Mechanistic investigations of glycosaminoglycan degrading enzymes

by

Seino Jongkees

B.Sc.(hons), The University of Otago, 2007 B.A., The University of Otago, 2007 Dip. Grad., The University of Otago, 2007

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

in

The Faculty of Graduate Studies

(Chemistry)

THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) June 2013 ©SeinoJongkees2013 Abstract

Glycosaminoglycans are the main structural polysaccharides of vertebrates, and rep- resent a major barrier to the spread of both bacterial infection and tumours. The enzymes by which mammals and pathogens degrade these polysaccharides use very different mechanisms, and may represent suitable therapeutic targets. In this thesis, work is presented towards an understanding of the mechanismsofClostridium per- fringens unsaturated glucuronyl hydrolase (UGL), the second enzyme in the bacterial

pathway for degradation of glycosaminoglycans, and human heparanase, the enzyme

by which the abundant glycosaminoglycan heparan sulfate is remodelled.

For UGL, evidence was presented for a hydration reaction scheme that had pre-

viously been proposed on the basis of crystallographic evidence. This was shown by

characterisation of products formed by reaction in D2Oand10%methanol,andby demonstrating hydrolysis of three compounds that are only expected to be turned

over by the enzyme if this reaction is correct. Investigationoftheeffectsofsubstitu-

ents on the transition state stability, by measurement of a linear free-energy rela-

tionship for a series of aryl glycosides, kinetic isotope effects, and rate determination

for heteroatom-substituted substrates, led to the proposalofalternatemechanisms.

Attempts to verify these mechanisms were made by testing of potential inhibitors,

rescue of a catalytic-residue mutant, trapping of a covalentglycosyl-enzymeinter-

mediate, or synthesis of a potential intermediate, but without success. The mechan-

ism that appears most likely proceeds through protonation ofthesubstrateC4-C5

ii double bond, with the resulting C5 positive charge being quenched by opening of the pyranose ring to give a C5 ketone and a C1-C2 epoxide. Subsequent hydration of the ketone and opening of this epoxide reforms the pyranoseringandgivesthe same product as direct hydration, but through a lower energy path.

For mammalian heparanase, several new potential substratesandapotential inactivator were synthesized and tested. While this work waslargelyunsuccessful,it indicated that optimisation of sulfation patterns without modification of the aglycone is likely a futile strategy. A redesigned aglycone was proposed, representing a new path towards the goal of studying this enzyme for its eventualuseasatherapeutic target in cancer therapy.

iii Preface

Aside from the exceptions noted below, all results presentedinthisthesisaremyown work. Analysis of these results was carried out in consultation with my supervisor,

Professor Withers. The work on natural substrate preferences and optimisation of enzyme conditions in Chapter 2, as well as the work on mutagenesis and testing of inhibitors in Chapter 4, was carried out together with Hayoung Yoo, a biochem- istry 449 student. Expression and purification of recombinant human heparanase and synthesis of potential substrates and inactivator for this enzyme, presented in

Chapter 5, was carried out in collaboration with Professor Jian Liu and co-workers at the University of North Carolina.

Material contained in Chapter 2 of this thesis was previouslyreportedinthe following publication:

Jongkees, S. A. K.;Withers,S.G.Journal of the American Chemical Society

2011, 133,19334–19337“GlycosideCleavagebyaNewMechanisminUnsaturated

Glucuronyl Hydrolases”

iv Table of Contents

Abstract ...... ii

Preface ...... iv

Table of Contents ...... v

List of Tables ...... x

List of Figures ...... xii

List of Schemes ...... xvi

List of Abbreviations ...... xx

Acknowledgements ...... xxii

1Introduction...... 1

1.1 General introduction ...... 1

1.2 Glycosaminoglycans ...... 3

1.2.1 Hyaluronan ...... 7

1.2.2 Keratan sulfate ...... 8

1.2.3 Chondroitin and dermatan sulfate ...... 10

1.2.4 Heparin and heparan sulfate ...... 12

1.2.5 Other uronic acid-containing polysaccharides ...... 17

v 1.3 Glycoside hydrolases (glycosidases) ...... 21

1.4 Hydration and elimination mechanisms in glycoside cleavage . . . . 26

1.4.1 Family GH4 and GH109 glycoside hydrolases ...... 27

1.4.2 Elimination and hydration in sialidases ...... 33

1.4.3 α-1,4-Glucan lyase ...... 35 1.4.4 Polysaccharide lyases ...... 39

1.4.5 Unsaturated glucuronyl and galacturonyl hydrolases ..... 43

1.4.6 N -Acetyl-muramic acid 6-phosphate hydrolase (MurQ) . . . 49

1.5 Thesis aims ...... 51

2 Confirmation of the hydration reaction ...... 53

2.1 Cloning and purification of Clostridium perfringens UGL ...... 54

2.1.1 Cloning ...... 54

2.1.2 Enzyme expression and purification ...... 54

2.2 Development of a chromogenic substrate ...... 57

2.2.1 Previous assays ...... 57

2.2.2 Substrate synthesis ...... 58

2.3 Enzyme optimisation ...... 61

2.4 Natural substrate variation ...... 65

2.5 Characterisation of UGL reaction products ...... 74

2.5.1 Reaction in D2O...... 76 2.5.2 Reaction in 10% methanol ...... 78

2.6 Unusual substrates ...... 82

2.6.1 Kdn2en ...... 83

2.6.2 Axial phenol ...... 87

2.6.3 Thiophenol ...... 89

2.7 Conclusions ...... 90

vi 3ProbingthemechanismofUGL...... 92

3.1 Detection of initial products by NMR ...... 92

3.2 Linear free-energy relationship ...... 96

3.2.1 Synthesis of aryl glycosides ...... 96

3.2.2 Kinetics ...... 98

3.3 Effects of heteroatoms ...... 102

3.3.1 α-andβ-ΔGlcA fluorides ...... 103

3.3.2 2,4-Dinitrophenyl 2F-ΔGlcA ...... 107 3.3.3 4-F substrate ...... 110

3.4 Kinetic isotope effects ...... 113

3.4.1 Synthesis of a substrate deuterated at carbon 1 ...... 114

3.4.2 Synthesis of a substrate deuterated at carbon 4 ...... 115

3.4.3 Kinetic isotope effect measurements ...... 116

3.5 Conclusions, and possible alternate mechanisms ...... 124

4Testingofalternativemechanisms ...... 130

4.1 Attempted rescue of D113G mutant ...... 130

4.2 Testing of potential inhibitor leads ...... 134

4.3 Anticipated trapping reagents for UGL ...... 141

4.3.1 2,3-Difluoro Kdn ...... 141

4.3.2 4-Deoxy-1,5-difluoro-iduronic acid ...... 145

4.3.3 1-Fluoro-ΔGlcAfluoride ...... 150 4.4 Attempted synthesis of proposed epoxide intermediate ...... 153

4.5 Conclusions and future directions ...... 159

5Heparanasesubstrateandinactivatortesting ...... 161

5.1 Chapter introduction ...... 161

vii 5.2 Synthesis of compounds ...... 164

5.3 Substrate testing ...... 166

5.3.1 Confirmation of heparanase activity ...... 166

5.3.2 Towards a fluorescent substrate ...... 168

5.4 Testing of a potential HPSE inactivator ...... 173

5.5 Conclusions ...... 176

6 Overall conclusions ...... 177

7Materialsandmethods ...... 185

7.1 Materials ...... 185

7.2 Synthesis ...... 186

7.2.1 General methods ...... 186

7.2.2 Development of chromogenic substrates ...... 188

7.2.3 Standards for UGL reaction in 10% methanol ...... 194

7.2.4 Unusual substrates for UGL ...... 197

7.2.5 Substrates for linear free-energy relationship ...... 206

7.2.6 Substrates for with varied heteroatoms ...... 221

7.2.7 Substrates for kinetic isotope effects ...... 235

7.2.8 Potential inhibitors of UGL ...... 242

7.2.9 Potential trapping reagents for UGL ...... 243

7.2.10 Glucuronides for extension to heparanase substrates. . . . . 256

7.3 Biochemistry...... 258

7.3.1 Cloning of UGL from Clostridium perfringens...... 258

7.3.2 Testing of expression conditions ...... 259

7.3.3 Heterologous expression of UGL in Escherichia coli...... 260

7.3.4 Michaelis-Menten kinetics ...... 261

viii 7.3.5 NMR monitoring of UGL-catalysed reaction ...... 262

7.3.6 Profile of UGL activity at varied pH...... 262

7.3.7 Titration of benzyl ΔGlcA...... 263 7.3.8 Effect of temperature on UGL...... 263

7.3.9 Reaction in D2O...... 264 7.3.10 Reaction in 10% methanol ...... 264

7.3.11 Testing of competitive inhibitors...... 265

7.3.12 Inactivator testing ...... 266

7.3.13 Kinetic isotope effects...... 266

7.3.14 Attempted Rescue of D113G mutant with nucleophiles . .. 268

7.3.15 Heparanase kinetics ...... 269

References ...... 272

Appendices

AUGLmultiplesequencealignment ...... 291

B2D-NMRspectraofUGLproductsandstandards ...... 295

CKineticisotopeeffects ...... 300

DPlotsforMichaelis-Mentenkinetics ...... 306

ix List of Tables

1.1 Michaelis-Menten kinetic parameters for previously characterised Ba-

cillus sp. GL1 UGL mutations ...... 48

2.1 Michaelis-Menten kinetic parameters for hydrolysis of the two aryl

glycoside substrates 6 and 10 by UGL under optimised conditions. . 65

2.2 Comparison of kinetic parameters for UGL from different source or-

ganisms with GAG-derived natural substrates ...... 68

2.3 Michaelis-Menten kinetic parameters for three unusual substrates ac-

cepted by UGL ...... 82

3.1 Conditions used and yields for the synthesis of unsaturated aryl

glucuronides ...... 97

3.2 Michaelis-Menten kinetic parameters for hydrolysis of aryl unsatur-

ated glucuronides by UGL...... 99

3.3 Kinetic parameters of UGL substrates with varied heteroatoms at the

anomeric carbon ...... 106

3.4 Kinetic isotope effects from deuterium incorporation at carbon 1 and

carbon 4 of 2,4,6-trichlorophenyl ΔGlcA, and solvent deuterium effect

with 4-nitrophenyl ΔGlcA...... 117

x 5.1 Rates of hydrolysis for the commercial pentasaccharide Arixtra and

heparin di- and tri-saccharide substrates with fluorescent leaving

groups ...... 169

xi List of Figures

1.1 Structures of the simplest repeating unit for each class of glycosa-

minoglycan ...... 4

1.2 Structures of types I and II keratan sulfate...... 9

1.3 Sub-types of chondroitin and dermatan sulfate, showing the main

sulfation patterns in each ...... 11

1.4 Known disaccharide units in heparan sulfate and heparin,withthe

archetypal disaccharides for the N- and S-regions indicated...... 14

1.5 Domain organisation of heparan sulfate and illustrationofitsdomain

structures ...... 15

1.6 Minimal structure for heparin anticoagulant activity ...... 16

1.7 Structures of xanthan and gellan...... 17

1.8 Structure of alginate...... 18

1.9 Structures of the pectins homogalacturonan and rhamnogalacturonan

I...... 19

1.10 Structure of rhamnogalacturonan II ...... 20

1.11 X-ray crystal structure of the Bacillus sp. GL1 UGL D88N active site

from two perspectives, showing an unsaturated hyaluronan disacchar-

ide substrate bound and all side-chains within 5 Å ...... 47

2.1 Expression of UGL under varying conditions ...... 55

xii 2.2 Purification of UGL ...... 56

2.3 Effect of salt concentration, pH, and temperature on the reaction rate

of UGL with 4-nitrophenyl ΔGlcA ...... 62

2.4 Determination of the pKa of benzyl ΔGlcA ...... 63 2.5 Dixon plot showing competitive inhibition of UGL by Mes.NaOH

buffer ...... 66

2.6 X-ray crystal structure of the S. agalactiae UGL active site . . . . . 71

2.7 Key fragments of a multiple sequence alignment of UGL sequences by

the MUSCLE algorithm ...... 73

2.8 1H-NMR spectrum of 24,thefinalproductsofUGLdegradationof

4-nitrophenyl ΔGlcA (6)...... 75 2.9 Overlay of expanded 1H-NMR spectra of the products from UGL-

catalysed hydrolysis of 4-nitrophenyl ΔGlcA (6)inD2OandH2O. . 77 2.10 Overlay of expanded 1H-NMR spectra of the products from UGL-

catalysed hydrolysis of phenyl ΔGlcA (10)in10%MeOHinH2O

(26)andacontrolinH2O...... 78 2.11 Overlay of expanded 1H-NMR spectra of the anomeric methanol syn-

thetic standard (31), the products from UGL-catalysed hydrolysis of

phenyl ΔGlcA (6)in10%MeOHinH2O(26)andthecarbon5meth- anol epimeric synthetic standard (28)...... 81

2.12 Dixon plot showing competitive inhibition of UGL by Kdn2en . . . . 86

3.1 Offset stack of 1H-NMR spectra showing reaction of thiophenyl ΔGlcA (46)withahighconcentrationofUGL...... 94

3.2 Plot of log(kcat), A, and log(kcat/Km), B, against leaving group pKa for hydrolysis of aryl unsaturated glucuronides by UGL...... 100

xiii 3.3 Minimal Dixon plot showing competitive inhibition of UGLbyaxial

ΔGlcAfluoride ...... 105

3.4 Inhibition of UGL by equatorial ΔGlcA fluoride ...... 106 3.5 Dixon plot showing competitive inhibition of UGL by 2,4-dinitrophenyl

2-deoxy-2-fluoro ΔGlcA ...... 110 3.6 Comparison of side-chain carboxyl placement in UGL from Bacillus

sp. GL1 and AroA from E. Coli ...... 121

4.1 Elution trace for UGL D113G ...... 132

4.2 Profile of first order rate for hydrolysis of 4-nitrophenyl ΔGlcA (6)by UGL D113G and wild-type ...... 133

4.3 X-ray crystal structure of the Bacillus sp. GL1 UGL active site show-

ing glycine and all side-chains within 5 Å, from two perspectives . . 136

4.4 Dixon plot showing competitive inhibition of UGL by shikimate . . . 138

4.5 Representation of substrate and shikimic acid interaction with D88/113

in the active site of UGL...... 139

4.6 Time-dependent inactivation of UGL by 2,3-difluoro Kdn (114)... 144

4.7 Time-dependent inactivation of UGL by ΔGlcA fluoride (70)....144 4.8 19F-NMR showing partial hydrolysis of 2,3-difluoro Kdn (114)with

UGL and in a non-enzymatic control ...... 145

4.9 Time-dependent inactivation of UGL by 4-deoxy-1,5-difluoro-iduronic

acid (120)...... 147

4.10 Overnight hydrolysis of 4-deoxy-1,5-difluoro-iduronic acid (120)with

UGL and in a non-enzymatic control as monitored by TLC and 19F-

NMR...... 148

4.11 Dixon plot showing competitive inhibition of UGL by 4-deoxy-1,5-

difluoro-iduronic acid (120)...... 149

xiv 4.12 Time-dependent inactivation of UGL by 1-fluoro-ΔGlcA fluoride (124)...... 153

4.13 X-ray crystal structure of Haemophilis influenzae sialic acid aldolase

(Neu5Ac aldolase) active site ...... 155

5.1 Sulfation in a substrate oligosaccharide as required forcleavageby

HPSE ...... 163

5.2 Overnight reaction of 144 with HPSE, monitored by 19F-NMR . . . 171

5.3 Inhibition of HPSE by TFMU substrate trisaccharides ...... 172

5.4 Attempted time-dependent inactivation of HPSE by 149 ...... 175

6.1 Hypothetical energy profile for UGL-catalysed ΔGlcA hydrolysis, com- pared to the non-enzymatic acid-catalysed reaction ...... 181

xv List of Schemes

1.1 The general mechanism of inverting α-glucosidases...... 22

1.2 The general mechanism of retaining β-glucosidases...... 24

1.3 The mechanism for hydration of glucal by a retaining β-glucosidase . 26

1.4 The general mechanism of family GH4 6-phospho-α-glucosidases . . . 28 1.5 The general mechanism of elimination and hydration by sialidases . . 34

1.6 The general mechanism of α-(1,4)-glucan lyases ...... 37

1.7 The anticipated mechanism of inactivation of α-(1,4)-glucan lyases by

1-fluoro-α-d-glucosyl fluoride, if a second residue were to be acting as catalytic base...... 39

1.8 The general mechanism of polysaccharide lyases acting onpectate. . 40

1.9 The general mechanism proposed for unsaturated glucuronyl

hydrolases...... 44

1.10 Bacterial metabolic pathway for catabolism of free ΔGlcA to the com- mon metabolites pyruvate and d-glyceraldehyde-3-phosphate . . . . 45

1.11 The general mechanism for N -acetyl-muramic acid 6-phosphate hy-

drolase (MurQ)...... 50

2.1 Final optimised 4-nitrophenyl glycoside substrate synthesis...... 59

2.2 Final optimised phenyl glycoside substrate synthesis...... 59

2.3 Formation of the final product of the UGL-catalysed reaction. . . . . 74

xvi 2.4 Reaction catalysed by UGL, showing the route by which deuterium

and methanol are incorporated ...... 77

2.5 Decomposition of the methanol adduct formed by UGL-catalysed re-

action of phenyl ΔGlcA (10)...... 79 2.6 Synthesis of a standard for the product of UGL-catalysed reaction in

10% methanol at carbon 5...... 80

2.7 Synthesis of a standard for the product of UGL-catalysed reaction in

10% methanol at carbon 1...... 81

2.8 Expected reaction pathway for UGL if a mechanism analogous to that

for glucal hydration were followed...... 82

2.9 Synthesis of 2,3-dideoxy-d-glycero-d-galacto-non-2-enopyranosonate

(Kdn2en, 40)...... 84

2.10 Synthesis of an axial phenyl ΔGlcA substrate for UGL...... 87

2.11 Synthesis of a thiophenyl ΔGlcA substrate for UGL...... 89

3.1 Mechanism for UGL catalysis of rearrangement of the hemiketal in-

termediate 22 to cleave the glycosidic bond...... 95

3.2 Synthesis of an equatorial ΔGlcA fluoride substrate...... 104

3.3 Synthesis of an axial ΔGlcA fluoride substrate...... 104 3.4 Synthesis of a 2-deoxy-2-fluoro substrate analogue for UGL...... 108

3.5 Attempted synthesis of a 4-fluoro substrate, failing at elimination by

DBU...... 111

3.6 Retrosynthetic analysis of 88 with late oxidation, and test of the first

step...... 112

3.7 Synthesis of a 1-deuterated subtrate...... 114

3.8 Synthesis of a 4-deuterated subtrate...... 116

xvii 3.9 Deduced conformation of the transition state that would give rise to

the observed KIE on kcat/Km,forthefirstirreversiblestep,leading to a hypothetical oxocarbenium ion intermediate ...... 119

3.10 Comparison of the reactions catalysed by AroA and UGL...... 120

3.11 Illustration of the analogous relationship of deuteriums in unsatur-

ated 1-deutero-glucuronides to 5-deutero-glucosides on formation of

an initial oxocarbenium-ion transition state ...... 122

3.12 SKIEs for hydration of vinyl ether acetals, proceeding by either initial

hydration of the vinyl ether or initial hydrolysis of the acetal. . . . . 123

3.13 Possible mechanisms for UGL to account for KIE and LFER observa-

tions...... 126

4.1 Illustration of structural analogies from UGL substrates, intermedi-

ates, and putative oxocarbenium ion-like transition statestopotential

inhibitors ...... 135

4.2 Synthesis of Neu2en...... 140

4.3 Rationale for attempted trapping of UGL with 2,3-difluoroKdn(114)142

4.4 Synthesis of 2,3-difluoro Kdn...... 143

4.5 Synthesis of 4-deoxy-1,5-difluoro-iduronic acid...... 146

4.6 Rationale for attempted trapping of UGL with 1-fluoro-ΔGlcA fluor- ide (124)...... 151

4.7 Synthesis of 1-fluoro-ΔGlcA fluoride ...... 152 4.8 Epoxide intermediate expected from Kdn2en (40)accordingtomech-

anism C of Scheme 3.13 ...... 154

4.9 Attempted synthesis of the epoxide intermediate expected from Kdn2en

(40)undermechanismCofScheme3.13...... 156

xviii 4.10 Hafnium tetrachloride and zinc iodide-catalysed rearrangement of d-

glucal in water...... 156

4.11 Alternate routes for synthesis of epoxide intermediate 133 from Kdn. 158

5.1 Starting glucuronic acid pseudo-disaccharides and general scheme for

chemoenzymatic synthesis of appropriately sulfated substrates and

inactivators for HPSE ...... 165

5.2 Compounds for the HPSE reducing sugar assay, WST-1 (141)and

Arixtra (142)...... 167

5.3 A potential synthesis for a HPSE substrate with a charged aglycone,

and illustration of two possible charge placements by this aglycone to

mimic the sulfation of optimal HPSE natural substrates ...... 173

5.4 Structure of the potential 2-deoxy-2-fluoro inactivatorofHPSE.. . . 174

6.1 Conformation of transition states in mechanism C of Scheme 3.13 to

account for experimental observations ...... 179

6.2 Precedent for mechanism C of Scheme 3.13 from non-enzymatic reac-

tions of glycosides...... 182

xix List of Abbreviations

BSA Bovine Serum Albumin

CaZY Carbohydrate active enzymes

COSY Correlation spectroscopy

DBU 1,8-Diazabicyclo[5.4.0]undec-7-ene

GAGs Glycosaminoglycans

Gal galactose

GalNAc N -acetyl-galactosamine

GlcA glucuronic acid

GlcNAc N -acetyl-glucosamine

GlcNTFA N -trifluoroacetyl-glucosamine

HPLC High-performance liquid chromatography

HPSE human heparanase

IdoA iduronic acid

IPTG isopropyl β-d-thiogalactoside

LB lysogeny broth, also known as Luria broth

xx List of Abbreviations

Man mannose

MU Methylumbelliferone

NAD+ Nicotinamide adenine dinucleotide (oxidised form)

Neu5Ac N -Acetyl-Neuraminic acid

O.D. optical density

PAPS 3!-phosphoadenosine 5!-phosphosulfate

PDB Protein Data Bank

PGs proteoglycans

Rha rhamnose

SDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis

SKIE solvent kinetic isotope effect

TFMU Trifluoromethylumbelliferone

TOCSY Total correlation spectroscopy

TYP tryptone-yeast-phosphate growth medium

UDP Uridine diphosphate

UGH unsaturated galacturonyl hydrolase

UGL Unsaturated glucuronyl hydrolase

UV ultraviolet

xxi Acknowledgements

Thanks first and foremost to Professor Stephen Withers, my supervisor, for all of his help and encouragement over the years, and also for believing in me after a less- than-stellar first semester. I am grateful to our collaborator Professor Jian Liu and group from the University of North Carolina for their willingness to synthesise the heparanase substrates and inactivator, and provide the enzyme to allow testing of these, as well as troubleshooting when nothing appeared to beworkingasitshould.

Thanks to all members of the Withers lab, past and present, fortheirhelp, discussions, and also for their distractions. In particular: Dr Hongming Chen for being the best chemistry labmate you could ask for (discussions, advice, keeping everything running, providing intermediates, generally being a fountain of knowledge in the lab); Emily Kwan for her help with molecular biology andorientationwithin the biochemistry lab; Hayoung Yoo, a biochemistry 449 student, for her hard work on natural substrate preferences, optimisation of conditions, inhibitor testing, and mutagenesis for UGL; Bojana Rakic and Su Hancock for getting meorientedinthe

NCE lab; Brian Rempel and Tom Morley for getting me oriented in the chemistry lab; Freya Chen for keeping me distracted (and helpful discussions); Ethan Goddard-

Borger and Jamie Rich for their many helpful suggestions at group meetings (as well as outside of these); Ricardo Resende and Tina Rasmussenforstealingmy spectrometer, which forced me to finish my chemistry when I didn’t want to; and

Miranda Joyce for her administrative support. The contributions of staffat UBC

xxii chemistry was also much appreciated, especially but not limited to the NMR and mass spectrometry teams. This document was created using thetemplateforLyx created by Christopher P. Barrington-Leigh.

Thanks to Dr Nancy (Ju-Wei) Liu of the University of Auckland and William

Archibald of the University of Victoria for sending me digital versions of the many articles for which the UBC library had only physical copies. Finally, thanks to my parents for all of their support over the years, and Gerritt Lighthart for proofreading several chapters of this thesis, and especially for stickingaroundtheselastfewyears until I finished this business.

xxiii Chapter 1

Introduction

1.1 General introduction

Carbohydrates are, by mass, the single most abundant class ofbiomoleculeinthe world. 1 This abundance arises in large part from the common role of polysacchar- ides in the structural features of many forms of life, such as glycosaminoglycans in vertebrates, cellulose in plants, chitin in fungi and arthropods, and peptidoglycan in bacteria. These structural polysaccharides are very diverse in their structures across species, and so a great diversity of enzymes is required in order to manipulate them.

Structural polysaccharides often play a crucial role in the defence of an organism both from the environment and during pathogenesis and competition, with the rigid layer preventing access to the otherwise vulnerable cell. However, because of the physical limitations of this structural layer, an organism also needs to be able to modify its own structural polysaccharides to be able to grow,eitherbyrecycling or replacing the structural layer. The balance between resistance to outside change while maintaining susceptibility to self-modification is animportantdrivingforcein carbohydrate-active enzyme evolution.

Adramaticexampleofthisisfoundintheevolutionaryhistory of plants, with the structural components of early trees giving them an enormousevolutionaryadvant- age in competition with other plants, allowing them to literally overshadow those unable to attain the same heights and leading to the formationofvastforests.The

1 1.1. General introduction new structures allowing this, in particular lignin impregnated with cellulose, were so successful at resisting degradation by other species that for millions of years there was a net increase in carbon sequestered as plant biomass (known as the carbonifer- ous era, for this deposition of carbon). This lasted until theeventualevolutionof fungal enzymatic systems able to degrade this plant matter that, among other en- vironmental factors, restored carbon balance.2–4 This co-evolution of new protective

features and new methods to bypass them is often referred to asanevolutionary

arms race, constantly occurring on a wide variety of time and size scales.

Carbohydrates also play a wide variety of roles in living organisms beyond struc-

tural features. One of the main energy reservoirs for living organisms is in the form

of polysaccharides, such as starch and glycogen. Carbohydrates also comprise one

half of the repeating unit of nucleic acids, providing the link between the backbone

and the nucleo-bases. In higher organisms, carbohydrates are important signalling

molecules, and glycosylation of lipids and proteins can drastically alter their role and

fate. Further roles in higher organisms include the formation of mucosal barriers and

lubrication of mammalian joints.

This diversity of roles is possible in large part because of the vast diversity of

structures available to carbohydrates, and the associated differences in physical and

chemical properties that arise from those differences. Whilefunctionallysimple,con-

taining only a few different functional groups, the permutations of branching and the

stereo-, and regio-chemistries in carbohydrates gives a class of compounds drastically

different from peptides and nucleic acids, the two other main classes of biopolymer.

As an example of this difference, a hexamer of nucleic acids allows for 46 (∼ 4 × 103)

unique structures, a hexamer of the natural amino acids, withalloftheirdifferent

side-chains, allows for 206 (∼ 6 × 107)uniquestructures,whileahexamerofsimple

reducing d-hexoses allows for a staggering ∼ 1.05 × 1012 structures.5 Accounting

2 1.2. Glycosaminoglycans for mirror image l-sugars, ketoses as well as aldoses, and variations in monomer size and functional groups (such as pentoses, amino sugars and uronic acids) would vastly increase this figure.

1.2 Glycosaminoglycans

Among primary structural carbohydrates of mammals are the glycosaminoglycans

(GAGs). These are composed of a linear repeating unit of two sugars, an amino-sugar and a uronic acid (with one exception, keratan sulfate), withmanyhavingextensive sulfation. The simplest repeating units for all GAGs are shown in Figure 1.1. The sulfation, together with the uronic acid carboxylate groups, gives GAGs a high degree of negative charge, with heparin having the highest density of negative charge of any bio-macromolecule.6 Classification of GAGs is based on the subunits present

(GlcNAc or GalNAc, GlcA or IdoA) and the linkages between these (α or β,1-3or1- 4), with further sub-classification based on sulfation pattern and other modifications.

The four main classes are hyaluronan, keratan sulfate, chondroitin and dermatan sulfates, and heparin and heparan sulfate. Each of these willbediscussedinmore depth in the following subsections. Note that, unless specified, all sugars are of the d-absolute stereochemistry, with the exception of idose derivatives, which occur as l-idose.

These GAGs primarily exist attached to proteins as proteoglycans (PGs), al- though one class exists as free polysaccharides.7 Almost all mammalian cells pro- duce PGs. These are either secreted, attached at the membrane, or stored in granules within the cell. The GAG chains are attached to the protein core by attachment at either serine or threonine (O-glycosylation) or asparagine (N -glycosylation) side- chains. Individual protein cores can have as few as one or as many as >100 GAG chains attached, with some core proteins only being temporarily modified with GAG

3 1.2. Glycosaminoglycans

OH OH OH -OOC O O O OH O HO O O O HO O O OH O OH NHAc NHAc 4)- GlcA -β-(1,3)- GlcNAc -β-(1, HO 3)-GalS-β-(1,4)-GalNAc-β-(1,

Hyaluronan Keratan sulfate

- OH OH O3SO OH -OOC O O O O O O HO O O HO O O OH NHAc -OOC OH NHAc 4)- GlcA -β-(1,3)- GalNAc -β-(1, 4)- IdoA -α-(1,3)-GalNAc4S-β-(1,

Chondroitin sulfate Dermatan sulfate

-OOC OH OH O O O O O O HO O HO O - OH HO OH HO OOC - AcHN O3SHN O O 4)- GlcA -β-(1,4)- GlcNAc -α-(1, 4)- IdoA -α-(1,4)- GlcNS -α-(1, Heparan sulfate Heparin

Figure 1.1: Structures of the simplest repeating unit for each class of glycosa- minoglycan. Sulfation is only shown in those cases where it isabsolutelyconserved. For descriptions of the variable sulfation patterns of each class see the relevant sub- section below.

4 1.2. Glycosaminoglycans chains (called part-time PGs).

There are no clear defining structural features of the proteincomponentofaPG, with many different types existing. Very broadly, these can beclassified7 into five main types: (1) large extracellular chondroitin sulfate-PGs, including a large number of extracellular matrix components from different tissues, (2) heparan sulfate-PGs of the extracellular membrane/basement membrane, (3) small homologous core proteins with a small number of GAG chains (chondroitin, dermatan or keratan sulfates), (4) membrane-intercalated cell-surface PGs, and (5) intracellular PGs, containing exten- ded sequences of alternating serine and glycine units (see the recognition sequence for xylosylation mentioned above) and heavily substituted with chondroitin sulfate and/or heparin. The first three of these are secreted proteins(e.g.aggrecan,ver- sican, and decorin), the fourth is cell-surface associated (e.g. syndecan), while the last is intracellular (e.g. serglycins).

Many of these proteins have domains that contain no GAGs, being involved

in other functions.7 In particular these are important for interactions with other

molecules, such as cell membranes, GAGs, or other PGs, in order to create a cross-

linked network. This networked structure is important for many of the structural

features of GAGs and PGs. The binding of PGs can be tight and specific, but is

often loose and general. Amino acids with basic side-chains are often important for

these interactions, as many of the interacting partners are poly-anionic. Interactions

of PGs, and the functions derived from this, can arise from theGAGchainalone,

the protein alone, or both together. The role of the protein can be as simple as

providing a simple scaffold to support and space the GAG chains, or it can provide

specific interactions such as that of the lectin-like domainsofaggrecanandversican,

PGs from cartilage and fibroblasts, respectively.

Synthesis of the GAG polysaccharide chain takes place in the Golgi apparatus,

5 1.2. Glycosaminoglycans usually starting with xylosylation of serine side-chains within a specific amino acid recognition sequence (broadly aaaGSGaba,wherea is D or E and b is G, D, or E, but

with many variations)8 in the endoplasmic reticulum. Not all recognition sites are

modified, and protein conformation appears to play a role in determining the extent

of this. This initial xylose is subsequently modified with twogalactoseunitsbefore

transfer of the first glucuronic acid residue. The subsequentadditionofanamino

sugar commits the chain to formation of a specific GAG.9 Following polymerisation

to the homopolymeric GAG chain, modification may take place starting with N -

deacetylation and followed by N -sulfation, O-sulfation, and finally epimerisation of

glucuronic acid to iduronic acid. These modification reactions do not occur in all

GAGs, or always proceed to completion, and the resulting patterns of modifications

define the sub-types of each GAG category, giving rise to different functions. Because

of this incomplete modification, GAG chains can exhibit a highdegreeofbothmicro-

(within a chain) and macro- (between chains) heterogeneity.GAGchainsfromthe

same tissue but on different core proteins may be different. Given that this process

gives a large number of diverse products and is not template driven, the regulation

of this synthesis is derived from complex interactions of enzymes and substrates and

competing rates between enzymes, and is poorly understood.

Glycosaminoglycans have three main roles in tissue.9 The majority of GAGs func-

tion as barriers to diffusion and as adhesion systems for cellsandotherbiomolecules.

GAGs are the primary carbohydrate components of the mammalian extracellular

matrix, forming the scaffolding between cells. A related roleisinprovidinglub-

rication and cushioning, with their high density of negativechargesoccupyinga

large hydrodynamic volume in solution and resisting compression. The viscosity of

GAG solutions is highly dependent on counter-ion nature and concentration. Fi-

nally, GAGs function as a reservoir of specific binding sites for proteins, regulating

6 1.2. Glycosaminoglycans or stabilising their activity. Examples 6 include cell-signalling and development, an-

giogenesis, axonal growth, tumour progression and metastasis, and anticoagulation.

GAGs also serve as adhesion points for pathogens, including viruses, protozoa, and

bacteria. Organisms defective in GAG metabolism show severeclinicalmanifesta-

tions, while defects in anabolism are embryonic lethal. 9

1.2.1 Hyaluronan

Hyaluronan is the only homopolymeric GAG, being composed of unmodified glucuronic

acid and N -acetyl-glucosamine in the sequence [GlcA-β-(1,3)-GlcNAc-β-(1,4)-]n,and is also the only GAG not bound to a protein core as a PG. Synthesis of hyaluronic

acid is achieved by three integral-membrane hyaluronan synthases, with subsequent

extrusion of the nascent chain into the extracellular space by a transport protein. 10

Chain lengths for hyaluronan vary from 500 to several thousand disaccharide units.

The simple repeating unit of hyaluronan belies its complex interactions, with

environmental factors such as counter-ions and concentration having dramatic ef-

fects on its structure, both in solution and in purified crystalline form.9 Calcium

ions, in particular, are able to bridge the negative charge ofcarboxylategroupsin

adjacent chains. At low Ca2+ concentrations hyaluronan forms a viscous, almost

gel-like, solution, while higher concentrations of Ca2+ give a much thinner solution.

Low concentrations of hyaluronan display non-Newtonian solution behaviour. At

higher concentrations, hyaluronan is able to form amphipathic helices, driving self-

aggregation into a double helix structure. Hyaluronan also associates with other

GAGs, especially keratan and chondroitin sulfate in cartilage. These interactions

are enhanced by high salt concentrations to mitigate charge-repulsion.

This GAG has roles as lubricant of synovial joints, space filler, wetting agent,

flow barrier in synovial fluid, and cartilage surface protector.

7 1.2. Glycosaminoglycans

1.2.2 Keratan sulfate

Keratan sulfate is unique among the GAGs in that it does not contain any uronic acids, being composed of alternating galactose and N -acetyl-glucosamine residues

11 (poly-lactosamine) in the sequence [Gal-β-(1,4)-GlcNAc-β-(1,3)-]n. Carbon 6 of one or both of these residues may be sulfated, with a non-random pattern of mono-

and di-sulfated subunits. The charged sulfate group on this position can fill many

of the roles of the uronic acid carboxylate, but notably does not allow cleavage

by polysaccharide lyase by an eliminative mechanism (refer to Subsection 1.4.4 on

page 39).

Keratan sulfate is broadly classed into three categories, with structures of the two well defined types shown in Figure 1.2. This classificationwasoriginallybased on the source tissue, but later revised to be based on linkage type to the core pro- tein.11 Type one, first located in the cornea, is linked by a branched oligosaccharide to an asparagine side-chain. Type two, first located in other tissues and previously referred to as skeletal, is linked by a different branched oligosaccharide to serine or threonine side-chains, and is fucosylated on the carbon 3 hydroxyl of some GlcNAc residues close to the protein core. Type three is poorly characterised, being found in PGs from the brain, and contains a mannose linkage between the keratan sulfate chain and serine or threonine side-chains. Keratan sulfatesaresynthesisedbyglyc- osyl transferases common to glycoproteins and glycolipids,unliketheotherglycosa- minoglycans, which have specific glycosyl transferases for their synthesis. Keratan sulfate is self-aggregating,9 and also interacts with collagen fibres. 12

The main role for type I keratan sulfate is in the cornea, in which it is the

primary GAG.12 Its role in this tissue is to establish and maintain the important

arrangement of collagen fibrils that allows light to pass through. Type II keratan

sulfate is a key component of cartilage, attached to the largeproteinaggrecan,along

8 1.2. Glycosaminoglycans Figure 1.2: Structures of types I and II keratan sulfate.

9 1.2. Glycosaminoglycans with the hyaluronan to which it binds.11

1.2.3 Chondroitin and dermatan sulfate

The disaccharide repeating unit in chondroitin sulfate, as well as the related dermatan sulfate, is composed of glucuronic (or iduronic) acid and N -acetyl-galactosamine in the sequence [GlcA-β-(1,3)-GalNAc-β-(1,4)-]n.ThesestructurescanbeO-sulfated on carbon 2 and very rarely carbon 3 of the uronic acid and on carbons 4 and 6 of

the galactosamine, with the chondroitin sulfates being further sub-classified based

on the sulfation patterns (see Figure 1.3). The difference between chondroitin and

dermatan sulfates is the presence of iduronic acid, the carbon 5 epimer of GlcA, in

the place of some glucuronic acid residues in dermatan sulfate. Dermatan sulfate

occurs in small PGs with a low number of short GAG chains (8 to 20chainsof

15 to 55 kDa each), while chondroitin sulfate typically occurs in larger structures

composed of longer chains (20 to 100 chains of 15 to 70 kDa each).9

These chains are attached to the core protein by the same GlcA-β-(1,3)-Gal-

β-(1,3)-Gal-β-(1,4)-Xyl-β-1-O-(Ser) linker sequence as heparan sulfate and heparin. 8 Synthesis of this linker initiates in the endoplasmic reticulum and is completed in the

cis/medial Golgi, followed by addition of the first uronic acid by a specialised glycosyl transferase common to chondroitin/dermatan sulfate and heparin/heparan sulfate.

Subsequent chain elongation steps are carried out by GlcA andGalNActransferases concomitantly with sulfation by sulfotransferases in the medial/trans Golgi. Some

sulfation also occurs in the linkage region of chondroitin/dermatan sulfate, which

has not been observed in heparin or heparan sulfate, even whenpresentonthesame

core protein.

Chondroitin-6-sulfate and some forms of dermatan sulfate are able to self-associate,

but chondroitin-4-sulfate is not.9 However, chondroitin-4-sulfate is able to associate

10 1.2. Glycosaminoglycans

- - O3SO OH O3SO OH -OOC O O O O O O HO O O HO O O OH NHAc -OOC OH NHAc 4)- GlcA -β-(1,3)-GalNAc4S-β-(1, 4)- IdoA -α-(1,3)-GalNAc4S-β-(1,

Chondroitin-4-sulfate Dermatan sulfate (formerly chondroitin sulfate A) (formerly chondroitin sulfate B)

OH OSO - OH OSO - -OOC 3 -OOC 3 O O O O O O HO O O HO O O - OH NHAc OSO3 NHAc 4)- GlcA -β-(1,3)-GalNAc6S-β-(1, 4)-GlcA2S-β-(1,3)-GalNAc6S-β-(1,

Chondroitin-6-sulfate Chondroitin-2,6-sulfate (formerly chondroitin sulfate C) (formerly chondroitin sulfate D)

- - O3SO OSO -OOC 3 O O O HO O O OH NHAc 4)- GlcA -β-(1,3)-GalNAc4,6S-β-(1,

Chondroitin-4,6-sulfate (formerly chondroitin sulfate E)

Figure 1.3: Sub-types of chondroitin and dermatan sulfate, showing the main sulfa- tion patterns in each.

11 1.2. Glycosaminoglycans with dermatan sulfate. Dermatan sulfate tends to adopt an extended conformation in solution, explained by the presence of iduronic acid units, for which the energy

4 1 2 6 difference between the C1, C4,and SO conformers is very small. This flexibility gives rise to poorly defined secondary structures.

Chondroitin and dermatan sulfate are synthesised by virtually all vertebrate cells.8 Their primary function is as a part of the extracellular matrix, providing support and connectivity to cells. They are also found in basement membranes and on cell surfaces, the latter having roles as receptors ratherthanstructuralfunctions.

1.2.4 Heparin and heparan sulfate

Heparan sulfate and heparin are the only glycosaminoglycanswithanalpha-linkage at the hexosamine, with the chain composed initially of glucuronic acid and N - acetyl-glucosamine in [GlcA-β-(1,4)-GlcNAc-α-(1,4)-]n.ThisclassofGAGsshows the highest degree of heterogeneity. The distinction between heparin and heparan

sulfate is one of degree, rather than type, with heparan sulfate showing a higher

proportion of acetylated glucosamine, less O-sulfation and a lower proportion of

iduronic acid. Counterintuitively, heparin is more heavilysulfatedthanheparan

sulfate. Heparan sulfate molecules, at 10–70 kDa, are also generally larger than

heparin, at 10–12 kDa.6 The other key difference between these molecules is their

location, with heparin being stored in granules of mast cellsandheparansulfatebeing

found on cell membranes or extracellularly. Overall, O-sulfation of these GAGs can

be found in varying proportions on carbon 2 of the uronic acid and carbons 3 and 6 of

the glucosamine, which can also be N -deacetylated and N -sulfated. A list of known

disaccharide units found in heparan sulfate and heparin are shown in Figure 1.4.13–15

Heparan sulfate is composed of long domains containing a low degree of sulfation

acting as spacers between shorter sections with high sulfation that resemble heparin,

12 1.2. Glycosaminoglycans with some intermediary mixed domains also present, as depicted in Figure 1.5. These mixed domains have important functional roles as they provide recognition sites for anumberofchemokines.

Synthesis of heparan sulfate and heparin proceeds in a similar manner to chon- droitin and dermatan sulfates. A core protein is xylosylatedonspecificserine residues, the linker region synthesised by specific transferases, then the glycosa- minoglycan extended by alternating additions of uronic acidandaminosugar.The divergence of these two classes of GAG is in the addition of thefirstaminosugar unit.15 As chain synthesis progresses, an aminosugar deacetylase/sulfotransferase removes acetyl groups from the GlcNAc residues and replaces these with sulfate groups, although some decoupling of these activities does occur, leading to free NH2 groups.16 The specificity of the subsequent enzyme, an epimerase actingoncarbon

5oftheglucuronicacidresiduesatthenon-reducingendofGlcNS residues, is the primary determinant of the various observed sulfation patterns, with less common patterns arising from its low rates of mis-recognition. Thisenzymecanactina reversible or irreversible mode, depending on the sequence context. The subsequent and final steps in heparan sulfate and heparin synthesis are sulfation by 2-O-, 6-O-, and 3-O-sulfotransferases. All of these modifying enzymes have been proposed to act as complexes, with the enzyme composition of a complex determining the nature of the final chain.17

Heparan sulfate adopts a helical conformation in solution, with the dimensions determined in part by the counter ion. Heparan sulfate chainscanself-associate under conditions of sufficiently high salt conditions, with chains showing higher af-

finity for those of the same charge density. Heparin, by contrast, tends to have an extended conformation, again a result of the many conformers of approximately equal energy available to iduronic acid as mentioned for dermatan sulfate.9 Specific

13 1.2. Glycosaminoglycans

-OOC OH -OOC OH O O O O O O HO O HO O OH HO OH HO N - AcHN O3SHN O O 4)- GlcA -β-(1,4)- GlcNAc -α-(1, 4)- GlcA -β-(1,4)- GlcNS -α-(1,

- - - OOC OSO - OOC OSO3 O 3 O O O O O HO O HO O OH HO OH HO - AcHN O3SHN O O 4)- GlcA -β-(1,4)-GlcNAc6S-α-(1, 4)- GlcA -β-(1,4)- GlcNS6S-α-(1,

-OOC OSO - -OOC OH O 3 O O O O O HO O HO O OH -O SO OH -O SO 3 - 3 - O3SHN O3SHN O O 4)- GlcA -β-(1,4)- GlcNS3,6S-α-(1, 4)- GlcA -β-(1,4)- GlcNS3S-α-(1,

OH -OOC OH O O O O O O HO O HO -O - OH HO OSO3 HO OOC - O3SHN AcHN O O 4)- IdoA -α-(1,4)- GlcNS -α-(1, 4)- GlcA2S -β-(1,4)- GlcNAc -α-(1,

OSO - O 3 OH O O O HO O O O - OH HO HO O OOC - - OSO - HO O3SHN OOC 3 - O O3SHN O 4)- IdoA -α-(1,4)-GlcNS6S-α-(1, 4)- IdoA2S -α-(1,4)- GlcNS -α-(1,

- OSO3 O OSO - O O O 3 HO O- O O -OOC OSO - O3SO HO O 3 -O SHN - OSO - HO S 3 OOC 3 - O O3SHN O 4)-IdoA2S-α-(1,4)GlcNS3,6S-α-(1, 4)-IdoA2S-α-(1,4)GlcNS6S-α-(1,

OSO - OH O 3 O O O O O HO O HO O - - -O SO - - -O SO OOC OSO3 3 OOC OSO3 3 - H2N O3SHN O O 4)-IdoA2S-α-(1,4)GlcNH3,6S-α-(1, 4)-IdoA2S-α-(1,4)GlcNS3S-α-(1,

Figure 1.4: Known disaccharide units in heparan sulfate and heparin, with the ar- chetypal disaccharides for the N- and S-regions indicated (refer Figure 1.5).

14 1.2. Glycosaminoglycans

S-region N-region

Figure 1.5: Domain organisation of heparan sulfate (upper) and illustration of its domain structures (lower). The relatively flexible N-regions contain mostly N- acetyl-glucosamine and little sulfation, the stiffer S-regions contain mostly 2-sulfated iduronic acid and 6- as well as N-sulfated glucosamine, whilemixedregionsrepresent the transitions between these.

15 1.2. Glycosaminoglycans recognition sequences exist in both heparin and heparan sulfate, with arginine in particular being important for binding interactions with these.6 An important ex- ample of such a sequence is that responsible for the anticoagulant activity of these molecules, a pentasaccharide with the sequence [GlcNS(Ac)6S-α-(1,4)-GlcA-β-(1,4)-

GlcNS3,(6)S-α-(1,4)-IdoA2S-α-(1,4)-GlcNS6S-α-(1,4)-] as shown in Figure 1.6 (the first glucosamine can be either N -sulfated or N -acetylated, and the second can be

with or without 6-O-sulfation). This must be contained within an oligosaccharide

of at least 12 to 16 monomers for full activity. The activity ofthismoleculearises

from binding to the blood coagulation factors thrombin and factor Xa, preventing

activation of the clotting cascade. 6

- OSO3 O O - HO (OSO3 ) -O SHN -OOC OH 3 O (AcHN)O O HO O OH -O SO 3 - - O3SHN OSO O 3 O O HO O - OSO - HO OOC 3 - O3SHN O -GlcNS(Ac)6S-α-(1,4)-GlcA-β-(1,4)-GlcNS3,(6)S-α-(1,4)-IdoA2S-α-(1,4)-GlcNS6S-α-(1,4)-

Figure 1.6: Minimal structure for heparin anticoagulant activity. Features in paren- theses show variants that retain activity.

Like all other classes of GAGs discussed, heparan sulfate hasimportantroles in the structure of many vertebrate tissues. However, heparin and heparan sulfate oligosaccharides also have a variety of signalling roles, such as in axonal growth, angiogenesis, and response to growth factors. 6

16 1.2. Glycosaminoglycans

1.2.5 Other uronic acid-containing polysaccharides

While not strictly GAGs, other polysaccharides exist in nature that contain a uronic acid in the repeating unit, making these amenable to degradation by polysaccharide lyases (see Subsection 1.4.4 on page 39) and so of relevance tothisthesis.Manyof the pathways for depolymerisation of these are thus related to those of GAGs.

The microbial gelling agents xanthan18 and the class of sphingans 19 (for example

gellan) both contain a repeating unit with glucuronic acid, among other sugars, in

the side-chain and the backbone respectively (Figure 1.7). Xanthan is a polymer of β- (1,4)-glucose (the same as cellulose), with every other glucose unit displaying a [Man-

α-(1,4)-GlcA-α-(1,2)-Man-α-(1,3)-] side-chain. The mannose unit proximal to the backbone can be modified by O-acetylation at carbon 6, while the distal mannose can

be modified by pyruvate, forming a (4,6) O-ketal. Gellan is a linear polysaccharide

with the sequence [Glc-β-(1,4)-GlcA-β-(1,4)-Glc-β-(1,4)-l-Rha-α-(1,3)]n .Examples of applications include use of both as thickening agents in food, use of gellan to

make particularly transparent gels for research, and use of xanthan in concrete to

increase its viscosity for use underwater.

Xanthan Gellan

4)- Glc -β-(1,4)- Glc -β-(1, 4)- Glc -β-(1,4)- GlcA -β-(1,4)- Glc -β-(1,4)- L-Rha -α-(1, OH O OH OH O O O -OOC O O O O OH OH HO O HO O O OH OH HO O OH OH HO O OH OH O Man-α-(1,3) O O OH OH O OH GlcA-α-(1,2) OH O OH O - COO OH OH O Man-α-(1,4) OH OH

Figure 1.7: Structures of xanthan and gellan.

Alginate is a polysaccharide from the cell wall of brown algae,20 and also pro- duced by some bacteria, that is capable of forming a viscous gum. It is composed of

17 1.2. Glycosaminoglycans

(1-4)-linked β-d-mannuronate and its C-5 epimer, α-l-guluronate, in discrete blocks of around 20 units, interspersed with short areas of alternating monomers (Fig-

ure 1.8).21 Some bacterially-sourced alginates are modified by acetylation.22 Like

xanthan and gellan, alginate is also used in food as a thickening agent, as well as a

number of specialised applications based on its rapid water absorption and physical

characteristics. 23

4)- ManA -β-(1, 4)- ManA -β-(1,4)- GulA -β-(1, 4)- GulA -β-(1,

-OOC OH O -OOC OH O O OH HO O O OH HO O O HO O O COO- HO COO- ~20 ~20

Figure 1.8: Structure of alginate.

Pectins are structural polysaccharides in plant cell walls.Theyarefurtherdivided into three classes, all of which contain uronic acids (Figure1.9andFigure1.10).24

Homogalacturonan, as the name might suggest, is composed solely of 1,4-linked α- d-galacturonic acid. This basic structure may be modified by formation of methyl esters on some carboxyl groups and O-acetylation at carbon 2 or carbon 3. Rham- nogalacturonans make up the other two classes, and are further divided into types I and II. Type I rhamnogalacturonan has a core chain of [GalA-α-(1,2)-Rha-α-(1,4)-]n. The galacturonic acid residues again may be O-acetylated at carbons 2 or 3, while the rhamnose units are heavily modified with linear or branched side-chains on the carbon 4 hydroxyl group (between 20 and 80%).24 The majority of these side-chains are composed of arabinose and galactose, while fucose and glucuronic acid may also be present. The type II rhamnogalacturonans are not at all similar to the type

I, as shown in Figure 1.10. The backbone of this is composed of 1,4-linked α-d- galacturonic acid, as for homogalacturonan, but unlike homogalacturonan it is short

18 1.2. Glycosaminoglycans rhamnogalacturonan I. Homogalacturonan: Rhamnogalacturonan I: Figure 1.9: Structures of the pectins homogalacturonan and

19 1.2. Glycosaminoglycans

Rhamnogalacturonan II:

O COO- O HO O O OR O O O B O O O RO O O OH O -OOC O

Figure 1.10: Structure of rhamnogalacturonan II core, with side-chain connections represented as blocks (upper), and an illustration of the cross-link formed between adjacent side-chain A moieties (lower), with one unit of the main galacturonic acid shown and the remainder of the side chain indicated by R.

20 1.3. Glycoside hydrolases (glycosidases) and heavily substituted with oligosaccharide side-chains.Onfourdifferentmonomers of this core are attached four different side-chains. Two of these side-chains attach to carbon 2 of a galacturonic acid, an octasaccharide and a nonasaccharide termed side-chains A and B, and two others attach to carbon 3 of a galacturonic acid, two disaccharides termed side-chains C and D. These rhamnogalacturonan II monomers are able to cross-link by formation of borate-diol esters through side-chain A.25

1.3 Glycoside hydrolases (glycosidases)

The enzymes responsible for hydrolysis of glycosides are termed glycoside hydrolases,

or glycosidases. Many of these are very efficient catalysts, with rate enhancements

on the order of 1017 fold over the non-enzymatic case.26 How these enzymes achieve

this catalytic feat has been the topic of extensive work over many decades, with the

latest insights capably covered in several recent reviews. 27–31

Glycoside hydrolases, along with the related glycosyl transferases, polysaccharide

lyases, carbohydrate esterases, and carbohydrate-bindingmodules,areclassifiedinto

families on the basis of sequence homology in the Carbohydrate Active Enzymes

(CaZY) database (available online at http://www.cazy.org/).32 For many of these

families the key mechanistic details have been elucidated, and representative crystal

structures of members have been solved. This database thus provides a framework

through which mechanistic similarities between these enzymes can be understood, as

well as reflecting structural features and evolutionary relationships. Conservation of

mechanistic features has also been found in some cases to extend beyond the family

classification, leading to the grouping of related families into clans.

Two general mechanisms have been found in most glycoside hydrolases, which

result in either a net retention or inversion of the stereochemistry at the anomeric

centre, and so the corresponding enzymes are called retaining and inverting glycosi-

21 1.3. Glycoside hydrolases (glycosidases) dases. The general features leading to these results were outlined in a seminal paper by Koshland, 33 with many more details being added over the subsequent years.For this reason, these are often referred to as Koshland mechanisms.

The inverting glycosidases catalyse hydrolysis by a single direct nucleophilic dis-

placement, with a water nucleophile attacking at the anomeric centre while the bond

to the nucleofuge is broken. Two key catalytic residues are involved in this mech-

anism; one acts as a base to deprotonate the water nucleophileandtheotheracts

as an acid to activate the nucleofuge. These are most often aspartate or glutamate

residues, one protonated and the other deprotonated in the resting state of the en-

zyme, with a spacing of approximately 10 Å to allow alignment of the nucleophile,

anomeric carbon, and nucleofuge between them in the active site. This reaction pro-

ceeds in a dissociative manner, with substantial formal positive charge developing

on the anomeric carbon, stabilised by lone-pair electrons oftheadjacentendocyclic

oxygen in an oxocarbenium-ion like transition state. These key mechanistic features

are illustrated in Scheme 1.1.

O δ- O O -O O HO H H HO HO δ- OH OH + O Oδ HO HO HO HO O HO + HO HO HO δ OH δ- HO OR OR OH

H HOR H O O δ- O-

O O O

Scheme 1.1: The general mechanism of inverting α-glucosidases.

Many of the mechanistic features of the inverting glycoside hydrolases are also present in the retaining glycoside hydrolases. These enzymes, however, catalyse a

22 1.3. Glycoside hydrolases (glycosidases) double inversion in order to achieve a net retention of stereochemistry at the ano- meric centre. Two key catalytic residues are again involved in this mechanism, but with slightly different roles, one acting as a nucleophile andtheotherasanacid/base catalyst, with a spacing of approximately 5.5 Å. The first inversion in the reaction se- quence is a replacement of the original nucleofuge in the glycoside by the nucleophile residue side-chain. This forms a covalent glycosyl-enzyme intermediate, with the acid/base residue protonating the nucleofuge to activate it. This first step is termed the glycosylation step. The glycosyl-enzyme intermediate is then hydrolysed, with a water nucleophile being deprotonated by the acid/base residue to activate it for nuc- leophilic attack, displacing the nucleophile residue and giving a hydrolysed product with the same anomeric configuration as the substrate glycoside. This second step is termed deglycosylation. Both steps pass through oxocarbenium-ion like transition states, similar to that of the inverting mechanism. Either step of this mechanism can be rate-determining, with some enzymes showing evidenceofachangeinrate- determining step with increasing activation of the nucleofuge in synthetic substrates.

These key mechanistic features are illustrated in Scheme 1.2.

Within the context of this general retaining mechanistic scheme, several vari- ations have been reported in both the nucleophile and acid/base residues.34 In the sialidases and trans-sialidases of clan GH-E the nucleophile residue is a tyrosine. This adaptation has been proposed to minimise charge repulsion between the nucleophile and the carboxylate at carbon 1 of the substrate sialosides, and also to provide a more stable intermediate for the generally more reactive sialosides (a 2-deoxy sugar), making its formation more favourable. Adjacent to the tyrosine nucleophile is a con- served carboxylate side-chain, potentially involved in abstracting the phenolic proton to improve nucleophilicity. Another variation in the nucleophile position is in en- zymes that cleave substrates with a 2-acetamido group adjacent to the leaving group,

23 1.3. Glycoside hydrolases (glycosidases)

O δ- O O O H HO H HO O HO OR δ+ HO OR HO O OH HO δ+ HO δ- O- O HOR O O

H2O

O -O

H HO OH O HO HO HO O

O

O δ- O O O H HO H HO O HO OH δ+ HO OH HO O OH HO δ+ HO δ- O- O

O O

Scheme 1.2: The general mechanism of retaining β-glucosidases.

24 1.3. Glycoside hydrolases (glycosidases) although not all glycoside hydrolases acting on such substrates show this mechanistic feature. This acetamide can act as an internal nucleophile, forming a non-enzyme- linked oxazoline or oxazolinium ion intermediate. An adjacent polarising residue is present in these enzymes, stabilising the intermediate either by charge interactions or by deprotonating it. An exogenous oxazoline analogue of the substrate has been shown to be catalytically competent in these enzymes, and thiazoline analogues have proven to be potent inhibitors.35 This variation may have evolved as a lowest-energy path for hydrolysis of these already-activated substrates.

Variations in the acid/base residue are less common, likely because of the lack

of other functional groups in amino acid side chains with an appropriate pKa.One example is found in the unusual inverting endo-sialidase from bacteriophage K1F,

for which it has been proposed that the carboxylate at carbon 1ofthesubstrate

sialosides acts as an internal catalytic base, with a glutamate residue acting as cata-

lytic acid.36 Another example comes from myrosinase, a retaining β-glucosidase that catalyses hydrolysis of plant thioglycosides.37,38 In this enzyme, the substrate leav-

ing group is already highly activated, meaning the formationofcovalentglycosyl

enzyme intermediate proceeds without acid catalysis. The subsequent deglycosyla-

tion, which is rate-limiting, is catalysed by the carboxylate group of an ascorbate

cofactor bound adjacent to the substrate. This may be functioning as a regulat-

ory system, releasing toxic aglycones from their benign glycosides when the plant

cell is in a state of poor redox homeostasis through cell lysis, as a defence against

herbivores. Finally, in a retaining exo-cellulase from family GH6 a lack of suitable

catalytic acid/base residue led to the postulation of a mechanism wherein a group of

nearby residues act through a network of water molecules indirectly to provide the

required protonation and deprotonation; a so-called Grotthus mechanism. 39 How-

ever, this is based on a homology model, so strong conclusionscannotbedrawn

25 1.4. Hydration and elimination mechanisms in glycoside cleavage based the placement of residues.

Afinalobservationontraditionalglycosidehydrolasesiswarranted in the context of the following section. It has been reported 40–43 that both α-andβ-retaining glucosidases are able to catalyse the stereospecific hydration of glucal. This hydration occurs by a syn-addition of the catalytic acid residue across the double bond of the substrate to give a covalently bound 2-deoxy-glycosyl-enzyme intermediate, followed by hydrolysis of this intermediate to give an overall trans-addition of water across the double bond, as illustrated in Scheme 1.3. This hydrationproceedsbyuseof the same catalytic residues and through a similar transitionstatetothatofthe hydrolysis reaction of these enzymes.

O O O -O -O O H HO HO O H O HO HO OH HO HO O HO OH HO H* HO H* O O- *H O O O O

Scheme 1.3: The mechanism for hydration of glucal by a retaining β-glucosidase. The proton transferred from the catalytic nucleophile is indicated as H* to emphasise the stereochemical result, as determined by NMR studies in D2O.

1.4 Hydration and elimination mechanisms in glycoside

cleavage

In addition to those using the inverting and retaining mechanisms outlined in Section

1.3, a number of enzymes have been reported to cleave glycosides by mechanisms

involving elimination, hydration, or both. These representasetofverydifferent

26 1.4. Hydration and elimination mechanisms in glycoside cleavage mechanisms, some with very different transition states, meaning that any attempts to inhibit or re-purpose these enzymes will likely require a very different set of strategies. Many, but not all, of these mechanisms have been observed in enzymes from bacterial sources.

1.4.1 Family GH4 and GH109 glycoside hydrolases

Both elimination and hydration reactions are found in family GH4 glycoside hydro- lases. These employ a fascinating mechanism wherein the hydroxyl group on carbon

3istransientlyoxidised,acidifyingtheadjacenthydrogenoncarbon2toallowan elimination across the carbon 2 to carbon 1 bond. In a second half-reaction, water at- tacks the α,β-conjugated system, followed by reduction of the carbon 3 ketone, giving an overall net hydrolysis of the anomeric bond, as shown in Scheme 1.4. This unique

and unusual mechanism, completely different from those of either the inverting or

retaining glycoside hydrolase mechanisms outlined in Section 1.3, proceeds through

aseriesofnegativelychargedintermediatesandtransitionstates.Thisreaction

sequence is achieved by the use of two obligate cofactors, a NAD+ to facilitate the

transient oxidation and a divalent metal, usually manganese, to stabilise the negative

charges developing during the course of the reaction. Evidence for this mechanism,

including crystal structures, kinetic isotope effects, and linear free-energy relation-

ships, was capably reviewed by Yip and Withers,44 with a focus here on work since

this publication.

The lack of involvement of the anomeric centre in this mechanism has allowed

this family to evolve to catalyse hydrolysis of both α-andβ-configured substrates. These reactions appear to proceed through the same general mechanism, but with

some differences in the residues used to achieve the key catalytic roles. Intriguingly,

the 6-phospho-α-glucosidase GlvA from Bacillus subtilis has been demonstrated to

27 1.4. Hydration and elimination mechanisms in glycoside cleavage ion. ible is O O - 2 OH H NH O O O HO H H H 2+ O N * H Mn H - O O PO 3 O 2 O O HO 2 H O 2- H (fast) (fast) HOR O O O O 2 2 H H H NH NH OR OH O O O O O O H H H H H HO HO H H 2+ 2+ O O N N * * Mn Mn H H H H O O - - O O PO PO 3 3 O O O O HO HO 2 2 O O 2- 2- H H -glucosidases, based on experimental results and calculat α . * (slow) rthoseinvolvingredoxtherelevanthalf-reactionrespons O O O O 2 2 H H NH NH OH OR O O O O O O HO HO H H H H H H H H 2+ 2+ O O N N * * H H Mn Mn H H - - O O O O PO PO 3 3 H O O O HO HO 2 HO O O 2- 2- H transfer) (slow) transfer) (slow) (slow) + - (H (H O O O O 2 2 H H NH OR OH NH O O O O O O HO HO H H H H - + 2+ 2+ + O O * N N H H H Mn Mn - - O O O PO PO HO 3 3 - - H O O HO HO O O 2- 2- HO HO Scheme 1.4: The general mechanism of family GH4 6-phospho- Partially rate-limitingindicated. steps The hydride are transferred indicated in redox is as denoted slow, by H and fo

28 1.4. Hydration and elimination mechanisms in glycoside cleavage cleave both α-andβ-glucosides within the same active site of the same enzyme, with similar kinetic parameters,45 provided the substrate contains a sufficiently activated aglycone. This anomerically blind reactivity is possible because the ability to bind substrate adjacent to the NAD+ is the primary determinant of hydrolysis, and agly- cone activation is less important. For less activated substrates, where aglycone pro- tonation becomes more important, the enzyme shows more specificity for its natural

α-configured substrates. GH4 enzymes have also been shown to cleave thio-linked substrates, for example the unactivated 4-deoxy-4-thio-d-cellobiose-6! -phosphate is cleaved by the β-glucosidase BglT from Thermotoga maritima with kinetic paramet- ers similar to those for the analogous O-glycoside.46 Incorporation of deuterium at carbon 2 of the glucose-6-phosphate product and determination that the Ki value equalled the Km provided evidence that this hydrolysis proceeded through the same active site and with the same mechanism as for hydrolysis of O-glycosides.

Much recent work has focused on the clarification of the rate-determining step(s) of family GH4, through determination of further kinetic isotope effects and linear free-energy relationships, as well as DFT calculations. Kinetic isotope effects in α- and β-glycosidases from GH445,47–50 generally agree well, and show that, in enzymes acting on both anomeric configurations, oxidation at carbon 3anddeprotonationat carbon 2 are both partially rate limiting. By contrast, glycosidic bond cleavage is shown to be kinetically unimportant by the lack of dependenceuponleavinggroup ability and the lack of a substantial kinetic isotope effect from deuteration at carbon

1. In MelA, an α-galactosidase from Citrobacter freundii,kineticisotopeeffects were measured on kcat/Km for substrates with deuterium at carbons 2 and 3, and these were compared to the effect on rate of substitutions on both carbons 2 and

3simultaneously.Thisdisubstitutionwasfoundtoshowanenhancement of the isotope effect by the same magnitude when compared to either substitution alone,

29 1.4. Hydration and elimination mechanisms in glycoside cleavage suggesting that these isotopes were influencing the same transition state. This result agrees with a concerted oxidation at carbon 3 and deprotonation at carbon 2 with no formation of a discrete ketone intermediate.49 It is unclear if this is a unique feature of MelA, or represents a more general rate determining step for these enzymes.

It is likely that a gradient exists between discrete oxidation/deprotonation and a concerted process, determined by minor variations in the active site. Interestingly, in this same enzyme the kinetic isotope effects on kcat (as opposed to kcat/Km, discussed above) were found to be unity for substitution at carbons 1 and 3 (i.e. kH/kD = 1.0), and inverse for substitution at carbon 2. By comparison, the 6- phospho-α-glucosidase GlvA exhibited normal KIEs for deuterium at carbons 2 and

45 3onbothkcat and kcat/Km, thereby suggesting that the overall rate-determining step and the first irreversible step were one and the same. Thissuggestsashiftin the overall rate-determining step in MelA to one of the steps after cleavage of the glycosidic bond. This step may be re-protonation at carbon 2,onthebasisofthe calculated reaction coordinate (vide infra), with the change in hybridisation from sp2 to sp3 consistent with the observed inverse effect.

An energy profile was calculated51 for the reaction catalysed by GH4 α-glycosidases, using the product bound crystal structure of the Bacillus subtilis 6-phospho-α- glucosidase GlvA 52 as a starting point. Overall, this profile was found to be surpris- ingly flat, with no single step being clearly rate determiningandallintermediates being lower in energy than the enzyme-bound starting material. On the basis of these calculations, the overall rate determining step is predicted to be re-protonation of carbon 2 following the Michael-type addition of water. However, the activation en- ergies for reprotonation of the carbon 3 hydroxyl of the product, deprotonation of carbon 2 in the substrate, abstraction of hydride from the substrate, and reproton- ation at carbon 2 of the product are all within 20 kJ.mol-1.Similarly,thebarrier

30 1.4. Hydration and elimination mechanisms in glycoside cleavage height relative to the starting material for reprotonation of the carbon 3 hydroxyl of the product, deprotonation of carbon 2 in the substrate, elimination of the agly- cone, and reprotonation at carbon 2 of the product are all within 17 kJ.mol-1.Given

these values, it seems likely that in any given GH4 enzyme there are likely several

steps that are at least partially rate-limiting, and subtle factors can easily influence

which steps these are. These values also provide an explanation for the difficulty in

assigning a single clear rate-determining step for the reaction catalysed by GH4 glyc-

osidases using kinetic isotope effects. Since many steps are very similar in energy, a

slight perturbation of any step can influence the overall rate. This calculated reac-

tion path also indicated a role for a manganese-bound hydroxyl group as a catalytic

base for the abstraction of a proton from the hydroxyl group oncarbon3duringthe

oxidation step and the involvement of aspartate 111 in activating tyrosine 265 as

acatalyticbasefordeprotonationofcarbon2,bymeansofanintermediary water

molecule. The oxidation step was found to proceed by an initial fast deprotonation

followed by a slower hydride transfer, while reduction was found to proceed in the

opposite order and with the opposite relative rates (fast hydride transfer then slow

protonation). Unfortunately, the model was built without a phosphorylated sub-

strate, which is the natural substrate of GlvA, so any effect from the phosphate on

the reaction coordinate remains unknown.

The identity of the catalytic base responsible for deprotonation of carbon 2 in

GH4 hydrolases acting on phosphorylated substrates, predicted on the basis of X-

ray crystal structures, has been experimentally confirmed as tyrosine 241 (BglT

numbering) by the careful measurement of kinetic isotope effects on kcat/Km for mutants at this position. 50 In a valiant set of experiments on deuterated substrates

with several mutants at varied pH, it was found that removal ofthisresidueablated

the kinetic isotope effect seen from substrates deuterated atcarbon3,whilethe

31 1.4. Hydration and elimination mechanisms in glycoside cleavage primary kinetic isotope effect from deuterium at carbon 2 is strengthened. This demonstrates that proton abstraction from carbon 2 has become completely rate limiting, a result of the retardation of this step by the removal of its catalytic base.

This effect was enhanced at lower pH, where deprotonation is generally more rate limiting. By contrast, at higher pH the isotope effect from deuterium at carbon 2 is decreased while deuterium substitution at carbon 3 remained silent, suggesting that in the presence of adequate exogenous base a later step inthereactionbecomes partially rate-limiting, but is silent in these kinetic isotope effects. Again, on the basis of the DFT calculations, this may be the re-protonationatcarbon2.An alternative explanation is that the Michael-type addition of water becomes rate- limiting, on the basis of a small inverse isotope effect that was observed for the

Y241A mutant of BglT at its optimal pH. Unfortunately, limitedsuppliesoflabelled compound prevented determination of effects at other pH values. Interestingly, the rate-determining nature of proton abstraction from carbon 2meansthatduring steady state turnover the enzyme has a constitutively bound NADH, rather than the

NAD+ of the resting state. The accumulation of this reduced cofactor on-enzyme could be observed spectroscopically for the Y241A mutant, since it binds this cofactor in a particularly tight fashion.

GH109, a new family of glycoside hydrolases that appear to follow a very similar mechanism to that of GH4, was recently founded on the basis of an α-GalNAc-ase discovered in Elizabethkingia meningosepticum.53 This enzyme showed a dependence on NAD+ for its catalytic activity, similar to GH4, but a surprising lack of require- ment for a metal cofactor. In GH4 enzymes this metal coordinates to hydroxyls on carbons 2 and 3, one of which is substituted by an N -acetyl group in GalNAc, so an- other means of activating the hydroxyl proton on carbon 3 is likely being employed in GH109. How this enzyme is able to catalyse a similar reaction, with turnover

32 1.4. Hydration and elimination mechanisms in glycoside cleavage equal to or better than that seen in GH4, is likely to be the subject of much study.

This enzyme, in contrast to the 6-phospho-α-glucosidase GlvA, only very slowly hy- drolysed activated substrates of non-optimal anomeric configuration or without an

N -acetyl moiety at carbon 2. Evidence for a similar redox/elimination/hydration mechanism came from the presence of a tightly-bound NAD+ cofactor appropriately positioned beside the substrate in the crystal structure, incorporation of deuterium at carbon 2 of the product when the reaction was carried out in D2O, and digestion of thio-linked substrates. A further unusual feature in thisenzymeisthelackofa

clear candidate in its crystal structure for the acid/base residue that activates the

aglycon for departure and water as a nucleophile.

Accurate prediction of the substrate preference of GH4 enzymes, of which there

are many in both the Archaeal and Bacterial domains, on the basis of sequence

identity and phylogenetic relationships was found to be difficult.54 The presence of

afouraminoaciddomainbeginningwiththemetal-ligatingactive-site cysteine, was

found to correlate with the substrate preference of the enzyme. However, mutation of

one such domain to another was not found to change substrate preference but rather

to ablate all activity, showing that these motifs are necessary but not sufficient for

each type of activity, with the active site context also playing a large role. This

finding may prove useful in predicting the substrate preference of newly identified

GH4 enzymes, but without further understanding of the effect of context is less useful

in any attempts at rationally redesigning the activity of these enzymes.

1.4.2 Elimination and hydration in sialidases

Asmallnumberofsialidaseenzymesfrombacterialandviralsources have been

observed to catalyse elimination55–58 and/or hydration56–59 reactions of sialosides, forming or degrading the sialidase competitive inhibitor Neu5Ac2en (also known as

33 1.4. Hydration and elimination mechanisms in glycoside cleavage

O O O- O H OH OH HO H O HO O HOR H O O 2 COO- O COO- AcHN AcHN HO HO HO HO OR H O H H H O O O O-

O O- OH O HO H COOH O AcHN HO HO H O O

O O O- O H OH H OH HO O HO O O COO- O COO- AcHN AcHN HO HO HO HO OH H O H H H O O O O-

Scheme 1.5: The general mechanism of elimination and hydration by sialidases. A given sialidase may catalyse either half reaction in isolation, both half reactions, or neither (giving only hydrolysis).

34 1.4. Hydration and elimination mechanisms in glycoside cleavage

2,3-dehydro-2-deoxy-N -acetylneuraminic acid, DANA). The mechanism for this is outlined in Scheme 1.5. Elimination proceeds through the same glycosyl-enzyme in- termediate as the hydration reactions catalysed by these enzymes. However, proton abstraction from carbon 3, to trigger elimination, competeswithhydrolysisatcar- bon 2 to varying degrees. A similar mechanism is also responsible for hydration of

Neu5Ac2en, with formation of the same glycosyl-enzyme intermediate preceding hy- drolysis to give overall hydration. This reaction is similartothehydrationofglucal by retaining glycosidases (refer to the end of Section 1.3), although notably tyrosine is unable to undergo a concerted syn-addition across the double bond as seen for as- partate/glutamate. The balance of elimination and hydration is thus determined by access of the water nucleophile to the anomeric carbon and theprotononcarbon3, as well as competition between substrates (Neu5Ac2en for hydration and sialosides for the formation of hydrolysed product and/or Neu5Ac2en). In most cases these elimination and hydration reactions represent minor pathways for hydrolases, and are correspondingly slow, but in one case it is the primary reaction catalysed and

-1 58 is reasonably efficient (kcat of 19 min in NanC from Streptococcus pneumoniae).

Evidence for this mechanism came from reaction in D2O, showing an overall trans- addition of water, catalytic competence of synthetic Neu5Ac2en for hydration, and

alackofreactionwiththio-sialosides,arguingagainstdirect elimination without a

covalent intermediate.58 Given that Neu5Ac2en is a reasonably potent inhibitor of

sialidases, its formation and decomposition has been proposed to be a regulatory

system for sialidase activity.58,60

1.4.3 α-1,4-Glucan lyase

An elimination mechanism, but no hydration, has also been found to be acting

in some enzymes of family GH31, the α-1,4-glucan lyases. As the name suggests,

35 1.4. Hydration and elimination mechanisms in glycoside cleavage these cleave the glycosidic bond of α-(1,4)-linked polymers through an elimination mechanism, and are predominantly involved in the metabolismofenergystorage polysaccharides such as glycogen and starch.61 The first step in this mechanism is the

displacement of the α-linked anomeric group by a catalytic nucleophile, in much the

same manner as the first step in the mechanism of a retaining α-glycosidase. However, the glycosyl-enzyme intermediate thus formed then undergoes a syn-elimination of

an adjacent proton and the nucleophile, rather than a second displacement by water,

to form anhydrofructose. On release from the enzyme active site, this tautomerises

to its keto form, and then hydrates to give a geminal alcohol. 62 The product of these

lyases can be used directly in energy metabolism, 61 or alternatively can be hydrated

by α-glucosidases from GH13 or GH31 to yield glucose in what is proposed to be an anhydrofructose scavenging system. 63 The mechanism for this hydration has not

been investigated, but presumably mimics that of glucal hydration (refer to the end

of Section 1.3. The mechanism of α-(1,4)-glucan lyases and the subsequent product rearrangements are shown in Scheme 1.6.

This mechanism uses the same basic catalytic machinery, and indeed the same

first step, as the α-glucosidases of the same family. The formation of the covalent

glycosyl-enzyme intermediate was observed by trapping with5-fluoro-β-l-idosyl flu- oride and detection of the labelled peptide by mass spectrometry.64 Interestingly, trapping could not be observed with 1-, 2- or 5-fluoro-α-d-glucosyl fluorides, which instead acted as slow substrates.65 The formation of this covalent glycosyl-enzyme intermediate was seen to be rate limiting, with a strong α-secondary kinetic isotope effect from 1-[2H] substrates showing a dissociative transition state, and ashallow

negative slope in the linear free energy plot of both kcat and kcat/Km showing sub- stantial proton donation to the leaving group. Consistent with this, substitution

with deuterium at carbon 2 revealed only a small β-secondary effect. For the partic-

36 1.4. Hydration and elimination mechanisms in glycoside cleavage

O O δ- HO O- O O HO HO δ+ HO HO O HO HO OR HO δ+ OR HOR H H δ- O O

O O

HO O H O HO O HO OH

O-

O

O O H δ- O δ- HO O O HO HO δ+ HO HO H O HO OH HO δ+

O- O-

O O

HO HO HO HO GH13/ + H2O O GH31 O +/- H O HO O HO HO HO HO HO H2O HO HO HO HO O H2O OH OH OH

Scheme 1.6: The general mechanism of α-(1,4)-glucan lyases (upper), with rearrange- ments and further reaction of the product in water also shown (lower).

37 1.4. Hydration and elimination mechanisms in glycoside cleavage ularly slow substrate 5-fluoro-α-d-glucosyl fluoride the deglycosylation/elimination step was found to be rate limiting, and a small primary kineticisotopeeffectwasseen from a 2-[2H] substrate, while a 1-[2H] substitution showed a fairly large α-secondary effect. Together these suggest that the proton abstraction step occurs in a concerted but asynchronous process. In other words, proton abstraction only occurs after sub- stantial bond cleavage at the anomeric carbon to generate an oxocarbenium ion-like transition state, the charge of which increases the acidity of the proton on carbon 2.

The catalytic base for this proton abstraction has been suggested to be the cata- lytic nucleophile itself. The asynchronous nature of this elimination makes this plausible, as at the transition state the glycosyl-enzyme bond has largely broken to give a proton, acidified by an adjacent carbocation, with a carboxylate in close proximity. Indeed, the placement of the carbonyl oxygen in GH31 α-glucosidases is over the proton on carbon 2 of glucose.66 Such a syn-elimination is the reverse of the syn-addition of the catalytic nucleophile in glucal hydration (Section 1.3).

Asimilareliminationhasbeenobservedintheformationandsubsequent hydra- tion of a minor product d-ribal in the 2-deoxy-ribosyltransferase from Lactobacillus leishmanii,67 and also in the observation of 1,5-anhydro-d-arabino-hex-1-enitol, an isomer of anhydrofructose, in the active site of an X-ray crystal structure of glycogen synthase from Escherichia coli,68 both formed by glycoside-synthesising enzymes in the absence of an appropriate acceptor. In the first case the same enzyme is re- sponsible for both the elimination and hydration reactions,emphasisingtheimplied relationship between the mechanisms of α-(1,4)-glucan lyases and glucal hydration by glycoside hydrolases. In the latter case the observed 1,5-anhydro-d-arabino-hex-

1-enitol did not seem to be catalytically competent, suggesting this is an off-path intermediate formed from the highly activated glycosyl donor in the absence of any suitable acceptor. If this assignment of the nucleophile residue as the catalytic

38 1.4. Hydration and elimination mechanisms in glycoside cleavage base for its own elimination is correct, this provides a convenient explanation for the lack of trapping observed with 1-fluoro-α-d-glucosyl fluoride, which is instead a slow substrate.65 These were anticipated to undergo a trans-elimination of HF from the glycosyl-enzyme intermediate to form a stable covalent species, as illustrated in

Scheme 1.7. However, if the departing nucleophile itself is the catalytic base for this step then the catalytic base would never be present at the sametimeasthecovalent glycosyl-enzyme intermediate, rendering such a trapping scheme impossible. While this explanation is appealing, other anticipated trapping reagents for this enzyme also did not provide trapped covalent glycosyl-enzyme intermediates, including the

C5 epimer of the successful 5-fluoro-α-d-idosyl fluoride, indicating that other explan- ations are also plausible.

B: B: BH+ BH+ O HO O- HF HO O HF HO O HO O H O O O +/- H+ O HO HO O HO O HO O HO F HO HO HO HO HO O F F HO

H O O- O- O-

O O O O

Scheme 1.7: The anticipated mechanism of inactivation of α-(1,4)-glucan lyases by 1-fluoro-α-d-glucosyl fluoride, if a second residue were to be acting as catalytic base.

1.4.4 Polysaccharide lyases

Polysaccharide lyases are a class of enzymes that catalyse aneliminationreaction that is superficially similar to that of the α-(1,4)-glucan lyases, but the mechanism by which this is achieved is very different. These enzymes are classified separately in the CaZY database, and substantially fewer of these familiesareknowncomparedto the glycoside hydrolases (22 compared to 131, respectively,asofMarch2013).The general mechanism of this class of enzymes involves three keycomponents.69 The

39 1.4. Hydration and elimination mechanisms in glycoside cleavage

O- O- O H O H O δ- O O O O O RO RO HO OR' HO OR' OH OH H H

-O O

O- O H O -O O RO HO OR' OH H O

O- O- O H O O H δ- O O O O RO O HO OR' HO OR' OH OH H -O O

Scheme 1.8: The general mechanism of polysaccharide lyases acting on pectate.

40 1.4. Hydration and elimination mechanisms in glycoside cleavage

first is the neutralisation of charge from the carboxylate at carbon 6, presumably occurring immediately on binding of substrate to the enzyme active site, followed by abstraction of the proton at carbon 5 to form a carbanion intermediate with resonance stabilisation by the adjacent carboxylate. Finally, the free electron pair forms a double bond between carbons 4 and 5, expelling the leaving group at carbon

4intheprocessandgivinganunsaturatedglucuronideastheproduct. This overall

E1cb reaction mechanism is outlined in Scheme 1.8. Because of themechanistic involvement of the carboxylate group, all substrates for polysaccharide lyases must contain this moiety. Indeed for some polyanionic saccharides, degradation by lyases is the only known path for catabolism.70 This mechanism is very similar to that of

the C5-epimerases acting on the same substrates, differing inthenatureofthefinal

step — protonation either occurring on the carbon 4 oxygen to give elimination or

on carbon 5 from the opposite face to give epimerisation. 69

On the basis of a set of experiments using chemically defined short substrates

with chondroitin AC lyase from Flavobacterium heparinum,71–74 the rate determining

step for this mechanism was found to be abstraction of the proton from carbon 5 of

the substrate.71 When this proton was substituted with deuterium a small primary

kinetic isotope effect was observed, while variation of the leaving group at carbon

4wasfoundtohavenoeffectontherateofreaction(forasetofactivated aryl

substrates). Consistent with this, a deuterium substitution at carbon 4 resulted in no

significant kinetic isotope effect (kH/kD =1.0).Deuteriumwasnotobservedtohave exchanged at carbon 5 in a partially-cleaved sample, suggesting that deprotonation

is irreversible, but may also indicate solvent inaccessibility of the active site during

catalysis (elimination is faster than reprotonation or solvent exchange). A corollary

of this E1cb mechanism is that unactivated thio-linked substrates areturnedover by these enzymes, but only very poorly. 75

41 1.4. Hydration and elimination mechanisms in glycoside cleavage

The various structural features of polysaccharide lyases have recently been very thoroughly reviewed by Garron and Cygler.76 These enzymes can generally be grouped into one of two types, depending on the residues responsible for the catalytic func- tions of charge neutralisation, base catalysis, and acid catalysis. One group, con- taining almost exclusively pectate and pectin lyases, employs a divalent metal for charge neutralisation, an arginine or a lysine as a catalyticbase,andwaterasacata- lytic acid. An example is the pectin lyase from Cellvibrio japonicus,whichcontains asinglecalciumbridgingthecarboxylatesofthe+1and-1subsites, an arginine catalytic base, and no clear acid residue, although a water ligand of the calcium ion or an aspartate acting via the carbon 3 hydroxyl were proposed as candidates for this role.77 Because of the high charge density of pectate, a poly-galacturonide, these enzymes may contain as many as four divalent metals. 78 By contrast, in en- zymes acting on pectins with methyl esters, many sites homologous to these metal binding sites are poorly conserved, reflecting the lack of a requirement for charge neutralisation in these substrates. The other group of polysaccharide lyases is more diverse, but generally employs an amide or acid side chain to protonate the sub- strate carboxylate, a histidine or tyrosine as the catalyticbase,andatyrosineasthe catalytic acid. In syn-eliminating enzymes, a single tyrosine can act as both acid and base.74,79 An interesting case within this class is exemplified by the chondroitin

ABC lyases from Bacteriodes thetaiotaomicron and Proteus vulgaris. 80 These en- zymes are able to catalyse both syn-andanti-elimination in the same active site. It appears that binding of each substrate type (glucuronides oriduronides)recruitsthe appropriate catalytic machinery for its degradation through conformational shifts, with a single substrate binding domain forming a part of two partially overlapping active sites. For syn-elimination, a single tyrosine acts as both base and acid, ashas been seen in the related chondroitin AC lyases.74 For the anti-elimination the same

42 1.4. Hydration and elimination mechanisms in glycoside cleavage tyrosine still acts as a catalytic acid, but the role of catalytic base is filled by a pair of histidine residues that are located 12 Å away in the enzyme’s resting state. A metal ion is also involved in the active site of this enzyme, and appears to mostly be important for turnover of dermatan sulfate, which contains iduronic acid residues and thus requires the catalytic residues for anti-elimination .

1.4.5 Unsaturated glucuronyl and galacturonyl hydrolases

Unsaturated glucuronyl and galacturonyl hydrolases (UGL and UGH, respectively)

degrade the products released by polysaccharide lyases by use of a hydration reac-

tion. UGL are the primary family of enzyme studied in this thesis. These were

identified in Bacillus sp. GL1 in a pathway for total degradation of xanthan (for the structure of this substrate, see Subsection 1.2.5),81 and have since been cloned

and expressed from a variety of bacteria. 82–85 These were initially thought to be a specialised group of hydrolases operating by a Koshland mechanism,86 but struc-

tural and limited biochemical evidence has suggested otherwise.87 The mechanism

based on this evidence comprises an initial hydration of the double bond between

carbons 4 and 5, followed by rearrangement of the hemiketal product, through an

intermediate hemiacetal, to cleave the glycosidic bond and afford a free unsaturated

uronic acid (a α-keto acid) as shown in Scheme 1.9. The free unsaturated glucuronic acid thus liberated is further catabolised in a pathway in which it is reduced and

tautomerised at carbon 2 over two steps, phosphorylated at what was carbon 1

(now 6, from a change in numbering), and finally cleaved by an aldolase to afford

two common metabolites, pyruvate and d-glyceraldehyde-3-phosphate, as shown in

Scheme 1.10.88 The expression of an unsaturated glucuronyl hydrolase and polysac-

charide lyase pair, along with a phosphoenolpyruvate-dependent phosphotransferase

system for import of unsaturated glucuronic acid disaccharides, are necessary for

43 1.4. Hydration and elimination mechanisms in glycoside cleavage growth of Streptococcus pneumoniae on hyaluronic acid as a sole carbon source. 89

O HN O HN

N - N O H O H H COO- COO- O H + O H O O HO OR O HO OR O OH OH O HN O H N O H N O- 2 O- 2 N OH -OOC O H O O HO OR NH2 OH O O-

O HN O HN

OH N OH N -OOC ROH -OOC H O O O O O O HO NH2 HO OR NH2 OH OH O O O- O-

- - COO H+ COO O OH (glycosidase) B: O O -OOC HO OR HO O O OH OH H OH

Scheme 1.9: The general mechanism proposed for unsaturated glucuronyl hydrolases (upper). Also shown is a rearrangement to form the same final product following the hypothetical cleavage of a ΔGlcA derivative by a glycoside hydrolase mechanism (lower).

Two related GH families have been identified that operate by this same gen- eral mechanism; GH88,83 containing unsaturated glucuronyl hydrolases (UGL), and

GH105,84 containing unsaturated galacturonyl hydrolases (UGH, alsoknownasURH for unsaturated rhamnogalacturonyl hydrolase). These names refer to the source polysaccharide of the substrates for each enzyme, with UGL substrates being de- rived from polymers containing β-glucuronide monomers in the repeating unit and

UGH substrates being derived from polymers containing α-galacturonide monomers in the repeating unit.90 These polymers are primarily glycosaminoglycans for UGL

44 1.4. Hydration and elimination mechanisms in glycoside cleavage

4-deoxy-5- 2-dehydro-3- keto-uronate deoxy-D-gluconate O OH isomerase O OH 5-dehydrogenase O OH

HOOC O HOOC OH HOOC OH OH O OH 2-keto-3-deoxy -D-gluconate kinase

2-keto-3-deoxy- O O 6-phosphogluconate O OH + aldolase 2- 2- HOOC OPO3 HOOC OPO3 OH OH

pyruvate D-glyceraldehyde- 3-phosphate

Scheme 1.10: Bacterial metabolic pathway for catabolism of free ΔGlcA to the common metabolites pyruvate and d-glyceraldehyde-3-phosphate. The site of modi- fication for each step is in red for emphasis. and pectins for UGH. Because the 4,5-unsaturated glucuronideandgalacturonide products are the same compound, with the distinction at carbon 4 being lost on formation of the double bond, the actual distinction in the substrate specificities of these enzyme families is the stereochemistry of the anomericbond.UGLarethusα-

ΔGlcAses, with pseudo-equatorial anomeric constituents, and UGH are β-ΔGlcAses,

with pseudo-axial anomeric constituents (the reference stereo-centre for α/β nomen- clature also being affected by the double bond between carbons4and5—toavoid

confusion, the anomeric configuration of unsaturated glucuronides will often be re-

ferred to as axial or equatorial in this thesis, derived from α-andβ-glucuronides respectively). Because of this specificity, enzymes from mammalian pathogens and

symbiotes predominate in GH88 and enzymes from plant pathogens and symbiotes

predominate in GH105, but many cases of both are found in soil bacteria for decom-

position of dead organic matter.90

AlargeproportionoftheworkcarriedoutonUGLandUGHhasfocused on

X-ray crystallography, with an excellent series of structures being determined by

45 1.4. Hydration and elimination mechanisms in glycoside cleavage

Hashimoto, Mikami, Murata and co-workers at Kyoto University. These have been determined for UGL from a non-pathogenic Bacillus strain with an inhibitor bound

(glycine),91,92 with substrates bound,87,93 and in apo-form,93 and also from a patho- genic Streptococcus species both in apo-form85 and with substrate bound.94 Simil- arly, structures for UGH have been determined in apo-form, with an inhibitor bound

(UGL substrate in the +1 subsite),84 and with substrate bound.95 The active sites

of these are very similar, with catalytic residues overlaying well. In the resting state,

the two catalytic aspartate residues hydrogen bond to one another, likely accounting

for the high pKa of the second residue. On substrate binding, the catalytic acid/base residue rotates through 70° to position itself adjacent to carbon 4 of the substrate.

No residue is positioned close to the anomeric oxygen. Stronginteractionsareseen

with the ΔGlcA in the -1 subsite, while less interactions take place in the +1 subsite. This gives UGL and UGH fairly wide substrate specificities, the strongest require-

ment being for a ΔGlcA moiety. An example active site structure of UGL with a hyaluronan substrate disaccharide bound is shown in Figure 1.11.

Aside from crystallography, little mechanistic work has been reported for GH88

or GH105. Two experiments were performed when the mechanism was first proposed,

showing a solvent kinetic isotope effect of 2.1 and 2.2 on kcat and kcat/Km,respect- ively, and incorporation of 18Oatcarbon5oftheproductbyGC/MSfollowing

18 reaction in H2 O. However, these SKIE values are consistent with many alternat- ive mechanisms, and incorporation of 18Oatcarbon5oftheproductispossibleby non-enzymatic means, for example by hydration/dehydrationoftheproductcarbon

5ketoneinwater(seeScheme1.9,lowerpanel).Somemutagenesis of conserved residues in the active site of the Bacillus sp. GL1 UGL has also been carried out, as summarised in Table 1.1. Conservative mutation of two aspartate residues (D88 and

D149, all residue numbers in this section from Bacillus sp GL1 UGL)92 was found

46 1.4. Hydration and elimination mechanisms in glycoside cleavage

Figure 1.11: X-ray crystal structure of the Bacillus sp. GL1 UGL D88N active site from two perspectives (top and looking at the ΔGlcA anomeric bond), showing an unsaturated hyaluronan disaccharide substrate bound and all side-chains within 5 Å(glycineandalanineomitted).Thesubstrateiscolouredbyatom(darkgrey, carbon; red, oxygen; blue, nitrogen), while amino acid side-chains are coloured by type: light grey, non-polar; yellow, polar; red, negativelycharged;blue,positively charged. (PDB: 2FV1)

47 1.4. Hydration and elimination mechanisms in glycoside cleavage to have a drastic effect on turnover, with little associated change in binding to the enzyme. Mutation of an arginine (R221)87 adjacent to the ΔGlcA carboxylate, ini- tially thought to be responsible for neutralisation of its charge, was found to have no effect on substrate binding. A histidine residue (H193)87 appears to be much more important for binding, but the main source of charge neutralisation for the substrate carboxylate was instead proposed to be the positive end of an inner α-helix dipole. Finally, a glutamine residue (Q211)87 was also found to be somewhat important for turnover. Table 1.1: Michaelis-Menten kinetic parameters for previously characterised Bacillus sp. GL1 UGL mutations, using unsaturated gellan tetrasaccharide as substrate.

kcat/Km -1 Enzyme kcat (s ) Km (µM) (rel., %) Wild type92 7.3 90 100 D88N92 0.00057 200 0.0036 D149N92 0.0059 60 0.12 R221A87 4.2 69 75 H193A87 0.42 980 0.53 Q211A 87 0.20 170 1.4

Based on the structures published, and the placement of residues within them, one aspartate (D149) has been proposed to act as the proton donor in the initial hydration step, and as the base in the subsequent addition of water. 85,87,94,95 An- other catalytically important aspartate, D88, was proposedtohydrogenbondtothe hydroxyl groups on carbons 2 and 3, preventing their interference with the role of

D149 by hydrogen bonding, and also to stabilise the oxocarbenium ion-like transition state for the hydration reaction. D88 was also proposed to modulate the acidity of

D149, ensuring it is protonated through the hydrogen bond seen in the apo-form.93

However, this role seems inadequate in explaining the large effect seen on turnover when this residue is mutated to asparagine. If the role of D88 is in aiding D149, its mutation to an amide should not have a larger effect on turnover than the same

48 1.4. Hydration and elimination mechanisms in glycoside cleavage mutation in the acid itself. Moreover, an amide at this position should still be able to form hydrogen bonds, and D88 is located on the opposite sideofthehexosering to the site of charge development, making any direct transition state stabilisation of this charge implausible. The role of this residue thus remains unclear.

1.4.6 N -Acetyl-muramic acid 6-phosphate hydrolase (MurQ)

While not strictly a glycoside hydrolase as it is cleaving an ether, the enzyme N -

acetyl-muramic acid 6-phosphate hydrolase (MurQ) catalyses an elimination and

hydration reaction in a monosaccharide, as shown in Scheme 1.11, and so warrants

mention here. This enzyme is involved in the scavenging and recycling of bacterial

cell wall components, and is required by Escherichia coli for growth on N -acetyl-

muramic acid as a sole carbon source, 96,97 forming GlcNAc-6-phosphate by cleavage

of the lactate ether at carbon 3 of MurNAc-6-phosphate. Interestingly, this enzyme

appears to have undergone duplication and fusion with itself, along with loss of most

but not all etherase activity, to become a mammalian glucokinase regulatory protein

that responds to glucose- and fructose-6-phosphate, using the original etherase active

site.98

In the mechanism of this enzyme,99 the aldehydic tautomer of the monosacchar-

ide substrate is used to activate the proton at carbon 2 for abstraction by a catalytic

base residue. The enolate product of this is then able to rearrange to cleave the

ether bond at carbon 3 with assistance from a catalytic acid residue, in a similar

manner to the mechanism of the polysaccharide lyases outlined in Subsection 1.4.4.

Following exchange of the leaving group for water, a Michael type addition to the

α/β unsaturated intermediate then occurs, with deprotonation by the former cata- lytic acid residue. In a final step the proton at carbon 2 is thenreplacedbythe

former catalytic base residue to give the hydrolysed product, which is then able to

49 1.4. Hydration and elimination mechanisms in glycoside cleavage

:B +HB 2- 2- 2- O3PO O3PO O3PO O H OH OH HO HO HO O OH O O O O- NHAc NHAc NHAc lactate H H - - - CO2 CO2 A CO2 A H2O

2- +HB O3PO OH HO O O H H NHAc A- + :B H B 2- 2- 2- O3PO O3PO O3PO O H OH OH HO HO HO HO OH HO O HO O- NHAc NHAc NHAc H H A A

Scheme 1.11: The general mechanism for N -acetyl-muramic acid 6-phosphate hy- drolase (MurQ). re-tautomerise to its cyclic form on release from the enzyme.Thismechanismis very similar to that established for the rapid non-enzymaticthermaldecomposition of the disaccharides Gal-(1,3)-GalNAc and Gal-(1,3)-GlcNAc at neutral pH, with a half-life on the order of minutes. This proceeds through an elimination half reaction to the same α/β unsaturated intermediate, but instead of re-hydrating thisforms intramolecular adducts.100 MurQ may have an important role in accelerating the

hydration reaction to prevent formation of such off-path products.

Evidence for this mechanism comes from an excellent set of isotope exchange,

NMR, and mutagenesis experiments.99 Abstraction of the proton on carbon 2 was

shown through the observation of its exchange with deuteriumwhentheMurQ-

18 catalysed reaction is carried out in D2O, while reaction in H2 Ogaveincorporation of 18Oatcarbon3,butnotcarbon1.Thissecondobservationindicated that a

Schiff’s base, a common enzymatic strategy for activating aldehydes and ketones,

50 1.5. Thesis aims likely does not form. A small primary kinetic isotope effect was also observed using asubstratedeuteratedatcarbon2,indicatingthatthisstepisrate-limitingforthe

first half-reaction. The α/β unsaturated intermediate was observed to accumulate in solution during the course of the reaction, showing that hydration is the slower of the two half-reactions. Finally, a homology model of MurQ built using the structure of a distantly related protein of unknown function, solved inastructuralgenomics project but never published, indicated two aspartate residues that could fill the roles of the two acid/base catalysts. Mutants of each of these were indeed found to be substantially reduced in activity. For one of these, D83, deuterium exchange at carbon 2 persisted in the mutant, suggesting that this is the catalytic acid residue responsible for the elimination reaction, and the other, D114, is thus likely the initial catalytic base residue.

1.5 Thesis aims

Given that unsaturated glucuronyl hydrolases are involved in the degradation of the extracellular matrix of mammalian tissue, and the role of this degradation both in providing an energy source and in removing a physical barriertobacterialmotility, they are clear pathogenicity factors. This family of enzymesrepresentsaparticu- larly promising target for therapeutic agents with minimal side-effects as they are not produced by vertebrates, and appear to be acting through acompletelynovel mechanism for glycoside hydrolysis. An eventual long-term goal of this work is to furnish either an inhibitor or inactivator of UGLs, allowingtheselectiveremovalof this activity in infected tissue, thus slowing or even halting its spread. In order to fa- cilitate design of such a bacteriostatic agent, a clear understanding is required of the exact reaction catalysed and the mechanism by which this is achieved. Furthermore, the vast majority of glycoside hydrolysing enzymes that havebeencharacterisedop-

51 1.5. Thesis aims erate through either the inverting or retaining mechanisms of Koshland,33 with very

few variations on this. The only other completely novel mechanism for glycoside

hydrolysis that has been established to date is that of the related families GH4 and

GH109. If the proposed hydration mechanism is shown to be correct it will repres-

ent a second, although with much more limited scope given its requirement for an

unsaturated uronide.

The first goal of this thesis is thus to clearly show that UGL does indeed catalyse ahydrationreaction.Whilesomeevidencehasbeenpublishedforthisreaction,87 alternate interpretations of the data remained possible. Following confirmation of the general reaction catalysed, a more in-depth study will beundertaken,aimingto characterise the rate determining step of the reaction and the structure of its associ- ated transition state. Knowledge of the charge and geometry of the transition state is essential for guiding design of a strongly binding competitive inhibitor that mimics these features, as enzymes are known to bind very strongly to the transition state in order to catalyse their reactions. This will be achieved by careful characterisation of

UGL reaction products, measurement of a linear free-energy relationship, determin- ing the effects of heteroatom substitution in substrates, kinetic isotope effects, and

finally the testing of some potential inhibitors and inactivators of UGL.

Afurthergoalofthisthesis,relatedtotheworkonUGL,istoinvestigate the mechanism of mammalian heparanase (HPSE). This is an enzyme responsible for degradation of heparan sulfate and heparin during tissue remodelling and growth, and has been strongly implicated in cancer metastasis. This thesis presents work to- wards the development of a convenient fluorogenic assay of HPSE, and an attempt at development of a 2-deoxy-2-fluoro-glucuronide inactivatorbothasaleadcompound for treatement and in order to allow confirmation of the catalytic nucleophile.

52 Chapter 2

Confirmation of the hydration reaction

In the simple hydration-initiated mechanism presented in Scheme 1.9 on page 44,

UGL is proposed to catalyse hydration of the carbon 4 to carbon5doublebond.

This hypothesis needs to be tested before any further investigation of the details of this mechanism can be carried out. Following cloning, expression, and optimisation of UGL, and investigation of the natural substrate preferences, the UGL-catalysed reaction of unsaturated glucuronides in D2O, to give an isotopic label for the site of protonation, and in methanol, as an alternate nucleophiletogiveastableinter-

mediate, were envisaged as first tests of this mechanism. Further confirmation was

sought in compounds that would only be anticipated to act as substrates under a

hydration-initiated mechanism.

Clostridium perfringens was chosen as a source organism for UGL as it is an organism of very high disease relevance, with infections often having very serious consequences — it is a common cause of food poisoning and post-surgical infection, with this infection necessarily involving degradation of the extracellular matrix and its GAGs.101–103 Because of this strong infectivity this organism was expectedto have a suitable copy of UGL, which was confirmed by homology searching of the genomic DNA. The commercial availability of genomic DNA for C. perfringens was also an advantage of using this organism, with DNA for the ATCC13124 strain

53 2.1. Cloning and purification of Clostridium perfringens UGL already being present in the laboratory at the time of commencing this study.

2.1 Cloning and purification of Clostridium perfringens

UGL

In order to investigate the reaction carried out by UGL, a reliable source of pure enzyme is required. To achieve this, the enzyme was cloned from C. perfringens and heterologously expressed in E. coli using the pET28a vector to provide a hexa- histidine tagged enzyme, and was purified by metal ion affinity chromatography.

2.1.1 Cloning

Primers for cloning of the gene for UGL from C. perfringens were designed to amplify the gene and to allow insertion with two of the restriction sites present in the multiple cloning site of the commercial pET28a(+) expression plasmid—XhoIandNheI.

The insertion site was selected in order to generate a fusion peptide with an N- terminal hexa-histidine tag. PCR from C. perfringens genomic DNA proved difficult initially, but longer primers to overcome the low GC content and optimisation of temperature and magnesium concentration allowed isolationofasmallamountof product. Gel purification of this product and its use as a template for a second round of amplification yielded sufficient DNA to insert the gene into the desired plasmid and, following direct transformation into BL21(DE3) cells byelectroporation,several colonies were picked and sequenced to confirm the insert.

2.1.2 Enzyme expression and purification

Expression was first tested at small scale in a variety of growth conditions, looking for a protein of the correct size by SDS-PAGE in the supernatant or pellet of lysed

54 2.1. Cloning and purification of Clostridium perfringens UGL cells (Figure 2.1). The effects of growth temperature, growthmedium,celldensity at induction, and concentration of IPTG used for induction onproteinlevelswere assessed. All conditions gave acceptable growth of cells andhighlevelsofsoluble protein, with some variability, so the choice of expression conditions was based largely on ease of workflow, opting for overnight expression following induction with a low level of IPTG at low cell density in TYP medium.

Figure 2.1: Expression of UGL under varying conditions. Calculated mass for UGL is 48 kDa. The red arrow indicates selected conditions for large-scale expression.

Protein expression was then scaled up to a half litre culture for purification of the enzyme. Cells were grown to mid log phase (OD600 of 0.5–1), induced, and the protein expressed overnight before harvesting. Lysing of cells and pelleting of cel- lular debris gave a clarified lysate that was purified using a nickel-affinity column.

Stepwise elution with increasing concentrations of imidazole gave protein that was deemed to be sufficiently pure for enzymatic work, as assessed by SDS-PAGE (see

Figure 2.2). The imidazole and salt from the purification wereremovedbyrepeated spin filtration in a 30 000 kDa cut offspin filter, with a final estimated dilution of

15 000 fold, bringing the imidazole to low micromolar concentrations. The concen- tration of enzyme was determined from a calculated extinction coefficient at 280 nm,

55 2.1. Cloning and purification of Clostridium perfringens UGL

Figure 2.2: Purification of UGL, with sample elution trace forwildtypeUGL(up- per), showing A280 in blue, conductivity in brown, % buffer B in green (up to 100% at its maximum), and fraction numbers in red. Axes are eluted volume in mL and ab- sorbance in mAU, with axes not shown for other traces. (Lower)SampleSDS-PAGE for purification of UGL, with fractions and marker masses as labelled.

56 2.2. Development of a chromogenic substrate with this process yielding around 45 mg of protein per litre ofculture.

2.2 Development of a chromogenic substrate

2.2.1 Previous assays

In earlier work,82,84–87,93 kinetic parameters for cleavage of natural substrate by

UGL were determined by monitoring the decreasing absorbanceat240nmfromthe double bond in the starting material. A phenyl unsaturated glucuronide has also been reported in a study of UGL anomeric specificity.104 Monitoring the substrate double bond, while successful for the limited characterisation carried out previously, has many restrictions. Because of the short wavelength of thisabsorptionpeak, there are many compounds that may interfere with these readings. Related to this, this method requires the measurement of a decrease in an initially high substrate concentration, meaning that there are practical limitations on how high the starting substrate concentration can be before this method becomes unsuitable. The assay has a useful range from high tens of micromolar, where signal is low, to low millimolar concentrations, where the initial absorption becomes too high. Finally, because this method monitors the consumption of starting material, and not the formation of product, this means that if the enzyme catalysed a fast reaction followed by a slower second step, which is a plausible scenario given the initially proposed hydration mechanism, this second step would not be observable using a method that only monitors reactants. In a mechanistic study it is useful to be able to monitor both substrates and products. Access to natural substrates is also very limited, coming from the degradation of glycosaminoglycans by polysaccharide lyases, and commer- cial sources are very expensive for small amounts. Availablequantitiesaresufficient for characterisation of basic kinetic parameters, but insufficient for in-depth study.

57 2.2. Development of a chromogenic substrate

While this older method remained useful for characterisation of natural substrates

(Section 2.4 on page 65), most kinetic measurements in this work were performed using aryl unsaturated glucuronide substrates. These provided a product that could be easily and accurately detected at wavelengths where thereisminimalbackground interference, are able to be synthesised in large quantities, and are easily variable at key positions with heteroatoms or isotopes, all while avoiding perturbation of the natural mechanism.

2.2.2 Substrate synthesis

Several sources in the literature had previously reported the synthesis of simple unsaturated glucuronides, for use either as substrates in early work on UGL104 or investigations of the bleaching process of paper from the Kraft process, where such unsaturated glucuronides are an important side product of the strong bases used and are responsible for brightness reversion of the paper andexcessconsumption of bleaching chemicals.105,106 Synthetic compounds reported were predominantly methyl or phenyl glycosides, and are largely unsuitable for in depth mechanistic studies because of difficulties in varying the anomeric group,lowyields,orrequiring an inconveniently large number of steps. A route was envisaged, based largely on that reported by Azoulay et al.,107 which comprised four key steps: global protection of the starting material, glycosylation of the desired anomeric alcohol, elimination across carbons 4 and 5, and global deprotection of the product. The final optimised routes for 4-nitrophenyl ΔGlcA (6)andphenylΔGlcA (10)aregiveninScheme2.1 and Scheme 2.2, respectively.

Global protection was achievable through several routes, with either methyl ester

formation or acetylation taking place first. The most convenient was found to be

from glucuronic acid-γ-lactone (1), opening the lactone with sodium methoxide in

58 2.2. Development of a chromogenic substrate

HO 1. NaOMe, MeOH MeOOC MeOOC O 2. Ac2O, HClO4 O HBr, AcOH O AcO AcO O OH AcO OAc AcO O OAc AcO Br 1 OH 2 3 58% Ag2O, 4-nitrophenol 97% ACN

1. NaOMe, MeOH COOH COOMe MeOOC 2. LiOH, THF/H2O DBU, DCM O AcO O O AcO O HO O AcO O OAc OH OAc NO NO NO2 6 2 5 2 4 80% 92% 54%

Scheme 2.1: Final optimised 4-nitrophenyl glycoside substrate synthesis.

HO 1. NaOMe, MeOH MeOOC MeOOC O 2. Ac2O, HClO4 O H2NNH2.AcOH O AcO AcO O OH AcO OAc DMF AcO OH O OAc AcO 2 7 1 OH 58% 65% 1. Trichloroacetonitrile/DBU, DCM 2. BF3.OEt2/phenol, DCM COOH COOMe MeOOC NaOH, acetone/H2O DBU, DCM O AcO O O AcO O HO O AcO O OAc OH OAc 10 9 8 77% 51% 30%

Scheme 2.2: Final optimised phenyl glycoside substrate synthesis. methanol then acetylating with acetic anhydride under acidic conditions (to avoid elimination side-products) to give a mixture of anomeric forms of per-O-acetylated methyl glucuronate (2).108 Purification of this product on the large scale proved to be problematic. The reactions appeared to go cleanly by TLC, but on workup and removal of solvent a thick slightly yellow syrup was formed. While this product could be used without further purification for many reactions, for long term storage and for reactions requiring more controlled stoichiometry a solid was more convenient.

The most effective means found to achieve this was triturationfromasmallvolume of methanol, yielding a white powder suspended in a residual amount of solvent

59 2.2. Development of a chromogenic substrate that could be conveniently filtered offand washed with a small additional portion of methanol then dried and stored. This process could be repeated several times with the filtrate to increase recovery, with no apparent loss in purity of the solid recovered. Crystallisation from ethanol, isopropanol, or toluene with petroleum ether gave a lower recovery, and the product remained sticky and difficult to handle. The overall yield for this protection was 58%, mostly limited by the number of times the evaporation from methanol was carried out, with each iteration giving diminishing recovery.

The method used for glycosylation by Azoulay, using tin tetrachloride and 4- nitrophenol directly with globally protected glucuronic acid (2), was relatively low- yielding. Improvement was sought through activation of the glycosyl donor us- ing two alternate methods. Deprotection of the anomeric position to give the free hemiacetal (7)andformationoftheSchmidtdonorwassomewhatsuccessful,but the activated donor was found to be unstable and it was difficulttoavoidsigni-

ficant hydrolysis when reacting with BF3 diethyl etherate and either phenol or 4- nitrophenol (Scheme 2.2). By contrast, formation of the glucuronyl bromide (3)using hydrobromic acid in acetic acid proceeded smoothly and gave averystableproduct that could be purified by flash column chromatography and stored for an extended period at -20 °C, although for routine preparations this was used without purifica- tion. Koenigs-Knorr glycosylation using this was also foundtoproceedsmoothly, giving the desired product in moderate to good yields for mostphenolderivatives

(Scheme 2.1), although notably not for phenol itself, with purification often achiev- able by crystallisation (see Subsection 3.2.1 on page 96 for further discussion).

Elimination of acetic acid from the protected aryl glucuronides (4 and 8)was achieved using DBU in dichloromethane overnight, and the reaction mixture could be loaded directly onto a flash column for purification (where inconvenient to purify

60 2.3. Enzyme optimisation immediately, the reaction was stopped by filtering through a small plug of silica then evaporating the solvent in vacuo). The success of this reaction was found to

be highly dependent on the purity of the starting material, with crystalline starting

materials often giving significantly better yields, but was otherwise found to be a

reliable process with moderate to good yields.

Deprotection of the resultant final products (5 and 9)wasattemptedbyseveral methods, with the most reliable found to be deacetylation using sodium methoxide in mixed methanol and dichloromethane followed by hydrolysis of the methyl ester by a slight excess of lithium hydroxide in tetrahydrofuran and water. This method produced very clean products, although a C18 Sep-pak was usedtoremovesalts and trace non-polar contaminants such as partially deprotected intermediates. An alternative using excess NaOH in 1:1 acetone and water then quenched with a slight excess of aqueous HCl was used in some cases, including phenyl ΔGlcA (9), but generated a large amount of salt that required subsequent removal. This method was found to be faster but more likely to give hydrolysis of theproduct.

The overall yield for the 4-nitrophenyl ΔGlcA substrate (6)was22%over7steps. Synthesis of the phenyl substrate (10)wasmuchlessefficientat4.4%.Thissynthesis

of 6 could be carried out with only one flash column — using the crudeproducts

of the first three steps to achieve 4,whichcanbecrystallisedfromethanol,thena

single column after the elimination step yields the protected final product 5,which

is deprotected in two steps with a final Sep-pak clean up.

2.3 Enzyme optimisation

With enzyme and substrate in hand, the enzymatic reaction conditions needed to be

optimised before careful kinetic measurements could be made. BSA was found to be

important for stabilising the enzyme, with around 0.1 % w/v proving sufficient and

61 2.3. Enzyme optimisation giving approximately twice the rate compared to a control without BSA. Inclusion of EDTA at 10 mM was found to have no effect on the rate of reaction, suggesting no metal ions are important in the reaction. A previous report82 had suggested that

sodium chloride may be important for the reaction, with 50–100 mM being required

in that case for maximal rate, but any concentration of added salt was found to be

detrimental to C. perfringens UGL (Figure 2.3A).

Figure 2.3: Effect of A, salt concentration; B, pH; and C, temperature, on the reaction rate of UGL with 4-nitrophenyl ΔGlcA (6). Y axes are a percentage of maximal kcat/Km for A and B, and of kcat/Km, kcat,andKm as noted for C. Enzyme stability was confirmed across the pH and temperature ranges tested (not shown).

Measurement of kcat/Km for hydrolysis of 4-nitrophenyl ΔGlcA (6)overarange of pH values yielded the curve shown in Figure 2.3B. This showedanoptimalpH

of 6.6 and two pKa values of 6.1 and 7.0. Since the pH of dependence of kcat/Km reflects ionisation in the free enzyme or free substrate it wasnecessarytodetermine

whether these could reflect the substrate ionisation state. The pKa of a represent- ative substrate benzyl ΔGlcA (53 from Section 3.2 on page 96) was determined to be 4.5 ± 0.2 (Figure 2.4). This substrate was chosen as it did not show interference

from the aryl group at low wavelengths, unlike the phenyl substrates. However, as

62 2.3. Enzyme optimisation measurements in this spectral region are inherently noisy, an average of the 200 to

220 nm spectral region with adjustment for the substrate double bond peak height at 235 nm was required to give a reasonable fit. This shows that ionisation of the substrate carboxylate group is not responsible for either oftheseinflectionpoints.

Given that the enzyme active site contains two aspartate residues that have been shown by mutagenesis to be critical for activity, it is parsimonious to attribute these pKa values to those two residues. The optimal pH of this enzyme is close to physiolo- gical pH, particularly that of new wounds, where the pH is slightly decreased relative

to normal tissue.109

Figure 2.4: Determination of the pKa of benzyl ΔGlcA (53). A, spectra of 53 from 260–200 nm at the following pH values: 2.7( ), 3.0(!), 3.5( ), 4.1( ), 4.6( ), 5.7( ), 6.5( ), and 7.3( ); B, fit of ratio (average A220 to 200)/A235 against pH.

Temperature was found to have an unusual effect on the rate of UGL hydrolysis,

since the optimal temperatures determined for kcat/Km and kcat were not the same

as each other, and together these show an increase in Km as temperature increases

(Figure 2.3C). The increase in kcat with temperature was close to linear, rather than

63 2.3. Enzyme optimisation exponential as would be predicted by the Arrhenius equation.Theenzymewas stable at all temperatures measured for longer than the assayrequired,showingno sign of denaturation. This is somewhat consistent with what was observed with F. heparinum UGL,82 where activity peaked at 30 °C using a substrate concentration slightly above Km,despitetheenzymebeingstabletomuchhighertemperatures. By contrast, UGL from Bacillus Sp GL186 showed a more conventional temperature

profile with an activity peak at around 50 °C but decreasing stability above 30 °C,

with a sharp drop in both above the activity maximum. A similartemperatureeffect

to that observed for C. perfringens UGL has been observed for other enzymes, such as 3-phosphoglycerate kinases,110 where enzymes sourced from mesophiles and ther- mophiles both showed this effect to varying extents, but at higher temperatures than for C. perfringens UGL (optima around 50–70 °C). In that case, the observation was explained by invoking an inactive intermediate I on the enzyme unfolding pathway in equilibrium with the native active form N, from which the enzyme then unfolds irreversibly to the denatured state D according to the equation N "−−−−! I −−→ D. The proportion of the I form increases with temperature, accounting for the apparent

increase in Km,withsaturatingsubstrate’S’stabilisingtheenzymeintheN.Sform, accounting for the trend in kcat readings. This reversible instability could be on a global or local scale, with potentially only the active site reversibly denaturing in astableglobalbackgroundtogivethisfoldedbutinactiveform.111 Given that an

enzyme active site is generally optimised for increased flexibility over the protein as

awholeinordertoallowforcatalysis,112,113 this is an appealing explanation.

The final reaction conditions determined to be used for further study involve use

of 40 mM Mes.NaOH buffer at pH 6.6 with 0.1% BSA w/v at 37 °C, to balance

optimisation of kcat and kcat/Km,orphosphatebufferatpH7.0forexperimentsin- volving NMR spectroscopy. The Michaelis-Menten kinetic parameters for hydrolysis

64 2.4. Natural substrate variation of the two aryl glycoside substrates discussed in section 2.2.2 (10 and 6)byUGL under these optimised conditions are given in Table 2.1. Thisbufferwasmuchlater found to act as a competitive inhibitor (Figure 2.5), with a Ki of 57 ± 5 mM. Given its use in the majority of experiments, use of this buffer was continued for all further experiments unless otherwise stated, with care taken to not vary its concentration from 40 mM to allow direct comparison of results.

Table 2.1: Michaelis-Menten kinetic parameters for hydrolysis of the two aryl glyc- oside substrates 6 and 10 by UGL under optimised conditions.

-1 -1 -1 Structure kcat (s ) Km (mM) kcat/Km (s .mM ) COOH O HO O OH NO2 (6) 2.05 ± 0.06 0.26 ± 0.02 7.9 ± 0.8 COOH O HO O OH (10) 4.3 ± 0.2 3.2 ± 0.4 1.3 ± 0.2

2.4 Natural substrate variation

The natural substrates for C. perfringens UGL are the products of reactions of poly- saccharide lyases acting on the GAGs of mammalian tissue, andassuchtheycan vary significantly in their glycosidic linkage and sulfationpattern,dependingonthe source GAG. Other UGLs that have previously been studied, from both pathogenic and non-pathogenic source organisms, have shown variability in their preferred sub- strates.82,85,86 In order to probe the natural substrate preferences of C. perfringens

UGL, a variety of unsaturated disaccharides of GAG origin were purchased and the

Michaelis-Menten kinetic parameters for their cleavage determined. Activity was highly variable with these substrates, ranging from the highest kcat detected with any substrate in this work at 112 s-1 (11)tonodetectableactivityatall(16), while

Km was less variable among the better substrates — see Table 2.2.Foronesub-

65 2.4. Natural substrate variation

Figure 2.5: Dixon plot showing competitive inhibition of UGLbyMes.NaOHbuffer with a Ki of 57 ± 5 mM. Substrate (6)wasatthefollowingconcentrations:25( ), 50(!), 170( ), and 600( )µM.

66 2.4. Natural substrate variation

strate (14), Km could not be determined as no sign of substrate saturation could be achieved within the limitations of both the assay and availability of compound.

Substrates derived from both chondroitin and heparin sulfate were accepted by

C. perfringens UGL, with no clear apriorideterminant of the level of activity. By far

the best substrate was ΔGlcA-β-1,3-GalNAc-6-sulfate (11), for which C. perfringens

UGL showed a kcat an order of magnitude higher than the next best, while the equivalently sulfated substrate from heparin sulfate (14)wasverypoorlyactedon

by C. perfringens UGL. By contrast, the N-sulfated compound derived from heparin

sulfate (12)wasthesecondbestofthenaturalsubstratestestedwithC. perfringens

UGL. For the other chondroitin-derived substrates, removing the sulfate still gives

areasonablelevelofactivity,whilewhenonthecarbon4hydroxyl it gives very ! low activity, while moving it onto the 2 -position of the ΔGlcA completely ablates this activity. This lack of activity with a 2!-sulfated substrate is not particularly surprising on an intuitive level, as it places this negatively charged sulfate in close proximity to Asp113, and the interaction of these two negative charges could very plausibly interfere with that residue’s catalytic function. However, it is notable that some strains were able to cleave this substrate, as the Streptococcal strains all showed asmalllevelofactivityandforBacillus this is a close third best substrate. Clearly, it is possible for UGL to accommodate a sulfate in this position, but how this occurs and what this means for the role of this residue is unclear. It is possible that the sulfate group itself takes over the role of Asp113 in these cases, since it has a similar charge and likely occupies a similar location.

The pattern of relative activity with these substrates is fairly similar across C.

perfringens, F. heparinum, Streptococcus pneumoniae, S. pyogenes,andS. agalac-

tiae,whiletheBacillus data are notably different. This likely reflects the different

environments each organism lives in and the metabolic requirements of this environ-

67 2.4. Natural substrate variation s F. , m 18 24 09 00 49 /K . . . . . nd 0 0 0 1 0 (rel.) cat k 82 C. perfringens (mM) 251 334 235 107 283 . . . . . 0 0 0 0 0 m K F. heparinum ) -1 7 3 9 8 4 . . (s . . . 4 8 2 11 15 cat k m 39 00 09 /K . . . nd 0009 0007 0 1 0 . . (rel.) cat 0 0 k sp. GL1. 2 7 0 5 . . . . etected; blank, no data). Data from (mM) 1 3 3 4 erent source organisms with GAG-derived natural substrate m Bacillus K C. perfringens ) ,and -1 5 (s 09 . . 14 8 112 0 cat k {continued on the following page} ) ) ) ) ) ) ) ) ) ) ) 19 20 21 11 12 13 14 15 16 17 18 ( ( ( ( ( ( ( ( ( ( ( + + + S. pneumoniae OH OH OH OH OH OH OH OH Na Na Na + + + + - - - , 3 3 3 2 OH OH OH Na Na Na + Na - - - - 3 3 3 + 3 NHAc NHAc NHAc NHAc O O O O O NH NHAc NHAc NHAc O O O Na - 3 Na O NHSO NHSO NHSO O O - OH 3 OSO OH OH OSO OH OSO OSO + + OH OSO OH Na Na + OH OH OSO OH OH OH - - 3 3 O O O O O O HO O Na OH OH O - O O O O O O O 3 HO O O HO O OH OH OSO OH OSO OH Structure O OH O O COOH COOH COOH COOH COOH COOH O COOH OH OSO OH OH COOH COOH COOH COOH S. pyogenes , HO HO HO HO HO HO HO HO HO HO HO 6S 2 SNS S6S S ! ! ! S. agalactiae , Ch. 2 Ch. 2 Ch. NS6S Ch. NH Hy. 0S He. NS Ch. 0S He. 6S Ch. 4S Ch. 2 Source GAG Ch. 6S Table 2.2:(Ch., Comparison chondroitin; He., of heparin: kinetic Hy., parameters hyaluronan; nd, for not UGL d from diff heparinum

68 2.4. Natural substrate variation m 00 20 22 98 /K . . . . nd nd 1 0 0 0 (rel.) cat k 86 38 (mM) . 19 0 m K Bacillus ) -1 1 . (s 50 14 cat k – m – 00 01 /K 003 . . 0008 003 . nd– 0001 . 1 . 0 (rel.) . 0 cat 0 0 0 k 85 , 94 ) 39 . 10 0 . 27 (mM) . , 0 ( 1 m 18 . 54 . 0 K 0 Streptococcus ) , 94 0 ) -1 . 2 4 . 7 (s . , 2 10 3 . ( cat 1 k 24 ) ) ) ) ) ) ) ) ) ) ) 21 11 12 13 14 15 16 17 18 19 20 ( ( ( ( ( ( ( ( ( ( ( + + + OH OH OH OH OH OH OH OH Na Na Na + + + + - - - 3 3 3 2 OH OH OH Na Na Na + Na - - - - 3 3 3 + 3 NHAc NHAc NHAc NHAc O O O O O NHAc NHAc NH NHAc O O O Na - 3 Na O NHSO NHSO NHSO O O - OH 3 OSO OH OH OH OSO OSO OSO + + OSO OH OH Na Na + OSO OH OH OH OH OH - - 3 3 O O O O O O HO O Na OH OH O - O O O O O O O 3 HO O O HO O OH OH OSO OH OSO OH Structure O OH O O COOH COOH COOH COOH COOH COOH O COOH OH OH OSO OH COOH COOH COOH COOH HO HO HO HO HO HO HO HO HO HO HO 6S 2 S6S S SNS ! ! ! Ch. 2 Ch. NS6S Ch. NH Hy. 0S Ch. 0S He. 6S Ch. 4S Ch. 2 Ch. 2 Source GAG Ch. 6S He. NS

69 2.4. Natural substrate variation ment. Interestingly, F. heparinum,asoil-isolatedbacterialstrain,showsasubstrate preference pattern in many ways similar to those of the pathogenic Streptococcus and Clostridium species, with high activity on sulfated substrates. The Bacillus species, which is also a soil isolate, has a quite different pattern of activity, however, but both do seem to have high activity on the unsulfated substrate. Since the UGL from Bacillus was initially isolated by growth on the unsulfated bacterialpolysac- charide gellan, while the F. heparinum UGL was isolated from growth on heparin sulfate and chondroitin sulfate, this may reflect diverse diets within the different soil-isolated samples, decomposing organic matter from different organisms. Inter- estingly, in these other species the Km values for the GAG-derived substrates were

lower than for C. perfringens UGL, with correspondingly lower kcat values to give

similar kcat/Km values. The X-ray crystal structure of the Streptococcus agalactiae UGL with substrate

bound revealed a binding motif, R-//-SXX(S)XK (where -//- isalargebreakinthe

sequence, and X is any amino acid), conserved in pathogenic bacteria-sourced UGL

that is important for binding of the 6-sulfate of chondroitin-derived substrate 11,

as shown in Figure 2.6.94 This motif is located on a dynamic loop that moves upon

substrate binding, and mutation of three key residues in thismotifwasobservedto

reverse the preference for 6-sulfated substrate over unsulfated substrate. Alignment

of UGL sequences for the four species shown in Table 2.2 (with S. agalactiae chosen as

arepresentativeStreptococcalspecies)proveddifficult,with different methods giving

very different alignments of the F. heparinum sequence. These often showed no

conservation of the otherwise very well conserved D113 and D173 catalytic residues

(C. perfringens numbering).

The result deemed the best was returned by the MUSCLE algorithm,114 and

the key results are given in Figure 2.7. Even in this best alignment, no residue is

70 2.4. Natural substrate variation

Figure 2.6: X-ray crystal structure of the S. agalactiae UGL active site showing the 6-sulfated chondroitin disaccharide substrate (11), the catalytic residues (D115 and D(N)175), and the sulfate binding domain (S365, S368, andK370),fromtwo perspectives (PDB: 3ANK). 71 2.4. Natural substrate variation aligned with D113, although there are several other residuescloseinsequence(if not necessarily space) that could potentially be filling the same role. The putative catalytic acid residue, D173, is conserved, although the sequence context that is preserved in the other species is changed for F. heparinum. A BLAST search using

the F. heparinum sequence returned GH88 sequences wherein the aspartate of the

’QNTRDART’ motif aligning with D113 is conserved in all but 4 of the top 99 hits,

suggesting this may be filling the same role in these species. The 6-sulfate binding

domain is present in its entirety in C. perfringens and S. agalactiae and completely

absent in Bacillus,whileF. heparinum contains one of the key residues and has a

conservative mutation for a second. This provides some explanation for the apparent

mixture of activities displayed by F. heparinum UGL, showing no strong preference

for sulfated or unsulfated substrates. Interestingly, the histidine immediately pre-

ceding D113 also appears to be highly conserved, and in the substrate-bound crystal

structure it is within hydrogen bonding distance, suggesting that these two residues

may be acting as a catalytic diad, although no mutagenesis of this histidine has

been reported. In F. heparinum the sequence aligning with D113 also has two ba-

sic residues (arginine) that appear to be well conserved in one or both positions

(arginine or lysine).

In two of the species presented, C. perfringens and Bacillus,the6-sulfatedsub-

strate derived from chondroitin is turned over much more efficiently than all other

substrates tested, as shown by their substantially higher kcat values. Given that the 6-sulfate binding motif has been identified by crystallography for only one of these

two species, this suggests that the sulfate group may also be influencing the reaction

in some other way, perhaps through long-range interactions that optimise the posi-

tioning of the catalytic residues. Alternatively, direct intramolecular catalysis of the

reaction may be taking place, with the negatively charged sulfate group somehow

72 2.4. Natural substrate variation

catalytic residue ↓ Clostridium_perfringens KDIELDHHD....LGFLY 118 Flavobacterium_heparinum QFIWVEQNTRDARTGLLY 204 Streptococcus_agalactiae NRIALDHHD....LGFLY 120 Bacillus_sp_GL_1 RFENLDHHD....IGFLY 93

catalytic residue ↓ Clostridium_perfringens YRFIID...... CLLN 177 Flavobacterium_heparinum WYDILDQPNRKGNYFES293 Streptococcus_agalactiae YRLIID...... CLLN 179 Bacillus_sp_GL_1 GRIIID...... CLLN 153

4- and 6-sulfate binding ↓ Clostridium_perfringens RGVTRQG YSWHSGKGV 370 Flavobacterium_heparinum RDGSYEY E!IFK"#TAK$QL560 Streptococcus_agalactiae KGVTRQG YSWHSGKGV 372 Bacillus_sp_GL_1 RGGTHQG YHVRGGISP346

Figure 2.7: Key fragments of a multiple sequence alignment ofUGLsequencesbythe MUSCLE algorithm. Identical residues are coloured yellow onapurplebackground, conserved residues are coloured white on a blue background, the putative catalytic residues D113 and D173 (C. perfringens numbering) are indicated with arrows, the 4- and 6-sulfate binding domains are indicated with an arrow and a brace, and the vertical red line indicates a break in the sequence. The full alignment is in Appendix Aonpage291.

73 2.5. Characterisation of UGL reaction products stabilising a partial positive charge in the transition state of the rate-determining step.

The arginine residue identified in the crystal structure of S. agalactiae UGL to

bind 4-sulfated substrate85,94 appears to be a poor predictor of high activity with

these substrates for UGLs from other species. A homologous residue is present in C. perfringens UGL, but the chondroitin 4-sulfate derived substrate is onlypoorlyhy- drolysed. While no kinetic parameters for the chondroitin 4-sulfate derived substrate were provided for S. agalactiae to allow quantitative comparison, TLC images in the above reference show complete hydrolysis of the compound by S. agalactiae UGL while Bacillus shows no conversion, suggesting a relatively high level of activity.

2.5 Characterisation of UGL reaction products

COOH HO O O OH H2O HOOC UGL (UGL?) O O HO OR HO OR HOOC + HO OH HOOC OH O (+/- H ) OH OH OH 22ROH 23 OH 24 + isomers

Scheme 2.3: Formation of the final product of the UGL-catalysed reaction.

The final products of UGL (GH88) and UGH (GH105) catalysed degradation of unsaturated glycuronides via the hydrated intermediate 22 have previously been described as the leaving group alcohol and free unsaturated glucuronic acid as the open chain 5-keto form (23),82,84,87,92,93,95 as demonstrated by mass spectrometry. 87

However, on observing the products of this reaction by 1H-NMR, the characteristic peaks from protons on carbon 1 and 4 of an unsaturated uronic acid are not visible.

Instead, a mixture of products is clearly visible, resultingfromadditionofwaterto and cyclisation of this open chain as shown in Scheme 2.3, withthehemiketaland hemiacetal both existing as an equilibrium mixture to give four compounds with both

74 2.5. Characterisation of UGL reaction products ). 6 GlcA ( Δ ,thefinalproductsofUGLdegradationof4-nitrophenyl 24 H-NMR spectrum of 1 Figure 2.8:

75 2.5. Characterisation of UGL reaction products stereochemistries at each of carbons 1 and 5. The peaks arising from the dominant form of this product, being that shown in Scheme 2.3 (24), are labelled in the 1H-

NMR spectrum given in Figure 2.8. These assignments were further confirmed by direct correlation (COSY, Appendix B on page 295) and total correlation (TOCSY,

Appendix B) spectra.

2.5.1 Reaction in D2O

The same UGL-catalysed reaction was carried out with D2Oasasolvent.Thisal- lowed any accessible exchangeable protons in the enzyme to become deuterated, in particular the catalytic acid. This means that, on catalysing the hydration reaction, adeuteriumwouldbeincorporatedintotheproductsatthesite of protonation. If asimpleglycosidehydrolysisreactionweretakingplace,this catalytic acid-sourced deuterium would be on the leaving group oxygen and exchangeable with bulk solvent, while subsequent keto-enol tautomerisation at carbon 4 is unlikely to effect stereo- specific protonation. However, in a hydration reaction, thisdeuteriumwouldbe stably incorporated at carbon 4 of the sugar product, and thiswouldoccurina stereospecific manner determined by the placement of the catalytic residue. Such an incorporation is detectable by 1H-NMR.

1 The H-NMR spectrum of the products of this enzymatic reaction in D2O, over-

laid on the spectrum from a control reaction in H2O, is shown in Figure 2.9. This clearly shows that all peaks are the same, except for those affected by exchange of

the axial proton at carbon 4 with deuterium. This signal itself is completely ab-

lated, the equatorial proton on the same carbon no longer shows geminal coupling,

and the signal from the proton on carbon 3 has lost a diaxial coupling partner to

become a doublet of doublets. Together these show stereospecific incorporation of

deuterium by the enzyme to the re face of the double bond at carbon 4 to form 25.

76 2.5. Characterisation of UGL reaction products

Figure 2.9: Overlay of expanded 1H-NMR spectra of the products from UGL- catalysed hydrolysis of 4-nitrophenyl ΔGlcA (6)inD2O(upper)andH2O(lower).

The spatial location of this deuterium correlates well with the aspartate residue 173

(149 in Bacillus numbering), which was proposed to be the principal catalyticacid

residue based on the crystal structure of UGL with substrate bound,87 as shown

in Figure 1.11 on page 47 and represented in Scheme 2.4.

D D173 DO COOD D (-PhOD) D OD COOD O OD O O O H H DO OPh DO OPh DO OD OD DOOC OD OD OD 10 25

Me D173 HO H OH COOH O OMe O O H HO OPh HO OPh OH HOOC OH 10 26

Scheme 2.4: Reaction catalysed by UGL, showing the route by which deuterium (upper) and methanol (lower) are incorporated.

77 2.5. Characterisation of UGL reaction products

2.5.2 Reaction in 10% methanol

To show that the solvent nucleophilic attack catalysed by theenzymeoccursat carbon 5, the UGL catalysed reaction was carried out in 10% methanol in H2O. This was the highest proportion of methanol in which the enzyme wasstable.Methanol can act as an alternate nucleophile, substituting for water in the active site and attacking in its place. This forms a product which has a methylketalinplaceofthe hemiketal in 22,andsocannotundergothesamerearrangementstepstocleavethe glycosidic bond shown in Scheme 2.3 on page 74. In practice, the product of this reaction was found to not be very stable, and attempts to purify it by HPLC or C-18

Sep-pak were unsuccessful. However, the product was clearlyvisibleasanadditional set of small peaks in the 1H-NMR spectra of the crude reaction mixture following

removal of enzyme, as shown in Figure 2.10. These peaks were further analysed by

correlation spectroscopy (COSY45 and NOESY, Appendix B on page 295).

Figure 2.10: Overlay of expanded 1H-NMR spectra of the products from UGL- catalysed hydrolysis of phenyl ΔGlcA (10)in10%MeOHinH2O(26,lower)anda control in H2O(upper).

The final structure determined for this product (26)resultsfromadditionof

methanol to the si face of the double bond at carbon 5, as shown in Scheme 2.4.

Thus an overall syn addition of methanol occurs, when taken together with the result

78 2.5. Characterisation of UGL reaction products

of the reaction with D2OdetailedinSubsection2.5.1.Thisproductstereochemistry was deduced from the NOESY correlation of the methyl singlet peak, which showed

acrosspeakwiththeaxialprotononcarbon4,butnottheequatorial proton at the

same position. This product conformation is likely the reason for the compound’s

instability, either through ground state destabilisation or intramolecular catalysis by

the C-6 carboxylate, which is now in close proximity to the anomeric carbon. It is

worth noting that the product formed from hydrolytic degradation of 26 initiated

at carbon 1 or carbon 5 is the same, as both will form the same hydrated product

seen in the enzymatic reaction (24), as illustrated in Scheme 2.5.

OMe OMe OMe O O O HO OPh HO HO OH OH HO - OH - - OOC COO O O O O UGL, (rotamers) (rotamers) O H2O HO OPh 10% MeOH OH OH -OOC - OH - OH 10 O OOC O+ OOC O

OMe OMe OMe OH OPh OH OH OH 26

OMe O HO OH OH - OMe O O COO- O O HO OPh HO OH OH OH OH O O- OH O 24 26 HO OPh OH O O-

Scheme 2.5: Decomposition of the methanol adduct formed by UGL-catalysed reac- tion of phenyl ΔGlcA (10), showing possible mechansims by which decomposition is accelerated (upper) and formation of the same product from decomposition initiated at either carbon 1 or carbon 5 (lower).

To further confirm the structure of this methanol adduct, a synthetic standard

was sought. Protracted treatment of protected phenyl ΔGlcA (9)inmethanoland HCl, generated in situ using acetyl chloride, followed by purification by HPLC af-

79 2.5. Characterisation of UGL reaction products forded the axial methyl acetal 27 (Scheme 2.6). This was then deprotected with sodium hydroxide in acetone/water to give the deprotected standard 28 in 49% yield over 2 steps. This compound was found to be the carbon 5 epimer of the en- zymatic product 26,andwascorrespondinglymuchmorestable.Thisstructurewas determined again by 1H(Figure2.11),TOCSY(notshown),COSY45,andNOESY

(Appendix B) NMR spectroscopy. The stereochemistry at carbon5isinferredfrom

NOE correlations between the methyl singlet and the protons on carbons 1 and 3 as well as the equatorial proton on carbon 4 — the opposite to that with which the methyl singlet correlates in the enzymatic reaction product.

COOMe AcCl, MeOH MeOOC O NaOH, acetone/H2O HOOC O O HO O HO O AcO O OMeOH OMeOH OAc 927 28 49% over 2 steps

Scheme 2.6: Synthesis of a standard for the product of UGL-catalysed reaction in 10% methanol at carbon 5.

Afurtherstandardformethanoladditiontotheanomericcarbon, the product

expected from a Koshland glycoside hydrolase mechanism, wasalsosynthesisedas

detailed in Scheme 2.7. Glucuronyl bromide 3 was reacted with silver carbonate

in methanol and acetone to give the methyl β-glucuronide 29 in low yield. This was then subjected to DBU-mediated elimination in good yield to 30,followedby

deprotection by NaOH in acetone and water to give 31 in 16% yield over 4 steps,

without optimisation. An overlay of 1H-NMR spectra for both 31 and 28 with the

enzymatic products is presented in Figure 2.11.

Finally, the observation of the syn adduct allows the exclusion of a mechanism

analogous to that acting in the hydration reaction of glucal by glycoside hydrolases,

wherein the catalytic acid residue itself is added in a syn- fashion across the double

bond to form a glycosyl enzyme intermediate, followed by attack of the eventual

80 2.5. Characterisation of UGL reaction products

MeOOC MeOOC COOMe COOH O O AcO Ag2CO3, AcO DBU NaOH, AcO AcO OMe O O AcO methanol/ OAc DCM AcO OMe acetone/ HO OMe acteone OAc OH 329Br H2O 56% 30 31 78% 36%

Scheme 2.7: Synthesis of a standard for the product of UGL-catalysed reaction in 10% methanol at carbon 1.

Figure 2.11: Overlay of expanded 1H-NMR spectra of the anomeric methanol syn- thetic standard (31,upper),theproductsfromUGL-catalysedhydrolysisofphenyl ΔGlcA (6)in10%MeOHinH2O(26,middle)andthecarbon5methanolepimeric synthetic standard (28,lower). nucleophile from the opposite face, to give an overall anti-addition (see the end of

Section 1.3 on page 21). If such a mechanism were acting in UGL,oneoftwopossible outcomes would be observed, depending on where the methanol nucleophile attacks.

In the case where methanol attacks the hypothetical glycosyl-enzyme intermediate at carbon 5, the product formed would reflect the overall anti addition, and would thus be the same as the synthetic standard 28.Thealternativewouldbeformethanol to attack at the acyl carbon of the catalytic acid/nucleophile residue. This would give an appropriate product stereochemistry, a syn addition product of water (not methanol) across the double bond, but would give a methyl ester of this catalytic

81 2.6. Unusual substrates

D173 OH D173 O HOOC O O D173 Me (A) O COOH (B) OH HO OR H O OMeOH O O O HO OR HO OR OH HOOC OH (A) (B) D173 OMe O HO O HO OR HOOC OH

Scheme 2.8: Expected reaction pathway for UGL if a mechanism analogous to that for glucal hydration were followed. residue, thereby preventing further turnover of substrate by the enzyme (paths A and B in Scheme 2.8, respectively).

2.6 Unusual substrates

Because UGL cleaves glycosidic bonds indirectly through hydration of the carbon 4- carbon 5 double bond, there are several compounds that are expected to be accepted by this enzyme as substrates that otherwise would not undergoreactionwithastand- ard hydrolase. Kinetic parameters for hydration of each of these are summarised in

Table 2.3, with discussion of each following.

Table 2.3: Michaelis-Menten kinetic parameters for three unusual substrates accep- ted by UGL (*, estimate based on Ki).

-1 -1 -1 Structure kcat (s ) Km (mM) kcat/Km (s .mM ) COOH OH O HO OH OH HO (40) 0.036 ± 0.0067 2.7 ± 0.4* 0.0133 ± 0.0002 COOH O HO HOO (43) 0.6 ± 0.1 38 ± 9 0.015 ± 0.006 COOH O HO S OH (46) 9.3 ± 0.2 4.8 ± 0.3 1.9 ± 0.2

82 2.6. Unusual substrates

2.6.1 Kdn2en

The first such unusual substrate, 2,3-dideoxy-d-glycero-d-galacto-non-2-enopyrano- sonate (Kdn2en, 40), is an unsaturated C-glucuronide analogue with a glycerol chain in place of a leaving group. The relationship between sialic acid and ΔGlcA deriv- atives has previously been exploited to design inhibitors for sialidases based on a

ΔGlcA scaffold,115,116 while in this case a sialic acid is being used for ease of access to a C-glycoside, a substrate that could only be acted on if a hydration mechanism were employed by UGL.

Asmallamountof40 could be purchased for initial testing, which confirmed that this compound was hydrated by UGL, with the formation of a hydration product being confirmed by mass spectrometry. In order to determine the Michaelis-Menten kinetic parameters for this substrate a larger quantity was required, so synthesis was attempted based on that of Schreiner and Zbiral 117,asoutlinedinScheme2.9.Kdn

(32)wasformedenzymaticallyfrommannoseandpyruvatebyNeu5Ac aldolase using amethodbasedonthatreportedforsynthesisofsialicacid,118 then protected by methanol/TFA followed by acetic anhydride/pyridine. For the elimination of 33 by trimethylsilyl trifluoromethanesulfonate, a more recent method from Chang et al. 119 was attempted first, with reaction at ambient temperature in ethyl acetate instead of at 0 °C in acetonitrile. This appeared to proceed smoothly by TLC, but gave a second set of NMR very similar to the published product characterisation and of a similar abundance to the product. This was initially assumed to be a rotamer of the product, given the apparent purity of the product by TLC analysis in several different solvent systems, so deprotection was undertaken. However, these additional peaks persisted.

Further investigation of the literature 120 revealed a propensity for this product to racemise at carbon 4 under these conditions, dependent on temperature, solvent, and reagent quantities, with the reaction proceeding through an allylic carbocation

83 2.6. Unusual substrates intermediate formed by the Lewis acid reagent.

OH OH 1. MeOH, TFA OAc OH OH OAc O Neu5Ac aldolase 2. Ac2O, pyridine HO O COOH O COOMe HO O HO HO AcOAcO OH HO AcO HO AcO COONa 32 33 48% 60%

TMSOTf EtOAc

HO Ph OH OAc COOMe COOMe O COOMe OH OAc O O O HO AcO O O HO AcO PhCH(OMe) NaOMe, MeOH O 38 2 HO 36 AcO 34 Ph pTSA, ACN 23% OH OH OAc + + + COOMe COOMe COOMe O O O O HO HO HO AcOAcO O HO HO HO AcOAcO Ph 39 37 35 77% overall 18% 80% overall

1. LiOH, THF/H2O 2. IEX H+

OH COOH COOH OH O O HO HO HO OH HO HO OH 40 HO 81%

Scheme 2.9: Synthesis of 2,3-dideoxy-d-glycero-d-galacto-non-2-enopyranosonate (Kdn2en, 40).

While it was possible to repeat the reaction where the racemisation had occurred

to try to avoid this side-product, this reaction had already been scaled up and

little starting material remained. It was decided to attemptseparationofthese

epimers, despite literature claims that they are inseparable. Chromatography and

crystallisation attempts using various solvent systems, onboththeacetylated(35

and 34)andde-acetylatedforms(37 and 36), largely confirmed this claim, although

some enrichment was possible with diligent flash column chromatography. A solution

to the lack of resolution was then sought on the basis of the differing reactivity of

84 2.6. Unusual substrates cis-andtrans-oriented vicinal diols in the formation of cyclic protecting groups, such as a benzylidene. Because the carbon 4 and 5 alcohols in theundesiredside- product are in a cis-oriented arrangement, it was anticipated that these could form

abenzylidenewherethetrans-oriented groups in the desired product could not, giving easily resolvable products. This proved to indeed be the case, although the yield of each product (39 and 38)waslow,owingtotheformationofvariousside products that were not characterised. These could then be deprotected by methyl ester hydrolysis using lithium hydroxide followed by protracted stirring under acidic conditions to hydrolyse the benzylidene and protonate the carboxylate at carbon

1. The yield for this deprotection likely could have been improved by a two step deprotection removing the benzylidene first, as some hydration of the double bond occurred, but with sufficient product in hand optimisation wasunwarranted.

Testing of this substrate indicated a high Km,whichwasnotsurprisinggiventhe nature of the group at the position equivalent to the anomericpositionofΔGlcA (a glycerol chain rather than an aryl group or sugar). This meant that substrate binding could not be saturated within the limits of the assay used (as discussed in Subsection 2.2.1 on page 57). Data from direct assay of thissubstrateatlow concentrations were fit to a linear equation to determine kcat/Km,andKi (see

Figure 2.12) was then approximated as a surrogate for Km to allow estimation of kcat

(see Table 2.3 on page 82). This estimation of Km derived from inhibition testing (2.7 ± 0.4 mM) agreed well with the value calculated from non-linear regression of data from reactions below 1 mM of 40 (2.4 ± 1.3 mM), despite this showing only slight curvature.

85 2.6. Unusual substrates

Figure 2.12: Dixon plot showing competitive inhibition of UGL by Kdn2en (40)with a Ki of 2.7 mM. Substrate (6)wasatthefollowingconcentrations:25( ), 100(!), 200( ), 400( ), and 600 ( )µM.

86 2.6. Unusual substrates

2.6.2 Axial phenol

Glycoside hydrolases are very specific for a particular anomeric configuration as the placement of catalytic residues needs to be very precise to effect catalysis. With ahydration-initiatedmechanism,suchasthatinUGL,theanomeric configuration is only important in determining binding. As long as the enzyme active site can accommodate the compound it should catalyse hydration by virtue of the catalytic residues around the double bond, the placement of which should be unaffected.

Hydrolysis of both anomers of a substrate has previously beenobservedinfamily

GH4 enzymes,45,121 the elimination/hydration mechanism of which also acts largely

independent of this configuration. To test if this is also possible in UGL, 43,theaxial

analogue of phenyl ΔGlcA (10)wassynthesised(seeScheme2.10)andsubjectedto UGL-catalysed hydrolysis.

HO 1. NaOMe, MeOH MeOOC ZnCl MeOOC O 2. Ac2O, HClO4 O 2 O AcO AcO O OH AcO OAc phenol, AcO O OAc 80 °C AcO 2 O 1 OH 41 58% 6% DBU, DCM

COOH COOMe NaOH, acetone/H2O O O HO AcO HOO AcOO 43 42 64% 44%

Scheme 2.10: Synthesis of an axial phenyl ΔGlcA substrate for UGL.

The phenyl leaving group was attached in an axial configuration by reaction of

acetylated methyl glucuronate (2)inmeltedneatphenolassolventwithzinc(II)

chloride as a Lewis acid catalyst. The yield of pure alpha product was very low, as

amixtureofanomersformedandresolutionofthepuredesiredproductfromthis

87 2.6. Unusual substrates as well as the excess phenol proved difficult. However, as only asmallamountof this compound was required for testing and a sufficient amount of the intermediate

41 was in hand to achieve this, it was not deemed important to repeat or improve

on this method. Subsequent elimination by DBU to 42 and deprotection by NaOH

in acetone/water proceeded smoothly, if with slightly loweryieldsthanwiththe

equatorial phenyl analogue, to afford 43.Whileothersynthesesofthisproduct

have been published, 104,106,122 with better yields, this route represented the smallest

number of discrete steps required to achieve a compound for testing from available

starting materials.

Testing of this compound (43)asasubstratewithUGLrevealedalowerkcat and

higher Km than for the equatorial phenyl substrate 10 (see Table 2.3 on page 82, and Table 2.1 on page 65). The observation of any activity clearly demonstrates a

reaction that is largely independent of anomeric configuration and consistent with a

hydration-initiated hydrolytic mechanism, with the kcat of the two phenyl anomers

differing only by a single order of magnitude. The Km is increased by approxim- ately an order of magnitude also, which is unsurprising sincetheactivesiteislikely

optimised for binding of a group with an equatorial configuration at the anomeric

carbon. Hydrolysis of this substrate was found to deviate significantly from the

classical Michaelis-Menten kinetics at higher substrate concentrations, possibly as a

result of either substrate inhibition or non-specific enzymeinactivation(asdiscussed

in Section 4.3 on page 141).

Hydrolysis of both anomers of phenyl ΔGlcA by UGL from F. heparinum may have previously been observed,104 but the extent to which this occurred and whether

or not this activity came from a single enzyme was not clear as the authors pursued

an enzyme with activity specific for one isomer from one sourceGAG.Theseauthors

dismissed this observation as the activity was low by comparison.

88 2.6. Unusual substrates

2.6.3 Thiophenol

Thioglycosides are generally very poor substrates for glycoside hydrolases, and are in common use as inert substrate analogues. However, carbohydrate-active enzymes that use eliminative mechanisms, such as family GH4 enzymes 46 and polysaccharide

lyases,75 have been shown to be able to hydrolyse these substrates. Because the

hydration-initiated hydrolytic mechanism of UGL acts distal to the thioglycosidic

bond, UGL was anticipated to be able to hydrolyse a suitable ΔGlcA thioglycoside

only if such a mechanism were acting. Thiophenyl ΔGlcA was thus synthesised and tested as a substrate for UGL.

HO 1. NaOMe, MeOH MeOOC MeOOC O 2. Ac2O, HClO4 O HBr, AcOH O AcO AcO O OH AcO OAc AcO O OAc AcO 2 3 Br 1 OH 58% 97% Na2HCO3, thiophenol, TBAB EtOAc/H2O

COOH COOMe MeOOC NaOH, acetone/H2O DBU, DCM O AcO O O AcO S HO S AcO S OAc OH OAc 46 45 44 86% 56% 78%

Scheme 2.11: Synthesis of a thiophenyl ΔGlcA substrate for UGL.

Synthesis of this substrate proceeded along a similar route to that for the 4-

nitrophenyl substrate 6,asshowninScheme2.11.Glycosylationwasaccomplished

in good yield under phase transfer conditions with base catalyst, taking advantage

of the good nucleophilicity of thiophenol to afford 44 without requiring expensive silver reagents. This was then subjected to DBU-mediated elimination with a modest yield. The protected thioglycoside 45 was deprotected in one step using NaOH in

acetone/water as for the other phenyl glycosides (10 and 43), in good yield, to afford

46.

89 2.7. Conclusions

Hydrolysis of 46 by UGL proceeded very efficiently, with both higher kcat and

Km compared to its phenyl analogue 10 (see Table 2.3 on page 82, and Table 2.1 on page 65), giving a slightly higher kcat/Km as a result.

2.7 Conclusions

These results demonstrate conclusively that UGL catalyses the hydrolysis of unsat- urated glucuronides through hydration of the carbon 4–carbon 5 double bond to generate a hemiketal. It remains unclear whether or not the subsequent rearrange- ment of this hemiketal to give the final products is catalysed by the enzyme or occurs spontaneously in solution, and what, if any, intermediate steps are involved in this hydration process.

Optimal conditions for UGL-catalysed hydrolysis were determined to be pH 6.6,

37 °C and with 0.1 % w/v BSA. The optimal temperature was seen be different for kcat and kcat/Km,andnotdeterminedbyirreversibleglobalthermalinstability, but likely rather by local reversible instability.

UGL from Cperfringensappears to show similar substrate preferences to most of the species previously profiled in the literature when presented with substrates derived from natural GAGs, and is also able to hydrolyse aryl substrates with similar efficiency. One substrate in particular, the chondroitin-6-sulfate derived 11,showed ahigherkcat than any other substrates assayed in this work or reported elsewhere.

The products of the UGL-catalysed hydrolysis of aryl glucuronides in D2Oand 10% methanol are completely consistent with the mechanism proposed, as well as the tentative assignment of D173 as a catalytic acid, based on crystallography. Proton addition takes place stereoselectively from the face where this residue is located, while nucleophilic attack takes place from the same face. Indeed, awellresolvedwater molecule is also located here in the structure and is presumedtobethenucleophile.

90 2.7. Conclusions

These results also rule out a syn addition of this residue over the double bond.

As a further test of the hydration mechanism, several compounds were synthes- ised and tested, each of which was predicted to be accepted as asubstratebyUGL only if this novel mechanism were acting — Kdn2en (40), axial phenyl ΔGlcA (43), and thiophenyl ΔGlcA (46). All of these were indeed seen to be turned over, con- firming the results of the above experiments.

91 Chapter 3

Probing the mechanism of UGL

With the overall reaction catalysed by UGL established in thepreviouschapter, attention next turned to probing of the details of how this reaction was catalysed.

Specific goals were identifying the nature of the transition state for the hydrolysis reaction, establishing whether or not the rearrangements ofthehemiketalwerecata- lysed on-enzyme, and determining the role of D113, the secondresidueidentifiedpre- viously as catalytically important by mutagenesis experiments in the literature.87,92

3.1 Detection of initial products by NMR

In their publication on the crystal structure of UGL from Bacillus sp. GL1 with

bound substrate,87 Itoh et al. claim that the rearrangements of the initial hemiketal

product (22 in Scheme 2.3 on page 74) occur on-enzyme, but provide no evidence of this. While that would be a plausible mechanism, it remains possible that this un- stable intermediate degrades non-enzymatically. Since theinitialhydrationreaction is likely the slowest step the system gains nothing by accelerating this subsequent step. One means by which this hypothesis can be tested is by monitoring the reac- tion using NMR as it progresses, looking for short-lived intermediates in solution.

By using a high concentration of enzyme in the reaction any enzymatic steps should be accelerated, while rates for non-enzymatic steps should be zero order in enzyme concentration. Thus by maximising the difference between theenzymaticandany

92 3.1. Detection of initial products by NMR non-enzymatic rates the likelihood of detecting intermediates is also maximised, if such intermediates are present.

This experiment was carried out using thiophenyl ΔGlcA (46)asasubstrate

because of its high kcat,givingahighrateofenzymaticreactionunderthesaturating substrate conditions used, and also because sufficiently large quantities were available

to allow monitoring of the reaction by 1H-NMR. The substrate/buffer/BSA reaction

mixture was used to tune the spectrometer and to give a baseline for the unreacted

substrate, then the enzyme in D2Owasadded,thesolutionmixed,andmonitoring initiated as quickly as possible. The resultant spectra are shown in Figure 3.1. New

peaks derived from the products are already apparent in the t =0minscan,from

reaction in the time taken to lower the sample into the spectrometer and re-establish

the lock. These peaks increase in intensity over the next 15 minutes, at which

point the starting material is almost completely consumed. Allowing the reaction to

continue for longer, up to 266 min, does not appear to change the spectrum of the

product. At no stage are there any peaks formed that are no longer present in the

final scan, as would be expected if an intermediate were being formed and released

into solution — the only peaks visible at any stage are those from the starting

material and those from the final products.

While not observing an intermediate is not evidence for no intermediate, these

results do show that either no intermediate is released into solution or that it is

extremely short-lived. Since this rearrangement proceeds in a similar fashion to

mutarotation of free sugars, it could be expected that the kinetics would be similar.

Half-life values for uncatalysed mutarotation of simple sugars range from around 5

minutes to an hour, and are highly dependent on temperature and pH,123 but the

process can generally be observed using 1H-NMR. The best model system for the

rearrangement in the UGL reaction is likely that of mutarotation in sialic acid, for

93 3.1. Detection of initial products by NMR ics )withahighconcentrationofUGL. 46 GlcA ( Δ est trace, thiophenol in the reaction buffer. Numbers in ital zontal lines below the lowest trace. H-NMR spectra showing reaction of thiophenyl 1 Figure 3.1:ff O set stack of Lowest trace, initial spectrum before addition of UGL; high are integrals for the spectral regions indicated by the hori

94 3.1. Detection of initial products by NMR which the half-life at pD 6.7 is 25 min,124 and appears to be similar for Kdn.125

This suggests that the rearrangement of the hemiketal intermediate 22 in the UGL- catalysed reaction occurs on-enzyme. Examination of the enzyme active site and associated mutagenesis 87 shows a potential pair of residues that may be involved in general base catalysis of this reaction, H193 and Q211 (Bacillus numbering), and likely proceeding through a mechanism as shown in Scheme 3.1,byanalogytobase- catalysed mutarotation.126 In this possible mechanism, these two residues act to deprotonate the newly-introduced hydroxyl group at carbon 5inafastequilibrium, the transient product of which then undergoes a rate-limiting rearrangement to the products. This rearrangement may occur in either a concertedmannerorintwo steps. However, in the substrate-bound crystal structure ofUGLfromS. agalactiae

the placement of the histidine invoked in this mechanism is not conserved, despite

strong sequence conservation, and has no obvious replacement in its suggested pos-

sible role as base catalyst, casting some doubt on this role.

Q211 Q211 H193 H193 R OH N NH C HN NH C + O NH2 O NH2 H - O OH O O O O HOOC O HO OR (fast) HO OR (slow) HOOC OH HOOC OH OH 22 23

O - via O ? HO HOOC OH OR

Scheme 3.1: Mechanism for UGL catalysis of rearrangement of the hemiketal inter- mediate 22 to cleave the glycosidic bond.

95 3.2. Linear free-energy relationship

3.2 Linear free-energy relationship

Measurement of a free energy relationship for the UGL-catalysed hydrolysis of aryl glycosides was undertaken as a way of probing the transition state of the hydration reaction. Free energy relationships show the effects of electron-withdrawing and/or

-donating groups on the activation energy of a reaction (ΔG‡), through measure- ment of the effect on the rate of a reaction.127 For example, a partial negative

charge developing on a phenolic oxygen at the transition state will be stabilised by

electron-withrdawing substituents on the phenol ring, decreasing the free energy of

this transition state relative to that of the unsubstituted phenol, and thus giving

faster reaction. In addition to informing on the polarity of charge development, the

degree to which different electron-withdrawing or -donatingsubstituentseffectthe

rate of reaction can also provide information on the magnitude of charge develop-

ment. In measurement of free energy relationships for glycoside hydrolases, the pKa of a phenyl group is commonly used as a measure of its electron withdrawing ability,

with a more withdrawing substituent stabilising a negative charge on the phenolic

oxygen and thus lowering the pKa.

3.2.1 Synthesis of aryl glycosides

To measure such a free energy relationship for UGL, a series ofarylglycosideswas

required with leaving groups of varying pKa.Theseweresynthesisedchemically using the same methods as for the phenyl and 4-nitrophenyl ΔGlcA substrates (6 and 10)outlinedinScheme2.1andScheme2.2onpage59,withasummary of the

reaction conditions used and yields for each substrate presented in Table 3.1.

Glycosylation of aryl groups was performed using three different methods, de-

pending on the pKa.LowpKa phenols, up to 3-nitrophenol at 8.39, were attached using Koenigs-Knorr glycosylation in moderate to good yields (as exemplified in

96 3.2. Linear free-energy relationship

Table 3.1: Conditions used and yields for the synthesis of unsaturated aryl glucuronides. TCA, trichloroacetonitrile (Schmidt donor); BnBr, acylation of free hemiacetal 7 (refer Scheme 2.2 on page 59) with benzyl bromide. Numbering for intermediates in parentheses.

Substrate pKa Glycosylation Elimination Deprotection COOH Koenigs- O HO O OH Knorr, NO2 O2N 79% 55% 47 3.96 (54) (55) HCl(aq.),32% COOH Koenigs- O NO2 HO O OH Knorr,

O2N 85% 50% 48 5.15 (56) (57) HCl(aq.),56% COOH Cl Koenigs- O HO O Knorr, OH Cl Cl 65% 72% 49 6.39 (58) (59) NaOH, 63%

COOH Koenigs- O Knorr, HO O NaOMe OH NO2 54% 92% then LiOH, 6 7.18 (4) (5) 80%

COOH Koenigs- O NO2 Knorr, HO O NaOMe OH 53% 72% then H2O, 50 8.39 (60) (61) 85% COOH O HO O OH Cl TCA, 20% 72% 51 9.38 (62) (63) NaOH, 46% COOH O HO O OH TCA, 30% 51% 10 9.99 (8) (9) NaOH, 77% COOH O HO O OH TCA, 28% 4% 52 10.37 (64) (65) NaOH, 54% COOH O HO O OH BnBr, 31% 73% 53 15.40 (66) (67) NaOH, 96%

97 3.2. Linear free-energy relationship

Scheme 2.1), while for all other phenols the free hemiacetal 7 was activated to the

Schmidt donor using trichloroacetonitrile and catalytic DBU, then glycosylation of the appropriate phenol using BF3 diethyl etherate (as exemplified in Scheme 2.2). The leaving group for the benzyl substrate was attached by alkylation of the same free hemiacetal 7 using benzyl bromide and silver carbonate, as the other methods

were found to not be effective. Eliminations were all performed using DBU, also with

moderate to good yield. Deprotection was more varied, but in general also depended

on the pKa of the aryl group — most were deprotected under basic conditions, in either one or two steps, while for the substrates with the mostactivatedleaving

groups aqueous acid was used as they were not stable under basic conditions.

Overall yields ranged from 2.5% for 4-chlorophenyl ΔGlcA to 22% for 4-nitrophenol. The halide-substituted phenols in particular were found to be more difficult to man-

age, with lower solubilities, less visibility on TLC, and smaller Rf changes during reactions, which made monitoring difficult.

3.2.2 Kinetics

Kinetic parameters for the hydrolysis of these substrates byUGLaresummarised

in Table 3.2. The best substrates are those with a higher pKa,largelyasaresult

of changes in kcat rather than kcat/Km.Logplotsofkcat and kcat/Km against pKa

are given in Figure 3.2, revealing slopes of βlg = 0.16 ± 0.04 and βlg = 0.04 ± 0.05,

respectively. The error range for the plot of log(kcat/Km)containszero,andso is considered to be a flat line, showing negligible effect from the leaving group,

while the slope of log(kcat)isslightlypositiveandexcludeszero,showingthatmore electron withdrawing groups make this step slower. Such a relationship would make

sense in the context of a transition state involving a a partial positive charge that is

destabilised by electron withdrawing groups, either a smallamountofchargeclose

98 3.2. Linear free-energy relationship

Table 3.2: Michaelis-Menten kinetic parameters for hydrolysis of aryl unsaturated glucuronides by UGL.

kcat Km kcat/Km -1 -1 -1 Substrate pKa (s ) (mM) (mM .s ) COOH O HO O OH NO2 O2N 47 3.96 0.140 ± 0.005 0.16 ± 0.02 0.9 ± 0.1 COOH O NO2 HO O OH

O2N 48 5.15 0.64 ± 0.04 0.59 ± 0.06 1.1 ± 0.2 COOH Cl O HO O OH Cl Cl 49 6.39 0.312 ± 0.007 0.57 ± 0.04 0.55 ± 0.05 COOH O HO O OH NO2 6 7.18 2.05 ± 0.06 0.26 ± 0.02 7.9 ± 0.8 COOH O NO2 HO O OH 50 8.39 14.7 ± 0.5 0.88 ± 0.03 16 ± 1 COOH O HO O OH Cl 51 9.38 3.9 ± 0.1 0.84 ± 0.09 4.6 ± 0.6 COOH O HO O OH 10 9.99 4.3 ± 0.2 3.2 ± 0.4 1.3 ± 0.2 COOH O HO O OH 52 10.37 3.27 ± 0.02 0.88 ± 0.02 3.7 ± 0.1 COOH O HO O OH 53 15.40 18.6 ± 0.6 9.0 ± 0.5 2.1 ± 0.2

99 3.2. Linear free-energy relationship to the leaving group or a larger charge at a greater distance. This second situation matches what would be expected from the proposed hydration mechanism, with the addition of a proton to the double bond between carbon 4 and carbon 5, generating significant carbocation character at carbon 5, which is then quenched by attack of the nucleophile. However, the observation of two different slopes for the plots of log(kcat)andlog(kcat/Km)suggeststhattheseparametersrepresenttwodifferent steps, meaning any mechanism must involve at least two distinct steps. Further

discussion of this will be taken up in Section 3.5 on page 124 after all experimental

data have been presented.

Figure 3.2: Plot of log(kcat), A, and log(kcat/Km), B, against leaving group pKa for hydrolysis of aryl unsaturated glucuronides by UGL.

An apparent outlier in these is 3-nitrophenyl ΔGlcA, which has an unusually high kcat,severalfoldhigherthanmostotherarylsubstrates,although still not as high as the best natural substrates. Given that there are several other nitrophenyl substrates presented in Table 3.2, it is unlikely that this is a specific effect of the nitro group, and 2,5-dinitrophenol has one of its two nitro groups in the same position on the phenyl ring as 3-nitrophenol. This higher kcat may arise from a specific interaction

100 3.2. Linear free-energy relationship of the 3-nitro group with a subdomain that interacts with and improves turnover of some natural substrates, but given that there is no improvement in binding over any of the other substrates, a simple binding sub-domain interaction seems unlikely. It is also possible that this group is involved in some fortuitous interaction that results in optimisation of the catalytic residue placements, as suggested for the 6-sulfated natural substrates in Section 2.4 on page 65, but again it is difficult to conceive of a mechanism whereby the 3-nitrophenol group achieves this butthe2,5-dinitrophenol does not. This source of the high activity with this substrateremainsunclear.

Free energy relationships for other glycoside hydrolases generally follow one of

two patterns, depending on the mechanism employed and which step within that

mechanism is rate determining. For inverting hydrolases andretaininghydrolases

128 with rate-limiting glycosylation the slope of log(kcat/Km)andlog(kcat)against

pKa is negative, as the leaving groups with a more stabilised negative charge are activated and accelerate the reaction. For retaining hydrolases with rate-limiting

deglycosylation the plot is usually flat, reflecting the lack of effect of charge sta-

bilisation on transition state energy as the charged group isnolongerpresentin

this step. In some cases, such as with many retaining β-glycosidases, the plot will show a shift between these two cases as the rate-limiting stepchangeswithleaving

group ability.129–131 It is also possible to observe a flat plot for enzymes with rate-

limiting glycosylation if there is substantial proton donation to the leaving group at

the transition state132 as this will minimise the developing charge that needs to be

stabilised. Polysaccharide lyases 71 and glycoside hydrolases from family 448 as well

as some members of family 3165 (those operating with eliminative mechanisms) also

follow a similar pattern, showing either a negative slope or aflatplot,depending

on the extent of charge developed and whether this occurs in the rate-limiting step.

Flat plots can also be indicative of a non-chemical step beingrate-limiting,suchas

101 3.3. Effects of heteroatoms domain movement in the enzyme or substrate conformational changes,133 but kin- etic isotope effects (see Subsection 3.4 on page 113) show thistonotbethecasefor

UGL. These data thus show that UGL exhibits a different trend toallofthesecases, reflecting the different charge development pattern in its unusual mechanism.

3.3 Effects of heteroatoms

To further investigate this trend, a substrate with an even more activated leaving group, and thus even stronger electron withdrawing effect, was sought in (4-deoxy-

α-l-threo-hex-4-enopyranosyl fluoride)uronic acid (70). The high electronegativity of fluorine generally makes glycosyl fluorides very good substrates for carbohydrate- active enzymes, and they are in fact considered good substrates for essentially all previous glycoside hydrolases, 134 including those with eliminative mechanisms such as glycoside hydrolases from family GH31, 65 while 4-fluoro glucuronides are good substrates for polysaccharide lyases.71 In the hydration mechanism, however, this is expected to be a very poor substrate.

Because fluorine is an almost isosteric substitution with hydrogen and activity had already been observed with both anomers of phenyl ΔGlcA (10 and 43), the alternate anomer of (70), (4-deoxy-β-l-threo-hex-4-enopyranosyl fluoride)uronic acid (73), was also synthesised to test as a substrate. In order to investigate the effect of the proximity of this highly electronegative group on the transition state energy, a substrate with fluorine substitution at carbon 2 (78)wasalsosynthesisedandtested, while synthesis of a substrate containing fluorine at carbon 4(88)wasattempted but unsuccessful.

102 3.3. Effects of heteroatoms

3.3.1 α-andβ-ΔGlcA fluorides

Previous syntheses of glucuronyl fluorides reported in the literature were accom- plished by selective oxidation of carbon 6 of glucosyl fluoride using TEMPO cata- lysis.135 However, to achieve the desired unsaturated glucuronides itwasdeemed more convenient to form the protected glucuronyl fluorides directly to allow the same elimination reaction by DBU used for all other compounds reported in this work, rather than trying to re-protect the reactive glucuronyl fluoride. The two anomers of ΔGlcA fluoride were synthesised by similar routes, as shown in Scheme 3.2 and Scheme 3.3, differing only in the means by which the fluorine wasincorporated.The equatorial fluorine in 68 was incorporated using silver fluoride displacement of the axial bromine in 3,overallanetdoubledisplacementattheanomericcarbonfrom acetylated methyl glucuronate (2), first with HBr then with AgF. This reaction se- quence was much higher yielding than for the axial anomer as the axial glucuronyl bromide appears to be a more stable product than the axial glucuronyl fluoride un- der their respective work-up conditions, and the equatorialfluorideisnotworked up in the presence of excess acid. The axial fluoride in 71 was incorporated dir- ectly from acetylated methyl glucuronate (2)byuseofHFinpyridineovernightat

4°C.Theyieldforthisreactionwaslowbothbecausethereaction was slow and the product was unstable during work-up; extending the reaction to drive complete product formation resulted in increased hydrolysis, likelybyatmosphericwatereven in the sealed reaction vessel. However, this reaction generated sufficient product to continue with the synthesis, and the hydrolysis by-product and recovered starting material were both useful intermediates in other syntheses.Theeliminationreaction of both these products was lower yielding than for other substrates (see Table 3.1 on page 97 for comparison), which may be a result of the electron-withdrawing fluor- ide accelerating formation of over-eliminated side-products, as was also seen for the

103 3.3. Effects of heteroatoms

4-fluoro unsaturated glucuronide 87 in Scheme 3.5 on page 111. Deprotection over two steps by sodium methoxide then lithium hydroxide proceeded smoothly, while acetyl chloride in methanol unsurprisingly led to decomposition of the product.

HO 1. NaOMe, MeOH MeOOC MeOOC O 2. Ac2O, HClO4 O HBr, AcOH O AcO AcO O OH AcO OAc AcO O OAc AcO 2 3 Br 1 OH 58% 97% AgF, ACN

1. NaOMe, MeOH COOH COOMe MeOOC 2. LiOH, THF/H2O DBU, DCM O AcO O O AcO F HO F AcO F OAc OH OAc 70 69 68 70% 33% 85%

Scheme 3.2: Synthesis of an equatorial ΔGlcA fluoride substrate.

HO 1. NaOMe, MeOH MeOOC MeOOC O 2. Ac2O, HClO4 O HF, pyridine O AcO AcO O OH AcO OAc 4 °C AcO O OAc AcO 2 71 F 1 OH 58% 20%

DBU, DCM

COOH 1. NaOMe, MeOH COOMe O 2. LiOH, THF/H2O O HO AcO HOF AcOF 73 72 79% 34%

Scheme 3.3: Synthesis of an axial ΔGlcA fluoride substrate.

Testing of these two substrates revealed that the equatorialfluorinecompound70 was a very poor substrate, while the axial anomer 73 showed no detectable hydrolysis by UGL. This lack of any activity with 73 is somewhat surprising given that 70 shows detectable hydrolysis and both phenyl isomers are hydrolysed by this enzyme, but

104 3.3. Effects of heteroatoms it is possible that the equatorial anomer is able to make a weakinteractionthat the axial cannot, and a small decrease under the already low activity of 70 renders

this substrate’s hydrolysis undetectable above background. It seems unlikely that

steric arguments are appropriate here, given both the size ofthefluorineatomand

the observation that much larger axial phenyl ΔGlcA (43)canbeaccommodated by the enzyme active site, and 73 was also shown to bind to the UGL active site

as a competitive inhibitor with a Ki of 6.4 ± 1.2 mM Figure 3.3). The low activity detectable with 70 could not be saturated within the limits of the assay used (as

discussed in Subsection 2.2.1). Data from reaction progresscurveswerefitbylinear

regression to determine kcat/Km,andKi was then approximated as a surrogate for

Km (Figure 3.4), as was done for Kdn2en (40), to allow estimation of the kcat given in Table 3.3.

Figure 3.3: Minimal Dixon plot showing competitive inhibition of UGL by axial ΔGlcA fluoride (73)withaKi of 6.4 mM: 1/Vmax is shown as a dashed line. Sub- strate (6)wasatthefollowingconcentrations:250(!), and 400( )µM.

Taken in the context of the other heteroatom-substituted substrates already

105 3.3. Effects of heteroatoms

Figure 3.4: Inhibition of UGL by equatorial ΔGlcA fluoride (70)showingaKi of 10.7 mM, determined from the intercept with 1/Vmax (dashed line). Substrate (6) was at 125 µM.

Table 3.3: Kinetic parameters of UGL substrates with varied heteroatoms at the anomeric carbon (*, estimate based on Ki). Electronegativity (Pauling kcat Km kcat/Km Substrate scale) (s-1) (mM) (mM-1.s-1) COOH O HO F OH 0.00047 0.000044 70 3.98 ±0.00007 10.7 ± 0.9* ±0.000002 COOH O HO O OH 10 3.44 4.3 ± 0.2 3.2 ± 0.4 1.3 ± 0.2 COOH O HO S OH 46 2.58 9.3 ± 0.2 4.8 ± 0.3 1.9 ± 0.2 COOH OH O HO OH OH HO 40 2.55 0.036 ± 0.006 2.7 ± 0.4* 0.0133 ± 0.0002

106 3.3. Effects of heteroatoms presented, namely thiophenyl ΔGlcA with sulfur at the anomeric position (46,Sub-

section 2.6.3 on page 89) and the reference case of phenyl ΔGlcA with oxygen at the anomeric position (10,Section2.3onpage61),atrendisseenwhereinmore

electronegative atoms at the anomeric position destabilisethetransitionstateof

the rate-determining step. However, Kdn2en, with carbon at the analogous posi-

tion to the anomeric carbon of ΔGlcA (40,Subsection2.6.1onpage83),isaclear exception to this trend. Because these compounds all have quite different groups

at the anomeric position, ranging from a single atom through aglycerolchainto

an aromatic ring, it is difficult to draw any strong conclusionsfromthesetrends,

especially since binding of these substrates to the enzyme (as reflected in Km)isdif- ferent in each case, and only 4 datapoints are available. However, these results are in

broad agreement with the previous section. Assuming that UGLdoescatalysethe

rearrangement of the intermediate hydrate 22,assuggestedbytheNMR-monitoring

results in Subsection 3.1 on page 92, this low kcat with Kdn2en may reflect the slower release of the intermediate hydrate, as this substrate is unique in not being able to

undergo this rearrangement.

3.3.2 2,4-Dinitrophenyl 2F-ΔGlcA

The class of 2-deoxy-2-fluoroglycosides are known mechanism-based inactivators of

glycosidases operating via retaining Koshland mechanisms.136 The fluorine on car-

bon 2 inductively destabilises the positive charge that develops on carbon 1 in the

transition state of both the glycosylation and deglycosylation steps, slowing both

steps. By also including a highly activated leaving group suchasfluorideor2,4-

dinitrophenol the glycosylation step is accelerated, and ifthebalanceofthesetwo

rates is appropriate this gives rise to a trapped intermediate that only turns over

slowly. Given that UGL has been shown in Chapter 2 to work through a hydra-

107 3.3. Effects of heteroatoms tion reaction, and that the positive charge in that reaction is understood to develop on carbon 5 rather than carbon 1, it was expected that an unsaturated 2-deoxy-2-

fluoroglucuronide would be a substrate for these enzymes.

MeOOC MeOOC MeOOC O NaH2PO4/Zn, O Selectfluor, O AcO AcO AcO AcO acetone AcO nitromethane/ AcO AcO F OAc 374Br AcOH 75 88% 16% (+23% manno) 1. HBr/AcOH 2. 2,4-dinitrophenol/Ag2O, ACN

COOH COOMe MeOOC LiOH, DBU, DCM O O O AcO HO O AcO O AcO O THF/H2O F F NO F NO 78 2 77 2 76 NO2 O N O N O N 55% 2 74% 2 37% 2

Scheme 3.4: Synthesis of a 2-deoxy-2-fluoro substrate analogue for UGL.

Synthesis of 2,4-dinitrophenyl 2-deoxy-2-fluoro ΔGlcA (78)proceededasshown in Scheme 3.4. Glycal 74 was formed from glucuronyl bromide 3 (see Scheme 2.1

on page 59) by reductive elimination with zinc dust, then fluorinated with the elec-

trophilic fluorinating reagent Selectfluor (1-chloromethyl-4-fluoro-1,4-diazoniabicyclo

[2.2.2]octane bis(tetrafluoroborate)) in 5:1 acetic acid/nitromethane to give protec-

ted 2-deoxy-2-fluoroglucuronate 75 as well as its manno-configured epimer, in modest

overall yield. Generation of the substrate analogue then proceeded in a similar man-

ner to synthesis of 2,4-dinitrophenyl ΔGlcA (47)asoutlinedinSubsection3.2.1on page 96, with activation by HBr in acetic acid then glycosylation of 2,4-dinitrophenol

under Koenigs-Knorr conditions to 76, elimination by DBU to 77 and finally de-

protection by LiOH in THF/water, instead of by aqueous HCl, to 78.Yieldswere

different from those of the non-fluorinated compound as might be expected. Glyc-

osylation proceeded much more slowly and in lower yield as both stages of this

process involve development of positive charge in the transition states, which the

fluorine destabilises. The anomeric stereochemistry in thisreactionmaybesetby

108 3.3. Effects of heteroatoms the influence of the 2-fluoro group on the oxonium ion conformation, giving pref- erential attack from one face,137–139 or the destabilising effect of the fluorine on

charged intermediates may push the reaction towards a more SN2-like mechanism. Elimination gave a better yield in a shorter reaction time, perhaps suggesting that

the carbon 2 fluorine’s inductive effect extends to carbon 5, ontheoppositesideof

the pyranose ring, increasing the acidity of the proton at this position. Finally, 78

was overall more stable than 47,allowingdeprotectioninonestepbyaqueousbase

rather than the slow acid hydrolysis used for the most activated leaving groups in

Subsection 3.2.1.

Surprisingly, no detectable hydrolysis of 78 by UGL was observed under standard

assay conditions. Upon addition of a much larger amount of enzyme a very small

amount of activity was observed over that in a control with no enzyme, with the

rate estimated at 0.006% of that with the 2-hydroxy analogue 47 (an approximately

15 000 fold decrease). In order to determine if this lack of activity arises from poor

binding or poor turnover, 78 was tested as an inhibitor of UGL-catalysed hydrolysis

of 4-nitrophenyl ΔGlcA (5), and was found to be a competitive inhibitor as shown

by the Dixon plot in Figure 3.5, with a Ki determined by non-linear regression to

be 150 ± 10 µM. This is within the error range of the Km of the analogous non- fluorinated substrate 47,at160 ± 20 µM (Section 3.2 on page 96). This indicates

that binding to the enzyme is essentially unchanged, and thatthelargedecreasein

activity is a result of very poor turnover. While it is possible that this low activity

is a cumulative effect of the two electron-withdrawing substituents on the pyranose

ring (2,4-dinitrophenol and fluorine) and loss of potential hydrogen bonding by the

hydroxyl group on carbon 2, these would not be expected to havesuchalargeeffect.

Substitution by fluorine at carbon 4 reduced rates of glycoside hydrolysis only 8–500

fold (i.e. 0.2% activity at least, with most substantially more than this),140–142 yet

109 3.3. Effects of heteroatoms this is an analogous substitution relative to the site of charge development. Another possible explanation for this effect is proposed in Section 3.5 on page 124.

Figure 3.5: Dixon plot showing competitive inhibition of UGLby2,4-dinitrophenyl 2-deoxy-2-fluoro ΔGlcA (78)withaKi of 150 µM. Substrate (6)wasatthefollowing concentrations: 52( ), 86(!), 260( ), 780( ), and 1300( )µM

3.3.3 4-F substrate

Given the apparent positive charge developing at carbon 5 in the transition state of the rate-determining step, and the strong effect of a fluorine at carbon 1 on this, it was of interest to synthesise a substrate with fluorine at carbon 4 (88). This fluorine

would be directly adjacent to the developing charge, and so would be expected to

drastically reduce, if not completely ablate, hydrolytic activity. Previous syntheses

of 4-fluoro glucuronides all report oxidation to the uronic acid after incorporation

of fluorine,71,74,75,143–145 so a route to the product was designed with this in mind,

given in Scheme 3.5.

110 3.3. Effects of heteroatoms

1. HBr, AcOH OH OBz OBz 2. Ag2O/pNP, OBz O Me-DAST O ACN O F F BzO DCM, -30 °C BzO BzO O BzO BzO 81 OBz OBz NO 79OBz 80 63% 2 96% NaOMe, DCM/MeOH 1. MeOH, AcCl + TEMPO/NaOCl/ OH MeOOC 2. Ac2O, IEX H HOOC O O NaBr, H2O O F F F AcO O HO O HO O 84 OAc 83 OH 82 OH NO2 NO2 NO2 73% 95% 44%

Br2/CaCO3, hν, CCl 4 1. NaOMe MeOOC DBU, DCM COOMe COOH O F 2. LiOH F F AcO O 0 °C O O OAc AcO O HO O Br OAc OH 85 NO2 87NO2 88 NO2 + Br O DABCO, hν, ACN F AcO O OAc COOMe NO2 86 85% (total)

Scheme 3.5: Attempted synthesis of a 4-fluoro substrate, failing at elimination by DBU.

Selectively benzoylated galactose 79 was fluorinated using Me-DAST to afford

80 in very good yield, followed by activation to the glycosyl bromide by HBr then glycosylation of 4-nitrophenol under Koenigs-Knorr conditions (81)withamore moderate yield to give the desired leaving group. Deprotection of this under Zemplén conditions to 82 proceeded poorly, for unknown reasons, but subsequent oxidation to the acid 83 by TEMPO-catalysed oxidation with sodium hypochlorite proceeded in very good yield, and because of the aryl aglycone this productcouldbeeasilypurified by Sep-pak. Protection of 83 under mild acid conditions afforded 84 in good yield. In order to allow a convenient elimination reaction, bromine was incorporated at carbon

5byradicalphotobrominationingoodoverallyield,withamixture of the intended

111 3.3. Effects of heteroatoms axial brominated product (85)andasmallproportionoftheequatorialbrominated

product (86). Attempted elimination of this mixture using DBU proceeded very

rapidly at 0 °C, with a significant proportion over-eliminating to form side-products

in the time taken to run TLC. A small amount of starting material was recovered from

this reaction, which was exclusively the equatorially brominated 86.Elimination

of this recovered starting material was attempted, using a radical-initiated syn- elimination,146,147 but also without success. This disappointing result, one step away from the desired product, indicated that the protected product (87)wasmuchtoo reactive under the standard DBU-mediated elimination conditions to allow isolation.

If a synthetic route is to have any hope of achieving 87 oxidation to the acid likely would need to occur as the last step. Control of the elimination reaction without the influence of the carboxyl group at carbon 6 then becomes a more significant challenge.

COOH OH OPG F F O O F HO O O PGO O HO O Br OPG OH NO2 OH NO2 NO2

Br2, hν or OBz OBz OPG NBS/(BzO) , Δ O 2 O O F F F BzO O CCl4 or BzO O PGO O OBz DCM Br OBz OPG 81NO2 89 NO2 NO2

Scheme 3.6: Retrosynthetic analysis of 88 with late oxidation, and test of the first step.

Retrosynthetic analysis of a route to 88 with oxidation as a last step is presented in Scheme 3.6. This was envisaged as proceeding with introduction of bromine to carbon 5 of a 4-deoxy-4-fluoroglucoside under radical conditions followed by elim- ination as a means of giving the desired double bond cleanly without a 6-carboxyl group, then deprotection and TEMPO-mediated oxidation to the desired compound.

112 3.4. Kinetic isotope effects

As a test of this, radical bromination of glucoside 81 to 89 was attempted, with both thermal and photochemical initiation, but without any success. The adjacent fluor- ine likely has too much of a destabilising effect on the radicalintermediate,which was overcome by the capto-dative stabilisation in 84.Whileitremainspossiblethat this reaction could proceed with different protecting groups, the choice of group(s) is not immediately apparent. Determining this empirically was decided to be more work than this compound is worth given that it was only sought as a confirmation of other results already obtained, and even if the desired result were obtained (no activity as a substrate) other possible explanations for this would also be plausible.

Synthesis of this compound was thus abandoned.

3.4 Kinetic isotope effects

Further investigation of the transition state and potentialstepsoftheUGL-catalysed hydrolysis of unsaturated glucuronides can be achieved through measurement of kinetic isotope effects, wherein a specific atom in a substrateisexchangedforone of its isotopes and the effect of this on the rate of reaction measured. Generally speaking, such differences in rate arise from the effect of the isotopic substitution on the relative stability of the transition state compared tothegroundstateforthe two isotopomers, and can arise from events such as breaking orformingofbondsto the isotope, changes in hybridisation or charge at or near theisotopicsubstitution, and occasionally steric effects. A full discussion of the theory behind kinetic isotope effects is presented in Appendix C on page 300. In this work, hydrogen was replaced by deuterium on carbons 1 and 4 independently in an aryl unsaturated glucuronide, and rates of reaction for each compared to those for the non-deuterated control. The reaction rate in D2OcomparedtothatinH2Oforanon-deuteratedsubstratewas also measured.

113 3.4. Kinetic isotope effects

3.4.1 Synthesis of a substrate deuterated at carbon 1

MeOOC MeOOC MeOOC O O O AcO DMP, DCM AcO NaBD4 AcO AcO OH AcO AcO OH D O/THF AcO AcO O 2 AcO 79091 D 83% 58%

Ac2O/TCA, DCM

MeOOC 2,4,6-trichlorophenol, MeOOC MeOOC O Cl O O AcO Ag2O AcO HBr/AcOH AcO AcO O AcO D AcO OAc AcO ACN AcO DCM AcO 94 D Cl 93 Br 92 D 50% Cl 86% 91%

DBU, DCM

COOMe COOH Cl Cl O NaOH O AcO O HO O Cl H O/acetone Cl AcOD 2 HOD 95 Cl 96 Cl 84% 67%

Scheme 3.7: Synthesis of a 1-deuterated subtrate.

Asubstratewithdeuteriumincorporatedatcarbon1ofanarylunsaturated glucuronide was synthesised as outlined in Scheme 3.7. Starting from methyl glucuro- nate hemiacetal (7), the deuterium was first incorporated in two steps by oxidation

to a 1,5-lactone (90)usingDess-Martinperiodinanethenreductionofthislactone

with sodium borodeuteride, selectively at the lactone in thepresenceoftheester

protecting groups by virtue of its higher activity due to the strain of a planar centre

in a 6-membered ring, giving the deuterated version of the hemiacetal (91)inmod-

erate yield. Attempts at oxidation using Moffatt conditions were unsatisfactory as a

large proportion of acetylation was found to occur, and this product was inseparable

from the lactone. The deuterated hemiacetal was then acetylated (92)andbrom-

inated (93)ingoodyieldstoactivateitforglycosylationof2,4,6-trichlorophenol.

114 3.4. Kinetic isotope effects

This was chosen as it absorbs light at a wavelength with a low background and also allows for a hydrogenation reaction required for synthesis of the substrate deuterated at carbon 4 (see Scheme 3.8 on the next page), which renders nitrophenyl groups unsuitable despite their superior optical properties. Glycosylation was achieved us- ing Koenigs-Knorr conditions as before (58 in Table 3.1 on page 97) to give the

deuterated 2,4,6-trichlorophenyl glucuronide 94 in slightly lower yield than for the

non-deuterated isotopomer. This was then subjected to DBU-catalysed elimination

to afford 95 in good yield, which was subsequently deprotected by NaOH in 1:1

water/acetone to give the desired compound (96)withanoverallyieldof11%over

7steps.

3.4.2 Synthesis of a substrate deuterated at carbon 4

Synthesis of a substrate with deuterium incorporated at carbon 4 was less straightfor-

ward. Oxidation of carbon 4 of a selectively protected galacturonide or glucuronide

was found to give a very unstable product, likely because of the 1,3-dicarbonyl moi-

ety formed. Use of a glucoside or galactoside may have overcome this, but without a

carboxyl group at carbon 6 to activate the proton at carbon 5 this would have neces-

sitated more extensive protecting group manipulations to achieve the double bond

between carbons 4 and 5. Together these discouraged incorporation of deuterium by

sodium borodeuteride as for 96,andanalternativemethodwassought.

The route eventually employed is that shown in Scheme 3.8. Deuterium was in-

corporated by hydrogenation of 59 using deuterium gas with palladium on carbon as

acatalysttoaffordthe4-deoxy-4,5-dideutero-glucuronide 97 in near-stoichiometric

overall yield but of two isomers, arising from syn-addition of D2 to either face of the double bond. These isomers were easily separable by flash column chromatography,

and the more abundant desired isomer was then brominated under radical conditions

115 3.4. Kinetic isotope effects

COOMe MeOOC Cl NBS, MeOOC Cl Cl O O D2, Pd/C D (BzO)2, Δ D O AcO O AcO O AcO O EtOAc CCl D OAc 4 Br OAc OAc Cl Cl Cl Cl Cl 59Cl 97 98 63% 65% (+ 37% 4,5-isomer) DBU, DCM

COOH COOMe D Cl NaOH, D Cl O acetone/H2O O HO O AcO O OH Cl OAc Cl 100 Cl 99 Cl 54% 72%

Scheme 3.8: Synthesis of a 4-deuterated subtrate. in moderate yield to give a bromine at carbon 5 anti-periplanar to the remaining hydrogen at carbon 4 in 98. Elimination of HBr using DBU proceeded in similar yield to elimination reactions for previous compounds to give the protected product

99,whichwasthendeprotectedinmodestyieldusingNaOHin1:1water/acetone as for the 1- and non-deuterated isotopomers 49 and 96,togive100 in 16% yield over 4 steps.

3.4.3 Kinetic isotope effect measurements

Using these two deuterated compounds (96 and 100)theeffectoftheisotopesub- stitutions on kcat and kcat/Km were measured. For effects on kcat substrate was used at a concentration well in excess of Km in order to saturate the enzyme, and initial linear rates were determined. For effects on kcat/Km substrate was used at a concentration well below Km,andfirstorderrateconstantsweredetermined(spec- trophotometrically, by depletion) to remove any effect from variations in substrate concentration between samples. Averages of multiple reactions were then calculated, along with standard errors, and used to determine the ratio ofactivitywithhy- drogen over that with deuterium. The results are presented inTable3.4.Forthe

116 3.4. Kinetic isotope effects

4-deuterated substrate the result on kcat/Km was verified by competition of equal amounts of the two substrates in one reaction, using integrals of H-1, H-4, and aryl

signals in 1H-NMR as an alternate method to follow the reaction. This method gave

the same result within error (1.08 ± 0.03). The SKIE on UGL has previously been

87 reported by Itoh et al. to be 2.1 and 2.2 for kcat and kcat/Km,respectively(errors not given, but estimated from errors on individual kinetic parameters to give SKIE

ranges that overlap), which agrees well with the results reported here. No directly

equivalent experiments to these kinetic isotope effects fromdeuteriumatcarbons1

and 4 have previously been reported, but analogies can be madetoseveralother

cases (vide infra).

Table 3.4: Kinetic isotope effects from deuterium incorporation at carbon 1 and carbon 4 of 2,4,6-trichlorophenyl ΔGlcA, and solvent deuterium effect with 4- nitrophenyl ΔGlcA.

Substrate kH /kD for kcat/Km kH /kD for kcat COOH Cl O HO O Cl HOD Cl (96) 1.03 ± 0.02 0.93 ± 0.02 COOH D Cl O HO O OH Cl Cl (100) 1.06 ± 0.02 1.46 ± 0.09 COOH O HO O (6,in OH NO 2 D2O) 2.51 ± 0.05 2.69 ± 0.12

The kinetic isotope effects on kcat and on kcat/Km are clearly different from one another, for both the 1- and 4-deuterated substrates, and theerrorrangesexclude aunityeffect.Thissupportstheconclusiondrawnfromthelinear free-energy rela- tionship in Section 3.2 on page 96, that these parameters represent two independent and kinetically important chemical steps in the mechanism, with different charge densities and/or distributions. If the first step was the overall rate-limiting step these parameters would be expected to be the same, for exampleaswasthecasefor

117 3.4. Kinetic isotope effects

α-1,4-glucan lyase.65 What these two steps might be are discussed in more detail in Section 3.5 on page 124.

Kinetic isotope effects on kcat/Km represent the effect of isotope substitution on the energy of the transition state for the first irreversible step.148 Deuterium substitutions at carbons 1 and 4 both give small normal effectsonthisstep.Inthe accepted mechanism for hydration of a vinyl ether the initialadditionofaproton to the carbon-carbon double bond is indeed an irreversible process.149–151 If, for the sake of discussion, the first irreversible step in UGL-catalysed hydration is assumed to be protonation of the substrate double bond between carbons 4 and 5, a plausible explanation for both of these effects can be given in terms of hyperconjugation and hybridisation changes as follows. The effect from deuterium at carbon 4 can be seen as arising from two competing effects. Because the orbitals at this centre change in hybridisation from sp2 to sp3 on going to the transition state, an inverse KIE would be expected on this basis. However, substantial carbocationcharacterdevelopsat carbon 5, which is stabilised by hyperconjugation from the hydrogen/deuterium on carbon 4, and this will give rise to a normal KIE.152,153 Because the overall net KIE observed is a small normal effect, the effect from hyperconjugation can be seen to be larger than that from hybridisation. From this it can be deduced that the deuterium at carbon 4 assumes a pseudo-axial geometry at the transitionstate,beingcloseto the optimal geometry to give a high degree of overlap of the C−D σ-orbital with the empty p-orbital of the carbocation at carbon 5. An exampleoftheimportance

of this geometry in determining the magnitude of a KIE can be seen in the effect

from substitution of both hydrogens with deuteriums at carbon 3 of a substrate

sialoside for Micromonospora viridifaciens sialidase; as only one of the hydrogens is appropriately aligned for hyperconjugation, the KIE for this hydrogen substitution alone is almost identical to that observed for both substitutions.154 The small to

118 3.4. Kinetic isotope effects

insignificant effect on kcat/Km from deuterium at carbon 1 can be attributed to the inability of the carbon-hydrogen bond at carbon one to stabilise positive charge on the endocyclic oxygen as there is no vacant orbital to allow hyperconjugation. A possible transition state geometry to account for these effects is given in Scheme 3.9.

Cl Cl Cl Cl COOH D Cl OH OH O H HO O O O δ+ H δ+ HO Cl O Cl O Cl D δ+ D δ+ D Cl D D OH OH

Scheme 3.9: Deduced conformation of the transition state that would give rise to the observed KIE on kcat/Km,forthefirstirreversiblestep,leadingtoahypothet- ical oxocarbenium ion intermediate. Deuteriums at both carbon 1 and carbon 4 are represented in one molecule for simplicity, but each substitution was made inde- pendently. The transition state for this first irreversible step is assumed to strongly resemble the oxocarbenium ion product, but whether this oxocarbenium ion actually forms as a stable species is unclear. The magnitude of charge thought to develop at each centre is qualitatively represented by font size.

Catalysis by the enzyme enolpyruvylshikimate 3-phosphate synthase (AroA)

provides an interesting comparison with UGL, as shown in Scheme 3.10. Extensive

calculations of the reaction coordinate for both the non-enzymatic (acid-catalysed)155 and AroA-catalysed156 cases showed a difference in the transition states, with exper- imental KIE data in good agreement. Both reactions proceed through an irreversible initial protonation followed by fast attack of the nucleophile on the carbocation, but the enzymatic case exhibited a much earlier transition stateinformationofthecar- bocation (C−Hbondorderof0.24vs.0.6inthenon-enzymaticcase).Thisearlier transition state arose from substantial stabilisation of the carbocation intermedi- ate, as well as the transition state leading to it, by an “electrostatic sandwich” of two carboxyl-containing sidechains. While UGL also contains two such sidechains of importance in the reaction, the arrangement in the active site is very different

(see Figure 3.6). The KIE observed in AroA for two deuteriums in the equivalent

119 3.4. Kinetic isotope effects position to carbon 4 in UGL was 0.990 in the enzymatic case and 1.04 in the non- enzymatic case, reflecting a similar balancing of hyperconjugation and hybridisation effects, with the lower contribution from hyperconjugation in the earlier transition state of the enzymatic case perhaps reflecting a more substrate-like than product-like conformation, which limits orbital overlap. In this case there are two deuteriums in the substrate giving an effect from hybridisation, while onlyonecanhavesubstantial orbital overlap to give an effect from hyperconjugation, explaining the net inverse effect.

AroA: COO- COO- COO- COO- (slow) (fast) (fast) OH 2- - 2- - 2- - 2- O3PO O COO O3PO O COO O3PO O COO O3PO OH OH OH OH OH2 + O

COO-

UGL: COO- HO ?? O O OH O HO OR - HO OR - OH OOC O OH OOC + OH

R OH

Scheme 3.10: Comparison of the reactions catalysed by AroA and UGL.

Kinetic isotope effects on kcat represent the effect of isotope substitution on the energy of the highest transition state, the overall rate-determining step.148 The

nature of this second step remains unclear, but several observations about the kinetic

isotope effects observed can still be made. The effect from deuterium incorporation

at carbon 4 is very large, around the limit for a secondary isotope effect.157 This

suggests substantial positive charge at the transition state and a conformation that

gives rise to a large degree of overlap of the C−D σ-orbital with the empty p-orbital

120 3.4. Kinetic isotope effects

Figure 3.6: Comparison of side-chain carboxyl placement in UGL from Bacillus sp. GL1 (A, PDB: 2FV1) and AroA from E. Coli (B, PDB: 4EGR), showing their distance to the carbon at which charge develops. (PEP, phosphoenolpyruvate).

at carbon 5, as discussed for the effect on kcat/Km,butwithoutacompensatory effect from change in hybridisation. The effect from deuteriumatcarbon1ismuch harder to explain. Common sources of inverse isotope effects include changes in hy-

2 3 bridisation from sp to sp ,asmentionedabovefortheeffectonkcat/Km,decreases in hyperconjugation on going from the ground state to the transition state, crowding of associative transition states in nucleophilic attack, and small effects arising from induction. None of these effects are expected to be present in asimpleone-step mechanism for the hydration reaction. Previous examples of the effect of deuterium substitution at carbon 5 in hydrolysis of a glycoside, in manywaysanalogoustocar- bon 1 in unsaturated glucuronide hydration (as illustrated in Scheme 3.11), include small inverse effects in acid-catalysed hydrolysis of methyl5-{2H}-α-andβ-glucosides (0.987 and 0.971, respectively) attributed to induction, 158 and no effect in a retaining

glycosyltransferase, despite substantial charge from a highly dissociative transition

state.159 An induction effect of this magnitude seems particularly unlikely given the

KIE arising from deuterium on carbon 4 that suggests the majority of the charge

121 3.4. Kinetic isotope effects resides on the carbon rather than the oxygen at the transitionstates.TheKIE from deuterium at carbon 1 on kcat thus seems to be arising from a very different transition state from that of most other glycoside hydrolases.

COOH HOOC HOOC O O O HO OR HO OR HO OR HO HO HOD D D

OH OH OH O O O HO HO HO HO OR HO HO D OH D OH D OH

Scheme 3.11: Illustration of the analogous relationship of deuteriums in unsaturated 1-deutero-glucuronides (upper) to 5-deutero-glucosides (lower) on formation of an initial oxocarbenium-ion transition state.

The effect from deuterium in the solvent (solvent kinetic isotope effect, SKIE) is also difficult to interpret, as there are many possible effects in play at the same time, some of which could be having opposite effects to others.Sincetheenzymeis incubated in D2Othemajorityofexchangeableprotonsareassumedtobereplaced with deuterium, and this will have effects on hydrogen-bonding, acid-base reactions, sterics to some extent, and all properties derived from interactions of these. This means that the effect of isotope substitution on transition state energies is unlikely to arise from a single such factor. Given this caveat, the mostcommoninterpretation of a solvent kinetic isotope effect in a reaction where proton transfer is expected to occur in the rate-determining step(s) is based on pKa effects and the ease of A−H bond breaking and formation (where A is any atom but hydrogen). The SKIE values

for UGL, 2.51 on kcat/Km and 2.69 on kcat,arereasonableforarate-determining hydrogen transfer in each step. These do not, however, informonwhichhydrogen

is being transferred, or even if it is the same hydrogen in bothkineticallyrelevant

steps. For non-enzymatic hydration of vinyl ethers SKIE values of 1.56–7.1 have been

122 3.4. Kinetic isotope effects reported,43,150,155,160–162 while for the hydration of maltal by β-amylase a SKIE value of 8 was reported,41 with the values for UGL falling toward the lower end of this

range.

SKIE = 1.56 - 7.1

R' R' OH2 R' R R'' R R'' R R'' O O O O O O O R' + + R'' H H2O +/- H R ++HO O R' R' R' R R'' R R'' R H O O O O O O H H R'' + HO SKIE = 0.37 - 0.55

Scheme 3.12: SKIEs for hydration of vinyl ether acetals, proceeding by either initial hydration of the vinyl ether or initial hydrolysis of the acetal.

In the UGL-catalysed hydrolysis of unsaturated glucuronides, the most likely proton transfer that this isotope substitution will affect istheinitialhydrogendona- tion by the catalytic acid residue to the double bond between carbons 4 and 5.

As noted earlier for the other kinetic isotope effects, a single step cannot explain the different isotope effects on kcat and on kcat/Km as observed here, meaning a second kinetically important proton transfer is taking place to account for the SKIE

on kcat.ItisworthnotingthatSKIEsreportedforthefewcasesofhydrolysis of vinyl ether acetals determined to proceed through initial hydrolysis of the acetal, rather than initial protonation of the vinyl group, gave inverse effects characteristic of acid-catalysed acetal hydrolysis in the order of 0.37–0.55,150,161,163 arising from a

pre-equilibrium protonation then rate-determining reaction of the protonated state

(see Scheme 3.12). This is clearly not the case for UGL, providing further evidence

for reaction at the vinyl group rather than the acetal.

123 3.5. Conclusions, and possible alternate mechanisms

3.5 Conclusions, and possible alternate mechanisms

While the simple hydration-initiated mechanism presented in Scheme 1.9 on page 44 is plausible, several of the experimental results outlined in this chapter appear to in- dicate that this picture is not complete. The linear free energy relationship, showing aflatlineforkcat/Km but a positive slope for kcat,andthekineticisotopeeffects observed, showing different effects from deuterium substitutions at carbons 1 and

4onkcat/Km and on kcat,togethersuggestthattherearetwokineticallyrelevant steps — an initial irreversible step and an overall rate-limiting step. The electron withdrawing effect of the anomeric group appears to have a bigger effect on the lat- ter, and KIE data suggest that both involve substantial increases in positive charge at the transition state over that at the relevant ground states. An exception to this is the inverse effect at carbon 1 on kcat,whichcannotbeeasilyexplainedbyany direct effect on a positive charge in the transition state of that step.

Although a formal oxocarbenium ion intermediate could be invoked between the transition states for these two steps, such an explanation would not be in agreement with the kinetic isotope effects observed. A full positive charge at the intermediate would be expected to give no KIE from hyperconjugation in the second step, as the degree of charge would not change substantially on going fromtheintermediateto the transition state (or would decrease, giving an inverse effect from deuterium at carbon 4). Quenching of this oxocarbenium ion is also not likely to involve rate- limiting proton transfer as indicated by the SKIE on kcat.However,giventhepo- tential complexity of such effects some mitigating factor, asdiscussedattheendof

Subsection 3.4.3 on page 116, could potentially allay this concern. Glycopyranosyl cations are very short-lived species in water.164–167 While 2-deoxy glucopyranosyl cations are slightly longer lived168,169 and the enzyme active site would no doubt be optimised to stabilise this hypothetical intermediate, it seems unlikely that the

124 3.5. Conclusions, and possible alternate mechanisms lifetime is increased in the UGL active site anywhere near enough for its hydration to become rate-limiting. Rate-limiting rearrangement of the hydrated intermediate

22 also cannot account for these results, as the breakdown of this intermediate is unlikely to proceed through a transition state with the substantial positive charge indicated by the KIE results. Also unexplained in the simple hydration-initiated mechanism is the role of the second catalytically important aspartate residue (D88 in Bacillus sp. GL1, D113 in C. perfringens,refertoTable1.1onpage48).The crucial catalytic importance of this residue, even more so than the putative catalytic acid (D149 in Bacillus sp. GL1, D173 in C. perfringens), suggests it may play a role in stabilisation of the transition state of the rate determining step over that in the uncatalysed reaction. It seems likely that this residue is involved in formation of the intermediate species implied by the kinetic studies.

Given this evidence for an uncharged semi-stable intermediate in the hydration

reaction, several possible alternative mechanisms were considered. These mechan-

isms are outlined in Scheme 3.13, each involving a clear role for D88/D113 in sta-

bilising the transition state, either as nucleophile at carbon 5 (A), a nucleophile at

carbon 1 (B), or as an acid/base residue to activate the substrate carbon 2 hydroxyl

as a nucleophile (C). These potential mechanisms will be assessed in the following

paragraphs in the context of the experimental evidence presented in this work as

well as earlier publications.

The simplest role for this residue in stabilising the positive charge at carbon

5isthroughdirectnucleophilicattackatthiscarbon,showninmechanismAof

Scheme 3.13. This provides a neutral covalent on-enzyme intermediate that can

then be degraded by nucleophilic attack of water at carbon 5 togivethehydrated

product. If this attack by water proceeds through a dissociative transition state, this

scheme can account for the KIE on kcat/Km and kcat from deuterium at carbon 4.

125 3.5. Conclusions, and possible alternate mechanisms

A D173 D173 D173 O COOH O H OH OH H HOOC HO O O O O O O HO OR HO OR OH HO OR O O OH OH O COOH D113 D113 O 22 O D113 O B D173 D173 D173 O COOH O H OH OH H HOOC O O HO O O O O HO OR HO OR HO HO OR O OH OH O O O COOH D113 O D113 D113 O 22 C D173 D173 D173 O COOH O H OH OH H HOOC O O HO O O O O HO OR HO OR O HO OR O O OH H O OH O COOH D113 O D113 D113 O 22

Scheme 3.13: Possible mechanisms for UGL to account for KIE and LFER observa- tions.

However, this mechanism suffers from an inability to explain the inverse KIE on kcat from deuterium at carbon 1, as this effect would be expected to be similar to that on kcat/Km,botharisingasaresultofhyperconjugationtostabilisecharge developing on the ring oxygen. A further significant objection to this mechanism arises from the distance and orientation of the D88/113 carboxylate group relative to carbon

5ofthesubstrate.Inthesubstrate-boundcrystalstructureofUGLfromBacillus

sp. GL1 this distance is 5.33 Å, or 5.71 Å in the structure from S. agalactiae,and

furthermore is positioned on the same plane as the pyranose ring but on the opposite

side from carbon 5, which makes this residue too far away and poorly placed for a

126 3.5. Conclusions, and possible alternate mechanisms nucleophile (see Figure 2.6 on page 71 and Figure 3.6 on page 121). While it is possible that it moves during catalysis, to invoke this explanation is a further ad hoc adjustment to the mechanism for which there is currently no evidence.

Asecondpotentialroleforthisresidue,presentedinmechanism B in Scheme 3.13, better accounts for the placement of residue D88/D113. This mechanism proceeds through a similar type of nucleophilic catalysis, but with the nucleophile adding at carbon 1, rather than carbon 5, of the pyranose ring. Concomitant breaking of the bond from carbon 1 to the endocyclic oxygen gives an intermediate with a ketone instead of a positive charge at carbon 5. In a subsequent step this ketone is then hydrated by water as a nucleophile while the endocyclic oxygen attacks at carbon 1 to reform the pyranose ring, expelling the D88/D113 nucleophile to give the hydrated product 22.Thismechanismaccountsfortheisotopeeffects on kcat/Km in the same manner as mechanism A and, assuming bond formation is sufficiently early on the reaction coordinate for the secondstep,itcanprovide abetterexplanationforthekineticisotopeeffectsonkcat.Thisearlyattackin the reaction coordinate would be expected to generate substantial positive charge at carbon 5, explaining the KIE on kcat from deuterium at carbon 4 by hyperconjugation to stabilise this charge, and gives steric crowding of carbon1inthetransitionstate, accounting for the inverse effect of deuterium at this position, while also allowing for aweakdestabilisationofthetransitionstatebyelectron-withdrawing leaving groups as shown in the LFER and heteroatom substitutions. If the carbon 1 to endocyclic oxygen bond breaking is sufficiently advanced in the transition state of the first step this provides an explanation for the lack of leaving group effect on this step, as seen in the flat LFER plot on kcat/Km.ThedistanceandalignmentofD88/D113, while better in mechanism B than in mechanism A, remains a potential problem for this mechanism as the residue is 4.01 Å from carbon 1 in the substrate-bound

127 3.5. Conclusions, and possible alternate mechanisms crystal structure of UGL from Bacillus sp. GL1, or 4.48 Å in the structure from

S. agalactiae,andisnotoptimallyalignedfornucleophilicattacktodisplace the

endocyclic oxygen (see Figure 2.6 and Figure 3.6).

Afinalmechanism,labelledCinScheme3.13,involvesD88/D113 acting as an

acid/base catalyst for activation of the carbon 2 hydroxyl group as a nucleophile.

The distance of 2.34 Å from the oxygen on carbon 2 in the crystalstructureof

UGL from Bacillus sp. GL1 or 2.62 Å in that from S. agalactiae (see Figure 2.6

and Figure 3.6) means that no movement of residues is requiredduringcatalysis.

In this mechanism, similar to in mechanism B, formation of a ketone at carbon 5

stabilises the positive charge developing there as a result of protonation at carbon

4, in this case with formation of an epoxide across carbons 1 and 2 by the hydroxyl

from carbon 2 stabilising the resulting charge on carbon 1. Inthesecondstep,

nucleophilic attack of water at carbon 5 re-forms the pyranose ring with opening of

the epoxide, further aided by re-protonation by D88/D113. This mechanism accounts

for the lack of reactivity of carbon 2-substituted substrateanalogues(see3.3.2),as

these are unable to form the required epoxide. While epoxidessuchasthatinthe

intermediate shown for mechanism C are known to have a short half-life, and thus

are not very stable,170,171 they are much more stable than would be a carbocation

or oxocarbenium ion. This instability may in fact be beneficial, as an overly stable

intermediate can act as a kinetic trap to decrease catalysis by slowing the subsequent

step. The formation of an unstable ido-configured hydrated intermediate (22)may

similarly promote fast rearrangement, as seen in the poor stability of the methyl

ketal analogue to this intermediate (26)formedbyUGL-catalysedreactionindilute

methanol (refer to Subsection 2.5.2) compared to its gluco-epimer (28).

These three mechanisms present possible explanations for the data presented in

this chapter, but further work is required to discriminate between them. Attempts

128 3.5. Conclusions, and possible alternate mechanisms at such a discrimination are presented in the following chapter.

129 Chapter 4

Testing of alternative mechanisms

In Section 3.5 at the end of the previous chapter, a set of alternative mechanisms for UGL was proposed and discussed . Several experiments werecarriedoutinan attempt to discriminate between them and these form the basisofthischapter.

Based on the apparent importance of, and lack of clear role for,thecatalyticresidue

D88/113, a mutant in which this residue had been replaced by glycine was generated, and rescue of the largely catalytically inactive mutant was attempted with various small molecules. Further testing of these mechanisms was sought in the analysis of various compounds as potential competitive inhibitors, matching the geometry and postulated sites of charge development in these mechanisms. For mechanisms invoking nucleophilic catalysis by the enzyme, A and B of Scheme 3.13 on page 126, attempts were made to kinetically trap these covalent glycosyl-enzyme intermedi- ates. Finally, an attempt was made to synthesise a small molecule intermediate as postulated in mechanism C.

4.1 Attempted rescue of D113G mutant

Alongside their publication of a crystal structure of UGL, Itoh et al. 92 presented kinetic data on two mutants in which active site aspartate residues had been replaced by asparagine. These mutations were seen to have a dramatic effect on kcat,butlittle

effect on Km.Oneoftheseresidues(aspartate149inBacillus sp. GL1 or aspartate

130 4.1. Attempted rescue of D113G mutant

173 in C. perfringens —D149/173)waslaterproposedtoactasacatalyticacid,87 donating a proton to the double bond between carbons 4 and 5. The other of these aspartate residues (D88 in Bacillus sp. GL1 or D113 in C. perfringens —D88/113) had no role assigned beyond hydrogen bonding to hydroxyl 2 and3forsubstrate binding and stabilising the transition state by some unknownmechanism.Several possible roles for this residue that would account for its great importance in catalysis are proposed in the concluding section of Chapter 3. Testing of these roles, as either anucleophileoracid/baseresidue,wasattemptedbymutation of this residue to glycine in C. perfringens UGL. This is a much less conservative mutant than the asparagine introduced into Bacillus sp. GL1 UGL, creating a relatively large space in the UGL active site. The ability of exogenous small molecules to occupy this space and rescue activity of the D113G mutant was investigated.

The C. perfringens UGL D113G mutant was generated using the Quikchange method.172 Primers were designed to cover this region of the UGL gene, with a single mis-match introduced to codon 113 to make it encode glycine instead of aspartate.

These primers were used for PCR of the pET28a::UGL plasmid in Subsection 2.1.1 on page 54, followed by selective digestion of the template on the basis of methylation by the restriction enzyme Dpn1. The resulting mutant plasmidwastransformed into TOP10 E. coli cells by heat shock, the cells allowed to recover, and the plasmid then isolated and transformed into BL21(DE3) E. coli cells for expression. The mutant UGL protein was expressed and purified in the same way asthewild-type

(see Subsection 2.1.2 on page 54), using a new column to avoid trace contamination with wild-type enzyme (see Figure 4.1). The yield of the mutant enzyme was similar to that of the wild-type.

Initial characterisation of this mutant showed very low activity at pH 6.6, with

-1 kcat = 0.0033 ± 0.0001 s and Km = 0.8 ± 0.1 mM for hydrolysis of 3-nitrophenyl

131 4.1. Attempted rescue of D113G mutant

Figure 4.1: Elution trace for UGL D113G, showing A280 in blue, conductivity in brown, % buffer B in green (up to 100% at its maximum), and fraction numbers in red. Axes are eluted volume in mL and absorbance in mAU, with axes not shown for other traces.

ΔGlcA (50), one of the best substrates with wild-type UGL ( Table 3.2 on page 99).

These parameters indicate little to no change in Km and a drastic reduction in kcat to only 0.02% of wild-type. This is similar to the 0.008% residual activity reported

by Itoh et al. for a conservative asparagine substitution at the analogousposition.92

This D113G mutant enzyme also exhibited an altered profile of activity with pH. The

mutant UGL showed greater relative activity in more acidic solutions, with a single

point of inflection in the pH profile at 4.99 ± 0.05,asshowninFigure4.2.Given

that this profile was determined by measuring kcat/Km,whichreportsonreaction of free enzyme with free substrate, it is important to note that this is above the pKa measured for the substrate, at 4.5 ± 0.2 (Figure2.4onpage63),andsolikely represents an intrinsic ionisation of the enzyme itself, andnotthesubstrate.This

shift in pKa arising from mutagenesis of one aspartate to glycine is dramatic, and represents a large change in the local environment of the remaining ionisable residue

132 4.1. Attempted rescue of D113G mutant

—assumedtobethecatalyticacidresidueD149/173.Suchashift in pKa is similar to that observed for the nucleophile residue of a retaining glycosidase on removal of

the adjacent charge on the acid/base residue.173 While this indicates that the low

residual activity reported at pH 6.6 can partially be overcome by reaction at a more

acidic pH, the highest activity observed with this mutant is still around two orders

of magnitude less than that of the wild-type at its optimum pH.

Figure 4.2: Profile of first order rate for hydrolysis of 4-nitrophenyl ΔGlcA (6) by UGL D113G ( ,rightscale)andwild-type( ,leftscale),plottingkcat/Km (µM-1.min-1)againstpH.

Rescue of catalytic activity in this mutant was attempted with a variety of small

molecules at pH 6.6, but with no success. Addition of sodium azide or sodium formate

buffer at pH 6.6 to the UGL-catalysed hydrolysis of 4-nitrophenyl ΔGlcA and mon- itoring of the reaction by spectrophotometry (detection of released phenol) showed

no enhancement of hydrolysis at 100 mM of either small molecule. Further testing

of a wider range of nucleophiles also showed no rescue, with UGL, 4-nitrophenyl

ΔGlcA (6), and the small molecules (at molar concentrations) incubated at 25 °C and monitored daily by TLC. The potential nucleophiles tested were sodium formate,

133 4.2. Testing of potential inhibitor leads sodium acetate (both adjusted to pH 6.6 with NaOH), methanol, β-mercaptoethanol, sodium azide, sodium cyanate, sodium thiocyanate, and sodium cyanide. While this shows no rescue with any exogenous nucleophile, formate and acetate could also be expected to rescue activity if the mutated residue functionsasanacid/baseresidue, and so this experiment provides a disappointing lack of information on the role of this residue.

4.2 Testing of potential inhibitor leads

Compounds that mimic the transition state of a reaction are often very potent com- petitive inhibitors, taking advantage of the enzyme’s strong binding to the transition state, which stabilises it and thereby catalyses the reaction.174 The mechanisms pro-

posed in Section 3.5 on page 124 involve different transition states, with charge devel-

oping in different locations and in different magnitudes during the course of the reac-

tion. In particular, both the simple hydration-initiated mechanism in Scheme 1.9 on

page 44 and direct nucleophilic catalysis at carbon 1 in mechanism A in Scheme 3.13

on page 126 invoke development of positive charge on the endocyclic oxygen of the

substrate. By contrast, the ring-opening mechanisms B and C ofScheme3.13may

involve development of negative charge on this same oxygen, depending on the tim-

ing of bond breaking and bond formation. Thus, inhibition of the reaction by a

compound with charge matching that which develops in the transition state would

provide evidence for that aspect of the mechanism. Scheme 4.1showsarepres-

entation of the first step in a direct hydration mechanism and potential inhibitors

mimicking structures in this mechanism. Inhibition of Streptococcal UGLs has pre-

viously been reported for the monosaccharides d-glucuronic acid and d-galacturonic acid, with 0.4–3.3 % maximal activity remaining in reactionswithsubstrateator

85 around Km (0.2 mM) and inhibitor at 1 mM (no Ki value reported), while glycine

134 4.2. Testing of potential inhibitor leads

has been reported to competitively inhibit Bacillus sp. GL1 UGL with a Ki of 6.5 mM.93

COOH COOH HOOC OH

O (UGL) O+ O H O HO OR 2 HO OR HO OR OH OH OH 22

COOH COOH HOOC OH

+ NH2 HO OH HO O 101 OH OH 102 103

COOH OH O R = OH HO R OH + NH3 .TFA- 108

Scheme 4.1: Illustration of structural analogies from UGL substrates, intermediates, and putative oxocarbenium ion-like transition states to potential inhibitors: d/l- proline (101), shikimate (102), shikimate’s biosynthetic precursor 3-dehydroquinate (103), and deacetylated DANA (Neu2en, 108).

Proline (101)wasselectedasaasimpleleadcompoundexhibitingpositivecharge at a centre analogous to the endocyclic oxygen at the assay pH. Both isomers of proline were tested as competitive inhibitors, with no inhibition observed up to 10 mM, clearly different from the case of glycine (vide supra). The inhibition by glycine is presumably the result of binding at the active site as seen in the X-ray crystal structure of wild-type UGL from Bacillus sp. GL1 with glycine and DTT bound

(Figure 4.3).92 In this structure glycine binds with its nitrogen situated between

D88 and D149, but substantially closer to the latter, presumably taking advantage

135 4.2. Testing of potential inhibitor leads

Figure 4.3: X-ray crystal structure of the Bacillus sp. GL1 UGL active site showing glycine (centre) and all side-chains within 5 Å, from two perspectives (PDB: 1VD5).

136 4.2. Testing of potential inhibitor leads of ionic interactions to enhance binding, and the positioning of the positive charge is not analogous to the position of the substrate endocyclic oxygen. This binding mode is perhaps not available to the sterically larger proline, making the latter a better mimic of the postulated charge. These results are consistentwithapositivelycharged nitrogen residue at a position analogous to the endocyclic oxygen of UGL substrates not giving inhibition, suggesting that such a positive charge may not develop on the endocyclic oxygen during catalysis. However, other possible explanations for this lack of inhibition are also possible, including the displacement of tightly bound water molecules by the alkyl chain of proline, which is unabletoformhydrogen bonds. Designing of a transition state mimic with positive charge at an analogous position to carbon 5 or negative charge in an analogous position to the endocyclic oxygen is more difficult, because of steric limitations of thiscentreandtheadjacent negative charge on the carbon 6 carboxylate group.

The natural product shikimic acid (102)isacarbocyclicanalogueofUGLsub-

strates, albeit with different stereochemistries for two hydroxyl groups. This com-

pound was found to exhibit competitive inhibition of UGL withaKi of 3.0 ± 0.4 mM (Figure 4.4), and was not hydrated by UGL. This affinity is slightly better than

anticipated for a substrate mimic, given the Km of substrates with a small anomeric group such as ΔGlcA fluoride (10.7 mM, refer Table 3.3 on page 106), and similar to the binding affinity of Kdn2en (40)andmostofthenaturalsubstratestested(1–3

mM, Table 2.2 on page 68). While the stereochemical mismatch would be expected

to decrease binding affinity, examination of the substrate-bound crystal structure of

UGL suggests that such differences in stereochemistry could be sterically accommod-

1 ated at the active site. 102 may also be able to bind in a H2 conformation to allow the hydroxyls on carbons 2 and 3 to interact with D88/113 in a similar manner to the

2 substrate in its H1 conformation, placing the remaining hydroxyl in a pseudo-axial

137 4.2. Testing of potential inhibitor leads configuration as illustrated in Figure 4.5. Derivatisation of a suitable hydroxyl group in 102 (on carbon 1 or perhaps carbon 3) with an aryl group would likely improve upon the binding observed, while its biosynthetic precursor3-dehydroquinate(103)

or suitable derivatives may also furnish improved binding asaresultofthehydroxyl

and carboxylate groups at carbon 5 acting as mimics of the hydrated intermediate

22 (see Scheme 4.1).

Figure 4.4: Dixon plot showing competitive inhibition of UGLbyshikimatewith a Ki of 3.0 mM. Substrate (6)wasatthefollowingconcentrations:52( ), 86(!), 260( ), 780( ), and 1300( )µM.

Asubstratemimicintendedtotakeadvantageoftheinteractions of the substrate hydroxyl on carbon 2 and the enzyme D88/113 residue was soughtinthedeacetylated form of sialidase inhibitor DANA, Neu2en (108).175 The amine group in 108 was predicted to form a strong ionic interaction with D88/113, sothiscompoundwasex- pected to bind much more tightly than the related substrate Kdn2en (40). Synthesis of 108 proceeded from acetylated Neu5Ac2en (DANA, 104)bythemethodofGer- vay et al.,176,177 through an N -acetyl-N -Boc intermediate (105)with79%yieldover

138 4.2. Testing of potential inhibitor leads

A B . COOH COOH O O H O O O H OR D88/113 - 2 HO OR O H1 OH

substrate (ΔGlcA)

OH COOH

COOH OH OH 2 H1 OH HO OH COOH OH H O O shikimic acid O H 1 H2 D88/113 O-

Figure 4.5: (A) Representation of substrate (upper) and shikimic acid (lower) inter- action with D88/113 in the active site of UGL. (B) X-ray crystalstructureofthe Bacillus sp. GL1 UGL active site showing ΔGlcA and side-chains of all residues within 3 Å (N(D)88, D143, and R221), from two different perspectives (PDB file 2FV0).

139 4.2. Testing of potential inhibitor leads

OAc OAc OH AcO Boc O/ AcO HO COOMe 2 COOMe COOMe DMAP, NaOMe, O DCM Ac O MeOH O AcHN N BocHN Boc AcOAcO AcOAcO HO HO 104 105 106

NaOH, MeOH/H2O

OH OH HO HO COOH TFA, COOH H O O 2 O - + TFA . H3N BocHN HO HO HO HO 108 107 79% (over 4 steps)

Scheme 4.2: Synthesis of Neu2en. four steps, as shown in Scheme 4.2. Surprisingly, 108 did not show any inhibition of UGL when tested in the micromolar concentration range. Given the anticipated proximity of the 108 amine group and D113, and the strong ionic interaction this would be anticipated to give, this lack of binding is perplexing. However, while

1-amino glycosides are reasonably good inhibitors of glycosidases (low micromolar affinity),178 2-deoxy-2-fluoro glycosides are not,179 despite the similar proximity of complementary charges on the amino group and catalytic carboxylates, weakening any conclusions that can be drawn from this. Incubation of 108 with UGL and sub- sequent spectrophotometric monitoring of the reaction showed that this compound also was not hydrated by this enzyme, in agreement with the lack of activity seen with a fluorine substituent on this same carbon (78,Subsection3.3.2).Onepoten- tial argument against the relevance of the inactivity of the 2-deoxy-2-fluoro ΔGlcA substrate is that it has lost the hydrogen bonds made by the 2-hydroxy group, which may be crucial for activity. The amine in Neu2en (108)shouldbeabletomakemore similar hydrogen bonds to the analogous 2-hydroxy substrateKdn2en(40,Subsec- tion 2.6.1), and so potentially provides a control for this mitigating factor. However,

140 4.3. Anticipated trapping reagents for UGL

2-deoxy-2-amino glucosides are poor substrates for glucosidases, showing that this substitution is clearly not able to compensate completely. 179

4.3 Anticipated trapping reagents for UGL

Mechanisms A and B in Scheme 3.13 on page 126, which involve nucleophilic cata- lysis, invoke covalent glycosyl-enzyme intermediates. Under appropriate conditions these may be expected to be trapped as stable species. In orderforthesespecies to have an appreciable lifetime, their rate of formation needs to be accelerated re- lative to the rate of decomposition. Because the reactivity oftheUGLsubstrate

(ΔGlcA) carbon-carbon double bond is difficult to modulate, thiswasattemptedby placement of fluorine substituents in product mimics, in the hope that the reverse half-reaction could be observed.

4.3.1 2,3-Difluoro Kdn

Given the activity of Kdn2en (40)asasubstrateforUGL(seeSubsection2.5.2

on page 78), it was hoped that a sufficiently activated leaving group at carbon 5

of a compound resembling the hydration product of this reaction (22)wouldbe

able to drive the reverse of the normal reaction, as illustrated in Scheme 4.3 for the

example case of mechanism A in Scheme 3.13. Addition of a second fluorine at the

adjacent position, carbon 3 in Kdn numbering or carbon 4 in ΔGlcA numbering, might inductively destabilise transition states leading toandfromthisintermediate,

in an analogous strategy to the use of 2-deoxy-2-fluoro glycosides, giving a stabilised

glycosyl-enzyme intermediate. Placement of this additional fluorine at carbon 3

in an axial stereochemistry would also prevent abstraction of this proton by the

enzyme, preventing elimination of the intermediate to give an unsaturated uronide

as a product, and thereby preventing completion of the reverse reaction.

141 4.3. Anticipated trapping reagents for UGL

A D173 D173 D173 O COOH O H OH OH H HOOC HO O O O O O O HO R HO R OH HO R O O OH OH O COOH D113 D113 O 22 O D113 O

B D173 D173 O D173 O COOH OH H F F O O COOH O O O F O HO R HO R OH HO R O O OH OH O COOH D113 D113 O O D113 O

Scheme 4.3: Rationale for attempted trapping of UGL with 2,3-difluoro Kdn (114), comparing one proposed mechanism involving nucleophilic catalysis (A) and the corresponding anticipated mechanism of trapping (B).

Synthesis of 2,3-difluoro Kdn is presented in Scheme 4.4, following the method of Watts et al.180 Briefly, 3-fluoro Kdn (109)wassynthesisedfromd-mannose and

3-fluoropyruvate by means of Neu5Ac aldolase, then protectedintwostepsbytri-

fluoroacetic acid in methanol followed by acetic anhydride inpyridinetoafford110 in 46% over 3 steps, with a 4.35:1 ratio of axial to equatorial fluorine at carbon 3. The anomeric position was subsequently deprotected using hydrazine acetate in methanol to give 111 (differing from the dichloromethane solvent used in the reference cited) in good yield, then the anomeric position fluorinated by meansofmethyl-DAST

(differing from the DAST used in the reference cited) in low yield to the protected product 112.Thiswasthendeprotectedintwosteps,deacetylatingtogive 113 by sodium methoxide in dichloromethane/methanol then saponifying the methyl ester to afford the desired compound 114 in 5.5% total yield over 6 steps.

On testing, 114 showed unusual inactivation behaviour with UGL. At a high concentration of inactivator (20 mM; well below the buffer concentration of 40 mM), rapid time-dependent inactivation was observed, but with a poor fit to first order kinetics. However, at lower concentrations no inactivationwasobserved,asshown

142 4.3. Anticipated trapping reagents for UGL

OH OH 1. MeOH, TFA OAc OH OH OAc O Neu5Ac aldolase 2. Ac2O, pyridine HO O COOH O COOMe HO F O HO HO AcOAcO OH HO AcO HO F AcO COONa 109 110 F 79% 58%

H2NNH2.HOAc, MeOH

OH OAc OAc COOMe COOMe OH NaOMe, Me-DAST, O O O HO F AcO F AcO COOMe HO DCM/MeOH AcO DCM AcO HO HO AcOAcO AcOAcO 113 F 112 F 111 F 65% 34% 70%

NaOH, THF/H2O

OH F COOH F O OH OH O F HO HO HO OH HOOC HO HO 114 F HO 78%

Scheme 4.4: Synthesis of 2,3-difluoro Kdn.

in Figure 4.6. The possibility that inactivation was due to pHshiftsintheinactiv-

ation mixture was discounted by checking the pH of reaction mixtures before and

after inactivation. No change in pH was observed. The resultsinFigure2.3on

page 62 also showed no inactivation of the enzyme as a result ofionicstrength.An

assay for time-dependent inactivation was subsequently carried out using the equat-

orial ΔGlcA fluoride substrate 70 at the same concentration, and a similar rate of inactivation was seen (Figure 4.7). This compound is not predicted to inactivate based on any mechanism proposed, suggesting that the inactivation seen for 114 is not mechanism-dependent. Indeed the axial phenyl substrate 43 (2.6.2) showed decreasing activity when testing hydrolysis of substrate at20mMorabove,which

143 4.3. Anticipated trapping reagents for UGL

Figure 4.6: Time-dependent inactivation of UGL by 2,3-difluoro Kdn (114)at0( ), 0.2( ), 2( ), and 20( )mM.Fitsaretofirstorderdecay.

Figure 4.7: Time-dependent inactivation of UGL by ΔGlcA fluoride (70)at20mM. Fit is to first order decay.

144 4.3. Anticipated trapping reagents for UGL may have arisen from the same non-specific inactivation phenomenon. The reason for this inactivation by 70 or 114 at high concentrations remains unclear. Further

testing also showed 114 to not be a substrate of UGL. Monitoring of the reaction over a period of days at 25 °C by TLC and 19F-NMR showed slow degradation of the compound to 3-fluoro Kdn 109 and fluoride at the same rate with or without enzyme, as illustrated in Figure 4.8.

Figure 4.8: 19F-NMR showing partial hydrolysis of 2,3-difluoro Kdn (114)withUGL (lower) and in a non-enzymatic control (upper) after 1 week. Labelled are fluorines 2and3of114 as well as fluorine 3 of 109 (labelled 3!)andfreefluoride.

4.3.2 4-Deoxy-1,5-difluoro-iduronic acid

Given the lack of inactivation by 2,3-difluoro Kdn, a better product mimic for use

as a mechanism-based inactivator was sought in a compound with fluorine at the

anomeric carbon, to stabilise the molecule, while retainingtheactivatedleaving

group fluorine at carbon 5. This compound, 4-deoxy-1,5-difluoro-iduronic acid (120),

was anticipated to be a potential mechanism-based inactivator for the same reasons

outlined in Scheme 4.3 of Subsection 4.3.1, although abstraction of a proton from

145 4.3. Anticipated trapping reagents for UGL carbon 4 of a covalent glycosyl-enzyme to give elimination toanunsaturateduronide product might be possible in this case.

O O O Ph Ph O O 1. H , Pd/C, O Ph 2 O O DBU EtOAc O O DCM O O AcO 2. PhCOCH2Br/ AcO F AcO F AcO F OAc OAc TEA, EtOAc OAc 115 116 117 84% 48%

1. NBS, hν, CCl4 2. AgF, ACN

F F F O NH3, MeOH O H2, Pd/C O HO F AcO F MeOH/H O AcO F OH OAc 2 OAc HOOC HOOC O O O 120 119 12% 69% 118 Ph 39%

Scheme 4.5: Synthesis of 4-deoxy-1,5-difluoro-iduronic acid.

Synthesis of this compound proceeded by the same route as thatusedforsyn- thesis of 1,5-difluoro iduronic acid by Wong et al., 181 with additional steps to gen- erate the 4-deoxy analogue as shown in Scheme 4.5. Starting from the protected glucuronyl fluoride intermediate 115 of this same reference, carbon 4 was deoxygen- ated by DBU-mediated elimination (116)followedbyhydrogenationandreprotec- tion of the phenacyl ester at carbon 6 (117)in40%yieldover3steps.Fluorinewas subsequently incorporated at carbon 5 by radical bromination with photochemical initiation followed by replacement of bromine by fluorine using silver (I) fluoride to afford the protected final product (118)inlowyield.Thiswasdeprotectedover2 steps, removing the phenacyl ester by hydrogenation (119)thendeacetylatingby ammonia in methanol to give the desired 120.Whilethedeprotectionappeared to proceed smoothly by TLC, this final product was unstable, decomposing during both Sep-pak and flash column chromatographic purification, but a small sample of

146 4.3. Anticipated trapping reagents for UGL pure product was isolated for testing. Unsurprisingly, an attempt at synthesis of an analogous compound with a phenyl group at the anomeric position by a similar path- way failed when the product decomposed completely on attempting deprotection by sodium methoxide followed by lithium hydroxide.

Figure 4.9: Time-dependent inactivation of UGL by 4-deoxy-1,5-difluoro-iduronic acid (120)at0( )and5(!)mM.Fitsaretofirstorderdecay.

Incubation of UGL with 5mM 120 resulted in no inactivation, as shown in Fig- ure 4.9. Nor was there any sign of hydrolysis or elimination, as determined by TLC and 19F-NMR (Figure 4.10). However relatively rapid non-enzymatic decomposi- tion was observed, as might be expected from its behaviour on deprotection, to give the same final product as UGL-catalysed hydrolysis (24 and free fluoride, shown in Scheme 2.3 on page 74). As with 114,pHwastestedbeforeandafterreactionat higher concentrations and confirmed to not change. This non-reactivity with UGL was determined to not be a result of poor binding, since 120 was seen to be a com- petitive inhibitor of UGL with a Ki of 7.5 ± 0.8 mM (see Figure 4.11 on page 149), demonstrating a slightly higher binding affinity than the analogous ΔGlcA fluoride substrate (70 at 10.7 mM, refer 3.3.1 on page 103).

The lack of inactivation or turnover of 114 and 120 by UGL appears to be

147 4.3. Anticipated trapping reagents for UGL

A .

B .

Figure 4.10: Overnight hydrolysis of 4-deoxy-1,5-difluoro-iduronic acid (120)with UGL (right lanes of TLC, upper spectrum of NMR) and in a non-enzymatic control (left lanes of TLC, middle spectrum of NMR) as monitored by TLC(A)and19F- NMR (B), with starting material in the lower spectrum.

148 4.3. Anticipated trapping reagents for UGL

Figure 4.11: Dixon plot showing competitive inhibition of UGL by 4-deoxy-1,5- difluoro-iduronic acid (120)withaKi of 7.5 mM. Substrate (6)wasusedatthe following concentrations: 25( ), 100( ), 300(!), and 600( )µM. aresultofinappropriatereactivityorplacementofreactive moieties. Given that

fluoride is a very good leaving group, and would certainly be expected to react if a suitably nucleophilic side-chain is appropriately placed to displace it, this suggests that mechanism A in Scheme 3.13 (in which a nucleophilic side-chain attacks at carbon 5) is not correct. Mechanism B (in which a nucleophilicside-chainattacksat carbon 1) and mechanism C (in which an acid/base side-chain activates the carbon

2hydroxylasanucleophiletoattackatcarbon1)bothremainplausible as the immediate leaving group for both cases is the endocyclic oxygen, which is expected to be only slightly activated by the inductive effect of the adjacent fluorine. In these mechanisms the driving force for nucleophilic attack is postulated to be donation of a proton to carbon 4 and the development of positive charge at carbon 5 (refer to discussion in Section 3.5), and thus would not be substantially influenced by a

fluorine at carbon 5 unless a suitable residue were present to accelerate its departure,

149 4.3. Anticipated trapping reagents for UGL such as by hydrogen bonding. The D88/D113 catalytic acid residue would possibly

fill this role, since the D88/D113 carboxylate to carbon 5 distance is 3.28 Å. However, the distance to a heteroatom substituent on this carbon in thehydrationproduct is presumably slightly greater (compared with 2.98 Å to carbon 4). This increased distance, combined with any difference in side-chain placement of this residue when the product is bound, rather than the substrate, may be sufficient to explain the lack of efficient proton donation to this fluorine.

4.3.3 1-Fluoro-ΔGlcA fluoride

The potential UGL inactivator 1-fluoro-ΔGlcA fluoride (124)wasdesignedasa further test for rearrangement of the initial hydrate product (22)intheUGLactive

site. Inactivation of UGL by this compound was expected to arise from hydrolysis

of one of the glycosyl fluorides to give an acyl fluoride at the anomeric carbon,

which can subsequently react with any nucleophile at or closetotheactivesitein

anon-specificmanner,givinginactivationoftheenzymeasshown in Scheme 4.6.

Although 124 was expected to only be very slowly hydrolysed by UGL, based on the results with the unsaturated glucuronyl fluorides 73 and 70 in Subsection 3.3.1 on page 103, each enzyme only needs to catalyse a single turnover to inactivate, and so reactivity does not need to be high for this probe to provideananswertothis question. If inactivation were to be seen with this molecule it would be indicative of rearrangement taking place in the active site, while no inactivation would be uninformative, as the acyl fluoride may be sufficiently long-lived to leave the active site, the probe may be too deactivated to form the acyl fluoride, or the rearrangement may be taking place outside of the enzyme active site.

Synthesis of this probe was accomplished over three steps from the glucuronyl

1,5 lactone 90 in 11% overall yield, as shown in Scheme 4.7. Both fluorines were

150 4.3. Anticipated trapping reagents for UGL

O COOH HO HOOC O O OH O UGL UGL? HO O UGL-Nu HO F HO F O +H O HO -HF HO -F- HO 2 COOH F F COOH F HO 124 Nu UGL

Scheme 4.6: Rationale for attempted trapping of UGL with 1-fluoro-ΔGlcA fluoride (124), showing reaction to form the hydrated product and subsequent rearrangement to form an acyl fluoride followed by trapping of an active site nucleophile (Nu). introduced in one step by protracted treatment with methyl-DAST to afford 121 in a low yield, but a sufficient amount to proceed with synthesis. A contaminant arising from elimination of a fluorinated intermediate was also observed (122). The desired di-fluorinated intermediate was then subjected to DBU-mediated elimination in good yield to give 123,followedbydeprotectionovertwostepsinverygoodyield by acid-catalysed trans-esterification in methanol then hydrolysis of the methyl ester by lithium hydroxide in a mixture of water and tetrahydrofuran to give the desired probe 124.Thisschemerepresentsanovelrouteforintroductionoftwofluorinesat the anomeric position, with previous routes typically usingasequentialschemeof halogenations, such as chlorination by hydrochloric acid, then radical bromination at the anomeric position, followed by replacement of these halogens with fluorine using silver fluoride.182–186 The previously employed reaction sequence was unsuitable for auronicacidbecauseofthehighreactivityofcarbon5underradical bromination conditions.

Previously, 1-fluoroglycosyl fluorides have been observed toactasslowsubstrates for inverting and retaining α-andretainingβ-glycosidases, as well as trehalase, 187 and α-1,4-glucan lyases of glycoside hydrolase family 31, 65 despite being anticipated to give mechanism-based inactivation similar to 2-deoxy-2-fluoroglycosides. This was attributed to a stabilising back-bonding effect from the fluorine p-orbitals compensat- ing for the destabilising inductive effects on transition state energies.134 Compounds

151 4.3. Anticipated trapping reagents for UGL

MeOOC MeOOC COOMe COOH O Me-DAST O DBU 1. AcCl, MeOH AcO AcO O O AcO DCM AcO F DCM 2. LiOH, AcO O AcO AcO F HO F F H2O/THF 90 121 AcOF HOF 16% 123 124 73% 98% MeOOC O AcO AcO + F AcO 122 10%

Scheme 4.7: Synthesis of 1-fluoro-ΔGlcA fluoride with additional fluorides at either carbon 2 or carbon 5 in addition to the geminal anomeric fluorines were observed to not react as either inactivators or substrates, despite evidence of their binding to the enzyme active site. 185 Unfortunately, 124 was seen to give non-specific inactivation of UGL, as shown in Figure 4.12, sim- ilar to that seen with 2,3-difluoro Kdn (114)andΔGlcA fluoride (70). The rate of this inactivation did not appear to vary in any meaningful way with changes in concentration above 20 mM, and fits to first order kinetics for this inactivation were relatively poor, while concentrations below this did not appear to give any inactiv- ation. Given the apparent binding of both anomers of ΔGlcA fluoride (70 and 73, see Subsection 3.3.1 on page 103), there is no reason why 124 should not bind. This inactivation is thus unlikely to be mechanism-based, and is more likely to either arise from some contaminant (no contaminant was detected, but it would not necessarily be expected to be, as the concentration of enzyme in an inactivation reaction was at around 0.02% that of the inactivator in a 20 mM reaction) or from non-specific effects such as an increase in surface adsorption as seen with yeast α-glucosidase with 1-fluoro-d-glucopyranosyl fluoride.187

152 4.4. Attempted synthesis of proposed epoxide intermediate

Figure 4.12: Time-dependent inactivation of UGL by 1-fluoro-ΔGlcA fluoride (124) at 0( ), 5.5( ), 11( ), 22.1( ), 36.8(!), and 51.5( )mM.Fitsaretofirstorder decay.

4.4 Attempted synthesis of proposed epoxide

intermediate

As a direct test of the epoxide-based mechanism C in Scheme 3.13 on page 126, a stable form of the proposed epoxide intermediate was sought to assay for catalytic competence. Reports of alkoxy oxiranes in the literature,170,171 which would be the

intermediates formed from O-glycosidic substrates if this mechanism were acting,

indicate half-lives in water on the order of minutes. While this would be consistent

with enzymatic intermediacy, it would be difficult to synthesise these for testing.

However, the epoxide formed from the C -glycoside analogue Kdn2en (40), as illus-

trated in Scheme 4.8, should be much more stable, and thus presents a viable target

for synthesis.

Given that synthesis of Kdn (32,Scheme2.9onpage84)wasachievedusing

153 4.4. Attempted synthesis of proposed epoxide intermediate

COOH HOOC HO OH UGL? OH UGL? O OH O OH O OH OH HO HO HO OH O OH COOH 40HO 133 HO HO

Scheme 4.8: Epoxide intermediate expected from Kdn2en (40)accordingtomech- anism C of Scheme 3.13

the enzyme Neu5Ac aldolase and mannose, as opposed to the natural substrate N -

acetyl-mannosamine, the active site of this enzyme was investigated to determine

if unsaturated substrates could be expected to be accommodated. A published set

of X-ray crystal structures of Neu5Ac aldolase with several product-like inhibitors

bound188 showed that the product and Schiff’s base intermediate of thisenzyme,

and presumably its substrate also, bind in the active site in an open chain extended

conformation, as illustrated in Figure 4.13. While the 4-hydroxy group of Neu5Ac

(Neu5Ac numbering), derived from the anomeric aldehyde of N -acetyl-mannosamine,

was found in the above publication to be important for the formation of the Schiff’s

base intermediate, the carbon 5 and 6 hydroxyl groups (Neu5Acnumbering)appear

to make few contacts with the enzyme and thus are plausibly less important for

binding and catalysis. Based on these observations, it was hoped that aldehydo-

2,3-dideoxy-erythro-trans-hex-2-enose (131,anα,β-unsaturated hexose enal) would also be accepted as a substrate by Neu5Ac aldolase and provideafacilemeansof constructing the desired carbon skeleton.

Asimplesynthesisof133 was envisaged in extension of the α,β-unsaturated hex- ose enal 131 (also known as a Perlin aldehyde) to an appropriate non-5-enulosic

acid by use of Neu5Ac aldolase followed by epoxidation to the desired compound, as

shown in Scheme 4.9. Rearrangement of acetyl and benzyl protected glucal (125 and

126)byhafniumtetrachlorideandzinciodideunderreflux,following the method

of Saquib et al.,189 proceeded in lower selectivity than reported to give a mixture of

154 4.4. Attempted synthesis of proposed epoxide intermediate

Figure 4.13: X-ray crystal structure of Haemophilis influenzae sialic acid aldolase (Neu5Ac aldolase) active site188 with the inhibitor 4-oxo-sialic acid bound as a stable Schiff’s base derivative of Lys164, showing all amino-acid side-chains within 3.5 Å (PDB: 1F7B).

cis (127 and 128)andtrans(129 and 130) α,β-unsaturated hexose enal products, while acetate migration from hydroxyl 4 to hydroxyl 5 was alsoobserved.These

products proved difficult to separate, especially those with benzyl protecting groups,

so deprotection was undertaken on the crude product mixture.Unfortunately,the

α,β-unsaturated hexose enals were found to be very difficult to deprotect and purify (vide infra). An attempted rearrangement of glucal (134)inwatertotheunprotec-

ted α,β-unsaturated hexose enal was unsuccessful, forming predominantly the over- eliminated product 135 and a small amount of the hydrated product 2-deoxy-glucose

(136)showninScheme4.10,asreportedforsimilarreactionscatalysed by indium

trichloride, iron trichloride, and acidified mercuric salts.190–193

For the acetyl-protected 127 and 129,deprotectionwasattemptedunderavari-

ety of standard conditions. Trans-esterification with methanol, catalysed by either

hydrochloric acid or sodium methoxide, gave rise to Michael-type addition to the

155 4.4. Attempted synthesis of proposed epoxide intermediate

OR O RO OH 127 R = Ac OBn: OR 128 H2, Pd/C OH HfCl /ZnI R = Bn O 4 2 or AlCl3 O RO + HO RO ACN/H2O, Δ OR OAc: NaOH/H O/acetone OH H O 2 RO or NH3/MeOH 125 R = Ac or NaOMe/MeOH 131 OH H 126 R = Bn or HCl/MeOH 129 R = Ac COOH 130 R = Bn O Neu5Ac aldolase O HO COOH mCPBA OH OH or COOH HO HO O O O OH OH O OH 133 132

Scheme 4.9: Attempted synthesis of the epoxide intermediateexpectedfromKdn2en (40)undermechanismCofScheme3.13

OH HO OH HfCl4/ZnI2 O O + O HO HO HO HO H2O HO OH 134 135 136

Scheme 4.10: Hafnium tetrachloride and zinc iodide-catalysed rearrangement of d- glucal in water.

α,β-unsaturated moiety. Saponification of the ester with sodiumhydroxideshowed signs of forming the desired product, but on quenching of the reaction and evap- oration of solvent (either by rotary evaporator or lyophilisation) a black insoluble compound was seen to form, presumed to be a result of self-polymerisation of the poly-ol with the α,β-unsaturated moiety, again by Michael-type addition. Finally, methanol saturated with ammonia gave a new product with mass consistent with

further elimination of acetate, with no sign of the desired deprotection product 131.

Deprotection of the benzyl-protected 128 and 130 was slightly more success-

156 4.4. Attempted synthesis of proposed epoxide intermediate

ful. Lewis acid-catalysed (AlCl3)benzyletherhydrolysisandpalladiumoncarbon- catalysed hydrogenation were attempted, the former giving no reaction but the latter showing a mixture of several products, with indications of the desired compound be- ing seen in a mass spectrum of the crude product mixture. However, this reaction proved not to be repeatable, and analysis of crude product mixtures was complic- ated by the formation of a hydrate from the aldehyde when dissolving the product in water, removing its characteristic peak from 1H-NMR spectra at around 9–10 ppm. TLC analysis of a small-scale reaction of Neu5Ac aldolase and pyruvate over a period of days showed complete consumption of most starting material spots. Fur- ther, a small peak of appropriate mass was detected by MS (along with several other peaks), suggesting that these unsaturated and/or 2,3-dideoxy hexoses can indeed be accommodated in the Neu5Ac aldolase active site, but synthesis of a pure starting material for this reaction proved to be problematic.

Two possible alternate synthetic routes to 133 are presented in Scheme 4.11.

Following selective protection of Kdn, 194 the first route (A in Scheme 4.11) takes advantage of an activated leaving group equatorially at the anomeric position to drive a similar epoxidation to that proposed in mechanism C ofScheme3.13,with catalysis by a suitable base such as DBU. If this attempted epoxidation reaction is found to give elimination, despite the poor overlap of the leaving group σ&*withad- jacent proton σ& orbitals, an alternate route (B in Scheme 4.11) may also be possible by reducing the anomeric ketone 195 then selectively protecting the resultant alco-

hol to allow chemistry at the endocyclic oxygen. This oxygen could subsequently

be activated as a nucleofuge for direct nucleophilic epoxidation with the adjacent

hydroxyl on carbon 5. However, this is clearly a more involvedroutecomparedto

that in Scheme 4.9 on the previous page, and elimination rather than epoxidation

remains a risk.

157 4.5. Conclusions and future directions

OH OBn OBn SPh BaO, Ba(OH)2, SPh PMB-Br SPh BnBr NaH BnO O COOBn O COOBn O COOBn HO HO BnO HO PMBO HO BnO BnO HO BnO BnO

NBS (A) OBn OBn OBn BnO COOBn DDQ BnO COOBn DAST BnO OH O F O F O COOBn HO PMBO PMBO BnO BnO BnO BnO BnO BnO

NaBH4 (B)

OBn OBn BnO OTBDMS BnO OH TBDMSCl OH COOBn OH COOBn PMBO imidazole PMBO BnOBnO BnOBnO

Tf2O

DBU OBn DDQ OBn BnO OTBDMS BnO OTBDMS OTf COOBn OTf COOBn PMBO HO BnO BnO BnO BnO

DBU

BnO BnO

BnO TBAF BnO OH OTBDMS OBn OBn O O COOBn COOBn BnO BnO

DMP

BnO HO

BnO HO HOOC OH O H2, Pd/C O O OH OBn OH HO O O COOBn COOH O BnO HO 133 HO

Scheme 4.11: Alternate routes for synthesis of epoxide intermediate 133 from Kdn.

158 4.5. Conclusions and future directions

4.5 Conclusions and future directions

Unfortunately, none of the attempted means of testing the mechanisms proposed in

Section 3.5 on page 124 provided clear answers. SubstitutionofD113,theresidue proposed to either act as a nucleophile directly or as an acid/base to activate the carbon 2 hydroxyl group to act as a nucleophile resulted in a dramatic decrease in hydrolytic activity at the optimal pH of wild-type UGL, butaproportionofthis activity could be recovered in sufficiently acidic solution. This clearly shows a shift in the pKa of the remaining carboxylate and a stronger requirement for acid catalysis. The activity of the D113G mutant could not be rescued by addition of exogenous small molecules such as azide or formate.

Further circumstantial evidence against the simple hydration-initiated mechan- ism, presented in Scheme 1.9 on page 44, was found in the lack ofinhibitionby proline. This was intended to be a mimic of the transition state charge of this mech- anism, with the proline amino group occupying the same position as the substrate endocyclic oxygen, although other explanations are possible for this lack of inhib- ition. Surprisingly, strong inhibition was also not seen from a compound with a positively charged substituent on carbon 2 (Neu2en, 108), which should be located very close to the negatively charged residue D88/113 (but similar charge placement has previously been seen to provide only poor inhibition of glycosidases). Inhibition was however observed with the substrate mimic shikimic acid,withahigheraffinity than expected given that the hydroxyl group stereochemistries did not match that of the substrate. It is postulated that binding of this compoundinaslightlydifferent conformation from that adopted by the substrate allows similar interactions of this compound with the adjacent D88/113.

The inactivation seen with all predicted mechanism-based inactivators was con- cluded to be a non-specific effect, and thus not mechanism based, as clearly demon-

159 4.5. Conclusions and future directions strated by the control in which UGL is inactivated by 20 mM equatorial ΔGlcA fluoride substrate (70). What the nature of this non-specific effect is remains un- known. Undesired variation in pH by the substrate/inactivator carboxylate group was ruled out by quenching each compound stock to pH 6.5 using sodium hydroxide, and by monitoring the pH of the inactivation reactions. In allcaseswhereinac- tivation was observed it occurred at or above 20 mM, with no clear concentration dependence above this. This suggests that the same non-specific effect is responsible for the inactivation by all of the compounds tested, and not a contaminant, as it is unlikely that the same contaminant is present at the same concentration in all of these independent samples. Below this threshold no inactivation was observed for any compound, despite all of these compounds being seen tobindascompet- itive inhibitors. Thus, no evidence was found for any mechanism involving direct nucleophilic catalysis.

Finally, an attempt at synthesis of a stable epoxide intermediate, as required by mechanism C of Scheme 3.13 on page 126, was unsuccessful, as deprotection of the key α,β-unsaturated hexose enal was found to be problematic. While apreliminary test reaction of Neu5Ac aldolase with this α,β-unsaturated hexose enal in impure form suggests that this synthetic route may still be feasible, an alternative method of synthesising the α,β-unsaturated hexose enal is required if mechanism C is to be tested directly.

Discussion of these results, together with those of other chapters, will be taken up in Chapter 6.

160 Chapter 5

Heparanase substrate and inactivator testing

5.1 Chapter introduction

Mammalian degradation of heparan sulfate structures is catalysed by the enzyme heparanase (HPSE) from family GH71. 196 This enzyme is expressed as a single chain, pre-pro-heparanase, which is then truncated N-terminally to remove a sig- nalling peptide, glycosylated, and finally cleaved in two locations to give the mature heterodimer (8 and 50 kDa subunits, excising a 6 kDa fragment). Proheparanase is typically stored in an intracellular pool in lysosomes untilrequired,atwhichpoint it is mobilised to either the cell surface or secreted. 197 HPSE is a retaining endo-β- glucuronidase that cleaves only at specific sites within heparin and heparan sulfate,

typically producing products of around 5–10 kDa (10–20 sugars). By homology mod-

elling, the catalytic residues of HPSE have been proposed to be Glu225 and Glu343

for the acid/base and nucleophile, respectively,196 but experimental evidence of these

assignments is still lacking. Consistent with its internal cleavage mode, HPSE has

an extended active site with recognition subsites for several sugar units.198

The substrate specificity of HPSE has been the subject of much recent study.

While earlier work was based on selective digestion of heparan sulfate with other

enzymes to form defined sets of oligosaccharides for testing,199,200 and thus neces-

161 5.1. Chapter introduction sarily used heterogeneous substrates, more recent work has benefitted from advances in chemoenzymatic synthesis of very specific oligosaccharides as substrates for test- ing.201,202 Further insights have been provided by docking of potential substrate

structures into the active site of homology models of HPSE. 198,203 Together, these

works show that HPSE has a complicated substrate specificity,anditisdifficultto

arrive at a single model to explain this. While all works agreethatasulfatedtrisac-

charide domain within a larger oligosaccharide is important(predominantlyfound

in the mixed N/S regions of heparan sulfate, as outlined in Figure 1.5 on page 15),

the exact effects of specific sulfations remain difficult to predict, and results are oc-

casionally contradictory. Sulfation of the hydroxyl group on carbon 2 of a glucuronic

acid residue (but not iduronic acid) in either the -3 or +2 subsites is beneficial for

cleavage,199 although not essential,200 while sulfation of the hydroxyl group on car-

bon 6 of glucosamine in the +1 subsite is also beneficial for cleavage. Sulfation of

the hydroxyl group on carbon 3 of the glucosamine in the +1 subsite appears to

be beneficial in the absence of other secondary sulfations201 (those coloured blue in

Figure 5.1), but it is inhibitory in their presence. 200 Recalcitrant substrates, such as

those with an N -acetyl group on the glucosamine in the -2 subsite, can be activated

by multiple adjacent sulfates, especially of the hydroxyl group on carbon 6 of the

N -acetyl-glucosamine in the +1 and -2 subsites. 200,202 Asimplerecognitionsequence

for heparanase cleavage based on these works is presented in Figure 5.1. Whether or

not a substrate is cleaved by HPSE appears to be determined by the way in which

the substrate binds to the active site, with appropriate substrates binding with the

scissile bond close to the two catalytic residues and inappropriate substrates binding

in such a way that these residues are slightly further away, byaslittleas1–2Å.203

In its normal function, HPSE has roles in growth and remodelling, especially in

foetal tissue and sites of injury. However, it is also highly expressed and active in

162 5.1. Chapter introduction

-OOC OSO - O 3 O O HO O - HO - OSO3 - OOC OSO - O3SHN O 3 -3 O O HO O - -2 OH O3SO - - OOC O3SHN O -1 O HO O +1 - OSO3 +2

Figure 5.1: Sulfation in a substrate oligosaccharide as required for cleavage by HPSE. Sulfates in boxes are vital for substrate recognition, sulfates underlined are beneficial for substrate recognition, although not all are necessary together, while the sulfate coloured in red can be beneficial or inhibitory, depending on other sulfates present. The scissile bond is indicated by an arrow. Subsites relativetothiscleavagesiteare numbered. many cancers, where it is seen as an important factor in determining malignancy.

This arises because the action of heparanase allows the tissue surrounding a grow- ing tumour to expand, and also because some oligosaccharide products of HPSE are angiogenesis stimulants.204 HPSE also has a direct role in signal transduction,

mediated by a C-terminal domain separate from its catalytic site, increasing gene

transcription associated with tumour progression. 205 There is only one known mam- malian heparanase enzyme, and activity is typically low in healthy adult tissue, so inhibition of its activity presents an appealing target for cancer treatment.

One of the largest current problems for HPSE research is the development of a simple and sensitive assay system for the activity of this enzyme on a short time- scale (minutes) to allow testing of inhibitors and inactivators. While much progress has been made in this area in recent years, an assay comparabletotheuseofaryl glycosides for other glycosidases has yet to be achieved. Presented here is work to- wards designing and testing of a short oligosaccharide substrate with a fluorescent aglycone, and also a potential active-site trapping reagentintendedtoallowunam-

163 5.2. Synthesis of compounds biguous identification of the catalytic nucleophile residue, the development of the former being intended to facilitate testing of the latter as well as further mechanistic work on this enzyme.

5.2 Synthesis of compounds

Given the size and complexity of heparan sulfate oligosaccharides, much previous synthesis of substrates, inhibitors, or inactivators has involved very laborious chem- ical synthetic routes (see, for example, Chen et al. 206 and Hu et al. 207). In this

work, substrates and an inactivator for HPSE were synthesised by a collaborative

chemo-enzymatic scheme, as represented in Scheme 5.1, 208 allowing much easier ac-

cess to these compounds. The starting pseudo-disaccharidesmethylumbelliferylβ- d-glucopyranosiduronic acid (137,commerciallyavailable),trifluoromethylumbel-

liferyl β-d-glucopyranosiduronic acid (138,synthesisedfrom3 by Koenigs-Knorr glycosylation to 140 and deprotected by HCl in methanol then lithium hydroxide)

and 2,4-dinitrophenyl 2-deoxy-2-fluoro-β-d-glucopyranosiduronic acid (139,depro- tected from 76 by HCl in methanol then lithium hydroxide) were sent to a collab-

orator for extension to appropriately sulfated trisaccharides, as outlined below.

Briefly, the pseudodisaccharide is first extended by addition of a glucosamine

residue using the N -acetylglucosamine transferase KfiA from E. coli,withUDP-

GlcNTFA as glycosyl donor. If the desired product requires an N -acetyl group then

UDP-GlcNAc is used instead. The oligosaccharide chain is subsequently elongated

by addition of a glucuronic acid residue using the glucuronicacidtransferasehep-

arosan synthase 2 from Pasteurella multocida,withUDP-GlcAasglycosyldonor.

These two steps can be repeated to give longer chains. The completed oligosacchar-

ide chain can then undergo a series of modifications, depending on the desired final

sulfation pattern. Trifluoroacetamide-protected glucosamine can be deprotected by

164 5.2. Synthesis of compounds

HOOC O HO HO OR OH

UDP-GlcNTFA, α-(1,4)-GlcNAc MnCl2, pH 7.2 transferase (E. coli KfiA) OH O HO HOOC HO -OOC O TFAHN O HO (AcHN)O HO O O O HO OR OH OH 137 UDP-GlcA, β-(1,4)-GlcA MnCl2, pH 7.2 transferase (Pasteurella multocida heparosan synthase 2)

-OOC OH CF3 O HO O HO O HOOC OH - O HO OOC HO TFAHN O HO O O O (AcHN)O OH HO OR OH 138 MeOH/H2O/TEA 2:2:1, 37 °C, 16 hr

-OOC OH O HO O HO O O2N NO2 OH - HOOC HO OOC O H2N O HO (AcHN)O HO O HO OR F OH 139 PAPS N-sulfotransferase

-OOC OH O HO O HO O OH HO - - OOC O3SHN O (AcHN)O HO OR OH PAPS 6-O-sulfotransferase

-OOC OSO - O 3 HO O HO O OH HO - - OOC O3SHN O (AcHN)O HO OR OH

Scheme 5.1: Starting glucuronic acid pseudo-disaccharides(left)andgeneralscheme for chemoenzymatic synthesis of appropriately sulfated substrates and inactivators for HPSE (right). Positions of sulfation are indicated in red, and N-acetylated gluc- osamine nitrogen derivatives are indicated in parentheses."R"mustbeanarylgroup or N -acetyl-glucosamine analogue, to occupy the GlcNAc transferase -2 subsite, for efficient transfer in the first step.

165 5.3. Substrate testing triethylamine in methanol and water at 37 °C for 16 hours, followed by treatment with an N -sulfotransferase using 3!-phosphoadenosine 5!-phosphosulfate (PAPS) as a source of activated sulfate, to give N -sulfated glucosamine. Further sulfation, of the oxygen on carbon 6 of glucosamine, is possible with a 6-O-sulfotransferase, again using PAPS as a source of activated sulfate. Finally, epimerisation of glucuronic acid to iduronic acid and subsequent sulfation of the oxygen on carbon 2 of iduronic acid is possible using a C5-epimerase and 2-O-sulfotransferase, but was not car-

ried out for the compounds presented in this work. Overall, this pathway largely

mimics the natural synthesis of heparan sulfate (see Subsection 1.2.4), but addi-

tional control is available by variation of the glucosamine donor nitrogen derivative

and omission of specific steps where desired. To obtain disaccharide substrates, a

trisaccharide substrate was initially synthesised, with the additional sugar being re-

quired for the activity of the sulfotransferase enzymes, then the terminal glucuronic

acid residue was removed by treatment with bovine β-glucuronidase. Unfortunately, alargeamountofenzymeandlongtreatmentwasrequiredtoobtain satisfactory

cleavage of this recalcitrant substrate. Heparanase enzymeasasingleconstitutively

active chain (fusing the 50 and 8 kDa subunits), expressed by insect cells using a

baculovirus vector, was sourced from the same collaborators.202

5.3 Substrate testing

5.3.1 Confirmation of heparanase activity

In order to provide a positive control for the activity of HPSEfollowingshipment as frozen aliquots on dry ice, the assay system of Hammond et al. was used. 209

This employs a synthetic pentasaccharide heparan sulfate fragment capped as its methyl anomer (Fondaparinux, 142,tradedasArixtrabyGlaxoSmithKline,U.K.,

166 5.3. Substrate testing

O2N - O3S N N - + SO3 N N

I 141

- OSO3 O HO HO - - - OOC OSO3 O3SHN O O O HO - O OH O3SO - - OSO3 O3SHN O O O HO O - - HO OOC OSO3 - 142 O3SHN OMe

Scheme 5.2: Compounds for the HPSE reducing sugar assay, WST-1 (141)and Arixtra (142). for use as an anticoagulant) as substrate in a reducing sugar assay, as detected by the dye WST-1 (water soluble tetrazole, 141)showninScheme5.2.Whilethis pentasaccharide can be cleaved by heparanase, it is not an ideal substrate as it is shorter than the optimal case (an octasaccharide), and contains a 2-sulfated iduronic acid in the +2 subsite, as opposed to the optimal 2-sulfated glucuronic acid, and 3- sulfation on the N -sulfo-glucosamine in the +1 subsite, which is inhibitory inthe

presence of other sulfates (see Subsection 5.1). Overnight incubation of up to 500

µM Arixtra with HPSE resulted in complete hydrolysis. A time course for cleavage

of 100 µM substrate over a shorter time frame showed a slow linear rate, from which

-1 -1 a kcat of 0.31 s can be derived, which is lower than the value of 3.5 s reported by Hammond et al. 209 These results clearly demonstrate that the heparanase enzyme

received had a detectable but relatively low level of activity. This discrepancy may,

at least in part, arise from the differences in the enzyme preparation, as the enzyme

167 5.3. Substrate testing used by Hammond et al. is a native heterodimer rather than a single constitutively

active chain as used here.

5.3.2 Towards a fluorescent substrate

Based on the publication by Pearson et al. 210 showing activity of HPSE on the fluor-

escent and chromogenic substrates GlcNS-α-1,4-GlcA-β-DNP (17 nmol.hr-1.mg-1)or -MU (48 nmol.hr-1.mg-1), the first compounds to be designed and tested for improved

HPSE substrates were the trisaccharides GlcA-β-1,4-GlcNS6S-α-1,4-GlcA-β-TFMU

(143)andGlcA-β-1,4-GlcNAc6S-α-1,4-GlcA-β-TFMU (144), as shown in Table 5.1. The additional glucuronic acid subunit, sulfation of the hydroxyl on carbon 6 of

the glucosamine, and a more sensitive fluorophore were together anticipated to give

asubstratethatallowsrapidquantificationofHPSEactivity. Unfortunately, very

little hydrolysis of these compounds was detected when 100 µMsubstratewasdiges-

ted overnight by HPSE. Indeed, quantification of this overnight digestion was only

possible by virtue of the extremely sensitive fluorophore. Activity was at an insig-

nificant level when compared to Arixtra, with 5 orders of magnitude less substrate

cleaved, and also substantially less than that reported by Pearson et al. for their

disaccharides, although this may partially reflect the low level of activity seen for this

source of HPSE. Cleavage of these trisaccharides to disaccharides, by use of bovine

liver β-glucuronidase, gave GlcNS6S-α-1,4-GlcA-β-TFMU (145)andGlcNAc6S-α-

1,4-GlcA-β-TFMU (146). These shorter substrates were, surprisingly, found to be slightly better substrates for HPSE, particularly 146,butagainstillhydrolysedat

much lower rates than required for a convenient assay.

As Pearson et al. also reported an effect from the nature of the aglycone in

their disaccharide substrates, further trisaccharides were designed that employed a

methylumbelliferyl aglycone, as this was the best leaving group reported in that

168 5.3. Substrate testing

Table 5.1: Rates of hydrolysis for the commercial pentasaccharide Arixtra and hep- arin di- and tri-saccharide substrates with fluorescent leaving groups. Key features of substrates are emphasised in red and the intended scissilebondisindicatedbyan arrow. Reactions were performed in acetate buffer at pH 5.5 with 100 µM substrate.

Rate Substrate (nmol.hr-1.mg-1 HPSE)

- OSO3 O HO HO - - - OOC OSO3 O3SHN O O O HO - O OH O3SO - - OSO3 O3SHN O O O HO O - - HO OOC OSO3 - O3SHN 4 OMe (142) 1.3 × 10 -OOC OSO - O 3 HO O CF3 HO O OH HO - - OOC O3SHN O O HO O O O OH (143) 0.35 -OOC OSO - O 3 HO O CF3 HO O OH HO -OOC AcHN O O HO O O O OH (144) 0.28 - OSO3 O CF3 HO HO - - OOC O3SHN O O HO O O O OH (145) 0.58 - OSO3 O CF3 HO HO -OOC AcHN O O HO O O O OH (146) 3.3 -OOC OSO - O 3 HO O HO O OH HO - - OOC O3SHN O O HO O O O OH (147) 5.9 -OOC OH O HO O HO O OH HO - - OOC O3SHN O O HO O O O OH (148) 2.3

169 5.3. Substrate testing work. The trisaccharides GlcA-β-1,4-GlcNS6S-α-1,4-GlcA-β-MU (147)andGlcA-β-

1,4-GlcNS-α-1,4-GlcA-β-MU (148)showedslightlyimprovedactivity,butstillvastly less than than seen with Arixtra, itself a sub-optimal substrate. While the hydrolysis

of these MU trisaccharides is sufficient for detection of activity following overnight

incubation, again by virtue of the sensitive fluorophore released, the level of signal

is still nowhere near sufficient for an assay on a timescale of minutes as desired.

Repeating the assays for 147 and 148 under identical conditions to those reported

by Pearson et al. (5 mM substrate) gave only marginal increases over the rates

reported above for 100 µM substrate concentrations (6.4 and 3.5 nmol.hr-1.mg-1,

respectively).

Several experiments were performed in order to test whether the enzyme is

cleaving at a site other than that between the fluorophore and the oligosacchar-

ide. Overnight HPSE digestion of GlcA-β-1,4-GlcNS6S-α-1,4-GlcA-β-TFMU (143), the anticipated best substrate, at 100 µM was assayed with WST-1 (141), but no

signs of free reducing sugars were detected above a background control. To test for

modification of the aglycone, particularly of the ester in TFMU, the reaction of 144

with HPSE was monitored by 19F-NMR. No change was detected in the spectrum

after overnight incubation, as shown in Figure 5.2. Finally,inordertotestifthe

substrates are able to bind to HPSE, 143 and 144 were tested as inhibitors of HPSE

hydrolysis of Arixtra. Figure 5.3 shows the measurement of initial hydrolysis rates

for a single Arixtra concentration and a range of TFMU trisaccharide concentra-

tions. Inhibition was calculated from these to be at 120 and 690 µM, respectively.

An assumption of competitive inhibition was made, as would beexpectedtoarise

from binding of a substrate to the enzyme active site, but insufficient compound

was available to test this assumption. These Ki values are similar to the Km of the Arixtra pentasaccharide at 46 µM, and indicates that the lackofactivitywiththese

170 5.3. Substrate testing substrates is not a result of inability to bind at the active site.

Figure 5.2: Overnight reaction of 144 with HPSE, monitored by 19F-NMR

Together these results clearly demonstrate that, in order toobtainafluorescent

or chromogenic substrate with sufficiently high activity to allow convenient assay-

ing of HPSE on a minute timescale, much further optimisation of the substrate is

required. Given the requirement for sulfation of the glucosamine in the +1 subsite

for appropriate positioning of a substrate in the HPSE activesite(asdiscussedin

Subsection 5.1), the nature of the leaving groups used here appears to be inappro-

priate. Future efforts likely need to focus on design of an activated leaving group

with appropriately placed charges, while sulfation of the hydroxyl on carbon 2 of the

GlcA in the -3 subsite may also help activity. Although two charged aglycones were

used by Pearson et al.,theplacementofthesechargesdidnotappeartobematched

to that in an ideal HPSE substrate, and no hydrolysis of these compounds by HPSE

was detected. One potential substrate incorporating these changes is illustrated in

Scheme 5.3.

171 5.3. Substrate testing

Figure 5.3: Inhibition of HPSE by TFMU substrate trisaccharides ( : 143 and !: 144)showingKi values of 120 and 690 µM, respectively, determined from the intercept with 1/Vmax (dashed line). Substrate (142)wasat100µM.

172 5.4. Testing of a potential HPSE inactivator

MeOOC HO COOMe O AcO + COOMe AcO COOMe AcO MeOOC O O O Br Br AcO AcO O OAc HO Cl HO O O

+ COOMe COOMe HO COOMe COO- -OOC O O HO - - HO O OOC OSO3 OH O O HO O HO O - OSO - HO - COO 3 - OOC O O3SHN O COO- O HO O OH O

COO-

- O O O- O S O O O O- O- O RO O RO O RO O- HO O O = O O NH O S O -O O O -O

Scheme 5.3: A potential synthesis for a HPSE substrate with a charged aglycone (upper), and illustration of two possible charge placementsbythisaglyconetomimic the sulfation of optimal HPSE natural substrates (lower).

5.4 Testing of a potential HPSE inactivator

Apotential2-deoxy-2-fluoro-glucuronideinactivatorofHPSE was also designed, syn- thesised and tested. Although the hydrolysis rates of the di-andtri-saccharide substrates in the previous section were disappointingly low, an inactivator needs only to catalyse half a turnover, the glycosylation half-reaction, to give a trapped intermediate. The potential inactivator synthesised was a trisaccharide, GlcA-β-1,4-

GlcNAc6S-α-1,4-2FGlcA-β-DNP (149,Scheme5.4),withN -acetyl-glucosamine be- ing used in the -2 subsite despite it giving poorer hydrolysisastheconditionsforde-

protection of N -trifluoroacetyl-glucosamine were found to cleave offthe dinitrophenyl

173 5.4. Testing of a potential HPSE inactivator leaving group. Time-dependent inactivation of HPSE by 149 was monitored by reac-

tion of an aliquot from the inactivation reaction with Arixtra at 250 µM for 30 min at

each timepoint, with the results shown in Figure 5.4. The initial loss and subsequent

recovery of HPSE activity seen with inactivator present overthefirst4hourswas

deemed to be an artefact, as the activity in the control variesbyasimilarmagnitude.

The inactivator was also present in a large excess over enzyme, so reactivation is not

expected without substantial amounts of dinitrophenol being released, which would

give an associated observable colour change. Insufficient Arixtra remained to repeat

this experiment to control for this noise. While the assay wasmuchmorenoisy

than desired, these results are still able to show that no inactivation of HPSE by

149 is taking place on this timeframe. If, for the sake of comparison, the rate of

inactivation with this compound were assumed to be the same asthatoftherelated

substrate GlcA-β-1,4-GlcNAc6S-α-1,4-GlcA-β-TFMU (144), inactivation would be expected to have a half life of around 40 hours. Given that 2-deoxy-2-fluoro com-

pounds react more slowly than their parent substrates, it is not surprising that no

inactivation was detected over 20 hours. Any future design ofsuchinactivatorsfor

HPSE will likely be dependent on the successful design of small molecule substrates,

particularly the effect of charge in the leaving group.

-OOC OSO - O 3 HO O HO O O2N NO OH HO -OOC 2 AcHN O O HO O 149 F

Scheme 5.4: Structure of the potential 2-deoxy-2-fluoro inactivator of HPSE.

174 5.4. Testing of a potential HPSE inactivator

Figure 5.4: Attempted time-dependent inactivation of HPSE by 149 at 0( )and 1(!)mM.

175 5.5. Conclusions

5.5 Conclusions

Presented here was an attempt at design and synthesis of smallmoleculesubstrates and an inactivator for HPSE. While literature precedent fromPearsonet al. 210 had suggested that substantial acivity from such compounds may be possible, the relev- ance of the level of activity detected in that work may have been overestimated by selection of a less appropriate native substrate for comparison, and by comparison to rates reported in other work rather than a control carried out under the same as- say conditions. The Arixtra pentasaccharide substrate 209 was used here to confirm activity of the HPSE enzyme, and as a native-like substrate toprovideabaseline comparison for the activity seen with other substrates. Thisclearlyindicatedthat the activity of HPSE with the six potential fluorogenic small molecule substrates tested here, which were anticipated to be better substrates than those of Pearson et al.,wasnegligible.Similarly,noinactivationofHPSEwasdetected from a 2-

deoxy-2-fluoro reagent based on these. Progress in this area is likely dependent on

the design of a chromophore that can make suitable binding interactions with the

+1 subsite of HPSE, and one potential candidate based on the placement of sulfates

in optimal substrates was suggested here.

176 Chapter 6

Overall conclusions

Overall conclusions

In this thesis, three chapters were presented on the mechanism of unsaturated glucuronyl hydrolases. The first chapter presented evidenceforahydrationreac- tion that had previously been proposed on the basis of crystallographic evidence.87

Careful characterisation of the products of reaction in D2Oand10%methanolled to the conclusion that this hydration is indeed the reaction catalysed by UGL, with

supporting evidence from reaction of UGL with three compounds that are only ex-

pected to be turned over by the enzyme if reaction occurs through such a hydration

mechanism.

The subsequent chapter presented a series of experiments designed to probe the

steps of this reaction, and the transition states of these steps. Evidence for rearrange-

ment of the hydrated product outside of the enzyme active sitewassought,butnot

found, leading to the tentative conclusion that the enzyme actively catalyses this

step. The effect of activated leaving groups and heteroatoms on the rate of UGL

hydrolysis supported the hypothesis that the transition state of the rate-determining

step involved development of positive charge at carbon 5, with the unexpected ob-

servation that a 2-deoxy-2-fluoro substrate was almost completely inert to hydrolysis

by UGL. Subsequent measurement of kinetic isotope effects showed that two kinetic-

ally important steps must be present in the mechanism, from observation of distinct

177 Overall conclusions

effects on kcat and on kcat/Km.Theeffectsonkcat/Km are consistent with an initial irreversible protonation step, while those on kcat clearly show that the intermediate formed is not an oxocarbenium ion. Clear solvent kinetic isotope effects were ob- served on both parameters, showing a likely importance of solvent-derived proton transfer in each step. On the basis of these results, three alternate mechanisms were proposed for UGL.

The third chapter presented work towards testing each of these hypotheses, but clear, incontrovertible evidence was not obtained for any one. Mutagenesis of D113, the aspartate residue without a clear role in the simple hydration mechanism, to glycine followed by subsequent unsuccessful attempts at rescue with exogenous nuc- leophiles confirmed the importance of this residue for turnover, but provided no clues as to its role. Inhibitors intended to mimic the charge distribution of trans- ition states in some of the candidate mechanisms were tested,butnonewerefound to bind strongly to the enzyme. Neu2en, a further compound modified at carbon 2

(numbering based on ΔGlcA), was also found not to be turned over by UGL. To test the mechanisms that involved nucleophilic catalysis, two potential trapping reagents were tested but neither inactivated UGL in a concentration-dependent manner, and indeed nor were they turned over by the enzyme. A further potential trapping re- agent, intended to test the hypothesis that the hydrated product rearranges in the active site, also did not inactivate the enzyme, and neither was it converted, thereby providing no clear answers. Finally, synthesis of a small molecule intermediate pro- posed for mechanism C was attempted, but unsuccessful.

Despite the lack of conclusive evidence, mechanism C from Scheme 3.13 on page 126 best accounts for all of the experimental observations presented in this work. Scheme 6.1 shows potential transition states that illustrate this. In the first irreversible step, bond formation is advanced between carbon 4 and its second proton

178 Overall conclusions

D173 D173 - D173 O COOH O H OH OH H HOOC O O O HO O O O HO OR HO OR HO OR O O OH O O H O OH COOH O- D113 O- 22 D113 D113 H OH O- HOOC D173 HOOC D173 OH OH O- δ+ O δ+ O OR O OR H H O D D D Oδ+ D Oδ+ H H O O Oδ- O- D113 D113

Scheme 6.1: Conformation of transition states in mechanism CofScheme3.13to account for experimental observations. Dashed lines represent partially formed or broken bonds at the transition state. Deuteriums for which kinetic isotope effects were measured are both shown in the one structure for simplicity, but each substi- tution was made independently. and between carbon 1 and the carbon 2 hydroxyl oxygen. Bond breaking is largely complete in the double bond between carbons 4 and 5, between the hydroxyl on carbon 2 and its proton, and between carbon 1 and the endocyclic oxygen, but form- ation of the double bond between this same oxygen and carbon 5 is less developed.

This lag in the formation of the ketone double bond gives the substantial positive charge at carbon 5 and large change in hybridisation at carbon4requiredtoaccount for the small normal overall KIE on kcat/Km from deuterium at carbon 4, and may arise from the requirement for a change in hybridisation at the endocyclic oxygen atom to give appropriate orbital overlap with the adjacent carbon. The low bond order between carbon 1 and the endocyclic oxygen isolates thedeuteriumatcarbon

1fromthischarge,explainingitssmallkineticisotopeeffect on this first step, and also the lack of detectable effect from electron withdrawing groups in the LFER.

In the second step, this situation is largely reversed. Bond formation is early between the epoxide oxygen and its proton, which provides a driving force for attack

179 Overall conclusions of the endocyclic oxygen (ketone) at carbon one. The double bond from the endocyc- lic oxygen to carbon 5 is also broken early in the reaction coordinate, while attack of the water nucleophile at carbon 5 occurs later in the reactioncoordinate.Thisagain gives a substantial positive charge at carbon 5 and, with no accompanying change in hybridisation at carbon 4 in this step, leads to a large normal KIE on kcat from deuterium at carbon 4. The higher bond order between the endocyclic oxygen and carbon 1 in this transition state means that the anomeric substituent is more able to influence its stability through electron withdrawing effects on the positive charge.

Attack of the endocyclic oxygen (ketone) at carbon one proceeds through an SN2-like mechanism, and the associative nature of this transition state at carbon 1 leads to steric crowding and limitation of the out-of-plane bending modes of the hydrogen at this position. This leads to an inverse KIE, with the lower energy vibration of the carbon-deuterium bond in this mode resulting in less destabilisation from crowding at the transition state than a carbon-hydrogen bond. Kineticisotopeeffectsonacid- catalysed epoxide opening211–214 are similar to the effect from deuterium at carbon

1onkcat,lendingfurthercredibilitytothisinterpretation.Protonation of the ep- oxide likely provides the main driving force for this step, astheketoneatcarbon5

is otherwise a poor nucleophile, and this proton transfer accounts for the SKIE on

kcat.AhypotheticalenergyprofileforreactionbymechanismCcompared to the non-enzymatic acid-catalysed case is given in Figure 6.1.

It is worth noting that some precedent exists for this mechanism in non-enzymatic

reactions of glycosides, as shown in Scheme 6.2.215 In non-enzymatic acid-catalysed

hydrolysis of glycosides an initial protonation can occur oneithertheanomericoxy-

gen, giving a cyclic oxocarbenium ion intermediate stabilised by the endocyclic oxy-

gen, or on the endocyclic oxygen, giving a linear oxocarbenium ion stabilised by the

anomeric oxygen. Ring-opening mechanisms for glycoside hydrolysis have previously

180 Overall conclusions O OH OH O E+P HOOC E.P OR OR OH O COOH OH O HO J' E.J OH HO HOOC HO GlcA hydrolysis (in black), compared to the non-enzymatic Δ OR O O OR E.I O + OH COOH I Reaction coordinate HOOC HO HO E.S OR O OH COOH E+S

HO Energy Figure 6.1: Hypotheticalacid-catalysed energy reaction profile (in for red). UGL-catalysed

181 Overall conclusions

HO O SMe HO OH +

+ O SMe + OH S Me O SMe H -H+

HO OH HO OH HO OH OH OH OH

+

HO O SMe HO OH

OH OH OH O OR OH- O MeOH O OMe O HO OH -ROH HO HO OH OH OH OH

O O O O + O H O H2O HO O O O O OMe HO OMe HO OMe OBn OBn OBn

Scheme 6.2: Precedent for mechanism C of Scheme 3.13 from non-enzymatic reac- tions of glycosides.

182 Overall conclusions been suggested, and distinguishing between these two mechanisms was the topic of much work. Indeed some rearrangements of thioglycosides in aqueous acid do appear to proceed through such linear intermediates. Furthermore,anepoxideintermediate has been proposed to accelerate non-enzymatic decomposition under basic conditions of activated glycosides with a 1,2-trans arrangement, while non-enzymatic opening

of an epoxide at carbons 3 and 4 of a hexose is accelerated by participation of an

ester substituent at carbon 6.

Further evidence for this mechanism comes from the importance of the hydroxyl

group at carbon 2, shown by the dramatically low hydrolysis rate for the 2-deoxy-2-

fluoro glucuronide (78,seeSubsection3.3.2)andthelackofanyhydrolysisdetected for Neu2en (108,seeSubsection4.2).Thesecompoundsareunabletoformthe epoxide required to stabilise the positive charge in the transition state, and so their hydrolysis proceed at a much lower rate than for their oxygen-bearing analogues DNP

ΔGlcA (47)andKdn2en(40). The very low level of residual activity seen with 78 may arise by an alternate mechanism, as the acid catalyst and water nucleophile are still appropriately placed but substantially less stabilisation is available for the trans- ition state. One likely candidate for this mechanism is direct hydration through an oxocarbenium ion-like transition state. The chondroitin-derived 2!-sulfated natural substrate 16,whichishydrolysedbyUGLfromBacillus sp. GL1 but not from any of the other organisms for which data have been reported (see Section 2.4), clearly cannot proceed through the epoxide mechanism. However, the sulfate group on the

2!-hydroxyl may be able to fill the same stabilisation role with no need for acid-base catalysis as a result of the already low pKa of the sulfate group. This explanation could be tested by determining the effect of the D88N mutant in Bacillus sp. GL1

UGL on hydrolysis of this 2!-sulfated substrate, which would be expected to not show the same dramatic decrease in activity seen for other substrates with this mutant.

183 Overall conclusions

One potential advantage for the remote charge stabilisationproposedinthismech- anism is that the identical charges of the substrate and enzyme carboxylates do not need to be in close proximity, so charge repulsion does not need to be overcome.

Asimilarexplanationhasbeeninvokedtoexplaintheactionof a neutral tyrosine residue as a catalytic nucleophile in sialidases, where a carboxylate is normally found in other glycoside hydrolases (refer to Section 1.3).

Finally, work towards simple substrates and an inactivator of mammalian hep- aranase was presented in the preceding chapter. While this work was not successful in achieving the desired assay sensitivity, it indicated that optimisation of sulfation patterns without modification of the aglycone is likely a futile strategy. A potential aglycone that takes advantage of knowledge of optimal HPSE sulfation patterns was presented, and represents a new path towards the goal of studying this enzyme for its eventual use as a therapeutic target in cancer therapy.

184 Chapter 7

Materials and methods

7.1 Materials

Chemicals were purchased from Sigma-Aldrich unless otherwise stated, and used without further purification. Natural substrates 11, 12, 13, 14, 15,and16 as well as asmallsampleofKdn2en(40)werepurchasedfromCarbosynth,UK(http://www. carbosynth.com/). TLC was performed on pre-coated 60F254 silica plates (Merck, Germany), with visualisation by UV light followed by charring with 10% ammonium molybdate in 2 M H2SO4.Flashcolumnchromatographywasperformedusing230– 400 mesh silica gel and an in-house compressed air system. Foranhydrousreactions solvents were freshly distilled (CH2Cl2 over CaH2,MeOHoverMg)andglassware dried in an oven. NMR spectra were recorded on Bruker Avance 300and400(with either an inverse or a direct probe) spectrometers at 300 and 400 MHz, respectively.

Chemical shifts are reported in δ scale in parts per million from tetramethylsilane (TMS) with internal reference to solvent for 1Hand13Cshifts,while19Fshiftsare

reported relative to an external standard of CFCl3 at 0 ppm. Low resolution mass spectra were recorded on a Waters ZQ equipped with ESCI ion source and Waters

2695 HPLC for sample delivery, while high resolution mass spectra were submitted

to the University of British Columbia mass spectrometry facility for analysis on a

Waters/Micromass LCT with electrospray ionisation and timeofflightdetectionin

either positive or negative mode. HPLC was carried out using an Agilent eclipse

185 7.2. Synthesis

XD-C18 column (5 µm pore size, 9.4 x 250 mm) on a Waters 600 at 4 mL.min-1 with aWaters2996photodiodearraydetector.Kineticmeasurements were performed in matched reduced-volume quartz cuvettes using a Varian Cary 4000 spectrophoto- meter with automatic cell changer and Peltier temperature controller, at 37 °C unless otherwise specified, and reaction mixtures were allowed to pre-incubate for 5 min before adding enzyme to start the reaction. All non-linear regression was performed using GraFit 5.0 (Erithacus software limited; www.erithacus.com/grafit).

7.2 Synthesis

7.2.1 General methods

General method for Koenigs-Knorr glycosylation.

The acceptor alcohol (1.3 eq.) was dried over acetonitrile where necessary, then dissolved along with the relevant glycosyl bromide in dry acetonitrile (to give ap- proximately 0.1 M of sugar). Ag2O(2-3eq.)wasadded,andthereactionmixture stirred vigorously overnight at ambient temperature in the dark. The reaction mix- ture was filtered through a plug of Celite and concentrated before being dissolved in ethyl acetate and washed extensively with sat. NaHCO3,waterandbrine. The organic phase was dried over MgSO4 then the solvent evaporated.

General method for DBU-catalysed elimination.

Protected aryl glucuronide was dissolved in dichloromethane (to give approximately

0.1 M of sugar), DBU (1.3 eq.) was added slowly through a septum and the reaction mixture stirred at room temperature overnight. Where found to be necessary, 4 Å molecular sieves were used and the solution flushed with argonornitrogentodry before addition of reagent. Once finished, the reaction mixture was concentrated and

186 7.2. Synthesis

filtered through a plug of silica (washing with 2:1 ethyl acetate/petroleum ether).

General method for ester saponification.

The protected final compound was dissolved in acetone to approximately 0.1 M, cooled to 0 °C, and an equal volume of NaOH added (1 M). The reaction mixture was stirred 5 min then quenched with a slight excess of HCl (1 M).

General method for Zemplén deprotection.

The acetyl- or benzoyl-protected compound was dissolved in 1:1 dichloromethane/- methanol to approximately 0.1 M, cooled to 0 °C, and either sodium methoxide in methanol (stock at 5.4 M) or a small piece of sodium metal addedtogiveafinal concentration of between 5 and 50 mM. Deacetylation was monitored by TLC and, upon completion, the reaction was quenched with Sephadex ionexchangeresin(H+ form) then filtered.

General method for acidic trans-esterification.

The acetyl- or benzoyl-protected compound was dissolved in 1:1 dichloromethane/- methanol to approximately 0.1 M, cooled to 0 °C, and acetyl chloride added (ap- proximately 1-5% v/v). Longer reactions were transferred toacoldcabinetat4

°C. Deacetylation was monitored by TLC and, upon completion,thesolventwas evaporated in vacuo for an extended period to remove HCl traces.

General method for hydrolysis by aqueous lithium hydroxide.

The methyl ester-protected compound was dissolved in tetrahydrofuran and water

(1.75:1, to approximately 0.1 M), cooled to 0 °C, and 2 eq. of 1 Mlithiumhydrox- ide added. The reaction was allowed to proceed for 5 min beforequenchingwith

187 7.2. Synthesis

Sephadex ion exchange resin (H+ form) then filtered.

7.2.2 Development of chromogenic substrates

Methyl (1,2,3,4-tetra-O-acetyl-α/β-d-glucopyranosid)uronate (2) MeOOC O AcO AcO OAc OAc

Glucuronic acid gamma lactone (9.936 g, 56.4 mmol) was suspended in MeOH

(500 mL), a sodium methoxide solution was added (1 mL at 5.4 M) at 0 °C, and the reaction mixture stirred until TLC indicated completion. The methoxide was then quenched with 500 µL acetic acid, and the solvents evaporated in vacuo. The intermediate was subsequently dissolved in Ac2O(100mL),cooledto0°C,perchloric acid (0.5 mL) added in small portions, and the reaction mixture allowed to warm to ambient temperature with stirring overnight. Upon completion, the reaction was quenched with ice water, extracted with 1 L ethyl acetate and the organic phase then washed extensively with water and saturated sodium bicarbonate, washed once with brine and dried with magnesium sulfate. The solvent was then evaporated to a syrup, which was allowed to stand at 4 °C until a slurry formed, the solid collected and washed with cold methanol, and the process repeated several times

(with diminishing returns) to give a white powder (12.32 g, 58%,α/β ratio 1.73:1).

1 HNMR(300MHz;CDCl3) δ 6.36 (d, J1,2 =3.6Hz,1H,H-1α), 5.74 (d, J1,2 =

7.7 Hz, 1H, H-1β), 5.49 (t, J2,3=J3,4 =9.9Hz,1H,H-3β), 5.32-5.06 (m, 3H, H-2,3,4),

4.38 (d, J4,5 =10.2Hz,1H,H-5α), 4.16 (d, J4,5 =9.3Hz,1H,H-5β), 3.72 (s, 3H, OMeα),3.71(s,3H,OMeβ), 2.16 (s, 3H, OAc), 2.09 (s, 3H, OAc), 2.01 (s, 6H, OAc),

2.00 (s, 4H, OAc), 1.98 (s, 3H, OAc); MS : Calcd. for C15H20NaO11:399.1;found: 399.3.

188 7.2. Synthesis

Methyl (2,3,4-tri-O-acetyl-α-d-glucopyranosyl bromide)uronate (3)

MeOOC O AcO AcO AcO Br

Globally protected glucuronic acid (2,1.18g,3.13mmol)wasdissolvedindichloro-

methane (4.5 mL) then acetic anhydride (0.5 mL) and 33% HBr in acetic acid (20

mL) were added at 0 °C. This was stirred at 4 °C until the reaction was complete as

determined by TLC (2:1 petroleum ether/ethyl acetate). The reaction was quenched

in ice/water, the aqueous layer extracted thrice with dichloromethane and this pooled

organic phase then extracted quickly with cold water, cold sat. NaHCO3 twice and

brine. After drying over MgSO4,concentratingandco-evaporatingfromtoluenethe colourless syrup (1.20 g, 3.02 mmol, 97%) was used without further purification.

Asampleforanalysiswaspurifiedbyflashcolumnchromatography (3:1 petroleum

1 ether/ethyl acetate), yielding a white powder. H-NMR (300 MHz; CDCl3): δ 6.67

(d, J1,2 =4.0Hz,1H,H-1),5.64(t,J2,3 =J3,4 =9.8Hz,1H,H-3),5.27(dd,J4,5 = 10.3 Hz, 1H, H-4), 4.88 (dd, 1H, H-2), 4.61 (d, 1H, H-5), 3.79 (s, 3H, OMe), 2.13 (s,

13 3H, OAc), 2.09 (s, 3H, OAc), 2.08 (s, 3H, OAc) ppm. C-NMR (75 MHz; CDCl3): δ 169.63 (OAc), 169.43 (OAc), 166.6 (C-6), 85.5 (C-1), 72.1 (C-3), 70.3 (C-2), 69.3 (C-4), 68.5 (C-5), 53.1 (OMe), 20.60 (2xOAc), 20.45 (OAc) ppm. MS : Calcd. for

C13H17BrNaO9:419.0/421.0;found:419.2/421.2

Methyl (4-nitrophenyl 2,3,4-tri-O-acetyl-β-d-glucopyranosid)uronate (4) MeOOC O AcO AcO O OAc NO2

4-Nitrophenol (1.14 g, 7.98 mmol) was reacted with globally protected glucuronyl bromide (3,2.51g,6.32mmol)bythegeneralmethodforKoenigs-Knorrglyc-

189 7.2. Synthesis osylation (page 186). Purification by flash column chromatography (1:1 petroleum ether/ethyl acetate) yielded a white powder (1.54 g, 3.38 mmol, 54%). 1H-NMR ! ! (300 MHz; DMSO-d6): δ 8.24 (d, J2!,3! =J5!,6! =9.3Hz,2H,H-2,6 ), 7.22 (d, 2H, ! ! H-3 ,5 ), 5.88 (d, J1,2 =7.8Hz,1H,H-1),5.48(t,J2,3 =J3,4 =9.6Hz,1H,H-3),5.16

(dd, 1H, H-2), 5.09 (br. t, J4,5 =9.9Hz,1H,H-4),4.76(d,1H,H-5),3.62(s,3H, OMe), 2.01 (s, 3H, OAc), 2.01 (s, 3H, OAc), 2.00 (s, 3H, OAc) ppm. MS : Calcd.

for C19H21NNaO12:478.1;found:478.2

Methyl (4-nitrophenyl 2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4- enopyranosid)uronate (5)

COOMe O AcO O OAc NO2

Globally protected 4-nitrophenyl β-d-glucuronide (4,0.521g,1.14mmol)was subjected to the general method for DBU-catalysed elimination (page 186). Puri-

fication by flash column chromatography (2:1 petroleum ether/ethyl acetate) gave a

1 white powder (0.416 g, 1.05 mmol, 92%). H-NMR (300 MHz; CDCl3): δ 8.23 (d, ! ! ! ! J2!,3! =J5!,6! =9.2Hz,2H,H-2,6 ), 7.21 (d, 2H, H-3 ,5 ), 6.34 (dd, J3,4 =4.5,J2,4

=1.6Hz,1H,H-4),5.96(dd,J1,2 =2.5,J1,3 =0.9Hz,1H,H-1),5.32-5.29(m,2H, H-2,3), 3.81 (s, 3H, OMe), 2.16 (s, 3H, OAc), 2.15 (s, 3H, OAc) ppm. MS : Calcd.

for C17H17NNaO10:418.1;found:418.2

4-Nitrophenyl 4-deoxy-α-l-threo-hex-4-enopyranosiduronic acid (6) COOH O HO O OH NO2

Globally protected 4-nitrophenyl unsaturated β-d-glucuronide (5,60mg,0.152

190 7.2. Synthesis mmol) was subjected to the general method for Zemplén deprotection (page 187) followed by the general method for hydrolysis by lithium hydroxide (page 187). The reaction was quenched with Sephadex ion exchange resin (H+ form) then filtered.

Purification by flash column chromatography (7:2:1 ethyl acetate/methanol/water) then 5 g C-18 Sep-pak, washed with water, 40% acetonitrile in water and 100% acetonitrile, gave a white powder following lyophilisation(36.1mg,0.122mmol,

1 ! ! 80%). H-NMR (D2O, 400 MHz): δ 8.09 (d, J2!,3! =J5!,6! =9.2Hz,2H,H-2,6 ), ! ! 7.17 (d, 2H, H-3 ,5 ), 6.22 (d, J3,4 =4.1Hz,1H,H-4),5.84(d,J1,2 =4.8Hz,1H,

13 H-1), 4.29 (br. t, J2,3 =4.4Hz,1H,H-3),4.07(br.t,1H,H-2)ppm. C-NMR

(D2O, 101 MHz): δ 176.8 (C-6), 165.1 (C-5), 161.1 (Ar), 142.7 (Ar), 126.1 (Ar), 116.9 (Ar), 112.5 (C-4), 97.3 (C-1), 69.3 (C-3), 65.7 (C-2) ppm. HRMS : Calcd. for

C12H10NO8:296.0406;found:296.0411

Methyl (2,3,4-tri-O-acetyl-α-d-glucopyran)uronate (7)

MeOOC O AcO AcO AcO OH

Globally protected glucuronic acid (2,1.902g,5.05mmol)wasdissolvedin1:1 dichloromethane/methanol (90 mL) and hydrazine acetate added (0.719 g, 1.6 eq.), and the reaction mixture then stirred at 0 °C followed by ambient temperature for 2 hours each. The solvent was subsequently evaporated in vacuo,thereactionmixture dissolved in ethyl acetate and the product washed with water,1MHClandbrine, then dried over MgSO4.Theproductwaspurifiedbyflashcolumnchromatography (3:2 to 1:1 petroleum ether/ethyl acetate) to give a white foam (1.09 g, 3.26 mmol,

1 65%). H-NMR (300 MHz; CDCl3) δ 5.50 (t, J2,3 =J3,4 =9.4Hz,1H,H-3),5.46(d,

J1,2 =3.0Hz,1H,H-1),5.09(t,J4,5 =10.1Hz,1H,H-4),4.82(dd,1H,H-2),4.52 (d, 1H, H-5), 3.66 (s, 3H, OMe), 2.01 (s, 3H, OAc), 1.96 (s, 3H, OAc), 1.95 (s, 3H,

191 7.2. Synthesis

13 OAc) ppm. C-NMR (75 MHz; CDCl3) δ 170.2 (OAc), 170.0 (OAc), 169.7 (OAc), 168.5 (C-6), 89.9 (C-1), 70.6 (C-3), 69.4 (C-2), 69.0 (C-4), 67.7 (C-5), 52.7 (OMe),

20.5 (2xOAc), 20.3 (OAc) ppm. MS : Calcd. for C13H18 NaO10:357.1;found:357.3

Methyl (phenyl 2,3,4-tri-O-acetyl-β-d-glucopyranosid)uronate (8)

MeOOC O AcO AcO O OAc

Globally protected glucuronyl hemiacetal (7,1.82g,5.45mmol)wasdissolvedin

dichloromethane (30 mL), 4 Å molecular sieves were added and the solution flushed

with argon, then cooled in a dry ice/acetone bath. Trichloroacetonitrile (10 mL,

10 eq.) then DBU (0.5 mL, 0.3 eq.) were added and the reaction mixture stirred

for 30 min at -78 °C. The reaction mixture was concentrated andfilteredthrough

aplugofsilicagel(elutingwith2:1petroleumether/ethylacetate). The resultant

crude trichloroacetimidate intermediate and phenol (0.66 g, 1.3 eq.) were dissolved

in dichloromethane (30 mL) and dried over 4 Å molecular sievesunderargon,then

cooled in a dry ice/acetone bath. Boron trifluoride diethyl etherate (0.22 mL, 0.3

eq.) was added and the reaction mixture stirred 30 min at -78 °Cbeforeallowing

to warm to ambient temperature while stirring overnight. Thereactionmixturewas

diluted with an equal volume of dichloromethane and washed with sat. NaHCO3

thrice then brine. The organic phase was dried over MgSO4 then purified by flash column chromatography (1:1 hexanes/ethyl acetate) to give awhitepowder(0.68g,

1 1.65 mmol, 30%). H-NMR (300 MHz; DMSO-d6): δ 7.33 (t, J2!,3! =J3!,4! =J4!,5! ! ! ! ! ! =J5!,6! =7.7Hz,2H,H-3,5 ), 7.06 (t, 1H, H-4 ), 6.99 (d, 2H, H-2 ,6 ), 5.65 (d, J1,2

=7.9Hz,1H,H-1),5.47(t,J2,3 =J3,4 =9.6Hz,1H,H-3),5.12-5.03(m,2H,H-2,4),

4.70 (d, J4,5 =9.9Hz,1H,H-5),3.63(s,3H,OMe),2.01(s,3H,OAc),2.00(s,3H,

OAc), 1.99 (s, 3H, OAc) ppm. MS : Calcd. for C19H22NaO10:433.1;found:433.1

192 7.2. Synthesis

Methyl (phenyl 2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4- enopyranosid)uronate (9)

COOMe O AcO O OAc

Globally protected phenyl β-d-glucuronide (8,678mg,1.64mmol)wassubjected to the general method for DBU-catalysed elimination (page 186). Purification by

flash column chromatography (3:2 hexanes/ethyl acetate) gave a white powder (295

1 ! ! mg, 0.84 mmol, 51%). H-NMR (300 MHz; CDCl3): δ 7.33-7.27 (m, 2H, H-3 ,5 ), ! ! ! 7.11-7.03 (m, 3H, H-2 ,4 ,6 ), 6.29 (d, J3,4 =5.9Hz,1H,H-4),5.82(dd,J1,2 =2.5,

J1,3 =1.3Hz,1H,H-1),5.31-5.29(m,2H,H-2,3),3.79(s,3H,OMe),2.13(s,3H,

13 OAc), 2.10 (s, 3H, OAc) ppm. C-NMR (75 MHz; CD3OD): δ 170.4 (OAc), 169.8 (OAc), 162.1 (C-6), 156.4 (C-5), 142.3 (Ar), 129.5 (Ar), 123.2 (Ar), 116.9 (Ar), 107.6

(C-4), 95.2 (C-1), 68.4 (C-3), 65.0 (C-2), 51.9 (OMe), 19.6 (OAc), 19.5 (OAc) ppm.

Phenyl 4-deoxy-α-l-threo-hex-4-enopyranosiduronic acid (10) COOH O HO O OH

Globally protected phenyl unsaturated β-d-glucuronide (9,84mg,0.24mmol) was deprotected by the general method for ester saponification (page 187). Puri-

fication was by 5 g C-18 Sep-pak, washed with water, 10% acetonitrile in water,

40% acetonitrile in water and 100% acetonitrile. All fractions determined by TLC

(3:3:1:1 toluene/ethyl acetate/methanol/acetic acid) to contain pure product were

pooled and lyophilised, while those containing impure product were pooled, lyophil-

ised and purified again to give a white powder (46 mg, 0.19 mmol,77%). 1H-NMR

193 7.2. Synthesis

(400 MHz; CD3OD) δ ! ! ! ! 7.29 (t, J2!,3! =J3!,4! =J4!,5! =J5!,6! =7.9Hz,2H,H-3,5 ), 7.08 (d, 2H, H-2 ,6 ), ! 7.05 (t, 1H, H-4 ), 5.81 (d, J3,4 =3.7Hz,H-4),5.61(d,J1,2 =5.2Hz,H-1),4.18(br.

13 t, J2,3 =4.5Hz,H-3),3.93(br.t,H-2)ppm. C-NMR (101 MHz; CD3OD) δ 157.3 (C-6), 129.3 (Ar), 122.8 (Ar), 117.2 (Ar), 117.1 (Ar), 112.4 (C-4), 99.5 (C-1), 70.8

(C-3), 67.0 (C-2) ppm (carbon 5 signal too weak to detect, but seen in protected form). HRMS : Calcd. for C12H11O6:251.0550;found:251.0556

7.2.3 Standards for UGL reaction in 10% methanol

Methyl (phenyl 4-deoxy-5-C-methoxy-β-d-glucopyranosid)uronate (27)

MeOOC O HO O OMeOH

Globally protected phenyl unsaturated β-d-glucuronide (9,10mg)wasdissolved in dry methanol (1 mL) under argon and acetyl chloride added (50 µL). The reaction

mixture was left at room temperature for 8 days then quenched with saturated

NaHCO3,andaprecipitatewasseentoform.Thiswasredissolvedbyaddition of glacial acetic acid, then the organic solvents evaporated and the product purified by HPLC, eluting with H2O(5min),5%acetonitrile(lineargradientover40min), 20% acetonitrile (linear gradient over 20 min) and finally 100% acetonitrile (linear

gradient over 10 min). Product fractions were identified by UV-vis absorbance on-

line and confirmed by TLC (3:3:1:1 toluene/ethyl acetate/methanol/acetic acid).

The title compound was seen to elute with a retention time of approximately 74

1 min. H-NMR (400 MHz; CD3OD): δ 7.31 (t, J2!,3! =J3!,4! =J4!,5! =J5!,6! =7.7 ! ! ! ! ! Hz, 2H, H-3 ,5 ), 7.18 (d, 2H, H-3 ,5 ), 7.06 (t, 1H, H-4 ), 5.05 (d, J1,2 =7.8Hz,

1H, H-1), 3.94 (ddd, J3,4ax =11.5,J2,3 =9.3,J3,4eq =5.1Hz,1H,H-3),3.79(s,

194 7.2. Synthesis

3H, OMe), 3.45 (dd, 1H, H-2), 3.25 (s, 3H, OMe), 2.32 (dd, J4ax,4eq =13.1Hz,1H,

H-4eq), 1.75 (dd, 1H, H-4ax)ppm.MS:Calcd.forC14H18NaO7:321.1;found:321.2

Phenyl 4-deoxy-5-C-methoxy-β-d-glucopyranosiduronic acid (28)

HOOC O HO O OMeOH

Methyl (phenyl 4-deoxy-5-C-methoxy-β-D-glucopyranosid)uronate (27)wasde- protected by the general method for ester saponification (page 187). The product was

purified by 5 g C-18 Sep-pak washed with H2O, 40% acetonitrile in water and 100% acetonitrile. The product was identified by TLC (3:2:2 1-butanol/acetic acid/water)

to be in the flowthrough and water washes, so these were pooled and lyophilised

before dissolving in ethanol to desalt, then filtering and removing the solvent to give

1 awhitepowder(4mg,14µmol,49%overtwosteps). H-NMR (600 MHz; D2O): δ ! ! ! ! 7.29 (t, J2!,3! =J3!,4! =J4!,5! =J5!,6! =7.9Hz,2H,H-3,5 ), 7.11 (d, 2H, H-3 ,5 ), ! 7.05 (t, 1H, H-4 ), 5.06 (d, J1,2 =8.0Hz,1H,H-1),3.90(ddd,J3,4ax =11.8,J2,3

=9.4,J3,4eq =5.1Hz,1H,H-3),3.44(dd,Hz,1H,H-2),3.07(s,3H,OMe),2.22

(dd, J4ax,4eq =13.1Hz,1H,H-4eq), 1.62 (dd, 1H, H-4ax)ppm.HRMS:Calcd.for

C13H15O7:283.0818;found:283.0813

Methyl (methyl 2,3,4-tri-O-acetyl-β-d-glucopyranosid)uronate (29)

MeOOC O AcO AcO OMe OAc

Globally protected glucuronyl bromide(3,0.564g,1.42mmol)wasdissolvedin

acetone (12 mL) and methanol (5 mL), Ag2CO3 (0.907 g, 2.3 eq.) was added, and the reaction mixture stirred overnight at room temperature in the dark. The solid was then removed by filtration through Celite, the solvents evaporated in vacuo,and

195 7.2. Synthesis the product purified by flash column chromatography (3:2 petroleum ether/ethyl acetate) to yield a colourless syrup (277 mg, 0.795 mmol, 56%). 1H-NMR (300

MHz, CDCl3): δ 5.16 (t, J2,3 =J3,4 =9.3Hz,1H,H-3),5.10(t,1H,H-4),4.89(dd,

J1,2 =7.8Hz,1H,H-2),4.41(d,1H,H-1),3.98(d,J4,5 =9.3Hz,1H,H-5),3.66 (s, 3H, COOMe), 3.42 (s, 3H, OMe), 1.95 (s, 3H, OAc), 1.92 (s, 3H, OAc), 1.91 (s,

13 3H, OAc) ppm. C-NMR (75 MHz, CDCl3): δ 169.9 (OAc), 169.26 (OAc), 169.18 (OAc), 167.2 (C-6), 101.5 (C-1), 72.4 (C-3), 72.0 (C-2), 71.0(C-4),69.4(C-5),57.2

(OMe), 52.8 (OMe), 20.53 (OAc), 20.45 (OAc), 20.36 (OAc) ppm.

Methyl (methyl 2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4-enopyranosid) uronate (30)

COOMe O AcO OMe OAc

Globally protected methyl glucuronide (29,277mg,0.80mmol)wassubjected

to the general method for DBU-catalysed elimination (page 186). Purification by

flash column chromatography (5:2 petroleum ether/ethyl acetate) gave a pale yellow

1 syrup (179 mg, 0.62 mmol, 78%). H-NMR (300 MHz, CDCl3): δ 6.09 (dd, J3,4 =

4.5, J2,4 =1.1Hz,1H,H-4),5.07(dd,J1,2 =4.3,J1,3 =1.9Hz,1H,H-1),5.01(dd, 1H, H-3), 4.97-4.95 (dd, 1H, H-2), 3.72 (s, 3H, OMe), 3.40 (s, 3H, COOMe), 1.98

(s, 3H, OAc), 1.97 (s, 3H, OAc) ppm. MS : Calcd. for C14H20NaO10:371.1;found: 371.3

196 7.2. Synthesis

Methyl 4-deoxy-α-l-threo-hex-4-enopyranosiduronic acid (31)

COOH O HO OMe OH

Globally protected methyl glucuronate (30,20mg,70µmol)wasdeprotectedas

per the general method for ester saponification (page 187). Purification was by 5 g

C-18 Sep-pak, washed with H2Oin5mLfractions.Allfractionscontainingproduct as determined by TLC (3:3:1:1 toluene/ethyl acetate/methanol/acetic acid) were

pooled and lyophilised to give a white powder (4.5 mg, 24 µmol,36%). 1H-NMR

(400 MHz, D2O): δ 6.05 (d, J3,4 =4.0Hz,1H,H-4),4.99(d,J1,2 =4.8Hz,1H,H-1),

4.10 (br. t, J2,3 =4.3Hz,1H,H-3),3.76(br.t,1H,H-2),3.44(s,3H,OMe)ppm.

MS : Calcd. for C7H9O6:189.0;found:189.3

7.2.4 Unusual substrates for UGL

3-Deoxy-d-glycero-d-galacto-2-nonulopyranosonic acid( 32) OH OH

O COOH HO HO HO HO

Pyruvic acid (2.55 g, 28.95 mmol, adjusted to pH 7 with 1 M NaOH)andD- mannose (10.161 g, 56.4 mmol) were dissolved in distilled water (200 mL), Neuraminic acid aldolase (260 mg) was added, and the reaction allowed to proceed at ambient temperature until TLC indicated consumption of the pyruvic acid (3:2:2 ethyl acet- ate/methanol/water). The product was then purified by Dowex 1XB ion exchange resin (pre-equilibrated with 6 M formic acid then water) eluting with water then 1

Mformicacid.Fractionscontainingproductwerepooledandthe solvent evapor- ated under reduced pressure until the volume was sufficiently low for lyophilisation,

197 7.2. Synthesis

yielding a white foam (3.75 g, 14.0 mmol, 48 %). MS : Calcd. for C9H15O9:267.2; found: 267.3

Methyl 2,4,5,7,8,9-hexa-O-acetyl-3-deoxy-d-glycero-d-galacto-2- nonulopyranosonate (33) OAc OAc

O COOMe AcOAcO AcOAcO

3-Deoxy-d-glycero-d-galacto-2-nonulopyranosonic acid (32,3.75g,14.0mmol) was dissolved in methanol (100 mL) and trifluoroacetic acid (0.5 mL), and stirred overnight at ambient temperature. The solvents were then evaporated in vacuo,and acetic anhydride (15 mL) and pyridine (75 mL) added and again stirred overnight at ambient temperature. The solvents were again evaporated in vacuo and the crude product dissolved in ethyl acetate before washing with water, saturated NaHCO3,

1NHClandbrinebeforedryingoverMgSO4.Filteringthroughaplugofsilica (eluting with 1:1 ethyl acetate/petrol) gave a pale yellow syrup of sufficient purity

1 for the subsequent reaction (4.47 g, 8.36 mmol, 60 %). H-NMR (300 MHz; CDCl3):

δ 5.39 (dd, J7,8 =6.2,J6,7 =2.3Hz,1H,H-7),5.25(ddd,J3ax,4 =11.6,J4,5 =10.0,

J3eq,4 =5.2Hz,1H,H-4),5.14(td,J8,9b =6.2,J8,9a =2.6Hz,1H,H-8),4.97(t,J5,6

=10.0Hz,1H,H-5),4.43(dd,J9a,9b =12.5Hz,1H,H-9a), 4.18 (dd, Hz, 1H, H-6),

4.11 (dd, Hz, 1H, H-9b), 3.78 (s, 3H, OMe), 2.62 (dd, J3ax,3eq =13.6Hz,1H,H-3eq),

2.15 (s, 3H, OAc), 2.11 (s, 3H, OAc), 2.06 (s, 3H, OAc), 2.06 (dd, 1H, H-3ax), 2.03 (s,

3H, OAc), 2.01 (s, 3H, OAc), 2.00 (s, 3H, OAc) ppm. MS : Calcd. for C22H30NaO15: 557.15; found: 557.3

198 7.2. Synthesis

Methyl 4,5,7,8,9-penta-O-acetyl-2,3-dideoxy-d-glycero-d-galacto-non-2-

enopyranosonate (35)

OAc COOMe O AcOAcO AcOAcO and

Methyl 4,5,7,8,9-penta-O-acetyl-2,3-dideoxy-d-glycero-d-talo-non-2-

enopyranosonate (34)

OAc OAc COOMe O AcOAcO AcO

To a solution of Methyl 2,4,5,7,8,9-hexa-O-acetyl-3-deoxy-d-glycero-d-galacto-2-non- ulopyranosonate (33,4.47g,8.36mmol)inethylacetate(40mL)at0°Cundernitro- gen was added Trimethylsilyl trifluoromethanesulfonate (3.36 mL, 2.2 eq.) dropwise, and allowed to react for 5 hours while warming to ambient temperature. The reac- tion was subsequently quenched with triethylamine (6 mL) andwater(60mL)at0

°C, then the aqueous phase extracted thrice with ethyl acetate. The pooled organic phases were then washed with brine and dried over MgSO4.Purificationbyflash column chromatography (2:1 petroleum ether/ethyl acetate)yieldedapaleyellow syrup, composed of the desired product and its C4 epimer in a 1.2:1 ratio, which proved to not be separable by further chromatography (3.07 g,6.47mmol,77%).

1 H-NMR (300 MHz; CDCl3): δ 6.04 (d, J3,4 =5.9Hz,1H,H-334), 5.95 (d, J3,4 = 2.9 Hz, 1H, H-3 35), 5.57-5.46 (m, 4H, H-4,7 35 & 34), 5.42-5.34 (m, 2H, H-8 35 &

34), 5.20 (dd, J5,6 =9.4,J4,5 =7.0Hz,1H,H-535), 5.00 (dd, J5,6 =11.0,J4,5 =

4.0 Hz, 1H, H-5 34), 4.67 (dd, J9a,9b =12.5,J8,9a =2.3Hz,1H,H-9a 34), 4.55 (dd,

199 7.2. Synthesis

J9a,9b =12.5,J8,9a =2.5Hz,1H,H-9a 35), 4.38 (dd, J6,7 =1.7Hz,1H,H-634),

4.32 (dd, J6,7 =3.0Hz,1H,H-635), 4.17 (dd, J8,9b =6.2Hz,2H,H-9b 35 & 34), 3.80 (s, 3H, OMe 34), 3.79 (s, 3H, OMe 35), 2.09 (s, 3H, OAc), 2.08 (s, 6H, 2xOAc),

2.07 (s, 3H, OAc), 2.06 (s, 3H, OAc), 2.05 (s, 3H, OAc), 2.04 (s,3H,OAc),2.03(s,

6H, 2xOAc), 1.99 (s, 3H, OAc) ppm. MS : Calcd. for C20H26NaO13:497.13;found: 497.2

Methyl 2,3-dideoxy-d-glycero-d-galacto-non-2-enopyranosonate (37) OH COOMe O HO HO HO HO and

Methyl 2,3-dideoxy-d-glycero-d-talo-non-2-enopyranosonate (36) OH OH COOMe O HO HO HO

Methyl 4,5,7,8,9-penta-O-acetyl-2,3-dideoxy-d-glycero-d-galacto-non-2-enopyra-

nosonate (35,2.08g,4.38mmol),contaminatedwithitsC4epimer(34), was sub-

jected to the general method for Zemplén deprotection (page 187). Purification by

flash column chromatography (18:1 ethyl acetate/methanol) yielded a white powder,

composed of the desired product and its C4 epimer in a 1.5:1 ratio, which proved to

not be separable by further chromatography or crystallisation (0.925 g, 3.50 mmol,

1 80%). H-NMR (300 MHz; CD3OD): δ 6.05 (d, J3,4 =5.9Hz,1H,H-336), 5.87 (d,

J3,4 =2.6Hz,1H,H-337), 4.27 (dd, J4,5 =7.7Hz,1H,H-437), 4.21 (dd, J4,5 =3.9

Hz, 1H, H-4 36), 4.13 (d, J5,6 =11.5Hz,1H,H-637), 4.10 (d, J5,6 =7.3Hz,1H,

200 7.2. Synthesis

H-6 36), 3.89-3.82 (m, 6H, H-7,8,9a 37 & 36), 3.78 (s, 3H, OMe 36), 3.78 (s, 3H,

OMe 37), 3.75-3.65 (m, 4H, H-5,9b 37 & 36)ppm.MS:Calcd.forC10H16NaO8: 287.07; found: 287.2

Methyl 6,8-O-phenylmethylene-2,3-dideoxy-d-glycero-d-galacto-non-2- enopyranosonate (39)

OH COOMe O O HO O HO Ph and

Methyl 4,5:6,8-bis-O-phenylmethylene-2,3-dideoxy-d-glycero-d-talo-non-

2-enopyranosonate (38)

HO Ph O COOMe O O O O Ph

Methyl 2,3-dideoxy-d-glycero-d-galacto-non-2-enopyranosonate (37,0.570g,2.16 mmol), contaminated with its C4 epimer (36), was dissolved in acetonitrile (40 mL), then benzaldehyde dimethyl acetal and p-toluenesulfonic acid (40 mg, 0.1 eq., mono- hydrate) were added and the reaction allowed to proceed at ambient temperature for 3.5 hours. The reaction was quenched with triethylamine and the solvent evap- orated in vacuo before purification by flash column chromatography (3:1 petroleum ether/ethyl acetate followed by neat ethyl acetate). Compound 38 was crystallised from ethanol to yield white plates (222 mg, 0.504 mmol, 23%), while the desired product 39 was purified on a second flash column (6% methanol in dichlorometh-

201 7.2. Synthesis

1 ane) to yield a white powder (138 mg, 0.392, 18%). 38: H-NMR (300 MHz; CDCl3):

δ 7.51-7.45 (m, 4H, Ph), 7.41-7.34 (m, 6H, Ph), 6.37 (d, J3,4 =4.3Hz,1H,H-3),

5.89 (s, 1H, CHPh), 5.56 (s, 1H, CHPh), 4.66 (dd, J4,5 =6.3Hz,1H,H-4),4.55

(dd, J5,6 =9.5Hz,1H,H-5),4.37(dd,J9a,9b =10.3,J8,9a =5.4Hz,1H,H-9a),

4.27 (dd, J6,7 =5.3Hz,1H,H-6),4.04-3.98(m,2H,H-7,8),3.81(s,3H,Me),3.67

(t, J8,9b =10.3Hz,1H,H-9b)ppm.MS:Calcd.forC24H24NaO8 :463.1;found: 463.2 39: 1H-NMR (400 MHz; acetone-d6): δ 7.51-7.46 (m, 2H, Ph), 7.38-7.31 (m,

3H, Ph), 5.85 (d, J3,4 =2.5Hz,1H,H-3),5.56(s,1H,CHPh),4.32(dd,J4,5 =7.7

Hz, 1H, H-4), 4.28 (dd, J9a,9b =10.4,J8,9a =4.8Hz,1H,H-9a), 4.21 (dd, J6,7 =0.9

Hz, 1H, H-6), 4.16-4.08 (m, 1H, H-7,8), 3.88 (dd, J5,6 =10.4Hz,1H,H-5),3.72(s,

3H, OMe), 3.64 (t, J8,9b =10.4Hz,1H,H-9b)ppm.MS:Calcd.forC17H20NaO8: 375.1; found: 375.3

2,3-Dideoxy-d-glycero-d-galacto-non-2-enopyranosonate (40) OH COOH O HO HO HO HO

Benzylidene-protected Kdn2en (39,45mg,0.128mmol)wassubjectedtothe

general method for hydrolysis by lithium hydroxide (page 187), at ambient tem-

perature. The reaction was then quenched with Sephadex ion exchange resin (H+

form) and stirred until the benzylidene was cleaved (7 days),stoppingshortofcom-

pletion to mitigate hydration of the final product under the acidic conditions. The

reaction mixture was then filtered, the organic solvent evaporated in vacuo,andthe

product purified by 5 g C-18 Sep-pak eluted with water. Lyophilisation yielded a

1 white powder (26 mg, 0.104 mmol, 81%). H-NMR (400 MHz; D2O): δ 5.97 (d, J3,4

=2.4Hz,1H,H-3),4.44(dd,J4,5 =7.9Hz,1H,H-4),4.19(d,1H,H-6),3.94-3.90

202 7.2. Synthesis

(m, 3H, H-7,8,9a), 3.80 (dd, J5,6 =10.4Hz,1H,H-5),3.70(dd,J9a,9b =13.3,J8,9b

=4.0Hz,1H,H-9b)ppm.HRMS:Calcd.forC9H13O8:249.0610;found:249.0607

Methyl (phenyl 2,3,4-tri-O-acetyl-α-d-glucopyranosid)uronate (41) MeOOC O AcO AcO AcO O

Globally protected glucuronic acid (2,2.013g,5.35mmol)wasdissolvedinphenol

(4.17 g) at 80 °C, ZnCl2 added (1 g) and the reaction mixture stirred under vacuum (aspirator) for two hours. The reaction mixture was allowed to cool, dissolved in

ethyl acetate and then extracted with 10% Na2SO4,1MNaCl,and10%Na2SO4

again. After drying over MgSO4 and concentrating, the residue was purified by flash column chromatography (2:1 hexanes/ethyl acetate) to give awhitefoamofthetitle

compound (0.135 g, 0.33 mmol, 6%), as well as the beta anomer (0.329 g, 0.80 mmol, ! ! 15%). 1H-NMR (300 MHz; DMSO-d6): δ 7.33 (m, 2H, H-3 ,5 ), 7.13-7.04 (m, 3H, ! ! ! H-2 ,4 ,6 ), 5.93 (d, J1,2 =3.6Hz,1H,H-1),5.53(t,J2,3 =J3,4 =9.8Hz,1H,H-3),

5.17-5.08 (m, 2H, H-2,4), 4.39 (d, J4,5 =9.8Hz,1H,H-5),3.59(s,3H,OMe),2.01 (s, 3H, OAc), 2.00 (s, 3H, OAc), 1.98 (s, 3H, OAc) ppm.

Methyl (phenyl 2,3-di-O-acetyl-4-deoxy-β-l-threo-hex-4- enopyranosid)uronate (42)

COOMe O AcO AcOO

Globally protected phenyl α-d-glucuronide (41,135mg,0.33mmol)wassub- jected to DBU-catalysed elimination by the general method . Purification by flash column chromatography (5:2 petroleum ether/ethyl acetate)gaveawhitefoam(50

203 7.2. Synthesis

1 mg, 0.14 mmol, 44%). H-NMR (300 MHz; CDCl3): δ 7.31 (dd, J2!,3! =J5!,6! =8.7, ! ! ! ! ! J3!,4! =J4!,5! =7.3Hz,2H,H-3,5 ), 7.11-7.05 (m, 3H, H-2 ,4 ,6 ), 6.14 (d, J2,4 =

2.8 Hz, 1H, H-4), 5.84 (dd, J1,2 =8.2,J1,3 =2.8Hz,2H,H-1),5.82(d,1H,H-3),

13 5.30 (dd, 1H, H-2), 3.78 (s, 3H), 2.11 (s, 6H) ppm. C-NMR (101 MHz; CD3OD): δ 170.6 (OAc), 170.3 (OAc), 161.9 (C-6), 156.6 (C-5), 141.8 (Ar), 129.5 (Ar), 123.3 (Ar), 116.9 (Ar), 108.4 (C-4), 95.9 (C-1), 68.6 (C-3), 66.4 (C-2), 51.8 (OMe), 19.5

(OAc), 19.3 (OAc) ppm.

Phenyl 4-deoxy-β-l-threo-hex-4-enopyranosiduronic acid (43)

COOH O HO HOO

Globally protected phenyl unsaturated α-d-glucuronide (42,25mg,0.071mmol) was deprotected by the general method for ester saponification (page 187). Puri-

fication was by 5 g C-18 Sep-pak, washed with water, 10% acetonitrile in water,

40% acetonitrile in water, 60% acetonitrile in water and 100%acetonitrile.All fractions containing product as determined by TLC (3:3:1:1 toluene/ethyl acet- ate/methanol/acetic acid) were pooled and lyophilised to give a white powder (11.5

1 ! ! mg, 0.046 mmol, 64%). H-NMR (400 MHz; D2O): δ 7.22 - 7.36 (m, 2H, H-3 ,5 ), ! ! ! 7.00 - 7.11 (m, 3H, H-2 ,4 ,6 ), 6.00 (d, J3,4 =3.1Hz,1H,H-4),5.70(d,J1,2 =2.5

13 Hz, 1H, H-1), 4.47 (dd, J2,3 =7.8Hz,1H,H-3),3.90(dd,1H,H-2)ppm. C-NMR

(151 MHz; D2O): δ 165.9 (C-6), 155.5 (C-5), 141.3 (Ar), 129.5 (Ar), 123.2 (Ar), 117.0 (Ar), 110.9 (C-4), 97.4 (C-1), 69.2 (C-3), 65.3 (C-2) ppm. MS : Calcd. for

C12H12NaO6:275.0;found:275.4

204 7.2. Synthesis

Methyl (phenyl 2,3,4-tri-O-acetyl-1-thio-β-d- glucopyranosid)uronate (44)

MeOOC O AcO AcO S OAc

Globally protected glucuronyl bromide (3,0.482g,1.21mmol)wasdissolvedin

ethyl acetate (10 mL), then tetra-n-butylammonium bromide (0.315 g), thiophenol

(200 µL, 1.6 eq.) and 1 M Na2CO3 (10 mL) were added. The reaction mixture was stirred vigorously at ambient temperature for 2 hours then diluted with ethyl acetate

and washed with 1 M NaOH, sat. NaHCO3,waterandbrinebeforedryingover

MgSO4.Purificationbyflashcolumnchromatography(3:1petroleumether/ethyl acetate) gave a white foam (405 mg, 0.95 mmol, 78%). 1H-NMR (300 MHz; DMSO- ! ! ! ! ! d6): δ 7.45-7.33 (m, 5H, H-2 ,3 ,4 ,5 ,6 ), 5.43 (t, J2,3 =J3,4 =9.8Hz,1H,H-3),5.40

(d, J1,2 =9.8Hz,1H,H-1),4.96(t,1H,H-2),4.86(dd,J4,5 =10.0Hz,1H,H-4), 4.58 (d, 1H, H-5), 3.65 (s, 3H), 2.03 (s, 3H), 1.98 (s, 3H), 1.95(s,3H)ppm.

Methyl (phenyl 2,3-di-O-acetyl-4-deoxy-1-thio-α-l-threo-hex-4- enopyranosid)uronate (45)

COOMe O AcO S OAc

Globally protected phenyl thio-β-d-glucuronide (44,163mg,0.38mmol)was subjected to the DBU-catalysed elimination as per the generalmethod.Purification by flash column chromatography (5:2 petroleum ether/ethyl acetate) gave a white

1 powder (78 mg, 0.21 mmol, 56%). H-NMR (300 MHz; CDCl3): δ 7.60-7.57 (m, 2H,

Ar), 7.37-7.32 (m, 3H, Ar), 6.35 (dd, J3,4 =4.8,J2,4 =1.4Hz,1H,H-4),5.71(dd,

J1,2 =2.3,J1,3 =1.5Hz,1H,H-1),5.32(br.q,J2,3 =1.5Hz,1H,H-2),5.20(dt,

205 7.2. Synthesis

Hz, 1H, H-3), 3.87 (s, 3H, OMe), 2.17 (s, 3H, OAc), 2.08 (s, 3H, OAc) ppm.

Phenyl 4-deoxy-1-thio-α-l-threo-hex-4-enopyranosiduronic acid (46) COOH O HO S OH

Globally protected phenyl unsaturated thio-β-d-glucuronide (45,65mg,0.177 mmol) was deprotected by the general method for ester saponification (page 187).

Purification was by 5 g C-18 Sep-pak, washed with water, 10%, 20%, 30%, 40%,

60% and 100% acetonitrile in water. All fractions containingpureproductasde- termined by TLC (3:3:1:1 toluene/ethyl acetate/methanol/acetic acid) were pooled and lyophilised to give a white powder (41 mg, 0.153 mmol, 86%). 1H-NMR (400 ! ! MHz; D2O): δ 7.58 (dd, J2,3 =J5,6 =6.5,J2,4 =J4,6 =3.2Hz,2H,H-2,6 ), 7.34 - ! ! ! 7.44 (m, 3H, H-3 ,4 ,5 ), 6.23 (dd, J3,4 =4.0,J2,4 =0.9Hz,1H,H-4),5.56(dd,J1,2

=4.9,J1,3 =0.9Hz,1H,H-1),4.21(td,J2,3 =4.0Hz,1H,H-3),3.98(ddd,1H,

13 H-2) ppm. C-NMR (101 MHz; D2O): δ 165.5 (C-6), 141.7 (C-5), 133.1 (Ar), 132.2 (Ar), 129.5 (Ar), 128.8 (Ar), 111.7 (C-4), 86.3 (C-1), 69.7 (C-3), 65.7 (C-2) ppm.

HRMS : Calcd. for C12H11NaO5S: 291.0303; found: 291.0304

7.2.5 Substrates for linear free-energy relationship

Methyl (2,4-dinitrophenyl 2,3,4-tri-O-acetyl-β-d- glucopyranosid)uronate (54)

MeOOC O AcO AcO O OAc NO2 O2N

2,4-Dinitrophenol (0.983 g, 5.34 mmol, 3.75 eq.) was reactedwithgloballyprotec-

206 7.2. Synthesis ted glucuronyl bromide (3,0.585g,1.47mmol)bythegeneralmethodforKoenigs-

Knorr glycosylation (page 186). The resulting off-white powder (0.580 g, 1.16 mmol,

79%) was used without further purification. 1H-NMR (300 MHz; DMSO-d6): δ 8.79 ! ! ! (d, J3!,5! =2.8Hz,1H,H-5), 8.54 (dd, J2!,3! =9.3Hz,1H,H-3), 7.65 (d, 1H, H-2 ),

5.97 (d, J1,2 =7.4Hz,1H,H-1),5.45(t,J2,3 =J3,4 =9.3Hz,1H,H-3),5.16(t,1H,

H-2), 5.13 (t, J4,5 =9.9Hz,1H,H-4),4.80(d,1H,H-5),3.62(s,3H,OMe),2.02(s, 3H, OAc), 2.01 (s, 3H, OAc), 2.00 (s, 3H, OAc) ppm.

Methyl (2,5-dinitrophenyl 2,3,4-tri-O-acetyl-β-d- glucopyranosid)uronate (56)

MeOOC O AcO NO2 AcO O OAc

O2N

2,5-Dinitrophenol (0.245 g, 1.33 mmol, 1.3 eq.) was reacted with globally protec- ted glucuronyl bromide (3,0.403g,1.02mmol)bythegeneralmethodforKoenigs-

Knorr glycosylation (page 186). The resulting off-white powder (0.442 g, 0.883 mmol, 85%) was used without further purification. 1H-NMR (300 MHz; DMSO-d6): ! ! ! δ 8.22-8.11 (m, 3H, H-3 ,4 ,6 ), 5.98 (d, J1,2 =7.4Hz,1H,H-1),5.41(t,J2,3 =J3,4

=9.2Hz,1H,H-3),5.16(t,J4,5 =9.5Hz,1H,H-4),5.14(dd,1H,H-2),4.83(d, 1H, H-5), 3.64 (s, 3H, OMe), 2.02 (s, 6H, 2xOAc), 1.99 (s, 3H, OAc) ppm.

207 7.2. Synthesis

Methyl (2,4,6-trichlorophenyl 2,3,4-tri-O-acetyl-β-d- glucopyranosid)uronate (58)

MeOOC O Cl AcO AcO O OAc Cl Cl

2,4,6-Trichlorophenol (0.34 g, 1.72 mmol) was reacted with globally protected glucuronyl bromide (3,0.527g,1.33mmol)bythegeneralmethodforKoenigs-

Knorr glycosylation (page 186). Purification by flash column chromatography (3:1 petroleum ether/ethyl acetate) yielded a white powder (0.441 g, 0.86 mmol, 65%).

1 ! ! H-NMR (300 MHz; DMSO-d6): δ 7.73 (s, 2H, H-3 ,5 ), 5.52 (d, J1,2 =7.9Hz,1H,

H-1), 5.46 (t, J2,3 =J3,4 =9.6Hz,1H,H-3),5.13(dd,1H,H-2),5.02(t,J4,5 = 10.0 Hz, 1H, H-4), 4.51 (d, 1H, H-5), 3.59 (s, 3H, OMe), 2.05 (s,3H,OAc),1.99(s,

3H, OAc), 1.97 (s, 3H, OAc) ppm. MS : Calcd. for C19H19Cl3NaO10:535.0/537.0; found: 535.1/537.1

Methyl (3-nitrophenyl 2,3,4-tri-O-acetyl-β-d- glucopyranosid)uronate (60)

MeOOC O AcO NO2 AcO O OAc

3-Nitrophenol (0.220 g, 1.58 mmol, 1.25 eq.) was reacted withgloballyprotected glucuronyl bromide (3,0.505g,1.27mmol)bythegeneralmethodforKoenigs-

Knorr glycosylation (page 186). Purification by flash column chromatography (2:1 petroleum ether/ethyl acetate) yielded an off-white powder (0.304 g, 0.668 mmol,

1 53%). H-NMR (300 MHz; DMSO-d6): δ 7.95 (ddd, J4!,5! =8.2,J2!,4! =2.2,J4!,6! ! ! =0.8Hz,1H,H-4), 7.80 (t, J2!,6! =2.2Hz,1H,H-2), 7.65 (t, J5!,6! =8.2Hz,1H,

208 7.2. Synthesis

! ! H-5 ), 7.48 (ddd, 1H, H-6 ), 5.87 (d, J1,2 =7.8Hz,1H,H-1),5.46(t,J2,3 =J3,4 =

9.6 Hz, 1H, H-3), 5.14 (dd, 1H, H-2), 5.09 (t, J4,5 =9.9Hz,1H,H-4),4.76(d,1H, H-5), 3.62 (s, 3H, OMe), 2.02 (s, 3H, OAc), 2.01 (s, 3H, OAc), 1.99 (s, 3H, OAc) ppm.

Methyl (4-chlorophenyl 2,3,4-tri-O-acetyl-β-d- glucopyranosid)uronate (62) MeOOC O AcO AcO O OAc Cl

Globally protected glucuronyl hemiacetal (7,0.4g,1.20mmol)wasdissolved

in dichloromethane (12 mL), 4 Å molecular sieves were added and the solution

flushed with argon, then cooled in a dry ice/acetone bath. Trichloroacetonitrile

(1.2 mL, 10 eq.) then DBU (0.25 mL, 1.5 eq.) were added and the reaction mixture

stirred for 30 min at -78 °C, then allowed to warm to ambient temperature. The

reaction mixture was concentrated and filtered through a plugofsilicagel(eluting

with 2:1 petroleum ether/ethyl acetate). The resultant crude trichloroacetimidate

intermediate and 4-chlorophenol (0.19 g, 1.48 mmol, 1.2 eq.)weredissolvedin

dichloromethane (25 mL) and dried over 4 Å molecular sieves under argon, then

cooled in a dry ice/acetone bath. Boron trifluoride diethyl etherate (0.05 mL, 0.3

eq.) was added and the reaction mixture stirred 30 min at -78 °Cbeforeallowing

to warm to ambient temperature while stirring overnight. Thereactionmixturewas

diluted with an equal volume of dichloromethane and washed with sat. NaHCO3

thrice then with brine. The organic phase was dried over MgSO4 then purified by flash column chromatography (3:1 petroleum ether/ethyl acetate) to give a white

1 powder (0.103 g, 0.232 mmol, 20%). H-NMR (300 MHz; CDCl3): δ 7.25 (d, J2!,3! ! ! ! ! =J5!,6! =9.0Hz,2H,H-3,5 ), 6.94 (d, 2H, H-2 ,6 ), 5.36-5.23 (m, 3H, H-2,3,4), 5.11

209 7.2. Synthesis

(d, J1,2 =7.2Hz,1H,H-1),4.19(d,J4,5 =9.5Hz,1H,H-5),3.73(s,3H,OMe),2.06 (s, 3H, OAc), 2.05 (s, 3H, OAc), 2.04 (s, 3H, OAc) ppm.

Methyl (4-tert-butylphenyl 2,3,4-tri-O-acetyl-β-d- glucopyranosid)uronate (64)

MeOOC O AcO AcO O OAc

Globally protected glucuronyl hemiacetal (7,0.4g,1.20mmol)wasdissolvedin

dichloromethane (12 mL), 4 Å molecular sieves were added and the solution flushed

with argon, then cooled in a dry ice/acetone bath. Trichloroacetonitrile (1.2 mL, 10

eq.) then DBU (0.25 mL, 1.5 eq.) were added and the reaction mixture stirred for 30

min at -78 °C, then allowed to warm to ambient temperature. Thereactionmixture

was concentrated and filtered through a plug of silica gel (eluting with 2:1 petro-

leum ether/ethyl acetate). The resultant crude trichloroacetimidate intermediate

and 4-tert-butylphenol (0.25 g, 1.66 mmol, 1.4 eq.) were dissolved in dichlorometh-

ane (25 mL) and dried over 4 Å molecular sieves under argon, then cooled in a dry

ice/acetone bath. Boron trifluoride diethyl etherate (0.025 mL, 0.3 eq.) was added

and the reaction mixture stirred 30 min at -78 °C before allowing to warm to ambi-

ent temperature while stirring overnight. The reaction mixture was diluted with an

equal volume of dichloromethane and washed with sat. NaHCO3 thrice then brine.

The organic phase was dried over MgSO4 then purified twice by flash column chro- matography (2:1 petroleum ether/ethyl acetate then 5:2 hexanes/ethyl acetate) to

1 give an amorphous solid (0.158 g, 0.339 mmol, 28%). H-NMR (300 MHz; CDCl3): ! ! ! ! δ 7.31 (d, J2!,3! =J5!,6! =8.9Hz,2H,H-3,5 ), 6.93 (d, 2H, H-2 ,6 ), 5.36-5.25 (m,

3H, H-2,3,4), 5.14 (d, J1,2 =7.2Hz,1H,H-1),4.19(d,J4,5 =9.5Hz,1H,H-5),3.74

210 7.2. Synthesis

(s, 3H, OMe), 2.06 (s, 3H, OAc), 2.05 (s, 3H, OAc), 2.05 (s, 3H, OAc) ppm.

Methyl (1-O-benzyl-2,3,4-tri-O-acetyl-β-d-glucopyranosid)uronate (66)

MeOOC O AcO AcO O OAc

Globally protected glucuronyl hemiacetal (7,0.343g,1.03mmol)wasdissolved in toluene (2 mL) and the solution flushed with nitrogen. Benzylbromide(0.245mL,

2eq.)andsilver(I)carbonate(0.885g,3eq.)wereaddedandthe reaction mixture stirred overnight at ambient temperature in the dark. The reaction was quenched with triethylamine, diluted with dichloromethane and filtered through Celite. The

filtrate was then washed with 1 M HCl, saturated NaHCO3,waterandbrinebefore drying over MgSO4.Purificationbyflashcolumnchromatography(7:2petroleum ether/ethyl acetate) followed by recrystallisation from toluene/petroleum ether yiel-

1 ded white plates (0.135 g, 0.318 mmol, 31%). H-NMR (300 MHz; CDCl3): δ ! ! ! ! ! ! ! 7.36-7.29 (m, 5H, H-2 ,3 ,4 ,5 ,6 ), 5.28-5.18 (m, 2H, H-7 a,7 b), 5.08 (br. t, J3,4 =

8.3, J4,5 =9.2Hz,1H,H-4),4.92(d,J1,2 =12.3Hz,1H,H-1),4.64-4.58(m,2H, H-2,3), 4.02 (d, 1H, H-5), 3.76 (s, 3H, OMe), 2.01 (s, 3H, OAc),2.01(s,3H,OAc),

13 1.99 (s, 3H, OAc) ppm. C-NMR (75 MHz; CDCl3): δ 170.3 (OAc), 169.51 (OAc), 169.35 (OAc), 167.4 (C-6), 136.6 (Ar), 128.6 (Ar), 128.2 (Ar), 127.9 (Ar), 99.4 (C-1),

72.8 (C-3), 72.2 (C-2), 71.3 (C-4), 71.0 (OCH2Ar), 69.6 (C-5), 53.1 (OMe), 20.74

(2xOAc), 20.64 (OAc) ppm. MS : Calcd. for C20H24NaO10:447.1; found: 447.4

211 7.2. Synthesis

Methyl (2,4-dinitrophenyl 2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4- enopyranosid)uronate (55)

COOMe O AcO O OAc NO2 O2N

Globally protected 2,4-dinitrophenyl β-d-glucuronide (54,0.275g,0.550mmol) was subjected to the general method for DBU-catalysed elimination (page 186).

Purification by flash column chromatography (5:2 petroleum ether/ethyl acetate) gave a white powder (0.133 g, 0.302 mmol, 55%). 1H-NMR (300 MHz; DMSO-d6): ! ! δ 8.80 (d, J3!,5! =2.8Hz,1H,H-5), 8.60 (dd, J2!,3! =9.3Hz,1H,H-3), 7.83 (d, ! 1H, H-2 ), 6.57 (dd, J3,4 =2.1,J2,4 =0.9Hz,1H,H-4),6.25(dd,J1,2 =4.8,J1,3 =

1.3 Hz, 1H, H-1), 5.27-5.25 (td, J2,3 =2.1Hz,1H,H-3),5.20(ddd,1H,H-2),3.73 (s, 3H, OMe), 2.10 (s, 3H, OAc), 2.09 (s, 3H, OAc) ppm.

Methyl (2,5-dinitrophenyl 2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4- enopyranosid)uronate (57)

COOMe O NO2 AcO O OAc

O2N

Globally protected 2,5-dinitrophenyl β-d-glucuronide (56,0.222g,0.444mmol) was subjected to the general method for DBU-catalysed elimination (page 186).

Purification by flash column chromatography (3:1 petroleum ether/ethyl acetate)

1 gave a white powder (0.098 g, 0.223 mmol, 50%). H-NMR (300 MHz; CDCl3): δ ! ! ! 8.40 (d, J4!,6! =2.2Hz,1H,H-6), 8.04 (dd, J3!,4! =8.8,1H,H-4), 7.90 (d, 1H, H-3 ),

6.35 (dd, J3,4 =4.9,J2,4 =1.4Hz,1H,H-4),6.05(dd,J1,2 =1.9,J1,3 =1.1Hz,1H,

212 7.2. Synthesis

H-1), 5.28 (ddd, J2,3 =1.1Hz,1H,H-2),5.24(dt,1H,H-3),3.80(s,3H,OMe),2.14 (s, 3H, OAc), 2.13 (s, 3H, OAc) ppm.

Methyl (2,4,6-trichlorophenyl 2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4- enopyranosid)uronate (59)

COOMe Cl O AcO O OAc Cl Cl

Globally protected 2,4,6-trichlorophenyl β-d-glucuronide(58,0.441g,0.858mmol) was subjected to the general method for DBU-catalysed elimination (page 186).

Purification by flash column chromatography (3:1 petroleum ether/ethyl acetate)

1 gave a white powder (0.279 g, 0.61 mmol, 72%). H-NMR (300 MHz; CDCl3): δ ! ! 7.28 (s, 2H, H-3 ,5 ), 6.36 (dd, J3,4 =5.1,J2,4 =1.4Hz,1H,H-4),5.79(dd,J1,2 =1.7,

J1,3 =1.1Hz,1H,H-1),5.44(br.q,J2,3 =1.7Hz,1H,H-2),5.20(br.dt,1H,H-3),

13 3.79 (s, 3H, OMe), 2.06 (s, 6H, 2xOAc) ppm. C-NMR (75 MHz; CD3OD): δ 171.5 (OAc), 170.6 (OAc), 163.4 (C-6), 148.8 (C-5), 144.0 (Ar), 132.1 (Ar), 131.1 (Ar),

130.2 (Ar), 107.9 (C-4), 98.2 (C-1), 68.3 (C-3), 64.5 (C-2), 53.1 (OMe), 20.74 (OAc),

20.58 (OAc) ppm MS : Calcd. for C17H15Cl3NaO8:475.0/477.0;found:475.2/477.2

Methyl (3-nitrophenyl 2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4- enopyranosid)uronate (61)

COOMe O NO2 AcO O OAc

Globally protected 3-nitrophenyl β-d-glucuronide (60,0.103g,0.232mmol)was subjected to the general method for DBU-catalysed elimination (page 186). Puri-

213 7.2. Synthesis

fication by flash column chromatography (3:1 hexanes/ethyl acetate) gave a white

1 powder (0.064 g, 0.166 mmol, 72%). H-NMR (300 MHz; CDCl3): δ 7.93-7.90 (m, ! ! ! ! 2H, H-2 ,4 ), 7.50-7.41 (m, 2H, H-5 ,6 ), 6.30 (dd, J3,4 =4.0,J2,4 =1.9Hz,1H,H-4),

5.88 (dd, J1,2 =1.9,J1,3 =1.1Hz,1H,H-1),5.30-5.26(m,2H,H-2,3),3.77(s,3H, OMe), 2.12 (s, 3H, OAc), 2.10 (s, 3H, OAc) ppm.

Methyl (4-chlorophenyl 2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4- enopyranosid)uronate (63) COOMe O AcO O OAc Cl

Globally protected 4-chlorophenyl β-d-glucuronide (62,0.103g,0.232mmol) was subjected to the general method for DBU-catalysed elimination (page 186).

Purification by flash column chromatography (3:1 hexanes/ethyl acetate) gave a

1 white powder (0.064 g, 0.166 mmol, 72%). H-NMR (300 MHz; CDCl3): δ 7.26 (d, ! ! ! ! J2!,3! =J5!,6! =9.0Hz,2H,H-3,5 ), 7.03 (d, 2H, H-2 ,6 ), 6.29 (dd, J3,4 =4.5,J2,4

=1.5Hz,1H,H-4),5.77(dd,J1,2 =2.7,J1,3 =1.0Hz,1H,H-1),5.29-5.26(m,2H, H-2,3), 3.80 (s, 3H, OMe), 2.13 (s, 3H, OAc), 2.11 (s, 3H, OAc) ppm. 13C-NMR (101

MHz; CD3OD): δ 170.4 (OAc), 169.7 (OAc), 162.1 (C-6), 154.9 (C-5), 142.1 (Ar), 129.4 (Ar), 128.1 (Ar), 118.5 (Ar), 107.7 (C-4), 95.1 (C-1), 68.2 (C-3), 64.9 (C-2),

51.9 (OMe), 19.47 (OAc), 19.32 (OAc) ppm.

214 7.2. Synthesis

Methyl (4-tert-butylphenyl 2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4- enopyranosid)uronate (65)

COOMe O AcO O OAc

Globally protected 4-tert-butylphenyl β-d-glucuronide (64,0.158g,0.338mmol) was subjected to the general method for DBU-catalysed elimination (page 186).

Purification by flash column chromatography (4:1 hexanes/ethyl acetate) gave a

1 white powder (0.061 g, 0.150 mmol, 44%). H-NMR (300 MHz; CDCl3): δ 7.31 (d, ! ! ! ! J2!,3! =J5!,6! =8.9Hz,2H,H-3,5 ), 7.02 (d, 2H, H-2 ,6 ), 6.28 (dd, J3,4 =4.2,J2,4

=1.6Hz,1H,H-4),5.78(dd,J1,2 =2.6,J1,3 =1.4Hz,1H,H-1),5.31-5.26(m, 2H, H-2,3), 3.81 (s, 3H, OMe), 2.13 (s, 3H, OAc), 2.11 (s, 3H, OAc), 1.29 (s, 9H,

13 tert-butyl) ppm. C-NMR (101 MHz; CD3OD): δ 171.7 (OAc), 171.0 (OAc), 163.4 (C-6), 155.4 (C-5), 147.4 (Ar), 143.6 (Ar), 127.5 (Ar), 117.8(Ar),108.7(C-4),96.8

(C-1), 69.7 (C-3), 66.4 (C-2), 53.1 (OMe), 35.1(tBu), 31.9 (tBu), 20.73 (OAc), 20.59

(OAc) ppm.

Methyl (1-O-benzyl-2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4- enopyranosid)uronate (67)

COOMe O AcO O OAc

Globally protected benzyl β-d-glucuronide (66,32mg,0.075mmol)wassubjec- ted to the general method for DBU-catalysed elimination (page186).Purificationby

flash column chromatography (5:2 petroleum ether/ethyl acetate) gave a colourless

1 syrup (0.020 g, 0.055 mmol, 73%). H-NMR (300 MHz; CDCl3): δ 7.37-7.28 (m, 5H,

215 7.2. Synthesis

! ! ! ! ! H-2 ,3 ,4 ,5 ,6 ), 6.23 (dd, J3,4 =4.5,J2,4 =1.1Hz,1H,H-4),5.30(d,J1,2 =2.5Hz,

! ! 1H, H-1), 5.20 (dd, J2,3 =1.9Hz,1H,H-3),5.14(br.q,1H,H-2),4.87(d,J7 a,7 b ! ! =12.3Hz,1H,H-7a), 4.68 (d, 1H, H-7 b), 3.82 (s, 3H, OMe), 2.08 (s, 3H, OAc),

13 2.06 (s, 3H, OAc) ppm. C-NMR (75 MHz; CDCl3): δ 170.2 (OAc), 169.5 (OAc), 162.3 (C-6), 142.4 (C-5), 136.8 (Ar), 128.5 (Ar), 128.0 (Ar),127.5(Ar),107.5(C-4),

95.7 (C-1), 70.6 (OCH2Ar), 68.5 (C-3), 64.4 (C-2), 52.7 (OMe), 21.03 (OAc), 20.88

(OAc) ppm. MS : Calcd. for C18H20NaO8:387.1;found:387.3

2,4-Dinitrophenyl 4-deoxy-α-l-threo-hex-4-enopyranosiduronic acid (47) COOH O HO O OH NO2 O2N

Globally protected 2,4-dinitrophenyl unsaturated β-d-glucuronide (55,30mg, 0.068 mmol) was dissolved in acetone (3.4 mL). Aqueous 1 M HCl (3.4 mL) was

added and the reaction mixture stirred at ambient temperature for 20 days, with

monitoring by TLC (3:3:1:1 toluene/ethyl acetate/methanol/acetic acid). The reac-

tion was stopped by removing the organic solvent under vacuumandimmediately

purified by HPLC over a C-18 stationary phase, eluting with water (5 min) then a

linear gradient to 100% acetonitrile over 1 hour. Product fractions were identified

by UV-vis absorbance on-line and confirmed by TLC (3:3:1:1 toluene/ethyl acet-

ate/methanol/acetic acid), then pooled and lyophilised to give an off-white powder

1 (7.8 mg, 0.022 mmol, 32%). H-NMR (600 MHz; CD3OD): δ 8.78 (d, J3!,5! =2.8Hz, ! ! ! 1H, H-5 ), 8.53 (dd, J2!,3! =9.3Hz,1H,H-3), 7.86 (d, 1H, H-2 ), 6.30 (dd, J3,4 =

4.2, J2,4 =0.7Hz,1H,H-4),6.07(d,J1,2 =4.1,J1,3 =0.7Hz,1H,H-1),4.15(br.

13 td, J2,3 =3.8Hz,1H,H-3),4.09-4.08(br.td,1H,H-2)ppm. C-NMR (151 MHz;

CD3OD): δ 164.67 (C-6), 155.14 (C-5), 143.26 (Ar), 141.15 (Ar), 141.07(Ar),130.16

216 7.2. Synthesis

(Ar), 122.67 (Ar), 119.74 (Ar), 114.11 (C-4), 99.78 (C-1), 70.56 (C-3), 66.89 (C-2) ppm. HRMS : Calcd. for C12H9N2O10:341.0257;found:341.0260

2,5-Dinitrophenyl 4-deoxy-α-l-threo-hex-4-enopyranosiduronic acid (48) COOH O NO2 HO O OH

O2N

Globally protected 2,5-dinitrophenyl unsaturated β-d-glucuronide (57,24mg, 0.050 mmol) was dissolved in acetone (2.5 mL). Aqueous 1 M HCl (2.5 mL) was

added and the reaction mixture stirred at ambient temperature for 16 days, with

monitoring by TLC (3:3:1:1 toluene/ethyl acetate/methanol/acetic acid). The reac-

tion was stopped by removing the organic solvent under vacuumandimmediately

purified by HPLC over a C-18 stationary phase, eluting with water (5 min) then a

linear gradient to 100% acetonitrile over 1 hour. Product fractions were identified

by UV-vis absorbance on-line and confirmed by TLC (3:3:1:1 toluene/ethyl acet-

ate/methanol/acetic acid), then pooled and lyophilised to give an off-white powder

1 (9.9 mg, 0.028 mmol, 56%). H-NMR (600 MHz; CD3OD): δ 8.43 (d, J4!,6! =1.0Hz, ! ! ! 1H, H-6 ), 8.03 - 8.11 (m, 2 H, H-3 ,4 ), 6.27 (d, J3,4 =4.1Hz,1H,H-4),5.97(d,

J1,2 =4.1Hz,1H,H-1),4.13(br.t,J2,3 =3.8Hz,1H,H-3),4.05(br.t,1H,H-2)

13 ppm. C-NMR (151MHz; CD3OD): δ 163.4 (C-6), 150.4 (C-5), 149.5 (Ar), 144.6 (Ar), 139.9 (Ar), 126.1 (Ar), 118.0 (Ar), 114.5 (Ar), 112.9 (C-4), 99.2 (C-1), 69.4

(C-3), 65.6 (C-2) ppm. HRMS : Calcd. for C12H9N2O10:341.0257;found:341.0251

217 7.2. Synthesis

2,4,6-Trichlorophenyl 4-deoxy-α-l-threo-hex-4- enopyranosid)uronate (49)

COOH Cl O HO O OH Cl Cl

Globally protected 2,4,6-trichlorophenyl unsaturated β-d-glucuronide (59,21mg, 0.044 mmol) was deprotected as per the general method for ester saponification (page

187). Purification was by HPLC over a C-18 stationary phase, eluting with water (5 min) then a linear gradient to 100% acetonitrile over 1 hour. Product fractions were identified by UV-vis absorbance on-line and confirmed by TLC (3:3:1:1 toluene/ethyl acetate/methanol/acetic acid) then pooled and lyophilisedtogiveawhitepowder

1 ! ! (9.9 mg, 0.028 mmol, 63%). H-NMR (400 MHz; D2O): δ 7.50 (s, 2H, H-3 ,5 ), 6.34

(dd,J3,4 =4.9,J2,4 =1.3Hz,1H,H-4),6.74(dd,J1,2 =2.2,J1,3 =0.9Hz,1H,H-1),

13 4.30 (br. q, J2,3 =1.8Hz,1H,H-2),4.88(br.dt,1H,H-3)ppm. C-NMR (101

MHz; D2O): δ 149.5 (Ar), 131.8 (Ar), 131.4 (Ar), 130.1 (Ar), 112.0 (C-4), 102.3 (C- 1), 70.1 (C-3), 65.9 (C-2) ppm (carbon 5 and 6 signals too weak to detect, but seen in protected form). MS : Calcd. for C12H8Cl3O6:376.9/378.9;found:377.1/379.1

3-Nitrophenyl 4-deoxy-α-l-threo-hex-4-enopyranosiduronic acid (50) COOH O NO2 HO O OH

Globally protected 3-nitrophenyl unsaturated β-d-glucuronide (61,44mg,0.111 mmol) was subjected to the general method for Zemplén deprotection (page 187), but using a slight excess of sodium methoxide instead of a catalytic amount. Deacetyla- tion was complete within 5 min, at which time water (200 µL) wasaddedandthe

218 7.2. Synthesis reaction mixture stirred at 0 °C for a further 60 min at 0 °C. Thereactionwas quenched with Sephadex ion exchange resin (H+ form) then filtered. Purification by

flash column chromatography (7:2:1 ethyl acetate/methanol/water) then 5 g C-18

Sep-pak, washed with water, 40% acetonitrile in water and 100% acetonitrile, gave aslightlyyellowpowderfollowinglyophilisation(28.1mg,0.095mmol,85%). 1H- ! ! ! ! NMR (600 MHz; D2O): δ 7.88 - 7.96 (m, 2 H, H-2 ,4 ), 7.45 - 7.55 (m, 2 H, H-5 ,6 ),

6.24 (d, J3,4 =4.1Hz,1H,H-4),5.84(d,J1,2 =4.6Hz,1H,H-1),4.27(t,J2,3

13 =4.1Hz,1H,H-3),4.10(br.t,1H,H-2)ppm. C-NMR (151 MHz; D2O): δ 164.7 (C-6), 155.6 (C-5), 148.1 (Ar), 139.9 (Ar), 130.1 (Ar),123.7(Ar),118.0(Ar),

111.9 (Ar), 111.8 (C-4), 97.4 (C-1), 68.6 (C-3), 64.9 (C-2) ppm. HRMS : Calcd. for

C12H10NO8:296.0406;found:296.0411

4-Chlorophenyl 4-deoxy-α-l-threo-hex-4-enopyranosiduronic acid (51) COOH O HO O OH Cl

Globally protected 4-chlorophenyl unsaturated β-d-glucuronide (63,32mg,0.083 mmol) was deprotected by the general method for ester saponification (page 187).

The product was purified by HPLC over a C-18 stationary phase, eluting with water

(5 min) then a linear gradient to 100% acetonitrile over 1 hour. Product frac- tions were identified by UV-vis absorbance on-line and confirmed by TLC (3:3:1:1 toluene/ethyl acetate/methanol/acetic acid), then pooledandlyophilisedtogivean

1 off-white powder (10.8 mg, 0.038 mmol, 46%). H-NMR (400 MHz; D2O): δ 8.09 ! ! ! ! (d, J2!,3! =J5!,6! =9.2Hz,2H,H-3,5 ), 7.17 (d, 2 H, H-2 ,6 ), 6.22 (d, J3,4 =4.1

Hz, 1 H, H-4), 5.84 (d, J1,2 =4.8Hz,1H,H-1),4.29(br.t,J2,3 =4.5Hz,1H,

13 H-3), 4.07 (br. t, 1 H, H-2) ppm. C-NMR (101 MHz; D2O): δ 165.1 (C-6), 161.1 (Ar), 142.7 (Ar), 126.1 (Ar), 116.9 (Ar), 112.5 (C-4), 97.3 (C-1), 69.3 (C-3), 65.7

219 7.2. Synthesis

(C-2) ppm (carbon 5 signal too weak to detect, but seen in protected form). HRMS

:Calcd.forC12H10ClO6:285.0166;found:285.0161

4-Tert-butylphenyl 4-deoxy-α-l-threo-hex-4-enopyranosiduronic acid (52)

COOH O HO O OH

Globally protected 4-tert-butylphenyl unsaturated β-d-glucuronide (65,33mg, 0.081 mmol) was deprotected by the general method for ester saponification (page

187). Purification was by 5 g C-18 Sep-pak, washed with 10% acetonitrile in wa- ter, 40% acetonitrile in water, 60% acetonitrile in water, 75% acetonitrile in water and 100% acetonitrile. All fractions determined by TLC (3:3:1:1 toluene/ethyl acet- ate/methanol/acetic acid) to contain pure product were pooled and lyophilised, while those containing impure product were pooled, lyophilised and purified again to give

1 awhitepowder(13.8mg,0.044mmol,54%). H-NMR (600 MHz; CD3OD): δ 7.34 ! ! ! ! (d, J2!,3! =J5!,6! =8.7Hz,2H,H-3,5 ), 7.10 (d, 2H, H-2 ,6 ), 6.13 (d, J3,4 =3.6Hz,

1H,H-4),5.52(d,J1,2 =5.6Hz,1H,H-1),4.20(br.t,J2,3 =4.9Hz,1H,H-3),

13 3.90 (br. t, 1 H, H-2), 1.30 (s, 9 H, tert-butyl) ppm. C-NMR (151 MHz; CD3OD): δ 165.38 (C-6), 156.26 (C-5), 146.90 (Ar), 127.28 (Ar), 117.95(Ar),113.17(C-4), 100.99 (C-1), 72.03 (C-3), 68.37 (C-2), 35.04 (s, tBu), 31.91 (s, tBu) ppm. HRMS :

Calcd. for C16H19O6:307.1186;found:307.1182

220 7.2. Synthesis

1-O-Benzyl-4-deoxy-α-l-threo-hex-4-enopyranosiduronic acid (53)

COOH O HO O OH

Globally protected benzyl unsaturated β-d-glucuronide(67,20mg,0.055mmol) was deprotected by the general method for ester saponification (page 187). Puri-

fication was by 5 g C-18 Sep-pak, washed with water, 40% acetonitrile in water twice and 100% acetonitrile. All fractions determined by TLC(3:3:1:1toluene/ethyl acetate/methanol/acetic acid) to contain pure product werepooledandlyophilised

1 to give a white powder (14.1 mg, 0.053 mmol, 96%). H-NMR (400 MHz; D2O): δ ! ! ! ! ! 7.35-7.28 (m, 5H, H-2 ,3 ,4 ,5 ,6 ), 6.06 (d, J3,4 =4.0Hz,1H,H-4),5.16(d,J1,2 =4.5 ! ! ! ! Hz, 1H, H-1), 4.77 (d, J7 a,7 b =10.5Hz,1H,H-7a), 4.71 (d, 1H, H-7 b), 4.06 (br. 13 t, J2,3 =4.1Hz,1H,H-3),3.80(br.t,1H,H-2)ppm. C-NMR (101 MHz; D2O): δ 136.86 (Ar), 128.82 (Ar), 128.51 (Ar), 128.46 (Ar), 111.63 (C-4), 99.53 (C-1), 71.54

(s, OCH2Ar), 69.57 (C-3), 65.76 (C-2) ppm (carbon 5 and 6 signals too weak to

detect, but seen in protected form). HRMS : Calcd. for C13H14NaO6:289.0688; found: 289.0690

7.2.6 Substrates for with varied heteroatoms

Methyl (2,3,4-tri-O-acetyl-β-d-glucopyranosyl fluoride)uronate (68)

MeOOC O AcO AcO F OAc

Globally protected α-d-glucuronyl bromide (3,0.142g,0.358mmol)wasdis- solved in acetonitrile (4 mL), then silver (I) fluoride (0.214g,1.687mmol,4.7eq.)

added and stirred at ambient temperature overnight in the dark under a nitrogen

atmosphere. The reaction mixture was filtered through Celite, then purified by flash

221 7.2. Synthesis column chromatography (2:1 petroleum ether/ethyl acetate)togiveawhitepowder

1 (0.103 g, 0.306 mmol, 85%). H-NMR (300 MHz; CDCl3): δ 5.41 (dd, J1,F =51.0,

J1,2 =5.2Hz,1H,H-1),5.39(t,J3,4 =J4,5 =8.2Hz,1H,H-4),5.22(t,J2,3 =8.2Hz,

1H, H-3), 5.08 (ddd, J2,F =9.9Hz,1H,H-2),4.27(d,1H,H-5),3.77(s,3H,OMe),

13 2.09 (s, 3H, OAc), 2.03 (s, 6H, 2xOAc) ppm. C-NMR (75 MHz; CDCl3): δ 169.86

(OAc), 169.34 (OAc), 169.14 (OAc), 167.00 (C-6), 105.93 (d, JC1,F =223.9Hz,C-1),

72.23 (d, JC5,F =3.6Hz,C-5),70.72(d,JC2,F =27.2Hz,C-2),70.48(d,JC3,F =3.0 Hz, C-3), 68.01 (C-4), 53.14 (OMe), 20.61 (2xOAc), 20.55 (OAc) ppm. 19F-NMR

(282 MHz; CDCl3): δ -135.35 (dd, F-1) ppm. MS : Calcd. for C13H17FNaO9:359.1; found: 359.3

Methyl (2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4-enopyranosyl fluoride) uronate (69)

COOMe O AcO F OAc

Globally protected β-d-glucuronyl fluoride (68,0.103g,0.306mmol)wassubjec- ted to the general method for DBU-catalysed elimination (page186).Purification

by flash column chromatography (4:1 petroleum ether/ethyl acetate) gave a colour-

1 less syrup (0.028 g, 0.101 mmol, 33%). H-NMR (400 MHz; CDCl3): δ 6.35 (dd,

J3,4 =4.8,J2,4 =1.6Hz,1H,H-4),5.93(dt,J1,F =49.7,J1,2 =J1,3 =1.6Hz, 1H, H-1), 5.17-5.16 (m, 2H, H-2,3), 3.85 (s, 3H, OMe), 2.09 (s,3H,2xOAc)ppm.

13 C-NMR (101 MHz; CDCl3): δ 169.83 (OAc), 169.21 (OAc), 161.51 (C-6), 141.98

(C-5), 107.47 (C-4), 101.38 (d, JC1,F =227.7Hz,C-1),66.18(d,JC2,F =36.7 Hz, C- 2), 62.31 (C-3), 52.94 (OMe), 20.81 (OAc), 20.66 (OAc) ppm. 19F-NMR (282 MHz;

CDCl3): δ -140.59 (d, F-1) ppm. MS : Calcd. for C11H13FNaO7:299.1;found:299.3

222 7.2. Synthesis

(4-Deoxy-α-l-threo-hex-4-enopyranosyl fluoride)uronic acid (70) COOH O HO F OH

Globally protected unsaturated β-d-glucuronyl fluoride (69,28mg,0.101mmol) was subjected to the general method for Zemplén deprotection(page187)followed

by that for hydrolysis by lithium hydroxide (page 187). The product was purified by

5gC-18Sep-pak,elutingwithwater,10%acetonitrileinwater, 20 % acetonitrile

in water, 40 % acetonitrile in water and 100% acetonitrile. All fractions determined

by TLC to contain product were pooled and purified further by flash column chro-

matography (7:2:1 ethyl acetate/methanol/water) then lyophilised to an off-white

1 powder (12.5 mg, 0.071 mmol, 70%). H-NMR (300 MHz; D2O): δ 6.18 (dd, J3,4 =

4.7, J2,4 =1.2Hz,1H,H-4),5.95(dd,J1,F =51.4,J1,2 =2.3Hz,1H,H-1),4.18

13 (dd, J2,3 =1.2Hz,1H,H-3),4.11(dt,1H,H-2)ppm. C-NMR (101 MHz; D2O):

δ 167.37 (C-6), 108.70 (C-4), 105.10 (d, JC1,F =222.7Hz,C-1),67.30(d,JC2,F =

19 30.3 Hz, C-2), 63.58 (C-3) ppm. F-NMR (282 MHz; CDCl3): δ -142.35 (d, F-1) ppm. MS : Calcd. for C6H6FO5:177.0199;found:177.0202

Methyl (2,3,4-tri-O-acetyl-α-d-glucopyranosyl fluoride)uronate (71) MeOOC O AcO AcO AcO F

Globally protected glucuronic acid (2,0.560g,1.488mmol)wasdissolvedin

70% HF in pyridine (10 mL) in a plastic container at 0 °C, then stirred overnight at 4 °C. The reaction was quenched with solid NaHCO3 on ice, then diluted with

EtOAc. The organic phase was washed with water, sat. NaHCO3 and brine before drying over MgSO4.Purificationbyflashcolumnchromatography(5:2petroleum

223 7.2. Synthesis ether/ethyl acetate) followed by crystallisation from toluene/n-heptane yielded sticky

1 white prisms (0.102 g, 0.303 mmol, 20%). H-NMR (300 MHz; CDCl3): δ 5.78 (dd,

J1,F =52.5,J1,2 =2.5Hz,1H,H-1),5.51(t,J2,3 =J3,4 =9.9Hz,1H,H-3),5.19

(t, J4,5 =9.9Hz,1H,H-4),4.94(ddd,J2,F =24.1Hz,1H,H-2),4.43(d,1H,H-5), 3.72 (s, 3H, OMe), 2.06 (s, 3H, OAc), 2.00 (s, 6H, 2xOAc) ppm. 13C-NMR (75

MHz; CDCl3): δ 169.89 (OAc), 169.79 (OAc), 169.48 (OAc), 166.99 (C-6), 103.61

(d, JC1,F =231.2Hz,C-1),69.97(d,JC3,F =4.4Hz,C-3),69.94(d,JC2,F =24.1 Hz, C-2), 68.68 (C-4), 68.54 (C-5), 53.13 (OMe), 20.63 (OAc),20.53(OAc),20.47

19 (OAc) ppm. F-NMR (282 MHz; CDCl3): δ -150.37 (dd, F-1) ppm. MS : Calcd.

for C13H17FNaO9:359.1;found:359.3

Methyl (2,3-di-O-acetyl-4-deoxy-β-l-threo-hex-4-enopyranosyl fluoride) uronate (72) COOMe O AcO AcOF

Globally protected α-d-glucuronyl fluoride (71,0.080g,0.238mmol)wassubjec- ted to the general method for DBU-catalysed elimination (page186).Purificationby

flash column chromatography (5:2 petroleum ether/ethyl acetate) gave a colourless

1 syrup (0.022 g, 0.080 mmol, 34%). H-NMR (300 MHz; CDCl3): δ 6.15 (d, J3,4 =

2.3 Hz, 1H, H-4), 5.88 (dd, J1,F =53.7,J1,2 =1.9Hz,1H,H-1),5.71(dd,J2,3 =8.9

Hz, 1H, H-3), 5.21 (ddd, J2,F =24.1Hz,1H,H-2),3.83(s,3H,OMe),2.14(s,3H,

13 OAc), 2.10 (s, 3H, OAc) ppm. C-NMR (75 MHz; CDCl3): δ 170.16 (OAc), 170.07

(OAc), 161.05 (C-6), 141.06 (C-5), 109.69 (C-4), 103.43 (d, JC1,F =232.7Hz,C-1),

68.39 (d, JC2,F =21.4Hz,C-2),65.64(d,JC3,F =4.6Hz,C-3),52.99(OMe),20.96

19 (OAc), 20.77 (OAc) ppm. F-NMR (282 MHz; CDCl3): δ -146.30 (dd, F-1) ppm.

MS : Calcd. for C11H13FNaO7:299.1;found:299.3

224 7.2. Synthesis

(4-deoxy-β-l-threo-hex-4-enopyranosyl fluoride)uronic acid (73) COOH O HO HOF

Globally protected unsaturated α-d-glucuronyl fluoride (72,22mg,0.080mmol) was subjected to the general method for Zemplén deprotection(page187).Theinter-

mediate was purified by flash column chromatography (9:1 dichloromethane/methan-

ol), then further deprotected by the general method for hydrolysis by lithium hy-

droxide (page 187). The final product was purified by flash column chromatography

(7:2:1 ethyl acetate/methanol/water) and lyophilised to a white powder (11.3 mg,

1 0.063 mmol, 79%). H-NMR (300 MHz; D2O): δ 5.92 (d, J3,4 =2.2Hz,1H,H-4),

5.85 (dd, J1,F =54.8,J1,2 =1.5Hz,1H,H-1),4.45(dt,J2,3 =8.7,J3,F =2.8Hz,1H,

13 H-3), 3.87 (ddd, J2,F =26.4Hz,1H,H-2)ppm. C-NMR (75 MHz; D2O): δ 169.41

(C-6), 109.84 (C-4), 105.39 (t, JC1,F =191.3Hz,C-1),69.72(d,JC2,F =21.7Hz,

19 C-2), 65.30 (d, JC3,F =5.2Hz,C-3)ppm. F-NMR (282 MHz; CDCl3): δ -147.46

(br. dd, F-1) ppm. HRMS : Calcd. for C6H6FO5:177.0199;found:177.0202

Methyl 3,4-di-O-acetyl-d-glucuronal (74)

MeOOC O AcO AcO

Globally protected α-d-glucuronyl bromide (3,3.488g,8.78mmol)wasdissolved in acetone (17.5 mL) and saturated monobasic sodium phosphate (35 mL), then zinc

dust (6.68 g) added and the reaction mixture stirred vigorously overnight at ambi-

ent temperature. The reaction slurry was then filtered through Celite, the solvent

evaporated in vacuo,andtheproductdissolvedinethylacetatebeforewashingwith

water, saturated NaHCO3 and brine. Purification by flash column chromatography

225 7.2. Synthesis

1 yielded an off-white solid (2.00 g, 7.75 mmol, 88%). H-NMR (300 MHz; CDCl3):

δ 6.57 (d, J1,2 =5.9Hz,1H,H-1),5.31(br.q,J3,4 =J4,5 =2.1,J2,4 =1.9Hz,

1H, H-4), 4.91 (td, J2,3 =5.6Hz,1H,H-2),4.87(br.dd,1H,H-3),4.74(dd,J3,5 =1.2Hz,1H,H-5),3.69(s,3H,OMe),2.01(s,3H,OAc),1.89(s,3H,OAc)ppm.

13 C-NMR (75 MHz; CDCl3): δ 169.32 (OAc), 169.09 (OAc), 167.05 (C-6), 146.29 (C-1), 97.09 (C-2), 72.17 (C-5), 67.25 (C-3), 62.43 (C-4), 52.21 (OMe), 20.74 (OAc),

20.68 (OAc) ppm. MS : Calcd. for C11H14NaO7:281.1;found:281.3

Methyl (2-deoxy-2-fluoro-1,3,4-tri-O-acetyl-α/β-d- glucopyranosid)uronate (75)

MeOOC O AcO AcO F OAc

Methyl 3,4-di-O-acetyl-d-glucuronal (74,1.75g,6.14mmol)wasdissolvedin

5:1 nitromethane/acetic acid (210 mL), then cooled to 0 °C. Selectfluor was added in portions, then the reaction mixture allowed to warm to ambient temperature while stirring over 4 days, then refluxed for 1.5hr. After cooling to ambient tem- perature, the solvents were evaporated in vacuo and the crude product dissolved

in dichloromethane for washing with water, saturated NaHCO3,wateragain,and brine. Purification by flash column chromatography (3:1, 2:1 then 3:2 petroleum

ether/ethyl acetate), with impure fractions being purified asecondtime(3:1petro-

leum ether/ethyl acetate), gave mixtures of alpha and beta gluco- (0.326 g, 0.969

mmol, 16 %, α/β 1:1.05) and manno- (0.481 g, crude, 23%, α/β 3:1) configured iso-

1 mers as colourless syrups. H-NMR (300 MHz; CDCl3): δ 6.37 (d, J1α,2α =3.9Hz,

1H, H-1α), 5.77 (dd, J1β,2β =7.8,J1β,F =3.5Hz,1H,H-1β), 5.49 (dt, J3α,F =12.1,

J2α,3α =J3α,4α =9.8Hz,1H,H-3α), 5.38 (dt, J3β,F =14.7,J2β,3β =J3β,4β =9.8Hz,

1H, H-3β), 5.06 (t, J4α,5α =J4β,5β =9.8Hz,1H,H-4α, β), 4.61 (ddd, J2α,F =48.3,

226 7.2. Synthesis

1H, H-2α), 4.38 (ddd, J2β,F =51.8,1H,H-2β), 4.28 (d, 1H, H-5α), 4.16 (d, 1H, H-5β), 3.63 (s, 6H, OMe α, β), 2.10 (s, 3H, OAc α), 1.99 (s, 6H, OAc α, β), 1.94 (s, 3H, OAc

19 α), 1.93 (s, 3H, OAc β), 1.93 (s, 3H, OAc β)ppm. F-NMR (282 MHz; CDCl3): δ

-201.25 (ddd, F-2β), -203.10 (dd, F-2α)ppm.MS:Calcd.forC11H14NaO7:281.1; found: 359.3

Methyl (2,4-dinitrophenyl 2-deoxy-2-fluoro-3,4-di-O-acetyl-β-d- glucopyranosid)uronate (76) MeOOC O AcO AcO O F NO2 O2N

Globally protected 2-deoxy-2-fluoro-glucuronic acid (75,326mg,0.969mmol)

was dissolved in dichloromethane (5 mL), cooled to 0 °C, then acetic anhydride (2

mL) and HBr in acetic acid (33% w/v, 10 mL) added. The reaction was allowed

to proceed at 4 °C overnight before being allowed to warm to ambient temperature

and stirred a further 24 hours. The solvent was subsequently evaporated in vacuo,

then the intermediate dissolved in dichloromethane and washed quickly with cold

water, saturated NaHCO3 and brine before drying over MgSO4 and again removing the solvent in vacuo.Dinitrophenol(439mg,2.46eq.)wasdriedthriceovertoluene

then dissolved in acetonitrile (15 mL) and added to the pale yellow brominated

intermediate along with Ag2O(1.00g,4.45eq.)andthereactionmixturestirred

vigorously at ambient temperature in the dark under an atmosphere of N2.After three days the reaction mixture was filtered through Celite, the solvent evaporated

in vacuo and the product dissolved in ethyl acetate and washed extensively with

sat. NaHCO3,waterandbrinebeforedryingoverMgSO4.Purificationbyflash column chromatography (5:2 then 2:1 petroleum ether/ethyl acetate) followed by

227 7.2. Synthesis crystallisation from toluene/hexanes yielded white plates(164mg,0.356mmol,37%

1 ! over two steps). H-NMR (300 MHz; CDCl3): δ 8.68 (d, J3!,5! =2.7Hz,1H,H-5), ! ! 8.41 (dd, J2!,3! =9.3Hz,1H,H-3), 7.51 (d, 1H, H-2 ), 5.68 (dd, J1,F =6.2,J1,2

=5.8Hz,1H,H-1),5.44-5.32(m,2H,H-3,4),4.71(dt,J2,F =47.8,J2,3 =5.8Hz,

1H, H-2), 4.41 (d, J4,5 =7.2Hz,1H,H-5),3.63(s,3H,OMe),2.09(s,3H,OAc),

13 2.04 (s, 3H, OAc) ppm. C-NMR (75 MHz; CDCl3): δ 169.58 (OAc), 169.44 (OAc), 166.65 (C-6), 153.46 (Ar), 141.96 (Ar), 139.71 (Ar), 128.84 (Ar), 121.56 (Ar), 117.71

(Ar), 97.06 (d, JC1,F =29.5Hz,C-1),86.94(d,JC2,F =188.7 Hz, C-2), 72.46 (C-5),

68.86 (d, JC3,F =23.1Hz,C-3),67.35(d,JC4,F =5.6 Hz, C-4), 53.05 (OMe), 20.49

19 (2xOAc) ppm. F-NMR (282 MHz; CDCl3): δ -197.40 (ddd, J3,F =12.2Hz,F-2)

ppm. MS : Calcd. for C17H17FN2NaO12:483.1;found:483.2

Methyl (2,4-dinitrophenyl 2,4-dideoxy-2-fluoro-3-O-acetyl-α-l-threo-hex- 4-enopyranosid)uronate (77)

COOMe O AcO O F NO2 O2N

Globally protected 2,4-dinitrophenyl 2-deoxy-2-fluoro-β-d-glucuronide (76,26 mg, 0.057 mmol) was subjected to the general method for DBU-catalysed elim-

ination (page 186), but reacted only 10 min at 0 °C. Purification by flash column

chromatography (3:1 petroleum ether/ethyl acetate) gave a colourless syrup (17 mg,

1 0.042 mmol, 74%). H-NMR (300 MHz; CDCl3): δ 8.71 (d, J3!,5! =2.6Hz,1H,H- ! ! ! 5 ), 8.49 (dd, J2!,3! =9.2Hz,1H,H-3), 7.76 (d, 1H, H-2 ), 6.37 (dd, J3,4 =4.8,J2,4

=1.0Hz,1H,H-4),6.15(d,J1,F =4.0Hz,1H,H-1),5.43(dd,J3,F =16.2Hz,1H,

H-3), 5.06 (dd, J2,F =43.2Hz,1H,H-2),3.81(s,3H,OMe),2.19(s,3H,OAc)ppm.

13 C-NMR (75 MHz; CDCl3): δ 170.17 (OAc), 161.22 (C-6), 152.96 (Ar), 142.40 (C-

228 7.2. Synthesis

5), 141.32 (Ar), 140.22 (Ar), 129.15 (Ar), 121.57 (Ar), 118.00 (Ar), 107.87 (C-4),

93.92 (d, JC1,F =34.8Hz,C-1),83.88(d,JC2,F =178.3Hz,C-2),61.89(d,JC3,F

19 =33.4Hz,C-3),53.02(OMe),20.75(OAc)ppm. F-NMR (282 MHz; CDCl3): δ

-196.59 (ddd, F-2) ppm. MS : Calcd. for C15H13FN2NaO10:423.1;found:423.3

2,4-Dinitrophenyl 2,4-dideoxy-2-fluoro-α-l-threo-hex-4- enopyranosiduronic acid (78)

COOH O HO O F NO2 O2N

Globally protected 2,4-dinitrophenyl unsaturated 2-deoxy-2-fluoro-β-d-glucuronide (77,17mg,0.042mmol)wasdeprotectedusingthegeneralmethodfor hydrolysis by lithium hydroxide (page 187). A significant proportion remained acetylated after quenching, so the crude partially deprotected product was subjected to the gen- eral method for Zemplén deprotection (page 187). Purification was by flash column chromatography (4:1 ethyl acetate/methanol) followed by 5 gC-18Sep-pak,eluting with water then 50% methanol. The solvent was evaporated in vacuo,theproduct dissolved in water and lyophilised to a slightly off-white powder (8 mg, 0.023 mmol,

1 ! 55%). H-NMR (300 MHz; D2O): δ 8.91 (d, J3!,5! =2.8Hz,1H,H-5), 8.60 (dd, ! ! J2!,3! =9.4Hz,1H,H-3), 7.84 (d, 1H, H-2 ), 6.40 (br. t, J3,4 =3.8,J2,4 =3.4Hz,

1H, H-4), 6.15 (d, J1,F =4.4Hz,1H,H-1),5.12(dbr.t,J2,F =45.1,J2,3 =3.1 Hz,

13 1H, H-2), 4.51 (d br. t, J3,F =17.9Hz,1H,H-3)ppm. C-NMR (101 MHz; D2O): δ 153.82 (C-6), 142.20 (Ar), 139.76 (Ar), 139.52 (Ar), 130.14 (Ar), 122.40 (Ar), 119.32

(Ar), 107.48 (C-4), 95.18 (d, JC1,F =33.4Hz,C-1),86.93(d,JC2,F =176.2Hz,C-2),

19 62.21 (d, JC3,F =29.2Hz,C-3)ppm. F-NMR (282 MHz; CDCl3): δ -196.3 (ddd,

F-2) ppm. HRMS : Calcd. for C12H8FN2O9:343.0214;found:343.0213

229 7.2. Synthesis

1,2,3,6-Tetra-O-benzoyl-4-deoxy-4-fluoro-α-d-glucopyranose (80) OBz O F BzO BzO OBz

(Dimethylamino)sulfur trifluoride (690 µL, 8 eq.) was added dropwise to a solu- tion of 1,2,3,6-Tetra-O-benzoyl-α-d-galactopyranose216,217 (79,0.516g,0.865mmol) in dichloromethane (10 mL) at -30 °C, the reaction mixture wasstirredfor5hours,

then the reaction was quenched with methanol. The reaction mixture was diluted

with dichloromethane then washed with ice cold water, saturated NaHCO3 and brine

before drying over MgSO4.Purificationbyflashcolumnchromatography(4:1pet- roleum ether/ethyl acetate) gave the product as a white foam (0.496 g, 0.829 mmol,

1 96%). H-NMR (300 MHz; CDCl3): δ 8.22-7.83 (m, 8H, Ar), 7.71-7.27 (m, 12H,

Ar), 6.81 (d, J1,2 =3.5Hz,1H,H-1),6.31(dt,J3,F =13.9,J2,3 =J3,4 =9.9Hz,1H,

H-3), 5.61 (dd, 1H, H-2), 4.98 (dt, J4,F =50.4,J4,5 =9.9Hz,1H,H-4),4.77-4.63

13 (m, 2H, H-5,6a), 4.55 (dd, J6a,6b =9.6,J5,6b =3.3Hz,1H,H-6b)ppm. C-NMR

(75 MHz; CDCl3): δ 166.16 (RCOOR), 165.80 (RCOOR), 165.49 (RCOOR), 164.38 (RCOOR), 134.15 (Ar), 133.73 (Ar), 133.68 (Ar), 133.46 (Ar),130.11(Ar),129.98

(Ar), 129.92 (Ar), 129.63 (Ar), 128.95 (Ar), 128.62 (Ar), 128.58 (Ar), 128.49 (Ar),

89.83 (C-1), 86.85 (d, JC4,F =188.0Hz,C-4),70.57(d,JC3,F =19.8Hz,C-3),70.09

19 (d, JC5,F =10.6Hz,C-5),69.88(d,JC2,F =4.9Hz,C-2),62.13(C-6)ppm. F-NMR

(282 MHz; CDCl3): δ -198.00 (dd, F-4) ppm. MS : Calcd. for C34H27FNaO9:621.2; found: 621.4

230 7.2. Synthesis

4-Nitrophenyl 2,3,6-tetra-O-benzoyl-4-fluoro-β-d-glucopyranoside (81)

OBz O F BzO O OBz NO2

Globally benzoylated 4-deoxy-4-fluoro-glucose (80,496mg,0.829mmol)was

dissolved in dichloromethane (2 mL), cooled to 0 °C, then acetic anhydride (100

µL) and HBr in acetic acid (33% w/v, 10 mL) added. The reaction was allowed

to proceed at 4 °C overnight. The reaction was quenched on ice/water, extracted

with four portions of dichloromethane and the pooled organicphasewashedwith

cold water, sat. NaHCO3,andbrinebeforedryingoverMgSO4 and removing the solvent in vacuo.Theintermediatebromideand4-nitrophenol(126mg,1.1eq.) were

dried twice over toluene then dissolved in acetonitrile (20 mL), Ag2O(620mg,3eq.) added, and the reaction mixture stirred vigorously overnight at ambient temperature

in the dark under an atmosphere of argon. The reaction mixturewassubsequently

filtered through Celite, the solvent evaporated in vacuo and the product purified

by flash column chromatography (6:1 petroleum ether/ethyl acetate) to afford white

1 powder (322 mg, 0.523 mmol, 63% over two steps). H-NMR (300 MHz; CDCl3):

δ 8.16-7.96 (m, 6H, OBz), 7.68-7.37 (m, 9H, OBz), 7.06 (d, J2!,3! =J5!,6! =9.2Hz, ! ! ! ! 2H, H-3 ,5 ), 6.94 (d, 2H, H-2 ,6 ), 5.99 (dt, J3,F =14.4,J2,3 =J3,4 =9.1Hz,1H,

H-3), 5.77 (dd, J1,2 =7.6Hz,1H,H-2),5.55(d,1H,H-1),4.94(dt,J4,F =50.6,J4,5

=9.1Hz,1H,H-4),4.85(br,t,J5,6a =12.1Hz,1H,H-5),4.65(dd,J6a,6b =6.4Hz, 13 1H, H-6a), 4.37 (td, J5,6b =2.9Hz,1H,H-6b)ppm. C-NMR (75 MHz; CDCl3): δ 166.19 (RCOOR), 165.76 (RCOOR), 165.25 (RCOOR), 162.25 (Ar), 161.00 (Ar), 143.28 (Ar), 133.85 (Ar), 130.01 (Ar), 129.94 (Ar), 129.77 (Ar), 128.71 (Ar), 128.65

(Ar), 128.59 (Ar), 126.30 (Ar), 125.78 (Ar), 116.86 (Ar), 115.74 (Ar), 98.21 (C-1),

87.25 (d, JC4,F =189.7Hz,C-4),72.78(d,JC5,F =20.1Hz,C-5),72.28(d,JC3,F

231 7.2. Synthesis

19 =23.8Hz,C-3),71.16(d,JC2,F =7.8Hz,C-2),62.52(C-6)ppm. F-NMR (282

MHz; CDCl3): δ -199.03 (dd, F-4) ppm. MS : Calcd. for C33H26FNNaO10:638.1; found: 638.5

4-Nitrophenyl 4-fluoro-β-d-glucopyranoside (82)

OH O F HO O OH NO2

Globally benzoylated 4-nitrophenyl 4-deoxy-4-fluoro-β-d-glucoside (81,288mg, 0.468 mmol) was subjected to the general method for Zemplén deprotection (page

187). The reaction was quenched with Sephadex ion exchange resin (H+ form) then

filtered. Purification by flash column chromatography (7% methanol in dichloro-

methane) gave a white powder (63 mg, 0.208 mmol, 44%). 1H-NMR(400 MHz; ! ! ! ! MeOD): δ 8.21 (d, J2!,3! =J5!,6! =9.2Hz,2H,H-3,5 ), 7.23 (d, 2H, H-2 ,6 ), 5.14

(d, J1,2 =7.8Hz,1H,H-1),4.36(dt,J4,F =50.8,J3,4 =J4,5 =9.2Hz,1H,H-4),

13 3.89-3.72 (m, 4H, H-3,5,6a,6b), 3.56 (dd, J2,3 =9.1Hz,1H,H-2)ppm. C-NMR (101 MHz; MeOD): δ 163.61 (Ar), 143.93 (Ar), 126.59 (Ar), 117.68 (Ar), 101.31

(C-1), 90.05 (d, JC4,F =181.0Hz,C-4),75.65(C-2),75.43(d,JC3,F =7.7Hz,C-3),

19 74.36 (d, JC5,F =8.4Hz,C-5),61.39(C-6)ppm. F-NMR (282 MHz; MeOD): δ

-201.66 (dd, J3,F =16.1Hz,F-4)ppm.MS:Calcd.forC12H14FNNaO7:326.1; found: 326.1

232 7.2. Synthesis

4-Nitrophenyl 4-fluoro-β-d-glucopyranosiduronic acid (83) HOOC O F HO O OH NO2

4-nitrophenyl 4-deoxy-4-fluoro-β-d-glucoside (82, 63 mg, 0.208 mmol), NaBr (15 mg, 0.7 eq.), and TEMPO (3 mg, 0.1 eq.) were dissolved in THF/water (1:2, 15

mL), cooled to 0 °C, then a solution of sodium hypochlorite (5%, 800 µL) added

slowly. Basicity of the reaction was maintained with 0.2 M NaOH. After 10 min,

the reaction was quenched with ethanol, acidified with Sephadex ion exchange resin

(H+ form), then filtered and the solvent evaporated in vacuo. Purification by 5 g

C-18 Sep-pak (eluting with 0%, 30%, 50%, and 100% methanol in water) followed by lyophilisation gave the product as a white powder (63 mg, 0.199 mmol, 95%).

1 ! ! H-NMR(400 MHz; MeOD): δ 8.19 (d, J2!,3! =J5!,6! =9.2Hz,2H,H-3,5 ), 7.22 (d, ! ! 2H, H-2 ,6 ), 5.24 (d, J1,2 =7.8Hz,1H,H-1),4.48(dt,J4,F =49.9,J3,4 =J4,5 =

9.2 Hz, 1H, H-4), 4.37 (dd, J5,F =4.1Hz,1H,H-5),3.85(dt,J3,F =16.0,J2,3 = 9.2 Hz, 1H, H-3), 3.61 (dd, 1H, H-2) ppm. 13C-NMR (101 MHz; MeOD): δ 170.67 (C-6), 163.23 (Ar), 144.07 (Ar), 126.61 (Ar), 117.65 (Ar), 101.11 (C-1), 91.50 (d,

JC4,F =185.5Hz,C-4),75.21(C-2),75.03(C-5),73.95(d,JC2,F =8.7Hz,C-3) ppm. 19F-NMR (282 MHz; MeOD): δ -200.57 (ddd, F-4) ppm. MS : Calcd. for

C12H11FNO8:316.1;found:316.3

233 7.2. Synthesis

Methyl (4-nitrophenyl 2,3-di-O-acetyl-4-fluoro-β-d- glucopyranosid)uronate (84)

MeOOC O F AcO O OAc NO2

4-nitrophenyl 4-deoxy-4-fluoro-β-d-glucopyranosiduronic acid (83,57mg,0.180 mmol) was dissolved in methanol (5 mL) and acetyl chloride (50µL)added.The reaction mixture was stirred for two hours then the solvent evaporated in vacuo

and replaced with acetic anhydride (5 mL) acidified with Sephadex ion exchange

resin (H+ form). After 2.5 hours the reaction mixture was filtered and the solvent

again evaporated in vacuo.Purificationbyflashcolumnchromatography(3:1pet-

roleum ether/ethyl acetate) yielded a white powder (63 mg, 0.152 mmol, 73%).

1 ! ! H-NMR(400 MHz; CDCl3): δ 8.20 (d, J2!,3! =J5!,6! =9.2Hz,2H,H-3,5 ), 7.08 (d, ! ! 2H, H-2 ,6 ), 5.44 (dt, J3,F =14.3,J2,3 =J3,4 =8.6Hz,1H,H-3),5.36(d,J1,2 =

6.9 Hz, 1H, H-1), 5.23 (dd, 1H, H-2), 4.93 (dt, J4,F =49.3,J4,5 =8.6Hz,1H,H-4),

4.36 (dd, J5,F =6.6Hz,1H,H-5),3.75(s,3H,Me),2.11(s,3H,OAc),2.06(s,3H,

19 OAc) ppm. F-NMR (282 MHz; CDCl3): δ -198.11 (ddd, F-4) ppm. MS : Calcd.

for C17H18FNNaO10:438.1;found:438.3

Methyl (4-nitrophenyl 2,3-di-O-acetyl-4-fluoro-5-bromo-β-d- glucopyranosid)uronate (85)

MeOOC O F AcO O Br OAc NO2

Globally protected methyl (4-nitrophenyl 4-deoxy-4-fluoro-β-d-glucopyranosid)-

uronate (84,63mg,0.152mmol),CaCO3 (72 mg), and bromine (100 µL, 15 eq.) were dissolved in carbon tetrachloride (10 mL) and irradiated with a 300 W light

234 7.2. Synthesis bulb. After one hour, the reaction mixture was diluted with dichloromethane and washed with water, 1 M Na2S2O3,saturatedNaHCO3,andbrinethendriedover

MgSO4.Purificationbyflashcolumnchromatography(4:1petroleumether/ethyl acetate) yielded a white foam (64 mg, 0.129 mmol, 85%). 1H-NMR (400 MHz; ! ! ! ! CDCl3): δ 8.25 (d, J2!,3! =J5!,6! =9.2Hz,2H,H-3,5 ), 7.15 (d, 2H, H-2 ,6 ), 5.80

(d, J1,2 =8.4Hz,1H,H-1),5.71(dt,J3,F =11.8,J2,3 =J3,4 =9.2Hz,1H,H-3),

5.37 (dd, 1H, H-2), 4.84 (dd, J4,F =48.2Hz,1H,H-4),3.92(s,3H,OMe),2.13(s,

19 3H, OAc), 2.09 (s, 3H, OAc) ppm. F-NMR (282 MHz; CDCl3): δ -185.92 (dd, F-4) ppm. MS : Calcd. for C17H17BrFNNaO10:516.0/518.0;found:516.2/518.2

7.2.7 Substrates for kinetic isotope effects

Methyl (1,5-anhydro-2,3,4-tri-O-acetyl-d-glucaropyran)uronate (90)

MeOOC O AcO AcO AcO O

Globally protected glucuronyl hemiacetal (7,0.748g,2.24mmol)wasdissolved in dichloromethane (25 mL), then Dess-Martin periodinane (1.432 g, 1.5 eq.) added and the reaction allowed to proceed overnight at ambient temperature. Following dilution with dichloromethane the product was washed with water, sat. NaHCO3 and brine, then dried over MgSO4.Purificationbyflashcolumnchromatography (2:1 petroleum ether/ethyl acetate) yielded a colourless syrup (0.620 g, 1.87 mmol,

1 83%). H-NMR (400 MHz; CDCl3): δ 5.48 (d, J4,5 =7.0Hz,1H,H-5),5.31(t,J2,3

=J3,4 =3.1Hz,1H,H-3),5.12(ddd,J2,4 =1.3Hz,1H,H-4),5.00(dd,1H,H-2), 3.80 (s, 3H, OMe), 2.08 (s, 3H, OAc), 2.05 (s, 3H, OAc), 1.98 (s,3H,OAc)ppm.

MS : Calcd. for C13H16NaO10:355.1;found:355.3

235 7.2. Synthesis

2 Methyl (2,3,4-tri-O-acetyl-1-{ H}-α-d-glucopyran)uronate (91)

MeOOC O AcO AcO D AcO OH

Globally protected glucuronic acid 1,5-lactone (90,0.620g,1.87mmol)inTHF

at 0 °C was slowly added an ice-cold solution of NaBD4 (40 mg, 0.956 mmol) in D2O (150µL), and the reaction then allowed to proceed with stirring for a further 5 min

before quenching with 300 µL acetic acid. The reaction mixture was then filtered

and the solvent evaporated in vacuo,followedbyco-evaporationwithmethanoltwice to remove boric acid. Purification by flash column chromatography (1:1 petroleum ether/ethyl acetate) yielded a white foam (0.364 g, 1.09 mmol, 58%). 1H-NMR (300

MHz; CDCl3): δ 5.59 (t, J2,3 =J3,4 =9.7Hz,1H,H-3),5.18(t,J4,5 =9.7Hz,1H, H-4), 4.90 (d, 1H, H-2), 4.60 (d, 1H, H-5), 3.75 (s, 3H, OMe), 2.09 (s, 3H, OAc),

2.05 (s, 3H, OAc), 2.04 (s, 3H, OAc) ppm. MS : Calcd. for C13H17DNaO10:358.1; found: 358.2

2 Methyl (1,2,3,4-tetra-O-acetyl-1-{ H}-α-d-glucopyranosid)uronate (92)

MeOOC O AcO AcO D AcO OAc

Globally protected 1-{2H}-glucuronyl hemiacetal (91,0.364g,1.09mmol)was

dissolved in dichloromethane (10 mL) then acetic anhydride (2.25 mL) followed by

tifluoroacetic acid (270 µL) added, and stirred at ambient temperature until TLC

indicated complete reaction (1:1 petroleum ether/ethyl acetate). The reaction was

then quenched with ice/water, diluted with dichloromethaneandwashedwithsat.

NaHCO3,andbrinethendriedoverMgSO4.Purificationbyflashcolumnchroma- tography (2:1 to 1:1 petroleum ether/ethyl acetate) yieldedawhitefoam(0.374g,

236 7.2. Synthesis

0.991 mmol, 91%). MS : Calcd. for C15H19DNaO11:400.1;found:400.3

2 Methyl (bromo 2,3,4-tri-O-acetyl-1-{ H}-α-d-glucopyranosid)uronate (93)

MeOOC O AcO AcO D AcO Br

Globally protected 1-{2H}-glucuronic acid (92,0.374g,0.991mmol)wasdis-

solved in dichloromethane (2.5 mL) then acetic anhydride (0.5 mL) and 33% HBr in

acetic acid (10 mL) were added at 0 °C. This was stirred at 4 °C until the reaction was

complete as judged by TLC (2:1 petroleum ether/ethyl acetate). The reaction was

quenched in ice/water, the aqueous layer extracted thrice with dichloromethane and

this pooled organic phase then extracted quickly with cold water, cold sat. NaHCO3

twice and brine, followed by drying over MgSO4.Purificationbyflashcolumnchro- matography (2:1 petroleum ether/ethyl acetate) yielded a colourless syrup that was

used immediately for the subsequent step (0.339 g, 0.852 mmol, 86%)

2 Methyl (2,4,6-trichlorophenyl 2,3,4-tri-O-acetyl-1-{ H}-α-d- glucopyranosid)uronate (94)

MeOOC O Cl AcO AcO O AcO D Cl Cl

2,4,6-Trichlorophenol (0.109 g, 0.564 mmol, 1.3 eq.) was reacted with glob-

ally protected 1-{2H}-glucuronyl bromide (93,0.173g,0.434mmol)bythegeneral

method for Koenigs-Knorr glycosylation (page 186). Purification by flash column

chromatography (2:1 petroleum ether/ethyl acetate) followed by recrystallisation

from ethanol yielded small white needles (0.112 g, 0.217 mmol, 50%). 1H-NMR (300

237 7.2. Synthesis

! ! MHz; CDCl3): δ 7.32 (s, 2H, H-3 ,5 ), 5.38-5.30 (m, 3H, H-2,3,4), 3.97-3.94 (m, 1H, H-5), 3.72 (s, 3H, OMe), 2.09 (s, 3H, OAc), 2.05 (s, 3H, OAc), 2.02 (s, 3H, OAc) ppm. MS : Calcd. for C19H18DCl3NaO10:536.0/538.0;found:536.1/539.1

2 Methyl (2,4,6-trichlorophenyl 2,3-di-O-acetyl-4-deoxy-1-{ H}-α-l-threo- hex-4-enopyranosid)uronate (95) COOMe Cl O AcO O Cl AcOD Cl

Globally protected 2,4,6-trichlorophenyl 1-{2H}-glucuronide (94,43mg,0.0835 mmol) was subjected to the general method for DBU-catalysed elimination (page

186). Purification by flash column chromatography (3:1 petroleum ether/ethyl acet-

1 ate) gave a white powder (32 mg, 0.070 mmol, 84%). H-NMR (300 MHz; CDCl3): ! ! δ 7.32 (s, 2H, H-3 ,5 ), 6.40 (dd, J3,4 =5.0,J2,4 =1.2Hz,1H,H-4),5.47(br.s,

1H, H-2), 5.23 (dd, J2,3 =1.2Hz,1H,H-3),3.83(s,3H,OMe),2.11(s,3H,OAc),

13 2.10 (s, 3H, OAc) ppm. C-NMR (75 MHz; CDCl3): δ 170.2 (OAc), 169.3 (OAc), 162.1 (C-6), 147.6 (C-5), 142.9 (Ar), 131.1 (Ar), 130.2 (Ar),129.1(Ar),106.9(C-4),

67.2 (C-3), 63.1 (C-2), 52.8 (OMe), 21.01 (OAc), 20.84 (OAc) ppm. MS : Calcd. for

C17H14DCl3NaO8:477.0/479.0;found:476.2/478.2

238 7.2. Synthesis

2 2,4,6-Trichlorophenyl 4-deoxy-1-{ H}-α-l-threo-hex-4- enopyranosiduronic acid (96) COOH Cl O HO O Cl HOD Cl

Globally protected 2,4,6-trichlorophenyl unsaturated 1-{2H}-glucuronide (95,32

mg, 0.070 mmol) was deprotected by the general method for ester saponification

(page 187). Purification was by 5 g C-18 Sep-pak, washed with water, 40% acetoni-

trile in water and 100% acetonitrile. All fractions containing product as determined

by TLC (3:3:1:1 toluene/ethyl acetate/methanol/acetic acid) were pooled and ly-

ophilised to give a white powder (16.7 mg, 0.047 mmol, 67%). Final deuterium

incorporation level measured to be 94.2% as determined by 1H-NMR integrals. 1H- ! ! NMR (400 MHz; CD3OD): δ 7.50 (s, 2H, H-3 ,5 ), 6.36 (dd, J3,4 =4.9,J2,4 =1.4

Hz, 1H, H-4), 4.30 (t, J2,3 =1.7Hz,1H,H-2),4.09(dd,1H,H-3)ppm.HRMS:

Calcd. for C12H8DCl3NaO6:377.9425;found:377.9423

2 Methyl (2,4,6-trichlorophenyl 2,3-di-O-acetyl-4-deoxy-4,5-{ H}-β-d- glucopyranosid)uronate (97)

MeOOC O Cl D AcO O D OAc Cl Cl

Globally protected 2,4,6-trichlorophenyl unsaturated β-d-glucuronide (59,97mg, 0.214 mmol) was dissolved in ethyl acetate and a suspension ofpalladiumoncarbon

(10% w/w) was added. The reaction vessel was purged with a vacuum aspirator,

flushed with D2 gas (three times each), then left sealed to react overnight. The catalyst was removed by filtration through Celite and the product purified by flash

239 7.2. Synthesis column chromatography (3:1 petroleum ether/ethyl acetate)togiveawhitepowder

(62 mg, 0.135 mmol, 63% C6 equatorial + 36 mg, 0.079 mmol, 37% C6axial;overall

1 ! ! quantitative). H-NMR (300 MHz; CDCl3): δ 7.31 (s, 2H, H-3 ,5 ), 5.29 (dd, J2,3 =

9.4, J1,2 =7.6Hz,1H,H-2),5.11(dd,J3,4 =11.6Hz,1H,H-3),5.11(d,1H,H-1), 3.74 (s, 3H, OMe), 2.11 (s, 3H, OAc), 2.07 (s, 3H, OAc), 1.93 (d,1H,H-4)ppm.

13 C-NMR (75 MHz; CDCl3): δ 170.4 (OAc), 169.7 (OAc), 168.6 (C-6), 148.29 (Ar), 130.9 (Ar), 130.5 (Ar), 129.12 (Ar), 101.5 (C-1), 72.01 (C-3), 70.1 (C-2), 60.53 (C-5),

52.7 (OMe), 21.0 (2xOAc), 14.35 (C-4) ppm. MS : Calcd. for C17H15D2Cl3NaO: 479.0/481.0; found: 479.2/481.2

Methyl (2,4,6-trichlorophenyl 2,3-di-O-acetyl-4-deoxy-5-bromo-4-{2H}-

β-d-glucopyranosid)uronate (98)

MeOOC O Cl D AcO O Br OAc Cl Cl

Methyl (2,4,6-trichlorophenyl 2,3-di-O-acetyl-4-deoxy-4,5-{2 H}-β-d-glucopyrano- sid)uronate (97,62mg,0.135mmol)wasdissolvedincarbontetrachloride,N - bromosuccinimide (29 mg, 1.2 eq.) and benzoyl peroxide (0.8 mg, 2.4 mol %) added, and the reaction mixture heated under reflux for 5 hours beforecoolingandconcen- trating. Purification by flash column chromatography (gradient 5:1 to 4:1 petroleum ether/ethyl acetate) gave a white powder (47 mg, 0.088 mmol, 65%). 1H-NMR (300 ! ! MHz; CDCl3): δ 7.33 (s, 2H, H-3 ,5 ), 5.59 (d, J1,2 =7.9Hz,1H,H-1),5.56(dd,

J3,4 =10.9,J2,3 =9.7Hz,1H,H-3),5.38(dd,1H,H-2),3.86(s,3H,OMe),2.32(d,

13 1H, H-4), 2.13 (s, 3H, OAc), 2.06 (s, 3H, OAc) ppm. C-NMR (75 MHz; CDCl3): δ 170.0 (OAc), 169.7 (OAc), 165.6 (C-6), 146.6 (Ar), 131.3 (Ar), 130.2 (Ar), 129.3 (Ar), 101.3 (C-1), 87.3 (C-5), 71.4 (C-3), 68.4 (C-2), 53.9 (OMe), 20.94 (C-4), 20.91

240 7.2. Synthesis

(2xOAc) ppm. MS : Calcd. for C17H15DBrCl3NaO8:555.9/557.9/559.9;found: 556.1/558.1/560.2

2 Methyl (2,4,6-trichlorophenyl 2,3-di-O-acetyl-4-deoxy-4-{ H}-α-l-threo- hex-4-enopyranosid)uronate (99)

COOMe D Cl O AcO O OAc Cl Cl

Methyl (2,4,6-trichlorophenyl 2,3-di-O-acetyl-4-deoxy-4-{2H}-5-bromo-β-d-gluco- pyranosid)uronate (98,29mg,54µmol)wassubjectedtothegeneralmethodfor

DBU-catalysed elimination (page 186). Purification by flash column chromatography

(4:1 petroleum ether/ethyl acetate) gave a white powder (15 mg, 33 µmol, 72%). 1H- ! ! NMR (300 MHz; CDCl3): δ 7.32 (s, 2H, H-3 ,5 ), 5.83 (dd, J1,2 =1.9,J1,3 =1.1

Hz, 1H, H-1), 5.49 (br. t, J2,3 =1.7Hz,1H,H-2),5.23(br.t,1H,H-3),3.83(s,

13 3H, OMe), 2.11 (s, 3H, OAc), 2.11 (s, 3H, OAc) ppm. C-NMR (75 MHz; CDCl3): δ 170.3 (OAc), 169.3 (OAc), 166.1 (C-6), 147.6 (C-5), 142.9 (Ar), 131.1 (Ar), 130.2 (Ar), 129.2 (Ar), 96.8 (C-1), 67.2 (C-3), 63.0 (C-2), 52.8 (OMe), 21.04 (OAc), 20.87

(OAc) ppm (carbon 4 signal too weak to detect, but seen in non-deuterated form).

MS : Calcd. for C17H14DCl3NaO8:476.0/478.0;found:476.1/478.2

241 7.2. Synthesis

2 2,4,6-Trichlorophenyl 4-deoxy-4-{ H}-α-l-threo-hex-4- enopyranosiduronic acid (100)

COOH D Cl O HO O OH Cl Cl

Globally protected 2,4,6-trichlorophenyl unsaturated 4-{2H}-β-d-glucuronide (99, 15 mg, 33 µmol) was deprotected by the general method for estersaponification(page

187). Purification was by 5 g C-18 Sep-pak, washed with water, 10% acetonitrile

in water, 40% acetonitrile in water and 100% acetonitrile. All fractions containing

product as determined by TLC (3:3:1:1 toluene/ethyl acetate/methanol/acetic acid)

were pooled and lyophilised to give a white powder (11 mg, 31 µmol, 54%). Fi-

nal deuterium incorporation level measured to be 90.5% as determined by 1H-NMR

1 ! ! integrals. H-NMR (400 MHz; CD3OD): δ 7.49 (s, 2H, H-3 ,5 ), 5.74 (dd, J1,2 =

2.3, J1,3 =0.9Hz,1H,H-1),4.30(t,J2,3 =2.3Hz,1H,H-2),4.08(dd,1H,H-

13 3) ppm. C-NMR (101 MHz; CD3OD): δ 172.19 (C-6), 149.57 (Ar), 131.72 (Ar), 131.44 (Ar), 130.07 (Ar), 102.25 (C-1), 69.98 (C-3), 65.90 (C-2) ppm (carbon 4 and

5signalstooweaktodetect,butseeninnon-deuteratedform). HRMS : Calcd. for

C12H8DCl3NaO6:377.9425; found: 377.9427

7.2.8 Potential inhibitors of UGL

2-Deoxy-2,3-didehydro-neuraminic acid (108)177 OH HO COOH O - + TFA . H3N HO HO

Protected 2-deoxy-2,3-didehydro-N -acetyl-neuraminic acid (104,36mg,0.076

242 7.2. Synthesis mmol) was deprotected by the method of Gervay et al.176,177 Final purification by

5gC8Sep-pak(elutedwithwaterin1mLfractions)andsubsequent lyophilisation yielded a white powder (21.9 mg, 60 µmol, 79% over 4 steps). NMRcharacterisation matched literature values.218

7.2.9 Potential trapping reagents for UGL

2-Keto-3-deoxy-3-fluoro-d-glycero-d-galactonononic acid (109)180 OH OH

O COOH HO HO HO HO F

β-Fluoropyruvic acid (0.504 g, 4.75 mmol) and D-Mannose (1.703 g, 2 eq.) were dissolved in water (45 mL), Neuraminic acid aldolase (36 mg) was added, and the reaction allowed to proceed at ambient temperature until 19F-NMR indicated con-

sumption of the fluoropyruvic acid (12 days, with a second 44 mgportionofaldolase

added after 4 days). The product was then filtered through a cotton plug and purified

by Dowex 1XB ion exchange resin (pre-equilibrated with 6 M formic acid then water)

eluting with water until no more mannose elutes, then 1 M formic acid to elute the

product. Fractions containing product were pooled and the solvent evaporated in

vacuo until the volume was sufficiently low for lyophilisation, which yielded a white

19 foam (1.07 g, 3.74 mmol, 79 %, 4.35:1 ax/eq F). F-NMR (282 MHz; H2O+D2O):

δ -199.85 (dd, JF,3 =50.0,JF,4 =13.3Hz,F-3eq), -207.09 (dd, JF,3 =48.5,JF,4 =

30.1 Hz, F-3ax)ppm.MS:Calcd.forC9H14FO9:285.1;found:285.3

243 7.2. Synthesis

Methyl 2,4,5,7,8,9-hexa-O-acetyl-3-deoxy-3-fluoro-d-glycero-d-galacto-2-

nonulopyranosonate (110)180

OAc OAc

O COOMe AcOAcO AcOAcO F

Asolutionof2-keto-3-deoxy-3-fluoro-d-glycero-d-galactonononic acid (109,1.07

g, 3.74 mmol) in methanol (13.4 mL) and trifluoroacetic acid (670 µL) was stirred

at ambient temperature for four hours. The solvents were thenevaporatedin vacuo,

the intermediate methyl ester dissolved in pyridine (13 mL) and acetic anhydride (4

mL), and allowed to react overnight at ambient temperature. Solvents were again

evaporated in vacuo,theproductdissolvedinethylacetate,andwashedwithwater,

saturated NaHCO3,1NHClandbrine.Purificationbyflashcolumnchromato- graphy (1:1 petroleum ether/ethyl acetate) yielded a white foam (1.19 g, 2.15 mmol,

1 58%). H-NMR (300 MHz; CDCl3): δ 5.39-5.27 (m, 3H, H-4,5,7), 5.16 (td, J7,8 =

J8,9b =5.8,J8,9a =2.1Hz,1H,H-8),4.96(dd,J3,F =48.7,J3,4 =2.0Hz,1H,H-3),

4.55 (dd, J9a,9b =12.5Hz,1H,H-9a), 4.17 (dd, 1H, H-9b), 4.12-4.08 (m, 1H, H-6), 3.84 (s, 3H, OMe), 2.18 (s, 3H, OAc), 2.12 (s, 3H, OAc), 2.09 (s,3H,OAc),2.04(s,

19 3H, OAc), 2.03 (s, 3H, OAc), 2.02 (s, 3H, OAc) ppm. F-NMR (282 MHz; CDCl3):

δ -208.19 (dd, JF,4 =29.4Hz,F-3)ppm.MS:Calcd.forC22H29FNaO15:575.1; found: 575.3

244 7.2. Synthesis

Methyl 4,5,7,8,9-penta-O-acetyl-3-deoxy-3-fluoro-d-glycero-d-galacto-2-

nonulopyranosonate (111)180 OAc OH

O COOMe AcOAcO AcOAcO F

Methyl 2,4,5,7,8,9-hexa-O-acetyl-3-deoxy-3-fluoro-d-glycero-d-galacto-2-nonulo- pyranosonate (110,0.960mg,1.74mmol)wasdissolvedinmethanol(15mL),hy- drazine acetate (0.331 g, 2 eq.) added at 0 °C and the reaction mixture stirred at ambient temperature for 1 hour. The solvent was evaporated in vacuo then the product dissolved in ethyl acetate and washed with water, 1 N HCl, and brine before drying over MgSO4.Purificationbyflashcolumnchromatography(3:2petroleum ether/ethyl acetate) yielded a white foam (0.620 g, 1.21 mmol, 70 %), while 85 mg

1 of unreacted starting material was recovered (9%). H-NMR (300 MHz; CDCl3): δ

5.45-5.15 (m, 4H, H-4,5,7,8), 4.91 (dd, J3,F =49.5,J3,4 =1.6Hz,1H,H-3),4.57

(dd, J9a,9b =11.9,J8,9a =1.3Hz,1H,H-9a), 4.27 (d, J5,6 =9.6Hz,1H,H-6),4.07

(dd, J8,9b =6.4Hz,1H,H-9b), 3.79 (s, 3H, OMe), 2.06 (s, 3H, OAc), 2.02 (s, 3H, OAc), 2.01 (s, 3H, OAc), 1.98 (s, 3H, OAc), 1.96 (s, 3H, OAc) ppm. 13C-NMR (75

MHz; CDCl3): δ 171.15 (OAc), 170.75 (OAc), 170.27 (br. s, 2xOAc), 169.60 (OAc),

167.48 (C-1), 94.43 (d, JC2,F =25.6Hz,C-2),87.73(d,JC3,F =184.9Hz,C-3),70.62

(C-6), 70.43 (d, JC4,F =16.9Hz,C-4),69.65(C-7),67.20(C-8),64.29(C-5),62.70 (C-9), 53.55 (OMe), 21.03 (OAc), 20.84 (OAc), 20.76 (OAc), 20.70 (br. s, 2xOAc)

19 ppm. F-NMR (282 MHz; CDCl3): δ -205.08 (dd, J4,F =26.8Hz,F-3)ppm.MS:

Calcd. for C20H27FNaO14:533.1;found:533.2

245 7.2. Synthesis

Methyl (fluoro 4,5,7,8,9-penta-O-acetyl-3-deoxy-3-difluoro-d-glycero-d-

galacto-2-nonulopyrano)sonate (112)180

OAc COOMe

O F AcOAcO AcOAcO F

(Dimethylamino)sulfur trifluoride (120 µL) was added dropwise to a solution of methyl 4,5,7,8,9-penta-O-acetyl-3-deoxy-3-fluoro-d-glycero-d-galacto-2-nonulopyran-

osonate (111,0.375g,0.735mmol)indichloromethane(10mL)at0°C,thereaction

mixture was stirred for 35 min, then the reaction was quenchedwithmethanol.The

reaction mixture was diluted with dichloromethane then washed with ice cold wa-

ter, saturated NaHCO3 and brine before drying over MgSO4.Purificationbyflash column chromatography (2:1 petroleum ether/ethyl acetate)gavetheproductasa

colourless syrup (0.164 g, 0.320 mmol, 44%), with the axial anomer also recovered

1 (0.129 g, 0.252 mmol, 34%). H-NMR (300 MHz; CDCl3): δ 5.31-4.98 (m, 5H, H- 2,3,4,7,8), 4.28-4.07 (m, 3H, H-6,9a,9b), 3.86 (s, 3H, OMe),2.06(br.s,6H,OAcx

2), 2.04 (s, 3H, OAc), 1.99 (s, 3H, OAc), 1.98 (s, 3H, OAc). 13C-NMR (75 MHz;

CDCl3): δ 170.55 (OAc), 169.90 (OAc), 169.78 (OAc), 169.60 (OAc), 169.16 (OAc), 164.22 (dd, J = 30.1, 4.1 Hz, C-1), 104.64 (dd, J = 227.7, 17.3 Hz, C-2), 85.80 (dd,

J=194.5,19.6Hz,C-3),72.04(d,J=4.3Hz,C-6),69.99(dd,J=16.6,5.5Hz,

C-4), 68.58 (C-7), 66.09 (C-8), 63.59 (d, J = 3.3 Hz, C-5), 61.73 (C-9), 53.98 (OMe),

20.98 (OAc), 20.77 (OAc), 20.64 (br. s, 2xOAc), 20.59 (OAc) ppm. 19F-NMR (282

MHz; CDCl3): δ -124.95 (d, JF2,F3 =10.5Hz,F-2),-216.13(ddd,J3,F3 =50.1,J4,F3

=24.2Hz,F-3)ppm.MS:Calcd.forC20H26F2NaO13:535.1;found:535.3

246 7.2. Synthesis

Methyl (fluoro 3-deoxy-3-difluoro-d-glycero-d-galacto-2-

nonulopyrano)sonate (113)180

OH COOMe

O F HO HO HO HO F

Methyl (fluoro 4,5,7,8,9-penta-O-acetyl-3-deoxy-3-difluoro-d-glycero-d-galacto-2-

nonulopyrano)sonate (112,55mg,0.107mmol)wassubjectedtothegeneralmethod

for Zemplén deprotection (page 187). Purification by flash column chromatography

(5:1 ethyl acetate/methanol) gave a colourless syrup (21 mg,0.069mmol,65%).

19 F-NMR (282 MHz; D2O): δ -122.61 (d, JF2,F3 =11.1Hz,F-2),-218.33(ddd,J3,F3

=50.3,J4,F3 =29.3Hz,F-3)ppm.MS:Calcd.forC10H16F2NaO8:325.1;found: 325.3

Fluoro 3-deoxy-3-difluoro-d-glycero-d-galacto-2-nonulopyranosonic

acid (114)180

OH COOH

O F HO HO HO HO F

To a solution of methyl (fluoro 3-deoxy-3-difluoro-d-glycero-d-galacto-2-nonulo- pyrano)sonate (113,21mg,0.069mmol)intetrahydrofuran(1.5mL)andwater

(5 mL) was added 1 N NaOH (150 µL), and the reaction allowed to proceed at ambient temperature for 10 min then quenched with Sephadex ion exchange resin

(H+ form). Following filtration and removal of the organic solvent in vacuo, the product was purified by 5 g C-18 Sep-pak eluted with water. Lyophilisation yielded

1 awhitefoam(15.4mg,0.054,78%). H-NMR (300 MHz; D2O): δ 5.19 (dt, J3,F3 =

19 50.9, J3,F2 =3.0Hz,1H,H-3),4.10-3.64(m,7H,H-4,5,6,7,8,9a,9b)ppm. F-NMR

247 7.2. Synthesis

(282 MHz; D2O): δ -122.27 (dd, JF2,F3 =11.1Hz,F-2),-217.57(ddd,J4,F3 =28.0

Hz, F-3) ppm. HRMS : Calcd. for C9H13F2O8:287.0578;found:287.0580

Phenacyl (2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4-enopyranosyl fluoride)uronate (116) O O O Ph O AcO F OAc

Phenacyl (fluoro 2,3,4-tri-O-acetyl-β-d-glucopyranosyl)uronate 181 (115,0.507g, 1.151 mmol) was subjected to the general method for DBU-catalysed elimination

(page 186). Purification by flash column chromatography (3:1 petroleum ether/ethyl

acetate) gave a colourless syrup (0.366 g, 0.962 mmol, 84%). 1H-NMR (400 MHz; ! ! CDCl3): δ 7.88 (d, J2!,3! =J5!,6! =7.4Hz,2H,H-2,6 ), 7.58 (t, J3!,4! =J4!,5! =7.4 ! ! ! Hz, 1H, H-4 ), 7.45 (t, 2H, H-3 ,5 ), 6.46 (d, J3,4 =4.9Hz,1H,H-4),5.93(d,J1,F

=49.6Hz,1H,H-1),5.52-5.43(m,2H,CH2C(O)Ar), 5.19-5.17 (m, 2H, H-2,3), 2.08

13 (s, 3H, Ar), 2.06 (s, 3H, Ar) ppm. C-NMR (101 MHz; CDCl3): δ 190.92 (C=O),

169.73 (OAc), 169.22 (OAc), 160.41 (C-6), 141.49 (d, JC5,F =1.4Hz,C-5),134.17

(Ar), 133.87 (Ar), 128.99 (Ar), 127.80 (Ar), 108.43 (C-4), 101.31 (d, JC1,F =228.3

Hz, C-1), 66.99 (CH2C(O)Ar), 66.14 (d, JC2,F =36.6Hz,C-2),62.22(C-3),20.71

19 (OAc), 20.59 (OAc) ppm. F-NMR (282 MHz; CDCl3): δ -140.41 (d, F-1) ppm.

MS : Calcd. for C18H17FNaO8:403.1;found:403.3

248 7.2. Synthesis

Phenacyl (2,3-di-O-acetyl-4-deoxy-β-d-glucopyranosyl fluoride) uronate (117) O Ph O O O AcO F OAc

Phenacyl (fluoro 2,3-di-O-acetyl-4-deoxy-α-l-threo-hex-4-enopyranosid)uronate (116,0.323g,0.849mmol)wasdissolvedinethylacetate(10mL)and a spatula tip

of palladium on carbon (10% catalyst) was added. The vessel was degassed under

vacuum then flushed with hydrogen, and allowed to react at ambient temperature

overnight. The crude intermediate was then filtered through Celite, the solvent evap-

orated in vacuo, and dissolved in fresh ethyl acetate (5 mL). 2-Bromoacetophenone

(208 mg, 1.2 eq.) and triethylamine (160 µL, 1.4 eq.) were added and the reaction

mixture stirred at ambient temperature for 3 hours. The reaction mixture was sub-

sequently diluted and washed with water, 1 N HCl, saturated NaHCO3,andbrine,

then dried over MgSO4 and purified by flash column chromatography (2:1 petroleum ether/ethyl acetate) and crystallisation to yield a white powder (0.157 g, 0.411, 48%

1 over two steps). H-NMR (300 MHz; acetone-d6): δ 8.02 (d, J2!,3! =J5!,6! =7.4Hz, ! ! ! ! ! 2H, H-2 ,6 ), 7.70 (t, J3!,4! =J4!,5! =7.4Hz,1H,H-4), 7.57 (t, 2H, H-3 ,5 ), 5.67 (d, ! ! ! ! J7 a,7 b =16.1Hz,1H,H-7a), 5.57 (d, 1H, H-7 b), 5.55 (dd, J1,F =52.2,J1,2 =6.2

Hz, 1H, H-1), 5.21 (ddd, J3,4ax =11.1,J2,3 =8.8,J3,4eq =5.3Hz,1H,H-3),4.98

(ddd, J2,F =11.7Hz,1H,H-2),4.79(dd,J4ax,5 =11.7,J4eq,5 =2.8Hz,1H,H-5),

2.60 (ddd, J4eq,4ax =13.2Hz,1H,H-4eq), 2.08 (s, 3H, OAc), 2.04 (ddd, 1H, H-4ax), 2.03 (s, 3H, OAc) ppm. 19F-NMR (282 MHz; acetone-d6): δ -139.95 (dd, F-1) ppm.

MS : Calcd. for C18H19FNaO8:405.1;found:405.3

249 7.2. Synthesis

Phenacyl (2,3-di-O-acetyl-4-deoxy-5-fluoro-α-l-idopyranosyl fluoride)uronate (118)

F O AcO F OAc O O O

Ph

N -Bromosuccinimide (0.274 g, 4 eq.) and phenacyl (fluoro 2,3-di-O-acetyl-4-

deoxy-β-d-glucopyranosyl)uronate (117,0.147g,0.384mmol)weredissolvedincar- bon tetrachloride (5 mL) and irradiated with a 300 W light bulb. After two hours,

the reaction mixture was diluted with dichloromethane and washed with water, sat-

urated NaHCO3,andbrinethendriedoverMgSO4.Purificationwasattemptedby column chromatography (4:1 to 3:1 petroleum ether/ethyl acetate) but the product

was found to have partially decomposed under these conditions and thus was not

completely pure; the partially purified intermediate was used immediately for the

next step. This intermediate (87 mg, 0.188 mmol) was dissolved in acetonitrile (2

mL) along with silver (I) fluoride (27 mg, 1.1 eq.), and stirredatambienttemperat-

ure overnight in the dark. On completion, the reaction mixture was filtered through

aplugofsilica(elutingwithethylacetate)andthenpurifiedbyflashcolumnchro-

matography (3:1 to 2:1 petroleum ether/ethyl acetate) to yield a colourless syrup

(60 mg, 0.150 mmol, 39% over 2 steps). This product was seen by 19F-NMR to

have contaminants remaining (approximately 10% by integrals) so crystallisation

was attempted from toluene/petroleum ether and from ethanol/water, with no suc-

cess. Further chromatography (5:2 petroleum ether/ethyl acetate, 1% methanol in

dichloromethane, 0.8% acetone in dichloromethane) also proved unsuccessful in re-

1 moving these contaminants. H-NMR (400 MHz; CDCl3): δ 7.90 (d, J2!,3! =J5!,6! = ! ! ! ! ! 7.4 Hz, 2H, H-2 ,6 ), 7.64 (t, J3!,4! =J4!,5! =7.4Hz,1H,H-4), 7.51 (t, 2H, H-3 ,5 ), ! ! 5.72 (dd, J1,F1 =48.3,J1,2 =0.9Hz,1H,H-1),5.55-5.47(m,2H,H-7a,7 b), 5.24-5.13

250 7.2. Synthesis

(m, 2H, H-2,3), 2.78 (ddd, J4ax,F5 =26.6,J4ax4eq =15.2,J3,4ax =4.2Hz,1H,H-4ax),

2.58 (ddd, J4eq,F5 =9.9,J3,4eq =5.6Hz,1H,H-4eq), 2.14 (s, 3H, OAc), 2.12 (s, 3H,

13 OAc) ppm. C-NMR (101 MHz; CDCl3): δ 190.33 (C=O), 170.00 (OAc), 169.32 (OAc), 164.72 (d, J = 33.1 Hz, C-6), 134.44 (Ar), 133.75 (Ar), 129.16 (Ar), 127.88

(Ar), 105.42 (d, JC5,F5 =239.3Hz,C-5),104.65(d,JC1,F1 =232.3Hz,C-1),67.53

(s, CH2C(O)Ar), 66.55 (d, JC2,F1 =35.3Hz,C-2),64.31(C-3),29.90(d,JC4,F5

19 =25.9Hz,C-4),21.00(OAc),20.82(OAc)ppm. F-NMR (282 MHz; CDCl3): δ

-102.64 (ddd, JF1,F5 =17.4Hz,F-5),-124.57(ddd,JF1,2 =5.6Hz,F-1),-140.41(d,

J=49.5Hz,unknowncontaminant)ppm.MS:Calcd.forC18H18F2NaO8:423.1; found: 423.3

(2,3-Di-O-acetyl-5-fluoro-4-deoxy-α-l-idopyranosyl fluoride)uronic acid (119) F O AcO F HOOC OAc

Phenacyl (fluoro 5-fluoro-2,3-di-O-acetyl-4-deoxy-α-l-idopyranosyl)uronate (118, 41 mg, 0.102 mmol) was dissolved in 8:1 methanol/water (4.5 mL) and a spatula tip

of palladium on carbon (10% catalyst) was added. The vessel was degassed under

vacuum then flushed with hydrogen, and allowed to react at ambient temperature

for 3 hours. The crude intermediate was then filtered through Celite, the solvent

evaporated in vacuo,andtheproductpurifiedbyflashcolumnchromatography(5

%methanolindichloromethane)togiveacolourlesssyrup(20mg,0.071mmol,

69%). The contaminant visible in 19F-NMR from the previous reaction remained.

1 H-NMR (300 MHz; CD3OD): δ 5.68 (d, J1,F1 =50.2Hz,1H,H-1),5.15-5.07(m,

2H, H-3,4), 2.63 (ddd, J4ax,F5 =21.7,J4ax,4eq =14.9,J3,4ax =5.9Hz,1H,H-4ax),

2.26 (ddd, J4eq,F5 =11.7,J3,4eq =6.1Hz,1H,H-4eq), 2.10 (s, 3H, OAc), 2.06 (s, 3H,

251 7.2. Synthesis

19 OAc) ppm. F-NMR (282 MHz; CD3OD): δ -100.68 (ddd, JF1,F5 =7.2Hz,F-5),

-124.89 (ddd, JF1,2 =14.9Hz,F-1),-141.55(d,J=50.3Hz,contaminant)ppm.

4-Deoxy-5-fluoro-α-l-idopyranosyl fluoride)uronic acid (120) F O HO F HOOC OH

Fluoro 2,3-di-O-acetyl-5-fluoro-4-deoxy-α-l-idopyranosyluronic acid (119,20mg, 0.071 mmol) was dissolved in dry methanol (5 mL) and ammonia gas bubbled

through until the solution was saturated. The reaction vessel was sealed and the

reaction mixture stirred at 4 °C over three days. The solvent and ammonia were

then evaporated in vacuo,andtheproductpurifiedby5gC-18Sep-pakelutedwith

water, passed through a plug of Sephadex ion exchange resin (H+ form), then by

flash column chromatography (17:2:1 then 7:2:1 ethyl acetate/methanol/water). The

product was observed to decompose on silica gel, but a single pure fraction was isol-

ated and lyophilised to give a white powder (1.7 mg, 8.5 µmol, 12%) with a further

1 4.5 mg of impure product recovered. H-NMR (400 MHz; D2O): δ 5.71 (dt, J1,F1 =

52.2, J1,2 =2.4Hz,1H,H-1),4.03(q,J2,3 =J3,4ax =J3,4eq =5.3Hz,1H,H-3),3.98

(ddd, J2,F1 =8.7Hz,1H,H-2),2.53(ddd,J4ax,F5 =31.0,J4ax,4eq =15.3Hz,1H,

19 H-4ax), 2.18 (ddd, J4eq,F5 =13.2Hz,1H,H-4eq)ppm. F-NMR (282 MHz; D2O):

δ -97.61 (ddd, JF1,F5 =16.3Hz,F-5),-124.72(ddd,F-1)ppm.HRMS:Calcd.for

C6H7F2O5:197.0262;found:197.0258

252 7.2. Synthesis

Methyl (2,3,4-tri-O-acetyl-1-fluoro-d-glucopyranosyl fluoride)

uronate (121) MeOOC O AcO AcO F AcO F and

Methyl (2,3,4-tri-O-acetyl-d-gluc-1-enopyranosyl fluoride)uronate (122) MeOOC O AcO AcO F AcO

Globally protected glucuronic acid 1,5-lactone (150,0.502g,1.511mmol)wasdis- solved in dichloromethane (15 mL), flushed with argon and cooled to -30 °C, then

(dimethylamino)sulfur trifluoride (0.29 mL, 2 eq.) added andthereactionmix- ture allowed to stir at ambient temperature for 4 days. A second equal portion of

(dimethylamino)sulfur trifluoride was added, and the reaction again allowed to pro- ceed at ambient temperature for 3 days before quenching with methanol at 0 °C.

The product was washed with saturated NaHCO3,waterandbrinebeforedrying over MgSO4.Partialpurificationbyflashcolumnchromatography(5:2petroleum ether/ethyl acetate) yielded the product 121 as a colourless syrup (0.088 g, 0.248

mmol, 16%) with the vinyl fluoride 151 as a minor contaminant (0.052 g, 0.156

mmol, 10%, by methyl peak integrals), along with 0.130 g (26%)startingmaterial

1 recovered. For 121, H-NMR (300 MHz; CDCl3): δ 5.40-5.24 (m, 3H, H-2,3,4), 4.39

(d, J4,5 =9.8Hz,1H,H-5),3.72(s,3H,OMe),2.06(s,3H,OAc),1.98(s,3H,OAc),

13 1.97 (s, 3H, OAc) ppm. C-NMR (75 MHz; CDCl3): δ 169.50 (OAc), 169.18 (OAc),

168.84 (OAc), 165.45 (C-6), 120.12 (dd, JC1,Fax =273.5,JC1,Feq =257.6Hz,C-1),

253 7.2. Synthesis

71.97 (dd, JC5,Fax =4.5,JC5,Feq =2.3Hz,C-5),69.98(t,JC3,Fax =JC3,Feq =9.0Hz,

C-3), 68.49 (t, JC2,Fax =JC2,Feq =30.0Hz,C-2),68.19(C-4),53.28(OMe),20.36

19 (OAc), 20.27 (OAc), 20.22 (OAc) ppm. F-NMR (282 MHz; CDCl3): δ -83.46 (d,

JFax,Feq =147.8Hz,F-1eq), -86.02 (dd, JFax,2 =15.8Hz,F-1ax)ppm.MS:Calcd. for C13H16F2NaO9:377.1;found:377.2

Methyl (2,3-di-O-acetyl-4-deoxy-1-fluoro-l-threo-hex-4-enopyranosyl

fluoride)uronate (123) COOMe O AcO F AcOF

Globally protected glucuronyl 1,1-difluoride (121,88mg,0.248mmol,contam- inated with globally protected 1-vinyl glucuronyl fluoride (122,52mg,0.156mmol) was subjected to the general method for DBU-catalysed elimination (page 186).

Purification by flash column chromatography (4:1 petroleum ether/ethyl acetate)

1 gave a white powder (53 mg, 0.180 mmol, 73%). H-NMR (300 MHz; CDCl3): δ

6.27 (d, J3,4 =3.6Hz,1H,H-4),5.53-5.43(m,2H,H-2,3),3.85(s,3H,OMe),2.14(s,

13 3H, OAc), 2.09 (s, 3H, OAc) ppm. C-NMR (101 MHz; CDCl3): δ 169.73 (OAc),

168.79 (OAc), 160.16 (C-6), 141.96 (C-5), 119.93 (t, JC1,Fax =JC1,Feq =264.2Hz,

C-1), 109.02 (C-4), 66.49 (t, JC2,Fax =JC2,Feq =31.4Hz,C-2),66.48(d,JC3,Feq = 8.9 Hz, C-3), 53.21 (OMe), 20.75 (OAc), 20.53 (OAc) ppm. 19F-NMR (282 MHz;

CDCl3): δ -83.81 (d, J = 158.6 Hz, F-1eq), -84.87 (ddd, JFeq,Fax =158.6,J2,Fax =

9.0, J3,Fax =4.5Hz,F-1ax)ppm.MS:Calcd.forC11H12F2NaO7:317.0;found: 317.3

254 7.2. Synthesis

(4-Deoxy-1-fluoro-l-threo-hex-4-enopyranosyl fluoride)uronic acid (124)

COOH O HO F HOF

Globally protected unsaturated glucuronyl 1,1-difluoride (123,17mg,0.058 mmol) was deprotected by the general method for acidic trans-esterification (page

187). The intermediate was purified by flash column chromatography (5 % methanol in dichloromethane) before further deprotection by the general method for hydrolysis by lithium hydroxide (page 187). The final product was purifiedbyflashcolumn chromatography (7:2:1 ethyl acetate/methanol/water) then5gC-18Sep-pak,elut- ing with water, and lyophilised to give a white powder (11.2 mg, 0.057 mmol, 98%).

1 H-NMR (400 MHz; D2O): δ 6.00 (d, J3,4 =3.2Hz,1H,H-4),4.45(dt,J2,3 =6.7,

J3,Fax =3.2Hz,1H,H-3),4.11(ddd,J2,Fax =11.6,J2,Feq =5.0Hz,1H,H-2)ppm.

13 C-NMR (101 MHz; D2O): δ 166.88 (C-6), 122.53 (dd, JC1,Fax =264.1,JC1,Feq =

260.0 Hz, C-1), 109.60 (C-4), 69.60 (t, JC2,Feq =JC2,Fax =27.4Hz,C-2),67.09(d,

19 JC3,Feq =3.2 Hz, C-3) ppm (C-5 signal too weak). F-NMR (282 MHz; D2O): δ

-86.07-85.94 (m, 2F, Fax,Feq)ppm.MS:Calcd.forC6H5F2O5:195.0;found:195.3

255 7.2. Synthesis

7.2.10 Glucuronides for extension to heparanase substrates.

Methyl (trifluoromethylumbelliferyl 2,3,4-tri-O-acetyl-β-d- glucopyranosid)uronate (140)

MeOOC O AcO AcO O OAc CF3

O O

Methyl (bromo 2,3,4-tri-O-acetyl-α-d-glucopyranosid)uronate (3,0.821g,2.067 mmol) was reacted with trifluoromethylumbelliferone (0.522g,1.1eq.)bythegen- eral method for Koenigs-Knorr glycosylation (page 186). Purification by flash column chromatography (2:1 petroleum ether/ethyl acetate) then crystallisation from eth- anol yielded a white powder (0.483 g, 0.884 mmol, 43%). 1H-NMR (300 MHz; ! ! ! CDCl3): δ 7.64 (d, J5!,6! =8.7Hz,1H,H-5), 7.00-6.97 (m, 2H, H-6 ,7 ), 6.67 (s, 1H, ! H-3 ), 5.45-5.29 (m, 4H, H-1,2,3,4), 4.27 (d, J4,5 =7.2Hz,1H,H-5),3.72(s,3H,

19 OMe), 2.04 (s, 9H, 3xOAc) ppm. F-NMR (282 MHz; CDCl3): δ -65.17 (s, CF3)

13 ppm. C-NMR (75 MHz; CDCl3): δ 170.07 (OAc), 169.41 (OAc), 169.21 (OAc), 166.64 (C-6), 159.93 (Ar), 158.95 (Ar), 155.86 (Ar), 126.73 (Ar), 114.84 (Ar), 113.94

(Ar), 113.90 (dAr), 113.78 (Ar), 109.13 (Ar), 104.75 (Ar), 98.11 (C-1), 72.74 (C-3),

71.58 (C-2), 70.85 (C-4), 68.88 (C-5), 53.21 (OMe), 20.68 (2xOAc), 20.59 (OAc) ppm. MS : Calcd. for C23H21F3NaO12:569.1;found:569.2

256 7.2. Synthesis

Trifluoromethylumbelliferyl β-d-glucopyranosiduronic acid (138)

HOOC O HO HO O OH CF3

O O

Globally protected trifluoromethylumbelliferyl unsaturated β-d-glucuronide(140, 30 mg, 54.9 µmol) was deprotected using the general methods for acidic trans- esterification followed by hydrolysis by lithium hydroxide (page 187). Purification by flash column chromatography (15:1 to 9:1 dichloromethane/methanol after trans- esterification and 7:2:1 ethyl acetate/methanol/water after methyl ester hydrolysis) yielded a white powder (5.3 mg, 13.0 µmol, 24% over two steps, with significant

1 losses to lactone hydrolysis on the second). H-NMR (400 MHz; CDCl3): δ 7.80 (d, ! ! ! ! J5!,6! =8.3Hz,1H,H-5), 7.20-7.18 (m, 2H, H-6 ,7 ), 6.84 (s, 1H, H-3 ), 5.28 (d, J1,2

=7.1Hz,1H,H-1),4.02(d,J4,5 =8.6Hz,1H,H-5),3.72-3.64(m,3H,H-2,3,4)ppm.

19 F-NMR (282 MHz; CDCl3): δ -65.21 (s, CF3)ppm.MS:Calcd.forC16H12F3O9: 405.0; found: 405.3

2,4-Dinitrophenyl 2-deoxy-2-fluoro-β-d-glucopyranosiduronic acid (139) HOOC O O2N HO HO O F NO2

Methyl (2,4-dinitrophenyl 2-deoxy-2-fluoro-3,4-di-O-acetyl-β-d-glucopyranosid)- uronate (76,18mg,39µmol)wasdeprotectedusingthegeneralmethodsforacidic trans-esterification followed by hydrolysis by lithium hydroxide (page 187). Puri-

fication was by 5 g C-18 Sep-pak, washed with water, 40% acetonitrile in water and 100% acetonitrile. All fractions determined by TLC (3:3:1:1 toluene/ethyl acet-

257 7.3. Biochemistry. ate/methanol/acetic acid) to contain pure product were pooled and lyophilised to give a white powder (10.6 mg, 29.3 µmol, 75% over 2 steps). 1H-NMR (400 MHz; ! ! D2O): δ 8.84 (d, J3!,5! =2.7Hz,1H,H-5), 8.48 (dd, J2!,3! =9.3Hz,1H,H-3), 7.58 ! (d, 1H, H-2 ), 5.77 (dd, J1,2 =7.4,J1,F =3.4Hz,H-1),4.62(ddd,J2,F =50.8,J2,3 =

9.3 Hz, H-2), 4.30 (d, J4,5 =9.3Hz,H-5),3.99(dt,J3,F =15.6,J3,4 =9.3Hz,H-3),

13 3.79 (t, H-4) ppm. C-NMR (101 MHz; D2O): δ 171.58 (C-6), 153.73 (Ar), 141.88

(Ar), 139.00 (Ar), 129.92 (Ar), 122.24 (Ar), 117.92 (Ar), 97.65 (d, JC1,F =25.5Hz,

C-1), 90.87 (d, JC2,F =186.4Hz,C-2),74.90(C-5),73.38(d,JC3,F =18.4Hz,C-3),

70.44 (d, JC4,F =9.1Hz,C-4)ppm.MS:Calcd.forC12H10FN2O10:361.0;found: 361.3

7.3 Biochemistry.

7.3.1 Cloning of UGL from Clostridium perfringens.

The gene for UGL from Clostridium perfringens strain ATCC13124 genomic DNA

(ATCC, Virginia) was amplified by PCR using Pwo ‘superyield’ polymerase in Pwo buffer (Roche, Switzerland) with 5 % DMSO and 3 mM MgCl2.Primersequences (Integrated DNA technologies, Illinois) were CACACAGCTAGCATGATTAAG-

GAAATAAGAGTTGAAGAGATTGC (forward) and CACACACTCGAGTTAC-

CAATAAAGGTTCCAATCTTTATAAAATCTTATTAAGG (reverse), showing re- striction sites (XhoI and NheI) in bold and start codon underlined. The thermocycler was run for 30 cycles of 30 s 95 °C melting, 30 s of 48 °C annealingand1minof72

°C extension, after which the product was gel purified and usedasatemplateforan- other round of PCR under the same conditions, except annealing at 51 °C and with a

final extension step of 7 min. The amplified gene was gel purified, digested with XhoI and NheI (Fermentas, Maryland) in NEB buffer 1 (New England Biolabs, Massachu-

258 7.3. Biochemistry. setts) and ligated in rapid ligation buffer using T4 ligase (Fermentas, Maryland) into pET28a(+), which had been digested with the same restrictionenzymes,andthe ligation product was again gel purified. This plasmid, termedpET28a::UGL,was transformed by electroporation in a Gene Pulser II (<5 ms, 25 µF, 200 Ω,2.5kV, 0.2 cm cuvette; Bio-Rad, California) into Escherichia coli BL21(DE3) which was plated on TYP kanamycin (50 μg.ml-1)andasinglecolonyselectedforovernight culture to store as a stock in 10% DMSO at -80 °C. Sequence was confirmed by commercial sequencing with T7 and T7terminal primers (NAPS unit, University of

British Columbia, Canada).

7.3.2 Testing of expression conditions

Expression was tested with variation in growth medium (TYP or LB), expression temperature (30 or 37 °C), concentration of IPTG (1 or 0.1 mM),andcelldensityat induction (exponential or stationary growth phase, low or high O.D. respectively).

Asetof8overnight3mLcultures,halfinTYPandhalfinLBmedia with 50

µg.mL-1 kanamycin, were subcultured 1 in 100 into a further set of 8 tubes with the same volume of fresh growth media and grown at 37 °C to an optical density of

0.5 (averaged across all tubes). Cells were induced with an appropriate amount of

IPTG then expression allowed to continue at the appropriate temperature for either

4hoursforhighO.D.inductionorovernightforlowO.D.induction. On completion of expression, 1.8 mL of each culture was pelleted at 13 200 rpmfor2min,the supernatant discarded, and the pellet frozen until all samples were ready. Cells were then resuspended in 300 µL of Bugbuster protein extraction reagent (Novagen,

Massachusetts) and incubated at ambient temperature for 20 min before centrifuging at 13 200 rpm for 20 min. The supernatant was separated from thepellet,towhich

20 µL of LDS loading buffer was added and then boiled for 10 min. Supernatant

259 7.3. Biochemistry. and partially-dissolved pellet were then analysed by SDS-PAGE.

7.3.3 Heterologous expression of UGL in Escherichia coli.

BL21(DE3) cells, transformed with pET28a::UGL, grown in an overnight culture

(1 mL TYP, 50 µg.mL-1 kanamycin) were sub-cultured 1 in 1000 into 500 mL TYP media with 50 µg.mL-1 kanamycin, and shaken (225 rpm) at 37 °C until middle to late log phase. The culture was then induced by adding IPTG (Invitrogen, California) to a final concentration of 100 µM and shaken overnight (225 rpm) at 37 °C. Cells were harvested by centrifugation at 3800 rcf, 4 °C, 25 min thenlysedbythreepasses of a 40 K manual fill French pressure cell at 1000 psi (SLM instruments, Illinois) in 15 mL lysis buffer (buffer A, below, with benzonase; Novagen,Germany,and

EDTA free protease inhibitor; Roche, Switzerland). The crude lysate was clarified by centrifugation at 1700 rcf, 4 °C, 30 min and filtered at 5.0 µmwithamillexSV sterile filter (Millipore, Massachusetts). Clarified supernatant was then purified by immobilised-metal affinity chromatography with a Qiagen NiNTA superflow 1 mL column on an Äkta 900 purifier as follows: The clarified supernatant was loaded using an auxiliary pump at 1 mL.min-1,andrinsedwith5mLbufferA.Thiswas then eluted stepwise: 20 mL buffer A, 15 mL 2.5% buffer B in A, 15 mL40%buffer

B in A and 5 mL buffer B, protein elution was monitored at 280 nm (buffer A: 20 mM Tris.HCl pH 8, 20 mM imidazole, 25 mM NaCl, 1 mM DTT, buffer B: 20mM

Tris.HCl pH 8, 400 mM imidazole, 25 mM NaCl, 1 mM DTT). Fractions confirmed to contain UGL by SDS-PAGE (4-20% in Tris/glycine buffer, Invitrogen, California) were pooled and concentrated then exchanged into Tris.HCl 20mMpH8,1mM

DTT in a 30 kDa cut offcentrifugal filter (final dilution estimated at 1 in 50 000;

Millipore, Massachusetts), then stored at 4 °C. Final protein purity was assessed by

SDS-PAGE and concentration determined by UV-vis absorbanceusinganε280 of

260 7.3. Biochemistry.

106 230 M.-1.cm.-1 calculated using the ProtParam tool from the ExPASy website

(http://web.expasy.org/protparam/).

7.3.4 Michaelis-Menten kinetics

Kinetic parameters were determined by measurement of initial rates in

MES.NaOH buffer 50 mM pH 6.6 with 1 mg.mL-1 BSA and 1 µM UGL. Extinc- tion coefficients were determined by allowing a 1 mM reaction toruntocompletion overnight, and wavelength was selected based on the largest difference in absorbance within the linear range of the spectrophotometer (total Abs≤3.5). Initial rates were measured at a range of at least five different substrate concentrations, ranging from at least Km/5 to 5 × Km (Km/7 to 7 × Km where possible), and the Michaelis- Menten equation (equation 7.1) fit by non-linear regression.Wherereactionswere linear over a sufficiently long time, several rates were measured simultaneously using an automated cell changer. To measure first order rates by substrate depletion, re- actions at low substrate concentrations ([S] ≤ Km/5)wereallowedtoproceedforat least 5 half lives and fit to a first order rate expression using the spectrophotometer’s proprietary software.

kcat[E][S] V0 = (7.1) Km +[S]

With the heparin- and chondroitin-derived natural substrates (11, 12, 13, 14,

15,and16), the axial phenyl substrate (43), Kdn2en (40), and the fluorinated

substrate (70)saturationkineticswerenotabletobeattained.Inthecaseofthe

axial phenyl substrate (43)thiswasaresultofapparentinhibitionathighercon-

centration giving a strong deviation from the curve, while with the non-chromogenic

substrates (11, 12, 13, 14, 15, 16, 40,and70)ratesforsubstrateconcentrations

over 5 mM could not be reliably measured due to high initial absorbance, since the

261 7.3. Biochemistry. reaction was monitored via decrease in absorbance from a highinitialpeak.The particularly poorly binding substrates heparan 6-sulfate disaccharide (14), Kdn2en

(40), and equatorial ΔGlcA fluoride (70)showedverylittledeviationfromlinear dependence of rate on substrate concentration in this range,sowasfitwithalinear model in GraFit 5.0 to determine kcat/Km,withKm approximated through meas- urement of Ki for the latter two (see section 7.3.11). Plots of rate againstsubstrate concentration are provided in Appendix D.

7.3.5 1H-NMR monitoring of UGL-catalysed reaction.

Hydrolysis of thiophenyl ΔGlcA (46,3.9mg)byUGL(14µM)wasmonitoredusing 1H-NMR in phosphate buffer (45 mM, pD 7.1 by direct pH meter measurement)

with β-mercaptoethanol (0.9 mM) and BSA (0.1 % w/v). Before addition of the enzyme to the substrate all locking, tuning, and shimming of the spectrometer was

performed on the enzyme/buffer/BSA mix, then the enzyme was added to the sub-

strate followed by fine-tuning of the shimming and recording of spectra after 0, 5, 10,

15, 45, 110, 170, and 266 minutes. The sample was then left at ambient temperature

over approximately 72 hours before a final spectrum was recorded, to allow time

for any slow equilibration. Each spectrum was recorded for 32scans.Aseparate

spectrum of the substrate in the same buffer conditions without enzyme was also

recorded.

7.3.6 Profile of UGL activity at varied pH.

Buffers at varied pH (3.0–9.0, with increments of 0.5) were madeusingcitricacid(0.1

M) and dibasic sodium phosphate (0.2 M), based on the table available from Sigma-

Aldrich (http://www.sigmaaldrich.com/life-science/core-bioreagents/biological-

buffers/learning-center/buffer-reference-center.html) with the exact pH determined

262 7.3. Biochemistry.

by pH meter. The pH-activity profile was determined by measuring kcat/Km at each pH by substrate depletion, as detailed earlier (section 7.3.4, page 261), using pNP

ΔGlcA (6,50µM)in50%buffer(above,finalconcentration50–100mM)with BSA (0.1 % w/v), and UGL at 1 µM.

The stability of UGL in each buffer was also tested by pre-incubation of the enzyme in the appropriate buffer for 20 min before 20× dilution into a solution of pNP ΔGlcA (6, 250 µM) in MES.NaOH (40 mM, pH 6.6) with BSA (0.1 % w/v) and the initial rate measured to determine the proportion of enzyme activity remaining.

7.3.7 Titration of benzyl ΔGlcA.

Benzyl ΔGlcA (53)wasdissolvedinthesamebuffersusedfortheenzymaticpH profile to a final concentration of 250 µM for a pH range from 2.7 to 7.2, then a scan taken of absorbance from 400–200 nm using the same buffer as a reference cell.

The main spectral changes were in the lowest end of this range,wheresubstantial noise was also seen. To control for this, scans were normalised for baseline using the averaged values from 340–340 nm then the ratio of the averagedabsorbanceforthe

220–200 nm range against the 235 nm peak was plotted against pHandfittoobtain pKa.

7.3.8 Effect of temperature on UGL.

Temperature effects on the UGL-catalysed reaction were determined using pNP

ΔGlcA (6)bymeasuringinitialratesatsaturatingsubstrateconditions (1.875 µM) for the effect on kcat,andsubstratedepletionatlowsubstrateconcentration(50µM) for kcat/Km,asdetailedearlier(section7.3.4,page261).Ratesweremeasured at 25–50 °C at 5 °C intervals in MES.NaOH buffer (40 mM, pH 6.6) containing BSA

(0.1% w/v) using 0.625 µM UGL. Stability of the enzyme at 50 °C was tested by

263 7.3. Biochemistry. incubating the enzyme at this temperature for 20 min, with a control kept on ice, before adding to 250 µM pNP ΔGlcA and measuring the initial rate.

7.3.9 Reaction of unsaturated glucuronides with UGL in D2O.

4-Nitrophenyl ΔGlcA (6,3.37mg,11.3µmol)wasdissolvedinD2O(500µLfinal), sodium phosphate buffer (100 mM final, pD 6.5), BSA (1 mg.mL-1 final). UGL stock was lyophilised thrice into D2Othenaddedtoafinalconcentrationof0.16 mg.mL-1/3.32 µM, and the reaction allowed to proceed at ambient temperature overnight. The reaction mixture was analysed directly by 1H-NMR with a water

2 suppression pulse sequence. The sample was then lyophilisedintoH2Ofor H-NMR.

7.3.10 Reaction of unsaturated glucuronides with UGL in 10% methanol.

4-Nitrophenyl or phenyl ΔGlcA (6, 10)(8.6mg,30µmol)wasdissolvedinwater, methanol (10 % v/v final), sodium phosphate buffer (50 mM final, pH 6.27), BSA

(1 mg.mL-1 final), UGL (0.051 mg.mL-1/1.05 µM final) to a total volume of 4 mL.

The reaction was allowed to proceed at ambient temperature for 3 hours, at which point TLC indicated the reaction was largely complete (3:3:1:1 toluene/ethyl acet- ate/methanol/acetic acid, substrate Rf = 0.4). Enzyme and BSA were removed by centrifugal filtration (30 kDa cutoff, 5 000 rcf, 4 °C, 10 min, reaction vessel and filter washed once with 2 mL H2O) then the flowthrough lyophilised and stored at -20 °C until ready for analysis. Reaction products were reconstituted in 700 µL D2Oand 1H-NMR, COSY and NOESY spectra recorded.

264 7.3. Biochemistry.

7.3.11 Testing of competitive inhibitors.

For those compounds which had been shown in other assays to actasasubstrate it was assumed that inhibition was competitive. With these compounds, inhibition was approximated by measurement of at least 5 rates with varied inhibitor and a

fixed substrate concentration at or slightly above Km.Plottingoftheinverseof these rates against the inhibitor concentration allowed determination of an intercept with the inverse of Vmax,definedas−1/Ki. All other compounds were assayed with a matrix of at least 4 different substrate and inhibitor concentrations (more where possible, but at times limited by com- pound amounts available), bracketing Km and a previously determined approximate

Ki (using the method outlined in the previous paragraph), respectively. These rates were then fit to modified Michaelis-Menten equations describing reaction in the pres- ence of competitive (equation 7.2), non-competitive (equation 7.3) and mixed type inhibition (equation 7.4) by non-linear regression. The equation giving the lowest errors was deemed to be the most appropriate, which was confirmed by plotting

1/rate against inhibitor concentration (a Dixon plot) and observing the intersection

−1 1 of all plots at X = Ki and Y = Vmax ,andonlythatvaluereported.

k [E][S] V = cat 0 [I] (7.2) Km(1 + Ki )+[S]

1 kcat[E][S] 1+[I]/Ki V0 = (7.3) Km +[S]

k [E][S] V = cat 0 [I] [I] (7.4) Km(1 + )+[S](1 + ! ) Ki Ki

265 7.3. Biochemistry.

7.3.12 Testing of compounds as mechanism-based inactivators.

Compounds designed as potential mechanism-based inactivators of UGL were tested for their ability to induce time-dependent loss of enzyme activity. Samples at varied inactivator concentration were incubated at 30 °C with enzyme (4 µM), using the buffer described for standard Michaelis-Menten kinetics, and aliquots were removed at timepoints to test for residual enzyme activity. For each such timepoint, 20 µL of inactivator mix was added to 180 µL of a pre-prepared substrate/buffer mix at a substrate concentration in a large excess over Km (pNP ΔGlcA, 6,1.875mM),and absorbance change was monitored for 1–2 minutes at 37 °C. Dataforeachinactivator

concentration were fit to an equation describing first order decay with offset. As no

clear inactivation at varied concentrations was observed, these first order rates were

not further processed. For all reactions where inactivationwasobserved,thepHwas

tested using a pH-fix strip (Macherey-Nagel, Germany) beforeandafterincubation.

For inactivator concentrations at or above the buffer concentration, the inactivator

was adjusted to pH 6.5 by mixing equimolar solutions of the inactivator as a free

acid and as its sodium salt.

7.3.13 Kinetic isotope effects.

Solvent kinetic isotope effect.

UGL was exchanged into D2Ousingrepeatedspinfiltrationthrougha30kDa cutoffspin filter at 4 °C, 4 000 rpm into 50 mM phosphate pD 7.1 with 1 mM

β-mercaptoethanol. The final volume was adjusted to match the initial volume, and

protein concentration was then determined by A280 to match the initial stock in

H2O. A stock of pNP ΔGlcA (152)wasalsopreparedinD2O. Reactions in H2O

were measured at pH 6.61 while those in D2OweremeasuredusingbufferatpD 7.05, using phosphate buffers (direct pH meter readings).

266 7.3. Biochemistry.

For isotope effects on kcat,initialratesweremeasuredandfittoalinearrate, with 8 replicates, using 2 mM substrate. Averaging of these linear rates for each substrate allowed calculation of a ratio and its standard error. For isotope effects on kcat/Km,thefirstorderrateconstantwasmeasuredbysubstratedepletion as detailed earlier (section 7.3.4, page 261) using 50 µM substrate. Reactions were monitored using a cell changer containing four reactions, with a total of 8 replicates for each substrate. These first order rate constants were alsoaveragedforeach substrate, allowing calculation of a ratio and its standard error.

Direct measurement of single isotope substitution effects.

Using the 1- and 4-deuterated substrates 96 and 100,aswellasthenon-deuterated form 49,ratesforisotopicallysubstitutedsubstratesweremeasured at high and low substrate concentrations relative to Km to determine isotope effects on kcat and kcat/Km.Measurementswerealternatedbetweensubstitutedandunsubstituted substrates using a single stock of each substrate and enzyme for all reactions to avoid bias. Stock solutions were pre-incubated at 37 °C between assays. In order to minimise pipetting errors, each stock was made to a concentration that allowed for mixing of large volumes relative to the total, typically 100 µL each of substrate and enzyme stock for a 200 µL reaction. Ratios of kH to kD were determined as detailed for the solvent kinetic isotope effects (with 5 mM substrate used for the effects on kcat).

Confirmation of effect from single isotope substitution at carbon 4 by competition in NMR.

An approximately 1:1 ratio of 4-1Hand4-2Hsubstrates(96 and 100)weredissolved in phosphate buffer 40 mM, pD 7.1 (direct pH meter reading) and BSA 1 mg.mL-1

267 7.3. Biochemistry.

all in D2Otoatotalsubstrateconcentrationof7.5mM.Thereactionwas started

1 by adding UGL stock to 2.4 μM(inH2O), and monitored by H-NMR. The fraction of reaction for the light isotope substrate in each spectrum was calculated from the integral of the proton signal at carbon 4 at 6.14 ppm over the sum of integrals of the aryl peaks at 7.53 ppm and 7.28 ppm, for starting material and product aryl groups respectively, all compared to that in the initial spectrum at time = 0. The corresponding ratio of substrate isotopomers was determined by the integral ratio of signals from protons at carbon 1 at 5.76 ppm and carbon 4 at 6.14 ppm for a given data point. The stated error represents the standard error from curve fitting by non-linear regression using equation 7.5, where F1 is the fraction of light isotope R substrate reacted, R0 is the isotope ratio at a given fraction of light isotope substrate reacted over that at t=0, and KIE is the kinetic isotope effect.

R − 1 − =10(log(1 F1)( KIE 1)) (7.5) R0

7.3.14 Attempted Rescue of D113G mutant with nucleophiles

UGLD113G (57 µM), purified as detailed earlier (section 7.3.3, page 260), was incub-

ated at ambient temperature in phosphate buffer (50 mM, ph 5.0) with BSA (0.1%

w/v), pNP ΔGlcA (6,10mM),andarangeofpotentialalternatenucleophiles.Nuc- leophiles tested were sodium formate (pH 5.0, 1 M), sodium acetate (pH 5.0, 0.5 M),

methanol (4 % v/v), β-mercaptoethanol (1 M), sodium azide (1 M), sodium cyanate (0.25 M), sodium thiocyanate (1 M), and potassium cyanide (1 M). Reactions were

monitored over a week using TLC (3:2:2 1-butanol/Acetic acid/water).

268 7.3. Biochemistry.

Testing in a spectrophotometer was carried out by monitoringofUGL-catalysed hydrolysis of 4-nitrophenyl ΔGlcA (6)throughreleaseof4-nitrophenolateinthe presence of sodium azide and sodium formate, pH 6.6, at a concentration of 100 mM. The enzyme was pre-incubated in the presence of azide/formate for 5 minutes before starting the reaction by addition of substrate to allow for the enzyme to equilibrate.

7.3.15 Heparanase kinetics

Arixtra

As a positive control for heparanase activity, the assay of Hammond et al. was used.209 Briefly, Arixtra (a kind gift from Prof. Jian Liu, University ofNorthCar-

olina, U.S.A.) at 100 µM (142)washydrolysedovernight(16–20hours)in40mM

sodium acetate buffer at pH 5.0 by 2 µg.mL-1 heparanase in a 100 µL reaction mix-

ture. The reaction was quenched with an equal volume of WST-1 (141,Toronto

Research Chemicals, Canada) at 1.69 mM in 0.1 M NaOH then developed for one

hour at 60 °C in a sealed 384 well plate followed by reading of A584.Astandard curve of galactose on the same plate allowed quantification ofthissignal.Thisas- say was also used on a shorter timeframe to monitor inhibitionandinactivationof other compounds. Linearity on a shorter timeframe was confirmed by taking ali- quots from a reaction over 20 minutes and plotting the concentration of reducing sugar equivalents against time.

Hydrolysis of trisaccharides with β-glucuronidase

Heparanase substrate trisaccharides (100–200 µM) were incubated with β-glucuron- idase from bovine liver (10 kU.mL-1)incitratebuffer(50mM,pH4.5)at37°C.

Reactions were monitored by HPLC using an analytic scale Zorbax SAX column

269 7.3. Biochemistry. eluting with a gradient of 0–100% 1 M potassium phosphate at pH6.6over60 min at 1 mL.min-1.Oncompletion,typicallyovernight,thereactionwasboiled for

5minutestoprecipitatetheenzyme,centrifuged,thenimmediately used to start heparanase digestion.

Fluorogenic substrate

Hydrolysis of fluorogenic substrates was assayed by reactionof100µMsubstrate overnight in 100 µL 40 mM sodium acetate pH 5.0 with 2 µg.mL-1 heparanase. Re-

actions with a trifluoromethylumbelliferyl leaving group were quenched by addition

of 100 µL Tris.NaOH pH 8.0 at 0.2 M then fluorescence read in a 384wellwith

excitation at 385 nm and emission at 502 nm. Reactions with a methylumbelliferyl

leaving group were quenched by addition of 100 µL glycine.NaOH pH 10.0 at 0.2 M

then fluorescence read in a 384 well with excitation at 355 nm and emission at 460

nm. A standard curve for each leaving group was prepared in thesamebuffers.For

comparison, certain substrates, as discussed in Subsection5.3.2,werealsomeasured

under conditions duplicating those of Pearson et al.,210 with pH 5.0 acetate buffer

at 60 mM, BSA at 0.1 mg.mL-1,substrateat5mM,andenzymeat1.2µg.mL-1 in

a50µLreaction,quenchingwith200µLofpH10.0glycineat0.2M.

Inactivation

Inactivation assaying of heparanase was carried out in a similar manner to UGL

(refer 7.3.12 on page 266), but at 37 °C. Inactivation was carried out in a 50 µL

reaction volume in sodium acetate buffer (40 mM, pH 5.0) with the DNP 2F 6SNAc

trisaccharide inactivator (149)at1mMandenzymeat20µg.mL-1 (10× the concen- tration used for substrate testing), along with a control reaction with no inactivator.

At each timepoint 5 µL of the inactivation reaction mixture was added to 45 µL of

270 7.3. Biochemistry.

Arixtra reaction mixture, in the same buffer and with the Arixtra substrate (142) at 250 µM, and reacted at 37 °C for 30 minutes before quenching with 30 µL of

0.2 M NaOH and freezing. After the final timepoint was taken, two sample Arixtra reactions with no enzyme and with and without inactivator were prepared as con- trols for a non-enzymatic background control in each case, and a standard curve of galactose prepared under the same conditions as the Arixtra reactions, then 30 µL of WST-1 at 3.38 mM was added to all timepoints and standard curve solutions and then incubated at 60 °C for one hour before reading A584 of 100 µL in a reduced- volume 384 well plate. The amount of Arixtra cleaved was quantified by subtracting the amount of reducing sugar equivalents in the relevant no enzyme background con- trol from that in each timepoint reaction, and these values plotted against time for the inactivation and no inactivator control, with non-linear regression fitting to first order decay with offset.

271 References

[1] Mathews, C. K.; van Holde, K. E.; Ahern, K. G. Biochemistry; Benjamin-

Cummings, 2000.

[2] Robinson, J. M. Geology 1990, 18,607–610.

[3] Robinson, J. Chemosphere 1996, 33,1641–1653.

[4] Taylor, T. N.; Osborn, J. M. Review of Palaeobotany and Palynology 1996,

90,249–262.

[5] Laine, R. A. Glycobiology 1994, 4,759–767.

[6] Gandhi, N. S.; Mancera, R. L. Chem Biol Drug Des 2008, 72,455–482.

[7] Kjellen, L.; Lindahl, U. Annual Review of Biochemistry 1991, 60,443–475.

[8] Silbert, J. E.; Sugumaran, G. IUBMB Life 2002, 54,177–186.

[9] Ernst, S.; Langer, R.; Cooney, C. L.; Sasisekharan, R. Critical Reviews in

Biochemistry and Molecular Biology 1995, 30,387–444.

[10] Schulz, T.; Schumacher, U.; Prehm, P. Journal of Biological Chemistry 2007,

282,20999–21004.

[11] Funderburgh, J. L. Glycobiology 2000, 10,951–958.

[12] Quantock, A. J.; Young, R. D.; Akama, T. O. Cellular and Molecular Life

Sciences 2010, 67,891–906.

272 [13] Santini, F.; Bisio, A.; Guerrini, M.; Yates, E. A. Carbohydrate Research 1997,

302,103–108.

[14] Chen, J.; Duncan, M. B.; Carrick, K.; Pope, R. M.; Liu, J. Glycobiology 2003,

13,785–794.

[15] Rudd, T. R.; Yates, E. A. Molecular Biosystems 2012, 8,1499–1506.

[16] Sheng, J.; Xu, Y.; Dulaney, S. B.; Huang, X.; Liu, J. Journal of Biological

Chemistry 2012, 287,20996–21002.

[17] Victor, X. V.; Nguyen, T. K. N.; Ethirajan, M.; Tran, V. M.; Nguyen, K. V.;

Kuberan, B. Journal of Biological Chemistry 2009, 284,25842–25853.

[18] Jansson, P.-E.; Kenne, L.; Lindberg, B. Carbohydrate Research 1975, 45,275–

282.

[19] Pollock, T. J. Journal of General Microbiology 1993, 139,1939–1945.

[20] Michel, G.; Tonon, T.; Scornet, D.; Cock, J. M.; Kloareg, B. New Phytologist

2010, 188,82–97.

[21] Percival, E. British Phycological Journal 1979, 14,103–117.

[22] Gacesa, P. Microbiology 1998, 144,1133–1143.

[23] Dettmar, P. W.; Strugala, V.; Richardson, J. C. Food Hydrocol loids 2011, 25,

263–266.

[24] Ridley, B.; O’Neill, M.; Mohnen, D. Phytochemistry 2001, 57,929–967.

[25] Kobayashi, M.; Matoh, T.; Azuma, J. Plant Physiology 1996, 110,1017–1020.

273 [26] Wolfenden, R. In Annual Review of Biochemistry;Kornberg,RDandRaetz,

CRH and Rothman, JE and Thorner, JW, Ed.; Annual Review of Biochem-

istry, Vol. 80, 2011; pp 645–667.

[27] Rye, C. S.; Withers, S. G. Current Opinion in Chemical Biology 2000, 4,

573–580.

[28] Withers, S. Carbohydrate Polymers 2001, 44,325–337.

[29] Vasella, A.; Davies, G. J.; Böhm, M. Current Opinion in Chemical Biology

2002, 6,619–629.

[30] Davies, G. J.; Gloster, T. M.; Henrissat, B. Current Opinion in Structural

Biology 2005, 15,637–645.

[31] Vocadlo, D. J.; Davies, G. J. Current Opinion in Chemical Biology 2008, 12,

539–555.

[32] Cantarel, B. L.; Coutinho, P. M.; Rancurel, C.; Bernard, T.; Lombard, V.;

Henrissat, B. Nucleic Acids Research 2009, 37,D233–D238.

[33] Koshland, D. E. Biological Reviews 1953, 28,416–436.

[34] Vuong, T. V.; Wilson, D. B. Biotechnology and Bioengineering 2010, 107,

195–205.

[35] Whitworth, G. E.; Macauley, M. S.; Stubbs, K. A.; Dennis,R.J.;Taylor,E.J.;

Davies, G. J.; Greig, I. R.; Vocadlo, D. J. Journal of the American Chemical

Society 2007, 129,635–644.

[36] Morley, T. J.; Willis, L. M.; Whitfield, C.; Wakarchuk, W.W.;Withers,S.G.

Journal of Biological Chemistry 2009, 284,17404–17410.

274 [37] Burmeister, W.; Cottaz, S.; Driguez, H.; Iori, R.; Palmieri, S.; Henrissat, B.

Structure 1997, 5,663–675.

[38] Burmeister, W.; Cottaz, S.; Rollin, P.; Vasella, A.; Henrissat, B. Journal of

Biological Chemistry 2000, 275,39385–39393.

[39] Vuong, T. V.; Wilson, D. B. FEBS Journal 2009, 276,3837–3845.

[40] Hehre, E.; Genghof, D.; Sternlicht, H.; Brewer, C. Biochemistry 1977, 16,

1780–1787.

[41] Hehre, E. J.; Kitahata, S.; Brewer, C. F. Journal of Biological Chemistry 1986,

261,2147–2153.

[42] Legler, G.; Roeser, K. R.; Illig, H. K. European Journal of Biochemistry 1979,

101,85–92.

[43] Legler, G. Advances in Carbohydrate Chemistry and Biochemistry 1990, 48,

319–384.

[44] Yip, V. L.; Withers, S. G. Current Opinion in Chemical Biology 2006, 10,

147–155.

[45] Yip, V. L. Y.; Thompson, J.; Withers, S. G. Biochemistry 2007, 46,9840–

9852.

[46] Yip, V. L. Y.; Withers, S. G. Angewandte Chemie-International Edition 2006,

45,6179–6182.

[47] Yip, V.; Varrot, A.; Davies, G.; Rajan, S.; Yang, X.; Thompson, J.; Anderson,

W.; Withers, S. Journal of the American Chemical Society 2004, 126,8354–

8355.

275 [48] Yip, V. L. Y.; Withers, S. G. Biochemistry 2006, 45,571–580.

[49] Chakladar, S.; Cheng, L.; Choi, M.; Liu, J.; Bennet, A. J. Biochemistry 2011,

50,4298–4308.

[50] Yip, V. L. Y.; Withers, S. G. Biochemistry 2012, 51,8464–8474.

[51] Huang, W.; Llano, J.; Gauld, J. W. Journal of Physical Chemistry B 2010,

114,11196–11206.

[52] Rajan, S.; Yang, X.; Collart, F.; Yip, V.; Withers, S.; Varrot, A.; Thompson,

J.; Davies, G.; Anderson, W. Structure 2004, 12,1619–1629.

[53] Liu, Q. P.; et al. Nature Biotechnology 2007, 25,454–464.

[54] Hall, B. G.; Pikis, A.; Thompson, J. Molecular Biology and Evolution 2009,

26,2487–2497.

[55] Burmeister, W.; Henrissat, B.; Bosso, C.; Cusack, S.; Ruigrok, R. Structure

1993, 1,19–26.

[56] Moustafa, I.; Connaris, H.; Taylor, M.; Zaitsev, V.; Wilson, J. C.; Kiefel,

M. J.; von Itzstein, M.; Taylor, G. Journal of Biological Chemistry 2004, 279,

40819–40826.

[57] Newstead, S. L.; Potter, J. A.; Wilson, J. C.; Xu, G.; Chien, C.-H.; Watts,

A. G.; Withers, S. G.; Taylor, G. L. Journal of Biological Chemistry 2008,

283,9080–9088.

[58] Xu, G.; Kiefel, M. J.; Wilson, J. C.; Andrew, P. W.; Oggioni, M. R.; Taylor,

G. L. Journal of the American Chemical Society 2011, 133,1718–1721.

[59] Flashner, M.; Kessler, J.; Tanenbaum, S. W. Archives of Biochemistry and

Biophysics 1983, 221,188–196.

276 [60] Parker, R. B.; McCombs, J. E.; Kohler, J. J. ACS Chemical Biology 2012, 7,

1509–1514.

[61] Yu, S. IUBMB Life 2008, 60,798–809.

[62] Yu, S.; Bojsen, K.; Svensson, B.; Marcussen, J. Biochimica et Biophysica Acta

(BBA) - Protein Structure and Molecular Enzymology 1999, 1433,1–15.

[63] Kim, Y.-M.; Saburi, W.; Yu, S.; Nakai, H.; Maneesan, J.; Kang, M.-S.; Chiba,

S.; Kim, D.; Okuyama, M.; Mori, H.; Kimura, A. Journal of Biological Chem-

istry 2012, 287,22441–22444.

[64] Lee, S. S.; Yu, S.; Withers, S. G. Journal of the American Chemical Society

2002, 124,4948–4949.

[65] Lee, S.; Yu, S.; Withers, S. Biochemistry 2003, 42,13081–13090.

[66] Lovering, A.; Lee, S.; Kim, Y.; Withers, S.; Strynadka, N. Journal of Biological

Chemistry 2005, 280,2105–2115.

[67] Smar, M.; Short, S.; Wolfenden, R. Biochemistry 1991, 30,7908–7912.

[68] Díaz, A.; Díaz-Lobo, M.; Grados, E.; Guinovart, J. J.; Fita, I.; Ferrer, J. C.

IUBMB Life 2012, 64,649–658.

[69] Gacesa, P. FEBS Letters 1987, 212,199–202.

[70] Sutherland, I. W. FEMS Microbiology Reviews 1995, 16,323–347.

[71] Rye, C. S.; Withers, S. G. Journal of the American Chemical Society 2002,

124,9756–9767.

[72] Rye, C. S.; Withers, S. G. Journal of Organic Chemistry 2002, 67,4505–4512.

277 [73] Rye, C.; Withers, S. Analytical Biochemistry 2002, 308,77–82.

[74] Rye, C.; Matte, A.; Cygler, M.; Withers, S. Chembiochem 2006, 7,631–637.

[75] Rye, C. S.; Withers, S. G. Carbohydrate Research 2004, 339,699–703.

[76] Garron, M.-L.; Cygler, M. Glycobiology 2010, 20,1547–1573.

[77] Charnock, S. J.; Brown, I. E.; Turkenburg, J. P.; Black, G. W.; Davies, G. J.

Proceedings of the National Academy of Sciences 2002, 99,12067–12072.

[78] Scavetta, R.; Herron, S.; Hotchkiss, A.; Kita, N.; Keen, N.; Benen, J.; Kester,

H.; Visser, J.; Jurnak, F. Plant Cell 1999, 11,1081–1092.

[79] Zheng, M.; Zhang, H.; Xu, D. The Journal of Physical Chemistry B 2012,

116,11166–11172.

[80] Shaya, D.; Hahn, B.-S.; Bjerkan, T. M.; Kim, W. S.; Park, N. Y.; Sim, J.-S.;

Kim, Y.-S.; Cygler, M. Glycobiology 2008, 18,270–277.

[81] Nankai, H.; Hashimoto, W.; Miki, H.; Kawai, S.; Murata, K. Applied and

Environmental Microbiology 1999, 65,2520–2526.

[82] Myette, J. R.; Shriver, Z.; Kiziltepe, T.; McLean, M. W.;Venkataraman,G.;

Sasisekharan, R. Biochemistry 2002, 41,7424–7434.

[83] Mori, S.; Akao, S.; Nankai, H.; Hashimoto, W.; Mikami, B.;Murata,K.Protein

Expression and Purification 2003, 29,77–84.

[84] Itoh, T.; Ochiai, A.; Mikami, B.; Hashimoto, W.; Murata, K. Journal of Mo-

lecular Biology 2006, 360,573–585.

[85] Maruyama, Y.; Nakamichi, Y.; Itoh, T.; Mikami, B.; Hashimoto, W.; Murata,

K. Journal of Biological Chemistry 2009, 284,18059–18069.

278 [86] Hashimoto, W.; Kobayashi, E.; Nankai, H.; Sato, N.; Miya, T.; Kawai, S.;

Murata, K. Archives of Biochemistry and Biophysics 1999, 368,367–374.

[87] Itoh, T.; Hashimoto, W.; Mikami, B.; Murata, K. Journal of Biological Chem-

istry 2006, 281,29807–29816.

[88] Richard, P.; Hilditch, S. Applied Microbiology and Biotechnology 2009, 82,

597–604.

[89] Marion, C.; Stewart, J. M.; Tazi, M. F.; Burnaugh, A. M.; Linke, C. M.;

Woodiga, S. A.; King, S. J. Infection and Immunity 2012, 80,1390–1398.

[90] Hashimoto, W.; Itoh, T.; Maruyama, Y.; Mikami, B.; Murata, K. International

Microbiology 2007, 10,233–243.

[91] Mori, S.; Akao, S.; Miyake, O.; Nankai, H.; Hashimoto, W.; Mikami, B.;

Murata, K. Acta Crystallographica Section D-Biological Crystallography 2003,

59,946–949.

[92] Itoh, T.; Akao, S.; Hashimoto, W.; Mikami, B.; Murata, K. Journal of Biolo-

gical Chemistry 2004, 279,31804–31812.

[93] Itoh, T.; Hashimoto, W.; Mikami, B.; Murata, K. Biochemical and Biophysical

Research Communications 2006, 344,253–262.

[94] Nakamichi, Y.; Maruyama, Y.; Mikami, B.; Hashimoto, W.; Murata, K.

Journal of Biological Chemistry 2011, 286,6262–6271.

[95] Itoh, T.; Ochiai, A.; Mikami, B.; Hashimoto, W.; Murata, K. Biochemical and

Biophysical Research Communications 2006, 347,1021–1029.

[96] Jaeger, T.; Arsic, M.; Mayer, C. Journal of Biological Chemistry 2005, 280,

30100–30106.

279 [97] Uehara, T.; Suefuji, K.; Jaeger, T.; Mayer, C.; Park, J. T. Journal of Bacteri-

ology 2006, 188,1660–1662.

[98] Veiga-da Cunha, M.; Sokolova, T.; Opperdoes, F.; Van Schaftingen, E. Bio-

chemical Journal 2009, 423,323–332.

[99] Hadi, T.; Dahl, U.; Mayer, C.; Tanner, M. E. Biochemistry 2008, 47,11547–

11558.

[100] Chiku, K.; Nishimoto, M.; Kitaoka, M. Carbohydrate Research 2010, 345,

1901–1908.

[101] Tweten, R. K. Veterinary Microbiology 2001, 82,1–9.

[102] Finsterer, J.; Hess, B. Infection 2007, 35,396–405.

[103] Boyd, S. D.; Mobley, B. C.; Regula, D. P.; Arber, D. A. International Journal

of Laboratory Hematology 2009, 31,364–367.

[104] Ototani, N.; Yosizawa, Z. Carbohydrate Research 1987, 159,25–36.

[105] Adorjan, I.; Jääskeläinen, A.-S.; Vuorinen, T. Carbohydrate Research 2006,

341,2439–2443.

[106] Tot, I.; Krainz, K.; Potthast, A.; Suess, U.; Rosenau, T. Macromolecular

Symposia 2008, 262,182–189.

[107] Azoulay, M.; Ivastchenko, N.; Florent, J.-C.; Monneret, C. Bioorganic and

Medicinal Chemistry letters 1997, 7,1307–1310.

[108] Fehlhaber, H. W.; Snatzke, G.; Vlahov, I. Liebigs Annalen der Chemie 1987,

637–8.

280 [109] Schneider, L. A.; Korber, A.; Grabbe, S.; Dissemond, J. Archives Of Derma-

tological Research 2007, 298,413–420.

[110] Thomas, T. M.; Scopes, R. K. Biochem. J. 1998, 330,1087–1095.

[111] Peterson, M. E.; Daniel, R. M.; Danson, M. J.; Eisenthal, R. Biochem J 2007,

402,331–337.

[112] Shoichet, B. K.; Baase, W. A.; Kuroki, R.; Matthews, B. W. Proceedings of

the National Academy of Sciences 1995, 92,452–456.

[113] Fields, P. A. Comparative Biochemistry and Physiology Part A: Molecular &

Integrative Physiology 2001, 129,417–431.

[114] Edgar, R. C. Nucleic Acids Research 2004, 32,1792–1797.

[115] Mann, M. C.; Thomson, R. J.; von Itzstein, M. Bioorganic & Medicinal Chem-

istry Letters 2004, 14,5555–5558.

[116] Mann, M. C.; Thomson, R. J.; Dyason, J. C.; McAtamney, S.; von Itzstein,

M. Bioorganic & Medicinal Chemistry 2006, 14,1518–1537.

[117] Schreiner, E.; Zbiral, E.; Kleineidam, R. G.; Schauer,R.Carbohydrate Research

1992, 216,61–66.

[118] Watts, A. G.; Withers, S. G. Canadian Journal of Chemistry 2004, 82,1581–

1588.

[119] Chang, C.-W.; Norsikian, S.; Guillot, R.; Beau, J.-M. European Journal of

Organic Chemistry 2010, 2010,2280–2294.

[120] Kok, G. B.; Mackey, B. L.; von Itzstein, M. Carbohydrate Research 1996, 289,

67–75.

281 [121] Pikis, A.; Immel, S.; Robrish, S. A.; Thompson, J. Microbiology-Sgm 2002,

148,843–852;Part3.

[122] Tosin, M.; Murphy, P. V. Organic Letters 2002, 4,3675–3678.

[123] Finch, P. Carbohydrates: Structures, Syntheses and Dynamics;Springer,1999.

[124] Friebolin, H.; Kunzelmann, P.; Supp, M.; Brossmer, R.; Keilich, G.; Ziegler,

D. Tetrahedron Letters 1981, 22,1383–1386.

[125] Telford, J. C.; Yeung, J. H. F.; Xu, G.; Kiefel, M. J.; Watts, A. G.; Hader,

S.; Chan, J.; Bennet, A. J.; Moore, M. M.; Taylor, G. L. Journal of Biological

Chemistry 2011, 286,10783–10792.

[126] Capon, B.; Walker, R. Journal of the Chemical Society — Perkin Transactions

2 1974,1600–1610.

[127] Hammett, L. P. Journal of the American Chemical Society 1937, 59,96–103.

[128] Tull, D.; Withers, S. Biochemistry 1994, 33,6363–6370.

[129] Bravman, T.; Zolotnitsky, G.; Belakhov, V.; Shoham, G.; Henrissat, B.;

Baasov, T.; Shoham, Y. Biochemistry 2003, 42,10528–10536.

[130] Vallmitjana, M.; Ferrer-Navano, M.; Planell, R.; Abel, M.; Ausin, C.; Querol,

E.; Planas, A.; Perez-Pons, J. Biochemistry 2001, 40,5975–5982.

[131] Bauer, M.; Kelly, R. Biochemistry 1998, 37,17170–17178.

[132] Cogoli, A.; Semenza, G. Journal of Biological Chemistry 1975, 250,7802–

7809.

282 [133] Comfort, D. A.; Bobrov, K. S.; Ivanen, D. R.; Shabalin, K.A.;Harris,J.M.;

Kulminskaya, A. A.; Brumer, H.; Kelly, R. M. Biochemistry 2007, 46,3319–

3330.

[134] Williams, S. J.; Withers, S. G. Carbohydrate Research 2000, 327,27–46.

[135] Wilkinson, S. M.; Liew, C. W.; Mackay, J. P.; Salleh, H. M.; Withers, S. G.;

McLeod, M. D. Organic Letters 2008, 10,1585–1588.

[136] Withers, S. G.; Rupitz, K.; Street, I. P. Journal of Biological Chemistry 1988,

263,7929–7932.

[137] Bucher, C.; Gilmour, R. Angewandte Chemie-International Edition 2010, 49,

8724–8728.

[138] Bucher, C.; Gilmour, R. Synlett 2011,1043–1046.

[139] Durantie, E.; Bucher, C.; Gilmour, R. Chemistry — A European journal 2012,

18,8208–8215.

[140] Withers, S. G.; MacLennan, D. J.; Street, I. P. Carbohydrate Research 1986,

154,127–144.

[141] McCarter, J. D.; Adam, M. J.; Withers, S. G. Biochemical Journal 1992, 286,

721–727.

[142] Namchuk, M. N.; McCarter, J. D.; Becalski, A.; Andrews, T.; Withers, S. G.

Journal of the American Chemical Society 2000, 122,1270–1277.

[143] Wei, G.; Kumar, V.; Xue, J.; Locke, R. D.; Matta, K. L. Tetrahedron Letters

2009, 50,5920–5922.

[144] Samuel, J.; Taylor, N. F. Carbohydrate Research 1984, 133,168–72.

283 [145] Esmurziev, A. M.; Reimers, A.; Andreassen, T.; Simic, N.; Sundby, E.; Hoff,

H. B. Molecules 2012, 17,820–835.

[146] Baldwin, J.; Kelly, D. Journal of the chemical society - chemical communica-

tions 1985.

[147] Cossy, J.; Ranaivosata, J.; Bellosta, V. Tetrahedron Letters 1994, 35,8161–

8162.

[148] Fersht, A. Enzyme structure and mechanism;W.H.Freeman&company,1984.

[149] Cooper, J. D.; Vitullo, V. P.; Whalen, D. L. Journal of the American Chemical

Society 1971, 93,6294–6296.

[150] Burt, R.; Chiang, Y.; Chwang, W.; Kresge, A.; Okuyama, T.; Tang, Y.; Yin,

Y. Journal of the American Chemical Society 1987, 109,3787–3788.

[151] Kresge, A.; Yin, Y. Canadian Journal of Chemistry 1991, 69,84–87.

[152] Schubert, W. M.; Lamm, B.; Keeffe, J. R. Journal of the American Chemical

Society 1964, 86,4727–4729.

[153] Kresge, A. J.; Weeks, D. P. Journal of the American Chemical Society 1984,

106,7140–7143.

[154] Chan, J.; Lu, A.; Bennet, A. J. Journal of the American Chemical Society

2011, 133,2989–2997.

[155] Lou, M.; Gilpin, M. E.; Burger, S. K.; Malik, A. M.; Gawuga, V.; Popovic,

V.; Capretta, A.; Berti, P. J. Journal of the American Chemical Society 2012,

134,12947–12957.

[156] Lou, M.; Burger, S. K.; Gilpin, M. E.; Gawuga, V.; Capretta, A.; Berti, P. J.

Journal of the American Chemical Society 2012, 134,12958–12969.

284 [157] Klinmann, J. P. Advances in Enzymology and Related Areas of Molecular Bio-

logy 1978, 46,415–494.

[158] Bennet, A. J.; Sinnott, M. L. Journal of the American Chemical Society 1986,

108,7287–7294.

[159] Lee, S. S.; Hong, S. Y.; Errey, J. C.; Izumi, A.; Davies, G. J.; Davis, B. G.

Nature Chemical Biology 2011, 7,631–638.

[160] Burt, R.; Chiang, Y.; Kresge, A. Canadian Journal of Chemistry 1980, 58,

2199–2202.

[161] Chiang, Y.; Chwang, W.; Kresge, A.; Yin, Y. Journal of the American Chem-

ical Society 1989, 111,7185–7190.

[162] Kresge, A.; Yin, Y. Canadian Journal of Chemistry 1987, 65,1753–1756.

[163] Kilpatrick, M. Journal of the American Chemical Society 1963, 85,1036–1038.

[164] Young, P. R.; Jencks, W. P. Journal of the American Chemical Society 1977,

99,8238–8248.

[165] McClelland, R. A.; Ahmad, M. Journal of the American Chemical Society

1978, 100,7027–7031.

[166] McClelland, R. A.; Ahmad, M. Journal of the American Chemical Society

1978, 100,7031–7036.

[167] Amyes, T. L.; Jencks, W. P. Journal of the American Chemical Society 1989,

111,7888–7900.

[168] Huang, X.; Surry, C.; Hiebert, T.; Bennet, A. J. Journal of the American

Chemical Society 1995, 117,10614–10621.

285 [169] Zhu, J.; Bennet, A. J. Journal of the American Chemical Society 1998, 120,

3887–3893.

[170] Sone, T.; Isobe, M.; Takabatake, E. Chemical & Pharmaceutical Bulletin 1989,

37,1922–4.

[171] Shipley, D. S.; Ross, A. M.; Mohan, R. S.; Whalen, D. L.; Sayer, J. M.; Serve,

P. M.; Jerina, D. M. The Journal of Organic Chemistry 1994, 59,977–983.

[172] Papworth, C.; Bauer, J. C.; Braman, J.; Wright, D. A. Strategies 1996, 9,3–4.

[173] McIntosh, L.; Hand, G.; Johnson, P.; Joshi, M.; Korner,M.;Plesniak,L.;

Ziser, L.; Wakarchuk, W.; Withers, S. Biochemistry 1996, 35,9958–9966.

[174] Schramm, V. L. Annual Review of Biochemistry 1998, 67,693–720.

[175] von Itzstein, M.; Thomson, R. Handb Exp Pharmacol 2009,111–154.

[176] Gervay, J.; Flaherty, T.; Nguyen, C. Tetrahedron Letters 1997, 38,1493–1496.

[177] Gregar, T. Q.; Gervay-Hague, J. Journal of Organic Chemistry 2004, 69,

1001–1009.

[178] Lai, H.-Y. L.; Axelrod, B. Biochemical and Biophysical Research Communica-

tions 1973, 54,463–468.

[179] Roeser, K.-R.; Legler, G. Biochimica et Biophysica Acta (BBA) - Enzymology

1981, 657,321–333.

[180] Watts, A. G.; Oppezzo, P.; Withers, S. G.; Alzari, P. M.; Buschiazzo, A.

Journal of Biological Chemistry 2006, 281,4149–4155.

[181] Wong, A. W.; He, S.; Withers, S. G. Canadian Journal of Chemistry 2001,

79,510–518.

286 [182] Praly, J. P.; Descotes, G. Tetrahedron Letters 1987, 28,1405–8.

[183] Praly, J. P.; Brard, L.; Descotes, G.; Toupet, L. Tetrahedron 1989, 45,4141–

52.

[184] Praly, J.-P.; Brendle, J.-C.; Klett, J.; Pequery, F. Comptes Rendus de

l’Academie des Sciences, Serie IIc: Chimie 2001, 4,611–617.

[185] Rempel, B. P.; Withers, S. G. Australian Journal of Chemistry 2009, 62,

590–599.

[186] Lee, S. S.; Greig, I. R.; Vocadlo, D. J.; McCarter, J. D.; Patrick, B. O.; Withers,

S. G. Journal of the American Chemical Society 2011, 133,15826–15829.

[187] Konstantinidis, A.; Sinnott, M. L. Biochemical Journal 1991, 279,587–93.

[188] Barbosa, J. A.; Smith, B. J.; DeGori, R.; Ooi, H. C.; Marcuccio, S. M.; Campi,

E. M.; Jackson, W. R.; Brossmer, R.; Sommer, M.; Lawrence, M. C. JMol

Biol 2000, 303,405–421.

[189] Saquib, M.; Sagar, R.; Shaw, A. K. Carbohydrate Research 2006, 341,1052–

1056.

[190] Hauser, F. M.; Ellenberger, S. R.; Ellenberger, W. P. Tetrahedron Letters 1988,

29,4939–42.

[191] Grapsas, I. K.; Couladouros, E. A.; Georgiadis, M. P. Polish Journal of Chem-

istry 1990, 64,823–6.

[192] Saeed, M.; Ilg, T.; Schick, M.; Abbas, M.; Voelter, W. Tetrahedron Letters

2001, 42,7401–7403.

[193] Porzelle, A.; Gordon, V. A.; Williams, C. M. Synlett 2007,1619–1621.

287 [194] Akai, S.; Nakagawa, T.; Kajihara, Y.; Sato, K.-i. Journal of Carbohydrate

Chemistry 1999, 18,639–654.

[195] Ooi, H. C.; Marcuccio, S. M.; Jackson, W. R. Australian Journal of Chemistry

2000, 53,171–174.

[196] Vreys, V.; David, G. Journal of Cellular and Molecular Medicine 2007, 11,

427–452.

[197] McKenzie, E. A. British Journal of Pharmacology 2007, 151,1–14.

[198] Zhou, Z.; Bates, M.; Madura, J. D. Proteins - Structure, Function and Bioin-

formatics 2006, 65,580–592.

[199] Pikas, D.; Li, J.; Vlodavsky, I.; Lindahl, U. Journal of Biological Chemistry

1998, 273,18770–18777.

[200] Okada, Y.; Yamada, S.; Toyoshima, M.; Dong, J.; Nakajima, M.; Sugahara,

K. Journal of Biological Chemistry 2002, 277,42488–42495.

[201] Peterson, S. B.; Liu, J. Journal of Biological Chemistry 2010, 285,14504–

14513.

[202] Peterson, S.; Liu, J. Journal of Biological Chemistry 2012, 287,34836–34843.

[203] Gandhi, N. S.; Freeman, C.; Parish, C. R.; Mancera, R. L. Glycobiology 2012,

22,35–55.

[204] Hammond, E.; Bytheway, I.; Ferro, V. New Developments in Therapeutic Gly-

comics ;Delehedde,M.;Lortat-Jacob,H.,Eds.;Trivandrum,India:Research

Signpost, 2006; Chapter Heparanase as a target for anticancer therapeutics:

New developments and future prospects, pp 251–282.

288 [205] Fux, L.; Ilan, N.; Sanderson, R. D.; Vlodavsky, I. Trends in Biochemical Sci-

ences 2009, 34,511–519.

[206] Chen, J.; Zhou, Y.; Chen, C.; Xu, W.; Yu, B. Carbohydrate Research 2008,

343,2853–2862.

[207] Hu, Y.-P.; Lin, S.-Y.; Huang, C.-Y.; Zulueta, M. M. L.; Liu, J.-Y.; Chang, W.;

Hung, S.-C. Nature Chemistry 2011, 3,557–563.

[208] Liu, R.; Xu, Y.; Chen, M.; Weïwer, M.; Zhou, X.; Bridges, A. S.; DeAngelis,

P. L.; Zhang, Q.; Linhardt, R. J.; Liu, J. Journal of Biological Chemistry 2010,

285,34240–34249.

[209] Hammond, E.; Li, C. P.; Ferro, V. Analytical Biochemistry 2010, 396,112–

116.

[210] Pearson, A. G.; Kiefel, M. J.; Ferro, V.; von Itzstein, M. Organic & Biomolecu-

lar Chemistry 2011, 9,4614–4625.

[211] Hanzlik, R.; Westkaemper, R. Journal of the American Chemical Society 1980,

102,2464–2467.

[212] Westkaemper, R.; Hanzlik, R. Archives of Biochemistry and Biophysics 1981,

208,195–204.

[213] Gillilan, R.; Pohl, T.; Whalen, D. Journal of the American Chemical Society

1982, 104,4482–4484.

[214] Hennig, C.; Oswald, R.; Schmatz, S. Journal of Physical Chemistry A 2006,

110,3071–3079.

[215] Capon, B. Chemical Reviews 1969, 69,407–498.

289 [216] Griffith, B. R.; Krepel, C.; Fu, X.; Blanchard, S.; Ahmed, A.; Edmiston,

C. E.; Thorson, J. S. Journal of the American Chemical Society 2007, 129,

8150–8155.

[217] Weiberth, F. J.; Gill, H. S.; Jiang, Y.; Lee, G. E.; Lienard, P.; Pemberton, C.;

Powers, M. R.; Subotkowski, W.; Tomasik, W.; Vanasse, B. J.; Yu, Y. Organic

Process Research and Development 2010, 14,623–631.

[218] Schreiner, E.; Zbiral, E.; Kleineidam, R. G.; Schauer,R.Carbohydrate Research

1991, 216,61–66.

290 Appendix A

Multiple sequence alignment of UGL from species shown in Table 2.2

Multiple sequence alignment of UGL by the MUSCLE algorithm. Identical residues are coloured yellow on a purple background, conserved residues are coloured white on a blue background, the putative catalytic residues D113 and D173 (C. perfringens

numbering) are indicated with arrows, while the sulfate binding domain is indicated

with a brace. Clostridium_perfringens M...... I 2 Flavobacterium_heparinum MNKSRTLIFSLLALMGTASSADAQIAKTPA30 Streptococcus_agalactiae M...... MKI 4 Bacillus_sp_GL_1 M...... 1

Clostridium_perfringens KEIRVEEIAKKDEFLKTKLLTRAEVKNAID32 Flavobacterium_heparinum KPLSDQMAATVMEIWPERAKKWSYDHGVVQ60 Streptococcus_agalactiae KPVKVESIENPKRFLDSRLLTKIEVEEAIE34 Bacillus_sp_GL_1 ...... W...... QQAIG7

Clostridium_perfringens LVIKQIDA...NMEYFK...... 46 Flavobacterium_heparinum DGMDALWKRSGNASYFKYIQNDMDGFISAD90 Streptococcus_agalactiae KALKQLYI...NIDYFG...... 48 Bacillus_sp_GL_1 DALGITAR...NLKKFG...... 21

291 Clostridium_perfringens ...EKFPSSA...... TK55 Flavobacterium_heparinum GIIDTYSQEHVNIDNVKNGTVLLDLYKITG 120 Streptococcus_agalactiae ...EEYPTPA...... TF57 Bacillus_sp_GL_1 ...DRFPHVS...... DG 30

Clostridium_perfringens NNQYGIIENIEWTD...... GFWTGL75 Flavobacterium_heparinum QQKYFKAATTLWEQLKIQPRTKQGSFWHKK150 Streptococcus_agalactiae NNTYKVMDNTEWTN...... GFWTGC77 Bacillus_sp_GL_1 SNKYVLNDNTDWTD...... GFWSGI50

Clostridium_perfringens L.....WLAYEYTGDEKYRELA..DKNVAS 98 Flavobacterium_heparinum IYPNQVWLDGLYMGQPFYAEYAALIGNKEA180 Streptococcus_agalactiae L.....WLAYEYNQDKKLKNIA..HKNVLS 100 Bacillus_sp_GL_1 L.....WLCYEYTGDEQYREGA..VRTVAS 73

catalytic residue ↓ Clostridium_perfringens FKNRVEKDIELDHHD....LGFLY...SLA121 Flavobacterium_heparinum FDDIANQFIWVEQNTRDARTGLLYHGWDES 210 Streptococcus_agalactiae FLNRINNRIALDHHD....LGFLY...TPS 123 Bacillus_sp_GL_1 FRERLDRFENLDHHD....IGFLY...SLS 96

Clostridium_perfringens TVSGYKLTGSEDARE...... ASIK 140 Flavobacterium_heparinum KTERWADPKTGLSPHIWARAMGWYAMALVE240 Streptococcus_agalactiae CTAEYRINGDAKALE...... ATIK 142 Bacillus_sp_GL_1 AKAQWIVEKDESARK...... LALD115

Clostridium_perfringens AANKLISRYQEKGEFIQAWGELG...... 163 Flavobacterium_heparinum TLDNFPKTHPKRQEMINILNRLAAAVKNTQ270 Streptococcus_agalactiae AADKLMERYQEKGGFIQAWGELG...... 165 Bacillus_sp_GL_1 AADVLMRRWRADAGIIQAWGPKG...... 138

catalytic residue ↓ Clostridium_perfringens .SKDH.YRFIID...... CLLNIPLLYWA 184 Flavobacterium_heparinum NNKTGVWYDILDQPNRKGNYFESSASSMFV300 Streptococcus_agalactiae .YKEH.YRLIID...... CLLNIQLLFFA 186 Bacillus_sp_GL_1 .DPENGGRIIID...... CLLNLPLLLWA 160

Clostridium_perfringens SDETGDAKYRNIANKHFVTSCN...... 206 Flavobacterium_heparinum YAIAKGVRLGYLPASYFVVASKGYKGIQQE330 Streptococcus_agalactiae YEQTGDEKYRQVAVNHFYASAN...... 208 Bacillus_sp_GL_1 GEQTGDPEYRRVAEAHALKSRR...... 182

292 Clostridium_perfringens NVIRDDASAFH...TFYMDNETGKPLRGVT 233 Flavobacterium_heparinum FIEQRAEGKINLKGTVSVSGLGGKPYRDGS360 Streptococcus_agalactiae NVVRDDSSAFH...TFYFDPETGEPLKGVT 235 Bacillus_sp_GL_1 FLVRGDDSSYH...TFYFDPENGNAIRGGT 209

4-sulfate binding ↓ Clostridium_perfringens RQGYSDDSAWARGQAWGVYGIPLNYRYTRN 263 Flavobacterium_heparinum YEYYMSEKVVSNDP.KGVGAFLMAANEMEI389 Streptococcus_agalactiae RQGYSDESSWARGQAWGIYGIPLSYRKMKD265 Bacillus_sp_GL_1 HQGNTDGSTWTRGQAWGIYGFALNSRYLGN 239

Clostridium_perfringens ESCFNLYEG....MTNYFLNRLPKD...NV 286 Flavobacterium_heparinum AALPKPGLGKTVLLDSYFNNESRKDQSGNL419 Streptococcus_agalactiae YQQIILFKG....MTNYFLNRLPED...KV 288 Bacillus_sp_GL_1 ADLLETAKR....MARHFLARVPED...GV 262

Clostridium_perfringens CYWDLIFND...... GDDHSKD 302 Flavobacterium_heparinum VSWHYKWDELANGGFSMWADQFNNAGFKTA449 Streptococcus_agalactiae SYWDLIFTD...... GSGQPRD 304 Bacillus_sp_GL_1 VYWDFEVPQ...... EPSSYRD 278

Clostridium_perfringens SSAAAIAVCGMHEMNKYL...PEVDENKEV329 Flavobacterium_heparinum TLKAAPTAANLKNASVYIIVDPDTEKETEK479 Streptococcus_agalactiae TSATATAVCGIHEMLKHL...PEVDPDKET331 Bacillus_sp_GL_1 SSASAITACGLLEIASQL...DESDPERQR305

Clostridium_perfringens YKYAMHNILRSLME...... 343 Flavobacterium_heparinum PNFVAQNDIKAIAEWVKGGGILVLMANDTG509 Streptococcus_agalactiae YKYAMHTMLRSLIE...... 345 Bacillus_sp_GL_1 FIDAAKTTVTALRD...... 319

Clostridium_perfringens ...... NYMNPE 349 Flavobacterium_heparinum NVELDHFNQLAKTFGIEFNKDSKGRVVKSQ539 Streptococcus_agalactiae ...... QYSNSE 351 Bacillus_sp_GL_1 ...... GYAERD325

6-sulfate binding domain

Clostridium_perfringens IEPGKPVLLHGVYSWHSGKGV...... 370 Flavobacterium_heparinum FEMGKVMVPAGNE!IFK"#TAK$QLYVKEYSSLK569 Streptococcus_agalactiae FIAGRPLLLHGVYSWHSGKGV...... 372 Bacillus_sp_GL_1 DGEAEGFIRRGSYHVRGGISP...... 346

293 Clostridium_perfringens ...... DEGNIW.376 Flavobacterium_heparinum LTTAAKAVLKDKDGDNVMAIAKYGKGAVFA599 Streptococcus_agalactiae ...... DEGNIW.378 Bacillus_sp_GL_1 ...... DDYTIW.352

Clostridium_perfringens .GDYFFLEALI...RFYKDWNLY...... 395 Flavobacterium_heparinum IGDPWLYNEYVDGRKLPADYQNFEAGQDLV629 Streptococcus_agalactiae .GDYYYLEALI...RFYKDWELY...... 397 Bacillus_sp_GL_1 .GDYYYLEALL...RLERGVTGY...... 371

Clostridium_perfringens .W...... 396 Flavobacterium_heparinum NWIGKQLLKK639 Streptococcus_agalactiae .W...... 398 Bacillus_sp_GL_1 .WYERGR...377

294 Appendix B

2D-NMR spectra of UGL products and standards

Figure B.1: Expanded TOCSY spectrum of the UGL catalysed reaction of phenyl ΔGlcA (10)in10%methanol.

295 Figure B.2: COSY45 spectrum of the UGL-catalysed reaction of phenyl ΔGlcA (10) in 10% methanol, to form 26.

296 Figure B.3: NOESY spectrum of the UGL catalysed reaction of phenyl ΔGlcA (10) in 10% methanol (26).

297 Figure B.4: COSY45 spectrum of 28,thecarbon5epimericsyntheticstandardfor the UGL catalysed reaction of phenyl ΔGlcA (10)in10%methanol.

298 Figure B.5: NOESY spectrum of 28,thecarbon5epimericsyntheticstandardfor the UGL catalysed reaction of phenyl ΔGlcA (10)in10%methanol.

299 Appendix C

Kinetic isotope effects

Kinetic isotope effects are an important tool that can be used to probe the transition state of a reaction, without changing the reaction path. These effects arise from the difference in energy of the vibrational states of the two isotopomers. Substitution of an isotope for a heavier or lighter analogue will have an effectonthereducedmassof adiatomicsystem,influencingtheenergyofitsvibrations,and thereby the ease with which its bond order can change. This influence on reduced massandvibrational energy can be approximated by the equation for the energy of a harmonic oscillator, equation C.1, and the associated equations for vibrational frequency, equation C.2, and reduced mass, equation C.3. From these it can be seen that changes in the mass of one atom in the pair can have a small influence on the vibrational energy states.

Example isotopologue energy levels for a quantum number of 0 are represented in the Morse potential curve in Figure C.1.

E =(n + 1/2)hv (C.1)

1 k v = (C.2) 2π %µ

m m µ = A B (C.3) mA + mB

(where E is energy, h is planck’s constant, n is quantum number, v is vibrational

300 frequency, k is vibrational force constant, µ is reduced mass, and m is mass) Energy

BDED BDEH

E0H

E0D

re internuclear distance, r

Figure C.1: Hypothetical Morse potential curve, showing thezeropointenergyfora hydrogen and deuterium deuterated compound. The optimal internuclear distance is denoted re, and bond dissociation energies are denoted BDE.

Any isotopic substitution can, in theory, lead to such an effect. However, the

most common is a deuterium for hydrogen substitution, as the effect on reduced

mass is substantial, synthesis is often reasonably simple, and no radio-isotopes are

involved. Other common isotopes used for such experiments include 13Cor14C,

15N, and 17Oor18O. It is important to note that, for an enzymatic reaction, if non-

chemical steps such as domain movements, substrate binding,orproductreleaseare

rate limiting then no isotope effect will be observed. Kineticisotopeeffectsarising

301 from bond making or bond breaking to the isotope in question are termed primary effects, while those arising from indirect effects are termed secondary.

Primary isotope effects can most easily be conceptualised from the difference in energy between the ground state, where a deuterated compound is lower in en- ergy than a protonated compound, and the hypothetical limit of a bond breaking transition state, where the bond is completely broken and so both deuterated and protonated compounds must necessarily be of the same energy.Thisisrepresented in Figure C.2. Since this energy gap is smaller in a protonatedthaninadeuterated compound, the protonated compound reacts more quickly. In cases where there is higher bond order in the transition state the effect will be smaller. Primary kinetic isotope effects usually manifest as rate differences of between 2 and 7 fold faster reaction for the protonated compound over the deuterated.

Secondary isotope effects can arise in a number of ways, but generally still come about from different effects on the energy at the ground state and at the transition state. Secondary effects are said to be α if the isotope is attached to the centre at which the reaction is taking place, and β if it is adjacent to this centre. Secondary effects have a less dramatic rate difference between isotopologues, typically ranging from 0.7 to 1.5 fold faster reaction of the hydrogen-substituted compound over that of the deuterium-substituted compound. Those cases where the effect is less than one are said to be inverse effects, with the deuterated compound reacting faster, while those above one are said to be normal effects.

The most common way in which an α-secondary effect can arise is through changes in hybridisation. Because the energy of vibrational modes is dependent on hybrid- isation, a change in hybridisation at the transition state relative to the ground state can lead to a difference in activation energy for the two isotopomers, and thus a difference in rate. Changing hybridisation from sp3 to sp2 or sp2 to sp results in

302 ΔEH1<ΔED1 ΔEH2<<ΔED2 small primary KIE large primary KIE

X- H+ X- D+ Transition states - + XHδ δ XDδ- δ+

ΔEH1 ΔEH2 Potential energy

ΔED1 ΔED2

XH

XD Ground states

Figure C.2: Example transition states leading to primary kinetic isotope effects. In the case represented by ΔEH1andΔED1, the transition states have a relatively high bond order and the differences in energy between the deuterated and protonated cases remain at the transition state, and so a small primary kinetic isotope effect would be observed. In the case represented by ΔEH2andΔED2, the transition states are shown at the extreme limit, with no bond remaining,andsoalarge primary kinetic isotope effect would be expected.

303 anormalkineticisotopeeffect,whiletheoppositechangeinhybridisation has the opposite kinetic isotope effect. These effects are illustrated in Figure C.3.

sp transition state sp3 transition state

EH

E H ED ED

2 Energy sp ground state ΔEH ΔEH ΔED ΔED

EH

ED

ΔED>ΔEH ΔED<ΔEH Normal KIE Inverse KIE

Figure C.3: Example energy differences between a hypothetic sp2 ground state and sp and sp3 transition states for a hydrogen and a deuterium substitutedcompound, which would give rise to a normal and an inverse KIE, respectively.

Acommonmeansbywhichanormalβ-secondary kinetic isotope effect can arise is by hyperconjugation. This is the donation of electron density from a carbon- or

heteroatom-hydrogen bond to an adjacent vacant orbital to stabilise it. Because

carbon-deuterium bonds are slightly stronger than a carbon-hydrogen bond, its elec-

trons are less available for hyperconjugation. This means that if there is an increase

in positive charge at the transition state at a carbon centre adjacent to a carbon-

hydrogen bond then substitution of this hydrogen for deuterium will give a slower

reaction, as the charged centre, and thus the transition state in general, will be less

stabilised. This electron donation is, however, very sensitive to the orientation of the

304 vacant orbital and the adjacent carbon- or heteroatom-hydrogen bond, as the bond- ing orbital must be able to overlap with the vacant atomic orbital for this stabilising electron donation to take place. This overlap is illustratedinFigureC.4.

+ H(D) H(D) X H(D) Y H(D) + H(D)

Figure C.4: Hypothetical kinetic isotope effect arising fromhyperconjugation.The positive charge at the transition state is better stabilisedbyanadjacenthydrogen, arising from its weaker C−Hbond.

Other sources of kinetic isotope effects exist, but are less common. These arise from the same general principle, of a difference in the activation energy arising from isotopic substitution. In the previous case, with a hydrogenadacenttoadeveloping charge, if the heteroatom-hydrogen bonding orbital is not able to overlap with the vacant orbital, an effect from induction can sometimes be seen. This gives a small inverse effect, as the deuterium atom is slightly less electron withdrawing compared to hydrogen. In a final example, steric effects can occasionally give rise to kinetic isotope effects, but only in highly restricted systems. For example in the bond rotation shown in Figure C.5, the larger deuterium atoms slowdowninterconversion of the two forms.

CD3 D3C D3C CD3

Figure C.5: Steric interactions leading to a kinetic isotopeeffect.

305 Appendix D

Plots for Michaelis-Menten kinetics

Plots of rate against substrate concentration, organised bycompoundnumber.Fits shown are to the Michaelis-Menten equation (equation 7.1 on page 261), with double recpirocal plots as insets (not used for analysis). 610

11 12

306 13 14

15 40

43 46

47 48

307 49 50

51 52

53 70

50 with D113G mutant

308