Glycochemistry: Principles: Synthesis, and Applications

ISBN: 0-8247-0538-6 This book is printed on acid-free paper. Headquarters Marcel Dekker, Inc. 270 Madison Avenue, New York, NY 10016 tel: 212-696-9000; fax: 212-685-4540 Eastern Hemisphere Distribution Marcel Dekker AG Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland tel: 41-61-261-8482; fax: 41-61-261-8896 World Wide Web http://www.dekker.com The publisher offers discounts on this book when ordered in bulk quantities. For more information, write to Special Sales/Professional Marketing at the headquarters address above. Copyright 䉷 2001 by Marcel Dekker, Inc.

All Rights Reserved.

Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming, and recording, or by any information storage and retrieval system, without permission in writing from the publisher. Current printing (last digit): 10 9 8 7 6 5 4 3

2

1

PRINTED IN THE UNITED STATES OF AMERICA

Preface

Ever since Emil Fischer established the stereochemical configuration of D-(⫹)-glucose at the end of the nineteenth century, carbohydrate chemistry has become an important branch of organic chemistry. Steady advances in carbohydrate synthetic and analytical methods have been made over the past hundred years. However, it is the past two decades that have seen remarkable new discoveries in the biology of carbohydrates and provided renewed impetus to the synthesis of complex glycoconjugates and pursuit of their medicinal and industrial applications. Glycobiology has emerged as an internationally recognized field of study and a potentially promising route to the discovery of novel medicines. Critical biological processes—including regulation of the growth and mobility of cells, immune responses, and reactions of cells to hormones and growth factors—all depend on carbohydrates. In addition, many viruses and bacteria use cell surface carbohydrates to get into cells and initiate infections. The diversity of roles played by carbohydrates makes them exciting new targets for the development of novel treatments for cancer, infectious diseases, inflammation, and cardiovascular disorders. The spur in glycobiology research resulted in the appearance of several enormously successful carbohydrate-based drugs including but certainly not limited to heparin (an anticoagulant) and Relenza (an antiviral). Increasing numbers of organic chemists are venturing into the field of glycochemistry to bring in their own chemical expertise and to expand the area into a truly multidisciplinary field. A major motivation among carbohydrate chemists is to develop tools for probing the biological functions of glycosylation. Genome sequencing projects present a vast repertoire of biosynthetic enzymes that may be targeted by selective carbohydrate-based drugs. Advances in automated oligosaccharide synthesis are highly significant as complex structures will soon be widely attainable and exploited in studies of biological phenomena. Their integration into complex assemblies (polymers, dendrimers, and liposomes) will provide cell surface mimics of unparalleled fidelity. Studies of glycobiology will benefit from the convergent application of chemical and biological techniques. Thus, in the past few years, glycochemistry has advanced from studies of chemical glycosylation reactions and synthesis of oligosaccharides to iii

iv

Preface

chemo-enzymatic and solid-phase synthesis of glycoconjugates, carbohydrate combinatorial libraries, multivalent glycoconjugates, carbohydrate–peptide hybrids, and much more. It is the aim of this book to capture the essence of these latest developments and to provide a comprehensive review of modern glycochemistry in one collected volume. In recognition of the great potential applications of glycochemistry in the pharmaceutical and chemical industries, we solicited chapters that contain pertinent examples of the development of carbohydrate-based pharmaceuticals and commercial polymers. Each chapter is written by experts in their respective fields. Many are young scientists who have just begun research programs in carbohydrate chemistry. Although some chapters include relatively detailed experimental procedures for novel synthetic methodology, the book is intended to provide a reference framework for the latest developments in the field of glycochemistry. Recent trends in glycochemistry are presented in three parts: synthesis, principles and applications. The first six chapters provide a comprehensive, up-to-date review on the chemical synthesis of complex carbohydrates for their potential use in biological systems. The following seven chapters reveal some fundamental principles that are used to design and exploit carbohydrates for their effects in biological settings. The remaining five chapters examine the applicability of enzymes towards the chemo-enzymatic synthesis and modification of carbohydrates and polysaccharides. In Chapter 1, Peter Seeberger reflects on the promising strategies for the solid support synthesis of oligosaccharides and glycoconjugates. The chapter puts in focus the glycal assembly method and reveals protecting group strategies for carbohydrate components, synthesis of biologically important oligosaccharides, and selective glycosylation strategies. Chapters 2 and 3 bring to light the potential of strategies for stereoselective glycosylation. David Gin examines a number of methodologies used in the glycosylation of 1-hydroxy donors and points out that in the direct dehydrative coupling with an appropriate dehydrating agent, the formation of undesired by-products of hemiacetal self-condensation can be minimized even when only a slight excess of the glycosyl acceptor components is employed. David Crich examines the glycosyl triflates as extremely reactive glycosyl donors. They are prepared from anomeric sulfoxides or thioglycosides upon activation with triflic anhydride or benzenesulfenyl triflate and provide access to the elusive ␤-mannopyranoside, which can be found in many biologically important oligomers. Hydrolytic vulnerability of O-linked glycosides makes the C-glycosides attractive sugar mimics and stable drug candidates. Maarten Postema and Daniel Calimente (Chapter 4) report on the many methods for the preparation of both ␣- and ␤-Cglycosides that have been developed, as well as some new approaches currently in the literature. One of the key issues surrounding the synthesis of C-glycosides concerns the stereochemistry of the C-glycosidation step, which has seen a remarkable improvement over the years. As C-glycosides make their way to the forefront of carbohydrate synthesis, many mild conditions will be examined, and these sophisticated synthetic methods will facilitate their production. Todd Lowary (Chapter 5) focuses on two major polysaccharides containing Darabinofuranose, namely, arabinogalactan and lipoarabinomannan, in a quest to identify new antibiotics. These glycopolymers are found as important components of the cell wall of the Actinomycete family. Although there is now an understanding of the

Preface

v

structure of mycobacterial arabinan, the biosynthetic pathway is yet to be understood. These investigations will depend critically on the access to synthetic oligosaccharides and their analogs, which are reviewed extensively. In Chapter 6, Biao Yu and Yongzheng Hui discuss the relevance of saponins and their chemical synthesis. This important class of glycosidic steroids is found mainly in plants used as herbal medicines. However, a major issue surrounding the saponins is their ambiguous mechanism of action. The authors discuss two important strategies for the construction of saponins, each involving protective group manipulation and various glycosylation methodologies. Increasing interest in traditional herbal medicine and in ‘‘carbohydrate drugs’’ should bring further attention to this group of natural glycosides. In Chapters 7–13, the focus shifts to the biochemistry of carbohydrates and the exploration of fundamental principles related to cell signaling and protein–carbohydrate interactions on a molecular level. Sialic acids are a class of nine-carbon monosaccharides found at the termini of oligosaccharides in many mammalian cellular systems. Randall Halcomb and Mark Chappell (Chapter 7) summarize the most current sialylation technology. Both chemical and enzymatic methodologies for coupling sialic acid to various carbohydrate moieties are examined while providing the reader with chemical detail that one requires to work in this field. In Chapter 8, David Mann and Laura Kiessling explore a number of fundamental binding principles such as the energetics surrounding hydrogen bonding, metal chelation, hydrophobic effects, coulombic interactions, and the role of water in protein–carbohydrate interactions. The information provides a prelude to the concept of multivalency in understanding the principles, methodologies, and techniques used to increase an overall binding affinity between carbohydrates and proteins. One technique used to achieve multivalency revolves around linking carbohydrates to polymeric supports so as to increase the number of binding sites as well as obtain optimal distal properties for effective binding. Rene´ Roy (Chapter 9), another pioneer in the application of multivalency effects in glycobiology, discusses some novel multivalent glycotools for biochemical investigations related to sialic acid. Roy describes methodology to produce sialyloligosaccharides, functionalized sialosides, sialoside clusters, N- and O-linked glycopeptides, amphiphilic calix[4]arene nanostructures, glycopolymers, and glycodendrimers. Hybrid dendrimer–polymer glycoclusters are also introduced as new and exciting complexes to show increased binding affinity. In Chapters 10 and 11, Jalal Haddad and Shahriar Mobashery, Lakshmi Kotra, and Mei-Zheng Liu examine the diversity of aminoglycosides and their significant role as antibiotics. They reveal the specificity of aminoglycosides for binding to RNA constructs derived from the HIV–RRE and TAR RNA activator regions and examine the varying inhibitory effects of aminoglycans on different microorganisms. They also introduce methodologies in the synthesis of aminoglycoside antibiotics by examining the chemical strategies that have been developed. In Chapter 12, Bryan Yeung, Pek Chong, and Peter Petillo survey the chemical preparation of the glycosaminoglycans, oligosaccharides of hyaluronan, chondroitin sulfate, dermatan sulfate, keratan sulfate, heparin, and heparan sulfate. Despite the biological importance of these ubiquitous carbohydrate polymers, there are few reports of the chemical syntheses of glycosaminoglycan (GAG) oligosaccharides. The multiple functionality of GAGs provides an excellent scaffold on which structure–

vi

Preface

activity relationships can be studied, but their syntheses present an unparalleled challenge to the synthetic carbohydrate chemist. The authors recognize the difficulty in the construction of these highly functionalized carbohydrates and present newly developed protecting groups as well as glycosylation procedures to synthetically access GAG fragments. In Chapter 13, Jacquelyn Gervay-Hague and Thomas Weathers, Jr., discuss some of the more recent chemistry behind pyranosyl sugar amino acids and the important role they play in biological systems. The authors make a complex group of carbohydrates seem relatively simple and interesting by addressing the important issues and illustrating them with clear, concise pictorial schemes. Included are the synthesis of C-aminoglycosides, protecting groups, the chemistry of glycosidic linkage, strategies for block synthesis, and the introduction of solid-phase synthesis of mixed sugar amino acid conjugates. The emphasis of this chapter is twofold: to describe the synthesis of aminoglycans oligomers and to examine the stable secondary structural characteristics of amido-linked oligomers. Chapters 14–18 are meant to pioneer a chemo-enzymatic approach to the construction of carbohydrates. One of the underlying themes of this book is the exploitation of the complexity of carbohydrates and applying various methodologies to simplify their production. In Chapter 14, Xiangping Qian, Keiko Sujino, Monica Palcic, and Murray Ratcliffe discuss the potential of glycosyltransferases in oligosaccharide synthesis. The authors focus on the enzymes responsible for creating various glycosidic linkages and the sugar donor requirements for glycosyltransferases as well as giving some detailed information for applying this knowledge to largescale carbohydrate synthesis. Chapter 15, written by H. N. Cheng and Qu-Ming Gu, discusses the application of biotransformations to polysaccharides modification. Polysaccharides are natural materials ideally suited for enzymatic modifications. The chapter introduces lipasecatalyzed and ␤-galactosidase–catalyzed modifications of carbohydrate polymers. Peter Andreana, Wei Zhang, and Peng George Wang (Chapter 16) take a close look at the ␣-gal epitope as a case study. This carbohydrate sequence is responsible for the hyperacute organ rejection associated with xenotransplantation. Efficient production of this carbohydrate epitope can lead to increased studies directed toward solving the xenotransplantation conundrum. The authors examine both chemical and enzymatic routes for synthesizing the epitope and present some important physical properties associated with binding of the carbohydrate to the protein (epitope–antibody interaction). Bacterial glycosyltransferases, the topic of Chapter 17, by Przemyslaw Kowal, Xi Chen, and Peng George Wang, are enzymes responsible for the assembly of bacterial cell walls (e.g., succinoglycan) and lipopolysaccharides or polysaccharide structures attached to the lipids of the outer membrane of gram-negative bacterial cells. Lipopolysaccharides are at the forefront of bacterial interactions with the outside world. These structures have been found to be essential in processes ranging from root nodulation to human pathogenicity. Therefore studies on the corresponding glycosyltransferases are of great importance and interest. The authors mention an emerging technology using biosynthetic pathway engineering to produce carbohydrates on a large scale. We envision that such technology will be rapidly developed into commercial use.

Preface

vii

Scarlett Goon and Carolyn Bertozzi close the book with a chapter discussing metabolic substrate engineering as a tool for glycobiology. In this approach, metabolic pathways are intercepted with unnatural monosaccharide substrates, leading to their incorporation into cell surface oligosaccharides. This provides an innovative method for studying the functions of the surface sugar structures. Modified substrates might also block biosynthetic enzymes, producing phenotypes similar to those induced by glycosyltransferase inhibitors. This chapter focuses on the flux of carbohydrate precursors and synthetic carbohydrate analogs through the metabolic pathways of the cell and the information that can be gained from investigating such processes. This book is intended for graduate students and researchers in carbohydrate chemistry, biochemistry, medicinal chemistry, and glycobiology in both academic and industrial laboratories. We feel privileged to have attracted such a distinguished group of investigators and express our sincerest gratitude for their time and effort in making this endeavor a meaningful contribution. Peng George Wang Carolyn R. Bertozzi

Contents

Preface Contributors Introduction: New Challenges in Carbohydrate Research

Chi-Huey Wong

1. Solid Phase Oligosaccharide Synthesis Peter H. Seeberger

iii xi xv 1

2. Dehydrative Glycosylation with 1-Hydroxy Donors David Gin

33

3. Chemistry of Glycosyl Triflates: Synthesis of ␤ -Mannopyranosides David Crich

53

4. C-Glycoside Synthesis: Recent Developments and Current Trends Maarten H. D. Postema and Daniel Calimente

77

5.

D-Arabinofuranosides from Mycobacteria: Synthesis and Conformation Todd L. Lowary

133

6. Chemical Synthesis of Bioactive Steroidal Saponins Biao Yu and Yongzheng Hui

163

7. Recent Developments in Technology for Glycosylation with Sialic Acid Randall L. Halcomb and Mark D. Chappell

177

8. The Chemistry and Biology of Multivalent Saccharide Displays David A. Mann and Laura L. Kiessling

221

9. Designing Novel Multivalent Glycotools for Biochemical Investigations Related to Sialic Acid Rene´ Roy 10. Aminoglycoside Antibiotics: Structures and Mechanisms of Action Jalal Haddad, Lakshmi P. Kotra, and Shahriar Mobashery

277 307 ix

x

Contents

11. Methodologies in Syntheses of Aminoglycoside Antibiotics Jalal Haddad, Mei-Zheng Liu, and Shahriar Mobashery

353

12. Synthesis of Glycosaminoglycans Bryan K. S. Yeung, Pek Y. C. Chong, and Peter A. Petillo

425

13. Pyranosyl Sugar Amino Acid Conjugates: Their Biological Origins, Synthetic Preparations, and Structural Characterization Jacquelyn Gervay-Hague and Thomas M. Weathers, Jr.

493

14. Glycosyltransferases in Oligosaccharide Synthesis Xiangping Qian, Keiko Sujino, R. Murray Ratcliffe, and Monica M. Palcic

535

15. Biotransformation of Polysaccharides H. N. Cheng and Qu-Ming Gu

567

16. From a Carbohydrate Sequence to Extensive Research Programs: The ␣-gal Epitope as a Case Study Peter R. Andreana, Wei Zhang, and Peng George Wang

581

17. Microbial Glycosyltransferases Przemyslaw Kowal, Xi Chen, and Peng George Wang

625

18. Metabolic Substrate Engineering as a Tool for Glycobiology Scarlett Goon and Carolyn R. Bertozzi

641

Index

675

Contributors

Peter R. Andreana Michigan

Department of Chemistry, Wayne State University, Detroit,

Carolyn R. Bertozzi Departments of Chemistry and Molecular and Cell Biology, Howard Hughes Medical Institute, and University of California, Berkeley, California Daniel Calimente Michigan

Department of Chemistry, Wayne State University, Detroit,

Mark D. Chappell Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado Xi Chen

Department of Chemistry, Wayne State University, Detroit, Michigan

H. N. Cheng

Hercules Incorporated, Wilmington, Delaware

Pek Y. C. Chong Department of Chemistry, University of Illinois at Urbana-Champaign, Urbana, Illinois David Crich Illinois

Department of Chemistry, University of Illinois at Chicago, Chicago,

Jacquelyn Gervay-Hague son, Arizona

Department of Chemistry, University of Arizona, Tuc-

David Gin Department of Chemistry, University of Illinois at Urbana-Champaign, Urbana, Illinois Scarlett Goon Departments of Chemistry and Molecular and Cell Biology, Howard Hughes Medical Institute, and University of California, Berkeley, California Qu-Ming Gu Jalal Haddad igan

Hercules Incorporated, Wilmington, Delaware Department of Chemistry, Wayne State University, Detroit, Mich-

Randall L. Halcomb Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado xi

xii

Contributors

Yongzheng Hui State Key Laboratory of Bio-organic and Natural Product Chemistry, Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, Shanghai, China Laura L. Kiessling Departments of Chemistry and Biochemistry, University of Wisconsin–Madison, Madison, Wisconsin Lakshmi P. Kotra Michigan

Department of Chemistry, Wayne State University, Detroit,

Przemyslaw Kowal Michigan

Department of Chemistry, Wayne State University, Detroit,

Mei-Zheng Liu Michigan

Department of Chemistry, Wayne State University, Detroit,

Todd L. Lowary Ohio

Department of Chemistry, The Ohio State University, Columbus,

David A. Mann Department of Chemistry, University of Wisconsin–Madison, Madison, Wisconsin Shahriar Mobashery Michigan Monica M. Palcic Alberta, Canada

Department of Chemistry, Wayne State University, Detroit,

Department of Chemistry, University of Alberta, Edmonton,

Peter A. Petillo Department of Chemistry, University of Illinois at Urbana-Champaign, Urbana, Illinois Maarten H. D. Postema Detroit, Michigan Xiangping Qian Alberta, Canada

Department of Chemistry, University of Alberta, Edmonton,

R. Murray Ratcliffe Rene´ Roy Canada

Department of Chemistry, Wayne State University,

SYNSORB Biotech Inc., Calgary, Alberta, Canada

Department of Chemistry, University of Ottawa, Ottawa, Ontario,

Peter H. Seeberger Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts Keiko Sujino Canada

Department of Chemistry, University of Alberta, Edmonton, Alberta,

Peng George Wang Michigan

Department of Chemistry, Wayne State University, Detroit,

Thomas M. Weathers, Jr. son, Arizona

Department of Chemistry, University of Arizona, Tuc-

Contributors

xiii

Bryan K. S. Yeung Department of Chemistry, University of Illinois at UrbanaChampaign, Urbana, Illinois Biao Yu State Key Laboratory of Bio-organic and Natural Product Chemistry, Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, Shanghai, China Wei Zhang

Department of Chemistry, Wayne State University, Detroit, Michigan

Introduction: New Challenges in Carbohydrate Research

Over the past 30 years, most of the important discoveries in the biological sciences have been associated with proteins and nucleic acids, and thus chemists have been most active in conducting their research to address the relationship between the structure and function of these macromolecules as well as to develop small molecules as mechanistic probes or drug candidates to target these two types of biopolymers. This rapid progress of development is undoubtedly due to—other than the importance of proteins and nucleic acids in biological functions—the availability of various powerful tools (such as the polymerase chain reaction, solid-phase synthesizers, sequence analysis, and sensitive assay systems) to both chemists and biologists, and the close interplay between these two disciplines. Carbohydrates, especially oligo- and polysaccharides and their conjugates with proteins and lipids, have been neglected. Although carbohydrates display enormous structural and functional diversity—fulfilling roles as energy sources and as elements for recognition and for the control of structure, function, and dynamics of proteins —their precise functions at the molecular level are poorly understood. This slow pace of development in carbohydrate research is due mainly to the lack of tools available to the biologist interested in carbohydrates, and to a lesser extent to the chemist. No amplification method is available to prepare functionally important oligo- or polysaccharides in quantities large enough for structural analysis; no general method is available for the synthesis of glycoproteins with well-defined carbohydrate structures to investigate the roles that carbohydrates play in glycoprotein structure and function; and no solid-phase synthesizer is available for the synthesis of oligoand polysaccharides. From a medicinal chemistry point of view, carbohydrates are an uninteresting class of molecules for drug development because they are too difficult to synthesize on a large scale and too hydrophilic to have good bioavailability, and they are generally orally inactive and unstable. Perhaps the most fundamental problem is that carbohydrates bind their receptors or enzymes with weak affinity, usually with dissociation constants in the millimolar range. Polyvalent interactions are, however, xv

xvi

Introduction

often used in sugar-mediated intercellular recognitions to increase affinity and specificity and to control or regulate function (such as signaling and cell surface protein shedding). The weak affinity in sugar–receptor interactions is attributed to a few causes: 1) there is a lack of hydrophobic groups in sugars which are often dominant in high-affinity receptor–ligand interactions; 2) the energy associated with multiply hydrogen-bond and other electrostatic interactions in sugar–receptor recognitions is significantly diminished by competition from bulk solvent; and 3) the flexible nature of many hydrogen-bond interactions results in a significant entropic penalty when they become constrained in space upon binding. In addition, it appears that not all functional groups of a carbohydrate ligand are essential for binding. Despite these bottlenecks, it is important to develop new strategies and methods to study carbohydrate-mediated biological recognition processes and to investigate their mechanisms of action, since understanding the mechanism of carbohydrate function may lead to the development of carbohydrate-based therapeutics. Recent advances in carbohydrate chemistry and biology have helped to solve some of the problems. Of particular significance are, for example, the development of new methods for the synthesis of complex carbohydrates, glycolipids, and glycoproteins, the design of carbohydrate mimetics including polyvalent mimetics and glycosyltransfer enzyme inhibitors, the manipulation of cell surface carbohydrates to probe intercellular interactions, the development of carbohydrate-based antibiotics targeting RNA and cell surface peptidoglycans, the study of sulfation in carbohydrate recognition, the study of carbohydrate effect on protein structure and function, and the development of carbohydrate-based vaccines. Future efforts perhaps may be directed to: 1) determine the structures of glycosyltransferases and postmodification enzymes such as sulfotransferases and carbohydrate receptors involved in metastasis, inflammation, and immunoresponse (e.g., CD1-glycolipid–TCR complex); 2) understand carbohydrate-mediated intercellular interaction and signaling; 3) design small molecules targeting carbohydrate–receptor and carbohydrate–enzyme interactions; and 4) develop efficient and rapid methods for the synthesis of oligosaccharides and glycoproteins and for their structural and functional analyses. In this new era of functional genomics, it is anticipated that more tools will be developed to facilitate discovery in glycobiology and that carbohydrate-based therapeutic breakthroughs will be realized. Chi-Huey Wong The Scripps Research Institute La Jolla, California

1 Solid Phase Oligosaccharide Synthesis Peter H. Seeberger Massachusetts Institute of Technology, Cambridge, Massachusetts

I.

INTRODUCTION

Polysaccharides are the structurally most complex of the major classes of biopolymers. While nucleic acids and proteins are linear assemblies, polysaccharides are structurally and stereochemically more diverse. Additional complexity is added to the polysaccharide structure by the formation of glycoconjugates such as glycolipids and glycoproteins [1]. Oligosaccharides in the form of glycoconjugates mediate a variety of events, including inflammation, immunological response, metastasis, and fertilization [2,3]. Cell surface carbohydrates act as biological markers of various tumors and as binding sites for other substances, including pathogens [4]. The increased understanding of the important roles oligosaccharides and glycoconjugates play in fundamental life-sustaining processes has stimulated a need for access to usable quantities of these materials. Glycoconjugates commonly exist as microheterogeneous mixtures and are difficult to isolate in homogeneous form, therefore yielding only small amounts of the desired material. The problems associated with isolation from natural sources give rise to opportunities for chemical synthesis [5]. The developments in synthetic carbohydrate chemistry have centered on finding solutions to two important challenges: the need to differentiate similar functionality (hydroxyl or amino) contained on each monosaccharide and the need to induce the selective formation of a variety of glycosidic linkages. Unlike linkages found in the other two classes of repeating biopolymers, each glycosidic bond to be fashioned in a growing oligosaccharide ensemble constitutes a new locus of stereogenicity. In response to these two key challenges, a wide variety of protecting group strategies 1

2

Seeberger

[6] and increasingly powerful and selective glycosylating agents have been developed [7]. Still, the synthesis of complex oligosaccharides and glycoconjugates remains a difficult and time-consuming task. The introduction of solid support synthesis methods for preparation of structurally defined oligopeptides [8] and oligonucleotides [9] has led to tremendous improvements in terms of synthesis speed and efficiency. The advantages of solid phase synthesis, such as maximized yields by use of excess reagents, ease of purification, and synthesis automation, are now well appreciated. The biopolymers obtained by these rapid synthetic methods significantly impacted the fields of peptide and nucleic acid biochemistry. Glycobiology would benefit in many ways from straightforward synthetic methods, which could be used by nonexperts to prepare oligosaccharides. Since the level of complexity associated with the synthesis of oligosaccharides on a polymer support is much greater than that associated with the other two classes of repeating biooligomers, this task seemed for a long time too difficult to tackle. The development of protocols for the solid support synthesis of oligosaccharides and glycopeptides requires the scientists in the field to find solutions to several problems: (1) selection of an overall synthetic strategy and development of methods for attachment of the carbohydrate to the polymeric support through the ‘‘reducing’’ or the ‘‘nonreducing’’ end, (2) choice of a solid support material, (3) selection of a linker (‘‘support-bound protecting group’’) that is stable during the synthesis but can be easily cleaved when desired, (4) a highly flexible protecting group strategy, (5) stereospecific and high-yielding coupling reactions, and (6) ‘‘on-resin’’ methods to monitor chemical transformations. Pioneering efforts in the area of solid support oligosaccharide synthesis were undertaken in the early 1970s by several research groups, utilizing the limited set of glycosylating agents available at the time. While some fundamental issues were explored, the lack of glycosylation reactions compatible with polymeric supports impeded further progress. This early work has been reviewed [10] and is not covered in this chapter. Since these initial attempts, a great deal of progress has been achieved in assembling relatively complex carbohydrate ensembles through simplification and refinement of protecting group strategies and the development of new and powerful glycosylation methodology. This progress is a result of the efforts that have focused on solid support oligosaccharide synthesis since the early 1990s. The different approaches to solid support oligosaccharide synthesis by chemical rather than enzymatic methods are summarized in this chapter. Much effort has been focused on the use of soluble, polyethylene glycol (PEG) based polymeric supports for the synthesis of oligosaccharides. Reaction development using these soluble polymers has proven quite facile, since the reactions are run under homogeneous conditions and conventional analytical methods may be employed [11]. A number of glycosylation strategies have been explored by Krepinsky and coworkers [12] and van Boom et al. [13]. An orthogonal glycosylation strategy in conjunction with soluble polymeric supports has been developed [14]. In 1996 a very innovative ‘‘gatekeeper’’ approach [15] to the synthesis of oligosaccharides containing ␤-mannosidic linkages was introduced. This work, which used a PEG polymer [16], was reviewed in 1998 [17] and is not covered in this chapter.

Solid Phase Oligosaccharide Synthesis

II.

3

SYNTHETIC STRATEGIES

Three major strategies have been explored for the solid support synthesis of oligosaccharides and glycoconjugates. In one variation, the first carbohydrate is anchored to the support via its ‘‘reducing’’ end (see Scheme 1, Case 1). The carbohydrate bound to the solid support functions as an acceptor in the coupling event to a solution-based donor (D). In the cycle, a unique acceptor hydroxyl must be exposed

Scheme 1

Glycosyl acceptor (Case 1) and donor (Case 2) bound to the solid support, and bidirectional synthesis (Case 3): S, solid support; P, unique protecting group; X, activating group; asterisk, uniquely differentiated hydroxyl group.

4

Seeberger

in the now elongated, resin-bound carbohydrate construct. In Case 1, this strategy virtually demands for the donor (D) employed in the preceding glycosidation step a uniquely removable blocking group at the site of the next proposed elongation. The need to expose the unique hydroxyl group in the context of the polymer support will necessitate multiple functional group manipulations in synthesizing D. Alternatively, the carbohydrate that is to undergo elongation may be mounted to the support somewhere in a ‘‘nonreducing’’ region, thereby making the reducing end available as a glycosyl donor for coupling with solution based acceptor A (Case 2). The use of A, of course, demands that the precise acceptor site be properly identified. In anticipation of the next coupling event, the reducing end of acceptor A is so functionalized that a new donor capability can be installed at the anomeric carbon of the elongated construct. This approach necessitates in the acceptor an anomeric group that does not serve as a glycosyl donor itself but rather may be converted in a straightforward manner into a glycosyl-donating moiety. A hybrid of both strategies is the bidirectional synthesis approach (Case 3), which is based on the orthogonal glycosylation concept. The first monosaccharide (A/D) may serve as glycosyl donor as well as glycosyl acceptor and is attached to the support matrix through a position other than the anomeric center. This glycoside contains an anomeric leaving group (as for Case 2), and in addition exhibits an uniquely distinguished acceptor site (analogous to Case 1). During the initial phase of the synthesis the support-bound sugar functions as a glycosyl acceptor A without compromising the chemical integrity of the anomeric moiety. In this fashion, oligosaccharide synthesis following the Case 1 paradigm may be carried out. If an orthogonal set of glycosylating conditions is used, the support-bound sugar may be activated to serve as a glycosyl donor in chain elongation to follow a Case 1 synthetic scheme. This chapter discusses the different approaches to solid phase oligosaccharide synthesis developed to date, grouped by the overall synthetic paradigm they obey.

III.

THE GLYCOSYL DONOR BOUND STRATEGY: THE GLYCAL ASSEMBLY APPROACH

A.

Overview of the Glycal Assembly Approach

While the donor-bound paradigm (Case 2) minimizes the number of protecting group manipulations that have to be carried out on the solid support, it mandates the presence of a latent glycosyl donor moiety in the solution-based glycosyl acceptor. Glycal building blocks fulfill these requirements, since they serve as glycosyl acceptors but may be converted into powerful glycosylating agents. Danishefsky and coworkers successfully explored the application of the glycal assembly method to the solid support synthesis of oligosaccharides and glycoconjugates [18]. The general principle of the glycal assembly method is outlined in Scheme 2. Attachment of a glycal to a solid support through the nonreducing end results in the presence of a terminal glycal function that may be readily converted into donor 2. Since glycals are employed as the solution-based acceptor, the protecting group scheme used for the identification of hydroxyls predestined for glycosylation remains relatively simple. Compound 2 could be a 1,2-anhydro sugar [19], derived by reaction with an epoxidizing agent, or it could be a transient species, derived by activation with an

Solid Phase Oligosaccharide Synthesis

5

Scheme 2 Use of the glycal assembly method in a general strategy for the synthesis of oligosaccharides on a solid support.

iodonium source [20]. Support-bound 2 acts as a glycosyl donor when treated with a solution of acceptor glycal 3, along with any necessary agents to promote the glycosylation, thereby forming 4. The process can be repeated to assemble the desired oligosaccharide, followed by retrieval from the support and purification by chromatographic methods. All studies on the solid support synthesis of oligosaccharides by the glycal assembly approach employed a polystyrene 1% divinylbenzene copolymer, which is commonly used in solid support peptide synthesis because of its high loading capacity, compatibility with a wide range of reaction conditions, and low price. The first glycal was linked to the solid support through a disilane linkage that could be cleaved rapidly and completely by treatment with fluoride [21]. The validity of the approach was first demonstrated by the synthesis of a linear tetrasaccharide [22] and a hexasaccharide 13 [23] as outlined in Scheme 3. Polymerbound galactal 5 was converted to the 1,2-anhydro sugar 6 by epoxidation with 3,3dimethyldioxirane [24]. Polymer-bound 6 acted as a glycosyl donor when reacted with a solution of 7 in the presence of zinc chloride, resulting in the formation of disaccharide 8a. Upon repetition, this glycosylation procedure accommodated the secondary alcohol glycosyl acceptor 10 as well as disaccharide acceptor 12. Fluoridolysis with tetrabutylammonium fluoride (TBAF) was used to cleave the desired products from the polymeric support and furnish hexasaccharide 13 in 29% overall yield from 5 [16]. B.

Solid Phase Synthesis of Blood Group Determinants

Carbohydate blood group determinants in the form of glycoproteins or glycolipids were found to play key roles in cell adhesion and other binding phenomena [25,26]. Furthermore, glycoconjugates related to these blood group substances have been recognized as markers for the onset of various tumors. These tumor-associated antigens are currently being studied in vaccines for cancer immunotherapy [27,28]. The

6

Seeberger

Solid Phase Oligosaccharide Synthesis

Scheme 3

Solid phase synthesis of a hexasaccharide by the glycal assembly method.

7

8

Seeberger

glycal assembly approach outlined above found its first applications in the synthesis of a H-type 2 tetrasaccharide (Scheme 4) [29,30]. Treatment of polymer-bound 1,2-anhydro sugar 6 with a solution of glucal acceptor 14 provided disaccharide 15. Upon opening of a 1,2-anhydro sugar during glycosylation, a C2 hydroxyl group is exposed, which may in turn serve as a glycosyl acceptor to form branched oligosaccharides [31]. Compound 15 was fucosylated using a solution of fucosyl donor 16 [32] to furnish trisaccharide 17. Treatment of 17a with TBAF provided trisaccharide glycal 17b in 50% overall yield from 5. Because of the lack of solid support methodology to fashion glycosidic linkages bearing C2-acylamino functions, solution phase chemistry had to be employed to access the H-type 2 blood group determinant glycal 18. Ready functionalization at the reducing end was achieved through the terminal glycal handle. The Lewisb blood group antigen (Leb) is of particular interest because it has been identified as a mediator for the binding of Helicobacter pylori to human gastric epithelium [33]. Clinical studies have identified H. pylori as a causative agent in gastric and duodenal ulcers [34], and antimicrobial treatments are an effective means to combat infection [35]. Since bacterial attachment is a prerequisite for infection [36], analogs of the Leb oligosaccharide may serve as therapeutic alternatives to broad spectrum antibiotics. The first approach to Leb addressed the synthesis of the core tetrasaccharide 22, which was assembled on the polymer support as depicted in Scheme 5 [37]. Polymer-bound galactal 19 was epoxidized with dimethyldioxirane and then reacted with a solution of glucal derivative 20 to give disaccharide diol 21. This reaction proceeded in highly regioselective fashion wherein glycosylation occurred at the allylic position at C3 of 20. Bisfucosylation of 21 using donor 16 provided polymerbound tetrasaccharide glycal 22a. Treatment of 22a with TBAF gave 22b, which

Scheme 4

Solid phase synthesis of an H-type blood group determinant by the glycal assembly method.

Solid Phase Oligosaccharide Synthesis

Solid phase synthesis of the Lewisb blood group determinant pentasaccharide glycal.

9

Scheme 5

10

Seeberger

was obtained in a 40% overall yield from 19. Initially, solution chemistry was used to further convert glycal 34b into a hexasaccharide of the Leb system, whereupon 34b was conjugated with human serum albumin to provide the desired neoglycoprotein. These approaches toward the synthesis of blood group determinants on a solid support were hampered by a serious shortcoming in the methodological arsenal of the glycal assembly method. While this strategy permitted rapid and concise access to ␤-glycosidic linkages, solution phase methodology solution had to be used for construction of N-acetylamine glucosidic linkages prevalent in biologically important blood group determinants, gangliosides, and N-linked glycopeptides [37]. To install the appropriate functionality at C2, solution phase chemistry has relied on a trans-diaxial addition of an iodonium electrophile in the presence of an amine to form an iodosulfonamide. Displacement of iodine proceeds presumably through an aziridine intermediate and may be induced by a thiolate nucleophile to fashion thioethyl 2-amidoglycosyl donors [38]. Successful transfer of this method to the solid support allowed polymer-bound glycals to be converted into thioethyl glycosyl donors. These donors were in turn coupled with a variety of glycosyl acceptors, including glycals [39]. After an efficient coupling protocol for the synthesis of ␤-2-amidoglucosidic linkages had been established, the difficulties encountered during the earlier synthesis of the Lewisb pentasaccharide glycal could be overcome. Branched tetrasaccharide 22a was converted into the thioethyl donor 23. Coupling to galactal acceptor 24 yielded 71% of the desired pentasaccharide 25a (Scheme 5). TBAF was used to effect retrieval of the pentasaccharide, affording 25b in 20% overall yield from 19 [32]. C.

Generation and Use of Thioethyl Donors on the Solid Support

While the use of glycals on the solid support allowed for the construction of ␤galactosyl linkages with great efficiency even with hindered glycosyl acceptors, the analogous ␤-glucosidic linkages could not be prepared reliably. In the galactose series, the anhydride is relatively stable to very mild Lewis acids, particularly anhydrous zinc chloride, because a conformationally constraining cyclic carbonate protecting group is used. The stability allows for galactosylation of even hindered acceptors such as C4 hydroxyls flanked by protecting groups at C3 and C6. No analogous constrained glucosyl epoxy donor is available, and glucosyl systems in the presence of zinc chloride are highly reactive and thus are prone to donor deterioration. To overcome the problems associated with the formation of ␤-glucosidic linkages, an approach that allowed for the conversion of glycals into thioethyl glycosyl donors was developed in solution phase [40]. These thioethyl glycosyl donors constitute a class of extremely powerful glycosylating agents upon activation with thiophilic reagents [41]. The glycal-derived donors were equipped with a C2 pivaloyl neighboring group and coupled to glycal acceptors to fashion a variety of glycosidic linkages with high efficiency. Pivaloyl neighboring groups had been shown by Kunz and others to prevent the formation of orthoester products during glycosylations [42]. Conversion of 26 to the protected thioethyl glycosyl donors 27 was achieved through epoxidation with dimethyldioxirane to yield the 1,2-anhydro sugar, followed

Solid Phase Oligosaccharide Synthesis

11

by opening of this intermediate by ethanethiol in the presence of a trace of acid. Thioethyl glycoside 27 was obtained in 91% yield and transformed into fully protected thioethyl donor 28a by pivaloylation in near quantitative yield. Support-bound thioglycoside 28a was activated using methyl triflate as a thiophile, while the nonnucleophilic base di-tert-butylpyridine (DTBP) provided stability for the glycal linkage during coupling. In model studies the formation of ␤-glucosyl (1→4), ␤-glucosyl (1→3), and ␤-glucosyl (1→6) linked disaccharides was achieved in good yield and complete selectivity [43]. After an efficient coupling protocol involving support-bound thioethyl glucosyl donors had been established, this methodology was applied to the synthesis of tetrasaccharide 31 containing exclusively ␤-(1→4) glucosidic linkages. Transformation of disaccharide glycal 29a into the C2 pivaloyl thioethyl glycosyl donor was followed by coupling to provide trisaccharide 30a in 45% overall yield from 26. Conversion of 30a to the thioethyl glycosyl donor was followed by coupling to glycal acceptor 14 to yield the desired tetrasaccharide 31a. After cleavage from the support 31b was obtained from 26 in 20% yield over nine steps, corresponding to 84% average yield per step (Scheme 6). D.

Solid Phase Synthesis of N-Linked Glycopeptides

N-linked and O-linked glycoproteins are the two major subgroups of glycoproteins, with the former being the most abundant in nature [44]. The biosynthesis of these glycoproteins results from cotranslational glycosylation usually occurring in the endoplasmic reticulum. The sugars of N-linked glycoproteins are usually attached by an oligosaccharyltransferase to an asparagine having the glycosylation sequence AsnX-Ser/Thr. Advances in glycopeptide synthesis have been achieved by several groups [45–47]. The synthesis of N-linked glycopeptides [48] on the solid support using a terminal glycal of a synthetic oligosaccharide domain aimed at a highly convergent synthetic strategy [49]. Polymer-supported trisaccharide 32 was reacted with anthracenesulfonamide and I(sym-coll)2ClO4 to form intermediate 33 (Scheme 7). Reaction of the iodosulfonamide 33 with tetra-n-butylammonium azide, followed by acetylation, provided the anomeric azide 34. The anthracenesulfonamide linkage can be cleaved under mild conditions such as 1,3-propanedithiol and Hu¨nig’s base, which concomitantly effected the reduction of the azide. The resulting amine was coupled with pentapeptide 35 in the presence of IIDQ to afford the protected glycopeptide 36. Orthogonal protecting groups on the C- and N-termini of the peptide provided the opportunity to extend the peptide chain while the ensemble was bound to the solid support. Alternatively, after removal from the support, the liberated peptide terminus may provide a functionality for linking to a carrier molecule to generate other glycoconjugates. The C-terminus of 36 was deprotected to give the acid 37, which was coupled to tripeptide 38 with a free N-terminus to give glycopeptide 39. Retrieval from the solid support afforded trisaccharide-octapeptide 40 in 18% overall yield from 19 [50]. IV.

STRATEGIES USING SUPPORT-BOUND GLYCOSYL ACCEPTORS

The solid support synthesis of oligopeptides and oligonucleotides is routinely carried out by reacting a support-bound structure (acceptor) with an excess of a solution-

12 Seeberger

Using glycal-derived polymer-bound thioethyl glucosyl donors in the synthesis of a ␤-(1→4) linked tetrasaccharide.

Scheme 6

Solid Phase Oligosaccharide Synthesis

Synthesis of N-linked glycopeptides on a solid support.

13

Scheme 7

14

Seeberger

based reactive species (donor). While the glycal assembly method, which utilizes a support-bound reactive species, has been used very successfully for the synthesis of oligosaccharides and glycopeptides, all other synthetic approaches explored to date have followed an acceptor-bound strategy. The acceptor-bound paradigm allows for the glycosyl donor to be used in excess during elongation of the growing oligosaccharide from the reducing to the nonreducing end. Because the donor is used in excess, side reactions that usually affect the glycosyl donor, resulting in inactive degradation products, do not impact the overall yield. Each hydroxyl group to serve as a glycosyl acceptor needs to be distinguished by a unique protecting group, which may be removed before each coupling cycle. In this section synthetic strategies employing a variety of glycosylating agents are reviewed. A.

Synthesis of Oligosaccharides Using Glycosyl Sulfoxides

The use of anomeric sulfoxides as glycosylating reagents was reported by Kahne and coworkers in 1989 [51]. Upon activation with triflic anhydride at low temperatures, anomeric sulfoxides are transformed into extremely reactive glycosyl donors, which can glycosylate very hindered acceptors. The excellent reactivity of the sulfoxide donors coupled with the selectivity achieved by use of a participating group in the C2 position prompted their use in the synthesis of oligosaccharides on the solid support [52]. Merrifield’s resin (polystyrene cross-linked with 1% divinylbenzene) was chosen as a polymer matrix, and an anomeric hydroxy thiophenyl ether moiety was selected as linker to the solid support (Scheme 8). The thiophenyl linkage is stable to all coupling and deprotection conditions but may be cleaved readily with mercuric trifluoroacetate. The first monosaccharide 42 was selectively deprotected and coupled by addition of 4 equiv of galactosyl sulfoxide 44, in the presence of DTBP as a base, and triflic anhydride as activator. Coupling at ⫺60⬚C for 30 min, filtration, and washing were followed by deprotection of the acetyl group. One further coupling and cleavage from the polymeric support by reaction with mercuric trifluoride furnished trisaccharide 50 in 52% overall yield. It was estimated that each coupling proceeded in 94–95% yield based on a detachment efficiency of 70–75% (as determined by solution phase studies). The glycosyl sulfoxide strategy successfully accommodated secondary hydroxyl acceptor groups, and ␣-fucosidic linkages were generated by the use of fucosyl sulfoxides [52]. These advances in the solid support synthesis of oligosaccharides using glycosyl sulfoxides were applied to the preparation of a combinatorial library of approximately 1300 disaccharides and trisaccharides [53]. A polystyrene–polyethylene glycol (PS-PEG) copolymer that swells in a wide range of solvents, including water, was employed in the syntheis of these libraries. Screening of the carbohydrate structures against a bacterial lectin from Bauthinia purpurea was performed with the carbohydrate structures still attached to the resin beads. Use of a chemical tagging system developed by Still and coworkers [54] allowed for rapid structure determination of the hits in the screening assay. Alternate binding structures with higher affinity than the natural ligand for the lectin of interest were identified in this manner. The glycosyl sulfoxide method has proven very efficient and flexible in constructing a variety of glycosidic linkages on the solid support. Combinatorial carbohydrate libraries prepared by this strategy hold great potential for the identification

Solid Phase Oligosaccharide Synthesis

Scheme 8

15

Solid phase synthesis of a trisaccharide by the glycosyl sulfoxide method.

of natural and nonnatural lectin ligands. While the synthesis of larger oligosaccharide structures using the glycosyl sulfoxide method still needs to be demonstrated, this strategy holds great potential for many applications. B.

Synthesis of Oligosaccharides Using Glycosyl Trichloroacetimidates

Among the multitude of glycosylating agents now at the disposal of the synthetic chemist, glycosyl trichloroacetimidates have become the most widely used building blocks. These donors have been used to prepare a large number of very complex oligosaccharides and glycoconjugates in solution phase [55]. High coupling yield, versatility, and excellent selectivity are hallmarks of this approach to oligosaccharide synthesis. Recently the first syntheses of oligosaccharides using trichloroacetimidate donors on the solid support have been reported. Schmidt and coworkers initially used Merrifield’s resin and explored a variety of ether and thioether linkers [56]. Attach-

16

Seeberger

ment of the first residue via a thioether linkage proved most reliable, and cleavage was successfully effected by reaction with a nucleophile. Reaction with water or methanol in the presene of N-bromosuccinimide yielded the lactol or the methyl glycoside, respectively. A two-step coupling cycle (Scheme 9) was used to assemble a series of (1→2) linked mannosides. Removal of the temporary acetyl-protecting group by sodium methoxide in methanol was followed by coupling of the trichloroacetimidate mannosyl donor 53 to the exposed axial C2 hydroxyl group. Repetition of this cycle and cleavage of the thioether linker resulted in assembly of the desired tetrasaccharide 56 in 34% overall yield. Penta- and hexamannosides obtained by using this method were characterized by mass spectrometry, but no yields were reported [57]. In addition to 1,2-trans-glycosides, disaccharides containing the synthetically more challenging 1,2-cis-glycosidic linkage were prepared. Fucosyl trichloroacetimidate was coupled to the C2 position of a support-bound mannosyl acceptor under the agency of trimethylsilyl triflate (TMSOTf) to yield 54% of the desired disaccharide [57]. Promising results were also obtained for the synthesis of a (1→2) trimannoside, which was prepared on a mercaptoethyl-functionalized controlled pore glass (CPG) solid support. This nonswelling support performed well under the described coupling and deprotection conditions [58]. More recently, the trichloroacetimidate strategy has been applied to the synthesis of a branched pentasaccharide unit common to most complex N-glycan structures (Scheme 10). Reaction of mannosyl donor 57 with thiol-functionalized resin resulted in attachment to the polymer matrix through the reducing end of the sugar 58. Removal of the benzoyl protecting groups on the C3 and C6 positions was effected by treatment with sodium methoxide to furnish diol 59. Dimannosylation was accomplished by reaction with trichloroacetimidate mannosyl donor 53 to afford solid support bound trisaccharide 60 in 38% overall yield. Cleavage of the acetyl protecting groups of the axial C2 hydroxyl functionalities revealed support-bound trisaccharide diol 61. Conversion of 61 to pentasaccharide 63 was achieved by reaction with glucosamine donor 62. Cleavage of the thioether linker with N-bromosuccinimide in the presence of methanol furnished the desired pentasaccharide methyl glycoside in 20% overall yield. Adinolfi et al. [59] explored the use of different solid supports for the synthesis of disaccharides employing trichloroacetimidate donors. The glycosyl acceptor was connected through the C2, C3, or C6 hydroxyl group to amino-functionalized solid support via a succinimide linker. Polystyrene, PS-PEG copolymer (TentaGel), and CPG supports were tested for their performance with trichloroacetimidate donors. Best coupling yields (95%) were obtained in these studies when the donor was reacted with acceptor bound to CPG or polystyrene support. PEG-containing copolymers performed significantly poorer in these glycosylations. Hunt and Roush [60] used a solid phase method to prepare 6-deoxy di- and trisaccharides. A sulfonate linker was employed to connect the first monosaccharide, a glycal, via its C6 position to the polymer resin. A galactosyl trichloroacetimidate donor was used in the first coupling reaction, followed by cleavage from the solid support by treatment with NaI to furnish pure disaccharides in 85–91% yield. Reduction with Bu3SnH (AIBN) provided the desired 6-deoxydisaccharide. These impressive initial accomplishments using trichloroacetimidate glycosyl donors on the solid support underscore the potential this strategy holds for the

Solid Phase Oligosaccharide Synthesis

Synthesis of ␣ (1→2) mannosides by the trichloroacetimidate strategy.

17

Scheme 9

18

Seeberger

Scheme 10

Use of trichloroacetimidate donors in the preparation of a branched pentasaccharide.

synthesis of complex oligosaccharide structures containing a variety of glycosidic linkages. Before the assembly of larger constructs may be contemplated, the overall yield and recovery will have to be improved. Novel linkers that are completely stable throughout the synthesis as well as improved coupling protocols should remedy these shortcomings.

Solid Phase Oligosaccharide Synthesis

C.

19

Solid Support Synthesis of Oligosaccharides Using Thioglycosides

Next to anomeric trichloroacetimidates, thioglycosides are the glycosyl donors most widely used in the synthesis of complex oligosaccharides and glycoconjugats. Thioglycosides may be prepared in high yield and are exceptionally stable, allowing for prolonged storage even at room temperature. Reaction with thiophiles such as methyl triflate or dimethylthiomethylsulfonium triflate (DMTST) ensures efficient activation of thioglycosides for the formation of glycosidic linkages [41]. In solution phase, thioethyl glycosides have found application in the synthesis of very complex oligosaccharides and in the preparation of a combinatorial carbohydrate library [61]. Straightforward synthetic access, stability, and high-yielding, selective glycosylation reactions made thioglycoside donors ideal for application to the solid support synthesis of oligosaccharides. Nicolaou and coworkers utilized phenolic polystyrene equipped with a photolabile, readily available o-nitrobenzyl linker in the synthesis of heptasaccharide 72 from mycelial walls of Phytophthora megasperma (Scheme 11) [62]. This heptasaccharide consists entirely of ␤ (1→6) or (1→3) glucosidic linkages and had previously been prepared by using solution phase methods [63] and block synthesis on a soluble PEG polymer [13]. Silyl ethers and fluorenylmethyloxy carbonyl (Fmoc) protective groups were employed to provide temporary protection of the hydroxyl functionalities to be glycosylated in the coupling steps that followed. Removal of the silyl ether protecting group by fluoridolysis to yield 64 was followed by coupling of 4 equiv of the solution-bound thiodonor 65 in the presence of the activator DMTST. The coupling to form 66 proceeded in greater than 95% yield according to material recovered after cleavage from the resin. Consecutive deprotection and coupling steps resulted in assembly of the desired heptasaccharide 71 still attached to the polymeric support. Photolytic cleavage of the linker, treatment with sodium methoxide to remove the ester protecting groups, and hydrogenolysis to cleave the benzyl ethers yielded unprotected heptasaccharide 72. Alternatively, fully protected heptasaccharide was obtained by acetylation of the anomeric hydroxyl group support in 20% overall yield from 64 after photolytic cleavage from the solid support [62]. After having established a reliable and efficient reaction sequence for the synthesis of oligosaccharides Nicolaou et al. further improved upon their initial strategy [64]. A 4-benzyloxybenzoic acid spacer was incorporated between the photolabile linker and the anomeric position of the first glycoside to provide access to either unprotected oligosaccharides or oligosaccharide glycosyl donors (Scheme 12). Support-bound acceptor 73 was converted into trisaccharide 77 following the synthetic strategy established earlier (see above). Cleavage from the resin was effected photolytically to furnish the fully protected trisaccharide 80 in 63% yield from 73. Alternatively, the protected trisaccharide was cleaved from the polymer with concomitant activation by exposure to PhSSiMe3/ZnI2/nBu4NI, furnishing phenylthiotrisaccharide donor 79 (Scheme 12; pp. 21 and 22). Support-derived trisaccharide building block 79 was used in a block synthesis of dodecasaccharide 82 (Scheme 13; p. 23). Reaction of support-bound glycosyl acceptor 78 with 79 furnished hexasaccharide 81. Removal of a silyl protecting group and coupling with trisaccharide donor 79 was repeated twice, followed by photolytic cleavage of the linker to furnish dodecasaccharide 82 in 10% yield from 73 [64].

20

Synthesis of a heptasaccharide employing thioglycoside donors.

Seeberger

Scheme 11

Solid Phase Oligosaccharide Synthesis

Preparation of trisaccharide thioglycoside building blocks on a solid support.

21

Scheme 12

22

Continued.

Seeberger

Scheme 12

Solid Phase Oligosaccharide Synthesis

Scheme 13

Assembly of a dodecasaccharide by the building block strategy on the solid phase.

23

24

Seeberger

The two examples of solid support oligosaccharide synthesis employing thioglycosides demonstrate the power of this strategy in the preparation of large carbohydrate structures. For the first time, a very large oligosaccharide was constructed via a block synthesis strategy in which every block itself was prepared on a solid support. The synthesis of combinatorial carbohydrate libraries using this strategy should certainly be possible. D.

Synthesis of Oligosaccharides Using Pentenyl Glycosides

Fraser-Reid and coworkers have established n-pentenyl glycosides as a class of powerful and versatile glycosyl donors [65]. Activation by a variety of reagents [e.g., Niodosuccinimide (NIS), TESOTf] furnished various glycosidic linkages in high yield and excellent selectivity in solution phase synthesis. More recently, the application of n-pentenyl glycosides to the synthesis of oligosaccharides on a solid support has been explored [66]. Both donor- and acceptor-bound synthetic strategies were investigated initially. Attachment of a pentenyl glycoside through the C6 hydroxyl group to the solid support was achieved via a photolabile 3-amino-3-(2-nitrophenyl)propionyl linker. Glycosylation using NIS and TESOTf as activating agents did not result in clean formation of the desired disaccharide product but also yielded lactol by donor degradation. An acceptor-bound synthetic strategy was more successful. Model studies with a glucose acceptor that was anchored to the solid support via its anomeric position showed that ␣- and ␤-glucosidic and C2 amino glucosidic linkages could be furnished in average coupling yields of greater than 90%. Reaction development was aided by 13C gel phase NMR spectroscopy to analyze the formed products [69]. After these initial studies, the synthesis of a support-bound, fully deprotected trisaccharide was achieved (Scheme 14). Chiron’s polystyrene-grafted ‘‘crowns’’ were chosen as solid support material because they are more amendable to parallel synthesis than traditional polymer supports. A photolabile o-nitrobenzyl linker was selected to attach the first amino glucoside 83 through its anomeric position to the solid support. Removal of the C6 dinitrobenzyl protecting group was followed by coupling with pentenyl mannoside 84 and resulted in formation of the desired disaccharide 85. Removal of the C2 chloroacetyl protecting group and coupling of pentenyl galactose donor 86 furnished trisaccharide 87. Global deprotection followed by peracetylation and photolytic cleavage from the solid support provided trisaccharide 88, but no yield was reported [66]. E.

Bidirectional Glycosylation Strategy for the Solid Phase Synthesis of Oligosaccharides

The preceding sections discussed the strengths and weaknesses of the donor- and acceptor-bound synthetic strategies developed to date for the synthesis of complex oligosaccharides. Boons et al. recently described an approach that combines both strategies in a bidirectional solid phase oligosaccharide synthesis [67]. The use of orthogonal glycosylating agents during the synthesis is an absolute prerequisite for the success of this bidirectional strategy. Trichloroacetimidate glycosides may be activated and function as glycosyl donors in coupling reactions with glycosyl acceptors containing an anomeric thioglycoside. The C6 hydroxyl group of thioethyl glycoside 89 (Scheme 15) was connected to TentaGel resin via a succinate

Solid Phase Oligosaccharide Synthesis

Scheme 14

Application of n-pentenyl glycosides to solid phase synthesis: preparation of a trisaccharide.

25

26

Bidirectional strategy for the synthesis of oligosaccharide libraries.

Seeberger

Scheme 15

Solid Phase Oligosaccharide Synthesis

27

linker. First, the exposed C4 hydroxyl functionality of 89 served as glycosyl acceptor in the reaction with solution-based trichloroacetimidate donor 90 in the presence of TMSOTf as promoter. Without the need for further protecting group manipulations, support-bound disaccharide thioglycoside 91 was activated with NIS/TMSOTf [68] to serve as a glycosyl donor in the reaction with solution-based glycosyl acceptor 92. This strategy served to generate a combinatorial library of trisaccharides while minimizing the necessary protecting group manipulations. This bidirectional approach stands to combine the advantages of the acceptorbound approach, such as use of excess donor during the synthesis, with the opportunity to fashion glycoconjugates through the anomeric position formerly possible only in a donor-bound paradigm. Other combinations of orthogonal glycosylating agents and the applicability of this strategy to the synthesis of larger oligosaccharide structures will have to be addressed in future studies. V.

‘‘ON-RESIN’’ ANALYTICAL METHODS

Development of a reliable solid phase methodology for the synthesis of oligosaccharides and glycopeptides has been greatly hampered by the lack of convincing ‘‘on-resin’’ analytical techniques. In most of the syntheses outlined above, it was necessary to cleave the products or intermediates of multistep syntheses from the resin to allow for the use of classical spectroscopic means (e.g., solution state NMR and mass spectrometry). The cleavage method analysis is time-consuming and wasteful in the context of multistep syntheses. Polystyrene–PEG copolymers may be analyzed by gel phase NMR spectroscopy but are quite expensive and allow only for relatively low loading [69]. Two NMR-based approaches for the on-resin analysis of polystyrene resins have been developed to aid reaction development by monitoring of the solid support matrix. A.

High-Resolution Magic Angle Spinning NMR Spectroscopy

The effectiveness of the glycal assembly approach to solid support oligosaccharide synthesis was impressively documented by on-resin analysis. The ‘‘crude product’’ of the multistep synthesis of trisaccharide 32 (Scheme 16) was monitored by highresolution magic angle spinning NMR (HR-MAS) [70].

Scheme 16

Solid phase synthesis of a trisaccharide by the glycal assembly method.

28

Seeberger

HR-MAS experiments proved to be an ideal way of monitoring the solid support synthesis by obtaining 1H NMR, 13C NMR, and 1H-13C NMR spectra of high quality. The 1H NMR (Scheme 17) of the crude product of this synthesis showed that only one product was obtained. Spin-echo techniques were used to suppress NMR peaks derived from the polymeric matrix. Since its introduction, this technique has greatly facilitated the development of novel synthetic schemes of oligosaccharides and glycopeptides on a solid support. B.

Gated-Decoupling–

13

C NMR Spectroscopy After 13C Enrichment

Wong et al. monitored the solid support synthesis of the sialyl Lewisx tetrasaccharide using 13C-enriched linker and protecting groups [71]. The 13C-labeled linker was attached to the polymer support and functioned as an internal integration standard for the 13C-enriched protecting groups of the growing oligosaccharide. After each coupling reaction, the NMR signal of the protecting group at the growing end of the molecule was compared with the signal of the internal standard to quantify the coupling. This method does not require special NMR hardware such as an HR-MAS probe but can be carried out in standard 5 mm NMR tubes. On the other hand, this method yields no detailed structural information regarding the anomeric purity of the compound prepared. The chemical integrity of the glycosidic linkages formed must be determined after cleavage of the target molecule from the solid support.

VI.

CONCLUSIONS AND OUTLOOK

Over the last 5 years a number of very promising strategies for the solid support synthesis of oligosaccharides and glycoconjugates have been developed. Several of the glycosylating reagents successfully used in solution have yielded promising results under the soild phase paradigm. While a number of different solid support matrices and linkers have been explored, much work in this area remains to be done. The rapid developments in combinatorial chemistry on the solid support have created novel support matrices, which will find application to solid phase oligosaccharide synthesis. Innovative linkers will be required to allow for the use of highly flexible protecting group strategies in the synthesis of branched oligosaccharides. Given the progress in the development of methods for the on-resin analysis of oligosaccharides, reaction development and optimization will be drastically accelerated. Novel glycosylating agents and orthogonal glycosylating schemes will further impact solid support synthesis. Alternative glycosylation methods such as the use of glycosyltransferase-catalyzed reactions on the solid support have also made great strides. For the synthesis of diverse structures, including natural and nonnatural glycoconjugates, chemical synthesis will be of paramount importance. While much progress has been made, many challenges remain before a flexible, high-yielding and absolutely selective strategy for the synthesis of oligosaccharides on the solid support becomes available. Once these problems have been solved, the construction of an automated oligosaccharide synthesizer may become feasible. Detailed studies concerning the structure and function of oligosaccharides and glycoconjugates will be possible once rapid access to complex glycoconjugates is a reality.

Solid Phase Oligosaccharide Synthesis

Spectrum of 32 obtained by high-resolution magic angle spinning 1H NMR spectroscopy.

29

Scheme 17

30

Seeberger

REFERENCES 1.

2. 3.

4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29.

30. 31. 32. 33.

(a) HJ Allen, EC Kisalius, eds. Glycoconjugates: Composition, Structure, and Function. New York: Marcel Dekker, 1992. (b) YC Lee, RT Lee, eds. Neoglycoconjugates: Preparation and Applications. London: Academic Press, 1994. (c) A Kobata. Acc Chem Res 26:319–324, 1993. A Varki. Glycobiology 3:97–130, 1993. (a) ML Phillips, E Nudelman, FCA Gaeta, M Perez, AK Singhal, S Hakomori, JC Paulson. Science 250:1130–1132, 1990. (b) LA Lasky. Science 258:964–969, 1992. (c) DJ Miller, MB Macek, BD Schur. Nature 357:589–593, 1992. (d) IT Schulze, ID Manger. Glycoconjugate J 9:63–66, 1992. (e) T Yaki, Y Hirabayashi, H Ishikawa, S Kon, Y Tanaka, M Matsumoto. J Biol Chem 261:3075–3078, 1986. (f) U Spohr, RU Lemieux. Carbohydr Res 174:211–237, 1988. DE Levy, PC Tang, JH Musser. In: WK Hagmann, ed. Annual Reports on Medical Chemistry, Vol. 29. San Diego, CA: Academic Press, 1994, pp 215–246. G Opdenakker, PM Rudd, CP Ponting, RA Dwek. FASEB J 7:1330–1337, 1993. SH Khan, RA O’Neill, eds. Modern Methods in Carbohydrate Synthesis. Amsterdam: Harwood Academic Publishers, 1996. K Toshima, K Tatsuta. Chem Rev 93:1503–1531, 1993. E Atherton, RC Sheppard. Solid phase peptide synthesis: A practical approach. Oxford: Oxford University Press, 1989. MH Caruthers. Scienec 230:281–285, 1985. JMJ Frechet. In: P Hodge, DC Sherrington, eds. Polymer-Supported Reactions in Organic Synthesis. Chichester: John Wiley & Sons, 1980, pp 407–434. DJ Gravert, KD Janda. Chem Rev 97:489–509, 1997. SP Douglas, DM Whitfield, JJ Krepinsky. J Am Chem Soc 117:2116–2117, 1995. R Verduyn, PAM van der Klein, M Douwes, GA van der Marel, JH van Boom. Recl Trav Chim Pays-Bas 112:464–466, 1993. O Kanie, Y Ito, T Ogawa. J Am Chem Soc 116:12073–12074, 1994. Y Ito, O Kanie, T Ogawa. Angew Chem Int Ed Engl 35:2510–2512, 1996. Y Ito, T Ogawa. J Am Chem Soc 119:5562–5566, 1997. Y Ito, S Manabe. Curr Opin Chem Biol 2:701–708, 1998. PH Seeberger, SJ Danishefsky. Acc Chem Res 31:685–695, 1998. RL Halcomb, SJ Danishefsky. J Am Chem Soc 111:6661–6666, 1989. RW Friesen, SJ Danishefsky. J Am Chem Soc 111:6656–6660, 1989. T-H Chan, W-Q Huang. J Chem Soc Chem Commun 909–911, 1985. SJ Danishefsky, KF McClure, JT Randolph, RB Ruggeri. Science 260:1307–1309, 1993. JT Randolph, KF McClure, SJ Danishefsky. J Am Chem Soc 117:5712–5719, 1995. RW Murray, R Jeyaraman. J Org Chem 50:2847–2853, 1985. (a) Reference 3a. (b) Y Hirabayashi, A Hyogo, T Nakao, K Tsuchiya, Y Suzuki, M Matsumoto, K Kon, S Ando. J Biol Chem 265:8144–8151, 1990. (c) Reference 3f. MJ Polley, ML Phillips, E Wagner, E Nudelman, AK Singhal, S Hakomori, JC Paulson. Proc Natl Acad Sci USA 88:6224–6228, 1991. For a review, see: KO Lloyd. Am J Clin Pathol 87:129–139, 1987. KO Lloyd. Semin Cancer Biol 2:421–431, 1991. For some recent syntheses of H-type determinants, see: (a) E Petrakova, U Spohr, RU Lemieux. Can J Chem 70:233–240, 1992. (b) TV Zemlyanukhina, NV Bovin. Bioorg Khim 16:1096–1104, 1990. RR Race, R Sanger. Blood Groups in Man. Oxford: Blackwell Science, 1975. JT Randolph, SJ Danishefsky. J Am Chem Soc 115:8473–8474, 1993. KC Nicoloau, CW Hummel, Y Iwabuchi. J Am Chem Soc 114:3126–3128, 1992. T Boren, P Falk, KA Roth, G Larson, S Normark. Science 262:1892–1895, 1993.

Solid Phase Oligosaccharide Synthesis 34. 35.

36.

37. 38. 39. 40. 41. 42. 43. 44. 45.

46. 47.

48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66.

31

J Alper. Science 260:159–160, 1993. (a) DY Graham, GM Lew, PD Klein, DG Evans, DJ Evans, ZA Saeed, HM Malaty. Ann Intern Med 116:705–732, 1992. (b) E Hentschel, G Brandstatter, B Dragosics, AM Hirschl, H Nemeg, K Schutze, M Taufer, H Wruzer. N Engl J Med 328:308–312, 1993. N Sharon. In: IE Liener, N Sharon, IJ Goldstein, eds. The Lectins: Properties, Functions and Applications in Biology and Medicine. New York: Academic Press, 1986, pp 494– 525. JT Randolph, SJ Danishefsky. Angew Chem Int Ed Engl 33:1470–1473, 1994. (a) DA Griffith, SJ Danishefsky. J Am Chem Soc 112:5811–5819, 1990. (b) T Hamada, A Nishida, O Yonemitsu. J Am Chem Soc 108:140–145, 1986. C Zheng, PH Seeberger, SJ Danishefsky. Angew Chem Int Ed Engl 37:789–792, 1998. PH Seeberger, M Eckhardt, CE Gutteridge, SJ Danishefsky. J Am Chem Soc 119:10064– 10072, 1997. For a review see: PJ Garegg. Adv Carbohydr Chem Biochem 52:179–205, 1997. (a) H Kunz, A Harreus. Liebigs Ann Chem 41–48, 1982. (b) H Kunz, A Harreus. Liebigs Ann Chem 717–732, 1986. C Zheng, PH Seeberger, SJ Danishefsky. J Org Chem 63:1126–1130, 1998. H Kunz. Angew Chem Int Ed Engl 26:294–308, 1987. SJ Danishefsky, JY Roberge. In: DG Large, CD Warren, eds. Glycopeptides and Related Compounds: Synthesis, Analysis and Applications. Marcel Dekker: New York, 1997, pp 245–293. M Meldal. Curr Opin Struct Biol 4:710–718, 1994. For recent selected examples, see: (a) ST Cohen-Anisfeld, PT Landsbury. J Am Chem Soc 115:10531–10537, 1993. (b) D Vetter, D Tumelty, SK Singh, MA Gallop. Angew Chem Int Ed Engl 34:60–63, 1995. (c) U Sprenghard, G Kretzschmar, E Bartnik, C Hu¨ls, H Kunz. Angew Chem Int Ed Engl 34:990–993, 1995. JY Roberge, X Beebe, SJ Danishefsky. Science 269:202–204, 1995. P Lloyd-Williams, F Albericio, E Giralt. Tetrahedron 49:11065–11133, 1993. JY Roberge, X Beebe, SJ Danishefsky. J Am Chem Soc 120:3915–3929, 1998. D Kahne, S Walker, Y Cheng, D Van Engen. J Am Chem Soc 111:6881–6882, 1989. L Yan, CM Taylor, R Goodnow, D Kahne. J Am Chem Soc 116:6953–6954, 1994. R Liang, L Yan, J Loebach, M Ge, Y Uozumi, K Sekanina, N Horan, J Gildersleeve, A Smith, K Biswas, WC Still, DE Kahne. Science 274:1520–1522, 1996. MHJ Ohlmeyer, RN Swanson, LW Dillard, JC Reader, G Asoline, R Kobayashi, M Wigler, WC Still. Proc Natl Acad Sci USA 90:10922–10926, 1993. RR Schmidt, W Kinzy. Adv Carbohydr Chem Biochem 50:21–123, 1994. J Rademann, RR Schmidt. Tetrahedron Lett 37:3989–3990, 1996. J Rademann, RR Schmidt. J Org Chem 62:3989–3990, 1997. A Heckel, E Mross, K-H Jung, J Rademann, RR Schmidt. Synlett 171–173, 1998. M Adinolfi, G Barone, L De Napoli, A Iadonisi, G Piccialli. Tetrahedron Lett 37:5007– 5010, 1996; 39:1953–1956, 1998. JA Hunt, WR Roush. J Am Chem Soc 118:9998–9999, 1996. C-H Wong, X-S Ye, Z Zhang. J Am Chem Soc 120:7137–7138, 1998. KC Nicolaou, N Winssinger, J Pastor, F DeRoose. J Am Chem Soc 119:449–450, 1997. BP Ossowski, A Pilotti, PJ Garegg, B Lindberg. Angew Chem Int Ed Engl 22:793–795, 1983. KC Nicolaou, N Watanabe, J Li, J Pastor, N Winssinger. Angew Chem Int Ed Engl 37: 1559–1561, 1998. B Fraser-Reid, UE Udodong, Z Wu, H Ottosson, JR Merritt, CS Rao, C Roberts, R Madsen. Synlett 927–942, 1992. R Rodebaugh, S Joshi, B Fraser-Reid, HM Geysen. J Org Chem 62:5660–5661, 1997.

32

Seeberger

67. 68. 69. 70.

T Zhu, G-J Boons. Angew Chem Int Ed Engl 37:1898–1900, 1998. A Demchenko, T Stauch, GJ Boons. Synlett 7:818–820, 1997. E Giralt, J Rizo, E Pedroso. Tetrahedron 40:4141–4154, 1984. PH Seeberger, X Beebe, GD Sukenick, S Pochapsky, SJ Danishefsky. Angew Chem Int Ed Engl 36:491–493, 1997. T Kanemitsu, O Kanie, C-H Wong. Angew Chem Int Ed Engl 37:3418–3420, 1998.

71.

2 Dehydrative Glycosylation with 1-Hydroxy Donors David Gin University of Illinois at Urbana-Champaign, Urbana, Illinois

I.

INTRODUCTION

The preparation of complex carbohydrates has emerged as a major focus in synthetic organic chemistry, and this is no doubt a result of the growing awareness of the many important roles of this class of molecules in biology [1]. Perhaps the most important reaction in the chemical synthesis of carbohydrates is the formation of the glycosidic bond, for this is the primary means for the controlled assembly of complex oligosaccharides and glycoconjugates from monosaccharide precursors. Thus a variety of methods have been developed to effect the glycosylation process. Much of the effort has focused on the general coupling strategy outlined in Scheme 1 [2]. In this strategy, one begins with a carbohydrate coupling partner (1), which is subjected to initial derivatization whereby the anomeric substituent is transformed into a latent leaving group (LG). The resulting intermediate 2, or glycosyl donor, is typically isolated, and in a second step the anomeric leaving group is activated with an appropriate glycosylation promoter or catalyst. This process usually takes place in the presence of a nucleophilic glycosyl acceptor (Nu-H), which undergoes an effective displacement of the leaving group to form the anomeric bond in the product glycoside (3). Within the last century, a variety of leaving groups have been developed for the generation of various glycosyl donors that can undergo efficient coupling in the second step. Some of these are listed in Scheme 1, and all have proven to be useful to some extent in complex carbohydrate synthesis [2]. A far less developed strategy for glycosidic bond formation is a direct dehydrative coupling procedure in which one begins with a 1-hydroxy carbohydrate (1) 33

34

Scheme 1

Gin

Traditional glycosylation strategies [2].

as the glycosyl donor (Scheme 2). This approach offers a complementary if not more efficient strategy for glycosylation in that a distinct anomeric derivatization step to generate an isolable glycosyl donor 2 is obviated. As such, all the operations of anomeric derivatization, activation, and bond formation are combined into a one-pot procedure. Despite its potential advantages, however, this strategy has not been extensively employed in complex oligosaccharide synthesis. The establishment of a viable synthetic method calls for the overcoming of such inherent difficulties associated with this approach as the reversibility of the process and the propensity for hemiacetal self-coupling, in addition to the common glycosylation obstacles such as coupling efficiency and high anomeric stereoselectivity. This chapter summarizes recent advances in the development of nonenzymatic direct dehydrative glycosylations with 1-hydroxy glycosyl donors. The concept of direct dehydrative glycosylation is not a new one. One of the earliest glycosylation methods is the Fischer procedure [3], currently adopted for preparation of simple glycosides (Scheme 3). In this process an unprotected monosaccharide (4) is treated with an excess of an alkyl alcohol in the presence of an acid catalyst, resulting in the net loss of water and substitution at the anomeric position by the alcohol acceptor. Usually a desiccant is not present in this hemiacetalto-acetal exchange process; as a result, the equilibrium can favor the formation of the glycoside product 5 only through the use of a large excess of the alcohol acceptor (typically employed as the reaction solvent or cosolvent). In the original glycosylation procedure, HCl was used as the acid catalyst, with the coupling event usually proceeding at elevated temperatures. Over the years, a number of other Brønsted acid catalysts have been found to be effective in this process, including various inorganic and sulfonic acids [4,5] as well as acidic resins [6]. In addition, a host of Lewis acid catalysts have been employed [7], giving rise to substrate-specific variants of the Fischer protocol. In fact, with selected Lewis acid promoters such as FeCl3

Scheme 2

Direct dehydrative glycosylations.

Glycosylation with 1-Hydroxy Donors

Scheme 3

35

Fischer glycosylation [3]. R = methyl, ethyl, n-propyl, i-propyl, amyl, allyl,

benzyl, etc.

[8], it is possible to favor the formation of the kinetic furanoside product over the thermodynamically more stable pyranoside adducts [9]. While it is not the goal of this section of the chapter to present a comprehensive summary of Fischer glycosylation methods, it is worth emphasizing that this protocol and its variants have, over the last century, remained one of the most popular methods for the preparation of simple alkyl glycosides. Indeed, this venerable method, with its simplicity and versatility, has frequently been chosen as the starting point for the preparation of C1-protected monosaccharide building blocks for complex molecule syntheses. Despite the widespread use of the Fischer method, it has yet to be shown to be effective in the controlled assembly of complex oligosaccharides. Because of the acidic medium under which the acetal exchange process occurs, only simple alcohols devoid of acid-labile functionality are employed as glycosyl acceptors. Moreover, the necessity of a large excess of acceptor to favor equilibrium formation of glycoside 5 precludes the use of complex or valuable molecules as nucleophilic acceptors. Typically the preparation of oligosaccharides by modified Fischer protocols has been limited to the preparation of carbohydrate oligomers of varying size and complexity [10,11]. To establish a method for controlled glycosylation with 1-hydroxy carbohydrates, mild dehydrative coupling conditions are required that favor the cross-condensation of distinct hemiacetal donor and nucleophilic acceptor substrates. One approach toward this end is to employ a set of reagents that can rapidly and completely activate the C1-hydroxyl functionality in 1 to generate, in situ, a highly reactive intermediate that incorporates a transient leaving group at the anomeric position. The requirement for rapid activation of the anomeric hydroxyl is obviously necessary to minimize the extent of hemiacetal self-condensation, and the in situ formation of an extremely potent leaving group at C1 would allow for facile glycosidic bond construction with a nucleophilic acceptor without the need for isolation of an intermediate glycosyl donor. II.

COUPLING VIA GLYCOSYL HALIDES

Koto and coworkers have developed a dehydrative coupling procedure employing the in situ generation of a glycosyl bromide intermediate (6, Scheme 4) [12,13]. In this procedure, a stoichiometric quantity of TMS-Br is introduced into a reaction mixture containing equimolar quantities of the 1-hydroxy glycosyl donor 1, the alcohol acceptor (R⬘OH), CoBr2, and tetrabutylammonium bromide. The glycosyl bromide 6, generated in situ, then proceeds to glycosylate the alcohol acceptor. Glycosyl bromides are well-established donors in the Koenigs–Knorr procedure [14], wherein these intermediates are typically isolated and activated with a halophilic promoter to

36

Gin

Scheme 4 Dehydrative glycosylations via glycosyl bromides [12,13,15]. * Using alternate method: MeSO3H, CoBr2, Et4NClO4.

effect glycosylation. However, the method of Koto and coworkers is distinct in that the intermediate 6 is both generated and consumed in situ under the reaction conditions, resulting in a net dehydration via a one-pot coupling process. The authors tentatively attribute the feasibility of this one-pot procedure to (1) the favorable ˚ molecular sieves acting both as a HBr scavenger and as a desiccant, effects of the 4 A and (2) the functioning of CoBr2 both as an effective desiccant and as a glycosyl bromide activator in the coupling stage with R⬘OH. This procedure was used to glycosylate a number of alkyl alcohols with 2,3,4,6tetra-O-benzyl-D-glucose, resulting in very good to moderate yields, depending on the complexity of the alcohol acceptor (e.g., 8–12). In these couplings, only trace amounts of products arising from hemiacetal self-coupling were detected, and the presence of Bu4NBr favored the formation of the ␣ anomer of the product glycosides. It is worth noting that in an earlier related report, the reagent combination of [MeSO3H, CoBr2, Et4NClO4] was also employed by Koto to effect the desired dehydrative glycosylation [15]. Under these reaction conditions, it is presumed that HBr is generated, leading to the formation of 6; however, this protocol generally led to diminished yields (e.g., 8 and 10) compared with the above-mentioned TMS-Br ⭈CoBr2 procedure.

Glycosylation with 1-Hydroxy Donors

III.

37

COUPLING VIA GLYCOSYL SULFONATES

The in situ generation and coupling of glycosyl sulfonate esters of 1-hydroxy glycosyl donors has also been investigated in the context of direct dehydrative glycosylation. Early work by Leroux and Perlin highlighted the generation of glycosyl trifluoromethanesulfonates 13 as useful reactive intermediates for glycosylation in the presence of Bu4NBr and the acid scavenger s-collidine (Scheme 5) [16]. Their initial attempts at the glycosylation of the acceptor R⬘OH directly with triflate 13 in the absence of Bu4NBr were unsuccessful (i.e., 1 → 13 → 7), presumably as a consequence of the instability of the highly reactive triflate species. However, the introduction of Bu4NBr served to rapidly convert 13 into the more stable glycosyl bromide 6, which then proceeds to glycosylate the acceptor R⬘OH. With this protocol, isolation of the glycosyl bromide 6 is again unnecessary, allowing for an overall one-pot process. The authors also highlight this as a useful method for the generation of glycosyl bromides under nonacidic conditions.

Scheme 5 Dehydrative glycosylations via glycosyl triflates [16]. * Using alternate method: (Me2SO2)2O, s-collidine; R⬘OH.

38

Gin

With this procedure, a number of simple alcohols were glycosylated to generate carbohydrate products such as 14–17. With glycosyl donors incorporating a C2-acyl functionality such as an acetate ester, the corresponding orthoester adduct (e.g., 18) is generated, the obvious result of acceptor addition to the bicyclic oxygen-stabilized carbocation that arises through neighboring group participation. It should be noted that the use of methanesulfonic anhydride in place of triflic anhydride also led to similar glycoside adducts (e.g., 14 and 18). When (MeSO2)2O is used as the dehydrating reagent, however, the addition of Bu4NBr is not necessary because the intermediate glycosyl methanesulfonate, which can directly glycosylate the acceptor at ambient temperature, has increased stability. Although the use of triflic anhydride and s-collidine in the absence of bromide ion was ineffective for glycosylation, the use of triflic anhydride alone was found to be useful for the coupling of 1-hydroxy glycosyl donors. Pavia et al. [17] have shown that the treatment of 1-hydroxy donors with triflic anhydride in the absence of any acid scavenger led to the formation of (1,1⬘)-linked disaccharides 19, the products of self-coupling of the 1-hydroxy donor (Scheme 6). The presumed mechanism [18] of this transformation involves (1) activation of the hemiacetal with trace amounts of triflic acid for conversion to the glycosyl triflate 13 (or the corresponding oxocarbenium triflate), (2) the rapid coupling of the highly reactive intermediate 13 with unactivated 1, and (3) the action of triflic anhydride as a desiccant to favor formation of the (1,1⬘)-disaccharide 19. Although this is a relatively efficient method for the preparation of symmetrical trehalose analogs (e.g., 20 and 21), it is often the case that the self-condensation product of the hemiacetal donor is an unwanted side product of dehydrative glycosylation. However, if this coupling procedure is performed in the presence of an excess (2–4 equiv) of an alcohol glycosyl acceptor R⬘OH, heterocouplings do take place (e.g., 22 and 23). Pavia has used this protocol to prepare complex galactosyl-O-glycopeptides [19]. Koto has shown that glycosyl arylsulfonates can also be used as intermediates in couplings with 1-hydroxy glycosyl donors. The direct glucosylation of several simple alcohol acceptors with 2,3,4,6-tetra-O-benzyl-D-glucose can be accomplished with the ternary mixture of p-nitrobenzenesulfonyl chloride, silver triflate, and triethylamine as the acid scavenger (Scheme 7) [20]. Simple alcohols such as methyl alcohol, cyclohexylmethyl alcohol, and dihydrocholesterol are glycosylated efficiently via the glycosyl sulfonate 24, and the formation of trehalose by-products is minimized as a result of rapid anomeric sulfonylation. In addition, couplings performed in the presence of N,N-dimethylacetamide generally lead to higher proportions of the ␣ anomer (26) [21]. This procedure can be used to prepare glucuronides (27) [22], as well as other complex oligosaccharides such as the branched-chain oligosaccharides of the sarsasaponins (e.g., 28–30) [23]. Other dehydrative glycosylations employing glycosyl arylsulfonates include the formation of transient glycosyl tosylates from 1-hydroxy donors, reported by Szeja (Scheme 8) [24]. For obvious reasons, all the preceding dehydrative coupling methods call for care to exclude moisture from the reaction. However, the method of Szeja is distinct in that it effects a dehydrative coupling with 2,3,4,6-tetra-O-benzylD-glucose in an aqueous cosolvent under phase transfer conditions, presumably via in situ formation of the intermediate tosylate 31. Using a mixture of p-toluenesulfonyl chloride, an excess (4 equiv) of the alcohol acceptor, a solvent mixture of CH2Cl2 and 40% NaOH(aq), and BnEt3NCl as the phase transfer catalyst, simple alcohols

Glycosylation with 1-Hydroxy Donors

Scheme 6

39

Dehydrative glycosylations via glycosyl triflates [17–19].

such as benzyl alcohol, cholesterol, and even tertiary alcohols such at t-butyl alcohol can be glucosylated with good ␣-selectivity (e.g., 15, 32, 33). IV.

COUPLING VIA GLYCOSYL OXOPHOSPHONIUM INTERMEDIATES

The in situ generation of glycosyl oxophosphonium salts has been shown to be useful for one-pot couplings with 1-hydroxy glycosyl donors. The most common method employed for the generation of the carbohydrate oxophosphonium intermediates 35 occurs by way of the Mitsunobu protocol [25] or variants thereof (Scheme 9). The method typically employs the initial reaction of a dialkyl azodicarboxylate with a phosphine substrate in the presence of a weak acid to generate a quaternary phosphonium salt, 34. With this reagent, it is known that a variety of alcohols can be

40

Gin

Scheme 7 Dehydrative glycosylations via glycosyl arylsulfonates [20–23]. * Coupling performed in the presence of 2.5 equiv AcNMe2.

activated to form the corresponding oxophosphonium species, which can then be displaced (SN2) by a variety of weak-acid nucleophiles, including carboxylic acids, phenols, N-hydroxyimides, imides, and oximes. The process, being a one-pot dehydration, was therefore studied in the context of activation of the carbohydrate C1hemiacetal function. Early reports by Jurczak et al. [26] focused on the preparation of N-glycosylphthalimides from various furanoses and pyranoses employing diethyl azodicarboxylate, triphenylphosphine, and phthalimide as glycosyl acceptor; however, yields were typically low (5–43%). The use of N-hydroxyphthalimide as the glycosyl acceptor, on the other hand, led to a dramatic increase in coupling efficiency (e.g., 36) [27] with both pyranose and furanose donors. The utility of this particular glycosylation was demonstrated in the preparation of a key alkoxyamino glycosyl linker in Nicolaou’s synthesis of the oligosaccharide portion of calicheamicin ␥ [28].

Glycosylation with 1-Hydroxy Donors

Scheme 8

41

Dehydrative glycosylations via glycosyl arylsulfonates [24].

Simple alkyl alcohols are typically poor acceptor substrates for this type of coupling owing to the insufficient acidity of the acceptor component to aid in the formation of the phosphonium salt 34. To overcome this difficulty, Szarek and coworkers [29] introduced a mercuric halide salt to facilitate the formation of 34, thereby effecting dehydrative glycosylation of alcohol acceptors (e.g., 37) with hemiacetal donors. Carboxylic acids are generally good nucleophiles in the Mitsunobu reaction and have also been demonstrated to function as efficient glycosyl acceptors with this protocol. Smith has refined and used this method for glycosyl ester formation [30] as one of the key convergent steps (i.e., 38) in the synthesis of a (⫹)-phyllanthoside [31]. Likewise, phenolic nucleophiles have also been shown to be useful acceptors. Early reports involved efficient glycosylations to prepare simple phenyl glycosides [32– 34] followed by the preparation of more complex ones such as 39 [35], an intermediate in Ogawa’s total of synthesis of hygromycin A [36]. Roush’s preparation of the olivomycin fragment 40 [37], highlighting the utility of the method in complex molecule synthesis and its tolerance to a wide variety of sensitive functionalities, is of particular interest. Mukaiyama and coworkers have developed an alternate method for the generation of glycosyl oxophosphonium intermediates. This method is based on Hendrickson’s earlier work concerning the development of phosphonium anhydride dehydrating reagents, prepared by the reaction of triphenylphosphine oxide (2 equiv) and triflic anhydride (1 equiv) [38]. With the reagent combination of tributyl phosphine oxide (2 equiv) and triflic anhydride (1 equiv), Mukaiyama reported that the resulting diphosphonium salt 41 (Scheme 10) [39] efficiently converted the hemiacetal functionality of 1-hydroxy carbohydrates to the anomeric oxophosphonium species 42, an effective glycosylating agent. In the presence of i-Pr2NEt as an acid scavenger, glycosylation of a number of simple alcohols, O-TMS-protected alcohol acceptors, and even azide acceptors, can be effected with 2,3,5-tri-O-benzyl-D-furanose to afford the product glycosides (e.g., 43–45).

42

Scheme 9

Gin

Dehydrative glycosylations via glycosyl oxophosphonium intermediates [26– 37]. DEAD = diethyl azodicarboxylate; DIAD = diisopropyl azodicarboxylate.

Glycosylation with 1-Hydroxy Donors

Scheme 10

V.

43

Dehydrative glycosylations via glycosyl oxophosphonium intermediates [39].

COUPLING VIA GLYCOSYL OXOSULFONIUM INTERMEDIATES

We have developed a new method for dehydrative glycosylation involving the in situ generation of glycosyl oxosulfonium species [40]. The use of reactive dimethylsulfonium reagents for hydroxyl activation has been investigated in depth, primarily in the context of hydroxyl oxidation to form a carbonyl functionality [41]. Similar activation of a carbohydrate hemiacetal should therefore lead to the formation of an anomeric oxosulfonium species such as 47 (Scheme 11), which would likely function as an efficient glycosyl donor. The key challenge is the development of a reaction protocol in which sulfoxide displacement (or dissociation) at the anomeric center of 47 is favored over other possible reaction manifolds such as oxidation or elimination. In our procedure (Scheme 11), initial activation of diphenyl sulfoxide with triflic anhydride presumably generates diphenyl sulfide bis(triflate) 46. In situ activation of the hemiacetal hydroxyl function in 1 by 46 would afford the oxosulfonium triflate 47, which then undergoes coupling with the appropriate acceptor to afford the glycosylated product 3. By employing diphenyl sulfoxide, which does not incorporate protons adjacent to the sulfur center, ylide formation from 47 is precluded, and thus oxidation via intramolecular proton transfer (i.e., Moffatt–Swern oxidation) is avoided. Furthermore, triflic anhydride serves as an ideal reagent for sulfoxide activation in that (1) activation of dimethyl sulfoxide with triflic anhydride to induce nucleophilic attack at sulfur has been reported in a limited number of cases in the contexts of Swern-type oxidations [42], sulfilimine synthesis [43], alkene functionalization [44], and quinone methide generation [45]; and (2) the anionic by-product of activation, triflate, is a weak nucleophile and thus should not obstruct glycosidic bond formation with the desired acceptor.

44

Scheme 11

Gin

Dehydrative glycosylations via glycosyl oxosulfonium intermediates [40].

The glycosylation method just described was used to couple a variety of acceptors with 2,3,4,6-tetra-O-benzyl-D-glucopyranose. For example, phenol, ethanethiol, and 1,3,5-trimethoxybenzene underwent efficient glycosylation to yield the corresponding O-aryl, S-alkyl, and C-aryl glycosides in good yield (48–50). Moreover, the N-glycosylation of amide functionalities, which has been reported to occur with only the most reactive of nonenzymatic glycosylation procedures [46], was

Glycosylation with 1-Hydroxy Donors

45

found to proceed smoothly with N-(trimethylsilyl)trimethylacetamide to afford 51. In addition, the glycosylation of tertiary alcohols (e.g., 33) was shown to be equally efficient when this new method was used. To determine whether C2-neighboring group effects would influence the glycosylation stereochemistry in this procedure, 2,3,4,6-tetra-O-benzoyl-D-glucopyranose, a donor incorporating multiple electronwithdrawing protective groups, was also employed as a coupling partner. This donor was coupled with carbohydrate acceptors to form the corresponding (1→6)- and (1→4)-linked disaccharides 53 and 54 with complete ␤-selectivity. Furanose donors are also compatible with this method, as evidenced by the formation of 52, the product of dehydrative N-glycosylation of bis(O-TMS)-thymine. Finally, the free NH functionality of a carbamate is compatible under these reaction conditions, allowing access to 2-amino-2-deoxy-glycosides (e.g., 55) with this protocol. VI.

COUPLINGS VIA GLYCOSYL OXOTITANIUM AND THIO-TIN INTERMEDIATES

Glycosyl oxotitanium intermediates were found to be useful in the furanosylation of various glycosyl acceptors. As long ago as 1989, Mukaiyama and coworkers had shown that [1,2-benzenediolato(2-)-O,O⬘]-oxotitanium (56) is a useful Lewis acid catalyst for aldol [47] and Michael reactions of ketene silyl acetals [48]. Used in combination with triflic anhydride, this Lewis acid was also found to be an effective dehydrating agent (Scheme 12) [49]. Treatment of 56 with triflic anhydride presumably leads to the formation of the bis(titanium) salt 57, which activates a 1-hydroxy glycosyl donor to form the intermediate anomeric oxotitanium species 58. This intermediate was shown to be an effective glycosyl donor to various alcohol and trimethylsilyl ether nucleophiles (e.g., 59–61). Mukaiyama et al. also found the use of diphenyltin sulfide and triflic anhydride to be suitable for furanoside formation with C1-hydroxy furanose donors (Scheme 13) [50]. Following activation of the hemiacetal with this pair of reagents, a glycosyl acceptor is introduced, in the form of an alcohol or a TMS-ether, to yield the product glycoside. Excellent anomeric selectivity can be achieved with 2,3,5-tri-O-benzyl-Dribofuranose as the donor (e.g., 60, 63), with the selectivity depending on the presence or absence of LiClO4 as an additive. The authors presumed that in the absence of LiClO4, the reactive intermediate 64 (Scheme 14) is generated, whereby the ␣ face is sterically shielded by virtue of coordination of tin(IV) with the C2-oxygen of the donor, leading to near exclusive information of the ␤ adduct 65. Conversely, the presence of LiClO4 (15 equiv) may serve to convert 64 into the ion pair 66, in which the perchlorate anion is situated on the ␤ face, trans to the C2 substituent, favoring acceptor approach onto the ␣ face. It is worth noting that the intermediacy of the corresponding C1-O-TMS derivative of the 1-hydroxy donor has also been suggested in these coupling reactions when TMS–ether acceptors are employed [51]. VII.

COUPLINGS VIA GLYCOSYL ISOUREA INTERMEDIATES

Carbodiimide reagents have long been known to be effective dehydrating agents in coupling reactions [52]. Tsutsumi et al. have shown that simple carbodiimide reagents are also effective in the context of dehydrative couplings with 1-hydroxy glycosyl donors (Scheme 15) [53]. Reactions of alcohols with carbodiimides have been known

46

Gin

Scheme 12

Dehydrative glycosylations via glycosyl oxotitanium intermediates [49].

to occur in the presence of CuCl to form O-alkyl isoureas [54]. This transformation was applied to C1-glycosyl hemiacetals 1 with dicyclohexyl carbodiimide to generate the corresponding O-glycosyl isourea 68, which proceeds to glycosylate appropriate acceptors via expulsion of dicyclohexyl urea from the anomeric center. This one-pot procedure was found to be effective at elevated temperatures (80–85⬚C) with a variety of phenolic nucleophiles (e.g., 69–72) in addition to several thiophenolic nucleophiles (e.g., 73). The high ␤-selectivity of the coupling reactions presumably arises from SN2 displacement of the corresponding ␣-glycosyl isourea intermediate, which is generated at elevated temperatures in the activation stage. VIII.

COUPLING VIA LEWIS ACID CATALYSIS

Acid catalysis in glycosylations with 1-hydroxy glycosyl donors has traditionally employed the Fischer glycosylation procedure or variants thereof in which simple glycosyl acceptors are employed in vast excess. Within the last several years, a series

Glycosylation with 1-Hydroxy Donors

Scheme 13

47

Dehydrative glycosylations via glycosyl thio-tin intermediates [50,51].

of dehydrative glycosylations using Lewis acid catalysts in the presence of a desiccant has been reported, employing equimolar or only slight excesses of various nucleophilic acceptors (Scheme 16, Table 1). Inanaga and coworkers reported the use of the catalyst combination of Yb(OTf)3 and methoxyacetic acid (0.1 equiv of each) employing only 1.2 equiv of the glycosyl acceptor (Table 1, entries 1–3) [55]. Anomeric mixtures were obtained with pyranose donors, while the ribofuranose donor 77 afforded high ␤-selectivities.

Scheme 14

Perchlorate ion and anomeric selectivity [50].

48

Gin

Scheme 15

Dehydrative glycosylations via glycosyl isoureas [53].

Scheme 16

Dehydrative glycosylations via Lewis acid catalysis: Selected coupling partners

[55–60].

Glycosylation with 1-Hydroxy Donors

Table 1

Entry

49

Dehydrative Glycosylations via Lewis Acid Catalysis Coupling partners

Catalyst(s)

Reagents

1

77 ⴙ 78

2

77 ⴙ 79

3

74 ⴙ 81

4

74 ⴙ 78

Yb(OTf )3, MeOCH2CO2H Yb(OTf )3, MeOCH2CO2H Yb(OTf )3, MeOCH2CO2H Zn(OTf )2

5

75 ⴙ 79

Zn(OTf )2

6

76 ⴙ 79

Zn(OTf )2

7

77 ⴙ 78

Sn(OTf )2, (TMS)2O

TMSCl (1.5 equiv), CH3CN (23⬚C) TMSCl (1.5 equiv), CH2Cl2 (23⬚C) TMSCl (1.5 equiv), CH2Cl2 (23⬚C) Drierite, MeNO2 (23⬚C)

8

77 ⴙ 79

Sn(OTf )2, (TMS)2O

Drierite, MeNO2 (23⬚C)

9

77 ⴙ 78

Sn(OTf )2, (TMS)2O

10

75 ⴙ 80

Sn(OTf )2TMSCl

11

77 ⴙ 78

[Ph3C]⫹ [B(C6F5)4]⫺

Drierite, LiClO4 (1.5 equiv), MeNO2 (23⬚C) Drierite, LiClO4 (3.5 equiv), PhH (23⬚C) Drierite, EtNO2 (0⬚C)

12

77 ⴙ 80

[Ph3C]⫹ [B(C6F5)4]⫺

Drierite, EtNO2 (0⬚C)

13

77 ⴙ 78

Ph2Sn — —S, AgClO4

PhH, 3A-MS (23⬚C)

14

77 ⴙ 79

Ph2Sn — —S, AgClO4

PhH, 3A-MS (23⬚C)

15

77 ⴙ 78

PhCH3, 3A-MS (23⬚C)

16

77 ⴙ79

Lawesson’s reagent, AgClO4 Lawesson’s reagent, AgClO4

4A-MS, CH2Cl2 (reflux) 4A-MS, CH2Cl2 (reflux) Cl(CH2)2Cl (53⬚C)

PhCH3, 3A-MS (23⬚C)

Yield (%) (␣ :␤ ) 92 (0:100) 98 (04:96) 85 (48:52) 76 (55:45) 61 (62:38) 43 (88:12) 97 (05:95) 97 (05:95) 96 (94:06) 86 (96:04) 90 (09:91) 92 (04:96) 89 (13:87) 90 (04:96) 79 (24:76) 90 (04:96)

Ref. 55 55 55 56 56 56 57 57 57 58 59 59 60 60 60 60

The authors found that simple thiols also are effective acceptors in this coupling ˚ molecular sieves, the glycoside yields were method, and, in the presence of 4 A uniformly high. The work of Susaki established that Zn(OTf)2 can be employed in dehydrative glycosylation [56]. The combination of Zn(OTf)2 and TMSCl was found to be useful for glycosidic bond formation with both glycosyl ester donors, as well as 1-hydroxy carbohydrate donors (entries 4–6). In these couplings, the ratio of donor to acceptor to TMSCl to Zn(OTf)2 is 1:2:1.5:0.3, with either acetonitrile or CH2Cl2 as the reaction solvent. Benzylated gluco-, galacto-, and mannopyranose donors generally afforded good yields of alkyl glycosides with anomeric selectivities that vary with

50

Gin

the nature of the coupling substrates. It is worth noting that couplings were significantly less efficient when an excess (1.5 equiv) of Zn(OTf)2 was employed. In a variety of the Lewis acid catalysts also surveyed by Mukaiyama and coworkers, benzylated furanose and pyranose substrates were found to be effective donors [57,58]. With the catalyst combination of the Sn(OTf)2 (0.01 equiv) and TMS2O (0.1 equiv) in the presence of anhydrous calcium sulfate, 2,3,5-tri-O-benzylD-furanose 77 can be used to directly glycosylate the alcohols 78 and 79 with high ␤-stereoselectivity (Table 1, entries 7, 8). The couplings are presumed to proceed via the TMS ether derivatives of the coupling partners, which are generated in situ, and the high ␤-selectivity is likely a result of product formation under thermodynamic control. In the presence of LiClO4 (1.5 equiv) as an additive, however, a dramatic reversal of the anomeric selectivity is observed, likely owing to the formation of the oxocarbenium perchlorate 66 (Scheme 14) as the active glycosylating agent (entries 9, 10). Various other glycosyl acceptors in addition to those in Table 1 were shown to be compatible with the Sn(OTf)2 /TMS2O method, including Cnucleophiles such as 1,3,5-trimethoxybenzene and O-silyl ketene acetals. Although the Sn(OTf)2 /TMS2O Lewis acid system is highlighted in Table 1, Mukaiyama has also demonstrated that other Lewis acid catalyst systems can be used, such as those incorporating Yb(OTf)3, La(OTf)3, and SnCl2, leading to comparable coupling efficiencies and stereoselectivities. In addition to metallic Lewis acid catalysts, it has been demonstrated that trityl salts can be used for the efficient ribofuranosylation of alcohols [59]. With 77 as the donor and 3 mol% of [Ph3C]⫹[B(C6F5)4]⫺ catalyst in the presence of anhydrous CaSO4, high yields of various furanosides were obtained with only 1.3 equiv of the alcohol acceptors, affording high selectivity for the thermodynamic ␤ isomer (entries 11, 12). In these studies, other counterions for the trityl cation were investigated ⫺ ⫺ in Mukaiyama’s survey of catalysts, including BF⫺ 4 , ClO4 , and SbCl6 , but ⫹ ⫺ [Ph3C] [B(C6F5)4] was found to be the most efficient in terms of yield, anomeric selectivity, and ease of catalyst handling. Two-component dehydrative catalysts have also been reported by Mukaiyama and coworkers for ribofuranosylation of alcohols. One such system is based on an earlier report by these investigators of the Ph2Sn — —S/Tf2O reagent combination for dehydrative coupling. When a catalytic amount of Ph2Sn — —S in combination with AgClO4 (0.2 equiv each) [60] were employed in the coupling of 77 with alkyl alcohols (1.2 equiv), good yields were obtained favoring the ␤ anomer of the product furanosides (entries 13, 14). The presumed active catalytic agent in the reaction is the Sn(IV) species 82. In like manner, the two-component catalyst system of Lawesson’s reagent and AgClO4 (0.2 equiv each) was found to be effective for coupling transformations (entires 15, 16), where the phosphonium species 83 is believed to be the active catalyst.

Glycosylation with 1-Hydroxy Donors

IX.

51

CONCLUSION

The importance of the development of new glycosylation methods cannot be disputed. Although many new variants of the traditional glycosylation processes have been developed over the years, comparatively fewer methods have been developed for direct dehydrative coupling with 1-hydroxy donors. The methods summarized in this chapter address many of the inherent obstacles associated with this strategy. For example, with the appropriate dehydrating agent in the glycosylation, the formation of the undesired by-products of hemiacetal self-condensation can be minimized, even when only a slight excess of the glycosyl acceptor component is employed in the coupling. Further development of these and other dehydrative coupling strategies will certainly establish this to be an efficient approach for the controlled assembly of complex carbohydrates. REFERENCES 1.

2.

3. 4. 5. 6. 7.

8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

(a) Synthetic Oligosaccharides. Indispensable Probes for the Life Sciences, Kovac, P., ed. ACS Symposium Series 560; American Chemical Society: Washington, DC, 1994. (b) Dwek, R. A. Chem. Rev. 1996, 96, 683–720. (a) Preparative Carbohydrate Chemistry; Hanessian, S., Ed.; Marcel Dekker: New York, 1997, Chapters 12–22. (b) Boons, G.-J. Tetrahedron 1996, 52, 1095–1121. (c) Danishefsky, S. J.; Bilodeau, M. T. Angew. Chem. Int. Ed. Engl. 1996, 35, 1380–1419. (d) Schmidt, R. R.; Kinzy, W. Adv. Carbohydr. Chem. Biochem. 1994, 50, 21–123. (e) Toshima, K.; Tatsuta, K. Chem. Rev. 1993, 93, 1503–1531. (f ) Sinay¨, P. Pure Appl. Chem. 1991, 63, 519–528. (a) Fischer, E. Chem. Ber. 1893, 26, 2400–2412. (b) Bochkov, A. F.; Zaikov, G. E. Chemistry of the O-Glycosidic Bond. Pergamon: Oxford, 1979. McCurry Jr., P. M.; Pickens, C. E. U.S. Patent 4,950,743, 1990. Wessel, H. P. J. Carbohydr. Chem. 1988, 7, 263–269. For example: Nepogod’ev, S. A.; Backinowsky, L. V.; Grzeszczyk, B.; Zamojski, A. Carbohydr. Res. 1994, 254, 43–60. For example: (a) Fischer, B.; Nudelman, A.; Ruse, M.; Herzig, J.; Gottlieb, H. E. J. Org. Chem. 1984, 49, 4988–4993. (b) Kuhn, M.; von Wartburg, A. Helv. Chim. Acta 1968, 51, 1631–1641. (c) Kochetkov, N. K.; Khorlin, A. J.; Bochkov, A. F. Tetrahedron Lett. 1964, 289–293. Lubineau, A.; Fischer, J.-C. Synth. Commun. 1991, 21, 815–818. Smirnyagin, V.; Bishop, C. T. Can. J. Chem. 1968, 46, 3085–3090. Defaye, J.; Gadelle, A. Carbohydr. Res. 1989, 186, 177–188. Brochette, S.; Descotes, G.; Bouchu, A.; Queneau, Y. J. Carbohydr. Chem. 1998, 17, 879–891. Koto, S.; Morishima, N.; Zen, S. Bull. Chem. Soc. Jpn. 1982, 55, 1543–1547. Koto, S.; Morishima, N.; Kusuhara, C.; Sekido, S.; Yoshida, T.; Zen, S. Bull. Chem. Soc. Jpn. 1982, 55, 2995–2999. Paulsen, H. Angew. Chem. Int. Ed. Engl. 1982, 21, 155–173. Koto, S.; Morishima, N.; Zen, S. Chem. Lett. 1976, 1109–1110. Leroux, J.; Perlin, A. S. Carbohydr. Res. 1978, 67, 163–178. Pavia, A. A.; Rocheville, J.-M.; Ung, S. N. Carbohydr. Res. 1980, 79, 79–89. Pavia, A. A.; Ung-Chhun, S. N. Can. J. Chem. 1981, 59, 482–489. Lacombe, J. M.; Pavia, A. A. J. Org. Chem. 1983, 48, 2557–2563. Koto, S.; Sato, T.; Morishima, N.; Zen, S. Bull. Chem. Soc. Jpn. 1908, 53, 1761–1762. Koto, S.; Morishima, N.; Owa, M.; Zen, S. Carbohydr. Res. 1984, 130, 73–83.

52

Gin

22.

Koto, S.; Miura, T.; Hirooka, M.; Tomaru, A.; Iida, M.; Kanemitsu, M.; Takenaka, K.; Masuzawa, S.; Miyaji, S.; Kuroyanagi, N.; Yagishita, M.; Zen, S.; Yago, K.; Tomonaga, F. Bull. Chem. Soc. Jpn. 1996, 69, 3247–3259. Koto, S.; Morishima, N.; Uchino, M.; Fukuda, M.; Yamazaki, M.; Zen, S. Bull. Chem. Soc. Jpn. 1988, 61, 3943–3950. Szeja, W. Synthesis, 1988, 223–224. (a) Hughes, D. L. Org. Prep. Proced. Int. 1996, 28, 127–164. (b) Mitsunobu, O. Synthesis 1981, 1–28. Jurczak, J.; Grynkiewicz, G.; Zamojski, A. Carbohydr. Res. 1975, 39, 147–150. Grochowski, E.; Jurczak, J. Carbohydr. Res. 1976, 50, C15–C16. Nicolaou, K. C.; Groneberg, R. D. J. Am. Chem. Soc. 1990, 112, 4085–4086. Szarek, W. A.; Jarrell, H. C.; Jones, J. K. N. Carbohydr. Res. 1977, 57, C13–C16. Smith, A. B. III; Hale, K. J.; Rivero, R. A. Tetrahedron Lett. 1986, 27, 5813–5816. Smith, A. B. III; Rivero, R. A.; Hale, K. J.; Vaccaro, H. A. J. Am. Chem. Soc. 1991, 113, 2092–2112. Grynkiewicz, G. Carbohydr. Res. 1977, 53, C11–C12. Kometani, T.; Kondo, H.; Fujimori, Y. Synthesis, 1988, 1005–1007. Roush, W. R.; Lin, X.-F. J. Org. Chem. 1991, 56, 5740–5742. Chida, N.; Ohtsuka, M.; Ogawa, S. Chem. Lett. 1988, 969–972. Chida, N.; Ohtuka, M.; Nakazawa, K.; Ogawa, S. J. Org. Chem. 1991, 56, 2976–2983. Roush, W. R.; Lin, X.-F. J. Am. Chem. Soc. 1995, 117, 2236–2250. (a) Hendrickson, J. B.; Schwartzman, S. M. Tetrahedron Lett. 1975, 277–280. (b) Hendrickson, J. B.; Hussoin, S. Md. J. Org. Chem. 1989, 54, 1144–1149. Mukaiyama, T.; Suda, S. Chem. Lett. 1990, 1143–1146. Garcia, B. A.; Poole, J. L.; Gin, D. Y. J. Am. Chem. Soc. 1997, 119, 7597–7598. Tidwell, T. T. Synthesis 1990, 857–870. Hendrickson, J. B.; Schwartzman, S. M. Tetrahedron Lett. 1975, 273–276. Coburn, M. D.; Hayden, H. H.; Coon, C. L.; Mitchell, A. R. Synthesis 1986, 490–492. Nenajdenko, V. G.; Vertelezkij, P. V.; Gridnev, I. D.; Shevchenko, N. E.; Balenkova, E. S. Tetrahedron 1997, 53, 8173–8180. Corey, E. J.; Gin, D. Y.; Kania, R. S. J. Am. Chem. Soc. 1996, 118, 9202–9203. Kahne, D.; Walker, S.; Cheng, Y.; Van Engen, D. J. Am. Chem. Soc. 1989, 111, 6881– 6882. Hara, R.; Mukaiyama, T. Chem. Lett. 1989, 1909–1912. Mukaiyama, T.; Hara, R. Chem. Lett. 1989, 1171–1174. Suda, S.; Mukaiyama, T. Chem. Lett. 1991, 431–434. Mukaiyama, T.; Matsubara, K.; Suda, S. Chem. Lett. 1991, 981–984. Mukaiyama, T.; Matsubara, K. Chem. Lett. 1992, 1041–1044. Williams, A.; Ibrahim, I. T. Chem. Rev. 1981, 81, 589–636. Tsutsumi, H.; Ishido, Y. Carbohydr. Res. 1981, 88, 61–75. Schmidt, E.; Moosmu¨ller, F. Liebigs Ann. Chem. 1955, 597, 235–240. Inanaga, J.; Yokoyama, Y.; Hanamoto, T. J. Chem. Soc. Chem. Commun. 1993, 1090– 1091. Susaki, H. Chem. Pharm. Bull. 1994, 42, 1917–1918. Mukaiyama, T.; Matsubara, K.; Hora, M. Synthesis, 1994, 1368–1373. Uchiro, H.; Miyazaki, K.; Mukaiyama, T. Chem. Lett. 1997, 403–405. Uchiro, H.; Mukaiyama, T. Chem. Lett. 1996, 79–80. Shimomura, N.; Mukaiyama, T. Bull. Chem. Soc. Jpn. 1994, 67, 2532–2541.

23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60.

3 Chemistry of Glycosyl Triflates: Synthesis of ␤-Mannopyranosides David Crich University of Illinois at Chicago, Chicago, Illinois

I.

INTRODUCTION

Among the numerous and varied glycosyl donors known, the glycosyl sulfonates have been somewhat neglected in recent years, despite their considerable potential in stereoselective glycosylation having been demonstrated over 20 years ago by Schuerch and coworkers. This unfortunate circumstance perhaps arose because of the difficulties encountered by the early workers in the isolation of pure samples. Recent work from our laboratories in Chicago has revealed that glycosyl triflates may be very readily accessed from either glycosyl sulfoxides or thioglycosides and that these substances have excellent reactivity toward a broad spectrum of glycosyl acceptors even at ⫺78⬚C. The main part of this chapter is therefore concerned with the development of the chemistry of the glycosyl triflates and, especially, with their applications in the synthesis of ␤-mannopyranosides. However, there are useful lessons to be learned from the precedents alluded to above, and it is with this in mind that we begin with an overview of the earlier work. II.

GLYCOSYL SULFONATES

The first reported preparation of a glycosyl sulfonate, in 1929, involved the heating of acetobromoglucose with silver toluenesulfonate in diethyl ether; Helferich and Gootz noted that the product was white and crystalline but decomposed in a matter of hours in chloroform solution at room temperature [1]. Many years after this inauspicious start, Schuerch and his coworkers turned their attention to an exploration of the uses of glycosyl sulfonates as glycosyl donors. Thus, in a 1973 paper that presaged many of the later developments, Kronzer and Schuerch suggested that the 53

54

Crich

treatment of 2,3,4,6-tetra-O-benzyl-␣-D-glucopyranosyl chloride (1) or bromide with silver triflate at ⫺78⬚C in dichloromethane or diethyl ether provided the anomeric triflate 2 (Scheme 1) [2]. They opined that, owing to the strongly electron-withdrawing nature of the triflate group, this substance probably had the ␣ configuration. It was also noted that subsequent couplings with methanol, conducted at ⫺78⬚C, were ␤-selective in dichloromethane but unselective in diethyl ether, and the difference was attributed to the superior shielding of the ␣ face provided by the tighter ion pair in dichloromethane [2]. The rapidity of the reactions, both the initial formation of the triflate and its subsequent reaction with methanol at ⫺78⬚C, were also noted at this time [2]. Prompted by the extreme reactivity of the triflates and the desire to work with isolable, characterized intermediates, Schuerch and Eby subsequently turned their attention to the use of toluenesulfonates and mesylates [3]. Thus, 2,3,4,6-tetra-Obenzyl-␣-D-glucopyranosyl chloride, or bromide, was allowed to react with silver toluenesulfonate in acetonitrile at room temperature. The authors were able to use standard vacuum line techniques to remove the silver halide by filtration and obtain an 1H NMR spectrum of the product in deuteriochloroform: the characteristics of the anomeric proton (␦ 6.1, 3J1,2 = 3.5 Hz) led to the attribution of the ␣ stereochemistry. In the case of the analogous 2,3,4-tri-O-benzyl-6-O-(N-phenylcarbamoyl) series, signals attributed to the ␤-toluenesulfonate (␦ 5.5, 3J1,2 = 8.0 Hz) accounted for approximately 15% of the reaction mixture [3]. In agreement with the earlier report of Helferich and Gootz, complete decomposition was noted after several hours at room temperature. Coupling reactions of the so-formed 2,3,4,6-tetra-O-benzyl-1-O-tosyl␣-D-glucopyranose, conducted with methyl 2,3,4,-tri-O-benzyl-␣-D-glucopyranoside in a range of solvents, gave moderate yields but little selectivity. Kinetic measurements, carried out polarimetrically, revealed no clear dependence of reaction rate on alcohol structure or concentration and suggested to the authors that the couplings were SN1 in character and involved a series of interchanging tight ion pairs [3]. Eby and Schuerch also noted that the anomeric toluenesulfonates formed from silver toluenesulfonate and peracylated (acetyl, benzoyl) glucosyl halides were considerably more stable, and less reactive, and upon exposure to methanol led to mixtures of ␣- and ␤-glucosides as well as to orthoesters. These latter results prompted them to suggest that the esterified toluenesulfonates would probably be of little use in glycosylation [3]. Schuerch and his coworkers also investigated the galactopyranosyl toluenesulfonates and triflates. As in the glucose series, these were formed from the corresponding bromides by metathesis reactions with the appropriate silver sulfonates [4,5]. With a series of differentially 6-O-protected 2,3,4-tri-O-benzylgalactopyranosyl

Scheme 1

Chemistry of Glycosyl Triflates

55

sulfonates, excellent yields and high ␣-selectivity were obtained upon coupling to a range of alcohols in various solvents. It was noted that the ␣-selectivity was minimized with the use of the triflate leaving group, albeit for reactions conducted at ⫺78⬚C, in contrast to the other members of the series when room temperature was used. In the case of the reaction of the 4,6-di-O-(N-phenylcarbamoyl)-2,3-di-O-benzylgalactopyranosyl triflate with methanol, high ␤-selectivity was exceptionally obtained. These results were again all interpreted in terms of more or less tight ion pairs [4,5]. Finally, Srivastava and Schuerch studied the formation of ␤-mannopyranosides and the related ␤-rhamnopyranosides, using a range of glycosyl sulfonates as donors [6,7]. In this most difficult series, the authors emphasized the increased anomeric effect resulting from the antiparallel dipoles of the C1—O and C2—O bonds and sought to maximize this contribution by installing a strongly electron-withdrawing, nonparticipating protecting group on O2. Thus, 3,4,6-tri-O-benzyl-2-O-mesyl-␣-Dmannopyranosyl chloride (4) was prepared and reacted with a range of silver sulfonates to give the corresponding ␣-mannosyl sulfonates. Subsequent exposure to methanol or cyclohexanol, usually in acetonitrile, then gave the mannopyranosides in high yield and excellent ␤-selectivity. Only one carbohydrate-based alcohol (6) was used as glycosyl acceptor, but this too gave excellent ␤-selectivity especially with the 2,2,2-trifluoroethanesulfonate 5 as donor (Scheme 2). In considering the mechanism of their mannosylation reaction, Srivastava and Schuerch suggested that the electron-withdrawing sulfonate ester at the 2-position served to render the reactive ion pair in the SN1 mechanism tighter on the ␣ face than was the case in the glucose and galactose series. In this manner the ␣ face was considered to be highly shielded toward approach of the nucleophile, with the displacement taking on the high stereoselectivity normally associated with SN2 processes [6,7]. Similar ␤-selectivities were obtained from a 3,4-di-O-benzyl-2-O-mesyl␣-L-rhamnopyranosyl tosylate [6,7]. Although the authors obtained excellent

Scheme 2

56

Crich

␤-selectivity and yields in their mannosylations and rhamnosylations, they repeatedly alluded to the high moisture sensitivity of the anomeric sulfonates, as in the earlier glucose and galactose series, and the need to work on vacuum lines. Presumably this, and the need to remove the somewhat unconventional 2-O-mesylate protecting group subsequent to coupling, prevented them from developing the full potential of the method and applying it in the synthesis of oligosaccharides. More or less contemporaneously with the work of Schuerch, Leroux and Perlin studied the sulfonylation of pyranoses. Thus, 2,3,4,6-tetra-O-benzyl-␣-D-glucopyranose was treated in cold dichloromethane with triflic anhydride in the presence of 2,4,6-collidine, leading to a purported anomeric triflate. However, the authors noted that subsequent addition of methanol or ethanol did not lead to the formation of the glycoside in acceptable yields [8,9]. Subsequently, it was discovered that conducting the triflation in the presence of tetrabutylammonium bromide resulted in the formation of the ␣-glucopyranosyl bromide and that addition of an alcohol then led cleanly to the glycoside. In these reactions Perlin and Leroux typically carried out the triflation/bromide displacement at ⫺70⬚C, noting that bromide formation was complete after 15–30 min; the reaction mixture was warmed to room temperature before addition of the alcohol. However, it was also noted that bromide formation could be conducted at room temperature without apparent detriment. These workers also investigated the use of 2,3,4,6-tetra-O-acetyl-␤-D-glucopyranose (8), with its potentially participating protecting groups, as substrate. In the presence of collidine but the absence of bromide ion, an orthoester (10) was isolated in 44% yield. When bromide ion was included in the reaction mixture, acetobromoglucose (11) was isolated in 50% yield and there was no indication of orthoester formation. It was suggested that both reactions proceeded by way of the ␤-triflate (9), which was trapped either by further pyranose or by bromide ion according to the conditions employed (Scheme 3) [8,9]. The reaction of methanesulfonic anhydride with 2,3,4,6-tetra-O-benzyl-␣-Dglucopyranose in dichloromethane in the presence of collidine, followed by addition of methanol, resulted in the formation of a 3:2 ␣/␤ mixture of the methyl glycosides, isolated in 87% yield. Since glycosylation was achieved without the need for addition of the quaternary ammonium bromide, unlike the case of triflic anhydride, it was concluded that the glucosyl mesylate was considerably more stable and allowed for the displacement reaction to take place. Moreover, the anomeric ratio suggested that the ␣ and ␤ anomers of this mesylate were in equilibrium, with displacement of the ␤ anomer occurring more rapidly. When 2,3,4,6-tetra-O-acetyl-␤-D-glucopyranose was treated with methanesulfonic anhydride, a crystalline product was obtained in 74% yield and assigned as the ␣-glucosyl mesylate. Indeed the characteristics of this substance were comparable to those of the compound obtained by Schuerch upon treatment of acetobromoglucose with silver methanesulfonate [8,9]. When methanesulfonyl chloride was allowed to react with 2,3,4,6-tetra-O-benzylglucopyranose and collidine in dichloromethane, the ␣-glucopyranosyl chloride was isolated regardless of whether the quaternary ammonium bromide was included. Addition of methanol to the reaction mixture resulted in the formation of an anomeric mixture of methyl glycosides. Similar results were obtained with toluenesulfonyl chloride, although it was noted that the initial sulfonylation was somewhat slower [8,9]. The use of tosyl chloride in the dehydrative coupling of alcohols with pyranoses was later revisited by Szeja and his coworkers, with the difference that aqueous

Chemistry of Glycosyl Triflates

57

Scheme 3

phase transfer conditions were used, and glycosyl toluenesulfonates were implied as intermediates [10,11]. Koto and coworkers investigated the coupling of tetra-O-benzyl-␣-D-glucopyranose and a range of acceptor alcohols with the aid of a mixture of 4-nitrobenzenesulfonyl chloride and silver triflate [12–15]. In the presence of triethylamine, the ␣-glycoside predominated, whereas the inclusion of N,N-dimethylacetamide resulted in the isolation of the ␤ anomer. It was suggested that the anomeric hydroxyl group was sulfonylated with the sulfonyl chloride to a glucopyranosyl 4nitrobenzenesulfonate and subsequently converted to the active glycosyl donor, the glucopyranosyl triflate, by the action of silver triflate. However, given the relative acidities of triflic and 4-nitromethanesulfonic acid and the related work of Szeja, carried out in the absence of silver triflate, the formation of a covalent glycosyl triflate in this work appears to be somewhat unlikely. The inversion of stereoselectivity on inclusion of the N,N-dimethylacetamide was explained by invoking the formation of an ␣-N,N-dimethylacetimidate ester [12,13]. Pavia et al. revisited the reactions of tetra-O-benzylglucopyranose (12) with trifluoromethanesulfonic anhydride, but in the absence of base. They discovered that the corresponding trehalose derivatives (13) were formed in good yield, predominantly as the ␣,␣ form (Scheme 4) [16]. Comparable results were obtained in the galacto-, manno-, and arabinopyranose series as with fructofuranose [16]. When trifluoromethanesulfonic anhydride was added to a mixture of a perbenzyl-protected glycopyranose and an acceptor alcohol, such as various serine, threonine and hydroxproline derivatives, coupling was achieved in good yield [17]. Pavia and coworkers carried out a careful study of the mechanism of these reactions using 19F-NMR spectroscopy. They concluded that they were observing simple acid-catalyzed dehydrative couplings in which water was removed from the

58

Crich

Scheme 4

equilibrium in the form of a salt with triflic acid (TfO⫺ H3O⫹), which is insoluble in dichloromethane. The authors specifically excluded the intermediacy of a covalently bound glucosyl triflate on the grounds that the reaction did not occur below 15⬚C, whereas Perlin’s reactions succeeded at ⫺70⬚C [18]. III.

MANNOSYL TRIFLATES FROM MANNOSYL SULFOXIDES AND THIOGLYCOSIDES: THE ␤-MANNOSYLATION REACTION

Several years ago in Chicago, we were engaged in developing a solution to the wellknown ␤-mannoside problem [19] involving the inversion of the much more readily accessible ␣-mannosides by a sequence of hydrogen atom abstraction, radical inversion, and diastereoselective quenching (Scheme 5) [20,21]. With the fundamental chemistry, which was developed using commercially available ␣-methyl mannopyranoside as substrate in hand, it became necessary to prepare a genuine ␣-disaccharide for inversion. We selected Kahne’s excellent sulfoxide method [22–24] for numerous reasons, which included the reported excellent yields for coupling to extremely hindered alcohols at low temperatures, the absence of any metal salt as promoter, and the implication that the coupling proceeded through quenching of the oxacarbenium ion, hence should provide the ␣-mannoside. Other than a brief footnote in their original communication noting that the anomeric stereoselectivity was a function of the stereochemistry at C2 of the sulfoxide donor [22], Kahne and his coworkers had not described the application of their method to the mannose series. Jarmila Brunckova thus prepared S-ethyl 4,6-benzylidene-␣-Dthiomannopyranoside by standard means and converted it to the 3-O-benzyl-2-Otert-butyldimethylsilyl derivative (14) by the aegis of dibutyltin oxide and benzyl bromide, then TBDMS triflate. Oxidation with mCPBA then gave the sulfoxide (15) required for coupling [21]. Brunckova noted the highly selective sulfoxidation process, which gave essentially a single diastereomer, in contrast to the unselective oxidations ␤-thioglycosides observed previously [23], but was unable to assign configuration at the time. In fact it was several years before Jan Mataka and Sanxing Sun were able to prepare crystalline derivatives suitable for X-ray analysis and so assign the configuration as SR [25]. Brunckova then mixed the sulfoxide with methyl

Scheme 5

Chemistry of Glycosyl Triflates

59

Scheme 6

2,3,4-tri-O-acetyl-␣-D-glucopyranoside (16) and 2,6-di-tert-butyl-4-methylpyridine (DTBMP), a hindered base, in diethyl ether at ⫺78⬚C and activated the sulfoxide by dropwise addition of triflic anhydride. In line with our expectations, a good yield of a 10:1 mixture of glycosides (17) favoring the ␣ anomer was obtained (Scheme 6) [21]. The TBDMS protecting group was removed and was replaced by the radical precursor, and the radical inversion procedure was conducted with moderate success, comparable to that seen with the ␣-methyl mannoside [21]. Sanxing Sun, a new student, sought to prepare more of the ␣-disaccharide (17) and subsequently to improve the radical inversion process. He repeated Brunckova’s preparation with the minor, but fortuitous, difference that the sulfoxide was activated with triflic anhydride before addition of the acceptor alcohol. To our amazement, an excellent yield was obtained of an anomeric mixture favoring the ␤-mannoside by a factor of roughly 10 (Scheme 7) [26,27]. Consideration of this unanticipated reversal of diastereoselectivity led us to propose a mechanism in which, under all conditions, the sulfoxide 15 (Scheme 8) undergoes rapid sulfonylation leading to a sulfonium ion 18. We then postulated that this sulfonium ion collapses to the oxacarbenium ion 19 which, when generated in

Scheme 7

60

Crich

Scheme 8

the presence of the donor a` la Brunckova, is trapped axially to give the anticipated ␣-mannoside (17␣). On the other hand, we suggested that, under the Sun conditions, the oxacarbenium ion is trapped by triflate anion to give an ␣-mannosyl triflate 20. Then, on subsequent addition of the acceptor, an SN2-like process occurs with formation of the ␤-mannoside (17␤) [27,28]. This mechanistic hypothesis also provided a reason for the poor ␤-selectivity observed with secondary alcohol acceptors, even under the Sun conditions, namely, the well-known retardation of SN2 reactions by steric hindrance, leading to the interference of a dissociative mechanism via a Curtin–Hammett type of kinetic scheme. We predicted therefore that reducing the size of the O2 protecting group on mannose would accelerate the SN2 process for any given alcohol and so lead to increased ␤-selectivity. We also predicted that a change in solvent from diethyl ether to dichloromethane would further shift any ion pair/covalent triflate equilibrium toward the covalent triflate and so similarly lead to enhanced ␤-selectivity [27,28]. Both hypotheses were readily tested and confirmed. Thus, the series of sulfoxides 15, 21 and 22, with decreasing bulk of the O2 protecting group, were prepared and coupled to the rhamnosyl acceptor 23. As seen from Table 1, selectivity increases both as the size of the protecting group decreases and as the solvent is changed from diethyl ether to dichloromethane. Ultimately, with the 2-O-benzyl donor in dichloromethane as solvent, the ␣ anomer of the product was not detectable [28]. These conditions were then applied to the ␤-mannosylation of a range of primary and secondary carbohydrate acceptors with considerable success (Table 2) [27,28]. Attention was next focused on the use of alternative, nonparticipating protecting groups for the mannosyl donor, and the allyl group was found to be satisfactory at both O2 and O3 (Scheme 9) [27]. Indeed, Zongmin Dai and Greg Barba subsequently used a 2-O-allyl protected donor to form the ␤-mannoside linkage in the Hyriopsis schlegelii trisaccharide and the caloporside disaccharide, respectively [29– 31]. Further work, however, revealed the 4,6-benzylidene group to be indispensable for high ␤-selectivity, inasmuch as a 2,3,4,6-tetra-O-benzyl protected mannosyl donor gave very poor selectivity (Scheme 10) [27,28].

Chemistry of Glycosyl Triflates

Table 1

61

Reaction of Glycosyl Donors with 23 in Ether and CH2Cl2 Product and yield (%)b

Donor

a

Protocola

␤-Mannoside

␣-Mannoside

␤:␣

Et2O

A

24␤, 49

24␣, 30

1.6 : 1

Et2O

A

25␤, 76

25␣, 15

5.1 : 1

Et2O

A

26␤, 74

26␣, 11

6.7 : 1

CH2Cl2

B

24␤, 82

24␣, 11

7.5 : 1

CH2Cl2

B

25␤, 92

25␣, 7

13.1 : 1

CH2Cl2

B

26␤, 90

26␣, 0

>25 : 1

Solvent

A, Protocol A: addition of ROH to premixed donor, Tf2O, and DTBMP in ether/benzene; Protocol B, addition of ROH to premixed donor, Tf2O, and DTBMP in CH2Cl2. b All except 26␤ and 26␣ isolated after treatment of the reacton mixture with TBAF.

Table 2

Coupling of Secondary Acceptors to 22 and 27 in CH2Cl2 Product and yield (%)

␤-Mannoside

␣-Mannoside

␤:␣

29␤, 93

29␣, 5

18.6 : 1

27

31␤, 90

31␣, 6

15.0 : 1

27

33␤, 94

33␣, 5

18.8 : 1

27

35␤, 31

35␣, 8

3.8 : 1a

27

37␤, 72

37␣, 13

5.5 : 1

27

39␤, 94

39␣, 3

31.3 : 1

33␤, 91

33␣, 7

13.0 : 1

Donor

a

Acceptor

The reaction mixture was allowed to come to room temperature and stirred there for 24 h before workup.

Chemistry of Glycosyl Triflates

63

Scheme 9

Scheme 10

With a series of successful couplings in hand, we returned to the question of mechanism and the hypothesis of glycosyl triflates as the key reaction intermediates. A donor 44, lacking any diastereotopic benzyl hydrogens, was prepared and its 1HNMR spectrum recorded in CD2Cl2 in the presence of DTBMP at ⫺78⬚C. Cold (⫺78⬚C) Tf2O was then added, and a rapidly recorded spectrum showed the sulfoxide to have been completely consumed and converted to a new carbohydrate species [32]. This new substance was characterized in the 1H-NMR spectrum by its anomeric proton, a broad singlet, which resonated at ␦ 6.20. In the 13C-NMR spectrum the anomeric carbon had a chemical shift ␦ of 104.6 and a 1JCH coupling of 184.5 Hz. These data indicated that a strongly electron-withdrawing group was covalently linked through oxygen to the ␣ position of the mannopyranose ring, providing strong support for the ␣-mannosyl triflate (45) hypothesis. Confirmatory evidence was obtained when treatment of the mannsoyl bromide 46 with AgOTf and DTBMP in CD2Cl2 gave indistinguishable spectra [32]. This experiment also provided strong support for Schuerch’s earlier hypothesis that glycosyl triflates were formed at low temperature upon treatment of glycosyl halides with AgOTf (see above). Addition of methanol at ⫺78⬚C to these NMR tube experiments resulted in the very rapid formation of mannosides (47) with high ␤-selectivity, in full agreement with the general mechanism proposed (Scheme 11) [32]. A tetra-O-methylmannosyl sulfoxide 48 was prepared as a surrogate for the unselective tetrabenzyl donor 42. Again, low-temperature 1H and 13C NMR experiments indicated clean formation of a covalently oxygen-linked intermediate and, again, the same spectra were obtained going out from the bromide and AgOTf [32]. A first clue to reasons underlying the differing selectivity of the tetramethyl (or benzyl) and 4,6-benzylidene series was obtained in the course of attempts to record a CH-gated, coupled 13C-NMR spectrum of 49. Substantial decomposition occurred

64

Crich

Scheme 11

over the acquisition time, unlike the case of the 4,6-benzylidene donor (44), which prevented us from obtaining the 1JCH coupling constant. This difference in stability was confirmed by a series of variable temperature experiments, which revealed the 4,6-benzylidene protected triflate (45) to decompose around ⫺10⬚C, whereas its tetramethyl congener (49) did so some 20 degrees lower (⫺30⬚C). These latter observations provide strong support for the notion that the ␣/␤-selectivity in these couplings is a function of the equilibrium between the covalently bound triflate and the ion pair. The less stable the triflate, as reflected in the lower decomposition temperature, the greater the population of the ion pair and the greater the likelihood that ␣-mannosylation will occur through a Curtin–Hammett type of kinetic scheme. The increased stability of the 4,6-benzylidene protected triflates may be rationalized in terms of Fraser-Reid’s concept of a torsionally disarming protecting group [33]. In effect, the sofa conformation of the oxacarbenium ion imposes a twist and torsional strain on the acetal ring, which increases the energy of the oxacarbenium ion with respect to that of the covalently bound triflate. This effect is not present in the perether protected systems.

The basic mechanistic hypothesis (Scheme 8) suggests that the ␤-mannosides are formed in an SN2-like manner from the ␣-triflate. The NMR experiments provide overwhelming support for the formation of these triflates and, in the 4,6-benzylidene series, for their stereochemistry. However, there remains the possibility, which we cannot rule out, that the covalently bound triflates simply serve as a reservoir for the storage of extremely reactive contact ion pairs. In this hypothesis, which echoes that of Schuerch (see above), the ␣-triflate dissociates to a contact ion pair in which the anion is intimately associated with the ␣ face of the oxacarbenium ion and so sterically prevents approach from that face. In this hypothesis, reduced selectivity, as

Chemistry of Glycosyl Triflates

65

Scheme 12

seen with the per-ether protected donors, arises from intervention of solvent and of solvent-separated ion pairs. This hypothesis therefore requires a change in the equilibrium between contact ion pairs and solvent-separated ion pairs in going from the 4,6-benzylidene series to the per-ether protected series. Again we fall back on the extra strain imposed on the oxacarbenium ion by the 4,6-benzylidene group, which will have the effect of shifting the CIP/covalent triflate equilibrium toward the covalently bound species, hence of reducing the likelihood of intervention of a solvent-separated ion pair. This variation of the original mechanism is presented in Scheme 12. Although we have provided compelling evidence for the formation of mannosyl triflates in our system, it is not implied that related triflates are formed in all sulfoxide glycosylation reactions. Indeed, Kahne and his coworkers subsequently investigated the mechanism of their reaction and found that in many cases the original sulfoxide is rearranged to isolable sulfenate esters, which also serve as glycosyl donors as shown, for example, in Scheme 13 [34]. The precise reasons for the different reactivity patterns are not clear and are the subject of ongoing research in our laboratory. In the course of our investigations into the mechanism of the reaction, we became aware of the extreme electrophilicity of the by-product benzenesulfenyl tri-

Scheme 13

66

Crich

Scheme 14

flate (53). As demonstrated by control experiments, this species is itself able to convert glycosyl sulfoxides into glycosyl triflates on a time scale comparable to that of the Tf2O activation (Scheme 14) [32]. Sanxing Sun was so struck by the electrophilicity of PhSOTf that he tested its ability to convert thioglycosides into triflates at ⫺78⬚C. His curiosity was immediately rewarded when low-temperature NMR experiments indeed showed this reagent to cleanly convert a mannosyl thioglycoside (54) into triflate (45) within the space of minutes at ⫺78⬚C (Scheme 15) [27,35]. Sun then went on and provided a series of examples of his new reaction, involving coupling to primary, secondary, and tertiary alcohols as set out in Table 3 [27,35]. The obvious advantage of this new method is that it eliminates the need to oxidize thioglycosides to sulfoxides. The disadvantage is that the reagent has to be freshly prepared in situ from AgOTf and benzenesulfenyl chloride which, itself, has only limited shelf life. In 1997 Gin and coworkers published an intriguing new method for the dehydrative coupling of pyranoses with alcohols in which the pyranose (61) is activated with a mixture of diphenyl sulfoxide and triflic anhydride before addition of the acceptor alcohol, leading to the formation of the coupled product (63) in excellent yield [36]. It was suggested that this chemistry, a development of that of Perlin and Pavia (see above), proceeds via the sulfonylation of the sulfoxide and attack of the pyranose on the so-formed sulfonium salt (62). In a NMR experiment, the mannosyl sulfonium salt (65) has been shown to be stable at low temperature in the absence of nucleophiles (Scheme 16). Indeed, it could even be generated by addition of diphenyl sulfoxide to the corresponding glycosyl triflate (49), which was generated from the fluoride (64) with trimethylsilyl triflate. Glycosyl triflates therefore appear not to be intermediates in this chemistry [37]. IV.

␤-THIOMANNOSIDES

Thiols are generally regarded as better nucleophiles than alcohols in substitution reactions. Accordingly, Hongmei Li investigated the use of thiols as nucleophiles in

Scheme 15

Chemistry of Glycosyl Triflates

Table 3

67

␤-Mannoside Formation from Thioglycosides with PhSOTf in CH2Cl2 Product and yield (%)

␤:␣

26, 95

>25 : 1

55

33, 95

<25 : 1

55

29, 95

23 : 1

55

31, 97

18 : 1

55

37, 85

5.1 : 1

55

58, 94

>25 : 1

Glycosyl donor

Glycosyl acceptor

(Table 3 continues on p. 68)

our ␤-mannosylation protocol, with the expectation that even higher ␤-selectivity would be observed. This proved to be the case with a range of primary, secondary, and tertiary thiols (Table 4) with no ␣-thiomannoside being detected in any case. The yields, although good, were nevertheless somewhat less than we had typically observed with alcohols [38]. We attribute this phenomenon to the partial capture of the soft nucleophiles by one or other of the several sulfur-based electrophilic byproducts present in the reaction mixture.

Table 3

Continued

Glycosyl donor 55

a

After treatment with TBAF.

Scheme 16 68

Glycosyl acceptor

Product and yield (%)

␤:␣

60, 90

>25 : 1

26,a 80

10 : 1

Chemistry of Glycosyl Triflates

Table 4 Acceptor

V.

69

Synthesis of ␤-Thiomannosides from Sulfoxide 27 Product

Yield (%)

69

74

70

77

72

54

Acceptor

Product

Yield (%)

74

71

76

63

␣-SELECTIVE MANNOSYLATION REACTIONS

In our continuing search for alternatives to the 4,6-benzylidene protecting group, and for 2,3-protecting groups that will confer enhanced selectivity even in the 4,6-benzylidene series, we have encountered several systems, which, contrary to our initial expectations, were highly ␣-selective. Zongmin Dai, recalling the very early work of Perlin in which the 2,3-O-carbonate was found to be highly ␤-selective when used in the mannosyl bromide/insoluble silver salt method [39], prepared the 2,3-O-carbonate (77) of S-phenyl 4,6-benzylidene-␣-D-mannothiopyranoside and investigated its reaction with 1,2;3,4-diacetone-D-galactose (78) by the benzenesulfenyl triflate method (Scheme 17). She found to her surprise that the only product isolated was the ␣-mannoside 79 [40]. Earlier Sanxing Sun had looked at the use of the 2,3;4,6diacetonide 80 as glycosyl donor in conjunction with methyl 2,3,4-tri-O-acetyl-␣-Dglucpyranoside as acceptor and had obtained a 1:1 ratio of the ␣- and ␤-mannosides [27]. Again, this result was surprising in view of the fact that Garegg had earlier successfully employed the 2,3;4,6-dicyclohexylidene protecting system in the synthesis of ␤-mannosides by the mannosyl bromide/insoluble silver salt method [41].

70

Crich

Scheme 17

We interpret both these results in terms of the general mechanism and the effect of the protecting groups on the covalent triflate/ion pair equilibrium. Thus, the fivemembered cyclic 2,3-O protecting groups lead to considerable flattening of the pyranose ring and so raise the ground state energy of the triflate. The barrier to formation of the flattened oxacarbenium is therefore reduced and the equilibrium shifted more in its favor. This effect is more pronounced in the carbonate than the acetonide, owing to the sp2 nature of the atoms in the bridge, and so leads to higher ␣-selectivity in that case. Zongmin Dai also investigated the use of a 2-O-TBDMS-3-O-benzoyl protected 4,6-O-benzylidene donor (81) but, again, excellent ␣-selectivity was observed [40]. This result must arise from neighboring group participation by the 3-O-benzoate. Models demonstrate that the twist–boat conformation (82) imposed by such participation on the pyranose ring is not unduly straining even though the system is 4,6benzylidene protected.

Weiling Cai prepared and investigated the Ley type [42,43] 3,4-bisacetal protected donor 83 in the expectation that the second ring would serve a similar rigidifying function as the 4,6-benzylidene group. This was not to be the case, and again excellent ␣-selectivity was observed in coupling to a number of acceptors (Table 5) [40].

Chemistry of Glycosyl Triflates

Table 5

71

␣-Mannoside Formation from Thioglycoside 83 with PhSOTf in CH2Cl2

Glycosyl acceptor

Product and yield (%)

␣:␤

84, 52

>95 : 5

85, 57

>95 : 5

86, 88

>95 : 5

87, 62

>95 : 5

88, 60

>95 : 5

72

Crich

With hindsight this result could have been predicted, inasmuch as Fraser-Reid’s earlier experimental and computational work on the formation of oxacarbenium ions from pentenyl glycosides showed this to be more facile with a six-membered ring bridging the 3- and 4-positions than in the absence of such a ring, and certainly much more facile than in the case of 4,6-benzylidene protected systems [33]. Thus, this system also fits the general mechanistic scheme, with the ␣-selectivity being again due to a shift in the covalent triflate/ion pair equilibrium. Furthermore, the reversal of selectivity seen across the series 4,6-benzylidene/tetrabenzyl/3,4-bisacetal is in full agreement with the increasing stability of the oxacarbenium ions, as predicted computationally by Fraser-Reid in the glucose series [33].

VI.

4,6-BENZYLIDENE PROTECTED GLUCOSYL TRIFLATES

Weiling Cai prepared a 4,6-O-benzylidene-2,3-di-O-benzyl-␣-D-glucopyranosyl thioglycoside 89 and oxidized it to the sulfoxide 90 in the expectation that activation under the standard conditions for ␤-mannosylation would lead to the formation of ␤-glucosides. Low-temperature studies in CD2Cl2 showed that both the sulfoxide and the thioglycoside, upon activation with Tf2O or PhSOTf, respectively, provided the ␣-glucosyl triflate 91 (␦H1 = 6.3; ␦C1 = 100.6; 3JH1,H2 = 3.5 Hz) in high yield (Scheme 18). However, all couplings except that to methanol were highly ␣-selective (Table 6) [44]. It is unlikely that this reversal of selectivity can be explained by reduced stability of the glucosyl triflate: variable temperature NMR studies showed that decomposition of the triflate did not set in until above 0⬚C (i.e., above the decomposition temperature of the corresponding mannosyl triflate). We believe that this result is a function of the differing extent of the anomeric effect in the two series. In effect it is known that the anomeric effect is considerably stronger in mannopyranose than glucopyranose derivatives [45,46]. Thus we suggest that the ␣- and ␤-triflates are in dynamic equilibrium and that the reduced anomeric effect in the glucose series permits a population of the more reactive ␤-triflate sufficient for it to serve as a donor in an SN2-like fashion, with Curtin–Hammett kinetics, leading preferentially to the ␣-glucosides observed. A similar rationale was advanced decades ago by Lemieux

Scheme 18

Chemistry of Glycosyl Triflates

Table 6

73

␣-Glucoside Formation Product and yield (%)

␣:␤

92, 98

>95 : 5

89

93, 87

>95 : 5

89

94, 70

>95 : 5

89

96, 85

>95 : 5

98, 89

>95 : 5

99, 72

>1 : 7.5

Glucosyl donor

90

Glycosyl acceptor

MeOH

74

Crich

for the formation of ␣-glucosides from ␣-acetobromoglucose in the presence of added bromide ion [45,47]. VII.

SUMMARY

Glycosyl triflates may be prepared rapidly and cleanly at ⫺78⬚C in CH2Cl2 solution from anomeric sulfoxides or thioglycosides upon activation with triflic anhydride or benzenesulfenyl triflate, respectively. These triflates are extremely reactive glycosyl donors and, when applied in conjunction with the 4,6-benzylidene protecting group, provide a very facile entry into the ␤-mannopyranosides. ACKNOWLEDGMENTS I thank the numerous talented students and postdoctorals whose hard work and diligence has contributed, and continues to contribute, to the development of glycosyl triflate chemistry in my laboratory. The National Institutes of Health (GM 57335) are gratefully acknowledged for their support of this work. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

20. 21. 22. 23. 24. 25. 26. 27.

B Helferich, R Gootz. Ber Dtsch Chem Ges 62:2788–2792, 1929. FJ Kronzer, C Schuerch. Carbohydr Res 27:379–390, 1973. R Eby, C Schuerch. Carbohydr Res 34:79–90, 1974. V Marousek, TJ Lucas, PE Wheat, C Schuerch. Carbohydr Res 60:85–96, 1978. TJ Lucas, C Schuerch. Carbohydr Res 39:39–45, 1975. VK Srivastava, C Schuerch. Carbohydr Res 79:C13–C16, 1980. VK Srivastava, C Schuerch. J Org Chem 46:1121–1126, 1981. J Leroux, AS Perlin. Carbohydr Res 47:C8–C10, 1976. J Leroux, AS Perlin. Carbohydr Res 67:163–178, 1978. W Szeja, J Bogusiak. Synthesis 224–225, 1988. W Szeja. Synthesis 223–224, 1988. S Koto, N Morishima, S Zen. Carbohydr Res 130:73–83, 1984. N Morishima, S Koto, S Zen. Chem Lett 1039–1040, 1982. S Koto, S Inada, T Yoshida, M Toyama, S Zen. Can J Chem 59:255–259, 1981. S Koto, T Sato, N Morishima, S Zen. Bull Chem Soc Jpn 53:1761–1762, 1980. AA Pavia, JM Rocheville, SN Ung. Carbohydr Res 79:79–89, 1980. JM Lacombe, AA Pavia, JM Rocheville. Can J Chem 59:473–481, 1981. AA Pavia, SN Ung-Chhun. Can J Chem 59:482–489, 1981. F Barresi, O Hindsgaul. Synthesis of ␤-D-mannose containing oligosaccharides. In SH Khan, RA O’Neill, eds. Modern Methods in Carbohydrate Synthesis. Amsterdam: Harwood Academic Publishers, 1996, pp 251–276. J Brunckova, D Crich, Q Yao. Tetrahedron Lett 35:6619–6622, 1994. D Crich, S Sun, J Brunckova. J Org Chem 61:605–615, 1996. D Kahne, S Walker, Y Cheng, DV Engen. J Am Chem Soc 111:6881–6882, 1989. L Yan, D Kahne. J Am Chem Soc 118:9239–9248, 1996. C Thompson, M Ge, D Kahne. J Am Chem Soc 121:1237–1244, 1999. D Crich, J Mataka, S Sun, K-C Lam, AR Rheingold, DJ Wink. J Chem Soc Chem Commun 2763–2764, 1998. D Crich, S Sun. J Org Chem 61:4506–4507, 1996. D Crich, S Sun. Tetrahedron 54:8321–8348, 1998.

Chemistry of Glycosyl Triflates 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.

47.

75

D Crich, S Sun. J Org Chem 62:1198–1199, 1997. D Crich, Z Dai. Tetrahedron Lett 53:1681–1684, 1998. D Crich, Z Dai. Tetrahedron 55:1569–1580, 1999. D Crich, GR Barba. Tetrahedron Lett 39:9339–9342, 1998. D Crich, S Sun. J Am Chem Soc 119:11217–11223, 1997. CW Andrews, R Rodebaugh, B Fraser-Reid. J Org Chem 61:5280–5289, 1996. J Gildersleeve, RA Pascal, D Kahne. J Am Chem Soc 120:5961–5969, 1998. D Crich, S Sun. J Am Chem Soc 120:435–436, 1998. BA Garcia, JL Poole, DY Gin. J Am Chem Soc 119:7597–7598, 1997. BA Garcia, DY Gin. J Am Chem Soc 122:4269–4279, 2000. D Crich, H Li. J Org Chem 65:801–805, 2000. PAJ Gorin, AS Perlin. Can J Chem 39:2474–2485, 1961. D Crich, W Cai, Z Dai. J Org. Chem 65:1291–1297, 2000. PJ Garegg, T Iversen, R Johansson. Acta Chem Scand B34:505–508, 1980. A Hense, SV Ley, H Osborn, DR Owen, J-F Poisson, SL Warriner, KE Wesson. J Chem Soc Perkin Trans 1:2023–2031, 1997. J-L Montchamp, F Tian, ME Hart, JW Frost. J Org Chem 61:3897–3900, 1996. D Crich, W Cai. J Org Chem 64:4926–4930, 1999. RU Lemieux, AR Morgan. Can J Chem 43:2214–2221, 1965. RU Lemieux. Rearrangements and isomerizations in carbohydrate chemistry. In P De Mayo, ed. Molecular Rearrangements, Part 2. New York: Wiley Interscience, 1964, pp 709–769. RU Lemieux, KB Hendriks, RV Stick, K James. J Am Chem Soc 97:4056–4062, 1975.

4 C-Glycoside Synthesis: Recent Developments and Current Trends Maarten H. D. Postema and Daniel Calimente Wayne State University, Detroit, Michigan

I.

INTRODUCTION

The chemistry of C-glycoside compounds has been of considerable interest over the last 20 years. This class of compounds is defined as compounds in which the exocyclic oxygen atom of an O-glycoside has been replaced by a carbon atom (Fig. 1). C-Glycosides are not solely man-made, but are the creation of nature, as attested by the occurrence of many C-glycoside natural products. Unlike the corresponding Oglycosides, C-glycosides are unaffected by hydrolytic or enzymatic cleavage and therefore would seem to be ideally suited for use as stable sugar mimics. They possess a level of chemical stability comparable to that of cyclic ethers. Several reviews have appeared on the subject of C-glycoside preparation, as well as two monographs. Many methods for the preparation of both the ␣- and ␤-C-glycosides have been developed, and new approaches are still appearing in the literature. This chapter covers the literature from the end of 1994 to early 1999, with other inclusions as deemed appropriate. Following the organization of earlier work [1], we select the mode reactivity of the anomeric center to define the category. The chapter discusses free radical chemistry, C1 anions, Wittig chemistry–cyclization chemistry, transition metal mediated chemistry, sigmatropic chemistry, and approaches based on cationic chemistry. The chapter focuses on the most recent and novel developments in the field, but all the relevant references are included whenever possible. 77

78

Postema and Calimente

Figure 1

II.

FREE RADICAL APPROACHES

A.

Introduction

The use of free radical chemistry at the anomeric center to produce carbon–carbon bonds, especially in an intramolecular fashion, remains a popular method for Cglycoside synthesis. A review entitled ‘‘C-Glycosidation Technology with Free Radical Reactions’’ appeared in early 1998, and the reader is referred to that source [2] as well as to the more traditional ones for complete overviews on the subject. Again, the coverage here focuses on the most recent developments. B.

Intermolecular Approaches

1.

Anomeric Radicals

Wang has studied the trapping of anomeric pyranosyl radicals with t-butylisocyanide to give anomeric cyanides. The readily available acetylated gluco derivative 1 when exposed to t-butylisocyanide in the presence of AIBN and tris(trimethylsilylsilane), gave a 73% yield of the ␣ isomer 2. Other sugars were examined and gave products with yields in the range of 22–71%, (Scheme 1) [3]. Junker and Fessner examined the addition of anomeric radicals to vinyl phosphonic dialkyl esters to deliver Cglycosyl-phosphonates directly. Thus exposure of 3 to standard radical conditions in the presence of vinyl phosphonic dimethyl ester gave 4 in 31% yield. Several other sugars were examined and generally gave yields of similar magnitude with excellent ␣ :␤ ratios, (Scheme 1) [4].

Scheme 1

C-Glycoside Synthesis

79

Scheme 2

The use and application to C-glycoside preparation of a different radical initiator also have been reported [5]. Schwartz has reported that treatment of glycosyl halides with (Cp2TiCl2)2 leads to glycals, presumably via the anomeric organometallic intermediate 6. However, if the reaction is carried out in the presence of a Michael acceptor, such as MVK, then the initially formed radical can be trapped and undergo carbon–carbon bond formation to give the ␣-C-glycoside 9. Methyl acrylate and acrylonitrile were also examined as radical traps. The product is presumably formed as the titanium enolate 8, which should be useful for further transformations (Scheme 2) [6]. Witzack et al. also carried out the following intermolecular radical addition to the levoglucosenone 11. The addition was highly exo-face selective and product 12 was obtained in 26% yield along with the remainder of product being that of direct reduction of 10, (Scheme 3) [7]. Allylations have continued to serve as useful reactions, and Bertozzi et al., found that radical allylation of the 2-deoxy-2-phthalimido derivatives 13 and 14 gives good to excellent selectivity of the ␤-allylated isomers 15 and 16, (Scheme 4). This is contrasted to allylations of the 2-deoxy-2-acetyl derivative, which gives mainly the ␣ product, which is the normal stereochemical course for additions of these types. Presumably the steric bulk of the phthalimido group plays a role in directing the stereochemistry of the addition [8]. Praly et al. cleverly used radical allylation of sugars to obtain ␤-C-glycosides. Treatment of the 1-bromoglucopyranosyl chloride 18 with tri-n-butylallyltin under photolytic conditions gave the intermediate ␤-allyl compound 19. When 19 was further exposed to standard radical reducing conditions (Bu3SnH, h␯) the expected ␤-allyl C-glycoside 20 was formed in good yield. Alternatively, 19 could be treated

Scheme 3

80

Postema and Calimente

Scheme 4

Scheme 5

Scheme 6

with DBU in acetonitrile to afford the product of elimination 21 in good yield (Scheme 5). The same chemistry was carried out on the D-manno and D-galacto derivatives which gave the analogous compounds in similar yields [9]. Work by Magnusson dealt with the application of allylic sulfones as radical acceptors for allylic carbon–carbon bond formation at the radical center. Various acceptors were utilized, and the products are shown in Scheme 6 [10]. The Kolbe electrolysis of sugar acids to give radicals and their subsequent trapping with a radical source gave the chain-extended C-glycoside 27 as the major product (Scheme 7). Compounds such as 27 are useful as stable nonionic tensides and are expected to show liquid crystalline properties [11]. 2.

Radical Addition to Sugar Olefins

Vismara, Nicotra, and their colleagues used glycoexenitol, 30 and 32 as radical acceptors in the preparation of precursors for isosteres of glycolipids and glycoproteins. In this case, the formed radical is anomeric, and reduction by tin hydride from the axial direction should then give the ␤-C-glycoside. The additions proceeded cleanly to give only 31 and 33 as the major products of (␣ :␤ = 10:4) noting that the yields

C-Glycoside Synthesis

81

Scheme 7

Scheme 8

Scheme 9

are quantitative when based on the amount of recovered glycoexenitol. Gervay has added thiolacetic acid to 32 to obtain 34 in good yield, (Scheme 8) [12]. Motherwell published the full paper corresponding to his initial work on radical additions to difluoromethylene-linked C-glycosides. Scheme 9 shows an application of the atom transfer work developed by Curran with iodide 34 to give the bis-C,Cdialkylglycoside in fair yield [13]. C.

Intramolecular Approaches

Fraser-Reid has published a full account of his work on serial radical cyclizations with carbohydrates. The example shown (38 → 40, Scheme 10) illustrates the complexity of the frameworks that can be accessed by careful application of this methodology [14]. Work by Beau, Skrydstrup, and their colleagues has focused on the generation of anomeric radicals from anomeric sulfones by treatment with samarium iodide [15]. If a radical trap is present, the second electron transfer to form the C1 anion does not have time to occur to any appreciable extent, and the formed radical can then

82

Postema and Calimente

Scheme 10

be trapped. The trapping with a sugar acetylene to give the ␣-C-glycoside 44 is illustrated in Scheme 11. Hung and Wong used glycosyl phosphates as radical precursors en route to Cglycosides via intramolecular samarium diiodide mediated cyclization as shown in Scheme 11 [16]. Work by Czernecki et al. focused on intramolecular additions of 2deoxy-2-amino sugars. Exposure of the diallylated species 47 to tin hydride in the presence of triethylborane gave compound 48 in 74% yield as the sole product [17]. Lee et al. carried out intramolecular cyclizations to C-glycoside-like structures using aldehydes such as 49. Exposure of 49 to tributyltin hydride and AIBN gave an Ostannyl ketyl intermediate that underwent 5-exo cyclization to give 50 and 51 in a 46:54 ratio in 91% combined yield, (Scheme 11) [18].

Scheme 11

C-Glycoside Synthesis

83

Scheme 12

Although not radical in nature, this class of species is electron deficient and their inclusion here is warranted. Vasella has continued his work on glycosyl carbenes. The carbene generated from 52, when trapped with dimethyl fumarate of Nphenylmaleimide, gave 53 and 54, respectively (Scheme 12) [19]. III.

C1-ANION APPROACHES

A.

Introduction

The chemistry associated with the anomeric center is cationic by nature. One interesting approach to C-glycoside synthesis is to reverse the electronic character of this center from electrophilic to nucleophilic. Hence anions at the anomeric center (glycosyl anions) have now become fairly common in the synthesis of C-glycosides. Of course, one limitation of this method was the presence of heteroatoms at C2, since any discrete anion at C1 would surely cause an elimination reaction to occur and in turn produce the corresponding glycal (55 → 58, Scheme 13). This is in fact a very common way of preparing sugar glycals from sugars. As a consequence of this limitation, the chemistry of 2-deoxy C1 anions (56 → 59) has been extensively explored. An interesting development in this area came with the observation that if the hydroxyl group at O2 is unprotected, deprotonation followed by anion formation at C1 can occur to give species 57, which can go on to react with a suitable electrophile to deliver 60, after workup. It should be noted that a publication dealing with the preparation of anomeric stannanes from 2,3-dihydro-4H-pyran-4-ones has appeared [20]. A related, and important observation was that anomeric organosamarium species possessing a protected oxygen atom at C2 are stable enough to react via the anomeric carbon atom with suitable electrophiles to give access to C-glycosides (61 → 62, Scheme 13) [21]. B.

Dianion Approaches

1.

O2 Sugars

Work by the groups of Vasella and Kessler (Scheme 14) focused on the preparation of dianions from the corresponding 1,2-anhydro sugars. Thus treatment of 63 with Ph3SnLi gave 64 as the major product in 65% yield. Exposure to an excess of nBuLi resulted in formation of dianion 65, which was then quenched with a variety

84

Postema and Calimente

Scheme 13

of electrophiles to give the ␤ isomers (66–69 ) as the major compounds in all cases, attesting to the configurational stability of these anions at low temperature [22]. This strategy was then applied to the preparation of the anomeric ␤ and ␣ acids 72 and 76. The ␤ isomer 72 was prepared by application of the epoxide-opening strategy above, while the ␣ acid 76 was synthesized by using the procedure originally developed by Kessler and coworkers. Once prepared, the acids were coupled with the protected asparagine amino acid 78 to give the unnatural C-glycosylated amino acids 73 and 74 (Scheme 15) [23]. Glycosyl phenyl sulfoxides also serve as a good source of dianions. The ␣sulfoxide was metallated to the dianionic species, treated with a suitable aldehyde to give alcohol 81 and reduced product 82 (Scheme 16) [24]. 2.

2-Deoxyamino Sugars

The Kessler group then investigated the question of whether similar dianion chemistry could be carried out with the corresponding 2-deoxy-2-amino sugars. Accordingly, compound 83 was treated with thionyl chloride to give the intermediate 2-deoxy-2-aminoglucosyl chloride, which was then treated sequentially with n-butyllithium and lithium naphthalenide to give intermediate 84. Quenching with a variety of electrophiles gave modest to good yields of the 2-deoxy-2-amino-␣-C-gly-

Scheme 14

C-Glycoside Synthesis

85

Scheme 15

Scheme 16

Scheme 17

cosides 85 stereoselectively (Scheme 17) [25]. Similar chemistry was also carried out in the galacto series [26]. The corresponding ␤ dianion 84 could be accessed by first converting 83 to the ␣-pyranosyl chloride followed by SN2 displacement with tributylstannyllithium to give 84. Deprotonation and subsequent halogen metal exchange was followed by quenching with various electrophiles to provide the ␤-isomeric C-glycosides 87 [25]. The chemistry was then applied to the preparation of a C-glycoside-based analog of N-glucoaspargine 88 (Scheme 18) [27]. C.

Anomeric Organosamarium Species: O2 Sugars

Work in the area of O2 sugars was spurred by the following observations: upon treatment of 89 with samarium iodide (to obtain an anomeric free radical that would

86

Postema and Calimente

Scheme 18

then cyclize onto the pendant acetylene), three products were obtained. The first, 91, was the expected product of free radical cyclization in 25% yield; the product of elimination 58 (11%) and the product of simple deoxygenation 90 (60%) also were isolated (Scheme 19). This was very surprising, since the elimination product is produced from the organometallic anomeric samarium species that then undergoes a 1,2-elimination, but 90 is derived from protonation of the same intermediate organometallic species! This implies that the organometallic species is stable enough to be deprotonated to give 90 [28]. The workers then took advantage of this reactivity to generate anions at the anomeric center of 2-oxygenated sugars and have succeeded in showing that pyridyl sulfones make excellent precursors for such reactions. When O2 is protected at a tbutyldimethylsilyl ether, acceptable yields of ␤-C-glycosides are obtained from the corresponding ␤-pyridyl sulfone (92 → 93, Scheme 20). When the ␣-manno derivative 94 was used, a good yield of the corresponding ␣-C-glycoside 95 was obtained [15]. The authors attribute the discrepancy in yields to the ability of the manno intermediate to attain a reactive boat conformation that is more stable than the gluco derivative. With anomeric phosphates such as 96 as precursors, Wong has used a similar reaction to prepare ␣-C-manno pyranosides. Good yields of ␣-C-glycosides were obtained (Scheme 20). In the absence of the electrophile, the 1-deoxy sugars were the main products formed. The use of furanoside 98 gave good yields of Cglycoside products [16]. The results were obtained in the gluco series paralleled those of Beau and coworkers. Scheme 21 shows the reaction of the 2-deoxy-2-amino sugar 100 reacting with samarium iodide and cyclohexanone to give the desired C-glycoside 101 as the major compound [29]. When the azide was used in lieu of the protected amine function, only the product of elimination was formed. This indicates a possible complexation between the acetimido oxygen atom (Scheme 21, second equation) and the samarium atom, resulting in stabilization of the ␣-organosamarium species, since normally,

Scheme 19

C-Glycoside Synthesis

87

Scheme 20 anomerization to the more stable ␤-organosamarium species would be expected to take place. In this case the complexation holds the organosamarium intermediate in the ␣ configuration. It may also be possible that a chair flip of 102 occurs to give 103, which can then react via the equatorial direction to ultimately give the ␣-Cglycoside [30]. The same workers applied their methodology to the preparation of a C-glycoside analog of the Tn antigen. The Tn antigen epitope is expressed in over 70%

Scheme 21

88

Postema and Calimente

Scheme 22

of human epithelial cancers. This epitope is a major tumor-associated O-linked glycoprotein. The same antigen has also been identified as a partial structure of the HIV envelope glycoprotein gp120. The workers designed an analog in which the interglycosidic serine oxygen atom was replaced with a methylene group. The anomeric pyridyl sulfone 100 was coupled with aldehyde 104 to give 105 as a mixture of isomers. Radical deoxygenation was followed by standard manipulations to deliver the analog building block 106 (Scheme 22) [31]. Lindhart and coworkers also relied on a samarium-mediated condensation reaction to prepare ␣-C-glycosides related to N-acetylneuraminic acid (Neu5Ac). Neu5Ac is frequently found as the terminating residue of cell surface glycoproteins and glycolipids and is involved in a number of important biological events such as intracellular interactions, adhesion, and aggregation. A stable analog that would be hydrolytically inert was desired, and thus sulfone 109 was exposed to standard samarium iodide conditions (Scheme 23). However, it gave only the net product of reduction 110 [32]. When the same reaction was carried out with pyridyl sulfone 111 in the presence of the shown cyclic ketone, a 90% yield of 112 was obtained. Several other carbonyl compounds were also used as electrophiles with the corresponding pyridyl sulfone and gave good yields of coupled products. However, the reaction did not proceed with hindered carbonyl compounds [32]. Matsuda and coworkers have found that if such 2-keto-thioglycosides are treated with samarium iodide in THF followed by quenching with a carbonyl compound, good yields of C-glycosides are obtained, with the ␣ isomer (114) predominating in a ratio of 9:1 (Scheme 24) [33]. Kocienski et al. showed that the ␣-glycosyl copper(I) compound 115 reacts with the molybdenum complex 116 to give an excellent yield of the allylic ␣-Cglycoside 117 (Scheme 25). The corresponding ␤-glycosyl copper(I) reagent gave a

Scheme 23

C-Glycoside Synthesis

89

Scheme 24

Scheme 25 49% yield of the corresponding ␤-C-glycoside. When the mirror image of the cationic complex was used, the stereochemistry of the newly formed stereogenic center was found to be reversed [34]. IV.

APPROACHES BASED ON WITTIG CHEMISTRY

A.

Introduction

The reaction of sugar lactols with Wittig-type reagents and cyclization to C-glycosides is a traditional method for preparing C-glycosides. This section addresses the reactions of sugar lactones, the Wittig reaction of sugar lactols with unstabilized ylides and their subsequent cyclizations, and reactions of lactols with stabilized ylides and their conversion into C-glycosides. B.

Reaction of Sugar Lactones

Chapleur and coworkers have published a full account on their work dealing with the dichloroolefination of sugar lactones [35]. They found that sugar-based lactones give good yields of dichloroolefins when treated with a carbon tetrachloride-triphenylphosphine reagent combination. Some of the products along with their yields are boxed in Scheme 26. The same group has also found that reaction of certain chloroolefins such as 121 with mCPBA in dichloromethane gives compounds of the general structure 122 [36]. The reaction is thought to proceed via formation from dichloromethane of chlorine radicals that do not undergo addition to chloroolefins, since they are poor radical acceptors, but instead combine to give molecular chlorine that then undergo ionic addition to the double bond to give the observed products.

90

Scheme 26

Postema and Calimente

C-Glycoside Synthesis

91

Scheme 27

The same group also courageously discovered that the reaction of sugar lactones with the stabilized ylide 125 under high-temperature sealed-tube conditions gives good yields of the corresponding Wittig products (Scheme 27)! No mechanistic explanation is offered, but the sealed-tube conditions are crucial for success. No reaction occurs if the products are heated under reflux at similar temperatures under atmospheric pressure [37]. When the same reaction was carried out on sugar lactones that had a judiciously placed free hydroxyl, cyclized structures were obtained. For example, reaction of 126 with 125 gave a 1:1 mixture of 127 and 128. Compound 127 results from Michael addition of the free hydroxyl addition onto the ␣,␤-unsaturated ester function (Scheme 27) [38]. C.

Wittig Reaction with Unstabilized Ylides Followed by Cyclization

The Wittig reaction of a sugar lactol with an unstabilized ylide to give a hydroxy olefin poised to undergo electrophilic-induced cyclization is a common method for the preparation of C-glycosides. Martin et al. converted 129 to olefin 130 and followed this by inversion and introduction of the nitrogen substituent to give 131. Mercury-induced cyclization was followed by treatment with iodine to give compound 132. Presumably the benzyl carbamate participates to form a new C—O bond during the treatment of the organomercurial intermediate with iodine. Compound 132 could also be debenzylated and treated with base go give 133, the ␣ homoanalog of galactostatin. Alternatively, 132 was transformed into 134, and exposure to potassium carbonate then gave 135, the 1,N-anhydro derivative, a potential inhibitor of both ␣- and ␤-galactosidases, (Scheme 28) [39]. Work from the same group showed that olefin 137, available via Wittig reaction of 136, could be reductively aminated to give a mixture of amines 138 and 139 in a 5:2 ratio (Scheme 29). NIS promoted cyclization of 138, and 139 then gave 140 and 141, respectively, in good yield. Similar chemistry was carried out in the galacto series [40].

92

Postema and Calimente

Scheme 28

Nicotra et al. observed that 144 or 145, obtained by mercury-induced cyclization of 142 and 143, could not be converted to the desired iodo compound 146. The workers had to resort to introducing the 2-deoxynitrogen substituent after the phosphonate had been introduced as shown in Scheme 30. The ketone was converted to an oxime and with diborane gave the gluco isomer 149 in 64% diastereomeric excess [41]. Schmidt et al. also used electrophilic cyclization to gain access to C-glycosides, in this instance to produce 2-deoxy C-aryl glycosides. Known aldehyde 150 was olefinated and gave a mixture of isomeric olefins 151, which were converted to 152 and 153 and separated. Compounds 153a–c and 152c were exposed to NIS in acetonitrile and gave good yields of the ␤ isomers exclusively. Protecting group manipulation and removal of the iodine atom then gave the ␤-C-aryl glycosides 155a–c (Scheme 31) [42]. Work by Tius has shown that the preference for formation of the axial organomercurial is not due solely to coordination of the incoming mercurio species to the C2 oxygen substituent. Oxymercuration of 156, a compound in which there is no C2 substituent, gave a 60:40 ratio of 157 and 158, respectively. Interestingly enough, cyclization of the TBS derivative 159 gave 160 as the exclusive product (Scheme 32). Compound 160 was transformed into the known triacetate 162 by standard methods. The axial OTBS groups and equatorial methyl group are preferred because van der Waals repulsions that occur between the OTBS groups when they are equatorial are of greater energy than the 1,3-diaxial interactions that occur when the groups are disposed axially [43]. Hydroxy olefin 164 was epoxidized with mCPBA to give a mixture of Cglycosides 165 and 166 in the yields shown in Scheme 33. Compound 165 then served as a building block for C-nucleoside synthesis [44]. Two groups reported a novel cyclization reaction when a suitably protected sugar derivative was treated with triflic anhydride in the presence of pyridine. Thus, exposure of 136 to the aforementioned conditions gave an excellent yield of the vinyl C-furanoside 169. Several other sugars were examined, and all gave good yields of the corresponding furanosides. The reaction proceeds via triflation of the free hydroxyl group to give 167, and displacement with the C3 benzyloxy group provided 168, which is then debenzylated to deliver 169 [45,46]. Sharma and coworkers found that exposure of

C-Glycoside Synthesis

Scheme 29

93

94

Postema and Calimente

Scheme 30

alcohol olefins such as 170 to Wacker oxidation conditions gave a good yield of the corresponding keto-furanoside 171 and the C-glycoside 172 in a 4:1 ratio (Scheme 33) [47]. D.

Wittig Reaction with Stabilized Ylides Followed by Cyclization

1.

With Unprotected or Partially Protected Sugars

The situation with stabilized ylides is rather different from that with unstabilized ones because the formed product has the potential to undergo Michael addition to the corresponding C-glycoside. It is not clear what factors control the question of cyclization selectivity, but it is certain that reaction conditions and substrate structure both play an important role for the reaction in question. It has been shown that free sugars can be converted to the free olefins when exposed to stabilized ylides (173 → 174, Scheme 34) [48]. This reaction was later applied by others [49] to the preparation of open-chain sugar ␣,␤-unsaturated ester derivatives. As shown in the second equation of Scheme 34, compound 175 was then efficiently converted to 177, a compound that has been previously converted to KDO. Demailly and coworkers have shown that when Reformatsky ‘‘type’’ conditions are applied to this mode of bond formation, a preponderance of open-chain com-

Scheme 31

C-Glycoside Synthesis

Scheme 32

Scheme 33

Scheme 34

95

96

Postema and Calimente

Scheme 35

pounds such as 179 can be isolated. These results are contrasted to those obtained when the same conditions were applied to the protected sugar derivatives, in which chelation of the lone free hydroxyl with the ester carbonyl causes a template effect favoring stereoselective cyclization to the ␤-C-glycoside as shown by 181. In the unprotected case, the formed zinc bromide presumably ties up two vicinal hydroxyl groups, thereby disfavoring cyclization to the C-glycoside (Scheme 35) [50]. 2.

With Protected Sugars

Similar chemistry was applied to the protected glucosamine derivative 183, and the ␤-C-glycoside compound 184 was produced in 50% yield (Scheme 36) [51]. Murai and coworkers used the Wittig reaction on a protected aldehyde to give the Z-olefin in 78% yield along with 15% of the E-olefin after acid-catalyzed deprotection of the acetal. Exposure to base then gave the C-glycoside derivative 187, along with some dimer in 92% combined yield (Scheme 37) [52]. The thioamidephosphonate was reacted with the manno derivative 188 and gave a good yield of C-glycoside 189 (␣ :␤ ratio of 21:78). Compound 189 was then converted into the glycine-C-glycosyl compound 190 (Scheme 37) [53]. Davis and coworkers carried out a Wittig–cyclization sequence with the partially protected 2-deoxyamino sugar 191 using an amino acid based phosphonate. Reaction of 191 with the cesium enolate 192 gave a 53% yield of epimers 193 and 194 (1:1 ratio) in which epimerization of the 2-amino group had taken place. The manno isomer 194 was then epimerized to the more stable gluco 193 derivative by simple treatment with t-BuOLi in methanol (Scheme 38) [54].

Scheme 36

C-Glycoside Synthesis

97

Scheme 37

Scheme 38

V.

TRANSITION METAL MEDIATED APPROACHES TO C-GLYCOSIDES

A.

Introduction

Several years ago transition metal mediated reactions in the area of C-glycoside synthesis were primarily limited to palladium and to a lesser extent nickel and manganese. Over the last few years several other metals, including chromium, molybdenum, tungsten, cobalt, and rhodium, have been utilized in C-glycoside synthesis. This section discusses the chemistry of palladium, which is divided into Stille-type couplings and ␲-allyl complexes. This is followed by considerations of the chemistry of chromium and the above-listed metals. A review by Frappa and Sinou entitled ‘‘Transition Metal Catalysed Functionalization at the Anomeric Center of Carbohydrates’’ appeared in early 1997 [55]. B.

Use of Palladium

1.

Cross-Coupling Reactions

Falck et al. used the anomeric stannane 195 under palladium/copper-mediated coupling conditions with phenyl chlorothionoformate and ethyl chlorothiolformate to deliver 196 and 197, respectively. Coupling with benzoyl chloride proved sluggish and gave only low yields of the desired C-glycosides [56]. Work from the same group relied on a similar reaction as the key step in the short synthesis of the cytotoxic natural product (⫹)-goniofufurone (202). Compound 199 (available in one

98

Postema and Calimente

step from commercial D-glucurono-6,3-lactone) was converted to 200 in eight steps, and exposure to the proper conditions gave a 60–65% yield of product 201. Selective reduction and deprotection completed the sequence to provide natural 202 (Scheme 39) [57]. Snieckus has used his lithiation methodology in tandem with Suzuki–Miyama cross-couplings to produce C-glycoside-like structures. Vinyl iodide 204 was coupled with a variety of aryl boronic acids to give the corresponding aryl derivatives 205– 208 in excellent yield [58]. In work directed at synthesizing the core of the papulacandin natural products, Brimble and Chan carried out a Suzuki coupling of the boronic acid 211 with the iodo glycal 212 and obtained a 25% yield of the desired product 213, along with large amounts of the product of protiodeboronation. The use of the Stille reaction (both with and without copper as cocatalyst) for the coupling of 210 and 212 gave only 11% of the same desired product 213 (Scheme 40) [59]. Work from the same group addressing a Stille approach to C-glycoside derivatives of kalafungin has been published [60]. Lowary, Bock, and coworkers have prepared carbon-linked glycopeptides, such as 216, via Sonogashira coupling of glycosyl acetylene 214 and aromatic iodide 215 (Scheme 41) [61]. 2.

␲-Allyl Complexes

Sinou and coworkers published a full account of their work on the synthesis of 2,3unsaturated aryl C-glycosides via ␲-allyl-type methodology [62]. They showed that the reaction of aryl Grignard reagents with 217␣ under transition metal catalysis leads to either the ␣- or ␤-C-glycoside in good yield. When palladium is used as the metal, ␲-allyl complex formation is followed by intermolecular attack of the Grignard reagent from the bottom ␣ face, leading to 220, the product of net retention. A complementary approach involves the use of nickel catalysis to give the product of net inversion. In this case transmetallation of the ␲-allyl intermediate with the Grignard reagent gives 221 and reductive syn elimination then delivers 222. As far as catalyst choice is concerned, the catalysts PdCl2(dppf ) and NiCl2(dppe) gave the best results. Numerous Grignard reagents were also examined as part of this work (Scheme 42). Daves has continued his work directed at using palladium-mediated processes to achieve C-nucleoside synthesis. Coupling of iodide 223 with glycal 224 gave a good yield of adduct 225, which was subsequently desilylated and oxidized to furnish 226. The coupling of several other iodides and glycals was also examined [63]. A Spanish group recently employed ␲-allyl chemistry to prepare a potentially useful anti-HIV agent. Olefin 227 was treated with two of 1,3-diketones (R = Me, Ph) to give the diketo-2,3-unsaturated C-glycosides 228. Condensation with a variety of hydrazine derivatives then gave access to a small library of 2⬘,3⬘-unsaturated pyrazole C-glycosides 229a–j (Scheme 43) [64]. Exposure of 230 and 231 to diethyl allylmalonate in the presence of sodium hydride and tetrakis(triphenylphosphine)palladium in tetrahydrofuran gave C-glycosides 232 and 233, respectively, in good yield (Scheme 44). Metallo-ene cyclization of both 232 and 233 gave the same cis product of cyclization 234 in 83–89% yield [65]. Genet has used ␲-allyl chemistry to extend the chain of a preexisting C-glycoside. The example shown in Scheme 44 is illustrative. Exposure of 235 (available from the corresponding ␤-glycosyl aldehyde by condensation with vinyl magnesium

C-Glycoside Synthesis

Scheme 39

99

100

Postema and Calimente

Scheme 40

Scheme 41

Scheme 42

bromide followed by acylation) to Pd(dppe)2 and sodio malonate gave the chainextended C-glycoside derivative 236 in 71% yield. Several other examples were studied and gave comparable yields of C-glycoside products [66]. C.

Use of Cobalt

Luengo and Gleason utilized Murai’s cobalt chemistry to produce the equatorial Cglycoside 238 in 85% yield, followed by desilylation and conversion of 239 to 241. The final step, standard coupling, then gave 244 the potential GDP-fucose inhibitor (Scheme 45). Several other analogs were also prepared [67].

C-Glycoside Synthesis

Scheme 43

Scheme 44

Scheme 45

101

102

Postema and Calimente

Scheme 46

Photolysis of the anomeric cobaloxime 245 in the presence of maleic anhydride and PhSSPh gave 246 in 46% yield. The reaction proceeds via radical intermediates and the formed ␣-carbonyl radical is trapped with PhSSPh to give the formed adducts. Then syn-␤-H elimination led to 247 in quantitative yield (Scheme 46) [68]. D.

Use of Rhodium

McDonald et al. used a rhodium-catalyzed cyclotrimerization reaction to gain access to aryl C-glycosides (Scheme 47). Their work began with the conversion of lactone 248 to alkynyl glycal 249 by standard methods. Exposure of 249 or 251 to 254 in the presence of ClRh(Ph3P)3 in protic solvent furnished the product glycals 252 and 253, respectively, in the indicated yields [69]. Dialkyne 255 was also subjected to similar conditions and gave C-aryl spiroglycoside 257, a compound structurally related to the papulacandin natural products. E.

Chromium, Tungsten, and Molybdenum

Do¨tz and coworkers prepared several interesting and novel glycosylidene carbenes (259–261 ) by reaction of lithiated glycal 258 with the appropriate metal carbonyl derivative. The synthetic utility of the formed carbenes was demonstrated by reaction with 3-hexyne to give a mixture of complexed and uncomplexed adducts 262 and 263, respectively. The anomeric chromium carbene 264 was converted to 265 by exposure to ethoxy ethyne (Scheme 48) [70]. The same workers have carried out a similar chromium-mediated benzannulation (266 → 268), this time with the chromium on the aromatic fragment [71]. Other sugar-based carbenes have also been prepared [72]. Postema and coworkers have used ring-closing metathesis to prepare C1-glycals, compounds convertible to C-glycosides (Scheme 49). The sequence is both general and convergent. Ester formation (269 → 270) is followed by methylenation to give 271, and ring-closing metathesis is then carried out with the Schrock catalyst 272 to deliver the generic C1-glycal 273. The reaction tolerates alkyl and aryl groups at the anomeric center. Very hindered groups (R = t-Bu) were not good substrates because the methylenation step did not proceed in good yield. This was presumably due to the steric hindrance [73]. The chemistry has been extended to the preparation of a number 1,6-linked C-disaccharides. The conversion of 274 to 277 exemplifies the strength and versatility of this approach to C-glycoside synthesis. Compound 277 was converted to the 2-deoxy ␤-C-disaccharide and also to the ␤-C-disaccharide by standard methods (not shown). A small library of glycal C-disaccharides was prepared by use of this novel methodology [74].

C-Glycoside Synthesis

Scheme 47

103

104

Postema and Calimente

Scheme 48

Scheme 49

VI.

SIGMATROPIC APPROACHES

A.

Introduction

This section focuses on the use of concerted processes in the preparation of Cglycosides. The topics covered are cycloaddition reactions and Wittig rearrangements as approaches to C-glycoside synthesis. B.

Cycloaddition Approaches

1.

Diels–Alder

Several types of Diels–Alder reactions have been applied to the preparation of Cglycosides. Some of these include approaches to prepare the sugar ring or to append

C-Glycoside Synthesis

105

Scheme 50

a ring on an existing C-glycosyl derivative. Dipolar cycloadditions have also been utilized. (a) From Sugar Dienes or Glycals. Lubineau et al. reported an aqueous Diels– Alder approach to the trehalo derivative 281. Cycloaddition of 279 in water at 140⬚C gave a mixture of four isomers, of which the major compound, 280a, was separated, debenzoylated, and reduced to give 281. Compound 281 showed only modest activity (IC50 = 47 mM) as an inhibitor or porcine kidney trehalase (Km = 3 mM) (Scheme 50) [75]. In attempting to prepare sugar-based radical acceptors, Martin and Xie found that treatment of 282 with base to generate enone 283 gave instead the dimerized product 284; the result of a [4⫹2] cycloaddition reaction. Compound 283 was never isolated from the reaction mixture (Scheme 51) [76]. Quayle and coworkers used a Stille coupling approach to prepare a number of sugar-based dienes that then under-

Scheme 51

106

Postema and Calimente

Scheme 52

went a facile cycloaddition reaction with Cookson’s reagent 287 to give good yields of the cycloadducts. The reaction shown in Scheme 51 is illustrative [77]. Work in this area has also been carried out with noncyclic carbohydrates. Reaction of the nitro sugar 289 with cyclopentanedienone gave a mixture of 291 and 290 in a 1.9:1 ratio (Scheme 52) [78]. Dondoni used his masked formyl derivative in a hetero-Diels–Alder approach to carbon-linked disaccharides. Cycloaddition of 293 with ethyl vinyl ether gave a mixture of isomers; the major compound, 294, is shown in Scheme 52. The cycloadduct was then converted to 295 via standard methods [79]. (b) De Novo Ring Formation. Varelis and Johnson utilized a hetero-Diels–Alder reaction to prepare the pyran ring system 300. The olefin diastereomers were separated, and reduction followed by acetylation gave 298. Ireland ester rearrangement of 299 proceeded with syn selectivity to deliver the 1,6-dialkyl C-glycoside-like structure 300 (Scheme 53) [80]. Hanna and coworkers used a [2⫹2] cycloaddition approach for the preparation of precursor en route to forskolin. Dichloroketene cycloaddition with glycal 301 was followed by reductive dechlorination to give 302 in 80% overall yield. Several steps were required to convert 302 into the forskolin (304) Diels–Alder precursor 303 (Scheme 54) [81]. (c) Other Cycloadditions. Holzapfel and van der Merwe utilized a palladium-mediated [3⫹2] cycloaddition of nitroglycal 305 to deliver the C-glycosides 306 and 307 in a 1:2.4 ratio (Scheme 55) [82]. The 1,3-dipolar nitrone cycloaddition has also been employed for the preparation of pseudo-aza-C-disaccharides from glycals. Ex-

C-Glycoside Synthesis

Scheme 53

Scheme 54

107

Scheme 55

108

Postema and Calimente

posure of nitrone 308 to 3 equiv of glucal 309 in toluene at 100⬚C for 3 days gave a 61% yield of the adduct 310. The protecting groups were removed by standard methods to yield an intermediate polyol and cleavage of the N—O bond then delivered to the target compound 311 (48% over four steps; Scheme 55) [83]. 2.

Wittig Rearrangements

In 1997 Nakai and coworkers wrote a mini-review on the general area of subject of Wittig rearrangements [84]. Scheme 56 shows an application of the Wittig rearrangement (312 → 313) with a benzyl ether used as the starting material. Scheme 56 also illustrates the use of the propargylic substrates 314 and 317. The rearrangement proceeds with retention of configuration. This high selectivity is attributed to the tightly bound lithium-coordinated transition state, as in 315 [85]. Nakai has also employed the Wittig rearrangement in the preparation of the zaragozic acid core (319 → 320, Scheme 57). This work disclosed a violation of the principle of retention of configuration: when 321 was treated with butyllithium, a product with a net inversion of configuration at the anomeric center 322 was obtained. The authors suggest that steric hindrance of the groups on the ring caused radical recombination to occur from the less hindered bottom ␣ face (Scheme 57) [86]. VII. A.

CATIONIC APPROACHES Introduction

This section, the largest of the entire chapter in terms of number of citations, relies on the natural mode of reactivity at the anomeric center. Because of the large number of references, only an overview is presented here. Relevant references are included

Scheme 56

C-Glycoside Synthesis

109

Scheme 57

whenever possible. The section is divided into the formation of C-glycosides from reaction of glycals and 1,2-anhydro sugars, heteroatomic glycosyl derivatives, and sugar lactones with various nucleophiles. Section VII.E deals with the preparation of aryl C-glycosides. B.

Use of Glycals

1.

Reaction Under Lewis Acid Catalysis

The reaction of glycals with allylsilanes, enol ethers, or silyl acetylenes has received considerable attention over recent years. In 1996 Toshima et al. employed environmentally friendly montmorillonite K-10 (clay) to catalyze the addition of allylsilanes to sugar glycals for the purpose of obtaining allyl C-glycosides. Reaction of the galactal derivative 323 as shown in Scheme 58 gave a good yield of the ␣- and ␤allyl C-glycosides, with the 324 as the major isomer [87]. Work from the same group has also shown that the glycal does not have to be protected for the allylation to work, at least with conventional Lewis acid catalysts. This is exemplified by the conversion of 325 to 326 (Scheme 58) [88].

Scheme 58

110

Postema and Calimente

Scheme 59

Panek and Schaus continued their work with crotyl silanes and their addition to glycals (319 → 320) as shown in Scheme 59 [89]. Grieco et al. utilized [LiCo(B9C2H11)2] as a catalyst for the condensation of an allylic acetate with glycal 331 to give the C-glycosides 332. The use of this catalyst is interesting because it mimics the effect of highly polar media (5 M LiClO4 in Et2O) for reactions of these types [90]. Portella et al. showed that the acylsilanes and the glycal 309, in the presence of CF3SiMe3 and TBAF, react to give the C-glycosides 334 as a mixture of anomers [91]. Hayashi et al. cyanated (335 → 336) unprotected glycals in the presence of a catalytic amount of palladium(II) acetate (Scheme 59) [92]. Csuk et al. also examined some reactions of various glycals with a hindered silyl enol ether under Lewis acid catalyzed conditions (not shown) [93]. Isobe has continued his explorations of the reactivity of silyl acetylenes with glycals. The sugar-based acetylene 338 is condensed with glycal 337 to give the disaccharide 339 [94]. This area was reviewed in 1998 [95]. Steel et al. used actinium perchlorate to catalyze the dimerization of glucal. Three products were isolated in the yields shown. Separate treatment of 342 with boron trifuluoride etherate and triethylsilane then gave 343 in good yield (Scheme 60) [96]. 2.

With ‘‘Organometallic’’ Nucleophiles

Dorgan and Jackson [97] extended the work of Thorn and Gallagher [98], reaction of organozinc with glycals to give the corresponding C1 enitols (309 → 344), to the synthesis of C-linked glycosyl amino acids. This is exemplified by the conversion of 309 to 347 (Scheme 61). It has been reported that the glycal 348 undergoes cyclopropanation to give 349 or 350 as the major products, depending on the choice of reagent [99]. Treatment of 348 with ethyl diazoacetate in the presence of rhodium diacetate gave a 59% yield

C-Glycoside Synthesis

Scheme 60

111

112

Postema and Calimente

Scheme 61

of the ester cyclopropane 351 [100]. Hoberg and Claffey showed that treatment of the related glycal 352 under similar conditions gave 353 as the major product in 87% yield, with only trace amounts of the other three possible isomers (Scheme 62) [101]. The formed cyclopropanes have proved to be useful in several cases. For instance, 350 has been ring-expanded to the bicyclo olefin 354 by exposure to TMSOTf [102]. Compound 355 was also found to undergo ring opening to the dihalide 356, albeit in low yield [101]. In addition, 351 was converted to the Fuc-GDP analog 357, a potential transition state based inhibitor (Scheme 63) [103]. Van Boom and Sinay¨ have collaborated to show that the reaction of acetylides with 1,2-anhydro sugars under the influence of zinc chloride catalysis gives the ␣ isomer as the major product. Accordingly, reaction of 358 and 359 under the aforementioned conditions gave 360 in good yield. Standard protecting group manipulations then gave the C-disaccharide 361. The reaction is postulated to proceed via intermediate C [104]. Evans and coworkers have found that allylstannanes serve as good nucleophiles for oxirane opening of the 1,2-anhydro sugars under Lewis acid catalysis. Treatment of 358 with either 362 or 363 in the presence of tributyltin triflate gave good yields of the corresponding ␤-allyl-C-glycosides 364 and 365 (Scheme 64) [105].

Scheme 62

C-Glycoside Synthesis

113

Scheme 63

Scheme 64

Gervay et al. reported reduction of the diastereomeric epoxides, giving regioisomeric mixtures of alcohols (369 and 370) (Scheme 65) [106]. Related work (not shown) has been reported by Panza et al. [107]. C.

Use of Lactones with Organometallic-Based Nucleophiles

Dondoni and Scherrmann published a full account of their thiazole-based approach to formyl C-glycosides. Addition of the thiazololithio species to lactone 371 gave

114

Postema and Calimente

Scheme 65 the ␣-C-glycoside 372 as the major product. Reduction under standard conditions followed reduction–epimerization, and demasking of the protected aldehyde group then gave 373 (Scheme 66) [108]. Geneˆt et al. used a similar protocol to access ␤C-glycosyl aldehydes. Both the dithiane group (375 → 377) and an acetylene (378 → 377) served well as masked aldehyde equivalents (Scheme 66) [109]. Heterocyclic nucleophiles have also been utilized with pyranose sugars to prepare potential glucosidase inhibitors. In the case shown in Scheme 67, Lewis acid mediated silane reduction did not work owing to chelation of the Lewis acid with the imidazole group, and a slightly longer route was required to effect the reduction [110]. Epoxide alkyl ␤-C-glycosides, potential inhibitors of ␤-glucan hydrolases, have also been prepared from sugar lactones (374 → 377) [111]. The reaction of 2-deoxy-2-azido sugar lactones seems to parallel the reactivity of the corresponding 2-oxygenated lactones as shown for the conversion of 389 to 392 (Scheme 68) [112]. Martin and Saavedra utilized a conceptually related approach, the reduction of hemiacetals to give ␤-alkyl products, in the preparation of aza sugar analogs. Addition of (methoxymethoxy)methyllithium to lactone 374 gave 394, and this was followed by conversion to diketone 395. Reductive amination then furnished the protected C-aza sugar 396 (Scheme 68) [113]. Wightman et al. showed that in furanose sugars, reduction of C-glycoside lactols gives 1,2-anti products predominantly. It seems that silane reduction occurs preferentially from the same face as the vicinal oxygen substituent (Scheme 69) [114]. The reaction of enolates with lactones has also been carried out to give chainextended sugars. Three approaches have been published, two on samarium diiodide mediated chemistry [115,116] and one on classical Claisen-type chemistry [117]. Representative examples are shown in Scheme 70. D.

Use of Glycosyl Halides, S-Glycosides, Lactols, and O-Glycosides

1.

Glycosyl Halides

Glycosyl halides are popular electrophilic coupling partners for C-glycoside synthesis. The area of ␣-monohalo ethers in synthesis has been reviewed [118]. Heteroatoms at the 2-position have been well tolerated, as can be seen from the examples shown in Scheme 71. Alkylation of 407 with the potassium salt of diethylmalonate

C-Glycoside Synthesis

Scheme 66

115

116

Scheme 67

Scheme 68

Scheme 69

Postema and Calimente

C-Glycoside Synthesis

117

Scheme 70

Scheme 71

gave the ␤-C-glycoside 408 after decarboxylation [119]. Under the influence of silver ion, the displacement of the anomeric bromide in 410 with the tin acetylide shown, gave the the ␣ product 411 exclusively. Here, the stereochemical outcome is rationalized on the basis of attack of the formed oxonium from the ␣ face under the influence of the anomeric effect [120]. Similar chemistry has also been carried out

118

Postema and Calimente

Scheme 72

on the 2-oxygenated compounds. Aza sugars have also been found to be reactive toward electrophilic C-glycosidation upon conversion to the anomeric fluorides. Several nucleophiles were examined, and Scheme 71 shows the allylation and reaction with silyl enol ethers of 413 [121]. Veyrie`res et al. found that the allylsilane 417 reacts with a variety of pyranose derivatives to give good yields of allylated and functionalized C-glycosides (Scheme 72) [122]. 2-Thioglycosyl chlorides, derived from ionic addition of RSCl to glycals, have been found to react with Grignard reagents to give the ␤-gluco C-glycosides 426. The same chlorides also react with silyl enol ethers and allyltrimethylsilane to give the corresponding C-glycosides, 425 (Scheme 73) [123,124]. Intramolecular C-glycosidations have also been explored. For example, exposure of 427 to AgBF4 with strict exclusion of water gave a 74% yield of 428. The benzylic positions were oxidized to furnish 429, and ester cleavage then gave the free C-glycoside 430 (Scheme 74) [125]. Lo´pez and Go´mez found that treatment of pyranosyl chlorides with a suitable organolithium gave acceptable yields of the corresponding C1-glycals 432–434 (Scheme 75). No mechanistic explanation was offered. One possibility could be elimination of the benzyloxy group to generate the C1-chloro glycal, followed by addition of the organometallic to the double bond at C1 and subsequent elimination of chloride [126].

Scheme 73

C-Glycoside Synthesis

119

Scheme 74

Scheme 75

2.

Thioglycosides

Thioglycosides have received some attention over the last few years. The ethyl thioglycoside 435 has been found to be a good substrate for the selective formation of the ␣-glycosyl cyanide 436 (Scheme 76) [127]. Craig has continued his work on template directed C-glycosidations. Work directed at the C-glycosidation of 437, shown in Scheme 77 [128], is highlighted by the preparation of 2,3-dideoxy-D-manno-2-octulopyranosonic acid (444). Accordingly, thioglycoside 437 was exposed to AgOTf and gave the cis-fused C-glycoside 438 in 64% yield. Oxidative cleavage followed by reduction led to 442, which was desulfonylated to give the fragmentation product 443. Standard manipulations then afforded the target acid 444 [129]. 3.

Anomeric Acetates and Imidates

The formate group has been used at the anomeric center to install acetylenic moieties (445 → 446, Scheme 78) [130], while an acetate function served admirably as the activating function on a tetrahydropyran ring to install an allyl group in the synthesis of the F-ring 448 of the spongistatins. The TBS groups were first replaced with acetates to ensure ␤-C-glycosidation [131]. C-Glycosidation of furanosyl acetates is still a popular method for introduction of carbon-based groups at the anomeric center. Scheme 79 shows some recent examples. The entries highlight the use of a diethylthiocarbamoyl function as a partic-

Scheme 76

120

Postema and Calimente

Scheme 77

ipating group to control the facial selectivity of C-glycosidation [132]. This is significant, since the lack of a group at C2 hinders stereocontrol in the 2-deoxyribofuranosides [133]. It has also been found that scandium perchlorate serves as a good catalyst for the selective ␣-C-glycosidation of perbenzylated ribofuranosyl acetates [134]. Heteroatom-based siloxy dienes have received attention in recent years as an efficient method for building up ring frameworks. The subject has been reviewed, and only a few applications are illustrated here [135]. This is a powerful process, since in a relatively small number of steps several rings can be joined together. The first two equations [136,137] of Scheme 80 show how a single butenolide unit can be attached, while the last sequence illustrates an iterative methodology [138]. It should be noted that in the addition of trimethylsiloxy furan to the anomeric acetate,

Scheme 78

C-Glycoside Synthesis

121

Scheme 79

a mixture of four compounds is possible. Usually the erythro product 463 was obtained in a 6:4 ratio. The two examples in Scheme 81 illustrate the use of a cyclic enol ether reacting with anomeric acetate 467 to give 468 [139], as well as the reaction of an olefin with the same acetate to give the novel ring-fused product 469 [140]. The final example demonstrates the use of amino acid derived enol ether 470 reacting with a glycosyl imidate 469 under Lewis acid catalysis to give a low yield of the carbon-linked threonine isostere 471 (Scheme 82) [141].

Scheme 80

122

Postema and Calimente

Scheme 81

Scheme 82

4.

Lactols

There are two reports of the use of lactols as precursors to C-glycosides. Wang and Gross found that when triol 472 was treated with nitroethane under the outlined conditions, a 50% yield of the C-glycoside 473 was obtained as the major product [142]. D-Glucose (474) was found to undergo condensation with the barbituric acid derivative 475 to afford 476. Alkylation then gave the C-glycosyl derivative 477 (Scheme 83) [143]. 5.

Alkyl Glycosides

During their synthesis of the F-ring (480) of halichondrin B, Yonemitsu and coworkers relied on allylation of the methyl glycoside derivative 478 to deliver 479, which was converted to the target (Scheme 84) [144]. Johnson has carried out a

Scheme 83

C-Glycoside Synthesis

123

Scheme 84

highly selective allylation on the nojirimycin analog 481 to provide 482, which was then converted to the indolizine analog 438 (Scheme 84) [145]. Schmidt converted the TCP-protected glucosamine derivative to 484 via a cationic allylation reaction. Compound 485 was then converted to the amino acid analog 486. The ␤-selective allylation is noteworthy and is presumably due to the steric bulk of the phthalidebased amino protecting group, (Scheme 84) [146]. Ley et al. found that exposure 487 to stannic chloride furnished a good yield of the disaccharide derivative 488 as a 1:1 mixture of isomers (Scheme 84) [147]. Two groups have prepared C-glycoside analogs of the natural product (⫹)hydantocidin via Lewis acid mediated C-glycosidation of the acetal derivative 489 [148]. Sano and coworkers succeeded in converting 490 to the succinimide analog 493 (Scheme 85) [149]. A synthesis of the spirohydrouracil nucleoside analog of (⫹)-hydantocidin has also appeared [150]. E.

Aryl C-Glycosides

1.

Lewis Acid Monoarylation

Several different catalyst systems, anomeric activators, and applications of Lewis acid catalyzed reactions for aryl C-glycoside synthesis have been published over recent years and a few are presented in Scheme 86. Inazu et al. used the dimethylphosphinothioate groups [151], while Satoh et al. introduced the use of AgOTfa/ SnCl4 as a novel catalyst system [152]. Schmidt et al. continued the use of the trichloroacetimidate group, this time directed at the preparation of bis(C-glycosyl) flavanoid precursors [153]. Sato’s group has also used glycosyl fluorides to prepare precursors for C-glycosyl flavanoids (Scheme 86) [154]. Mallareddy and Rao prepared analogs of monocerin via intramolecular aryl C-glycosidation, (Scheme 86)

124

Scheme 85

Scheme 86

Postema and Calimente

C-Glycoside Synthesis

125

Scheme 87

[155]. Andrews and Larsen also carried out Lewis acid catalyzed glycosidations of naphthalene derivatives with anomeric acetates to gain access to the ring systems of the Angucycline antibiotics [156]. Suzuki has also published an account of his work on novel Lewis acid catalysts in organic synthesis [157]. Toshima et al. found that unprotected 2-deoxy sugars, as either reducing sugars or methyl glycosides, will undergo C-glycosidation to give aryl C-glycosides [158]. The reaction has been catalyzed by TMSOTf, and it has also been found that montmorillonite K-10, used in excess, also works with unprotected sugars (Scheme 87) [159]. Work by Park et al. deals with regioselective acylation of the 2⬘-hydroxyl group on a free aloesin glucopyranoside (not shown) [160]. The normal course for these O → C-glycoside rearrangement is to deliver the ␤ or equatorial product preferentially. Suzuki et al. found that if O4 and O3 are protected by extremely bulky groups, such as t-butyldiphenylsilyl, the ␣ anomer is the major product of the reaction. This result is contrasted to the case of the use of smaller groups, such as TBS groups, which produces ony the ␤ isomer (Scheme 88). So the stereochemistry of the glycosidation can be controlled by judicious choice of protecting groups [161]. 2.

Lewis Acid Mediated Diarylation

A few groups have carried out multiple arylation reactions to give bis-C-glycosyl compounds (Scheme 89). Schmidt et al. used their trichloroacetimidate methodology

Scheme 88

126

Scheme 89

Postema and Calimente

C-Glycoside Synthesis

127

Scheme 90

in a sequential process to deliver 522 [153]. Kumazawa et al. employed anomeric fluorides under boron trifluoride catalysis to give the corresponding bis-C-glycosyl flavanoid precursor 526 [162]. Satoh, using his SnCl4 /AgOTfa catalyst system, was able to introduce a second sugar unit onto an existing aryl C-glycoside (Scheme 90) [163]. VIII.

SUMMARY

This chapter provides an updated overview of recent developments in the synthesis of C-glycoside compounds. The stereoselectivity of the C-glycosidation step has improved over recent years, and milder conditions for C—C bond formation are still being developed. The use of unprotected C-glycosides is another important advance that has recently come to fruition [164]. Also the use of modern methods based on metals seems to be increasing in popularity. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

MHD Postema C-Glycoside Synthesis, 1st ed. Boca Raton, FL: CRC Press, 1995. H Togo, W He, Y Waki, M Yokoyama. Synlett 700–717, 1998. J Martin, GLM Jaramillo, PG Wang. Tetrahedron Lett 39:5927–5930, 1998. H-D Junker, W-D Fessner. Tetrahedron Lett 39:269–272, 1998. Y Kita, K Gotanda, A Sano, M Oka, K Murata, M Suemara, M Matsugi. Tetrahedron Lett 38:8345–8348, 1997. RP Spencer, J Schwartz. J Org Chem 62:4204–4205, 1997. ZJ Witzack, R Chhabra, J Chojnacki. Tetrahedron Lett 38:2215–2218, 1997. BA Roe, CG Boojamra, JL Griggs, CR Bertozzi. J Org Chem 61:6442–6445, 1996. J-P Praly, G-R Chen, J Gola, G Hetzer, C Raphoz. Tetrahedron Lett 38:8185–8188, 1997. F Ponte´n, G Magnusson. J Org Chem 61:7463–7466, 1996. M Harenbrock, A Matzeit, HJ Scha¨fer. Liebigs Ann 55–62, 1996. L Cipolla, L Liguori, F Nicotra, G Torri, E Vismara. J Chem Soc Chem Commun 1253–1254, 1996. TF Herpin, WB Motherwell, MJ Tozer. Tetrahedron Asymm 5:2269–2282, 1994. JC Lo´pez, AM Go´mez, B Fraser-Reid. J Org Chem 60:3871–3878, 1995. T Skrydstrup, D Maze´as, M Elmouchir, G Doisneau, C Riche, A Chiaroni, JM Beau. Chem Eur J 3:1342–1356, 1997. S-C Hung, C-H Wong. Angew Chem Int Ed Engl 35:2671–2674, 1996. S Czernecki, E Ayadi, J Xie. Tetrahedron Lett 37:9193–9194, 1996. E Lee, JS Tae, YH Chong, YC Park, M Yun, S Kim. Tetrahedron Lett 35:129–132, 1994. SE Mangholz, A Vasella. Helv Chim Acta 78:1020–1035, 1995.

128

Postema and Calimente

20. 21. 22.

A Kirschning, J Harders. Synlett 772–774, 1996. J-M Beau, T Gallagher. Top Curr Chem 187:1–50, 1997. O Frey, M Hoffmann, V Wittmann, H Kessler, P Uhlmann, A Vasella. Helv Chim Acta 77:2060–2069, 1994. O Frey, M Hoffmann, H Kessler. Angew Chem Int Ed Engl 34:2026–2028, 1995. C Jaramillo, G Corrales, A Fernandez-Mayoralas. Tetrahedron Lett 39:7783–7786, 1998. M Hoffmann, H Kessler. Tetrahedron Lett 35:6067–6070, 1994. F Burkhart, H Kessler. Tetrahedron Lett 39:255–256, 1998. F Burkhart, M Hoffmann, H Kessler. Angew Chem Int Ed Engl 36:1191–1192, 1997. D Maze´as, T Skrydstrup, J-M Beau. Angew Chem Int Ed Engl 34:909–912, 1995. D Urban, T Skrydstrup, J-M Beau. J Org Chem 63:2507–2516, 1998. D Urban, T Skrydstrup, C Riche, A Chiaroni, J-M Beau. J Chem Soc Chem Commun 1883–1884, 1996. D Urban, T Skrydstrup, J-M Beau. J Chem Soc Chem Commun 955–956, 1998. IR Vlahov, PI Vlahova, RJ Lindhart. J Am Chem Soc 119:1480–1481, 1997. S Ichikawa, S Shuto, A Matsuda. Tetrahedron Lett 39:4525–4528, 1998. JA Christopher, PJ Kocienski, MJ Procter. Synlett 425–427, 1998. M Lakhrissi, Y Chapleur. J Org Chem 59:5752–5757, 1994. M Lakhrissi, G Carchon, T Schlama, C Mioskowski, Y Chapleur. Tetrahedron Lett 39: 6453–6456, 1998. M Lakhrissi, Y Chapleur. Angew Chem Int Ed Engl 35:750–752, 1996. M Lakhrissi, C Taillefumier, A Chaouch, C Didierjean, A Aubry, Y Chapleur. Tetrahedron Lett 39:6457–6460, 1998. OR Martin, F Xie, L Liu. Tetrahedron Lett 36:4027–4030, 1995. OR Martin, L Liu, F Yang. Tetrahedron Lett 37:1991–1994, 1996. F Casero, L Cipolla, L Lay, F Nicotra, L Panza, G Russo. J Org Chem 61:3428–3432, 1996. AT Khan, W Ahmed, RR Schmidt. Carbohydr Res 280:277–286, 1996. MA Tius, J Busch-Petersen. Tetrahedron Lett 35:5181–5184, 1994. N Hossain, N Blaton, O Peeters, J Rozenski, PA Herdewijn. Tetrahedron 52:5563– 5578, 1996. B-H Yang, J-Q Jiang, K Ma, H-M Wu. Tetrahedron Lett 36:2831–2834, 1995. OR Martin, F Yang, F Xie. Tetrahedron Lett 36:47–50, 1995. GVM Sharma, AS Chander, K Krishnudu, PR Krishna. Tetrahedron Lett 38:9051– 9054, 1997. T Henk, A Giannis, K Sandhoff. Liebigs Ann Chem 167–168, 1992. CJ Railton, DLJ Clive. Carbohydr Res 281:69–77, 1996. C Lie`vre, C Fre´chou, G Demailly. Tetrahedron Lett 36:6467–6470, 1995. A Mbongo, C Fre´chou, D Beaupe`re, R Uzan, G Demailly. Carbohydr Res 246:361– 370, 1993. K Fujiwara, S Amano, T Oka, A Murai. Chem Lett 2147–2150, 1994. F Sandrinelli, S Le Roy-Gourvennec, S Masson, P Rollin. Tetrahedron Lett 39:2755– 2758, 1998. RM Werner, LM Williams, JT Davis. Tetrahedron Lett 39:9135–9138, 1998. I Frappa, D Sinou. J Carbohydr Chem 16:255–276, 1997. YY Belosludtsev, RK Bhatt, JR Falck. Tetrahedron Lett 36:5881–5882, 1995. J Ye, RK Bhatt, JR Falck. Tetrahedron Lett 1993:8007–8010, 1993. JF Bower, D Guillaneux, T Nguyen, PL Wong, V Snieckus. J Org Chem 63:1514– 1518, 1998. MA Brimble, SH Chan. Aust J Chem 51:235–242, 1998. MA Brimble, MR Nairn, H Prabaharan, NB Walters. Aust J Chem 50:711–718, 1997.

23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60.

C-Glycoside Synthesis 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103.

129

T Lowary, M Meldal, A Helmboldt, A Vasella, K Bock. J Org Chem 63:9657–9668, 1998. C Moineau, V Bolitt, D Sinou. J Org Chem 63:582–591, 1998. H-C Zhang, M Brakta, GD Daves Jr. Nucleosides Nucleotides 14:105–116, 1995. P Borrachero-Moya, F Cabrera-Escribano, M Go´mez-Guille´n, M del Rocı´o ParedesLeo´n. Carbohydr Res 308:181–190, 1998. CW Holzapfel, L Marais. J Chem Res (S) 60–61, 1998. V Michelet, J-P Geneˆt, G Dujardin. Tetrahedron Lett 38:7741–7744, 1997. JI Luengo, JG Gleason. Tetrahedron Lett 33:6911–6914, 1992. RM Slade, BP Branchaud. J Org Chem 63:3544–3549, 1998. FE McDonald, HYH Zhu, CR Holmquist. J Am Chem Soc 117:6605–6606, 1995. KH Do¨tz, R Ehlenz, D Paetsch. Angew Chem Int Ed Engl 36:2376–2378, 1997. D Paetsch, KH Do¨tz. Tetrahedron Lett 40:487–488, 1999. KH Do¨tz, R Ehlenz. Chem Eur J 3:1751–1756, 1997. D Calimente, MHD Postema. J Org Chem 64:1770–1771, 1999. MHD Postema, D Calimente, L Liu, TL Behrmann. J Org Chem 65:6061–6068, 2000. A Lubineau, E Grand, M-C Scherrmann. Carbohydr Res 297:169–174, 1997. OR Martin, F Xie. Carbohydr Res 264:141–146, 1994. A Abas, RL Beddoes, JC Conway, P Quayle, CJ Urch. Synlett 1264–1266, 1995. JL Del Valle, S Marcaccini, T Torroba. Liebigs Ann 1377–1379, 1995. A Dondoni, L Kniezo, M Martinkova. J Chem Soc Chem Commun 1963–1964, 1994. P Varelis, BL Johnson. Aust J Chem 50:43–51, 1997. J Pan, I Hanna, J-Y Lallemand. Bull Chim Soc Fr 131:665–673, 1994. CW Holzapfel, TL van der Merwe. Tetrahedron Lett 37:2307–2310, 1996. F Cardona, S Valenza, A Goti, A Brandi. Tetrahedron Lett 38:8097–8100, 1997. K Tomooka, H Yamamoto, T Nakai. Liebigs Ann Recl 1275–1281, 1997. K Tomooka, H Yamamoto, T Nakai. J Am Chem Soc 118:3317–3318, 1996. K Tomooka, M Kikuchi, K Igawa, P-H Keong, T Nakai. Tetrahedron Lett 40:1917– 1920, 1999. K Toshima, N Miyamoto, G Matsuo, M Nakata, S Matsumura. J Chem Soc Chem Commun 1379–1380, 1996. K Toshima, T Ishizuka, G Matsuo, M Nakata. Tetrahedron Lett 35:5673–5676, 1994. JS Panek, JV Schaus. Tetrahedron 53:10971–10982, 1997. PA Grieco, WJ DuBay, LJ Todd. Tetrahedron Lett 37:8707–8710, 1996. T Brigaud, O Lefebvre, R Plantier-Royon, C Portella. Tetrahedron Lett 37:6115–6116, 1996. M Hayashi, H Kawabata, O Arikita. Tetrahedron Lett 40:1729–1730, 1999. R Csuk, M Schaade, C Krieger. Tetrahedron 52:6397–6408, 1996. S Hosokawa, B Kirschbaum, M Isobe. Tetrahedron Lett 39:1917–1920, 1998. M Isobe, R Nishizawa, S Hosokawa, T Nishikawa. J Chem Soc Chem Commun 2665– 2676, 1998. ALJ Byerley, AM Kenwright, PG Steel. Tetrahedron Lett 37:9093–9096, 1996. BJ Dorgan, RFW Jackson. Synlett 859–861, 1996. SN Thorn, T Gallagher. Synlett 185–187, 1996. R Murali, CV Ramana, M Nagarajan. J Chem Soc Chem Commun 217–218, 1995. CM Timmers, MA Leeuwenburgh, JC Verheijen, GA van der Marel, JH van Boom. Tetrahedron Asymm 7:49–52, 1996. JO Hoberg, DJ Claffey. Tetrahedron Lett 37:2533–2536, 1996. JO Hoberg, JJ Bozell. Tetrahedron Lett 36:6831–6834, 1995. CM Timmers, M Dekker, RC Buijsman, GA van der Marel, B Ethel, G Anderson, B Burchell, GJ Mulder, JH van Boom. Bioorg Med Chem Lett 7:1501–1506, 1997.

130

Postema and Calimente

104.

MA Leeuwenburgh, CM Timmers, G van der Marel, JH van Boom, JM Mallet, PG Sinay¨. Tetrahedron Lett 38:6251–6254, 1997. DA Evans, BW Trotter, B Coˆte´. Tetrahedron Lett 39:1709–1712, 1998. J Gervay, TM Flaherty, D Holmes. Tetrahedron 53:16355–16364, 1997. L Lay, F Nicotra, L Panza, G Russo. Synlett 167–168, 1995. A Dondoni, M-C Scherrmann. J Org Chem 59:6404–6412, 1994. MEL Sa´nchez, V Michelet, I Besnier, JP Geneˆt. Synlett 705–708, 1994. T Granier, A Vasella. Helv Chim Acta 78:1738–1746, 1995. WM Best, V Ferro, J Harle, RV Stick, DMG Tillbrook. Aust J Chem 50:463–472, 1997. E Ayadi, S Czernecki, J Xie. J Chem Soc Chem Commun 347–348, 1996. OR Martin, OM Saavedra. Tetrahedron Lett 36:799–802, 1995. E Calzada, CA Clarke, C Roussin-Bouchard, RH Wightman. J Chem Soc Perkin Trans I 517–518, 1995. S Hanessian, C Girard. Synlett 865–867, 1994. R Csuk, U Ho¨ring, M Schaade. Tetrahedron 52:9759–9776, 1996. R Csuk, M Ku¨hn, D Stro¨hl. Tetrahedron 53:1311–1322, 1997. T Benneche. Synthesis 1–27, 1995. K-I Kim, RI Hollingsworth. Tetrahedron Lett 35:1031–1032, 1994. C Leteux, A Veyrie`res. J Chem Soc Perkin Trans I 2647–2655, 1994. T Fuchss, H Streicher, RR Schmidt. Liebigs Ann Recl 1315–1321, 1997. A Je´gou, C Pacheco, A Veyrie`res. Synlett 54:81–83, 1998. IP Smoliakova, R Caple, D Gregory, WA Smit, AS Shashkov, OS Chizov. J Org Chem 60:1221–1227, 1995. IP Smoliakova, R Caple, JW Brenny, WA Smit, YK Kryschenko, AS Shashkov, OS Chizhov, MZ Krimer, GV Morar, YB Kalyan. Synlett 275–276, 1995. P Verlhac, C Leteux, L Toupet, A Veyrie`res. Carbohydr Res 291:11–20, 1996. AM Go´mez, M Casillas, S Valverde, JC Lo´pez. J Chem Soc Chem Commun 2357– 2358, 1996. Y Igarashi, T Shiozawa, Y Ichikawa. Bioorg Med Chem Lett 7:613–616, 1997. D Craig, JP Tierney, C Williamson. Tetrahedron Lett 38:4153–4156, 1997. D Craig, MW Pennington, P Warner. Tetrahedron Lett 36:5815–5818, 1995. E Alwarez, R Pe´rez, M Rico, RM Rodrı´guez, MC Sua´rez, JD Martı´n. Synlett 1082– 1084, 1996. S Lemaire-Audoire, P Vogel. Tetrahedron Lett. 39:1345–1348, 1998. T Mukaiyama, H Uchiro, N Hirano, T Ishikawa. Chem Lett 629–630, 1996. T Mukaiyama, T Ishikawa, H Uchiro. Chem Lett 389–390, 1997. I Hachiya, S Kobayashi. Tetrahedron Lett 35:3319–3320, 1994. G Casiraghi, G Rassu. Synthesis 607–626, 1995. M Pichon, B Figade`re, A Cave´. Tetrahedron Lett 37:7963–7966, 1996. G Rassu, F Zanardi, L Battistini, G Casiraghi. Tetrahedron Asymm 6:371–374, 1995. B Figade`re, J-F Peyrat, A Cave´. J Org Chem 62:3428–3429, 1997. T Minehan, Y Kishi. Tetrahedron Lett 38:6815–6818, 1997. DE Levy, F Dasgupta, PC Tang. Tetrahedron Asymm 5:2265–2268, 1994. A Dondoni, A Marra, A Massi. J Org Chem 64:933–944, 1999. X Wang, PH Gross. Liebigs Ann 1367–1370, 1995. G Wulff, G Clarkson. Carbohydr Res 257:81–95, 1994. K Horita, Y Sakurai, M Nagasawa, D Yonemitsu. Chem Pharm Bull 45:1558–1572, 1997. CR Johnson, A Golebiowski, H Sundram, MW Miller, RL Dwaihy. Tetrahedron Lett 36:653–654, 1995. T Fuchss, RR Schmidt. Synthesis 753–758, 1998.

105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146.

C-Glycoside Synthesis 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163. 164.

131

MF Buffet, DJ Dixon, GL Edwards, SV Ley, E Tate. Synlett 1055–1056, 1997. MP Dillon, H Maag, DM Muszynski. Tetrahedron Lett 36:5469–5470, 1995. H Sano, S Mio, N Tsukaguchi, S Sugai. Tetrahedron 51:1387–1394, 1995. H Sano, S Mio, J Kitagawa, S Sugai. Tetrahedron Asymm 5:2233–2240, 1994. T Yamanoi, A Fujioka, T Inazu. Bull Chem Soc Jpn 67:1488–1491, 1994. T Kuribayahsi, N Ohkawa, S Satoh. Tetrahedron Lett 39:4537–4540, 1998. E El Telbani, S El Desoky, MA Hammad, AH Abdel Rahman, RR Schmidt. Carbohydr Res 306:463–467, 1998. T Kumazawa, K Ohki, M Ishida, S Sato, J Onodera, S Matsuba. Bull Chem Soc Jpn 68:1379–1384, 1995. K Mallareddy, SP Rao. Tetrahedron 52:8535–8544, 1996. FL Andrews, DS Larsen. Tetrahedron Lett 35:8693–8696, 1994. K Suzuki. Pure Appl Chem 66:1557–1564, 1994. K Toshima, G Matsuo, T Ishizuka, Y Ushiki, M Nakata, S Matsumura. J Org Chem 63:2307–2313, 1998. K Toshima, Y Ushiki, G Matsuo, S Matsumura. Tetrahedron Lett 38:7375–7378, 1997. MK Park, JH Park, SY Cho, YG Shin, J-K Jung, Y-G Suh. Tetrahedron Lett 38:6411– 6414, 1997. T Hosoya, Y Ohashi, T Matsumoto, K Suzuki. Tetrahedron Lett 37:663–666, 1996. T Kumazawa, M Ishida, S Matusba, S Sato, J Onodera. Carbohydr Res 297:379–383, 1997. T Kuribayahsi, N Ohkawa, S Satoh. Tetrahedron Lett 39:4541–4542, 1998. K Toshima, G Matsuo, M Nakata, S Matsumura. J Synth Org Chem Jpn 56:841–850, 1998.

5 D-Arabinofuranosides

from Mycobacteria: Synthesis and Conformation Todd L. Lowary The Ohio State University, Columbus, Ohio

I.

INTRODUCTION

Unlike the majority of monosaccharides, the pentose arabinose exists in nature in all possible absolute and ring configurations, namely, D-arabinofuranose (D-Araf ), L-arabinofuranose (L-Araf ), L-arabinopyranose (L-Arap), and D-arabinopyranose (DArap). The pyranose forms are more rare and are found primarily either in protozoan parasites [1] or as constituents of plant saponins [2]. Oligo- and polysaccharides comprised of L-arabinofuranose are also widespread in the plant kingdom, where they are present as constituents of arabinoxylans [3], pectins [4], and hydroxyprolinerich glycoproteins (HPRGs) [5]. This chapter focuses on polysaccharides containing D-arabinofuranose and in particular the most prominent examples of these polymers, which are found as important components of the cell wall of members of the Actinomycetes family including the genera Mycobacteria, Corynebacteria, Nocardia, and Rhodococcus [6]. Although the majority of actinomycetes are benign to humans, among them are two important human pathogens, Mycobacterium tuberculosis and Mycobacterium leprae [7]. Infection by these organisms causes, respectively, tuberculosis and leprosy. Mycobacterial infections have attracted renewed attention in recent years owing to their increasing incidence in the industrialized world as well as the emergence of drug-resistant strains of these organisms [8]. Additionally, AIDS patients and others with compromised immune systems are susceptible to opportunistic infections caused by ‘‘atypical’’ mycobacteria including M. avium and M. kansasii [9]. 133

134

Lowary

The total number of people infected with M. tuberculosis has been estimated at a third of the world’s population and almost 3 million deaths from tuberculosis occur each year, making it the single most lethal bacterial disease [10]. Treating mycobacterial infections is difficult and usually requires a 6–12 month course of antibiotic therapy [11]. The need for antibiotics over an extended period is due largely to the structure of the mycobacterial cell wall, which serves as a formidable barrier to the passage of antibiotics into the organism as well as into the immune system of the human host. Drug regimens that are in current use generally include one or two antibiotics that inhibit cell wall biosynthesis (e.g., ethambutol or isoniazid), in combination with others that have intracellular targets (e.g., rifampicin or streptomycin). When administered together, such combinations destroy the integrity of the cell wall, which in turn allows other antibiotics to pass to the cell more efficiently [12]. Even ignoring the issue of drug resistance, the arsenal of drugs that are active against mycobacteria is relatively small [8], and thus there is a pressing need for the identification of new antibiotics. When considered along with the emergence of drug-resistant mycobacterial strains, this need becomes even more urgent. In the search to identify new antibiotics, one area receiving particular attention is the identification of compounds that act by inhibiting mycobacterial cell wall biosynthesis. An increasing number of papers on this topic have appeared in the last few years [13–17]. However, a detailed understanding of the cell wall structure has really become available only in the last decade [6,13], and an in-depth understanding of its biosynthesis is unavailable [13]. Many more fundamental chemical and biochemical studies need to be completed before the rational design of such antibiotics can be realized. The mycobacterial cell wall is composed, to a very large degree, of polysaccharides and lipids [6]. The two major polysaccharides are an arabinogalactan (AG) and a lipoarabinomannan (LAM). The predominant lipid species are mycolic acids, branched long-chain lipids characteristic to the actinomycetes. What is particularly unique about both AG and LAM is that all the arabinose and galactose residues are present in the furanose form. Since oligosaccharides containing furanose residues are xenobiotic to mammalian biochemistry, the enzymes that assemble these polysaccharides are ideal targets for drug action. This chapter focuses on the arabinan portions of the AG and LAM. Sections II–IV outline the structure, biochemical importance, and biosynthesis of these arabinans. Section V discusses the chemical syntheses of oligosaccharides containing D-arabinofuranose residues. Such compounds are important tools in biochemical studies leading to a more detailed understanding of cell wall biosynthesis. Section VI focuses on work that we and others are doing to elucidate the solution conformation of oligo- and polysaccharides containing D-arabinofuranose residues, and Section VII offers some concluding remarks. This review covers work published up to the end of 1998. II.

ARABINAN STRUCTURE

Only recently has a detailed picture of the structure of the mycobacterial cell wall emerged [6,18,19]. Although the constituent species present have been known for many years [20,21], the assembly of these constituents into a coherent structure has required lengthy investigations and has relied on a number of techniques including,

D-Arabinofuranosides from

Mycobacteria

135

in addition to methylation analysis, both NMR spectroscopy and fast-atom bombardment high-resolution mass spectrometry [6,18,19]. These studies were also greatly facilitated by the discovery [22] of hydrolytic enzymes capable of digesting both AG and LAM into small fragments, which are more amenable to these characterization methods. A.

Arabinogalactan

In common with gram-positive bacteria, mycobacteria have a layer of peptidoglycan immediately external to the cytoplasmic membrane. This peptidoglycan serves as the scaffold to which the AG is attached. Figure 1 is a schematic drawing of the overall structure of a single AG molecule [6]. A number of structural motifs are worthy of mention, and these are illustrated in Figure 2. Linkage of the AG to peptidoglycan is by way of a phosphodiester through a disaccharide comprised of rhamnose and N-acetylglucosamine (1). Attached to the 4-hydroxyl group of the rhamnose moiety is a linear chain of approximately 30 D-galactofuranose (D-Galf ) residues with alternating ␤-(1→5) and ␤-(1→6) linkages (2). Along the galactofuranose chain are branch points at which arabinan chains are attached. This arabinan contains in total approximately 70 D-arabinofuranose residues and consists primarily of linear ␣(1→5)-linked chains (3). However, there are periodic branch points at which another linear arabinan chain is attached via an ␣-(1→3) linkage (4). At the distal, nonreducing end of these chains is the branched hexasaccharide 5. In a single AG molecule, approximately two-thirds of these hexasaccharide motifs are esterified to mycolic acids at the four primary alcohols to provide 6. These lipids, which are unique to mycobacteria and other actinomycetes, are branched fatty acids containing 70–90 carbons as well as cyclopropane, ketone, hydroxyl, and methoxyl functionalities [23,24]. An example of one type of mycolic acid, a ketomycolate, 7, is shown in Figure 2. B.

Lipoarabinomannan

LAM, which has a less homogeneous structure than the AG, is attached noncovalently to the cytoplasmic membrane through the lipid portion of a phosphatidylinositol (PI) linker. A schematic drawing of LAM is presented in Figure 3 [6]. A mannan, comprised of ␣-(1→6) linked D-mannopyranose residues, is covalently bound via a glycosidic linkage to the inositol O6. Approximately half these mannose residues carry an ␣-(1→2) D-mannopyranosyl branch. An arabinan very similar to that found in the AG is attached at the termini of the mannan. In common with the AG arabinan, the major structural motif is an ␣-(1→5)-linked chain with periodic branch points, and hexasaccharide 5 at the nonreducing termini. However, there are fewer branch points, and thus the ratio of linear to branched arabinofuranose in LAM is greater than in AG. In addition, while the hexasaccharide motifs (5) found in the AG are esterified to mycolic acids, in LAM these moieties are found either unsubstituted or, in particularly virulent strains, capped with short mannopyranosyl oligosaccharides (8). When capped with the mannose residues, the polymer is referred to as ManLAM. Other structural features have been discovered more recently. For example, hexasaccharide 5 found in LAM of M. smegmatis is capped with inositol phosphates (9), thus providing a new type of LAM termed PI-GAM (phosphoinositol-glyceroarabino-mannan) [25,26].

136

Schematic representation of mycobacterial arabinogalactan (AG).

Lowary

Figure 1

D-Arabinofuranosides from

Figure 2

III.

Mycobacteria

137

Structural motifs present in mycobacterial AG and LAM.

ROLE OF THE AG AND LAM IN DISEASE PROGRESSION

The AG together with the mycolic acids comprise the major structural component of the mycobacterial cell wall; intercalated within this framework is LAM. These polymers play paramount roles in enabling these organisms to survive in the human host. For example, in contrast to many other microorganisms, mycobacteria can thrive in human macrophages instead of being killed by them [27]. This property is believed to be due, in large part, to the impenetrable nature of the cell wall envelope. In addition to protecting the organism from macrophage destruction, the cell wall is believed to substantially reduce the flow of antibiotics into the organism [12,28] and to be involved in the initial stages of the infection by the binding of cell wall oligosaccharides to host receptors [29]. A.

Arabinogalactan

The mycolic esters at the nonreducing ends of the AG present a formidable hydrophobic barrier that prevents the passage of antibiotics into the organism [28]. The currently accepted structural model of the mycolyl–AG complex was initially proposed by Minnikin in 1982 [24] and is illustrated in Figure 4. In this model the mycolic acid chains are packed side by side, perpendicular to the plane of the cy-

138

Schematic representation of mycobacterial lipoarabinomannan (LAM).

Lowary

Figure 3

D-Arabinofuranosides from

Figure 4

Mycobacteria

139

Schematic drawing of the mycolyl–AG complex.

toplasmic membrane. The tight packing of these chains results in very low fluidity of the cell wall, which in some cases is essentially crystalline [30,31]. The fluidity is dependent on the mycolic acids present, which in turn are species specific [31]. However, in all cases a major problem in treating mycobacterial infections is the passage of drugs through this lipid barrier into the host. This model was initially criticized by some, who suggested that the AG was too large to accommodate the tight packing of the lipid chains [28]. However, a more detailed look at the structure of the AG provides a solution to this problem. A distinguishing feature of mycobacterial AG is the almost exclusive presence of furanosyl (Araf or Galf ) residues. Furanoside rings are known to be much less conformationally well defined than their pyranoside counterparts [32]. It has been suggested that this inherent increased flexibility, in addition to the predominant coupling of these residues through linkages that involve primary hydroxyl groups, results in the polysaccharide being a scaffold of high flexibility. This in turn allows the polymer to adopt a conformation that facilitates the orientation of the mycolic acids into tightly packed parallel arrays. Additionally, this proposal provides a reasonable explanation for what, from an evolutionary standpoint, seems odd. Polysaccharides containing furanosyl residues are expected to be less thermodynamically stable than those possessing pyranose rings, and it therefore appears that these organisms have developed a system whereby glycan stability is traded for the protection afforded by the tightly packed mycolic acids. B.

Lipoarabinomannan

The major antigenic component of the mycobacterial cell wall is LAM. It should be pointed out that although not explicitly shown in Figure 4, LAM is interspersed throughout the mycolyl–AG framework, and this polymer is believed to be long enough that any groups found at the nonreducing terminus (e.g., the mannopyranosyl oligosaccharides) extend through the mycolic acid layer and are expressed on the outer surface of the cell wall [33]. LAM has been suggested to play critical roles in many immunomodulatory events occurring during progression of the disease. Among these are the inhibition of macrophage activation [34], the neutralization of potentially cytotoxic O2 free radicals [35], the induction of cytokines [36–38], the inhi-

140

Lowary

bition of protein kinase activities [39], and the induced expression of collegenases that destroy the extracellular matrix of the lung [40]. LAM in which the primary hydroxyl groups of 5 are capped with mannopyranosyl oligosaccharides (8, ManLAM) has been suggested to be involved in the infection process. It is believed that these terminal mannose residues initiate infection by adhering to human cells through their recognition by human mannose binding proteins [41–43]. In contrast to the capping of 5 by mannopyranosyl residues, mycobacterial species (e.g., M. smegmatis) in which this motif is capped with inositol phosphate residues (8) are avirulent. It has been proposed that the lack of pathogenicity of this species arises from the induction of tumor necrosis factor ␣ (TNF-␣) by the inositol phosphate moieties, which in turn results in the killing of the bacteria by macrophages [26]. Perhaps the most important recent discovery is that T cells recognize LAM via antigen presentation pathways that are independent of the major histocompatibility complex (MHC) [44,45]. It was shown that LAM is initially recognized by a mannose receptor and the protein CD14 before it is processed and then complexed with CD1b. This complex is then expressed on the surface of the antigen-presenting cell. Recognition by the T cell requires both the carbohydrate and lipid portion of LAM, and two distinct cell lines have been shown to differentiate between structurally different LAM fragments. Thus it appears that there is a significant amount of specificity in T-cell responses mediated by the LAM-complexed CD1 glycoproteins. IV.

ARABINAN BIOSYNTHESIS

Given the critical importance of AG and LAM to the survival of mycobacteria, there is increasing interest in developing new drugs for the treatment of these diseases, which act by inhibiting the enzymes involved in their biosynthesis. Inhibitors of the glycosyltransferases involved in the assembly of the furanosidic oligosaccharide portions of the glycoconjugates are particularly attractive targets. Glycans of this type are unknown in humans, and therefore blocking these biosynthetic pathways will have little deleterious effect on the host. However, for rational drug design to succeed, more details about the biosynthetic pathways leading to the formation of these polysaccharides are required. Although the overall features of this process are known [13,18,19], many of the details are not. The ␤-(1→2)- and ␣-(1→5)-linked arabinose residues are incorporated into the polymer from the activated polyprenyl sugar phosphate 10, which is in turn synthesized from glucose via 5-phosphoribose pyrophosphate (pRpp) [46–48]. Elongation of the polymer chain is believed to involve a family of arabinosyltransferases (AraT’s) that recognize both 10 and arabinofuranoside-based acceptors of differing structures (Fig. 5) [18,19,49,50]. In AG biosynthesis, the entire polysaccharide appears to be assembled as a polyprenol diphosphate intermediate, which is transferred to peptidoglycan prior to the addition of the mycolate esters [18]. In LAM biogenesis, the arabinan portion is believed to be synthesized as a polyprenol phosphate that is transferred to lipomannan [51]. Incorporation of arabinose into AG has been demonstrated through the use of synthetic, radiolabeled 10 and mycobacterial membrane extracts [50]. These reactions are believed to involve membrane-bound substrates and enzymes and to occur extracellularly. While to date none of these putative glycosyltransferases have been purified to homogeneity, the gene cluster that encodes for one or more of these AraT’s

D-Arabinofuranosides from

Figure 5

Mycobacteria

141

A typical reaction catalyzed by mycobacterial arabinosyltransferases (AraT’s).

has been cloned and expressed [52–54]. In addition, an assay for their activity that uses mycobacterial membrane preparations as the enzyme source has been developed [49,50]. The assay measures the incorporation of radiolabeled arabinose from 10 into arabinan, and it has been used to screen both potential substrates and inhibitors [46,47,55,56]. Although the natural acceptor substrates for these enzymes are lipid-bound intermediates, these AraT’s also recognize small arabinofuranosyl oligosaccharides. A few of the oligoarabinosides that have been shown to be substrates for these enzymes are illustrated in Figure 6 [49,55]. Arabinofuranose disaccharides appear to be the minimum epitopes. Furthermore, investigation of the AraT recognition of the disaccharide ␣-D-Araf-(1→5)-␣-D-Araf-R (13–15 ) and the trisaccharide ␣-D-Araf(1→5)-␣-D-Araf-(1→5)-␣-D-Araf-R (16, 17) revealed that glycosides with R = octyl (14, 17) are better substrates and those with R = methyl (13, 16). Surprisingly, the dodecyl glycoside 15 was a poorer substrate than 13. In the investigation of 14 and 17, the products of the enzymatic reactions were isolated and their structures elucidated [49]. In both cases, mixtures of two products, corresponding to addition of an Araf residue to either the 2- or the 5-position of the terminal residue in the oligosaccharide acceptor, were obtained. From this study it was concluded that 10 was the source of the ␤-(1→2)-, and ␣-(1→5)-linked Araf residues. Less is known about the formation of the ␣-(1→3) branch points than about the ␤-(1→2) and ␣-(1→5) linkages. Only ␤-(1→2) and ␣-(1→5) linkages are formed by the incubation of small oligosaccharide substrates with 10 in the presence of a mycobacterial membrane preparations [49]. However, additional work [57] demonstrated that when radioactive 10 is incubated with endogenous acceptors and a mycobacterial membrane preparation, a polymer essentially identical to native AG can be obtained. Additionally the radioactivity in the polymer was equally distributed

142

Lowary

Figure 6 Arabinofuranosyl oligosaccharides tested as substrates for mycobacterial arabinosyltransferases.

throughout, which suggests that 10 is the major, and possibly sole, source of Araf residues. The lack of ␣-(1→3) linkages formed in the initial study [49] could be attributed to a number of factors: (1) the instability or absence of ␣-(1→3) AraT activity in the membrane preparation, (2) the possibility that this enzyme recognizes oligosaccharide substrates larger than those investigated, or (3) the possibility that another activated donor (e.g., a sugar nucleotide) is used by this AraT. Although the presence of UDP-Araf in mycobacteria has been reported [58], the incorporation of arabinose from this source into arabinan has not been demonstrated. The range of donor substrates that are tolerated by these enzymes has also been explored. A series of analogs was prepared, each one differing from 10 only in the identity of the lipid chain, and the set was screened for biological activity [56]. It was discovered that conjugates of C50 and C55 prenols were substrates, while those compounds with shorter (
Action of Ethambutol

From the discussion above is it clear that both AG and LAM play paramount roles in the infection and survival of mycobacteria in the host. Drugs that act by inhibiting the biosynthesis of these polysaccharides are expected to show antimycobacterial action. One of the commonly used antituberculosis agents, ethambutol [(S,S⬘)-2,2⬘(ethylenediimino)di-1-butanol, 22] (Fig. 7), has been used for the treatment of tuberculosis since 1961, when it was first reported to have antimycobacterial activity

D-Arabinofuranosides from

Figure 7

Mycobacteria

143

Structure of the antituberculosis drug ethambutol.

[59]. This drug is active only against mycobacteria [8], and while it was thought for many years to be involved in inhibiting cell wall biosynthesis [60,61], its precise target was not elucidated until recently. The structure of the drug is relatively simple and bears little resemblance to any cell wall component, which have complicated the identification of its precise biochemical target. It is now known that the drug acts as an AraT inhibitor and disrupts the biosynthesis of the arabinan portions of both AG and LAM [43,47,62–64].

V.

SYNTHESIS OF D-ARABINOFURANOSIDES

In comparison with the synthesis of oligopyranosides, the preparation of oligosaccharides containing furanose residues has received relatively little attention. However, this will likely change as the biological importance of glycans containing these residues becomes more widely appreciated. The methodologies used to date for assembling these molecules have closely paralleled the methods used for the assembly of pyranosidic linkages. Nevertheless, these interesting structures also open up the possibility for new methodologies to be developed, and these will undoubtedly emerge. In this section are outlined the chemical synthesis of D-arabinofuranosides that have been reported. As explained in Section IV, although the normal acceptor substrates for mycobacterial arabinosyltransferases are high molecular weight glycolipids, some (and probably most, if not all) of these enzymes will recognize small oligosaccharide fragments of these large glycans [49,55]. In this regard, these enzymes appear to be similar to mammalian glycosyltransferases [65]. Therefore, these small oligosaccharides are invaluable tools not only for mapping out substrate specificities of these enzymes but also as templates for the design of potentially inhibitory analogs. A.

Route to Arabinofuranose Donors

At equilibrium in solution, most sugars exist predominantly in the pyranose forms. For example, in water, D-arabinose is found in approximately a 95:5 pyranose-tofuranose ratio [66]. Therefore, the initial step in any synthesis of an oligofuranoside is the conversion of this mixture of pyranoses to a furanosyl derivative that can be incorporated into oligosaccharides. This can be done either directly or indirectly as outlined below.

144

1.

Lowary

Fisher Glycosylation

Furanose glycosides of simple alcohols are conveniently obtained by subjecting a reducing sugar to a controlled Fisher glycosylation reaction. The furanosides are the kinetic products of this reaction, and excellent yields of the product (as a mixture of anomers) are usually obtained if the proper conditions are chosen [67]. The alcohol generally used in this reaction is methanol, which serves as both a reactant and the solvent. The methyl glycosides thus obtained must then be protected and converted to a suitable glycosylation agent. It is also possible to synthesize glycosides of other alcohols by this method; however, since the alcohol is usually employed as the solvent, the choice is somewhat limited. Therefore, from a practical point of view, this method is most useful for the preparation of glycosides of relatively volatile alcohols (e.g., methanol, ethanol, and allyl alcohol). This method works well for the preparation of methyl D-arabinofuranosides, and an excellent preparation is available [68] that can be used to prepare multigram quantities of the product (Fig. 8). Treatment of D-arabinose (23) with acidic methanol at room temperature provides a 1:1 ␣ : ␤ mixture of methyl arabinofuranosides (24). Benzoylation provides 25, again as a mixture of anomers; however, the ␣ isomer can be selectively crystallized from ethanol. With either 24 or 25 in hand, it is possible to prepare, in 1–3 steps, the glycosyl donors shown in Figure 9. These include glycosyl acetates 26 [69] and 27 [70,71], glycosyl halides 28 [68,72], 29 [49], 30 [73], 31 [73], and thioglycosides 32–34 [55,74–76]. 2.

Cyclization of Acyclic Dithioacetals and Related Derivatives

An old [67,77] but recently rediscovered [78] method for synthesizing furanose glycosides involves the cyclization of acyclic sugar S,O-acetals (e.g., 35) in the presence of mercuric salts (Fig. 10). This method is highly selective, producing furanosides in the absence of pyranosides, and is especially useful for the synthesis of galactofuranosides (e.g., 36) of both simple and carbohydrate alcohols [67,78]. This approach has not yet been used for the synthesis of D-arabinofuranosides; however, two related syntheses have been reported (Fig. 11). Work by Zinner and coworkers [79] demonstrated that the cyclization of the 5-O-benzoyl derivative of D-arabinose diethyl dithioacetal (37) in mercuric chloride afforded modest yields of thioglycoside 38. More recently, the isopropylidene derivative 39 has been prepared in good yield from 37 by treatment with mercuric chloride in acetone [80]. Conversion of 37 to any of the derivatives shown in Figure 9 can be achieved in a few steps. 3.

Ozonolysis of Glycals

The most recently reported route to the D-arabinofuranose ring nucleus involves an oxidative cleavage of protected glucal derivatives as illustrated in Figure 12 [74].

Figure 8

Fisher glycosylation route 2,3,5-tri-O-benzoyl-␣-D-arabinofuranoside (25).

D-Arabinofuranosides from

Figure 9

Mycobacteria

Synthesis of commonly employed D-arabinofuranosyl donors.

145

146

Lowary

Figure 10

Synthesis of ␤-galactofuranosides from acylic S,O-sugar acetals.

Ozonolysis of 3,4,6-tri-O-acetyl D-glucal (40), followed by reductive workup with dimethyl sulfide, affords the formyl aldehyde 41. The formate ester can be selectively cleaved in the presence of the acetates by refluxing in acidic methanol yielding the reducing sugar 42. Attempts to use 42 directly in glycosylation reactions via in situ activation of the anomeric hydroxyl group [81,82] failed, but this derivative could be converted in two steps and excellent yield to thioglycoside 33 [74]. Protected D-glucal derivatives are readily available, and consequently this methodology is an attractive alternative to more conventional routes to arabinofuranosides. In particular, the conversion of fully 13C-labeled glucose to labeled arabinose oligosaccharides by this route has been demonstrated [79]. For the synthesis of 13C-labeled oligoarabinosides, this method is preferable to a strategy involving Fisher glycosylation of labeled D-arabinose because the cost of the starting material (13C-labeled D-glucose) is fourfold cheaper, while the overall yields of the transformations are similar. B.

Glycosylation Reactions with Arabinofuranose Donors

The glycosyl donors illustrated in Figure 9 have been the ones most commonly employed in the preparation of oligosaccharides containing D-arabinofuranosyl residues. The vast majority of glycosidic bonds in mycobacterial arabinan are ␣ link-

Figure 11

Cyclization of arabinose dithioacetals to give furanose derivatives.

D-Arabinofuranosides from

Mycobacteria

147

Figure 12

Synthesis of protected arabinofuranose reducing sugars by ozonolysis of acetylated D-glucal.

ages, and thus the relationship between the groups at C1 and C2 are trans. Consequently, the synthesis of these linkages can be achieved through the use of glycosyl donors possessing acyl protecting groups on O2. As would be expected, neighboring group participation resulting in the formation of an acyloxonium ion provides, in most cases, good to excellent yields of the 1,2-trans-glycoside. Donors 26–29 , 32, and 33 have been used effectively for these glycosylation reactions; however, with some donors the stereocontrol is poorer than in others. For example, it has been demonstrated that while thioglycoside 33 provides good yields of the desired product, peracetate 27 gave significant amounts of the 1,2-cis product [74]. Similar observations have been made with other arabinofuranosyl glycosyl donors [71,83]. In addition to 1,2-trans-D-arabinofuranosyl linkages, mycobacterial arabinan also contains ␤-D-arabinofuranosyl residues (1,2-cis-glycosides). These linkages are directly analogous to ␤-D-mannopyranosides, which are among the most difficult of all glycosidic linkages to prepare [84]. However, in contrast to the ␤-D-mannoside case, no significant effort has been directed toward the stereocontrolled synthesis of ␤-D-arabinofuranosides. A minimum prerequisite to the synthesis of any such linkage are donors possessing a nonparticipating group at O2. For this purpose, the 2-Obenzyl derivatives 30, 31, and 34, have been used [73,85,86] to produce, in most cases, mixtures from which the desired compounds could be isolated. Characterizing the stereochemical course of these glycosylation reactions is easily done by means of standard one-dimensional 13C and 1H NMR experiments [83]. The anomeric carbon of the ␣ anomers resonate between 107 and 110 ppm; the ␤ anomers between 97 and 104 ppm. Furthermore, in the ␣ anomers 3JH1,H2 is small (0–2 Hz), whereas in the ␤ anomers the magnitude of this coupling is larger (3–5 Hz). It should also be mentioned that although 1JC1,H1 values are unambiguous determinants of anomeric stereochemistry in pyranosides [87], they cannot be used to make these assignments in arabinofuranosides as their magnitudes have been shown to be insensitive to anomeric configuration [83]. 1.

Syntheses of 1,2-trans-Linked Oligosaccharides

The disaccharide ␣-D-Araf-(1→5)-␣-D-Araf comprises the core structure of the linear sections of the arabinan. A number of syntheses of this glycan nucleus, as glycosides of a variety of alcohols have been reported. The first synthesis was of the allyl glycoside 48, which was prepared as illustrated in Figure 13 [72]. Orthoester 43 was prepared from the bromide 29 and then rearranged to give the allyl glycoside 44. A series of protection and deprotection steps afforded the trityl ether 45, which was subsequently glycosylated with orthoester 46 (also prepared from 29). Deprotection afforded the target, 48. In the same paper, this route was used to prepare the benzyl glycoside starting from a benzyl orthoester analogous to 43.

148

Synthesis of disaccharide 48: All, allyl; Tr, triphenylmethyl.

Lowary

Figure 13

D-Arabinofuranosides from

Mycobacteria

149

As outlined in Figure 14, a more recent synthesis [55,74] of this disaccharide as the methyl glycoside (13, Fig. 6) made use of thioglycoside donor 33. This synthesis began with the preparation of alcohol 49 from 24␣, through a sequence of silylation, benzoylation, and desilylation. Glycosylation promoted by N-iodosuccinimide and silver triflate afforded the disaccharide 50, which could be deprotected with sodium methoxide to provide 13. A number of other derivatives of this disaccharide have been synthesized, namely, the 2⬘-chloroethyl, octyl (14), dodecyl (15), and hexyl glycosides [49,71]. These syntheses have employed either acetate 27 or chloride 28 as the glycosylating agent. The trisaccharide ␣-D-Araf-(1→5)-␣-D-Araf-(1→5)-␣-D-Araf is also a core fragment of the linear arabinan, and two synthetic derivatives have been reported. The chloride 28 has been used to prepare the S-octyl glycoside analog (17), as well as the methyl glycoside (16) [49]. Another route to 16 is illustrated in Figure 15 [55]. Monosaccharide 33 was converted to the differentially protected thioglycoside 51 in three steps. Glycosylation of 49 with 51 provided the protected disaccharide 52, which upon treatment with tetrabutylammonium fluoride was transformed to alcohol 53. The trisaccharide was then obtained by glycosylation of 53 with 33 and then deprotection. Four other oligosaccharides (18–21 , Fig. 6) corresponding to fragments of either the linear or the branched hexasaccharide moieties of the arabinan have been synthesized as methyl glycosides [55,74]. These were obtained in a manner similar to the disaccharide and trisaccharide via glycosylation of methyl glycoside acceptors with thioglycoside donors.

Figure 14

Synthesis of disaccharide 13: TBDPS, t-butyldiphenylsilyl; NIS, N-iodosuccinimide; AgOTf, silver trifluoromethanesulfonate.

150

Lowary

Figure 15 Synthesis of trisaccharide 16: TBDPS, t-butyldiphenylsilyl; NIS, N-iodosuccinimide; AgOTf, silver trifluoromethanesulfonate.

2.

Syntheses of 1,2-cis-Linked Oligosaccharides

The stereocontrolled synthesis of ␤-arabinofuranosides has received little attention since the reports, over 30 years ago, that solvolyses of 2-O-benzylated ␣-D-arabinofuranosyl halides (e.g., 30 or 31) with methanol gave mixtures of glycosides in which the ␤ isomer predominates [86,87]. It is of tangential interest to note that kinetic studies were done, and it was proposed that this reaction proceeds through an ion pair SN1, not SN2 mechanism. More recently, this method has been used to synthesize octyl ␤-D-arabinofuranosides from 30 and 31 [73]. As outlined in Figure 16, reaction of these labile chlorides in the presence of octanol without a promoter gave modest yields of ␤glycosides 54 and 55. The reaction of 30 under these conditions gave only the desired product; however, in the case of 31, traces of the ␣ isomer were also detected. Although the yields are modest, the ease by which the starting material can be prepared (two steps from commercially available 2,3,5-tri-O-benzyl arabinofuranose, 56) makes this an attractive route to these products. Unfortunately, while this method can be used for the synthesis of glycosides of simple primary alcohols, attempts to glycosylate secondary carbohydrate alcohols by this method failed, and thus the method appears to be of limited utility for the synthesis of oligosaccharides. On the other hand, reaction of the pyridyl thioglycoside 34 with alcohol 57 has been shown [76] to provide good yields of the ␤-linked disaccharide 58 (Fig. 17). No mention was made of the formation of any ␣-glycoside product in this

D-Arabinofuranosides from

Figure 16

Mycobacteria

151

Synthesis of ␤-D-arabinofuranosides by alcoholysis of glycosyl chlorides.

glycosylation reaction. This disaccharide product was then used as a synthon for the preparation of a pentasaccharide fragment (62) of the hexasaccharide motif found at the nonreducing terminus of AG and LAM. Coupling of 58 with alcohol 59 followed by deprotection of the silyl group afforded trisaccharide 60. Glycosylation of 60, again with 58, provided the protected pentasaccharide 61, which was subsequently hydrogenated to afford the product. In this synthesis the first glycosylation proceeded stereoselectively to afford the ␣-linked product, but in the second case, a separable 3:2 ␣ : ␤ mixture was produced. Unquestionably, the stereocontrolled synthesis of ␤-D-arabinofuranosides is an area ripe for investigation. Indeed, no general methods for their preparation currently exist. Although new strategies will likely appear, it is probable that many of the methodologies used for ␤-mannoside synthesis can be applied to this problem. Indeed, recent work [88,89] has shown that Ogawa’s [90,91] intermolecular aglycone delivery method for ␤-mannoside synthesis can be used for the stereoselective synthesis of ␤-D-fructofuranosides (e.g., 65) (Fig. 18). Given the very close structural similarities between D-arabinofuranose and D-fructofuranose, it is likely that the approach can be used, but this has yet to be demonstrated.

VI.

CONFORMATIONAL STUDIES

A detailed appreciation of the biological function of a molecule requires that its conformation be understood. Therefore, in recent years a great deal of effort has been directed toward elucidating the solution conformation of oligosaccharides [92]. The work, however, has been heavily focused on oligopyranosides. At the same time, a wealth of information detailing the conformation of furanose rings in nucleotides and nucleic acids has accumulated [93]. However, except for sucrose [94,95], only recently have conformational investigations of oligosaccharides containing furanose residues been reported [96,97]. Outlined in this section are some recent investigations that have explored the solution conformation of methyl ␣-D-arabinofuranoside and oligosaccharides containing ␣-D-arabinofuranosyl residues.

152

Lowary

Figure 17

Synthesis of 62: IDCP, iodine dicollidine perchlorate; TBDPS, t-butyldiphen-

ylsilyl.

A.

Background

In oligosaccharides containing pyranosyl residues, most, if not all, of the constituent rings are rigid, existing in one of two possible chair conformations. Thus, the major degrees of freedom available to an oligopyranoside are rotation about glycosidic bonds, which can be described by two torsion angles (⌽, ⌿), and rotation about the exocyclic C—C bond (torsion angle ␻), as indicated in Figure 19 [98]. Three possible staggered conformations of ␻ are possible (gg, gt, and tg, Fig. 20) and usually a mixture is present in solution [99]. NMR studies focused on elucidating oligosaccharide structure have relied heavily on the measurement, of 3JH,H to determine ring conformations and rotamer populations of exocyclic hydroxymethyl groups of the

D-Arabinofuranosides from

Figure 18

Mycobacteria

153

Synthesis of ␤-D-fructofuranosides by intramolecular aglycone delivery.

constituent monosaccharide residues [92,99]. The value of ⌽ is fixed by the exoanomeric effect, which place one of the lone pairs of the glycosidic oxygen antiperiplanar to the C1-ring oxygen bond, and the value of ⌿ is determined largely by steric constraints. The ⌽, ⌿ angles for a glycosidic linkage cannot be determined by means of 3JH,H values. Therefore, to orient the individual residues together in three-dimensional space, most investigations have employed either interresidue nuclear Overhauser effect (NOE) constraints [92] or 3JC,H values [92,100–104] measured across glycosidic linkages. In oligofuranosides, the problem is more complex. In addition to rotation about glycosidic and exocyclic C—C bonds, the rings themselves are flexible. Similar to substituted cyclopentanes, furanose moieties exist in either envelope (E) or twist (T) conformations [105]. At the monosaccharide level, often two of these conformations are of roughly equal energy and are thus present in equilibrium. These puckered conformers generally interconvert through pseudorotation, a process whereby, through a series of small molecular reorganizations, two conformations can equilibrate. An alternate pathway, inversion through the planar ring form, is disfavored because in the planar form, all substituents on the ring are eclipsed, which results in a significant energy penalty due to torsional strain [106]. A particular ring conformer can be described by two parameters, the puckering amplitude (␶m) and the pseudorotational phase angle (P), which can be illustrated by the pseudorotational wheel shown in Figure 21 for a D-aldofuranose [106]. The radius of the circle is ␶m and P, which defines the part of the ring that is most puckered,

Figure 19 furanosides.

Glycosidic, hydroxymethyl group, and ring torsion angles in pyranosides and

154

Figure 20

Lowary

Staggered rotamers about the C5–C6 bond in a pyranose ring (C4–C5 in

furanoses).

is indicated along with the associated envelope or twist conformer. A consequence of ring pucker is that in addition to the ⌽, ⌿, and ␻ torsion angles, furanoses possess a ring torsion angle, ␹ (Fig. 19). In solution, furanose rings typically exist as an equilibrating mixture of conformers, one in the northern hemisphere and another in the southern, respectively termed the north (N) and south (S) conformers. Given two low-energy conformations, for example 3T2 and 2T3, equilibration can occur through one of two pathways, either via the east or west. In the east pathway, usually favored for D-pentofuranoses, equilibration proceeds through the following series of conformers: 3T2 ↔ 3E ↔ 3T4 ↔ E4 ↔ ⬚E ↔ ⬚T1 ↔ E1 ↔ 2T1 ↔ E2 ↔ 2T3 [106]. In the case of furanose rings, where a mixture of conformers is equilibrating rapidly on the NMR time scale, conformational analysis becomes complicated because averaging of coupling constants is observed. To facilitate analysis, a commer-

Figure 21 Pseudorotational wheel for a D-aldofuranosyl ring: T, twist; E, envelope; superscript numbers indicate the atoms lying above the plane, while subscript numbers are atoms displaced below the plane.

D-Arabinofuranosides from

Mycobacteria

155

cially available least-squares minimization program (PSEUROT 6.2) can be used, given the observed intracyclic ring 3JH,H and the ␶m for each conformer, to calculate the N/S ratio and to provide P values for both conformers from which the ring forms (e.g., 3T2) can be determined [107]. Similarly, equations have been derived for calculating populations of rotamers about the exocyclic C—C bond from coupling constant data [99,108,109]. B.

NMR Studies

Mindful of the discussion above, it is clear that when one is probing the conformation of an oligofuranoside, the critical first issue to be addressed is that of the conformational equilibrium of the constituent monosaccharide rings. Once this equilibrium is understood, the three-dimensional orientation of the sugar residues relative to each other can be assessed to provide an overall picture of the shape of the molecule. In recent investigations [110], the equilibrium populations of ring conformers in methyl ␣-D-arabinofuranoside (24␣, Fig. 8) and a series of related di-, tri-, and tetrasaccharides (13, 16, 18–21 , Fig. 6) have been studied. These investigations have led to a better understanding of D-arabinofuranose ring conformation and have allowed the determination of the effect of ring substitution on conformer equilibria. High-field (600 and 800 MHz) NMR spectroscopy has been used for these studies. The oligosaccharides of interest contain only ␣-D-arabinofuranose residues, and consequently the potential for spectral overlap is high. However, in all the compounds studied to date, the spectral resolution at these field strengths has been good enough that standard one- and two-dimensional experiments (e.g., 1H-1H COSY, 1H-1H TOCSY, HMQC) can be used not only in assigning all the 1H and 13C resonances, but also in the measurement of coupling constants. Once the couplings have been obtained, PSEUROT analysis has provided conformer identities and their populations. As illustrated in Figure 22, this analysis has identified that the ring in 24␣ adopts a roughly equimolar distribution of the ⬚T4 (N) and 2T3 (S) conformations. In both these conformers the anomeric methoxy group is oriented in a pseudoaxial orientation, as would be expected on the basis of the endo anomeric effect. Furthermore, in both conformers the hydroxymethyl group lies in a pseudoequatorial fashion, as would be favored on the basis of steric interactions. In the N conformer, both the secondary OH groups adopt sterically favored pseudoequatorial orientations; however, in the S conformer, these groups are pseudoaxial. Although conformers possessing this orientation of substituents might not be expected to be energy minima on the basis of sterics, gauche effects [111] between these hydroxyl oxygens stabilize this conformer. Similar effects have been observed in the furanose rings of nucleic acids [112]. Comparison of the two solution conformers with that of the crystal structure [113] of octyl ␣-D-arabinofuranoside (66) and 24a [114] reveals that the rings in these crystals adopt a conformation (E4) very similar to the solution N conformer. Extension of these investigations to the oligosaccharides showed that both the conformations present and their populations can, in some cases, be affected by glycosylation. Figure 23 illustrates the N and S conformers present for each ring in disaccharides 13 and 18. In the ␣-(1→5)-linked disaccharide (13), each ring adopts the same two conformations as the monosaccharide, and in essentially the same populations. However, in the ␣-(1→3)-linked disaccharide (18), while the nonre-

156

Lowary

Figure 22

Solution N and S conformers for 24␣ as determined by PSEUROT analysis.

ducing end residue behaves like the monosaccharide, the reducing end residue is profoundly affected by glycosylation. In the 3-O-substituted ring, the S conformer remains at 2T3 but the identity of the N conformer changes, slightly, to ⬚E. In addition and more strikingly, the N conformer is favored with an equilibrium population of 86%. Clearly, glycosylation of a D-arabinofuranoside at O3 has a substantial influence upon the conformers available to the ring. Steric arguments can be used to rationalize these results, and similar trends are observed (Fig. 24) when these studies are carried out on the larger oligosaccharides (16, 19–21 ). From the data available to date, some general conclusions cn be drawn. 1.

Glycosylation, and likely any form of substitution, of a primary hydroxyl group in an arabinofuranose ring does not significantly alter the conformers present or their populations.

Figure 23 ulations.

Ring conformers present in disaccharides 13 and 18 and their relative pop-

D-Arabinofuranosides from

Mycobacteria

Figure 24

Ring conformers present in oligosaccharides 16 and 19–21 and their relative populations. 157

158

Lowary

2.

3.

4.

Similarly, replacement of a methyl group at the anomeric center with a monosaccharide residue does not alter the equilibrium. Hence, all terminal Araf residues behave like the monosaccharide. Glycosylation of O3 on an arabinofuranose ring does alter the conformational equilibria of the ring significantly. The identity of the N conformation changes from ⬚T4 to ⬚E, and this conformation is favored at equilibrium. In cases in which a ring is glycosylated at both O3 and O5, as in 19 and 21, favoring of the N conformation becomes even more pronounced.

The populations of rotamers about the C4—C5 bond group are relatively constant and are insensitive to ring conformation. Rotamer populations about this bond have been calculated using 3J4,5R and 3J4,5S values, and in all the oligosaccharides, the gg orientation is most heavily populated (47–53%). The gt rotamer is also present in significant amounts (34–40%), while the tg rotamer is the least populated (12– 15%). In the gg and gt orientations a gauche effect [111] between the ring oxygen and O5 stabilizes these rotamers over the tg conformational isomer, in which O5 is trans to the ring oxygen. C.

Computational Work

The initial computational studies on D-arabinofuranosides were molecular mechanics calculations of methyl ␤-D-arabinofuranoside done by Pe´rez and coworkers [115]. Through a process by which the MM3 force field was used to optimize 360 unique conformations, two energy minima, the 3T2 (N) and 2T3 (S) conformations, were identified. The N conformer is about 1.3 kcal/mol more stable than the S, and the favored conformation about the C4—C5 bond was the gt orientation, which was favored over the gg rotamer by less than 2 kcal/mol. Additionally, the most favorable pathway for pseudorotation was shown to be through the east. More recently, gas phase ab initio (HF/6-31G*, HF/6-31⫹G**, MP2/631⫹G**, MP3/6-31⫹G**) and density functional theory (B3LYP/6-31G*, B3LYP/ 6-31⫹G**) calculations have been carried out on methyl ␣-D-arabinofuranose [116]. A protocol developed by Serianni and coworkers [117,118], was used to construct the 10 possible envelope forms by fixing four atoms in a plane; then, upon minimization, the other parameters (bond lengths, bond angles, etc.) were optimized. These studies identified 3E and 2E as the N and S minima, respectively, with the latter being the global minima (Fig. 25). B3LYP/6-31⫹G** calculations indicated that in the gas phase, the N conformer was 1.5 kcal/mol less stable than the S conformer. Additionally, comparison of the relative energies of the conformers showed, as expected, that pseudorotation through the east was preferred. The highest energy conformer along this pathway is ⬚E (3.2 kcal/mol less stable than the global minima), whereas pseudorotation through the west would require proceeding through the E conformer, which is 7.0 kcal/mol less stable than the global minima. Inter⬚ estingly, these studies also showed that the planar species is approximately equivalent in energy to the E conformer, and thus inversion through the planar ring form is a ⬚ viable alternative to pseudorotation through the west. However, neither pathway is expected to be important because the east pseudorotation pathway is more favorable. In 9 of the 10 envelopes, the favored rotamer about the C4—C5 bond is the gg orientation; the gt rotamer is favored in the other. In the gg orientation O5 can

D-Arabinofuranosides from

Mycobacteria

159

N (3E) and S (2E) conformers for 24␣ as determined by gas phase density fashion theory (B3LYP/6-31G*) calculations.

Figure 25

form an intramolecular H bond with OH2 and this capability appears to be of major importance in stabilization of the conformations in the gas phase. Indeed, in the two most stable ring conformers, 2E and E1, an intramolecular H bond is formed between these two groups (OH2 ⭈ ⭈ ⭈ O5) in addition to another such interaction between OH3 and O1 (OH3 ⭈ ⭈ ⭈ O1). These H bonds undoubtedly stabilize the conformers to a significant degree. This observation is consistent with other recent gas phase calculations on sugars, which revealed that the formation of intramolecular H bonds is critically important in the stabilization of these molecules [119]. VII.

CONCLUDING REMARKS

There is now a thorough understanding of the structure of mycobacterial arabinan, and biosynthetic studies are underway that will lead to the elucidation of the pathways by which it is assembled. These investigations will depend critically on the access to synthetic oligosaccharides and oligosaccharide analogs that can be used as potential substrates or inhibitors of these enzymes. Consequently, the synthesis of arabinofuranosides and more generally oligofuranosides will prove to be an interesting and important research area in the future. Additionally, if inhibitors of the enzymes that biosynthesize these glycans are to be identified, a much better understanding of the conformational preferences of oligofuranosides will be necessary. In this area too, although there have been some fundamental investigations, a great deal more must be accomplished before even a rudimentary understanding of the conformation of these molecules is available. REFERENCES 1. 2. 3.

P Schneider, A Treumann, KG Milne, MJ McConville, N Zitzmann, MA Ferguson. Biochem J 313:963–971, 1996. M Straubinger, H Knapp, N Watanabe, N Oka, H Washio, P Winterhalter. Nat Prod Lett 13:5–10, 1999. MC Jarvis. Plant Cell Environ 7:153–164, 1984.

160

Lowary

4.

JA De Vries, FM Rombouts, AGJ Voragen, W Pilnik. Carbohydr Polym 1:117–127, 1981. FM Kils. New Compr Biochem 29a:511–520, 1995. PJ Brennan, H Nikaido. Annu Rev Biochem 64:29–63, 1995. C Ratledge, J Stanford. In: C Ratledge, J Stanford, eds. The Biology of Mycobacteria, vol. 1. London: Academic Press, 1982, pp 1–2. JS Blanchard. Annu Rev Biochem 65:215–239, 1996. MM Braun, RH Byers, WL Heyward, CA Ciesielski, AB Bloch, RL Berkelman, DE Snider. Arch Intern Med 150:1913–1916, 1990. BR Bloom, CJJ Murray. Science 257:1055–1064, 1992. DA Mitchison. Tubercle 66:219–225, 1985. V Jarlier, H Nikaido. FEMS Microbiol Lett 123:11–18, 1994. GS Besra, PJ Brennan. J Pharm Pharmacol 49(suppl 1):25–30, 1997. JA Maddry, N Bansal, LE Bermudez, RN Comber, IM Orme, WJ Suling, LN Wilson, RC Reynolds. Bioorg Med Chem Lett 8:237–242, 1998. RE Lee, MD Smith, RJ Nash, RC Griffiths, M McNeil, RK Grewal, WX Yan, GS Besra, PJ Brennan, GWJ Fleet. Tetrahedron Lett 38:6733–6736, 1997. JA Maddry, WJ Suling, RC Reynolds. Res Microbiol 147:106–112, 1995. C Bouix, P Bisseret, J Eustache. Tetrahedron Lett 39:825–828, 1998. PJ Brennan, GS Besra. Biochem Soc Trans 25:188–194, 1997. RE Lee, PJ Brennan, GS Besra. Curr Topics Microbiol 215:1–27, 1996. PJ Brennan. Rev Infect Dis 11:S420–S430, 1989. MR McNeil, PJ Brennan. Res Microbiol 142:451–463, 1991. MR McNeil, KG Robuck, M Harter, PJ Brennan. Glycobiology 4:165–173, 1994. K Takayama, N Qureshi. In: GP Kubica, LG Wayne, eds. Mycobacteria: A Sourcebook, Part A. New York: Marcel Dekker, 1984, pp 315–344. D Minnikin. In: C Ratledge, J Stanford, eds. The Biology of Mycobacteria, vol 1. London: Academic Press, 1982, pp 95–184. KH Khoo, A Dell, HR Morris, PJ Brennan, D Chatterjee. J Biol Chem 270:12380– 12389, 1995. M Gilleron, N Himoudi, O Adam, P Constant, A Venisse, M Rivie`re, G Puzo. J Biol Chem 272:117–124, 1997. MJ Fenton, MW Vermeulen. Infect Immun 64:683–690, 1996. ND Connell, H Nikaido. In: BR Bloom, ed. Tuberculosis: Pathogenesis, Protection and Control. Washington DC: American Society for Microbiology, 1994, pp 333–352. LS Schlesinger. Curr Top Microbiol 215:71–96, 1996. J Liu, CE Barry III, GS Besra, H Nikaido. J Biol Chem 271:29545–29551, 1996. H Nikaido, SH Kim, EY Rosenberg. Mol Microbiol 8:1025–1030, 1993. E Westhof, M Sundaralingam. J Am Chem Soc 105:970–976, 1983. D Chatterjee, KH Khoo. Glycobiology 8:113–120, 1998. LD Sibley, SW Hunter, PJ Brennan, JL Krahenbuhl. Infect Immun 56:1232–1236, 1988. J Chan, X Fan, SW Hunter, PJ Brennan, BR Bloom. Infect Immun 59:1755–1761, 1991. TI Roach, CH Barton, D Chatterjee, JM Blackwell. J Immun 150:1886–1896, 1993. Y Zhang, M Broser, H Cohen, M Bodkin, K Law, J Reibman, WN Rom. J Clin Invest 95:586–592, 1995. A Yoshida, Y Kodie. Infect Immun 65:1953–1955, 1997. KL Knutson, Z Hmama, P Herrera-Velit, R Rochford, NE Reiner. J Biol Chem 273: 645–652, 1998. JC Chang, A Wysocki, KM Tchou-Wong, N Moskowitz, Y Zhang, WN Rom. Thorax 51:306–311, 1996.

5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40.

D-Arabinofuranosides from

41. 42. 43. 44.

45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74.

Mycobacteria

161

LS Schlesinger, SR Hull, TM Kaufman. J Immunol 152:4070–4079, 1994. LS Schlesinger, TM Kaufman, S Iyer, SR Hull, LK Marchiando. J Immunol 157:4568– 4575, 1996. BK Kang, LS Schlesinger. Infect Immun 66:2769–2777, 1998. PA Sieling, D Chatterjee, SA Porcelli, TI Prigozy, RJ Mazzaccaro, T Soriano, BR Bloom, MB Brenner, M Kronenberg, PJ Brennan, RL Modlin. Science 269:227–230, 1995. SA Porcelli, CT Morita, RL Modlin. Curr Opin Immunol 8:510–516, 1996. BA Wolucka, MR McNeil, E de Hoffmann, T Chojnacki, PJ Brennan. J Biol Chem 269:23328–23335, 1994. MS Scherman, A Weston, K Duncan, A Whittington, R Upton, L Deng, R Comber, JD Friedrich, M McNeil. J Bacteriol 177:7125–7130, 1995. MS Scherman, L Kalbe-Bournonville, D Bush, Y Xin, L Deng, M McNeil. J Biol Chem 271:29652–29658, 1996. RE Lee, PJ Brennan, GS Besra. Glycobiology 7:1121–1128, 1997. RE Lee, K Mikusova´, PJ Brennan, GS Besra. J Am Chem Soc 177:11829–11832, 1995. GS Besra, CB Morehouse, CM Rittner, CJ Waechter, PJ Brennan. J Biol Chem 272: 18460–18466, 1997. AE Belanger, GS Besra, ME Ford, K Mikusova´, JT Belisle, PJ Brennan, JM Inamine. Proc Natl Acad Sci USA 93:11919–11924, 1996. A Telenti, WJ Philip, S Sreevatsan, C Bernasconi, KE Stockbauer, BE Wieles, JM Musser, WR Jacobs Jr. Nature Medicine 3:567–570, 1997. S Sreevatsan, KE Stockbauer, X Pan, BN Kreiswirth, SL Moghazeh, WR Jacobs Jr, A Telenti, JM Musser. Antimicrob Agents Chemother 41:1677–1681, 1997. JD Ayers, TL Lowary, CB Morehouse, GS Besra. Bioorg Med Chem Lett 8:437–442, 1998. RE Lee, PJ Brennan, GS Besra. Bioorg Med Chem Lett 8:951–954, 1998. Y Xin, RE Lee, MS Scherman, KH Khoo, GS Besra, PJ Brennan, M McNeil. Biochim Biophys Acta 1335:231–234, 1997. S Singh, SE Hogan. Microbios 77:217–222, 1994. JP Thomas, CO Braughn, RG Wilkinson, RF Shepherd. Am Rev Respir Dis 83:891– 893, 1961. JO Kilburn, K Takayama. Antimicrob Agents Chemother 20:401–404, 1981. JO Kilburn, K Takayama. Antimicrob Agents Chemother 33:1493–1499, 1989. L Deng, K Mikusova´, KG Robuck, M Scherman, PJ Brennan, MR McNeil. Antimicrob Agents Chemother 39:694–701, 1995. K Mikusova´, RA Slayden, GS Besra, PJ Brennan. Antimicrob Agents Chemother 39: 2484–2489, 1995. KH Khoo, E Douglas, P Azadi, JM Inamine, GS Besra, K Mikusova´, PJ Brennan, D Chatterjee. J Biol Chem 271:28682–28690, 1996. MM Palcic, O Hindsgaul. Trends Glycosci Glycotechnol 8:37–49, 1996. SJ Angyal. Adv Carbohydr Chem Biochem 42:15–68, 1984. JW Green. Adv Carbohydr Chem Biochem 21:95–142, 1966. HG Fletcher Jr. Methods Carbohydr Chem 2:228–230, 1963. RL Tolman, DA Baker. Methods Carbohydr Chem 7:59–62, 1976. BL Kam, JL Barascut, JL Imbach. Carbohydr Res 69:135–142, 1979. AK Pathak, YA El-Kattan, N Bansal, JA Maddry, RC Reynolds. Tetrahedron Lett 39: 1497–1500, 1998. K Hatanaka, H Kuzuhara. J Carbohydr Chem 4:333–345, 1985. V Subramaniam, TL Lowary. Tetrahedron 55:5965–5976, 1999. WD D’Souza, PE Cheshev, JD Ayers, TL Lowary. J Org Chem 63:9037–9044, 1998.

162

Lowary

75. 76. 77. 78. 79. 80. 81. 82. 83.

JB Houseknecht, TL Lowary. Unpublished results. HB Mereyala, S Hotha, MK Gurjar. J Chem Soc Chem Commun 685–686, 1998. ML Wolfrom, DI Weisblat, AR Hanze. J Am Chem Soc 66:2065–2068, 1944. JC McAuliffe, O Hindsgaul. J Org Chem 62:1234–1239, 1996. H Zinner, A Koine, H Nimz. Chem Ber 93:2705–2712, 1960. WJ Guilford, SD Copley, JR Knowles. J Am Chem Soc 109:5013–5019, 1987. J Inanga, Y Yokoyama, T Hanamoto. J Chem Soc Chem Commun 1090–1091, 1993. BA Garcia, JL Poole, DY Gin. J Am Chem Soc 119:7597–7598, 1997. K Mizutani, R Kasai, M Nakamura, O Tanaka, H Matsuura. Carbohydr Res 185:27– 38, 1989. F Baressi, O Hindsgaul. In: SH Khan, RA O’Neil, eds. Modern Methods in Carbohydrate Synthesis, vol 1. Amsterdam: Harwood Academic Publishers, 1996, pp 251– 276. CPJ Glaudemans, HG Fletcher Jr. J Am Chem Soc 87:2456–2461, 1965. CPJ Glaudemans, HG Fletcher Jr. J Am Chem Soc 87:4636–4641, 1965. K Bock, CJ Pedersen. Chem Soc Perkin Trans 2:293–297, 1974. C Krog-Jensen, S Oscarson. J Org Chem 61:4512–4513, 1996. C Krog-Jensen, S Oscarson. J Org Chem 63:1780–1784, 1998. Y Ito, T Ogawa. Angew Chem Int Ed Engl 33:1765–1767, 1994. A Dan, Y Ito, T Ogawa. J Org Chem 60:4680–4681, 1995. SW Homans. New Compr Biochem 29a:67–86, 1995. AN Lane. ACS Symp Ser 682:106–121, 1998. K Bock, RU Lemieux. Carbohydr Res 152:63–74, 1986. B Adams, L Lerner. J Am Chem Soc 114:4827–4829, 1992. K Mazeau, S Pere´z. Carbohydr Res 311:203–217, 1998. Q Xu, R Gitti, CA Bush. Glycobiology 6:281–288, 1996. Nomenclature of Carbohydrates. Adv Carbohydr Chem Biochem 52:47–177, 1997. K Bock, JØ Duus. J Carbohydr Chem 13:513–543, 1994. B Mulloy, TA Frenkiel, DB Davies. Carbohydr Res 184:39–46, 1988. I Tvaroska, M Hricovı´ni, E Petra´kova´. Carbohydr Res 189:359–362, 1989. I Tvaroska. Carbohydr Res 206:55–64, 1990. I Tvaroska, FR Taravel. Adv Carbohydr Chem Biochem 51:15–61, 1995. JM Duker, AS Serianni. Carbohydr Res 249:281–303, 1993. C Altona, M Sundaralingam. J Am Chem Soc 94:8205–8212, 1972. E Westhof, M Sundaralingam. J Am Chem Soc 102:1493–1500, 1980. FAAM de Leeuw, C Altona. J Comp Chem 4:428–437, 1983. GD Wu, AS Serianni, R Barker. J Org Chem 48:1750–1757, 1983. JA Gerlt, AV Youngblood. J Am Chem Soc 102:7433–7438, 1980. WD D’Souza, JD Ayers, PR McCarren, TL Lowary. J Am Chem Soc 122:1251–1260, 2000. S Wolfe. Acc Chem Res 5:102–111, 1972. J Plavec, C Thibaudeau, J Chattopadhyaya. Pure Appl Chem 68:2137–2144, 1996. JC Gallucci, PR McCarren, TL Lowary. Manuscript in preparation. AG Evdokimov, AJ Kalb, TF Koetzle, WT Klooster, JML Martin. J Phys Chem A 103: 744–753, 1999. S Cros, C Herve´ du Penhoat, S Pe´rez, A Imberty. Carbohydr Res 248:81–93, 1993. MT Gordon, TL Lowary, CM Hadad. J Am Chem Soc 121:9682–9692, 1999. DM Chipman, AS Serianni. J Am Chem Soc 109:5297–5303, 1987. TJ Church, I Carmichael, AS Serianni. J Am Chem Soc 119:8946–8964, 1997. B Ma, HF Schaefer III, NL Allinger. J Am Chem Soc 120:3411–342, 1998.

84.

85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119.

6 Chemical Synthesis of Bioactive Steroidal Saponins Biao Yu and Yongzheng Hui Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, Shanghai, China

I.

INTRODUCTION

Saponins, glycosides of steroids or triterpenes, exist widely in terrestrial plants and usually in considerable amounts [1]. They are constituents of many significant foods, forage crops, and herbal medicinal plants. The foodstuffs containing steroidal saponins include such familiar items as oats, leeks, onions, garlic, chives, asparagus, sarsaparillas, and yams. Saponins are of special importance in traditional Chinese herbs, and the list of those containing saponins as active components is long [2]. The herb genera commonly used in China that contain steroidal saponins as active principles include Dioscorea, Paris, Polygonatum, Ophiopogon, Liriope, Asparagus, Yucca, Allium, Smilax, Solanum, and Ruscus. Saponin extracts from these herbs have been used in China to treat various human diseases with surprisingly good precision. However, the precise mechanisms of the actions of these saponins remain ambiguous by modern standards. One reason is that each saponin extract from a herbal plant normally contains several dozens saponin components. These components have such similar structural patterns, and some components exist in such tiny amounts, that separation of each component is, if not impossible, an extremely formidable task. The structural diversity of saponins lies mainly in their ‘‘glycoforms’’ (a term used to describe a set of sugars attached to the same peptide chain in animals) [3]. And many investigations have shown that the ‘‘glycoform’’ of a saponin is important to its bioactivity [1]. Chemical synthesis should provide a realistic way to determine the availability of homogeneous saponins, thus affording new opportunities for understanding and applying this important group of natural products. However, only sporadic studies 163

164

Yu and Hui

on the chemical synthesis of saponins have been conducted and reported [4]. Two recent examples are noted. Randolph and Danishefsky used their ‘‘glycal assembly’’ process to synthesize a tigogenyl tetrasaccharide from Digitalis purpurea [5], and Yamada and Nishizawa synthesized a sweet saponin (osladine) from Polypodium vulgare [6]. We set out to synthesize saponins in recent years as one of numerous research projects in China focusing on traditional Chinese herbs. This chapter presents our study of the chemical synthesis of bioactive steroidal saponins. II.

TWO BASIC STRATEGIES

Structurally, steroidal saponins have two quite distinct parts: a steroid part and a sugar part. The chemical synthesis of steroids was one of the hottest fields during 1950s to 1970s after the discovery of steroidal hormones and contraceptives [7]. Almost all the famous synthetic chemists of that period did something related to steroids. And then a big steroid industry was established. Some steroids of the steroidal saponins are important industrial materials (e.g., diosgenin, tigogenin, and hecogenin, etc.); others, which can be readily prepared from them and their industrial degradation products, include pregnenolone, dehydroisoandrosterone, and progesterone (Scheme 1). It is also practical to isolate considerable amounts of some steroids from the corresponding plants after removal of the ‘‘glycoforms.’’ The chemical synthesis of oligosaccharides has been a major focus since the 1980s [8] as a result of our increased understanding of the importance of these sugars in many life processes [3]. Although with low efficiency and sometimes with tremendous difficulties, almost all the sugar linkages known in nature have been synthesized. Since the chemical syntheses of oligosaccharides and steroids have been studied intensively, the key to synthesizing a saponin turns out to be the construction of the glycosidic bond between the oligosaccharide and the steroid. Accordingly, there are two basic strategies, outlined in Scheme 2: one is to fabricate an oligosaccharide donor and then attach it to the aglycone (strategy 1); the other is to begin by con-

Scheme 1

Some important industrial steroids.

Synthesis of Bioactive Steroidal Saponins

Scheme 2

165

Two basic strategies for the synthesis of saponins.

necting the first monosaccharide to the aglycone, then manipulate the protecting groups of the sugar moiety, and extend the sugar chain sequentially (strategy 2). Strategy 1 looks straightforward; however, it is risky, considering the unavailability of generally applicable glycosylation protocols [8]. In employing strategy 2, the formation of the glycosidic bond between the sugar and the steroid can be sought to be stereospecific and in high yields at the monosaccharide level, but the subsequent protecting group manipulation becomes lengthy and complicated. From another point of view, a series of saponins with identical oligosacchride chains but different steroids can be readily prepared by means of strategy 1, by connecting the oligosaccharide donor to various steroids; if strategy 2 is used, a series of saponins with the same steroid but different oligosaccharide chains similar in pattern can be readily prepared by employing a number of monosaccharide donors in each glycosylation steps. The systematic synthesis of both series of saponins would facilitate the evaluation of the respective biological roles of the sugar part and the steroid part of saponins. A.

Strategy 1: Synthesis of OSW-1 (1)

Saponin OSW-1 (1) is the major component of a small group of cholestane saponins isolated by Sashida et al. from the bulbs of Ornithogalum saudersia. This molecule, while showing little toxicity to normal cells in vitro, exhibited remarkable activities against a broad spectrum of malignant tumor cells, with IC50 between 0.1 and 0.7 nM, which is about 10–100 times more potent than such clinical anticancer agents as mitomycin C, adriamycin, cisplatin, camptothecin, and taxol [9]. According to strategy 1, OSW-1 (1) was broken down into steroid 2 and disasccharide donor 3 (Scheme 3). Glycosyl trichloroacetimidate with a neighboring participation group (3) is believed to be one of the most reliable donors [8b]. Since

166

Scheme 3

Yu and Hui

Retrosynthesis of OSW-1 (1).

either acidic or basic conditions might cause migration or cleavage of the acetyl (Ac) and 4-methoxybenzoyl (MBz) groups and the degradation of the whole molecule of OSW-1, protecting groups [tert-butyldimethylsilyl (TBS) and triethylsilyl (TES) for hydroxyls and ethylene glycol ketal for the keto] were selected to allow complete removal under neutral or near neutral conditions. Steroid 2 was prepared from dehydroisoandrosterone in a total of 12 steps and 10% yield (Scheme 4) [10]. The stereochemistries involved in the transformations were completely controlled by the steroid substrates themselves (4 → 5, 5 → 6, 8 → 9, 9 → 2). The 3-OH was first protected by a tert-butyldiphenylsilyl (TBDPS) group because the TBS protection was found to be susceptible in the ene reaction (4 → 5). Then, because it was found in a model reaction that the conditions for removal of the 3-OTBDPS group resulted in considerable migration and cleavage of the Ac and MBz groups in the sugar moiety of the target molecule (1), the 3OTBDPS was transferred into TBS protection (7 → 8). The disaccharide donor 3 was prepared as shown in Scheme 5. Coupling of the arabinopyranoside diol 11 with xylopyranosyl imidate donor 10 gave 12, the 3OH (equatorial hydroxyl) glycosylated product preferentially, in 70% yield. Since

Synthesis of Bioactive Steroidal Saponins

167

Scheme 4

Preparation of the aglycone of OSW-1 (2). (a) Ph3PEtBr, tert-BuOK, 78%; TBDPSCl, 100%. (b) (CH2O)n, BF3OEt2, 75%. (c) Dess–Martin periodinane, 86%; 3-methylbutylmagnesium bromide, 96%. (d) PDC, 83%; HOCH2CH2OH, 96%. (e) TBAF, 95%; TBSCl, 96%. (f) OsO4, 41%. (g) ClCOCOCl, DMSO, 78%; NaBH4, CeCl3, 93%.

Scheme 5 Preparation of the disaccharide donor 3. (a) BF3OEt2. (b) TESOTf, 70% (two steps). (c) H2, Pd/C, 50%. (d) CCl3CN, DBU, 65%.

168

Yu and Hui

anomeric benzyl group was very difficult to remove (13 → 14), forcing conditions (50 atm H2, Pd/C, 50⬚C, 3 days, 50%) were used. Glycosylation of steroid 2 with disaccharide imidate donor 3 under the promotion of trimethylsilyl trifluoromethanesulfonate (TMSOTf) provided the desired glycoside 15 stereospecifically and in good yield (69%). Finally, all the protecting groups (one TBS, three TES, and one ethylene glycol ketal) were removed cleanly in one step by employing Pd(MeCN)2Cl2 as a catalyst to furnish the target OSW-1 (1) (Scheme 6) [11]. When strategy 1 was used, the first total synthesis of OSW-1 (1) was accomplished from dehydroisoandrosterone, L-arabinose, and D-xylose in a total of 27 steps, with the longest linear sequence being 14 steps, and in 6% yield [11]. The present route, at least after some optimization, would provide a practical entry to various OSW-1 analogs, especially those with the same sugar moiety as OSW-1, which can be readily prepared by coupling disaccharide donor 3 to various steroids. B.

Strategy 1: Synthesis of Maidong Saponin C (16)

Maidong, one of the most commonly used Chinese herbs, is held to nourish ‘‘yin’’ (tonic effect; ‘‘yin’’ is one of the basic philosophical concepts in traditional Chinese medicine). Saponins are believed to be the active principles in Maidong plants, which are found in several species, including some in the Ophiopogon and Liriope genera. Thus far more than 60 steroidal saponins have been isolated from these species [2c]. Yu et al. isolated Maidong saponin C (16) from the tuber of Liriope muscari [12]. Pharmacological studies showed that 16 possessed strong anti-inflammatory activity [13]. According to strategy 1, Maidong saponin C (16) was broken down into trisaccharide donor 17 and steroid 18 (Scheme 7). Again, a trichloroacetimidate donor (17) was used; without a neighboring participation group, an SN2 reaction is expected to achieve the formation of a ␤-glycosidic bond [8b]. Steroid 18 was conceivably derived from diosgenin by introducing a 1␤-OH. The 3-OH of 18 was believed to be more active than the 1␤-OH, and therefore protected as a TBS ether. Steroid 18 was prepared from diosgenin in 6 steps and 14% yield (Scheme 8). The configuration of the 1␣-OH (22) could not be inverted by using SN2-type reactions, probably owing to the stereohindrance of this position. Therefore an oxidation–reduction sequence was used to epimerize the 1␣-OH (22 → 18). The trans-

Scheme 6

Assembly of OSW-1 (1). (a) TMSOTf, 69%. (b) Pd(MeCN)2Cl2, 79%.

Synthesis of Bioactive Steroidal Saponins

Scheme 7

169

Retrosynthesis of Maidong saponin C (16).

formation of 20 to diol 21 under Li/NH3 reduction was found to be capricious; the ⌬6 product or other products sometimes dominated. Schmidt glycosylation [8b] was used twice to prepare the trisaccharide donor 17 (Scheme 9). Attempts to use xylopyranosyl imidate donor 28 in the regioselective glycosylation of diol 26, which has an equatorial 3-OH and an axial 4-OH as diol 11, were not successful. The deprotection of the anomeric allyl group of trisaccharide 29 was found to be problematic. When reactions were catalyzed by palladium spe-

Scheme 8 Preparation of aglycone 18. (a) DDQ, 69%. (b) H2O2, 68%. (c) Li/NH3, 53%. (d) TBSCl, 90%. (e) PDC, 92%. (f) NaBH4, 70% (for 18).

170

Yu and Hui

Scheme 9 Preparation of the trisaccharide donor 17. (a) BF3OEt2, 73%; 50% HOAc, 93%. (b) CH3C(OEt)3, 87%. (c) BF3OEt2, 90%. (d) I(CF2)6CL, Na2S2O4, NaHCO3; Zn, NH4Cl, 50%. (e) Cl3CCN, DBU, 83%.

cies, a mixture of three inseparable products, including the expected deallylation product and two Wacker oxidation products, resulted. We recently developed a new procedure for the deprotection of allyl group: the allyl group was converted into a ␤-iodide-␥-perfluoroalkane derivative using the perfluoroalkylation reaction and then reductive eliminated [14]. This procedure worked well on trisaccharide 29, giving 30 in good yield (50%). With donor 17 and steroid 18 at hand, we sought to effect the final glycosylation to assemble the target molecule 16. However, at this time we were not as lucky as we were in the OSW-1 synthesis. The final glycosylation was found to be difficult and capricious. Under very carefully controlled conditions, the corresponding glycosides were produced in moderate yields and in a mixture of their anomers (␣ : ␤ = 1:4.6). Deprotection of the ␤ anomer (31) gave the target molecule 16 (Scheme 10) [15].

Scheme 10 Assembly of Maidong saponin C (16). (a) TMSOTf (0.2 equiv). (b) TMSOTf (2.0 equiv); NaOMe, 85% (for two steps).

Synthesis of Bioactive Steroidal Saponins

171

It is obvious that the present synthetic approach to Maidong saponin C (16) is of no practical use because the final glycosylation is so inefficient. From this example, we sensed the disadvantage in applying strategy 1 to the saponin synthesis, especially when the final glycosylation would involve either an oligosaccharide in the absence of a neighboring group (containing 1→2 linkage) or a steroid with a stereohindered hydroxyl group. C.

Strategy 2: Synthesis of Two Series of Diosgenyl Saponins

Diosgenyl saponins are the most abundant existing steroidal saponins [1,16]. A typical structural pattern of the ‘‘glycoforms’’ in this family is one with a ␤-D-glucopyranoside as the first sugar attched to diosgenin, which in turn has an ␣-L-rhamnopyranose substituted at 2-OH and another sugar or sugar chain at either the third (series 1) or the fourth (series 2) hydroxyl of the glucopyranose moiety (Scheme 11). Many members of these two series of diosgenyl saponins have been found in a number of traditional Chinese herbs, such as those in species of Paris, Ophiopagon, Dioscorea, and Allium. Some members, which have been isolated in sufficient amounts to permit investigators to conduct bioactivity evaluations, have demonstrated various interesting antitumor, cardiovascular, and antifungal activities. According to strategy 2, the glucopyranose residue is coupled to diosgenin first. At this stage, a large array of glucopyranosyl donors and the glycosylation conditions are readily screened. Some results are listed in Scheme 12. The glycosylation of a steroid was known to be a rather special glycosylation reaction [17]. Two side reactions—namely, acetyl group transfer and orthoester formation—took place easily and under some conditions dominated. It was found that acyl group transfer could be avoided completely by substituting a benzoyl group for the acetyl group as neighboring group; orthoester formation depended on the promoter used. Finally, we disclosed that the combined use of benzoyl-protected trichloroacetimidate as donor and TMSOTf as promoter produced the normal glycoside quantitatively (entry 6 of Scheme 12). Moreover, this reaction was conveniently conducted at a 40 g scale without any loss in yield [18]. Diosgenyl disaccharides 44 and 45 with only 3⬘-OH and 4⬘-OH free, respectively, were two key intermediates in the synthesis of the two series of diosgenyl saponins. Their preparations are shown in Scheme 13. Coupling of the benzoylprotected glucopyranosyl imidate 34 with diosgenin, followed by removal of the benzoyl protection, afforded the diosgenyl ␤-D-glucopyranoside (trillin, 37) quantitatively and on a large scale. Attempts to regioselectively glycosylate diol (38) with rhamnopyranosyl imidate 41 were not successful. Therefore, a bulky protecting group

Scheme 11

Two series of diosgenyl saponins.

172

Scheme 12

Yu and Hui

Glycosylation of diosgenin with various glucopyranosyl donors.

[TBS or trimethylacetyl (Piv)] was used to monoprotect one of the two hydroxyls of 38. The Piv group was selectively protected at the 3-OH, producing 40 in 64% yield, with the corresponding 2-OPiv product in 9% yield and 2,3-di-OPiv product in 5% yield. The regioselectivity of TBS in this case was moderate, producing 39 in 52% yield with the corresponding 2-OTBS product in 35% yield [19,20]. Glycosylation of diosgenyl disaccharide 44 with monosaccharide trichloroacetimidates (24, 28, 41, 46) produced the corresponding fully protected diosgenyl trisaccharides (47–50 ) in excellent yields (84–98%). Cleavage of the benzylidene group under 80% HOAc followed by removal of the acetyl groups under NaOMe provided four series 1 diosgenyl saponins (51–54 ) (Scheme 14) [19]. Saponin 51 (gracillin), which has been isolated from many medicinal plants, including species of Paris, Dioscoea, and Costacea, was shown to be cardioactive and hemostatic [21,22]. Saponin 52 has been isolated from Ophiopogon, 53 from Taccacheancer, and 54 from Paris [1,16]. Glycosylation of diosgenyl disaccharide 45 with monosaccharide trichloroacetimidates (41, 46) and thiodisaccharide donors (55–57 ) produced the corresponding fully protected diosgenyl trisaccharides (58, 59) and tetrasaccharides (60–62 ) in good to excellent yields (65–93%). Removal of the acyl groups under NaOH provided the corresponding five series 2 diosgenyl saponins (63–67 ) (Scheme 15) [20]. Saponin 63 (dioscin), which exists widely in plants, including many species intensively used in traditional Chinese herbal medicine, itself exhibits antitumor activity (GI50 = 0.40 ␮g/mL against P-388) and antifungal activity [23]. Saponin 64 (polyphyllin D)

Synthesis of Bioactive Steroidal Saponins

173

Scheme 13

Preparation of the key intermediates 44 and 45. (a) PhCH(OMe)2, 89%. (b) TBSCl, 52%. (c) PivCl, 64%. (d) BF3OEt2, 100%. (e) TBAF, 80%. (f) TsOH, 80%; BzCl, 75%.

Scheme 14 Synthesis of series 1 saponins (51–54 ). (a) BF3OEt2, 84–98%. (b) 80% AcOH; NaOMe, 75–91%.

174

Yu and Hui

Scheme 15

Synthesis of series 2 saponins (63–67 ). (a) BF3OEt2 (for 41, 46); NIS/AgOTf (for 55–57 ), 65–93%. (b) NaOH, 75–99% (for 67, 80% HOAc was used to remove the benzylidene).

has been isolated from species of Paris and other genera and shows very promising cardioactive and antitumor properties [21,22]. Saponins 65 and 66 are components of Paris polyphylla [16]. Saponin 67 (balanitin 7) is one of the cytostatic components isolated from the East African medicinal plant Balanites aegyptica, with the ED50 of 0.22 ␮g/mL against P-388 tumor cells [24]. When strategy 2 was used, as long as the steroid has been connected to a sugar moiety, normal chemistry for the oligosaccharide synthesis could be applied to extend the sugar chain. As shown in Scheme 16, a facile ‘‘one-pot’’ glycosylation protocol [25] was efficiently used: treatment of the thioglycoside acceptor 68 with imidate donors (41 and 34, respectively) and TMSOTf, followed by the addition of diosgenyl disaccharide acceptor 45 and N-iodosuccinimide (NIS), generated the corresponding diosgenyl tetrasaccharides 69 and 70 in excellent yields (96 and 98%, respectively). Deprotection of 69 and 70 produced saponins 65 and 66, respectively [26]. III.

SUMMARY

Saponins are distributed widely in plants, including many significant foodstuffs and medicinal plants. In traditional Chinese herbal medicine, saponin components are of special importance; many of them exhibit various interesting bioactivities. However, research on saponins has been limited to the isolation of these compounds and the determination of their structure. Pharmacological evaluation of saponins has been mainly retained at the folkloric level. The difficulty in isolation of saponins from

Synthesis of Bioactive Steroidal Saponins

Scheme 16

175

‘‘One-pot’’ glycosylation in the synthesis of saponins.

plants, where they exist as complicated ‘‘glycoforms,’’ is one reason, and another is the long neglect of their importance, due to insufficient understanding and application of saponins by modern standards [27]. The increased interest in traditional herbal medicine and in ‘‘carbohydrate drugs’’ [28] should bring broad attention to this important group of natural glycosides. And chemical synthesis, as shown in the foregoing examples, can make homogeneous saponins readily available. ACKNOWLEDGMENT Appreciation is expressed to our excellent coworkers whose names are included in the references. Financial support was provided by the Ministry of Science and Technology of China. B. Y. is also grateful to a NSFC grant (No. 29925203). REFERENCES 1.

2.

3.

(a) K Hosttettmann, A Marston. Saponin. Cambridge: Cambridge University Press, 1995. (b) GR Wall, K Yamasaki, eds. Advances in Experimental Medicine and Biology, vol 404: Saponins Used in Traditional and Modern Medicine. New York: Plenum Press, 1996. (c) GR Wall, K Yamasaki, eds. Advances in Experimental Medicine and Biology, vol 405: Saponins Used in Food and Agriculture. New York: Plenum Press, 1996. (a) C Huang. The Pharmacology of Chinese Herbs. Boca Raton, FL: CRC Press, 1993. (b) LG Guo, J Ying, eds. Contemporary Research and Clinical Application on Chinese Herbs (in Chinese). Xueyuan Press, 1994. (c) GJ Xu, LS Xu, eds. Species Systemization and Quality Evaluation of Commonly Used Traditional Chinese Herbs (in Chinese), vol 1–2. Fujian Science Press, 1995, 1997. (d) ZC Lou, B Xu, eds. Species Systemization and Quality Evaluation of Commonly Used Traditional Chinese Herbs (in Chinese), vols 1–3. Beijing: Beijing Medical University and Union Medical University Press, 1995, 1996. (a) RA Dwek. Chem Rev 96:683–720, 1996. (b) A Varki. Glycobiology 3:97–127, 1993.

176 4.

5. 6. 7. 8.

9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.

24. 25. 26. 27. 28.

Yu and Hui (a) T Ikeda, T Kajimoto, J Kinjo, K Nakayama, T Nohara. Tetrahedron Lett 30:3513– 3516, 1998. (b) LN Atopkina, NI Uvarova, GB Elyakov. Carbohydr Res 303:449–451, 1997. (c) S Saito, Y Sasaki, K Kuroda, Y Hayashi, S Sumita, Y Nagamura, K Nishida, I Ishiguro. Chem Pharm Bull 41:539–543, 1993. JT Randolph, SJ Danishefsky. J Am Chem Soc 117:5693–5700, 1995. H Yamada, M Nishizawa. J Org Chem 60:386–397, 1995. B Asbell. The Pill: A Biography of the Drug That Changed the World. New York: Random House, 1995. (a) H Paulson. Angew Chem Chem Int Ed Engl 21:155–224, 1982. (b) RR Schmidt. Angew Chem Chem Int Ed Engl 25:212–235, 1986. (c) K Toshima, K Tatsuta. Chem Rev 93:1053–1531, 1993. (d) GJ Boons. Contem Org Synth 173–200, 1995. Y Mimaki, M Kuroda, A Kameyama, Y Sashida, T Hirano, K Oka, R Maekawa, T Wada, K Sugita, JA Beutler. Bioorg Med Chem Lett 7:633–636, 1997. C Guo, PL Fuchs. Tetrahedron Lett 39:1099–1102, 1998. SJ Deng, B Yu, Y Lou, YZ Hui. J Org Chem 64:202–208, 1999. BY Yu, Y Hirai, J Shoji, GJ Xu. Chem Pharm Bull 38:1931–1935, 1990. Q Xu, R Wang, BY Yu. J China Pharm Univ 24:98–101, 1993. (a) B Yu, JB Zhang, SF Lu, YZ Hui. Synlett 29–30, 1998. (b) B Yu, B Li, JB Zhang, YZ Hui. Tetrahedron Lett 39:4871–4874, 1998. (a) MZ Liu, B Yu, YZ Hui. Tetrahedron Lett 39:415–418, 1998. (b) MZ Liu, B Yu, XY Wu, YZ Hui, KP Fung. Carbohydr Res 329:745–754, 2000. SB Mahato, AN Ganguly, NP Sahu. Phytochemistry 21:959–978, 1982. (a) G Wulff, G Rohle. Angew Chem Int Ed Engl 13:157–216, 1974. (b) FJ Urban, BS Moore, R Breitenbach. Tetrahedron Lett 31:4421–4424, 1990. SJ Deng, B Yu, JM Xie, YZ Hui. J Org Chem 64:7265–7266, 1999. (a) C Li, B Yu, MZ Liu, YZ Hui. Carbohydr Res 306:189, 1998. (b) C Li, B Yu, YZ Hui. J Carbohydr Chem 18:1107–1120, 1999. (a) SJ Deng, B Yu, YZ Hui. Tetrahedron Lett 39:6511, 1998. (b) SJ Deng, B Yu, YZ Hui, H Yu, XW Han. Carbohydr Res 317:53–62, 1999. J Zhou. Pure Appl Chem 61:457–460, 1989. T Namba, XL Huang, YZ Shu, SL Huang, M Hattori, N Kakiuchi, Q Wang, GJ Xu. Planta Med 55:501–505, 1989. (a) T Nakamura, C Komori, YY Lee, F Hashimoto, S Yahara, T Nohara, A Ejima. Biol Pharm Bull 19:564–566, 1996. (b) CD Hufford, S Liu, AM Clark. J Nat Prod 51:94– 98, 1988. GR Pettit, DL Doubek, DL Herald, A Numata, C Takahashi, R Fujiki, T Miyamoto. J Nat Prod 54:1491–1502, 1991. (a) H Yamada, T Harada, T Takahashi. J Am Chem Soc 116:7919–7920, 1994. (b) H Yamada, T Harada, H Miyazaki, T Takahashi. Tetrahedron Lett 35:3979–3982, 1994. (a) B Yu, H Yu, YZ Hui, XW Han. Tetrahedron Lett 40:8591–8594, 1999. (b) H Yu, B Yu, XY Wu, YZ Hui, XW Han. J Chem Soc Perkin Trans 1 1445–1453, 2000. AM Rouhi. Chem Eng News 73 (Sept 11):28–35, 1995. JC McAuliffe, O Hindsgaul. Chemistry & Industry (London) 5:170–175, 1997.

7 Recent Developments in Technology for Glycosylation with Sialic Acid Randall L. Halcomb and Mark D. Chappell University of Colorado, Boulder, Colorado

I.

INTRODUCTION

Sialic acids are a class of nine-carbon monosaccharides found at the termini of oligosaccharides in many mammalian cellular systems [1]. This class is represented by the prototypical congener N-acetylneuraminic acid (NeuAc, 1, Scheme 1). These unique sugars are ubiquitous, and they are present as components of both glycolipids and glycoproteins. Some examples of sialic acid containing oligosaccharides are shown in Scheme 2. Sialic acids are found in variety of glycosidic linkages, some more common of which are ␣-2,3- or ␣-2,6-linkages to galactose residues. Additionally, they frequently exist as ␣-2,8-linked oligomers or polymers. The lack of efficient technology to accomplish glycosylations with sialic acid is one of the long-standing deficiencies in carbohydrate chemistry [2]. Owing to the central role of sialic acids in carbohydrate recognition events, the development of high-yielding and operable methods to synthesize sialic acid glycosides has been the subject of considerable research. The development of such methods allows the construction of complex sialic acid containing glycoconjugates for the investigation of their roles in biochemical and cellular processes. Most classical methods for synthesizing sialic acid glycosides are based on the reaction of an activated sialic acid such as 2 with an oligosaccharide glycosyl acceptor bearing a hydroxyl group nucleophile (Scheme 3) [2]. The leaving group is typically a halogen, such as a chloride or bromide, and the activator is typically a heavy metal salt [3,4]. A regioselective union of the two reacting partners of course relies on appropriate protecting group patterns for both the glycosyl donor and acceptor components. Many of the early methods are generally plagued with side reactions, low yields, and poor stereoselectivity. There are several reasons for these 177

178

Halcomb and Chappell

Scheme 1

shortcomings. The electron-withdrawing carboxylate at the anomeric carbon disfavors the formation of the oxocarbenium ion 6 (7) en route to glycoside bond formation (Scheme 4). This oxocarbenium ion intermediate is also somewhat sterically hindered, so attack of hydroxyl nucleophiles can be slow. Therefore an elimination to provide 9 can be a significant competing reaction pathway. Additionally, since the reacting carbon is insulated from the nearest stereogenic center by either O6 or C3, there is very little steric biasing of one face of this oxocarbenium ion over the other. As a consequence, the stereoselectivity commonly observed is less than desirable. Complicating the issue further, the anomeric effect causes ␣-glycosides to be thermodynamically less stable than the corresponding ␤ anomers. Heating a sialic acid ester in the presence of a primary alcohol and an acid catalyst results in exclusive formation of the ␤-glycoside (Scheme 5) [5]. As a result of these factors, glycosylation reactions with sialyl donors often provide low yields of the desired sialic acid ␣-glycosides. Thus the synthesis challenge of efficient glycosylation with sialic acid is a daunting one indeed. More recently, this long-standing problem has been revisited in a number of laboratories. This renewed focus was influenced significantly by the recent explosion of discoveries concerning the roles of oligosaccharides in biochemical processes of all types. Reviewed in this chapter are some selected recent advances in both chemical and enzymatic glycosylation with sialic acid. II.

IMPROVED VARIATIONS OF THE CLASSICAL METHODS

A.

Optimized Leaving Groups and Activation Methods

␣-Glycosides of sialic acid are classically prepared by the Koenigs–Knorr [3] or Helferich [4] method, each of which involves the activation of a glycosyl halide by a metal salt and subsequent displacement by an alcohol (Scheme 6). Typically a metal salt such as silver(I) (Koenigs–Knorr) or mercury(II) (Helferich) complexes to a ␤-glycosyl halide 5 to form an activated intermediate 12. This intermediate can then be directly displaced to give the product 8 or dissociate to form an ion pair 6. It has been hypothesized that an ion pair may block the top face of the molecule and force bottom face attack to also provide the desired ␣-glycoside 8. solvent polarity plays an important role in determining the mechanistic course of the reaction. One hypothesis is that more polar solvents favor the ion pair 6 because they can solvate the charged species, while the SN2 pathway is preferred in nonpolar solvents. Many variations of the metal promoters have been examined to probe the effect of the activator on the Koenigs–Knorr glycosylation yield and stereoselectivity. The

Glycosylation with Sialic Acid

179

Scheme 2

180

Scheme 3

Scheme 4

Scheme 5

Halcomb and Chappell

Glycosylation with Sialic Acid

Scheme 6

181

182

Halcomb and Chappell

promoters display a wide range of activity depending on the metal and counterion present. The order of reactivity among promoters with glycosyl halides was experimentally determined to be AgOTf > Ag2CO3 > silver salicylate > HgBr2 > Hg(CN)2 [6]. Generally, the more reactive silver promoters furnish higher stereoselectivities than the mercury salts, but give lower overall yields. In the majority of cases examined, the compound of choice was the most active promoter, AgOTf, or the least reactive, Hg(CN)2. Glycosyl halides are the typical sialyl donors in these processes. Sialyl chlorides are generally less reactive than bromides, but they are not as susceptible to decomposition and are sufficiently active under silver ion promotion to make them the most common halogenated glycosyl donor [2]. Sialyl fluorides are useful for glycosylation reactions but are usually activated under Lewis acidic conditions. An advantage of using fluoride glycosyl donors is their inherent preference to form ␤glycosides. Under Koenigs–Knorr conditions, ␤-sialyl chloride 13 leads to predominantly the ␣-glycoside 15 in a glycosylation with primary alcohol 14 [7a]. However, when the ␤-fluoride 16 is activated with BF3 ⭈ OEt2, the ␤-glycoside 17 is afforded in a 5:1 ␤ :␣ mixture of anomers (Scheme 7). Methods derived from the Koenigs–Knorr approach are useful for glycosylation reactions involving reactive alcohols, such as primary alcohols. Unfortunately,

Scheme 7

Glycosylation with Sialic Acid

183

when more hindered secondary alcohols are employed, the greater steric bulk of the nucleophile leads to lower stereoselectivities and yields. This is an important factor because many naturally occurring sialosides contain linkages to secondary alcohols. Sialyl bromides can be employed as more reactive donors in these instances. As a representative example, Paulsen and Von Reessen used a sialyl bromide to apply the Koenigs–Knorr method to the synthesis of an ␣-2,3-linked disaccharide (Scheme 8) [7b]. Glycosylation of the galactose acceptor 19 with sialyl bromide 18 led to the desired disaccharide 20, although with poor stereoselectivity (ca. 1:1 ␣ :␤). The low yields of Koenigs–Knorr glycosylations with secondary alcohols are partly due to elimination of the sialyl halide and formation of the 2,3-dehydro derivative 21. Many of the methods of activation using Koenigs–Knorr conditions require heterogeneous promoters and long reaction times at room temperature. In addition, the expense and toxicity of silver and mercury makes these methods impractical for large-scale syntheses. Consequently much research has focused on the development of glycosyl donors that could be activated at low temperature, under mild conditions, and without the need for heavy metal promoters. Sialic acid thioglycosides, although initially synthesized for use as biological probes, have been found to be excellent glycosyl donors. Hasegawa and coworkers utilized thioglycosides in the synthesis of several ␣-2,3- and ␣-2,6-linked disaccharides. The methylsulfide donor 22 was activated by using the thiophile dimethyl(methylthio)sulfonium triflate (DMTST) at ⫺15⬚C in the presence of either acceptor 23 or 25 to provide compounds 24 and 27, respectively, as exclusively the ␣-glycosides (Scheme 9) [8]. High stereoselectivities were observed for these glycosylations when the reactions were performed in acetonitrile, even though the thioglycoside starting material was a 1:1 mixture of anomers. This finding suggests that the stereoselectivity was not derived from the donor configuration as in Koenigs– Knorr glycosylations, but must depend on the reaction conditions. In an attempt to determine the source of the stereocontrol, Sina¨y and coworkers examined glycosylations of the xanthate donor 27 in polar and nonpolar solvent systems (Scheme 10) [9,10]. Glycosylation of 28 in acetonitrile under DMTST activation afforded the ␣-2,3-linked disaccharide 29 in 26% yield as a 7:1 ␣ :␤ mixture of diastereomers. However, the same reaction in CH2Cl2 gave exclusively the ␤glycoside 30 in 25% yield (Scheme 10). The major by-product in both reactions was the 2,3-dehydro NeuAc derivative. Sina¨y proposed that the glycosylation in CH2Cl2 probably involved an oxocarbenium ion 31 that underwent axial attack to form the ␤ product (Scheme 11). However in acetonitrile, formation of a ␤-nitrilium intermediate 32, which undergoes SN2 attack to afford the ␣-glycoside, may be the stereocontrolling element. The effect of acetonitrile has been studied in glucose and galactose systems, where the presence of axial nitrilium ions was supported by trapping and spectroscopic studies, both performed at low temperatures [10]. The Whitesides group achieved a more efficient synthesis of ␣-sialosides from sialyl xanthates by modifying the reaction conditions of Sina¨y’s system [11]. Previously, Lo¨nn and coworkers had demonstrated that lower temperatures and the use of milder promoter, methylsulfenyl triflate, increased the product yield and reaction stereoselectivity [12]. Whitesides and coworkers introduced an even milder promoter, phenylsulfenyl triflate (PST), which itself was prepared in situ from silver triflate and phenylsulfenyl chloride [11]. When sialylations of xanthate donor 27 were pro-

184

Halcomb and Chappell

Scheme 8

Glycosylation with Sialic Acid

Scheme 9

Scheme 10

185

186

Halcomb and Chappell

Scheme 11

moted by PST at ⫺70⬚C in a 2:1 ratio of CH3CN to CH2Cl2, isolated yields of 60– 80% and stereoselectivities of 19:1 ␣ :␤ were observed. A representative example is shown in Scheme 12. Interestingly, when the reaction was conducted under high dilution (0.01 M in donor) the stereoselectivity improved to better than 99:1 ␣ :␤, albeit at the expense of overall yield (52%). Regarding the reaction mechanism, Whitesides and coworkers also postulated that the presence of a ␤-nitrilium ion was the source of stereocontrol. In their proposed mechanism, the xanthate 27 initially reacts with the PST promoter to form 35 (Scheme 13). Subsequent loss of 36 leads to the oxocarbenim intermediate 31. Attack by acetonitrile is presumed to occur from the less hindered ␣ face and leads to the presumably thermodynamically more favorable equatorial nitrilium ion 37 because of the reverse anomeric effect [13,14]. However the equatorial nitrilium ion is believed to be in equilibrium with the more reactive ␤ species 32. Acetonitrile is then displaced from this species by the glycosyl acceptor in an SN2-like manner for form the ␣-glycosidic linkage in compound 38 The aforementioned experimental data do support this mechanism to some extent. First of all, the stereoselectivity is independent of the anomeric configuration of the starting material. A 1:1 mixture of xanthate anomers reacts to give predominantly the ␣-glycoside in acetonitrile, and the reaction by-product 36 was isolated and fully characterized. These data suggest that the xanthate is activated by the thiophilic promoter and glycosylation does not proceed through an SN2-type mechanism. Finally, Whitesides proposes that the gain in stereoselectivity that was observed by performing the reaction dilute in acetonitrile corresponds to an increase in the [MeCN]/[acceptor] ratio. The authors believe that a competition exists between acetonitrile and the acceptor alcohol for the oxocarbenium ion intermediate. When the concentration of the glycosyl acceptor is high enough, the alcohol reacts directly with oxocarbenium ion 32 to form the ␤-glycoside. However, an increase in the amount of acetonitrile favors nitrilium ion formation over direct glycosylation, and since nitrilium ion formation is the stereocontrolling element, higher ␣ :␤ ratios are observed. Takahashi and coworkers examined a novel approach to stereoselective sialylations by incorporating an auxiliary into the ester [15]. An electron-donating group with an appropriate linker to the carboxylate carbon was anticipated to stabilize an oxocarbenium ion through long-range participation. Activation of compound 39, where X is an electron-donating group, would produce oxocarbenium ion 40. This intermediate could be stabilized by X to provide an equilibrating mixture of ␤ and

Glycosylation with Sialic Acid

Scheme 12

187

188

Halcomb and Chappell

Scheme 13

␣ ‘‘onium’’ isomers, 41 and 42, respectively (Scheme 14). The authors predicted that the ␤-onium species 41 would preferentially undergo attack to afford the ␣glycoside 44. Preliminary experiments revealed that a sialyl donor containing a methyl sulfide and a 2-carbon spacer within the ester was the optimum substrate for these studies. This sialyl donor 48 was synthesized as an approximately 1:1 mixture of anomers and was activated with N-iodosuccinimide and trifluoromethanesulfonic acid. Glycosylations of compound 48 were compared with the methyl ester control 45 in various solvents to determine the extent of the enhancement of stereoselectivity (Scheme 15). Interestingly, glycosylations of 2-[2-(2-azidoethoxy)ethoxy]ethanol with donor 45 in acetonitrile or dichloromethane gave the same product ratios (entries 1 and 2), suggesting that for this system acetonitrile does not provide an increase in stereoselectivity. However, the tethered donor 48 predominantly favored formation of the ␣-glycosides (entrites 3, 4, and 5), with the solvent dimethoxyethane (DME) providing the highest ␣ :␤ ratios. Next, the sialyl donor 48 was utilized in the synthesis of ␣-2,6- and ␣-2,3-linked disaccharides under the optimal reaction conditions (entires 6 and 7). The selectivities observed were similar to those in the earlier

Glycosylation with Sialic Acid

189

Scheme 14

examples, although increased formation of products derived from elimination caused the reaction yields to drop significantly. A useful and powerful method that has been developed independently by Wong [16] and by Schmidt [17] and their colleagues utilizes sialyl phosphites as glycosyl donors. Phosphites are synthesized from the corresponding anomeric alcohols by reacting the latter with a phosphoramidite. This greatly shortens the reaction sequence needed to obtain the sialyl donors because many transformations that ordinarily would be needed to install a nonoxygen atom at the anomeric carbon are precluded, and several phosphoramidite reagents are readily available. An additional and significant advantage of these donors is that they are activated by many commonly used Lewis acids such as TMSOTf. Lewis acid activation is now a standard for glycoside synthesis, thus making these methods compatible with the most commonly used protecting group patterns and assembly strategies. Furthermore, sialyl phosphites are activated at low temperatures, and as a result generally provide products in high stereoselectivity. Some typical examples of the use of phosphites for

190

Halcomb and Chappell

Scheme 15

sialylation are the Wong lab’s synthesis 53 [18], an intermediate en route to the sialyl Lexis X antigen (Scheme 16) and the Schmidt lab’s synthesis of 56 (Scheme 17) [17a]. B.

Protection Strategies to Optimize Yields and Regiocontrol

The course and outcome of glycosylation reactions with sialic acid can be greatly influenced in some instances by seemingly insignificant modifications of the protecting groups on the glycosyl donor or acceptor. For example, simply employing a benzyl ester rather than the more commonly used methyl ester to protect the carboxylic acid of NeuAc, in combination with a sialyl chloride as donor using Koenigs–Knorr-type conditions, has profound consequences on the stereoselectivity of the glycosylation [19]. This fact was recorded by Ratcliff and coworkers, and has been widely utilized [19]. A representative example is shown in Scheme 18. The Boons laboratory has observed a dependence of the yield of glycosylation on the nature of the protecting group of the sialic acid C5 nitrogen [20]. When the nitrogen was engaged in an N,N-diacetylimide 60, yields significantly increased relative to the simple acetamide, the latter bearing only one acetyl group. This observation was utilized to efficiently construct NeuAc-␣-2,8-NeuAc linkages such as that in compound 62 (Scheme 19). The glycosylation of O8 of sialic acid by a sialyl

Glycosylation with Sialic Acid

Scheme 16

191

192

Scheme 17

Halcomb and Chappell

Scheme 18

Glycosylation with Sialic Acid

193

Scheme 19

donor is notoriously difficult to accomplish, and the donor 60 was nicely used to construct this linkage. A similar observation has been made by Hossain and Magnusson [21].

III.

USING DIRECTING GROUPS FOR STEREOCONTROL

Although several improved glycosyl donors have been developed to accomplish ‘‘direct’’ sialylations of acceptors, elimination reactions of the sialyl oxocarbenium ion intermediate often are significant competing nonproductive reaction pathways. Additionally, control of the stereochemistry at the NeuAc anomeric center is difficult and nontrivial. To alleviate these problems, several laboratories have investigated the use of stereodirecting auxiliaries at C3 of NeuAc. The postulate is that this group Y in 63 (Scheme 20) could, after activation, assist in stabilizing the positive charge of the oxocarbenium intermediate 64. In doing so, an intermediate such as 65 would be formed. Not only would this suppress elimination reactions, but one could imagine that the trajectory of attack of the glycosyl acceptor would be guided anti to the three-membered onium ion. As a result, the anomeric configuration might be assembled in a more predictable and ␣-selective fashion. Goto and coworkers, who were early contributors, synthesized compounds of this nature from the 2,3-dehydro sialic acid 21 [22]. The Goto group initially investigated hydroxyl groups at C3 as directors and used the reliable Koenigs–Knorr method for glycosylation (Scheme 21) [22]. The prototypical donor was the bromide 67. As a means of examining the versatility of this sialyl donor, the ␣-2,9 and ␣-2,8 NeuAc dimers were chosen as the synthetic targets. Treatment of 67 with silver triflate and a selectively protected sialic acid glycal 68 afforded 42% of the ␣-2,9 dimer 69 along with a 21% of the ␤ anomer. The secondary alcohol 70 was also glycosylated with 67 to give 26% of disaccharide 71. Although the yield of the glycosylation was rather modest, the stereoselectivity was good (3:1 ␣ :␤). The Goto lab later found that bromides at C3 could also be used as directing groups [23]. This prompted a great deal of research into more convenient C3 substituents that would provide higher stereocontrol.

194

Halcomb and Chappell

Scheme 20

Ogawa and coworkers introduced C3 phenylselenyl and phenylthio substituents as stereocontrolling auxiliaries [24,25]. These directing groups had already been developed in the context of stereoselective syntheses of 2-deoxyglycosides by several groups [26]. Phenylthio and phenylselenyl neighboring groups were anticipated to provide better charge stabilization than a C3 hydroxyl substituent because of the increased polarizability of sulfur and selenium [26]. Initial glycosylation attempts with the phenylselenyl substituent proceeded with good stereoselectivity, although yields were low owing to elimination of a cationic selenium species to provide the corresponding 2,3-dehydro NeuAc derivative as the major product of the reaction [24]. The less polarizable phenylsulfide auxiliary was found to circumvent his problem [25]. Mercury-promoted glycosylation of the acceptor 73 with the donor 72 afforded the GM3 precursor 74 in good yield with no trace of the ␤ isomer (Scheme 22). An additional benefit of using a sulfur auxiliary over a hydroxyl was its relative ease of removal in one step by a radical reduction. Treatment of 74 with Bu3SnH and AIBN provided the trisaccharide 75 in 75% yield. The phenylsulfide auxiliary was also used to synthesize an ␣-2,8 NeuAc dimer [25b]. The Ogawa group immobilized these types of sialyl donor on solid support resins and demonstrated their ability to act as donors in solid phase glycosylations [25c]. Thioglycoside donors combined with C3 phenylthio directing groups have been effectively employed by the Magnusson laboratory [27]. A disadvantage of auxiliary-directed glycosylation is that additional steps are required for the synthesis of the sialyl donor. To address this problem, the Whitesides group developed an expeditious route to a sialyl donor in two steps from the NeuAc glycal 21 (Scheme 23) [28]. Treatment of 21 with 2,4-dimethylbenzenesulfenyl chloride afforded the crystalline intermediate 76 in 85% yield. This compound was quan-

Glycosylation with Sialic Acid

Scheme 21

195

196

Halcomb and Chappell

Scheme 22

Glycosylation with Sialic Acid

197

Scheme 23

titatively converted to the thioglycoside 77 by treatment with sodium methanethiolate. Sialylations of 77 were achieved in good yield and stereoselectivity by using phenylsulfenyl triflate (PST) as a promoter and di-tert-butylpyridine (DTBP) as a proton scavenger. The most impressive application of this donor was for the synthesis of the ␣-2,3-linked disaccharide 78, which proceeded in 78% yield with no trace of the ␤ isomer (Scheme 24). It was also demonstrated that the auxiliary could be efficiently removed by reduction with Bu3SnH and AIBN in 94% yield. In 1998 the Schmidt group looked to improve the efficiency and simplicity of sialyl glycosylations through the development of a phosphite donor containing a directing auxiliary at C3 [29]. The primary objective was to develop an auxiliary that could be more easily introduced and removed and would more effectively direct the stereochemistry while minimizing side reactions. It was postulated that a better auxiliary might be one that could form a cyclic five-membered onium intermediate rather than a three-membered one. Compound 79 was found to meet the requirements. Acid-catalyzed activation of sialyl phosphite 79 should lead to oxocarbenium ion 80 without the need for stoichiometric promoters (Scheme 25). The intermediate oxocarbenium ion should readily undergo attack by the thionobenzoate auxiliary to form the intermediate 81. Incorporation of the thionobenzoate auxiliary, which provides anchimeric assistance through a larger, more favorable five-membered ring, should provide higher stereocontrol than the three-membered episulfonium ions. Glycosylation should then proceed from the ␣ face to provide the desired sialoside 82. Following this reaction, the thionobenzoate auxiliary should be easily removed by treatment with Bu3SnH/AIBN. Many of the earlier syntheses of NeuAc derivatives containing auxiliaries were long and tedious because it was so difficult to obtain the correct stereochemistry at

198

Scheme 24

Scheme 25

Halcomb and Chappell

Glycosylation with Sialic Acid

199

Scheme 26

C3. However, Schmidt and coworkers discovered that the reaction of glycal 21 with OsO4 led to stereoselective dihydroxylation from the ␤ face, providing diol 83 as the only product (Scheme 26). Subsequent regioselective introduction of the thiobenzoyl auxiliary was accomplished by treatment with N,N-dimethyl-␣-chlorobenzimidium chloride followed by dihydrogen sulfide. Compound 84 was obtained as the product of this sequence in 92% yield. Finally, the alcohol was converted to the glycosyl phosphite 79 in a straightforward manner. The utility of this donor was demonstrated in the synthesis of an ␣-2,3-disaccharide and an ␣-2,8-linked NeuAc dimer. Glycosylation of the protected lactose derivative 73 with donor 79 afforded an 88% yield of the protected GM3 analog 85, with no trace of the ␤-glycoside (Scheme 27). Likewise, glycosylation of glycal 86 provided an unprecedented 83% yield of the ␣-2,8 NeuAc dimer 87, once again with no trace of the undesired ␤ diastereomer. In both cases the auxiliary was removed with Bu3SnH/AIBN in good yield. IV.

USING SIALYLTRANSFERASES FOR THE SYNTHESIS OF SIALOSIDES

Enzyme-mediated glycosylations are powerful methods for the synthesis of complex carbohydrates, including those that contain sialic acid [30,31]. In mammalian systems, Leloir pathway glycosyltransferases are responsible for the biosynthesis of most glycoconjugates [32]. Sialyltransferases are a subset of the glycosyltransferases that transfer the NeuAc component of cytidinyl-5⬘-monophospho-␤-N-acetylneuraminic acid (88, CMP-NeuAc) to acceptor hydroxyl groups with inversion of configuration at the anomeric center (Scheme 28) [30,31]. Transferase-mediated sialylations are not burdened by some of the pitfalls of chemical glycosylations, such as un-

200

Halcomb and Chappell

Scheme 27

Glycosylation with Sialic Acid

201

Scheme 28

wanted reaction pathways and the need for tedious protecting group schemes or stereocontrolling auxiliaries. In addition, sialyltransferases have been shown to tolerate several acceptor modifications [33], although less is known about their donor specificity because the requisite CMP-NeuAc sugar donors are so difficult to synthesize. Kinetic isotope effect studies with a particular transferase, namely, rat liver ␣2,6-sialyltransferase, have shown that there is significant charge buildup at the anomeric center in the transition state [34]. This suggests that the glycosylation proceeds through an SN1-like mechanism. CMP-NeuAc appears to be the common donor for all the sialyltransferases. However, each enzyme varies in regioselectivity and to some degree in acceptor specificity. Under the premise that an enzyme exists for the formation of each natural sialoside, there should be at least nine sialyltransferases responsible for the synthesis of the following disaccharides: NeuAc-␣-2,6-Gal, NeuAc-␣-2,3-Gal, NeuAc-␣-2,6-GalNac, NeuAc-␣-2,4-Gal, NeuAc-␣-2,4-GlcNac, NeuAc-␣-2,6-GlcNAc, NeuAc-␣-2,6-Man, NeuAc-␣-2,8-NeuAc, and NeuAc-␣-2,9NeuAc. However, only a small subset of the known and postulated sialyltransferases has been isolated and used for synthesis purposes. The sialyltransferases are membrane-bound proteins located in the endoplasmic reticulum (ER) and in the Golgi apparatus. Information about their sequence homology is limited, but they do appear to share a common topography [35]. A catalytic domain resides at the C-terminus followed by an N-terminal segment that anchors the enzyme into the ER or Golgi membrane. Soluble, catalytically active sialyltransferases that lack the anchor segment have been isolated from milk, serum, and other body fluids, suggesting that this N-terminal anchor is not necessary for the enzyme to retain catalytic activity. However, the ability to obtain from natural sources quantities of most sialyltransferases that would be needed for synthesis applications is hampered by low tissue concentrations and difficult purifications. The genes for members of some of the most common classes of sialyltransferases have been cloned and expressed [33a]. Expressing and isolating membranebound enzymes in high catalytic activity can be difficult; however, Paulson and coworkers replaced the anchor segment with a cleavable peptide in the expression of Gal-␤-1,4-GlcNAc ␣-2,6-sialyltransferase [35a,b]. This allowed the enzymes to be secreted from the cell into the medium, thus simplifying the process of isolation. This technology has been used to express several sialyltransferases in quantities

202

Halcomb and Chappell

Scheme 29

suitable for synthesis. Gaining sialyltransferase accessibility is becoming a less serious problem as these enzymes become commercially available. A.

Synthesis of CMP-NeuAc and Related Derivatives

The primary method for the synthesis of CMP-NeuAc 88 is through the use of CMPNeuAc synthetase, an enzyme that catalyzes the condensation of cytidine triphosphate (CTP) with sialic acid to produce CMP-NeuAc (Scheme 29) [30]. CMP-NeuAc synthetase has been cloned from microbial sources and has been isolated from mammalian tissues [36,37]. The substrate specificity of each synthetase has been studied to some degree [36,37]. The mammalian version accepts C9 and some C8 modifications of NeuAc, as well as variations at the C5 position, such as replacement of the acetamide with OH (KDN) or hydroxylation of the acetamide (NeuAc). The microbial CMP-NeuAc synthetase has a high activity for C9-modified sialic acids, but does not tolerate alterations at the C5 position. Unfortunately, the microbial version is the more readily available enzyme, thus limiting the variety of analogs that can be prepared in this manner. Scheme 30 gives a representative sample of some important C9-derivatized CMP-NeuAc analogs [37]. Modifications at the C5 position are somewhat limited by the specificity of CMP-NeuAc synthetase for an amide at this position. Nevertheless, several analogs were prepared that incorporate sterically and electronically diverse substituents at C5 (Scheme 31) [37].

Scheme 30

Glycosylation with Sialic Acid

203

Scheme 31

Wong and coworkers first attempted the nonenzymatic synthesis of CMPNeuAc by employing phosphoramidite chemistry in a key step involving the ligation of a sialyl phosphoramidite and a selectively protected cytidine analog (Scheme 32) [38]. Treatment of NeuAc derivative 99 with 2-cyanoethyl chlorophosphoramidite resulted in the formation of the ␤-sialyl phosphoramidite 100 in 89% yield. The sialyl donor was then coupled to the protected cytidine 101 under promotion by 1H tetrazole to afford the intermediate phosphite, which was immediately oxidized with tert-butyl hydroperoxide to provide the protected CMP-NeuAc 102. Deprotection of this intermediate was not reported; however, the Wong group demonstrated that the CMP-NeuAc core structure could be successfully synthesized utilizing phosphoramidite chemistry.

Scheme 32

204

Halcomb and Chappell

The Schmidt group utilized a sialyl phosphite in a very different synthesis strategy (Scheme 33) [39]. Upon treatment of sialyl donor 54 with cytidine phosphoric acid 103, a phosphite–phosphate exchange reaction occurred to give compound 104 exclusively as the ␤ isomer. Deacylation by treatment with sodium methoxide followed by ester saponification through the addition of water provided CMP-NeuAc 88. This method circumvented the need for an oxidation step or phosphorus deprotection. This method was also applied to the synthesis of another naturally occurring CMP-NeuAc derivative 105 [40]. Hata and coworkers took a different approach to the ligation by utilizing cytidine phosphoramidite 107 and the tertiary anomeric alcohol of 106 as coupling partners (Scheme 34) [41]. The coupling reaction successfully provided a phosphite intermediate, which was subsequently oxidized with tert-butyl hydroperoxide to provide the trialkyl phosphate 108. Attempts to purify the phosphotriester failed, but the production of 108 was supported by 31P NMR spectroscopy of the crude reaction mixture. The protecting groups were then removed under mild conditions by treatment with tetrakis(triphenylphosphine)palladium(0) in the presence of the allyl scavenger n-butylammonium bicarbonate. Following purification by size exclusion chromatography, CMP-NeuAc 88 was obtained in acceptable overall yield (25% for three steps). In a related synthesis of CMP-NeuAc, Kajihara and coworkers also used a cytidine phosphoramidite in the coupling reaction, although with different protecting group patterns on the coupling partners (Scheme 35) [42]. The reaction of 99 with phosphoramidite 109 provided the phosphite 110 as a mixture of phosphorus diastereomers. Phosphite oxidation and subsequent treatment with methoxide led to decomposition. As an alternative, the cyanoethyl protecting group was first removed with DBU; then complete deprotection by treatment with sodium methoxide and sodium hydroxide afforded CMP-NeuAc 88 in good yield. This stragety was also applied to the synthesis of CMP-NeuAc-␣-2,8-NeuAc 111 [43]. Halcomb and Chappell developed a route to CMP-NeuAc 88 that promises to be general for the synthesis of virtually any derivative thereof [44,45]. The route (Scheme 36) utilizes a condensation of sialic acid derivative 99 with the phosphoramidite 112 to afford the phosphite 113 in 62% yield. Oxidation of the phosphite provided the phosphotriester 114 [46], which was taken directly to the next transformation without purification (owing to its instability to chromatography). Deallylation of the phosphate gave compound 115 (61% for two steps), which was stable to silica gel chromatography. Compound 115 was deacylated with methoxide, and its methyl ester was subsequently saponified with NaOH to provide CMP-NeuAc 88. The derivatives shown in Scheme 37 were synthesized according to this protocol and were investigated as substrates for sialyltransferases (see below). The synthesis of a CMP-NeuAc derivative that was bound to a solid support through the 9-position of the sialic acid has been reported by the Kajihara group [47]. This derivative is quite useful in that it can be utilized to immobilize glycoproteins onto a solid support by transferring the sialic acid to the terminus of the carbohydrate chain of the glycoprotein. B.

Synthesis of Wild-Type and Mutant Sialosides

The Brossmer and Paulson groups have studied the sialyltransferase donor specificity with a series of C5 and C9 conjugates, all of which were prepared through the CMP-

Glycosylation with Sialic Acid

205

Scheme 33

206

Halcomb and Chappell

Scheme 34

NeuAc synthetase route [37]. The relative rates of these C9-modified CMP-NeuAc derivatives shown in Scheme 30 were compared against the natural donor CMPNeuAc in sialyltransferase assays that utilized enzymes from different sources with their appropriate natural acceptors. The rat liver ␣-2,6-sialyltransferase tolerated a wide range of functional groups without significant decreases in the relative rates. Synthetically useful relative rates were observed for most of the CMP-NeuAc analogs with porcine sialyltransferase and rat liver ␣-2,3-sialyltransferase. The exception was the 9-amino analog, which was a poor substrate for both enzymes. Overall, these assays demonstrated that a wide variety of modifications at the C9 position are tolerated by these sialyltransferases. The relative rates of sialylation for the C5-modified CMP-NeuAc derivatives shown in Scheme 31 were also determined [37]. All the C5-modified sialyl donors were accepted by the transferase enzymes, but the variants containing free amines were poor substrates for all the sialyltransferases examined. Generally, higher relative

Glycosylation with Sialic Acid

Scheme 35

207

208

Halcomb and Chappell

Scheme 36

Glycosylation with Sialic Acid

Scheme 37

209

210

Halcomb and Chappell

rates were observed for both ␣-2,6-sialyltransferases over the porcine ␣-2,3-sialyltransferase, although all the rates are high enough to be synthetically useful. This study has shown that numerous C5-modified CMP-NeuAc sialyl donors can be successfully utilized in the synthesis of modified glycoproteins. The sialyltransferases investigated were found to accept many variations at C5 and C9 of the sialic acid moiety. Halcomb and Chappell confirmed this fact with studies of sialyltransferases from different sources [45]. The C9- and C5-modified substrates in Scheme 37 were good substrates. However no activity was detected when compounds 111 and 117 were assayed. In addition to these examples, other CMP-NeuAc analogs that have been prepared and successfully transferred include C9 fluorescent compounds, C9 thioacetyl, C5 thioacetyl, and C4 dexoy [37]. A particularly useful modification introduced by the Wong lab incorporates a mercury atom at C9 (Scheme 38) [48]. Transfer of this sialic acid to glycoproteins could greatly aid in X-ray crystallographic analysis of these biomolecules. The field of solid phase synthesis has also benefited from enzyme technology. Several reports have described the use of sialyltransferases in the solid-supported synthesis of oligosaccharides that bear sialic acid [49]. Additionally, metabolic pathways have been harnessed by Bertozzi and coworkers to synthesize glycoconjugates on cell surfaces that bear modified sialic acids [50]. C.

Cofactor Regeneration

Enzymatic methods are efficient for the assembly of complex oligosaccharides on a small scale, although the high cost of sugar nucleotides and problems with product inhibition detract from this approach in large-scale reactions. The synthesis of CMPNeuAc by CMP-NeuAc synthetase requires stoichiometric amounts of expensive reagents, such as CTP, while the CMP by-product of the sialyltransferase reaction is an enzyme inhibitor. Ichikawa and Wong used the concept of in situ cofactor regeneration to alleviate these problems and increase the efficiency of sialylation (Scheme 39) [51]. The cascade begins with stoichiometric amounts of phosphoenolpyruvate (PEP), ␤-allyl-N-acetyl lactosamine 120, NeuAc 1, and catalytic quantities of ATP and CMP. Initially, CMP is converted to CDP by nucleoside monophosphate kinase (NMK) in the presence of ATP. The CDP produced reacts with PEP under pyruvate kinase (PK) catalysis to form CTP. Next, CMP-NeuAc synthetase catalyzes the in situ formation of the sialyl donor from NeuAc and CTP. The pyrophosphate byproduct is decomposed to inorganic phosphate by inorganic pyrophosphatase (PPase). Subsequently, the ␣-2,6-sialyltransferase accomplishes the sialyation of the lactosamine acceptor 120 and produces the transferase inhibitor CMP as a by-product. The CMP concentrations are kept low by conversion to CDP, and in so doing the problem of product inhibition is minimized. The cycle afforded 21% of the sialylated trisaccharide 121, which is remarkable considering the complexity of the system and number of synthetic steps that can be avoided. V.

USING SIALIDASES FOR THE SYNTHESIS OF SIALOSIDES

A.

Sialylhydrolases

Sialidases are another class of enzymes that have been used in the synthesis of sialylated oligosaccharides. In addition to hydrolysis activity, sialidases are capable

Glycosylation with Sialic Acid

Scheme 38

211

212

Halcomb and Chappell

Scheme 39

Glycosylation with Sialic Acid

213

of sialoside transfer to alcohol donors. Using the 4-nitrophenylglycoside 122 as a sialyl donor, sialylidases can catalyze a transsialylation to an alcohol acceptor to generate the product 89, or hydrolyze to NeuAc 1 (Scheme 40). To complicate matters further, the product 89 may also undergo sialidase-catalyzed hydrolysis to form NeuAc. So if any product is to be isolated, the rate of sialyl transfer must be much faster than either hydrolysis rate. Thiem and Sauerbrei examined this concept to determine whether various sialidases could be used synthetically [52]. The rate of condensation, or reverse hydrolysis, was found to be negligible. However, the product hydrolysis rate was competitive with the rate of transsialylation to an alcohol acceptor. In an attempt to minimize product loss, reactions were stopped after 65–75% of the starting material had been consumed. Interestingly, a mixture of ␣-2,3 and ␣-2,6 regioisomers was obtained for reactions with an immobilized Vibrio cholerae sialidase (Scheme 41). In all cases, the ␣-2,6 isomer predominated, probably because of a combination of faster hydrolysis of the ␣-2,3 products (123 → 1 ⫹ 125) and lesser steric hinderance of the primary alcohol. Variations in the donor/acceptor ratio (1:7 optimized) had an effect on both reaction yield and regioselectivity, although most of the examples afforded only 2–3:1 ␣-2,6/␣-2,3 product ratios in 14–20% overall yield. Ajisaka et al. examined different sialidase sources and found that Newcastle disease virus (NDV) sialidase afforded predominantly the ␣-2,3 regioisomers, while Arthrobacter ureafaciens and Clostridium perfringens sialidases, in addition to the Vibrio cholerae sialidase examined by Thiem, favored the ␣-2,6-linked products [53]. Unfortunately, the reaction yields did not improve for the new enzymes, varying from 0.8% to 3.6% isolated yield. In the case of NDV sialidase, the high selectivity for ␣-2,3-sialosides stemmed from a large ␣-2,6/␣-2,3 hydrolysis ratio. Hydrolysis of the ␣-2,6 products was found to be 28 times faster than the ␣-2,3 isomers. Inter-

Scheme 40

214

Halcomb and Chappell

Scheme 41

estingly, the ␣-2,6 preference of the other three enzymes was not correlated to product hydrolysis rates. B.

trans-Sialidases

The Trypanosoma cruzi trans-sialidase catalyzes the reversible transfer of NeuAc from a NeuAc-␣-2,3-Gal-␤-OR1 sequence to an acceptor bearing the Gal-␤-OR2 motif (Scheme 42) [54]. The enzyme is a particularly useful sialidase because it has very little hydrolytic activity and tends to almost exclusively catalyze transsialylation to a galactose. However a major drawback to this method is that to drive the glycosylation to completion, there is a need for large quantities of complex ␣-2,3-linked sialyl donors, which are generally difficult to obtain from natural sources. Other natural donors with ␣-2,6- or ␣-2,8-linked sialic acids have been examined but were discovered to be poor sialyl donors for ␣-2,3-sialylations catalyzed by T. cruzi transsialidase [55]. Simple aryl ␣-sialosides, such as the 4-nitrophenyl glycoside 122 and methylumbelliferone glycoside 130 (Scheme 43), have been found to be excellent

Glycosylation with Sialic Acid

215

Scheme 42

substrates because of the irreversibility of the sialyl transfer, and these have become the most utilized sialyl donors for trans-sialidase-catalyzed glycosylations. The donor specificity of T. cruzi trans-sialidase was examined with various side chain modified NeuAc glycosides. Vandekerckhove and coworkers found that methoxy or deoxy modifications to C9 of a NeuAc-␣-2,3-Gal-␤-OR donor did not affect trans-sialidase activity, although the same modifications at C4, C7, or C8 completely prevented transfer of the sialic acid. In addition, the C4- and C8-modified compounds were found to be mild inhibitors of the enzyme [56]. Lee and Lee extended these studies to more drastically modified NeuAc aryl glycosides [57]. The triol side chain of the methylumbelliferone-␣-ketoside 130 was cleaved with periodate to afford the C7 aldehyde, which underwent reductive amination with different amines to provide several novel sialyl donors. Three of these derivatized sialic acids (those of 132, 134, and 136) were successfully transferred to a lactose acceptor on analytical scales (50 nmol), albeit in much lower yield than NeuAc from aryl glycoside 130 (Scheme 44). NeuAc analogs containing longer alkyl chain and terminal amine substituents were found not to be substrates for the trans-sialidase. Nonetheless, these studies helped to define the substrate requirements for trans-sialidase acceptance, and the latter work provides a novel method for the synthesis of chromophore labeled sialylated oligosaccharides. Ito and Paulson designed a cofactor regeneration system that overcomes several limitations of trans-sialidase-catalyzed sialylations (Scheme 45) [58]. The trans-sialidase is used in conjunction with ␣-2,3-sialyltransferase to effectively broaden the

Scheme 43

216

Halcomb and Chappell

Scheme 44

Glycosylation with Sialic Acid

Scheme 45

217

218

Halcomb and Chappell

substrate specificity of the sialyltransferase. The ␤-trimethylsilyl galactoside 139 is known to be a poor substrate for ␣-2,3-sialyltransferase; however, it is readily accepted by T. cruzi trans-sialidase. The sialylated oligosaccharide 138 was synthesized by the CMP-NeuAc regeneration system developed by Ichikawa and Wong [51]. Then 139 underwent a transsialylation with 138 to provide the desired GM4 precursor 141. The regeneration of 138 from the by-product 140 by sialyltransferase drives the equilibrium toward the product 141. This example demonstrates that trans-sialidases have some synthetic utility on their own but are much more general tools for oligosaccharide synthesis when coupled with sialyltransferases.

REFERENCES 1.

2.

3. 4. 5. 6. 7. 8. 9. 10.

11. 12.

13. 14. 15. 16. 17. 18. 19. 20. 21.

(a) Varki, A. Glycobiology 1993, 3, 97. (b) Varki, A. FASEB J. 1997, 248. (c) Schauer, R.; Kamerling, J. P. In Glycoproteins II, Montreuil, J., Vliegenhart, J. F. G., Schachter, H., eds. Elsevier: Amsterdam, 1997, pp. 243–402. (a) Okamoto, K; Goto, T. Tetrahedron 1990, 46, 5835. (b) DeNinno, M. Synthesis 1991, 583. (c) Hasegawa, A.; Kiso, M. In Carbohydrates: Synthetic Methods and Applications in Medicinal Chemistry, Ogura, H., Hasegawa, A., Suami, T., eds. VCH: New York, 1992, pp. 243–266. (a) Koenigs, W.; Knorr, E. Chem. Ber. 1901, 34, 957. (b) Igarashi, K. Adv. Carbohydr. Chem. Biochem. 1977, 34, 243. Helferich, B.; Zirner, J. Chem. Ber. 1962, 95 2604. Julina, R.; Muller, I.; Vasella, A.: Wyler, R. Carbohydr. Res. 1987, 164, 415. Binkley, R. W. Modern Carbohydrate Chemistry. Marcel Dekker: New York, 1988, pp. 297–322. (a) Kunz, H.; Waldmann, H. J. Chem. Soc., Chem. Commun. 1985, 638. (b) Paulsen, H.; Von Reessen, U. Carbohydr. Res. 1988, 175, 283. Hasegawa, A.; Ohki, H.; Nagahama, T.; Ishida, H.; Kiso, M. Carbohydr. Res. 1991, 212, 277. Pougny, J.-R.; Sina¨y, P. Tetrahedron Lett. 1976, 45, 4073. (a) Ratcliffe, A. J.; Fraser-Reid, B. J. Chem. Soc., Perkin Trans. I 1990, 747. (b) Braccini, I.; Derouet, C.; Esnault, J.; Herve du Penhoat, C.; Mallet, J.-M.; Michon, V.; Sina¨y, P. Carbohydr. Res. 1993, 246, 23. Martichonok, V.; Whitesides, G. M. J. Org. Chem. 1996, 61, 1702. (a) Birberg, W.; Lo¨nn, H. Tetrahedron Lett. 1991, 32, 7453. (b) Birberg, W.; Lo¨nn, H. Tetrahedron Lett. 1991, 32, 7457. (c) Lo¨nn, H.; Stenvall, K. Tetrahedron Lett. 1992, 33, 115. Lemieux, R. U.; Morgan, A. R. Can. J. Chem. 1865, 43, 2214. Perrin, C. L. Tetrahedron 1995, 51, 11901. Takahashi, T.; Tsukamoto, H.; Yamada, H. Tetrahedron Lett. 1997, 38, 8223. Sim, M. M.; Kondo, H.; Wong, C.-H. J. Am. Chem. Soc. 1993, 115, 2260. (a) Martin, T.; Schmidt, R. R. Tetrahedron Lett. 1992, 33, 6123. (b) Lassaletta, J. J.; Schmidt, R. R. Tetrahedron Lett. 1995, 36, 4209. Kondo, H.; Aoki, S.; Ichikawa, Y.; Halcomb, R. L.; Ritzen, H.; Wong, C.-H. J. Org. Chem. 1994, 59, 864. Ratcliffe, M. R.; Venot, A. P.; Abbas, Z. S. European Patent Application EP 319,253 (Chem. Abstr. 1990, 112, 175281a). (a) Demchenko, A. V.; Boons, G.-J. Tetrahedron Lett. 1998, 39, 3065. (b) Demchenko, A. V.; Boons, G.-J. Chem. Eur. J. 1999, 5, 1278. Hossain, N.; Magnusson, G. Tetrahedron Lett. 1999, 40, 2217.

Glycosylation with Sialic Acid 22.

23. 24. 25.

26.

27. 28. 29. 30.

31. 32. 33.

34. 35.

36. 37.

38. 39. 40.

219

(a) Okamoto, K.; Kondo, T.; Goto, T. Bull. Chem. Soc. Jpn. 1987, 60, 637. (b) Okamoto, K.; Kondo, T.; Goto, T. Tetrahedron Lett. 1986, 27, 5233. (c) Okamoto, K.; Kondo, T.; Goto, T. Tetrahedron Lett. 1986, 27, 5229. (d) Okamoto, K.; Kondo, T.; Goto, T. Tetrahedron Lett. 1987, 28, 5919. Oamoto, K.; Kondo, T.; Goto, T. Tetrahedron 1987, 43, 5909. Ito, Y.; Ogawa, T. Tetrahedron Lett. 1987, 28, 2723. (a) Ito, Y.; Ogawa, T. Tetrahedron Lett. 1988, 29, 3987. (b) Ito, Y.; Nunomura, N.; Shibayama, S.; Ogawa, T. J. Org. Chem. 1992, 57, 1821. (c) Konov, L. O.; Ito, Y.; Ogawa, T. Tetrahedron Lett. 1997, 38, 1599. (a) Nicolaou, K. C.; Laddowahetly, T.; Randall, J. L.; Chucholowski, A. J. Am. Chem. Soc. 1986, 108, 2466. (c) Jaurand, G.; Beau, J.-M.; Sina¨y, P. J. Chem. Soc., Chem. Commun. 1981, 572. (d) Ercegovic, T.; Magnusson, G. J. Org. Chem. 1995, 60, 3378. (a) Ercegovic, T.; Magnusson, G. J. Chem. Soc., Chem. Commun. 1994, 831. (b) Ercegovic, T.; Magnusson, G. J. Org. Chem. 1995, 60, 3378. Martichonok, V.; Whitesides, G. M. J. Am. Chem. Soc. 1996, 118, 8187. Castro-Palomino, J. C.; Tsvetkov, Y. E.; Schmidt, R. R. J. Am. Chem. Soc. 1998, 120, 5434. (a) Wong, C.-H.; Halcomb, R. L.; Ichikawa, Y.; Kajimoto, T. Angew. Chem., Int. Ed. Engl. 1995, 34, 412. (b) Wong, C.-H.; Halcomb, R. L.; Ichikawa, Y.; Kajimoto, T. Angew. Chem., Int. Ed. Engl. 1995, 34, 521. Wong, C.-H.; Whitesides, G. M. Enzymes in Synthetic Organic Chemistry. Elsevier: New York, 1994. Leloir, L. F. Science 1971, 172, 1299. (a) Ito, Y.; Gaudino, J. J.; Paulson, J. C. Pure Appl. Chem. 1993, 65, 753. (b) Sabesan, S.; Paulson, J. C. J. Am. Chem. Soc. 1986, 108, 2068. (c) Ito, Y.; Paulson, J. C. J. Am. Chem. Soc. 1993, 115, 1603. (d) Kren, V.; Thiem, J. Angew. Chem., Int. Ed. Engl. 1995, 34, 893. (e) Palcic, M. M.; Venot, A. P.; Ratcliffe, R. M.; Hindsgaul, O. Carbohydr. Res. 1989, 190, 1. (f ) Auge´, C.; Fernandez-Fernandez, R.; Gautheron, C. Carbohydr. Res. 1990, 200, 257. Lubineau, A.; Basset-Carpentier, K.; Auge´, C. Carbohydr. Res. 1997, 300, 161. (g) Lubineau, A.; Auge´, C.; Francois, P. Carbohydr. Res. 1992, 228, 137. (h) Liu, K. K.-C.; Danishefsky, S. J. J. Am. Chem. Soc. 1993, 115, 4933. Bruner, M.; Horenstein, B. A. Biochemistry 1998, 37, 289. (a) Colley, K. J.; Lee, E. U.; Adler, B.; Browne, J. K.; Paulson, J. C. J. Biol. Chem. 1989, 264, 17619. (b) Paulson, J. C.; Colley, K. J. J. Biol. Chem. 1989, 264, 17615. (c) Kornfeld, R.; Kornfeld, S. Annu. Rev. Biochem. 1985, 54, 631. Liu, J. L.-C.; Shen, G.-J.; Ichikawa, Y.; Rutan, J. F.; Zapata, G.; Vann, W. F.; Wong, C.-H. J. Am. Chem. Soc. 1992, 114, 3901. (a) Gross, H. J.; Brossmer, R. Glyconjugate J. 1995, 12, 739. (b) Brossmer, R.; Gross, H. J. Methods Enzymol. 1994, 247, 177. (c) Brossmer, R.; Gross, H. J. Methods Enzymol. 1994, 247, 153. (d) Gross, H. J. Eur. J. Biochem. 1992, 203, 269. (e) Gross, H. J.; Rose, U.; Krause, J. M.; Paulson, J. C.; Schmid, K.; Feeney, R. E.; Brossmer, R. Biochemistry 1989, 28, 7386. (f ) Gross, H. J.; Brossmer, R. Eur. J. Biochem. 1988, 177, 583. (g) Gross, H. J.; Buensch, A.; Paulson, J. C.; Brossmer, R. Eur. J. Biochem. 1987, 168, 595. (h) Paulson, J. C.; Higa, H. H. J. Biol. Chem. 1985, 260, 8838. (i) Ichikawa, Y. In Glycopeptides and Related Compounds, Large, D. G.; Warren, C. D., eds. Marcel Dekker: New York, 1997, Chapter 3, pp. 79–205. ( j) Gross, H. J.; Brossmer, R. Glyconjugate J. 1988, 5, 411. (k) Rogers, G. N.; Herrier, G.; Paulson, J. C.; Klenk, H.-D. J. Biol. Chem. 1986, 261, 5947. Kondo, H.; Ichikawa, Y.; Wong, C.-H. J. Am. Chem. Soc. 1992, 114, 8748. Martin, T. J.; Schmidt, R. R. Tetrahedron Lett. 1993, 34, 1865. Yoshino, T.; Schmidt, R. R. Carbohydr. Lett. 1995, 1, 329.

220 41. 42. 43. 44. 45. 46. 47. 48. 49.

50.

51. 52. 53. 54.

55. 56. 57. 58.

Halcomb and Chappell Makino, S.; Ueno, Y.; Ishikawa, M.; Hayakawa, Y.; Hata, T. Tetrahedron Lett. 1993, 34, 2775. Kajihara, Y.; Koseki, K.; Ebata, T.; Kodama, H.; Matsushita, H.; Hashimoto, H. Carbohydr. Res. 1994, 264, C1. Kajihara, Y.; Ebata, T.; Koseki, K.; Kodama, J.; Matsushita, H.; Hashimoto, H. J. Org. Chem. 1995, 60, 5732. Chappell, M. D.; Halcomb, R. L. Tetrahedron 1997, 53, 11109. Chappell, M. D.; Halcomb, R. L. J. Am. Chem. Soc. 1997, 119, 3393. Chappell, M. D.; Halcomb, R. L. Tetrahedron Lett. 1999, 40, 1. Kajihara, Y.; Ebata, T.; Kodama, H. Angew. Chem., Int. Ed. Engl. 1998, 37, 3166. Martin, R.; Witte, K. L.; Wong, C.-H. Bioorg. Med. Chem. 1998, 6, 1283. (a) Halcomb, R. L.; Huang, H.; Wong, C.-H. J. Am. Chem. Soc. 1994, 116, 11315. (b) Schuster, M.; Wang, P.; Paulson, J. C.; Wong, C.-H. J. Am. Chem. Soc. 1994, 116, 1135. (c) Yamada, K.; Nishimura, S.-I. Tetrahedron Lett. 1995, 36, 9493. (d) Blixt, T.; Norberg, T. J. Carbohydr. Chem. 1997, 16, 143. (e) Blixt, O.; Norberg, T. J. Org. Chem. 1998, 63, 2705. (f ) Osborn, H. M. I.; Khan, T. H. Tetrahedron 1999, 55, 1807. (g) Kanemitsu, T.; Kanie, O.; Wong, C.-H. Angew. Chem., Int. Ed. Engl. 1999, 37, 3415. (a) Yarema, K. J.; Mahal, L. K.; Bruehl, R. E.; Rodriguez, E. C.; Bertozzi, C. R. J. Biol. Chem. 1998, 273, 31168. (b) Lemieux, G. A.; Yarema, K. J.; Jacobs, C. L.; Bertozzi, C. R. J. Am. Chem. Soc. 1999, 121, 4278. (a) Ichikawa, Y.; Shen, G.-J.; Wong, C.-H. J. Am. Chem. Soc. 1991, 113, 4698. (b) Ichikawa, Y.; Liu, J. L.-C.; Shen, G.-J.; Wong, C.-H. J. Am. Chem. Soc. 1991, 113, 6300. Thiem, J.; Sauerbrei, B. Angew. Chem., Int. Ed. Engl. 1991, 30, 1503. Ajisaka, K.; Fujimoto, H.; Isomura, M. Carbohydr. Res. 1994, 259, 103. (a) Schenkman, S.; Jiang, M.-S.; Hart, G. W.; Nussenzweig, V. Cell 1991, 65, 1117. (b) Nahori, M. A.; Etievant, M.; Terzidistrabelsi, H.; David, B.; Lambre, C. R. Immunol. Lett. 1994, 42, 67. Ferrero-Garcia, M. A.; Trombetta, S. E.; Sanchez, D. O.; Reglero, A.; Frasch, A. C. C.; Parodi, A. J. Eur. J. Biochem. 1993, 213, 765. Vandekerckhove, F.; Schenkman, S.; Decarvalho, L. P.; Tomlinson, S.; Kiso, M.; Yoshida, M.; Hasegawa, A.; Nussenzweig, V. Glycobiology 1992, 2, 541. Lee, K. B.; Lee, Y. C. Anal. Biochem. 1994, 216, 358. Ito, Y.; Paulson, J. C. J. Am. Chem. Soc. 1993, 115, 7862.

8 The Chemistry and Biology of Multivalent Saccharide Displays David A. Mann and Laura L. Kiessling University of Wisconsin–Madison, Madison, Wisconsin

I.

INTRODUCTION

Recognition of the importance of carbohydrate binding proteins in regulating biological systems is fueling efforts to understand and manipulate these processes. Lectin–ligand interactions have been found to play key roles in a wide variety of life processes: bacterial and viral attachment, immune system function, transplant rejection, fertilization, cell targeting, and the inflammatory response [1–7]. Despite these significant functions for protein–carbohydrate interactions, observed dissociation constants are typically on the order of 10⫺4 –10⫺3 M for a monovalent carbohydrate adhesion to a single lectin binding site [3,8–10]. The low affinity of these interactions raises a critical question: How do natural systems exploit weak binding events? An attractive idea is that protein–carbohydrate interactions garner the requisite strength and specificity required to mediate important biological processes through multivalent contacts. Thus, multiple simultaneous interactions can occur between an array of associated lectin binding sites and a multivalent presentation of saccharide ligand epitopes [8,11–15]. This chapter describes recent efforts to investigate and manipulate multivalent binding events involving protein–carbohydrate interactions. We focus on how synthetic ligands have served as critical tools for dissecting, illuminating, and controlling these complex recognition processes and their functional consequences. There are several excellent reviews that discuss other aspects of multivalent protein–carbohydrate interactions [9,15–17]. II.

STRUCTURAL FEATURES AND ENERGETICS

A number of specific components contribute to the complex energetics of lectin– ligand interactions. The observed affinities (functional affinities, see Section II.B.1) 221

222

Mann and Kiessling

for these processes arise from the features of the monovalent and multivalent binding interactions. As a prelude to an examination of specific multidentate interactions, this section describes the features of monovalent protein–saccharide complexation and the factors that contribute to the high functional affinities observed for many multivalent interactions. A.

Monovalent Interactions

The structural features of monovalent lectin–ligand complexes have been elucidated through X-ray crystallographic analysis [1,18–20]. These complexes have helped facilitate an understanding of the energetics of monovalent carbohydrate binding to proteins; they have facilitated structure-based free energy calculations, as well as providing a basis for interpretation of titration microcalorimetry experiments [10,21– 23]. Although proteins can bind carbohydrates tightly, lectin binding sites are often solvent-exposed indentations rather than deep binding pockets. In these shallow binding sites, contacts are made with a limited portion of the carbohydrate ligand (Figs. 1 and 2). Monovalent interactions appear to rely on hydrogen bonding, metal chelation, water displacement and reorganization, hydrophobic contacts, and Coulombic interactions. We will use specific protein–lectin interactions to illustrate these energetic contributions. 1.

Hydrogen Bonding

The density and diversity of stereochemical orientations available to saccharide hydroxyl groups suggests that hydrogen bonding interactions might play a role in their recognition by proteins. Although the energetic contributions of specific hydrogen bonding interactions are difficult to dissect, the directionality of these interactions is important for recognition and discrimination of saccharide ligands. For example, in the structure of the lectin concanavalin A (ConA) complexed with ␣-methyl mannopyranoside, hydrogen bonding interactions occur between the protein and the 3-, 4-, and 6-position hydroxyl groups of the saccharide [19]. There is no direct contact with the 2-position hydroxyl group, and ConA interacts with saccharides of both the gluco and manno configurations. The hydrogen bonding interactions, however, allow the lectin to discriminate between galactose and mannose (or glucose) derivatives. The amino acid side chains as well as the peptide backbone can contribute to recognition (Fig. 2). The side chain amide hydrogen of asparagine 14 can form a hydrogen bond with the 4-position hydroxyl group, whereas the backbone amide hydrogens of leucine 99 and tyrosine 100 are within reach of the ring and 6-position hydroxyl groups, respectively. This structure also illustrates another common feature of lectin–saccharide complexes: many side chains form bidentate hydrogen bonds. For example, the aspartic acid 208 carboxylate oxygens can interact with both the 4- and 6-position hydroxyl groups of mannose. In addition to direct contacts between the lectin and the saccharide, water-mediated hydrogen bonds commonly occur [22,24]. 2.

Metal Chelation

Many lectins depend on divalent metal cations, especially Ca2⫹, for carbohydrate binding. Two roles have been observed for metal cations. First, to preorganize side chains for saccharide binding, and second, to directly ligate saccharide hydroxyl

Multivalent Saccharide Displays

223

Figure 1

The main chain of concanavalin A as determined by X-ray crystal structure analysis. The homotetramer and monomeric subunits are shown with bound ligand.

groups. Examples of the former can be found in the legume lectin structures, as observed in ConA, where Ca2⫹ and Mn2⫹ assist in organizing local structure and also in orienting protein functional groups for saccharide interaction (Fig. 2) [19]. Both modes of metal ion function are important in C-type lectin recognition. Direct coordination of a metal to both the complexed lectin and the target saccharide ligand is observed in C-type lectin structures [4,5] such as that of rat serum mannose binding protein A (MBP-A). In the model derived from X-ray crystallographic analysis, the

224

Mann and Kiessling

Figure 2

Concanavalin A binding site showing metal chelation and hydrogen bond

networks.

saccharide hydroxyl groups act as ligands for the protein-bound calcium ion (Fig. 3). In addition, Ca2⫹ binding organizes protein side chain conformation to facilitate saccharide complexation [25]. The addition to these metal-cation-dependent lectins of chelating agents, such as EDTA, abrogates ligand binding, a result that emphasizes the importance of divalent metal ions in lectin activity. 3.

Hydrophobic Effects

Lectin binding sites, like the saccharides that bind to them, are amphiphilic. In addition to the hydrogen bonds that are important for orienting saccharides, the binding sites contain many hydrophobic groups that can contribute to saccharide binding affinity. The structure of MBP-A derived from X-ray crystallography illustrates the importance of hydrophobic binding regions complementary to those on the ligand (Fig. 4) [1]. For example, isoleucine 207 is located proximal to the 6-position methylene group of mannose, which should result in favorable van der Waals contacts. In many lectin–saccharide complexes, aromatic residues have been found to pack against one or more faces of the saccharide [26]. In the structure of MBP-A, the side chain of histidine 189 is near the 4-position C—H and 6-position O—H bonds, suggesting that interactions between the aromatic residue and nonpolar areas of the saccharide contribute to mannose binding in this system. 4.

Coulombic Interactions

Coulombic interactions between carbohydrate binding proteins and their ligands are pervasive. Carbohydrates often bear anionic charges, which are installed through attachment of sialic acid residues or by phosphorylation or sulfation of uncharged

Multivalent Saccharide Displays

225

Figure 3

The mannose binding protein A (MBP-A) binding site, featuring calcium coordination to protein side chains as well as the 3- and 4-position hydroxyl groups of mannose.

Figure 4

The mannose binding protein A (MPB-A) binding site featuring van der Waals contacts with the mannose ligand.

226

Mann and Kiessling

saccharides [27]. For proteins that bind sulfated polysaccharides such as heparin, salt bridges have been observed between positively charged lysine and arginine side chains and the anionic sulfate and carboxylate groups of the saccharide [28]. In contrast, some lectins bind anionic ligands without formation of salt bridges. For example, in the structure of sialic acid bound to influenza virus hemagglutinin, the carboxylate group of sialic acid hydrogen bonds to water and to a backbone amide proton [18]. Structural studies of the protein–saccharide complex in a related example, an MBP-A variant that binds the tetrasaccharide sialyl Lewis acid x (sLex) [29], also revealed that interactions between the protein side chains and the anionic carboxylate of sialic acid occurred indirectly rather than through a direct salt bridge. 5.

The Role of Water

The ability of water to act as both donor and acceptor of hydrogen bonds makes it an excellent mediator in protein–carbohydrate complexes. Results of numerous Xray crystallographic studies of protein–carbohydrate complexes attest to the pervasiveness of water-mediated hydrogen bonds in protein–carbohydrate interactions. Lectin-bound water molecules are proposed to act as an extension of the protein surface itself [1]. This idea is supported by the fact that water molecules occupy conserved sites in the structures of related lectins [24]. In addition, water molecules can mediate similar hydrogen bonding interactions in the complexes of different proteins with a specific carbohydrate ligand or in complexes of the same lectin with different saccharide ligands [1]. The energetic contribution of water to carbohydrate binding is controversial. From crystallographic and thermodynamic data and molecular modeling calculations, Lemieux [21,22] and Toone [10] have speculated that changes in hydration are a driving force for carbohydrate binding. They postulate that because carbohydrates and their binding sites are amphiphilic, an advantageous reorganization of water occurs upon binding. To gather experimental support of the importance of hydration in binding, Lemieux’s research group has generated oligosaccharide derivatives with polar groups attached to positions outside the binding site. These modifications, which should not disrupt critical binding interactions but do alter the saccharide solvation shell, affect ligand binding [22,30]. Most of the changes in saccharide solvation were detrimental to binding, a result that underscores the significant and impalpable role of water in carbohydrate–protein complexation. B.

Modes of Multivalent Ligand Function

Saccharides constitute one class of biologically important molecules whose function depends on multivalent presentation. In physiological settings, saccharide epitopes are not usually encountered in isolation, but rather in multivalent arrays. For example, high ligand density glycoproteins can serve as scaffolds for the presentation of saccharides. Alternatively, saccharides can be displayed in multivalent arrays through glycolipid clustering. Moreover, many saccharide binding proteins are multivalent because they form oligomeric quaternary structures or are present in multiple copies on a cell surface. To understand carbohydrate function, therefore, we need insight into how multivalent saccharide presentation affects recognition. Natural systems rely on assembling multiple weak monovalent binding events

Multivalent Saccharide Displays

227

into multivalent arrays, and this strategy can lead to enhancements in the strength and specificity of the subsequent interaction. The importance of ligand multivalency, which is also referred to as the ‘‘cluster effect’’ [3], is widely recognized, but additional investigations are required to understand the increases in activity that are observed. Several types of interaction that can augment the observed activities of multivalent ligands are discussed in the following sections. Because of the difficulties associated with dissecting the mechanisms that contribute to the activity of a multivalent ligand, we use the term ‘‘functional affinity’’ to refer to the measured activity of a particular ligand. Functional affinity is a descriptor that has been used in studies of antibody–antigen interactions, which are also multivalent. Functional affinity is useful for the purposes of comparing monovalent and multivalent ligand activities because it describes a measurable quantity: the apparent association constant (or inhibition constant). The conditions under which an assay is conducted, however, must also be carefully considered. Many multivalent binding events, both in vivo and in vitro, do not occur under equilibrium conditions. Thus, the functional affinities of multivalent ligands can vary between assays, even when those of their monovalent counterparts do not (see Section III). 1.

Functional Affinity

As noted, monovalent binding energies are often weak, displaying dissociation constants in the millimolar range. The relative affinity preference of a lectin for two structurally related monovalent saccharide ligands can be slight. Presenting lectin binding sites in an array that can simultaneously recognize multiple clustered carbohydrate epitopes facilitates multivalent binding, a process that enhances not only the strength of the resulting interaction, but also the specificity. For example, a fourfold affinity preference for monovalent mannose over glucose binding to ConA is magnified to a 160-fold preference for a multivalent display of mannose over an analogous glucose polymer [31]. This enhancement in affinity and avidity, functional affinity, is thought to be the result of the combination of numerous multivalent binding modes. Characterizing the molecular features of these binding modes is a critical step in understanding lectin–ligand interactions. 2.

Antibody–Antigen Binding: A Context for Exploring Multivalent Saccharide Recognition

Multivalent interactions involving antibody binding to multivalent antigens have been studied extensively. Antibodies themselves can be multivalent, existing as the bivalent immunoglobulin D, E, and G types, the tetravalent IgE, or the immunoglubulin M type, displaying 10 associated binding sites. Models for multivalent antibody binding were developed by Crothers and Metzger in the 1970s [32–34]. These early studies emphasized that for multidentate ligands to simultaneously occupy multiple binding sites, they must present recognition epitopes that are both properly spaced and sufficiently flexible. As expected, these variables are also important in multivalent carbohydrate binding. Still, an understanding of how much increase in activity is to be gained by such presentation and how the structures of multivalent ligands should be optimized for maximal potencies remains a mystery. The following sections describe some of the molecular features and general mechanisms that lead to enhanced activities.

228

3.

Mann and Kiessling

The Chelate Effect

The simultaneous placement of saccharide residues from a multivalent ligand into multiple binding sites on an oligomeric lectin can lead to high functional affinities if the ligand epitopes are properly oriented to interact favorably with appropriate sites on the receptor. This situation, in which multiple receptor sites are occupied by a multidentate ligand, is an example of the chelate effect. The term ‘‘chelate effect’’ arises from the observation that ligands possessing more than one donor group are more effective at coordinating electrophilic metal centers than those that possess a single donor. Thus, the chelate effect describes multiple interactions, occurring after the first contact is made, as effectively intramolecular (Fig. 5A) [35]. The observed free energy of interaction (⌬Gobs) for an oligomeric receptor R1 —R2, . . . , —Rn with an oligomeric ligand, L1 —L2, . . . , —Ln will be related to the sum of the free energies of binding for each individual receptor. Thus, the binding is often estimated as follows: ⌬Gobs = ⌬G(R1 —L1) ⫹ ⌬G(R2 —L2) ⫹ ⭈ ⭈ ⭈ ⫹ ⌬G(Rn —Ln) This model predicts that the formation of additional contacts should give exponential affinity increases because only the first binding event exacts a translational entropy penalty. However, the entropic costs related to conformational flexibility are likely to be much higher for a multivalent protein–carbohydrate interaction. Page and Jencks recognized the potential utility of the chelate effect for devising highly effective inhibitors of target enzymes [35,36]. This idea was extended to devising multidentate ligands as potent inhibitors for multivalent receptors. For example, a synthetic nonnatural trimer of the glycopeptide antibiotic vancomycin binds with high functional affinity to the corresponding trimer of d-Ala-d-Ala, its dipeptide ligand. The affinity of this complex, stabilized by the chelate effect, is similar to that of the avidin–biotin interaction [37]. However, large distances (>40 ˚ ) separate the saccharide binding sites in many oligomeric lectins; therefore, long A spacers are required to link saccharide epitopes for multivalent binding. Thus, many attempts to design effective multivalent saccharide derivatives have been thwarted by the increases in conformational entropy inherent in the long linkers required for multivalent binding. The conformational entropy term is affected by the structural features of the multivalent ligand, such as the rigidity of the scaffold displaying the binding determinants and the relative orientation of each recognition element [15]. The functional affinity of multivalent ligands will reflect both the advantages of translational entropy gains and the disadvantages of conformational entropy penalties. 4.

Rebinding Leads to Decreased Off-Rates

The display of a receptor or ligand at high density can result in its partner exhibiting increased functional affinity, even if multipoint binding does not occur. The increase in apparent affinity can be rationalized by modeling the binding event as a two-step process. In the first phase, receptor and ligand form a transitory encounter complex. The second stage involves the transitory species proceeding to generate a stable complex or dissociating as a result of nonproductive binding (Fig. 5B). In the case of a multivalent partner, the probability of rebinding increases and that of dissociation decreases. Because the off-rates are slower for such an interaction, the observed

Multivalent Saccharide Displays

Figure 5

Modes of multivalent binding. 229

230

Mann and Kiessling

functional affinities are higher. For example, ligand affinities for proteins clustered on a membrane or immobilized on a surface can appear to be significantly higher than that of the corresponding species in solution [38,39]. Like the chelate effect, this mode of statistical enhancement benefits from a high effective concentration of the binding partner, although the magnitude of enhanced activity is likely to be smaller than that for systems that exploit chelation. An advantage of such a statistical effect, however, is that optimal spacing of the ligand epitope is not crucial. 5.

Receptor Clustering

For a multivalent ligand, an alternative to occupying multiple saccharide binding sites within an oligomeric receptor is to bind multiple receptors at the cell surface (Fig. 5C). This mode of interaction is similar to the chelate effect, although the entropic costs are different. Moreover, the resulting effects of complexation can be quite distinct. For example, once the initial translational penalty for ligand binding has been paid, additional interactions will result in some restriction of receptor mobility. Nevertheless, this cost is moderated because receptors in a fluid membrane diffuse in two dimensions rather than three. An advantage over the chelate effect is that the ligand scaffold need not precisely orient epitopes for binding as proper spacing can be adopted by receptor movement in the membrane. Consequently, there will be more orientations in which multiple receptor–ligand complexes can form. Finally, the interaction of multivalent ligands with membrane-bound receptors can lead to their clustering. Bringing receptors together via liganded clustering may have useful functional consequences [40,41]. 6.

Subsite Binding

Some lectins have demonstrated the presence of secondary binding sites that aid in ligand discrimination [1,42–46]. In binding to these lectins, ligand substituents may either interact with a second independent saccharide binding site or with regions adjacent to the primary binding site (Fig. 5D). Secondary saccharide binding sites can impose ligand selectivity that results in specificity for complexation of one oligosaccharide (or one multivalent ligand) over another [43]. Similarly, an extended building site can engage in further interactions with substituents on the ligand, such as an appended carbohydrate residue or a portion of the scaffold upon which the primary saccharide ligand is displayed [47–49]. Although the physiological relevance of such sites remains unknown, multivalent ligands that can target subsites and extended sites will be more effective and selective inhibitors. 7.

Steric Interference

Steric stabilization can result in increases in the activities of multivalent ligands designed to inhibit the interaction of receptors and ligands on opposing surfaces, as in the case of cell adhesion. Whitesides and coworkers have suggested that the binding of high molecular weight multivalent ligands to one surface (e.g., a virus) can preclude the attachment of a competing surface (e.g., a cell) by sterically blocking access to neighboring binding sites (Fig. 5E) [15,50]. This mechanism may be especially important for ligands that target cell surface receptors displayed at high densities (or high local densities). Moreover, a large multivalent ligand that can bind to one receptor may also block access to other receptors located at interstitial sites. For multivalent ligands operating by this mechanism, the nature of the scaffold em-

Multivalent Saccharide Displays

231

ployed will play a role in steric repulsion in addition to providing conformation flexibility or rigidity. 8.

The Polyelectrolyte Effect

Many naturally occurring multivalent saccharide ligands are anionic. Placement of specific anionic groups can be critical for saccharide recognition (see Section II.A.4), but the charge density resulting from a multivalent display of such groups can influence their binding activities. In analogy to protein–DNA interactions, protein binding to highly anionic saccharide derivatives (such as heparan sulfate) can benefit from the polyelectrolyte effect. Charged functional groups on both lectin and ligand mobilize counterions when solvated. Ligand binding releases these ions into the bulk solution, a process that is entropically favorable (Fig. 5F). Additional energetic gains will result from multiple successive attachments [51]. III.

ANALYZING MULTIVALENT BINDING EVENTS

Publications focusing on assessing multivalent interactions often highlight the difficulties associated with assessing binding kinetics and thermodynamics for such processes [11,15,39]. Determining the true equilibrium constant for a multivalent binding event is complicated because, as described above, multiple types of binding interaction may contribute to the interaction under investigation. Thus, an inability to dissect the contributions of monovalent binding, multiple site binding, subsite binding, rebinding of multiple epitopes within one or more binding site, and other effects such as steric stabilization renders the analyses of these systems difficult. The evaluation of protein–carbohydrate interactions is complicated by the tendency of these substances to participate in multivalent binding and the low affinities of their monovalent interactions. These features dictate that the assay used to measure ligand activities be chosen carefully. Assays of several different types may be required to properly evaluate ligand potency. Moreover, generating data relevant to ligand function under physiological conditions can be a difficult and onerous task. Because throughput can be limited in assays that closely mimic physiological conditions, the best assay for screening potential inhibitors might necessarily be artificial, and also may not identify the best ligand (see Sections VI.4 and VII.B.2). Some of the more effective methods that have been employed to evaluate inhibitors of protein–saccharide interactions are described in the subsequent sections. A.

Fluorescence Anisotropy

Fluorescence anisotropy is a technique that can be used to monitor low-affinity monovalent lectin–ligand interactions [52–54]. An advantage of this technique is that it can be used directly to determine ligand binding constants and also to confirm that ligands are competing for a specific saccharide binding site. The basis of the method is the detection of changes in the fluorescence anisotropy of a ligand-bound fluorophore. When a protein binds the labeled ligand, there is an increase in the rotational correlation time of the fluorophore because the resulting complex is large and tumbles more slowly than the free ligand. An increase in fluorescence anisotropy results (Fig. 6), and these changes can be analyzed to afford a binding constant. Moreover, binding affinities of unlabeled ligands can also be obtained accurately and

232

Mann and Kiessling

Figure 6 Schematic depiction of a fluorescence anisotropy assay. Fluorescently tagged molecules are excited by plane-polarized light; only molecules in the proper orientation are excited. Emitted light is detected in the original plane and in the perpendicular plane. The quantity of fluorescence observed in the two orientations is determined by the rate of tumbling, which depends on particle size and relates to binding.

efficiently by monitoring their ability to compete with a fluorescently labeled control ligand. The disadvantage of this method is that high concentrations of protein are needed to accurately determine the weak association constants typically observed for protein–carbohydrate interactions. Even when the protein can be readily produced or isolated, the high concentrations required can result in protein aggregation and/or precipitation. Overall, the method has many advantages, but its applicability depends on the system. B.

Turbidimetric Analysis

Turbidimetric analysis is useful for the study of multivalent interactions that result in the formation of large protein aggregates. The method is typically employed to evaluate the ability of multivalent saccharide derivatives to promote lectin crosslinking (Fig. 7). To conduct the assay, multivalent ligand and lectin are mixed and allowed to form cross-linked complexes; the formation of large protein aggregates is monitored by changes in the optical density of the solution [55–57]. The abilities of various multivalent ligands to cross-link lectin is correlated with their relative affinities. As a measure of monovalent ligand function, inhibitory potencies of these species can be obtained by evaluating their abilities to inhibit lectin cross-linking by a specific multivalent ligand. This analysis is complicated, however, by the possibility

Multivalent Saccharide Displays

233

Figure 7 Multivalent ligand-induced lectin cross-linking. Multivalent ligands, which are able to simultaneously span two lectin molecules, can form cross-linked aggregates. These aggregates are detectable by spectroscopic means in a turbidimetric assay.

that the monovalent ligands could initially contribute to optical density increases. In principle, they could occupy binding sites on the lectin that do not directly result in cross-linking, thereby releasing epitopes on the multivalent ligand that then attract more lectin to the aggregate. The complexities of comparing monovalent and multivalent ligands by turbidimetric analysis are not the only disadvantages of this method. One potential problem is that it does not effectively report on the activities of multivalent ligands that act by simultaneously binding to multiple sites within a lectin oligomer. In addition, the relative activities of a multivalent ligand for two related lectins cannot be determined because the inherent solubilities of the lectins and their cross-linked complexes will differ. Importantly, because precipitation of the lectin–ligand complex occurs, the process is generally under kinetic control; consequently, the relative potencies of ligands may not correlate with their intrinsic affinities for the lectin. As with most assays of multivalent ligand function, the species being detected is not the initial protein–ligand complexation event. Still this assay is widely used because it can be conveniently and rapidly performed. C.

Enzyme-Linked Immunosorbant Assays

A wide range of variations on the enzyme-linked immunosorbant assay (ELISA) have been used to evaluate monovalent and multivalent inhibitors owing to the convenience and utility of these methods (Fig. 8) [58,59]. Generally, one binding partner is immobilized on the bottom of a microtiter plate, and a binding counterpart, labeled with a reporter, is allowed to adhere to the immobilized partner. Monovalent and multivalent compounds can be tested for their ability to inhibit the binding interaction. The reporter group can be a radiolabel, fluorophore, biotin, conjugated antibody, peroxidase, or other enzyme conjugate. In the case of a protein–saccharide interaction, either the lectin or the ligand can be immobilized. Inhibitory potencies obtained from ELISA and related assays provide an estimate of the functional affinity of various ligands, although confounding parameters can complicate the analysis. The immobilized partner is often displayed in a random mixture of orientations, some of which may not be optimal for binding. All soluble ligands will encounter these various receptor orientations; therefore a comparison of ligand activity generally affords the proper qualitative relative ranking. One problem with the variety of orientations observed is that they may not reflect those found in natural settings. Consequently, the artificial ligand spacing inherent in the ELISA can introduce a bias to

234

Mann and Kiessling

Figure 8 An example of an ELISA. Monomeric lectin is immobilized in the bottom of a microtiter well. Labeled multivalent ligand binds to the lectin. The assay can also be set up to observe lectin binding to immobilized ligand.

the assay. ELISAs, however, are simple to perform and can be used for rapid screening of multiple inhibitors. D.

Cell Agglutination

The activities of both monovalent and multivalent ligands can be assessed in a cell agglutination assay if the targeted receptor has the ability to agglutinate cells. This assay is generally applied to evaluate the potencies of ligands for a multivalent lectin that recognizes saccharide epitopes on red blood cell surfaces, thereby causing hemagglutination (Fig. 9) [60]. The lectin and isolated cells are combined in a microtiter plate with conical wells. In the absence of lectin, or in the presence of an inhibitor, the cells eventually pellet by gravity in the bottom of the well. In contrast, when the cells are cross-linked they form a large aggregate network that covers the well. Inhibition of agglutination is detected by visual inspection. The human component of the analysis renders this assay subject to potential bias, and it also makes it difficult to rapidly screen ligand libraries. Nevertheless, hemagglutination is a convenient method of identifying either monovalent or multivalent inhibitors of lectin–ligand binding. E.

Surface Plasmon Resonance

Surface plasmon resonance (SPR) is an analytical method that can be used to evaluate protein–ligand interactions. It has only recently been applied to probe lectin–ligand interactions [11,61–66]. In SPR, the amount of material bound to a surface is quantified by monitoring the absorption of an evanescent wave. One binding partner is attached to the surface by covalent or noncovalent interactions, and its counterpart

Multivalent Saccharide Displays

235

Figure 9 Hemagglutination of red blood cells. Red blood cells are combined with a lectin that can recognize cell surface carbohydrate ligands. In the presence of inhibitors, the cells pellet, forming a visually identifiable spot in the well. Because cross-linked cells cannot pellet; the well maintains a homogeneous red tint.

is allowed to flow over the surface (Fig. 10). Receptor–ligand association is monitored to generate an apparent on-rate, and dissociation to afford the apparent offrate. From these values, an equilibrium constant can be obtained. To overcome the complications associated with monitoring ligand function in a heterogeneous system, a competition assay can be employed in which inhibitors are evaluated for their ability to block the target binding interaction. Thus, multiple ligands can be screened rapidly with a single binding surface [67–69]. One advantage of SPR is that only small amounts of material are required for the analysis. In addition, the availability of automated sample injection and monitoring facilitates the rapid screening of large numbers of compounds. Since confounding interactions can complicate the analysis, the design and implementation of an optimal binding surface is critical for the generation of meaningful data. IV.

PLANT LECTIN–SACCHARIDE INTERACTIONS

The carbohydrate binding properties of plant lectins have been widely studied and exploited. Because many plant lectins can be readily isolated, numerous studies of

236

Mann and Kiessling

Figure 10 Surface plasmon resonance analysis. A surface displaying ligand is examined under conditions of flow. Injection of lectin results in an increase in signal observed in the association phase of the curve. At equilibrium the surface achieves maximal binding. Measurement of the dissociation can yield off-rates. Competing ligands in solution can also be assayed.

these receptors have been conducted. The results serve as a foundation for understanding protein–carbohydrate interactions. Plant lectins can be classified by their source, and these groups include the legume, cereal, Amaryllidaceae and related families, and Moraceae and Euphorbiaceae lectins [5]. The lectins in all these classes assemble into multimeric proteins and can adopt a range of different quaternary structures including di- and tetrameric arrangements. The multimers formed afford multivalent displays presenting saccharide binding sites in a variety of spacings and orientations with documented multiplicities of 2 to as many as 12 binding sites, as in the case of the snowdrop lectin [70]. X-ray crystallographic studies have revealed the spatial relationships in many of these multiple binding site arrangements [1,5,42,71,72]. With the diversity of multivalent binding site displays and the large amount of structural and biophysical data available on them, highly accessible plant lectins serve as excellent receptors for the study of multivalent protein–carbohydrate interactions. A.

Concanavalin A

One of the most widely studied lectins to date is the glucose/mannose binding legume lectin from Concanavalia ensiformis (ConA) [73]. ConA is a tetramer at neutral pH, and X-ray crystallographic analyses revealed the tetrahedral arrangement of the protomers (Fig. 1) [19]. Each subunit contains a saccharide binding site; therefore, ConA can bind up to four saccharide epitopes simultaneously. Significant insight into the

Multivalent Saccharide Displays

237

molecular basis for ConA–saccharide interactions has been obtained through X-ray crystallographic analyses of ConA and its saccharide complexes (Figs. 1 and 2) [19]. The binding site is organized by side chain interactions with two divalent metal cations, Mn2⫹ and Ca2⫹ (Fig. 2), and the minimum saccharide determinants that bind ConA are glucose and mannose residues [73]. The lectin exhibits a three- to fourfold binding preference for mannose over glucose and does not bind fucose or galactose appreciably [74–76]. Hydrogen bonding contacts from the protein to the 3-, 4-, and 6-position hydroxyl groups of the carbohydrate dictate ConA saccharide binding specificity. ConA binds relatively weakly to monosaccharides, but it interacts with higher affinities with oligosaccharides and glycoproteins [77]. For example, ConA binds the trimmanoside methyl 3,6-di-O-(␣-D-mannopyranosyl)-␣-D-mannopyranoside with approximately 100-fold higher affinity than methyl-␣-D-mannopyranoside [20,23,78,79]. The well-characterized molecular features of ConA–saccharide interactions and the protein’s ability to participate in multivalent recognition events render ConA an excellent subject for the study of multivalent recognition. 1.

Synthetic Multivalent Ligands Are Effective Inhibitors of Concanavalin A

A number of natural multidentate ligands, including oligosaccharides and glycoproteins [73,77,80,81], serve as highly effective ligands for tetrameric ConA. While these ligands provide useful information about ConA binding, the structures are so difficult to vary systematically that it is difficult to analyze binding data from such substrates. Synthetic ligands have been used, therefore, to address fundamental issues with regard to multivalent recognition by ConA. Major ligand classes that have been employed include low molecular weight derivatives, dendrimeric ligands, and linear polymers. Some of the most potent inhibitors of ConA are multidentate ligands synthesized by the ring-opening metathesis polymerization (ROMP). ROMP has emerged as an attractive approach to the synthesis of functional polymers [82]. The advent of defined, functionalized group tolerant, metal alkylidine initiators has offered new opportunities for the creation of tailored materials with significant biological properties. A generalized mechanism for the polymerization involves an initiation through coordination of the metal alkylidine with a strained cyclic alkene monomer (Fig. 11). A [2 ⫹ 2] cycloaddition between the initiator (2) and a monomer (1) unit generates a metallocyclobutane intermediate (3) that undergoes a ring opening to relieve strain, thereby generating a new metal alkylidene. This intermediate metallocarbene (4) can react with another monomer, and the sequence of events can continue until the monomer is consumed. The reaction can be terminated by the addition of an electron-rich enol ether (5), which results in the production of an unreactive metal carbene (7) and a terminated polymer chain (6). Because ROMP can be a living process and because the rate of initiation exceeds that of propagation, oligomers of controlled length and narrow molecular mass distribution can be generated. In addition, both the catalyst and the electron-rich enol ether afford an opportunity to attach unique end labels to either polymer terminus. Thus, ROMP offers advantages for the production of biologically active polymers [83]. To explore interactions of multivalent ligands with ConA, polymers with pendant mannose or glucose residues have been generated by ROMP. The saccharide epitopes were attached to norbornene monomers through C-glycoside linker units

238

Figure 11

Mann and Kiessling

General scheme for ring-opening metathesis polymerization (ROMP). Monomer 1 and catalyst 2 continue to react until the monomer has been consumed yielding polymer 4, which can be end-capped with functionalized allyl ether 5.

Multivalent Saccharide Displays

239

and polymers of high molecular mass were produced. These materials were highly effective inhibitors of ConA-facilitated red blood cell agglutination: they were up to 105-fold more potent than ␣-methyl glucopyranoside [31]. Other nonnatural saccharide-substituted polymers have also been found to be effective inhibitors of ConA [84,85]. In addition, dendrimers [86–88] and low molecular weight ligands [44,89– 93] show levels of activity that exceed those of mannose, although they do not approach those of the ROMP-generated polymers. Together, these results highlight the range of inhibitory activities that can be manifested by synthetic multidentate ligands. 2.

Specificity in Multivalent Recognition: Concanavalin A Discriminates Between Related Multidentate Ligands

The specificity of ligand binding for a multivalent protein–carbohydrate interaction was explored by using ConA. ConA binds both glucose and mannose residues, but it binds monovalent C-mannoside 9 with a slightly (0.2 kcal/mol) more favorable free energy than the corresponding C-glucoside 8 (Fig. 12) [54]. Whether the small difference observed in the monovalent binding event would be manifested with the corresponding multivalent ligands has been explored. By examining the ability of glucose- and mannose-substituted polymers to inhibit ConA-mediated agglutination, the inhibitory potencies of mannose-modified polymer 11 could be compared with the corresponding glucose counterpart 10 in a hemagglutinaton assay (Fig. 12). Not surprisingly, no difference in potency was detected between the glucose and mannose monovalent C-glycosides 8 and 9. The activity difference between the glucose- and mannose-substituted polymers, however, is greater than 100-fold [31]. Consequently, multivalent interaction yields higher specificity as well as affinity. An enhancement of specificity through multivalent saccharide display is also observed in the binding of cluster glycosides of N-acetylgalactosamine to the mammalian hepatic Gal/ GalNAc specific receptor with subnanomolar binding constants (see Section V.C) [94]. When evaluated for binding to the lectin from Bauhinia purpurea, resin-bound saccharides also elicited a dramatic increase in observed functional affinity [38]. Enhanced specificity through polyvalent binding to multivalent carbohydrates has also been observed in interactions with cell surface L-selectin (see below: Sections VII.B.2–VII.B.3) [95]. Enhancement of specificity through multivalent recognition is likely to be an important feature of saccharide recognition events in nature. 3.

The Chelate Effect: Multivalent Ligands That Can Span Two Saccharide Binding Sites Within the ConA Tetramer Are the Most Potent Inhibitors

Multipoint binding leads to an amplification in specificity, perhaps at least partly accounting for the high selectivity of multidentate ligand binding by ConA. In multipoint binding, multiple saccharide epitopes would coincidentally contact ConA, and if this model is important, ligands that could simultaneously occupy more than one saccharide binding site within the ConA tetramer would be expected to be more potent inhibitors. The large distance between the binding grooves, predicted to be ˚ from X-ray crystallographic analysis, is easily spanned by linear approximately 65 A polymers. Although few polymerization strategies offer control over polymer size, ROMP can be used to generate polymers of defined length.

240

Mann and Kiessling

Figure 12 Kiessling’s C-gluco- (8) and mannopyranosides (9) demonstrate increased functional affinity and specificity when displayed as bivalent mannose (13) and glucose (12) ligands.

Kanai et al. set out to prepare and determine the ConA inhibitory potencies of series of mannose-substituted oligomers differing in length (Fig. 13) [84]. Studies by the Grubbs group suggested that ruthenium carbene complexes can carry out living polymerization with the initiation step approximately eightfold faster than propagation [82]. Therefore, by varying the ratio of monomer 14 to initiator 15 in the reaction mixture, multidentate mannose ligands 16a–c of controlled length could be assembled. The effect of polymer length on inhibitory potency was assessed in a ConA-mediated cell agglutination assay. Interestingly, the relative potency of the

Multivalent Saccharide Displays

241

Figure 13

ROMP polymers of defined length. Varying the ratio of monomer (14) to catalyst (15) affords control over the average polymer 16 length (n).

multivalent displays (on a saccharide residue basis) initially increased significantly with increasing polymer length, but eventually reached a plateau. Molecular modeling studies suggest that when polymer 16 is composed of about 25 repeat units, it ˚ required to simultaneously place saccharide residues in two bindcan span the 65 A ing sites within the ConA tetramer (Fig. 14). As predicted, the potency of the oligomers beyond that length did not increase. An independent assay, in which surface plasmon resonance (SPR) was used to monitor inhibition of ConA binding to a glycolipid surface, reproduced the trends [68]. Specifically, a maximal functional affinity increase was observed for polymer 16b, which consisted of 25 monomer units. ConA structural data provides a context for interpreting mechanisms of multivalent ligand function; consequently, these studies represent strong evidence that lectins can use the chelate effect to increase the strength of a protein–carbohydrate interaction. Such mechanisms are likely also important in vivo.

Figure 14

(A) Multivalent polymer binding within one binding site. (B) Polymers of sufficient length can span two binding sites simultaneously. (C) Longer polymers yield no increase in potency.

242

4.

Mann and Kiessling

High Local Ligand Concentration Contributes to ConA Binding

Despite the importance of the chelate effect in the increased functional affinities of ConA for the multidentate ligands described above, the chelate effect is not the only contributing mechanism. If chelation was the only important process, oligomers composed of more residues than are required to span the binding sites should display decreased potencies when evaluated on a saccharide residue basis because noninteracting residues would detract from the observed affinity. This was not the case, however, inasmuch as potency did not decrease. Therefore, the potencies of the longer polymers are greater than the chelation model predicts (Fig. 14). We suspect that the high local concentration of saccharide residues available with these polymers decreases the rate of dissociation of the multivalent ligands, making rebinding more favorable. This statistical enhancement mode of multivalent binding also contributes to the increase in the observed affinity of ligands unable to span two sites (i.e., <25 residues in length). Moreover, the large number of mannose residues available for binding with these ligands suggests that a number of different lectin–ligand complexes can form. Finally, ConA clustering and steric stabilization (see Section VI.C) may also contribute to the efficacies of the longer polymers (Fig. 5). 5.

Spacing and Orientation of the Saccharide Epitopes Is Important for ConA Interaction

Low molecular weight divalent and trivalent ligands have also been generated to probe ConA ligand recognition. As had been shown with the hepatic lectin (see Section V.C) [96,97] and influenza virus hemagglutinin (see Sections VI.E–VI.F) [98], low molecular weight ligands can achieve high potencies, displaying inhibitory potencies 100- to 1000-fold higher than those of the corresponding monovalent saccharides. To investigate the effect of epitope spacing and orientation in ConA binding, Roy and coworkers constructed a series of low molecular weight bivalent ligands in which the saccharide epitopes were appended through linkers differing in length and composition (Fig. 15) [89,90]. The ability of these bivalent molecules to inhibit ConA binding to immobilized mannan (a yeast glycoprotein) was tested in an enzyme-linked lectin assay (ELLA) (Fig. 8). The structure of the linking group had subtle effects on the ConA inhibitory properties of the bivalent molecules. The bivalent ligand with a six-carbon linker, 17c, found to be superior in this study, was 5.3 times more effective per saccharide at inhibition than ␣-p-nitrophenyl-mannopyranoside [89]. Roy and coworkers also examined the effects of higher order valency in a related series of related ligands (Fig. 16) [57]. Bi-, tri-, and tetravalent ligands, 18, 19, and 20, were 138, 210, and 231 times more active than ␣-Memannose, respectively, when analyzed on a molar scale. However this translates to 69-, 70-, and 57.8-fold increases in potency when determination is made on a persaccharide-residue basis. On a multivalent display, additional ligands that do not contribute productively to binding can detract from the observed potency [86]. This has also been observed in the case of polyacrylamide displays targeting influenza virus, where maximal activity is observed at a low ligand density (see Sections VI.A–VI.B) [99–101]. The abilities of small bi-, tri-, and tetravalent molecules to inhibit ConA likely arise from their propensities to cross-link lectin multimers. Molecular modeling of Roy’s ligands suggests that the spacers linking the saccharide recognition epitopes

Multivalent Saccharide Displays

243

Figure 15

Roy’s bivalent mannose ligands directed against ConA. To investigate spacing effects, the ligands are designed with variable length linkers.

˚ required to simultaneously occupy two binding are not long enough to span the 65 A sites within a single ConA tetramer (D. Mann, unpublished results, 1999). Another potential mode of action for these ligands is suggested by Brewer’s finding that many oligosaccharides and glycopeptides can form cross-linked lectin complexes [102]. A turbidimetric assay monitoring the formation of cross-linked ConA precipitates supports this model for Roy’s ligands (Fig. 7). Thus, the best small-molecule inhibitors of ConA may be those that can cross-link the lectin. Dendrimers that exhibit enhanced inhibitory activity toward ConA also appear to function by this mechanism [87,88]. Regardless of their mode of action, data generated with the bivalent ligands highlight the necessity for proper spacing between ligands to elicit optimal multivalent interactions. 6.

Multivalent Saccharide Display on Beads Alters the Ligand Preference

The role of ligand presentation in affinity is also highlighted by the fact that functional affinity trends observed with monovalent ligands do not necessarily mirror the effects observed when these ligands are assembled into multivalent displays. Seeking to identify highly active monovalent ligands, Kahne and colleagues developed a combinatorial synthesis yielding a library of disaccharide ligands on TentaGel beads [38]. A mixed population of the beads was assayed for activity against peroxidaselabeled Bauhinia purpurea lectin (also a hemagglutinin). Beads bearing the two most active ligands identified from the library, compounds 21a and 22a, were shown to have significantly higher functional affinities than beads bearing the known Bauhinia purpurea ligand Gal␤1,3GalNAc, compound 23a (Fig. 17). However, when the monovalent analogs were tested in competition experiments inhibiting Bauhinia purpurea lectin binding to Gal␤1,3GalNAc-coated beads (23a), known ligand 23b was more active than the two best compounds from the library, 21b and 22b. One interpretation of these results is that ligand presentation on the beads altered the relative potencies of the ligands; however, other factors, such as saccharide residue density

244

Mann and Kiessling

Multivalent Saccharide Displays

Figure 16

Roy’s bi-, tri-, and tetravalent ligands 18, 19, 20 all demonstrated similar potencies on a per-saccharide basis

for ConA.

245

246

Mann and Kiessling

Figure 17

When presented on beads, Kahne’s carbohydrate-based ligands had relative activities different from those for corresponding monomers in solution.

and hydrophobicity of the saccharides and beads themselves, might influence the activities of the immobilized library members. 7.

Bivalent Ligands Distinguish Between Related Proteins ConA and the Lectin from Dioclea grandiflora

Many lectins can bind the same carbohydrate epitopes, but small molecules that can selectively target one lectin over another are desirable. To explore this issue, several groups have compared the specificity of carbohydrate binding to ConA with that of the lectin from Dioclea grandiflora [44,76,103]. Examination of the fine saccharide binding specificity of Dioclea grandiflora lectin and ConA reveals a great deal of similarity [103,104]. The two proteins have amino acid sequences that are more than 75% identical, and both bind glucose and mannose residues, exhibiting similar preferences for most naturally occurring saccharides. When assayed in fluorescence anisotropy competition experiments, synthetic bivalent C-glycoside derivatives were able to discriminate between ConA and the D. grandiflora lectin, even though monovalent C-allyl mannose and glucose analogs behaved similarly. Specifically, bivalent mannose and glucose, compounds, 13 and 12, bound selectively to ConA but not to the D. grandiflora lectin (Fig. 12) [44]. Turbidimetric analysis suggested that ConA cross-linking was not occurring, although soluble cross-linked complexes could be ˚ between primary present [102]. The divalent ligands are unable to span the 65 A binding sites on the tetramer, but the molecular basis for this discrimination may lie in the subtle structural differences between ConA and D. grandiflora lectin in regions proximal to the saccharide binding site [79,105,106]. These results indicate that changes in carbohydrate presentation, such as switching from a monovalent to a bivalent ligand, can have dramatic effects on the specificity of protein–carbohydrate recognition. Multivalent ligands can thus be used to distinguish between highly related lectins. V.

HEPATIC LECTIN

The mammalian hepatic lectin (also referred to as the asialoglycoprotein receptor) is found on the surface of liver cells. The protein directs the clearance from the serum

Multivalent Saccharide Displays

247

of glycoproteins that possess terminal galactose or N-acetylgalactosamine residues. Upon recognition by the hepatic lectin, asialoglycoproteins are endocytosed and degraded [107]. Ligands for the protein can be generated from tri- and tetra-antennary N-linked carbohydrate epitopes by release of terminal sialic acid residues to expose galactose residues or by chemical synthesis [97,108–110]. The physiological function of the receptor and the identity of its relevant ligands are matters still under investigation. We will focus on efforts to utilize the hepatic lectin mediated recognition and endocytosis to target various therapeutic agents to the liver. A.

Multivalent Hepatic Lectin Is a Heteromultimer

The hepatic lectin is displayed on hepatocytes as a heteromultimer; in the human and the rabbit, the protein is composed of two subunits, and in mice and rats it consists of three subunits [107,111,112]. Hepatic lectin is a member of the C-type lectin family, a group of mammalian lectins requiring calcium for ligand binding [113]. The subunits act in concert to recognize branched saccharide ligand epitopes, which accounts for the saccharide epitope selectivity described above [108,114–116]. The relationship between multivalent ligand binding and endocytosis is unclear because both monovalent and multivalent peptide-based ligands can lead to an increase in the rate of receptor internalization [114]. It is clear, however, that characterization of the recognition and binding properties of the hepatic lectin is critical for understanding its function and devising strategies for selectively targeting liver tissues. B.

Low Molecular Weight Trimeric Galactose Displays Can Be Potent Ligands for the Hepatic Lectin on the Cell Surface

Multivalent displays of galactose and N-acetylgalactosamine epitopes have been synthesized to identify the features important for glycoconjugate recognition and endocytosis by the hepatic lectin [117,118]. Initial studies, performed with galactosesubstituted neoglycoproteins and polyacrylamide gels, demonstrated that the receptor binds more efficiently to multivalent ligands [119]. Since many natural saccharides contain branched or multiantennary structures, several groups postulated that highaffinity recognition of carbohydrates in biological systems requires the presentation of clustered saccharide residues [108,118]. To test this idea and to elucidate structure–function relationships, Lee’s group and others have synthesized a number of nonnatural low molecular weight multivalent compounds [97,108,109,120]. The compounds were evaluated for their ability to inhibit the binding to hepatocytes of radiolabeled asiolomucoid, a ligand for the hepatic lectin. A synthetic trivalent compound, YEE(ah-GalNAc)3 (24, Fig. 18) in which GalNAc (N-acetylgalactosamine) residues were appended to a glutamic acid template (YEE), was found to be 10,000fold more active than monovalent GalNAc in the assay. Other trivalent derivatives were less effective. For example, displays based on the aminotris(hydroxymethyl) methane template were only 70-fold more active than their monovalent analogs at inhibiting asialoorosomucoid binding to isolated hepatocytes. The modest potency increases suggest that the spacing of the saccharide residues presented on this scaf˚ ) is not complementary to that of the carbohydrate binding sites on the fold (ca. 9 A cell surface. This result is consistent with investigations suggesting that the best ˚ [3]. inhibitors display galactose residues separated at distances between 15–30 A These studies established that low molecular weight synthetic multivalent ligands

248

Mann and Kiessling

could serve as highly potent ligands. Moreover, they emphasize the importance of template design in generating effective multidentate ligands. C.

Low Molecular Weight Trimeric Galactose Arrays More Effectively Inhibit Cell-Surface-Bound than Soluble Hepatic Lectin

The large increases in functional affinities of multivalent Gal- and GalNAc-substituted ligands (such as 24) for the hepatocyte cell surface are not observed in assays involving inhibition of the isolated hepatic lectin. Specifically, the best multivalent inhibitors of soluble lectin activity are approximately 100-fold more active than their monovalent counterparts; in a similar comparison of potencies using cell-surfacebound lectin, however, the multivalent displays exhibit a 10,000-fold enhancement [96,108]. Two potential rationales for the differences are as follows: the subunit organization of the lectin is different on the cell surface, and/or the multivalent ligands bridge more than one lectin on the cell surface. Interestingly, multivalent ligands directed against influenza virus hemagglutinin display similar binding properties; they show greater potency in cell surface binding assays than when evaluated against isolated trimeric lectin (see Section VI.F). Together, these results suggest that small multivalent ligands may give rise to high functional affinities because they can bridge or cluster lectins at the cell surface (Fig. 5). D.

Ligand Spacing Allows Selective Targeting of the Hepatic Lectin over a Related Receptor

As described above, ligands with the proper saccharide epitope spacing and flexibility are more effective at binding the hepatic lectin [108,121,122]. Altering the scaffold for ligand display can also give rise to selectivity for one protein over another, as discussed in the case of ligands that discriminate between ConA and the lectin from Dioclea grandiflora. For the hepatic lectin to function in cell-selective targeting, the carrier ligands must discriminate between this receptor and the related macrophage Gal/GalNAc-specific lectin. Moreover, galactose receptors are also found on hepatocarcinoma cells [123]; demonstrating the potential for cell selectivity would provide support for the concept of designing agents that target hepatocarcinoma cells over hepatocytes. To identify selective ligands, Ozaki et al. compared the binding of a series of multivalent GalNAc presenting cluster ligands for their abilities to inhibit asialomucoid binding to different cell lines. This study revealed that YEE(ahGalNAc)3, 24, can selectively target cell lines producing recombinant hepatic lectin. Specifically, a 20-fold increase in activity for trivalent derivative 24 over monovalent Gal was observed with the cell lines displaying the macrophage Gal/GalNAc-specific lectin, while an increase of approximately 106-fold was obtained with cells displaying the hepatic lectin [94,124]. These studies suggest that although these cell surface lectins possess similar amino acid sequences, their quaternary structures differ. Significantly, ligands can be generated that exploit these differences in lectin, quaternary structure and/or cell surface organization. E.

Applications of Ligands for the Hepatic Lectin: Exploiting Hepatic Lectin for Selective Targeting

The hepatic lectin is a key model system for the development of receptor-mediated drug delivery [125–131]. The observations that the hepatic lectin can endocytose

Multivalent Saccharide Displays

249

ligands is restricted to selected cell types and can be inhibited by potent smallmolecule ligands prompted efforts to exploit this receptor for cell-specific targeting. Several research groups have explored the scope of molecules that can be targeted and endocytosed through a hepatic lectin binding ligand to develop novel therapeutic strategies. For example, oligonucleotide–saccharide ligand conjugates have been generated to deliver genes or antisense oligonucleotides to hepatocytes [132–137]. A similar approach has been employed to direct drugs to the liver [138,139]. In addition, approaches that take advantage of the degradative events that result from hepatic lectin mediated endocytosis have been investigated as a means of targeting and clearing unwanted proteins and associated molecules from circulation [134,140,141]. The examples discussed below focus on the use of low molecular weight multidentate ligands to achieve selective targeting of associated conjugates to hepatocytes through the hepatic lectin. 1.

Using Synthetic Saccharide Determinants for Gene Delivery via Hepatic Lectin

Wu and coworkers conducted pioneering studies in which they used polylysine– asialoglycoprotein conjugates to deliver genes to hepatocytes through the hepatic lectin [132,138,142–145]. The nucleic acid component binds the polylysine portion through Coulombic interactions, and the glycoprotein moiety interacts with the hepatic lectin for delivery of the conjugate. The feasibility of using this receptor to achieve selective uptake of nucleic acids in vivo was demonstrated, and this result launched efforts to optimize this approach. Since obtaining homogeneous galactosesubstituted glycoproteins is difficult, other ligands for the hepatic lectin are under investigation as potential targeting agents. These include glycoprotein and glycopeptide conjugates [146,147] and liposomes [141,148–151]. Although most applications of the targeting strategy have employed macromolecular targeting agents, defined small multivalent displays have valuable properties as drug delivery agents. They are homogeneous, their structures can be more easily optimized, and they are not immunogenic. Nonnatural multivalent galactose derivatives can also function to target genes to cells displaying the hepatic lectin. This approach was illustrated by the studies of Plank et al., in which a tetrameric galactose derivative was linked to unsubstituted polylysine [152]. This conjugate was able to selectively target cell lines displaying the hepatic lectin. Another study assessed in vivo as well as in vitro delivery of the luciferase (Luc) gene. The high-affinity ligand YEE(ah-GlcNAc)3, 24, was used to target hepatocytes [137]. In a strategy that is similar to Wu’s design, Merwin and colleagues covalently linked the YEE(ah-GalNAc)3 cluster ligand to polylysine by conjugation through human serum albumin (Fig. 18). The Luc gene then associated with the conjugate through Coulombic interactions, 28 (Fig. 19). Expression of the Luc gene was observed both in vitro and in vivo. These results indicate the promise of these targeting strategies to improve the design of gene–saccharide conjugates that may result in therapeutically useful delivery agents. 2.

Selective Uptake by the Hepatic Lectin of an Antisense Oligonucleotide–Saccharide Epitope Conjugate

In analogy to the results obtained for targeting genes, it is expected that the endocytotic abilities of the hepatic lectin could also delivery oligonucleotides for antisense

250

Mann and Kiessling

Figure 18 Covalently linked oligonucleotides could be delivered to hepatocytes via attachment to the GlcNAc cluster ligand YEE(ah-GalNAc)3 24.

Multivalent Saccharide Displays 251

Figure 19

Construct 28 was used for luciferase gene delivery to hepatocytes by noncovalent association of the gene to GalNAc cluster 24.

252

Mann and Kiessling

applications. Building on their most tissue-selective trivalent galactosamine cluster, Lee and coworkers covalently attached a radiolabeled oligonucleotide methylphosphonate analog to YEE(ah-GalNAc)3 via the heterobifunctional linkage SMCC (25) to generate (27), which was devised to examine cell targeting specificity (Fig. 18) [135]. These investigations revealed that the oligonucleotide analogs could be selectively delivered to hepatocytes. After a 24 h incubation, the conjugate associated preferentially with a cell line that displayed the hepatic lectin, HepG2, over those that did not. The observed selectivity in uptake was 260-fold over leukemic HL60 and 49-fold over fibrosarcoma HT1080 cells. The fate of the conjugate and its location within the cell were not investigated in this initial study. The results demonstrate selective targeting, but whether this strategy can be used to deliver an antisense agent to the proper cellular location remains to be determined. 3.

Cholesterol-Lowering Conjugates Designed to Function Through the Hepatic Lectin

In contrast to the approaches described above in which molecules designed to function within the liver are delivered through their interaction with the hepatic lectin, a strategy to use the hepatic lectin to eliminate unwanted species from serum has also been investigated. Specifically, Biessen et al. attempted to take advantage of hepatic lectin mediated endocytosis and subsequent ligand proteolysis to lower serum cholesterol levels [134]. It was anticipated that targeting the proteins responsible for serum cholesterol transport, high-density lipoprotein (HDL) and low-density lipoprotein (LDL), to hepatocytes via the hepatic lectin would lead to their degradation, thereby lowering serum cholesterol. Bifunctional ligand 29 was designed, in which cholesterol was linked to a trivalent galactose derivative (Fig. 20). When added to serum, compound 29 formed complexes with HDL and LDL. Intravenous injection of 29 into rats resulted in a decrease in detected serum cholesterol levels by 44%. This approach may prove useful for patients who do not respond to the commonly

Figure 20 LDL in rats.

The cholesterol-linked galactoside cluster 29 was used to lower serum HDL and

Multivalent Saccharide Displays

253

employed treatment, which involves administering inhibitors of cholesterol biosynthesis. Moreover, this strategy may be generally useful to recruit serum proteins to specific sites in vivo. VI.

TARGETING INFLUENZA VIRUS

Many pathogens use protein–carbohydrate interactions to seek out host cells; influenza virus is one example [15,153]. Because influenza virus infection can cause serious illness and even death, the development of strategies to inhibit it is clearly significant. The virus infects its host, using the sialic acid binding lectin hemagglutinin (HA) to bind sialic acid residues on the cell surface. HA exists as a homotrimer that is densely displayed on the viral surface [15,18]. HA binds terminal sialic acid residues found on target cell glycoproteins and is involved in membrane fusion of the virus with the target cell [154,155]. Although structures of HA complexed with different sialic acid derivatives have been determined by X-ray crystallographic analysis [18,156], progress in identifying high-affinity, selective, monovalent inhibitors has been slow [52]. As an alternative to generating high-affinity monovalent inhibitors, researchers have investigated the ability of multivalent ligand arrays to inhibit influenza virus HA. A variety of different ligand classes have been investigated, ranging from low molecular weight ligands [157–161] to dendrimers [9,162–164] to liposomes and high molecular weight polymers [99–101,162,165–173]. The data amassed in the course of these studies illuminate the structural features important in multivalent ligand design and the mechanisms underlying multivalent ligand function. A.

Sialic Acid Substituted Polymers Are Highly Effective Inhibitors

The medical relevance of HA inhibitors has prompted several efforts to generate, characterize, and understand the recognition properties of sialic acid substituted polymers. In a multivalent display of epitopes convalently linked to a scaffold, receptor binding determinants are presented at a high effective concentration and often give rise to high functional affinities. Thus, ligand off-rates decrease because binding is favored by the multiplicity of recognition elements available in the vicinity of the binding site, a statistical binding enhancement. Linear polymeric displays also present recognition units in a variety of distances and orientations. This feature increases the probability that a multivalent ligand will interact favorably with a cell surface receptor through the chelate effect. Large polymeric displays can also benefit from steric stabilization. Moreover, with this range of variation, a multivalent ligand may simultaneously take advantage of several different multivalent binding modes. The following examples illustrate the functional properties of linear arrays. Several groups have generated sialic acid substituted acrylamide polymers and investigated the abilities of these materials to inhibit influenza virus HA [101,162, 166,170,174]. In these investigations, the mole fraction of N-acetylneuraminic acid (NeuAc) residues displayed by the different acrylamide polymers has been varied. The potency of the materials generated depends on the degree of substitution on the polymers. Maximal polymer activity was observed when approximately 10–20% of the units were substituted with NeuAc. Lower amounts of substitution resulted in decreased functional affinities, presumably because these materials were less likely

254

Mann and Kiessling

to bind through the chelate effect and/or related entropically favorable mechanisms. When the mole fraction of NeuAc was higher than 0.2, the potency of the resulting materials typically decreased. This drop in activity may be due to the addition of binding epitopes that do not contribute to productive binding interactions. Moreover, the high charge density of a sialic acid bearing polymer may impede binding, in addition to saccharide substituents that could sterically impede binding. Thus, multivalent, high molecular weight polymers are extremely effective HA inhibitors, and their inhibitory activities are dependent on the degree of saccharide substitution. B.

Method of Polymer Synthesis Affects Activity

Typically, saccharide-substituted copolymers are synthesized through the reaction of a carbohydrate-substituted monomer along with an unsubstituted derivative; many variations have been described [16,175–181]. Alternatively, acrylamide polymers have been generated by postsynthetic modification, in which an electrophilic acrylic acid derivative is polymerized and saccharides are coupled to the resulting template [101,182,183]. The Whitesides group directly compared the activities of materials with identical compositions assembled by these different synthetic methods. First, they generated sialic acid substituted polymers by copolymerization. Reactions were conducted with a sialic acid functionalized acrylamide monomer and different acrylamide monomers displaying other functional groups such as amides, alcohols, glycerols, glucose residues, carboxylic acids, and amines (Fig. 21) [99,100,168–170]. Sialylated acrylamide monomer 30 was reacted with functionalized monomer 31 to generate a series of copolymers with varied steric bulk and backbone flexibility. The most active polymers in a viral hemagglutination assay were those generated by reaction of 30 with simple acrylamide 31 (R = NH2). They found that the materials exhibiting maximal potencies were those generated from the reaction of 0.2 mole fraction sialic acid monomer 30 and acrylamide. The optimized polymer was 6600fold more effective than monomeric sialic acid at inhibiting agglutination [168]. For comparison, polymers with activated esters, such as 36 were also made to explore the effect of postsynthetic modification on polymer activity (Fig. 22) [99]. When two polymers of the same composition with different synthetic histories were compared, the material produced by the postsynthetic modification route was 75-fold more potent. Because monomers differ in their rates of polymerization, the postsynthetic modification strategy can give rise to materials with broader distributions of epitope substitution patterns. A statistically random distribution of binding epitopes may be advantageous as the resulting materials have the capacity to interact with their receptors by a variety of mechanisms. C.

Polymers Act by Chelation and Steric Stabilization

To explore structure–function relationships and optimize their activities, NeuAcsubstituted acrylamide polymers were prepared with a variety of appended functional groups (Fig. 22) [99]. These studies generated polymers with very high inhibitory potencies toward HA. For example, an acrylamide copolymer substituted with 10% benzylamine and 20% NeuAc residues had a KI of 0.6 nM [100]. Both the chelate effect and steric stabilization were mechanisms suggested to contribute to the potent activity of this and related materials (Fig. 5). The NeuAc-substituted acrylamide

Multivalent Saccharide Displays

Figure 21

Whitesides’s copolymerization of SA acrylamide 30 and various copolymerants 31 yielded mixed copolymers of altered carbohydrate density.

255

256

Mann and Kiessling

Figure 22

Two routes to polyfunctional acrylamide polymers. (A) Formation of acrylamide polymers by reacting functionalized monomers 33, 34, and acrylamide. (B) Postpolymerization coupling of activated polyacrylate ester 36.

polymers inhibited the adhesion of polyclonal antivirus antibodies that do not bind HA, thus supporting the hypothesized role of steric stabilization [169]. Additional support was provided by the observation that the addition of monovalent neuraminidase inhibitors to a solution of virus-bound polymer increased the potency of the polymer [50]. Neuraminidase binds sialic acid residues but does not contribute measurably to agglutination; consequently, the addition of neuraminidase inhibitors could promote the release of polymer side chains involved in nonproductive neuraminidase binding. The result would be an increase in the effective volume of the polymer, which should increase the observed activity of the polyvalent ligand through steric stabilization.

Multivalent Saccharide Displays

D.

257

Hemagglutination Inhibition Data Fails to Correlate with Viral Infectivity Data

Liposomes are another class of multivalent ligands that have been used to inhibit HA [167,173,184]. One advantage of this form of multivalent ligand presentation is that glycolipids can diffuse laterally to adopt proper spacing for optimal binding. Therefore, the need to precisely engineer the distance required for maximal binding by the multivalent ligand is circumvented. In addition, the ligand density can be easily varied by adjusting the proportion of nonglycosylated lipid included in the liposomal preparation. To assess the ability of multivalent liposomal displays to inhibit viral hemagglutination, Whitesides et al. synthesized the sphingosine-inspired, sialic acid functionalized glycolipid 38 (Fig. 23A) [167]. The introduction of 38 into liposomes containing phosphatidylcholine and cholesterol created assemblies that were approximately 104-fold more effective than monovalent inhibitors, on a per-saccharideresidue basis, at inhibiting agglutination. The high inhibitory potency exhibited by this molecular assembly is likely the result of a combination of the chelate effect and steric inhibition as well as other mechanisms. The size of the liposomes may interfere with red blood cell binding because when bound to the virus, these assemblies occupy a large surface area. As observed with the acrylamide glycopolymers, liposomes too may garner added potency by obscuring free neighboring HA sites from interacting with cells. The relative inhibitory potencies of multivalent liposomal displays in a viral infectivity assay did not correspond to those determined in a hemagglutination assay. Bednarski and coworkers designed NeuAc displays composed of glycolipids possessing a dialkyne group within the lipid portion (Fig. 23B) [173]. The liposomal array could be covalently cross-linked by irradiation at 254 nm, which limits saccharide epitope reorganization within the mosaic. Glycolipid 39 was introduced into liposomes along with the corresponding lipid 40 at varying densities. Testing these cross-linked multivalent displays for their abilities to inhibit influenza virus in a hemagglutination assay and a viral plaque forming assay revealed different activity trends for the two assays. Liposomal preparations consisting of 10% glycolipid proved the most potent inhibitors of agglutination, exhibiting a 30,000-fold increase in activity over monovalent analogs. Liposomes containing 1% glycolipid did show some activity, albeit about 10-fold less than the 10% preparation, and they were unable to inhibit agglutination completely. Significantly, the preparation containing 1% glycolipid proved to be the most active at inhibiting viral attachment in the plaque assay. The two assays measure viral attachment to different cell types, red blood cells and MDCK cells, and they monitor two different functions, hemagglutination inhibition and cell lysis. The difference in potencies observed could be explained by differences in ligand presentation on the cells being studied, which affects both virus–cell recognition and the functional consequences of this process. E.

The Affinity of Bivalent Ligands Depends on Linker Length and Flexibility

Because HA is a protein bound to viral membrane, multivalent ligands can interact with a single trimer or by spanning neighboring trimers. Molecules that can place multiple sialic acid residues in binding sites within a trimer might be expected to

258 (A) Whitesides’s sphingosine-inspired glycolipid 38. (B) Bednarski’s cross-linkable liposomal constructs 39 and

Mann and Kiessling

Figure 23 40.

Multivalent Saccharide Displays

259

have more potent activities. Bivalent compounds designed to bind HA in this mode have been synthesized. Several compounds in which the NeuAc residues were connected by various linker units were assembled to explore the role of linker length and flexibility in HA binding [98,160]. Sabesan and coworkers investigated oligosaccharide templates to generate bivalent sialic acid presenting ligands that could occupy multiple sites within a single HA trimer [160]. Heptasaccharides 41–45 , differing only in the linkage pattern to the core galactosyl residue, were assembled using a chemoenzymatic synthesis (Fig. 24). Nuclear Overhauser effect measurements made by 1H NMR spectroscopy and Monte Carlo simulations were used to determine the distances separating the NeuAc residues within the bivalent ligands. Compounds 44 and 45 had the shortest NeuAc ˚ , and the NeuAc residues were spaced 17–19 separation of approximately 9–10 A ˚ A apart in the other analogs. The heptasaccharides were assayed for their ability to inhibit hemagglutination by influenza virus [60]. Modest increases in activity were observed; the most effective compounds in the series, the 3,6-linked (44) and 4,6galactose-linked (45) heptasaccharides, were 10- and 8-fold more potent than monovalent analogs, respectively, when analyzed per sialic acid residue. Given the distances separating the saccharide epitopes, it is unlikely that these compounds bind to different primary sialic acid binding sites within an HA trimer. They may, however, occupy both a primary and a secondary site on HA [18,185]. The importance in multivalent binding of distance spanned and relative ligand geometries observed

Figure 24 Sabesan’s heptasaccharide influenza virus hemagglutinin (HA) inhibitors present bivalent sialic acid linked through disubstituted galactose core.

260

Mann and Kiessling

suggests that proper ligand spacing is important, a result echoed in other lectin– ligand interactions. F.

Bivalent Ligands Function by Binding Neighboring HA Trimers on the Viral Surface

Glick et al. used nonnatural linkers to synthesize a series of bivalent sialosides [98]. Three series of bivalent compounds were generated in which different linker groups were used to connect two NeuAc residues (Fig. 25). The linkers employed were based on ethylene glycol (50), glycine (52), and piperazine (51) and were used to ˚ , possessing varying degrees of assemble bivalent compounds spanning 27 to 65 A conformational flexibility. The binding of these molecules to free HA trimer was measured by an 1H NMR shift assay [186]. None of the compounds had activities greater than that of ␣-methyl NeuAc when tested against free HA trimer. Potency was also measured by inhibition of virus-mediated hemagglutination. The most effective bivalent ligand was derived from the glycine series and possessed a linker with four amino acid residues between each NeuA group and the aromatic ring, 53 (n = 4). This derivative was approximately 400-fold more effective than NeuAc at inhibiting cell agglutination when analyzed on a per-saccharide-residue basis. The best polyethylene-based inhibitor, 50 (n = 2, m = 1) was only eight-fold more potent than NeuAc. These inhibitors ˚ span between the sialic acid moieties. Interestingly, the piperapproximate a 57 A azine series did not yield any active inhibitors even though compound 44 (n = 2, m ˚ span. Conformational flexibility, which is influenced by = 1) also modeled a 57 A the structure of the linker, appears to have an important effect on multivalent HA binding. The distances required for optimal activities in this series are appropriate for occupying multiple binding sites within an HA trimer or for bridging multiple binding sites between HA trimers. Given that different results obtained when the inhibitor activity was compared against HA trimer in solution versus that against the intact virus, the active ligands do not bind to multiple sites within a single HA trimer. Rather, they appear to be capable of spanning trimers on the surface of the virus and/or crosslinking viral particles. This result is similar to that obtained with multivalent ligands for the hepatic lectin (see Section V.C). VII.

TARGETING THE SELECTINS

The selectins are a family of C-type lectins that play key roles in leukocyte recruitment into extravascular tissues during the inflammatory response [187–193]. Leukocytes routinely patrol the vascular endothelial cell walls scanning for sites of tissue damage or infection. When action is required, signals are sent that elicit cellular tethering, rolling, firm adhesion, and subsequent transmigration [194,195]. Selectins E-, L-, and P-mediate the initial leukocyte interaction with the endothelium and facilitate leukocyte rolling. L-selectin is found on leukocytes and recognizes glycoprotein ligands on endothelial cells; E- and P-selectin are found on endothelial cells and recognize glycoproteins on leukocytes. All three selectins weakly bind the tetrasaccharide determinant sialyl Lewis acid x (sLex), 54 (Fig. 26). Unchecked leukocyte extravasation is detrimental to the surrounding tissues and can result in inflammatory

Multivalent Saccharide Displays

261

Figure 25 Bivalent sialic acid dimers target HA by varying linker length. (A) Route to polyethylene glycol and piperazine linked compounds. (B) Route incorporating glycine spacers.

262

Mann and Kiessling

Figure 26

The sialyl Lewis acid x (sLex) determinant 54 is recognized by the selectins.

disorders. The possibility of modulating the inflammatory response at the initial stages has fueled efforts to generate inhibitors targeting this family of lectins. Several reviews highlight efforts to develop selectin inhibitors [17,196,197]. Our focus is on new applications of multivalent displays to target the selectins and inhibit their functions. A.

Multivalent Selectin Ligands

Although soluble forms of the selectins appear to be monomeric in solution, the selectins are transmembrane proteins and their state of clustering on the cell surface in vivo is unknown. Many of the physiological selectin ligands identified to date are glycoproteins that possess multiple copies of their saccharide epitopes [191,192,198,199]. For example, GlyCAM-1, the best-characterized glycoprotein ligand for L-lectin, presents multiple copies of sialylated and sulfated carbohydrate epitopes [200–203]. In addition, L-selectin is clustered on microvilli, suggesting that this family member is poised to interact through multipoint attachment [204,205]. A variety of multivalent ligand mimics that target the selectins have been synthesized to dissect the structural requirements for selectin binding and create high-affinity inhibitors. These structures range from large macromolecular structures to low molecular weight multivalent ligands [175,206–225]. 1.

Low Molecular Weight Multivalent Ligands Show Some Affinity Enhancement for the Selectins

Low molecular weight multivalent ligands have been designed to target the selectins, with most efforts focusing on E-selectin. Overall, these bi- or trisubstituted derivatives are slightly more effective than their monovalent counterparts in static binding assays, with typical affinity enhancements of three- to five-fold. For example, Defrees et al. used a chemoenzymatic approach to synthesize bivalent sialyl Lewis acid x derivatives in which the saccharides were linked by either a carbohydrate or other acyclic spacer [207]. The most effective inhibitor of E-selectin in a static cell adhesion assay (inhibition of HL-60 cells binding to immobilized E-selectin) was that in which two sLex epitopes were linked through the 3- and 6-positions of galactose, 55 (Fig. 27). This ligand was approximately five-fold more potent than monovalent sLex. Ligands in which the sLex groups were tethered with more flexible polyethylene

Multivalent Saccharide Displays

263

Bivalent sLex ligands generated by Wong. The selectin epitopes were linked through the 3- and 6-positions of galactose (55) and via a polyethylene glycol chain (56).

Figure 27

glycol spacers exhibited similar activities in a related assay; the best ligand, 56, was approximately five-fold more active than monovalent sLex (Fig. 27) [226]. Other bivalent derivatives with flexible linking groups afforded similar results [221]. Trivalent sLex derivatives reported by Kretzschmar and coworkers afforded similar enhancements in affinity (Fig. 28). Molecules displaying sLex epitopes on glycopeptide backbones exhibited higher activities than did those possessing more flexible scaf-

264

Figure 28

Mann and Kiessling

Kretzchmar’s trivalent sLex conjugate.

folding. For example, trivalent glycopeptides were up to 10-fold more active than their monovalent counterparts [209,216,223]. Bivalent ligands displaying nonnatural residues, mannose biphenyl glycosides, were devised by Kogan et al. [225] to mimic di-sLex, an oligosaccharide that appears to bind more tightly to E-selectin than sLex. Compounds in a series with variable spacer residues connecting the mannose moieties were tested against all three selectins in a static cell binding assay. The most active ligand, 58, demonstrated increased potencies (ca. sixfold for E- and L-selectin) consistent with binding within a single lectin domain (Figure 29) [225]. Trivalent ligands in this series were no more potent than their bivalent counterparts. The modest inhibitory potency increases observed with the small multivalent ligands explored to date are likely due to secondary interactions near the binding site.

Figure 29

Kogan’s biphenyl-based sLex mimic 58, made bivalent through an alkyl linker.

Multivalent Saccharide Displays

265

Attempts to target L-selectin with large multivalent oligosaccharides have resulted in the identification of ligands with higher functional affinities. Renkonen and colleagues have prepared bi- and tetravalent sLex ligands in which the selectin epitope is linked to N-acetyllactosamine and polylactosamine through the 3- and 6positions of galactose (Fig. 30) [215,217]. Whereas bivalent ligand 59 demonstrated modest inhibitory potency increases, approximately twofold, against L-selectin-mediated adhesion of lymphocytes to endothelial cells, the tetravalent sLex analog, 60, had increased potency, of approximately 50- to 100-fold [217]. The tetravalent sLex derivative, composed of 22 saccharide residues, may be able to span the distance required to bind to multiple L-selectin receptors on the cell surface. Alternatively, the tetravalent ligand may be able to occupy distal binding sites within a single Lselectin molecule. Nonfucosylated derivatives did not inhibit binding, suggesting that inhibition is dependent on the presence of the sLex epitope. 2.

Multivalent Ligands Are Effective Selectin Inhibitors

Although low molecular weight displays have not exhibited high functional affinities for the selectins, larger multivalent ligands are more effective inhibitors. For exam-

Renkonen’s bi- and tetravalent sLex displays 59 and 60, made using chemoenzymatic synthesis.

Figure 30

266

Mann and Kiessling

ple, bovine serum albumin (BSA) conjugates bearing 16 sLex/BSA surpassed monovalent sLex at inhibiting HL-60 cells. They bound to immobilized E-selectin with reported IC50 values of 1 ␮M (16 ␮M saccharide residue concentration) versus 1 mM for sLex [208]. The BSA conjugate also inhibited neutrophil rolling on human umbilical vein endothelial cells (HUVECs) more effectively than sLEx (an IC50 value of 16 ␮M vs. ⬃0.3 mM). Similarly, a polyacrylamide substituted with 3⬘-sulfoLewisx epitopes inhibited the binding of an L-selectin–IgG chimera to immobilized 3⬘-sulfo-Lewisx with IC50 values around 1 ␮M (vs. >1 mM for sLex) [211,212,224]. ROMP generated polymers bearing the simple sulfated saccharide derivatives 3sulfo- and 3,6-disulfogalactose were effective inhibitors of E-, P-, and L-selectin in a cell binding assay. Significantly, the 3,6-disulfogalactose-substituted derivative specifically targeted P-selectin. The IC50 of the polymer was 84 ␮M, as determined on a per-saccharide-residue basis, while the IC50 of sLex was 3.4 mM [227]. Large liposomal arrays are also highly effective inhibitors of the selectins. For instance, multivalent liposomes bearing sLex inhibited binding of HL-60 cells to immobilized E-selectin with potencies 750-fold higher than monovalent analogs [228]. Other liposomal sLex displays with enhanced affinities also have been reported [210,223,229]. These higher order multivalent displays can act by several mechanisms including multisite binding and statistical and steric effects. Factors governing specificity and affinity for the lectins discussed earlier are also relevant in these examples. B.

Inhibitory Potencies of Multivalent Ligands for the Selectins Are Influenced by Assay Method

It is difficult to develop an appropriate assay for all potential selectin ligands because multivalent ligands function by means of a very broad range of mechanisms. An added complication is that the selectins function under conditions of shear flow. Conducting high-throughput assays under conditions of flow is complex, and therefore the activities of potential ligands are generally evaluated under static conditions. A widely employed method involves monitoring the abilities of molecules to inhibit cell binding to an immobilized selectin (Fig. 8). This assay is convenient and can lead to the identification of potential lead structures; however, the cell surface presentation of the selectins may be quite important for their function [205,230–233]. Moreover, the static assay conditions may not reproduce ligand activities that might be observable under flow conditions [95]. 1.

The Presentation of the Receptor Is Important for Assessing Multivalent Ligand Activity

The importance of selectin presentation was illustrated in experiments conducted by Stahn et al., who compared the E-selectin inhibitory activity of BBA2, an anti-Eselectin monoclonal antibody, to that of sLex-decorated liposomes in two different static assays [223]. When the agents were evaluated for inhibition of HepG2 cell binding to E-selectin immobilized in a microtiter well, the antibody and the liposomes had similar activities. The antibody, however, was approximately 100-fold more active than the liposomal preparation at inhibiting binding of HepG2 cells to activated HUVEC cells expressing E-selectin. These results demonstrate the need to examine multiple assays and to critically evaluate the data obtained.

Multivalent Saccharide Displays

2.

267

Static Assays Afford Results Different from Those Carried Out Under Dynamic Conditions

Many biological processes take place under dynamic rather than static conditions. For example, the ability of the selectins to mediate the dynamic process of leukocyte rolling is a critical aspect of their function; yet most selectin inhibitors have been evaluated in static assays. A study evaluating multivalent L-selectin inhibitors revealed dramatic differences in their activities in static cell free versus dynamic cell rolling assays. To explore the recognition of multivalent ligands by L-selectin, the ring-opening metathesis polymerization method (ROMP) was used to generate materials presenting Lex epitopes differing in sulfation pattern [95]. These compounds were designed to mimic the glycoprotein mucins that bind L-selectin, such as GlyCAM-1, by displaying multiple copies of sulfated saccharide epitopes on an extended polypeptide backbone. Oligomers bearing 3⬘,6⬘-disulfo Lewis x(Glc) (61) and 3⬘,6-disulfo-Lewis x(Glc) (62) were more potent than their monovalent counterparts, 63 and 64, in assays to determine their ability to inhibit an L-selectin–IgG fusion protein binding to immobilized heparin (Fig. 31). However, their abilities to inhibit the rolling of L-selectin-transfected cells on GlyCAM-1 were very different. Only the 3⬘,6disulfo-Lewis x(Glc) polymer 62 (Fig. 31) was able to prevent rolling, and it did so with high efficacy (IC50 = 1 ␮M; 15 ␮M based on saccharide residue concentration). These results suggest that fluid flow may play an important role in the specificity of selectin–ligand interactions. Intriguingly, multivalent ligand 62 may bind cell surface L-selectin through multipoint attachment. Thus, multivalent synthetic ligands along with their natural counterparts may possess the ability to cluster L-selectin on the cell surface (Fig. 5C). 3.

Synthetic, Multivalent Ligands Designed to Induce L-Selectin Shedding

The release of a cell surface protein, often termed shedding, can serve to regulate its density on the cell surface and also affects the concentration of the soluble form in circulation [234]. L-selectin shedding is believed to occur during the inflammatory response, and this process may be involved in regulating cell rolling rates [187]. We hypothesized that multivalent ligand binding may cluster L-selectin and thereby trigger shedding. To test this idea, the ability of the 3⬘,6⬘-disulfo-Lewis x polymer 62 to cause L-selectin release from the cell surface was evaluated [41]. The amount of L-selectin on the cell surface of neutrophils after treatment with ligand was assessed by flow cytometry using the anti-L-selectin antibody, DREG56. Although the monovalent compound 64 had no effect on L-selectin levels, exposure of neutrophils to polymeric compound 62 led to a reduction in the amount of cell surface L-selectin. The ability of multivalent ligand 62 to induce L-selectin shedding suggests that natural ligands might function similarly. The demonstration that synthetic ligands can be designed to promote the shedding of a cell surface receptor has implications for understanding the mechanisms cells use to control protein display and for manipulating the concentration of cell surface receptors. The ability to trigger this response by exposure to a synthetic multivalent ligand also suggests a new therapeutic strategy for targeting the inflammatory response.

268 Mann and Kiessling

Figure 31 Kiessling’s disulfated Lex constructs 63 and 64 elicited different physiological effects when appended on multivalent displays to afford 61 and 62.

Multivalent Saccharide Displays

VIII.

269

SUMMARY

Multivalent protein–carbohydrate interactions are prevalent in biological recognition processes. Despite their important roles, an understanding of these interactions is only beginning to emerge. This understanding is facilitated by new developments in the chemical synthesis of multivalent ligands and the use of those ligands to explore and manipulate multivalent recognition processes. Applications for such ligands are widespread, ranging from the investigation of cellular recognition to agents for tissue-selective drug delivery to molecules that can be used to evoke cellular responses. With a growing appreciation of the importance of multivalent interactions, these systems offer intriguing opportunities and challenges for chemistry and biology.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.

WI Weis, K Drickamer. Annu Rev Biochem 65:441–473, 1996. A Varki. Glycobiology 3:97–130, 1993. YC Lee, RT Lee. Acc Chem Res 28:321–327, 1995. K Drickamer, ME Taylor. Annu Rev Cell Biol 9:237–264, 1993. H Lis, N Sharon. Chem Rev 98:637–674, 1998. RA Dwek. Chem Rev 96:683–720, 1996. HJ Gabius. Eur J Biochem 243:543–576, 1997. LL Kiessling, NL Pohl. Chem Biol 3:71–77, 1996. R Roy. Curr Opin Struct Biol 6:692–702, 1996. EJ Toone. Curr Opin Struct Biol 4:719–728, 1994. CR MacKenzie, T Hirama, SJ Deng, DR Bundle, SA Narang, NM Young. J Biol Chem 271:1527–1533, 1996. M Mammen, EI Shakhnovich, JM Deutch, GM Whitesides. J Org Chem 63:3821– 3830, 1998. PR Crocker, T Feizi. Curr Opin Struct Biol 6:679–691, 1996. SYC Wong. Curr Opin Struct Biol 5:599–604, 1995. M Mammen, SK Choi, GM Whitesides. Angew Chem Int Ed Engl 37:2755–2794, 1998. NV Bovin, HJ Gabius. Chem Soc Rev 24:413–421, 1995. EE Simanek, GJ McGarvey, JA Jablonowski, C-H Wong. Chem Rev 98:833–862, 1998. W Weis, JH Brown, S Cusack, JC Paulson, JJ Skehel, DC Wiley. Nature 333:426–431, 1988. Z Derewenda, J Yriv, JR Helliwell, AJ Kalb, EJ Dodson, MZ Papiz, T Wan, J Campbell. EMBO J 8:2189–2193, 1989. R Loris, D Maes, F Poortmans, L Wyns, J Bouckaert. J Biol Chem 271:30614–30618, 1996. RU Lemieux. ACS Symp Ser 519:5–18, 1993. RU Lemieux. Acc Chem Res 29:373–380, 1996. DK Mandal, N Kishore, CF Brewer. Biochemistry 33:1149–1156, 1994. R Loris, PPG Stas, L Wyns. J Biol Chem 269:26722–26733, 1994. KK-S Ng, S Park-Snyder, WI Weis. Biochemistry 37:17965–17976, 1998. FA Quiocho. Annu Rev Biochem 55:287–315, 1986. A Varki. Glycobiology 6:707–710, 1996. S Faham, RJ Linhardt, DC Rees. Curr Opin Struct Biol 8:578–586, 1998. O Blanck, ST Iobst, C Gabel, K Drickamer. J Biol Chem 271:7289–7292, 1996. RU Lemieux, M-H Du, U Spohr. J Am Chem Soc 116:9803–9804, 1994.

270

Mann and Kiessling

31. 32. 33. 34.

KH Mortell, RV Weatherman, LL Kiessling. J Am Chem Soc 118:2297–2298, 1996. DM Crothers, H Metzger. Immunochemistry 9:341–357, 1972. GI Bell. Nature 248:430–431, 1974. JA Berzofsky, IJ Berkower, SL Epstein: In: WE Paul, ed. Fundamental Immunology. New York: Raven Press, 1993. MI Page, WP Jencks. Proc Natl Acad Sci USA 68:1678–1683, 1971. WP Jencks. Proc Natl Acad Sci USA 78:4046–4050, 1981. JH Rao, J Lahiri, L Isaacs, RM Weis, GM Whitesides. Science 280:708–711, 1998. R Liang, J Loebach, N Horan, M Ge, C Thompson, L Yan, D Kahne. Proc Natl Acad Sci USA 94:10554–10559, 1997. KM Muller, KM Arndt, A Pluckthun. Anal Biochem 261:149–158, 1998. EJ Gordon, LE Strong, LL Kiessling. Bioorg Med Chem 6:1293–1299, 1998. EJ Gordon, WJ Sanders, LL Kiessling. Nature 392:30–31, 1998. JM Rini. Annu Rev Biophys Biomol Struct 24:551–577, 1995. CS Wright. J Mol Biol 178:91–104, 1984. RV Weatherman, KH Mortell, M Chervenak, LL Kiessling, EJ Toone. Biochemistry 35:3619–3624, 1996. DK Mandal, L Bhattacharyya, SH Koenig, RD Brown, S Oscarson, CF Brewer. Biochemistry 33:1157–1162, 1994. T Sokolowski, T Peters, S Perez, A Imberty. J Mol Graph 15:37–38, 1997. Y Bourne, J Mazurier, D Legrand, P Rouge, J Montreuil, G Spik, C Cambillau. Structure 2:209–219, 1994. D Sako, KM Comess, KM Barone, RT Camphausen, DA Cumming, GD Shaw. Cell 83:323–331, 1995. T Pouyani, B Seed. Cell 63:333–343, 1995. SK Choi, M Mammen, GM Whitesides. Chem Biol 3:97–104, 1996. CF Anderson, MT Record. Annu Rev Biophys Biophys Chem 19:423–465, 1990. EG Weinhold, JR Knowles. J Am Chem Soc 114:9270–9275, 1992. GS Jacob, C Kirmaier, SZ Abbas, SC Howard, CN Steininger, JK Welply, P Scudder. Biochemistry 34:1210–1217, 1995. RV Weatherman, LL Kiessling. J Org Chem 61:534–538, 1996. LM Killingsworth, J Savory. Clin Chem 18:335–339, 1972. A Boyden, E Botton, D Gemeroy. J Immunol 57:211, 1947. R Roy, D Page´, SF Perez, VV Bencomo. Glycoconjugate J 15:251–263, 1998. M-C Shao, CQ Chin. In: YC Lee, RT Lee, eds. Methods in Enzymology, vol 247. San Diego, CA: Academic Press, 1994, pp 253–262. V Horesji, V Matousek. Mol Immunol 22:125–133, 1985. TJ Pritchett, R Brossmer, U Rose, JC Paulson. Virology 160:502–506, 1987. O Hayashida, C Shimizu, T Fujimoto, Y Aoyama. Chem Lett 1:13–14, 1998. NL Kalinin, LD Ward, DJ Winzor. Anal Biochem 228:238–244, 1995. GM Kuziemko, M Stroh, RC Stevens. Biochemistry 35:6375–6384, 1996. DG Myszka. Curr Opin Biotechnol 8:50–57, 1997. MW Nicholson, AN Barclay, MS Singer, SD Rosen, PA van der Merwe. J Biol Chem 2:763–770, 1998. Y Shinohara, H Sota, F Kim, M Shimizu, M Gotoh, M Tosu, Y Hasegawa. J Biochem 117:1076–1082, 1995. MM Morelock, RH Ingraham, R Betageri, S Jakes. J Med Chem 38:1309–1318, 1995. DA Mann, M Kanai, DJ Maly, LL Kiessling. J Am Chem Soc 120:10575–10582, 1998. L Nieba, A Krebber, A Plu¨ckthun. Anal Biochem 234:155–165, 1996. G Hester, CS Wright. J Mol Biol 262:516–531, 1996. WI Weis. Curr Opin Struct Biol 7:624–630, 1997. CS Wright. Curr Opin Struct Biol 7:631–636, 1997.

35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72.

Multivalent Saccharide Displays 73. 74. 75. 76. 77. 78. 79. 80. 81.

82. 83.

84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106.

271

H Bittiger, HP Schnebli, eds. Concanavalin A as a Tool. London: John Wiley & Sons, 1976. IJ Goldstein. In: H Bittiger, HP Schnebli, eds. Concanavalin A as a Tool. London: John Wiley & Sons, 1976, pp 55–65. BA Williams, MC Chervenak, EJ Toone. J Biol CHem 267:22907–22911, 1992. MC Chervenak, EJ Toone. Biochemistry 34:5685–5695, 1995. DK Mandal, CF Brewer. Biochemistry 31:12602–12609, 1992. F Brewer, L Bhattacharyya, RD Brown, SH Koenig. Biochem Biphys Res Commun 127:1066–1071, 1985. JH Naismith, RA Field. J Biol Chem 972–976, 1996. DK Mandal, CF Brewer. Biochemistry 32:5117–5120, 1993. IJ Goldstein, RD Poretz. In: IE Liener, N Sharon, IJ Goldstein, eds. The Lectins: Properties, Functions and Applications in Biology and Medicine. New York: Academic Press, 1986, pp 33–244. DM Lynn, S Kanaika, RH Grubbs. J Am Chem Soc 118:784–790, 1996. LL Kiessling, LE Strong. In: A Fu¨rstner, ed. Alkene Metathesis in Organic Synthesis, Topics in Organometallic Chemistry, vol 1. New York: Springer-Verlag, 1998, pp 200– 231. M Kanai, KH Mortell, LL Kiessling. J Am Chem Soc 119:9931–9932, 1997. K Tagawa, N Sendai, K Ohno, T Kawaguchi, H Kitano. Bioconjugate Chem 10:354– 360, 1999. PR Ashton, EF Hounsell, N Jayaraman, TM Nilsen, N Spencer, JF Stoddart, M Young. J Org Chem 63:3429–3437, 1998. D Page´, S Aravind, R Roy. J Chem Soc Chem Commun 1913–1914, 1996. D Page´, R Roy. Bioconjugate Chem 8:714–723, 1997. D Page´, R Roy. Bioorg Med Chem Lett 6:1765–1770, 1996. D Page´, R Roy. Glycoconjugate J 14:345–356, 1997. D Page´, R Roy, SF Perez, VV Bencomo. Glycoconjugate J 15:251–263, 1998. H Imata, K Kubota, K Hattori, A Masaaki, C Jindoh. Bioorg Med Chem Lett 7:109– 112, 1997. H Yuasa, Y Kamata, S Kurono, H Hironobu. Bioorg Med Chem Lett 8:2139–2144, 1998. RT Lee, YC Lee. Glycoconjugate J 4:317–328, 1987. WJ Sanders, EJ Gordon, O Dwir, PJ Beck, R Alon, LL Kiessling. J Biol Chem 274: 5271–5278, 1999. RT Lee, P Lin, YC Lee. Biochemistry 23:4255–4261, 1984. YC Lee, RR Townsend, MR Hardy, J Lonngren, J Arnarp, M Haraldsson, H Lo¨nn. J Biol Chem 258:199–202, 1983. GD Glick, PL Toogood, DC Wiley, JJ Skehel, JR Knowles. J Biol Chem 266:23660– 23669, 1991. M Mammen, G Dahmann, GM Whitesides. J Med Chem 38:4179–4190, 1995. SK Choi, M Mammen, GM Whitesides. J Am Chem Soc 199:4103–4111, 1997. MN Matrosovich, LV Mochalova, VP Marinina, NE Byramova, NV Bovin. FEBS Lett 272:209–212, 1990. A Dessen, D Gupta, S Sabesan, CF Brewer, JC Sacchettini. Biochemistry 34:4933– 4942, 1995. D Gupta, S Oscarson, TS Raju, P Stanley, EJ Toone, CF Brewer. Eur J Biochem 242: 320–326, 1996. TK Dam, S Oscarson, CF Brewer. J Biol Chem 273:32812–32817, 1998. DA Rozwarski, BM Swami, CF Brewer, JC Sacchettini. J Biol Chem 273:32818– 32825, 1998. DN Moothoo, JH Naismith. Glycobiology 8:173–181, 1998.

272

Mann and Kiessling

107. 108.

M Spiess. Biochemistry 29:1009–1018, 1990. DT Connolly, RR Townsend, K Kawaguchi, WR Bell, YC Lee. J Biochem 257:939– 945, 1982. RT Lee, KG Riec, NBN Rao, Y Ichikawa, T Barthel, V Piskarev, YC Lee. Biochemistry 28:8351–8358, 1989. MS Kuhlenschmidt, E Schmell, CW Slife, TB Kuhlenschmidt, F Sieber, YC Lee, S Roseman. J Biol Chem 257:3157–3164, 1982. J Bischoff, HF Lodish. J Biol Chem 262:11825–11832, 1987. J Bischoff, S Libresco, MA Shia, HF Lodish. J Cell Biol 106:1067–1074, 1988. K Drickamer. Curr Opin Struct Biol 3:393–400, 1993. MD Bider, M Spiess. FEBS Lett 434:37–41, 1998. NI Ruiz, K Drickamer. Glycobiology 6:551–559, 1996. L Burrows, ST Iobst, K Drickamer. Biochem J 324:673–680, 1997. JU Baenziger, Y Maynard. J Biol Chem 255:4607–4613, 1980. JU Baenziger, D Fiete. Cell 22:611–620, 1980. PH Weigel, RL Schnarr, MS Kuhlenschmidt, EL Schmell, RT Lee, YC Lee, S Roseman. J Biol Chem 254:10830–10838, 1979. RT Lee, YC Lee. Bioconjugate Chem 8:762–765, 1997. D Yi, RT Lee, P Longo, ET Boger, YC Lee, WA Petri, RL Schnaar. Glycobiology 8: 1037–1043, 1998. A Kichler, F Schuber. Glycoconjugate J 12:275–281, 1995. AL Schwartz, SE Fridovich, HF Lodish. J Biol Chem 257:4230–4237, 1982. K Ozaki, RT Lee, YC Lee, T Kawasaki. Glycoconjugate J 12:268–274, 1995. MS Wadhwa, KG Rice. J Drug Targeting 3:111–127, 1995. G Molema, DKF Meijer. Adv Drug Delivery Rev 14:25–50, 1994. Y Kato, Y Sugiyama. Ther Drug Carrier Syst 14:287–331, 1997. S Gabius, K Kayser, NV Bovin, N Yamazaki, S Kojima, H Kaltner, HJ Gabius. Eur J Pharm Biopharm 42:250–261, 1996. H Sato, Y Sugiyama, A Tsuji, I Horikoshi. Adv Drug Delivery Rev 19:445–467, 1996. DKF Meijer, G Molema. Semin Liver Dis 15:202–256, 1995. M Monsigny, A-C Roche, P Midoux, R Mayer. Adv Drug Delivery Rev 14:1–24, 1994. RM Smith, GY Wu. Semin Liver Dis 19:83–92, 1999. SI Michael, DT Curiel. Gene Ther 1:223–232, 1994. EAL Biessen, H Broxterman, JH Vanboom, TJC Vanberkel. J Med Chem 38:1846– 1852, 1995. JJ Hangeland, JT Levis, YC Le, POP Tso. Bioconjugate Chem 6:695–701, 1995. JJ Hangeland, JT Levis, YC Lee, JJ Frost, POP Tso. Biophys J 70:TU210–TU210, 1996. JR Merwin, GS Noell, WL Thomas, HC Chiou, ME Derome, TD McKee, GL Spitalny, MA Findeis. Bioconjugate Chem 5:612–620, 1994. R Plourde, AT Phillips, CH Wu, RM Hays, JR Chowdhury, NR Chowdhury, GY Wu. Bioconjugate Chem 7:131–137, 1996. L Fiume, C Busi, GD Stefano, A Mattioli. Adv Drug Delivery Rev 14:51–65, 1994. TJC Van Berkel, JK Kruijt, HH Spanjer. JF Nagelkerke, L Harkes, HJ Kempen. J Biol Chem 260:2694–2699, 1985. HH Spanjer, GL Scherphof. Biochim Biophys Acta 734:40–47, 1983. GY Wu, CH Wu. J Biol Chem 262:4429–4432, 1987. GY Wu, CH Wu. J Biol Chem 267:12436–12439, 1992. GY Wu, CH Wu. J Biol Chem 263:14621–14624, 1988. GY Wu, CH Wu. Biochemistry 27:887–892, 1988. P Midoux, C Mendes, A Legrand, J Raimond, R Mayer, M Monsigny, AC Roche. Nucleic Acids Res 21:871–878, 1993.

109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146.

Multivalent Saccharide Displays 147. 148. 149. 150. 151.

152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179.

273

P Erbacher, AC Roche, M Monsigny, P Midoux. Bioconjugate Chem 6:401–410, 1995. J Haensler, F Schuber. Glycoconjugate J 8:116–124, 1991. JS Remy, A Kichler, V Mordvinov, F Schuber, JP Behr. Proc Natl Acad Sci USA 92: 1744–1748, 1995. N Murahashi, A Sasaki. Biol Pharmacol Bull 19:418–423, 1996. L Sliedregt, PCN Rensen, ET Rump, PJ van Santbrink, MK Bijsterbosch, A Valentijn, GA van der Marel, JH van Boom, TJC van Berkel, EAL Biessen. J Med Chem 42: 609–618, 1999. C Plank, K Zatloukal, M Cotten, K Mechtler, E Wagner. Bioconjugate Chem 3:533– 539, 1992. TL Lentz. J Gen Virol 71:751–766, 1990. T Stegmann, JM White, A Helenius. EMBO J 9:4231–4241, 1990. DC Wiley, JJ Skehel. Annu Rev Biochem 56:365–394, 1987. NK Sauter, JE Hanson, GD Glick, JH Brown, RL Crowther, SJ Park, JJ Skehel, DC Wiley. Biochemistry 31:9609–9621, 1992. S Sabesan, JO Duus, P Domaille, S Kelm, JC Paulson. J Am Chem Soc 113:5865– 5866, 1991. GD Glick, JR Knowles. J Am Chem Soc 113:4701–4703, 1991. PL Toogood, PK Galliker, GD Glick, JR Knowles. J Med Chem 34:3138–3140, 1991. S Sabesan, JO Duus, S Neira, P Domaille, S Kelm, JC Paulson, K Bock. J Am Chem Soc 114:8363–8375, 1992. Y Ito, J Gaudino, JC Paulson. Pure Appl Chem 65:753–762, 1993. A Gamian, M Chomik, CA Laferriere, R Roy. Can J Microbiol 37:233–237, 1991. R Roy, D Zanini, SJ Meunier, AV Romanowska. J Chem Soc Chem Commun 1869– 1872, 1993. D Zanini, R Roy. J Org Chem 61:7348–7354, 1996. R Roy, FO Andersson, G Harms, S Kelm, R Schauer. Angew Chem Int Ed Engl 31: 1478–1481, 1992. LV Mochalova, AB Tuzikov, VP Marinina, AS Gambaryan, NE Byramova, NV Bovin, MN Matrosovich. Antiviral Res 23:179–190, 1994. JE Kingerywood, KW Williams, GB Sigal, GM Whitesides. J Am Chem Soc 114: 7303–7305, 1992. WJ Lees, A Spaltenstein, JE Kingerywood, GM Whitesides. J Med Chem 37:3419– 3433, 1994. GB Sigal, M Mammen, G Dahmann, GM Whitesides. J Am Chem Soc 118:3789– 3800, 1996. A Spaltenstein, GM Whitesides. J Am Chem Soc 113:686–687, 1991. MA Sparks, KW Williams GM Whitesides, 778–783, 1993; J Med Chem 36:778–783, 1993. JO Nagy, P Wang, JH Gilbert, ME Schaefer, TG Hill, MR Callstrom, MD Bednarski. J Med Chem 35:4501–4502, 1992. W Spevak, JO Nagy, DH Charych, ME Schaefer, JH Gilbert, MD Bednarski. J Am Chem Soc 115:1146–1147, 1993. R Roy, CA Laferriere. Carbohydr Res 177:C1–C4, 1988. NE Nifantev, YE Tsvetkov, AS Shashkov, LO Kononov, VM Menshov, AB Tuzikov, NV Bovin. J Carbohydr Chem 15:939–953, 1996. S Aravind, WKC Park, S Brochu, R Roy. Tetrahedron Lett 35:7739–7742, 1994. T Furuike, N Nishi, S Tokura, SI Nishimura. Macromolecules 28:7241–7247, 1995. K Kobayashi, A Tsuchida, T Usui, T Akaike. Macromolecules 30:2016–2020, 1997. WJ Zhou, ME Wilson, MJ Kurth, YL Hsieh, JM Krochta, CF Shoemaker. Macromolecules 30:7063–7068, 1997.

274

Mann and Kiessling

180.

K Ohno, Y Tsujii, T Miyamoto, T Fukuda, M Goto, K Kobayashi, T Akaike. Macromolecues 31:1064–1069, 1998. R Roy, FD Tropper, A Romanowska. J Chem Soc Chem Commun 1611–1613, 1992. PH Weigel, RL Schnaar, MS Kuhlenschmidt, E Schmell, RT Lee, YC Lee, SJ Roseman. J Biol Chem 254:10830–10838, 1979. RL Schnaar, PH Weigel, MS Kuhlenschmidt, YC Lee, S Roseman. J Biol Chem 253: 7940–7951, 1978. H Kamitakahara, T Suzuki, N Nishigori, Y Suzuki, O Kanie, C-H Wong. Angew Chem Int Ed Engl 37:1524–1528, 1998. NK Sauter, GD Glick, RL Crowther, S-J Park, MB Eisen, JJ Skehel, JR Knowles. Proc Natl Acad Sci USA 89:324–328, 1992. NK Sauter, MD Bednarski, BA Wurzburg, JE Hanson, GM Whitesides, JJ Skehel, DC Wiley. Biochemistry 28:8388–8396, 1989. GS Kansas. Blood 88:3259–3287, 1996. LA Lasky. Annu Rev Biochem 64:113–139, 1995. D Vestweber. J Cell Biochem 61:585–591, 1996. KC Chou. FEBS Lett 363:123–126, 1995. SD Rosen, CR Bertozzi. Curr Opin Cell Biol 6:663–673, 1994. A Varki. Proc Natl Acad Sci USA 91:7390–7397, 1994. RP McEver. Curr Opin Immunol 6:75–84, 1994. EC Butcher. Cell 67:1033–1036, 1991. TA Springer. Cell 76:301–314, 1994. HC Kolb, B Ernst. Pure Appl Chem 69:1879–1884, 1997. CR Bertozzi. Chem Biol 2:703–708, 1995. A Varki. J Clin Invest 99:158–162, 1997. RP McEver, KL Moore, RD Cummings. J Biol Chem 270:11025–11028, 1995. S Hemmerich, SDB Rosen. Biochemistry 33:4830–4835, 1994. S Hemmerich, CR Bertozzi, H Leffler, SD Rosen. Biochemistry 33:4820–4829, 1994. CR Bertozzi, S Fukuda, SD Rosen. Biochemistry 34:14271–14278, 1995. D Crommie, SD Rosen. J Biol Chem 270:22614–22624, 1995. UH von Andrian, SR Hasslen, RD Nelson, SL Erlandsen, EC Butcher. Cell 82:989– 999, 1995. RE Bruehl, TA Springer, DF Bainton. J Histochem Cytochem 44:835–844, 1996. SD DeFrees, FCA Gaeta, Y-C Lin, Y Ichikawa, C-H Wong. J Am Chem Soc 115: 7549–7550, 1993. SD Defrees, W Kosch, W Way, JC Paulson, S Sabesan, RL Halcomb, DH Huang, Y Ichikawa, C-H Wong. J Am Chem Soc 117:66–79, 1995. JK Welply, SZ Abbas, P Scudder, JL Keene, K Broschat, S Casnocha, C Gorka, C Steininger, SC Howard, JJ Schmuke, M Graneto, JM Rotsaert, ID Manger, GS Jacob. Glycobiology 4:259–265, 1994. U Sprengard, M Schudok, W Schmidt, G Kretzschmar, H Kunz. Angew Chem Int Ed Engl 35:321–324, 1996. W Spevak, C Foxall, DH Charych, F Dasgupta, JO Nagy. J Med Chem 39:1018–1020, 1996. G Weitz-Schimdt, D Stokmaier, G Scheel, NE Nifantev, AB Tuzikov, NV Bovin. Anal Biochem 238:184–190, 1996. SM Game, PK Rajapurohit, M Clifford, MI Bird, R Priest, NV Bovin, NE Nifant’ev, G O’Beirne, ND Cook. Anal Biochem 258:127–135, 1998. G Kretzschmar, U Sprengard, H Kunz, E Bartnik, W Schmidt, A Toepfer, B Horsch, M Krause, D Seiffge. Tetrahedron 51:13015–13030, 1995. A Seppo, JP Turunen, L Penttila, A Keane, O Renkonen, R Renkonen. Glycobiology 6:65–71, 1996.

181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194. 195. 196. 197. 198. 199. 200. 201. 202. 203. 204. 205. 206. 207. 208.

209. 210. 211. 212. 213. 214.

Multivalent Saccharide Displays 215. 216. 217. 218. 219. 220. 221. 222. 223. 224. 225. 226. 227. 228. 229.

230. 231. 232. 233. 234.

275

H Maaheimo, R Renkonen, JP Turunen, L Penttila, O Renkonen. Eur J Biochem 234: 616–625, 1995. G Baisch, R Ohrlein. Angew Chem Int Ed Engl 35:1812–1815, 1996. O Renkonen, S Toppila, L Penttila, H Salminen, J Helin, H Maaheimo, CE Costello, JP Turunen, R Renkonen. Glycobiology 7:453–461, 1997. G Thoma, F Schwarzenbach, RO Duthaler. J Org Chem 61:514–524, 1996. G Thoma, JL Magnani, R Ohrlein, B Ernst, F Schwarzenbach, RO Duthaler. J Am Chem Soc 119:7414–7415, 1997. H Miyauchi, M Tanaka, H Koike, N Kawamura, M Hayashi. Bioorg Med Chem Lett 7:985–988, 1997. H Miyauchi, M Yuri, M Tanaka, N Kawamura, M Hayashi. Bioorg Med Chem Lett 7: 989–992, 1997. MM Palcic, H Li, D Zanini, RS Bhella, R Roy. Carbohydr Res 305:433–442, 1997. R Stahn, H Schafer, F Kernchen, J Schreiber. Glycobiology 8:311–319, 1998. R Roy, WKC Park, OP Srivastava, C Foxall. Bioorg Med Chem Lett 6:1399–1402, 1996. TP Kogan, B Dupre, H Bui, KL McAbee, JM Kassir, IL Scott, X Hu, P Vanderslice, PJ Beck, RAF Dixon. J Med Chem 41:1099–1111, 1998. V Wittmann, S Takayama, KW Gong, G Weitz-Schmidt, C-H Wong. J Org Chem 63: 5137–5143, 1998. DD Manning, LE Strong, X Hu, PJ Beck, LL Kiessling. Tetrahedron 53:11937–11952, 1997. SA DeFrees, L Phillips, L Guo, S Zalipsky. J Am Chem Soc 118:6101–6104, 1996. T Ikami, N Tomiya, T Morimoto, N Iwata, R Yamashita, T Jomori, T Usui, Y Suzuki, H Tanaka, D Miyamoto, H Ishida, A Hasegawa, M Kiso. J Carbohydr Chem 17:499– 518, 1998. JV Stein, GY Cheng, BM Stockton, BP Fors, EC Butcher, UH von Andrian. J Exp Med 189:37–49, 1999. K Sasaki, Y Okouchi, HJ Rothkotter, R Pabst. Acta Anat 162:33–39, 1998. JG Leid, CA Speer, MA Jutila. J Leukocyte Biol 64:104–107, 1998. SR Hasslen, UH Vonandrian, EC Butcher, RD Nelson, SL Erlandsen. Histochem J 27: 547–554, 1995. NM Hooper, EH Karran, AJ Turner. Biochem J 321:265–279, 1997.

9 Designing Novel Multivalent Glycotools for Biochemical Investigations Related to Sialic Acid Rene´ Roy University of Ottawa, Ottawa, Ontario, Canada

I.

INTRODUCTION

Sialic acids constitute a family of more than 30 neuraminic acid derivatives varying by the nature of their substituents (acetyl, glycolyl, phosphate, sulfate) and by their relative positioning around the pyranose ring [1]. The most common member of this rather unusual nine-carbon amino acid sugar (5-amino-3,5-dideoxy-D-glycero-D-galacto-non-2-ulopyranosonic acid) is represented by N-acetylneuraminic acid (Neu5Ac, 1) (sometime abusively coined sialic acid itself ) (Scheme 1). Sialic acids are common mammalian sugars that usually end oligosaccharide sequences of glycolipids (gangliosides), N- and O-linked glycoproteins, and some proteoglycans. They are also found as ␣-(2-8/9)-linked polysaccharides in encapsulated bacteria. As such, sialic acids are forefront carbohydrate haptens responsible for a wide range of recognition events [2]. Cell surface sialosides are involved as anchoring motifs for microbial attachment. Various pathogenic agents such as viruses (influenza, coronavirus, Sendai, polyoma, rotaviruses), bacteria and bacterial toxins (Pseudomonas aeruginosa, Helicobacter pylori, E. coli, Vibrio cholerae, Bordella pertussis; cholera, tetanus toxins), and parasites (Streptococcus suis, Plasmodium falciparum, Trypanosoma cruzi) can adhere to and colonize host tissues after binding to sialosides [3]. It has also been demonstrated that bacterial infections [4] and cancer metastasis [5] can be prevented by blocking receptor sites with high serum carbohydrate concentrations including sialic acid. It is also involved in selectin-mediated cell adhesion related to leukocytes over recruitment to infected or damaged tissues [6]. 277

278

Roy

Scheme 1

Unfortunately, medicinal applications of carbohydrate binding protein inhibitors suffer from major drawbacks. Except for a few marginal cases, carbohydrates show deceptively low binding affinities toward their receptor counterparts [7]. Moreover, simple or more complex oligosaccharides are poorly bioavailable and are rapidly catabolized through various glycosidases. A few strategies are being pursued to overcome these difficulties. Among these, the syntheses of carbohydrate analogs (deoxy, fluoro, epi, etc.), conformationally restricted analogs, and glycomimetics have been more or less successful. One promising example of rational drug design has recently emerged in the field of influenza flu virus neuraminidase inhibitors [8,9]. Compounds that mimic the transition states involved in ␣-sialoside hydrolyses have surpassed the classical lead Neu5Ac2en (2) (Ki 4 ␮M). Thus, Glaxo Wellcome’s Zanamivir (4guanidino-Neu5Ac2en, GG167, 3) with a Ki of 0.1 nM [8] and Gilead Sciences prodrug GS4104 (4, Ki 1 nM) [9] have reached clinical phase II for a long-awaited drug treatment against flu virus infections. GS4104 proved effective against both influenza A and B strains in test animals. Another useful strategy for the design of potent microbial antiadhesins may rely on the rational syntheses of various multivalent glycoforms that can mimic multiantennary glycoproteins. It can be speculated that common oligosaccharide sequences might express further specificity and enhanced avidity through precise multivalent architectural antigen presentation. Numerous cases exist to point toward multivalency or ‘‘cluster’’ effects to ‘‘boost’’ carbohydrate–protein bindings [10]. Multivalent macromolecules would be particularly appealing for respiratory, gastrointestinal, and urinary tract infections where bioavailability does not constitute a major problem. Such novel, nonimmunogenic biopolymers may also find applications as carriers for drug delivery and as affinity adsorbents, both in vitro and in vivo, in cases calling for natural killer cell and macrophage activation. We and others have previously demonstrated that multivalent neoglycoconjugates constitute powerful inhibitors in a number of generally low-affinity carbohydrate–protein binding interactions [11–16]. Potent multivalent neoglycoconjugates

Multivalent Glycotools and Sialic Acid

279

have been scaffolded onto a wide range of carriers including polymers [11], polyamino acids [12], cyclodextrins [13], calix[4]arenes [14], carbohydrates [15], and dendrimers [16]. This last class of scaffolds is particularly appealing because the effects of size, shape, and valency can be controlled at will [17]. This chapter focuses on ongoing activities regarding the syntheses of multivalent sialosides. II.

SOURCES OF SIALIC ACID

The cost for commercially available sialic acid has dramatically declined over the last 10 years. Depending on the amount purchased, the cost can be as low as approximately $CDN 20–$CDN 30/g. A practical source, requiring no particular skills, consists of isolating sialic acid through a mild acid hydrolysis of the Collocalia mucin isolated from edible bird’s nest [18], which can be purchased from any Chinese grocery [$CDN 100/$CDN 200 g). The mucin is essentially a sialic acid–rich glycoprotein. Thus, 10 g (5% w/w) can be obtained in less than a week. Alternatively, sialic acid can be isolated from egg yolk [19], from cow’s milk [20], or from the complete acid hydrolysis of colominic acid, a bacterial polysaccharide produced by both E. coli K1 and Neisseria meningitidis serogroup B or C [␣-(2,8/9)]-polysialic acid [21]. It is also possible to produce several discrete oligomers by controlled acid hydrolysis of the polysaccharides [22]. Additionally, it has been feasible to produce the analogous N-glycolylneuraminic acid (NAc replaced by NCOCH2OH, Neu5Gc) and their corresponding oligomers by strong base hydrolysis of the N-acetyl groups of the polysaccharide (2 M NaOH, 110⬚C, 7 h); treatment with acetoxyacetyl chloride; followed by de-O-acetylation, and hydrolysis. Alternatively, the de-N-acetylated polysaccharide can be treated with acryloyl chloride and then reductively ozonolyzed [23]. This N-glycolylsialic acid analog is the key determinant epitope of the Hanganutziu–Deicher antigen [24] found in some patients who have been treated with animal serum (serum sickness). Moreover, it has been found in large quantity in tumor cells having abnormally high N-glycolyl GM3-ganglioside. A.

Useful Sialic Acid Derivatives as Precursors in Neoglycoconjugate Syntheses

As mentioned above, the sialic acid residue itself constitutes the key immunodominant epitope in several biological interactions. It was therefore appealing to generate a wide range of neoglycoconjugates containing solely sialic acid or its analogous Nglycolylated derivative. Nowadays, sialic acid can be readily prepared as O-, S-, N-, and C-glycosides. Stereospecific access to the first three forms can be achieved with complete anomeric stereocontrol. However, great difficulty still attends the synthesis of pure ␣-linked C-sialosides [25]. Fortunately, reports describing the use of samarium diiodide (SmI2) (Barbier’s conditions) seem to open the door toward improved stereocontrol [26]. The following subsections briefly highlight aspects of our approach toward the preparation of suitably functionalized sialosides. 1.

Stereospecific Anomeric Functionalization Under PTC Conditions

Generally, N-acetylneuraminic acid is ␣-glycosidically linked to D-galactose or to DN-acetylgalactosamine residues through ␣-(2–3/6) linkages or to another N-acetylneuraminic acid residue in ␣-(2–8/9) linkages [1]. In most cases, terminal Neu5Ac

280

Roy

residues constitute the immunodominant epitopes; that is, most binding interactions occur through intrinsic Neu5Ac functionality, with the carboxyl group constituting the key polar group [27]. Based on cumulative observations [1], initial efforts were centered around the syntheses of ␣-sialoside derivatives alone. Anomeric phase transfer catalysis (PTC) has been systematically shown to take place through inversion of configuration through a seemingly SN2-type mechanism [28,29]. In spite of its quaternary nature and the absence of a potential participating group, acetochloroneuraminic acid (5), having an axial ␤-chloro substituent, has always provided clean and high-yielding anomeric inversion with a wide range of nucleophiles under mild PTC conditions [29]. When tetrabutylammonium hydrogen sulfate (TBAHS) was used as phase transfer catalyst, 1 M sodium carbonate as aqueous phase, and ethyl acetate (or dichloromethane) as organic phase, ␤-chloride 5 provided exclusively ␣-sialylated derivatives 6–11 in excellent yields (Scheme 2). The only by-product occasionally formed appeared to be peracetylated Neu5Ac2en (2) derivative, resulting from dehydrochlorination through an elimination side reaction (E2). For instance, treatment of chloride 5 with thioacetic acid, sodium azide, 4nitrophenol, 4-hydroxybenzaldehyde, 4-nitrothiophenol, or thiophenol afforded derivative 6 (66%) [30], 8 (94%) [31], 9a (90%) [32], 9b (65%) [33], 10 (81%) [34], or 11 (80%) [34], respectively. Chemoselective (35) de-S-acetylation of 6 (H2NNH2HOAc, DMF, room temperature < 30 min) or NaOMe in methanol at ⫺40⬚C [20] afforded 1-thioderivative 7 in 88% yield. The foregoing ␣-sialo derivatives were suitably functionalized for coupling to various multivalent carriers (see below). B.

Using Novel ‘‘Active–Latent’’ Glycosylation Strategy to Produce Sialyloligosaccharides

The biological significance of N-acetylneuraminic acid containing oligosaccharides has stimulated remarkable progress in devising glycosylation strategies [36]. However, efficient protocols for building sialylated oligosaccharides from monosaccharide components is still a major challenge. Initially, acetochloroneuraminic acid (5) has been the glycosyl donor most widely used for the synthesis of ␣-sialosides, but its efficacy necessitates the use of a temporarily incorporated participating group at C3 [37]. Phosphite [38] and S-ethyl xanthate [39] derivatives have met some successes, while thioglycosides remain the method of choice [40]. The ‘‘armed’’ (ether pro-

Scheme 2

Multivalent Glycotools and Sialic Acid

281

tecting groups at C2) and ‘‘disarmed’’ (ester protecting groups at C2) n-pentenyl glycosylation strategy introduced by Madsen and Fraser-Reid [41] has proven to be a powerful tool for complex oligosaccharide syntheses that also is applicable to thioglycosides and few other glycosyl donors (reviewed in Ref. 42). Despite the versatility of this approach, there remains opportunity to further fine-tune the leaving ability of the glycosyl donor and thus to realize a greater potential for oligosaccharide syntheses. Another versatile strategy, coined the ‘‘active–latent’’ glycosylation strategy, in which the reactivity of the carbohydrate units can be directly controlled by the aglycone, together with the differential protection of the glycone itself, enabled the synthesis of various oligosaccharide donors [34,43–47]. Two realizations were found to be analogous: that the nature of the substituents on the aryl thioglycosides could be used to modulate the nucleophilicity of the anomeric sulfur atom and that protecting groups that could be used as ‘‘armed’’ or ‘‘disarmed’’ glycosyl donors. Thus an electron-donating group (EDG: e.g., Me, OMe, NHAc, H) placed in the para position of the aryl moiety of an aryl thioglycoside provided ‘‘active’’ or ‘‘armed’’ glycosyl donors, while electron-withdrawing groups (EWG: e.g., NO2, Br, Cl) afforded ‘‘latent’’ or ‘‘disarmed’’ thioglycosyl donors that could be used as building blocks in blockwise oligosaccharide syntheses. ‘‘Latent’’ thioglycosyl derivatives, possessing one free hydroxyl group and corresponding to ‘‘temporary inactive’’ species, serves as glycosyl acceptors. Their reactivities can be ‘‘turned on’’ by transforming their electron-withdrawing thioaryl substituents (e.g., NO2) into electron-donating groups (e.g., NHAc), thus ‘‘reactivating’’ the sulfur atoms toward electrophilic promoters usually employed in glycosylation reactions. Alternatively, the ‘‘active’’ thioglycosyl donors should already possess EDG substituents on their aryl moieties. The ‘‘latent’’ thioglycosyl acceptors can be glycosylated with ‘‘active’’ thioglycosyl donors by using thiophilic promoters. The disaccharide can then be transformed into an ‘‘active’’ glycosyl donor by modification of the aryl substituents. Taking advantage of both ‘‘active–latent’’ and ‘‘armed–disarmed’’ concepts, it was possible to modulate the reactivities of thioglycosyl donors and acceptors by changing both the nature of the substituents in the para position of the aglycone’s thiophenyl group and the protecting groups at the C2 position [34,43–46]. This has expanded the reactivity differences between thioglycosyl donors and acceptors and thus provided improved flexibility for the proper choice of glycosyl donors and acceptors toward building complex sialylated oligosaccharides. Initially, the method was based on the differential reactivity conferred on each of the partners by the nature of the protecting groups and by the intrinsic nucleophilicity of the thioaryl leaving groups. However, as the method evolved, it became clear that the choice of the promoter was also an important factor to take into consideration. Indeed, it was demonstrated that a ‘‘disarmed’’ thioglycoside could be activated in the presence of powerful thiophilic promoters such as N-iodosuccinimide/trifluoromethanesulfonic acid (NIS/TfOH), whereas it remained inactivated in the presence of weak thiophilic reagent such as iodonium dicollidine perchlorate or methyl triflate [43,45]. An example of this strategy is illustrated in Scheme 3, which describes the synthesis of a positional isomer of sialyl LewisX. For instance, armed glycosyl donor phenyl 2-thio-␣-sialoside 11 was chemoselectively activated with NIS/TfOH in the

282

Roy

Scheme 3

presence of latent 4-nitrophenyl thiolactoside 12 to provide sialyllactose derivative 13 in 47% yield. Then armed–active perbenzylated ethyl thiofucoside 14 was used to further regioselectively glycosylate trisaccharide 13 at O2⬘ of the galactose residue to give sialylated tetrasaccharide 15 in 63% yield (46). 4-Nitrophenyl sialyl-1-thiolactoside 13 still possesses an aglycone that can be further transformed into an active glycosyl donor (NO2 → NH2 → NHAc) or into a reactive acrylamido monomer (NH — COCH — —CH2) for copolymerization purpose [48]. C.

Syntheses of Suitably Functionalized Sialosides

It is clear that multivalent sialosides may offer numerous opportunities for ‘‘medicinal glycobiology.’’ They can be used to increase receptor binding interactions in areas such as flu virus inhibition of hemagglutination, anti-inflammatory agents (selectin antagonists), cancer vaccines and immunodiagnostics, and in treating gastrointestinal infections. Moreover, inasmuch as sialic acid receptor themselves might be organized as clusters, it appeared sound to synthesize multivalent glycoforms varying in molecular weights, shapes, valencies, and geometries to ‘‘scan’’ wide ranges of

Multivalent Glycotools and Sialic Acid

283

topographical areas. To this end, suitably functionalized N-acetylneuraminic acid derivatives were required. A convergent approach in which various sialoside haptens and other carbohydrates of interest could be attached to multivalent scaffolds at a late stage was chosen. This strategy offers the advantage of permitting further optimization of binding interactions where appropriate. General approaches in which N-acetylneuraminic acid derivatives could be used as either electrophiles, nucleophiles, or comonomers were considered. Using derivatives already obtained through PTC (Scheme 2), it was possible to provide the necessary building blocks. Thus, thioacetate 6 was readily de-S-acetylated into potent nucleophile 7 (NaOMe, MeOH, ⫺40⬚C, 30 min, 88%), which had been treated with 4-nitrobenzyl bromide (16), a wide variety of N-chloroacetamides 18, or galactosyl derivative 20 (Z. Gan, R. Roy, unpublished results) to afford key precursors 17 (74%), 19, or 21 (70%), respectively (Scheme 4). N-Chloroacetamides 18 represent typical examples of multivalent scaffolds onto which were built various N-acetylneuraminic acid dendrimers (see below). Multiple chloride substitutions were found to be very efficient and high yielding, together with providing a key 1H NMR signal (␦ 4.2 ppm) that was used to evaluate coupling efficiency. 4-Nitrophenyl O/S-sialosides 9 and 10 were reduced to amine derivatives 22 (HCO2NH4, 10% Pd-C, MeOH, reflux) or 23 (SnCl2, EtOH, reflux) which were directly transformed into comonomers 24 or 25 with acryloyl chloride (CH2 — — CHCOCl, CH2Cl2, Et3N, 86–88%) [49] (Scheme 5). Additionally, 4-aminophenyl 2thiosialoside 23 was converted into 4-isothiocyanatophenyl derivative 26 (CSCl2, DIPEA, CH2Cl2, 87%) for direct conversion into thiourea-linked PAMAM dendrimers [50]. Because it was also of interest to synthesize glycopeptidomimetics and glycopeptoids (see below) by either solution or solid phase synthesis, sialosyl azide 8 was reduced into amine 27 (10% Pd-C, MeOH, quant) and treated with sebacoyl

Scheme 4

284

Roy

Scheme 5

chloride (28) or with Cbz-protected 6-aminocaproyl chloride 29 to give acid 30 [51] or amine 31 [52] after standard deprotection. Acid 30 was used in solid phase synthesis of hyperbranched dendrimers [51], while blockwise approach was used to transform amine 31 into oligopeptoids up to an octamer level [52]. With efficient (active–latent) and stereospecific (PTC) methodologies in hand for key building block syntheses, the stage was ready for multivalent sialoside preparations. III.

GLYCOFORMS, OLIGOPEPTOIDS, AND AMPHIPHILIC NANOSTRUCTURES

As stated, the search for carbohydrate ligands of high affinity and specificity is of prime interest in glycobiology. While classical structure–activity relationships (SARs) have led to the development of only a limited number of effective inhibitors through exhaustive enterprises, the rational design of glycoclusters or glycomimetics spanning cooperative pharmacophores has furthered our fundamental understanding of carbohydrate–protein interactions. Moreover, multivalent glycoforms may assist unraveling cooperative binding interactions as they really appear in in vivo experiments, a situation not always prevailing with isolated protein receptors or their recombinant forms. It is also stimulating to imagine the implications of glycoclusters in signal transduction. Neoglycoproteins, liposomes, and glycopolymers have been successfully used to demonstrate that multivalency does indeed amplify carbohydrate–protein binding interactions by factors as high as thousands. However, by their very nature, these neoglycoconjugates have ill-defined chemical structures. They are heterogeneous in size and carbohydrate contents. Additionally, neoglycoproteins have been shown to

Multivalent Glycotools and Sialic Acid

285

be immunogenic [53], and the same may hold for high molecular weight glycopolymers, which can mimic repeating units found on bacterial capsular polysaccharides. Thus, while such glycopolymers can demonstrate the role played by multivalency in recognition processes, they fail to allow precise biophysical analyses of the cluster effect. The following paragraphs illustrate our approach toward the rational design of sialic acid containing clusters. Sialoside clusters, like their other neoglycoconjugate analogs (liposomes, polymers, proteins), have been mainly used as inhibitors of flu virus hemagglutination [54] and as selectin antagonists [55]. Natural tetravalent [56] and semisynthetic divalent [57] sialyloligosaccharides have been prepared by means of chemoenzymatic sialylation of preisolated multiantennary glycans obtained by degradation of glycoproteins or simply by chemical sialylation of tethered glycosides. In recent examples, tetrameric and dimeric sialyl LewisX clusters showed inhibitory IC50 values of less than 50 nM and 0.15 mM in L-selectin-mediated lymphocyte endothelium interactions, respectively. The enzymatically prepared tetramer was a 60-fold better inhibitor than its corresponding monomer. The description of sialylated clusters was reviewed in 1997 [58] and is covered in detail here. Several pieces of evidence suggest that cooperative binding interactions from small clusters depend on both the overall number of sialic acid ligands and their relative positioning with respect to one another. For instance, we have shown that carbohydrate dimers can be nicely oriented and interspaced with very short spacers to provide efficient ‘‘cross-linkers’’ without the need to reach two binding sites from the same lectins [59]. Interestingly, analogous trimers were shown to be less effective on a per-saccharide basis in similar interactions when measured by turbidimetric and inhibition experiments. It was concluded that the ‘‘third’’ unbound ligand of a trimer acted counter-productively in the overall binding associations. These observations seem to hold when densely packed glycodendrimers are used as inhibitors (see below). As deeper appreciation of multivalent carbohydrate-protein interactions was gained, it became obvious that clusters of various shapes, size, and orientation would provide powerful ligands from which fundamental informations about receptors topography would be obtained. To address these issues, the systematic development of novel multivalent glycoclusters was deemed essential. A.

Neoglycopeptidomimetics

Solid phase glycopeptide synthesis is now routine operation in laboratories and has been used to prepare multivalent sialyl LewisX by chemoenzymatic combination [60]. Such a strategy allows investigators to obtain oligomers of varied valency (up to 8) and interspaced by determined numbers of amino acid residues. While this strategy is conceptually appealing, it still lacks the advantages conferred to peptidomimetics if therapeutically valuable products are foreseen. The design of peptide isosteres would be highly preferable, following common practice in drug developments. The recent development of peptoids derived from simple N-substituted oligoglycines seems to fulfill the required criteria of metabolic stability and synthetic simplicity [61]. Our laboratory has been involved in the design of related N- [62] and O-linked [63] glycopeptoids. Once model glycopeptoid syntheses had been established, we decided to prepare sialic acid oligopeptoids having controlled distances both between the backbone

286

Roy

and between the repeating units [52]. To this end, a convergent blockwise approach with orthogonally protected derivatives was chosen. The following example illustrates the synthesis of linear sialic acid oligomers interspaced by two N-substituted glycine residues onto which was attached a 6-aminocaproic acid spacer (Scheme 6). ␣-Sialosyl azide 8 was transformed into Cbz-protected amine derivative, which upon hydrogenolysis afforded amine 31. A similar strategy has been used by Sabesan [64] to prepare saccharopeptides. This approach allows the preparation of the required ␣ anomer because this configuration is absolutely critical for the sialic acid’s biological activity. Amine 31 was then transformed into key building block 32 following initial treatment with tert-butylbromoacetate (75%) and N-bromoacetylation with bromoacetic anhydride (92%). This orthogonally protected intermediate 32 can be elongated from either direction depending on the strategy chosen. N-Alkylation of 32 by amine 31 afforded dimer 33 which has been N-acetylated to end group 34 (AcCl, 91%) or to middle group 35 after Cbz protection (89%). Sequential deprotection of tert-butyl ester or Cbz group followed by amide coupling (DCC) of the resulting amino acid derivatives afforded tri-, tetra-, hexa-, and octamers such as 36, after protecting group removal under standard conditions. The resulting ‘‘sialopeptoids’’ were thus made available for biological evaluation. Unfortunately, their inhibitory properties against influenza virus hemagglutination were rather negligible compared with dendrimers or polymers (see below). B.

Amphiphilic Calix[4]arene Nanostructures

Guided by the need to prepare well-defined clusters having good binding properties in solid phase enzyme immunosorbent assays (ELISA) while having readily exposed carbohydrate ligands, we became interested in synthesizing glycosylated calix [n]arene derivatives. In many ways, glycocalix[n]arenes, which possess guest–host capabilities that can be used as drug vectors, are structurally related to cyclodextrins. Additionally, they have advantages unsurpassed by the cyclodextrins because they can be readily modified at either ‘‘upper’’ or ‘‘lower’’ rim [65]. Commercially available p-tert-butylcalix[4]arene (37) is already equipped with a hydrophobic tail that has been useful in coating polystyrene microtiter plates [66]. Thus, by freezing 37 in the cone conformation upon treatment with ethyl bromoacetate, ester hydrolysis, and acid chloride formation, known [67] tetraacid chloride 38 was readily made available [68] (Scheme 7). Treatment of 38 with a slight excess of mono-Boc-protected 1,4-butanediamine gave 39 in 62%. Trifluoroacetolysis of the Boc protecting groups resulted in tetramine 40 quantitatively. N-Chloroacetylation ((ClCH2CO)2O, Et3N, CH2Cl2) gave 41 in 63% yield which, when treated with peracetylated thiosialoside 7 (Et3N, CH3CN, N2, 16 h, room temperature) and protecting group hydrolysis (NaOMe, then NaOH) gave 42 in 65% yield. In spite of having high hydrophobic content, tetrameric thiosialoside 42 was fairly water soluble (4.8 mM, 13 mg/mL). It showed strong binding affinity to the plant lectin wheat germ agglutinin (WGA) in a microtiter plate assay. Moreover, it formed insoluble cross-linked lattices with WGA as demonstrated by turbidimetric experiments. The insoluble complex could be inhibited by monomeric phenylthio ␣-sialoside 11 (free OH), thus demonstrating the specificity of the binding interaction. Following several observations leading to the easy formation of well-organized cross-linked lattices with simple clusters [59,68], including dimers [69], we became

Multivalent Glycotools and Sialic Acid

Scheme 6

287

288

Scheme 7

Roy

Multivalent Glycotools and Sialic Acid

289

intrigued by the possibility of generating ‘‘sugar rods.’’ Several of these molecules could be synthesized via olefin self-metathesis reactions catalyzed by Grubb’s catalyst [(PCy3)2Cl2Ru — —CHPh] [70]. Additionally, they can be prepared from ␣-alkynyl sialosides (Z. Gan, R. Roy, unpublished data) by using palladium(0)-catalyzed crosscoupling chemistry (Sonogashira reaction) (71). These novel derivatives fall outside the scope of this chapter, so their synthesis will be described in due course.

IV.

GLYCOPOLYMERS

We introduced the term ‘‘glycopolymer’’ to identify water-soluble polymers onto which carbohydrate haptens are covalently appended [72]. In this respect, glycopolymers should differ from pseudopolysaccharides, which refer to chemically modified polysaccharides and to insoluble materials used in affinity chromatography. Curiously, we initially made the first sialylated glycopolymers [73] to screen anti–sialic acid antibodies obtained from a sialylated neoglycoproteins [18b,73–75]. The original aim was to generate an antigen deprived of cross-reactive hapten (Neu5Ac), which inevitably would have been produced from other protein carriers. It was then quickly realized that these copolymers offered great potential as inhibitors in cell adhesion processes. Patents for a cancer diagnostic kit consisting of a sialylated protein (vaccine) and an ELISA screening antigen (polymer) were filed several years ago [76]. While this activity was ongoing, we reported the inhibitory potential of these novel sialylated copolymers in influenza flu virus inhibition of hemagglutination. These early observations were then similarly made by several other groups [77,78]. Earlier reviews [10,11,58,79] had described numerous polymerization methods that have been used in synthesizing a wide range of glycopolymers. Amazingly, even though most methodologies can afford better organized copolymers than the one initially used (i.e., by random acrylamide copolymerization or modification), very few alternative methods have been exploited for sialosides. The syntheses and applications of glycopolymers are now covered in several reviews and book chapters [79–82], and only recent developments are highlighted. Reducing sugars can be reductively aminated with ammonia or other amines. The resulting amine derivatives can then be transformed into N-acrylamide monomers useful in copolymerization strategy. Unfortunately, acrylamide copolymerization affords polymers that may vary greatly in their batch-to-batch molecular weight distributions. An improved protocol consists of synthesizing polyacrylates (43) having active ester functionality NHS [83] or 4-NO2Ph [84] (Scheme 8). After aminolysis or hydrolysis, the molecular weight of the resulting polyacrylamides or polyacrylic acids can be compared against commercially available polymer standards used in HPLC. This allows the synthesis of reproducible lots of a given copolymer. Once established, the starting polyacrylates 43 can be treated with various amounts of any amine-containing sugars. Quenching the residual reactive esters with amine 1 (R1NH2) followed by amine 2 (R2NH2), or simply with ammonia or water, afforded copolymers having desired biophysical properties [77,83]. The second and third amines may include probes (e.g., biotin, fluorescamine), lipid groups, other sugars, and peptides. The strategy has also been elegantly used to generate sialopolymer libraries [85]. Sialic acid copolyacrylamides such as 44 were obtained with O/S-aryl

290

Roy

Scheme 8

spacers [32,49], C-glycosides [86], and other related spacers [78]. GM3-type copolymer 45 was obtained using the foregoing procedure [48]. Thioaryl sialoside comonomer 46 [49] is easier than C-glycoside [25] to produce in a stereocontrolled manner. It is also resistant to sialidases that are simultaneously present on flu virions. Interestingly, it could be directly incorporated onto both poly-L-lysine and proteins by 1,4-conjugate additions (Michael addition) at pH 10 to provide biocompatible random copolymer 47 (Scheme 9) [87]. By analogy, reductively aminated ␣-(2,8)-polysialic acid (colominic acid) 48 can be N-acryloylated into compounds 49, which undergo 1,4-conjugate addition onto poly-L-lysine to provide copolymer 50, isolated as its biotin conjugate [87]. In 1998 Wong et al. [88] reported an analogous strategy whereby a lysoganglioside derivative was amidated to poly-L-glutamic acid (DP 540) together with the fluorescent tag 4,4-difluo5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionyl group (BODIPY) to give 51. The copolymer showed picomolar inhibition of H1N1 influenza hemagglutinin that corresponded to an improved binding of 1000-fold over gangliosides GM3 or lyso-GM3 and 105-fold relative to the monosaccharide sialyl lactose. Lipidlike copolymer 51 is thus as active as a polymerized liposome published in 1993 [89]. All the strategies discussed so far included sialylated copolymers having randomly distributed sialic acid residues. As these approaches failed to generate regular copolymers onto which the hapten distributions would be interspaced at constant distances, it was deemed necessary to prepare copolymers by addition polymerization [90]. Chapter 8 by Mann and Kiessling offers an alternative strategy for the use of ring-opening metathesis polymerization (ROMP) of norbornene derivatives to generate analogous polymers. The novel strategy described herein depends on a sialic acid monomer (53) having two amine groups that can be copolymerized by a reiterative addition process onto a bisisocyanate (Scheme 10) (R. Roy, Y. Makimura, unpublished data). Thus, 4-nitrophenyl sialoside 9a was reduced and treated with

Multivalent Glycotools and Sialic Acid

291

Scheme 9

bromoacetyl chloride. After N-alkylation with Cbz-protected 3,3⬘-iminobis(propylamine) [91], intermediate 52 was obtained in 75% yield. Hydrogenolysis provided diamine 53 quantitatively. Finally, addition polymerization of 53 with either 1,4-butanediamine or 1,6-hexanediamine bisisocyanates and ester hydrolysis gave low molecular weight copolymers 54 and 55 in ⬃50% yield, with a degree of polymerization of ⬃15. It is worth mentioning that all known glycopolymers have shown strong inhibitory properties when used in conjunction with carbohydrate binding proteins. The exact origins of the increased associative forces conferred on glycopolymers are not clearly understood. Aside from the individual binding site’s affinity constants (KD), which obviously remained unchanged, entropic penalties that occur upon binding are minimized with multivalent ligands. The ligands’ local high concentrations are certainly also affecting association/dissociation’s kinetics (kon /koff ). Some external fac-

292

Roy

Scheme 10

tors such as steric stabilization have also been invoked to support the observations [92]. It is clear, however, that a closer examination of the active site’s topography is a critical factor to be elucidated for the sake of better designing potent ligands/ inhibitors. With this criterion in mind, increasing activity is being addressed in attempts to design neoglycoconjugates of intermediate size between small clusters (di-/trimers) and glycopolymers. Glycodendrimers seem to fulfill the foregoing requirement. Each neoglycoconjugate has its own strength and weaknesses and ultimately, it is the targeted application that will dictate which glycoforms will be favorable. Section V provides a brief overview of the progress and limitations encountered during the development of glycodendrimers since their first synthesis in 1993 [93]. V.

GLYCODENDRIMERS

Several reviews cover dendrimer syntheses and applications [94], but very few describe glycodendrimers as such [16,52,82,95,96]. Dendrimers can basically adopt two shapes: spherical, globular-like structures (58) and monodendritic (59–61 ) architectures (Scheme 11). The last family is particularly appealing because it can mimic complex multiantennary glycans found at the tips of natural glycoproteins. Moreover, from cumulative observations, spherical dendrimers, particularly large ones (i.e., ⱖ16–32-mer) have started to show their intrinsic structural limitations; that is, they suffer from severe steric accessibility. This situation is further amplified by the presence of complete bacterial and viral particles wherein the receptors are themselves clustered and congested. Alternatively, they have shown excellent inhibitory properties with soluble or surface-bound lectins and antibodies.

Multivalent Glycotools and Sialic Acid

Convergent (59 ⫹ 60) and divergent (56,57) strategies toward glycodendrimer 58.

293

Scheme 11

294

Roy

Glycodendrimers can be synthesized by both convergent and divergent strategies. Ideally, they can be simply prepared by conjugation of active carbohydrate derivatives onto preformed dendrimers (Scheme 11). Given the commercial availability of poly(amidoamine) (PAMAM, 56) and poly(propyleneimine) (Dab, 57) dendrimers, these amine-ending dendrimers are the most heavily exploited. Even though dendrimer surfaces can be constructed to exhibit all possible functionalities, amine-terminating groups are synthetically more appealing and have been used most extensively. The potentially useful thiolated dendrimers self-oxidize, while carboxylated dendrimers tend to form intramolecular anhydrides once activated. This last situation may cause defects upon carbohydrate attachment. Although alcohols seem also attractive, a priori, their direct use in glycosylation chemistry is hampered by potentially difficult complete anomeric stereocontrol. Amine-functionalized dendrimers have been used in several instances. Reducing sugars (62) can be directly anchored to PAMAM dendrimers by reductive amination [97], and sugar lactones (63), readily prepared from reducing sugars by oxidation with basic iodine solutions, can be amidated [98]. Aryl [50] or glycosyl [99] isothiocyanato derivatives (64) also react rapidly and efficiently with polyaminated dendrimers, even under aqueous conditions [100]. Incorporation of chloro- or bromoacetamido groups onto PAMAM dendrimers [XCH2COCl or (ClCH2CO)2O] afforded highly electrophilic species that react readily with thio sugars (65) (Scheme 12). The last approach has been successfully applied in double N-alkylation when bromoacylated carbohydrate derivatives (66) were used. It simultaneously allowed increasing surface group density [66,101]. These versatile strategies gave to glycodendrimers such as 67–71 high-yielding accesses that obviously are applicable to higher oligosaccharides. Sialylated dendrimers having modest activity in inhibition of hemagglutination of flu viruses have been prepared by using Fmoc-chemistry and L-lysine core in solid phase synthesis reactions [30,93]. Similarly, the foregoing dendrimers, together with analogous peptidomimetic-like dendrimers (72) built on 3,3⬘-iminobispropylamine cores [102,103] showed up to ⬃32-fold inhibition of binding of human ␣1-acid glycoprotein (orosomucoid) to the plant lectin wheat germ agglutinin or the slug lectin from Limax flavus (Scheme 13). Similar dendrimers having aryl ␣-D-mannopyranosides as surface group (L-lysine core) showed ⬃2000-fold increased inhibition of binding of yeast mannan to concanavalin A or pea lectin [104]. Nanomolar IC50s values have been reported for sialodendrimers 73 obtained by condensation of PAMAM 56 with p-isothiocyanatophenyl sialoside 26 [50] (Scheme 14). Additionally, dendritic 3⬘-sulfo-LewisX (Glc) bound to poly-L-lysine backbones inhibited the binding of E-selectin to sialyl–LewisX glycolipids 600 times better than the corresponding monomer [105]. The growth valency of the foregoing dendrimers was based on a 2n progression, where n is the generation (i.e., 2-, 4-, 8-, 16-mer, etc.). It has been deemed of interest to construct dendrimers with a 3n progression (i.e., 3-, 9-, 27-mer, etc.). To this end, we constructed sialodendrimers (74) based on a gallic acid core [106] (Scheme 15). The inhibition of certain interactions by large variations in glycodendrimer efficacy points to the need to design multivalent neoglycoconjugates with defined geometry, valency, and shapes. The seemingly modest results observed with some plant lectins further exemplify the problem encountered in glycobiology as opposed to other drug–protein interactions. A good model to clarify these observations has

Multivalent Glycotools and Sialic Acid

Scheme 12

295

Divergent construction of glycodendrimers built on polyamine scaffolds using reductive amination (67), amide formation (68), thiourea bridging with isothiocyanates (64), thioethers (70), and novel double N-alkylation strategy (71).

296

Scheme 13

Roy

16-Mer dendritic sialoside constructed by solid-phase synthesis using 3,3⬘iminobis(propylamine) scaffold [91].

Multivalent Glycotools and Sialic Acid 297

Scheme 14 16-Mer sialodendrimer synthesized using divergent anchoring of isothiocyanotophenyl sialoside onto PAMAM backbone.

298

Roy

Scheme 15 Dendritic 9-mer sialoside (3n growth) built on gallic acid core using hydrophilic oligoethyleneglycol spacer.

been recognized with cell adhesion molecules (selectins), where it was established that mediocre in vitro results with sialyllactose do not necessarily translate to poor in vivo experiments [107]. As seen from the synthesis of glycodendrimers derived from the oligosaccharide portion of ganglioside GM1 (Gal␤(1–3)GalNAc␤(1– 4)[Neu5Ac␣(2–3)]Gal␤(1–4)Glc␤) built on PAMAM and poly(propyleneimine) dendrimers [108], the inhibitory properties against cholera toxin B subunit and E. coli heat-labile enterotoxin showed IC50s values 15-fold lower than that of GM1 itself and 1000-fold lower than that of the oligosaccharide. Just as glycopolymers can be further glycosylated by classical enzymatic reactions via the necessary sugar nucleotides and the corresponding enzymes [109], glycodendrimers can be treated analogously. In this way, it has been straightforward to produce on a dendritic poly-L-lysine core N-acetylglucosamine-ending dendrimers sequentially elongated with UDP-galactose, CMP-Neu5Ac, and finally with GDP-

Multivalent Glycotools and Sialic Acid

299

fucose and the appropriate glycosyltransferases [110]. Notably, all eight branches of the preformed glycodendrimer could be fully glycosylated to provide octameric sialyl LewisX tetrasaccharides in excellent yields.

VI.

HYBRID DENDRIMER–POLYMERS

The take-home lesson to date is that several small glycoforms exhibit strong binding affinity. The best examples are derived from the pioneering work of Lee et al. [111], who showed that N-acetylgalactosaminide trimers may be few thousand times better than the monosaccharide in binding to hepatic asialoglycoprotein receptors (ASGRs) [112]. There is also a growing body of evidence suggesting that glycoclusters, when properly designed, showed increased affinity. Given the observations and interpretation by Whitesides et al. [92] that steric stabilization (i.e., wrapping of virion particles by sialopolymers) may be responsible for some of the observed increased affinity, it appeared tempting to combine the best of the accumulated observations to build dendritic-like glycopolymers. Such glycoforms already exist and were described in 1995 [113,114]. A fascinating example has been proposed by Fan et al. [113], who chemoenzymatically synthesized a dendritic mannoside (9-mer) ending with an N-acetylglucosaminide acrylamide derivative. The dendrimer was built by the transglycosylaiton of an asparagine, Man9GlcNAc2Asn glycopeptide, obtained by pronase digestion and transferred onto the GlcNAc monomer by means of the endo␤-N-acetylglucosaminidase isolated from Arthrobacter protophormiae. The resulting multiantennary copolymer was a better ligand than soybean agglutinin toward a recombinant rat mannose binding protein (MBP). An analogous strategy has been used by Furuike et al. [114] in the synthesis of an acrylamide-ending N-acetyllactosamine trimer that had been copolymerized with acrylamide. In a recent attempt to generate nontoxic polysialosides that would be powerful influenza flu virus hemagglutinin inhibitors as well as inhibiting infectivity, we synthesized the novel nanostructures shown in Scheme 16 [115]. To this end, poly(ethyleneimine) was used as scaffolding for the attachment of 4-isothiocyanatophenyl sialoside (deprotected 26) to give ‘‘classical’’ randomly substituted copolymers such as 75 (Scheme 16). The poly(ethyleneimine) (PEI) backbone was prepared by leaving-ring-opening cationic polymerization of 2-ethyl-2-oxazoline, with methyl tosylate as an initiator and morpholine as a terminator, followed by acid hydrolysis [116]. Alternatively, the secondary amines of the poly(ethyleneimine) were used as initiator for the second-generation cationic polymerization [117], which after sialoside conjugation gave comb–branch (G0) structures 76. For comparison purposes, PAMAM dendrimers having partial amine substitutions and interspacing hydroxyl groups [115] were prepared according to our published protocol (Scheme 14) to provide structures such as 77. Rod-shaped, cylindrical dendrimers synthesized by means of the PAMAM reiterative strategy onto PEI backbone [116] were also used to prepare dendritic copolymers like 78. Finally, use of hyperbranched PEI/PEI backbones gave dendrigraft structures 79 (PEI G1–G3). Preliminary results showed comb–branched (76) and dendrigraft (79) to be the most effective sialopolymers made to date (50,000-fold better than the monomer). They also blocked infection of mammalian cells in vitro. The results also varied significantly with various strains of influenza A (H2N2, X-31) and Sendai viruses.

300

Roy

Scheme 16

Schematic representation of novel hyperbranched glycostructures illustrating: regularly interspaced ligands onto poly(ethyleneimine) backbone (75), comb-branch (G0) structure (76), interspaced spheroidal dendrimer (77), rod-shaped, cylindrical polymer with dendritic branch (78), and dendrigraft (79).

VII.

CONCLUSIONS

The construction of neoglycoconjugates has reached an unprecedented level of sophistication and imagination. This is undoubtedly the result of increased interest from the traditional glycobiology community as well as by a wide range of synthetic chemists now entering the field, a trend facilitated by access to modern analytical tools. The degree of refinement in the techniques employed to quantitate polyvalent interactions has also steadily increased over the last few years (reviewed in Ref. 92). The need for multivalent sialosides is just emerging as their implications in novel biological interactions are being continuously unraveled. Our ignorance about the in vivo receptors’ topography and valency requirements is amazingly challenging. It is our opinion that the synthetic demand for glycoclusters, glycopolymers, and hybrid molecules thereof will significantly increase within the next few years. The

Multivalent Glycotools and Sialic Acid

301

medicinal chemistry dogma for small-molecule therapeutics will also need to be revisited and adapted to glycopharmaceuticals. ACKNOWLEDGMENTS Financial support from the Natural Sciences and Engineering Research Council of Canada (NSERC) is gratefully acknowledged. I am particularly indebted to my past and present graduate students and postdoctoral fellows who had courageously tackled the chemistry of polysialosides: C. A. Laferrie`re, D. Zanini, S. J. Meunier, S. Cao, W. K. C. Park, M. Letellier, Q. Wu, S.-N. Wang, Z. Gan, M. Llinares, and U. K. Saha, and F. Hernandez Mate´o (visiting professor from Granada). Generous donation of sialic acid by Dr. M. Kawase from NGK Insulators, Ltd (Handa, Japan) is also acknowledged. REFERENCES 1.

2. 3. 4. 5. 6. 7. 8.

9.

10. 11. 12. 13.

14. 15. 16. 17. 18. 19. 20.

(a) R Schauer. Sialic Acids—Chemistry, Metabolism & Functions. Cell Biology Monographs, vol 10. Vienna: Springer-Verlag, 1992. (b) M von Itzstein, MJ Kiefel. In: ZJ Witczak, KA Nieforth, eds. Carbohydrates in Drug Design. New York: Marcel Dekker, 1997, pp 39–82. A Varki. Glycobiology 3:97–130, 1993. KA Karlsson. Curr Opin Struct Biol 5:622–635, 1995. H Lis, N Sharon. Chem Rev 98:637–674, 1998. J Beuth, HL Ko, G Pulverer, G Uhlenbruck, H Pichlmaier. Glycoconjugate J 12:1–6, 1995. SD Rosen, CR Bertozzi. Curr Opin Cell Biol 6:663–673, 1994. EJ Toone, Curr Opin Struc Biol 4:719–728, 1994. M von Itztein, WY Wu, GB Kok, MS Pegg, JC Dyason, B Jin, TV Phan, ML Smythe, HF White, SW Oliver, PM Colman, JN Varghese, DM Ryan, JM Wood, RC Bethell, VJ Hotham, JM Cameron, CR Penn. Nature 363:418–423, 1993. CU Kim, W Lew, MA Williams, H Liu, L Zhang, S Swaminathan, N Bischofberger, MS Chen, DB Mendel, CY Tai, WG Laver, RC Stevens. J Am Chem Soc 119:681– 690, 1997. YC Lee, RT Lee. Acc Chem Res 28:321–327, 1995. R Roy. Trends Glycosci Glycotechnol 8:79–99, 1996. (a) R Duncan, J Kopecek. Adv Polym Sci 57:51–101, 1984. (b) M Monsigny, AC Roche, P Midoux, R Mayer. Adv Drug Delivery Rev 14:1–24, 1994. JJ Garcı´a-Lo´pez, F Herna´ndez-Mate´o, J Isac-Gracı´a, JM Kim, R Roy, F SantoyoGonza´lez, A Vargas-Berenguel. J Org Chem 64:522–531, 1999, and references cited therein. A Marra, A Dondoni, F Sansone. J Org Chem 61:5111–5158, 1996. M Dubber, TK Lindhorst. Carbohydr Res 310:35–41, 1998. R Roy. Polym News 21:226–232, 1996. D Zanini, R Roy. In: Y Chapleur, ed. Carbohydrate Mimics: Concepts and Methods. Weinheim: Verlag Chemie, 1998, pp 385–415. (a) MF Czarniecki, ER Thornton. J Am Chem Soc 99:8273–8279, 1977. (b) R Roy, CA Laferrie`re. Can J Chem 68:2045–2054, 1990. M Koketsu, LR Juneja, H Kawanami, M Kim, T Yamamoto. Glycoconjugate J 9:70– 74. RW Veh, JC Michalski, AP Corfield, M Sander-Wewer, D Gies, R Schauer. J Chromatogr 212:313–322, 1981.

302

Roy

21. 22. 23. 24.

HJ Jennings, AK Bhattacharjee. Carbohydr Res 55:105–112, 1977. HJ Jennings, R Roy, F Michon. J Immunol 134:2651–2657, 1985. R Roy, RA Pon. Glycoconjugate J 7:3–12, 1990. H Higashi, M Naiki, S Matuo, K Okouchi. Biochem Biophys Res Commun 79:388– 395, 1977. (a) H Paulsen, P Matschulat. Liebigs Ann Chem 487–495, 1991. (b) JO Nagy, MD Bednarski. Tetrahedron Lett 32:3953–3956, 1991. IR Vlahov, PI Vlahova, RJ Linhardt. J Am Chem Soc 119:1480–1481, 1997. RU Lemieux. Acc Chem Res 29:373–380, 1996. R Roy, In: Y Sasson, R Neumann, eds. Handbook of Phase Transfer Catalysis. New York: Blackie Academic & Professional, 1997, pp 244–275. R Roy, FD Tropper, S Cao, JM Kim. ACS Symp Ser 659:163–180, 1997. R Roy, D Zanini, SJ Meunier, A Romonowska. ACS Symp Ser 560:104–119, 1994. FD Tropper, FO Andersson, S Braun, R Roy. Synthesis 618–620, 1992. R Roy, FO Andersson, G Harms, S Kelm, R Schauer. Angew Chem Int Ed Engl 31: 1478–1481, 1992. R Roy, FD Tropper, A Romanowska, M Letellier, L Cousineau, SJ Meunier, J Boratynski. Glycoconjugate J 8:75–81, 1991. S Cao, SJ Meunier, FO Andersson, M Letellier, R Roy. Tetrahedron: Asymm 5:2303– 2312, 1994. WKC Park, SJ Meunier, D Zanini, R Roy. Carbohydr Lett 1:179–184, 1995. (a) K Okamoto, T Goto. Tetrahedron 46:5835–5857, 1990. (b) MP DeNinno. Synthesis 583–593, 1991. (a) Y Ito, T Ogawa. Tetrehedron 46:89–102, 1990. (b) T Erce´govic, G Magnusson. J Org Chem 60:3378–3384, 1995. TJ Martin, RR Schmidt. Tetrahedron Lett 33:6123–6126, 1992. A Marra, P Sinay¨. Carbohydr Res 195:303–308, 1990. A Hasegawa, M Kiso. In: H Ogura, A Hasegawa, T Suami, eds. Carbohydrates. Synthetic Methods and Applications in Medicinal Chemistry. Weinheim: VCH, 1992, pp 243–266. R Madsen, B Fraser-Reid. In: SH Khan, RA O’Neil, eds. Modern Methods in Carbohydrate Synthesis. Amsterdam: Harwood Academic Publishers, 1996, pp 155–170. GJ Boons. Tetrahedron 52:1095–1121, 1996. S Cao. ‘‘Active–Latent’’ Glycosylation Strategy in Oligosaccharide and Glycoconjugate Syntheses. PhD dissertation, University of Ottawa, Ottawa, ON, Canada, 1996. R Roy, FO Andersson, M Letellier. Tetrahedron Lett 33:6053–6056, 1992. S Cao, F Herna´ndez-Mate´o, R Roy. J Carbohydr Chem 17:609–631, 1998. S Cao, Z Gan, R Roy. Carbohydr Res 318:75–81, 1999. S Raghavan, D Kahne. J Am Chem Soc 115:1580–1581, 1993. S Cao, R Roy. Tetrahedron Lett 37:3421–3424, 1996. CA Laferrie`re, R Roy, FO Andersson. Methods Enzymol 242:271–280, 1994. D Zanini, R Roy. J Org Chem 63:3486–3491, 1998. M Llinares, R Roy. J Chem Soc Chem Commun 2119–2120, 1997. (a) R Roy. Topics Curr Chem 187:241–274, 1997. (b) UK Saha, JM Kim, R Roy. Syntheses of glycoforms of biological interest. Proceedings of the Eighth European Carbohydrate Symposium, Spain, 1995, C IL-5. R Roy, CA Laferrie`re, RA Pon, A Gamian. Methods Enzymol 247:351–361, 1994. (a) GD Glick, JR Knowles. J Am Chem Soc 113:4701–4703, 1991. (b) S Sabesan, JØ Duus, P Domaille, S Kelm, JC Paulson. J Am Chem Soc 113:5865–5866, 1991. (c) S Sabesan, JØ Duus, S Neira, P Domaille, S Kelm, JC Paulson, K Bock. J Am Chem Soc 114:8363–8375, 1992. (d) SA DeFrees, W Kosch, W Way, JC Paulson, S Sabesan, RL Halcomb, DH Huang, Y Ichikawa, CH Wong. J Am Chem Soc 117:66–79, 1995.

25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40.

41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52.

53. 54.

Multivalent Glycotools and Sialic Acid 55. 56. 57. 58. 59. 60. 61.

62.

63.

64. 65. 66. 67.

68. 69. 70.

71.

72. 73. 74. 75. 76. 77. 78.

79. 80.

303

EE Simaek, GJ McGarvey, JA Jablonowski, C-H Wong. Chem Rev 98:833–862, 1998. A Seppo, JP Turunen, L Penttila¨, A Keane, O Renkonen, R Renkonen. Glycobiology 6:65–71, 1996. H Maaheimo, R Renkonen, JP Turunen, L Penttila¨, O Renkonen. Eur J Biochem 234: 616–625, 1995. R Roy. In ZJ Witczak, KA Nieforth, eds. Carbohydrates in Drug Design. New York: Marcel Dekker, 1997, pp 84–136. (a) R Roy, D Page´, S Figueroa Perez, V Verez Bencomo. Glycoconjugate J 15:251– 265, 1998. (b) See also: CF Brewer. Chemtracts-Biochem Mol Biol 6:165–179, 1996. C Unverzagt, S Kelm, JC Paulson. Carbohydr Res 251:285–301, 1994. RN Zuckermann, EJ Martin, DC Spellmeyer, GB Stauber, KR Shoemaker, JM Kerr, GM Figliozzi, DA Goff, MA Siani, RJ Simon, SC Banville, EG Brown, L Wang, LS Richter, WH Moos. J Med Chem 37:2678–2685, 1994. (a) UK Saha, R Roy. Tetrahedron Lett 38:7697–7700, 1997. (b) UK Saha, R Roy. J Chem Soc Chem Commun 201–202, 1996. (c) UK Saha, R Roy. Tetrahedron Lett 36: 3635–3638, 1995. (d) UK Saha, R Roy. J Chem Soc Chem Commun 2571–2573, 1995. (a) JM Kim, R Roy. Carbohydr Res 298:173–179, 1997. (b) JM Kim, R Roy. Tetrahedron Lett 38:3487–3490, 1997. (c) JM Kim, R Roy. Carbohydr Lett 1:465–468, 1996. S Sabesan. Tetrahedron Lett 38:3127–3130, 1997. (a) V Bo¨hmer. Angew Chem Int Ed Engl 34:713–745, 1995. (b) A Ikeda, S Shinkai. Chem Rev 97:1713–1734, 1997. R Roy, JM Kim. Angew Chem Int Ed Engl 38:369–372, 1999. F Arnaud-Neu, G Barrett, S Cremin, M Deasy, G Fergusson, SJ Harris, AJ Lough, G Lourdes, MA McKervey, MJ Schwing-Weill, P Schwinte. J Chem Soc Perkin Trans 2: 1119–1125, 1992. SJ Meunier, R Roy. Tetrahedron Lett 37:5469–5472, 1996. D Page´, R Roy. Biorg Med Chem Lett 6:1765–1770. (a) R Dominique, SK Das, R Roy. J Chem Soc Chem Commun 2437–2438, 1998. (b) SK Das, R Dominique, C Smith, J Nahra, R Roy. Carbohydr Lett 3:361–368, 1999. (c) YJ Hu, R Roy. Tetrahedron Lett 40:3305–3308, 1999. (a) K Sonogashira, Y Tohda, N Hagihara. Tetrahedron Lett 4467–4470, 1975. (b) R Roy, SK Das, R Dominique, MC Trono, F Herna´ndez-Mate´o, F Santoyo-Gonza´lez. Pure Appl Chem, 565–571, 1999. R Roy, FD Tropper, A Romanowska. Bioconjugate Chem 3:256–261, 1992. R Roy, CA Laferrie`re, A Gamian, HJ Jennings. J Carbohydr Chem 6:161–165, 1987. R Roy, CA Laferrie`re. Carbohydr Res 177:C1–C4, 1988. CA Laferrie`re. Synthesis of Sialic Acid Antigens. PhD dissertation, University of Ottawa, Ottawa, Ontario, Canada, 1990. R Roy, CA Laferrie`re, U.S. Patents 5,034, 516, 1991; 5,192,661, 1993. (a) A Gamian, M Chomik, CA Laferrie`re, R Roy. Can J Microbiol 37:233–237, 1991. (b) A Spaltenstein, GM Whitesides. J Am Chem Soc 113:686–687, 1991. (a) NE Nifant’ev, AS Shashkov, YE Tsvetkov, AB Tuzikov, IV Abramenko, DF Gluzman, NV Bovin. ACS Symp Ser 560:267–275, 1994. (b) MA Sparks, KW Williams, GM Whitesides. J Med Chem 36:778–783, 1993. (c) JO Nagy, P Wang, JH Gilbert, ME Schaefer, TG Hill, MR Callstrom, MD Bednarski. J Med Chem 35:4501–4502, 1992. R Roy. In GJ Boons, ed. Carbohydrate Chemistry. London: Chapman & Hall, 1998, pp 243–321. G Magnusson, AY Chernyak, J Kihlberg, LO Kononov. In YC Lee, RT Lee, eds. Neoglycoconjugates: Preparation and Applications. San Diego, CA: Academic Press, 1994, pp 53–143.

304

Roy

81.

(a) NV Bovin, HJ Gabius. Chem Soc Rev 24:413–421, 1995. (b) NV Bovin. Glycoconjugate J 15:431–446, 1998. R Roy. In SH Khan, R O’Neil, eds. Modern Methods in Carbohydrate Synthesis. Amsterdam: Harwood Academic Publishers, 1996, pp 378–402. (a) GB Sigal, M Mammen, G Dahmann, GM Whitesides. J Am Chem Soc 118:3789– 3800, 1996. (b) SK Choi, M Mammen, GM Whitesides. Chem Biol 3:97–104, 1996. NE Byramova, LV Mochalova, JM Belyanchikov, MN Matrosovich, NV Bovin. J Carbohydr Chem 10:691–700, 1991. SK Choi, M Mammen, GM Whitesides. J Am Chem Soc 119:4103–4111, 1997. JE Kingery-Wood, KW Williams, GB Sigal, GM Whitesides. J Am Chem Soc 114: 7303–7305, 1992. (a) R Roy, CA Laferrie`re. J Chem Soc Chem Commun 1709–1711, 1990. (b) R Roy, RA Pon, FD Tropper, FO Andersson. J Chem Soc Chem Commun 264–265, 1993. (c) A Romanowska, SJ Meunier, FD Tropper, CA Laferrie`re, R Roy. Methods Enzymol 242:90–101, 1994. H Kamitakahara, T Suzuki, N Nishigori, Y Suzuki, O Kanie, CH Wong. Angew Chem Int Ed Engl 37:1524–1528, 1998. W Spevak, JO Nagy, DH Charych, ME Schaefer, JH Gilbert, MD Bednarski. J Am Chem Soc 115:1146–1147, 1993. D Page´. Effect of Shape, Size, and Valency of Multivalent Mannopyranosides on Their Binding Properties to Phytohemaggutinins. MSc dissertation, University of Ottawa, Ottawa, Ontario, Canada, 1997. SI Murahashi, T Naota, N Nakajima. Chem Lett 879–882, 1987. M Mammen, SK Choi, GM Whitesides. Angew Chem Int Ed Engl 37:2754–2794, 1998. R Roy, D Zanini, SJ Meunier, A Romanowska A. J Chem Soc Chem Commun 1869– 1872, 1993. (a) HF Chow, TKK Mong, MF Nongrum, CW Wan. Tetrahedron 54:8543–8660, 1998. (b) DK Smith, F Diederich. Chem Eur J 4:1353–1361, 1998. (c) D Seebach, PB Rheimer, G Greiveldinger, T Butz, H Sellner. Top Curr Chem 197:125–164, 1998. (d) D Astruc. CR Acad Sci Paris 322 (Ser IIb):757–766, 1996. (e) N Ardoin, D Astruc. Bull Soc Chim Fr 132:875–909, 1995. (f ) A Archut, F Vo¨gtle. Chem Soc Rev 27:233–240, 1998. (g) F Zeng, SC Zimmerman. Chem Rev 97:1681–1712, 1997. (h) A Tomalia, HD Durst. Top Curr Chem 165:193–313, 1993. (i) GR Newkome, C Moorefield, F Vo¨gtle. Dendritic Molecules: Concepts, Syntheses, Perspectives, 2nd ed. Weinheim, Wiley-VCH, 1998. ( j) JMJ Fre´chet, CJ Hawker. Compr Polym Sci 140–201, 1996. (a) R Roy. Curr Opin Struct Biol 6:692–702, 1996. (b) TK Lindhorst. Nachr Chem Tech Lab 44:1073–1079, 1996. (a) T Toyokuni, AK Singhal. Chem Soc Rev 231–242, 1995. (b) N Jayaraman, SA Nepogodiev, JF Stoddart. Chem Eur J 3:1193–1199, 1997. R Roy, J Thompson, H Sashiwa, SK Das, S Tripathy, HJ Gabius. Pre´paration et proprie´te´s de ne´oglycoconjugue´s implique´s lors du rejet chronique de xe´notransplantation. Proceedings of 67th ACFAS Congress, Ottawa, 1999. K Aoi, K Itoh, M Okada. Macromolecules 28:5391–5393, 1995. (a) D Page´, S Aravind, R Roy. J Chem Soc Chem Commun 1913–1914, 1996. (b) TK Lindhorst, C Kieburg. Angew Chem Int Ed Engl 35:1953–1956, 1996. (c) D Page´, R Roy. Glycoconjugate J 14:345–356, 1997. C Kieburg, TK Lindhorst. Tetrahedron Lett 38:3885–3888, 1997. R Roy, JM Kim. Polym Mater Sci Eng 77:195–196, 1997. D Zanini, R Roy. J Am Chem Soc 119:2088–2095, 1997. D Zanini, R Roy. J Org Chem 61:7348–7354, 1996. D Page´, D Zanini, R Roy. Bioorg Med Chem 4:1949–1961, 1996.

82. 83. 84. 85. 86. 87.

88. 89. 90.

91. 92. 93. 94.

95. 96. 97.

98. 99.

100. 101. 102. 103. 104.

Multivalent Glycotools and Sialic Acid 105.

106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117.

305

(a) R Roy, WKC Park, D Zanini, C Foxall, OP Srivastava. Carbohydr Lett 2:259–266, 1997. (b) R Roy, WKC Park, OP Srivastava, C Foxall. Bioorg Med Chem Lett 6:1399– 1402, 1996. (a) SJ Meunier, Q Wu, SN Wang, R Roy. Can J Chem 75:1472–1482, 1997. (b) R Roy, WKC Park, Q Wu, SN Wang. Tetrahedron Lett 36:4377–4380, 1995. MS Mulligan, JC Paulson, S DeFrees, ZL Zheng, JB Lowe, PA Ward. Nature 364: 149–151, 1993. JP Thompson, CL Schengrund. Glycoconjugate J 14:837–745, 1997. SI Nishimura, K Yamada. J Am Chem Soc 119:10555–10556, 1997. MM Palcic, H Li, D Zanini, RS Bhella, R Roy. Carbohydr Res 305:433–442, 1998. See also: D Zanini, R Roy. Bioconjugate Chem 8:187–192, 1997. RT Lee, YC Lee. In: YC Lee, RT Lee, eds. Neoglycoconjugates: Preparation and Applications. San Diego, CA: Academic Press, 1994, pp 23–50. YC Lee. In: G Bock, S Harnett, eds. Carbohydrate Recognition in Cellular Function. New York: John Wiley & Sons, 1989, pp 80–95. JQ Fan, MS Quensenbery, K Takegawa, S Iwahara, A Kondo, I Kato, YC Lee. J Biol Chem 270:17730–17735, 1995. T Furuike, N Nishi, S Tokura, SI Nishimura. Chem Lett 823–824, 1995. JD Reuter, A Myc, MM Hayes, Z Gan, R Roy, D Q, R Yin, LT Piehler, R Esfand, DA Tomalia, JR Baker Jr. Bioconjugate Chem 10:271–278, 1999. R Yin, Y Zhu, DA Tomalia. J Am Chem Soc 120:2678–2679, 1998. DA Tomalia, DM Hedstrand, MS Ferrito. Macromolecules 24:1435–1438, 1991.

10 Aminoglycoside Antibiotics: Structures and Mechanisms of Action Jalal Haddad, Lakshmi P. Kotra, and Shahriar Mobashery Wayne State University, Detroit, Michigan

I.

INTRODUCTION

The vast majority of compounds bearing amino sugars as the main component of their structures belong to the family of aminoglycoside antibiotics. Another term used is aminocyclitol antibiotics, because the most important members of this class with high antimicrobial activity contain an aminocyclitol moiety in their structures. This chapter briefly describes some important noncyclitol aminoglycosides, but our main focus is on aminocyclitol antibiotics, which are more important in terms of biological activity. Aminoglycoside antibiotics are antimicrobial agents potently active against both gram-positive and gram-negative bacteria. They show rapid concentration-dependent cidal action [1,2], and their activity is independent of the size of the bacterial inoculum [3]. Furthermore, their pharmacokinetics is predictable and they synergize with ␤-lactam antibiotics. It is interesting to note that even though a number of resistance enzymes to these have been identified, such resistance cases have not disseminated widely in the clinic, as has been done with respect to ␤-lactam antibiotics [4,5]. Aminoglycosides have relatively poor oral absorption; therefore they are generally administered intravenously [6]. All aminoglycosides possess the potential for renal and otovestibular toxicity, but despite these problems they remained useful for the treatment of serious infections and continue to be the antibiotic of choice in clinics because of their many desirable traits. Streptomycin [7], the first aminoglycoside to find clinical use, was second only to penicillin as an antibiotic in so doing. It was the very first antibiotic with activity against Mycobacterium tuberculosis, the microorganism that causes tuberculosis. In307

308

Haddad et al.

deed, it was used in treatment of this disease just one year after its discovery [8]. Streptomycin remains effective as a key component of modern antimycobacterial therapy [9]. This initial finding led to the identification and characterization of several other clinically useful aminoglycosides over the subsequent 50 years. Aminoglycoside antibiotics are used for the treatment of infections. They exert their antibacterial properties by binding to the particular sites of RNA molecules in the bacterial ribosome [10–12]. As discussed later, this binding interferes with the fidelity of mRNA translation, results in miscoding, and leads ultimately to bacterial cell death [13–16]. Compounds in this class can interact with a variety of other biologically relevant RNA sequences. For instance, it is known that they inhibit growth of the human immunodeficiency virus (HIV-I) by blocking the binding of the regulatory protein Rev to its response element [17–19]. During the past five decades, numerous aminoglycosides of microbial origin have been isolated, their structures elucidated, and the compounds synthesized. Extension of these synthetic strategies in combination with the understanding of the mechanisms of bacterial resistance and the knowledge of binding ability of aminoglycoside antibiotics to various RNA molecules have opened new avenues for preparation of a large number of semisynthetic aminoglycoside antibiotics with remarkable potency against antibacterial-resistant organisms. Reviews by Hanessian et al. [20], Umezawa [21], and Cox et al. [22] deal with the chemistry of aminoglycosides; Price [23] has reviewed structure–activity relationships, and Umezawa et al. [24,25] have discussed the structure, synthesis, and biosynthesis of aminoglycoside antibiotics. This chapter summarizes what is known of the structural classes of aminoglycosides to date and presents state-of-the-art knowledge of the mechanisms for manifestation of the biological activities of these versatile antimicrobials. II.

AMINOGLYCOSIDE STRUCTURES

Naturally occurring aminoglycosides are amino sugars consisting of an aglycon unit, which can be cyclic or open chain. The aglycon is either streptamine, its related aminocyclitols, cyclitol, monoaminocyclitol, or a noncyclitol moiety. Therefore, these natural products can be divided into three major classes: (1) aminoglycosides containing streptamine and its related aminocyclitols, (2) aminoglycosides containing cyclitol and/or monoaminocyclitol, and (3) aminoglycosides having a noncyclitol moiety in their structures. A.

Aminocyclitol Aminoglycosides Containing Streptamine and Its Derivatives

The aglycon unit in aminocyclitol aminoglycosides that contain streptamine and its derivatives is 2-deoxystreptamine (compounds 1–3 ) 1 (2-DOS), streptamine 2, or streptidine 3. Most of the aminocyclitol aminoglycosides with high antibacterial activity have the 2-deoxystreptamine as the aglycon unit. The carbon atom bearing the amino group with the R configuration is assigned number 1 in the structure. The most active aminoglycosides are those with their sugar units substituted at positions 4 and 5, or 4 and 6, of the aglycon unit. By convention, the sugar unit attached to the 4-position

Aminoglycoside Antibiotics

309

Compounds 1–3

of 2-DOS is designated by the primed (⬘) numbers and those attached to the 5- or 6-positions are indicated by the double-primed (⬙) numbers. 1.

Aminoglycosides Containing 2-Deoxystreptamine

(a) 4-Substituted 2-Deoxystreptamine. Among 4-substituted-2-deoxystreptamine aminoglycosides, three pseudotrisaccharides, apramycin, oxyapramycin (14) [26], and saccharocin (15) (compounds 13–15 ) [27,28], show more potency than neamine (4) and are active against many resistant bacterial strains. Members of this group of pseudodisaccharides (compounds 4–12 ) are minor metabolites of Actinomycetes and show weak antibacterial activity [29–35]. (b) 5-Substituted 2-Deoxystreptamine. These aminoglycoside antibiotics have an unusual spiro ring junction in their structures. They show weak antibacterial activity

Compounds 4–12

310

Haddad et al.

Compounds 13–15

and are relatively more toxic than neamine. Of these antibiotics, hygromycin B (16), produced by Streptomyces hygroscopicus [36], and destomycin A (17), produced by Streptomyces rimofaciens [37], are used as anthelmintics for poultry and swine. The antibiotic SS-56C is structurally related to A-396-I, having streptamine (2) instead of 2-deoxystreptamine (1) in its structure [38]. Another member of this group, produced by Saccharopolyspora hirsuta, has been characterized a 1-N-amidino-1-Ndemethyl-2-hydroxydestomycin A and is relatively weaker than destomycin A (17) [39]. In general, antibacterial activity of this class of aminoglycoside antibiotics against Staphylococcus aureus and Escherichia coli is lower than that of neamine, but these drugs are relatively active against Pseudomonas aeruginosa [38]. Compounds 19 and 20 are discussed in the literature [40,41] (compounds 16–20 ). (c) 6-Substituted 2-Deoxystreptamines. These pseudodisaccharides show weak antibacterial activity. Compound NK-1012-2 (21) is a fermentation product of Streptomyces kanamyceticus [32], and garamine (22) is one of the minor components of gentamicin-producing organisms [35]. (d) 4,5-Disubstituted 2-Deoxystreptamines. The most active members of this group are neomycins B and C (compounds 23 and 24, respectively), which are pseudotetrasaccharides produced by Streptomyces fradiae [42]. These antibacterial agents are clinically important and are used as a mixture for treatment of infectious diseases caused by both gram-positive and gram-negative bacteria. Paromomycins I and II (25 and 26, respectively) are clinically useful drugs, and as a mixture they are less toxic than neomycins. Neomycin B and paromomycin I (having five and six charged amino groups, respectively) show strong interactions with 16S ribosomal RNA in vitro and in vivo; they are good candidates for structure modification and synthesis of new drug [43]. Two pseudopentasaccharides, lividomycin A and mannosylparomomycin, and 4⵮-O-␣-D-mannopyranosyl-substituted derivatives of lividomycin B (27) and paromomycin I (25), respectively. They are produced by the same organism that produces

Aminoglycoside Antibiotics

311

Compounds 16–20

lividomycin B and paromomycin [34]. 3-N-Methylparomomycin I has been isolated from the culture filtrate of Streptoverticillium sp. A1-R2827. The antibacterial activity of this compound is a little weaker than that of paromomycin I [44]. 6⵮-Deamino6⵮-hydroxyneomycin and 6⵮-deamino-6⵮-hydroxyparomomycin have been obtained from strains of Streptomyces (S. fradiae and S. rimosus, respectively). They are mixtures of C5⵮ epimers, suggesting a status as intermediates in the biosynthesis of the parent antibiotics [45] (compounds 23–27 ; see pg. 312).

Compounds 21 and 22

312

Haddad et al.

Compounds 23–27

The ribostamycin aminoglycosides are pseudotrisaccharides compounds 28–36 that lack the ring IV (neosamine) of neomycins in their structure [45–52]. They show antibacterial activity greater than neamine, but less than neomycins. (e) 4,6-Disubstituted 2-Deoxystreptamine. These pseudotrisaccharides (compounds 37–61 ) are effective against a variety of gram-positive and gram-negative bacteria but are highly susceptible to the drug resistance enzymes (see below). Kanamycins B and C are minor components of the kanamycin A producing organism Streptomyces kanamyceticus [53]. In comparison to kanamycins A and B, kanamycin C (39) has a much weaker antibacterial activity against S. aureus and E. coli. Tobramycin 42 or 3⬘-deoxykanamycin B is a clinically useful antibiotic because of its ability to avoid phosphorylation at the 3⬘-position by the resistance enzymes aminoglycoside 3⬘-phosphotransferases. Another important class of antibiotics in this group is the gentamicins (compounds 45–60 ), of which about 25 different compounds have been isolated from the same or different culture filtrates [35,57,58,59,60,61]. Commercially available gentamycin C, which is a valuable therapeutic agent, is a mixture of gentamicins C1, C1a, and C2. Gentamicins A3, B, and B1 bear a hydroxyl group instead of an amine at the C2⬘-position. Sisomicin 61 and related compounds are derivatives of gentamicin C, which have an unsaturated amino sugar moiety in their structures. Sisomicin is more active than the gentamicin C mixture, but it is also twofold more toxic (compounds 61–65 ; see pg. 317).

Aminoglycoside Antibiotics

313

Compounds 28–36

The aminoglycoside antibiotic 66-40G produced by Micromonospora inyoensis is 3⬙-N-demethylsisomicin [70], and the G-367S (compound 66; see pg. 317) produced by a strain of Dactylosporangium, has been characterized as 2⬘-N-formylsisomicin [71]. Seldomycins (compounds 67–69 ; see pg. 318) are another set of structurally related gentamicin aminoglycosides in this group. Seldomycin 5 is the most potent, having a methoxy group at the 4⬙-position of xylopyranose ring [72]. Seldomycin 2 is a pseudodisaccharide lacking the xylopyranose moiety. Seldomycin 5 is the only naturally occurring aminoglycoside antibiotic that has an amino group at the 2⬙position next to the 3⬙-amino group. 2.

Aminoglycosides Containing Streptamine, Streptidine, and Their Derivatives

(a) Streptamine-Containing Aminoglycosides. One of the aminoglycosides bearing a streptamine ring in its structure is SS-56C (Section II.A.1.b) [38], which is struc-

314

Haddad et al.

Compounds 37–44

turally related to the destomycin family. Spectinomycins (compounds 70–72 ; see pg. 318) are another class in this series that contained actinamine (N,N⬘-dimethyl-2-epistreptamine) in their structures. Acmimycin (72), a spectinomycin analog having a cis–cis ring junction, is believed to have the following structure (see pg. 318) [44]. (b) Streptidine-Containing Aminoglycosides. Streptomycins (compounds 73–78 ; see pg. 319), an important class of aminoglycoside antibiotics, are pseudotrisaccharides or pseudotetrasaccharides that contain a 4-substituted streptidine moiety. Streptomycin, a potent chemotherapeutic agent, was the first aminoglycoside antibiotic, discovered in 1944 in the culture filtrate of Streptomyces griseus [7], as described earlier. Glebomycin [82] is an aminoglycoside antibiotic, that is structurally related to dihydrostreptomycin, having a carbamoyloxy group (NH2COO) instead of the guanidino group at the 1-position of the streptidine moiety. Antibiotic AC4437 is the deN-methyl-L-glucosamine analog of dihydrostreptomycin that shows weaker antibacterial activity than dihydrostreptomycin [83]. Ashimycins A and B are two streptomycin analogs (modified in the L-glucosamine unit) isolated from the fermentation broth of S. griseus [84]. Ashimycin A (compound 79; see pg. 319) carries a glycosidically linked sugar moiety having an

Aminoglycoside Antibiotics

315

Compounds 45–50

acid functional group in its structure, but the stereochemistry has not been established yet. 3.

Aminoglycosides Containing Cyclitol, Monoaminocyclitol, Diaminocyclitol, and Related Derivatives

(a) Cyclitol- and Monoaminocyclitol-Containing Aminoglycosides. Several aminoglycoside antibiotics containing monoaminocyclitol or cyclitol moiety are known. Among these, kasugamycins [85], myomycins [86,87], validamycins [88], and inosamycins [89] are of clinical interest (compounds 80–84 ; see pg. 320). The revised structure of validamycin A 85 (compounds 85–92 ; see pg. 320) was reported by Suami, Ogawa, and Chida [90] in 1980 and it turned to be 4-O-␤-Dglucopyranosyl validamycin A, instead of the 3-O-glucosyl derivative previously reported by Kameda in 1972. Inosamycins (compounds 93–96 ; see pg. 321) are monoaminocyclitol-containing aminoglycosides produced by S. hygroscopicus; they are structurally related to the neomycin, paromomycin, and ribostamycin families but contain a 2-deoxy-scylloinosamine instead of 2-deoxystreptamine 1 as the aglycon unit [89]. Inosamycin E is a ribostamycin analog in which the amino group at position 1 is replaced by a hydroxyl function [89]. Inosamycin A 93 exhibits antibacterial activity comparable to that of neomycin, with lower toxicity than neomycin; however, inosamycin E is less active than ribostamycin. Boholmycin (compound 97; see pg. 321) is a pseudotetrasaccharide from S. hygroscopicus H617-25. Its structure consists of a heptose, two amino sugars, and dicarbamoyl-scyllo-inositol. It shows

316

Compounds 51–57

Compounds 58–60

Haddad et al.

Aminoglycoside Antibiotics

317

Compounds 61–65

weak antibacterial activity but exhibits a broad spectrum of activity against grampositive and gram-negative bacteria, including aminoglycoside-resistant strains [95]. Antibiotic X-14847 (compound 98; see pg. 321), produced by Micromonospora echinospora, has been identified at 1-O-(2-amino-2-deoxy-␣-D-glucopyranosyl)-Dmyo-inositol showing weak antibiotic activity [96]. (b) 1,4-Diaminocyclitol-Containing Aminoglycosides. This class includes fortimicins and related analogs. They are active antibiotics against most gram-positive and gram-negative bacteria, including those with resistant strains, but they are not active against Pseudomonas. In this class, the glycyl moiety plays an important role in activity of the compound, and those without this component (fortimicin B, sporaricin

Compound 66

318

Compounds 67–69

Compounds 70 and 71

Compound 72

Haddad et al.

Aminoglycoside Antibiotics

319

Compounds 73–78

B, istamycins A0 and B0) show less potency (compounds 99–111 ; see pgs. 322 and 323). Sannamycin C (compound 112; see pg. 323) is a fortimicin analog, and its 4N-glycyl derivative inhibits gram-positive and gram-negative bacteria resistant to other aminoglycoside antibiotics [108]. In sannamycin C the methoxy group at position 3 is equatorial, whereas in sannamycin B the methoxy group holds an axial position.

Compound 79

320

Compounds 80–84

Compounds 85–92

Haddad et al.

Aminoglycoside Antibiotics

Compounds 93–96

Compound 97

Compound 98

321

322

Haddad et al.

Compounds 99–109

In this series, (compounds 113 and 114; see pg. 323) lysinomicin (113) [109] and 2⬘-de-N-L-lysyllysinomocin [110] showed good antibacterial activity against several strains. Fortimicin KG3 (114) and its related analogs are another branch of this family that have an unsaturated amino sugar moiety similar to that of sisomicins in their structure. Among these, fortimicin KG3 has potent antibacterial activity [111]. Another fortimicin analog is sporaricin E or 2-deoxy fortimicin B, which shows weak antibacterial activity [112]. In a report by Morioka et al. it was demonstrated that a blocked mutant of istamycin producing Streptomyces tenjimariensis converts fortimicin B to 1-epi-dactimicin with inversion of the amino group at position 1 and N-4-acylation [113].

Aminoglycoside Antibiotics

323

Compounds 110 and 111

Compound 112

Compounds 113 and 114

B.

Noncyclitol Aminoglycosides

The noncyclitol aminoglycosides include several monosaccharides, disaccharides, and those with the open-chain aglycon units. Compounds 115–117 (see pg. 324), nojirimycin (115) [114], streptozotocin (116) [115], prumycin (117) [116], and a few others are monosaccharides with only weak antibacterial activity. Trehalosamines are representative of the disaccharides in this class of aminoglycosides and are relatively weak antibiotics. Trehalosamine (compound 118) [117], its 2-epimer mannosylglucosaminide [118], and the 4-amino derivative compound 119 [119] were isolated from culture filtrate of Streptomyces and showed weak an-

324

Haddad et al.

Compounds 115–117

Compounds 118–120

tibacterial activity. The mannosylglucosaminide is a trehalosamine with an axial hydoxyl group at the 2-position, and the BMY-28251 compound 120 is the first example of an ␣,␤-trehalose antibiotic obtained from Bacillus pumilis [120]. The sorbistins (compounds 121–124 ) are monosaccharides having an openchain aglycon unit (1,4-diaminosorbitol) in their structures. Although they show weak antibacterial activity, they exert their antibiotic activity against many of the aminoglycoside-resistant strains. Sorbistin A1 is the most active compound in the class, showing activity 1/50 of that of kanamycin A against E. coli, but exhibiting some activity against resistant strains [121,122,123]. Trehazolin (compound 125) is a pseudodisaccharide that serves as a trehalase inhibitor from Micromonospora sp. strain SANK 62390 [124]. Salbastatin compound

Compounds 121–124

Aminoglycoside Antibiotics

325

Compounds 125 and 126

126 is another trehalase inhibitor [125], isolated from Streptomyces albus. These natural products are inhibitors of glycosidases, which play critical roles in intra- and intercellular processes including cell adhesion, membrane transportation, and signal transduction. III.

ENZYMATIC MODIFICATION OF AMINOGLYCOSIDES

Resistance to antibiotics manifests itself in different ways. The target of the antibiotic may be altered such that affinity for the drug may be reduced [126,127,128]. Or, the rate of the uptake of the drug into the bacterium may be attenuated such that the critical concentration necessary for saturating the binding site cannot be reached [129,130,131]. Alternatively, specific enzymes may evolve to alter the structure of the drug in bacterium, whereby the site for the binding shows lower affinity for the modified drug, thus manifesting resistance to the original drug. This third method for acquired resistance is the most common for aminoglycoside antibiotics [132,133,134,135]. Three classes of enzymes have evolved for aminoglycoside modification for the manifestation of the resistance phenotype. These enzymes are aminoglycoside acetyltransferases (Aac), aminoglycoside nucleotidyltransferases (Ant), and aminoglycoside phosphotransferases (Aph). The acetyltransferases transfer the acetyl group from acetyl coenzyme A to specific amines in aminoglycosides. The sites of these modifications are 1-, 3-, 2⬘-, and 6⬘-amino groups. The other two classes of enzymes transfer the adenyl or the phosphoryl group from ATP to hydroxyl groups in aminoglycosides. The common sites of adenyl transfer are hydroxyl groups at 2⬙- and 4⬘-hydroxyls of kanamycins and 3⬙- and 6-hydroxyls of streptomycin. The sites of phosphotransferase reactions are the hydroxyl groups at positions 3⬘ and 2⬙ of kanamycins and 3⬙ and 6 of streptomycin. The products of these enzymatic modifications are invariably poorer antibiotics because their binding to their ribosomal target sites has been impaired. The nature of these enzymes and state of the art in their mechanistic understanding were reviewed comprehensively in 1998 [134]. IV.

MECHANISM OF ACTION OF AMINOGLYCOSIDE ANTIBIOTICS

Aminoglycosides exhibit their antibiotic action primarily by entering the cell irreversibly by active transport, binding to the ribosomal RNA (rRNA), interfering with protein synthesis (translation process), and causing membrane damage resulting in leakage of the cell membrane [134,136,137]. These actions result in disruption of

326

Haddad et al.

the integrity of the bacterial cell, ultimately leading to cell death. There are exceptions, however, to this mode of action, and a few of the aminoglycosides such as spectinomycin and kasugamycin are bacteriostatic; that is, they stop the growth of bacteria rather than killing the cell, although they too interact with rRNA. It is now well established that aminoglycosides bind to the RNA major groove at various important sites such as the A site (ribosomal acceptor site) and hammerhead regions (see below). Aminoglycosides enter bacteria in a biphasic and energy-dependent process. The initial phase (phase I) allows for transport of small quantities of aminoglycoside, although the amount that is required for antibacterial action is not known [134,138]. Phase I has been described as an energy-dependent active transport and entails a slow, linear uptake process [138,139,140]. Phase II of uptake is both energy and concentration dependent, and it shows a behavior similar to the diffusion of small molecules through channels. After uptake during phase I, aminoglycoside binds to the chain-elongating ribosomes and causes misreading of mRNA, resulting in malformed proteins. A few of these miscoded proteins become part of the cell surface and disrupt the integrity of the membrane. The damaged membrane subsequently allows more antibiotic molecules to enter the cell, where ultimately they reach a concentration that can inhibit the initiation of the translation process. Eventually, the combination of disrupted membrane and irreversible blocking of the protein biosynthesis kills the organism [136]. Ribosome plays an important role in selecting the correct EF-Tu-GTP–aminoacyl–tRNA ternary complex at the A site, which supplies the next amino acid to the elongating polypeptide during translation [141]. In the A-site region of ribosomal RNA, the highly conserved A1492 and A1493 point toward the minor groove in the A site [142]. It was proposed that when a complementary tRNA binds to the codon, the 2⬘-OH moieties on mRNA may interact with the N1-position of the conserved adenines, and perhaps this process is a beginning step in the communication needed for the codon–anticodon complex. It was suggested that binding of aminoglycosides to the rRNA decreases the dissociation rates of aminoacyl–tRNA, and stabilizes a high-affinity conformation of tRNA–mRNA complex [142]. This in turn may result in inhibition of the translation process and/or favor miscoding. In fact, in early 1960s, ribosomes were identified as the target sites of action for aminoglycosides such as streptomycin. However, the limitations of scientific knowledge at the time prevented the establishment of the relationship between the ribosome and the bactericidal activity [136]. A few years later, it was understood that binding to the ribosome of small molecules such as aminoglycosides can influence not only the catalytic activity of the ribosome but also its specificity; that is, such binding can lead to improper information processing in translation, transcription, and replication. The interactions of aminoglycosides with ribosome are concentration dependent [140]. At lower concentrations, streptomycin blocks the abundant polysomal ribosomes (chain elongating), whereas at higher concentrations, these molecules can bind all ribosomes, including the ones involved in initiation [136]. It was proposed that binding of streptomycin to the ‘‘initiating ribosome’’ is irreversible. However, this may be due to the irreversible uptake of aminoglycoside into the cell as well. The bactericidal activity of aminoglycosides is now supported by the involvement of almost every step of the mechanism given above [134]. However, there are still several unanswered questions relating to, for example, the nature of the miscoded

Aminoglycoside Antibiotics

327

proteins that are transported to the membrane, the enzymes/channels in the active transport of aminoglycosides into the cell, the nature of miscoding, and their effect on protein folding and membrane integration. Lately, there is increasing evidence of inhibition of self-splicing of group I introns by a number of aminoglycosides such as streptomycin, tobramycin, neomycin B, kanamycin, neamine, 5-epi-sisomicin and gentamicin [143,144,145,146]. It was shown that at reasonably low (micromolar) concentrations, several of these aminoglycosides can inhibit noncompetitively the thymidylate synthase group I intron RNA splicing in vitro. It appears that the ability of various aminoglycosides to inhibit intron self-splicing is purely structure dependent; and depending on the concentration of the aminoglycoside, one or both steps in the catalysis of group I intron selfsplicing are inhibited. It is very interesting, however, to learn that new activities are being discovered for these older antibiotics, and perhaps these findings will shed more light on the mechanism of action of aminoglycosides and their toxicity mechanisms. The similarities in the binding of aminoglycosides to ribosomal RNA and the catalytic self-splicing introns have led to the hypothesis that perhaps aminoglycosides were modulators of RNA activity in biologically evolving systems. Furthermore, during the coevolution of RNA and aminoglycosides, modern ribosomes have formed on which aminoglycosides exhibit similar binding properties [146,147]. V.

SITES OF RNA–AMINOGLYCOSIDE INTERACTIONS

Aminoglycoside antibiotics, as discussed in the preceding section, interact with RNA. Ribosome is the target for this class [10–12,148], and the antibiotics exert their antibacterial effects by binding to ribosomal RNA (rRNA) in the A site, causing misreading of the genetic code and inhibiting the process of translocation [149,150] (see above). Aminoglycoside antibiotics also interact with a variety of other biologically relevant RNA molecules. These include mRNA from human immunodeficiency virus (HIV-1), transactivation response element (TAR) [151], group I self-splicing introns [152], the hammerhead region of rRNA [153,154,155], and the hepatitis delta virus (HDV) ribozyme [156]. Next, we discuss recent developments in the binding of aminoglycoside antibiotics to RNA. A.

Binding of the Neomycin Class of Aminoglycosides to the A Site of 16S rRNA

Aminoglycoside antibiotics exert their antibacterial activity by interfering with protein biosynthesis. The common binding site for the structurally related aminoglycosides neamine (4), ribostamycin (28), paromomycin (25), and neomycin B (23) is localized in the A site of the prokaryotic 16S ribosomal RNA (rRNA) in the 30S subunit (Fig. 1a) [13–16,157]. These 2-deoxystreptamine-containing aminoglycoside antibiotics show a characteristic miscoding pattern that is distinct from that of streptomycin 73 [158], having a streptidine moiety in place of the 2-deoxystreptamine in its structure. They affect the codon–anticodon interaction in the A site of the 30S subunit by binding to the same site on rRNA, causing codon misreading and inhibiting translocation [149,150]. A specific interaction between the tRNA anticodons and mRNA codons is required for an accurate translation in the decoding region, which is formed by two short conserved sequences near the 3⬘ end of 16S rRNA (the nucleotide 1400–1500 region, Fig. 1b).

328

Haddad et al.

Figure 1

(a) Secondary structure of the 16S rRNA from E. coli. (b) The sequence of the decoding region [arrow in (a)]. (c) The sequence of a 27-nucleotide RNA used to characterize aminoglycoside–RNA interaction, with the highly conserved A-site portion of the 16S boxed. (d) Structures of aminoglycoside antibiotics.

To investigate the interactions between paromomycin and rRNA, Puglisi and colleagues [15] designed a small RNA fragment that mimics the structure of the A site of E. coli 16S rRNA. From these investigations, it appears that the universally conserved C1407 ⭈ G1494 base pair, A1408, and A1493 are required for the binding of paromomycin to the wild-type rRNA. An asymmetrical internal loop produced by a universally conserved adenine residue at position 1492 must also be present. In addition, base pairing is required in the lower stem for aminoglycoside binding.

Aminoglycoside Antibiotics

329

Positions 1406 and 1495 are universally conserved uridines that form a U ⭈ U pair in the ribosome. Paromomycin typically binds to the U1406 ⭈ U1495 base pair. It binds to two binding sites, the A1406 ⭈ U1495 and U1406 ⭈ G1495 base pairs, but binds only weakly to the U1406 ⭈ A1495 base pair variant. Quantitative footprinting of oligonucleotide variants identified critical nucleotides for binding of paromomycin to the A-site oligonucleotide. These include the C1407 ⭈ G1494 base pair, the A⭈ U base pair at positions 1410 and 1490, and nucleotides A1408, A1493, and U1495. Puglisi and coworkers further characterized the structure and binding of paromomycin to the A-site rRNA by using NMR spectroscopy to determine the solution structure of the RNA–paromomycin complex (Fig. 2) [142]. Some salient features of this structure are as follows. 1. Paromomycin binds in the major groove of the A-site rRNA, within the internal loop (see Fig. 1). Distortion of the RNA backbone by the presence of the bulged nucleotide A1492 and noncanonical A1408 ⭈ A1493 base pair leads to the formation of a distinct binding pocket for paromomycin. 2. Rings I and II of paromomycin adopt chair conformations, and their equatorial amino and hydroxy moieties make specific contacts with RNA, stabilizing the RNA–antibiotic complex.

Figure 2 The RNA–paromomycin NMR structure. Paromomycin is shown by the cappedstick presentation. A portion of the RNA structure at the aminoglycoside binding site is shown as a surface.

330

Haddad et al.

3. The two amino groups at positions 1 and 3 of the 2-deoxystreptamine moiety (ring II), found in all aminoglycosides that bind to the A site, make hydrogen bonds to U1495 and G1494, respectively, and are essential for specific binding of these antibiotics to the rRNA. 4. Ring I stacks above the base moiety of G1491 such that its hydroxyl groups at positions 3⬘ and 4⬘ are directed toward the phosphate moieties of A1493 and A1492, respectively, but they are not essential for antibiotic function. 5. Rings III and IV weakly contribute to specific binding, but the amino and hydroxyl groups of ring IV may have weak electrostatic interaction with the phosphate backbone of U1406, C1407, and U1490. 6. The A1408 ⭈ A1493 pair is essential for antibiotic binding in prokaryotic ribosomes and leads to formation of the specific binding pocket for ring I, whereas in the case of eukaryotic ribosomes having G1408 in place of A1408, a base pair of equivalent geometry cannot be formed, and paromomycin binds weakly to the G1408 ⭈ A1493 base pair. This RNA–paromomycin structure also explains a number of features that help show why modification of the aminoglycoside antibiotic by the resistance enzymes prevents the modified drug from binding to the ribosomal site. These features are as follows. 1. Disruption of the antibiotic binding pocket, formed by a base pair at positions 1409 and 1491, leads to aminoglycoside resistance in prokaryotes. 2. Enzymatic N1-methylation of A1408 is known to impair binding of the antibiotic to rRNA [128], a base that is a hydrogen bond acceptor in the A1408 ⭈ A1493 base pair. This modification gives resistance to kanamycins by disrupting the critical hydrogen bonding interactions between these two bases, which are required for aminoglycoside binding. 3. Enzymatic modification by acetylation of the amino group at position 3 of the 2-deoxystreptamine moiety would disrupt the essential hydrogen bonding between this amino group and G1494, as well as specific interactions with the A1493 phosphate. Any modification in ring I, which is tightly ensconced in the A1408 –A1493 pocket (e.g., such as the common 3⬘-phosphorylation or 6⬘-acetylation of aminoglycoside) prevents binding to rRNA by steric hindrance or undesired electrostatic interaction. Subsequently, the structure of the A-site region of rRNA in the absence of aminoglycoside was determined by Fourmy et al. by NMR spectroscopy [159]. Comparison of this structure with the paromomycin–rRNA complex indicated a local conformational change in the A-site RNA upon paromomycin binding. In the absence of paromomycin, the asymmetrical internal loop is closed by a Watson–Crick base pair (C1407 ⭈ G1494) and by two noncanonical base pairs (U1406 ⭈ U1495 and A1408 ⭈ A1493). The nucleic base A1492 stacks below A1493 and is intercalated between the upper and lower stems. Paromomycin binding stabilizes the conformation of A1492 and A1493, which are less well defined in the free RNA. It also changes the hydrogen bonding pattern of the A1408 ⭈ A1493 base pair such that hydrogen bonding increases from 1 in the free form to 2 in the bound form. Comparison of these free and bound conformations reveals that the two universally conserved residues, A1492 and A1493, are displaced toward the minor groove of the rRNA helix in the paromomycin–rRNA complex such that their N1-positions point into the minor groove of the A-site rRNA. These changes in the rRNA conformation, induced by the bound aminoglycoside,

Aminoglycoside Antibiotics

331

indicate a mechanism for the action of aminoglycoside antibiotics on translation, as elaborated shortly. Further studies on binding of the neomycin class of aminoglycosides to the A site of 16S rRNA revealed that neomycin B (23) binds to the rRNA with affinity similar to that of paromomycin (25) in slow exchange with the free form (Fig. 1d) [160]. This suggests that the change of the hydroxyl group with the amino group at position 6⬘ of ring I has little effect on aminoglycoside binding affinity. Other observations showed NMR chemical shift changes for ribostamycin (28) and neamine (4) similar to those collected for neomycin and paromomycin complexes, but with smaller magnitudes, indicating that rings I and II of the neomycin class of aminoglycosides are sufficient to direct their binding to a unique binding pocket on the model A-site RNA. The specificity of ribostamycin and neamine interactions with the A-site RNA also was demonstrated by footprinting on mutant oligonucleotides. The U1495 → A mutation resulted in significant decreases of binding affinity in paromomycin, ribostamycin, and neamine due to disruption of the ring II hydrogen bonding, indicating specific targeting to the A site of at least ring II of these aminoglycoside antibiotics. In a similar study, several functionalities in the major groove of the A-site RNA, including G1405 (N7), G1491 (N7), G1494 (N7), A1408 (N7), A1493 (N7), A1408 (N1), A1492 (N1), and A1493 (N1), as well as the pro-R phosphate oxygens of A1492 and A1493 in 16S rRNA (Fig. 1c) were identified as essential for a highaffinity paromomycin binding to the 16S rRNA from E. coli (161). Wong and colleagues investigated the specificity of ring IV of neomycin (Fig. 1) in binding to the A-site RNA [162]. They used the A-site RNA sequence previously employed by Puglisi and coworkers [15], which mimics the A site of the prokaryotic 16S ribosomal RNA as a binding target for antibiotic, to design a series of neomycin B derivatives modified in the idose ring (Fig. 3). The sequence of RNA with U1495 mutated to A (Fig. 3) was used as a negative control for in vitro binding studies because U1495 is a critical base for specific aminoglycoside binding, and replacement of this base with A leads to complete loss of specific binding [15,16]. The binding results showed similar specificity for neamine (4) and ribostamycin (28), but threefold lower affinity was observed for 28. Addition of an uncharged idose moiety to 28 to generate 129 does not change the affinity or specificity of binding, indicating the importance of the charged amino groups of the idose ring for binding. A 40-fold higher activity for compound 130, compared to 129, and further improvement of affinity in neomycin B (23) compared to that of 130, prove that the amino groups of idose ring are necessary for specific binding in vitro and that they cannot be replaced by the amino groups attached to flexible linkers. The basis for this assertion is that the addition of an amino group to ribostamycin via a flexible linker to furnish 127 increases the affinity to some extent, but specificity remains unaffected. Addition of one more amino group to generate 128 results in diminishing returns on the affinity without affecting the specificity. However, the minimum inhibitory concentration (MIC) data of these compounds showed comparable antibacterial activity to that of neomycin B, indicating that the in vivo activity does not always correlate well with the in vitro binding results. These discrepancies reflect probable conformational difference of the ribosome in vivo. Wong and coworkers in a recent investigation [163] studied the A-site-binding aminoglycosides, including the 4,5-disubstituted 2-deoxystreptamine derivatives (neomycin B, paromomycin, and ribostamycin) and the 4,6-disubstituted 2-deoxystrep-

332

Haddad et al.

Figure 3 (a) Synthetic aminoglycosides to probe the role of ring IV of neomycin B. (b) The sequence of the model for the A-site RNA. (c) The sequence of RNA used as a negative control.

tamine aminoglycosides (kanamycins and gentamicins), to probe their specificities for the A site of the decoding region of rRNA. Two other structurally dissimilar Asite-binding antibiotics, hygromycin B (16) and apramycin (13), were tested as well. Figure 4 shows the structures of some aminoglycoside antibiotics. One common characteristic of all these A-site binders in chemical footprinting is the strong protection of G1494 (N7) of the 16S ribosomal RNA, although there are significant differences in terms of other footprints and resistance patterns. For example, methylation of G1405 (N7) confers resistance to gentamicins and kanamycins, but not to neomycin B. Surface plasmon resonance (SPR) [164] was used for direct observation of aminoglycoside interactions with the A-site RNA. These results indicated poor specificity for 4,6-disubstituted 2-deoxystreptamine aminoglycosides (kanamycins and gentamicins), good specificity for 4,5-disubstituted (neomycin class) antibiotics, moderate specificity for neamine, marginal specificity for apramycin (13), and none for hygromycin B (16). Likewise, streptomycin (73), which does not target the A site of the ribosome, shows no specificity for the A-site RNA sequence. These results suggest that the aminoglycosides target different regions of the same ribosomal RNA involving different RNA tertiary structural features, which cannot be well understood by the simple model for the A-site RNA.

Aminoglycoside Antibiotics

Figure 4

333

Structures of some aminoglycoside antibiotics.

Extending their work to search for small molecules that recognize RNA, Wong et al., considered the 1,3-hydroxyamine motif as a core for such molecules [162,165]. 1,3-Hydroxyamines can be effective motifs for molecular recognition of phosphodiesters because an interesting and common feature of almost all aminoglycoside antibiotics is the presence of a trans-1,3-hydroxyamine or cis-1,3-diamine motif. In fact, a model study by NMR spectroscopy [165] showed that the glyco-type 1,3hydroxyamines interact via multiple hydrogen bonds, with phosphodiesters and the Hougsteen face of guanine, which is typically stronger than the cyclic guanidine group (Fig. 5). To test these findings, new aminoglycoside derivatives were synthesized in a series bearing either one or both of these recognition motifs for RNA interaction

334

Haddad et al.

Figure 5 Strong interactions of ␤-1,3-hydroxyamine motif found in aminoglycoside antibiotics with (a) phosphodiesters and (b) Hougsteen face of guanine, compared to that of standard cyclic guanidine (c).

[162]. Some of the new compounds showed strong antibiotic activities and were effective in binding to the A-site RNA sequence (Fig. 3b) and to a drug-resistant mutant RNA in the micromolar range. For instance, compound 128 (Fig. 3a) with the neamine moiety as the recognition motif showed a Kd value of 0.26 ␮M [162], and the pyranoside derivative compound 131 bearing the 1,3-hydroxyamine motif as the core exhibited a Kd value in the range of 10 ␮M in binding to the A-site RNA model [162,166]. Application of this strategy in designing small molecules that target RNA could lead to development of potent sequence-specific RNA binders and potential discovery of new aminoglycoside antibiotics with less toxicity. In a recent communication by Rando and colleagues [167], simple aminols were reported to be effective A-site RNA binders with a potency similar to or greater than the potent RNA binder aminoglycoside paromomycin (25). Discovery of these simple 1,3-(2)-aminol-containing molecules (e.g., 132 and 133), which act as potent substituents for the structurally complicated aminoglycosides in binding to the bacterial A-site decoding region, would pave the way for the design and synthesis of a new class of potent antibiotics.

Compound 131

Aminoglycoside Antibiotics

335

Compounds 132 and 133

The measured dissociation constant for compounds 132 and 133 were 1.01 ⫾ 0.03 and 2.95 ⫾ 0.04 ␮M, respectively, which are in the range of the Kd value measured for paromomycin (Kd = 1.90 ␮M) binding to the A-site decoding region. These data actually show twofold greater potency for aminol 132 than for paromomycin, despite the structural simplicity of 132. It is noteworthy that compound 132, which contains three stereogenic centers, could exhibit more potency for the binding with the A site when it is used as a diasteriomerically pure isomer. Perhaps 1,3-(2)aminol moiety represents the essential structural element of aminoglycosides required for RNA interaction. B.

Specificity of Aminoglycosides for Binding to RNA Constructs Derived from the HIV-RRE and TAR RNA Activator Regions

Aminoglycoside binding to RNA is not limited to binding to prokaryotic 16S ribosomal RNA. These positively charged molecules interact with the HIV-1 Rev Response Element (RRE) transcriptional activator region [17] and TAR RNA [151], interfere with intron splicing [146], and bind to hammerhead ribozymes [153]. The HIV-1 RRE is an important region of the viral RNA, required for accurate and successful transcription of the HIV-1 genome [168,169], and replication of RNA viruses is dependent on specific interactions between the viral RNAs and viral proteins, such as Rev and Tat. The function of the arginine-rich Rev protein appears to be transportation of the RNA through the nuclear membrane into the cytoplasm, where the RNA is translated into viral proteins [168,169,170]. Therefore, inhibitors that interfere with the Rev-RRE binding could lead to the development of drugs effective in the treatment of AIDS [17]. A particularly useful fragment of Rev protein is Rev30–50 (a 16-mer peptide derived from the Rev protein; Fig. 6), which interacts with RRE IIB much as the full Rev protein does [171]. It exists in an ␣-helical conformation [172], and modifications of the peptide that increase the ␣-helical content and affinity of the protein, led investigators to conclude that the Rev protein contains an ␣-helical domain, which specifically interacts with the RRE region of the viral RNA [173]. Rando and coworkers employed fluorescence anisotropy methods to quantitatively measure aminoglycoside–RNA interactions [174,175]. These investigators used the RRE IIB construct of the HIV-1 RRE transcriptional activation region to determine the affinities of aminoglycoside to this biologically relevant construct. In this study, a fluorescent analog of Rev34–50 (Fl-Rev34–50) was prepared and shown by

336

Figure 6

Haddad et al.

Structures of HIV-1 RRE IIB RNA and Rev peptides.

fluorescence anisotropy measurements to bind to RRE IIB in a stoichiometric fashion, with a dissociation constant of 7.6 nM. On the other hand, when neomycin B (23) and tobramycin (42), as well as a series of 6⬘-substituted tobramycin analogs were bound to RRE IIB, in competition experiments with Fl-Rev34–50, binding affinities in the range of 1–2 ␮M were reported. These results indicate that binding of aminoglycosides to the RRE IIB region of the construct is not highly selective. Similarly, aminoglycoside antibiotics block the binding of HIV-1 Tat-derived peptide to its viral RNA target in a transactivator-responsive region (TAR). In a study by Mei and colleagues [151], Tat, an 86 amino acid regulatory protein of HIV-1 that contains an arginine-rich basic domain (residues 47–58) involved in binding to TAR RNA (residues 1–59 of HIV-1 mRNA), was replaced by a synthetic 40 amino acid (Tat40) peptide consisting of the TAR binding domain of the Tat protein. Interactions of this peptide (Tat40) with a 31-nucleotide RNA (TAR31), which contained residues 18–44 of the native TAR RNA, in the presence and absence of aminoglycosides, were monitored by gel electrophoresis. Three classes of aminoglycoside antibiotics, neomycin B (23: a 4,5-disubstituted 2-deoxystreptamine aminoglycoside), gentamicin C (51: a 4,6-disubstituted 2-deoxystreptamine), and a structurally dissimilar aminoglycoside, streptomycin (73), were chosen for inhibition of Tat40 binding TAR31. Among these selected antibiotics, neomycin B exhibited an IC50 of 0.92 ⫾ 0.09 ␮M, whereas IC50 values of streptomycin and gentamicin C were 9.5 ⫾ 0.8 and 45 ⫾ 4 ␮M, respectively, suggesting that the IC50 values are not simply correlated to the number of charged amino groups in these antibiotics. For instance, streptomycin (⫹3 charge), which is 10-fold less effective than neomycin (⫹6 charge), is 5-fold more effective than gentamicin (⫹5 charge) in inhibiting Tat40 binding.

Aminoglycoside Antibiotics

C.

337

RNA Molecules That Specifically Bind Aminoglycoside Antibiotics with High Affinities

Despite a paucity of information at the molecular level on the principles associated with specific recognition of RNA sequences by aminoglycosides, targeting RNA sequences and recognition of specific sites of RNA molecules by aminoglycoside antibiotics are topics of salient interest. However, recent developments in in vivo selection methods resulted in identification of particular RNA sequences that specifically bind to certain aminoglycoside antibiotics with high affinity [174,176,177]. Rando and colleagues [178] have shown recently that a small (40-nucleotide) stem–loop (J6f1), derived from the high-affinity RNA aptamer J6 [174], can still bind tobramycin (42) specifically and with high affinity, as is the case with J6 itself. This construct (Fig. 7), which is quite specific for tobramycin, contains two bulges, one of three nucleotides and another of one that are essential for high-affinity tobramycin binding. The J6f1 aptamer can stoichiometrically bind tobramycin with a dissociation constant in the range of 5 nM, and it discriminates among similar aminoglycosides for binding. The aminoglycoside binding site of this construct was defined by a combination of chemical interference studies, chemical modification, and mutational studies. It was found that specific interactions of tobramycin with certain nucleotide bases in the stem–loop region are responsible for recognition of the aminoglycoside by J6f1, and specific interactions of tobramycin with the phosphate backbone are apparently not important for binding. Rando and colleagues [178] concluded that interactions of their positively charged amino groups with the negatively charged phosphate backbone of the RNA may not result in high-affinity binding in the nanomolar range. However, most of the aminoglycoside binding RNA molecules described so far appear to bind aminoglycosides in the micromolar range. To generate strong binders in the nanomolar range, specific contacts between aminoglycoside and nucleotide bases, rather than

Figure 7

(a) Structures of tobramycin and the 40-nucleotide RNA sequence (J6f1). (b) Structures of the 27-nucleotide RNA aptamer I and the 26-nucleotide RNA aptamer II.

338

Haddad et al.

interaction with phosphate backbone, appear to be essential in RNA–aminoglycoside binding. Patel and coworkers [179,180] undertook an NMR–molecular dynamics study to solve the solution structure of the tobramycin complexed with two RNA aptamers. The 27-nucleotide RNA construct I (Fig. 7b) specifically binds to tobramycin (42) with affinity in the nanomolar range [174]. These investigators solved a low-resolution structure for tobramycin bound to the RNA aptamer I [179]. Their initial findings showed that tobramycin binds to the major groove centered about a stem– loop junction site, such that significant segments of the antibiotic are encapsulated between the floor of the major groove and a looped-out cytosine residue that forms a flap over the binding pocket in the complex. The high-resolution solution structure of the tobramycin–RNA aptamer II complex [180] revealed that the ring II of tobramycin is centrally positioned in the deep groove (about the G9 ⭈ A18 mismatch pair), such that its hydroxyl groups face toward the floor of the groove (Fig. 8). The 2-deoxystreptamine moiety (ring I) has its hydrophobic face directed toward the floor of the deep grove centered about the U10 ⭈ A17 and positioned closer to the C9-U10-U11 bases of the RNA construct. The amino sugar ring II, which is centered about the U12 base, interacts with the RNA deep groove through its positively charged amino group at position 2⬘, whereas its amino group at position 6⬘ is directed toward the milieu. The RNA aptamer II (Fig. 7b) discriminates against closely related aminoglycoside antibiotics. For instance, kanamycin B (38), which contain a hydroxyl group at the 3⬘-position instead of a hydrogen, as in tobramycin 42, forms a 1:1 complex with the RNA aptamer II. Whereas kanamycin A (37: replacement of amino group at the 2⬘-position of kanamycin B with a hydroxyl group) binds poorly to the same RNA target. Similarly, gentamicin C (51), having 3⬙-NH2 replaced by NHCH3, shows poor binding ability for this RNA aptamer. These findings demonstrate the importance of the amino groups at positions 2⬘ and 3⬙ for specific interactions with the edges of the base pairs of the deep groove of this particular RNA aptamer, which result in high specificity of binding for tobramycin. D.

Aminoglycoside Binding to the Hammerhead Ribozyme

The hammerhead ribozymes, derived from self-cleaving sequences found within natural RNA replicons, are like other self-cleaving RNA molecules in that they undergo a transesterification reaction that generates the 5⬘-hydroxyl and 2⬘,3⬘-cyclic phosphate termini. Crystal structures of two hammerhead ribozymes have been reported [181,182,183], and they have been shown to be efficiently inhibited by aminoglycoside antibiotics, particularly neomycin B (23) [153]. The activities of the hammerhead ribozymes characterized to date are strongly stimulated by the presence of divalent metal ions, typically Mg2⫹, and it is suggested that competition of positively charged aminoglycosides with Mg2⫹ is relevant to aminoglycoside binding and inhibition [184]. On the other hand, Murray and colleagues [185] have demonstrated that metal ions are not obligatory chemical participants in the catalytic reactions and self-cleavage of hammerhead ribozyme. These findings suggest that the catalytic function of the ribozyme is influenced by the folded structure of the RNA itself, as well as the presence of a relatively dense positive charge, than by divalent metal ions.

Aminoglycoside Antibiotics

339

Figure 8

The NMR structure of the tobramycin–RNA complex. Tobramycin is shown by the capped-stick presentation. A portion of the RNA structure at the aminoglycoside binding site is shown as a surface.

Aminoglycoside binding to the hammerhead ribozyme 16 (HH16; Fig. 9a) [153] was found to be specifically effective with neomycin B. Interactions of the antibiotic with the enzyme–substrate complex were suggested to reduce the cleavage rate by stabilizing the ground state and destabilizing the transition state of the complex, leading to phosphodiester cleavage. These workers studied the inhibition as a function of pH and Mg2⫹ concentration and concluded that the positively charged antibiotic inhibits the hammerhead ribozyme by displacing critical magnesium ions at physiological pH. Wang and Tor [186], who probed the role of electrostatic interactions in aminoglycoside–RNA binding, found that the positively charged amino groups play an important role in affinity of binding and displacement of Mg2⫹ ions. These workers made a series of aminoglycosides with increasing number of amino groups (compounds 134–136 , Fig. 9b) by simple conversion of the hydroxymethyl group in the natural aminoglycosides to the basic aminomethyl group. All these synthetic amino

340

Haddad et al.

Figure 9

(a) The secondary structure of the hammerhead ribozyme 16 (HH16) showing the enzyme (E16) and the substrate (S16). (b) Structures of some semisynthetic and natural aminoglycosides.

derivatives showed higher binding affinity for the hammerhead ribozyme than the parent compounds. For instance, 6⬙-amino-6⬙-deoxykanamycin A (134) showed enhanced affinity compared to kanamycin A (37), a poor RNA binder, and was as effective as kanamycin B (38), a natural product containing five amino groups. Likewise, 6⬙-amino-6⬙-deoxytobramycin 135 and 5⬙-amino-5⬙-deoxyneomycin B (136) showed about fivefold enhanced binding affinities for the hammerhead ribosome, suggesting the importance of the overall charge of these small molecule ligands for RNA binding affinity. These investigators [187] have also shown that the binding affinities of aminoglycosides for the hammerhead ribosome correlate with the basicity of the amino groups in these RNA binders. They compared binding affinities of two aminoglycoside antibiotics tobromycin (42) and kanamycin B (38), which have same number of amino groups (five). Tobramycin, which is 3⬘-deoxykanamycin B, inhibits the

Aminoglycoside Antibiotics

341

hammerhead ribozyme more effectively than kanamycin B itself. Therefore, the absence of a hydroxyl group next to an amino group might increase the basicity of that amino group inductively and enhance the RNA binding affinity of the compound. To probe this possibility, Wang and Tor [187] prepared a number of deoxygenated tobramycin analogs (compounds 137–140 ) and found increased inhibitory activity for 4⬘-, 2⬙-, and 4⬙-deoxy analogs, which involved hydroxyl group deletion on a carbon vicinal to an amino group. In contrast, 6⬙-deoxytobramycin (140), with hydroxyl deletion remote from an amino group, exhibited poorer inhibitory activity than the parent antibiotic. These results support the critical role of electrostatic interactions in RNA binding, indicating the importance of the basicity of the amino groups in RNA binding affinity of aminoglycoside antibiotics at physiological pH. Hermann et al. [188,189] have demonstrated that the positively charged amino groups of an aminoglycoside antibiotic are capable of displacing three or four magnesium ions located within the cavity formed by the deep grooves of stems I and II in the hammerhead crystal structure [182,183]. These workers used molecular dynamics simulations on complexes of the hammerhead RNA and aminoglycoside antibiotics and showed that the interactions between RNA and amino groups of aminoglycosides are almost identical with the observed interactions for Mg2⫹ in the hammerhead crystal structure [188]. It has also been noted that more than one solution conformation of aminoglycoside antibiotics could be docked with the hammerhead RNA. These would lead to a set of spatially oriented, positively charged amino groups that could be complementary in various ways to the field created by an electronegative cavity of the hammerhead RNA, which results in high-affinity binding. These high-affinity interactions of different solution conformations of aminoglycosides could be facilitated by water-mediated interaction between the antibiotic and the RNA molecules. This diversity in recognition and binding, derived from water-assisted, spatially positioned charges, suggests a three-dimensional electrostatics model rather than a highly spe-

Compounds 137–140

342

Haddad et al.

cific drug–RNA interaction, for interaction between these charged antibiotics and different RNA molecules [190]. E.

Inhibition of Group I Introns by Aminoglycoside Antibiotics

Inhibition of the first step of splicing by binding of an exogenous guanosine to the G-binding site (guanosine binding site) in the ribozyme results in cleavage of the 5⬘ splice site via a transesterification process, leading to attachment of the guanosine to the first nucleotide of the intron. The second step of splicing, which involves a configurational transition for proper accommodation of the terminal nucleotide of the intron (always a guanosine) into the G-binding site, occurs via a second transesterification process that cleaves the 3⬘ splice site, resulting in ligation of exons and releasing of the intron [191,192]. Von Ahson and colleagues [144,146] demonstrated that aminoglycoside antibiotics inhibit self-splicing of group I intron RNA in vitro at micromolar concentrations. These workers tested a series of 4,5- and 4,6-disubstituted 2-deoxystreptamine aminoglycoside antibiotics, including neomycins, kanamycins, and gentamicins [146]. Of these, neomycin B (23) was found to be the most active aminoglycoside in inhibiting the splicing reaction. In the kanamycin family, the order of activity for inhibition of splicing was shown to be tobramycin (42) > kanamycin B (38) > kanamycin A (37) > kanamycin C (39). However, tobramycin and gentamicin C (51) exhibited inhibitory activities comparable to that of neomycin B, in the range of 1 ␮M. It was thought that aminoglycoside antibiotics inhibit only the second step of the splicing of the td intron, but further study of structure–activity relationships between aminoglycoside and self-splicing of group I introns revealed that antibiotics other than gentamicin (e.g., 5⬘-epi-sisomicin) also inhibit the first step of splicing with inhibitory activity concentration similar to that of neomycin B, in the range of 0.5 ␮M [144]. The known inhibitors of group I intron RNA self-splicing, including dGTP, ddGTP, arginine, and streptomycin, which exert their inhibitory effects by competing with the substrate guanosine, show Ki values in the millimolar range [143,193,194]. Therefore, guanosine (estimated intracellular concentration of 1 mM) would compete with these inhibitors, such that no inhibition would take place in vivo. On the other hand, these new inhibitors inhibit splicing by a different mechanism at micromolar concentration, and they show noncompetitive inhibition at high concentrations of guanosine [144]. To elucidate the mechanism of splicing, a mutant group I intron, having the conserved guanosine of the G-binding site replaced by an adenosine, was employed in splicing inhibition studies. This mutant variant was as sensitive to aminoglycoside antibiotics as the wild type, providing direct evidence that the antibiotics do not interact with the G-binding site. Furthermore, the results suggested that the binding site(s) might be in a region near the catalytic core of the group I introns and not necessarily at the G-binding site [144]. Further studies by von Ahsen and Noller [145] on footprinting the sites of interaction of neomycin with the catalytic group I intron RNA suggested that neomycin prevents docking of the p1 stem (bearing the 5⬘ splice site) to the catalytic core and inhibits cofactor binding. Aminoglycoside antibiotics as specific inhibitors of splicing might be of interest should they inhibit growth of the organisms [195], with possible applications in development of a new class of antifungal agents.

Aminoglycoside Antibiotics

F.

343

Inhibition of the Human Hepatitis Delta Virus (HDV) by Aminoglycoside Antibiotics

The human hepatitis delta virus (HDV), a 1700-nucleotide single-stranded RNA virus, like other self-cleaving RNAs, undergoes a transesterification reaction that generates a 5⬘-hydroxyl group [196]. Both genomic and antigenomic sequences contain ribozymes that are important for viral replication, and it appears that the HDV ribozyme requires divalent metal ions for self-cleavage [185]. In 1996 Rogers et al. [156] used genomic and antigenomic plasmid constructs of the HDV ribozyme and demonstrated that autocatalytic activity of both HDV ribozymes can be inhibited by aminoglycosides at micromolar concentrations. A variety of aminoglycosides and nonaminoglycoside antibiotics known to inhibit the group I intron splicing were tested for their inhibitory activity against these selfcleaving HDV ribozymes. Among aminoglycoside antibiotics tested, only a small number, including gentamicin C (51), neomycin (23), and 5⬘-epi-sisomicin were active as inhibitors, yet their inhibitory activities were less than those for inhibition of the group I intron ribozyme. Kanamycin A (37) was a poor inhibitor, and paromomycin (25: having 6⬘-NH2 of neomycin B replaced by a hydroxyl group) showed poor activity relative to neomycin B. The mechanism whereby neomycin B inhibits ribozyme function has yet to be elucidated. However, existing results [156] suggest a competitive inhibition of this catalytic RNA by neomycin B, implying that this antibiotic inhibits HDV ribozyme cleavage by disruption of the divalent cationic binding near the catalytic core of the ribozyme.

VI.

TOXICITY OF AMINOGLYCOSIDE ANTIBIOTICS

It has been known for a long time that aminoglycoside antibiotics cause ototoxicity (i.e., mid-ear damage resulting in loss of hearing). It is known that aminoglycoside treatment results in the destruction of cochlear cells in the mid-ear that are responsible for the reception and analysis of sound. The cases of vestibular toxicity and loss of hearing are well documented owing to long-term use of aminoglycosides, in particular gentamicin [197]. Aminoglycosides are also known to cause vestibular nerve damage, causing loss of balance (ataxia), dizziness, and so on. Ototoxicity has been a serious problem associated with aminoglycosides. The mechanism of such toxicity has been elusive. Recently, a hypothesis for the aminoglycoside toxicity was proposed [198,199]. Observations that gentamicin chelates with iron and also forms free radicals led to the hypothesis that iron chelation and free radical generation might be sources of gentamicin toxicity. In addition, it was noted that free radical scavengers and iron chelators reduce the toxicity of gentamicin, as well as that of other aminoglycosides such as kanamycin and streptomycin [200]. Further experimentation will be necessary to confirm either mode of toxicity manifestation. Various studies with aminoglycosides have also led to the hypothesis that even after the use of aminoglycoside antibiotics has been discontinued, the compounds accumulate in the biological system and may manifest their toxic effects in symptoms such as ototoxicity. Loss of hearing, a result of hair cell toxicity, appears to be due to mitochondrial mutation in hair cells in individuals who are susceptible to this side effect [201].

344

Haddad et al.

Aminoglycosides possess affinity toward prokaryotic RNA, due to which they exhibit strong binding to various sites on RNA. However, eukaryotic mitochondrial ribosomes (such as in hair cells) are structurally similar to their ancestral prokaryotic ribosomes. This perhaps explains the affinity of aminoglycosides to mitochondrial RNA and their ability to interfere in the process of protein synthesis in mitochondria [201]. From a study of 41 human families, it was concluded that the hypersensitivity toward aminoglycosides is transmitted via maternal lineage. Studies involving several hundred individuals indicated that a genetic mutation in mitochondrial 12S subunit of ribosomal RNA at position 1555 to a ‘‘G’’ was perhaps responsible for the hypersensitivity toward aminoglycosides [201]. From the comparison of secondary structures of prokaryotic rRNA and mitochondrial 16S rRNA, it was concluded that the G1555 mutation lies in a very highly conserved region of the stem–loop that is involved in aminoglycoside binding in bacteria. It was argued that the G1555 mutation forms an additional base pair with a cytosine on the opposite strand of rRNA, and this leads to shrinkage of the volume of rRNA at this site, allowing more space for aminoglycoside binding [201]. Based on this mechanism of mitochondrial toxicity, Hutchin and Cortopassi proposed a molecular mechanism of the toxicity to hair cells in conjunction with the foregoing rRNA mutation [201]. According to their hypothesis (Fig. 10), binding of aminoglycoside to the mitochondrial rRNA leads to mistranslation of mitochondrial proteins and a deficiency of mitochondrial complex I. These events lead to the deficiency of superoxide dismutase and cause overproduction of superoxide in mitochondria. Thus, the mitochondria are subjected to oxidative damage, leading to cell death. Evidence exists that intrastriatal administration of neomycin to rats causes damage to the glial cells in the central nervous system [202]. Upon coadministration of N-methyl-D-aspartate (NMDA), gliosis was enhanced, indicating that neomycin is acting at the NMDA receptors. Further, in the presence of NMDA receptor antagonists, the toxicity exhibited by neomycin was attenuated. However, paromamine and 2-deoxystreptamine did not produce such damage. This study shows that aminoglycosides such as neomycin may cause excitation of NMDA receptors, which may, at least in part, be the cause of ototoxicity of aminoglycosides. When the blood– brain barrier does not deter aminoglycosides, these compounds may cause central nervous system damage.

Figure 10 The hypothetical mechanism of damage to hair cells due to the aminoglycosideinduced mistranslation of mitochondrial genes [201].

Aminoglycoside Antibiotics

345

Aminoglycosides seem to function like polyamines at the NMDA receptors [203]. Endogenous polyamines such as spermine and spermidine and various derivatives of these compounds serve as positive modulators of NMDA receptors, and their interactions are complex. It is proposed that the primary and secondary amines present on the aminoglycoside molecules render the polyamine-like properties to aminoglycosides, and it appears that all the clinically useful aminoglycoside antibiotics exhibit polyamine-like properties with the NMDA receptor [203]. Surprisingly, individual sugar moieties in aminoglycosides did not produce potent modulation of NMDA receptor activity, indicating that the NMDA receptor binding is the property of the entire molecule rather than the effect of individual amino sugars. It also appears that an increased number of amino groups on an aminoglycoside does not necessarily increase the NMDA binding ability of aminoglycosides or increase the ototoxicity, although this remains a controversial issue. Recently, Sha and Schacht performed a study with gentamicin in guinea pigs and showed that simple iron chelators such as malic acid, tartaric acid, and even dimethyl sulfoxide, can protect these animals from the hearing loss caused by gentamicin [204]. Malic acid and tartaric acid are dicarboxylic acids that can chelate iron, and dimethyl sulfoxide scavenges free radicals such as hydroxyl radical. The most important and common acute toxicity caused by aminoglycoside treatment is renal dysfunction or nephrotoxicity, and comprehensive review articles on the molecular mechanisms of nephrotoxicity of aminoglycosides have been published [205,206,207]. Clinically, renal toxicity is manifested by the increase of blood urea and serum creatinine levels, as well as by alteration of the glomerular filtration rate (GFR). Although the effects of aminoglycosides on the kidney are reversible and rarely cause death, since renal clearance is affected, kidney dysfunction leads to complications in patients who are on multiple drugs and in other hospitalized patients. Aminoglycosides are not metabolized in the body significantly and are excreted by the kidneys. In kidneys, aminoglycosides are filtered through the glomerulus, and some amount is reabsorbed in the proximal tubule. Aminoglycosides bind to the brush-border membranes of proximal tubules and enter the tubular cells by endocytosis. Thus, aminoglycosides accumulate in the proximal tubular cells and distal parts of nephron, reaching concentrations much higher than those in serum. Following the uptake of aminoglycosides, proximal tubular epithelial cells show structural alterations leading to tubular necrosis by a number of mechanisms [205,206]. Aminoglycosides are known to be associated with lysosomes to form ‘‘myeloid bodies’’ and result in swelling of lysosomes. Additionally, aminoglycosides are implicated in altering the plasma membrane structure by binding to the phospholipids. Phospholipid catabolism in the proximal tubular cells is impaired by aminoglycosides owing to decreased activity of sphingomyelinase and phospholipase A1, leading to the accumulation of phospholipids in lysosomes. Aminoglycosides are protonated in the acidic environment of the lysosome and bind to the phospholipids. There is no experimental evidence on how the binding of aminoglycosides to the phospholipids inhibits the hydrolytic activity of phospholipases, but it is thought that aminoglycosides create a less favorable environment for the cleavage of phospholipids by the hydrolytic enzymes in the lysosome [207]. There are also hypotheses that aminoglycosides damage mitochondrial respiratory mechanisms, leading to decreased levels of ATP, and may alter the activity of Na⫹/K⫹-ATPase [206]. It is not clear which

346

Haddad et al.

mechanism accounts for the proximal cell death, but damage of these cells leads to the altered GFR. Most of the data presented so far show that the phospholipidosis due to aminoglycosides leads to the tubular cell death, ultimately impairing renal function. Although the impairment of the kidneys by aminoglycosides is serious, the aminoglycoside-damaged cells in the proximal tubules are regenerated eventually, restoring the renal tissue architecture [205,208]. The extent of nephrotoxicity is a characteristic of the individual aminoglycoside antibiotics and appears to be directly related to the number of cationic groups on the aminoglycoside [207]. REFERENCES 1. 2. 3. 4.

5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.

JE Kapusnik, CJ Hackbarth, HF Chambers. J Infect Dis 158:7–12, 1988. B Vogelman, W Craig. J Pediatr 5:835–840, 1986. BD Davis. Microbiol Rev 57:341–350, 1987. K Bush, S Mobashery. In: BP Rosen, S Mobashery, eds. Resolving the Antibiotic Paradox: Progress in Understanding Drug Resistance and Development of New Antibiotics. New York: Plenum Press, 1998, pp 71–98. LP Kotra, S Mobashery. Bull Inst Pasteur 96:139–150, 1998. RS Edson, CL Terrell. Mayo Clin Proc 66:1158–1164, 1991. A Schatz, E Bugie, SA Waksman. Proc Soc Exp Biol Med 55:66–69, 1944. HC Hinshaw, WH Feldman. Mayo Clin 20:313–318, 1945. JM Musser. Clin Microbiol Rev 8:496–514, 1995. N Tanaka. In: H Umezawa, IR Hooper, eds. Handbook of Experimental Pharmacology. New York: Springer-Verlag, 1982, pp 221–266. E Cundliffe. In: WE Hill et al, eds. In the Ribosome. Washington, DC: American Society for Microbiology, 1990, pp 479–490. HF Noller. Annu Rev Biochem 60:191–227, 1991. D Moazed, HF Noller. Nature 327:389–394, 1987. P Puhroit, S Stern. Nature 370:659–662, 1994. MI Recht, D Fourmy, SC Blanchard, KD Dahlquist, JD Puglisi. J Mol Biol 262:421– 436, 1996. H Miyaguchi, H Narita, K Sakamoto, S Yokoyama. Nucleic Acids Res 24:3700–3706, 1996. ML Zapp, S Stern, MR Green. Cell 74:969–978, 1993. G Werstuck, ML Zapp, MR Green. Chem Biol 3:129–137, 1996. Y Wang, K Hamasaki, RR Rando. Biochemistry 36:768–779, 1997. S Hanessian, TH Haskell. In: H Pigmann, D Horton, eds. Antibiotics Containing Sugars. New York/London: Academic Press, 1970. pp 139–211. S Umezawa. Adv Carbohydr Chem Biochem 30:11–182, 1974. DA Cox, K Richardson, BC Ross. In: PG Sammes, ed. The Aminoglycosides. New York: John Wiley & Sons, 1977, pp 1–90. KE Price. In: D Perlman, ed. Structure–Activity Relationships Among the Semisynthetic Antibiotics. New York: Academic Press, 1977, pp 1–84. H Umezawa, IR Hooper. Handbook of Experimental Pharmacology. New York: Springer-Verlag, 1982. S Umezawa, S Kondo, Y Ito. Biotechnology 4:309–357, 1986. CE Higgens, RE Kastner. Antimicrob Agents Chemother 1967:324–331, 1968. A Awata, S Satoi, N Muto, M Hayashi, H Sagai, H Sakakibara. J Antibiot 36:651– 655, 1983. K Kamia, T Deushi, A Iwasaki, I Watanabe, H Itoh, T Mori. J Antibiot 36:738–741, 1983.

Aminoglycoside Antibiotics 29. 30. 31. 32.

33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58.

347

EA Swart, D Hutchinson, SA Waksman. Arch Biochem 24:92–103, 1949. M Kojima, Y Yamada, H Umezawa. Agric Biol Chem 32:467–473, 1968. S Sato, S Takasawa, T Sato, M Yamamato, R Okachi, I Kawamoto, T Iida, A Morikawa, T Nara. J Antibiot 30:25–30, 1977. M Murase, T Ito, S Fukatsu, H Umezawa. Studies on kanamycin-related compounds produced during fermentation by mutants of Streptomyces kanamyceticus. Isolation and Properties. Progress in Antimicrobial and Cancer Chemotherapy, Proceedings of Sixth International Congress on Chemotherapy, vol II. Baltimore: University Park Press, 1970, pp 1098–1110. KF Koch, JA Rhoades. Antimicrob Agents Chemother 10:309–313, 1970. T Oda, T Mori, H Ito, T Kunieda. J Antibiot 24:333–338, 1971. J Berdy, HK Pauncz, ZM Vajna, G Horvath, J Gyimesi, I Kaczka. J Antibiot 30:945– 954, 1977. RL Mann, WW Bromer. J Am Chem Soc 80:2714–2716, 1958. S Kondo, M Sezaki, M Koike, M Shimura, E Akita, K Satoh, T Hara. J Antibiot A18: 38–42, 1965. S Inouye, T Shomura, H Watanabe, K Totsugawa, T Niida. J Antibiot 26:374–385, 1973. Y Ikedo, S Kondo, F Kanai, T Sawa, M Hamada, T Takeuchi, H Umezawa. J Antibiot 38:436–438, 1985. M Shimura, Y Sekizawa, K Iinuma, H Naganawa, S Kondo. J Antibiot 28:83–84, 1975. J Shoji, S Kozuki, M Mayama, Y Kawamura, K Matsumoto. J Antibiot 23:291–294, 1970. SA Waksman, HA Lechevalier. Science 109:305–307, 1949. PB Alper, M Hendrix, P Sears, CW Wong. J Am Chem Soc 120:1965–1978, 1998. S Rengaraju, S Narayanan, PL Ganju, MA Amin, MRS Iyengar. J Antibiot 39:1598– 1601, 1986. D Autissier, P Barthelemy, N Mazieres, M Peyre, L Penasse. J Antibiot 34:536–543, 1981. TH Haskell, JC French, QR Bartz. J Am Chem Soc 81:3482–3483, 1959. T Shomura, N Ezaki, T Tsuruoka, T Niwa, E Akita, T Niida. J Antibiot 23:155–161, 1970. S Horii, I Nogami, N Mizokami, Y Arai, M Yoneda. Antimicrob Agents Chemother 5: 578–581, 1974. JP Kirby, DB Borders, GE Van Lear. J Antibiot 30:175–177, 1977. HW Dion, PWK Woo, NE Willmer, DL Keru, J Onaga, SA Fusari. Antimicrob Agents Chemother 2:84–88, 1972. H Tsukiura, K Saito, S Kobaru, M Konishi, H Kawaguchi. J Antibiot 26:386–388, 1973. H Kawaguchi, K Tomita, T Hoshiya, T Miyaki, K Fujisawa, M Kimeda, K Numata, M Konishi, H Tsukiura, M Hatori, H Koshiyama. J Antibiot 27:460–470, 1974. H Umezawa, K Ueda, K Maeda, K Yagishita, S Kondo, Y Okami, R Utahara, Y Osato, K Nitta, T Takeuchi. J Antibiot 10:181–189, 1957. K Maeda, M Ueda, K Yagishita, S Kawaji, S Kondo, M Murase, T Takeuchi, Y Okami, H Umezawa. J Antibiot A10:228–232, 1957. KF Koch, FA Davis, JA Rhoades. J Antibiot 26:745–751, 1973. KF Koch, KE Merkel, SC O’Connor, JL Occolowitz, JW Paschal, DE Dorman. J Org Chem 43:1430–1434, 1978. MJ Weinstein, GM Leudemann, EM Oden, GH Wagman, JP Rosselet, JA Marquez, CT Coniglio, W Charney, HL Herzog, J Black. J Med Chem 6:463–464, 1963. MJ Weinstein, GM Leudemann, EM Oden, GH Wagman. Antimicrob Agents Chemother 3:1–7, 1964.

348

Haddad et al.

59.

MJ Weinstein, GM Leudemann, RT Testa, GH Wagman, EM Oden, JA Waitz. J Antibiot 23:551–554, 1970. PJL Daniels. Aminoglycoside Antibiotics, Drug Action and Drug Resistance in Bacteria, vol. 2. Baltimore: University Park Press, 1975, pp 77–111. TL Nagabhushan, PJL Daniels, RS Jaret, JB Morton. J Org Chem 40:2835–2836, 1975. RT Testa, B Tilley. Abstract 13th Interscience Conference on Antimicrobial Agents and Chemotherapy, Washington, DC, 1973, no 46. R Okachi, I Kawamoto, S Takasawa, M Yamamoto, S Sato, T Sato, T Nara. J Antibiot 27:793–800, 1974. PJL Daniels, RS Jaret, TL Nagabhushan, WN Turner. J Antibiot 29:488–491, 1976. GH Wagman, JA Marques, JV Bailey, D Cooper, J Weinstein, R Tkach, PJL Daniels. J Chromatogr 70:171–173, 1972. DJ Cooper, RS Jaret, H Reimann. J Chem Soc Chem Commun 285, 1971. MJ Weistein, GH Wagman, JA Marquez, RT Testa, JA Waitz. Antimicrob Agents Chemother 7:246–249, 1975. JA Marquez, GH Wagman, RT Testa, JA Waitz, MJ Weinstein. J Antibiot 29:483–487, 1976. BK Lee, RG Condon, GH Wagman, MJ Weinstein, E Katz. J Antibiot 29:677–684, 1976. M Kugelman, RS Jaret, S Mittelman, W Gau. J Antibiot 31:643–645, 1978. S Satoi, M Awata, N Muto, M Hayashi, H Sagai, M Otani. J Antibiot 36:1–5, 1983. JB McAlpine, AC Sinclair, RS Egan, RL DeVault, RS Stanaszek, M Cirovic, SL Mueller, PC Goodley, RJ Mauritz, NE Wideburg, LA Mitscher, K Shirahata, H Matsushima, S Sato, T Iida, J Antibiot 30:39–49, 1977. DJ Mason, A Dietz, R Smith. Antibiot Chemother 11:118–122, 1961. ME Bergy, TE Eble, RR Herr. Antibiot Chemother 11:661–664, 1961. H Hoeksema, JC Knight. J Antibiot 28:240–241, 1975. S Tatsuoka, T Kusaka, A Miyake, M Inoue, H Hitomi, Y Shiraishi, H Iwasaki, M Imanishi. Pharm Bull 5:343, 1957. RG Benedict, FH Stodola, OL Shotwell, AM Borud, LA Lindenfelser. Science 112: 77–78, 1950. S Hosoya, M Soeda, N Komatsu, Y Sonoda. J Antibiot A3:66–71, 1949. H Heading. Acta Chem Scand 22:1649–1654, 1968. J Fried, E Titus. J Biol Chem 168:391–392, 1947. GF Arcamone, G Cassinelli, G D’Amico, P Orezzi. Experientia 24:441–442, 1968. T Miyaki, H Tsukiura, M Wakae, H Kawaguchi. J Antibiot A15:15–19, 1962. M Awata, N Muto, M Hayashi, S Yaginuma. J Antibiot 39:724–726, 1986. S Tohma, H Kondo, J Yokotsuka, J Iwamoto, G Matsuhashi, T Ito. J Antibiot 42:1205– 1212, 1989. H Umezawa, Y Okami, T Hashimoto, Y Suhara, M Hamada, T Takeuchi. J Antibiot A18:101–103, 1965. J French, Q Bartz, H Dion. J Antibiot 26:272–283, 1973. WJ McGahren, BA Hardy, GO Morton, FM Lovell, NA Perkinson, RT Hargreaves, DB Borders, GA Ellestad. J Org Chem 46:792–799, 1981. T Iwasa, H Yamamoto, M Shibata. J Antibiot 23:595–602, 1970. M Tsunakawa, M Hanada, H Tsukiura, M Konishi, H Kawaguchi. J Antibiot 38:1313– 1321, 1985. T Suami, S Ogawa, N Chida. J Antibiot 33:98–99, 1980. T Iwasa, Y Kameda, M Asai, S Horii, K Mizuno. J Antibiot 24:119–123, 1971. S Horii, Y Kameda, K Kawahara. J Antibiot 25:48–53, 1972. Y Kameda, NA Sano, T Yamaguchi, K Matsui, S Horii, H Fukase. J Antibiot 39:1491– 1494, 1986. N Asano, Y kamede, K Matsui, S Horii, H Fukase. J Antibiot 43:1039–1041, 1990.

60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72.

73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94.

Aminoglycoside Antibiotics 95. 96. 97. 98.

99. 100. 101. 102. 103. 104. 105.

106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123.

349

K Saitoh, M Tsunakawa, K Tomita, T Miyaki, M Konishi, H Kawaguchi. J Antibiot 41:855–861, 1988. H Maehr, JM Smallheer, JF Blount. J Org Chem 46:378–381, 1981. T Nara, M Yamamoto, I Kawamoto, K Takayama, R Okachi, S Takasawa, T Sato, S Sato. J Antibiot 30:533–540, 1977. RS Egan, RS Stanaszek, M Cirovic, SL Mueller, J Tadanier, JR Martin, P Collum, AW Goldstein, RL Deuault, AC Sinclair, EE Fager, LA Mitscher. J Antibiot 30:552–563, 1977. T Ida, M Sato, I Matsubara, Y Mori, K Shirahata. J Antibiot 32:1273–1279, 1979. T Deushi, A Iwasaki, K Kamiya, T Kunieda, T Mizoguchi, M Nakayama, H Itoh, T Mori, T Oda. J Antibiot 32:173–179, 1979. T Deushi, M Nakayama, I Watanabe, T Mori, H Naganawa, H Umezawa. J Antibiot 32:187–192, 1979. Y Okami, K Hotta, M Yoshida, D Ikeda, S Kondo, H Umezawa. J Antibiot 32:964– 966, 1979. T Deushi, A Iwasaki, K Kamiya, T Mizoguchi, M Nakayama, H Itoh, T Mori. J Antibiot 32:1061–1065, 1979. D Ikeda, Y Horiuchi, M Yoshida, T Miyasaka, S Kondo, H Umezawa. Cabohydr Res 109:33–45, 1982. S Inouye, K Ohba, T Shomura, M Komjima, T Tsuruoka, J Yoshida, N Kato, M Ito, S Amano, S Omoto, N Ezaki, T Ito, T Niida, K Watanabe. J Antibiot 32:1354–1356, 1979. S Kondo, Y Horiuchi, D Ikeda, S Gomi, K Hotta, Y Okami, H Umezawa. J Antibiot 35:1104–1106, 1982. S Inouye, T Shomura, K Ohba, H Watanabe, S Omoto, T Tsuruoka, M Kojima, T Niida. J Antibiot 33:510–513, 1980. T Deushi, T Yamaguchi, K Kamuja, A Iwasaki, T Mizoguchi, M Nakayama, I Watanabe, H Itoh, T Mori. J Antibiot 23:1274–1280, 1980. P Kurath, W Rosenbrook, DA Dunnigan, JB MeAlpine, RS Egan, RS Stanaszek, M Cirovic, SL Mueller, WH Washburn. J Antibiot 37:1130–1143, 1984. P Kurath, RS Stanaszek, M Cirovic. J Antibiot 35:1338–1344, 1982. KL Rinehart Jr, T Suami. Aminocyclitol Antibiotics. Washington, DC: American Chemical Society, 1980, pp 309–320. A Iwasaki, T Deushi, I Watanabe, M Okuchi, H Iton, T Mori. J Antibiot 35:517–519, 1982. M Morioka, K Hotta, D Ikeda, H Naganawa, Y Okami. J Antibiot 42:831–833, 1989. S Inouye, T Tsuruoka, T Niida. J Antibiot A19:288–292, 1966. JJ Vavra, C Deboer, A Dietz, LJ Hannka, WT Sokolsi. Antibiot Ann 1959/60:230–235, 1960. T Hata, S Omura, M Katagiri, K Atsumi, J Awaya, S Higashikawa, K Yasui, H Terada, S Kuyama. J Antibiot 24:900–901, 1971. F Arcamone, F Bizioli. Gazz Chim Ital 87:896–902, 1957. M Uramoto, N Otake, H Yonehara. J Antibiot A20:236–237, 1967. H Naganawa, N Usui, T Takita, M Hamada, K Maeda, H Umezawa. J Antibiot 27: 145–146, 1974. T Tsuno, C Ikeda, KI Numata, K Tomita, M Konishi, H Kawaguchi. J Antibiot 39: 1001–1003, 1986. H Tskiura, M Hanada, K Saito, K Fujisawa, T Miyaki, H Koshiyama, H Kawaguchi. J Antibiot 29:1137–1146, 1976. JP Kirby, GE VanLear, GO Morton, WE Gore, WV Curran, DB Borders. J Antibiot 30:344–347, 1977. K Nara, Y Sumino, K Katamoto, S Akiyama, M Asai. Chem Pharm Bull 26:1075– 1082, 1978.

350

Haddad et al.

124.

O Ando, H Satake, K Itoi, A Sato, M Nakajima, S Takahashi, H Haruyama, Y Ohkuma, T Kinoshita, R Enokita. J Antibiot 44:1165–1168, 1991. L Vertesy, HW Fehlhaber, A Shultz. Angew Chem Int Ed Engl 33:1844–1846, 1994. B Matkovic, W Piendl, A Bock. FEMS Microbiol Lett 24:237–276, 1984. J Thompson, PA Skeggs, E Cundliffe. Mol Gen Genet 201:168–173, 1985. AA Beauclerk, E Cundliffe. J Mol Biol 93:661–671, 1987. MH Miller, SC Edberg, LJ Mandel, CF Behar, NH Steigbigel. Antimicrob Agents Chemother 18:722–729, 1980. R Humbert, K Altendorf. J Bacteriol 171:1435–1444, 1989. ML Young, M Bains, A Bell, RE Hancock. Antimicrob Agents Chemother 36:2566– 2568, 1992. KJ Shaw, PN Rather, RS Hare, GH Miller. Microbiol Rev 57:138–163, 1993. GH Miller, FJ Sabatelli, RS Hare, Y Glupczynski, P Mackey, D Shlaes, K Shimizu, KJ Shaw. Clin Infect Dis 24:S46–S62, 1997. GD Wright, AM Berghuis, S Mobashery. In: BP Rosen and S Mobashery, eds. Resolving the Antibiotic Paradox: Progress in Understanding Drug Resistance and Development of New Antibiotics. New York: Plenum Press, 1998, pp 27–69. M-P Mingeot-Leclercq, Y Glupczynski, PM Tulkens. Antimicrob Agents Chemother 43:727–737, 1999. BD Davis. Microbiol Rev 51:341–350, 1987. BD Davis, L Chen, PC Tai. Proc Natl Acad Sci USA 83:6164–6168, 1986. LE Bryan, HM Van den Elzen, MS Shahrabadi. Microb Drug Resist 475–490, 1976. HS Fraimow, JB Greenman, IM Leviton, TJ Dougherty, MH Miller. J Bacteriol 173: 2800–2808, 1991. LE Bryan, S Kwan. Antimicrob Agents Chemother 23:835–845, 1983. T Powers, HF Noller. J Mol Biol 235:156–172, 1994. D Fourmy, MI Recht, SC Blanchard, JD Puglisi. Science 274:1367–1371, 1996. U von Ahsen, R Schroeder. Nucleic Acids Res 19:2261–2265, 1991. U von Ahsen, J Davies, R Schroeder. J Mol Biol 226:935–941, 1992. U von Ahsen, HF Noller. Science 260:1500–1502, 1993. U von Ahsen, J Davies, R Schroeder. Nature 353:368–370, 1991. J Davies. Mol Microbiol 4:1227–1232, 1990. EF Gale, E Cundliffe, PE Reynolds, MH Richmond, MJ Waring. In: The Molecular Basic of Antibiotic Action. London: John Wiley & Sons, 1981, pp 402–547. J Davies, L Gorini, BD Davis. Mol Pharmacol 1:93–106, 1965. J Davies, BD Davis. J Biol Chem 243:3312–3316, 1968. H-Y Mei, AA Galan, NS Halim, DP Mack, DW Moreland, KB Sanders, HN Truong, AW Czarnik. Bioorg Med Chem Lett 5:2755–2760, 1995. H Wank, J Rogers, J Davies, R Schroeder. J Mol Biol 236:1001–1010, 1994. TK Stage, KJ Hertel, OC Uhlenbeck. RNA 1:95–101, 1995. T Hermann, EJ Westhof. J Mol Biol 276:903–912, 1998. B Llano-Sotelo, CS Chow. Bioorg Med Chem Lett 9:213–216, 1999. J Rogers, AH Chang, U von Ahsen, R Schroeder, J Davies. J Mol Biol 259:916–925, 1996. J Woodcock, D Moazed, M Cannon, J Davies, HF Noller. EMBO J 10:3099–3103, 1991. J Davies, DS Jones, HG Khorana. J Mol Biol 18:48–57, 1966. D Fourmy, S Yoshizawa, JD Puglisi. J Mol Biol 277:233–345, 1998. D Fourmy, MI Recht, JD Puglisi. J Mol Biol 277:347–362, 1998. SC Blandchard, D Fourmy, RG Eason, JD Puglisi. Biochemistry 37:7716–7724, 1998. PB Alper, M Hendrix, P Sears, C-H Wong. J Am Chem Soc 120:1965–1978, 1998. C-H Wong, M Hendrix, ES Priestley, WA Greenberg. Chem Biol 5:397–406, 1998.

125. 126. 127. 128. 129. 130. 131. 132. 133. 134.

135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163.

Aminoglycoside Antibiotics 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194. 195. 196. 197. 198. 199. 200. 201. 202. 203. 204. 205. 206. 207. 208.

351

M Hendrix, ES Priestley, GF Joyce, C-H Wong. J Am Chem Soc 119:3641–3648, 1977. M Hendrix, PB Alper, ES Priestley, C-H Wong. Angew Chem Int Ed Engl 36:95–98, 1997. C-H Wong, M Hendrix, DD Manning, C Rosenbohm, WA Greenberg. J Am Chem Soc 120:8319–8327, 1998. JB-H Tok, RR Rando. J Am Chem Soc 120:8279–8280, 1998. DA Mann, J Karn. J Mol Biol 241:193–207, 1994. U Fischer, J Huber, WC Boelens, IW Mattaj, R Luhrmann. Cell 82:475–483, 1995. TJ Daly, RC Doten, JR Rusche, M Auer. J Mol Biol 253:243–258, 1995. J Kjems, BJ Calnan, AD Frankel, PA Sharp. EMBO J 11:1119–1129, 1992. R Tan, L Chen, JA Buettner, D Hudson, AD Frankel. Cell 73:1031–1040, 1993. K Harada, SS Martin, AD Frankel. Nature 380:175–179, 1996. Y Wang, RR Rando. Chem Biol 2:281–290, 1995. Y Wang, J Killian, K Hamasaki, RR Rando. Biochemistry 35:12338–12346, 1996. SM Lato, AR Boles, AD Ellington. Chem Biol 2:291–303, 1995. MG Wallis, U von Ahsen, R Schroeder, M Famulok. Chem Biol 2:543–552, 1995. J Cho, K Hamasaki, RR Rando. Biochemistry 37:4985–4992, 1998. L Jiang, AK Suri, R Fiala, DJ Patel. Chem Biol 4:35–50, 1997. L Jiang, DJ Patel. Nat Struct Biol 5:769–774, 1998. HW Pley, KM Flaherty, DB McKay. Nature 372:68–74, 1994. WG Scott, JT Finch, A Klug. Cell 81:991–1002, 1995. WG Scott, JB Murray, JRP Arnold, BL Stoddard, A Klug. Science 274:2065–2069, 1996. B Clouet-d’Orval, TK Stage, OC Uhlenbeck. Biochemistry 34:11186–11190, 1995. JB Murray, AA Seyhan, NG Walter, JM Burke, WG Scott. Chem Biol 5:587–595, 1998. H Wang, Y Tor. Angew Chem Int Ed Engl 37:109–111, 1998. H Wang, Y Tor. J Am Chem Soc 119:8734–8735, 1997. T Hermann, E Westhof. J Mol Biol 276:903–912, 1998. T Hermann, P Auffinger, E Westhof. Eur Biophys J 27:153–165, 1998. Y Tor, T Hermann, E Westhof. Chem Biol 5:R277–R283, 1998. F Michel, M Hanna, R Green, DP Bartel, JW Szostak. Nature 342:391–395, 1989. TR Cech. Annu Rev Biochem 59:543–568, 1990. BL Bass, TR Cech. Biochemistry 25:3377–4473, 1986. M Yarus. Science 240:1751–1758, 1988. ML Sogin, JC Edman. Nucleic Acids Res 17:5349–5359, 1989. L Sharmeen, MY Kuo, G Dineter-Gottlieb, J Taylor. J Virol 62:2674–2679, 1988. J Marais, JA Rutka. Clin Otolaryngol 23:360–367, 1998. EM Priuska, J Schacht. Biochem Pharmacol 50:1749–1752, 1995. J Schacht. Otolaryngol Head Neck Surg 118:667–674, 1998. B Ryals, E Westbrook, J Schacht. Hear Res 112:44–48, 1997. T Hutchin, G Cortopassi. Antimicrob Agent Chemother 38:2517–2520, 1994. JA Segal, BD Harris, Y Kustova, A Basile, P Skolnick. Brain Res 815:270–277, 1999. JA Segal, P Skolnick. Eur J Pharmacol 347:311–317, 1998. S-H Sha, J Schacht. Neurorep 9:3893–3895, 1998. G Laurent, BK Kishore, PM Tulkens. Biochem Pharmacol 40:2383–2392, 1990. GB Appel. Am J Med 88 (suppl 3C):16S–20S, 1990. M-P Mingeot-Leclercq, R Brasseur, A Schanck. J Toxicol Environ Health 44:263–300, 1995. G Laurent, G Toubeau, JA Heuson-Stiennon, P Tulkens, P Maldague. Crit Rev Toxicol 19:147–183, 1988.

11 Methodologies in Syntheses of Aminoglycoside Antibiotics Jalal Haddad, Mei-Zheng Liu, and Shahriar Mobashery Wayne State University, Detroit, Michigan

I.

INTRODUCTION

Many aminoglycoside antibiotics that are being used as therapeutic agents are semisynthetic products. These compounds show excellent antibiotic properties, yet they suffer from ototoxicity and renal toxicity. Furthermore, bacterial resistance to aminoglycosides is widespread among pathogens, resulting in failure of these antibiotics in the treatment of infectious diseases. As a consequence, these antibiotics have been the targets of structure alteration to expand their spectrum of activity, to limit their susceptibility to resistance enzymes, and to reduce their undesirable toxic properties. Since, however, these compounds are multifunctional molecules, they present challenges in synthetic transformations, which have stimulated the development of chemistry special to them. The decade of 1970s witnessed a burst of activity in the development of synthetic methodology for aminoglycosides [1–4]. There has been renewed interest in these antibiotics because of the new properties that have been shown to be associated with aminoglycosides [5]. Yet another reason is the recent availability of three-dimensional structures for the aminoglycoside resistance enzymes [6] and RNA binding sites for these drugs. To make this chapter a definitive review of the chemistry of aminoglycosides, we discuss only briefly the most important contributions from the earlier literature. The major coverage centers on findings from the early 1980s to the present. II.

SELECTIVE N-PROTECTION AND MODIFICATION OF AMINOGLYCOSIDE ANTIBIOTICS

Aminoglycoside antibiotics that are being used currently for treatment of serious infections by Enterobacteriaceae or Pseudomonas aeruginosa, are mostly semisyn353

354

Haddad et al.

Compounds 1–5

thetic compounds. These have been synthesized from their natural product parent precursors to increase their antibacterial activities and reduce the renal and otovestibular toxicity associated with these drugs [7]. Many of these semisynthetic antibiotics (compounds 1–5 ), such as netilmicin 1 (1-N-ethylsisomicin) [8], amikacin 2 [1-N-(S)-4-amino-2-hydroxybutanoylkanamycin A] [9], and isepamicin 3 [1-N-(S)-3amino-2-hydroxypropanoylgentamicin B] [10] are aminoglycosides modified at position 1 of the 2-deoxystreptamine moiety, prepared by selective protection and modification of the amino groups of the antibiotics. These synthetic targets are challenging because aminoglycosides often bear four to six amino groups with comparable reactivities. To overcome this problem, methods have been developed to take advantage of specificity of the amino groups of aminoglycosides for regioselective chemical modification, due to their varying nucleophilicity. In one method, regioselective modification can be achieved based on the differences in reactivities of the amino groups due to varying basicity or steric environment of these functional groups. The second and more important method, application of transition metal ions in temporary protection of some amino groups by complexation, leads to deactivation of the coordinated amino groups, leaving the others free for further chemical manipulation. A.

Selective Modification of Aminoglycosides Based on Varying Reactivities and Basicities of the Amino Groups

Kawaguchi et al. [9] selectively modified aminoglycosides to prepare amikacin 2, a valuable and potent analog of kanamycin A, with activity against many strains of

Syntheses of Aminoglycoside Antibiotics

355

the Enterobacteriaceae and P. aeruginosa that show resistance to most other aminoglycoside antibiotics [11]. The order of reactivity of the amino groups of kanamycin A (4), toward (benzyloxycarbonyloxy)-succinimide and other carbamoylating reagents is 6⬘-NH2 > 1NH2 > 3-NH2 > 3⬙-NH2. Therefore, the amino group at position 6⬘ was first protected by the reaction of benzyloxycarbonyloxy succinimide in low yield, and subsequently the (S)-4-amino-2-hydroxybutanoyl (ABH) group was introduced to the amino group at position 1 with poor selectivity. Removal of the Cbz group at position 6⬘ furnished amikacin. The same order of reactivity has also been observed for kanamycin B (5), which possesses an extra amino group at position 2⬘ with reactivity comparable to that of 6⬘-amino group (6⬘-NH2 > 2⬘-NH2 > 1-NH2 > 3-NH2 > 3⬙-NH2) [12]. Therefore, the 1-N-ABH derivative of kanamycin B was prepared by selective protection of the 2⬘and 6⬘-amino groups, and the resulting product was acylated by ABH to give the 1N-ABH analog with poor selectivity [12]. A similar sequence of reactions in the hands of several other investigators led in general to poor selectivity and formation of regioisomeric mixtures of products [13,14]. However, this method offers procedures that rely only on varying reactivities and basicities of the amino groups for selective protection, an approach that results in poor selectivity and does not seem to be a reliable means of regioselective modification of aminoglycoside antibiotics. B.

Application of Transition Metal Ions in Regioselective Modification of Aminoglycosides

Aminoglycoside antibiotics have long been known to make complexes with transition metal ions. It was discovered in 1966 that the addition of a solution of ammoniacontaining copper(I or II) salts to these amino sugars can be useful for analytical purposes [15]. In fact, addition of the ammonia solution of copper led to significant changes in optical rotation of the amino sugars bearing vicinal hydroxyl and amino groups [15]. After this early finding, transition metal salts were found to be useful for synthetic purposes when Hanessian et al. [16] took advantage of complexation of Cu(II) ion with the vicinal amino alcohol functions to selectively acylate the uncoordinated amino groups of kanamycin A via temporary chelation of the vicinal aminohydroxyl groups with Cu(II) ions. Soon after these initial results, the same chemistry was applied to other classes of aminoglycoside antibiotics, and a variety of other metal ions (Co2⫹, Ni2⫹, Cd2⫹, etc.) were employed as chelating agents for proximal (vicinal or nonvicinal) aminohydroxyl functions. For instance, sisomicin, gentamicins, and kanamycin A were selectively acylated by various acylating agents in the presence of Co2⫹, Ni2⫹, or Cu2⫹ by Nagabhushan et al. [17,18]. These workers demonstrated that Ni2⫹ and Co2⫹ temporarily mask the amino groups at 1- and 3⬙-positions, leaving the 3- and 6⬘-amino groups free for acylation. The data indicate that chelation of the vicinal (3⬙ and 4⬙), as well as nonvicinal (1 and 2⬙) aminohydroxyl groups, with these divalent ions led to protection of 1- and 3⬙-amino groups, allowing selective acylation of the uncoordinated sites. A similar strategy was used by Daniels and colleagues [19] to prepare 1-NAHB analogs of gentamicin B (6), kanamycin A (4), and 3⬘,4⬘-dideoxykanamycin B through their corresponding 3,6⬘-di- or 3,2⬘,6⬘-tri-N-acyl-protected derivatives. These investigators were able to prepare isepamicin (3) in high yield (60%) via chelation

356

Haddad et al.

of the 1- and 3⬙-amino groups of gentamicin B (6) with cobalt(II) acetate. Acylation of the 3- and 6⬘-amino functions in complex 7 followed by the decomplexation of the coordinated amino groups with hydrogen sulfide gave compound 8, which then was converted to 3 by selective introduction of the aminohydroxypropanoyl group at position 1, and deprotection of the protecting groups (compounds 6–8 ). Among the other divalent transition metals, Zn2⫹ ion proved to be a very efficient chelating agent for work on the preparative scale [20–22]. It was employed by Tsuchiya et al. to prepare amikacin (2) by a sequence of reactions involving temporary protection to the 1- and 3⬙-amino groups of kanamycin A (4) by zinc complexation and acylation of the free amines at the 3- and 6⬘-positions to form the Cbz-protected derivative 9, followed by selective trifluoroacetylation at the 3⬙-position to furnish 10 (compounds 9, 10). The selective acylation at the 3⬙-position was believed to be directed by the neighboring hydroxyl group participation. Finally treatment of 10 with an active ester of (S)-4-(benzyloxycarbonylamino)-2-hydroxybutyric acid, followed by hydrogenolysis with palladium black, gave the product in high yield (>60%). The same procedure was also applied to prepare a 1-N-AHB derivative of kanamycin B [20]. A combination of metal ions can be used in selective acylation of amino groups in consecutive sequence of reactions as is described in the synthesis of 1-C-hydroxymethylkanamycin B [23]. In this synthesis, the amino groups of kanamycin B (5) at positions 3, 2⬘, and 6⬘ were first protected with di-tert-butyldicarbonate in the presence of Zn(II), and then the amino group at the 1-position was selectively protected

Compounds 6–8

Syntheses of Aminoglycoside Antibiotics

357

Compounds 9 and 10

in the presence of Cu(II), leaving the 3⬙-position free for further manipulation. The same combination of transition metals (Cu2⫹ and Zn2⫹) was also employed for regioselective protection of the amino groups in the syntheses of a series of deaminated derivatives of neamine [24]. When neamine (11) was treated with N-butoxycarbonyloxy-5-norbornene-endo-2,3-dicarboxamide (16) in the presence of Zn(II), the diBoc-protected derivative 12 was obtained, which was subsequently protected at position 3 with N-benzyloxycarbonyloxy-5-norbornene-endo-2,3-dicarboxamine (17) (compounds 11–17 ) in the presence of copper(II) to give the neamine derivative 13, having the 2⬘-amino group free. On the other hand, when initial protection of neamine 11 was performed in the presence of Cu(II), the neamine derivative 14 was formed, which was then converted to compound 15 by further manipulations. The hindered reagents 16 and 17 have also shown good selectivity for the seemingly unhindered amino groups in the absence of transition metal ions [25,26]. For instance, when neamine 11 was treated with 16 (1 equiv.), 6⬘-N-protected neamine derivative was exclusively formed in 65% isolated yield and 22% of the unreacted neamine was recovered. In a similar manner kanamycin A (4) was selectively protected at the 6⬘-position to give the monoprotected product in high yield (70%) [26]. Metal-directed protection of amino groups can even be further improved by using a mixture of divalent transition metals. For instance, treatment of kanamycin B (5) with di-tert-butyldicarbonate in the presence of Zn(II) acetate leads to protection at 3-, 2⬘-, and 6⬘-positions (see above), whereas when 5 was subjected to the same reagent in the presence of a mixture of Zn(II) and Cu(II) ions, compound 18 resulted, which is protected only at the 3- and 6⬘-positions. Subsequent treatment of this intermediate with Zn(II) acetate yielded a complex having the 1- and 3⬙-amino groups coordinated, allowing selective introduction of the Cbz group at the 2⬘-position to afford the derivative 19 [27] (compounds 18, 19). Among transition metal ions employed in regioselective modification of aminoglycosides, Zn2⫹ and Cu2⫹ are the most common. They have been used repeatedly

358

Compounds 11–17

Compounds 18 and 19

Haddad et al.

Syntheses of Aminoglycoside Antibiotics

359

in the syntheses of a variety of potent semisynthetic antibiotics, mainly 1-N-acylated derivatives, by several different research groups [23,28–31]. III.

GLYCOSYLATION METHODS

A.

1,2-trans Glycosylation

As known in the general oligosaccharide syntheses [32,33], 1,2-trans glycosylation is more convenient than 1,2-cis glycosylation, and it can be achieved by neighboring group participation of the acyl group protected hydroxyl or amino group at the C2 position of the glycosyl donor. Formation of a cyclic intermediate between the C1 and C2 atoms of the glycosyl donor in the presence of promoters prevents the ␣face attack and leaves the ␤ face open for the approach of the glycosyl acceptor. In a recent synthesis of neomycin analogs [34], the glycosyl donor 21, having the Op-nitrobenzoyl group at position 2, was employed. In this synthesis, the neamine derivative 20 was used as the glycosyl acceptor to couple with 21 in the presence of borane trifluoride etherate as the promoter, to afford the key intermediate ␤glycoside 22 (compounds 20–22 ). A similar strategy was also applied in several other examples of aminoglycoside synthesis [35–37]. B.

1,2-cis Glycosylation

Compared to 1,2-trans glycosylation, it is more difficult to construct 1,2-cis-glycosidic linkage in high selectivity and yield. However, this is what is needed in the syntheses of the majority of aminoglycoside antibiotics. A solution to this problem is found in the popular strategy of the use of glycosyl donors with nonparticipating protective groups as the substituent for the functionality at the position 2. This strategy gives 1,2-cis glycosides selectively due to the anomeric effect. For the amino sugars with the amino group at position 2, four glycosyl donors with various substituents at this position have been developed. The first strategy uses a Schiff base as the amino protective group in glycosylation reaction. Umezawa and colleagues [4] have successfully accomplished the total synthesis of paromamine, neamine, neomycin C, streptomycin, dihydrostreptomycin, and lividomycin B by the use of 4-methoxybenzylimino as the 2-amino substituent. For example, in total synthesis of streptomycin 26, the L-acetylbromoglucosamine derivative 23 was coupled with the acceptor alcohol 24 to afford the ␣-L-glycopyraniside 25 [38,39] (compounds 23–26 ).

Compounds 20–22

360

Haddad et al.

Compounds 23–26

Alternatively, employment of the glycosyl donors bearing the 2,4-dinitrophenylamino substituent at position 2 in the presence of insoluble promoters such as AgOTf and AgClO4, led to the formation of the glycosidic linkage in favor of the 1,2-cis anomer in moderate yields. Thus, in total syntheses of fortimicin B (30) and ribostamycin, glycosylation of the glycosyl chloride 27 with the aminoinositol derivative 28 gave the corresponding ␣-glycoside 29 [40] (compounds 27–30 ).

Compounds 27–30

Syntheses of Aminoglycoside Antibiotics

361

The azido group is another nonparticipating protective group as an amine surrogate in the glycosyl donor counterpart for 1,2-cis glycosylation reaction. ␤-Glycosyl chlorides, but not bromides owing to their instability, are the most commonly used donors in such condensation in the presence of soluble promoters. However, in situ formation of ␤-glycosyl bromides from their corresponding chloride derivatives may be used for these glycosylation reactions. In the total synthesis of neomycin B (34), the 2-azido derivative 31 was condensed with the acceptor alcohol 32 to give the 1,2-cis-glycoside 33 in the presence of mercuric cyanide and mercuric bromide, which was subsequently reduced to afford the 2-amino product [41] (compounds 31–34 ). The final useful method for the 1,2-cis glycosylation reaction involves an addition–elimination–reduction sequence, developed by Lemieux in the 1970s [42– 44]. This strategy was successfully employed in the syntheses of streptomycin and gentamicins. For example, the addition of nitrosyl chloride to the acetylated glucal 35 resulted in the formation of the dimer adduct 36, which underwent reaction with the acceptor alcohol 37 to yield ␣-oximino glycoside intermediate 38. Compound 38 was subsequently reduced by hydrogenation and deprotection to afford gentamicin X2 (39) [43] (compounds 35–39 ). In the case of amino sugars with the hydroxyl group at position 2, the use of the benzyl group as the hydroxyl protective group leads to stereoselective formation of 1,2-cis-glycoside. Thus, the 2-O-benzyl-protected glycosyl bromide 40 was coupled with the alcohol derivative 41 to produce conveniently the ␣-glycosidic product 42 in 44% yield [45] (compounds 40–42 ). Besides the bromo and chloro functions as the anomeric leaving group, several other glycosyl donors have been utilized. For example, ␤-glucopyranosyl fluoride was employed in the total synthesis of apramycin 48 and saccharocin (KA-5685) 49

Compounds 31–34

362

Haddad et al.

Compounds 35–39

[46]. The glycosylation of the fluoride derivatives 43 or 44 with the aprosamine derivative 45 under modified Mukaiyama conditions selectively provided the 1,2cis-glycosides 46 and 47, respectively, in 40 and 24% yields. Also, the thioglycoside 50, activated by methyl trifluoromethanesulfonate, has been used for the preparation of sorbistin analogs by treatment with the glycosyl acceptor 51, to yield mainly ␣glycoside 52 [47] (compounds 43–52 ). It should be noted that the addition of aminocyclitol derivatives to glycals catalyzed by boron trifluoride etherate also affords 1,2-cis-aminoglycosides stereoselectively. Thus, the glycosyl product 54 with an ␣-anomeric linkage was obtained in 65% yield by addition of the alcohol 53 to the glycal 35 [48] (compounds 53, 54). In summary, although a large number of naturally occurring aminoglycoside antibiotics and their derivatives have been synthesized, glycosylation methods in this area have not been well developed. This is a major problem especially in 1,2-cis

Compounds 40–42

Syntheses of Aminoglycoside Antibiotics

363

Compounds 43–52

glycosylation, which often results in a mixture of anomeric products with low yield. In most cases, the modified Koenigs–Knorr condensation is still the most widely used method. IV.

TOTAL SYNTHESIS AND CHEMICAL MODIFICATION OF THE AMINOGLYCOSIDE ANTIBIOTICS

A.

4-Substituted 2-Deoxystreptamine-Containing Aminoglycosides: Paromamine, Neamine, Paromamine, Apramycin, and Related Compounds

1.

Neamine

The quinic acid derived 2,5,6-trideoxystreptamine analogs 61 was used to prepare 5,6-dideoxyneamine (62) biosynthetically from Streptomyces fradiae [49]. The crucial intermediate for synthesis of 61 was the hydroxyketone 55, which is readily available from quinic acid. Treatment of 55 with p-toluenesulfonyl chloride in pyridine gave enone 56, which after catalytic hydrogenation generated the saturated ketone 57. Stereoselective reduction of 57, followed by tosylation, furnished deriv-

364

Haddad et al.

Compounds 53 and 54

ative 58, which underwent cyclohexylidene deprotection and selective tosylation to afford 59. Conversion of 59 to the diazo compound 60, followed by subsequent hydrogenation gave the streptamine derivative 61, which was employed to make the 5,6-dideoxyneamine 62. The intermediate 59 was also used to prepare the 3⬘,5,6trideoxykanamine analog 66 by a glycal coupling procedure; the glycal 63 was used for the glycosylation in the presence of a catalytic amount of boron trifluoride etherate to afford 64 and 65 as a mixture of ␣ and ␤ isomers, respectively, in a 7:3 ratio. The ␣ isomer 64 was converted to 66 by sequential azide treatment, deacetylation, and catalytic reduction (compounds 55–66 ). Periodate oxidation has been used to prepare the sorbistin analog 68 from the appropriately protected neamine derivative 67 [50]. Likewise the acetal analog of 2deoxystreptamine (69) was prepared by selective cleavage of the C3—C4 bond of the neosamine moiety by subsequent oxidation, followed by borohydride reduction and deprotection (compounds 67–69 ). In an attempt to prepare potent analogs of neamine, Sitrin and colleagues [51] have synthesized a series of amino derivatives of neamine including 3⬘- and 4⬘amino analogs, as well as their 3⬘- and 4⬘-epimers. However, all these amino neamine derivatives were found to be less active than the parent compound. Reductive radical elimination (Barton deoxygenation) of the 3⬘,4⬘-xanthate derivative of neamine (70) has been used to prepare the key intermediate 71 (compounds 70, 71) for the synthesis of the 3⬘,4⬘-dideoxyneamine analog gentamine C1a [52]. In this synthesis, a precursor possessing free hydroxyl groups at positions 3⬘ and 4⬘ was treated with a mixture of carbon disulfide, aqueous sodium hydroxide, and methyl iodide in DMSO to afford compound 70. Subsequent reduction of 70 with tributyltin hydride furnished the olefin 71, which after catalytic hydrogenation and deprotection was converted to gentamine C1a. The 2⬘-nitro-2⬘-deaminoneamine 73, together with a 2⬘-nitro derivative of kanamycin B, have been synthesized as the first mechanism-based inactivators for aminoglycoside phosphotransferases [27]. Compound 73 was prepared from an appropriately protected neamine derivative (72) (compounds 72, 73), using trimethylsilyl chloride for protection of 2⬘-amino and hydroxyl groups. Then, selective deprotection of the silyl-protected amino group followed by oxidation of the free amino group with m-chloroperoxybenzoic acid (mCPBA), and deprotection of all protective groups with trifluoroacetic acid gave the title compound 73. Another report describes syntheses of 6⬘-N-substituted amino acid derivatives of neamine with alanine, phenylalanine, and lysine. These neamine analogs were

Syntheses of Aminoglycoside Antibiotics

365

Compounds 55–66

prepared by using either an active ester method under controlled conditions, or a mixed anhydride method by blocking of all functional groups of neamine except the 6⬘-amino function, leaving it free for modification [53]. The same group has also reported the synthesis of 5⬘-epi-neamine by conversion of the C5⬘-aminomethyl moiety to the aldehyde function, followed by reductive amination [54]. Roestamadji et al. [24] synthesized a series of deaminated derivatives of neamine and kanamycin A to investigate the effect of electrostatic interactions in binding ability and activity of substrates bound to the active sites of aminoglycoside-modifying enzymes. The four deaminated analogs of neamine (compounds 74–77 ) were prepared by selective protection–deprotection of the amino groups to get tri-protected derivatives (e.g., compound 78). Reductive deamination of the corresponding free amino group via an intermediary formylated derivative (e.g., compound 79) using the Barton procedure, followed by deprotection of the protecting groups, resulted in the title compounds.

366

Haddad et al.

Compounds 67–69

Compounds 70 and 71

Compounds 72 and 73

The neamine-derived aldehyde 81 has been prepared and used in combinatorial synthesis of a library of neomycin B mimetic aminoglycoside antibiotics, a strategy for the discovery of new inhibitors for binding to the Rev-responsive element of HIV mRNA [55]. The synthesis involves allylation of the selectively protected neamine derivative 80, followed by ozonolysis of the allylic double bond to generate the aldehyde functional group (compounds 80, 81). Recently an azo derivative of neamine was used by the same group to prepare a series of ribostamycin and neomycin analogs [34]. The 4⬘,4⬘-difluoroneamine (compound 82) was recently synthesized from an appropriately protected neamine derivative having the 3⬘-hydroxyl group free for

Syntheses of Aminoglycoside Antibiotics

367

Compounds 74–79

Compounds 80 and 81

Compound 82

functional group manipulation. The presence of the powerful electron-withdrawing fluorine atoms next to the hydroxyl group at the 3⬘-position prevents phosphorylation of this hydroxyl group by aminoglycoside phosphotransferases, which are the cause of clinical resistance to aminoglycoside antibiotics. This compound can be used as a potential inhibitor for aminoglycoside phosphotransferases [APH(3⬘)s] [56]. In a study to explore the mechanism of phosphorylation of aminoglycoside antibiotics by phosphotransferases and to develop potent inhibitors for these en-

368

Haddad et al.

Compounds 83–86

zymes, the conjoint molecules (compounds 83–86 ) were prepared. These compounds incorporate the structure of neamine to that of adenosine via carbon tethers. These molecules would take advantage of the availability of two distinct binding subunits for ATP and aminoglycosides in the active site of the enzymes. Hence, the conjoint molecules would bind the enzymes more effectively than the individual components for entropic reasons [57]. In another synthesis, neamine was linked by a spacer to a known isoquinoline sulfonamide protein kinase inhibitor to make the conjunct molecule (compound 87). A molecule such as 87 should bind the active site of these enzymes with great affinity [58]. 2.

Paromamine

Yoshikawa et al. [59] described the preparation of paromamine 92 from D-glucosamine. They used Pb(OAc)4 for decarboxylation and sodium borohydride for deacetoxylation in the key steps of the synthesis. The D-glucosamine-derived alcohol 88 was oxidized by the Jones reagent to furnish the corresponding 5-carboxyl analogs, which were decarboxylated with Pb(OAc)4 to give compound 89 as a mixture of ␣ and ␤ isomers. Treatment of 89 with CH3NO2 in the presence of 1% NaOMeMeOH afforded the scyllo-nitrocyclitol glycoside 90. After protection of the hydroxyl groups and acetylation at position 2, this compound was reductively deacetoxylated with NaBH4 in EtOH to furnish 91, which was converted to paromamine 92 by subsequent deprotection, Raney nickel reduction, and catalytic hydrogenation (compounds 88–92 ).

Compound 87

Syntheses of Aminoglycoside Antibiotics

369

Compounds 88–92

trans-4-Aminocyclohexanol has been used to prepare the paromamine analog 93 (compounds 93–99 ) and the neamine derivatives 94, 95, and 96. In these syntheses the corresponding 2⬘-amino-␣-glycosides were prepared by the method of Lemieux using the dimeric 2-nitroso ␣-D-hexopyranosyl chlorides 97 and 98 with appropriately protected cyclohexanols to afford 2-oximino ␣-D-hexopyranosides (e.g., 99) [60]. Reductive deprotection in aqueous hydrazine gave the title compounds 93, 94, 95, and 96. However, none of these analogs exhibited minimum inhibitory concentration (MIC) values of less than 200 mg/mL. In another report, paromamine 92 was converted to the pseudodisaccharide 100, which is a constituent of the structure of oxyapramycin 101 (compounds 100, 101). Two-carbon chain elongation was performed by Wittig reaction on a 5⬘-aldehyde derivative followed by osmium tetraoxide hydroxylation, which after lactonization in the basic condition, followed by reduction and deprotection, afforded compound 100 [61]. Paromamine and its 3⬘-deoxy analog (lividamine) have been converted to the corresponding ethenic 6⬘-aldehyde derivatives 102 and 103, respectively. Reductive amination of the aldehydes followed by deprotection yielded the pseudodisaccharides 104, 105, 106, and 107, which showed less in vitro activity than the parent neamine [62] (compounds 102–107 ). 3.

Aspramycin

Abu et al. [63] described syntheses of two apramycin analogs, 6-O-(3-amino-3deoxy-␣-D-glucopyranosyl)- and 5-O-(␤-D-ribofuranosyl)-apramycin (108 and 109, respectively). To prepare these analogs, glycosylation reactions of a suitably protected derivative of apramycin with corresponding glucosyl or furanosyl chloride

370

Haddad et al.

Compounds 93–99

were performed in the presence of mercuric cyanide. Antimicrobial activity results revealed that compound 108 was as active as apramycin against most gram-positive and gram-negative bacteria, and more active than apramycin against strains producing aminoglycoside-modifying enzymes. However, 109 was less active than 108 [63] (compounds 108, 109). Selective acylation of amino groups of apramycin and related aminoglycosides has been achieved in the presence of metal acetates [64], and paromamine has been used to synthesize the 4-O-(aminooctodiosyl)-2-deoxystreptamine (compound 110), a close analog of the pseudodisaccharide present in oxyapramycin [61,65].

Compounds 100 and 101

Syntheses of Aminoglycoside Antibiotics

371

Compounds 102–107

Compounds 108 and 109

Compound 110

Total syntheses of apramycin 48 and saccharosin 49 from neamine 11 have been described [46,66]. The two apramycin analogs 111 [67] and 112 [68] were prepared from an appropriately protected 2-deoxystreptamine derivative; BF3-catalyzed addition of alcohol (e.g., alcohol 113 to glycals 114 and 115) was used for construction of the second and third rings. The 2-deoxystreptamine derivative 113

372

Haddad et al.

Compounds 111–116

was prepared from a suitably protected neamine analog by sequential periodate oxidation and base-catalyzed ␤-elimination. Coupling of 113 with 115 in the presence of BF3 followed by catalytic hydrogenation afforded the precursor 116, which was employed for glycosylation with 115 to make the apramycin analog 111 (compounds 111–116 ). A series of N-acyl derivatives of apramycin, including ␥-amino-␣-hydroxybutyryl (AHB) and glycyl analogs have been prepared by Allen et al. [29], who used different transition metal ions, including Ni2⫹, Cu2⫹, and Zn2⫹, for regioselective acylation of the amino groups. These investigators demonstrated that acylation of the amino group of apramycin at position 2⬘ yields analogs with enhanced uptake, which can effectively penetrate the cell wall and kill bacteria. B.

5-Substituted 2-Deoxystreptamine-Containing Aminoglycoside: Destomycin and Related Compounds

Yoshimura and colleagues [69,70] effected a total synthesis of destomycin C 117. These workers used the pseudodisaccharide 118, which resulted from mannose and deoxystreptamine, for coupling with the destomic acid 119 (6-amino-6-deoxy-L-glycero-D-galacto-heptonic acid) to construct the orthoester linkage present at position 1⬙ of the molecule. Inversion of the configuration at the 4⬘-position was performed by sequential oxidation–reduction of the 4⬘-hydroxyl group in the desilylated derivative in two steps, which after removal of the protecting groups led to the formation of the antibiotic (compounds 117–119 ). C.

4,5-Disubstituted 2-Deoxystreptamine-Containing Aminoglycosides: Neomycins, Paromomycins, Lividomycins, and Related Compounds

1.

Neomycin B

A total synthesis of neomycin B (34) from condensation of neobiosamine and neamine derivatives has been described [71].

Syntheses of Aminoglycoside Antibiotics

373

Compounds 117–119

Diazotization by sodium nitrite in a dilute solution of sulfuric acid has been used to convert amino groups of neomycin B at the 6⬘- and 6⵮-positions to a hydroxyl group to make paromomycin I (120), the 6⵮-deamino-6⵮-hydroxyparomomycin I (121), and the 6⵮-deamino-6⵮-hydroxyneomycin B (122). These compounds were found to be less active than the parent antibiotic, particularly the tetra-amino derivative 121, which was the least active compound, showing about one-tenth the activity of penta-amino derivative 120 or 122 (compounds 120–122 ). These results indicated

Compounds 120–122

374

Haddad et al.

that deletion of the amino group at position 6⵮ reduces the intrinsic activity to a greater extent than the deamination at the 6⬘-position [72]. The 5⬙-hydroxy group of neomycin B was converted to an amino group by treatment of the 5⬙-tosylated derivative with sodium azide, followed by reduction of the azide to the amino group by triphenylphosphine in the presence of an aqueous solution of sodium hydroxide. The resulting 5⬙-amino neomycin B was shown to be a better RNA binder, slowing down the ribozyme cleavage more effectively than the parent compound [73]. A number of neomycin B derivatives have recently been synthesized to probe the role of ring IV (2,6-dideoxy-2,6-diamino-␤-L-idopyranose) of neomycin B in the binding ability of the antibiotic to the ribosomal RNA. To address this role, analogs were synthesized in which the idose ring was replaced with a flexible mono- or diamino tail (123 and 124), or idose derivatives with no amino groups (125) or one amino group (126) were synthesized. The RNA binding data profile of these compounds showed that the idose ring by itself (without amino groups) does not contribute to the affinity or specificity of binding, indicating that the positively charged amino groups presented on the idose ring are necessary for specific binding. Likewise, these amino groups cannot be replaced by those that have flexible linkers [34]. To synthesize 123 and 124, reductive amination of the neamine-derived aldehyde 127 with the corresponding linear amines was employed, and the glycosylation method was used to prepare analogs 125 and 126 from coupling of the glycosyl acceptor 128 with corresponding idose derivative as glycosyl donors (compounds 123–128 ). 2.

Paromomycin

4⬘-Deoxy analogs of paromomycin (120) and paromamine (92) [74], as well as derivatives of paromomycin modified at the 4⬘-, 5⬙-, and 4⵮-positions [75], have been

Compounds 123–128

Syntheses of Aminoglycoside Antibiotics

375

made to examine their antibiotic activities against several sensitive and resistant strains. These include the 4⬘-deoxy-4⬘-epichloro 129, 4⬘-deoxy 130, 4⬘,4⬙-dideoxy 131, 5⬙-amino-4⬘,5⬙-dideoxy-4⬘-epichloro 132, and 5⬙-amino-4⬘,5⬙-dideoxyparomomycin 133, some of which showed good antibacterial activity against Staphylococcus-resistant strains (compounds 129–133 ). Among these semisynthetic paromomycins, compounds 131 and 132 showed activity comparable to the parent antibiotic, but together with analog 129 they were more active against 4⬘-adenyltransferase-producing organisms. Replacement of the 5⬙-hydroxyl with an amino group, to generate the derivatives 132 and 133, resulted in loss of activity. In another communication the same group described the preparation of a series of paromomycin analogs modified at the 3⬘-, 4⬘-, 1-, 6-, and 5⬙-positions. All the new antibiotics (3⬘,4⬘-deoxy 134, 5⬙-amino-3⬘,4⬘,5⬘-trideoxy 135, 3⬘-deoxy-4⬘-epi 136, 5⬙-amino-3⬘,5⬙-dideoxy-4⬘-epi 137, 1-deamino-6,3⬘,4⬘-trideoxy,-1,6-␤-epi 138, and 4⬘-amino-4⬘-deoxyparomomycin 139) (compounds 134–139 ) showed more potency than paromomycin against organisms that produced 3⬘-phosphotransferase and 4⬘-adenyltransferase, and they were more effective than the parent compound in inhibition of the protein synthesis [76]. 3.

Lividomycin B

2⵮-Hydroxy-2⵮-deoxylividomycin B (143) and its 2⬙-regioisomer 144 have been synthesized by condensation of the protected idopyranosyl halide 140 with the pseudotrisaccharide derivative 141 [77]. The antibiotic activities of these compounds were in the order of lividomycin B 142 > 143 > 144, indicating the importance of 2⵮amino group in antibacterial activity. Consequently, to address the effect of the aminoacyl substituent at the 1-amino group in antibacterial activity of compounds 145 and 146, two 1-N-[(S)-4-amino-2-hydroxybutanoyl] derivatives of 143 and 144 were prepared. The acylamide analog of 143 (compound 145) showed enhanced activity

Compounds 129–133

376

Haddad et al.

Compounds 134–139

compared to 143, whereas the acylamido derivative of the 2⬙-regioisomer (146) showed marked diminution in activity [78] (compounds 140–146 ). The same group also synthesized 5⬙-deoxy-5⬙-fluorolividomycin B, using diethylaminosulfur trifluoride (DAST) as the fluorinating agent. The antibacterial activity of this fluorinated compound was reported to be less than that of the parent antibiotic [79]. 4.

Ribostamycin

The 2⬘,3⬘-epimino derivative of ribostamycin (149) was prepared from the corresponding 3⬘-tosyl derivative 148, using sodium hydride for the cyclization step. However, no intrinsic antimicrobial activity was observed for this epimino analog [80] (compounds 147–149 ). A glycosylation approach was used to prepare ribostamycin 147 and its 4⬙-thio analog from neamine and appropriately protected glycosyl donors ribofuranosyl or 4-thioribofuranosyl derivatives [81]. In another synthesis, D-glucosamine and ribose were used to prepare ribostamycin via an aminotrisaccharide intermediate [82]. A number of ribostamycin analogs (150–158 ) have been synthesized by glycosylation of various ribofuranosyl derivatives with the three protected sisamine analogs 159, 160, and 161 (compounds 150–161 ). These ribostamycin analogs showed only modest antibacterial activity, compared to ribostamycin [83]. 5.

Butirosins

A series of deoxy derivatives of butirosin A (162), including the 3⬘-deoxybutirosin A (163), 5⬙-amino-3⬘,5⬙-dideoxybutirosin (164), and 5⬙-amino-4⬘,5⬙-dideoxybutirosin A (165) were prepared by Woo et al. [84] via sequential deoxygenation of the corresponding trifluoromethylsulfonyl derivatives by nucleophilic substitution with sodium salt of thiophenol followed by reduction with sodium in liquid ammonia and

Syntheses of Aminoglycoside Antibiotics

377

Compounds 140–146

catalytic hydrogenation. These deoxy analogs showed improved activities over butirosin and gentamicin against certain phosphotransferase-producing organisms, particularly Pseudomonas aeruginosa and Staphylococcus aureus. A number of 5⬙-amidino and 5⬙-guanidino analogs of butirosin A and its 4⬘deoxy derivative (compounds 165–168 ) were also prepared by another research group. Among these analogs, 166 showed the best activity against most strains tested, but it was similarly more toxic than the 5⬙-hydroxyl derivatives (e.g., butirosin A, 162 and 163) [85] (compounds 162–168 ). D.

4,6-Disubstituted 2-Deoxystreptamine-Containing Aminoglycosides: Kanamycins, Gentamicins, and Related Compounds

1.

Kanamycins

(a) Deoxykanamycins. Kanamycin A (4) has been used to prepare 2⬘-deoxykanamycin A (178), and 2⬘-epi-kanamycin A and B (179 and 180, respectively) by selective 2⬘-de-O-benzoylation and then 2⬘-triflylation [86]. In this synthesis, the hexa-O-benzoyl-protected derivative 169 was selectively deprotected at the 2⬘-position with hydrazine hydrate in pyridine to afford compound 170, which was treated with trifluoromethanesulfonic anhydride in pyridine to furnish the triflyl derivative 171. This compound was separately treated with benzenethioxide, sodium benzoate,

378

Haddad et al.

Compounds 147–149

and sodium azide to prepare compounds 172, 173, and 174, respectively. Treatments of 172 with NaOMe in MeOH and then Raney nickel, 173 with NaOMe in MeOH, and 174 with NaOMe in MeOH, then Pd-C and hydrogen, afforded the kanamycin analogs 175, 176, and 177, respectively, which were subjected to Ba(OH)2 to furnish the title compounds 178–180 (compounds 169–180 ). 3⬘-Deoxykanamycin A, an antibiotic resistant to 3⬘-phosphotransferase-modifying enzymes [(APH)3⬘-Ia’s], has been prepared via two different routes. One route used Barton’s deoxygenation protocol [87], and in another a mixture of 2⬘,3⬘- and 3⬘,4⬘-epoxy intermediates of kanamycin A derivatives resulted in the formation of a single product (3⬘-deoxy derivative) upon reduction with Raney nickel or sodium borohydride [88]. In another synthesis, Iwasawa and colleagues [89] used 3⬘-deoxyamikacin (181), which has shown better antibiotic activity than amikacin (2), to prepare a number of amikacin analogs including 5,3⬘-dideoxy- (182), 5,3⬘,6⬙-trideoxy- (183), and 3⬘,6⬙-dideoxyamikacin (184). The 5-deoxyamikacin 185 was also synthesized from amikacin (2). Among these analogs, 3⬘,6⬙-dideoxyamikacin (184) showed the strongest antibacterial activity compared to amikacin and the other deoxy derivatives (compounds 181–185 ). Syntheses of 2⬘,3⬘-dideoxykanamycin A, 2⬘,3⬘-dideoxyamikacin, and their related derivatives have been described [22]. 2⬘,3⬘-Dideoxykanamycin A was prepared by two different methods. In one, a 2⬘,3⬘-unsaturated intermediate was obtained by a modified version of the Tipson and Cohen method, which proceeds via triflylation

Syntheses of Aminoglycoside Antibiotics

379

Compounds 150–161

of the vicinal hydroxyl groups at the 2⬘- and 3⬘-positions, followed by treatment of the product with sodium iodide in DMF in the absence of zinc dust. In the other method, a 2⬘,3⬘-epoxy derivative was employed and converted to a 2⬘,3⬘-iodohydrin derivative via hydrogen iodide assisted ring opening of the epoxide, which then was mesylated at the 2⬘-position to furnish the 2⬘,3⬘-unsaturated intermediate. The 2⬘,3⬘deoxyamikacin showed slightly enhanced antibacterial activity compared to amikacin. A number of kanamycin analogs, including 3⬘-deoxy-2⬘-epi-kanamycin A [90], 5-deoxy-, 5,4⬘-dideoxy-, 5-deoxy-5-epi-chloro-, and 5,4⬘-dideoxy-5,4⬘-di-epi-chlorokanamycin A [91], as well as 3⬘-deoxykanamycin B (tobramycin) [92], were prepared and tested against several sensitive and resistant organisms. Antibacterial activity data profile revealed that deoxygenation of kanamycin at the 5-position results in the loss of activity and that the 4⬘-epichlorination of kanamycin A results in lack of activity. 5,6⬙-Deoxykanamycin B has also been prepared by reduction of the corresponding dichloro derivative with tributylstannane, showing improved activity against Staphylococcus epidermidis ATCC 12228 and Klebsiella pneumoniae 602. It was slightly more active than kanamycin B against strains of S. aureus Apo-1, K. pneu-

380

Compounds 162–168

Compounds 169–180

Haddad et al.

Syntheses of Aminoglycoside Antibiotics

381

Compounds 181–185

moniae, and E. coli, but showed a marked decrease in activity against P. aeruginosa [93]. Two other analogs of kanamycin A including 3⬘,4⬘-dideoxy and 4⬘-deoxy derivatives were synthesized by Yoneta and colleagues [94]. These compounds were synthesized from the 3⬘,4⬘-epoxy intermediate 186, which was derived from treatment of its 4⬘-O-mesylated precursor with sodium methoxide in methanol. Treatment of 186 with sodium iodide in the presence of sodium acetate and acetic acid afforded the iodohydin 187, which after tosylation at 3⬘-position was converted to the 3⬘-ene derivative 188 under thermal conditions. Removal of the protecting groups of the olefin 188, followed by catalytic hydrogenation, gave the 3⬘,4⬘-dideoxykanamycin A. Similarly catalytic hydrogenation of the iodohydrin 187 followed by treatment with methanolic sodium methoxide and trifluoroacetic acid afforded the 4⬘-deoxykanamycin A analog (compounds 186–188 ). An attempt to make 3⬘,4⬘-dideoxy analogs of kanamycin A and B by mutational biosynthesis in the presence of gentamicin-producing Micromonospora organisms

Compounds 186–188

382

Haddad et al.

yielded combimicins (189), a mixture of aminoglycosides. These compounds are 3⬘,4⬘-dideoxy derivatives of 4⬙-C-methyl and 3⬙-N-methyl kanamycins and are structurally related to gentamicin C1a (190) (compounds 189–190 ). Combimicins have strong antibacterial activity against gram-negative and gram-positive bacteria, and combimicin B2 (R1 = NH2, R2 = H) has shown less ototoxicity than the gentamicin complex [95]. 3⬘,4⬘-Dideoxykanamycin B (dibekacin) 194 was synthesized by Matsuno and coworkers [96] and exhibited good antibacterial activity against both sensitive and resistant bacteria, including P. aeruginosa. The 4⬙,6⬙-O-cyclohexylidene kanamycin B derivative 191 was employed in this synthesis. Treatment of this compound with benzylsulfonyl chloride in pyridine afforded the derivative 192, which was subjected to sodium iodide in DMF to furnish the 3⬘,4⬘-unsaturated derivative 193. Cyclohexylidene deprotection with aqueous acetic acid (80%) followed by reductive desulfonylation of benzylsulfonyl groups under the Birch condition and catalytic hydrogenation of the double bond gave the 3⬘,4⬘-dideoxykanamycin B (194) in high yield. A multistep synthesis of 194 also was reported by Yoshikawa et al. [97], who used D-glucosamine to construct rings I and II, as well as D-glucose to assemble the idose moiety (ring III) of the antibiotic (compounds 191–194 ). Amino groups play an important role in antibacterial activity of aminoglycoside antibiotics. To probe this role, 3⬘,4⬘-dideoxykanamycin B (dibekacin: 194) was used to prepare hexa- and heptadeoxy derivatives (5,2⬘,3⬘,4⬘,4⬙,6⬙ and 5,2⬘,3⬘,4⬘,2⬙,4⬙,6⬙) of kanamycin A (4) and amikacin (2). The antibacterial activity data profile of these antibiotics revealed that the amino groups are essential for antibacterial activity, and that the 2⬙-hydroxy group and 1-N-(S)-4-amino-2-hydroxybutanoyl (AHB) moiety markedly increase the activity of the antibiotic [98]. 3⬘,4⬘,2⬙-Trideoxy- and 2⬙-deoxykanamycin B have also been synthesized, the former being more active than kanamycin B against resistant strains of P. aeruginosa [99]. Another report, by Albert and colleagues [100], describes the synthesis of 4⬙deoxykanamycin A together with 4⬙-epi-chloro-, 4⬙-epi-bromo-, and 4⬙-epi-iodokanamycin A. (b) epi-Kanamycins. The 4⬘-deoxy-5⬘-epi derivative of kanamycin B (197) was made from its 1,4-epoxy intermediate 195 using the sequences of reactions outlined

Compounds 189 and 190

Syntheses of Aminoglycoside Antibiotics

383

Compounds 191–194

below to get the unsaturated derivative 196 (compounds 195–197 ). Reduction of the double bond followed by deprotection afforded the product. However, the resulting compound showed weak antibacterial activity [101]. 1-epi-Kanamycin A together with 1- and 3-epi-tobramycin have been synthesized by either catalytic hydrogenation over Raney nickel or electrochemical reduction of their corresponding 1- or 3-hydroxyimino derivatives. 1-epi-Kanamycin A showed lower activity than kanamycin A against most bacteria tested, but it was more active against P. aeruginosa. 1-epi-Tobramycin was slightly less active than the parent antibiotic, whereas the 3-epi derivative showed marked decrease in activity against most sensitive and resistant bacteria [102]. 1-epi-Kanamycin A (200) and its 1-N-acylamino derivative 201 were prepared by Takahashi et al. [103] to investigate the effect of C1 epimerization on antibiotic activity of kanamycin A. In this synthesis, oxidation of the kanamycin derivative 198 with hydrogen peroxide in the presence of sodium tungstate afforded the 1hydroxyimino derivative 199, which after deprotection and reduction with hydrogen– Raney nickel in aqueous ammonia furnished 1-epi-kanamycin A (200). This compound was converted to the 1-N-[(S)-4-amino-2-hydroxybutyryl] analog by regiospecific acylation in the presence of zinc acetate. Antibacterial activity results revealed that epimerization of the amino group at position 1 greatly reduces the original activity of the antibiotic (compounds 198–201 ). Several other kanamycin A analogs, including 4⬙-epi-kanamycin A together with 4⬙,5⬙-unsaturated, and 3⬙,4⬙-cyclic urethane by products [104], 4⬙-deoxy-4⬙,5⬙didehydro, and 4⬙-deoxy-5⬙-epi-kanamycin A [105], as well as a number of 4⬙-epianalogs (OH, F, N3) of amikacin [106] have also been prepared, all of which are much less active than the parent antibiotics.

384

Haddad et al.

Compounds 195–197

(c) Halokanamycins. In an attempt to make more active antibiotics, Umemura and colleagues [107] synthesized halogenated derivatives of amikacin, including 6⬙chloro-6⬙-deoxyamikacin (compound 202), 6⬙-iodo-6⬙-deoxyamikacin (compound 203), and some other analogs, including compounds 204 and 205. Among these, the 6⬙-chloro-3⬘,6⬙-dideoxyamikacin showed the best activity, and the study concluded that substitution of the 6⬙-hydroxyl group of amikacin with chlorine enhances the antibacterial activity of the antibiotic against all bacteria except P. aeruginosa. On the other hand, it has been demonstrated that introduction of fluorine at the 6⬙position of amikacin lowers the activity to some extent [106]. Several other derivatives including 5-fluoro-, 6⬙-fluoro-, and 5,6⬙-difluorokanamycin A were prepared by Albert and coworkers [108]. In another synthesis by these investigators, 4⬙-bromo-, 4⬙-fluoro-, and 4⬙-aminokanamycin A were prepared by double inversion substitution reaction using the 4⬙-O-triflate 206 as the key syn-

Compounds 198–201

Syntheses of Aminoglycoside Antibiotics

385

Compounds 202–205

thetic intermediate. This intermediate was first treated with sodium nitrite in DMF to give the corresponding 4⬙-epi-hydroxy derivative 207, which was subjected to triflic anhydride to furnish the 4⬙-epi-O-triflate 208. This compound was then converted to the corresponding bromo, fluoro, and azido derivatives 209–211 (compounds 206–211 ) by another nucleophilic displacement, using tetrabutylammonium bromide and fluoride in acetonitrile or sodium azide in DMF. Antibacterial activity screening showed no dramatic changes in activity for these compounds compare to those of kanamycin A [109]. The same group has also synthesized a number of 4⬙-epi-kanamycin A analogs, including 4⬙,6⬙-dideoxy-4⬙,6⬙-difluoro-, 4⬙,6⬙-dideoxy-4⬙-fluoro-, 6⬙-deoxy-6⬙-fluoro-,

Compounds 206–211

386

Haddad et al.

and 6⬙-deoxy-4⬙-epi-kanamycin A, using a 4⬙,6⬙-ditriflyl derivative as the key intermediate. These compounds showed comparable or better activity for some strains tested [110]. 3⬘-Deoxy-3⬘-fluorokanamycins A and B have been synthesized and showed strong activities against both sensitive and resistant bacteria [111]. The 3⬘-deoxy-3⬘fluorokanamycin A (214), was prepared by condensation of the 3-fluoro-␣-glucopyranosyl derivative 212 with the appropriately protected compound 213 in the presence of mercury(II) cyanide followed by deprotection and hydrogenolysis (compounds 212–214 ). The same group also has described the syntheses of 6⬙-deoxy-6⬙-fluorokanamycin A, 6⬙-deoxy-6⬙-fluoroamikacin, and 1-N-[(R)- and (RS)-3-amino-2-fluoropropanoyl] kanamycin A, using the DAST reagent for fluorination of the corresponding hydroxyl groups of kanamycin A and amikacin derivatives [112]. The 2⬘,3⬘-dideoxy-2⬘-fluorokanamycin A 217 and its 1⬘-epimer have been prepared by coupling reaction of the fluoro sugar 215 and the pseudodisaccharide 216. The resulting analog 217 was only slightly less active than 3⬘-deoxykanamycin A, whereas the 1⬘-epimer was completely inactive [113] (compounds 215–217 ). A series of halogenated analogs of kanamycins, including 3⬘,4⬘,6⬘-trideoxy-6⬘fluorokanamycin C (218), 3⬘,4⬘-dideoxy-6⬘-C-fluoromethyl kanamycin B [114], 6⬙chloro-6⬙-deoxyamikacin (202) and some of its derivatives [115], as well as 3⬘deoxy-3⬘-fluorokanamycin A (219) and its 3⬘,4⬘-dideoxy-3⬘-fluoro analog 220 [45] were also prepared by the same research group. Antibacterial activity results showed weak activity for 6⬘-fluoro, moderate activity for 6⬘-fluoromethyl, and good activity (better than the parent antibiotics) for 3⬘-fluoro derivatives. Several other halogenated derivatives of kanamycins were synthesized by the Umezawa group. These include 3⬘-deoxy-3⬘-fluoro (219) and 3⬘-chloro-3⬘-deoxykanamycin A (221) [116], 4⬘-deoxy-4⬘-fluorokanamycin A and B (222 and 223, respectively) [117], 5-deoxy-5-fluoro-(224), 5,3⬘-dideoxy-5-fluoro-(225), 5,3⬘,4⬘-trideoxy5-fluoro-(226), and 5,5-difluoro-(227) derivatives of kanamycin B [118]. The antibacterial activity of the 3⬘-chloro analog 221 was one-sixth that of the 3⬘-fluoro derivative 219. Also the 4⬘-fluoro derivatives 222 and 223 were inactive against resistant bacteria producing phosphotransferase enzymes [APH(3⬘)s], slightly less active than the parent compounds against common bacteria, but very active against the kind that produce [AAD(4⬘)]. The 5-fluoro and 5,5-difluoro derivatives (com-

Compounds 212–214

Syntheses of Aminoglycoside Antibiotics

387

Compounds 215–217

pounds 224–227 ) showed comparable or better activities and were markedly less toxic than the parent compounds (compounds 218–229 ). 3⬘-Deoxy-3⬘-fluoro (228) and 3⬘,4⬘-dideoxy-3⬘-fluorokanamycin B (229) [119], 5-deoxy-5-fluoro (230), and 5-deoxy-5,5-difluoro (231) derivatives of netilmicin (1) [120], as well as fluoroarbekacins (234–238 ) [121], were prepared by Umezawa’s group. Both fluorinated derivatives of kanamycin B were active against resistant bacteria producing 3⬘-phosphotransferase enzymes [119]. In the synthesis of fluorinated netilmicins (230 and 231), two other analogs, 5-epi-netilmicin (232) and 5epi-6⬘-N-methylnetilmicin (233) were also prepared. Compounds 230 and 232 showed activity similar to that of netilmicin, 231 was slightly less active than netilmicin, but 233 showed markedly reduced activity. In terms of toxicity, the 5-fluoro derivative 230 showed half to one-third the acute toxicity of netilmicin [120]. The fluorinated arbekacins (234–238 ) exhibited activity similar to that of arbekacin and were remarkably less toxic than the parent compound (compounds 230–238 ). These observations were suggestive of the fact that the presence of fluorine at the 5-position decreases the basicity of the 3-amino group and lowers the toxicity of the antibiotic [121]. A few other fluorinated kanamycins, including 5-deoxy-5-epi-fluoroamikacin, 5-deoxy-5-epi-fluoroarbekacin, and their related analogs, have been prepared to study the fluorination–toxicity relationship. In contrast to the low toxicities of the 5-fluoro derivatives [121], these epi-fluoro compounds showed acute toxicity values identical to those of arbekacin and amikacin [122]. This indicates the importance of stereoelectronic effects of the fluoro group at position 5 of the 2-deoxystreptomine moiety in toxicity of aminoglycoside antibiotics. The 3⬘-oxokanamycin A (239) has been recently synthesized. Compound 239, which adopts a hydrate chair conformation in aqueous solution, was shown to be an aminoglycoside antibiotic that can undergo a self and continuous phosphorylation– dephosphorylation process to generate the active antibiotic [123].

388

Compounds 218–229

Compounds 230–238

Haddad et al.

Syntheses of Aminoglycoside Antibiotics

389

Compound 239

(d) N-Acylkanamycins. Umezawa and coworkers [124] further modified several analogs of 5,3⬘,4⬘-trideoxykanamycin B (240), including chloro, deoxy, and N-methyl compounds (241–245 ), to 1-N-[(S)-4-amino-2-hydroxybutyryl] derivatives (246– 251) (compounds 240–251 ). Antibacterial activity results were the best activity for compound 250 and were improved for 6⬙-chlorinated compounds against gram-positive bacteria, with a concomitant increase in toxicity, compared to the related 6⬙hydroxy derivatives, and marked increase in the activity by 1-N-acylation [124]. The same group has also reported the synthesis of 1-N-(2-aminoethoxycarbonyl) and 1-N-(3-aminopropoxycarbonyl) kanamycin A, the latter being less active than amikacin (2). However, the former showed activity similar to that of amikacin, implying that the (S)-4-aminohydroxybutyryl residue of amikacin can be replaced

Compounds 240–251

390

Haddad et al.

by a simple residue such as aminoethoxycarbonyl and retain its activity [125]. Reactions reported include enzymatic O-phosphorylation of kanamycin A and N-acetylation of tobramycin [126], regiospecific synthesis of 1-N-[(S)-4-amino-2-hydroxybutyryl]-2⬙-deoxykanamycin B from neamine [127], and preparation of 1-N-(2-aminoethanesulfonyl) derivatives of kanamycin A and B, ribostamycin, and dibekacin [128]. In contrast to earlier observations for 2⬙-deoxyaminoglycosides, the 1-N-acylamido-2⬙-deoxy derivative of kanamycin B showed loss of activity relative to amikacin [127]. All the aminoethanesulfonyl derivatives showed strong activity against both gram-positive and gram-negative bacteria, suggesting that the (S)-4amino-2-hydroxybutanoyl group can be replaced by a simple group such as aminoethanesulfonyl moiety and still be an effective antibiotic [128]. In another report, kanamycin A was biologically converted to amikacin by a mutant strain of the butirosin-producing organism Bacillus circulans [129]. 6⬘-N-Formimidoylamikacin (253), 6⬘-N-acetimidoylamikacin (254), and 6⬘-Nformimidoyldibekacin (256) have been synthesized from the appropriately protected amikacin and dibekacin derivatives 252 and 255, respectively (compounds 252–256 ). Compared to the parent antibiotics, all these compounds showed improved activity against resistant strains producing AAC(6⬘) enzyme; but in general, the 6⬘-N-amidination slightly decreased the antibacterial activity of kanamycins [21]. Another report described the synthesis of N-guanyl derivatives (at positions 6⬘, 2⬘, 3⬘) of kanamycin A and gentamicins C1, C1a, and C2 [130]. Likewise, a series of 3⬙-N-trifluoroacetyl derivatives of kanamycin A, possessing C5–C22 acyl groups, or related alkoxy, amino, and thiocarbonyl substituents at the C1 amino group have been synthesized and tested for their activity against herpes simplex virus type I (HSV-I) and influenza virus. The optimum antiviral activity was found for the C14– C16 acyl derivatives [131,132]. The same group also has prepared N-palmitoylkanamycin A derivatives modified at the 1-, 3-, 6⬘-, or 3⬙-amino groups having a 3⬙-

Compounds 252–256

Syntheses of Aminoglycoside Antibiotics

391

amino group free or acylated with trifluoroacetyl group. The antiviral activities of these compounds were excellent and almost similar to each other, but those with trifluoroacetyl group at the 3⬙-amino group exhibited reduced cytotoxicity on Vero cell [133]. Likewise, N-palmitoyl derivatives of kanamycin B, tobramycin (3⬘-deoxykanamycin B), dibekacin (3⬘,4⬘-dideoxykanamycin B), and gentamicin (C1, C1a, and C2 complex), and a few other aminoglycosides were prepared, all of which showed excellent antiviral activity [134]. Aminoglycoside antibiotics (257), including kanamycin A, tobramycin, and gentamicin C complex, have been acylated with 1 mole of disodium carbenicillin (258) to give monoacylated derivatives (e.g., 259) (compounds 257–259 ). The resulting compounds were devoid of antibacterial activity against several strains tested, however, and showed no toxicity [135]. The 1-N-(D-threo- and racemic erythro-3-amino-2-hydroxylbutanoyl)-2⬘,3⬘-dideoxykanamycins A have been synthesized by using a standard esterification procedure to acylate the amino group of kanamycin A at position 1 with the corresponding 3-azido-2-hydroxybutanoic acid. The antibacterial activities of these compounds were less than that of amikacin having a 4-amino-2-hydroxybutanoyl moiety at the 1-amino group [136]. In a study on the fluorination–toxicity relationship, investigators prepared 1N-acylamido derivatives of kanamycins, including (2R,3R)- and (2R,3S)-4-amino-3fluoro-2-hydroxybutanoylkanamycins A and B (compounds 262–265 ). The 1-N[(2R,3R)-4-amino-3-fluoro-2-hydroxybutanoyl] derivatives 262–264 showed activity similar to that of the related compounds [amikacin (2), arbekacin (260), and 1-N[(S)-4-amino-2-hydroxybutanoyl]-3⬘-deoxykanamycin B 261], whereas the (2R,3S)derivative 265 (compounds 260–265 ) showed decreased activity relative to 260. In toxicity, these compounds were similar to the parent antibiotics, and fluorination did not show any influence in the toxicity of the antibiotics [137]. 6⬘-N-Benzamido derivatives of kanamycin A and tobramycin having a chromium complex in the benzene ring were prepared for use as metallotracers in immunoassay studies. To make these compounds, N-hydroxysuccinimide-activated esters of these organochromiums were treated with kanamycin A or tobramycin to selectively produce the 6⬘-N-acylated derivative [138]. Different amino acid and peptide derivaties were used to synthesize a series of aminoacyl and dipeptidyl derivatives of kanamycin A and netilmicin modified at the 1-, and 6⬘-amino groups by selective acylation of the copper complexes of these antibiotics. However, none of these modified compounds were active against either gram-positive or gram-negative bacteria [139]. Recently, in a report from Abbott

Compounds 257–259

392

Haddad et al.

Compounds 260–265

Laboratories, impurities of bulk amikacin were identified as 1,6⬘- and 1,3⬙-di-N-[(S)4-amino-2-hydroxybutyryl] kanamycin A [140]. In an attempt to probe structure–toxicity relationships of aminoglycoside antibiotics via fluorination, the fluoroacyl derivatives 1-N-[(2S,4S)- and (2S,4R)-5amino-4-fluoro-2-hydroxypentanoyl]dibekacin (266 and 267, respectively) were synthesized. These fluorinated analogs showed antibacterial activity almost similar to that of amikacin, but they were less toxic. Toxicities of these compounds were compared with those of arbekacin analogs 268–270 , and the investigators concluded that the observed decrease in toxicity is a function of the chain length rather than of the introduction of fluorine [31]. The fluorinated derivatives 266 and 267 showed toxicities analogous to 269, but lower than arbekacin 260 (compounds 266–270 ). The 3⬙-N-acetylation of arbekacin 260 and amikacin 2, the first example of 3⬙N-acetylation of aminoglycoside antibiotics by enzymatic modification, was reported in 1988. Subjection of kanamycin group antibiotics to an aminoglycoside 3-N-ace-

Compounds 266–270

Syntheses of Aminoglycoside Antibiotics

393

tyltransferase, AAC(3)-X, resulted in formation of 3⬙-N-acetyl derivatives of arbekacin and amikacin, but 3-N-acetyldibekacin was formed in the case of dibekacin 194, which lacks the (S)-4-amino-2-hydroxybutyryl moiety in its structure, showing the importance of this side chain in the enzymatic modification of kanamycins. Interestingly, 3⬙-N-acetylarbekacin showed antibacterial activity as high as that of 2⬘-Nacetylarbekacin, whereas the 3⬙-N-acetylamikacin exhibited no substantial activity. These results illuminate a novel aspect of arbekacin chemistry, distinct from the other aminoglycoside antibiotics [141]. (e) N-Hydroxy and C-Hydroxymethylkanamycins. Tsuchiya et al. prepared [142] 6⬘-N-hydroxykanamycin A and 6⬘-N-hydroxydibekacin by reduction of the corresponding 6⬘-aldoximes with sodium cyanoborohydride. However, these compounds showed very weak antibacterial activity. In other reports and in extensive work by van Schepdael et al. [143,144] the 6⬙-triisopropylbenzenesulfonyl ester derivative of kanamycin B (compound 271) was used to prepare a series of 6⬙-substituted kanamycin B analogs. In these compounds, the 6⬙-hydroxyl group was replaced with Cl, N3, NH2, NHMe, NMe2, alkylamido, thioalkyl, alkoxy, and a few other substituents [143]. Likewise, a number of analogs having the 1-position modified by the hydroxymethyl group axial to the 1-amino group were prepared [144]. These include the 6⬙chloro, 6⬙-amino, and 6⬙-acetamido derivatives 280, 281, and 282, respectively. To introduce the hydoxymethyl function at position 1, the Cbz-protected derivative 272 was deprotected by hydrogenolysis, and the resultant free amino group was oxidized to a nitro group with m-chloroperbenzoic acid to afford compound 273. This intermediate was subjected to paraformaldehyde in the presence of triethylamine to furnish the hydroxymethyl analog 274, which after hydrogenolysis in the presence of Raney nickel was converted to the amino derivative 275. Amino group protection followed by nucleophilic displacement at the 6⬙-position with lithium chloride or azide in anhydrous DMF afforded the chloro and azido derivative 276 and 277, respectively. Reduction of the azide function with triphenylphosphine in anhydrous pyridine followed by treatment with aqueous ammonia gave the amino derivative 278, which after acetylation with acetic anhydride afforded the acetamido analog 279. Removal of the protecting groups in 276, 277, and 279 with trifluoroacetic acid resulted in the formation of the title compounds 280, 281, and 282 (compounds 271– 282). However, these new antibiotics showed activity comparable to that of kanamycin B and were only slightly less toxic than the parent compound [145]. ( f ) Amino Kanamycins. 2⬙-Amino-2⬙-deoxyarbekacin (283), together with five related analogs (284–288 ) were synthesized by reductive amination from their corresponding 2⬙-oxo derivatives. For example, oxidation of the 2⬙-hydroxyl group of 289 by the Pfitzner–Moffatt procedure, followed by reductive amination, gave exclusively an equatorial 2⬙-amino group, which after protection with the Boc group afforded compound 290. Treatment of the Cbz-deprotected derivative 291 with N-hydroxysuccinimide ester of (S)-2-hydroxy-4-(p-methoxybenzyloxycarbonylamino)butyric acid in THF mainly gave the 1-N-acylated derivative, which after deprotection furnished the title compound 283 (compounds 283–291 ). All the new antibiotics showed good antibacterial activities against staphylococci and gram-negative bacteria with improved anti-MRSA activity. Among these analogs, compounds 283 and 286 exhibited excellent anti-MRSA activity and were less toxic than the parent antibiotic [30,146].

394

Haddad et al.

Compounds 271–282

A series of 2⬙-amino-2⬙-deoxy and 2⬙-acylamido-2⬙-deoxy derivatives of 5deoxy-5-epi-fluorodibekacin (compounds 292–298 ) were also prepared via an oxidation–methoxyimination–reduction sequence to introduce the 2⬙-NH2 group. All the new compounds except 292 showed only limited or no antibacterial activity, and the epi-amino derivative 293 exhibited much lower activity than 292 [147] (compounds 292–298 ). In another synthesis, Wang et al. [73] prepared 6⬙-amino-6⬙-deoxykanamycin A, together with 6⬙-amino-6⬙-deoxytobramycin and 5⬙-amino-5⬙-deoxyneomycin B, by nucleophilic displacement of an azido group with a tosyl group in the O-tosylated derivatives. These amino-aminoglycosides were tested for their binding ability to RNA and showed improved binding ability, being much stronger RNA binders than their corresponding parent counterparts. (g) Deamino Kanamycins. A series of neamine and kanamycin A derivatives involving amino group deletions have been synthesized to investigate the importance of electrostatic interactions for both substrate recognition and enzymatic modification by resistant enzymes. The three deaminated kanamycin A analogs compounds 299– 301, together with those of neamine (see Section IV.A.1: compounds 74–77 ) were

Syntheses of Aminoglycoside Antibiotics

395

Compounds 283–291

tested against resistant organisms harboring APH(3⬘)-Ia and APH(3⬘)-IIa, and the same organisms without the resistance enzymes. The resulting data showed virtually unmodified products in the face of these enzymes, indicating the importance of electrostatic interactions for enzymatic modification [24]. (h) Other Kanamycin Derivatives. Cyclic carbamate derivatives of kanamycin A have been prepared [148], and in another report regiospecific methylation and stereospecific hydrogenation were used to synthesize three kanamycin B analogs with the general formula of compound 302 [149]. The same group [150] has also reported conversion of the neosamine ring of kanamycin B to the bicyclo-oxazoctene unit compound 303 (part structure). Metal acetates were used for selective protection of the amino groups in the preparation of 2⬘-deamino-2⬘-nitrokanamycin B together with 2⬘-nitroneamine analog

396

Compounds 292–298

Compounds 299–301

Compounds 302 and 303

Haddad et al.

Syntheses of Aminoglycoside Antibiotics

397

(see Section IV.A.1: compound 73). These compounds are the first mechanism-based inactivators for bacterial 3⬘-phosphotransfer enzymes [APH(3⬘)s] [27]. Synthesis of kanamycin A analogs having 6-amino-3-oxo-2,3,4,6-tetradeoxy-Dand L-glycero-hexopyranose in their structures (e.g., 306) were described by Kuwahara and Tsuchiya [151]. To make compound 306, the 1-thioglycoside 304 was condensed with appropriately protected disaccharide 305, using N-iodosuccinimide (NIS) in a slightly acidic solution (compounds 304–306 ). This compound was only slightly active against common bacterial strains, and the other isomers were devoid of antibacterial activity. Synthesis of 2⬙-modified dibekacin and arbekacin derivatives including 5deoxy-5,2⬙-di-epi-5-fluorodibekacin 310 and their 2⬙-oxo analogs (compounds 311 and 312, as hydrated ketones) also was reported by the same group [152]. To make compound 310, the intermediate 307 was fluorinated by DAST at position 5 of the 2-deoxystreptamine moiety and converted to fluoro derivative 308 after deacetylation of the acetyl group at the 2⬙-position by 1.0 M methanolic sodium methoxide. Triflylation of the 2⬙-hydroxyl group followed by treatment of the resulting triflate with sodium acetate in DMF afforded the 2⬙,3⬙-cyclic carbamate 309 with inversion of configuration at 2⬙-position. Sequential removal of the protecting groups led to the formation of the title compound 310. The oxo compound 311 was prepared from the intermediate 308 by pyridinium dichromate (PDC) oxidation, followed by deprotection with trifluoroacetic acid. The same synthetic route was also employed to prepare the 2⬙-oxoarbekacin 312 (compounds 307–312 ). The 2⬙-oxo derivatives 311 and 312 were collected as the hydrated form after removal of the protecting groups and upon treatment with water. All three analogs were less active than the parent antibiotics, indicating that the equatorial 2⬙-hydroxyl group is essential for antibacterial activity. In a study of ribozyme inhibitory effect, Wang and Tor [153] prepared covalently linked dimeric aminoglycosides derived from kanamycin A, tobramycin, and neomycin B. The resulting symmetrical or asymmetrical dimeric aminoglycosides showed enhanced affinity in binding to RNA in comparison with their monomeric counterparts. In another investigation, Sreedhara and Cowan [154] prepared a variety of metallokanamycin A compounds and tested them for their ability to cleave DNA at physiological pH. Only a mixture of Cu2⫹ and kanamycin A was found to be efficient in effecting the rapid degradation of plasmid DNA at 1 mM concentration. To improve the antibacterial activity of arbekacin, addition salts of this antibiotic with (⫺)-cis-1,2-epoxypropylsulfonic acid (compound 313) were prepared by making a hemisalt of 313 with (⫹)-␣-phenylethylamine, followed by treatment with

Compounds 304–306

398

Haddad et al.

Compounds 307–312

arbekacin free base. Some of these compounds showed MIC as low as 0.10 mg/mL against S. aureus [155]. 2.

Gentamicins

An extensive report by a group from the Schering-Plough Corp., describes the synthesis of a series of hexopyranosyl, hexofuranosyl, pentopyranosyl, and pentofuranosyl derivatives of gentamicin C1 and C1a with general formula of 314 [156,157]. The same group also prepared several 1- or 3-substituted and 1- or 3-epi-substituted

Compound 313

Syntheses of Aminoglycoside Antibiotics

399

analogs of sisomicin and gentamicin (e.g., 315–324 ). These compounds were synthesized from the corresponding 1- and 3-oxo derivatives by suitable reductive techniques, and the oxo compounds themselves were prepared from selectively protected gentamicin and sisomicin derivatives by means of the Corey procedure, using 3,5di-tert-butyl-1,2-benzoquinone for deamination–oxidation. The 1-hydroxy, 1-epi-hydroxy, 1-alkylamino, and 1-epi-alkylamino derivatives (compounds 316, 317, 320, and 321 respectively) were all highly potent antibiotics, and epimerization at the 1position has not significant effect on biological activity. On the other hand, the 3hydroxy, 3-epi-hydroxy, and 3-epi-amino derivatives were essentially devoid of any antibacterial activity [158]. Likewise, a number of 1-N-carbonyl derivatives of gentamicins, sisomicin, and kanamycin A, including amino and thio analogs with the general formula (partial structure) of 325 (compounds 314–325 ), have been synthesized. Some of these, particularly the 1-N-aminoalkoxycarbonyl and 1-N-aminoalkylcarboxamido analogs, exhibited highly potent antibacterial activity [159]. In another report, by Paulson and Jansen [160] described synthesis of the sisomicin analog 328 by glycosylation of the suitably protected disaccharide 326 with furanosyl bromide 327 in the presence of mercury cyanide. Likewise, gentamicin X2 analogs (compounds 329–331 ) were prepared by glycosylation of an appropriate 2azido-glycosyl chloride with a garosaminyl-deoxystreptamine derivative. The antibacterial activity screening showed no activity for 330 and 331 [161]. The same group has also reported the synthesis of the tetra-N-acetylsisamine 333 (compounds

Compounds 314–325

400

Haddad et al.

326–333 ); the mechanism was selective degradation of the 1,2⬘,3,6⬘-tetra-N-acetylsisomicin (332) by periodate oxidation [162], and it was used, as well, for syntheses of sisomicins B and D and 5⬙-C-methylsisomicin B by the Koenigs–Knorr glycosylation method with appropriately protected glycosyl halides [163]. Aminoglycoside adenyltransferases were used by Coombe and George [126] to prepare adenylated sisomicin. In another report Kim-Wright and colleagues [164] described a glycosylation method used to prepare 2⬘-deoxy- and 2⬘,3⬘-dideoxygentamicin B, 2⬘-deaminogentamicin C1a, and 2⬘-deaminosisomicin from garamine and an appropriately protected glycal; all the products showed potent antibacterial activity against resistant strains. The same group has also prepared the 6⬙-aminogentamicin C2 (compound 336) by the intramolecular insertion reaction of the acyl nitrene group into the C-methyl group in compound 334 (partial structure). This thermal reaction

Compounds 326–333

Syntheses of Aminoglycoside Antibiotics

401

proceeds through the oxazolidinone intermediate 335, which after deprotection affords the title compound 336 [165] (compounds 334–337 ). Philippe et al. [166] reported the use of a quinic acid derived cyclic ␥-amino␣-hydroxy acid related to the 1-N-4-amino-2-hydroxybutyric acid (AHBA) to prepare a 1-N-acyl derivative of gentamicin C1a (compound 337). This new antibiotic showed less activity than gentamicin C1a against gentamicin-sensitive gram-negative organisms, but it was more active against gentamicin-resistant bacteria. To probe the role of the 1-amino group in aminocyclitol antibiotics, the same group used the sequence of reactions outlined below to prepare the 1-deaminogentamicin C2 derivative 341. The free amino group of the protected gentamicin C2 derivative 338 at position 1 was formylated with p-nitrophenyl formate to give 339 and dehydrated to isocyanide 340 in the presence of p-toluenesulfonyl chloride, and then phosphorus oxychloride and triethylamine. This compound was finally converted to the deamino derivative 341 (compounds 338–341 ) by reduction with the tributyltin hydride–AIBN procedure, followed by deprotection. The deaminated analog was neither a substrate nor an inhibitor for resistance enzymes, indicating that the 1-amino group is both essential for antibacterial activity and binding to the ribosomes [167]. A 3⬘,4⬘-dideoxy-3⬘-unsaturated derivative of neamine, made by Barton and coworkers [52], from its 3⬘,4⬘-bisxanthate intermediate via radical elimination reaction, was used for the synthesis of gentamicin C1a. Streicher et al. [130] elsewhere described the synthesis of N-guanyl derivatives of gentamicins C1, C1a, and C2 at positions 2⬘,3⬘, and 6⬘. Also, metal complexion has been used by Nagabhushan and colleagues [168] to prepare 1-N-ethylsisomicin (netilmicin) from a selectively protected sisomicin derivative. Other investigators have recently reported the synthesis of 1-N-ethylsisomicin by selective protection of the 3-, 2⬘-, and 6⬘-amino groups;

Compounds 334–337

402

Haddad et al.

Compounds 338–341

Zn(OAc)2 ⭈ 2H2O was used in methanol for selective acetylation with Ac2O and Et3N in THF, followed by decomplexation with NH3-EtOH solution and ethylation with NaBH(OAc)3 in AcOH-CHCl3 [169]. E.

Streptamine-Containing Aminoglycosides: Spectinomycins

The keto moiety of spectinomycin 342 has been converted to an amino group via a hydroxyl intermediate, leading to the (S) isomer 343 [170], or through reduction of its oxime, resulting in both (S) and (R) isomers 343 and 344 (compounds 342–344 ), respectively. The (R) epimer 344 exhibited activity comparable to that of spectinomycin, but the (S) epimer 343 showed no activity [171]. A number of 3⬘-N-acetylated and 3⬘-N-alkylated derivatives were prepared from the active isomer 344 and tested for antibacterial activity. Some of these analogs showed improved activity, and the most active analog, the N-ethyl derivative, was superior to spectinomycin in terms of activity and toxicity [172].

Compounds 342–344

Syntheses of Aminoglycoside Antibiotics

403

␣-Keto rearrangement of the N-protected spectinomycin 345 to the ␣-hydroxyl lactone 346 in the presence of bis(tributyltin) oxide and bromine or mild acid was reported by Hanessian and Roy [173]. This product (compound 347), which is related to spectinoic acid, was further degraded to 6-deoxy-D-isosaccharino-1,4-lactone 348 (compounds 345–348 ) by methonolysis (3% MeOH-HCl). Partially and fully N-demethylated spectinomycin analogs were also prepared by using oxygen on platinum black for catalytic N-demethylation. However, the resulting products showed reduced or no antibacterial activity [174]. A series of 6⬘-alkylated spectinomycin analogs (compound 349), synthesized by White et al. [175], showed an increase of activity with increasing of the lipophilicity of the antibiotic. Total synthesis of spectinomycin with stereocontrolled rearrangement of its analogs has been reported [176]. There are also reports of the preparation of modified spectinomycin at the 3⬘-position, including mono- and dihalogenated derivatives, as well as branched-chain analogues via diazoketone intermediates [177,178], and synthesis of racemic spectinomycin from a cyclohexylbutadienyl ether precursor [179]. Some of the 3⬘-modified derivatives showed improved activity relative to the parent antibiotic. 3⬘-Aminomethyldihydrospectinomycins have been synthesized via spectinomycin 3⬘-cyanohydrin intermediates [180], and the Tiffeneau–Demjanov diazonium ion rearrangement of the resulting 3⬘-aminomethyl derivative compound 350 has led to the preparation of a series of spectinomycin analogs (compounds 351,

Compounds 345–348

Compound 349

404

Haddad et al.

Compounds 350–353

352, and 353) incorporating a ring-expanded seven-membered ring sugar (e.g., 351). The 3⬘-(S) isomer of 350 gave only the corresponding epoxide 352 [181]. A synthesis of spectinomycin together with the preparation of its 5⬘-n-butyl analog (replacement of the methyl group with n-butyl group) has been described. The 5⬘-n-butyl derivative (trospectomycin or 6⬘-n-propylspectinomycin) showed potent antibacterial activity and is the only aminocyclitol antibiotic that is active against anaerobic bacteria [182]. F.

Striptidine-Containing Aminoglycosides: Streptomycins

Tsuchiya, Umezawa, and their colleagues have described the synthesis of the 3⬙deoxydihydrostreptomycin 355 by Barton deoxygenation of the 3⬙-O-imidazolylthiocarbonyl derivative with tributyltin hydride [183], 6-deoxydihydrostreptamin 356 by displacement of the tosyloxy group of a 6-O-tosyl derivative with chloride followed by hydrogenolysis with liquid ammonia [184], and 3⬙-epi-dihydrostreptamycin 357 by epimerization of the 3⬘-hydroxyl group via the N,O-carbonylallopyranoside intermediate 358 (part structure) [185]. The 3⬙-deoxy analog 355 exhibited remarkable activity against most resistant and sensitive bacteria, except P. aeruginosa. The 6deoxy derivative 356 showed less activity than dihydrostreptomycin, and the 3⬙-epi analog 357 was as active as the 3⬙-deoxy derivative 355 [186] (compounds 354– 358). Methylation of a selectively protected derivative with diazomethane in the presence of tin(II) chloride has been used to prepare 5-O- and 6-O-methyldihydrostreptomycin. However, these antibiotics were only slightly active, indicating the importance of C5 and the hydroxyl groups at positions 5 and 6 for binding to the bacterial ribosome [187]. A series of dihydrostreptomycin analogs modified at the guanidino groups was synthesized. Among these, an analog with a methyl group in the guanidine at position 1 was nearly as active as dihydrostreptomycin, but all analogs having a substituted guanidino group at position 3 were devoid of activity [188]. Photosensitive derivatives of streptomycin (nitroguaiacol derivatives) have also been prepared by modification of the streptose unit, to study the mode of action of the antibiotic [189]. G.

Monoaminocyclitol-Containing Aminoglycosides: Validoxylamines and Validomycins

In extensive work, Ogawa and colleagues have used glycosylation methods to prepare monoaminocyclitol-containing aminoglycoside antibiotics and their analogs. For

Syntheses of Aminoglycoside Antibiotics

405

Compounds 354–358

instance, condensation of DL-validamine with a bromocyclenitol derivative furnished stereoisomers of validoxylamine A [190], and DL-validoxylamines A and B were prepared from condensation of the epoxycyclohexane 359 with the valienamine derivative 360 under thermal conditions [191,192]. Likewise, racemic stereo- and regioisomers of validoxylamine A were prepared [193], and ␤-glycosylation of validoxylamine A (361) with an ␣-D-glucopyranosyl chloride derivative led to the formal synthesis of validamycin A (365) [194] (compounds 359–366 ).

Compounds 359–366 (Part A)

406

Haddad et al.

Compounds 361–368 (Part B)

The same group also reported a total synthesis of 6⬙-epi-validamycin A 366 and its diastereomer in which the racemic cyclohexene epoxide 363 was used for condensation with the protected ␤-D-glucopyranosyl validamine derivative 364. The oxirane 363 has also been used to prepare a DL-6⬘-epi-validoxylamine analog [195,196]. These workers extended their previous work to prepare a DL-2-deoxyvalidoxylamine B analog [197] and to report the first total synthesis of validamycin A via glycosylation of a partially protected validoxylamine A with an ␣-D-glucopyranosyl chloride [198]. Total syntheses of (⫹)-validamycin B (compound 367) [199,200], (⫹)-validamycin A (365) [201,202], validoxylamine B (362) [200], (⫹)-validoxylamine A (361) and (⫹)-validamycin E (compound 368) [202] have also been reported by the same group. In another report, cleavage of the imino bonds of validoxylamine A derivatives with N-bromosuccinimide was demonstrated. This method reports the preparation of synthetically useful derivatives of (⫹)-validamine and valienamine [203]. The same group has also reported the syntheses of (⫹)-validoxylamine G [204], (⫹)-validamycins C, D, and F [35], and (⫹)-validamycin H [36] by standard glycosylation procedures. Asano et al. [205] used a combination of enzymatic and chemical reactions to synthesize eight possible mono-␤-D-glucosides of validoxylamine A. In another synthesis by Ogawa and coworkers [206], treatment of validoxylamine A or its per-Obenzylated derivative with NBS under different conditions resulted in the formation of mixtures of keto- and aminocyclitol derivatives (e.g., 369 and 370). These compounds were used to prepare the dimeric aminocyclitol 371, which is a potent trehalose inhibitor (compounds 369–371 ).

Compounds 369–371

Syntheses of Aminoglycoside Antibiotics

H.

1,4-Diaminocyclitol-Containing Aminoglycosides: Fortimicins, Istamicins, and Related Compounds

1.

Fortimicins

407

The total synthesis of fortimicin B was achieved by glycosylation of a protected ␣sugar chloride with a biscarbamate cyclitol [207]. (a) epo-Fortimicins. 2-O-Mesylfortimicin B has been used to prepare 2-deoxyfortimicins A and B, and 1-deamino-2-deoxy-2-epi-aminfortimicins A and B. The 2deoxyfortimicin A was more active than fortimicin A (compound 372), whereas the 2-epi-amino derivative was devoid of any activity [208]. The 2-epi-fortimicin B (compound 373) was prepared by solvolysis of a 1-N-acetyl-2-O-methanesulfonyl derivative of fortimicin B [209]. Also 1-epi-fortimicins A and B (compounds 374 and 375) were synthesized by epimerization of a 2-keto derivative of fortimicin A with subsequent borohydride reduction [210]. In other work, the 6⬘-epi-fortimicins A and B (compounds 377 and 378) were prepared by oxidative deamination of the 6⬘-amino group with N-chlorosuccinimide to get a 6⬘-keto analog (e.g., 376, partial structure), followed by reductive amination with sodium cyanoborohydride [211]. The 1-epi-fortimicin A (374) showed attenuation of activity against AAC(3)-I producing organisms, and the 6⬘-epi-fortimicin A (377) showed less activity than fortimicin A (372). On the other hand, the 2⬘-epi-fortimicin A prepared from 2⬘-epifortimicin B (373) showed improved antibacterial activity. Several other derivatives of fortimicins, including 1-deamino-2-deoxyfortimicins A and B, 1,2-di-epi-fortimicins, and 2-amino-1-deamino-2-deoxy-1-hydroxyfortimicin A [212], 3-O-demethyl-2,3-di-epi-, and 3-O-demethyl-3-epi-fortimicin A and B [213] have been prepared and tested for their antibacterial activities. The 1,2-di-

Compounds 372–378

408

Haddad et al.

epi- and 3-O-demethyl-2,3-di-epi-fortimicins A were almost half as active as fortimicin A, and the other analogs were devoid of activity. (b) N-Substituted Fortimicins. Base-catalyzed rearrangement of 3-O-demethylfortimicin A has led to formation of 1-N-glycyl-3-O-demethylfortimicin B and 3-Odemethylisofortimicin [214]. Other reports have described the syntheses of a wide variety of 4-N-acyl derivatives of fortimicin A [215] and 4,2⬘-di-N-substituted derivatives of fortimicin B. These include 2-aminoethyl, glycyl, (S)-4-amino-2-hydroxybutyl, and (S)-4-amino-2-hydroxybutyryl substituted analogs (compounds 379–384 ) [216]. Among these analogs, the 4-N-substituted derivative 381 was the most potent antibiotic and showed slightly better activity than fortimicin A (372). The other derivatives showed activity comparable to that of fortimicin A but were more active against resistant strain producing AAC(3)-I enzyme. The literature contains descriptions of the 4-N-glycylation of fortimicin B 30 via selective amino group protection [217] and the selective acylation of fortimicin B and 2-epi-fortimicin B derivatives with N-(N-benzyloxycarbonylglycyloxy)-succinimide [218], as well as 4-N-acyl to 2⬘-N-acyl migration of fortimicins under basic conditions [219]. The resulting 2-O-glycyl and 1-N-glycyl analogs, as well as 5deoxy derivatives of epi-fortimicin A, were devoid of antibacterial activity. Other reports describe the syntheses of 6⬘-N-methyl and 6⬘,6⬘-di-N-methylfortimicins A and B by reductive alkylation in the presence of aldehydes [220], alkylated and acylated derivatives of fortimicins modified at the 1-, 2⬘-, and 6⬘-amino groups [221], and 4de-N-methyl analogs of fortimicins A and B by the Ruschig degradation method [222]. Of these new analogs, the 2⬘-N-[(S)-4-amino-2-hydroxybutyl] fortimicin A was more potent than fortimicin A, 6⬘-N-methylfortimicin A was half as active as the parent compound, and the others were less active or devoid of any activity. Methods have been developed to prepare 4-N-arenesulfonyl and 4-N-alkanesulfonyl fortimicin B (compounds 386 and 387) from the 1,5-carbamate derivative 385. However, these analogs were found to be devoid of antibacterial activity [223] (compounds 385–390 ). Synthesis of 2-deoxy-3-demethyoxyfortimicin A (390) has been described by Tadanier and Hallas [224]. This compound was prepared from the methanesulfonyl derivative 388 by application of the Tipson–Cohen elimination protocol to give the

Compounds 379–384

Syntheses of Aminoglycoside Antibiotics

409

Compounds 385–390

2,3-unsaturated derivative 389 (partial structure) upon mild acid-catalyzed hydrolysis, which after glycylation with N-(N-benzyloxycarbonylglycyloxy)-succinimide, and removal of Cbz groups by hydrogenolysis, afforded the title compound 390. The overall activity of this antibiotic was approximately twice that of fortimicin A and showed improved activity against Pseudomonas species. Fortimicin B has been converted to dactimicin (391) by selective formimidation of the 2⬙-amino group with formimidate hydrochloride in dry ethanol [225]. Another report described the synthesis of a number of fortimicin KG2 analogs, including 3,4didemethyl, 4-demethyl, and 3-amino-4-demethoxy-4-deoxy derivatives from sisamine [226]. (c) Other Fortimicin Analogs. 2-Substituted fortimicins, including 2-amino-2deoxy and 2-chloro-2-deoxyfortimicin A and its 2-epimer (compounds 393–395 ), have been made by stereo- and regiospecific aziridine ring opening of the 2-deoxy1,2-epimino-2-epi-fortimicin B (392) with chloride or azide. However, all these new analogs were less active than fortimicin A [227] (compounds 391–395 ). The literature provides descriptions of the preparation of a series of fortimicin analogs modified in the sugar moiety [228], the formation of mono- and biscarbamate

410

Haddad et al.

Compounds 391–395

of fortimicin B [229], the synthesis of 2-amino-3-O-demethyl-2-deoxy- and 3-amino3-demethoxy fortimicin A [230], and a number of fortimicin analogs modified in the sugar ring, including 4⬘-hydroxyfortimicin D [231], 7⬘-C-propylfortimicin A [232], 7⬘-(3-hydroxypropyl)fortimicin A and its 6⬘-epimer [233], 7⬘-phenylfortimicin A and its 6⬘-epimer [234], and 3⬘-enofortimicin D [235]. Of these analogs, only 2-amino3-O-demethyl-2-deoxy- and 3-amino-3-demethoxyfortimicin A showed activity comparable to that of fortimicin A; the others were either inactive or less active than the parent compounds. Several synthetic methodologies have been employed to convert the 1,6-anhydromaltose derivative 396 to the 1,4-diaminocyclitol aminoglycoside 397 a fortimicin analog. However, this compound was approximately 25% as potent as fortimicin A [236]. The same precursor was also used to prepare fortimicin A analogue 398 via regio- and stereoselective introduction of azido and methylamino groups and transformation of the 1,6-anhydrohexopyranose moiety into a cyclitol ring [237] (compounds 396–398 ). 2.

Istamycins

3⬘,4⬘-Dideoxyneamine has been used to prepare demethylated analogs of istamycin A (399). Antibacterial activity screening of these compounds revealed that the 4-Nmethyl group is essential for antibacterial activity, since 4-N-demethylated derivatives exhibited weak activity, whereas 6⬘-N-3-O-didemethylistamycin A still showed activity [238]. The compounds 2⬙-N-formimidoylistamycin (403) and 2⬘-N-formimidoylstamycin B (404) have been prepared by treatment of appropriately protected istamycin A0 and B0 derivatives with N-tert-butoxycarbonylglycine (compounds 399–404 ). De-

Syntheses of Aminoglycoside Antibiotics

411

Compounds 396–398

protection of the 2-N-tert-butoxycarbonyl group followed by formimidation of the 2⬙-amino function with ethylformimidate hydrochloride and deprotection of the protecting groups led to the formation of the title compounds. These naturally occurring aminoglycosides have strong antibacterial activity, similar to istamycin A and B [239]. A number of istamycin B analogs, including 5-epi-, 3-O-demethyl-5-epi-, 3-Odemethyl-3-epi-, 3-demethoxy, 3-demethoxy-2⬙-N-formimidoyl [240], 3-O-demethyl, and several 2⬘-N-acyl and N-amidinyl [241] derivatives have been synthesized. Of these, the 3-demethoxy and 3-demethoxy-2⬙-N-formimidoyl analogs showed more potency than istamycin B and its 3-O-demethyl derivative. Various istamycin derivatives (compound 405) bearing different aminoacyl groups at the 4-amino group have been synthesized. These analogs, which have a combination of hydroxyl and amino groups for X and Y in their structures [e.g., (X, Y) = (NH2, OH), (OH, OH), (H, OH), (OH, NH2)], showed promising antibiotic activity. Of these, the antibiotic (compound 406) was tested with 54 bacteria and showed potent antibacterial activity, much higher than that of istamycin B [242]. 3.

Sporaricins

The three fluorinated derivatives of sporaricin A 407, including 3-demethoxy-3-fluoro (408), 3-demethoxy-epi-fluoro (409), and 3-demethoxy-3,3-difluoro (410), have been prepared from a protected 3-demethylsporaricin derivative. Direct fluorination of the

Compounds 399–404

412

Haddad et al.

Compounds 405 and 406

3-hydroxyl group with DAST gave the 3-fluoro derivative, and oxidation of the free hydroxyl group followed by sodium borohydride reduction and the subsequent fluorination afforded the epi-fluoro compound. Likewise, PCC oxidation of the hydroxyl group at position 3 followed by fluorination of the resulting oxo derivative furnished the difluoro analog. Of these analogs, the 3-fluoro derivative 408 showed less toxicity and more activity than sporaricin A [243]. Replacement of the 3-O-methyl group of sporaricin A with allyl group via hydroiodic acid assisted removal of the methyl group, followed by allylation of the free 3-hydroxyl group, led to preparation of the 3-O-substituted analogs 411–413 (compounds 407–413 ). These compounds were synthesized by epoxidation of the allylic double bond with mCPBA, followed by sequential treatment with sodium azide, water, or potassium cyanide. All these analogs exhibited high antibacterial activity against AAC(3)-I producing organisms. Among these analogs, compounds 411 and 413 showed enhanced activity against P. aeruginosa compared to the parent compound [244]. Total synthesis by a glycosylation method of the semisynthetic antibiotic 3-Omethylsporaricin A has been described [245]. Other reports describe the preparation of a series of fluorinated sporaricin A, including 3-fluoro and 3,3-difluoro derivatives

Compounds 407–413

Syntheses of Aminoglycoside Antibiotics

413

of 3-demethoxysporaricin A [246], as well as 5-deoxy-5-fluorosporicin A [247]. In this work, the direct fluorination method was used with the DAST reagent to make the fluorinated analogs. 4.

Lysinomicin

4-N-Glysyllysinomicin (415) and several 4-N-aminoacyl analogs of lysinomicin derivatives (compounds 416–418 ) have been prepared to examine the effect of 4-Naminoacylation on antibacterial activity of this class of aminoglycoside antibiotics (compounds 414–418 ). In contrast to lysinomicin, which is slightly more potent than fortimicin A (372), the new analogs showed lower activity than the 4-N-unsubstituted lysinomicin derivatives, indicating that the structure–activity relationship observed with fortimicins does not apply to lysinomicin derivatives [248]. 5.

Sannamycins

Ludin et al. [249] have described the synthesis of the pseudodisaccharide sannamycin 419, its 2-epimer, and a variety of other isomers including the enantiomers of sannamycin. These workers also prepared various other sannamycin analogs including 6⬘-de(N-methyl) sannamycin A (420) and its 2⬘-epi isomer 421, as well as derivatives 422 and 423, using the glycosyl donors 424 or 425 for glycosylation with sannaminetype glycosyl acceptor 426 (compounds 419–426 ). Among these analogs, compound 420 showed no activity, and 421 was slightly active, but much less active than the parent compound, sannamycin (419) [250]. I.

Other Aminocyclitol Aminoglycosides

Three ␣-glycosidase inhibitors of 2,5,6-trideoxystreptamine (compounds 427–429 ) have been prepared by glycosylation of the quinic acid derived ditosyloxycyclohexanol derivative 430, with the corresponding glycal derivatives (e.g., 431 and 432)

Compounds 414–418

414

Haddad et al.

Compounds 419–426

Compounds 427–432

as glycosyl donors. However these compounds were devoid of antibacterial activity, indicating that the presence of an ether linkage at the 4⬘-position (in comparison with apramycin, 48) is probably essential for antibacterial activity in this class of aminoglycoside antibiotics [251].

Syntheses of Aminoglycoside Antibiotics

J.

Noncyclitol Aminoglycosides

1.

Trehalosamines

415

Trehalosamine (433), an ␣,␣-(1→1)-linked disaccharide produced by Streptomyces lavendulae, is an aminoglycoside antibiotic composed of D-glucose and 2-amino-2deoxy-D-glucose [252]. Ogawa and colleagues [253,254] used a racemic inosamine derivative to prepare the trehalosamine analogs 434–436 by glycosylation with an appropriately protected glucopyranosyl bromide. Likewise, a number of pseudosugar analogs, including the ␣-D-glucosides 437 and 438, were synthesized by glycosylation of the ␤-D-glucopyranose derivative 439 with the ␣-DL-glucopyranose 440 in the presence of trimethylsilyl trifluoromethanesulfonate (compounds 433–440 ). However, these compounds exhibited no antibacterial activity [37,254]. 2.

Acarviosins

Acarviosin 441, an ␣-glycosidase inhibitor, is a maltose-type pseudodisaccharide that can be prepared from methanolysis of acarbose 442 [255]. It has been shown to have five times more inhibitory activity than its precursor acarbose, for which the first total synthesis has been described by Ogawa’s group in 1989 [256]. The methylacarviosin 447, a core structure of acarbose and related pseudooligosaccharidic ␣-amylase inhibitors, together with its 6-hydroxy analog 448, were prepared by Shibata et al. [257] from the corresponding oligobiosaminide derivatives 443 and 444. The aziridine intermediates 445 and 446 were treated with hydrochloric acid to give the corresponding 6⬘-chloro derivative, and then elimination of the chlorine by base furnished the title compounds 447 and 448 as peracetylated derivatives (compounds 441–448 ). The same group has also prepared several other ␣-glucosidase inhibitors, including 1⬘,6⬘- and 3⬘,6⬘-anhydro derivatives of acarviosin [258], methylacarviosin

Compounds 433–440

416

Haddad et al.

Compounds 441–448 derivatives [259], and analogs of acarviosin and methyl acarviosin having ␣-D-manno configurations and saturated cyclohexane rings [260,261]. Another report by this group describes the synthesis of methyl 1⬘-epi-acarviosin and its 6-hydroxy analog [262]. Various other 1,6-anhydro acarviosins, including 2⬘-substituted derivatives having the 2⬘-hydroxyl group replaced with NH2 or F, as well as a diastereomeric methylacarviosin and its hydroxylated analog, have also been synthesized by the same group [263,264]. The total synthesis of the trehalose inhibitor salbostatin 451 was achieved by the coupling reaction of the ␣-valienamine derivative 449 with 1,5:2,3-dihydro-Dmannitol (450) under forcing conditions (sealed tube, 120⬚C) in a protic solvent [265] (compounds 449–451 ). K.

Aminocyclopentitol Aminoglycosides: Trehazolin and Trehalostatin

The total syntheses of the trehalose inhibitors trehazolin (452) [266,267] and its 5epi analog trehalostatin (453) [266,268] have been described by two different groups. The 5-epi derivative 453 possesses much weaker inhibitory activities than trehazolin toward two trehalases. Another publication by Boiron et al. [269] reports a synthesis of trehazolin 452 from D-glucose. In this synthesis three chiral centers of the starting compound, D-glucose, are conserved and the remaining two chiral centers of trehazolamine 456 are formed stereoselectively by reductive cyclization of the ketoxime ether 454. Reduction of this oxide ether leads to formation of the title compound 456 (compounds 452–456 ). In addition, different research groups have described syntheses of (⫹)-trehazolin [270,271], (⫹)-6-epitrehazolin [271], a tetrahydropyrano-[2,3-d] oxazol analog

Syntheses of Aminoglycoside Antibiotics

417

Compounds 449–451

Compounds 452–456

of trehazolin [272], and a series of trehazolin derivatives in which the ␣-D-glucopyranose unit was replaced by other mono- and disaccharide or carba sugar units [273]. The inhibitory activity results revealed that the four hydroxyl groups of the D-glucopyranosyl moiety of trehazolin are topologically essential for the hydrogen binding of the inhibitor to the active site of the enzyme. Furthermore, it appeared that the ␤-D-glucopyranosyl residue is likely to facilitate the uptake of the compound into the cell [273].

REFERENCES 1. 2. 3. 4. 5. 6.

7. 8. 9.

S Hanessian, TH Haskell. In: H Pigmann, D Horton, eds. Antibiotics Containing Sugars. New York/London: Academic Press, 1970, pp 139–211. S Umezawa. Adv Carbohydr Chem Biochem 30:11–182, 1974. DA Cox, K Richardson, BC Ross. In: PG Sammes, ed. The Aminoglycosides. New York: John Wiley & Sons, 1977, pp 1–90. S Umezawa, T Tsuchiya. In: H Umezawa, IR Hooper, eds. Handbook of Experimental Pharmacology. New York: Spring-Verlag, 1982, pp 37–110. J Haddad, LP Kotra, S Mobashery, Chapter 10, this volume. GD Wright, AM Berghuis, S Mobashery. In: BP Rosen and S Mobashery, eds. Resolving the Antibiotic Paradox: Progress in Understanding Drug Resistance and Development of New Antibiotics. New York: Plenum Press, 1998, pp 27–69. SA Lerner, RP Gaynes. In: SL Gorbach, JG Bartlett, NR Blacklow, eds. Infectious Diseases. Philadelphia: WB Saunders, 1992, pp 188. JJ Wright, A Cooper, PJL Daniels, TL Nagabhushan, D Rane, WN Turner, J Weinstein. J Antibiot 29:714–719, 1976. H Kawaguchi, T Naito, S Nakagawa, K Fujisawa. J Antibiot 25:695–708, 1972.

418

Haddad et al.

10.

TL Nagabhushan, AB Cooper, H Tsai, PJL Daniels, GH Miller. J Antibiot 31:681–687, 1978. A Kucers, NMcK Bennett. In: A Kucers, NMcK Bennett, eds. The Use of Antibiotics. London: William Heinemann Medical Books, 1987, pp 619. S Kondo, K Iinuma, H Yamamoto, K Maeda, H Umezawa. J Antibiot 26:412–415, 1973. T Naito, S Nakagawa, Y Abe, S Toda, K Fujisawa, T Miyaki, H Koshiyama, H Ohkuma, H Kawaguchi. J Antibiot 26:297–301, 1973. T Naito, S Nakagawa, Y Narita, S Tada, Y Abe, M Oka, H Yamashita, T Yamasaki, K Fujisawa, H Kawaguchi. J Antibiot 27:851–858, 1974. S Umezawa, T Tsuchiya, K Tatsuta. Bull Chem Soc Jpn 39:1235–1243, 1966. S Hanessian, G Patil. Tetrahdedron Lett 12:1035–1038, 1978. TL Nagabhushan, AB Cooper, WN Turner, H Tsai, S McCombie, AK Mallams, D Rane, JJ Wright, P Reichert, DL Boxler, J Weinstein. J Am Chem Soc 100:5253–5254, 1978. TL Nagabhushan, WN Turner, AB Cooper. US Patent 4,230,847, 1980. PJL Daniels, AB Cooper, SW McCombie, TL Nagabhushan, DF Rane, JJ Wright. Jpn J Antibiot [suppl] 32:S195–S204, 1979. T Tsuchiya, Y Takagi, S Umezawa. Tetrahedron Lett 141:4951–4954, 1979. H Iwasawa, D Ikeda, S Kondo, H Umezawa. J Antibiot 37:428–430, 1984. Y Kobayashi, T Tsuchiya, S Umezawa, T Yoneta, S Fukatsu, H Umezawa. Bull Chem Soc Jpn 60:713–720, 1987. A van Schepdael, R Busson, HJ Vanderhaeghe, PJ Claes, L Verbist, MP MingeotLeclercq, R Brasseur, PM Tulkens. J Med Chem 34:1483–1492, 1991. J Roestamadji, I Grapsas, S Mobashery. J Am Chem Soc 117:11060–11061, 1995. Bristol-Myers, British Patent 1,486,450, 1977. I Grapsas, YJ Cho, S Mobashery. J Org Chem 59:1918–1922, 1994. J Roestamadji, I Grapsas, S Mobashery. J Am Chem Soc 117:80–84, 1995. K Matsuda, N Yasuda, H Tsutsumi, T Takaya. J Antibiot 40:843–854, 1987. NE Allen, WE Alborn, HA Kirst, JE Toth. J Med Chem 30:333–340, 1987. S Kondo, Y Ikeda, T Takeuchi, T Usui, M Ishii, T Kudo, S Gomi, S Shibahara. J Antibiot 47:821–832, 1994. Y Takahashi, J Kohno, T Tsuchiya. Carbohydr Res 306:349–360, 1998. K Toshima, K Tatsuta. Chem Rev 93:1503–1531, 1993. J Banoub, P Boullanger, D Lafont. Chem Rev 92:1167–1195, 1992. PB Alper, M Hendrix, P Sears, C-H Wong. J Am Chem Soc 120:1965–1978, 1998. Y Miyamoto, S Ogawa. J Chem Soc Perkin Trans I 2121–2129, 1991. Y Miyamoto, S Ogawa. Carbohydr Res 223:299–301, 1992. S Ogawa, K Nishi, Y Shibata. Carbohydr Res 206:352–360, 1990. S Umezawa, T Tsuchiya, T Yamasaki, H Sano, Y Takahashi. J Am Chem Soc 96:920– 921, 1974. T Suami, Y Honda. Chem Lett 641–644, 1980. T Usui, S Umezawa. Carbohydr Res 174:133–143, 1988. H Paulsen, P Stadler, F Tidter. Chem Ber 110:1925–1930, 1977. RU Lemieux, RV Stick. Aust J Chem 31:901–905, 1978. M Kugelman, AK Mallams, HF Vernay. J Chem Soc Perkin Trans I 1126–1134, 1976. M Kugelman, AK Mallams, HF Vernay, DF Crowe, G Detre, M Tanabe, DM Yasuda. J Chem Soc Perkin Trans I 1097–1113, 1976. Y Takahashi, T Tsuchiya, S Umezawa, H Umezawa. Carbohydr Res 210:221–232, 1991. K Tatsuta, K Akimoto, H Takahashi, T Hamatsu, M Annaka, M Kinoshita. Bull Chem Soc Jpn 57:529–538, 1984.

11. 12. 13. 14. 15. 16. 17.

18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.

Syntheses of Aminoglycoside Antibiotics 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87.

419

G Medgyes, G Jerkovich, J Kuszmann. Carbohydr Res 186:225–239, 1989. A Canas-Rodriguez, A Martinez-Tobed. Carbohydr Res 68:43–53, 1979. C Colas, B Quiclet-Sire, J Clephax, J-M Delaumeny, A-M Sepulchre, SD Gero. J Am Chem Soc 102:857–858, 1980. S Tohma, T Yoneta, S Fukatsu. J Antibiot 33:671–676, 1980. RD Sitrin, FR Pfeiffer, JR Rosenbloom, DJ Cooper, SJ Schmidt, D Peterson, G Wellman, JRE Hoover, JA Weisbach. J Antibiot 33:383–392, 1980. DHR Barton, DK Zhong, SD Gero. J Carbohydr Chem 1:105–118, 1982. MP Georgiadis, V Constantinov-Kokotouv. J Carbohydr Chem 10:739–748, 1991. MP Georgiadis. Chem Chron 22:109–123, 1993. WKC Park, M Auer, H Jaksche, C-H Wong. J Am Chem Soc 118:10150–10155, 1996. J Haddad, S Mobashery. Unpublished results. M-Z Liu, J Haddad, E Azucena, LP Kotra, M Kirzhner, S Mobashery. J Org Chem 65: 7422–7431, 2000. M-Z Liu, S Mobashery. Unpublished results. M Yoshikawa, Y Ikeda, H Kayakiri, K Takenaka, I Kitagawa. Tetrehedron Lett 23: 4717–4720, 1982. FR Pfeiffer, TW Ku, DC Peterson. J Antibiot 34:5–12, 1981. OR Martin, WA Szarek. Carbohydr Res 130:195–219, 1984. J-M Girodeau, R Pineau, M Masson, F LeGoffic. J Antibiot 37:143–149, 1984. Y Abu, S Nakagawa, T Naito, H Kawaguchi. J Antibiot 34:1434–1446, 1981. HA Kirst, BA Truedell, JE Toth. Tetrahedron Lett 22:295–298, 1981. OR Martin, WA Szarek. J Chem Soc Chem Commun 926–928, 1983. K Tatsuta, K Akimoto, H Takahashi, T Hamatsu, M Annaka, M Kinoshita. Tetrahedron Lett 24:4867–4870, 1983. A Canas-Rodriguez, SG Ruiz-Poveda, LA Coronel-Borges. Carbohydr Res 159:217– 227, 1987. A Canas-Rodriguez, LA Coronel-Borges. Carbohydr Res 165:129–133, 1987. J Yoshimura, S Horito, J-I Yamura, H Hashimoto. Chem Lett 1335–1338, 1985. J-I Tamura, S Horito, H Hashimoto, J Yoshimura. Carbohydr Res 174:181–199, 1988. T Usui, S Umezawa. J Antibiot 40:1464–1467, 1987. S Toda, S Nakagawa, T Naito, H Kawaguchi. J Antibiot 36:87–91, 1983. H Wang, Y Tor. Angew Chem Int Ed Engl 37:109–111, 1998. S Hanessian, J-M Vatele. J Antibiot 33:675–681, 1980. C Battistini, G Gassinelli, G Franceschi, F Arcamone, R Mazzoleni. J Antibiot 34:240– 242, 1981. C Battistini, G Franceschi, F Zarini, G Cassinelli, F Arcamone. J Antibiot 35:98–101, 1982. T Torii, T Tsuchiya, I Watenabe, S Umezawa. Bull Chem Soc Jpn 35:1178–1182, 1982. T Torii, T Tsuchiya, S Umezawa. J Antibiot 35:58–61, 1982. T Torii, T Tsuchiya, I Watenabe, S Umezawa, H Umezawa. Bull Chem Soc Jpn 56: 1522–1526, 1983. V Kumar, GS Jones, I Blacksberg, WA Remers. J Med Chem 23:42–49, 1980. V Kumar, WA Remers. J Org Chem 46:4298–4300, 1981. M Yoshikawa, Y Ikeda, K Takenaka, M Torihara, I Kitagawa. Chem Lett 2097–2100, 1984. J-M Girodeau, R Pineau, M Masson, F LeGoffic. J Antibiot 37:150–158, 1984. PWK Woo, TH Haskell. J Antibiot 35:692–702, 1982. Y Narita, S Masuyoshi, T Yamasaki, T Naito, H Kawaguchi. J Antibiot 44:86–92, 1991. N Sakairi, T Yoneta, MdD Rahmad, Y Ishido. J Carbohydr Chem 3:51–65, 1984. T Jikihara, T Tsuchiya, H Umezawa, S Umezawa. Carbohydr Res 89:91–101, 1981.

420

Haddad et al.

88. 89. 90.

T Yoneta, T Matsuno, H Nanahoshi, S Fukatsu. Chem Pharm Bull 29:3469–3473, 1981. H Iwasawa, D Ikeda, S Kondo, H Umezawa. J Antibiot 35:1715–1718, 1982. T Yoneta, T Matsuno, H Nanahoshi, S Fukatsu. Nippon Kagaku Kaishi 1730–1736, 1982. T Miyake, T Tsuchiya, T Nishi, S Umezawa. Bull Chem Soc Jpn 56:1149–1154, 1983. T Miyake, Y Takahashi, T Tsuchiya, S Umezawa, H Umezawa. Nippon Kagaku Kaishi 1706–1712, 1982. T Suami, K Nakamura. Bull Chem Soc Jpn 53:3655–3657, 1980. T Yoneta, T Matsuno, S Fukatsu. Chem Pharm Bull 29:3464–3468, 1981. Y Oka, H Ishida, M Morioka, Y Numasaki, T Yamafuji, T Osono. J Antibiot 34:777– 781, 1981. T Matsuno, T Yoneta, S Fukatsu. Carbohydr Res 109:271–275, 1982. M Yoshikawa, M Torihara, T Nakae, BC Cha, I Kitogawa. Chem Pharm Bull 35:2136– 2139, 1987. H Umezawa, H Iwasawa, D Ikeda, S Kondo. J Antibiot 36:1087–1091, 1983. RJ Reid, SA Mizsak, LM Reineke, GE Zurenko, KF Stren, BJ Magerlerin. J Med Chem 24:1487–1492, 1981. R Albert, K Dax, AE Stu¨tz, H Weidmann. Tetrahedron Lett 23:2645–2646, 1982. Y Nishimura, H Umezawa. Tetrahedron Lett 27:77–80, 1981. K Igarashi, T Sugawara, T Honda, Y Tada, H Miyazaki, H Nagata, M Mayama. Carbohydr Res 109:73–88, 1982. Y Takahashi, T Tsuchiya, Y Suzuki, S Umezawa. H Umezawa, S Fukatsu. Bull Chem Soc Jpn 56:1807–1811, 1983. R Albert, K Dax, AE Stu¨tz, H Weidmann. J Carbohydr Chem 3:289–299, 1982–1983. R Albert, K Dax, NA Gassner, AE Stu¨tz, H Weidmann. Liebigs Ann Chem 644–649, 1985. R Albert, K Dax, AE Stu¨tz. J Antibiot 38:275–278, 1985. E Umemura, T Tsuchiya, K Tanaka, S Umezawa. J Antibiot 43:1611–1614, 1990. R Albert, K Dax, AE Stu¨tz. Tetrahedron Lett 24:1763–1766, 1983. R Albert, K Dax, AE Stu¨tz. Carbohydr Res 132:162–167, 1984. R Albert, K Dax, AE Stu¨tz. J Carbohydr Chem 3:267–277, 1984. T Tsuchiya, Y Takahashi, Y Kobayashi, S Umezawa. J Antibiot 38:1287–1290, 1985. Y Tagaki, T Tsuchiya, S Umezawa. Nippon Kagaku Kaishi 10:2001–2009, 1985. T Tsuchiya, Y Takahashi, M Endo, S Umezawa, H Umezawa. J Carbohydr Chem 4: 587–611, 1985. E Umemura, T Tsuchiya, S Umezawa. Bull Chem Soc Jpn 63:3583–3589, 1990. E Umemura, T Tsuchiya, K Tanaka, S Umezawa. J Antibiot 43:1161–1163, 1990. E Umemura, T Tsuchiya, Y Kobayashi, K Tanaka. Carbohydr Res 224:141–163, 1992. Y Takahashi, S Tsuneda, T Tsuchiya, Y Koyama, S Umezawa. Carbohydr Res 232:89– 105, 1992. T Shitara, Y Kobayashi, T Tsuchiya, S Umezawa. Carbohydr Res 232:273–290, 1992. Y Kobayashi, T Tsuchiya, T Ohgi, N Taneichi, Y Koyama. Carbohydr Res 230:89– 105, 1992. E Umemura, T Tsuchiya, Y Koyama, S Umezawa. Carbohydr Res 238:147–162, 1993. T Tsuchiya, T Shitara, S Umezawa, T Takeuchi, M Hamada, N Tomono, E Umemura. Carbohydr Res 240:307–312, 1993. T Shitara, E Umemura, T Tsuchiya, T Matsuno. Carbohydr Res 276:75–89, 1995. J Haddad, Sergei Vakulenko, S Mobashery. J Am Chem Soc 121:11922–11923, 1999. H Umezawa, T Miyasaka, H Iwasawa, D Ikeda, S Kondo. J Antibiot 34:1635–1640, 1981. Y Takagi, C Komuro, T Tsuchiya, S Umezawa, M Hamada, H Umezawa. J Antibiot 34:1–4, 1981.

91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125.

Syntheses of Aminoglycoside Antibiotics 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156.

157. 158. 159. 160. 161.

421

RG Coombe, AM George. Aust J Chem 34:547–554, 1981. MN Sharma, V Kumar, WA Remers. J Antibiot 35:905–910, 1982. E Akita, Y Horiuchi, T Miyazawa. J Antibiot 36:745–748, 1983. LM Cappelletti, R Spagnoli. J Antibiot 36:328–330, 1983. W Streicher, H Loibner, J Hildebrandt, F Turnowsky. Drugs Exp Clin Res 9:591–598, 1983. K Matsuda, N Yasuda, H Tsutsumi, T Takaya. J Antibiot 38:547–549, 1985. K Matsuda, N Yasuda, H Tsutsumi, T Takaya. J Antibiot 38:1050–1060, 1985. K Matsuda, N Yasuda, H Tsutsumi, T Takaya. J Antibiot 39:1479–1482, 1986. K Matsuda, N Yasuda, H Tsutsumi, T Takaya. J Antibiot 40:843–854, 1987. BS Iyengar, V Kumar, TP Wunz, WA Remers. J Med Chem 29:611–614, 1986. E Umemura, T Tsuchiya, S Umezawa. J Antibiot 41:530–537, 1988. Y Takahashi, C Ueda, T Tsuchiya, Y Kobayashi. Carbohydr Res 249:57–76, 1993. J Szymoniak, B El Mouatassim, J Besancon, C Moı¨se, D Brossier. Tetrahedron 49: 3109–3116, 1993. S Kotretsou, MP Mingeot-Leclercq, V Constantinou-Kokotou, R Brasseur, MP Georgiadis, PM Tulkens. J Med Chem 38:4710–4719, 1995. G. Gunawardana, C Childress, M Tripp, X Zhang, P West. J Antibiot 50:887–889, 1997. H Kunimoto, S Atsuko, I Jun, M Satoshi, I Yoko, K Shinichi. J Antibiot 51:735–742, 1988. T Tsuchiya, M Nakano, T Torii, Y Suzuki, S Umezawa. Carbohydr Res 136:195–206, 1985. A van Schepdael, J Delcourt, M Mulier, R Busson, L Verbist, HJ Vanderhaeghe, MP Mingeot-Leclercq, PM Tulkens, PJ Claes. J Med Chem 34:1468–1482, 1991. A van Schepdael, R Busson, HJ Vanderhaeghe, PJ Claes, L Verbist, MP MingeotLeclercq, R Brasseur, PM Tulkens. J Med CHem 34:1483–1492, 1991. M-P Mingeot-Leclercq, A van Schepdael, R Brasseur, R Busson, HJ Vanderhaeghe, PJ Class, PM Tulkens. J Med Chem 34:1476–1482, 1991. S Kondo, S Shibahara, T Usui, T Kudo, A Tamura, S Gomi, Y Ikeda, D Ikeda, T Takeuchi. J Antibiot 46:531–534, 1993. R Kuwahara, T Tsuchiya. Carbohydr Res 299:271–279, 1997. Y Takegi, C Komuro, T Tsuchiya, S Umezawa. Bull Chem Soc Jpn 54:1834–1837, 1981. Y Nishimura, S Umezawa. Tetrahedron Lett 23:81–84, 1982. Y Nishimura, S Umezawa. Tetrahedron Lett 23:85–86, 1982. R Kuwahara, T Tsuchiya. Carbohydr Res 286:107–122, 1996. Y Kobayashi, T Tsuchiya. Carbohydr Res 298:261–277, 1997. H Wang, Y Tor. Bioorg Med Chem Lett 7:1951–1956, 1997. A Sreedhara, JA Cowan. J Chem Soc Chem Commun 1737–1738, 1998. S Masayuki, U Takayuki, N Tetsutaro, AS Reiei, G Shuichi, I Shigeharu, K Shinichi, T Tomio. Jpn Kokai Tokkyo Koho JP 09, 104, 694 [97, 104, 694]. PJL Daniels, GE Luce, AK Mallams, JB Morton, SS Saluja, H Tsai, J Weinstein, JJ Wright, G Detre, M Tanabe, DM Yasuda. J Chem Soc Perkin Trans I 2137–2150, 1981. DH Davies, M Kugelman, P Lee, CE Luce, AK Mallams, JB Morton, SS Saluja, JJ Wright, G Detre, M Tanabe, DM Yauda. J Chem Soc Perkin Trans I 2151–2167, 1981. DL Boxler, R Brambilla, DH Davies, AK Mallams, SW McCombie, JB Morton, P Reichert, HF Vernay. J Chem Soc Perkin Trans I 2168–2185, 1981. AK Mallams, JB Morton, P Reichert. J Chem Soc Perkin Trans I 2186–2208, 1980. H Paulsen, R Jansen. Carbohydr Res 92:305–309, 1981. H Paulsen, B Schro¨der, H Bottcher, R Hohlweg. Chem Ber 114:322–332, 1981.

422

Haddad et al.

162. 163. 164. 165. 166.

H Paulsen, R Jansen, P Stadler. Chem Ber 114:837–842, 1981. H Paulson, R Jansen. Chem Ber 114:843–857, 1981. JJ Kim Wright, P Lee. J Carbohydr Chem 1:157–190, 1982. JJ Kim Wright, JA Albarella, P Lee. J Org Chem 47:523–527, 1982. M Philippe, AM Sepulchre, SD Gero, H Loibner, W Streicher, P Stu¨tz. J Antibiot 35: 1507–1512, 1982. M Philippe, B Quiclet-Sire, AM Sepulchre, SD Gero, H Loibner, W Streicher, P Stu¨tz, N Moreau. J Antibiot 36:250–255, 1983. TL Nagabhushan, WN Turner, H Tsai, RS Jaret, D Schumacher, JK Jenkins, J Chiu. Carbohydr Res 130:243–249, 1984. Jpn Kokai Tokkyo Koho JP 08, 217, 788 [96, 217, 788]. S Hanessian, R Roy, M Therien, D Delorme. J Antibiot 34:350–352, 1981. R Maier, E Woitun, A Reuter, W Reuter, B Wetzel. J Antibiot 34:16–21, 1981. E Woitun, R Maier, B Wetzel, W Reuter, U Lechner. J Antibiot 34:22–27, 1981. S Hanessian, R Roy. Tetrahedron Lett 22:1005–1008, 1981. RC Thomas. J Antibiot 35:1202–1207, 1982. DR White, CJ Maring, GA Cain. J Antibiot 36:339–342, 1983. S Hanessian, R Roy. Can J Chem 63:163–172, 1985. RC Thomas, EL Fritzen. J Antibiot 38:197–207, 1985. RC Thomas, EL Fritzen. J Antibiot 38:208–219, 1985. S Danishefsy, J Aube´, M Bednarski. J Am Chem Soc 108:4145–4149, 1986. RC Thomas, EL Fritzen. J Antibiot 41:1439–1444, 1988. RC Thomas, EL Fritzen. J Antibiot 41:1445–1451, 1988. DR White, GA Cain. Tetrahedron Lett 30:1469–1472, 1989. T Usui, T Tsuchiya, H Umezawa, S Umezawa. Bull Chem Soc Jpn 54:781–786, 1981. T Tsuchiya, T Kishi, S Kobayashi, Y Kobayashi, S Umezawa. Carbohydr Res 104:69– 77, 1982. T Tsuchiya, T Shitara. Carbohydr Res 109:59–72, 1982. T Tsuchiya, S Sakamoto, T Yamasaki, S Umezawa, H Umezawa. J Antibiot 35:639– 641, 1982. T Kishi, T Tsuchiya, N Koga, Y Kobayashi, S Umezawa. Bull Chem Soc Jpn 58:3087– 3091, 1985. DL Delaware, MS Sharma, BS Iyengar, WA Remers, JA Pursiano. J Antibiot 39:251– 258, 1986. JP Abad, A Amils. Antimicrob Agents Chemother 34:1908–1914, 1990. S Ogawa, T Toyokuni, T Suami. Chem Lett 947–950, 1981. S Ogawa, T Toyokuni, Y Iwasawa, Y Abe, T Suami. Chem Lett 279–282, 1982. S Ogawa, T Ogawa, N Chida, T Toyokuni, T Suami. Chem Lett 749–752, 1982. T Toyokuni, S Ogawa, T Suami. Bull Chem Soc Jpn 56:2999–3004, 1983. S Ogawa, T Ogawa, T Nose, T Toyokuni, Y Iwasawa, T Suami. Chem Lett 921–922, 1983. S Ogawa, M Inoue, Y Iwasawa, T Toyokuni, T Suami. Chem Lett 1085–1088, 1983. S Ogawa, M Suetsugu, T Toyokumi, T Suami. Nippon Kugaku Kaishi 1721–1726, 1982. S Ogawa, H Ito, T Suami. Bull Chem Soc Jpn 56:2326–2329, 1983. S Ogawa, T Nose, T Ogawa, T Toyokuni, Y Iwasawa, T Suami. J Chem Soc Perkin Trans I 2369–2374, 1985. S Ogawa, Y Miyamoto. J Chem Soc Chem Commun 1843–1844, 1987. S Ogawa, Y Miyamoto, T Nose. J Chem Soc Perkin Trans I 2675–2680, 1988 S Ogawa, Y Miyamoto. Chem Lett 889–890, 1988. Y Miyamoto, S Ogawa. J Chem Soc Perkin Trans I 1013–1018, 1989. S Ogawa, Y Miyamoto, A Nakajima. Chem Lett 725–728, 1989.

167. 168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194. 195. 196. 197. 198. 199. 200. 201. 202. 203.

Syntheses of Aminoglycoside Antibiotics 204. 205. 206. 207. 208. 209. 210. 211. 212. 213. 214. 215. 216. 217. 218. 219. 220. 221. 222. 223. 224. 225. 226. 227. 228. 229. 230. 231. 232. 233. 234. 235. 236. 237. 238. 239. 240. 241.

423

Y Miyamoto, S Ogawa. J Chem Soc Chem Commun 999–1000, 1990. N Asano, Y Kameda, K Matsui. J Antibiot 44:1406–1416, 1991. S Ogawa, A Nakajima, Y Miyamoto. J Chem Soc Perkin Trans I 3287–3290, 1991. Y Honda, T Suami. Bull Chem Soc Jpn 55:1156–1162, 1982. JR Martin, P Johnson, J Tadanier, AW Goldstein. J Antibiot 33:810–815, 1980. J Tadanier, R Hallas, JR Martin, M Cirovic, RS Stanaszek. Carbohydr Res 92:191– 205, 1981. H Sano, Y Mori, K Shirahata. J Antibiot 34:474–477, 1981. J Tadanier, R Hallas, JR Martin, RS Stanaszek. Tetrahedron 37:1309–1316, 1981. JR Martin, P Johnson, J Tadanier, M Cirovic, RS Stanaszek. J Antibiot 35:46–57, 1982. J Tadanier, R Hallas. J Antibiot 36:256–266, 1983. PA Lartey, DJ Grampovnik. J Antibiot 33:1071–1076, 1980. J Tadanier, JR Martin, P Kurath, AW Goldstein, P Johnson. Carbohydr Res 79:91–102, 1980. M Sato, K Mochida, S Yoshiie, Y Mori, K Shirahata, K Kitaura. J Antibiot 34:530– 535, 1981. M Sato, K Mochida, Y Mori, K Shirahata. J Antibiot 34:513–520, 1981. J Tadanier, R Hallas, JR Martin, M Cirovic, RS Stanaszek. Carbohydr Res 92:207– 218, 1981. K Shirahata, T Iida, M Sato, K Mochida. Carbohydr Res 92:168–175, 1981. J Tadanier, DA Dunnigan, JR Martin, L Freiberg, M Cirovic. J Antibiot 34:193–201, 1981. K Mochida, M Sato, S Yoshie, Y Mori, K Shirahata, K Kitaura. Carbohydr Res 95: 522–529, 1981. P Kurath, J Tadanier, P Johnson, D Grampovnik, RS Egan, RS Stanaszek, M Cirovic, WH Wahsburn, JH Leonard. J Antibiot 34:691–700, 1981. J Tadanier, R Hallas. J Antibiot 34:403–411, 1981. J Tadanier, R Hallas. J Antibiot 35:688–691, 1982. K Atsumi, E Akita, T Niida. J Antibiot 35:90–91, 1982. H Paulsen, H Schmidt-Lewerku¨hne. Liebigs Ann Chem 959–973, 1985. J Tadanier, JR Martin, AM Nadzan, P Johnson, J Holms, R Hallas, R Stanaszek, M Cirovic, D Grampovnik, AW Goldstein. Carbohydr Res 96:185–203, 1981. W Rosenbrook, JS Fairgrieve. J Antibiot 34:681–690, 1981. J Tadanier, JR Martin, R Hallas, R Rasmussen, D Grampovnik, W Rosenbrook, W Arnold, E Schuber. Carbohydr Res 98:11–23, 1981. J Tadanier, R Hallas, J Holms, LA Freiberg, D Bacino. J Antibiot 36:267–268, 1983. K Kanai, I Sakamoto, Y Miyamoto, S Ogawa, T Suami. Bull Chem Soc Jpn 60:255– 259, 1987. K Kanai, J Nishigaki, S Ogawa, T Suami. Bull Chem Soc Jpn 60:261–266, 1987. K Kanai, J Nishigaki, T Taki, S Ogawa, T Suami. Carbohydr Res 170:47–55, 1987. K Kanai, S Ogawa, T Suami. Bull Chem Soc Jpn 60:2079–2083, 1987. K Kanai, S Yokoi, S Ogawa, T Suami. Bull Chem Soc Jpn 60:1532–1533, 1987. N Sakairi, M Hayashida, H Kuzuhara. Tetrahedron Lett 28 2871–2874, 1987. N Sakairi, M Hayashida, A Amano, H Kuzuhara. J Chem Soc Perkin Trans I 1301– 1313, 1990. D Ikeda, Y Horiuchi, S Kondo, H Umezawa. J Antibiot 23:1281–1288, 1980. Y Horiuchi, D Ikeda, S Gomi, S Kondo, H Umezawa. Carbohydr Res 109:25–31, 1982. D Ikeda, S Gomi, S Kondo, H Umezawa. J Antibiot 36:331–334, 1983. D Ikeda, S Gomi, M Hamada, S Kondo, T Takeuchi. Drugs Exp Clin Res 18:205– 216, 1992.

424

Haddad et al.

242.

I Daishiro, T Masaki, K Shinichi, T Tomio. Jpn Kokai Tokkyo Koho JP 09, 278, 787 [97,278,787]. T Tsuchiya, T Torii, S Umezawa. J Antibiot 35:1245–1247, 1982. I Watanabe, K Kamiya, T Yamaguchi, T Mori. Carbohydr Res 109:47–57, 1982. N Yasuda, K Matsuda, H Tsutsumi, T Takaya. J Antibiot 38:1512–1525, 1985. T Tsuchiya, T Torii, Y Suzuki, S Umezawa. Carbohydr Res 116:277–286, 1983. T Torii, T Tsuchiya, S Umezawa. Carbohydr Res 116:289–294, 1983. P Kurath, RS Stanaszek, M Cirovic. J Antibiot 35:1338–1344, 1982. C Ludin, T Weller, B Seitz, W Meier, S Erbeck, C Hoenke, R Krieger, M Keller, L Knothe, K Pelz, A Witmer, H Prinzbach. Leibigs Ann Chem 291–316, 1995. S Erbeck, XF Liang, D Hunkler, R Krieger, H Prinzhach. Eur J Org Chem 9:1935– 1948, 1998. N Rolland, G Vass, J Cleophax, A-M Sepulchre, SD Gero. Helv Chim Acta 65:1627– 1636, 1982. T Suami, S Ogawa, T Toyokuni. Chem Lett 611–612, 1983. S Ogawa, T Toyokuni, Y Hattori, T Nose, T Suami. Carbohydr Res 146:167–173, 1986. S Ogawa, Y Shibata. Carbohydr Res 170:116–123, 1987. B Junge, F Heiker, J Kurz, L Mu¨ller, DD Schmidt, C Wunsche. Carbohydr Res 128: 235–268, 1984. Y Shibata, S Ogawa. Carbohydr Res 189:309–322, 1989. Y Shibata, S Ogawa, T Suami. Carbohydr Res 200:486–492, 1990. S Ogawa, Y Shibata, Y Kosuge, K Yasuda, T Mizukoshi, C Uchida. J Chem Soc Chem Commun 1387–1388, 1990. Y Shibata, Y Kosuge, T Mizukoshi, S Ogawa. Carbohydr Res 228:377–398, 1992. S Ogawa, H Suguzaki. Chem Lett 1977–1980, 1986. S Ogawa, Y Nakamura. Carbohydr Res 226:79–89, 1992. S Ogawa, C Uchida, Y Shibata. Carbohydr Res 223:279–286, 1992. JC McAuliffe, RV Syick. Aust J Chem 50:219–224, 1997 JC McAuliffe, C Joseph, VR Stick, D Mutthew, G Tilbrook, GA Watts. Aust J Chem 51:91–95, 1998. T Yamagishi, C Uchida, S Ogawa. Chem Eur J 1:634–636, 1995. C Uchida, T Yamagishi, S Ogawa. J Chem Soc Perkin Trans 1 589–602, 1994. Y Kobayashi, H Miyazaki, M Shiozaki. J Org Chem 59:813–822, 1994. Y Kobayashi, H Miyazaki, M Shiozaki. J Antibiot 47:932–938, 1994. A Boiron, P Zillig, D Faber, B Giese. J Org Chem 63:5877–5882, 1998. BE Ledford, EM Carreira. J Am Chem Soc 117:11811–11812, 1995. J Li, F Lang, B Ganem. J Org Chem 63:3403–3410, 1998. M Shiozaki, T Mochizuki, H Hanzawa, H Haruyama. Carbohydr Res 288:99–108, 1996. C Uchida, H Kitahashi, S Watanabe, S Ogawa. J Chem Soc Perkin Trans I 1707–1717, 1995.

243. 244. 245. 246. 247. 248. 249. 250. 251. 252. 253. 254. 255. 256. 257. 258. 259. 260. 261. 262. 263. 264. 265. 266. 267. 268. 269. 270. 271. 272. 273.

12 Synthesis of Glycosaminoglycans Bryan K. S. Yeung, Pek Y. C. Chong, and Peter A. Petillo University of Illinois at Urbana-Champaign, Urbana, Illinois

I.

INTRODUCTION

The glycosaminoglycans (GAGs) are an important family of highly functionalized, linear, and negatively charged bioactive oligosaccharides that are ubiquitous components of animal connective tissue. Except for hyaluronan and heparin, which are also found in their free form, they exist as long chains covalently bound to a protein core, giving rise to macromolecular assemblies known as proteoglycans (PGs) [1– 7]. As major structural components of PGs, GAGs play both diverse and critical roles in lymphocyte trafficking [2], inflammatory response [3], wound repair and healing [4], and smooth muscle cell migration [5], and in conferring structural stability and resistance to deformation in cartilage [6]. The presence of GAGs on the surface of cells has been described in a number of systems and explains how GAGs achieve the diversity of roles played in various biological processes [7]. Moreover, their polyanionic character makes GAGs ideal cell surface receptors that bind circulating molecules in the extracellular matrix [8]. Glycosaminoglycan polymers consist of repeating disaccharide units usually composed of a 2-deoxy-2-amino sugar and an uronic acid partner. Single GAG chains may reach weights in excess of 1 MDa (Table 1). Mutual repulsion between negatively charged carboxylate groups contributes to the rigid structure of GAGs. Sulfate groups heterogeneously situated along the chain contribute to the inherent anionic character of the polymer and are often the sites of protein interactions. Depending on the type of amino sugar found in the polymer, GAGs can be classified into two broad categories: the glucosaminoglycans, based on D-glucosamine (hyaluronic acid, keratan sulfate, heparin, and heparan sulfate), and the galactosaminoglycans, based on D-galactosamine (chondroitin 4-sulfate, chondroitin 6-sulfate, and dermatan sulfate). 425

426

Table 1

Summary of Glycosaminoglycan Structure and Occurrence

Glycosaminoglycan

Uronic acid

Hexosamine

Occurrence

N-Acetyl-D-glucosamine

Synovial fluid, vitreous body of the eye, loose connective tissue

D-Glucuronate

N-Acetyl-D-galactosamine Sulfation at C4 or C6

Cartilage and the intervertebral disc

L-Iduronate

N-Acetyl-D-glucosamine Sulfation at C4

Fibrous connective tissues (e.g., tendons, sclera, skin)

D-Galactose

N-Acetyl-D-glucosamine Sulfation at C4 or C6

KS(I): cornea, embryonic liver, lung KS(II): skeletal tissue

Heterogeneous mixture of D-glucuronate and L-iduronate

N-Acetyl-D-glucosamine Sulfation is usually incomplete and random

Heparin: found in granules of mast cells Heparan sulfate: cell surface component of many cell types

Yeung et al.

D-Glucuronate

Synthesis of Glycosaminoglycans

427

GAGs and their derivatives are widely used in biomedical applications, with heparin and heparan sulfate being the most widely used in clinical settings [9,10]. In particular, heparin’s remarkable anticoagulant activity has led to its use as an antithrombotic drug. Other biomedical applications of GAGs, including their potential as antiviral agents, are enormous. For example, a heparin decasaccharide and the polysulfonated heparin analog suramin were both reported to inhibit dengue virus infection of host cells [11]. Other viruses such as HIV also appear to be susceptible to polysulfated, negatively charged carbohydrate oligomers such as curdlan sulfate [12] and kakelokelose [13]. Curdlan sulfate is a sulfated semisynthetic polysaccharide that inhibits HIV-1 infection of human peripheral lymphocytes. Kakelokelose, a related polysulfated ␤(1,6)-mannose polymer isolated from a marine source, also displays moderate anti-HIV activity. Although no consensus on the mode of activity currently exists, it is clear that the long, negatively charged chains of the polysaccharide bind to specific domains on the viral surface proteins, thereby neutralizing entry into the host cell. With the development of potential therapies based on the control of protein–GAG interactions for modifying cell–cell interaction, viral infection, and cell growth, the chemical preparation of GAG fragments and their analogs becomes increasingly important. This chapter surveys the chemical preparation of the glycosaminoglycan oligosaccharides of hyaluronan, chondroitin sulfate, dermatan sulfate, keratan sulfate, heparin, and heparan sulfate. Despite the biological importance of these ubiquitous carbohydrate polymers, there are surprisingly few reports of the chemical syntheses of GAG oligosaccharides. The multiple functionality of GAGs provides an excellent scaffold on which structure–activity relationships can be studied, but their syntheses present an unparalleled challenge to the synthetic carbohydrate chemist. The de novo construction of these highly functionalized carbohydrates has proven difficult, and only limited synthetic methodologies, exist for their assembly. For a long time, the lack of well-defined synthetic targets had discouraged organic chemists from the chemical synthesis of GAG oligosaccharides. Only in recent decades have the structures of these biologically active carbohydrate oligosaccharides been elucidated. This, together with the development of new carbohydrate synthetic methodology that utilizes newly developed protecting groups as well as glycosylation procedures, has made it possible to synthetically access GAG fragments that allow them to be analoged for new drug development. A.

Synthetic Strategies for GAG Preparation

As a result of their high degree of functionality, GAGs represent challenging synthetic targets. All GAGs have structural similarities that form the focal point of any synthetic effort. The disaccharide repeating units are composed of either ␤-D-glucopyranosiduronic acid or ␣-L-idopyranosiduronic acid and a hexosamine residue of either ␤-D-glucosamine or ␤-D-galactosamine that is usually N-acetylated. Additionally, the linear, polymeric chains may be O-sulfated and N-sulfated to varying degrees. The order of the glycosidic bond forming events, the choice of starting amino functionality, and the timing of oxidation state adjustment of C6 all represent important considerations that must be addressed at the onset of any synthetic effort. As with any carbohydrate synthesis, multiple orthogonal protecting groups are required to allow reaction to be isolated to a particular site. The choice of protecting groups is crucial inasmuch as the final deprotection sequence may prove problematic [14].

428

Yeung et al.

Formation of the ␤(1,4)-glycosidic linkage between C4 of the uronic acid moiety and the hexosamine residue is generally more difficult than formation of the corresponding ␤(1,3)-linkage. In addition to the steric deactivation of the C4 hydroxyl group, its nucleophilicity is lowered by the electron-withdrawing properties of the C6 ester in the uronic acid derivative [15]. Therefore, historically, the ␤(1,4)linkage is formed early on in the synthetic sequence to yield disaccharide fragments, which are then further elaborated (Scheme 1). The uronic acid residue is typically masked as a selectively protected form of D-glucose or L-idose prior to glycosidic bond formation. When glycosylation is complete, the uronic acid moiety is produced by selective C6 deprotection and oxidation. Newly described and efficient glycosylation methodologies that utilize highly reactive glycosyl donors, such as Schmidt’s trichloroacetimidate method [16], Kahne’s sulfoxide chemistry [17], and the pentenyl glycosylation technique of FraserReid [18], now offer high-yielding glycosylation strategies to offset the poor nucleophilicity of uronic esters in GAG synthesis. The synthesis of GAGs requires the installation and protection of a 2-deoxy2-amino functionality. The hexosamine unit is either N-acetyl-␤-D-glucosamine or N-acetyl-␤-D-galactosamine, with the former being more extensively employed, largely because of its ready availability and relatively low cost. Galactosamine, a rare and expensive sugar, is less attractive for use as a starting material. Consequently, several methodologies have been developed for the installation of the 2amino functionality into D-glucal and D-galactal building blocks, both of which are readily available at modest cost. Regardless of the strategy, the N-protecting group employed must survive several chemical manipulations while maintaining facile conversion to the corresponding acetamide at the conclusion of the synthesis. Presumably, the most convenient N-protecting group for GAG synthesis would be the acetate group, since this would eliminate the need for an N-deprotection and acetylation sequence at the conclusion of the synthesis. However, the N-acetate group generally imports poor solubility to the sugar; therefore, a range of alternative N-protecting groups including phthalamide [19], azide [20], trichloroacetyl [21], trichloroethoxycarbonyl [22], and benzenesulfonamide [23] are currently employed. The widely used 2-deoxy-2-azido group has been accessed by the azidonitration [24] of D-glucal or D-galactal. In general, this route is more attractive for the preparation of N-acetyl galactosamine derivatives because it produces predominantly two isomers owing to the strong preference for the C2 azido group to exist in the equatorial orientation (presumably due to the sterically disfavored axial approach). On the other hand, azidonitration of D-glucal gives rise to four isomers: 1:1 mixtures of both the gluco- and mannoazidonitrate derivatives (Scheme 2). In either case, the C1 nitrate can then be transformed into a variety of functional groups such as halides, acetates, or methyl ethers. Additionally, the azide is stable under a wide range of glycosylation conditions and can readily be reduced and acetylated at the conclusion of the synthesis. The iodosulfonamidation method pioneered by Danishefsky and coworkers offers an alternative method of introducing a 2-deoxy-2-amino moiety into D-glucal or D-galactal scaffolds [25–29]. Briefly, trans-diaxial addition of a glycal is achieved by treatment with an iodonium ion source (sym-collidine iodonium perchlorate) followed by trapping with an arylsulfonamide (Scheme 3). The active glycosylating intermediate formed under basic conditions is believed to be the 1,2-sulfonylaziri-

Synthesis of Glycosaminoglycans

Scheme 1

A typical retrosynthetic analysis for GAGs.

429

430

Scheme 2

Yeung et al.

The synthesis of 2-deoxy-2-amino glucosides from glycals.

dine, which is an extremely powerful electrophile that promotes the ␤-attack of a nucleophile at the anomeric carbon. Chain elongation can be achieved if the nucleophile is another glycal or a suitably differentiated glycosyl acceptor. Several procedures for the conversion of the 2-sulfonamide to the free amine have been reported and depend on the arylsulfonamide employed. Benzenesulfonamide can be converted to the corresponding amine by treatment with excess sodium in ammonia [26]. When 9-anthracenesulfonamide is used, conversion to the free amine can be achieved by treatment with thiophenol, 1,3-propanedithiol, or diisopropylethylamine. The milder reaction conditions required to cleave 9-anthracenesulfonamide derivatives are more likely to be compatible with other functional groups on the carbohydrate [30]. Since D-galactose is the C4 epimer of D-glucose, the selective inversion of the configuration at C4 on D-glucosamine derivatives provides efficient access to the corresponding D-galactosamine derivatives (Scheme 4). In this way, construction of GAG oligosaccharides may incorporate D-glucosamine units, which can then be inverted at C4 to afford the corresponding galactosaminoglycan. This C4 inversion has been achieved by formation of the C4 triflate followed by nucleophilic displacement

Scheme 3

Iodosulfonamidation of glycals to form 2-deoxy-2-amino glucosides.

Synthesis of Glycosaminoglycans

Scheme 4

431

Epimerization of D-glucose to D-galactose.

[31b,32] and also by oxidation to the ketone followed by stereoselective reduction [33]. II.

HYALURONAN (HYALURONIC ACID)

Meyer and Palmer first isolated hyaluronan (hyaluronic acid) in 1934 from the vitreous body of bovine eyes, and its structure was elucidated by a combination of chemical and enzymatic methods [31]. Hyaluronan (HA) is a linear, unbranched repeating polymer of 2-acetamido-2-deoxy-D-glucopyranose (GlcNAc) linked ␤(1,4) to D-glucuronic acid (GlcUA). The disaccharide repeating units are linked ␤(1,3) to form the HA chain (Fig. 1). HA has been found in all mammalian tissues including the extracellular matrix [32,33], connective tissue, and synovial fluid [34]. Its intrinsic viscoelasticity confers mechanical properties to cartilage, where it interacts with proteoglycans to produce aggregates [35] that enable cartilage to resist compressive loads. Loss or collapse of this construct reduces the ability of the surrounding tissue to withstand mechanical stress, which in turn may accelerate the breakdown of cartilage [36]. Hyaluronan is becoming increasingly important in biomedical applications as more biological functions for the polymer are discovered. HA regulates such biological processes as cellular proliferation [37], cell–cell recognition, and cell migration and cell adhesion [38]. High concentrations of hyaluronan oligosaccharides have been shown to suppress the initial vascularization of blood vessels [39], while lower concentrations of low molecular weight HA fragments, produced by enzymatic cleavage, can stimulate the formation of blood capillaries [40]. This dual role and its nonimmunogenic properties have sparked interests in hyaluronan as a potentially

Figure 1

Polymeric HA, its degradation product, and methyl hyalobiuronic acid.

432

Yeung et al.

useful antitumor drug [41]. HA is found in bacteria and can be produced in larger quantities than can be achieved by extraction methods [42]. Historically, the chemical preparation of small HA fragments aided in the structure elucidation of HA degradation products as well as providing units that can be used to probe the biological pathways and activities of HA. A step toward this goal was the synthesis of an HA disaccharide fragment, hyalobiuronic acid, which was repeatedly isolated as a degradation product of acid hydrolysis [43]. Identified as 2deoxy-2-amino-3-O-(␤-D-glucopyranosyluronic acid)-␣-D-glucose (II.2), the compound most likely existed as a mixture of ␣ and ␤ anomers. The chemical preparation of II.2, reported simultaneously by Flowers and Jeanloz [44] and Takanashi et al. [45] in 1962, employed mercury-mediated Koenigs–Knorr glycosylation methodology. Subsequently, Jeanloz and Flowers reported the synthesis of methyl hyalobiuronic acid (II.3) by condensation of 1-bromo-2,3,4,6-tetra-O-acetyl-␣-D-glucopyranoside (II.4) with II.5 in the presence of mercuric cyanide [Hg(CN)2], followed by deacetylation to produce II.6 (Scheme 5). Removal of the 4,6-O-benzylidene with acid afforded II.7, methyl 2-acetamido-2-deoxy-3-O- (␤-D-glucopyranosyl)-␣-D-glucopyranoside, in 35% yield. Attempts to selectively oxidize C6 of the D-glucose moiety in II.7 to the corresponding carboxylic acid with platinum oxide proved unsuccessful. The inability to effect selective oxidation of C6 on the D-glucose moiety led to the use of II.8, methyl(2,3,4-tri-O-acetyl-␣-D-glucopyranosyl uronate) bromide, as an alternate glycosyl donor (Scheme 6). Glycosylation with II.5 in the presence of Hg(CN)2 afforded disaccharide II.9 in 54% yield, and subsequent removal of the benzylidene followed by acetylation gave the hexa-acetyl derivative II.10. Treatment of II.10 with lithium borohydride afforded 54% of a product identical to II.3. Warren and coworkers reported the synthesis of the protected tetrasaccharide of HA with a ␣-D-glucose derivative at the reducing end [46]. Their strategy utilized the glycosylation of two disaccharides, which were both derived from a common disaccharide precursor. Starting from the known glycosyl bromide II.11 [47], silver triflate–collidine mediated coupling of 4-penten-1-ol in CH2Cl2 afforded the acetylated pentenyl glycoside II.12 (Scheme 7). Deacetylation under Zemple´n conditions followed by treatment with 2,2-dimethoxypropane furnished the 4,6-O-isopropyli-

Scheme 5

The initial attempt to synthesize hyalobiuronic acid.

Synthesis of Glycosaminoglycans

Scheme 6

433

The use of a glucuronic acid derivative to synthesize hyalobiuronic acid.

dene compound II.13. Subsequent bromoacetylation of the 3-OH gave the fully protected glycoside donor II.14 in 55% yield over two steps. Glycosyl acceptor II.15 was prepared by benzoylation of methyl 4,6-O-benzylidene-␣-D-glucopyranoside, removal of the benzylidene, and selective bromoacetylation at C6 of the resulting diol II.16. Glycosylation of II.15 with II.14 was carried out in the presence of N-iodosuccinamide (NIS) and a catalytic amount of silver triflate (AgOTf), to give the ␤(1,4)-linked disaccharide II.16 in 68% yield (Scheme 8). The conversion of II.16 into the glycosyl acceptor II.17 was achieved by treatment with thiourea in methanol [48], followed by reinstallation of a bromoacetyl moiety at C6. Alternatively, conversion of II.16 into glycosyl donor II.18 was ef-

Scheme 7

Preparation of monomers used in the Warren synthesis of HA.

434

Scheme 8

Yeung et al.

Warren synthesis of the fully protected HA tetrasaccharide.

fected by treatment with dichloromethylmethyl ether and freshly fused zinc chloride [49]. Donor II.18 proved to be fairly unstable, so after filtration of the insoluble salts, the crude syrup was dried azeotropically with toluene and used immediately in the next step. The target tetrasaccharide II.19 was obtained in 41% yield with silver triflate mediated coupling of II.17 and II.18 in collidine. The anomeric configuration of II.19 was determined by 1H NMR coupling constants, which indicated the presence of a single ␣ and three ␤ linkages. Ogawa and coworkers have synthesized a series of hyaluronan fragments ranging from the di- to the tetrasaccharide with both N-acetylglucosamine and glucuronic acid at the reducing ends. The first HA target described was the tetrasaccharide II.20 with N-acetylglucosamine at the reducing end [19]. The strategy employed the use of two glycosyl donors II.21 and II.22, and a glycosyl acceptor II.23 (Scheme 9). Compound II.23 was prepared by glycosylation of donor II.24 with II.25 using trimethylsilyl triflate (TMSOTf) as a promoter to afford 87% of the corresponding disaccharide II.26. De-isopropylidenation followed by acetylation and removal of allyloxycarbonyl (Aloc) group afforded acceptor II.23 in 82% yield over three steps. Condensation of II.23 with donor II.22 in the presence of boron trifluoride etherate (BF3⭈Et2O) gave the corresponding trisaccharide II.24 in 88% yield (Scheme 10). Deprotection of the allyloxycarbonyl afforded the trisaccharide acceptor II.25. Tri-

Synthesis of Glycosaminoglycans

Scheme 9

The Ogawa synthesis of the HA tetrasaccharide with N-acetylglucosamine at the reducing end. 435

436

Scheme 10

Yeung et al.

Deprotection of the HA tetrasaccharide with N-acetylglucosamine at the re-

ducing end.

methylsilyl triflate promoted glycosylation of II.25 with donor II.21 gave the tetrasaccharide II.26 in 87% yield. Following hydrolysis of the isopropylidene and acetylation, removal of both levulinoyl groups furnished diol II.27. Oxidation of the primary alcohols to the corresponding diacid II.28 was achieved in two steps in 86% overall yield. Finally, deacylation of II.28 with methylamine in methanol followed by selective N-acetylation afforded the target tetrasaccharide II.20 in 82% yield over the two steps. Ogawa and coworkers also reported the synthesis of the corresponding HA tetrasaccharide having a glucuronic acid at the reducing end [50]. The strategy employed the prior formation of two ␤(1,4)-linked disaccharides, followed by coupling of these disaccharides through the ␤(1,3)-linkage to produce the target tetrasaccharide. The synthesis describes the preparation of two separate, orthogonally protected disaccharide units and highlights the use of Schmidt’s trichloroacetimidate glycosylation methodology. Construction of the target tetrasaccharide utilized the monomer units II.29 [51], II.30 [19], and II.31 (Scheme 11) [52]. Stereocontrolled glycosylation of II.30 with II.29 in the presence of boron trifluoride etherate (BF3⭈Et2O) afforded the corresponding disaccharide II.32 in 81% yield. Subsequent conversion of the methoxy-

Synthesis of Glycosaminoglycans

Scheme 11

The Ogawa synthesis of the HA tetrasaccharide with glucuronic acid at the reducing end.

437

438

Yeung et al.

phenyl group into the trichloroacetimidate gave the glycosyl donor II.33 in 88% yield over two steps. Similarly, glycosylation of II.30 with II.31 under the same conditions produced the corresponding ␤(1,4) dimer II.34 in 81% yield. Following this, treatment with tetrakis(triphenylphosphine)palladium ((Ph3P)4Pd) [53] and morpholine unblocked the 3⬘-OH and produced the disaccharide acceptor II.35 in 88% yield. Trimethylsilyl triflate promoted coupling of II.33 with II.35 gave, after acid hydrolysis of the isopropylidene, tetrasaccharide II.36 in 77% yield (Scheme 12). Compound II.36 was acetylated, and removal of the levulinoyl groups with hydrazine acetate unmasked the 6-OH on each of the glucuronic acid precursors. A Swern oxidation of the two primary hydroxyl groups was achieved in two steps to give the diacid II.37 in 76% overall yield. Finally, following deacylation of II.37 with methylamine in methanol, selective N-acetylation afforded the target tetrasaccharide II.38. Using the strategy developed and described by Ogawa and coworkers in the preparation of tetrasaccharides II.20 and II.38, the methodology was expanded to an additional set of HA-related di-, tri-, and tetrasaccharides having N-acetylglucosamine at the reducing end [54]. For the synthesis of II.39, II.40, and II.41, synthons II.42–II.46 were employed. Imidates II.42 and II.45 are precursors for the D-glucuronic acid moiety at the nonreducing terminal and internal position, respectively. Similarly, II.43, II.44, and II.46 [52] are precursors for the N-acetyl-D-glucosamine residue (Fig. 2). The preparation of each of the synthons is detailed and fully described by Ogawa and coworkers. The preparation of disaccharide II.39 is outlined in Scheme 13. Condensation of II.42 with II.43 using TMSOTf as a promoter afforded disaccharide II.47 in 81% yield. Treatment with aqueous TFA followed by acetylation replaced the isopropylidene moiety with two acetate groups, and subsequent de-levulinoylation with hydrazine acetate gave the primary alcohol II.48. A Swern oxidation of the primary

Scheme 12 end.

Deprotection of the HA tetrasaccharide with glucuronic acid at the reducing

Synthesis of Glycosaminoglycans

Figure 2

439

Other HA fragments prepared by Ogawa and coworkers.

hydroxyl group was carried out in two steps with oxalyl chloride and dimethyl sulfoxide [55], followed by treatment with sodium chlorite [56] to give II.49 in 70% yield. Subsequent treatment with methylamine followed by selective N-acetylation afforded a disaccharide with an O-acetyl group present, indicating that acetylation

Scheme 13

The Ogawa synthesis of the HA disaccharide.

440

Scheme 14

Yeung et al.

The Ogawa synthesis of the HA trisaccharide.

was not selective at the 2-deoxy-2-amino position. Consequently, addition of sodium methoxide in methanol was required to produce the desired disaccharide II.39 (65%). The preparation of the trisaccharide II.40 was carried out by condensation of imidate II.45 with acceptor II.43 in the presence of TMSOTf to afford disaccharide II.50 in 87% yield (Scheme 14). Conversion of II.50 into glycosyl acceptor II.51 was achieved by de-isopropylidination and acetylation followed by removal of the allyloxycarbonyl group. Glycosyl acceptor II.51 was subsequently condensed with imidate II.44 in the presence of BF3⭈Et2O to give the trisaccharide derivative II.52 in 81% yield. The deprotection, oxidation, and N-acetylation sequence was achieved as described for II.39 to afford II.40 in 66% yield over four steps. The preparation of tetrasaccharide II.41 is outlined in Scheme 15. Boron trifluoride etherate mediated glycosylation of II.51 with imidate II.46 produced the corresponding trisaccharide (88%), which was readily converted to the glycosyl acceptor II.53 upon removal of the allyloxycarbonyl group. Under TMSOTf-catalyzed

Scheme 15

The Ogawa synthesis of the HA tetrasaccharide.

Synthesis of Glycosaminoglycans

441

conditions, acceptor II.53 and imidate II.42 were coupled to produce the fully protected tetrasaccharide II.54 in 87% yield. De-isopropylidenation, acetylation, and delevulinoylation provided the diol II.55. Conversion of II.55 into target tetrasaccharide II.41 was performed as described for the di- and trisaccharides. As demonstrated by Ogawa and coworkers, the preparation of HA oligosaccharides can be achieved in a stereocontrolled and high-yielding manner (80–90% yield). Although requiring two steps, the Swern oxidation is consistently high yielding, thereby demonstrating its utility over other types of oxidation that occasionally lead to low yields and mixtures of products. Danishefsky’s iodosulfonamidation methodology offers an alternate route to 2deoxy-2-amino functionalized sugars, although the reported use of N-protecting groups is limited to benzenesulfonamides. Carter and coworkers demonstrated the use of this methodology in the construction of the protected ␤(1,4)-HA disaccharide (Scheme 16) [14]. Glycosylation of iodosulfonamide II.56 and II.57 in the presence of lithium tetramethylpiperidide (LTMP) and silver triflate (AgOTf) afforded disaccharide II.58 in 51% yield. Subsequent removal of the 4-methoxybenzyl ether with CAN followed by Jones oxidation of C6 and esterification with CH2N2 produced II.59 in 43% overall yield. However, all attempts to remove the benzenesulfonyl and benzyl protecting groups to yield the target disaccharide resulted in incomplete deprotection and/or decomposition. The problems associated with the deprotection of II.59 prompted the use of an alternate glycosyl donor, II.60 [57]. The use of II.61 eliminates the need to oxidize C6 after glycosylation. Moreover, II.61 incorporates a phthalamide as the N-protecting group, which is more easily removed. Glycosylation of II.60 and II.61 was carried out in the presence of AgOTf and sym-collidine [58] at ⫺30⬚C and, upon warming to room temperature, provided the protected disaccharide II.62 in 94% yield. Hydrogenolysis with 30% Pd/C afforded 75% of diol II.63. Unfortunately, removal of the tert-butyl ester with formic acid produced, in addition to II.64, an unidentified side product that could not be removed. Glycosylation of II.60 with alternate acceptor II.57 via AgOTf and sym-collidine afforded the corresponding

Scheme 16

The Carter synthesis of the ␤(1,4)-HA disaccharide.

442

Yeung et al.

disaccharide that could be purified only after removal of the 4-methoxybenzyl ether to give II.65 (Scheme 17). Jones oxidation followed by esterification with diazomethane provided II.66 in 58% yield. Complete deprotection and N-acetylation were achieved in four steps and 52% overall yield to give the target disaccharide II.67. Although the final published route of II.67 by Carter and coworkers utilized classic glycosylation methodology, their prior synthetic approach involving the iodosulfonamidation methodology demonstrates its own utility in effecting glycosidic bond formation. Unfortunately, the subsequent deprotection of the aromatic sulfonamides usually requires strongly reducing conditions that are incompatible with other functionalities on the GAG backbone. Deprotection of II.59 with Na0/NH3 led to less than 10% of the amine, while the majority of the resulting mass balance consisted of decomposed monomer fragments. It is apparent that the success of Danishefsky’s iodosulfonamidation methodology in GAG synthesis depends on the ability to remove the benzenesulfonamide group under mild conditions. Vedejs and Lin reported the use of SmI2 for the deprotection of arenesulfonamides, citing excellent yields without epimerization [59]. Based on these results, Hill and coworkers investigated the application of SmI2mediated deprotection of arylsulfonamides in the N-tosyl-2-deoxy-2-amino- and Nsulfonyl-2-deoxy-2-amino-glycosides, II.68 and II.58, made by the iodosulfonidation methodology (Scheme 18) [23]. Synthesis of the protected ␤(1,4)-hyaluronan disaccharide II.58 was achieved as previously reported in 38% yield and the protected ␤(1,3)-hyaluronan disaccharide II.68 was prepared by condensation of II.56 and II.69 in the presence of LTMP and AgOTf in 82% yield. Cleavage of the N-sulfonyl bond to the corresponding free amines (II.70 and II.71), achieved by using SmI2 and 1,3-dimethylpropyleneurea (DMPU), occurred in 48 and 60% yield, respectively. These reductions took longer than those reported by Vedejs (2–3 days vs. 24 h) and, in general, the phenyl sulfonamides are reduced more quickly than the corresponding tosylamides. Fukuyama and coworkers have found 2- and 4-nitrobenzenesulfonamides to be efficient N-protecting groups for both primary and secondary amines, undergoing facile deprotection with thiophenol or mercaptoacetic acid [60]. These sulfonamides may find utility in GAG synthesis as N-protecting groups. The preparation of the largest synthesized fragments of HA was reported by Blatter and Jacquinet [21]. The tetra- (II.72), hexa- (II.73), and octasaccharides (II.74) were prepared, with each fragment containing a methyl ␤-D-glucopyranosi-

Scheme 17

Deprotection sequence to the ␤(1,4)-HA disaccharide.

Synthesis of Glycosaminoglycans

Scheme 18

443

Iodosulfonamidation and deprotection with SmI2.

duronic acid residue at the reducing end. The approach differs from the earlier preparations of HA in that a direct coupling at C4 of D-glucuronic acid derivatives was used. The poor nucleophilicity of the coupling units was offset by the efficiency of the trichloroacetimidate glycosylation methodology. The synthesis of targets II.72, II.73, and II.74 was achieved from precursors II.75, II.76, and II.77, which were, in turn, constructed from disaccharide units II.78 and II.79 (Scheme 19). The key to this strategy was the synthesis of a common dimeric building block, II.80, which could be converted into both the glycosyl donor (II.78) and the acceptor (II.79). Disaccharide II.80 was obtained by coupling glycosyl donor (II.81) and acceptor (II.82) in the presence of TMSOTf (89% yield). Conversion of II.80 into imidate II.78 was achieved by removal of the 4-methoxyphenyl group with CAN to form the hemiacetal, followed by treatment with trichloroacetonitrile and DBU. Disaccharide II.78 was converted into the corresponding methyl glycoside (II.83) by condensation with methanol; however, the reaction proved to be irreproducible (60–90% yield). Consequently, an alternate acceptor (II.84) was employed and glycosylated with imidate II.81 to give disaccharide II.83 in 91% yield (Scheme 20). Removal of the chloroacetate (ClAc) group with thiourea provided acceptor II.79 in 96% yield. With II.78 and II.79 in hand, condensation in the presence of TMSOTf produced the tetrasaccharide II.75 in 87% yield. Removal of the trichloroacetyl group gave tetrasaccharide II.85. Further glycosylation of II.85 with imidate II.78, as described for the preparation of II.75, afforded the hexasaccharide derivative II.76 in 93% yield. Subsequent deprotection of II.76 gave alcohol II.86, which could be further condensed with imidate II.78 to afford octasaccharide II.77 in 93% yield, which upon dechloroacetylation gave II.87. Deprotection of II.85, II.86, and II.87 was carried out as follows. Conversion of the N-trichloroacetyl groups to the corresponding N-acetyl groups was achieved by treatment with tributylstannane and azoisobutylnitrile (AIBN) to give the corre-

444

Yeung et al.

Scheme 19

Retrosynthesis of the tetra-, hexa-, and octasaccharides of HA.

Scheme 20

Preparation of the tetra-, hexa-, and octasaccharides.

Synthesis of Glycosaminoglycans

445

sponding acetamides II.88–II.90 in 88, 91, and 92% yield, respectively. Treatment of II.88–II.90 with aqueous acetic acid followed by saponification with aqueous sodium hydroxide afforded the target oligosaccharides II.72–II.74 , as their sodium salts in 80–83% overall yields. Pursuing our interest in probing the intramolecular hydrogen bonding network and the bond mobilities of the glycosidic linkages in hyaluronan, our group synthesized two complementary HA trisaccharides, II.91 and II.92 (Fig. 3) [61]. These trisaccharides represent the smallest fragments that contain all the structural features of polymeric hyaluronan. The synthesis of II.91, having N-acetylglucosamine at the reducing end, required the use of monomers II.93, II.94, and II.95 (Scheme 21). Condensation of sulfoxide II.93 and II.94 in the presence of Tf2O produced the corresponding ␤(1,3)disaccharide II.96. Selective ring opening of the 4-methoxybenzylidene with sodium cyanoborohydride and TFA [62] revealed the 4-OH; however, all attempts to glycosylate the resulting alcohol with imidate II.95 were unsuccessful. Presumably the steric bulk of the pivaloyl ester at C3 in addition to the low reactivity of the 4-OH precluded the formation of the glycosidic bond. Alternatively, disaccharide acceptor II.97 could be readily obtained from II.96 by conversion of the pivaloyl esters to benzyl ethers followed by regioselective ring opening to reveal the 4-OH. Subsequent glycosylation of II.97 with imidate II.95, using TMSOTf as a catalyst, provided the fully protected trisaccharide II.98 in 86% yield. Reduction of both the trichloroethoxycarbonyl (troc) carbamate and the azide and subsequent conversion to the corresponding acetamido groups was carried out in one pot by treatment with cadmium dust in acetic acid/DMF [63], followed by reduction of the azide with thiol acetic acid (Scheme 22). Subsequent treatment with acetic anhydride provided the diacetamido derivative II.99 in 60% yield over three steps. Removal of the p-methoxybenzyl ether followed by oxidation of C6, using a catalytic amount of 2,2,6,6tetramethyl-1-piperidinyloxy (TEMPO) in the presence of sodium hypochlorite [64] (NaOCl), provided the fully functionalized trisaccharide (II.100) in 51% yield over two steps. Finally, hydrogenolysis using Pearlman’s catalyst followed by saponification with lithium hydroxide provided the target trisaccharide II.91 in 13 steps and 16% overall yield. The complementary trisaccharide, II.92, having glucuronic acid at the reducing end, was prepared from monomers II.95, II.101, and II.102 (Scheme 23). Condensation of methyl glycoside II.99 with imidate II.95, using TMSOTf as a catalyst, afforded the corresponding disaccharide II.103 in 93% yield. Saponification of II.103 followed by benzylidenation provided disaccharide acceptor II.104. However, saponification under Zemple´n conditions (methanolic sodium methoxide) resulted in the conversion of the troc group to the corresponding methyl carbamate. Consequently, a milder deacetylation method that used a guanidine/guanidinium nitrate solution [65] was adopted, and near-quantitative deacetylation was achieved in 20 min. Condensation of II.104 with the glycosyl donor II.102 with TMSOTf provided the fully protected trisaccharide II.105 in 87% yield. Reduction of the troc carbamate to the free amine was accomplished with cadmium in AcOH/DMF followed by acetylation to provide II.106. Removal of the p-methoxybenzyl ether and subsequent saponification yielded the pentaol, II.107 (Scheme 24).

446

Figure 3

The shortest HA fragments that contain structural features of polymeric HA.

Yeung et al.

Synthesis of Glycosaminoglycans

Scheme 21

Preparation of the HA trimer with N-acetylglucosamine at the reducing end. 447

448

Yeung et al.

Scheme 22

III.

The deprotection sequence for the HA trimer.

CHONDROITIN SULFATE

Chondroitin sulfates occur in tissues attached to proteoglycans and form a family of polysaccharides that differ in the degree and position of sulfation. Cartilage and the intervertebral disc contain the highest concentration: as much as 10% of the wet weight can be due to chondroitin sulfate [66]. Chondroitin sulfate is a repeating copolymer of D-glucuronic acid (GlcUA) and 4-O- or 6-O-sulfated 2-acetamido-2deoxy-D-galactose (GalNAc). Three different sulfated variants of chondroitin are known: the 4-O-sulfated, the 6-O-sulfated, and the 4,6-di-O-sulfated variants (Fig. 4). Several strategies have been developed and implemented to synthesize various oligomers of chondroitin sulfate. Jacquinet reported the synthesis of the methyl glycosides of chondroitin disaccharides (III.1, III.2, III.7, and III.8), which represent the four possible repeating units of chondroitin 4-O- and 6-O-sulfate [20]. The 4-O-sulfate (III.1) and 6-O-sulfate (III.2) of methyl 2-acetamido-2-deoxy3-O-(␤-D-glucopyranosyluronic acid)-(1,3)-␤-D-galactopyranoside were obtained by condensation of glycosyl acceptors III.3 and III.4 with donors III.5 and III.6 (Scheme 25). The corresponding ␤(1,4)-linked disaccharides III.7 and III.8 were obtained by condensation of glycosyl donors III.9 or III.10 and III.11 or III.12 with acceptors III.13 and III.14 (Scheme 26). The design of monomer building blocks for sulfated GAGs requires that the 4-OH and 6-OH of the galactosamine unit be orthogonally protected to facilitate regiospecific monosulfation. This resulted in the need for an increased number of protecting group transformations. Glycosyl acceptors III.3 and III.4, used in the preparation of the ␤(1,3)-disaccharides, were obtained from methyl 2-azido-2-deoxy-␤-D-galactopyranoside [67]. Selective 3,6-di-O-silylation followed by treatment with benzyl bromide led to the installation of the benzyl ether at C4. Desilylation occurred in situ under the benzylation reaction conditions to produce III.15 and subsequent 6-O-acetylation of III.15 with 1-acetylimidazole produced III.3 in 78% yield (Scheme 25). The preparation of III.4, carried out as reported earlier by Jacquinet and Sina¨y [67], was obtained from III.16 by regioselective ring opening of the isopropylidene to unmask the 3-OH. Glycosyl donors III.5 and III.6 were derived from the corresponding acetylated derivative III.17. Treatment of III.17 with dibutyltin oxide [68] afforded the hemi-

Synthesis of Glycosaminoglycans

Scheme 23

Preparation of the HA trimer with glucuronic acid at the reducing end.

449

450

Scheme 24

Yeung et al.

The deprotection sequence of the HA trimer.

acetal, which was then converted to the corresponding ␤-imidate III.5. Alternative treatment of III.17 with hydrogen bromide gave the glycosyl bromide III.6. Condensation of III.3 with III.5 by means of TMSOTf gave III.18 (75% yield) as the precursor to the 4-O-sulfated chondroitin disaccharide. Similarly, condensation of III.3 with III.6 by means of AgOTf as a promoter also afforded III.18 in 90% yield. The disaccharide precursor to the 6-O-sulfated chondroitin disaccharide (III.19) was also obtained by the use of either glycosyl donor. Glycosylation of III.4 with imidate III.5 or bromide III.6 gave the desired disaccharide III.19 in 72 or 80% yield, respectively. Interestingly, for these specific substrates, the Koenigs– Knorr glycosylations for the preparation of III.18 and III.19 proceeded in higher yield than the corresponding trichloroacetimidate glycosylations.

Figure 4

The four possible repeating units of chondroitin sulfate.

Synthesis of Glycosaminoglycans

Scheme 25

The Jacquinet synthesis of the 4- and 6-O-sulfated chondroitin disaccharides.

451

452

Scheme 26

Yeung et al.

Azidonitration for the preparation of chondroitin sulfate disaccharides.

Simultaneous reduction of the benzyl ethers and the azide proved unsatisfactory in the presence of methyl esters. An alternate procedure utilized thioacetic acid to reductively N-acetylate the azide. Thus, treatment of III.18 and III.19 with thioacetic acid afforded the corresponding 2-deoxy-2-acetamido derivatives III.20 and III.21, respectively. Hydrogenolysis followed by sulfation with the sulfur trioxide–trimethylamine complex gave the corresponding 4- and 6-O-sulfated disaccharides III.22 and III.23. Final saponification of III.22 and III.23 with aqueous sodium

Synthesis of Glycosaminoglycans

453

hydroxide in methanol gave the target disaccharides III.1 and III.2 in 83 and 86 yield, respectively. Four different galactosamine imidate donors were designed and used in the preparation of the target ␤(1,4)-linked disaccharides. These glycosyl donors, III.9– III.12, were all prepared by azidonitration of D-galactal (Scheme 26). Selective 3,6di-O-tert-butyldimethylsilylation of D-galactal followed by treatment with methoxyethoxymethyl chloride (MEMCl) and benzyl bromide gave III.24. Azidonitration of III.24 proceeded cleanly, and the product was subsequently converted to the corresponding hemiacetal III.25 with sodium nitrite [69]. The 4-(2-methoxyethoxymethyl) group in III.25 was replaced by an acetate group by treatment with aqueous TFA followed by acetylation. Subsequent l-O-deacetylation with benzylamine [70] gave III.26. Conversion of III.25 and III.26 to the corresponding imidates III.9 and III.10 was achieved in the usual manner. Similarly, preparation of the 6-O-sulfated galactosamine donors III.11 and III.12 utilized the selective 6-O-silylation of D-galactal followed by benzylation to afford alcohol III.27. Subsequent treatment with either MEMCl or benzoyl chloride provided the corresponding galactal derivative, III.28 or III.29. Azidonitration followed by hydrolysis to the hemiacetal and conversion into the corresponding trichloroacetimidate afforded donors III.11 and III.12. Glycosyl acceptors III.13 and III.14 were obtained by 6-O-selective chloroacetylation and tert-butyldimethylsilylation of methyl 2,3-di-O-benzyl-␤-D-glucopyranoside, respectively. Glycosylations involving acceptors III.9–III.12 and donors III.13 and III.14 were carried out by using BF3⭈Et2O as a promoter and are summarized in Table 2. Most coupling products could be isolated only after deprotection of the 6-OH on the glucose unit. The reported yields for the glycosylations and the stereoselectivity of the (1,4)-linkages are moderate, at best. Lower selectivities observed may be due to the lack of a participating group at C2 on the glycosyl donors. Of the four different disaccharides prepared, only the deprotection/oxidation sequence of disaccharides III.32 and III.34 led to targets III.7 and III.8, respectively. All attempts at removing the methyl ether protecting groups in disaccharides III.31 and III.33 were unsuccessful. Preparation of the target disaccharide III.7 is described as follows. Transformation of azide III.32 to the N-acetate III.35 was achieved in 81% yield by sodium

Table 2

Glycosylation Conditions for the Chondroitin Disaccharides

Donor

Acceptor

Solvent

Product

Yield (%)

␣/␤ Ratio

III.9 III.9 III.9 III.10 III.10 III.11 III.11 III.12 III.12 III.12

III.13 III.13 III.14 III.13 III.14 III.13 III.14 III.13 III.14 III.14

1,2-Dichloroethane Toluene Toluene Toluene Toluene Toluene Toluene Toluene 1,2-Dichloroethane Toluene

III.31 III.31 III.31 III.32 III.32 III.33 III.33 III.34 III.34 III.34

30 66 40 50 65 35 32 40 32 60

1:6 1:10 1:9 1:3 1:20 2:3 1:3 2:3 1:8 1:11

454

Yeung et al.

borohydride reduction followed by N-acetylation. Oxidation of the primary alcohol with pyridinium dichromate followed by saponification then gave III.36. Sulfation of the 4-OH on galactosamine as described above, and subsequent catalytic hydrogenation (Pd/C) afforded the target disaccharide III.7. The 6-O-sulfated derivative III.8 was prepared from disaccharide III.34, where oxidation followed by esterification with diazomethane gave the methyl ester III.37. Reductive N-acetylation, saponification, and sulfation produced derivative III.38, and subsequent hydrogenation afforded the target disaccharide III.8. More recently, Ogawa and coworkers have prepared 4-O-sulfated chondroitin di- and tetrasaccharides fragments that allow for chain elongation by the condensation of a dimeric repeating unit [71]. The key intermediate used for chain elongation was the ␤(1,4)-disaccharide III.39, which was prepared from monomer building blocks III.40 and III.41 (Scheme 27). The glycosylations were carried out by using the trichloroacetimidate methodology with BF3⭈Et2O as the promoter. Condensation of III.40 and III.41 afforded the corresponding disaccharide III.42. To aid in purification, the tert-butyldimethylsilyl group was removed by subsequent treatment of the crude disaccharide with tetrabutylammonium fluoride and acetic acid to give III.43 [72]. Oxidation of C6 to the methyl ester by means of a Swern oxidation [73] followed by esterification with diazomethane produced the key intermediate III.39 in 96% yield over three steps. Disaccharide III.39 could then be transformed into both glycosyl donor III.44 and glycosyl acceptor III.45 (Scheme 28). Treatment of III.39 with cerium ammonium nitrate gave the corresponding hemiacetal, which was easily converted to the ␣-imidate III.44 in 95% yield. Conversion of III.39 to III.45 was achieved quantitatively by simple removal of the levulinoyl group with hydrazine acetate [74]. With III.44 and III.45 in hand, glycosylation furnished the fully protected tetrasaccharide III.46 in 50% yield.

Scheme 27

The key intermediate in the Ogawa synthesis of chondroitin sulfate derivatives.

Synthesis of Glycosaminoglycans

Scheme 28

Preparation of the tetrasaccharide of chondroitin sulfate.

455

456

Scheme 29

Yeung et al.

Deprotection and sulfation of the tetrasaccharide.

Deprotection and sulfation was achieved by reduction of the azide to the corresponding acetamide with thioacetic acid [75] to afford III.47 (Scheme 29). Subsequent de-levulinoylation followed by acetylation gave III.48 in 64% yield over two steps. Reductive opening of the bisbenzylidene acetals with sodium cyanoborohydride and acid [76] afforded III.49 in 51% yield and by-product III.50, formed by acetyl migration from C3 to C4 (31%). Sulfation of III.49 was carried out with the sulfur trioxide–trimethylamine complex to give III.51 in 92% yield. Deacetylation followed by hydrogenolysis provided the target tetrasaccharide III.52. The preparation of the disaccharide III.53 from III.45 was achieved in a similar manner (Scheme 30). Transformation of the azide to the corresponding acetamide was carried out with thioacetic acid. Acetylation and reductive opening of the benzylidene unblocked the 4-OH to give III.54. Sulfation, de-esterification, and hydrogenolysis afforded the target 4-O-sulfated disaccharide III.53 in 67% yield over three steps. In 1998 Ogawa and coworkers revisited the synthesis of chondroitin sulfate and prepared the di-, tri-, and tetrasaccharides in their nonsulfated, 4-O-sulfated, 6O-sulfated, and 4,6-di-O-sulfated forms (Fig. 5). As before, disaccharide III.39 was used as the primary building block for chain elongation.

Scheme 30

Preparation of the chondroitin sulfate disaccharide.

Synthesis of Glycosaminoglycans

Figure 5

Chondroitin sulfate derivatives prepared by Ogawa and coworkers.

457

458

Yeung et al.

Chondroitin disaccharides III.56–III.59 were prepared by conversion of the azide in building block III.39 to the acetamide, which was achieved with thioacetic acid. Complete deprotection then afforded the nonsulfated disaccharide III.56. An exchange of the levulinoyl group in III.39 for the pivaloyl group to produce III.60 was utilized for the preparation of the sulfated variants (Scheme 31). Conversion of the azide to the acetamide (III.61) followed by regioselective benzylidene ring opening gave III.62. Subsequent sulfation (71%), saponification, and hydrogenolysis (88% in two steps) gave the target chondroitin 4-sulfate disaccharide (III.57). The corresponding 6-O-sulfated derivative (III.58) was prepared as follows. Starting from disaccharide III.62, acetylation followed by hydrogenolysis unmasked the 6-OH to give the corresponding alcohol with no acetate migration observed. Subsequent sulfation and saponification afforded III.58 in 87% yield over three steps. Quantitative preparation of the 4,6-di-O-sulfated disaccharide III.59 was achieved by treatment of the diol III.63 with the sulfur trioxide–trimethylamine complex. Alternatively, sulfation of III.62 followed by hydrogenolysis unmasked the 6-OH and subsequent sulfation provided III.59 in 71% yield. Chondroitin trisaccharides III.64–III.67 were obtained by coupling the disaccharide acceptor III.45 with methyl glucuronate trichloroacetimidate III.68 [77] in the presence of BF3⭈OEt to give the corresponding trisaccharide III.69. The azide was transformed into the corresponding acetamide III.70 by Lindlar reduction followed by acetylation (Scheme 32). Hydrogenation was favored over the use of thioacetic acid as it generally gave higher yields for oligomers larger than the disaccharide. Acid hydrolysis of the benzylidene acetal followed by saponification gave the nonsulfated trisaccharide (III.64). The 4-O-sulfated derivative (III.65) was prepared by reductive ring opening of III.70 to provide the 6-O-benzyl derivative III.71 in 69% yield with formation of 12% of the 4-O-benzylated product. Sulfation of the free 4-OH, saponification, and hydrogenolysis furnished the 4-O-sulfated trisaccharide III.65. The same synthetic strategy used to prepare the 6-O-sulfated disaccharide was applied to the corresponding trisaccharide. Acetylation of III.71 followed by hydrogenolysis afforded the primary alcohol III.72. Subsequent sulfation and saponification gave the target 6-O-sulfated trisaccharide III.66. The 4,6-di-O-disulfate (III.67) could not be obtained directly from the 4,6-diol. Consequently, hydrogenolysis of the 4-O-sulfated derivative III.73 unmasked the 6-OH, which was then sulfated to give target III.67 after saponification. The prolonged reaction time required for O-sulfation of the 6-OH (6 days, 71% yield) was attributed to the electronegativity of the neighboring sulfate. Tetrasaccharide derivatives (III.74–III.77 ) incorporate an internal and a terminal D-galactosamine residue with varying degrees of sulfation. The strategy for selective sulfation was identical to that used for the di- and trisaccharides. The tetrasaccharides were derived from the fully protected derivatives III.46 (Scheme 33). Reduction of the azides to the corresponding acetamides with thioacetic acid gave III.78 in 43% yield. As with the trisaccharides, higher yields were obtained when the azide was reduced in two steps: hydrogenation with Lindlar catalyst followed by acetylation (60% yield). Saponification of III.78 (88%), followed by hydrogenolysis (68%), gave the target nonsulfated tetrasaccharide III.74. The 4-O-sulfated derivative III.75 was obtained by first converting the levulinoyl ester into the corresponding privaloate, followed by regioselective opening of the benzylidene to give the desired secondary diol (III.79) in 65% yield, as well as a small amount of the 4-O-benzylated

Synthesis of Glycosaminoglycans

Scheme 31

The disaccharide synthesis. 459

460

Scheme 32

The trisaccharide synthesis. Yeung et al.

Synthesis of Glycosaminoglycans

Scheme 33

461

The tetrasaccharide synthesis.

regioisomer (15%). No acyl migration was observed when the pivaloyl ester was substituted for the corresponding acetate. Sulfation of the 4-OH, saponification, and hydrogenation furnished III.75 in 70% yield. When the regioselective sulfation of the tetraol III.80 was attempted for the chondroitin 6-sulfate tetrasaccharide, it proceeded slowly, yielding a single product after 2 days that was identified as the tetrasulfated derivative III.77. Therefore, the 6-O-sulfated derivative was obtained stepwise by acetylation of III.79 followed by hydrogenation of the two 6-O-benzyl ethers to afford the corresponding diol. Conventional sulfation and saponification afforded III.76 in 82% yield over three steps. Chondroitin differs from hyaluronan in two ways: the presence of sulfate esters and the configuration of the amino sugar (i.e., D-galactosamine instead of D-glucosamine). Since D-galactosamine is a rare and expensive starting material, it is usually prepared by the azidonitration of D-galactal as demonstrated in the earlier preparations of chondroitin and dermatan fragments. An alternate way of accessing D-galactosamine is by inversion of C4 in D-glucosamine, and indeed such conversions have been reported for both monomers [78] and neutral disaccharides [79] containing D-glucosamine. Coutant and Jacquinet used this approach to access saccharides con-

462

Yeung et al.

taining uronic acid moieties and reported the preparation of chondroitin 4-O-sulfate trisaccharide III.81 from the hyaluronic acid trisaccharide III.82 (Scheme 34) [80]. A trisaccharide derivative, in which the central D-glucosamine residue is flanked by two D-glucuronic acid residues, was chosen to test the validity of the strategy. Construction of the protected hyaluronic acid trisaccharide was achieved from the following monosaccharides: III.83, III.84, and III.85. The glucuronic acid precursor III.85 was glycosylated with glucosamine moiety III.84 by using N-iodosuccinamide (NIS) and trimethylsilyl triflate in dichloromethane to afford the corresponding disaccharide III.86 in 90% yield. In general, uronic esters are poor nucleophiles when glycosylation is at the C4 position. Presumably the ester moiety significantly reduces the nucleophilicity of the 4-OH, and as a result, necessitates the use of a D-glucose unit, where the C6 is selectively oxidized after coupling. Conversion of disaccharide III.86 into glycosyl acceptor III.87 was achieved in 92% yield by removal of the chloroacetyl ester with thiourea in pyridine–ethanol. Condensation of acceptor III.87 with imidate III.83 in the presence of trimethylsilyl triflate afforded the crystalline trisaccharide III.88 in 92% yield. Removal of the isopropylidene with aqueous acid followed by selective benzoylation of C6 gave III.82 (90%, two steps). Inversion of the configuration at C4 was carried out by treatment of III.82 with triflic anhydride in pyridine to form the 4-O-triflyl derivative followed by reaction with tetrabutylammonium nitrite, a reagent known to give the epi-hydroxyl analog [81], to afford the D-galacto product III.89 in 87% yield (Scheme 35). Transformation of the N-trichloroacetyl group to the acetamide was carried out with tributylstannane and azoisobutyronitrile [82] to give the crystalline acetamide III.90 in 92% yield. Sulfation of the free hydroxyl with the sulfur trioxide–trimethylamine complex gave 93% of III.91, which was saponified with sodium hydroxide in aqueous methanol to afford the target chondroitin 4-O-sulfate trisaccharide III.81 in 87% yield. While the syntheses of derivatives of the chondroitin sulfate disaccharides dominate the existing literature, Jacquinet and coworkers have reported the successful syntheses of the reducing disaccharides of chondroitin 4- and 6-sulfates on a multigram scale [83]. The synthetic sequence (Scheme 36) utilizes a silver triflate mediated glycosylation between bromide III.92 and benzyl glycoside III.93 to afford the disaccharide III.94 in 70% yield. After treatment with hot aqueous acetic acid, diol III.95 was obtained in 87% yield. The strategic choice of functionality in this common intermediate minimizes the number of postglycosylation transformations. Regioselective benzoylation of diol III.95 was achieved in 93% yield by treatment with benzoyl cyanide in pyridine. Consequent O-sulfation with the sulfur trioxide– trimethylamine complex followed by ion exchange chromatography afforded the sodium salt III.96 in 90% yield. Saponification of III.96 with lithium hydroperoxide and methanolic sodium hydroxide provided the disodium salt III.97 in 83% yield, which was subsequently subjected to hydrogenolysis to afford the chondroitin 4sulfate disaccharide III.98 in 97% yield. Preparation of chondroitin 6-sulfate from the common intermediate III.95 was achieved by regioselective sulfation at C6 with the sulfer trioxide–trimethylamine complex. The resulting monosulfated disaccharide III.99 was isolated in 90% yield. Saponification to the disodium salt III.100 occurred in 82% yield, and subsequent hydrogenolysis provided the chondroitin 6-sulfate disaccharide III.101 in 96% yield.

Synthesis of Glycosaminoglycans

Scheme 34

The use of C4 inversion in the preparation of the chondroitin sulfate trimer.

463

464

Scheme 35

The use of C4 inversion in the preparation of the chondroitin sulfate trimer.

Yeung et al.

Synthesis of Glycosaminoglycans

Scheme 36

Multigram-scale synthesis of chondroitin sulfate disaccharides.

465

466

Scheme 37

The combinatorial approach for chondroitin sulfate derivatives.

Yeung et al.

Synthesis of Glycosaminoglycans

Scheme 38

467

The combinatorial approach for chondroitin sulfate derivatives.

Recognizing the need for the facile generation of chondroitin sulfate disaccharides with varying sulfation patterns, Lubineau and Bonnaffe´ recently reported the use of a combinatorial approach to access eight chondroitin sulfate disaccharides (III.102 and III.103) from the key chondroitin sulfate disaccharide scaffold (III.104) bearing orthogonal protecting groups [84]. The synthesis of key disaccharide III.104 (Scheme 37) was achieved by the TMSOTf-mediated glycosylation of acceptor III.105 with trichloroacetimidate III.106 to produce disaccharide III.107 in 74% yield. Following methanolysis of the acetal- and stannylene-promoted alkylation, the disaccharide III.108 was obtained. Inversion of the configuration of C4 was achieved by Swern oxidation followed by K Selectride reduction, providing the galacto counterpart III.104, illustrating this inversion method as an effective alternative to triflate displacement. Orthogonal protecting group manipulations at C4 and C6 of benzyl glycoside III.104 followed by sulfation led to the four different sulfoforms III.109, consisting of the unsulfated, two mono-O-sulfated, and the di-O-sulfated derivatives (Scheme 38). These disaccharides were then hydrolyzed and fully deprotected to give the four sulfoforms III.102, which are unsulfated at the C2 position of the glucose moiety. Alternatively, sulfoforms III.109 were deprotected and sulfated at the C2 position of the glucose moiety prior to deprotection to afford the other four sulfoforms III.103. This combinatorial split-and-pool approach represents a powerful route to the preparation of sulfoforms of a given oligosaccharide. Additionally, it illustrates the use of O-sulfate esters as base and low-temperature, acid-stable protecting groups. IV.

DERMATAN SULFATE

Dermatan sulfate is an O-sulfated glycosaminoglycan first isolated from pigskin [85]. Structurally related to chondroitin, in that repeating disaccharide units consist of Dglucuronic acid and N-acetylgalactosamine, dermatan sulfate incorporates the C5 epimer of D-glucuronic acid, L-iduronic acid, in addition to N-acetylgalactosamine. Position and degree of sulfation are dependent on the source [86], but in general, the D-galactosamine residues are mainly sulfated at C4 and the adjacent L-iduronic

468

Yeung et al.

residue can be sulfated or not. Dermatan sulfate possesses anticoagulant, profibrinolytic, and antithrombotic properties. Although dermatan sulfate does not directly inhibit thrombin, it accelerates heparin cofactor II (HCII) mediated inhibition of the thrombin [87]. The antithrombin activity of dermatan sulfate has been narrowed down to a binding domain on HCII. Biological activity could be induced with a hexasaccharide of dermatan sulfate [88] and, more recently, a nonasaccharide [89]. Since dermatan sulfate is a hybrid polymer comprising two types of disaccharide unit, N-acetylchondrosine and N-acetyldermosine (Fig. 6), the heterogeneity of the polymer complicates biological studies that use smaller fragments as probes. Moreover, the smaller fragments generally have a lower activity with respect to the parent polymeric dermatan sulfate [87], and the longer plyanionic chains are required for the formation of the ternary complex with thrombin. Fragments of dermatan that sulfate have been synthesized include the methyl glycoside derivatives of the basic monosulfated disaccharide, a hexasaccharide containing three disulfated residues and most recently, a basic disulfated disaccharide. Jacquinet and Sina¨y have reported the preparation of the methyl ␤-glycosides of dermatan sulfate fragments starting from D-galactal and the readily available 1,6anhydro-␤-L-idopyranose [90]. This report described the syntheses of disaccharides IV.22, IV.23, and IV.26 (see below: Schemes 41 and 42) and the determination of their anomeric configurations by 1H NMR spectroscopy. In addition, the syntheses of the target monosaccharides IV.7, IV.10, and IV.11 (Scheme 39), which differ in the position and degree of sulfation, were described and their observed NMR signals were used to constitute a firm basis for the structural assignment of the target disaccharides. The model monosaccharides just listed were prepared from common precursor IV.1 (Scheme 39), which was readily obtained by azidonitration of 3,4,6-tri-O-acetylD-galactal followed by deacetylation with sodium methoxide. Treatment of IV.1 with acetone and toluene p-sulfonic acid monohydrate at room temperature led to predominant formation of the thermodynamically favored 3,4-O-isopropylidene (IV.2) in 61% yield while also producing 27% of the 4,6-O-isopropylidene derivative IV.3. The position of the isopropylidene IV.2 was verified by the use of NMR chemical shift analysis to confirm the position of the acetate group in the resultant acetylated adduct IV.4. Synthesis of the 4-O-sulfate derivative (IV.7) from IV.2 utilized a step that differentiated the 3-OH and 4-OH positions: after benzylation and de-isopropylidination of IV.2, a selective methylation at the 3-OH of diol IV.5 was achieved via a tin procedure [91] to give methyl glycoside IV.6. Conversion of the azide into

Figure 6

Typical disaccharide units that comprise dermatan sulfate.

Synthesis of Glycosaminoglycans

Synthesis of the dermatan sulfate monosaccharides.

469

Scheme 39

470

Yeung et al.

an acetamido group followed by O-sulfation with sulfur trioxide–trimethylamine and subsequent removal of the benzyl ether afforded the desired sulfate IV.7 as its sodium salt. Alternatively, kinetic isopropylidination of IV.1 was demonstrated, using 2methoxypropene in dimethylformamide and toluene-p-sulfonic acid monohydrate conditions to produce the 4,6-O-isopropylidene derivative IV.3 in 85% yield. Subsequent methylation of the 3-OH with methyl iodide followed by removal of the isopropylidene protecting group by treatment with aqueous 90% trifluoroacetic acid afforded crystalline IV.8. Catalytic hydrogenolysis (Pd/C) of IV.8 produced the corresponding amine, which was then N-acetylated to provide 90% of crystalline IV.9. Selective O-sulfation of the 6-OH on IV.9 with the sulfur trioxide–trimethylamine complex in dimethylformamide and subsequent ion exchange chromatography produced the crystalline sulfate IV.10 as its sodium salt in 86% yield. Complete 4,6-Osulfation occurred in the presence of excess sulfating agent to give the disulfate IV.11 in 85% yield. Construction of the L-idopyranosyluronic acid donor moiety began with the benzylation of 1,6-anhydro-␤-L-idopyranose to give 1,6-anhydro-2,3,4-tri-O-benzyl␤-L-idopyranose (IV.12). Acetolysis of IV.12 with acetic anhydride–trifluoroacetic acid gave 91% of IV.13 (Scheme 40). Subsequent treatment of IV.13 with methanolic sodium methoxide provided 2,3,4-tri-O-benzyl-L-idopyranose, which was monochloroacetylated with chloroacetyl chloride to give IV.14 in 84% yield, as a 5:2 mixture of ␣ and ␤ anomers. Addition of dichloromethane saturated with hydrogen chloride gave 90% of the corresponding chloride IV.15, which was immediately condensed with methyl 4-O-acetyl-2-azido-6-O-benzyl-2-deoxy-␤-D-galactopyranoside (IV.16). Compound IV.16 was prepared by regioselective ring opening [92] of a methyl orthoester (prepared from IV.5 by treatment with trimethyl orthoacetate and toluene p-sulfonic acid monohydrate). The glycosylation was carried out in the presence of silver triflate and 2,4,6-trimethylpyridine and, following O-chlorodeacetylation, afforded 58% of the ␣-linked disaccharide IV.17 and 30% of the corresponding ␤-linked disaccharide IV.18. Reduction of the azide in the ␣-linked disaccharide IV.17, by using sodium borohydride in the presence of nickel dichloride hexahydrate and boric acid, followed by N-acylation, gave IV.19 in 81% yield (Scheme 41). Oxidation of the 6-OH on the idopyranosyl moiety was achieved with chromium trioxide in acetone–sulfuric acid and gave, after deacetylation, the crystalline acid IV.20. The free acid was converted to the sodium salt and O-sulfated to afford IV.21 in 81% yield. Finally, catalytic hydrogenolysis (Pd/C) provided 83% of the target disaccharide of N-acetyldermosine (IV.22). Additionally, catalytic hydrogenolysis of IV.20 provided the nonsulfated analog IV.23 in 86% yield. Similarly, the ␤-linked disaccharide IV.18 was converted to the corresponding N-acetylated disaccharide IV.24 (81%) followed by oxidation, deacetylation, and Osulfation to give IV.25 (Scheme 42). Catalytic hydrogenolysis resulted in IV.26 in 87% yield after purification. Sina¨y and coworkers explored different approaches to the synthesis of the Nacetyldermosine disaccharide (IV.22) that resulted in the report of an improved synthetic route [93]. The strategy investigated initially used diol IV.5 (as an alternative glycosyl acceptor to IV.16) that relied on regioselective glycosylation of the more reactive equatorial 3-OH. This would eliminate the need for a deprotection sequence

Synthesis of Glycosaminoglycans

Scheme 40

Synthesis of the dermatan sulfate disaccharides.

471

472

Synthesis of N-acetyldermosine.

Yeung et al.

Scheme 41

Synthesis of Glycosaminoglycans

Scheme 42

Synthesis of N-acetylchondrosine.

473

474

Yeung et al.

prior to sulfation of the axial 4-OH. However, glycosylation of diol IV.5 with methyl (2,3,4-tri-O-acetyl-␣-L-idopyranosyl bromide) uronate [94] (IV.27) was reported to produce a low yield (16%) of desired disaccharide IV.28 (Scheme 43). The major product isolated (66%) was the intermediate orthoester IV.29 [95]. The presence of the vicinal 4-OH appeared to complicate the rearrangement of orthoester IV.29 into IV.28. All attempts to effect this rearrangement with tin tetrachloride were unsuccessful, and an alternate strategy was then employed. Glycosyl acceptor IV.30, obtained by benzylidenation of IV.1 with ␣,␣-dimethoxytoluene and camphorsulfonic acid in nitromethane, was used instead. Silver triflate mediated glycosylation of IV.30 with IV.27 in the absence of 2,3,4-trimethylpyridine afforded the disaccharide IV.31 in 54% yield. Alternatively, trimethylsilyl triflate promoted glycosylation of IV.30 with the ␣-trichloroacetimidate IV.32 (obtained from glycosyl bromide IV.27) gave an improved yield of the disaccharide IV.31 (68%). The 4-OH of the galactopyranoside unit was selectively unmasked [96] to give 70% of IV.28, and then O-sulfated to produce disaccharide IV.33 in 94% yield. Saponification of IV.33 with sodium hydroxide followed by catalytic hydrogenolysis and N-acetylation provided the target disaccharide (IV.22) as the disodium salt. Using the same synthetic approach, Sina¨y and coworkers also reported the synthesis of the second repeating copolymer of dermatan sulfate, N-acetylchondrosine (IV.37) (Scheme 44). Silver triflate mediated glycosylation of IV.30 with methyl(2,3,4-tri-O-acetyl-␣-D-glucopyranosyl bromide)uronate (IV.34) gave the corresponding disaccharide IV.35. Compound IV.35 was then converted to the target disaccharide (IV.37) as described above, in 60% overall yield. Ogawa and Goto reported the synthesis of a dermatan sulfate hexasaccharide (IV.38) that utilized trichloroacetimidate glycosylations in a regio- and stereocontrolled manner according to the retrosynthetic route shown in Scheme 45 [97]. The target hexasaccharide IV.38 is derived from benzyl-protected precursor IV.39, which is in turn derived from IV.40 through oxidation, N-acetylation, and sulfation. The central tetrasaccharide was constructed by the repeated use of the imidate disaccharide IV.42, and the nonreducing end and reducing end units utilized monosaccharides IV.41 and IV.43 [98], respectively. The preparation of glycosyl donors IV.41 and IV.42 is detailed in the corresponding reference. The linear synthesis of the hexasaccharide started at the reducing end by the TMSOTf-promoted glycosylation of IV.43 with trichloroacetimidate IV.42 to produce trisaccharide IV.44 in 86% yield (Scheme 46). Removal of the levulinoyl group with hydrazine acetate and coupling of the resultant glycosyl acceptor IV.45 with another unit of IV.42 in the presence of tert-butyldimethylsilyl triflate afforded the pentasaccharide IV.46 in 87% yield. Following de-levulinoylation of IV.46 to IV.47, the terminal L-idose moiety was installed by glycosylation of IV.47 with IV.41 in the presence of tert-butyldimethylsilyl triflate to afford 99% of the hexasaccharide IV.48. Regioselective ring opening of the benzylidene followed by acetylation gave precursor IV.40 in 64% overall yield. Transformation of IV.40 to IV.49 was achieved sequentially by treatment with thioacetic acid [99], deprotection [100], and subsequent Swern oxidation and esterification. Deacetylation and saponification of the methyl esters with sodium hydroxide afforded IV.50. Finally, sulfation with sulfur trioxide–triethylamine complex gave IV.39, which upon hydrogenolysis (Pd/C) produced the target hexasaccharide IV.38.

Synthesis of Glycosaminoglycans

Scheme 43

The Sina¨y synthesis of sulfated N-acetyldermosine.

475

476

Scheme 44

The Sina¨y synthesis of sulfated N-acetylchondrosine.

Yeung et al.

Synthesis of Glycosaminoglycans

Scheme 45

477

The Ogawa synthesis of the dermatan sulfate hexasaccharide.

Most recently, Jacquinet and Rochepeau-Jobron reported the preparation of the disulfated ␣(1,3)-linked disaccharide (IV.51) of dermatan sulfate [see later-Scheme 48) [101]. Earlier approaches to the D-galactosamine moiety relied on azidonitration of D-galactal to introduce the 2-amino functionality. Here, Jacquinet and RochepeauJobron demonstrated the efficiency of using D-glucosamine derivatives as D-galac-

478

The Ogawa synthesis of the dermatan sulfate hexasaccharide.

Yeung et al.

Scheme 46

Synthesis of Glycosaminoglycans

479

tosamine precursors. Benzyl D-galactosamine derivatives IV.54 and IV.56 were thus obtained by selective inversion of the configuration at C4 of the corresponding Dglucosamine analogs in a three-step sequence (Scheme 47). Treatment of benzyl 2acetamido-2-deoxy-␣-D-glucopyranoside IV.52 [102] with pivaloyl chloride afforded the 3,6-di-O-pivaloyl derivative in 87% yield. Treatment of the unblocked 4-OH with trifluoromethanesulfonyl anhydride and pyridine followed by addition of water and subsequent heating at 90⬚C produced the 3,6-di-O-pivaloyl-D-galactopyranosyl intermediate IV.53 [103]. Deacylation of IV.53 afforded the crystalline triol IV.54 in 72% yield over the three steps. The same approach was used to achieve conversion of the ␤-D-glucopyranoside (IV.55) to the corresponding galactal (IV.56) in 75% overall yield. Standard protecting group manipulation of galactopyranosides IV.54 and IV.56 gave the respective glycosyl acceptors IV.57 and IV.58. Glycosylations of IV.57 and IV.58 using both the trichloroacetimidate IV.59 and the chloride IV.60 as L-iduronic acid donors demonstrated that the trichloroacetimidates are superior candidates for the anomeric activation of L-iduronic acid moieties to form the desired ␣-linked disaccharide (Scheme 48). Condensation of IV.57 with IV.59 in the presence of trimethylsilyl triflate afforded the ␣-linked disaccharide IV.61 in 63% yield and the ␤-linked anomer IV.62 in 27% yield. Interestingly, glycosylation of IV.57 or IV.58 with chloride IV.60 in the presence of silver triflate afforded very slowly, but exclusively, the corresponding ␤-linked disaccharides in 70 and 60% yield, respectively. O-Dechloroacetylation of IV.61 by treatment with thiourea gave IV.62, which was subsequently reprotected as the hydrogenolyzable 4-methoxybenzyl ether with 4-methoxybenzyl trichloroacetimidate and triflic acid under phase transfer catalysis conditions [104]. Saponification of the benzoate and methyl esters with lithium hydroperoxide followed by methanolic sodium hydroxide and acidification then gave the acid IV.63. O-Sulfonation of IV.63 was achieved with the sulfur trioxide–trimethylamine complex to give the disulfate IV.64 as the sodium salt. Finally, hydrogenolysis of IV.64 with Pd/C in aqueous methanol afforded the target disaccharide IV.51.

V.

KERATAN SULFATE

Keratan sulfate occurs in two principal forms, keratan sulfate I and keratan sulfate II, which are distinguished by their linkages to the core protein. Keratan sulfate I, found exclusively in the cornea and the oligosaccharide, is linked to protein by an N-glycosylic linkage between N-acetyl-D-glucosamine and L-asparagine [106]. Found in skin, cartilage, and bone, keratan sulfate II differs from the first variant by an Oglycosidic linkage between N-acetyl-D-galactosamine and L-serine or L-threonine [107]. Generally, the length of keratan sulfate chains range from short (5–10 disaccharides), to medium (20–30 disaccharides). Structurally, keratan sulfate differs from the other glycosaminoglycans in that it contains a nonacidic residue. The uronic acid moiety is replaced by a neutral D-galactose residue, thereby simplifying its synthesis by obviating the need for elaboration at C6. Additionally, sulfation can occur at the 6-OH on either the D-galactose or the N-acetylglucosamine residue. In cases of low sulfation, keratan sulfates can have a very low anionic character, attributable to the

480

Scheme 47

Yeung et al.

Synthesis of the galactosamine derivatives.

lack of a carboxylate group. The copolymer subunit is internally ␤(1,4)-linked between D-galactose and N-acetylglucosamine residues with ␤(1,3)-linkages between subunits (Fig. 7). Ogawa and coworkers have synthesized a tetrasaccharide fragment of keratan sulfate I, V.1 [108]. The target V.1 was derived from the fully protected tetrasaccharide V.2, which was in turn assembled by glycosylations involving key components V.3, V.4, and V.5 (Scheme 49). Starting from the known monomers V.6 [109] and V.7 [110], silver triflate promoted condensation afforded 92% of the allyl disaccharide V.8 (Scheme 50). Deactylation of V.8 was carried out with lithium hydroxide and hydrogen peroxide [111] in tetrahydrofuran to give 91% of the corresponding diol. Subsequent treatment with benzyl bromide in the presence of potassium iodide and silver(I) oxide afforded V.9 in 90% yield. Deallylation of V.9 was achieved with a rhodium catalyst and 1,4-diazbicyclo[2.2.2]octane followed by mercury(II) oxide and mercury(II) chloride in 10% aqueous acetone [112] to give the crude hemiacetal. Acetylation afforded V.10 as a mixture of ␤ and ␣ anomers (11:1) in 71% overall yield. Chemoselective deacetylation was achieved with hydrazine acetate [113] to afford 73% of the hemiacetal, which was then converted to the ␤-imidate V.4 (87%) as described earlier. Glycosylation of V.4 with the known compound V.5 [114] in the presence of BF3⭈OEt2 gave the desired trisaccharide V.11 in 83% yield. Deacetylation of V.11 to the diol followed by treatment with tert-butylchlorodiphenylsilane [115] and imidazole gave the monosilyl ether V.12 in 78% yield. The final glycosylation step was carried out with imidate V.3 in the presence of BF3⭈OEt2 to afford 48% of the desired tetrasaccharide V.13. De-phthaloylation followed by acetylation gave 63% of V.2. Conversion of the target V.1 from V.2 was achieved as follows: removal of the p-methoxyphenyl protecting group was carried out with cerium(IV) ammonium nitrate to give the diol. Subsequent desilylation with tetrabutylammonium fluoride afforded triol V.14 in 78% yield. Sulfation of V.14 with sulfur trioxide– triethylamine complex produced 93% of the tri-O-sulfated derivative V.15. Finally, hydrogenolysis (Pd/C) of V.15 provided the tetrasaccharide fragment of keratan sulfate I (V.1) in 92% yield.

Synthesis of Glycosaminoglycans

Scheme 48

The use of galactosamine in the synthesis of dermatan sulfate derivatives.

481

482

Yeung et al.

Figure 7

VI.

Keratan sulfate.

HEPARIN AND HEPARAN SULFATE

Heparin was originally isolated from liver in 1916 by Maclean, and its anticoagulant properties were immediately realized. However, structural complexity prevented the unambiguous establishment of the accepted chemical structure of heparin until the late 1960s [116]. Heparin’s remarkable pharmacological properties have resulted in a large body of research, including the chemical synthesis of heparin fragments and related analogs [117]. The literature pertaining to the synthesis of heparin, heparan sulfate, and their analogs is vast and has been reviewed [118]. Heparin (VI.1) (Fig. 8) is found in a variety of mammalian tissues and consists of a repeating, linear copolymer of ␤(1,4)-linked uronic acid and glucosamine residues, where the uronic acid moiety consists of 90% iduronic acid and 10% glucuronic acid [119]. The most common copolymer of heparin is the trisulfated disaccharide, where sulfation is solely on glucosamine at the C2, C3, and C6 positions. In a number of structural variants, the 2-amino functionality can be sulfated, acetylated or unsubstituted, and consequently heparin formations are microheterogeneous [120]. The source of antithrombotic activity, hence the anticoagulant activity of heparin, is the result of specific affinity for the serine protease inhibitor antithrombin III (AT-III). Inhibition of AT-III in turn inactivates serine proteases factor Xa and thrombin (factor IIa), which are downstream members of the coagulation cascade [121]. AT-III is a weak inhibitor of factor Xa and thrombin, which is considerably enhanced upon the binding of heparin [122]. Different molecular weight fragments of heparin display different anticoagulant properties. Specifically, heparin polysaccharides above 5 kDa inhibit thrombin and factor Xa in the presence of AT-III, while lower molecular weight heparin fragments only inhibit factor Xa [123]. Studies on heparin fragments obtained from chemical or enzymatic degradation revealed that approximately one-third of the heparin chains had the ability to bind to AT-III [124]. These experiments suggested that a limited number of heparin fragments possessed the structural features that are the source of its biological activity. Subsequently, a unique pentasaccharide domain was identified as necessary and sufficient for binding and activation of AT-III (Fig. 9) [125]. The sequence contains three monosaccharide units that rarely occur in heparin: a 6-Osulfate-N-acetyl-␣-D-glucosamine (unit D in Fig. 9), a ␤-D-glucuronic acid (unit E), and a 3,6-di-O-sulfate-N-sulfate-␣-D-glucosamine moiety (unit F). The D unit can be either N-acetylated (VI.3) or N-sulfated (VI.4) depending on the source of the polysaccharide. The discovery that this pentasaccharide was responsible for the biological activity of the larger polysaccharide chain was a considerable breakthrough,

Synthesis of Glycosaminoglycans

Scheme 49

The Ogawa synthesis of the keratan sulfate I tetrasaccharide, part 1.

483

484

Scheme 50

The Ogawa synthesis of the keratan sulfate I tetrasaccharide, part 2. Yeung et al.

Synthesis of Glycosaminoglycans

Figure 8

485

Heparin and heparan sulfate.

and it highlights the role that synthesis of GAGs can play in revealing the molecular origin of their biological activities. The charged groups on the heparin pentasaccharide were originally thought to be required for AT-III activation [125]. To test this hypothesis, the methyl glycosides of heparin analogs that lacked charged groups at defined positions, or their de-Osulfated derivatives, were chemically synthesized and used to probe the binding properties of the pentasaccharide to AT-III. As expected, the presence and position of the charged sulfate groups are essential for binding to and activation of AT-III (summarized in Fig. 10). Anionically charged groups indicated with an asterisk are essential, inasmuch as their removal leads to more than 95% (***) or 75% (**) loss of anti-Xa activity. The sulfates indicated by a single asterisk (*) are sites that contribute weakly to or are nonessential for AT-III activation [121,126]. The introduction of an extra 3-O-sulfate group (VI.6) at the reducing end (unit H of Fig. 10) significantly increased the affinity for AT-III, and thus the bioefficacy over that of the natural pentasaccharide [127]. The identification of the specific sites of interaction within the heparin pentasaccharide catalyzed the synthesis of a myriad of heparin analogs (Fig. 11). Among these were a series of ‘‘nonglycosaminoglycan’’ pentasaccharide analogs containing

Figure 9

The heparin pentasaccharide necessary for binding and activation to AT-III.

486

Yeung et al.

Figure 10

Essential and nonessential sites on heparin required for AT-III activation.

only O-sulfate esters and O-methyl ethers to varying degrees (VI.7) [128]. van Boeckel and coworkers combined structural properties that were shown not to affect AT-III activation by methylation of the free hydroxyls and replacement of N-sulfates for O-sulfates [129]. As expected, the methylated and O-sulfated heparin analogs displayed increased biological activity. Moreover, the synthesis of this class of analogs is greatly simplified because no amino sugars need be introduced, and protecting group chemistry is simplified because all nonsulfated hydroxyl groups are methylated. Heparan sulfate (VI.2) is a related repeating linear copolymer of variably sulfated uronic acid and glucosamine content. It contains an average of only one sulfate per disaccharide and is predominantly composed entirely of ␤(1,4)-linked glucuronic acid (10–50%) to glucosamine. Whereas heparin is an intracellular polysaccharide, heparan sulfate is a common cell surface component of many tissue types. Heparan sulfate contains all the disaccharide copolymer variants that are found in heparin, with a higher frequency of the minor sequences. Additionally, heparan sulfate contains domain structures that have high levels of sulfation, which resemble heparin and may be involved in its binding to heparin binding proteins [130,131]. The increased microheterogeneity of complexity may explain its presence on the cell surface. VII.

CONCLUSION

The arsenal of synthetic methodology of carbohydrates has increased significantly in recent decades and coincides with the discovery and characterization of biologically active polysaccharides. Glycosaminoglycans showcase the intertwining of biology and chemistry, both of which are required to completely understand protein–GAG interactions that are essential to biological systems. Small GAG oligomers and analogs are useful probes of the specific biological interactions responsible for activity and the various strategies utilized in the synthesis of GAGs represent state-of-theart synthetic methodology. Early synthetic efforts employed Koenigs–Knorr-type

Synthesis of Glycosaminoglycans

Figure 11

487

The nonglycosaminoglycan analogs of heparin sulfate.

glycosylations that were often low yielding with moderate ␣,␤-selectivities. Although halo sugars are still widely used today, new glycosylation strategies have been developed that achieve consistently high yields and selectivities. The strategy most widely utilized in GAG synthesis is Schmidt’s trichloroacetimidate methodology. The high efficiency and substrate generality of the method allow for the use of uronic acid building blocks. The design and installation of the hexosamine unit has also benefited from recent advances in synthetic methodology. The 2-deoxy-2-amino functionality was normally derived from glucosamine or galactose amine; however, azidonitration and sulfonamidation of glycals are particularly suited for GAG synthesis. The chemical synthesis of GAG oligomers, their derivatives, and analogs is an important tool in determining their biological roles. High degree of functionalization

488

Yeung et al.

coupled with diversity among individual GAG members continue to make them challenging synthetic targets. As with heparin, the chemical preparations of GAGs play a critical role in probing and elucidating specific protein–GAG interactions. Chemical syntheses of heparin clearly established the anionic carboxylate and sulfate groups as the sites of protein interaction. The application of this methodology to other GAGs is forthcoming, but it is clear that only synthesis allows for the specific modifications needed to reveal the sites responsible for biological activity. REFERENCES 1. 2. 3. 4. 5.

6. 7. 8. 9. 10.

11. 12.

13. 14. 15. 16. 17. 18.

19. 20. 21.

(a) Folkman, J.; Shing, Y. Adv. Exp. Med. Biol. 1992, 313, 355. (b) Riegel, A.T.; Wellstein, A. Breast Cancer Res. Treat. 1994, 31, 309. Hascall, V. C.; Kimura, J. H. Methods Enzymol., 1982, 82, 769. Baldwin, H. S.; Lloyd, T. R.; Solursh, M. Circ. Res. 1994, 74, 224. Tan, S. W.; Johns, M. R.; Greenfield, P. F. Nucl. Med. Biol. 1994, 22, 251. (a) Hardwick, C.; Hoare, K.; Owens, R.; Hohn, H. P.; Hook, M.; Moore, D.; Cripps, V.; Austen, L.; Nance, D. M.; Turley, E. A. J. Cell. Biol. 1992, 117, 1343. (b) Underhill, C. J. Cell Sci. 1992, 103, 293. Hascall, V. C.; Hascall, G. C. In Cell Biology of the Extracellular Matrix, E. H. Hay, ed. Plenum Press, New York, 1991, p. 39. Lindhart, R. J.; Toida, T. In Carbohydrates as Drugs, Z. B. Witczak and K. A. Nieforth, eds. Marcel Dekker, New York, 1997, p. 277. Hook, M.; Kjellen, L.; Johansson, S.; Robinson, J. Annu. Rev. Biochem. 1984, 53, 847. Templeton, D. M. Crit. Rev. Clin. Lab. Sci. 1992, 29, 141. (a) Lindahl, U.; Hook, M. Annu. Rev. Biochem. 1978, 47, 385. (b) Kjellen, L.; Lindahl, U. Annu. Rev. Biochem. 1991, 60, 443. (c) Hassell, J. R.; Kimura, J. H.; Hascall, V. C. Annu. Rev. Biochem. 1986, 55, 539. Chen Y.; Maguire, T.; Hileman, R. E.; Fromm, J. R.; Esko, J. D.; Linhardt, R. J.; Marks, R. M. Nat. Med. 1997, 3, 866. (a) Jagodzinski, P. P.; Wustner, J.; Kmieciak, D.; Wasik, T. J.; Fertala, A.; Sieron, A. L.; Takahashi, M.; Tsuji, T.; Mimura, T.; Fung, M. S.; Gorny, M. K.; Kloczewiak, M.; Kaneko, Y.; Kozbor, D. Virology 1996, 226, 217. (b) Gordon, M.; Guralnik, M.; Kaneko, Y.; Mimura, T.; Goodgame, J.; Lang, W. J. Med. 1995, 26, 217. Ricco, R.; Kinnel, R. B.; Bifulco, G.; Scheuer, P. J. Tetrahedron Lett. 1996, 37, 1979. Carter, M. B.; Petillo, P. A.; Anderson, L.; Lerner, L. E. Carbohydr. Res. 1994, 258, 299. (a) Nakahara, Y.; Ogawa, T. Tetrahedron Lett. 1987, 28, 2731. (b) Nakahara, Y.; Ogawa, T. Carbohydr. Res. 1987, 167, c1. (a) Schmidt, R. R. Angew. Chem. Int. Ed. Engl. 1986, 25, 212. (b) Schmidt, R. R.; Kinzy, W. Adv. Carbohydr. Chem. Biochem. 1994, 50, 21. (a) Kahne, D.; Walker, S.; Cheng, Y.; van Engen, D. J. Am. Chem. Soc. 1989, 111, 6881. (b) Yan, L.; Kahne, D. J. Am. Chem. Soc. 1996, 118, 9239. (a) Mootoo, D. R.; Date, V.; Fraser-Reid, B. J. Am. Chem. Soc. 1988, 110, 2662. (b) Mootoo, D. R.; Konradsson, P.; Udodong, U.; Fraser-Reid, B. J. Am. Chem. Soc. 1988, 110, 5583. (c) Fraser-Reid, B.; Konradsson, P.; Mootoo, D. R.; Udodong, U. J. Chem. Soc. Chem. Commum. 1988, 823. (d) Ratcliffe, A. J.; Konradsson, P.; Fraser-Reid, B. J. Am. Chem. Soc. 1990, 112, 5665. (d) Mootoo, D. R.; Fraser-Reid, B. Tetrahedron Lett. 1989, 30, 2363. Slaghek, T. L.; Nakahara, Y.; Ogawa, T. Tetrahedron Lett. 1992, 33, 4971. Jacquinet, J.-C. Carbohydr. Res. 1990, 199, 153. Blatter, G.; Jacquinet, J.-C. Carbohydr. Res. 1996, 288, 109.

Synthesis of Glycosaminoglycans 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33.

34. 35.

36. 37. 38. 39. 40.

41. 42.

43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54.

489

Yeung, B. K. S.; Petillo, P. A. Unpublished results. Hill, D. C.; Flugge, L. A.; Petillo, P. A. J. Org. Chem. 1997, 62, 4864. Lemieux, R. U.; Ratcliffe, R. M. Can. J. Chem. 1979, 57, 1244. Danishefsky, S. J.; Randolph, J. T.; Roberge, J. Y.; McClure, K. F.; Ruggeri, R. B. E. Schering Research Foundation Lecture, 1995, Vol. 26. Griffith, D. A.; Danishefsky, S. J. J. Am. Chem. Soc. 1990, 112, 5811. Griffith, D. A.; Danishefsky, S. J. J. Am. Chem. Soc. 1991, 113, 5863. Danishefsky, S. J.; Koseki, K.; Griffith, D. A.; Gervay, J.; Peterson, J. M.; McDonald, F. E.; Oriyama, T. J. Am. Chem. Soc. 1992, 114, 8331. Roberge, J. Y.; Beebe, X.; Danishefsky, S. J. Science 1995, 269, 202. Seeberger, P. H.; Bilodeau, M. T.; Danishefsky, S. J. Aldrichim. Acta 1997, 30, 75. Meyer, K.; Palmer, J. W. J. Biol. Chem. 1934, 107, 629. Lindahl, U.; Hook, M. Annu. Rev. Biochem. 1978, 47, 385. (a) Hassell, J. R.; Kimura, J. H.; Hascall, V. C. Annu. Rev. Biochem. 1986, 55, 539. (b) Hascall, V. C.; Kimura, J. H. Methods Enzymol. 1982, 82, 769. (c) Kjellen, L.; Lindahl, U. Annul. Rev. Biochem. 1991, 60, 443. (a) McDonald, J. A. Annu. Rev. Cell Biol. 1988, 4, 183. (b) Laurent, T. C.; Fraser, J. R. E. FASEB J. 1992, 6, 2397. (a) Toole, B. P. Curr. Opin. Cell Biol. 1990, 2, 839. (b) Toole, B. P. In Cell Biology of the Extracellular Matrix, E. H. Hay, ed. Plenum Press, New York, 1991, p. 305. (c) Toole, B. P.; Jackson, G.; Gross, G. Proc. Natl. Acad. Sci. USA 1972, 69, 1384. Bayliss, M. T. In Articular Cartilage Biochemistry, K. Kuettner, R. Schleyerbach, and V. C. Hascall, eds. Raven Press, New York, 1986, p. 295. Yoneda, M.; Shimizu, S.; Nishi, Y.; Yamagata, M.; Susuki, S.; Kimata, K. J. Cell Sci. 1988, 90, 275. Ellis, I.; Grey, A. M.; Schor, A. M.; Schor, S. L. J. Cell Sci. 1992, 102, 447. Feinberg, R. N.; Beebe, D. C. Science, 1983, 220, 1177. (a) West, D. C.; Hampson, I. N.; Arnold, F.; Kumar, S. Science, 1985, 228, 1324. (b) West, D. C.; Kumar, S. In The Biology of Hyaluronan, Ciba Foundation Symposium 143. Wiley Press, Chichester, England, 1989, p. 187. Banerjee, S. D.; Toole, B. P. J. Cell Biol. 1992, 119, 643. Mashimoto, M.; Saegusa, H.; Chiba, S.; Kitagawa, H.; Myoshi, T. Japanese patent 63,123,392, 1988. (b) Akasaka, H.; Seto, S.; Yanagi, M.; Fukushima, S.; Mitsui, T. J. Soc. Cosmet. Chem. Jpn. 1988, 22, 35. (a) Rapport, M. M.; Weissmann, B.; Linker, A.; Meyer, K. Nature 1951, 168, 996. (b) Isikawa, T. Tohoku J. Expl. Med. 1951, 53, 217. Flowers, H. M.; Jeanloz, R. W. J. Am. Chem. Soc. 1962, 84, 3030. Takanashi, S.; Hirasaka, Y. L.; Kawada, M.; Ishidate, M. J. Am. Chem. Soc. 1962, 84, 3029. Klaffke, W.; Warren, C. D.; Jeanloz, R. W. Carbohydr. Res. 1992, 244, 171. Lemieux, R. U.; Takeda, T.; Chung, B. Y. ACS Symp. Ser. 1976, 39, 90. (a) Kovac, P. Carbohydr. Res. 1986, 153, 237. (b) Bertolini, M.; Glaudemans, C. P. J. Carbohydr. Res. 1970, 15, 263. (a) Gross, H.; Farkas, I.; Bognar, R. Z. Chem. 1978, 18, 201. (b) Ziegler, T.; Kovac, P.; Glaudemans, C. P. J. Carbohydr. Res. 1989, 194, 185. Slaghek, T. L.; Hypponen, T. K.; Ogawa, T. Tetrahedron Lett. 1993, 34, 7939. Schmidt, R. R.; Michel, J.; Roos, M. Liebigs Ann. Chem. 1984, 1343. Grundler, G.; Schmidt, R. R. Carbohydr. Res. 1985, 135, 203. (a) Kunz, H.; Waldman, H. Angew. Chem. 1984, 96, 49. (b) Hayakawa, Y.; Kato, H.; Uchiyama, M.; Kajino, H.; Noyori, R. J. Org. Chem. 1986, 51, 2400. Slaghek, T. L.; Nakahara, Y.; Ogawa, T.; Kamerling, J. P.; Vliegenthart, J. F. G. Carbohydr. Res. 1994, 255, 61.

490

Yeung et al.

55. 56. 57.

Omura, K.; Swern, D. Tetrahedron 1978, 34, 1651. Lindgren, B. O.; Nilsson, T. Acta Chem. Scand. 1973, 27, 888. (a) Baker, B. R.; Joseph, J. P.; Schaub, R. E.; Williams, J. H. J. Org. Chem. 1954, 19, 1786. (b) Lemieux, R. U.; Takeda, T.; Chung, B. Y. ACS Symp. Ser., 39, 1976, Ch. 6. Chernyak, A. Y.; Kononov, L. O.; Krishna, P. R.; Kochetkov, N. K.; Rama Rao, V. A. Carbohydr. Res. 1992, 225, 279. Vedejs, E.; Lin, S. J. Org. Chem. 1994, 59, 1602. Fukuyama, T.; Jow, C.-K.; Cheung, M. Tetrahedron Lett. 1995, 36, 6373. Yeung, B. K. S.; Hill, D. C.; Janicka, M.; Petillo, P. A. In preparation. Johansson, R.; Samuelsson, B. J. Chem. Soc. Perkin Trans. I 1984, 2371. Handcock, G.; Galpin, I. J., Morgan, B. A. Tetrahedron Lett. 1982, 23, 249. (a) Davis, N. J.; Flitsch, S. L. Tetrahedron Lett. 1993, 34, 1181. (b) Garegg, P. J.; Oscarson, S.; Tedebark, U. J. Carbohydr. Chem. 1998, 17, 587. Ellervik, U.; Magnusson, G. Tetrahedron Lett. 1997, 38, 1627. Fransson, L.-A. In The Polysaccharides, vol. 3, G. O. Aspinall, ed. Academic Press, New York, 1985, p. 337. Jacquinet, J.-C.; Sina¨y, P. Carbohydr. Res. 1987, 159, 229. Herzig, J.; Nudelman, R.; Gottlieb, H. E. Carbohydr. Res. 1988, 177, 21. Grundler, G.; Schmidt, R. R. Liebigs Ann. Chem. 1984, 1826. Helferich, B.; Portl, W. Chem. Ber. 1959, 86, 604. Tamura, J.; Neumann, K. W.; Ogawa, T. Bioorg. Med. Chem. Lett. 1995, 5, 1351. Ogilvie, K. K.; Beaucage, S. L.; Entwistle, D. W. Tetrahedron Lett. 1976, 16, 1255. Nakahara, Y.; Ogawa, T. Carbohydr. Res. 1990, 205, 147. Nakano, T.; Ito, Y.; Ogawa, T. Tetrahedron Lett. 1991, 32, 1569. Rosen, T.; Lico, I. M.; Chu, D. T. W. J. Org. Chem. 1988, 53, 1580. Garegg, P. J.; Hultberg, H.; Wallin, S. Carbohydr. Res. 1982, 108, 97. Goto, F.; Ogawa, T. Tetrahedron Lett. 1992, 33, 6841. (a) El-Sokkary, R. I.; Silwanis, B. A.; Nashed, M. A.; Paulsen, H. Carbohydr. Res. 1990, 203, 319. (b) Ito, Y.; Nunomura, S.; Shibayama, S.; Ogawa, T. J. Org. Chem. 1992, 57, 1821. (c) Lay, L.; Nicotra, F.; Panza, L.; Russo, G.; Adobati, E. Helv. Chim. Acta 1994, 77, 509. (a) Wong, T. C.; Hague, W.; Abbas, S. Z.; Noujaim, A. A. J. Carbohydr. Chem. 1990, 9, 745. (b) Lubineau, A.; Bienayme, H. Carbohydr. Res. 1991, 212, 267. Coutant, C.; Jacquinet, J.-C. J. Chem. Soc. Perkins Trans. I 1995, 1573. Albert, R.; Dax, K.; Link, R. W.; Stutz, A. E. Carbohydr. Res. 1983, 118, c5. Blatter, G.; Beau, J. M.; Jacquinet, J.-C. Carbohydr. Res. 1994, 260, 189. Jacquinet, J.-C.; Rochepeau-Jobron, L.; Combal, J.-P. Carbohydr. Res. 1998, 314, 283. Lubineau, A.; Bonnaffe´, D. Eur. J. Org. Chem. 1999, 2532. Meyer, K.; Chaffee, E. J. Biol. Chem. 1941, 138, 491. Poblacion, C. A.; Michelacci, Y. M. Carbohydr. Res. 1986, 147, 87. Tollefsen, D. M.; Pestka, C. A.; Monafo, W. J. J. Biol. Chem. 1983, 258, 6713. Maimone, M. M.; Tollefsen, D. M. J. Biol. Chem. 1990, 265, 18263. Mascellani, G.; Liverani, L.; Prete, A.; Bergonzoni, G. L.; Bianchini, P.; Silvestro, L.; Torri, G.; Bisio, A.; Guerrini, M.; Casu, B. J. Carbohydr. Chem. 1995, 14, 1165. Jacquinet, J.-C.; Sina¨y, P. Carbohydr. Res. 1987, 159, 229. David, S.; Hanessian, S. Tetrahedron 1985, 41, 643. Lemieux, R. U.; Driguez, H. J. Am. Chem. Soc. 1975, 97, 4069. Marra, A.; Dong, X.; Petitou, M.; Sina¨y, P. Carbohydr. Res. 1989, 195, 39. Chiba, T.; Sina¨y, P. Carbohydr. Res. 1986, 151, 379. Banoub, J.; Bundle, D. R. Can. J. Chem. 1979, 57, 2091. Garegg, P. J.; Hultberg, H.; Wallin, S. Carbohydr. Res. 1982, 108, 97. Goto, F.; Ogawa, T. Bioorg. Med. Chem. Lett. 1994, 4, 619.

58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78.

79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97.

Synthesis of Glycosaminoglycans 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118.

119. 120. 121. 122. 123. 124. 125.

491

Horito, S.; Lorentzen, J. P.; Paulsen, H. Liebigs Ann. Chem. 1986, 1880. Rosen, T.; Lico, I. M.; Chu, D. T. W. J. Org. Chem. 1988, 53, 1580. Fukuyama, T.; Laird, A. A.; Hotchkiss, L. M. Tetrahedron Lett. 1985, 26, 2691. Rochepeau-Jobron, L.; Jacquinet, J.-C. Carbohydr. Res. 1998, 305, 181. Shulman, M. L.; Khorlin, A. Ya. Carbohydr. Res. 1973, 27, 141. Belot, F.; Jacquinet, J.-C. Carbohydr. Res. 1996, 290, 79. Patil, V. J. Tetrahedron Lett. 1996, 37, 1481. Lucas, H.; Basten, J. E. M.; van Dinther, T. G.; Meuleman, D. G.; van Aelst, S. F.; van Boeckel, C. A. A. Tetrahedron 1990, 46, 8207. Stein, T.; Keller, R.; Stuhlsatz, H. W.; Grieling, H.; Ohst, E.; Muller, E.; Scharf, H.-D. Hoppe-Seyler’s Z. Physiol. Chem. 1982, 363, 825. Hopwood, J. J.; Robinson, H. C. Biochem. J. 1974, 141, 57. Kobayashi, M.; Yamazaki, F.; Ito, Y.; Ogawa, T. Carbohydr. Res. 1990, 201, 51. Ito, Y.; Ogawa, T. Agric. Biol. Chem. 1986, 50, 3227. Yamazaki, F.; Kitajima, T.; Nukada, T.; Ito, Y.; Ogawa, T. Tetrahedron Lett. 1989, 30, 4417. Corey, E. J.; Kim, S.; Yoo, S.; Nicoloau, K. C.; Melvin Jr., L. S.; Brunelle, D. J.; Flack, J. R.; Trybulski, E. J.; Lett, R.; Sheldrake, R. W. J. Am. Chem. Soc. 1978, 100, 4620. (a) Corey, E. J.; Suggs, R. J. Org. Chem. 1973, 38, 3224. (b) Gent, P. A.; Gigg, R. J. Chem. Soc. Chem. Commun. 1974, 227. Excoffier, G.; Gagnaire, D.; Uttile, J.-P. Carbohydr. Res. 1975, 39, 368. Liptak, A. Tetrahedron Lett. 1976, 3551. Hanessian, S.; Lavalle, P. Can. J. Chem. 1975, 53, 2975. Casu, B. Adv. Carbohydr. Chem. Biochem. 1985, 43, 51. Petitou, M.; van Boeckel, C. A. A. Prog. Chem. Org. Nat. Prod. 1992, 60, 143. (a) Petitou, M.; Duchaussoy, P.; Lederman, I.; Choay, J.; Sina¨y, P.; Jacquinet, J.-C.; Torri, G. Carbohydr. Res. 1986, 147, 221. (b) Petitou, M.; Duchaussoy, P.; Lederman, I.; Choay, J.; Jacquinet, J.-C.; Sina¨y, P.; Torri, G. Carbohydr. Res. 1987, 167, 67. (c) Ichikawa, Y.; Monden, R.; Kuzuhara, H. Carbohydr. Res. 1988, 172, 37. (d) Chiba, T.; Jacquinet, J.-C.; Sina¨y, P.; Petitou, M.; Choay, J. Carbohydr. Res. 1988, 174, 253. (e) Ichikawa, Y.; Monden, R.; Kuzuhara, H. Tetrahedron Lett. 1986, 27, 611. (f) Wessel, H. P.; Labler, L.; Tschopp, T. B. Helv. Chim. Acta 1989, 72, 1268. (g) van Boeckel, C. A. A.; Beetz, T.; Vos, J. T.; de Jong, A. J. M.; van Aelst, S. F.; van den Bosch, R. H.; Mertens, J. M. R.; van der Vlugt, F. A. J. Carbohydr. Chem. 1985, 4, 293. (h) Paulsen, H.; Huffziger, A.; van Boeckel, C. A. A. Liebigs Ann. Chem. 1988, 419. (i) Jaurand, G.; Basten, J.; Lederman, I.; van Boeckel, C. A. A.; Petitou, M. Bioorg. Med. Chem. Lett. 1992, 2, 897. (j) Petitou, M.; Jaurand, G.; Derrien, M.; Duchaussoy, P.; Choay, J. Bioorg. Med. Chem. Lett. 1991, 1, 95. (a) Lane, D. A.; Lindahl, U.; Heparin, Chemical and Biological Properties. CRC Press, Boca Raton, FL, 1989. (b) Casu, B. Adv. Carbohydr. Chem. Biochem. 1985, 43, 51. Griffin, C. C.; Linhardt, R. J.; Van Gorp, C. L.; Toida, T.; Hileman, R. E.; Schubert, R. L.; Brown, S. E. Carbohydr. Res. 1995, 276, 183. Van Boeckel, C. A. A.; Grootenhuis, P. D. J.; Meuleman, D.; Westerduin, P. Pure Appl. Chem. 1995, 67, 1663. Bjork, I.; Olsen, S. T.; Shore, J. D. In Heparin, D. A. Lane, and U. Lindahl, eds. Edward Arnold, London, 1989, p. 229. Andersson, L. O.; Barrowcliffe, T. W.; Holmer, E.; Johnson, E. A.; Sims, G. E. C. Thromb. Res. 1976, 9, 575. Basten, J.; Jaurand, G.; Olde-Hanter, B.; Duchaussoy, P.; Petitou, M.; van Boeckel, C. A. A. Biomed. Chem. Lett. 1992, 2, 905. (a) Lindahl, U.; Backstrom, G.; Thunberg, L.; Leger, I. G. Proc. Natl. Acad. Sci. USA 1980, 77, 6551. (b) Thunberg, L.; Backstrom, G.; Lindahl, U. Carbohydr. Res. 1982,

492

126. 127. 128. 129. 130. 131.

Yeung et al. 100, 393. (c) Choay, J.; Lormeau, J. C.; Petitou, M.; Sina¨y, P.; Fareed, J. Ann. NY Acad. Sci. 1981, 370, 644. (a) Van Boeckel, C. A. A.; Beetz, T.; van Aelst, S. F. Tetrahedron Lett. 1988, 29, 803. Meuleman, D. G.; Hobbelen, P. M. J.; van Dinther, T. G.; Vogel, G. M. T.; van Boeckel, C. A. A.; Moelker, H. C. T. Semin. Thromb. Hemostasis 1991, 17, 112. van Boeckel, C. A. A.; van Aelst, S. F.; Beetz, T.; Meuleman, D. G.; van Dinther, Th. G.; Moelker, H. C. T. Ann. N. Y. Acad Sci. 1989, 556, 489. Jaurand, G.; Basten, J.; Lederman, I.; van Boeckel, C. A. A.; Petitou, M. Biomed. Chem. Lett. 1992, 2, 897. Hileman, R. E.; Fromm, J. R.; Weiler, J. M.; Linhardt, R. J. BioEssays 1998, 20, 156. Lindahl, U. Pure Appl. Chem. 1997, 69, 9, 1897.

13 Pyranosyl Sugar Amino Acid Conjugates: Their Biological Origins, Synthetic Preparations, and Structural Characterization Jacquelyn Gervay-Hague and Thomas M. Weathers, Jr. University of Arizona, Tucson, Arizona

I.

INTRODUCTION

Sometimes it is true that everything old is new again, which can be disheartening for the originators of an idea who see others receiving credit years later. In some cases, an idea consistently evolves over time and early insights are ‘‘rediscovered’’ many years later. In other situations, ideas are born before their time, when the tools required to assure their continued development are lacking. And, in some circumstances, the creators of an idea fail to fully realize its potential. There are elements of all these factors in this chapter on the use of pyranosyl sugar amino acids as amino acid equivalents. Perusal of the literature suggests that the idea of making pyranosyl sugar amino acids originated in the laboratories of Heyns and Paulsen, who synthetically prepared the first pyranosyl sugar amino acid in 1955. The accuracy with which the literature records the moment of inception is somewhat suspect, since reports of naturally occurring pyranosyl sugar amino acids quickly followed the syntheses. Natural product identification was facilitated by comparisons to synthetic materials presenting the possibility that collaborative efforts prompted the synthetic work. This seems especially plausible because there were relatively few syntheses published between 1955 and the late 1960s. This period was primarily dominated by reports on the isolation of several different pyranosyl sugar amino acids from natural sources. Hanessian and Haskell authored a prophetic disclosure describing the isolation and characterization 493

494

Gervay-Hague and Weathers

of a pyranosyl sugar amino acid amidically linked to alanine. The possibility that the alanine could bridge sugar amino acids in a similar fashion to ␣-peptides did not go unnoticed by these researchers. In the late 1960s and into the mid-1970s new synthetic reports appeared in the literature. Most of these focused on the total synthesis of naturally occurring pyranosyl sugar amino acid conjugates. Then, in 1976, Fuchs and Lehmann published the first synthesis of sugar amino acids amidically linked in a fashion similar to ␣amino acids. Surprisingly, this disclosure received little attention for nearly 20 years. In the past 5 years, the chemistry of pyranosyl sugar amino acids has evolved into a new science at the forefront of chemical design, synthesis, and structural characterization. Parallel work by Fleet and coworkers has elegantly exploited furanosyl sugar amino acids. At the same time, the sugar amino acids are only a small part of a larger science encompassing the use of unnatural amino acids in the synthesis of new materials with defined secondary structures. II.

NATURALLY OCCURRING PYRANOSYL SUGAR AMINO ACIDS

Glycopeptides are a large class of naturally occurring molecules containing a carbohydrate glycosidically linked to an ␣-amino acid, which is typically a component of a peptide or protein. There are both N- and O-linked glycopeptides. N-Acetylgalactosamine linked ␣ to serine is the major O-glycosidic linkage, whereas N-acetylglucosamine linked ␤ to asparaginine characterizes N-linked glycopeptides (Fig. 1) [1]. Sugar amino acids are carbohydrates containing both amine and carboxylic acid functionalities in place of hydroxyls and are structurally distinct from glycopeptides. Naturally occurring sugar amino acids come in several types, consisting of both amino furanosiduronic and pyranosiduronic acids. This chapter includes the latter class of compounds, often referred to as amino hexuronic acids. Hexuronic acids are six-carbon monosaccharides containing both an aldehyde and a chain-terminating carboxylic acid. The amino hexuronic acids have amine functionalities in place of hydroxyl groups normally found in sugars. 2-Acetamido-2-deoxy-D-galacturonic acid, a major immunogenic component of Salmonella typhosa [2], was the first naturally occurring amino hexuronic acid identified [3]. In 1962 Perkins reported that the cell wall of Micrococcus lysodeikticus

Figure 1

Pyranosyl

495

contained 2-amino-2-deoxymannuronic acid [4], and shortly thereafter, Hemophilus influenzae type d was shown to contain 2-acetamido-2-deoxy-D-glucuronic acid [5]. At about the same time, an antigenic staphylococcal polysaccharide was shown to be composed of 2-acetamido-2-deoxy-glucuronic acid glycosidically linked to 2-[(Nacetylalanyl)amino]-2-deoxyglucuronic acid (Fig. 2) [6]. It would appear that this was the first identification of a naturally occurring sugar amino acid amidically conjugated to an ␣-amino acid [7]. All possible hexose configurations have been found as amino hexuronic acids in bacteria, with the possible exception of idose [8]. In some cases, more than one hydroxyl is replaced by amine functionality; for example, 2,3-diamino-2,3-dideoxyglucuronic acid was isolated from the cell wall of Propionibacterium acnes [9]. Several more polysaccharides containing hexuronic acids linked to ␣-amino acids such as alanine, serine, and threonine have been characterized [7]. There are also examples of amino sugars linked to the carboxy terminus of ␣-amino acids (Fig. 3) [10]. Ezomycin A1 and gougerotin are natural nucleoside antibiotics comprising sugar amino acids conjugated to ␣-amino acids through amide bonds. Ezomycin A1 contains two pyranoses, both having ␥-amino acid functionalities, whereas gougerotin is a carbohydrate-based ␤-amino acid (Fig. 4) [11]. Ezoaminuroic acid was the first naturally occurring 3-amino-3-deoxy-hexuronic acid to be identified [12]. Structure elucidation studies, chemical syntheses, and biochemical properties of these and related compounds have been reviewed [13,14]. Amino glyculosonic acids are members of another class of sugar amino acids that are 2-keto sugars with a C1 carboxylic acid. One of the most abundant representatives of this class of molecules is neuraminic acid, a nine-carbon sugar amino acid. There are several known derivatives of neuraminic acid, which are collectively called the sialic acids [15]. They are components of bacterial and viral capsular polysaccharides, and they are important constituents of mammalian cellular recognition elements. Besides the sialic acids, two other amino nonulosonic acids have been isolated from bacteria. These compounds are 5,7-diamino-3,5,7,9-tetradeoxynonulosonic acids. The amine functionalities of pseudaminic acid (Fig. 5) have been found acylated, formylated, and conjugated to serine in nature [16].

III.

SYNTHETIC STUDIES OF PYRANOSYL SUGAR AMINO ACIDS

A.

Syntheses of Natural Amino Hexuronic Acids

Several syntheses targeting amino hexuronic acids have been reported. Heyns and Paulsen synthesized the first sugar amino acids in 1955 [17]. In the event, benzyl2-N-carbobenzyloxy-2-deoxy-␣-D-glucopyranoside was reacted with oxygen and platinum to provide 2-N-carbobenzyloxy-2-deoxy-D-glucuronic acid, which after removal of the nitrogen protecting group yielded a ␦-sugar amino acid (Fig. 6). 2-Amino-2-deoxy-galacturonic acid was prepared in a similar fashion [18]. It is noteworthy that these synthetic efforts predated the identification of sugar amino acids in nature. The syntheses of ␣- and ␤-phenylglycosides of 2-N-acetyl-2-deoxyglucuronic acid were described in 1958 [19], and in 1961 Weidmann and Zimmerman reported several different reactions of 2-amino-2-deoxyglucuronic acids [20]. These combined

496

Gervay-Hague and Weathers

Figure 2

Pyranosyl

Figure 3

497

498

Gervay-Hague and Weathers

Figure 4

Pyranosyl

Figure 5

499

500

Gervay-Hague and Weathers

Figure 6

reports set the foundation for subsequent studies by Yoshimura et al., who prepared several derivatives of 2-amino-2-deoxyglycuronic acids [21]. Tsuji and coworkers were the first to synthesize a 3-amino-3-deoxyhexuronic acid, in 1968 [22]. They began with an isopropylidene glucofuranosiduronic ester and introduced the nitrogen functionality via hydrazone formation and subsequent reduction. Removal of the isopropylidene resulted in formation of amino allouronic acid (Fig. 7). Paulsen and coworkers used a very similar strategy in making 3-amino3-deoxyglucuronic acid [23]. Ogawa later published the synthesis of a 3-amino-3,4-dideoxyhexuronic acid in route to ezomycin [24]. In that account, a 1,6-anhydro epoxy sugar was reacted with sodium azide followed by antimony pentachloride to give methyl 3-azido-2-Obenzoyl-␣-D-glucospyranoside. Oxidation of the primary alcohol was achieved with potassium permanganate, and reduction of the azide was accomplished with hydrogenation (Fig. 8). Finally, in 1979 Horton reported the synthesis of 3-amino-2,3-dideoxyhexuronic acids. In six steps, methyl ␣-D-mannopyranoside was converted to a highly functionalized 3-acetamido-6-azido-2,3,6-trideoxy derivative. Photochemical activation of the azide provided an imine, which was subsequently hydrolyzed to the aldehyde. Bromine was used to oxidize the aldehyde to the acid, which was esterified with methyl iodide (Fig. 9). In their studies directed toward the synthesis of gougerotin, Watanabe and coworkers prepared methyl-4-O-mesyl galactopyranoside and reacted it with sodium azide; after deprotection, methyl 4-azido-4-deoxyglucopyranoside was obtained [25]. This compound was oxidized with platinum and oxygen to afford the 4-azido glucuronic acid (Fig. 10) [26]. The focus of these early synthetic studies was to prepare amino hexuronic acid derivatives that could be joined through O-glycosidic linkages, which are found in nature. B.

Syntheses of Unnatural Sugar Amino Acids

Fuchs and Lehmann demonstrated that C-glycoside amino acids could be prepared from selective ring opening of sugar-based anhydrides [27]. In the reaction, the anhydride was regioselectively reacted with ammonia to give a C-glycoside amide, which was subsequently converted to a nitrile by the action of tosyl chloride in pyridine. The nitrile was reduced with catalytic hydrogenation, and the major product resulted from migration of the C3 acetate to the C1 amine. The O- and N-acetyls were removed by saponification, using 2 N sodium hydroxide at 100⬚C (Fig. 11). In

Pyranosyl

Figure 7

Figure 8

501

Figure 9

502

Figure 10

Gervay-Hague and Weathers

Figure 11

Pyranosyl

503

Figure 12

1994 a related heptonic amino acid was prepared by nitromethane addition to glucose, reduction of the nitro group, and oxidation of the primary hydroxyl group (Fig. 12) [28]. Fuchs and Lehmann also prepared 7-amino C-glycoside carboxylic acids from the corresponding C-glycoside nitriles, which were hydrolyzed to carboxylic acids. The C6 hydroxyls were selectively converted to azides and subsequently reduced with catalytic hydrogenation [29]. Although Hanessian and Haskell mentioned the possibility that naturally occurring sugar amino acids could be linked via amide bonds rather than glycosides, Fuchs and Lehmann were the first to reduce this to practice in a synthetic arena (see below). Kim and Hollingsworth prepared a C-glycoside of N-acetylglucosamine by alkylation of a pyranosyl bromide with malonate anion, giving an N-acetyl ␥-sugar amino acid after decarboxylation (Fig. 13) [30]. Galantinic acid is a somewhat related compound that can be considered to be a C-glycoside pyranosyl amine. It is an ␧-amino acid that was thought to be a component of Galantin I, a naturally occurring peptide-based antibiotic. In 1992, however, Sakai and Ohfune showed that galantinic acid is a by-product of the isolation procedure, rather than a component of the natural product [31]. In earlier work, Ohfune and Kurokawa reported a stereocontrolled synthesis of galantinic acid from a serine-derived epoxide. Cuprate opening of the epoxide gave a conjugated ester that underwent Michael addition upon deprotection (Fig. 14) [32]. This example illustrates a subtle relationship between ␣-amino acids and sugar amino acids, since galantinic acid has structural entities related to both classes of compounds. Perhaps the interplay between sugar amino acids and ␣-amino acids is more clearly demonstrated in the synthesis of ␣-D-glucosyl-(R)-alanine reported by Axon and Beckwith [33]. In the reaction, (2R)-methyleneoxazolidinone was treated with 2,3,4,6-tetra-O-acetyl glucosyl iodide in the presence of sodium cyanoborohydride and tributyltin chloride to give the ␣-C-glycoside in 88% yield (Fig. 15).

504

Figure 13

Gervay-Hague and Weathers

Figure 14

Pyranosyl

Figure 15

505

506

Gervay-Hague and Weathers

␤-C-Glycosides have also been prepared by the Lewis acid catalyzed addition of ketene acetals, derived from ␣-amino acids, to glycosyl bromides [34]. The acyl protecting group at C2 was critical in this reaction because it directed the ␤ incorporation of the amino acid. However, benzoyl protecting groups were trapped by addition to the carbonyl rather than the anomeric carbon. This problem was overcome by using a pivaloyl (OCOt-butyl) protecting groups instead (Fig. 16). N-Linked ␤-amino acid conjugates of sugars were reported by Kunz et al. [35]. Mannich reaction of bis(O-trimethylsilyl) ketene acetals and N-galactosylimines gave ␣-branched ␤-amino acids in high diastereomeric ratios (Fig. 17). 1-Amino-1-deoxyglucuronic acids have also been prepared. In one account, 1,2,3,4-tetra-O-acetyl glucuronic acid was treated with iodine at 0⬚C, followed by the addition of trimethylsilyl azide. The acetates were removed with hydrazine and the azide was reduced with 1,2-ethanedithiol [36]. Other azido sugars have been used in the preparation of sugar amino acids. For example, Fleet and coworkers converted the isopropylidene of D-glucuronolactone into an ␣-azido lactone, which was subsequently reduced to an ␣-amino lactone [37]. Removal of the acetal protecting group unmasked the aldehyde, which underwent reductive amination and hydrolysis of the lactone to give a trihydroxypipecolinic acid (Fig. 18). Fleet developed this elegant methodology as a rapid entry into several picolinic acid derivatives, which are naturally occurring L-amino acids with known biological activity. These molecules further illustrate the difficulty in defining the difference between sugar amino acids and ␣-amino acids. In this case, they are arguably indistinguishable. C.

Syntheses of Neuraminic Acid Derived Amino Acids

In his review on complex carbohydrates, Nathan Sharon stated: ‘‘Neuraminic acid is a nine carbon sugar acid, with an amino group in the molecule’’ [38]. This simple and obvious declaration was an epiphany for Gervay-Hague, whose prior experience with NeuAc had been limited to the challenges of O-glycosylations. The realization that neuraminic acid is an amino acid presented new possibilities for its utilization in the production of novel materials. Sharon’s writing inspired a program in the Gervay-Hague laboratories directed to the synthesis of amino acid equivalents derived from neuraminic acid. Since N-acetyl neuraminic acid is the most abundant form of the sialic acids, it was important to first establish a method for removing the acyl group. This turned out to be remarkably difficult, since both acid and base hydrolyses led to retro aldol products rather than the desired amine. Borrowing from work published by Roy and Pon [39], the ␤-methyl glycoside of NeuAc was prepared and successfully N-deacylated using 2 N sodium hydroxide at 100⬚C for 48 h. Realizing that these conditions would not be suitable for a wide variety of substrates, the investigators sought milder conditions. After much experimentation, it was found that treatment of the amide with tert-butoxycarbonyl (Boc) anhydride followed by mild hydrolysis with sodium methoxide provided the Boc-protected sugar amino acid in high yield [40]. This general strategy provided a reliable route to several N-protected neuraminic acid analogs, including ␣- and ␤-O-methyl glycosides and a 2,3-dehydro derivative (Fig. 19). 2-Deoxy analogs of N-protected neuraminic acids were also prepared. The ␤hydrido derivative was obtained by hydrogenation of N-acetyl-N-Boc-4,7,8,9-tetra-

Pyranosyl

Figure 16

Figure 17 507

508

Gervay-Hague and Weathers

Figure 18

Pyranosyl

509

Figure 19

O-acetyl-2,3-dehydroneuraminic acid followed by treatment with sodium methoxide. The ␣-hydrido analog was synthesized from N-acetyl-2,4,7,8,9-penta-O-acetyl neuraminic acid benzyl ester under the action of hydrogen iodide in acetic acid. In the reaction, the anomeric acetate was converted to an iodide, which subsequently underwent reductive elimination to give the ␣-hydrido compound under thermodynamic conditions (Fig. 20) [41]. These combined syntheses efficiently provided five new sugar amino acids with functionality suitable for typical peptide coupling reactions. The foregoing synthetic efforts, targeting the design and syntheses of novel carbohydrate-based amino acids over the past 45 years, have established a solid foundation for an entirely new area of scientific inquiry—the use of sugar amino acids in the synthesis of novel and unnatural products. IV.

SYNTHESIS OF SUGAR AMINO ACID CONJUGATES

The first synthetic reports of sugar amino acid conjugation appeared in the early 1970s. Most of those studies were directed to the synthesis of nucleoside antibiotics containing hexuronic acids conjugated to ␣-amino acids [11]. The first suggestions that unnatural compounds were of synthetic interest appeared in two separate papers published in 1976. Yoshimura and coworkers demonstrated that a hexosaminuronic acid could be conjugated to a glucosamine via an amide bond [42]. These scientists were clearly influenced by earlier work on nucleoside antibiotics. Nonetheless, they

510

Gervay-Hague and Weathers

Figure 20

did make amido-bonded disaccharides for the first time. Because one sugar was an acid and the other an amine, these systems were self-terminating (Fig. 21). Fuchs and Lehmann first demonstrated that homo-oligomers of sugar amino acids could be prepared. In their reaction, a glucose-derived sugar amino acid possessing an anomeric carboxylate and a C4 amino group was polymerized under basic conditions [43]. After the reaction mixture had been heated for 4 h at 100⬚C, a waterinsoluble precipitate formed (Fig. 22). The products were not rigorously characterized, but there was evidence that the precipitate was composed of dimers, trimers, and tetramers. Although these two reports clearly demonstrated the feasibility of using sugar amino acids in the production of novel and unnatural compounds, it was nearly 20

Figure 21

Pyranosyl

511

Figure 22

years before an aggressive exploration of this rich field of chemistry began. The art that followed the pioneering work of Yoshimura, Fuchs, and Lehmann can be classified into two categories: (1) syntheses involving mixed congeners, having either self-terminating residues and/or a combination of sugar amino and ␣-amino acids, and (2) homo-oligomeric syntheses composed of one sugar amino acid joined in consecutive amido linkages. A.

Syntheses of Mixed Sugar Amino Acid Conjugates

In 1994 Kessler used a sugar amino acid as a dipeptide isostere [28]. He and coworkers proposed that the C-glycoside amine of glucuronic acid would have torsion angles comparable to those of dipeptides composed of glycine and either serine or threonine. They incorporated the sugar amino acid into different peptides and measured their ability to inhibit the release of growth hormone as somatostatin analogs. The results of these experiments suggested that sugar-based peptides hold promise as potent peptidomimetics. Toth and coworkers also prepared sugar amino acid derived peptides in route to modified enkephalins [44]. Two conjugates were prepared and studied. An azido glucuronic acid was immobilized on a trityl resin, and, after reduction, it was further conjugated to five ␣-amino acids (Leu-Phe-Gly-Gly-Tyr). A second derivative consisted of a sugar amino acid dimer conjugated to the same five ␣-amino acids. The pharmacology of these compounds was evaluated and the analog containing only one sugar amino acid was shown to be a potent and selective agonist of the ␦-opioid receptor (Fig. 23). Very recently, amino deoxyglucuronic acids were used in the construction of disaccharide libraries [45]. In the report, methyl 3-azido-3-deoxy-4-O-methyl-␤-Dglucuronic acid was prepared and glycosidically linked to a protected glucosamine, yielding a disaccharide with both amino and carboxamide functionalities. The azide was subsequently reduced and reacted with several isocyanates, providing libraries of ␤-linked disaccharides (Fig. 24). One incentive for making amido-linked sugars as mimics of glycosidically linked disaccharides was the possibility of effecting enzymatic resistance to glycosidic bond cleavage. Sabesan nicely illustrated this point in his work on amide-linked disaccharides containing NeuAc [46]. The 2-azido sugar was reduced and subsequently condensed with an activated galacturonic acid to afford only the ␣-aminolinked disaccharide. This compound was designed to be an isostere of the naturally occurring ␣-D-NeuAc-(2-6)-␤-D-Gal glycosidic linkage (Fig. 25).

512

Figure 23

Figure 24

Gervay-Hague and Weathers

Pyranosyl

513

Figure 25

Gervay and coworkers used two different strategies to further explore the incorporation of NeuAc into mixed sugar amino acid conjugates. In the first report, sialyllactones were prepared and condensed with glycine under thermal conditions [47]. Both five- and six-membered lactones were reacted, anticipating that the ring strain in the 1,4-lactones would promote the reaction. Unfortunately, this was not the case, and only low yields of the desired conjugates were obtained. The six-membered lactones gave improved yields, but, in general, the reactions were not efficient. In later studies, NeuAc was efficiently conjugated to several ␣-amino acids using BOP and HOBT activation of the acid followed by addition of an ␣-amino acid [48]. This protocol provided a number of sugar amino acid peptides in nearly quantitative yields (Fig. 26). B.

Oligomeric Sugar Amino Acid Syntheses

In 1995 Nicolaou and coworkers initiated a renaissance in the construction of sugar amino acid conjugates with their synthesis of carbonucleotoids [49]. Although they did not prepare amide-linked carbohydrates, they did introduce the term ‘‘carbopeptoid’’ to designate such materials. Shortly thereafter, a number of papers directed toward the synthesis of carbopeptoids appeared. One of the earliest was reported by Wessel et al., who used nor-muramic acid derivatives and condensed them in solution by means of 2-chloro-4,6-dimethoxy-1,3,5-triazine (CDMT) in DMF to construct a tetramer (Fig. 27). Wessel later used solid phase synthesis to prepare a tetramer of amido-linked 2-amino-2-deoxyglucuronic acids [50]. In that disclosure, benzyl 2-fluoren-9-ylme-

514

Gervay-Hague and Weathers

Figure 26

Pyranosyl

515

Figure 27

516

Gervay-Hague and Weathers

thoxycarbonyl (Fmoc) amino-2-deoxy-␣-D-glucuroniside was immobilized on Rink resin. After deprotection of the Fmoc group with piperidine, an Fmoc-protected amino glucuronic acid was added and amide formation was promoted by TATU [O(7-azabenzotriazol-1-yl)-1,1,3,3-tetramethyluronium tetrafluoroborate]. Iteration of this procedure culminated in the construction of a tetramer. Some difficulties encountered in immobilizing the first sugar residue may have been due to steric congestion at the resin surface. Subsequent couplings readily occurred, suggesting that the first residue may have served to extend the reaction centers away from the polymer bead (Fig. 28). The Ichikawa group also made amido-linked oligomers derived from amino Cglycoside carboxylic acids [51]. A Boc-protected ␤-sugar amino acid was conjugated to phenylalanine and subsequently deprotected to give the free amine, which was coupled to another Boc-protected monomer. Diethylphosphoryl cyanide and triethylamine were used to activate the acid for coupling in the solution phase. A tetramer was synthesized and subsequently sulfated to increase solubility and to introduce negative charge (Fig. 29). This material was designed as a potential inhibitor of HIV replication, since it was known that sulfated polymeric carbohydrates inhibit HIV entry into T cells. Syncytium formation was completely blocked when CD4 cells were infected with

Figure 28

Pyranosyl

517

Figure 29

518

Gervay-Hague and Weathers

HIV in the presence of 50 ␮M concentrations of the sulfated tetramer. Later, an even more potent inhibitor (IC50 = 1 ␮M) was prepared from a sulfated tetramer comprised of amido-linked C7-amino-C1 heptonic acid monomers [52]. Goodnow et al. targeted anti-sense agents in their syntheses of amido-linked pyranosyl nucleosides [53]. Carbohydrate-based nucleic acids were sought because peptide nucleic acids self-aggregate, creating solubility and cellular entry problems. It was hoped that the rigid backbone of the amide-linked sugars would provide a scaffold allowing the bases to align in a Watson–Crick base-pairing fashion with DNA and/or RNA. Solid phase methods were used to prepare two nucleoside analogs of 2-amino-2-deoxyglucuronic acid and incorporate them into oligomers. Both the 10-residue and the 13-residue oligomers were water soluble (Fig. 30). The binding affinities of the two oligomers for selected DNA and RNA oligomers were determined from duplex formation and melt temperature measurements. The 10-mer bound complementary antiparallel DNA approximately 5⬚C lower than a known peptide nucleic acid, strongly supporting the Watson–Crick model. The amide-linked oligomers showed binding affinities for DNA that were greatly enhanced over glycosidically linked analogs. The van Boom laboratories investigated the possibility of replacing glycosidic linkages with amide linkages to enhance the biological activity of a phytoalexin elicitor [54]. Earlier, they had shown that an amidically terminated, branched hexasaccharide was as potent as the naturally occurring sugar terminating in a reduced sugar moiety. From a series of structure–activity relationship studies, they deduced that the conformation of the glycosidic linkages along the backbone was critical for the observed activity. Based on this information, they questioned whether an amidically linked backbone structure would show improved phytoalexin elicitor activity. A glycosidically linked dimer was converted to an amino acid equivalent and amidated with a terminating sugar at the anomeric position. The C6 carboxyl functionality was readied for chain elongation with a monomeric sugar amino acid, which was subsequently condensed with a glycosidically linked dimeric sugar amino acid. Finally, the carboxy terminus was capped (Fig. 31). Subsequent biological studies showed that the amide-linked analogs were completely inactive. These results, when contrasted with the work of Ichikawa and Goodnow, suggest that amide-linked carbohydrates may serve better as unique molecular scaffolds than as glycosidic linkage mimics. V.

SYNTHESES AND STRUCTURAL CHARACTERIZATION OF AMIDO-LINKED OLIGOMERS WITH STABLE SECONDARY STRUCTURE

A.

Non-Carbohydrate-Based Materials

Although this chapter is focused on carbohydrate-based compounds, it is important to briefly describe the pioneering work of Gellman and Seebach directed to the syntheses of amido-linked oligomers derived from ␤- and ␥-amino acids that are not based on carbohydrates. These researchers and others [55] have engineered systems that adopt stable helical, sheet, and turn conformations in solution. Surprisingly, in many cases as few as four residues is sufficient to stabilize the conformation. This completely contrasts with oligomers derived from ␣-amino acids which typically require many more residues before conformational stability is established [56].

Pyranosyl

519

Figure 30

Gellman introduced the term ‘‘foldamer’’ to refer to ‘‘any polymer with strong tendency to adopt a specific compact conformation’’ [57]. Early work from his group exemplified the rational design process. Computer-assisted analysis of several amidolinked ␤-amino acids led to the proposal that trans 2-amino hexanoic acid and trans 2-amino pentanoic acid would form stable helical structures. Modeling predicted that the hexanoic system would form a 14-helix and the pentanoic system a 12-helix (Fig. 32). These structures were synthesized and their conformations were experimentally verified using a combination of NH/ND exchange rates, circular dichroism (CD), and X-ray crystallography. Seebach et al. reported the synthesis of ␤- and ␥-peptides with stable secondary structures [58] and, independently, Hanessian et al. reported the synthesis of ␥-peptides that adopt helical structures in solution [59]. These combined studies showed important structural characteristics for ␤- and ␥-peptides in contrast with ␣-peptides. For example, ␣-peptides adopt helical conformations stabilized by a 13-membered hydrogen bond between the carbonyl of the amino terminus and the amide NH of the fourth residue toward the C-terminus (3.6 residues per turn). ␤-Peptides formed

520

Gervay-Hague and Weathers

Pyranosyl

Figure 31

521

522

Gervay-Hague and Weathers

Figure 32

a 14-membered hydrogen bond between the carbonyl of the C-terminus and the NH of the amide two residues toward the N-terminus (3 residues per turn) (Fig. 33) [60]. Seebach’s ␥-peptides formed a 14-membered hydrogen bond between the carbonyl of the N-terminus and the carbonyl two residues toward the C-terminus (2.6 residues per turn) [61]. Of the many elegant studies reported by Gellman and coworkers, two are particularly relevant to the discussion at hand. The first was the formation of watersoluble ␤-amino acid hexamers with stable helices in water [62]. Four different systems containing positively charged amine functionalities were prepared (Fig. 34). The charge promoted water solubility and, at the same time, prevented aggregation [63]. Structural analyses using the aforementioned techniques confirmed that these oligomers adopted a 14-helix conformation in aqueous solution. Demonstrating that conformationally stable, water-soluble candidates could be prepared was the first step toward using these new materials to solve biologically relevant problems. The Gellman laboratories later reported the chemical synthesis of a 17-mer based on the 12-helix forming trans-2-aminocyclopentanoic acids [64]. Charged amine functionalities in combination with neutral monomer units were incorporated into the oligomer to provide an amphiphilic secondary structure (Fig. 35). When tested for bactericidal and bacteriostatic activity against a vancomycin-resistant bacterial strain, this material exhibited activity similar to that of a known 23-mer

Pyranosyl

Figure 33

523

524

Gervay-Hague and Weathers

Figure 34

Pyranosyl

Figure 35

525

526

Gervay-Hague and Weathers

Figure 36

␣-peptide inhibitor. Moreover, the ␤-peptide was less destructive to red blood cells than was the ␣-peptide. In addition to their many structural studies on ␤- and ␥-peptides, Seebach and coworkers showed that human receptors accept ␤-peptides that mimic natural ␣peptides [65]. They modeled a cyclic ␤-tetrapeptide placing phenylalanine, tryptophan, lysine, and threonine-like side chains in key positions designed to mimic a cyclic ␣-octapeptide derived from somatostatin (Fig. 36). Standard peptide chemistry that relied on Boc and benzyl protecting group strategies was used in the synthetic construction of the tetrapeptide. The protected form of the cyclic peptide was not soluble under normal conditions required for hydrogenolysis; however, the researchers were able to reduce the benzyl protecting group in THF with 6 equiv of lithium chloride. Radioligand binding assays were conducted to measure human somatostatin receptor affinities for the ␤-tetrapeptide relative to the ␣-octapeptide. The ␤-peptide was active in the micromolar range, whereas only nanomolar concentrations were required for the ␣-peptide. Although the ␤-peptide was less active, it was shown to be peptidase resistant, an important consideration in terms of pursuing related compounds in the future. B.

Carbohydrate-Based Oligomers with Secondary Structure

Amide-linked carbohydrate-based oligomers with defined secondary structures were first engineered by Szabo et al. [66]. ␤-Methoxy neuraminic acid was used as the monomer unit, and solid phase techniques were employed in the construction of oligomers ranging from dimer to octamer. Caproamide was incorporated at the carboxy terminus to prevent solvation of the terminal residue, a phenomenon commonly referred to as fraying (Fig. 37). The hydrophobic end group provided directionality

Pyranosyl

527

Figure 37

to the growing chain, fashioned after a membrane-bound oligomer. It was anticipated that these combined factors would stabilize backbone hydrogen bonding interactions and lead to relatively small molecules with stable secondary structures. Caproamide was immobilized on Rink resin, and Fmoc-protected ␤-methoxy neuraminic acids were condensed by using BOP and DIEA in N-methylpyrrolidinone (NMP). Flaherty et al. had earlier reported that solution phase coupling of neuraminic acids was sluggish, and only poor yields were realized (48 h, 25–30%) [40]. In contrast, the solid phase syntheses were completed in a few hours, as determined by the Kaiser test, and combined yields ranged from 44 to 55% indicating that each coupling step was far more efficient (Fig. 38). ␤-Methoxy NeuAc was used as a sugar amino acid in the construction of oligomers for three primary reasons. First, it was readily available from a naturally occurring ␦-amino acid. Second, it was hypothesized that the trihydroxy side chain would increase water solubility in higher order oligomers. Finally, O-glycoside oligomers of NeuAc were known to have stable secondary structure in aqueous solution, and it seemed possible that amide-linked analogs would exhibit similar properties [67]. The second hypothesis was proved by the chemical syntheses of the oligomers, which were all shown to be highly water soluble. The oligomer secondary structures were probed using a combination of NH/ND exchange studies and CD, patterned after the studies of Gellman and Seebach. The exchange studies were originally performed in dimethyl sulfoxide (DMSO) because they were too rapid to be observed in water. The half-lives of NH/ND exchange were determined for the series of oligomers ranging from dimer to octamer. The dimer exchanged rapidly (half-life ⬃ 30 min). Two different exchange rates were observable for the trimer; the amino terminus amide exchanged fastest, on the same order as the dimer. The half-life of the internal amide was approximately 6 h. This was also true of the tetramer; the reducing end amide exchanged relatively quickly, but the internal amides took several hours. The rapid exchange of the reducing end amide was attributed to fraying. For the most part, as the oligomer length increased, the exchange rates of the internal amides slowed (the octamer was the slowest).

528

Gervay-Hague and Weathers

Figure 38

CD spectra were recorded in water at neutral pH, and they correlated with the NH/ND studies (performed in DMSO) surprisingly well. The dimer, lacking an internal amide, did not show a signature CD. The other oligomers displayed an absorbance maximum at ⬃200 nm with a zero crossover at ⬃212 nm and a minimum at 200 nm, returning to zero at ⬃240 nm. The intensity of the maximum at 200 nm consistently increased with increasing length. These data completely correlated increasing secondary structural stability with increasing oligomer length. Shortly after this report, Fleet and coworkers reported that furanosyl-derived amido-linked sugar amino acids also adopt stable secondary structures in solution (Fig. 39) [68]. Gregar and Gervay-Hague recently prepared a series of ␣-methoxy NeuAc amido-linked oligomers and compared them to the ␤-methoxy series [69]. The first noticeable difference was the NH/ND exchange rates, which were slow enough in water (pH = 3.0 phosphate buffer) to be measured. The exchange rates were fast for dimer and trimer, and essentially disappeared before the NMR spectra were acquired.

Pyranosyl

529

Figure 39

The rates slowed for the tetramer, which exchanged similarly to the pentamer through octamer. CD spectra were also recorded in pH = 3.0 phosphate buffer solution. The CD signatures of these compounds showed a reverse trend from the ␤-methoxy series, with an absorbance minimum at ⬃195 nm and a maximum at ⬃230 nm, returning to zero at ⬃260 nm (Fig. 40). Interestingly, there was a dramatic increase in the intensity of the absorbencies in going from trimer to tetramer, but the spectra of the longer oligomers (5-mer through 8-mer) were similar to the tetramer, which correlated exactly with the NH/ND studies. The CD and NMR studies indicated that the ␣-methoxy series formed stable secondary structures with as few as four residues.

Figure 40 CD spectra of ␣-methoxy NeuAc amido-linked oligomers recorded in phosphate buffer solution (pH 3).

530

Gervay-Hague and Weathers

CD spectra of ␤-methoxy NeuAc amido-linked oligomers recorded in phosphate buffer solution (pH 3).

Figure 41

For comparison purposes, CD spectra of the ␤-methoxy series were recorded in pH 3.0 phosphate buffer solution. In contrast to the ␣-methoxy series, the octamer was clearly distinguishable from shorter oligomers, suggesting that it is significantly more stable (Fig. 41). NH/ND exchange studies in pH 3.0 buffer showed that all oligomers exchanged rapidly with the exception of the octamer. These combined studies clearly showed that both ␣- and ␤-methoxy neuraminic acid-derived oligomers adopt stable secondary conformations in aqueous solution. However, the ␣ series required only four residues for stability, whereas the ␤-series required eight. VI.

THE FUTURE

There are many intended applications of the scientific inquiries described in this chapter, including (1) use of unnatural amino acids in the construction of libraries, (2) use of the oligomers as peptido mimetics, and (3) incorporation of the oligomers in artificial protein engineering (see below). Artificial amino acids have routinely been incorporated into natural peptides in pharmacophore drug design. However, rapid elimination half-lives and biochemical degradation often contribute to poor efficacy for many of these compounds. The development of nonnatural peptides may lead to drug candidates with improved bioavailability profiles, since they would be resistant to biochemical degradation through the action of proteases and peptidases. Application to protein engineering requires the interaction of secondary structures to form tertiary structures. DeGrado and coworkers recently accomplished de novo design of helical bundles composed of natural ␣-peptides [70]. In earlier work, they described three principles that guide the design process: (1) complementary packing within the hydrophobic core, (2) specific interfacial interactions to promote

Pyranosyl

531

interhelical associations, and (3) core and interfacial interactions to destabilize potential alternate conformations [71]. The distinctive packing of the amino acid side chains in the core of the bundle gives rise to a coiled coil, or superhelix. Coiled coils have biological relevance. For example, they are found in fibrous proteins such as fibrinogen and keratin [72] and are structural components of viral proteins [73]. De novo design of these macromolecular structures offers insight into the process of protein folding, in addition to providing new materials. The same would be true of de novo design of coiled coils composed of nonnatural peptides. As of this writing, 3⬚ structure in nonnatural peptides has not been reported. However, Gellman has offered a blueprint for the design of coiled coils based upon ␤-peptides [74]. The possibility of creating carbohydrate-based ␦-peptides that adopt stable 3⬚ structures in solution is also of great interest. Determining the conditions that promote oligomer aggregation is an important step toward engineering artificial proteins. Beyond the obvious biomedical uses as drug delivery agents, biosensors, and affinity materials, compounds with unique structural motifs could serve as chemical catalysts. For example, just as natural proteins provide a hydrophobic medium in aqueous solution, properly designed carbohydrate-based proteins could provide a hydrophilic environment in an organic medium, opening up new possibilities for chemical transformations. It is clear that an idea born nearly 50 years ago has evolved into a rich science at the frontiers of synthetic design, structure elucidation, and materials production. Over the years, old chemistry was rediscovered and applied in new directions resulting in the birth of entirely new ideas. Many factors have contributed to the successes, including the introduction of combinatorial chemistry, the availability of biological screening assays, and the development of new and improved technologies for structure determination. Pyranosyl sugar amino acid conjugates will surely continue to provide a fertile field for scientific inquiry and discovery.

REFERENCES 1. 2.

Kornfeld, R.; Kornfeld, S. In The Biochemistry of Glycoproteins and Proteoglycans, Vol. XIV, William J. Lennarz, ed. Plenum Press: New York, 1980, pp. 1–33. Several other sources have also been identified: Torii, M.; Sakakiba, K.; Kuroda, K. Eur. J. Biochem. 1973, 37, 401–405. Karakawa, W. W.; Kane, J. A. J. Immunol. 1975, 114, 310–315. Liau, D.; Melly, M. A.; Hash, J. H. J. Bacteriol. 1974, 119, 913–922. Eguchi, H.; Kaya, S.; Araki, Y. Carbohydr. Res. 1992, 231, 147–158. Romanows, E.; Reinhold, V. Eur. J. Biochem. 1973, 36, 160–166. Reddy, G. P.; Hayat, U.; Xu, Q. W.; Reddy, K. V.; Wang, Y. H.; Chiu, K. W.; Morris, J. G.; Bush, C. A. Eur. J. Biochem. 1998, 255, 279–288. Williamson, A. R.; Zamenhof, S. J. Biol. Chem. 1963, 238, 2255–2258. Jeanes, A.; Burton, K. A.; Cadmus, M. C.; Knutson, C. A.; Rowin, G. L.; Sandford, P. A. Nat. New Biol. 1971, 233, 259. Fournier, J. M.; Vann, W. F.; Karakawa, W. W. Infect. Immun. 1984, 45, 87–93. Lee, J. C.; Michon, F.; Perez, N. E.; Hopkins, C. A.; Pier, G. B. Infect. Immun. 1987, 55, 2191–2197. Moreau, M.; Richards, J. C.; Fournier, J. M.; Byrd, R. A.; Karakawa, W. W.; Vann, W. F. Carbohydr. Res. 1990, 201, 285–297. Ichihara, N.; Ishimoto, N.; Ito, E. FEBS Lett. 1974, 40, 309–311. Konig, H.; Kandler, O. Arch. Microbiol. 1979, 123, 295–299. Kochetkov, N. K.; Kholodkova, E. V.; Stanislavsky, E. S. Eur. J. Biochem. 1987, 167, 549–561. Wu, T. C. M.; Park, J. T. J. Bacteriol. 1971, 108, 874.

532 3.

4. 5.

6.

7.

8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.

Gervay-Hague and Weathers Clark, W. R.; McLaughlin, J. J. Biol. Chem. 1958, 230, 81. Williamson, A. R.; Zamenhof, S. J. Biol. Chem. 1963, 238, 2255. Hanessian, S.; Haskell, T. H. J. Biol. Chem. 1964, 239, 2758. Heyns, K.; Kiessling, G.; Lindenberg, W.; Paulsen, H.; Webster, M. Chem. Ber. 1959, 92, 2435. Perkins, H. R. Biochem. J. 1963, 86, 475. Williamson, A. R.; Zamenhoff, S. J. Biol. Chem. 1963, 238, 2255. Several other sources were later cited: Braneforshelander, P.; Kenne, L.; Lindberg, B.; Petersson, K.; Unger, P. Carbohydr. Res. 1981, 97, 285–291. Tsui, F. P.; Schneerson, R.; Boykins, R. A.; Karpas, A. B.; Egan, W. Carbohydr. Res. 1981, 97, 293–306. Karakawa, W. W.; Kane, J. A. J. Immunol. 1972, 108, 1199. Watson, P. R.; Sandford, P. A.; Burton, K. A.; Cadmus, M. C.; Jeanes, A. Carbohydr. Res. 1976, 46, 259–265. Fisher, M. W.; Devlin, H. B.; Erlandson, A. L. Nature 1963, 199, 1074. Haskell, T. H.; Hanessian, S. Nature 1963, 199, 1075. Haskell, T. H.; Hanessian, S. Biochim. Biophys. Acta 1964, 83, 35. Hanessian, S.; Haskell, T. H. J. Biol. Chem. 1964, 9, 2758. Subsequently several other examples were reported: Hofmann, P.; Jann, B.; Jann, K. Carbohydr. Res. 1985, 139, 261–271. Iwasaki, H.; Araki, Y.; Kaya, S.; Ito, E. Eur. J. Biochem. 1989, 178, 643–648. Dengler, T.; Jann, B.; Jann, K. Carbohydr. Res. 1986, 150, 233–240. Lindberg, B. Adv. Carbohydr. Chem. Biochem. 1990, 48, 279–318. Cummins, C. S.; White, R. H. J. Bacteriol. 1983, 153, 1388. Dimitriev, B. A.; Lvov, V.; Tochtamysheva, N. V.; Shashkov, A. S.; Kochetkov, N. K.; Jann, B.; Jann, K. Eur. J. Biochem. 1983, 134, 517. Fox, J. J.; Kuwada, Y.; Watanabe, K. A. Tetrahedron Lett. 1968, 6029. Watanabe, K. A.; Kotick, M. P.; Fox, J. J. Chem. Pharm. Bull. 1969, 17, 416. Sakata, K.; Sakurai, A.; Tamura, S. Agric. Biol. Chem. 1973, 37, 697; 1974, 38, 1883; 1975, 39, 885. Lichtenthaler, F. W.; Morino, T.; Winterfeldt, W.; Sanemitsu, Y. Tetrahedron Lett. 1975, 41, 3527, and references therein. Knapp, S. Chem. Rev. 1995, 95, 1859. Knirel, Y. A.; Vinogradov, E. V.; Shashkov, A. S.; Dmitriev, B. A.; Kochetkov, N. K.; Stanislavsky, E. S.; Mashilova, G. M. Eur. J. Biochem. 1986, 157, 129. Kenne, L.; Lindberg, B.; Schweda, E. Carbohydr. Res. 1988, 180, 285. Heyns, K.; Paulsen, H. Chem. Ber. 1955, 88, 188. Heyns, K.; Beck, M. Chem. Ber. 1957, 90, 2443. Marsh, C. A.; Levvy, G. A. Biochem. J. 1958, 68, 617. Weidmann, H.; Zimmerman, H. K., Jr. Ann. Chem. 1961, 639, 198. Yoshimura, J.; Sato, T.; Ando, H. Bull Chem. Soc. Jpn. 1969, 42, 2352. Yoshimura, J.; Sakai, H.; Oda, N.; Hashimoto, H. Bull. Chem. Soc. Jpn. 1972, 45, 2027. Tsuji, A.; Kinoshita, T.; Maeda, M. Chem. Pharm. Bull. 1968, 16, 539. Paulsen, H.; Koebernick, W.; Autschbach, E. Chem. Ber. 1972, 105, 1524. Ogawa, T.; Akatsu, M.; Matsui, M. Carbohydr. Res. 1975, 44, C22. Kotick, M. P.; Klein, R. S.; Watanabe, K. A.; Fox, J. J. Carbohydr. Res. 1969, 11, 359. Paulsen, H.; Koebernick, W.; Autschbach, E. Chem. Ber. 1972, 105, 1524. Fuchs, E.-F.; Lehmann, J. Carbohydr. Res. 1975, 45, 135–141. Graf von Roedern, E.; Kessler, H. Angew. Chem. Int. Ed. Engl. 1994, 33, 687. Fuchs, E.-F.; Lehmann J. Chem. Ber. 1975, 108, 2254–2260. Kim, K.-I.; Hollingsworth, R. I. Tetrahedron Lett. 1994, 35, 1031. Sakai, N.; Ohfune, Y. J. Am. Chem. Soc. 1992, 114, 998. Ohfune, Y.; Kurokawa, N. Tetrahedron Lett. 1984, 25, 1587. Axon, J. R.; Beckwith, A. L. J. J. Chem. Soc. Chem. Commun. 1995, 549. Simchen, G.; Pu¨rkner, E. Synthesis 1990, 525. Kunz, H.; Burgard, A.; Schanzenbach, D. Angew. Chem. Int. Ed. Engl. 1997, 36, 386.

Pyranosyl 36. 37. 38. 39. 40.

41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58.

59. 60. 61.

62. 63. 64.

533

Drouillat, B.; Kellam, B.; Dekany, G.; Starr, M. S.; Toth, I. Bioorg. Med. Chem. Lett. 1997, 17, 2247. Bashyal, B. P.; Chow, H.-F.; Fleet, G. W. J. Tetrahedron Lett. 1986, 27, 3205. Sharon, N. Complex Carbohydrates Their Chemistry, Biosynthesis and Function. Addison-Wesley, London, 1975, p. 16. Roy, R.; Pon, R. A. Glycoconjugate J. 1990, 7, 3. Gervay, J.; Flaherty, T. M.; Nguyen, C. N. Tetrahedron Lett. 1997, 38, 1493. Nguyen, C. Master’s thesis, University of Arizona, 1998. Flaherty, T. M. Ph.D. dissertation, University of Arizona, 1997. Gervay, J.; Gregar, T. Q. Tetrahedron Lett. 1997, 34, 5921. Yoshimura, J.; Ando, H.; Sato, T.; Tsuchida, S. Bull. Chem. Soc. Jpn. 1976, 49, 2511. Fuchs, E.-R.; Lehmann, J. Carbohydr. Res. 1976, 49, 267. Drouillat, B.; Kellam, B.; Dekany, G.; Starr, M. S.; Toth, I. Bioorg. Med. Chem. Lett. 1997, 17, 2247. Silva, D. J.; Wang, H.; Allanson, N. M.; Jain, R. K.; Sofia, M. J. Org. Chem. 1999, 64, 5926. Sabesan, S. Tetrahedron Lett. 1997, 38, 3127. Gervay, J.; Ramamoorthy, P. S.; Mamuya, N. N. Tetrahedron 1997, 53, 11039. Ramamoorthy, P. S.; Gervay, J. J. Org. Chem. 1997, 62, 7801. Nicolaou, K. C.; Flo¨rke, H.; Egan, M. G.; Barth, T.; Estevez, V. A. Tetrahedron Lett. 1995, 36, 1775. Mu¨ller, C.; Kitas, E.; Wessel, H. P. J. Chem. Soc. Chem. Commun. 1995, 2425. Suhara, Y.; Hildreth, J. E. K.; Ichikawa, Y. Tetrahedron Lett. 1996, 37, 1575–1578. Suhara, Y.; Ichikawa, M.; Hildreth, J. E. K.; Ichikawa, Y. Tetrahedron Lett. 1996, 37, 2549–2552. Goodnow, R. A. Jr.; Tam, S.; Pruess, D. L.; McComas, W. W. Tetrahedron Lett. 1997, 38, 3199. Timmers, C. M.; Turner, J. J.; Ward, C. M.; van der Marel, G. A.; Kouwijzer, M. L. C. E.; Grotenhuis, P. D. J.; van Boom, J. H. Chem. Eur. J. 1997, 3, 920. Hanessian, S.; Luo, X.; Schaum, R.; Michnick, S. J. Am. Chem. Soc. 1998, 120, 8569. Rohl, C. A.; Baldwin, R. L. Biochemistry, 1994, 31, 7760. Gellman, S. H. Acc. Chem. Res. 1998, 31, 173–180. (a) Seebach, D.; Ciceri, P. E.; Overhand, M.; Jaun, B.; Rigo, D.; Oberer, L. U.; Amstutz, R.; Widmer, H. Helv. Chim. Acta 1996, 79, 2043–2066. (b) Seebach, D.; Matthews, J. L.; Meden, A.; Wessels, T.; Baerlocher, C.; McCusker, L. B. Helv. Chim. Acta 1997, 80, 173–182. (c) Seebach, D.; Abele, S.; Sifferlen, T.; Ha¨nggi, M.; Gruner, S.; Seiler, P. Helv. Chim. Acta 1998, 81, 2218–2243. (d) Abele, S.; Guichard, G.; Seebach, D. Helv. Chim. Acta 1998, 81, 2141. (e) Gademann, K.; Jaun, B.; Seebach, D.; Perozzo, R.; Scapozza, L.; Folkers, G. Helv. Chim. Acta 1999, 82, 1–11. (a) Hannesian, S.; Luo, X.; Schaum, S.; Michnick, S. J. Am. Chem. Soc. 1998, 120, 8569. (b) Hannessian, S.; Luo, X.; Schaum, R. Tetrahedron Lett. 1999, 40, 4925–4929. Appella, D. H.; Barchi, J. J., Jr.; Durell, S. R.; Gellman, S. H. J. Am. Chem. Soc. 1999, 121, 2309–2310. Seebach, D.; Abele, S.; Gademann, K.; Guichard, G.; Hintermann, T.; Jaun, B.; Matthews, J. L.; Schreiber, J. V. Helv. Chim. Acta 1998, 81, 932–982. (b) Seebach, D.; Abele, S.; Gademann, K.; Jasun, B. Angew. Chem. Int. Ed. Engl. 1999, 38, 1595–1597. Appella, D. H.; Barchi, J. J., Jr.; Durell, S. R.; Gellman, S. H. J. Am. Chem. Soc. 1999, 121, 2309. For a related study on water-soluble ␤-peptides see: Gung, B. W.; Zou, D.; Stalcup, A. M.; Cottrell, C. E. J. Org. Chem. 1999, 64, 2176. Porter, E. A.; Wang, X.; Lee, H.-S.; Weisblum, B.; Gellman, S. H. Nature, 2000, 404, 565.

534 65. 66. 67. 68.

69. 70. 71. 72. 73. 74.

Gervay-Hague and Weathers Gademann, K.; Ernst, M.; Hoyer, D.; Seebach, D. Angew. Chem. Int. Ed. Engl. 1999, 38, 1223. Szabo, L.; Smith, B. L.; McReynolds, K. D.; Parril, A. L.; Morris, E. R.; Gervay, J. J. Org. Chem. 1998, 63, 1074. Brisson, J.-R.; Baumann, H.; Imberty, A.; Pe´rez, S.; Jennings, H. J. Biochemistry 1992, 31, 4996–5004. Fleet and coworkers have reported a number of successes in this area. The reader is referred to the rich literature provided by this laboratory, which is outside the scope of the present chapter: Estevez, J. C.; Arcron, H.; Wormald, M. R.; Brown, D.; Fleet, G. W. J. Tetrahedron Lett. 1994, 47, 8885. Estevez, J. C.; Arcron, H.; Wormald, M. R.; Brown, D.; Fleet, G. W. J. Tetrahedron Lett. 1994, 47, 8889. Kru¨lle, T. M.; Watson, K. A.; Gregoriou, M.; Johnson, L. N.; Crook, S.; Watkin, D. J.; Griffiths, R. C.; Nash, R. J.; Tsitsanou, K. E.; Zographos, S. E.; Oikonomakos, N. G.; Fleet, G. W. J. Tetrahedron Lett. 1995, 45, 8291. Brandstetter, T. W.; Wormald, M. R.; Dwek, R. A.; Butters, T. D.; Platt, F. M.; Tsitsanou, K. E.; Zographos, S. E.; Oikonomakos, N. G.; Fleet, G. W. J. Tetrahedron Asymm. 1996, 157. Estevez, J. C.; Smith, M. D.; Lane, A. L.; Crook, S.; Watkin, D. J.; Besra, G. S.; Brennan, P. J.; Nash, R. J.; Fleet, G. W. J. Tetrahedron Asymm. 1996, 387. Estevez, J. C. Burton, J. W.; Estevez, R. J.; Ardron, H.; Wormald, M. R.; Dwek, R. A.; Brown, D.; Fleet G. W. J. Tetrahedron Asymm. 1998, 2137. Smith, M. D.; Long, D. D.; Marquess, D. G.; Claridge, T. D. W.; Fleet, G. W. J. J. Chem. Soc. Chem. Commun. 1998, 2039. Gregar, T. Q.; Gervay-Hague, J. J. Am. Chem. Soc. Submitted. Walsh, S. T.; Cheng, H.; Bryson, J. W.; Roder, H.; DeGrado, W. F. Proc. Natl. Acad. Sci. USA 1999, 96, 5486–5491. Betz, S. F.; Liebman, P. A.; DeGrado, W. F. Biochemistry 1997, 36, 2450–2458. (a) Pauling, L.; Corey, R. B. Nature 1953, 171, 59–61. (b) Crick, F. H. C. Acta Crystallogr. 1953, 6, 685–697. Wilson, I. A.; Skehel, J. J.; Wiley, D. S. Nature 1981, 289, 366–373. Appella, D. H.; Christianson, L. A.; Karle, I. L.; Powell, D. G.; Gellman, S. H. J. Am. Chem. Soc. 1999, 121, 6206–6212.

14 Glycosyltransferases in Oligosaccharide Synthesis Xiangping Qian, Keiko Sujino, and Monica M. Palcic University of Alberta, Edmonton, Alberta, Canada R. Murray Ratcliffe SYNSORB Biotech Inc., Calgary, Alberta, Canada

I.

INTRODUCTION

Recent demonstrations that oligosaccharides play important roles in diverse biological events have resulted in renewed interest in the synthesis of oligosaccharides and their analogs. The availability of such molecules can facilitate studies on carbohydrate–protein recognition and help to elucidate molecular mechanisms of oligosaccharide-mediated biological processes [1,2] that could eventually lead to rationally designed carbohydrate-based therapeutics [3,4]. Despite many advances that have been made over the past decades, the chemical synthesis of oligosaccharides remains a challenge [5,6]. This is attributed to the inherent chemical difficulties presented by this class of molecules. Each monosaccharide carries at least three hydroxyl groups that must be protected and deprotected during synthesis. Also, glycosylation generates a new stereocenter at the anomeric carbon, and there are no general methods for the introduction of all types of glycosidic linkage in a manner that is both stereocontrolled and high yielding. The chemical synthesis of oligosaccharides is therefore very time-consuming and requires specialized expertise. The synthesis of oligosaccharide analogs containing modified sugars is an even more complex task than the preparation of natural structures. Almost twice as many steps are usually required for analog synthesis and the steps are more difficult, since most chemical protocols in the literature have been optimized for natural sugars. In nature, glycosyltransferase enzymes accomplish the ‘‘daunting’’ task of the construction of diverse and complex oligosaccharide [7–9]. These enzymes catalyze 535

536

Qian et al.

the transfer of a monosaccharide from a glycosyl donor (usually a sugar nucleotide) to a glycosyl acceptor in a regio- and stereospecific manner (Fig. 1). Pioneered by Barker [10,11], Whitesides [12] and Auge´ [13] and their colleagues, enzymatic synthesis by means of glycosyltransferases opened up a new venue for the preparation of oligosaccharides, avoiding many of the problems encountered in traditional chemical synthesis. Multistep protection and deprotection sequences are no longer required, since the glycosyl transfer catalyzed by glycosyltransferases is stereo- and regiospecific; thus the configuration of the newly formed anomeric center is absolute. Glycosyltransferases, unlike many catalysts used in organic synthesis, are environmentally benign and operate best in aqueous solution. Although glycosyltransferases exhibit high specificity for both the sugar nucleotide donor and acceptor, they have been shown to tolerate certain structural changes on both donor and acceptor substrates. An enzymatic approach to unnatural oligosaccharides greatly simplifies the synthetic scheme because the requirement for the chemistry is then reduced to the synthesis of more readily accessible smaller ‘‘primers’’ (mono- to trisaccharides), which can be elongated in a regio- and stereospecific manner using glycosyltransferases without additional protection and deprotection steps.

Figure 1 Reactions catalyzed by Leloir glycosyltransferases (B = uracil, guanine, cytosine). The acceptor glycosylation site is highlighted.

Glycosyltransferases

537

With advances in molecular biology and biotechnology, more than 30 glycosyltransferases from mammalian [9,14], bacterial [15–17], and even viral [18] sources have been cloned, and many are now readily available in large multiunit quantities. For synthetic applications, the major sugar nucleotide donor substrates for mammalian glycosyltransferases are commercially available, and cost limitations for large-scale synthesis are rapidly being overcome with nucleotide donor recycling systems [12,19] and pathway engineering [20]. Glycosyltransferase-catalyzed synthesis of both natural and unnatural oligosaccharides is an important component of glycochemistry, as evidenced by the numerous comprehensive reviews that have appeared [21–25]. II.

GLYCOSYLTRANSFERASES

Glycosyltransferases can be divided into the Leloir and non-Leloir types according to the type of glycosyl donors they use [26]. Non-Leloir glycosyltransferases typically use glycosyl phosphates as donors, while Leloir glycosyltransferases utilize sugar nucleotides as donors and transfer the monosaccharide with either retention (retaining enzymes) or inversion (inverting enzymes) of the configuration of the anomeric center (Fig. 1). Most of the glycosyltransferases responsible for the biosynthesis of mammalian glycoproteins and glycolipids are Leloir glycosyltransferases. They are typically type II membrane proteins with a short cytoplasmic Nterminal domain, a hydrophobic transmembrane domain, a luminally oriented stem, and a large C-terminal catalytic domain [27]. With a few exceptions [28,29], each glycosyltransferase produces a unique glycosidic linkage. According to the ‘‘one enzyme–one linkage’’ hypothesis [30], there are estimated to be hundreds of glycosyltransferases responsible for the production of the diverse and complex oligosaccharide structures found in nature. Glycosyltransferases are classified according to their nucleotide donor sugars, the type of glycosidic linkage formed (␣ or ␤), and the specific hydroxyl group in the acceptor to which the monosaccharide is transferred. Mammalian glycosyltransferases use only nine main sugar nucleotides as building blocks to construct complex oligosaccharides. They are activated sugar donors of uridine diphosphate (UDP-Gal, UDPGlcNAc, UDP-Glc, UDP-GlcNAc, UDP-Xyl, UDP-GlcA), guanosine diphosphate (GDP-Fuc, GDP-Man), and cytidine monophosphate (CMP-sialic acid or CMPNeu5Ac). All the sugar units have the D configuration except for fucose, which has an L configuration. This chapter reviews the literature on enzymatic synthesis utilizing Leloir glycosyltransferases, with focus on galactosyltransferases, fucosyltransferases, and sialyltransferases, most of which are commercially available or widely accessible. Also, they are well studied in terms of their donor and acceptor specificity and have been used for the synthesis of both natural oligosaccharides and their analogs. As well, two practical examples for the synthesis of trisaccharide analogs are given. III.

GALACTOSYLTRANSFERASES

A.

␤ 1,4-Galactosyltransferase

␤1,4-Galactosyltransferase (␤1,4-GalT, EC 2.4.1.22/38/90) has been commercially available for many years in unit quantities, where a unit is the amount of enzyme

538

Qian et al.

that converts one micromole of substrate to product per minute. It is the most widely studied glycosyltransferase with regard to substrate specificity and use in preparative synthesis and the first mammalian glycosyltransferase for which an X-ray crystal structure was determined [31]. Historically, ␤1,4-GalT was also the first enzyme used for the preparation of oligosaccharides on a large scale (>10 g) [12]. ␤1,4-GalT catalyzes the transfer of Gal from UDP-Gal to OH-4 of terminal ␤-linked GlcNAc to form N-acetyllactosamine (LacNAc) (Fig. 2). The acceptor specificity shifts to glucose in the presence of the protein modifier ␣-lactalbumin, giving lactose as the major product. Since Berliner’s pioneering work demonstrating that ␤1,4-GalT could transfer glucose, 4-deoxy-Gal and arabinose from their corresponding UDP donors [32], donors with modifications at every OH group have been employed for the synthesis of numerous LacNAc analogs (Fig. 3). While the enzyme transfers 2-deoxy-Gal at a rate comparable to that of UDP-Gal [33], 3-deoxy [34], 6-deoxy [35,36], and 6deoxy-6-fluoro Gal [35,36] are transferred at reduced rates. The enzyme also utilizes UDP-GalNAc, UDP-GalNH2, and UDP-GlcNH2 donors [37,38] (Fig. 3). Replacement of the oxygen of Gal with sulfur in UDP-5⬘-thio-Gal is tolerated by ␤1,4-GalT [39], including transfer in the presence of ␣-lactalbumin [40]. The acceptor specificity of ␤1,4-GalT is equally relaxed, since numerous modifications in acceptors including the ring oxygen are possible as long as the 4-OH remains available for glycosylation (Fig. 3). The 4-OH is a ‘‘key polar group’’ [41] where replacement or modification yields analogs that no longer bind to the enzyme. The 2-NHAc group can be replaced with N-propanoyl [42,43], N-butanoyl [42], azido [38,44], allylcarbamate [45], and other amide derivatives, including bulky heterocycles, charged groups, and glycuronamides [46]. Acceptors with the 3-OH group deoxygenated [47,48], alkylated with a methyl or allyl group, or oxidized to the ketone are active as acceptors with reduced reaction rates [48]. The 6-OH group of GlcNAc can be methylated [43,48], deoxygenated [47], fucosylated [43], or replaced with F or SH [47]. GlcNAc bearing ␣-linked sialic acid at 6-OH is not a substrate;

Figure 2

The reaction catalyzed by ␤1,4-galactosyltransferase.

Glycosyltransferases

Figure 3

539

Unnatural donors and acceptors for ␤1,4-GalT.

however, if the carboxylic acid of NeuAc is derivatized with a methyl ester, the resulting compound is an acceptor with 4% the rate of transfer compared to the parent acceptor [43]. ␤1,4-GalT will utilize acceptors with the ring oxygen modified, including 5⬘thio-Glc and 1-deoxy-nojirimycin [48]. Glucal with a flattened ring [48] and (⫾)conduritol B are substrates; the latter mixture can be resolved because only a single galactosylated product (⫺)-conduritol B is produced [49]. In an exception to the

540

Qian et al.

‘‘one enzyme–one linkage’’ hypothesis, the enzyme transfers galactose to the ␤anomeric position of 3-acetamido-3-deoxy-D-glucose acceptors giving a ␤1→1 trehalose type of linkage [29]. This atypical regiochemistry of galactosylation was also reported for N-acetyl-gentosamine, N-acetyl-5⬘-thiogentosamine, and xylose acceptors [50,51]. A variety of immobilized acceptors have been employed in solid phase or soluble polymer synthesis [52–58], and even UDP-GlcNAc is an acceptor substrate, yielding the nucleotide disaccharide UDP-N-acetyllactosamine [59]. Reactions at scales of 500 g or more with ␤1,4-GalT have been reported with in situ donor recycling [60]. Also ␤1,4-GalT enzymes have been found in bacterial sources, including Helicobacter pylori and Neisseria meningitidis [61–63] and utilized in 100 g reactions for N-acetyllactosamine production through bacterial coupling [64]. B.

␣1,3-Galactosyltransferase

␣1,3-Galactosyltransferase (␣1,3-GalT, EC 2.4.1.151) has recently attracted much attention because it catalyzes the transfer of Gal from UDP-Gal to 3-OH of the Gal residue in Gal␤1 → 4GlcNAc-R to form Gal␣1 → 3Gal␤1 → 4GlcNAc epitopes on glycoproteins and glycolipids (Fig. 4) [65–67]. This is the major xenoactive antigen that is responsible for hyperacute rejection in xenotransplantation [68,69]. ␣1,3-Galactosyltransferases from porcine and bovine tissues and recombinant bovine ␣1,3-galactosyltransferase have all been used for preparative synthesis, the latter for gram-scale reactions [66,70]. Recombinant porcine ␣1,3-GalT is now commercially available in unit quantities, and a fusion protein comprising UDP-Gal epimerase and ␣1,3-GalT has been engineered for large-scale synthesis [71]. ␣1,3-Galactosyltransferase transfers 3-deoxy-, 4-deoxy-, and 6-deoxy-Gal from the corresponding donor at very low rates (<2%), while UDP-2-deoxy-Gal is a better

Figure 4

The reaction catalyzed by ␣1,3-galactosyltransferase.

Glycosyltransferases

541

substrate than UDP-Gal [72] (Fig. 5). However, unlike the case of ␤1,4-GalT, UDPGlc, GalNAc, GlcNAc, and glucuronic acid are not transferred by ␣1,3-GalT [73]. Acceptors other than Gal␤1 → 4GlcNAc are utilized by ␣1,3-GalT, including Gal␤1 → 3GlcNAc and Gal␤1 → 4Glc [67,74], and acceptors immobilized on polymers [71]. Modifications on 4-OH of Gal in acceptors give inactive compounds, suggesting that this is a key polar group essential for binding to ␣1,3-GalT [67]. The N-acetyl group in acceptors can be replaced with azido, succinimido, and others with a large number of acyl groups [67,75] (Fig. 5). However, replacement of the 2-NHAc with an amino group abolishes activity [73]. As shown in Figure 5, analogs with modifications on 6-OH of GlcNAc, or 2-OH and 6-OH of the terminal Gal residue are substrates, while only deoxygenation of the 3-OH of GlcNAc residue is tolerated [67]. The enzyme can catalyze the transfer of a Gal residue to a very hindered tertiary alcohol acceptor where the carbon-bonded hydrogen at the glycosylation site is replaced with a methyl group, demonstrating that glycosyltransferases can overcome inherent steric limitations to produce oligosaccharide analogs that cannot be easily prepared by chemical methods [76]. C.

␤ 1,3- and ␣1,4-Galactosyltransferases

Though not widely available yet, the ␤1,3-galactosyltransferase that synthesizes Gal␤1 → 3GlcNAc (type I) from GlcNAc acceptors has been cloned and expressed [77,78]. Mapping has indicated that N-acyl analogs can be utilized as acceptors by

Figure 5

Unnatural donors and acceptors for ␣1,3-galactosyltransferase.

542

Qian et al.

the enzyme [79]. ␣1,4-Galactosyltransferase from Neisseria meningitidis transfers Gal from UDP-Gal donors solely to lactose acceptors yielding globotriose [62,72]. The enzyme will transfer arabinose, 2-deoxy-, 4-deoxy-, and 6-deoxy-Gal from modified donors from 2 to 28% of the rate of UDP-Gal; however, 3-deoxy-Gal is not transferred [72]. Large-scale production of UDP-Gal and globotriose (188 g/L) has been achieved by metabolic engineering of bacteria [20]. D.

Human Blood Group A and B Glycosyltransferases

Human blood group A and B glycosyltransferases (GTA and GTB: EC 2.4.1.40 and 2.4.1.37) are responsible for the biosynthesis of A and B blood group antigens, which are important in cell development, cell differentiation and oncogenesis [80,81]. Both GTA and GTB are retaining enzymes. GTA catalyzes the transfer of GalNAc from UDP-GalNAc to the (O)H antigen (Fuc␣1 → 2Gal␤-OR) to give the A antigen GalNAc␣1 → 3[Fuc␣1 → 2]Gal␤-OR. GTB uses the same acceptor but catalyzes the transfer of Gal from UDP-Gal to form the B antigen, Gal␣1 → 3[Fuc␣1 → 2]Gal␤-OR (Fig. 6). Chemical mapping studies reveal that OH-4 of the Gal residue is a key polar group for both GTA and GTB, whereas OH-3 of the Gal unit to which the glycosyl residue transfers is not essential for recognition by either enzyme [82,83]. As shown in Figure 7, both enzymes tolerate deoxygenation, substitution, and derivatization of the 6-OH group of the Gal unit [82]. Deoxygenation of any of the hydroxyl groups on the Fuc residue is tolerated by GTA. Methylation of O3⬘ and O4⬘ is tolerated by both GTA and GTB. The arabino derivative, where the CH3 group of the Fuc residue

Figure 6

Reactions catalyzed by blood group A and B glycosyltransferases.

Glycosyltransferases

Figure 7

543

Unnatural acceptors for GTA (left) and GTB (right).

is replaced by H, was found to be an acceptor for GTA [84]. Both GTA and GTB, like ␣1,3-GalT and human milk ␣1,3/4-FucT, can also act on highly hindered tertiary alcohol acceptors and will even accommodate a large propyl group at the glycosylation site [76]. While crossovers in donor specificity have been reported (i.e., GTA can utilize UDP-Gal and vice versa) [85,86], UDP-Glc is a donor only for GTB and UDP-GlcNAc only for GTA [87]. Donor monosaccharide mapping of GTB shows that modifications to Gal are tolerated at all positions. The rates of transfer relative to UDP-Gal are 2-deoxy-Gal (175%), 3-deoxy-Gal (0.1%), 4-deoxy-Gal (0.2%), 6deoxy-Gal (18%), and arabinose (0.1%) [72].

IV.

FUCOSYLTRANSFERASES

Many antigenic oligosaccharides on the cell surface such as blood group antigens are fucosylated. Along with sialylation, fucosylation by fucosyltransferases is often the last in vivo modification of oligosaccharides. The fucosylated oligosaccharides on cell surfaces are involved in numerous intercellular recognition events [28,88,89]. ␣1,2-Fucosyltransferases transfer a fucosyl residue from GDP-Fuc to the 2-OH group of ␤-D-galactosides to form Fuc␣1 → 2Gal structures, while ␣1,3/4-fucosyltransferases catalyze the transfer of a fucose residue from GDP-fucose to the 3-OH or 4OH group of ␤-D-galactosides in a variety of acceptors to form the blood-grouprelated antigenic determinants Lea and Lex [28,90–92]. ␣1,3/4-Fucosyltransferases are a multigene family of enzymes with differences in substrate specificities toward Gal␤1 → 4GlcNAc, Gal␤1 → 3GlcNAc, and NeuAc␣2 → 3Gal␤1 → 4GlcNAc acceptors; there are differences, as well, with respect to inhibitor sensitivity, pH optima, and tissue distribution. The most common source for the isolation of preparatively useful ␣1,3/4-fucosyltransferases is human milk [93]. There are two different human milk fucosyltransferases, ␣1,3/4-fucosyltransferase and ␣1,3-fucosyltransferase. Five different human ␣1,3/4-fucosyltransferases have been cloned, FucT III to FucT VII. The in vivo specificities of all five of these human fucosyltransferases expressed in baby hamster kidney (BHK) cells have been elucidated [94]. FucTs IV– VI are now commercially available in 100 mU quantities, sufficient for preparative synthesis.

544

A.

Qian et al.

Human Milk ␣1,3/4-Fucosyltransferase

Human milk ␣1,3/4-FucT (EC 2.4.1.65) is another exception to the ‘‘one enzyme– linkage’’ hypothesis. This enzyme catalyzes the transfer of fucose to both type I and type II acceptors to produce Lea and Lex structures, respectively (Fig. 8). The enzyme was found to tolerate several structural changes on the donor substrate. The enzyme utilizes GDP-3-deoxy-fucose, GDP-arabinose, and GDP-L-galactose donors in transfers to Gal␤1 → 3GlcNAc␤-O-(CH2)8COOMe as an acceptor (Fig. 9) [95]. In 1998 a group used GDP-3-deoxy-Fuc and GDP-L-Gal in the preparative synthesis of Lex analogs, with Gal␤1 → 4GlcNAc␤-O(CH2)8COOMe as an acceptor [96]. The enzyme was also shown to tolerate the addition of a propyl group at 6-OH of L-Gal in GDP-L-Gal. Even larger substitutents can be introduced at C6 of the Fuc residue in GDP-Fuc, including tethered blood group B trisaccharide [97]. Virtually any group can be attached to the C6 position of Fuc, as was demonstrated by the transfer of modified fucose residues containing biotin and blood group A trisaccharide (Fig. 9) [98]. Human milk ␣1,3/4-FucT uses both Gal␤1 → 3GlcNAc (type I) and Gal␤1 → 4GlcNAc (type II) acceptors. As shown in Figures 8 and 9, modifications are tolerated at every hydroxyl group in the sugar rings except for two key polar groups, 6-OH of the Gal and 3-OH or 4-OH of the GlcNAc residue to which fucose is transferred [99]. The 2-NHAc group of GlcNAc residue in NeuAc␣2 → 3Gal␤1 → 3GlcNAc␤-O(CH2)8COOMe or NeuAc␣2 → 3Gal␤1 → 4GlcNAc␤-O(CH2)8COOMe can be replaced with azido, amino, or propionamido groups [100]. Thiolinked N-acetyllactosamine, in which the interglycosidic oxygen is replaced by sulfur, is also a good acceptor for the enzyme [101]. Ether- and imino-linked octyl N-acetyl5a⬘-carba-␤-lactosamides were also found to be acceptors for human milk ␣1,3/4FucT [102]. As with ␣1,3-GalT and blood group B galactosyltransferase, ␣1,3/4FucT can tolerate the introduction of a large methyl group directly at the site of fucosylation. It transfers fucose to the hindered tertiary alcohol acceptor 3-C-methylN-acetyllactosamine to produce a carbon-branched Lex analog [103]. B.

Fucosyltransferase V

FucT V is similar to human milk ␣1,3/4-FucT because it will utilize both type I and type II acceptors (Fig. 8). This important industrial enzyme has been used for the large production of sialyl Lewisx. FucT V also has similar acceptor specificity as human milk ␣1,3/4-Fuc, requiring 6-OH of Gal and 3-OH or 4-OH of GlcNAc for substrate binding [99]. Gal␤1 → 4Glucal, Gal␤1 → 4(-5-S)-Glc, 3-sulfo-Gal␤1 → 4GlcNAc, and Gal␤1 → 4-(6-sulfo)-GlcNAc are acceptors for FucT V [44,104]. C.

Fucosyltransferases III and VI

FucT III transfers fucose onto 4-OH of GlcNAc residues of Gal␤1 → 3GlcNAc or NeuAc␣2 → 3Gal␤1 → 3GlcNAc to form Lea or sialyl Lea, respectively (Fig. 10). FucT VI uses Gal␤1 → 4GlcNAc or NeuAc␣2 → 3Gal␤1 → 4GlcNAc to form Lex or sialyl Lex products (Fig. 10). Despite their close sequence homologies, they exhibit differences in their recognition of unnatural donors [105]. FucT III transfers L-Gal, arabinose, L-Glc, 2-amino-2-deoxy-fucose, and 2-fluoro-2-deoxy-fucose (Fig. 11). FucT VI does not tolerate modifications on the 2-OH group of the fucose but will

Glycosyltransferases

Figure 8

Reactions catalyzed by human milk ␣1,3/4-FucT and FucT V. 545

546

Figure 9

Qian et al.

Unnatural donors and acceptors for human milk ␣1,3/4-FucT.

Glycosyltransferases

Figure 10

Reactions catalyzed by FucT III and FucT VI.

547

548

Unnatural donors and acceptors for FucT III.

Qian et al.

Figure 11

Glycosyltransferases

549

transfer L-Gal and arabinose from their corresponding donors [105] (Fig. 12). Both enzymes accept replacements and modifications of the N-acetyl group of the GlcNAc unit [106–108]. FucT VI has been shown to tolerate replacement of the GlcNAc unit with glucal and cyclohexane diol [109]. FucT III and FucT VI have further been used to construct libraries of sialyl Lea or sialyl Lex derivatives using both unnatural donors and unnatural acceptors [110–112]. FucT III and FucT VI have been probed with unnatural donors with a modified purine base. They can tolerate exchange of the guanine by other purine bases such as adenine [113]. These unnatural donors proved to be preparatively efficient in the enzymatic synthesis of Lea or Lex, suggesting that natural sugar nucleotide donors can be replaced with inexpensive ones to lower the cost of enzymatic synthesis. V.

SIALYLTRANSFERASES

Sialyltransferases (SialTs) catalyze the transfer of N-acetyl-neuraminic acid from CMP-Neu5Ac to acceptors at or near the nonreducing terminus of oligosaccharide chains of glycoproteins or glycolipids [114–116]. Cell surface sialic acid residues are important as antigenic determinants, as ligands for cellular adhesion, and as receptors for the binding of viruses and bacteria [117]. Sialic acids are usually found in terminal positions linked through an ␣-glycosidic linkage. The stereoselective synthesis of ␣-sialosides remains a challenge, since the glycosides have a kinetically and thermodynamically unfavorable equatorial orientation and there is no participating group at C3 to direct the stereochemistry in their synthesis [118]. Enzymatic sialylation therefore offers an attractive alternative to prepare ␣-sialosides in an efficient and stereocontrolled manner. To date, 16 different mammalian sialyltransferases, several bacterial sialyltransferases, and a viral sialyltransferase have been identified. They are ␣2,3-, ␣2,6-, and ␣2,8-sialyltransferases. A systematic classification for sialyltransferases has been proposed on the basis of the regiochemistry of the sialylated reaction product and the acceptor substrate preference [116]. Two different recombinant ␣2,3-sialyltransferases and an ␣2,6-sialyltransferase from rat liver are now commercially available in sufficient quantities (100 mU) for use in preparative synthesis. A.

␣ 2,3-Sialyltransferase and ␣ 2,6-Sialyltransferase

␣2,3-SialT (␣2,3-SialT, EC 2.4.99.6, ST3Gal III) from rat liver transfers a sialic acid unit from CMP-sialic acid to 3-OH of the terminal Gal residue in Gal␤1 → 4GlcNAc or Gal␤1 → 3GlcNAc sequences (Fig. 13) while ␣2,6-SialT (␣2,6-SialT, EC 2.4.99.1) transfers the sialic acid to 6-OH of the terminal Gal residue in Gal␤1 → 4GlcNAc (Fig. 14) [115,116]. The finding that CMP–sialic acid synthetase can accept many sialic acid analogs has facilitated the production of donor analogs for sialyltransferase specificity studies [119–121]. Both ␣2,3-SialT and ␣2,6-SialT tolerate substitutions at C9 of CMP–sialic acid (Figs. 15 and 16) [120,122,123]. The 9-OH group can be replaced with fluoro, azido, amino, acetamido, hexanoylamido, benzamido, and fluorescent labels. 9-O-Acetylsialic acid is also transferred. Donors with the 5-NHAc of sialic acid replaced with OH, NHC(O)CH2OH, or NHCbz (Cbz = benzyloxycarbonyl) are also utilized by ␣2,3-SialT [124]. CMP-4-deoxysialic acid was found to be a donor substrate for ␣2,6-SialT [125].

550 Qian et al.

Figure 12

Unnatural donors and acceptors for FucT VI.

Glycosyltransferases

Figure 13

551

Reactions catalyzed by ␣2,3-SialT.

Chemical mapping studies of acceptor binding have indicated that 6-OH of Gal and 2-NHAc are required for recognition by rat liver ␣2,6-SialT, while rat liver ␣2,3SialT requires an intact 3,4,6-triol system on the Gal residue [126]. Analogs of Gal␤1 → 4GlcNAc␤1 → 2Man␣-OOct, where the 3-OH or 4-OH group of the Gal residue is deoxygenated or substituted with a fluoro atom, were found to be active as substrates for ␣2,6-SialT (Fig. 16) [127]. The 4⬙-O-methyl derivative of this trisaccharide is an acceptor for ␣2,3-SialT [127]. A variety of substitutions on the N-acetyl group of either type I or type II acceptors are tolerated by ␣2,3-SialT [128,129]. ␣2,3-SialT also transfers sialic acid to lactal, lactose, and 2-O-pivaloyl lactose acceptors (Fig. 15) [44]. ␣2,3-SialTs from N. meningitidis [17,130] and myxoma virus [18] and an

552

Qian et al.

Figure 14

Reaction catalyzed by ␣2,6-SialT.

Figure 15

Unnatural donors and acceptors for ␣2,3-SialT.

␣2,6-SialT from Photobacterium damsela [16] have acceptor specificities somewhat different from those of their mammalian counterparts, allowing for broader synthetic applications. Large-scale production of CMP-NeuAc and sialylated oligosaccharides has been achieved by bacterial coupling [131]. VI.

PRACTICAL BIOCATALYSIS WITH GLYCOSYLTRANSFERASES

There are no fundamental differences between chemical catalysts and biological enzyme catalysts. Enzymes accelerate the rates of chemical reaction without being changed in the overall transformation; they reduce the activation energies of chemical

Glycosyltransferases

Continued

Figure 15

553

554

Unnatural donors and acceptors for ␣2,6-SialT.

Qian et al.

Figure 16

Glycosyltransferases

555

reactions. The only practical differences between chemical and enzyme catalysts are that enzymes are generally sensitive to temperature, pH, and shear. The most critical parameter for using an enzyme in synthesis is defining the quantity or activity of enzyme that will be added to ensure that the reaction goes to completion. This amount should first be determined in a defined standard kinetic assay and then with the substrate that will be employed in the reaction. The amount of enzyme activity is defined by units (U), where one U of enzyme activity is the amount producing 1 ␮mol of product per minute. Substrate and product inhibition are frequently observed in enzyme catalysis; therefore it is desirable to carry out a full kinetic analysis by varying substrate concentrations and determining the Vmax (maximal rate of reaction at saturating substrate) and KM for the substrate. This analysis will show whether substrate inhibition occurs at high concentrations that would reduce the reaction rate. As well, by estimating the rate of reaction at a given substrate concentration, the time required for completion can be estimated. A standard assay can also be used to check the viability of the enzyme catalyst throughout the reaction by removing a small aliquot from the reaction mixture and assaying for remaining activity in a standard assay. Water is the usual reaction medium for enzymes; however, organic solvents or cyclodextrins can be added to increase the solubility of poorly soluble compounds. While multienzyme donor recycling systems can be used for in situ donor production [12,19], the following examples utilize stoichiometric donors. A.

Glycosyltransferase Assays

After the glycosyltransferase of interest has been purchased or isolated, it is desirable to verify the activity of the enzyme with a standard substrate and the compound of interest under the conditions of the preparative synthesis. Methods for assaying glycosyltransferase activity are summarized in Table 1, and a suitable method should be selected based on the characteristics of the substrate and the required sensitivity of the assay. Kinetic studies to obtain KM and Vmax values for the substrate and optimization of reaction conditions, including buffer and pH, should also be carried out. If the compound of interest has a very low reaction rate (<1% of the activity of a known standard substrate), further evaluations with concentrated enzyme are recommended. There are examples of traces of reactive substrates remaining at these low levels in inactive compounds. Since 0.1% conversion of a reactant to product can easily be detected in radiochemical glycosyltransferase assays, confirmation that the low rate of conversion is not due to the presence of contaminants is essential. This level of contamination in modified synthetic substrates is sometimes difficult to detect even by high-field NMR spectroscopy. Therefore before the actual preparative synthesis is performed, 10% conversion of substrate to product should be verified by assay to ensure that the low rate of reaction is not due to traces of substrate in a largely inactive compound. B.

Enzyme and Substrate Concentrations

The activity obtained in an assay allows an estimation of the amount of enzyme that will be required to completely convert substrate to product in a reasonable reaction time (e.g., 6–12 h). Reaction rates will decrease as the substrate is consumed and

556

Table 1

Assays for Glycosyltransferases

Acceptor

Donor

1 Glycosides

Radioactive sugar nucleotides

2 Glycosides 3 Glycosides

Additives

Method

Ref.a

C18 Sep-Pak

1

⬙

HPLC

2

⬙

Ion exchange column chromatography

3

4 Glycosides

⬙

Thin-layer chromatography

4

5 Glycoprotein

⬙

Affinity chromatography

5

6 Glycoprotein

⬙

Particle exclusion assay

6

7 Long alkyl glycosides

⬙

Organic solvent extraction

7

8 Glycosides

⬙

GC and mass spectometry

8

9 Glycoprotein

⬙

10 Glycosides with high molecular weights

Radioactive or labeled sugar nucleotides

11 Glycosides-conjugates

Sugar nucleotides

Filtration with membranes

9

Size-exclusion column chromatography

10

Antibody (or antiserum, lectin)

ELISA

11

⬙

Coupling enzymes (NADH)

Coupled enzyme assays

12

13 Glycosides

⬙

Fluorochrome

14 Labeled glycosides

⬙

12 Glycosides

15 Labeled glycosides

⬙

16 Biotin-labeled glycan

⬙

HRP-avidin

Microtiter-based fluorescence assay

13

HPLC

14

Capillary electrophoresis

15

Affinity chromatography

16

⬙

Mass spectrometry, HPLC

17

⬙

HPLC, size exclusion column chromatography

18

Qian et al.

17 Glycosides 18 Radioactive acceptor

Table references as follows. 1. (a) Palcic, M. M., Heerze, L. D., Pierce, M., Hindsgaul, O., Glycoconjugate J. 1988, 5, 49. (b) Kaur, K. J., Hindsgaul, O., Glycoconjugate J. 1991, 8, 90. (c) Srivastava, G., Hindsgaul, O., Carbohydr. Res. 1992, 224, 83. (d) Wilkes, C., Pierce, M., Anal. Biochem. 1995, 224, 173. (e) Sato, T., Furukawa, K., Bakker, H., Van Den Eijinden, D. H., Van Die, I., Proc. Natl. Acad. Sci. USA 1998, 95, 472. 2. (a) Salvucci, M. E., Crafts-Brandner, S. J., Anal. Biochem. 1991, 194, 365. (b) Ng, K., Handman, E., Bacic, A., Glycobiology 1994, 4, 845. (c) Misawa, H., Tsumuraya, Y., Kaneko, Y., Hashimoto, Y., Plant Physiol. 1996, 110, 665. (d) Ethell, B. T., Anderson, G. D., Beaumont, K., Rance, D. J., Burchell, B., Anal. Biochem. 1998, 255, 142. (e) Yongxin, Z., Biggins, J. B., Thorson, J. S., J. Am. Chem. Soc. 1998, 120, 12986. 3. (a) Sadler, J. E., Beyer, T. A., Oppenheimer, C. L., Paulson, J. C., Prieels, J.-P., Rearick, J. I., Hill, R. L., Methods Enzymol. 1982, 83, 458. (b) Sichel, F., Malas, J. P., Gauduchon, P., Aubert, M., Bar, E., Le Talaer, J. Y., Clin. Chim. Acta 1990, 188, 49. (c) Parker, G. J., Handley, C. J., Robinson, H. C., Anal. Biochem. 1995, 226, 154. (d) Yamamoto, F., McNeill, P. D., J. Biol. Chem. 1996, 271, 10515. 4. (a) Gu, X. B., Gu, T. J., Yu, R. K., Anal. Biochem. 1990, 185, 151. (b) Lynch, D. V., Criss, A. K., Lehoczky, J. L., Bui, V. T., Arch. Biochem. Biophys. 1997, 340, 311. 5. (a) Dakour, J., Zopf, D., Lundblad, A., Anal. Biochem. 1992, 204, 210. (b) Srivatsan, J., Smith, D. F., Cummings, R. D., J. Parasitol. 1994, 80, 884. 6. Spicer, A. P., Olson, J. S., McDonald, J. A., J. Biol. Chem. 1997, 272, 8957. 7. (a) Shukla, G. S., Radin, N. S., Arch. Biochem. Biophys. 1990, 283, 372. (b) Abe, N., Abe, E., Yuasa, A., J. Vet. Med. Sci. 1992, 54, 125. (c) Matern, H., Heinemann, H., Matern, S., Anal. Biochem. 1994, 219, 182. 8. White, A. R., Xin, Y., Pezeshk, V., Biochem. J. 1993, 294, 231. 9. (a) Nakamura, M., Tsunoda, A., Saito, M., Anal. Biochem. 1991, 198, 154. (b) Weston, A., Nassau, P. M., Henly, C., Marriott, M. S., Eur. J. Biochem. 1993, 215, 845. (c) Hooper, L. V., Baenziger, J. U., Anal. Biochem. 1993, 212, 128. (d) ref 8. (e) Maga, T., Mizuno, M., Tanaka, S., Yoshinaga, F., Mikami, Y., Kihara, Y., Kiso, T., Tomoda, J., Okada, H., Tsuji, Takano, T., Digestion 1997, 58, 389. 10. (a) Ref. 3(a). (b) Gross, H. J., Sticher, Y., Brossmer, R., Anal. Biochem. 1990, 186, 127. 11. (a) Crawley, S. C., Hindsgaul, O., Alton, G., Pierce, M., Palcic, M. M., Anal. Biochem. 1990, 185, 112. (b) Taki, T., Nishiwaki, S., Ishii, K., Handa, S., J. Biochem. 1990, 107, 493. (c) Palcic, M. M., Ratcliffe, R. M., Lamontagne, L. R., Good, A. H., Alton, G., Hindsgaul, O., Carbohydr. Res. 1990, 196, 133–140. (d) Stults, C. L., Macher, B. A., Arch. Biochem. Biophys., 1990, 280, 20. (e) Tachikawa, T., Yazawa, S., Asao, T., Shin, S., Yanaihara, N., Clin. Chem. 1991, 37, 2081. (f ) Zatta, P. F., Nyame, K., Cormier, M. J., Mattox, S. A., Prieto, P. A., Smith, D. F., Cummings, R. D., Anal. Biochem. 1991, 194, 185. (g) Mengeling, B. J., Smith, P. L., Stults, N. L., Smith, D. F., Baenzinger, J. U., Anal. Biochem. 1991, 199, 286. (h) Keshvara, L. M., Newtron, E. M., Good, A. H., Hindsgaul, O., Palcic, M. M., Glycoconjugate J. 1992, 9, 16. (i) Fukushima, K., Okada, T., Ochiai, K., Infect. Immun. 1993, 61, 323. ( j) Keshvara, L. M., Gosselin, S., Palcic, M. M., Glycobiology 1993, 3, 416. (k) Yazawa, S., Nakajima, T., Kameyama, N., Saga, K., Tachikawa, T., Carbohydr. Res. 1993, 239, 329. (l) Mobassaleh, M., Mishra, K., Keusch, G. T., Anal. Biochem. 1993, 214, 295. (m) Chia, J. S., Lin, R. H., Lin, S. W., Chen, J. Y., Yang, C. S., Infect. Immun. 1993, 61, 4689. (n) Meier, W., Leon, D. R., Miatkowski, K., Lobb, R., Goelz, S. E., Biochem. J. 1993, 294, 25. (o) Taki, T., Nishiwaki, S., Handa, N., Hattori, N., Handa, S., Anal. Biochem. 1994, 219, 104. (p) Dohi, T., Hashiguchi, M., Yamamoto, S., Morita, H., Oshima, M., Cancer 1994, 73, 1552. (q) Yan, L., Smith, D. F., Cummings, R. D., Anal. Biochem. 1994, 223, 111. (r) Keusch, J., Lydyard, P. M., Isenberg, D. A., Delves, P. J., Glycobiology 1995, 5, 365. (s) Tomita, Y., Zhu, X., Ochiai, K., Namiki, Y., Okada, T., Ikemi, T., Fukushima, K., FEMS Microbiol. Lett. 1996, 145, 427. (t) DeBose-Boyd, R. A., Nyame, A. K., Smith, D. F., Cummings, R. D., Arch. Biochem. Biophys. 1996, 335, 109. Continues

Glycosyltransferases

a

557

558 [Table references continued] (u) Suzuki, A., Shibata, N., Suzuki, M., Saitoh, F., Taketa, Y., Oshie, A., Oyamada, H., Kobayashi, H., Suzuki, S., Okawa, Y., Eur. J. Biochem. 1996, 240, 37. (v) Borsig, L., Kleene, R., Dinter, A., Berger, E. G., Eur. J. Cell Biol. 1996, 70, 42. (w) Schwientek, T., Narimatsu, H., Ernst, J. F., J. Biol. Chem. 1996, 271, 3398. (x) Ogiso, M., Komoto, M., Hoshi, M., Biochim. Biophys. Acta 1996, 1315, 29. (y) Rabina, J., Smithers, N., Britten, C. J., Renkonen, R., Anal. Biochem. 1997, 246, 71. (z) Yoshiki, J., Kubushiro, K., Tsukazaki, K., Udagawa, Y., Nozawa, S., Iwamori, M., Jp. J. Canc. Res. 1997, 88, 669. (aa) Alaluusua, S., Gronroos, L., Zhu, X., Saarela, M., Matto, J., Asikainen, S., Fukushima, K., Arch. Oral Biol. 1997, 42, 417. (bb) Oubihi, M., Kitajima, K., Kobayashi, K., Adachi, T., Aoki, N., Matsuda, T., Anal. Biochem. 1998, 257, 169. (cc) Nogrady, N., Pocsi, I., Paffard, S. M., Katona, E., Miles, R. J., Balazs, A., Sandor, E., Fachet, J., Price, R. G., Szentirmai, A., J. Biotechnol. 1998, 60, 15. (dd) Borsig, L., Katopodis, A. G., Bowen, B. R., Berger, E. G., Glycobiology 1998, 8, 259. (ee) Keusch, J., Lydyard, P. M., Berger, E. G., Delves, P. J., Glycoconjugate J. 1998, 15, 1093. 12. (a) Gosselin, S., Alhussaini, M., Streiff, M. B., Takabayashi, K., Palcic, M. M., Anal. Biochem. 1994, 220, 92. (b) Murray, B. W., Wittmann, V., Burkart, M. D., Hung, S.-C., Wong, C.-H., Biochemistry 1997, 36, 823. (c) Trubetskoy, O. V., Shaw, P. M., Drug Metab. Dispos. 1999, 27, 555. 13. (a) Shedletzky, E., Unger, C., Delmer, D. P., Anal. Biochem. 1997, 249, 88. 14. (a) Higuchi, T., Tamura, S., Takagaki, K., Nakamura, T., Morikawa, A., Tanaka, K., Tanaka, A., Saito, Y., Endo, M., J. Biochem. Biophys. Methods 1994, 29 135. (b) Uozumi, N., Teshima, T., Yamamoto, T., Nishikawa, A., Gao, Y. E., Miyoshi, E., Gao, C. X., Noda, K., Islam, K. N., Ihara, Y., Fujii, S., Shiba, T., Taniguchi, N., J. Biol. Chem. 1996, 120, 385. (c) Palmerini, C. A., Datti, A., Vanderelst, I. E., Minuti, L., Orlacchio, A., Glycoconjugate J. 1996, 13, 631. (d) Limberg, G., Slim, G. C., Compston, C. A., Stangier, P., Palcic, M. M., Furneaux, R. H., Liebig. Ann. Chem. 1997, 1773. (e) Luukkanen, L., Elovaara, E., Lautala, P., Taskinen, J., Vainio, H., Pharmacol. Toxicol. 1997, 80, 152. (f ) Morita, N., Hase, S., Ikenaka, K., Mikoshiba, K., Ikenaka, T., J. Biochem. 1998, 103, 332. (g) Higai, K., Masuda, D., Matsuzawa, Y., Satoh, T., Matsumoto, K., Biol. Pharm. Bull. 1999, 22, 333. 15. (a) Lee, K. B., Desai, U. R., Palcic, M. M., Hindsgaul, O., Linhardt, R. J., Anal. Biochem. 1992, 205, 108. (b) Bogan, D. P., Killard, A. J., O’Kennedy, R., J. Capillary Electrophor. 1995, 2, 241. (c) Gilbert, M., Watson, D. C., Cunningham, A. M., Jennings, M. P., Young, N. M., Wakarchuk, W. W., J. Biol. Chem. 1996, 271, 28271. (d) Kanie, Y., Kirsch, A., Kanie, O., Wong, C. H., Anal. Biochem. 1998, 263, 240. (e) Snow, D. M., Shaper, J. H., Shaper, N. L., Hart, G. W., Anal. Biochem. 1999, 271, 36. (f ) Sujino, K., Uchiyama, T., Hindsgaul, O., Seto, N. O. L., Wakarchuk, W., Palcic, M. M., J. Am. Chem. Soc. 2000, 122, 1261. 16. Liu, A. H., Liu, F., Li, Z., Gu, J. X., Chen, H. L., Cell Biol. Int. 1998, 22, 545. 17. (a) Kodama, H., Kajihara, Y., Endo, T., Hashimoto, H., Tetrahedron Lett. 1993, 34, 6419. (b) Willenbrock, F. W., Neville, D. C., Jacob, G. S., Scudder, P., Glycobiology 1991, 1, 223. (c) Wu, J., Takayama, S., Wong, C.-H., Siuzdak, G., Chem. Biol. 1997, 4, 653. 18. Verostek, M. F., Trimble, R. B., Glycobiology 1995, 5, 671.

Qian et al.

Glycosyltransferases

559

its concentration drops. Therefore reaction times to achieve quantitative conversions are always three to ten times longer than those calculated, and even longer for substrates with high KM values. Preparative synthesis should start with substrates at kinetically saturated concentrations, about five times their KM values. In the case of substrates with a severe solubility problem, the addition of organic solvents to the buffer may aid dissolution [46]. Plastic containers (polypropylene falcon tubes, microcentrifuge tubes, or scintillation vials) are preferred to glass for glycosyltransferase reactions, since these enzymes are inactivated on glass surfaces. The addition of bovine serum albumin (BSA: ⬃1 mg/mL) or glycerol (ⱕ50%) to glycosyltransferase reaction mixtures can stabilize the enzymes during long reactions. Reactions can be carried out at ambient temperature or at 37⬚C. In stoichiometric reactions, nucleotides are produced after the monosaccharide is transferred from the donor to the acceptor saccharide. Nucleotides are potent inhibitors of glycosyltransferases; however, they can be hydrolyzed to their corresponding weakly inhibitory nucleosides by adding alkaline phosphatase to the reaction mixtures. C.

Reaction Monitoring

Thin-layer chromatography (TLC) is a convenient means of monitoring the progress of reactions throughout the preparative synthesis. Both donor and acceptor can be monitored by using Silica Gel 60-F254 plates. After aqueous enzymatic reactions have been spotted, the plates should be completely dried under high vacuum. Then the plates are developed with appropriate solvent mixtures. For UDP-Gal donor, CHCl3 /MeOH/H2O/AcOH (4:5:2:1) will differentiate between the donor and its reaction products UDP, UMP, and uridine. Both UV and spray reagents [anisaldehyde/ H2SO4 /EtOH (1:1:18) with heating] can be used for detection. If the presence of nucleotides is observed during the reaction, the alkaline phosphatase is probably not active. In this case, the addition of more alkaline phosphatase or adjustment of the pH with base will be required. Sugar nucleotides also can degrade in reactions, especially in the presence of Mn2⫹ ions. This requires the addition of sugar nucleotide donors to drive reactions to completion. Aliquots of reaction mixtures can also be analyzed by high-performance liquid chromatography (HPLC), matrix-assisted laser desorption ionization (MALDI) mass spectrometry, or NMR spectroscopy to monitor the reaction progress. At the end of the reaction it is often difficult to establish that all the acceptor has been completely converted to product. An aliquot of the reaction mixture can be removed and used in place of an acceptor substrate solution in a standard radiochemical assay. Additional glycosyltransferase and radiolabeled donor must be added. If no activity (radiolabeled product) is detected, the acceptor has been depleted from the reaction mixture. The activity of glycosyltransferases can also be monitored throughout the reaction course by removing a small aliquot from the reaction, followed by dilution and assay of activity by one of the standard assays with defined substrates in Table 1. It is desirable to monitor the pH of the reaction closely, especially when the concentration of the substrate is relatively high compared to molarity of the buffer. The pH of glycosyltransferase reactions drops as reactions proceed as a result of the production of phosphoric acid by alkaline phosphatase. Acidic pH values can inactivate glycosyltransferases and substrates, and decrease the efficiency of alkaline phosphatase.

560

D.

Qian et al.

Product Isolation

For hydrophobic products like glycolipids or other aliphatic aglycones, C18 reversed phase supports can be used for their isolation from enzyme reactions. In milligramscale synthesis, C18 Sep-Pak cartridges from Waters perform well. The resin or cartridge should be preconditioned before use by washing with MeOH, then equilibrated with water before aqueous enzymatic reaction samples are loaded. Organic solvents such as glycerol or DMSO in enzymatic reactions prevent substrates from binding to the C18 resin. In these cases, a dilution with water to less than 1% organic solvent in the enzymatic samples prior to loading on columns is recommended. Ion exchange chromatography can be used for the isolation of sialylated products and oligosaccharides with amine groups. Size-exclusion chromatography serves well for the isolation of reducing sugars or glycoconjugates with high molecular weights. Adsorption on charcoal is another practical method for the purification of oligosaccharides. A recent publication focuses on the separation of oligosaccharides [132]. E.

Example 1: Synthesis of ␣-D-Galp (1→ →3)-␤-D-Galp(1→ →4)-␤-DGlcp-O-(CH2)8CO2Me

Recombinant ␣1,3-GalT was assayed by standard radiochemical assay. In this assay, acceptor (␤-D-Galp(1→4)-␤-D-Glcp-O-(CH2)8CO2Me, 47 nmol), donor (UDP-Gal, 10 nmol), and radioactive donor (UDP-[3H]Gal, about 70,000 dpm) were incubated with enzyme in assay buffer (100 mM sodium cacodylate, 20 mM MnCl2, 0.1 mg/ mL BSA, pH 6.5) in a final volume of 20 ␮L for 10 min at 37⬚C. The reaction mixture was diluted with water to 200 ␮L and loaded onto a reversed-phase SepPak C18 cartridge that had been preequilibrated with MeOH (10 mL) and then with water (10 mL). After loading, the cartridge was washed with water to remove buffer components and unreacted donor until the counts in the eluent were background (about 50 mL). Radiolabeled trisaccharide product was eluted with MeOH (3.5 mL). The MeOH eluates were quantitated for radioactivity by liquid scintillation counting after addition of EcoLite (⫹) scintillation cocktail (from ICN, 10 mL). The enzyme activity of the stock solution was 4.4 U/mL. The components of a 4.5 mL mixture of acceptor (␤-D-Galp(1→4)-␤-D-GlcpO-(CH2)8CO2Me, 122 mg, 238 ␮mol, 50 mM in the reaction), UDP-Gal (dipotassium salt, 168 mg, 262 ␮mol, 55 mM in the reaction), ␣1,3-GalT (1 mL, 4.4 U/mL), and alkaline phosphatase (Boehringer Mannheim, 23.8 ␮L, 1 U/␮L) were combined in a 15 mL plastic falcon tube. The buffer was 124 mM MOPS, pH 7.5, containing 2 mM MnCl2 and 1 mg/mL BSA. The mixture was incubated at room temperature with gentle rotation. After 24 h, the pH of the solution was 6.5. TLC (CHCl3 /MeOH/H2O, 65:35: 8) showed that almost all the starting disaccharide had been converted to trisaccharide product. Unreacted, excess UDP-Gal and uridine were also observed by TLC (CHCl3 /MeOH/H2O/AcOH, 4:5:2:1). When an aliquot of the reaction mixture (6 ␮L, 1/30 diluted enzymatic reaction mixture) was assayed by the standard assay described above, the MeOH eluent gave 13,000 dpm, indicating that ␣1,3-GalT in the reaction mixture was still active. The same assay in the absence of additional acceptor with 6 ␮L (undiluted) of the reaction mixture gave 430 dpm in the methanol eluent, indicating that a trace of unreacted disaccharide remained.

Glycosyltransferases

561

After 48 h, the pH of the reaction was 6.5. The standard radiochemical assay for activity mixture gave 12,000 dpm in the MeOH eluent, showing minimal loss of enzyme activity in the reaction mixture. Assay in the absence of added acceptor solution gave 211 dpm, close to background 100–200 dpm, indicating that all the disaccharide had been converted to trisaccharide product. The reaction mixture was centrifuged (1450 ⫻ g), and the resultant clear su˚ ) that had been pretreated with pernatant was loaded onto a C18 column (5 g, 125 A MeOH and conditioned with water (30 mL) prior to the loading of the centrifuged sample. The column was washed with water (100 mL) then 1% acetonitrile (70 mL). The desired trisaccharide product was eluted with MeOH (HPLC grade, 120 mL). This eluate was concentrated under reduced pressure, and the product was dissolved in water (10 mL). This solution was passed through a Millex-GV filter (Millipore) to remove any residues of C18 resin. The filtrate was lyophilized to yield product as a fluffy white powder (133 mg, 196 mmol, 82%). NMR spectroscopy confirmed that the conversion was complete. F.

Example 2: Synthesis of ␣-D-Galp(1→ →3)-␤-D-Galp(1→ →4)-␣-DGlcp-O-(CH2)8CO2Me

Based on the standard radiochemical assay described above, an ␣-lactose derivative (␤-D-Galp(1→4)-␣-D-Glcp-O-(CH2)8CO2Me) has a relative rate of transfer by ␣1,3GalT that is 0.03% compared to that of the corresponding ␤-lactoside (␤-D-Galp (1→4)-␤-D-Glcp-O-(CH2)8CO2Me). Therefore, ␣1,3-GalT has 1 mU of activity for the ␣-lactoside compared to 3.3 U of activity for the ␤-lactoside. Expression cloning of ␣1,3-GalT [66,71] provides sufficient quantities for conversion of substrates with such low activities. For the following reaction, standard enzymatic activity was calculated for the ␣-lactoside. A mixture of acceptor (␤-D-Galp(1→4)-␣-D-Glcp-O(CH2)8CO2Me, 20 mg, 39 ␮mol), UDP-Gal (dipotassium salt, 30 mg, 47 ␮mol), ␣(1→3) GalT (152 mU) in 3.8 mL of 190 mM MOPS buffer, pH 7.5 containing 3.8 mM DTT, 1 M sodium chloride, MnCl2 solution (1 M, 98 ␮L), BSA solution (100 mg/mL, 40 ␮L), and alkaline phosphatase (1 U/␮L, 20 ␮L) was incubated in a 15 mL plastic falcon tube at room temperature with gentle rotation. After 4 days, the enzyme was still active when assayed with an aliquot of the enzymatic reaction and the ␤-lactoside as an acceptor. After 5 days, TLC was performed (CHCl3 /MeOH/ H2O, 65:35:8) and showed that all the starting disaccharide had been converted to trisaccharide. The reaction mixture was loaded onto a preconditioned Sep-Pak reversed phase cartridge. The cartridge was washed with water (100 mL) and 1% acetonitrile (20 mL); then trisaccharide product was eluted in 20% acetonitrile (100 mL). This eluate was concentrated, and the residue was loaded on a 75 mL Iatrobead column (Iatron Laboratories) in CHCl3 /MeOH/H2O (65:35:5). Chromatography was performed under gravity, with collection of 400 drop fractions. The fractions containing the desired products (nos. 21–36) were collected and concentrated to dryness under reduced pressure. The dried sample was loaded on a preconditioned Sep-Pak cartridge, and the cartridge was washed with water (20 mL) to remove Iatrobeads. Product was eluted with MeOH (20 mL) and this eluate was concentrated. Water (10 mL) was added to the residue and this solution was passed through a Millex-GV filter to remove residues of C18 resin. The filtrate was lyophilized to yield product as a fluffy white powder (18 mg, 26 ␮mol, 66%).

562

Qian et al.

ACKNOWLEDGMENT Funding from the Natural Sciences and Engineering Research Council of Canada for an operating and a partnership grant supported by SYNSORB Biotech Inc. is gratefully acknowledged.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33.

Varki, A., Cummings, R., Esko, J., Freeze, H., Hart, G., Marth, J., eds. Essentials of Glycobiology. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, 1999. Fukuda, M., Hindsgaul, O., eds., Molecular Glycobiology. New York: Oxford University Press, 1994. McAuliffe, J. C., Hindsgaul, O., Chem. Ind. 1997, 170. Witczak, J. Z., Nieforth, K. A., eds., Carbohydrates in Drug Design. New York: Marcel Dekker, 1997. Schmidt, R. R., Jung, K.-H., Carbohydr. Eur. 1999, 27, 13. Boons, G.-J., Tetrahedron 1996, 52, 1095. Palcic, M. M., Methods Enzymol. 1994, 230, 300. Kleene, R., Berger, E. G., Biochim. Biophys. Acta 1994, 1154, 283. Schachter, H., in Molecular Glycobiology, M. Fukuda, and O. Hindsgaul, eds. New York: Oxford University Press, 1994, pp. 88. Nunez, H. A., Barker, R., Biochemistry 1980, 19, 489. Rosevear, P. R., Nunez, H. A. Barker, R., Biochemistry 1982, 21, 1421. Wong, C.-H., Haynie, S. L., Whitesides, G. M., J. Org. Chem. 1982, 47, 5416. Auge´, C., David, S., Mathieu, C., Gautheron, C., Tetrahedron Lett. 1984, 25, 1467. Malissard, M., Zeng, S., Berger, E. G., Glycoconjugate J. 1999, 16, 125. Blixt, O., van Die, I., Norberg, T., van den Eijnden, D. H., Glycobiology 1999, 9, 1061. Kajihara, Y., Akai, S., Nakagawa, T., Sato, R., Ebata, T., Kodama, H., Sato, K.-I., Carbohydr. Res. 1999, 315, 137. Gilbert, M., Bayer, R., Cunningham, A. M., DeFrees, S., Gao, Y., Watson, D. C., Young, N. M., Wakarchuk, W. W., Nat. Biotechnol. 1998, 16, 769. Sujino, K., Jackson, R. J., Chan, N. W. C., Tsuji, S., Palcic, M. M., Glycobiology 2000, 10, 313. Butler, T., Elling, L., Glycoconjugate J. 1999, 16, 147. Koizumi, S., Endo, T., Tabata, K., Ozaki, A., Nat. Biotechnol. 1998, 16, 847. David, S., Auge´, C., Gautheron, C., Adv. Carbohydr. Chem. Biochem. 1991, 49, 175. Wong, C.-H., Halcomb, R. L., Ichikawa, Y., Kajimoto, T., Angew. Chem. Int. Ed. Engl. 1995, 34, 521. Guo, Z., Wang, P. G., Appl. Biochem. Biotechnol. 1997, 6, 1. ¨ hrlein, R., Top. Curr. Chem. 1999, 200, 227. O Palcic, M. M., Curr. Opin. Biotechnol. 1999, 10, 616. Leloir, L. F., Science 1971, 172, 1299. Paulson, J. C., Colley, K. J., J. Biol. Chem. 1989, 264, 17615. Macher, B. A., Holmes, E. H., Swiedler, S. J., Stults, C. I., Srnka, C. A., Glycobiology 1991, 1, 577. Nishida, Y., Wiemann, T., Sinnwell, V., Thiem, J., J. Am. Chem. Soc. 1993, 115, 2536. Roseman, S., Chem. Phys. Lipids 1970, 5, 270. Gastinel, L. N., Cambillau, C., Bourne, Y., EMBO J. 1999, 18, 3546. Berliner, L. J., Robinson, R. D., Biochemistry 1982, 21, 6340. Srivastava, G., Hindsgaul, O., Palcic, M. M., Carbohydr. Res. 1993, 245, 137.

Glycosyltransferases 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44.

45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67.

563

Hindsgaul, O., Kaur, K. J., Gokhale, U. B., Srivastava, G., Alton, G., Palcic, M. M., ACS Symp. Ser. 1991, 466, 38. Kodama, H., Kajihara, Y., Endo, T., Hashimoto, H., Tetrahedron Lett. 1993, 34, 6419. Kajihara, Y., Endo, T., Ogasawara, H., Kodama, H., Hashimoto, H., Carbohydr. Res. 1995, 269, 273. Palcic, M. M., Hindsgaul, O., Glycobiology 1991, 1, 205. Wong, C.-H., Wang, R., Ichikawa, Y., J. Org. Chem. 1992, 57, 4343. Yuasa, H., Hindsgaul, O., Palcic, M. M., J. Am. Chem. Soc. 1992, 114, 5891. Tsuruta, O., Shinohara, G., Yuasa, H., Hashimoto, H., Bioorg. Med. Chem. Lett. 1997, 7, 2523. Lemieux, R. U., Acc. Chem. Res. 1996, 29, 373. Berliner, L. J., Davies, M. E., Ebner, K. E., Beyer, T. A., Bell, J. E., Mol. Cell. Biochem. 1984, 62, 37. Palcic, M. M., Srivastava, O. P., Hindsgaul, O., Carbohydr. Res. 1987, 159, 315. Ichikawa, Y., Lin, Y.-C., Dumas, D. P., Shen, G.-J., Garcia-Junceda, E., Williams, M. A., Bayer, R., Ketcham, K., Walker, L. E., Paulson, J. C., Wong, C.-H., J. Am. Chem. Soc. 1992, 114, 9283. ¨ hrlein, R., Ernst, B., Berger, E. G., Carbohydr. Res. 1992, 236, 335. O ¨ hrlein, R., Ernst, B., Bioorg. Med. Chem. Lett. 1996, 6, 749. Baisch, G., O Kajihara, Y., Kodama, H., Endo, T., Hashimoto, H., Carbohydr. Res. 1998, 306, 361. Wong, C.-H., Ichikawa, Y., Krach, T., Gautheron-Le Narvor, C., Dumas, D. P., Look, G. C. J. AM. Chem. Soc. 1991, 113, 8137. Yu, L., Cabrera, R., Ramirez, J., Malinovskii, V. A., Brew, K., Wang, P. G., Tetrahedron Lett. 1995, 36, 2897. Nishida, Y., Wiemann, T., Thiem, J., Tetrahedron Lett. 1993, 34, 2905. Wiemann, T., Nashida, Y., Sinnwell, V., Thiem, J., J. Org. Chem. 1994, 59, 6744. Nishimura, S.-I., Lee, K. B., Matsuoka, K., Lee, Y. C., Biochem. Biophys. Res. Commun. 1994, 199, 249. Kopper, S., Carbohydr. Res. 1994, 265, 161. Wiemann, T., Taubken, N., Zehavi, U., Thiem, J., Carbohydr. Res. 1994, 257, C1. Schuster, M., Wang, P., Paulson, J. C., Wong, C.-H., J. Am. Chem. Soc. 1994, 116, 1135. Meldal, M., Auzanneau, F.-I., Hindsgaul, O., Palcic, M. M., J. Chem. Soc. Chem. Commun. 1994, 1849. Matsuoka, K., Nishimura, S.-I., Macromolecules 1995, 28, 2961. Blixt, O., Norberg, T., J. Org. Chem. 1998, 63, 2705. Elling, L., Zervosen, A., Gallego, R. G., Nieder, V., Malissard, M., Berger, E. G., Vliegenthart, J. F., Kamerling, J. P., Glycoconjugate J. 1999, 6, 327. Bayer, R. J., DeFrees, S., Ratcliffe, R. M., U.S. patent 5,728,254 (1998). Chan, N. W. C., Stangier, K., Sherburne, R., Taylor, D. E., Zhang, Y., Dovichi, N. J., Palcic, M. M., Glycobiology 1995, 5, 683. Wakarchuk, W., Cunningham, A., Watson, D. C., Young, N. M., Protein Eng. 1998, 11, 295. Logan, S. M., Conlan, J. W., Monteiro, M. A., Wakarchuk, W. W., Altman, E., Mol. Microbiol. 2000, 35, 1156. Endo, T., Koisumi, S., Tabata, K., Kakita, S., Ozaki, A., Carbohydr. Res. 1999, 316, 179. Joziasse, D. H., Shaper, N. L., Salyer, L. S., Van den Eijnden, D. H., van der Spoel, A. C., Shaper, J. H., Eur. J. Biochem. 1990, 191, 75. Fang, J., Li, J., Chen, X., Zhang, Y., Wang, J., Guo, Z., Zhang, W., Yu, L., Brew, K., Wang, P. G., J. Am. Chem. Soc. 1998, 120, 6635. Sujino, K., Malet, C., Hindsgaul, O., Palcic, M. M., Carbohydr. Res. 1998, 305, 483.

564

Qian et al.

68. 69. 70.

Galili, U., Immunol. Today 1993, 14, 480. Joziasse, D. H., Oriol, R., Biochim. Biophys. Acta 1999, 1455, 403. Hokke, C. H., Zervosen, A., Elling, L., Joziasse, D. H., van den Eijnden, D. H., Glycoconjugate J. 1996, 13, 687. Wang, J.-Q., Chen, X., Zhang, W., Zacharek, S., Chen, Y., Wang, P. G., J. Am. Chem. Soc. 1999, 121, 8174. Sujino, K., Uchiyama, T., Hindsgaul, O., Seto, N. O. L., Wakarchuk, W., Palcic, M. M., J. Am. Chem. Soc. 2000, 122, 1261. Stults, C. L. M., Macher, B. A., Bhatti, R., Srivastava, O. P., Hindsgaul, O., Glycobiology 1999, 9, 661. Blanken, W. M., van den Eijnden, D. H., J. Biol. Chem. 1985, 260, 12927. ¨ hrlein, R., Kolbinger, F., Streiff, M., Bioorg. Med. Chem. Lett. 1998, 8, Baisch, G., O 1575. Qian, X., Sujino, K., Otter, A., Palcic, M. M., Hindsgaul, O., J. Am. Chem. Soc. 1999, 121, 12063. Hennet, T., Dinter, A., Kuhnert, P., Mattu, T. S., Berger, E. G., J. Biol. Chem. 1998, 273, 58. Kolbinger, F., Streiff, M. B., Katopodis, A. G., J. Biol. Chem. 1998, 273, 433. ¨ hrlein, R., Kolbinger, F., Streiff, M., Bioorg. Med. Chem. Lett. 1998, 8, Baisch, G., O 751. Yamamoto, F.-I., Hakomori, S.-I., J. Biol. Chem. 1990, 265, 19257. Greenwell, P., Glycoconjugate J. 1997, 14, 159. Lowary, T. L., Hindsgaul, O., Carbohydr. Res. 1993, 249, 163. Lowary, T. L., Hindsgaul, O., Carbohydr. Res. 1994, 251, 33. Mukherjee, A., Palcic, M. M., Hindsgaul, O., Carbohydr. Res. 2000, 326, 1. Greenwell, P., Yates, A. D., Watkins, W. M., Carbohydr. Res. 1986, 149, 149. Seto, N. O. L., Palcic, M. M., Compston, C. A., Li, H., Bundle, D. R., Narang, S. A., J. Biol. Chem. 1997, 272, 14133. Seto, N. O. L., Compston, C. A., Szpacenko, A., Palcic, M. M., Carbohydr. Res. 2000, 324, 161. Hakomori, S.-I., Adv. Cancer Res. 1989, 52, 257. Feizi, T., Nature 1985, 314, 53. Natsuka, S., Lowe, J. B., Curr. Opin. Struct. Biol. 1994, 4, 683. Oriol, R., Mollicone, R., Cailleau, A., Balanzino, L., Breton, C., Glycobiology 1999, 9, 323. Watkins, W. M., Carbohydr. Res. 1986, 149, 1. Eppenberger-Castori, S., Lotscher, H., Finne, J., Glycoconjugate J. 1989, 6, 101. Grabenhorst, E., et al., J. Biol. Chem. 1998, 273, 30985. Gokhale, U. B., Hindsgaul, O., Palcic, M. M., Can. J. Chem. 1990, 68, 1063. Stangier, K., Palcic, M. M., Bundle, D. R., Hindsgaul, O., Thiem, J., Carbohydr. Res. 1998, 305, 511. Srivastava, G., Kaur, K. J., Hindsgaul, O., Palcic, M. M., J. Biol. Chem. 1992, 267, 22356. Hallgren, C., Hindsgaul, O., J. Carbohydr. Chem. 1995, 14, 453–464. Gosselin, S., Palcic, M. M., Bioorg. Med. Chem. 1996, 4, 2023. Nikrad, P. V., Kashem, M. A., Wlasichuk, K. B., Alton, G., Venot, A. P., Carbohydr. Res. 1993, 250, 145. Ding, Y., Hindsgaul, O., Li, H., Palcic, M. M., Bioorg. Med. Chem. Lett. 1998, 8, 3199. Ogawa, S., Matsunaga, N., Li, H., Palcic, M. M., Eur. J. Org. Chem. 1999, 631. Qian, X., Hindsgaul, O., Li, H., Palcic, M. M., J. Am. Chem. Soc. 1998, 120, 2184.

71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103.

Glycosyltransferases 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118.

119. 120. 121. 122. 123. 124. 125. 126. 127.

128. 129. 130. 131. 132.

565

Chandrasekaran, E. V., Jain, R. K., Larsen, R. D., Wlasichuk, K., DiCioccio, R., A., Matta, K. L., Biochemistry 1996, 35, 8925. ¨ hrlein, R., Katopodis, A., Streiff, M., Kolbinger, F., Bioorg. Med. Chem. Baisch, G., O Lett. 1997, 7, 2447. ¨ hrlein, R., Katopodis, A., Ernst, B., Bioorg. Med. Chem. Lett. 1996, 6, Baisch, G., O 759. ¨ hrlein, R., Streiff, M., Bioorg. Med. Chem. Lett. 1998, 8, 161. Baisch, G., O ¨ hrlein, R., Katopodis, A., Bioorg. Med. Chem. Lett. 1997, 7, 2431. Baisch, G., O ¨ hrlein, R., Top. Curr. Chem. 1999, 200, 227. O ¨ hrlein, R., Katopodis, A., Bioorg. Med. Chem. Lett. 1997, 7, 2431. Baisch, G., O ¨ hrlein, R., Streiff, M., Kolbinger, F., Bioorg. Med. Chem. Lett. 1998, 8, Baisch, G., O 755. ¨ hrlein, R., Bioorg. Med. Chem. 1998, 6, 1673. Baisch, G., O ¨ hrlein, R., Katopodis, A., Bioorg. Med. Chem. Lett. 1996, 6, 2953. Baisch, G., O Paulson, J. C., Colley, K. J., J. Biol. Chem. 1989, 264, 17615. Harduin-Lepers, A., Recchi, M.-A., Delannoy, P., Glycobiology 1995, 5, 741. Tsuji, S., Datta, A. K., Paulson, J. C., Glycobiology 1996, 6, v. Kelm, S., Schauer, R., Int. Rev. Cytol. 1997, 175, 137. Hasegawa, A., Kiso, M., in Carbohydrates—Synthetic Methods and Applications in Medicinal Chemistry, H. Ogura, A. Hasegawa, and T. Sumai, eds., Kodansha Publishers, Tokyo, 1992, pp. 243–246. Higa, H. H., Paulson, J. C., J. Biol. Chem. 1985, 260, 8838. Gross, H. J., Bunsch, A., Paulson, J. C., Brossmer, R., Eur. J. Biochem. 1987, 168, 595. Brossmer, R., Gross, H. J., Methods Enzymol. 1994, 247, 153. Gross, H. J., Rose, U., Krause, J. M., Paulson, J. C., Schmid, K., Feeney, R. E., Brossmer, R., Biochemistry 1989, 28, 7386. Gross, H. J., Brossmer, R., Eur. J. Biochem. 1988, 177, 583. Chappell, M. D., Halcomb, R. L., J. Am. Chem. Soc. 1997, 119, 3393. Gross, H. J., Brossmer, R., Glycoconjugate J. 1987, 4, 145. Wlasichuk, K. B., Kashem, M. A., Nikrad, P. V., Bird, P., Jiang, C., Venot, A. P., J. Biol. Chem. 1993, 268, 13971. Van Dorst, J. A. L. M., Tikkanen, J. M., Krezdorn, C. H., Streiff, M. B., Berger, E. G., Van Kuik, J. A., Kamerling, J. P., Vliegenthart, J. F. G., Eur. J. Biochem. 1996, 242, 674. ¨ hrlein, R., Streiff, M., Ernst, B., Bioorg. Med. Chem. Lett. 1996, 6, 755. Baisch, G., O ¨ hrlein, R., Streiff, M., Bioorg. Med. Chem. Lett. 1998, 8, 157. Baisch, G., O Gilbert, M., Cunningham, A. M., Watson, D. C., Martin, A., Richards, J. C., Wakarchuk, W. W., Eur. J. Biochem. 1997, 249, 187. Endo, T., Koizumi, S., Tabata, K., Ozaki, A., Appl. Microbiol. Biotechnol. 2000, 53, 257. Fang, J., Chen, X., Wang, J., Andreana, P. R., Wang, P. G., J. Org. Chem. 1999, 64, 4089.

15 Biotransformation of Polysaccharides H. N. Cheng and Qu-Ming Gu Hercules Incorporated, Wilmington, Delaware

I.

INTRODUCTION

Polysaccharides are among the world’s most abundant raw materials. They are structurally diverse and exhibit a wide range of properties. As a result, they are extensively utilized for a large number of applications [1–4]. In the industrial chemicals business, polysaccharides have been used for many years as specialty products and as functional additives in food, coatings, construction, paper, pharmaceutical, personal care, and other niche applications [1–6]. Because polysaccharides are natural materials, they are ideally suited for enzymatic and microbial reactions. This field has been growing in the past 10 years, and a large number of papers have been published [7–9]. Much of the research and development is targeted toward pharmaceutical or agrichemical applications. Less emphasis has been placed on the specialty chemicals area. This chapter discusses a number of reactions being studied at Hercules, a specialty chemicals company. II.

POLYSACCHARIDES AND PRODUCT DEVELOPMENT

Figure 1 gives a simplified product development scheme. The raw material may come from seeds (e.g., guar, locust bean gum, corn, wheat starch), tubers and roots (e.g., potato and tapioca starch, konjac), fruits (pectin), seaweeds (e.g., agar, alginate, carrageenan, furcellaran), plant exudates (e.g., ghatti, gum arabic, karaya, tragacanth), and various plant sources (cellulose), or microorganisms (dextran, gellan, pullulan, xanthan). These substances are collected through appropriate means, cleaned, purified, and often modified. The modification may be chemical (e.g., to form substituted celluloses or modified starches), or microbial (e.g., to convert sucrose to dextran). For a polysaccharide to have commercial value as a specialty chemical, it must have some functional attributes. These attributes are related to the polysaccharide 567

568 Product development scheme for polysaccharides as specialty chemicals. (Courtesy Cheng and Gu,

Cheng and Gu

Figure 1 1999.)

Biotransformation of Polysaccharides

569

structure, and indeed much work has been done to understand the polysaccharide structure–property relationships [1–6]. Structural features of interest include the types of sugar the polysaccharide contains, linkage positions and configurations, presence or absence of charged groups, molecular weight, branching, and substitution pattern (if applicable). Sometimes the polysaccharide in question contains proteins and/or fatty acids, which can modify its properties. In addition, many polysaccharides have secondary or tertiary structures (e.g., helix, intramolecular hydrogen bonding, and interchain association). These can also affect the polysaccharide properties. In a commercial context, the end use dictates the properties desired. For solution applications, the rheology of the polysaccharide, its ability to retain water, and its gelling tendency are often the most important. For solids applications, the thermal properties (e.g., Tg and Tm), the mechanical properties (e.g., stiffness, tensile, texture, and adhesion), and other features such as water content, crystallinity, and spatial heterogeneity are relevant. Many polysaccharides are used in interfacial applications, in which case the surface-active properties of the polysaccharide are important. Sometimes a polysaccharide is desirable in some respects but lacking in others for a given application. In this case a producer may consider improving the polysaccharide. Improvements may entail simple operations such as purification, fractionation, blending, or the addition of an inorganic salt or other organic/polymeric additives. Alternatively, chemical modification or biotransformation (both enzymatic and microbial) can be carried out to alter the chemical or the physical structures of the polysaccharide. III.

BIOTRANSFORMATION

Biotransformation can be viewed as a ‘‘toolbox’’ to produce polysaccharides with targeted properties [7–10]. One can use hydrolase enzymes to lower the molecular weight. One can use enzymatic reactions to add or modify a substituent on a polysaccharide. One can add a monosaccharide or a second polysaccharide to produce a larger molecule. Some polysaccharides contain proteins or fatty acids, and these can be cleaved or otherwise altered to produce different properties. Some general comments can be made concerning the biotransformation of polysaccharides. First, polysaccharides are usually soluble only in water and a few selected solvents. Water is good for hydrolysis reactions, but usually not for organic synthesis and modification reactions when hydrolytic enzymes are used as catalysts. For the latter reactions, we must either choose an enzymatic reaction that can be carried out in water or use polar aprotic solvents (e.g., dimethyl sulfoxide, N,Ndimethylformamide, and N,N-dimethylacetamide), in which polysaccharides are often soluble. Occasionally, polysaccharides can be suspended in suitable organic solvents, such as pyridine and toluene, to carry out enzymatic reactions. Many hydrolytic enzymes are active and stable in organic solvents [11–13]. Second, some enzymes are expensive and may not be stable at high temperatures. Thus, these reactions should be designed to minimize enzyme use and to avoid more extreme reaction conditions. The recent developments in directed evolution of enzymes have alleviated this problem [14]. Third, enzyme reactions should always be compared in practical terms with equivalent chemical reactions (if available). Relative to chemical methods, enzymes can impart high chemo-, enantio-, and regioselectivity, reduce by-product yields, and require milder reaction conditions; moreover, they are environmentally

570

Cheng and Gu

friendly. If these features are not needed, the enzymatic process should be compared with the corresponding chemical process on a cost-competitive basis. We shall describe two types of specialty polysaccharides. The first type consists of commercial polysaccharides that have been modified with special functional groups. Those in the second type are produced by the synthesis of several oligosaccharides and low molecular weight polysaccharides.

IV.

MODIFIED POLYSACCHARIDES

Modifications are sometimes desired to modulate the solubility or the viscosity of a given polysaccharide. Two reactions illustrate this approach: pectin/glycerol ester and succinated guar. A.

Pectin/Glycerol Ester

Pectin is extracted from citrus peel, apple pomace, sugar beet, and other plant materials. Structurally it comprises polygalacturonic acid and its methyl ester, and a number of neutral sugars [15]. It is widely used in food applications (e.g., jams and jellies, confectionery, beverages, desserts) and in some pharmaceutical applications. It is susceptible to enzyme reactions, and the following three pectolytic enzymes are well known: polygalacturonase (cleaving the galacturonan backbone), pectin methylesterase (hydrolyzing the methyl ester), and pectin lyase (which catalyzes the ␤scission of the backbone [16]. It is of interest to use a nonpectin enzyme to carry out biotransformation of pectin. One reaction is to convert the pectin to its glycerol ester through transesterification (Scheme 1). Since glycerol is very hydrophilic, this modification should make the pectin more water soluble at low pH values. To find a suitable enzyme to catalyze the transesterification reaction (methyl ester to glycerol ester) under mild conditions, we screened a number of lipases and esterases. We found two lipases (Candida cylindracea lipase and porcine pancreatic lipase) to exhibit catalytic activity for this reaction. The reaction conditions entailed elevated temperatures, with the use of glycerol as the solvent and an acyl acceptor. The enzyme catalyzed the esterification of the carboxyl group of the pectin to the C1 (primary) alcohol of the glycerol to form the monoester. No substantial glycerol

Scheme 1

Synthesis of pectin/glycerol ester.

Biotransformation of Polysaccharides

571

diester (i.e., cross-linked structure) was found. Pectin methylesterase (PME) was also studied and found not to have catalytic activity under the same reaction conditions. B.

Succinated Guar

Guar is a galactomannan derived from the seed of the guar plant. It is extensively used in food and industrial applications as a thickener, sizing agent, flocculent, and gelling agent. Since guar is a neutral polysaccharide, sometimes it is desirable to convert it to an anionic polymer. This can be done via the reaction of guar with succinic anhydride in the presence of a suitable lipase (Scheme 2). Previously, acid anhydrides such as succinic anhydride [17,18] and acetic anhydride [19] were used for enzymatic acylation of small organic compounds or monosaccharides via lipases, but not much has been published on polysaccharides. Three enzymes, lipase from Pseudomonas sp. (lipase AK, Amano), lipase from Candida cylindracea (Sigma), and lipase OF-360 (Meito-Sangyo Ltd.), have been studied for this acylating reaction. Among the three enzymes, lipase AK showed the highest catalytic activity and Candida cylindracea lipase the lowest activity. Interestingly, the succinated guar shows somewhat enhanced viscosity relative to that of the guar [20]. V.

LOW MOLECULAR WEIGHT POLYSACCHARIDES

Low molecular weight polysaccharides and oligosaccharides are common synthetic targets [21]. Many of them have found applications as pharmaceuticals [22,23] and nutriceuticals [24,25] (also known as functional foods). As specialty chemicals, these materials have been used, for example, as binders, encapsulants, protective colloids, bulking agents, and film formers. The preparations of three low molecular weight polysaccharides are given here as examples: galactosylated polyethylene glycol, galactose-containing oligosaccharide, and biomodified guar. The first two examples rely only on transglycosylation reactions, and the last example uses a combinatorial mixture of hydrolase enzymes. A.

Galactosylated Polyethylene Glycol and Glycidol

The reaction of lactose and ␤-galactosidase permits galactose to be appended at the end of polyethylene glycol. We also discovered that galactose can be attached to the primary alcohol of glycidol while the epoxide functional group remains intact (Scheme 3). The use of lactose and ␤-galactosidase for glycosylation has been shown for a number of disaccharides and trisaccharides [26]. The reaction of Scheme 3 was found to be efficient for ␤-galactosidase from Aspergillus oryzae, but not for the enzyme from E. coli. The reaction was relatively facile and entailed mixing glycidol, galactosidase, and lactose, followed by stirring for a period of time at room temperature. B.

Galactose-Containing Oligosaccharides

In a systematic study carried out recently on the E. coli ␤-galactosidase-catalyzed transgalactosylation reaction of oligosaccharides, lactose served as a galactose donor [27]. It was found that the galactosylation yield was 40–60% when disaccharides

572

Scheme 2

Lipase-catalyzed synthesis of succinated guar.

Cheng and Gu

Biotransformation of Polysaccharides

Scheme 3

Galactosylated polyethylene glycol and glycidol.

573

574

Cheng and Gu

were used as the galactose acceptor. The yields were around 20% for trisaccharides, about 8% for tetrasaccharides, and only 1–5% for higher oligosaccharides. Interestingly, no transgalactosylation was observed when sucrose was used as the galactose acceptor. The transgalactosylation reactions for two disaccharides are shown in Scheme 4. C.

Biomodified Guar

In an earlier work [27], several enzymes (lipase, hemicellulase, and protease) were shown to have noticeable effects on the viscosity of guar solutions. Guar typically contains more than 80% soluble galactomannans, 7% proteins, 2–3% fatty acids, and small amounts of insoluble galactomannans and cellulosics [28]. In this work the effects of four enzymes and their combinations are examined. The enzymes are as follows: A, lipase from Mucor miehei (Amano, lipase M, 10 mg); B, hemicellulase (Sigma, 10 mg); C, pectinase (Novo Nordisk, Pectinex 3XL, 20 ␮L); D, protease N (Amano, 10 mg). Single component: Two components: Three components: Four components:

A, B, C, D AB, AC, AD, BC, BD, CD ABC, BCD, ABD, ACD ABCD

The results of single-component enzyme treatments are given in the top part of Table 1. In each case, the treatment was carried out on 60 g of 1% solutions (w/w) of neutral guar (Supercol威 U, Hercules) at pH 6.95. The guar solutions were measured before and after the treatment for Brookfield viscosity, molecular weights (Mn and MW), and optical density. It is clear that the hemicellulase has the largest effect on Brookfield viscosity and molecular weight. Presumably it contains mannanase, which can cleave the galactomannan backbone. Interestingly, it increases the optical density of the guar solution. The polydispersity (MW/Mn) of lipase-treated guar solution is very large, with a multimodal molecular weight distribution, indicating that only part of the guar is affected (probably the fatty acid containing portion). Pectinase and protease treatments appear to give similar results. The results for enzyme treatment with pairs of enzymes are also shown in Table 1. The combinations AC and AD only marginally decrease the molecular weight, relative to A alone; however, the optical density is noticeably lower. Since B is so effective in lowering the molecular weight, the combinations AB, BC, and BD all yield low molecular weights, similar to B alone. The combination CD is interesting in that the overall molecular weight decreases, but the polydispersity increases substantially (MW/Mn = 20.6). The two enzymes seem to work on different parts of guar. From the data it is clear that D has a large impact on optical density. All the combinations involving D give low optical densities. The results of three- and four-component enzyme treatments are depicted in the lower part of Table 1. Again the ability of B to degrade the molecular weight is clear in the combinations involving B (viz., ABC, ABD, BCD, and ABCD). The ability of D to lower the optical density is also apparent in the combinations involving D (i.e., ABD, ACD, BCD, and ABCD). Through such combinatorial experiments, we can determine the optimal enzyme combinations needed for targeted molecular weight, viscosity, or optical den-

Biotransformation of Polysaccharides

␤-Galactosidase catalyzed synthesis of oligosaccharides.

575

Scheme 4

576

Cheng and Gu

Table 1

Effect of Enzymesa on Guarb in Solution

Sample

BVc

Mn (Da)d

Mw (Da)d

MW/Mnd

ODe

control A B C D AB AC AD BC BD CD ABC ABD ACD BCD ABCD

15,920 1,720 30 880 1,160 40 530 520 40 70 320 40 50 280 35 50

107,300 37,600 12,600 151,600 179,200 12,100 35,400 33,500 11,600 9,800 41,300 10,100 9,900 28,470 96,600 9,800

1,811,000 1,336,200 161,100 1,280,500 1,414,700 161,700 1,111,100 1,092,600 146,200 118,600 849,500 119,500 123,700 780,300 1,010,400 121,900

16.9 35.6 12.8 8.5 7.9 13.1 31.4 32.6 12.6 12.2 20.6 11.9 12.5 27.4 10.5 12.4

2.05 1.90 2.14 1.61 1.60 1.87 1.46 1.44 2.10 1.57 1.60 1.81 1.47 1.46 1.59 1.46

a

The enzymes are as follows: A, lipase; B, hemicellulase; C, pectin methylesterase; D, protease. Neutral guar used (Supercol威 from Hercules Incorporated). c Brookfield viscosity at room temperature for 1% solution at 3 cpm using spindle 3. d SEC measurement, PEO standard. e Optical density measured by UV absorbance at 360 nm. b

sity. If the lowest molecular weight and optical density are desired, the combination ABD gives the best results. If the narrowest molecular weight distribution is needed, then the single enzyme C or D gives the best performance. If the lowest optical density is the goal, then the combination AD is the best, followed by AC, ACD, and ABCD. It may be noted that this experiment represents only the starting point for optimization. Additional combinatorial experiments can be done to optimize the proportions of the four enzymes and to fine-tune the enzyme formulation.

VI.

EXPERIMENTAL SECTION

A.

Pectin/Glycerol Ester

HM-pectin (1 g, 73% DE, GENU) was partially swollen in 20 mL of glycerol. Then 50 mg of lipase from Candida cylindreacea (Lipase OF-360, Meito-Sangyo) was added to the solution and the mixture was stirred a 55⬚C for 24 h. The modified pectin was precipitated in methanol and recovered by filtration. After being washed extensively with methanol, the product was obtained as a powder. The yield was 0.98 g. Analysis of the product (MW = 125,000, MW/Mn = 6.8) by size-exclusion chromatography indicated around 2–3% increase in the average molecular weight from the starting material (MW = 123,000, MW/Mn = 6.4), corresponding roughly to a degree of esterification of 2% (with glycerol). 1H NMR analysis showed the presence of glycerol moiety. The aqueous solution viscosity of the modified product slightly increased in comparison to the starting pectin.

Biotransformation of Polysaccharides

B.

577

Succination of Guar

As a typical procedure, crude lipase from Pseudomonas sp. (AK) (100 mg) was added to a mixture of flake guar (0.5 g) and succinic anhydride (0.2 g) in 2 mL of dimethylformamide (DMF) at 50⬚C, and the mixture was stirred for 24 h. The solid residue was washed three times with isopropanol and dried under vacuum to give 0.49 g of product in 98% yield. An IR spectrum of the product showed the presence of a strong ester bond at 1745 cm⫺1, whereas the spectrum of the control sample (without enzyme under the same conditions) gave the same band at 1745 cm⫺1 but at much less intensity. C.

Galactosylation of Polyethylene Glycol

Lactose (10 g) and polyethylene glycol (10 g, Fluka, MW, 8000) were dissolved in 100 mL of water, and ␤-galactosidase from Aspergillus oryzae (50 mg, Sigma) was added. The reaction mixture was stirred at room temperature overnight. The enzyme was inactivated by heating the reaction mixture up to 80⬚C for 5 min, and the resulting solution was concentrated under reduced pressures. After the unreacted starting materials and by-products had been removed, the product was isolated. The galactosylation yield was about 30–50% based on NMR analysis. D.

Galactosylation of Glycidol

Lactose (500 mg) and glycidol (1.5 mL) were dissolved in 4 mL of 0.1 M phosphate buffer, pH 7.0. ␤-Galactosidase from Aspergillus oryzae (69 units, Sigma) was added. The reaction mixture was stirred at 37⬚C for 24 h. Analysis by thin-layer chromatography (TLC) showed a new spot with a higher Rf value than the starting materials. 1 H NMR analysis indicated the formation of the product. The yield was estimated to be around 20% by the TLC analysis. E.

Galactose-Containing Cellobiose

A mixture containing lactose (8 mmol), cellobiose (3 mmol), and ␤-galactosidase from E. coli (50 units) in 10 mL of 0.1 M, phosphate buffer (pH 7.0) was incubated at 37⬚C. The progress of the reaction was monitored via TLC (isopropanol/H2O/ NH4OH, 7:4:1). A small aliquot of the reaction mixture was also withdrawn at certain time intervals and analyzed by reversed-phase high-performance liquid chromatography (HPLC). After 40 min, the transgalactosylation yield reached 50–60% based on HPLC analysis. The reaction was stopped and the product was purified by flash column chromatography (silica gel, eluted by isopropanol/H2O/NH4OH, 7:4:1). The isolation yield was 37%. Structure and conformation of the product were confirmed by 13C and 1H NMR analyses. F.

Biomodified Guar

Guar (1 g, Supercol, Hercules) was dissolved in 100 mL of 0.2 M phosphate buffer (pH 7.0). Enzymes in different combinations were added. The pH of the resulting solution was adjusted if necessary to reach 7.0. The mixture was incubated at room temperature (22⬚C) for 18 h. The Brookfield viscosities of the solutions were measured before isolation and purification. The products were obtained by precipitating

578

Cheng and Gu

in and washing with isopropanol followed by drying in air. The Brookfield viscosity was measured using an LVDV Brookfield viscometer (spindle 3) at room temperature. The samples were also analyzed by size-exclusion chromatography using polystyrene standards. No corrections were made for the difference in hydrodynamic volumes of polystyrene and guar. Thus, the molecular weights reported in Table 1 are not absolute values. The turbidity of each guar solution was determined by measuring its optical density at 360 nm. ACKNOWLEDGMENT We thank Dr. Barry L. Marrs for helpful discussions, and R. W. Kliment and S. Mital for their assistance in performing many of the experiments. REFERENCES 1. 2. 3. 4. 5.

6. 7. 8. 9. 10. 11. 12. 13. 14.

15. 16. 17. 18. 19. 20. 21.

R. L. Whistler and J. N. BeMiller, eds., Industrial Gums, 3rd ed. Academic Press, New York, 1991. R. L. Davidson, ed., Handbook of Water Soluble Gums and Resins. McGraw-Hill, New York, 1990. A. M. Stephens, ed., Food Polysaccharides and Their Applications. Marcel Dekker, New York, 1995. H. N. Cheng, in Polymeric Materials Encyclopedia, J. C. Salamone, ed. CRC Press, Boca Raton, FL, 1996, pp. 2568–2574. G. O. Phillips, D. J. Wedlock, and P. A. Williams, eds., Gums and Stabilizers for the Food Industry, Pergamon Press, Oxford, vol. 1, 1982; vol. 2, 1984; Elsevier, London, vol. 3, 1986; IRL Press, Oxford, vol. 4, 1988; vol. 5, 1990; vol. 6, 1992. A. Imeson, ed., Thickening and Gelling Agents for Food. Blackie Academic, London, 1992. C.-H. Wong, G. M. Whitesides, Enzymes in Synthetic Organic Chemistry. Pergamon Press, London, 1994. S. M. Roberts, N. J. Turner, A. J. Willetts, M. K. Turner, Introduction to Biocatalysis. Cambridge University Press, Cambridge, 1995. K. Faber, Biotransformations in Organic Chemistry, 3rd ed. Springer-Verlag, Berlin and Heidelberg, 1997. J. N. Saddler, M. H. Penner, eds., Enzymatic Degradation of Insoluble Carbohydrates (ACS Symp. Ser., 618). American Chemical Society, Washington, DC, 1995. A. Zaks, A. M. Klibanov, Proc. Natl. Acad. Sci. USA, 82:3192–3196 (1985). A. M. Klibanov, Acc. Chem. Res., 23:114–120 (1990). J.-M. Fand, C.-H. Wang, SynLett, 339–340, 1994. For example, (a) Z. Lin, T. Thorsen, F. H. Arnold, Biotechnol. Prog., 15, 467–471 (1999). (b) F. H. Arnold, Acc. Chem. Res., 31, 125–131 (1998). (c) W. P. C. Stemmer, Nature, 370, 389–391 (1994). Genu威 Pectin—General Description . Hercules Incorporated, commercial brochure. J. Visser, A. G. J. Voragen, eds., Pectins and Pectinases. Elsevier, Amsterdam, 1996. J. Hiratake, K. Yamamoto, Y. Yamamoto, J. Oda, Tetrahedron Lett., 30, 1555–1556 (1989). K. Faber, S. Riva, Synthesis, 895–910 (1992). A. Uemura, K. Nozaki, J.-I. Yamashita, M. Yasumoto, Tetrahedron Lett., 30, 3819–3820 (1989). Q.-M. Gu, U. Wandel, H. Vaessen, unpublished results. P. Gagneux, A. Varki, Glycobiology, 9, 747–755 (1999).

Biotransformation of Polysaccharides 22. 23. 24. 25. 26.

27. 28.

579

D. Zopf, P. Simon, R. Barthelson, D. Cundell, I. Idanpaan-Heikkila, E. Tuomanen, Adv. Exp. Med. Biol., 408:35–38 (1996). T. Mizuochi, M. Nakata, Nippon Rinsho, 53(9):2340–2349 (1995). D. S. Newburg, Curr. Med. Chem., 6(2):117–127 (1999). S. T. Yang, E. M. Silva, J. Dairy Sci., 78(11):2541–2562 (1995). P. McVeagh, J. B. Miller, J. Paediatr. Child Health, 33(4):281–286 (1997). For example: (a) K. G. Nilsson, Carbohydr. Res., 67, 95–103 (1987), (b) W. H. Binder, H. Kahlig, W. Schmid, Tetrahedron Lett., 50, 10407–10418 (1994), (c) A. Jamczuk, J. Li, Zhang, W., X. Chen, Y. Chen, J. Fang, J. Wang, P. G. Wang, Curr. Med. Chem., 6, 155–164 (1999), (d) D. E. Stevenson, R. A. Stanley, R. H. Furneaux, Biotechnol. Bioeng., 42:657–667 (1993), (e) C. Schechermann, F. Wagner, L. Fisher, Enzyme Microbiol. Technol., 20:629–634 (1997), (f) Y. Ooi, T. Hashimoto, N. Mitsuo, T. Satoh, Chem. Pharm. Bull., 33:1808–1814 (1985). Q.-M. Gu, J. Environ. Polym. Degrad., 7(1), 1–7 (1999). Supercol威 Guar Gum, Hercules Incorporated, Commercial brochure.

16 From a Carbohydrate Sequence to Extensive Research Programs: The ␣-gal Epitope as a Case Study Peter R. Andreana, Wei Zhang, and Peng George Wang Wayne State University, Detroit, Michigan

I.

INTRODUCTION

Hyperacute rejection presents a formidable barrier to xenotransplantation, the transplantation of organs from nonprimate to human. The major xenoactive antigens responsible for such rejection have recently been identified as carbohydrate structures containing terminal ␣-D-Galp-(1→3)-␤-D-Galp sequence (␣-gal epitopes). The most common ␣-gal epitopes are disaccharide ␣-D-Galp-(1→3)-␤-D-Galp, trisaccharides ␣-D-Galp-(1→3)-␤-D-Galp-(1→4)-␤-D-GlcNAcp, and ␣-D-Galp-(1→3)-␤-D-Galp(1→4)-␤-D-Glcp, and pentasaccharide ␣-D-Galp-(1→3)-␤-D-Galp-(1→4)-␤-DGlcNAcp-(1→3)-␤-D-Galp-(1→4)-␤-␤-D-Glcp. The ␣-gal epitopes are abundantly expressed on the cells of most mammals, with the exception of humans, apes, and Old World monkeys (monkeys of Asia and Africa). The unique enzyme responsible for its formation is ␣1,3-galactosyltransferase (␣1,3-GT), which is inactive in Old World primates. Conversely, the natural antibody with specificity to ␣-gal epitope (anti-Gal) exists only in humans and other Old World primates. The discovery of the interaction of ␣-gal and anti-Gal has led to experimental attempts to overcome hyperacute rejection by either depleting the recipient’s anti-Gal through an ␣-gal containing affinity column (anti-Gal immunoadsorption approach) or antagonizing anti-Gal by infusing soluble synthetic ␣-gal oligosaccharides (anti-Gal neutralization approach). Moreover, ␣-gal–antigen conjugates were shown to enhance antigen presentation by the natural human anti-Gal antibody, and ␣-gal–immunotoxin glycoconjugates are being used to target B cells, which generate anti-Gal antibody. All these biomedical applications require access to a substantial amount of ␣-gal oli581

582

Andreana et al.

gosaccharides as well as synthetically derived ␣-gal analogs and mimetics with high affinity to anti-Gal antibodies. Through the interpretation of NMR experiments of ␣gal epitopes, the answer behind the structural motif for antibody/antigen complexation may be unraveled as varying ␣-gal derivatives are used to probe the binding site of the anti-Gal antibody.

II.

GLYCOBIOLOGY OF ␣-GAL AND ANTI-GAL

A.

Xenotransplantation

In the early years of this century the development of vascular anastomosis, a surgical technique allowing the suturing of cut ends of blood vessels, prompted the first attempts to transplant organs from one individual to another [1]. The successful clinical application of organ transplantation had, however, to wait a further 50 years until the development of immunosuppressive drugs allowed physicians to control the immune response of the recipient against the graft. Now, with modern methods of immunosuppression, the transplantation of tissues and organs between human subjects has become a regular part of medical practice and the preferred treatment for chronic failure of the kidneys, heart, liver and lungs. Nonetheless, the severe shortage of human organs for transplantation (e.g., kidney, liver, heart) has increased the interest in the possible use of xenogenetic organs for transplantation in humans. The use of primate donors is restricted for two main reasons: there are a limited number of suitable animals in captivity, and there are concerns regarding animal rights. This has prompted intense research in the possible use of nonprimate mammals (in particular, pigs) as xenograft donors. Aside from overcoming the shortage of human organs and tissues, xenotransplantation could offer some advantages over the use of human organs. One advantage is that a xenotransplant can be planned in advance; the organ can be transplanted while it is still fresh and undamaged, and planned transplantation allows therapeutic regimes that call for the pretreatment of the recipient. Because disease pathogenesis is often species specific, the use of animal organs or tissues may also help avoid the recurrence of disease in the graft. Yet another advantage is the possibility that animal donors could be genetically or biochemically manipulated in order to lower the risk of rejection, or even to express new genes or alter biochemical processes for the benefit of the patient. One might intuitively think that the most suitable species to use as a source of organs or tissues for xenotransplantation would be a species phylogenetically close to humans, the baboon being one primate species that has been used. It is evident, however, that lower, phylogenetically more distant mammals are preferred, the most favorable one being the pig. Pigs are available in large numbers, and their organs are of a size suitable for use in adult patients, whereas baboons and other nonhuman primates are relatively scarce and too small to provide organs for full-grown human subjects. Also, unlike primates, pigs can be genetically manipulated to express heterologous genes; this procedure can be used to address some biological hurdles to xenotransplantation or to confer new physiological properties on the graft. The first hurdle to xenotransplantation is hyperacute rejection (HAR), which can destroy a transplanted organ in minutes to a few hours [2]. HAR of a pig organ in humans is triggered by the binding of xenoreactive natural antibodies to the endothelial lining of the blood vessels within the graft. In 1992 Cooper and coworkers

The ␣-gal Epitope

583

found that the xenoactive antigens responsible for HAR are oligosaccharides with a terminal ␣-D-Galp-(1→3)-␤-D-Galp moiety [3]. Such an oligosaccharide epitope is subsequently termed ␣-galactosyl epitope (␣-gal), and the natural antibody with specificity to ␣-gal is called anti-Gal [4]. In the initial stage of hyperacute rejection, antiGal IgG binds to ␣-gal epitopes expressed on glycolipids and glycoproteins of xenograft cells and results in antibody-dependent cell-mediated cytotoxicity by human blood monocytes and macrophages. The IgM isotype of anti-Gal fixes the complement, and a cascade of events that follows leads to complement-mediated lysis of the xenograft cells. By inhibiting anti-Gal binding to the graft or by suppressing the activation of the complement system, HAR can be reliably prevented in experimental animal models. However, long-term graft survival has not been achieved owing to other immunological barriers such as acute vascular rejection, cellular rejection, and chronic rejection [5,6]. The interaction of ␣-gal and anti-Gal also plays a role in the delayed rejection following abrogation of HAR, but the precise mechanism has not yet been defined [7]. B.

Anti-Gal, ␣-gal, and ␣1,3-Galactosyltransferase

Human beings acquire, during their lifetime, a variety of natural antibodies specific for carbohydrate epitopes that are either phenotypically restricted or of more general occurrence. They include antibodies against a diverse number of blood group substances, which in most cases have been defined with respect to their structural requirements. Anti-Gal is the most abundant natural polyclonal antibody in human sera (including IgG, IgM, and IgA) [8]. Anti-Gal IgG comprises 1.0–2.4% of total serum IgG, whereas anti-Gal IgM comprises 3.9–8.0% of total serum IgM [9,10]. Human anti-pig natural antibodies are known to be highly specific for glycoproteins and glycolipids bearing terminal ␣-D-Galp-(1→3)-␤-D-Galp-based residues [11–13]. This moiety is part of the blood group B epitope [␣-D-Galp-(1→3)-␣-D-Fucp-(1→2)-␤D-Galp], but a subspectrum of ␣-D-Galp-(1→3)-␤-D-Galp-specific antibodies with low cross-reactivity has been shown to exist, explaining why the human population responds to ␣-D-Galp-(1→3)-␤-D-Galp irrespective of the B phenotype [14]. Most interestingly, man and Old World monkeys possess the natural antibodies but not New World monkeys (monkeys of Central and South America) and other mammalian species, in accordance with the presence of the ␣-D-Galp-(1→3)-␤-D-Galp epitope in the latter species. Not surprisingly, it does not cross-react with other known mammalian carbohydrate epitopes, or with proteins lacking the ␣-gal epitope. Anti-Gal does not bind to glycolipids that have terminal ␤-galactosyl, sialic acid, fucose, or N-acetyllactosamine carbohydrate units [15]. The highly restricted specificity of antiGal and its overabundant production in humans indicate that this natural antibody belongs to a different category from polyreactive natural antibodies, which react with various antigens and are present in small quantities in the circulatory system. Inhibitory effects using ␣-D-Galp-(1→3)-␤-D-Galp-based sugars are several orders higher than that using melibiose (␣-D-Galp-(1→6)-␤-D-Glcp) [16]. Recently, McKane and coworkers have demonstrated that this degree of specificity has an even higher level of sophistication [7]. Upon analysis of 77 individual sera with neoglycoprotein conjugated with disaccharide ␣-D-Galp-(1→3)-␤-D-Galp (1), trisaccharide ␣-D-Galp(1→3)-␤-D-Galp-(1→4)-␤-D-GlcpNAc (2) and pentasaccharide ␣-D-Galp-(1→3)-␤D-Galp-(1→4)-␤-D-GlcpNAc-(1→3)-␤-D-Galp-(1→4)-␤-D-Glcp (4) to probe the precise antigenic requirements for the binding of anti-Gal, it was found that 66% of

584

Andreana et al.

individuals have IgG that recognize all ␣-D-Galp-(1→3)-␤-D-Galp di- (1), tri- (2), and pentasaccharide (4) (D, T, and P, respectively), termed DTP phenotype (Fig. 1). The frequency of other phenotypes was -P, 13%; -TP, 12%; D-P, 8%; and DT-, 1%. The IgG subclasses found were IgG2 (95%), IgG3 (34%), IgG1 (31%), and IgG4 (17%). IgM in 91% of individuals recognized all three antigens. It is hypothesized that elevated levels of anti-Gal persist as a result of constant antigenic stimulation by ␣-gal-like epitopes found on the surface of bacteria constituting the normal intestinal flora [17]. Anti-Gal is also abundant in apes and Old World monkeys, which, like humans, lack ␣-gal epitopes. In contrast, New World monkeys and nonprimate mammals synthesize an abundance of ␣-gal epitopes on cell surfaces and thus do not produce anti-Gal antibodies. The enzyme responsible for the existence of ␣-gal is ␣1,3-galactosyltransferase (␣1,3-GT), which transfers the terminal galactose from the sugar donor, uridine diphosphate galactose (UDPGal), to the acceptor N-acetyllactosaminyl group [18,19]. This enzyme is inactivated in humans and Old World primates [20,21]. It was first purified from bovine thymus by use of affinity chromatography on UDP-Sepharose [22] and cloned in 1989 [23] to reveal an encoded 368 amino acid protein (that of the murine was simultaneously completed) [24]. The murine ␣1,3-GT cDNA was also used as a probe in the cloning of homologous sequences from marmoset (a New World monkey) [25] and pig [26,27]. Since that time, over 40 distinct galactosyltransferases have been cloned from humans alone. The remarkable similarity that exists between these enzymes from varied species extends into the amino acid sequence, where a high incidence of overlap occurs. Each enzyme is organized as a type II transmembrane protein, with an N-terminal end within the cytosol and the C-terminal end protruding into the lumen of the endoplasmic reticulum or the Golgi apparatus. C.

Use of ␣-gal Oligosaccharides in Xenotransplantation

The relevance of glycoproteins extends into the fundamentals of many important biological processes including fertilization, immune defense, viral replication, parasitic infection, cell growth, cell–cell adhesion, degradation of blood clots, and inflammation [28–31]. Anti-Gal immunoadsorption and neutralization are two practical uses of ␣-gal oligosaccharides in xenotransplantation. For immunoadsorption, a potential recipient is depleted of anti-Gal by passing the blood through an affinity column containing ␣-gal epitopes [32,33]. A number of pig-to-primate organ xenotransplantations following anti-Gal immunoadsorption were reported. This procedure was shown to extend the survival of pig organs from 1 to 23 days. Recently it was discovered that antibody heterogeneity made the depletion of all anti-Gal in human serum difficult [7]. McKane and coworkers found that immunoadsorbents derived from pentasaccharide 4 or trisaccharide 3 were effective for IgG removal in all individuals. For IgM, no single immunoadsorbent achieved complete removal. For example, pentasaccharide 4 immunoadsorbent removes anti-Gal IgM in only 30% of individuals. Interestingly, 3-deoxy-trisaccharide ␣-D-Galp-(1→3)-␤-D-Galp-(1→4)␤-D-[3deoxy]GlcNAcp and 6-deoxy-trisaccharide ␣-D-Galp-(1→3)-␤-D-Galp(1→4)-␤-D-[6deoxy]GlcNAcp were the best single immunoadsorbents for IgM removal. The two deoxy derivatives were completely successful in 73% of individuals. The polymorphic nature of the anti-Gal repertoire poses a considerable challenge when one is trying to develop an efficient antibody removal system. So far the best

The ␣-gal Epitope

Figure 1

Xenotransplantation hindered by ␣-gal epitopes (1–4 ) and the anti-Gal antibody.

585

586

Andreana et al.

system is the combination of disaccharide 1 and trisaccharide 3 agarose, which appeared to efficiently remove anti-Gal IgG and IgM in all individuals [7]. For anti-Gal neutralization, synthetic ␣-gal epitopes are infused into the recipient’s circulation so that the anti-Gal antibody in the recipient will be neutralized by the epitopes and no longer free to attack the transplanted pig organ. Owing to insufficient amounts of ␣-gal oligosaccharide, experiments utilizing the intravenous infusion of synthetic ␣-gal oligosaccharides are extremely limited, with a total of three experimental studies performed to date [34–36]. Lin, Cooper, and coworkers [37,38] reported the pharmacokinetic parameters of the intravenous administration of tri- and pentasaccharide (2 and 4) in the inhibition of hyperacute rejection of pig hearts transplanted into baboons. The circulatory pharmacokinetics of 2 and 4 were typical of small water-soluble molecules. Their serum half-lives of 45–50 min indicated rapid renal clearance. In the period during which blood oligosaccharide concentration was above 1 mM (initially maintained at 3–12 mM for more than 4 h), the serum cytotoxic activity against porcine cells was completely abolished. Hyperacute rejection of the xenograft was inhibited during the infusion. This work pointed out several obstacles that need to be overcome before this approach becomes practical. First, large-scale production of ␣-gal oligosaccharides is needed in support of further biological testing. Secondly, the avidity of ␣-gal oligosaccharides to anti-Gal antibodies is not high (normally Kd is in the millimolar range). Third, the ␣-gal oligosaccharides are cleared from the blood too rapidly. Fourth, anti-Gal antibodies have some heterogeneity, which leads to inefficient binding. Aside from anti-Gal immunoadsorption and neutralization, several other approaches are being investigated to abate hyperacute rejection for prolonged survival of the xenogenic organ. These include the depletion or inhibition of the complementmediated lysis in the recipient [39,40], genetic engineering of a donor pig such as ␣-gal ‘‘knockout’’ [41], expression of human complement-inhibiting proteins [42], and expression of differentiated carbohydrate structures to mask the ␣-gal epitopes [41]. In the process of coping with finely detailed medical situations, the final solutions to the immunological barriers in xenotransplantation may be expressed as a combination of some or all these approaches. In addition to the applications for xenotransplantation, ␣-gal–antigen conjugates were shown to enhance antigen presentation of influenza virus hemagglutinin by the natural human anti-Gal antibody. With this discovery, it is conceivable that certain ␣-gal–antigen conjugates can be used as vaccines [43,44] for the treatment of various viral infections. Moreover, since ␣-gal receptors express on the surface of B cells that generate anti-Gal, ␣-gal–immunotoxin (e.g., ␣-gal–ricin conjugate) is being tested to specifically target and lyse such B cells [45]. This may provide another solution for prolonged survival of the xenogenic organ against hyperacute rejection.

III.

DEVELOPMENT OF SYNTHETIC STRATEGIES FOR ␣-GAL EPITOPES

A.

Biosynthetic Approach: Enzymatic Synthesis of ␣-gal Epitopes

Issues of chemoselectivity and regioselectivity pose an enormous strain on the modern-day synthetic organic chemist. Most natural products are enantiomerically or

The ␣-gal Epitope

587

diastereomerically pure, meaning that their conformation plays a specific role in the recognition of proteins for inhibition, suppression, or activation. Nature has developed an ingenious system to combat these problems and has allowed for an unsurpassable specificity in the biological system through the creation of enzymes. Enzymes are highly specific both in binding chiral substrates and in catalyzing reactions. This stereospecificity arises because enzymes, by virtue of their inherent chirality (proteins consist of only L-amino acids), form asymmetrical active sites. The required synthetic methodology for studying the structural features necessary for anti-Gal binding, and for the subsequent preparation of analog structures is made available through initial synthetic studies targeting the ␣-gal structure itself. The overall scheme of synthesis must remain flexible enough to accommodate varying structural motifs to allow for the preparation of analog structures, afford suitable quantities of material for biological analysis, and incorporate a minimum number of synthetic operations. Traditional organic synthesis of oligosaccharides requires multiple protection and deprotection steps, including tedious separation procedures at each step. Glycosyltransferases play important roles in obtaining oligosaccharides, glycopeptides, and glycolipids under mild conditions. The high specificity in the formation of glycosides of glycotransferases makes them a viable strategic choice for the preparative synthesis of complicated oligosaccharides and glycoconjugates. Nevertheless, the limitations of quantity and purity of the glycosyltransferases available from natural sources have directed a great interest toward the cloning of the glycosyltransferase genes into convenient expression systems by recombinant DNA technology. Although several studies on cloning and characterization of ␣1,3-galactosyltransferase (␣1,3GT) have been reported, [25,26,46–49], no practical production of the recombinant enzyme has been accomplished. There has recently been developed an efficient chemoenzymatic approach based on the use of a recombinant ␣1,3-GT for the synthesis of xenoactive ␣-galactosyl epitopes. A truncated bovine ␣1,3-GT (80-368) was cloned into pET 15b vector and subsequently transformed into an E. coli BL21 strain (Fig. 2) [50]. This expression system efficiently produced the soluble recombinant enzyme on a large scale with high specific activity. Subsequent to the synthesis of galactosyl disaccharide acceptors, via chemical methods and enzymatic approaches involving glycosidase-catalyzed reactions, the recombinant enzyme was used to synthesize a variety of ␣1,3galactosylated trisaccharides on a preparative scale. UDP-Glc was used as the donor, along with commercially available UDP-Gal 4-epimerase, to avoid direct use of the costly UDP-Gal. UDP-Gal-4-epimerase (galE), from E. coli, is a dimer of identical subunits, with an overall molecular weight of 79 kDa, which plays a key role in the Leloir pathway for galactose metabolism. Members of a series of glycosylation acceptors were examined, including lactose 6, ␤-lactosylazide 8, [51] ␤-thiophenyl lactoside 9 [52], N-acetyllactosamine 5, ␤-allyl-N-acetyllactosamine 10, and lactosamine 7 (Fig. 3, p. 589). These dissaccharides served as acceptors to produce the trisaccharide ␣-gal epitopes and derivatives 11–16 in good yields. However, one of the main issues surrounding glycosyltransferase’s commercial viability is the limited dynamics with the type of reaction it is involved in. The high cost factor of the sugar nucleotides (donor) required for biocatalysis necessitates the use of UDP-glucose 4epimerase to alleviate some of the expense [53,54].

588

Andreana et al.

Figure 2

Genetic map of an ␣1,3-galactosyltransferase expression plasmid.

Genetic engineering has allowed for the generation of bifunctional proteins, which are formed from enzymes acting in sequence of cofactor regeneration. The construction of bifunctional enzymes consisting of a truncated bovine recombinant ␣1,3-GT (BGT) and an E. coli UDP-galactose 4-epimerase has made the ␣-gal epitope synthesis feasible by allowing for the conversion of an inexpensive sugar nucleotide (UDP-glucose) followed by glycosidation by ␣1,3-GT. The bifunctional enzyme, galE-␣1,3-GT, was constructed by in-frame fusion of the Escherichia coli gene galE to the 3⬘-terminus of bovine ␣1,3-GT gene within a high-expression plasmid. The incorporation of two enzymes into one plasmid vector reduces the amount of exhaustible materials required for the overexpression of particular proteins. This methodology allows for the reversal of genetic sequences producing two bifunctional proteins, termed galE-␣1,3-GT and ␣1,3-GT-galE (Fig. 4), capable of catalyzing the same reaction with similar kinetic rates [55]. Since various immobilization techniques are available for the use of enzymes, the bifunctional protein technology can be utilized in a one-pot, enzyme-immobilized reaction system, facilitating preparation protocols. Bifunctional enzymes are also thought to have a greater reaction rate. This increase in the overall kinetics of the reaction may occur as a result of the substrate proximity to the binding sites of both enzymes. This technology can be exploited in constructing various bifunctional enzyme systems where the number of genetic sequences incorporated into the plasmid may depend solely on the product desired. B.

CLONEZYME姠: Using a Glycosidase Library to Build Disaccharides

Over the past few years, glycosidases have been widely used for the preparation of a broad spectrum of glycoconjugates and derivatives employing either transglycosylation or reverse hydrolysis reactions [56–60]. To find biocatalysts with unprece-

The ␣-gal Epitope

Figure 3

Synthesis of ␣1,3-gal trisaccharides achieved by using ␣1,3-galactosyltransferase and UDP-galactose-4-

epimerase.

589

590

Figure 4

Andreana et al.

Bifunctional proteins ␣1,3-GT-galE and galE-␣1,3-GT.

dented properties from vast untapped resources of extremophiles, Diversa Inc. developed recombinant technologies and robotic screening approaches and ultimately produced a novel glycosidase library, namely, the CLONEZYME娃 library, which contains nine unique, thermostable glycosidases [61–63]. Glycosidases are attractive for the construction of disaccharides primarily because of their commercial availability, low cost, and high stability [64–70]. N-Acetyllactosamine (5) not only is one of the most fundamental oligosaccharide sequences in glycoproteins and glycolipids but also serves as an important building block for enzymatic synthesis of ␣-gal epitopes and other complex sequences. With O-nitrophenyl-␤-galactopyranoside (17) as the glycosylation donor and N-acetylglucosamine (18) as the acceptor (Fig. 5), the enzymatic reactions were carried out at 77⬚C; nine different CLONEZYME glycosidases, which all exhibited ␤-galactosidase activities, were used, as well as ␤galactosidases from other sources for comparison. The results indicated that three out of the nine thermophilic enzymes in the library, Gly001-06, -07, and -09, had superior activity in catalyzing the glycosylation. No other regio- or stereoisomers were detected from the reactions using these three enzymes, and the hydrolysis rates of the disaccharide were considerably lower. Some of the catalytic yields were as high as 60%. The same CLONEZYME library was examined for catalyzing the tandem transglycosylation of lactose (6) and glucosamine hydrochloride salt (19) to form lactosamine (7) [71]. Enzymes Gly-06 and Gly-09 exhibited excellent regio- and stereoselectivity in the formation of lactosamine as well as moderate yields of 19 and 23%, respectively. It was discovered that when glucosamine hydrochloride (19) was used as an acceptor in a similar transgalactosylation reaction, the product lactosamine 7 could be synthesized on a multigram scale (Fig. 6). One of the drawbacks associated with the glycosidase-catalyzed glycosylation reactions is the difficulty of separating the product from the reaction mixture, which may contain very similar compounds (molecular weight and corresponding linkages)

Figure 5

Synthesis of N-acetyllactosamine with CLONEZYME娃 glycosidase library.

The ␣-gal Epitope

Figure 6

Synthesis of lactosamine via glycosidase-catalyzed transgalactosylation reaction.

591

592

Andreana et al.

[70]. Various separation methodologies, such as size-exclusion chromatography (SEC), ion exchange chromatography (IEC), HPLC, and charcoal adsorption, have been evaluated for isolation efficiency [53,72–77]. SEC, which is conceivably the most used method for separating enzyme-synthesized oligosaccharides, is limited to compounds that have significant molecular weight differences. In contrast, IEC is frequently used to isolate biomolecules such as nucleic acids and proteins. The use of IEC in carbohydrate chemistry was reported for exploiting the weakly acidic character of polyhydroxyl groups of ‘‘neutral’’ sugars [78]. This large-scale synthesis was facilitated through the development of a resin column for cation (for purification) and anion (for neutralization) exchange. When positively charged amino sugars at neutral pH were used, separation was accomplished by means of a cation exchange resin with dilute HCl as eluent. Effective separation of the charged species from the noncharged species allowed for purification/separation from the residual by-products. The versatility of the cation and anion exchange resin column extends into the purification of a mono- and disaccharide mixture by adjusting the concentration of the eluent HCl (Fig. 7). C.

In Situ Cofactor Regeneration in the Synthesis of an ␣-gal Pentasaccharide

Full-length bovine ␣1,3-GT is a type II membrane protein with a short N-terminal cytosolic domain, a membrane-spanning region, a stem, and a C-terminal catalytic region [46]. This recombinant technology has been successfully used in the synthesis of a variety of ␣-gal epitope derivatives as well as in an in situ cofactor regeneration system for the synthesis of ␣-D-Galp-(1→3)-␤-D-Galp-(1→4)-␤-D-GlcNAcp-(1→3)␤-D-Galp-(1→4)-␤-D-GlcN3p pentasaccharide [50]. One of the most elegant yet simplistic ways to synthesize long-chain carbohydrates is via enzymes. To synthesize

Figure 7 resins.

Separation of lactosamine from an enzymatic reaction mixture with ion exchange

The ␣-gal Epitope

593

the ␣-D-Galp-(1→3)-␤-D-Galp-(1→4)-␤-D-GlcNAcp-(1→3)-␤-D-Galp-(1→4)-␤-DGlcp-R pentasaccharide (22), a one-pot enzymatic synthesis was designed that included two sequential enzymatic glycosylations using ␣1,3-galactosyltransferase and ␤1,4-galactosyltransferase that share a common UDP-galactose donor. In situ cofactor regeneration cycles of the donor were applied to avoid stoichiometric use of the expensive nucleotide sugar donor and product inhibition of the transferases. The starting trisaccharide was prepared through an established stereospecific glycosylation [79]. It is noteworthy that almost all the starting trisaccharide 20 was consumed in this high-yielding and efficient one-pot synthesis. The corresponding intermediate 21 and product 22 were obtained in 53 and 35% yield, respectively (Fig. 8). D.

A Unique Chemoenzymatic Synthesis of ␣-gal Epitope

Amino sugars (e.g., glucosamine, galactosamine, mannosamine) are widely distributed in living organisms, where they constitute building blocks of glycoconjugates such as glycopeptides and glycolipids [80,81]. They are found in milk, in blood group substances, and in lipopolysaccharide antigens, where they serve as part of the cell surface antigenic determinants (epitopes) [1,32,82–85]. The chemical synthesis of oligosaccharides containing amino sugar moieties (amino oligosaccharides) relies on glycosylation employing the oxazoline method or phthalimido protecting scheme [86]. Enzyme-catalyzed synthesis of oligosaccharides has evolved into a powerful shortcut for those chemical strategies [53,72–74]. Thus we have focused our attention on the chemoenzymatic synthesis of ␣-gal analogs containing versatile handles, such as free amino (NH2) groups, which can be used for further synthetic manipulation to generate carbohydrate diversity [50]. These derivatives would present the possibility of identifying unnatural ligands with enhanced binding affinity toward anti-Gal antibodies. The synthesis of the intermediate 24 was accomplished by both enzymatic and chemical methods. The azido group was purposefully introduced for its synthetic flexibility in the solid phase synthesis of glycopeptides, glycopolymers, and glycodendrimers [87,88]. Another benefit for using the azido functionality is the ease with which functional group interconversion can allow for a variety of useful glycosylation functionalities such as glycosyl fluorides for orthogonal oligosaccharide synthesis. Compound 23 did not undergo glycosylation with UDP-Gal and ␤1,4-galactosyltransferase; however, it still served as a substrate for the thermophilic enzyme Gly001-09 to yield the important intermediate amino tetrasaccharide 24 in 11% after purification with IEC (Fig. 9). In comparison, tetrasaccharide 27 was also prepared on the basis of chemical transformation, which could be thought of as a precursor of 24. The various chemical steps introduced led to the disaccharide acceptor 26, which was utilized in conjunction with donor 25 to form the important tetrasaccharide via a Schmidt-type glycosylation procedure (Fig. 10) [89]. The relevance of this methodology allows for a chemoenzymatic approach to the synthesis of an important intermediate. Bifunctional protein (galE-␣1,3-GT) containing both uridine-5⬘-diphospho-galactose 4-epimerase (galE) and ␣1,3-galactosyltransferase (␣1,3-GT) was used as a catalyst with UDP-glucose (UDP-Glc) as a glycosylation donor for the enzymatic approach to pentasaccharide 28 (Fig. 11). The important characteristic to 28 is the unprotected amino group, which was synthesized in an attempt to determine the inhibition activities toward human anti-Gal IgG. Bifunctional protein galE-␣1,3-GT

594

Figure 8 Synthesis of ␣-D-Galp-(1-3)-␤-D-Galp–(1-4)-␤-D-GlcpNAc–(1-3)-␤-D-Galp–(1-4)-␤-D-GlcpN3 pentasaccharide 22 with in situ cofactor regeneration. Andreana et al.

The ␣-gal Epitope

Figure 9

Utilizing CLONEZYME library for the enzymatic synthesis of intermediate 24.

595

596

Andreana et al.

Figure 10

Schmidt-type glycosylation employed in the chemical preparation of interme-

diate 24.

was also used to synthesis compound 13 from the building block 7 of CLONEZYME technology. The results were compared with their parent N-acetylated carbohydrates (Fig. 12) in order to realize the effects, if any, of binding affinity toward the anti-Gal antibody. The results indicated that the unprotected amino group did not significantly change the binding ability of ␣-gal epitopes to the antibody, leading to the conclusion that the carbohydrate moiety may attach to the Fab portion of anti-Gal in such an organized, structured manner that there are no bonding interactions possible with the polar amino groups. E.

Large-Scale Synthesis of ␣-gal Trisaccharide Derivative

Although ␣-D-Galp-(1→3)-␤-D-Galp-(1→4)-␤-D-Glcp and ␣-D-Galp-(1→3)-␤-DGalp-(1→4)-␤-D-GlcNAcp sequences had been synthesized chemically [90–94], these synthetic methods failed to produce a large quantity of oligosaccharides effectively. In addition to our successful efforts in the enzymatic synthesis of several ␣gal oligosaccharides using a recombinant ␣1,3-GT, we have developed a practical and efficient route for chemical synthesis of derivatives of ␣-gal trisaccharide 2 on a 50 g scale. Synthesis of ␣-gal epitope derivatives 35 and 37 was achieved by utilizing a stereoselective and high-yielding glycosylation method, with the acceptor prepared through an efficient dibutylin oxide mediated regioselective protecting scheme [95]. In preparation of the glycosylation acceptor 33 (Fig. 13), lactosyl bromide 29 was converted to azide 30 [51] and then deacetylated to generate lactosyl azide 8. The regioselective monoalkylation of the C3 hydroxy group of galactose with p-methoxybenzyl chloride (MPMCl) was achieved through a one-pot reaction

The ␣-gal Epitope

Figure 11

Synthesis of ␣-gal tri- and pentasaccharide epitopes with the bifunctional protein galE-␣1,3-GT. 597

598

Andreana et al.

Figure 12 Comparing inhibitory activity of N-acetylated ␣-gal epitopes and those bearing a free amino group.

sequence involving a dibutylstannylene acetal intermediate. The selectively protected azide 31 was peracetylated to give compound 32, which was selectively deprotected at C3-OH through oxidative cleavage of the MPM ether with cerium(IV) ammonium nitrate (CAN) [96] to yield acceptor building block 33 (Fig. 13). Large scale (50 g) glycosylation between perbenzylated phenyl thiogalactoside donor 34 [97] and acceptor 33 (Fig. 14) was carried out under activation of N-iodosuccinimide/triflic acid

Figure 13

Synthesis of glycosylation acceptor 33.

The ␣-gal Epitope

Figure 14

599

Synthesis of trisaccharide derivatives 35 and 37.

[98–100] to form the protected trisaccharide 35 in 94% yield and with high level of stereoselectivity (␣/␤:>20:1). Trisaccharide 35 serves as an important intermediate for the synthesis of ␣-gal-containing conjugates and multivalent ␣-gal epitope clusters. For example, as shown in Figure 14, trisaccharide 35 was readily converted to N-acetyl derivative 37 for conformational analysis by using the following sequence: selective hydrogenolysis of azido group with Adam’s catalyst (PtO2) [101] followed by N-acetylation gave compound 36, which was deacetylated and debenzylated to form N-linked ␣-gal trisaccharide 37. The overall yield of this synthesis (from 29 to 37) was 41%.

IV.

CONFORMATIONAL ANALYSIS OF AN ␣-GAL TRISACCHARIDE

The ␣(1→3) linkage at the nonreducing end of ␣-gal epitopes is the major structural motif that triggers the hyperacute rejection through ␣-gal–anti-Gal interaction upon xenotransplantation. The remaining parts of the carbohydrate sequences at the reducing end are considered only to enhance the binding toward antibodies [102]. Hence, determination of the solution conformations of ␣-gal epitope, especially the ␣(1→3) linkage, provides the structural basis for understanding antibody recognition. Very little is known about the solution conformations of ␣-gal epitopes, although discrete knowledge can be obtained from several reports on NMR investigation and molecular modeling of ␣-D-Galp-(1→3)-␤-D-Galp disaccharide [103–106] and conformational study of isoglobotriaosylceramide (␣-D-Galp-(1→3)-␤-D-Galp-(1→4)-␤D-Glcpceramide) in nonaqueous media [107,108]. Recently, a variety of NMR techniques in combination with molecular modeling were used in the conformational analysis of 37, an N-linked ␣-D-Galp-(1→3)-␤-D-Galp-(1→4)-␤-D-Glcp derivative, in aqueous solution [109–117]. Complete 1H and 13C NMR spectra were first as-

600

Andreana et al.

signed, followed by the determination of each monosaccharide residue and hydroxymethyl exocyclic group conformations. Three-bond heteronuclear long-range coupling constants 3JCH from inter- and intraglycosidic residues were measured to determine crucial dihedral angles of ␾/␺. The experimental data were compared with theoretical calculations. The flexibility and dynamics of the ␣(1→3) linkage were evaluated by temperature-dependent interglycosidic long-range coupling constants 3 JCH obtained from NMR measurements and molecular modeling. The relevance to this conformational information may provide a structural base for the design of ␣gal analogs and mimetics useful in xenotransplantation research and pharmaceutical development. A.

Proton and Carbon-13 Resonance Assignments

The proton chemical shifts and coupling constants of the ␣-gal and ␤-glc spin systems of trisaccharide 37 were established by using DQF-COSY (double-quantumfiltered correlational spectroscopy) [118], TOCSY (total correlational spectroscopy) [119], and 1D TOCSY [120] spectra. The assignment of H-5⬘ of ␤-gal was obtained from difference double-resonance (DDS) spectrum, upon homodecoupling irradiation of the H-4⬘. The proton coupling connectivity between H-5⬘, H-6⬘, and H-6b⬘ was established by observation of cross-peaks in the DQF-COSY spectrum. The assignments of all carbon signals of 37 were confirmed by DEPT (distortionless enhancement by polarization transfer), INAPT (insensitive nuclei assigned by polarization transfer) [121,122], and HMQC (heteronuclear multiple quantum correlation) experiments [123–125]. By comparing complete data of 3JHH scalar coupling constants with the theoretical vicinal proton–proton coupling constants predicted by higher level architecture (HLA) empirical equations [126,127], it was confirmed that ␣-gal, ␤-gal, and ␤-glc residues were all in the 4C1 chair conformations. B.

Conformational Distribution About the Exocyclic C5—C6 Bond

Rotational populations of primary hydroxyl groups on each residue of trisaccharide 37 were studied by analyzing coupling constants 3J5,6R and 3J5,6S that are correlated with H5-C5-C6-H6 dihedral angles through a modified Karplus equation [126,127]. The results indicated that the primary hydroxymethyl group of ␣-gal, ␤-gal, and ␤glc residue adopted predominantly tg, gt, and gg orientation, respectively, in trisaccharide 37. The tg orientation for ␤-glc was rarely occupied, but it was the most populated conformer for the terminal ␣-gal. For ␤-gal, an intramolecular hydrogenbonding formation between OH-6 and the ring oxygen presumably contributed to the predominance of gt orientation. Such conformational preferences for glucose and galactose residues also had been observed in earlier studies [128]. C.

Glycosidic Linkage Conformations and Rigidity

A selective 2D J-␦ INEPT experiment [129–131] was used for measuring the threebond, long-range proton–carbon coupling constants, 3JCH [132]. This approach allowed unequivocal detection of doublets for each carbon three-bond, long-range coupling with the selected proton. The coupling constants of JH1⬙-Cl⬙-O-C3⬘ and JH1⬘-Cl⬘-O-C4 were readily determined through the polarization transfer from the well-isolated an-

The ␣-gal Epitope

601

omeric protons (Fig. 15). The measuring of the coupling constant JH3⬘-C3-O-Cl⬙ required selective irradiation of H-3⬘ of ␤-gal, which was otherwise overlapped with other protons. The 3JCH coupling constants and the corresponding interglycosidic torsional angles were calculated by using the equation [133] 3JCH = 5.7 cos2␺ ⫺ 0.6 cos ␺ ⫹ 0.5. The results showed the crucial dihedral angles of ␣(1→3) linkage ␾/␺ = ⫺43⬚/ ⫺40⬚ and dihedral angle for ␤(1→4) ␾ ⫹ 38⬚. The literal interpretation would have the trisaccharide 37 assuming a U-shaped conformation in solution state. In general, the rigidity of the glycosidic linkage can be deduced from the temperature dependence of the long-range 3JCH coupling constants [134]. The interglycosidic coupling constants 3JCH for ␣(1→3) linkage were measured at different temperatures by performing selective 2D J-␦ INEPT experiments. The intraresidue 3 JCH coupling constants for ␣-gal were also obtained in the same 2D experiments at different temperatures. It was found that the NMR-derived average dihedral angles were almost independent of the temperature changes, which suggests that the ␣(1→3) linkage is relatively rigid. Interresidue nuclear Overhauser effects (NOEs) were also measured to substantiate the structural information obtained from long-range coupling constant measurement (Fig. 16).

Figure 15 charide 37.

Selective 2D J-␦ INEPT spectra and cross sections of ␣(1-3) linkage of trisac-

602

Andreana et al.

Figure 16

D.

Observed nuclear Overhauser effects (NOEs) in U-shaped trisaccharide 37.

Theoretical Calculations

Because oligosaccharides are so flexible, structural and dynamic properties are difficult to define unambiguously by evaluating the ensemble-averaged NMR data alone. Computational protocols [135,136], therefore, have been developed and employed to calculate experimentally accessible NMR parameters as well as to simulate internal motions that can truly reproduce the corresponding ensemble-average properties. The GEGOP (Geometry of Glycoproteins) program [137], which employs a modified version of the hard sphere exo anomeric (HSEA) force field [109] and a Metropolis Monte Carlo (MMC) [138] stochastic simulation procedure, was used for conformational calculation of trisaccharide 37. The conformations of the oligosaccharide were defined by dihedral angles of glycosidic linkage (␾ and ␺), the exocyclic H5-C5-C6-H6 torsional angle (␻), and the glycosidic bond angle (␶). Global minimum conformation of ␣-D-Galp-(1→3)␤-D-Galp-(1→4)-␤-D-Glcp trisaccharide 2 was obtained by performing the potential energy minimization with GEGOP (Fig. 17). The global minimum conformation (E = ⫺3.24631 kcal/mol) was located around ␾/␺ = ⫺50⬚/⫺34⬚ and 52⬚/4⬚ for the corresponding ␣(1→3) and ␤(1→4) linkages. Grid search of the dihedral angles ␾ and ␺ around this linkage versus energy was also performed to locate all the local minima in the conformational space. Only one global minimum was located for ␣(1→3) linkage, along with an additional local minimum. V.

␣-GAL-CONTAINING NEOGLYCOPOLYMERS FOR ENHANCED BINDING

A.

Receptor-Specific Glycopolymers for Bacteria Targeting

Antibody-based immunotherapy is often limited by a weak immunogenetic response. The complexity associated with the development of humanized antibodies has hindered progress for immunosuppressive therapy. Recently the direct use of human natural antibodies such as anti-Gal for the specific deletion of targeted bacterial cells has become very attractive [139,140]. The feasibility of using antibody targeting

The ␣-gal Epitope

603

Figure 17 Energy grid map of ␣(1-3) linkage in trisaccharide 2. Isoenergy contours are scaled at 1 kcal/mol intervals up to 6 kcal/mol above the global minimum energy.

followed by mediated immune response against bacterial cells was established by Bertozzi et al. [141,142]. Numerous bacterial strains produce surface lectins, which are in the form of fimbriae (or pili), which are filamentous appendages of protein subunits. They are found to be responsible for cell–cell recognition and infectious diseases. However, many bacteria undergo genetic changes to eschew the host immune system. In spite of these changes, the binding specificity of bacterial cell surface lectins is conserved [143]. Lectins are a group of nonenzymatic proteins with high affinities for mono- and oligosaccharides. They have been found in animals, plants, microorganisms, and viruses; in many cases, their sugar specificity has been determined [144]. Among the best-characterized fimbrial lectins are the type 1 fimbrial lectins, which are specific for ␣-mannosides [145,146]. The binding affinity of the mannose receptor to a mannose ligand falls in the millimolar range. Only recently has the literature suggested that the avidity of carbohydrate ligands binding to proteins can be increased dramatically when carbohydrate ligands are organized as multivalent clusters [88,147,148]. The mannose-containing glycopolymers and glycodendrimers have been shown to have ideal inhibitory effects of binding on yeast mannan to concanavalin A and pea lectins [149]. The results of the experiment conclusively illustrated that the well-defined glycodendrimers were not as effective as the glycopolymers. In an attempt to capitalize on these preliminary studies, new types of ␣-gal epitope and mannose-containing glycopolymer were synthesized for use in detailed studies on inhibitory affects toward anti-Gal antibody (Fig. 18). To synthesize the complete glycopolymer 52A-B, a convergent-type synthetic scheme was designed. For the synthesis of the ␤-lactosyl acrylimide portion, a com-

604

Andreana et al.

Figure 18

Schematic representation of human immune defense against pathogen through a synthetic ␣-gal–ligand mediator.

mercially available 2-[2-(2-chloroethoxy)-ethoxy]ethanol 38 (Fig. 19) was chosen as a spacer building block to provide for the convergent polymer distal flexibility that would facilitate transfer efficiency in a proposed final enzymatic transfer modification of the glycoconjugate. Complete transformation of the chloride 38 to the corresponding azide 39 was accomplished via an SN2 type addition in DMF at 80⬚C according to a well-known procedure [150]. A Ko¨enigs–Knorr glycosylation method employing ␣-lactosyl bromide heptaacetate 29 with azide 39 gave compound 40 in an 82% isolated yield. As a result of neighboring group participation, from the 2-acetate protecting group, the desired ␤ configuration was established and confirmed by the coupling constant of the anomeric proton (J12 = 7.8 Hz). The corresponding intermediate product 42 was accomplished by a simple quantitative yielding Zemple´n method to give 41, followed by azide reduction via hydrogenation and treatment with acryloyl chloride to give 42 in 71% yield. ␣-Mannosyl acrylamide 45 (Fig. 20) was synthesized according to a similar approach to that described for 42. The glycosylation reaction provided compound 43 in moderate yield (49%). The desired ␣ configuration was determined by the anomeric coupling constant (J12 = 1.5 Hz) along with a characteristic feature that there were no carbon chemical shifts around 76 ppm. (In peracetylated mannoside, the C5 chemical shift is 76 ppm for the ␤ anomer, while it exists at 70 ppm for the ␣ anomer.) ␣-Mannosyl bromide tetraacetate was more reactive than ␣-lactosyl bromide heptaacetate, as denoted by the excess of by-products and the diminished yield. In an attempt to further derivatize the projected glycopolymer, an enzymatic synthesis of an ␣-gal disaccharide containing a monomer precursor was designed. The acrylomide disaccharide 48 was accomplished via a one-pot enzymatic transglycosylation reaction catalyzed by ␣-galactosidase from green coffee beans [152]. Although the yield of this enzymatic reaction, like the conventional transglycosylation (Fig. 21), is relatively low, the concise synthetic approach is still more attractive than the typical protecting–deprotection chemical synthesis. Transformation of the nitro group of compound 46 to the N-acryloyl function was then effected by hydro-

The ␣-gal Epitope

Figure 19

605

Synthesis of the lactosyl intermediate 42.

genation, followed by the treatment with acryloyl chloride in methanol to give compound 48. Glycopolymers bearing ␣-gal epitopes were constructed through a well-known polymeric synthetic route. The method utilized epitope 48 in an attempt to copolymerize with mannose monomer precursor 45 and acrylamide in an established terpolymerization method (Fig. 22) [153]. Polymerization was initiated by ammonium persulfate and N,N,N⬘,N⬘-tetramethylethylenediamine in aqueous solution. The final polymer was obtained as a white fluffy material after separation by Bio-Gel P-2 permeation column and freeze-drying (76% yield). Proton NMR spectroscopy in D2O verified the incorporation of all three monomers in the final glycopolymer, 49. Once these glycopolymers had been well characterized, it was our intention to take advantage of the stereo- and regioselectivity of galactosyltransferases to transfer a galactose monomer from uridine 5⬘-diphosphate galactose donor (UDP-Gal) to a lactosyl acceptor in the glycopolymer-containing mannose sugar moieties. The relevance of this methodology would extend to illustrating the potential viability of selective modification of glycopolymers bearing different sugar moieties. It is important to note that there was precedent in the literature for enzymatic transformations of glycopolymers; however, the modification was accomplished with polymers bearing only a single type of carbohydrate ligand [154–156]. As anticipated, the enzymatic transformation generated the modified glycopolymers 52A-B (Fig. 23) in nearly quantitative yields (confirmed by the integration ratio of anomeric protons).

606

Synthesis of the mannosyl intermediate 45.

Andreana et al.

Figure 20

The ␣-gal Epitope

Figure 21 Polymerizable disaccharide 48 synthesized via a one-pot enzymatic transglycosylation catalyzed by ␣-Dgalactosidase from green coffee beans.

607

608

Synthesis of lactose/␣-mannose copolymer 49.

Andreana et al.

Figure 22

The ␣-gal Epitope 609

Figure 23

Synthesis of ␣-gal/␣-mannose copolymer 52A-B.

610

Andreana et al.

All the glycopolymers obtained from terpolymerization have a high molecular weight (MW > 870 kDa) with a mean hydrodynamic diameter of polymer chain in aqueous solution grater than 69 nm at ambient temperature. The choice of using aromatic 50 and nonaromatic 45 ␣-mannoside ligands for the glycopolymers was based on the work completed by Firon et al. [157,158]. They discovered that close to mannose binding sites, E. coli has a hydrophobic region that favors aromatic ␣-mannosides such as p-nitrophenyl-␣-D-mannopyranoside. They also noticed that Salmonella species possess a binding site smaller than that of E. coli and devoid of a hydrophobic region. To test the avidity of binding toward specified bacteria cells, a control (56, in comparison with the multivalent counterpart 52B) containing both an ␣-gal monomer moiety and a mannose unit was synthesized. The same Ko¨enigs–Knorr glycosylation method was utilized to prepare 53. It was sequentially deacetylated by the Zemple´n method followed by saponification to give water-soluble 54 with a free carboxylic group. The lactosyl derivative was coupled with p-aminophenyl-␣-D-mannopyranoside by using diphenylphosphoryl azide (DPPA) as a promoter, which was demonstrated to activate glycopeptide coupling in the presence of free hydroxyl groups [159]. The reaction was conducted in DMF at room temperature to give conjugate 55 in 66% yield. Enzymatic galactosylation of compound 55 by using ␣1,3-galactosyltransferase provided the ␣-gal epitope containing conjugate 56. The ␣-gal–mannose glycopolymers were tested for their ability to function as multivalent ligands for the binding of E. coli K-12 HB101 bacterial cells. This E. coli strain contains mannose binding sites located on the surface of bacteria, which can bind yeast (Saccharomyces cerevisiae), resulting in visible agglutination. To evaluate the activity of glycopolymers 49 and 52A-B, their inhibitory effects on agglutination were compared with several controls, including the monovalent methyl-␣D-mannopyranoside, p-nitrophenyl ␣-D-mannopyranoside, ␣-gal-mannose conjugate 56, and polyvalent mannose copolymers 52C-D, with striking results. These results indicated that the glycopolymers 49 and 52A-B were effective in preventing agglutination at concentrations as low as 100 ␮M. In comparison to methyl-␣-D-mannopyranoside, a dramatic increase in the inhibitory efficacy was observed with polyvalent mannose. However, p-nitrophenyl ␣-D-mannopyranoside was found to be the most potent inhibitor. This compound showed strong inhibition at concentrations as low as 90 ␮M. Monovalent conjugate 56 and polymer 52B, bearing a hydrophobic aromatic ring, had comparable inhibitory effects. The polyvalency was unexpectedly overshadowed by van der Waals forces resulting from the packing of the aromatic ring against a hydrophobic binding pocket on the surface of the bacteria. It is noteworthy that no inhibition was observed with p-nitrophenyl ␣-D-galactopyranoside. This indicated that the mannose binding sites were highly specific. The ␣-gal residue was not able to compete with mannose in binding to the bacterial mannose receptors. To assess the activity of ␣-gal–mannose glycoconjugates binding to the antiGal antibodies, a previously reported inhibition-type enzyme-linked immunosorbent assay (ELISA) was conducted [21]. Mouse laminin having ␣-gal epitopes were fixed on ELISA plate as solid phase antigens. Test glycoconjugates were then incubated with human anti-Gal antibodies on the ELISA plate. The plate was washed and incubated with horseradish peroxidase (HRP) conjugated anti-human IgG antibody.

The ␣-gal Epitope

611

After additional washing, a chromogenic compound was used to develop the color by an HRP reaction. Monovalent conjugate 56 (Fig. 24) and polymer 52B were tested with two controls: ␣-gal epitope ␣-D-Galp-(1→3)-␤-D-Galp-(1→4)-␤-D-Glcp and mannose copolymer 52D (Table 1). As illustrated in Table 2, conjugate 56 and polymer 52B bind effectively to human anti-Gal antibody. The polyvalency of the ␣-gal epitope on polymer 52B was clearly demonstrated by comparing the inhibition of compound 56 with polymer 52B at the same concentration of ␣-gal epitopes. Not surprisingly, polymer 52D, without having an ␣-gal epitope, did not bind to human anti-Gal antibodies. The results obtained from biological testing of anti-Gal antibodies in yeast mannan to E. coli K12 cells and human natural anti-Gal antibodies prove the efficacy of the constructed glycopolymers and glycoconjugates as potent inhibitors. Crossreactivities of mannose binding to anti-Gal antibodies and ␣-gal residue binding to bacteria were not observed in any of the experiments. These results demonstrate a potential for the viability of ligands to redirect human natural immunity against bacterial pathogens coated with xenoactive epitopes.

Figure 24

Synthesis of a monomeric ␣-gal/␣-mannose conjugate 56.

612

Andreana et al.

Table 1

Agglutination Inhibition Assays

Inhibitor

Inhibiting concentration (M)a

Methyl-␣-D-mannopyranoside p-Nitrophenyl ␣-D-mannopyranoside p-Nitrophenyl ␣-D-galactopyranoside Compound 56 Polymer 49 Polymer 52A Polymer 52B Polymer 52C Polymer 52D

1.5 9.0 >2.0 1.0 1.5 1.0 1.0 1.0 1.0

a

⫻ ⫻ ⫻ ⫻ ⫻ ⫻ ⫻ ⫻ ⫻

10⫺2 10⫺5 10⫺2 10⫺4 10⫺3 10⫺3 10⫺4 10⫺3 10⫺4

Molar concentrations of mannose unit were used in all polymers.

B.

Amplified Binding Differentiation of anti-Gal Isotypes by ␣-gal Polymers

The existence of differentiating anti-Gal antibody isotypes has allowed so-called isotype switching, a distinct process of recognition for antigen–antibody complexation [160]. Recent studies involving newborn baboons heterotropically transplanted with baby pig hearts have elucidated the isotype switch effect. The porcine hearts were rejected 4–6 days posttransplantation [161]. Six days postrejection of the graft, distinct anti-Gal IgM and IgG antibodies were detected. After an additional week, anti-Gal IgG activity increased, whereas that of the IgM isotype decreased. The findings suggest that in course of xenograft rejection by maternal anti-Gal IgG molecules, the immune system of the neonatal baboon responds to ␣-gal epitopes on the xenografts, first producing anti-Gal IgM and subsequently anti-Gal IgG antibody molecules.

Table 2

Binding of the Human Natural Polyclonal anti-Gal Antibody IgG to ␣-gal–Mannose Glycoconjugates as Measured by ELISA

Glycoconjugate Laminin ␣-D-Galp-(1→3)-␤-D-Galp-(1→4)-D-Glcp (0.1 mM) ␣-D-Galp-(1→3)-␤-D-Galp-(1→4)-D-Glcp (1.0 nM) Compound 56 (0.1 nM) Compound 56 (1.0 nM) Polymer 52Ba (0.1 mM) Polymer 52Bb (1.0 mM) Polymer 52Db (0.1 mM) Polymer 52Db (1.0 mM)

OD655 (% inhibition) 0.799 0.427 0.279 0.386 0.278 0.241 0.183 0.784 0.729

(0%) (47%) (65%) (52%) (65%) (70%) (77%) (2%) (9%)

The molar concentrations used in polymer 52B were ␣-gal trisaccharide concentrations. b The molar concentrations used in polymer 52D were mannose concentrations. a

The ␣-gal Epitope

613

In a continuation to determine the ␣-gal and anti-Gal binding affinity, neoglycopolymers of polyacrylamide backbone conjugated with varying densities of ␣-DGalp-(1→3)-␤-D-Galp-(1→4)-␤-D-Glcp trisaccharide epitopes (␣-gal epitopes) were designed and synthesized [162]. It is known that carbohydrates are typically expressed on the cell surface in clusters; thus their overall binding capacity with protein receptors (commonly with multiple binding sites) is enhanced over the affinity of individual monovalent ligands through cooperative multiple interactions. Therefore it is a logical choice to use multivalent ligands as potent synthetic inhibitors to effectively block the recognition process. Lee [163–165], Whitesides [147,166,167], Roy [168,169], and others [148,150] in recent years have demonstrated that the multivalent forms of carbohydrate ligands, either polymers or dendrimers, often have amplified inhibitory effects over their monovalent counterparts, although the levels of enhancement vary. Our group has been successful in illustrating that the overall avidity of certain ␣-gal-containing polymers toward anti-Gal antibodies is significantly enhanced by incorporating their ‘‘multivalent’’ or ‘‘polyvalent’’ effects. The synthesis of the ␣-gal-containing polymers was achieved by reacting preactivated poly [N-(acryloyloxy)succinimide] (pNAS) 58 with an ␣-gal trisaccharide derivative 59, followed by capping of the active esters with aqueous ammonia (Fig. 25) [166,167,170–173]. All ␣-gal polymers 60A–60F were prepared from a single batch of pNAS, which was obtained by polymerization of N-(acryloyloxy)succinimide (57). To determine the ‘‘parent’’ molecular weight of the polymer, gel filtration chromatography was preformed after the complete hydrolysis to poly(acrylic acid) sodium salt. The average molecular weight of the hydrolyzed polymer was 252 kDa, with a relatively narrow molecular weight distribution (Mw/Mn = 1.5). By varying the ratio of ␣-gal trisaccharide to active esters in pNAS, a series of polymers with different densities of ␣-gal was obtained. The ratios of ␣-gal unit to acrylamide unit obtained were determined by integrating the proton NMR peaks from the trisaccharide and acrylamide signals. To achieve polymerization of ␣-gal polymers, the important intermediate {N␣-D-Galp-(1→3)-␤-D-Galp-(1→4)-␤-D-Glucp-5-aminopentamide}, 59, was synthesized through a series of well-known reactions beginning with the glycosylation of a thioglycoside donor, 34, and the previously prepared acceptor, 33 [174]. Upon mild hydrogenation utilizing PtO2 in methanol for azido reduction and subsequent reaction with 5-chlorovaleryl chloride, compound 61 was obtained in 87% yield. Debenzylation followed by acetylation and subsequent azide addition to the anomeric valeryl chloride afforded compound 64, which was subjected to the Zemple´n method for deacetylation, yielding compound 65. The ␣-gal trisaccharide 59 was achieved by a very mild reduction procedure. Compound 63 (Fig. 26) was also obtained via an enzymatic glycosylation utilizing bifunctional protein technology, as mentioned earlier (Section III.D). In addition, an even more attractive synthesis of 59 [50,175–178] was achieved by using the bifunctional enzyme [179]. Reaction of disaccharide derivative 67 with UDP-glucose gave compound 59 (Fig. 27) in 52% yield. Polylactose 71 [180–182], with a lactose/acrylamide ratio of 1:1.25, was synthesized as a negative control for bioassays. The synthetic route is illustrated in Figure 28. Compound 68 was obtained in 79% yield by using the same transformation from 35 to 61 illustrated in Figure 25. Azide addition gave 69 in 84% yield.

614

Andreana et al.

Chemosynthesis of ␣-gal-derivatized carbohydrate polymer 60. Reagents: (a) NIS, TfOH, CH2Cl2, 4A MS, ⫺30⬚C, 90%; (b) PtO2, H2, MeOH; (c) 5-chlorovaleryl chloride, Et3N, CH2Cl2, 87% (b and c); (d) Pd/C, H2, MeOH; (e) Py/Ac2O, DMAP, 89%; (f) NaN3, DMF, 70⬚C, 90%; (g) NaOMe, MeOH; (h) PtO2/H2, H2O-MeOH, 93% (g and h); (i) AIBN, PhH, reflux; (j) DMF, rt, 24 h; 65⬚C, 6h; rt, 24 h then NH3 ⭈ H2O, rt, 24 h.

Figure 25

The fully deprotected lactose derivative 67 was then obtained from 69 in 91% yield after deacetylation and hydrogenation. Polylactose 71 with a lactose/acrylamide ratio of 1:2.5 was prepared by reaction of pNAS (58) and intermediate 67, utilizing the same methodology as for the preparation of polymer 60B. Later, ELISA was used in an attempt to quantify the binding affinities of the newly synthesized ␣-gal-containing carbohydrate polymers 60 to anti-Gal antibodies [183]. Purified human (male, blood type AB) anti-Gal antibody was the primary antibody and mouse laminin as a natural source of ␣-gal were employed in the protocol. The concentrations of ␣-gal polymers at 50% inhibition (IC50) of anti-Gal antibody binding to ␣-gal epitopes on mouse laminin were quantified, and the IC50 data were summarized (Table 3). All the IC50 data presented in the text are the net ␣-gal trisaccharide concentrations (micromolar) calculated from the degree of functionalization and polymerization of each polymer, for comparison with the corresponding ␣-gal monomer explicitly. Results indicated that polymers 60A–60D inhibit the antibody with a higher binding affinity than the corresponding ␣-gal trisaccharide ␣-D-Galp-(1→3)-␤-D-

The ␣-gal Epitope

Figure 26

Synthesis of 63 with galE-␣1,3-GT.

615

616

Figure 27

Alternative route to the synthesis of compound 59.

Andreana et al.

The ␣-gal Epitope

617

Figure 28 Synthesis of polylactose 71, with a lactose/acrylamide ratio of 1:2.5, as a control for bioassays. Reagents: (a) PtO2, H2, MeOH, then 2.5-chlorovaleryl chloride, Et3N, 79%; (b) NaN3, DMF, 70⬚C, 84%; (c) NaOMe, MeOH; (d) PtO2/H2, H2O-MeOH, 91% (two steps); (e) pNAS, DMF, rt, 24 h; 65⬚C, 6h; rt, 24 h, then NH3⭈H2O, 24 h.

Galp-(1→4)-␤-D-GlcNAc 11 in all cases. The efficacy can be illustrated by the IC50 result for polymer 60C of 5.6 nM, which is 5.0 ⫻ 104-fold better than its trisaccharide analog 11 in inhibiting anti-Gal IgA. The enhancement proved to be greater for antiGal IgM and IgA than for anti-Gal IgG. For example, the activity enhancement of polymer 60B was 246-, 1.0 ⫻ 104-, and 4.4 ⫻ 104-fold better toward anti-Gal IgG, IgA, and IgM, respectively. This observation is consistent with the increasing numbers of binding sites from IgG to IgA to IgM. In human serum, IgM exists as a pentamer with 10 equivalent binding sites in one molecule; 80% of IgA exists as a monomer with two binding sites and 20% as higher order oligomers with multiple binding sites; IgG exists as a monomer with two binding sites. Therefore, a ‘‘mul-

Table 3 Inhibition of ␣-gal Polymers and Monomer Against the Binding of Purified Human anti-Gal Antibody (male, blood type: AB) to Mouse Laminin IC50 (␮M)b Compound

Ratioa

IgG

Monomer 11 Polymer 60A Polymer 60B Polymer 60C Polymer 60D Polymer 60E Polymer 60F Polylactose 71

— 1:1.8 1:2.5 1:3.3 1:6.4 1:15 1:29 1:2.5

69 ⫾ 5 0.37 ⫾ 0.08 0.28⫾ 0.05 2.8 ⫾ 0.9 7.2 ⫾ 2.7 60 ⫾ 5 93 ⫾ 16 N/I

IgA 55 0.016 0.0053 0.0070 0.043 0.86

⫾ 16 ⫾ 0.006 ⫾ 0.0023 ⫾ 0.0032 ⫾ 0.017 ⫾ 0.28 — N/I

IgM 277 0.0068 0.0063 0.0056 0.027 165

⫾ 24 ⫾ 0.0016 ⫾ 0.0015 ⫾ 0.0014 ⫾ 0.018 ⫾ 13 — N/I

Ratio of the unit of ␣-gal epitope (or lactose) to that of acrylamide in polymer. The concentration of inhibitor at 50% inhibition of the binding between purified anti-Gal antibody (16 ␮g/mL) and mouse laminin in inhibition ELISA. Errors are reported as 2␴. N/I = no inhibition. a

b

618

Andreana et al.

tivalent’’ effect is certainly more pronounced for antibodies (IgM, IgA) with increased protein–carbohydrate interaction sites. The effectiveness of ␣-gal polymers varied not only with antibody isotype, but also with the density of ␣-gal epitopes conjugated to the polymer. There was a steady increase inhibition efficacy for IgG and IgA from polymer 60E to 60B with increased ␣-gal densities in the polymers as denoted in Table 3. The same trend was observed for IgM from polymer 60E to 60C. However, polymer 60A, with a higher density of ␣-gal residues, exhibited a lower inhibition than polymer 60B for IgA binding, and the same level as 60B for IgG and IgM. The decrease in activity was rationalized by the thermodynamics of polyvalent interactions in biological systems in accordance with work by Whitesides et al. [147,167]. Negative cooperative interactions and the enthalpically diminished binding can be caused by unfavorable spatial conformation of polymers at high ligand concentration. The enhancements of inhibition by introducing more ␣-gal epitopes in 60A were obviously overcome by the simultaneous formation of noncooperative spatial conformation of the polymer as well as steric hindrance. Similar to the positive cooperative multivalent interactions, this negative effect is prevalent in antibodies with increased binding sites, such as IgA and IgM. Research is in progress to try to elucidate the detailed physical mechanism of the enhancement. To assess the practical application of ␣-gal polymers, inhibition ELISA was used to test intact human serum (male, blood type AB) instead of purified anti-Gal antibody (Table 4). The results also indicated the activity enhancement of the ␣-gal polymers in comparison to the monomer 11. The same trends were also observed in interaction of ␣-gal polymers with different isotypes of the antibody and with the varied densities of the ␣-gal epitope conjugated to the polymer. Interestingly, the IC50 results observed with the purified antibodies were consistently lower than the IC50 results with human sera. One explanation is that the purified antibodies were obtained from affinity column immobilized with ␣-gal trisaccharide similar in structure to the epitope on the polymer. Therefore, subsets of antibodies selected during the purification would bind most tightly to the polymer. Since the purified antibodies

Table 4 Inhibition of ␣-gal Polymers and Monomer Against the Binding of Human Serum (male, blood type: AB) to Mouse Laminin IC50 (␮M)b Compound

Ratioa

IgG

IgA

Monomer 11 Polymer 60A Polymer 60B Polymer 60C Polymer 60D Polymer 60E Polylactose 71

— 1:1.8 1:2.5 1:3.3 1:6.4 1:15 1:2.5

74 ⫾ 11 12 ⫾ 3 9.4 ⫾ 0.9 18 ⫾ 5 52 ⫾ 8 >500 N/I

306 ⫾ 22 1.14 ⫾ 0.26 0.12 ⫾ 0.04 0.091 ⫾ 0.003 3.3 ⫾ 0.7 >1000 N/I

IgM 268 0.070 0.035 0.031 0.38 89

⫾ 31 ⫾ 0.001 ⫾ 0.002 ⫾ 0.009 ⫾ 0.02 ⫾ 27 N/I

Ratio of the unit of ␣-gal epitope (or lactose) to that of acrylamide in polymer. The concentration of inhibitor at 50% inhibition of the binding between human serum (10 ␮L) and mouse laminin in inhibition ELISA. Errors are reported as 2␴. N/I = no inhibition.

a

b

The ␣-gal Epitope

619

from the affinity column contained 89% of IgM, 50% of IgA, and 42% of IgG of corresponding anti-Gal in human sera [184,185], the increased proportion of IgM in the purified anti-Gal would contribute to the increased inhibitory effect of the ␣-gal polymers. In an attempt to mimic cell transplantation situations, a flow cytometry assay [186,187] was carried out in which the inhibition of synthetic polymers against the binding of human natural anti-Gal antibodies in intact serum to ␣-gal epitopes on pig kidney cells from the American Type Culture Collection (PK15, ATCC) was measured. As expected, results from flow cytometry analysis demonstrated that ␣gal polymers exhibited enhanced inhibition activities in comparison to the monomer 59. With IC50 results of approximately 5.2 and 2.2 ␮M, respectively, high levels of inhibition were achieved with polymers 60A and 60B against anti-Gal IgM binding to PK15 cells. In sharp contrast, the ␣-gal monomer 11 has no significant inhibitory effect up to 1 mM, whereas polylactose did not exhibit any inhibition up to 1 mM (Table 5). In all concentrations tested, in the range from 0.10 to 100 ␮M, no inhibition was observed with monovalent ␣-gal trisaccharide. Clearly, the enhancement of inhibitory activity of the polymer is concentration dependent. At net trisaccharide concentration of 0.1 ␮M, the enhancement of the polymer over the monomer was minimal, while at concentrations of 100 ␮M, the inhibition of anti-Gal binding to PK15 cells by the polymer jumped to 68%. Our results also demonstrated that IC50 results of the polymers against antibody binding to cells were much higher (orders of magnitude) than the values for binding to laminin. This may be ascribed to the lower density of epitopes on the cell surface [20,184,188–190]. Moreover, the concentration of human sera used in cell assay was 2.5 times higher than that in laminin assay, which would bind more ␣-gal epitopes on the polymer. Other factors such as non-␣-gal antigen–antibody binding may contribute to the discrepancy in tandem [191]. In summary, the synthetic ␣-gal-conjugated polymers significantly enhanced activities in inhibition of human anti-Gal antibody binding to mouse laminin glycoproteins and mammalian PK15 cells. The significance to these findings extends into the amplified binding differences among the three anti-Gal isotypes, which can be selectively inhibited as a means for deciphering the various isotype antibodies.

Table 5 Inhibition of ␣-gal Polymers and Monomer Against the Binding of Human Serum (male, blood type: AB) to Pig Kidney (PK15) Cells IC50 (␮M)b Compound

Ratioa

IgG

IgA

IgM

Monomer 11 Polymer 60A Polymer 60B Polymer 60C Polylactose 71

— 1:1.8 1:2.5 1:3.3 1:2.5

⬃1000 595 407 774 N/I

>1000 — 63 — N/I

>1000 5.2 2.2 >1000 N/I

Ratio of the unit of ␣-gal epitope (or lactose) to that of acrylamide in polymer. The concentration of inhibitor at 50% inhibition of the binding between human sera and pig kidney cells in flow cytometry assay. N/I = no inhibition.

a

b

620

Andreana et al.

The flow cytometry analysis proved that certain ␣-gal polymers were effective in inhibiting anti-Gal antibodies in human serum binding to pig kidney cells. Thus, such carbohydrate polymers can serve both as tools for evaluation of ␣-gal–anti-Gal interactions and as immunosuppressors in cell xenotransplantations.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33.

JL Platt. Nature 392:11–17, 1998. CG Groth, O Korsgren, A Tibell. Lancet 344:1402, 1994. AH Good, DKC Cooper, AJ Malcolm. Transplant Proc 24:559, 1992. U Galili. Sci Med Sept/Oct:28–37, 1998. B Benda, A Karlsson-Parra, A Ridderstad, O Korsgren. Transplantation 62:1207, 1996. A Karlsson-Para, A Ridderstad, AC Wallgren, E Moller, HG Ljunggren, O Korsgren. Transplantation 61:9:1313, 1996. W McKane, J Lee, R Preston, A Hacking, P Simpson, S Lynds, L Goldberg, T Cairns, D Taube. Transplantation 66:5:626–633, 1998. U Galili, L Wang, MZ Radic. J Immunol 155:1276–1285, 1995. U Galili. Springer Semin Immunopathol 15:155, 1993. MS Sandrin, HA Vaughan, PL Dabkowski, FC McKenzie. Proc Natl Acad Sci USA 90:11391, 1993. U Galili, EA Rachmilewitz, A Peley, I Flechner. J Exp Med 160:1519–1531, 1984. U Galili, BA Macher, J Buchler, SB Shohet. J Exp Med 162:573–582, 1985. U Galili, J Buchler, SB Sohet, BA Macher. J Exp Med 165:693–704, 1987. J Wieslander, O Mansson, E Kallin, A Gabrielli, H Nowack, R Timpl. Glycoconjugate J 7:85–100, 1990. U Galili, P Andrews. J Human Evol 29:433–442, 1995. HA Vaughen, KR Oldenburg, MA Gallop, JD Atkin, IFC McKenzie, MS Sandrin. Xenotransplantation 3:18–23, 1996. HC Krivan, GF Clark, DF Smith, DT Wilkins. Infect Immun 5:573–581, 1986. M Basu, S Basu. J Biol Chem 248:1700–1706, 1973. WM Blanken, DH Van den Eijnden. J Biol Chem 260:12927–12934, 1985. U Galili, SB Shohet, E Kobrin, CLM Stults, BA Macher. J Biol Chem 263:17755– 17762, 1988. A Thall, U Galili. Biochemistry 29:3959–3965, 1990. WM Blanken, DH Eijnden. J Biol Chem 260:12927–12934, 1985. DH Joziasse, JH Shaper, DH Van den Eijnden, AJ Van Tunen, NL Shaper. J Biol Chem 264:14290–14297, 1989. RD Larsen, VP Rajan, MM Ruff, J Kukowska-Lattallo, RD Cummings, JB Lowe. Proc Natl Acad Sci USA 86:8227–8231, 1989. TR Henion, BA Macher, F Anaraki, U Galili. Glycobiology 4:193–201, 1994. KM Strahan, F Gu, AF Preece, I Gustavsson, L. Andersson, K Gustafsson. Immunogenetics 41:101–105, 1995. PL Dabkowski, HA Vaughan, IFC McKenzie, HS Sandrin. Transplant Proc 25:2921, 1993. RA Dwek. Chem Rev 96:683–720, 1996. K Drickamer, ME Taylor. Annu Rev Cell Biol 237–264, 1993. TW Rademacher, RB Parekh, RA Dwek. Annu Rev Biochem 57:785–838, 1998. A Varki. Glycobiology 3:97–130, 1993. U Galili. Immunol Today 14:480, 1993. R Rieben, E Von Allmen, E Korchagina, U Nydegger, FA Neethling, M Kujundzic, E Koren, N Bovin, DKC Cooper. Xenotransplantation 2:98, 1995.

The ␣-gal Epitope 34. 35. 36. 37. 38. 39. 40. 41. 42.

43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56.

57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74.

621

JC Magee. J Clin Invest 96:2404–2412, 1995. KR McCurry. Nat Med 1:423–427, 1995. E Cozzi, DJG White. Nat Med 1:964–966, 1995. DKC Cooper. J Heart Transplant 7:238–246, 1988. SS Lin. Transplant Immunol 5:212–218, 1997. JR Leventhal. Transplantation 55:857–866, 1993. SK Pruitt. Transplantation 57:363–370, 1994. AD Thall, D Maly, JB Lowe. J Biol Chem 270:21437–21440, 1995. T Nagasaka, T Kobayashi, H Muramatsu, H Fujimoto, I Matsuo, K Ajisaka, K Kadomatsu, S Hayashi, I Yokoyama, A Hayakawa, T Muramatsu, H Takagi. Biochem Biophys Res Commun 232:731–736, 1997. U Galili, DC La Temple. Immunol Today 18:281–285, 1997. Y Takeuchi, CD Porter, KM Straham. Nature 379:85–88, 1996. RP Rother, SP Squinto. Cell 36:185–188, 1996. DH Joziasse, JH Shaper, DH Eijnden, AJV Tunen, NL Shaper. J Biol Chem 264:14290, 1989. DH Joziasse, NL Shaper, LS Salyer, DH Eijnden, AC Spoel, JH Shaper. Eur J Biochem 191:75, 1990. RD Larsen, VP Pajan, MM Ruff, J Kukowska-Latello, RD Cummings, JB Lowe. Proc Natl Acad Sci USA 86:8227, 1989. DH Joziasse, NL Shaper, D Kim, DH Eijnden, JH Shaper. J Biol Chem 267:5534, 1992. J Fang, J Li, X Chen, Y Zhang, J Wang, Z Guo, W Zhang, L Yu, K Brew, PG Wang. J Am Chem Soc 120:6635, 1998. FD Tropper, FO Andersson, S Braun, R Roy. Synthesis 618, 1992. R Ferrier, R Furneaux. Methods Carbohydr Chem 8:251, 1980. C-H Wong, SL Haynie, GM Whitesides. J Org Chem 118:1836–1840, 1996. C Auge´, C Mathieu, C Merienne. Carbohydr Res 151:147–156, 1986. X Chen, P Kowal, S Hamad, H Fan, PG Wang. Biotechnol Lett 21:1131–1135, 1999. KGI Nilsson. Synthesis with glycosidases. In Modern Methods in Carbohydrate Synthesis, SH Khan and RA O’Neill, eds. Harwood Academic: Amsterdam, 1996, pp 518– 547. HJM Gijsen, L Qiao, W Fitz, C-H Wong. Chem Rev 96:443, 1996. A Ferna´dez-Mayoralas. Top Curr Chem 186:1, 1997. T Murata, T Usui. Biosci Biotechnol Biochem 61:1059, 1997. DHG Crout, G Vic. Curr Opin Chem Biol 2:98, 1998. MB Brennan. Chem Eng News Oct 14, 1996, p 31. MWW Adams, RM Kelly. Chem Eng News December 18, 1995, p 32. DE Robertson, EJ Mathur, RV Swanson, BL Marrs, JM Short. SIMS News 46:3, 1996. KGI Nilsson. Trends Biotechnol 6:256, 1998. GI Co`te, BY Tao. Glycoconjugate J 7:145, 1990. S David, C Auge´, C Gautheron. Adv Carbohydr Chem Biochem 49:175, 1991. S David. Chemtracts-Org Chem 7:92, 1994. A Vetere, S Paoletti. Biochem Biophys Res Commun 219:6, 1996. S Takayama, M Shimazaki, L Qiao, C-H Wong. Bioorg Med Chem Lett 6:1123, 1996. T Kimura, S Takayama, H Huang, C-H Wong. Angew Chem Int Ed Engl 35:2348, 1996. J Fang, W Xie, J Li, PG Wang. Tetrahedron Lett 39:919, 1998. C-H Wong, RL Halcomb, Y Ichikawa, T Kajimoto. Angew Chem Int Ed Engl 34:412, 521, 1995. EL Toone, ES Simon, MD Bednarski, GM Whitesides. Tetrahedron 45:5365, 1989. S Takayama, C-H Wong. Bioorg Med Chem Lett 6:3, 1996.

622

Andreana et al.

75. 76.

KGI Nilsson, A Eliasson, U Lorek. Biotechnol Lett 17:717, 1995. A Iwanowitsch, H Friebolin, R Carganico, M Errenst, G Weissharr, H Brunner, H Mann. Biol Chem 379:737, 1998. E Samain, S Drouillard, A Heyraud, H Driguez, RA Geremia. Carbohydr Res 302:35, 1997. RD Rochlin, CAJ Pohl. J Liquid Chromatogr 6:1577, 1983. S Abbas, KL Matta. Carbohydr Res 123:53, 1983. A Varki. Glycobiology 3:97, 1993. T Ferzi, JC Solomon, CT Yuen, KCG Jeng, LG Frigeri, DK Hsu, FT Liu. Biochemistry 33:6342, 1994. DKC Cooper, E Koren, R Oriol. Immunol Rev 141:31, 1994. MS Sandrin, HA Vaughan, IFC McKenzie. Transplant Rev 8:134, 1994. BE Samuelsson, L Rydberg, ME Breimer, A Backer, M Gustavsson, J Holgersson, E Karlsson, AC Uyterwaal, T Cairns, K Welsh. Immunol Rev 141:151, 1994. Nature (special issue) January, 391:309, 1998. J Banoub, P Boullanger, D Lafont. Chem Rev 92:1167, 1992. H Kunz, B Dombo. Angew Chem Int Ed Engl 27:711, 1988. R Roy. Curr Opin Struct Biol 6:692, 1996. KK Sadozai, T Nukada, Y Ito, Y Nakahara, T Ogawa. Carbohydr Res 157:101, 1986. JC Jacquinet, D Duchet, ML Milat, PJ Sina¨y. J Chem Soc Perkin Trans I 326, 1981. K Koike, M Sugimoto, S Sato, Y Ito, Y Nakahara, T Ogawa. Carbohydr Res 163:189, 1987. Y Matsuzaki, Y Ito, Y Nakahara, T Ogawa. Tetrahedron Lett 34:1061, 1993. GV Reddy, RK Jain, BS Bhatti, KL Matta. Carbohydr Res 263:67, 1994. T Zhu, GJ Boons. J Chem Soc Perkin Trans I 857, 1998. S David, S Hanessian. Tetrahedron 41:643, 1985. R Johansson, BJ Samuelsson. J Chem Soc Perkin Trans I 2371, 1984. PJ Garegg, H Hultberg, C Lindberg. Carbohydr Res 83:157, 1980. P Konradsson, DR Mootoo, RE McDevitt, B Fraser-Reid. J Chem Soc Chem Commun 270, 1990. P Konradsson, UE Udodong, B Fraser-Reid. Tetrahedron Lett 31:4313, 1990. GH Veeneman, SH van Leeuwen, JH van Boom. Tetrahedron Lett 31:1331, 1990. H Garg, RW Jeanloz. Adv Carbohydr Chem Biochem 43:135, 1985. F Neethling, D Joziasse, N Bovin, DKC Cooper, R Oriol. Transplant Int 9:98, 1996. RU Lemieux, K Bock, LTJ Delbaere, S Koto, VS Rao. Can J Chem 58:631, 1980. K Bock, A Brignole, BWJ Sigurskjold. J Chem Soc Perkin Trans II 1711, 1986. CA Stortz, ASJ Cerezo. Carbohydr Chem 13:235, 1994. F Bizid, I Tvaroska. Chem Pap 49:202, 1995. L Poppe, J Dabrowski, CW Von der Lieth, K Koike, T Ogawa. Eur J Biochem 189: 313, 1990. J Dabrowski, K Trauner, K Koike, T Ogawa. Chem Phys Lipids 49:31, 1988. K Bock. Pure Appl Chem 55:605, 1983. DA Cumming, JP Carver. Biochemistry 26:6676, 1987. RU Lemieux. Chem Soc Rev 18:347, 1989. SW Homans. Progr NMR Spectrosc 22:55, 1990. B Meyer. Top Curr Chem 154:141, 1990. CA Bush, P Cagas. In Advances in Biophysical Chemistry, CA Bush. eds. JAI Press, Greenwich, CT, 1992, p 149. S Walker, J Murnick, D Kahne. J Am Chem Soc 115:7954, 1993. T Peters, BM Pinto. Curr Opin Struct Biol 6:710, 1996. A Poveda, J Jimenez-Barbero. Chem Soc Rev 27:133, 1998. U Piantinis, OW Sorensen, RR Ernst. J Am Chem Soc 104:6800, 1982.

77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118.

The ␣-gal Epitope 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161.

623

L Braunschweiler, RR Ernst. J Magn Reson 53:521, 1983. H Kessler, H Oschkinat, C Griesinger. J Magn Reson 70:106, 1986. AJ Bax. J Magn Reson 57:314, 1984. A Bax, JA Ferretti, N Nashed, DM Jerina. J Org Chem 50:3029, 1985. L Muller. J Am Chem Soc 101:4481, 1979. A Bax, S Subramanian. J Magn Reson 67:565, 1986. FJ Cassels, HM Fales, J London, RW Carlson, H van Halbeek. J Biol Chem 265: 14127, 1990. CAG Haasnoot, FAAM De Leeuw, C Altona. Tetrahedron 36:2783, 1980. C Altona, CAG Hasnoot. Magn Reson Chem 13:417, 1980. K Bock, JO Duus. J Carbohydr Chem 13:513, 1994. P Ladam, J Gharbi-Benarous, M Piotto, M Delaforge, JP Girault. Magn Reson Chem 32:1, 1994. H Hricovı´ni, I Tvaroska, D Uhrin, GY Batta. J Carbohydr Chem 8:389, 1989. T Jippo, O Kamo, K Nagayama. J Magn Reson 66:344, 1986. I Tvaroska, FR Taravel. Adv Carbohydr Chem Biochem 51:15, 1995. I Tvaroska, M Hricovini, E Petrakova. Carbohydr Res 189:359, 1989. M Hricovini, I Ivaroska, J Hirsch. Carbohydr Res 198:193, 1990. CW Von der Lieth, T Kozar, WE Hull. J Mol Struct 395:225, 1997. RJ Woods. Glycoconjugate J 15:209, 1998. R Stuike-Prill, B Meyer. Eur J Biochem 194:903, 1990. N Metropolis, AW Rosenbluth, MN Rosenbluth, AH Teller, E Teller. J Chem Phys 21: 1087, 1953. AR Lussow, R Buelow, P Pouletty. US patent 5, 714, 332, 740, 166, 11490, 1998. PG Wang. Proposal submitted to the US National Institutes of Health, February 1, 1998. CR Bertozzi, MD Bednarski. J Am Chem Soc 114:2242, 1992. CR Bertozzi, MD Bednarski. J Am Chem Soc 114:5543, 1992. M Virji, JE Heckels. Infect Immun 49:621, 1985. EE Simanek, GJ McGarvey, JA Jablonowski, C-H Wong. Chem Rev 98:833–862, 1998. N Sharon. FEBS Lett 217:145, 1987. I Ofek, RJ Doyle. In Bacterial Adhesion to Cells and Tissues. Chapman & Hall, New York, 1994, pp 94–135. M Mammen, SK Choi, GM Whitesides. Angew Chem Int Ed Engl 37:2754, 1998. LL Kiessling, NL Pohl. Chem Biol 3:71, 1996. YC Lee. In Neoglycoconjugates: Preparations and Applications, YC Lee, and RT Lee, eds. Academic Press: San Diego, 1994, pp 3–21. D Page´, D Zanini, R Roy. Bioorg Med Chem 4:1949, 1996. PH Amvan-Zollo, P Sina¨y. Carbohydr Res 150:199, 1986. KGI Nilsson. Carbohydr Res 167:95, 1987. R Roy, FD Tropper, A Romanowska. J Chem Soc Chem Commun 1611, 1992. SI Nishimura, K Yamada. J Am Chem Soc 119:10555, 1997. K Matsuoka, SI Nishimura. Macromolecules 28:2961, 1995. M Oubihi, K Kitajima, K Kobayashi, T Adachi, N Aoki, T Matsuda. Anal Biochem 257:169, 1998. N Firon, I Ofek, N Sharon. Carbohydr Res 120:235, 1983. N Firon, I Ofek, N Sharon. Infect Immun 43:1088, 1984. T Tsuda, SI Nishimura. J Chem Soc Chem Commun 2779, 1996. U Galili, OP Minanov, RE Michler, KR Stone. Xenotransplantation 4:127–131, 1997. OP Minanov, S Itescu, FA Neethling, AS Morgenthau, P Kwiatkowski, DCK Cooper, RE Michler. Transplantation 63:182–186, 1997.

624

Andreana et al.

162.

JQ Wang, X Chen, W Zhang, S Zacharek, Y Chen, PG Wang. J Am Chem Soc 121: 8174–8181, 1999. YC Lee, RR Townsend, MR Hardy, J Lo¨nngren, J Arnarp, M Haraldson, H Lo¨nn. J Biol Chem 258:199, 1983. TC Lee, RT Lee, K Rice, Y Ichikawa, TC Wong. Pure Appl Chem 63:499, 1991. YC Lee, RT Lee. Acc Chem Res 28:321, 1995. A Pollak, H Blumenfeld, M Wax, RL Baughn, GM Whitesides. J Am Chem Soc 102: 6324, 1980. M Mammen, G Dahmann, GM Whitesides. J Med Chem 38:4179, 1995. D Zanini, R Roy. In Carbohydrate Mimics: Concepts and Methods. Y Chapleur, ed. Verlag-Chemie: Weinheim, Germany, 1998, Ch 20, pp 385–415. R Roy. Top Curr Chem 187:241, 1997. NV Bovin. Glycoconjugate J 15:431, 1998. Y Nishida, H Dohi, H Uzawa, K Kobayashi. Tetrahedron Lett 39:8681, 1998. J Li, S Zacharek, X Chen, J Wang, W Zhang, A Janczuk, PG Wang. Bioorg Med Chem 7:1549–1588, 1999. NV Bovin, EY Korchagina, TV Zemlyanukhina, NE Byramova, OE Galanina, AE Zemlyakov, AE Ivanov, VP Zubov, LV Mochalova. Glycoconjugate J 10:142, 1993. W Zhang, JQ Wang, J Li, LP Yu, PG Wang. J Carbohydr Chem 18:1009–1017, 1999. J Fang, X Chen, W Zhang, J Wang, PR Andreana, PG Wang. J Org Chem 64:4089– 4094, 1999. G Vic, M Scigelova, JJ Hastings, OW Howarth, DHG Crout. J Chem Soc Chem Commun 1473, 1996. G Vic, CH Tran, M Scigelova, HG Crout. J Chem Soc Chem Commun 169, 1997. J Thiem, T Wiemann. Synthesis 141, 1992. X Chen, H Fan, Y Zhang, K Brew, PG Wang. In preparation. WKC Park, S Aravind, A Romanowska, J Renaud, R Roy. Methods Enzymol 247:325, 1994. S Aravind, WKC Park, S Brochu, R Roy. Tetrahedron Lett 35:(42), 7739, 1994. R Roy, FD Tropper, A Romanowska Bioconjugate Chem 3:256, 1992. U Galili, DC LaTemple, MZ Radic. Transplantation 65:(8), 1129, 1998. IM McMorrow, CA Comrack, DH Sachs, H DerSimonian. Transplantation 64:(3), 501, 1997. Y Xu, T Lorf, T Sablinski, P. Gianello, M Bailin, R Monroy, T Kozlowski, M Awwad, DKC Cooper, DH Sachs. Transplantation 65:(2), 172, 1998. MR Melamed, T Lindmo, and ML Mendelsohn, eds. Flow Cytometry and Sorting, 2nd ed., Wiley-Liss: New York, 1990. A Jacquemin-Sablon, ed: Flow Cytometry, New Development. Springer-Verlag: Berlin, 1993. R Timpl. Eur J Biochem 180:487, 1989. K Beck, I Hunter, J Engel. FASEB J 4:148, 1990. U Galili, F Anaraki, A Thall, C Hill-Black, M Radic. Blood 82:(8), 2485, 1993. DKC Cooper. Xenotransplantation 5:6, 1998.

163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191.

17 Microbial Glycosyltransferases Przemyslaw Kowal, Xi Chen, and Peng George Wang Wayne State University, Detroit, Michigan

I.

INTRODUCTION

Polysaccharides, besides nucleic acids and proteins, are the third class of macromolecules, yet they constitute the largest part of the planet’s biomass [1]. Unlike the linear nucleic acid and protein molecules, polysaccharides exhibit highly branched structures and a multiplicity of different, although related, types of linkage known as glycosidic bonds. Enzymes that catalyze the formation of these glycosidic bonds are referred to as glycosyltransferases (GTs). These enzymes use activated sugar molecules (donors) to attach saccharide residues to a variety of substrates (acceptors). The donor molecules most often are nucleotide diphospho sugars (e.g., UDP-Glc) but can also be sugar phosphates or disaccharides. The most common acceptor molecules are growing carbohydrate chains; however, a variety of other compounds such as lipids, proteins, or steroids can also serve this function [2]. Generally, glycosyltransferases are divided into two groups: processive and nonprocessive. Processive group enzymes are usually found in eukaryotic systems and include enzymes such as cellulose synthase and chitin synthase that are capable of sequentially attaching multiple residues to a growing carbohydrate chain. Most microbial transferases are nonprocessive and are responsible for the addition of single sugar residues. They are involved in the synthesis of complex polysaccharide structures with few or no repeating residues. These include various forms of lipopolysaccharides (LPS), the repeat units of succinoglycan and O-antigens on the surfaces of bacterial cells. The biological significance and functions of these complex structures have only recently begun to be unraveled. Polysaccharide synthesis does not follow a linear template, and the mechanisms that govern this synthesis are difficult to elucidate. Generally, glycosyltransferases display flexibility in their recognition of donor and acceptor substrates, a fact that 625

626

Kowal et al.

has been exploited in the synthesis of natural and unnatural oligosaccharides. Glycosyltransferase specificities come from the recognition of some critical hydroxyls in their substrates where modifications are not tolerated. This is seen in the case of glucosyltransferases and galactosyltransferases that distinguish their respective substrates solely on the basis of the orientation of the hydroxyl at the fourth carbon. However, the exact in vivo substrate specificities of majority of GTs have not been determined. In the majority of cases, the formation of each type of glycosidic bond requires a different enzyme, and sometimes, different enzymes (encoded by separate genes) can form the same bond type. Bacterial glycosyltransferases, the topic of this chapter, are enzymes responsible for the assembly of bacterial cell walls (e.g., succinoglycan) and lipopolysaccharides or polysaccharide structures attached to the lipids of the outer membrane of gram-negative bacterial cells. Referred to as endotoxins because of their physiological and pharmacological properties, lipopolysaccharides are at the forefront of bacterial interactions with the outside world. These structures have been found to be essential in processes ranging from root nodulation to human pathogenicity. Therefore studies on the corresponding glycosyltransferases are of great importance and interest. II.

GLYCOSYLTRANSFERASES

Glycosyltransferases are enzymes involved in the synthesis of oligosaccharides and the transfer of sugar moieties to free hydroxyls or amine groups of a variety of nonsugar substrates. The reaction catalyzed by these enzymes is: Nucleotide—sugar ⫹ R—OH → R—O—sugar ⫹ nucleotide where R can be a free saccharide, glycoprotein, glycolipid (aglycone), or protein and lipid. The enzyme nomenclature comes from the donor specificity of these molecules; these enzymes that transfer glucose are termed glucosyltransferases, those utilizing galactose are called galactosyltransferases and so forth. Inversion or retention of anomeric configuration in conversion from donor to acceptor can be used as a means of further classification of this enormously varied group of molecules. A.

Catalytic Mechanism

Although the details of catalysis by glycosyltransferases have not been elucidated per se, a likely mechanism has been proposed by analogy to glycosyl hydrolases [3,4]. Glycosyltransferases share two major catalytic mechanisms [5]. The reaction can result in either retention or inversion of configuration at the anomeric carbon. Since nucleotide diphospho sugars (the most common donors) have the ␣ configuration, ␣-glycosyltransferases are said to proceed with retention, while ␤-glycosyltransferases proceed with the inversion of the anomeric configuration. A single nucleophilic attack is necessary to generate the inversion of configuration in the so-called inverting enzymes (Fig. 1A). The retention mechanism is likely to involve a glycosyl enzyme intermediate (Fig. 1B), which forms from an oxocarbenium transition state [6]. B.

Processive or Nonprocessive: Structure–Function Relationship

Glycosyltransferase enzyme structure is divided into two domains: N-terminal domain A and C-terminal domain B. It has been noted that all glycosyltransferases

Microbial Glycosyltransferases

Figure 1

627

The catalytic mechanism of glycosyltransferases. (A) Inversion mechanism consisting of a single nucleophilic attack by the acceptor on the activated donor anomeric carbon. (B) Retention mechanism involving double nucleophilic substitution. The enzyme became transiently glycosylated and serves as the sugar donor in the subsequent steps [5]. The reaction goes through a proposed oxocarbenium transition state (middle left structure) [6]. Since aspartate and glutamate residues have been shown to be conserved and necessary for catalysis (see text), the general base and general acid functionalities shown are carboxylate groups. However, other functional groups could also perform these functions.

628

Kowal et al.

share domain A, while the processive enzymes contain both domains A and B [5]. This has led to an assignment of functions for the respective domains. Domain A is thought to contain two sites, one responsible for binding the activated sugar donor and the other for binding the acceptor molecule. Domain B is thought to bind a second activated donor molecule so that the processive enzymes can add two sugars to the growing chain either at the same time or in quick succession. The double addition is important because it restores the orientation of the polysaccharide’s end. Polysaccharides such as cellulose or chitin adopt a twofold screw axis [7–9], and the addition of a single residue would result in a different conformation of the reducing end with respect to the rest of the polymer. A double addition could, in principle, also be accomplished if the processive enzyme used an activated disaccharide donor such as UDP-cellobiose. However, this has not been observed in living systems.

C.

Glycosyltransferase Structure

Other studies on the structure of glycosyltransferases have uncovered that groups of amino acid residues necessary for catalysis are usually conserved within families of proteins. This is exemplified by two ␣-glucosyltransferases: ␣-1,3-glucosyltransferase (RfaI) [10] and ␣-1,2-glucosyltransferase (RfaJ) [11]. Both proteins have been analyzed by means of hydrophobic cluster analysis (HCA) [12,13] and site-directed mutagenesis. The results revealed the existence of four highly conserved regions in RfaI, RfaJ, and other related proteins (Table 1). Within these regions there are four critical aspartic acid residues which, when replaced by site-directed mutagenesis, abolish the activity of the enzymes. These latest studies were further supplemented with the publication of the crystal structure of a nucleotide diphospho sugar transferase SpsA from Bacillus subtilis [14], the only currently available microbial glycosyltransferase structure. (The other two glycosyltransferases structures published are DNA-modifying ␤-glucosyltransferase from bacteriophage T4 [15] and bovine ␤-1,4-galactosyltransferase catalytic domain [16].) SpsA protein is a member of family 2 glycosyltransferases (see Ref. 17 for family classification) and is an inverting enzyme. The 256-residue nonprocessive protein is most likely involved in the production of mature spore coat of B. subtilis, but its natural substrate has not been determined. The structure reveals that residues 2–100 form the donor nucleotide binding site, which consists of four parallel ␤ strands flanked by two ␣ helices. The remaining residues are folded into a ␤ sheet, which is flanked by ␣ helices and forms a groove that binds the acceptor molecule. A cocrystal of SpsA and UDP shows the interactions between the protein and the donor nucleotide. The uracil stacks with Tyr 11. The binding of the base is further stabilized by a hydrogen bond between an invariant Asp 39 and N3 of uracil. Asp 99 coordinates to the Mn2⫹ ion bound to the phosphate moieties. Further binding interaction is provided by Lys 13, which forms a hydrogen bond with the ␣ phosphate. Two other aspartate residues, Asp 158 and Asp 191, hydrogen-bond to a glycerol molecule that is thought to occupy the acceptor binding site. Owing to its ideal placement, Asp 191 is suspected of being the catalytic residue. However, the bound glycerol molecule makes that assignment somewhat uncertain.

Alignmment of RfaI, RfaJ, and Related Proteinsa

Microbial Glycosyltransferases

Table 1

a

The four highly conserved regions are shown with the number of the residue at which they begin. Shaded areas correspond to the residues that deviate from the conserved sequences. Symbol x indicates residues that, when replaced, abolish the catalytic activity [11].

629

630

D.

Kowal et al.

Classification of Microbial Glycosyltransferases

Microbial glycosyltransferases are an extremely diverse group of enzymes. Their classification is made difficult by the lack of knowledge about their structure, functions and substrate specificities. Formally, glycosyltransferases are listed by the Enzyme Commission under EC 2.4.x.y registry numbers. They can be classified based on the reaction catalyzed and the substrate specificity according to the recommendations of the International Union of Biochemistry and Molecular Biology (IUBMB) [18]. A complementary classification scheme has been proposed that also considers sequence and structural feature similarities and how they relate to functions [17]. Such schemes have already been applied to a related group of enzymes, the glycosyl hydrolases [19]. To classify GTs, researchers [17] retrieved 35 known hexosyltransferase enzyme sequences and used them to perform BLAST similarity searches [20] with sequences available from the Swiss-Prot and EMBL/GenBank databases. The sequences were further compared by means of the hydrophobic cluster analysis method [12,13] and classified into 26 families on the basis of HCA similarity over a sequence of at least 100 amino acids (see Ref. 17). Hydrophobic cluster analysis is a sequence comparison and alignment method introduced by Gaboriaud and others in 1987 [12]. The method relies upon twodimensional representation of protein sequences to determine regions of probable structural similarity. The method has been successfully applied [5,17,21] to detect regions of structural similarity even in proteins with little sequence identity and when little or no three-dimensional information is available.

III.

GLYCOSYLTRANSFERASES IN NATURE: STRUCTURE OF THE BACTERIAL LIPOPOLYSACCHARIDES

Bacterial cells express an enormous variety of polysaccharide structures. Usually found protruding from the outer membrane, these entities include exopolysaccharide (EPS), lipopolysaccharide (LPS), and lipo-oligosaccharide (LOS). EPS can be found in both gram-positive and gram-negative cells and is made of repeating polysaccharide units. Membrane-attached EPS forms the cell capsule. However, it can also be released into the surrounding medium as slime. LPS is composed of lipid A, core polysaccharide, and O-antigen, whereas LOS is a variation of LPS that lacks the repeating units of the O-antigen. In some gram-negative cells (e.g., E. coli and Salmonella typhimurium), the outer leaflet of the outer membrane can be composed almost exclusively of lipid A molecules. Lipid A in E. coli [22,23], Salmonella typhimurium [24,25], and Neisseria meningitidis [26] consists of two ␤,1-6-linked 2,3-diacylglucosamine units. The fatty acid chains attached to the 2- and 3-positions are generally 2–6 carbons shorter than their counterparts in glycerophospholipids. In addition, the 2,3-diacyl chains contain an R-3-hydroxyl substituent. The R3-hydroxyls of the nonreducing sugar are esterified with laurate and myristate fatty acids. The structure is phosphorylated at the 1- and 4⬘-positions. The core polysaccharide is attached at the 6⬘-position of lipid A. The inner core consists of a KDO (3-deoxyD-manno-octulosonic acid) region and a heptose (Hep) region (Fig. 2). The number of residues varies between species.

Microbial Glycosyltransferases

631

Figure 2 Schematic representation of the LOS structure of N. meningitidis and N. gonorrhoeae. Variations in the largest ␣ chain are responsible for most of the immunogenic properties of the LOS. Abbreviations: KDO, 3-deoxy-D-manno-2-octulosonic acid; Hep, L-glyceroD-manno-heptose; Glc, glucose; GlcNAc, N-acetylglucosamine; PEA, phosphoethanolamine. Note: C11 indicates an additional 11 carbon atoms, not the total chain length.

A.

LOS Biosynthesis in N. meningitidis and N. gonorrhoeae: A Case Study

The pathogens Neisseria gonorrhoeae and Neisseria meningitidis have received much attention in the last few years. Neisserial lipo-oligosaccharide (LOS) is distinct from the LPS of the enteric gram-negative bacilli. The differences include phosphorylation patterns, composition and attachment of acyl chains, and the lack of the repeating O-antigen characteristic of the enteric species. In Neisseria species, the core region consists of Hep2-Kdo2 unit that is further elaborated with oligosaccharide chains ␣, ␤, and ␥. The ␣ chain is attached to Hep I of the core polysaccharide through a glucose residue. Chains ␤ and ␥ are attached to the Hep II residue. The ␣ chain is of the most interest because it is the largest of the three and is responsible for most of the antigenic properties of the pathogens. It consists of only 2–5 sugar residues attached to the inner core structure, and the repeating units of the O-antigen are lacking. Length and composition of the ␣ chain are strain dependent. Antigenic variation caused by changes in the expression of immunotypes on the surface of the cell enables the cells to invade a host organism and evade its immune system [27,28].

632

Kowal et al.

Various immunotypes express differing LOS structures. The N. gonorrhoeae strain F62 ␣-chain LOS structure and the enzymes involved in its biosynthesis are shown in Fig. 3. B.

Genes Involved in LOS ␣-Chain Synthesis

The genes encoding the proteins involved in LPS synthesis are spread throughout the bacterial chromosome. Many have been found in loci that contain clusters of genes. Interestingly, many of these genes have been found to be coexpressed with protein involved in aspects of cellular metabolism. For example, the KDO biosynthesis genes are found together with genes for glycolysis enzymes. Morse et al. [29] have observed that the changes in glucose metabolism are correlated to the changes in the LOS profile of N. gonorrhoeae. These changes have also altered complementmediated killing of the cells by human sera. Further, Tsai et al. [30] have observed that modifications in LOS profiles in N. meningitidis can be afforded by changes in growth media composition. All the transferase genes involved in the synthesis of the ␣ chain of meningococcal and gonococcal LOS are found in the lgt (lipo-oligosaccharide glycosyl transferase) cluster [27,31]. The lgt cluster contains up to five genes designated lgtA through lgtE. The functions of these genes have been elucidated and are listed in Fig. 3. C.

Poly-G Tracts and Variation of LOS Expression in Neisseria Species

The genetic mechanism of the variability of LOS expression in N. gonorrhoeae has been uncovered [32–35]. Three genes, lgtA, lgtC, and lgtD, in the lgt locus contain polyguanine (poly-G) tracts that code for glycine residues in the enzymes. It has been found that the length of the poly-G tracts in lgtA and lgtC is variable and undergoes slipped-strand mispairing. This mechanism inserts or deletes a G residue, leading to the rest of the gene being out of frame and therefore resulting in a truncated, inactive transferase. The variability in the expression of the transferase genes is reflected in the changes in LOS structure present on the surface of the cells. The variation has been found to occur at a frequency of once in 100–1000 generations [32].

IV.

SYNTHETIC APPLICATIONS OF GLYCOSYLTRANSFERASES

Over the past few years, glycosidases have been widely used for the preparation of a broad spectrum of glycoconjugates and derivatives employing either transglycosylation or reverse hydrolysis reactions [36]. The synthetic use of glycosyltransferases, especially of microbial origin, has so far been very limited. This is mainly due to two factors: low availability and high cost of both the enzymes and the nucleotidesugar donor cofactors. Currently, only five glycosyltransferases, all eukaryotic in origin, are commercially available: ␣1,3-glucosyltransferase, ␤1,4-galactosyltransferase, ␣1,2-mannosyltransferase, ␣2,3-sialyltransferase, and ␣2,6-sialyltransferase. These range in price from $25/unit to as much as $1800/unit (unit is defined as the transfer of 1 ␮mol of sugar from donor to acceptor in 1 min at 37⬚C). However, the

Microbial Glycosyltransferases 633

Figure 3 Structures of the two alternative ␣ chains of meningococcal LOS. The residues are attached in sequence by the transferases as shown. Transferases encoded by the IgtA, IgtC, and IgtD genes contain the poly-G tracts and are responsible for the phase variation in LOS. In fact, structure II is made only if IgtA is inactive. At certain stages of meningococcal infections, the chain may further be sialylated.

634

Kowal et al.

number of microbial glycosyltransferases that have been overexpressed is continually increasing (Table 2) [14,37–62]. The commercially available enzymes have been employed in the synthesis of a large variety of natural and unnatural oligosaccharides [63–65]. The cost of nucleotide-sugar donors is also very high. UDP-Gal, used in the synthesis of the wellknown ␣-gal (Gal␣1-3Gal-R) epitopes is listed at $4000/g, making its stoichiometric use prohibitively expensive. A.

Making Oligosaccharides Accessible

There are three general approaches to oligosaccharide preparation. The first approach is the isolation of the desired compound from its natural source. This is, however, a very laborious and time-consuming method and may yield only small quantities of the desired product. The second and most well-established approach is to use chemical synthesis. There are a variety of methods available, but the synthesis is also time-consuming and requires much expertise. Multiple protection and deprotection steps for both donor and acceptor components are necessary to ensure desired regioand stereoselectivity. Enzymatic synthesis, as a third approach, is becoming a viable alternative to chemical synthesis, especially with increasing availability of recombinant and natural glycosyltransferase enzymes. The obvious advantages of enzymatic preparations are that they can be carried out in mild conditions, do not require protection and deprotection steps, and ensure required regio- and stereospecificity. As more and more glycosyltransferases became available, the remaining problem will be the cost of the nucleotide-sugar cofactors. Two approaches are being pursued to circumvent this predicament. The first is to change the donor requirement of the transferase by coexpressing it with an epimerase. For example, we have engineered a bacterial strain [unpublished data] that contains a plasmid-encoded fusion protein that consists of UDP-Gal-4-epimerase and bovine ␣-1,3-galactosyltransferase, the latter being a key enzyme in the biosynthesis of a terminal structure Gal␣1,3Gal␤1,4Glc(NAc)-R, known as ␣-Gal. This epitope has been implicated in the hyperacute rejection of tissue grafts in xenotransplantation [66–68]. The fusion with the epimerase enables us to use UDP-Glc instead of UDP-Gal, to enzymatically attach a galactose to lactose of N-acetyllactosamine as shown in Fig. 4. This affords a 40-fold cost reduction in production of the ␣-Gal trisaccharide. The second approach to the production of oligosaccharides is pathway engineering. This methodology has already been applied to the synthesis of variety of interesting compounds. Its use in the production of oligosaccharides was pioneered by researchers at the Tokyo Research Laboratories, Kyowa Hakko Kogyo Co., Ltd. [69]. The authors engineered three strains of bacteria and coupled them to produce UDP-galactose and globotriose. A recombinant strain of E. coli was engineered with UDP-Gal biosynthetic genes (galactose-1-phosphate uridyltransferase, galT; galactokinase, galK, glucose 1-phosphate uridyltransferase, galU; and pyrophosphatase, ppa). Corynebacterium ammoniagenes was used to convert inexpensive orotic acid to UTP. Finally, the production of globotriose (Gal␣1,4Lac) was accomplished in another recombinant E. coli strain that expressed the Neisseria gonorrhoeae lgtC gene encoding ␣-1,4-galactosyltransferase. The same methodology was also applied to the large-scale production of Nacetyllactosamine [70]. As before, three bacterial strains were coupled together. C.

Microbial Glycosyltransferase Enzymes for Which an Overexpression System Has Been Developed

Description

a

2.1.4.182

2.4.1.4

2.4.1.10

2.1.4.10

Organism

Accession numbera

Ref.

Acetobacter xylinum Bacillus subtilis Bacillus subtilis Bradyrhizobium japonicum Chlamydia pneumoniae Erwinia herbicola Erwinia herbicola Escherichia coli Haemophilus influenzae Helicobacter pylori Mycobacterium avium Neisseria gonorrhoeae Neisseria meningitidis Neisseria meningitidis Neisseria meningitidis Neisseria meningitidis Neisseria polysaccharea Photobacterium damsela JT0160 Pseudomonas aeruginosa Pseudomonas syringae Rhizobium leguminosarum Rhizobium leguminosarum Rhizobium leguminosarum Sinorhizobium meliloti Sinorhizobium meliloti Streptococcus equisimilis Streptococcus pyogenes Synechocystis sp. Zymomonas mobilis

U37258 P54166 P39621 L22756 Z31593 P21686 Q01330 P17443 P45011 AF008596 AF060183 U60664 U60663 U25839 U25839 U65788 AJ011781 AB012285 L28170 O68609 X94963 X94963 AF050103 P33695 P04341 AF023876 L20853 D90911 2107190A

37 38 39,14* 40 41 42 42 43 44 45 46 47 47,48 49 50 50 51 52 53,54 55 56 56 56 57 58 59,60 60 61 62

Accession numbers starting with P-Q are from the Swiss-Prot databank, others from the EMBL/GenBank (*crystal structure reference).

635

␣-Mannosyl transferase AceA Glc transferase YpfP (processive) SpsA protein Fucosyltransferase NodZ KDO-transferase KdtA (multifunctional) Zeaxanthin glucosyltransferase Zeaxanthin glucosyltransferase N-Acetylglucosamine transferase Lipid A disaccharide synthase LpxB ␣-1,3-Fucosyltransferase (HpfucT) Rhamnosyltransferase RtfA ␣-2,3-Sialyltransferase ␣-2,3-Sialyltransferase ␤-1,3-GlcNAc transferase LgtA ␤-1,4-Gal transferase LgtB ␣-1,4-Gal transferase LgtC Amylosucrase (processive) ␣-2,6-Sialyltransferase Bst Rhamnosyltransferase rhlAB Levansucrase lsc Galactosyltransferase LpcA KDO transferase LpcB Mannosyltransferase LpcC ␤-1,4-Glc transferase ExoM GlcNAc transferase NodC Hyaluronan synthase 1 Hyaluronan synthase 1 Sucrose-phosphate synthase SpsA Levansucrase SucZE2

EC number

Microbial Glycosyltransferases

Table 2

636

Kowal et al.

Figure 4 The fusion of epimerase (UDP-Gal-4-epimerase) and ␣1,3GalT (bovine ␣1,3 galactosyltransferase-EC 2.4.1.151) enables the use of UDP-Glc instead of UDP-Gal, a 40fold reduction in the cost of production of the ␣-Gal epitopes.

ammoniagenes was used to produce UTP from orotic acid, and an E. coli strain was engineered to express the UDP-Gal biosynthetic pathway. Another E. coli strain expressed recombinant lgtB gene from N. gonorrhoeae, encoding a ␤-1,4-galactosyltransferase. The authors were able to obtain 107 g/L of product after a 38 h incubation period by supplying the cells with orotic acid, galactose, and N-acetylglucosamine (Fig. 5). This production system has the obvious advantages of not requiring purification of the enzymes involved and generating a final product in high yields from inexpensive starting materials. The only drawback seems to be the need to use multiple bacterial strains. V.

CONCLUSION

Carbohydrate research has lagged behind work done on the other two well-known macromolecules, proteins and nucleic acids. This was mostly due to the enormous variety and complexity of oligo/polysaccharide structures found in nature. With the discovery of the multiplicity of roles now assigned to carbohydrate molecules, there is an increasing need for methodologies that will allow researchers access to these structures. Organic synthetic methods are available for the production of oligosaccharides but carry disadvantages that make them difficult to apply. Enzymatic synthesis of carbohydrates is still in its infancy. Low availability of transferase enzymes and high cost of the activated sugar donor molecules make these methodologies prohibitively expensive. However, new methodologies are being developed that are focusing on two fronts: increasing the availability of glycosyltransferases and lowering the cost of cofactors. Finally, whole-cell systems are being developed that address both of these targets. These do not require the purification of the enzymes or the input of costly sugar donor molecules, and they can produce desired oligosaccharides in high quantity. Besides the functions they serve in the natural environment, novel applications are being sought for carbohydrates. Oligosaccharide moieties are used as vaccines in prevention of variety of microbial infections. Recently, new polysaccharide conjugate vaccines have been developed [71]. Polysaccharides have been discovered to have antibacterial properties. One such example is the modified disaccharide moiety of vancomycin, which alone exhibits novel antibacterial

Microbial Glycosyltransferases The production systems for globotriose (Gal␣1,4Lac) [69] and N-acetyllactosamine [70] as designed by researchers at the Tokyo Research Laboratories, Kyowa Hakko Kogyo Co. Ltd. Both IgtB and IgtC were from Neisseria gonorrhoeae.

Figure 5

637

638

Kowal et al.

properties by inhibiting the transglycosylation step in peptidoglycan synthesis [72]. Immunoreactive oligosaccharides, such as the ␣-Gal epitopes have been used to enhance the immunogenicity of a tumor vaccine [73]. The need for oligosaccharides will grow in proportion to the number of their applications. Cost-effective methods of their synthesis are clearly needed. It is also clear that much further work will be required in this field. Only three glycosyltransferase crystal structures are currently available, and only one of these is a microbial glycosyltransferase [14]. The exact mechanism of action of these enzymes still awaits elucidation. However, as more and more of glycosyltransferase enzymes became available, their synthetic usefulness will inevitably increase and, as the first wholecell attempts [69,70] show, the use of microbial glycosyltransferases can be facile, efficient, and full of possibilities. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

19. 20. 21. 22. 23. 24. 25. 26.

R Kleene, EG Berger. Biochim Biophys Acta 1154:283–325, 1993. BS Hundle, DA O’Brien, M Alberti, P Beyer, JE Hearst. Proc Natl Acad Sci USA 89: 9321–9325, 1992. B Henrissat, A Bairoch. Biochem J 316:695–696, 1996. B Henrissat, GJ Davies. Curr Opin Struct Biol 7:637–644, 1997. IM Saxena, RMJ Brown, M Fevre, RA Geremia, B Henrissat. J Bacteriol 177:1419– 1420, 1995. D Kapitanov, RK Yu. Glycobiology 9:961–978, 1999. KH Gardner, J Blackwell. Biopolymers 13:1975–2001, 1974. KH Gardner, J Blackwell. Biopolymers 14:1581–1595, 1975. R Minke, J Blackwell. J Mol Biol 120:167–181, 1978. K Shibayama, S Ohsuka, T Tanaka, Y Arakawa, M Ohta. J Bacteriol 180:5313–5318, 1998. K Shibayama, S Ohsuka, K Sato, K Yokoyama, T Horii, M Ohta. FEMS Microbiol Lett 174:105–109, 1999. C Gaboriaud, V Bissery, T Benchetrit, JP Mornon. FEBS Lett 224:149–155, 1987. L Lemesle-Varloot, B Herrissat, C Gaboriaud, B Bissery, A Morgat, JP Mornon. Biochemie 72:555–574, 1999. SJ Charnock, GJ Davies. Biochemistry 38:6380–6385, 1999. A Vrielink, W Ruger, HPC Driessen, PS Freemont. EMBO J 13:3413–3422, 1994. LN Gastinel, C Cambillau, Y Bourne. EMBO J 18:3546–3557, 1999. JA Campbell, GJ Davies, V Bulone, B Henrissat. Biochem J 326:929–942, 1997. International Union of Biochemistry and Molecular Biology. Enzyme Nomenclature: Recommendations of the Nomenclature Committee of the International Union of Biochemistry and Molecular Biology. Academic Press, San Diego, 1992. B Henrissat. Biochem J 280:309–316, 1991. SF Altschul, W Gish, W Miller, EW Myers, DJ Lipman. J Mol Biol 21:403–410, 1990. A Geremia, AE Petroni, L Ielpi, B Henrissat. Biochem J 318:133–138, 1996. K Takayama, N Qureshi, P Mascagni, MA Nashed, L Anderson, CRH Raetz. J Biol Chem 258:7379–7385, 1983. M Imoto, S Kusumoto, T Shiba, ET Rietschel, C Galanos, O Luderitz. Tetrahedron Lett 26:1907–1908, 1985. N Qureshi, K Takayama, D Heller, C Fenselau. J Biol Chem 258:12947–12951, 1983. N Qureshi, K Takayama, E Ribi. J Biol Chem 257:11808–11815, 1982. VA Kulshin, U Zahringer, B Lindner, CE Frasch, CM Tsai, BA Dmitriev, ET Rietschel. J Bacteriol 174:1793–1800, 1992.

Microbial Glycosyltransferases 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.

37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60.

639

MP Jennings, DW Hood, IRA Peak, M Virji, ER Maxon. Mol Microbiol 18:729–740, 1995. FP de Vries, A van der Ende, JPM van Putten, J Dankert. Infect Immun 64:2998–3006, 1996. SA Morse, CS Mintz, SK Sarafian, L Bartenstein, M Bertram, MA Apicella. Infect Immun 41:74–82, 1983. C-M Tsai, R Boykins, CE Frasch. J Bacteriol 155:498–504, 1983. EC Goschlich. J Exp Med 180:2181–2190, 1994. JMP van Putten. EMBO J 12:4043–4051, 1993. RJ Danaher, JC Levin, D Arking, CL Burch, R Sandlin, DC Stein. J Bacteriol 177: 7275–7279, 1996. Q-L Yang, EC Gotschlich. J Exp Med 183:323–327, 1996. CL Burch, RJ Danaher, DC Stein. J Bacteriol 179:982–986, 1997. KGI Nilsson. Synthesis with glycosidases. In Modern Methods in Carbohydrate Synthesis, SH Khan and RA O’Neill. eds. Harwood Academic: Amsterdam, 1996, pp 518– 547. RA Geremia, M Roux, DU Ferreiro, R Dauphin-Dubois, AC Lellouch, L Ielpi. Mol Gen Genet 261:933–940, 1999. P Jorasch, FP Wolter, U Zahringer, E Heinz. Mol Microbiol 29:419–430, 1998. SJ Charnock, GJ Davies. Acta Crystallogr D Biol Crystallogr 55:677–678, 1999. C Quinto, AHM Wijfjes, GV Bloemberg, L Blok-Tip, IM Lopez-Lara, BJ Lugtenberg, JE Thomas-Oates, HP Spanik. Proc Natl Acad Sci USA 94:4336–4341, 1997. S Lobau, U Mamat, W Brabetz, H Brade. Mol Microbiol 18:391–399, 1995. BS Hundle, DA O’Brien, M Alberti, P Beyer, JE Hearst. Proc Natl Acad Sci USA 89: 9321–9325, 1992. M Crouvoisier, D Mengin-Lecreulx, J van Heijenoort. FEBS Lett 449:289–292, 1999. S Servos, S Khan, D Maskell. Gene 175:137–141, 1996. Z Ge, NW Chan, MM Palcic, DE Taylor. J Biol Chem 272:21357–21363, 1997. TM Eckstein, RS Silbaq, D Chatterjee, NJ Kelly, PJ Brennan, TJ Belisle. J Bacteriol 180:5567–5573, 1998. M Gilbert, DC Watson, AM Cunningham, MP Jennings, NM Young, WW Wakarchuk. J Biol Chem 271:28271–28276, 1996. M Gilbert, AM Cunningham, DC Watson, A Martin, JC Richards, WW Wakarchuk. Eur J Biochem 249:187–194, 1997. O Blixt, I van Die, T Norberg, DH van den Eijnden. Glycobiology 9:1061–1071, 1999. WW Wakarchuk, A Cunningham, DC Watson, NM Young. Protein Eng 11:295–302, 1998. GP De Montalk, M Remaud-Simeon, RM Willemot, V Planchot, P Monsan. J Bacteriol 181:375–381, 1999. T Yamamoto, M Nakashizuka, I Terada. Biochem (Tokyo) 123:94–100, 1998. UA Ochsner, A Fiechter, J Reiser. J Biol Chem 269:19787–19795, 1994. HL Rocchetta, LL Burrows, JC Pacan, JS Lam. Mol Microbiol 28:1103–1119, 1998. U Hettwer, FR Jaeckel, J Boch, M Meyer, K Rudolph, MS Ullrich. Appl Environ Microbiol 64:3180–3187, 1998. JL Kadrmas, D Allaway, RE Studholme, JT Sullivan, CW Ronson, PS Poole, CR Raetz. J Biol Chem 273:26432–26440, 1998. AC Lellouch, RA Geremia. J Bacteriol 181:1141–1148, 1999. E Kamst, J Pilling, LM Raamsdonk, BJ Lugtenberg, HP Spanik. J Bacteriol 179:2103– 2308, 1997. K Kumari, PH Weigel. J Biol Chem 272:32539–32546, 1997. VL Tlapak-Simmons, BA Baggenstoss, K Kumari, C Heldermon, PH Weigel. J Biol Chem 274:4246–4253, 1999.

640 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73.

Kowal et al. L Curatti, E Folco, P Desplats, G Abratti, V Limones, L Herrera-Estrella, G Salerno. J Bacteriol 180:6776–6779, 1998. H Yanase, J Fujimoto, M Maeda, K Okamoto, K Kita, K Tonomura. Biosci Biotechnol Biochem 62:1802–1805, 1998. S David, C Auge´. Adv Carbohydr Chem Biochem 49:175–237, 1991. C-H Wong. Agnew Chem Intl Ed Engl 34:521–529, 1995. MM Palcic, O Hindsgaul. Trends Glycosci Glycotechnol 8:37–41, 1996. AH Good, DK Cooper, AJ Malcolm, RM Ippolito, E Koren, FA Neethling, Y Ye, N Zuhdi, LR Lamontagne. Transplant Proc 24:559–562, 1992. U Galili, Immunol Today 14:480–482, 1993. MS Sandrin, HA Vaughan, PL Dabkowski, IF McKenzie. Proc Natl Acad Sci USA 90: 11391–11395, 1993. S Koizumi, T Endo, K Tabata, A Ozaki. Nat Biotechnol 16:847–850, 1998. T Endo, S Koizumi, K Tabata, S Kakita, A Ozaki, Carbohydr Res 316:179–183, 1999. M Moreau, Ann pharm Fr 57:223–231, 1999. M Ge, Z Chen, RH Onishi, J Kohler, LL Silver, R Kerns, S Fukuzawa, C Thompson, D Kahne. Science 284:507–511, 1999. DC LaTemple, U Galili. Subcell Biochem 32:361–379, 1999.

18 Metabolic Substrate Engineering as a Tool for Glycobiology Scarlett Goon and Carolyn R. Bertozzi Howard Hughes Medical Institute and University of California, Berkeley, California

I.

INTRODUCTION

The biological importance of oligosaccharides was first recognized in the context of their role in metabolism and energy storage. During the past three decades, it has become apparent that complex oligosaccharides also regulate many of the fundamental processes occurring in a biological system. Of special interest are the oligosaccharides found on the surfaces of cells (Fig. 1), which guide their social behavior. Among many other responsibilities, oligosaccharides mediate cell–cell interactions, regulate the serum half-life of glycoproteins, and serve as specific ligands for bacteria, viruses, and parasites [1]. Progress toward the elucidation of oligosaccharide function has historically been impeded by their structural complexity and heterogeneity on cells [2]. The biosynthesis of glycoconjugates is determined by various factors, including the relative abundance and specificity of glycosyltransferases (the enzymes that build oligosaccharides one saccharide at a time), the spatial localization of these enzymes, the availability of nucleotide sugar donors, and the nature of the underlying substrate. Collectively, these factors result in a diverse array of heterogeneous structures that are difficult to control by using conventional genetic and biochemical tools. Considerable creativity has been exercised in the development of strategies for modulating oligosaccharide structures on cell surfaces in a well-defined fashion. For example, specific glycosidase and glycosyltransferase inhibitors [3–8] have provided insight into the biosynthesis of glycans as well as their roles in biological processes. Glycosyltransferases have been used to modify cell surface glycans by exogenous treatment [9–11]. Another approach, the subject of this chapter, is to intercept metabolic pathways with unnatural monosaccharide substrates. If incorporated into cell 641

642

Goon and Bertozzi

Figure 1

The landscape of the surfaces of cells.

surface oligosaccharides, modified sugars can aid in the elucidation of biological function (Fig. 2). Alternatively, modified substrates might block biosynthetic enzymes, producing phenotypes similar to those induced by glycosyltransferase inhibitors. We summarize here the historical development of metabolic substrate engineering as a tool for glycobiology. This chapter focuses on the journey of carbohydrate precursors and synthetic carbohydrate analogs through the metabolic pathways of the cell and the information that can be gained from investigating such processes. We begin with an overview of the structures of glycoconjugates on cell surfaces (for more comprehensive reviews, see Refs. 2, 12–14, and 24). Next, the pathways of monosaccharide interconversions inside the cell are summarized, providing a framework for opportunities in unnatural monosaccharide metabolism. Specific examples of modified sugars and their effects on cells are then presented. Finally, the biosynthetic pathway of sialic acid is discussed with respect to its utility for cell surface oligosaccharide engineering. II.

PROTOTYPICAL GLYCOCONJUGATES ON CELL SURFACES

Cell surface glycoconjugates can be grouped into four main classes: N-linked glycoproteins, O-linked glycoproteins, glycosyl phosphatidylinositol (GPI)-anchored proteins, and glycolipids. These are the major cell surface hosts of monosaccharides derived from intracellular metabolism. N-Linked glycosylation in eukaryotes is a cotranslational event that occurs in the endoplasmic reticulum (ER) [2,16]. It is found almost exclusively on proteins that contain the consensus sequence AsnXaaSer/Thr (where Xaa is any amino acid except Pro). Glycosylation of the asparagine residue in the appropriate sequon is mediated by a membrane-bound enzyme oligosaccharyl-

> Figure 2

Metabolic substrate engineering as a strategy for modulating cell surface oligosaccharide structure. Modified metabolic substrates can intercept a biosynthetic pathway in two ways: the pathway might be inhibited, leading to truncated structures on the cell surface, or the modified substrate might be incorporated into oligosaccharides in place of the normal substrate.

Metabolic Substrate Engineering

643

644

Goon and Bertozzi

transferase. This enzyme transfers an oligosaccharide with the structure Glc3Man9GlcNAc2 from a dolichol diphosphate precursor to the side chain of particular asparagine residues in the polypeptide chain (Fig. 3). During its journey through the secretory pathway, this tetradecasaccharide is first trimmed and then elaborated to form an N-linked glycan with a high-mannose, hybrid, or complex-type structure [16,17]. Examples of these structures are shown in Fig. 4 [25]. Unlike N-glycosylation, O-glycosylation begins with the transfer of a single monosaccharide residue, usually GalNAc, to a serine or threonine residue of the polypeptide chain [12]. The GalNAc residue is then further elongated through stepwise enzymatic modifications by glycosyltransferases, giving rise to specific core structures. To date, eight core structures have been identified by NMR spectroscopy or mass spectrometry (Fig. 5) [26–30]. These glycans can then be further elaborated by the addition of a sialic acid or Fuc residue, sulfate, methyl, acetyl, or poly-N-acetyllactosamine units. A subset of O-glycoproteins are the proteoglycans, which present long glycosaminoglycan chains from a peptide-proximal xylose residue (Fig. 6). GPI anchors are involved in cell signaling, protein targeting, and protein secretion [15,23]. The addition of a GPI anchor to a polypeptide is a posttranslational event that occurs in the endoplasmic reticulum (ER). The structures of GPI anchors are very complex. The core structure, consisting of ethanolamine, Man residues, GlcN, and phosphatidylinositol, is conserved from protozoan to mammalian organisms (Fig. 7). However, the peripheral structures vary through species and cell type. The conserved GPI core can be further modified by the addition to the core mannoses of ethanolamine phosphate residues, GalNAc residues, and Man residues. This entire structure is then bound to a lipid that is embedded in the membrane. Similar to Nglycosylation, GPI modification of proteins involves the transfer of a preassembled precursor, en bloc, to the C-terminus of a target protein by amide bond formation with the ethanolamine group. Glycolipids present oligosaccharide epitopes immediately proximal to the plasma membrane [18–22]. Most glycolipids can be classified into three major groups: sphingolipids, gangliosides, and glycero- and isoprenol-glycolipids [31]. The biosynthesis of glycolipids proceeds by stepwise addition of monosaccharide units to a lipid carrier in the compartments of the secretory pathway. As in the case of glycoprotein biosynthesis, the transfer of nucleotide-activated monosaccharides is mediated by glycosyltransferases. An example of a ganglioside, GM3, is shown in Fig. 8 [32].

III.

OVERVIEW OF METABOLIC SUBSTRATE ENGINEERING

The complex glycoconjugates expressed on eukaryotic cell surfaces (Figs. 3–8) comprise primarily 10 monosaccharides: Glc, Gal, Man, Fuc, GlcN, GlcNAc, GalNAc, Xyl, GlcUA, and sialic acid (Neu5Ac). These sugars can be biosynthesized de novo within a cell, often from Glc; or, in many cases, key metabolic intermediates can be supplied exogenously to a cell. The pathways for these interconversions are summarized in Figure 9 [33–36]. If the enzymes in these biosynthetic pathways could tolerate unnatural substrates, subtle modifications, such as replacement of a hydroxyl group with a hydrogen atom or halogen, could be introduced into cell surface glycans. This possibility attracted the attention of many groups interested in understanding how specific carbohydrate structures dictate molecular and cellular interactions. (text continues on p. 652)

Metabolic Substrate Engineering 645

Figure 3 The oligosaccharide Glc3Man9GlcNA2 is a common precursor in the biosynthesis of the three types of N-linked oligosaccharides.

646 Examples of the three types of N-linked oligosaccharides found in human ␣-galactosidase A. Their common core structure is shown

Goon and Bertozzi

Figure 4 in gray.

Metabolic Substrate Engineering 647

Figure 5 The eight O-linked core structures characterized to date: core 1, Gal␤1,3GalNAc; core 2, Gal␤1,3(GlcNAc␤1,6)GalNAc; core 3, GlcNAc␤1,3GalNAc; core 4, GlcNAc␤1,3(GlcNAc␤1,6)GalNAc; core 5, GalNAc␣1,3GalNAc; core 6, GlcNAc␤1,6GalNAc; core 7, GalNAc␣1,6GalNAc; core 8, Gal␣1,3GalNAc.

648

Figure 6

The core region of heparin and heparan sulfate. Goon and Bertozzi

Metabolic Substrate Engineering

Figure 7

The conserved core structure of GPI anchors found in mammals. 649

650

GM3.

Figure 8

Goon and Bertozzi

Metabolic Substrate Engineering Monosaccharide interconversions in eukaryotic cells. Structures in squared boxes serve as exogenous substrates. Structures in rounded boxes are the activated monosaccharide donors. For simplicity, reactions are shown to progress in the forward direction.

651

Figure 9

652

Goon and Bertozzi

Some early attempts at metabolic oligosaccharide engineering using halogenated and deoxy sugars resulted in cell death. In retrospect, we now know that some of these compounds are inhibitory of intracellular biosynthetic pathways [37,38]. However, the toxicity of halogenated and deoxy sugars could possibly be exploited therapeutically for antiviral, antibacterial, or anticancer therapy. Examples are presented in detail later in this chapter. Remarkably, some unnatural monosaccharides were shown to access the metabolic pathways within cells. Some analogs were transformed only to an intermediate stage in the biosynthesis, never reaching the cell surface as a glycoconjugate component. Nevertheless, through the study of the imbalances in metabolic intermediates induced by these compounds, much has been learned about the specificities and roles of the enzymes involved. Perhaps most interesting of all has been a small, but growing number of compounds that successfully navigate glycoconjugate biosynthetic pathways and are presented on the cell surface. Such analogs, discussed in detail later, fall into two major categories. First, deoxy sugars, when incorporated into glycans, can lack the necessary hydroxyl groups required for further elaboration of the glycoconjugate, thereby resulting in cell surfaces deficient in some distal epitopes. The other category of compounds, exemplified by ManNAc derivatives that are converted to cell surface sialic acids, decorate the cell surface with novel structural and functional motifs. A discussion of these will come at the end of this chapter. A.

Metabolic Studies with Modified Amino Sugars and N-Acetamido Sugars

The initiatory metabolic interference studies of the 1970s utilized amino sugars. Early work demonstrated that even the naturally occurring compound GlcN causes specific inhibition of viral glycoprotein synthesis at high concentrations [39]. Removal of exogenous GlcN from cell culture media resulted in a reversal of cytotoxicity. NBromoacetyl-GlcN and its tetra-O-acetyl derivative produced strong in vitro cytotoxicity against both Ehrlich and L1210 leukemia cells at micromolar concentrations (Table 1) [40]. Indeed, injection of the same compounds into mice bearing Ehrlich ascites tumors cured 85% of the mice, although successive injections ultimately produced resistance. The molecular basis for this interesting series of results was not explored, but it might reflect the nonspecific alkylating activity of the ␣-bromoacetamide. Accordingly, the corresponding fluoro and chloro derivatives, which are less reactive alkylating agents, were found to be inactive both in vitro and in vivo, with a toxicity level 10-fold lower than that of the bromo analog (Table 1). In a separate study, Fondy et al. demonstrated that although the toxicity of fluoroacetamido analogs of GlcNAc and GalNAc was not affected by the presence or absence of O-acetyl protecting groups, a fourfold increase in toxicity was observed for tetra-O-acetyl N-fluoroacetamido-ManN [41,42]. The toxic effects of the chloroand bromoacetamido analogs of GlcNAc, ManNAc, and GalNAc were more similar to that of the N-fluoroacetamido-ManN analog (Table 2). In general, the tetra-Oacetylated derivatives were markedly more toxic than their nonacetylated counterparts. Fondy et al. postulated that while the bromo and chloro analogs exert their toxic effects as lipophilic alkylating agents, the fluoroacetamido analogs might have a different mechanism of toxicity. Again, the molecular basis of this suggestion was not explored at the time.

Metabolic Substrate Engineering

653

Table 1

Growth Inhibitory Activity of GlcN and GalN Analogs Against Erlich and L1210 Leukemia Cells

IC50 (␮M) R NH3Cl NHCOCH2F NHCOCH2Cl NHCOCH2Br

1

2

3

4

N/A ⱖ100 ⱖ100 ⱖ100

N/A 20 30–60 2–8

>1000 ⱖ100 ⱖ100 ⱖ100

220 30–80 20–30 4

The effects of amino sugars on cell viability have been evaluated using bacteria as well. The growth of Escherichia coli ML308 and K12 was inhibited by N-iodoacetamido-GlcN [43]. Further experiments indicated this compound inhibits the uptake of methyl-␣-D-glucoside by the bacterial cells, although complete inhibition was never observed. This effect is not surprising, considering that iodoacetamide reacts with several amino acid residues, such as Cys and His, and also with certain amino groups. It is possible that the iodoacetamide analogs irreversibly alkylated a key glucose transporter. Korytnyk and coworkers tested several pentaacetylhexosamine analogs for tumor cell toxicity and also evaluated their effects on glycoprotein biosynthesis [44]. Peracetylated GlcNAc inhibited the incorporation of [14C]GlcN into glycoproteins to a much greater extent than peracetylated GalNAc or ManNAc. However, protein biosynthesis was unaffected, as determined by levels of [3H]leucine incorporation.

Table 2

Toxicity of ManN, GlcN, and GalN Derivatives to B6D2F1 Mice

LD50 (mmol/kg) R

R⬘

ManN

GlcN

GalN

NHCOCH2F NHCOCH2F NHCOCH2Cl NHCOCH2Cl NHCOCH2Br NHCOCH2Br

H Ac H Ac H Ac

1.6 0.42 >3.7 1.6 >3.2 0.36

0.18 0.17 2.0 1.12 2.4 0.36

0.74 0.98 1.5 0.83 1.7 0.24

654

Goon and Bertozzi

Interestingly, replacement of the N-acetyl group of GlcNAc with a trifluoroacetyl group depressed [14C]GlcN incorporation by 31% within 5 h, while [3H]leucine incorporation was unaffected. Additionally, treatment of P288 cells with 1 mM GlcNAc resulted in no detectable change in ribonucleotide content, while treatment of cells with N-trifluoroacetyl-GlcN more than doubled the sugar-nucleotide content in the cell. The difference in the biological effects and metabolism of these compounds remains poorly understood, but a difference in cell permeability may be responsible. Derivatives in which the anomeric position of GlcNAc was modified were also evaluated as potential metabolic inhibitors of glycoconjugate biosynthesis [45,46]. In these fully acetylated compounds, the hydroxyl group at the anomeric position was replaced by a variety of groups including amino, N-chloroacetyl, N-bromoacetyl, N-iodoacetyl, N-trifluoroacetyl, and N-trifluoromethanesulfonyl groups. These GlcNAc derivatives were potent growth inhibitors of either mouse mammary adenocarcinoma TA3 or L1210 leukemia cells with IC50 values in the micromolar range (Table 3). Toxicity of these compounds might simply reflect their alkylating activity. Further modification of GlcNAc to S-glycosyl derivatives revealed that the 1-chloroacetylthio, 1-S-diphenylmethyl, and 1-S-trityl, inhibited the growth of TA3 cells at micromolar concentrations (Table 3). In yet another study, a ManN derivative con-

Table 3

Growth Inhibitory Activity of GlcNAc Derivatives Displaying Unnatural Anomeric Substituents Against TA3 Mammary Cells

IC50 (mM) R OH NH2 NHAc (␤) NHAc (␣) N3 NHCOCH2Cl NHCOCH2Br NHCOCH2I NHCOCF3 NHSO2CF3 SCOCH2Cl SCHPh2 STr SCPh2( p-OMeC6H4) a

R⬘ = Ac a

0.27 0.32 0.19 0.22 1.0 0.26 0.019a 0.026a 0.2 0.007 0.25 0.8 N/A N/A

R⬘ = H >1a >1 N/A N/A >1 >1 0.62a N/A N/A N/A N/A N/A 0.053 0.016

IC50 values were determined using L1210 leukemia cells.

Metabolic Substrate Engineering

655

taining a nitrosourea group at C2 was found to have antitumor activity [47]. The compound displayed an IC50 value against L1210 leukemia cells in the high micromolar range and increased the life span of leukemic mice by 35%. Although little mechanistic information was gleaned in these early studies, the differential effects of various unnatural substrates suggested possible engagement in metabolic pathways. B.

Halogenated Monosaccharides

Halogenated compounds have been widely used to study biological processes [48]. The halogens (F, Cl, Br, and I) are readily installed on synthetic analogs and might mimic a polar hydroxyl group. Among the halogens, fluorine has attracted the most attention. A comparison of bond lengths, van der Waals radii, and electronegativities indicates that the C—F bond quite closely resembles the C—OH bond [49]. The similar properties of the fluorine atom and hydroxyl group suggest that substitution might be tolerated by biosynthetic enzymes. The introduction of a fluorohexose or hexosamine analog into a biosynthetic pathway might lead to metabolic incorporation or metabolic disruption. An advantage to the use of fluorosugar analogs is the availability of additional analytical techniques, such as 19F NMR spectroscopy and radiolabeled 18F tracers, that facilitate characterization of downstream metabolic products [50–53]. Independent studies by Korytnyk and coworkers and May and Sartorelli have probed the effects of halogenated fucose (Fuc) and galactose (Gal) analogs on glycoconjugate biosynthesis [54–56]. Synthetic analogs bearing fluoro, chloro, bromo, or iodo groups at the 2- or 5-positions of Fuc, and the 6-position of Gal were shown to have inhibitory effects on the incorporation of the corresponding natural sugars, [3H]Fuc or [3H]Gal, into glycoproteins (Tables 4 and 5). The growth of murine L1210 leukemia cells was inhibited by 2-iodo-Fuc with an IC50 value of 0.06 mM (Table 4). Similarly, when mice with L1210 ascites tumors were injected with either 6fluoro-Gal (Table 5) or 2-bromo-Fuc (Table 4), their life span was prolonged by 32 and 25%, respectively. The 6-methanesulfonyl (Ms) analog of Gal had similar toxicity against L1210 leukemia cells. Other halogenated derivatives of Fuc such as 2bromo-Fuc, 6-chloro-Fuc, and 6-bromo-Fuc were much less toxic, with very little activity at concentrations above 1 mM (Table 4). The 6-halogenated analogs of Gal had inhibitory effects similar to those of their enantiomers, the corresponding 6halogenated-Fuc analogs (Table 5). Some halogenated analogs displayed cell-typespecific effects: 6-fluoro-Fuc did not inhibit the growth of L1210 leukemia cells at 1 mM, although it did reduce the incorporation of [3H]Fuc by 91% in human mammary tumor cells. Other compounds, such as 6-fluoro-Gal, although appearing nontoxic at 1 mM, had significant inhibitory effects toward the incorporation of [3H]Gal in P2888 murine lymphoma cells [55]. Westwood and coworkers have investigated the antitumor activity of fluorohexoses in vivo. From a panel of 6-halogenated Gal and Glc analogs, only 6-fluoro-Glc exhibited significant antitumor activity, blocking the growth of R-1 lymphoma up to 90% in mice [57]. Fluorohexosamine analogs that have been used in metabolic studies include 6fluoro-ManNAc, 6-fluoro-ManN, methyl 3-fluoro-ManNAc, 3-fluoro-GlcNAc, 4fluoro-GlcNAc, and 4-fluoro-GalNAc [58–60]. Many of these compounds were evaluated in both acetylated and nonacetylated forms. Among several 6-fluoro analogs

656

Goon and Bertozzi

Table 4

Toxicity or Growth Inhibitory Activity of Fuc Derivatives Against L1210 Leukemia Cells

R2

R3

R4

R6

IC50 (mM)

H Cl Br I Cl Br OH OH OH OH

OH OH OH OH OAc OAc OH OH OH OH

OH OH OH OH OAc OAc OH OH OH OH

H H H H H H F Cl Br I

1.0 0.2 No effect at 1.0 0.06 0.04 0.06 Slight effect at 1.0 Slight effect at 1.0 0.4 0.18

tested in murine L1210 leukemic cells, fully acetylated 6-fluoro-ManNAc had an IC50 value of 100 ␮M, 6-fluoro-ManN had an IC50 value of 1 mM, and 6-fluoroManNAc was inactive (Table 6). The 3-fluoro analogs both inhibited cell growth at micromolar concentrations and reduced tumor growth by 50%. The 3-fluoro GlcNAc derivative also appeared to inhibit glycoconjugate biosynthesis. In the presence of the 3-fluoro analog, incorporation of [3H]GlcN was reduced by 41% and [3H]thymidine by only 7%. This indicates that 3-fluoro-GlcNAc competes in the

Table 5

Growth Inhibitory Activity of Various Halogenated Gal Derivatives Against L1210 Leukemia Cells

X

IC50 (mM)

F Cl Br I Ms

>1 1.0 0.13 >1 0.075

Metabolic Substrate Engineering

657

Table 6

Growth Inhibitory Activity of Various ManNAc, GlcNAc, and GalNAc Derivatives Against L1210 Leukemia Cells

IC50 (mM) R1 OH OAc OH OMe (␣) OAc OAc OH OAc

R2

R3

R4

R6

ManN

GlcN

GalN

NHAc NHAc NH3Cl NHAc NHAc NHAc NHAc NHAc

OH OAc OH F F OH OAc OAc

OH OAc OH OAc OAc F F F

F F F OAc OAc OH OAc F

Inactive 0.1 1.0 0.043 N/A N/A N/A N/A

N/A N/A N/A N/A 0.027 Inactive 0.034 N/A

N/A N/A N/A N/A N/A N/A 0.035 0.024

pathway as GlcNAc itself. Contrary to the results in vitro, the 3-fluoro derivatives were relatively inactive in vivo, increasing life span by only 14% at 100 mg/kg. 4Fluoro-GalNAc, on the other hand, exhibited antitumor activity, increasing the life span of mice by 68% when administered a dosage of 50 mg kg⫺1 d⫺1 for 5 consecutive days. Since GalNAc is primarily contained in O-linked structures (Fig. 5) this might reflect an important role for O-linked glycoproteins in tumor growth. Despite the diverse behavior exhibited by the various halogenated Fuc and Gal derivatives, two clear trends emerged. First, the incorporation of 3H-labeled natural sugar analogs decreased with the increasing size of the halogen atom on the derivatized counterpart, and this was accompanied by an increase in cytotoxicity. This pair of effects suggests that more sterically demanding substituents are not tolerated by biosynthetic enzymes, thus minimizing biosynthetic competition with native substrates. Second, acetylated versions of all the derivatives were much more toxic than their analogous nonacetylated counterparts. This increase in toxicity could be attributed to the greater cellular permeability of these, and in general, any acetylated sugars [61]. Once inside a cell, esterases enzymatically remove the acetyl groups to produce a free monosaccharide that has full biological activity. Some reports suggest that partially deacetylated sugars can also enter the biosynthetic pathways [62]. Depending on the degree of deacetylation, these sugars may not be recognized by all biosynthetic enzymes for metabolism to cell surface glycans, thereby perhaps also contributing to the inhibitory and toxic effects of these compounds. Other studies addressed the conversion of fluorosugars to metabolic products, and ultimately to cell surface glycoconjugates. This section discusses the various fluorinated analogs that have been used in incorporation studies and the effects they had on the cell compared with their nonfluorinated counterparts. An analog of GlcNAc with a single fluorine atom on the N-acetamido group has been shown to

658

Goon and Bertozzi

be incorporated into glycosaminoglycans [63]. Glycosidase digestion of macromolecules isolated from rabbit tracheal explants that were incubated with radiolabeled N-fluoroacetyl-GlcN indicated the presence of these compounds within hyaluronic acid. These findings indicate that N-fluoroacetyl-GlcN traverses a pathway similar to GlcNAc and does not block the biosynthetic enzymes. Furthermore, cultivation of mouse embryo cells with N-fluoroacetyl-GlcN caused a 50% inhibition in the conversion of GlcNAc to UDP-GlcNAc, perhaps the result of alternate substrate competition [64]. Unlike the metabolism of the previous fluoro analogs, 4-fluoro-Man inhibits the synthesis of glycoproteins found on the vesicular stomatitis virus, probably through disruption of the biosynthesis of Glc3Man9GlcNAc2-PP-Dol [65]. At a concentration of 1 mM, 4-fluoro-Man specifically inhibited protein glycosylation of BHK21 cells infected with the virus. When the concentration was raised to 10 mM, general protein synthesis was also affected. It is possible that 4-fluoro-Man was incorporated into the oligosaccharide-Dol substrate and acted as a chain terminator, blocking further elaboration. Indeed, in Saccharomyces cerevisiae S288C, 4-fluoroMan is converted to the 1,6-bisphosphate, 6-phosphate, and GDP analogs [65]. Furthermore, 4-fluoro-Man was transferred by mannosyltransferase into the cell wall polysaccharide. The 3-fluoro analog, on the other hand, was transported into the cells at a lower rate and was not incorporated into cell wall glycoproteins. Although conversion of 3-fluoro-Man to its GDP derivative was much less efficient than the 4-fluoro analog, similar metabolites were isolated from the cells. 3-Fluoro-Glc is an inhibitor of glucose and galactose metabolism in yeast [67]. This analog also interferes with polysaccharide synthesis, as well as serving as a phosphate trap, as indicated by significant decreases in intracellular levels of phosphoglyceric acids, ATP, and UDP-Glc [68]. Exposure to 2-fluoro-Glc has a drastic effect on S. cerevisiae [69]. This toxic compound is a potent inhibitor of cell wall glucan biosynthesis and leads to cell wall degradation within a matter of hours. Similarly, Datema and Schwarz showed that 2-fluoro-Glc and 2-fluoro-Man inhibit formation of lipid-linked oligosaccharides in vivo [70]. 2-Fluoro-Man is a more potent antiviral agent that 2-fluoro-Glc in many virus– host cell systems [71]. Moreover, addition of exogenous Man, but not Glc, to culture media containing the fluorosugars, restored virus multiplication. A look at the metabolism of these analogs revealed that they are both activated by GTP or UTP, respectively, to yield their corresponding GDP or UDP derivatives [72]. 2-FluoroMan and 2-fluoro-Glc are incorporated into glycoproteins in both yeast and chicken embryo cells [71]. 2-Fluoro-Glc and 2-fluoro-Man were also found to be interconverted in vitro, presumably by epimerization. Furthermore, 2-fluoro-Glc appears to be converted to 2-fluoro-Gal. Therefore, these fluorinated sugars have multiple avenues for incorporation into glycoproteins. The metabolic fates of 2-fluoro-Gal have also received attention. Several groups have used 19F NMR spectroscopy and positron emission tomography (PET) with an 18 F label to track the substrate. These workers have determined that 2-fluoro-Gal is converted to 2-fluoro-Glc and also to phosphorylated and nucleotide-bound 2-fluoroGal analogs [51–53,71]. Thus, it is likely that 2-fluoro-Gal can be directly incorporated into glycoconjugates. 6-Fluoro derivatives of several hexoses have been investigated and found to have biological effects that significantly differed from the 2-fluoro derivatives men-

Metabolic Substrate Engineering

659

tioned earlier. In particular, 6-fluoro-Gal inhibited the incorporation of [3H]Gal, but not of Man, Glc, or leucine into L1210 leukemic cells. Electron microscope autoradiography was used to show that this analog is activated to form the nucleotide analog and then localized to the Golgi compartment [73,74]. Moreover, upon treatment with 6-fluoro-Gal, cells showed a reduction in available sites for external modification using an ␣2,6-sialyltransferase, providing evidence that the analog was incorporated into cell surface glycoconjugates. Because of its importance in tumor glycosylation and blood antigen structures, Fuc has attracted attention as a target for metabolic structural modification [75]. 2Fluoro-Fuc was shown to be incorporated into glycoproteins in mouse fibroblasts [76]. Specific localization of 6-fluoro-Fuc in the Golgi apparatus, nuclear membrane, plasma membrane, and cytoplasm of human mammary tumor cells indicated glycoprotein incorporation there as well [54]. Likewise, the metabolites GDP-6-fluoroFuc and 6-fluoro-Fuc-1-phosphate have been directly observed in these cells. C.

Deoxy Sugars as Inhibitors of Biosynthesis

The Glc and Man analog 2-deoxy-D-glucose has been shown to be a general inhibitor of cellular metabolism. Early work by Woodward et al. showed that this deoxy sugar is a strong inhibitor of both the growth of yeast cells and their fermentation of Glc [77]. It blocks glycoconjugate biosynthesis and has antiviral activity [39,78–83]. Since 2-deoxy-Glc uses the same transport system as Glc, it might simply block Glc uptake, depriving cells of a key metabolic substrate [84–87]. This has been directly demonstrated at high concentrations in yeast, chick fibroblasts, and Ehrlich ascites carcinoma cells [88,89]. However, at lower concentrations, 2-deoxy-Glc is taken up and converted to its UDP and GDP analogs. Incorporation into cell wall glucans in yeast and glycoconjugates in mammalian cells has been demonstrated [81,90–94]. Within glycoconjugates, 2-deoxy-Glc might serve as a chain terminator for further elongation. Studies of altered cell surface architecture induced by treatment of 2-deoxy-Glc have been performed [90]. Lectin binding studies suggest that 2deoxy-Glc actually replaces Man, not Glc. Agglutination of cells by the lectins Ricinus communis (RCA) and Phaseolus vulgaris (PHA) [78], both of which bind to Gal and GalNAc residues, is significantly decreased during incubation with 2-deoxyGlc. This result indicates that the underlying mannosyl residues are unable to support further chain elongation to include elaboration with Gal or GalNAc, residues necessary for binding to these lectins. Another possibility is that the underlying Man residues are simply missing. However, binding of concanavalin A (ConA) [78,95], a Man-specific lectin, increases on cells incubated with 2-deoxy-Glc, suggesting that Man residues are present, perhaps as terminal residues if the 2-deoxy modification prevents further elongation. Other studies provide more direct evidence that 2-deoxyGlc functions as a Man substitute rather than a Glc substitute in cells. The addition of low doses of Man, but not Glc, to cells grown with 2-deoxy-Glc can reverse the effects of this anti-metabolite [96,97]. Addition of Man to cells grown with 2-deoxyGlc restores natural lectin agglutinating properties, further confirming that the compound functions as a Man mimic. Similar to 2-deoxy-Glc, it has been established that 2-deoxy-Gal is incorporated into glycoproteins in both rat and human cells [98–100]. When rats to which 2deoxy-Gal had been administered were injected with [3H]Fuc, a significant decrease

660

Goon and Bertozzi

in Fuc incorporation was observed. This result suggested that the missing 2-OH group on Gal prevented 2-fucosylation, and this was confirmed by treatment with an ␣1,2-fucosidase. Interestingly, analysis with ␣1,3- and ␣1,4-fucosidases revealed a corresponding increase in ␣1,3/␣1,4-fucosylation. Incubation of AS-30D rat ascites hepatoma cells with 2-deoxy-[1-14C]Gal [101] or the administration of this compound to rats [102,103] resulted in a large accumulation of 2-deoxy-Gal-1-phosphate. This metabolite comprised 78% of the total radioactivity detected in the liver, with the remaining metabolites being 20% UDP-2-deoxy-Gal and 0.7% UDP-2-deoxy-Glc [104]. The formation of the uridylate intermediates occurred only after high levels of 2-deoxy-Gal-1-phosphate had been achieved. The accumulation of UDP-2-deoxyGal-1-phosphate was attributed to the limited epimerization of UDP-2-deoxy-Gal-1phosphate to UDP-2-deoxy-Glc-1-phosphate. This phenotype is characteristic of uridylyltransferase-deficient cells and tissues induced by Gal and may serve as an excellent model for studying the cellular injury in galactosemia. 2-Deoxy-Fuc, a residue found to be a constituent of several natural products, was tested for inhibition and cytotoxicity properties in L1210 leukemia, mouse mammary adenocarcinoma (TA3), and P288 leukemia cells (Table 7) [105]. This compound had no significant effect on the growth of any of the cell lines tested. However, the fully acetylated version of this substrate was active as a growth inhibitor, and it exhibited significant influence on glycoprotein biosynthesis in cultured P288 leukemia cells. The methyl glycoside of the fully acetylated 2-deoxy-Fuc had an even greater effect on cell viability, with an IC50 of 500 ␮M in L1210 leukemia cells; cell growth was inhibited to 43% of control cells. Furthermore, this compound decreased cellular incorporation of [3H]GlcN and [3H]leucine by 20 and 9%, respectively. In summary, much early work with subtly modified monosaccharide substrates suggested the capacity to alter cell surface glycosylation through metabolic processes.

Table 7

Growth Inhibitory Activity of 2-Deoxy Fuc Analogs Against L1210 Leukemia Cells

R1 OH OMe OMe OAc

R3

R4

IC50 (mM)

OH OH OAc OAc

OH OH OAc OAc

>1 >1 0.5 >1a

a IC50 value was determined using TA3 mammary cells.

Metabolic Substrate Engineering

661

Direct application of the technique to the study of cell surface carbohydrate recognition was realized later, mostly using sialic acids, as summarized in Section IV. IV.

INTRODUCTION TO SIALIC ACIDS

Sialic acid is a unique monosaccharide typically found at the nonreducing end of oligosaccharides. Sialic acids serve as modulators for cell adhesion [106–108], ligand determinants for specific viral and bacterial receptors, and epitopes for lectin binding [109,110]. They can also mask underlying antigenic sites [111,112]. The metastatic potential of tumor cells has been positively correlated with the total sialic acid content of the cell [113]. Their biological significance suggests that manipulation of the sialic acid content of complex oligosaccharides, as well as modification of the sialic acid structure, may provide insight into the molecular basis of cell surface interactions. The remainder of this chapter focuses on the biosynthesis of sialic acids and incorporation of unnatural analogs into cell surface glycoconjugates. Sialic acid is a generic term used to represent a family of over 30 neuraminic acid derivatives [114–116]. The most abundant sialic acids are N-acetylneuraminic acid (Neu5Ac) and N-glycolylneuraminic acid (Neu5Gc). These sialic acids, in addition to a few other commonly found sialic acids, are shown in Table 8. While both eukaryotes and prokaryotes possess similar sialic acids, there exists some variation in the biosynthetic pathway leading to the expression of these compounds on the cell surface. The biosynthetic pathway of Neu5Ac in mammals is shown in Figure 10. In mammalian cells, exogenous ManNAc and Neu5Ac can be taken up, metabolized through the sialic acid biosynthetic pathway, and then expressed on the cell surface as sialosides [117]. Thus, cell surface glycoconjugates could be altered if unnatural ManNAc or sialic acid analogs exogenously introduced to the cell were successfully converted into sialoglycoconjugates.

Table 8

Some Commonly Found Sialic Acids

Common name

Abbreviation

N-Acetylneuraminic acid

Neu5Ac

N-Glycolylneuraminic acid

Neu5Gc

9-O-Acetyl-N-acetylneuraminic acid

Neu5,9Ac2

2-Keto-3-deoxynononic acid

Kdn

Structure

662

Goon and Bertozzi

Metabolic Substrate Engineering

A.

663

The Unnatural Substrate Tolerance of the Enzymes Involved in Sialic Acid Biosynthesis

The possibility that sialic acid biosynthesis might be amenable to interception with unnatural substrates was first suggested by studies of the individual enzymes in the pathway. Several groups have analyzed the tolerance of various enzymes for unnatural substrates, revealing sites on metabolic intermediates at which functional group substitutions might be tolerated. Of the enzymes involved in sialic acid biosynthesis, the two that have attracted the most attention are CMP-Neu5Ac-synthetase and various sialyltransferases. Both bacterial and mammalian homologs from these two enzyme families have been characterized. CMP-Neu5Ac-synthetase catalyzes the condensation of free Neu5Ac with CTP to make CMP-Neu5Ac. The enzymes from bovine brain and rat liver have been investigated with respect to their activity with unnatural sialic acids [118]. Sialic acids modified at C5 and C9 have been extensively studied; several of these are accepted by the enzyme and are successfully converted into the corresponding CMP– sialic acid analogs in yields of 40–90% [119] (Table 9). The efficiency of the reaction depended on the analog. While the enzyme tolerates rather large substituents at C9, structural perturbation at C5 must be conservative. Examples of acceptable modifications are shown in Table 9 [119–124]. The 4-deoxy derivative of Neu5Ac is also a substrate for CMP-Neu5Ac synthetase [125]. The large variety of compounds that have been successfully converted into their CMP analogs indicate that bovine CMPNeu5Ac-synthetase might be useful for in vitro enzymatic synthesis of unnatural CMP–sialic acid analogs. In eukaryotes, sialyltransferases are located in the Golgi compartments [126,127]. They transfer a sialic acid residue from CMP–sialic acid to a nonreducing Gal-, GalNAc-, GlcNAc-, or sialic acid residue via an ␣-glycosidic linkage. The possible linkages include ␣2,3, ␣2,6, or ␣2,8, and each is the product of a different sialyltransferase. Sialyltransferases from rat and human liver have been studied extensively [128]. Similar to CMP-Neu5Ac-synthetase, the broad specificity of sialyltransferases has made them useful for the synthesis of neoglycoconjugates containing a variety of sialic acid analogs. Examples of CMP–sialic acid analogs that have been successfully converted into glycoconjugates by sialyltransferases are shown in Table 10 [118–120,125,128–131]. Even perturbations as large as a fluorescein group at C5 or C9 are tolerated by some sialyltransferases, allowing one to install fluorescent probes in cell surface glycoconjugates [132,133]. This approach has provided a method for kinetic analysis of the rates of different sialyltransferases within cells [134,135]. Furthermore, the C9 fluorescein derivative has been used to localize sialyltransferases to the Golgi compartments in rat liver [127]. Although alterations at C5 and C9 can reduce efficiency of sialyltransferases, the reactions occur at a reasonable rate and are therefore useful for studying cellular processes.

< Figure 10

The sialoside biosynthetic pathway in eukaryotic cells: A, UDP-GlcNAc-2epimerase; B, ManNAc-6-kinase; C, Neu5Ac-9-PO2⫺ synthase; D, Neu5Ac-9-PO2⫺ phospha4 4 tase; E, CMP-Neu5Ac-synthetase; F, sialyltransferase. Enzymes A and B function together as a bifunctional enzyme.

664

Goon and Bertozzi

Table 9

A Limited Number of Sialic Acid Derivatives Are Recognized as Substrates for CMP–Sialic Acid Synthetase from Bovine Brain and Rat Liver

R1 H

R2 Ac

R3 OH

OH

OH

OH

OH

OH

OH

OH OH

Ac Ac

OH

Ac

OH

Ac

OH

Ac

OH

Ac

OH

Ac

N3 NH⫹3

SHgMe

Sialic acid biosynthesis in bacteria is less complex than in mammals. There are only three enzymes involved in Neu5Ac biosynthesis; they are the Neu5Ac– synthase, CMP-Neu5Ac-synthetase, and sialyltransferase. Although these enzymes were cloned only within the past 10 years, their straightforward overexpression in bacteria has allowed extensive studies to be performed. It was not until 4 years ago that the first bacterial Neu5Ac-synthase was cloned [136]. This enzyme, derived from E. coli K1, is unlike the mammalian enzyme in that it does not accept the ManNAc-6-PO2⫺ as a substrate. Rather, it condenses 4 ManNAc with phosphoenolpyruvate (PEP) to produce Neu5Ac. Furthermore, 6-N3 ManNAc can be converted to 9-N3 Neu5Ac, although only with 59% efficiency compared to the native substrate [137]. The E. coli synthase has also been shown to accept ManN as a substrate, albeit with only a minimal production of the sialic acid

Metabolic Substrate Engineering

665

Table 10 A Limited Number of CMP–Sialic Acid Derivatives Are Accepted as Substrates by ␣2,3- and ␣2,6-Sialyltransferases from Bovine and Human Liver

R1 H OH OH

R2 Ac Ac Gc

R3 OH OAc OH

OH

OH

OH

OH

OH

OH

OH

OH

OH

OH

OH OH OH

Ac Ac Ac

OH

Ac

OH

Ac

OH

Ac

OH

Ac

F N3 NH⫹3

666

Table 10

Goon and Bertozzi Continued

R1

R2

OH

Ac

OH

Ac

R3

analog. This lack of specificity was also seen with the purified Neu5Ac-synthase from Neisseria meningitidis [138]. Since its initial identification in E. coli, the Neu5Ac-synthase gene has also been cloned from numerous microbial organisms including Helicobacter pylori. The availability of these enzymes promises to facilitate the enzymatic synthesis of sialic acid analogs. In contrast to the eukaryotic CMP-Neu5Ac-synthetases, the bacterial counterparts have not been extensively characterized. Therefore, it would be interesting to see whether the bacterial enzymes exhibit the same broad substrate specificity. Thus far, only 9-N3 and 9-NH⫹3 modified sialic acids have been tested with the E. coli and N. meningitidis CMP-Neu5Ac-synthetase [138,139]. While the 9-N3 analog was similar to Neu5Ac as a substrate for the enzyme, the 9-NH⫹3 derivative was a poor substrate. Unfortunately, many of the sialic acid analogs that were active as substrates for the mammalian enzyme have not yet been examined by means of bacterial enzymes. A few C5-modified analogs have been tested, namely, N-propanoyl sialic acid (SiaProp), N-glycolylneuraminic acid (Neu5Gc), and N-carbomethoxyneuraminic acid (Neu5CMe), and these were tolerated by the enzyme. 5-Azidoneuraminic acid was not a substrate for the enzyme, and N-carbobenzyloxyneuraminic acid (Neu5Cbz) coupled at a much slower rate than Neu5Ac [140]. While it would appear that the bacterial enzyme is more restrictive in its substrate specificity, it does show some tolerance for unnatural C5 and C9 substituted sialic acid derivatives. Sialylated glycoconjugates found in bacteria often mimic oligosaccharides found on the surface of mammalian cells [141–143]. The similarity in structure between the glycoconjugates of mammalian and bacterial cells might reflect a mechanism by which bacterial cells evade the host immune response. Given the similarity of bacterial and mammalian sialoglycoconjugate structures, the bacterial sialyltransferases might have substrate recognition properties similar to those of their mammalian counterparts. Unfortunately, few bacterial sialyltransferases have been cloned or characterized [144]. However, it is known that similar to mammalian sialyltransferases, bacterial sialyltransferases are membrane-associated proteins that are responsible for terminating glycans by transferring a sialic acid residue to a terminal

Metabolic Substrate Engineering

667

GlcNAc, GalNAc, or Gal residue. Like their mammalian counterparts, they too are capable of producing ␣2,3-, ␣2,6-, or ␣2,8-linked sialosides. An ␣2,6-sialyltransferase isolated from the bacterium Photobacterium damsela is capable of sialylating 3⬘-sialylLacNAc oligosaccharides, generating a rare 3⬘,6⬘-disialylated structure [145]. Bacterial sialyltransferases are also capable of forming ␣2,9-linked structures. The ␣2,8- and ␣2,9-linked sialic acids are generally found as homopolymers that make up the capsule of pathogenic bacteria like N. meningitidis, E. coli K1, and E. coli Bos 12. Of the few bacterial sialyltransferases cloned, even fewer have been studied with respect to donor–substrate specificity. CMP-Neu5Ac, CMP-Neu5Gc, CMPSiaProp, and CMP-9-fluoro-Neu5Ac have been tested as substrates for the N. meningitidis ␣2,3-sialyltransferase [146–148]. While these substrates were effectively recognized, the small panel does not provide enough information to generalize the specificity of the enzyme. If the enzyme is to be useful in chemoenzymatic synthesis or in studying cellular processes, knowledge of its specificity with a broader panel of substrates would be useful. B.

Metabolic Conversion of ManNAc Analogs to Cell Surface Sialosides

The notion that unnatural analogs of ManNAc might be metabolized to sialic acids in living cells was first explored in the 1970s. Initial studies of the sialic acid biosynthetic pathway focused on the inhibitory properties of metabolic precursors [62,149,150]. It was discovered that some ManNAc analogs, while exhibiting a degree of toxicity to the cells, competed with native ManNAc and were incorporated into the sialic acid biosynthetic pathway. This finding, along with the earlier knowledge that cells can take up exogenous ManNAc and convert it into sialic acid [117,151], opened up many possibilities for cell surface modification. Figure 11 shows the different ManNAc analogs that have been successfully metabolized by cells and converted into cell surface sialosides. For example, peracetylated N-trifluoroacetyl ManN, although toxic to cells at high micromolar concentrations, competed with ManNAc in glycoconjugate biosynthesis [62]. The expression of metabolically modified sialic acids on the cell surface has been used to modulate several biological processes. Schmidt et al. showed that Npropanoylmannosamine (ManProp), which is metabolized to SiaProp, stimulates proliferation in specific cells found in the rat central nervous system (CNS) [152]. Incubation of cells from neonatal rat brain with ManProp successfully stimulated proliferation of astrocytes and microglia, but not of oligodendrocyte progenitor cells in culture. The basis for this proliferation is unclear, since all cells examined converted ManProp into cell surface SiaProp. Similarly, human diploid fibroblast cells grown with ManProp, N-butanoylmannosamine (ManBut), or N-pentanoylmannosamine (ManPent) lost their sensitivity to contact inhibition of growth [153]. Virus–receptor interactions have also been shown to be affected by these ManN derivatives [154,155]. Treatment of human B-lymphoma BJA-B cells or African green monkey kidney epithelium cells with either ManProp, ManBut, or ManPent resulted in structural modification of about 50% of total cell surface sialic acids. Polyoma viruses, which use sialic acids as ligands for binding prior to infection, show either reduced or enhanced ability to infect cells carrying these modified sialic

668

ManNAc derivatives that are metabolically converted to the corresponding cell surface sialic acid analogs in mammalian cells.

Goon and Bertozzi

Figure 11

Metabolic Substrate Engineering

669

acids, depending on the ManN analog used. The expression of modified sialic acids might therefore be a new mechanism to prevent viral infection. An analog similar to ManPent, but containing a ketone group on the N-acyl side chain, has been synthesized as a sialic acid precursor [156]. This derivative, called N-levulinoylmannosamine (ManLev), is converted to the corresponding sialic acid (SiaLev), delivering reactive ketone groups to cell surface glycoconjugates. The biosynthesis of N-levulinoyl sialic acid has been demonstrated in a variety of human cell lines (e.g., Jurkat, HeLa, and HL-60 cells). Scatchard analysis of ManLev-treated cells indicated the presence of at least several million derivatized sialic acid residues on the cell surface [157]. The expression of SiaLev on the cell surface introduces an orthogonal reactive group, the ketone, into cell surface glycans. The ketone group, which can undergo a number of coupling reactions under physiological conditions, can be used to introduce new recognition epitopes onto the cell surface, a method called ‘‘glycoform remodeling’’ [157]. In this fashion, large biomolecules can be displayed on cell surfaces by using chemical methods rather than conventional genetics tools. For example, it has been demonstrated that aminooxy- and hydrazidefunctionalized oligosaccharides can be chemoselectively reacted with SiaLev to further elongate the oligosaccharide chain. Modified sialic acids bearing ketones have been exploited in new approaches to selective drug delivery [156] and for targeting of diagnostic agents [158]. Other functionalized N-acyl side chains on ManN, such as the N-azido acetyl group, are also tolerated by the biosynthetic enzymes [159]. Discovery of this property opens up new opportunities for modifying sialic acids and reacting them with exogenous agents. V.

CONCLUSION

The field of metabolic substrate engineering has evolved from initial cytotoxicity studies using unnatural sugars, which produced empirical observations, to cell surface engineering using chemically defined epitopes. The transformation of this rapidly growing field is the result of molecular cloning efforts of the past decades, which have enabled detailed studies of enzyme specificities. Moreover, analytical tools such as mass spectrometry and HPLC have advanced to the point that characterization of unnatural metabolites within complex glycoconjugates is now possible. With the availability of powerful genetic and analytic methods, it should be possible to apply metabolic substrate engineering more broadly to fundamental studies of glycobiology. ACKNOWLEDGMENTS The authors thank Bristol-Myers Squibb and the National Institutes of Health (R01 GM58867-01) for generous funding. REFERENCES 1. 2. 3. 4.

Varki, A. Glycobiology 1993, 3, 97–130. Rudd, P. M.; Dwek, R. A. Crit. Rev. Biochem. Mol. Biol. 1997, 32, 1–100. Peyrieras, N.; Bause, E.; Legler, G.; Vasilov, R.; Claesson, L.; Peterson, P.; Ploegh, H. EMBO J. 1983, 2, 823–832. Tulsiani, D. R. P.; Touster, O. J. Biol. Chem. 1983, 258, 7578–7585.

670

Goon and Bertozzi

5.

Pan, Y. T.; Hori, H.; Saul, R.; Sanford, B. A.; Molyneux, R. J.; Elbein, A. D. Biochemistry 1983, 22, 3975–3984. McDowell, W.; Schwarz, R. T. Biochimie 1988, 70, 1535–1549. Elbein, A. D. FASEB J. 1991, 5, 3055–3063. Schwarz, R. T.; Datema, R. Trends Biochem. Sci. 1980, 65–67. Srivastava, G.; Kaur, K. J.; Hindsgaul, O.; Palcic, M. M. J. Biol. Chem. 1992, 267, 22356–22361. Yamamoto, H.; Kaneko, Y.; Rebbaa, A.; Bremer, E. G.; Moskal, J. R. J. Neurochem. 1997, 68, 2566–2576. Prieto, P. A.; Mukerji, P.; Kelder, B.; Erney, R.; Gonzalez, D.; Yun, J. S.; Smith, D. F.; Moremen, K. W.; Nardelli, C.; Pierce, M.; Li, Y.; Chen, X.; Wagner, T. E.; Cummings, R. D.; Kopchick, J. J. J. Biol. Chem. 1995, 270, 29515–29519. Van den Steen, P.; Rudd, P. M.; Dwek, R. A.; Opdenakker, G. Crit. Rev. Biochem. Mol. Biol. 1998, 33, 151–208. Lis, H.; Sharon, N. Eur. J. Biochem. 1993, 218, 1–27. Candy, D. J. Complex Carbohydrate Polymers. New York: John Wiley, 1980, pp. 152– 170. Thomas, J. R.; Dwek, R. A.; Rademacher, T. W. Biochemistry 1990, 29, 5413–5422. Kornfeld, R.; Kornfeld, S. Annu. Rev. Biochem. 1985, 54, 631–664. Snider, M. D. In Biosynthesis of Glycoproteins: Formation of N-Linked Oligosaccharides, Vol. 2; V. Ginsburg and P. W. Robbins, eds. New York: John Wiley, 1984; pp. 163–288. Brady, R. O.; Fishman, P. H. Biochim. Biophys. Acta 1974, 355, 121–148. Lloyd, K. O.; Furukawa, K. Glycoconjugate J. 1998, 15, 627–636. Fishman, P. H. Chem. Phys. Lipids 1974, 13, 305–326. Roseman, S. Chem. Phys. Lipids 1970, 5, 270–297. Morell, P.; Braun, P. J. Lipid Res. 1972, 13, 293–310. Yeh, E. T. H.; Kamitani, T.; Ming Chang, H. Immunology 1994, 6, 73–80. Bill, R.; Revers, L.; Wilson, I. B. H. Protein Glycosylation. Boston: Kluwer Academic Publishers, 1998. Matsuura, F.; Ohta, M.; Ioannou, Y. A.; Desnick, R. J. Glycobiology 1998, 8, 329–339. Hounsell, E. F.; Davies, M. J.; Renouf, D. V. Glycoconjugate J. 1996, 13, 19–26. Savage, A. V.; Donoghue, C. M.; D’arcy, S. M.; Koeleman, C. A. M.; van den Eijnden, D. H. Eur. J. Biochem. 1990, 192, 427–432. Hounsell, E. F.; Lawson, A. M.; Feeney, J.; Gooi, H. C.; Pickering, N. J.; Stoll, M. S.; Lui, S. C.; Feizi, T. Eur. J. Biochem. 1985, 148, 367–377. Chai, W.; Hounsell, E. F.; Cashmore, G. C.; Rosankiewicz, J. R.; Bauer, C. J.; Feeney, J.; Feizi, T.; Lawson, A. M. Eur. J. Biochem. 1992, 203, 257–268. Kurosaka, A.; Nakajima, H.; Funakoshi, I.; Matsuyama, M.; Nagayo, T.; Yamashina, I. J. Biol. Chem. 1983, 258, 11594–11598. Wiegandt, H. In Glycolipids. A. Neuberger and L. L. M. van Deenen, eds., Vol. 10. Amsterdam: Elsevier Science Publishers, 1985. Jhon, G.-J.; Park, S.-Y.; Han, S.-Y.; Lee, S.; Kim, Y.; Chang, Y.-S. Chem. Pharm. Bull. 1999, 47, 123–127. Martin, A.; Rambal, C.; Berger, V.; Perier, S.; Louisot, P. Biochimie 1998, 80, 75–86. Kornfeld, S.; Kornfeld, R.; Neufeld, E. F.; O’Brien, P. J. Biochemistry 1964, 52, 371– 379. Yanagishita, M.; Salustri, A.; Hascall, V. Methods Enzymol. 1989, 179, 435–445. Varki, A. FASEB J. 1991, 5, 226–235. Bekesi, J. G.; Winzler, R. J. J. Biol. Chem. 1969, 244, 5663–5668. Keppler, D. O. R.; Rudigier, J. F. M.; Bischoff, E.; Decker, K. F. A. Eur. J. Biochem. 1970, 17, 246–253.

6. 7. 8. 9. 10. 11.

12. 13. 14. 15. 16. 17.

18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38.

Metabolic Substrate Engineering 39. 40. 41. 42. 43. 44. 45. 46. 47.

48. 49.

50. 51. 52. 53.

54.

55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68.

671

Kaluza, G.; Scholtissek, C.; Rott, R. J. Gen. Virol. 1972, 14, 251–259. Simon, P.; Burlingham, W. J.; Conklin, R.; Fondy, T. P. Cancer Res. 1979, 39, 3897– 3902. Fondy, T. P.; Emlich, C. A. J. Med. Chem. 1981, 24, 848–852. Fondy, T. P.; Roberts, S. B.; Tsiftsoglou, A. S.; Sartorelli, A. C. J. Med. Chem. 1978, 21, 1222–1225. Kent, P. W.; Ackers, J. P.; White, R. J. Biochem. J. 1970, 118, 73–79. Bernacki, R. J.; Sharma, M.; Porter, N. K.; Rustum, Y.; Paul, B.; Korytnyk, W. Supramol. Struct. 1977, 7, 235–250. Paul, B.; Bernacki, R. J.; Korytnyk, W. Carbohydr. Res. 1980, 80, 99–115. Paul, B.; Korytnyk, W. Carbohydr. Res. 1984, 126, 27–43. Bernacki, R. J.; Korytnyk, W. In Development of Membrane Sugar and Nucleotide Sugar Analogs as Potential Inhibitors or Modifiers of Cellular Glycoconjugates, M. I. Horowitz, ed., Vol. IV. New York: Academic Press, 1982, pp. 245–263. Elliott, K.; Birch, J. Carbon-Fluorine Compounds: Chemistry, Biochemistry and Biological Activities. Amsterdam: Elsevier Scientific Publishers, 1972. Barnett, J. E. G. Fluorine as a Substituent for Oxygen in Biological Systems: Examples in Mammalian Membrane Transport and Glycosidase Action. Amsterdam: Ciba Foundation Symposium; Associated Scientific Publishers, 1972, pp. 95–115. Kanazawa, Y.; Kuribayashi, S.; Kojima, M.; Haradahira, T. Chem. Pharm. Bull. 1988, 36, 4213–4216. Gru¨n, B. R.; Berger, U.; Oberdorfer, F.; Hull, W. E.; Ostertag, H.; Keppler, D. Adv. Enzyme Regul. 1990, 30, 231–242. Gru¨n, B. R.; Berger, U.; Oberdorfer, F.; Hull, W. E.; Ostergag, H.; Friedrich, E.; Lehmann, J.; Keppler, D. Eur. J. Biochem. 1990, 190, 11–19. Ishiwata, K.; Yamaguchi, K.; Kameyama, M.; Fukuda, H.; Tada, M.; Matsuzawa, T.; Muraishi, K.; Itoh, J.; Kawashima, K.; Takahashi, T.; Ido, T. Nucl. Med. Biol. 1989, 16, 247–254. Sufrin, J. R.; Bernacki, R. J.; Porter, C. W.; Korytnyk, W. Synthesis and metabolic effects of halogenated L-fucose and D-galactose analogs. In Glycoconjugates Research, Vol. II. New York: Academic Press, 1979, pp. 809–811. Sufrin, J. R.; Bernacki, R. J.; Morin, M. J.; Korytnyk, W. J. Med. Chem. 1980, 23, 143–149. May Jr., J. A.; Sartorelli, A. C. J. Med. Chem. 1979, 22, 971–976. Bessell, E. M.; Courtenay, V. D.; Foster, A. B.; Jones, M.; Westwood, J. H. Eur. J. Cancer 1973, 9, 463–470. Sharma, M.; Korytnyk, W. Carbohydr. Res. 1980, 83, 163–169. Sharma, M.; Bernacki, R. J.; Hillman, M. J.; Korytnyk, W. Carbohydr. Res. 1993, 240, 85–93. Sharma, M.; Bernacki, R. J.; Paul, B.; Korytnyk, W. Carbohydr. Res. 1990, 198, 205– 221. Sarkar, A. K.; Fritz, T. A.; Taylor, W. H.; Esko, J. D. Proc. Natl. Acad. Sci. USA 1995, 92, 3323–3327. Schwartz, E. L.; Hadfield, A. F.; Brown, A. E.; Sartorelli, A. C. Biochim. Biophys. Acta 1983, 762, 489–497. Winterbourne, D. J.; Barnaby, R. J.; Kent, P. W.; Mian, N. Biochem. J. 1979, 182, 707–716. Schultz, A. M.; Mora, P. T. Carbohydr. Res. 1975, 40, 119–127. Grier, T. J.; Rasmussen, J. R. J. Biol. Chem. 1984, 259, 1027–1030. Grier, T. J.; Rasmussen, J. R. Biochem. J. 1983, 209, 677–685. Woodward, B.; Taylor, N. F.; Brunt, R. V. Biochem. Pharmacol. 1971, 20, 1071–1077. Woodward, B.; Taylor, N. F.; Brunt, R. V. Biochem. J. 1969, 114, 445–447.

672

Goon and Bertozzi

69. 70. 71.

Biely, P.; Kovarik, J.; Bauer, S. J. Bacteriol. 1973, 115, 1108–1120. Datema, R.; Schwarz, R. T. Biochem. J. 1979, 184, 113–123. Schmidt, M. F. G.; Biely, P.; Kra´tky´, Z.; Schwarz, R. T. Eur. J. Biochem. 1978, 87, 55–68. Schmidt, M. F. G.; Schwarz, R. T.; Ludwig, H. J. Virol. 1976, 18, 819–823. Morin, M. J.; Porter, C. W.; Petrie III, C. R.; Korytnyk, W.; Bernacki, R. J. Biochem. Pharmacol. 1983, 32, 553–561. Porter, C. W.; Hui, S. W.; Bernack, R. J.; Korytnyk, W. J. Cell Biol. 1979, 83, 280a. Mullen, J. L.; Rosato, F. E.; Allen, T. R.; Miller, E. E.; Roseman, J.; Rosato, E. F. J. Surg. Oncol. 1973, 61–69. Winterbourne, D. J.; Butchard, C. G.; Kent, P. W. Biochem. Biophys. Res. Commun. 1979, 87, 989–992. Cramer, F. B.; Woodward, G. E. J. Franklin Inst. 1952, 253, 354–360. Hughes, R. C.; Meager, A.; Nairn, R. Eur. J. Biochem. 1977, 72, 265–273. Klenk, H.-K.; Scholtissek, C.; Rott, R. Virology 1972, 49, 723–734. Schwarz, R. T.; Klenk, H.-D. J. Virol. 1974, 14, 1023–1034. Steiner, S.; Courtney, R. J.; Melnick, J. L. Cancer Res. 1973, 33, 2402–2407. Courtney, R. J.; Steiner, S. M.; Benyesh-Melnick, M. Virology 1973, 52, 447–455. Kilbourne, E. D. Nature 1959, 183, 271–272. Schmidt, M. F. G.; Schwarz, R. T.; Scholtissek, C. Eur. J. Biochem. 1974, 49, 237– 247. Wick, A. N.; Drury, D. R.; Nakada, H. I.; Wolfe, J. B. J. Biol. Chem. 1957, 224, 963– 969. Barban, S.; Schulze, H. O. J. Biol. Chem. 1961, 236, 1887–1890. Schwarz, R. T.; Schmidt, M. F. G.; Lehle, L. Eur. J. Biochem. 1978, 85, 163–172. Steiner, S.; Lester, R. L. J. Bacteriol. 1972, 109, 81–88. Letnansky, K. Biochim. Biophys. Acta 1964, 87, 1–8. Myers, M. W.; Sartorelli, A. C. Biochem. Biophys. Res. Commun. 1975, 63, 164–171. Meager, A.; Nairn, R.; Hughes, R. C. Eur. J. Biochem. 1977, 72, 275–281. Biely, P.; Kra´tky´, Z.; Bauer, S. Biochim. Biophys. Acta 1972, 255, 631–639. Steiner, S.; Steiner, M. R. Biochim. Biophys. Acta 1973, 296, 403–410. Steiner, M. R.; Somers, K.; Steiner, S. Biochem. Biophys. Res. Commun. 1974, 61, 795–801. Steiner, S.; Altenburg, B.; Melnick, J. L. J. Natl. Cancer Inst. 1974, 52, 617–620. Havell, E. A.; Vilcek, J.; Falcoff, E.; Berman, B. Virology 1975, 63, 475–483. Kaluza, G.; Schmidt, M. F. G.; Scholtissek, C. Virology 1973, 54, 179–189. Bu¨chsel, R.; Hassels-Vischer, B.; Tauber, R.; Reutter, W. Eur. J. Biochem. 1980, 111, 445–453. Geilen, C. C.; Kannicht, C.; Orthen, B.; Heidrich, C.; Paul, C.; Grunow, D.; Nuck, R.; Reutter, W. Arch. Biochem. Biophys. 1992, 296, 108–114. Bu¨chsel, R.; Reutter, W. Hoppe-Seyler’s Z. Physiol. Chem. 1978, 359, 1066–1077. Smith, D. F.; Keppler, D. O. R. Eur. J. Biochem. 1977, 73, 83–92. Starling, J. J.; Keppler, D. O. R. J. Neurochem. 1977, 29, 603–604. Starling, J. J.; Keppler, D. O. R. Eur. J. Biochem. 1977, 80, 373–379. Starling, J. J.; Keppler, D. Hoppe-Seyler’s Z. Physiol. Chem. 1977, 358, 311–312. Korytnyk, W.; Sufrin, J. R.; Bernacki, R. J. Carbohydr. Res. 1982, 103, 170–175. Berwick, L.; Coman, D. R. Cancer Res. 1962, 22, 982–987. Weiss, L. Expl. Cell Res. 1963, 30, 509–520. Kemp, R. B. J. Cell Sci. 1970, 6, 751–766. Wiley, D. C.; Skehel, J. J. Annu. Rev. Biochem. 1987, 56, 365–394. Traving, C.; Schauer, R. Cell Mol. Life Sci. 1998, 54, 1330–1349. Bagshawe, K. D.; Currie, G. A. Nature 1968, 218, 1254–1255.

72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111.

Metabolic Substrate Engineering 112. 113. 114. 115.

116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139.

140. 141. 142. 143. 144.

673

Rogentine, J., G. Nicholas; Plocinik, B. A. J. Immunol. 1974, 113, 848–858. Yogeeswaran, G.; Salk, P. L. Science 1981, 212, 1514–1516. Reuter, G.; Schauer, R. Glycoconjugate J. 1988, 5, 133–135. Schauer, R.; Kelm, S.; Reuter, G.; Roggentin, P.; Shaw, L. Biochemistry and role of sialic acids. In Biology of Sialic Acids, A. Rosenberg, ed. New York: Plenum Press, 1995, pp. 7–67. Varki, A. Glycobiology 1992, 2, 25–40. Harms, E.; Kreisel, W.; Morris, H. P.; Reutter, W. Eur. J. Biochem. 1973, 32, 254–262. Conradt, H. S.; Bu¨nsch, A.; Brossmer, R. FEBS Lett. 1984, 170, 295–300. Brossmer, R.; Gross, H. J. Methods Enzymol. 1994, 247, 153–176. Gross, H. J.; Bu¨nsch, A.; Paulson, J. C.; Brossmer, R. Eur. J. Biochem. 1987, 168, 595–602. Kosa, R. E.; Brossmer, R.; Gross, H.-J. Biochem. Biophys. Res. Commun. 1993, 190, 914–920. Martin, R.; Witte, K. L.; Wong, C.-H. Bioorg. Med. Chem. 1998, 6, 1283–1292. Herrler, G.; Gross, H.-J.; Brossmer, R. Biochem. Biophys. Res. Commun. 1995, 216, 821–827. Herrler, G.; Gross, H.-J.; Imhof, A.; Brossmer, R.; Milks, G.; Paulson, J. C. J. Biol. Chem. 1992, 267, 12501–12505. Gross, H. J.; Brossmer, R. Glycoconjugate J. 1987, 4, 145–156. Brossmer, R.; Gross, H. J. Methods Enzymol. 1994, 247, 177–193. Chammas, R.; McCaffery, M.; Klein, A.; Ito, Y.; Saucan, L.; Palade, G.; Farquhar, M. G.; Varki, A. Mol. Biol. Cell 1996, 7, 1691–1707. Sticher, U.; Gross, H. J.; Brossmer, R. Glycoconjugate J. 1991, 8, 45–54. Gross, H. J.; Brossmer, R. Glycoconjugate J. 1995, 12, 739–746. Kelm, S.; Schauer, R.; Manuguerra, J.-C.; Gross, H.-J.; Crocker, P. R. Glycoconjugate J. 1994, 11, 576–585. Gross, H. J.; Rose, U.; Krause, J. M.; Paulson, J. C.; Schmid, K.; Feeney, R. E.; Brossmer, R. Biochemistry 1989, 28, 7386–7392. Gross, H. J. Eur. J. Biochem. 1992, 203, 269–275. Gross, H. J.; Brossmer, R. Eur. J. Biochem. 1988, 177, 583–589. Gross, H. J.; Sticher, U.; Brossmer, R. Anal. Biochem. 1990, 186, 127–134. Gross, H. J.; Brossmer, R. Clin. Chim. Acta 1991, 197, 237–248. Annunziato, P. W.; Wright, L. F.; Vann, W. F.; Silver, R. P. J. Bacteriol. 1995, 177, 312–319. Vann, W. F.; Tavarez, J. J.; Crowley, J.; Vimr, E.; Silver, R. P. Glycobiology 1997, 7, 697–701. Brossmer, R.; Rose, U.; Kasper, D.; Smith, T. L.; Grasmuk, H.; Unger, F. M. Biochim. Biophys. Res. Commun. 1980, 96, 1282–1289. Brossmer, R.; Rose, U.; Unger, F. M.; Grasmuk, H. Enzymatic synthesis of cytidine 5⬘-monophospho-9-azido-9-deoxy-N-acetylneuraminic acid. Studies with sialyltransferase. In Glycoconjugates, R. Schauer, P. Boer, E. Buddecke, M. F. Kramer, J. F. G. Vliegenthart, and H. Wiegandt, eds. Stuttgart: Georg Thieme, 1979, pp. 242–243. Shames, S. L.; Simon, E. S.; Christopher, C. W.; Schmid, W.; Whitesides, G. M.; Yang, L.-L. Glycobiology 1991, 1, 187–191. Mandrell, R. E.; McLaughlin, R.; Kwaik, Y. A.; Lesse, A.; Yamasaki, R.; Gibson, B.; Spinola, S. M.; Apicella, M. A. Infect. Immun. 1992, 60, 1322–1328. Mandrell, R. E.; Griffiss, J. M.; Macher, B. A. J. Exp. Med. 1988, 168, 107–126. Moran, A. P.; Prendergast, M. M.; Appelmelk, B. J. FEMS Immunol. Med. Microbiol. 1996, 16, 105–115. Gilbert, M.; Watson, D. C.; Cunningham, A.-M.; Jennings, M. P.; Martin Young, N.; Wakarchuk, W. W. J. Biol. Chem. 1996, 271, 28271–28276.

674

Goon and Bertozzi

145.

Kajihara, Y.; Yamamoto, T.; Nagae, H.; Nakashizuka, M.; Sakakibara, T.; Terada, I. J. Org. Chem. 1996, 61, 8632–8635. Gilbert, M.; Cunningham, A.-M.; Watson, D. C.; Martin, A.; Richards, J. C.; Wakarchuk, W. W. Eur. J. Biochem. 1997, 249, 187–194. Gilbert, M.; Bayer, R.; Cunningham, A.-M.; DeFrees, S.; Gao, Y.; Watson, D. C.; Young, N. M.; Wakarchuk, W. W. Nat. Biotechnol. 1998, 16, 769–772. Miyazaki, T.; Sakakibara, T.; Sato, H.; Kajihara, Y. J. Am. Chem. Soc. 1999, 121, 1411– 1412. Gru¨nholz, H.-J.; Harms, E.; Opetz, M.; Reutter, W. Carbohydr. Res. 1981, 96, 259– 270. Hadfield, A. F.; Mella, S. L.; Sartorelli, A. C. J. Pharm. Sci. 1983, 72, 748–751. Monaco, F.; Robbins, J. J. Biol. Chem. 1973, 248, 2072–2077. Schmidt, C.; Stehling, P.; Schnitzer, J.; Reutter, W., Horstkorte, R. J. Biol. Chem. 1998, 273, 19146–19152. Raimund Wieser, J.; Heisner, A.; Stehling, P.; Oesch, F.; Reutter, W. FEBS Lett. 1996, 395, 170–173. Keppler, O. T.; Stehling, P.; Herrmann, M.; Kayser, H.; Grunow, D.; Reutter, W.; Pawlita, M. J. Biol. Chem. 1995, 270, 1308–1314. Kayser, H.; Zeitler, R.; Kannicht, C.; Grunow, D.; Nuck, R.; Reutter, W. J. Biol. Chem. 1992, 267, 16934–16938. Mahal, L. K.; Yarema, K. J.; Bertozzi, C. R. Science 1997, 276, 1125–1128. Yarema, K. J.; Mahal, L. K.; Bruehl, R. E.; Rodriguez, E. C.; Bertozzi, C. R. J. Biol. Chem. 1998, 273, 31168–31179. Lemieux, G. A.; Yarema, K. J.; Jacobs, C. L.; Bertozzi, C. R. J. Am. Chem. Soc. 1999, 121, 4278–4279. Saxon, E.; Bertozzi, C. R. Science 2000, 287, 2007–2010.

146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159.

Index

ABH, 355 Acarbose, 415 Acarviosins, 415 Acetylation, 440 Acmimycin, 314 Acrylamide copolymerization, 289 Actinomycete, 134, 309 Acyloxonium ion, 147 Adam’s catalyst, 599 Adenine, 549 Aglycon, 308 AIBN (2,2⬘-azobisisobutyronitrile), 78 Alanine, 494, 495 Aliphatic aglycone, 560 Alkyl glycosides, 122 Allium, 163, 171 Allyl C-glycoside, 109 Allylic sulfones (radical acceptor), 80 Allylic ␣-C-glycoside, 88 ␣-amino acids, 494, 495 Amikacin, 354 Amino deoxyglucuronic acids, 511 glyculosonic acids, 495 hexuronic acids, 494, 495 kanamycins, 393 nonulosonic acids, 495 sugar, 593, 652 Aminocyclitol antibiotic, 307 Aminoglycoside, 307, 353 acetyltransferase, 325 adenyltransferase, 400 nucleotidyltransferase, 325 phosphotransferase, 325

Aminoglycoside-modifying enzyme, 365 Angucycline antibiotics, 125 1,2-Anhydro sugar, 5, 8, 83, 112 Anomeric effect, 72, 359 Anomeric sulfoxides, 14 Anthelmintics, 310 Antiadhesins, 278 Antibiotic therapy, 134 Anticancer agents, 165 Anticoagulant, 427, 482 Antifungal activity, 172 Anti-Gal IgG, 583 Anti-Gal IgM, 583 Anti-HIV agent, 98 Anti-inflammatory activity, 168 Anti-MRSA, 393 Antisense oligonucleotide-saccharide conjugates, 249 Anti-sialic acid antibodies, 289 Antithrombotic, 427, 482 Antitumor activity, 172, 174 Antiviral agents, 427 Antivirus antibodies, 256 Aparamycin 309, 333, 361, 415 Arabinan chain, 135 Arabinogalactan (AG), 134 Arabinopyranoside diol, 166 Arabinose, 133, 538, 542, 543 Arabinosyltransferases (AraT’s), 140 Arabinoxylans, 133 Aryl C-glycosides, 123 Arylsulfonamides, 428, 442 Ashimycin, 314 Asparaginine, 494 675

676 Asparagus, 163 Aspergillus oryzae, 577 Auxiliary-directed glycosylation, 194 Azide, 428 Azido glucuronic acid, 511 Azidonitration, 428, 461 Azoisobutylnitrile (AIBN), 443 Baby hamster kidney, 543 Bacillus circulans, 323, 390 Bacterial lipopolysaccharide, 630 Balanites aegyptica, 174 Barbituric acid, 122 Barton’s deoxygenation, 378 Bauhinia purpurea lectin, 243 BBA2, 266 Benzannulation, 102 Benzenesulfenyl triflate, 65 Benzenesulfonamide, 428 Benzoyl cyanide, 462 Benzoylation, 144 Bidirectional glycosylation strategy, 24 in solid phase synthesis, 24 Bifunctional enzyme, 588 Biomodified guar, 577 Biosynthesis of glycoconjugates, 640 Biotransformation of Polysaccharide, 567, 569 Blood group determinants, 5, 10 Boric acid, 470 Boron trifluoride etherate, 434 Bovine serum albumin (BSA), 266 Brookfield viscometer, 577 Butirosin, 333, 376, 380 Camphorsulfonic acid, 474 CAN, 443 Cancer diagnostic kit, 289 Cancer metastasis, 277 Caproamide, 526, 527 Capsular polysaccharides, 285, 495 Carba, 417 Carbenicillin, 391 Carbodiimide, 45 Carbohydrate drugs, 175 Carbohydrate-based nucleic acids, 518 therapeutics, 535 Carbohydrate-protein binding interactions, 278 Carbon-linked disaccharides, 106 hetero-Diels–Alder approach, 106

Index Carbonucleotoids, 513 Carbopeptoid, 513 Cardioactive properties, 174 Catalytic hydrogenation, 454, 470 CD spectra, 528–530 Cell adhesion, 5 Cerium(IV) ammonium nitrate, 480 Chelate effect, 228 Chinese herbal medicine, 172, 174 Cholera toxin B, 298 Cholestane saponins, 165 Chondroitin trisaccharides, 458 Clonezyme, 588, 596 Cluster effect, 227 CMP-NeuAc regeneration, 210 CMP-Neu5Ac, 298, 537 synthetase, 663–666 CMP-NeuAc, 199, 201 synthesis of, 202–204 CMP-sialic acid, 537 C-NMR Spectroscopy, 28 C-nucleoside synthesis, 98 Cobalt, 100 in C-glycoside synthesis, 100 Cofactor regeneration, 210 Coiled coils, 531 Collocalia mucin, 279 Combimicins, 382 Combinatorial synthesis, 243 Computer-assisted analysis, 519 ConA, 223, 239, 293 ConA-saccharide interactions, 236 Condensation reaction, 88 Cookson’s reagent, 106 Costacea, 172 Coulombic interactions, 224, 249 CPG (controlled pore glass) solid support, 16 Cross-Coupling reactions, 97 C-type lectin recognition, 223 Cyclization, 92 mercury-induced, 91, 92 Cycloaddition reaction, 104, 106 in C-glycoside synthesis, 104 Cyclodextrin, 555 Cytidine monophosphate, 537 Cytidine triphosphate (CTP), 202 Dactimicin, 322, 409 Danishefsky’s iodosulfonamidation method, 428

Index DBU (1,8-diazabicyclo[5.4.0]-7-undecene), 80, 443 De Novo Ring formation, 106 Deallylation, 480 Deamino kanamycin, 394 Dehydrative glycosylation, 36 via glycosyl bromides, 36 Dehydroisoandrosterone, 164–168 De-N-Acetyl GM3, 179 Density functional theory, 158 Deoxy sugar, 652, 659 DEPT, 600 Dermatan sulfate hexasaccharide, 474 Dermatan sulfate, 425, 467 Destomycin A, 310 Destomycin C, 372 D-glucuronic acid (GlcUA), 431, 443 Dianion approach, 83 Diazotization, 373 Dibekacin, 382, 390 Dibutyltin oxide, 448 Diels–Alder reaction, 104 Digitalis purpurea, 164 Dihydrostreptomycin, 314 Dioclea grandiflora lectin, 248 binding specificity, 246 Dioscorea, 163, 171, 172 Diosgenin, 164, 168, 169, 172 Diosgenyl saponins, 171, 172 DMTST (dimethylthiomethylsulfonium triflate), 19 Dodesaccharide, 19 block synthesis, 19 DQF-COSY, 600 DREG56 (anti-L-selectin antibody), 267 Drug regimens, 134 Drug-RNA interaction, 342 DTBMP (2,6-di-tert-butyl-4-methylpyridine), 59 DTBP (di-tert-butylpyridine), 11, 14 E. coli, 577 heat-liabile enterotoxin, 298 ELISA (enzyme-linked immunosorbant assay), 232, 242, 268, 610, 614 Endocytosis, 247, 252 Endothelial cells, 260 Enkephalins, 511 Enterobacteriaceae, 353 Enzymatic synthesis, 634 Esterase, 570 Eukaryote, 641

677 Exoanomeric effect, 153 Ezoaminuroic acid, 495 Ezomycin A1, 495, 498 Facial selectivity, 120 of C-glycosidation, 120 Fast-atom bombardment high-resolution mass spectrometry, 135 Fibringen, 531 Fibrosarcoma HT1080, 252 Fibrous proteins, 531 Fimbriae, 603 Fischer glycosylation, 35, 144, 146 Fischer protocol, 34 Flow cytometry analysis, 620 Flu virus hemagglutination, 293 inhibitors of, 285, 299 Fluorescence anisotropy, 231 Fluorinated netilmicin, 387 Fluorinated sporaricin A, 412 Fluorination-toxicity relationship, 391 Fluoro sugar, 386 Fluoroarbekacin, 387 Fluorohexosamine, 655 Fluorohexose, 655 Foldamer, 519 Formimidation, 409 Fortimicin, 317, 360 Fraser–Reid pentenyl glycosylation technique, 428 Fucosyltransferase, 543, 544 Functional affinity, 242 of ConA, 242 Furanose donor, 45 Galactosaminoglycans, 425 Galactose, 587 galactose-containing cellobiose, 577 metabolism, 587 Galactosemia, 660 Galactosidase, 577 Galactosyltransferase, 537, 540, 544, 584, 626 GalNAc, 540, 542 GalNAc-Ser, 494 Ganglioside, 10, 179, 644 Garamine, 30, 400 GDP-Fuc, 298, 537 inhibitor, 100 GDP-Man, 537 GEGOP, 602 Gene-saccharide conjugates, 249

678 Genetic engineering, 588 Gentamicin, 312, 333, 382, 398 GFR, 346 GlcNAc, 540, 541, 654 GlcNAc-Asn, 494 Glebomycin, 314 Gliosis, 344 Globotriose, 541, 542, 634 Glucotransferase, 626 Glucuronic acid, 434, 540 GlyCAM-1, 267 Glycocalix[n]arenes, 286 Glycoconjugate, 1, 4, 140, 560, 593 Glycodendrimer, 292, 293, 298, 299 Glycoform remodeling, 669 Glycolipid, 5, 177, 540, 549, 560, 644 Glycomimetics, 278 Glycopeptides, 2, 494 Glycopeptidomimetics, 283 Glycopolymer, 298, 603 Glycoproteins, 140, 177, 540, 549, 658 biosynthesis of, 653 Glycosaminoglycans (GAGs), 425, 479, 658 Glycosidase, 324, 588, 632 Glycosyl donor, 33 halide coupling, 35 halides, 114 electrophilic coupling reagents, 114 phosphite, 199 sulfonate, 37, 53 trichloroacetimidate, 15, 165 triflate, 39, 53, 55 Glycosylation, 181 enzyme-mediated, 199 with sialic acid, 181 Glycosyltransferase, 299, 535–538, 552, 555, 559, 587, 625, 640 Glycosyltransferase-catalyzed synthesis, 537 GM3-type copolymer, 290 Golgi apparatus, 584 Golgi compartment, 659 Gougerotin, 495, 498 Grignard reaction, 98 Grubb’s catalyst, 289 GS4104, 278 GTA, 542, 543 GTB, 542, 543 Guanidino, 314, 404 Guanine, 549 Guanosine diphosphate, 537 Guar, 571, 574, 578

Index HA, 256, 257 inhibitors of, 254 Halokanamycins, 384 Hanganutziu–Deicher antigen, 279 HAR, 582, 583 HDV, 343 Hecogenin, 164 Helicobacter pylori, 8, 540 Hemagglutination, 234, 259, 260 Hemagglutinin(HA), 253 Hemisalt, 397 Heparin, 226, 267, 425, 482 Hepatocyte lectin, 248 Hepatocytes, 249, 252 HepG2 cell, 266 Heptonic amino acid, 503 Heteomultimer, 247 Heterocoupling, 38 Hexasaccharide, 135 Hexosamine, 655 Hexosaminuronic acid, 509 High-density lipoprotein (HDL), 252 High-performance liquid chromatography, 559 (see also HPLC) High-throughput assays, 266 HIV replication, inhibitor of, 516 HL-60 cells, 266 HMQC, 600 Horseradish peroxidase, 610 HPLC, 559, 592 HRP, 611 HUVECs (human umbilical vein endothelial cells), 266 Hyaluronan, 425, 461 Hyaluronic acid, 425, 462 Hydroxyproline-rich glycoproteins (HPRGs), 133 Hygromycin, 310, 333 Hyperacute rejection, 582 IC50, 618 IEC, 592 Immunoadsorption, 584 Immunomodulatory, 139 Immunoreactive oliogosaccharide, 638 Immunosuppressive drug, 582 Immunosuppressor, 620 INAPT, 600 Inflammatory response, 425 Influenza A, 299 Influenza flu virus neuraminidase inhibitor, 278

Index Influenza virus hemagglutinin, 242, 248 Influenza virus, 253, 259 Inosamycin, 315, 321 Iodohydin, 381 Ion exchange chromatography, 560, 592 Isepamicin, 354 Isoniazid, 134 Istamycin, 317, 410, 411 Jones oxidation, 442 Kakelokelose, 427 Kanamycin, 312, 324, 354, 380 Karplus equation, 600 Kasugamycin, 315 Keratan sulfate, 425 Keratin, 531 Ketomycolate, 135 Klebsiella pneumoniae, 379 Ko¨enigs–Knorr glycosylation, 432, 604, 610 Kolbe electrolysis, 80 of sugar acids, 80 LacNAc, 538 LAM, 139 Lectin, 603 Lectin-ligand interaction, 210, 227 Leloir glycosyltransferase, 537 Leloir pathway glycosyltransferases, 199 Leloir types, 537 Leukocyte rolling, 267 Levoglucosenone, 79 Levulinoyl, 436 Lewis acid catalysis, 109 mediated diarylation, 125 monoarylation, 123 Lewis acids, 34 Lewis blood group antigen (Le b), 8 Ligand proteolysis, 252 Lindlar’s catalyst, 458 Lipase, 570, 571, 574 Lipoarabinomannan (LAM), 134 Lipopolysaccharide, 625 Liposomes, 249, 257, 284 Liriope muscari, 168 Lividomycin, 310 LOS, 632 Low-density lipoprotein(LDL), 252 Luciferase(Luc) gene, 249 Lukemic HL60, 252 Lymphocyte trafficking, 425

679 Lysine, 526 Lysinomicin, 319, 413 Lysoganglioside, 290 Lysosome, 345 Lysyllysinomocin, 319 Maidong, 168 Maidong saponin C, 168–171 Major histocompatibility complex (MHC), 140 MALDI, 559 ManLAM, 135 ManNAc, 661 Mannosyl triflates, 65 ManPent, 669 ManProp, 667 Matrix-assisted laser desorption ionizaton mass spectrometry, 559 MBP (mannose binding protein), 299 Merrifield’s resin, 14 Metallo-ene cyclization, 98 Metallokanamycin A, 397 Methoxyethoxymethyl chloride (MEMCl), 453 Methylacarviosin, 415 Methylation analysis, 135 Methylation, 542 Michael addition, 91 Microbial glycosyltransferase, 630 Micrococcus lysodeikticus, 494 Micromonospora, 381 Millex-GV filter, 561 Mitochondrial mutation, 343 Mitsunobu protocol, 39 reaction, 41 MM3, 158 Modified CMP-NeuAc sialyl donors, 210 Moffatt–Swern oxidation, 43 Monosaccharide, 569 Mycobacterium smegmatis, 135 Myomucin, 315 Myxoma virus, 551 N-acetylation, 436 Neamine, 309 Neighboring hydroxyl group participation, 356 Neisseria meningitidis, 279, 540 Neisserial lipo-oligosaccharide, 631 Neoglycoconjugates, 278, 292, 300 Neomycin, 310, 315 Nephrotoxicity, 345, 346

680 Netilmicin, 354, 387 Neu5Ac2en, 278 Neu5Ac-synthetase, 666 NeuAc glycal, 194 Neuraminidase, 256 inhibitor of, 256 Nickel dichloride hexahydrate, 470 Nitroglycal, 106 NMDA, 344 NMR spectroscopy, 27, 135, 559 Nojirimycin, 3222 Non-Leloir types, 537 Nuclear Overhauser effect (NOE), 153, 601 Nucleoside antibiotics, 495, 509 Nucleotide, 559 O2 sugars, 85 O-glycosylation, 644 Olefin self-metathesis reactions, 289 Oligobiosaminide, 415 Oligofuranoside, 155 Oligonucleotides, 2 Oligopepoids, 284 Oligopeptides, 2 Oligosaccharide, 1, 4, 143, 425, 535, 536, 560, 634 preparation of, 634 solid support synthesis, 19 synthesis of, 15, 24 Oncogenesis, 542 One-pot glycosylation, 174 Ophiopagon, 163, 168, 171, 172 Ornithogalum saudersia, 165 Otovestibular, 307 Oxazolidinone, 401 Oxirane, 406 Oxyapramycin, 309, 369, 370 Palladium, 97 in coupling reactions, 97 Palladium(II) acetate, 110 PAMAM, 283, 293 Paromamine, 333, 359 Paromomycin, 310, 315, 329 Pea lectin, 293 Pearlman’s catalyst, 445 Pectin lyase, 570 Pectin, 133, 570, 576 methylesterase, 570 Pentasaccharide, 592 Pentenyl galactose donor, 24 Peptidases, 530

Index Peptide coupling, 512 Peptidoglycan synthesis, 638 Peptidomimetics, 511 Perfluoroalkylation reaction, 170 Pfitzer–Moffatt procedure, 393 Phase transfer catalysis (PTC), 280 Phenolic nucleophiles, 46 Phenyl glycosides, 41 Phenylalanine, 526 Phenylsulfide auxiliary, 194 Phosphatidylcholine, 257 Phosphatidylinositol, 135 Phosphoinositol-glyceroarabino-mannan, 135 Phospholipid, 345 Photobacterium damsela, 552 Phthalamide, 428 Plasma membrane, 345, 644 PME, 570 pNAS, 613 Polyelectrolyte effect, 231 Polygalacturonase, 570 Polygonatum, 163 Polylactose, 613 Polylysine-asialoglycoprotein conjugates, 249 Polymeric hyaluronan, 445 Polyphyllin D, 172 Polyprenyl sugar phosphate, 140 Polysaccharide, 1, 133, 134, 567, 569, 570 Progesterone, 164 Propionibacterium acnes, 495 Proteases, 530 Proteoglycans (PGs), 425, 448 Prototypical glycoconjugates, 641 Prumycin, 323 Pseudomonas aeruginosa, 310, 353, 377, 381 PSEUROT 6.2, 155 PS-PEG (polystyrene-polyethylene glycol) copolymer, 14 PTC, 283 Purine, 549 Pyranosiduronic acids, 494 Pyranosyl sugar amino acids, 493 Quinic acid, 363, 401 Radical addition, 79 allylation, 79 cyclization, 81, 86 deoxygenation, 88

Index Reductive amination, 393 deamination, 365 Regiospecific acylation, 383 Renal dysfunction, 345 Reticulum, 584 Rev-RRE, 335 Rhodium catalyst, 480 Ribostamycin, 312, 315 Rifampicin, 134 Rink resin, 516, 527 ROMP (ring-opening metathesis polymerization), 290 Ruscus, 163 Saccharosin, 309, 371 Salbastatin, 325, 416 Salmonella typhosa, 494 Sannamycin, 319, 413 Saponification, 445 Saponin OSW-1, 165 Saponins, 133, 163 Sarsasaponins, 38 Schmidit glycosylation, 169 Schrock catalyst, 102 Schuerch’s hypothesis, 63 SEC, 592 Seldomycin, 313, 318 Selectin antagonists, 285 inhibitors, 265, 267 Selectins, 260, 298 Selective allylation, 123 Sendai viruses, 299 Serine protease, 482 Serine, 494, 495 Sialglycoconjugate, 661 Sialic acid, 177, 178, 253, 259, 277, 289, 495, 549, 551, 652, 661, 669 biosynthesis, 663, 664 synthesis of, 177 Sialidases, 210 Sialodendrimers, 293 Sialopeptoids, 286 SialTs, 549 Sialylated glycoconjugate, 666 Sialyl glycosylations, 197 Lewis acid x (sLex), 260 Lewisx, 281, 299, 582 Lexis X antigen, 190 Sialylactones, 513

681 Sialyllactose, 282, 298 Sialyl T antigen, 179 Sialyltransferase, 199, 201, 549, 663 Sisamine, 371 Sisomicin, 312, 399 Size-exclusion chromatograghy, 560, 592 SLex epitopes, 262 SLex, 263–267 Slug lectin, 293 SMCC, 252 SmI2 (Barbier’s conditions), 279 Smilax, 163 Snti-sense agents, 518 Solanum, 163 Solid phase synthesis, 2, 5, 8, 285 Somatostatin, 526 Sorbistin, 324, 362 Spectinoic acid, 403 Spectinomycins, 402 Sporaricins, 411 SPR (surface plasmon resonance), 241 Staphylococcus aureus, 377 Staphylococcus epidermidis, 379 Steroidal saponins, 163, 164 Stille reaction, 98 Streptamine, 308 Streptidine, 314 Streptomyces, 311 Streptomycin, 134, 307, 336, 342 Streptose, 404 Streptoverticillium, 311 Streptozotocin, 323 Sugar olefins, 80 Superhelix, 531 Support-bound glycosyl acceptors, 11 Surface plasmon resonance (SPR), 234 Suzuki–Miyama cross-coupling, 98 Swern oxidation, 438 Taccacheancer, 172 TFA, 445, 453 Thin-layer chromatography, 559 Thioaryl sialoside, 290 Thioethyl donors, 10, 11 Thioethyl glycosides, 11 Thioglycoside donors, 194 Thioglycosides, 119, 144 Threonine, 495, 526 Thrombin, 468, 482 Thymidylate synthase, 327 Tiffeneau–Demjanov diazonium ion, 403 Tigogenin, 164

682 Tipson–Cohen elimination protocol, 408 TMSOTf (Trimethylsilyl triflate), 16, 112 Tn antigen epitope, 87 Tobramycin, 333, 379 TOCSY, 600 Trehazolin, 324, 416 Tributylstannane, 379, 443 Trifluoroacetic acid (TFA), 470 Trityl resin, 511 Trospectomycin, 404 Tryptophan, 526 Tuberculosis, 307 Turbidimetric analysis, 232 UDP, 559 UDP-cellobiose, 628 UDP-Gal, 537, 542, 543, 559, 560, 605 UDP-galactose, 298 UDP-Glc, 537, 540, 543 UDP-GlcA, 537 UDP-GlcNAc, 537, 543 UDP-N-acetyllactosamine, 540 UDP-Xyl, 537 UMP, 559 Uridine, 560 Uronic acid, 425, 428 Vaccine, 586 Validamycin A, 405

Index Validamycin, 315 Validoxylamine, 405 Valienamine, 320, 405 Valiolamine, 320 van der Waals force, 610 Viral infection, 669 Wacker oxidation, 94, 170 Watson–Crick base-pairing, 518 WGA (wheat germ agglutinin), 286, 293 Wittig reaction, 89, 91, 369 of sugar lactol, 91 with protected sugars, 96 rearrangement, 108 with unprotected sugars, 94 Wittig-cyclization sequence, 96 Xenobiotic, 134 Xenotransplantation, 540, 581, 582, 584, 586 Yeast mannan, 293 YEE (ah-GalNAc)3, 247–249, 252 Ylide, 91, 94 Yucca, 163 Zanamivir (4-guanidino-Neu5Ac2en, GG167), 278 Zemple´n method, 445, 604