UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Mechanistic studies on phosphoglucomutase Percival, Michael David 1988

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

Item Metadata

Download

Media
831-UBC_1988_A1 P47.pdf [ 13.57MB ]
Metadata
JSON: 831-1.0060300.json
JSON-LD: 831-1.0060300-ld.json
RDF/XML (Pretty): 831-1.0060300-rdf.xml
RDF/JSON: 831-1.0060300-rdf.json
Turtle: 831-1.0060300-turtle.txt
N-Triples: 831-1.0060300-rdf-ntriples.txt
Original Record: 831-1.0060300-source.json
Full Text
831-1.0060300-fulltext.txt
Citation
831-1.0060300.ris

Full Text

MECHANISTIC STUDIES ON PHOSPHOGLUCOMUTASE By MICHAEL DAVID PERCTVAL B.Sc. (Hons), The University of Otago, 1982  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Chemistry)  We accept this thesis as conforming toihe required standard  THE UNIVERSITY OF BRITISH COLUMBIA September 1988 © Michael David Percival, 1988  In presenting  this thesis in partial fulfilment of the requirements for an advanced  degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may department or by  his or her  representatives.  be granted by the head of  It is understood that copying or  publication of this thesis for financial gain shall not be allowed without my permission.  Department of  Chemistry  The University of British Columbia Vancouver, Canada Date  DE-6 (2/88)  12 October, 1988  my  written  ii  ABSTRACT. The mechanism ofrabbitskeletal muscle phosphoglucomutase (EC.2.7.5.1) has been investigated using fluorinated and deoxygenated substrate analogues. Each of the analogues in which the non-acceptor hydroxyls arereplacedby fluorine or hydrogen are substrates of the enzyme. The kinetic constants of these substrates are reported. The rate of the mutase reaction of each substrate analogue in the presence of glucose 1,6-diphosphate is the same as that of the half reaction involving production of the fluorinated and deoxygenated glucose 1,6-diphosphate species. The exceptions are 3-fluoro- and 3-deoxy-glucose 1-phosphate, in which cases the rates of the half reactions are 8 times that of the overall mutase reaction. The K of 3-fluoro-glucose m  1,6-diphosphate is approximately 90 times that of glucose 1,6-diphosphate and the other deoxy and fluoro analogues. The inhibition of phosphoglucomutase byfluorinatedand deoxygenated substrate analogues has been investigated. The synthesis of a series of novel disubstituted inhibitors (based on glucose 1-phosphate) in which the C-6 hydroxyl is replaced byfluorineand a sugar ring hydroxyl is replaced by either hydrogen orfluorineis described. The inhibition constants show that the hydroxyl distal to the acceptor hydroxyl is most important in the formation of a strong enzyme-inhibitor complex. The synthesis is described of three phosphorofluoridate analogues of glucose phosphate substrates. These analogues were found to only weakly inhibit phosphoglucomutase. No evidence of any phosphoryl transfer between the phosphoenzyme and the phosphorofluoridate analogues could be detected. Thus phosphoglucomutase has a strict requirement for a doubly negatively charged substrate phosphate group. The interaction of phosphoglucomutase with fluorinated substrates and inhibitors has been investigated by l^F-nmr. Large downfield changes in the chemical shifts of the inhibitors 6-fluoro-glucose 1-phosphate and a-glucosylfluoride6-phosphate were found to accompany binding to the phosphoenzyme. The effects of the binding of activating and non-activating metal ions on these spectra were investigated. The different effects observed may be directly related to the chemical basis for the metal induced activation of the enzyme. ^F-Nmr data consistent with  ii i  a 10 to 103 fold increase in the tenacity with which phosphoglucomutase binds substrates and 2  inhibitors in the presence of L i were observed in the spectra of the phosphoenzyme with +  difluorinated glucose 1-phosphate inhibitors. Two enzyme bound species were detected in the l^F-nrnr spectra of the complexes formed by reaction of the Cd^+.phosphoenzyme with 2- and 3-fluoro-glucose phosphates. These species are tentatively assigned as the fluoro-glucose 1,6diphosphate species bound in two different modes to the dephosphoenzyme. Only one bound species was observed in the case of 4-fluoro-glucose phosphates. The environment of each substrate glucose hydroxyl in the active site was probed using l^p-rimr and the fluorinated glucose phosphate substrates. Data inconsistent with a minimal motion type of mechanism (WJ. Ray, A.S. Mildvan & J.W. Long, Biochemistry 1973,12, 3124) were obtained. Theresultsof the nmr and kinetic studies are consistent with an exchange type of mechanism in which the C-3 hydroxyl plays an important role in the reorientation of the glucose 1,6-diphosphate. The data also suggest that there are two distinct glucose binding sites, one for each substrate and glucose 1,6-diphosphate bound in the same mode.  iv TABLE OF CONTENTS. Abstract  ii  List  of  Tables  vii  List  of Figures  viii  List of Abbreviations  xi  Glossary of Enzymic Terms  xiii  Acknowledgements  xiv  I.  Background 1.1. Enzymes-A General Background and Perspective 1.2. Phosphoglucomutase-A Brief Introduction to the Enzyme 1.2.1. History and Biological Significance 1.2.2. The Mechanism of the Reaction 1.2.3. Metal Ion Effects 1.2.4. Other Phosphomutases 1.2.5. Aims of this Study  1 1 3 3  4  9 10  12  n. Kinetic Studies on the Utilization of Fluorinated and Deoxygenated Substrate Analogues 14 II.A. Introduction 14 II.A.l. The Use of Enzyme-Substrate Binding Energy 14 II. A.2. The Source of Enzyme-Substrate Binding Energy and its Measurement 16 II.A.3. The Effects of Ligand Modification 19 II.A.4. The Energetics of Hydrogen Bond Formation 24 II.A.5. Summary and Overview 25 II.B. Results and Discussion 27 II.B.1. The Synthesis of Fluorinated and Deoxygenated Analogues of Glucose 1Phosphate 27 II.B. 1.1.  2,6-Dideoxy-2,6-difluoro-a-D-glucopyranosyl phosphate  .... 27  II.B. 1.2. 3,6-Dideoxy-3,6-difluoro-a-D-glucopyranosyl phosphate  .... 28  II.B. 1.3. 3,6-Dideoxy-6-fluoro-a-D-ribohexopyranosyl phosphate  31  II.B. 1.4. 4,6-Dideoxy-4,6-difluoro-a-D-glucopyranosyl phosphate  .... 33  II.B. 1.5. 4,6-Dideoxy-6-fluoro-a-D-xylohexopyranosyl phosphate .... 35 II.B. 1.6. 2-Deoxy-a-D-arabinohexopyranosyl phosphate 37 II.B.2. Analysis of the Solution Conformations of Glucose 1-Phosphate Analogues 40 II.B.3. Interaction of Fluoro and Deoxy-Substrates and Inhibitors with Phosphoglucomutase 43 II.B.3.1. Methods of the Kinetic Parameter Analysis 43 II.B.3.2. Substrate Activity of Fluoro and Deoxy-Analogues 44 II.B.3.3. Determination of Substrate Kinetic Parameters 52 II.B.3.4. Determination of Inhibitor Kinetic Parameters 59 II.B.3.5. Spectral Studies on Enzyme-Ligand Complexes 66 II.B.3.6. Evaluation of Equilibrium Constants of Fluoro-Substrates .... 69 II.B.4. Summary and Implications for the Mechanism 70 II.C. Experimental Procedures 74 II.C.l. General Synthetic Methods and Purification of Reagents 74  V  II.C.2. II.C.3.  Synthetic Methods Biological Methods  77 92  HI. Charge State Analogues and Potential Covalent Inhibitors 95 III.A. Introduction 95 III.A.l. Substrate Ionization States 95 III.A.2. Phosphite Esters as Phosphate Analogues 98 III.A.3. Phosphorofluoridate Esters as Phosphate Analogues 100 III.A.4. Cyclic-Phosphate Esters as Phosphate Analogues 105 III.A.5. Sulfate Esters as Phosphate Analogues 106 III.A.6. Enzyme-Phosphate Electrostatic Interactions in Phosphoglucomutase 107 III.A.7. Summary and Overview 108 III.B. Results and Discussion 109 III.B.l. Synthesis of Phosphorofluoridate Analogues 109 III.B.2. Synthesis of Cyclic-Phosphate Analogues 112 III.B.3. Biological Activity of Phosphorofluoridate and Cyclic-Phosphate Analogues 113 III. C. Experimental Procedures 121 III.C.l. General Methods and Purification of Reagents 121 III. C.2. Synthetic Methods 122 IV. NMR Studies of Enzyme-Ligand Complexes 1 2 6 IV. A. Introduction 1 2 6 IV. A.l. F-NMR in Macrolecular Systems 126 IV.A.2. F-NMR Chemical Shifts in Macromolecular Systems 128 IV.A.3. Relaxation in Macromolecular Systems 130 IV.A.4. Nuclear Overhauser Effects in Macromolecular Systems 132 IV.A.5. Summary and Overview 133 IV.B. Results and Discussion 134 IV.B.1. Synthesis of Fluorinated and Deoxygenated Analogues of Glucose 6Phosphate 134 IV.B.2. Investigation of Enzyme-Ligand Complexes by ^ F-NMR 137 IV.B.2.1. Complexes of 6-Fluoro-Glucose 1-Phosphate and oc-Glucosyl Fluoride 6-Phosphate 137 IV.B.2.2. Effect of Binding of Activating Metal Ions 152 IV.B.2.3. Ternary Enzyme-Cd -Fluoro-Substrate Complexes 161 IV.B.2.4. Complexes of Difluorinated Inhibitors 1 7 0 IV.B.2.5. Ternary Phosphoenzyme-Li-Ruoro-Substrate Complexes 177 IV.B.3. Summary and Implications for the Mechanism 194 IV.B.4. 1 9 . 1 H . Nuclear Overhauser Effects 196 IV.B.4.1. Investigation of a Ligand-Enzyme iH-f^F} NOE 198 IV.B.4.2. Investigation of a Ligand-Enzyme F - { ! H } NOE 201 IV.C. Experimental Procedures 205 IV.C.l. General 205 IV.C.2. Synthetic Methods 205 IV.C.3. Enzyme Isolation and Purification 207 IV.C.4. Demctallation of NMR Solutions and Additives 211 IV.C.5. Demctallation of Phosphoenzyme 212 IV.C.6. Acquisition of F-NMR Spectra 214 IV.C.7. Deuteration of Phosphoenzyme and Reagents for ^H-NMR 215 IV.C.8. Acquisition of ifl-NMR Spectra 215 V. Summary 217 19  19  9  2+  +  F  1 9  19  vi  Appendix  221  References  229  vii L I S T  O F  T A B L E S .  Table II. 1. Comparison of the sizes of some funtional groups 20 Table 1X2. ^H, l^F and 31p-nmr chemical shifts of substituted a-D-glucopyranosyl phosphates 41 Table 1X3. lH-lH coupling constants of substituted a-D-glucopyranosyl phosphates 41 Table 1X4. ^H-^lp and ^H-l^F coupling constants of substituted a-D-glucopyranosyl phosphates 42 Table 1X5. Kinetic constants of fluorinated and deoxygenated substrate analogues of glucose 1-phosphate with phosphoglucomutase 54 Table 1X6. Dissociation constants of glucose phosphate inhibitors of phosphoglucomutase. 61 Table 1X7. Equilibrium ratios of fluorinated glucose phosphates at 30° C and pH 7.4 70 Table DX1. Dissociation constants (25° C) of phosphoric and phosphorous acids 99 Table HX2. The properties of sulfate and phosphate ions 114 Table DX3. Dissociation constants of phosphorofluoridate inhibitors of phosphoglucomutase 117 Table IV. 1. The properties of some nuclei 127 Table TV.2. Kinetic parameters of yeast hexokinase with deoxy and fluoro-analogues of glucose 134 Table PV.3. % spin-lattice relaxation data of enzyme-bound and free species in the Ep.6FGlc 1-P complex 143 Table IV.4. 19p spin-spin relaxation data of enzyme-bound andfreespecies in the Ep.6FGlc 1-P complex 144 Table TV.5. Summary of ^F-nmr chemical shifts of complexes of phosphoenzyme and 6-fluoro-glucose 1-phosphate and a-glucosyl fluoride 6-phosphate 147 Table TV.6. Summary of ^F-mnr chemical shifts of free fluoro-substrates and complexes produced by reaction with the phosphoenzyme and the phosphoenzyme-Cd2+ complex 168 Table IV.7. Summary of l^F-nmr chemical shifts of binary and ternary complexes of phosphoglucomutase and difluorinated glucose 1-phosphate inhibitors 172 Table IV.8. Surrimary of l^F line widths of the free and enzyme-bound species in spectra of difluorinated glucose 1-P complexes 175 Table IV.9. Summary of l^F-nmr chemical shifts offreeand liganded fluoro-glucose 1phosphate analogues in the ternary phosphoglucomutase-Li-ligand complexes 189 Table IV. 10. Summary of the reaction conditions and yields obtained in the synthesis of each fluorinated and deoxygenated derivative of glucose 6-phosphate 206 Table IV.l 1. Summary of 31p and l^F-nmr data of fluorinated and deoxygenated analogues of glucose 6-phosphate 207 +  viii LIST OF FIGURES.  Figure 1.1. Reaction scheme for the interconversion of glucose 1-phosphate and glucose 6phosphate by phosphoglucomutase 5 Figure 1.2. Schematic representation of a minimal motion type of mechanism for phosphoglucomutase 6 Figure 1.3. Schematic representation of an exchange type of mechanism for phosphoglucomutase 7 Figure 1.4. The reactions catalyzed by phosphoglycerate mutase and bisphosphoglycerate mutase 11 Figure 1.5. Proposed reaction mechanism of wheat germ phosphoglycerate mutase 12 Figure II. 1. Synthetic route for the preparation of 2,6-dideoxy-2,6-difluoro-a-Dglucopyranosyl [bis(cyclohexylammonium) phosphate] 28 Figure H.2. Synthetic route for the preparation of 3,6-dideoxy-3,6-difluoro-a-Dglucopyranosyl [bis(cyclohexylammonium) phosphate] 29 Figure n.3. Mechanism of the reaction of DAST with 1,2-O-isopropylidene-a-D-glucofuranoseand3-a^xy-3-fluoro-l,2-0-isopropyUdene-a-D-glucofuranose ... 30 Figure H.4. Synthetic route for the preparation of 3,6-dideoxy-6-fluoro-oc-D-ribohexopyranosyl [bis(cyclohexylammonium) phosphate] 32 Figure II.5. Synthetic route for the preparation of 4,6-dideoxy-4,6-difluoro-a-Dglucopyranosyl [bis(cyclohexylammonium) phosphate] using a single fluorination step 34 Figure II.6. Synthetic route for the preparation of 4,6-dideoxy-4,6-difluoro-a-Dglucopyranosyl [bis(cyclohexylammonium) phosphate] using two fluorination steps 35 Figure H.7. Synthetic route for the preparation of 4,6-dideoxy-6-fluoro-a-D-xylohexopyranosyl [bis(cyclohexylammonium) phosphate] 36 Figure H.8. First step of the enzymatic synthesis of 2-deoxy-a-D-arabinohexopyranosyl [bis (ammonium) phosphate] 38 Figure n.9. Second step of the enzymatic synthesis of 2-deoxy-a-D-arabinohexopyranosyl [bis(ammonium) phosphate] 39 Figure BMO. Kinetic mechanism of the reaction of phosphoglucomutase with a fluoroglucose 1-phosphate analogue and glucose 1,6-diphosphate 45 Figure n. 11. Time courses of the reaction of 4-fluoro-glucose 1-phosphate with phosphoglucomutase in the presence of varied concentrations of glucose 1,6diphosphate 47 Figure 11.12. Time courses of the reaction of 3-fluoro-glucose 1-phosphate with phosphoglucomutase in the presence of varied concentrations of glucose 1,6diphosphate 48 Figure 11.13. Schematic representation oftimecourses measured by the coupled assay of the reaction of 4-£luoro-glucose 1-phosphate with phosphoglucomutase in the presence of varied concentrations of glucose 1,6-diphosphate 51 Figure 11.14. Schematic representation oftimecourses measured by the coupled assay of the reaction of 3-fluoro-glucose 1-phosphate with phosphoglucomutase in the presence of varied concentrations of glucose 1,6-diphosphate 51 Figure 11.15. Schematic representation oftimecourses measured by the coupled assay of the reaction of 2-fluoro-glucose 1-phosphate with phosphoglucomutase in the presence of varied concentrations of glucose 1,6-diphosphate 52 Figure 11.16. Proposed ground state enzyme-substrate hydrogen bonding scheme for phosphoglucomutase 64 T  IX  Figure n.17. Gibbs energy changes associated with catalysis and binding of substrates and inhibitors of phosphoglucomutase 65 Figure n.18. Schematic representation of complete and partial substrate recognition by an enzyme substrate binding site 67 Figure n.19. Inhibitor induced UV difference spectrum produced by the binding of 3,6difluoro-glucose 1-phosphate to demetallated phosphoenzyme 68 Figure 11.20. Inhibitor induced UV difference spectrum produced by the binding of methyl phosphonate to demetallated phosphoenzyme 69 Figure 11.21. Schematic diagram of the proposed Rose exchange mechanism for phosphoglucomutase 73 Figure III. 1. The structure of phosphoric and phosphorous acids 98 Figure DX2. Reaction mechanism for the formation of phosphorofluoridate monoesters by reaction with 2,4-dinitrofluorobenzene 100 Figure m.3. Mechanism of covalent inhibition of serine proteases by diisopropylphosphorofluoridate 101 Figure III.4. Proposed mechanisms of phosphoryl transfer 102 Figure DI.5. Proposed covalent inhibition of phosphoglucomutase by a mixed phosphorofluoridate-phosphate ester substrate analogue 104 Figure m.6. Proposed covalent inhibition of phosphoglucomutase by a cyclic phosphate substrate analogue 106 Figure DI.7. Proposed reaction mechanism for formation of an acetoxonium ion during reaction of tetra-O-acetyl-a-D-glucose 1-phosphate and 2,4dinitrofluorobenzene 110 Figure m.8. Synthetic route for the preparation of glucose 4:6-cyclic phosphate 1phosphate 112 Figure III.9. Newmann projection of the conformation of the 4:6-cyclic phosphate group of glucose 4:6-cyclic phosphate about the C-O bonds 118 Figure HI. 10. Newmann projection of the conformation of the phosphate group of glucose 6phosphate about the C-O bond 119 Figure IV. 1. Graph of the relationship between correlationtimeand spin-lattice and spin-spin relaxation times 131 Figure IV.2. Attempted chemical synthetic route to cc-glucosyl fluoride 6-phosphate 136 Figure IV.3. Titration of phosphoenzyme with 6-fluoro-glucose 1-phosphate followed by F-nmr 139 Figure IV.4. Effect of decoupling on the l F-nmr spectrum of the phosphoenzyme-6fluoro-glucose 1-phosphate complex 141 Figure IV.5. Effect of binding of L i on the phosphoenzyme-6-fluoro-glucose 1-phosphate complex followed by l^F-nmr 146 Figure IV.6. Titration of phosphoenzyme with a-glucosyl fluoride 6-phosphate and L i followed by F -nmr 148 Figure IV.7. Effect of Mg2+ on the phosphoenzyme-6-fluoro-glucose 1-phosphate complex followed by 19 -nmr 153 Figure rv.8. Titration of the phosphoenzyme-6-fluoro-glucose 1-phosphate complex with Mg2+ followed by l F-nmr 154 Figure IV.9. Effect of Mg2+ on the phosphoenzyme-Li-6-fluoro-glucose 1-phosphate complex followed by l F-nmr 156 Figure IV. 10. Effect of C d on the enzyme-4-fluoro-glucose-phosphate complex followed by 19 -nmr 163 Figure IV. 11. Effect of Cd on the enzyme-3-fluoro-glucose-phosphate complex followed by 19 -nmr 165 Figure IV. 12. Effect of Cd"*" on the enzyme-2-fluoro-glucose-phosphate complex followed by 19 -nmr 167 19  9  +  +  1 9  F  9  +  9  2+  F  2+  F  2  F  X  Figure IV.13. Titration of phosphoenzyme with 4,6-difluoro-glucose 1-phosphate and L i followed by 19p_ 171 Figure IV. 14. Titration of phosphoenzyme with 3,6-difluoro-glucose 1-phosphate andLi followed by 19p_ 173 Figure IV.15. Titration of phosphoenzyme with 2,6-difluoro-glucose 1-phosphate and L i followed by l^F-nmr 174 Figure IV.16. Binding of 4-fluoro-glucose 1-phosphate to the phosphoenzyme-Li complex followed by ^F-nmr 180 Figure IV. 17. Binding of 4-fluoro-glucose 6-phosphate to the phosphoenzyme-Li complex followed by 19p_ 181 Figure rv.18. Binding of 3-fluoro-glucose 1-phosphate to the phosphoenzyme-Li complex followed by F-nmr 184 Figure IV. 19. Binding of 3-fluoro-glucose 6-phosphate to the phosphoenzyme-Li complex followed by 19p_ 185 Figure IV.20. Binding of 2-fluoro-glucose 1-phosphate to the phosphoenzyme-Li complex followed by 19p_ 187 Figure IV.21. Binding of 2-fluoro-glucose 6-phosphate to the phosphoenzyme-Li complex followed by F-nmr 188 Figure IV.22. Probable mechanism of phosphoryl transfer showing participationfromacid and base catalysts 196 Figure IV.23. lH-nmr spectrum of phosphoglucomutase at 270 MHz 200 Figure IV.24. Time dependance of ^ F- {^H} noe development at two *H decoupling frequencies in the Ep.6FGlc 1-P complex 202 Figure IV.25. Basic steps of isolation and purification of phosphoglucomutase 208 Figure IV.26. Modifications to the Bruker HXS-270 spectrometer to allow observation and 1 ° F decoupling 216 +  nmr  +  nmr  +  +  +  nmr  +  19  +  nmr  +  nmr  +  19  9  xi L I S T  O F  A B B R E V I A T I O N S .  ctGlcF 6-P a-D-glucosyl fluoride 6-phosphate AIBN azobisisobutyronitrile ATP adenosine 5'-triphosphate Bo magnetic field strength Bz benzyl DAST diemylammosulfur trifluoride 1- deoxy-glucose 6-P 1,5-anhydro-D-glucitol 6-phosphate 2- deoxy-glucose 1-P 2-deoxy-a-D-arabinohexopyranosyl phosphate 3- deoxy-glucose 1-P 3-deoxy-a-D-ribohexopyranosyl phosphate 4- deoxy-glucose 1-P 4-deoxy-a-D-xylohexopyranosyl phosphate 6-deoxy-glucose 1-P 6-deoxy-a-D-glucopyranosyl phosphate 3- deoxy-6-fluoro-glucose 1-P .... 3,6-dideoxy-6-fluoro-a-D-ribohexopyranosyl phosphate 4- deoxy-6-fluoro-glucose 1-P .... 4,6-dideoxy-6-fluoro-a-D-xylohexopyranosyl phosphate 2,6-difluoro-glucose 1-P 2,6-dideoxy-2,6-difluoro-a-D-glucopyranosyl phosphate 3,6-difluoro-glucose 1-P 3,6-dideoxy-3,6-difluoro-a-D-glucopyranosyl phosphate 4,6-difluoro-glucose 1-P 4,6-dideoxy-4,6-difluoro-a-D-glucopyranosyl phosphate DMAP 4-(d^emylainmo)-pyrio!ine DNA (*H) decoupler on during acquisition ED dephosphoenzyme EDTA ethylenediamine tetraacetic acid Ep phosphoenzyme FAB Fast atom bombardment fid free induction decay 2FGlc 1-P 2-deoxy-2-fluoro-a-D-glucopyranosyl phosphate 3FGlc 1-P 3-deoxy-3-fluoro-a-D-glucopyranosyl phosphate 4FGlc 1-P 4-deoxy-4-fluoro-a-D-glucopyranosyl phosphate 6FGlc 1-P 6-deoxy-6-fluoro-a-D-glucopyranosyl phosphate 2FGlc 6-P 2-deoxy-2-fluoro-D-glucose 6-phosphate 3FGlc 6-P 3-deoxy-3-fluoro-D-glucose 6-phosphate 4FGlc 6-P 4-deoxy-4-fluoro-D-glucose 6-phosphate 2,6F2Glc 1-P 2,6-dideoxy-2,6-difluoro-a-D-glucopyranosyl phosphate 3,6F2Glc 1-P 3,6-dideoxy-3,6-difluoro-a-D-glucopyranosyl phosphate 4,6F2Glc 1-P 4,6-dideoxy-4,6-difluoro-a-D-glucopyranosyl phosphate 4FMeGlc methyl 4-deoxy-4-fluoro-a-D-glucopyranoside 6FMeGlc methyl 6-deoxy-6-fluoro-a-D-glucopyranoside 2-fluoro-glucose 1-P 2-deoxy-2-fluoro-a-D-glucopyranosyl phosphate 2- fluoro-glucose 1-PF 2-deoxy-2-fluoro-a-D-glucopyranosyl phosphorofluoridate 3- fluoro-glucose 1-P 3-deoxy-3-fluoro-a-D-glucopyranosyl phosphate 4- fluoro-glucose 1-P 4-deoxy-4-fluoro-a-D-glucopyranosyl phosphate 6-fluoro-glucose 1-P 6-deoxy-6-fluoro-a-D-glucopyranosyl phosphate 3-fluoro-glucose 1,6-diP 3-deoxy-3-fluoro-a-D-glucose 1,6-diphosphate y gyromagnetic ratio  xii  G glucose 1-P glucose 1-PF glucose 6-P glucose 6-PF glucose 1,6-diP mannose 1-PF lit M Me MOM m.p. M NADP nmr noe OAc Ph o tc Tl T 2 TFA THF TMP TMS UDPG UTP UV coo  Gibb's free energy a-D-glucopyranosyl phosphate a-D-glucopyranosyl phosphorofluoridate D-glucose 6-phosphate D-glucose 6-phosphorofluoridate a-D-glucose 1,6-diphosphate a-D-mannopyranosyl phosphorofluoridate literature mass methyl methoxy methyl melting point relative molecular weight (^-nicotinamide adenine dinucleotide phosphate nuclear magnetic resonance nuclear Overhauser effect acetate phenyl shielding constant correlation time spin-lattice relaxation time spin-spin relaxation time trifluoroacetic acid tetrahydrofuran trimemylpyridine tetramethylsilane uridine diphosphoglucose uridine 5'-triphosphate ultraviolet spectrometer frequency (rads" in nmr assignments  ax b d eq m s t  axial broad doublet equatorial multiplet singlet triplet  r  xiii GLOSSARY OF ENZYMIC  TERMS.  This brief glossary of terms is included for the reader not familiar with enzyme kinetics. The mathematical derivation and meaning of each of these terms can be found in almost any biochemistry textbook. kcat > simple systems kcat is the first order rate constant for the conversion of the enzymem  substrate complex into products. In more complex systems, it is a function of all the first order rate constants. The kcat value represents the number oftimesthe enzyme turns over per unit time. Kd; The enzyme-ligand dissociation equilibrium constant. It generally refers to an enzymesubstrate complex. Ki; The dissociation equilibrium constant of an enzyme-inhibitor complex. Km; The substrate concentration which produces half the enzyme's maximal velocity (Vmax)- hi simple cases, K = Kd; however, in more complex cases it is a function of equilibrium and rate m  constants. Vmax; The (maximal) velocity of an enzyme when saturated by substrate. Vmax = kcat [E]» where E is the total enzyme concentration.  xiv  ACKNOWLEDGEMENTS. I would like to thank Dr. Steve Withers for his advice and suggestions in the the past "few" years and also for his help in the preparation of this thesis. I am also indebted to Dr. Bill Ray for giving me the opportunity to learn the techniques involved in the isolation of phosphoglucomutase. Thanks is also due to the staff of the nmr laboratory who measured many spectra, the electronics workshop staff who were always available to fix the 270, everytimeit broke down and to Neil for the loan of his Mac. Lastly to Jill for her patience.  1 CHAPTER I: BACKGROUND.  1.1: Enzymes-A General Background and Perspective. Enzymes are proteins which are specialized to catalyze biological reactions. All living cells contain enzymes and life as we know it would not be possible without them. The majority of the chemical reactions which occur within the Uving cell are inherently slow and unlikely processes, and except in living cells they hardly ever occur. Enzymes are the key to life as they are the catalysts which speed up the chemical reactions needed for the process of living. The fact that these many chemical reactions occur only in living cells gave rise in the nineteenth century to the idea of a "vital force" (Dixon, 1970). This "vital force" resided in cells, directing their chemical activities and although the reactions of several enzymes had been discovered in the previous sixty years, it was not until 1897 when Buchner showed that sugar could be fermented by a cell free yeast extract, that our present concepts of enzymes began to emerge. However not until 1926 when the enzyme urease was crystallized by Sumner was strong evidence presented for the protein nature of enzymes. Sumner's results were challenged, some believing that the enzyme was an impurity adsorbed on the crystals of some inert protein. In the next ten years J. Northrop's group isolated and crystallized several proteolytic enzymes and it became generally accepted that enzymes are in fact proteins. All the enzymes which have been purified since (several thousand) have been shown, with the very recent exception of RNA enzymes (Cech & Bass, 1986), to be proteins. Many other advances have led to our modern understanding of enzymes. The great degree of specificity of each enzyme for its particular substrate led Emil Fischer in 1894 to suggest that the substrate molecule fits the active site of the enzyme in a "lock and key" or complementary relationship. A kinetic description of the mode of action of enzymes was published in 1913 (Michaelis & Menten). They showed that enzymes display "saturation kinetics" and therefore distinguish themselves from ordinary chemical reagents. The equations of Michaelis and Menten have been extended to more complex systems by Briggs and Haldane (1925) and more recently to equations for multi-substrate systems by Dixon and Webb (Segel, 1975).  2 The first complete sequencing of the amino acids of a polypeptide was that of insulin (actually a hormone) achieved by Sanger in 1953. But not until 1966 when Phillips constructed a 3-dimensional model of the enzyme lysozyme, by X-ray crystallography, was the way in which a protein chain is folded up to produce an active site determined (Phillips, 1967). Even more recently, advances in DNA technology are producing a new age of enzymology. Cloning of the genes for enzymes into bacteria for over-production has facilitated the study of enzymes which in the past were difficult to obtain in large amounts (Fersht, 1985). Site -directed mutagenesis has allowed the amino acid sequence of native enzymes to be altered at will. These changes in the primary structure of the enzyme allow enzymologists to tackle the fundamental questions of how substrate binding and catalysis occur. In the last one hundred years most of the enzymes concerned with basic cell metabolism have been identified. However many questions remain to be answered, including how enzyme synthesis is genetically controlled, how enzyme activity is regulated and how these enzymes manage to catalyze reactions with such efficiency, precision and specificity. There are many practical reasons why the above questions should be answered, apart from satisfying our enquiring minds. Many drugs are inhibitors of the enzyme systems of pathogenic bacteria. Clearly if our own metabolic system is not to be adversely affected then these inhibitors must be specific for the enzymes of the pathogen (Ferdinand, 1976). This requires us to have an intimate knowledge of how the enzymes of both man and microbe operate. There are many human diseases that are due to the absence or overproduction of enzymes. A better understanding of how drugs interact with the enzymes of the system will allow the development of more powerful and efficient methods of controlling the disease. Enzymes will probably be of greater use in the chemical and pharmaceutical industries of the future. Many reactions that are difficult to carry out by classical organic chemistry can be easily carried out with the correct enzyme under less extreme conditions. Because enzymes operate at lower temperatures than many industrial processes there would be a consequent savings in energy. Common enzymes can be altered to make their industrial and household use a more attractive proposition. For example, many washing detergents contain a genetically altered  3  proteolytic enzyme that has been engineered so it is stable to high temperatures and high peroxide concentrations. The wild type enzyme would be immediately inactivated under the same conditions. This thesis describes an investigation into the mechanism of the enzyme phosphoglucomutase. Although no human disease has been associated with the enzyme, nor has it at present any industrial application, it is an example of an important group of enzymes involved in phosphoryl transfer and for this reason merits study. This enzyme and its mechanism have been studied intensively for a number of years. However, the means by which it carries out its chemical transformation is still a matter of some debate. It was in an attempt to resolve some of these unanswered questions that this study was undertaken. 1.2: Phosphoglucomutase-A Brief Introduction to the Enzyme. Phosphoglucomutase (a-D-glucose 1,6-bisphosphate; oc-D-glucose 1-phosphate phosphotransferase; EC 2.7.5.1) is an enzyme which catalyzes the interconversion of a-Dglucose 1-phosphate (glucose 1-P) and D-glucose 6-phosphate (glucose 6-P). The equihbrium of the reaction lies 17 to 1 in favor of the 6-isomer at pH 7.5 and 30°.  1.2.1: History and Biological Significance. The existence of an enzyme that interconverts the 1- and 6-isomers of glucose phosphates was discovered by CF. and G.T. Cori in the 1930's during their investigation of glycogen metabolism. Phosphoglucomutase is ubiquitous in most, if not all living cells. It has been isolated and characterizedfroma wide range of sources including human and shark muscle, baker's yeast, Escherichia coli, rat liver and numerous plants (Najjar, 1962). Most of the work on phosphoglucomutase has been conducted with rabbit skeletal muscle enzyme and it is with this enzyme that this study has been undertaken. One of the two main functions of the enzyme is as part of the system in which glycogen or starch is converted into glucose and utilized as an energy source. Glycogen is the storage form of glucose in animals. It is a large branched polymer primarily linked by a-l,4-glycosidic  4  linkages, with some branch points created by a-l,6-linkages. In plants, glucose is stored as starch which differs from glycogen in the degree of branching. The path of glucose metabolism in animals involves phosphorolysis of glycogen, catalyzed by a phosphorylase, to produce glucose 1-P. Glucose 1-P is then isomerized to glucose 6-P by phosphoglucomutase. In muscle, the glucose 6-P is then fed into the glycolytic pathway which results in the net production of pyruvic acid and two moles of adenosine 5'-triphosphate (ATP). In the liver, which is the other major site of glycogen storage in animals, the glucose 6-P is hydrolyzed by a phosphatase to produce glucose and inorganic phosphate. Thus the liver releases glucose into the blood during periods of activity and between meals. After being absorbed by the cell, the glucose is rephosphorylated by hexokinase to produce glucose 6-P which feeds the glycolytic pathway. The other main function of phosphoglucomutase is to provide glucose 1-P for the formation of uridine diphosphoglucose (UDPG). UDPG is the precursor of many biosynthetic pathways including the synthesis of lactose, polysaccharides (such as glycogen), vitamin C and some detoxification reactions. 1.2.2: The Mechanism of the Reaction. Rabbit skeletal muscle phosphoglucomutase is a monomelic enzyme consisting of a single polypeptide chain with a relative molecular weight of 61600. The complete amino acid sequence (561 residues) has been determined by chemical means (Ray et al., 1983). The enzyme exists in two forms, a catalytically active phosphoenzyme (Ep) in which the serine-116 residue is phosphorylated and an inactive dephosphoenzyme (ED). The enzyme requires a divalent metal ion for activity. The physiological activator is Mg2+, but others such as Ni , Mn^+, Co and 2+  2+  Cd2+ will activate the enzyme although to a lesser extent (Ray, 1969). 31p-rimr has shown that the activating metal ion is most probably coordinated to the enzymic phosphate group (Rhyu et al, 1984). The enzyme also requires the presence of a catalytic amount of cc-D-glucose 1,6diphosphate (glucose 1,6-diP). The enzyme operates by donating its enzymic phosphate to the glucose monophosphate substrate to produce glucose 1,6-diP and the dephospho form of the enzyme ( E D . G I C 1,6-diP). The bound glucose 1,6-diP then donates the original phosphate group  5  of the substrate back to the enzyme to reproduce the phosphoenzyme and the alternate glucose monophosphate (Figure 1.1).  Ep.Glc 1-P  Ep  I  E G l c 1,6-diP D  I  T  Glc 1,6-diP  Ep.Glc 6-P Central complexes Figure 1.1: Reaction scheme for the interconversion of glucose 1-P phosphoglucomutase. From Ray and Peck, 1972.  and glucose 6-P by  Isotope exchange studies have shown that the free dephosphoenzyme is only rarely formed during the catalytic cycle since the second phosphoryl transfer step (transfer back to the dephosphoenzyme) occurs more rapidly than glucose 1,6-diP can dissociate from the enzyme (Ray & Roscelli, 1964; Britton & Clarke, 1968). Thus the role of glucose 1,6-diP is to ensure that the enzyme is present in the active phospho form The kinetic mechanism of phosphoglucomutase and its structure has been intensively studied over the last 25 years. Work up until 1972 has been reviewed by Ray and Peck (1972). What is still not understood about the enzyme's mechanism is how the dephosphoenzymeglucose 1,6-diP complex rearranges to allow the shuttling of the phosphate groups between the two positions whilst the glucose 1,6-diP remains bound in the active site. Two mechanisms, in which two successive phosphoryl transfers occur within the central complexes, have been postulated to account for the enzyme's activity (Ray et a/., 1973). A minimal motion mechanism would involve a single glucose ring binding site as depicted in Figure 1.2.  Ep.Glc 1-P  E .Glc 1,6-diP D  E .Glc 6-P P  Figure 1.2: Schematic representation of a minimal motion type mechanism for phosphoglucomutase. From Ray et al. (1973). Here it is envisaged that the phosphoenzyme is a mixture of rotameric species in which the enzymic phosphate can interact with one of two phosphate bmding sites, in addition to being covalently bonded to the enzyme. These two phosphate binding sites would be occupied by the phosphate group of the alternate substrate monophosphates in the two phosphoenzyme-glucose monophosphate complexes (Ep.Glc 1-P and Ep.Glc 6-P). Thus the single dephosphoenzymeglucose 1,6-diP complex (ED-Glc 1,6-diP) is structured so that when the 6-phosphate group is in a position to be transferred, only a small protein structural change is required to bring the 1phosphate into position for transfer, and vice versa. An exchange mechanism (Figure 1.3) would involve two rather different dephosphoenzyme-glucose 1,6-diP complexes that could be interconverted without complete dissociation of the diphosphate from the enzyme. In one complex, the 6-phosphate is in position for transfer to the enzyme. In the other complex, the 1-phosphate occupies the same position. Thus the enzyme must be able to accommodate the glucose ring of glucose 1-P and glucose 6-P in two different modes. An exchange mechanism would involve the enzyme having two distinct types of phosphate binding sites, one where the substrate monophosphate binds and the other where phosphoryl transfer occurs.  Ep. Glc 1-P  ED .Glc 1,6-diP  ED .Glc 1,6-diP  E  P  .Glc 6-P  Figure 1.3: Schematic representation of an exchange type mechanism for phosphoglucomutase. From Ray et al. (1973). Experimental evidence has been obtained to support both of these mechanisms. Isomerization of the free phosphoenzyme, at a rate of 4.5 x 10^ s"*, has been detected using isotope transport experiments (Britton & Clarke, 1968). These results have been interpreted in favor of a minimal motion type of mechanism, in which the rapid isomerization was considered to be an enzymic conformational change allowing the serine-116 hydroxyl to take up the two positions to which phosphoryl transfer of either the C-l or C-6 phosphate could occur. This type of mechanism was considered to have an evolutionary advantage over a dual site exchange mechanism since only a single glucose binding site would need to have evolved (Lowe & Potter, 1981). Model studies have shown that when the distance between the 1- and 6-phosphate groups of glucose 1,6-diP is minimized, a hydroxymethyl group can fit between the phosphate groups such that either could be transferred to the hydroxyl with minimal structural changes (Ray & Peck, 1972). However although glucose ring binding is simplified in a minimal motion mechanism, the phosphoryl transfer steps are made more complex since the direction of nucleophilic attack on the phosphate group depends on whether the 1- or 6-phosphate group is being transferred. In contrast, Knowles (1980) has suggested that the construction of a recognition (binding) site is a much simpler problem than the creation of a catalytic site. If this is the case,  8  then an exchange type of mechanism would appear to be the simpler of the two as a minimal motion mechanism would require the enzyme to have two phosphoryl transfer sites. Other evidence does indeed appear to be more consistent with an exchange type of mechanism. Ma and Ray (1980) showed that the changes induced in the UV spectrum of the enzyme on substrate binding are due to the bmding of the glucoseringrather than the phosphate group and that clifferent spectral changes are induced by the bmding of the two substrates. These data are inconsistent with a rninimal motion mechanism as it would be expected that similar spectral changes would occur on binding of either substrate since only a single glucose binding site exists. An examination of the UV difference spectrum produced by the binding of glucose 1,6diP to the dephosphoenzyme showed that it was very similar to that produced by the binding of glucose 1-P to the phosphoenzyme. These results suggest that the glucose ring of glucose 1,6diP binds to the same enzymic site as does the glucose ring of glucose 1-P. Furthermore, these results suggest that the major dephosphoenzyme-glucose 1,6-diP species is that which is bound in the same mode as that in which glucose 1-P binds to the phosphoenzyme (see Figure 1.3). The proposal that this species would dominate the putative dephosphoenzyme complexes is suggested by the fact that glucose 1-P binds to the phosphoenzyme approximately 2.5timesmore tightly than does glucose 6-P (Lowry & Passonneau, 1969). Thus the enzyme-substrate interactions would be stronger in the dephosphoenzyme-glucose 1,6-diP complex in which the 6-phosphate occupies the phosphoryl transfer site than the alternate complex, in which the 1-phosphate occupies the same site. Consequently the equilibrium between the two species would be such that glucose 1,6-diP bound in the glucose 1-P mode would be present in a higher proportion. The minimal motion mechanism is also inconsistent with the results of inhibition studies. Only a single strong phosphate binding site was detected in the phosphoenzyme whereas the presence of two sites, one strong and the other weak, was inferred for the dephosphoenzyme. This was rationalized in terms of an exchange mechanism, the strong site being the substrate binding site and the weak site being that at which the enzymic phosphate interacts and where phosphoryl transfer occurs (Ray et ah, 1973).  9  The enzyme-ligand interactions have also been investigated by 31p-nmr spectroscopy (Rhyu et al., 1984,1985a). These studies showed that only a single dephosphoenzyme-glucose 1,6-diP complex was observable, a result which would appear to be consistent with a minimal motion mechanism. However, other evidence suggested that the observed complex is probably that in which the 6-phosphate occupies the phosphoryl transfer site. The alternate complex, in which the 1-phosphate occupies the phosphoryl transfer site, was considered to be a minor componant. Thus the results of the 31p-nmr studies are consistent with those obtained in the aforementioned UV study. The X-ray crystal structure of the phosphoenzyme, determined to a resolution of 2.7 A, has allowed the structure of the active, site to be determined. The active site is situated at a position where all four domains that make up the enzyme converge and produce a large cleft, the degree of resolution allowing the position of three positively charged arginine residues deep in the active site to be determined. The position of these charged groups is such that a strong interaction could occur between them and the phosphate groups of glucose 1,6-diP. This strong salt bridge type of attractive force, plus the deep crevice in which the active site is situated, could provide a favorable environment for the interchange of the two glucose 1,6-diP species without complete dissociation from the enzyme (Lin et al., 1986). The structure of the enzyme therefore appears to be consistent with an exchange type of mechanism. Thus it appears that the majority of experimental evidence is in accord with an exchange type of mechanism. However, although a great deal of study has been undertaken to unravel the enzyme's mechanism, the nature of the problem is such that a definitive result is difficult to achieve. Consequendy, although the results of investigations to date have been very suggestive of an exchange mechanism they fall short of either the proof of one mechanism or the ruling out of the other.  1.2.3: Metal Ion Effects. The role of the metal ion in the activation of phosphoglucomutase has been studied in detail (Ray & Peck, 1972; Ray et al, 1978; Ray, 1978). In the absence of metal,  10  phosphoglucomutase is at least 10 times less active as when Mg^+ is bound at the activating 7  metal site. However, the way in which the binding of a divalent metal ion causes such a large activation is not well understood. As noted above, other divalent metal ions do activate the enzyme, those studied ranging from 70% to 0.2% of the activity elicited by Mg2+ The monovalent ion, L i also binds to the enzyme although the enzyme's activity in its presence is +  less than 10~8 that in the presence of Mg^+ (Ray et ah, 1978). L i and Mg 2+ bind +  competitively, with similar affinities (in the presence of substrates) and also produce similar UV spectral changes in the enzyme on binding. It has been proposed that L i and the divalent metal +  ions produce a similar enzymic spectral change as the effective charge of the divalent ions is reduced by ligation to a carboxylic acid. This is supported by the observation that Be , which 2+  has a reduced tendency to coordinate with carboxylic acids, elicits nearly double the difference intensity of other ions on binding to the enzyme. This has led to the proposal that a metal dependent enzymic structural change, perhaps involving the aforementioned carboxylic acid ligand, causes the greater than 10 fold activation of the enzyme (Ray, 1978). An ancillary metal 7  binding site on the enzyme has been detected by nmr experiments (Ray & Mildvan, 1970; Rhyu et al., 1985a). It appears that the ancillary site is weaker than the activating site and has no known function, as the enzyme is maximally active in the presence of a 1:1 molar ratio of activating metal ion (Ray, 1969).  1.2.4: Other Phosphomutases. Phosphoglucomutases isolated from some bacterial sources cannot be phosphorylated and appear to catalyze the isomerization by a cufferent mechanism from that of the animal and E. coli enzymes (Hanabusa et al., 1966). Other sugar phosphates such as N-acetylglucosamine phosphate, mannosamine phosphate, ribose and deoxy-ribose phosphates and fi-D-glucose phosphate all have enzymes specific for their isomerization which have been isolated from a variety of sources. However, none of the aforementioned enzymes have been characterized to the extent of rabbit skeletal muscle phosphoglucomutase (Ray & Peck, 1972).  11  Two other phosphomutases which catalyze the isomerization of phosphoglycerate substrates are phosphoglycerate mutase and bisphosphoglycerate mutase. The reactions of these enzymes are shown in Figure 1.4.  CH OP CHjOH  Phosphoglycerate mutase .  2  CH OH CHjOP 2  coo-  Aoo-  3-phosphoglycerate  2-phosphoglycerate  CH OP 2  Bisphosphoglycerate mutase  CH OH 2  COOP 1,3-bisphosphoglycerate Figure 1.4:  CH^OP CH OP + 2  H  +  coo2,3-bisphosphoglycerate  The reactions catalyzed by phosphoglycerate mutase and bisphosphoglycerate  mutase.  The reaction of phosphoglycerate mutase has been studied in detail. It has been found that the animal and yeast enzymes require the presence of the cofactor 2,3-bisphosphoglycerate and that both are phosphoenzymes, a single histidine residue bearing the phosphoryl group (Han & Rose, 1979). The reaction proceeds with the overall retention of the phosphate group's configuration (as does phosphoglucomutase, Lowe & Potter, 1981), which is interpreted in terms of two phosphoryl transfers and a single phosphoenzyme intermediate (Blatter & Knowles, 1980). Thus the enzyme's mode of action is similar to that of phosphoglucomutase, an exchange type of mechanism being inferred from inhibition studies (Knowles, 1980). In contrast, phosphoglycerate mutase from wheat germ does not require the bisphosphate cofactor, nor has a phosphoenzyme been isolated. The reaction does however proceed with overall retention of configuration and a phosphoenzyme intermediate has been implicated by other studies. The mechanism is believed to be that shown in Figure 1.5, the intermediate  12  phosphoenzyme-glycerate complex not dissociating to form free phosphoenzyme as is the case with phosphoglucomutase and animal and yeast phosphoglycerate mutases (Knowles, 1980).  COOFigure 1.5: Proposed reaction mechanism of wheat germ phosphoglycerate mutase. (From Knowles, 1980).  1.2.5: Aims Of This Study. The aim of the present study is to investigate the mechanism of phosphoglucomutase and to attempt to determine whether the enzyme does in fact operate by a minimal motion or exchange type of mechanism Our modus operandi has been the use of substrate analogues which have been modified by fluorination and deoxygenation. The binding and substrate activity of each fluorinated and deoxygenated analogue of the substrates glucose 1-P and glucose 6-P have been investigated in order to determine which of the substrates features are particularly important in the enzyme-catalyzed isomerization process. The modifications have taken two general forms. The first involves the replacement of each of the sugar hydroxyls by fluorine or hydrogen. Bmding energy derived from the interaction of enzyme and substrate at the ground state, but in particular at the transition state, is thought to be important in lowering activation free energies and therefore effecting catalysis. In light of this, the forces involved in the enzyme-substrate interactions in this system have been  13  investigated using the aforementioned substrate analogues with the aim of probing the types and magnitudes of the interaction with each glucose hydroxyl group. The synthesis of suitable analogues has allowed the interactions between the enzyme and substrate in both the ground state and transition state complexes to be investigated. The second form of modification has involved changing the charge state of the substrate's phosphate functionality by replacement of a non-bridging oxygen by fluorine. These analogues were used to define the charge state of the substrate with which the enzyme binds. These analogues were also of interest as potential suicide substrates which could provide covalently linked enzyme-substrate complexes for 31p or l F-nmr studies. 9  The nature of the enzyme-ligand interactions has been further investigated by the use of 19F-nmr to probe enzyme-substrate and enzyme-inhibitor complexes. The aim of this investigation was to use the great sensitivity of  chemical shifts to external effects to probe  the environment into which the fluorinated probe is placed on binding to the enzyme. Using this approach and a series of fluorinated analogues of both glucose 1-P and glucose 6-P, it was anticipated that the results would distinguish between the minimal motion and exchange mechanisms. The effects of the binding of the metal ions Mg , C d and L i on the enzyme2+  2+  +  fluoro-substrate complexes were also investigated by l F-nmr. The aim was to examine the way 9  in which each metal ion elicits such a different enzymic catalytic activity.  14 C H A P T E R II: K I N E T I C S T U D I E S O N T H E U T I L I Z A T I O N O F F L U O R I N A T E D AND DEOXYGENATED SUBSTRATE ANALOGUES B Y PHOSPHOGLUCOMUTASE.  HA: INTRODUCTION.  II. A. 1: The Use of Enzyme-Substrate Binding Energy. Present theory on the manner in which enzymes catalyze chemical transformations with such efficiency, precision and specificity involves the use of enzyme-substrate bmding energy. The great specificity of enzymes is made possible by the favorable binding interaction energy between the active site and its specific substrates and the unfavorable binding or exclusion of incorrect substrates. It is believed that some of the intrinsic bmding energy of the enzymesubstrate complex, which is the maximum possible binding energy of a substrate or functional group, is utilized by the enzyme to effect catalysis by lowering the activation energy of the reaction. This means that the observed enzyme-substrate bmding energy, as measured by the dissociation constant, is what remains after part of it has been utilized. This can be probed by modifying the substrate and deterrmning its enzymic catalytic parameters. A classic example of this is the study carried out on the hydrolysis of a series of synthetic peptides by the enzyme pepsin. All the peptides were found to have a K that varied within a factor of about four; m  whereas the values of kcat ranged by a factor of approximately one thousand (Sachdev & Fruton, 1975). The alternative approach to the modification of substrates for mvestigating the effects on Km and kcat is the modification of the enzyme itself. By the use of site specific mutagenesis, selected amino acids may be changed so that the new mutant enzyme lacks a side chain involved in binding the substrate. This approach has been applied to mechanistic studies of the enzyme tyrosyl-tRNA synthetase which catalyzes a two step reaction resulting in the aminoacylation of tRNA. The first step of the reaction involves the formation of a stable, enzyme-bound, aminoacyl adenylate complex, the structure of which has been determined by X-ray diffraction.  15  E + Tyr + ATP  • E-Tyr-AMP + PPj  Mutant enzymes have been prepared in which residues involved in hydrogen bonding with the substrate (determined by analysis of the crystal structure) were replaced by residues to which no hydrogen bonding is possible. When the catalytic efficiencies of the mutant enzymes were measured, it was found that only small increases in K occurred, but these were accompanied m  by large decreases in kcat. The deletion of another side chain, which did not interact with either substrate but was positioned to make a hydrogen bond with the proposed reaction transition state, resulted in an even larger decrease in kcat but had no effect on the K of either substrate m  (Winter et al, 1982). This last experimental observation is evidence (along with many tightbinding transition-state analogues) that enzymes have evolved to bind the transition state of the reactiontighterthan the substrate itself. When the enzyme active site is complementary in structure to the resultant transition state, then kcat/K , th second order rate constant for e  m  catalysis, is maximized. In fact, contrary to what would be intuitively expected, the maximal catalytic rate of an enzyme occurs when the K is maximized, i.e., substrate binding is m  weakened. This is subject to kcat/Km being kept constant which occurs if the enzyme and transition state are complementary. From the equation (Fersht, 1985) kqat v = [E][S]^m where E is the concentration of free enzyme and S that of the substrate, it can be seen that if the substrate's K is high, the concentration of free enzyme will be maximized as will the rate of m  reaction. However the enzyme can only go so far in weakening the binding of the substrate before this will be matched by a weakening of transition state binding. Three mechanisms have been postulated to account for ways in which enzymes convert binding energy into chemical activation energy and how K is maximized. m  1. Strain. The concept of the enzyme's active site being complementary to the transition state rather than the substrate was proposed by Haldane (1930) and Pauling (1946). On binding, the substrate is strained towards the structure of the transition state, thus K is increased due to the m  16  substrate distortion. The modern version of the strain mechanism postulates a much stronger enzyme-transition state interaction than that of the enzyme and substrate. This means that the maximum binding interaction is not attained until the transition state is reached. 2. Induced fit. This concept of Koshland (1958) was formulated to account for specificity in simple enzymes such as phosphoglucomutase and hexokinase. In this case the enzyme does not have a structure complementary to that of the transition state in the absence of substrate. However, on formation of the enzyme-substrate complex, the conformation of the enzyme is changed such that enzyme-transition state complementarity is achieved. Therefore the binding energy is used to distort the enzyme resulting in an increased Km- This theory has been used to account for the greater than lCjlO fold difference in the rate of transfer of the enzymic phosphate group of phosphoglucomutase to glucose 1-P, over that of the chemically similar hydroxyl of water. 3. Nonproductive Binding. This concept accounts for the relative reactivities of different sized substrates. Larger substrates bind in the productive mode whereas smaller substrates bind more weakly in both the productive mode and in other nonproductive modes thereby lowering Km and kcat- However it does not provide a mechanism for the utilization of enzyme-substrate binding energy. Investigations on individual enzymes have provided evidence for all three of the preceding concepts. However, since an enzyme's specificity for competing substrates is determined by the ratio kcat/Km, neither mechanism can determine specificity, as each alters kcat and K in a compensating manner without altering kcat/Km itself. m  II.A.2: The Source of Enzyme-Substrate Binding Energy and its Measurement Due to the importance of enzyme-substrate binding energy in catalysis, much interest has centered on both the sources of such interactions and how they can be measured. Four types of interactions, all electrostatic in nature, act to hold the enzyme-substrate complex together (Fersht, 1985).  17 1. Coulombic interactions. These forces arise from an interaction between charged or dipolar molecules. Four varieties of Coulombic interactions exist, the energy of each having a different dependence on the distance (r) between the two centers and the dielectric constant of the medium (D). 1. Between ions of opposite charges (1/Dr) 2. Between randomly oriented permanent dipoles (l/Drfy 3. Between an ion and an induced dipole (1/Dr ) 4  4. Between a permanent dipole and an induced dipole (l/Dr^) 2. Van der Waals forces. These non-polar interactions are due to instantaneous dipoles caused by local fluctuations in electron density. Large polarizable atoms have large Van der Waals interactions whereas little interaction of this type occurs between small atoms such as hydrogen and oxygen. These forces are generally weak, being proportional to l/r^, but the large number of Van der Waals forces over a whole enzyme-substrate complex make the resultant attractive force significant. 3. Hydrogen bonds. Elements with an electronegativity of 3.0 or greater on the Pauling scale are judged able to form a hydrogen bond (Barnett, 1972). Hydrogen bonds occur between two electronegative atoms, one of which is bonded to a proton. The largest interaction occurs when the arrangement is linear, although only small energy losses are caused by bending. The length of the bond of the hydrogen to the atom with which it is formally bonded is the normal covalent distance, but the second bond to the other electronegative atom is shorter than the normal Van der Waals contact distance. The total energy of a hydrogen bond has been estimated at between 12 and 38 kJ/mol (Fersht, 1985). A hydroxyl functional group will be able to form three hydrogen bonds, two through the oxygen (one on each lone pair) to other X-H groups and atiiirdthrough its own hydrogen atom. Hydrogen bonding is believed to be a major force involved in enzyme-substrate interactions (Fersht et al., 1985).  1 8 4. Hydrophobic bonds. These are responsible for the transfer of non-polar compounds from a polar to a nonpolar environment. This is due to the entropically disfavored structuring of water molecules in the vicinity of the non-polar compound. The hydrophobic molecule is driven to the hydrophobic region of a protein by the destructuring of the water.  Measurement of binding energies in enzyme-substrate complexes. Interactions between non-bonded atoms such as those noted above may be calculated or extrapolated from various calorimetric or hydrophobicity measurements. This may allow the calculation of the energy of interaction between two single molecules. However in general this is not possible for enzyme-substrate interactions because binding is an exchange process. In each case (hydrogen bonding etc.) the substrate exchanges its solvation shell of water for the binding site of the enzyme. Similarly, the enzyme binding site will also be solvated. Consequently, the net binding energy is the difference between the bmding energies of the substrate and enzyme with water and that of the enzyme with substrate. In many cases it is possible to use the kinetic parameters ( K , kcat, Kj) of a ligand with m  the enzyme to measure free energy changes associated with binding interactions. The intrinsic free energy of binding of a group B in a ligand A - B (AAG°B) can be determined by comparison of the free energies of binding of A and A - B , i.e., AAG°B=AG°AB-AG°A where AG° = -RT ln K and K is the enzyme-substrate dissociation constant. This is possible since the translational and rotational entropy decrease on the binding of A will be only slightly different from that on binding of A - B , especially if B is a small functional group. Free energies of binding of a specific group may be determined in this manner only if the binding of one substituent does not affect the binding of another, by for example, reducing strain or forcing a tighter fit. Thus the mtrinsic binding energy of B will be free from entropic complications only if A and A - B do not show differences in strain or rotational or translational entropy loss on  19 binding to the enzyme (Jencks, 1975,1981). Therefore the contribution of the substituent B in a substrate A - B to binding will be equal to  MG-  B  = R T l n ^  However, as discussed earlier in this chapter, enzymes can use binding energy to raise kcat rather than decrease K , and that kcat/Km is maximized when the enzyme is complementary m  to the transition state. It can be shown (Fersht, 1985) that  A A G ° t B = - R T l n (Kcat/^m)A V ^? : (  at/  l)A  B  l  where - A A G ^ B is the incremental Gibbs free energy of transfer of B from the enzyme (at the transition state) to water. The smallest functionality that can be substituted for the group B is a f  hydrogen atom. Therefore the binding free energy of the functional group B may be evaluated by comparison of the kinetic parameters of the analogue A - H with that of the native substrate A-B. However the effect of deletion of a ligand's functional group and its replacement by either hydrogen or some other functional group can result in changes other than just the loss of binding free energy; for example, steric, electronic and conformational effects can be important. The change in the catalytic activity of the enzyme with the modified substrate can also be informative in the elucidation of the enzyme's mechanism. Modification of a ligand by substitution of hydroxyl or hydrogen by fluorine has become a popular means of this type of investigation (Barnett,  1972; Taylor, 1972; Penglis, 1981; Walsh, 1983).  n.A.3: The Effects of Ligand Modification. For an analogue to be of use in the determination of an enzyme's mechanism, it is important that the effects of modification of the ligand be understood. The rationalization of the observed effects of the analogue on the enzyme's catalytic activity is facilitated if the differences between the native substrate and the analogue are limited to a single effect. For example, the substitution of a hydroxyl group in a substrate by a fluorine atom may result in steric, conformational, electronic and hydrogen bonding differences compared to the native substrate. If all these effects are found to be operating in the analogue, then elucidation of the effect which  20  actually causes the change in the enzyme's catalytic activity would be difficult. Substitution of a hydroxyl group by fluorine or hydrogen will manifest itself in the four types of effects noted above. The results of each type of effect on enzymic systems is discussed in the remainder of this introduction. 1. Steric effects. Fluorine is a small element having a Van der Waals radius only slightly larger than that of hydrogen (Table ILL). The slight reduction in size of both the Van der Waals radius and the carbon-fluorine bond length makes fluorine substitution (for hydroxyl) a conservative change.  Group  Bond Length (A)  Van der Waals radius (A) Total (A)  C-H  1.09  1.20  2.29  C-F  1.39  1.35  2.74  C-O (C-OH)  1.43.  1.40  2.83  C-OH  1.43  2.10  3.53  Table II. 1: Comparison of the sizes of some functional groups. Data from Walsh (1983). Comparison of the total group size of fluorine with that of hydrogen shows a small increase of approximately 0.5 A. The ability of an enzyme to distinguish C-F from C-H has been investigated by the stereochemical analysis of the product of an enzymic reaction on fluorine containing substrate analogues. It appears that citrate synthase can distinguish C-F from C-H whereas malate synthase cannot. A similar analysis, involving glycerol kinase, in which fluorine was substituted for a hydroxyl group showed that the C-F moiety was treated as a C-OH rather than a C-H equivalent (Walsh, 1983). The reduction in functional group size caused by the substitution of a fluorine or hydrogen for hydroxyl would remove the possibility that any disruption of the enzyme-analogue interaction is due to a steric effect. However a water molecule may be able to fit in a gap  21  produced by deletion of a hydroxyl and its replacement by hydrogen, whereas this probably would not be possible in the case of its replacement by fluorine (Street et ah, 1986). 2. Electronic effects. Ruorine is the most electronegative element (4.0 on the Pauling scale, Pauling, 1960), being considerably more electronegative than oxygen (3.5) and even more so than hydrogen (2.1). The results of this are manifested in the types of chemistry which a C-F fragment can undergo. The C-F bond strength is very high and is only cleaved very slowly by normal SN2 type displacements. Enzymes that do cleave C-F do so by a /p HF (or C02, F ) elimination or -  formation of fluorohydrins which decompose to fluoride and the carbonyl compound (Walsh, 1983). The substitution of fluorine for a proton would suppress any reaction involving a proton abstraction step, since F+ is unstable. Similarly, fluorine substitution for hydroxyl will suppress any oxidation to carbonyl since C=F does not exist and hence aldol chemistry will be +  suppressed. The mechanisms of many enzymes have been elucidated by incorporation of fluorine at a site adjacent to the reaction center. Production of a carbanion may result in the elimination of fluoride ion whereas carbonium ion formation is greatly slowed down due to the inductive effect of the adjacent fluorine. Conversely substitution of hydroxyl by hydrogen will increase the rate of carbonium ion formation. The inductive effect of fluorine will also alter pKa values of nearby acids, bases and hydroxyl groups. 3. Conformational effects. It is essential that the conformation of the fluorine or hydrogen-substituted analogue be the same as that of the parent species for the analogue to be of use. Any differences in conformation of the analogue may bring about unknown steric effects which would complicate conclusions drawn. The solution conformations of monosaccharides are readily determined by analysis of *H or 19p.  nmr  spectra and use of the Karplus equation (1963). The major conformational effect of  fluorine substitution for a hydroxyl is observed at C-l (Hall & Manville, 1969). These authors observed that the fluorine took up an axial orientation in a series of fJ-per-acetylated and perbenzoylated pentopyranosyl fluorides. The change in conformation from the normally more  stable ^Ci to the less stable IC4, in which there are severe destabilizing 1,3 diaxial interactions, is due to the strong anomeric effect of fluorine caused by the large dipole moment of the C-F bond. However the same series of P-hexopyranosyl fluorides took the normal conformation, presumably due to the destabilizing effect of an axial hydroxymethyl group (Hall et al., 1969). A study on the conformation of 2-fluoro-2-deoxy-p-D-mannopyranosyl fluoride determined by nmr and X-ray crystallography showed that the sugar took the normal ^Ci conformation, the ring being essentially undistorted (Withers et al., 1986b). On the basis of dipolar repulsions between the fluorine substituents and the ring oxygen, the species may have been expected to exist in a distorted conformation. The solution conformations of the analogues of glucose in which the C-1,2,3,4 and 6 hydroxyls were replaced by fluorine have been determined by *H and l^F-nmr (Phillips & Wray, 1971). No distortions from the normal ^Cl conformation were detected. 4. Hydrogen bonding effects. Since elements with an electronegativity greater than approximately 3.0 are believed able to form hydrogen bonds (Barnett, 1972), it would be expected that fluorine would be able to act as a hydrogen bond acceptor in the same way as oxygen does. In fact the strongest hydrogen bond known is that in the hydrogen difluoride ion, HF2". The energy of this bond has been evaluated as 240 kJ/mol or approximately tentimesas great as any other hydrogen bond (Pauling, 1960). However the ability of fluorine attached to carbon to form hydrogen bonds is not clear. Fluorine's high electronegativity causes electrons to betightlyheld and consequently any bonds formed by the sharing of electrons will be weak. A study of hydrogen bonding by fluorine-containing compounds by l^-nmr failed to show any change in chemical shift that could be attributed to hydrogen bonding (Muller, 1976). However hydrogen bonding through fluorine has been attributed as the cause of the relatively high boiling points of some fluorocarbons and the presence of intramolecular hydrogen bonds has been proposed for the existence of the exclusively gauche conformation of P-fluoroethanol up to a temperature of 60° (Hudlicky, 1976).  23 X-ray crystallographic studies have shown several instances of the involvement of fluorine in hydrogen bonding. The existence of a hydrogen bond is recognized by a reduction in the Van der Waals contact distance between the donor and acceptor pair. In a review of over 260 structures of small molecules containing a C-F fragment (Murray-Rust et al, 1983), nine hydrogen bonds involving fluorine were detected, most involving C-F—H-N type interactions, while only one hydrogen bond of the type C-F—H-0 was detected. The authors concluded that the C-F fragment was capable of forming significant hydrogen bonds, being similar in character, although weaker than those involving C-O- and C-N=. They suggested that these weak hydrogen bonds of C-F fragments may become significant in ligand-macromolecule interactions because of the more stringent geometrical constraints in these systems compared to those of small molecules. The results of an X-ray crystallographic study on 2-fluoro-2-deoxy-|3-Dmannopyranosyl fluoride showed several weak interactions involving fluorine. F-2 was shown to be involved in a weak intramolecular interaction with OH-3 and F-l was weakly bonded with the C-H proton of C-2 (Withers et al., 1986b). Normally C-H protons are not involved in hydrogen bonding (as carbon has a Pauling electronegativity of 2.5) but in this case the proton is probably activated by the two electronegative fluorine substituents. The formation of hydrogen bonds by C-F fragments in biological systems has been indicated by several studies. In a study of the substrate activity of a series of fluoro and deoxy galactose analogues for the enzyme galactokinase, Thomas et al. (1974) found that 2-fluorogalactose bound five fold weaker than the parent substrate but 2-deoxy-galactose bound over a hundred fold weaker. These results were taken as evidence that the role of OH-2 is to accept a hydrogen bond. The fact that 2-fluoro-galactose bound strongly compared to its deoxy analogue is evidence for the formation of an acceptor hydrogen bond between the protein and F-2. The inhibition of the enzyme phosphorylase b by a complete series of fluoro and deoxy glucose analogues was analyzed in a similar manner (Street et al., 1986). From a comparison of the inhibition afforded by each fluoro and deoxy glucose pair, the authors were able to construct a hydrogen bonding network defining the types and strengths of hydrogen bonds between the enzyme and each hydroxyl group of the native inhibitor glucose. A comparison with the X-ray  24  crystallographic data of the enzyme-glucose complex showed that the distances between the hydrogen bonding pairs and the type of protein residue involved in the hydrogen bond corresponded well with the conclusions drawn from the inhibition data. II. A.4: The Energetics of Hydrogen Bond Formation. The formation of a hydrogen bond on protein-ligand bmcling does not result in a gain of energy equal to that of one hydrogen bond. This is because both the ligand and the enzyme are hydrogen bonded to water and therefore complex formation results in an exchange of hydrogen bonds. This can be expressed, for a single site of enzyme-ligand interaction as (Fersht, 1985), E-H- • • OH + 2  HOH —-A-L  "  [E-H-—A-L]  +  HOH-OHj  where E and L represent enzyme and ligand, and H and A are substituents providing donor and acceptor hydrogen bonds. Since the number of hydrogen bonds is unchanged on complex formation, the reaction is probably isoenthalpic. However, complex formation does result in an increase in entropy. Once the ligand is immobilized in the active site of the protein there will be no further decrease in entropy on the formation of further protein-ligand hydrogen bonds. Thus complex formation is favoured by the gain in entropy from the release of water molecules which were bound to the free protein and ligand. The energies of hydrogen bonds have been evaluated in a number of biological systems (Fersht, 1987a; Street etai, 1986; Bartlett & Marlowe, 1987). Fersht and coworkers, using mutants of the enzyme tyrosyl tRNA synthetase have evaluated the strengths of hydrogen bonds between uncharged donor-acceptor pairs as approximately 2-6 kJ/mol (0.5-1.5 kcal/mol). A much stronger interaction of 13-25 kJ/mol (3-6 kcal/mol) is formed when one member of the donor-acceptor pair is charged. The complementary approach, whereby the hydrogen bonding capacity of the ligand was removed by substitution of ligand hydroxyl groups by fluorine and hydrogen (Street et ai, 1986), gave similar values to those of Fersht. A rather different value of 17 kJ/mol (4 kcal/mol) for the binding energy of an uncharged hydrogen bonding pair was obtained by Bartlett and Marlowe (1987). They investigated the binding of two isosteric transition state analogues of zinc endopeptidase thermolysin. They found  25  that a series of inhibitors containing a phosphonarnidate group, (P02"-NH-) bound approximately 840 times tighter than the analogous phosphonate-containing series (P02"-0-). The difference in binding was attributed to the presence of a donor hydrogen bonding functionality (-NH-)in the phosphonarnidate series. No such hydrogen bonding is possible in the phosphonate series of inhibitors. X-ray crystallographic studies revealed only extremely slight conformational differences between the enzyme-inhibitor complexes of both inhibitors (Tronrud et al., 1987) and indicated that the phosphonarnidate group is neither doubly protonated nor zwitterionic. The authors suggest that this approximately four fold increase in hydrogen bond interactions over that observed by Fersht and coworkers, for a similar system, is due to the flexibility of side chains in the latter case and that in general the magnitude of hydrogen bonding may be reflected by the mobility of the interacting groups as much as by their charge. However, it has been pointed out that the oxygen of the phosphonate inhibitor can act as a hydrogen bond acceptor with water (but not the enzyme) and therefore as shown in the hydrogen bond inventory below, complex formation is no longer isoenthalpic. E-A-'-OHj +  HOH---A-L "  [E-A A-L]  +  HOH-'-OHj  The difference in the number of hydrogen bonds on each side of the equihbrium causes the large decrease in affinity of the enzyme for the phosphonarnidate inhibitor (Fersht, 1987b). In the case of a functional group modification such that no hydrogen bonding is possible {e.g., replacement of hydroxyl by hydrogen), then complex formation will be isoenthalpic. The poorer binding that results from the deletion of a ligand functional group which participates in a hydrogen bond is therefore not equal to a full hydrogen bond but is due to the difference in hydrogen bond strengths formed with the protein and with water.  II.A.5: Summary and Overview. This introductory section has briefly described how enzymes utilize enzyme-substrate binding energy to bring about their chemical transformations. The different types of forces behind this binding interaction and the ways in which these forces can be probed have been discussed. This chapter describes how two series offluorinatedand deoxygenated glucose 1-P  26  analogues have been used to probe the hydrogen bonding interactions of the phosphoglucomutase-glucose 1-P complex. The first series of substrate analogues are those in which a single "non-acceptor" (i.e., not C-l or C-6) hydroxyl has been replaced by fluorine and hydrogen. These analogues, which were available in our laboratory, have been used to probe the enzyme-substrate transition state complex. The (lifferences in the substrate activities of each of these analogues has enabled some proposals to be made on the mechanism of the reaction. The second series of fluorinated and deoxygenated substrate analogues, in which two sugar hydroxyls have been replaced, was synthesized for this study. These inhibitors provided information on the interactions in the enzyme-substrate ground state complex. The synthesis and conformational analysis of these analogues is described.  27 II.B: RESULTS AND DISCUSSION. n.B.l: The Synthesis of Fluorinated and Deoxygenated Analogues of Glucose 1-Phosphate. The synthesis of the series of disubstituted fluoro and deoxy glucose 1-P analogues in which the C-6 hydroxyl is replaced by fluorine and one glucose ring hydroxyl is replaced by either fluorine or hydrogen was conducted by the use of standard synthetic procedures. Each of the novel products and intermediates was characterized by ^H, 19p and 31p-nmr (where applicable) and in many cases melting points were determined. The P-peracetate of each disubstituted sugar and the final products was also characterized by elemental analysis. Many of the starting materials used in these syntheses were intermediates prepared for the synthesis of each fluoro and deoxy-glucose 1-P by Mr. I. Street (U.B.C.) or provided by Dr. D. Dolphin.  II.B. 1.1: 26-Dideoxy-2,6-difluoro-a-D-glucopyranosyl [bis(cyclohexylammonium) phosphate] >  , (6), (2,6-difluoro-glucose 1-P). The synthetic route used for the preparation of thetitlecompound is given in Figure ILL Trifluoromethyl 3,4,6-tri-0-acetyl-2-deoxy-2-fluoro-a-D-glucopyranoside (1), (Adamson et al, 1971) was deacetylated with sodium methoxide in methanol to give (2). Direct fluorination of this partially protected compound with cUethylaminosulfur trifluoride (DAST) in dichloromethane by the method of Card and Reddy (1983) gave the 2,6-difluoro derivative (3) in 56% yield. The trifluoromethyl group was removed by acid hydrolysis by refluxing with Dowex 50W-X8 (H ) +  resin to afford the free sugar (4). Acetylation of 2,6-dideoxy-2,6-difluoro-D-glucose (4) by reaction with acetic anhydride and pyridine gave an anomeric mixture of the peracetates. Quenching of the acetic anhydride with methanol and removal of the solvents gave a gum which was brominated at C-l by 45% hydrogen bromide in glacial acetic acid containing 10% v/v acetic anhydride (Wolfram & Thompson, 1963a). The bromination reaction was slow, requiring eight hours for completion. An extractive work up gave the a-bromide which was converted to the P-per-acetate (5) by reaction with mercuric acetate in glacial acetic acid. The overall yield from (4) was 71%. The P-per-acetate was phosphorylated by fusion with anhydrous phosphoric  28  acid (MacDonald, 1972) followed by saponification of the acetate groups by lithium hydroxide. The fusion reaction required nine hours for optimum yield. After conversion to the biscyclohexylammonium form the salt (6) was isolated in 61% yield.  Figure HI: Synthetic route for the preparation of'2,6-dideoxy-2,6-difluoro-a-Dglucopyranosyl [bis(cyclohexylammonium) phosphate].  II.B.1.2: 3,6-Dideoxy-36-difluoro-a-D-glucopyranosyl [bis(cyclohexylammonium) phosphate] >  , (14), (3,6-difluoro-glucose 1-P). The route used for the synthesis of the title compound is given in Figure n.2. Controlled acid hydrolysis of 3-deoxy-3-fluoro-l;2:5;6-di-0-isopropylidene -a-D-glucofuranose (7) (Withers et al., 1986a) using dilute sulfuric acid (Schmidt, 1963) resulted in the selective removal of the 5;6-0-isopropylidene group. Attempts were made to selectively fluorinate C-6 of the partially protected (8) with DAST in dichloromethane by the method of Card and Reddy (1983). However no difluorinated species was obtained and the products were shown to consist of an equal mixture of the endo and exo isomers of the 5,6-cyclic sulfite derivatives (15,16).  29  Ph CO—i _| o 3  Ph CCl pyridine  R 0  3  I. CH 0CH C1, iPrEtN II. HCOOH, Et^O 3  2  2  HO—KJ?2  F—j CH OCH 0— O,  F  3  (12)  m  X  2  THF, TFA, H 0  x  2  OH  2  I. Ac 0, pyridine II. HBr,HOAc HI Hg(OAc)  (11)  HO—. CH OCH 0—| O DAST ™ 3  (O 0--^\  2  P  2  2  \sA^OAc  AcO  n. LiOH  (13)  Figure H2: Synthetic route for  O—p-o(14)  ±_  the preparation of'3,6-dideoxy-3,6-difluoro-a-D-  glucopyranosyl [bis(cyclohexylammonium) phosphate].  Such by-products have been previously observed in the reaction of DAST with compounds containing a vicinal diol (Baillargeon & Reddy, 1986). However the results were surprising in this case since fluorination of 1,2-O-isopropyhcbne-a-D-glucofuranose in the same manner gives the 6-fluoro derivative in 70% yield (Card & Reddy, 1983). This reaction was confirmed in this study. Consequently it would appear that the C-3 hydroxyl plays an important role in the fluorination of C-6, presumably reacting with DAST and providing a source for the  30 intramolecular delivery of fluoride to the C-6 position. Since the 3-fluoro derivative cannot form this intermediate species, the side reaction involving the formation of the 5,6-cyclic sulfite is favored (Figure II.3).  (15,16) Figure II.3: Mechanism of the reaction of DAST with 1 ;2-0-isopropylidene-a-Dglucofuranose and 3-deoxy-3-fluoro-l;2-0-isopropylidene-a-D-glucofuranose.  31  Presumably reaction of the allose derivative (the C-3 epimer of glucose) would result in the same 5,6-cyclic sulfite product. In order to remove the possibility of 5,6-cyclic sulfite formation it was necessary to selectively protect the C-5 hydroxyl, leaving the C-6 hydroxyl unprotected for the fluorination reaction. This was achieved by protection of the C-6 hydroxyl as its triphenylmethyl ether by reaction of (8) with triphenylmethyl chloride in pyridine (Barker, 1963). This was followed by reaction of (9) with chloromethyl methyl ether and cWsopropylethylamine in dichloromethane (Hanessian et al, 1984) to afford the methoxymethyl (MOM) ether at C-5. This reaction was particularly slow,requiringten days for the complete reaction of starting material. The slow rate of reaction is probably due to the steric hindrance afforded by the tertiary carbon (C-4) and the large triphenylmethyl group at C-6. Selectiveremovalof the triphenylmethyl group by formic acid in diethyl ether (Bessodes et al., 1986) gave the partially protected compound (10). Fluorination of (10) with DAST and 2,4,6-trimethyl pyridine (TMP) in dichloromethane (Withers et al., 1986a) gave the difluorinated species (11) in 63% yield. The protecting groups wereremovedusing a mixture of trifluoroacetic acid (TFA), tetrahydrofuran (THF) and water (Dauben et al, 1982) to yield 3,6-dideoxy-3,6-difluoro-D-glucose (12) as a non crystallizable gum. The B-per-acetate (13) was preparedfrom(12) in the same manner as for the analogous 2,6-difluoro compound (5), the bromination reactionrequiringonly one hour for completion. Phosphorylation of the B-per-acetate (13) for 3 hours according to MacDonald (1972) followed by saponification and crystallization as the bis-cyclohexylammonium salt gave the product (14) in 88% yield.  II.B.1.3: 3,6-Dideoxy-6-fluoro-a-D-ribohexopyranosyl [bis(cyclohexylammonium) phosphate], (26), (3-deoxy-6-fluoro-glucose 1-P). The synthetic route used for thetitlecompound is shown in Figure H.4. In light of the production of the 5,6-cyclic sulfite by-products observed in the fluorination of (7), a similar route to that finally used for the fluorination of (7) was chosen, in which all the hydroxyls except that of C-6 were protected prior to fluorination. Thus 3-chloro-3-deoxy-l;2:5;6-di-0-  32 isopropylidene-a-D-glucofuranose (17) (Street et al, 1986) was selectively hydrolyzed (Schmidt, 1963) to produce the diol (18). HO—1  Nfe CCOO—, 3  HO—  (17)  HO—I  (18)  (19) CH 0CH C1 3  CH OCH 0—| 3  2  3  3  2  PAST  2  MeOMe MeOH  TMP  (22)  TFA, THF, H 0  2  CH OCH C  CH OCH <  O  iPr EtN  Nfe3CCOO—1  HO—,  F—j  2  (20)  (21)  2  H  0  CHjF  I. Ac^O, *.™^ ™.pyridine FJ  II. HBr, HOAc ^  A  d  r ^ C ^ P C l ^ — ^  AIBN  AcO—  CH F 2  Bu.SnH  m. Hg(OAc)  2  )H  (23)  (25) (24)  L H P0 H. L D H 3  4  (26)  Figure II.4:  Synthetic route for the preparation of3,6-dideoxy-6-fluoro-a-D-  ribohexopyranosyl [bis(cyclohexylammonium) phosphate].  Selective esterification of C-6 by reaction with trimethylacetyl chloride in pyridine followed by protection of C-5 as its methoxymethyl ether gave the fully protected compound (20). Trimethylacetyl chloride was used to protect the C-6 hydroxyl in order to increase the rate of formation of the methoxymethyl ether at C-5 relative to that when the C-6 hydroxyl is protected  33  as the triphenylmethyl ether. Deprotection of the C-6 ester with methoxide gave the suitably protected compound (21) which was fluorinated with DAST in the presence of TMP (Withers et al, 1986a) to give the 6-fluoro-3-chloro derivative (22) in 49% yield. Removal of the protecting groups by acid hydrolysis gave the free sugar (23) which was converted to the p-per-acetate (24) in the same manner as in the synthesis of (13). Reduction of the C-3 chloro group by reaction with tributyltin hydride using aa'-azobisisobutyronitrile (AEBN) as the radical source gave the deoxy sugar (25) in 81% yield. The reduction was not performed until this stage due to the propensity of 3-deoxy-D-glucose to assume furanose forms.This would result in poor yields of the desired pyranose form of the per-acetylated sugar. Phosphorylation by fusion with anhydrous phosphoric acid, followed by saponification and crystallization as the biscyclohexylammonium salt gave the title compound (26) in 67% yield. II.B.1.4: 4,6-Dideoxy-4,6-difluoro-a-D-glucopyranosyl [bis(cyclohexylammonium) phosphate] , (34), (4,6-difluoro-glucose 1-P). The synthesis of thetitlecompound was achieved by two routes (shown in Figures 11.5 and II.6) with approximately equal efficiency. Reaction of methyl oc-D-galactopyranoside (27) with benzaldehyde and zinc chloride catalyst (Bell & Greville, 1955) gave the 4,6-O-benzylidene derivative (28). Acetylation of this compound with acetic anhydride and pyridine gave the fully protected species (29). Removal of the benzylidene group by treatment with 80% aqueous acetic acid (Arita et al, 1972) yielded (30). Reaction of the diol (30) with DAST and TMP in dichloromethane (Withers et al, 1986a) afforded the difluorinated product (31) in 31% yield. The methyl glycoside was converted to the a-chloride (32) by the action of a,a-dichloromethyl methyl ether and zinc chloride (Kovac et al, 1976). Conversion to the P-per-acetate (33) was accomplished by reaction with mercuric acetate in glacial acetic acid. Phosphorylation and saponification (MacDonald, 1972) followed by crystallization as the bis-cyclohexylammonium salt gave the title compound (34) in 58% yield. Alternatively, the difluorinated sugar (38) was obtained by the direct fluorination of the partially protected methyl 4-deoxy-4-fluoro-a-D-glucopyranoside (37) according to Card and  Reddy (1983). Thus methyl 23,6-tri-O-benzoyl-a-D-galactopyranoside (35) was fluorinated by reaction with DAST in the presence of 4-(dimemylamino)-pyridine (DMAP) in dichloromethane.  Figure n.5: Synthetic route for the preparation of4,6-dideoxy-4,6-difluoro-a-Dglucopyranosyl [bis(cyclohexylammonium) phosphate] using a singlefluorinationstep.  Deacetylation of the product (36) with sodium methoxide in methanol yielded the methyl glucoside (37). Huorination of (37) by reaction with DAST in dichloromethane gave the difluoro compound (38) in a yield of 67%. Acetylation of (38) with acetic anhydride and pyridine gave (31).  35  OH  Figure n.6: Synthetic route for the preparation of4,6-dideoxy-4,6-difluoro-a-Dglucopyranosyl [bis(cyclohexylammonium) phosphate] using twofluorinationsteps.  II.B.1.5: 4,6-Dideoxy-6-fluoro-a-D-xylohexopyranosyl [bis(cyclohexylanimonium) phosphate] (45), (4-deoxy-6-fluoro-glucose 1-P). The synthetic route used for the title compound is shown in Figure n.7. The synthesis of methyl 4,6-dideoxy-6-fluoro-a-D-xylohexopyranoside was attempted direcdy from methyl 4deoxy-a-D-xylohexopyranoside (39). However yields of less than 20% were achieved and the purification of the desired product from a major side product was very difficulL As noted in Section n.B.1.4 the direct fluorination of the 4-fluoro methyl glucoside (37) by the same method gives the difluorinated product (38) in a yield of approximately 70%. These results and those observed in Section n.B.1.2 show that the reaction of partially protected substrates with DAST is determined to a large extent by other ring substituents. Consequently each sugar exhibits a unique reactivity affording unique fluorodeoxy (or other) products. In order to reduce side reactions, the hydroxyls at C-2 and C-3 were protected prior to the fluorination step. Methyl 4deoxy-a-D-xylohexopyranoside (39) (Street et al., 1986) was selectively protected at C-6 by reaction with triphenylmethyl chloride in pyridine and acetylated in a "one pot" procedure by  36 addition of acetic anhydride. The triphenylmethyl ether containing compound (40) was cleaved with formic acid in diethyl ether (Bessodes et al., 1986) to give (41). Fluorination of (41) with DAST and TMP gave the primary fluoride (42) in 66% yield. Conversion to the a-chloride (43) was accomplished by reaction with cx,a-dichloromethyl methyl ether and zinc chloride (Kovac et al., 1976). Conversion to the P-per-acetate (44) was accomplished by reaction with mercuric acetate in glacial acetic acid. Phosphorylation according to MacDonald (1972) resulted in a low yield (25%) of the title compound (45) due to charring during the fusion reaction. The discoloration was removed by treatment with Norite prior to crystallization as the biscyclohexylammonium salt.  H. LiOH  Figure II.7: Synthetic route for the preparation of4,6-dideoxy-6-fluoro-a-Dxylohexopyranosyl [bis(cyclohexylammonium) phosphate].  37 II.B.1.6: 2-Deoxy-a-D-arabinohexopyranosyl [bis (ammonium) phosphate], (46), (2-deoxyglucose 1-P). The successful synthesis of 2-deoxy-a-D-arabinohexopyranosyl phosphate (2-deoxyglucose 1-P) by fusion of l,3,4,6-tetra-0-acetyl-2-deoxy-B-D-arabinohexopyranose  with  anhydrous phosphoric acid has been claimed by Shibaev et al. (1973). However attempts in our laboratory to prepare the title compound by this method were unsuccessful, the products consisting of a mixture of phosphate esters as determined by 31p-nmr. Anion exchange chromatography at neutral pH on this mixture failed to yield the desired product. Equally unsuccessful were attempts to add phosphoric acid across the double bond of 3,4,6-tri-O-acetylD-glucal (LP. Street & M.D. Percival, unpublished results). 2-Deoxy-glucose 1-P has also been prepared, but not isolated, by the reaction of glycogen phosphorylase with D-glucal (Klein et al, 1984). The difficulty in the preparation of this phosphate ester probably stems from its predicted extreme acid lability resultingfromthe replacement of the electronegative C-2 hydroxyl by hydrogen. Since acid catalyzed hydrolysis of glycosyl phosphates proceeds via C-O bond cleavage (Bunion & Humeres, 1969), the rate of hydrolysis will depend on the stability of the oxocarbonium transition state. Hydrogen at C-2 will not destabihze this transition state so effectively as hydroxyl and in fact 2-deoxy-glucosides are hydrolyzed approximately 2000 fold faster than the corresponding glucosides (Mega & Matsushima, 1983). Due to the extreme lability of the target compound, an alternative synthetic route involving the use of enzymes at neutral pH was employed. It has been shown that 2-deoxy-glucose 6-P is a substrate of phosphoglucomutase (Egyud & Whelan, 1963) and thus 2-deoxy-glucose 1-P could be preparedfromcommercially available 2-deoxy-glucose 6-P by the action of phosphoglucomutase. However, since the equilibrium of the interconversion of such sugar phosphates lies heavily in favor of the C-6 linked phosphate and separation of the two isomers would be extremely difficult, a two step procedure in which the reaction was pulled to completion was employed. This second step is based on the fact that 2-deoxy-glucose can be converted into its uridine-diphospho derivative by partially purifiedtissuehomogenates  38 (Schwartz & Schmidt, 1976). This implies that 2-deoxy-glucose 1-P is a substrate for the enzyme uridine-diphosphoglucose pyrophosphorylase. Thus 2-deoxy-glucose 6-P (45) was converted by the action of phosphoglucomutase into 2-deoxy-glucose 1-P (46), which was then converted to uricline-diphospho-2-deoxy-glucose (47) by uricline-diphosphoglucose pyrophosphorylase and uridine 5'-triphosphate (UTP). A small amount of glucose 1,6-diP was included in the reaction mixture in order to satisfy the cofactor requirements of phosphoglucomutase. The reaction was pulled to completion by inorganic pyrophosphatase which catalyzes the hydrolysis of inorganic pyrophosphate (Figure n.8).  »  H O — K W > H  Phosphoglucomutase (45)  (46)  —O  Uridine-diphosphoglucose pyrophosphorylase, UTP H O - ^ C H  H 0  2  O H  A_T \ <  (47)  Uridine  H O — f — O " O "  +  Inorganic pyrophosphatase  HO—P—-O" i -  Figure II.8: First step of phosphate.  -o-LJLo i o-  the enzymic synthesis of2-deoxy-a-D-arabinohexopyranosyl  39 After completion of the reaction, which was monitored by the release of inorganic phosphate, the products were chromatographed on a DE-52 cellulose anion exchange column at 4° C and pH 8.0. The desired product, the first UV absorbing and phosphate-containing species to be eluted, was then treated with lithium hydroxide to pH 13. This step was included in order to precipitate inorganic phosphate produced by the reaction which was not separated from the desired dianionic product by the chromatography. The high pH also causes the decomposition of any uridine diphosphoglucose into uridine 5'-monophosphate and glucose 1,2-cyclic phosphate by the intramolecular attack of the C-2 sugar hydroxyl on the phosphate moiety (Paladini & Leloir, 1952). Small amounts of uridine diphosphoglucose would arisefromthe added cofactor glucose 1,6-diP and probably in larger amounts from contaminating glucose 6-P present in the 2-deoxyglucose 6-P starting material. The 2-deoxy derivative, lacking a C-2 hydroxyl, would be unable to undergo this decomposition pathway and therefore be stable at high pH. The alkaline solution was neutralized by addition of Dowex 50W-X8 (H ) resin to pH 7.6. 31p-nmr (proton +  decoupled) of this product showed two doublets (coupled by 21 Hz) having a chemical shift almost identical to that of uridine diphosphoglucose. The reformation of the desired product, 2deoxy-glucose 1-P (46), was achieved by treatment of the uridine diphospho-2-deoxy-glucose (47) with more uridine diphosphoglucose pyrophosphorylase plus a large excess of inorganic pyrophosphate (Figure n.9).  Figure n.9: Second step of the enzymic synthesis of2-deoxy-a-D-arabinohexopyranosyl phosphate.  40  The equilibrium constant of the interconversion of uridine diphosphoglucose and glucose 1-P plus UTP is close to unity (Biochemica Information, Boehringer Mannheim Inc.,1973) and thus the addition a large excess of inorganic pyrophosphate would push the equilibrium over in favour of glucose 1-P. It was assumed that any difference in the equilibrium constants of the reaction of uridine diphosphoglucose and its 2-deoxy analogue would not be large. The progress of the reaction was monitored by 31p-nmr. On completion, the solution was re-treated with lithium hydroxide to precipitate any inorganic phosphate produced by hydrolysis of the product. Only a slight cloudiness was observed. After filtration and neutralization the mixture was chromatographed as described above, elution of the product being monitored by the appearance of phosphate according to the method of Bartlett (1959). The product was successfully characterized by *H and 31p-nmr, the data being consistent with those previously published (Shibaev et al., 1973; Klein et al., 1984); however, a satisfactory microanalysis could not be obtained. n.B.2: Analysis of the Solution Conformation of Glucose 1-Phosphate Analogues. The solution conformation of ct-glucose 1-P has been elucidated by a combination of *H and l^C-nmr (O'Connor et al, 1979). An analysis of the *H trans-diaxial coupling constants gave values of between 9 and 10 Hz, similar to those of glucose, thus indicating an undistorted ^Ci conformation. The coupling between P and H-2 of 1.8 Hz and the results of  I ^ C - I H  coupling constants indicate that the phosphate takes up a preferred orientation trans to C-2 in aglucose 1-P. The *H, 1 F and 31p chemical shifts and coupling constants of each disubstituted 9  glucose 1-P analogue and 2-deoxy-glucose 1-P are shown in Tables II.2, n.3 and II.4. The configuration of the phosphate group'of each analogue is confirmed as being a (axial) by the small vicinal coupling of 3.0-3.6 Hz between H-l and H-2. The conformation of each phosphate group is probably the same as that of a-glucose 1-P since a coupling between H-2 and P of 2.02.2 Hz is observed in each case.  41 Substituted  Chemical shift 8(ppm)  glucose 1-P  H-l  H-2  H-3  H-4  H-5  H-6  H-6'  2,6-difluoro  5.60  4.35  u  3.55  u  4.74  4.68  235.9 201.0  -4.63  3,6-difluoro  5.45  u  u  u  4.00  u  u  235.2 202.1  -4.67  4,6-difluoro  5.44  3.52  4.02  4.42  4.46  4.63  4.63  236.1 200.2  -4.76  3-deoxy6-fluoro  5.39  u  1.90  u  3.89  4.70  4.64  235.6 2.18  -4.58  4-deoxy6-fluoro  5.51  3.41  4.27  4.62  4.48  230.3 2.02e  2-deoxy  5.55  1.65  u  u  u  u  a  e  a  4.03 1.58a u  u  F-6  H-X  P  e  2.15  e  -4.61 -1.28  Unassigned due to peak overlap. Axial proton in methylene system. Equatorial proton in methylene system.  Table H.2: H, F and P -nmr chemical shifts of substituted a-D-glucopyranosyl 1  19  31  phosphates. The conditions under which the spectra were measured are detailed in Section II.C.l. Substituted  iH-iHcoui pling constants (H[z)  glucose 1-P  Jl,2  J2,3  2,6-difluoro  3.6  3,6-difluoro  3,4  J4,5  J5,6  J5,6'  J6,6'  9.0  9.5  9.5  2.9  1.5  10.0  3.5  u  u  10.8  u  u  u  4,6-difluoro  3.3  10.0  9.0  9.0  u  u  u  3-deoxy6-fluoro  3.0  4.8  4.8C  10.0  3.0  1.5  10.0  12.0  12.0  4-deoxy6-fluoro  3.4  9.9  5.0C  2.5C  4.0  1.5  10.0  12.0  12.0  2-deoxy  u  5.0  u  u  u  u  u  12.0  12.0  C  T  Jyi  a C  Jgem  b  Undetermined due to signal overlap. Vicinal couplings with axial proton at deoxy center. h Geminal coupling at deoxy center. Coupling with equatorial proton at deoxy center.  u  a  c  Table II.3: ^H-^H coupling constants of substituted a-D-gluopyranosyl phosphates.  42 i H - ? and ! H - F coupling constants (Hz)  Substituted glucose 1-P  3 1  3  JP,H-1  4  Jp,H-2  1 9  2  JF,H-6  3  JF,H-5  2  JF,H  a  3  JF,H  b  JF,H  2,6-difluoro  7,8  2.2  48.0  u  50.0  u  3,6-difluoro  7.2  2.0  48.7  30.6  53.8  u  3.5b  4,6-difluoro  7.3  2.0  48.0  30.7  _  —  —  3-deoxy6-fluoro  7.1  2.2  48.0  30.7  4-deoxy6-fluoro  7.2  2.2  48.0  26.0  2-deoxy  6.8  2.0  u  a  b4  c  Undetermined due to signal overlap. Coupling with ring fluorine. JF-3,H-lJF-4,H-L 3  Table n.4: lH-31p and ^H-^F coupling constants of substituted a-D-glucopyranosyl phosphates.  Trans-diaxial couplings between the protons of C-2,3,4 and 5 of 9.0-10.8 Hz suggest a relatively undistorted 4 Q conformation of the sugar ring of each analogue. The l^F chemical shifts of each analogue are similar to.those of otherfluorinatedcarbohydrates (Penglis, 1981), fluorine at primary centers resonating approximately 35 ppm upfield from those at secondary centers. Geminal I^F-IH couplings of approximately 50 Hz were observed, whilst vicinal couplings ranged from zero to 31 Hz, the magnitude depending on the relative orientations of the coupled nuclei and the presence of electronegative substituents. The absence of a coupling between H-l and F-2 has been previously observed in 2-fluoro-glucose and has been attributed to the electronegativity of the ring oxygen (Phillips & Wray, 1971). The two long range couplings of F-3 and F-4 with H-l were also observed in the analogousfluoroglucose species by the same authors. The four bond coupling has been attributed to the W conformation of the two nuclei and the five bond coupling to the participation of the ring oxygen in the coupling pathway. The conformation of the other mono-substituted deoxy andfluoroglucose 1-P  43 analogues used in this study have been analyzed in the same manner and were found to exist in an undistorted C i conformation (Withers et al., 1986a, 1988). 4  II.B 3: Interaction of Fluoro and Deoxy Substrates and Inhibitors with Phosphoglucomutase.  n.B.3.1: Measurement of the Kinetic Parameters. Two different types of assay have been used to determine the kinetic parameters of each substrate ( K and V ax) and inhibitor (Ki) of phosphoglucomutase. A colorimetric assay, used m  m  for substrates, makes use of the difference in acid lability of the C-l and C-6 esters of sugar phosphates, the glucosyl phosphates (C-l linked) being fairly acid labile whereas their C-6 isomers are all acid stable. All enzymic assays were performed using glucose 1-P or analogues thereof as the substrate because of the ease of the determination of the rate of reaction in that direction. The reaction of glucose 1-P is linear withtimeto approximately 60% conversion, whereas that of the reverse clirection is curved at 20% conversion (Ray & Peck, 1972). This high degree of linearity for glucose 1-P results from the favorable equilibrium constant (17 :1) and the ratio of K values of substrate and product (5 : 1). As a first approximation this is probably true m  for other substrates although it has not been verified for the cases described in this thesis. In any case, product-time plots of the reaction of each substrate were measured to determine the linear region. The principle of the colorimetric assay involves the quenching of the reaction after a given period with acid followed by the hydrolysis of the remaining substrate to inorganic phosphate and determination of the phosphate with Fiske-Subbarow (1925) reagent. The product, glucose 6-P, is stable under the hydrolysis conditions. Subtraction of the optical density from that of a reaction blank, in which enzyme was added to acidified substrate, gives the amount of substrate consumed. Quantitation of the optical density change is achieved by construction of a standard curve. The colorimetric assays used in this investigation are those of Ray and Roscelli (1964) and Peck and Ray (1971). The fluorinated glucose 1-P analogues investigated were all found to be completely hydrolyzed under the assay conditions whereas the corresponding C-6 phosphates  44 were completely stable. The deoxygenated glucose 1-P analogues are all less acid stable than glucose 1-P (Withers et al, 1988) and hence the assays of Ray et al could be used with only rninor modifications. A coupled enzymic assay was used to determine the K and Vmax of the substrate m  glucose 1-P in the presence of inhibitors. The product of the isomerization, a-glucose 6-P, is oxidized by the enzyme glucose 6-phosphate dehydrogenase in the presence of NADP (0nicotinamide adenine dinucleotide phosphate) to produce gluconolactone 6-P and NADPH (the reduced form of NADP). The molar extinction coefficient difference of NADPH-NADP at 340 nm is 6220 and production of glucose 6-P is quantitated by the increase in absorbance at this wavelength. Glucose 6-phosphate dehydrogenase is specific for p%glucose 6-P so it is important that the rate of anomerisation is not the limiting factor in the assay. Controls were carried out to ensure that this was not the case. The coupled assay procedure used in this study is a modification of that of Lowry and Passonneau (1969). In all cases the substrates and inhibitors used in this study were present as their biscyclohexylammonium salts, with the exceptions of 2-deoxy-glucose 1-P and 3-fluoro-glucose I, 6-diP which were isolated as the bis-ammonium salts. The activity of phosphoglucomutase with bis-sodium glucose 1-P was compared with that of the bis-cyclohexylammonium salt. No evidence of any counter ion effect was detected. This was particularly important as each cyclohexylammonium salt had been isolated in the bis-lithium form prior to conversion to the cyclohexylammonium form and it has been shown that Uthium is a potent inhibitor of the phosphoglucomutase system (Ray et al, 1978). II. B.3.2: Substrate Activity of Fluoro and Deoxy Analogues. All of the members of the series of glucose 1-P analogues, viz., 2-, 3- and 4-deoxyglucose 1-P and 2-, 3- and 4-fluoro-glucose 1-P were found to be substrates of phosphoglucomutase. Previous studies (Egyud & Whelan, 1963) had shown that 2- and 3deoxy-glucose 6-P were both substrates of phosphoglucomutase although no further investigation of their kinetic properties was made. The substrate activity of 3-fluoro-glucose 1-P  45  and the corresponding 6-phosphate with phosphoglucomutase from Saccharomyces cervisiae has also been investigated (Wright et al., 1972). The authors found that 3-fluoro-glucose 1-P was not a substrate although 3-fluoro-glucose 6-P was found to be a weak competitive inhibitor with a Ki of 40 mM compared to a K for glucose 6-P of 60 nM. Whether this difference from m  the results of this study is due to the different source of enzyme or rather a problem with the coupling enzymes used in the system cannot be determined. To be an effective catalyst for the interconversion of glucose 1-P and glucose 6-P, phosphoglucomutase requires the presence of catalytic amounts (in the order of \iM) of a cofactor, glucose 1,6-diP. However in this study, the deoxy and fluoro-glucose 1,6-diP analogues corresponding to each deoxy and fluoro-glucose 1-P were not available. Instead glucose 1,6-diP was used as the cofactor for each substrate.  Glc 6-P  Glc 1,6-diP  Figure 11.10: The reaction ofphosphoglucomutase with a fluoro-glucose 1-P analogue and glucose 1,6-diP.  The use of glucose 1,6-diP instead of the deoxy or fluoro-glucose 1,6-diP cofactor has an effect on the initial sequence of reactions carried out by the enzyme. The modified pathway of the reaction of phosphoglucomutase with a fluoro or deoxy-glucose phosphate in the presence of glucose 1,6-diP is shown in Figure HIO. The following description of the reaction is given for a fluoro-glucose 1-phosphate but could be applied equally well to a deoxy substrate analogue. The initial reaction will involve the phosphorylation of fluoro-glucose 1-P (FGlc 1-P) by the phosphoenzyme (Ep) to produce the fluoro-glucose 1,6-diP (FGlc 1,6-diP) and dephosphoenzyme (ED) (Figure 1 1 . 1 0 , reactions 1 and 3 ) . Normally, in the reaction with glucose phosphates, glucose 1,6-diP is included in the reaction mixture at saturating levels. The rate of glucose 1,6-diP dissociation is slow compared to the rate of its dephosphorylation and formation of the alternate monophosphate (Ray & Roscelli, 1964). This results in glucose 1,6-diP only dissociating from the dephosphoenzyme about once in every twenty turnover cycles. However, in the case of the fluoro substrate, there is no corresponding fluoro-glucose 1,6-diP included in thereactionmixture and so the dephosphoenzyme-fluoro-glucose 1,6-diP complex will dissociate to produce dephosphoenzyme and fluoro-glucose 1,6-diP (Figure 1 1 . 1 0 ,reaction9). The dephosphoenzyme will be immediately rephosphorylated by the glucose 1,6-diP present to produce glucose 6-P (Figure n.10,reaction1 1 ) . This step is very rapid and would probably not be rate limiting (Lowry & Passonneau, 1969). The netresultof thisreactionis the conversion of glucose 1,6-diP into glucose 6-P and the coproduction of the fluoro-glucose 1,6-diP species. FGlc 1-P + Glc 1,6-diP  • FGlc 1,6-diP + Glc 6-P  This reaction will continue until the concentration of the fluoro-glucose 1,6-diPreachesa level at which the specificity, defined as kc^t/Km.[FGlc 1,6-diP], of phosphoglucomutase for fluoroglucose 1,6-diP is greater than its specificity for glucose 1,6-diP. At this stage the fluoro-glucose 1,6-diP will react with the dephosphoenzyme, one of its phosphate groups being transferred back to the enzyme to form the alternate fluoro-glucose monophosphate species and phosphoenzyme (Figure n.10, reactions 10,5 and 7). In order to examine the initial rate ofreactionof each fluoro and deoxy-glucose 1-P analogue with glucose 1,6-diP and phosphoglucomutase, experiments were performed in which  47 the time course of the reaction was monitored at two different concentrations of glucose 1,6-diP (0.04 and 2.5 uM), at a constant concentration of each fluoro and deoxy substrate analogue. Linear product-time plots were obtained for all the analogues except 3-deoxy and 3-fluoroglucose 1-P. The slopes of each pair of linear plots (except those of 3-deoxy and 3-fluoroglucose 1-P) were essentially the same at low and high glucose 1,6-diP concentrations. An example, shown in Figure ELI 1, is the time course of the reaction of 4-fluoro-glucose 1-P.  Time, minutes Figure ELI 1: Time courses of the reaction of 4-fluoro-glucose 1-P (25 pM) with phosphoglucomutase in the presence of varied concentrations of glfcose 1,6-diP. The  glucose 1,6-diP concentrations were; a, 2.5 |J.M;+, 0.04 ^ M ; * , 0.0 pM.  The rates of reaction {i.e., the slopes) at glucose 1,6-diP concentrations of 2.5 and 0.04 |iM are only slightly different, implying that a concentration of 0.04 jxM of 4-fluoro-glucose 1,6diP is sufficient to saturate the enzyme and produce a maximal isomerization rate. As noted above, the glucose 1,6-diP present initially is used to build up a saturating concentration of 4fluoro-glucose 1,6-diP. Obviously, if the initial concentration of glucose 1,6-diP is 0.04 uM, then the maximal concentration of 4-fluoro-glucose 1,6-diP that can be produced is also 0.04 uM. The concentration of phosphoglucomutase used in thesetimecourse studies was approximately 0.001 pM and thus the amount of fluoro-glucose 1,6-diP species that could be produced by the reaction of phosphoenzyme with monophosphate (Figure EL 10, reactions 1, 3  48 and 9) would be very low (approximately 0.001 |iM) and would have a minimal effect on the reaction rate. This was confirmed by the almost complete lack of reaction of each substrate in the absence of any added glucose 1,6-diP (Figure nil). Studies have shown that very high concentrations of phosphoenzyme can catalyze the isomerization of glucose monophosphates in the absence of glucose 1,6-diP because of the formation of the cofactor from the phosphoenzyme as described above. (Ray & Roscelli, 1964). Measurement of the reaction rates of 3-fluoro and 3-deoxy-glucose 1-P with two concentrations of glucose 1,6-diP (0.04 and 2.0 |iM) gave rather different results to those observed with the other substrate analogues. Typical product-time plots obtained for 3-fluoroglucose 1-P are shown in Figure 11.12. Similar plots were obtained for 3-deoxy-glucose 1-P.  Time, minutes Figure n.12: Time courses of the reaction of 3-fluoro-glucose 1-P (25 pM) with phosphoglucomutase in the presence of varied concentrations of glucose 1,6-diP. The  glucose 1,6-diP concentrations were.*4s 0.04 p.M; • , 2.0 |iM.  The time courses depicted in Figure n.12 show that a concentration of glucose 1,6-diP of 0.04 |iM produces a reaction rate approximately one eighth of that obtained by a concentration of 2.0 nM. This suggests that a 3-fluoro-glucose 1,6-diP concentration of 0.04 \JM is insufficient to saturate the enzyme and produce a maximal rate of isomerization. The evidence from the previous product-time plots suggests that the Km values of the deoxy and fluoro-glucose 1,6-diP  49  species (apart from 3-deoxy and 3-fluoro) are in the order of 0.01|iM (or less) since inclusion of 0.04 and 2.5 |iM glucose 1,6-diP both produce the same rate of isomerization. The K  m  of  glucose 1,6-diP is also 0.01 p.M (Peck et al., 1968). Conversely, it would appear that the K  m  values of both 3-deoxy and 3-fluoro-glucose 1,6-diP are well in excess of 0.01 ^M. The apparent rough equivalence of the K values of glucose 1,6-diP and the four other deoxy and m  fluoro-diphosphates is not unreasonable since Posternak and Rosselet (1953) reported that the Km of mannose 1,6-diP is about the same as glucose 1,6-diP. However a value of 3.7 \iM. for the K of mannose 1,6-diP has also been reported (Mulhausen & Mendocino, 1970) compared m  to a glucose 1,6-diP K of 0.1 uM. Thus the results of these two investigations are rather m  contradictory. The time course of the reaction of 3-fluoro-glucose 1-P with 2.0 (xM glucose 1,6-diP has a biphasic course (Figure n.12), the change in rates occurring after the consumption of approximately 2.0 (iM acid labile phosphate. It would appear that the enzyme is using glucose 1,6-diP to convert 3-fluoro-glucose 1-P into 3-fluoro-glucose 1,6-diP with the coproduction of glucose 6-P (Figure n.10, reactions 1,3,9 and 11). Thus it seems that the change in reaction rate occurs when the enzyme runs out of glucose 1,6-diP and the reaction changes from the production of 3-fluoro-glucose 1,6-diP to 3-fluoro-glucose 6-P (Figure n.10, reactions 1,3,5 and 7). In effect 3-fluoro-glucose 1-P is converted to 3-fluoro-glucose 6-P at a slower rate than that at which it releases glucose 6-P from glucose 1,6-diP. This same change in reaction pathway also occurs for the other fluoro and deoxy-glucose 1-P substrates (as will be shown later). This is suggested by the differences in the specificity constants (kcat/Km) of the enzyme for the two competing substrates, glucose 1,6-diP and fluoroglucose 1,6-diP. Although it appears that the K values of these two substrates are similar, it m  will be shown in Section H.B.3.3 that the Vmax (or kcat) values of the fluoro-substrates are much lower than that of glucose 1,6-diP. Consequently the enzyme must continue to utilize glucose 1,6-diP until it is almost gone. However, the absence of an observed change in rate when the glucose 1,6-diP is totally consumed suggests that the rate at which the C-2 or C-4 modified glucose 1-P analogues undergo isomerization (Figure n.10, reactions 1, 3,5 and 7) is  50 the same as that at which they are able to release glucose 6-P from glucose 1,6-diP (Figure n.10, reactions 1,3,9 and 11). These observations that a change of rate occurs only in the cases of 3deoxy and 3-fluoro-glucose 1-P and not the C-2 and C-4 modified substrates can be rationalized in terms of a different rate-determining step for the two classes of analogues. Thus production of the fluoro-glucose 1,6-diP species (Figure n.10, reactions 1,3 and 9) is rate determining (or at least no faster than reactions (10), 5 and 7) for the C-2 and C-4 modified substrates, whereas the production of the alternate monophosphate is rate determining for the C-3 modified substrates (Figure 11.10, reactions (10), 5 and 7). Further product-time plots of the reaction of phosphoglucomutase with the fluorosubstrates (a single concentration, 100 [IM) were measured with varied concentrations (2-20 |JM) of glucose 1,6-diP. Due to the postulated lower specificity of the enzyme for the fluoro-glucose 1,6-diphosphates, the enzyme must continue to utilize glucose 1,6-diP until this substrate is almost gone. Only at thistimewill the specificity of the enzyme for fluoro-glucose 1,6-diP be higher than for glucose 1,6-diP. In the cases of 2- and 4-fluoro-glucose 1-P, no change in rate was evident over the wholetimecourse. In each case the reaction was monitored to a loss of acid labile phosphate of 40 \IM. However, in the case of 3-fluoro-glucose 1-P, a large decrease in reaction rate occurred at a concentration corresponding to the amount of glucose 1,6-diP included in the reaction mixture. The experiments described in the previous paragraph were performed using the direct colorimetric assay. The same experiments were repeated using the coupled assay, a large amount of coupling enzyme (glucose 6-phosphate dehydrogenase) being included to allow for the reduced activity of this enzyme for the fluoro-glucose 6-phosphates. Essentially the same results were obtained as those noted above. Schematic representations of the spectrophotometer traces are shown in Figures n. 13 and 11.14.  51  15  Time, minutes Figure 11.13: Time courses of the reaction of 4-fluoro-glucose 1-P (50 pM) with phosphoglucomutase in the presence of varied concentrations of glucose 1,6-diP  (schematic representation). The glucose 1,6-diP concentrations were 5 and 10 \iM. Both plots followed essentially the sametimecourse.  15  0  10 Time, minutes  20  Figure 11.14: Time courses of the reaction of 3-fluoro-glucose 1-P (50 pM) with phosphoglucomutase in the presence of varied concentrations of glucose 1,6-diP  (schematic representation). The glucose 1,6-diP concentrations were 5 and 10 \iM.  The exception was the case of 2-fluoro-glucose 1-P. The rate at which 2-fluoro-glucose 6-P is oxidized by glucose 6-phosphate dehydrogenase is many fold less than that of the native  52 substrate glucose 6-P (Bessel & Thomas, 1973). Thus the coupling enzyme can discriminate between glucose 6-P and 2-fluoroglucose 6-P. This discrirnination was used to show that i  glucose 6-P is indeed produced until all of the glucose 1,6-diP has been used up, at which time the enzyme switches to the production of 2-fluoro-glucose 6-P. These results are shown in Figure 11.15.  0  10 Time, minutes  20  Figure 11.15: Time courses of the reaction of 2-fluoro-glucose 1-P (50 fiM) with phosphoglucomutase in the presence of varied concentrations of glucose 1,6-diP  (schematic representation). The glucose 1,6-diP concentrations were 5 and 10 \iM.  Time course experiments with 3-fluoro-glucose 1-P in which a saturating concentration of the synthetically prepared 3-fluoro-glucose 1,6-diP (5|iM) was included in the assay mixture, (no glucose 1,6-diP present) eliminated the break point and produced a monophasic plot, the rate of which depended on the concentration of 3-fluoro-glucose 1,6-diP. This confirmed that the biphasic plots were indeed due to the glucose 1,6-diP conversion to glucose 6-P as described above.  n.B.3.3: Determination of Substrate Kinetic Parameters. The K and V a x values of each deoxy and fluoro-glucose 1-P substrate were m  m  determined with phosphoglucomutase at pH 7.4 and 30° C using the colorimetric assay  53  procedure of Ray and Roscelli (1964). Each determination was made with a glucose 1,6-diP concentration of 1.0 |iM except those of the 3-deoxy and 3-fluoro analogues which are described later in this section. The assay interval was such that conversion of substrate to product was approximately 20% in all cases. A time course of the lowest substrate concentration used in the assay was found to be linear at least until this point. The results were plotted on double reciprocal (Lineweaver & Burk, 1934) plots in order to detect any deviationfromlinearity, none of which was observed. The values of K and Vmax were calculated by a computer program based on the m  statistical method of Wilkinson (1961). The determination of Km and Vmax values of 2-deoxyglucose 1-P was difficult due to the extreme lability of this substrate. The rate of hydrolysis of this compound was measured to be 2700timeshigher than that of glucose 1-P in 1.0 M perchloric acid (Percival & Withers, 1988). Assay intervals were limited to one to two minutes in order to minimize hydrolysis which would both reduce the substrate concentration and produce the competitive inhibitor inorganic phosphate. This, combined with the difficulties involved in measuring the concentration of the extremely labile phosphate monoester in the presence of a background of inorganic phosphate, means that the results of the assay are very approximate and should probably only be interpreted in terms of a "range finder". The kinetic parameters of 3-deoxy and 3-fluoro-glucose 1-P were determined by measuring the extent of the reaction over a range of assaytimeintervals, i.e., a time course, each at a series of substrate concentrations. The glucose 1,6-diP concentration was kept constant at 5 u\M in each assay. The rates of the linear regions of both phases of the product-time plots for each substrate concentration were plotted on double reciprocal plots in order to detect any nonlinearity. The actual values of K and Vmax of the substrates in each phase were calculated by m  computer. The K and Vmax of 3-fluoro-glucose 1-P were also determined in the presence of 5 \iM m  3-fluoro-glucose 1,6-diP. The inclusion of the appropriate diphosphate cofactor resulted in a monophasic linear plot and hence the K and Vmax values were determined exactly as for the m  "linear" substrates.  54 The double reciprocal plots obtained for each substrate and both phases of the data for 3deoxy and 3-fluoro-glucose 1-P are shown in the Appendix. The kinetic parameters of each substrate and the cofactor 3-fluoro-glucose 1,6-diP are summarized in Table H.5.  K (^M)a  Vmax (|imol/min/mg)  glucose 1-P  13±2  875 ± 64  2-fluoroglucose 1-P  222 ± 66  24 ± 5  16  128 ± 22  64±6  12  113 ± 9  7.6 ± 0.3  17  4-fluoroglucose 1-P  217 ± 4 4  31 ± 5  15  2-deoxyglucose 1-P  60 ±15  43 ± 8  153 ±40  68 ± 1 2  13  220 ± 2 0  9 ±0.6  19  4-deoxyglucose 1-P  153 ± 26  17 ± 2  16  3-fluoroglucose 1,6-diP  0.9 ± 0 . 1  4.5 ± 0.4  Substrate  m  3-fluoroglucose 1-P (first phase) 3-fluoroglucose 1-P (second phase)  a  -AAG * (kJ/mol) 0  b  0  3-deoxyglucose 1-P (first phase) 3-deoxyglucose 1-P (second phase)  a At pH 7.4 and 30°. Values calculatedfromRT ln[(Vmax/Km)l/(Vmax/Km)2], where T = 303 K, R = 8.314 J/K/mol, 1 refers to the native substrate glucose 1-phosphate and 2, the modified substrate. This value was determined with 3-fluoro-glucose 1,6-diP as the cofactor. A similar value was obtained when glucose 1,6-diP was used instead, this value being obtained from the second phase of the biphasic plots. b  c  Table II.5: Kinetic constants of fluorinated and deoxygenated substrate analogues of glucose 1-phosphate with phosphoglucomutase.  55  The apparent Km and Vmax of 3-fluoro-glucose 1,6-diP were determined at a nonsaturating 3-fluoro-glucose 1-P concentration of 503 (iM by the same method used to determine the kinetic parameters of the "linear" substrates. The K and Vmax of the cofactor (for a m  saturating substrate concentration) were calculated by the equations (Ray & Roscelli, 1964) K app (Glc 1,6-diP) = {K (Glc l,6-diP)[Glc l-P]}/{K (Glc 1-P) + [Glc 1-P]} m  m  m  V = Vmax(K (Glc l,6-diP)/[Glc 1,6-diP] + K (Glc-lP)/[Glc 1-P] + l)' m  1  m  The Lineweaver-Burk plot of the raw data is shown in the appendix but the recalculated K and m  Vmax values are listed in Table II.5. The value of Vmax obtained for 3-fluoro-glucose 1,6-diP should be the same as that obtained for 3-fluoro-glucose 1-P (4.5 ± 0.4 and 7.6 ± 0.3 |imol/min/mg, respectively). This difference cannot be accounted for. The value of K obtained for glucose 1-P, 13 ± 2 |iM, is slightly higher than the m  generally accepted value of 8 \iM (Ray & Roscelli, 1964; Lowry & Passonneau, 1969; Ray et al., 1973). A value of 5 |J.M has also been obtained (Ray et al., 1966). Studies carried out prior to the investigation of these deoxy and fluoro-substrate analogues also gave a value of 8 |J.M. Thus it seems that this value is somewhat variable by batch of enzyme. The results of the determination of the kinetic parameters show that the K values of each fluoro and deoxym  substrate are similar, ranging from approximately 100 to 200 \iM. The Km values obtained from both thefirstand second phases of the reaction of 3-fluoro-glucose 1-P were found to be the same within experimental error. This may be expected as 3-fluoro-glucose 1-P was the substrate in each phase, only the products differ. However since K is not a true dissociation constant but m  rather a function of equilibrium and rate constants, different K values for each phase may have m  been obtained. The Vmax values obtained for each phase differed by a factor of about nine, the same outcome being observed for both phases of the reaction of 3-deoxy-glucose 1-P. These  results show that the rate limiting step for complete turnover is in the second phase rather than the first as was suggested by the product-time plots. The consequences of this with respect to the mechanism are discussed later in this section.  56 The results in Table n.5 show that the values of Vmax are also similar amongst the fluoro and deoxy analogues, the Vmax values of the second phase of 3-deoxy and 3-fluoro-glucose 1-P being substantially less than the others (except 4-deoxy-glucose 1-P). There also does not appear to be a significant difference between the K and Vmax values of the fluoro and deoxym  substrates determined in this study and those of other subtrates determined by previous workers. Of the other substrates of phosphoglucomutase studied, mannose 1-P (the C-2 epimer of glucose) is the most efficiently utilized, having a K of 245 |J.M and V a x 0.06 times that of m  m  glucose 1-P. The V a x value, however, is not the rate at which mannose 1-P is converted to m  mannose 6-P, but rather the rate at which mannose 1-P releases glucose 6-P from glucose 1,6diP. The true rate of isomerization is claimed to be about 0.04timesthe above rate, thereby giving an overall rate of 0.0024 that of glucose 1-P. An accurate K value of galactose 1-P (the m  C-4 epimer of glucose) has not been determined but is probably in excess of 200 |iM (Lowry & Passonneau, 1969). However, at a concentration of 5 mM (probably saturating), a mutase rate 0.0025timesthat of glucose 1-P has been measured (Posternak & Rosselet, 1954). Thus the deoxy and fluoro analogues all bind slightly tighter than the C-2 and C-4 epimeric analogues of glucose 1-P and turn over at a somewhat greater rate. Comparisons between the kinetic constants of mannose 1-P, galactose 1-P and glucose 1P with the aim of making deductions on the nature of binding and catalysis are probably not warranted. This is due to the possibility of unfavorable steric interactions between the enzyme and the axial hydroxyl of mannose and galactose phosphates, plus the prevention of any potential hydrogen bonding between the equatorial C-2 or C-4 hydroxyls and the enzyme. Since the steric effect and the hydrogen bonding effect can neither be distinguished nor quantitated individually, no meaningful comparison can be made of this kinetic data with that of glucose 1-phosphate. However, in the case of a series of deoxy and fluoro-glucose 1-P substrates, the steric factor has been removed by replacement of the hydroxyl group with a smaller hydrogen or fluorine atom. The only difference between the deoxy and fluoro analogues and glucose 1-P will be a loss of bonding interactions to the enzyme at that position. Assuming that hydrogen bonding is the most important, there will be a loss of both acceptor and donor hydrogen bonding in the deoxy cases  57 and donor hydrogen bonding in thefluoroanalogues. Inductive effects caused by the substitution of hydrogen or fluorine for the hydroxyl group will probably be oftittleconsequence mechanistically. The mechanism of the phosphoryl transfer reaction of phosphoglucomutase involves cleavage of the O-P bond, rather than C-O, most probably via an "in-line associative" mechanism (see Section in.A.3). Thus the transition state of the reaction would not involve any substantial build up of charge on the anomeric carbon atom (see Figure D1.4). This contrasts sharply with glycosyl transfer reactions proceeding via oxocarbonium ion-like transition states in which the positive charge is destabilized by the replacement of hydroxyl with fluorine at C-2 and stabilized by substitution with hydrogen. The sensitivity of the stability of the oxocarbonium ion formed by the cleavage of the C-O bond in glucose 1-P to substitution of each sugar hydroxyl by fluorine or hydrogen has been investigated (Withers et al., 1986a, 1988; Percival & Withers, 1988). It was found that 2-fluoro-glucose 1-P is hydrolyzed in 1 M perchloric acid 60 times slower than glucose 1-P, whereas 2-deoxy-glucose 1-P is hydrolyzed 2700timesfaster. At low pH values, this hydrolysis reaction proceeds via C-O bond cleavage. In contrast, at pH values greater than 5, the reaction proceeds by the "metaphosphate" mechanism (O-P bond cleavage, Section HI.A.3) which does not involve charge build up on the anomeric carbon. It was found that the hydrolysis rate of 2-fluoro-glucose 1-P at neutral pH is the same as that of glucose 1-P (Withers et al., 1986a). This sensitivity to hydroxyl substitution on hydrolysis rates would be expected to be mimicked in enzymic systems which involve C-O bond cleavage as part of their mechanism, for example, glycogen phosphorylase. It has been found that the rate of reaction of 2-fluoro-glucose 1-phosphate is 3.1 x 10~6 times that of the native substrate, glucose 1-P (LP. Street, unpublished results). However, as an oxocarbonium ion-like intermediate is not part of the mechanism of phosphoglucomutase, a great reduction in the reaction rate of 2-fluoro-glucose 1-P or increase in that of 2-deoxy-glucose 1-P compared to that of glucose 1-P (due to inductive effects) was neuther expected nor observed. Substitution of fluorine or hydrogen for hydroxyl will probably have a minimal effect on the charge state of the substrate phosphate group. The pKa2 of 2-fluoro-glucose 1-P is 5.9 compared to 6.1 for glucose 1-P (Withers et ah, 1986a) and thus at pH 7.4 approximately 97% of the 2-fluoro-glucose 1-P will be present in the dianionic  58 form, compared to 95 % for glucose 1-P. The pKa2 for 2-deoxy-glucose 1-P has not been measured but is probably slightly higher than 6.1 on the basis of inductive effects, which will reduce the concentration of the dianionic phosphate species at pH 7.4 very slightly. Effects of substitution at more remote centers would probably not be detectable in terms of the phosphate group's p K . a  The sttiking feature of the kinetic constants measured in this study is the large apparent difference in the Km values of 3-fluoro-glucose 1,6-diP and 3-deoxy-glucose 1,6-diP compared to those of the other fluoro and deoxy-glucose diphosphates. The kinetic constants of the latter species were not measured but were inferred on the basis of the results of the product-time plots. The data also suggest that the rate cktermining step in the isomerization of 3-deoxy and 3-fluoroglucose 1-P occurs after the diphosphate intermediate has been produced. This is in contrast to the other fluoro and deoxy substrate analogues where it appears that production of the diphosphate is rate limiting, i.e., the rate at which the fluoro or deoxy-glucose 1-P is converted to the 6-phosphate is the same as the rate at which the monophosphate can release glucose 6-P from glucose 1,6-diP. In addition, the rate of steady state isomerization of the C-3 modified substrates is substantially less than the other deoxy and fluoro substrates, with the exception of 4-deoxy-glucose 1-P where the factor of difference is only 2. The apparent 90 fold difference in the K  m  values of the 3-fluoro and 3-deoxy-glucose 1,6-diP analogues compared with the other  diphosphate species and the differences in the rates of isomerization may be explained in terms of the importance of the C-3 hydroxyl in the interconversion of the two putative forms of the enzyme-glucose 1,6-diP complexes. The existence of two diphosphate complexes has been postulated by WJ. Ray (Ray et al., 1973) in what is known as the exchange mechanism (Figure 1.3). LA. Rose has suggested (1987) that the interconversion of these two enzyme-bound diphosphate species, one with the C-6 phosphate facing the serine hydroxyl and the other with the C-l phosphate in the same position, may occur via a rotation of the glucose diphosphate about its C-3 hydroxyl. Substitution of the C-3 hydroxyl by fluorine or hydrogen would remove some or all of the hydrogen bonding interactions between the enzyme and the modified diphosphate at that crucial position. The effect of this would be to increase the dissociation  59  constant of the glucose diphosphate and reduce the rate at which the interconversion of the two bound species would occur. The rate of conversion of the 3-fluoro or 3-deoxy-glucose 1-P into the corresponding diphosphate would not be affected to any greater degree than the other fluoro or deoxy analogues, since the interconversion of the two bound diphosphate species is not involved in this part of the reaction mechanism. The validity of this argument does not necessarily hang on the fact that the K values of 3-fluoro and 3-deoxy-glucose 1-P are not m  higher than those of the other analogues, as it is possible that enzyme-substrate interactions in the phosphoenzyme-monophosphate complex could be somewhat different from those of the dephosphoenzyme-diphosphate complex. The apparent loss of binding free energy at the transition state of each substrate caused by the replacement of aringhydroxyl by either a fluorine or hydrogen atom is also listed in Table n.5. These data show that the apparent loss of binding free energy for each substrate .analogue amounts to approximately 12-17 kJ/mol (3-4 kcal/mol). There does not appear to be any graded difference in the values amongst the analogues, within experimental error, to indicate the importance of any single ring hydroxyl in binding and catalysis. The exceptions are the values of 3-deoxy and 3-fluoro analogues, the kinetics of the second phase indicating a larger loss of binding energy than the other analogues. The values of AAG'-i- of each analogue (12-17 kJ/mol) are similar to those obtained by Fersht et al. (1985) for the deletion of a single hydrogen bond in which one of the two groups is charged or several hydrogen bonds with neutral species. The results also correspond well with those of Street et al. (1986). Whether the hydrogen bonding of each side chain does in fact involve charged amino acid side chains or several neutral hydrogen bonds at each hydroxyl, as suggested by the magnitude of the loss of binding free energy, would require information from the as yet unsolved X-ray crystal structure of the enzyme-substrate complex. n.B.3.4: Determination of Inhibitor Kinetic Parameters. In order to assess the hydrogen bonding interactions of the substrate hydroxyl groups without the added complication of the analogue's substrate activity, a series of deoxy and fluoro-  60 glucose 1-P analogues was synthesized, each having the C-6 (acceptor) hydroxyl replaced by fluorine. The rationale for their synthesis was based on the observation that the Ki of 6-fluoroglucose 1-P is the same as that of the K of the parent substrate, glucose 1-P. Ray et al. (1976) m  showed that the dissociation constant (Kd) of glucose 1-P is the same as its K . Hence it would m  appear that no binding energy is lost on the replacement of the C-6 hydroxyl by afluorineto produce an inhibitor. In light of the additive nature of binding energies (Jencks, 1975,1981), an inhibitor produced by the replacement of hydroxyls on both the sugar ring and C-6 by fluorine should bind to the enzyme with a loss of binding energy equivalent to that for the substrate monofluoro-glucose 1-P. Thus the inhibition constants, which are true enzyme-ligand dissociation constants, were measured for a series of inhibitors based on 6-fluoro-glucose 1-P, in which each sugar ring hydroxyl was replaced by either hydrogen orfluorineto produce a disubstituted glucose 1-P analogue. The synthesis of the analogue in which the C-2 hydroxyl is replaced by a hydrogen was not attempted due to its probable lability. As well as the disubstituted inhibitors, the inhibition constants were determined for the monosubstituted glucose phosphate analogues in which the C-l or C-6 hydroxyls were replaced by bothfluorineand hydrogen. The inhibition constants of each analogue were determined by measurement of the apparent K of glucose 1-P in the presence of a range of concentrations of the inhibitor using the m  coupled assay system. The results were fitted to the Michaelis-Menten equation by the same computer program described in Section n.B.3.3 and were plotted on double reciprocal plots in order to identify the type of inhibition. All the inhibitors produced linear double reciprocal plots, the lines of which intersected at the same position on the y-axis, thereby demonstrating the competitive nature of the inhibition. Plots of apparent K versus inhibitor concentration were m  constructed to allow the determination of Ki. The points were fitted by a least squares regression analysis and extrapolated to the x-axis (-Ki). The double reciprocal and replots are presented in the Appendix. The inhibition constants and values of AAG° for each inhibitor are shown in Table n.6. The errors in these values estimated from the replots are approximately ±10% in each case.  61 Inhibitor  Ki (|xM)  AAG° (kJ/mol)  6-fluoro-glucose 1-P  lfjc  0  6-deoxy-glucose 1-P  73  3  2,6-difluoro-glucose 1-P  2430  12  3,6-difluoro-glucose 1-P  1710  11  3-deoxy-6-fluoro-glucose 1-P  1350  11  4,6-difluoro-glucose 1-P  720  9  4-deoxy-6-fluoro-glucose 1-P  740  9  a-glucosyl fluoride 6-P  41  2  1-deoxy-glucose 6-P  32  1  a  b  AtpH7.4 and 30°. Values calculated from RT ln(Ki/K2), where T = 303 K, R = 8.314 J/K/mol and Kl is the Ki of the inhibitor and K2 is the K of glucose 1-P, 20 (iM. Compared to a K of glucose 1-P of 10 u\M. a  b  m  c  m  Table n.6: Dissociation constants of glucose phosphate inhibitors of phosphoglucomutase.  During the initial investigation of the inhibition of phosphoglucomutase by 6-fluoroglucose 1-P, the K of glucose 1-P was measured by the coupled assay with m  phosphoglucomutase supplied by Sigma Chemical Co. and found to be 8-10 [iM, which is in good agreement with the established value of 8 [iM. The value of Ki for 6-fluoro-glucose 1-P was also determined as 10 |iM. However, several months later, when the complete series of inhibitors had been synthesized, the K of glucose 1-P was measured as 20 [iM. Several m  measurements were made, each producing a value of 20 ± 2 uM. The enzyme used in these experiments was prepared at Purdue University and had a specific activity approximately 10 fold greater than the commercially prepared enzyme. The K of glucose 1-P measured by the m  colorimetric assay was 13 ± 2 |iM which is in fair agreement with the established value. A range finding measurement of the Ki of 6-fluoro-glucose 1-P by the coupled assay also gave a value of 20 |J.M, i.e., the same as the value of K obtained for glucose 1-P. No explanation of this m  62 variation in the value of Km for glucose 1-P can be offered, but the matter does not effect this study since the analysis of the data involves comparison of relative rather than absolute values and all values quoted were (ktermined with the same batch of enzyme (except where noted). The results show that there is no significant binding energy lost on the replacement of either the C-l or C-6 hydroxyl by fluorine. This indicates that the acceptor hydroxyls (C-l and C-6) are not involved in any donor type hydrogen bonding with the enzyme. The Km of aglucose 6-P has been estimated at approximately 20 |iM or 2.5 times the value of Km of glucose 1-P (which is 8 )iM). The K  m  value of a-glucose 6-P was determined from a knowledge of the  equilibrium ratio between the a and B forms, the measured Km (47 \iM), and the assumption that the B-anomer does not bind tightly to the enzyme (Lowry & Passonneau, 1969). Therefore in the assay system used in this study .where K study, the K  m  m  values are higher than in the aforementioned  of a-glucose 6-P would be approximately 50 pM. Consequently cc-glucosyl  fluoride 6-P and 1-deoxy-glucose 6-P actually bind slightly tighter than their parent substrate. Observations of this kind have been previously made with other fluorinated substrate analogues in different systems (e.g., Bessel et al., 1972) and are believed to be due to the increased hydrogen bond donor potential of the hydroxyl group vicinal to a fluorine substitution. However this rationale cannot be used with the deoxy-inhibitor and the cause of this effect is unknown, but is, in any case, very small. 6-Deoxy-glucose 1-P has a slightly higher Ki value than its fluorinated analogue, indicating that the C-6 hydroxyl may be involved in a weak acceptor hydrogen bond with the enzyme. The value of AAG° obtained for 6-deoxy-glucose 1-P (see Table H..6) is in good agreement with the value obtained for xylose 1-P by Ray and Long (1976b). (Xylose is equivalent to glucose with the C-6 hydroxymethyl group removed). This shows that the attractive force between the C-6 hydroxymethyl group of glucose 1-P and the enzyme is probably due to hydrogen bonding alone and not a combination of hydrogen bonding and Van der Waals forces. The absence of a strong interaction of these acceptor hydroxyls with the enzyme may be predicted in light of the theory that the enzyme-substrate interactions are maximized in the transition state and (relatively) minimized in the enzyme-substrate ground state complex (see  63 introduction to this chapter). A strong ground state interaction between the acceptor hydroxyl group of the glucose monophosphate and the enzyme would need to be broken in order to approach the transition state in which the acceptor hydroxyl attacks the enzymic phosphate, thereby increasing the activation energy and reducing the reaction rate. However, it may be expected that the "acceptor" hydroxyls (C-l and C-6) are involved in a hydrogen bonding interaction with an enzymic general base catalyst which becomes protonated during the phosphoryl transfer reaction. Thus some interaction might have been expected. It is quite possible however that this interaction does not become important until the transition state of the phosphoryl transfer reaction is achieved and information on this is unavailable. In contrast to thetightbinding of the monosubstituted inhibitors, the disubstituted inhibitors all bind relatively weakly to the phosphoenzyme. Thetightestbinding inhibitors were found to be 4,6-difluoro and 4-deoxy-6-fluoro glucose 1-P, both of which have Ki values of approximately 730 |iM. The fact that both Kj values are the same suggests that the principal hydrogen bonding interaction is of the type in which neither a C-F nor C-H fragment can participate and therefore probably involve (a) donor hydrogen bond(s). The pair of inhibitors 3,6-difluoro and 3-deoxy-6-fluoro-glucose 1-phosphate have Ki values of 1700 and 1350 u\M, respectively. These two values are the same within experimental error and therefore the hydrogen bond interaction is probably also of the donor type. The weakest binding inhibitor with a Ki of 2400 \iM is 2,6-difluoro-glucose 1-P. These kinetic data indicate that the importance of each glucose ring hydroxyl group of the substrate glucose 1-P in providing binding interactions in the ground state enzyme-substrate complex decreases in the order C-2 > C-3 > C-4. For example, replacement of the C-2 hydroxyl by a fluorine produces a weaker inhibitor than the similar replacement of the C-4 hydroxyl. These results are consistent with those of Ma and Ray (1980) who measured the UV spectra of complexes of phosphoglucomutase and various sugar phosphates. They found that the differences between the spectra produced on the binding of the sugar 6-phosphates were less sensitive to changes in the hydroxyl groups at C-l, C-2 and C-3 than C-4. Similarly, spectral changes between the 1-phosphates were less sensitive to changes at C-4 and C-6 than C-2 and  64 C-3. The authors interpreted these findings in terms of the greater importance of the hydroxyl groups distal to the acceptor group for the production of a normal complex. The sugar phosphates used in this study were the C-2 and C-4 epimers (mannose and galactose) of glucose phosphates and ribose 1- and 5-P. The ribose phosphates were considered as an approximation to hexoses in which C-3 and its hydroxyl are missing. Due to the difficulty in ascribing observed spectral differences to the removal of the particular hydroxyl group and not to steric effects caused by the presence of an axial hydroxyl, the conclusions made were fairly tentative. However they are in good agreement with the present, more rigorous study. The results of the kinetic studies on each inhibitor can be used to predict the type and strength of hydrogen bonding to each sugar hydroxyl in the enzyme-inhibitor complex (Figure 11.16).  P? (3)  Figure n.16: Proposed ground state enzyme-substrate hydrogen bonding scheme for phosphoglucomutase. The values in brackets represent the strength of hydrogen bonds (in kJ/mol) at that position.  This analysis cannot predict the presence of multiple hydrogen bonding since it cannot detect a weak hydrogen bond in the presence of a strong one. Inspection of the AAG° values obtained for each of the disubstituted inhibitors (Table n.6) shows a range of 9-12 kJ/mol (2-3 kcal/mol) loss of binding energy compared to the parent inhibitor 6-fluoro-glucose 1-P. Comparison with the AAG°i values of the fluoro and deoxy glucose 1-phosphate substrates (15-19 kJ/mol, 3-4 kcal/mol, Table II.5) indicates a difference in the apparent substituent binding energy values of approximately 4-8 kJ/mol (1-2 kcal/mol) for  65  each substrate/inhibitor pair. The AAG" values for the disubstituted. inhibitors are a measure of the differences in AG" values between the two ground state enzyme-ligand complexes. However, AAG *' for the fluoro and deoxy-substrates gives the difference between the two activation 0  energies. This is the case since the difference in Gibb's free energy between the free enzyme and free substrate, and the enzyme-substrate transition state complex is proportional to the second order rate constant, kc t/K (Fersht, 1985).This concept is shown ^grammatically in Figure a  m  n.17. These schemes are greatly simplified, showing only the attainment of the transition state and not showing the energy barrier involved in the binding step. In addition it is assumed that the ground state energies of the substrate and the fluorinated analogue are the same and that the energies of the various transition states of the reaction are equivalent. Glucose 1-P  Fluoro or deoxy substrate E.S* A G  .0*  G  6-Fluoro-glucose 1-P  G  Difluorinated inhibitor  E+I  G  |  Figure n.17:  E+I  >o  A A G  Gibb's energy changes associated with catalysis and binding of substrates  and inhibitors.  66  Thus this difference in apparent substituent binding energy values of 4-8 kJ/mol (1-2 kcal/mol) represents an estimate of the increase in enzyme-substrate binding energy found in the transition state complex compared to the ground state complex. It would appear that each sugar ring hydroxyl contributes approximately 4-8 kJ/mol more bmding energy in the transition state than in the ground state complex. The transition state of the phosphoryl transfer reaction in phosphoglucomutase most probably involves the formation of a pentacoordinate phosphate species (see Section HI. A3) with little involvement of the remainder of the sugar molecule. No chemical or structural changes occur to the sugar portion of the molecule and thus a large change in the interactions between the sugar ring and the enzyme during the approach to the transition state would not be expected, nor is it observed. The importance of this pentacoordinate phosphorus transition state has been implied by the strong competitive inhibition afforded by the C-6 vanadate esters of glucose and glucose 1-P (Doherty, 1987). Vanadate can easily take up a pentacoordinate trigonal bipyramidal geometry and it was proposed that the vanadate esters mimic the transition state of the substrate, thus binding very tighdy.  II.B.3.5: Spectral Studies on Enzyme-Ligand Complexes. The poor binding of the disubstituted inhibitors to the phosphoenzyme may suggest that the enzyme is not recognizing the sugar moiety at all and that essentially all the interaction occurs at the phosphate binding site (shown schematically in Figure n.18). This seems plausible in view of the fact that inorganic phosphate and methyl phosphonate themselves bind to the enzyme with a dissociation constant approximately twice that of the poorest inhibitor 2,6-difluoro-glucose 1-P (Ray et al., 1973). The values from this reference were adjusted to take into account the differences in the K values of glucose 1-P determined in each study. A similar phenomenon m  was observed with the enzyme phosphorylase b and its substrate cc-glucose 1-P, which has a Km of 2 mM. The authors found that the competitive inhibitors a-mannose 1-P and phenyl phosphate both possessed similar dissociation constants (35 and 19 mM, respectively). The similarity of their dissociation constants and their diversity of structure was interpreted in terms of the recognition of the phosphate moiety only. Thus it was postulated that all the enzyme-  67  inhibitor interaction occurred through the phosphate subsite rather than both phosphate and sugar subsites (Sprang et al., 1982).  Complete Recognition  Partial Recognition  Figure 11.18: Schematic representation showing complete and partial substrate recognition by enzyme binding sites.  In order to test this possibility, structural comparisons of the enzyme complexes of the inhibitors 3,6-difluoro-glucose 1-P and methyl phosphonate were made by means of UV spectroscopy. Methyl phosphonate is believed to bind at the same phosphate binding site (Kd = 1.5 mM) as that of substrate monophosphates (Ray et al., 1973). The UV spectra of the phosphoenzyme and the binary phosphoenzyme-inhibitor complexes were measured between 250 and 330 nm. The difference spectrum produced by subtraction of the spectrum of the phosphoenzyme from that of the phosphoenzyme-3,6-difluoro-glucose 1-P complex (Figure n.19) was close in shape and detail to that produced by glucose 1-P (Ma & Ray, 1980), although the difference in molar absorptivity obtained for the disubstituted inhibitor was half of that measured for glucose 1-P. Ma and Ray showed that the difference spectra produced by binding of 6-deoxy-glucose 1-P (or glucose 1-P) to the phosphoenzyme in the absence of metal and that produced by the binding of glucose 1-P in the presence of L i were identical. They concluded +  that these spectra would be similar to that produced by the binding of glucose 1-P to the Mg +2  phosphoenzyme complex (an experiment which cannot be conducted due to turnover).  68  1000  250  270  290 310 Wavelength, nm  330  Figure 11.19: Inhibitor induced difference spectrum produced by the binding of 3,6difluoro-glucose 1-P to demetallated phosphoenzyme. The concentration of enzyme was 38.1 |iM and that of inhibitor, 20 mM.  The same observations and conclusions were made for the binding of 1-deoxy-glucose 6-P and glucose 6-P although the difference spectra differed in detail from that produced by glucose 1-P. The similarities in the difference spectra produced by 3,6-difluoro-glucose 1-P and glucose 1-P indicate that the disubstituted inhibitor is binding in the same mode as that of glucose 1-P. This conclusion is supported by the results of the difference spectrum produced by the binding of methyl phosphonate to the phosphoenzyme (Figure 11.20). The details of this difference spectrum differ from that produced by 3,6-difluoro-glucose 1-P although the major peak at approximately 298 nm is present in both spectra. The difference spectrum produced by methyl phosphonate has a greater similarity to that produced by glucose 6-P, the peak at 288 being almost as large as that of the major broad peak at 298 nm However, the large trough at 270 nm present in the spectra of both substrates and disubstituted substrate analogue is missing. Thus it appears that there is some recognition of the sugar moiety in the disubstituted glucose 1-P analogues and not just of the phosphate moiety. A more detailed study of the spectral changes on binding of non-glucose phosphate esters would need to be carried out in order to  69 establish whether the similar spectral changes produced by the substrate phosphate esters (at 287 and 298 nm) are in fact due to the binding of the phosphate moiety.  1000  -1000 « — ' — i — — i — — i — 250 270 290 310 Wavelength, nm 1  1  1  — I 330  Figure n.20: Inhibitor induced difference spectrum produced by the binding of methyl phosphonate to demetallated phosphoenzyme. The concentration of enzyme was 24.6 |iM and that of inhibitor, 30 mM.  In accord with the earlier assumption that the binding mode of 3,6-difluoro-glucose 1-P is equivalent to that of 3-fluoro-glucose 1-P, it was found that the difference spectrum produced on binding of the latter species to demetallated phosphoenzyme was identical to that of the former.  II.B.3.6: Evaluation of Equihbrium Constants of Fluoro Substrates. The equihbrium constant for the isomerization of each fluoro-glucose 1-P substrate was determined by the use of l^F-nmr. Due to the high chemical shift dispersion of l^F-nmr, the three signals arising from the a- and B-anomers of the 6-phosphate and the a-anomer of the 1phosphate were sufficiently resolved to allow their relative intensities to be evaluated by peak height The proton decoupled l^F-nmr spectrum of a solution containing an equiUbrium mixture of the two isomers of each fluoro-glucose phosphate and a catalytic amount of phosphoglucomutase was measured at 30° C and pH 7.4. A small amount of glucose 1,6-diP  70 was included in the assay mixture in order to satisfy the enzyme's cofactor requirement. The amount added was approximately 1% of the amount of monophosphate present so that the fluoro-glucose 1,6-diP species produced would not be evident in the spectrum The relative concentrations of each species were determined by measurement of the peak heights. The spectrum was re-measured approximately one hour later to ensure that equilibrium had in fact been attained. A third spectrum was measured in which the delay time between acquisitions was tripled in order to determine whether differences in Tl values of the species would result in differences in the signal intensities. No change was observed. The values of the equilibrium constants for each fluoro-glucose 1-phosphate analogue are shown in Table U.7. ,  Glucose P derivative  Equihbrium ratio (Glc 6-P/Glc 1-P)  _  a  17a  2-fluoro  23  3-fluoro  14  4-fluoro  9  From Atkinson et al. (1961).  Table n.7: Equilibrium ratios of fluorinated glucose phosphates at30"C and pH 7.4. Measurement of the equilibrium constant by the use of l^F-nmr has the advantage over an apparently simpler colorimetric method in that it avoids complications brought about by the presence of low concentrations of inorganic phosphate and the fJ-anomer of the fluoro-glucose 1P. Although the concentrations of these contaminants are only in the order of several percent, they become substantial non-reacting, acid-labile, impurities when the equilibrium mixture contains only a small amount of the a-glucose 1-P.  n.B.4: Summary and Implications fey the Mechanism. The experimental results of this investigation indicate that the rate ctetermining step in the isomerization of 3-fluoro- and 3-deoxy-glucose 1-P is not the same as that of the other fluoro and  71 deoxy substrate analogues. Evidence that the rate determining step for the isomerization of the C3 modified substrates occurs after the production of the corresponding diphosphate species comes from the observation that the Vmax of the reaction involving production of the diphosphate ("first phase") is eight times greater than the overall mutase rate ("second phase"). The fact that the rate (ktermining step for the isomerization of the other C-2 and C-4 modified substrates occurs during the production of the corresponding diphosphate species is indicated by the absence of a change in the rate on the depletion of glucose 1,6-diP, when the course of the reaction must change. It is also observed that the Vmax values of the overall isomerization of the C-3 modified analogues is substantially less than that of the other fluoro and deoxy substrates, whereas the V ax for the production of the C-3 modified diphosphate species are similar to the m  overall mutase rate of the other analogues. These observations have been rationalized in terms of a decreased affinity of the C-3 modified glucose diphosphates for the dephosphoenzyme. This decrease in affinity and reduction in the mutase rate is due to the removal of potential hydrogen bonding interactions between the C-3 hydroxyl and the enzyme. However it could also be due to a slower phosphoryl transfer from 3-fluoro-glucose 1,6-diP to the phosphoenzyme. This question is being addressed independently (LA.Rose). The fact that the C-3 hydroxyl plays a more important role than the other ring hydroxyls is evidence for the suggestion that the reorientation of the glucose diphosphate in the putative "exchange mechanism" occurs via a rotation around the C-3 hydroxyl. If this were indeed the case then the removal of this hydroxyl would decrease the rate at which this reorientation could occur. However, the initial formation of this diphosphate species would not be affected to any greater degree than the other deoxy or fluoro substrates. The rate at which the C-2 and C-4 modified glucose 1,6-diPdephosphoenzyme complexes interconvert will not necessarily be affected since the hydroxyl around which the rotation occurs is still present. The result of this reduction in rate of reorientation of the 3-fluoro or 3-deoxy-glucose 1,6-diP is the change in rate limiting step from the production of glucose 1,6-diP ("first phase") to the reformation of the alternate monophosphate species ("second phase").  72 However, these results could also be considered consistent with a minimal motion type of mechanism. The protein-substrate interactions of the glucose 1,6-diP-dephosphoenzyme complex could be such that the C-3 hydroxyl plays a more important role than the other sugar ring hydroxyls. Hence loss of the C-3 hydroxyl would make these diphosphate species bind poorly relative to the other diphosphates, or in an incorrect mode thereby causing a reduction in the rate of phosphoryl transfer back to the enzyme. The C-3 hydroxyl could also be important in any protein conformation change necessary for the phosphoryl transfer step. The results of the investigation of the disubstituted inhibitors show that there is a graded effect on the binding of glucose 1-P produced by the deletion of each sugar hydroxyl. The conclusion can be drawn that the hydroxyl groups distal to the acceptor hydroxyl (C-6) are more important in creating a normal complex than that in its proximity. Thus in glucose 1-P, the most important hydroxyl appears to be C-2 which is farthest (distal) from the acceptor hydroxyl (C-6). These results are in accord with the tentative conclusions of Ma and Ray (1980). These authors also found the same effect with glucose 6-P, the C-4 hydroxyl which is furthermost from the acceptor hydroxyl (C-l) is more important than that of C-3 and C-2. These findings support the suggestion that both substrates have the same binding site in the protein. However, the interactions are reversed, the C-2 hydroxyl of glucose 1-P interacting with the same protein site as the C-4 hydroxyl of glucose 6-P. Similarly, the C-4 hydroxyl of glucose 1-P interacts with the same site as the C-2 hydroxyl as glucose 6-P. It may be predicted that the C-3 hydroxyls of both substrates interact with the same protein site. This concept is shown schematically in Figure 11.21. If the enzyme were to utilize a minimal motion mechanism, it would be expected that the hydroxyls of both substrates would interact with the same protein amino acids. The observed effect would be that the deletion of the same glucose ring hydroxyl in glucose 1-P and glucose 6P would produce a similar weakening in the binding of both substrates. Clearly the present results and those of Ma and Ray are not in accord with this type of mechanism.  73  Figure 11.21: Schematic diagram of the proposed exchange type of mechanism for phosphoglucomutase. The symbols S (strong), M (medium) and W (weak) indicate the strength of the interactions between the protein and the substrate hydroxyl groups at that position.  74  UC: EXPERIMENTAL PROCEDURES.  n.C.l: General Synthetic Methods and Purification of Reagents. Nuclear magnetic resonance (nmr) spectroscopy was carried out in the Fourier transform mode on the following instruments. Many of the spectra were measured by members of the nmr staff, (U.B.C); lH: 270 MHz spectra on a Bruker HXS-270 instrument interfaced with a Nicolet 1180 computer, 300 MHz spectra on a Varian XL-300 and 400 MHz spectra on a Bruker WH-400. l^F: 254 MHz spectra on the Bruker HXS-270 described above but equipped with a 5 mm 19p high resolution probe. 31p: 121 MHz spectra on a Varian XL-300 equipped with a 5 mm tuneable probe. The results of all spectra are given in parts per million from the following reference compounds; lH: Signal positions are given on the delta (8) scale, signals occurring at a lower field than the reference having a positive chemical shift. Samples dissolved in deuterochloroform (CDCI3) and deuteromethanol (CD3OD) are internally referenced to tetramethylsilane (8 = 0.00). Samples dissolved in deuterium oxide (D2O) are referenced to external 2,2-dimethyl-2silapentane-5-sulfonate (8 = 0.015). 1 F: Signal positions are given on the delta (8) scale, signals occurring at higher field 9  than the reference CCI3F (8 = 0.00) are given a positive chemical shift Spectra were measured referenced to external trifluoroacetic acid (8 = 76.53) and then converted to the delta scale. 31p: Signal positions are given on the delta (8) scale, signals occurring at higher field than the external reference, 85% H 3 P O 4 in D 2 O , are given a positive chemical shift. Due to the sensitivity of phosphate chemical shifts to pH and the line broadening effect of paramagnetic metal ions, 31p-nmr spectra were measured in a pH 6.8 buffer containing 100 mM triethanolamine, 1 mM EDTA and 50% D 2 O unless otherwise noted. No adjustment of the pH meter reading to account for the presence of D 2 O was made.  75  Fast atom bombardment (FAB) mass spectrometry was performed by the Mass Spectrometry Laboratory staff at the University of British Columbia on a AEIMS-9 instrument operating with a 8 kV Xe atom beam source. Microanalyses were performed by Mr. P. Borda, Microanalytical laboratory, University of British Columbia. Melting points were determined on a Fisher-Johns apparatus and are corrected using a calibration curve supplied with the instrument. Thin layer chromatography was carried out on aluminum backed plates of Kieselgel 60 F254 (Merck). The developed plates were visualized by either fluorescence quenching under a UV light source or spraying with 10% sulfuric acid in methanol followed by strong heating. Generally, deprotected or partially protected sugars were eluted with a solvent mixture of ethyl acetate: ethanol: water (7:2:1) and fully protected sugars were eluted with a mixture of n-pentane: ethyl acetate: ethanol (20:9:1). Flash chromatography was carried out on silica gel (Kieselgel 60,180-230 mesh, Merck) according to Still et al. (1978), the solvent system employed is given in parentheses. Solutions were evaporated on a Buchi rotary evaporator with a bath temperature below 40°. Crystallizations were generally performed (unless stated otherwise) by dissolving the impure material in a minimum volume of the more polar solvent and careful addition of the less polar solvent until a lasting opalescence was observed. Many crystallizations were carried out at 4°. Sodium methoxide solution was prepared by adding preweighed, hexane washed, sodium metal to a volume of anhydrous methanol. DE-52 cellulose (Whatman) for anion exchange chromatography was pretreated by suspending the exchanger in 0.5 M ammonium bicarbonate solution and bubbling carbon dioxide through the mixture until the pH reached 8.0. The suspension was allowed to settle and the supernatant removed in order to reduce fines. After the column was poured, it was equihbrated by elution with the noted concentration of ammonium bicarbonate. Equihbration was considered complete when both the pH and conductivity of the effluent were identical to that of the eluent. Solvents and reagents were purified as follows. Pyridine, diisopropylethylamine, cyclohexylamine and 2,4,6-trimethyl pyridine were pre-dried over potassium hydroxide pellets for several days and distilled from barium oxide under nitrogen. The solvents were stored over  76 solid potassium hydroxide. Dichloromethane was washed twice with concentrated sulfuric acid, followed by water (twice) and saturated sodium bicarbonate solution. The solvent was pre-dried over sodium sulfate and distilled under nitrogen from calcium hydride. The solvent was stored over 4 A molecular sieves. Methanol was treated with magnesium turnings and iodine to produce magnesium methoxide from which it was distilled under nitrogen. The solvent was stored over 3 A molecular sieves under nitrogen. Toluene (1000 ml) was washed twice with ice-cold concentrated sulfuric acid (100 ml), followed by water (twice) and saturated sodium bicarbonate solution. The solvent was pre-dried over magnesium sulfate and distilled from phosphorous pentoxide under nitrogen. aa'-Azobisisobutyronitrile was crystallized from chloroform and stored in a desiccator. Triphenylmethyl chloride was crystallized from benzene containing several drops of acetyl chloride by addition of two volumes of petroleum ether and stored at 4°. Crystalline phosphoric acid (BDH) was dried in vacuo over magnesium perchlorate for several days prior to use. Other solvents and reagents were generally Certified or Analar grade and used as supplied. Diethylaminosulfur trifluoride (DAST), chloromethyl methyl ether, a,a-dichloromethyl methyl ether and tributyltin hydride were obtained from Aldrich Chemicals and used without further purification. Methyl a-D-galactopyranoside, uridine 5'-triphosphate, uridine 5'diphosphoglucose pyrophosphorylase, inorganic pyrophosphatase, glucose 1,6-diP, 2-deoxy-Dglucose 6-P and bovine serum albumin were obtained from Sigma Chemical Co. and were used without further purification. Trifluoromethyl 3,4,6-tri-0-acetyl-2-deoxy-2-fluoro-a-Dglucopyranoside was a gift from Dr. D. Dolphin (U.B.C.). 3-Deoxy-3-fluoro-l,2:5,6-di-0isopropyhdene-a-D-glucofuranose, 3-deoxy-3-chloro-l,2:5,6-di-0-isopropyUdene-a-Dglucofuranose, methyl 2,3,6-tri-O-benzoyl-a-D-galactopyranoside and methyl 4-deoxy-oc-Dxylohexopyranoside were gifts from Mr. Ian Street (U.B.C.). The preparation of cc-glucosyl fluoride 6-P and 1-deoxy-glucose 6-P are both described in the experimental procedures section of Chapter in.  77 n.C.2: Synthetic Methods. Trifluoromethyl 2-deoxy-2-fluoro-a-D-glucopyranoside (2). Trifiuoromethyl 3,4,6-tri-0-acetyl-2-deoxy-2-fluoro-a-D-glucopyranoside (1,4.0 g,10.6 rnmol) was deacetylated by treatment with dry methanol (20 mL) containing 50 mM sodium methoxide. After 15 minutes at room temperature the base was neutralized by addition of Dowex 50W-X8 (H ) resin. Filtration and concentration yielded a yellowish gum 2 (2.5 g, 10.0 +  mmol, 94%); lH-nmr data (300 MHz, CD30D):8 5.84 (d,l H, J  1 > 2  4.0 Hz, H-l), 4.38 (ddd,l  H, J2,F 50.0, J2,3 9.0, J2,l 4.0 Hz, H-2), 3.60-3.90 (m,4 H, H-3,5,6,6'), 3.46 (dd,l H, J  4 > 3  10.0, J4,5 10.0 Hz, H-4).  Trifluoromethyl 2,6-dideoxy-2,6-difluoro-a-D-glucopyranoside (3). Dry, emanol-free dichloromethane (20mL) was added to 2 (1.20 g, 4.80 mmol) and the suspension cooled to -40°. DAST (3.73 mL, 28.8 mmol) was added with stirring and the reaction allowed to warm to room temperature under anhydrous conditions. After 1 hour the o  solution was cooled to -40 and quenched by addition of methanol (10 mL). The solvent was removed and the residue purified by flash chromatography (1:1 hexane, ethyl acetate) yielding a colorless oil, 2 (0.67 g, 2.66 mmol, 56%); lH-nmr data (400 MHz,  CDCI3): 8 5.76 (d,l H,  Jl,2 3.8 Hz, H-l),4.71 (ddd,l H, J6JJ 47.0, J6.6' 10.5, J6,5 3.0 Hz, H-6), 4.61 (ddd,l H, J6',F 48.0, J6',6 10.5, J6',5 2.0 Hz, H-6'), 4.50 (ddd.l H, J2,F 48.0, J2,3 9.5, J2,l 3.8 Hz, H-2), 4.10 (ddd,l H, J3JJ 12.0, J3 9.2,13,4 9.2 Hz, H-3), 3.94 (dddd,l H, J >2  5  F  27.0,15,4  10.0, J ,6 3.0, J ,6' 2.0 Hz, H-5), 3.75 (dd,l H, J 4 3 9.5, J ,5 9.5 Hz, H-4). 5  5  4  2,6-Dideoxy-2,6-difluoro-D-glucopyranose (4). Washed Dowex 50W-X8 (H ) resin was added to a solution of 3 (0.67 g, 2.66 mmol) in +  water (30 mL) and the mixture refluxed for 60 minutes. After filtration and evaporation the resulting oil was filtered through silica gel (ethyl acetate) which after evaporation yielded a clear gum 3 (0.37 g, 2.0 mmol, 76%); 19F-nmr data (254 MHz, D2O): 5 200.47 (ddd, JF,2 51.3,  78 JF.3 15.0, JF,1 2.4 Hz, F-2), 200.59 (dd, J F , 2 49.3, J F , 3 13.4 Hz, F-2), 235.94 (dt, J F , 5  25.7, JF.6+6' 47.5 Hz, F-6), 236.50 (dt, J F , 5 28.6, JF,6+6'47.5 Hz, F-6).  1,3,4-Tri-0-acetyl-2,6-(iideoxy-2,6-cu^uoro-f5-D-glucopyranose (5). A solution of 4 (0.37 g, 2.0 mmol) in 3 mL of dry pyridine was acetylated by addition of 2 mL of acetic anhydride. After 20 hours at room temperature, methanol (5 mL) was added to remove excess anhydride and one hour later the solvents were evaporated. The residue was further dried in vacuo . The resulting gum was converted directly to the per-acetylated a-bromide by dissolution in 45% hydrogen bromide in glacial acetic acid (5 mL) containing acetic anhydride (10% v/v). After 8 hours at room temperature, dichloromethane was added and the mixture was washed thrice with ice water followed by saturated sodium bicarbonate solution. The organic phase was dried (magnesium sulfate) prior to evaporation and further drying in vacuo . The ccbromide was converted to the (3-per-acetate by reaction with mercuric acetate (0.63 g, 4.0 mmol) in glacial acetic acid (7 mL). After stirring for one hour in the dark, dichloromethane was added and the mixture was twice extracted with water, followed by saturated sodium bicarbonate solution. After drying (magnesium sulfate), the product 5 was crystallized from ethyl acetate, pentane (0.44 g,1.41 mmol, 71%); m.p. 135-135.5°; lH-nmr data (400 MHz, CDCI3): 8 5.75 (dd,l H, Ji,2 8.1, Ji,F 3.3 Hz, H-l), 5.35 (ddd.l H, J3JF 14.3, J3,4 9.1, J3,2 9.1 Hz, H-3), 5.06 (dd,l H, J 4 3 9.6, J4,5 9.6 Hz, H-4), 4.50 (ddd,l H, J6,F 47.0, J6,6' 10.0, J6,5 2.3 Hz, H-6), 4.45 (ddd,l H, J6',F 47.0, J6',6 10.0, J - 1 0 Hz, H-6"), 4.43 (ddd,l H, J2.F 50.0, 6  J 2 , l 9.0, J2,3 9.0 Hz, H-2), 4.07 (dddd,l H, J  5  5 j F  23.0,15,4 10.2, J5,6 3.5, J5,6' 2.0 Hz, H-  5), 2.18, 2.07, 2.04 (3 s, 9 H, 3 OAc); l F-nmr data (254 MHz, C D C L 3 ) : 8 201.46 (dd, JF,2 9  51.6, JF.313.3 Hz, F-2), 234.35 (dt, JF,5 22.8, JF.6+6' 47.2 Hz, F-6). Anal. Calc. for C 1 2 H 6 F 2 O 7 : C.46.46; H.5.20. Found: C,46.65; H,5.20.  79 2,6-Dideoxy-2,6-difluoro-a-D-glucopyranosyl [bis(cyclohexylarnmoniurn) phosphate] (6). (2,6-difluoro-glucose 1-P). The peracetate 5 (0.40 g,1.29 mmol) was stirred with anhydrous phosphoric acid (0.89 g, 9.0 mmol) under vacuum for 9 hours at 50-55Mce cold aqueous 2 M lithium hydroxide (19 mL) was added and the mixture kept overnight at room temperature. After filtration through celite to remove precipitated lithium phosphate the solution was run through a cold column of Dowex 50W-X8 (H ) resin into an excess of cyclohexylamine. Lyophilisation of this solution gave a +  powder which was dissolved in a minimal amount of water. Acetone was added dropwise to the point of incipient opalescence. Cooling to 4° allowed the crystallization of a small amount of biscyclohexylammonium phosphate. The product itself could not be crystallized and was isolated as a freeze dried powder 6 (0.363 g, 0.71 mmol, 61%). Anal. Calc. for Cl8H37N2O7F2P.H 0: C,45.00; H.8.18; N.5.83. Found: C.45.15; 2  H.8.19; N.5.74.  3-Deoxy-3-fluoro-1,2-O-isopropylidene-a-D-glucofuranose (8). Methanol (12.5 mL) and dilute sulfuric acid (0.8% v/v,12.5 mL) were added to 3-deoxy3-fluoro-l,2:5,6-di-0-isopropylidene-a-D-glucofuranose (7,2.35 g, 8.97 mmol) and the mixture stirred for 8 hours at room temperature. Solid barium carbonate was added to neutrality and after boiling and filtration through celite, the residue was evaporated to leave a colorless oil 8 (1.86 g, 8.38 mmol, 93%); lH-nmr data (270 MHz, CDCI3): 5.85 (d,l H, Ji,2 4.0 Hz, H-l), 5.00 (dd,l H, J3,p 49.5,13,4 2.5 Hz, H-3), 4.60 (dd,l H, J2,F 10.5, J , l 4.0 Hz, H-2), 4.06 2  (ddd,l H, J 4 , F 34.5, J 5 9.5, J 4 3 2.5 Hz, H-4), 3.58-3.91 (m,3 H, H-5,6,6'), 2.95 (s,2 H, 4)  OH-5,6),1.43,1.26 (2 s, 6H, 2 Me).  3-Deoxy-3-fluoro-l,2-0-isopropyUdene-6-0-triphen^  (9).  To a solution of 8 (1.41 g, 6.36 mmol) in dry pyridine (12 mL) was added triphenylmethyl chloride (1.95 g, 7.0 mmol) and the solution heated to 40° for 16 hours. Ice (0.1 g) was added and after 2 hours the pyridine was removed by co-evaporation with toluene and the  80 oil dissolved in chloroform and washed with 10% sodium bicarbonate solution. After drying (magnesium sulfate) and filtration, the resulting oil was purified by flash chromatography (4:1 hexane, ethyl acetate) to yield a clear gum 9 (2.21 g, 4.76 mmol, 75%); lH-nmr data (300 MHz, C D C I 3 ) : 5 7.50-7.20 (m,15 H, 3 Ph), 5.94 (d,Jl,2 4.0 Hz, H-l), 5.12 (dd,l H J  3 ) F  50.1, J3.4  1.3 Hz, H-3), 4.70 (dd,l H, J2,F 10.4, J 2 , l 3.8 Hz, H-2), 4.27 (ddd,l H, J4,F 29.4, J4.5 9.2, J4,3 1.3 Hz, H-4), 4.02 (m,l H, H-5), 3.41 (dd,l H, J6,6' 10.0, J6,5 3.3 Hz, H-6), 3.37(dd,l H, J6',6 10.0, J6',5 5.0 Hz, H-6'), 2.46 (d,l H, JOH,5 6.0 Hz, OH-5), 1.49,1.33 (2 s, 6 H , 2 Me).  3-Deoxy-3-fluoro-l,2-0-isopropyMene-5-0-methoxymethyl-a-D-glucofuranose (10). Monochloromethyl methyl ether (1.08 mL,14.3 mmol) was added to a mixture of 9 (2.21 g, 4.76 mmol) and dry diisopropylethylamine (3.32 mL,18 mmol) in dry dichloromethane (60 mL).After 10 days at room temperature the solvent was removed,the residue dissolved in chloroform and extracted with ice cold 1 M hydrochloric acid followed by saturated sodium bicarbonate solution. After drying (magnesium sulfate) and evaporation, the oil was dissolved in diethyl ether (20 mL) and an equal volume of 70% formic acid added. After 20 minutes, more ether (100 mL) was added and the solution extracted with 10% sodium chloride solution followed by saturated sodium bicarbonate solution. Drying (magnesium sulfate) and evaporation followed by flash chromatography (1:1 hexane, ethyl acetate) yielded a clear oil 8.(0.79 g, 2.96 mmol, 62%); lH-nmr data (300 MHz, CDCI3): 8 5.97 (d, 1 H, Jl,2 4.0 Hz, H-l), 5.04 (dd, 1 H, J3,F 50.8, J3,4 2.3 Hz, H-3), 4.73 (m, 3 H.CH2, H-2), 4.21 (ddd, 1 H, J4,F 29.3, J4,5 9.5, J4,3 2.3 Hz, H-4), 3.88 (m, 2 H, H-6,6*), 3.70 (dd, 1 H, 15,4 11.0, J5,6 5.4 Hz, H-5),  3.44 (s, 3 H, OMe),1.49,1.32 (2 s, 6 H, 2 Me).  3,6-Dideoxy-3,6-mfluoro-l,2-0-isopropylidene-5-0-methoxymethyl-q-D-glucofurano (11). A solution of 10 (0.78 g, 2.93 mmol) and dry 2,4,6-trimethylpyridine (1.16 mL, 8.8 mmol) in dry, ethanol-free dichloromethane (8 mL) was treated with DAST (1.16 mL, 8.8 mmol) at -20° under anhydrous conditions. The solution was allowed to warm to room  81 temperature and stirred for 48 hours. Dichloromethane (30 mL) was added and the solution extracted with cold 1 M hydrochloric acid followed by saturated sodium bicarbonate solution. After drying (magnesium sulfate) and evaporation the residue was purified by flash chromatography (2:1 hexane, ethyl acetate) to yield a yellowish oil 11, (0.50 g,1.85 mmol, 63%); iH-n.m.r data (300 MHz, CDCI3): 5 5.98 (d, 1 H, Ji,2 3.9 Hz, H-l), 5.07 (dd, 1 H, J3,F 50.4,J34 2.1 Hz, H-3), 4.52-4.86 (m, 5 H, C H 2 , H-2,6,6'), 4.32 (ddd, 1 H, J4,F 29.1, J ,5 9.4, J ,3 2.2 Hz, H-4), 3.98 (dddd, 1 H, T 4  4  5 ) F  25.9, J5,4 9.4, J5,6 3.8, J5,6' 1.9 Hz, H-  5), 3.47 (s, 3 H, OMe), 1.53,1.36 (2 s, 6 H, 2 Me); 19F-nmr data (254 MHz, CDCI3): 8 202.51 (ddd, JF,3 50.0, JF,4 28.9, JF,2 10.5, F-3), 232.62 (dt, JF,5 25.2, JF,6+6* 47.3 Hz, F-6).  3,6-Dideoxy-3,6-difluoro-D-glucopyranose (12). A mixture of tetrahydrofuran, water and trifluoroacetic acid (1:1:4 v/v, 20 mL) was added to 11 (0.50 g,1.85 mmol) and the solution left at room temperature for 8 hours. The solvents were removed and the productfilteredthrough silica gel (ethyl acetate) yielding, after evaporation, a colorless gum 12 (0.29 g,1.58 mmol); l^-nmr data (254 MHz, D2O): 8 187.22 (ddd, JF,3 52.7, JF,4 13.6, JF,2 13.6 Hz, F-3),192.18 (ddd, JF,3 54.0, JF,4 13.3, JF,2 13.3 Hz,F-3), 227.15 (dt, JF,5 26.3, JF,6+6" 47.3, F-6), 227.67 (dt, JF,5 28.5, JF,6+6' 47.3 Hz, F-6).  1,2,4-Tri-0-acetyl-3,6-dideoxy-3,6-difluoro-B-D-glucopyranose (13). A solution of 12 (0.28 g,1.52 mmol) in dry pyridine was treated, appropriately scaled, exactly as in the synthesis of 5. The reactiontimefor the bromination reaction was 1 hour. Crystallization from ethyl acetate, pentane yielded colorless crystals 13 (0.37 g,1.19 mmol, 79%); m.p.l27-127.5°; lH-nmr data (300 MHz, CDCI3): 8 5.68 (d, 1 H, Ji,2 8.5 Hz, H-l), 5.26 (m, 2 H, H-2,4), 4.63 (ddd,l H, J3,F 48.8,13,4 9.0, J3,2 9.0 Hz, H-3), 4.55 (ddd.l H, J6,F 48.0, J6,6' 10.2, J6,5 3.6 Hz, H-6), 4.45 (ddd,l H, Jtf.F 48.0, J6',6 10.2, J 6  >5  5.4  Hz, H-6') 3.76 (dddd,l H, J5,F 21.2, J5,4 13.1, J5,6" 5.4, J5,6 3.6 Hz, H-5), 2.11,2.12,2.13  82 (3 s, 9 H, 3 OAc); ^F-nmr data (254 MHz, CDCI3): 5 196.53 (ddd, JF,3 51.8, JF,4 12.6, JF,2 12.6, F-3), 232.95 (dt, JF,5 21.3, JF.6,6' 47.0, F-6). Anal.  Calc. for C 1 2 H 1 6 O 7 F 2 : C.46.46; H.5.20. Found: C,46.67; H,5.39.  3,6-Dideoxy-3,6-difluoro-a-D-glucopyranosyl [bis(cyclohexylarnmonium) phosphate] (14). (3,6-difluoro-glucose 1-P). The peracetate 13 (0.25 g, 0.81 mmol) was treated with anhydrous phosphoric acid (0.56,5.7 mmol) for 3 hours exactly as in the synthesis of 6. Repeated crystallization yielded the colorless biscyclohexylammonium salt 14 (0.33 g, 0.72 mmol, 88%). Anal.  Calc. for C 1 8 H 3 7 N 2 O 7 F 2 P . H 2 O : C,45.00; H.8.20; N,5.86. Found: C.45.00;  H.8.18; N.5.83.  3-Deoxy-3-fluoro-l,2-0-isopropylidene-5,6-sulfite-a-D-glucofuranose (15,16). Reaction of 8 (0.15 g, 0.68 mmol) with DAST (0.36 mL, 2.72 mmol) in dry, ethanolfree dichloromethane (1.5 mL), performed exactly as in the preparation of 3, gave two major products in equal proportions. The two compounds were separated by flash chromatography (2:1 hexane, ethyl acetate, r.f. 0.36, 0.30) which was repeated twice to yield the oils 15 and 16. Isomer 1; lH-nmr data (400 MHz, CDCI3): 8 5.98 (d, 1 H, Ji,2 3.6 Hz, H-l), 5.10 (dd, 1 H, J3,F 50.0, J3,4 2.0 Hz, H-3), 4.60-4.78 (m, 4 H, H-2,5,6,6'), 4.50 (ddd, 1 H, J F 4  27.2, J 4 5 8.6, J 4 3 2.0 Hz, H-4), 1.50, 1.33 (2 s, 6 H, 2 Me); 19F-nmr data (254 MHz, CDCI3): 8 210.91 (JF,3 49.9, JF,4 27.1, JF,2 10.4 Hz); FAB mass spectrum (M+l) = 269. Anal.  Calc. for C 9 H 1 3 O 6 S F : C, 40.30; H, 4.88. Found: C, 40.60; H, 5.02.  Isomer 2; lH-nmr data (400 MHz, CDCI3): 8 5.98 (d, 1 H, Ji,2 3.6 Hz, H-l), 5.10 (ddd, J  5 > 4  7.6, J 6 6.5, J5,6'4.0 Hz, H-5), 5.04 (dd, 1 H, J 5  3 J  F  50.0, J3,4 2.0 Hz, H-3),  4.79 (dd, 1 H, J6.6' 9.0, J6,5 6.5 Hz, H-6), 4.73 (dd, 1 H, J2,F 10.3, J2,l 3.6, H-2), 4.58 (dd, 1 H, J6',6 9.0, J6',5 4.0 Hz, H-6'), 4.18 (ddd, 1 H, J4,F 28.3,14,5 7.6, J 4 3 2.0 Hz, H4), 1.50, 1.33 (2 s, 6 H, 2 Me); 19F-nmr data (254 MHz, CDCI3): 8 210.66 (JF,3 50.0, JF,4 28.3, JF,2 10.4 Hz); FAB mass spectrum (M+l) = 269.  83 Anal. Calc. for C 9 H 1 3 O 6 S F : C, 40.30; H, 4.88; S, 11.95. Found: C, 40.99; H, 5.05;  S, 11.58.  3-Chloro-3-deoxy-1,2-Q-isopropylidene-a-D-glucofuranose  (18).  3-chloro-3-deoxy-l,2:5,6-di-0-isopropylidene-a-D-glucofuranose (17, 2.9 g,10.4 mmol) was treated with dilute sulfuric acid and methanol exacdy as described in the synthesis of 8, yielding the oil 18 (2.2 g, 9.2 mmol, 89%); lH-nmr data (270 MHz, CDCI3): 8 5.92 (d,l H, Jl,2 3.1 Hz, H-l), 4.71 (d,l H, J2,l 3.7 Hz, H-2), 4.50 (d,l H, 13,4 3.1 Hz, H-3), 4.26 (dd,l H, J4,5 8.9, J4,3 3.1 Hz, H-4), 3.95 (m,l H, H-5), 3.87 (bd, 1H, J6,6' 10.0 Hz, H-6), 3.72 (bd, J6',6 10.0 Hz, H-6'), 3.39,3.13 (2 s, 6 H, OH-5,6), 1.50,1.32 (2 bs, 6 H, 2 Me).  3-Chloro-3-deoxy-1,2-Q-isopropylidene-6-0-trimethyl acetyl-a-D-glucofuranose (19). To a solution of 18 (2.0 g, 8.8 mmol) in dry pyridine (40 mL) at -40° was added trimethylacetyl chloride (1.23 mL,10.0 mmol) with stirring. The solution was allowed to warm to room temperature and after 15 minutes quenched with methanol (5 mL). The pyridine was removed with the aid of toluene.The residue was dissolved in dichloromethane and washed with cold 0.5 M hydrochloric acid followed by saturated sodium bicarbonate solution. The solution was dried (magnesium sulfate), filtered and after evaporation purified by flash chromatography (2:1 hexane, ethyl acetate) to yield a colorless oil 19 (2.2 g, 6.8 mmol, 77%);lH-nmr data (400 MHz, C D C I 3 ) : 8 5.95 (d,l H, Ji,2 3.5 Hz, H-l), 4.73 (d,l H , J2,l 3.5 Hz, H-2), 4.46 (m, 2  H, H-6,6'), 4.27 (dd,l H, J 9.2, 2.8 Hz), 4.22 (dd,l H, J 12.0, 5.5 Hz), 5.12 (m,l H, H-5),  3.52 (d,l H, JOH.5 5.0 Hz, OH-5), 1.50,1.31 (2 s, 6 H, 2 Me), 1.22 (s, 9 H, M e 3 C ) .  3-CMoro-3-cleoxy-l,2-0-isopropyh\iene-5-0-m  acetyl-a-D-  glucofuranose (20). Dry m^sopropylethylamine (9.3 mL, 54 mmol) and 19 (2.16 g, 6.7 mmol) were dissolved in dry dichloromethane (50 mL). After flushing with nitrogen and cooling to -20° monochloromethyl methyl ether (3.05 mL, 40.2 mmol) was added dropwise. After 3 days at  84  room temperature the solvent and ether were removed, the residue dissolved in dichloromethane, and then washed with cold 0.5 M hydrochloric acid followed by saturated sodium bicarbonate solution before drying (magnesium sulfate). The evaporated residue was purified by flash chromatography (4:1 hexane, ethyl acetate). The product was crystallized from hexane, ethyl acetate yielding 20 (2.18 g, 5.9 mmol, 89%); m.p. 92-93°; lH-nmr data (400 MHz, CDCI3): 8 5.95 (d,l H, Ji,2 3.6 Hz, H-l), 4.75 (dd, 1 H, J 10,10 Hz), 4.74 (d, J C H 2 , 5 6.1 Hz, O C H 2 O ) , 4.62 (dd,l H, J 12.3,1.9 Hz), 4.46 (m, 2 H, H-6,6'), 4.13 (dd, J 12.2, 3.2 Hz),  4.05 (dm,l H, 15,4 9.5 Hz, H-5), 3.40 (s, 3 H, OMe),1.50,1.32 (2 s, 6 H, 2 Me), 1.23 (s, 9H, M e 3 Q .  3-CMoro-3-deoxy-l,2-0-isopropylidene-5-0-methoxymethyl-q-D-glucofuranose (21). Dry methanol (10 mL) containing 0.2 M sodium methoxide was added to 20 (2.18 g, 5.9 mmol) and the reaction left at room temperature for 2 days. After neutralization as described in the preparation of 2 the residue was purified by flash chromatography (1:1 hexane, ethyl acetate) to give the oil 21 (1.53 g, 5.4 mmol, 93%); lH-nmr data (270 MHz, CDCI3): 8 5.96 (d,l H, Jl,2 3.5 Hz, H-l), 4.75 (d, 2 H, J C H 2 . 5 4.0 Hz, O C H 2 O ) , 4.71 (d,l H, J 3.5 Hz), 4.42 (d,l H, J 3.5 Hz), 4.31 (dd,l H, J 9.0, 3.5 Hz), 3.88 (m, 2 H, H-6,6'), 3.70 (m,l H, H-5), 3.44  (s, 3 H, OMe), 3.02 (dd,l H, JOH,6 4.0, JOH,6' 4.0 Hz, OH-6), 1.50,1.32 (2 s, 6 H, 2 Me).  3-Chloro-3,6-dUdeoxy-6-fluoro-l,2-Q-isopropylidene-5-Q-methoxymethyl-a-D-glucofuranose (22). A solution of 21 (1.53 g, 5.4 mmol) and 2,4,6-trimethylpyridine (2.16 mL,16.2 mmol) was treated with DAST (2.16 mL,16.2 mmol) exactly as for the preparation of 11 with a reaction time of 21 hours. Flash chromatography yielded a colorless gum 22 (0.75 g, 2.6 mmol, 49%); lH-nmr data (400 MHz, CDCI3): 8 5.95 (d,l H, J  1 ( 2  3.4 Hz, H-l), 4.78 (ddd,l H, J6,F 47.6,  J ,6' 10.2, J6,5 1.9 Hz, H-6), 4.62 (ddd,l H, J ',F 46.5, J6',6 10.2, J6',5 3.5 Hz, H-6'), 6  6  4.80 (d, 2 H, J C H 2 . 5 4.8 Hz, O C H 2 O ) , 4.75 (d,l H, J 2 , l 3.4 Hz, H-2), 4.46 (m, 2 H, H-  85 3,4), 3.96 (ddm,l H, J  5 > F  26.0, J5,4 9.0 Hz, H-5), 3.44 (s,3 H, OMe), 1.55,1.33 (2 s, 3 H, 2  Me).  3-Chloro-3,6-dideoxy-6-fluoro-D-glucopyranose (23). Deprotection of 22 (0.53 g,1.87 mmol) was achieved exactly as described for the synthesis of 12. After 5 hours at room temperature the solvent was removed to give crystalline material. Recrystalhzation from ethyl acetate, hexane yielded 23 (0.325 g,1.63 mmol, 87%); m.p. 159-161°; ^F-nmr data (254 MHz, D2O): 8 236.05 (dt, JF,5 26.2, JF,6+6' 47.3 Hz), 236.66 (dt, JF,5 28.6, JF,6,6' 47,4 Hz).  l,2,4-Tri-0-acetyl-3-chloro-3,6-dideoxy-6-fluoro-B-D-glucopyranose (24). A solution of 23 (0.31 g,1.55 mmol) in dry pyridine was treated as in the synthesis of 5, appropriately scaled. The reaction time for the bromination reaction was 1 hour. Crystallization from ethyl acetate, hexane yielded 24 (0.405 g,1.24 mmol, 80%); m.p. 145-146°; lH-nmr data (300  MHz, C D C I 3 ) : 8 5.68  (d,l H, Ji,2 8.6 Hz, H-l), 5.22  (dd,l H, J2,l 8.4, J2,3 8.4 Hz,  H-2), 5.19 (dd,l H, J 4 3 9.0, J4,5 9.0 Hz, H-4), 4.53 (ddd,l H, J6,F 46.6, J6,6' 10.0, J6,5 2.7 Hz, H-6), 4.39 (ddd, J6',F 46.6, J&,6 10.0, J6',5 4.4 Hz, H-6'), 4.03 (dd, J3,4 10.0, J3,2 10.0 Hz, H-3), 3.78 (dddd,l H, J5,F 21.1, J5,4 10.0, J5,6' 4.6, J5,6 2.7 Hz, H-5), 2.17, 2.12 (2 s, 9 H, 3 OAc); 19F-nmr data (254 MHz,  CDCI3):  8 232.49 (dt, JF,5 20.7,  JF,6+6' 47.0 Hz).  1,2,4-Tri-0-acetyl-3,6-dideoxy-6-fluoro-B-D-ribohexopyranose (25). The peracetate 24 (0.20 g, 0.61 mmol) was dissolved dry, distilled toluene (5 mL) containing aa'-azobisisobutyronitrile (0.003 g). The reaction vessel was flushed with dry nitrogen before the addition of tributyltin hydride (0.21 mL, 0.78 mmol). The mixture was heated at 80° for 16 hours prior to the removal of solvent The residue was purified by flash chromatography (1:1 hexane, ethyl acetate) and the product crystallizedfromethyl acetate, hexane to yield 25 (0.15 g, 0.49 mmol, 81%), m.p. 130.5-131°; lH-nmr data (270 MHz,  86 CDC13): 6 5.72 (d, 1 H, Ji,2 8.0 Hz, H-l), 4.90 (m, 2 H, H-2,4), 4.54 (ddd, 1 H, J6,F 47.8, J6,6' 10.0, J6,5 2.7 Hz, H-6), 4.44 (ddd,l H, J 6 \ F 47.4, J6\6 10.0, J6',5 4.0 Hz, H-6'), 3.82 (dddd,l H, J5,F 23.0, J5,4 9.5, J5,6' 4.0, J5,6 2.7 Hz, H-5), 2.68 (ddd, 1 H, J3eq,3ax  12.5, J3eq,4 5.0, J3eq,2 5.0 Hz, H-3 ), 2.11,2.06,2.04 (3 s, 9 H, 3 OAc), 1.69 (ddd, 1 H, eq  J3ax,3eq 12.0, J3ax,4 12.0, J3ax,2 12.0, H-3 ). ax  Anal. Calc. for C12H17O7F: C.49.32; H,5.86. Found: C.49.54; H,5.91.  3,6-Dideoxy-6-fluoro-a-D-ribohexopyranosyl [bis(cyclohexylarnmonium) phosphate] ( 2 6 ) , (3-deoxy-6-fluoro-glucose 1-P). The per-acetate 2 5 (0.15 g, 0.51 mmol) was treated with anhydrous phosphoric acid (0.35 g, 3.6 mmol) exactly as for the preparation of 6 with a reaction time of 2.5 hours. Three recrystallizations yielded the product 2 6 (0.15 g, 0.34 mmol, 67%). Anal. Calc. for C 1 8 H 3 8 N 2 O 7 F P : C,48.64; H.8.62; N.6.30. Found: Q48.28; H.8.46;  N,6.00.  Methyl 2,3-di-O-acetyl-a-D-galactopyranoside ( 3 0 ) . Methyl 2,3-di-0-acetyl-4,6-0-benzylidene-a-D-galactopyranoside  (29,5.36  g, 14.6  mmol) was treated with 80% aqueous acetic acid (70 mL) and the mixture heated to 60° for two hours. The solvent was removed and the gum further dried in vacuo. The product was crystallized from ethyl acetate, hexane to yield colorless crystals 3 0 (3.52 g, 12.6 mmol, 86%); m.p. 94-95°; lH-nmr data (300 MHz, CDCI3): 8 5.27 (m, 2 H, H-2,3), 5.02 (d, Ji,2 2.5 Hz,  H-l), 4.28 (bs, 1 H, H-4), 4.00-3.85 (m, 3 H, H-5,6,6'), 3.40 (s, 3 H, OMe), 2.17 (s, 2 H, OH-4,6), 2.12,2.09 (2 s, 6 H, 2 OAc).  Methyl 2,3-m-0-acetyl-4,6-dideoxy-4,6-difluoro-a-D-glucopyranoside ( 3 1 ) . A solution of 2 2 (1.50 g, 5.4 mmol) and 2,4,6-trimethylpyridine (4.3 mL, 32.6 mmol) in dry, ethanol-free dichloromethane (20 mL) was treated with DAST (4.31 mL,32.6 mmol) as described in the synthesis of 1 1 . After 20 hours the reaction was worked up and purified by  87 flash chromatography (50:1 dichloromethane, ethyl acetate) to give the gum 31 (0.47 g,1.69 mmol, 31%); lH-nmr data (270 MHz, CDCI3): 8 5.65 (ddd,l H, J F 14.0, J3,4 9.4, J  3 ) 2  3  9.4  Hz, H-3), 4.97 (dd,l H, J I , F 2.9, Jl,2 2.9 Hz, H-l), 4.84 (dd,l H, J2,3 9.7, J2,l 3.0 Hz, H2), 4.68 (bd, 2 H, J6+6',F 46.8, H-6,6'), 4.57 (ddd,l H, J4,F 50.0, J4,5 9.5, J4,3 9.5 Hz, H4), 4.00 (dm,l H, J  5  F  25.5, H-5), 3.46 (s, 3 H, OMe), 2.15, 2.12 (2 s, 6 H, 2 OAc);  1  9  F -  nmr data (254 MHz, C D C I 3 ) : 8 198.69 (dd, JF,4 50.5, JF,3 14.0 Hz, F-4), 236.24 (dt, JF,5 25.8, JF,6+6' 47.1 Hz, F-6).  2,3-Di-0-acetyl-4,6-dideoxy-4,6-difluoro-a-D-glucopyranosyl chloride (32). A solution of 31 (0.53 g, 1.88 mmol) in l,l'-dichloromethyl methyl ether (10 mL) was stirred with a catalytic amount of freshly fused zinc chloride under nitrogen for eight hours.at 65°. The remaining ether was evaporated and the residue was dissolved in chloroform. After washing twice with saturated sodium bicarbonate solution and drying (sodium sulfate), the product was purified byflashchromatography (dichloromethane) and crystallized from ethyl acetate, hexane to yield colorless crystals 32 (0.24 g, 0.84 mmol, 45%); m.p. 109-110°; lHnmr data (400 MHz, CDCI3): 8 6.20 (dd,l H, J I , F 3.5, Ji,2 3.5 Hz, H-l), 5.70 (ddd,l H,  J3,F 13.6, J3,2 8.5, J3,4 8.5 Hz, H-3), 4.87 (ddd,l H, J2,3 10.0, J2,l 4.0, J2,4 1.0 Hz, H2), 4.62 (dddd, 1 H, J F 46.0, J4,3 8.5, J4,5 8.5, J4,2 1.0 Hz, H-4), 4.46-4.71 (m, 2 H, H4 J  6,6'), 4.21 (dm, 1 H, J F 27.0 Hz, H-5), 2.05,2.03 (2 s, 6 H, 2 OAc). 5 >  1,2,3-Tri-0-acetyl-4,6-dideoxy-4,6-difluoro-B-D-glucopyranose (33). A solution of 32 (0.24 g, 0.84 mmol) in glacial acetic acid (3 mL) was treated with mercuric acetate (0.53 g,1.69 mmol) in the dark for one hour. Chloroform was added and the mixture was washed thrice with water, followed by saturated sodium bicarbonate solution. After drying (sodium sulfate) and removal of solvent, the product 33 was crystallized from diethyl ether, pentane (0.18 g, 0.60 mmol, 73%); m.p. 116-119°; lH-nmr data (270 MHz, CDCI3): 8 5.74 (d, 1 H, Ji,2 8.0 Hz, H-l), 5.39 (ddd, 1 H, J  3 j F  14.8, J3,4 9.2,13,2 9.2 Hz, H-3), 5.08  88 (dd, 1 H, J2,3 8.8, J2,4 8.8 Hz, H-2), 4.50-4.80 (m, 3 H, H-4,6,6'), 3.81 (dm, 1H, J 5 , F 26.0 Hz, H-5), 2.11,2.10,2.04 (3 s, 9 H, 3 OAc).  4,6-Dideoxy-4,6-difluoro-a-D-glucopyranosyl [bis(cyclohexylammonium) phosphate] (34). (4,6-difluoro-glucose 1-P). The per-acetate 33 (0.18 g, 0.60 mmol) and anhydrous phosphoric acid (0.41 g, 4.2 mmol) were treated for 3 hours exactly as described in the synthesis of 6. Three recrystaUizations yielded the product 34 (0.16 g, 0.34 mmol, 58%). Anal. Calc. for C 1 8 H 3 7 N 2 O 7 F 2 P : C,46.75; H,8.00; N.6.06. Found: C.46.47; H.8.04; N,6.10.  Methyl 2,3-di-0-acetyl-4-deoxy-a-D-xylohexopyranoside (40). To a solution of methyl 4-deoxy-a-D-xylohexopyranoside (39,0.20 g, 1.12 mmol) in dry pyridine (2 mL) was added triphenylmethyl chloride (0.51 g, 1.84 mmol) and the solution heated to 40° under anhydrous conditions for 24 hours. More pyridine (3 mL) was added, the solution cooled to 0° and acetic anhydride (3 mL) added. After 2 days the solution was poured into saturated aqueous sodium bicarbonate and stirred 2 hours. The solution was extracted with chloroform and dried (magnesium sulfate). The solvents were removed with the aid of toluene and the residue dissolved in diethyl ether (3 mL). Formic acid (3 mL,70%) was added and after 20 minutes more ether (20 mL) was added. The mixture was extracted with water followed by saturated sodium bicarbonate solution, dried (magnesium sulfate) and the solvent removed. The residue was purified by flash chromatography (4:1 hexane, ethyl acetate) yielding the gum 40 (0.10 g, 0.38 mmol 34%); lH-nmr data (300 MHz, CDCI3): 8 5.31 (ddd, 1 H, J3,4ax 10.4, J3,2 10.4, J , 4 e q 5.2, H-3), 4.94 (d, 1 H, h,2 3.5 Hz, H-l), 4.84 (dd, 1 H, J 3 10.4, J , l 3  2 >  2  3.5 Hz, H-2), 3.95 (m, 1 H, H-5), 3.69 (d, 1 H, J6,6' 12.0 Hz, H-6), 3.56 (d, 1 H, J6',6 12.0 Hz, H-6'), 3.39 (s, 3 H, OMe), 2.02-2.18(m, 7 H, 2 OAc, H 4 ) , 1.63 (ddd, 1 H, T e q  11.0, J4ax,4eq H A J4ax,5 H-0 Hz, H - 4 ) . ax  4 a x 3  89  Memyl23-di-0-acetyl-4,6-cUdeoxy-6-fluoro-a-D-xylohexopyranoside (41). Compound 40 (0.46 g, 1.75 mmol) and 2,4,6-trimethylpyridine (0.46 mL, 3.51 mmol) were dissolved in dry, ethanol-free dichloromethane (10 mL). The mixture was treated with DAST (0.46 mL, 3.51 mmol) as described in the synthesis of 11, worked up after 28 hours and purified by flash chromatography (3:2 hexane, ethyl acetate) to yield the gum 41 (0.31 g,1.17 mmol, 66%); lH-nmr data (270 MHz, CDCI3): 8 5.31 (ddd, 1 H, J3,4ax 10.5, J3,2 10.5, J3,4eq 5.0 Hz, H-3), 4.95 (d,l H, Ji,2 4.0 Hz, H-l), 4.87 (dd,l H, J2.3 10.5, J2,l 4.0 Hz, H-2), 4.42 (dm, 2 H, J6+6',F 48.0 Hz, H-6,6'), 4.08 (m,l H,H-5), 3.40 (s, 3 H, OMe), 2.18 (dm,l H, J4eq,4ax 12.0 Hz, H-4eq), 2.02,2.08 (2 s, 6 H, 2 OAc), 1.64 (ddd,l H, J4ax,3 ;  12.0, J4ax,5 12.0, J4ax,4eq 12.0 Hz, H4ax).  2,3-Di-0-acetyl-4,6-dideoxy-6-fluoro-a-D-xylohexopyranosyl chloride (43). A solution of 42 (0.30 g,1.14 mmol) in dichloromethyl methyl ether (6 mL) was treated exactly as in the preparation of 32. Purification was achieved by flash chromatography (2:1 hexane, ethyl acetate) to yield the gum 43 (0.22 g, 0.82 mmol, 72%); lH-nmr data (270 MHz, CDCI3): 8 6.35 (d, 1 H, Ji,2 4.0 Hz, H-l), 5.40 (ddd,l H, J3,4ax 10.5, J3,2 10.5, J3,4eq 5.0 Hz, H-3), 4.96 (dd,l H, J2,3 10.5, J2,l 4.0 Hz, H-2), 4.33-4.61 (m, 3 H, H-5,6,6'), 2.28 (ddd,l H, J4eq,4ax 12.0, J4eq,3 4.8, J4eq,5 2.1 Hz, H-4 ), 2.12,2.07 (2 s, 6H, 2 OAc), eq  1.80 (ddd,l H, J x,3 12.0, J x,5 12.0, J4ax,4eq 12.0 Hz, H4 ). 4a  4a  ax  1,2,3-Tri-0-acetyl-4,6-dideoxy-6-fluoro-P-D-xylohexopyranose (44). A solution of the chloride 43 (0.30 g, 1.13 mmol) and mercuric acetate (0.54 g, 1.7 mmol) in glacial acetic acid (5 mL) was treated exactly as described for the preparation of 33. The product 44 was crystallized from diethyl ether, pentane (0.25 g, 0.85 mmol, 76%); m.p. 102.5-103°; lH-nmr data (270 MHz, CDCI3): 8 5.68 (d, 1 H, Ji,2 8.0 Hz, H-l), 5.02-5.08 (m, 2 H, H-2,3), 4.45 (dm, 2 H, J6+6',F 47.0 Hz, H-6,6'), 3.91 (m, 1 H, H-5), 2.04-2.18 (m, 7 H, H-4eq, 2 OAc), 1.74 (ddd, 1 H, J  4ax  ,3 12.0, J4ax,5 12.0, J4ax,4eq 12.0 Hz, H4 ).  Anal Calc. for C12H17O7F: C, 49.32; H, 5.86. Found: C, 49.44; H, 6.00.  ax  90  4,6-Dideoxy-6-fluoro-a-D-xylohexopyranosyl [bis(cyclohexylammonium) phosphate] (45). (4-deoxy-6-fluoro-glucose 1-P). The per-acetate 44 (0.15 g, 0.51 mmol) and anhydrous phosphoric acid (0.35 g, 3.6 mmol) were treated for 2.5 hours exactly as described in the synthesis of 6. The brown biscyclohexylammonium salt was recrystallized thrice to yield an off white powder 45 (0.056 g, 0.13 mmol, 25%). Anal.  Calc. for Ci8H38N20i7FP: C, 48.64; H, 8.62; N, 6.30. Found: C, 48.71; H,  8.58; N, 6.37.  Uiidine-5'-diphospho-2-deoxy-q-D-glucopyranose (47). To a solution of 100 mM Tris buffer (20 mL), pH 7.6 containing 30 mM magnesium chloride was added uridine-5'-triphosphate (0.168 g, 0.274 mmol), 2-deoxy-D-glucose 6-P (45, 0.070 g, 0.228 mmol), ct-D-glucose 1,6-diP (0.1 njnol) and bovine serum albumin (0.15 mg/mL). After readjustment of the pH, phosphoglucomutase (100 units), uridine-51  diphosphoglucose pyrophosphorylase (25 units) and inorganic pyrophosphatase (10 units) were added and the reaction incubated at 25°. The course of the reaction was monitored by the appearance of inorganic phosphate using the Fiske-SubbaRow method. After 6 days the solution was diluted with water to 150 mL and the pH adjusted to 8.0. The mixture was chromatographed at 4° on a DE-52 cellulose column (1.8 x 30 cm) which had been previously equilibrated with 30 mM ammonium bicarbonate. After washing with 200 mL water, the product was eluted with a gradient of 0 to 0.2 M ammonium bicarbonate at a flow rate of 2 mL/min in a total volume of 2 L. The desired product was the first UV active peak, eluting at 60-70 mM salt, a concentration characteristic of dianions. After repeated freeze drying to remove the volatile buffer, inorganic phosphate, which coeluted with the product, was precipitated by addition of 2 M hthium hydroxide to pH 13. The precipitated lithium phosphate was removed by filtration through celite. The pH of the filtrate was readjusted to pH 7.6 by addition of Dowex 50 W-X8 H resin and a +  91  white solid obtained by lyophilization. P-nmr (121.4 MHz, D 2 O , {^H}): 8 8.50 (d, Jp,p 21 31  Hz), 10.53 (d, Jp,p 21 Hz).  2-Deoxy-a-D-glucopyranosyl [bis(ammonium)phosphate] (46) (2-deoxy-glucose 1-P). The uridine-5'-diphospho-2-deoxy-a-D-glucopyranose (47) was added to 100 mM Tris (20 mL), pH 7.6 containing 30 mM magnesium chloride, sodium pyrophosphate (1.0 g, 2.25 mmol) and bovine serum albumin (0.15 mg/mL). Uridine-5'-diphosphoglucose pyrophosphorylase (25 units) was added and the solution incubated at 25°. The course of the reaction was monitored by lp-nmr. After 3 days the inorganic phosphate was removed as 3  described above, the pH readjusted to 8.0 and the mixture chromatographed on a DE-52 cellulose column (2.8 x 40 cm) exactly as described for previous compound. The product was identified by the presence of phosphate according to the method of Bartlett (1959), and the buffer removed by repeated freeze drying to yield a white foam. 31p-nmr was performed at pH 5.0, the solution being unbuffered.  3- deoxy-3-fluoro-a-D-glucopyranosyl 6-phosphate [tetra (ammonium)phosphate], (3-fluoroglucose 1,6-diP). This was prepared by a modification of the method of Hanna and Mendocino (1970). A solution of 3-fluoro-glucose 6-P bis-ammonium salt (0.155 g, 0.52 mmol) was converted to the pyridinium salt via a column of Dowex 50W-X8 (H ) resin and the solution lyophilized.Dry +  pyridine (5 mL) was added and then evaporated thrice in order to completely dry the gum. More dry pyridine (6 mL) and acetic anhydride (2 mL) was added and the mixture stirred rapidly at room temperature for two hours to effect dissolution. The reaction was then kept at 4° under anhydrous conditions for two days. The solvents were evaporated and the resulting gum kept in vacuo over phosphorous pentoxide for 24 hours. The dried gum was dissolved in anhydrous tetrahydrofuran (10 mL) and anhydrous phosphoric acid (0.35 g, 3.6 mmol) added. After removal of solvent the resulting gum was heated to 55° in vacuo for 2.5 hours. Ice cold 2 M lithium hydroxide solution was added and the pH of the thoroughly mixed solution adjusted to  10.0. The pH was readjusted to 10.0 over the next several hours before being left over night at room temperature. The precipitated lithium phosphate was removed by filtration through celite and the pH adjusted to 8.0 with Dowex 50-W X8 (H ) resin. The product was purified at 4° on +  a DE-52 cellulose column (1.6 cm x 30 cm) which had been previously equilibrated with 50 mM ammonium bicarbonate.The chromatography was performed exactly as in the purification of 47 except the product was eluted with a linear gradient of 0-0.3 M ammonium bicarbonate. The desired product eluted at approximately 0.15 M ammonium bicarbonate. Repeated lyophilzation yielded an off white powder which contained approximately 16% p-anomer as ctetermined by 31  P  -nmr (0.055 g, 0.13 mmol, 26%). The product contained approximately 10% of an unknown impurity and was rechromatographed in the same manner to effect its removal, the eluting salt gradient being 0.05-0.25 M; a-anomer, lH-nmr data (300 MHz, D2O): 8 5.25 (ddd,l H, Ji,p 7.0, Ji,2 3.3 J l , F 3.3 Hz, H-l), 4.65 (ddd, 1-H, J 3 , F 51.0, J3,4 8.9, J3,2 8.9 Hz, H-3), 4.2-3.8 (m, 5 H, H-2, 4, 5, 6, 6'), P-nmr data (121 MHz, D2O, { lH}): 8 -7.20 31  (P-6), -5.04 (P-l), ^-nmr data (254 MHz, D2O,): 8 201.29 (ddd, J F , 3 55, J F , 2 13, J F , 3 13 Hz); P-anomer, 31p-nmr data (121 MHz, D2O, { H}): 8 -7.20 (P-6),-5.48 (P-l), F-nmr data l  19  (254 MHz, D2O): 8 196.03 (ddd, J F , 3 54, J F , 2 13, J F , 3 13 Hz). II.C.3: Biological Methods. In all cases phosphoglucomutase was pre-activated prior to assaying by the method of Peck and Ray (1971). All kinetic measurements were made at pH 7.4 and 30°. The monofluoro and monodeoxy glucose 1-P analogues were gifts from Mr. Ian Street. All reagents used in these procedures which were not synthesized or obtained as already described were obtained from the Sigma Chemical Co. and used without further purification. All were of the highest quality available. Absorbance measurements were conducted on Pye-Unicam PU 8800 UV/Visible recording spectrophotometer attached to a Julabo VI circulating thermostat bath. All volumes were measured by grade A volumetric pipette or Hamilton pi syringes. Weights were measured on micro or semi-micro balances. The pH of solutions were determined by a Radiometer PHM82  93  pH meter. The concentration of phosphoglucomutase for kinetic measurements was determined by absorbance at 278 nm, a l%w/v solution having an absorbance of 7.7 (Peck & Ray, 1971).  Determination of substrate and inhibitor concentrations. The concentrations of stock solutions of glucosyl phosphates were measured by the determination of acid labile phosphate by the method of Peck and Ray (1971). The small amount of inorganic phosphate present was determined by the method of Fiske and Subbarow (1925) and the relative concentration of any p-anomer present was determined by l^F or 31p-nmr. Acid stable phosphate esters were determined using the more forcing hydrolysis conditions of Bartlett (1959).  Determination of kinetic parameters. 1.Substrates: The activity of phosphoglucomutase at saturating substrate and cofactor concentrations was determined by a colorimetric assay (Peck & Ray, 1971). Determination of Km and Vmax for each substrate was made by the colorimetric method of Ray and Roscelli (1964). The use of new test tubes, washed thrice with doubly deionised water was essential to ensure the production of reproducible results. Time courses of the substrates were determined by the same assay procedure. Hydrolysistimesof each fluoro-glucose 1-P at 100° were as follows: 2-fluoro, 30 min; 3-fluoro, 20 min and 4-fluoro, 20 min. Hydrolysistimesfor the deoxysubstrates were all 10 min. 2: Inhibitors: The apparent K of glucose 1-P was determined in the presence of a range m  of inhibitor concentrations, plus one deterrrtination with no inhibitor. The determinations were made by the coupled enzymic assay rather than the technically more demanding colorimetric assay. The assay mixture used was: 0.4 mL assay buffer containing 25 mM Tris buffer, pH 7.4, 2.5 mM MgCl2,1.3 mM EDTA, 30 uM NADP, 0.4 U/mL glucose 6-P dehydrogenase (Sigma, G5760), 1.3 |J.M glucose 1,6-diP; 0.1 mL dilute phosphoglucomutase (approx. 0.006 U/mL) in activation buffer, 0 to 50 uX, inhibitor in 25 mM Tris, pH 7.4. The reactants were incubated for ten minutes in cuvettes in the spectrophotometer prior to the initiation of the reaction by addition  94  of substrate. The total volume was made to 0.60 mL prior to the addition of substrate. The change in absorbance was monitored at 340 nm. Control experiments were performed to ensure that none of the reagents was present in rate hmiting quantities and that the observed rate was not hmited by the rate of anomerisation of glucose 6-P.  UV Spectral studies. Demetallated enzyme and inhibitors were prepared as described in the experimental procedures section of Chapter rV, procedures for the handling of demetallated solutions are described therein. All spectra were measured at 25° using a spectrophotometer band path width of 0.5 nm and a scanning speed of 0.2 nm/s. Difference spectra were measured in 20 mM Tris buffer, pH 7.5 containing 1 mM EDTA, the final enzyme concentration being approximately 30 UM. Baseline corrections were performed with the split cell containing buffer only and were subtracted from the protein spectrum by the spectrophotometer. Solutions of enzyme and inhibitor were centrifuged in a benchtop Eppendorf microcentrifuge prior to addition to each half of the split cell. The concentration of the ligand was sufficient to effectively saturate the enzyme. After equilibration for ten minutes, the spectra were measured and the data transferred for storage on disk via an interfaced Apple He computer. After thorough mixing and re-equilibration, the spectrum of the enzyme-ligand complex was measured and the data stored. The difference spectrum was obtained by subtraction of the spectrum of enzyme and inhibitor from that of the binary complex using a computer graphics package "Vidichart". Increased signal to noise ratio of the difference spectrum was obtained by measuring individual spectra (before and after mixing) several times and subtraction of the sums of these spectra. Each individual spectrum measured prior to and after mixing was identical and did not change over a period of one hour.  95 C H A P T E R HI: C H A R G E S T A T E A N A L O G U E S A N D P O T E N T I A L C O V A L E N T  INHIBITORS.  PLA: INTRODUCTION. in.A.l: Substrate Ionization States. A complete uncterstanding of the specificity and mechanism of phosphoglucomutase requires a knowledge of the charge state of the bound glucose-phosphate substrate. The pK  a  values of the two ionizations of both glucose 1-P and glucose 6-P are 1.1,6.13 and 0.94, 6.11, respectively (Corbridge, 1985) and hence at physiological pH, (near pH 7), the phosphate species will be present in solution in both mono and dianionic forms. Thus the enzyme could in principle bind either the mono or the dianionic species of either substrate. Several methods have been used to determine the charge state of bound ligands that have several possible ionization states. Studies of the pH dependence of an enzyme's kinetic parameters (K and kcat) have been m  of use in determining ionizations of both free enzyme and substrate, and the binary complex. Plots of pH versus kcat/Km follow the ionizations of both free enzyme and free substrate thus enabling the relative specificity of the enzyme towards different ionic states of the substrate to be measured. However, because the method is also a measure of the ionization of the free enzyme, the results can become ambiguous due to the coincidence of several ionizations. Moreover, because of a disregard for many assumptions necessary for correct interpretation, the entire validity of many pH dependence studies has been questioned (Knowles, 1976). A second method which is useful when the substrate contains a phosphate group, is the use of31p-nmr of the enzyme-ligand complex. The chemical shift of phosphate and phosphate esters is strongly dependent on their ionization state; for example, a pH change of 7.4 to 6.2 produces a change in chemical shift of 1.4 ppm for inorganic phosphate (Meyer et al, 1982). Hence the construction of atitrationcurve of the ligand's chemical shift and comparison with the chemical shift of the ligand in the binary complex can give information on the ionization state of the bound species. However, this method requires on the order of gram amounts of pure enzyme  96 and again can produce ambiguous results. A phosphate group bound in an active site can be forced into a strained conformation and it is known that change of the O-P-0 bond angle can cause a much larger chemical shift change than that due to a change of ionization state (Chlebowski et al., 1976). In fact the argument has become rather circular as it has been proposed (Gorenstein, 1984) that a change in chemical shift caused by ionization is in fact attributable to an O-P-0 bond angle effect. A more useful approach which yields on the whole less ambiguous information is the use of chemically modified ligands. Ideally a modification would only alter the charged group itself, either by changing the pK values or by addition or deletion of a whole a  charge, and not affect the remainder of the molecule to any great degree. Analysis and comparison of the enzyme-ligand dissociation constants of both parent and modified ligands (at the same pH) should give easily interpretable information on the enzyme's requirements for the charge state of the ligand. The pK values of phosphate esters have been modified by a variety of chemical a  substitutions. 1. Thiophosphates. The substitution of a non-bridging oxygen of a phosphate ester by sulfur results in a thiophosphate analogue. The substitution by sulfur lowers pKa2 of phosphate from 7.2 to 5.75 (Yount, 1975) and thus a thiophosphate ester analogue would be expected to be more negatively charged (more acidic) than its parent compound at pH 7. The long P-S bond length (2.1 A) and different bond angle would make a bridging sulfur containing species rather more unlike its oxygen containing parent than the non-bridging sulfur analogue and hence complicate any interpretation made from binding data. 2. Imidophosphates. The replacement of a bridging oxygen by a nitrogen results in an imidophosphate. The pKa2 of the terminal phosphate of the B-7 nitrogen-linked ATP analogue is 7.0 compared to 7.8 for ATP. Thus the terminal phosphoramidate is approximately a ten fold weaker acid than the terminal phosphate of the parent compound.  97 3. Phosphonates. The replacement of the bridging oxygen of a phosphate ester by a methylene group results in an extremely stable phosphonate species. The second pKa of a-D-glucose 1-phosphonate is 7.35 (LP. Street, unpublished results) compared to 6.1 for its phosphate parent and hence the net charge of the phosphonate species will be less than that of glucose 1-P at physiological pH. Crystallographic data have shown that simple phosphonate and phosphate esters have very similar bond lengths and bond angles (Engels, 1977). However differences between the two 'isosteres' may arisefromthe loss of the hydrogen bonding potential of the esterified oxygen resultingfromits absence in the analogue, and any steric effect brought about by the replacement of oxygen by a methylene group. Halophosphonates have found applicability as hydrolysis resistant, isopolar (i.e., having the same p K values) analogues of pyrophosphoric acid and ATP (Blackburn et al., 1981a, a  1981b). Although the phosphonate analogues of the above species are stable and isosteric, in some cases it is desirable to use an isbpolar, non-hydrolyzable analogue. Fluorophosphonate analogues meet this criterion as the electronegativity of the C F 2 fragment is approximately the same as that of oxygen. The preceding types of chemical modification have found wide use in the generation of ATP analogues but so far have had relatively little use in the area of sugar phosphate analogues.  4. q-Halophosphates. Substitution of the functionality a to the phosphate group with a halide, generally fluorine, results in a more acidic a-halophosphate. The effect is due to the large inductive effect of fluorine which can extend over several bonds. For example, the pKa2 of glucose 1-P is 0.2 units higher than that of its 2-fluoro analogue, a P-halophosphate (Withers et al., 1986a). A more direct approach than the previous examples is the complete removal of one of the phosphate's hydroxyl groups and its replacement by the non-ionizable hydrogen or fluorine atom.  98 III.A.2: Phosphite Esters as Phosphate Analogues. Phosphorous acid (H3PO3) and phosphoric acid (H3PO4) are closely related structurally, despite their differences in valence state (+3 and +5 respectively). X-ray diffraction studies have shown that phosphate trianion (PO$-) and phosphite dianion (HPO32-) have a tetrahedral structure. Phosphorous acid and its mono and diesters can exist in two tautomeric forms, (Figure DXl) although from spectroscopic evidence it appears that the form with a direct hydrogen-phosphorus bond is the major species in most cases (Corbridge, 1974). O  [ HOQ^OH Phosphoric acid  OH  O  J  J  HO^^OH  " HO^J^OH  Phosphorous acid  Figure DXl: The structure of phosphoric and phosphorous acids.  An investigation of the crystal structure of 45 phosphorous acid salts (HPO32-) gave an average P-0 bond length of 1.51 A and an O-P-O bond angle range of 110-115°. This compares with an average P-O bond length of 1.54 A and O-P-O bond angle range of 103-115° for 38 salts of phosphoric acid (PO^-) (Corbridge, 1974). The crystal structure of potassium glucose 1-P has been determined and the phosphate monoester functionality was shown to have a distorted tetrahedral structure (Beevers & Maconochie, 1965). The bridging P-0 bond length was found to be slightly larger than that of the non-esterified oxygens and the RO-P-0 bond angle slightly smaller than the O-P-O bond angles. By analogy with the correspondence between phosphate salts and monoesters, phosphite monoesters most likely have a distorted tetrahedral structure. Thus a phosphate monoester will only differ significantly from its phosphite analogue in that the former possesses a P-0 bond whereas the latter has a P-H bond. The dissociation constants of phosphoric and phosphorous acid are compared in Table DI.1.  99  pK l  pKg2  pKa3  H3PO4  2.1  7.1  12.1  H2PO3  1.3  6.7  a  Table IH.1: Dissociation constants (25 V) ofphosphoric and phosphorous acids. (From Corbridge, 1985).  Esterification of a hydroxyl group of phosphoric or phosphorous acid results in the loss of the last ionization. Thus a phosphate monoester has two ionizations of approximately pK 1 a  and 6. By analogy with the pK values of phosphate monoesters, phosphite monoesters will a  have a single ionization of pK approximately 1 and therefore at physiological pH will be a  completely ionized. D-Glucose 6-phosphite has been synthesized and its biological activity investigated (Robertson & Boyer, 1956). The authors found that glucose 6-phosphite was neither a substrate nor inhibitor of hexokinase, glucose 6-phosphatase or glucose 6-phosphate dehydrogenase. However the phosphite was found to be fairly unstable, being hydrolyzed to the extent of 10% in six hours at room temperature. a-D-Glucose 1-phosphite has also been synthesized (Ogawa & Seta, 1982) although no study on its biological activity has been published. In our laboratory we have been unable to repeat the synthesis, the phosphite group being hydrolyzed on deprotection of the per-acylated and per-benzylated glucosyl-phosphite intermediates (Johnston, 1984; M.D. Percival, unpublished results). Thus phosphite esters seem to have limited use as charge state analogues due to their instability. Stability is an important criterion for the selection of a substrate analogue. If the analogue can decompose to a substrate, then even a small amount of decomposition will lead to some activity. In such cases it can be very difficult to distinguish between activity due to the substrate analogue itself, or that due to its decomposition product  100  III.A.3: Phosphorofluoridate Esters as Phosphate Analogues. Phosphorofluoridates (replacement of hydroxyl byfluorine)on the other hand are relatively stable and are structurally similar to phosphites. X-ray crystal structure investigations of BaP03F and PbP03F indicate that the salts are approximately tetrahedral in structure (Corbridge, 1974). The replacement of the hydroxyl byfluorineis a sterically conservative change (see Section n.A.3), and hence no adverse repulsive steric interactions between the enzyme and modified phosphorofluoridate ligand should occur. Phosphorofluoridic acid is a strong acid and hence a phosphorofluoridate monoester will carry a single charge over the complete range of physiological pH values. The synthesis of phosphorofluoridate monoesters is easily achieved by the treatment of the parent phosphate monoester with 2,4-dinitrofluorobenzene in N,N-dimethylformamide in the presence of a hindered base (Wittmann, 1962). Other workers have shown that the Malkylamine generally used has no catalytic effect on the rate of reaction (Johnson et al, 1975). The same authors investigated the mechanism of thefluorinationof a nucleoside phosphate ester by 2,4dinitrofluorobenzene. They found that the mechanism involves the formation of a phosphate diester followed by displacement of 2,4-dinitrophenolate byfluorideion (Figure JJJ.2).  0N 2  R-  N0  2  R—O—P—F  Figure in.2: Reaction mechanism for the formation ofphosphorofluoridate monoesters by reaction with 2,4-dinitrofluorobenzene.  101  Phosphorofluoridate esters have found most of their biological applications as nucleoside phosphate analogues. They are not cleaved by acid or alkaline phosphatases but are defluoridated by snake venom phosphodiesterase to release the nucleoside phosphate and fluoride (Johnson et ai, 1975). Phosphorofluoridic acid and phosphorofluoridate salts are slowly hydrolyzed by water to produce phosphate (Corbridge, 1985). Phosphorofluoridate esters have been used in studies to determine the charge state of the bound parent ligand in a number of enzyme systems. In the study of Withers and Madsen (1980) it was found that adenosine 5'-phosphorofluoridate has a very low affinity (Ki = 2.7 mM) for the enzyme phosphorylase b compared to that of the parent phosphate (K = 40 \iM) and did not activate the enzyme to any extent. The results were a  interpreted as a strict requirement for a dianionic phosphate moiety. Because of the structural similarity between phosphorofluoridate monoesters and diisopropylphosphorofluoridate it has been suggested that the former could behave as affinity labels of their respective enzymes. Diisopropylphosphorofluoridate is an affinity label towards many serine esterases causing irreversible inhibition by forming a stable serine phosphate bond by nucleophilic attack of the serine hydroxyl on the phosphorus center and displacement of fluoride (Balls & Jansen, 1953; Walsh, 1979) (Figure HI.3).  CH—Od  >•  CH3  HC 3  N  HgC^  j?  .CH  jL  CH3  CH—O—P—O—CH  3  + F  E—OH  Figure LTI.3: Mechanism of  covalent inhibition of serine proteases by  diisopropylphosphorofluoridate.  However no phosphorofluoridate monoesters have been shown to afford any covalent inhibition towards enzymes. The nature of enzymic phosphoryl transfer reactions may have a bearing on the unreactivity of phosphorofluoridate monoesters in enzymic systems. Four mechanisms for phosphoryl transfer have been proposed and are summarized in Figure HIA  102  O R  —  O  —  P  —  O O  R - O -  N  O—V—N / O  O R  —  O  —  P  ^  -  R—0"  O  L d 0 J V  A .Dissociative via a monomeric metaphosphate.  O  P R  —  O  —  P  —  O  •  R  —  O  -  d o  —  N  v  B. Associative via a pentacoordinate transition state.  O  R—O  R -  t  N  C. Associative via a pentacoordinate intermediate. O O  R — O — P ^ - 0 N  o  •  D. Associative via a pentacoordinate intermediate and pseudorotation. Figure III.4: Proposed mechanisms of phosphoryl transfer. (From Knowles, 1980).  The metaphosphate mechanism, A, in which bond breaking occurs before bond formation is analogous to that of an S N I type of mechanism in carbon chemistry. The metaphosphate ion is a highly reactive electrophilic species and its existence has only been proven in a few cases although much mdirect evidence for its importance does exist. This dissociative mechanism can in principle proceed with inversion of configuration or racemization, depending on the ufetime of the metaphosphate species. The in-line associative mechanism, B,  103  proceeding via a pentacoordinate transition state is analogous to the SN2 type of mechanism of carbon chemistry. It proceeds via a trigonal bipyramidal species, as do the other associative mechanisms, and leads to inversion of configuration. The second associative mechanism, C, proceeds via a pentacoordinate species which may have a lifetime long enough to undergo one or more pseudorotations before the leaving group departs. A pseudorotation involves a reorganization of the phosphorus ligands to allow the best leaving group to take up the apical position. Apical ligands generally have rather weaker bond strengths than those of equatorial ligands and consequently apical ehmination is the more favored process (Westheimer, 1968; Corbridge, 1985). Thus this mechanism can either proceed with retention (if accompanied by a pseudorotation) or inversion of configuration. The adjacent associative mechanism, D, involves attack at a position adjacent to the leaving group. The pentacoordinate intermediate undergoes a pseudorotation and the leaving group departs from an apical position. This mechanism results in retention of configuration. To date, all enzymic phosphoryl transfer reactions have been found to involve an inversion of configuration. This and other evidence has led to the proposal that enzymic phosphoryl transfer reactions occur via the in-line associative mechanism whereby the leaving group must leave from an apical position, 180° from the mcoming nucleophile with no pseudorotations. In a reaction involving phosphoryl transfer to an enzyme, the phosphate ester substrate would probably be held rigidly in a fixed orientation relative to the mcoming nucleophile. The orientation of the substrate would be correct for phosphoryl transfer, i.e., the leaving group is in-line with the nucleophile. Obviously in the case of a phosphorofluoridate monoester substrate analogue, the nucleophile would have to attack from a different angle in order for the reaction to result in the expulsion of fluoride and the covalent attachment of the substrate to the enzyme. Presumably in the case of diisopropylphosphorofluoridate the inhibitor can attain an orientation in which the fluorine is in-line with the serine hydroxyl. An added advantage in the use of phosphorofluoridate monoesters is the introduction of a second more sensitive nmr nucleus. l^F has a relative sensitivity thirteentimesthat of 31p (at constant field strength) and hence binding studies using l F-nmr make more efficient use of both 9  protein and ligand. In a recent study (Monasterio, 1987) ^F-nmr of the ternary tubulin-Mn 2+  104 guanosine 5'-(y-fluorophosphate) complex was used to measure the distance between the positions of the metal and the fluorine and also to place a limit on the rate of nucleotide exchange. In the present study a series of novel phosphorofluoridate analogues of the phosphoglucomutase substrates has been synthesized. The aim of their synthesis was to provide information on which ionic form of the substrate is bound initially in the enzyme's active site. It was also envisaged that the phosphorofluoridate monoesters may behave as suicide substrates resulting in the covalent attachment of the glucose moiety to the serine-116 residue of the active site. This is shown schematically in Figure III.5.  6—P—O—E  A Figure III.5: Proposed covalent inhibition ofphosphoglucomutase by a mixed phosphorofluoridate-phosphate ester substrate analogue.  For the sugar phosphorofluoridate to be covalently linked to the enzyme through the active site serine residue, it must bind to the dephospho form of the enzyme (otherwise the serine is  105  already esterified to a phosphate group). Since glucose monophosphates bind very poorly to the dephosphoenzyme (the Kd of glucose 1-P for the dephosphoenzyme is 0.5 mM, Ray et al., 1966), the  sugar phosphorofluoridate would have first to be phosphorylated at the alternate  acceptor hydroxyl by the phosphoenzyme to produce a mixed phosphate-phosphorofluoridate ester and dephosphoenzyme. If the mixed ester species were to re-bind to the dephosphoenzyme (for which it should have an increased affinity), and presents the phosphorofluoridate functionality to the serine hydroxyl, then attack of the serine hydroxyl on the phosphorofluoridate ester may result in the displacement of fluoride and the covalent attachment of the glucose 1,6-diP to the enzyme, The covalent attack of the serine hydroxyl on the mixed phosphorofluoridate-phosphate ester was considered to be likely since the normal reaction of the enzyme involves attack of the serine hydroxyl on the substrate glucose 1,6-diP. Thus, the sugar phosphorofluoridate would be a member of the class of covalent inhibitors known as suicide substrates since the enzyme must first cause some chemical change in the substrate to produce the inactivating species.  II.A.4: Cyclic-Phosphate Esters as Phosphate Analogues. Glucose l:2-cyclic phosphate has been used as a non-reacting analogue of glucose 1-P in the study of the enzyme glycogen phosphorylase (Withers etal., 1981). This analogue, a phosphate Chester, has a single ionizable proton, the pK of which is probably around a  1-2.  The  cyclization to form the 5-membered phosphate ring produces an analogue in which the conformation of the phosphate is restricted, and of course has an esterified sugar C-2 hydroxyl. However, the conformation of the glucose ring is only slightly flattened compared to that of glucose 1-P, this evidence coming from the analysis of al, 1979). By  the analysis of  ^JH-P  3  JH-H  coupling constants (O'Connor et  coupling constants, the conformation of the cyclic  phosphate was determined to be that of an envelope in which C-l, C-2,0-l and P are planar and 0-2 is above the plane of the ring. Glucose 4;6-cyclic phosphate provides a cyclic phosphate analogue of glucose 6-P. In contrast to glucose l;2-cyclic phosphate no study of its biological activity has been made.  106  In this study the mixed ester glucose 4:6-cyclic phosphate 1-phosphate has been prepared in order to attempt to covalently label the dephospho form of phosphoglucomutase in an analogous manner to that of the mixed phosphorofluoridate-phosphate esters (Figure in. 6). Attack of the active site serine-116 hydroxyl on the 4:6-cyclic phosphate group could result in the cleavage of the C-6 ester linkage and result in a covalently linked glucose 1,6-diP-enzyme complex.  Figure m.6: Proposed covalent inhibition ofphosphoglucomutase by a cyclic phosphate substrate analogue.  m.A.5: Sulfate Esters as Phosphate Analogues. The sulfate monoester analogue of glucose 6-P has been used to study the thermodynamics of the phosphoglucomutase reaction (Peck et a/.,1968) and also as a monoanionic charge state analogue (Ray et al., 1976). A sulfate monoester will have a single negative charge at all but the most acidic pH values due to the extremely high acidity of sulfuric acid. The sulfate dianion is structurally very similar to phosphate, having a tetrahedral structure and similar bond lengths although sulfate has a greater amount of covalent character than phosphate (Corbridge, 1974). It was found that the phosphoenzyme reacted with glucose 6sulfate to produce the dephosphoenzyme and the mixed diester glucose 6-sulfate 1-phosphate. However, this mixed diester did not further react with the dephosphoenzyme to produce the sulfoenzyme and glucose 1-P, nor did the mixed diester bind appreciably to the  107 dephosphoenzyme. The kinetic parameters of the reaction of glucose 6-sulfate with phosphoglucomutase have been evaluated. The K was ctetermined to be 40 mM and the Vmax m  120 s"l. Thus glucose 6-sulfate binds about 0.0015 as well and reacts about 0.4 as fast as glucose 6-P under the conditions of the assay (Ray et al., 1976).  ffl.A.6: Enzyme-Phosphate Electrostatic Interactions in Phosphoglucomutase. X-ray crystal structure analysis of phosphoglucomutase (Lin et al., 1986) has shown the presence of three cationic arginine residues in the active site. The above authors proposed that strong electrostatic interactions between the cations and the anionic phosphate moieties of the reaction intermediate, glucose 1,6-diP, provide the attractive force to hold the diphosphate species in the active site whilst allowing its reorientation without dissociation. Thus the enzyme uses salt bridges to provide the attractive forces for a strong interaction with the substrate phosphate groups. The evaluation of the energy of such an interaction is difficult for a number of reasons. The ions of substrate and enzyme are solvated prior to binding and the energies involved in solvation are high but not readily predictable. Also, the energy of a salt bridge depends on the dielectric constant of the medium which in this case is the interior of a protein. A final factor is that of entropy. Because of the solvation of each charged group, binding will be very favorable due to the release of water of solvation (Fersht, 1985). E—NH (H 0) + (HjCOnPO^-S +  3  2  m  "  „  E—NH  + 3  Po|-S + (HjCtym  +  n )  However, by making a comparison of the ratio of the dissociation constants of the charged substrate and uncharged modified substrate, an estimation of the contribution of the substrate's charge to the binding energy can be made. This approach was taken in Chapter II for the estimation of hydrogen bonding interactions at each sugarringhydroxyl group. The changes in free energy of binding due to the presence of salt bridges in enzymic systems has been measured in much the same ways as was used to determine the energies of hydrogen bonds. Alteration of the substrate to delete a charge has led to a value of 18 kJ/mol for  108 the relative bmding of tyrosine and its deaminated derivative to the enzyme tyrosyl tRNA synthetase (Santi & Pena, 1973). The use of pH studies on the enzyme chymotrypsin gave a value of 12.1 kJ/mol for a buried salt bridge between two buried amino acid side chains (Fersht, 1972). Mutagenesis of the enzyme tyrosyl tRNA synthetase in which charged side chains involved in salt bridges (to charged groups of the substrate) were altered to uncharged analogues, showed a loss of stabilization energy greater than 8.4 kJ/mol (Fersht, 1987a). The ionic strength of the medium will also affect enzyme-ligand interactions. The potential of an ion is lowered in a solution of high ionic strength and therefore an electrostatic interaction with a particular counter ion will be reduced. The ionic strength of the cell cytoplasm is approximately 0.15 M whereas most biochemical measurements are performed in buffers having ionic strengths of approximately 20-50 mM. Thus, enzyme-ligand dissociation constants in vivo  will, in general, be higher than those measured in vitro. The local pH of the enzyme's active site can also have an affect on electrostatic  interactions. The presence of weakly acidic and basic enzymic groups in the active site could result in a local pH rather different to that of the bulk solution. Thus the ionic state of a ligand in the close proximity of the active site may be different from that in the solution.  rn.A.7: Summary and Overview. This introduction has briefly discussed the various means by which the charge state of a substrate (or ligand) bound to an enzyme can be ctetermined. The potential of phosphorofluoridate and cyclic phosphate substrate analogues as covalent inhibitors has also been discussed. This chapter describes the synthesis and use of phosphorofluoridate substrate analogues to determine the ionic state of the (substrate) phosphate group when bound to phosphoglucomutase. The results of the attempts to covalently inhibit the enzyme have been rationalized in terms of the enzymic mechanism of phosphoryl transfer.  109 IJI.B: RESULTS AND DISCUSSION.  in.B.l: Synthesis of Phosphorofluoridate Analogues. A series of three phosphorofluoridate monoester analogues of glucose 1-P and glucose 6P plus a potential irreversible inhibitor based on glucose 4:6-cyclic phosphate was prepared. These novel compounds were characterized by ^H, l^p and 31p-nmr, elemental analysis and fast atom bombardment mass spectrometry. Glucose 6-phosphorofluoridate (glucose 6-PF) was prepared by the reaction of the bistributylammonium salt of glucose 6-P in anhydrous, amine free N,N-dimemylformamide with anhydrous, primary amine-free tributylamine and 2,4-dinitrofluorobenzene (DNFB) in a 1: 3 :2 molar ratio under anhydrous conditions. After an extractive workup to remove excess 2,4dinitrofluorobenzene and tributylamine, the product was purified by chromatography on a DE-52 cellulose anion exchange column followed by repeated lyophilizations to remove the volatile eluting buffer (ammonium bicarbonate). The product was stable for 3-6 months at -20° under anhydrous conditions, but partially decomposed over several weeks at room temperature with exposure to air, turning black and becoming highly acidic. Its 31p-nmr spectrum showed that the major decomposition product was the phosphate monoester with a smaller amount of inorganic phosphate. The decomposition products did not contain any glucose 4:6-cyclic phosphate which would have been formed if decomposition were caused by the intramolecular attack of the C-4 glucose hydroxyl on the phosphorofluoridate group with displacement offluoride.It would appear that decomposition of phosphorofluoridate monoesters occurs by hydrolysis of the P-F bond by water, although a mechanism involving the formation of the 4:6-cyclic phosphate followed by acid hydrolysis cannot be ruled out It was not possible to determine the fate of the glucose moiety by lH-nmr spectroscopy because of signal overlap due to the presence of a and B-anomers. It has been shown (S.G. Withers, unpublished results) that the formation of a-glucose 1phosphorofluoridate (glucose 1-PF) is not possible by the analogous reaction with 2,4dinitrofluorobenzene. Reaction of glucose 1-P with DNFB results in the formation of glucose  110 l:2-cyclic phosphate and no phosphorofluoridate monoester. The formation of the cyclic phosphate could occur by attack of the C-2 hydroxyl on either the 2,4-dinitrophenyl phosphate diester reaction intermediate and subsequent displacement of 2,4-dinitrophenolate, or on the phosphorofluoridate product with displacement of fluoride. If the former is the case, then the reaction of a blocked glucose 1-P derivative with DNFB followed by deprotection should yield the desired phosphorofluoridate product. If the latter is the case, then the desired compound would not be obtainable. This was investigated by the synthesis of 2,3,4,6-tetra-O-acetyl-a-Dglucose 1-P. This compound was prepared by two methods; direct acetylation of a-glucose 1-P with acetic anhydride in pyridine, or phosphorylation of fJ-per-acetyl-glucose by the method of MacDonald (1972) followed by careful neutralization (Warren et al., 1978). The successful synthesis of the acetylated glucose 1-P was evidenced by 31p-nmr spectroscopy. Fluorination of this compound was carried out exactly as in the synthesis of glucose 6-PF. Examination of the crude fluorination product by 31p-nmr spectroscopy revealed a complex mixture of compounds, none of which contained a P-F splitting of approximately 950 Hz. The lack of formation of a phosphorofluoridate monoester is probably due to attack of the C-2 acetate on the 2,4dinitrophenol phosphate diester intermediate to form an acetoxonium ion derivative (Figure m.7).This species could be neutralized by either fluoride or 2,4-dinitrophenolate.  c  O CH —OAcA 2  c  O  —  ^  .  ^  A  c  Q  " . 2CH OAc 2  •NO,  Figure D3.7: Proposed reaction mechanism for formation of an acetoxonium ion during reaction of tetra-O-acetyl-a-D-glucose 1-phosphate and 2,4-dinitrofluorobenzene.  Extensive attempts to prepare and isolate the tetra-O-benzyl blocked glucose 1-P in which no neighboring group participation would be possible were unsuccessful.  111 In order to circumvent the problem of the participation of the C-2 hydroxyl in the fluorination reaction, two other a-glycosyl phosphates were used as phosphorofluoridate substrate analogues. Mannose 1-P is a substrate of phosphoglucomutase albeit with a rather lower affinity (see Section U.B.3.3), but it has the advantage that no participation of the C-2 hydroxyl is possible because of the trans diaxial relationship between it and the phosphate group. Similarly, 2-fluoroglucose 1-P is also a substrate but has a non participating fluorine substituted for hydroxyl. Both the phosphorofluoridate derivatives of mannose 1-P and 2-fluoro-glucose 1-P were synthesized in exactly the same manner as glucose 6-PF but proved to be more stable. As mentioned earlier, the mechanism of fluorination of mymidine 5-phosphate by 2,4dinitrofluorobenzene takes place via a 2,4-dinitrophenyl diester intermediate (Johnson et al., 1975) rather than via the concerted mechanism proposed by Wittman (1962). A study was therefore undertaken to establish the likelihood that the mechanism of the reaction with glycosyl phosphates involves the same intermediate, using mannose 1-P as a model. The postulated intermediate mannose 1-(2,4-dinitrophenyl) phosphate was prepared by the condensation of mannose 1-P and 2,4-dinitrophenol using dicyclohexylcarbodiimide (DCC). The product was purified by anion exchange chromatography, as in the preparation of glucose 6-PF. However the diester was very unstable, decomposing over several hours with release of 2,4-dinitrophenol. 31  P  -nmr of the crude product showed the formation of a phosphate diester as evidenced by an up-field shift of 10.1 ppm from that of the starting material. The crude diester was dissolved in water and a large excess of potassium fluoride was added, the solution turning bright yellow from the release of dinitrophenolate. A 31p-nmr spectrum of the solution after several hours was identical to that of authentic mannose 1-PF. Thus mannose 1-PF can be prepared from the corresponding 2,4-dinitrophenyl diester although the synthesis via this route would not be particularly attractive due to the difficulty in separating the product from the large excess of fluoride ion. These results also imply the intermediacy of the 2,4-dinitrophenyl diester in the fluorination reaction by 2,4-(iinitrofluorobenzene, although a more rigorous description would require a detailed kinetic investigation.  112 II.B.2: Synthesis of Cyclic Phosphate Analogues. The synthetic route for the preparation of glucose 4:6-cyclic phosphate 1-phosphate is shown in Figure III.8.  Figure ITI.8. Synthetic route for the preparation of glucose 4:6-cyclic phosphate 1phosphate.  Glucose 6-P was cyclized by the method of Zmudzka and Shugar (1964) using (ticyclohexylcarbochimide as the coupling reagent Shugar reported that the method produced small amounts of the 3:6-cyclic phosphate but 31p-nmr spectroscopy showed only the presence of two products corresponding to the a and P-anomers of glucose 4:6-cyclic phosphate. To remove any small contaminant of starting material, the product was purified by anion exchange chromatography exactly as in the purification of glucose 6-PF. This step would have removed any traces of dianionic glucose 6-P as the monoanionic product elutes at an ammonium bicarbonate concentration of 30 mM whereas dianions elute at approximately 90 mM. The method of introduction of the 1-phosphate group was that of Hanna and Mendocino (1970) which is a variation of the MacDonald method. The pyridinium salt of glucose 4:6-cyclic phosphate was acetylated at 4° C with acetic anhydride in pyridine prior to fusion with anhydrous  113 phosphoric acid. The saponification of the acetate groups was conducted at pH 11 because of the instability of cyclic phosphate esters to base hydrolysis. The product was purified by anion exchange chromatography as previously described except that the product eluted at an ammonium bicarbonate concentration of approximately 0.12 M.  m.B.3: Biological Activity of Phosphorofluoridate and Cyclic-Phosphate Analogues. Each phosphorofluoridate substrate analogue was tested for irreversible inhibition of phosphoglucomutase. Phosphoglucomutase was incubated at 30° with the inhibitor (5 mM), glucose 1,6-diP (0.1 mM), magnesium chloride (2 mM) and EDTA (1 mM), plus a control in which no inhibitor was included. Aliquots were removed, diluted and assayed for activity over a period of 6 hours. The activity of the phosphoglucomutase after incubation with each inhibitor was identical to that of the control after the same incubation period. Thus no irreversible inhibition was observed. An investigation was carried out to determine whether the phosphoenzyme does in fact phosphorylate each phosphorofluoridate monoester to produce the putative irreversible inhibitor species. The coupled assay (see Section n.B.3.1) was used to measure thereleaseof glucose 6-P from glucose 1,6-diP by the phosphorofluoridate monoester. Glc 1,6-diP + 2FGlc 1-PF  •  Glc 6-P + 2FGlc 1-PF.6-P  Theresultsshowed that even at phosphoglucomutase concentrations of 1 mg/mL and phosphorofluoridate monoester concentrations of 5 mM no detectable production of glucose 6-P took place. Because of the sensitivity of the assay system a phosphorylation rate of approximately 5 x 10"^ ujnol/min/mg would have been detected. This corresponds to a rate of phosphorylation of approximately 10 timesthat of glucose 1-P. -7  In the presence of xylose 1-P, phosphoglucomutase becomes an inefficient glucose 1,6diphosphatase since the water that occupies the vacant C-6 hydroxymethyl site in the phosphoenzyme-xylose 1-P complex can act as an efficient phosphate acceptor. Thus the phosphoenzyme is converted to the dephosphoenzyme which then can be rephosphorylated by glucose 1,6-diPresultingin the production of glucose 6-P (Ray et ah, 1976).  114 E + Xyl 1-P  •  P  E + Glc 1,6-diP D  E + Pi + Xyl 1-P D  •  E + Glc 6-P P  This reaction allowed the viability of the above assay system to be demonstrated by addition of xylose 1-P. The results showed that phosphoglucomutase does not phosphorylate the monoanionic phosphorofluoridate substrate analogues. This is in direct contrast to the results of Ray et al., (1976) who showed that the monoanionic glucose 6-P analogue glucose 6-sulfate was phosphorylated at almost one half of the rate at which the enzyme reacts with glucose 6-P. The difference in reactivity must be related to differences between the sulfate and phosphorofluoridate monoester groups. There is a great similarity in both structure and size of the two groups as noted in the introductory section and several phosphorofluoridate salts have solubilities and structures very similar to those of the corresrxmcling sulfates (Corbridge, 1985). The properties of phosphate and sulfate anions are summarized in Table m.2.  Sulfate (SO4 -)  Phosphate (PO4 -)  2/6  2/5  Bond length (nm)  0.149  0.154  Sum of covalent radii (nm)  0.170  0.176  Corrected sum (nm)  0.161  0.163  Ionic character of bond  22%  39%  -3  2  Property  2  Electrostatic bond strength  Acid strength pK l a  3  Table III.2: The properties of sulfate and phosphate ions. (Corbridge, 1985).  The increased covalency of sulfate compared to phosphate is due to the decreased difference in electronegativity between oxygen and the central atom and is reflected in the difference in acid strengths of each compound. Similar information on phosphorofluoridate is not available. The differences in ionic character of sulfate and phosphate are reflected by differences  115 in their affinity for different metal ions. Thus phosphate forms insoluble aluminum salts whereas aluminum sulfate is soluble. Differences exist between the biological function of phosphate and sulfate ions. For example, alkaline phosphatase is inhibited by phosphate with a Ki of 1.5 mM, whereas sulfate does not observably inhibit the enzyme (Georgatsos, 1967). One would expect that the differences in the ionic and coordination chemistry of sulfate and phosphate would produce a sulfate analogue of glucose 6-P that is less active than a phosphorofluoridate analogue. The measured dissociation constants of the two types of phosphate analogues do reflect this as it will be shown that glucose 6-PF binds moretightlyto the phosphoenzyme than does glucose 6sulfate. However, it is difficult to rationalize why the phosphorofluoridate analogues bind better than the sulfate analogue, the structure and conformation of which would be expected to be very similar, but do not undergo phosphorylation. Glucose 4:6-cyclic phosphate was phosphorylated chemically to give the diphosphorylated sugar. This avoided the problem encountered with the phosphorofluoridate substrate analogues where it appeared that the putative active inhibiting species was not formed. The synthesis of a mixed phosphorofluoridate-phosphate ester would be extremely difficult, if possible at all. The mixed cyclic phosphate-phosphate ester was tested as an irreversible inhibitor of phosphoglucomutase by the same method used for the phosphorofluoridate monoesters. However, the enzyme was dephosphorylated (Ma & Ray, 1980) prior to incubation with the inhibitor and no glucose 1,6-diP was included in the solution. Notime-dependentloss of activity was observed over several hours using an inhibitor concentration of 3.5 mM. An estimate of the reversible inhibition constant (Ki) for this glucose 1,6-diP analogue was obtained by measuring the rate of reaction using the coupled assay at a series of inhibitor concentrations (0-25 mM) with a constant concentration of glucose 1-P and a subsaturating glucose 1,6-diP concentration. Plotting the results according to Dixon (Dixon, 1953) allowed an estimation of Ki of approximately 65 jiM to be obtained. As well as competitively inhibiting phosphoglucomutase, this mixed ester can also phosphorylate the dephosphoenzyme. This was shown by the observation that the enzyme had no activity in the absence of glucose 1,6-diP but activity  116 appeared slowly after addition of glucose 4:6-cyclic phosphate 1-phosphate. The possibility of the activity being due to a slight contamination with glucose 1,6-diP cannot be completely ruled out, but is unlikely as the product and its precursor were purified by anion exchange chromatography. Hence glucose 4:6-cyclic phosphate 1-phosphate is both an inhibitor with respect to glucose 1,6-diP and an activator of the enzyme. A similar effect on phosphoglucomutase has been previously observed with 1,3-diphosphoglyceric acid (Alpers, 1967, Alpers & Lam, 1968) and so the present system was not investigated to any further extent. The inhibition constant (Ki) of each of the three phosphorofluoridate charge state analogues was determined in the same manner as for the disubstituted glucose 1-P analogues (Section H.B.3.4) using the coupled assay and computer-generated values of apparent K . In m  the cases where only Dixon plots were obtained, the K and Vmax of glucose 1-P were m  determined prior to assaying the inhibitor. The line of best fit was determined by least squares linear regression analysis. Obtaining precise data was extremely difficult for inhibition of phosphoglucomutase by glucose 6-PF as the inhibitor is a substrate of the coupling enzyme glucose 6-phosphate dehydrogenase. Thus measurement of the rate involved subtracting the rate of the reaction before addition of phosphoglucomutase to the assay solution from that after its addition. Therefore the error in these rates is double that of the other phosphorofluoridate inhibitors. Because of the inaccuracy of the assay with glucose 6-PF, a full (ietermination of Ki was not attempted. A full extermination of Ki for 2-fluoro-glucose 1-PF was not carried out due to a shortage of material. Inhibition by mannose 1-PF produced double reciprocal plots indicative of competitive inhibition. It is assumed that glucose 6-PF and 2-fluoro-glucose 1-PF inhibit the enzyme in the same manner as mannose 1-PF does. The graphical results of inhibition by the phosphorofluoridate analogues are presented in the Appendix. The values of Ki for each inhibitor and the calculated loss of binding energy compared to that of the parent substrate are presented in Table m.3. In this case the dissociation constant (Kd) of the parent substrate was taken as being equal to its K although this is not necessarily so. The dissociation constants of glucose 1- and m  6-P from the phosphoenzyme have in fact been determined as approximately the same values as that of their K (Ray & Long, 1976b). m  117 Inhibitor  Ki(mM)  AAG° (kJ/mol)  Glucose 6-phosphorofluoridate  7  12.6b  Mannose 1-phosphorofluoridate  19  11.0 c  2-Fluoro-glucose 1phosphorofluoridate  17  10.9d  a  Values calculatedfromRT ln(Ki/K ), where T = 303 K, R = 8.314 J/K/mol and K is the Ki of the phosphorofluoridate substrate analogue and Ki is the K of the corresponding substrate. b K glucose 6-P = 47 |J.M. DatafromLowry & Passonneau (1969). K mannose 1-P = 245 |i.M. Data from Lowry & Passonneau (1969). K 2-fluoro-glucose 1-P = 222 PM. a  2  2  m  m  c  m  d  m  Table m.3. Dissociation constants ofphosphorofluoridate inhibitors of phosphoglucomutase.  The results of the binding studies of phosphoglucomutase with each phosphorofluoridate substrate analogue show that a loss of enzyme-ligand bindingfreeenergy (AAG°) of approximately 11 kJ/mol occurs on the deletion of a negative chargefromthe substrate's phosphate functional group. This value compares favorably with the values of previous studies for salt bridge interactions noted in the introduction. The results also show the importance of the doubly negatively charged phosphate group on catalysis since no substrate activity of any singly charged substrate analogue was detected. However, such diverse acceptors as methanol, 1,4butanediol phosphate and glucose have been found to be phosphorylated (albeit at extremely slow rates) by the phosphoenzyme (Ray et al., 1976). To measure these extremely slow rates of phosphoryl transfer the authors used a very sensitive radioactive 32p label assay. On this pretext phosphoglucomutase would also be expected to transfer its phosphate to the monoanionic substrate analogues. It seems however that the rate of transfer is too low to be detected by the assay system used in this study. This lack of substrate activity of the monoanionic substrate analogues provides a noteworthy example of how substrate binding energy is inexorably linked with catalytic events. This is particularly impressive in this case as the substrate's phosphate group, which provides much of the overall binding energy, is quite far removed from the active site where actual phosphoryl transfer occurs.  118 The lack of observable irreversible inhibition afforded by the phosphorofluoridate monoesters is not surprising in light of the absence of formation of the mixed phosphatephosphorofluoridate ester, the putative inhibiting species, by reaction with the phosphoenzyme. The phosphorofluoridate monoesters themselves probably bind even more weakly to the dephosphoenzyme than do glucose monophosphates. On the basis that the phosphate binding site at which phosphoryl transfer occurs is the weaker of the two (Ray et al., 1973), it would be unlikely that the phosphorofluoridate monoesters would bind to the dephosphoenzyme in the correct orientation for attack of the serine hydroxyl on the phosphorofluoridate to occur. This argument presupposes that an exchange type of mechanism does operate. In addition, the justification for the non-reactivity of phosphorofluoridate substrate analogues in other systems (outlined in the introduction) may have a bearing on the lack of covalent inhibition observed in this study. The irreversible inhibition of phosphoglucomutase by glucose 4:6-cyclic phosphate 1phosphate would necessitate the 4:6-cyclic phosphate group taking up the same conformation in the active site as that of the C-6 phosphate of glucose 1,6-diP. The cyclic phosphate group, being part of a trans-fused ring system, has a rigid chair conformation. Its conformation relative to its vicinal hydrogens is shown in Figure m.9. The dihedral bond angle between the phosphorus and one of the C-6 protons is close to 180° and this results in a large spin-spin coupling of approximately 20 Hz between the two nuclei. H  H  4  6  P  C  5  »6  Figure 111.9: Newman projection of the conformation of the 4:6-cyclic phosphate group of glucose 4:6 cyclic phosphate about the C-O bonds.  119 An analysis of the Jp-H coupling constants of glucose 6-P shows a single 6 Hz coupling to the two C-6 protons (cf., O'Connor et al., 1979). Newman projections along the C-O bond (Figure DXIO) show that a single preferred conformer would account for the small observed three bond coupling constant.  II V H 6  =  2  4  H  Z  V H 6 =  III 2  4  H  z  3  J -H6 = P  2  -  4  H  Z  Figure DXIO: Newman projection of the conformation of the phosphate group of glucose 6-phosphate about the C-O bond. From O'Connor et al. (1979).  Thus the conformer with the largest population, DI, is that in which the C-5-C-6 and O-P bonds have a trans antiplanar arrangement. The phosphate group has a second degree of freedom due to rotation about the C-5-C-6 bond. The absence of a Jp_H5 coupling (such as is observed 3  in the spectrum of a-glucose 1-P between H-2 and P) suggests that the rotamer in which the atoms between the C-5 proton and phosphorus has a trans antiplanar arrangement is a minor constituent. A comparison of the preferred conformation of glucose 6-P with that of glucose 4:6cyclic P shows a difference in the C-6-O bond angle of 120°. Thus the phosphate of glucose 6-P would rarely take up a conformation similar to that of the cyclic phosphate. This difference in phosphate conformation can rationalize the unreactivity of the 4:6-cyclic phosphate species. Since all enzymic phosphoryl transfer reactions have so far been shown to occur by an in-line associative mechanism, the serine hydroxyl would be unable to attack the constrained cyclic phosphate in such a way as to cause the cleavage of the C-4 or C-6 ester linkage. This analysis  120 presupposes that the conformation in which the 6-phosphate of glucose 6-P is bound in the active site is the same as its preferred solution conformation and is not distorted to any great extent. If the difference in energies between the solution conformation and a different enzyme-bound conformation is small compared to the overall binding energy of the substrate, then the above argument would have little validity. Alkaline hydrolysis of some cyclic phosphates in non-enzymic systems has been found to involve adjacent attack followed by a pseudorotation resulting in a product whose configuration is retained (Hall & Inch, 1980). However, all enzymic systems which involve nucleophilic attack on cyclic phosphate centers have been shown to proceed with inversion of configuration, thereby ruling out the possibility of an adjacent-pseudorotation mechanism (Knowles, 1980). The likelihood of an enzyme-bound substrate undergoing a pseudorotation would probably be remote since each phosphate oxygen and the remainder of the substrate would be involved in interactions with amino acid side chains and this would hamper any reorganization of the phosphorus ligands. This would appear to rule out the possibility of a mechanism involving the attack of the serine hydroxyl on the cyclic phosphate in an adjacent fashion, followed by a pseudorotation and cleavage of the C-4 or C-6 ester linkage.  121 m.C: EXPERIMENTAL PROCEDURES. IJI.C.1: General Methods and Purification of Reagents. Solvents and reagents were purified as follows. N,N-dimethylfonnarnide was kept for several days over anhydrous magnesium sulfate, vacuum distilled at low temperature and stored over 4 A molecular sieves under an atmosphere of dry nitrogen. Tributylamine was shaken with p-toluenesulfonyl chloride (10% w/w) in aqueous 15% sodium hydroxide at 4° C. After one day at room temperature the amine was washed with dilute aqueous sodium hydroxide solution, then water, and dried overnight over potassium hydroxide pellets. The amine was then distilled from barium oxide under nitrogen and stored at 4°. 2,4-Dinitrophenol was crystallized from ethanol and stored over anhydrous calcium sulfate in a desiccator. Glucose 6-phosphate, 2,4dinitrofluorobenzene and dicyclohexylcarbodiimide were purchased from Sigma Chemical Co. and used without purification. Thin layer chromatography of phosphate and phosphorofluoridate esters was carried out using cellulose PEI-F plates (J. T. Baker Co.). The plates were developed with an aqueous solution of Uthium chloride. For mono and dianions the concentration of hthium chloride used was 1 M, however higher charged species required a more concentrated eluting solution. The compounds were visualized by the use of a dipping solution specific for phosphates (Burrows et al, 1952) followed by development with UV light. The color of phosphorofluoridate-containing species developed more slowly than that of phosphates. The concentrations of phosphorofluoridate ester and cyclic-phosphate solutions were determined as described for the phosphate esters in Chapter JJ. However in the case of C-l linked esters the hydrolysis period was extended to 30 minutes to ensure the complete hydrolysis of the P-F bond. Other procedures and reagent purifications are described in the experimental procedures section of Chapter U.  122 in.C.2: Synthetic Methods.  D-Glucose 6-[ammonium phosphorofluoridate], D-glucose 6-[disodium phosphate] (1.0 g, 2.8 mmol) was dissolved in water and passed through a column (4° C) of Dowex 50W-X8 (H+) resin into an excess of tributylamine. The solution was evaporated and then further dried in vacuo . The gum was dissolved in dry, distilled N,N-dimethylformamide (10 mL) containing distilled tributylamine (2.0 mL, 8.4 mmol) and 2,4dinitrofluorobenzene (0.60 mL, 4.8 mmol). The mixture was protected from moisture with a calcium chloride guard tube and stirred for 24 hours. After this time diethyl ether was added to cloud point and the product precipitated with cyclohexylamine (1.1 mL, 10 mmol). The solvent was decanted off and the gum was washed thrice with diethyl ether. The precipitate was dissolved in water and extracted with diethyl ether fourtimesbefore being lyophilized to yield a gummy yellow solid. The solid was dissolved in a minimum volume of methanol and precipitated by addition of diethyl ether. This procedure was repeated thrice, or until most of the color was removed from the precipitate. The precipitate was then dissolved in water (500 mL), the pH adjusted to 8.0 and the solution applied to a column (1.8 x 30 cm) of DE-52 cellulose at 4° C which had been previously equilibrated with 10 mM ammonium bicarbonate. After washing the column with water (200 mL), the products were eluted with a linear gradient of 0-50 mM ammonium bicarbonate in a total of 2 L. Fractions (20 mL) were collected at aflowrate of 2 mL/min. The fractions containing the desired product, eluting at a salt concentration of approximately 30 mM, were identified by a "spot" test for reducing sugars (Trevelyan et ai, 1950). After pooling the fractions, the buffer was removed by repeated lyophilization to yield a pale yeUow foam (0.41 g, 1.5 mmol, 53%); P-nmr data (121 MHz, D2O): 8 2.93 (dt, Jp,F 31  932, ]p 6+6' 6.0 Hz), 2.98 (dt, Jp,F 932, Jp,6+6' 6.0 Hz); 19F-nmr data (254 MHz, D2O): 8 t  78.63 (d, J F , P 931 Hz), 78.72 (d, J F , P 931 Hz); FAB mass spectrum (M + glycerol) = 371. Anal. Calc. for C 6 H 1 5 O 8 N P F : C, 25.82; H, 5.42; N, 5.02. Found: C, 25.58; H, 5.55; N, 5.61.  123 q-D-Mannopyranosyl [bis (cyclohexylarnmonium)] phosphate. 1,2,3,4,6-penta-O-acetyl-q-D-mannopyranose (Wolfram & Thompson, 1963a) (18.0 g, 46 mmol) was phosphorylated according to the method of MacDonald (1972) and the biscyclohexylammonium salt was crystallized from water, acetone to yield colorless crystals (10.6 g, 23 mmol, 50%); iH-nmr data (400 MHz, D2O): 8 5.31 (dd, 1 H, Ji,p 8.6, Jl,2 1.6 Hz, H1), 3.95-3.66 (m, 5 H, H-2,3,5,6,6'), 3.58 (dd, 1 H, 14,5 9.5, J 4 3 9.5, H-4), 3.13 (m, 2 H, 2 NH3+-CH), 2.00-1.18 (m, 20 H, 2 cyclohexyl); 31p-nmr data (121 MHz, D2O): 8 -4.70 (d, Jp,l 8.5 Hz).  q-D-Mannopyranosyl [tributylammonium] 2,4-dinitrophenyl phosphate. A solution of oc-D-mannopyranosyl [bis (cyclohexylammonium)] phosphate (1.4 g, 2.2 mmol) was converted to the bis-tributylammonium salt as described in the synthesis of glucose 6-phosphorofluoridate. The dried gum was dissolved in dry, distilled N,N-dimethylformamide (25 mL) containing 2,4-dinitrophenol (4.05 g, 22 mmol) and dicyclohexylcarbodiimide (9.06 g, 44 mmol). After stirring for 2 hours under anhydrous conditions the solvent was removed, the residue suspended in water and the mixture extracted four times with diethyl ether. After filtration the solution was lyophilized. 31p. mr data (32.4 MHz, D2O): 8 6.12 (d, Jp,i 6.8 Hz). n  q-D-Mannopyranosyl [ammonium phosphorofluoridate]. q-D-Mannopyranosyl [bis(cyclohexylammonium) phosphate] (1.5 g, 3.2 mmol) was converted to the bis-tributylammonium salt and was treated, appropriately scaled, exactly as in the synthesis of glucose 6-phosphorofluoridate. The product was purified by DE-52 cellulose column chromatography, the fractions containing the desired products were identified by colorimetric assay for acid labile phosphate (Peck & Ray, 1971). Repeated lyophilizations yielded a white foam (0.49 g, 1.76 mmol, 55%); iH-nmr data (400 MHz, D2O): 8 5.51 (dd, 1 H, Jp,i 6.8, Ji,2 1-8 Hz, H-l), 4.02 (dd, 1 H, Ji,2 2.0, J 2 3 2.0 Hz, H-2), 3.92-3.74 (m, 4 H, H-3,5,6,6'), 3.70 (dd, 1 H, J 4 3 9.5, J4,5 9.5 Hz, H-4); 31p-nmr data (121 MHz, D2O): 8  124 5.79 (dd, Jp,F 940, Jp,i 6.7 Hz); Wp-nmr data (254 MHz, D 0): 8 76.00 (d, J F , P 939 Hz); 2  FAB mass spectrum (M+l) = 279, (M + glycerol) = 371. Anal. Calc. for C 6 H 1 5 O 8 N P F : C, 25.82; H, 5.42; N, 5.02. Found: C, 25.64; H, 5.70;  N, 5.30.  2-Deoxy-2-fluoro-a-D-glucopyranosyi [ammonium phosphorofluoridate]. 2-Deoxy-2-fluoro-a-D-glucopyranosyl [bis(cyclohexylammonium) phosphate] (0.205 g, 0.44 mmol, 20% B-anomer) was treated, appropriately scaled, exactly as in the synthesis of glucose 6-phosphorofluoridate. DE-52 column chromatography yielded a white foam (0.088 g, 0.30 mmol, 68%), but did not separate the a and B anomers. oc-anomer; ^H-nmr data (400 MHz, D 0): 8 5.75 (dd, 1 H, Ji,p 6.8, J l , 3.6 Hz, H-l), 4.46 (dddd, 1 H, 2  2  J  2  J  F  48.5, J ,3 2  9.5, J , l 3.2, J2,P 3.2 Hz, H-2), 3.98 (ddd, 1 H, J3,F 13.1, J3,2 9.4, J3,4 9.4 Hz, H-3), 2  3.90-3.75 (m, 3 H, H-5,6,6'), 3.51 (dd, 1 H, J 4 3 9.5, J4,5 9.5 Hz, H-4); P-nmr data (121 31  MHz, D2O): 8 5.41 (ddd, Jp,F 940, Jp,l 6.6, Jp,2 2.7 Hz); 19F-nmr data (254 MHz, D2O): 8 66.77 (d, J ,p 942 Hz), 192.86 (dd, JF,2 49, JF,3 13 Hz), B-anomer; iH-nmr data (400 MHz, F  D2O):  8 5.25 (ddd, 1 H, Ji,p 7.5, J i , 7.5, Ji,F 2.8 Hz, H-l), 4.22 (ddd, 1 H, J ,F 51.0, 2  2  J2,l 7.5, J2.3 7.5 Hz, H-2), 3.90-3.75 (m, 4 H, H-3,5,6,6'), 3.48 (dd, 1 H, J 4 3 9.5, J4,5 9.5 Hz, H-4); 31p- mr data (121 MHz, D2O): 8 5.83 (dd, Jp,F 942, Jp,i 7.6 Hz); 19F-nmr n  data (254 MHz, D2O): 8 66.96 (d, J F , P 935 Hz), 192.37 (dd, JF,2 51, JF,3 15 Hz); FAB mass spectrum (M+l) = 282. Anal.  Calc. for C 6 H 1 4 O 7 N P F 2 : C, 25.63; H, 5.02; N, 4.98. Found: C, 25.19; H, 5.17;  N, 5.01.  D-Glucose [ammonium 4:6-cyclic phosphate]. D-glucose 6-(dihydrogen phosphate) (0.86 g, 3.3 mmol) was cyclized according to the method of Zmudzka & Shugar (1964) and the product chromatographed exactly as in the synthesis of glucose 6-phosphorofluoridate. Repeated lyophilizations yielded a colorless foam  125 (0.72 g, 2.8 mmol, 85%); P-nmr data (121 MHz, D 2 O , { ! H } ) : 8 -5.05, -5.08 (a and p31  anomers).  q-D-Glucopyranosyl 4:6-cyclic phosphate [tris (ammonium) phosphate]. A solution of D-glucose [ammonium 4:6-cyclic phosphate] (0.70 g, 2.73 mmol) was converted to the pyridinium salt via a Dowex 50W-X8 (H ) resin column and the water +  evaporated. The gum was further dried in vacuo and anhydrous pyridine added and evaporated thrice in order to completely dry the gum. Anhydrous pyridine (6 mL) and acetic anhydride (2 mL) were added and the solution stirred for 2 hours to effect dissolution. The solution was kept for 2 days at 4" and the solvent was then removed. The gum was further dried in vacuo . Anhydrous phosphoric acid (2.0 g, 20 mmol) was added and the mixture was heated under vacuum at 55° for 2 hours. Ice cold 1 M lithium hydroxide (60 mL) was added and the pH adjusted with further base to 11.0. The solution was left at room temperature for 2 days and the pH was occasionally readjusted to 11.0. After filtration through celite and adjustment of the pH to 8.0 with Dowex 50W-X8 (H ) resin, the solution was applied to a column (1.8 x 30 cm, 4°) +  of DE-52 cellulose which had been previously equilibrated with 40 mM ammonium bicarbonate. The column was developed exactly as in the preparation of glucose 6-phosphorofluoridate, however the salt gradient was 0-0.25 M ammonium bicarbonate. The desired product eluted at approximately 0.13 M salt and was identified by colorimetric assay for acid labile phosphate (Peck & Ray, 1971). After pooling the fractions, the buffer was removed by multiple lyophilizations to yield a white foam (0.32 g, 0.86 mmol, 31%); iH-nmr data (300 MHz, D2O): 8 5.42 (dd, 1 H, Ji,p 7.0, Ji,2 2.5 Hz, H-l), 4.15-3.90 (m, 2 H), 3.72 (dd, 1 H,J3,4 9.0, J3,2 9.0 Hz, H-3), 3.60-3.45 (m, 3 H); P-nmr data (121 MHz, D2O): 8 -4.12 (dd, Jp.l 7.2, 31  Jp,2 1.5 Hz), -4.5 l(m).  126  CHAPTER IV: NMR STUDIES OF ENZYME-LIGAND COMPLEXES.  IV.A: INTRODUCTION.  IV.A.l: 19p-NMR in Macromolecular Systems. In the past few years, the use of lH-nmr as a tool for determining the structural characteristics of smaller proteins has increased dramatically. High field spectrometers (corresponding to a proton resonance frequency of 400 MHz and higher) and the advent of 2Dnmr techniques have allowed the amino acid sequence and 3-dimensional structures of many small proteins to be determined (e.g., Wuthrich, 1986). However in diamagnetic proteins, having a relative molecular weight of above about 20000, the *H spectrum, even at high field, generally consists of a broad envelope between 0 and 7 ppm. On the low field side of that envelope, the histidine Ce-H resonances are often well resolved and on the high field side, methyl resonances which have been shifted by ring currents are often visible (Jardetzky & Roberts, 1981). The functional significance of histidine residues and the ease with which they are resolved has lead to a large number of studies of this spectral region (e.g., Rhyu et al., 1985b). However, apart from these specific residues, the large majority of protein resonances are usually a set of ill defined peaksfromwhich the extraction of any physical characteristics is difficult. Alternate approaches which overcome the relatively poor resolution inherent in lH-nmr have been utilized in order to obtain structural information on proteins and protein-ligand complexes. Other nuclei found in proteins (having a non-zero nuclear spin) have advantages that make them suitable nmr probes. The nuclei 1 C, I ^ N and 31p all have large chemical shift 3  ranges and have the advantage of being less numerous than protons. However as shown in Table IV. 1 their abundance and insensitivity as nmr nuclei make them difficult to study. Enrichment of both protein and ligand with 13c, H and 2  make the study of these nuclei attractive;  however, the time and expense involved have reduced their usefulness. Phosphorylated proteins and substrates are common in enzymic systems and 31p-nmr has been used extensively as a  127  probe of enzyme mechanisms, especially in the last few years with the advent of high field, high sensitivity instruments to offset the low sensitivity of the 31p nucleus (e.g., Gorenstein, 1984).  Nucleus  Relative sensitivity  Relative sensitivity  Chemical shift range (ppm)  lH  1.0000  1.000  10  2  0.0096  2x10-°  10  13  0.0159  2xl0-  340  15  0.0010  3 x lO"  620  19  0.8330  0.833  960  31p  0.0663  0.0663  700  H  C  N  F  a  0  a  0  4  6  at constant number of nuclei. at natural isotopic abundance of each nuclide.  Table IV. 1: The properties of some nuclei. (Data from Gerig, 1978.) Another alternative is the use of 19F, a nucleus which has many advantages for use as an nmr probe. The sensitivity of 19p is 83% that of H and the chemical shift range of 19F is l  almost 1000 ppm. Thus 19p chemical shifts are much more sensitive than H shifts to changes in l  the environment of the nucleus. Consequently signals from 19p win b highly dispersed. The e  19F isotope is present at 100% natural abundance and so no enrichment procedures are necessary for its incorporation in the protein or ligand. Furthermore, 19F is extremely rarely present naturally in biological systems and so the only resonances observed will be those introduced by the investigator. As noted in Table II. 1, 19F is a structurally conservative substitution for hydroxyl and hence any secondary effects due to steric interactions can probably be neglected. Fluorine has been introduced into macromolecules by two general methods. The first involves the covalent modification of the protein by afluorine-containingreporter molecule. Fluorine can be introduced into the protein either by incorporation offluorinatedamino acids into the diet of the animal or bacteria producing the protein (Sykes & Weiner, 1980), or by the covalent modification of the isolated native protein with afluorine-containingreagent. The  128  protein can be specifically labelled with afluorine-containingsuicide substrate, a fluorinated cofactor (Chang et al., 1987) or a site directed covalent inhibitor such as one of the aryl sulfonyl fluorides which modify the active site serine residue of chymotrypsin (Ando et al., 1986). Alternatively it can be non-specifically labelled by reagents such as S-ethylthiotrifluoroacetate which trifluoroacetylates amino groups under mild conditions (Goldberger, 1967). The second general method offluorineintroduction into a protein is by the formation of a reversible complex with afluorine-containingligand. This is the method used in this study. From both types of experiments, various parameters characteristic of the fluorinated probe such as chemical shift, spin-spin coupling, spin lattice (Tl) and spin-spin (T2) relaxation times and nuclear Overhauser effects (noe) can potentially be determined. These can potentially give information on the environment, conformation and mobility of the probe, as well as the rates of any exchange processes.  IV.A.2:19p-NMR Chemical Shifts in Macromolecular Systems. As noted above, the range of l^F chemical shifts approaches 1000 ppm although the range in biologically important molecules will be somewhat less than this. Thus the chemical shift of a l^F nucleus is extremely sensitive to the effects of the environment in which it is placed. The chemical shift of a nucleus relative to that of a reference resonance is determined by the shielding constant, a, which relates the applied magnetic field, Bo, to the effective field at the nucleus, B ff. e  B ff = Bo.(l-o) e  The shielding constant is itself the sum of two terms, a = ad + Op where ad is the diamagnetic term and ap is the paramagnetic term. The term ad is the shielding due to the magnetic field induced by the electrons of the isolated atom when placed in an external magnetic field. However, in a molecule, the electrons are not freely circulating and their motion is anisotropic. This effect tends to reduce the local magnetic field at the nucleus, the value of Op  129 depending on the symmetry of the electrons' orbitals. For example, electrons in (symmetric) Is orbitals make a very small op contribution, whereas the distribution of p electrons in a covalendy bonded * F atom is unsymmetrical resulting in a large Op contribution. Consequently the range 9  of *H chemical shifts is much less than that of shift of a nucleus  (i.e.,  (Dwek, 1973). Thus the observed chemical  the shielding) is determined by the chemical nature of the bond which  links it to the remainder of the structure. The paramagnetic term will be altered by any changes in, 1) substituents elsewhere in the molecule, 2) conformation, and 3) solvent (i.e., any intermolecular effects). No change in the chemical shift of a ligand on binding to a macromolecule would be expected to arise from 1 or 2, assuming that there is no serious deformation of bond lengths and angles on complex formation. Thus solvent effects remain to rationalize any macromolecule induced fluorine chemical shift effects. Solvent effects have contributions from four independent sources. a = Oa + Ow + o"e + o"c The term c»a arises from the magnetic anisotropy of nearby groups such as aromatic and carbonyl functionalities.This effect can produce downfield or upfield shifts depending on their spatial relationship with the reporter nucleus, varying as l/r , but the effects on 19p nuclei are expected 3  to be similar in magnitude to those on ^H. Consequently the largest change in chemical shift caused by an aromatic amino acid will be approximately 2 ppm and about 0.3 ppm from a carbonyl functionality (Webb, 1986; Campbell et al., 1985). Effects on the nucleus n which arise from Van der Waals interactions (with nuclei i), aw, are expected to be described by the following equation T~, v (3.aj.Ij) o = BL — w  where B is a constant, and the terms within the summation depend on the polarizability, ai, and the ionization potential, Ii, of the atoms near the nucleus n as well as the distance, r, between n and i (Gamcsik et al., 1987). The l/r° dependance renders any shielding contribution negligible from nuclei at a distance greater than about 3.3 A although at small distances, r < 2.5 A, a  130  downfield shift of greater than 10 ppm has been predicted (Kimber et al., 1977). A correlation has been observed between chemical shift changes and internuclear distances in fluoro tyrosine alkaline phosphatase (Hull & Sykes, 1976). These authors noted a direct relationship between the chemical shift of each resonance and the internuclear F-H distances which were determined from dipolar relaxation measurements. The changes in chemical shift were proportional to El/r^ thereby suggesting the importance of Van der Waals interactions in the origin of deshielding effects. The electric field gradient contribution, o*e, is due to the presence of charged or polar groups such as P04 and OH in the proximity of the nucleus. Both Van der Waals and static 2-  electric field effects are expected to give rise to deshielding (Kimber et al., 1977). Effects caused by hydrogen bonding to fluorine, a , are expected to result in deshielding c  although attempts to show the effect of hydrogen bonding on l^F chemical shifts in small molecules were unsuccessful (Muller, 1976). However, the solvent isotope shift is believed to be due to hydrogen bonding effects. This effect is demonstrated by the shift to lower field of resonances arising from solvent-exposed fluorines on changing the solvent from H 2 O to D2O. The effect is, however, small, solvent isotope shifts generally being in the range up to 0.2 ppm (Hull & Sykes, 1976). Chemical shift changes induced by the binding of a fluorinated ligand to a protein or incorporation of a fluorinated compound into a protein generally range from -8 to +4 ppm, although the majority of shifts are only up to several ppm downfield (Gerig, 1978,1982). These downfield shifts have been interpreted as a change to a more hydrophobic or buried environment since surface probes generally appear to be affected to a lesser degree (Sykes & Weiner, 1980). This interpretation is supported by the fact that chemical shifts of fluorinated compounds shift downfield on changing to a less polar solvent (Robertson et al., 1977).  IV.A.3: Relaxation in Macromolecular Systems. In a solution nmr experiment on small molecules (molecular weight less than several thousand), the rates of reorientation of the nuclei are very rapid resulting in the equalization of the  131 spin-spin (T2) and spin lattice (Tl) relaxationtimes.However, in the cases of macromolecules, which have much larger correlationtimes,XQ,  (i.e.,  slower rate of reorientation), the values of Tl  and T2 become rather different. Figure IV. 1 shows the relationship between XQ and Tl and T2. The effect generally observed on incorporation of a nucleus into a macromolecule is a decrease in Tl and increase in T2, resulting in a significant broadening of the signal from the bound nucleus.  overall (sec)  T  Figure IV. 1: Graph of the relationship between correlation time (T) and spin-lattice (Tl) and spin-spin (T2) relaxation times. The relationships are shown at two spectrometer frequencies. Graph from Roberts and Jardetzky (1981). Relaxation of the nuclear spins results from the action of a local magnetic field oscillating at the same frequency as the Larmor frequency of the nucleus. The relaxation mechanisms commonly encountered in l^F-nmr of diamagnetic macromolecular systems are dipole-dipole interactions between like or unlike nuclei, such as 19p_19p  19p.ljj interactions and  chemical shift anisotropy (CS A). If chemical shielding is anisotropic (i.e., 0"xx * Oyy * o"zz) then the local fields due to the motion of these electrons will change as the molecule rotates. This fluctuating magnetic field can provide a relaxation mechanism for nuclei with a large chemical shift anisotropy such as  and ^P. At low magnetic fields Gess than 100 MHz) relaxation is 3  132  governed by I ^ J J - I H dipolar interactions. However, since relaxation due to CSA increases as the square of Bo, the effect of higher spectrometer fields is to broaden lines and reduce resolution. It has been shown that the same resolution of the l^F signals of fluorotyrosine alkaline phosphatase (relative molecular weight 80 000) can be obtained from spectrometers operating at 94 and 254 MHz (Hull & Sykes, 1975a).  IY.A.4: Nuclear Overhauser Effects in Macromolecular Systems. The nuclear Overhauser enhancement (noe) is the fractional change in intensity of one nmr resonance when another is irradiated. The noe is due to cross relaxation, which is the mutual effect of two spins on each others relaxation.Since relaxation is dependent on the spectrometer frequency (coo, in rad s"*) and the correlationtimeof the molecule (Te), so is the noe. For small molecules in which (tccoo) « 1 (the extreme narrowing limit), the limiting noe factor, (T|i{j}), 2  for spin i upon irradiation of spin j is (Wuthrich, 1986), Tli(j}=^2yi  The intensity of the signal that is observed is equal to 1 + T|i{j}. In a fluorine-proton system, in which the gyromagnetic ratios (y) of the two nuclei are almost the same, the maximal noe obtained by irradiation of either proton or fluorine will be approximately 0.5. In this case (rapid motion), the transitions between the energy levels of spin i that are induced by saturation of spin j result in a decrease in the population of the upper energy level and hence an increase in signal intensity. However in the case where (tcCOo) » 1 ('•£•, slow motion, the non-extreme 2  narrowing limit) the previous transition becomes unimportant and a second transition which causes the equalization of the populations of the upper and lower energy levels ctorriinates. Upon equalization of the two energy levels absorption ceases and the signal will disappear. The probability of cross relaxation is dependent on the distance (r) between the two nuclei, varying as l/A Therefore an noe measurement can establish proximity although a distance calculation from a single measured noe is not possible. Generally the effects of an noe can be detected up to a distance of approximately 5 A.  133  The presence of an observed noe in macromolecular systems does not always prove that the irradiated nucleus is in close proximity to the observed nucleus. This is so because of spin diffusion which tends to equalize all of the relaxation rates in a system. Spin (liffusion becomes important in the non-extreme narrowing hmit and results in a completely non-specific noe. In this case the saturation of any resonance produces an noe on all other resonances, independent of the distance between them (Roberts & Jardetzky, 1981).  IV.A.5: Summary and Overview. The I ^ F nucleus is well suited to investigations of small molecule/macromolecule interactions. It has a high sensitivity, a wide chemical shift dispersion and is only present when and where it is introduced by the investigator. Thus the fluorinated substrates and inhibitors used in this study are good candidates for this type of investigation. The main theme behind this chapter is the relationship between the chemical shift change of the ligand on bmding to the enzyme and the environment into which the reporter nucleus has been placed. Thus the environment of each enzyme-bound glucose hydroxyl has been investigated by the use of the appropriate fluorinated analogue. In order to obtain data on each of the glucose 6-P hydroxyls, the complete series of mono-fluorinated glucose 6-P analogues has been synthesized. As well as investigating the binary enzyme-ligand complexes, the ternary enzyme-metalligand complexes have also been probed. Three metal ions, viz., Mg , C d and L i have 2+  2+  +  been used in this study. All three bind at the activating metal site, but all elicit a very different activity from the enzyme. Mg + is highly activating, Cd + is poorly activating (0.8% that of 2  2  Mg2+) and L i elicits no activity (< 10"^ that of Mg ). +  2+  Using these metal ions, the mono-fluorinated substrates and the mono and difluorinatedinhibitors, the interactions between the enzyme and its two substrates have been probed with the aim of providing evidence for the two types of mechanism postulated for this enzyme.  134 IV.B: RESULTS AND DISCUSSION.  IV.B.l: Synthesis of Fluorinated and Deoxygenated Analogues of Glucose 6-Phosphate. The complete series of fluorinated analogues of glucose 6-P, viz.,1-,  3- and 4-fluoro-  2-,  glucose 6-P and 1-deoxy-glucose 6-P was synthesized by reaction of the corresponding fluoro or deoxy-glucose with hexokinase and adenine 5'-triphosphate (ATP). Hexokinase catalyzes the phosphorylation of the C-6 hydroxyl of glucose and a number of its epimers using ATP as the phosphate donor. The specificity of yeast hexokinase for each fluoro-glucose analogue and some fluorinated epimers of glucose has been investigated in detail (Bessel et al., 1973, Drueckhammer & Wong, 1985). The authors found that most of the analogues were substrates of the enzyme although the efficiency of catalysis varied greatly. The exceptions were a and Pglucosyl fluoride (cx/P-l-fluoro-glucose) which were found to be neither substrates nor inhibitors. However 1-deoxy-glucose is a substrate of hexokinase, although reacting at a much slower rate than the fluorinated analogues of glucose (Machado de Domenech & Sols, 1980). The kinetic parameters of the substrates of hexokinase used in this study are shown in Table IV.2  Substrate  a  Km(mM)  Vmax (relative)  glucose  0.17  1.0  2-fluoro-glucose  0.20  0.5  3-fluoro-glucose  70  0.1  4-fluoro-glucose  84  0.1  1-deoxy-glucose  20  0.02  a  relative to glucose.  Table IV.2: Kinetic parameters of yeast hexokinase with deoxy and fluoro analogues of glucose. (Data from Bessel et al., 1972 and Machado de Domenech and Sols, 1980).  135  The catalytic efficiency of hexokinase with each of the glucose analogues is such that the corresponding 6-phosphate ester was readily prepared by incubation (25 °C) of a solution containing the sugar with a 10% excess of ATP and hexokinase at pH 7.6. The progress of the reaction was monitored by l°p-nmr and the change in pH. After completion, the mixture was chromatographed at pH 8.0 on a DE 52 cellulose ion exchange column. Each glucose 6-P analogue was characterized by ^ F and/or lp-nmr. 9  3  The synthesis of a-glucosyl fluoride 6-P was more complicated since it has been reported that a-glucosyl fluoride is not a substrate of hexokinase. This observation is surprising in light of the fact that 1-deoxy-glucose is phosphorylated by the enzyme. The a and pVglucosyl fluorides are the only glucosyl halides stable enough to exist as deprotected sugars. They are however fairly unstable, the fluorine being susceptible to eUmination, with participation from the ring oxygen (Micheel & Klemer, 1961). Whether the proposed phosphorylated glucosyl fluoride would be stable enough to be synthesized via a chemical route and whether it would be stable enough for the intended enzymic studies was unknown. A useful method for the preparation of 2,3,4,6-tetra-O-acetyl-a-D-glucosyl fluoride is by the reaction of penta-O-acetyl-P-D-glucose with Olah's reagent (pyridinium poly(hydrogen fluoride) Hayashi et al, 1984). Adaptation of this method to the synthesis of a-glucosyl fluoride 6-P leads to the synthetic scheme shown in Figure IV.2. However, this approach was unsuccessful as the reaction of the per-acetylated glucose 6P with Olah's reagent resulted in cleavage of the phosphate ester. An attempt to fluorinate 1,2,3,4-terra-O-acetyl-pVD-glucose at C-l with Olah's reagent with the aim of phosphorylating the product at the unprotected primary hydroxyl by reaction with diphenyl phosphorochloridate was also unsuccessful. Three products of the fluorination reaction were obtained, the major one being identified by ^H-nmr as tri-0-acetyl-l,6-anhydro-p-Dglucopyranose, the other two were not identified but contained no fluorine.  136 ,OP0 Hi v Hi  2-  2.OPO^  2  3  AC2O  HF/pyridine 'OAc NaOMe MeOH ,OPO^2Hl  Hi  Figure rv.2: Attempted chemical synthetic route to a-glucosylfluoride6-P. Finally an attempt was made based on the observation that 3-fluoro-glucose 6-P can be prepared by reaction of benzyl 3-fluoro-a-D-glucopyranoside with the more selective phosphorylating agent dibenzyl phosphorochloridate. Following hydrogenolysis to remove the benzyl groups, the desired product was obtained in 90% yield (Wright et al., 1972). This method was attempted with a-glucosylfluorideas a substrate, using 2,6-dimemylpyridine as solvent rather than pyridine which had been used by the original authors. 31p-nmr of the reaction mixture revealed a complex mixture of products and 19p-nmr showed that the glucosyl fluoride had been cleaved. Due to the lack of success in the chemical synthesis of a-glucosyl fluoride 6-P, it was decided to reinvestigate a possible enzymic synthesis using hexokinase. It appeared possible that a-glucosyl fluoride is an extremely poor substrate of hexokinase, its low activity being overlooked by previous investigators. The reaction of a high concentration (0.22 M) of aglucosylfluoridewith hexokinase (140 units) was monitored by l^F-nmr (proton decoupled). The appearance of a new signal, shifted slightly downfield from that of a-glucosyl fluoride, over a period of 24 hours suggested that phosphorylation was occurring. However, the integrated area of the new peak showed that only a small fraction of the a-glucosylfluoridehad been  137  phosphorylated, the greater proportion being hydrolyzed with the production of fluoride. The rate of spontaneous hydrolysis of a-glucosyl fluoride was measured as approximately 0.3% h"  1  at pH 7.6 and 30° by coupling the production of glucose with glucose dehydrogenase and NADP. Control experiments showed that the comparatively rapid hydrolysis of a-glucosyl fluoride in the presence of hexokinase was caused by slight contamination with an aglucosidase. a-Glucosidases hydrolyze a-glucosidic linkages of polysaccharides but also have been shown to efficiently hydrolyze a-glucosyl fluoride (Barnett et al., 1967). A search was conducted to find the commercially available hexokinase preparation having the lowest aglucosidase activity. Each commercial preparationfromSigma Chemical Co. (5 in total) was tested for a-glucosidase activity with the substrate p-nitrophenyl-a-D-glucopyranoside by monitoring the release of p-nitrophenol at 400 nm. The activity of each preparation ranged from approximately 4 x 10"H to 0.08 x 10" H mol/min/unit hexokinase. The a-glucosidase activity of the latter preparation (H-5875) was reduced to an undetectable level by addition of the potent glucosidase inhibitor acarbose, to a concentration of 2 mM. Thus a-glucosyl fluoride 6-P was synthesized in the same manner as the other fluoro and deoxy analogues; however, the concentration of the substrate was increased to attempt saturation of the enzyme. Following completion of the reaction at 25°, all manipulations and chromatography were carried out at 4° to reduce the amount of non-enzymic hydrolysis. The product, containing approximately 15% glucose 6-P, was characterized by ^ F and P-nmr 9  3l  spectroscopy. A solution of a-glucosyl fluoride 6-P frozen at -20° in buffer at pH 7.5 was found to be fairly stable, about 10% decomposing to produce glucose 6-P and fluoride over one month. IV.B.2: Investigation of Enzyme-Ligand Complexes by l F-nmr. 9  IV.B.2.1: Complexes of 6-Fluoro-Glucose 1-Phosphate and a-Glucosyl Fluoride 6-Phosphate. The initial l F-nmr experiments were carried out on the binary phosphoenzyme-6-fluaro9  glucose 1-P complex (Ep.6FGlc 1-P). This ligand was chosen for the initial studies since it bindstightlyto the phosphoenzyme (Kd = 10 (J.M, Section E.B.3.4) but cannot undergo any  138  phosphoryl transfer reaction since the acceptor hydroxyl is replaced by fluorine. The advantage of this is that since no turnover can occur, a slight contamination by an activating metal ion will have no effect, as long as its concentration is below 5-10% of the enzyme concentration, i.e., no ternary metal-enzyme-ligand complex will be observable. The initial experiments were conducted with enzyme concentrations of 1.4 and 1.2 mM. These concentrations of enzyme and ligand allowed spectra with a signal to noise ratio of greater than 10 to be acquired in one hour. In order to conserve valuable enzyme and to standardize variables such as viscosity, the remaining experiments were carried out with an enzyme concentration of 1.0 mM. Thereportedchemical shifts are referenced to trichlorofluoromethane (Freon, 8 = 0.0 ppm), upfield shifts having a positive value. These values were calculated from the internalreference4-fluoro-a-methyl glucoside (4FMeGlc) which has a chemical shift of 199.2 ppm relative to Freon. 4-Fluoro-amethyl glucoside would be expected to have very Utile affinity for the enzyme since the K of m  glucose itself is 0.7 M (Layne & Najjar, 1978). Indeed no evidence of binding (i.e., no broadening of the reference signal) was observed. All spectra were measured at pH 7.5 in a solution containing 20 mM Tris-HCl and 50% D2O. The temperature was 20° unless noted otherwise. The l^F-nmr spectrum of a 1:1 molar ratio of phosphoenzyme (Ep) and 6-fluoro-glucose 1-P (6FGlc 1-P) (Figure IV.3.B) shows aresonancethat is shifted 14.2 ppm downfield from that of ligand in the absence of enzyme (Figure IV.3. A). This shift is of such a magnitude that initial attempts to observe theresonancewere hampered by the narrower spectral width initially used to collect the data! Addition of a second equivalent of 6-fluoro-glucose 1-P produced a secondresonanceat the same chemical shift as that of the free ligand (Figure IV.3.C.) The resonance occurring at 223.2 ppm can be assigned as the Ep.6FGlc 1-P complex and the peak at 237.4 ppm, appearing on addition of a second equivalent of ligand, as free 6-fluoroglucose 1-P. Since both peaks are relatively sharp (At) 1/2 = 150 and 120 Hz, respectively) and the peak of the free ligand is not shifted, it is concluded that exchange between enzyme-bound and free species is slow on the nmrtimescale withrespectto chemical shift Thus the data are adequately described by the simple equilibrium  139  4FMeGlc Reference  B.  170  180  190  200 210 220 230 Chemical shift 8(ppm) from freon.  240  250  Figure IV.3: Titration of phosphoenzyme with 6-fluoro-glucose 1-P followed by Fnmr. All solutions were at 28°: (A) Spectrum of "free" 6FGlc 1-P; (B) 1.4 mM Ep, 1.4 mM 6FGlc 1-P; (C) 1.4 mM Ep, 2.8 mM 6FGlc 1-P. 19  140 E +L "  p.  EL  However the possibility cannot be excluded that the real situation is E +L  w  EL  -  EL*  where exchange between E and EL is slow, but exchange between EL and EL* is rapid so that the observed bound resonance corresponds to a fast exchange-averaged signal from EL and EL*. This binary complex was found to be very stable, only after several weeks at 4° were small additional peaks observed in the spectrum. These were assigned as a and f3-6-fluoroglucose (8 = 236.5 and 236.0 ppm, respectively) and may be due to a contaminating phosphatase or a small phosphatase activity of phosphoglucomutase itself. This phenomenon has been previously reported (Rhyu et ai, 1985a). Broadband proton decoupling causes the complete disappearance of the signal from the bound ligand (Figure IV.4.B). These results are in accord with those of Sykes et al. (1974) who observed the same phenomenon for signalsfromfluorotyrosine-labelled alkaline phosphatase. The disappearance of signal is due to a negative nuclear Overhauser effect (noe) caused by the longer correlationtimeof the bound ligand. The signal from the free ligand was narrowed as expected but was reduced to approximately 20% of its intensity in the coupled spectrum, the magnetization of the bound ligand being transferred to thefreesignal by chemical exchange. Negative nuclear Overhauser effects are discussed further in Section rV.B.4. Fluorine-proton spin-spin splitting can also be removed by the gated pulse sequence, DNA (Decoupler oN during Acquisition). In this sequence the *H decoupler is gated on during acquisition of the free induction decay (fid) to collapse any proton-fluorine spin-spin splitting and gated off during a time period between successive fid acquisitions to allow the fluorine spin system to re-equihbrate before the next pulse. In this way any multiplets are collapsed but no noe is induced. If the noe is to be completely removed, then the delay between pulses must be severaltimesthe Tl of the resonance and hence the time required to acquire the spectrum is lengthened. DNA decoupling of the sample consisting of a 2:1 molar ratio of ligand to enzyme (Figure IV.4.C) resulted in a collapse of the multiplets of the signalsfromthe reference (AD 1/2 = 80 to 5 Hz) andfreeligand (Al) 1/2 = 105 to 40 Hz), but only a relatively small reduction in linewidth of the bound ligand  141  4FMeGlc Reference A.  170  180  190  200 210 220 230 Chemical shift 8(ppm) from freon.  240  250  Figure IV.4: Effect of^H decoupling on the l^F-nmr spectrum of the phosphoenzyme-6fluoro-glucose 1-P complex. All solutions were at 28°: (A) 1.2 mM Ep, 1.2 mM 6FGlc 1-P, no decoupling; (B) same as (A) but with continuous but with gated (DNA) *H decoupling.  decoupling; (C) same as (A)  142 (AD 1/2 = 130 to 100 Hz). The narrow linewidth of the reference peak (Aui/ = 5 Hz) confirms 2  that the broadening of the free ligand is not due to a viscosity effect, but rather exchange with the bound species. Inspection of the integrated intensities of the three signals (reference, bound and free ligand) in the absence and presence of DNA decoupling (Figure IV.4.A and C) showed that the relative intensities of the bound and free ligand compared to the reference are reduced in the presence of decoupling. To a certain extent this is due to the effects of the differences in acquisition times and the different spin-lattice relaxation times (Tl) of each species. However after calculation of the equihbrium intensities of each signal, based on a knowledge of the Tl of each species and the spectral acquisition times (Sykes & Hull, 1978), the intensities of the enzyme-bound and free ligand are still reduced by 65% and 50% .respectively, in the presence of DNA decoupling. Even when the delay period between pulses was lengthened, several fold the same reduction in the intensity of the bound peaks was observed. This phenomenon could not be explained at the time this study was undertaken. However, during later testing of the spectrometer for experiments using gated fluorine decoupling, it was discovered that the electronic gate in the decoupling hardware was faulty and allowed a small current to pass continuously. This current was insufficient to decouple any spin-spin splitting, but could explain the negative noe since a much smaller decoupling power is used in such experiments compared to that needed to remove spin-spin splitting. The Ti values of the internal reference, free and bound ligand of the Ep.6FGlc 1-P complex were measured by the progressive saturation method under a variety of conditions of temperature, D2O concentration and presence of protein. The steady state progressive saturation method was used to measure Tl as it does not require the 5Ti waiting period between pulses necessary with the more accurate inversion recovery method. The results are shown in Table IV.3.  143 Species  Enzyme concentration (mM)  Temperature (°C)  Solvent ratio  4FMeGlc  0  20  1:1  1.04 ± 0.05  4FMeGlc  1.0  20  1:1  1.05+0.02  4FMeGlc  1.0  6  1:1  0.84 ± 0.02  4FMeGlc  1.0  7  99:1  0.99 ± 0.02  6FGlc 1-P  0  20  1:1  0.61 ± 0.02  6FGlc 1-P  1.0  20  1:1  0.42 ± 0.02  6FGlc 1-P  1.0  7  1:1  0.39 ± 0.01  E .6FGlc 1-P  1.0  20  1:1  0.21 ± 0.01  E .6FGlc 1-P  1.0  6  1:1  0.29 ± 0.03  E .6FGlc 1-P  1.0  7  99:1  0.30 ± 0.03  P  P  P  Table IY.3: F 19  Ti (s)  D2O/H2O  spin lattice relaxation data of enzyme-bound andfree species in the  Ep.6FGlc 1-P complex.  The result of a Tl determination is formally simple only in the presence of lH irradiation in which case the ^p-nuclei will relax in a simple exponential manner (Sykes & Weiner, 1980). However, saturation of the proton signals by continuous irradiation causes the complete disappearance of the bound signal due to the negative noe. The Ti experiment can be carried out with gated decoupling, but the time necessary for the experiment and the observation that relative intensity changes did occur on gated decoupling made this technique unsuitable. A theoretical study of  I^F-IH  dipolar relaxation in macromolecules (Hull & Sykes, 1975b) has shown,  however, that Ti will be the same in the presence or absence of ^H decoupling. This has been verified experimentally (Opella et al, 1979, Ando et al., 1986). In all cases in this study, the relaxation data did fit a simple exponential decay curve. The difference between the Ti values of the reference, 4-fluoro-a-methyl glucoside (1.0 s) and that of the ligand, 6-fluoro-glucose 1-P (0.61 s) in the absence of any protein is due to the dominance of the dipolar spin-lattice relaxation mechanism. The fluorine in 6-fluoro-glucose 1-P  144 has two geminal protons whereas that in 4-fluoro-a-methyl glucoside has only one. Consequently thefluorinewith the greater number of dipolar neighbors (protons) is more efficiendy relaxed. On binding to the protein the observed Tl value of the ligand decreases from 0.61 s (in the absence of protein) to 0.21 s. The large increase in the efficiency of the spin lattice relaxation process arises because correlation times in the bound state are much longer. The effect of the increased correlationtimeof the enzyme-bound ligand is also observed in spin-spin relaxation times (T2). Thus T2 of the ligand in the absence of protein is 0.064 s but is reduced to 0.003 s on binding to the enzyme (Table IV.4).  A\)i/2 (Hz) a  T2 (s)  4FMeGlc  5  0.064  6FGlc 1-P  40  0.008  E .6FGlc 1-P  100  0.003  Species  P  a b  b  line broadening of 20 Hz subtracted from value measured from the spectrum. calculated by T2 = l/(TIA\)i/2).  Table IV.4: l^F spin -spin relaxation data of enzyme-bound and free species in the Ep.6FGlc 1-P complex. Data from the spectrum shown in Figure IV.4.C. The T of each nucleus can also be affected by the temperature of the sample. A reduction c  in temperature increases the viscosity of the solution and hence the Tc is increased. This effect is demonstrated by the reduction in Tl value of the reference 4-fluoro-a-methyl glucoside from 1.05 to 0.84 s on lowering the temperature from 20° to 6°. In contrast, the value of Tl of the bound ligand increases from 0.21 to 0.29 s thus demonstrating that the value of Tc of the bound ligand is such that any further reduction in local motion causes a decrease in the efficiency of spin lattice relaxation. This is as expected for the non-extreme narrowing limit. The Tl of a nucleus may also be dependent on the concentration of H2O in the solution if dipole-dipole relaxation with solvent is an important process. Dipolar relaxation is proportional to  145  the square of the gyromagnetic ratio of the relaxing nucleus. The gyromagnetic ratio of hydrogen is 26752 rad G"* s* compared to 4107 rad G^s"* for deuterium. Consequently, dipolar -  relaxation due to deuterium will only be 2% as efficient as that from hydrogen (Gerig, 1982). This effect is observed in the Ti values of the reference in the presence of 50% and 99% D 2 O (Table IV.3), the Tl being increased from 0.84 s to 0.99 s on increasing D 2 O concentration from 50 to 99.8%. However, the value for Tl of the bound ligand under the same conditions was not affected. This is evidence that the fluorine nucleus (the acceptor hydroxyl in the active substrate) of the bound ligand is largely inaccessible to solvent. Similar experiments using 31p-nmr, Tl and noe studies (Rhyu et ai, 1984) suggested that the enzymic phosphate group is accessible to water in the absence of ligand, whereas in the ternary complex of phosphoenzyme, glucose 6-P and Li , (Ep.Li .Glc 6-P) the enzymic phosphate is probably inaccessible to water. These +  +  results are in accord with the present findings since it would be expected that the enzymic phosphate group and the fluorine at C-6 would be in close proximity to each other. Solvent accessibility of the probe nucleus could also be probed by measurement of a possible solvent isotope shift (SIS), the chemical shift of nuclei exposed to D 2 O often being different to that in H2O. However, as the SIS is generally very small, in the order of 0.2 ppm (Hull & Sykes, 1976), a difference of this magnitude would be difficult to detect in this system, the spectral width of which is 79 ppm. Addition of 10 mM L i to the binary Ep.6FGlc 1-P complex caused a further shift of the +  bound ligand 1.5 ppm downfield, bringing the total shift from that of free ligand to -15.9 ppm (Figure IV.5). A slight narrowing in the linewidth of the bound resonance accompanies the shift caused by the binding of Li . This slight reduction in linewidth (Ai)l/2 130 to 120 Hz) is most +  likely due to the decreased rate of ligand exchange resulting from the binding of Li . This +  phenomenon is more fully discussed in Section IV.B.2.4. In addition to the changes in chemical shift and linewidth, there is also an increase in the intensity of the bound resonance on the binding of Li . This could be caused by a decrease in Ti although this has not been confirmed. +  146  4FMeGlc Reference  ^ ^ ^ ^  \^fMfJ  B.  170  180  190  200  210  220  230  240  250  Chemical shift 8(ppm) from freon.  Figure rV.5:  Effect of binding of Li on the phosphoenzyme-6-fluoro-glucose 1-P complex followed by F-nmr: (A) 1.0 mM Ep, 1.0 mM 6FGlc 1-P; (B) 1.0 mM Ep, +  19  1.0 mM 6FGlc 1-P, 10 mM LiCl, 1.0 mM EDTA.  147  The next series of experiments that are described are those involving a-glucosyl fluoride 6-P (aGlcF 6-P). This ligand is of interest since it is complementary to 6-fluoro-glucose 1-P and is a non-reacting analogue of glucose 6-P. The l^F-nmr spectrum of a-glucosyl fluoride 6-P and the internal reference 4-fluoro-a-methyl glucoside is shown in Figure YV.6.A. The l^p-nmr spectrum of the complex formed by addition of approximately 1.2 equivalents of a-glucosyl fluoride 6-P to 1.0 equivalent phosphoenzyme is shown in Figure IV.6.B. The bound ligand at 131.7 ppm is shifted 19.7 ppm downfield from that of the excess unbound ligand. The larger than anticipated excess of free ligand is due to the presence of contaminating glucose 6-P which bound to the phosphoenzyme, thus reducing the amount of phosphoenzyme (with which aglucosyl fluoride 6-P can bind) by approximately 25%. The sharp peak at 121.4 ppm is fluoride ion arising from decomposition of a-glucosyl fluoride 6-P. Addition of a second equivalent of a-glucosyl fluoride 6-P causes an increase in the intensities of the peaks assigned to the free ligand and fluoride (spectrum not shown). The addition of 10 equivalents of L i shifts the +  resonance of the bound peak 1.5 ppm upfield towards that of the free ligand and also appears to cause a decrease in linewidth (Aoi/2 ~ 140 to 115 Hz, Figure IV.6.C). The chemical shifts and changes in chemical shifts of the complexes of phosphoenzyme and 6-fluoro-glucose 1-P and a-glucosyl fluoride 6-P are summarized in Table IV.5. Species  Chemical shift 5(ppm) Change in chemical shift (ppm) a  6FGlc 1-P  237.4  Ep.6FGlc 1-P  223.2  -14.2  221.5  -15.9  Ep.Li+.6FGlc 1-P  1  aGlcF 6-P  151.3  Ep.aGlcF 6-P  131.6  -19.7  Ep.Li+.aGlcF 6-P  133.1  -18.2  a  changes in chemical shift are from the free ligand.  Table IV.5: Summary  of 19F-nmr chemical shifts of complexes of phosphoenzyme and  6-fluoro-glucose 1-P and a-glucosyl fluoride  6-P.  148  4FMeGlc Reference  100  120  140  160  180  200  220  Chemical shift 8(ppm) from freon. Figure rv.6: Titration ofphosphoenzyme with a-glucosyl fluoride 6-P and Li followed +  by F-nmr: All spectra are DNA decoupled (A) Spectrum of "free" aGluF 6-P; (B) 1.0 19  mM Ep 1.2 mM aGluF 6-P; (C) 1.0 mM Ep, 2.4 mM aGluF 6-P, 10 mM LiCl, 1.0 mMEDTA.  149  To our knowledge the magnitudes of the ^^p chemical shift changes caused by binding of each substrate analogue are the largest yet observed for a small molecule/macromolecule interaction. As noted in the introduction, the incorporation of fluorine reporter groups in a protein or nucleic acid generally results in downfield shifts of up to approximately 8 ppm. The size of the shift observed in this system and a knowledge of the binding site into which the nucleus has been introduced may shed some light on the cause of fluorine chemical shifts in such macromolecular systems. The origin of these large shifts is most probably either Van der Waals interactions or electrical field effects from the proximal enzymic phosphate group. The large size of the shift rules out anisotropic ring currents as the major cause, although a fraction of the observed shift could result from this type of effect. Likewise, hydrogen bonding is unlikely to be the cause of the large shift, since inhibition studies (Section JJ.B.3.4) showed that at most only very small hydrogen bonding interactions occur between the "acceptor" hydroxyls and the protein. The large downfield shift on binding of both ligands to the enzyme is consistent with observations by previous workers that covalent incorporation offluorinein a protein, or the non-covalent association with a protein, reflects a change to a hydrophobic (buried) environment. However, this may not be consistent with the expected structure of the active site of phosphoglucomutase. X-ray crystallographic studies (Lin et al., 1986) have shown that the active site of phosphoglucomutase consists of a deep crevice in which three charged arginine residues have been identified. In addition, thefluorinewould be expected to be very close to the enzymic phosphate groups as well as the activating metal ion. The remainder of the active site probably consists of polar side chains which form the hydrogen bonds to the sugar ring hydroxyl groups. Thus the environment into which the fluorine has been introduced is probably hydrophilic rather than hydrophobic. The conclusions by previous workers that changes in chemical shifts produced on incorporation offluorinein a protein or polynucleotide are mainly due to hydrophobic effects, i.e., Van der Waals interactions (Gerig, 1978,1982, Sykes & Weiner, 1980), may reflect the nature of the reporter group used. Generally thefluoro-aminoacids incorporated in proteins and polynucleotides have been analogues of tyrosine, phenylalanine,  150  tryptophan and uracil. Similarly, many of the non-covalent probes have also involved ligands that have been fluorinated in hydrophobic regions of the molecule. Consequently, it would be expected that these hydrophobic molecules would lie in hydrophobic regions of the macromolecule where Van der Waals interactions dominate. However, in this system the probe will clearly be in a hydrophilic region of the protein and it can be suggested that the large downfield shifts observed are due to the electrostatic fields arising from the proximal phosphate group. This suggestion is supported by the much smaller downfield chemical shift changes that are observed on the binding of the glucose 1-P analogues in which the glucose ring hydroxyls have been replaced by fluorine (Section rV.B.2.4). These fluorines are involved in hydrogen bond interactions with the protein but will be relatively distal to the enzymic phosphate group. The two inhibitors used in this nmr study are both structurally virtually identical to their parent substrates. The expectation that the enzyme will bind both of these inhibitors in the same manner as it would the parent substrates, is corifirmed by the fact that they bind competitively with similar affinities (compared to the parent substrates). Thus the fluorine reporter nuclei of each of the two inhibitors would be expected to occupy the same position(s) in the active site of the enzyme as would the "acceptor" hydroxyls of both substrates, but phosphoryl transfer is blocked. If the "acceptor" hydroxyl groups of the native substrates glucose 1-P and glucose 6-P were to occupy the same site in the enzyme-substrate complex, as may occur in an exchange type of mechanism (Figure 1.3), then the changes in chemical shifts on binding of both fluorinated inhibitors, caused by local environmental effects, would be expected to be very similar. In contrast, if a minimal motion type of mechanism were to operate, then the two fluorines (of 6FGlc 1-P and aGlcF 6-P) would occupy slightly different sites (Figure 1.2) This may result in two different chemical shift changes. The results obtained show that binding of 6-fluoro-glucose 1-P causes a downfield shift of 14.2 ppm, whereas binding of a-glucosyl fluoride 6-P causes a larger downfield shift of 19.7 ppm. This suggests that the environments of the fluorine reporter nuclei of the inhibitors are not identical. A difference in the structures of the two phosphoenzyme-glucose phosphate complexes has also been suggested by 31p and ^Li-nmr studies. Binding of glucose 1-P causes a downfield shift of the enzymic phosphate, whereas  151  binding of glucose 6-P produces a shift in the opposite direction (Rhyu et al, 1984). Similarly the chemical shift of the bound metal ion L i differs in the presence of glucose 1-P and glucose +  6-P (Rhyu et al, 1985a). All three of these reporter nuclei, viz., Li , enzymic phosphate and +  fluorine would be situated in the same region of the active site and all undergo different chemical shift changes depending whether the ligand is glucose 1-P or glucose 6-P. It is difficult to rationalize whether the differences in the chemical shift changes of the two ligands observed in this study provide evidence for one mechanism over the other. The changes in chemical shift are not identical, but they are both very large shifts in the same direction. An additional complication is the differences in the carbon atoms to which each fluorine is bonded. Whereas one fluorine is bonded to a primary carbon, the other is bonded to a secondary center. It is not known whether an identical external electric or magnetic field would cause the same change in chemical shift of the two different types of fluorine atom. The binding of L i to the Ep.6FGlc 1-P and Ep.ccGlcF 6-P complexes causes a further +  small shift of approximately 1.5 ppm. However the directions of these shifts are different (Table IY.5). This may suggest that the positions in the active site of the fluorine nuclei of 6-fluoroglucose 1-P and a-glucosyl fluoride 6-P are not the same, since the added effect of a nearby L i  +  ion would be expected to have a similar shielding or (tesruelding effect on both nuclei if they occupied the same site, assuming that there is only a single activating metal site for both complexes. It is not a simple matter to determine unequivocally whether the change in chemical shift which occurs on binding of L i is due to the direct electric field effect of the ion, or rather an +  indirect effect caused by a conformational change in the protein. Kimber et al, (1977) have suggested a rationale for deciding between the two alternatives. They considered that if a chemical shift change arises from an electric field effect then it will be downfield (see introduction). On the other hand, a conformational change can result in either an upfield or downfield shift depending on whether the interacting groups are moved closer to or further away from the fluorine nucleus. Thus the upfield shift produced on the bmding of L i to the +  Ep.aGlcF 6-P complex is most probably due to a conformational effect that moves charged or  152  polar groups away from the fluorine nucleus. One would expect then that the downfield shift observed on the binding of L i to the alternate Ep.6FGlc 1-P complex is also due to a +  conformational effect, although a direct electric field effect cannot be ruled out since the possibility exists that the fluorine nuclei in the two binary complexes occupy at least slighdy different sites. Thus the results of this investigation in terms of providing evidence for an exchange or minimal motion type of mechanism are rather ambiguous. Ideally an exchange mechanism would have produced two identical chemical shift changes, whereas a minimal motion mechanism would produce two opposite chemical shift changes. In reality the shifts, -14.2 and -19.7 ppm are neither identical, nor very different.  IV.B.2.2: Effect of Binding of Activating Metal Ions. The effect of two activating metal ions, Mg2+ and C d on the binary Ep.6FGlc 1-P 2+  complex was followed by ^p-nmr. The addition of 1.1 equivalents of Mg2+ to the Ep.6FGlc 1P complex at 20° (1:1 molar ratio) causes a large decrease in signal intensity of the bound peak and the resonance is broadened (AD 1/2 ~ 1600 Hz) and shifted upfield to 227.7 ppm (AS = +4.5 ppm) (Figure IV.7.B). When the same sample is cooled to 6° the broad peak sharpens (AD 1/2 " 500 Hz) but the chemical shift is unchanged within experimental error (Figure  TVJ.C).  The spectra obtained when the Ep.6FGlc 1-P complex (1:2 molar ratio) istitratedwith Mg2+ at 28° are shown in Figure IV.8.A-F. The spectrum is essentially unaffected by the addition of 0.5 equivalent of Mg2+ (Figure IV.8.B) but after addition of a second 0.5 equivalent ofMg + a new broad shoulder appears downfield on the peak of the free ligand at 236.2 ppm. A 2  possible broad peak also appears at approximately 227.2 ppm (Figure IV.8.C). On cooling to 6° (Figure IV.8.D) the resonance from the free ligand sharpens and the broad resonance ( AD 1/2 ~ 500 Hz) at approximately 227.2 ppm increases in intensity. Finally, after addition of a total of 2.0 equivalents of Mg + at 28° the peaks from the free and bound ligand at 223.2 and 237.4 2  ppm have disappeared and are replaced by two very broad peaks at 227.2 and 236.2 ppm (Figure IV.8.E). Cooling to 6° sharpens the resonance at 227.2 ppm but the resonance at 236.2 ppm  153  170  180  190  200  210  220  230  240  250  Chemical shift 5(ppm) from freon. Figure IV.7: Effect of Mg? on the phosphoenzyme-6-fluoro-glucose 1-P complex followed by F-nmr: (A) 1.0 mM Ep, 1.0 mM 6FGlc 1-P, 20°; (B) 1.0 mM Ep, 1.0 mM 6FGlc l-P.l.l mM MgCl2,20°; (C) same as (B) except at 6°. +  19  154  4FMeGlc Reference A.  B.  c.  170"  180  190  200 210 220 230 Chemical shift 5(ppm) from freon.  240  250  Figure IV.8: Titration of the phosphoenzyme-6-fluoro-glucose 1-P complex with Mg followed by ^F-nmr: (A) 1.4 mM Ep, 2.8 mM 6FGlc 1-P, 28°; (B) 1.4 mM Ep, 2.8 mM 6FGlc 1-P, 0.7 mM MgCl2,28°; (C) 1.4 mM Ep, 2.8 mM 6FGlc 1-P, 1.4 mM MgCl2, 28°. (cont.)  2+  Figure IV.8 (cont.): (D) 1.4 mM Ep, 2.8 mM 6FGlc 1-P, 1.4 mM MgCl2, 6°; (E) 1.4 mM Ep, 2.8 mM 6FGlc 1-P, 2.8 mM MgCl2, 28°; (F) same as (E) except at 6°.  156  4FMeGlc  Reference  180  190  B  170  200 210 220 230 Chemical shift 8(ppm) from freon.  240  250  Figure IV.9: Effect ofMg^ on the phosphoenzyme-Li+-6-fluoro-glucose 1-P complex followed by ^F-nmr: (A) 1.0 mM Ep, 1.0 mM 6FGlc 1-P, 10 mM LiCl, 1.0 mM EDTA. (B) 1.0 mM Ep, 1.0 mM 6FGlc 1-P, 10 mM LiCl, 1.0 mM EDTA, 20 mM MgCl2+  157  appears to split into two peaks having chemical shifts of approximately 236.2 and 237.4 ppm (Figure IV.8.F). This set of spectra was reproducible. Similar results were obtained when the Ep.Li .6FGlc 1-P complex was titrated with +  Mg + at 20° (Figure IV.9.B). After the addition of 20 mM Mg + (a 2:1 Mg /Li+ molar ratio), 2  2  2+  the resonances at 221.5 and 237.4 ppm, which had been assigned as enzyme-bound and free ligand, were reduced in intensity and two broad peaks appeared at 227.1 and 236.3 ppm. These data suggest that Mg2+ binds slightly weaker to the enzyme-inhibitor complex than does Li , as +  less than 50% of the L i was displaced in the presence of a 100% excess of Mg . This +  2+  contrasts with the results of Rhyu et al. (1985a), in which it was found that C d readily 2+  displaces a large excess of L i from the activating metal site. It appears that poorly activating +  metal ions, such as Cd , bindtighterthan Mg2+ (Ray, 1969). 2+  Treatment of the Ep.6FGlc 1-P complex (2:1 molar ratio) with 1.0 equivalent of Cd2+ caused the bound and free peaks to broaden but no change in chemical shift was detected (spectrum not shown). The apparent broadening and shifts of the resonances of both free and bound ligand suggest that the presence of Mg2+ is affecting some exchange process. The observed effect is probably not related to an exchange of Mg2+ between enzyme-bound and free forms since it has been shown that the dissociation rate of Mg2+ from the Mg2+-enzyme-substrate and the Mg2+enzyme complexes (activating site) is very slow, k-i = 0.03 s"*, giving a half life of 23 s in the enzyme-bound state (Ray, 1969, Ray & Peck, 1972). The slow exchange of Mg2+ and C d  2+  has been demonstrated by the observation of separate *P enzymic phosphate signals from 3  metal-bound and metal-free forms of the enzyme in the presence of less than a stoichiometric amount of metal (Rhyu et al., 1984). Also, no contribution to the observed processes is made by any exchange effect caused by chelation of Mg2+ by the phosphate ester of the free ligand, as no line broadening of the free ligand was observed in the absence of protein at a concentration of Mg ten times that used in the above experiments. Line broadening due to rapid exchange of a 2+  small amount of contaminating paramagnetic metal ion between the enzyme and solution is also unlikely since it has been demonstrated that addition of 0.5 equivalent of Co to 2+  158  phosphoenzyme decreases the 31p-nmr signal intensity from the enzymic phosphate by 50% without any change in linewidth (Rhyu et al, 1984). To remove any doubt about contamination of paramagnetic species, a saturated solution of MgCl2 was examined by electron spin resonance. The spectrometer was set up to detect Mn which is the most likely contaminating 2+  paramagnetic metal ion. No Mn was detected and a calibration curve showed that a level of 1 2+  ppm would have been observed. As the observed effects are probably not due to any rapid exchange of metal ion between enzyme-bound and free forms, the alternative, a metal induced 6fluoro-glucose 1-P exchange process, must be considered. It has been shown that the equilibrium constant for dissociation of glucose 1-P from the Ep.Mg .Glc 1-P complex is approximately 8 \iM and that the same value can be approximated 2+  for the Ep.Glc 1-P complex (Ray  etai,  1966,1978). However, although the equilibrium  constants for the two complexes are the same, the dissociation and association rate constants may differ. Thetineshape expected for a two site exchange process in which the populations of the two environments are equal and are undergoing exchange at an intermediate to fast rate is a broad resonance equidistant between the resonances in the absence of exchange. However, at shghtly slower exchange rates, the two peaks will be broadened and shifted towards the mean of the chemical shifts. Exchange on thistimescale causes the line shapes to become distorted and nonLorentzian. Thus the broadening andinward shift of the resonances of both enzyme-bound and free ligand on addition of Mg2+ is consistent with an increase in the exchange rate between the free and enzyme-bound forms. The facts that a single resonance is observed in the presence of a 1:1 molar ratio of Mg -enzyme to 6-fluoro-glucose 1-P (Figure IV.7.B) and that two 2+  resonances of similar intensity are observed when a second equivalent of 6-fluoro-glucose 1-P is added (Figure IV.8.E) suggest that the equilibrium constant for the binding of Ep and 6-fluoroglucose 1-P is not greatly affected by the binding of Mg2+. Unfortunately, the signal to noise ratio of the spectra of Figure IV. 8 is too low for the line shape to be determined with any accuracy. Cooling the complex from 28" to 6° should result in a decrease of any exchange rate and a consequent sharpening of signals and a shift back towards the chemical shift of the  f  159  complex in the absence of exchange. Figures IV.7.C and IV.8.F show that a sharpening of the signals does occur although no change in the chemical shifts can be reliably determined. It is not simple to rationalize the observation that the addition of 1.0 equivalent of Mg2+ to a 1:1 molar ratio of the Ep.6FGlc 1-P complex (Figure IV.7.B) has the same kind of effect as the addition of 2.0 equivalents of Mg2+ to a 1:2 molar ratio of the Ep.6FGlc 1-P complex (Figure IV.8.E). These results suggest that free 6-fluoro-glucose 1-P binds the Mg2+ preferentially to the enzyme. This seems unlikely since the dissociation constant of Mg2+ from the Ep.Mg .Glc 1-P complex is 8 \iM (Ray et al, 1966,1978) whereas that from the 2+  Mg +Glc 1-P complex is 10-20 mM (Ray & Roscelli, 1966). 2  The line-shape of the peak at 236.2 ppm (Figure IV.8.E) suggests that the resonance is composed of a broad and a sharper component and that cooling to 6° causes the broad component to sharpen and shift upfield to 237.4 ppm, which is the chemical shift of thefreeligand (Figure IV.8.F). The linewidth of the resonance at 227.2 ppm (Figure IV.8.F) is similar to that of the enzyme-bound ligand in Figure IV.8.A. This suggests that the resonance at 227.2 ppm may represent the Ep.Mg2+.6FGlc 1-P complex in slow exchange. Thus, this resonance has been shifted =4 ppm upfield by the binding of Mg2+ The observation of what appears to be 3 peaks in Figures IV.8.E and F is not consistent with a simple two site exchange process and it appears that some other exchange process is occurring as well as that of 6-fluoro-glucose 1-P betweenfreeand enzyme-bound forms. The present data do not allow any further conclusions to be drawn on the nature of this process but several possibilities can be considered. 1. An exchange is occurring between two forms of the Ep.Mg2+6FGlc 1-P complex arisingfromeither a conformational change in the enzyme or an exchange of the reporter nucleus between two environments in the active site. 2. An exchange of Mg2 between the ancillary site of the Ep.Mg2+6FGlc 1-P complex +  and the free form is occurring. A shift of 1.0 ppm of the 31p-nmr signals of the E D . G I C 1,6-diP complex is caused by the binding of Cd2+ at the ancillary site (Rhyu et al., 1985a).  160 3. The possibility exists that the resonance at 237.4 ppm (Figure rv".8.A) is not that of free 6-fluoro-glucose 1-P, but rather is due to the rapid exchange of free ligand with a second, much more weakly bound site. Evidence for this second site comes from the broadening of the "free" ligand on addition of the first equivalent of Mg2+ (Figure IV.8.C) (c.f., Gerig & Klinkenborg, 1980). On binding of Mg2+, either at the activating or ancillary metal sites, this exchange process is slowed and the equilibrium is altered, resulting in the formation of two resonances at 237.4 and 236.2 ppm. Alternatively, this second site may be created by the binding of Mg2+ However, no second substrate binding site on phosphoglucomutase has been detected by either 31p-nmr or kinetic studies. In light of this it may seem more attractive to predict that the apparent second site is in fact a second environment for the fluorine nucleus in the single substrate binding site. This proposal may be related to the presence of a small unexplained resonance at 235.8 ppm which only appears on addition of a second equivalent of 6-fluoroglucose 1-P (Figures IV.3.C and IV.8.A). The fact that the presence of both Mg2+ and Cd2 causes such a marked difference in the +  spectra of the ternary complexes compared to L i may be related to the activating ability of each +  metal ion. Although C d is a poor activator of phosphoglucomutase, being about 0.8% as 2+  effective as Mg2+ (Ray, 1969), it still activates the enzyme some 10^ fold better than L i which +  binds at the same site (Ray et al., 1978). Since the fluorine reporter of 6-fluoro-glucose 1-P is probably very close to the enzymic phosphate (and therefore the bound metal ion, Rhyu et al, 1984), it is an effective reporter on events occurring right at the site of phosphoryl transfer. It can be theorized that the observed effects are due to a Mg2+ and C d induced localized 2+  conformational change which allows the phosphoryl transfer reaction to occur. This conformational change is not induced by the binding of L i and therefore the enzyme is inactive +  in its presence. Alternatively, L i may induce a different protein conformational change as was +  suggested by the changes in the chemical shift caused by the binding of L i to the Ep.oFGlc 1-P +  complex.  161  IV.B.2.3: Ternary Enzyme-Cd -Fluoro Substrate Complexes. 2+  The equilibrium distribution amongst the central complexes of phosphoglucomutase and its substrates is dependent on the identity of the activating metal ion (Ray & Long, 1976c). The predominant species formed (approximately 90%) by a 1:1:1 ratio of phosphoenzyme, glucose phosphate and C d is the dephosphoenzyme-Cd+-grucose diphosphate (Eo.Cd^+.Glc 1,62+  2  diP) complex. This complex has been observed directly by 31p-nmr (Rhyu et al., 1984,1985a); however, only a single complex of this type was observed. The distribution of glucose phosphates in the central complex was determined by rapidly quenching an equihbrium solution and subjecting the released rnixture of free glucose phosphates to analysis. Since the turnover number (kcat) of the enzyme in the presence of M g is 1000 s~l, an extremely rapid quenching 2+  procedure using perchloric acid was employed in order to ensure that no shift in the equilibrium occurred during the quenching process. Analysis of released glucose phosphates for glucose 6-P involved treating the mixture, the pH of which had been adjusted to neutrality, with NADP and glucose 6-phosphate dehydrogenase and monitoring the absorbance change at 340 nm. Glucose 1-P was then determined by addition of active phosphoglucomutase to the same solution. Finally glucose 1,6-diP was determined by the addition of xylose 1-P. This causes the hydrolysis of the phosphoenzyme which is rephosphorylated by glucose 1,6-diP, resulting in the overall conversion of glucose 1,6-diP into glucose 6-P (see Section IH.B.3). The glucose 6-P released in each case is quantitated by the reaction with glucose 6-phosphate dehydrogenase. This same approach was used in an attempt to determine the equilibrium distribution of the central complexes of each fluoro-glucose phosphate in the presence of Cd . The rather sophisticated 2+  rapid quench approach employed by Ray and Long (1976a) was not necessary since the activity of the C d form of phosphoglucomutase is 0.8% that of M g (Ray, 1969). Furthermore, the 2+  2+  Vmax of each fluoro-glucose 1-P is in the order of several percent of that of glucose 1-P (Section II.B.3.3), thereby reducing the turnover number to approximately 0.1 s"l. It could be argued, however, that the interconversion of the central complex is much faster and the rate deternnning step involves some association or dissociation process. The quenched equihbrium mixture was analyzed by the aforementioned coupled assay. However the combination of a high concentration  162  of perchlorate salt arising from the perchloric acid used to quench the enzyme, the high Km and low Vmax of each fluoro-glucose 6-P with the coupling enzyme (Bessel & Thomas, 1973), and the low concentration of the substrate made the assay unusable. Different salts such as NaCI and Na2S04 were also effective competitive inhibitors of the coupling enzyme and no activity could be detected even at a glucose 6-phosphate dehydrogenase concentration of 0.1 mg/mL. An alternative procedure for the determination of each glucose-phosphate, which was not available in our laboratory, would involve the use of 32p-ibelled enzyme. The quenched equilibrium a  mixture of labelled mono and diphosphates could be separated by anion exchange chromatography and their relative amounts quantitated from their radioactivity. The species formed in an equilibrium mixture of each fluoro-glucose phosphate and Cd -enzyme were examined by l F-nmr. The mixture was obtained by treatment of the 2+  9  phosphoenzyme-fluoro-glucose phosphate complex (1:1 molar ratio) with 1.1 equivalents of Cd2+ 1. 4-Fluoro-glucose phosphates. The proton decoupled l^F-nmr spectrum of free 4-fluoro-glucose 6-P (4FGlc 6-P, Figure IV.4.10.A) in the absence of protein shows the internal reference, 6-fluoro-a-methyl glucoside (6FMeGlc, 8 = 236.2 ppm) and the a and B-anomers. The small peak between the resonances of a and B-4-fluoro-glucose 6-P at 199.1 ppm is a-4-fluoro-glucose 1-P, presumably produced by a contaminant of phosphoglucomutase in the hexokinase used in the preparation of 4-fluoroglucose 6-P. The coupled spectrum produced by an equimolar ratio of phosphoenzyme and 4fluoro-glucose 6-P is shown in Figure IV.10.B . An identical spectrum to that of Figure IV.10.B was produced when 4-fluoro-glucose 1-P was used as the ligand instead of 4-fluoro-glucose 6P. The fact that the two spectra are identical shows that there was sufficient activating metal ion present to allow 4-fluoro-glucose 1-P to be isomerized to the thermodynamically more stable a and B-4-fluoro-glucose 6-P. Gated proton decoupling (DNA) of the complex enabled the linewidths of the two peaks (Figure IV.10.B) at 197.9 and 200.0 ppm to be determined as approximately 50 and 40 Hz, respectively (spectrum not shown). The linewidth of the reference was 5 Hz. The narrow linewidths of these peaks suggest that these species are in rapid  163  170  180  190  200  210  220  230  240  250  Chemical shift 6(ppm) from freon.  Figure IV. 10: Effect ofCd?+ on the enzyme-4-fluoro-glucose-phosphate complex followed by F-nmr. (A) Spectrum of "free" 4FGlc 6-P, decoupled; (B) 1.0 mM Ep, 19  1.0 mM 4FGlc 6-P; (C) 1.0 mM Ep, 1.0 mM 4FGlc 6-P, 1.1 mM Cd(OAc)2.  164 exchange between enzyme-bound and free forms. This was confirmed by the addition of a second equivalent of ligand, the spectrum of which (not shown) showed an increase in area of the same two resonances at 197.9 and 200.0 ppm.However, no change of chemical shift within experimental error (± 0.1 ppm) was detected. The slightly broader linewidth of the oc-anomer suggests that it has a higher affinity for the enzyme than the P-anomer as it better reflects the intrinsic T2 of the bound state. The relativelytightbinding of the P-anomer of glucose 6-P was also concluded from 31p-nmr T2 relaxation studies of the enzyme-glucose phosphate complex (Gadian et al, 1974). Addition of 1.1 equivalents of C d to the Ep.4FGlc 6-P complex (1:1 2+  molar ratio) resulted in a change to a single resonance at 198.2 ppm (Figure IV.10.C). The linewidth of this peak was shown by gated decoupling to be 130 Hz (spectrum not shown). The appearance of a single resonance, the linewidth of which suggests that it is exchanging on a slow time scale, is evidence that the complex is in fact ED.Cd2+.4FGlc 1,6-diP. The slow rate of dissociation of a diphosphate species would be expected in light of itstightbinding to the dephosphoenzyme (Kd ~ 10" 8 M) 2. 3-Fluoro-glucose phosphates. The proton decoupled 19p spectrum of 3-fluoro-glucose 6-P (3FGlc 6-P, a and panomers) and reference in the absence of protein is shown in Figure IV. 11. A. The coupled spectrum of 1.0 mM phosphoenzyme and 1.0 mM 3-fluoro-glucose 1-P is shown in Figure IV.l l.B. The spectrum indicates that the observed resonances are due to a rapid exchange of a and P-3-fluoro-glucose 6-P between enzyme-bound and free forms, a contamination of metal ion having allowed the isomerization of 3-fluoro-glucose 1-P to 3-fluoro-glucose 6-P to occur. Addition of a second equivalent of 3-fluoro-glucose 1-P caused the appearance of no additional peaks, just the intensification of the peaks at 200.1 and 195.2 ppm (spectrum not shown). On addition of one equivalent of Cd2+ two broad peaks appeared at 199.2 and 200.9 ppm in addition to the peaks arising from the free a and P-3-fluoro-glucose 6-P at 195.2 and 200.1 ppm (Figure IV. 1 l.C). It seems unlikely that the two broader resonances are due to the Ep.Cd2+a3FGlc 6-P and Ep.Cd .P-3FGlc 6-P complexes since binding of Cd2+ is not known to 2+  decrease the enzyme-substrate dissociation constant. The possibility must therefore be considered  165  6FMeGlc Reference  B.  180  i I 200 210 220 230 Chemical shift 5(ppm) from freon.  190  I 240  ~ 250  Figure IV. 11: Effect of Cd on the enzyme-3-fluoro-glucose-phosphate complex followed by F-nmr: (A) Spectrum of "free" 3FGlc 6-P, decoupled; (B) 1.0 mM Ep, 1.0 mM 3FGlc 1-P; (C) 1.0 mM Ep, 1.0 mM 3FGlc 1-P, 1.1 mM Cd(OAc)22+  19  166  that the two resonances are from two forms of the ED.Cd .3FGlc 1,6-diP complex, the 2+  existence of which has been proposed in the exchange mechanism. This spectrum was reproducible. 3. 2-Fluoro-glucose phosphates. The proton decoupled l^F-nmr spectra of 2-fluoro-glucose 6-P (2FGlc 6-P, a and pV anomers) and 2-fluoro-glucose 1-P (2FGlc 1-P) in the absence of protein are shown in Figure IV. 12.A and B. The small peak at 199.7 ppm in Figure IV.12.B is a contaminant of (J-2-fluoroglucose 1-P. The proton decoupled spectrum produced by the addition of either 2-fluoro-glucose 1- P or 2-fluoro-glucose 6-P to phosphoenzyme (1:1 molar ratio) is shown in Figure IV.12.C. Close inspection of this spectrum shows that the single peak at 200.0 ppm is composed of two resonances at 199.9 and 200.0 ppm, corresponding to the fast exchange of a and P-2-fluoroglucose 6-P between enzyme-bound and free forms. Addition of Cd2+ to the Ep.2FGlc 6-P complex causes the appearance of two peaks at 199.9 and 200.9 ppm (Figure IV. 12 D). An identical spectrum was obtained when 2-fluoro-glucose 1-P was used as the substrate instead of 2- fluoro-glucose 6-P, thereby demonstrating that equihbrium had been achieved. The two resonances at 199.9 and 200.9 ppm may represent two Ep.Cd .2FGlc 1,6-diP complexes. 2+  A summary of the chemical shifts of the species observed in the complexes of each fluoro substrate, phosphoglucomutase and Cd2+ is given in Table rv.6. The preceding spectra show that a single major bound species is present in the equihbrium mixture of 4-fluoro-glucose phosphates in the presence of a small excess ofCd + 2  whereas two bound species are present in the equilibrium mixtures of 2- and 3-fluoro-glucose phosphates. As noted above, a single ED.Glc-1,6-diP complex was detected by 31p-nmr whether in the absence of metal ion or in the presence of Li+ or Cd + (Rhyu et al, 1984). On 2  the basis of the differences in chemical shifts of each phosphate with different metal ions, the authors concluded that the signals were unlikely to representtimeaveraged signals from the two clifferent forms of the diphosphate species proposed as intermediates in the exchange mechanism (Figure 1.3). They also suggested, on the same basis, that the observed glucose 1,6-diP complex is that in which the 6-phosphate is close to the bound metal ion and the serine-116 hydroxyl.  167 6FMeGlc Reference  A.  B.  c.  D.  170  180  190  Figure IV. 12:  200 210 220 230 Chemical shift 8(ppm) from freon.  240  250  Effect of Cd on the enzyme-2-fluoro-glucose-phosphate complex followed by F-nmr: (A) Spectrum of "free" 2FGlc 6-P, decoupled; (B) Spectrum of 2+  19  "free" 2FGlc 1-P, decoupled; (C) 1.0 mM Ep, 1.0 mM 2FGlc 1-P; (D) 1.0 mM Ep, 1.0 mM 2FGlc 1 -P, 1.1 mM Cd(Q Ac)2-  168  Chemical shift 8(ppm) Ligand  Free  4FGlc 1-P  199.1  4FGlc 6-P (a/B) 3FGlc 1-P 3FGlc 6-P (a/B) 2FGlc 1-P 2FGlc 6-P (a/B)  198.3, 200.2  E .L  E .Cd2+.L  197.9, 200.1  198.0  200.1, 195.2  199.2, 200.9  199.9, 200.0  199.9, 200.9  P  P  201.2 200.6, 195.7 200.2 200.3, 200.1  Table IV.6: Summary of^F  chemical shifts of freefluorosubstrates and complexes  produced by reaction with the phosphoenzyme and the phosphoenzyme-Cd complex. 2+  These results were in accord with those of UV spectral studies in which the spectrum of the ED.Li .Glc 1,6-diP complex was found to be similar to that of Ep.Li .Glc 1-P but differing in a +  +  manner expected if a small fraction were present in the form in which the diphosphate was bound in the glucose 6-P mode. The observation that the major diphosphate species appears to be the one in which the 6-phosphate is in close proximity to the serine-116 hydroxyl, i.e., is bound in the same mode as glucose 1-P binds to Ep, is rationalized by the fact that glucose 1-P binds the enzyme substantially better than glucose 6-P (Kd = 8.4 and 57 \iM, respectively, Ray and Long, 1976a). Therefore, as suggested by Ma and Ray (1980), the enzyme-glucose ring interactions in the E D - G I C 1,6-diP complex will be stronger when the diphosphate species is bound in the same manner as glucose 1-P. In the present study it has been shown that replacement of a sugar ring hydroxyl by fluorine causes a decrease in the affinity of the enzyme for each fluoro-substrate. The results obtained by the investigation of the disubstituted fluoro-glucose 1-P inhibitors showed that replacement of the C-2 hydroxyl by fluorine produced a poorer inhibitor than the same replacement of C-3 or C-4 hydroxyls. These observations supported the suggestions of Ma and  169 Ray that the sugar ring hydroxyls distal to the acceptor hydroxyl are more important in producing a normal complex. Thus, one would expect that 2-fluoro-glucose 1-P would bind to the enzyme poorly compared to glucose 1-P, whereas binding of 2-fluoro-glucose 6-P would be affected to a lesser degree. Similarly, 4-fluoro-glucose 1-P would bind relatively well compared to glucose 1P, whereas 4-fluoro-glucose 6-P would bind poorly compared to glucose 6-P. Therefore, on the basis of the rationalizations of Ma and Ray (1980) and Rhyu et al. (1984) one would predict that 2-fluoro-glucose 1,6-diP would bind in such a manner that the glucose 6-P mode is present in a greater proportion than it is when the substrate is glucose 1,6-diP. In the case of 4-fluoroglucose 1,6-diP, the major diphosphate species would be expected to be that which binds in the glucose 1-P mode since 4-fluoro-glucose 6-P is predicted to bind relatively poorly compared to 4-fluoro-glucose 1-P. It would be difficult to predict the relative bmding of 3-fluoro-glucose 1-P and 3-fluoro-glucose 6-P as the modification is equidistant from the C-l and C-6 phosphate. Consequently no prediction can be made on which of the two putative diphosphate species would dominate. Thus, the  ^V-ncox data of  the Cd -enzyme complexes appear to confirm the predictions 2+  based on both the inhibition data of this thesis and the suggestions of Ma and Ray (1980). Only a single diphosphate species is observed in the case of the 4-fluoro substrate analogue, whereas two species arising from the two alternate forms of the diphosphate complex are detected with the 2- and 3-fluoro analogues. Alternatively, the single diphosphate species observed by 31p-rimr (Rhyu et al., 1984, 1985a) may represent atimeaverage of the two diphosphate species. In the cases of the fluorinated substrates, the interconversion of the two putative diphosphate complexes may be slowed to such an extent that the two species are observable as separate signals. The single resonance observed in the 4-fluoro case may be due to signal coincidence. Obviously these interpretations are at best tentative and cannot be regarded as conclusive until the equilibrium amongst the central complex of each fluoro-substrate in the presence of C d has been determined. If it is determined that each fluoro-glucose 1,6-diphosphate does 2+  make up the majority of the central complexes in the presence of Cd , then the results of the 2+  170 present study will be heavily supportive of an exchange mechanism involving two different diphosphate binding modes. IV.B.2.4: Complexes of Difluorinated Inhibitors. The effect of L i on the binding of substrates to phosphoglucomutase was further studied +  using each of the difluorinated glucose 1-P analogues. Under the conditions of proton decoupling the l^F-nmr spectrum of 4,6-difluoro-glucose 1-P (4,6F2Glc 1-P) consists of two sharp peaks at 201.7 ppm (F-4) and 237.6 ppm (F-6) (Figure IV.13.A). When phosphoglucomutase is present in a 1:1 molar ratio with the inhibitor, the gated proton decoupled signalfromthe F-6 is broadened (AD 1/2 = 70 Hz) and shifted slightly downfield by 0.2 ppm to 237.4 ppm. Similarly, F-4 is broadened (AD 1/2 = 65 Hz) and is also shifted downfield by approximately 0.2 ppm (Figure IV.13.B). The small shift and only slight broadening of the signals suggest that the free ligand is in fast exchange with the enzymebound form. (The line width of enzyme-bound 6-fluoro-glucose 1-P which bindstightlyto enzyme and is in slow exchange with the free form was measured as 100 Hz under the same conditions, Table IV.4). The value of the Kd of 4,6-difluoro-glucose 1-P was determined as 730 \IM by inhibition of the substrate reaction (Table n.6). From this value and the concentrations of enzyme and ligand (both 1.0 mM), the fraction of bound ligand can be calculated as 44%. The fact that there are negligible changes in the chemical shift of both exchange averaged peaks suggests that either the Kd is in fact much higher than 730 \IM or that the chemical shifts of the enzyme-bound species are not very different from those free in solution. If the latter is the case, then it would be in sharp contrast to the change in chemical shift observed on the binding of 6fluoro-glucose 1-P to the phosphoenzyme (-14.2 ppm). Addition of Li to the binary enzyme-inhibitor complex to a concentration of 10 mM +  caused the ligand peaks to broaden and shift downfield. The total chemical shift from that of free ligand for F-4 is -7.0 ppm and for F-6 is -14.2 ppm (Figure IV.13.C). On addition of a second equivalent of 4,6-difluoro-glucose 1-P two new resonances appeared at the same chemical shifts as that of the ligand in the absence of protein (Figure IV.13.D). The observation of only a single  171  Chemical shift 8(ppm) from freon. Figure IV.13: Titration of phosphoenzyme with 4,6-difluoro-glucose 1-P andLi followed by F-nmr: (A) Spectrum of "free" 4,6F2Glc 1-P, decoupled; (B) 1.0 mM 19  Ep, 1.0 mM 4,6F2Glc 1-P, DNA decoupled; (C) 1.0 mM Ep, 1.0 mM 4,6F2Glc 1-P, 10 mM LiCl, 1.0 mM EDTA; (D) 1.0 mM Ep, 2.0 mM 4,6F2Glc 1-P, 10 mM LiCl, 1.0 mM EDTA.  +  172  pair of resonances from the bound ligand in the Ep.Li .4,6F2Glc 1-P complex (Figure IV.13.C, +  1:1 ligand/enzyme ratio) shows that the equilibrium constant for the bmding of the ligand has been reduced by at least two orders of magnitude so that the percentage of free ligand is less than approximately 10%. Concurrently, the exchange rate between free and enzyme-bound forms of the ligand has been reduced so that separate resonances of free and bound ligand are observed. Similar results were obtained when the phosphoenzyme was titrated with 2,6- and 3,6difluoro-glucose 1-P and L i (Figures IV.14 and rv.15). However, in the presence of enzyme +  (1:1 molar ratio, no Li ), the linewidths of the ligand peaks were rather narrower than those of +  the same complex with 4,6-difluoro-glucose 1-P. This is consistent with the Kd values of the inhibitors, measured by kinetic methods (Chapter H), being larger than that of 4,6-difluoroglucose 1-P. On addition of excess Li , downfield shifts of approximately 4-6 and 14-15 ppm +  were observed for the glucose ring and F-6 fluorines respectively. A summary of the chemical shifts, changes in chemical shift and linewidths measured from the spectra of Figures IV.13-15 is shown in Tables IV.7 and IV.8.  Chemical shift 8(ppm) Free  Ep.L  Ep.Li+.L  Shift Ep.Li .L +  F-X  F-6  F-X  F-6  F-X  F-6  F-X  4,6F2Glc 1-P 237.7  199.5  237.5  199.3  223.5  192.5  -14.2  -7.0  3,6F2Glc 1-P 237.7  201.5  237.6  201.2  222.9  197.7  -14.8  -3.8  2,6F2Glc 1-P 237.4  200.3  237.3  200.3  222.2  194.2  -15.2  -6.1  Ligand (L)  a  F-6  a  Chemical shift change from free ligand.  Table IV.7: Summary of^F-nmr  chemical shifts of binary and ternary complexes of  phosphoglucomutase and difluorinated glucose 1-P inhibitors. (L)  173  180  190  200  210  220  230  240  250  Chemical shift 8(ppm) from frcon. Figure rv.14: Titration of phosphoenzyme with 3,6-difluoro-glucose 1-P andLi followed by F-nmr. (A) Spectrum of "free" 3,6F2Glc 1-P, decoupled; (B) 1.0 mM Ep, 1.0 mM 3,6F2Glc 1-P; (C) 1.0 mM Ep, 1.0 mM 3,6F2Glc 1-P, 10 mM LiCl, 1.0 mM EDTA; (D) 1.0 mM Ep, 2.0 mM 3,6F2Glc 1-P, 10 mM LiCl, 1.0 mM EDTA. +  19  174  IS)  190  200 210 220 230 Chemical shift 8(ppm) from freon.  240  250  Figure IV.15: Titration of phosphoenzyme with 2,6-difluoro-glucose 1-P andLi followed by F-nmr: (A) Spectrum of "free" 2,6F2Glc 1-P, decoupled; (B) 1.0 mM Ep, 1.0 mM 2,6F2Glc 1-P; (C) 1.0 mM Ep, 1.0 mM 2,6F2Glc 1-P, 10 mM LiCl, 1.0 mM EDTA; (D) 1.0 mM Ep, 2.0 mM 2,6F2Glc 1-P, 10 mM LiCl, 1.0 mM EDTA. +  19  175  Linewidth » (AD 1/2 Hz) a  D  Ep.Li+.L  Ep.L  F-X  Ligand (L)  F-6  F-X  Ref.  F-6  4,6F2Glc 1-P  70  65  10  110  80  100  70  10  3,6F2Glc 1-P  30  20  10  160  25  125  20  15  2,6F2Glc 1-P  30  20  10  220  160  200  150  20  B  F-6  F-X  F  B  F  Ref.  a At 20°. The linewidths of the ligands in the absence of protein were all 10 Hz. 0  Table IV.8: Summary of  F  19  linewidths of the free and enzyme-bound species in spectra  These linewidths were measured from DNA decoupled spectra. L denotes ligand, B bound species, F free species and X, the ring fluorine.  of difluorinated glucose 1-P complexes.  A comparison of the ligand linewidths of the Ep.Li .3,6F Glc 1-P complex with those +  2  of the other ternary complexes shows that a much larger difference exists between enzyme-bound and free species. The linewidth of the free ligand in Ep.Li .3,6F2Glc 1-P is only slightly greater +  than that of the reference, suggesting that the ligand is in very slow exchange between enzymebound and free forms and that the strength of the binding of 3,6F2Glc 1-P is increased by the presence of L i to a greater degree than the other difluorinated inhibitors. This is an interesting +  phenomenon, but at present no explanation can be offerred. There is some doubt as to the extent by which L i increases the tenacity of ligand +  binding. Earlier kinetic studies have indicated that the binding of L i in place of the activating +  metal, Mg2+, increases the affinity of the enzyme for glucose 1-P by a factor of about 1000. Thus, the Kd of glucose 1-P in the Ep.Li .Glc 1-P complex is approximately 10 nM compared +  to 8 p.M in Ep.Mg .Glc 1-P or Ep.Glc 1-P O^ay etai, 1978). These suggestions were based 2+  on the observation that L i binds to the Ep.Glc 1-P complex 1000timesmore tenaciously than to +  Ep alone (Kd = 10 [iM and 10 mM respectively). However, more recently, Li-nmr studies have 7  shown that the equilibrium constant for dissociation of L i from Ep.Li .Glc 1-P is +  +  176  approximately 0.8 mM. Similar results were calculated for the dissociation constant from the Ep.Li .Glc 1,6-diP and Ep.Li .Glc 6-P complexes. Thus, L i appears to bind much less +  +  +  tightly under the conditions of the nmr experiment than previously estimated on the basis of L i inhibition of the initial velocity reaction. Consequently, it would be predicted that under the same conditions, the binding of glucose 1-P would be only 10timestighterin the presence of L i  +  instead of the previously estimated 1000 fold. However, the experimental observations made in this study support the earlier suggestion that bmding of L i to the enzyme increases the +  equihbrium constant for the binding of substrates by about 2-3 orders of magnitude. Each of the disubstituted inhibitors binds rather poorly to the Ep.Mg2+ complex, having Kd values in the range 0.7-2.5 mM (see Table II.6). These weakly binding species would be predicted to exchange rapidly between enzyme-bound and free forms as was noted previously. This appears to be the case as the bound ligand resonances are narrow and only slightly shifted (0-0.3 ppm) from that of the free ligand. However, on addition of an excess of L i the exchange process is +  greatly slowed down and separate free and enzyme-bound ligand peaks are observed. The chemical shift changes of F-6 of the difluorinated inhibitors which occurred on binding to the Ep.Li complex («-15 ppm, Table IV.7) are of a similar magnitude to that +  observed for the binding of 6-fluoro-glucose 1-P to Ep.Li (-15.9 ppm). The slight reduction in +  the size of the chemical shift changes of the difluorinated inhibitors compared to 6-fluoro-glucose 1-P could be due to the loss of enzyme-glucose ring interactions so that F-6 is not positioned quite so close to the enzymic phosphate, the electric field of which was earlier predicted to be the cause of the large chemical shift change. However, this suggestion is not supported by the differences in chemical shift change observed for F-6 of each difluorinated inhibitor. Thus, the chemical shift change of F-6 of 4,6-difluoro-glucose 1-P is 1.0 ppm less than that of 2,6difluoro-glucose 1-P which on the basis of inhibition constants should be bound more weakly, thereby reducing the interaction between F-6 and the enzyme. An alternative explanation which fits the observed data is as follows. Replacement of the C-4 and C-6 hydroxyls by fluorine reduces the interaction of the protein and the inhibitor in that region of the sugar so that F-6 has more mobility in the active site than if the C-2 and C-6 hydroxyls were replaced by fluorine. In  +  177  either event, the similarities between the changes in chemical shift of F-6 of all three inhibitors in the presence of L i show that the disubstituted inhibitors do interact with the enzyme in a similar +  fashion to 6-fluoro-glucose 1-P (and therefore by extrapolation, glucose 1-P) and that binding does not just involve recognition of the phosphate functional group. These observations are therefore in accord with those of the UV spectral study described in Chapter II. However, the very small changes in ligand chemical shifts which accompany bmding in the absence of L i  +  suggest that the sugar rings of the difluorinated inhibitors are not binding in the same mode as they do in the presence of Li . These results are consistent with the observations that almost no +  changes in chemical shifts of the ring fluorines occur on binding of the mono-fluorinated glucose 6-P substrates to the phosphoenzyme (Section IV.B.2.3). Thus, one is left with the possibility that the substrates (and inhibitors) bind to the enzyme in a different mode from that in the presence of Li . +  Since the chemical shift changes of F-6 of each difluorinated inhibitor compare well with that of 6-fluoro-glucose 1-P in the presence of Li , it would be expected that the chemical shift +  changes of each glucose ring fluorine (of the difluorinated inhibitor) should be similar to that of the corresponding monofluorinated glucose 1-P substrate bound to the Ep.Li complex. The +  changes in chemical shift of the glucose ring fluorine of each difluorinated inhibitor on binding to the Ep.Li"" complex are smaller and show a somewhat greater variation than those of F-6 (Table 1  IV.7). However, they do follow the trend observed for incorporation of a fluorine nucleus into a macromolecule and are therefore shifted downfield. The different chemical shift changes of the ring fluorine of each inhibitor suggest that each reporter nucleus is exposed to a different environment in the active site. Whether these differences are due to different hydrogen bonding interactions or other magnetic or electric field effects cannot be determined by the present work.  IV.B.2.5: Ternary Phosphoenzyme-Li-Fluoro-Substrate Complexes. +  In the presence of 10 mM Li , the activity of phosphoglucomutase is less than 2 x 10'8 +  times that of the Mg form of the enzyme. This slight activity is even less than that in the 2+  apparent absence of any activating metal ion (Ma & Ray, 1980). The turnover rate in the presence  178  of L i is so slow that the three enzyme bound intermediates, Ep.Li .Glc 6-P, Ep.Li .Glc 1-P +  +  +  and ED-Li .Glc 1,6-diP, can be maintained as separate entities for several hours at room +  temperature (Ma & Ray, 1980, Rhyu et al, 1984,1985a). This has allowed the individual complexes to be studied by 31p and Li-nmr (Rhyu et al, 1984,1985a). Since the relative rate 7  at which phosphoglucomutase isomerizes the fluorinated glucose 1-P substrates is only approximately 1-3% of that of the normal substrate, and assuming that activating metal ions can be reduced to the same level as was achieved by Rhyu et al, then the individual fluoro-glucose phosphate-phosphoenzyme complexes should be stable for up to one week. Experimental evidence was presented in Chapter II in support of the suggestion of LA. Rose (1987) that the interconversion of the two putative diphosphate complexes (Figure 1.3) occurs by a rotation around the C-3 hydroxyl. It was also suggested that this rotation causes a reversal of the interactions between the enzyme and the ligand, such that an amino acid interacting with the C-2 hydroxyl in one of the two diphosphate species, now interacts with C-4 hydroxyl in the other and vice versa.. In addition, Ma and Ray (1980) suggested that the interactions of the sugar ring of each of the two diphosphate species with the enzyme are the same as that of the corresponding monophosphate. For example, the C-2 hydroxyl of glucose 1P would be located in the same position in the active site as that of glucose 1,6-diP bound in the glucose 1-P mode. Therefore, according to Rose's suggestion, this same enzyme site would be occupied by the C-4 hydroxyl of both glucose 6-P and glucose 1,6-diP bound in the glucose 6-P mode. The l^F-nmr study of the chemical shifts of the Ep.Li .difluorinated glucose 1-P +  inhibitors (Section IV.B.2.4) showed that each sugar ring fluorine is subjected to a slightly different environment in the enzyme's active site. Thus the sensitivity of l^F-nmr chemical shifts to differences in environment may provide a means of determining the vahdity of the aforementioned hypothesis about the enzyme's mechanism. If the hypothesis was correct, then it would be expected (simplistically) that the change in chemical shift observed on bmding of 2fluoro-glucose 1-P to the phosphoenzyme would be the same as that caused by the binding of cc4-fluoro-glucose 6-P since the environments into which the fluorines are introduced would be the  179  same. Similarly, the chemical shift change of a-2-fluoro-glucose 6-P should be the same as that of 4-fluoro-glucose 1-P on binding to the phosphoenzyme. Conversely, the chemical shift change caused by the binding of 3-fluoro-glucose 1-P and a-3-fluoro-glucose 6-P should be the same since this is the site on the enzyme about which the reorientation of the diphosphate species takes place. In light of the comparatively weak binding of the fluoro-glucose phosphate substrate analogues, which would result in a rapid exchange between free and enzyme-bound forms, the Ep.Li complex was used in these experiments as it was shown that L i increases the tenacity of +  +  the enzyme for the substrates by at least two orders of magnitude. In addition to decreasing the ligand dissociation rate, the presence of L i should reduce the difficulty of mamtaining the +  complexes as separate entities with only minimal turnover. 1. 4-Fluoro-glucose phosphates. The proton decoupled l F-nmr spectrum of 4-fluoro-glucose 1-P (8 = 199.0 ppm) is 9  shown in Figure IV.16.A. When 1.0 equivalent of 4-fluoro-glucose 1-P was added to 1.0 equivalent of phosphoenzyme in the presence of a ten fold excess of L i the resonance was +  shifted downfield to 192.9 ppm, a shift of -6.1 ppm (Figure Pv\16.B). This complex can be reasonably assigned as Ep.Li .4FGlc 1-P. Over a period of approximately 10 hours the +  resonance at 192.9 ppm disappeared and two new resonances at 195.5 and 198.9 ppm gradually built up. This complex was quite stable and did not change over the following 24 hours (Figure IV.16.C). These resonances can be assigned as Ep.Li .a-4FGlc 6-P and Ep.Li .B-4FGlc 6-P +  +  (see below). The proton decoupled spectrum of 4-fluoro-glucose 6-P consists of two peaks due to the a and B-anomers at 198.3 and 200.2 ppm with relative intensities of 35 and 65% respectively (Figure IV.17.A). Addition of 1.0 equivalents of 4-fluoro-glucose 6-P to 1.0 equivalents of phosphoenzyme in the presence of 10 mM L i caused the appearance of two peaks at 195.6 and +  198.8 ppm in the same ratio of intensities (Figure IV.17.B). This complex was stable and no change was observed over a period of 24 hours. The two resonances can be assigned as Ep.Li .a-4FGlc 6-P and Ep.Li .B-4FGlc 6-P, respectively. It is most likely that the complex +  +  180  Figure IV. 16: Binding of 4-fluoro-glucose 1-P to the phosphoenzyme-Li complex followed by F-nmr: (A) Spectrum of "free" 4FGlc 1-P, decoupled; (B) 1.0 mM Ep, 1.0 mM 4FGlc 1-P, 10 mM LiCl, 1.0 mM EDTA; (C) same as (B) except after 10 hours. +  19  181  170  180  190  200  210  220  230  240  250  Chemical shift 8(ppm) from freon.  Figure IV.17: Binding of 4-fluoro-glucose 6-P to the phosphoenzyme-Li complex followed by F-nmr. (A) Spectrum of "free" 4FGlc 6-P, decoupled; (B) 1.0 mM Ep, 1.0 mM 4FGlc 6-P, 10 mM LiCl, 1.0 mM EDTA; (C) 1.0 mM Ep, 2.0 mM 4FGlc 6-P, 10 mM LiCl, 1.0 mM EDTA. +  19  182  having a chemical shift of 195.6 ppm is in fact the enzyme bound a-anomer (rather than the Banomer) since the ratio of intensities is very similar to that of the a and B-anomers free in solution (Figure IV. 17.A). Moreover, since both anomers would be expected to bind to the enzyme in a similar fashion, it would be unlikely for one resonance to be shifted 4.6 ppm downfield whilst the other is shifted 0.4 ppm upfield. The fact that the relative proportions of the a and B-anomers are no different from that free in solution and the equivalence of the two linewidths (ADI/2 ~ 125 Hz) shows that the dissociation constants for each anomer from the Ep.Li .4FGlc 6-P complex are approximately equal. This is in contrast to the results obtained +  for the binding of a and B-4-fluoro-glucose 6-P to the phosphoenzyme which indicated that the a-anomer binds slightlytighterthan the B-anomer (Section IV.B.2.3). This latter result would be predicted on the basis of the fact that the product of the isomerization reaction is the a-anomer (Lowry & Passonneau, 1969). Addition of a further 1.0 equivalent of 4-fluoro-glucose 6-P caused the appearance of a single new peak at 200.2 ppm which is assigned as free B-4-fluoroglucose 6-P. The resonance from a-4-fluoro-glucose 6-P appears at 198.3 ppm and is coincident with the enzyme-bound B-4-fluoro-glucose 6-P (Figure IV.17.C). The observation of two sets of peaks corresponding to enzyme-bound and free ligand again supports the conclusion gained from the investigation of the difluorinated inhibitors (Section IV.B.2.4) that L i does gready +  decrease the enzyme-substrate dissociation constant These results can be contrasted with those of the Ep.4FGlc 6-P complex where only two resonances, corresponding to the a and Banomers, were observed in the presence of an excess of ligand (Section IV.B.2.3). The observation that the spectrum of Figure IV.16.C is exactly the same as that of Figure IV.17.B supports the assignments made for these complexes. It is unlikely that both spectra represent two dephosphoenzyme-glucose 1,6-diP complexes since no change in the spectrum of Figure IV.17.B was observed over 24 hours. Moreover, the results of the investigation of the C d  2+  complexes suggests that only a single complex of this type would be observed. 2. 3-Fluoro-glucose phosphates. The proton decoupled l F-nmr spectrum of 3-fluoro-glucose 1-P (8 = 201.1 ppm) and 9  internal reference 6-fluoro-a-methyl glucoside is shown in Figure IV. 18.A. Addition of 1.0  183  equivalents of 3-fluoro-glucose 1-P to 1.0 equivalent of the Ep.Li complex (10 fold excess of +  Li ) caused the appearance of a single resonance at 197.4 ppm (Figure IV.18.B). This complex +  was stable for at least 24 hours. After this time a small aliquot was removed and subjected to the analysis procedure of Ray and Long (1976b) described in Section IV.B.2.3. Instead of quenching the enzyme with acid, the mixture was heated in a microcentrifuge tube at 100°C for 10 min. As L i was present in excess, a rapid quenching procedure was considered unnecessary +  since any turnover by phosphoglucomutase would still be extremely slow even at the elevated temperatures before the enzyme denatured. In the absence of a high concentration of salt, the coupling enzyme, glucose 6-phosphate dehydrogenase, reasonably efficiently oxidizes 3- and 4fluoro-glucose 6-P, each reaction taking approximately 30 min to be completed. The results of the analysis of the mixture showed that 3-fluoro-glucose 1-P comprised at least 90% of the glucose phosphates present. The conclusion can therefore be made that the complex observed in Figure IV. 18.B is that of Ep.Li+.3FGlc 1 -P. The proton decoupled spectrum of the 3-fluoro-glucose 6-P used in this section of the study is shown in Figure IV.19.A. The four resonances can be assigned as fi-3-fluoro-glucose 6-P, 195.7 ppm, f5-3-fluoro-glucose 1-P, 195.8 ppm, ct-3-fluoro-glucose 6-P, 200.6 and cc-3fluoro-glucose 1-P, 201.3 ppm. This mixture of 3-fluoro-glucose phosphates was used as it was found that the 3-fluoro-glucose 6-P prepared by enzymic phosphorylation of 3-fluoro-glucose had decomposed. It appears that this species is less stable than its 2- and 4-fluoro isomers and should be kept desiccated at -20°. Unfortunately, no more 3-fluoro-glucose was available to repeat the preparation; thus, 3-fluoro-glucose 6-P was prepared by the action of phosphoglucomutase on 3-fluoro-glucose 1-P. Consequently, a small amount of oc-3-fluoroglucose 1-P is present in the mixture as well as p-3-fluoro-glucose 1-P which was a contarninant in the 3-fluoro-glucose 1-P starting material. The spectrum produced by a 1:1 molar ratio of 3fluoro-glucose 6-P to Ep.Li (10 fold excess of Li ) is shown in Figure IV.19.B. The poor +  +  quality of the spectrum probably reflects the number of different species present. The sharper resonance at 195.8 ppm can be assigned as P-3-fluoro-glucose 1-P which is non-reacting and has a very low affinity for the enzyme. An aliquot of this sample was analyzed for each of the 3-  184  B.  170  190  180  200  210  220  230  240  250  Chemical shift 5(ppm) from freon. Figure IV.18: Binding of 3-fluoro-glucose 1-P to the phosphoenzyme-Li+ complex followed by F-nmr: (A) Spectrum of "free" 3FGlc 1-P, decoupled; (B) 1.0 mM Ep, 1.0 mM 3FGlc 1-P, 10 mM LiCl, 1.0 mM EDTA. 19  185  Chemical shift 5(ppm) from freon. Figure IV. 19: Binding of 3-fluoro-glucose 6-P to the phosphoenzyme-Li complex followed by F-nmr: (A) Spectrum of "free" 3FGlc 6-P, decoupled; (B) 1.0 mM Ep, 1.0 mM 3FGlc 6-P, 10 mM LiCl, 1.0 mM EDTA. +  19  186  fluoro-glucose phosphates as previously described. 3-Fluoro-glucose 6-P comprised at least 90% of the phosphate esters present. The two broad peaks at 197.0 and 201.1 ppm can be assigned as Ep.Li+P-3FGlc 6-P and Ep.Li .a-3FGlc 6-P respectively. The same rationale as +  that used for assigning the a and P-4-fluoro-glucose 1-P complexes was used to assign the anomeric composition of these bound species. Any bound a-3-fluoro-glucose 1-P would be coincident with the peak at 197.0 ppm. 3. 2-Fluoro-glucose phosphates. The proton decoupled spectra of 2-fluoro-glucose 1-P and 2-fluoro-glucose 6-P are shown in Figures IV.20.A and IV.21.A, respectively. As the three signals are almost coincident , the spectra of the enzyme-ligand complexes were measured with gated proton decoupling (DNA) in order to attempt to resolve the resonances. The disadvantages of this is the greater acquisition time required and the previously noted (Section IV.B.2.1) decrease in signal intensity. The spectra obtained ontitrationof the Ep.Li complex (10 fold excess of Li ) with 2+  +  fluoro-glucose 1-P are shown in Figure 1Y.20.B and C. The broad resonance at 198.0 ppm (Figure IV.20.B, Aui/2 = 70 Hz) can be assigned as Ep.Li .2FGlc 1-P and that at 200.1 ppm +  (Figure IV.20.C) as excess free 2-fluoro-glucose 1-P (At) 1/2 = 25 Hz). The smaller sharp peak at 199.7 ppm is that of P-2-fluoro-glucose 1-P. The narrow linewidth (Aui/2 = 5 Hz) of this peak is the same as that of the reference and shows the lack of any affinity of phosphoglucomutase for any |3-glucose 1-phosphates. The spectra obtained by titration of the Ep.Li complex with 2-fluoro-glucose 6-P are +  shown in Figure IV.21. The broad peak at 198.0 ppm (ADI/2 = 70 Hz) can be assigned as a coincidence of the resonances from the Ep.Li .oc-2FGlc 6-P and Ep.Li+P-2FGlc 6-P. Addition +  of a second equivalent of 2-fluoro-glucose 6-P causes the appearance of two sharper peaks having the same chemical shift as that of free a and p-2-fluoro-glucose 6-P (AD 1/2 = 20 Hz). As the two anomers of 2-fluoro-glucose 6-P are easily resolved in Figure IV.21.C, it can be concluded that the resonance at 200.1 ppm in Figure IV.20.C is indeed excess 2-fluoro-glucose 1-P and not a mixture of a and P-2-fluoro-glucose 6-P which would have appeared if turnover had occurred. These nmr samples could not be analyzed by the coupled assay system described  187  170  180  190  200  210  220  230  240  250  Chemical shift 8(ppm) from freon.  Figure IV.20: Binding of 2-fluoro-glucose 1-P to the phosphoenzyme-Li complex followed by F-nmr: All spectra are DNA decoupled.(A) Spectrum of "free" 2FGlc 1P; (B) 1.0 mM Ep, 1.0 mM 2FGlc 1-P, 10 mM LiCl, 1.0 mM EDTA; (C) 1.0 mM Ep, 2.0 mM 2FGlc 1-P, 10 mM LiCl, 1.0 mM EDTA. +  19  188  Chemical shift 5(ppm) from frcon.  Figure IV.21: Binding of 2-fluoro-glucose 6-P to the phosphoenzyme-Li complex followed by F-nmr: All spectra are DNA decoupled. (A) Spectrum of "free" 2FGlc 6P; (B) 1.0 mM Ep, 1.0 mM 2FGlc 6-P, 10 mM LiCl, 1.0 mM EDTA; (C) 1.0 mM Ep, 2.0 mM 2FGlc 6-P, 10 mM LiCl, 1.0 mM EDTA. +  19  189  previously since the activity of yeast glucose 6-phosphate dehydrogenase for 2-fluoro-glucose 6P is about 10 fold less than that of the 3- or 4-fluoro isomers (Bessel & Thomas, 1973). A summary of the chemical shifts and the changes in chemical shifts of each of the species observed in the Li -liganded enzyme-substrate complexes is presented in Table IV.9. +  Chemical shift 8(ppm)  a  Ligand (L)  Free  Ep.Li+.L  Shift Ep.Li+.La  a-2FGlc 1-P  200.2  198.0  -2.2  6-2FGlc 1-P  199.7  -  -  a-2FGlc 6-P  200.3  197.9  -2.4  B-2FGlc 6-P  200.1  197.9  -2.2  a-3FGlc 1-P  201.2  197.4  -3.8  a-3FGlc 6-P  200.6  201.1  +0.5  P-3FG1C 6-P  195.7  197.0  +1.3  a-4FGlc 1-P  199.0  192.9  -6.1  a-4FGlc 6-P  198.3  195.6  -2.7  B-4FG1C 6-P  200.2  198.8  -1.4  Chemical shift changefromfree ligand.  Table IV.9: Summary of^F-nmr  chemical shifts of free and ligandedfluoro-glucose 1-  phosphate analogues in the ternary phosphoglucomutase-Li -ligand complexes. L +  indicates the fluoro substrate.  It was suggested at the end of Section IV.B.2.4 that the chemical shift changes of the difluorinated inhibitors on binding to the Ep.Li complex should be similar to those of the +  mono-fluorinated substrates and inhibitor. It was found that the change in the chemical shift of  190 6-fluoro-glucose 1-P (-15.9 ppm) was very similar to that of F-6 of all three disubstituted inhibitors (-14.2 to -15.2 ppm). The present results show that the changes in chemical shift of 4and 3-fluoro-glucose 1-P on binding to Ep.Li , (-6.1 and -3.8 ppm, respectively, Table IV.9) +  compare favorably with those of the difluorinated glucose 1-P species (-7.0 and -3.8 ppm, Table IV.7). However, the chemical shift changes of 2-fluoro-glucose 1-P and 2,6-difluoro-glucose 1P(-2.2 and -6.1 ppm, respectively), do not suggest that the environments of the two fluorine reporter nuclei are the same. The present results can also be used to test the vahdity of the proposals made at the beginning of this section. It was suggested that an exchange mechanism involving a reorientation of glucose 1,6-diP about the C-3 hydroxyl would result in a predictable pattern of chemical shift changes for each glucose ringfluorine.Thus, the chemical shift change of 2-fluoro-glucose 1-P on bmding to the Ep.Li complex should be similar to that of a-4-fluoro-glucose 6-P as should +  be the chemical shift changes of cc-2-fluoro-glucose 6-P and 4-fluoro-glucose 1-P. The chemical shift changes of cc-3-fluoro-glucose 6-P and 3-fluoro-glucose 1-P were also predicted to be similar. However, inspection of Table IV.9 shows that this is not the case. The chemical shift changes of 4-fluoro-glucose 1-P, (-6.1 ppm) and a-2-fluoro-glucose 6-P, (-2.4 ppm) are not similar, as are neither those of 3-fluoro-glucose 1-P, (-3.8 ppm) and a-3-fluoro-glucose 6-P (+0.5 ppm). The only pair of results that would fit the proposals made at the beginning of this section are those of a-4-fluoro-glucose 6-P, (-2.7 ppm) and 2-fluoro-glucose 1-P, (-2.2 ppm). The alternative to this particular exchange mechanism, the minimal motion mechanism, does not seem to fare any better from these results. If a simple form of this mechanism were to operate, there would be only a single'glucose ring bmding site, but two glucose monophosphate phosphate binding sites. (Figure 1.2). Thus, each hydroxyl of glucose 1-phosphate would interact with the same amino acids as the corresponding hydroxyl of glucose 6-P. However, the data obtained from the present investigation do not support this mechanism. For example, the change in chemical shift on binding of 4-fluoro-glucose 1-P to Ep.Li , (-6.1 ppm) is not similar +  to that of ct-4-fluoro-glucose 6-P, (-2.7 ppm). Nor is that of 3-fluoro-glucose 1-P, (-3.8 ppm) similar to that of a-3-fluoro-glucose 6-P, (+0.5 ppm).  191 The present results are also inconsistent with the suggestion of WJ. Ray (1986) that interconversion of the two putative dephosphoenzyme-glucose 1,6-diP complexes (exchange mechanism) involves a swapping of the interactions of the C-3 and C-4 hydroxyls. The inability of the results obtained from this study to provide evidence consistent with either of the two proposed mechanisms leaves one to consider a number of alternatives. 1. Since the lithium-enzyme is completely inactive, the interactions between the enzyme and substrate in the inactive Ep.Li.substrate complex may bear little resemblance to those of the +  active Mg form.This is suggested by the fact that the binding of L i causes the enzyme to 2+  +  bind substrates much more tenaciously than does binding of Mg . Thus binding of L i must 2+  +  cause some conformational change which increases the strength of the interactions of some protein side chains with either the glucose hydroxyls and/or the substrate phosphate group. The changes in chemical shift of fluorinated glucose phosphate/enzyme complexes upon binding of L i have been measured in two cases. The chemical shift of 6-fluoro-glucose 1-P was +  shifted 1.5 ppm downfield whereas that of a-glucosyl fluoride 6-P was shifted 1.5 ppm upfield. These changes were considered fairly minor compared to the large shifts of -14.2 and -19.7 ppm, respectively, which occurred on binding to the apoenzyme. However, it is not clear whether the changes in chemical shift of each glucose monophosphate ring fluorine, caused by L i binding, are much greater or smaller than those observed for the C-l or C-6 fluorines. The +  spectra of the Ep.fluoro-glucose 6-P complexes may suggest that only small changes of chemical shifts occur in the absence of L i (Section IV.B.2.3). Inspection of these chemical shifts (Table +  IV.6) show that the exchange averaged resonances are only shifted -0.1 to -0.5 ppm from that of the free species. As in the difluorinated inhibitor cases, this suggests that either the chemical shift of the bound species is not very different from that of the free ligand or that the enzyme-substrate dissociation constants are larger than previously thought In contrast, the study of Ma and Ray (1980) showed that there was no or little difference in the UV spectra of the enzyme/substrate complex, either completely demetallated, or in the presence of Li . These results suggested that +  the enzyme-ligand interactions are not changed to any great degree by the bmcting of Li . +  However, this technique is often considered to be very non-specific and insensitive.  192 There are two alternatives to the use of L i in order to deactivate the enzyme. The first of +  these is to use demetallated enzyme, since the metal free enzyme has an activity 10" times that of 7  the Mg formand is expected to bind substrates in the same manner as the Mg2+ form (Ray & 2+  Peck, 1972). Attempts were made to measure the l^F-nmr spectra of each binary Ep.fluorosubstrate complex but were unsuccessful as it appeared that turnover had occurred (Section IV.B.2.3). Except in the cases of substitution of the "acceptor" hydroxyls, the replacement of a substrate hydroxyl group causes a large decrease in affinity for the enzyme, resulting in a rapid exchange between enzyme-bound and free forms. Thus, the use of the Li -enzyme complex is +  also advantageous in this respect as it increases the enzyme's affinity for the ligand and slows down the chemical exchange. In the absence of Li , the detenmnation of the limiting chemical +  shift of the fully bound species would require atitrationof the enzyme with the fluorinated substrate. Plotting the concentration of ligand versus the reciprocal of the change in chemical shift allows the determination of the chemical shift in the absence of exchange (Dwek, 1973). Obviously, this is a much more difficult procedure than in the slow exchange regime, especially as the system is so extremely sensitive to any contamination by metal ions. The second alternative to the use of L i to inactivate the enzyme is that of disubstituted +  inhibitors in which the acceptor hydroxyl has been removed.As shown in Section IV.B.2.3, this also results in a rapid exchange of ligand between enzyme-bound and free forms. However, the  titration to determine the chemical shift in the absence of exchange would be technically simple as any slight contamination by metal would cause no problem since the possibility of turnover is removed. The disubstituted inhibitors analogous to glucose 6-phosphate, in which a glucose ring hydroxyl is replaced by fluorine and the C-l hydroxyl is replaced by hydrogen, could be used in the same manner as were the disubstituted glucose 1-P analogues were (Section rV.B.2.4) or in the absence of Li . These inhibitors could possibly be prepared by the action of hexokinase on +  the corresponding free sugar, although the reaction would certainly be very slow. Alternatively, they could be prepared by reaction of the free sugar with a selective phosphorylating agent such  193  as dibenzyl phosphorochloridate. A difluorinated glucose 6-P analogue based on a-glucosyl fluoride would be too labile for general use. 2. The possibility must be considered that the discrepancies between the results obtained and those predicted at the beginning of this section are due to the wrong assignment of the resonances. It would appear that each complex is correctly assigned, except perhaps those of the 2- fluoro analogues in which it was not possible to use the enzymic analysis to determine the identity of the bound species. This may explain the discrepancy observed between the chemical shift change of 2,6-difluoro-glucose 1-P, (-6.1 ppm) and 2-fluoro-glucose 1-P, (-2.2 ppm) on binding to Ep.Li . +  3. The fact that l^F chemical shifts are very sensitive to changes in environment is well known. However, the actual basis of these chemical shift changes in macromolecular systems is not well understood. It may be that l^F chemical shifts are so sensitive to environmental effects that any small difference in orientation or distance of the nucleus with respect to the protein could manifest itself in a large change in chemical shift. This large angle and/or distance dependence may be such that two complexes could be considered very similar, but still show considerable differences in their l^F chemical shifts. This could be an important factor in light of the l/r^ dependence on Van de Waals interactions which are considered to be important in 1 F chemical 9  shifts. The unknown orientation dependence of fluorine chemical shifts could also be of importance in this study. It was considered that the interactions of each fluorinated glucose 1-P substrate with the protein would be opposite to those of each fluorinated glucose 6-P (the Rose mechanism, detailed earlier in this section). However, the interconversion of the two putative diphosphate species must result in a different angle of interaction between the protein side chains and each fluorine nucleus (of the two diphosphate species). Thus, it may be naive to expect that the chemical shift changes of two fluorine atoms which occupy the same site would be similar since the angles of interaction with the protein side chains must be different. The possibility that this is the case is suggested by the observation of two resonances in the spectrum produced from 3- fluoro-glucose phosphate and the Ep.Cd complex (Section IV.B.2.3). These two bound 2+  resonances were tentatively assigned as 3-fluoro-glucose 1,6-diP bound to the  194 dephosphoenzyme in two different modes. The Rose mechanism suggests that both of these fluorine nuclei would occupy the same site and therefore should have the same chemical shift. However it does not appear that this is the case (199.2 and 200.9 ppm). These suggestions could lead to the proposal that the data obtained in this study do not rule out either of the proposed exchange mechanisms and that the differences in the chemical shifts observed are due to relatively minor differences in the environments of the reporter nuclei. The data do, however, appear to be less consistent with a minimal motion type of mechanism since the binding modes of each substrate would be essentially identical.  IV.B.3: Summary and Implications for the Mechanism. The results of this section have shown that l^p-nmr f the complexes of fluorinated 0  substrate analogues is a useful tool for the study of this enzyme. Perhaps most interesting are the extremely large changes in chemical shift that accompany the binding of 6-fluoro-glucose 1-P and a-glucosyl fluoride 6-P to the phosphoenzyme (-14.2 and -19.7 ppm, respectively). These reporter nuclei are most likely situated right in the heart of the active site and these large shifts probably reflect this abnormal environment. As such, these analogues are extremely sensitive probes of the active site. This is exemplified by the vasdy different effects that the activating metal ions, Mg2+ and C d have on the spectra, compared to that of the non-activating metal 2+  ion, Li . It is conceivable that these differences are related to the different activities elicited by +  these metal ions and therefore this aspect deserves attention in future studies. Two apparently enzyme-bound species were observed in the l^F-nmr spectra of the complexes formed by reaction of the phosphoenzyme, C d and 2- and 3-fluoro-glucose 2+  phosphates. In contrast only one was observed in the same complex involving 4-fluoro-glucose phosphates. If the dominant central complex species of the fluorinated substrates and the Cd 2+  enzyme is the diphosphate, as it is with the native substrates, then these results would be consistent with an exchange type of mechanism in which there are two binding modes for glucose 1,6-diP. An argument based on enzyme-inhibitor dissociation constants (Chapter IT) was  195  presented to rationalize the number of fluoro-glucose 1,6-diphosphate species observed in each case. The series of spectra obtained from the complexes of the difluorinated analogues and phosphoenzyme (Section IV.B.2.4) provided evidence that L i does indeed gready increase the +  tenacity with which phosphoglucomutase binds its substrates.The correlation between the chemical shift changes of the Ep.6FGlc 1-P complex and those of the difluorinated inhibitors (F6) suggests that, at least in the L i ternary complex, these inhibitors do bind to the +  phosphoenzyme in the same mode as their parent substrates (with the possible exception of the C-2 position). The main aim of this investigation was, however, to probe the environments of the "nonacceptor" glucose ring hydroxyls and hopefully to obtain data consistent with a suggested mechanism. The results of that investigation (Section IV.B.2.5) are consistent with neither a simple concept of the minimal motion mechanism nor the exchange mechanisms proposed by WJ. Ray and LA. Rose. The results of this study are, however, not inconsistent with an exchange mechanism in which there are two different glucose ring binding sites, but only a single glucose monophosphate phosphate binding site (as required by the results of Ray et al., (1973)). In this variation of the exchange mechanism, two different dephosphoenzyme-glucose 1,6-diP complexes would exist, each having the same glucose ring-protein interactions as their corresponding phosphoenzyme-monophosphate complexes. However, since there are two different glucose bmding sites, one for the glucose ring in the glucose 1-P mode and the other for the glucose ring in the glucose 6-P mode, the environments of the reporter fluorines of each fluoro substrate (glucose 1-P and glucose 6-P type) would be different. This mechanism is supported by the nmr evidence of this study as each fluorine underwent a different change in chemical shift and therefore occupies a different environment A scheme representing this concept and showing the interactions between each substrate glucose ring and the enzyme is the same as that of the original concept of the exchange mechanism (Figure 1.3).  196 IV.B.4: F-iH Nuclear Overhauser Effects. 19  The observation of an noe between two nuclei is often good evidence for their close proximity. This technique has been used recently in enzymic systems to identify the protein residues which lie in the proximity of a fluorinated probe (Hammond, 1984, Jones et al., 1987). A l F-lH noe cannot identify the proximal protein residues per se. It can however provide a 9  spectral footprint, which, if lying in a particular characteristic spectral region, may enable the identification of the residue(s) causing the effect The identities of some of the amino acids close to the active site of phosphoglucomutase have been determined from the enzyme's primary sequence and the X-ray crystal structure. However, those involved in the actual catalytic process have not been identified. This section describes an attempt to identify the protein residues close to the fluorine nucleus of the Ep.6FGlc 1-P complex by means of a l F-lH noe. 9  The mechanism of phosphoryl transfer in phosphoglucomutase most likely requires the presence of two amino acid side chains which act as general base and acid catalysts in order to increase the nucleophilicity of the attacking hydroxyl and to protonate the alkoxy leaving group as shown in Figure IV.22.  Figure IV.22: Probable mechanism of phosphoryl transfer showing participation acid and base catalysts.  from  197  Thus, the mechanism requires that an amino acid remove the proton from the substrate acceptor hydroxyl at some stage in the phosphorylation step and then donate it back during the reverse reaction. In consequence, this protein side chain must lie in close proximity to the 'acceptor' hydroxyl. Presumably, there is also a similar side chain in the proximity of the serine-116 hydroxyl to perform the same task. The identities of these amino acids are unknown. It is likely that the fluorine atom of 6-fluoro-glucose 1-P in the Ep.6FGlc 1-P complex occupies the same site in the enzyme's active site as the acceptor hydroxyl of glucose 1-P and therefore it is probable that this fluorine reporter is also in the proximity of the aforementioned base catalyst. The observation of a  I^F-IH  (protein) noe could help in the identification of this base catalyst  In addition to the acid and base catalytic groups in the active site, the presence of a histidine residue has been demonstrated by lH-nmr. The pHtitrationof the protein histidine CgH residues of the ED-Cd^+.Glc 1,6-diP and Eo.Li .Glc 1,6-diP complexes showed that the +  chemical shift of one residue did not change whilst the remainder titrated normally. This contrasted with the Ep and Ep.Cd^"*" complexes in which all the histidine residues titrated normally (Rhyu et al, 1985b). This and other evidence suggested that the non-titrating histidine becomes coordinated to the metal ion* replacing the departing enzymic phosphate group on the metal on formation of the ED.M.G1C 1,6-diP complex ( M = Cd^+  o r  Li+). The metal ion, the  enzymic phosphate and the "acceptor" fluorine of 6-fluoro-glucose 1-P are all expected to he in close proximity in the active site. Thus, the observation of an noe between a histidine Ce-H residue, which lies in a characteristic *H spectral region, and the fluorine would be further evidence for the involvement of this amino acid in the aforementioned process.  198  IV.B.4.1: Investigation of a Ligand-Enzyme ! H - {  1 9  F}  NOE.  noe can be measured in two different ways. The first more direct approach is to observe protons whilst irradiating fluorine. Any noe induced can then be measured by subtraction of the spectrum in the absence of decoupling from that in its presence. The second, less direct method, which is discussed in Section IV.B.4.2, involves observation of fluorine whilst selectively decoupling protons. The Bruker HXS-270 spectrometer was modified to allow *H observation through the *H decoupling coils and 19F decoupling through the ^ observation coils of the high resolution 5 mm 19F probe. These modifications are detailed in the experimental procedures section. Control experiments with fluorine-containing small molecules showed that the *H spectra obtained through the decoupling coils were of a similar sensitivity and resolution to that obtained with a high resolution dedicated 5 mm *H probe. These controls also showed that complete decoupling of l^F-lH spin-spin coupling could be achieved by gated 19F decoupling. The external 1F-frequency synthesizer was calibrated using a sample of 6-fluoro-glucose 1-P in 9  D2O. Using a decoupling bandwidth of 100 Hz, the frequency of 6-fluoro-glucose 1-P (8 = 237.4 ppm) was determined as 253.99630 MHz. By varying the decoupling power, a level was determined that was sufficient to decouple completely the l?F-lH spin-spin coupling to H-6 and H-5 with a contact time of 50 ms and a decoupling band width of 1000 Hz. A sample of 1.0 mM Ep.6FGlc 1-P (1:1 molar ratio) was prepared exactly as for the 1 F experiments except that the 9  protein was dissolved in 99.8% D20and no internal reference was added. The 1 F spectrum of 9  this complex showed a single resonance at 223.2 ppm. The lH-{ 1F} noe was measured at 20° 9  using the following pulse sequence. After an initial delay of 5.0 s, the l^F decoupler was turned on for 200 ms. After a short delay of 100 (is, a 90° *H sampling pulse (18 \is) was applied and the free induction decay acquired for 1.02 s with a 4000 Hz spectral width, followed by a 0.5 s delay. This sequence was repeated for a block of 16 transients during whichtimethe l^Fdecoupler was on resonance for enzyme-bound 6-fluoro-glucose 1-P (253.9999 MHz) followed by a block of 16 transients in which the l^F saturation was turned off. This sequence was continued 200 times, the two blocks (decoupler on and off) being stored in two files on the  199 computer disk. The decoupler was turned off for the control spectrum, rather than being shifted off-resonance which would have been more desirable, because the frequency of the external frequency synthesizer could not be controlled by the computer. The small block size of 16 transients was necessary because of instrument instability which resulted in a gradual loss of sensitivity over several hours. The *H spectrum of the enzyme complex in the absence of decoupling is shown in Figure IV.23.A. The expansion of the low field region shows some of the histidine Ce-H protons (Figure IV.23.B). The noe was observed by subtraction of the fid in the absence of ^  decoupling from that in its presence, followed by fourier transformation. The  resulting spectrum was featureless, apart from two small peaks at 4.5 and 3.5 ppm arising from residual H2O and Tris buffer, and it showed no detectable noe. Further experiments were conducted in which the decouplingtimeand the irradiating power were varied. In no case was an noe detected, even with the H-5 and H-6 protons of the ligand which are close enough to the fluorine to give an observable noe in the absence of protein. The viability of the system was tested with a sample of 6-fluoro-glucose 1-P in the absence of protein. A small positive *H{l^F} noe was detected to the H-6, H-5 and H-4 protons. Spin-spin coupling between the irradiated fluorine and the H-5 and H-6 protons resulted in one half of each multiplet being inverted. The triplet of H-4 had a positive intensity as there is no coupling between the fluorine and this proton. The absence of a detectable lH-{ l^F} noe in a similar macromolecular system has been previously observed and investigated theoretically by Hull and Sykes (1975b). A full theoretical description of the reasons why no H-{ F) noe is observed in such systems is beyond the l  l9  scope of this thesis. However, it can be noted that the absence of a detectable noe is due to the similarity of the *H and l^p Tl values which are (ktermined by their correlationtimesand distances to neighboring dipoles. This situation occurs in large macromolecules (large Tc) in which the protons neighboring the fluorine are themselves surrounded closely by a large number of protons. Irradiation of the fluorine perturbs the spins of the proximal protons, but this perturbation is rapidly distributed amongst the remainder of the protons in the protein by spin diffusion and is diluted to such an extent that no effect is observed. A lH-{ * F} noe may be 9  200  A.  i 12  I  I  10  ,  ,  8.6  8.4  8  .  i i 4 2 Chemical shift 5(ppm) from TMS.  i 0  6  —I  8.2  1  8.0  1  7.8  r-  -2  -T-  7.6  Chemical shift 5(ppm) from TMS.  Figure IV.23:1 H-nmr spectrum ofphosphoglucomutase. The solution contained 1.0 mM phosphoglucomutase, 1.0 mM 6FGlc 1-P and 20 mM Tris-HCl pH 7.5 in 99.8% D 0. (A) Full spectrum; (B) Expansion of the low field region showing histidine Ce-H resonances. Measured at 400 MHz. 2  201 observable if the protons (dipolar) coupled with the fluorine are isolated from the body of the protein, such that spin diffusion cannot occur, or at least occurs very slowly. Hull and Sykes derived the following equation to estimate the *H-{ 1F} noe in 9  macromolecular systems, where N F and Ni are the number of fluorine and hydrogen atoms in the system, yis the gyromagnetic ratio of the nucleus and /H(F) is the fractional enhancement of the proton signal. /  H  (  F  )  . •» N P /  T,H  For the broad envelope observed in the *H spectrum of phosphoglucomutase the experimental Tl is 0.6 s, which is close to that of the Ti of the bound fluorine, 0.2 s. Since the number of protons in a protein of molecular weight 61600 will be several thousand and the gyromagnetic ratios of *H and l^F are similar, the calculated noe is negligible. The observation of a *H-{1F} 9  noe between the enzyme and fluorinated ligand in both the acyl carrier protein system (Jones et al., 1987) and chymotrypsin (Hammond, 1984) indicates that spin diffusion is a much less important phenomenon in these smaller proteins, the relative molecular weights of which are 8800 and 24500, respectively.  IV.B.4.2: Investigation of a Ligand-Enzyme F-{ JHINOE. 19  It was noted in Section IV.B.2.1 that broad band continuous *H decoupling of the Ep.6FGlc 1-P complex resulted in the total loss of the ^ F signal from the bound ligand. In light 9  of the unsuccessful attempt to observe a *H-{ l^F} noe, the alternate experiment was attempted in which selective *H decoupling was applied whilst 19F was observed. Experiments were conducted in which various *H frequencies were selectively saturated using a range of contact times and the noe observed. The results are shown for irradiation in the region of the H-6 ligand protons, 4.7 ppm and for irradiation at 8.1 ppm which is the chemical shift of one of the protein's histidine Ce-H signals (Figure IV.24).  202  0 f*"^ i 0  |—i 100  1  200  1—i  1  300  1  400  1— 500  Irradiationtime(ms) Figure IV.24:  Time dependence of^F-{^H} noe development at two  frequencies in the Ep.6FGlc 1-P complex.  decoupling  The *H decoupling frequencies were: n 4.7  ppm; 0 8.1 ppm.  It has not been established that the chemical shifts of the H-6 protons of the enzyme-bound 6fluoro-glucose 1-P are the same as that of the free ligand. However, since proton chemical shift changes on binding to macromolecules are usually less than 0.5 ppm (Dwek, 1973) and since the frequency width was approximately 200 Hz, it is most probable that the H-6 protons of the bound ligand were saturated. The graph shows that a large negative noe is produced on *H irradiation at 4.7 ppm. The noe develops rapidly, a contacttimeof 50 ms producing 50% of the maximal effect. In contrast, irradiation at 8.1 ppm causes a smaller noe that develops more slowly. Thus, a contact time of 50 ms produces an noe of only 10% of the maximal effect observed after a contacttimeof 400 ms. Plots such as these are characteristic of a direct and indirect noe, respectively (e.g., Rosevear et al., 1987, Jardetzky & Roberts, 1981). The rapid development of the noe at 4.7 ppm indicates that the effect is direct and not transferred as a result of spin diffusion. This would be expected since the H-6 protons are by definition very close to the fluorine nucleus. In contrast it would appear that the histidine residue at 8.1 ppm is not sufficiendy close (i.e., within 5 A ) to the fluorine nucleus to produce a direct noe. The  203  experiment was repeated at 7.8 and 8.4 ppm which are the centers of the other three sets of histidine Ce-H resonances visible in the 400 MHz *H spectrum of the protein. Only a very small negative noe was detected at 7.8 ppm whereas an noe of approximately the same magnitude as that at 8.1 ppm was detected at 8.4 ppm. Of the 10 histidine residues present in phosphoglucomutase (Ray et al, 1983), at least 6 were observed in the *H nmr spectrum at 400 MHz and were clustered at approximately 8.4, 8.1 and 7.8 ppm. The noe results suggest that none of the above Ce-H histidine protons are close enough to participate in a direct noe with the bound ligand. The possibility cannot be ruled out that another histidine residue having a different chemical shift from those noted above and therefore not irradiated is in the proximity of the ligand fluorine and would show an noe. A series of l F-{ *H} noe experiments was carried out on the same Ep.6FGlc 1-P 9  complex at different decoupling frequencies with a contact time of 50 msec. On the basis of the above results this short contact time should reduce the effects of spin diffusion and perhaps allow some "footprints" of neighboring protons to be obtained. The same pulse sequence as that used in the lH-{ 1F} noe experiments was used except that the decoupler frequency was computer 9  controlled. Blocks of 200 transients at 25 decoupling frequencies (0 - 8.5 ppm) were repeated 10 times to give a total of 2000 scans at each decoupling frequency. The frequency width of the decoupling pulse was approximately 100 Hz as determined by control experiments on 6-fluoroglucose 1-P in the absence of protein. A plot of decoupling frequency versus peak height was constructedfromthe data to enable the determination of the chemical shift of any protein residues giving an noe. The results showed that an noe of approximately 40% existed across the whole proton spectrum with the exceptions of the regions 5-6 ppm and 7.5 - 8.5 ppm. Comparison with the *H spectrum of phosphoglucomutase (Figure 1Y.23.A) shows that few resonances are present in the two regions. This almost totally non-specific noe does not allow any information on groups proximal to fluorine to be obtained and once again indicates the importance of spin diffusion in this system. The effects of correlation time and internuclear distance on l^F-f *H} noe in macromolecular systems have been investigated (Gerig, 1977). The results of theoretical  204 calculations showed that for a protein up to a molecular weight of about 20000, a selective noe should be observable. For larger macromolecules, the observed effect would depend on the number of interacting nuclei and their stereochemical relationship. These limitations were derived for a system in which the reporter nucleus and its immediate environment have the same correlation time as that of the macromolecule. Any rotations or motions of the fluorine reporter, or the nuclei with which it directly interacts, that are more rapid than the overall tumblingtimeof the macromolecule will operate to give rise to a more specific noe. Thus, if there are few nuclei interacting directly with the fluorine and they are isolated from the remainder of the protein, and/or the fluorine and its immediate environment have a smaller correlation time than that of the macromolecule, then a selective  1H} noe may be observed. However, this does not appear  to be the case in this system, this same conclusion being obtained from the *H-{ l^F} noe experiments.  IV.C: EXPERIMENTAL PROCEDURES.  IV.C.1: General. Pyridinium poly(hydrogenfluoride)was obtained from Aldrich Chemical Co. Magnesium chloride was obtained from Fisher Scientific and was of "Certified ACS" grade. Cadmium acetate was from MaUinckrodt and was Analar grade and the hthium chloride was obtained from Johnson Mattey Chemicals and was Puratronic grade, the highest quality available. Ammonium sulfate was obtained from Schwartz/Mann Biotech and was Ultra Pure grade. All other buffers, reagents and enzymes, except where stated, were obtained from Sigma Chemical Co. Other materials and procedures are as described in the Experimental Procedures section of Chapter II.  IV.C.2: Synthetic Methods. Hexokinase (Sigma H5500) was added to 0.25 M triethanolamine buffer, pH 7.6 (1 mL) and was dialysed (twice) against the same buffer (50 mL) for two hours to remove traces of glucose. The sugar to be phosphorylated and a 10% excess of adenosine 5'-triphosphate were dissolved in 0.25 M triemanolamine buffer pH 7.6 containing 5 mM magnesium chloride and the pH readjusted to 7.6. The volume of buffer added was such that the concentration of sugar was approximately 50 mM. The solution of dialysed enzyme was added and the mixture incubated at room temperature. During the incubation period the pH was readjusted severaltimesand aliquots were removed for analysis by ^F-nmr to determine whether the reaction had reached completion. After completion, the protein was removed by a Centricon microconcentrator and the solution diluted with water to a volume of 200 mL. After adjustment of the pH to 8.0 the solution was applied to a column of DE-52 cellulose (1.8 x 30 cm) which had been previously equilibrated with 40 mM ammonium bicarbonate. The column was first washed with 200 mL of water and the products were eluted by a linear gradient of 0-0.2 M ammonium bicarbonate in a total volume of 2 L. Fractions (20 mL) were collected at aflowrate of 2 mL/min. The desired product was eluted at a salt concentration of about 80 mM and was identified by the  206 phenol/sulfuric acid test for carbohydrates (Hodge & Hofreiter, 1962). Adenosine 5'diphosphate was identified by its absorption at 260 nm. After pooling the desired fractions, the volatile buffer was removed by repeated lyophilizations. The product was converted to its biscyclohexylammonium salt via a Dowex 50W-X8 (H ) resin column and crystallization attempted +  from water/acetone. This was only successful for the 2-fluoro derivative. The remaining phosphate esters were isolated as freeze dried powders. The procedure for the phosphorylation of a-glucosyl fluoride was slightly different. Hexokinase was the type having a low aglucosidase activity (Sigma H-5875) and thefinalconcentration of the sugar in the reaction mix was approximately 0.2 M, the solution also containing 2 mM acarbose. The ion exchange chromatography was performed at 4" on a smaller column (1.8 x 10 cm) and the volumes of the eluting solution and fractions were reduced proportionally. The reaction conditions and yields for the individual reactions are given in Table IV.10. A summary of the 31p and l F-nmr 9  characterization data for each analogue is given in Table IV. 11.  Glucose derivative  Amount of sugar (K>  Hexokinase (mg)  Reaction Time (h)  Yield  2-fluoro-  0.146  0.5  2  77  3-fluoro-  0.300  3.0  20  82  4-fluoro-  0.045  0.6  20  96  1-deoxy-  0.075  1.0  72  91  a-glucosyl fluoride  0.02  0.3  22  20  %  Table IV. 10: Summary of the reaction conditions and yields obtained in the synthesis of eachfluorinatedand deoxygenated derivative of glucose 6-phosphate.  207  Glucose 6-phosphate derivative  a b c  Chemical shift 8(ppm). 3lp(121MHz) 19F (254 MHz) a  2-fluoro-  -7.31, -7.25  200.08, 200.38  3-fluoro-  -7.26, -7.17  196.35, 200.93  4-fluoro-  -6.32, -6.35  200.67, 198.80  1-deoxy-  -6.79°  a-glucosyl fluoride  -6.83  b  156.98  c  Proton decoupled unless noted. t, Jp,6,6' 6.0 Hz. dd, JF,1 53.4, JF,2 25.9 Hz.  Table IV. 11: Summary of  31p and WF-nmr data of fluorinated and deoxygenated  analogues of glucose 6-phosphate.  IV.C.3: Enzyme Isolation and Purification. Commercial preparations of phosphoglucomutase could not be used for the 19p_  nmr  experiments in view of the prohibitive cost of this quantity of enzyme and also because the specific activity of commercial enzyme is only 10-20% that of freshly isolated enzyme. For these reasons phosphoglucomutase was isolated and purified in gram quantities by the method of W. J. Ray. This procedure has not been published in its most updated form and persons requiring details should contact W.J. Ray. However, in the interests of completeness and in order to show the basic principles of the purification, the important steps of the procedure are presented here in condensed form by way of a flow diagram (Figure IV.25).  208  2 kg Rabbit muscle ground and extracted with water  Extract heated to 65° over 4 niinutes, then cooled to 25° over 6 minutes  Solution brought to 20% saturation with solid ammonium sulfate  t  Centrifuge Discard precipitate  Supernatant  Solution brought to 65% saturation with solid ammonium sulfate  Centrifuge Store precipitate overnight at 4°  Discard supernatant  Precipitate dissolved in buffer and brought to 32% saturation by addition of ammonium sulfate solution II  Figure IV.25: Basic steps of isolation and purification ofphosphoglucomutase.  209  I  Centrifuge Supernatant  Discard precipitate.  Solution made to 52% saturation by addition of saturated ammonium sulfate  Centrifuge Precipitate  Discard supernatant  Precipitate dissolved in buffer and ammonium sulfate added to 40% saturation  J  Centrifuge Discard precipitate  Supernatant  I I  Solution made to 51% saturation by addition of saturated ammonium sulfate  ure IV.25: (continued).  210  !  Centrifuge Store precipitate over night at 4°  Discard supernatant  Precipitate dissolved in buffer Sephadex G-25 desalting column  I  DE 52 cellulose column, during with 10mMTrisHCLpH7.0  J  CM Sephadex column, overnight pH and salt gradient elution  I  Fractions pooled and brought to 65% saturation with ammonium sulfate  Cfentrifuge Precipitate  I  Dissolve in 30% ammonium sulfate at 30 mg/mL, yield 500 mg Figure IV.25: (continued).  Discard supernatant  211  The enzyme used in these nmr experiments was prepared at Purdue University, Indiana, by Dr. J. Puvathingal and myself and was repeated, on a half scale, at U.B.C. with the assistance of Ms. K. Rupitz. In our own preparation, 2.0 kg of rabbit skeletal muscle yielded 524 mg of phosphoglucomutase having a specific activity of 1100  \imoV min/mg.  SDS-gel  electrophoresis of the purified enzyme showed that the enzyme was essentially pure, the major product and the very minor impurities having exactly the same electrophoretic pattern as that of the enzyme prepared at Purdue University. The purified phosphoenzyme was stored at 4° at a concentration of 30 mg/ml in a solution of 30% saturated ammonium sulfate and a buffer containing 50 mM MES, 16 mM MgS04, and 1 mM EDTA, pH 6.5. The enzyme was found to be completely stable under these conditions with no loss of activity detectable over a period of one year. However, after one year, the presence of about 10% dephosphoenzyme was detected by the method of Lowry and Passonneau (1969).  IV.C.4: Demetallation of NMR Solutions and Additives. The complete demetallation of the enzyme is essential for obtaining meaningful and reproducible results in nmr studies. For this reason, extreme care was taken that all solutions and glassware with which the demetallated enzyme would come into contact were clean and demetallated. All glassware, nmr tubes and plastic disposable pipettetipswere soaked at least overnight in a solution of 0.1 mM EDTA and were rinsed with distilled, double deionized water obtained from a Millipore water purification system. Stock solutions of 0.5 M Tris-HCl pH 7.5 and D 2 O were demetallated by passage through equilibrated Chelex-100 columns. The Chelex column (1.5 x 28 cm) for demetallation of the Tris-HCl solution was equilibrated as follows; 150 mL 1 M HCI, 150 mL deionized water, 150 mL IM Tris base, 0.5 M Tris-HCl pH 7.5 until the pH of effluent reached 7.5. The demetallated effluent was stored at 4° in Nalgene plastic bottles which had been treated with EDTA. The D 2 O for use as the nmr lock additive was demetallated as follows through a Chelex100 column (1.6 x 8 cm); 50 mL 1 M HCI, 50 mL deionized water, 50 mL Tris base, 50 mL 0.5  212  M demetallated (DM) Tris-HCl pH 7.5,50 mL DM 70 mM Tris-HCl pH 7.5 and DM 20 mM Tris-HCl pH 7.5 until the effluent pH was 7.5. D2O was then added, the first 20 mL discarded and the remainder collected in a clean Nalgene botde. A slurry of DM Chelex-100 resin was prepared in the same manner as for the D2O. However, after equilibration with the DM 20 mM Tris-HCl, the resin was slowly forced out the top of the column by compressed ahy the first 30% being discarded and the remainder collected in a wide mouth Nalgene bottle. Each substrate, inhibitor and internal reference compound was demetallated by passage of the solution in 20 mM Tris-HCl pH 7.5, through a small column of the above Chelex slurry in a 1 mL disposable pipette tip. The column was washed with a small volume of DM 20 mM TrisHCl pH 7.5, and the effluent collected in an Eppendorf microcentrifuge tube. The solution was diluted to approximately 30-40 mM with 50% Chelex slurry/20 mM Tris-HCl, pH 7.5. The exact concentration was determined by removing a small aliquot and analyzing it for phosphate in the same manner as described in the experimental section of Chapter U. The solutions were stored at -20°. All stock solutions of metal ions were made up by weight The concentration of MgCl2 was checked bytitrationwith EDTA using calmagite as the indicator. MgCl2 and Cd(OAc)2 were made up in DM 20 mM Tris-HCl, pH 7.5, at a concentration of 20 mM. LiCl was made up at 0.3 M in the same buffer and also contained 30 mM EDTA. The EDTA was recrystallized from 2M HC1. DM 1M Tris base was added to dissolve the EDTA and LiCl and the solution was made to the appropriate volume. A small volume of Tris base was added to bring the pH to 7.5. The volume was determined with a solution made up in tandem that was later discarded. The direct determination of the pH was undesirable since the pH electrode was likely contaminated with metal ions.  rV.C.5: Demetallation of Phosphoenzyme. Sufficient enzyme was demetallated for 2 or 3 nmr experiments at a time (« 60-80 mg protein). An aliquot of the stock enzyme solution was treated with 1.6 volumes of 90% saturated (NH4)2S04 in pH 6.5 buffer at 0°. After 1 hour, the precipitate was centrifuged at 17000 rpm in  213  a SS-34 rotor for 15 min. The supernatant was discarded and the precipitate dissolved in pH 8.5 buffer (1.2 rnL/100 mg protein) containing 0.4 M Tris-HCl and 40 mM EDTA (demetallation buffer). After 1 hour at 20° the solution was transferred to a dialysis bag (12-14000 M cutoff) r  and was dialysed against 300 mL of 20 mM Tris-HCl, 10 mM EDTA, pH 7.5 at 4° for 4 hours. The dialysis was repeated with fresh buffer for 8 and 12 hours. The dialysis bag was then transferred to a solution of 300 mL DM 20 mM Tris-HCl pH 7.5 containing 2 mL of the Chelex slurry. From this point on, all solutions, glassware and other apparatus that came into contact with the enzyme had been subjected to the aforementioned demetallation procedures. The solution was dialysed for 5 hours and the process repeated thrice with fresh buffer. The demetallated protein was transferred to a weighed 2 mL Centricon microconcentrator (30000 M  r  cutoff) and centrifuged at 7000 rpm in a SS-34 rotor until the volume was 1.0 mL. Demetallated D 2 O (1.0 mL) was added, the solution mixed by inverting several times and then centrifuged until the protein concentration was approximately 1.2 mM. The concentrated protein was then centrifuged into the collecting device. The exact concentration was determined from the absorbance of an aliquot using the extinction coefficient El%278 = 7.0 (Ray et ah, 1983). About 10% of the protein was lost during the manipulations of the demetallation procedure. Each nmr sample was made up as follows. An aliquot of 0.4 fxmol enzyme (approximately 0.35 mL of the 70 mg/mL solution) was transferred to an Eppendorf microcentrifuge tube and the internal reference and substrate analogue added. The volume was made to 0.40 mL, to give a protein concentration of 1.0 mM, with deionized water and DM D2O, the ratio of which was such that the final D 2 O concentration was 50%. After mixing and centrifugation, the solution was transferred to a 5 mm high precision nmr tube (Norrel 507-HP) using a piece of plastic tubing attached to a syringe. This allowed the viscous solution to be pipetted into the bottom of the tube with minimal losses on the sides. This demetallation procedure was found to be sufficiently complete for use with inhibitors with which a small amount of metal ion would have no effect. For use with substrates, a slightly different procedure was used. After the enzyme was treated with demetallation buffer it was transferred to a dialysis thimble (Sartorius collodion bag) and was dialysed against 300 mL 20  214  mM MES, 10 mM EDTA , pH 6.2 for 5 hours in a Sartorius vacuum concentrating flask. During this time and the subsequent three dialysis steps with fresh buffer, the protein was subjected occasionally to a vacuum of approximately 50 mm Hg in order to keep the volume constant. The solution was then dialysed thrice against DM 20 mM Tris-HCl pH 7.5 and Chelex slurry. During the final dialysis the vacuum was applied to increase the protein concentration to approximately 3 mM. The demetallated protein solution was removed from the dialysis thimble using a cleaned pyrex pipette and an aliquot transferred to a niirocentrifuge tube for making up of the nmr sample. The volume of D2O added was such that its final concentration was 50%. In the cases of the Ep.Li complexes, the LiCl /EDTA solution was added to the protein and mixed thoroughly +  prior to addition of the substrate.  IV.C.6: Acquisition of ^F Spectra. l^F-nmr spectra (254 MHz) were obtained in the fourier transform mode on a Bruker HXS-270 spectrometer equipped with a  high resolution probe and interfaced with a Nicolet  1180 computer. 90" Pulses (14 (is) were used, the spectral width was 20000 Hz in each case except for the solutions including a-glucosyl fluoride 6-P in which case a spectral width of 34000 Hz was employed. The spectra were digitized with 8K data points and quadrature phase detection was used. The transmitter offset was -22000 Hz and the spectral filters were set at ±12500 Hz. In the absence of *H decoupling a pulse delay of 0.5 s was employed, giving a repetition rate of 0.7 s. A delay of 2.0 s was employed when using gated Q3NA) decoupling. The frequency offset for *H decoupling was 4000 Hz. The temperature of the sample was regulated by an internal heating element and by blowing nitrogen cooled to -70" through the probe. In order to obtain a flat baseline it was essential to delete thefirst4 data points of the fid prior to fourier transformation (type "LS" 4 times). This was necessary because of a slight transmitter pulse breakthrough into the first few microseconds of the fid. The line broadening factor used in exponential apodizations was 20 Hz. This line broadening factor has already been subtracted for all the linewidths quoted in the text  215  IV.C.7: Deuteration of Phosphoenzyme and Reagents for ^H-rimr. Solutions of demetallated (DM) 0.5 M Tris-HCl pH 7.5 and DM 38.2 mM 6-fluoroglucose 1-P in 20 mM Tris-HCl pH 7.5 were deuterated by lyophilization followed by reconstitution to the original volume with DM D2O. This procedure was repeated to ensure complete replacement of exchangeable protons. The protein solution was demetallated exactly as detailed for the l F-nmr samples. The dialysed solution was then transferred to a Centricon 9  microconcentrator and the volume reduced to 0.5 mL. The volume was then made up to 2.0 mL with DM D2O containing 20 mM deuterated Tris-HCl pH 7.5 and the volume reduced to 0.5 mL. This process was repeated until the D2O concentration was calculated as 99.8%. The exact concentration of protein was ctetermined by absorbance and an equimolar amount of deuterated 6fluoro-glucose 1-P addecLThe protein concentration was adjusted to 1.0 mM with D20/deuterated buffer.  IV.C.7: Acquisition of lH-nmr Spectra. The 5 mm  high resolution probe from the Bruker HXS-270 spectrometer was used  to observe *H with gated ^ F decoupling. The normal *H observation channel was connected to 9  the *H decoupling coils without further modification. An external frequency synthesizer was connected to the ^ F observation coils as detailed in Figure IV.26 to provide gated 19p 9  decoupling. These modifications to the spectrometer hardware were made by Mr. K. Sukul of the Chemistry Department Electronics workshop. The *H spectra (270 MHz) were obtained in the fourier transform mode using quadrature phase detection. 90" Pulses (18 |xs) were used with a spectral width of 4000 Hz. 8K data points were used to digitize the spectra. Spectra were collected at a sample temperature of 20°. The noe was measured using a pulse sequence described in the results and discussion section. The pulse sequence in which the decoupler was turned on was programmed as follows; NOEON #1: D6 #2: DO 8 1= l.ILIM  216  #3: D3 20000 #4:D4 #5:P2 1 #6: A 40 #7:D2 JUMP TO #3 The sequence NOEOFF in which the decoupler was turned off was programmed by the same commands except. #3: D3 was substituted for #3; D3 20000.  Variable Frequency Synthesizer 0-150 MHz  Gate  Frequency Doubler  Filter 270 MHz stop  Noise Generator  HXS-270 Output BF-BU17  Wide Band Amplifier  Probe 19,Fcoil  Bruker 254 MHz Pass Filter  Attenuator  Figure IV.26: Modifications to the Bruker HXS-270 spectrometer to allow ^H observation and ^F decoupling.  217  CHAPTER V: SUMMARY.  The mechanism of phosphoglucomutase has been investigated using fluorinated and deoxygenated substrate analogues. Each of the analogues in which the non-acceptor hydroxyls are replaced by fluorine or hydrogen are substrates of the enzyme. The K values of the glucose m  1-P analogues all fall in the range of approximately 100 - 200 JIM and their Vmax values (relative to glucose 1-P) between 0.009 and 0.05. Time courses of the reactions of the fluorinated and deoxygenated glucose 1-P substrates with phosphoglucomutase in the presence of glucose 1,6-diP showed that the rate at which the overall mutase reaction occurs is the same as that of the half reaction, involving production of the fluorinated and deoxygenated glucose 1,6-diP species. The exceptions were 3-fluoro- and 3deoxy-glucose 1-P. In these cases the rates of the half reactions are both 8timesthat of the overall mutase reaction. In addition, it was determined that the K of 3-fluoro-glucose 1,6-diP is m  approximately 90 fold higher than that of glucose 1,6-diP and the other deoxygenated and fluorinated diphosphate species. These observations are consistent with the hypothesis of LA. Rose (1987) that the enzyme operates by an exchange type of mechanism in which the glucose 1,6-diP reorientates itself in the active site by a rotation about the C-3 hydroxyl. Removal of this hydroxyl (and its replacement by hydrogen or fluorine) disrupts the hydrogen bonding interaction between the enzyme and the substrate at that critical site. Thus the rate of the interconversion of the two diphosphate species is reduced, as is that of the overall mutase reaction. The inhibition of phosphoglucomutase by various fluorinated and deoxygenated glucose 1-P inhibitors has also been investigated. Replacement of (either of) the acceptor hydroxyls (i.e., C-l or C-6) created inhibitors that bind the phosphoenzyme with essentially the same dissociation constants as that of their respective parent substrates. Thus, no significant hydrogen bonding (or other) interactions present in the enzyme-substrate complex are interrupted in these enzyme-inhibitor complexes. A series of novel disubstituted glucose 1-P analogues has been synthesized in which the C-6 hydroxyl was replaced by fluorine and one of the glucose ring  218  hydroxyls by either hydrogen orfluorine.These inhibitors were used to investigate enzymesubstrate interactions in the ground state complex. The results showed that replacement of a hydroxyl functionality distal to the acceptor hydroxyl (C-6) produced a worse inhibitor than the same replacement proximal to the acceptor hydroxyl. Thus 4,6-difluoro-glucose 1-P is a better inhibitor than 3,6-difluoro-glucose 1-P, which in turn is a better inhibitor than 2,6-difluoroglucose 1-P. These results are in good agreement with the tentative findings of Ma and Ray (1980). A comparison was made of the apparent loss of binding free energies of each fluorinated substrate and difluorinated inhibitor. These energies correspond to the transition state and ground state, respectively. The results showed that each sugar ring hydroxyl contributes approximately 4-8 kJ/mol more binding energy in the transition state than in the ground state complex. These relatively small differences would be expected in view of the non-involvement of the glucose moiety in the phosphoryl transfer reaction. The charge state of the enzyme-bound substrate phosphate group has been investigated using phosphorofluoridate substrate analogues. These analogues, which can only bear a single negative charge at physiological pH, were found to only weakly inhibit phosphoglucomutase. No evidence of any phosphoryl transfer between the phosphoenzyme and the phosphorofluoridate analogues could be detected. These results were interpreted in terms of a strict requirement for a doubly negatively charged substrate phosphate group. The phosphorofluoridate substrate analogues were also tested for covalent inhibition of phosphoglucomutase. The possibility that these analogues may act as covalent inhibitors resides in their similarity to diisopropylphosphorofluoridate, a covalent inhibitor of other enzymes with a nucleophilic serine residue at their active site. However, no covalent inhibition was observed with this, or with another potential covalent inhibitor, glucose 4:6-cyclic phosphate 1-phosphate. These results were rationalized in terms of the "stereochemically demanding" phosphoryl transfer reaction. The interaction of phosphoglucomutase withfluorinatedsubstrates and inhibitors has been investigated by F-nmr. Large downfield changes in the chemical shifts of the inhibitors 19  219 6-fluoro-glucose 1-P and a-glucosylfluoride6-P were found to accompany binding to the phosphoenzyme. The further binding of L i to these complexes altered the chemical shifts of +  these bound species by 1.5 ppm. However, binding of the activating metal ions Mg2+ and C d  2+  produced a dramatic broadening (and shift in the case of Mg2+) of both bound and free species. This phenomenon, which could not be entirelyrationalized,may be related to the way in which these metal ions cause the 10^ to 10 fold activation of the enzyme. 7  The effect of Li+ was further studied using the aforementioned difluorinated inhibitors. Binding of each of these inhibitors to the phosphoenzyme produced spectra consistent with a rapid exchange between bound and free species. Addition of L i caused the exchange process to +  be greatly slowed and caused the resonances of the bound species to shift downfield by 3-15 ppm. These data show that binding of L i greatly increases the tenacity with which the +  phosphoenzyme binds substrates and inhibitors. The species formed by reaction of the Cd -phosphoenzyme with each of the fluoro 2+  substrates was investigated by l°F-nmr. In the presence of Cd2+, the equihbrium amongst the central complexes of glucose phosphates is such that glucose 1,6-diP is the dominant species. Whether this is the case with thefluorosubstrates could not be independently determined. The l^F-nmr spectra of these complexes showed two bound species in the cases of 2- and 3-fluoroglucose phosphate, whereas only one was observed in the case of 4-fluoro-glucose phosphate. These species were tentatively assigned as thefluoro-glucose1,6-diP species bound in two different modes to the dephosphoenzyme. A rationale was presented, based on the inhibition constants of the difluorinated inhibitors, to explain the numbers offluoro-glucose1,6-diP species observed in each case. Thus these data would appear to be consistent with an exchange type of mechanism involving two different glucose-1,6-diP-dephosphoenzyme complexes. The environment of each substrate glucose ring hydroxyl in the active site has been probed using F-nmr and thefluorinatedglucose phosphate substrates. The changes in l9  chemical shift observed on the binding of eachfluorosubstrate to the Ep.Li complex were used +  as an indicator of thefluorinereporter groups' environment in the active site. The data obtained suggested that there is not a single glucose ring binding site for both substrates of the enzyme.  220  This result is clearly inconsistent with a minimal motion type of mechanism and may also be inconsistent (although the results are less clear) with an exchange type of mechanism involving the swapping of interactions between the C-2 and C-4 hydroxyls (as proposed by LA. Rose). An attempt was made to identify the protein amino acid residues in the active site by way of a I ^ F - I H nmr noe experiment. However, no noe could be detected between 6-fluoro-glucose 1-P and the protein, this result being attributed to rapid and efficient spin diffusion. Thus, it would appear that noe techniques such as this have Umited use in such large proteins. The overall aim of this study was to determine the type of mechanism by which phosphoglucomutase operates. The results obtained are more consistent with an exchange type of mechanism in which the C-3 hydroxyl plays an important role in that species' reorientation. The data also suggest that there are two distinct glucose binding sites, one for each substrate and glucose 1,6-diP bound in the same mode. Although these results are not conclusive by themselves, they do add to the growing number of reports in the literature consistent with an exchange type of mechanism.  221  APPENDIX.  0.6-  -a  •  0.5" 0.40.3' 0.2-  •  •  0.1 • o.o- —•  i • i • i • i • i • -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 1/Substrate concentration, 1/JLM  Lineweaver-Burk plot of phosphoglucomutase activity versus 2-fluoro-glucose 1-P concentration in the presence of 1.0 |xM glucose 1,6-diP.  0.06T  1  * ~  0.05 • 0.04"  2  *  0.03-  i  •  •  ^  0.02-  ^  0.01 • o.oo-l—•—i—•—i—'—i—•—i • i •— -0.010-0.0050.000 0.005 0.010 0.015 0.020 1/Substrate concentration, l/|iM  Lineweaver-Burk plot of phosphoglucomutase activity versus 3-fluoro-glucose 1-P concentration in the presence of 5.0 |iM glucose 1,6-diP. First phase.  222 1.0  0.0 T T — I — — I — 1  1  l —  I— —I 1  • I—'—r—*—  -0.02-0.010.00 0.01 0.02 0.03 0.04 0.05 1/Substrate concentration, 1/uM  Lineweaver-Burk plot of phosphoglucomutase activity versus 3-fluoro-glucose 1-P concentration in the presence of 5.0 |iM 3-fluoro-glucose 1,6-diP. Second phase.  o.ooo -0.010.00  0.01 0.02 0.03 0.04 0.05 0.06 1/Substrate concentration ,1/uM  Lineweaver-Burk plot of phosphoglucomutase activity versus 4-fluoro-glucose 1-P concentration in the presence of 1.0 |J.M glucose 1,6-diP.  -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 1/Substrate concentration, 1/uM  Lineweaver-Burk plot of phosphoglucomutase activity versus 2-deoxy-glucose 1-P concentration in the presence of 1.0 iiM glucose 1,6-diP.  223  -0.02 -0.01 0.00 0.01 0.02 0.03 1/Substrate concentration, 1/uM  0.04  Lineweaver-Burk plot of phosphoglucomutase activity versus 3-deoxy-glucose 1-P concentration in the presenceof 5.0 PM glucose 1,6-diP. First phase.  -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 1/Substrate coimcentration, 1/uM  Lineweaver-Burk plot of phosphoglucomutase activity versus 3-deoxy-glucose 1-P concentration in the presence of 5.0 |J.M glucose 1,6-diP. Second phase.  0.250.20" 0.15i  0.100.05" o.oo-0.010-0.0050.000 0.005 0.010 0.015 0.020 1/Substrate concentration, 1/uM  Lineweaver-Burk plot of phosphoglucomutase activity versus 4-deoxy-glucose 1-P concentration in the presence of 1.0 nM glucose 1,6-diP.  224  l/Substrate concentration, l/|iM  Lineweaver-Burk plot of phosphoglucomutase activity versus 3-fluoro-glucose 1,6-diP concentration in the presence of 503 |J.M 3-fluoro-glucose 1-P.  l/Glc 1-P 1/u.M  Inhibitor Concentration,  Lineweaver-Burk plot and replot of inhibition of phosphoglucomutase by 6-fluoro-glucose 1-P.  Lineweaver-Burk plot and replot of inhibition of phosphoglucomutase by a-glucosyl fluoride 6-P.  225  0.014 • 0.012" oo  I  "o  o.oio0.008" 0.006" 0.004" 0.002-  -o.ooo-0.05  -r  0.00  0.20 0.05 0.10 0.15 l/[Glc 1-P], 1/uM  -60 -40 -20 0 20 40 60 80 100 120 Inhibitor concentration, | i M  Lineweaver-Burk plot and replot of inhibition of phosphoglucomutase by 1-deoxyglucose 6-P.  0.004  0.000 -0.06 -0.03 0.00 0.03 0.06 0.09 0.12 l/[Glc 1-P]. 1/uM  -i—'  -  3  i  - 2 - 1 0 1 2 Inhibitor concentration, m M  3  Lineweaver-Burk plot and replot of inhibition of phosphoglucomutase by 2,6-difluoroglucose 1-P.  226  0.006 -a o.oo5 .1 0.004  I 1 0.003 S 0.0021 - 0 11 0. .0 00 00 -0.08  -0.04 0.00 0.04 0.08 l/[Glc 1-P], 1/u.M  0.12  -2.0 -1.5 -1.0 -0.5  0.0  0.5  1.0  1.5  2.0  Inhibitor concentration. m M  Lineweaver-Burk plot and replot of inhibition of phosphoglucomutase by 3,6-difluoroglucose 1-P.  0.000 -0.05  -«—i— 0.00 0.05 0.10 0.15 l/[Glc 1-P], 1/uM  0.20  -0.5 0.0 0.5 1.0 Inhibitor concentration, m M  Lineweaver-Burk plot and replot of inhibition of phosphoglucomutase by 4,6-difluoroglucose 1-P.  227  0.010 "So 0.008  60-  I  50-  ! | 0.006" o t  . 40-  3  1  30"  0.004-  CO V  ^  5  0.002 0.000 -0.12 -0.08 -0.04 0.00 0.04 0.08 0.12 l/[Glc 1-P], 1/uM  20-  ioo-1.0  T  '  1  '  1  • 1 '  I  '  -0.5 0.0 0.5 1.0 1.5 2.0 Inhibitor concentration, mM  Lineweaver-Burk plot and replot of inhibition of phosphoglucomutase by 4-deoxy-6fluoro-glucose 1-P.  0.03  o.oo  -0.15 -0.10 -0.05 0.00 0.05 l/[Glc 1-P], 1/n.M  1*  0.10 0.15  I ' I ' I ' I-' I  -30 -20 -10 0 10 20 30 40 50 60 Inhibitor concentration, mM  Lineweaver-Burk plot and replot of inhibition of phosphoglucomutase by mannose 1phosphorofluoridate.  -200  i  •  i  -100 0 100 200 Inhibitor concentration, uM  300  Dixon plot of inhibition of phosphoglucomutase by glucose 4:6-cyclic phosphate 1phosphate. [Glucose 1-P] = 70 uM, [Glucose 1,6-diP] = 0.1 ^M, V = 36 |imoVmin/mg, K (glucose 1,6-diP) = 0.1 \iM. m a x  m  228  -20  -10 0 10 20 Inhibitor concentration, mM  30  Dixon plot of inhibition of phosphoglucomutase by glucose 6-phosphorofluoridate. [Glucose 1-P] = 9.7 iiM, [Glucose 1,6-diP] = 1 |xM, V = 81 limol/min/mg, K (glucose 1-P) = 8.5 \iM. m a x  0.030 T «  m  y JC  0.025" 0.020"  2 0.015"  X  ~ o.oio«  0.0050,000 - F - ^ — i — • — i — • — i — • — i • i—'— -30 -20 -10 0 10 20 30 Inhibitor concentration, mM  Dixon plot of inhibition of phosphoglucomutase by 2-fluoro-glucose 1phosphorofluoridate. [Glucose 1-P] = 9.7 ^iM, [Glucose 1,6-di-P] = 1 iiM, Vmax = 138 (imol/min/mg, Km (glucose 1-P) = 14 |iM.  229  REFERENCES. Titles of journals are abbreviated according to Periodical Title Abbreviations, 5 edition, L.G. Alfire. m  Adamson, J., Foster, A.B., Hall, L.D., Johnson, R.N. & Hesse, R.M. (1971) 15, 351.  Carbohyd. Res.  Alpers, J.B. (1967) J. Biol. Chem. 243, 1698. Alpers, J.B. & Lam, G.K.H. (1968) /. Biol. Chem. 244, 200. Ando, M.E., Gerig, J.T. & Luk, K.F.S. (1986) Biochem. 25, 4772. Arita, H., Ueda, N. & Matsushima, Y. (1972) Bull. Chem. Soc. Japan 45, 567. Atkinson, M.R., Johnson, E. & Morton, R.K. (1961) Biochem. J.  79,  Baillargeon, D.J. & Reddy, G.S. (1986)  Carbohyd. Res. 154, 275.  Balls, A.K. & Jansen, E.F. (1953) Adv.  Enzymol. 13,  12.  321.  Barker, G.R. (1963) in Methods in Carbohydrate Chemistry II (Whistler, R.L. & Wolfrom, M., Eds.) pp 168, Academic Press, New York. Barnett, J.E.G., Jarvis, W.T.S. & Munday, K.A. (1967) Biochem. J.  105,  669.  Barnett, J.E.G. (1972) in Carbon-Fluorine Compounds, Chemistry, Biochemistry and Elsevier, Amsterdam.  Biological Activities,  Bartlett, G.R. (1959) /. Biol. Chem. 234, 466. Bartlett, P.A.& Marlow, C.K.(1987) Science 235, 569. Beevers, CA. & Maconochie, G.H. (1965) Acta. Cryst.  18, 232.  Bell, D.J. & Greville, G.D. (1955) /. Chem. Soc. 1136. Bessel, E.M., Foster, A.B. & Westwood, J.H. (1972) Biochem. J.  128,  199.  Bessel, E.M. & Thomas, P. (1973) Biochem. J. 131, 83. Bessodes, M., Komiohs, D. & Antanokis, K. (1986) Tetrahedr. Lett. 47, 579. Biochemica Information 1 (1973) Boehringer Mannheim Information. West Germany. Blackburn, G.M., Kent, D.E. & Kolkmann, F. (1981a) /. Chem. Soc. Chem. Commun. 930. Blackburn, G.M., England, D.A. & Kolkmann, F. (1981b) /. Chem. 1188. Blattler, W.A. & Knowles, J.R. (1980) Biochem. 18, 3927. Bunton, CA. & Hameres, E. (1969) /. Org. Chem. 34, 572.  Soc. Chem. Commun.  230  Burrows, S., Grylls, F.S.M. & Harrison (1952) Nature 170, 800. Briggs, G.E. & Haldane, J.B.S. (1925) Biochem. J. 19, 338. Britton, H.G. & Clarke, J.B. (1968) Biochem. J. 110, 161. Campbell, I.D., Dobson, CM. & Williams, R.J.P. (1985) Biochem. J. 231, 1. Card, P. (1983) /. Org. Chem. 48, 393. Card, P. & Reddy, G.S. (1983) /. Org. Chem. 48, 4734. Cech, T.R. & Bass, L. (1986) Annu. Rev. Biochem. 55, 599-629. Chang, Y.C, Scott, R.D. & Graves, DJ. (1987) Biochem. 26, 360. Cheblowski, J.A., Armatage, I.M., Tusa, P.P. & Colman, J.E. (1976) /. Biol. Chem. 251, 1207. Corbridge, D.E.C. (1974) The Structural Chemistry of Phosphorous, Elsevier, Amsterdam. Corbridge, D.E.C. (1985) Phosphorous; An Outline to its Chemistry, Biochemistry and Technology (3rd. ed.), Elsevier, Amsterdam. Dauben, W.G., Kessel, C.R., Kishi, M., Somei, M., Tada, M. & Guillerm, D. (1982) J. Am. Chem. S. 104, 303.  Dixon, M (1953) Biochem. J. 55,170. Dixon, M., (1970) in The Chemistry of Life (Needham, J. Ed.), Cambridge University Press. Doherty, K. (1987) M.Sc. Thesis, Chemistry Department, Simon Eraser University. Drueckhammer, D.G. and Wong, C. (1985) /. Org. Chem. 50, 5913. Dwek, R.A.(1973) Nuclear Magnetic Resonance in Biochemistry, Clarendon Press, Oxford. Engels, R. (1977) Chem. Rev. 77, 349-367. Eygud, L.G. & Whelan, W.J. (1963) Biochem. J. 86, IIP. Ferdinand, W. (1976) The Enzyme Molecule, Wiley, New York. Fersht, A.R. (1972) /. Molec. Biol. 64, 497. Fersht, A.R. (1985) Enzyme Structure and Mechanism, 2nd ed., Freeman. New York. Fersht, A.R., Shig, J., Knill-Lones, J., Lowe, D.M., Wilkinson, A.J., Blow, D.M., Brick, P.,Carter, P.,Waye, M.M.Y. & Winter, G. (1985) Nature 314, 235. Fersht, A.R. (1987a) Biochem. 26, 8031. Fersht, A.R. (1987b) TIBS 12, 301. Fiske, CH. & Subbarow, Y. (1925) / . Biol. Chem. 66, 375.  231  Gaddian, D.G., Radda, G.K. & Richards, R.E. (1974) Biochim. Biophys. Gamcsik, M.P., Gerig, J.T. & Gregory, D.H. (1987) Biochim. Biophys. Georgatsos, J.G. (1967) Arch. Bioch. Gerig, J.T. (1977) J. Am.  121,  Chem. S. 99,  Gerig, J.T. (1978) in Biological  Acta 358,  Acta 912,  57.  303.  619.  1721.  Magnetic Resonance, Vol 1  Gerig, J.T. & Klinkeborg, J.C. (1980) /. Am.  (Berliner, L.T. & Reuben, J. Eds.)  Chem. S. 102, 4267.  Gerig, J.T. (1982) in Biomedicinal Aspects of Fluorine Chemistry (Filler, R. Ed.). Goldberger, R.F. (1967) Methods Enzymol.  11,  317.  Gorenstein, D.G. (1984) Phosphorous-31 NMR. Principles and Applications, Chapter 1, Academic Press, Orlando. Haldane, J.B.S. (1930) Enzymes, Longman, Green and Co. MIT press, 1965. Hall, CR. & Inch, T.D. (1980) Tetrahedron, 36,2059. Hall, L.D. & Manville, J.F. (1969)  Can. J. Chem. 47,  Hall, L.D. & Manville, J.F. & Bacca, M.S. (1969) Hammond, S.J. (1984) /.  19.  Can. J. Chem. 47,  Chem. Soc. Chem. Commun.  Han, C.H. & Rose, Z. (1979) /. Biol.  Chem. 254,  1.  713.  8836.  Hanabusa, K., Dougherty, H.W., del Rio, C , Hashimoto, T. & Handler, P. (1966) /. Biol. Chem. 241, 3930. Hanessian, S., Delorme, D. & Dufresne, Y. (1984) Tetrahedr. Lett. 25,2515. Hanna, R. & Mendocino, J. (1970) /. Biol. Chem. 245, 4031. Hayashi, S., Hashimoto, R. & Noyori, R. (1984)  Chem. Lett..  1747.  Hodge, J.E. & Hofreiter, B.T. (1963) in Methods in Carbohydrate Chemistry II (Whisder, R.L. & Wolfram, M., Eds.) pp 380, Academic Press, New York. Hudlicky, M. (1976) Chemistry of Organic Fluorine Compounds, 2nd Ed., Halsted Press, New York. Hull, W.E. & Sykes, B.D. (1975a) J. Mol. Biol. 98, 121. Hull, W.E. & Sykes, B.D. (1975b) /. Chem.  Phys. 63,  867.  Hull, W.E. & Sykes, B.D. (1976) Biochem. 15, 1535. Jardetzky, O. & Roberts, G.C.K.(1981) NMR in Molecular Biology, Academic Press, New York. Jencks, W.P. (1975) Adv.  Enzymol. 43,  219.  232  Jencks, W.P. (1981)  Proc. Natl. Acad. Sci. USA 78, 4046.  Johnson, P.W., Tigerstrom, R. & Smith, M. (1975) Nucleic Acids Res.  2,  1745.  Johnston, J.W. (1984) B.Sc. Thesis, Chemistry Dept., University of British Columbia. Jones, P.J., Cioffi, E.A. & Prestegard, J.H. (1987) /. Biol. Chem. 262, 8963. Karplus, M. (1963) /. Am.  Chem. S. 85,  2870.  Kimber, B.D.V., Griffiths, B., Birdsall, J., King, R.W., Scudder, P., Fenney, J., Roberts, G.C.K. & Burgen, A.S.V. (1977) Biochem. 16, 3492. Klein, H.W., Im, M.J., Palm, P., Helmreich, E.J.M. (1984) Biochem. 23, 5853. Knowles, J.R. (1976)  CRC. Crit. Rev. Biochem. 4,  Knowles, J.R. (1980) Annu. Rev. Knowles, J.R. (1982) Fed. Koshland, D.E. (1958)  Biochem. 49,  Proc. 41,  165-173.  877.  2424.  Proc. Natl. Acad. Sci. U.SA. 44,  98.  Kovac, P., Farkas, I., Mihalov, V., Palovcik, R & Bognar, R. (1976) /.  Carbohyd. Nucl.  Nucl. 3, 57.  Layne, P.P. & Najjar, V.A. (1978) Biochim. Biophys. Acta 526, 429. Lin, Z., Konno, M., Abad-Zapatero, C, Wierenga, R., Murthy, M.R.N., Ray, W.J., Jr. and Rossmann, M. (1986) /. Biol. Chem. 261, 264. Line weaver, H. & Burk, D. (1934) /. Am.  Chem. S. 56,  Lowe, G. & Potter, B.V.L. (1981) Biochem. J.  199,  658.  693.  Lowry, O.H. and Passonneau, J.V. (1969) /. Biol. Chem. 244, 910. Ma, C. & Ray, W.J., Jr. (1980) Biochem. 19, 751. MacDonald, D.L. (1972) in Methods in Carbohydrate Chemistry VI (Whistler, R.L. & BeMiller, J.N., Eds.) pp 389, Academic Press, New York. Machado de Domenech, E.E. & Sols, A. (1980) FEBS Lett. 119,174. Mega, T. & Matsushima, Y. (1983) /. Biochem. (Tokyo) 94, 1637. Meyer, R.A., Kushmerick, M.J. & Brown, T.R. (1982) Am.  J. Physiol. 242,  Michaelis, L. & Menten, M.L. (1913) Biochem. Z. 49, 333. Micheel, F. & Klemer, A. (1961) Adv.  Carbohyd. Chem. 16,  85.  Monasterio, O. (1987) Biochem. 26V 6099. Monasterio, O. & Timasheff, S.N. (1987) Biochem. 26, 6091.  Cl-Cll.  233  Mulhausen, H. & Mendocino, J. (1970) /. Biol. Chem. 245,4038. Muller, N.(1976) /. Magn. Reson. 23, 327. Murray-Rust, P., Stallings, W.C, Monti, C.T. Preston, R.K. & Clusker, J.P. (1985) /. Am. Chem. S. 105, 3206. Najjar, V.A. (1962)  The Enzymes, 2nd Ed. 6,  161.  O'Connor, J.V., Nunez, H.A. & Barker, R. (1979) Biochem. O'Connor, J.V. & Barker, R. (1979) Ogawa, T. & Seta, A. (1982)  Carbohyd. Res. 73,  Carbohyd. Res. 110,  500.  18,  227-234.  CI-4.  Opella, S.J., Friedman, R.A. Jarema, M.C. & Lu, P. (1979) J. Magn. Res. 36, 81. Paladini, A.C. & Leloir, L.F. (1952) /. Biol. Chem. 51, 426. Pauling, L. (1946)  Chem. Engng. News  24, 1375.  Pauling, L. (1960) The Nature ofthe Chemical ;  Bond, 3rd Ed.,  Cornell University Press, Ithaca.  Peck, E.J., Jr., Kirkpatrick, D.S. & Ray, W.J., Jr. (1968) Biochem. 7, 152. Peck, E.J., Jr. & Ray, W.J., Jr. (1971) in Specifications and Criteria for Biochemical Compounds, 2nd Ed., Natl. Res. Council Washington D.C. Penglis, A. (1981) Adv.  Percival, M.D. & Withers, S.G. (1988) Phillips, D.C. (1967)  195.  Carbohyd. Chem. Biochem. 38,  Can. J. Chem. In  Proc. Natl. Acad. Sci. U.SA . 57,  press.  484.  Phillips, L. & Wray, V. (1971) /. Chem. Soc. (B), 1618. Posternak, T. & Rosselet, J.P. (1953) Helv. Chim. Acta 36, 1614. Posternak, T. & Rosselet, J.P. (1954) Helv. Chim. Acta 37, 246. Ray, W.J., Jr. & Roscelli, G.A.(1964) /. Biol.  Chem. 239,  1228.  Ray, W.J., Jr. & Roscelli, G.A. (1966) J. Biol.  Chem. 241, 2596.  Ray, W.J., Jr., Roscelli, G.A. & Kirkpatrick, D.S. (1966) / . Biol. Ray, W.J., Jr. (1969) /. Biol.  Chem. 244,  Chem. 241, 2603.  3740.  Ray, W.J., Jr. & Mildvan, A.S. (1970) Biochem. 9, 3886. Ray, W.J., Jr. & Peck, E.J., Jr. (1972)  The Enzymes, 3rd Ed. 6,  Ray, W.J., Jr., Mildvan, A.S. & Long, J.W. (1973) Biochem. Ray, W.J., Jr. & Long, J.W. (1976a) Biochem.  15, 3990.  407.  12, 3724.  234  Ray, W.J., Jr. & Long, J.W. (1976b) Biochem.  15,  3993.  Ray, W.J., Jr. & Long, J.W. (1976c) Biochem.  15,  4018.  Ray, W.J., Jr., Long, J.W. & Owens, J.D. (1976) Biochem.  15,  4006.  Ray, W.J., Jr. (1978) Biochem. 77, 1554. Ray, W.J., Jr., Szymanki, G.L. & Ng, L. (1978) Biochim.  Biophys. Acta 522,  434.  Ray, W.J., Jr., Hermodson, M.A., Puvathingal, J.M.& Mahony, W.C. (1983) / . Biol. Chem. 9166.  258,  Ray, W.J., Jr. (1986) Personal communication to S.G. Withers. Rhyu, G.I., Ray, W.J., Jr. & Markley, J.L. (1984) Biochem.  23, 252.  Rhyu, G.L, Ray, W.J., Jr. & Markley, J.L. (1985a) Biochem.  24,  2536.  Rhyu, G.L, Ray, W.J., Jr. & Markley, J.L. (1985b) Biochem.  24,  4746.  Robertson, D.E., Kroon, P.A. & Ho, C. (1977) Biochem.  1443.  16,  Robertson, H.E. & Boyer, P.D. (1956) Arch. Biochem. Biophys. 62, 380. Rose, LA. (1987) Personal communication to S.G. Withers. Rosevear, P.R., Powers, V.M., Dowhan, D., Mildvan, A.S. & Keynon, G.L. (1987) 5338.  Biochem. 26,  Sachdev, G.P. & Fruton, J.S. (1975)  Proc. Nat. Acad. Sci. USA. 72,  3424.  Santi, D.V.& Pena, V.A. (1973) /. Med. Chem. 16, 273. Schmidt, O.Th. (1963) in Methods in Carbohydrate Chemistry II (Whistler, R.L. & Wolfrom, M., Eds.) pp 318, Academic Press, New York. Schwartz, R.T. & Schmidt, M.F.G. (1976)  Eur. J. Bioch. 62,  181.  Segel, LH. (1975) Enzyme Kinetics, Wiley, New York. Shibaev, N.M., Kusov, Y.Y., Kuchar, S. & Kochetkov, N.K. (1973) Izv. Akad.  Navk.  SSSR, Ser. Khim. 4, 922.  Sprang, S.R., Goldsmith, E.J., Fletterick, R.J., Withers, S.G. & Madsen, N.B. (1982) 5362.  Biochem. 21,  Still, W.E., Kahn, M. & Mitra, A. (1978) /. Org. Chem. 43, 2923. Street, LP., Armstrong, C.R. & Withers, S.G. (1986) Biochem. Stubbe, J.A. and Abeles, R.H. (198i) Biochem.  19,  25, 6021.  5505.  Sykes, B.D., Weingarten, H.I. & Schlesinger, J. (1974) Proc. Nat. Acad. Sci. U.SA. 71, 469. Sykes, B.D. & Hull, W.E. (1978) Methods Enzymol. 49, 270.  235  Sykes, B.D. and Weiner, J.H. (1980) in Magnetic Resonance in Biochemistry, J.S., Ed.) Taylor, N.F. (1972) in Carbon-Fluorine Elsevier, Amsterdam.  Vol. 1,  (Cohen,  Compounds, Chemistry, Biochemistry and Biological  Activities,  Thomas, P., Bessel, E.M. & Westwood, J.H. (1974) Biochem. J.  139,  661.  Trevelyan, W.E., Procter, D.P. & Harrison, J.S. (1950) Nature 164, 444. Tronrud, D.E., Holden, H.M. & Mathews, B.W. (1987) Science 235, 571. Walsh, C. (1979) Enzymatic Reaction Mechanisms, W.H. Freeman, San Fransisco. Walsh, C. (1983) Adv.  197.  Enzymology 55,  Warren, CD., Nasir-ud-Din, Jeanloz, R.W. (1978) Webb, G.A. (1986) in NMR Westheimer, F.H. (1968)  in living Systems  Carbohyd. Res. 64,  43.  (Axenrod, T. & Ceccarelli, G., Eds.) Reindel.  Acc. Chem. Res. 1,70.  Wilkinson, G.N. (1961) Biochem. J.  324.  80,  Winter, G., Fersht, A.R., Wilkinson, A.J., Zoller, M. & Smith, M. (1982) Nature 299, 756. Withers, S.G. & Madsen, N.B. (1980) Biochem. Biophys. Res.  Commun. 97,  513.  Withers, S.G., Madsen, N.B. & Sykes, B.D. (1981) Biochem. 20, 1748. Withers, S.G., MacLennan, D.J. & Street, LP. (1986a) Withers, S.G., Street, LP. & Rettig, S.J. (1986b)  Can. J. Chem. 64,  Withers, S.G , Percival, M.D. & Street, LP. (1988) Wittman, R. (1963)  Chem. Ber. 96,  Carbohyd. Res. 154,  Carbohyd. Res.  127.  232. In Press.  771.  Wolfrom, M.L. & Thompson, A. (1963a) in Methods in Carbohydrate Chemistry II (Whisder, R.L. & Wolfrom, M., Eds.) pp 211, Academic Press, New York. Wolfrom, M.L. & Thompson, A. (1963b) in Methods in Carbohydrate Chemistry II (WhisUer, R.L. & Wolfrom, M., Eds.) pp 215, Academic Press, New York. Wright, J.A., Taylor, N.F., Brunt, R.V. & Brownsey, R.W. (1972) J.Chem.Soc. Commun. 691.  Wuthrich, K. (1986) NMR Yount, R.G. (1975) Adv.  of Proteins and Nucleotides,  Enzymol. 43,  Wiley, New York.  1.  Zmudzka, B. and Shugar, D. (1964) Acta Biochim. Pol.  11,  509.  Chem.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0060300/manifest

Comment

Related Items