UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Mechanism-based inhibitors as in vitro and in vivo probes of glycosidase structure and mechanism McCarter, John D. 1995

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1995-060187.pdf [ 5.18MB ]
JSON: 831-1.0059845.json
JSON-LD: 831-1.0059845-ld.json
RDF/XML (Pretty): 831-1.0059845-rdf.xml
RDF/JSON: 831-1.0059845-rdf.json
Turtle: 831-1.0059845-turtle.txt
N-Triples: 831-1.0059845-rdf-ntriples.txt
Original Record: 831-1.0059845-source.json
Full Text

Full Text

MECHANISM-BASED INHIBITORS AS IN VITRO AND IN VIVOPROBES OF GLYCOSIDASE STRUCTURE AND MECHANISMByJOHN D. McCARTERB.Sc., The University of Victoria, 1988M.Sc., The University of British Columbia, 1991A THESIS SUBMI’IThD IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Department of Chemistry)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAJuly, 1995© John D. McCarter, 1995In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of ChernistxyThe University of British ColumbiaVancouver, CanadaDate 14_September, 1995DE-6 (2/88)ABSTRACTTwo new classes of specific, mechanism-based glycosidase inactivators weredeveloped: 2,2-dihalo glycosyl chlorides and 5-fluoro glycosyl fluorides. Both classeswere effective against x-glucosidases, which had been hitherto resistant to similarinactivation strategies. Incubation of yeast -glucosidase with 2-chloro-2-deoxy-2-fluoro-a-D-glucopyranosyl chloride or 2-deoxy-2,2-difluoro-c-D-arabinohexopyranosyl chlorideresulted in time-dependent inactivation of the enzyme, presumably by formation ofextremely stabilized 2,2-dihalo glycosyl-enzyme intermediates that are essentially incapableof turnover. Similar inhibition of Agrobacterium faecalis f3-glucosidase and yeast aglucosidase was seen with the corresponding 5-fluoro glycosyl fluorides. 5-Fluoro-f3- and5-fluoro-a-D-glucosyl fluorides form catalytically competent intermediates with theappropriate glucosidases that are capable of turnover, but at rates reduced 10- and 1 0-fold, respectively, with respect to the f3- and oc-D-glucosyl fluoride parent substrates. Thecorresponding 5-fluoro-L-idosyl fluorides, the C5 epimers of the glucosyl compounds,show even greater reductions in turnover rates, the kcat values of the 5-fluoro-cc- and 5-fluoro-(3-L-idosyl fluorides with the appropriate enzymes being reduced a further 1.5- and3000-fold. The spontaneous hydrolysis rates of these 5-fluoro glycosyl fluorides, andthose of the corresponding 2-deoxy-2-fluoro compounds, were determined to probe theeffects of the various fluorine substitutions on the transition states for hydrolysis.A novel mass spectrometric technique for the identification and sequencing oflabelled active site peptides without the need for radiolabels has been developed. Briefly,the technique involves enzyme inactivation, proteolytic digestion of the enzyme, andidentification of the modified peptide using electrospray tandem mass spectrometry byexploiting the lability of the inhibitor-peptide bond which is selectively cleaved by collision-induced fragmentation. The key catalytic nucleophiles in the clinically important humanlysosomal f-glucosidase (deficient in Gaucher disease), human acid f3-galactosidase11(deficient in GM1 gangliosidosis), and yeast ct-glucosidase were identified, as Glu-340,Glu-268, and Asp-214, respectively.3-Glucosidase and 3-mannosidase inhibitors, labelled with the positron-emittingisotope 18F, were synthesized for use in a novel approach to the in vivo imaging ofglycosidase activity using positron emission tomography (PET). This may be useful in thediagnosis and treatment of abnormal glycosidase activity associated with disease (e.g.Gaucher disease, the inherited deficiency of lysosomal 3-g1ucosidase). Initially,Agrobacterium faecalis f3-glucosidase was labelled in vitro with an 18F-labelledmechanism-based enzyme inactivator 2-deoxy-2-[18F]fluoro--D-mannosyl18F]fluoride,the first such labelling of a glycosidase with this positron-emitting isotope, and reactivationof the labelled enzyme was observed by monitoring release of radioactivity. Non-labelledversions of these inhibitors were administered to rats. Rapid inactivation of the appropriateenzymes was observed in each tissue assayed, and the clearance of the inhibitors wasdemonstrated by their slow release from the enzymes both in vitro and in vivo. Uptake andclearance of the inhibitors was probed using 19F NMR and18F-radiolabelled compounds,revealing little hydrolysis but non-specific uptake of the inhibitor. Preliminary imagingresults were obtained in rats using PET, demonstrating the potential of this approach forimaging glycosidase activity in vivo.TABLE OF CONTENTSABSTRACT iiTABLE OF CONTENTS ivLIST OF FIGURES ixLIST OF SCHEMES xiiiLIST OF TABLES xixLIST OF ABBREVIATIONS xvACKNOWLEDGEMENTS xviiCHAPTER IGENERAL INTRODUCTION1.1 Glycosidases 21.2 Classification of Glycosidases 21.3 The Catalytic Mechanism of Retaining Glycosidases 41.4 Evidence for a Double Displacement Mechanism in RetainingGlycosidases 51.4.1 Presence of a carboxylate nucleophile 51.4.2 The nature of the glycosyl-enzyme intermediate 71.4.3 Oxocarbenium ion-like transition states 91.4.4 General acid catalysis 121.4.5 Non-covalent enzyme-substrate interactions 141.5 Glycosidase Inhibitors as Structural and Mechanistic Probesin Biomedicine 161.5.1 Affinity labels 171.5.2 Conduritol epoxides 191.5.3 Enzyme-activated inhibitors 211.5.4 Trapping of a catalytic intermediate:2-deoxy-2-fluoro glycosides 231.6 Aims of this Thesis 25ivCHAPTER II5-FLUORO GLYCOSIDES AS MECHANISM-BASED INACTIVATORSOF BOTH I- AND cc-GLYCOSIDASES2.1 Introduction 282.2 Specific Aims 29Results and Discussion 312.3 Synthesis 312.4 Spontaneous Hydrolysis of 5-Fluoro Glycosyl Fluorides 382.5 Kinetic Studies of the Reaction of 5-Fluoro Glycosyl Fluorideswith Agrobacterium faecalis -Glucosidase and Yeast(Saccharomyces cerevisiae) c-G 1 ucosi dase 432.5.1 Inactivation of Agrobacterium faecalis -glucosidaseand yeast cc-glucosidase by the 5-fluoro D-glucosides5Fj3GIuF (2.3) and 5FxGJuF (2.4), respectively 482.5.2 Inactivation of Agrobacterium faecalis 3-gi ucosidaseand yeast a-giucosidase by the 5-fluoro L-idosides5FxIdoF (2.9) and 5Fl3IdoF (2.14), respectively 532.5.3 Interpretation of Kinetic Results with 5-FluoroGlycosyl Fluorides 602.6 Identification of the Catalytic Nucleophiles of both3- and a-G)ycosidases using 5-Fluoro Glycosyl Fluoridesand Mass Spectrometry 652.6.1 Confirmation of the mode of action of 5-fluoroglycosyl fluorides by labelling of the catalyticnucleophile of Agrobacterium faecalis 3-glucosidase 652.6.2 The catalytic nucleophile of yeast(Saccharomyces cerevisiae) ec-glucosidase 68CHAPTER III2,2-DIHALO GLYCOSIDES AS MECHANISM-BASED INACTIVATORSOF o-GLYCOSIDASES3.1 Introduction 813.2 Specific Aims 82VResults and Discussion.833.3 Synthesis 833.4 Kinetic Studies with Yeast cL-glucosidase and Jack Beana-mannosidase 883.4.1 Inactivation of yeast -glucosidase with2Cl2FaGluCl (3.1) and 2,2FxAraCl (3.3) 913.4.2 Attempted inactivation of Jack bean x-mannosidase 973.5 Mass Spectrometry of Labelled cx-Glucosidase 1003.6 Conclusions 102CHAPTER IVMECHANISM-BASED GLYCOSIDASE INHIBITORS AS PROBES OFHUMAN ENZYMES AND AS POTENTIAL DIAGNOSTICS4.1 Introduction 1054.2 Specific Aims 107Results and Discussion 1084.3 Identification of the Catalytic Nucleophiles of HumanLysosomal Glycosidases by 2-Deoxy-2-Fluoro-GlycosylFluorides and Mass Spectrometry 1084.3.1 The catalytic nucleophile of human glucocerebrosidase 1084.3.2 The catalytic nucleophile of human lysosomal acid3-galactosidase precursor 1194.4 Specific In Vivo Inhibition of -Glucosidase andf3-Mannosidase, but Not 3-Galactosidase Activity in Ratsby 2-Deoxy-2-Fluoro- f3-Glucosyl and -Mannosyl Fluoridesand Recovery of Activity In Vivo and In Vitro 1314.5 Radiosynthesis of 2-deoxy-2-[18F]flu ro-3-D-18 . 18mannopyranosyl [ F]fluorade and F-Labellingof Agrobacterium faecalis -Glucosidase In Vitro:a Model Study 1434.6 Biodistribution and preliminary imaging results of 2-deoxy-2-[18F]fluoro-3-D-mannopyranosy1[18F]fluoride in rats 147viCHAPTER VMATERIALS AND METHODS5.1 Synthesis 1575.1.1 General 1575.1.2 General Procedures 1585.1.3 Syntheses of 5-Fluoro Glycosyl Fluorides 1595.1.4 Syntheses of 2,2-Dihalo Glycosyl Chlorides 1635.2 Enzyme Kinetics 1695.2.1 General Methods and Materials 1695.2.2 Time-Dependent Inactivation 1705.2.3 Protection Against Inactivation 1715.2.4 Reactivation of Inactivated Enzyme 1715.2.5 Test for Inhibitory Contaminant in 5FIdoF Preparation 1725.2.6 Test for Inhibitory Contaminant in SFcddoF Preparation 1725.2.7 Determination of Apparent K1’ Values 1735.2.8 Fluoride Electrode Kinetics 1735.3 Determination of the Stereochemical Course of GlycosideHydrolysis by 1H NMR Spectroscopy 1745.4 Spontaneous Hydrolysis of Glycosyl Fluorides 1755.5 Inactivation of Glycosidases for Proteolysisand Electrospray MS Analysis 1755.6 Proteolysis 1765.7 Electrospray MS conditions for Peptide Analysis 1775.8 Isolation of the 2FG1u-Iabelled Peptide of GCase 1785.9 Chemical Sequencing of 2FGIu-labelled Peptide of GCase 1795.10 Aminolysis 1795.11 Stoichiometry of Inactivation by Electrospray MS 1795.12 Liquid Secondary Ion Mass Spectrometry (LSIMS) 1795.13 Preparation of Tissue Homogenates 1805.14 Assay of Tissue 3-Glycosidase Activity 1805.15 In Vitro Reactivation of Tissue 3-Glycosidase Activity 1815.16 19F NMR Spectroscopy of 2-Deoxy-2-Fiuoro-f3-D-Mannopyranosyl Fluoride in Rat Tissue Extracts 1815.17 Radiosynthesis of 2-Deoxy-2-[18FJFIu ro-f3-D-Mannosyl[18F]Fluoride 182vii5.18 Labelling of Agrobacterium faecalis 3-glucosidase with 18F 1835.19 Reactivation of the 2-Deoxy-2-[’8F]Flu ro-a-D-Mannosyl-Enzynie 1835.20 Biodistribution of 2-Deoxy-2-[18F]Fluoro-f3-D-Mannosyl[18F]Fluoride in Rats 1845.21 Pre-treatment of Rats with 2-Deoxy-2-Fluoro-3-D-Mannosyl Fluoride (2F3ManF) 1845.22 PET Imaging 185APPENDIX I Graphical Representation of Spontaneous HydrolysisData 186REFERENCES 190‘/111LIST OF FIGURESFigure 1.1. Reaction normally catalyzed by a glycosidase 2Figure 1.2. Hydrolyses catalyzed by retaining and inverting enzymes(shown for J3-glycosidases) 3Figure 1.3. Preswnedmechanisms of retaining and inverting f3-glucosidases 4Figure 1.4. Presumed mechanism ofa retaining 13-glucosidase 6Figure 1.5. Inactivation ofAgrobacterium J3-glucosidase by trappingof a 2 -deoxy-2 -fluoro-glucosyl-enzyme 9Figure 1.6. Comparison of transition state glucosyl oxocarbenium ionand ground state glucoside 10Figure 1.7. Resonance structures of aldonolactones (1.1) and aldonolactams (1.2) 11Figure 1.8. Nojirimycin (1.3), in both hydrated and dehydratedforms,and nojiritetrazole (1.4) 12Figure 1.9. Hydration ofD-galacto-octenirol (1.5) and ofD-galactal (1.6)by E. coli /3-galactosidase 13Figure 1.10. Galactosylpyridinium salt 14Figure 1.11. N-Bromoaceiyl glycosylamines (1.8) and glycosylisothiocyanates (1.9), affinity labels of glycosidases 18Figure 1.12. Reaction of an epoxyalkyl glucoside (1.10) with a /3-glucosidase 19Figure 1.13. Reaction of conduritol B epoxide (1.11) with a /3-glucosidase 20Figure 1.14. Generation ofa reactive glycosyl carbonium ionfroma glycosylmethyl triazene 22Figure 1.15. Generation ofa reactive acyifluoride and a quinone methidefrom duoroalkyl glycosides (1.13) and duorotolylglycosides (1.14), respectively 23Figure 1.16. Turnover of a 2-deoxy-2-fluoro-glucosyl-enzyme intermediateby transglycosylation 25Figure 2.1. Structures of valiolamine (2.1), and nucleocidin (2.2) 30Figure 22 Structures of5-fluoro /3—D-glucopyranosylfluoride, 5FJ3G1uF (2.3)and 5-fluoro cx-D -glucopyranosylfluoride, 5FcxGluF (2.4) 31Figure 2.3. Partial map ofpyranoid ring interconversions 35Figure 2.4. Presumed mechanism of spontaneous glycoside hydrolysis 38Figure 2.5. Presumed mechanism ofspontaneous 5-fluoro glycoside hyrolysis 39Figure 2.6. Kinetic schemefor retaining glycosidases 44ixFigure 2.7. Kinetic schemefor inactivation of retaining glycosidasesby accumulation ofa covalent intermediate 46Figure 2.8. Inactivation ofAgrobacterium J3-glucosidase with 2.3 49Figure 2.9. Reactivation of isolated 5-fluoro D-glucopyranosyl-Agrobacteriumf3-glucosidase at 37°C 50Figure 2.10. ReversibleK1’for 2.3 with Agrobacterium /3-glucosidasedetermined under steady state conditions 50Figure 2.11. Hydrolysis of2.4 catalyzed by yeast a-glucosidase at 37°C 52Figure 2.12. ReversibleK1’for2.4 with yeast a-glucosidasedetermined under steady state conditions 52Figure 2.13. Inactivation ofAgrobacterium J3-glucosidase with 2.9 55Figure 2.14. Reactivation of isolated 5-fluoro L-idopyranosyl-AgrobacteriumJ3-glucosidase at 4°C and 37°C 56Figure 2.15. Inactivation ofyeast cz-glucosidase with 2.14 58Figure 2.16. Reactivation of isolated 5-fluoro L-iclopyranosyl-yeasta-glucosidase at 4°C and37°C 59Figure 2.17. ESMS experiments on Agrobacterium /3-glucosidase proteolyticdigests 67Figure 2.18. ESMS experiments on a-glucosidaseproteolytic digests 72Figure 2.19. Neutral loss Total Ion Chromatograms of a-glucosidaseproteolytic digests 73Figure 2.20. Tandem MS/MS daughter ion spectum of the 5FIdo-labelledactive site peptide (m/z 925 in the singly-charged state) 74Figure 221. A) Alignment of region containing the catalytic nucleophile aspartatelabelled by 5-fluoro glycosyifluorides in selected Family 13glycosyl hydrolases B) Alignment of region containingactive site aspartate labelled by conduritol B epoxide (CBE) inselected Family 3lglycosyl hydrolases 77Figure 3.1. Structures of2 -chloro-2 -deoxy-2 -fluoro-a-D-glucopyranosyl chloride,2Cl2FaGluCl (3.1), 2 -chloro-2 -deoxy-2 -fluoro-a-D-mannopyranosylchloride, 2Cl2FaManCl (3.2), and 2-deoxy-2,2 -dWuoro-a-Darabinohexopyranosyl chloride, 2 ,2FaAraCl (3.3) 82Figure 3.2. Possible mechanism for the formation of 3.10 89Figure 3.3. Synthesis of3-methyl-i -butenefrom 2-methyl-i -butanol,possibly via a 1,3 hydride shift 89xFigure 3.4. Kinetic schemefor retaining glycosidases 90Figure 35. Kinetic schemefor inactivation of retaining glycosidosesby accumulation ofa covalent intermediate 90Figure 3.6. Inactivation ofyeast a-glucosidase by 2C12FxG1uC1 93Figure 3.7. Protection against inactivation by 2C12FaG1uC1 94Figure 3.8. Inactivation ofyeast a-glucosidase by 2,2FaAraCl 95Figure 3.9. Protection against inactivation by 2,2FaAraCl 96Figure 3.10.1H-NMR determination of the stereochemical course of hydrolysisofp-nitrophenyl a-D-mannopyranoside by Jack bean a-mannosidase 99Figure 4.1. Structure ofglucosyl ceramide 109Figure 4.2. Inactivation ofGCase by 2-deoxy-2-fluoro-J3-D-glucosylfluoride 112Figure 4.3. Inactivation ofGCase by 2-deoxy-2-fluoro-J3-D-mannosylfluoride 113Figure 4.4. Reactivation of2 -deoxy-2 -fluoro-D -glucosyl-GCase 114Figure 45. Reactivation of2-deoxy-2 -fluoro-D-mannosyl-GCase 114Figure 4.6. ESMS experiments on GCase proteolytic digests 116Figure 4.7. TandemMS/MS daughter ion spectrum of the 2FGlu4abelledpeptide (m/z 688, FASEA+2FG1u) 1171Figure 4.8. H NMR spectra showing the stereochemical course of hydrolysisof2,4-dinitrophenyl galactopyranoside by human acid J3-galactosidase.... 121Figure 4.9. Inactivation of acid /3-galactosidase with 2F/3Ga1DNP 122Figure 4.10. ESMS Experiments on acid /3-galactosidase peptic digest 124Figure 4.11. ESMS Experiments on acid /3-galactosidasepeptic/tryptic digest 126Figure 4.12. Tandem MS/MS daughter ion spectrum of the 2FGa1-labelledpeptide,m/z 1040, GPLINSEF + 2FGal 127Figure 4.13. Selected ion chromatograms ofacid f3-galactosidase peptic/trypticdigest in normal mode 129Figure 4.14. Conserved residues among f3-galactosidases ofFamily 35glycosyl hydrolases and relatedproteins 130Figure 4.15. In vivo inhibition of J3-glucosidase activity by 2FJ3G1uFand recovery ofactivity in vivo 134Figure 4.16. In vivo inhibition of J3-mannosidase activity by 2FJ3ManFand recovery ofactivity in vivo 134Figure 4.17. Specflcity of inhibition of /3-glycosidase activities by 2F/3G1uFand 2FJ3ManF 135Figure 4.18. Inhibition of /3-glucosidase activity by 2FJ3G1uF and 2FJ3ManF 135xFigure 4.19. In vitro reactivation of f3-glucosidase activity in rat brainhomogenates at 37°C 137Figure 420. In vitro reactivation of /3-glucosidase activity in rat kidneyhomogenates at 37°C 137Figure 421. Radiochromatogramfrom gel permeation HPLC column ofincubated mixture of excess 2-deoxy-2-(‘8F]fluoro-13-D-mannopyranosyl[18F]fluoride and Agrobacterium J3-glucosidase 146Figure 422. Plots ofenzyme-bound label remaining (%) versus timeforthe isolated2-deoxy-2-[18F]fluoro mannosyl-enzyme in buffer,and in the presence ofglucosyl benzene (197 mM) at 37°C 146Figure 4.23. ‘H-decoupled’9F NMRspectra ofwater-extractablefraction ofA) liver and B) kidney homogenates, and C) urine of rat administered2FJ3ManF (70 mg/kg) 153Figure 424. Transaxial, sagittal, and coronal views ofa rat. PET imageafter injection of— 1 mCi of 2-deoxy-2-[18F]fluoro-J3-D-mannopyranosyl[18F]fluoride 155Figure ALl. Spontaneous hydrolysis of 2FcxGluF (10.7 mM) at 50.0°C, pH 6.8 187Figure A12. Spontaneous hydrolysis of 2F/3G1uF (9.24 mM) at 50.0°C, pH 6.8 187Figure AI.3. Spontaneous hydrolysis of5FJ3IdoF (7.65 mM) at 50.0°C, pH 6.8 188Figure AI.4. Spontaneous hydrolysis of 5FciJdoF (7.90 mM) at 50.0°C, pH 6.8 188Figure MS. Spontaneous hydrolysis of 5FaGluF (4.96 mM) at 50.0 C, pH 6.8 189Figure AJ.6. Spontaneous hydrolysis ofSFf3GluF (8.10 mM) at 50.0°C, pH 6.8 189xl’LIST OF SCHEMESScheme 2.1. Synthesis of5-fluoro f3-D-glucosyl fluoride (2.3) and5-fluoro a-L-idosyl fluorides (2.9) 32Scheme 2.2. Synthesis of5-fluoro a-D-glucosylfluoride (2.4) and5-fluoro J3-L-idosyl fluorides (2.14) 34Scheme 3.1. Synthesis of2-chloro-2-deoxy-2-fluoro-a-D-glucopyranosylchloride (3.1) and attempted synthesis of2-chloro-2-deoxy-2-fluoro-a-D-mannopyranosyl chloride (3.2) 85Scheme 3.2. Synthesis of 2,2 -dWuoro- a-D-arabinohexopyranosyl chloride (3.3) 87xJnLIST OF TABLESTable 2.1. and 19FNMR chemical shjfts for 5-fluoro D-glucosylfluorides 36Table 2.2. Coupling constantsfor 5-fluoro D-glucosylfluorides 36Table 2.3. 1H and 19FNMR chemical shifts for 5-fluoro L-idosylfluorides 37Table 2.4. Coupling constants for 5-fluoro L-idosylfluorides 37Table 2.5. Spontaneous hydrolyses rates of5-fluoro glycosylfluorides in 50 mMphosphate buffer, pH 6.8, IMNaC1O4 41Table 2.6. Kinetic parameters of J3GluF, 2FJ3G1uF, 2dJ3PNPG1u, 5F/3GluF,and SFcxJdoF with Agrobacterium /3-glucosidase 61Table 2.7. Kinetic parameters of aGluF, 2FaGluF, 5FaGluF, and SFJ3IdoFwith yeast a-glucosidase 63Table 3.1. Size comparison of C-H, C-F, C-Cl, and C-OH groups 83Table 3.2. Kinetic parameters of2Cl2FaGluCl and 2,2FaAraClwith yeast a-glucosidase and Jack bean a-mannosidase 102Table 4.1. Edman degradation ofactive site peptide in GCase 118Table 4.2. Tissue distribution of2 -deoxy-2 [18F]fluoro-J3-D-mannopyranosyl[18F]fluoride in male Wistar rats 148Table 4.3. Tissue distribution of2 -deoxy-2 [18F]fluoro-J3-D-mannopyranosyl[18F]fluoride infemale Sprague-Dawley rats at 60 mm post-injection,control andpre-treated with 2FJ3ManF 151xivLIST OF ABBREVIATIONS2,2FaAraCl: 2-Deoxy-2,2-difluoro-a-D-arabinohexopyranosyl chloride2Cl2FocGluCl: 2-Chloro-2-deoxy-2-fluoro-a-D-glucopyranosyl chloride2C12FaManC1: 2-Chloro-2-deoxy-2-fluoro-a-D-mannopyranosyl chloride2d[3PNPG1u: p-Nitrophenyl 2-deoxy-3-D-arabinohexopyranoside2FcGluF: 2-Deoxy-2-fluoro (x-D-glucopyranosyl fluoride2FccManF: 2-Deoxy-2-fluoro x-D-mannopyranosyl fluoride2FI3Ga1DNP: 2 ‘,4 -Dinitrophenyl 2-deoxy-2-fluoro-3-D-galactopyranoside2Ff3G1uF: 2-Deoxy-2-fluoro f3-D-glucopyranosyl fluoride2F[3ManF: 2-Deoxy-2-fluoro-f3-D-mannopyranosyl fluoride2FG1u: 2-Deoxy-2-fluoro D-glucopyranosyl5FaG1uF: 5-Fluoro-a-D-glucopyranosyl fluoride5FaIdoF: 5-Fluoro-z-L-idopyranosyl fluoride5F[3GIuF: 5-FIuoro--D-glucopyranosyl fluoride5Ff3IdoF: 5-Fluoro--L-idopyranosyl fluoride5FG1u: 5-Fluoro-D-glucopyranosyl5FIdo: 5-Fluoro-L-idopyranosylaPNPG1u: p-Nitrophenyl a-D-glucopyranosidexPNPMan: p-Nitrophenyl ct-D-mannopyranosidej3DNPGa1: 2’, 4’-Dinitrophenyl f-D-galactopyranosideI3DNPG1u: 2’, 4’-Dinitrophenyl f3-D-glucopyranosideI3IPTG1u: Isopropylthio 3-D-glucopyranosideI3PNPFuc: p-Nitrophenyl D-fucopyranoside3PNPG1u: p-Nitrophenyl 3-D-g1ucopyranosideBSA: Bovine Serum AlbuminCBE: Conduritol B EpoxideDCI-MS: Desorption Chemical Ionization Mass SpectrometryDCME: 1,1 -Dichioromethyl methyl etherDMF: N,N-Dimethyl formamideDNJ: 1 -DeoxynojirimycinE.C.: Enzyme Commission (classification number) of the InternationalUnion of BiochemistryEDTA: Ethylenediamine tetraacetic acid, disodium saltEOB: End of BombardmentES/MS: Electrospray Ionization Mass SpectrometryxvESMS: Electrospray Ionization Mass SpectrometryFDG: 2-Deoxy-2-fluoro-D-glucoseGCase: GlucocerebrosidaseG1uNAc: N-Acetyl-glucosamineHPLC: High Performance Liquid ChromatographyID/g: Injected Dose per g of TissueISV: Ion Source VoltageLC: Liquid ChromatographyLSIMS: Liquid Secondary Ionization Mass SpectrometryMan: MannoseMSIMS: Tandem Mass SpectrometryMS: Mass SpectrometryNMR: Nuclear Magnetic Resonance SpectroscopyOR: Orifice Plate EnergyPET: Positron Emission TomographyPNP: p-NitrophenylTFA: Trifluoroacetic AcidTIC: Total Ion ChromatogramTLC: Thin Layer ChromatographyUV: UltravioletKinetic Constantsk2: Glycosylation rate constantk3: Deglycosylation rate constantk: Maximum rate constant for catalysis (turnover number)K1: Dissociation constant for enzyme-inhibitor complexk1: Maximum rate constant for inactivationK’: Apparent dissociation constant under steady-state conditionsKm Michaelis-Menten constant of a substratekobs: Pseudo-first order rate constant for inactivationVm: Maximum rate of enzyme-catalyzed reactionxviACKNOWLEDGEMENTSI wish to thank my supervisors, Dr. Steve Withers and Dr. Mike Adam, for theiradvice, guidance, and encouragement. The financial support of the Natural Sciences andEngineering Research Council is gratefully acknowledged. I wish to thank Ms. KarenRupitz, Ms. Salma Jivan, and Ms. Kelly Hewitt, for expert instruction and technicalassistance; Dr. Neil Hartman for advice and help with the animal studies; Drs. RuediAebersold, Shichang Miao, David Burgoyne, and Curtis Braun for assistance withelectrospray mass spectrometry; Dr. Lawrence McIntosh for assistance with NMRspectroscopy; Dr. Marie Grace (Mt. Sinai Medical Centre, New York) and Dr. JohnCallahan (Hospital for Sick Children, Toronto) for providing human lysosomal f3-glycosidases; and Mr. Wai Yeung for the syntheses and kinetic analyses of severalcompounds. I would also like to thank everyone in the Withers group for their friendships,for the many stimulating and helpful discussions we have had, and for helping to make mytime here an enjoyable one.Thanks especially to my family and friends for their support.xviiCHAPTER IGENERAL INTRODUCTION11.1 GlycosidasesGlycosidases, or glycoside hydrolases, constitute a large family of enzymes whichcatalyze the hydrolysis of glycosidic linkages (Figure 1.1) by the cleavage of the C-O bondbetween a sugar (glycone) and an aglycone (which may be a second sugar, or other group).The glycoside substrates of glycosidases thus comprise two structural units: the sugarglycone, which is bound in the catalytic site of the enzyme; and the aglycone, which isbound in an adjacent subsite. Although the aglycone in many natural substrates is anothersugar residue, many glycosidases exhibit a liberal aglycone specificity, catalyzing thecleavage of a variety of anomeric substituents, including C-O-Aryl, C-S, C-N, and C-Fbonds. Interest in these enzymes is fuelled by their applications in industry (e.g. invertasefor the production of ‘invert’ sugar, cellulases for the degradation of cellulosic biomass,xylanases in pulp-and-paper production) and by their relevance to cancer metastasis,influenza, AIDS and other viral diseases, diabetes, and lysosomal storage disorders.GlycosidaseOR HO OH+ ROHGlycone AglyconeFigure 1.1. Reaction normally catalyzed by a glycosidase.1.2 Classification of GlycosidasesGlycosidases may be divided into smaller subgroups according to:(1) Anomeric configuration (a or $) of the substrate. For example, in general, af-glucosidase will catalyse the hydrolysis of f3-glucosides but not of cz-glucosides, whilean a-glucosidase is specific for a-glucosides.2(2) Glycone specificity. A particular glycosidase usually exhibits maximal activitywith a specific glycone, and is generally classified according to the glycone against which itis most active. An a-mannosidase, for example, is most active against x-mannosides, butother a-glycosides may be substrates as well.(3) Stereochemical outcome of the reaction. A glycosidase is termed “inverting” or“retaining” depending on the relative anomeric configurations of the substrate and of theproduct initially released from the enzyme. Hydrolysis catalyzed by a retaining glycosidaseleads to the formation of a product with the same anomeric configuration as the substrate,while hydrolysis via an inverting enzyme gives a product of the opposite anomericconfiguration to that of the substrate (Figure 1.2).Figure 12. Hydrolyses catalyzed by retaining and inverting enzymes (shown for J3-glycosidases).These different stereochemical outcomes reflect the existence of two distinct mechanisticclasses of glycosidases. The two mechanisms differ in that inverting glycosidases operatevia a direct displacement of the leaving group by water, while retaining glycosidases utilizea double-displacement mechanism involving a glycosyl-enzyme intermediate. Despite thesedifferences, it is noteworthy that both classes employ a pair of carboxylic acids in the activeRetainingHO.Inverting3site. With inverting enzymes, one residue acts as a general acid and the other as a generalbase, whereas with retaining enzymes one functions as a general acid and a general basewhile the other acts as a nucleophile and a leaving group (Figure 1.3).RETAINING MECHANISMHOO-o\INVERTING MECHANISMOH0..HHFigure 1.3. Presumed mechanisms of retaining and inverting J3-glucosidases.1.3 The Catalytic Mechanism of Retaining GlycosidasesA general mechanism for retaining glycosidases, which are the focus of this study,was proposed in 1953 (Koshland, 1953) and is shown in detail in Figure 1.4 (in this case,for a 3-glucosidase). The first step involves displacement of the aglycone by anappropriately positioned carboxylate nucleophile in the active site to form an cz-glycosylenzyme intermediate (the ‘glycosylation’ step). In a second step, the intermediate is-ooHOHO\,4hydrolysed to yield a-glucose as the initial product, regenerating the free enzyme (the‘deglycosylation’ step). Both the formation and the hydrolysis of the glycosyl-enzymeintermediate occur via oxocarbenium ion-like transition states. Cleavage of the C-O bondin this model is assisted by general acid catalysis by an appropriately positioned carboxylicacid residue while hydrolysis of the intermediate likely occurs with general base catalysisby the same residue. Although not explicitly shown in Figure 1.4, much of the enzymaticrate acceleration is proposed to be derived from non-covalent binding interactions betweenthe enzyme and the substrate.A variation on the retaining mechanism involving stabilization of an ion-pair ratherthan formation of a covalent intermediate has long been popular (Phillips, 1967) and cannotbe discounted for some glycosidases. However, considerable evidence in favour of acovalent intermediate exists, including kinetic isotope effect studies and actual trapping ofsuch an intermediate. This evidence is described in a number of reviews (Legler, 1990;Sinnott, 1990; Svensson & Sogaard, 1993). An alternative ring-opening mechanism forretaining glycosidases has been recently resurrected (Franck, 1992), but lacks seriousexperimental support (Sinnott, 1993).1.4 Evidence for a Double Displacement Mechanism in RetainingGlycosidases1.4.1 Presence of a carboxylate nucleophileX-ray crystallographic studies have invariably revealed the presence of glutamic oraspartic acid residues in the active sites of glycosidases. Active site carboxylic acids havebeen identified, for example, in the retaining (3-glycosidases hen egg white lysozyme(Imoto et al., 1972), Cellulomonasfimi exoglycanase (White et al., 1994) and in E. coli fgalactosidase (Jacobsen et al., 1994), and in the retaining a-glycosidase Taka c.t-amylase(Matsuura et al., 1984) which are appropriately located to function as catalytic nucleophiles(or charge stabilizers). More compelling evidence derives from X-ray5HOHOOOHO%IcFigure 1.4. Preswnedmechanism of a retaining f3-glucosidase.11Oo‘0H16structures of glycosidases co-crystallized with either inhibitors or substrates in the activesite (recent structures, e.g. (Campbell et al., 1993; Divne et a!., 1994), are reviewed in(McCarter & Withers, 1994)). In such structures, however, although the residues whichare important to catalysis are often apparent and the identities of the residues in close spatialproximity to the substrate or inhibitor are known, their specific roles in catalysis frequentlycannot be predicted from X-ray structural data alone. Other evidence for active sitecarboxylates is derived from labelling and mutagenesis studies, and will be discussed later.1.4.2 The nature of the giycosyl-enzyme intermediateThe double displacement mechanism for retaining glycosidases initially proposedby Koshland described the formation of a covalent glycosyl-enzyme intermediate. Indeed,low temperature and rapid-quench studies with radiolabelled versions of natural substrateshave resulted in the trapping of covalent glycosyl-enzyme intermediates in porcinepancreatic x-amylase (Tao et at., 1989) and in Streptococcus sobrinus cxglucosykransferase (Mooser et a!., 1991). However, it has been suggested (Blake et al.,1967), based on X-ray structural studies with hen egg white lysozyme, that a negativelycharged active site carboxylate (Asp-52) could stabilize a positively charged oxocarbeniumion intermediate sufficiently to allow diffusion of the leaving group out of the active siteand attack of this ion-pair by a glycosyl acceptor (water or another sugar). While ion-pairintermediates are still regularly invoked, particularly for lysozyme, there is considerableevidence which favours the existence of a covalent glycosyl-enzyme intermediate for manyretaining glycosidases.Strong evidence for the covalent nature of a glycosyl-enzyme intermediate isderived from a-secondary deuterium kinetic isotope effect measurements with substrates(deuterated at C-i) for which deglycosylation is rate-determining. Values of kH/kD = 1.20-1.25 have been determined for E. coli (lac Z) 3-galactosidase (Sinnott et a!., 1978), and7kH/kD = 1.10 - 1.12 for Agrobacterium 3-glucosidase (Kempton & Withers, 1992). Theselarge and positive kinetic isotope effects reflect significant sp3 to sp2 rehybridization at theanomeric centre as the reaction proceeds from a ground state glycosyl-enzyme intermediateto the transition state of the deglycosylation step. This strongly implies that theintermediate has considerably more sp3 character than the subsequent transition state,consistent with a covalent glycosyl-enzyme which is hydrolyzed via a transition state withsignificant oxocarbenium ion character. If the intermediate were truly an ion pair, aninverse kinetic isotope effect would be expected.Additional evidence for the covalent nature of the intermediate is derived from theinteraction of several 3-g1ycosidases with activated 2-deoxy-2-fluoro-3-D-glycosides(Withers et a!., 1988). These compounds inactivate retaining -glycosidases by theformation of a stabilized glycosyl-enzyme intermediate. The presence of the electronegativefluorine at C-2 destabilizes both oxocarbenium ion-like transition states, thereby slowingboth the formation and the hydrolysis of the intermediate, while the presence of the goodleaving group (fluoride or 2,4-dinitrophenolate) accelerates the glycosylation step, leadingto the accumulation of the intermediate and inactivation of the enzyme (Figure 1.5). In thecase of Agrobacterium f3-glucosidase, such glycosyl-enzyme intermediates can beextremely stable (with half-lives ranging from one to 500 hours at 37°C in buffer),permitting demonstration of the covalent nature and of the stereochemistry (a-linked) of the19 .mtermediate by F-NMR (Withers & Street, 1988), as well as identification of thenucleophilic residue as Glu-358 (Withers et at., 1990). Upon incubation with a suitableglycosyl acceptor, the enzyme may be reactivated and a disaccharide product isolated,demonstrating that the intermediate is catalytically competent. The interaction of retainingglycosidases with this class of inhibitiors will be discussed in further in greater detail inSection NO2Figure 1.5. Inactivation of Agrobacterium /3-glucosidase by trapping of a 2-deoxy-2-fluoro-glucosyl-enzyme.Finally, spontaneous (non-enzymatic) hydrolysis studies of various glycosidesindicate that glycosyl cations are very unstable species with vanishly brief lifetimes-10 -12(estimated to be 10 to 10 s in aqueous solution) (Banait & Jencks, 1991; Bennet &Sinnott, 1986) while the lifetimes of covalent glycosyl-enzymes at ambient temperatures are1 to 100 ms (Weber & Fink, 1980). It seems unlikely that an unstable ion-pair would existlong enough, even within an extremely favourable enzyme active site microenvironment, toallow diffusion of the leaving group out of the active site and attack by water prior tocollapse to a covalent species (Sinnott & Souchard, 1973).1.4.3 Oxocarbenium ion-like transition statesKinetic isotope effect studies provide evidence for oxocarbenium ion-like transitionstates associated with both the glycosylation and deglycosylation steps of glycosidasecatalyzed hydrolysis. By selection of appropriate aglycones, it is possible to obtainsubstrates for which either step is rate-limiting, and the isotope effect on each step may thenbe determined. The E. coli (lacZ) -galactosidase-catalyzed hydrolysis of -galactosylpyridinium salts exhibited a-secondary deuterium kinetic isotope effects of kHJkD = 1.15 -1.20, indicating substantial sp2 character in the transition state for formation of thegalactosyl-enzyme since glycosylation is rate-limiting for these substrates (Sinnott &Withers, 1974). Even larger isotope effects are reported for this enzyme whenFF9deglycosylation is rate-determining, with k/kD = 1.2 - 1.25 for 2,4 -dinitrophenyl 3-galactosides (Sinnott & Souchard, 1973). Similar results were obtained withAgrobacterium 3-g1ucosidase (Kempton & Withers, 1992), with kH/IcD values for theglycosylation and deglycosylation steps of kR/lcD = 1.06 and kH/kD = 1.11, respectively.Glycosidases are strongly inhibited by compounds which appear to mimic aglycosyl oxocarbenium ion in conformation or charge, presumably functioning as transitionstate analogues. In a glucosyl oxocarbenium ion (Figure 1.6), in contrast to a ground stateglucoside, C-i and 0-5 share a full positive charge, while C-i, C-2, C-5, and 0-5 arerequired to be co-planar (Sinnott, 1987). To fulfill these requirements, the glycosyloxocarbenium ion may theoretically adopt either a half-chair (4H3 or or a boat (2’5BorB2,5) conformation.8+Glucosyl cationFigure 1.6. Comparison of transition state glucosyl oxocarbenium ion and ground stateglucoside. The 4F13 half chair conformer of the transition state oxocarbenium ion isshown.Putative transition state analogues include aldonolactones (1.1) and aldonolactams(1.2) which resemble the glycosyl oxocarbenium ion both in geometry, and to someextent, in charge (Figure 1.7). K values for these compounds are often i02 to 104-foldlower than K values for the corresponding hexoses (Legler, 1990).Glucoside10Figure 1.7. Resonance structures of aldonolactones (1.1) and aldonolactams (1.2).Other potent glycosidase inhibitors of this type include the amino sugar nojirimycin (1.3,4Figure 1.8) and its vanous analogues. Such compounds may bind up to 10 -fold tighterthan the corresponding ground state analogues (Legler, 1990). K values of theseinhibitors with the corresponding glycosidases are typically in the micro- or submicromolarrange. The K1 values of nojirimycin, galactononojirimycin, and mannonojirimycin with A.wentii -g1ucosidase, E. coli -ga1actosidase, and Jack bean ct-mannosidase are 0.36,0.045, and 1.2 M, respectively (Legler, 1990). Nojirimycins are presumed to resemble aglycosyl oxocarbenium ion (Figure 1.6) in geometry and charge. Upon protonation,nojirimycin is isoelectronic with a glucosyl oxocarbenium ion; the dehydrated form wouldappear to be nearly isosteric as well. However, for such inhibitors to be true transitionstate analogues, their inhibitory efficiencies should correlate with the glycone specificitiesof the corresponding substrates (Fersht, 1985; Wolfenden, 1972). This is not always thecase, and tight binding may result from fortuitous interactions between the inhibitors andresidues in the enzyme activite site. However, a correlation has indeed been shownOH1.111between a related class of transition state analogues, nojiritetrazoles (1.4, Figure 1.8), anda series of glycosidases (Ermert et a!., 1993).HO—%.,.c -H20.OHHOL..NVOH1.3Figure 1.8. Nojirimycin (1.3), in both hydrated and dehydrated forms, andnojiritetrazole (1.4).1.4.4 General acid catalysisIn addition to the enzymic nucleophile, a second carboxylic acid in the active site ofretaining glycosidases located on the face of the sugar opposite to the nucleophile, isproposed to assist departure of the aglycone by general acid catalysis in the Koshlandmechanism (Figure 1.4). Several crystal structures of both [3- and a-glycosidases,including complexes of hen egg white lysozyme (Imoto, 1972) and porcine pancreatic aamylase with appropriate inhibitors (Qian et a!., 1994), show suitably disposed carboxylicacids (Glu-35 and Glu-233, respectively) within hydrogen bonding distance to theglycosidic oxygen (or ‘glycosidic’ nitrogen) of bound inhibitors. In the amylase structure,a third residue, Asp-300 is also apparently hydrogen-bonded to Glu-233 and to the+OHOH1.412‘glycosidic’ nitrogen of the pseudotetrasaccharide inhibitor acarbose, and is equally likelyto play a role in acid/base catalysis.Supportive evidence for acid catalytic assistance in glycoside hydrolysis is derivedfrom the glycosidase-catalysed hydration of octenitol derivatives. E. coli 3-galactosidasecatalyzes the hydration of D-galacto-octenitol (1.5) to a galactooctulose derivative; theproduct of this reaction was characterized by 1H-NMR (Lehmann & Schlesselmann,1983). Based on the stereochemistry of the hydrated product, protonation occurred fromthe [i-face of the galactose ring, consistent with the placement of a residue in this enzymethat could function as an acid catalyst in the normal hydrolysis reaction (Figure 1.9).Glycals are also hydrated by glycosidases to yield 2-deoxy derivatives (Wentworth &Wolfenden, 1974). In contrast to the results with the exocyclic double bond in the octenitolderivative, deuteration (or protonation) occurs from the a-face of the endocycic doublebond in D-galactal (1.6) (Lehmann & Zieger, 1977).H20____HO.1 -GaIactosidasecDL (.cv D20DO.1.6Figure 1.9. Hydration ofD-galacto-octenitol (1.5) and ofD-galactal (1.6) by E. coli J3-galactosidase. Protonation (or deuteration) occurs from the J3-face of the sugar in theoctenitol derivative andfrom the a-face in galactal.OHD13With most glycosidases studied, deuteration (or protonation) of glycals by retainingglycosidases occurs on the face of the sugar opposite to that on which the acid catalyst isassumed to be located. Hydration of cellobial, for example, catalyzed by Irpex lacreus exo-(1,4)-celluIase and Aspergillus niger endo--(1,4)-cellulase in 1)20 led to deuterationfrom the x-face of cellobiose (Kanda et a!., 1986). With glycals, it is believed that thedeuteron (or proton) is delivered by the catalytic nucleophile (Hehre et a!., 1977; Kanda eta!., 1986) via a cyclic six-membered transition state.Although a general acid catalytic residue is implicated in the mechanism of catalysisof many glycosidases, acid catalysis is apparently not essential for hydrolysis of allsubstrates with all glycosidases. The efficient E. coli 3-ga1actosidase-cata1ysed hydrolysis(108 to 103-fold rate enhancements) of 3-D-galactopyranosyl pyridinium salts (1.7,Figure 1.10), which cannot accept a proton in a manner which will assist aglyconedeparture, indicates that acid catalysis is not essential to the catalytic activity of this enzyme(Sinnott & Withers, 1974). However, this enzyme also requires Mg2 for maximal activitywith aryl galactosides (although not with galactosyl pyridinium salts) (Sinnott, 1978). Themetal ion may have a role in orienting an acid/base catalytic residue (Sinnott & Souchard,1973) or may act directly as an electrophilic catalyst (Sinnott, 1990).OHHOOFigure 1.10. Galacrosyl pyridinium salt.1.4.5 Non-covalent enzyme-substrate interactionsBoth stabilization of a developing oxocarbenium ion by an enzymic nucleophile andacid catalytic assistance contribute substantially to the rate accelerations achieved by14retaining glycosidases, but these do not fully account for the catalytic power of theseenzymes. In general, enzymes achieve spectacular (possibly up to 107-fold (Guthrie,1977)) rate enhancements, in some cases approaching the limit of diffusion control, bystabilization of the transition states for each of the steps involved, a concept elaborated byPauling (Pauling, 1946). This stabilization of the transition state and consequent decreasein activation energy may be directly attributed to tighter binding of the transition state thanthe ground state by the enzyme, preferentially stabilizing the transition state throughspecific binding interactions (see (Fersht, 1985) for a detailed discussion).It has been shown that tyrosyl t-RNA synthetase utilizes a complex network ofhydrogen bonds to achieve ground state specificity and to stabilize the transition state of thereaction (Fersht, 1987). In that study, mutant enzymes were prepared using site-directedmutagenesis to delete specific enzyme/substrate interactions (identified from the availableX-ray crystal structure). Comparison of kinetic data for these mutants with that for thewild-type enzyme afforded information about the contribution of each mutated residue’snon-covalent binding interactions to catalysis. Such interactions likely account for virtuallyall of the rate increase afforded by this particular enzyme since no other residues which canparticipate catalytically (e.g. as nucleophiles, acid catalytic groups etc.) were identified.In glycosidases, these non-covalent binding interactions likely comprisehydrophobic interactions and hydrogen bonding. The numerous hydroxyl groupsassociated with most carbohydrates suggest that hydrogen bonding plays an important rolein binding and catalysis by enzymes with carbohydrate substrates. Indeed, elaboratehydrogen bonding networks between proteins and sugar ligands have been elucidated fromX-ray crystal data in complexes of, for example, L-arabinose binding protein with Larabinose (Quiocho et a!., 1989), Aspergillus awamori glucoamylase withdeoxynojirimycin (Harris et a!., 1993), f-amylase with maltotetmose (Mikami et al.,1994), and ct-amylase with acarbose (Qian et al., 1994).15Deoxygenated glycosides have been shown to be useful probes of hydrogenbonding interactions in a number of glycosidases and glycosyltransferases, for example in,rabbit muscle glycogen phosphorylase (Street et al., 1989), glucoamylase (Sierks et al.,1992), and E. coli 13-galactosidase (McCarter et at., 1992). Replacement of a hydroxyl at aparticular position on the normal substrate by hydrogen, a stericafly conservativesubstitution, results in the abolition of all hydrogen bonding interactions with the enzyme atthat position. This will be reflected in increased activation energies for the catalytic steps,and consequent rate decreases relative to the parent compounds. These binding interactionswere estimated to range from 10 to 18 kJ/mol for each of the 4’- and 6’-hydroxyls ofmaltose (Sierks et at., 1992), and from about 20 kJ/mol for the 3-, 4-, and 6-hydroxyls toabout 33 kJ/mol for the 2-hydroxyl of galactose in 3-galactosidase (McCarter et at., 1992),indicating that non-covalent interactions with each of the hydroxyls contributedsubstantially to enzymic rates. The very significant contribution of the 2-hydroxyl totransition state stabilization in retaining glycosidases has been noted previously (seeWolfenden & Kati, 1991, and references therein).1.5 Glycosidase Inhibitors as Structural and Mechanistic Probes inBiomedicineGlycosidase activity is relevant to a number of clinically important processes. Thevarious lysosomal storage diseases, characterized by accumulation of specific glycolipids inthe lysosomes of affected individuals, result from genetic deficiencies in the specificlysosomal glycosidases which normally catabolize these glycolipids. Diverse disorders,always debilitating and often fatal, such as Gaucher disease, Tay-Sachs disease, and GM1gangliosidosis, are associated with various defects in a great number of different lysosomalglycosidases (Neufeld, 1991). One such enzyme, glucocerebrosidase (glucosylceramidaseor acid f3-glucosidase), the deficiency of which causes Gaucher disease, has been usedwith some success to treat patients with Type 1 Gaucher disease by enzyme replacement16therapy (Barton et a!., 1990). Glycosidase inhibitors have been useful in producing animalmodels of lysosomal storage diseases in rats and mice by inhibition of normal lysosomalglycosidase activity in vivo (Atsumi et at., 1993; Kanfer et at., 1982). They have also beenused as anti-cancer and anti-viral agents by interference with cell surface and proteinglycosylation (Bemacki et a!., 1985; Elbein et at., 1984). Another therapeutic applicationof such inhibitors is the management of diabetes by the inhibition of intestinal aglucosidase activity (Martin et at., 1991; Rios, 1994).The kinetic evaluation of glycosidase inhibitors with wild-type or mutant enzymescan provide valuable mechanistic insights, which in the case of enzymes of clinicalimportance, may aid in the design of more effective therapeutic drugs. Such inhibitors areextremely valuable in providing information about the roles of specific residues in catalysisthrough high resolution X-ray structures of enzyme-inhibitor complexes, but are even moreindispensable in the absence of such structural data. Non-covalent transition-state analogueinhibitors (Section 1.4.3) may afford insights into transition state structure. Additionally,covalent inhibitors may enable identification of specific amino acid residues involved incatalysis, in conjunction with sequence alignments of homologous enzymes and detailedkinetic analyses of site-specific mutants. Active site residues which may play catalyticallyessential roles in the mechanism of a glycosidase may be identified by derivatization with avariety of covalent inhibitors, discussed below.1.5.1 Affinity labelsAffmity labels are composed of an inherently reactive functional group which canreact with derivatizable residues on the enzyme, and a sugar moiety, which confersspecificity for the active site of the glycosidase. As the reactive group in the molecule mayreact with a number of residues, use of affinity labels often results in labeffing of multipleresidues on the enzyme. Because of this lack of specificity, affinity labels in general are oflimited utility for the identification of specific catalytic residues or elucidation of their roles,17other than to indicate their presence in or near the active site. The reactive nature of suchcompounds renders them inappropriate for use as drugs. Examples of this class ofcovalent inhibitors include N-bromoacetyl glycosylamines (1.8) and glycosylisothiocyanates (1.9) (Figure 1.11).HO—Figure 1.11. N-bromoaceiyl glyco.sylamines (1.8) and glycosyl isothiocyantes (1.9),affinity labels ofglycosidases.Appropriate N-bromoacetyl glycosylamines have been used to inactivate E. coli 3-galactosidase (Naider et al., 1972), Agrobacterium faecalis f3-glucosidase and C. fimiexoglycanase (Black et al., 1993). The presence of several Agrobacterium f3-glucosidasespecies derivatized by one, two and three N-acetylglucosaminyl moieties was shown byelectrospray mass spectrometry, indicating modification of several residues on this enzyme.However, on occasion N-bromoacetyl glycosylamines may react specifically with a singleresidue, as in the case of E. coli 3-galactosidase and C. fimi exoglycanase. In E. coli f3-galactosidase, the single active site amino acid labelled was identified as Met-502, whichwas, however, later shown not to participate in catalysis. In the case of C. fimiexoglycanase, the labelled amino acid was identified as Glu-127, a conserved residue inthis glycosidase family, whose coniribution to catalysis was demonstrated by site-directedmutagenesis. Sweet almond -glucosidase was inactivated by the glucosyl isothiocyanate(Shulman et al., 1976) but no labelled residues were identified.A large number of epoxyalkyl derivatives have been used to inactivate a variety ofglycosidases. Upon protonation of the epoxide either by the acid catalyst or by some otherH1.818residue, an enzymic nucleophile may then attack, resulting in irreversible inactivation of theenzyme (Figure 1.12). Various epoxyalkyl glycosides (1.10) have been used to label anumber of glycosidases, including hen egg white lysozyme (Moult et al., 1973), sweetalmond and Aspergillus wentii 3-glucosidases (Bause & Legler, 1974), and cellulases fromAspergillus niger, Aspergillus wentii, Oxyporus sp (Legler & Bause, 1973) and Bacillusmacerans (Keitel et al., 1993). In many instances, the rate of inactivation varies with thelength of the alkyl chain. In lysozyme, the residue labelled by an epoxyalkyl chitobiosidederivative was identified as Asp-52 (Moult et a!., 1973), the residue previously proposedon the basis of structural data and mutagenesis to act as the catalytic nucleophile (or charge-stabilizer) in the normal mechanism.HO_AFigure 1.12. Reaction of an epoxyalkyl glucoside (1.10) with a 13-glucosidase.1.5.2 Conduritol epoxidesThe conduritol epoxides are a related class of inhibitors which incorporate anendocylic epoxide within a cyclitol ring, this structure resembling the sugar ring of thenatural substrate. A number of these epoxides, with structures, corresponding to those ofthe substrates of a range of glycosidases, have been synthesized (e.g. conduritol B epoxide(1.11) is the “gluco” analogue). The mechanism of inactivation is similar to that of theepoxyalkyl glycosides, with protonation of the epoxide and attack by an enzymicnucleophile leading to formation of a covalent adduct. These inactivators are likely to beAHHO.1.1019more specific since nucleophilic attack of the endocyclic epoxide occurs at a centreanalogous to the anomeric centre of the substrate (Figure 1.13). A wide range ofglycosidases has been successfully inactivated by these compounds (Legler, 1990). Thesehave also been used in vivo to inactivate glucocerebrosidase in mice, providing a possibleanimal model for Gaucher disease (Kanfer et al., 1982).AHo_)Figure 1.13. Reaction of conduritol B epoxide (1.11) with a J3-glucosidase. Theresidues involved in protonation and nucleophilic attack may not necessarily be the catalyticresidues in the normal mechanism.Since rotation of the molecule by 180° about an axis bisecting the epoxide ringplaces the epoxide and all hydroxyls in the correct orientations to mimic an a-glucoside(Figure 1.13), conduritol B epoxide is an effective inactivator of both 13- and aglucosidases. With 13-glycosidases, attack of the catalytic nucleophile from the a-face ofthe cyclitol ring would give products with the favoured trans-diaxial stereochemistry, whilewith a-glycosidases, attack of this catalytic residue from the 13-face would produce the lessfavoured trans-diequatorial products. Indeed, 13-glycosidases are more effectivelyinactivated by these compounds than are a-glycosidases, (Legler, 1990). However,conduritol epoxides have shown an unfortunate propensity to specifically, but ‘incorrectly’,label active site residues other than the normal catalytic nucleophiles, on occasion leading toincorrect assignment of the catalytic nucleophile. For example, inactivation of E. coli 13-20galactosidase with the the appropriate conduritol epoxide resulted in labelling of Glu-461(Herrchen & Legler, 1984). Subsequent mutagenesis studies revealed that the glycinemutant modified at this position displayed a significant level of activity inconsistent withremoval of the catalytic nucleophile, suggesting that this epoxide had originally labelledanother residue in the active site. Re-investigation of this enzyme using a new class ofinactivators, the 2-deoxy-2-fluoro glycosides (Section 1.5.4), resulted in the identificationof the actual nucleophile, Glu-537, mutants modified at this position exhibiting theexpected kinetic behaviour (Gebler et a!., 1992; Yuan et a!., 1994). The absence of the C-6hydroxymethylene substituent in conduritol epoxides present in the sugar substrates likelyaccounts for the ‘incorrect’ labelling observed, by permitting alternative binding modes inthe enzyme active site, and subsequent reaction with other residues.1.5.3 Enzyme-activated inhibitorsEnzyme-activated, or mechanism-based, inhibitors are relatively chemically inertspecies which require an activation process based on the normal mechanism of the enzymeto subsequently react covalently at the active site. Except for their inherent reactivity,particularly at less than neutral pH, conduritol epoxides could arguably be included in thiscategory. Similarly, a glycosylmethyl triazene (1.12, Figure 1.14) requires protonation(either non-specifically or by a specific active site residue) before yielding a reactive specieswhich decomposes rapidly to produce nitrogen, an arylamine and a highly reactiveglycosylmethyl carbonium ion in the active site. This electrophiic species may then rapidlyreact with active site residues. Two active site residues in E. coil -galactosidase, Met-SO 1and Glu-461, both of which were also identified by other labels, were labelled andidentified by use of these inactivators (Marshall et a!., 1980; Sinnott & Smith, 1976).21NO2+ + N2NH2Figure 1.14. Generation of a reactive glycosyl carbonium ion from a glycosylmethyltriazene.A different mechanism by which a reactive species is produced in the active site isby enzymatic cleavage of the glycosidic bond of an inherently unreactive glycoside,releasing an aglycone which rearranges into a chemically reactive species. Inactivators ofthis type include difluoroalkyl glycosides (1.13) which can undergo enzyme-catalyzedcleavage, releasing a fluorohydrin which rapidly eliminates hydrogen fluoride to yield areactive acyl fluoride (Halazy et a!., 1989); or difluorotolyl glycosides (1.14) whoseaglycone, upon enzymatic release, again liberates hydrogen fluoride, but this time yieldinga reactive quinone methide (Halazy eta!., 1990) (Figure 1.15). A limitation of this strategyis that the reactive species, once generated, may react non-specifically with various residuesif it does not react immediately. Even if immediate reaction occurred, the labelled residuewould reflect the active site location at which the reactive species was generated and thereactivity of the labelled residue itself, and would not necessarily provide insights into thenormal mechanism. This rather indiscriminate reactivity also reduces their attractiveness aspotential drug candidates.1.1222HOOHHO—,— HOç4R1.13Figure 1.15. Generation of a reactive acyl fluoride and a quinone methide fromdifluoroalkyl glycosides (1.13) and dfluorotolyl glycosides (1.14), respectively.1.5.4 Trapping of a catalytic intermediate: 2-deoxy-2-fluoro glycosidesA new class of mechanism-based inactivators, 2-deoxy-2-fluoro glycosides withgood leaving groups, has been developed which function via the formation, andaccumulation of a stabilized glycosyl-enzyme intermediate analogous to that formed in thenormal enzymic mechanism (Withers et a!., 1987; Withers et al., 1988). Unlike theinhibitors described previously which contain a reactive group present or latent in themolecule, these compounds are 2-deoxy-2-fluoro analogues of normal, albeit artificial,substrates, that are less chemically reactive than the parent substrates. The substitution offluorine for the hydroxyl at C-2 slows both the formation and the hydrolysis of theintermediate, while the presence of the good leaving group (e.g. fluoride or 2,4-HOOH1.1423dinitrophenolate) accelerates only its formation, leading to trapping of this intermediate(Figure 1.5).The 2-hydroxyl group contributes substantially to transition state stabilisation inglycosidases, by involvement in non-covalent interactions with residues in the enzymeactive site (likely via hydrogen bonding) that are worth at least 30 kJ/mol (McCarter et al.,1992; Wolfenden & Kati, 1991). These interactions are diminished or abolished uponreplacement of this hydroxyl by fluorine, a substituent which may (weakly) act as ahydrogen bond acceptor but which cannot possibly act as a donor, resulting indestabilization of both transition states. Abolition of these interactions may also beachieved by the use of 2-deoxy glycosides, in which the hydroxyl is substituted by ahydrogen. Indeed, the deglycosylation step of p-nitrophenyl 2-deoxy f3-glucoside isslowed sufficiently to permit denaturation trapping of a 2-deoxy-glycosyl-enzymeintermediate in Aspergillus wentii -g1ucosidase and identification of the labelled residue(Roeser & Legler, 1981). In the 2-deoxy-2-fluoro glycoside, the highly electronegativefluorine also electronically destabilizes both electron-deficient transition states through polarand inductive effects to a greater extent than the hydroxyl of the parent substrate, slowingboth steps further. Both effects (binding and electronic) combined produce a massive (10to 108-fold) reduction in the rates of both steps (Street et a!., 1992; Street et a!., 1989).The good leaving group increases the rate of the glycosylation step relative to thedeglycosylation step, resulting in accumulation of a 2-deoxy-2-fluoro-glycosyl-enzymeintermediate.This inactivation mechanism is supported by demonstration of the stoichiometricreaction of inhibitor and enzyme both by electrospray mass spectrometry and bymeasurement of the magnitude of the “burst” of dinitrophenolate released (Street et a!.,1992); by demonstration (by 19F-NMR) of a covalent a-D-glycopyranosyl-enzymeintermediate (Withers & Street, 1988); by identification of the residue labelled afterproteolytic digestion and sequencing of the labelled peptide in several glycosidases, and24subsequent mutagenesis (Gebler eta!., 1992; Miao et a!., 1994; Tull et a!., 1991; Witherset al., 1990). Further, the catalytic competence of such intermediates is demonstrated bythe turnover of the intermediate in the presence of a suitable glycosyl acceptor which canbind in the aglycone subsite (Street et al., 1992; Withers et a!., 1990). A disaccharide isformed via a reaction analogous to a normal transglycosylation reaction, with attack by oneof the ring hydroxyls of the glycosyl acceptor at the anomeric centre of the 2-deoxy-2-fluoro-glucosyl-enzyme intermediate, giving a disaccharide product and active enzyme(Figure 1.16. Thus the enzyme is not irreversibly inactivated.-oO OHFigure 1.16. Turnover of a 2-deoxy-2-fluoro-glucosyl-enzyme intermediate bytransglycosylation.However, a significant limitation of such 2-deoxy-2-fluoro inhibitors is that they are onlyeffective against retaining -glycosidases, which form an a-glycosyl-enzyme intermediatein the normal enzymic mechanism. A number of retaining a-glycosidases,. whichpresumably form a f3-glycosyl-enzyme intermediate, were not effectively inactivated by thecorresponding a anomers of these 2-deoxy-2-fluoro glycosides (Withers et a!., 1988).1.6 Aims of this Thesis1) To develop new mechanism-based inhibitors of both 3- and a-glycosidases asprobes of enzyme mechanism and transition-state structure. These will be used in thederivatization of enzymes hitherto resistant to previous mechanism-based strategies in orderHO.25to identify specific active site residues. Such inhibitors may also be therapeutic drugcandidates.2) To examine the inhibition of specific glycosidases by mechanism-basedglycosidase inhibitors in vivo in order to develop new diagnostic probes for disorders ofglycosidase metabolism (e.g. lysosomal storage diseases), and to develop possible animalmodels of these diseases.26CHAPTER II5-FLUORO GLYCOSIDES AS MECHANISM-BASED INACTIVATORSOF BOTH - AND a-GLYCOSIDASES272.1 IntroductionGlycosidase inhibitors have proven to be valuable probes of enzymic mechanism(Lalegerie eta!., 1982; Legler, 1990) and show considerable promise as therapeutic drugs(Bernacki et al., 1985; Hughes & Rudge, 1994; Rios, 1994). Those described to dateinclude both reversible, non-covalent types and irreversible, covalent types. Inhibitors inthis latter category have proved particularly useful in identifying active site residues (seeSections 1.5.1 to 1.5.3). However, most such compounds are inherently chemicallyreactive, thereby limiting their usefulness in the specific identification of catalyticallyimportant residues. Further, their irreversibility, resulting in the permanent modification ofthe enzyme labelled, limits their potential as drugs. Retaining glycosidases are generallybelieved to follow a double-displacement mechanism in which a covalent glycosyl-enzymeintermediate is formed and hydrolysed via oxocarbenium ion-like Iransition states(Kempton & Withers, 1992; Koshland, 1953; Sinnott, 1990) (Figure 1.4). A successfulstrategy for inactivation of retaining 3-glycosidases involves the use of activated 2-deoxy-2-fluoro glycosides which form a stabiised 2-deoxy-2-fluoroglycosyl-enzyme intermediatethat turns over only very slowly (Section 1.5.4). Unfortunately, this approach has beennotably unimpressive with all a-glycosidases tested (McCarter et a!., 1993; Withers et a!.,1988). Further, the requirement for a fluorine at C2 limits the utility of these inhibitors ifthe enzyme is intolerant of substitution at this position (e.g., the C2 N-acetamido group ofN-acetylglucosaminidases). A novel approach to obviate both these problems and possiblyallow inhibition of both J3- and cx-glycosidases through accumulation of a covalentglycosyl-enzyme intermediate, without compromising specificity through substitution ofany ring hydroxyl, was attempted.5-Fluoro glycosides with good leaving groups might be expected to inactivateretaining glycosidases by formation of a stabilized 5-fluoro glycosyl-enzyme intermediatethrough a trapping mechanism analogous to that of the 2-deoxy-2-fluoro glycosides (seeFigure 1.5). A fluorine at C5 of a glycosyl oxocarbenium ion exerts electronic effects28similar to or greater than those of a C2 fluorine, both atoms being adjacent to cenires ofdeveloping positive charge (05 in the case of the 5-fluoro oxocarbenium ion and Cl in thecorresponding 2-deoxy-2-fluoro species). Indeed, modeling studies have indicated that thegreatest difference in partial charge between a ground state sugar and the correspondingglycosyl oxocarbenium ion is at 05 rather than Cl (Kajimoto et al., 1991; Winkler &Holan, 1989), thus the electronic effects of fluorine substitution at C5 might be expected tobe even greater than those at CiHowever, the reduction in the rates of both glycosylation and deglycosylation stepswith 2-deoxy-2-fluoro glycosides presumably results from both the electronic effects of theC2 fluorine and disruption of transition state binding interactions between the enzyme andthe usual C2 substituent. Such binding interactions should still be possible for the 5-fluoroglycosides since the C2 substituent is unaltered. Interactions at C2 in retainingglycosidases are of substantial importance (McCarter et al., 1992; Roeser & Legler, 1981;Wentworth & Wolfenden, 1974) and would be expected to significantly increase the ratesof both formation and turnover of a 5-fluoro glycosyl-enzyme intermediate relative to ananalogous 2-deoxy-2-fluoro species. Further, a C5 fluorine is sufficiently small thatrepulsive interactions involving this substituent would not be expected to be so large as topreclude binding to the enzyme. Indeed, valiolamine (2.1), a potent ct-glucosidaseinhibitor (Kameda et aL, 1984) produced by Streptomyces hygroscopicus, has an axial C5hydroxyl group. Another natural product, nucleocidin (2.2), has a fluorine situated at theanalogous C4 position on the furanose ring of this nucleoside analogue.2.2 Specific AimsInitially, the syntheses of the novel 5-fluoro glycosyl fluorides, 5-fluoro (-Dglucopyranosyl fluoride (2.3), and 5-fluoro x-D-glucopyranosyl fluoride (2.4) will beattempted.29The 5-fluoro -D-g1ucosyl fluoride (2.3) may be a mechanism-based inactivator of-glucosithses, like the analogous 2-deoxy-2-fluoro compound, in this case functioning bythe accumulation of a stabilized 5-fluoro glycosyl-enzyme intermediate. Similarly, the cxglucosyl fluoride (2.4) may be an inactivator of the corresponding cx-glucosidase.Inhibitors of x-glycosidases are of particular interest as potential agents for the treatment ofviral diseases and cancer (Hughes & Rudge, 1994), and for the management of diabetes(Rios, 1994).The rates of spontaneous hydrolysis of these and related compounds will bedetermined, and compared with those of the parent and 2-deoxy-2-fluoro glycosyl fluoridesto assess the effect of the CS and C2 fluorines on the electron-deficient transition statesinvolved in the spontaneous (and enzyme-catalyzed) hydrolysis of these glycosyl fluorides.No catalytic nucleophile in any a-glycosidase has yet been unequivocally identifiedby accumulation of a species analogous to the normal glycosyl-enzyme intermediate,without resort to denaturation trapping of a substrate (Mooser et al., 1991). Derivatizationand identification of the catalytic nucleophiles in an x-glucosidase by a novel techniqueutilizing electrospray mass spectrometry will be attempted. Such a technique removes theneed for the synthesis of radiolabelled inactivators. The derivatized enzymes will beproteolytically digested, and the labelled peptides identified by a variety of massspectrometric techniques.OHHOHOHOOH2.1 2.2Figure 2.1. Structures of valiolamine (2.1), and nucleocidin (2.2).F30OH2.4Figure 2.2 Structures of 5-fluoro J3-D-glucopyranosyl fluoride, 5FGluF (2.3) and 5-fluoro a-D-glucopyranosyl fluoride, 5FaG1uF (2.4).Results and Discussion2.3 Synthesisa) Synthesis of 5-fluoro 3-D-glucopyranosyl fluoride, 5FG1uF (2.3), and 5-fluoroct-L-idopyranosyl fluoride, 5Fc.ddoF (2.9)Synthesis of the 5-fluoroglycosyl fluorides hinged upon the known radicalphotobromination reaction at C5 of glucosides and glucosyl halides (Ferrier & Tyler, 1980;Praly & Descotes, 1987; Somsák & Ferrier, 1991). Broniination occurs mainly at eitherC5 or Cl since radicals at both centres are stabilized. Substitution at Cl can affect therelative product distribution, with electron-withdrawing groups disfavouring Clsubstitution. Thus photobromination of per-O-acetylated f3-glucosyl fluoride 2.5 (hv, Nbromosuccinimide, CCI4)yielded the protected 5-bromoglucosyl fluoride 2.6 in 41% yield(Scheme 2.1). Fluorination (Igarashi et al., 1969) of this compound with retention ofconfiguration at C5 (AgBFtoluene) afforded the protected 5-fluoro-3-D-glucosyl fluoride2.7 in low (11%) yield. Deacetylalion (NH3/COH) and chromatography (27:2:1EtOAc/CH3OH H0)on silica gel yielded 2.3 in 59% yield. Fluorination of 2.6 withAgF/CH3CNproceeded with inversion of configuration at C-5, affording the protected 5-fluoro-cz-L-idosyl fluoride 2.8 in 40% yield. Note that the anomeric configuration of thisF F2.331sugar is formally a, despite the fact that the configuration of the Cl fluorine is unchanged,since inversion at C5 formally renders this compound an L-idoside. Deacetylation andpurification of 2.8 as for the gluco compound gave 2.9 in 47% yield. All 5-fluorocompounds were characterized by ‘H and ‘9F NMR, elemental analysis, and highresolution mass spectrometry (Tables 2.1 to 2.4, and Chapter 5).OAchv/ NBS/ CC14, 5 h41%BTOAcF2.6AgF/CH3N, 6 h,/4O%2.7Scheme 2.1. Synthesis of 5-fluoro f3-D-glucosyl fluoride (2.3) and 5-fluoro a-L-idosylfluoride (2.9).OAc2.52.8NH3!CH3O , 0°C, 2.5 h 47%AgBF4/toluene,11%\OoC, 1 hNH3!CH3O , 0°C, 2.5 h 59%OOFHO2.932b) Synthesis of 5-fluoro ct-D-glucopyranosyl fluoride, 5FczGluF (2.4), and 5-fluoro3-L-idopyranosyl fluoride, 5FI3IdoF (2.14)Prolonged reaction of 2,3,4,6-tetra-O-acetyl a-D-glucosyl fluoride 2.10 with NBSeventually afforded the corresponding protected 5-bromo a-glucosyl fluoride 2.11 in 34%yield (Scheme 2.2). Treatment of this compound with AgBF4or with HF/pyridine resultedin extensive decomposition, but the 5-fluoro a-fluoride 2.12 was synthesized via the 5-fluoro f3-L-idosyl fluoride 2.13. Fluorination of 2.11 with silver fluoride (AgF,CH3N) proceeded with inversion of the C5 configuration, affording the 5-fluoro (3-L-idoside 2.13 in 45% yield. Treatment of the idosyl fluoride 2.13 overnight withHF/pyridine gave the evidentally thermodynamically more stable 5-fluoro gluco compound2.12 in 25% yield. Each of 2.12 and 2.13 were smoothly deprotected with anhydrousammonia in methanol, giving the deprotected 5-fluoro glycosyl fluorides 2.4 and 2.14,upon purification, in 81% and 98% yields, respectively.Both acetylated D-glucosides 5F[3G1uF (2.7) and 5FaG1uF (2.12) exhibit the4couplmg constants expected for D-glucosides and appear to be essentially undistorted C1chairs. However, the coupling constants of the acetylated 5-fluoro L-idosides indicate thatthese sugars are clearly not in the conformations adopted by the 5-fluoro D-glucosides.Analysis of the observed coupling constants and comparison with those of othersaccharides indicate that the conformations of both compounds lie on the boat/skew-boatpseudorotational itinerary (Figure 2.3). In the a-L-idoside (2.8), H-2, H-3 and H-4appear to be pseudoaxial and the observed coupling constants (J23 = J34 = 6.3 Hz)compare well with those measured for a series of a-D-glucosylpyridimum bromides (e.g.,= 6.8 Hz, J34 = 6.2 Hz), which have been assigned the 1S3 skew-boat conformation(Hosie et at., 1984; Hosie & Sinnott, 1985). In the -L-idoside (2.13), the magnitude ofJ2,3 (8.3 Hz) indicates an approximately anti-periplanar orientation of H-2 and H-3, whilethe small value of .134 (1.6 Hz) for this compound indicates a gauche orientation of H-333AcO hvl NBS/ CC14, 26 hOAcF 2.11Br0210 FAgBF4IAgF/CH3N, 48 h/’45% to1ueneHF/py, 15 h7—O OAc25% oAc2.5h 98% NHCH3O ,4h 81:OHF OHHOThcH ,’OH FOH1F F2.14 2.4Scheme 2.2. Synthesis of 5-fluoro a-D-glucosylfluoride (2.4) and 5-fluoro f3-L-ido.sylfluoride (2.14).and H-4. Similar coupling constants have been noted in 2,3,4-th-O-acetyl-D-xylono-1,5-lactone (J2,3 = 8.9 Hz, J34 = 2.5 Hz) from which a 2’5B boat conformation has beeninferred (Nelson, 1979). Idosides have been known to exhibit considerable confonnationalflexibility (Augé & Serge, 1984), and the possibility that the observed coupling constantsare an average of several slightly different conformations cannot be excluded. Relativelyminor changes in the chemical shift values and in the most of the measurable couplingconstants of the deacetylated compounds indicate that drastic conformational changes havenot occurred upon deprotection, and these compounds, 5FccIdoF (2.9) and 5F(MdoF34(2.14) are depicted in the ‘S3 and 2’5B conformations likely adopted by their acetylatedprecursors.1k 1ki,4B - iS3 B3,0 2S - 25B1k 1k5H0‘C4Figure 2.3. Partial map of pyranoid ring interconversions. An additional eight half-chair(H), four skew-boat (S), and three boat (B) conformations are not shown. Adapted from(Stoddart, 1971).35•OAc•OHQb.Table2.1.and19FNMRchemicalshiftsforthe5-fluoroD-glucosylfluorides(ppm,8)!2.32.4AcOFFAcO.2.7HO.2.12FFFH-iH-2H-3H-4H-6H-6F-iF-5Compound2.75.655.235.435.474.374.12-146.0- fluorides(Hz).tH1,H2H1,FH1,F5H2,H3H2,F1H3,H4H4,F5H6,F5H6-,F5H6,H-F1,F5Compound2.76.052.0——010.323.010.323.———026.021.2Thespectraof2.7and2.12wererecordedinCDC13,thespectraof2.3and2.4inD20.’9FsignalsarereferencedtoCFC13.FOAc2.8Table2.3.and‘9FNMRchemicalshiftsforthe5-fluoroL-idosylfluorides(ppm,H-iH-2H-3H-4H-6H-6’F-iF-5Compound2.85.655.355.185.444.384.30-123.7- for5-fluoroL-idosylfluorides(Hz)!H1,H2H1,FH1,F5H2,H3H2,F1H3,H4H4,F5H6,F5H6-,F5H6,H-F1,F5Compound2.83.552.81.46.311.76.39.417.511.512.—08.324.,&_/2.142.9Thespectraof2.8and2.13wererecordedinCDC13,thespectraof2.9and2.14signalsarereferencedtoCFC13.2.4 Spontaneous Hydrolysis of 5-Fluoro Glycosyl FluoridesThe mechanism of glycoside hydrolysis is thought to involve rate-limitingformation of a short-lived oxocarbenium ion, followed by interception of this species bysolvent to give products of both inverted and retained configuration (Banait & Jencks,1991; Sinnott & Jencks, 1980) (Figure 2.4). The lifetime of a glucosyl oxocarbenium ionin water is estimated to be 10’ to 1(112 s (Banait & Jencks, 1991). Considerable preassociation of the incoming nucleophile therefore likely occurs, and the existence of aglucosyl oxocarbenium ion as a discrete intermediate is questionable, at least with anionicnucleophiles or leaving groups (Zhang, Y. et a?., 1994).H20HO.%.(OH +1OHOHFigure 2.4. Presumedmechanism ofspontaneous glycoside hydrolysis.The presumed mechanism of 5-fluoro glycoside hydrolysis is somewhat morecomplex. As shown in Figure 2.5, initial hydrolysis of the Cl aglycone would produce a5-fluoro glycopyranose species. This will undergo ring opening to produce a reactivefluorohydrin, which spontaneously decomposes with release of fluoride and a 5-ketoHOROH38HROHH20 HO(OHHo_\.Z>FOHHo%LJL2.15Figure 2.5. Presumed mechanism ofspontaneous 5-fluoro glycoside hyrolysis. Althoughthe initial departure of the leaving group is depicted as occuring a Cl, followed by ringopening and loss of the CS fluoride, the reaction could also occur through initial departureof the CS fluoride.derivative, D-xylo-hexos-5-ulose 2.15. Though it is possible with extremely good leavinggroups that ring-opening may become rate limiting, the rate-detemilning step in thisreaction would be expected to be the formation of the oxocarbenium ion upon loss of theinitial aglycone. Although the first leaving group departure is depicted in Figure 2.5 asoccurring at Cl, the reaction could also occur through the initial departure of the C5fluoride.HO.F F +HOH39It is difficult to predict which pathway would be followed. Hydrolysis likelyfollows the pathway involving initial loss of the C5 fluoride in compounds with leavinggroups at Cl of similar or lesser nucleofugacity than fluoride. Loss of the C5 fluoridewould formally produce a tertiary oxocarbenium ion, which would be expected to be morestable than the secondary oxocarbenium ion produced from loss of the Cl fluoride;however, nucleophilic attack at C5 would be more hindered than that at Cl. Since themechanism is on the border between SN1 and SN2, it is not clear which factor will be moreimportant. In addition, differences in the electronic effects of the initially non-departingfluorines at Cl or C5 on the stabilities of the two transition states are difficult to estimate.With similar or identical leaving groups, both processes may occur. Ultimately, regardlessof which fluoride is lost first, two equivalents of fluoride would expected to be releasedupon hydrolysis of a 5-fluoro glycosyl fluoride, along with a hexos-5-ulose species.Spontaneous hydrolyses of the 2-deoxy-2-fluoro- and 5-fluoro glycosyl fluorideswere carried out at 50.0°C in 50 mM phosphate buffer, pH 6.8 containing 1 M NaC1O4 tomaintain a constant ionic strength, according to Konstantinidis & Sinnott, 1991. Thehydrolysis of glycosyl fluorides is pH-independent between pH — 4 to 10 (Banait &Jencks, 1991; Konstantinidis & Sinnott, 1991). Solutions of the glycosyl fluorides wereincubated in screw-top plastic vials, and aliquots were removed at intervals, frozenimmediately, and assayed together after completion of the reaction. In cases for whichhydrolysis was sufficiently rapid to allow monitoring of the reaction for four to five half-lives, rates of fluoride release were determined by plotting the concentration of fluorideversus time and fitting the data to a first order rate equation. If the rate of hydrolysis wasslow, so that only partial hydrolysis was observed over the assay period (- 7 days), therate constant was determined by a linear fit of the initial rate of hydrolysis in which <5% ofthe glycosyl fluoride had been consumed. Hydrolysis rate constants (k) for all compoundsrefer to glycosyl fluoride consumed. Note that the value given is one-half the value initiallydetermined based upon F released for those 5-fluoro glycosides for which initial rates of40fluoride release were measured since two equivalents of F are produced upon thehydrolysis of these compounds. The results are presented in Table 2.5. Plots are shownin Appendix I.Table 2.5. Rates of spontaneous hydrolysis of glycosyl fluorides in 50 mM phosphatebuffer, pH 6.8, 1 M NaC1O4.Compound k at 50.0°C (h1) t1,2 (h)I3GluFa1.4 0.52FfGluF 0.046 ± 0.003 155FfGluFb1.4 x 10 ± 0.05 x 10 5105FcddoF 0.063 ± 0.005 11ocGluFa0.036 192FaG1uFb2.8 x lO ± 0.09 x 10 2 5005FcG1uFb1.1 x 10 ± 0.01 x 10 6505Ff3IdoF 0.048 ± 0.001 14aData taken from (Konstantinidis & Sinnott, 1991).bDetermined from initial rates.The hydrolysis rates of the fluorine-substituted D-glucosyl fluorides wereconsiderably reduced compared to those of the parent o- and f3-D-glucosyl fluorides. Asexpected, incorporation of an additional electronegative fluorine into the sugar ringapparently leads to destabilization of the electron-deficient transition states involved.Fluorine substitution at C2 results in a 130-fold reduction in the rate of hydrolysis of x-Dglucosyl fluoride and a 30-fold reduction in the 3 anomer. In addition, it has beenpreviously shown (Konstantinidis & Sinnott, 1991) that the anomeric configuration of D41glycosyl fluorides has a significant effect on their hydrolysis rates, the rate for a-Dglucosyl fluoride being some 40-fold slower than for the corresponding (3 anomer. The 2-deoxy-2-fluoro-D-glucosyl fluorides exhibit an even greater difference in hydrolysis rates,the hydrolysis rate for the Ct anomer being 160-fold slower than for the corresponding (3compound. However, the cx and (3 anomers of 5-fluoro D-glucosyl fluoride undergohydrolysis at almost the same rate (1.1 x 10 h’ and 1.4 x 10 h1, respectively),suggesting that loss of the axial C5 fluoride (common to both cx and (3 anomers) initiallyoccurs with each of these compounds, followed by rapid expulsion of the remainingfluoride. A similar phenomenon is observed in the L-ido series, the rates for 5-fluoro-(3-L-idosyl fluoride and its a anomer being 0.048 h’ and 0.063 h’, respectively. Thus, it isimpossible to directly compare the effects of fluorine substitution at C2 and C5 in the glucoseries since the rate-determining steps in the hydrolysis of the 2-deoxy-2-fluoro glucosidesand of the 5-fluoro glucosides are likely different processes (necessarily loss of Fl in the 2-deoxy-2-fluoro compounds and presumably loss of F5 in the 5-fluoro analogues).However, a minimum estimate of the effect of the C5 fluorine is possible since loss of theCl fluorine must be slower than loss of the C5 fluorine. This suggests at least a 1000-foldreduction in rate for the (3-D-glucosyl fluoride (cf 30-fold for the 2-deoxy-2-fluorocompound) and at least a 30-fold reduction in rate for the a anomer (cf 130-fold for the 2-deoxy-2-fluoro compound). The apparently greater effect of fluorine substitution at C5than at C2 is possibly to due to a greater proportion of partial positive charge at thetransition state residing at 05 rather than at Cl, or to the greater net electronic effect ofsubstitution of a hydrogen, rather than a relatively electronegative hydroxyl, by fluorine.A mechanism for 5-fluoro glycosyl fluoride hydrolysis involving loss of bothfluorides and formation of D-xylo-hexos-5-ulose (Figure 2.5) is supported by the availableevidence. All of the 2-deoxy-2-fluoro glycosyl fluorides that were followed to totalhydrolysis released one equivalent of fluoride, while all of the 5-fluoro glycosyl fluoridesfollowed to total hydrolysis released two equivalents of fluoride. The products of42hydrolysis of 2Ff3G1uF and 5FaIdoF were analyzed by desorption chemical ionization(DCI) mass spectrometry. Both samples were followed to >95% completion, as assayedby removing aliquots and measuring the fluoride released. The hydrolysates were freezedried, acetylated with acetic anhydride in pyridine, and the products analyzed by desorptionchemical ionization mass spectrometry after work-up. A strong peak at m/z 368 wasobserved in the spectrum of the 2FfGluF hydrolysate, consistent with the mass of 1,3,4,6-tetra-O-acetyl-2-deoxy-2-fluoro-D-glucose (M÷NH4). The high resolution spectrumsupported this assignment (calcd. forC14H9F09+ NH: 368.1356; found: 368.1351).The spectrum of the 5Fc.ddoF hydrolysate showed a weak peak at m/z 364, consistent withthe mass of 2,3,4,6-teira-O-acetyl-D-xylo-hexos-5-ulose (M+NI{4). Again, thisassignment is supported by the high resolution spectrum (calcd. forC14H800+ NH4:364.1237; found: 364.1248). The low intensity of the peak derived from D-xylo-hexos-5-ulose (2.15) is consistent with the known instability of this compound (Kiely & Fletcher,1969), which likely underwent further decomposition prior to acetylation and analysis.Indeed, solutions of hydrolysed 5-fluoro fluorides presumably containing 2.15 becamepale brown upon prolonged incubation, in contrast to solutions of the hydrolysis productsof the 2-deoxy-2-fluoro fluorides, which remained colourless.2.5 Kinetic Studies of the Reaction of 5-Fluoro Glycosyl Fluorides withAgrobacterium faecalis -Glucosidase and Yeast (Saccharomycescerevisiae) ct-GlucosidaseThe general kinetic mechanism for a retaining o- or f3-glycosidase is shown inFigure 2.6. Assuming only the chemical steps are kinetically significant, the mechanismmay be depicted as follows.43k1 k2 k3E + Sk1E•S E-P ‘ E+PHX H20Figure 2.6. Kinetic schemefor retaining glycosidases. E is the enzyme, S is the substrate,E.S is the non-covalent enzyme-substrate complex, HX is the aglycone, E-P is the covalentglycosyl-enzyme complex, and P is the product. k1 and k1 are the rare constants for rapidand revesible formation of the initial Michaelis complex, and k2 and k3 are the first orderrate constants for glycosylarion and deglycosylation, respectively.Assuming a steady state concentration of E•S and E-P is reached during the reactionthenk2[E.SJ=k3[E-PJ (2.1)and dfE•S1 =k1[E][SJ -k1[E•Sj -k2[E.S] (2.2)dt=0The total concentration of enzyme, E0 is the sum of the concentrations of free enzyme andall enzyme-bound species[E0] = [E] + [E.S] + [E-P] (2.3)Therefore, by substituting for [E-P] and rearrangingk1[E0][S] = (k1 +k2)[E•S] +k1[E•S][S] +k12[E•S1[S1 (2.4)k3= (k1 +k2)[E’S] + (k1 +k1Ik3)[E•S][S]Solving for [E.S][E.S] = k1JE0IS1 (2.5)k1 + + kXk2-±- 1 [SI44Since at steady state the rate of production of P. v iS given bydEEl=k3[ -P1 (2.6)dt=k2[E.SJthen Jc2k3_[EQJ [SIv,= jç________ (2.7)_k3 k1..±k2 + [SIk2+k3 k1which follows the standard form of the Michaelis-Menten equationv =kcatf1[S1 (2.8)Km + [SITherefore the kinetic parameters for this mechanism arekcat = 23— (2.9)k2 + k3Km =k1+2 1(3 (2.10)k1 k2+k3kcatfK = .j1k2_. (2.11)k1 + k2It can be shown that k is the rate constant for the rate-determining step of the reaction andwill always be associated with the highest free energy step in the pathway. kcat/Km Willalways be the second-order rate constant for the free enzyme and free substrate proceedingto the transition state of the first irreversible step (Fersht, 1985).a) Substrates with k2 > k3, k3 >> 0. When the aglycone is a highly activatedleaving group (e.g. fluoride or 2,4-dinitrophenolate), it would be expected that if the rate ofglycosylation were increased sufficiently relative to deglycosylation, then deglycosylationwould become rate determining, i.e. k2 > k3. Assuming further a rapid, reversible45association of enzyme and substrate, i.e. k1 >> k2, the kinetic parameters kcat, Km, andk /K then becomecau mkcat = 1 (2.12)Km =k1 1(3 (2.13)k1 k2+k3kcat/K = (2.14)k1This kinetic model applies to substrates for which deglycosylation is rate-limiting, such asf3-D-glucosyl fluoride with Agrobacterium f-g1ucosidase.b) Inactivators with k2 >> k3, k3 0. At the extreme, if k3 is very much smallerthan k2, such that k3 (or keat) approaches zero, then an extremely stable glycosyl-enzymeintermediate will accumulate and the enzyme will be inactivated. The kinetic model thenbecomesk1 k2E+I E•I E-Ik1HXFigure 2.7. Kinetic schemefor inactivation of retaining glycosidases by accumulation of acovalent intermediate. E is the enzyme, I is the inactivaror, E•I is the noncovalent enzymeinactivator complex, HX is the aglycone, E-I is the covalent glycosyl-enzyme intermediate.Rate constants are as in Figure 3.4.This kinetic model predicts a time-dependent inactivation of the enzyme. If [I] is muchgreater than [E0j, [I] can be assumed to be essentially constant during the reaction, andpseudo first-order kinetics with respect to enzyme concentration wifi be observed. InMichaelis-Menten form, the equation for this process is:v1=]j1[fl (2.15)K + [I]46where v1 is the inactivation rate, k1 is the rate constant for inactivation, and K1 is anapparent dissociation constant for all forms of enzyme-bound inactivator. From Figure2.7, it can be seen that k1 = k2, and K = k..1/k. Equation (2.15) can be rewritten asVi = kobs[O] (2.16)wherekObS=.jIjj1_ (2.17)Kj + {IJThe rate of inactivation, v1 is equal todm01= lSbs[EO1 (2.18)dtThus ln[E0j= kobst (2.19)Note that if K>> [11, then Equation (2.17) becomeskobs = (k1fK)[]] (2.20)Note that since k/K is equivalent to (k1k2)/1c,k1/K is analogous to kcat/Km in terms ofthe individual rate and equilibrium constants for each step. This kinetic model applies toinactivators for which keat is very small relative to k1, such as 2-deoxy-2-fluoro-3-D-glucosyl fluoride with Agrobacterium 3-g1ucosidase.c) Inactivators or slow substrates with k2 > Ic3, Ic3 > 0. The scheme represented inFigure 2.7 holds if k2 (or k) is very much greater than k3 (or kcat). In such a situation,essentially all of the enzyme will exist as the accumulated glycosyl-enzyme intermediate,and complete inactivation will be observed. K and k1 may be then be directly determined.However, the extent to which the intermediate accumulates depends primarily on the47relative values of k and kcat. If the magnitude of keat approaches (but is still less than) k1,then the intermediate will slowly turn over and a steady-state rate will be established (cfFigure 2.6). An apparent K’ under steady-state conditions Legler, 1990; Wentworth &Wolfenden, 1974) is given byK1’= K1_ (2.21)1 + ki/kcatThis equation takes the form of the Michaelis-Menten expression for Km (Equation 2.13)with K =k1/k,k1 = k2 and k3 = kcat. In such a situation, K1’ represents a minimumvalue for the actual dissociation constant K. This kinetic model applies to inactivators (orslow substrates) for which keat is significant relative to k, such as D-galactal with E. coli[3-galactosidase (Wentworth & Wolfenden, 1974), which exhibits tight apparent bindingand subsequent slow turnover of the 2-deoxy-hexosyl enzyme formed (see Figure 1.9).2.5.1 Inactivation of Agrobacterium faecalis f3-glucosidase and yeast aglucosidase by the 5-fluoro D-glucosides 5F3GluF (2.3) and5FaGIuF (2.4), respectivelya) Inactivation of Agrobacterium /3-glucosidase. Incubation of 5-fluoro--D-glucosyl fluoride (5FfGluF) 2.3 with Agrobacterium faecalis 3-glucosidase resulted inrapid time-dependent inactivation according to essentially pseudo-first order kinetics,though inactivation, particularly at lower inactivator concentrations, did not proceed tocompletion (Figure 2.8A). This is consistent with a kinetic model (Street et al., 1992) inwhich at high concentrations of inactivator, the rate of formation of the intermediate (=k[5F(3GluFJ) is significantly greater than that of its breakdown (= k[5FGlu-enzyme]),but not at low inactivator concentrations. A re-plot of the rate constants from the initialexponential phase showed no saturation at concentrations up to 3.7 p.M and at higherconcentrations, inactivation became too rapid to accurately measure. Nonetheless, a second480.032A0.0240.0160.008B2.421.[5FG1uF] (pM)Figure 2.8. A) Inactivation of Agrobacterium J3-glucosidase with 2.3. Enzyme wasincubated with the following concentrations of2.3 and aliquots assayed with /3PNPFuc athe indicated times. 3.68 jiM (V), 1.84 pM (C)), 0.921 pM (X), 0.736 pM (•), 0.368pM (0). B), Replot of rate constants from above.0 4 8 12 16 20 24 28Time (mm)1 2 3 449Figure 2.9. Reactivation of isolated 5-fluoroglucosidase at 37°C.glucopyranosyl-Agrobacterium 13--2 0 2 4 6 8 10 12 14 16[5FG1uF] (iM)Figure 2.10. ReversibleK1’for2.3 with Agrobacterium J3-glucosidase determined understeady state conditions.100IE0 20 40 60Time (mm)80400300200100050order rate constant k1fK = 660 ± 43 min’mlvf’ was obtained from the slope of this plot(Figure 2.8B). The expected level of protection was afforded by 4.7 tM castanospermine(K = 3 p.M, (Namchuk, 1993)), the pseudo-first order rate constant at 1.84 .tM 5FGluFbeing reduced from 1.07 to 0.18 mm4. When freed of excess inhibitor by ultrafiltration,followed by incubation in buffer at 37°C, a first-order recovery of enzyme activity wasobserved (k3 (keat) = 0.082 ± 0.006 min1, corresponding to t1 = 8.5 mm, Figure 2.9),indicating a catalytically competent intermediate that is capable of normal turnover, but atgreatly reduced rates. Consistent with the proposed kinetic model, an apparent K1’ understeady-state conditions, representing a minimum value for the true dissociation constant K1,was determined to be 0.30 ± 0.08 p.M (Figure 2.10).As a further demonstration that inactivation had occurred by formation of arelatively stable 5-fluoro glucosyl-enzyme intermediate, an electrospray mass spectrum ofthe 5FG1u-inactivated f3-glucosidase was obtained. This showed that the mass of theprotein increased from 51192 ± 6 Da to 51 363 ± 6 Da upon inactivation, an increase of171 ± 12 Da, which is consistent, within experimental error, with the covalent attachmentof one 5-fluoro glucosyl moiety (181 Da).b) Inactivation of yeast (Saccharomyces cerevisiae) a-glucosidase. Incubation ofyeast ct-glucosidase with 5-fluoro-a-D-glucosyl fluoride (5FaG1uF) 2.4 also resulted ininhibition of the enzyme by accumulation of a 5-fluoro glycosyl-enzyme intermediate.Inhibition was veiy rapid, and no time dependence was observed, even when assaying atthe shortest possible time intervals (— 30 s), indicating that the inactivation step (k1) wasvery rapid. However, complete inhibition was not observed since turnover of theintermediate (keat) was also rapid, allowing the establishment of a small, but significantsteady-state rate. This compound is perhaps best described as a slow substrate rather thanan inactivator. Incubation of various concentrations of 5FaG1uF with enzyme resulted inslow rates of fluoride release until the compound was consumed, ultimately releasing two51Figure 2.11. Hydrolysis of 2.4 (99 uM (V), 198by yeast a-glucosidase at 37°C.uM (zi), and 990 jiM (0)) catalyzed8006004002000II)I I I I I I I I F 1 ‘I’ll’-2 0 2 4 6 8 10 12 14 16 18 205FxG1uF (IIM)Figure 2.12.state conditions.Reversible K. ‘for 2.4 with yeast a-glucosidase determined under steady—180016001400120010008006004002000I I I I I I I II I I I I I I0 2000 ‘4000Time (s)—1”•I I I i I i I i I i I i I i I i I I I I I52(2.0 ± 0.3) equivalents of fluoride (Figure 2.11). These rates were constant over a range-1of concentrations, enabling determination of a vaiue of 6.6 ± 0.4 nn . Times coursesof fluoride release were linear over virtually the entire course of the hydrolysis reaction,even at gtM concentrations, indicating an extremely low Km value. This veiy low Km value(resulting from significant accumulation of an intermediate) precluded its determination dueto the insensitivity of the F--selective electrode at the low concentrations required.However, an apparent K under steady-state conditions was determined to be 1.40 ± 0.09j.iM (Figure 2.12). Among inhibitors of yeast a-glucosidase, 5FaGluF thus shows thetightest apparent binding yet described.2.5.2 Inactivation of Agrobacterium faecalis f-glucosidase and yeast cxglucosidase by the 5-fluoro L-idosides 5FctIdoF (2.9) and 5FIdoF(2.14), respectivelya) Inactivation of Agrobacterium 13-glucosidase. Perhaps surprisingly, the C-5epimers of the 5-fluoro gluco compounds above, in which the sugar hydroxymethylene isof the L-idosyl configuration, are also effective inactivators of the correspondingglycosidases. In the case ofAgrobacterium 3-gIucosidase, a non-fluorinated p-nitrophenylcx-L-idoside (synthesized and kinetically evaluated by Dr. Evelyn Rodriguez in thislaboratory) is a correspondingly reasonable substrate of this enzyme, consistent withinactivation by the 5-fluoro idosides occurring by recruitment of the normal catalyticmechanism. The kinetic parameters for hydrolysis of p-nitrophenyl a-L-idopyranosidebyAgrobacterium 3-glucosidase are keat = 47 ± 2 mm4 and Km 2.55 ± 0.2 mM. Thevalue is only 200-fold reduced from the analogous 3-D-glucoside (k = 10000 mind,Km = 0.078 mM), and the Km is increased 30-fold. Remarkably, considering thesignificant structural and conformational differences between L-idosides and D-glucosides,this reduction in keat is less than that seen with either D-allosides (1300-fold) or Dmannosides (1000-fold) in a series of 2,4-dinitrophenyl glycosides (Namchuk, 1993). The53corresponding D-galactoside in this series is, however, a good substrate with a 1.3-foldhigher kcat than the D-glucoside.Incubation of Agrobacterium -glucosidase with 5FctIdoF 2.9 resulted in time-dependent inactivation of the enzyme. As with the 5-fluoro gluco analogue 2.3 above, lossof enzyme activity initially followed first order kinetics, yet incomplete inactivation and asteady-state phase were observed (Figure 2.13A). The phenomenon was more extreme inthis case, the steady-state rate at the lowest concentrations of inhibitor corresponding toonly — 25% inactivation. A re-plot of the rate constants derived from the initial exponentialphase versus inhibitor concentration showed no saturation at concentrations up to 1 mM.Unfortunately, at higher concentrations, inactivation became too rapid to accuratelymeasure at 370C, and it was not possible to obtain values of k1 and K•. Nonetheless, asecond order rate constantk1/K = 3.0 ± 0.09 min1M- was obtained from the slope ofthe plot of kobs versus inhibitor concentration (Figure 2.1 3B). The inactivation wastherefore re-examined at a lower temperature (5°C) to slow the inactivation and allowstudies to be performed at higher inhibitor concentrations. Under these conditions,saturation was indeed observed, allowing the determination of kinetic parameters of k1 at5°C = 8.1 ± 0.6 min’, K1 = 3.0 ± 0.5 mM. The expected level of protection was affordedby 4.5 mM isopropylthio 3-D-glucopyranoside (K1 = 4 mM, (Withers et a!., 1987)) thepseudo-first order rate constant at 0.36 mM 5FaIdoF being reduced from 0.94 to 0.28mm-’. Again, the intermediate formed was capable of turnover. When freed of excessinhibitor and incubated at 4°C or 37°C, a rapid first-order recovery of enzyme activity wasobserved, with k3 (kcat) values of 1.1 x i02 ± 0.2 x i02 min1 and 5.8 x i02 ± 1.2 x i02min1, corresponding to t112 = 65 mm and 12 mm, respectively (Figure 2.14). At 37°C,both the glycosylation (reflected byk1IK) and the deglycosylation (reflected by kcat) stepsof the enzyme were slower for the 5-fluoro idosyl compound 2.9 than for the glucoanalogue 2.3, by 200- and 1:5-fold, respectively.54A0. 200 400 600 800 1000 1200[5FocIdoF] (riM)Figure 2.13. A) Inactivation of Agrobacterium f3-glucosidase with 2.9. Enzyme wasincubated with thefollowing concentrations of2.9 and assayed with J3PNPFuc at the timesshown: • = 0.014 mM, 13 = 0.029 mM, N = 0.058 mM, A = 0.072 mM, C) = 0.151 mM,A = 0.359 mM, 17 = 1.08 mM. B) Replot of rate constants from above.0 2000 4000Time (s)550.048i“—‘ 0.032.—-S0.0160 2000 4000 6000 8000 10000 12000Time (s)Figure 2.14. Reactivation of isolated 5-fluoro idopyranosyl-Agrobacterium J3-glucosidaseat 4°C (0) and 37°C (•).As with the gluco compound, an electrospray mass spectrum of the 5-fluoro idosylinactivated f-glucosidase showed that the mass of the protein increased from 51 205 ± 6Da to 51 384 ± 5 Da upon inactivation, an increase of 179 ± 11 Da, which is consistent,within experimental error, with the covalent attachment of one 5-fluoro idosyl moiety (181Da).b) Test for inhibitory contaminant in SFcxJdoF 2.9. A 1.3-fold molar excess of5FoIdoF (95 .tM) was incubated with Agrobacterium -glucosidase (70 p5M). Aliquotswere removed at appropriate intervals, diluted, and assayed for activity. At thisconcentration of 5FczIdoF, — 70% inactivation of the enzyme after 17 minutes waspreviously observed (see Figure 2.13). In the present experiment, — 50% inactivation wasobserved after 14.5 minutes. These values are very close, indicating that 5FczIdoF is thetrue inhibitor. Indeed, exactly the same level of inactivation is not expected since with nearstoichiometric amounts of enzyme, the inactivation is a second order process. Further, the56rapid turnover of the intermediate, releasing active enzyme, also results in consumption ofthe inhibitor, and would reduce the level of inactivation. However, if this — 50% reductionin activity was due to a possible contaminant, it would have to have constituted almost halfof the preparation, which is clearly not the case, as indicated by NMR or mc.c) Inactivation of yeast a-glucosidase. 5-Fluoro -L-idosyl fluoride (5FfIdoF)2.14 inactivated yeast a-glucosidase according to essentially pseudo-first order kinetics,the inactivation of this enzyme by 2.14 being the most complete of all the 5-fluoroglycosyl fluorides tested (Figure 2.15A). A replot of the rate constants obtained from theexponential portion revealed reversible, saturable binding (K1 = 1.8 ± 0.4 mM) andallowed determination of an inactivation rate constant, k1 = 1.4 ± 0.1 min1, correspondingto a half-life at saturation of = 0.49 mm (Figure 2.15B). The competitive inhibitor 1-deoxynojirimycin (K = 13 tM, (Legler, 1990)) afforded protection from inactivation asexpected; a concentration of 58 I.LM reducing kobs for 1.25 mM 5F1doF from 0.51 mm4to 0.29 min1. When freed of excess inhibitor by ultrafiltration, followed by incubation inbuffer at 40C or 370c, a first-order recovery of enzyme activity was observed, with k3(kcat) values of 3.2 x 10 ± 1.0 x 10 min’ and 2.1 x 10 ± 0.2 x 10 min1,corresponding to t1 = 2200 mm and 330 mm, respectively, indicating a catalyticallycompetent intermediate that is capable of normal turnover, but at greatly reduced rates(Figure 2.16). The reactivation of the 5-fluoro-cz-L-idosyl enzyme was the slowest of anyof the 5-fluoro glycosyl-enzymes observed, consistent with the almost completeinactivation seen. Indeed, the of 2.14 was reduced 3000-fold from the gluco analogue2.4, sufficient to allow isolation and monitoring of the reactivation of the glycosyl-enzymeintermediate. The glycosylation step was also slow for the 5-fluoro L-idoside, the k valueof the idoside (1.4 mint) being less than even the kcat value for the 5-fluoro D-glucoside(6.6 min1), which reflected the deglycosylation step. However, in the absence of pre57AI:0 2 4 6 8Time (mm)ço.8.‘ 2 4 6 8 10 12[5FIdoF] (mM)Figure 2.15. A) Inactivation ofyeast a-glucosidase with 2.14. Enzyme was incubatedwith the following concentrations of 2.14 and aliquots assayed with czPNPGIu at thetimes shown: O = 0.313 mM; 13=0.625 mM; = 1.25 mM; LI = 3.13 mM; V = 11.3mM. B) Replot of rate constants from A).580.040.02• 0.010Time (mm)Figure 2.16 A) Reactivation of isolated 5-fluoro idopyranosyl-yeast a-glucosidase ci4°C (0) and37°C (0).steady-state data on the 5-fluoro D-glucoside, the degree of the reduction in theglycosylation step cannot be inferred.d) Test for inhibitory contaminant in SFj3IdoF 2.14. In this experiment, arelatively large amount of enzyme is used to remove possible inhibitory contaminantswhich might be present in small amounts in the inhibitor preparation. This “purified”inhibitor sample is then re-tested as an inactivator. An 11-fold molar excess of 5FIdoF(0.31 mM) was incubated with yeast a-glucosidase (0.028 mM) for 20 minutes. At thisconcentration of 5FIdoF, inactivation of the enzyme was previously determined to be>90% complete at this time, corresponding to a kobS = 0.20 min1 (see Figure 2.15). Theinactivated enzyme was removed by ulirafiltration, and the filtrate was re-tested forinactivation of c-glucosidase, by adding an appropriate dilution of fresh enzyme, removingaliquots at intervals and assaying for residual activity with ctPNPG1u. Time-dependentinactivation was observed with kobS = 0.21 min1, a value that compares well with that0 2000 400059previously determined. Thus any possible contaminant in the original preparation thatmight have been responsible for the inactivation observed would have to have constituted atleast 10% of the preparation, an amount detectable by NMR and TLC, which was notobserved.e) Anomeric specificity of the L-idosyl fluorides. Despite significant differencesbetween the ground state conformations of these L-idosides and the corresponding Dglucosides, a high level of anomeric specificity for the appropriate a- or f3-D-glucosidasewas exhibited by these L-idosyl fluorides. Less than 4% inactivation of a-glucosidase inthe presence of 5-fluoro a-L-idosyl fluoride 2.9 (1.58 mM) and 2% inactivation of f3-glucosidase in the presence of the f3-L-idosyl fluoride 2.14 (1.25 mM) was observed afterone hour at 370C, although 39% and 7% inactivation of these enzymes, respectively, wasobserved after 24 hours incubation with these compounds. It is likely that these muchslower inactivation processes are due to very small amounts of contaminating 5F3IdoF and5FaIdoF, respectively, in preparations of the a-L- and -L-idosyl fluorides.2.5.3 Interpretation of Kinetic Results with 5-Fluoro Glycosyl FluoridesThe kinetic parameters for reaction of the 5-fluoro glycosyl fluorides with each ofAgrobacterium 3-glucosidase and yeast a-glucosidase are summarized in Tables 2.6 and2.7, and compared with those of the analogous 2-deoxy-2-fluoro compounds. The kineticbehavior of the 5-fluoro glycosides, which cause rapid inactivation and exhibit turnover at asignificant rate, is analogous to the slow hydration and tight apparent binding of variousglycals by some glycosidases (Legler, 1990; Wentworth & Wolfenden, 1974). Like someglycals, which display a slow onset of inhibition, the formation of most of the 5-fluoroglycosyl-enzyme intermediates is relatively slow, sufficient in some cases to permitdetermination of a k1 value. However, the formation of the 5-fluoro-D-glucosyl-ct-60Table 2.6. Kinetic parameters for reaction of f3GluF, 2FI3G1uF, 2dI3PNPG1u, 5FJ3G1uF,and SFcddoF with Agrobacterium J3-glucosidase.k3 or kcat Km or K’ k1 K kca/Km or(1)(mM)(1)(mM) k1/K(min’mlvf)CompoundGluFa 11 000 3.7 3 0002FI3G1uFb1.2 x i(1g 8.1 xC 5.9 0.40 14.82d[3PNPG1ud 1.5 1.5 x 1(121005FI3G1uF 8.2 x 1(12g3.0 x 10e --- - 6602 -4 —±0.6x1C1 ±0.lxlO5FcddoF 5.8 x o2 g 8.i’ 3.0 3.0±1.2x1C1 — —aData taken from (Day & Withers, 1986).bData taken from (Street et al., 1992; Withers et al., 1988).CK1’ calculated according to K1’ =K1! (1 +k1fk).dData taken from (Namchuk, 1993).eExperimentally determined K1‘value.“At 5°C.g Reactivation of isolated glycosyl-enzyme.glucosidase from 5FaG1uF is very rapid, and thus no k1 was determined. Furthermore,the turnover of the 5-fluoro glycosyl-enzyme intermediates is generally slower than theturnover of the 2-deoxy-hexosyl-enzyme intennediates formed with glycals (e.g. kcat =0.28 min1 for D-galactal with (3-galactosidase (Wentworth & Wolfenden, 1974)). Lowerapparent K1’ values are therefore observed. The behavior of the 5-fluoro L-idosides,which inactivate more slowly, and turn over at an even lower rate, is similar to that of 2-61deoxy-2-fluoro compounds with 3-glucosidases, permitting isolation of the corresponding5-fluoro L-idosyl-enzyme intermediates with both Agrobacterium 3-glucosidase and yeasta-glucosidase.Comparison of the k3 (kcat) values for the 2FG1u- and 5FG1u-labelled (3-glucosidase reveals that the turnover of the 2-deoxy-2-fluoro-glycosyl-enzyme intermediateis slowed some 109-fold (compared to the parent substrate) presumably due to acombination of electronic effects and the loss of critical binding interactions at the 2-position, while turnover of the 5-fluoro glucosyl-enzyme intermediate is only (!) slowed5 . ..some 10 -fold. Since the 2-position hydroxyl is intact in the latter compound, littletransition state binding energy would be expected to have been lost in that case. The keatvalue for the 2-deoxy hexoside (2dI3PNPG1u) is 1.5 min1, which if deglycosylation israte-limiting, represents at least a 104-fold reduction in the turnover of the 2-deoxy-hexosyl-enzyme intermediate. This reduction could be even greater if the intermediate isalso destabilized. It is tempting to speculate from these data that the contributions ofelectronic and binding factors to the enormous decrease in the rate of 2-deoxy-2-fluoroglycosyl-enzyme hydrolysis are approximately equal. However, several factors must beconsidered. Firstly, the rate for the 2-deoxy compound represents a minimum estimate ofthe contribution of the 2-position binding interactions, since the hydrolysis rate of thiscompound is considerably increased by the electronic effect of a relatively electron-richcentre adjacent to the developing oxocarbenium ion-like transition state. Secondly, thedistributions of partial positive charge in the oxocarbenium ion-like enzymic transitionstates may be different in the 2-deoxy-fluoro and 5-fluoro species, and the electronic effectson transition state stabilities of the fluorines at C5 and at C2 may be different. Indeed, theeffect of incorporation of a C5 fluorine on the spontaneous hydrolysis rate of (3-glucosylfluoride is some 30-fold greater than that of a C2 fluorine (Table 2.5). Thirdly, theturnover rates will also be influenced by destabilization of the 2-deoxy-2-fluoro glucosylor 5-fluoro glucosyl-enzyme intermediates by loss of specific binding interactions and/or62Table 2.7. Kinetic parameters of aGluF, 2FaG1uF, 5FaG1uF, and SFj3IdoF with yeasta-glucosidase.k3 or kcat Km or K - k kcat/Km or(min’) (mM) (min’) (mM) k/K1(min’mM’)CompoundaGluFa 1500 0.93 --- --- 16002FaG1uFb 96 4.8 --- --- 205FctGluF 6.6 1.4 x 10± 0.4±0.09x 105FIdoF 2.1 xe2.7 xd 1.4 1.8 0.77±0.1 ±0.4 ±0.22±0.2x 10aData taken from (Konstantinidis & Sinnott, 1991).bData taken from (Braun, 1995).CExperimentally determined K’ value.dK calculated according to K1 = K•/ (1 + kjfk).Reactivation of isolated glycosyl-enzyme.steric factors, which may be different in the two cases. However, these data indicate thatthe electronic effects of fluorine substitution at C2, and to an even greater extent at C5,results in substantial destabilization of the enzymic transition states.With yeast c-g1ucosidase, substitution of the 2-hydroxyl for fluorine has notresulted in the great destabilization of the deglycosylation transition state observed with the-glucosidase, and 2-deoxy-2-fluoro a-D-glucosyl fluoride is merely a substrate for theformer enzyme. There are several possible explanations for this. One possibility is thatthis enzyme is able to recoup binding energy at the 2-position via fluorine acting as ahydrogen bond acceptor. A second, perhaps more likely scenario, is that the more rapid63turnover is due to the inherently greater lability of a 2-deoxy-2-fluoro j3-D-glucosyl-enzymeintermediate. The smaller electronic effect of fluorine substitution on the hydrolysis rate ofsuch a species (cf the spontaneous hydrolysis rates of 2FGluF and 2FaG1uF relative tothe parent compounds, Table 2.5) suggests a lesser degree of oxocarbenium ion characterat this transition state, consistent with the smaller a-secondary deuterium kinetic isotopeeffects seen with f’-methyl glucoside and 3-glucosyl fluoride relative to the a anomers(Bennet & Sinnott, 1986; Zhang, Y. et al., 1994). In the enzymic reaction, the 5-fluorosubstituent reduces the kcat value by a further 15-fold relative to the 2-deoxy-2-fluorospecies, consistent with a greater electronic effect of the C5 fluorine (Section 2.4).In conclusion, 5-fluoro 3- and a-glycosides are potent mechanism-based inhibitorsof - and a-glycosidases, respectively. Both the 5-fluoro-3- and 5-fluoro-a-D-glucosylfluorides form catalytically competent intermediates with the appropriate glucosidases thatare capable of turnover, at rates reduced i- and 103-fold, respectively, from the f3- (Day& Withers, 1986) and a-D-glucosyl fluoride (Konstantinidis & Sinnott, 1991) parentsubstrates. The corresponding L-idosyl fluorides are hydrolyzed even more slowly, thekeat values of the 5-fluoro-a- and 5-fluoro-[3-L-idosyl fluorides with the appropriateenzymes being reduced a further 1.5- and 3000-fold, respectively. Evidently, and quitereasonably, in light of the 200-fold reduction in the rate of p-nitrophenyl L-idoside withAgrobacterium -glucosidase, both 5-fluoro idosyl-enzyme intermediates turn over moreslowly than the analogous 5-fluoro glucosyl species. These 5-fluoro compounds should bevaluable tools to probe glycosidase mechanisms and to identify active site residues,particularly those of enzymes which are intolerant of substitution of the C2 substituent (e.g.N-acetylglucosaminidases) or which are resistant to inactivation by 2-deoxy-2-fluoroglycosides (e.g. a-glycosidases). Further, the fact that successful inactivation is seenprovides further substantial evidence (Lehmann & Reinshagen, 1970) against an enzymicmechanism involving initial 05-Cl endocyclic cleavage (See Section 1.3, and (Fleet, 1985;Franck, 1992; Post & Karplus, 1986)). Such a process, rather than resulting in enzyme64inactivation by accumulation of a cyclic 5-fluoro glycosyl-enzyme intermediate, wouldliberate a reactive fluorohydrin at C5 as the proposed acycic intermediate is formed. Thiswould presumably eliminate fluoride rapidly, releasing a 5-ketoglucose derivative.2.6 Identification of the Catalytic Nucleophiles of both 13- and cxGlycosidases using 5-Fluoro Glycosyl Fluorides and MassSpectrometry2.6.1 Confirmation of the mode of action of 5-fluoro glycosyl fluorides bylabelling of the catalytic nucleophile of Agrobacterium faecalis 1-glucosi dasePreviously, the catalytic nucleophile of Agrobacterium f3-glucosidase has beenidentified as Glu-358 by inactivation with[3H]-labelled 2,4-dinitrophenyl 2-deoxy-2-fluoro-3-D-glucopyranoside, followed by peptic hydrolysis of the 2FG1u-labelled enzyme,isolation of the radiolabelled peptides by liquid chromatography, and sequencing by Edmandegradation (Withers et al., 1990). Radioactivity was detected in two peptides which,upon isolation and sequencing, proved to have the sequences Y1TENGAC and ITENGAC.The nucleophile, Glu-358, is absolutely conserved in all glycosidases of this family, andmutation of this residue to alanine resulted in a mutant with severely compromised activity(106 of wild-type activity) (Withers et a!., 1992).Inactivation of the enzyme by the 5-fluoro f3-D-glucosyl- and 5-fluoro ct-L-idosylfluorides would be expected to result in derivatization of the same residue as the 2-deoxy-2-fluoro glycoside. Such a fmding would indicate that the 5-fluoro glycosides are notreacting through a different mode with the enzyme, perhaps by displacement of the C5fluorine by another enzymic nucleophile. To test this, (3-glucosidase was inactivated withnon-radiolabelled 5FI3G1uF or 5FczIdoF, proteolyzed with pepsin, and the resultingmixture of peptides subjected to liquid chromatography as for the 2FGlu-labelled enzyme.Identification of the site of attachment of the 5-fluorosugars, thus of thenucleophilic amino acid residue, was achieved without resort to radiolabelling by using anew mass spectrometric technique to identify the labelled peptide in a peptic hydrolysate of65the labelled enzyme. Peptic hydrolysis of 5FG1u- and 5FIdo-labelled -glucosidaseresulted in a mixture of peptides which was separated by reverse phase-HPLC using theESMS as detector. When the spectrometer was scanned in the normal LC/MS mode, thetotal ion chromatogram (TIC) of the 5FIdo-labelled enzyme digest displayed a large numberof peaks, which arise from every peptide in the mixture (Figure 2.17A). The peptidebearing the 5-fluoroidosyl label was then identified in a second run by using the tandemmass spectrometer in the neutral loss mode. In this technique the ions are subjected tolimited fragmentation by collisions with an inert gas (Ar) in a collision cell. Since the esterlinkage between the sugar inhibitor and the peptide is one of the more labile linkagespresent, relatively facile homolytic cleavage of this bond occurs resulting in the loss of aneutral sugar residue of known mass and leaving the peptide moiety with its originalcharge. Quadrupoles Qi and Q3 can then be scanned in a linked manner in which the twoanalyzers are offset by the mass of the anticipated “lost” neutral species such that only ionsdiffering in m/z by the mass of the lost sugar moiety (181 Da) can pass through bothquadrupoles and be detected (Busch & Cooks, 1983). When the spectrometer was scannedin the neutral loss tandem MS/MS mode searching for the mass loss m/z 181,corresponding to the loss of the 5FIdo label from the labelled active-site peptide in thesingly charged state, a greatly simplified chromatogram was obtained (Figure 2. 17B).Several major peptides are seen, plus a number of minor species. However, when asample of non-labelled enzyme was subjected to peptic digestion and subsequentESMS/MS neutral loss mode analysis, the chromatogram shown in Figure 2. 17C wasobtained. Most of the features of Figure 2. 17B are seen here also, with the exception ofpeptides 1 and 2, indicating that these peptides are the species of interest. The other signalsarise from non-labelled peptides which undergo an equivalent fragmentation, the mostlikely being elimination of a tyrosine residue (181 Da) from several different peptides.Since the stoichiometry of inactivation is 1:1 enzyme/inhibitor, peptides 1 and 2, whichwere only detected in the digest of the inhibited enzyme, are presumably overlapping66A18 20 2212 14C18 20 2212 14 16 18 20 22100‘ 75050.z 25Caa)100755025a)100755025Caa)1 OC505005Ô0 750 1000 1250 1500 1750m/zFigure 2.17. ESMS experiments on Agrobacterium f3-glucosid.ase proteolytic digests: (A)labelled with 5FIdo, TIC in normal MS mode, (B) labelled with 5FIdo, TIC in neutral lossmode, and (C) unlabelled, TIC in neutral loss mode, (D) mass spectra of peptides 1 and 2in Fig.2.17B.67D 888I.500E750 1000 1250 1500 1750mlz1051I _I I—peptides containing the covalently modified active-site residue. These singly-chargedpeptides were measured at m/z 1051 ± 1 and 888 ± 1 (Fig. 2.17D), thus since the 5FG1u or5FIdo moieties have a mass of 181 Da, the molecular weights of the unlabelled active-sitepeptides are 870 ± 1 and 707 ± 1. The neutral loss chromatogram of the 5FG1u-labelleddigest was essentially identical, with peaks of identical masses eluted at nearly the sameretention times (not shown).Consistent with these findings, the calculated masses of the Y1TENGAC andITENGAC species previously identified by labelling with the 2-deoxy-2-fluoro derivativeare indeed 870.37 and 707.30 Da, respectively. A similar neutral loss experiment, on2FG1u-labelled enzyme, again produced an essentially identical chromatogram, but withmasses of 1035 ± 1 and 872 ± 1, consistent with the attachment of 2FGIu (165 Da)moieties to the YITENGAC and ITENGAC peptides (Tull et at., 1995). These results,taken together, confirm that the same residue in Agrobacterium f-glucosidase is labelled bythe 2-deoxy-2-fluoro, 5-fluoro glucosyl, and 5-fluoro idosyl derivatives, and providesfurther evidence that inactivation of this enzyme by these inhibitors occurs by theaccumulation of stabilized 2-deoxy-2-fluoro- or 5-fluoro-glycosyl-enzyme adducts whichare analogous to the glycosyl-enzyme intermediate in the normal catalytic mechanism. Thisrapid, sensitive and non-radioisotopic approach utilizing mass spectrometry to identifyinglabelled residues in enzymes offers broad applicability to other systems (see followingSection, and Chapter 4).2.6.2 The catalytic nucleophile of yeast (Saccharomyces cerevisiae) aglucosi dasea) Background and signcance. Glycosyl hydrolases of Family 13 (the aamylase family (Henrissat, 1991; Henrissat & Bairoch, 1993)), which include cz-amylases,and some ct-glucosidases and oligo-1 ,6-glucosidases, are important enzymes involved inthe digestion of starch and other a-linked oligosaccharides in bacteria, plants and animals.68These glycosyl hydrolases cleave a-glycosidic bonds with net retention of configuration,and thus are retaining ct-glucosidases. The active sites of these enzymes are known toinclude a trio of conserved carboxylic acids (Kadziola et at., 1994; Klein et al., 1992;McCarter & Withers, 1994; Qian et a!., 1993; Qian et at., 1994). Presumably, one residuefunctions as the catalytic nucleophile, one as a general acid/base and the third possibly toafford additional stabilization of developing positive charge or to modulate theP1<a’S of theother catalytic residues.Considerable progress has been made towards elucidating the roles of thesecarboxylates in Family 13 enzymes, involving kinetic analyses of site-specific mutants(Svensson & Sogaard, 1993, and references therein) and X-ray crystallography (Kadziolaet a!., 1994; Klein eta!., 1992; Qian eta!., 1993; Qian eta!., 1994). Labelling studies havealso been useful. For example, an aspartate which functions as a catalytic nucleophile in asucrose: cL-glucan glycosyltransferase, has been identified in Streptococcus sobrinus ciglucosyltransferase by denaturation trapping using radiolabelled sucrose (Mooser et a!.,1991). Similarly, in a second group of cx-glycosyl hydrolases (Family 31), Asp-505 andAsp-1394 have been identified as active site residues in each of the two homologous activesites in the sucrase-isomaltase complex by affmity labelling with conduritol B epoxide(CBE) and have been suggested to be the catalytic nucleophiles (Quaroni & Semenza,1976). However, lysosomal ct-glucosidase and sucrase-isomaltase (Family 31) and the oamylases (Family 13) belong to different glycosidase families (Henrissat, 1991; Henrissat& Bairoch, 1993) with only weak active site similarities. Another approach was thereforerequired to unequivocally demonstrate the covalent involvement of a specific active sitecarboxylate as the nucleophile in the normal catalytic mechanism of Family 13 glycosidasesand related enzymes of the a-amylase family catalyzing the hydrolysis of a-glycosidiclinkages.69b) Inactivation of yeast a-glucosidase. Incubation of the enzyme with either5FcGluF 2.4 or 5F3IdoF 2.14 resulted in inhibition of the enzyme by accumulation of a5-fluoro glycosyl-enzyme intermediate, as described previously (Section 2.5). For ES/MSexperiments, the enzyme was incubated with 5F3IdoF (8.7 mM) for one hour or 5FaG1uF(2.7 mM) for 10 to 15 s. Because of the significant turnover of the 5-fluoro glucosylenzyme, the 5FctGluF mixture was immediately added to pH 2 buffer. Both mixtures wereproteolyzed with pepsin and analyzed immediately after digestion (— 30 mm). Relativelyrapid digestion and prompt analysis of the 5-fluoro glycosyl-labelled peptides is necessarysince the C5 fluorine is labile at pH 2.c) Identification of the Labelled Active Site Peptides by Mass Spectrometry. Pepticdigestion of the 5FGlu- or 5FIdo-labelled enzymes resulted in a mixture of peptides whichwas separated by HPLC using the mass spectrometer as a detector. When scanned inLC/MS mode, the total ion chromatogram (TIC) displays a complex mixture of peaksarising from every peptide in the digest (Figure 2.18A). The labelled peptides wereidentified in a second run using tandem mass spectrometry in neutral loss mode, byselective fragmentation of the inhibitor-peptide bond. As before, the two quadrupoles alescanned in a linked mode so that only those ions differing by the mass of the label could bedetected. For a singly-charged peptide, this m/z difference is the mass of the label (181Da); for a doubly-charged peptide, the m/z difference is one-half of the mass of the label(90.5 Da), and so on.Scanning in neutral loss mode for the mass loss m/z 181 from a singly-chargedpeptide revealed two peaks (Figure 2.1 8B) in the 5FIdo-labelled digest that were absent inan unlabelled, control digest (Figure 2.18C). The strong peak at 18.9 mm was due to asingly-charged labelled peptide measured at 925 ± 1 (Figure 2.1 8D), corresponding to asingly-charged unlabelled peptide of mass 745 ± 1 (i.e., 925 - 181 + 1). The less intensepeak at 20.5 mm was due to a singly-charged labelled peptide measured at 1073 ± 170(Figure 2.18E), corresponding to a singly-charged unlabelled peptide of mass 893 ± 1.Inspection of the protein sequence without any consideration of protease specificityrevealed only six possible candidate singly-charged peptides of mass 745 ± 1, and onlytwelve of mass 893 ± 1. However, only two individual glutamic or aspartic acids, residueswhich are known to act as catalytic nucleophiles in glycosidases, were present in candidatepeptides of both masses. These residues were Asp-214 in peptides RIDTAGL (residues212-218) and FRIDTAGL (211-218), and Glu-390 in peptides GQEIGQI (388-394) andVYQGQEIG (385-392). However, only the former aspartic acid residue, Asp-214, ishighly conserved among this family of proteins as would be expected of a catalyticallyessential residue, and the following data show that the peaks observed at m/z 925 and 1073indeed correspond to labelled peptides with the sequences RIDTAGL and FRIDTAGL.Neutral loss chromatograms of the 5FIdo- and the 5FG1u-labelled digests weresimilar, both giving the same two masses (shown in Figure 2.1 8D and 2.1 8E) at nearly thesame retention times, indicating that the L-idosyl and D-glucosyl moieties had labelled thesame residue in two overlapping peptides (Figure 2.19). Both peaks were stronger in the5FIdo-labelled digest, probably because of the greater accumulation of theintermediateresulting from the less rapid turnover of the 5-fluoro idosyl moiety.Interestingly, despite identical masses, the 5FG1u- and 5FIdo-labelled peptides appeared toelute slightly differently. Though retention times can vary between runs, the difference inretention times between the two peptides was 1.6 mm in the 5FIdo-labelled digest, and afull 2 mm in the 5FGlu-labelled digest, suggesting that the structural differences betweenthe 5-fluoro glucosyl esters and their idosyl analogues influence the elution of these heptaand octa-peptides from the column. Similar results were obtained with Agrobacteriwn -glucosidase (Section 2.6.1, not shown).The first technique applied to the defmitive identification of these candidate peptidesinvolves tandem mass spectrometry to fragment the selected ions by collision-induceddissociation (CII)) and detect the fragments so generated (Busch & Cooks, 1983).7116 18 20 22 24Time (mm)925D.i500 1000 1500 2000m’zE 1073. I Ii I1500nlzFigure 2.18. ESMS experiments on a-glucosidase proteolytic digests. (A) Labelled with5FIdo, TIC in normal MS mode, (B) labelled with 5FIdo, TIC in neutral loss mode, and(C) unlabelled, TIC in neutral loss mode. (D) Mass spectrum of peptide at 18.9 mm inpanel B (E) Mass spectrum ofpeptide at 20S mm in panel B.A16 18B22 2416 18 20 22 24Time bin)C100.75;so250,_. 100:..E 5001007550250- 100‘ 75502501007550250500 moo 200072100Figure 2.19. Neutral loss Total Ion Chromatograms of a-glucosidase proteolytic digests.(A) Labelled with SFIdo, (B) labelled with 5FG1u, (C) unlabelled. The mass spectra ofthe peaks at 18.9 and 20.5 mm in panel A, and at 18.5 and 20.5 mm in B, show identicalmasses and are displayed in Figure 2.19D and E. Chromatograms were obtained onconsecutive runs on the same day.755025A0C-. 7510014 16 18 20Thi(min)B22 2450Time (mm)C. 75502514 16 18 20 22Time (mm)2473Daughter ion scans at increased collision gas energy of the 5FIdo-derived peak at mlz 925,which gave the most intense neutral loss signal, resulted in loss of the label giving rise tothe peak at m/z 745 (Figure 2.20). The peak at mlz 906 is consistent with the eliminationof I{F (-20) from the 5-fluoro glycosyl-labelled peptide. Ions at m/z 887 and 727 areconsistent with elimination of water (-18) from threonine-containing peptides at m/z 906and 745. The peak at mlz 633 is likely due to side chain loss from leucine and isoleucine.Further peptide bond fragmentation was also observed, with ions at m/z 385 (b2) and 270(b3) corresponding to arginylisoleucylaspartyl and arginylisoleucyl species resulting fromlosses of TAGL and DTAGL fragments, respectively, from the C-terminus of the parentpeptide. Typical of arginine-containing peptides, a ions were observed, botharginylisoleucyl and arginyl species at m/z 225 (a2) and 112 (a1) having also lost ammonia(-17). These data indicate that the labelled peak at m/z 925 has the sequence RIDTAGL.RIDTAGLtoo75 RIDTAGL+5FIdo92650 RRIa1a2 -HF112 RI RID906b2385 -LIt 72770 270 633 887t.[.J. i 11111111 liii i i I iii____________ _____________200 400 600 800 1000fl%/ZFigure 2.20. Tandem MS/MS daughter ion spectum of the SFIdo-labelled active sitepeptide (m/z 925 in the singly-charged state).74Further confirmation of the sequence was obtained by high resolution liquidsecondary ion mass spectrometry (LSIMS) of the 5FIdo-derived peptides. The accuratemasses, determined by LSIIvIS, of the peaks at mlz 925 and 745 (putatively the labelled andunlabelled RIDTAGL peptides in the singly-charged state) were 925.46193 and745.4 1832, respectively. No possible peptides derived from any type of proteolyticcleavage of the native protein have a mass within 5 ppm of the accurate 925.46193 mass,indicating that this peak can only be due to a labelled peptide. Indeed, a 5-fluoro glycosyllabelled RIDTAGL species (calculated mass 925.46423, Am = 2.48 ppm) is consistentwith this mass. Only one possible peptide has a mass within 5 ppm of the measured745.41832 mass of the unlabelled peptide: R11)TAGL, with a calculated mass of745.42082 (Am = 3.36 ppm), confirming that the smaller peptide labelled must have thesequence RIDTAGL.Similarly, the accurate masses of the peaks at mlz 1073 and 893 (putatively thelabelled and unlabelled FRII)TAGL peptides in the singly-charged state) were 1072.53332and 892.489 14, respectively. The accurate mass of the peak at mlz 1072.53332 is indeedconsistent with a 5-fluoro glycosyl-labelled FRII)TAGL species (calculated mass1072.53264, Am = 0.63 ppm). Only one other possible peptide has a mass within 5 ppmof this mass: SDQIFSFTK (residues 519-527, calculated mass 1072.53 150, Am = 1.70ppm). However, a peptide of this mass at the same retention time was not present in theunlabelled digest. Similarly, only two possible peptides have masses within 5 ppm of theaccurate 892.48914 mass of the unlabelled peptide: FRIDTAGL, with a calculated mass of892.48925 (Am = 0.12 ppm), and RAIFESAV (residues 193-200, calculated mass892.48924, Am = 0.11 ppm). However, a RAIFESAV peptide is inconsistent with theMS/MS results above, indicating that the larger peptide labelled has the sequenceFRIDTAGL.The attachment site of the 5-fluoro glycosyl moiety, thus the nucleophile, wasconfirmed by aminolysis of the labelled peptide. A labelled digest treated with 30%75ammonium hydroxide for fifteen minutes at 500C was examined by mass spectrometry andcompared to an untreated, labelled control digest. In the aminolyzed sample, the labelledpeptide of m/z 925.5 present in the untreated digest was absent and was replaced by a newpeptide of m/z 745.0, consistent, within error, with aminolysis of a glycosyl ester toglutamine, to give a RINTAGL peptide of one mass unit less than the parent, unlabelledpeptide.The aspartic acid equivalent to Asp-214 is absolutely conserved in all members ofFamily 13 glycosyl hydrolases (Figure 2.21). It has been suggested previously that thisresidue was the catalytic nucleophile in this family (Mooser, 1991; Svensson and Sogaard,1993). Interestingly, however, this suggestion was based in part on the supposedhomology of residues equivalent to Asp-214 and the surrounding sequence in Family 13enzymes with the short active site regions surrounding Asp-505 or Asp-1394 in thesucrase-isomaltase complex (the enzyme has two homologous active sites), both of whichare labelled by the active site-directed inhibitor CBE. Mutagenesis results by Hermans eta!. of the equivalent Asp-5 18 residue in lysosomal a-glucosidase expressed in COS- 1 cellsare equivocal (Hermans et at., 1991). The asparagine mutant exhibits activity 2 to 6% ofwild-type, a level some one hundred- to one thousand-fold greater than expected fornucleophile position mutants which have been observed to show greatly reduced activity inmechanistically similar enzymes (typically, < 10-5 of wild type activity in purified mutantsof several retaining 3-glycosidases) (Miao er a!., 1994; Withers et at., 1992; Yuan et at.,1994). However, the — 5% level of wild-type enzymatic activity reported by Hermans etal. represents the background in the assay system employed. This assay system wastherefore too insensitive to measure the true activity of a mutant with greater than ahundred-fold reduction in activity. Moreover, this region in sucrase-isomaltase is notconvincingly homologous to the region surrounding Asp-214 in the yeast enzyme (Figure2.21), and indeed, sucrase-isomaltase (Family 31) and yeast cz-glucosidase (Family 13)belong to different, albeit likely related, glycosidase families (Henrissat, 1991; Henrissat &76A.Saccharomyces cerevisiae cz-glucosidase 203 WL2HVD.ER IDTAGLYSCandida albicans cc-glucosidase 195 WLKV.QER IDTAGMYSBacillus thermoglucosidasius sucrase-isomaltase 188 WLKVD_ER MDVINMI SPediococcuspentosaceus a-glucosidase 190 WNKVR MDVINQISStreptococcus mutans glucodextranase 183 W.mKIGER MDVIDMIGHomo sapiens pancreatic cx-amylase 186 L2IVAR LDASKHMWBarley cz-amylase, Type A 193 KS2L.FAWR LDFARGYSTaka amylase, A. oryzae 216 VSNYS IDLR IDTVKHVQBacillus subtilis cc-amylase 165 ANDADR FDAAEHIEBHomo sapiens intestinal sucrase-isomaltase*isomaltase 494 HQEVQYilGLW IDMNEV*sucrase 1383 NEKMK2GLW IDMNEPHomo sapiens lysosomal ct-glucosidase 507 HDQVP2GMW IDMNEP.aNCandida tsukubaensis a-glucosidase 515 SEIVDESGIW LDMNEPSchwanniomyces occidentalis glucan 459 YELTPE.GIW ADMNEVcx- 1 ,6-glucosidase (glucoamylase 1)Aspergillus niger cc-glucosidase chain P2 213 SKKVAGVW YDMSEVFigure 221. A) Alignment of region containing the catalytic nucleophile aspartate labelledby 5-fluoro glycosyl fluorides in selected Family 13 glycosyl hydrolases (boxed). B)Alignment of region containing active site aspartate labelled by conduritol B epoxide (CBE)in selected Family 31 glycosyl hydrolases. The sequences shown are. sp P07265 yeast aglucosidase, sp Q02751 C. albicans a-glucosidase, gp D10487 Bacillusthermoglucosidasius sucrase-isomaltase, gp L32093 Pediococcus pentosacêus aglucosidase, sp Q99040 Streptococcus mutans glucodextranase, sp P00693 Barley aamylase, Type A, sp P04746 Homo sapiens pancreatic a-amylase, gp M33218 A. oryzaeTaka-amylase A sp P14410, sp P00691 Bacillus subtilus a-amylase, Homo sapiensintestinal sucrase-isomaltase, sp P10253 Homo sapiens lysosomal cx-glucosidase, spP29064 Candida tsukubaensis a-glucosidase, sp P22861 Schwanniomyces occidentalisglucan a-1,6-glucosidase (glucoamylase 1), and pir JC1200 Aspergillus niger aglucosid.ase chain P2. Accession numbers correspond to those in the SWISS-PROT,GenPept, or PIR databases. Absolutely conserved residues in either Family 13 (A) orFamily 31 enzymes (B) are shown in bold; other conserved residues are underlined.• Labelled by 5-fluoro glycosylfluorides* Labelled by conduritol B epoxide77Bairoch, 1993). Furthermore, CBE may have labelled other active site residues in sucraseisomaltase, possibly the acid/base catalysts, as has been observed with conduritol epoxidederivatives in other glycosidases (Gebler, 1992; Miao, 1994).While the identity of the catalytic nucleophiles in sucrase-isomaltase and otherFamily 31 enzymes remains equivocal, that of yeast ct-glucosidase and Family 13 enzymes((x-amylase family) is clearly established by the present work and is consistent with theavailable mutagenesis and structural data, as described below. Consistent with theassignment of a catalytic nucleophilic role to Asp-214, mutagenesis of the equivalentresidue (Asp-176) in the Bacillus subtilis ct-amylase of Family 13 to asparagine results inthe expected drastic reduction in activity, from 8500 Units in the wild-type to < 0.069Units in the mutant with starch as substrate, a > 105-fold reduction in wild-type activity(Svensson & Sogaard, 1993; Takase et a!., 1992). By contrast, mutagenesis of theremaining residues of the conserved trio in the same enzyme (Glu-208 and Asp-269) toglutamine or asparagine, respectively, result in less severe reductions in activity.As well as defining the role of a key catalytic residue, the unequivocal identificationof Asp-214 as the catalytic nucleophile is of considerable usefulness in the elucidation ofthe roles of other active site residues of Family 13 glycosidases. In particular, the presentresults aid in rationalizing the available X-ray structural data, which with regard to defmingthe catalytic roles of the conserved active site residues, is unfortunately inconclusive.Inspection of the X-ray structure of porcine pancreatic ct-amylase complexed with thepseudotelrasaccharide inhibitor acarbose (Qian et a!., 1994) reveals that Asp-197 (theequivalent residue to Asp-214) lies on the side of the active site cleft opposite to Glu-233and Asp-300, the latter two residues forming hydrogen bonds with the “glycosidic” amineof the acarviosine moiety. In the acarbose-complexed structure, Glu-233 and Asp- 197 arenearly equidistant to the “anomeric” carbon of the acarviosine moiety (3.5 versus 3.3 A,respectively). However, the present results clearly implicate Asp-197 as the nucleophile,suggesting that Glu-233 and Asp-300 (which are apparently hydrogen-bonded to each78other via an intervening water, as well as to the “glycosidic” amine) may be involved inacid/base catalysis. Furthermore, the use of this novel class of inactivators to specificallylabel and inhibit this important class of glycosyl hydrolases may have implications for themanagement of diabetes or obesity, conditions in which it is desirable to limit the digestionof a-linked oligosaccharides by intestinal a-glucosidases.79CHAPTER III2,2-DIHALO GLYCOSIDES AS MECHANISM-BASED INACTIVATORSOF ot-GLYCOSIDASES803.1 Introduction2-Deoxy-2-fluoro glycosyl fluorides have been shown to be effective mechanism-based inactivators of retaining -glycosidases. Inactivation of these enzymes occurs by theformation of a stabilized 2-deoxy-2-fluoro-a-D-glycosyl-enzyme intermediate which turnsover only very slowly. Electionic destabilization by the electronegative fluorine of theelectron-deficient transition states involved in both the formation and hydrolysis of thisintermediate slows both steps in the enzymic reaction. Incorporation of a good leavinggroup (fluoride, PKa = 3.45, or 2,4-dinitrophenolate, PKa = 3.9) renders this intermediatekinetically accessible. However, the corresponding cx anomers of these compounds wereless successful at inactivating a variety of cx-glycosidases, at best partial inactivation beingseen (Withers et al., 1988). For example, incubation of Jack bean a-mannosidase with 2-deoxy-2-fluoro-cx-mannosyl fluoride resulted in no time-dependent inactivation of theenzyme, while incubation of yeast cx-glucosidase with the 2-deoxy-2-fluoro-cx-D-glucosylfluoride resulted in only partial inactivation according to complex kinetics. It was latershown that even this inactivation of the yeast enzyme was due to a contaminant in theinactivation preparation, and that 2-deoxy-2-fluoro-cx-glucosyl fluoride is in fact a slowsubstrate for the enzyme (see Chapter 2, and (Braun, 1995)).The failure of 2-deoxy-2-fluoro-cx-D-glucosyl fluoride to inactivate yeast cxglucosidase is due to turnover of a 2-deoxy-2-fluoro f3-D-glucosyl-enzyme intermediate at arate too great to permit its significant accumulation. To overcome this problem, it wasproposed to further retard the turnover of the glycosyl-enzyme intermediate by introductionof a second fluorine at C-2. Since the presence of these geminal fluorines would beexpected to result in a similar large decrease in the rate of formation of the intermediate,exceptionally good leaving groups such as chloride (pKa -7) would be incorporated intothese inactivators, to ensure that the glycosylation step is not too compromised to precludeformation of the intermediate at reasonable rates.813.2 Specific AimsThe initial goal is the synthesis of novel 2-deoxy-2,2-dihalo-a-D-glycopyranosylhalides, specifically 2-chloro-2-deoxy-2-fluoro-a-D-glucopyranosyl chloride (3.1), 2-chloro-2-deoxy-2-fluoro-cx-D-mannopyranosyl chloride (3.2), and 2-deoxy-2,2-difluoro-cL-D-arabinohexopyranosyl chloride (3.3). These compounds may be effectivemechanism-based inactivators of the corresponding a-glycosidases, functioning by theformation of extremely stabilized 2-deoxy-2,2-dihalo glycosyl-enzyrne intermediates. Theywill therefore be kinetically evaluated with yeast cx-glucosidase and Jack bean xmannnosidase.HO HOFigure 3.1. Structures of 2-chloro-2-deoxy-2-fluoro-a-D-glucopyranosyl chloride,2C12FaG1uC1 (3.1), 2 -chloro-2 -deoxy-2 -fluoro-a-D-mannopyranosyl chloride,2C12FaManC1 (3.2), and 2 -deoxy-2 ,2 -duoro-a-D-arabinohexopyranosyl chloride,2,2Fcv.AraCl (3.3).The 2-deoxy-2,2-difluoro compound, since it has two small, identical substituentsat C-2, might be expected to be an inactivator of either an c-mannosidase (whosesubstrates have an axial hydroxyl at C-2) or an (x-glucosidase (whose substrates have anequatorial hydroxyl at C-2). By contrast, the 2-chloro-2-deoxy-2-fluoro mannosyl chloridewould not be expected to be sterically accommodated by and thereafter inactivate an xglucosidase since the axial chlorine at C-2 is significantly larger than the hydrogen at thatposition in a glucoside substrate (fable 3.1). Similarly, the 2-chloro-2-deoxy-2-fluoroglucosyl chloride would not be expected to inactivate an a-mannosidase. However, the 2-chloro-2-deoxy-2-fluoro glucosyl and mannosyl chlorides may inactivate the ‘correct’enzymes since chJorine is very similar in size to a hydroxyl, and should not cause greatsteric interference with binding. However, such compounds would likely not participate in3.182significant hydrogen bonding interactions at that position. Such possibly selectivemechanism-based inactivators of ct-glycosidases may be useful probes of mechanism andmay be of therapeutic interest for the inhibition of a-mannosidases and cx-glucosidases.Table 3.1: Size comparison of C-H, C-F, C-Cl, and C-OH groups (Adapted from Walsh,1983).Group Bond length (A) Van der Waals Radius of Total (A)C-X Substituent (A)C-H 1.09 1.20 2.29C-F 1.35 1.35 2.74C-Cl 1.77 1.80 3.57C-OH 1.43 2.10 3.53Finally, derivatization and identification of the active site nucleophiles in theseenzymes by electrospray mass spectrometry, as described previously, will be attempted.The residue labelled in yeast cz-glucosidase by these compounds would be expected to beidentical to the residue identifed by labelling with the 5-fluoro glucosyl and idosyl fluorides(Chapter 2).Results and Discussion3.3 SynthesisMuch of the work on synthesis and enzyme kinetic analysis of the following 2,2-dihalo glycosyl chlorides was carried out by an undergraduate student, Mr. Wai Yeung,under my supervision.a) Synthesis of 2-chloro-2-deoxy-2-fluoro-x-D-glucopyranosyl chloride,2C12FctG1uC1 (3.1)83The syntheses of both the 2-chloro-2-deoxy-2-fluoro-glucosyl and -mannosylchlorides were attempted from the same acetylated 2-fluoro glucal precursor (Scheme 3.1).3,4,6-Tri-O-acetyl-2-deoxy-2-fluoro-cL-D-glucopyranosyl fluoride (3.4) (Shelling et al.,1984) was converted to the protected glycosyl bromide by treatment with 45% HBr/HOAc.After workup, the crude bromide was treated with triethylamine in refluxing acetonitrile togive 3,4,6-th-O-acetyl-2-deoxy-2-fluoro-D-glucal (3.5) in 98% yield. This glycal waschlorinated (Bradley & Buncel, 1968) by stirring a cooled solution of the glycal in carbontetrachioride saturated with chlorine to afford 3,4,6-tri-O-acetyl-2-chloro-2-deoxy-2-fluoro-cx-D-glucopyranosyl chloride (3.6) in 34% yield. The per-O-acetylated chloride wasdeacetylated by treatment with anhydrous ammonia in freshly distilled methanol to giveonly the gluco product, 2Cl2FxGluCl (3.1) in 50% yield.Only one other deprotected stable glycosyl chloride has been reported (Bradley &Buncel, 1968). Not surprisingly, this compound is another 2,2-dihalo glycosyl chloride,2,2-dichloro-2-deoxy-cL-D-arabinohexopyranosyl chloride. The instability of free glycosylchlorides is due to facile hydrolysis of the anomeric chloride, but in these cases the twohalogens at C-2 provide sufficient electionic stabilization to permit their deprotection andisolation.b) Attempted synthesis of 2-chloro-2-deoxy-2-fluoro--D-mannopyranosyl chloride,2Cl2FcxManCl (3.2)The synthesis of 2-chloro-2-deoxy-2-fluoro-ct-D-mannopyranosyl chloride (3.2)was attempted in a similar manner from the same 2-fluoro glucal precursor. Studiesinvolving the addition of chlorine to D-glucal tri-O-acetate, in which the reaction solventwas varied, revealed that the proportions of the manno and gluco products produced weredependent principally on the polarity of the solvent used (Igarashi et a!., 1970). Nonpolarsolvents such as carbon tetrachioride yielded more cis-addition products and a higherproportion of glucosides, whereas polar solvents such as nitromethane resulted in841. 45%HBr/HOAc,6h2. Et3N/CH3N, reflux 18 hFQ2/CC14,0°C, 1812/CH3N0,0°CNH9!CH3O , 0°C, 4.5 h 50%Scheme 3.1. Synthesis of 2 -chloro-2 -deoxy-2 -fluoro-cc-D-glucopyranosyl chloride (3.1)and attempted synthesis of 2-chloro-2-deory-2-fluoro-a-D-mannopyranosyl chloride(3.2).more trans-addition products and a higher proportion of mannosides. Reaction of theglycal (3.5) in a number of polar solvents was therefore investigated. However, reactionof 3.5 in acetonitrile or nitromethane yielded complex mixtures from which no mannocompounds and only low yields (7%) of the acetylated a-glucosyl chloride (3.6) could beisolated in several attempts, possibly due to the instability of the manno compounds onsilica gel. This may be due to the more facile formation of a chioronium ion in the 2-3.43.185chloro-2-deoxy-2-fluoro a-mannosyl chloride. The synthesis of 2Cl2FctManCl (3.2) wasnot pursued further.c) Synthesis of 2-deoxy-2,2-difluoro-a-D-arabinohexopyranosyl chloride,2,2FctAraCl (3.3)The synthesis of the 2,2-difluoro chloride (3.3) proceeded from the same commonintemiediate 2-fluoro glucal used for the 2-chloro-2-deoxy-2-fluoro compounds (Scheme3.2). The 2-fluoro glucal was fluorinated by addition of acetyl hypofluorite across thedouble bond, as described previously (McCarter et at., 1993). A slurry of sodium acetate,glacial acetic acid and CFC13was cooled to -78 C and a 20% fluorine/helium mixture wasbubbled through the slurry to produce acetyl hypofluorite in situ. The glucal dissolved inCFC13was then added to the slurry to produce, after workup, 1,3,4,6-tetra-O-acetyl-2,2-difluoro-cz-D-arabinohexopyranose (3.8) in 76% yield. 1,1-Dichioromethyl methyl etherwith a Lewis acid catalyst has been employed to synthesize 3,4,6-th-O-acetyl-2-chloro-2-deoxy-cx-glucopyranosyl chloride from 1 ,3,4,6-tetra-O-acetyl-2-chloro-2-deoxy-3-glucopyranose (Farkas et a!., 1977). Therefore, the synthesis of the protected 2,2-difluoro-c-D-arabinohexopyranosy1 chloride (3.9) was attempted directly from the per-U-acetate 3.8. However, several attempts under various conditions with different Lewisacids (ZnCl2,SnCl2)produced only a low (7%) yield of 3.9 and a compound identified as3,4,6-tri-O-acetyl- 1 ,5-anhydro-5-chloro-2-deoxy-2,2-difluoro-D-glucitol (3.10) in 15%yield after chromatography.The final successful synthetic pathway involved a two-step process in which 3,4,6-tri-O-acetyl-2,2-difluoro-x-D-arabinohexopyranose (3.11) was synthesized first, and thentreated with thionyl chloride. Selective deacetylation of the anomeric centre of 3.8 withhydrazine acetate in dimethylformamide for 4 days afforded 3.11 in 23% yield. Thishemiacetal was then added to freshly distilled thionyl chloride and refluxed for three days,861. 45%HBr/HOAc,6hOAc2. Et3N/CH3N, reflux 18 hAcAcO1 98% F3.4 FAcOF/CFC13,-78° 76%AFOAcF iC12HOCH3/ZnC12,OC7% AcOL..L+ AcO.ç03.8FOAc3.10 NH23Ac0 / DMF/ 50°C, 4 d 23%15%-OAc -OMSOC12/50° ,3d AcO1%4jAcOL_62%AcOLLFlFl39 ci 3.11 OH83% NH3!CH3O , OOC, 4.5 h-OHF IciScheme 3.2. Synthesis of 2 ,2-difluoro-a-D-arabinohexopyranosyl chloride 3.3.87yielding the desired product 3.9 in 62% yield after chromatography. The extendedreaction times required for this synthesis are a testament to the deactivating effects of thegeminal fluorines at C-2 on reactivity at the anomeric centre. An additional complication ofthe synthesis of these 2,2-difluoro compounds is their difficult visualization by TLC. Mostchar extremely poorly when visualized with sulphuric acid or ammonium molybdate, andare not of course, UV-active. However, 3.9 was smoothly deacetylated using anhydrousammonia in dry methanol to yield 2,2FczAraCl (3.3) as a white solid in 83% yield.The unintended synthesis of 3.10 in the 1,1-clichioromethyl methyl ether reactionwas a complete surprise. A possible explanation for the formation of 3.10 could be theformation of C5-05 oxocarbenium ion via a 1,3 hydride shift (Figure 3.2). The presenceof the two electronegative fluorine substituents at C-2 might well promote such arearrangement by destabilization of the initially formed C1-05 oxocarbenium ion. There issome precedent for such 1,3 hydride shifts (Skell & Maxwell, 1962). 2-Methyl-l-butanolreacted with KOH and bromoform results in a mixture of products, including 3-methyl-i-butene, presumably via a 3-methyl-2-butyl cation (Figure 3.3). However, 2-methyl-2-butanol gives none of this product, suggesting that the 3-methyl-2-butyl cation in theprevious reaction of 2-methyl-1-butanol is not formed via a 1,2 shift from an initial 2-methyl-2-butyl cation, but rather directly from the 2-methyl-1-butyl cation via a 1,3 hydrideshift.3.4 Kinetic Studies with Yeast a-glucosidase and Jack Bean -mannosidaseRetaining glycosidases are proposed to follow the kinetic scheme shown in Figure3.4, according to the Koshland mechanism:881,3 hydride shiftC)___Cf attackOAcAcO3.9 3.10Figure 3.2. Possible mechanism for the formation of3.10.2-methyl-2-butyl cation+ 1,2 hydride shift1,2 hydride shift1,3 hydride shift +2-methyl-1-butyl cation 3-methyl-2-butyl cationFigure 3.3. Synthesis of 3-methyl-i -butene from 2-methyl-i -butanol, possibly via a 1,3hydride shift.OH 4*89k1 k2 k3E + Gly-X E.Gly-X E-Gly E + GlyOHk1HX H20Figure 3.4. Kinetic scheme for retaining glycosidases. E is the enzyme, Gly-X is thesubstrate, E.Gly-X is the noncovalent enzyme-substrate complex, HX is the aglycone, EGly is the covalent glycosyl-enzyme complex. k1 and k.1 are the rate constants for rapidand revesible formation of the initial Michaelis complex, and k2 and k3 are the first orderrate constantsfor glycosylation and deglycosylation, respectively.To trap the intermediate requires that k3 <<k2. The fact that 2-deoxy-2-fluoro-a-D-glucosyl fluoride is a substrate for cz-glucosidase indicates that with a single C2 fluorine,the 2-deoxy-2-fluoro 3-D-glucosy1-enzyme intermediate turns over at a rate too great topermit its significant accumulation (see Chapter 2). Incorporation of a secondelectronegative C2 halogen was proposed to further slow k3. If the aglycone is anexceptional leaving group (e.g. chloride, PKa — -7), then the rate of glycosylation might beexpected to be much greater than the rate of deglycosylation, i.e. k2 >> k3. In suchcircumstances, an extremely stabilized glycosyl-enzyme intermediate will accumulate,resulting in enzyme inactivation and the kinetic model becomesk1 k2E+I El E-IkHXFigure 3.5. Kinetic schemefor inactivation of retaining glycosi&zses by accunudation of acovalent intermediate. E is the enzyme, I is the inactivator, E•I is the noncovalent enzymeinactivator complex, HX is the aglycone, E-I is the covalent glycosyl-enzyme intermediate.Rate constants are as in Figure 3.4.This kinetic model predicts a time-dependent inactivation of the enzyme. If [ii ismuch greater than [E0], [I] can be assumed to be essentially constant during the reaction,and pseudo first-order kinetics with respect to enzyme concentration will be observed. InMichaelis-Menten form, the equation for this process is:90v=kjjffi (3.1)K1 + [I]where v1 is the inactivation rate, k1 is the rate constant for maximal inactivation, and K1 isan apparent dissociation constant for all forms of enzyme-bound inactivator. From Figure3.5, it can be seen that k1 = k2, and K1 =k..1/k. Equation (3.1) can be rewritten asV = (3.2)wherekObS=......kflL (3.3)K1 + [I]The rate of inactivation, v1 is equal to=k0b[E] (3.4)dtThus ln[E0] =- l(obst (3.5)Note that if K>> [1], then Equation (3.3) becomes‘Sbs = (k/K)[Ij (3.6)3.4.1 Inactivation of yeast x-glucosidase with 2CI2FctGIuCI (3.1) and2,2F(xAraCi (3.3)To determine the kinetic parameters for reaction of each of the above compoundswith a-glucosidase, the enzyme was incubated in the presence of various concentrations ofeach of 2Cl2FctGluCl and 2,2FczAraCl. Aliquots were removed from the inactivationmixtures at intervals and assayed for remaining enzyme activity by dilution into cellscontaining p-nitrophenyl c-glucopyranoside. This prevents further inactivation by diluting91the inactivator and providing a large excess of a competing ligand. The observed pseudo-first order rate constants for the inactivation process,k0 at each inactivator concentrationwere determined by plotting the residual enzyme activity against time and fitting the data toa weighted single-exponential equation, using the curve-fitting program GraFit(Leatherbarrow, 1990). Values for k and K1 were calculated by fitting the kobs values sodetermined to Equation (3.3) using GraPh.Time dependent inactivation of yeast a-glucosidase by 2Cl2FcxGluCl (3.1) wasobserved. Inactivation followed pseudo-first order kinetics at each of a series of inactivatorconcentrations, allowing determination of a series ofk0 values. Saturation of the enzymewas observed at higher 2Cl2FctGluCl concentrations, allowing determination of theindividual kinetic parameters k1 = 0.25 ± 0.06 min1 and K1 = 47 ± 19 mM (Figure 3.6).The corresponding half-life at saturating concentrations of inactivator is 2.8 mm. The k/K1value is 5.3 x 10 ± 3.4 x 10 min’mIvf’. Inactivation of ct-glucosidase by thiscompound is in keeping with expectations since the equatorial chlorine at C-2 of2Cl2FczGluCl is similar in size to, and of the same configuration as, the equatorialhydroxyl of the natural glucoside substrate, while the axial fluorine is only slightly largerthan the axial hydrogen in the substrate (Table 3.1). Protection from inactivation with acompetitive ligand, in this case 1-deoxynojirimycin, DNJ (K1 = 9.6 IIM (Legler, 1990)),indicates that the inactivation is active-site directed (Figure 3.7). The presence of DNJ (63p.M), reduced the kobs at 11.3 mM 2Cl2IkxGluCl from 0.056 min1 to 0.020 min1 (Figure3.7)Inactivation of retaining glycosidases by the trapping of a stabilized glycosylenzyme intermediate represents the first step in the turnover of a very slow substrate.According to this kinetic model, when inactivated enzyme is freed of excess inactivator,incubated at an appropriate temperature and aliquots removed for assay, the activity of theenzyme would be expected to slowly return as the glycosyl-intermediate is hydrolyzed,releasing free enzyme and 2-chloro-2-deoxy-2-fluoro-glucose. However, no reactivation92A0 40 80 120Time (mm)[2C12FaG1uC1] (mM)Figure 3.6. Inactivation of yeast cc-glucosidase by 2C12FaG1uC1. A) Plot of residualactivity versus time at the indicated inhibitor concentrations: 3.78 mM (X), 5.67 mM (V),1134 mM (0), 22.7 mM (N), 34 mM (0), and 60 mM (A). B) Replot ofk0bS valuesfrom above.20 40 60930. 40Figure 3.7. Protection against inactivation by 2C12FaG1uC1. Inactivation a a singleconcentration of2,2FaAraCl (11.3 mM) in the presence of the indicated concentrations of1-deoxynojirimycin: 0 ,nM (0), 63 pM (3).of ct-glucosidase that had been inactivated by 2C12FczG1uC1 was observed, even afterincubation for ten days at 37°C in buffer (results not shown). Such an exceedingly slowreactivation rate is consistent with the enormous destabilization of the deglycosylationtransition state afforded by the two geminal halogens at C-2, resulting in an extremelystable 2-chloro-2-deoxy-2-fluoro-3-glucosyl-enzyme intermediate.Inactivation of yeast a-glucosidase by 2,2FcLAraC1 (3.3) was also observed.Saturation of the enzyme was observed in the range of inhibitor concentrations tested (11 to93.5 mM). Values of kobS were plotted against time, fitted to Equation (3.3), and thekinetic parameters determined to be k1 = 8.8 x 10± 1.0 x min’ and K1 = 9.7 ± 2.7mM. The corresponding value fork1fK is therefore 9.1 x 1(1 ± 3.3 x 10 min’mM’(Figure 3.8). The inactivation of the enzyme by 2,2FaAraCl is thus 58-fold slower thanthat by 2C12FG1uC1 in terms ofk1/K, and 280-fold slower in terms of k1. Protectionagainst inactivation was observed with DNJ (200 p.M) reducing the kobs at 11 mlvi2,2FaAraCl from kobs = 3.7 x 10 mint to kObs = 9.9 x 10 mint (Figure 3.9).10 20 30Time (mm)94AI0 400 800 1200 1600 2000 2400Time (mm)0.10080.060.040.02B0.00080.00060.00040.000200 100[2,2FaAraCl] (mM)Figure 3.8. Inactivation of yeast a-glucosidase by 2,2FcxAraCl. A) Plot of residualactivity versus time at the indicated inhibitor concentrations: 9.9 mM (C)), 11 mM (A),19.8 mM (•), 49.5 (D), 715 (N), and 93.5 mM (a). B) Replot of k0b values fromabove.20 40 60 8095As with the 2C12RxG1uCI-inactivated enzyme, no reactivation of the 2,2FxAraCl-inactivated enzyme was observed over 3 days at 37°C, in buffer alone or in the presence of100 mM cz-methyl glucoside as a possible transglycosylation acceptor (results not shown),again consistent with formation of an extremely stabilized glycosyl-enzyme intermediate.0.1C0.08. 0.060.040 400 800 1200 1600 2000 2400Time (mm)Figure 3.9. Protection against inactivation by 2,2FcrAraCl. Inactivation at a singleconcentration of2,2FaAraCl (11 mM) in the presence of the indicated concentrations of 1-deoxynojirimycin: 0 pM (•), 200 pM (0).Recall that the kinetic parameters (k and Km) for 2FczGluF with yeast aglucosidase are 96 rnin1 and 4.8 mM, respectively (versus 1500 min1 and 0.93 mlvi forthe parent substrate (x-glucosyl fluoride, respectively. See Chapter 2). The incorporationof a second halogen at C-2 has thus resulted in a reduction in the rate of the deglycosylationstep by at least 108-fold over that of the 2-deoxy-2-fluoro glucosyl derivative. This is inaddition to the one to two order of magnitude reduction afforded by the substitution of theequatorial hydroxyl by a fluorine. However, it is difficult to quantify the full effect of thesecond halogen since no reactivation was observed. It is not certain whether the poorbinding of 2Cl2FctGluCl with ct-glucosidase is due principally to the equatorial chlorine at96C-2, which is slightly larger than a hydroxyl, or to the axial fluorine at this position, whichis slightly larger than a hydrogen. However, the poorer binding seen with the 2-chloro-2-deoxy-2-fluoro glycosyl chloride compared to the 2-deoxy-2,2-difluoro analog suggeststhat the relatively bulky equatorial chlorine is largely responsible for the poor binding of theformer compound with this enzyme. Although the 2-deoxy-2,2-difluoro compound lacksthe steric bulk of the 2-chloro-2-deoxy-2-fluoro compound, and is apparently bound moretightly, it results in much slower inactivation of the enzyme. This is likely a consequenceof the greater electronic effect of the equatorial fluorine (rather than chlorine) in the 2,2-difluoro compound. With both compounds, the colossal reduction in the rate of thedeglycosylation (and glycosylation) steps upon introduction of the second halogen(chlorine or fluorine) is principally a result of very substantial electronic factors.3.4.2 Attempted inactivation of Jack bean cx-mannosidaseJack bean a-mannosidase was tested for inactivation by the 2,2-dihalo glycosylchlorides. No inactivation of Jack bean cx-mannosidase by either 2Cl2FczGluCl (3.1) or2,2F(xAraCl (3.3) was observed. Enzyme was incubated with 2C12FaG1uC1 (20.0 mM)for three days, or 2,2FLAraC1 (20.0 and 10.0 mM) for 15 hours and two days,respectively. No time-dependent loss of activity which could be attributed to the presenceof the inhibitors was detected. It was expected that 2Cl2FczGluCl would not inactivate an(x-mannosidase since binding of this compound would be expected to be sterically hinderedby the equatorial chlorine atom at C-2 in the gluco configuration. However, no stericeffects of equal severity would be expected to have significantly hindered the binding of2,2lkzAraCl by the mannosidase since a fluorine atom is smaller than the axial hydroxylgroup and only slightly larger than the equatorial hydrogen in a mannoside substrate (Table4.).97The lack of inactivation of Jack bean cz-mannosidase by either 2C12FaG1uC1 or2,2FxAraCl prompted an investigation to confirm the stereochemical outcome ofhydrolysis by the cz-mannosidase. This enzyme has been shown to catalyzetransglycosylation reactions, suggestive of a retaining mechanism, but its mechanism hasnot been unequivocally elucidated. ‘H-NMR speciroscopy was used to determine whetherthe glycosidic bond of a-mannosides is cleaved with retention or inversion of anomericconfiguration by this enzyme (Figure 3.10). Spectrum A shows the anomeric protonregion of the1H-NMR spectrum of p-nitrophenyl a-D-mannopyranoside in DO. Thedoublet at 3 = 5.73 ppm (J12 — 1.4 Hz) is due to the equatorial anomeric proton of the amannoside substrate. The large resonance at 6=4.75 ppm is due to HOD. Spectra B andC were recorded at intervals after the addition of enzyme and clearly reveal first theappearance of a resonance at 6 = 5.13 ppm (J12 — 1.4 Hz) due to the equatorial anomericproton of a-mannose, followed by the appearance of a resonance at 6 = 4.85 ppm (j12 —1.0 Hz) from f3-mannose formed by mutarotation of the cc-anomer (Angyal & Pickles,1972). After 90 minutes (Spectrum D), the substrate has been completely consumed andthe fmal equilibrium anomeric ratio (60:40 a43) has been established. This enzyme clearlycatalyzes the hydrolysis of mannoside substrates with net retention of anomericconfiguration, thus the lack of inactivation by 2,2FctAraCl in particular cannot be thesimple consequence of this enzyme catalyzing hydrolyses via an inverting mechanism.The ‘H-NMR results clearly establish that Jack bean ct-mannosidase is a retainingenzyme. Assuming a Koshland mechanism, the lack of inactivation of this enzyme by2,2FaAraCI must be due either to: 1) a glycosylation step which is rate-limiting, in whichcase 2,2FczAraCl is either a slow substrate, or in the extreme, is not turned over by theenzyme; or 2) a deglycosylation step which is too fast to accumulate an intermediate, andhence 2,2FcxAraCl is again simply a substrate. ‘9F-NMR analysis of samples of2,2FcxAraCl incubated for several days in the presence of a large amount of a-mannosidaserevealed no detectable hydrolysis of 2,2FaAraCl, indicating that the second possibility, a9899Figure 3.10. 1H-NMR determination of the stereochemical course of hydrolysis of p..nitrophenyl a-D-mannopyranoside by Jack bean a-mannosidase. Spectra are of theanomeric proton region of the substrate andproducts before addiron of enzyme (A), and cZ15,45, and 90 mm after addition (B, C, and D, respectively). See text and ChapterS forexperimental details.2,2-difluoro-f3-D-arabinopyranosyl-enzyme intermediate that is turned over rapidly, isunlikely. To further probe this behavior, 2-deoxy-2-fluoro-cx-D-mannosyl fluoride2FcManF, which is not an inactivator of this enzyme (Withers et al., 1988), was evaluatedas a substrate by monitoring the initial rates of release of fluoride using a fluoride electrode.2FaManF was an exceedingly poor substrate and no saturation was observed. However,an estimate for VmfKm of 9.6 x i08 ± 2.0 x i08 Lmin’mg’ at pH 6.5 was determined.This represents a 12 000-fold reduction in rate relative to the parent compound (x-Dmannosyl fluoride (Vm = 6.0 ± 0.6 j.tmolmin’ mg1,Km = 5.2 ± 0.4 mM, Vm/Km = 1.2 Xi0 ± 0.1 x 10 Lmin1mg’ (Howard, S., McCarter, J., Withers, S., unpublishedresults). In contrast to the results with yeast c-gIucosidase, where the rate of enzymatichydrolysis of 2-deoxy-2-fluoro-cx-D-glucosyl fluoride is only reduced by an order ofmagnitude relative to that of the parent a-D-glucosyl fluoride (see Chapter 2), substitutionof the 2-hydroxyl by fluorine in the analogous mannosyl fluoride results in this much largerrate decrease. Possibly, crucial binding interactions involving the axial 2-hydroxyl areformed at the transition state in the ct-mannosidase which are abolished with either the 2-deoxy-2-fluoro or the 2-deoxy-2,2-difluoro glycosyl halides, or this enzyme is moresensitive than is the cx-glucosidase to the electronic effects of even a single fluorine at C2.Unfortunately, the 2-chloro-2-deoxy-2-fluoro mannosyl chloride was not available to testwith the cx-mannosidase, but the results above would suggest that this compound isunlikely to be either an inactivator or a substrate.3.5 Mass spectrometry of Labelled a-GiucosidaseIdentification of the site of attachment of the 2-deoxy-2,2-dihalo sugars, thus of thenucleophilic amino acid residue of yeast a-glucosidase, by electrospray mass spectromeirywas attempted. The labelled residue would be expected to be identical to that identified by100the 5-fluoro glycosyl fluorides, namely Asp-214 (Chapter 2). Peptic hydrolysates of thelabelled and unlabelled enzyme were prepared and analyzed by ESMS as previouslydescribed in Chapter 2.Repeated attempts to obtain a neutral loss signal resulting from loss of the mass ofeither the 2-chloro-2-deoxy-2-fluoro glucosyl moiety (m/z = 199) or the 2-deoxy-2,2-difluoro-arabinopyranosyl moiety (m/z = 183) from peptic digests of inactivated (Xglucosidase, as described previously, were unsuccessful. No significant neutral loss peakswere observed. Apparently, and quite reasonably, the presence of the two halogens at C-2destabilizes the formation of an anomeric radical arising from homolytic cleavage of theinhibitor-enzyme ester bond.The LCMS spectra showing all peptides in digests of x-glucosidase inactivated by2C12FctG1uCI and 2,2FccAraCl were then searched for the same major peptide that waslabelled by the 5-fluoro glucosyl and idosyl fluorides (RJDTAGL, m/z 745.5, see Chapter2). This peptide, when derivatized by a 5-fluoro glycosyl moiety, has a mass of 925.5.When labelled instead by a 2-chloro-2-deoxy-2-fluoro glucosyl moiety or a 2-deoxy-2,2-difluoro-arabinopyranosyl moiety, the expected masses of these peptides would be 943.5and 927.5, respectively. These peptides were not reproducibly identified in peptic digestsof a-glucosidase inactivated by 2C12FxG1uC1 or 2,2FaAraCl, respectively. However,these peaks were identified in a single, brief and partial peptic digest of inactivated aglucosidase, but they decomposed prior to isolation (Braun, 1995). In contrast to the greatstability of the intact 2,2-dihalo glycosyl-enzymes at near neutral pH, cleavage of 2,2-dihalo glycosyl esters on solvent-accessible small peptides at pH 2 may be very facile.Such cleavage probably occurs via attack of water on the ester carbonyl, rather than at theanomeric centre of the sugar. This might be facilitated by the presence of the two C-2halogens, possibly rendering the sugar a better leaving group.1013.6 ConclusionsThe results described herein are significant as they represent only the second modeby which an a-glycosidase has been inactivated by the accumulation of a glycosyl-enzymeintermediate analogous to that formed in the normal catalytic mechanism. Mechanism-based inactivation by these 2,2-dihalo glycosides with exceptional leaving groups thusprovides further proof that the catalytic mechanism of a-glucosidases involves such aglycosyl-enzyme intermediate. Yeast a-glucosidase was inactivated by both 2C12FzG1uC1and 2,2FaAraCl, with 2C12FaG1uC1 being a superior inactivator in terms of k1 and k.IK,but poorer in terms of binding. Neither of these compounds inactivated Jack bean cxmannosidase, despite the apparent mechanistic similarity of these enzymes. The results forboth enzymes are summarized below (Table 3.2).Table 32: Kinetic parameters of 2C12FcxG1uCZ and 2,2FcxAraCl with yeast cc-glucosid.aseand Jack bean a-mannosidase.EnzymeCompoundyeast cx-glucosidase Jack bean cz-mannosidasek1 K1 k1fK(min1) (mM)2C12FaGJuC1 0.25 47 5.3 x No Inactivation±0.06 ±19 ±3.4x102,2FcxAraCl 8.8 x 10 9.7 9.1No Inactivation±1.0x104 ±2.7 ±3.3x105Protection of cz-glucosidase against inactivation by both inactivators by DNJ wasobserved, indicating that these inactivators are active site-directed. After freeing theinactivated cz-glucosidase of excess inactivator, the enzyme did not exhibit significant102reactivation, even after ten days at 37°C, consistent with formation of an extremelystabilized glycosyl-enzyme intermediate. Inactivation of yeast a-glucosidase, and not Jackbean a-mannosidase, by 2,2FxAraCl suggests significant differences in transition statebinding interactions, geometry, or charge development between these enzymes, andpossibly, between retaining ct-glucosidases and retaining a-mannosidases in general.103CHAPTER IVMECHANISM-BASED GLYCOSIDASE INHIBITORS AS PROBES OFHUMAN ENZYMES AND AS POTENTIAL DIAGNOSTICS1044.1 IntroductionGlycosidic linkages in mammalian tissues occur in both glycolipids andglycoproteins. Catabolism of these species occurs principally in the lysosome. Forexample, the f3-glucosidic bond of glucosyl ceramide, which forms the core structure ofgangliosides, is cleaved by -glucocerebrosidase (EC.3.2. 1.45), a membrane-associatedacid glucosidase localized to this organelle. Gaucher’s disease, the most prevalentlysosomal storage disorder, is caused by an inherited deficiency in this enzyme (Brady etat., 1965; Grabowski et al., 1990) which results in accumulation of glucosyl ceraniideprincipally in macrophages of the spleen, liver and bone marrow. A large number oflysosomal storage diseases, each resulting from various genetic deficiencies in a specificglycosidase, have been characterized.Animal models to study the pathology and evaluate treatments for Gaucher’sdisease have been produced in mice using the irreversible glycosidase inhibitors conduritolB epoxide (CBE, 1 ,2-anhydro-myo-inositol) and the related, naturally occuring epoxidebased inhibitor, cyclophellitol (Atsumi etal., 1992; Kanferet at., 1975) (see Figure 1.13).Assays of f3-glucocerebrosidase in tissues of animals administered these compoundsshowed marked inhibition of enzyme activity in spleen, liver, kidney and brain, and aresultant accumulation of glucosyl ceramide was found in these tissues. Related cycitolshave not, however, been used to generate animal models of other lysosomal storage diseasearising from similar glycosidase deficiencies, for two principal reasons. Firstly, thechemical syntheses of epimeric analogues of these epoxide-based cycitol inhibitors (e.g.manno or galacto configurations) or incorporation of appropriate radionucides into thesestructures for tracer studies are non-trivial tasks. Secondly, the inherent reactivity andrelative lack of specificity of this class of inhibitors tends to limit their usefulness asspecific inhibitors of a given class of glycosidases. Indeed, CBE is also an effectiveinhibitor of a-glucosidases (Legler, 1990). More specific covalent inhibitors that may bereadily labelled with appropriate radioisotopes are required.105Recall that retaining glycosidases catalyze the hydrolysis of glycosidic bonds withoverall retention of anomeric configuration. The mechanism involves the formation(glycosylation step) and hydrolysis (deglycosylation step) of a glycosyl-enzymeintermediate via transition states with substantial oxocarbenium ion character (Figure 1.4).2-Deoxy-2-fluoro glycosides with good leaving groups have been shown to act ascovalent, mechanism-based inhibitors of retaining 3-glycosidases by forming a relativelystable 2-deoxy-2-fluoro-glycosyl-enzyme intermediate (See Section 1.5.4, Withers et at.,1988; Withers & Street, 1988). The C-2 fluorine of the inhibitor electronically destabilizesthese positively charged transition states, slowing the rates of both glycosyl-enzymeformation and hydrolysis. In addition, the limited hydrogen bonding capability of fluorinewould be expected to result in loss of significant transition state binding interactions,further slowing these processes. The presence of a good leaving group increases only therate of glycosyl-enzyme formation, thereby resulting in accumulation of the intermediateand inactivation of the enzyme. Turnover of the intermediate and reactivation of theenzyme does, however, occur by hydrolysis or by transglycosylation to acceptor ligands,with half-lives ranging from one to 500 hours (Street et at., 1992). Transglycosylationinvolves binding of the acceptor ligand to the aglycone site of the inactivated enzyme andreaction of one of its hydroxyl groups with the trapped intermediate, a process which isessentially the microscopic reverse of the glycosylation step with the natural substrate.The production of animal models of glycosidase deficiency diseases is but onepotential application of specific inhibitors of mammalian glycosidases. With thedevelopment of positron emission tomography (PET), the non-invasive study ofmetabolism and physiological processes involved in health and disease in the living humanbody has become feasible. The technique is based on the labelling of biologically activemolecules with a short-lived positron-emitting isotope, and the quantitation of regionaltissue concentrations of the administered radiopharmaceutical by computed tomography.PET has been extensively applied in the investigation of regional glucose metabolism in106heart and brain (e.g. Reivich et a!., 1979; Strauss, 1989), and in receptor- and enzyme-binding studies by non-covalent and covalent ligands (e.g. Wagner, 1986). The potentialfor labelling 2-deoxy-2-fluoro glycosyl fluorides with ‘8F (t112 110 mm) suggests theapplication of 2-deoxy-2-[18F]fluoro glycosyl fluorides to the diagnostic imaging of (3-glycosidase activity in vivo using positron emission tomography (PET). Irreversiblecovalent inhibitors of enzymes of the dopaminergic system (e.g. inhibition of monoamineoxidase with deprenyl) have been similarly used for the in vivo imaging of enzyme activityfor diagnosis and treatment of several psychiatric disorders (MacGregor et a!., 1985).2-Deoxy-2-fluoro glycosyl fluorides are promising candidates for the specificinactivation of retaining mammalian (3-glycosidases in vivo, especially since they are closestructural mimics of the parent sugars, and thus will likely be transported across cellmembranes. They may therefore be useful in producing animal models of glycosidasedeficiency diseases. As potential imaging agents, a key advantage of this class of inhibitorsis the relatively rapid reactivation of the 2-deoxy-2-fluoro glycosyl-enzyme intermediate viahydrolysis or transglycosylation, thus facilitating prompt clearance after imaging andminimizing adverse effects due to irreversible enzyme inactivation. No such reactivationwould occur with the cyclitol-based inhibitors. Such selective inhibition of mammalianretaining glycosidases may aid in defining the roles of specific glycohydrolases involved inglycolipid and glycoprotein processing or catabolism.4.2 Specific AimsSpecific, mechanism-based inhibitors will be used as structural and mechanisticprobes of mammalian glycosidases, with an emphasis on those lysosomal enzymesinvolved in genetic disease. Several approaches will be taken:The appropriate 2-deoxy-2-fluoro glycosyl fluorides will be employed in an attemptto inactivate purified human lysosomal (3-glucosidase and human lysosomal (3-galactosidase in vitro by accumulation of stable intermediates. The derivatized enzymes107will be proteolytically digested, and the labelled peptides identified and characterized by avariety of mass spectrometric techniques. Thus the residue which functions as the catalyticnucleophile in the normal catalytic mechanism of each enzyme will be identified.These inhibitors will be administered to rats, and glycosidase activities of varioustissues will be assayed to determine the degree and specificity of glycosidase inhibition invivo. Reactivation of the inactivated enzyme activity will be examined, and compared tothe in vitro results obtained above.Inhibitors will be labelled with the posiiron-emitting isotope 18F, and used to labelglycosidases in vitro. Radiolabelled inhibitors will then be administered to rats to evaluatethe biodistribution and clearance of the labelled compound in vivo. Specific uptake of theseinhibitors will be assessed by ligand blocking experiments. Finally, imaging of theadministered inhibitor in rats using PET will be attempted to demonstrate the feasibility ofthis approach to imaging glycosidase activity in vivo.Results and Discussion4.3 Identification of the Catalytic Nucleophiles of Human LysosomalGlycosidases by 2-Deoxy-2-Fluoro-Glycosyl Fluorides and MassSpectrometry4.3.1 The catalytic nucleophile of human glucocerebrosidasea) Background and signficance. Human glucocerebrosidase (acid 3-glucosidase,N-acyl-sphingosyl-l-O-3-D-glucoside: glucohydrolase, GCase) cleaves the 3-glucosidiclinkage of glucosylceramide (Figure 4.1). It is a membrane-associated glycoprotein (67kDa, 497 amino acids) with both high mannose and complex oligosaccharides (Takasaki eta!., 1984; Grabowski, 1990), whose crucial role in glycolipid catabolism has beenestablished by disruption of the GCase gene in mice, such mice dying within hours of birth(Tybulewicz et a!., 1992). Various point mutations of the human enzyme result inphenotypically diverse Gaucher disease variants. This disease has served as a prototype108for the development of enzyme replacement therapies using specifically oligosaccharidemodified enzyme for targetting to the major sites of pathologic involvement (Barton et at.,1991; Barton, 1990; Fallet et a!., 1992; Figueroa et aL, 1992).OHFigure 4.1. Structure ofgluco.syl ceramide.Even though a great variety of mutations in the structural gene coding for GCasehas been associated with the many phenotypic variants of Gaucher disease, few residues inthe enzyme have been suggested as crucial to catalysis (Grace et at., 1994). Labelling ofAsp-443 with[3H]-labelled bromoconduritol B epoxide (Br-CBE), a putative mechanism-based inactivator, had previously led to the suggetion of this residue being the catalyticnucleophile in GCase (Dinur et at., 1986). However, recent kinetic analyses of GCasemutants produced by site-directed mutagenesis of Asp-443 or Asp-445, including that inwhich Asp-443 was converted to glycine, revealed activities similar to or greater than wildtype (Grace et at., 1994). These observations are incompatible with Asp-443 being thecatalytic nucleophile since replacement of a nucleophiic carboxylate side chain by hydrogenwould have a profoundly deleterious effect upon the reaction rate. Indeed, kinetic analysesof mutants of the mechanistically similar Agrobacteriwn faecalis 3-glucosidase mutated atits catalytic nucleophile residue Glu-358 have shown enormous (> 106-fold) reduction incatalytic ‘activity for all substitutions except aspartic acid, which reduced the activity 10-fold (Withers, 1992). An alternative approach to the identification of this nucleophile wastherefore required.109GCase has been assumed previously (Dinur et al., 1986; Legler, 1990) tohydrolyse the glycosidic linkage of glucosyl ceramides with overall retention of theanomeric configuration. 2-Deoxy-2-fluoro glycosides with good leaving groups aremechanism-based covalent inactivators for glycosidases of this type. Therefore, theinactivation of human glucocerebrosidase using 2-deoxy-2-fluoro--D-glucosyl fluorideand the subsequent identification of the labelled residue by means of a novel combination ofmass spectromeiric techniques which does not require the use of radiolabels will beattempted (see also Chapter 2). Such a non-radioisotopic approach may have valuableapplication to the identification of active site residues in other enzymes.b) Inactivation of GCase. Inactivation of GCase by 2-deoxy-2-fluoro-3-D-glucosyl fluoride occurred in a rapid, time-dependent manner as shown by the activity vs.time plots in Figure 4.2A. The inactivation followed the expected pseudo-first orderkinetics, allowing pseudo-first order rate constants at each inactivator concentration to becalculated. No saturation was observed at the inactivator concentrations studied, andhigher concentrations could not be investigated due to the rapidity of inactivation whichprecluded accurate sampling. Reliable values for the inactivation rate constant (k1) or thereversible dissociation constant (K1) could not, therefore, be determined as can be seenfrom the replot in Figure 4.2B. However, a reliable second order rate constant of kfK1 =0.023 ± 0.00 1 min4lvl was calculated from the slope of this plot. Poor binding of2FGluF is unsurprising given that the enzyme undoubtedly ordinarily develops many ofits key binding interactions with the ceramide moiety (Figure 4.1). Indeed, a K1 value of75 mM for 2-deoxy-2-fluoro-glucose has been reported previously (Osiecki-Newman eta!., 1988). Inclusion of castanospermine (8.3 .tM), a known (Osiecki-Newman et al.,1988) competitive inhibitor of GCase (K1 = 7 .tM) in an inactivation mixture containing10.4 mlvi 2FGluF reduced k0bs, the pseudo-first order inactivation rate constant, from0.23 min1 to 0.15 min1. This is the expected degree of protection against inactivation if110the two reagents bind at the same site, thereby indicating that the inactivation is active-sitedirected. Additional evidence that this inactivation is due to stabilization and trapping of thenormal intermediate in catalysis was obtained by demonstration of its catalytic competence.Following removal of excess inactivator from the labelled enzyme, the sample wasincubated at 37°C and the return of enzymatic activity was monitored. Return of activitywas a first order process, with a spontaneous reactivation rate constant, k3 of 5.3 x 10 ±0.4 x i04 mm-1, corresponding to a half-life of t112 = 1300 minutes (Fig. 4.4). Themanno epimer, 2-deoxy-2-fluoro-3-D-mannosyl fluoride, was also an inactivator of theenzyme with kfK1 = 0.0019 ± 0.0001 min4M (Figure 4.3A and B). As might beexpected, the inactivaton was slower with this species, reflecting less facile accomodationof an axial fluorine by the enzyme at a position which normally binds a hydroxyl of thegluco configuration. Reactivation of the intermediate was also less rapid, with k3 = 2.8 x-4 -4.-I10 ± 0.4 x 10 mm corresponding to a t112 = 2500 mm (Figure 4.5).c) Identification of the catalytic nucleophile of GCase. Identification of the site ofattachment of the 2-fluorosugar, thus of the nucleophilic amino acid residue, was achievedby using a combination of mass spectrometric techniques to first identify the labelledpeptide in a peptic hydrolysate, and then to determine the sequence of this peptide. Peptichydrolysis of 2FGlu-labelled GCase resulted in a mixture of peptides which was separatedby reverse phase-HPLC using the ESMS as detector. When the spectrometer was scannedin the normal LC/MS mode, the total ion chromatogram (TIC) of the 2FG1u-labelled GCasedigest displayed a large number of peaks, which arise from every peptide in the mixture(Fig. 4.6A). The peptide bearing the 2-fluoroglucosyl label was then identified in a secondrun by using the tandem mass spectrometer in the neutral loss mode. When thespectrometer was scanned in the neutral loss tandem MS/MS mode searching for the mass111A00 2 4 6 8 10 12 14 16 18 20[2FI3G1uF] (mM)Figure 42. Inactivation ofGCase by 2-deoxy-2-fluoro-J3-D-glucosyl fluoride: (A) Semi-logarithmic plot of residual activity versus rime at the indicated inactivator concentrations:0, 2.07 mM; 13, 5.18 mM,•, 10.4 mM; 0, 15.5 mM; A, 18.6 mM. (B) Re-plot offirst-order rate constants from (A).00-0.4-0.8-1.2- 8 12 16 20BTime (mm)112A 0-0.2-0.4-0.6e— -0.8—1-1.2-1.4100B0. 40[2FManF] (mM)Figure 4.3. Inactivation of GCase by 2-deoxy-2-fluoro-f3-D-mannosyl fluoride: (A)Semi-logarithmic plot of residual activity versus time at the indicated inactivatorconcentrations: A, 3.91 mM; M, 9.78 mM; I 19.6 mM; •, 293 mM; 0, 35.2 mM. (B)Re-plot offirst-order rate constants from (A).0 20 40 60 80Time (h)10 20 30113CICI:0 20 40 60Time (h) 4.4. Reactivation of2-deoxy-2-fluoro-glucosyl-GCase: See text and Experimentalchapterfor details. 4.5. Reactivation of2-deoxy-2-fluoro-mannosyl-GCase.0 20 40 60Time (mm)80114loss m/z 165, corresponding to the loss of the 2FG1u label from the labelled active-sitepeptide in the singly charged state, a greatly simplified chromatogram was obtained (Figure4.6B). Two major peptides (peptides 1 and 2 in Fig. 4.6B) are seen, plus a number ofminor, later eluting species. However, when a sample of non-labelled GCase wassubjected to peptic digestion and subsequent ESMS/MS neutral loss mode analysis, thechromatogram shown in Figure 4.6C was obtained. Most of the features of Figure 4.6Bare seen here also, with the exception of peptide 1, indicating that peptide 1 is the species ofinterest and that the other signals arise from non-labelled peptides which undergo anequivalent fragmentation. This most likely involves elimination of a phenylalanine residue(165 Da) from several different peptides. Therefore peptide 1, which was only detected inthe digest of the inhibited enzyme, is presumably the covalently modified active-sitepeptide. This singly-charged peptide was measured at mlz 688 ± 1 (Fig. 4.6D), thus sincethe 2FG1u moiety has a mass of 165, the molecular weight of the unlabelled active-sitepeptide must be 523 ± 1.Candidate peptides were then identified by inspection of the amino acid sequence ofthe enzyme (Sorge et al., 1985) and searching for all possible peptides with this mass.Only four peptides with a mass 523 ± 1 could be identified, namely LGTFS (residues 34-38), GIGYN (113-117), DDFQ (140-143) and FASEA (337-341), of which only the lattertwo peptides contain the acidic amino acid residues which would be expected for thecatalytic nucleophile of a glycosidase (Sinnott, 1990).Glu-340 was identified as the 2FG1u binding site, thus the catalytic nucleophile, byindependent methods. Firstly, the amino acid sequence of the labelled peptide (mlz 688)was determined, without a need for further purification, by collision induced fragmentationof the peptide of interest and analysis of the daughter ions. The parent ion of mlz 688 wasselected in the first quadrupole, subjected to collision induced fragmentation at a higherenergy than used in the neutral loss mode, then the masses of the daughter ions producedwere detected in the third quadrupole. The family of daughter ions produced is shown in115100 A. 750Ca,5025i0075Ca)0100.755o25a)0.100 D075000 600 800 1000 1200m/zFigure 4.6. ESMS experiments on GCase proteolytic digests: (A) labelled with 2FG1u,TIC in normal MS mode, (B) labelled with 2FG1u, TIC in neutral loss mode, and (C)unlabelled, TIC in neutral loss mode, (D) mass spectrum ofpeptide. 1 in Fig.4.6B.C10 12 1468818 mm.116Figure 4.7. Peaks at m/z 688 and 523 confirm that the m/z 523 peptide was derived bycollision-induced fragmentation of the m/z 688 peptide. The peak at m/z 435 is consistentwith loss of a C-terminal alanine (88 Da) from the unlabelled peptide, while those at m/z306 and mlz 219 correspond to subsequent losses of glutamyl and seiyl species,respectively, leaving a fragment (m/z 219) consistent with a phenylalanylalanyl species.The sequence FASEA is therefore predicted. The C-terminal fragments are not observed inthis mode since the loss of the charged N-terminal amino acid produces neutral peptideswhich are not detected.I1007550250Figure 4.7. Tandem MS/MS daughter ion spectrum of the 2FG1u-labelled peptide (mlz688, FASEA +2FG1u).Secondly, chemical pulsed-liquid-phase sequence analysis of the purified specieswith mlz 688 yielded the sequence FASXA, where X is an unidentified residue. Thefourth residue could not be identified since the putative phenylthiohydantoin-Glu-2FGIuspecies is resistant to extraction from the sequencer under the conditions used (Gebler,1992; Withers, 1990).688FASEA+2FGIcFASFA 306219FASEAFASE435300m/z500 600 700117Table 4.1. Edinan degradation ofactive site peptide in (iCaseCycle PTH Derivative Yield (pmol) eDNA Sequencea1 F 34 F2 A 30 A3 S 6 S4 X - E5 A 13 Aa Sequence of peptide derived from the cDNA starting at residue 337X, unidentified residueThirdly, site-directed mutagenesis of Glu-340 to glycine and expression in Sf9 cellscarried out by Dr. Marie Grace at the Mt. Sinai Medical Centre produced a stable protein,but one with greatly reduced activity. The correct folding of this mutant was demonstratedby the finding that the mutant protein bound to a deoxynojirimycin-based affmity columnwith a similar affinity to the wild type. The specific activity of the E340G mutant was l0-fold reduced compared to the expressed normal acid 3-glucosidase. This large reduction inspecific activity in the mutant enzyme is consistent with the key role of this residue incatalysis. Further, even the small amount of residual activity observed is likely due to thecontamination with wild type arising from translational misreading. Equivalent levels ofmisincorporation have been observed previously (Schimmel, 1990). Indeed the activityobserved is completely inactivatable by 2FfGluF at rates comparable to those of wild typeenzyme, a result which is consistent with a mutant in which the attachment site has beenremoved.d) Conclusions. These findings establish Glu-340 as the catalytic nucleophile inthe active site of human acid f-glucosidase. The previous assignment of Asp-443 as thecatalytic nucleophile was based upon labelling of the enzyme with CBE and derivatives.118However, such conduritol derivatives lack the hydroxymethylene group present at the C-5position of glucose, and thus have reduced specificity. As a consequence, CBE bindingand enzyme inactivation probably occurred via an alternate binding mode, as has beenobserved in other enzymes (Gebler et a!., 1992). The critical importance of the catalyticnucleophile and the surrounding structure for enzyme activity is reflected in the fact thatGlu-340 is located in the sequence within the longest stretch of 100% amino acid identity(3 15-375) between the human and murine enzymes (O’Neill et a!., 1989). Furthermore,several of the mutations responsible for the variants of Gaucher disease have been locatedto this region (Takasaki, 1984).4.3.2 The catalytic nucleophile of human lysosomal acid 3-galactosidaseprecursora) Background and significance. Lysosomal acid f3-galactosidase (EC isan essential catabolic enzyme which catalyzes the hydrolysis of terminal non-reducing 3-galactosyl residues from a variety of substrates, including ganglioside GM1lactosylceramide, lactose, and some galactose-containing oligosaccharides. The inheriteddeficiency of the enzyme in humans results in GM1 gangliosidosis, a severe neurologicaldisease, or Morquio Syndrome B, primarily a skeletal disorder. The enzyme is synthesizedas an 88 kDa glycoprotein precursor (677 amino acids), then processed to the 64 kDamature enzyme, a process thought to involve proteolytic cleavage of the C-terminal portionof the protein (lYAgrosa et a!., 1992; Zhang, S. et al., 1994). The precursor is kineticallyand functionally identical to the mature form of the enzyme, with pH optimum and kineticparameters identical to the mature enzyme, and is taken up and localized to the lysosomesof GM1 gangliosidosis fibroblasts, correcting the enzymic deficiency (Zhang, S. et a!.,1994).Though a number of mutations in the enzyme resulting in loss of in vitro activityhave been associated with different disease phenotypes, no residues have been establishedas essential to the catalytic mechanism by either labelling or mutagenesis. The inactivation119of the human f-galactosidase precursor with an analogous mechanism-based inactivator ofgalacto configuration, 2,4-dinitrophenyl 2-deoxy-2-fluoro--D-galactopyranoside(2FGalDNP), and identification of the catalytic nucleophile through the application andextension of the mass spectrometric techniques pioneered with GCase (Miao et al., 1994)was therefore attempted.b) Determination of the stereochemical course of hydrolysis of acid 13-galactosidase. NMR spectroscopy was used to determine whether the glycosidic bond of asubstrate was cleaved by the enzyme with retention or inversion of configuration as shownin Figure 4.8. Spectrum A (Figure 4.8) shows the anomeric proton region of the ‘H NMRspectrum of 2,4-dinitrophenyl f-D-galactopyranoside inD20. The doublet at 5.38 ppm(J1,2 7.8 Hz) is due to the axial anomeric proton of the -galactoside substrate, and thelarge resonance at 4.75 ppm is due to HOD. Spectra B and C were recorded at intervalsafter the addition of enzyme and clearly reveal first the appearance of a resonance at 4.59ppm (J,2 = 7.9 Hz) due to the axial anomeric proton of -galactose, followed by a newresonance at 5.27 ppm (J,2 = 3.2 Hz) from CL-galactose formed by mutarotation of the -anomer. After 5 h (Spectrum D), the substrate has been consumed and establishment of theequilibrium pyranose anomeric ratio (-- 2:1 p/ct) indicates that the enzyme catalyzeshydrolysis with net retention of the substrate anomeric configuration.c) Inactivation of acid f3-galactosidase. Inactivation of -galactosithse by2FGalDNP occurred in a rapid, time-dependent manner. Kinetic parameters forinactivation were determined by measuring inactivation rates at several different inhibitorconcentrations (Figure 4.9), and value of k1 = 2.2 ± 0.2 min’ and K1 = 0.17 ± 0.03 mMthus determined. Protection against inactivation was afforded by isopropyl f-Dthiogalactoside (5 mM), reducing the k0bS at 0.079 mM 2F3GalDNP from 0.67 to 0.38mm1.12055 MINUTESFigure 4.8. 1H NMR spectra showing the stereochemical course of hydrolysis of 2,4-dinitrophenyl galactopyranoside by human acid /3-galactosidase. The anomeric protonregion of the substrate andproducts before addition of enzyme (A), and at 25, 55 and 300mm cjter addition (B, C, and D, respectively). For details, see text and Experimentalchapter.25 M[NUESBpC5.5 5.0 4.5ppmTIME ZERO5.5 4.5ppmA5 HOURSD5.5 5.0 4.5ppm5.5 4.5ppm121A 2 4 6 8 10 12 14Time (mm)j’ 0.4[2F(Ga1DNP] (mM)Figure 4.9. Inactivation of acid f3-galactosidase with 2F/3Ga1DNP. A) Enzyme wasasssayed with 2,4-DNPGa1 following incubation with 0.011 mM (A), 0.025 mM (4),0.049 mM (0), 0.079 mM (11), 0.16 mM (•), and 0.32 mM (0) 2FJ3Ga1DNP at the timesindicated. B) Re-plot of rate constants from (A).0.1 0.2 0.3122d) Identification of the catalytic nucleophile of acid f3-galactosidase by massspectrometry. Peptic digestion of the 2FGa1-labelled enzyme (after treatment withirichioroacetic acid to facilitate proteolytic digestion) resulted in a mixture of peptides whichwas separated by HPLC using the mass spectrometer as detector. When scanned inLC/MS mode, the TIC displays a complex mixture of peaks arising from every peptide inthe digest (Figure 4. bA). The labelled peptides were identified in a second run usingtandem mass specirometry in neutral loss mode, by selective fragmentation of the inhibitorpeptide ester bond. As before, the two quadrupoles are scanned in a linked mode so thatonly those ions differing by the mass of the label could be detected. For a singly-chargedpeptide, this m/z difference is the mass of the label (165 Da); for a doubly-charged peptide,the mlz difference is one-half of the mass of the label, and so on.Scanning in neutral loss mode for the mass loss m/z 165 from a singly-chargedpeptide revealed no significant peaks. However, scanning for mass loss m/z 82.5 revealeda peak at 18 mm (Figure 4.1OB) that was absent in an unlabelled, control digest (Figure4.1OC). This peak, due to three doubly-charged peptides measured at 950.5, 1058.5, and1114.5 ± 1 (Figure 4.1OD) corresponded to unlabelled peptides of masses 1735, 1951 and2063 ± 2 (e.g. [950.5 x 2] - 2 -165 + 1). Therefore, the singly-charged peptides hadmasses of 1736, 1952 and 2064 ± 2. These peptides, presumably containing overlappingsequences which include the catalytic nucleophile, are selectively detected because each hasundergone loss of the inhibitor mass. The significant background observed in the controlis again likely due to losses of phenylalanine residues from several different peptides.Inspection of the protein sequence revealed 26, 35 and 31 candidate peptides of masses1736, 1952 and 2064 ± 2, respectively. However, only seven different glutamic oraspartic acids, residues which are known to act as catalytic nucleophiles in glycosidases,were present in candidate peptides of all three masses. Increasing the collision gas energyand scanning in daughter ion mode in an attempt to cause further amide bond fragmentationand identify the labelled peptides was unsuccessful. This resulted only in loss of the123Time (nun)1002Time (mm)100•Tmn (mmn)950.5100 D75 1058.5501114.5250••. . I500 1000 1500 2000nh/zFigure 4.10. ESMS Experiments on acid J3-galactosidase peptic digest. A, labelled with2FGa1, TIC in normal MS mode; B, labelled with 2FGa1, TIC in neutral loss mode; C,unlabelled, TIC in neutral loss mode; D, mass spectrwn of labelledpeptide at 18 mm in B.124inhibitor with no significant peptide backbone fragmentation, presumably because therelatively large size of these peptides precludes their fragmentation in the collision cell.The mixture of peptic peptides was therefore treated with a second protease, in anattempt to cleave these large peptides into smaller species which would undergofragmentation by tandem mass spectrometry. Solid ammonium bicarbonate was added to apeptic digest of the labelled enzyme prepared as above to increase the pH to — 8 and themixture was treated with irypsin for fifteen minutes. The mixture was then re-acidifiedwith trifluoroacetic acid to pH — 2 since the inhibitor-peptide ester bond is much moresusceptible to hydrolysis above pH 7 than at acidic pH (Bause & Legler, 1974). Acomplex mixture of peptides was once again observed upon scanning in the LCIMS mode(Figure 4.11A). In neutral loss mode, a search for a mass loss of m/z 165 revealed a singlepeak at 19.5 mm (Figure 4.1 1B) which was absent in a control, identically-treated, butunlabelled digest (Figure 4.11C). This singly-charged labelled peptide had a mass of1040.5 ± 1 (Figure 4.1 1D), corresponding to a singly-charged peptide of mass 876.5 ± 1(1040.5 - 165 + 1). Thirteen candidate peptides had a mass of 876 ± 1, but only oneglutamic or aspartic acid was present in candidate peptides of mass 876 ± 1, and of masses1736, 1952 and 2064 ± 2. This residue is Glu-268, located within a GPLINSEF (262-269) sequence of mass 876 ± 1, a EPKGPLINSEFYTGW (259-273) sequence of mass1736 ± 2, a CEPKGPLINSEFYTGWL (258-274) sequence of mass 1952 ± 2, and aKGPLINSEFYTGWLDHW (261-277), or a PLINSEFYTGWLDHWGQ (263-279), or aLINSEFYTGWLDHWGQP (264-280) sequence of mass 2064±2. Since the GPLINSEFpeptide is the sole peptide detected in the neutral loss scan of the peptic/tryptic digest, thelatter two candidate peptides of mass 2064 ± 2 may be excluded because they do notcontain the N-terminal glycyl and prolyl residues present in the GPLINSEF peptide, whichmust be derived from tryptic digestion of the larger peptic peptides.Daughter ion scans of peak 1040.5 at increased collision gas energy resulted in lossof the label, giving rise to the peak at 876.5 (Figure 4.12). Further peptide bond125100A755025012 14 16Tine (nm)18 20 22 241007550BCTune (mm)Time (mm)D104012 14 16 18 20250100750100. 7550250mJzFigure 4.11. ESMS Experiments on acid J3-galactosidase peptic/tryptic digest. A, labelledwith 2FGa1, TIC in normal MS mode; B, labelled with 2FGa1, TIC in neutral loss mode;C, unlabelled, TIC in neutral loss mode; D, mass spectrum of labelled peptide at 195 mmmB.22 24500 1Ô00 1500 2000126fragmentation was also observed, with peaks at 711, 581, 496, 381, and 268,corresponding to C-terminal losses of phenylalanine, and of EF, SEF, NSEF, and INSEFspecies from the parent peptide, confirming that the labelled peptide does indeed have thesequence GPLINSEF. A number of ions in the corresponding a series were alsoobserved.100 1040GPL GPLINSEF+ 2FGa1‘-‘75GPLINSEF50 a4 GPLI876.5GPLJNa3GPLINS25 352.5 381 GPLINSE240I 5810I L I 710.50 - ailAn I Iø’ y’ L4IdIA ii iIiiIiiI. iinii i A_____________j....200 400 600 800 1000m/zFigure 4.12. Tandem MS/MS daughter ion spectrum of the 2FGa1-labelled peptide, m/z1040, GPLINSEF + 2FGa1.Further confirmation of the identity of the labelled peptide was obtained by highresolution liquid secondary ionization mass spectrometry (LS1MS), in which the accuratemasses of the 1040.5 and 876.5 peaks were measured at 1040.49500 and 876.44696,respectively. No possible peptides, regardless of protease specificity, may be generatedfrom the native protein that have masses within 5 ppm of the accurate 1040.49500 mass,indicating that this peak can only be due to a derivatized peptide. Indeed, a 2-deoxy-2-fluoro galactosyl-labelled GPLINSEF species (calculated mass 1040.495 19, 1m = 0.18ppm) is consistent with this mass. Only two possible peptides have masses differing byless than 5 ppm from the accurate 876.44696 mass: GPLINSEF, with a calculated mass of127876.44672 (z\m = 0.28 ppm), and RNATQRM (25-31), with a calculated mass of876.44740 (m = 0.50 ppm). However, the latter candidate peptide contains no glutamicor aspartic acids, is inconsistent with the neutral loss and MS/MS results above, and maytherefore be excluded.The attachment site of the 2-fluoro galactosyl moiety, thus the nucleophile, wasconfirmed by aniinolysis of the labelled peptide. A labelled digest treated with 30%ammonium hydroxide for fifteen minutes at 50°C was examined by mass spectrometry andcompared to an untreated, labelled control digest. In the aminolyzed sample, the peak ofm/z 1040.5 and retention time 19.5 mm present in the untreated digest (Figure 4.13A) wasabsent (Figure 4.13B) and was replaced by a new peak, absent in the untreated digest(Figure 4.13C), of mass 875.5 and retention time 19.5 mm (Figure 4.13D), consistentwith aminolysis of a glycosyl ester to glutamine, giving a GPLINSQF peptide one massunit less than the parent, unlabelled peptide (Figure 4.1 3E). A peak of mass 876.5 andretention time 15 mm present in both untreated and treated samples (Figures 4.13C and4. 13D) was also considerably increased in the aminolyzed digest, and is presumably theunlabelled peptide resulting from concomitant hydrolysis of the 2-fluoro glycosyl esterunder the aminolysis conditions.e) Conclusions. GIu-268 and much of the surrounding sequence is conservedbetween the human and mouse lysosomal f3-galactosidases, both of which belong toFamily 35 of glycosyl hydrolases (Figure 4.14). Several other f-galactosidases fromMacropus, asparagus, apple, carnation, and Aspergillus niger show significant homology.In each case, the corresponding glutamic acid is absolutely conserved, as would beexpected of the essential catalytic nucleophile. Four other carboxylic acids, Glu- 186, Glu188, Asp-332, and Glu-339 (human enzyme numbering), are also absolutely conserved inthis family of proteins and may be important to catalysis. Presumably, at least one of theseresidues is involved in acid/base catalysis. Site-directed mutagenesis of these residues,128100 1040/1041ih .. .i.I, II.flI. .1. I . i.Figure 4.13. Selected ion chromatograms of acid J3-galactosidase peptic/tryptic digest innor,nal mode: A, labelled with 2FGa1, peaks with m/z 1040-1041, B, labelled, but treatedwith NH3,peaks with ni/z 1040-1041; C, labelled, peaks with m/z 875-876 , D, labelled,but treated with NH3,peaks with m/z 875-876,,• E, mass spectrwn of 19.5 mm peak in D.A. . 112 14 16 18 20 22Time (mm)1040/1041B1112 14 16 18 20 22Tin (mm)75502501007550250100755025010075502501007550250875/876C. .20 22Time (mm)875/876DJI’12 14 16Time (mm)E18 20 22875.5500 1000 1500 2000m/z129human 176 NGGPVITVQV ENEXSYFAC DFDYLRFLmouse 177 NGGPVITVQV ENEXSYFAC DYDYLRFLAsparagus officianalis 171 QGGPISQ I ENEXPVEYY DGAAGKSXMalusdomestica 170 QGGPIXLLSQI ENEFPVEWE IGAPGKAXDianthus SR12 protein 175 QGGPIIJ.LNQI ENEXPVEWE IGAPGKAXAspergillusniger 188 NGGPIZILYQP ENEXTSGCCG VEFPDPVXhuman 259 EPKGPLINSE EYIGLWIG QPHSmouse 260 EPKGPLINSE .YIGLHLG KPHSMacropus eugeniil 1 SE E.YIGWLHWG EAHQAsparagus officianalis 243 KDNKPKMWTE AWIGLFTGFG GAVPMalusdomestica 242 KDYKPKMWTE VW.IGWYTEFG GAVPDianthus SR12 protein 248 DKSKPKMWTE NW.IGWYTEYG KPVPAspergillus niger 289 SPTTPYAIVE EQGGSYPG GPGFhuman 327 QPTSYDY LSEADLIEK YFALRmouse 328 QPTSYDY LSEADLIKK YFALRMacropuseugeniit 60 QPTSYDY LSEADLIEK YFALRAsparagus officianalis 308 I STSYDY IDEYLJ.LRQP WGHLMalus domestica 307 MATSYDY LDEYLPREP KWGHLDianthusSRl2protein 312 VSTSYDY&. DEYLPREP YTHAspergillusniger 358 GYTSYDYGSA VTESRNIIRE YSELLFigure 4.14. Conserved residues among 13-galactosidases of Family 35 glycosylhydrolases and relatedproteins. The putative catalytic nucleophile region is boxed, and theconserved glutamate labelled by 2FJ3Ga1DNP is indicated by a ‘•‘. Residues in bold areabsolutely conserved in all aligned proteins. Underlined residues are conserved in themajority of aligned proteins. The sequences shown are: sp P162 78 human acid j3-galactosidase precursor, gp M75122 mouse acid /3-galactosidase, gp L15561 Macropuseugenii f3-galactosidase, gp X77319 Asparagus officianalis J3-galactosidase, gp L29451Malus domestica (Granny Smith) J3-galactosidase-related protein, sp Q00662 Dianthuscaryophyllus 5R12 protein, and gp A00968 Aspergillus niger /3-galactosidase. Accessionumbers correspond to those in the SWISS-PROT or GenPept databases.I Partial cDNA sequenceincluding that of the putative catalytic nucleophile Glu-268, is being performed currently byDr. John Callahan at Sick Children’s Hospital in Toronto in an ongoing collaboration.Deficiencies in this enzyme resulting in GM1 gangliosidosis or Morquio SyndromeB are characterized by a number of known genotypes, including the following pointmutations located predominantly in the C-terminal portion of the enzyme: R49C, 15 iT,G123R, R2O1C, W273L, Y316C, R457Q, R482H, and W509C (Mosna et al., 1992;130Nishinioto et al., 1991; Oshima et at., 1991; Yoshida et al., 1991). It has been assumedthat these mutations are responsible for defects in protein folding or trafficking leading torapid degradation of the mature protein. However, some enzymic deficits may be due tomore direct effects of certain mutations on binding interactions of the catalytic residues orother residues in the active site. The deleterious effect of the W273L mutation (8% ofnormal activity (Oshima et a!., 1991)), in particular, is possibly due to altered secondaiystructure within the enzyme’s active site in close proximity to the catalytic nucleophile.Other mutant residues, though perhaps distant in primary sequence, may be in close spatialproximity to the active site and may have similar effects. The identification of Glu-268 asthe catalytic nucleophile should facilitate a greater understanding of the molecular basis ofGM1 gangliosidosis and Morquio B disease.4.4 Specific In Vivo Inhibition of f3-Glucosidase and f-Mannosidase, butNot 3-GaIactosidase Activity in Rats by 2-Deoxy-2-Fluoro-3-Glucosyl and .Mannosyl Fluorides and Recovery of Activity InVivo and In VitroBackground and significance. Glucocerebrosidase (GCase), the lysosomal enzymewhose nucleophile is identified above, is not the only mammalian 3-glucosidase. Another13-glucosidase (E.C. with more neutral pH optimum, different sub-cellularlocalization and broader substrate specificity (exhibiting 3-D-galactosidase, (x-Larabinosidase, 3-D-xy1osidase and f3-D-fucosidase activities) has been identified andcharacterized (Glew et a!., 1993; Glew et al., 1976; Peters et at., 1976). Furthermore,virtually every conceivable type of glycosidic linkage is present in mammalian tissues. Thehydrolysis of these linkages in the catabolism of glycoproteins and glycolipids in thelysosomal compartment is catalyzed by specific glycosidases. For example, 3-mannosidicbonds occur most commonly within the ubiquitous Man-(131,4)-GluNAc-([31,4)-G1uNAcmoiety that comprises the core oligosaccharide structure of N-linked glycoproteins.Deficiency of the f3-mannosidase (E.C.3.2. 1.25) that cleaves the terminal mannose residue131of this lrisaccharide, which like -glucocerebrosidase is localized in the acidic environmentof the lysosome, leads to the accumulation of mannose-terminated di- or trisaccharides andis the basis of -mannosidosis in goats (Jones & Dawson, 1981), cattle (Abbit et al., 1991)and humans (Wenger et al., 1986). As with -glucocerebrosithse, this acid -mannosidase activity is complemented by that of a neutral, non-lysosomal enzyme(Dawson, 1982). While the lysosomal acid glycosidases play an essential catabolic role,the functions of the neutral cytosolic enzymes are unclear.The inhibition of GCase by the appropriate 2-deoxy-2-fluoro-3-D-glycosyl fluoridehas been demonstrated in vitro and the catalytic nucleophile of this enzyme was identifiedby derivatization of the labelled residue (vide supra). Similar inhibition of 3-glucosidaseand -mannosidase activity in vivo by 2-deoxy-2-fluoro-[3-glucosyl and -mannosylfluorides in rats was therefore attempted. Specific and mechanism-based inhibition wouldfacilitate the production of animal models of the appropriate lysosomal storage diseases andsuggests the application of such inhibitors, labelled with 18F, to the diagnostic imaging ofglycosidase activity in vivo using PET. A key feature of such inhibitors is reactivation ofthe labelled enzyme by hydrolysis or transglycosylation either in vivo or in vitro, whichwould be expected to occur independently of protein synthesis and on a time-scale similarto that of the purifed enzymes.a) Inhibition of J3-Glucosidase Activity by 2FJ3G1uF and In Vivo Recovery.Figure 4.15 shows the inhibition and in vivo recovery of -glucosidase activity in brain,kidney, liver and spleen after administration of a single dose of 2FGluF (10 mg/kg).Inhibition of f3-glucosidase activity after one hour was most pronounced in the spleen (20%of control activity), while that in the kidney was the least (40% of control activity). fGlucosidase activity steadily increased and fully recovered in brain, liver and spleen after48 hours in vivo while that in the kidney, after a slight further decrease in activity at 20hours, had regained 60% of its control level after 48 hours.132b) Inhibition of J3-Mannosid.ase Activity by 2F/3ManF and In Vivo Recovery.Figure 4.16 shows the inhibition and in vivo recovery of -mannosidase activity in brain,kidney, liver and spleen after administration of a single dose of 2FManF (10 mg/kg). 3-Mannosidase activity was not as effectively inhibited by 2F3ManF as -g1ucosidaseactivity by 2F3GluF, f3-mannosidase levels at one hour after injection ranging from 35 to45% of control levels in all tissues. In vivo recovery of [3-mannosidase activity was slowerthan 3-glucosidase, but 60 to 75% of control activity in all tissues was regained by 48hours. Like the l3-glucosidase activity in the 2Ff3GIuF-treated animals, the slowest rate ofrecovery of 3-mannosidase activity was observed in the kidney, which also suffered afurther slight decrease in activity at 20 hours, before recovering to 60% of its control levelafter 48 hours.c) Specificity ofGlycosidase inhibition by 2-Deoxy-2-Fluoro GlycosylFluorides. Figure 4.17 shows the effect of 2FI3G1uF and 2F3ManF (both 10 mg/kg) on[3-galactosidase, 3-g1ucosidase and 3-mannosidase activities in the kidney (4.17a) and thebrain (4. 17b) one hour after administration. Neither 2Ff3G1uF nor 2F3ManF significantlyaffected 3-galactosidase activity in either brain or kidney at this dosage, nor did 2Ff3G1uFaffect 13-mannosidase activity in either tissue. However, 293ManF inhibited -glucosidaseactivity in both brain and kidney, but less effectively than 2Ff3G1uF one hour afteradministration (65% versus 35% of control levels for 2FI3GIuF in the brain; 60% versus40% of control levels for 2F3CuluF in the kidney).d) Inhibition of J3-Glucosidase Activity by 2FJ3ManF and In Vivo Recovery. Theinhibition of 3-glucosidase activity by 2F3ManF (see above) was further investigated.Figure 4.18 shows the effect of 2F3ManF and 2FI3G1uF (both 10 mg/lcg) on f3-glucosidaseactivity in brain, kidney, liver and spleen one hour after administration. Inhibition by133Time (h)Figure 4.15. In vivo inhibition of 13-glucosidase activity by 2FI3G1uF and recovery ofactivity in vivo. f3-Glucosidase activities in tissue homogenates from male Wistar ratsfollowing a single injection of2FJ3GluF (10 mg/kg) at zero time. J3-Glucosidase activity inthe indicated tissues was assayed as described in text: brain, 0; spleen, D,• liver, N;kidney,I.Cr:C1406040200 20 401201008060402010 20 30 40 50Time (h)Figure 4.16. In vivo inhibition of ,l3-mannosidase activity by 2FI3ManF and recovery ofactivity in vivo. f3-Mannosi&zse activities in tissue homogenates from male Wistar ratsfollowing a single injection of2Ff3ManF (10 mg/kg) at zero time. 13-Mannosidase activityin the indicated tissues was assayed as described in text: : brain, 0; spleen, L liver, Nkidney,.013430Galactosidase Glucosidase Mannosidase4=20C 100 0Control2FGluF2FManFFigure 4.17. Specificity of inhibition of /3-glycosidase activities by 2F/3G1uF and2FJ3ManF. Tissue /3-galactosido.se, /3-glucosidase, and f3-mannosidase activities in rat a)kidney and b) brain one hour after injection of 2Ff3G1uF or 2FI3ManF (10 mg/kg in 0.9%saline), or saline alone.Galactosidase Glucosidase Mannosidase6428242161208brain 8pleen liver idchiey40Figure 4.18. Inhibition of /3-glucosidose activity by 2FISG1uF and 2Ff3ManF. Tissue /3-glucosidase activities in rat brain, spleen, liver and kidney one hour after injection of2F/3G1uF or 2F/3ManF (10 mg/kg in 0.9% saline), or saline alone. Note change in scale ofkidney /3-glucosidase levels.1352F3ManF was less effective than by 2FGluF in all tissues examined; indeed, nosignificant inhibition was observed in spleen or liver with 2Ff3ManF, the organs in whichinhibition of 3-glucosidase activity by 2FGluF was greatest. In vivo recovery of (3-glucosidase activity inhibited by 2F[3ManF in kidney (not shown) was slightly faster thanthat inhibited by 2Ff3G1uF, with higher levels of (3-glucosidase activity recovered (80% ofcontrol activity for 2F(3ManF at 48 hours versus 60% for 2F(3G1uF at 48 hours).e) In Vitro Reactivation of 2FI3G1uF- and 2Ff3ManF-Inhibited J3-GlucosidaseActivity In Brain and Kidney Homogenates. Figure 4.19 shows an investigation of the invitro reactivation at 37°C of 2F[3G1uF-inhibited (3-glucosidase in brain homogenates fromanimals sacrificed after one hour. The rates of in vitro and in vivo reactivation (see Figure4.15) were similar, essentially full activity being regained in 48 hours. The in vitro ratewas enhanced in the presence of the transglycosylation ligands ceramide (2 mM) orj3IPTGIu (20 mM). Figure 4.20 shows an investigation of the in vitro reactivation at 37°Cof 2F(3G1uF- and 2Ff3ManF-inhibited f3-glucosidase activity in kidney homogenates fromanimals sacrificed after one hour. As observed in vivo, the 2FJ3ManF-inhibited (3-glucosidase regained higher levels of activity than the 2Ff3GluF-inhibited enzyme over 48hours. The 2Ff3ManF-inhibited f3-glucosidase recovered essentially to control levels(versus 80% of control levels in vivo) and showed an enhanced rate of reactivation withceramide (2 mM). However, in contrast to the situation in vivo, no further decreases inactivity in the kidney were observed at 20 hours. No significant recovery in activity wasobserved for the 2F(3G1uF-inhibited (3-glucosidase activity from kidney after 48 hours invitro, either in the presence or absence of ceramide, activity remaining at 50% of controllevels. In vivo recovery in this tissue was also slow, however, with only 60% of controlactivity regained by 48 hours.136‘IcuO)wECl)CuCl)0ocTime (h)Figure 4.19. In vitro reactivation of /3-glucosidase activity in rai brain homogenates ci3 7°C. J3-Glucosidase activity was originally inhibited by 2FJ3G1uF (10 mg/kg) in vivo.Homogenates were incubated in buffer (0), in presence of 20 mM f3IPTG1u (X), or inpresence of 2 mM ceramides (A). Aliquots were removed periodically and assayed asdescribed in text.C.)Cua)C’)CuCl)0Time (h)Figure 4.20. In viiro reactivation of /3-glucosidase activity in rat kidney homogenates ci3 7°C. f3-Glucosidase activity originally inhibited by 2Ff3GluF (10 mg/kg) in vivo andhomogenates incubated in buffer (0), or in presence of 2 mM ceramides (A); /3-glucosidase activity originally inhibited by 2Ff3ManF (10 mg/kg) in vivo and homogenatesincubated in buffer (•), or in presence of2 mM ceramides (LI).4200 10 20 30 40 500)E0EC4030201000 10 20 30 40 50137f) In Vitro Reactivation of 2FI3ManF-Inhibited J3-Mannosidase Activity In KidneyHomogenates. The in vitro reactivation at 37°C of 2F3ManF-inhibited -mannosidase inkidney homogenates was also examined (not shown). Rates of recovery of 2F3ManF-inhibited 13-mannosidase activity in vivo (Figure 4.16) and in vitro at 37°C in kidney over48 hours were similar, in each case recovering to — 60% of control levels. Like the2F3GluF-inhibited (3-glucosidase activity, no further decrease in activity was observed at20 hours as was seen in vivo. Neither ceramide (2 mM) nor G1uNAc (20 mM)significantly affected the rate of reactivation in vitro.g) Interpretation of Results of 2-Deoxy-2-Fluoro Glycosyl Fluoride Inhibition inRats. 2-Deoxy-2-fluoro-3-glucosyl- and 3-mannosyl fluorides were shown to be potentinhibitors of 3-glucosidase and f3-mannosidase activities in rats (Figures 4.15 and 4.16).At a single dose of 10 mglkg, levels of 3-glucosidase inhibition achieved with the glucocompound in some tissues were comparable to those obtained in mice using the cyclitolinhibitors CBE (Hara & Radin, 1979; Kanfer et a!., 1975) or cyclophellitol (Atsumi et a!.,1992) at similar or higher dosages. The significant inhibition of glycosidase activity in thebrain indicates that these glycosides cross the blood-brain barrier intact and are sufficientlyresistant to spontaneous hydrolysis in the brain and other tissues to result in the observedlevels of inhibition. All tissues except the kidney showed maximal inhibition by one hourand subsequent gradual recovery. The further drop in both glucosidase and mannosidaseactivities in the kidney by 20 hours (when glycosidase activities in all other tissues assayedhad generally recovered significantly) is consistent with higher inhibitor concentrations inthis organ than in other tissues during this period, resulting in greater inhibition. The lackof such a drop in the in vitro experiments provides further support for this notion. Further,the fact that most of the 2-deoxy-2-fluoro glycosyl fluoride is recovered intact in the urine(see Section 4.6c) suggests such a renal clearance.138Specificity of glycosidase inhibition. Greater than 50% of the f3-glycosidaseactivity in all tissues was inhibited by the appropriate mechanism-based inhibitor, indicatingthat the majority of the total tissue f-glycosidase activity as assayed is accounted for byretaining glycosidases since these inhibitors would not be expected to be effective againstinverting enzymes. Interpretation of the results is complicated by the number of enzymespecies in the various tissues. “-Galactosidase”, “f-glucosidase” or “13-mannosidase’activities in this study refer to the total hydrolytic activity in the tissue homogenates towardsthe appropriate PNP 3-D-glycopyranoside under the assay conditions described (seeExperimental chapter). Though the assay condiditons (low pH, presence of detergents) areexpected to favour the lysosomal acid glycosidases, activity of neutral enzymes may bereflected in the measured activities as well. Both 3-glucocerebrosidase and the neutral -glucosidase are reported to be retaining enzymes (Grabowski, 1990; Legler & Bieberich,1988) and Cavanagh et a!. have suggested that the non-lysosomal -mannosidase fromgoat liver is a retaining enzyme, while the lysosomal -mannosidase is inverting (Cavanaghet al., 1985). If this is true, 2FManF would thus be expected to selectively inhibit thenon-lysosomal retaining 3-mannosidase. The residual activity observed when -glucosidase or -mannosidase activities are maximally inhibited may be due to invertingenzyme activity, or simply to incomplete inactivation at the dosage of inhibitors used.The 2-deoxy-2-fluoro glycosyl fluorides tested were specific inhibitors of theappropriate 3-glycosidase activity (Figure 4.17). Thus, neither 2FfGluF nor 2FManFinhibited -galactosidase activity in brain or kidney; nor did 2F3GluF inhibit 3-mannosidase activity in either of these tissues. However, 2F3ManF inhibited J3-glucosidase activity in brain and kidney, but not as effectively as 2F3GluF. Levels of 13-glucosidase activity in spleen and liver, organs in which inhibition of glucosidase activityby 2Ff3G1uF was greatest, were not significantly affected by 2F13ManF (Figure 4.18).This (weaker) inhibition of 13-glucosidase activity by 2F13ManF likely reflects the relativelack of specificity of the neutral 13-glucosidase for the glycone portion of its substrate139(Glew et a!., 1976), while f-glucocerebrosidase is reportedly highly specific for f3-glucosides, with possibly some 13-xylosidase activity (Grabowski, 1990). Indeed, it hasbeen demonstrated with a purified preparation of f-glucocerebrosidase that although2Ff ManF inactivates this enzyme, it is 12-fold less effective in terms of k1/K than2F3GluF (Figure 4.4). The lack of significant -glucosithse inhibition by 2F[3ManF inspleen and liver, organs for which in rats and mice most of the 3-glucosidase activity isattributed to 3-glucocerebrosidase (Glew et a!., 1993) is consistent with this less efficientinhibition of f3-glucocerebrosidase by 2FfManF. Consistent with the apparent inhibitionof the neutral 3-glucosidase by 2FfManF, the organ with the most significant inhibition off3-glucosidase activity by 2FJ3ManF (60% of control levels) is the kidney, which in rats hasa large proportion of the less specific neutral enzyme (Glew, 1993; Legler & Bieberich,1988). Indeed, in experiments with purified neutral f-glucosidase (in collaboration withDr. Robert Glew at the University of New Mexico Health Sciences Centre), incubationwith identical concentrations (0.25 mM) of 2F[3G1uF or 2FfManF resulted in similar levelsof inactivation, 54% and 35%, respectively, after 20 minutes (McCarter et al., 1994).Recovery of glycosidase activitily. Significantly faster in vivo recovery ofglycosidase activity was observed with glycosidases inactivated by 2-deoxy-2-fluoroglycosyl fluorides compared to those inactivated by the active site-directed irreversibleinhibitors CBE (Hara & Radin, 1979) or cyclophellitol (Atsumi et al., 1992) in mice, or byglycosylmethyl p-nitrophenykriazenes in cultured human skin fibroblasts (Van Diggelen &Galjaard, 1980). Recovery of f-glucocerebrosidase activity in mouse tissues following asingle dose of cyclophellitol (50 mg/kg) occurred in seven and fourteen days, in liver andbrain, respectively (Atsumi et a!., 1992). Similarly, turnover times of human skinfibroblast 3-glycosidase activities inactivated by glycosylmethyl p-nitrophenyliriazenesvaried from five to thirteen days, depending on the enzyme (Van Diggelen and Galjaard,1980). Recovery of enzyme activity in these instances is presumed to be due to proteinsynthesis, as would be expected if the inactivation of glycosidase activity were truly140irreversible. Such rates of reactivation alone cannot account for the relatively rapid in vivorecovery of glycosidase activity (within 24 to 48 hours in some tissues) observed afterinactivation by 2-deoxy-2-fluoro glycosyl fluorides (Figures 4.15 and 4.16). Sincereactivation of 2-deoxy-2-fluoro-glycosyl-enzyme intermediates can occur by hydrolysis ortransglycosylation to an appropriate ligand in vitro, such processes could possibly accountfor the relatively rapid recovery of glycosidase activity observed in vivo. Indeed, purified(-glucocerebrosidase from calf spleen has previously been shown to undergo an equivalentreaction with[‘4C]-ceramide as the transglycosylation ligand when PNP f3-glucoside or 4-methyl umbelliferyl 3-glucoside are used as glucosyl donors (Kanfer et al., 1975). In vitrorates of glycosidase reactivation in tissue homogenates (where recovery via proteinsynthesis cannot occur) were therefore explored.-Glycosidase activity in brain and kidney tissue homogenates that had beeninhibited in vivo underwent spontaneous reactivation upon incubation at 370C in buffer,recovering partially or fully to control levels after —50 hours in vitro. No further decreasesin activity at 20 hours in the kidney as seen in vivo were observed, presumably sincefurther concentration and hence greater inhibition in the kidney cannot occur. In vitroreactivation rate enhancements of 2FfGluF-inhibited f-glucosidase activity by both thehydrolytically stable glucoside 3lPTGlu (20 mM) and ceramides (2 mM) were observed inbrain homogenates (Figure 4.19). The heterogenous nature of the tissue homogenatesagain complicated interpretation of the results. However, enhanced reactivation of a 2-deoxy-2-fluoro-glucosy1--glucocerebrosidase adduct in the presence of ceramide as atransglycosylation ligand is fully expected since glucosyl ceramide is the natural substrateof this enzyme. The natural substrate(s) of the neutral 3-glucosidase is not known, but thewell-known ability of neutral enzyme to catalyze transglycosylation to sugars or aliphaticalcohols via a hydrophobic aglycone site (Glew et a!., 1993; Legler & Bieberich, 1988)suggests that enhanced reactivation of a 2-deoxy-2-fluoro-glucosyl-enzyme adduct by3IPTGlu (Figure 4.19) or ceramides (Figure 4.20) is quite reasonable. Further,1412F3ManF-inhibited -glucosidase activity in kidney recovered faster in vitro than thatinhibited by 2FG1uF (Figure 4.20). Since with purifed glucocerebrosidase, the 2-deoxy-2-fluoro mannosyl adduct exhibits slower reactivation than the glucosyl-enzymeintermediate (Figures 4.3 and 4.5), this again indicates that the kidney 3-glucosidaseactivity assayed is due largely to the neutral enzyme. Faster turnover of glucosidaseactivity inhibited by a 2-deoxy-2-fluoro glycoside of the manno configuration has beenseen previously in vitro with Agrobacterium f3-glucosidase where the 2-deoxy-2-fluoro-mannosyl enzyme reactivates two orders of magnitude faster than the 2-deoxy-2-fluoro-glucosyl adduct (Street et a!., 1992).Similarly, 2F3ManF-inhibited f3-mannosidase activity in kidney homogenatesrecovered to — 60% of control activity upon incubation in buffer at 37°C (not shown). A 2-deoxy-2-fluoro-mannosyl-f-mannosidase adduct might be expected to exhibit an increasedreactivation rate in the presence of G1uNAc moieties since the natural substrates ofmammalian -mannosidases are the di- or irisaccharides Man-([31,4)-G1uNAc or Man(f31,4)-G1uNAc-(1,4)-GluNAc. Faster reactivation was not observed at the concentrationof G1uNAc used (20 mM) but it is possible either that a disaccharide is required foradequate binding and full realization of any rate enhancement, or that reactivation of the 2-deoxy-2-fluoro mannosyl-mannosidase(s) is inherently very slow (indeed, in vivo recoveryof j3-mannosidase activity was significantly slower than recovery of -gIucosidase activityin most tissues examined; cf Figures 4.15 and 4.16). Moreover, in contrast to -glucosidase activity, reactivation of 2F3ManF-inhibited (3-mannosidase activity was notenhanced by ceramide (2 mM), as expected. Transglycosylation to ceramide by a 2-deoxy-2-fluoro-mannosyl-3-mannosidase, an enzyme which normally binds the sugars G1uNAcor chitobiose in its aglycone site, would not be expected to be an effective process.The rates of in vivo glycosidase recovery (Figures 4.15 and 4.16). and the rates of invitro reactivation (Figures 4.19 and 4.20) were similar. This similarity in rates, coupledwith the relatively rapid recovery of enzyme activity, suggests that hydrolysis or142transglycosylation of 2-deoxy-2-fluoro glycosyl-enzyme adducts are the primarymechanisms involved in recovery of both 3-mannosidase and 3-g1ucosidase activitiesinhibited by these 2-deoxy-2-fluoro glycosides in vivo. While it is possible that theincreased reactivation rates observed with ceramides or I3IPTG1u in vitro are due toprocesses other than transglycosylation, possibly stabilization of the enzyme(s) involved, itis nonetheless clear that the relatively rapid recovery of glycosidase activity in vivo can bemimicked in vitro either in the presence or absence of ligands.h) Conclusions. 2-Deoxy-2-fluoro-f-glucosyl or f-mannosyl fluoridesadministered to rats in a single dose (10 mg/kg) inhibited 3-glucosidase or [3-mannosidaseactivity after one hour in brain, spleen, liver and kidney tissues. This inhibition,presumably due to accumulation of 2-deoxy-2-fluoro-glycosyl-enzyme intermediates,indicates that intact 2-deoxy-2-fluoro glycosyl fluorides are distributed to these organs andin the case of brain, that they cross the blood/brain barrier. 3-Glucosidase activityrecovered completely or partially in brain, spleen, liver and kidney by 20 to 48 hours. 3-Mannosidase activity partially recovered in all tissues by 48 hours. 3-Galactosidase activityin brain and kidney was not significantly affected by administration of either the gluco ormanno compounds at this dosage, indicating that these inhibitors are directed towardsspecific glycosidases. Observation of similar, relatively rapid rates of 3-glycosidasereactivation in vivo and in tissue homogenates in vitro at 37°C suggests that hydrolysis oriransglycosylation of 2-deoxy-2-fluoro-glycosyl-enzymes, not protein synthesis, are theprimary mechanisms involved in the recovery of glycosidase activity inhibited by this classof compounds in vivo (McCarter et al., 1994).4.5 Radiosynthesis of 2-deoxy-2-[’8F]fIuoro-3-D-mannopyranosyl[18F]fluoride and 18F-Labelling of Agrobacterium faecalis 3-Glucosidase In Vitro: a Model Study14318The labelling of one of the glycosadase mhibitors described above with F wasundertaken for use as a potential diagnostic agent to image glycosidase activity withpositron emission tomography (PET). 2FfManF was chosen for two reasons: firstly, theradiosynthesis of this compound may be readily carried out by the addition of[18F]-F2 toD-glucal; secondly, 2F3ManF is an effective mactivator of the retaining f-glucosidase fromAgrobacteriwn faecalis (Withers et at., 1988). The inactivation and derivatization of thiswell-characterized enzyme with an 18F-labelled inhibitor, and reactivation of the enzymemay serve as an in vitro model for the labelling ofmammalian enzymes.a) Radiosynthesis. The synthesis of a mixture of 2-deoxy-2-fluoro-a-glucosyl-and 3-mannosyl fluorides was achieved by modification of previously reported synthesesof 2-deoxy-2-fluoro-D-glucose and 2-deoxy-2-fluoro-D-mannose (Satyamurthy et a!.,1985), except that the glycosyl fluorides initially produced were separated withoutsubsequent hydrolysis. [‘8FJ-F2with neon as carrier was bubbled through a solution ofD-glucal in acetonitrile. The resulting 2:1 mixture of 2-deoxy-2-fluoro-cc-glucosyl- and(-mannosyl fluorides was evaporated to dryness in vacuo and the residue chromatographedon silica gel with diethyl ether as the eluant. Each compound was chromatographicallyhomogenous and ‘H and 19F nmr spectra were fully consistent with literature values(Diksic & Jolly, 1985; Satyamurthy et at., 1985). The radiochemical yield of the 2-deoxy-2-fluoro--mannosyl fluoride was 12% and the overall time from end of bombardment(EOB) for synthesis and purification was — 40 mm. The specific activity was 500mCiJmmol.b) LoJ,elling of Agrobacterium faecalis f3-glucosidase with ‘8F. The fractioncontaining the -mannosyl fluoride was evaporated, dissolved in 50 mM sodium phosphatebuffer, pH 6.8 and an aliquot incubated with a solution of Agrobacterium f3-glucosidase(2.7 mg/mL) for 15 minutes at 37°C, sufficient to ensure > 99.9% inactivation of the144enzyme at the inhibitor concentration used (— 1 mM). This rate is consistent with theinactivation parameters determined previously for the unlabelled 2-deoxy-2-fluoro-f3-mannosyl fluoride (K1 = 1.29 mM, k1 = 5.6 min’ (Withers et a!., 1988)). Theenzyme/inhibitor mixture was then chromatographed on an HPLC gel permeation column(300 A pore-size) using phosphate buffer as the eluant. The radiochromatogram of theenzyme/inhibitor mixture showed two peaks (Figure 4.21). The first, smaller peak elutedat the same time as uninhibited enzyme (detected by UV absorbance at 214 nm) andtherefore comprised the radiolabelled inhibited enzyme, while the second, larger peakcorresponded to excess unbound inhibitor. This is the first demonstration of the labellingof a glycosidase with a PET isotope. The labelled enzyme peak was collected to monitorturnover of the isolated 2-deoxy-2-[18F]-fluoromannosyl-enzyme.18c) Reactivation of the 2-deoxy-2-[ F]fluoro-a-D-mannosyl-enzyme. Thecollected glycosyl-enzyme complex was incubated in buffer at 37°C. in the absence andpresence of 197 mM l-deoxy--D-glucosyl benzene, which promotes reactivation via afacile transglycosylation reaction yielding a disaccharide product (Withers, 1990). Aliquotswere removed at intervals and re-injected onto the HPLC column. After correction fordecay of ‘8F (t1 = 110 mm), the decrease in radioactivity of the glycosyl-enzyme peakand the concomitant increase in radioactivity of the labelled sugar peak (which co-elutedwith a 2-deoxy-2-fluoro-D-mannose standard) were determined by collecting theappropriate fractions and measurement of radioactivity by scintillation counting. The 2-deoxy-2-[’8Flfluoro mannosyl moiety was released from the enzyme considerably fastervia transglycosylation with glucosyl benzene (k3 = 4.7 x 10 ± 0.4 x 10 min1) than viahydrolysis (k3 = 6.8 x 10 ± 0.3 x 10’ min’), as expected (Figure 4.22). In both cases,the labelled hydrolysis or transglycosylation products eluted significantly later than theglycosyl-enzyme complex, as would be anticipated for small molecules such as mono- ordisaccharides on a gel permeation column. These data are in good agreement with turnover145Excess Inhibitor0Figure 4.22. Plots of enzyme-bound label remaining (%) versus time for the isolated 2-deoxy-2-(18F]fluoro mannosyl-enzyme in buffer (A), and in• the presence of glucosylbenzene (197 mM) () at 37°C.Enzyme-boundInhibitor6 4Time (mm)Figure 421. Radiochromatogramfrom gel permeation HPLC column of incubated mixtureof excess2-deoxy-2-[8F]fluoro-J3-D-mannopyranosyl[‘8F]fluoride and Agrobacterium f3-glucosidase.__100806040220040 80 120 160 200 240Time (mm)146rates previously obtained by assaying activity of the iactivated enzyme with p-nitrophenyl3-D-glucopyranoside (k3 = 5.7 x103min1 for iransglycosylation at this concentration ofglucosyl benzene and k3 = 10 x 10 min’ for hydrolysis (Street et al., 1992)), confirmingthat the radiolabelling and subsequent release of radiolabel has occurred by the formationand turnover of a glycosyl-enzyme intermediate (McCarter et a!., 1992).4.6 Biodistribution and preliminary imaging results of 2-deoxy-2-[18F]fiuoro--D-mannopyranosyl[18F]fluoride in rats.a) Background and significance. Positron emission tomography (PET) has thepotential for imaging glycosidase activity in lysosomal storage diseases and possibly, intumour imaging. The proposed imaging strategy hinges on the covalent inhibition ofspecific glycosidases, resulting in labelled enzymes. Unbound inhibitor might be expectedto rapidly clear from tissues, and the remaining radioactivity would then reflect the leveland regional distribution of glycosidase activity. Further, the amount of labelled inhibitoradministered in a typical tracer study would likely be too low to elicit a pharmacologicalresponse and turnover of the labelled enzyme would ultimately occur by hydrolysis of theglycosyl-enzyme. Irreversible covalent inhibitors of monoamine oxidase (deprenyl andclorgyline) (MacGregor et a!., 1985) and of L-amino acid decarboxylase (a-fluoromethyl6-fluoro-L-dopa) (Chirakal eta!., 1989) have been labelled with “C or 18F for the imagingof these enzyme activities in vivo using PET. The biodistribution, metabolic fate andpreliminary in vivo imaging results of 2-deoxy-2-[’Fjfluoro-[3-D-mannopyranosylI’8F]fluoride in rats are herein reported.b) Biodistribution of2-deoxy-2-118F]fluoro-J3-D-mannopyranosyl[18F]fluoride inrats. The biodistribution of2-deoxy-2-[’8F]fluoro-J3-D-mannopyranosyl [‘8F]fluoride inWistar rats at 2, 15, 30, 60 and 120 mm is shown in Table 4.2. Aliquots of the labelledcompound prepared as above were administered by intravenous injection to a group of rats,147Table 4.2. Tissue distribution of 2-deoxy-2 f’8F]fluoro-f3-D-mannopyranosyl[‘8F]fluoride in male Wistar rats.Uptake (% JDIg)2 mm 15 mm 30 mm 60 mm 120 mmBlood 1.72 ± 0.25 0.48 ± 0.10 0.35 ± 0.08 0.51 ± 0.07 0.55 ± 0.06Heart 0.94 ± 0.10 0.41 ± 0.11 0.48 ± 0.08 0.53 ± 0.06 0.66 ± 0.06Lung 0.91 ± 0.12 0.33 ± 0.06 0.31 ± 0.02 0.39 ± 0.02 0.48 ± 0.07Brain 0.25 ± 0.01 0.20 ± 0.05 0.23 ± 0.03 0.30 ± 0.04 0.36 ± 0.04Spleen 0.89 ± 0.12 0.34 ± 0.06 0.31 ± 0.03 0.37 ± 0.05 0.45 ± 0.16Liver 1.72 ± 0.11 0.47 ± 0.12 0.43 ± 0.06 0.53 ± 0.03 0.54 ± 0.05Kidney 3.18 ± 0.43 0.77 ± 0.07 0.68 ± 0.08 0.67 ± 0.04 0.93 ± 0.06Bone --- 0.18 ± 0.05 0.18 ± 0.02 0.31 ± 0.05 0.26 ± 0.08Expressed as % ID (injected dose)/g tissue ± standard deviation for 5-6 animals.and 5-6 animals were sacrificed at each of the indicated times. The organs were removed,blotted, weighed, and tissue concentrations of injected radiolabel were determined byscintillation counting. The highest concentrations were found in the kidneys, reflecting arenal clearance of this compound consistent with the further drop in kidney glycosidaseactivity after 20 h (Figures 4.15 and 4.16). Clearance of an intravenous bolus of manydrugs is often a bi-exponential process, comprising an initial rapid distribution phase and aslower elimination phase (see, e.g. (Rowland & Tozer, 1989)). After an initial rapiddistribution phase, the label in this study shows a very slow clearance from all tissues, thetissue concentrations remaining essentially constant, within error, over the course of thestudy. In this, the pharmacokinetics resemble those of the glycosidase inhibitor conduritolB epoxide (CBE) labelled with thtium, which has a half-life of - 7 h in mice (Kanfer,1982). The slow clearance of these compounds may be due to binding to proteins in148plasma and tissues, or to reabsorption from the glomerular filtrate of the kidneys back intothe blood by the glucose transport system (Goodman et a!., 1981; Rowland & Tozer,1989). Such a slow clearance cannot unfortunately be followed accurately with the short-lived isotope 18F.Unlike CBE, which only slowly penetrates the brain, reaching a maximum level ofuptake and resulting in the greatest degree of inhibition of glycosidase activity after 12 h,2F3ManF appears to rapidly penetrate the blood-brain barrier, resulting in maximuminhibition of activity by 1 h. Indeed, significant recovery of activity occurs by 20 h(Figures 4.15 and 4.16). The slow uptake of CBE to the brain is likely a consequence ofthis compound being a poor substrate for brain glucose transport proteins, due to the lackof a C-5 hydroxymethylene group and the presence of the epoxide ring. The importance ofthe C-5 substituent for efficient transport of glucose analogues has been shown inerythrocyte (GLUT1) (Barnett et a!., 1973) and liver-type (GLUT2) (Rees & Holamn,1981) glucose transporters, xylose binding 10-fold more weakly than glucose. Bycontrast, 2-deoxy-2-fluoro glucose, mannose, and f3-glucosyl fluoride bind with K1 valuescomparable to that of glucose (Barnett et a!., 1973), and thus the efficient transport of2FManF is expected.A structurally related glycosyl fluoride 2-deoxy-2-[’8F]fluoro-c -D-glucosyl[‘8F]fluoride has been synthesized and its biodistribution determined in mice (Shiue et a!.,1984). The tissue distributions of2-deoxy-2-[’8F]fluoro-c-D-glucosyl[18F]fluoride and2-deoxy-2-[’F]fluoro-3-D-mannopyranosyl [18FJfluoride are significantly different,reflecting the different metabolism of these compounds. Unlike 2-deoxy-2-[’8F]fluoro-f3-D-mannopyranosyl [18F]fluoride, the cL-gluco compound was apparently rapidlyhydrolyzed in vivo, likely by the action of cz-glycosidases. Recall that 2-deoxy-2-fluoro-a-glucosyl fluoride is a substrate for yeast a-glucosidase in vitro. After 120 mm, much ofthe label was present in bone, presumably as[18F]fluoride, and considerable uptake inheart and brain were detected, as expected, since the 2-deoxy-2-fluoro-D-glucose released149would be phosphorylated by hexokinase and be retained in tissues of high glucose• 18 18metabolism. Most of the label from 2-deoxy-2-[ F]fluoro-a-D-glucosyl [ F] fluoridewas recovered in the urine as fluoride, in contrast to the results observed in the presentstudy with 2-deoxy-2-fluoro-3-D-mannopyranosyl fluoride (see Section 4.6c).One of the aims of this study was to determine if the biodistribution of 2-deoxy-2-[18F]fluoro-f-D-mannopyranosyl [‘8Fjfluoride was specifically associated withglycosidase activity in vivo or if non-specific uptake would mask specific binding.Binding and inhibition of various glycosidases is clearly demonstrated by the significantinhibition of glycosidase activity in tissue homogenates (Section 4.5), but the labelledinhibitor may also be distributed non-specifically to tissues due to non-specific interactionswith other proteins or some other reason which results in slow clearance. After pretreatment with 2FI3ManF (20 mg/kg), the uptake of 2-deoxy-2-[’8F]flu ro-3-D-mannopyranosyl[18F]fluoride in all tissues after 60 mm was the same as in control animals(Table 4.3). Since the pre-treatment dose is twice that required to produce 55 to 65%inhibition of 3-mannosidase activity in kidney, liver, spleen and brain homogenates (andsignificant inhibition of f-glucosidase activity in these tissues) after 60 mm (see Figure4.16 and Figure 4.18), this pre-treatment dose would be expected to be sufficient to blockthe majority of relevant enzyme active sites. Since this blocking is not observed, nonspecific uptake of the label, at least at this time after injection, appears to be masking uptakedue to specific enzyme inactivation. The uptaken label may be in the form of intact2FI3ManF or a number of metabolites. Interestingly, pre-treatment with unlabelledcompound similarly failed to block uptake of the 11C-labelled competitive mannosidaseinhibitor N-[C]methyl-1-deoxymannonojirimycin, though uptake of the gluco analogueN-[’C]methyl-1-deoxynojirimycin in some tissues was significantly reduced by pretreatment with N-methyl-1-deoxynojirimycin (Ishiwata et al., 1993).150Table 4.3. Tissue distribution of 2-deoxy-2-[’8F]fluoro-3-D-mannopyranosyl[‘8F]fluoride in female Sprague-Dawley rats at 60 mm post-injection, control and pretreated with 2Ff3ManF.Uptake (% ID/g)Control Pre-treated (+20 mg/kg 2F[ManF)tBlood 0.23 ± 0.02 0.28 ± 0.03Heart 0.42 ± 0.01 0.43 ± 0.19Lung 0.24 ± 0.01 0.29 ± 0.04Brain 0.24 ± 0.01 0.22 ± 0.06Spleen 0.32 ± 0.03 0.32 ± 0.05Liver 0.24 ± 0.02 0.32 ± 0.05Kidney 0.58 ± 0.09 0.60 ± 0.03Bone 0.13 ± 0.01 0.10 ± 0.03Expressed as % ID (injected dose)/g tissue ± standard deviation for 3 animals.Pre-treated animals received 20 mg/kg wilabelled 2F(3ManF 60 mm prior to injection ofthe18F-labelled compound.c) 19F NMR spectroscopy of 2-deoxy-2-fluoro-f3-D-mannopyranosyl fluoride inrat tissues. The metabolic fate of the administered 2FI3ManF was assessed by ‘9F NMR.An initial concern was that the glycosyl fluoride would undergo rapid spontaneous orenzymatic hydrolysis (possibly by the action of inverting enzymes) prior to the desiredinactivation of retaining glycosidases. The 2-deoxy-2-fluoro mannose produced wouldthen be phosphorylated by hexokinase (Bessell et al., 1972) and accumulate as glycosylphosphates in tissues of high metabolic glucose demand (heart and brain), in much thesame way as FDG, masking specific glycosidase labeffing. In the absence of suchhydrolysis, it is unlikely that an intact glycosyl fluoride would be phosphorylated by151hexokinase since it has been shown that both a- and [3-glucopyranosyl fluorides areexceedingly poor substrates for yeast hexokinase (Bessell et at., 1972).Unlabelled 293ManF was administered to a rat (at a dosage of 70 mgllcg), and theanimal was sacrificed after 60 mm, a time suitable for imaging with ‘8F and within therange of times examined in the biodistribution study above. The organs were removed andhomogenized as described previously, and the brain, liver, and kidney homogenates werecentrifuged to remove cell debris, filtered, and the supernatant freeze-dried. Urine in thebladder was also removed and freeze-dried. The residues were dissolved in D20 and 19FNMR spectra were acquired.The amounts of fluorinated compounds in the tissue samples (brain, liver, andkidney) were, as expected, very small (cf Table 4.2). Indeed, no fluorine signal wasdetected in the brain sample. However, measurable signals were detected after overnightdata acquisition of both the liver and kidney samples. Most of the administered compoundwas detected in the urine, as judged by the much weaker 19F NMR signal intensities of thebrain, liver and kidney samples compared to the urine sample, necessitating many morescans for the tissue samples. These relative amounts compare favourably with thebiodistribution values above (Table 4.2). More importantly, spectra in all tissues indicatedthat little hydrolysis or other metabolic transformation of the 2F3ManF resulting in thegeneration of new fluorinated species had occurred at this time. No major peaks other thanthe two signals due to the C-i and C-2 fluorines of 2F[3ManF were detected (Figure 4.23)in this water-soluble fraction of the tissue homogenates. In particular, major signals due toinorganic fluoride or to the a or anomers of 2-deoxy-2-fluoro-D-mannose at - 44.9, -123.7, and - 142.8 ppm, respectively (trifluoroacetic acid reference) were not observed.The chemical shifts and fluorine-fluorine coupling constants of the peaks at -71.9 and -146.7 ppm (F1,F2 = 12 Hz) were identical to those of 2F3ManF (Diksic & Jolly, 1985;Satyamurthy et a!., 1985).152ACb*.—80 —100 —120 —140PPM1 19Figure 423. H-decoupled F-NMR spectra of A) liver and B) kidney homogenates,and C) urine of rat athninistered 2FJ3ManF (70 mg/kg). lnsets. Expansion of signals at -71.9 and -146.7 ppm. Data was acquired over 150,000 scans for the liver and kidneysamples, and over 256 scansfor the urine sample. Spectra are referenced to TFA.B-fl.5 —72.0 —72.5PPNV153The high dose of 2I3FManF used in the NMR study is unavoidably much greaterthan the tracer amounts employed in the biodistribution study. It is possible thatmetabolism of the compound is different in the NMR study due to saturation of variousenzyme systems. Comparison of tracer and NMR biodistribution studies of FDG are,however, in good agreement (Kanazawa et a!., 1986; Reivich er a!., 1979). In addition,the possibility that protein-bound metabolites that were present in the tissues were removedduring centrifugation cannot be excluded. However, it appears that significant hydrolysisof 2FfManF has not occurred and that the majority of the radiolabel in tissues at this time ispresent as intact inhibitor.d) Imaging experiment in a rat. 2-Deoxy-2-[’8F]fluoro-f3-D-mannopyranosyl[18F]fluoride (— 1 mCi), prepared as in Section 4.5a, was administered intravenously in abolus injection to a single anaesthetized rat immobilized in a prone position. PET scanswere recorded continuously from the time of radioisotope injection with a Siemens/CTIECAT 953B. Figure 4.24 shows images reconstructed from data acquired -40 mm afterinjection of the compound. The image likely reflects the blood distribution since thebiodistribution shows low tissue-to-blood ratios and the blocking experiment indicates thatuptake at this time is non-specific (Section 4.6b). The distribution is not homogenous,however, and some features are apparent, although their meaning is unclear.e) Conclusions. Though apparently only a fraction of labelled 2F(ManF in thetissues at up to 120 mm post-injection is bound to inactivated glycosidases, the resultssuggest that PET imaging of tissue glycosidase activity may be feasible. No interferencefrom glucose metabolism would be expected since spontaneous hydrolysis of this inhibitorin the time course of the experiment appears to be insignificant. However, analogues of2F[3ManF which clear more rapidly and which result in more rapid inactivation of specificglycosidases are desirable, in order to develop a useful imaging agent which reflects tissue154glycosidase activity. 2-.Deoxy-2-fluoro glycosyl fluorides labelled with 18F thus may beused for imaging abnormal levels of glycosidase activity associated with lysosomal storagediseases (e.g. Gaucher’s disease).y: 431 zc: 2Figure 4.24 Transaxial, sagittal, and coronal views ofa rat. PET image after injection of—1 mCi of2 -deoxy-2 f’8F]fluoro-f3-D-mannopyranosyl[18F]fluoride.155CHAPTER 5MATERIALS AND METHODS1565.1 Synthesis5.1.1 GeneralMelting points were determined with a Mel Temp II melting point apparatus and areuncorrected. Thin layer chromatography (TLC) was performed on aluminum-backedsheets of silica gel 60F4 (Merck) of thickness 0.2 mm. Compounds were visualizedunder UV light and/or by charring with 10% H2S04 in methanol or 10% ammoniummolybdate in 2 M sulfuric acid. Solvents and reagents used were either reagent, certified orspectral grade. Methanol was distilled prior to use from Mg(OCH3)2under nitrogen.Pyridine was distilled from calcium hydride. Distilled thionyl chloride was obtained fromDr. Brian Cliff.Nuclear magnetic resonance (NMR) spectroscopy was performed on the followinginstruments: 1H- and‘9F-NMR spectra on a 200 MHz BrUker AC-200 at 200 MFIz and188 MHz, respectively;1H-NMR spectra at 300 MHz or 400 MHz on a Varian XL-300 ora Brüker WH-400, respectively; 13C-NMR at 75 MHz on a Varian XL-300. For samplesdissolved in CDC13, ‘H-NMR chemical shifts are referenced to internal teiramethylsilane (6= 0.00 ppm). For samples dissolved inD20, the reference was external 2,2-dimethyl-2-silapentane-5-sulphonate (6= 0.015 ppm). 13C spectra are proton-decoupled with CDC13as reference. ‘9F-NMR chemical shifts are reported using the 6 scale referenced to CFC13(6 = 0.00 pppm), although the samples were referenced to external irifluoroacetic acid(TFA) (6 = - 76.53 ppm). Note that, where relevant, Fe represents an equatorial fluorineand Fa an axial fluorine. Routine low- and high-resolution desorption chemical ionizationmass spectrometry (DCI-LRMS and DCI-HRMS) were performed on a Delsi-NermagR10-1OC mass spectrometer using ammonia as the reagent gas.Micro-analyses were performed by Mr. Peter Borda, Micro-analytical Laboratory,University of British Columbia. Some protein and peplide electrospray mass (ES MS)spectra were recorded by Dr. David Burgoyne or Dr. Shichang Miao in Dr. RuediAebersold’s laboratory, Biomedical Research Centre, University of British Columbia. The1572,2-dihalo glycosyl chlorides were synthesized by an undergraduate student, Mr. WaiYeung, under my supervision. 3,4,6-Tri-O-acetyl-2-deoxy-2-fluoro-a-D-glucopyranosylfluoride was generously provided by Dr. David Dolphin.Column chromatography was performed using Silica Gel 60 (230-400 mesh) fromBDH Inc. The following solvent systems were employed: (A) 1:1 petroleum ether/diethylether, (B) 2:1 petroleum ether/diethyl ether, (C) 3:1 petroleum ether/diethyl ether, (D) 27:2:1ethyl acetate/methanol/water; (B) 2:1 petroleum ether/ethyl acetate; (F) 3:1 petroleumether/ethyl acetate; (G) 4:1 petroleum ether/ethyl acetate; (H) 5:1 petroleum ether/ethylacetate; (I) 8:1 petroleum ether/ethyl acetate; (J) 147:2:1 ethyl acetate/methanol/water, and(K) 199:2:1 ethyl acetate/methanol/water.5.1.2 General Proceduresa) General work-up procedure.Unless otherwise stated, the following work-up procedure was employed. To thereaction mixture was added dichloromethane, and the organic layer was washed withsaturated aqueous sodium bicarbonate until the washings remained basic, then washed withdoubly deionized water until neutrality. The organic layer was dried over magnesiumsulphate for — 20 minutes, suction-filtered, and the solvents evaporated in vacuo using arotary evaporator. The residual oil was purified by flash chromatography using the solventsystems indicated for the individual compounds.b) Photobromination with N-bromosuccinamide.The method is taken from (Praly et al., 1989). The per-O-acetylated glycosyl halide(2 -2.5 mmol) was dissolved in Cd4 (30 - 40 mL) in a 100 niL round bottom flaskequipped with a condenser. NBS (4 equivalents) and 2-3 boiling chips were added. Theresulting mixture was irradiated with a 500 W tungsten bulb situated directly beneath the158flask, and the mixture was refluxed for the times indicated for the individual compounds.A shroud of aluminum foil loosely enclosed the upper portion of the apparatus, but did nottouch the bulb surface. After the reaction was complete as indicated by TLC, the reactionmixture was allowed to cool, worked up according to the general procedure, and theresulting oil purified by chromatography.c) Deacetylation with ammonia in methanol.The acetylated glycosyl halide was dissolved in freshly distilled methanol, and thesolution was cooled to 0°C in an ice bath. Anhydrous ammonia was bubbled through thesolution until the solution was saturated (—5 mm), and the mixture was allowed to warm toroom temperature until the reaction was complete as judged by TLC. Solvent wasevaporated in vacuo, and the resulting syrup was purified by chromatography on silica gel.5.1.3 Syntheses of 5-Fluoro Glycosyl Fluorides2,3,4,6-Tetra-O-acetyl-5-fluoro-/3-D-glucopyranosylfluoride (2.7)The 5-bromo 3-g1ucosyl fluoride 2.6 (0.64 g, 1.49 nimol) (Praly et a!., 1989),prepared by the general photobromination procedure, was dissolved in toluene (10 niL)under a N2 atmosphere. The solution was cooled to 0°C, and AgBF4 (0.48 g, 2.47 mmol)was added. After stirring for one hour, the reaction was worked up according to thegeneral procedure and chromatographed (solvent F) to afford 2.7 as a colourless oil (0.060g, 0.163 minol, 11%). ‘H-NMR and‘9F-NMR data: see Table 2.1. DCI-HRMSCalcd. forC14H809F2+N}{4: 386.12627. Found: 386.12674.F1592,3,4,6-Tetra-O-aceryl-5-fluoro-a-D-glucopyranosylfluoride (2.12)The protected 5-fluoro idosyl fluoride 2.13 (0.20 g, 0.54 mmol) was dissolved inHF/pyridine (3 mL, Aldrich) in a dry plastic vial to which a catalytic amount of AgF wasadded, and cooled to -78°C. The mixture was alowed to warm to room temperature, andstirred for 15 h. The reaction mixture was diluted with dichioromethane, and addedcarefully to an ice-filled separatory funnel. The organic layer was washed with saturatedsodium bicarbonate, water and dried (MgSO4). Evaporation of the solvent, followed bychromatography (solvent F) afforded 2.12 (0.050 g, 0.14 mmol, 25%) along withrecovered 2.13 (0.013 g, 0.035 mmol, 7%).1H-NMR and 19F-NMR data: see Table2.1. AnaL Calcd forC14H89F2:C, 45.66; H, 4.93. Found: C, 45.82; H, 4.86.2,3,4,6-Tetra-O-acetyl-5-fluoro-a-L-idopyranosylfluoride (2.8)OAcThe 5-bromo 3-glucosyl fluoride 2.6 (0.226 g, 0.526 mmol) (Praly et a!., 1989),prepared by the general photobromination procedure, was dissolved in acetonitrile (5 mL)and AgF (0.405 g, 3.2 mmol) was added. The mixture was stirred under a N2 atmosphereat room temperature over 4A molecular sieves with the exclusion of light for 6 h. Themixture was dissolved in chloroform, and washed with saturated sodium chloride solution,saturated sodium bicarbonate, water and dried (MgSO4). Evaporation of solvent andchromatography (solvent E) afforded 2.8 (0.078 g, 0.212 mmol, 40%). ‘H-NMR andAcO.F160‘9F-NMR data: see Table 2.3. AnaL Calcd. forC14H809F2:45.66; 4.93. Found: C,45.36; H, 4.95.2 ,3,4,6-Tetra-O-acetyl-5 -fluoro-f3-L-idopyranosylfluoride (2.13),,OAcThe 5-bromo a-glucosyl fluoride 2.11 (0.52 g, 1.21 mmol) (Praly et a!., 1989),prepared by the general photobromination procedure, was dissolved in acetonitrile (7.5mL) and AgF (0.2 15 g, 1.69 mmol) was added. The mixture was stirred under a N2atmosphere over 4A molecular sieves in the dark for 48 h. The reaction mixture was takenup in chloroform, and washed with saturated sodium chloride solution, saturated sodiumbicarbonate, water and dried (MgSO4). Evaporation of solvent and chromatography(solvent A) afforded 2.13 (0.202 g, 0.549 mmol, 45%). ‘H-NMR and‘9F-NMRdata: see Table 2.3. Anal. Calcd. forC14H809F2:45.66; 4.93. Found: C, 45.40; H,4.95.5-Fluoro-J3-D-glucopyranosylfluoride (2.3)HOThe acetylated 5-fluoro glucopyranosyl fluoride 2.7 (0.102 g, 0.277 mmol) wasdissolved in methanol (5 mL) and deacetylated according to the general procedure for 3 h.Column chromatography (solvent D) gave 2.3 as a colourless oil (0.032 g, 0.160 mmol,59%). ‘H-NMR and‘9F-NMR data: see Table 2.2. Anal. Calcd. forC6H1005F2•0.5H2: C, 34.46; H, 5.30. Found: C, 34.57; H, 5.72.1615-Fluoro- a-D-glucopyranosylfluoride (2.4)OHHO0HOFOHFThe acetylated 5-fluoro glucopyranosyl fluoride 2.12 (0.050 g, 0.136 mmol) wasdissolved in methanol (5 mL) and deacetylated according to the general procedure for 4.5h. Column chromatography (solvent D) gave 2.4 as a colourless oil (0.022 g, 0.110mmol, 81%). H-NMR and 19F-NMR data: see Table 2.2. Anal. Calcd. forC6H10O5F2:C, 36.01; H, 5.04. Found: C, 35.56; H, 5.20.5-Fluoro-a-L-idopyranosylfluoride (2.9)HO>(><The acetylated 5-fluoro idopyranosyl fluoride 2.8 (0.078 g, 0.212 mmol) wasdissolved in methanol (15 mL) and deacetylated according to the general procedure for 4 h.Column chromatography (solvent D) gave 2.9 as a colourless oil (0.020 g, 0.10 mmol,47%). ‘H-NMR and‘9F-NMR data: see Table 2.4. Anal. Calcd. forC6H1005F2:C, 36.01; H, 5.04. Found: C, 35.92; H, 5.235-Fluoro-f3-L-idopyranosylfluoride (2.14)HO_,k_/162The acetylated 5-fluoro idopyranosyl fluoride 2.13 (0.125 g, 0.339 mmol) wasdissolved in methanol (15 mL) and deacetylated according to the general procedure for 2.5h. Column chromatography (solvent D) gave 2.14 as a colourless oil (0.067 g, 0.335mmol, 98%). ‘H-NMR and‘9F-NMR data: see Table 2.4. Anal. Calcd. forC6H1005F2:C, 36.01; H, 5.04. Found: C, 35.99; H, Syntheses of 2,2-Dihalo Glycosyl Chlorides3,4,6-Tri-O-aceryl-2 -fluoro-D-glucal (3.5)AcO3,4,6-Tri-O-acetyl-2-deoxy-2-fluoro-x-D-glucopyranosyl fluoride 3.4 (2.63 g,8.47 mmol) (Shelling et a!., 1984), obtained from Dr. D. Dolphin, was dissolved in 45%HBr/glacial acetic acid (10 mL) and acetic anhydride (2 mL) and stined for six hours,followed by the general workup procedure. The crude bromide was dissolved inacetonitrile (75 mL) and triethylamine (15 mL), and heated in a water bath at reflux for 24hours. After evaporation of the solvent in vacuo, thethylammonium bromide was removedby column chromatography (solvent E) to give the glucal 3.5 as a yellow oil (2.31 g, 7.96mmol, 98%). ‘H-NMR data (CDCI3,200 MHz) : 8 6.78 (d, 1 H, lF 4.2 Hz, H-i),5.63 (dt, 1 H, J34 4.0, 3,F 4.0 Hz, H-3), 5.21 (dd, 1 H, J4,3 4.0, J4,5 5.5 Hz, H-4),4.30 (m, 1 H, H-5), 4.29 (AB multiplet, 2 H, J66’ 12.0, J56 7.0, J56’ 3.8 Hz, H-6, H-6’), 2.12 (s, 3 H, OAc), 2.11 (s, 3 H, OAc), 2.10 (s, 3 H, OAc). 19F-NMR data(CDC13,188 MHz) : 8 - 167.1 (t, Jp1 J 3.9 Hz, F2).1633,4,6-Tri-O-acetyl-2-chloro-2 -deoxy-2-fluoro-a-D-glucopyranosyl chloride (3.6)OAcDry 3.5 (O.80g, 1.38 mmol) was dissolved in carbon tetrachioride (150 mL) over4A molecular sieves and cooled in a CCI4/dry ice bath (-23°C). Chlorine was bubbledthrough the solution until it turned a yellowish green (5 minutes). The flask was wrappedin aluminum foil to exclude light, and stined for 18 hours. Upon completion of thereaction, excess chlorine was purged with a stream of chy nitrogen for several minutes untilthe solution was colourless, and the solvent was evaporated in vacuo. The resultingmaterial was chromatographed twice (with solvent B, then solvent C) giving theglucopyranosyl chloride 3.6 as a yellow oil (0.17 g, 0.47 mmol, 34%). ‘H-NMR data(CDC13,400 MHz): 6 6.06 (d, 1 H, l,F 6.0 Hz, H-i), 5.71 (dd, 1 H, 3F 22.8, J34 9.7Hz, H-3), 5.28 (dt, 1 H, J43 9.7 = 45 JJ 1.5 Hz, H-4), 4.4 - 4.05 (m, 3 H, H-5,H-6, H-6’), 2.14 (s, 3 H, OAc), 2.07 (s, 3 H, OAc), 2.03 (s, 3 H, OAc).‘9F-NMRdata (CDC13, 188 M1-Iz) : 6 -120.4 (dd, Fl 6.0, F,3 22.8 Hz, F-2). Anal. Calcd. forC12H5O7FC12:C, 39.91; H, 4.19; Found: C, 39.73; H, 4.08.2-Chloro-2 -deoxy-.2-fluoro-a-D-glucopyranosyl chloride (3.1)The acetylated glucopyranosyl chloride 3.6 (0.12 g, 0.33 mmol) was dissolved inmethanol (10 mL) and deacetylated according to the general procedure for 4.5 h. Columnchromatography (solvent D) gave a yellow solid which was further purified using ethyl164acetate/methanol (24:1) affording a yellow oil (0.039 g, 0.12 mmol, 50%). ‘H-NMRdata (D20, 200 MHz): 6 6.36 (d, 1 H, l,F 6.6 Hz, H-i), 4.21 (dd, 1 H, 3,F 24, J34 9Hz, H-3), 4.10 - 3.69 (m, 4 H, H-4, H-5, H-6, H-6’). ‘9F-NMR data (D20, 188MHz): 6 -123.25 (dd, F3 24, F,l 6.6 Hz, F-2). Anal. Calcd. forC6H904FC12:C,30.66; H, 3.86; Found: C, 30.91; H, 3.93.Attempted syntheses of 3,4,6-tri-O-acetyl-2-chloro-2-deoxy-2-fluoro-a-D-manno-pyranosyl chloride (3.7)•OAci) Dry 3.5 (0.057g, 0.20 mmol) was dissolved in carbon tetrachloride (23 mL)over 4A molecular sieves and cooled in an ice bath (0°C). Chlorine was bubbled throughthe solution until it turned a yellowish green (5 minutes). The solution was re-saturatedafter stirring for 4.5 hours in the dark at room temperature and left to stir at roomtemperature for a total of 23 hours. The mixture was then purged with nitrogen and thesolvent was removed by evaporation in vacuo to yield a green oil. TLC and NMR analysisshowed that all the starting material 3.5 had been consumed, but none of the expectedproduct 3.7 was detected.ii) Dry 3.5 (0.0799g, 0.275 mmol) was dissolved in nitromethane (25 mL)containing 4A molecular sieves and the mixture was cooled to 0°C. The solution wassaturated with chlorine as before, allowed to warm to room temperature, and stirred in thedark for 18 hours. The solution was purged with nitrogen and the solvent was evaporated.The general work-up procedure afforded a light yellow oil which by TLC contained threecompounds, none of which was the starting material. After chromatography (solvent B),only the glucopyranosyl chloride 3.6 (7 mg, 0.02 1 mmol, 7%) was obtained.1651 ,3,4,6-Tetra-O-aceiyl-2 -deoxy-2 ,2 -difluoro-a-D-arabinohexopyranose (3.8)OAcOAcSodium acetate (280 g), glacial acetic acid (3.5 mL) and CFC13 (35 mL) weremixed in a 3-necked round bottom flask and cooled in a dry ice/acetone bath (-78 °C). Agas target (— 1 L) was filled with 20% fluorine in neon (10 psi) which was diluted 4-foldwith helium to 40 psi. This mixture was then bubbled through the slurry. The flask wasthus charged twice before the glucal 3.5 (0.800 g, 0.9 mmol) dissolved in CFC13 (10 mL)was added to the mechanically-stirred mixture. The loosely-stoppered flask was allowed towam to room temperature. After general work-up, column chromatography (solvent G)afforded the 2,2-difluoro per-O-acetate 3.8 (0.77 g, 2.1 mmol, 76%). ‘H-NMR data(CDC13,200 MHZ): 6 6.19 (t, 1 H, l,Fe = l,Fa 3.0 Hz, Wi), 5.50 (dt, 1 H, 3,Fa 20,3,Fe =J34 10 Hz, H-3), 5.20 (t, 1 H, J43 =J4,5 10 Hz, H-4), 4.3 - 4.0 (m, 3 H, H-5, H-6, H-6’), 3.88 (s, 3 H, OAc), 3.85 (s, 3 H, OAc), 3.81 (s, 3 H, OAc), 3.74 (s, 3 H,OAc). ‘9F-NMR data (CDC13,188 MHz) : 6 -121.0 (F-2e, F-2a coincident).3,4,6-Tri-O-aceiyl-1 ,5-anhydro-5-chloro-2-deoxy-2 ,2 -duoro-D-glucitol (3.10)(Attempted synthesis of 3.9 from 3.8 using DCME)•OAcClThe 2,2-difluoro per-O-acetate 3.8 (0.28 g, 0.76 mmol) was refluxed in 1,1-dichloromethyl methyl ether (2 mL) over 4A molecular sieves with a few crystals of freshlyfused zinc chloride for 5 hours at 68°C under nitrogen. After general work-up, columnchromatography (solvent G) afforded an amorphous white solid, 3.10 (0.039 g, 0.11mmol, 15%). ‘H-NMR data (CDCI3,400 MHz): 6 5.65 (ddd, 1 H, la,Fa 17.8, la,le16610.3, la,Fe 4.8 Hz, H-ia), 5.48 (dd, 1 H, lale 10.3, le,Fe 1.2 Hz, H-le), 4.45 (d, 1 H,12.2 Hz, H-6), 4.30 (ci, 1 H, J6’ 12.2 Hz, H6’), 4.23 (dd, 1 H, 3Fa 29.3, J3412.7 Hz, H-3), 4.13 (t, 1 H, 4,Fa 12.8, J43 12.8 Hz, H-4), 2.14 (s, 3 H, OAc), 2.11 (s,3 H, OAc), 2.10 (s, 3 H, OAc).‘9F-NMR data (CDC13,188 MHz): 6 -117.2 (dd, Fa,Fe254, Fela 4.8 Hz, F-2e), -119.6 (dddd, Fa,Fe 254, Fa,3 29.3, Fa,la 17.8, Fa,4 12.8 Hz,F-2a).‘3C-NMR data (CDC13, 75 MHz): 6 169.809 (s, C=O), 169.313 (s, C=O),168.965 (s, C=O), 115.504 (t, 2,Fa 249.9, J2Fe 249.9 Hz, C-2), 101.838 (s, C-5),68.673 (dd, 3,Fe 21.2, 3,Fa 18.7 Hz, C-3), 67.893 (d, 4,Fe 7.3 Hz, C-4), 65.038 (dd,l,Fa 36.0, l,Fe 25.0 Hz, C-i), 64.944 (s, C-6), 20.552 (s, CH3), 20.431 (s, CH3),20.358 (s, CH3). DCI-LRMS: 364(C12H507F237C1+NH4); 362(C1257F3C1+ ++ NH4 ); 309 (C12H57F2).3,4,6-Tri-O-acetyl-2-deoxy-2 ,2-dWuoro-a-D-arabinohexopyranose (3.11)OAcAcO F—oAcOFOHHydrazine acetate (100 mg) in DMF (2 mL) was added to a solution of the per-Oacetate 3.8 (0.39 g, 1.1 mmol) dissolved in DMF (10 mL) and stirred at 52°C for 4 days.This mixture was then dissolved in ethyl acetate (40 mL), washed twice with doublydeionized water, dried (MgSO4), and filtered. DMF was removed by successive coevaporations with toluene. Column chromatography (solvent E) yielded the hemiacetal3.11 (0.079 g, 0.24 mmol, 23%). ‘H-NMR data (CDC13,200 MHz): 6 5.56 (ddd, 1H,3,Fa 19, J3.4 10,3,Fe 6.0 Hz, H-3), 5.24 - 5.10 (m, 2 H, H-i, H-4), 4.87 (br ci, 1 H,l,OH 3 Hz, 1-OH), 4.32 - 4.00 (m, 3 H, H-5, H-6, H-6’), 2.09 (s, 3 H, OAc), 2.06 (s, 3H, OAc), 2.00 (s, 3 H, OAc). ‘9F-NMR data (CDC13, 188 MHz): 8 120.7 (dd, Fe,Fa251, Fe3 6.0 Hz, F-2e), -122.7 (ddd, Fa,FC 251, Fa,3 19, Fa,l 5.0 Hz, F-2a).1673,4,6-Tri-O-acetyl-2 -deoxy-2 ,2 -dWuoro- a-D-arabinohexopyranosyl chloride (3.9)(from 3.11 using SOd2)OAcThe hemiacetal 3.11 (0.079 g, 0.23 mmol) was dissolved in freshly distilledthionyl chloride (1 mL, 11 mmol) and stirred at reflux (— 65°C) for 80 hours. Columnchromatography (solvent H) yielded the acetylated glycosyl chloride 3.9 (0.049 g, 0.14mmol, 62%). 1H-NMR data (CDC13,200 MHz): 6 5.92 (dd, 1 H, lFa 4.8, l,Fe 2.0Hz, H4), 5.68 (dt, 1 H, 3Fa 15, = 3,Fe 10 Hz, H-3), 5.19 (t, 1 H, J45 = J43 10Hz, H-4), 4.2 - 4.0 (m, 3 H, H-5, H-6, H-6’), 2.40 (s, 3 H, OAc), 2.36 (s, 3 H, OAc),2.32 (s, 3 H, OAc).‘9F-NMR data (CDC13, 188 MHz): 6 115.5 (ddd, Fa,Fe 249.8,Fa,3 15, Fa,l 4.8 Hz, F-2a), -116.9 (ddd, Fa,Fe 249.8, Fe,3 10 Hz, F-2e). Anal. Calcd.forC12H50721:C, 41.81; H, 4.39; Found: C, 41.76; H, 4.50.2 -Deoxy-2 ,2 -dfluoro- a-D-arabinohexopyranosyl chloride (3.3)The acetylated glycosyl chloride 3.9 (0.049 g, 0.14 mmol) was dissolved inmethanol (3 mL) and deacetylated according to the general procedure for 5.5 h. Theresulting light orange oil was purified by column chromatography (solvent J) to afford 3.3as a colourless oil (0.016 g, 0.085 mmol, 83%). 1H-NMR data (D20, 200 MHz): 86.21 (d, 1 H, lFa 6.0 Hz, H-i), 4.25 (ddd, 1 H, 3,Fa 21.0, .13,4 9.5,3,Fe 5.0 Hz, H-3),3.92-3.62 (m, 4 H, H-4, H-5, H-6, H-6’). ‘9F-NMR data (D20, 188 MHz): 6 -116.1168(dd, Fe,Fa 248.7, Fe,3 5.0 Hz, F-2e), -119.45 (ddd, Fa,Fe 248.7, Fa,3 21.0, Fa,l 6.0,Hz, F-2a). Anal. Calcd. forC6H9O4F21:C, 32.97; H, 4.15; Found: C, 33.15; H, Enzyme Kinetics5.2.1 General Methods and MaterialsBuffer chemicals and reagents for kinetic measurements were obtained from theBDH, Aldrich or Sigma chemical companies. 2-Deoxy-2-fluoro-3-g1ucopyranosy1 fluoride(2Ff3G1uF) and 2-deoxy-2-fluoro--mannosy1 fluoride (2FfManF) were synthesized asreported previously (Hall et al., 1971; Shelling et al., 1984; Withers et al., 1988) and hadsatisfactory 1H and ‘9F NMR spectra and elemental analysis. All kinetic studies werecarried out on a Pye Unicam PU8800 UV/VIS spectrophotometer equipped with aNESLAB RTE-210 circulating water bath. Quartz or plastic cuvettes with a pathlength of 1cm path length were used. Measurements were taken at 400 nm, the wavelength ofmaximal absorbance of 4-nitrophenol or 2,4-dinitrophenol. Because the intact nitrophenylglycopyranosides do not absorb at 400 nm, the rate of change of absorbance (iA4/min)is proportional to the rate of release of nitrophenol. Enzyme concentrations and reactiontimes were chosen so that less than 10% of the total substrate was hydrolyzed to ensurelinear kinetics.Yeast ct-glucosidase (EC Type III from yeast) and Jack Bean (Xmannosidase (EC were obtained from the Sigma Chemical Co. ClonedAgrobacterium faecalis 3-glucosidase was prepared as described (Kempton & Withers,1992). A commercial preparation of human GCase (Genzyme, Framingham, MD, USA)obtained through Dr. Marie Grace, Mt. Sinai Medical Centre, New York, was used forthese studies. Samples of GCase employed in kinetic studies contained human serumalbumin, whereas those used in peptide sequencing did not. Cloned human lysosomal 3-galactosidase in 10 mM phosphate buffered saline, pH 7.4, was obtained from Dr. John169Callahan, Hospital for Sick Children, Toronto. Kinetic studies with the above enzymeswere performed in the following buffers at the temperatures indicated: yeast a-glucosidase0(50 mM phosphate buffer, pH 6.8, containing 0.1% BSA at 37 C); Jack bean amannosidase (14.5 mM citrate! 71 mM phosphate buffer, pH 6.5, containing 1 mM ZnSO4at 25°C); Agrobacterium faecalis f3-glucosidase (50 mM phosphate buffer, pH 6.8,containing 0.1% BSA at 37°C or 5°C); human GCase (20 mM citrate/60 mM phosphatebuffer, pH 5.5, containing 4 mM [-mercaptoethanol, 1 mM EDTA, 0.25% (v/v) Triton X100, and 0.25% taurocholate at 37°C); human lysosomal 3-galactosidase (50 mM citrate,100 mM phosphate, pH 4.3 at 37°C). The appropriate 4-nitrophenyl or 2,4-dinitrophenylglycosides were used to assay the various glycosidases in the buffers and at thetemperatures indicated.5.2.2 Time-Dependent InactivationThe kinetic parameters for the inactivation of the various glycosidases by differentinhibitors were determined as follows. The enzyme was incubated in the appropriate bufferat the appropriate temperature in the presence of various concentrations of the inhibitor.Aliquots (5 or 10 iL) of these inactivation mixtures were removed at time intervals anddiluted into assay cells containing a large volume (— 1 mL) of substrate (at saturatingconcentrations, 7 x Km. or at least - Km). This effectively stops the inactivation both bydilution of the inactivator and by competition with an excess of substrate. The residualenzymatic activity was determined from the rate of hydrolysis of the substrate, which isdirectly proportional to the amount of active enzyme. The process was monitored until 80 -90% of the enzymatic activity was inactivated. Pseudo-first order rate constants (kObs) foreach inactivator concentration were calculated from the slopes of the plots of naturallogarithm of the residual enzymatic activity versus time or by fitting plots of the residualactivity versus time to a single exponential equation using GraFit (Leatherbarrow, 1990).If inactivation did not go to completion and a significant steady-state rate was observed,170these data were fitted to a single exponential equation plus offset using GraFit. For theinactivation of 3-glucosidase with 5FaIdoF or 5FGluF, residual activity was assayed at18°C with j3PNPFuc to minimize reactivation in the assay cell over the course of themeasurement. Values of k and K1 were determined from these kobs values by fitting to theequationkobs = icm (5.1)K1 + [I]If, however, due to very large values of Ki, rapidity of inactivation, paucity or limitedsolubility of inactivator, saturation was not observed, ki/Ki may be calculated according tokOb= iLil (5.2)K1whereK1>> [I]. In this case, ki/Ki is given by the slope of the plot of kobs versus [II,which was determined by linear regression using GraFit, as above.5.2.3 Protection Against InactivationProtection against inactivation was investigated as follows. Samples of enzymewere incubated in the appropriate buffer containing the inactivator and in the absence orpresence of a competitive inhibitor (at a concentration of — Ki or higher). Aliquots wereremoved at various time intervals, diluted into assay cells containing saturatingconcentrations of substrate and the residual enzyme activity monitored by following therelease of the nitrophenolate at 400 nm as described above. Pseudo-first order rateconstants for inactivation at the same inactivator concentration, but in the absence orpresence of the competitive inhibitor, were determined.5.2.4 Reactivation of Inactivated Enzyme171Reactivations of inactivated enzymes were studied as follows. An appropriatedilution of the inactivated enzyme (— 400 pL) was concentrated at 4°C using 10 kDanominal cut-off centrifugal concentrators (Amicon Corp., Danvers, MD) to a volume ofapproximately 50 iL, then diluted with 400 pL of buffer. This was repeated twice, and theretentate was diluted to a fmal volume of buffer containing 1 mg/mL BSA, and appropriatetransglycosylation ligands, if any. The inactivated enzyme was then incubated at 37°C andreactivation was monitored by removal of aliquots (5 or 10 iL) at appropriate time intervalsand assaying as described above. Measured activities were coffected for decreases inactivity due to denaturation over this time course using data for non-inhibited controlsamples. The spontaneous reactivation rate constant, k3, was determined by fitting the datato a first order rate equation, as described above.5.2.5 Test for Inhibitory Contaminant in 5FIdoF PreparationYeast ct-glucosidase (3.75 mg/mL, 50 IlL, 2.75 nmol) was incubated with5F3IdoF (0.31 mM, 31 nmol) in a total volume of 100 tL for 25 mm at 37°C. The molarratio of inhibitor/enzyme is thus — 11/1. The inactivated enzyme was removed byultracenthfugation with 10 kDa nominal cut-off centrifugal concentrators (Amicon Corp.,Danvers, MD) for 30 mm, and this enzyme-treated filtrate was re-evaluated as an inhibitor,by adding an appropriate dilution of fresh enzyme, removing aliquots and assaying withaPNPGIu. A kobs value was derived by fitting the residual activity versus time to anexponential decay, and compared to that determined without pre-treatment with enzyme.5.2.6 Test for Inhibitory Contaminant in 5FcxIdoF Preparationf-Glucosidase (8.8 mg/mL, 40 IlL, 7.0 nmol) was incubated with 5FcxIdoF (95I.LM, 9.5 nmol) in a total volume of 100 pL at 37°C. The molar ratio of inhibitor/enzyme isthus — 1.3/1. Aliquots were removed at intervals, diluted 1000-fold and assayed withI3PNPG1u. The level of inhibition after 15 minutes incubation was compared to that172previously determined after the same incubation time with the same concentration ofSFcxIdoF, but with a much larger excess of inhibitor.5.2.7 Determination of Apparent K1’ ValuesAn apparent K1’ value for 5RxG1uF with yeast c-glucosidase under steady-stateconditions was detemined by assaying cz-glucosidase with zPNPG1u in the presence ofvarious concentrations of 5FctGluF. Initially, a Vm and Km determination for thisparticular dilution of enzyme was done. An appropriate dilution of enzyme (5 .iL) wasadded to cells (1 mL) each containing aPNPG1u (0.1 mM) and various concentrations of5FaG1uF (1.98 to 39.6 .iM), and the steady-state enzymic rates were determined bymonitoring release of nitrophenolate. A plot of 1/V versus [5FxGluF] intersects a linegiven by 1/Vm at - K. An analogous value for 5F3GluF with Agrobacterium -glucosidase was similarly determined by Ms. Karen Rupitz by assaying the steady stateactivity of the enzyme with [WNPFuc (0.13 mM) in the presence of various concentrationsof the inhibitor (2.28 to 16.0 j.tM). Lower concentrations of 5FcxGluF or 5Ff3GluF couldnot be employed since substrate depletion occurred prior to attainment of a steady-state.5.2.8 Fluoride Electrode KineticsThe kcat value for 5FcGluF was determined by monitoring the release of fluorideusing an Orion 96-09 combination fluoride ion electrode. Stock enzyme (10 pL) wasadded to glass cells containing various concentrations of 5FctGluF in 50 mM sodiumphosphate buffer, pH 6.8 in a final volume of 250 .tL, incubated at 37°C. As initial ratesof fluoride release at all of the concentrations assayed were identical, this rate was taken asVm, andk was calculated according to= V/[E0J (5.3)173Rates at lower 5FccGluF concentrations could not be investigated because of theinsensitivity of the fluoride electrode at the low concentrations required. The calculatedis expressed as substrate hydrolyzed, and is one half that initially measured as fluoridereleased since two equivalents of fluoride are produced upon hydrolysis.The kinetic parameters of 2FaManF with Jack bean cz-mannosidase were similariydetermined, except that the buffer was 14.5 mM citrate! 71 mM phosphate buffer with 1mM ZnSO4,pH 6.5 and the assays were conducted at 25.0°C. Initial rates of fluoriderelease were measured, spontaneous hydrolysis rates subtracted, and the kinetic parametersk and Km determined using GraFit, as described above.5.3 Determination of the Stereochemical Course of Glycoside Hydrolysisby ‘H NMR SpectroscopyThis experiment was carried out with human lysosomal f-galactosidase and withJack Bean a-mannosidase. Solutions of the substrates (lDNPGa1 and ctPNPMan,respectively) were dissolved in the appropriate buffers and freeze-dried twice fromD20 tominimize ‘H signals from the solvent. The enzymes were transferred to the appropriatebuffers containing D20 by ultrafiltration using 10 kDa nominal cut-off centrifugalconcentrators (Amicon Corp., Danvers, MD).For the f-galactosidase, ‘H-NMR spectra were recorded by Dr. LawrenceMcintosh on a Varian Unity 500 MHz spectrometer at 30°C using a 5 mm sample tube.The reaction mixture, in a total volume of 0.6 mL, contained 1.8 mlvi [DNPGa1. Afteracquisition of the initial spectrum, enzyme (10 jiL) was added in sufficient quantity toensure complete hydrolysis in one hour. Residual H20 was supressed by weakpresaturation during the 2 s recycle delay. The stereochemical course of the enzyme-catalyzed hydrolysis was monitored by collecting spectra at intervals to first detect the initialproduct, and then observe the formation of the other anomer by mutarotation.174For the -mannosidase, specira were recorded by Mr. Curtis Braun on a BrukerWH-400 spectrometer at ambient temperature. The reaction mixture, in a total volume of0.5 mL, contained 5 mM xPNPMan. This solution did not contain 1 mM ZnSO4(although ZnSO4at this concentration was present in the buffer of the enzyme aliquot (10p.L) added subsequently) since substantial amounts of this metal ion were found to catalyzethe mutarotation of the initially released mannose product, resulting in a rapidly equilibratedmixture. After addition of enzyme, the stereochemical course of hydrolysis was monitoredas described previously.5.4 Spontaneous Hydrolysis of Glycosyl FluoridesSolutions of the parent, 2-deoxy-2-fluoro-, and 5-fluoro glycosyl fluorides (— 10mM) were prepared in 50 mM phosphate buffer, pH 6.8, containing 1 M NaClO4.Samples (1 mL) were incubated at 50.0°C in parafilmed 1.5 mL screw-top plastic vialswith 0-ring seals, and aliquots (50 or 100 L) were removed at appropriate time intervalsand diluted 4- or 2-fold into the same buffer, frozen immediately, and the fluorideconcentrations determined at the conclusion of the experiment. In cases for whichhydrolysis was sufficiently rapid to allow monitoring of the reaction for at least four or fivehalf-lives, rates of fluoride release were determined by plotting the concentration of fluorideversus time and fitting the data to a first order equation using GraFit, as describedpreviously. If the reaction was slow, so that only partial hydrolysis was observed over theassay period (—7 days), the rate constant was determined by a linear fit of the initial rate ofhydrolysis in which <5% of the glycosyl fluoride had been consumed. Values of k referto glycosyl fluoride consumed. Thus for initial rate determinations of the 2-deoxy-2-fluorofluorides, k = rate/[F]; for the 5-fluoro fluorides, k = 1/2(rate/[F]).5.5 Inactivation of Glycosidases for Proteolysis and Electrospray MSAnalysis175Agrobacteriwn faecalis [-glucosidase (40 iLL, 2.2 mg/mL, in 50 mM phosphatebuffer, pH 6.8) was incubated with 5FGluF (10 ilL, 16.2 mM) or with 5FaIdoF (40 ilL,15.8 mM) at 37°C for 30 mm.Dialyzed yeast x-glucosidase (25 IlL, 10 mg/mL, in 50 mM phosphate buffer, pH6.8), was incubated with 5FI3IdoF (5 IlL, 52 mlvi) at 37°C for one hour. Completeinactivation was confirmed by enzyme assay. Labelling with 5FctGluF was accomplishedby incubation of cz-glucosidase (25 IlL, 10 mg/mL, in 50 mlvi phosphate buffer, pH 6.8)with 5FaGluF (50 ilL, 4.0 mM in 50 mM phosphate buffer) for 10 to 15 s. This mixturewas immediately digested with pepsin as described below.GCase (1 mL, 0.49 mg/mL, in 40 mlvi phosphate/citrate pH 5.5 buffer, 0.9%NaCl, 0.3% mannitol, and 0.005% Tween-80), purified as described above, was incubatedin the presence of 3.45 mlvi 2FG1uF at 37°C for 16 hours. Complete inactivation wasconfirmed by enzyme assay.Lysosomal f-galactosidase (50 p.L, 5 mg/mL, 50 mM citrate, 100 mM phosphate,pH 4.3) was incubated with 2FDNPGa1 (1.58 mM, 10 p.L) at 37°C for 15 mm.5.6 ProteolysisTo 13-glucosidase (100 IlL, 2.2 mg/mL, native, or 5FGlu- or 5FIdo-labelled) wasadded 50 mM phosphate buffer pH 2 (100 iiL) and pepsin (100 p.L, 0.1 mg/mL in 50 mMpH 2 phosphate buffer). The mixture was incubated at 37°C for 30 mm, then frozen andanalyzed immediately by ES/MS upon thawing.a-Glucosidase (30 jiL, 10 mg/mL, native or 5FIdo-labelled) was mixed with 50mlvi phosphate buffer pH 2 (60 ilL) and pepsin (30 I.LL, 1 mg/mL in 50 mM pH 2phosphate buffer). To the 5FG1u-labelled enzyme (75 i.IL, prepared as described above)was immediately added 50 mlvi phosphate buffer, pH 2 (125 IlL) and pepsin (25 pE, Img/mL in 50 mlvi pH 2 phosphate buffer. Both mixtures were incubated at roomtemperature for 15 mm, then frozen and analyzed immediately by ES/MS upon thawing.176GCase (10 i.iL, native or inactivated, 5 mg/mL) was mixed with 50 mM phosphatebuffer pH 2 (20 I.IL) and pepsin (10 p.L, 0.5 mg/mL in 50 mM pH 2 phosphate buffer).The mixture was incubated at room temperature until digestion was complete(approximately 40 hours), as indicated by SDS-PAGE eletrophoresis.Lysosomal f3-galactosidase (native or inactivated, 60 .tL, 5 mg/mL) was incubatedwith trichioroacetic acid (120 iL, 100 mM) in pH 2 buffer for 5 mm at room temperature.Pepsin (60 iiL, 0.5 mg/mL) was added, and the mixture was incubated at room temperaturefor 12 h. For further tryptic proteolysis of the peptic peptides, solid ammoniumbicarbonate (10 mg) was added to 200 p.L of this digest, and trypsin (Sigma, Type XIII,TPCK-treated, 5 .tL, 5 mg/mL in 50 mlvi phosphate buffer, pH 6.8) was immediatelyadded. The digest was incubated at room temperature for 15 mm, then re-acidified with50% trifluoroacetic acid (TFA) (20 pL).5.7 Electrospray MS conditions for Peptide AnalysisMass spectra were recorded on a PE-Sciex API III triple quadrupole massspectrometer (Sciex, Thornhill, Ont., Canada) equipped with an ionspray ion source.Peptides were separated by reverse phase HPLC on an Ultrafast Microprotein Analyzer(Michrom BioResources Inc., Pleasanton, CA) directly interfaced with the massspectrometer. In each of the MS experiments, the proteolytic digest was loaded onto a Cl 8column (Reliasil, 1 x 150 mm), then eluted with a gradient of 0-60% HPLC solvent B over20 minutes followed by 100% HPLC solvent B over 2 minutes at a flow rate of 50mi/minute (HPLC solvent A: 0.05% TFA, 2% acetonitrile in water; HPLC solvent B:0.045% TFA, 80% acetonitrile in water). A post-column splitter was present in allexperiments, splitting off 85% of the sample into a fraction collector and sending 15% intothe mass spectrometer. Spectra were obtained in either the single-quadrupole scan mode(LC/MS), the tandem MS neutral loss mode, or the tandem MS daughter scan mode(LC/MSIMS).177In the single quadrupole mode (LCIMS) the quadrupole mass analyzer was scannedover a m/z range 300-2400 Da with a step size of 0.5 Da and a dwell time of 1 ms per step.The ion source voltage (ISV) was set at 5 kV and the orifice energy (OR) was 80 V.In the neutral loss scanning mode, MS/MS spectra were obtained by searching forthe mass loss of mlz 165 or 181, corresponding to the losses of 2-deoxy-2-fluoro-glycosylor 5-fluoro-glycosyl moieties from a peptide ion in the singly charged state. To detectlosses of the labels from doubly-charged peptides, the mass loss of one-half the mass ofthe label used (m/z 82.5 or 90.5, respectively) was searched for. Thus, scan range: nVz300-2400; step size: 0.5; dwell time: 1 ms/step; ion source voltage (ISV): 5 kV; orificeenergy (OR): 80; RE1 = 115; DM1 = 0.16; Ri = 0 V; R2 = -50 V; RE3 = 115; DM3 =0.16; Coffision gas thickness: 3.2-3.6 x IO’4 molecules/cm2 (CGT 320-360). Tomaximize the sensitivity of neutral loss detection, normally the resolution (RE and DM) iscompromised without generating artifact neutral loss peaks.In the tandem MS daughter scan mode, the spectrum was obtained by selectivelyintroducing the m/z 688 peptide from the first quadrupole (Qi) into the coffision cell (Q2)and observing the daughter ions in the third quadrupole (Q3). Thus, Qi was locked on m/z688; Q3 scan range: 200-800; Step size: 0.5; Dwell time: 1 ms; Ion source voltage (ISV): 5kV;Orificeenergy(OR):80;RE1=112;DM1=0.i8;R1=OV;R2= -50 V;RE3 = 112;DM3 0.18; Collision gas thickness: 4.5 x 1O’4molecules/cm2(CGT = 450).5.8 Isolation of the 2FGIu-labelled Peptide of GCaseThe (3Case proteolytic digest was separated by reverse phase HPLC under theabove conditions and the labelled active-site peptide was collected via the post-column flowsplitter by following the neutral loss signal in the mass spectrometer as described. Thispeptide was further purified on the same HPLC C18 column by using a shallower gradientprofile (0-15% HPLC solvent B over 50 mm), using the mass spectrometer as detector.1785.9 Chemical Sequencing of 2FGIu-labelled Peptide of GCaseThe amino acid sequence of the 2FGlu-labelled peptide of GCase was determinedby Dr. Hans Nika using standard pulsed liquid phase protocols and instrumentation on aModel 477A sequencer, Model 120A PTH analyser (Applied Biosystems, Foster City, CA)as described elsewhere (Wang et al., 1993).5.10 AminolysisTypically, aminolysis of labelled peptides was carried out with 15 j.iL of the labelledpeptide digest (-4 mg/mL), to which was added 30% ammonium hydroxide (7.5 .tL). Themixture was incubated at 50°C for 15 mm, re-acidified with 50% TFA (5 p.L), and resubjected to ES/MS.5.11 Stoichiometry of Inactivation by Electrospray MSThe stoichiometry of Agrobacterium faecalis 3-g1ucosidase inactivation by5FGluF or 5FaIdoF was determined by subjecting samples of inactivated enzyme anduntreated enzyme to mass spectrometric analysis. This involved introduction of the proteinsample into the mass spectrometer through a microbore PLRP column (1 x 50 mm) on theMichmm HPLC system using a gradient of 20-100% HPLC solvent B over 10 minutesfollowed by 100% HPLC solvent B over 2 minutes. The mass spectrometer, in the singlequadrupole mode, was scanned over a m/z range of 300-2400 Da. Protein molecularweights were determined from this data using deconvolution software supplied by Sciex.5.12 Liquid Secondary Ion Mass Spectrometry (LSIMS)Samples for analysis were collected as eluted from the post-column flow splitter, bypooling fractions eluting approximately one minute prior to and one minute following thedetection of the peak. Pooled fractions (- 100 pL) were concentrated on a Speed-Vac to5 or 10 p1 prior to LSIMS analysis. Low and high resolution liquid secondary ionizationmass spectrometry (LSIMS) was perfomed on a Kratos Concept II HQ mass spectrometer179using a thioglycerol matrix, 8 kV ion source, and a 12 kV LSIMS gun using Cs ions. Themass reference for high resolution spectra was potassium iodide/polyethylene glycol(KIPEG).5.13 Preparation of Tissue HomogenatesAn appropriate concentration of. the 2-deoxy-2-fluoro glycosyl fluoride (10 mg/kg)in 0.9% saline was administered by lateral tail vein injection in a single dose to 200-250 gmale Wistar rats after anesthesia with phenobarbital (65 mg/kg, intraperitoneal). With theassistance of Dr. Neil Hartrnan or Ms. Kelly Hewitt, and following an approved protocolof the animal care facility at the University of British Columbia, the animals were sacrificedby carbon dioxide asphyxiation at 1, 20 and 48 hours after injection. The liver, spleen,kidney and brain were removed and portions weighing — 1 g were rinsed in distilled water,blotted and weighed. To each sample was added 2.5 mL of 25 mM citrate/50 mMphosphate buffer, pH 5.0 containing EDTA (1 mM), dithiothreitol (4 mM), and Triton X100 (0.1% v/v), and the tissues were treated for one minute with a mechanical tissuehomogenizer. Control tissues were treated identically, but the animals received only 0.9%saline.5.14 Assay of Tissue -GIycosidase Activityf3-Glucosidase, -mannosidase and 3-galactosidase activities in the various tissuehomogenates were determined essentially as described previously (Van Diggelen et a!.,1980; Van Hoof & Hers, 1968) with slight modification, using the appropriate PNP f3-D-glycopyranosides. Aliquots (0.0125 or 0.025 mL) of the tissue homogenates preparedabove were added to plastic Eppendorf vials containing 0.1 or 0.2 mL of the appropriatePNP glycoside (5 mM) in the buffer described above. The vials were capped, mixed andincubated at 37°C for 30 minutes, 1.0 mL of 200 mM sodium carbonate buffer (pH 10.7)was added and the vials were centrifuged at 15 000 x g at 4°C for 15 minutes. The180supernatant was removed by pipette and its absorbance at 400 nm determined immediately.Incubation times and substrate concentrations were chosen so that less than 10% of thesubstrate was hydrolyzed. Measured absorbances were corrected for spontaneoushydrolysis of the PNP glycoside and the background absorbance of the tissue supernatant.f3-Glycosidase activity is expressed as nmol PNP glycoside hydrolyzed/h/mg tissue (E =1.82 x IO4M1cmfor p-nitrophenol at pH 10.7). 3-Glycosidase assays were performedon fresh tissue samples (kept on ice prior to assay). For each tissue, results of duplicateassays on three rats (for a total of six determinations) were averaged; standard deviationsare denoted by error bars.5.15 In Vitro Reactivation of Tissue 3-GIycosidase ActivityTissue homogenates showing maximum -glycosidase inhibition (from animalssacrificed one hour after injection of inhibitor) were frozen and stored at -20°C. For eachtissue, pooled samples (0.30 or 0.75 mL) were diluted two-fold with buffer and incubatedat 37°C in the presence or absence of appropriate transglycosylation ligands (20 mMflPTGlu, 20 mM G1uNAc or — 2 mM ceramides, the maximal concentration of the lattercompound attainable). Aliquots (0.0 125 or 0.025 mL) were removed periodically andassayed as described above. f-Glycosidase activities were corrected for decreases inactivity over this time due to denaturation using data for non-inhibited control samples ofthe appropriate tissue homogenate.5.16 ‘9F NMR Spectroscopy of 2-Deoxy-2-Fluoro-f3-D-MannopyranosylFluoride in Rat Tissue ExtractsUnlabelled 2F[ManF at a dosage of 70 mg/kg in 0.9% saline was administered to a200-250 g male Wistar rat by lateral tail vein injection after anesthesia with phenobarbital(65 mg/kg, intraperitoneal). The animal was sacrificed after 60 minutes, and the organswere removed and homogenized as described previously. Brain, liver and kidneyhomogenates were centrifuged (15 000 x g, 15 mm, 4°C) to remove cell debris, and the181supernatants were filtered by passage through a 0.4 tM filter, and freeze-dried. Urine wasremoved by bladder puncture and freeze-dried. The residues were dissolved in D20 and19F NMR spectra were acquired. Data was acquired over 150 000 scans for the brain, liverand kidney samples, and over 256 scans for the urine sample. Acquisition times were0.328 and 0.655 s, respectively.185.17 Radiosynthesis of 2-Deoxy-2-[ F]FIuoro-3-D-MannosyI18[ F]FluorjdeUsing facilities at TRIUMF, [‘8F]-F2(0.1% in neon) was bubbled at a flow rate of150 mL/min through a solution of D-glucal (—20mg) in acetonitrile (10 mL) for — 20 mm.The reaction vessel was a large test tube with side-arm (sealed with a septum) attached to arotary evaporator. Fluorine was introduced via 1.6 mm Teflon tubing running down theinterior of the cooling jacket. A second section of Teflon tubing (3 mm) also running downthe cooling jacket connected the reaction vessel to the top of a 7.5 x 150 mm glass columnof silica gel. A vent was provided into a tube of soda lime. Upon completion of thefluorination, the tubing was withdrawn, the vent sealed, and the resulting 2:1 mixture of 2-deoxy-2-fluoro-a-glucopyranosyl- and --mannopyranosyl fluorides was evaporated todryness under reduced pressure. Thethyl ether (not anhydrous) was used to wash thereaction vessel and elute the column. 2 x 1.5 mL of this solvent was first introduced viasyringe through the septum, and the contents of the reaction vessel were then transferred tothe top of the silica column. The transfer was accomplished by creating a partial vacuumwith a 50 mL syringe at the column head to draw over the contents of the reaction vessel.After loading in this way, eluant was gently forced through the column using positivepressure on the same syringe. The 2-deoxy-2-[’8Fjfluoro cx-glucosyl [‘8F]fluoridetypically eluted in the first 30 mL, followed by the 2-deoxy-2-[1FJfluoro--D-mannosyl[18F]fluoride, typically in the subsequent 25 mL. The fraction containing the 3-mannosy1fluoride, collected in a small wide-mouth vial, was evaporated in a 40°C oil bath with the182aid of a stream of air, then dissolved in either 50 mM phosphate buffer (for in vitro studies)or sterile 0.9% saline (for animal studies). The overall time from end of bombardment(EOB) for synthesis and purification was — 40 mm and the specific activity was — 500mCi/mmol. The entire procedure was conducted behind 50 mm blocks of lead, and alloperations were done as remotely as possible, after first determining reaction times, columnsize and the fractions in which the products eluted in non-radioactive trials.5.18 Labelling of Agrobacterium faecalis 3-gIucosidase withTo the chromatographed 2-deoxy-2-[’8F]fluoro-f-D-mannosy1 [‘8F]fluoridefraction prepared above (— 2 - 2.5 mg) was added degassed aqueous 50 mM phosphatebuffer, pH 6.8 (5 mL). An aliquot (100 p.L) was added to an equal volume ofAgrobacteriumfaecalis 3-glucosidase solution (2.7 mg/mL), and the mixture was incubatedfor 15 mm at 37°C. A portion (100 .iL) was injected onto an HPLC (Waters) equippedwith a W-Porex GP-300 gel permeation column (4.6mm x 250 mm) (Phenomenex, Inc.).The eluant was the same buffer described above, at a flow rate of 1.0 mL/min. UV-activecomponents were detected at 214 nm and radioactivity was detected with an on-line NaI(Tl)scintillation detector.5.19 Reactivation of the2-Deoxy-2-[18F]Fluoro-a-D-Mannosyl-EnzymeLabelled enzyme, freed of excess inhibitor, prepared as above was collected aseluted from the column in a total volume of - 1 mL. This was concentrated to a volume of— 100 p.L using 10 kDa nominal cut-off centrifugal concentrators (Amicon Corp., Danvers,MD). The sample was divided in half. To one aliquot was added 50 p.L buffer. The otherwas diluted with an equal volume of glucosyl benzene in buffer to a fmal glucosyl benzeneconcentration of 197 mM. Both samples were incubated at 37°C, and aliquots (15 pL)were removed at appropriate time intervals and injected onto the HPLC column. Fractions,corresponding to the previously determined elution times of the labelled enzyme and a 2-183deoxy-2-fluoro-mannose standard, were collected and counted using a Beckman Gamma8000 well counter. Reactivation of the radiolabelled enzyme was monitored for — 4 hours.The % labelled enzyme remaining was calculated by dividing the labelled enzyme activityby the total radioactivity eluted. All measurements were corrected for activity in blanks and18for decay of F (t112 = 110 nun).5.20 Biodistribution of 2-Deoxy-2-[’8F]F1uoro-3-D-Mannosy1[18F]Fluoride in RatsMale 200 - 250 g Wistar rats (anesthetized with 65 mg/lcg phenobarbital,intraperitoneal) were administered — 0.05 - 0.1 mCi 2-deoxy-2-[18F]fluoro-3-D-mannosy1[18F]fluoride by lateral tail vein injection in a single dose. 5 - 6 animals were sacrificed bycarbon dioxide asphyxiation at 2, 15, 30, 60 and 120 mm post-injection. The organs wereremoved, blotted free of blood, weighed in pre-weighed counting vials, and tissueconcentrations of injected radiolabel were determined by scintillation counting on aBeckman Gamma 8000 well counter. The % injected dose! g tissue (%ID/g) is calculatedaccording to% ID!g = Tissue cpm/(Standard cpm x Dilution’) x 100 (5.4)g TissueTypically, the standard is diluted 50 - 500-fold prior to counting. All measurements werecorrected for activity in blanks and for decay of ‘8F over the course of the counting of thesamples.5.21 Pre-treatment of Rats with 2-Deoxy-2-Fluoro-3-D-MannosyI Fluoride(2Fj3ManF)An appropriate concentration of 2-deoxy-2-fluoro 3-D-mannopyranosy1 fluoride(20 mgfkg) in 0.9% saline was administered by lateral tail vein injection in a single dose tothree 200 - 250 g female Sprague-Dawley rats after anesthesia with phenobarbital (65184mg/kg, intraperitoneal). Three animals received only 0.9% saline, and served as controls.After 60 mm, each of the animals was intravenously injected with — 0.05 mCi 2-deoxy-2-[‘8F]fluoro--D-mannosyl[‘8F]fluoride in 0.9% saline. After a further 60 mm, all of theanimals were sacrificed, and the organs were removed, weighed and counted as describedabove.5.22 PET Imaging2-Deoxy-2-[’8F]fluoro-I-D-mannosyl[18F}fluoride (0.95 mCi in 0.9 mL 0.9%saline), prepared as described previously, was injected intravenously to an anesthetized 400- 450 g male Wistar rat. PET scans were made continuously from the time of radioisotopeinjection. The instrument used for these studies was a Siemens/CT! ECAT 953B (Spinkset at., 1992), operated by personnel from the PET imaging team at Vancouver Hospital,UBC site. The interslice separation is 3.375 mm and the slice thickness is approximately4.5 mm. The radial resolution is 5.0 mm in the centre, 6.0 mm at 4.5 cm, and 7.3 mm at9.0 cm off-centre.185APPENDIX IGRAPHICAL REPRESENTATIONOF SPONTANEOUS HYDROLYSIS DATA1860.00120.0010.00080.00060.010.0080.0060.0040.0020Time (h)Time (h)I,.-0 20 40 60 80 100 120 140 160 180 200Figure All. Spontaneous hydrolysis of 2FaGluF (10.7 mM) at 50.0°C, pH 6.8.11111111111111 ‘I • I ‘I. — -ItI,ItIIIIIIIIIIIII•I10 20 40 60 80 100 120 140 160 180 200Figure AJ2. Spontaneous hydrolysis of 2FJ3G1uF (924 mM) at 50.0°C, pH 6.8.187I I I I I I’I I I I I I I I I0.0160.014 = =0.012- =0.01 =0.008 = =0.0060.004-0.002 r0_II I I i I I I I I I I i I i I I I0 20 40 60 80 100 120 140 160 180Time (h)Figure AI.3. Spontaneous hydrolysis ofSFJ3IdoF (7.65 mM) at 50.0°C, pH 6.8.016_IIiiIIlIIiIIIIIII—0.014 =0.012 =0.01 -—0.008— =0.006 =0.004- =0.002w-—I I i I I I i I i I i I i I i I I0 20 40 60 80 100 120 140 160 180Time (h)Figure AJ.4. Spontaneous hydrolysis of5FcvJdoF (7.90 mM) at 50.0°C, pH 6.8.1880.0010.00080.00060.00040.00020Time (h)0Figure AI.5. Spontaneous hydrolysis of 5FaG1uF (4.96 mM) at 50.0 C, pH 20 40Time (h)Figure AJ.6. Spontaneous hydrolysis of 5FJ3G1uF (8.10 mM) at 50.0°C, pH 6.8.0 20 40 60189REFERENCESAbbit, B., Jones, M. Z., Kasari, T. R., Storts, R. W., Templeton, 3. W., Holland, P. S.,& Castenson, P. E. (1991) J. Am. Vet. Med. Assoc., 198, 109.Angyal, S. J. & Pickles, V. A. (1972) Aust. J. Chem., 25, 1695.Atsumi, S., Nosaka, C., linuma, H., & Umezawa, K. (1992) Arch. Biochem. Biophys.,297, 362.Atsumi, S., Nosaka, C., linuma, H., & Umezawa, K. (1993) Arch. Biochem. Biophys.,304, 302.Augé, J. & Serge, D. (1984) Tetr. Leit., 40, 2101.Banait, N. S. & Jencks, W. P. (1991) J. Am. Chem. Soc., 113, 7958.Barnett, J. E. 0., Holman, G., & Munday, K. (1973) Biochem. .1., 131, 211.Barton, N. W., Brady, R. 0., Dambrosia, J. M., Di Bisceglie, A. M., Doppek, S. H.,Hill, S. C., Mankin, S. H., Murray, G. J., Parker, R. I., Argoff, C. E., Grewal, R. P.,& Yu, K.-T. (1991) New Engi. J. Med., 324, 1464.Barton, N. W., Furbish, F. S., Murray, G. J., Garfield, M., & Brady, R. 0. (1990)Proc. Nati. Acad. Sci. U.S.A. 87, 1913.Bause, E. & Legler, 0. (1974) Hoppe-Seyler’s Z. Physiol. Chem., 355, 438.Bennet, A. & Sinnott, M. L. (1986) J. Am. Chem. Soc., 108, 7287.Bernacki, R. J., Niedbala, M. J., & Korytnyk, W. (1985) Cancer Metastasis Rev., 4,81.Bessell, E. M., Foster, A. B., & Westwood, J. H. (1972) Biochem. J., 128, 199.Black, T., Kiss, L., Tull, D., & Withers, S. G. (1993) Carbohydr. Res., 250, 195.Blake, C. C. F., Johnson, L. N., Mair, G. A., North, A. C. T., Phillips, D. C., &Sarma, V. R. (1967) Proc. R. Soc. Lond. B, 167, 365.Bradley, P. R. & Buncel, E. (1968) Can. J. Chem., 46, 3001.Brady, R. 0., Kanfer, J. N., & Shapiro, D. (1965) J. Blot. Chem., 240, 39.Braun, C. (1995) Ph.D. Thesis. University of B.C.Busch, K. L. & Cooks, R. 0., Tandem Mass Spectrometry. ed. MacLafferty, F.W. 1983,New York: John Wiley & Sons.Campbell, R., Rose, D., Wakarchuk, W., To, R., Sung, W., & Yaguchi, M., Acomparison of the structures ofof the 20 kD xylanasesfrom Trichoder,na harzianum andBacillus circulans, in Proceedings of the 2nd TRJCEL symposium on Trichoderma reesel190cellulases and other hydrolases. Suominen, P. and Reinikainen, T., eds. 1993, Foundationfor Biotechnical and Industrial Fermentation Research: Helsinici, Finland. p. 63.Cavanagh, K. T., Fisher, R. A., legler, G., Herrchen, M., Jones, M. Z., Julich, E.,Sewell-Alger, R. P., Sinnott, M. L., & Willcinson, F. E. (1985) Enzyme, 34, 75.Chirakal, R., Firnau, G., & Garnett, E. (1989) J. Label. Cmpds. Radiopharm., 26, 228.D’Agrosa, R. M., Hubbes, M., Zhang, S., Shankaran, R., & Callahan, J. W. (1992)Biochem. J., 285, 833.Dawson, G. (1982) J. Biol. Chem., 257, 3369.Day, A. G. & Withers, S. G. (1986) Can. J. Biochem., 64, 914.Diksic, M. & Jolly, D. (1985) J. Carb. Chem., 4, 265.Dinur, T., Osiecki, K. M., Legler, G., Gatt, S., Desnick, R. J., & G., G. (1986) Proc.Nat!. Acad. Sci. U.S.A., 83, 1660.Divne, C., Stahlberg, J., Reinikainen, T., Ruohonen, L., Petterson, G., Knowles, J. K.C., Teen, T., & Jones, T. A. (1994) Science, 265, 524.Elbein, A. D., Legler, 0., Tlusty, A., McDowell, W., & Schwarz, R. (1984) Arch.Biochem. Biophys., 235, 579.Ermert, P., Vasella, A., Weber, M., Rupitz, K., & Withers, S. 0. (1993) Carbohydr.Res., 250, 113.Fallet, S., Grace, M. E., Sibille, A., Mendelson, D. S., Shapiro, R. S., Hermann, G., &Grabowski, G. A. (1992) Ped. Res., 31, 496.Farkas, I., Szabó, I. F., & Bognár, R. (1977) Carbohydr. Res., 58, 478.Fernier, R. J. & Tyler, P. (1980) J. C. S. Perkin I, 1528.Fersht, A., Enzyme Structure and Mechanism. 2nd ed. 1985, New York: Freeman. 97.Fersht, A. R. (1987) Biochemistry, 26, 8031.Figueroa, M. L., Rosenbloom, B. E., Kay, A. C., Garver, P., Thurston, D. W., Koziol,3. A., Gelbart, T., & Beutler, E. (1992) New Engi. J. Med., 327, 1632.Fleet, 0. W. 1. (1985) Tetr. Lett., 5073.Franck, R. W. (1992) Bioorg. Chem., 20, 77.Gebler, J. C., Aebersold, R., & Withers, S. 0. (1992) J. Biol. Chem., 267, 11126.Glew, R. H., Gopalan, V., Forsyth, G. W., & Van der Jagt, D. J., f3-Glucosidases:Biochemistry and Molecular Biology. ACS Symposium Series, 1993, Washington, D.C.p. 83.191Glew, R. H., Peters, S. P., & Christopher, A. R. (1976) Biochim. Biophys. Acta, 422,179.Goodman, M. M., Elmaleh, D. R., Kearfott, K. J., Ackerman, R. H., Hoop, B. J.,Brownell, G. L., Alpert, N. M., & Strauss, H. W. (1981) J. Nuci. Med., 22, 138.Grabowski, G. A., Gatt, S., & Horowitz, M. (1990) Biochem. Molec. Biol., 25, 385.Grace, M., Newman, K., Scheinker, V., Berg-Fussman, A., & Grabowski, G. (1994)J. Biol. Chem., 269, 2283.Guthrie, J. P. (1977) J. Am. Chem. Soc., 99, 3391.Halazy, S., Berges, V., Ehrhard, E., & Danzin, C. (1990) Bioorg. Chem., 18, 330.Halazy, S., Danzin, C., Ehrhard, A., & Gerhart, F. (1989) J. Am. Chem. Soc., 111,3484.Hall, L., Johnson, R., Adamson, 3., & Foster, A. (1971) Can. J. Chem., 49, 118.Hara, A. & Radin, N. S. (1979) Biochem. Biophys. Acta, 582, 412.Harris, E. M. S., Aleshin, E. E., Firsov, L. M., & Honzatko, R. B. (1993)Biochemistry, 32, 1618.Hebre, E. J., Genghof, D. S., Stemlicht, H., & Brewer, C. F. (1977) Biochemistry, 16,1780.Henrissat, B. (1991) Biochem. J., 280, 309.Henrissat, B. & Bairoch, A. (1993) Biochem. J., 293, 781.Hermans, M. P. P., Kroos, M. A., van Beeumen, J., Oostra, B. A., & Reuser, A. J. 3.(1991) J. Biol. Chem., 266, 13507.Herrchen, M. & Legler, G. (1984) Eur. J. Biochem., 138, 527.Hosie, L., Marshall, P. J., & Sinnott, M. L. (1984) J. C. S. Perkin Trans. II, 1121.Hosie, L. & Sinnott, M. L. (1985) Biochem. J., 226, 437.Hughes, A. B. & Rudge, A. 3. (1994) Nat. Prod. Rep., 135.Igarashi, K., Honma, T., & Imagawa, T. (1970) J. Org. Chem., 35, 610.Igarashi, K., Honma, T., & Irisawa, J. (1969) Carbohydr. Res., 11, 577.Imoto, T., Johnson, L., North, A., Phillips, D., & Rupley, 3., The Enzymes. Boyer, P.ed. 1972, New York: Academic Press. p. 666.Ishiwata, K., Seki, H., Sasaki, T., Ishii, 5.-I., Nozaki, T., & Senda, M. (1993) Nuci.Med. Biol., 20, 843.192Jacobsen, R. H., Zhang, X.-J., DuBose, R. F., & Matthews, B. W. (1994) Nature,369, 761.Jones, M. Z. & Dawson, G. (1981) J. Biol. Chem., 256, 5185.Kadziola, A., Abe, 3., Svensson, B., & Haser, R. (1994) J. Mol. Biol., 239, 104.Kajimoto, T., Liu, K. K.-C., Pederson, R. L., Zhong, Z., Ichikawa, Y., Porco, J. A. 3.,& Wong, C.-H. (1991) J. Am. Chem. Soc., 113, 6187.Kameda, Y., Asano, N., Yoshilcawa, M., Takeuchi, M., Yamaguchi, T., Matsui, K.,Horii, S., & Fukase, H. (1984) J. Antibiot., 37, 1301.Kanazawa, Y., Momozono, Y., Ishikawa, M., Yamada, T., Yamane, H., Haradahira, T.,Maeda, M., & Kojima, M. (1986) Life Sciences, 39, 737.Kanda, T., Brewer, C. F., Okada, G., & Hehre, E. J. (1986) Biochemistry, 25, 1159.Kanfer, J. N., Legler, G., Sullivan, J., Raghavan, S. S., & Mumford, R. A. (1975)Biochem. Biophys. Res. Com,n., 67, 85.Kanfer, 3. N., Raghaven, S. S., & Mumford, R. A. (1975) Biochim. Biophys. Acta,391, 129.Kanfer, J. N., Stephens, M. C., Singh, H., & Legler, 0. (1982) Prog. CIin. Biol. Res.,95, 627.Keitel, T., Simon, 0., Borriss, R., & Heinemann, U. (1993) Proc. Nail. Acad. Sci.U.S.A., 90, 5287.Kempton, J. B. & Withers, S. 0. (1992) Biochemistry, 31, 9961.Kiely, D. & Fletcher, H. (1969) J. Org. Chem., 34, 1386.Klein, C., Hollender, 3., Bender, H., & Schulz, G. E. (1992) Biochemistry, 31, 8740.Konstantinidis, A. & Sinnott, M. L. (1991) Biochem. J., 279, 587.Koshland, D. E. (1953) Biol. Rev., 28, 416.Lalegerie, P., Legler, G., & Yon, 3. M. (1982) Biochemie, 64, 1977.Leatherbarrow, R. 3., GraFit Version 3.0. 1990, Erithacus Software Ltd.: Staines, U.K.Legler, G. (1990) Adv. Carb. Chem. Biochem., 48, 319.Legler, G. & Bause, E. (1973) Carbohydr. Res., 28, 45.Legler, G. & Bieberich, E. (1988) Arch. Biochem. Biophys., 260, 427.Lehmann, 3. & Reinshagen, H. (1970) Liebigs Ann. Chem., 732, 112.Lehmann, 3. & Schlesselmann, P. (1983) Carbohydr. Res., 113, 93.193Lehmann, J. & Zieger, B. (1977) Carbohydr. Res., 58, 73.MacGregor, R. R., Halidin, C., Fowler, J. S., Wolf, A. P., Amett, C. D., Langstrom,B., & Alexoff, D. (1985) Biochem. Phar,nacol., 34, 3207.Marshall, P. J., Sinnott, M. L., Smith, P. J., & Widdows, D. (1980) J. C. S. Perkin I,366.Martin, J. L., Veluraja, K., Ross, K., Johnson, L. N., Fleet, G. W. J., Ramsden, N. G.,Bruce, I., Orchard, M. G., Oikonomakos, N. G., Papageorgiou, A. C., Leonidas, D. D.,& Tsitoura, H. S. (1991) Biochemistry, 30, 10101.Matsuura, Y., Kusunoki, M., Harada, W., & Kakudo, M. (1984) J. Biochem., 95, 697.McCarter, J., Adam, M., Braun, C., Namchuk, M., Tull, D., & Withers, S. G. (1993)Carbohydr. Res., 249, 77.McCarter, J., Adam, M., & Withers, S. G. (1992) Biochem. J., 286, 721.McCarter, J., Adam, M. J., Hartman, N. G., & Withers, S. G. (1994) Biochem. J.,301, 343.McCarter, J. & Withers, S. G. (1994) Curr. Opin. Struct. Biol., 4, 885.McCarter, J., Adam, M. J., & Withers, S. G. (1992) J. Label. Cmpds. Radiopharm.,31, 1005.Miao, S., McCarter, J., Grace, M., Grabowski, G., Aebersold, R., & Withers, S. G.(1994) J. Biol. Chem., 269, 10975.Mikami, B., Degano, M., Hehre, E., & Sacchettini, J. (1994) Biochemistry, 33,7779.Mooser, G., Hefta, S. A., Paxton, R. I., Shively, I. E., & Lee, T. D. (1991) J. Biol.Chem., 266, 8916.Mosna, G., Fattore, S., Tubiello, G., Brocca, S., Trubia, M., Gianazza, E., Gatti, R.,Danesino, C., Minelli, A., & Piantanida, M. (1992) Hum. Genet., 90, 247.Moult, J., Esgdat, Y., & Sharon, N. (1973) J. Mo!. Biol., 75, 1.Naider, F., Bohak, Z., & Yariv, J. (1972) Biochemistry, 11, 3202.Namchuk, M. (1993) PhD. Thesis. University of B.C.Nelson, C. N. (1979) Carbohydr. Res., 68, 55.Neufeld, E. F. (1991) Ann. Rev. Biochem, 60, 257.Nishimoto, J., Nanba, E., Inui, K., Okada, S., & Suzuki, K. (1991) Am. J. Hum.Genet., 49, 566.O’Neill, R. R., Tokoro, T., Kozak, C. A., & Brady, R. 0. (1989) Proc. Nat!. Acad.Sci. USA, 86, 5049.194Oshima, A., Yoshida, K., Shimmoto, M., Fukuhara, Y., Sakuraba, H., & Suzuki, Y.(1991) Am. J. Hum. Genet., 49, 1091.Osiecki-Newman, K., Legler, G., Grace, M., Dinur, T., Gatt, S., Desnick, R. J., &Grabowski, G. A. (1988) Enzyme, 40, 173.Pauling, L. (1946) Chem. Engng. News, 24, 1375.Peters, S. P., Coyle, P., & Glew, R. H. (1976) Arch. Biochem. Biophys., 175, 569.Phillips, D. C. (1967) Proc. Nail. Acad. Sci. U.S.A., 57, 484.Post, C. & Karplus, M. (1986) .1. Am. Chem. Soc., 108, 1317.Praly, J.-P., Brard, L., Descotes, G., & Toupet, L. (1989) Tetr., 45, 4141.Praly, J. P. & Descotes, G. (1987) Tetr. Lett., 28, 1405.Qian, M., Haser, R., Buisson, 0., Duee, E., & Payan, F. (1994) Biochemistry, 33,6284.Qian, M., Haser, R.,.& Payan, F. (1993) .1. Mol. Biol., 231, 785.Quaroni, A. & Semenza, 0. (1976) J. Biol. Chem., 251, 3250.Quiocho, F. A., Wilson, D. K., & Vyas, N. K. (1989) Nature, 340, 404.Rees, W. & Holamn, G. (1981) Biochim. Biophys. Acta, 646, 251.Reivich, M., Kuhl, D., Wolf, A., Greenberg, J., Phelps, M., Ido, T., Casella, V.,Fowler, J., Hoffman, E., Alavi, A., Som, P., & Sokoloff, L. (1979) Circ. Res., 44,127.Rios, M. (1994) Eur. J. Clin. Invest., 24, Suppi. 3, 36.Roeser, K.-R. & Legler, G. (1981) Biochim. Biophys. Acta, 657, 321.Rowland, M. & Tozer, T., Clinical Phar,nacokinetics: Concepts and Applications. 1989,Philadelphia: Lea & Febiger. p. 297.Satyamurthy, N., Bida, G. T., Padgett, H. C., & Barrio, J. R. (1985) J. Carb. Chem.,4, 489.Schimmel, P. (1990) Biochemistry, 29, 9495.Shelling, J. G., Dolphin, D., Wirz, P., & Einstein, F. W. B. (1984) Carbohydr. Res.,132, 241.Shiue, C.-Y., Arnett, C. D., & Wolf, A. P. (1984) Eur. J. Nucl. Med., 9, 77.Shulman, M. L., Shiyan, S. D., & Khorlin, A. Y. (1976) Biochim. Biophys. Acta, 445,169.Sierks, M. R., Bock, K., Refn, S., & Svensson, B. (1992) Biochemistry, 31, 8972.195Sinnott, M. L., Glycosyl group transfer, in Enzyme Mechanisms, Page, M.I. andWilliams, A., eds. 1987, Royal Society of Chemistry: London. p. 259.Sinnott, M. L. (1990) Chem. Rev., 90, 1171.Sinnott, M. L. (1993) Bioorg. Chem., 21, 34.Sinnott, M. L. & Jencks, W. P. (1980) J. Am. Chem. Soc., 102, 2026.Sinnott, M. L. & Smith, P. 3. (1976) J. C. S. Chem. Commun., 223.Sinnott, M. L. & Souchard, I. 3. L. (1973) Biochem. J., 133, 89.Sinnott, M. L. & Withers, S. G. (1974) Biochem. J., 143, 751.Sinnott, M. L., Withers, S. G., & Viratelle, 0. M. (1978) Biochem. J., 175, 539.Skell, P. S. & Maxwell, R. J. (1962) J. Am. Chem. Soc., 84, 3962.Somsák, L. & Ferrier, R. J. (1991) Adv. Carb. Chem. Biochem., 49, 37.Sorge, 3., West, C., Westwood, B., & Beutler, E. (1985) Proc. Nail. Acad. Sd.U.S.A., 82, 7289.Spinks, T., Jones, T., Bailey, D., Townsend, D., Grootoonk, S., Bloomfield, P.,Gilardi, M.-C., Casey, M., Sipe, B., & Reed, J. (1992) Phys. Med. Biol., 37, 1637.Stoddart, J. F., Stereochemistry ofCarbohydrates. 1971, New York: Wiley-Interscience.p. 57.Strauss, H. W. (1989) J. Nuci. Med., 30, 1123.Street, I. P., Kempton, 3. B., & Withers, S. G. (1992) Biochemistry, 31, 9970.Street, I. P., Rupitz, K., & Withers, S. G. (1989) Biochemistry, 28, 1581.Svensson, B. & Sogaard, M. (1993) J. Biotechnol., 29, 1.Takasaki, S., Murray, G. J., Furbish, F. S., Brady, R. 0., Barranger, J. A., & Kobata,A. (1984) J. Biol. Chem., 268, 10112.Takase, K., Matsumoto, T., Mizuno, H., & Yamane, K. (1992) Biochim. Biophys.Acta, 1120, 281.Tao, B. Y., Reilly, P. 3., & Robyt, 3. F. (1989) Biochim. Biophys. Acta, 995, 214.Tull, D., Miao, S., Withers, S. G., & Aebersold, R. (1995) Anal. Biochem., 224, 509.Tull, D., Withers, S. G., Gillces, N. R., Kilburn, D. G., Warren, R. A. 3., & Aebersold,R. (1991) J. Biol. Chem., 266, 15621.196Tybulewicz, V., Tremblay, M., LaMarca, M., Willemsen, R., Stubblefield, B., Winfield,S., Zablocka, B., Sidransky, E., Martin, B., Huang, S., Mintzer, K., Westphal, H.,Mulligan, R., & Ginns, E. (1992) Nature, 357, 407.Van Diggelen, 0. P. & Galjaard, H. (1980) Biochem. J., 188, 337.Van Diggelen, 0. P., Galjaard, H., Sinnott, M. L., & Smith, P. 3. (1980) Biochem. J.188, 337.Van Hoof, F. & Hers, H. G. (1968) Eur J. Biochem. 7, 34.Wagner, H. N. (1986) Semin. Nuci. Med., 16, 51.Walsh, C. (1983) Adv. Enzymol., 55, 197.Wang, Q., Tull, D., Meinke, A., Gilkes, N. R., Warren, R. A. 3., Aebersold, R., &Withers, S. 0. (1993) J. Biol. Chem., 268, 14096.Weber, J. P. & Fink, A. L. (1980) J. Biol. Chem., 255, 9030.Wenger, D. A., Sujansky, E., Fennessey, P. V., & Thompson, 3. N. (1986) New Engi.J. Med., 315, 1201.Wentworth, D. F. & Wolfenden, R. (1974) Biochemistry, 13, 4715.White, A., Withers, S., Gillces, N., & Rose, D. (1994) Biochemistry, 33, 12546.Winkler, D. A. & Holan, G. (1989) J. Med. Chem., 32, 2084.Withers, S. G., Rupitz, K., & Street, I. P. (1988) J. Biol. Chem., 263, 7929.Withers, S. G., Rupitz, K., Trimbur, D., & Warren, R. A. 3. (1992) Biochemistry, 31,9979.Withers, S. G. & Street, I. P. (1988) J. Am. Chem. Soc., 110, 8551.Withers, S. G., Street, I. P., Bird, P., & Dolphin, D. H. (1987) J. Am. Chem. Soc.,109, 7530.Withers, S. G., Warren, R. A. 1., Street, I. P., Rupitz, K., Kempton, 3. B., &Aebersold, R. (1990) J. Am. Chem. Soc., 112, 5887.Wolfenden, R. (1972) Acc. Chem. Res., 5, 10.Wolfenden, R. & Kati, W. M. (1991) Acc. Chem. Res., 24, 209.Yoshida, K., Oshima, A., Shimmoto, M., Fukuhara, Y., Sakuraba, H., Yanagisawa, N.,& Suzuki, Y. (1991) Am. J. Hum. Genet., 49, 435.Yuan, J., Martinez-Bilbao, M., & Huber, R. E. (1994) Biochem. J., 299, 527.Zhang, S., McCarter, 3., Okamura-Oho, Y., Yaghi, F., Hinek, A., Withers, S. G., &Callahan, J. W. (1994) Biochem. J., 304, 281.197Zhang, Y., Bommuswamy, J., & Sinnott, M. L. (1994) J. Am. Chem. Soc., 116, 7557.198


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items