drugs. The second section contains five cases studies of peptides from diverse therapeutic fields, and the lessons to be learned from them, while the final part looks at new targets and opportunities, discussing several drug targets and diseases for which peptide drugs are currently being developed.

Leseprobe

Contents

Preface

List of Contributors

Part I: The Academia – Market Bouncing of Peptide Drugs – Challenges and Strategies in Translational Research with Peptide Drugs

Chapter 1: Peptides as Leads for Drug Discovery

1.1 Introduction

1.2 Overview of Process for Transforming Peptides to Peptidomimetics

1.3 HCMV Protease

1.4 HCV Protease

1.5 Herpes Simplex Virus

1.6 Renin

1.7 HIV

1.8 Conclusions

Chapter 2: Marketing Antimicrobial Peptides: A Critical Academic Point of View

2.1 Introduction

2.2 Basic Research: Antimicrobial Peptides

2.3 Patents

2.4 Potential Applications of AMPs

2.5 Technology Transfer: Valorization, Licensing, or Spin-Off Creation

2.6 Spin-Off Creation: An Academic Point of View

Chapter 3: Oral Peptide Drug Delivery: Strategies to Overcome Challenges

3.1 Introduction

3.2 Challenges Associated with Oral Peptide Delivery

3.3 Strategies to Overcome the Barriers of the Gastrointestinal Tract

3.4 Conclusions

Chapter 4: Rational Design of Amphipathic α-Helical and Cyclic β-Sheet Antimicrobial Peptides: Specificity and Therapeutic Potential

4.1 Introduction to Antimicrobial Peptides

4.2 Antimicrobial and Hemolytic Activities of Amphipathic α-Helical Antimicrobial Peptides: Mechanisms and Selectivity

4.3 Structure–Activity Relationship Studies of Amphipathic α-Helical and Cyclic β-Sheet Antimicrobial Peptides: Optimization of Pathogen Selectivity and Prevention of Host Toxicity

4.4 Commercialization of Antimicrobial Peptides

4.5 Therapeutic Potential

Chapter 5: Conotoxin-Based Leads in Drug Design

5.1 Introduction

5.2 Conotoxin Synthesis, Folding, and Structure

5.3 Conotoxins as Drug Leads

5.4 Conclusions

Chapter 6: Plant Antimicrobial Peptides: From Basic Structures to Applied Research

6.1 Introduction

6.2 The Diversity of Plant Antimicrobial Peptides: Focusing on Tissue Localization and Plant Species Distribution

6.3 Possible Structural Folds Found in Plant AMPs to Date

6.4 New Biotechnological Products Produced from Plant Peptides

Part II: Peptide Drugs’ Translational Tales – Peptide Drugs Before, Through and After Industry Pipelines

Chapter 7: Omiganan Pentahydrochloride: A Novel, Broad-Spectrum Antimicrobial Peptide for Topical Use

7.1 Omiganan: A Novel Anti-Infective Agent for Topical Indications

7.2 Structure and Mechanism of Action

7.3 Spectrum of Activity

7.4 Preclinical Efficacy Studies

7.5 Preclinical Toxicology Studies

7.6 Clinical Studies

7.7 Conclusions

Chapter 8: Turning Endogenous Peptides into New Analgesics: The Example of Kyotorphin Derivatives

8.1 Introduction

8.2 Peptides as Future Drug Candidates

8.3 Central Nervous System Analgesic Peptides

8.4 Endogenous Opioid System

8.5 Strategies to Deliver Analgesic Peptides to the Brain

8.6 Development of New Opioid-Derived Peptides

8.7 Kyotorphin – the Potential of an Endogenous Dipeptide

8.8 New KTP Derivatives

8.9 Assessing BBB Permeability with Peptide–Membrane Partition Studies

8.10 Kyotorphins: Partition to the Membrane and Enhanced Analgesic Activity

8.11 Academia and Pharmaceutical Industry: Friends or Foes?

Chapter 9: The Development of Romiplostim – a Therapeutic Peptibody Used to Stimulate Platelet Production

9.1 Introduction

9.2 Thrombopoietin and c-Mpl

9.3 Discovery and Optimization of Romiplostim

9.4 Pharmacodynamics (PD) and Pharmacokinetics (PK) of Romiplostim

9.5 A Brief ITP Primer

9.6 Romiplostim Clinical Data

9.7 Safety and Other Insights Gained from Romiplostim Design and Development

Chapter 10: HIV vs. HIV: Turning HIV-Derived Peptides into Drugs

10.1 Introduction

10.2 HIV-1 Envelope Protein

10.3 HIV Entry and Its Inhibition

10.4 HIV-1 Fusion Inhibitors: from Bench to Clinical Administration

10.5 New Strategies for Creating New HIV Fusion Inhibitor Peptides

10.6 Drug-Resistance and Combination Therapy

10.7 Concluding Remarks

Chapter 11: Sifuvirtide, A Novel HIV-1 Fusion Inhibitor

11.1 Ideal Drug Target HIV-1 gp41

11.2 Structure-Based Drug Design of Sifuvirtide

11.3 High Potency of Sifuvirtide

11.4 Limited Drug Resistance

11.5 Enhancement of the Efficiency of Sifuvirtide by Biomembrane Selectivity

11.6 Pharmacokinetics of Sifuvirtide with Long Half-Life

11.7 Stratification of Monotherapy

11.8 20 mg Sifuvirtide Once Daily vs. 100 mg T20 Twice Daily

11.9 Conclusions and Discussion

Part III: Whither Peptide Drugs? Peptides Shaping the Future of Drug Development

Chapter 12: Endogenous Peptides and Their Receptors as Drug Discovery Targets for the Treatment of Metabolic Disease

12.1 Centrally Secreted Neuropeptide Systems

12.2 Peripherally Secreted Neuropeptides

12.3 Summary

Chapter 13: Translation of Motilin and Ghrelin Receptor Agonists into Drugs for Gastrointestinal Disorders

13.1 Introduction

13.2 Motilin and Ghrelin Receptor Agonists Under Development

13.3 Translational Value of Preclinical Assays

13.4 Clinical Translation: Selecting the “Right” Patient Population

13.5 Clinical Development of Motilin and Ghrelin Receptor Agonists

13.6 Conclusions

Chapter 14: Of Mice and Men: Translational Research on Amylin Agonism

14.1 Overview of Amylin Physiology

14.2 Pramlintide: An Amylin Agonist

14.3 Amylin Agonism: Translational Research in Insulin-Dependent Diabetes

14.4 Amylin Agonism: Translational Research in Obesity

Chapter 15: Peptides and Polypeptides as Immunomodulators and Their Consequential Therapeutic Effect in Multiple Sclerosis and Other Autoimmune Diseases

15.1 Introduction

15.2 Peptides as Antigens and Vaccines

15.3 Peptides as Immunomodulators

15.4 Development of Copolymer 1 – a Polypeptide Immunomodulator Drug for the Treatment of Multiple Sclerosis

15.5 Additional Immunomodulatory Peptides as Drug Candidates

15.6 Summary and Concluding Remarks

Chapter 16: Development of Antibody Fragments for Therapeutic Applications

16.1 Antibodies

16.2 Conclusions

Index

Related Titles

Jameel, F., Hershenson, S. (eds.)

Formulation and Process Development Strategies for Manufacturing Biopharmaceuticals

976 pages

Hardcover

ISBN: 978-0-470-11812-2

Jorgenson, L., Nielson, H. M. (eds.)

Delivery Technologies for Biopharmaceuticals

Peptides, Proteins, Nucleic Acids and Vaccines

442 pages

Hardcover

ISBN: 978-0-470-72338-8

Jensen, K. (ed.)

Peptide and Protein Design for Biopharmaceutical Applications

306 pages

Hardcover

ISBN: 978-0-470-31961-1

Rathore, A. S., Mhatre, R. (eds.)

Quality by Design for Biopharmaceuticals

Principles and Case Studies

288 pages

Hardcover

ISBN: 978-0-470-28233-5

Groner, B. (ed.)

Peptides as Drugs

Discovery and Development

242 pages with approx. 38 figures and approx.

8 tables

2009

Hardcover

ISBN: 978-3-527-32205-3

Cavagnaro, J. A. (ed.)

Preclinical Safety Evaluation of Biopharmaceuticals

A Science-Based Approach to Facilitating Clinical Trials

approx. 1064 pages

Hardcover

ISBN: 978-0-470-10884-0

The Editors

Prof. Dr. Miguel Castanho

University of Lisbon

Institute of Biochemistry

Av. Egas Moniz

1649-028 Lisboa

Portugal

Prof. Dr. Nuno C. Santos

University of Lisbon

Institute of Biochemistry

Av. Egas Moniz

1649-028 Lisboa

Portugal

All books published by Wiley-VCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate.

Library of Congress Card No.: applied for

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.

Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at <http://dnb.d-nb.de>.

© 2011 Wiley-VCH Verlag & Co. KGaA, Boschstr. 12, 69469 Weinheim, Germany

All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – by photoprinting, microfilm, or any other means – nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law.

Print ISBN: 978-3-527-32891-8

ePDF ISBN: 978-3-527-63675-4

oBook ISBN: 978-3-527-63673-0

ePub ISBN: 978-3-527-63674-7

Mobi ISBN: 978-3-527-63676-1

Preface

The Timeliness of Peptide Drugs

Peptides make good drugs in challenging situations. They may be more expensive and time-consuming to produce than traditional small molecules, have low oral bioavailability, fast clearance in the body, and even, in some cases, be immunogenic. Yet, their ability to be very active, very specific, present very low toxicity, and often to be developed from natural endogenous scaffolds with known biological activity, makes them a desirable solution for unmet complex medical problems. This is not wishful thinking for the future, only. There are examples of very successful peptide drugs in clinical use. The key to this success is very much related to the mutual co-development of peptide biochemistry/biophysics, peptide synthetic chemistry, peptide pharmacology, and peptide biotechnology. This is the basic principle that underlies this book. Peptides have been the passion of many scientists and the investment of many entrepreneurs. Together they made a fantastic world with a very positive contribution to medicine. This book lets the reader know about this world, where peptides bounce between the bench and the bedside.

The numbers that reveal the achievements of peptide drugs are impressive. In 2008 six peptide drugs had attained global sales of US $750 million. Reichert and colleagues say this figure is a consequence of the widespread acceptance of protein therapeutics by both physicians and patients, together with improvements in tackling problems such as a short half-life and challenges with the delivery of these molecules [1]. In the same study, these authors state that the number of peptide drugs entering clinical trials per year was 1.7 in the 1970s, 4.6 in the 1980s, 9.7 in the 1990s and 16.9 in the 2000s up to 2009. Until now, at least 55 therapeutic peptides have been approved for human clinical use by at least one regulatory agency, although 6 of them were withdrawn from their markets afterwards. The approval success rates for peptides that entered clinical trials from 1984 to 2000 were 21–24%. More than 15 peptide candidates were in phase III trials or under regulatory review in 2009; thus, it is expected the field of peptide drugs will continue its positive trend of progress during the years to come. Given the effort of both academic and industrial R&D to overcome the pitfalls of peptide drug candidates [2, 3] and the “scalability challenges” of peptide production [4], it is predictable that peptide drug discovery and development will continue to prosper. We are not alone in our enthusiasm and optimism [5, 6].

From 2000 to 2007, peptides entering trials were most frequently treatments for metabolic disorders (26%). During the 1990s this fraction was less than 10% [1]. This variation shows how fast peptide drugs are evolving and their range of application being broadened. Areas which are today among the less frequent, such as infection and central nervous system, are being intensively researched and have a large potential to grow [4, 7–9]. They may be the future trend in peptide drug pipelines.

Despite all the peptide R&D figures above, the most important numbers are yet to be clearly ascertained: the number of lives saved by peptide drugs and the improvement in the patients’ quality of life. These are the figures truly worth working for and we hope this book will transmit knowledge, power and enthusiasm to the readers so that this endeavour is reinforced.

Finally, we wish to thank all contributors to the book. We are proud of their collaboration and engagement to turn this book into something different: a book about academic research translated into pharma industrial development, written by people directly involved and/or privileged inside witnesses. The mission of authors and editors has been to involve the reader in this world. Be welcome!

Miguel Castanho

Nuno C. Santos

Lisbon, March 14, 2011

References

1 Saladin, P.M., Zhang, B.D., and Reichert, J.M. (2009) Current trends in the clinical development of peptide therapeutics. Idrugs, 12 (12), 779–784.

2 Sato, A.K., Viswanathan, M., Kent, R.B., and Wood, C.R. (2006) Therapeutic peptides: technological advances driving peptides into development. Curr. Opin. Biotechnol., 17 (6), 638–642.

3 Lien, S. and Lowman, H.B. (2003) Therapeutic peptides. Trends Biotechnol., 21 (12), 556–562.

4 Marx, V. (2005) Watching peptide drugs grow up. Chem. Eng. News, 83 (11), 17–24.

5 Ayoub, M. and Scheidegger, D. (2006) Peptide drugs, overcoming the challenges, a growing business. Chim. Oggi – Chem. Today, 24 (4), 46–48.

6 Pichereau, C. and Allary, C. (2005) Therapeutic peptides under the spotlight. Eur. Biopharm. Rev., winter issue.

7 Naider, F. and Anglister, J. (2009) Peptides in the treatment of AIDS. Curr. Opin. Struct. Biol., 19 (4), 473–482.

8 Schwarze, S.R., Ho, A., Vocero-Akbani, A., and Dowdy, S.F. (1999) In vivo protein transduction: delivery of a biologically active protein into the mouse. Science, 285 (5433), 1569–1572.

9 Davis, T.P., Abbruscato, T.J., Brownson, E., and Hruby, V.J. (1995) Conformationally constrained peptide drugs targeted at the blood-brain barrier. in Membranes and Barriers: Targeted Drug Delivery (ed. R.S. Rapaka), NIDA Research Monograph 154, NIH, Rockville, MD, pp. 47–59.

List of Contributors

Rina Aharoni

Weizmann Institute of Science

Department of Immunology

PO Box 26

76100 Rehovot

Israel

Frederico Aires da Silva,

Technophage, SA

IMM

Rua Prof. Egas Moniz

Edifício Egas Moniz, Piso 2

1649–028 Lisboa

Portugal

Muharrem Akcan

The University of Queensland

Division of Chemistry and Structural

Biology

Institute for Molecular Bioscience

QLD 4072 Brisbane

Australia

David H Alpers

Washington University School of

Medicine

Department of Internal Medicine

Box 8031, 660 S Euclid Ave

MO 63110 St Louis

USA

Ruth Arnon

Weizmann Institute of Science

Department of Immunology

PO Box 26

76100 Rehovot

Israel

Eduard Bardají

Universitat de Girona, Campus

Montilivi

Laboratori d’Innovació en Processos i

Productes de Síntesi Orgànica

(LIPPSO), Chemistry Department

17071 Girona

Spain

John Broad

Queen Mary, University of London

Barts & The London School of

Medicine and Dentistry, Wingate

Institute of Neurogastroenterology

26 Ashfield Street

E1 2AJ London

UK

Sofia Côrte-Real

Technophage

IMM

Rua Prof. Egas Moniz

Edifício Egas Moniz, Piso 2

1649–028 Lisboa

Portugal

David J. Craik

The University of Queensland

Institute for Molecular Bioscience

QLD 4072 Brisbane

Australia

Simoni C. Dias

Universidade Católica de Brasília

Centro de Análises Proteômicas e

Bioquímicas Programa de Pós-

Graduação em Ciências Genômicas e

Biotecnologia, SGAN Quadra 916,

Modulo B

Av. W5

70790-160 Brasília-DF

Brazil

Dominique Dugourd

BioWest Therapeutics Inc.

Suite 1320–885 West Georgia

Vancouver, BC, V6C 2G2

Canada

Paul J. Edwards

Boehringer Ingelheim (Canada) Ltd

Research & Development

2100 Rue Cunard

Laval, QC, H7S 3G5

Canada

Octavio L. Franco

Universidade Católica de Brasília

Centro de Análises Proteômicas e

Bioquímicas Programa de Pós-

Graduação em Ciências Genômicas e

Biotecnologia

SGAN Quadra 916, Modulo B

Av. W5

70790-160 Brasília-DF

Brazil

Henri G. Franquelim

Universidade de Lisboa

Instituto de Medicina Molecular

Faculdade de Medicina

Av. Prof. Egas Moniz

1649-028 Lisboa

Portugal

João Gonçalves

Universidade de Lisboa

Instituto de Medicina Molecular

Faculdade de Farmácia

Av. Prof. Egas Moniz

1649-028 Lisboa

Portugal

Josias H. Hamman

Tshwane University of Technology

Department of Pharmaceutical

Sciences

Private Bag X680, Arcadia Campus,

0001 Pretoria

South Africa

and

North-West University

Unit for Drug Research and

Development

Private Bag X6001

2520 Potchefstroom

South Africa

Wendy J. Hartsock

University of Colorado

School of Medicine, Department of

Biochemistry & Molecular Genetics

CO 80045 Aurora

USA

Robert S. Hodges

University of Colorado

School of Medicine

Department of Biochemistry & Molecular Genetics

CO 80045

Aurora

USA

Steven R. LaPlante

Boehringer Ingelheim (Canada) Ltd

Research & Development

2100 Rue Cunard Laval

QC

H7S 2G5

Canada

Ning Lee

Bristol Myers Squibb

Department of Metabolic Diseases

NJ 08534

Hopewell

USA

Christine M. Mack

Amylin Pharmaceuticals Inc.9360 TowneCentre Drive

CA 92121

San Diego

USA

Pedro M. Matos

Universidade de Lisboa

Instituto de Medicina Molecular Faculdade de Medicina

Av. Prof. Egas Moniz1649-028 Lisboa

Portugal

Graham Molineux

Amgen Inc

Hematology and Oncology Discovery Research

One Amgen Center Drive

Mailstop 15-2-A

Thousand Oaks

CA 91320

USA

David G. Parkes

Amylin Pharmaceuticals Inc.9360 Towne Centre Drive

CA 92121

San Diego

USA

Mary Ann Pelleymounter

Bristol Myers Squibb

Department of Metabolic Diseases

NJ 08534

Hopewell

USA

Suzana M. Ribeiro

Universidade Católica de Brasília

Centro de Análises Proteômicas e

Bioquímicas, Programa de Pós-

Graduação em Ciências Genômicas e

Biotecnologia SGAN Quadra 916

Modulo B

Av. W5

70790-160

Brasília-DF

Brazil

Marta M.B. Ribeiro

Universidade Lisboa,

Instituto de Medicina Molecular,

Faculdade de Medicina

Av. Prof. Egas Moniz

1649-028 Lisboa

Portugal

Jonathan D. Roth

Amylin Pharmaceuticals Inc.

9360 Towne Centre Drive

CA 92121 San Diego

USA

Evelina Rubinchik

BioWest Therapeutics Inc.

Suite 1320–885 West Georgia

Vancouver, BC, V6C 2G2

Canada

Gareth J. Sanger

Queen Mary, University of London

Barts & The London School of

Medicine and Dentistry Wingate

Institute of Neurogastroenterology

26 Ashfield Street

E1 2AJ London

UK

Sónia Sá Santos

Universidade Lisboa,

Instituto de Medicina Molecular,

Faculdade de Medicina

Av. Prof. Egas Moniz

1649-028 Lisboa

Portugal

Michael Sela

Weizmann Institute of Science

Department of Immunology

PO Box 26,

76100 Rehovot

Israel

Isa D. Serrano

Universidade Lisboa,

Instituto de Medicina Molecular,

Faculdade de Medicina

Av. Prof. Egas Moniz

1649-028 Lisboa

Portugal

Jan H. Steenekamp

North-West University

Department of Pharmaceutics, School

of Pharmacy

Private Bag X6001

Potchefstroom Campus,

2520 Potchefstroom

South Africa

James L. Trevaskis

Amylin Pharmaceuticals Inc.

9360 Towne Centre Drive

CA 92121

San Diego

USA

A. Salomé Veiga

National Cancer Institute

Chemical Biology Laboratory

376 Boyles Street

MD 21702 Frederick

USA

and

Universidade Lisboa,

Instituto de Medicina Molecular,

Faculdade de Medicina

Av. Prof. Egas Moniz

1649-028 Lisboa

Portugal

Fengshan Wang

Shandong University

Institute of Biochemical and

Biotechnological Drug, School of

Pharmaceutical Sciences

Jinan

250012 Shandong

P. R. China

Yuren Wang

Bristol Myers Squibb

Department of Metabolic Diseases

NJ 08534 Hopewell

USA

Ping Wei

Amgen Inc

Hematology and Oncology Discovery

Research

One Amgen Center Drive, Mailstop

15-2-A

CA 91320 Thousand Oaks

USA

Hao Wu

Beijing You’an Hospital, Capital

Medical University

Department of Infectious Diseases

100069 Beijing

P. R. China

Xiaobin Zhang

FusoGen Pharmaceuticals, Inc.

Ping’an Building B19-A

59 Machang Road

Hexi District

300203 Tianjin

P. R. China

Part I

The Academia – Market Bouncing of Peptide Drugs – Challenges and Strategies in Translational Research with Peptide Drugs

Chapter 1

Peptides as Leads for Drug Discovery

Paul J. Edwards, and Steven R. LaPlante

1.1 Introduction

Peptides have long been used as a source of active material for drug discovery but their use as marketed medicaments is often limited to intravenous administration. The received opinion for peptide drugs delivered via the oral route is that this is a highly challenging endeavor due to the peptides propensity for proteolytic degradation, high clearance, and resulting problems with its delivery, such as low oral bioavailability; all arising from the inclusion of a number of peptide bonds. Scientists have, therefore, sought peptide mimics (peptidomimetics) [1] where degradation of the peptide bonds is hindered (through, for example, N-methylation of the amide nitrogen) and de-peptidization through morphing of the amide (peptide) bonds into peptoids, for example, or through making the molecule more like a small molecule than a peptide. The peptidomimetic would have similar secondary structure as well as other structural features analogous to that of the original peptide, which allows it to displace the original peptide, or protein, from receptors or enzymes [2]. It is in the context of peptidomimics that this work will focus, on peptide-based discovery that has led to advanced (pre-)clinical candidates which are delivered by the oral route of administration.

Fortunately, advances in technologies such as phage-display screening, have enabled the high-throughput discovery of peptides that can inhibit a desired biological reaction for drug discovery purposes [3]. It is now also possible to enable the rapid optimization of peptide mimics to satisfy the many hurdles of drug development. This review intends to expose the fascinating properties and handles that peptides offer and how, through “sensemaking” – that is, the process of gathering and interpreting a body of information relevant to a problem [4] – leading to knowledge building within the project, coupled to synthetic and analytical strategies, successful processes may be adopted that can deliver peptide mimetic (pre-) clinical candidates. We will emphasize our experience from in-house programs where peptide leads were successfully advanced to pre-clinical and clinical candidates (see Figure 1.1). Critical lessons, novel strategies, and examples will be explored at various stages of this process, ranging from the discovery of peptide leads to those that have entered clinical trials. Particularly, we propose that our strategy and work flow (see Schemes 1.1 and 1.2) can represent an expeditious way to render peptides to drugs. This review will be based on exposing major findings/lessons that have commonality among systems and that can advance drug discovery rapidly, if used appropriately. Central for this purpose, it is demonstrated in Figure 1.1 that there can be a significant degree of similarity between the original peptide hit and the advanced analog. Figure 1.1 shows that one can conserve major segments and structural features of relatively weak lead peptides that are required for achieving potency in drugs (colored as red in Figure 1.1). Also critical are truncations and alterations (colored black) and the addition of new features (colored in blue). In summary, we believe that if one can exploit the latent structural functionality on the peptide starting points effectively, then delivery of orally bioavailable drugs derived from a peptide lead becomes possible.

Figure 1.1 Structures of peptide starting points (leads) and their advanced peptide mimics (drugs). The conserved bonds/atoms in both the lead and drug are colored red, whereas the blue and black represent new and modified or deleted sections, respectively.

Scheme 1.1 An overview of the peptide-derived, small-molecule drug discovery. Highlighted in bold text are discovery periods where “sensemaking” and knowledge building cycles can be employed during peptide optimization.

Scheme 1.2 An overview of the drug discovery periods that impact and fuel the “sensemaking,” knowledge building, and SAR exploitation cycles.

1.2 Overview of Process for Transforming Peptides to Peptidomimetics

The pharmaceutical industry has developed complex and fascinating processes for discovering and optimizing leads that become drugs. Scheme 1.1 depicts an overview of how one might progress a peptide to peptidomimetic drug project, indicating the various stages (bold text) that will be exemplified in the projects that follow. Of central importance is the “sensemaking” phase (Scheme 1.2), supported by knowledge building, including mapping of the critical binding parts of the peptide with model creation and peptide truncation, and matching of the free state of the peptide to the bioactive conformation through, for example, rigidification, and de-peptidization.

This chapter will consider a number of case studies against different targets where, after hit identification, the minimal peptide fragments were elucidated and then subjected to conformational rigidification. It is well understood that the binding of a ligand to a macromolecule involves numerous recognition events that are strongly influenced by forces such as van der Waal contacts, electrostatic interactions, solvation effects, and also by ligand to macromolecule shape complementarity. Less well appreciated, but no less important and as critical to these recognition events, are the necessary structural and flexibility adaptations of the ligand and receptor to attain the bioactive complex. Therefore, when considering utilizing a peptide hit as a starting point for drug discovery, the tactics utilized should move beyond the classical “lock-and-key” model to a more holistic approach that incorporates the effects of dynamics and conformational changes. In doing so, rational drug design efforts could be accelerated from the knowledge of these adaptive processes. However, to date, few reports of the application of dynamics and conformational changes have appeared in the literature. In part this is due to the paucity of experimental methods that can provide the type of atomic-level information required. Thus, their importance and impact in drug design have not yet been fully realized.

The process we propose may be summarized (Scheme 1.2) as follows:

Identify lead peptides.Understand properties of the protein.Map critical binding elements of substrate peptide.Understand (by X-ray/NMR) protein–ligand interactions.Increase potency, for example, by using a warhead if necessary, to provide meaningful structure–activity relationships (SAR).Truncation to minimally active peptide (allowing for initial losses in potency as needed).Elucidate free versus bound conformations and ensure SAR designs produce compounds with free conformations matching the bioactive one.In parallel, de-peptidize molecule (e.g., by including bulky side-chains or altering the backbone) and remove any warhead present.

This process will be exemplified through the following examples.

1.3 HCMV Protease

1.3.1 HCMV Protease: Identification and Characterization of Antiviral Inhibitors Targeting the Serine Protease Domain of the Human Cytomegalovirus (HCMV Protease)

Human Cytomegalovirus (HCMV) is a pathogen and member of the herpesvirus family that is highly prevalent in the human population [5]. This virus poses a significant risk to immunocompromized individuals, organ transplant recipients and neonates who acquire the infection congenitally [6, 7]. HCMV encodes a unique protease involved in capsid assembly and this protease enzyme is responsible for processing the assembly protein; the latter protein’s function is analogous to that of the “scaffolding” protein of bacteriophages [8] and its activity is essential to the production of infectious virions [9–12].

The full-length HCMV protease precursor contains 708 amino acids encoded by the UL80 gene. It was discovered that the enzyme can process its own C-terminus and that the protease can also undergo self-processing at the release site near its amino terminus. This cleavage liberates the 256 amino acid catalytic domain, or HCMV protease. Although this enzyme belongs to the serine protease family, differences between familial members exist, as evidenced through X-ray crystallographic analyses [13–16]. These analyses have shown that it possesses a unique protein fold and an unusual catalytic triad (a histidine replacing the more common aspartate). Additionally its activity arises exclusively from its dimer form [17, 18]. Spectroscopic studies [19] have demonstrated that the binding of substrate-based competitive inhibitors results in a conformational change in the enzyme and that catalysis by HCMV protease is performed through an “induced fit” model [20, 21]. Faced with a need to develop a potent HCMV protease inhibitor, the following research process was undertaken.

1.3.2 Mapping Essential Elements of the Substrate Peptides and Determining Structures of Ligands Bound to HCMV

As substrate hydrolysis by HCMV protease was essential for viral capsid assembly, the first task was to decipher the minimal structural elements of the substrates that were required for recognition and hydrolysis. Enzymological studies revealed that peptides which corresponded to 17 amino acids of the release- and maturation-sites (R-site and M-site peptides) were sufficient to induce hydrolysis by HCMV protease (Figure 1.2) [21]. Substitutions of amino acids of the P′ residues (using standard nomenclature [22]) had less of an effect on oligopeptide substrate hydrolysis rates than those of P-side residues [23].

Figure 1.2 Amino acid sequences of peptides/inhibitors with enzymological (kcat/KM) and inhibitory activity (IC50) data [21]. The nomenclature used to denote amino acid positions is given above (i.e., P9 to P4′) [22]. The peptides labeled as R-site, R-mutant, M-site, and M-mutant are 17 amino acids in length, but only the sequence to P4′ is provided for space-saving. The full sequence can be found elsewhere [21].

Reproduced with kind permission from Springer Science+Business Media: Top. Curr. Chem., Exploiting Ligand and Receptor Adaptability in Rational Drug Design Using Dynamics and Structure-Based Strategies, 272, 2007, 264, Steven R. LaPlante, Figure 2.

Differential line-broadening (DLB) NMR was then valuable for understanding which residues were playing a direct role in the binding of the substrate and product recognition by the enzyme [21]. The DLB method [24] provides atomic resolution data and was used as a tool within the project to assess ligand binding. As Figure 1.3a shows, using the N-terminal product peptide (R-product) of the release-site, albeit a weak inhibitor (IC50~3000 μM), it nevertheless bound to the protease, as indicated by the selective resonance perturbations observed when the hydrogen resonances of the peptide were compared in the absence versus presence of HCMV protease (Figure 1.3a). The changes in the peak shape and intensity resulted from fast-exchange averaging between the free and bound states (Figure 1.3b). The broadened resonance of the methyl group of Ala 1 (comparing “A” with “B” in Figure 1.3a) was due to this group becoming pocket exposed upon binding to HCMV (see the illustration in Figure 1.3b). In contrast, the 1H resonance of the methyl group of Thr 9 changed little, as expected for a group that was predominantly solvent exposed in the free and bound states. Using this method, the ensembles of DLB patterns were monitored for the R-product and M-product peptides, and it was discovered that the P4 to P1 residues directly contacted the protease whereas the P9 to P5 residues were solvent exposed [21].

Figure 1.3 DLB mapping that distinguished between residues that were solvent exposed versus those that directly contact the receptor. Shown is the methyl region of the 1H NMR spectrum of R-product showing the P1 and P9 CH3 side-chain doublets [21]. (a) A, free R-product; B, R-product after the addition of HCMV protease at a ratio of 7 : 1; and C, R-product after the addition of HCMV protease and the more potent inhibitor 6 for displacement purposes at a 7 : 1 : 2 ratio. For the sake of clarity, the P1 methyl signals have been skewed slightly. (b) Demonstrates that the DLB method requires fast-exchange binding (averaging) between the free and bound states (on the NMR timescale). (c) The 3D structure of the P5–P1 sequence of R-product when bound to HCMV protease. The 32 overlapped structures determined were derived using transferred NOESY restraints and simulated annealing calculations.

Reproduced with kind permission from Springer Science+Business Media: Top. Curr. Chem., Exploiting Ligand and Receptor Adaptability in Rational Drug Design Using Dynamics and Structure-Based Strategies, 272, 2007, 265, Steven R. LaPlante, Figure 3.

This was also consistent with enzymology findings that peptides spanning the P4–P1′ M-site core were capable of competitively inhibiting catalysis with binding affinity only fivefold less than that of the P4–P4′ substrate [25]. Overall, the ensemble of data indicated that the structural elements of the substrate which were N-terminal to the scissile bond were clearly crucial for complexation to the enzyme. Thus, the first step in the process of identifying the critical binding parts of the substrate peptides has been completed (i.e., P4–P1).

There was also enzymology effort applied to identifying the (peptide) source of the different processing rates observed between the R-site and M-site peptides (Figure 1.2). Enzymology studies followed that mutated the R-site and M-site peptide substrates (i.e., P5 and P4 residues were separately exchanged), as well as through substitution of the P2 residue of the R-site peptide to that of the M-site (Lys to Asn, and called R-mutant in Figure 1.2). It was discovered that the P2 side-chain played a major role in the observed variation in cleavage rates, suggesting that this P2 side-chain influenced the catalytic triad reactivity and so was integral to modulating the catalytic machinery (i.e., shielding the catalytic triad from solvent effects). It was also envisioned that a better understanding of this phenomenon could be exploited for inhibitor potency improvements. Although the promise of this phenomenon was not sufficiently explored in our HCMV protease program, it was successfully exploited in our efforts at optimizing the P2 position in our HCV protease peptidomimetic program, as described later in this chapter.

1.3.3 Improving Peptide Activity to Allow SAR Studies

Serine proteases are a well-studied class of enzyme [26–28]. Despite significant differences in global protein architecture, they possess similar catalytic machinery (triad), which is thought to arise from convergent structural evolution at the enzyme level. The frequently considered “warhead” strategy was, therefore, employed to create substrate-based activated carbonyl inhibitors [29] to boost potency (i.e., compound 1 in Figure 1.2). The improvement in activity also allowed the generation of meaningful SAR. Warheads involved the synthesis of electrophilic ketones which replaced the C-terminus acid of the N-terminal cleavage products [30]. By allowing attack upon the active-site serine, covalent hemiketal adducts were formed that mimic the transition state of the tetrahedral intermediate formed during the catalytic reaction. In this way, a boost in potency was observed with compound 1 (HCMV protease IC50 1.8 μM) [19, 30], compared to the corresponding M-product peptide containing a C-terminal carboxylate (IC50 > 1000 μM; Figure 1.2).

With this improved potency, meaningful SAR then became possible. N-terminal truncation of P5 gave an inhibitor with similar potency (e.g., compare compound 1 versus 2 in Figure 1.4), but losses in potency were observed upon further truncation of the P4 and P3 residues (e.g., compare compound 2 with 3 and 4 in Figure 1.4) [30]. Thus, the P4 to P1 peptidyl segment, as suggested by the NMR experiments described above, played a critical role in ligand binding to the active-site of HCMV protease. Further chemistry efforts focussed on optimizing each of the P1–P4 substituents in turn; once one position had been improved significantly, this moiety was incorporated into optimization of the next position along in the sequence. In this way, the best P2 group was as indicated in compound 2 (IC50 3 μM) and the best P3 was found to be the tert-butyl group (compound 5; IC50 1.1 μM).

Figure 1.4 Inhibitors 1–6 with inhibition constants. The amino-acid positions are designated on top as P5–P1.

1.3.4 Elucidation of the Binding Mode of the Optimized Peptidyl Segment

Contemporaneous with medicinal chemistry efforts to determine which substituents controlled binding and activity, structural research efforts focussed on determining the binding mode of these compounds when bound to HCMV protease. It did not prove possible to use the technique of transferred NOESY (nuclear Overhauser spectroscopy) to determine bound conformation, so ligands were designed that as closely as possible resembled the peptidyl portion of the inhibitors without providing the slow exchange phenomenon arising from inactivation of the enzyme caused by highly electrophilic warhead groups. Thus, methyl ketones 6 [31] and 7 (Figures 1.4 and 1.5, respectively) were designed as NMR-friendly structural probes that could not form covalent complexes when bound to the enzyme, and exhibited fast exchanging binding attributes. In contrast, fluoroketone 5 (Figure 1.4) formed a slow-exchange covalent complex, as determined by 13C NMR experiments [19].

Figure 1.5 (a) Overlap of NMR-derived bound conformations of compound 7 (colored green; 29 conformations), R-product peptide (gray; 32 structures), and the X-ray crystallographic conformations of bound compound 8 (yellow; 4 conformations). (b) Zoomed view of the P3–S3 interaction of compound 8 bound to HCMV protease.

Reproduced with kind permission from Springer Science+Business Media: Top. Curr. Chem., Exploiting Ligand and Receptor Adaptability in Rational Drug Design Using Dynamics and Structure-Based Strategies, 272, 2007, 269, Steven R. LaPlante, Figure 7.

Thus, replacing the C-terminal CF3 with a CH3 provided compounds unreactive toward attack by the active-site serine and so provided a useful structural probe of the bioactive conformation. Transferred NOESY data on compound 7 (vide infra, Figure 1.5a) and the derived distance restraints were applied to determine the family of bound structures shown in green in Figure 1.5a [31]. These studies indicated that they all bound in the extended conformation with a zigzagged backbone, with the P1 and P3 side-chains lying close to one another, and similarly for the P2 and P4 side-chains. The commonality of this structural feature for all three compounds suggested that this bioactive conformation played an important role for binding and activity. The dramatic losses in potency observed upon N-terminal truncation of P4 and P3 was consistent with this observation.

1.3.5 Ligand Adaptations upon Binding

There followed a stage in the project where compounds were designed that preferentially adopted the bioactive conformation in the free-state. Dramatic improvements in potency were attained (Figure 1.6) when a glycine at P3 (compound 9, IC50 > 300 μM) was replaced with an alanine (compound 10, IC50 48 μM) or a tert-butyl group (compound 11, IC50 0.1 μM) [30]. These increases in potency could not be explained by direct contacts with the protease pocket alone. The more overwhelming source for the improvements in potency was a result of the incorporation of the bulkier side-chains which helped to rigidify the compounds to resemble the bioactive conformation.

Figure 1.6 Inhibitors 9–11 with inhibition constants. 13C T1 relaxation data are given for the P3 Cα carbon. For the methylene carbon which has two covalently attached hydrogens, NT1 values are provided, where N is the number of attached hydrogens, to help in interpreting the relative flexibility between different carbon types.

Reproduced with kind permission from Springer Science+Business Media: Top. Curr. Chem., Exploiting Ligand and Receptor Adaptability in Rational Drug Design Using Dynamics and Structure-Based Strategies, 272, 2007, 270, Steven R. LaPlante, Figure 8.

The relative rigidity (or flexibility) of these compounds was monitored by the NMR technique 13C spin–lattice relaxation measurements (13C T1) [20]. Overall, shorter relaxation times are indicative of less flexibility. These experiments indicated that the conformational restriction induced by the bulkier P3 side-chain resulted in a minimization of the overall entropic cost of binding. The bulky group forced the critical P3 backbone into the bioactive, extended conformation in the free state which aided in the formation of two hydrogen bonds that were required upon complexation with HCMV protease (see Figure 1.5b) [20, 32].

In considering the ensemble of mounting information and given the highly related findings in our HSV RNR, HCV protease, and HIV protease programs, as described below, the principle of an easily synthesizable, Val or tert-butyl side chain group, certainly proved to be a valuable medicinal chemistry tool, allowing significant improvements in inhibitor potency, where its incorporation is appropriate.

Later in the HCMV program, a peculiar observation was noted where a significant adaptation of the receptor was seen upon substrate or peptidomimetic binding, as supported by X-ray structure and fluorescence methods (Figure 1.7). In this respect, the process bears elements of protein conformer selection [33]; the peptide guiding the reorganization of the protein around it after association. At this stage of the project, the “sensemaking” indications were that a prerequisite rigidification of the pocket upon peptidyl ligand binding could represent a minimal ceiling in terms of the entropic cost of binding. Thus, this intrinsic feature may ultimately limit any further potency improvements.

Figure 1.7 Peptidyl ligand-induced conformational changes of HCMV protease. Views of the active site region of (a) the inhibited enzyme (in green) covalently bound with ketoamide 8 shown in yellow and (b) the apoenzyme (in gray). Spheres indicate the points at which there was no electron density further along the L9-loop.

Reproduced with kind permission from Springer Science+Business Media: Top. Curr. Chem., Exploiting Ligand and Receptor Adaptability in Rational Drug Design Using Dynamics and Structure-Based Strategies, 272, 2007, 273, Steven R. LaPlante, Figure 11.

In fact, many of our further efforts, and those of competing companies, failed to improve the potency of the peptidyl-activated carbonyl inhibitors. This failure was consistent with the concept of an energetic penalty for protein reorganization on ligand binding. Thus, in this case, many companies aborted efforts at discovering peptidomimetics targeted at this protease (virus). We believe that we went one step further by reasoning that a promising path would be to screen for inhibitors favoring the flexible apo receptor, as we found for beta-lactam inhibitors [5, 20].

1.3.6 Strategic Summary for HCMV Peptide Mimic Design Process

Inhibition of HCMV protease by the peptidyl compounds described above involved the binding of the peptidyl portion to morph the transition state to a catalytically active or activated form of the enzyme. With HCMV protease, an induced-fit catalysis results in binding energy that is expended to compensate for the energy required to convert the enzyme to a thermodynamically less favorable state.

This then explains our strategy of identifying the first lead inhibitor, whereby a C-terminal warhead was incorporated onto N-terminal product peptides. The warhead induced a reversible, covalent mode of binding that mimicked the transition state of substrate cleavage and was utilized to facilitate meaningful SAR studies. However, due to toxicity concerns of deleterious interactions between warheads and proteins, the removal of the warhead from inhibitors was sought, once a level of inhibitor potency had been obtained that allowed the generation of meaningful SAR within the program.

First, the core scaffold of the ligand involved in direct binding to the protease was discovered, followed by utilization of NMR methods to monitor the bioactive conformation of these inhibitors, including the dependence of potency on the free-state flexibility. More potent compounds exhibited similar free and bound (bioactive) conformations, resulting in a reduction in the entropic cost of binding.

Another important finding was that the active-site underwent conformational adaptations upon binding the peptidomimetic ligands and substrates, characterizing HCMV protease as an induced-fit enzyme. The resultant entropic cost required to induce the “activated” state meant there was an intrinsic cost involving the receptor due to changes upon binding. Consistent with this, we and others could not improve inhibitor potencies. In the end, potent ligands were successfully obtained, but the necessity for protein rearrangement precluded further progression of this series of compounds. Throughout this campaign, novel strategies were developed to monitor the bioactive conformation and changes in flexibility of the ligands and receptor. Fortunately, the general nature of these strategies was later applied with success to a highly related campaign that targeted the HCV protease.

1.4 HCV Protease

1.4.1 HCV Protease as an Antiviral Target

Up to 200 million people around the world are infected with the hepatitis C virus [34]. The majority of individuals with persistent HCV infection will develop chronic hepatitis C, a progressive liver disease that can lead to cirrhosis and hepatocellular carcinoma [35–38]. HCV is an enveloped RNA virus belonging to the Flaviviridae family and Hepacivirus genus. As is typical for this family, its positive-sense RNA genome (9.5 kb) encodes a single precursor polyprotein which undergoes proteolytic maturation by enzymes that include host signalases and the viral NS2/3 protease and NS3 protease. NS3 protease (also referred to as HCV protease) is responsible for cleaving four of its non-structural (NS) proteins.

1.4.2 NS3 Serine Protease Possesses a Chymotrypsin-Like Fold

The cleavage sequences have little homology, with the following exceptions: three of four sites have a serine at P1′ and a cysteine at P1, and all four sites have an acidic amino acid at P6 (colored blue, Figure 1.8). Given this permissivity, we employed a peptide substrate having the sequence DDIVPC-SMSYTW [39] for in vitro enzymology studies/assays (colored orange, Figure 1.8).

Figure 1.8 (a) An illustration of the polyprotein encoded by the HCV genome, together with the processing cleavage sites. (b) Sites and sequences cleaved by NS3 protease. The amino-acid positions are designated on top as P6–P3′. The numbers to the left correspond to the numbered sites in (a). Also shown is a model sequence used for enzymology studies and assays (colored orange). Consensus residues are colored blue.

Reproduced with kind permission from Bentham Science Publishers Ltd.: Curr. Med. Chem. – Anti-Infective Agents, Dynamics and Structure-based Design of Drugs Targeting the Critical Serine Protease of the Hepatitis C Virus – From a Peptidic Substrate to BILN 2061, 4, 2005, 113, Steven R. LaPlante and Montse Llinàs -Brunet, Figure 2(A).

1.4.3 Discovery of the Peptide DDIVPC as an Inhibitor of NS3 Protease

Initial efforts focussed on the approach of designing substrate-based activated carbonyl inhibitors which involved the synthesis of N-terminal cleavage products in which the acid of the C-terminus is replaced with an electrophilic ketone. Upon attack by the active-site serine, a stable covalent hemiketal adduct is formed that mimics the transition state of the tetrahedral intermediate formed during the catalytic reaction [26–28]. Concurrent to this work, we sought to determine the bioactive conformation of bound peptidyl ligands through transferred NOESY NMR methods. Since slow binding ligands, such as activated carbonyl inhibitors, are not suitable for such studies [31], our efforts were directed to the synthesis of N-terminal cleavage peptides having an acid C-terminus. Surprisingly, the activity of the N-terminal product peptide DDIVPC of NS3 protease was more active than expected (IC50 71 μM), thus this became the initial inhibitor lead for further SAR efforts [40]. The next efforts were to replace the reactive cysteine at P1 of DDIVPC with a chemically more stable residue and norvaline was found to be a stable replacement with only a fivefold loss in potency [41]. This loss was then recovered by the replacement of P5 by D-Asp and the addition of a P6 acetyl, resulting in compound 12 having an IC50 value of 17 μM. Efforts were then refocussed on replacing the C-terminal acid with an activated-carbonyl warhead to provide an expected boost in activity. However, the C-terminal acid and trifluoromethylketone analogs exhibited comparable activity (compounds 12 and 13 in Figure 1.9) [41].

Figure 1.9 Inhibitors of HCMV and HCV are shown with inhibition constants determined using various enzymes. The amino acid positions are designated on top as P6–P1. The inhibition constants involving HCV were determined using the NS3 protease domain and an NS4A peptide [41]. The other binding constants were determined as described elsewhere [30, 40]. The abbreviations for the proteases are defined as follows: HCV, hepatitis C virus; HLE, human leukocyte elastase; PPE, porcine pancreatic elastase; BPC, bovine pancreatic chymotrypsin; HCMV, human cytomegalovirus protease. Selectivity targets were chosen because they represented closely-related chymotrypsin-like serine proteases, when compared to HCV protease.

Unlike the case of HCMV protease, where only compounds with an activated carbonyl were active enough to be considered as a lead for optimization (compound 15 is much more active than 14, as shown in Figure 1.9) [20, 30], both product peptide and activated carbonyl inhibitors of NS3 protease (compounds 12 and 13) [40–42] had similar potency and were viable starting points as leads. However, the C-terminal acid (compound 12) had superior attributes to the trifluoromethylketone N-peptide (compound 13). For example, compound 12 had a superior selectivity profile, as compared to other serine proteases shown in Figure 1.9. Figure 1.9 shows that compound 13 inhibits human leukocyte elastase (HLE) with an IC50 < 0.06 μM. In contrast, the C-terminal carboxylate inhibitor (compound 12) provides high specificity for NS3 protease with an IC50 17 μM versus IC50 > 1000 μM for other typical serine proteases (e.g., human leukocyte elastase, porcine pancreatic elastase, and bovine pancreatic chymotrypsin) [41]. Warheads would be expected to show reduced selectivity toward other proteins and, hence, would be expected to contribute to greater inhibitor toxicity than the carboxylic acid group. The C-terminal carboxyl derivatives also exhibit other favorable qualities, such as chemical stability and aqueous solubility at neutral pH. For all of the above reasons, the carboxylic acid group was preferred over a warhead for inclusion in inhibitor design for this project.

1.4.4 “Sensemaking” and Knowledge Building: Mapping of the Critical Binding Residues of the Peptide and Creation of an Inhibitor-Protease Model

In trying to improve inhibitor potency by modifying the side-chains and reducing the peptidic nature of our early peptides, medicinal chemists undertook a synthetic study in which single amino acid changes (natural and unnatural) were incorporated into hexapeptides, and the effect on potency was monitored. This exercise led to the finding that a benzylmethoxy proline at P2 (compound 16, Figure 1.10) resulted in a 21-fold improvement in potency (compound 17 versus 18, Figure 1.11) [43, 44].

Figure 1.10 “Sensemaking” using NMR-based knowledge building that probed both structure and dynamics information. (a) The first model of the complex between compound 16 and NS3 protease, with a summary of DLB perturbation mapping data, and transferred 13C T1 data. The complex was determined by docking the bound structure of compound 16 (experimentally determined by transferred NOESY NMR data) to an apo X-ray structure of NS3 protease. For the DLB mapping data, hydrogens of compound 16 are colored blue for resonances in which no broadening perturbations were observed upon binding protease, and hydrogens are colored red when significant resonance broadening was observed upon binding. P5 and P6 were determined to be relatively flexible in the bound state and are not shown above. A summary of the transferred 13C T1 data (placed next to each carbon) is also displayed as the percentage change in 13C T1 before and after the addition of NS3 protease. (b) DLB perturbation data. Selected 1H NMR resonances of compound 16 are shown (with the exception that P1 CH2 is of DDIVPC) when free (colored blue) and after adding small amounts of NS3 protease (colored red). (c) A comparison of free-state (ROESY NMR) versus bound-state (tr-NOESY) conformation and dynamics. Note the similarities of the backbone and differences of the side-chains. (d) Inhibitor 16 is shown with its inhibition constant and free-state 13C T1 relaxation times (next to each carbon). The amino acid positions are designated on top as P6–P1. The inhibition constant was determined using an assay involving the NS3 protease domain and an NS4A peptide [44]. 13C T1 relaxation data are given next to each protonated carbon. In cases where a carbon has more than one covalently attached hydrogen, NT1 values are provided, where N is the number of attached protons and NT1 is the product to help in interpreting the relative flexibility between different carbon types.

Reproduced with kind permission from Springer Science+Business Media: Top. Curr. Chem., Exploiting Ligand and Receptor Adaptability in Rational Drug Design Using Dynamics and Structure-Based Strategies, 272, 2007, 287, Steven R. LaPlante, Figure 20.

Figure 1.11 Inhibitors 17–21 with inhibition constants determined using an assay that included the NS3 protease domain and an NS4A peptide [44]. The amino acid positions are designated on top as P6–P1.

To help guide medicinal chemistry efforts, a ligand-focussed NMR strategy was undertaken to determine which sites of the peptides contacted the protease and which were solvent exposed in the bound state [43, 45]. The differential line-broadening (DLB) NMR experiment [24, 45] was used and, in general, ligand sites that contacted the protease could be identified by specific broadening of the corresponding NMR resonances upon addition of small amounts of protease. When applied to DDIVPC and longer sequences spanning P10–P1, it was found that only specific resonances of P4–P1 experienced broadening (Figure 1.10a and b). No broadening was observed for peptides corresponding to the P′ sequences. This suggested that smaller compounds spanning only P4–P1 should retain a reasonable binding affinity [43].

1.4.5 Knowledge Building: Monitoring Ligand Flexibility in the Free-State and Changes Upon Binding – P3 Rigidification

Next we sought to identify any differences that may exist for compound 16 between the free and bound states. Distance information from the bound state was monitored by the transferred NOESY experiment (Figure 1.10c), and the free state was probed using an NMR ROESY experiment (Figure 1.10c) [43]. This comparison of the distance-related cross-peaks indicated that compound 16 adopts an extended backbone conformation in both states, with important differences observed for the side-chains.

A better means of monitoring dynamic attributes was sought and led the researchers to consider 13C NMR spin–lattice relaxation experiments (13C T1) [43]. 13C T1 relaxation is sensitive to segmental flexibility in the picosecond to nanosecond timescales, and the internal flexibility of drug-like ligands in the free state, which influences the binding affinity to macromolecules via entropic costs, typically occurs within these timescales. The direct correlation of 13C T1 relaxation data with molecular flexibility can be made qualitatively for protonated carbons of free ligands where longer relaxation times are generally indicative of increased segmental flexibility [20, 43].

This work indicated that segmental fluctuations of the norvaline group in the free state was evident given the long and incremental increases of 13C T1 times for the α to δ-carbons (Figure 1.10d, α 0.39, β 0.60, γ 0.94, and δ 3.66 s). Thus, it was reasoned that P1 replacements, which chemically rigidify this side-chain to resemble the bound conformation shown in Figure 1.10a, would likely be more potent owing to a lower entropic cost of binding.

The P2 substituent of 17 also exhibited significant flexibility in the free state, indicating that this aromatic ring underwent fast rotation or spinning along the benzylic/para-carbon axis. Replacement of the phenyl group with a larger naphthyl resulted in an 18-fold improvement in potency (compare compounds 17 and 19, Figure 1.11) [44], which is likely due, in part, to a reduction in rotational rate and the associated entropic cost of binding (vide infra).

The ensemble of data in Figure 1.10 revealed features relevant to the role of P3 for potency. The P3 side-chain played an indirect role in the binding affinity by sterically rigidifying the P3 backbone in the free state to resemble that of the bound extended conformation [20, 43, 45], as was found for the HCMV protease inhibitors. The P3 side-chain had no DLB (Figure 1.10b, and blue-colored hydrogens in Figure 1.10a) indicating that it had no direct binding to the pocket, despite the fact that its removal resulted in significant loss in potency. Thus, given the similarities, and as exploited in the HCMV program, it was suggested to again replace the P3 side-chain with a bulky tert-butyl side-chain (vide infra).

1.4.6 N-Terminal Truncation and Improved P1, P2 and P5 Substituents

DLB and transferred NOESY mapping suggested that the principle binding residues spanned P4–P1. At first, this appeared to contradict earlier attempts to reduce the size of non-optimized hexapeptides by the removal of N-terminal residues. This resulted in shorter peptides with no significant activity. However, a similar exercise was successful by first improving the potency of the hexapeptide series, to better anchor the inhibitor, using beneficial P1, P2, and P5 replacements [44]. For example, 21- and 384-fold improvements were observed when the P2 proline was substituted with a benzyl-methoxyproline (compound 17 versus 18, Figure 1.11) and a naphthyl-methoxyproline (compound 19 versus 18, Figure 1.11), respectively. A further threefold improvement in potency was observed when the P1 norvaline was replaced with a 1-aminocyclopropyl carboxylic (ACCA) (compound 21 versus 20, Figure 1.11). Replacement of the P5 L-aspartic acid with a D-glutamic acid resulted in a 20-fold gain in affinity (compound 20 versus 18, Figure 1.11) [44]. Combining these substitutions simultaneously into a single hexapeptide, resulted in an inhibitor with an IC50 of 0.013 μM (compound 22, Figure 1.12) [44].

Figure 1.12 Inhibitors 22–24 with inhibition constants determined using an assay that included the NS3 protease domain and an NS4A peptide [44]. The amino acid positions are designated on top as P6–P1.

Reproduced with kind permission from Bentham Science Publishers Ltd.: Curr. Med. Chem. – Anti-Infective Agents, Dynamics and Structure-based Design of Drugs Targeting the Critical Serine Protease of the Hepatitis C Virus – From a Peptidic Substrate to BILN 2061, 4, 2005, 119, Steven R. LaPlante and Montse Llinàs -Brunet, Figure 7.

This potent compound then served as the starting compound for a renewed effort at N-terminal truncation. Removal of the P6 residue and the P5 amide resulted in a 69-fold loss in potency (compound 23 versus 22, Figure 1.12) and a tetrapeptide having a simple acetyl capping group resulted in a loss in affinity by 269-fold (compound 24 versus 22, Figure 1.12) [44]. As a result, N-terminal truncation successfully resulted in tetrapeptides that had measurable activity, and that were also more drug-like. The affinity imparted by the P2 naphthyl methoxy and P1 ACCA likely also helped to “anchor” the C-terminal end in the bound state.

Further improvements in potency were sought. It was found that an ethyl appendage on the P1 ACCA provided beneficial contributions to potency (compound 26 is twofold more potent than compound 24, Figure 1.13) and further improvements could be gained by a vinyl appendage (compound 27 is 12-fold more potent than compound 24) [46]. The requirement for a specific stereochemistry of the appendage was observed, given that a fourfold loss in affinity was measured for compound 25 as compared to 24 (Figure 1.13). Overall, it is noteworthy that the P1 ACCA group both improved potency and had a more rigidified, bioactive conformation in the free state as compared to the more flexible P1 Nvl (Figure 1.10d). Like P2, the P1 position played multiple roles in the bimolecular interaction. Due to these multiple roles, “sensemaking” exercises were required to qualitatively deconvolute the roles as much as possible, allowing further exploitation.

Figure 1.13 Inhibitors 24–27 with inhibition constants determined using an assay that included the full-length NS3–NS4A protein [46, 47]. The amino acid positions are designated on top as P4–P1.

A combinatorial chemistry approach was also undertaken to identify alternative P2 substituents [48]. A large variety of aromatic groups was appended to the oxy-prolyl group at P2, resulting in two promising lead compounds 28 and 29 (Figure 1.14). The bound structure of each compound was then determined by transferred NOESY and conformational search methods [49]. These structures are shown on the right in Figure 1.14. Given that the biphenyl group of compound 29 and the quinoline ring of compound 28 occupy different physical space when bound to NS3 protease, it was hypothesized that a compound having three rings would better span both regions of space and could result in an improved affinity (see the theoretical superposition of compounds 28 and 29 on the right of compound 30). A >10-fold improvement in potency was observed (compound 30), but transferred NOESY data for a compound related to 30 showed that the tricyclic group actually bound in the opposite orientation from that predicted. This was understood when considering the multiple roles played by the P2 substituent (vide infra), including its effect on free-state rigidification, solvent shielding of the catalytic triad and electrostatic interactions.

Figure 1.14 Inhibitors 28–30 with inhibition constants determined using an assay that included the NS3 protease domain and an NS4A peptide [40, 49]. The amino acid positions are designated on top as P4–P1. The protease-bound structures of inhibitors 28 and 29 are provide on the right and were determined using transferred NOESY data and a conformational search protocol. Both are overlaid on the right of inhibitor 30 to illustrate the design concept.

Reproduced with kind permission from Bentham Science Publishers Ltd.: Curr. Med. Chem. – Anti-Infective Agents, Dynamics and Structure-based Design of Drugs Targeting the Critical Serine Protease of the Hepatitis C Virus – From a Peptidic Substrate to BILN 2061, 4, 2005, 122, Steven R. LaPlante and Montse Llinàs-Brunet, Figure 10.

1.4.7 Macrocyclization: Linking the Flexible P1 Side-Chain to P3

At this stage of the project, the following considerations were given to improving the properties of the series. The transferred NOESY model of the complex involving compound 16 (Figure 1.10a) revealed that the P3 side-chain lies on the solvent-exposed surface of the protease and in close proximity to the P1 norvaline side-chain [43, 50]. Transferred 13C T1 data (Figure 1.10a) indicated that the P1 side-chain underwent rigidification upon binding the protease [51]. It was speculated that intramolecular linking of the P1 side-chain to the P3 side-chain with a hydrocarbon bridge would lead to a macrocyclic inhibitor which would, in the free state, preferentially adopt the bound conformation observed for compound 16 in Figure 1.10a. A rigid macrocyclic scaffold would also ensure that the P2–P3 amide bond would adopt exclusively the trans-geometry observed in the bound conformation, unlike linear peptides which exist as a mixture of cis- and trans-rotamers.

As an example of the impact that macrocyclization can have on potency [50] and free-state flexibility, the macrocyclic compound 32 in Figure 1.15 (15-membered ring) is 10-fold more potent than the acyclic compound 31. 13C T1 data are shown for both compounds in Figure 1.15, indicating that a reduction in the flexibility of the P1 side-chain (cyclopropyl and vinyl) was achieved by macrocyclization, as shown by the shorter 13C T1 relaxation times for this residue in compound 32 (Figure 1.15, 0.21–0.26 s) as compared to the acyclic compound 31 (Figure 1.15, 0.29–0.32 s).

Figure 1.15 Inhibitors 31 and 32 with inhibition constants determined using an assay that included the full-length NS3–NS4A protein. The amino acid positions are designated on top as P3–P1. The free-state 13C NT1 relaxation times are also provided next to each carbon position (see also the legend of Figure 1.10).

Reproduced with kind permission from Bentham Science Publishers Ltd.: Curr. Med. Chem. – Anti-Infective Agents, Dynamics and Structure-based Design of Drugs Targeting the Critical Serine Protease of the Hepatitis C Virus – From a Peptidic Substrate to BILN 2061, 4, 2005, 123, Steven R. LaPlante and Montse Llinàs-Brunet, Figure 12.

The employed strategies described above [40–57] included an early “sensemaking” and knowledge building phase in which structural and dynamics data were acquired to, (i) understand the bioactive conformation of lead peptides when bound to HCV protease, (ii) identify the important substituents that directly contact the protease pocket, and (iii) determine the differences in conformational flexibility between the free and bound states of ligands. With the rational use of this information, medicinal chemists identified potent hexapeptide compounds with improved P1, P2, and P5 substituents. Efforts to reduce the size and peptidic character resulted in N-terminal truncation to tetra- and tri-peptidic compounds that had novel P1 and P2 substituents. The macrocyclic scaffold was then designed to chemically rigidify the free-state conformation to further resemble the bound-like state, which resulted in a reduction in entropic costs of binding. Having extensive information regarding the binding mode of compounds, medicinal chemists exploited this knowledge in their campaign that eventually led to the BILN 2061 family of compounds. Further SAR efforts at P2 and P4 delivered the first clinical candidate BILN 2061 (ciluprevir) [58, 59] which was the first compound to show proof of concept in humans for a direct acting anti-HCV protease inhibitor, see Figure 1.16. However, the further development of this compound was discontinued due to the observation of cardiotoxicity in high-dose monkey toxicology studies.

Figure 1.16 Structures of the lead peptide DDIVPC and the clinical compound BILN 2061.

Reproduced with kind permission from Springer Science+Business Media: Top. Curr. Chem., Exploiting Ligand and Receptor Adaptability in Rational Drug Design Using Dynamics and Structure-Based Strategies, 272, 2007, 278, Steven R. LaPlante, Figure 14.

1.4.8 HCV Protease Inhibitor BI00201335

With the discovery of cardiotoxicity and subsequent discontinuation of development of the HCV NS3 protease inhibitor BILN 2061 [55], this caused a re-evaluation of active material within the project in an effort to discover novel, non-covalent NS3 protease inhibitors [60].

Work continued on the C-terminal carboxylic acid, (1R,2S