*3 l LII um; mzllyujml in i"! w my; ”i ll)! [l1 u Mi ' igan State Univcmr)’ ,W a This is to certify that the thesis entitled The Cyanohydrin (Kiliani) Reaction: Mechanism Studies by 13C NMR Spectroscopy and Application to the Synthesis of Isotopically-Enriched Carbohydrates presented by Anthony Stephen Serianni has been accepted towards fulfillment of the requirements for Ph.D. Biochemistry degree in 1%wa . Major professor Date 12/17/79 0-7639 ‘ZJP‘T‘AVVA', -...:_2:! Y ~. ...-I -. :y- "(Out-4155» . OVERDUE FINES: 25¢ per day per item RETURNING LIBRARY MATERIALS: Place in book return to remove charge from circulation records- THE CYANOHYDRIN (KILIANI) REACTION: MECHANISM STUDIES BY 1 1JC NMR SPECTROSCOPY AND APPLICATION TO THE SYNTHESIS OF ISOTOPICALLY—ENRICHED CARBOHYDRATES By Anthony Stephen Serianni A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Biochemistry 1979 ABSTRACT THE CYANOHYDRIN (KILIANI) REACTION: MECHANISM STUDIES BY 13C NMR SPECTROSCOPY AND APPLICATION TO THE SYNTHESIS OF ISOTOPICALLY-ENRICHED CARBOHYDRATES By Anthony Stephen Serianni The classical Kiliani (cyanohydrin) reaction was studied by 13C NMR and GLC. 13C NMR studies were facilitated by the use of [13C]cya- nide and/or [13C]-enriched aldoses. The effects of aldose configura- tion, carbon—chain length and derivatization on the rate and extent of cyanide consumption, and on the overall rate of aldononitrile disap- pearance, were investigated. Hydrolytic intermediates in the reaction of KlBCN with D-erythrose were identified and characterized by 13C NMR and GLC with the use of standard compounds and/or NMR parameters. Reaction sequences, at several pH values, were determined from time- resolved reaction profiles produced from 13C NMR spectral data. At high pH (10.5, 12.7), the reaction sequence appears to be cyanide + D-erythrose-+ aldononitriles + imido-l,4-lactones + carbinolamines + aldonamides. Aldonamides hydrolyze (via carbinolamines and aldono- lactones) to aldonates. At lower pH (7, 8.5), the direct conversion 0f imido-l,4-lactone to aldono-l,4~lactone becomes appreciable. Am- monia, which is released in this reaction, can react with imido-l,4~ lactones to yield amidines. A by-product reaction between imido- l lactones and aldononitriles is proposed. 3C NMR parameters (5 and J) Anthony Stephen Serianni and GLC retention times for the reactants, intermediates and products are tabulated. From observations made during the study of the Kiliani synthesis, a new method for the preparation of isotOpically—enriched carbohydrates was developed. Cyanohydrins can be formed rapidly and essentially quantitatively at pH 8.0 i 0.5 with minimal hydrolysis, and they are stable at pH 4.0. The nitriles can be hydrogenolyzed to aldoses with palladium-barium sulfate (5%) at 1 atm to 60 lb in-2 and pH 1.7 to 4.2 depending on the structure of the nitrile. The mixed aldononitrile epimers are reduced without purification and the product aldoses puri- fied by chromatography. Aldononitrile phosphate epimers are separated, prior to reduction, by chromatography at pH 3.9. Using the above pro- cedure and KlBCN, C -C aldoses and C2-C5 aldose phosphates were pre- 2 6 . l3 . . pared with [ C]-enrichment at various carbon atoms. In addition to the introduction of carbon isotopes, catalytic hy- drogenolysis of cyanohydrins provided a route to carbohydrates enriched with hydrogen and oxygen isotopes. The technique permitted the simul— taneous incorporation of carbon and hydrogen isotopes at C-l and H—1, reSpectively, and oxygen isotOpes at 0-2 for each cycle of cyanide ad- dition and catalytic reduction, as shown in the following scheme. CHO H ao cnao KbCN bCN Pd, CH2 bCCHO I -2... _._. I I R -— R --— (H)(l3(aOH) _'—’ (mclxaoa) R R 13 C NMR and 1H NMR parameters for several [l3C]— and [2H]-enriched carbohydrates and their derivatives are tabulated and discussed in terms of configuration and solution conformation. To my parents, Elizabeth and Anthony, For For For The For And For For For the love and patience they showed, the home they created, the values in life they gave, and sacrifices they made. ears that heard, for their smiles. many an unselfish deed; carrying the load when their backs were tired. giving me life. ii ACKNOWLEDGEMENTS I am deeply grateful to Dr. Robert Barker for the guidance, friend- ship and encouragement he provided during the course of this study. I thank my colleagues in the laboratory for the countless discus- sions, where constructive criticism and encouragement were ideally interwoven. I am indebted to Ms. Terry Auld, who risked her sanity and sac- rificed her social life by patiently and meticulously typing this thesis. I would also like to thank the Biochemistry Department at Cornell University for permitting me to complete this thesis during the first few months of my appointment. I acknowledge financial support by a grant from the National Institute of General Medical Sciences, and thank the Stable Isotopes Resource of the Los Alamos National Laboratory for the supply of potassium [l3C]cyanide. iii TABLE OF CONTENTS LIST OF TABLES . LIST OF FIGURES LIST OF ABBREVIATIONS I. II. INTRODUCTION . A. B. EXPERIMENTAL . A. Carbon-113C) NMR Spectroscopy . B. Proton ( H) NMR Spectroscopy . 13. . . l . . . . . . . C- Computer Simulation of Complex C and H NMR Spectra D. Gas—Liquid Chromatography . . . . . . . E. Phosphate and Radioactivity Assays . F. pH Measurements . . . . . . . . . . . .'. G. Assays for Reducing Sugars, Aldonic Acids. Glycollic 73.9th Statement of the Problem . Survey of the Literature . l. Cyanide in chemistry and biochemistry 2. The Kiliani (cyanohydrin)reaction 3. Synthesis of monosaccharides . a. Aldoses and amino-aldoses b. Aldose phosphates . c. Deuterated carbohydrates . 4. NMR spectroscopy of carbohydrates a. 1H NMR spectroscopy b. 130 NMR spectroscopy . Acid, Formaldehyde and Cyanide . Catalytic Hydrogenolysis . . Chemicals . . . . . . General Syntheses and Purifications Analysis of Cyanohydrin Reaction Mixtures l. Gas-liquid chromatography (GLC) 2. 13C NMR spectroscopy . . . Preparation of Carbon-l3 Enriched Aldoses l. Pentoses and hexoses . iv Page vii ix xii l4 17 20 22 22 26 35 35 35 35 36 36 36 37 38 38 40 44 44 45 46 III. Page a. General method for the preparation of aldononitriles . . . . , , , , , . . 46 b. Catalytic reduction of aldononitriles . . . . . . 46 c. Purification of the reduction mixture . . . . . . 49 2. Formaldehyde, glycolaldehyde, glyceraldehyde and the tetroses . . . . . . . . . . . . . . . . . . . 50 a. Preparation of aldononitriles . . . . . . . . . 50 b. Reduction of short—chain aldononitriles . . . . . 50 c. Purification of the reduction mixture . . . . . . 51 3. Triose, tetrose and pentose phosphates . . . . . . . . 52 a. Preparation and purification of aldono— nitrile phosphates . . . . . . . 52 b. Hydrogenolysis of aldononitrile phosphates . . . . 54 c. Characterization of aldose phosphates . . . . . . 55 4. Enzymatic preparation of [13C]-enriched carbo- hydrates . . . . . . . . . . . . . . . . . . . . . . . 55 13 . - a. D-[Z- C Ribulose 1,5-P . . . . . . . . . . . . . 35 b. L-[3,4-l C]Sorbose 1,6— 2 . . . . . . . . . . . . 56 c. D—[z-l C]Fructose 1,6-P2 . . . . . . . . . . . . . 55 d. L-[2-13C]Glyceraldehyde 3-P . . . . . . . . . . . 57 M. Preparation of [2H]-Enriched Aldoses and Derivatives . . . 57 l. D-[l-l C, Clac, 2H]Erythrose and threose . . . . . . . 57 2. D-[2-—13C,14C,2H]Ribose and arabinose . . . . . . . 38 3. Methyl a-D—[2-13C,14C, 2H]ribofuranoside and methyl B-D-[2-13C,14C, 2H]ribofuranoside . . . . 59 4. Other compounds enriched with C, C and/or 2H . . . 59 N. List of Isotopically—Enriched Compounds 60 1- [13C1Pentoses and hexoses . . . . . 20 2. [13C]Lower- carbon aldoses (Cl-C4) and derivatives 61 3. [13C1Aldose and ketose phosphates and derivatives 62 4. Compounds with [2H]—enrichment . . . . . 63 RESULTS AND DISCUSSION . . 63 A. The Cyanohydrin Reaction . 1. Aldonic acid formation . . . . . . . . . . . . . . . . 22 2. Aldononitrile formation . . . . . . . . . . . . . 68 3. Rates of aldononitrile disappearance . . . . . . . 4 ants, intermediates Characterization of react 74 and products . . . . . . . . . . . 5. 6. 7. B. Preparation of [ The cyanohydrin reaction applied to D-erythrose Imidolactone formation . Conclusions l3Cl-Enriched Aldoses, Aldose Phos- phates and Their Derivatives . bump—4 Preparation of C6a1dononitriles . Hydrogenolysis o%% C5 and C6 aldononitriles . Hydrogenolysis of C2, C3 and C4 aldononitriles . Preparation of C3, C4 and C5 aldononitrile and aldose phosphates 13C NMR parameters . a. Short-chain aldoses and derivatives b. Aldose phosphates and derivatives i. Solution structure . ii. Assignment of chemical shifts iii. Carbon-phosphorus coupling constants . iv. Carbon-hydrogen coupling constants . C. Preparation of [2H]-Enriched Aldoses, Aldose Phos- phates and Their Derivatives . . . . 1. 2. 2 Hydrogenolysis with H2 NMR parameters . 13 a. C NMR parameters . b. 1H NMR parameters . 13 . . D. Enzymatic Conversions USing [ C]-Enriched Aldoses and Aldose Phosphates 1. 2. 3. 4. D-[2-13C1Ribose S-P to D—[2-13C1ribulose 1 5-P2 DL- [1-13 C]Glycera1dehyde 3-P to L- [3,’ 4—13CC]Sor- bose 1 6- P . . D- [2-13C]GIucose to D—[2-i3C]fructose 1. 6-P2 Action of glycerol kinase on DL— [2-1C]g1ycer- aldehyde . . . . . 13 . E. Miscellaneous Applications of [ C]—Enriched Carbohydrates . . . . 1. 2. Detection of aldehydo forms in solutions of [1- -13C]a1doses . . Use of 2JC2 H1 to determine aldose configuration and conformation . LIST OF REFERENCES APPENDIX . vi Page 86 104 113 124 126 126 133 138 146 146 156 156 161 164 164 168 168 169 169 172 183 183 186 189 192 195 195 198 201 211 Table I“) ‘1‘ 10 11 LIST OF TABLES Aldonate formation: Effect of pH and reactant concen- tration on the distribution of epimers . . . . . . . . GLC retention times of pertr imethylsilylated carbohy— drates and der -vatives . . . . . . . . . . . 13 - . . C NMR parameters or reactants, intermediates and products of the cyanohydrin reaction . . . . . . . . . Total percentage of intermediates and products in the cyanohydrin reaction (D-erythrose) at various pH values 0 O O O I O O O O I I O D O O 6 O O O O O O O 0 Relative amounts of arabino epimers produced during the cyanohydrin reaction (D-eryt hrose) at various PH val ues . . O O . O O O C O . . O 0 O O O O I I O 0 Specific reaction conditions and results for the preparation of aldononitriles . . . . . . . . . . . . Aldose yields based on weights, borohydride reduc- tion products, and radioactivity . . . . . . . . . . . Yields of aldoses from hydrogenolysis of two-, three-, and four-carbon aldononitriles . . . . . . . . Purification and epimeric distribution of aldono- nitrile phosphates . . . . . . . . . . . . . . . . . . 13 . ., - . . ~ . . C Chemical shirts or snort-chain carsonydrates and derivatives . . . . . . . . . . . . . . . . . . . l3, , . . . L Coupling constants of enriched caroohydrates and derivatives 0 I O O l I O O O O 3 O I C I I O O 0 13 .- . . . C Chemical shirts of aldose phosphates and re- lated compomds I O O I O O O O O 0 O O l O O O O O I Structural forms of the pentose 5-phosphates in aqueous solution . . . . . . . . . . . . . . . . . . . l 3C-P and 13C-H Coupling conscants of aldose phos- phates and related compounds . . . . . . . . . . . . . 13 13 13- 2 . C Chemical shifts,l3 C313 and c- H coupling con- stants for several [ C,‘H]-enriched carbohydrates and derivatives . . . . . . . . . . . . . . . . . . . ’ U {D 00 (D \J Ul 130 135 Table Page 1 16 Apparent and intrinsic H chemical shifts for :h getroses and some methyl pentofuranosides in ‘Sr‘o I C I O I O I I 9 O O O O I 0 O Q 0 O O O 0 d C a 0 I O 1.31. I. V o o a 9 o o q 10-, L- 17 Apparent and intrin51c geminal and Vicinai :~ 1 coupling constants for the tetroses and some methyl pentofuranosides in “H O . . . . . . . . . . . . . . . . . . 132 2 13 3C2,Hl Values for several carbohydrates and their derivatives . . . . . . . . . . . . . . . . . . . . . . . . 199 Figure [0 10 11 12 13 14 15 LIST OF FIGURES Cyanide-catalyzed decarboxylation of o-keto acids . Thiamine-catalyzed benzoin condensation and de- carboxylation of pyruvic acid . . . . . . Varma and French (46) mechanism of the Kiliani synthesis applied to n-D-arabinose Preparation of D—glucosamine Comparison between the traditional Kiliani—Fischer pathway (reactions 5, 6, 7) and the new pathway (reactions 1, 2) . . . . . . . Possible forms of D-erythrose in solution . 15.08 MHz 13C NMR and 1 0.04 MHz 1H NMR spectra of D—erythrose in H20 and H20, respectively . Newman projections of methyl B-D-mannopyranoside (A) and methyl o-D-galactopyranoside (B) . . Reaction vessel for the preparation of aldono- nitriles D C U U . O Q O O O O Q C O I l 0 Change in pH during the Kiliani reaction applied to D-erythrose . . . . . . . . . . . . . Disappearance of aldononitriles in cyanohydrin reaction mixtures . . . . . . . . . . . . . . . Differences in the rates of aldononitrile disap- pearance between diastereomers during the cyano- hydrin reaction at pH 8.5 t 0.2 . . . . . Gas-chromatograph of pertrimethylsilylated parent aldose, intermediates and end—products from the addition of cyanide to D-erythrose lH-Decoupled 15.08 MHz 13C NMR spegtrum of inter- mediates with 90.7 atom percent Il C]—enrichment at (2‘1 0 o o o o o o g o o e o a a o o o 1 H-Decoupled 15.08 MHz 13C NMR spectrum of inter- mediates with 90.7 atom percent [1 C]-enrichment at C-1 and C-2 ix Page 12 16 18 28 29 33 48 67 69 73 80 81 88 Page Figure 16 Cyanohydrin reaction profile at pH 12.7 . . . . . . . . . . 90 17 Cyanohydrin reaction profile from pH-quenched reactions at pH 12.7 . 93 18 lH-Decoupled 15.08 MHz 13C NMR spectrum of [13(3lintermediates after 11 min at pH 10.5, showing imido-l,4-1actone formation . . . . . . . . . . . . 94 19 Cyanohydrin reaction profile at pH 10.5 . . . . . . . . . . 95 20 Conversion of imido—1,4-1actones to aldono-1,4- lactones at pH <4 . . . . . . . . . . . . . . . . . . . . . 98 21 Cyanohydrin reaction profile at pH 8.5 . . . . . . . . . . 100 22 Effect of NHACl on the formation of intermediates . . . . . 103 23 Addition of [13C]cyanide to DL-[2-13C1erythrose . . . . . . 107 24 Addition of [13C]cyanide to DL—[3-13CJerythrose . . . . . . 109 25 Addition of cyanide to D-[l-lBCJerythro, threo-2,3- dihydroxybutanal . . . . . . . . . . . . . . . . . . . . . 112 26 Mechanism of the cyanohydrin reaction applied to D-erythrose o o o o o o o o o o I o o o o o o a e o o o o o 114 27 Rates of amidine formation from C, and C, aldono- nitriles . . . . . . . . . . . .4. . . 9 . . . . . . . . . 122 28 Change in H pressure during hydrogenolysis 3f hexononitriIes at pH 4.2 i 0.1 and 60 1b in . . . . . . . 128 . 13 13 29 Preparation of D—[Z- C]g1ucose and D-[Z- C]~ mannose from D-[l- C]arabinose and KCN . . . . . . . . . . 132 30 Preparation of D—[1-13C1erythrose and D-[1-13C1- threose from D-glyceraldehyde and K CN . . . . . . . . . . 137 31 Demonstration of the serial application of the synthesis . . . . . . . . . . . . . . . . . . . . . . . . . 140 32 Separation of DL-[1-13C]xylononitrile S-P and DL- [1-13C11yxononitrile 5-? and 13C NMR analyses of the products after hydrogenolysis over palladium . . . . . 143 33 Various forms of glyceraldehyde in aqueous solution . . . . 148 34 . 13 l . Determination of heteronuclear C- H coupling by C NMR . . . . 154 x Figure 36 37 38 39 40 41 42 43 13 . .. Structural and C NMR spectral relationsnips bet- ween the furanose forms of the pentose 5—phosphates and the tetroses Incorporation of 2H into a [13C hydrate . 180.04 MHz 1H NMR spectra of the H-2 to H-4 regions of D-threose and DL-[3-2H . . . . . . ]-enriched carbo- Jthreose . 180.04 MHz 1H NMR spectra of the H-2 to H-S regions 3f methyl S—D—ribofuranoside and methyl B-D—[Z-lJC, H]ribofuranoside . The enzymatic conVersion of D- [2— D-[2-13CC]ribulose 1, 5- P 2 l3C}ribose 5—? to as followed by 13C NMR . The enzymatic convef;~ ion of DL-[1-13Cngycera1de- hyde 3- P to L- [3, 4- by 1 C NMR “C]sorbose l, 6- P as followed 2 The enzymatic conversion of D-[2-13C]glucose to (FDP) as followed by D-[2-13C]fructose 1,6-P 130 NMR. .. 13 The C NMR spectra of DL-[2-13C 2 C]glyceraldehyde (A), 35 mM in water, and the mixture (b) produced by treatment with 1. 5 molar equivalents of Mg2+ -ATP and glycerol kinase at pH 7.5 for 3 h at 36° C . 13 The C NMR spectrum of D- [1—13C dioxane-water . xi C]erythrose in 80/ Page 188 191 194 197 ADP AMP ATP BSTFA 13C NMR FDP GLC H NMR Me Si Me Si NOE P P2 Pd/BaSO4 POPOP PPO TMCS TMSCN UMP LIST OF ABBREVIATIONS adenosine 5'-diphosphate adenosine 5'-monophosphate adenosine 5'-triphosphate N,0-bis(trimethylsilyl)trifluoroacetamide l3C nuclear magnetic resonance spectroscopy D-fructose 1,6-bisphosphate gas-liquid chromatography H nuclear magnetic resonance spectroscopy trimethylsilyl tetramethylsilane nuclear Overhauser enhancement phosphate bisphosphate palladium—barium sulfate 1,4—bis[2-(5-phenyloxazolyl))benzene 2,5-diphenyloxazole chlorotrimethylsilane trimethylsilylcyanide uridine 5'-monoph08phate xii I. INTRODUCTION The development of Fourier-transform 13C NMR spectroscopy in the late 1960's provided a powerful and practical new tool to aid in the elucidation of chemical structure and in the quantitation of the dyna— mics of chemical and biochemical reactions and interactions. The prob- lem of low C resonance sensitivity, caused by the low natural abun- dance of 1.1% (1) and a magnetic moment one-quarter that of 1H, prompted the use of Fourier-transform methods of detection and broad-band lH decoupling. The latter technique not only resulted in the elimination of l3C-lH spin multiplets but also produced a threefold enhancement of C resonances derived from a change in the Boltzmann distribution of H energy level populations during lH excitation. This enhancement is known as nuclear Overhauser enhancement (NOE). For studies that re— quire the determination of l3C-lH coupling constants, gated lH-de- coupling techniques are employed which permit lH-coupled spectra with NOE. Other alternatives to improve sensitivity include increasing the field strength of the spectrometer, lowering the temperature of the sample, increasing sample size and/or increasing the isotopic abundance of the 13C isotope. The last alternative, [13Cl-enrichment, is gener- ally time—consuming and expensive, and high-yield chemical and/or bio- chemical synthetic routes must be available. In addition, selective [13C]—enrichment is preferred, since 13C NMR spectra of uniformly en- riched compounds (>70 atom Z isotopic enrichment) are complicated by extensive homonuclear l3C coupling with a concomitant loss in sensi- tivity. However, [13C1-enrichment provides the only means of 1 .44 .f' D‘- . 13 13 evaluating spin-spin coupling between carbon, J13 13 , Since C— C C C coupling cannot be measured at natural abundance levels. Selective enrichment also allows easier evaluation of the fate of the enriched nucleus in chemical and biochemical conversions. Carbohydrates occupy a unique role in the study of organic chemis- try and a central role in the chemistry of biological systems. In the first sense, they are a class of poly-hydroxylic compounds especially suited for systematic study of configuration-conformation and structure- reactivity relationships since complete groups are available composed of compounds with similar empirical formula and carbon skeletons but different carbon stereochemistry. In the second sense. these compounds are involved in energy-producing catabolic reactions in biological sys- tems, in cell-cell interactions in the form of oligosaccharides on gly- coproteins and glycolipids, in the immune response as components of glycosylated antibodies, in biological structuresixnthe form of poly- saccharides like cellulose, and in energy storage in polysaccharides like starch and glycogen. 13C NMR is particularly suited for the study of carbohydrates. H-Decoupled 13C NMR spectra are not complicated by multiplets and are characteristically measured over 200 ppm of the applied field, which significantly decreases resonance overlap and facilitates the observa- tion of the various tautomeric forms. By comparison, 1H NMR spectra are measured over 10 ppm of the field, and are further complicated by H-lH coupling and non-first-order behavior. 1H NMR spectra of carbo- hydrates can be interpreted, however, with the aid of selective lH- decoupling, selective deuteration, computer simulation and/or high- field spectrometers. In cases where both 13C NMR and 1H NMR spectra are interpretable for [13C]—enriched compounds, chemical shifts of the protons and carbons, and homonuclear (lH—lH, l3C-l3 (lH—13C) coupling constants can be compared and evaluated in terms of C) and heteronuclear structure and conformation. 13 . . - . [ C]-Enrichment lowers the concentration or sample required for convenient detection into the range of most biological applications (uM) by significantly decreasing the acquisition time to obtain 13C NMR spectra of the enriched carbon. A. Statement of the Problem 13C NMR spectroscopy offers several advantages over traditional methods in the study of reaction mechanisms and intermediates. Spectra are frequently less complex than 1H spectra, while the tedium often associated with radioactive tracer work is eliminated in many instances. Specific [13C1-enrichment permits easy observation of a single nucleus during the course of a reaction by 13C NMR, and the simultaneous use of [13C]— and [ZHl-enriched reactants in tracer studies has been well established. Rates of chemical exchange (101-106 sec-l) can be calculated from the line-widths of the participating nuclei (2). Several studies of reaction mechanisms and reactive intermediates have been reviewed by Stothers (3). During this study, 13C NMR was employed to elucidate the mechanism of the classical Kiliani reaction. [13C]-Enriched reactants were em- Ployed to facilitate the detection of intermediates. The Kiliani reac- tion, which involves the addition of cyanide to an aldose,with subse- quent alkaline hydrolysis of the intermediates,was examined using 13 I Clcyanide and several aldoses. 'Identification and characterization of all intermediates and elucidation of the overall mechanism was ac- complished for the reaction as applied to ggerythrose. A new method for the incorporation of carbon, hydrogen and oxygen isotopes into carbohydrates was developed from observations made during the study of the Kiliani reaction. Aldoses, aldose phosphates and their derivatives were enriched with l3 l3 . C and H NMR parameters were measured and related to chemical struc- 14 2 C, C and/or H, and several ture, configuration and conformation. Several [13C]-enriched carbo- hydrates were converted enzymatically to biologically-important com- pounds to demonstrate their chemical integrity as enzyme substrates and the versatility of combining chemical and biochemical methods in the preparation of [13CI—enriched carbohydrate derivatives. B. Survey of the Literature 1. Cyanide in chemistry and biochemistry The chemistry of the cyano group is extensive and has been thoroughly examined by several authors (4). It is the intent of this brief review to discuss the salient features of chemical and biochemi- cal reactions involving cyanide in particular. Hydrogen cyanide, or hydrocyanic acid (HCN), is a weak acid (pKa = 9.21) (5). Cyanide ion is an ambident nucleophile, that is, it can react with electrophilic centers from the more electronegative nitrogen or from the more nucleophilic carbon. Silver cyanide, AgCN, generally reacts at the nitrogen to produce isocyanides. Boullanger, Marmet and Descotes (6) have recently reported the preparation of glycosyl iso- cyanides from the glycosyl halides by Walden-inversion at C-1. Gly- cosyl isocyanides, as well as other isocyanides, thermally rearrange In. to cyanides at elevated temperatures (>140°C). Alkali salts of hydrogen cyanide (KCN, NaCN) generally react as N 5 C-' with carbonyls in 1,2-addition to form cyanohydrins, or with conjugated carbonyl compounds in 1,4—addition to produce S-cyanoketones or Secyanoaldehydes. The latter reaction, known as hydrocyanation, has been recently reviewed by Nagata and Yoshioka (3). Cyanide adds readily to aldehydes in 1,2-addition, and racemic mix- tures result. Prelog and Wilhelm (7) have shown that stereoselectivity (M102 optical purity) can be achieved in the reaction of HCN with benz- aldehyde in chloroform by adding an optically—active alkaloid to the reaction mixture. Ketones are generally less reactive than aldehydes toward 1,2—addition. However, silyl derivatives of cyanide [trimethyl- silylcyanide (TMSCN) (8) and t-butyldimethylsilylcyanide (9)] react rapidly and completely with ketones to yield 2—0-trimethylsilyl cyano- hydrins. Other reactions include Strecker synthesis of a-amino acids (10, 11), and the asymmetric synthesis of amino acids by addition of cyanide to Schiff bases (12). The former reaction, shown in Scheme 1, is ter- molecular \\ x’ N + coon C e H3O l 3+NaCN+NH4Cl——’ -C— -—+-+ —C- , + NH2 N33 Scheme 1 and involves reaction between a carbonyl compound, NHACI and NaCN. Am- monia addition occurs either by displacement of OH-Z of the initially- formed cyanohydrin with inversion, or by formation of a Schiff base With the carbonyl prior to cyanide addition. The resulting d-amino \‘.. nitrile is hydrolyzed in acid to the o-amino acid. The asymmetric prep- aration of o-amino acids involves the formation of a Schiff base with an aliphatic aldehyde and an optically-active benzylamine. After addi- tion of cyanide and hydrolysis of the a«aminonitrile to the carboxylic acid, the 2°-benzylamine is reduced catalytically with Pd/C to the primary amine. Cyanide ion will displace halides in 8N2 reactions. For example, Bayly and Turner (13) prepared 2—deoxy-3,5-0-ethy1idene-D-[1-14C] ribononitrile from l—deoxy-Z,4-O-ethylidene-l-iodo-D—erythritol and K14CN in dimethylformamide at 45°C in 68 percent yield. Cyanide ion acts catalytically in benzoin condensation (14, 15) and in chemical decarboxylation of o—keto acids (16), as shown in Figure 1. This catalysis is dependent on the reversibility of cyanide addition and the ability of the -C E N moiety to stabilize a negative charge on the o-carbon through resonance. In biological systems, the thiazolium ring of thiamine acts as a -C E N equivalent (17-19), catalyzing biological benzoin condensation and decarboxylation reactions (Figure 2). The cyanogenic plant glyco— side, amygdalin (1), is composed of cellobiose CHZOH o 0 CH2 CN l 0 O-—-C I H 1 glycosidically-linked to 0H-2 of mandelonitrile. Two B-glucosidases (20) have been identified that cleave 1 to two molecules of glucose and mandelonitrile. A nitrilase or hydroxynitrile lyase (21) catalyzes the .mmaom Ouexlo mo :ofiumazxonumoon cmw>amumuummwcmxo .H ouswfih I s ,wmzollt Ibwzo o oo ,. 1 11 oz.o 2w 2w JV .mzo owwzo I - mo -mzo a IO \ IO - IO 0 2w 2w woeful]. oooo-.o£o lid. oooo-w-m:o oo :0 oo o 0 \e cult-("Loope \@ \ s 4.___.____. N . N N I ll —"’ l [I ——-- [I j Gk HO I I S //i\s Yj%\5 H ‘J\s HO'9°CH3 | . / ‘\N O l l \N HO-C-H ’5 HO\ ' I' l /C\ S e S \N + )I\ ll 00H It I CH3€H H0\ 5 1.__ CH3-C-C°CH3 ll CH3-C H O CH Thiamine-catalyzed benzoin condensation and decarboxy- Figure 2. lation of pyruvic acid (18). conversion of mandelonitrile to HCN and benzaldehyde as shown in Scheme II. 0 nitrilase ('3: <__'————-______-> -H+HCN H é—CN a. Scheme II The biological functions and formation of the cyano group have been re- viewed by Ferris (22). Hydrogen cyanide assumes a central role in primitive or pre-biotic chemistry, as discussed by Calvin (23). It has been detected as a first-transformation product in Miller electric-discharge experiments using methane-ammonia-water-hydrogen mixtures. HCN, its oligomers and polymers have been implicated in the pre-biotic formation of o-amino acids (Strecker synthesis), purines like adenine [(HCN)5], pyrimidines and porphyrins. 2. The Kiliani (cyanohydrin) reaction The preparation of 2-hydroxyacids from carbonyl compounds through cyanohydrins was an early discovery in organic chemistry. Simpson and Gautier (24) prepared DL-lactic acid from acetaldehyde and HCN in 1867. Staedeler (25) prepared a—hydroxyisobutyric acid by condensing HCN with acetone and hydrolyzing acetone cyanohydrin with HCl. In a series of papers, Kiliani (26) first applied the cyanohydrin reaction to the reducing sugars (Scheme 111). C-2 Fpimeric aldonic acids are formed from CN + coon CHO l H l | + HCN 1:: (H)—C-—(OH) -—-> (H)-C-(OH) R l. l R = (CHOH) CH OH n 2 Scheme III 10 the addition of cyanide to both faces of the planar carbonyl carbon. Kiliani (27-35) prepared epimeric aldonic acids from the following monosaccharides: L-arabinose, D-glucose, D-galactose and D—fructose. He also demonstrated that aldonic acids could be lactonized. Rupp (36) and Hudson (37) later modified the Kiliani reaction by using aqueous solutions of alkali cyanide and aldose in the presence of CaCl2 or BaCl2 instead of slightly basic liquid HCN. The cyano- hydrins (aldononitriles) were hydrolyzed in situ in the alkaline solu— tion. The Kiliani reaction has been applied analytically by Militzer (38) as a direct measure of reducing groups. In addition, Mednieks and Ninzler (39) reacted the following carbohydrate derivatives and complex carbohydrates with KIACN during [IACI-labeling studies on mucopolysac— charides and related substances: N-acetylglucosamine, fucose, glucuro- nate, hyaluronate hexasaccharide, chondroitin sulfate and hyaluronate. Militzer (40) completed the first systematic study of the effects of aldose structure, temperature and pH on the rate of cyanide consump- tion. Arabinose, galactose, glucose and. Z-ketogluconate reacted quan- titatively at 50 mM with a twofold excess of cyanide at pH 9.1, but at different rates. Reaction times for complete addition at 25°C varied from 3 h for arabinose to 72 h for 2-ketogluconate. Glucose reacted with cyanide completely after 18 h at 25°C. Militzer (40) noted that reducing disaccharides gave rates similar to glucose, but that di- saccharides with B-linkages (cellobiose, lactose) gave noticeably slower rates than those with a-linkages (maltose, melibiose). Dif- ferences in the rates of reaction were explained in terms of the amount of aldehydo form in solution. 11 Militzer (40) observed that increasing temperature from 25°C to 40°C caused a significant increase in the rate of cyanide consumption while lowering the pH of the reaction from pH 9.2 to pH 6.5 caused a significant decrease in the rate. At acid pH ( .m ounwflh m mefizomdé m WZOBUQH 02084 m HQH2¢ZOQQ4 Mmmuu.l m MZHE OZH mmdo oN m + mmzoeofiuméuoonH llllllulsl maHmeHzozooad ao 1r 13 order of appearance and disappearance of the prOposed intermediates and the effect of pH on these events, (e) explain the observed effect of pH on the ratio of epimeric products (41), and (f) provide evidence for the existence of tetrahedral (carbinolamine) intermediates. The present study of the Kiliani reaction resulted from the desire to prepare [13C]-enriched aldoses and their derivatives in high yield based on [13(3]cyanide. The reaction of several carbohydrates and their derivatives with cyanide was examined under a variety of con- ditions of pH, temperature and concentration of reactants using 13C NMR and gas-liquid chromatography (GLC). The effects of aldose configura- tion, carbon-chain length and derivatization on the rate and extent of cyanide condensation, and on the rate of aldononitrile disappearance were investigated. To study the intermediates in the hydrolysis reaction, the conden- sation of K13 CN with D-erythroseanzl8°C and 0.3 M was examined. The appearance of hydrolytic intermediates was recorded as a function of time by 3C NMR analysis of reaction mixtures. Reaction intermediates, when feasible, were prepared by standard routes with [13C1-enrichment at specific sites to provide standard l3C chemical shifts and l3C-13C coupling constants, to provide standard GLC retention times, and to per— mit examination of their hydrolysis individually. D-[1-13C], [241%3]and [3-l3tllErythrose were used as parent aldoses to assist in the charac- terization of the intermediates. Data are interpreted in terms of fore mation and stability of intermediates, and their routes of hydrolysis under various reaction conditions. 14 3. Synthesis of monosaccharides a. Aldoses and amino-aldoses Classical routes for the preparation of monosaccharides include Wohl degradation (47), Ruff degradation (48), controlled oxidation of polyols with lead tetraacetate [Pb(OAc)4] and sodium periodate (NaIOA), Kiliani-Fischer synthesis (49, 50),nitromethane synthesis (51. 52), dia— zomethane synthesis (53), enzymatic syntheses, and epimerizations. The Kiliani-Fischer synthesis of aldoses is perhaps the most widely recognized method. The Kiliani reaction, discussed in the previous sec- tion, was extended by Fischer (49), who discovered that aldonolactones could be reduced with sodium amalgam (Na/Hg) to aldoses. The Kiliani- 4C]- Fischer reaction has been used extensively for the preparation of [1 labeled aldoses from [14(31cyanide by Isbell and his colleagues (41-45). This application and others that employ isotopically-enriched cyanide have been reviewed by Pichat (54). The method has been applied to the preparation of the higher-carbon aldoses, that is, aldoses having more than a six-carbon linear skeleton. The utility of the method is demon- strated by the work of Philippe (55). who lengthened D-glucose up to D- glucodecose (Clo) by serial application of the synthesis. The prepara— tion of higher carbon aldoses and alditols has been reviewed by Hudson (50). In recent years, aldonolactones. or their acylated derivatives, have been reduced with a variety of reagents, including diborane in THF (56) and disiamylborane in THF (57). In this regard, Guidici and Fluharty (58) reduced D-erythrono-l,4—1actone with disiamylborane in THF to produce D-erythrose in 60% yield. Although acylated aldoses are obtained in high yield (>90%)liyreduction of protected aldonolactones .44 15 with disiamylborane, removal of the acyl groups is often accompanied by degradation of the product aldose. It is apparent that [13C1-enriched pentoses and hexoses can be pre- pared in good yield by the Kiliani-Fischer synthesis, and D—[lvl3(3]g1u- 13 cose, D-[1-13(3]mannose and D-[l- (Ilgalactose have been prepared in this fashion (59). However, only [1—13CI- and [2-13CJ-enriched hexoses l3Claldoses cannot be prepared by are accessible, since the C2-C4 [1- the Kiliani-Fischer reaction. .The amino-sugar, glucosamine, was originally prepared (60) by the addition of HCN to D-arabinosylamine, hydrolysis of the 2-aminonitrile with concentrated HCl to the corresponding acid, dehydration to the 2- aminolactone. and reduction to the 2-amino-2-deoxya1dose with Na/Hg. The reaction scheme is shown in Figure 4. The overall yield is about 1 percent. The reaction was modified by preparing the 2-aminonitriles from the starting aldose with HCN in the presence of ammonia or other amines (Strecker addition). Kuhn and his associates (61-64) later de- monstrated that 2-aminonitriles, which are isolable, could be reduced with palladium directly to 2-amino-2-deoxyaldoses , as shown in Figure 4, in yields based on the aldose of more than 70 percent. Kuhn and Klesse (65) demonstrated that D-glucono- and D—mannono- nitriles could be prepared in PYridine and reduced catalytically in di- lute acid with palladium—barium sulfate (Pd/BaSOa) to D-glucose and D- mannose, respectively. Bayly and Turner (13) have used platinum oxide to prepare 2-deoxy-D-erythro-[l-JA’Clpentose from the corresponding nitrile. During the investigation of the mechanism of the Kiliani reaction, we observed that aldononitriles form rapidly and essentially 16 .AHOV womanhoozawxn mo coaumnmeowm .q ownwwm :o~:n.o illll.mumV++ Ti . ® :8: 0:62 IOOI NI M . o:zo: :ou: IIIIIIIIL .01 .NI\Un_ :oN:H.o :ofw :ofmo :8: :8: :8: low: :8: on: :8: :8: 1% :oo: 14%... :00: m. . m. . N . e:zn.v: e26: :zc: Flouo :08 20 17 quantitatively at pH 8.0 i 0.5 with minimal hydrolysis, and that they are stable at pH 4.0. Aldononitriles were hydrogenolyzed to aldoses in 70 to 80 percent yield with palladium-barium sulfate at pH 1.7 to 4.2 and 1 atm to 60 1b. in.2 H2, depending on the structure of the cyanohy- drin. The reaction was used to prepare [13C1—enriched C2 to C6 aldoses. The reaction scheme is shown in Figure 5 and is compared with the tradi- tional Kiliani-Fischer reaction. b. Aldose phosphates The formation of phosphoric esters of carbohydrates is an essential stage in biological synthesis, interconversion and degradation of pen- toses and hexoses. For example, D-glucose is phosphorylated by hexo- kinase and Mg2+-ATP to form D-glucose 6-P during glycolysis. D-Glucose 6~P is isomerized by phosphoglucoisomerase to D-fructose 6-P or is con- verted to D-glucose 1—P by phosphoglucomutase. D-Glucose l-P is a pre- cursor in the enzymatic synthesis of the polysaccharides amylase, amylo— pectin and glycogen. Studies on these and other metabolic events re— quire the appropriate phosphate esters to measure enzyme activity and inhibition, to investigate enzyme-substrate interactions, and to exa- mine the solution structurescifthe phosphate esters. Chemical syntheses of aldose phosphates commonly involve the phos- phorylation of protected carbohydrate derivatives. For example, 1,2— epoxides have been used in the preparation of glucose 3-P (66) and glu- cose 6-P (67). l,2-0-ISOpropy1idene-D—xylose has been treated with di- phenyl phosphochloridate to yield the 5—pheny1 phosphate, which is hydrolyzed in alkali and acid to produce a mixture of D—xylose 3-P and D-xylose S-P (68). Stverteczky et a1. (69) prepared D-arabinose 5-P from 5,6-anhydro-3-0-benzy1-l,2-0-isopropylidene-D-glucofuranose (2) by 18 ll . . . a .muosmouatxn cowuuwmu ohm : mom 0 mmc:0QEOG dwwwWwM m .chuomaocomnm n w .mwom owcomum u w .HOanHw>xooptho:HEmnm u o umomam Mm3MDmn 3o: .acHEweam u e .oamtnacoeoeam n a .wnoeae meantenm u a .Am a meowtuaetw conflueaeoo .m seamen use mcm Ac .m .c mcomuumonv zmzcumn umzomwmlficmwawx Hmcofiuflmmwu mzu :mm3u n r. ll ; a a n :o~:o e a a son: angel 0 at + 5.: ~:z~:o :08 m o :o o n. .v ~__ g 20.: 1...! eoizvieeoihlly 1.. o :25 2o 05 1. 2o: e o a w 19 /’C32 o\| CH O OBn Bn = benzyl 0 0% L CH3 2 CH 3 phosphorylation of the latter compound with KZHPO4 to produce the 6- phosphate. Acid hydrolysis, periodate cleavage and hydrogenolysis yields D-arabinose 5-P. Michelson and Todd (70) prepared D-ribose S-P by phosphorylating 2,3v0~isopropylidene-D—methyl ribofuranoside. 1,3,4-Tri-0-acetyl-N-acetyl-B-D-glucosamine yields, after treatment with diphenyl phosphochloridate, hydrogenolysis with Pro and acid by- 2 drolysis, D-glucosamine 6-P (71). Maehr and Smallheer (72) have recently prepared D-arabinose 5-P and D—xylose S-P from methyl o—D-arabinofuranoside and 1,2-0-isopro- pylidene-D-xylofuranose, respectively, by treatment with dibenzyl or diphenylphosphochloridate. The resulting 5-esters were hydrogenolyzed and hydrolyzed in acid to afford the product aldose phosphates. Ballou and Fischer (73) and Ballou et a1. (74) prepared D-glycer- aldehyde 3-P from mannitol in nine steps and D-erythrose 4-P from D- erythrose in eight steps, respectively. These syntheses yield the C3 and C4 aldose phosphate as the dimethyl acetal derivative which, unlike the free aldose phosphate, can be stored without decomposition. Treat- ment with Dowex 50 X8 (H+) affords the free aldose phosphate. Klybas et al. (75) prepared D-erythrose 4-P directly by lead tetraacetate oxi- dation of D-glucose 6-P. Aldose phosphates can be prepared enzymatically in many instances through the use of isomerases and kinases, as discussed by Leloir and 20 Cardini (76)- . l3 . . - The preparation of [ CJ-enriched aldose phosphates, therefore, could be accomplished by these methods, but often from [13C1-enriched precursors frequently more complex than the aldose phosphate. However, addition of [13(31cyanide to C C and C aldose phosphates at pH 8.0 2’ 3 4 yields C3, C4 and C5 aldononitrile phOSphates almost quantitatively. Hydrogenolysis of the aldononitrile phosphates with Pd/BaSO4 at pH 13 C]aldoses with terminal 1.7 r 0.1 and atmospheric pressure afford [1- phosphate esters in 75-85% yield. The method was applied serially to produce aldose phosphates enriched with 13C at positions other than C-l. For millimolar preparations of aldose phosphates, procedures are des- cribed to prepare glycolaldehyde—P and D-glyceraldehyde 3«P in 80-90% yield from economical starting materials for use as parent aldose phos- phates in cyanide condensation. c. Deuterated carbohydrates The preparation of deuterated carbohydrates is frequently accom- plished by reduction of carbonyl derivatives with NaBzHa. Lemieux and Stevens (77) prepared 1,2:5,6-di-0—isopropylidene-a-D-glucofuranose-6, 6Ld2 by reduction of l,2—0-isopropylidene-a-D-glucofuranurono-6,3-lac- tone with LiAlea. Gray and Barker (78) prepared D-glyceraldehyde-3, 3'--d2 3-phosphate by reduction of Z-O-benzyl-D—arabinono-l,4-lactone with NaBZH to produce the intermediate 2-0-benzyl-D-arabinitol-l,l'—d 4 d-D-Galactose-6,6'-d 2. 2 was prepared by reduction of methyl D-galacturo- nisides with NaBzH4 (79). Other methods of deuteration include base- catalyzed lH-ZH exchange in 2H20 using 2-0-benzyl derivatives (77), re- 2 duction of aldonolactones with sodium amalgam in H20 (80), and enzyme- catalyzed solvent exchange (81~84). These methods have been 21 discussed by Barnett and Corina (85). Koch and Stuart (86) and Balza et a1. (87) have recently prepared deuterated carbohydrates (methylglycopyranosides, oligosaccharides, nucleosides) by lH-ZH exchange in 2H20 in the presence of Raney nickel. Nucleosides are fully deuterated in the base portion, with no incorpora— tion of H detected in the glycosyl moiety. Position(s) of deuteration in the carbohydrate derivatives were determined by 13C or 1H NMR since incorporation was not always predictable or specific. The traditional methods of deuteration, therefore, are limited by the availability of the appropriate carbonyl compound or by the non- specificity of incorporation. In addition, reduction of keto deriva- tives is complicated by the formation of two diastereomeric products, which must be separated. Hydrogenolysis of aldononitriles provides a convenient and simple method for the incorporation of 2H into the simple aldoses and aldose phosphates in 70 to 80 percent yield. An aldose is condensed with cya- nide at pH 8.0 1 0.5 to produce the 2—epimeric cyanohydrins, which are stabilized by lowering the pH to 4.0. The cyanohydrins are hydrogeno- lyzed in 2H20 solution with 2H2 over palladium to yield the [1-2H1- enriched 2-epimeric aldoses having one more carbon than the parent com- pound. The aldoses can be separated as described previously (88, 89). Izhl-Enriched aldose phosphates can be prepared from aldononitrile phos- phates in a similar fashion. IsotOpes of oxygen (170, 180) are commonly introduced into carbo— hydrates by oxygen exchange between a carbonyl derivative and isotopi_ tally-labeled water (90) and are trapped by reduction of the carbonyl With NaBHa. A complementary method involves oxygen-exchange between 22 0-1 of an aldose and isotopically-labeled water followed by trapping with cyanide. Hydrogenolysis of the [2-170,l80] cyanohydrins affords the corresponding aldoses enriched with the oxygen isotope at OH—2. The method of preparing aldononitriles and reducing them with palladium-barium sulfate provides a means for the independent or si- multaneous incorporation of carbon, hydrogen and oxygen (91) isotOpes into the carbohydrate molecule, as shown in Scheme IV. cHo Hzao caao KbCN pr Pd, CH2 bICCHO R """"' R (H)C(aOH) (H)(|3(aOH) l R R Scheme IV Starting materials are readily available and the reaction can be applied serially, permitting the preparation of a wide variety of isotopically- enriched carbohydrates and their derivatives. 4. NMR spectroscopy of carbohydrates a. 1H NMR spectroscopy A major interest in the study of carbohydrates lies in the deter- mination of the various forms present in solution and in the solid state. The advent of 1H NMR spectroscopy was of utmost importance for the determination of anomeric configuration and conformation of carbo— hydrates and their derivatives in solution. 1H NMR provides a means to distinguish a- and B—forms of the pyranosyl ring based primarily on H-l chemical shifts and H-1 —-H-2 coupling constants. The predominant con— formers (chair) of the pyranose ring (1C4 and 4C1) are separated by large energy barriers, and, therefore, can usually be distinguished from the magnitudes of the coupling constants between H—1 and H-2 23 (Scheme V). Conformational analysis of the furanose ring is complicated by a dynamic equilibrium between the various forms (pseudorotation) with small energy barriers between individual conformers (W12-20 KJ mole -1). The two major forms are the enveIOpe (E) and twist (T) forms, where one or two atoms are outside the plane of the ring, respectively. If the rate of interconversion between conformers is sufficiently rapid, the observed chemical shifts and couplings represent weighted averages of chemical shifts and couplings of the individual conformers. Studies on the configuration and conformation of carbohydrates in solution using 1H NMR are based on the dihedral-angle dependence of vicinal (three-bond) proton coupling constants discovered by Karplus (92). This relationship states that dihedral angles of 0° and 180° produce maximal lH-lH coupling (8-10 Hz) while a dihedral angle of 90° produces a minimum value (0-1 Hz). The equation had the following form, 3 J = A + Boos 0 + Ccos 2 0 H,H' where 0 = dihedral angle, A = 4.42 Hz, 8 = -O.5 Hz and C = 4.5 Hz. The equation was calculated based on a carbon—carbon bond length of 1.543 A with both carbons sp3-hybridized. Karplus (93) has emphasized, however, that the relationship between coupling constant and dihedral angle is subject to parameters other than dihedral angle, namely, the electro- negativity of substituents and carbon-carbon and carbon-hydrogen bond lengths. Estimations of dihedral angles to an accuracy of one or two degrees is tenuous at best. The use of 1H NMR to determine anomeric configuration and 24 conformation of carbohydrates and other cyclic compOunds has been dis- cussed by several investigators (94-98). The composition and conforma- tion of carbohydrates in solution has been reviewed by Angyal (99). Conformations of deoxyribose and ribose moieties in nucleosides. nu- cleotides, and oligonucleotides were recently discussed by Davies(100). In brief review, Hall (101) examined the conformations of several benzoylated ribofuranosides by 1H NMR using a modified Karplus equa- tion for application to the carbohydrate molecule. 2 J = J cos 6 - 0.28 where Jo = 9.26 Hz for 0‘: 6‘: 90° and Jo = 10.36 Hz for 90° 2 8 1 180°. Results from this study showed that l,3,5—tri-0-benzoy1-a—D-ribofuranose has the El conformation, that is, C-2, C-3, C-4 and 0-4 are essentially co-planar with C-l above the plane defined by these nuclei. In 1966, Lemieux and Stevens (77) examined the 100 MHz 1H NMR spectra of D—xylose, D~1yxose, D-arabinose, D-ribose, D-glucose, D- mannose and Degalactose. Anomeric protons were observed at 4.6-5.3 ppm relative to tetramethylsilane (MeASi), and the remaining protons were ob- served upfield (3.3-4.2 ppm). Generally, axial protons on the pyrano- syl ring were found to produce signals at higher field than equatorial protons when the compounds considered were epimeric and both in the same chair conformation. This correlation facilitated the determina- tion of the proportions of anomeric forms in solution. Vicinal c0up~ ling between H-1 and H-2 was used to establish preferred conformations, since this parameter is often sensitive to conformational change, as shown in Scheme V. Conformational preferences based on 1H NMR data 25 are in close agreement with those determined from free energy calcula— tions (99). H1 0 O H A 2 <:— H2 4C1 = as H1 1C4 = ee dihedral angle = 180° dihedral angle = 60° . 3 _ . 3T * predicted JH1,H2 — 8.6 11.5 Hz predicted ”H1,H2 0.6 3.5 Hz Scheme V 1H NMR In 1967, Stevens and Fletcher (102) examined the 60 MHz spectra of several furanoid derivatives of D-arabinose, D—lyxose, D— ribose and D-xylose. As discussed above, conformational analysis of the furanoid ring is complicated by pseudorotation, resulting in uncertain- ties in the dihedral angles between vicinal hydrogen nuclei. Despite these difficulties, preferred conformations were assigned to the various pentofuranosides, although conclusions were highly speculative. The study was also complicated by near magnetic-equivalence of coupled nu- clei causing second or greater order behavior in the spectra. Equations . . . 1 l . were required to determine intrinSic H— H coupling constants from the experimental (apparent) values. In addition, compounds were studied in deuterated chloroform and extrapolation to conformation in aqueous media. where water may play a role in ring stabilization through hydrogen bond— ing, cannot easily be made. Stevens and Fletcher (102) also observed that, generally, furanoid rings with OH—l and OH-2 trans have H-l at higher fields than those With OH-l and OH-2 cis. In 1971, Alfoldi et al. (103) studied the conformationsof several methyl O-methyl-D-xylofuranosides in pyridine by 1H NMR and concluded -a .m: ..o 26 - 2 2 2 . . that the d-anomers preter Tl, E and T3 conformations, while the 8- anomers prefer 2T3, 3E and 3T4 conformations. Angyal and Pickles studied the equilibria between the pyranoses and furanoses for the aldoses (104) and deoxyaldoses (105) by 1H NMR, and, recently, De Bruyn and Anteunis (106) established the conformation of S-L-arabinopyranose (4C1) in aqueous solution by 1H NMR at 300 MHz . Horton and his colleagues (107) have used 1H NMR to determine the conformation of linear carbohydrates and their derivatives in solution. In general, 1H NMR spectra of the free aldoses are complicated by the presence of two or more forms in solution. Line-multiplicity and resonance overlap hinder full interpretation, especially for the pen- toses and hexoses, which have six and seven non-exchangeable protons for each form, respectively. Selective lH-decoupling, selective deutera— tion and analysis at high magnetic fields have been utilized to facili- tate the analysis of complex 1H NMR spectra. It is noted that anomeric configuration and conformation cannot be easily determined by 1H NMR of the cyclic ketoses, since these compounds do not possess an anomeric proton. b. 13C NMR spectroscopy 13C NMR spectroscopy offers an alternative to 1H NMR spectroscopy for configurational and conformational analysis of carbohydrates, but more often complements 1H NMR. In contrast to protons which typically resonate over 10 ppm of the applied magnetic field, carbons typically resonate over 200 ppm, which minimizes apparent magnetic-equivalence between dissimilar carbons. 13C NMR spectra are normally obtained with broad-band 1H-decoupling (108) which effectively removes line- multiplicity due to 13C-lH coupling. Although information from 13C-lH 27 coupling is lost, lH—decoupled 13C NMR spectnaaresimple and usually more easily interpreted than 1H NMR spectra. lH-Coupled spectra can be obtained, however, when l3C-lH coupling constants need to be evaluated. Moreover, 13C NMR spectra of un-enriched compounds are not complicated by 13C_13 C coupling, since 13C has a low natural abundance (1.1 percent). Carbon spectra can also be obtained in water, a convenient and common solvent for carbohydrates. without solvent-line interference. 130 NMR has been used to determine the various forms of carbohy- drates present in solution. For example, D-erythrose can potentially exist as five or more forms in solution, as shown in Figure 6: d- furanose, B-furanose, aldehydo, hydrate and dimers and/or oligomers. The 15.08 MHz 13C NMR and 180.04 MHz 1H NMR spectra of D-erythrose are shown in Figure 7. The carbon spectrum is noticeably simpler. The anomeric regions for C-l (89-105 ppm) and H-1 (5.00-5.30 ppm) are easily identified as dowhfield from the remaining nuclei. Three major tauto- meric forms are observed in both spectra: d—furanose, B-furanose and hydrate (h). Resonance assignments were made as described in later sections. The lH-decoupled 130 NMR spectrum shows one resonance for each carbon, whereas the 1H spectrum shows multiplets for each proton that arise from lH-lH coupling. The proton spectrum of D-erythrose is relatively simple compared with those of other aldoses. 13C NMR offers a means to determine the tautomeric equilibria of .the ketoses, since anomeric carbons are observed in the spectra. Que and Gray (109), Angyal et al. (110) and Angyal and Bethell (111) have studied equilibrated mixtures of the ketohexoses, l-deoxyhexuloses, and 13 3-hexuloses, respectively, by C NMR. Furanose, pyranose and keto forms were detected.but keto hydrate forms were not observed in the 28 0 H OH H H H H 0H OH OH OH (3 - fu ranose 0k - furanose 9 ‘3“ C-H HC-OH HCOH HCOH HCOH HCOH CHZOH CH20H aldehydo hydrate 0H HC-OH 9H HC- 0 —-—- CH HCOH HCOH CHZOH HCOH CHZOH dimers and oligomers Figure 6. Possible forms of D-erythrose in solution. 29 .>~o>fiuoommou N N .o =~ see c = a“ mmou;u>u01a mo wuuoodm «:2 2 ~22 «0.0ma mom mzz 0 we: wo.ma .A beamed H ma .:33 mm .a S n 3.4. a: one .81.“. mmn3 5: : e a 3533-: «<33? 3 2, a: 5 €14 :11 i 6.3. BA: ._ _: . 5 .2 6-2.. 6N2 aw: 3.... «5° :32 :— \3 .. v) ‘4‘ .1 (x .<3./. ... ea 8 Au 93 Q -3 sanded szo 2 ha mo— ld a z) fi)n\/\33\Jvl .tilll». Q._u 30 compounds studied. Assignments of 13C chemical shifts are established by several methods: (a) the effect of substitution (derivatization) on chemical shift, (b) the magnitudes of scalar or one-bond C- H coupling con- stants, (c) deuterium isotope shifts, and (d) selective [ 3CI-enrich- Rules derived from the effects of substitution on 13C chemical (S9). and ment . shifts are not always reliable, as discussed by Walker et al. incorrect assignments have been made based on these rules (112. 113). Ritchie et al. (114) examined the carbon spectra of the methyl glyco- furanosides and cyclopentanols and assigned 13C chemical shifts based on substituent effects and shielding factors. For the furanoid ring, however, changes in shielding cannot be totally attributed to configura- tional factors since conformational components, presently unknown, pro- bably make a major contribution to shielding. Bock et al. (115), Bock and Pedersen (116) and Bock and Pedersen l3 (117) have measured direct C-lH coupling constants (lJC H) in the hexopyranoses, pentopyranoses and their derivatives and have specifi- cally related lJCl Hl to anomeric configuration and conformation. l JCl H1 is approximately 10 Hz smaller (N160 Hz) when H-l is axial than 13 l . C- H coupling constants for other methine carbons of carbohydrateS(e-8-, ljcz H2 when H-l is equatorial. One-bond ) are approximately 20-25 1 H2 smaller than JCl.Hl‘ Gorin (79) and Gorin and Mazurek (118, 119) have exploited the ef- 13 fects of deuterium substitution on 13C and 1H NMR spectra to assign C chemical shifts. A signal from a 13C nucleus directly bound to a deuteron either disappears or is converted to a triplet at O l-0.5 ppm higher field. Progressively smaller upfield shifts are observed for 31 signals of the B-carbon(s) (%O.l ppm) and y-carbon(s) (No.01 ppm). From isotope shifts, carbon nuclei that are one-,two-and three—bonds from the deuteron can be assigned. Carbon—l3 chemical shifts are easily assigned by specific [13C]- enrichment. The [13C]-enriched nucleus is assigned based on its en- hanced detection. An d-carbon will be coupled to the I Clvenriched nucleus, causing splitting of its resonance into a doublet. The en- riched nucleus and carbon(s) a to it are, thereby, unequivocally as- signed. Walker et al. (59) prepared [l—l3(3]hexoses and hexopyrano- sides and assigned the 13C chemical shifts of C-1 and C-2 based on these effects. 13 Tweebond (ZJC,H) and three-bond (3JC,H) C-lH coupling constants have been measured and related to configuration and conformation. Schwarcz and Perlin (120), Schwarcz et al. (121) and Cyr et al. (122) demonstrated with several carbohydrates and their derivatives that the magnitude and sign of two-bond coupling between 13C and 1H depends on the orientation and electronegativity of the substituents appended to the 13C nucleus relative to the proton. Generally, ZJCl H2 is large (“5 HZ) when H—2 is gauche to both oxygen atoms on C—l. Perlin and his colleagues (121, 122) have observed that the sign of 2JC,H depends on configuration, with an oxygen anti to the coupled proton making a posi- tive contribution to coupling and a gauche oxygen making a negative contribution. Walker et al. (59) measured 2JC1,H2 coupling constants in [l-l3(3]hexoses and hex0pyranosides and in [1-l3(3]2-amino-2- deoxyaldoses (123) and found them to be consistent with these rules. 7 Bock and Pedersen (124) have recently related 'JC H to configura- tion and conformation by considering all the oxygen substituents on \.no 32 both carbons. For example, consider the Newman projection of methyl B-D-mannopyranoside (Figure 8A) viewed from C-2 to C-1. If the projec- tion of the C-2-rOH-2 bond on an axis trans to the C-l-H-l bond is given the value +1.0 (cos 0°), then the projections of C-1-OH-l and C-l-O—5 will each be +0.5 (cos 60°) and the projection-sum will equal +2.0. Similar projections obtained from the Newman projection of methyl a-D—galactopyranoside (Figure SB) give a sum of +0.5 (+0.5 from C-l-0-5 and C-l-0Me and —0.5 from C-2-0H-2). Projection-sums were related to coupling constants as follows: +0.5 = +0 Hz; +1.0 = 0-1 Hz; +1.5 = +5.5 Hz; +2.0 = +9.0 Hz. The projection-sum rule (124) was determined from the study of several pyranosyl carbohydrate deriva— tives and the furanose rings of 3-0-acetyl-l,2:5,6-di-O-isopropylidene- d-D-allofuranose and 3-0-acety1—l,2:5,6-di-0-isopr0pylidene-a-D-gluco- furanose. Schwarcz and Perlin (120) have established a Karplus relationship between 3JC H and dihedral angles for the carbohydrates: 60° 2 2 Hz, 100° = 0 Hz, 180° = 6 Hz. Walker et 1 (59) easured 2J and 2J in the [l-lBC]- a ' m c1,c3 c1,c5 hexoses and hex0pyranosides and discussed these parameters in terms of a dihedral angle dependence, where the angle is defined by the rela- tive orientations of C-3 and C-5 and the electronegative oxygen sub— stituents. BJCOCC coupling between C-1 and C-6 was also observed. Marshall and Muller (125) and Barfield et al. (126) have discussed the angular dependence and substituent dependence of 3JC C for ali~ Phatic carboxylic acids and alcohols, and alicyclic alcohols, reSpec- tiVely. Dihedral angles of 0°, 80° and 180° produce BJC C values of 2 HZ: 0 Hz and 4 Hz, respectively. 3JC C in the carbohydrates has not .Amv wwwmosmu>10uumamwla 32¢ 33 to Hazuma cam Amv mwwmocmuznoccmslalm Hazuma mo menauummoun :mEBmz .m munwflm < 34 been systematically examined. In summary, the analysis of carbohydrates by 1H and 13c NMR has 9 been limited to a large extent by the availability of [‘ Cl- and [“HI- , 13 . .. . enriched derivatives. C Chemical snifts can be unequivocally esta- blished with [13C1-enriched compounds. Tautomeric forms in solution . . . , . l3 . can be determined with carbohydrates enricned with C at the anomeric carbon. Orientational and substituent dependencies of JC H and JC C’ which are useful probes of configuration and conformation in solution, can be ascertained only with model compounds enriched at specific sites 13 2 with C and/or H. The full interpretation of complex 1H NMR spectra . 2 . . . . of carbohydrates is greatly facilitated by [ H]-enricned derivatives. With this need in mind, a method was developed to incorporate isotopes of carbon, hydrogen and/or oxygen at almost any site into carbohydrates and their derivatives. A vast array of isotopically-enriched compounds can be prepared for chemical, biochemical, physical and biomedical uses. II. EXPERIMENTAL A. Carbon-l3 (13C) NMR Spectroscopy 13C NMR spectra were obtained using a Bruker WP-60 Fourier- transform spectrometer equipped with quadrature detection and operating at 15.08 MHz for carbon. Spectra were obtained with a K real Spectral points and spectral widths of 2400 Hz or 3000 Hz. Filter widths of 2400 Hz or 6000 Hz were employed. The spectrometer was locked exter- nally to the resonance of 2H20 in a capillary. Chemical shifts are re- ported relative to external tetramethylsilane(Me431)3uu1areaccurate to i0.l ppm. B. Proton (1H) NMR Spectroscopy 1H NMR spectra were obtained at 30°C in 2H20 using a Bruker WH-lBO Fourier-transform spectrometer operating at 180.04 MHz for 1H. Spectra were obtained with 8 K real spectral points and aspectral width of 1300 Hz. Chemical shifts are reported in ppm downfield from internal sodium 3- (trimethylsilyl)-1-propanesulfonate and are accurate to £0.01 ppm. 2B 0 Solutions were treated with 2H O-washed Chelex resin to remove 2 2 paramagnetic species (127) prior to 1H NMR analysis. C. Computer Simulation of Complex 13C and 1H NMR Spectra 13 l . . . . 13 C and H chemical shifts and coupling constants in complex C l and H NMR spectra were obtained by comparison of the experimental data with the theoretical spectra generated by the ITRCAL program available from the Nicolet Computer Company, Madison, WI. This program permits the calculation of 13C and 1H NMR spectra by first entering reasonable 35 36 estimates of the chemical shifts and coupling constants for the nuclei in question and entering the actual frequency of the lines in the ex- perimental spectrum. By changing the chemical shifts, coupling con- stants, or both, iteration to a best fit of the theoretical with the real spectrum is obtained. The rms error between a simulated and ex- perimental spectrum was typically 0.15 Hz. D. Gas-Liquid Chromatography (GLC) Gas—chromatographic analysis was performed on a Varian Aerograph 1200 equipped with flame-ionization detection. A 1.8 m x 2 mm column of UV—l7 (3%) on High Performance Chromosorb W-Aw (100—200) from Applied Science was used with a temperature program of 100-230°C at $°/min. For derivatization, aqueous samples (6 uL) were added rapidly to a mixture of 150 uL of N,0-bis(trimethylsily1)trifluoroacetamide (BSTFA) containing 1% of chlorotrimethylsilane (TMCS) and 150 uL of dry pyridine. and the mixtures were analyzed after 25 min at 60°C. Reten- tion data are reported relative to the pertrimethylsilylated (Me3Si) derivative of D-gluconate, and GLC peak-areas were calculated from the product of peak-height and peak-width at one-half height. E. Phosphate and Radioactivity Assays Quantitative inorganic and organic phosphate assays were performed according to the procedure described by Leloir and Cardini (128). Radioactivity was assayed on a Beckman LS-lOO scintillation counter by using a Triton X-100 (1000 mL)-PPO(8 g)~POPOP(0.2 g)-toluene(2000 mL) cocktail. Aqueous sample (0.2 mL) was dissolved in 2.3 mL of the cocktail for analysis. 37 F. pH Measurements pH Measurements were performed with a Corning Digital 110 pH meter equipped with a Corning semi-micro combination electrode. pH Adjust- ments on solutions of aldose phosphates were made at 25°C prior to NMR analysis at the reported temperatures. pH Measurements in 2H 0 solu- 2 tions were corrected using the equation, pH = pD - 0.4 (129). G. Assays for Reducing Sugars, Aldonic Acids, Glycollic Acid, Formaldehyde and Cyanide Tetrose solutions were standardized as follows: aliquots (1—4 mL containing 0.1-1.0 mmol of aldose) taken from a stock tetrose solution were added at 4°C to 5 mL solutions containing 1 mmole KCN. Reaccion mixtures were adjusted to 10 mL volumes with H20. After 1 h at 4°C, the reaction mixtures were incubated at 25°C for 50 h, and excess cya- nide was determined in each reaction mixture by the Liebig-Déniges method (40, 130,131). Stock solutions of the tetroses standardized in this fashion were subsequently used to prepare standard curves for the Nelson reducing sugar assay (132). All other quantitative reducing sugar assays were conducted with standard, crystalline compounds using Nelson's assay (132) or Park and Johnson's reducing sugar determination (133). Formaldehyde was assayed according to the method of Walker (134). Aldonic acids were assayed essentially according to the method of Frisell and MacKensie (135). Neutral samples (1 mL) were treated with 0.1 mL of 0.1 M sodium periodate and incubated at room temperature for 10 minutes at which time 0.1 mL of a 10% solution of NaHSO was added 4 followed by 5.0 mL of chromotropic acid (0.2% in 10 M 32804). The 38 samples were placed in a boiling water bath for 20 minutes, cooled. and treated with 1.0 mL of 5% thiourea. Samples were read against a reagent blank at 570 nm. Glycollic acid was assayed by the method of Lewis and Weinhouse (136). Glycosides were assayed with phenol-sulfuric acid (137). H. Catalytic_§ydrogenoly§is Catalytic hydrogenolysis at >1 atm was carried out with a Parr pressure-apparatus and 250-mL reduction flasks. Reductions at atmospheric pressure were conducted in an apparatus described by Vogel (138). The reaction vessel consisted of a side-arm flask equipped with an addition funnel tightly secured with a rubber stopper, and stirring was provided magnetically. All catalytic hydrogenolysis reactions were performed at room temperature. 1. Chemicals Glycolaldehyde,DL-glyceraldehyde, D—arabinose, D—lyxose, D-ribose, D-xylose, D-glucose, D-mannose, D-galactose, D-ribose 5—P, D-arabinose 5-P, D- and DL-glyceraldehyde 3-Pdimethylacetal,1,3-dihydroxy-2- propanone-P dimethyl ketal, D-fructose 1,6—P DL-glycerol 1—P, di- 2, sodium adenosine 5'-triphosphate, DL-calcium glycerate, D—ribono-1,4— lactone, DL-sodium 2-hydroxybutyrate,D-gulono-l,4—lactone, palladium~ barium sulfate (5%), and deuterium oxide (ZHZO) (99.8 atom percent) were obtained from Sigma Chemical Company and used without further purification. Formaldehyde (37 percent aqueous solution) was purchased from Mallinckrodt. L-Threonine (allo—free) was obtained from the United 39 States Biochemical Corporation. D—Calcium galactonate and D-calcium glu- conate were purchased from Pfanstiehl Laboratories, Inc. Potassium [13C] cyanide (K13CN) was supplied by the Los Alamos Scientific Laboratory, University of California, Los Alamos, New Mexico with 99.64 percent purity and 90.7 atom percent [13C1-enrich- ment. Potassium [14C] cyanide (K14CN) was obtained from New England Nuclear and had a specific activity of 45-55 mCi/mmol. Lead tetra- acetate,deuterium chloride (ZHCl) (20%, 99 atom percent) and sodium deuteroxide (NaOZH) (30%, 99 atom percent) were obtained from Aldrich Chemical Company. Acetic acid-2H4 (99.5 atom percent) and deuterium gas (2H2) (99.5 atom percent) were obtained from Merck Sharpe and Dohme Canada Limited. Ion-exchange resins were purchased from Sigma Chemical Company and converted to the appropriate forms. N,0-Bis(trimethylsilyl) trifluoroacetamide (BSTFA) containing 1% chlorotrimethylsilane (TMCS) was obtained from Pierce Chemical Company. Pyridine for gas chromato- graphic analyses was distilled from barium oxide and stored over 4 A molecular sieves. Glycerol kinase (EC 2.7.1.30) from E. coli, D-fructose-l,6—P2 aldolase (EC 4.1.2.13) from rabbit muscle, alkaline phosphatase (EC 3.1.3.1), acid phosphatase (EC 3.1.3.2) from potato, triose— phosphate.isomerase (EC 5.3.1.1) from yeast, D-ribose S—P isomerase (EC 5.3.1.6) from yeast, D-ribulose S-P kinase (EC 2.7.1.19) from spinach, hexokinase (EC 2.7.1.1) from yeast, phosphoglucose isomerase (EC 5.3.1.9) from yeast, phosphofructokinase (EC 2.7.1.11) from rabbit muscle and myokinase (EC 2.7.4.3) from rabbit muscle were pur- chased from Sigma Chemical Company. 40 J. GeneralgSyntheses and Purifications D-Glyceraldehyde was prepared from D-fructose by oxidation with lead tetraacetate (139). D-Lactaldehyde was prepared from L-threonine by the method of Zagalak et al. (140). 2,4-0-Ethylidene-D-erythrose was prepared according to Perlin (141). Aqueous solutions adjusted to pH N8 with dilute NaOH contained approximately 95 percent monomer as determined by 13C NMR. Hydrolysis of this compound with 0.12 M sul- furic acid at 90°C for 35 min yielded D-erythrose (87%). The acidic solution was neutralized with barium carbonate, the mixture filtered through Celite, and the filtrate decolorized with charcoal and de- ionized with Dowex 1 X8 (0Ac‘) and Dowex 50 X8 (H+). Aqueous solu- tions of D-erythrose at 0.5 M contain approximately 5 percent dimers and/or higher—order structures. 2,4-0-Ethylidene-D-threose was pre- pared according to Ball (142). D-Threose was prepared from the acetal as described for the preparation of D-erythrose and was estimated to be greater than 95 percent by 13C NMR. D-Sodium xylonate and DL-[l-lBC]glyceric acid were prepared by hypoiodite oxidation of D-xylose and DL-[1-13C]glyceraldehyde, reSpec— tively (143). D-[l-13C]Arabinonamide and D-[l-13C]ribonamide were pre- pared from the corresponding lactones by the method of Hudson and Komatsu (144). D-Arabinono—, D-ribono-, D—lyxono- and D-xylononitriles were pre- pared from the corresponding aldononitrile phosphates (preparation described in Section II, L3a). The aldononitrile 5-phosphates were de- PhOSphorylated by incubating the phosphate overnight at 34°C with pota- to acid phosphatase (5~10 mg) at pH 4.0. When inorganic P was greater than 90% of the total P present. the reaction mixture was treated with up ..- DV -a a! nd an equal volume of hot ethanol, incubated for 15 min at 34°C, and cen- trifuged at 12,000 rpm to remove protein. The supernatant was treated consecutively with Dowex 1 X8 (OAc-) and Dowex 50 X8 (H+) to remove organic and inorganic P. The pH of the nitrile solution was maintained below 5 to prevent epimerization. The final solution was concentrated in vacuo at 30°C prior to assay by GLC and 13C NMR. Epimeric purity was greater than 95 percent by 130 NMR. Mixtures of D-[1—13C]sodium ribonate and arabinonate (5 mmol) were purified by chromatography at 25°C on a 2.2 x 51 cm column packed with Dowex 1 X8 (200—400 mesh) resin in the acetate form. Aldonates were applied at pH 9-10 and the column was developed with 0.5 M acetic acid. Fractions (5 mL) were collected at a flow rate of 0.5 mL per min and D- [1-13C1ribonic acid eluted between fractions 200-215. D—[1-13C1Arabinonic acid eluted between fractions 80-120 after changing the eluent to 1 M acetic acid. Aldonic acids were detected by radioactivity or by chromotropic acid assay. Fractions containing the aldonic acids were pooled and evaporated at 35°C in vacuo to remove acetic acid. Gas-liquid chromo- tography and 13C NMR of the corresponding C-2 epimeric sodium salts and standard sodium D-ribonate established configuration and purity (>95%). Mixtures of D-[l-13C130dium xylonate and lyxonate (2 mmol) were purified on a similar column at 4°C. The column was developed with a linear gradient of acetic acid (0.3 M - 0.5 M, 4000 mL) at 7.5 mL per 15 min fraction, followed by 0.5 M acetic acid. D-[1—13C1Xylonic acid eluted first (fractions475-520), followed by D-[l-IBCllyxonic acid (fractions 545-620). Fractions containing the aldonic acids were - . . 13 pooled and evaporated at 35°C in vacuo to remove acetic acid. C NMR 42 of the corresponding C-2 epimeric sodium salts and standard sodium D- xylonate established configuration and purity (>95%). Mixtures of D-sodium erythronate and threonate (2.5 mmol) were purified by chromatography at 4°C on a 2.2 x 60 cm column packed with Dowex 1 X8 (200-400 mesh) resin in the acetate form. The column was developed with a linear gradient acid (3 L, 0.1-0 6 M) followed by 0.611 acetic acid until the acids eluted as two peaks. Fractions (4 mL) were collected at 0.4 mL per min. Fractions containing the aldonic acids were pooled, and evaporated at 35°Ctkzvacu0 to remove acetic acid. Gas-liquid chromatography and 13C NMR of the corresponding C-2 epimeric sodium salts and standard sodium threonate established configuration . and purity (peak 1, >95% argtkro; peak 2, >95% threo). 13 D-[l C]Ribono—l,4-lactone,ll-[l-13CJarabinono-1,4-lactone:and D— 13 [l- C]lyxono-—l ,4-1actone were prepared by passage of the corresponding purified [1-13C ]a1donates through columns containing a tenfold excess of Dowex 50 X8 (200-400 mesh) in the H+ form and eluting with deionized H20. Solutions of the free acids were concentrated to gums at 30°C in vacuo and the residues were stored in vacuo at 25°C over MgClO4. Lactonization was determined by 13C NMR analysis and, in most cases, was complete in 2-5 days. Preparation of D-[l-13C]xylono-l,4—lactone by the above method resulted in a mixture containing lactone and other intermolecular esterification products. The 1,4-lactone was prepared by incubating D-[l-13C]xylonic acid at 50°C and pH <1 (HCl) for 30 h. The resulting mixture contained 35% aldono-l,4-lactone and 65% free acid. This mix- ture was analyzed by 13C to determine l3C chemical shifts and J for C,C the 1.4—1actone. l3C]ribono-l,5~-lactone l3 D-[1-13C1Arabinono-l,5-lactone and D-[l— 13C] were prepared by catalytic oxidation of D—[l- C]arabinose and D—[l- ribose, respectively, with Pt/O as described by Conchie et al. (145). 2 D- and DL-Glyceraldehyde 3-P were prepared from the acetals as described by Ballou and MacDonald (146). 1,3-Dihydroxy-2-propanone phosphate was prepared from the ketal as described by Ballou (147). Silver carbonate (Ag2C03) was prepared from silver nitrate and sodium carbonate as described by McCloskey and Coleman (148). Glycolaldehyde Phosphate. Disodium DL-glycerol l-P hexahydrate (8.6 g, 27 mmol) was moistened with 5 mL of H O and dissolved in 400 2 mL of glacial acetic acid with efficient stirring. Upon dissolution of the salt, 1.7 mL of 18 M sulfuric acid was added (149), and addi- tion of lead tetraacetate (24 g, 54 mmol) was made during 15 min. After 2 h, oxalic acid (4.5 g, 50 mmol) was added and stirring was continued for an additional 30 min. The suspension was filtered through Celite and the filtrate was concentrated at 30°C in vacuo to approximately 30 mL. The filter cake was washed with 200 mL of H20, the concentrate and washings were combined, and barium acetate (13 g, 50 mmol) was added with efficient stirring at 4°C for 15 min. The white suspensirnx was filtered through Celite, the filter was washed with H20, and the filtrate and washings were treated with ex- cess Dowex SO (H+). The suspension was filtered, the solution con- centrated as before to about 200 mL, and the concentrate extracted overnight at 4°C with diethyl ether in a continuous liquid-liquid ex- traction apparatus. The aqueous solution was recovered, concentrated as before to about 30 mL, and stored at ~20°C. Yield: 25 mmol (93%) by total P with a trace of inorganic P. Purity: at least 95% by 44 13C NMR. D-Glyceraldehyde 3—P. Disodium D-fructose 6vP dihydrate (3.4 g, 10 mmol) was treated with lead tetraacetate (18 g, 40.5 mmol) in the manner described for the preparation of glycolaldehyde—P, except that 1.1 mL of 18 M sulfuric acid (20 mmol) was added prior to the addi— tion of the oxidant. After Dowex 50 (H+) treatment, the acidic solu- tion of 2-0—glycoloyl-D—glyceraldehyde 3—P was concentrated to 10 mL and stored at 25°C for 18 h to yield D-glyceraldehyde 3-P and glycollic acid. Alternatively, hydrolysis can be carried out by incubating the acidic solution at 40°C for 6 h. The resulting solution was adjusted to pH 5.5 with 2 M NaOH and applied to a DEAE-Sephadex A—25 (40-120 mesh) column at 4°C in the acetate form, washed with a small amount of H20 and eluted with a linear gradient of sodium acetate (1500 mL, 0.05-0.6 M, pH 5.5 i 0.1). Glyceraldehyde 3-P eluted at 0.15 M sodium acetate and was preceded by glycollflzacid. Fractions were as- sayed for D-glyceraldehyde 3—P by organic P analysis and for glycollic acid by the method of Lewis and Weinhouse (136). Fractions containing D-glyceraldehyde 3-P were pooled, treated with excess Dowex 50 (H+), and concentrated twice in vacuo at 30°C to approximately 5 mL to re- move acetic acid. Yield: 8.1 mmol (81%) by organic P analysis with a trace of inorganic P. Purity: at least 95% by 13C NMR. K. Analysis of Cyanohydrin Reaction Mixtures l. Gas-liquid chromatography (GLC) Reactions were carried out in lS-mL centrifuge tubes, reaction mixtures were incubated at 18° i 1°C (water bath) and stirring was provided magnetically. For reactions without pH control, the tube was 45 charged with KCN (0.6 mmol in 1.2 mL H20), followed by the aldose solution (0.6 mmol in 0.8 mL H20). For reactions at controlled pH, the vessel containing the cyanide solution (1.1 mL) was sealed with a stopper fitted with a pH electrode and three polyethylene tubing inlets for adding 4 M HCl, 1 M NaOH and the aldose solution. The pH of the solution (N11) was adjusted to the desired value with 4 M HCl. The aldose solution (0.8 mL) was then added over a period of 0.5 min with efficient stirring. The pH was controlled to $0.2 units by the addition of 4 M HCl and/or 1 M NaOH. Aliquots were withdrawn at various time intervals , derivatized with BSTFA, and analyzed by GLC. 2. 13C NMR spectroscopy Reaction mixtures were prepared as described above except that KlBCN was used. Reactions without pH control were studied by mixing equal volumes (1 mL) of equimolar solutions of KISCN and aldose in a 10 mm NMR tube. Reaction mixtures at controlled pH were assayed at various times by transferring a 0.5-mL aliquot from the reaction vessel to a 0.5-mL coaxial NMR insert tube. The PrObe W35 main~ tained at 18°C i 1°C during data acquisition. To improve detection and quantitation, 55° pulses and 5 sec delay times were employed. Computer integrated peak areas were used to provide quantitation. The areas of all peaks were summed and the proportion of each peak taken in relation to the total. 46 L. Preparation of Carbon—l3 Enriched Aldoses l. Pentoses and hexoses a. General method for the preparation of aldononitriles A three—neck, 25-mL round-bottom flask was constructed as shown in Figure 9. In a typical preparation, the flask is immersed in a water bath at 18°C and charged with an aqueous solution of sodium cyanide (13 mL) prior to sealing the center neck. Port A is clamped and a 10 mL volume of air withdrawn from port 8 with a syringe. Port B is sealed. Acetic acid (3 M) is added from syringe C until the de- sired initial pH of the cyanide solution is reached. Port A is opened and the solution of aldose (6 mL) is added slowly from a syringe with efficient stirring. The pH is adjusted during the reaction by addi- tiOn of 3 M acetic acidcnrlbisodium hydroxide from syringes C and D as required. Samples are withdrawn from port A at 10 min time intervals and the extent of reaction determined by GLC. After the reaction is complete, the pH is lowered to 4.2 i 0.1 and the contents of the flask are reduced over palladium-barium sulfate. Concentrations of sodium cyanide and starting aldose,pH require- ments, yields and ratios of the epimeric aldononitriles are listed for specific aldoses in Table VI of Results and Discussion. GLC retention times for the Me3Si derivatives of aldononitriles relative to that of the D-gluconate derivative are given in Table II of Results and Dis~ cussion. b. Catalytic reduction of aldononitriles Palladium—barium sulfate (5%, 62 mg per mmol of nitrile) was weighed into a 250-mL flask, 10 mL of water was added, and the suspen- -9 sion was reduced with hydrogen at 4.2 kg/cm2 (60 lb in 7) for 10 min . 47 Figure 9. Reaction vessel for the preparation of aldononitriles. The necks were sealed with rubber stoppers. Inlets for ports A and B were constructed from lB-gauge needles and PE—60 polyethylene tubing. Inlets for syringes C and D were constructed from 25-gauge needles and PE-20 tubing for finer regulation during additions. Port A contained the reactant aldose solution. Port B was used to withdraw air from the sealed flask. Syringes C and D contained acetic acid (3 M) and sodium hydroxide (M) solutions. The entire assembly was immersed in a water bath at 18° i 1°C, and stirring was provided magnetically. A, port A; B, port B; C, syringe C; D, syringe D; E, pH electrode; F, 25-mL flask; G, stirring bar; and H, pinch clamps. 49 with shaking. The solution of the C-2 epimeric aldononitriles at pH 4.2 i 0.1 was transferred to the reduction vessel, which was evacuated twice and then filled with hydrogen to an initial pressure of 4.2 kg/ cm2 (60 1b in-z). The reduction was allowed to proceed for 2 h at 25°C with vigorous agitation, after which time a decrease in hydrogen pres- sure of 0.14 kg/cm2 (2 lb in-Z) per mmol of nitrile was observed. After completion of the reaction, the solution was filtered through Celite to remove the catalyst. The pH of the filtrate was 4.8 i 0.2. c. Purification of the reduction mixture The filtrate containing the reduction products was acidified to pH 2.8 i 0.2 with batchwise addition of Dowex 50 X8 (H+) to remove amine byproducts. The resin was recovered by filtration and washed with dilute acetic acid, and the filtrate and washings were concen- trated to a syrup. Hydrolysis products formed during the preparation of the nitriles were removed by dissolving this syrup in water and ad- justing the solution to pH 9.5 with dilute sodium hydroxide. After 30 min at room temperature, the alkaline solution was applied to a column of Dowex 1 X8 (OAc-) that was eluted with water. The effluent was collected in a flask containing an excess of Dowex 50 X8 (R+). This solution contained the product aldoses and unreacted starting aldose, which are separable by chromatography on ion—exchange resins (88. 89). Aldonic acid byproducts were recovered from the resin by elution with l M acetic acid. Neutral products were analyzed by 13C NMR spectroscopy, and re- duced with sodium borohydride for GLC analysis of the alditols. 50 2. Formaldehyde, glycolaldehyde, glyceraldehyde, and the tetroses a. Preparation of aldononitriles Two-, three- and four—carbon aldononitriles were prepared by using the apparatus described in Figure 9. The flask is charged with 0.15 M K13CN, sealed, and the pH adjusted to 8.5 i 0.1 with 2.0 M acetic acid. The starting aldose is added so that the final concentration of both K13CN and aldose is 0.1 M. The pH is maintained at 8.5 i 0.1 by addi- tion of 2.0 M acetic acidcnrllisodium hydroxide as required. Samples are taken for analysis of the aldononitriles by GLC or 13C NMR at 10 min intervals. All condensations were complete (>95%) in 20—30 min and virtually no hydrolysis had occurred. The pH was adjusted to 4.0 with 6.0 M acetic acid and then to 1.7 i 0.1 with 6.0 M hydrochloric acid. At this pH, the aldononitriles are stable for several weeks at 4°C. The preparation and purification of glycolonitrile has been described previously (150). In the present study, this compound and the other aldononitriles were used without purification. [13C]Formaldehyde was prepared from a solution (0.1 M) of K13CN at 5°C after adjusting the pH as already described, prior to reduction. Caution should be exercised in the preparation of formaldehyde and glycolaldehyde, as hydrogen chloride and formaldehyde on contact can produce the potent carcinogen, bis(chloromethyl)ether (151). b. Reduction of short-chain aldononitriles Palladium-barium sulfate (5%, 62 mg per mmol of nitrile) was weighed into a side-arm flask, 5-10 mL of water was added, and the suspension reduced with hydrogen for 15—20 min at atmospheric pressure and 25°C with efficient stirring. During this period, the su5pension 51 changed from brown to a whitish-gray tint. The solution of aldono- nitriles at pH 1.7 i 0.1 was added from an addition funnel, and the reduction vessel was evacuated three times prior to a final charging with hydrogen. Reduction times varied with the aldononitrile: hydro- gen cyanide, 8 h; glycolonitrile, 18 h; glyceronitrile, 8—10 h; erythrono- and threononitriles, 18 h. Completion of the reaction was determined by 13C NMR and GLC. GLC Retention times for the Me3Si deri- vatives of the short-chain aldoses relative to that of the D-gluconate derivative are given in Table II in Results and Discussion. After re- duction, the mixture was freed of catalyst by filtration through Celite. For the preparation of [13C1formaldehyde, the reduction flask was cooled to 5°C prior to the addition of hydrogen cyanide, and the re- duction was started immediately after addition, omitting evacuation. The reaction vessel was allowed to warm to room temperature slowly during the reduction. c. Purification of the reduction mixture The filtrate containing the reduction products was treated with an excess of Dowex 50 X8 (H+) resin to remove amine byproducts. This step should be completed immediately after reduction to minimize reac- tion between byproduct amines and the desired aldose. The resin was removed by filtration and the acidic solution treated with an excess of silver carbonate with stirring until the pH was approximately 6.5. The precipitate (AgCl and Ag2003) was removed by filtration through Celite and the filtrate treated with an excess of Dowex 50 X8 (R+) resin with stirring for 60 min at 25°C. After removal of the resin, the clear solution was concentrated at 30°C in vacuo to 30 mL and 52 treated again with Dowex 50 X8 (H+) for 30 min. The mixture was fil- tered through Celite, concentrated at 30°C to 30 mL, and treated with an excess of Dowex 1 X8 (200-400 mesh) resin in the acetate form for 10 min. The solution was filtered, concentrated to 3 mL at 30°C, and analyzed by 13C NMR and GLC. Erythrose and threose were separated at 25°C on a column (2.2 x 92 cm) of Dowex 50 X8 (200-400 mesh) resin in the barium form (88); 4 mL fractions were collected at 0.4 mL per min. Threose was eluted between fractions 55 and 65, and erythrose between fractions 105 and 122. Unreacted C4 nitriles and glyceraldehyde eluted with threose. The column capacity exceeded 4 mmol of tetrose. Concentration of acidic solutions (pH <2) of erythrose and threose produced oligomeric mixtures of these sugars. Oligomerization may be reversed by heating a 50-70 mM solution of oligomers for 30 min at 90°C with the addition of dilute sulfuric acid to a final concentra- tion of “0.02 M. The acidic solution, after being neutralized with barium carbonate, and deionized with Dowex 1 (OAc-) and Dowex 50 (3+) resins, contains only monomeric tetroses, as evaluated by 130 NMR Spectroscopy. 3. Triose, tetrose and pentose phosphates a. Preparation and purification of aldononitrile phosphates The K13CN solution (0.15 M) containing 107 cpm of K1°CN was placed in the sealed flask described in Figure 9 and cooled to 5°C with an ice bath prior to adjustment to pH 8.0 i 0.1 with 2 M acetic acid. The solution of aldose phosphate at pH 7.5 was added while maintaining the pH of the reaction mixture between 7.5 and 8.0 with additions of 2 M acetic acid and/or 1 M NaOH. Stoichiometric amounts of aldose phosphate and cyanide were used, and the final concentration of 53 reactants was 0.05-0.1 M. After 15 min at 5°C and pH 7.5-8.0, the ice bath was removed and the reaction mixture allowed to warm to 25°C over a period of 30-40 min. The pH was then adjusted to pH 4.0 i 0.2 with 2 M acetic acid. Condensation was complete (>95%) when assayed by 13C NMR using a short pulse width (10 us, 55°) and long delay time (10 s) to facilitate aldononitrile detection. The racemic mixture of glyceronitrile phosphate was adjusted to pH 1. 7 i 0.1 with Dowex 50 (11+) and hydrogenolyzed directly to DL-glyceral‘ dehyde 3-P without further purification. Epimeric tetrono- and pento- nonitrile phosphates were purified by ion-exchange chromatography on a 2.2 x 51 cm Dowex 1 X8 (200-400 mesh) column in the formate form at 4°C. Solutions of aldononitrile phosphates were adjusted to pH 6.5- 7.0 prior to application to the column bed. Columns were developed with linear gradients of sodium formate: for 4-carbon aldononitrile phosphates, 3000 mL, 0.2-0.9 M sodium formate, pH 3.9; for 5-carbon aldononitrile phosphates, 3000 mL, 0.05-0.8 M sodium formate, pH 3.9. Fractions (7 mL) were collected at 0.5 mL per min and were assayed by radioactivity or phosphate. The epimeric aldononitrile phosphates typically elute in separate peaks between 1500 mLe2000 mL of the gradient. The nitrile phosphate with Ei§72,3-hydroxyl groups is the major product and is eluted last under these conditions. Column capa- city exceeds 6 mmol of aldononitrile phosphate. Fractions containing the aldononitrile phosphates were pooled and adjusted to pH 1.5 with Dowex 50 X8 (H+). After filtration, the acidic solutions were concentrated to 100 mL in vacuo at 30°C and ex- tracted continuously with diethyl ether overnight at 4°C to remove formic acid. The aqueous acidic solutions were recovered, concentrated 54 in vacuo at 30°C to approximately 10 mL, and adjusted to pH :2 with Dowex 50 X8 (H+) for storage prior to hydrogenolysis. b. Hydrogenolysis of aldononitrile phosphates Palladium-barium sulfate (5%, 62 mg per mmol of nitrile) was weighed into a side-arm flask, 5—10 mL of H20 was added, and the sus— pension was reduced for 15-20 min at atmospheric pressure and 25°C with efficient stirring. During this period, the catalyst changed from a brown to a whitish gray color. The solution of aldononitrile phosphate adjusted to pH 1.7 as described below was added from an addition funnel into the reduction vessel which was filled and eva- cuated three times prior to a final charging with hydrogen. The con- centration of aldononitrile phosphate solution during hydrogenolysis varied between 50 and 100 mM. Purified four- and five-carbon aldononitrile phosphate solutions were treated with 1 mL of glacial acetic acid per mmol of nitrile phos- phate and then with Dowex 50 X8 (8+) or 2 M NaOH to pH 1.7 i 0.1. Re- ductions were carried out as described for the three-carbon homologue. Aldononitrile phosphates were typically reduced for 6—8 h at 25°C. In a few instances, incomplete reduction was noted. In these cases, the spent catalyst was removed by filtration through Celite and a second reduction was performed to complete the conversion to the aldose phosphate. Hydrogenolysis products were assayed by 13C NMR to determine the extent of reduction to l—amino-l—deoxyalditol phOSphates and the amount of unreacted aldononitrile phosphates. After hydrogenolysis, the catalyst was removed by filtration through Celite and the solution treated with excess Dowex 50 X8 (H+). 55 After filtration, the solution was concentrated to 10 mL. Typically, the reaction mixture contains product aldose phosphate. l-amino-l- deoxyalditol phosphate, and a small amount of aldononitrile phOSphate. This solution was adjusted to pH 4.5 I 0.1 with dilute NaOH and applied to a 1.2 x 50 cm DEAE-Sephadex A—25 (OAc-) column at 4°C which had been equilibrated with 0.05 M sodium acetate at pH 4.5 i 0.1. The column was developed with a linear acetate gradient (1500 mL, 0.05- 0.8 M sodium acetate, pH 4.5 i 0.1). Fractions (6 mL) were collected with a flow rate of 0.5 mL per min. The l-amino—l-deoxyalditol phos- phate eluted near the void volume, followed in order by aldose phosphate and the aldononitrile phosphate. Fractions containing aldose phos- - + phate were pooled, treated with excess Dowex 50 (H ), and concen— trated in vacuo at 30°C to approximately 10 mL. Aldose phosphate solutions were stored at pH 4.0 and ~15°C. c. Characterization of aldose phosphates Purified [1-13C1-enriched aldose phosphates (100 umol) in 2 mL of 50 mM Tris-HCl buffer at pH 9.0 were incubated with alkaline phospha- 13 tase for 1 h at 36°C. Carbon-13 NMR spectra of the resulting [l- C]- enriched aldoses were compared with standard spectra. I 4. Enzymatic preparation of [13CI-enriched carbohydrates a. D—[2-13CIRibulose 1,5-P2 (D-[2-13C1Erythropentulose 1,5-P2) D—[2-13C]Ribose S—P (0.2 mmol) was incubated with 245 units of D- ribose S-P isomerase for 30 min at pH 7.5 and 36°C. The resulting mixture containing 28% D—[2-13C1-ribulose S-P was converted to D- [2-l3C]ribulose 1,5-P, by the method of Horecker et al. (152) and purified by the method of Byrne and Lardy (153). 56 b. L-[3,4-13C]Sorbose 1,6-P2 DL—[l-13C]Glyceraldehyde 3—P (75 mM) was incubated for 30 min at pH 7.5 and 34°C with triosephosphate isomerase, producing a mixture of l L-[1-13C1glyceraldehyde 3-P and [3- 3CJdihydroxyacetone phosphate. D- Fructose—1,6-P2 aldolase was added and the mixture incubated for 4 h at 34°C. The reaction mixture was made 50% in hot ethanol and centri— fuged to remove protein. The supernatant was concentrated in vacuo at 30°C to 2 mL, adjusted to pH 7.5 r 0.1, and analyzed by 13C NMR. c. D-[l—lac,2-13C ]Fructose 1,6-P2 (FDP) D-[l-14C,2-13C]Glucose (1.4 mmol) was dissolved in 17 mL H20, 1.25 g ATP and 0.15 g MgCl2 were added, and the pH was adjusted to 7.4 with 2 M NaOH. Myokinase (500 units), hexokinase (500 units), phos- phoglucoisomerase (250 units), and phosphofructokinase (300 units) were added, and the mixture was incubated for 60 min at 34°C while maintain- ing the pH of the reaction at pH 7.4. Assay by 13C NMR showed complete conversion of glucose to FDP. The reaction mixture was applied to a 1.7 x 26 cm Dowex 1 X8 (200-400 mesh) column in the chloride form. The column was eluted with 6 L of 0.01 M HCl to remove AMP and ADP. The column was then eluted with 4 L of 0.02 M HCl, 0.02 M LiCl,and 11 mL fractions werecol- lected at 0.55mL/min. FDP eluted between fractions 130-240 as deter- mined by radioactivity. Fractions containing the phosphate were pooled and concentrated in vacuo at 30°C to 200 mL. The solution was neutra- lized with 0.5 M NaOH to pH 7.0 and Ba(0Ac)2 (5.6 mmol) was added with stirring followed by 240 mL of ethanol. The solution was stored at 4°C overnight to facilitate the precipitation of the barium salt of D- [l-laC,2-13C]fructose 1,6-P2. The precipitate was collected by 57 centrifugation, suspended in H20, and treated with Dowex 50 X8 (H+). The solution was filtered to remove the resin. concentrated to 3 mL at 30°C in vacuo, and adjusted to pH 6.0 with 0.1 M NaOH prior to analysis by 13C NMR. Yield: 1.2 mmol (86%) based on P assay. 13 d. L-[Z- C]Glycera1dehyde 3-P DL-[2-13C1G1yceraldehyde (0.1 mmol) in 2 mL H 0 was incubated 2 + with 1.5 molar equivalents of Mg2 -ATP and 250 units of glycerol kinase at pH 7.5 and 36°C. After 1.5 h, the pH was adjusted to pH 7.5 and incubated for an additional 60 min. 13C NMR analysis showed the presence of unreacted D—[2-13C]glyceraldehyde and the product, L- [2-13C]glyceraldehyde 3-P. The phosphorylated product can be purified by chromatography on DEAE-Sephadex (OAc-) as described in Section II, L3b. The unreacted D-[2—13C]glyceraldehyde, which elutes at the void volume of the column, was shown to be 91% D-isomer by reaction with 1,3—dihydroxy-2-propanone phosphate and D-fructose 1,6—bisphosphate aldolase to produce 91% D—[5-13C}fructose 1-phosphate (70.5 and 81.7 ppm) and 9% L-[5-13C]sorbose l—phosphate (70.9 ppm). Rabbit-muscle aldolase can act on both D- and L-glyceraldehyde, and gives an equi- molar mixture of the ketose l-phosphates when the original mixture of DL-[2—13C]glyceraldehyde is used. M. Preparation of [2H1-Enriched Aldoses and Derivatives 14C, 2HlErythrose and threose A solution of K13CN (2 mmol, 13 mL 2H20) at 20°C containing KIACN 1. D-[l-13C, 7 (10 cpm) was added to a 25-mL sealed flask (Figure 9) and adjusted to PH 8.0 t 0.1 with 0.7 M acetic acid—2H4. D—Glyceraldehyde (2 mmol) 2 . . was concentrated from 3 mL of H20 several times at 30°C Ln vacuo. 58 The residual gum was dissolved in 4-5 mL 2H20 and added to the solu- tion of Kl3CN. The pH of the reaction mixture was maintained between 8.0 and 8.3 with additions of 0.7 M acetic acid-2H4 and/or 1.0 M NaOZH. After 20-25 min, the pH was lowered to 4.0 i 0.2 with 17 M acetic acid-2H4. A further adjustment of pH to 1.7 r 0.2 was made with 3 M 2HCl. Palladium-barium sulfate (5%, 62 mg per mmol of nitrile) was weighed into a 50-mL side-arm flask, 5 mL 2H O was added, and the 2 system was evacuated and charged three times with N After the last 2. 2 evacuation, the system was charged with H and the catalyst reduced 2 for 15-20 min at atmospheric pressure and 25°C with efficient stir- ring. The ballast containing 2H was filled with light mineral oil 2 to prevent entry of H 0 into the reduction apparatus. The [13C]- 2 enriched aldononitriles were then added and the reduction was con- tinued for 10 h, or until the nitriles were completely reduced as determined by GLC. The product epimeric aldoses were deionized and separated as described in Section II, L2c. Products were characterized by 1H and 13C NMR. Yield: 70% based on product weight as gums and on the recovery of radioactivity after separation. 2. D-[2-13C, 1°C, 2H]Ribose and arabinose ’3 -l3c.l4c.231 L D~[2 Ribose and arabinose were prepared from D—[l- C, 14C,2H]erythrose according to a modified procedure described fortfluapre- paration11 is difficult to determine since the change in aldono- nitrile concentration is determined by two rates, the rate of forma- tion from aldose and the rate of hydrolysis to aldonamides and other products. Both the rate of aldonamide formation and the rate of hy- drolysis are high at high pH values. The disappearance of aldono- nitrile at 18°C was measured by 13C NMR for the [1-13C1aldononitri1es 3—8 at several pH values (Chart I, Figure 11). At pH 8.5, the rate of DL-glyceronitrile (3)disappearance is negligible (not shown). How- ever, 3 disappears slowly at pH 12.7, with completion in 100 h. The 13C NMR spectrum of the products from the total hydrolysis of DL- [1-13C1g1yceronitrile shows a resonance for DL-[l-13Clg1ycerate and four resonances corresponding to the C-1 carbons of DL-(l—13011yxonate , xylonate, arabinonate, and ribonate. GLC confirms the presence of these pentonates and, after removal of excess cyanide as H13CN by . l . . . aeration, 3C NMR analySis indicates that the pentonates are present in the following proportions: 41% lyxo, 18% xylo, 29% arabino, OsZuXZ). CCZCCGAI C. .CfiULOQ 69 ( 5(IO.5) IOO‘ : :2 1+4 1 1+ «9.07 90- 6(8.5) 803- 70' Percent in Reaction Mixture V Figure 11. 5 ( I2.7 ) . V 20 2.5 3.0 3.5 Time (h) Disappearance of aldononitriles in cyanohydrin reaction mix- tures. Reaction mixtureijwere incubated at 18° : 1°C and data were obtained from C NMR spectra. Concentrations of reactants were 0.3 M except for reactions involving 5, which were conducted at 0.4 M. Experiments were conducted at the pH value in parenthesis. Zero-time points for 4 (pH 9.0) and 5 (pH 10.5 and 12.7) were determined from the amount of aldononitrile in each reaction mixture after 4 h at pH 8.0, pH 8.0 and pH 10.5, respectively. The remaining reactants were conducted with both cyanide addition and hydrolysis at the indicated pH values. 2 = D-[l-l Cleryth— rono- and threononitrile; 3 = DL—[l-13C]2—hydroxybutyroni- trile; 4 = D—[1-13C1arabinono- and ribononitrile. 7O 12% ribo. DL-Glycerate accounts for approximately 5 percent of the —* total products. CM CM CM CM CN CN (H)C(OH) (H)C(OH) (H)C(OH) (H)C(OH) (H)C(OH) (H)C(OH) CH CH HCOH CH HCOH HOCH HCO CHOOH CH3 HCOH HCOH HCOH 3 " ' . L (H) CHZOH CHZOH LHZO 3 4 5 6 7 8 Chart I 13 Formation of the DL-[l- C]pentonates is due to reversal of cya- nide condensation at alkaline pH to produce free glycolaldehyde, which undergoes aldol condensation to form the DL-tetroses. DL-Pentonates derived from the ££§g§_a1dol product, DL-threose, compose about 60% of the mixture, and those derived from the gig product, DL-erythrose, compose the remaining 40%. The reaction to be considered in the alka- line hydrolysis of 3 is shown in Scheme VIII. 1, 7' m L\ glycerate £—§— 3 =13 glycolaldehyde + cyanide K2 / \K3 erythrose threose + + cyanide cyanide ribono-, arabinono- lyxono-, xylono- nitriles nitriles k6 1 k7 l ribonate, arabinonate lyxonate, xylonate Scheme VIII 71 The amount of glycerate formed depends on the rate of hydrolysis, k8, relative to the rates of competing reactions. It is reasonable to as- sume that Kl = K4 = K5 and it is known that k6 and k The results indicate, therefore, that aldol condensation is favored 7 over hydrolysis of 3 at pH 12.7. Although trans aldol products predominate in most aldol condensa- tions (159-161), 40 percent of the products in this case are derived from the cis product, DL-erythrose. The amounts of individual pento- nates depends on K2, K3, K4, K5 and the overall rates of aldononitrile hydrolysis, k6 and k7. It is reasonable to assume that R4 = K5 and that K3 > K2, so that k6 must be greater than k7 for the cis products to be found in the high proportion observed. An examination of the rates of disappearance of aldononitriles derived from D-erythrose and from D-threose shows that this is the case (i.e., k > k7) (Figure 12). 6 Rates of disappearance of C aldononitriles 4 are negligible at 4 pH 9.0 and 18°C (Figure 11). At 18°C, with no pH control, disappear- ance is more rapid (t1/ = l h) than for the C homologue 3 (t 2 3 1/2 1 25 h). However, the rate of disappearance of 4 is comparable with that of[mpz—hydroxybutyronicrile:5(t1/2=l.h). The difference in rates ob- served between 3 and 4 suggests that cyclization may be an important factor in facilitating hydrolysis. The similarity in the rates of disappearance of 4 and 5,however, suggests that cyclization is not es- sential and may be no more facile than the direct, unassisted hydroly- sis of the nitrile. Examination of the products from the hydrolysis of 5 at pH 12.7 and 18°C indicates the presence of 75 percentIHrQ-hydroxybutyrateydth the remaining aldonates probably arising from aldol condensation >> k8 (Figure 11). 72 products of propionaldehyde. The ease of aldolization is a function of the acidity of H—2 of the aldehyde. Removal of the OH group a to the carbonyl would be expected to decrease the acidity of H—2, making carbanion formation more difficult than for glycolaldehyde, thereby favoring hydrolysis over aldolization. CS Cyanohydrins 5 and 7 (tl/2 z 1.5 h) (Cl/2 = 4.5 h) disappear more quickly than 3, 4 and 5, at pH 8.5 and 18°C. In contrast, com— pound 8 is stable at pH 8.5 and 18°C. Pentoses produced from 3 are predominantly hydrates in aqueous solution (6 s 90.7 ppm), demon- C-l strating that OH-4 is unavailable for cyclic hemiacetal formation. Thus, 3 probably cannot form cyclic intermediates, which accounts for its slower_ rate of disappearance relative to the underivatized homo- logue 6. Data in Figure 11 suggest that aldononitriles capable of cyclization have significantly greater rates of disappearance and that cyclization to form six-membered rings facilitates hydrolysis better than the cyclization to form five-membered rings. The latter conclu- sion is not valid, however, and will be discussed later. Configuration also affects the rate of aldononitrile disappear- ance (Figure 12). The overall rate of disappearance of the C5 cyano- hydrins derived from D-erythrose (ribo, arabino; 6) is greater than that of the C5 aldononitriles derived from D-threose (lyxo, xylo; 7), 1/2 = 1.5 h and tl/z = 4.5 h. respectively, at pH 8.5. Within epimeric pairs, arabinononitrile hydrolyzes more rapidly than ribo- with t nonitrile, while lyxono- and xylononitriles appear to hydrolyze at ap— proximately the same rate. The ratio of the epimeric C5 aldononitriles formed initially at pH 8.5 does not always correspond to the ratio of aldonates produced. For example, the ratio of lyxono- to xylononitrile 73 zzz s scum mosaEHouov our: mum: .usacmxo . .mwuowam . H _ a ox~ m pearance“ mum: nouzuxwa :omuosuz .~.m w m.m z; um comuomou sawm>:o:w>o as“ mzquzm nuoeouoenmma: :uszuua oocmumszzzrwm m—duudcocomam mo momma use :a noo:mu5wefia t: we: Om Om ON ON 0.. 0.. m0 L .NH unawa: 0. ON Om Ov On 00 ON aJnmw uouooeg ug maniac} 74 is 3:2 at pH 8.5. Because their rates of disappearance are similar, the corresponding aldonates are formed in the same ratio (Table 1). 0n the other hand, arabinono— and ribononitriles are formed in a 2:3 ratio, but hydrolysis yields arabinonate and ribonate in a 7:3 ratio. Clearly, arabinononitrile hydrolyzes more rapidly than the pibg epimer. Data in Figure 12 do not reflect these differences accurately, since rapid equilibration of the epimeric aldononitriles by reversal of cyanide condensation increases the apparent rate of disappearance of ribononitrile and decreases the apparent rate of hydrolysis of the arabino epimer. 4. Characterization of reactants, intermediates and products Before discussing the sequence of intermediates in the hydrolysis of cyanohydrins, evidence for the identification of the ob« served intermediates is presented. GLC and 13C NMR parameters of the compounds involved in this study are listed in Tables 2 and 3. As shown in Figure 13, unreacted aldose and the epimeric aldono- nitriles, aldonolactones, aldonates and aldonamides are resolvable by GLC. However. derivatization changes the structure and distribution of intermediates so that GLC alone was not a reliable method to exa- mine changes in concentration of intermediates as a function of time. 13 . . . . C NMR spectroscopy is an ideal tool to assay reaction mixtures, . 13 . . . . . . . espeCially when [ C]-enricnment is employed to increase senSitiVity l 13 and decrease acquisition times. [1- 3C]Intermediates observed by C NMR in the reaction of K13CN with D-erythrose are shown in Figure 14. The assignments of resonances to aldononitriles, aldonolactones, aldo— namides and aldonates were made by comparison with spectra of [l—l3C]- enriched standards prepared by alternative routes. Table 2. and derivatives. GLC retention times of pertrimethylsilylated carbohydrates Compound Retentiona D-gluconate 1.00 glycolaldehyde 0.25. 0.26, 0.27 D-glyceraldehyde 0.36, 0.83. 0.85. 0.87. 0.90 2,4-0-ethylidene-D-erythrose 0.50, 1.32 D—erythrose 0.37 D-threose 0.33, 0.36 D-arabinose 0.59, 0.64, 0.67 D-lyxose 0.60, 0.65. 0.68 D-ribose 0.60. 0.64 D-xylose 0.57, 0.66, 0.72 D—glycerate 0.41 D-erythronate 0.52 D-threonate 0.56 D-arabinonate 0.77 D-ribonate 0.73 D-lyxonate 0.75 D-xylonate 0.75 D-allonate 0.94 D-altronate 0.99 D-gulonate 0.93 D-idonate 1.02 D-mannonate 0.94 D-galactonate 0.99 D-talonate 0.99 D-glyceronitrile 0.17 D~erythrononitrile 0.41 D-threononitrile 0.41 3,S-O-ethylidene-D-arabinononitrile 0.63 0.63 3,5-Osethy1idene~D-ribononitrile 76 Table 2 (cont’d). Compound Retentiona D-arabinononitrile 0.63 D—ribononitrile 0.66 D-lyxononitrile 0.65 D-xylononitrile 0.65 D-glucononitrile 0.92 D—mannononitrile 0.89 D-galactononitrile, D-talononitrile 0.87, 0.91b D-allononitrile, D—altrononitrile 0.89, 0.90b D-gulononitrile 0.90 D—idononitrile 0.92 D-arabinono-l,4-lactone 0.68 D-ribono-1,4-1actone 0.75 D-arabinonamide 0.85 0.80 D-ribonamide a . . . . . . Retention times are relative to the (the) Si derivative of D- gluconate; column conditions are described in Instrumentation. Retention times were not assigned. 77 o.ama meHUHEm ~.N~a m.m locouzuzmm.locomu:ulo ea.m m.qm m.mo am.qm am.mm H.q~ ~.oma m.m mumcoamxua co m.em m.qc n~.mm n~.- «.mm q.ow~ m.oH mumcoxxauo s.am A.mo A.~A s.AAH o.a swam oweoasxua «.mm o.so A.NA o.AAa ~.H ease oasoxsaua n.0m m.oo A.Nm a.mm m.mma economauc.alocoalen wn m.~ on m.om N.Hc o.mw o.Hm ¢.Hm o.msa ocouomauq.alo:oxxalo a.a m.w~ s.sa m.~ma mumususssxoueseum-an o.o~a m.m manpoweoeoasxaa m.a~a m.m manpoweocoxsa3a m.q~a mmaawuac .o.m~a m.~H Ionowsu>umro:omwsula c.0ma mawuuwcouxusnxxowcznwm .A.oNH n.A um.~uou;usum.uompzuuo ¢.oma m.m~ m.-~ m.oH mawuuaeousuan «.mma A.m nsxoueseumuun c.¢o N.mo o.H~H m.w A.mNa m.~a mawuuaeowmosawuua m.am o.o Hmamuassxoueseae nm.~|ow£uxuo .omwzuln H.ma m.aa m.sa o.o women»; .mmzzmcamuomalo ¢.ooa m.aa zuz «.maa o.m 20: a a a a mo How so How mo #05 No How no «a no we an H Aw.oa1 szv mucmumcoo wsaflnaoo omauoma Aanav muuazm amowsmso o In masonsoo .c0wuomuw saucxzo:m>o msu mo muozmoun mam moumwmmEHmuCH .mucmuommu mo mwmumsmmma “:2 c . m m ma m an e 78 m.mma m.ma all. 333. III. m.mm m.qc m.mm m.¢m o.mm m.mma n.m commonauua 3:3- .333 .III q.om m.¢c ¢.Hm N.qm m.m~ m.cna m.H meow oficonwwln m.¢ma c.~ occuomaum.aloconfiulo 3:3. m.~ ~.N m.mm o.~o N.mw o.am «.ON o.owH occuomqu.Hloconfiwlo ~.~ma m.m~ nil. nil. 33!. m.~m q.qc m.Nm H.¢m m.mm m.wna N.w mmfismconfiwno en N.N m.Hm N.@@ m.- 0.05 m.~ma m.oH occuoma :s.a-oeaea-oeonauue ~.m~H m.~H 33- 3:3. 3:3, H.oc m.mm o.mm m.qo H.0NH m.m mHHuuHcoconHula q.m¢ ~.mma o.m mchHEmococflnmumuo o.Hm~ m.ma III. a.~ .333 N.¢m o.qo c.Nn o.mm o.mm m.owH m.m oumcocfinmwmla Ill. N.~ 33!. H.oo q.qc m.Hm N.mm N.Hn «.mna o.~ mfiom oHcocanmumla N.wma o.m encuomatm.alo:ocwnmumum u: .311 m.m m.mm m.oo o.mm w.mm «.mm o.mma occuomalc.alococfinmumla w.~w~ m.ma 13!. o.~ III. m.mm q.qo o.~m m.~m «.mm m.owa H.w omwfimcocanmwmla III. o.o m.~m c.~w m.¢m ~.mm «.mma m.oH occuomH|¢.H nomafialococfinmumlc ~.mma m.- ¢.N .III c.Ho ~.Hm o.~m w.~o ¢.H~H m.w oawuuacococfinmwmla o.¢c o.m~ m.qm w.om eunuch: m.c¢ «.mm ~.H~ m.m~ «.NOH m m.m¢ m.mm 0.05 «.mm w.om o o.c omouzuxumla mu How «c How mo How NC ~05 no so mo we do Am.ow1 szv mucwumcoo womansoo UMHIUMH Asmav mumwcm Hmoweono and ma vcsoneou .A.m.ucoov m manme 79 . .N: A.m as uaaam m .Nz s.s an uaaam 6 .mo Huwm eowm on >3: .ammum>o mocmcommu cu one um: wo.mH um mousmmos on uo: fiasco wcfiaasou o .mmmuo>ow on me «no pmm MID mo unmecwamm< n .mmcwewmump uo: mum: >uuam 0: sad: muwfizm HmoHEmco mom mwsaaasoo .mocmcommu mmcmmmoun u up new w: m.o cmcu mmmH mum Amoco m cuaz mucmumaoo wcflaaaoo .w: 5.0+ Om nucmuncoo wswansoo cam Eng H.0H Ou oumuooom mum muuwzm HmowEocu .ooH H oma um mmc«Eumqu Aw w.m~ m.o~ newsmconflu>xoomsmao w.Hm m.o~ spasmsocwnmwmxxommlmln m.¢m m.OH accuumHIQ.HIO@HEH locoazxzxommlmlm m.Nm m.OH OGOuumH!¢.HIOfiwEM locoxxaaxomcnmlo m.oN n.3H o:0uomalq.HaOpHE« locoafiu>xoomlmua m.mm m.oH pseuomanq.alomwaw nocoawmmumzxoomlmlo o.om o.oA ma.maa o.m 33 m.¢m m.~m m.-a m.m N: m.ms m.am mm.aoa o.m .3. w.~q m.~w o.o- m.m a: ©.HNH o.m unavwEmoconwuua no How «U Hum mo How NU HLHH szv mocmumsoo mafianzou omdqoMH Asnnv mumwzm HmUAEmcu Una 2a conceaou no we no NU Ho '5 II...- 5355' '1! 5 I'll. I'll "tll .A.e.u=oov m magma 80 .ucommmu wcaumazafim onu mo mucopoasoo one x mxmom .cOAumeuw>Humu mcfiamcomfim scum omqum ou mwmoaam nan sacs: 1:: as w xmma mo zufiucmcfi one .omwamaoafiamumtn .w umcfismconaulo .n “mumnocanmumla .w monouomalc.alo:onauln .w "mumconfiuln .0 “occuomalq.auoaocfinmumta .m “maauuficoconfiuln .o “mafiuuficoconfinmumln .n "mmownu>uoln .m .mmounu>uoln ou awesome «0 :ofiuammm mnu scum muosvowdumco can mmumumoswmuafi .omomam acumen mommahafimamnumafiuuumm mo nameOumsousonmmo Act—.125... em cm or up w v d d d .ma museum 81 .xao>fiuomamow .mumconHHHUmHIH11o can mcouoma I¢.Hloconww_UnH3~1|a and u can a mxmma comzuon oocmcommw Hones mam umpaaozm .mawuufi: noconww_0malfi_na. m mofiqwuacococflnmwm_o Hua135 .H mm: .2 ”a: .w mmcwmwamloconwu_UmHIH_|a .w usawmwemnococfinmwmaomala.1: .m mono omHIq.Hlococwamwmmomana_ln .m “spasmcwawu umaiaa 3: .o mandamcocwnmmeUmHIH139 .n “panaceanmwQHUmHIH110 .m .ooH .fl owd .: AH mafia commommw Hmuou ”q.o H o.m rd nonficmxomo H. can omownuxwmln :fi 2 m.o umcowumuucmocoo :oHuomom .Amcmom cmmmv mconwmo mozoflwco ozm xaco m30:m sswuomnm mze .Hlo um unmasoflwcm Imvma_ unwound EOum m.cm :ufiB mmumwmchoucw mo aawuommm mzz UmH ~22 mo.ma moannoUmA—I:fl .ca mwswwm Ema . cm 9. oo om oo. our 9: one 09 d a q q d 1 — 313 q 4 l— 4 d d d a — a «1.. (b7): (5.;1/2‘? fléaiéig- g5.§3%?§§3)7.4 1.;251végjxiV/‘fi1 : M u 0 a a _ u o avenge—Emu:— o la 2 82 The 13C NMR parameters of the reaction components are important to establish theirpmesence:UIthereaction sequence. Thechemicalshift of cyanide is pH-dependent, with HCN having 6 = 113 ppm and CN- with 6 = 166.4 ppm. The short-chain aldoses that exist principally as hy- drates have C-l resonances at 90 i 2 ppm. Changing OH-3 to H-3' pro- duces a downfield shift in C-1 of 3-4 ppm. Aldononitriles have C-l resonances at approximately 120 ppm at pH 8.5 that shift downfield N5 ppm as the pH is increased to pH 12.7. C3 Aldononitrile chemical shifts were assigned by preparing and separating the Z—epimeric [1-13C]aldononitrile phosphates and re- moving the phosphate group with acid phosphatase. Assignments were also based on observed differences in magnitude of 3JCl,C4 for stan- dard linear rib2_compound (NO Hz) and arabino compounds (W2.5 Hz) (Table 3). This difference probably reflects the preferred conforma- tions of these compounds in solution (107). Compounds having the arabino configuration are expected to have a planar, extended confor- mation, where C-l and C-4 are anti-planar (180°) and maximal coupling occurs based on the Karplus relationship (125). Solution conformation of linear compounds having the £322 configuration are expected to have C-l to C-4 dihedral angles of approximately 90°, where l3C-13C three- bond coupling is at, or close to, a minimum. Chemical shifts at 175.4 ppm and 177.3 ppm are assigned to C—1 resonances of imido-l,4-lactones having the arabino and £323 configura- tion , respectively. These assignments are based on comparison of NMR parameters (5 and J) of these compounds with structurally-related stan- dard 1,4-lactones of known configuration, on their time of appearance during the alkaline hydrolysis of aldononitriles, and on the structure 83 and proportion of products found after acid hydrolysis (aldono-1,4- lactones). The C-2, C-3 and C-4 chemical shifts of D-ribono- and D- arabinono-1,4—1actones and the respective imido-l,4-lactones are very similar (Table 3), while C-1 of the 1,4-lactones is 2.2-2.7 ppm down- field from the corresponding imido-1,4—1actones. The chemical shift of C-4 is sensitive to ring form, i.e., ring formation involving OH-4 causes downfield shifts (>10 ppm) in C—4 from 71-73 ppm to 83-88 ppm. For example, the C—4 chemical shifts of D-ribono-l,4-lactone and D- ribono-imido-l,4-lactone are found at 88.2 ppm, while those for linear D—ribononitrile, D—ribonamide, D—ribonic acid and D—ribonate are found between 71.9 and 72.9 ppm (Table 3). Although C-A chemical shifts of the pentose phosphates (Section III, 35b) and tetroses (Section III, BSa) are relatively insensitive to configuration at C-2 and C-3, the C—4 resonances of the 1,4-lactones and imido-1,4-lactones having the £322 configuration (OH-2 and OH'3.E£§) are downfield (88.2 ppm) from those having the arabino configuration (OH—2 and OH-3 trans) (82.6 ppm). JC1,C3 and JC1,C4 for standard aldono—l,4-lactones and for imido— 1,4-1actones having the arabino configuration are about 7 Hz and <0.8 Hz, respectively. and J for the ribo isomers are JCl,c3 c1,c4 m2.0 Hz and ml.5 Hz, respectively. Acid hydrolysis (.__# AI ”‘0 g RAd MN UI (\g { p——r¢-———1 F-- UL KAAS HAS 1 Led“ ..__. L— ;———J——__y f____L____ IO 3.7 ' 5. Time (h) ()1 Cyanohydrin reaction profile at pH 10. 5. Data were obtained from 13C NMR spectra of [1-13C1—enriched intermediates. Reaction parameters are the same as those described in Figure318. AI, D—[1-13C]arabinono-imido-l, 4-1actone; RI, D-[1-13C]ribono-imido-l,4-1actone; AAd, D-[1-13C]arabinon— amide; RAd, D-[l-13C1ribonamide; UI and 02, dimer 12. 96 Dimer Z2 is observed (g1,.§2) and appears to arise from imidolactone. Aliquots taken from reaction mixtures at pH 10.5 during the first hour were quenched to pH 4 to determine whether the amount of lactone observed at pH 4 corresponded to the amount of imidolactone present at pH 10.5. As shown in Figure 20, imidolactones present at pH 10.5 are quantitatively converted to aldono-l,4-lactones at pH 4. Assign- ment of the broad C-l resonance at 175.3 ppm (Table 2 and Figures 18A, 188) to D-[l-lBC]arabinono-imido-l,4-lactone is based on the observa- tion that D-[1-l3C]arabinono—1,4-lactone is the predominant lactone produced in this experiment. The formation of five—membered imido- lactones in the cyanohydrin reaction is suggested by the production of aldono—l,4-1actones during acid hydrolysis. At pH 10.5. the reac~ tion sequence appears to be nitrile + imido-l,4-lactone + amide + aldonate. Aldononitrile hydrolysis at pH 8.5 is slow. During the first 30 min, aldononitriles comprise 95 percent of the reaction mixture (Figure 21). Imidolactone (arabino), observed next, reaches its maximum con- centration (5%) after 40 min. Amides are produced slowly and are stable. Dimer 12 is produced concomitantly with amide. D-Arabinono- 1,4-lactone forms by hydrolysis of D—arabinono—l,4-imidolactone at pH 8.5. At this intermediate pH value, the products of imidolactone hydrolysis are both lactone and amide, as described by Schmir and Cunningham (162). At pH 8.5, aldonolactones hydrolyze slowly to aldo- nates. The latter are not formed from amide hydrolysis, which occurs at a negligible rate. At pH 8.5, ammonia released from the hydrolysis of imidolactones reacts with the protonated imidolactones to produce amidines 13 (%15%). 97 Figure 20. Conversion of imido—l,4-lactones to aldono—l,4-lactones at pH <4. Th3 reaction at pH 10.5, 0.3 M and 18° 1 1°C with D-erythrose and [1 C]cyanide was examined three times during the first 60 min by 13C NMR to determine the amount of aldonamide (I) and imido-l,4-1actone (O). From the same reaction, three 0.5 mL aliquots were quenched with HC1 (0.7 mg, 0.3 M), incubated for 20 min, adjusted to pH 5 and exa— mined by l C NMR to determine the amount of aldonamide (O) and aldono- l,4-lactone (:1). Percentage of intermediates is based on the total peak areas in the spectra arising from the aldonamides, imido-l,4- lactones and l,4~1actones, and therefore, does not contain a contribu- tion from dimer 12 (U1, U2). 98 IOO 90— 80- . _ p _ _ O O O 0 83x22 cozoomm E Emocmn. 30 4O 50 60 Time (min) 20 IO Figure 20. 99 Figure 21. Cyanohydrin reaction profile at pH 8.5. Data were obtained from 13C NMR spectra of [l-lBC]-enriched inter- mediates. Reaction conditions are the same as those described in Figure 16. RN, D-[l-13C]ribononitrile; AN, D-[1-13c arabinononitrile; AI, D—[1-13C]arabinono-imido-l,4-1actone° RAd, D—[l- C]ribonamide; AAd, D-[1-13C]arabinonamide; AAm D-[1-13C]arabinono-amidine; AAs, D-[l-13CJarabinonate; RAs, D-[1-13C]ribonate; AL, D—[1—13C]arabinono- 1,4-lactone; U1 and U2, dimer 12. 100 Zwm 1 Om -Avw flue ainixuN uouooeg UIiUGOJad Figure 21. 101 When stoichiometric amounts of NH4C1, D-erythrose and [13C]cyanide I»: Cu are mixed at pH 8.5 or 9.5, amidine formation is stimulated. At pH 8.5, amidine formation is slow (55% in 50 min), whereas,at pH 9.5, it is rapid and almost quantitative. The arabino epimer (SC-1 = 173.2 ppm) and ribo epimer (5 = 171.6 ppm) are produced in both cases in C-l a 3:1 ratio (Figure 22B). At pH 8.5, standard cyanohydrin reaction mixtures contain arabinonoamidine (13%) while the £292 isomer is barely observable (<22). These observations are consistent with those of Hand and Jencks (164) who observed that the rate of ethyl benzimidate aminolysis is pH-dependent, with the maximum rate occur- ring at pH 8.9. Assignment of the arabino configuration to the amidine C-l re- sonance at 173.2 ppm is based on the chemical shift of the principal product from aminolysis of authentic D-[l—lBC]arabinononitrile at pH 9.5. When D-[1-13C]arabinononitrile is incubated at pH 9.5, a small amount of D-[1-13C]ribononitrile is formed by reversal of the conden- sation reaction. Thus, when D-[l-l3C]arabinononitri1e is treated with ammonia at pH 9.5, approximately 10 percent of the amidine formed has the ribo configuration. Dimer 12 comprises 35 percent of the reaction mixture after 4 h at pH 8.5. It appears to arise from imidolactone, since solutions 102 Figure 22. Effect of NH4C1 on the formation of intermediates. (A) 13C NMR spectrum of the reaction mixture at pH 10.5 after 1 h, showing D-[l- 3C]arabinonamide (AAd), D-[l-13C]ribonamide (RAd), and dimer 12 (U1, U2). (B) C NMR spectrum of the reaction mixture at pH 9.5 with one equivalent of NH4C1 after 1 h, showing D—[l- C]ara— binono-amidine (AAm), D—[1—13C]ribono-amidine (RAm), D—[l-l C]ara- binonamide (AAd) and D—[l-l C]ribonamide (RAd). Reaction conditions are the same as those described in Figure 16. 103 En: ow ov ow . co co. ca. 03 om. owi—i . _ . . 4 . . _ _ _ . _ . a _>stttt. is: ...??>$22L3e31..\t} z}...lii1$i.fs its? r§.5ar.ii$§>»i&(§ki;. #4. i t. 3... p5. E amide aldono—1,4-1actone-—-O aldonate The condensation of HCN with D-erythrose at pH 7 produces aldono- nitriles in the ratio of 3:2 ribozarabino, and their hydrolysis is slow. After 11 h, aldononitriles comprise 38 percent of the reaction mixture (Figure 14). Hydrolysis products at this time include aldo- A nates, amides, 1,4-lactones, amidines and dimer 2. Imidolactones are barely detectable (<22) , and their direct hydrolysis to lactones is faster than at higher pH. The arabino epimers predominate in all of the in- termediates except for the aldononitriles. Aldonates are formed by 1,4-lactone hydrolysis since amide hydrolysis is negligible at pH 7. The apparent sequence of the reaction appears similar to that at pH 8.5 with differences occurring only in the relative amounts of intermediates produced. 6. Imidolactone formation At all pH values, imidolactones appear to play an important role in the hydrolysis of those aldononitriles that can form them. Varma and French (46) have proposed that six-membered imidolactones are intermediates in the reaction of cyanide with D~arabinose. The slow rate of hydrolysis of C aldononitriles (4) relative to C5 4 105 aldononitriles (6, 7) suggests that the formation of five- membered imidolactones is not favorable, although there are many examples where ring-closure involving an spZ-hybridized carbon pro- ceeds to yield the more stable five-membered ring with an exocyclic double bond (166). For these reasons, the ringrforms of the imido- lactone and lactone intermediates required further investigation. The C-l chemical shifts of 1,4-1actones and 1,5—1actones are not identical, as shown in Table 3. In addition, 1.5-lactones do not revert to 1,4-lactones under conditions where imidolactones hy- drolyze to lactones. Therefore. the production of 1,4—lactones pro- bably reflects the presence of imido-l,4—1actones. Direct evidence for imido—l,4-lactone intermediates was obtained by preparing [1,3-13C]a1dononitriles from DL-[2-13C1erythrose and K13CN, and [1,4-13C]aldononitriles from DL-[3-13C1erythrose and K13CN. With these compounds, the chemical shifts of C-3 and C-4 and 13 13 C- C coupling constants between the enriched nuclei are readily determined (Figures 23A and 24A, Table 3). Following condensation at pH 8.5, the 13C NMR spectra of the aldononitriles were obtained. The reaction mixture was then adjusted to pH 10.5 where rapid conversion of aldononitriles to imidolactones occurs. Results are shown in Figures 23B, 23C and 24B. The 13C chemical shifts and l3C-l3C coup- ling constants of imidolactones obtained from these experiments are similar to those for 1,4-lactones prepared by standard methods. The large downfield shift of C-4 observed in the imidolactones esta- blishes that 0H-4, and not OH-5, is involved in imidolactone ring formation. Further evidence supporting formation of imido-l,4-lactone was 106 Figure 23. Addition of [13C]cyanide to DL-[2—13CJerythrose. 13C NMR spectra show only [13CJ—enriched C—1 and C-3 carbons. (A) NMR spectrum of the reaction mixture after 13.5 min at pH 8. 5 showing the formationCHFDL- [1, 3-13CCJarabinonon1trile (a) and DL— [L 3-i3g1ribon- onitrile (b). ZJCl C3 is <0. 7 Hz in the aldononitriles. (B) C NMR spectrum of the reaction mixture from A 3after 11 min at pH 10. 5. DL'il 3-13C]arabinonamide; b, DL— [1, 3-13CJribonamide; c, DL-{l, 3—13CJ- ribono-imido-l, 4- -lactone; d. DL- [1, 3-13C C]arabinono— imido- l, 4- -1actone; e, [13C]cyanide. (C)13C NMR spectrum of the expanded C-3 region of B. showing JCl C3 for the imido- l, 4- lactones. Like the linear aldononi- triles in A, JCl C3 is <0. 7 Hz in the aldonamides. 107 Al C 1 C-3 region his... } ,‘ AAd i i RI RAd % 1 €1.63 hi m» L_ L 1 l . l 76 74 72 70 PPM B d] b a . C 8 bed A b a a b l A I r 41 A l 4 I J A 1 180 160 140 120 100 80 60 ppm Figure 23. 108 Figure 24. Addition of [13C]cyanide to DL-[3—13C]erythrose. 13C NMR spectra show only the [13CJ-enriched carbon at C- 4, since the acquisition parameters [16 usec (90°) pulse with no delay time do not permit the detection of the unprotonated carbon at C-1. (A)1C NMR spectrum of the reaction mixture after 21.5 min at pH 8. 0, showing the C- 4 resonances of DL-[1,4-13C]ribononitrile (a) and DL—[1,4-13Clarabinono- nitrile (b). Inset shows 3JC1 C4 for the arabino epimer. (B) 3C NMR spectrum of the reaction mixture from A after 10 min at pH 10. 5. 3C]cyanide; b, DL- [1, 4-13C]ribono-imido-1, 4-—lactone; c, DL-[l, 4-13C1— arabinono-imido—l, M lactone; d, DL- [1, 4-13C1ribonamide; e, DL- [1, 413 C]- arabinonamide. Inset shows 33C1 C4 for DL-[1,4-13C]arabinonamide, and a broadened C— 4 resonance for DL- [1, 4-13C]ribono-imido-l, 4—1actone. 109 [1.4136] Intermediates A A ‘ A ‘ t . .1 . I-A 11.1 m. 1 IL A: ‘ I. .‘v 7 A 1.. l A A.“ A u .... ¢ up: ”AIM“ -rm ‘W‘r‘r “”7. w. H w «v11 Ier v: vrv'ymvvv- RN AN A *IJCIC4 a w$bfltw4 \'MMWL b 74 70 WW ‘ w " "wa- 1 A“ x 1 l 1 . 1 . L 1 | 1 ..L_ A 1 . l 180 160 140 120 100 80 60 40 20 pp"! Figure 24. 110 obtained by preparing the four diastereomeric [2—13C1aldononitriles from D—[l-lBCJerythro, threo—2,3-dihydroxybutanal (Figures 25A, 253) and KCN having structure 14. Aldononitriles 14 (Figure 25C) readily convert to imido—l,4-lactones 15 at pH 10.3 (Figure 25D). The C—2 resonances of the S-deoxyimido-l,4—lactones were assigned based on their close agreement with C-2 of the corresponding S—oxy analogues (Table 3) and on the ratio of epimeric imidolactones produced from addition of cyanide to D—erythrose and D-threose. The S-deoxyimido- l,4-lactones hydrolyze readily to the corresponding aldonamides at pH 10.5,as do the corresponding S-hydroxy compounds. EN (H) mm CH 0 (H)C(OH) NH2+ I H ( ) (H H?OH CH3 (OH) (OH) 14 25 Brown, et al. (166) have discussed the behavior of five- and six- membered rings having egg double bonds in a variety of chemical reac— tions. They concluded that the formation of five-membered rings is more facile due to their greater stability while the reactivity of six-membered rings is greater due to their instability. This differ- ence in ease of formation and in chemical reactivity affects the in- terpretation of data concerning ring formation in the cyanohydrin reaction. The presence of imido-l,4-lactone in reaction mixtures does not demonstrate that aldononitriles hydrolyze solely through this intermediate. An undetectable amount of imido—l,S-lactone may be present, with hydrolysis proceeding through the 1,3-ring, as shown 111 Figure 25. Addition of cyanide to D-[1-13CJerythro, threo-2,3-dihy- droxybutanal. 13C NMR 5 ectra show only the enriched carbons. (A) 13C NMR spectrum of D-[l--l C]erythro, threo-2,3-dihydroxybutanal. The aldose is hy- drated (h) in dilute aqueous solution g<0.l M). Peaks x are contami~ nants. (B) 13C NMR spectrum of D—[l-l CJerythro, threo-2,3-dihydroxy- butanal at W1 M, showing the presence of dimers and/or oligomers and hydrate (h). (C) 13C NMR spectrum of the reaction mixture after 12 min at pH 8.0. The C-2 resonances of the four diastereomeric D- [2- C] S—deoxyaldononitriles are observed, with peak b having twice the area of peak a and peak c. (D) 13C NMR spectrum of the reaction mixture from C after 8 min at pH 10.5, showing the formation of imido- l,4-lactones. Inset shows the C-2 resonances of D-[2-13C] S—deoxy- arabino- (a), xVlo—(b), lyxo—(d) and ribo-(f) imido-l,4-1actones, D- [2- C] S-deoxyaldonamides (c, e) D-[2-13C] S—deoxyaldononitriles (g) and contaminants x (h, i). (E) 3C NMR spectrum of the reaction mix- ture from D after 47 min at pH 10.5, showing increased amounts of D- [2-13C] S-deoxyaldonamides and the absence of D-[2-13C] 5—deoxyaldono- nitriles. 112 H i“ M38014 (H0)§(H) A HCOH CH3 w.“ ’5 ' a \ 1 h W x UL L__ ———- (b ‘ a c +HCN: pH 8.0,” min Figure 25. 113 in Scheme IX. A significant proportion of aldononitrile will hydro- lyze through the 1.5-ring if klO >> k9. Under these conditions, never- theless, it would appear that hydrolysis proceeded through the 1.4- ring, since K6 >> K7. The possible involvement of 1.5-rings can be excluded by examining the reaction at pH 7, where aldononitriles cy- clize very slowly and imidolactones decompose almost equally to amide and lactone. If the imido-l,5-lactone was hydrolyzing, reaction pro- ducts would include l,5-lactone or, more likely, its decomposition product (aldonate) early in the course of reaction. The results show that aldono—l,4-lactone accumulates 2-3 h before k K imido-l,4-lactone'-;2* products % aldononitrile K‘7\ k10 imido-l,5-lactone --+ products Scheme IX aldonate is detected and that 1,5-lactone is not observed. Thus, aldonate appears to arise from 1,4-lactone hydrolysis. Imido- l,4-lactone is, therefore, not only formed predominantly from aldononitrile, but appears to be the predominant substrate for hydrolysis at pH 7. 7. Conclusions The reactions that occur following the addition of cyanide to D- erythrose are summarized in Figure 26. The percentages of products at several DH values are listed in Tables 4 and S. In the follow1ng discussion, the effects of pH and chemical structure on reactions 1-10 of Figure 26 are summarized. The presence of tetrahedral intermediates 114 .mmouzuzuoan cu mmfiaaam aofiuummu afiuvznocwmu man no Emficmnuwz .om musmwm meHEOn—Ad mgdzgd = m umo .— 3 @mfiZHgOZng fin mflZHgOZHmmé bl Jm JN mflZOBUsI ¢.HIOZO§ IEO «32 03+ g m r +z .. mmzoaufiu 1782.: 11v mmzHasz m “my mmEHQ WZHZAN mam—OBOE! v.HIOQHZH w =. 3250266 4” mmommsammnm + . 26: a 115 Table 4. Total percentage of intermediates and products in the cyano- hydrin reaction (D—erythrose) at various pH values, . . . a , Compound Percent in reaction mixture (:34) pH 7.0b pH 8.5C pH 10.5d pH 12.7e nitrile 67 9 imido—l,4-lactone amide 8 36 84 l,4-lactone lO 2 aldonate ‘ 2 S 100 amidine 10 13 y;, g; (dimer 12)f 3 35 l6 l3 Determined by C NMR. The absence of an entry indicates that <2Z was observed. b 6 h of reaction. C 4 h of reaction. d 55 min of reaction. e 3.5 h of reaction. Percentage of dimer 12 was deter- mined by multiplying the fraction of the total spectral area_under El +.§§ x 100. 116 Table 5. Relative amounts of arabino epimers produced during the cyanohydrin reaction (D-erythrose) at various pH values. Compound Relative_percent of arabino epimera (:32) pH 7.0b pH 8.5C pH 10.5d pH 12.78 D-arabinononitrile 33 37g N40 D-arab1nono-1m1do—l,4— f 84g 86 lactone D-arabinono-amidine 90 N95 D-arab1nono—l,4- N90 b lactone D-arabinonamide 56 65 73 82 l3 Determined by C NMR. served. 100. b 6 h of reaction. Determined from pH-quench experiments. Absence of an entry indicates <22 was ob- Relative percent 8 (arabino epimer/arabino + ribo epimers) x C 4 h of reaction. 55 min of reaction. Amide percent is that ob- served after 45 min of reaction while the imidolactone percent is that observed after 15 min of reac ion and estimated from the 1,4-lactone (acid hydrolysis) products. at 40 min. Only the arabino epimer is observable g After 10 min of reaction. 117 (carbinolamines) in the reaction is discussed. With stoichiometric amounts of reactants. aldononitrile formation (reaction 1) is essentially complete at pH values between 7 and 10.5. while at pH > 11, D-erythrose remains during the first phase of the reaction. The rate of aldononitrile hydrolysis increases with in- creasing pH. After 6 h at pH 7, 67 percent of the reaction mixture is composed of aldononitriles (Table 4) compared with 9 percent after 4 h at pH 8.5 and none after 1.7 h at pH 12.7 (Figure 16). Imidolactone formation (reaction 2), while not essential, is the major route of hydrolysis of aldononitriles with hydroxyl groups situated to permit ring-closure. For pentononitriles, imido-l,4- lactone rings are formed in preference to 1.5-rings. At pH values be- low 9.5, imidolactone formation is slow,whereas,at higher pH (10.5),it is rapid and essentially complete after 10 min (Figure 18A, 183). At low pH (<4),imidolactones hydrolyze rapidly to lactones via reactions 3, 6 and 9. This reaction was used to establish the presence of imido- lactone in reaction mixtures at pH 12.7 (Figure 17). As pH is in- creased. hydrolysis to aldonamides is favored and occurs via reactions 3, 6, 7 and 8. At pH :_10.5, amides are the major products of imido- lactone hydrolysis. Reaction 4 produces the proposed dimer 12, yielding the greatest amount of product at pH 8.5 and lesser amounts at higher and lower pH values (Table 4). This result is due to the effect of pH on the reactions that affectthe availability of nitrile and imidolactone needed for the formation of 12. At pH 10.5, reaction 2 is essentially complete. leaving a smaller amount of nitrile for reaction 4 than is present at pH 8.5. At pH 7, reaction 2 is slower and reaction 6 is 118 probably faster, resulting in lower concentrations of imidolactone and faster imidolactone hydrolysis than at pH 8.5. At pH 7, amino- lysis (reaction 5) also competes for imidolactone to limit the production of 12. Increased imidolactone hydrolysis to lactone (reac- tions 3, 6 and 9) at pH 7 yields more ammonia to Stimulate aminolysis, further decreasing the amount of imidolactone available for reac- tion 4. The amount of 12 formed during the reaction should be dependent on the concentration of reactants as discussed earlier, since reaction 4 is bimolecular. In support of the proposed reaction, condensation of cyanide with D—erythrose at pH 8.5 and 0.9 M produces 50 percent more 12 than the same reaction at 0.3 M after 4 h. Reaction 5 was observed only at pH 7 and 8.5 and depends on reac- tion 9 for ammonia. At pH 8.5, imidolactones hydrolyze predominately to amides (3:1 amidezlactone + aldonate after 4 h) (Table 4). Aldo- nates produced at pH < 10.5 arise primarily from the sequence imido- lactone + lactone + aldonate, and are, therefore, added to the per- centage of lactone when partitioning at pH < 10.5 is discussed. Amide hydrolysis proceeds slowly with the release of ammonia. In contrast, imidolactones hydrolyze to amides and lactones in 2:3 proportion at pH 7 after 6 h. liberating ammonia for reaction 5 and leading to in- creased production of amidines (30 percent of the non-nitrile product after 6 h). Reaction 6 is implicated by the behavior of imidolactones in acidic and basic solution. Schmir and Cunningham (162) proposed tetrahedral intermediates in the hydrolysis of N-substituted imido- lactones, and explained the partitioning between lactone and 119 N-substituted amide in terms of the presence of different ionic forms of carbinolamine. Neutral carbinolamine (or its zwitterion) was pro- posed to yield lactone, while anionic carbinolamine yields amide (Figure 26). Although carbinolamine resonances are not observed by 13 . . . C NMR, 1t 18 reasonable to propose reactions 6 and 7 based on the observed behavior of the imido-l,4—lactones in acidic and basic media and on results reported previously for N-substituted imidolactone hydrolysis (162). The hydrolysis of amides is facilitated when hydroxyl groups parr ticipate to produce cyclic carbinolamine intermediates (167-169). Cunningham and Schmir (169) studied the alkaline hydrolysis of 4- hydroxybutyranilide and concluded that reactions 7, 8, 9 and 10 or 8. 9, and 10 are involved. Aldonamides are stable over several hours at pH < 10, while hydrolysis can be followed at pH > 11. The alkaline hydrolysis of purified D—ribonamide or D-arabinonamide proceeds with retention of configuration at C-2, indicating that the overall reaction does not reverse to reaction 1, where stereochemis- try at C-2 can be altered. It is assumed that reaction 1 establishes C-Z configuration, and that the predominant epimer is determined by reactions 1 and 2. Isomerization of imidolactone to eneimine could provide another route for epimerization. This has been ruled out, however. since reactions at pH 8.5auu112.7 in 3H O produce products 2 containing only 0.003% and 0.014% tritium, respectively. In addition, only aldonates are observed by 13C NMR during amide hydrolysis, de— monstrating that reaction 6 is not appreciably reversible. If pre- sent (33), imidolactone would be detected under these conditions. The irreversibility of reaction 6 for the hydrolysis of N-substituted 120 imidolactones has been established by Schmir and Cunningham (162). The rate of amide hydrolysis appears to depend on amide configura— tion. Hydrolysis of an equimolar mixture of D—ribonamide and D- arabinonamide at pH 11 shows that D-ribonate is formed about 2.5 times faster than D-arabinonate. This result probably reflects dif- ferences in the ratesof cyclization of the amides (reaction 8) and/or the rates of carbinolamine breakdown (reactions 7 and 9). This rate difference between C-2 epimers is not anticipated from direct hydrox- ide ion attack on the carbonyl of linear amides,whereas reactions in- volving cyclic intermediates would probably show such differences (170). Reaction 10 occurs rapidly at pH > 10.5 and lactones are not ob- served in reaction mixtures at these pH values. Lactone hydrolysis is slower at pH 8.5 and lactones are detected in reaction mixtures under these conditions. Lactones in reaction mixtures at these pH values are derived from the hydrolysis of imidolactones, and not from amide hydrolysis. Differences in the proportions of arabino and ribo epimers are apparent at various stages of the reaction,as shown in Table 5. At all pH values, ribononitrile predominates (63-67Z). In comparison, D- erythrose 4-phosphate, which cannot cyclize and exists predominantly as hydrate in aqueous solution, reacts with cyanide at pH 8.0 to yield 58% ribononitrile S-P. Nitrile cyclization (reaction 2) favors the arabino configuration with arabinono-imido—l,4—lactone accounting for ”85% of the imidolactones under conditions where hydrolysis is slow. The predominance of arabinono-imido-l,4-lactone is reflected in acidic (1.4—lactone) and alkaline (amide) hydrolysis products and in amino- lysis products, with 70—90% having the arabino configuration (Table 5). 121 Reaction 6 may be sensitive to configuration, since the percen— tage of arabinono-imido-l,4-1actone (%8SZ) does not equal the epimeric percentage of arabinonate after total hydrolysis, with the latter varying between 68—79 percent (Table 1). As established by the demon- stration that aldonamides do not undergo epimerization during hydroly- sis, this difference cannot be ascribed to differences in the rates of reactions 7-10. The slower rate of disappearance of C aldononitrile compared to 4 C5 aldononitrile (Figure 11) cannot be explained by differences in ring- forms of intermediate imidolactones, since alkaline hydrolysis of C5 aldononitriles involves the formation of imido-l,4-lactones. The rela- tive ease of cyclization of C4 and C5 aldononitriles can be estimated by measuring rates of amidine formation. Ammonia does not react with DL-glyceronitrile at pH 9.5, in agreement with previous studies show- ing that amidine formation occurs readily from imidates (164), and that the addition of ammonia to acyclic nitriles to form amidines occurs only under more rigorous conditions. Conversion of imido-l,4-lactones to amidines (aminolysis) occurs readily at pH 9.5 and rates of amino- lysis of C and C5 imido-l,4-1actones are not expected to be signifi- 4 cantly different. Thus, a difference in the overall rate of conver— sion of 4 and 6 to amidines will reflect a difference in their rates of cyclization (cyclization is the rate determining step in this process). Rates of amidine formation are shown in Figure 27. A fivefold faster rate is observed for C5 aldononitriles than for the C4- homologue. We conclude that imido-1,4-lactones form less readily when a l°-OH is participating than when a 2°-OH is involved. 122 | p 00 AAm + mm A A 90 - .vfi’—______JP 80L ‘////”EAm+TAm ‘1’ :3; 70+ 2 C .e 8 O Q) G: .s 2.5 U E 0. 0.5 LO l5 2.0 2.5 3.0 Figure 27. Time (h) Rates of amidine formation from C4 and C5 aldononitriles. Data were obtained from.l3C NMR spectra (loo-200 scans). Reaction conditions: 0.3 M in aldose (D—glyceraldehyde or D-erythrose), [ 3C1c§anide and NHACI; 18° 2 1°C; pH‘ 9.5 2 0.2. AAm, D-[l-1 C]arabinono—amidine; RAm, D- [1-13C ribono-amidine; TAm, D-[l-13C]threono-amidine; EAm, D—[l—1 C]arythrono-amidine. 123 The presence of -CH7OH or -CH substituents on C-4 may orient the 4-OH 3 group for easier attack on the nitrile carbon. The slower rate of C4 aldononitrile hydrolysis (Figure 11) is primarily due to a slower rate of cyclization to imidolactone. In summary, we have presented evidence for the following: a) At pH 7-9, cyanide condenses stoichiometrically with C a1- 1'C4 doses to form aldonitriles quantitatively. Aldoses that form pyranose rings require a threefold excess of cyanide under similar conditions to promote complete condensation. b) The extent of cyanide addition decreases with increasing pH. c) The rate of aldononitrile disappearance decreases as pH de- creases and is affected by aldononitrile structure. The overall rate of reaction is faster for aldononitriles derived from D-erythrose than for those derived from D-threose. Aldononitriles having the arabino configuration hydrolyze more readily than those having the ‘gibg configuration; aldononitriles having lyxo and xylg_configura- tions hydrolyze at approximately similar rates. d) Hydroxyl groups in the prOper position(s) facilitate aldono- nitrile hydrolysis through the formation of imidolactones. Cycliza- tion of C5 aldononitriles occurs at pH 10.5 to form imido—l,4-lactones almost quantitatively. Aldononitriles having the arabino configuration cyclize faster than those having the Eibg_configuration. e) C Aldononitriles (erythro. threo) cyclize, and therefore, 4 hydrolyze slower than C5 aldononitriles (ribo, arabino),a1though five- membered imido—l,4-lactones are involved in the latter case. f) Imido-l,4—lactones hydrolyze in basic (pH > 10.5) and acidic (PH < 4) solution to yield aldonamides and aldono-1.4-lactones, 124 respectively, in agreement with the behavior of N-substituted imido-l,4- lactones (162). g) D-Arabinonamide and D-ribonamide appear to hydrolyze .at dife ferent rates at pH 11. with the latter approximately 2.5 times faster than the former. h) Amidines are formed in 85-95 percent yield from the C4 and C5 aldononitriles by reaction with one equivalent of ammonia at pH 9.5. 1) Linear compounds having arabino and ribo configurations can 3 be distinguished by 3JC1 C4° Furanoids (with an sp"-hybridized C-l carbon)having arabino and ribo configuration can be distinguished by J and J and b 13C chemical shift*' 13C NMR arameters of c1,c3 c1,c4 Y 5’ “ P 1,4-lactones and imido-l,4-1actones are similar. j) The following epimeric intermediates and products are ob- 13 . 13 . . .. served by C NMR u31ng K CN 1n the reaction w1tn D-erythrose: nitriles, imido-l,4-lactones, amidines, amides, a1dono-l,4-lactones and aldonates. A dimer(s) from reaction between imido-l,4-lactones and nitrile is proposed as an intermediate. B. Preparation of [13CJ—Enriched Aldoses,_Aldose Phosphates and Their Derivatives The synthetic route described in this study utilizes the condensa- tion of cyanide with an aldose or aldose derivative, as first described by Kiliani (26). In this classical reaction, a cyanide salt and aldose are mixed in aqueous solution at high pH to produce cyanohydrins which, in the alkaline solution, hydrolyze to aldonic acid salts. We observed, however, that cyanohydrins can be formed rapidly and essentially quanti- tatively at pH 8.0 i 0.5 with minimal hydrolysis, and that they are 125 stable at pH 4.0. As described by Kuhn (64) for the preparation of 2-amino-2-deoxyaldoses from 2-benzylamino-2-deoxyaldononitriles, aldononitriles can be hydrogenolyzed to aldoses in 70-80 per- cent yield. Furthermore, condensation with Kl3CN provides a convenient route for the preparation of [13C]-enriched derivatives of all the al- doses and of the C3 to CS aldose phosphates. The mixed aldononitrile epimers produced from cyanide condensation are reduced without purifi- cation and the product epimeric aldoses purified by chromatography (88, 89). Aldononitrile phosphate epimers can be separated prior to reduc- tion by chromatography at pH 3.9, illustrating the stability of cyano- hydrins and the absence of cyanide exchange at low pH values. Hydrogen pressure, pH and the structure of the nitrile all affect the ease of hydrogenolysis of cyanohydrins. Hydrogenolysis occurs smoothly in the absence of excess cyanide, which appears to poison the catalyst. In the course of this study it was observed that chloride ion promotes "over-reduction" to aminoalditols. Sulfuric acid rather than HC1 is preferred since sulfate ion does not inhibit hydrogeno- lysis, gives higher yields of aldose, and can be removed more easily than chloride ion. Iodate ion, at low concentrations, inhibits reduc- tion completely. In addition to the introduction of carbon isotopes, the catalytic hydrogenolysis of cyanohydrins provides a route to carbohydrates en- riched with hydrogen and oxygen isotopes. The technique permits the separate or simultaneous incorporation of carbon and hydrogen isotopes at C-1 and H-1, respectively, and oxygen isotopes at 0-2 for each cycle of cyanide addition and catalytic reduction. Successive appli- cation of condensation and reduction permits the synthesis of derivatives enriched at sites other than C-1 and C—2. A wide variety of selec- tively-enriched carbohydrates and their derivatives, which were diffi- cult to prepare previously, are now accessible. 1. Preparation of C7—C6 aldononitriles 2, C3 and C4 a1- doses proceeds almost quantitatively at low pH with a 1:1 ratio of The addition of cyanide to formaldehyde and the C reactants, as shown in Table 6. Aqueous solutions of the short chain cyanohydrins at pH 4.3 are stable and can be stored at low temperature (-15°C) for extended periods of time. The formation of aldononitriles is strongly favored by the inabi- lity of the starting aldose to exist as a pyranose. The complete con- version of C5 aldoses to C6 aldononitriles using stoichiometric amounts of cyanide is hindered by an unfavorable equilibrium. With 1:1 ratios of cyanide to aldose, at least 10% of the starting aldose remains un— reacted. Use of a threefold excess of cyanide at pH 7.8-8.0 in these reactions produces aldononitriles in better than 90% yield with lit- tle hydrolysis to the aldonates. When cyanide is the limiting reactant during the preparation of [13C1-enriched compounds, an excess of cya- nide is used and the unreacted reagent recovered efficiently and almost quantitatively by aerating the acidic mixture (pH 4.2) with nitrogen and trapping the hydrogen cyanide released in 8 M alcoholic potassium hydroxide. The nitriles are stable during removal of cyanide. Excess cyanide, C orl3C, should be removed by aeration prior to hydrogeno- lysis, since cyanide inhibits reduction. 2. Hydrogenolysis of C5 and C6 aldononitriles The catalytic hydrogenolysis of C6 aldononitriles at pH 4.2 i 0.1 and 60 lb in-2 H2 proceeds readily to yield hexoses. Figure 28 .o>qumucmu ma acoecwfimm< o .pmumowpca mm upms muw3 mummEumSnpm :d “:0wu uzaom 20:2 “0 :1 Hmuuwcw u fizz .mfimxamcm Hmuuomam mzz Una zn mama oumz mcofiumcfisuwuww .mmmmo ommzu cu .o:~m> vzu :uwz newsman sz dons m mzu mmmacs mwoumtxmma UAU Scum pocfianoump muwz mOMumu paw mwmucvuuom m so: Huw.a n no 0H o.m omocammwAIQ 7, me m.o M“ :mmm;mmmmms-vi:li.mm-Ll--£:!};.mw.l:li-n-;MW%:l:lt;m.w u fl:n. :1 mmoswnmuHMIa m am am o.m meszmcamuumaua sz and n ma ON m.w mpxcmpamuQUAHolo m mm on m.@ mpxsmpamaouxHo n no ow m.w opazmoamauom umwzm mafluudz Aswav muawuufiz venomous: mo meme mo memm Lo ucuoumm usoouom cowuomem 2e mew:m_ mzumz_ ucmuommx ummsm mnea~uaqcosom~m Lo sodomuwesue osu now mu~smou paw assauwpcoo :oHuomou owmwooam .o maame 128 Figure 28. Change in Hz pressure during hydrogenolysis of hexono- nitriles at pH 4.2 e 0.1 and 60 lb 1n-2. D-[2-13C]Glucono- and mannononitriles (5-6 mmol) were hydrogenolyzed over Pd/BaSO4 in a 250-mL reduction vessel. The drop in H2 pressure corresponds to an uptake of ~5-6 mmoles of H2. The 13C NMR spectrum of the hydrogenolysis products is shown in Figure 29B. 129 shows the uptake of H2 during the hydrogenolysis of 6 mmoles of D- [,_13 C]g1ucono- and mannononitriles at 60 lb in-2 in a 250-ml reduction vessel. Reduction is complete after 1 h. Typical yields, determined from product weights, GLC analysis and 14C incorporation are presented in Table 7. In reactions using lzl ratios of cyanide and C5 aldose, the ex- tent of nitrile hydrolysis during their preparation varies from 15 to 20% (Table 7% and 10% of the aldononitriles are reduced to l-deoxy—l- aminoalditols during catalytic hydrogenolysis. The final mixture of compounds, after treatment with Dowex 1 X8 (OAc-) contains 70-75% pro- duct aldose, with an overall yield based on the starting aldose of about 50%. Total incorporation of cyanide is 80% or better; that is, 50% in product aldoses, 20% in aldonic acids, and 10% in aminoalditols, all of which can be readily separated by ion-exchange chromatography and recovered. The remaining 20% of cyanide remains unreacted due to the cyanohydrin-aldose equilibrium. The overall yield of hexoses from pentoses is greatly improved when a threefold excess of cyanide is used in the condensation reac- tion (Table 6). Residual cyanide after aeration is reduced to methyl- amine or formaldehyde, both of which are separated readily from the products. The 13C NMR spectra obtained at various stages during the 13 13 synthesis of D-[2—13C1glucose and D-[2- C]mannose from D—[l- C]ara- binose and a threefold excess of KCN are shown in Figure 29. Spectra were obtained of the reaction mixture after cyanide condensation (Figure 29A), of the products after hydrogenolysis (Figure 293) and of the purified hexoses (Figures 29C and 29D). The intermediate 1 13 . . [2- 3C]aldononitriles, byproduct [2— C] l-amino-l-deoxyalditols, and 130 .o>wumuomu me ucoacwwmm< o .opfismzomoea_ mo aceumuoeuoocw :o momma anagrams mo mama» a .040 an pocwEwouop one: mHOquHm wcfiuaomou one .mpfippzs IOuoa somoow and: .ucveuwuuu A u m 1.! 1'1 mm as and m.H N.H Hum mmoxsauo an an me new Hum.e e.a m.H are mmoasxuo me an an odea Hum.~ eamev e.e names A.H Hue mmonee-o on oe mm Ame Hum.e em.o m.e and wmoxsau: on es On an: Hum.~ ~.H nAmAv o.e Hue mmoeanmuaua Aw m.e Aw w.H Aucmuuomv Auswoummv Aucmoummv muomzuv >uomzuv I29 120 mzpammz va va umwsm monomwm :o Hmuoe :0 ma xmBOQ mom0p~< H xmzoa om xmBOQ cu tzo AHOEE OHV momma mfimfiy momma mama» :H mumwsm mospoum Hmuwm umuwm mo mmOpH< HHmum>o aamuo>o moppOHm mo ofiumm uzwwmz unwwm3 Owumm wcwuumum .III'I'.'I'I-!-¢ -.I i'li‘l'l'll 1.-I.-C'i - .xufi>wuom0wpmu mam .muusmoee :owuozpmu mpfiup>20uon .muswfimz so momma mpamw» mmOpH< .N wanna 131 Figure 29. Preparation of D—[2-13C]glucose and D-[2-13C]mannose from D-[l—13C1arabinose and KCN. lH-Decoupled 15.08 MHz 13C NMR spectra were taken of the intermediate aldononitriles (A), crude hydrogenolysis products (B), and purified hexoses (C and D). (A) 13C NMR spectrum of the reaction mixture con- taining D-[2-13Cngucononitrile (GN) (62.8 ppm) and D-[2-13C]mannono- nitrile (MN) (62.4 ppm) at pH m4. Ac = natural abundance resonances of acetic acid. (B) l C NMR spectrum of the products from hydrogeno- lysis of the epimeric [2-13C]hexononitriles: A, D-[2-13C] l—amino—l- deoxyalditols (68.6 ppm and 70.6 ppm); resonances between 72-76 ppm are due to C-2 of the product [2—13C]hexoses. (C) 13C NMR spectrum of puri— fied D-[2-13C]glucose. showing C-2 of the a- and B-pyranoses at 73.0 ppm and 75.7 ppm, respectively. (D) 13C NMR spectrum of purified D-[2-13C]- mannose, showing C—2 of the d- and B-pyranoses at 72.2 ppm and 72.8 ppm, respectively. 132 MN A 6" Ac _, 14L fl 3 AA 1 Ac fi° g, 1 11L 1L 3 c Q G o D 11 is 1- .1.“ 1 - 1 1 L 1 L _L 1 L J . l g L L 180 160 140 120 100 80 50 40 20 Figure 29. 133 the product [2-13C]aldoses are readily identified by their character- istic chemical shifts. Quantitation of unreacted starting [1-13C1a1- dose and [2-13C] products is not hindered by the effect of different 13C relaxation times, since the enriched carbons in the starting al- dose (C-1) and in the products (C—Z) are methine carbons. The absence of starting [1—13C1aldose after reaction with a threefold excess of KCN is shown in Figure 29A. C-l Resonances for u- and 8- D-arabino- pyranose would be observed at 98.2 ppm and 94.0 ppm, respectively. The mixture of crude products of hydrogenolysis (Figure 29B) is pre- l3C] l—Amino-l-deoxyalditols are dominantly the desired aldoses. D-[Z- quantitatively removed by treatment of the crude products with Dowex 50 X8 (H+).and the epimeric [2-13C]aldoses are purified on Dowefo1X8 (200—400 mesh) in the barium form (88) (Figures 29C and 290). The preparation of C5 aldononitriles is conducted with stoichio- metric amounts of tetrose and cyanide (Table 6). Hydrogenolysis of the CS aldononitriles derived from D-erythrose at pH 4.2 i 0.1 and 60 lb in"2 H2 produces about 15% l-amino-l-deoxyalditols. The overall yield of D-ribose and D-arabinose after purification on Dowex 50 X8 (200-400 mesh) (Ba2+) (88) is about 75%. Hydrogenolysis of the C5 aldononitriles derived from D-threose under the same conditions typi— cally yields ”40% l-amino-l-deoxyalditols. At lower H2 pressure and pH (1 atm — 20 lb in“2 H2, pH 2-3), less amine (IO-15%) is produced. 3. Hydrogenolysis of C2, C3 and C4 aldononitriles Hydrogenolysis of C2, C3 and C4 aldononitriles proceeds differ- ently than that of the C5 and C6 cyanohydrins. Apparently, in the synthesis of the pentoses and hexoses, the intermediate five- and six- carbon imines can cyclize, facilitating formation of the aldose, 134 presumably through an intermediate glycosylamine (171). In the syn— thesis of shorter-chain homologues, cyclized intermediates can form only from the four-carbon nitriles. It appears that this process does not occur readily, however, as the reduction of erythrono- and threono-nitriles at pH 4.2 i 0.1 and 60 lb in-2 over palladium- barium sulfate at 25°C affords l-amino-l-deoxyalditols as the princi- pal products. In this regard, blocking groups that restrict the for- mation of cyclic intermediates effect the extent of reduction to l— amino-l-deoxyalditols. Hydrogenolysis of 3,5—0-ethylidene-ID-ribono- and arabinononitriles at pH 4.2 i 0.1 and 60 lb in”2 H2 yields, in addition to the corresponding aldoses, larger percentages (>40%) of amines than the unblocked homologues. Low yields of aldose are also obtained from hydrogenolysis of C2 and C3 aldononitriles under these conditions. The ease with which aldononitriles can be hydrogenolyzed to al- doses at high pressure parallels exactly their susceptibility to al- kaline hydrolysis. In both processes, cyclization has been impli- cated. Slower rates of hydrolysis of D-xylononitrile and D-lyxono- nitrile compared with D-ribononitrile and D-arabinononitrile,and slower rates of cyclization of the tetrononitriles relative to the C5 aldono- nitriles would be predicted based on the quantities of l-amino-l- deoxyalditols produced during hydrogenolysis. These differences have been observed (Figures 11 and 12). Aldononitriles that yield imines which cannot cyclize, however, are hydrogenolyzed smoothly and almost quantitatively at 25°C to the corresponding aldehydes at pH 1.7 i 0.1 and atmospheric pressure over palladium-barium sulfate (Table 8). Presumably, these conditions 135 permit the intermediate imine to dissociate from the catalyst and to hydrolyze without the intramolecular participation that appears to oc- cur during the preparation of the hexoses and pentoses. Table 8. Yields of aldoses from hydrogenolysis of two-, threes, and four-carbon aldononitrilesa. Compound Conditions Pd—BaSO4, PtOz, _9 Pt02, _2 Pd-BaSOa, atm. press. 60 lb in ' 30 lb in 60 1b in"2 pH 1.7 pH 0-8 pH 0.8 pH a.2 Glycolaldehyde 90(80)b DL-Glyceraldehyde 85(75)b tracec <15 D-Threose and D- 85(70)d tracec <10 erythrose a Based on 13C NMR peak areas. Spectra were obtained by using a 55° pulse and 10 sec delay time to minimize relaxation effects. Spectra were obtained of reduction mixtures. Integration was performed by computer. b Percent yield based on weight of products as gums. De- termined by gas chromatography. 9 Percent yield based on weight of products as gums after separation by chromatography on Dowex 50 X8 (200—400 mesh) (Ba2+). The 13C NMR spectra obtained at various stages during the synthe- sis of D-[l-13C]erythrose and D-[l-13C1threose from D—glyceraldehyde and K13CN are shown in Figure 30. Spectra were obtained of the reac- tion mixture after cyanide condensation, of the products after hydro- genolysis, and of the purified tetroses. The C-l resonances of the intermediate epimeric [1-13C]tetrononitriles are found at 120.9 ppm (Figure 30A). Hydrogenolysis at atmospheric pressure and pH 1.7 r 1 produces a mixture of the tetroses and small (8, purified aldononitrile phosphates revert to epimeric mixtures. Analysis of the hydrogenolysis products from DL-[l-lBC] 142 Figure 32. Separation of DL-[l-lBC xylononitrile 5—P and DL-[1-13C]- lyxononitrile S—P and l C NMR analyses of the products after hydrogenolysis over palladium. (A) Chromatography of the 2-epimeric pentononitrile phosphates on a 2.2 x 51 cm Dowex 1 X8 (200-400 mesh) column in the formate form at 4°C deve10ped with a linear gradient of sodium formate (3000 mL, 0.05-0.8 M, pH 3.9). Column effluent was assayed for radioactivity and total phos— hate. The xylo epimer was eluted before the lyxo epimer. (B and C) C NMR analyses showing resonances due to the enriched carbons of the reduction products from DL-[l-13C]xylononitrile 5-P (B) and DL-[l-13C]- lyxononitrile 5-P (C); C-l resonances of the [1-13C]-enriched u- and S- furanose and hydrated forms of DL-aldose 5—P appear at approximately 100 ppm, [l—l3C]-l—amino-l-deoxyalditol 5-P (a) appears at approximately 43 ppm, and resonances due to natural abundance 13C of acetic acid (Ac), used to adjust pH prior to hydrogenolysis, appear at approximately 23 and 180 ppm. Spectra were obtained at 13 t 1°C with a sweep width of 3000 Hz and a filter width of 2400 Hz. 143 12 30- A g S o 20. ‘8 X . I? ,'. Q 11.. a ‘5 |0r 3. ‘4 0 . '. O " '. .9 5 1 t, , .. c ,o . ‘ 1, CE l). -\ 1’74; " 1.\\ EW_,HW»’AWI “LIV-77:. ‘ 1000 2000 3000 Eluont Volume (ml) 3 C B h o A; 111 .. _. ,..L A: A. ,1, “ ° I I 1 1 1 l 1 1 I80 160 I40 120 100 80 60 40 PPM Figure 32. (----.) (WLU) snmoquoqd .0101 144 Table 9. Purification and epimeric distribution of aldononitrile phosphates. Parent Aldose Chromatography on Ratio of Phosphates Dowex 1 X8 EpimersC (formate)a Peak ib Peak 2b Glyceraldehyde 3—P threo erythro 1.3:1 erythro Erythrose 4-P arabino ribo 1.4:1 ribo Threose 4-P xylo lyxo 1.5:1 lyxo a A 2.2 x 51 cm Dowex 1 X8 (200-400 mesh) column in the formate form was employed. Solutions of aldononitrile phosphates were adjusted to pH 6.5-7.0 prior to application to the column bed. Gradients: for 4-carbon aldononitrile phosphates, 3000 mL, 0.2-0.9 M sodium formate, pH 3.9; for 5-carbon aldononitrile phosphates, 3000 mL, 0.05-0.8 M sodium formate, pH 3.9. Temperature = 4°C; flow rate 0.5 mL/min; 7 mL per fraction. b Aldononitrile-P configurations were determined by reduction to aldose phosphates and incubation with alkaline phosphatase. The l C NMR spectra of the resulting aldoses were compared with those of standard pentoses. c Deter- mined by computerized integration of 1 C NMR spectra of epimeric mixtures and by quantitation of organic P in purified preparations. 145 xylononitrile S-P and DL«[l-13C]lyxononitrile 5—P by 13C NMR is shown in Figure 32B and 32C, respectively. DL-[l-13C] l-Amino-l-deoxyalditol 5—phosphates are usually formed in 5-10% yield during reduction. They are removed from the product aldose phosphates by chromatography on DEAF-Sephadex A-25 (OAc’) at 4°C using linear gradients of sodium acetate at pH 4.5. Recovery from DEAF-Sephadex chromatography is about 90% based on P-assay. For the preparation of millimolar quantities of aldose phosphates, glycolaldehyde-P and D-glyceraldehyde 3—P were prepared from DL- glycerol 1-P and D-fructose 6-P, respectively, by lead tetraacetate oxidation. Sodium metaperiodate oxidation was also examined, but traces of iodate interfere with hydrogenolysis of the aldononitrile phosphates, and careful chromatographic purification was required. On the other hand, aldononitrile phosphates prepared from lead tetra- acetate oxidation products hydrogenolyze smoothly. Aldose phosphates, particularly the triose and tetrose phosphates, should be handled at low pH to avoid base-catalyzed isomerizations and B-elimination. The acyclic triose and tetrose phosphates isomerize to give mixtures which include keto compounds when chromatographed on Dowex 1 X8 (formate). Purification of the alkali-sensitive aldose phosphates and the pentose phosphates can be achieved by anion-ex- change chromatography on DEAE-Sephadex A-25 at 4°C using linear gradients of acetic acid at pH 4.5 r 0.1. The tetrose 4-phosphates consistently yielded skewed peaks with notable tailing, whereas triose and pentose phosphates yielded symmetric peaks. Isomerization to keto compounds on DEAF-Sephadex was not observed under the condi- tions used. 146 It is usual to prepare aldose phosphates having four or fewer car— bons as acetals to protect the base sensitive aldehydic function (73, 74). We find, however, that the free aldose phosphates are stable during long-term storage at pH 1.0-2.0. When stored at -15°C as 50 mM solutions, no detectable changes occur over a two month period as de- termined by 13C NMR analysis of the [l-l3C]-enriched compounds and by inorganic phosphate analysis. Storage at higher temperatures, however, results in degradation of these compounds even in acidic solution. 5. 13C NMR parameters a. Short-chain aldoses and derivatives The various forms of the aldoses in solution can be determined with ease from 13C NMR spectra of the [1-13C]—enriched compounds, as shown in Figures 29-32. Formaldehyde (172), glycolaldehyde (173) and glyceraldehyde (174) appear to exist predominantly as ggmfdiols in dilute, aqueous solution. The observation of carbon-l3 resonances of the [1-13C]—enriched derivatives at approximately 90 ppm is not typi- cal of aldehydes or hemiacetals, and has been shown to be characteris- tic of ggmfdiols of aldohexose derivatives (175). A significant pro- portion of gemydiol (hydrate) exists in aqueous solutions of erythrose (”12%) and threose (W12%). The remainder of the tetroses is present in aqueous solution as B-furanose (37% threose, 63% erythrose) and a— furanose (51% threose, 25% erythrose) forms. Assignments of the carbon-13 resonances to a- and B-furanoses were made by analogy with those of the methyl c- and B-glycosides (114). The presence of dimers and/or oligomers of glycolaldehyde, DL- glyceraldehyde (Figure 33), D-erythrose (Figures 30C and 31D) and D— erythro, threo-2,3-dihydroxybutana1 (Figure 25) has been observed in 147 Figure 33. Various forms of glyceraldehyde in aqueous solution. The natural-abundance 13C NMR Spectrum of polymeric D-glyceraldehyde (A) prepared according to Perlin (139), and the l C NMR Spectrum of DL- [1— 3C]glyceraldehyde (B) showing only C-l resonances. Peak (h) is the linear, ggmediol carbon at C-1; the remaining downfield peaks arise from the C-1 resonances of dimers and higher polymers. 148 l1 1- “c l l l 1 110 90 70 50 mm Figure 33. 149 concentrated solutions or in solutions freshly prepared from syrups. These structures hydrolyze rapidly to the gemfdiol monomer in the case of glycolaldehyde (173). The depolymerization of glyceralde- hyde is slower but eventually yields monomer (176). Concentrated, neutral solutions of D—[l-l3C]erythrose contain dimers and/or oligo- mers having C—l chemical shifts at 98.5, 97.9, 92.5 and 92.1 ppm. Carbon-l3 chemical shifts obtained from specific [13C]-enrich- ment of the short-chain aldoses are presented in Table 10. As demon— strated for erythrose in Figure 31, [13C]-enrichment permits un- equivocal assignment of chemical shifts of all carbon atoms. Several observations regarding these l3C chemical shifts are noted. The nitrile carbon of the aldononitriles is found at approxi— mately 120 ppm, and C-2 of aldononitriles is more shielded than C-2 of the corresponding aldoses and aldonates. For example, C-2 for DL- glyceraldehyde hydrate resonates at 75.5 ppm while C-2 for DL- glyceronitrile is found at 63.2 ppm. The same effect is observed when C-2 is a l°-hydroxylic carbon. Similar shielding is observed for the protons of acetylene (177). Cyclization causes a downfield shift in the resonance of the car- bon involved in ring-formation. The C—4 resonance of D-erythrose hy- drate is found at 64.0 ppm while the C—4 resonances for c- and 8- D- erythrofuranoses are observed at 72.9 and 72.4 ppm, respectively. This effect was exploited in determining the ring size of the inter- mediate imidolactones formed during the cyanohydrin reaction (Figure 24). 13C_l3 . l C coupling constants ( J were measured for Direct C,C) several short-chain aldoses and their derivatives and are found in 150 Table 10. 13C Chemical shifts of short-chain carbohydrates and derivatives. Compound Carbon Position, ppma C-1 C-2 C-3 C-4 Formaldehyde, hydrate 83.3 Glycolaldehyde, hydrate 91.2 66.0 DL-Glyceraldehyde, hydrate 91.2 75.5 63.4 DL-Erythrose u-furanose 96.8 72.4 70.6 72.9 B-furanose 102.4 77.7 71.7 72.4 hydrate 90.8 74.9 73.0 64.0 DL-Threose d-furanose 103.4 82.0 76.4 74.3 B-furanose 97.9 77.5 76.2 71.8 hydrate 91.1 74.6 72.2 64.4 Glycolonitrile 120.6 49.2 DL-Glyceronitrile 121.0 63.2 64.4 DL-Erythrono- and 120.9: 63.2 73.2: 62.9: -threono-nitriles 120.9 63.6 73.4 63.0 DL—Glyceric acid 177.2 72.8 64.8 D-Erythronate, sodium 179.7 75.1 74.9 63.5 D-Threonate, sodium 180.2 73.7 74.2 64.5 a All chemical shifts are relative to MeASi as external standard. Tentative assignment. 151 Table 11. As expected (178), the magnitude of direct coupling between C-1 and C-2 generally increases as the s-character of the C-l-C-Z bond increases. JC1,C2 is greater for the aldononitriles (N60 Hz) and a1- donates (m53 Hz) than for the furanoses («44 Hz) and hydrates (N48 Hz). As noted for other aldonates and aldonic acids (Table 3), the values of lJCl C2 for the aldonic acids (”59 Hz) is about 5 Hz larger than 1 JC1,C2 for the aldonates (~34 Hz). Vicinal 130-130 coupling is observed only in the case of threo de- rivatives; the nitrile and aldonic acid salts have 3JC1 C4 values of 4.0 t 1.5 Hz and 3.7 t 0.7 Hz, respectively. This result is surprising as, in the acyclic forms of both erythro and Ehggg_derivatives, a EEEEE relationship should exist between C-1 and C-4, and be somewhat more stable in the erythro isomers. The observation of coupling only in the 32532 forms indicates that the arrangement of the hydroxyl substituents along the coupling pathway is important. This dependence on the ar- rangement of the hydroxyl substituents was observed for several linear compounds having the £122,399 arabino configurations, where 3JC1,C4 was only observed for those compounds having the latter configuration (Table 3, Figure 24). The magnitude of 33 was explained in terms Cl,C4 of preferred solution conformations. Direct 13C—lH coupling constants were determined from [13C1-en- riched compounds and are found in Table 11. Heteronuclear l3C-lH coup- ling constants can be determined from lH-coupled 13C NMR spectra, as 13 . l 13 shown for D-[l- C]threose in Figure 34. JCl H1 for the [1— C]tet— 9 roses varies with configuration at C-1 and with sugar conformation, with values ranging from 172-174 Hz for cyclic forms and from 162-164 Hz for acyclic hydrates. Direct C-2-H-2 coupling in the [2-13C1- 152 Table 11. 13C Coupling constants of enriched carbohydrates and derivatives. Compound CouplinggConstant, Hza C-er-Z C—l—H—l C-2—H-2 C—3—H-3 C-l—H-Z C-l—C-4 Glycolaldehyde, hydrate A 163.9 A b Glyceraldehyde, hydrate A 162.1 143.3 A b Erythrose, hydrate 48.4 164.2 141.5 145.9 b A Threose, hydrate 49.0 162.8 141.5 145.0 4.4 A a-Erythrofuranose 43.3 172.3 150.3 154.0 b A d-Threofuranose 45.9 172.3 152.5 152.8 Nl.9 A B-Erythrofuranose 46.9 172.3 150.3 152.1 N3.4 A B Threofuranose 42.3 173.8 151.8 152.8 4.4 A Glyceronitrile 59.4 A A d Erythrononitrile 60.8 153.0 A d c Threononitrile 60.8 153.1 A d m4.0 Glyceric acid 59.4 A A A Erythronate, sodium 52.8 A A 2.9 c Threonate, sodium 54.2 A A 3.7 3.7 Coupling constants are accurate to within 10.7 Hz. The letter (A) indicates that no experiments were performed to evaluate coupling bet- ween the designated atoms. b Broadened peaks. c No coupling ob- served. Coupling observed, but could not be reliably measured. 153 Figure 34. Determination of heteronuclear l3C-lH coupling by C NMR. The proton-decoupled (A) and coupled (B) 13C NMR spectra of the C-1 re- gion of o- and B-D—[1-13C]threose and D-[1-13C]threose hydrate, showing Cl-Hl couplings for the three forms in aqueous solution. 154 C45 J C'IHI J [ CIHI C1H2 CIH2 110 100 90 Figure 34. 155 enriched tetroses ranges from 150-153 Hz for cyclic forms and 141-143 Hz for the hydrates. Direct C-3-H-3 coupling in the [3-13C1-enriched tetroses ranges from 152-154 Hz for cyclic forms and 145-146 Hz for acyclic hydrates. Two-bond C-l to H-Z coupling appears to be present in the coupled 13C NMR spectra of [1—13C]-enriched compounds (Figure 34), and differ- ences appear to exist between erythro and threo furanoses (Table 11). Although these differences probably reflect the conformational pre- ferences of the various forms, analysis of the 13 NMR spectra will be required to estab1ish the magnitude of 2J more precisely before C1,H2 conformational inferences are made. 2 3 . It is noted that J and J cannot, in contrast to be C,H C,H Jc,u’ distinguished solely on the basis of lH-coupled 13C NMR spectra of the enriched 13C nucleus. Interpretation of the 1H NMR spectra of the 13 . . . 13 1H . [ C]-enriched compounds is required to establish C- coupling pathways with certainty. Differences in 2J are observed between D-erythrose hydrate Cl,HZ and D—threose hydrate. The Ehrgg_isomer has 2JC1,H2 = 4.4 Hz, where- as the erythro isomer gives broadened resonances. Complex carbon- hydrogen geminal and vicinal couplings were also observed in the [1-13C]-enriched three— and four-carbon aldononitriles. Again, these differences probably reflect the conformational preferences of the acyclic forms, but additional data will be required to confirm and interpret them. 156 b. Aldose phosphates i. Solution structure The structures of the aldose and ketose phosphates in aqueous solu- tion are of considerable interest because of their involvement in enzyme-substrate reactions. Swenson and Barker (179) examined sev- eral sugar phosphates by infrared, ultraviolet and circular dichroic spectroscopy to determine the percentage of carbonyl forms present in aqueous solution. Fructose 1-P and glucose 6-P have no, or only trace amounts of, free keto or aldehydo forms in solution. D—Fructose:6—P. D- fructose 1,6-P and DL-glyceraldehyde 3-P have approximate1y lZ-loz 2 free carbonyl in solution, depending on the compound. 1,3-Dihydroxy- 2-pr0panone-P is 55% keto form in solution. Using 31P NMR spectroscopy to examine stereochemical analogs of D- fructose 1,6-P2 (FDP),the methylglycosides of FDP, and specifically deuterated FDP, Gray (180) concluded that four forms of FDP are pre- sent in aqueous solution in the following percentages: e-furanose (D90), a-furanose QulO), keto (<1.7), and hydrated keto (<0.1). Koerner et al. (181) investigated the tautomeric composition of natural abundance FDP and D-fructose 6-P (F6?) in solution by 13C NMR spectroscopy. In addition to the assignment of 13C chemical shifts, integration of signal intensities yielded the following equilibrium compositions: F6P, a-furanose (19 t 2%), B-furanose (81 t 2); FDP, a-furanose (23 i 4%) and B-furanose (77 i 4%). Less than 1.5% keto or hydrated forms was reported in solutions of either fructose phos- phate. Midelfort, Gupta and Rose (182) prepared [UL-13C] F6? and FDP en- 13 zymatically and examined them in H20 and HZO-DMSO mixtures by C NMR. 157 Despite complications arising from l3C-13C coupling, keto forms were observed at N215 ppm downfield fnmnHeQSL.They concluded that F6? and FD? contain 4.1% and 2.070 keto isomer at room temperature, respectively. From 13C line-broadening studies , ring-opening rates for the cit-furanose (8 3‘1) and B-furanose (35 5.1) forms of FD? were measured and related to the kinetic properties of the aldolases. In this study, the solution structures of the aldose phosphates are compared with those of the simple aldoses from which they are de- rived and with those having one less carbon atom. The terminal phos- phate eliminates the possibility ~ VM.MM UM.MM ~.0~ o.m~ 0.00 0.0 mmocmHSWIm n 0 m 05 N.~w N.NoH 0.0 mmocmhsmld . . ml0 mmocfianMIn 0m.mw v0.0m H.0m m.om 0.0 woman»: .mlc mmomunuada 0 0¢.Hm 0.05 m.o¢ 0.H mount»: .mlq mmounuzumlgo eo so eo.sa m.H¢ m.a omumuese .aum messoeampmosawuo 0~.w0 0~.oa 0.5 omumuvzn .mlmchsmvamaooxam mIU owlU MIU NIU HID mAEmmv :oHuwmom conumo 0:9 masca80u .mwcsanoo doumduu mam mmumzmmoza mmomdm mo 0000zm Hmofiamzo 00H .NH manme 159 .mmmmzusmumm :H CSozm m .wsaaanoo o Imam Eoum wcflmwum umansou m mm mummaam .ooa H 00 um mama mum3 mcowumcwspmomv mmmza o .mwmxmmcm mzz ou uoapq uo0m um Auaa: H.OHV mm cm>fiw osu cu wmumznvm meB mcowumaom 0 .vmusmmoa uoc mum: mmumucm uoc mum 000:3 nonficm Hmoflamnu .N= ooqm 00 50003 umuaww m mam N: 000H 00 :upHB awash m Sofia 00H H 0H om nmcweuoumn oum3 mumwsm Hmofieosu m mzam> =m mso um pmuzmmma who: muwficm Hmofismzo mocmcommx v .A.0.u:oov NH mflnms 160 in dilute aqueous solution at pH 2-5, but chemical shifts at 97.9 and 98.4 ppm, typical of hemiacetals, indicate the presence of up to 15% of dimers and/or oligomers in solutions of [1--1 CJerythrose 4—P. Blackmore et al. (183) examined the spontaneous dimerization of ery— throse 4-? in aqueous solution and prOposed a structure for the dimer based on data from mass spectrometry. In concentrated solutions of D-[l—IBC]erythrose, C-l resonances at 98.5, 97.9, 92.5 and 92.1 ppm have been attributed to dimeric or higher order structures (Figures 30C and 31D). Aqueous solutions of the [1—13C1pentose S-phosphates at pH 5.5 contain predominantly a- and B-furanose forms, with a small proportion (1—52) of linear gem—diol (Table 13). In comparison, solutions of Table 13. Structural forms of the pentose S—phosphates in aqueous solutiona. Pentose 5-? Z Composition, t3Zb B a Hydrate Arabinose 5-? 4O 58 2 Ribose 5-? 64 34 =1 Xylose 5-? 45 51 4 Lyxose 5-? 25 70 5 a Determined at 3 t 1°C and pH 5.5 t g.1. b Determined by computer integration of C NMR spectra of [l-1 C -enriched pentose 5—phosphates. Aldehydo forms were not observed. [1-13C1tetroses contain approximately 12% hydrate. Assignment of the chemical shifts to the a- and B-furanose forms of the pentofuranose phosphates was made by analogy to those assigned to the a- and B-methyl furanosides(1l4). 161 Furanose rings having OH-l t£§g§_to OH-Z (Figure 35) are generally more stable than those having the corresponding gig arrangement (99), and the a—lyxo, a-arabino-, and B-ribofuranose phosphates which have this arrangement are the predominant anomers in aqueous solution at pH 5.5 (Table 13). D- lo-Pentofuranose phosphate is an exception to this rule, indicating that destabilization arising from two gi§71,3 inter~ actions in the B-anomer (Figure 35) is equivalent to that arising from a single £3§71,2 interaction in the a-anomer. As observed by Ritchie et al. (114), a useful relationship exists between the chemical shift of C-1 and the anomeric configuration in the furanose ring. As in the methyl furanosides. the resonances for C-1 of anomers of the aldofuranose phosphates having OH-l trans to 0H~2 are typically downfield (2105 ppm) from those with the cis arrangement (=97 ppm) (Table 12, Figure 35). It is well established that the relative positions of hydroxyl groups in pyranoses and furanoses determine the relative stabilities of anomers and various ring forms (99). Although the conformations of the furanose rings in the aldose phosphates are not established, cer- tain relative configurations of hydroxyl groups appear to be preferred leading to the observed anomeric preferences listed in Table 13. Spe- cifically, the pentose phosphates with 0H~2 EEEE§.t° OH-3 generally have a greater proportion of the anomer with OH-l gi§_to OH—Z than do those pentose phosphates with 03-2 gig to 0H-3 (Figure 35). ii. Assignment of chemical shifts (Table 12) 'Chemical shifts for C-l have been discussed above in terms of the various forms present in aqueous solution. Assignments of carbons bearing the phosphate group and adjacent carbons have been made on the 162 100 man muuqfim Hmufifioao .u: 5.0“ 00 oumuau fimHQSOO mfiu mo hufixwdmfioo way ou 05¢ a: m.# .3 n 34.2.33 2. 0000000 0:0 00 mEuou 00000000 050 000300 I lall'"! 1’ Il..l|. I'll! I |ll|t1l\. ‘1 III-.OIII.'llIJ. .u {'4' 9N wd 04;. mm... 0 Dwm aka 0\ m /0 m $3.. 002 m..O. VNO. 0x». 00.. maN-£o UHH I. ANT—HIMfiUV .momso n a on own mwcfiaasoo mic umauo HH< . 0:000:00 mafiamaoo 00:0 0“ uouuo 0:9 0 .500 0.0“ 00 wumuau .mumcwam 0.0 ”N... me my 0215 0 a: Head .0 .momouuou 0:0 0:0 000000000010 0 A 000:0:0000000 #0000000 mzz and 0:0 Hmuzuoauum .00 on: «m 0.. ow 0.. “13-9.3 v.2. 0.3. 0.2.. Er... 0.00 new 0.8 59. ..-o x y m y N01:0 0 m. A0 a mil. nNh. 0N... 32.0.“ 0.8. NS. v.8. can ..-0 m J x M $4.00.. 0 m. xnv m 0 0.»: 00.00 .0 00:... n.~-m00.. \lici II, ..I. II C 163 basis of comparison with standards and on the predictable coupling pattern of phosphorus to carbon (184, 185). C-2 Resonances were as- signed on the basis of l3C-13C coupling to C~1 in [1-13C]—enriched compounds. The [2-13C1-enriched compounds prepared also permit un- equivocal assignment of C-2 and of C-3 chemical shifts by difference (the close proximity of C-2 and C-3 resonances causes the C-3 reson- ance to be obscured in several of the [2-13C]-enriched compounds). As observed by Ritchie et al. (114) for g1ycosides,and for the tetrofura— noses (Table 10), the C-2 and Cv3 chemical shifts for the pentofura- nose 5-phosphates in which OH-Z and OH-3 are Eran§_(arabino, xylo) are downfield from those of the Ei§_compounds (ribo, lyxo) (Table 12 and Figure 35). The relative configuration at C-2 and C-3 does not ap— parently affect the chemical shifts of C—1, C-4 and C-5. Ritchie et al. (114) have observed that substitution of CHZOH for H at C-4 of a tetrofuranoside to produce a pentofuranoside causes a downfield shift of the C-4 resonance of approximately 10 ppm. The C-4 resonance in the tetrofuranoses (Table 10) also shifts downfield approximately 10 ppm when a 4-H is replaced by CHZO? to produce the pentose 5-?, suggesting that phosphorylation at the primary alcohol may not significantly alter electron density at C-4. Alternately, the CHZOP group may alter the through-bond effect on a neighboring group and, at the same time, pro- pagate a through-space effect that cancels the through-bond effect. The nitrile group shields the a-carbon in the aldononitrile phos- phates, producing an upfield shift of approximately 10 ppm from the a- carbon of the aldose phosphates. This effect is expected and is ana- logous to proton shielding observed in the acetylene system by Pople (177). 164 iii. Carbon-phosphorus coupling constants Two—bond coupling between carbon and phoSphorus is not very sen- . n 9 c v 2 Sitive to Changes in structure of the phosphate ester. Two-bond ( JPOC) coupling constants observed in this study (Table 14) are typical of those observed in other phosphate esters, ranging from 4.4 Hz for threose 4-? (pH 4.5) to 5.5 Hz for a-ribose andCi-arabinose 5—phosphates - 2 - . - . . . (pH 5.5). The measured J of 4.8 Hz for s—D-ribose 5-? 15 Similar FCC to the reported ZJPOC value of 4.7 Hz for the B—D—ribose 5-? moieties in UMP and AM? (186) at pH 6.3. Three-bond (3JPOCC) coupling (Table 14) is dihedral-angle depen- dent and can be used to determine molecular orientations in solution. For S-D-ribose 5-? and the B-D-ribose S—P moieties in 5'~UM? and 5'- AMP, BJPOCC = 8.4 Hz (pH 5.5), 8.5 Hz (pH 6.3), and 8.7 Hz (pH 6.3), respectively. The magnitude of these couplings indicates that the pre- ferred position of the phosphate group is gragg to C-4 and gauche to H—5' and H—5" (186). The values of 3JPOCC for both cyclic and acyclic aldose phosphates (Table 14) indicate that in these compounds a EEEEE, arrangement is preferred. It should be noted that BJPOCC couplings for several a-glycosyl l-phosphates indicate that the preferred posi- tion of the phosphate is Egggg to C-2 and gauche to C-1 and to 0-5, but substantial amounts of other rotamers may be present since the maximum values for BJPOCC may be 12 Hz or more when the EEEEE arrange- ment is fixed (187). iv. Carbon—hydrogen coupling constants Carbon-hydrogen coupling constants can provide useful information about carbohydrate structures,as shown by Bock et al. (115), Walker et a1. (59), and Schwarcz and Perlin (120). These couplings are 0.0 0.00H 0.m 01m mdfiuuwcocoxsfi100 0 N.00H N.m 010 mawuuwcosoH>x1ac 0.00H 0.0 010 0000000000000100 0.N0~ «.0 010 00H0000000000000140 0.0 0.0 m.H 010 mafiuuwsocovuzulqn 0.0 0.0 0.0 010 0000000000000000100 0.0 0.0 m.~ 01m mafipuwcoumosam100 0.0 0.000 0.000 0.0 0 0000002010 0 N.00~ 0.~m~ 0.0 0000000010 m10 mmoxzalgo 0 0.000 0.H~H 0.0 0000000010 0 m.00H 0.000 0.0 0000000010 010 0mo~>x100 0 0.000 0.050 0.0 0.0 0.0 0000000010 0.0 0.NOH 0.000 0.0 0.0 0.0 0000000010 .3 010 000000100 ”m 0 0.000 a.sa~ 0.0 0.0 0.0 mmocmhau1m o 0.000 0.NNH 0.0 0.0 0.0 00o0000010 010 000000000100 .H.Nv 00.0 A~.N0 00.~0H ~.m 0.0 0.0 00000»; .010 0000000100 0 0.000 ~.m H.0 0.H 0o000>£ .010 0mouzu>u01qa 0 0.00H N.0 H.0 0.0 000000: .010 000:00H00000Hw140 . . . 0 . m m H 0 0 N 000000»; ml00>z00H0Hoo>Hw N=.Hu ~=.N0 00.00 0000, 000 Wm ”H 5H mm WN Auzv 00000000100000200 mHUm~11 0A02v 00000000 0 0:000:00 010ma :0 00300000 .000000500 0000H0u 000 0000000000 000000 00 000000000 0:000:09 :1OMH 100 m1oma .00 manme 166 .00000000000 00 03000 =0 000 00 00000000 0003 000000000 0 .0o0 H 00 00 0000 0003 00000000000000 000:9 .00>00000 00000000 02 .00>00000 0000000000 .00000000 000 0003 00000 100 0 0 0 . 1 0 0> :0 000 00 000 000 00003 000000000 .0: 0.00 000003 00 00000000 000 000 000000000 00000 0 0+0 00 0 0 000 H 00 00 0: 000m 00 00003 000000 0 000 0: 0000 00 00003 00030 0 0003 00000000 0003 000000000 0000 000 0 .A.u.u0oov 00 00008 167 particularly easy to observe in [13C1—enriched compounds, and some use— ful correlationsof magnitude of coupling constants to Structure emerge l3, 1 L— from this study. The H couplings observed in the [13C1—enriched aldose phosphates and their derivatives are listed in Table 14. l . .. . _ JCl,Hl coupling for the gem—dial triose and tetrose phosphates range from 160 to 164 Hz. Formation of the furanose phosphate ring in- creases the value of lJCl H1 by about 10 Hz. This increase upon cycli- zation has been observed for erythro- and threofuranoses (Table ll) and provides another parameter, in addition to chemical shift, for the identification of linear hydrates in solution by 13C NMR and 1H NMR. Bock et al. (115), Bock and Pedersen (116, 117), and Walker et al. (59) have shown that lJCl H1 in the pyranoses is dependent on the con- figuration at C-l and is useful in assigning anomeric configuration 0f carbohydrates. In these rings, 1%: for an axial H-l is approxi- ,H mately 10 Hz smaller than lJCHfor an equatorial H-l. One-bond C-l to H—l coupling in the pentose S-phosphates and tetroses is also sensitive to the configuration at C-1. The gi§71,2 anomers of threofuranose and arabino-, xylo-, and lyxofuranose phosphates have larger C-l to H-l coupling constants than the respective £g§3§71,2 anomers (Table 14 and Figure 35), although differences between anomers are smaller (AlJCl Hl = 1.5 - 2.6 Hz) than those observed for the pyranoses. lJCl Hl for ery- throfuranose and ribofuranose phosphate is not as sensitive to config- uration at C—1. Differences in lJCl H1 for the furanose ring probably reflect conformational preferences (188) which must be determined be- fore a full interpretation of l3C-1H coupling with respect to furanose configuration can be made. 1 - - -9 JC2,H2 couplings are typically 20 -5 Hz less than JCl,Hl 168 couplings for both the linear and furanose forms of phosphorylated (Table 14) and simple aldoses (Table 11). Examinations of lJCH in . . . . , l the pyranoses have snown Similar differences between J 1 Jc2,az Cl,Hl and (116, 117). 7 Several geminal "J coupling constants for the [1-13C1aldose C1,HZ phosphates were observable (Table 14). Threose 4-P and threose hydrate (Table 11) show ZJCCH couplings of 3.5 and 4.4 Hz, reSpectively, where— as erythrose 4-P and erythrose hydrate do not exhibit coupling. Proton- 13 13 13 coupled C NMR spectra of [1— C]- and [2- Clpentofuranose S-phosphates vary in complexity with furanose configuration. For example, geminal 13 l . . . , 13 and longer range C- H coupling is not apparent in 2- and 3-[1- C]- arabinofuranose S—P, whereas a- and 3-[1-13C]lyxofuranose S-P show com— plex coupling patterns. Analysis by 1H NMR will be required to identi- . . l3 1 l 1 . fy speCific C- H and H- H couplings and relate the value of these couplings to furanose conformation and configuration. C. Preparation of [2H1-Enriched Aldoses and Their Derivatives l. Hydrogenolysis with 2H2 For the incorporation of 2H into the aldoses, cyanide condensation and hydrogenolysis are carried out in 2H20 instead of H20 to avoid exchange of 1H for 2H on the catalytic surface. This exchange decreases the incorporation of 2H at H-l. Under the conditions used for hydrogenolysis, lH-ZH exchangedoesrmn;occur at other positions, as determined by 1H NMR. In several cases, nitrile (”15%) remained in the reduction mixtures after 10 h of hydrogenolysis. Complete conver- sion to products was achieved by adding new catalyst and continuing the reduction. 169 The methods for the preparation of aldononitriles in 2H20 and their hydrogenolysis with 2H2 are the same as those for the preparation of the undeuterated compounds. Yields are also comparable. The use of [13C, 2HJ-enriched compounds facilitates the observa- tion of the deuterated carbon (Figure 36). The 13C NMR spectrum of D- "\ [l-lJC1threose in aqueous solution shows the presence of three major tautomeric forms, namely, the G- and B-furanoses, and a linear gem-diol (hydrate) (Figure 36A). The proton-decoupled 13C NMR spectrum of D- l3C,2H]threose (Figure 3bB) shows four lines for each tautomeric [l- - . . - 13 2 . . form, three ariSing from C- H coupling, and one from the reSidual protonated carbon. The percent isotOpic incorporation is not reflected in peak areas,since nuclear Overhauser enhancement is smaller for deu- terated than for protonated methine carbons (189). For example, the _13 2 1H NMR spectrum of the same preparation of D—[l C, H]threose shows no resonance for H-l, indicating deuterium enrichment of at least 97 1 percent, whereas the proton—decoupled 3C NMR spectrum (Figure 368) gives the appearance of a significant prOportion of 1H at H-1. 2. NMR parameters a. 13C NMR parameters . 2 , l . . - 13 Substitution of H for H permits the aSSignment of C resonances of the directly-bound and nearby carbons due to decreased nuclear Over- hauser effects and characteristic isotope shifts. In addition,13C-2H coupling can be measured. The effect of 2H substitution for 1H on the 13C chemical shift of the derivatized carbon is shown in Table 15 for several carbohydrates. The values observed are similar to those observed by Gorin (79) and Gorin and Mazurek (118). The use of [13P 2H]-enriched compounds permits an easier evaluation of the V) 170 Figure 36. Incorporation of 2H into a [13C1-enriched carbohydrate. (A) The 15.08 MHz proton—decoupled 13C NMR spectrum of the enriched re- gion for D-[l-13C1threose. The three predominant tautomeric forms in aqueous solution are a- and B—furanose (103.4 and 97.9 ppm, respec- tively) and acyclic hydrate (91.1 ppm). (B) The 15.08 MHz roton- decoupled 13C NMR spectrum of the enriched region of D-[l-l C,2H]- threose, showing the splitting of C-1 of each form by the directly— bound deuteron. Isotope shift is shown for each Species as the dif- ference in the positions between the protonated C-1 and the center of gravity of the triplet arising from the deuterated C-l. 171 Cwlct M P Cfilh. JW/\”vfim\\m/A CD 108 104 100 96 92 88 Figure 36. 172 magnitudes of 1J13 2 and the isotope shift, as shown in Figure 36. A C, H directly-bound deuterium nucleus, with a spin of 1, will Split a carbon . . . l , l3 2 . resonance into a triplet with J13 7 = 24 Hz. The C- H isotOpe C,°H shift can be estimated from the difference in resonance position bet- ween the protonated carbon and the center of gravity of the deuterated carbon triplet. The deuterated carbon is upfield (NS Hz) from the pro- tonated carbon. 1 J 13C,2H . . . . . l bonydrates and derivatives. The data indicate that J In Table 15, coupling constants are listed for several car- is larger l3Cl,2H1 when OH—l and OH-Z are cis in the furanose ring, and that this coupling is larger in the ring forms than in the acyclic hydrates. The same re- . . 1 lationships were observed for J (Tables 11 and 14, Figure 35). 13C 1H Colli et al. (190) have shown for several non-carbohydrate compounds . l .1 .g . that the ratio J13 1 . J13 7 is very close to the value predicted C, H C,”H from the magnetogyric ratios for 1H and 2H. The value of AJ indicates the extent of variation between observed and predicted values of lJ13 2 ,and is zero for perfect agreement. As shown in Table 15, C, H values for AJ are within the error of the determinations. b. 1H NMR parameters Replacement of 1H with 2H often simplifies 1H NMR spectra, facili- tating the assignment of chemical shifts and coupling constants. For example, whereas the 180 MHz 1H NMR spectrum of D-erythrose is essen— tially first-order, that of D-threose is complex (Figure 37A). The complexity was eliminated in the spectrum of DL-[3-2H1threose and it was shown that,in D-threose,the 1H NMR spectrum is complicated by the near magnetic equivalence of H-3a and H-Aa which perturbs the reson- ances of these nuclei and produces a complex multiplet for H-A'a. 173 7 Table 15. 13C Chemic l CShifts, 13C- 1H and 13C- -H coupling constants for several [1,H]-enriched carbohydrates and derivatives. Compound Carbon Chemical Shift, ppma l 1 l3 1 13 2 ‘ c A d JC 1H JC,2H C- H C-‘H uV -J (H2) (H2) (H2) (15.3) a-D-[l-IBC,2H]- Threoie 172.3 26.0 103.5 103.2 5.9 2.9 B-D-[l- 3C, 2H]- Thregse2173.8 26.4 98.0 97.7 5.1 1.8 D-[l-12H1- Thre0f3, hydrate 162.8 24.9 91.0 90.8 4.8 0.6 a-D-[l- C, 2H]- Erythrose 172.3 26.8 96.9 96.5 5.1 -2.3 8-0-[1-13C,2H]- Erythrose 172.3 26.0 102.5 102.1 5.9 2.9 D-[1—13c, 2H]- Erythrose, h drage 164.2 25.3 90.8 90.5 5.1 -O.6 Methyl a-D-[2-3H1- Ribofuranosidg 2 22.5 72.5 72.1 5.1 Methyl 8-D-[2- H]- Ribofuranoside 23.8 75.7 75.3 5.1 a-D- [2-13C, 2H]- Arabinopyranose 22.4 73.4 72.9 6.6 3-D-[2-13C,2H]- Arabinopyranose 22.0 70.0 69.6 5.9 d-D-[l- 13C 2H]- Ribose S-Phosphate 173.6 26.8 97.6 97.3 4.4 -l.O B-D-[l-13C, Zn]— . Ribose 5-Phosphate 173.0 26.0 102.3 102.0 4.4 3.6 DL-[1-13C,2H]- Glyceraldehyde 3—Phos- 159.8 24.9 90.8 90.5 5.1 -2.4 __phate Chemical shifts are given relative to external Me Si and are accurate at $0.1 ppm. Carbon spectra were obtained with broad-band proton de- coupling at 30°C. Coupling constants are accurate to 10.7 Hz. c Av V(1H ) - v 2 , where v (1H and v 2 are equal to the resonance frequen- cies of the protonated (and deutérafed carbons, respectively. Av is positive since the observed isorgpeJ shifts upon deuterption are upfield from the protonated homologue. 1J13C 1H YlH J13 2 C, H 2H’ 1H and Y2 are the magnetogyric ratios for 1H and 2H, respec- where y tively. 174 Figure 37. 180.04 MHz 1H NMR spectra of the H-2 to H-4 regions of D- threose and DL-[3-2H1threose. (A) The 180.04 MHz 1H NMR spectrum of the H2—H4 region of D-threose. (B) The 180.04 MHz ]-H NMR spectrum of the same region of DL-[3-2H1- threose. Deuteration at C-3 simplifies the spectrum so that assignment of resonances can be made as shown. The upfield half of the doublet from H—28 (due to coupling to H-l) is observed, with the other half hid- den by the H—20 doublet. The magnitude of the coupling is confirmed by observation of H-lB (not shown). The chemical shifts of H-30 and H-38, determined by computer simulation, are 4.20 ppm and 4.30 ppm, respec- tively. Lines between the resonances due to H-4'8 appear to arise from H-4 and H-4' of the acyclic hydrate. H-4‘ was arbitrarily designated as the more shielded H-4 of each form. 176 Substituting 2H for 1H at H-3 in threose removes three vicinal 1H- 1H couplings and the resonances due to H-3 from the spectrum, greatly simplifying assignment and analysis. D-Threose exists primarily as three tautomeric forms in aqueous solution in the ratio a—furanose: S-furanosezhydrate of 4.2:3.l:1 (Figure 36A). Although H—2 can be iden- tified by selective homonuclear decoupling of H-l or by [IBCI-enrichment at C-2, neither technique permits the unequivocal assignment of H—3, H-4 and H-4' for each tautomer. Figure 378 shows the H2-H4 region of the 180 MHz 1H NMR spectrum of DL-[3-2H]threose. The H—2, H-4 and H—4' resonances are easily identified for the furanose forms on the basis of their proportions in solution. In addition, the spectrum is essentially first-order, permitting direct determination of several geminal and vicinal lH—lH coupling constants. The H-2 resonance of the hydrate is at 3.46 ppm, but resonances due to H-3, H—4 and H-4' for this form are not readily assigned. However, from the 1H NMR spectrum of the 3‘2H derivative (Figure 37B), it is clear that H—4 and H-4' for the linear hydrate lie in the same region as H-4' for the S-furanose, permitting the position of these resonances to be estimated by computer simulation. Chemical shifts for H—3 for the cyclic forms of D-threose were estimated by comparison of the normal and deuterated compounds and refined by computer simulation. Apparent and intrinsic proton chemical shifts and H-lH coupling constants for the tetroses are given in Tables 16 and 17. In both tetroses, H-l of the hydrate is more shielded than H-l for the furanoses, while H-Z of the hydrate is the most shielded nucleus. Heteronuclear 2H-lH coupling has been observed in both high reso— lution 1H and 2H spectra (190, 191). However, the 1H NMR Spectra of DL-[3-2H] threose (Figure 378) and DL-[3-2H1erythrose at 180 MHz 177 do not exhibit 2H-lH couplings. Only vicinal or long-range lH-ZH coup— ling pathways would be expected in these compounds. Since lH-ZH coup- ling constants are about 15% (1/6.5144) of their lH-lH analogs, triplets with 3J1 2 values of 0.15-0.77 Hz would be difficult to re- solve. As discusgeduby Mantsch et al. (192), 1H spectra often exhibit an average lH—2H coupling because the remaining lH-lH couplings are con- siderably larger than the lH-2H couplings or the 2H-induced isotope shifts in the H spectrum, and the 1H spectrum is consequently decep— tively simple (193). Line broadening («0.7 Hz) due to pseudorotation of the furanose ring also hinders observation of the smaller lH-ZH couplings. Specific deuteration has been useful in establishing long-range H—lH coupling in the furanose ring. For example, deuteration of D- erythrose at either H—l or H-3 simplifies the H-3 or H-l multiplets, respectively, for the 8-anomer, indicating that a small (m0.6 Hz) coup- ling exists between these nuclei. The a-anomer shows no such coupling. Interestingly, methyl a—D-arabinofuranoside has = 0.5 Hz, JH1,H3 whereas the 8-anomer shows no coupling (102). Heteronuclear spin-spin coupling between 13C and 1H is valuable in examining conformations of carbohydrates in solution. Generally, two- and three-bond l3C-H coupling constants are difficult to obtain from 1 l3 l3 . H-coupled C NMR spectra, even when [ C]-enriched compounds are used. These studies are greatly facilitated at high fields (67.89 MHz for carbon) as demonstrated recently (124), but, even in this case, deuterated analogues and heteronuclear selective lH decoupling were used to confirm assignments. An alternative approach has been to ana- lyze [13C1-enriched compounds by 1H NMR (120, 121, 194-198). In cases 178 where 1H NMR spectra are complex, the synthesis of compounds with [2H]- and/or [13CI-enrichment may aid interpretation. For example, Figure 38A shows the 180 MHz 1H NMR spectrum of the H2 to H5 region of methyl B-D- -13C,2H]-enriched derivative ribofuranoside. The spectrum of the [2 (Figure 388) is simplified by the loss of a quartet at 4.02 ppm due to H-2. The multiplet centered at 4.14 ppm is altered. In Figure 38A. JH3,H4 and 3JH2,H3 can be assigned. Inspection of the coupling pat- terns for H-l (not shown) and H-2 confirms the assignment of the multi- plet at 4.14 ppm to H-3. In Figure 388, the H-3 multiplet contains 3 JH3 H4 and a new coupling, 2 = 1.6 Hz. Loss of the H-2 multi- Jc2,H3 plet in this spectrum permits H—4 to be assigned and 3HH4,H5 and 3JH4,H5' to be evaluated from the H-4 multiplet. Note that H—4 is slightly broadened in the [13C]-enriched compound, suggesting a small three-bond coupling of this nucleus to C-2. This broadening probably does not arise from lH-ZH coupling (see above). Resonancesdue to H-5 and H-5' are quartets centered at 3.79 and 3.59 ppm, respectively. In 13 [2- C,2H]methyl a—D-ribofuranoside, is small,producing a J(22,113 broadening of the H-3 doublet. . . . 1 . .- 1 l . The apparent and intrinSic H chemical shifts and H- H coupling constants for methyl ribo and arabino furanosides determined from the experimental and computer-simulated data, respectively, are listed in Tables 16 and 17. Although in several instances the use of [2H]- enriched compounds was not required to make these assignments, there is no doubt of the value of multiply-enriched derivatives for use in more complex instances. 179 Figure 38. 180.04 MHz 1H NMR spectra of the H-2 to H-S regions of methyl B-D-ribofuranoside and methyl B-D-[2-13C,2H]ribo- furanoside. (A) The 180.04 MHz 1H NMR spectrum of the H-Z—H-S region of methyl B-D- ribofuranoside, showing the assignment of multiplets. H-S' was arbi- trarily designated as the more shielded H-S. (B) The 180.04 MHz 1H NMR spectrum of the same region of methyl B-D-[2-13C,2H]ribofuranoside, showing the loss of the H-2 multiplet and coupling of H-3 to 13C-2. The H-4 multiplet is broadened while H-5 and H-5' are unchanged. Re- sidual [2-13C,2H] compound would produce two H-2 multiplets Split by lch H2 = 152 Hz (Tables 11 and 14). One of these multiplets would appear at approximately 3.59 ppm in this spectrum. It is not observed. 180 Figure 38. Table 16. Apparent and intrinsic 181 and some methyl pentofuranosides in "H2 0. 1 . .- H chemical shifts for the tetroses Chemical Shifts, ppma Compound H1 H2 H3 4 H4' H5 H5' CH3 a-D—Erythrose 5.270 4.10 4.28 4.03 3.92 (4.02) S-D—Erythrose 5 25 4.02 4.39 4.20 3.79 D-Erythrose, 5.09 3.54 3.79 3.64 hydrate a—DL-Threose 5.24 4 05 4.20 3.95 (4.20) B-DL-Threose 5.40 4.04 4.18 3.65 (4.30) DL-Threose 5 02 3.46 3.67C 3.63C hydrate Methyl a-D-Ribo- 4 99 4.11 4.02 4.09 3.74 3.64 3.43 furanoside (4.03) (3.73) (3.66) Methyl B-D-Ribo— 4.88 4 02 4.14 3.99 3.79 3.59 3.38 furanoside (4.00) (3.78) Methyl a-D-Ara- 4 91 4.04 3.92 4.02 3.81 3.68 3.40 binofuranoside (3.93) (3.80) (3.69) Methyl B-D-Ara- 4.89 4 13 4.00 3.88 3.76 3.60 3.41 binofuranoside a Chemical shifts are given relative to internal sodium silyl)-l prOpanesulfonate and are accurate to £0.01 ppm. (3.61) 3-(trimethyl- Spectra of reducing sugars and g1ycosides were taken at pH 6.5 and 8.0, respec- tively. Intrinsic chemical shifts determined by computer simulation are given in parentheses when they are significantly different from experimental values. Assignment aided by 360 MHz Biochemical Magnetic Resonance Laboratory, Department of Chemistry, C Values accurate to $0.02 ppm. Purdue University. Assignments of H4, H4' and H5, H5' are arbitrary. 1H NMR spectrum obtained at the Purdue 182 Apparent and intrinsic geminal and vicinal H-lH coupling Table 17. constants for the tetroses and some methyl pentofuranosides in 2H 0. Compound Coupling Constant, Hza 1,2 2,3 3,4 3,4' 4,4' 4,5 4,5' 5.5' b b a-D—Erythrose 4.7 5.0 5.0 3.1 -10.0 (4.7) (5.2) (5.1) (3.0) (-10 1) B-D-Erythrose 3.4 4.8 5.0 3.4 - 9.7 (3.4) (4.8) (4.9) (3.5) (- 9.7) D—Erythrose, 4.0 6.6 7.6 -12.ib hydrate a-DL-Threose 1.2 1.8 b b -10.1 (1.2) (1.8) (5.6) (2.6) (*10.l) B-DL-Threose 4.2 4.1 - 9.6 (4.0) (4.0) (5.3) (3.6) (- 9.6) DLvThreose, 6.2 2.7 hydrate Methyl a-D—Ribo- 4.2 6.2 3.1 3.4 4.4 -12.3 furanoside (4.3) (6.2) (3.3) (3.1) (4.8) (-12.4) Methyl S-D-Ribo- 1.2 4.7 6.8 3.7 6.4 -12.3 furanoside (1.2) (4.6) (6.9) (3.1) (6.6) (-12.2) Methyl a-D-Ara- 1.7 3.3 5.8 3.4 5.6 -12.2 binofuranoside (1.7) (3.3) (5.9) (3.1) (6.1) (-12.2) Methyl S-D-Ara- 4.5 7.9 6.7 3.4 6.7 -12.1 binofuranoside (4.6) (8.0) (7.1) (3.2) (7.4) (-12.0) a . P Coupling constants are accurate to $0.15 Hz. theses are intrinsic coupling constants (t0.2 Hz) determined by com- uter simulation. Values are accurate to within i0.3 Hz. Assignments of H4, H4' and H5, H5' are arbitrary. Values found in paren- 183 D. Enzymatic Conversions Using [13C1-Enriched Aldoses and Aldose Phosphates . . . . l3 . . . . The prinCipal interests in [ CJ-enriched compounds lie in their value as useful derivatives to study carbohydrate structure and be- 13. . . . C NMR prooes to examine enzyme«substrate inter- l3C havior in solution, as actions (199), and as tracers to follow enzymatic conversions by NMR spectroscopy (200). In this regard, we have prepared several [13C]-enriched carbohydrates enzymatically from (13CJ-enriched sub- strates synthesized according to methods described in this report pri- marily to demonstrate the biological activity of these substrates and to emphasize the versatility of combining chemical and biochemical syn— thetic routes for the preparation of [13C]-1abeled carbohydrates. 1. D-[2-13C]Ribose 5-P to D-[2-13C]ribulose 1,5-P2 The 130 NMR spectrum of D-[2-13C]ribose 5-P has strong resonances at 71.9 and 76.5 ppm due to C—2 of the a— and B-furanose forms, res- pectively (Figure 39A). Addition of D-ribose 5-P isomerase causes the appearance of a resonance downfield (213.7 ppm) (Figure 398) which is characteristic of the free keto form of D-[2-13C]ribulose 5-P. The equilibrium favors the aldose phosphate (72%) at 36°C as previously observed by Axelrod and Jang (201). Addition of D-ribulose-S-P kinase and Mg2+-ATP converts the downfield singlet into a doublet arising from carbon-phosphorus coupling and shifts the equilibrium toward the product,I}{2-13C]ribulose 1,5-P2. The 13C NMR spectrum of purified D- [2-13C]ribulose 1,5-P2 (Figure 39C) at pH 7.6 shows doublets centered , 3 _ a 3 = at 211.7 ppm (88%, JPOCC 7.3 Hz) and 97.6 ppm (12%, JPOCC 6.6 Hz), indicating that aqueous solutions of D-ribulose 1,5-P2 at pH 7.6 con- tain 88% keto and 12% hydrated forms. This result compares favorably 184 Figure 39. The enzymatic conversion of D-[2-13C]ribose 5—P to D- [2-13C]ribulose 1,5-P2 as followed by 13c NMR. Peaks identified by "X" are unidentified com onents. Only resonances of the enriched nuclei are shown. (A) The l C NMR spectrum of D- [2-13C]ribose 5—P at 36°C, showing a- and B-furanose forms. (B) The addition of phosphoriboisomerase to A produces a downfield resonance (k) originating from the keto form of D-[2-l C]ribulose S-P. (C The addition of phosphoribulokinase and Mg2+-ATP to B produces D-[2- 3C]- ribulose 1,5-P2, whose 3C NMR spectrum is shown after purification. The doublet at approximately 210 ppm originates from the keto form, whereas the doublet at approximately 95 ppm originates from the keto- hydrate form. Splitting of these resonances is caused by P C coupling. Spectra were obtained at 36 i 1°C with a sweep width of 3000 Hz and a filter width of 6000 Hz. 185 3% cm _ . . Om. ON N 1.5. + omoc_xo_:atocamocd+ x x; mmEmEOm_o£co:amozd+ x3; an 32E 89mins. Figure 39. 186 with a determination made by infrared spectroscopy (202). D-[2-13C]- Ribulose 1,5-P2 provides a standard for the chemical shift of a keto hydrate and may serve as a useful probe of the active site of D- ribulose-l,5-P2 carboxylase/oxygenase. 2. DL—[l-13C]G1ycera1dehyde 3-P to L-[3,4-l3C]sorbose 1,6-P2 Incubating DL-[1-13C]glyceraldehyde 3—P with triose phosphate iso- merase produces a mixture of L-[l-l3C1g1yceraldehyde 3-P (91.3 ppm) and [3-13C]dihydroxyacetone-P (66.7 and 65.6 ppm) (Figure 40A). The two chemical shifts for [3-13C]dihydroxyacetone-P arise from the keto (65%) and hydrated keto (35%) forms,in agreement with earlier estima- tions based on 1H NMR spectroscopy by Gray and Barker (202). The addi- tion of D-fructose 1,6-P2 aldolase to this mixture causes the rapid formation of both L-[3,413C]sorbose 1,6-P2 (77.9 and 77.5 ppm) and D-[3, 4-13C]fructose 1,6-P2 (76.7 and 76.2 ppm) (Figure 408). After incuba- tion for an additional 4 h and removal of protein, four resonances are observed at 77.8, 77.2, 77.0, and 76.5 ppm (Figure 40C) due to the major equilibrium product, L-[3,4-13C]sorbose 1,6-P2. The chemical shifts for C-3 and C-4 of sorbose 1,6-P2 are expected to be similar to those of fructose 1,6-P2 reported by Koerner et al. (181). At 15.08 MHz, the difference in frequency between C-3 and C-4 is smal- ler (approximately 30 Hz) than the expected l3C-13C coupling constant (40-50 Hz) and a complex spectrum is observed from which the coupling constant cannot be estimated readily (Figure 40C). Using computer simulation, we estimated JC3 C4 to be 47.7 r 1.0 Hz. It should be noted that this parameter can only be measured in doubly-enriched com- pounds. In singly-enriched compounds, the resonance due to the en- riched carbon will obscure the resonance due to the unenriched adjacent 187 Figure 40. The enzymatic conversion of DL-[l-13C]glycera1dehyde 3-P to L—[3,413C]sorbose 1,6-P2 as followed by 13C NMR. Peaks designated by "X" are unidentified components. Only the re- sonances of enriched nuclei are shown. (A) The 13C NMR spectrum of DL-[1-13C]g1yceraldehyde 3—P hydrate (a) after the addition of triose- phosphate isomerase, producing resonances due to C-3 of the keto (b) and keto-hydrate (c) forms of [3-13C]dihydroxyacetone-P. (B) The ad- dition of D-fructose-l,6-P2 aldolase tx> A causes the appearance after 10 min of resonances due to C-3 and C-4 of D—[3,4-13C]fructose 1,6— P2 (3), with a smaller amount of L-[3,4—13C]sorbose 1,6-P2 (d). (C) The same reaction mixture as analyzed by 13C NMR after 4 b shows little unreacted [1-13C]g1yceraldehyde 3-P and [3-13C]dihydroxy- acetone-P and the major product, L-[3,4-13C]sorbose 1,6-P2. Spectra were obtained at 34 i 1°C with a sweep width of 3000 Hz and a filter width of 2400 Hz. 188 ON 0? Om 41 J1 \vl‘ {{{li‘hllj‘lv‘ t: S 82626 2230:330333... + «5:. o: 3292: o.ofiuogau3:o.o:i+ {(34%th7 332:3. 2233.332; +a.m 823.2685 89-3.3... a SEQ 0%. ON. OS 00. 00. < — q — Figure 40. 189 carbon. The fundamental resonances of coupled carbons must be at least 100 Hz apart to permit easy estimation of coupling in singly enriched compounds. 3. D-[2vl3C]Glucose to D-[2-13C1fructose 1,6-P2 The 13C NMR spectrum of D-[2-13C]glucose, prepared as shown in Figure 29, has two strong resonances at 73.0 and 75.7 ppm due to C-2 of the a- and B-pyranose forms, respectively (Figure 41A). Addition of hexokinase, myokinase, phosphoglucoisomerase and Mg2+-ATP causes the appearance of two new resonances at 106.0 and 103.1 ppm (Figure 418) due to C-2 of n- (13%) and B-furanose (87%) forms of D-[2-13C1fructose 6—P, respectively. Resonances at 73.1 and 75.8 ppm are the C—2 re- sonances of a- and S—D-[2—13CIglucose 6—P, respectively. Addition of phosphofructokinase and Mg2+-ATP converts the 6-phosphates to D-[2-13C1- fructose 1,6-P2 (Figure 41C). Doublets centered at 105.8 and 102.1 ppm are the C-2 resonances of d- (15%) and B-furanose (85%) forms of FDP, respectively. Splittings of 8.8 and 8.1 Hz for the c- and B-anomers, respectively, are caused by three-bond l3C-31P coupling (:3ch Pl)’ Ad- dition of aldolase and triosephosphate isomerase causes the formation of D-[2,5-13C]fructose 1,6-P2 (Figure 41D) which has essentially 45 atom % 13C at C-2 and C-5 based on the summation of the atom percent 13C isotope in four distinct [13CJ-enriched species. Doublets centered at 82.9 and 80.8 ppm are the C-5 resonances of the d- and 8-furanose forms of FDP. Splittings of 8.8 and 8.1Hz for the'oz- and B-anomers, res- pectively,.arecaused by three-bond l3C-31P coupling (3JC5,P6)' 13C NMR spectra in Figure 41 A and B were obtained at 34°C, where- as spectra in Figure 41 C and D were obtained at 18°C to minimize line- broadening caused by anomerization of the diphosphate (182). It is 190 Figure 41. The enzymatic conversion of D-[2-13C] lucose to D-[2—13CJ- fructose 1,6-P2 (FDP) as followed by 3C NMR. 13C NMR 3 ectra show only the enriched nuclei. (A) The 13C NMR spectrum of D-[2--l C]g1ucose, showing a- and B-pyranose forms. (8) Addition of hexokinase, myokinase, phosphoglucoisomerase and M 2+-ATP to A produces a mixture of D-[2-13C]g1ucose 6-P (G6P) and D--[2-l C]fructose 6-P (F6P). Resonances between 73-76 ppm and between 106-103 ppm are the a- and B-forms of G6P and F6P, respectively. (C) Addition of phosphofructo- kinase to B produces a set of doublets originating from the a— (105.8 ppm) and B-furanose (102.1 ppm forms of D-[2-13C]FDP. Splitting of these resonances is caused by 3C-31P coupling. (D) Addition of aldo- lase and triosephosphate isomerase to C produces D-[2,5—13C]FDP. Re- sonances between 80—83lppm are C-5 of the n— and B-furanose forms, with splitting due to C-31P coupling. 191 m Q-[2-‘3CJGlucou A in L L M _~_ \M VV\.VHA_ '3 13 +hoxokinou, myokinau, “I i, 8 2+ - ’ PGI, M9 NP .. «my mm a. 1 "1, L; W VWW I,“ +PFK,Mg”’-AIP c 1! l C. 1 \muwu»~$u . L . l L - l . 1 . 1 - 1 . 1* - 1 i _L 190 170 150 130 110 90 7O 50 30 10 ppm +aldolase, TPI CS D C2 H h A .1 11 1 h .1 a ,J '1 M 7 L‘l 1 WJJ M 1 " win 3 9. .u ‘JMA M}; ‘r'.’ Jwfiu‘Mr-VMJ Al anm‘r W )1 W4 'N'h! Jm‘fffift'W'WvK 1 4 . 1 i 1 . 1 i 1 1 . 120 110 100 90 80 70 610 99'“ Figure 41. 192 interesting that line-widths of the anomeric carbons of D-[2-13C]fruc- tose 6-P and the D-[1-13c]pentose 5-phosphates are also temperature— dependent, with high temperatures (40°-50°C) producing line-widths greater than 5 Hz. Rates of anomerization as well as activation ener- gies for ring—Opening can be determined from line—width analysis of these [13CJ-enriched derivatives since line—widths can be related to rates of chemical exchange (2, 182). 13 4. Action of glycerol kinase on DL-[2- C]glyceraldehyde DL-[2—l3C]Glycera1dehyde (Figure 42A) was incubated with glycerol kinase and Mg2+eATP at pH 7.5 for 2.5 h. The 13C NMR spectrum of the products is shown in Figure 428. Approximately 45 percent of the gly- ceraldehyde CSC2== 75.5 ppm) is converted to the 3-phosphate whose C-2 resonance at 74.9 ppm hssplit by phosphorus (33 = 6.6 Hz). The C2,P3 product 3-phosphate has been shown to have the L—configuration (203, 204). In proof, the unreacted glyceraldehyde was separated from the 3-phosphate by chromatography on DEAE—Sephadex A—25 in the acetate form and incubated with aldolase and 1,3-dihydroxy-2-propanone phos- phate. D-[5-13CIFructose l-P (91%) and L-[5—13C]sorbose l-P (9%) were produced, indicating that 91% of the unreacted glyceraldehyde is the 13 D-isomer. Rabbit muscle aldolase acts on DL-[2- C]glyceraldehyde to produce an equimolar mixture of D-[5-13C1fructose l-P and L-[5-13C]- sorbose l—P. 2 1 By the use of glycerol kinase and Mg +-ATP, [ 3C]-enriched L- glyceraldehyde 3-P can be prepared from [13C1-enriched DL-glyceralde- hyde. A convenient preparation of [13C1-enriched D-glyceraldehyde 3-P, however, is not yet available. but the enzyme, triokinase (205),has been reported to convert D-glyceraldehyde to D—glyceraldehyde 3-P in 193 Figure 42. The 13C NMR spectra of DL-[2-13C]glycera1dehyde (A), 35 mM in water, and the mixture (8) produced by treatment with 1.5 molar equivalents of Mg2+—ATP and glycerol kinase at pH 7.5 for 3 h at 36°C. Approximately 45% of the starting glyceraldehyde is conver ed to the L—3 phosphate, and generates a doublet at 74.9 ppm having J13 31p = 6.6 Hz. The [2—13C]glyceraldehyde remaining after phosphoryla ion was found to be 91% D-enantiomer. Peak (C is an unidentified con— taminant. The resonance at 91 pm is [l-1 C]glyceraldehyde remaining from the preparation of DL-[2-l C]glyceraldehyde. 194 9() 7C) ppm Figure 42. 195 + 2 -ATP. the presence of Mg E. Miscellaneous Applications of (13C1-Enriched Carbohydrates 1. Detection of aldehydo forms in solutions of aldoses The equilibrium solution of D-glucose contains only 0.0026% alde- hydo form, as determined by polarography (206). Aldehydo forms have 1 also been detected by H NMR where the aldehydo proton appears as a 136 mm of [1-13C]- weak Signal at very low fields (207-209). enriched aldoses provides another tool for the detection of aldehydo forms in solution. In this regard, D-[1-13C]erythrose (0.25 M) was dissolved in 80% v/v dioxane-water and analyzed by 13C NMR at 20°C. The specrrum of the anomeric region is shown in Figure 43A. Several forms of D— erythrose are observed: B-furanose (75 t 3%), a-furanose (25 i 3%), hydrate (0.7% t 3%) and dimers and/or oligomers (1.5% i 3%). A signi- ficant decrease in the proportion of hydrated form is observed in 80% dioxane-HZO in comparison to the proportion found in aqueous solution. In aqueous solution, the hydrate comprises 12 t 3% of the solution, with the remaining forms contributed by 8-(63 : 3%) and d-furanoses (25 i 3%) (Figure 31A). Decreasing the concentration of water from 55.5?1toll1M causes an increase in the proportion of dimers and/or oligomers and a decrease in the amount of hydrate. Changes in the proportions of the tautomeric forms in solution may be important when considering the forms of carbohydrates in the living cell, where the concentration of water is not 55.5 M. 1 13 The same solution was examined by H—gated-decoupled C NMR (Figure 438). In this spectrum, aldehydo form was observed at 203.7 196 Figure 43. The 13C NMR spectrum of D-[l-13C]erythrose in 80% dioxanee water . 13C NMR Spectra show only the enriched nuclei. (A) lH-Decoupled 13C NMR spectrum of the anomeric region of D-[l-13CJerythrose in 80% dioxane-water at 20°C and 0.25 M. The a- and B-furanose forms are present with small proportions of hydrate (h) and dimers and/or oligo- mers (a, b, c, d). (B) lH-Gated decoupled 13C NMR spectrum of A show- ing the aldehydo and anomeric regions. C—l chemical shift of the aldehydo form is found at 203.7 ppm with lJC1,H1 . 179.1 2 0.8 Hz. lJCl H1 (B-furanose) = 170.7 1 0.8 Hz and lJCl,Hl (o-furanose) = 174.0 i 0.8 Hz. 197 B 1 l. } \PCLHI B aldehydo CL ’ HCLH! x4 H 3“ W ”a 1 l . l . l J _l 1 200 180 . 160 140 PP"1 B .A l !L L [_ 1 l l L L 105 102 99 96 93 90 37 PP“! Figure 43. 198 . l _ 1 _ ppm with JC1,H1 179.1 : 0.8 Hz. JC1,H1 for the a and 8 furanose 1 forms are 174.0 : 0.8 Hz and 170.7 a 0.8 Hz, respectively. JCl Hl the aldehydo form is larger than that for the furanose forms, in agree- ment with predictions based on the s-character of C-1 (178). 13 for C1,H1 for the furanoses in 80% dioxane—HZO is different from lJCl H1 in water (Table 11) and probably reflects conformational differences caused by solvent effects and/or differences in rates of anomerization. 2. Use of 2JC2 H1 to determine aldose configuration and con- formation ’ 80 k nd Pede (124) have re ntl relat d 23 and 2J c a rsen 1 ce y e C1,H2 C2,Hl in several carbohydrate derivatives to the orientation of oxygen atoms along the coupling pathway through a projection-sum. We have exa- mined JC2,H1 for several furanoses, furanosides and pyranoses (Table 18) and have found that the predicted couplings, based on projection- sums, and observed couplings are in good agreement. Furanose rings in which OH-l and OH—2 are trans Show small 2J couplings (0 Hz - 1 Hz) C2,Hl Calculation of the projection-sums from models for all conformational forms that affect the relative orientation of substituents on C-1 and C-2 (82, 80, 3T2, 82, 80, 2T3) for trans-1,2 furanose rings predicts a range of values for ZJCZ H1 of approximately 0 Hz - 2 Hz. A similar calculation for cis-1,2 furanose rings predicts a range of values for 2 JCZ H1 of approximately 2 Hz to 5 Hz. As shown in Table 18, the ex- ’ perimental results are in close agreement. These results suggest that 2 JCZ H1 may be a more sensitive probe than lJ (Figure 35) in the C131 determination of relative configuration about the C-l-C-Z bond in fura- nose rings. 2JC2 H1 for a—D- and S-D-arabinopyranoses and a-D— and B-D—ribo- ’ pyranoses were also determined (Table 18). Projection-sums for . 2 . Table 18. JC2,H1 Values derivatives. for several carbohydrates and their 199 Compound 2 Jc2,11 a 1 (Hz) cis l Furanoses ,2 trans Hydrates 1,2 Pyranoses a-D-Erythrose S-D-Threose a-D-Ribofuranose B-D—Arabinofuranose Methyl a-D-Ribofuranoside 3—0-Acetyl-1,2:5,6—di-0- isoprOpylidene-a-D- allofuranose 3-O-Acety1-1,2:5,6-di-O— isopropylidene-n—D- glucofuranose 8-D-Erythrose i-D—Threose S-D-Ribofuranose Methyl B-D-Ribofuranoside D-Erythrose, hydrate D-Threose, hydrate 8-D-Ribopyranose d-D-Arabinopyranose a-D—Ribopyranose B-D-Arabinopyranose to IN) U1 0 Unit-‘9») OOH Ul <1 br br br br 4.1 br a Coupling constants were determined by analysis of the H-1 region of the H NMR spectra obtained at 180.04 MHz, and are accurate to within t0.2 Hz. These values were taken from reference 124. 200 (”0.5 Hz) for both 1C and o . 2 B-D—ribopyranose predict a small JC2,H1 4 4 , . . . . Cl conformations. For a-D—ribopyranose, pr0jection sums predict a 4 small value for ZJC2 H1 (NO.5 Hz) for the Cl conformation and a large value Ck? Hz) for the 1C4 conformation. As shown in Table 18, 2JC2 H1 for a-D- and 8-D-ribopyranose are 4.1 Hz and