CARBOHYDRATE PHOSPHATES By A. B. FOSTER YH.D. (UNIVERSITY OF BIRMINGHAM) and W. G. OVEREND D.Sc. (BIRKBECK COLLEGE UNIVERSITY OF LONDON) IN their classical studies of the effect of inorganic phosphate on cell-free sugar (glucose) fermentation Harden and Young showed that carbohydrate phosphate is formed in addition to alcohol and carbon dioxide according to the equation 2 Glucose + 2 Phosphate = 1 Hexose &phosphate + 2 Alcohol + 2 Carbon dioxide . - (1) The phosphoric ester which accumulates is fructose 1 6-diphosphate fre- quently termed the Harden-Young ester. These observations led to other important investigations for example the isolation of glucose monophos- phates by Robison and Embden studies of glycolysis in muscle extracts by Meyerhof and Lohmann and Lundsgaard's observations on the chemical events which accompany the alactacid muscle contraction.This work clearly indicated that phosphoric esters play a central part in the biological world by linking processes of respiration and fermentation with other essential cellular reactions. Quite a large number of esters of this class are now known and the group is still growing as further compounds are isolated. Amongst these phosphoric esters those of the carbohydrates form a principal class and are known to function as intermediates in the network of enzymic reactions associated with the breakdown and interconversion of carbo- hydrates in plants and animals. Consequently it is not surprising that this class of compound has attracted widespread interest and there is a flood of papers annually on their biological function.Although strictly chemical studies are less numerous all facets of the subject could not be condensed adequately into a Review of the present type so we propose to emphasise chemical synthesis isola.tion and reactions of sugar phosphates and to mention only briefly their biological r81e. For more detailed account's reference should be made to recent re~iews.l-~ Detection and Estimation Both in the intact cell and in the isolated enzyme systems in which biological reactions are studied phosphoric esters usually occur as mix- tures. Colour reactions are used to identify the sugar component of carbo- hydrate phosphates ; e.g. the reaction with resorcinol permits the estimation Leloir Portschr. Chem. org. Naturstoffe 1951 8 47. 2Foster Overend and Stacey Die Starke 1953 11 285.Benson " Phosphorylated Sugars " in " Moderne Methoden der Pflanzenanalyse " 1'01. 11 Springer-Verleg Berlin 1955. 61 62 QUARTERLY REVIEWS of ketose esters,4 and pentose esters are detected with orcinol. The rate of colour development with the latter reagent serves to differentiate ribose 3- and 5-phosphate,5 * and also compounds containing phosphoribose residues (and possibly related pentose phosphates). (The method is rapid and can be used on as little as 10 i ~ g . of phosphate ester. It cannot be used precisely on crude plant and bacterial extracts containing polysac- charides since these alter the rate of colour development.) As in other branches of carbohydrate chemistry paper chromatography provides a valuable micromethod for identification. Removal of inter- fering ions by ion-exchange resins from hydrolysates of hexosephosphates improves the chromatograms.6 Hanes and Isherwood 7 demonstrated that it is €easible to separate phosphoric esters including compounds of very similar constitution on a filter-paper chromatogram and to detect them by spraying the papers with an acid molybdate solution and then heating thein under conditions which hydrolyse the esters without unduly decom- posing the paper.The orthophosphoric acid produced fornis a phospho- molybdate complex and this is reduced to an intensely blue compound on exposure to hydrogen sulphide. Various solvent mixtures have been detailed for the chromatographic separation,8 and lists of R values for sugar phosphates have been published. Addition of boric acid to the solvents helps to separate esters with cis-hydroxyl groups from esters in which this grouping is absent .Q The unidimensional chromatography described by Hanes and Isherwood does not always adequately resolve the complex mixtures obtained from some plant materials (cf.Mortimer 8) and so two-dimensional chromato- graphy with successive development in an acid and in a basic solvent has been worked out.1° I n addition? modifications of the original Hanes- Isherwood method have been described. It is claimed that by upward migration at 4" on acid-washed paper with appropriate solvents it is possible to adopt shorter running times and achieve higher R values. This method gives more discrete spots than the two-dimensional procedure and these spots are detected by dipping rather than spraying the papers.11 Irradia- tion with ultraviolet light resulting in colour differences,1° has been used to differentiate between organic compounds containing bound phosphorus and those containing inorganic phosphate.Two drawbacks to the widely used Hanes-Isherwood method are that 6Albaum and Umbreit ibid. 1947 167 369. Roe J . Biol. Chern. 1934 107 15. Dulberg Roessler Sanders and Brewer ibid. 1952 194 199. Hanes and Isherwood Nature 1949 164 1107. Mortimer Canad. J . Chem. 1952 30 653 ; see also Wright and Khorana J . Amer. Chem. SOC. 1956 '78 811 and Loring Levy and Moss Analyt. Chem. 1956 28 539. Scott and Cohen J . Riol. Chem. 1951 188 509 ; Science 1950 111 543. lo Bandurski and Axelrod J . Biol. Chem. 1951 193 405. l1 Burrows Grylls and Harrison Nature 1952 1'70 800. * In this Review nomenclature of the type ribose 5-phosphate is used as customary when it is not desired t o specify whether the compound is present as free acid R*O.PO,H or as salt.For part'icular derivatives the Anglo-American agreed nomenclature is used [see J. 1952 51 11 rule 1 l(d)] e.g. cr-D-ribose &(barium phosphate). FOSTER AND OVEREND CARBOHYDRATE PHOSPHATES 63 the prolonged initial " digestion " necessary to break down the more resistant esters often leaves the paper in a fragile state and that further analyses cannot be carried out on the spot after the detection treafment. To overcome the former difficulty Fletcher and Malpress 1 2 used an enzyme (alkaline phosphomonoesterase) to break down the esters resolved on the chromatogram. To counteract the latter a method has been used depend- ing on fixation of ferric ions by the esters and reaction of the free ferric ion with " salicylsulphonic acid ".13 The phosphates appear as white spots on a pale mauve background orthophosphoric acid having a band of deeper mauve surrounding it.Other spot indicators of ferric ions also have given good results but in some cases the colours fade in light. In experiments with diphenylphosphoric esters ultraviolet contact prints have been used to find the position of the spots on a chromstogram,l4 and radiograms of sugar phosphates labelled with 32P have been examined by Calvin and his colleagues 15 during their work on photosynthesis. Ion-exchange resin chromatography has been used to separate mixtures of sugar phosphates ; e.g. Horecker and Smyrniotis 16 used Dowex-1 formate for the separation of pentose phosphates formed from 6-phospho- gluconate by yeast enzyme.Separations have also been achieved by ion- exchange with the aid of the borate c0mplex,~75 l8 and by ionophoresis l4 in borate buffer a t pH 10 and in acetate buffer at pH 5 at 800 v. Different phosphate esters have been separated by counter-current distribution the solubility in the organic phase was increased by addition of long-chain amines.lg Procedures other than chromatography have also been used to estimate sugar phosphates. A method described by Slater 2o depends on enzymic conversion of these compounds into dihydroxyacetone phosphate which subsequently reacts with reduced diphosphopyridine nucleotide (DPN) in the presence of glycerol phosphate dehydrogenase. The amount of reduced nucleotide undergoing reaction is determined spectrophotometrically.The method is highly sensitive-0.05 millimole of phosphorylated sugar can be measured with an accuracy of a few per cent. Methods for the estimation of fructose diphosphate 21 and glucose 6-phosphate 22 have been outlined and very small amounts of glucose diphosphate can be estimated by taking advantage of its coenzyme activity for phosphoglucomi~tase. 23 A method l2 Fletcher and Malpress Nature. 1953 1'71 838. l3 Wade and Morgan ibid. p. 529. l4 Matthews and Overend unpublished results. l5 Benson Bassham Calvin Goodale Haas and Stepka J . Amer. Chem. Xoc. 1950 l6 Horecker and Smyrniotis Arch. Biochem. Biophys. 1950 29 232. l7 Khym and Cohn J . Amer. Chem. SOC. 1953 75 1153. l8 Khym Doherty Volkin and Cohn ibid. p.1262. Plaut Kuby and Lardy J. Biol. Chem. 1950 184 243. 2o Slater Biochem. J. 1953 53 157. 21 Meyerhof and Wilson Arch. Biochem. Biophys. 1948 17 153. 2 2 Haas J . Biol. Chem. 1944 155 333. 23 Carclini Yaladini Caputto Leloir and Trucco Arch. Biochenz.. Biophys. 1949 72 1710. 22. 87. 64 QUARTERLY REVIEWS proposed for the estimation of fructose diphosphate is based on the deter- mination of the phosphate groups liberated during osazone formation. 24 Differences in the rate of hydrolysis of various sugar phosphates provide in some cases a method for their estimation in simple mixtures. Optical rotation has been used for distinguishing between ribose 3- and 5-phosphate and other pentose phosphates. Isolation from Natural Sources and Preparation by Enzymic Methods The pioneer investigations of Harden and his colleagues stimulated work on the isolation of sugar phosphates from natural sources.In addition to the changes formulated in equation (1) (the Harden-Young equation) it is possible under different conditions to obtain by the use of dried yeast or yeast- juice fermentations hexose monophosphate in amounts varying from 20 to 50% or more. The diphosphates can be separated from the monophosphates 25 and can be further differentiated by fractional crystal- lisation of their brucine salts.26 I n 1937 Cori Colowick and Cori 27 showed that a-D-glucose 1-phosphate (Cori ester) is formed when a solution of glycogen inorganic phosphate and adenylic acid is incubated with a dialysed muscle extract. Phosphorylase is now known to be widespread in Nature. The reverse of this reaction namely the enzymic conversion of the Cori ester into 1 4-a-glucosans is well known (cf.equation 2) and a recent review in this series by Barker and Bourne 28 on the enzymic synthesis of polysaccharides includes a full dis- cussion of the formation of amylose and glycogen from glucose 1-phosphate. (CgH1005)n + nK2HPOd + nC,H,1O,.OPO,K2 . - (2) 1 4-a-Glucosan K salt of Cori ester At equilibrium the ratio of total inorganic phosphate to total glucose 1-phosphate depends on the pH value of the system but the ratio of the bivalent ions [HP0,]2-/[C,H,,0,~O*P03] 2- is independent of pH and is always constant 299 30 a t 2.2. Hence the conversion of an unbranched 1 4-a-glucosan into a-glucose ]-phosphate can be carried to virtual com- pletion if the polysaccharide is treated with phosphorylase in the presence of a sufficiently large excess of inorganic phosphate to ensure that the equilibrium ratio of the bivalent ions is not attained before all the poly- saccharide is degraded.31-33 Since the enzymic degradation of amylose is so effective and easy to control the preparation of a-glucose 1-phosphate by this method is popular. Glucose 1-phosphate can be rearranged by 2 4 Deuticke and Hollman Z. physiol. Chem,. 1939 258 160. 25Robison and Morgan Bioc1io.m. J. 1930 24 119. 26 Robison and King ibid. 1931 25 323. 2 7 Cori Colowick and Cori J . Biol. Clhem. 1935 123 375 381. 28 Barker and Bourne Quart. Rezj. 1953 7 56. 29 Hanes Nature 1940 145 348 ; Proc. Boy. Soc. 1940 By 128 421 ; 129 174. 3O Trevelyan Mann and Harrison Arch. Rioclbetn. Biophys. 1952 39 419 440.31 Swanson J . Biol. C'h,ena. 1948 172 805 825. 32Bourne Sitch mid Peat J. 1949 1448. 33 Hestrin J . Biol. Ch,ern. 1949. 179 943. FOSTER AND OVEREND CARBOHYDRATE PHOSPHATES 65 phosphoglucornutase to glucose 6-phosphate,34 which can also be obtained in good yield directly frpm starch by using phosphorylase and phospho- glucomutase in conjunction. 35 In Escherichia coli a biosynthesis has been detected in which a tmnsphosphorylation between two glucose 1 -phosphate molecules gives glucose 1 6-diphosphate and free glucose.36 Glucose 1 6-diphosphate has been isolated in small amount from crude fructose diphosphate preparations 37 obtained by fermentation procedures the diphosphates being separated by destroying the fructose ester with alkali which leaves the glucose analogue unchanged.Hydrolysis of fructose 1 6-diphosphate with phosphatase splits off both phosphate residues at the same rate and thus half of the monophosphate formed is fructose l-phos- hate.^^ It is not advisable to use highly purified phosphatase since the crude enzyme also changes fructose 6-phosphate into glucose 6-phosphate. The glucose derivative can be oxidised to 6-phosphogluconic acid and separ- ated as its insoluble barium salt. Consequently less fructose 6-phoaphate remains to be separated from the l-isomer than is the case if purified enzyme is used. Fructose l-phosphate was later obtained by an aldolase-induced condensation of phosphodihydroxyacetone and D-glyceraldehyde. 39 If DL-glyceraldehyde is used the products are fructose 1 -phosphate and sorbosc l-phosphate.39 To cite one example fructose 6-phosphate is formed when fructose and adenosine triphosphate are incubated with yeast hexokinase.40 41 This ester is also produced by the action of a specific enzyme (phosphomannose isomerase) on mannose 6-phosphate itself obtained by phosphorylation of mannosc with hexokinase. There are many references to the enzymic preparation of other hexose phosphates but the products have not in all cases been fully purified or satisfactorily characterised. Kalckar 42 observed that enzymic phosphorolysis of some ribonucleosides (inosine guanosine) leads to the formation of a pentose phosphate considered to be D-ribofuranose l-phosphate. The yield is very low possibly owing Inosine + Phosphate T Ribose l-phosphate + Hypoxanthine . (3) to losses by acid hydrolysis t o specific and non-specific contaminant phos- phatase action during the incubation with the enzyme and to retention on the bulky barium phosphate precipitate during working-up.Moreover in reaction (3) the equilibrium favours formation of the nucleoside rather a4 Colowick and Sutherland J . Biol. Chem. 1942 144 423 ; Sutherland Colowick and Cori ibid. 1941 140 309. 36 Swanson ibid. 1950 184 647. 36 Lelok Trucco Cardini Paladini and Caputto Arch. Biochem. Biophys. 1949 37 Idem ibid. 1948 19 339; 1949 22 87. 38MacLeod and Robison Biochem. J . 1933 27 286. 39Meyerhof Lohmann and Schuster Biochem. Z. 1936 286 301 319. 40 Kunitz and McDonald J . Gem. Physiol. 1946 20 393. 4lBerger Slein Colowick and Cori ibid. p. 379. r a Kalckar J. Biol. Chem. 1946,158,723 ; 1947 167,477 ; Fed.Proc. 1946,4,248 ; The action of kinases on the sugars is well established. 24 65. Symp. Boo. Expt. Biol. 1947 1 38. XI 66 QUARTERLY REVIEWS than of the pentose phosphate. Addition of xanthine-oxidase to the system leads t o the removal of hypoxanthine by conversion into xanthine and uric acid the pentose phosphate is then isolable as its barium salt Naturally occurring nucleosides are derivatives of P-ribofuranose 43 and apparently nucleoside phosphorylase produces inversion and this ribose 1 -phosphate has been shown to have the or-c~nfiguration.~~~ Synthetic ribopyranose l-phosphate 45 will not serve as substrate for the enzyme producing nucleo- sides a result which suggests that the pentose phosphate produced according to equation (3) is of the furanose type. This has been confirmed by the chemical synthesis of or-D-ribofuranose 1 -phosphate 44b which was found to be identical with enzymically prepared samples and to be fully active as a substrate for the fish-muscle purine-nucleoside phosphorylase.Phosphorolysis of deoxyribonucleosides has also been achieved. Enzyme preparations from calf-thymus gland and rat liver act on hypoxanthine deoxyriboside to give an acid-stable phosphate ester ; this is 2-deoxyribose 5-phosphate and is formed from deoxyribose 1 -phosphate by mutase action.46 From the enzymic phosphorolysis product of guanine deoxy-D- riboside Friedkin 47 isolated %deoxy-~-ribose 1 -phosphate as the crystalline cyclohexylamine salt. Recently a simplified procedure for the isolation of deoxyribose 1 -phosphate has been developed it involves phosphorolysis of thymidine in the presence of ammonium dicycEohexy1 hydrogen phosphate followed by a fractionation with butan-1 -01-diethyl ether which yields crystalline dicyclohexylammonium deoxyribose 1 -phosphate after a single filtrati0n.4~ This ester is even more unstable than ribose l-phosphate and is hydrolysed by the acid used in methods for phosphate estimation it therefore appears in analyses as “ inorganic phosphate ”.By mutase action ribose l-phosphate can be converted into ribose 5-phosphate. Klenow and Larsen 4s have shown that phosphoglucomutase acting with glucose 1 6-diphosphate (and probably ribose 1 5-diphosphate) as coenzyme will bring about this change. Preparations from liver also contain a mutase capable of transforming ribose l-phosphate into the 5-is0mer.~~ Levene et aL51 claimed to have prepared ribose 3-phosphate from nucleo- tides (xanthylic and yeast adenylic gcid) but more recent work has shown that they were handling mixtures.In the light of present knowledge con- cerning the migrations of phosphate esters it is obvious that the experimental conditions employed by the Levene school could not have resulted in the retention of isomeric integrity in the compounds studied but would lead 43Davol1 Lythgoe and Todd J. 1946 833. 4 4 (a) Wright and Khorana J . Amer. Chem. Xoc. 1956 78 811 ; ( b ) Tener Wright 4 5 Kalckar Biochim. Biophys. Acta 1950 4 232. ‘6Manson and Lampen J . Biol. Chem. 1951 191 95. 4 7 Friedkin ibid. 1950 184 449. 481i.riedkin and Roberts ibid. 1954 207 257. 49 Klenow and Larsen Arch. Biochem. Biophys.1952 37 488. bo Wajzer and Baron Bull. SOC. Chim. biol. 1949 31 750. 61 Levene and Harris J . Biol. Chenz. 1932 95 755 ; 98 9 ; 1933 101 419. and Khorana ibid. p. 506. FOSTER AND OVEREND CARBOHYDRATE PHOSPHATES 67 to mixtures. Ribose phosphates have been obtained in ingenious fashion by Khym et al.18 Hydrolysis of the glycosylamine nitrogen-carbon linkage in adenylic acid " a " and " b " was achieved with the hydrogen form of a polystyrenesulphonic acid resin at a rate comparable with the rate of isomerisation. The ribose phosphates were released from the resin at the moment of formation (in contrast to adenine and most of the adenylic acid) and little or no isomerisation takes place subsequent to their formation. In this way ribose 2-phosphate was obtained from adenylic acid " a " and ribose 3-phosphate from adenylic acid " b ".Subsequently the method was developed to obtain pure ribose 2- and 3-phosphate by hydrolysis of adenylic acids with a polystyrenesulphonic acid cation-exchange resin. The mixture of phosphate esters was separated by ion-exchange chromato- graphy with borate complex-formation. Khym et al. also prepared ribose 5-phosphate by treating adenosine-5' phosphate with resin [Dowex-SO(H+)] at 100" for 4 minutes. This ester had been obtained by Levene and Jacobs 62 by subjecting the barium salt of inosinic acid to acidic hydrolysis thereby cleaving the sugar-base linkage. An improved method for the preparation from muscle of inosinic acid and then of ribose 5-phosphate has been de- scribed recently. Optimum conditions were determined for the hydrolysis.53 This ester is also obtainable by acidic hydrolysis of cozymase 64 and it can be prepared in a high degree of purity from adenosine triphosphate by ion-exchange.55 A fraction containing 70-80y0 of ribose 5-phosphate is afforded when xylose and adenosine triphosphate are incubated with a pentose phosphate isomerase from extracts of Lactobacillus pent0sus.56~ The enzymic conversion of 6-phosphogluconic acid into ribulose 5-phosphate and then ribose 5-phosphate is now well established.In the past to obtain ribose phosphates from ribonucleotides it has been necessary to work with purine nucleotides but very recently Cohn and Doherty 56b have developed a method for obtaining ribose from pyrimidine nucleosides and ribose phosphates from pyrimidine nucleotides. The accessibility of sugars (and derivatives) of pyrimidine nucleosides and nucleotides is severely limited by the resistance of the glycosylamine link- age to acid hydrolysis.It has long been known that this stability is depend- ent on the ethylenic unsaturation between the adjacent carbon atoms in the ring and that reduction or bromination of the 4 5-double bond renders the glycosylamine linkage susceptible to acid hydrolysis. Cohn and Doherty completely hydrogenated pyrimidine ribonucleotides under mild conditions with a rhodium catalyst and cleaved the product by dilute alkali a t room temperature to the phosphate of /?-ribosylureidopropionic acid. Dilute acid 'at room temperature hydrolyses this substance to ribose phos- phate and @-ureidopropionic acid without appreciable isomerisation of the 62Levene and Jacobs Ber.1908 41 2703; 1911 44 746. 53Marmur Schlenk and Overland Arch. Biochem. Biophys. 1951 34 209. 54Schlenk J . Biol. Chem. 1942 146 619. 5 5 Groth Mueller and LePage ibid. 1952 199 389. 66 (a) Lampen ibid. 1953 203 999 ; ( b ) Cohn and Doherty J . Amer. Chm. SOC. 1956 78 2863 ; (c) Bergmann and Burke Angew. Chem. 1955 67 127. 68 QUARTERLY REVIEWS phosphate group thus making available the sugar phosphates of pyrimidine nucleotides. No previous isolation of a sugar phosphate from a pyrimidine nucleotide had been reported and even the reduction of such substances to achieve labilisation of the glycosylamine linkage has been achieved only rarely. The sodium-ethanol-liquid ammonia procedure so effective with nucleosides is seemingly ineffective with n ~ c l e o t i d e s .~ ~ ~ From uridylic acids " a " and " b " ribose 2- and 3-phosphate respectively were obtained thus confirming the identity of the pyrimidine nucleotide isomers. The method is also applicable to deoxyribonucleotides and has been used with deoxycytidylic and thymidylic acid Evidence has been presented to show that xylose is phosphorylated a t the expense of adenosine triphosphate by extracts of Pseudomonas Chemical Syntheses Intrigued by the problems presented and no doubt stimulated by the biological implications of sugar phosphates organic chemists have developed chemical syntheses for many members of this class. I n early experiments it was usual to phosphorylate unprotected sugars and the products were probably mixtures. As far as we can trace the first phosphorylation of a carbohydrate was carried out in 1858 by Berthelot,58 who treated glucose with syrupy phosphoric acid a t 140".I n the past the most widely used reagent in synthesis of sugar phosphates was phosphoryl chloride. It was used by Neuberg and Pollak 59a to prepare sucroseand dextrose phosphates by Fischer 6o to obtain a phosphoric ester of methyl glucoside and by Helferich et aL61 to phosphorylate an unprotected disaccharide (trehalose). Neuberg and Pollak attempted to control the reaction by adding alkali to absorb the hydrogen chloride formed. Substances which have been added by others for the same reason include sodium hydroxide magnesium oxide anhydrous pyridine and quinoline. Inconsistencies have been noted and it has been reported that phosphorylation of glucose was unsuccessful when barium or calcium hydroxide was replaced by calcium carbonate as the added base.59b 62 More examples need to be studied before all the incon- sistencies can be satisfactorily explained.Many phosphorylations have been carried out with suitably protected sugars. The following are a few representative examples reaction between methyl 2 3-O-isopropylidene-~-ribofuranoside and phosphoryl chloride in pyridine a t -40° followed by hydrolysis of the isopropylidene and glycoside residues yielded ribose 5-phosphate ; 63 arabinose 5-phosphate has also been prepared ; 64 phosphorylation of 1 2-5 6-di-O-isopropylideneglucose hydrop h i h .57 67 Hochster and Watson Nature 1952 170 357. 5BBerthelot Ann. Chim. (France) 1858 54 81. 59Neuberg and Pollak (a) Biochem. Z .1910 23 515; 26 514; ( b ) Ber. 1910 61 Helferich Lowa Nippe and Riedel 2. physio2. Chem. 1923 128 141. 6 2 Fawaz and Zeile ibid. 1940 263 176. 63Levene and Stiller J. Bid. Chern. 1934 104 299. srLevene and Christman ibid. 1938 123 607. 43 2060. 6o Fischer Ber. 1914 47 3193. FOSTER AND OVEREND CARBOHYDRATE PHOSPHATES 69 and 1 2 3 6-tetra-O-acetylglucose and subsequent removal of the pro- tecting groups afforded glucose 3- G 5 66 and 4-phosphate 67 respectively. Another reagent which has been used to some extent is phosphoric oxide. Di-0-isopropylidene-D-fructopyranose was phosphorylated with phosphoric oxide in ether and the intermediate product presumably a mixture of tri- di- and mono-0-isopropylidenefructose 1 -phosphate was subjected to hydrolysis fructose 1-phosphate was isolated as the cyclo- hexylammonium salt.68 Although it is usual and frequently necessary to use protected sugars in phosphorylations selective reaction of free sugars can be achieved in some instances. After the conversion of glucose into glucose 6-phosphate 69 by metaphosphoric acid Percival and Anderson 70a directly phosphorylated glucosamine at position 6 with metaphosphoric acid in the presence of acetonitrile. Phosphate residues on other positions were removed by hydrolysis of the crude product with N-hydrochloric acid at 100". (A purer product has since been prepared by these workers by an alternative route ; 70b cf. Maley and Lardy.70c) Amino-sugar phosphate esters had previously only been obtained enzymically . In addition to working with protected sugars nowadays it is usual to use protected phosphorylating agents to eliminate undesirable side reactions.The value of such reagents was realised many years ago since Langheld 7 l in 1910 used ethyl metaphosphate in chloroform. To prevent the formation of di- and tri-esters it has become customary to use a disubstituted phos- phoryl monochloride usually in pyridine as the phosphorylating agent. It is essential of course that the protecting groups should be removed easily under mild conditions. Compounds which have been suggested as useful include phosphorochloridic dianilide the catechol ester of phos- phorochloridic acid and dibenzyl and diphenyl phosphorochloridate. The first-named compound was used for the phosphorylation of a series of com- pounds including a sugar ; 7 2 the aniline residues were removed as acetanilide by hydrolysis with acetic acid.It is claimed that catechol can be eliminated from the catechol ester of phosphorochloridic acid merely by treatment with water.73 The most useful and widely used phosphorylating agents are dibenzyl and diphenyl phosphorochloridate. The benzyl or phenyl groups can be readily cleaved by hydrogenolysis. The diphenyl derivative which is a stable liquid (for preparative details see Brigl and Muller 7 4 6 5 Nodzu J . Biochem. (Japan) 1926 6 31 ; Chem. Abs. 1927 21 924. 66 Levene and Raymond J . Biol. Chem. ( a ) 1928 79 621 ; ( b ) 1929 83 619; 6 7 Raymond ibid. 1936 113 375. 68Pogell ibid. 1953 201 645. 69 Viscontini and Olivier Heh. Chim. Acta 1953 36 466. 70 (a) Percival and Anderson Chenz,. and Id. 1954 1018 ; ( b ) Anderson and Percival J .1956 814; (c) Maley and Lardy J . Amer. Chem. SOC. 1956 78 1393. 'lLangheld Ber. 1910 43 1857. 7 2 Zetzsche and Buttiker Ber. 1940 B 73 47. 73Reich Nature 1946 157 133. 7 4 Brigl and Muller Ber. 1939 73 2121. (c) 1930 89 479. 70 QUARTERLY REVIEWS or Baer 75) has been used for the synthesis of monophosphates of ald0-,7~ 77 keto-,66b 74 78 2-deoxy-,79 and 2-amino-2-deoxy-hexoses,7~b~ and of pen- toses 809 81 and 2-deo~ypentoses.~~ In addition it has been used to prepare some aZdehydo-sugar phosphates.83 The initial reaction between the pro- tected sugar and the phosphorylating agent proceeds in good yield and the products are frequently crystalline. The phenyl residues can be removed not only by hydrogen and a catalyst but also by dilute sodium hydroxide and in some cases by sodium in liquid ammonia.Illustrative of the use of this reagent are the following phosphorylation of benzyl 3 4 6-tri-O- acetyl-P-D-glucoside yielded the 2-(diphenyl phosphate) which was treated with hydrogen over Adams catalyst to afford hexahydrobenzyl 3 4 6-tri- O-acetyl-P-D-ghcoside 2-phosphate from which the free glycoside phosphate was obtained by deacetylation.77 Similar phosphorylation of 1 3 4 6- tetra-0-acetyl-P-D-glucose followed by treatment of the product with potassium methoxide in methanol yielded glucose 2-( dipotassium phosphate). 1 2-O-koPropylidene-~-xylose when phosphorylated in anhydrous 2 6- lutidine at - 20" with diphenyl phosphorochloridate afforded pure crystalline 1 2-O-~sopropylidene-~-xylofuranose 5-(diphenyl phosphate) ; hydrogeno- lysis in glacial acetic acid over Adams catalyst then quantitatively removed the phenyl groups ; mild hydrolysis in acetic acid cleaved the isopropylidene grouping and D-xylofuranose 5-phosphate was obtained in 72% yield from D-xylose.81 Phosphorylation of 2 3 4 5-tetra-O-acetyl-~-galactose di- ethyl mercaptal with diphenyl phosphorochloridate in pyridine proceeded readily at Oo yielding crystalline 2 3 4 5-tetra-O-acetyl-~-galactose diethyl mercaptal6- (diphenyl phosphate) which on scission of the ethylthio- residues afforded 2 3 4 5-tetra-O-aCetyl-Uldehydo-D-galaCtOSe 6-(diphenyl phosphate).A similar reaction sequence was successfully completed with the 2-deoxygalactose anal0gue.8~ 1 3 4-Tri-O-acetyl-N-acetyl-P-~-gluco- samine with the reagent yielded the 6- (diphenyl phosphate) which after hydrogenolysis and acidic hydrolysis of the acetyl groups afforded crystalline D-glucosamine 6-phosphate 70b (cf.ref. 70c). Diphenyl phosphorochloridate was used in the nucleotide field by Bredereck and his collaborator^.^^ Monophenyl 85 phosphorochloridate (and phosphorochloridic monoanilide 7 2 has been used for the production of phosphate esters but shows no advantage over the corresponding disubstituted derivative. Although Zervas 86 mentioned the use of dibenzyl phosphorochloridate ' 5 Baer " Biochemical Preparations " Wiley and Sons New York 1949 Vol. I p. 51. 76 Reithel and Claycomb J. Amer. Chern. SOC. 1949 71 3669. 7 7 Farrar J. 1949 3131. 78Mann and Lardy J. Biol. Chern. 1950 187 339. 79Foster Overend and Stacey J. 1951 980. 8o Parker Ph.D.Thesis Birmingham 1952. Barnwell Saunders and Watson Chem. and Ind. 1955 173 ; Canad. J. Chem. 1955 33 711. 8zAllerton Overend and Stacey Chem. and Id. 1952 952. 83 Barclay Foster and Overend J. 1955 2505. 8 4 Bredereck Berger and Ehrenberg Bey. 1940 73 269. 8 5 Gulland and Hobday J. 1940 746. Se Zervas Naturwks. 1939 27 317. FOSTER AND OVEREND CARBOHYDR-4TE PHOSPHATES 71 he considered it too unstable to be of practical value. The reagent has been developed by Todd and his co-workers and is used extensively by them. If the sole purpose is the preparation of monoesters then possibly the more stable diphenyl analogue is more convenient but the use of dibenzyl phosphorochloridate is not limited to the preparation of simple phosphoric esters and can be applied to the preparation of esters of pyrophosphoric acid and triphosphoric acid (see p.76). A synthesis of ribose 5-phosphate provides an example of the use of this reagent. Methyl 2 3-O-isopropyli- dene-D-ribofuranoside with this phosphorylating agent in pyridine a t low temperature affords methyl 2 3-O-isopropylidene-~-ribofuranoside 5-(di- benzyl phosphate) from which protecting groups were removed by the usual methods to give ribose 5-phosphate in high yieldqs7 In the nucleotide field thymidine-3' phosphate was synthesised by the phosphorylation of 5'-triphenylmethylthymidine with this reagent and subsequent elimination of the triphenylrnethyl and benzyl residues.*8 Some sugar phosphates have been prepared by phosphorylation a t one site the ester grouping being then caused to migrate to another.Levene and Raymond 89 tried to prepare xylose 3-phosphate by phosphorylation of 5-O-benzoyl-1 2-O-isopropylidenexylose but the product was xylose 5-phosphate. Likewise phosphorylation of 5-O-benzyloxycarbonyl- or 5-O-acetyl- 1 2-O-isopropylidenexylose also yielded xylose &phosphate after removal with mineral acid of the acyl and isopropylidene groups and obviously a phosphate migration had occurred. Recently it was claimed by Watson and Barnwell 90a; that migration in the reverse direction (i.e. from position 5 to position 3) afforded xylose 3-phosphate xylose 5-phos- phate was merely heated in water a t pH 6.4 a t 50" for 2 hours. This claim was soon shown to be incorrect by Moffatt and K h ~ r a n a ~ ~ who successfully prepared and fully characterised D-xylose 3-phosphate.Crystalline 1 2-0- isopropylidene-D-xylofuranose 5-(diphenyl phosphate) was converted by alkali into the 1 2-O-isopropylidenexylofuranose 3 5-( cyclic phosphate) which was hydrolysed quantitatively to a mixture of 1 2-O-isopropylidene xylose 3- and 5-phosphate from which the isopropylidene groups were readily cleaved by the aqueous acids a t 100' for 10 minutes a t their own pH. The xylose 3- and 5-phosphates were separated satisfactorily on a Dowex- 2(formate) resin column and the products differentiated structurally by standard carbohydrate reactions. The 3-isomer was obtained in 15 yo yield. The nature of the reaction responsible for the change in the optical rotation of a solution of D-XylOSe 5-phosphate was re-investigated because the properties of the sample of D-xylose 3-phosphate prepared as described above were completely different from those of solutions of D-xylose 5-phos- phate treated according to Watson and Barnwell's procedure.90a Moreover a migration under neutral conditions as postulated by Watson and Barnwell Levene and Raymond J .Biol. Chem. 1934 107 75 ; cf. ibid. 1933 102 317 331 347. O0 ( a ) Watson and Barnwell Chem. and Ind. 1955 1089 ; ( b ) Moffatt and Khorana J . Amer. Chem. SOC. 1956 78 883 ; (c) Axelrod and Jang J . Biol. Chem. 1954 209 847. 87 Michelson and Todd J. 1949 2476. BBIdern J. 1953 951. 72 QUARTERLY REVIEWS seemed highly improbable. The change was found to be really due to the formation of xylulose 5-phosphate from xylose 5-phosphate a transformation analogous to that previously observed by Axelrod and Jang,s*c who reported that ribose &(barium phosphate) can be partially converted a t room temperature into a ribulose-containing compound.When ribose 2- or 3-phosphate is heated for 2 hours with Dowex 50(H+) resin or for 45 minutes with 0-1N-sulphuric acid it forms ribose 4-phosphate in low yield. A method of preparing phosphoric esters which might be further exploited in carbohydrate chemistry is that employing ethylene oxide derivatives of sugars as initial materials. Lampson and Lardy 91 treated 5 6-anhydro- 1 2-O-~sopropy~idene-~-g~ucofuranose in water with dipotassium or disodium hydrogen phosphate and cleaved the anhydro-ring. The phosphate residue was located at the terminal carbon atom of the sugar molecule and by removal of the isopropylidene group glucose 6-phosphate was obtained.Although the yield was lower than by other methods the authors recommend this procedure in the special case when it is desired to introduce labelled phosphate because it avoids the use of special phosphorylating agents. Todd and his co-workers s2 studied the action of dibenzyl hydrogen phosphate on mcthyl 2 3-anhydro-4 6-O-benzylidene-a-~-alloside (I). The product was a mixture of methyl benzylidenehexoside dibenzyl phosphates. After elimination of the benzyl and benzylidene residues this mixture was separ- ated into methyl or-D-altropyranoside 2-phosphate (II),* which was the main product and methyl or-D-glucopyranoside 3-phosphate (III).* The Me (II) (I) R PO,H general conclusion drawn by Todd and his colleagues was that the epoxide route is feasible for carbohydrate esters of phosphoric acid and compounds of the nucleotide type but is limited in its application.The limitations were considered to be inaccessibility of appropriate anhydro-compounds and the tendency to formation of more than one product from other than 5 6-anhydro-sugar derivatives. The method probably warrants further study however especially in the light of the newer methods available for the separation of sugar phosphates. Although the preparation of sugar phosphate esters with the substituent located at the glycosidic centre of the sugar is frequently achieved by en- zymic methods chemical syntheses have been developed. The reaction glLampson and Lardy J . Biol. Chem. 1949 181 693. B2Harvey Michelksi and Todd J. 1951 2271. * Depiction of this and other sugar phosphates as free acids does not imply that they were always isolated as such.FOSTER AND OVEREND CARBOHYDRATE PHOSPHATES 73 between the acetobromo-sugar and sodium or (more usually) silver phos- phate or silver diphenyl or dibenzyl phosphate is frequently employed. Depending on the experimental conditions and reagent either the a- or the @-derivative is formed. The a-form of glucose 1 -phosphate was successfully obtained by treating trisilver phosphate with acetobromoglucose in ben- ~ e n e . ~ ~ 9 93 The initial product tris(tetra-0-acetylglucose-1) phosphate was hydrolysed by acid in methanol until about 20% of the organic phosphate was liberated and deacetylation was completed with alkali. The method has been described in detail by Krahl and CorL9* In like fashion a-gal- actose l - p h ~ s p h a t e ~ ~ xylose l-pho~phate,~O and maltose 1 -phosphate Q8 have been prepared.Usually a-acetobromoglucose reacts with inversion of configuration at the glycosidic centre and in this respect the preparation of a-glucose 1-phosphate is anomalous. Posternak 97 treated acetobromo-aldoses with silver diphenyl phosphate and cleaved the phenyl groups from the product by hydrogenolysis. Treat- ment with alkali resulted in deacetylation and the glycoside phosphate was isolated. a-Glucose 1 -phosphate and cc-galactose 1 -phosphate were synthe- sised in this way and the yields are reported to be five times those obtained by the trisilver phosphate procedure. In similar fashion a-D-glucose 1 6-diphosphate was prepared from 2 3 4-tri-O-acetyl-l-bromo-l-deoxy- a-D-glucose 6- (diphenyl phosphate) .989 99 Other compounds prepared by this route include a-D-mannose l-phosphate and 1 6-diphosphate and a-lactose l-phosphate.If a-acetobromo-D-glucose (IV) is treated with silver dibenzyl phosphate reaction occurs with inversion of configuration and after elimination of protecting groups ,!%D-glucose l-phosphate (VII) can be isolated.86 loo The compound is formed via the intermediates (V) and (VI). O 3 Cori Colowick and Cori J . Biol. Chem. 1937 121 465. 9 4 Krahl and Cori '' Biochemical Preparations " Wiley and Sons New York 1949 95Colowick J. Biol. Chem. 1938 124 557. Q6Meagher and Hassid J . Arner. Chern. SOC. 1946 68 2135. " Posternak ibid. 1950 72 4824. "Posternnk J. Bid. Chem. 1949 180 1269. O9 See also Leloir Repetto Cardini Paladini and Caputto Andes Asoc.quim. loo Wolfroin Smith Pleteher and Brown J . Amer. Chem. SOC. 1942 64 23. VOl. I p. 33. argentim 1949 37 187. E" 74 QUARTERLY REVIEWS P-D-Galactose 1 -phosphate can also be obtained by this procedure,lo1 but a-D-mannose 1 -phosphate is formed when acetochloromannose is treated with silver diphenyl phosphate or silver dibenzyl phosphate with subsequent removal of protecting groups. loZu Likewise acetobromo-D- xylose affords finally a-D-XylOSe 1 -phosphate when treated with either silver diphenyl or dibenzyl phosphate. loZb Khorana and his colleagues have successfully synthesised both a- and P-D-ribofuranose 1 -pho~phate.lo~~? 2 3 Ei-Tri-O-benzoyl-~-~-ribose was converted into the corresponding ribofuranose l-bromide ( V I I I ) to which the P-configuration has been assigned.At low temperature this compound underwent some reaction with silver dibenzyl phosphate in a medium of chloroform and methylene dichloride and chromatography of the product after hydrogenation showed the presence of a fast-moving labile phosphate but much inorganic phosphate was also present. After debenzoylation only very small yields of ribofuranose l-phosphate were obtained. To reduce losses a much shorter reaction period appeared advisable and to achieve this advantage was taken of the high solubility in benzene of tri- ethylammonium dibenzyl phosphate. When a cooled benzene solution of this salt was added to a precooled solution of compound (VIII) a rapid reaction ensued. As expected the product (IX) was extremely labile and direct hydrogenation appeared desirable in order to secure some stabilisation of the ester by the creation of phosphoryl dissociation.This viewpoint was borne out by experiment and after removal of the benzoyl groups a considerably improved yield of P-ribofuranose 1 -phosphate (X) was obtained. (The P-configuration was based on enzymic studies and methods described later.) In this case we have the formation of a 0-glycose l-phosphate from a /3-glycose l-halide on reaction with a salt of dibenzyl phosphoric acid. The importance of " neighbouring group ' ' participation in the synthesis of purine nucleosides is well known and the configuration at appears to depend on the position of the 2-hydroxyl substituent in that in all known cases the base is on the opposite side of the ring from this 2-substituent regardless of the relative configuration at positions 1 and 2 in the original halogeno-sugar (see Baker et ~ 1 .l ~ ~ ~ for a fuller discussion of this point). lolReithel J . Amer. Chem Soc. 1945 67 1056. lo2 (a) Posternak and Rosselet Helv. Chim. Acta 1953 36 1614 ; (b) Antia and Wataon Chem. and Ind. 1956 1143. Io3 (a) Tener Wright and Khorana J . Amer. Chem. SOC. 1956 78,506 ; (b) Wright and Khorana ibid. 1955 77 3423 ; 1966 78 811 ; ( c ) cf. Baker Joseph Schaub and Williams J . Org. Chem. 1954 19 1786 ; ( d ) Maley Maley and Lardy J. Amer. (Ihern. SOC.. 1956. 78 6303. FOSTER AND OVEREND CARBOHYDRATE PHOSPHATES 75 An analogous explanation can be entertained for the nature of the products formed when salts of dibenzyl phosphoric acid are treated with acylglycose 1-halides. To prepare the other form of the anomeric pair of phosphates the blocking group a t should not exercise the important neighbouring- group influence in the replacement reaction a t position 1 and in addition should be readily removed at a later stage in the synthesis.cc-D-Ribo- furanose 1 -phosphate was synthesised by taking account of these require- ments. Methyl 5-O-benzyl-~-ribofuranoside 2 3-carbonate (XI) was con- verted by hydrogen bromide in acetic acid into an oily bromide which was directly treated in benzene with one equivalent of triethylammonium dibenzyl phosphate. Hydrogenation of the product followed by mild alkaline treatment afforded a-D-ribofuranose 1 -phosphate (XII). The use of triethylammonium dibenzyl phosphate appears to be most promising for the synthesis of labile glycosidic phosphates.Lardy and his co-workers have recently prepared a-D -glucosamine 1 -phosphate (and its N-acetyl derivative) by treating acetobromoglucosamine hydrobromide with the triethylamine salt of diphenyl phosphoric acid and subsequent removal of the protecting groups.10M Reactions of acetohalogeno-sugars with monosilver phosphate (for pre- paration see Lipmann and Tuttle lo4) usually proceed with inversion and lead to the formation of p-glycosides. After removal of the protecting groups from the product of reaction of acetobromogalactose and monosilver phosphate p-D-galactose 1-phosphate was obtained.lol Methyl 2 3 4-tri- O-acetyl- 1 - bromoglucuronate with monosilver phosphate gave finally /I-glucuronic acid 1 -phosphate 105 (see also Pippen and McCready 106 for other attempts to prepare hexuronic acids with 1 -phosphate substituents).A thorough study of the reactions for the preparation of aldose l-phosphates would provide useful information. Although the nature of the products formed from acylglycosyl halides and salts of dibenzylphosphoric acid can be explained the reactions with salts of diphenylphosphoric acid appear anomalous. Likewise configurational assignments are demonstrated only for the aldose l-phosphates finally isolated and not on the initial pro- ducts of reaction of the acylglycosyl halides and phosphoric acid diester salts. * lo4Lipmann and Tuttle J . Biol. Chem. 1944 153 571. 105Touster and Reynolds ibid. 1952 19’7 863. lo6Pippen and McCready J . Org. Chem. 1951 16 262. * The reaction between the silver salt of these phosphoric acids and a halogeno-sugar in which the halogen grouping is located at positions other than 1 has apparently not been used to give simple sugar phosphates but has been used for nucleotides.Uridine-5’ phosphate was prepared by reaction of silver dibenzyl phosphate and 5’-deoxy-5’-iodo- 2’ 3’-O-isopropylideneuidine with subsequent debenzylation. The method is not 76 QUARTERLY REVIEWS In general for the various syntheses described the site of the phosphoryl residue in the sugar molecule has been confirmed by the classical methods of carbohydrate chemistry involving inter abia glycosidisation methylation periodate oxidation optical rotation and ion-exchange in the presence and absence of borate and differences in the decomposition rates in alkali. Work on the synthesis of esters of pyrophosphoric and triphosphoric acids has been limited to preparations of the nucleotides.Although in these compounds it is the sugar portion of the molecule which is esterified this work will be described only briefly as it is more appropriately included in a review of nucleotides. Methods have been developed which render it possible to eliminate selectively only one of the benzyl residues from the dibenzyl phosphate esters of sugars and nucleotides. If an alcohol of the general formula (XIII) is allowed to react with dibenzyl phosphorochloridate it affords the ester (XIV) which on hydrogenolysis yields a monoester (XV). Selective de- benzylation of compounds of type (XIV) can be accomplished by “ quaterni- sation”-a process depending on the transfer of a benzyl residue from oxygen to nitrogen with formation of a quaternary salt.A strong tertiary base such as 4-methylmorpholine 1O7 is satisfactory but the method has been extended to include all classes of amines. Debenzylation can also be brought about by a base hydrochloride. l08 Lithium chloride in 2-ethoxyethanol proved most efficient and was recommended for the preparation of mono- benzyl esters of the general formula (XVI). An equilibrium is set up between the triester and lithium chloride on the one hand and the lithium salt of the diester and benzyl chloride on the other. Precipitation of this lithium salt from the solution leads to quantitative reaction. In both methods of debenzylation the monobenzyl ester is produced as an anion and therefore a second debenzylation which would produce a doubly charged anion is not favoured.Treatment of the silver salt of the mono- benzyl ester (XVI) with dibenzyl phosphorochloridate gives the tribenzyl ester (XVII) and subsequently by hydrogenolysis the diphosphate (XVIII). Repetition of this sequence of reactions commencing with compound (XVII) yields the tetrabenzyl ester (XXI) and thence the triphosphate (XXII). generally applicable because of the difficulties encountered in the preparation of halogeno-sugar moieties owing to formation of cyclonucleoside salts. An unsymmetrical djester of phosphoric acid [a diribonucleoside phosphate (A)] has been synthesised by this reaction sequence. The silver salt of 2’ 3’-O-isopropylideneadenosine-5’ benzyl phosphate was treated in boiling toluene with 5’-deoxy-5’-iodo-2’ 3’-O-isopropylidene- uridine t o give after removal of the protecting groups adenosine-5’ uridine-5’ phosphate (Elmore and Todd J.1952 3681). lo7 Baddiley Clark Michalski and Todd J. 1949 815. losClark and Todd J . 1950 2023 2030. FOSTER AND OVEREND CARBOHYDRATE PHOSPHATES 77 By such methods Todd and his co-workers log synthesised adenosine di- phosphate (ADP) and adenosine triphosphate (ATP). It might be expected that a mixture of isomers would be obtained from the monodebenzylation. 3 9 R-OH + ~ph~CH~62pOCI - R+P(0.CHph)2 - R-O*P(0H2 (XIII) R; N ow I LlCl ( XV) r R O ~ - ~ - ~ . O . C H p t l (XVII) Phn-0 O*CH,Ph 1 I k . R-08-Of-OH R-08-0- 0 P-OCH2h and R-0.P-0 O Q - P-OCH,Ph H6'- OH HC) b€H2Ph PhCHsd OH ' 9 0 9 % 9,O.P (OCH2P h) R-0.7- 0 -b- 0-7 *O*CH,Ph R-0-P Ph.CH2.0 OCH2Ph (OCHfhL 4 CHph (XXI) I R-O('-O-f-O-F.OH Q O (XXIII) O Q/ & O H ) HO OH OH (XXII) \ t R-0-P 0g(W2 -.\ .A ' $,OH ?/Om 10 (xx I V) R-O-b ox<*.o(xv) For example the substance (XVII) could yield the triester (XIX) or (XX) Whereas further reaction of compound (XX) with dibenzyl phosphoro- chloridate would lead finally to the " unbranched " triphosphate (XXII) the isomer (XIX) would be expected to afford finally the " branched " substance (XXIV). If in compounds (XIX) and (XX) R were adenosine esterified a t position 5' then (XXII) would be natural ATP and (XXIV) an isomer of it. that the disilver salt of adenosine-5' phosphate reacts with an excess of dibenzyl phosphorochloridate to give after debenzylation natural ATP in far better yield than is obtained by the original alternative procedure.log 0 bviously a rearrangement is involved and probably compound (XXIV) is converted into ATP (XXII) via a cyclic intermediate (XXV).(At temperatures of 50" or above benzyl pyrophosphates are rapidly debenzylated by phenol with the production of nuclear-benzylated phenols.ll1 Practically it has been shown lo9Todd and co-workers J . 1947 648; 1949 582. 110 Michelson and Todd J. 1949 2487. ll1 Quoted by Christie Kenner and Todd J. 1954 46. 78 QUARTERLY REVIEWS This is an acid-catalysed reaction whereas anionic debenzylation occurs under neutral or alkaline conditions it is an alternative to hydrogenolysis as a method of debenzylation.) Other methods for the preparation of pyrophosphoric and triphosphoric esters have also been developed. Syntheses of ribonucleoside-5’ phosphites have been achieved.l12 Chlorination of phosphites can be effected with N-chlorosuccinimide and the chloro-derivative is a valuable intermediate for further stages in the synthesis of ribonucleotides.N 2 4-Trichloro- acetanilide can also be used to chlorinate the phosphites and although it is less reactive it might be a useful reagent for the preparation of water- soluble phosphates and pyrophosphates from phosphites since both N 2 4- trichloroacetanilide and 2 4-dichloroacetanilide produced from it are virtually insoluble in water and can be readily separated from the desired reaction products. Uridine-5’ pyrophosphate has been syrithesised in this way as shown in Scheme I. R is uracil and the reagents are (B) 2’ 3’-O-isopropylidene- uridine in acetonitrile containing 2 6-lutidine (C) N-chlorosuccinimide (D) triethylammonium dibenzyl phosphate and (E) lithium chloride hydrogenation and hydrolysis.B s CHjDVH I CH~O~-O-~.OCrCph 1 9 9 CHiO-?-O-?*OH I OCH2Ph OCH2Ph PhCHiO O€H2Ph HO OH Scheme I Esters of pyrophosphoric and triphosphoric acid can be synthesised by reactions of the following types 8 9 9 9 (a) R-CHd + Acf{-O.r-O-P.O.CH,Ph - R-CHiOf)-O-rOCH2Ph PhCHiO OCH,Ph PhCHiO OCH,Ph O Q - R-cH~o-P-o-~ OH HO OH 112 Corby Kenner and Todd J . 1952 3669; Kenner Todd and Weymouth J. 1952 3675. FOSTER AND OVEREND CARBOHYDRATE PHOSPHATES 79 Again the methods have been developed for use in nucleotide syntheses. In reactions of type (a) with a variety of carbohydrate derivatives in which the substituent to be replaced is on the terminal carbon atom it was found that derivatives of open- chain aldehyde-sugars always react readily whereas those containing a lactol ring (e.g.methyl ribofuranoside derivatives) were so sluggish in reaction that they were of little preparative value. Alternative methods of preparing these esters have also been developed. Adenosine monophosphate has been treated with phosphoric acid in the presence of dicyclohexylcarbodi-imide to afford the di- and tri-phosphates 113 the use of protected intermediates is avoided. Employment of carbodi- imides as reagents has proved remarkably effective for the synthesis of symmetrical pyrophosphates and to a smaller degree of unsymmetrical pyrophosphates to which class most of the natural coenzymes belong. Although the method has been applied to the synthesis of inter alia uridine- diphosphate-glucose and the 5'-triphosphates of adenosine and uridine the unsymmetrical esters are always produced as components of complex mixtures with the corresponding symmetrical pyrophosphates.Recently attempts have been made to overcome this difficulty by the use of imidoyl phosphates,l14 which are analogous in structure to the hypothetical inter- mediates in the synthesis of pyrophosphates by use of carbodi-imides and consequently undergo phosphorolysis with the production of pyrophos- phates e.g. (R R' R1 R2 R3 and R4 are suitable protecting residues) No doubt this method will be further exploited. Exchange reactions with trifiuoroacetic anhydride can be used for pyrophosphate syntheses and fully esterified pyrophosphates can also be prepared by exchange reactions between diesters of phosphoric acid and a suitably reactive pyrophosphate.Exchange reactions with nucleosides were less successful than those with simple model compounds. Although cyclic esters of phosphoric acid have been made from glycols and their existence has been postulated as intermediates in various re- arrangements not much work has been done on the synthesis of such esters from simple sugars. Again examples generally must be drawn from nucleo- tide chemistry. Sometimes direct phosphorylation of the sugar moiety leads to a cyclic ester. For example treatment of riboflavin with phos- phoryl chloride in pyridine containing a small amount of water yields a cyclic 4' 5'-phosphate.lf5 An attempt 116 to synthesise a monobenzyl ester of flavin-adenine-dinucleotide consisted in bringing about an exchange llSKhorana J .Amer. Chem. Soc. 1954 76 3517. 114Atherton Morrison Cremlyn Kenner Todd and Webb Chem. and Id. 1955 115Forest and Todd J. 1950 3295. 116 Forest Mason and Todd J. 1952 2630 1183. 80 QUARTERLY REVIEWS reaction between riboflavin-5’ phosphate and 2’ 3’-O-isopropylidene adeno- sine-5’ (benzyl diphenyl pyrophosphate) e.g. (Ad = adenosine residue used as 2’ 3’-O-isopropylidene derivative ; F1 = riboflavin residue.) Prom many reactions in all cases the product was riboflavin-4’ 5‘ cyclic phosphate a compound which could also be obtained by treating ribo- flavin-5’ phosphate with tetraphenyl or tetrabenzyl pyrophosphate in the presence of bases. It may be reasonably assumed that in these reactions the desired exchange did in fact occur and that the pyrophosphate of ribo- flavin initially produced then behaved in the presence of a base as a phos- phorylating agent towards the adjacent hydroxyl group of the riboflavin residue.Riboflavin-5’ phosphate and trifluoroacetic anhydride afford 3 2’ 3’-tristrifluoroacetylriboflavin-4’ 5’ cyclic phosphate.l16 Uridine-diphosphate-glucose on treatment with alkali yields glucose 1 2-(hydrogen phosphate) as a cleavage product.l17 The cyclic 2’ 3’- phosphates derivable from the “ a ” and the “ b ” type of ribonucleotides have been well studied. The cyclic phosphates of this type were prepared by Brown et d118 from adenosine cytidine and uridine. The “ a ” and ‘‘ b ” nucleotides were treated with excess of trifluoroacetic anhydride followed by ethanolic ammonia to remove the trifluoroacetyl residues.There is no doubt that intramolecular reaction occurs as intermolecular reaction would have given diadenosine pyrophosphate. Reaction proceeds by the initial formation of a mixed anhydride of the phosphate with trifluoro- acetic acid and the mixed anhydride can react in intramolecular reaction as a phosphorylating agent towards the adjacent hydroxyl group. Adenylic acid “ a ” or “ b ” with dicyclohexylcarbodi-imide yields the cyclic phos- phate although subsequent opening of the ring may occur in further reactions.ll9 Recently the synthesis was reported of six-membered cyclic phosphates derived from sugars.120 Methyl a-D-ghcoside and phenyl phosphorodi- chloridate afforded a crystalline neutral ester (XXVI) in l0-20% yield O-H& 0 -H2C PhO-9-0 I 0 Me HO-P-0 I QOMr (XXVI) 0 OH 0 OH u<XVlD from which a phenyl group was removed by hydrogenolysis thereby afford- ing methyl ct-D-glucoside 4 &(hydrogen phosphate) (XXVII).From phenyl p-D-glucoside a better yield (40%)) of phenyl P-D-glucoside 4 6-(phenyl 117 Paladini and Leloir Biochem. J. 1952 51 426. 118Brown Magrath and Todd J. 1952 270s. l19Dekker and Khorana J . L4mer. Chern. SOC. 1954 76 3522. 120 Baddiley Ruchanan and Szabb J. 1964 3826. FOSTER AND OVEREND CARBOHYDRATE PHOSPHATES 81 phosphate) was obtained which on hydrogenolysis afforded mainly glucose 4 6-(hydrogen phosphate). Properties and Reactions According to Leloir sugar monophosphates are stronger acids than free phosphoric acid and both pK and pK have smaller values. (Por a dis- cussion of this point see Kumler and Eiler.121) Aldose l-phosphates are very sensitive t o acid and in this respect resemble the glycosides and glycosylamines.Further like the glycosides the p-anomers are usually more acid labile than the or-forms (e.g. in comparative experiments hydrolysis constants for the a- and the @-form of glucose l-phosphate are 5 x and 15 x respectively.loO) Reasons for this difference are probably the same as those put forward for differences in hydrolysis rates of anomeric glycosides.122 The rate of hydrolysis of the glycosidic phosphate residue in a-glucose 1 -phosphate is greater than in cc-glucose 1 6-dipho~phate.~~ The phosphate substituent at C(s) reduces the rate of hydrolysis again probably for the same reasons as in Eydroly- sis of methyl a-D-glucoside methyl 6-deoxy-a-~-glucoside and methyl cc-D-xyloside (XXVIII ; R = CH,*OH Me and H respectively) where the rate increases as the bulk of R diminishes.lz2 Ribose 1 -phosphate is sufficiently acid-labile to undergo hydrolysis at the acidity employed in estimation of phos- phate.The ester is somewhat more acid-labile than The pyranose form of ribose l-phosphate is more stable towards acid than the furanose f0rm.lO3~ %Deoxy-~-ribose 1 -phos- phate is even more acid-labile than the ribose analogue. The mechanism of hydrolysis of aldose l-phosphate has been studied by various workers. The curve of first-order rate coefficient against acidity for the hydrolysis of a-D-glUCOSe l-phosphate is quite different from that obtained for a simple phosphate such as methyl phosphate. At 72.9" and in the range pH 1-4 the logarithm of the rate coefficient is proportional to the pH of the medium.At higher acidities the rate increases more rapidly than the stoicheiometric acidity and at 25" in aqueous perchloric acid the logarithm of the rate coefficient is accurately proportional to Hammett's acidity function H,. Isotope experiments at about pH 4 and in strong perchloric acid showed fission of the carbon-oxygen bond.123 These results are consistent with a single unimolecular mechanism operative over the whole range of acidities studied. The first step must be a rapid and reversible proton-transfer to the a-D-glucose 1 -phosphate followed by a slow reaction not involving a water molecule. There are two possible formulations one of which involves an opening of the hexose ring e.g. sequences (A) and (B).The two mechanisms possibly have different Me phosphocreatine but less so than acetyl phosphate. UXVl II) 121Kumler and Eiler J . Arner. Chem. SOC. 1943 65 2355. 122Foster and Overend Chem. and Id. 1955 566. 123 Barnard Bunton Llewellyn Oldham Silver and Vernon ibid. 1955 760 ; cf. Cohn J . Biol. Chem. 1949 180 771. a2 QUARTERLY REVIEWS stereochemical consequences mechanism (B) necessarily involves the production under kinetic control of the equilibrium mixture of a- and ,&glucose but for mechanism (A) this is not necessarily so. Further informa- tion on this point would be desirable. Since under the experimental con- ditions the mutarotation of glucose to produce the equilibrium mixture is extremely rapid this possible stereochemical distinction has no diagnostic value.In methanol however where the methyl glucosides produced are stable under the experimental conditions study of the steric course of the reaction may throw considerable light on the mechanism. To achieve acidic hydrolysis of the phosphate ester resulting from the esterification of the primary hydroxyl group in a sugar fairly drastic treat- ment is required which may lead to some decomposition of the sugar. Levene and Stiller 63 demonstrated that a pentose esterified at C(3) is hydrolysed more rapidly than the C(5 )-isomer. Thus hydrolysis of 5-0- methyl- 1 2-O-isopropylidenexylose 3-phosphate is many times faster than that of xylose 5-phosphate. Ribose 3-phosphate is hydrolysed 5-9 times faster than the 5-phosphate and 3-phosphoribonic acid is hydrolysed about twice as rapidly as 5-phosphoribonic acid.51 This rate difference has been used to determine whether substances containing a ribose phosphate moiety have the phosphate residue a t position 3 or 5.For the reasons given on p. 66 reservations must be made regarding the rates reported for the ribose phosphates obtained by Levene and his co-workers. An examination 124 of the hydrolysis of fructose 6-phosphate revealed that the hydrolysis is markedly slower than that of the l-phosphate and indeed it is possible to obtain a good yield of fructose 6-phosphate by hydrolysis of the 1 6-diphosphate 125 with hydrochloric or hydrobrornic acid a t 35" under special conditions. Comparison of the rate constants for hydrolysis of fructose 6-phosphate7 -pyrophosphate and -hiphosphate has shown that cleavage of t'he phosphate entity in the first subst'ance is slower by a factor of lo2-lo3 than is hydrolytic cleavage of a single phosphate group from either of the other two compounds.The hydrolysis of hexahydrobenzyl b-glucoside 2-phosphate by O*lN-sulphuric acid a t 100" was followed and it was found that k calculated for a unimolecular reaction increased from 2-9 x 10-5 after 30 minutes to 6.1 x 10-5 after 540 minutes. It might be inferred that the glycoside phosphate is more slowly hydrolysed l Z 4 Friess J . Amer. Chem. Soc. 1952 74 6521. Neuberg Lustig and Rothenberg Arch. Biochem. Biop?tgs. 1943 8 33. FOSTER AND OVEREND CARBOHYDRATE PHOSPHATES 83 than glucose 2-phosphate (k = 8.4 x t o which it will give rise on cleavage of the glycosidic substituent. Glucose 2-phosphate is far less acid-labile than methyl 3 5 6-tri-0-methylglucoside 2-phosphate which is probably at least 80% hydrolysed by O-lN-sulphuric acid after one hour at 100".This difference is probably accounted for at least partly by the difference in lactol-ring forms in the two compounds because generally sugar phosphates without a glycosidic substituent and with a 2-phosphate group are more readily hydrolysed. It has been suggested that this is possibly due to migration of the phosphate residue from position 2 to position 1 but proof has not been presented. Rates of hydrolysis of esters of 2-deoxygalactose have been compared with those for analogous derivatives of galactose.126 Hydrolysis is faster in the 2-deoxy-series. The rates of hydrolysis (N-hydrochloric acid a t 100") of the phosphate groups in glucose 6-phosphate and glucosamine 6-phosphate were compared by Anderson and Per~ival.7~~ Whereas glucosamine 6-phosphate was only 50% hydrolysed during 73 hours glucose 6-phosphate was hydrolysed to the same extent in only 23 hours.A list of acid-hydrolysis constants for some carbohydrate phosphates can be found in the review by Le1oir.l For example D-ribofuranose l-phosphate is completely stable to 06~-sodiurn hydroxide at 80" for one hour.103 On the other hand other sugar phosphates are rapidly changed glucose 2-phosphate is 50% hydrolysed in 97 minutes by 0-1N-alkali at loo" and glucose 6-phosphate is 60% hydrolysed by O.2~-alkali a t 100" in 3 minutes. In the ribose series the decreasing order of stability towards alkali (0.OlN-sodium hydroxide at 22") is ribose 2-phosphate (which is scarcely attacked) 3-phosphate and 5-phosphate.It is reported 127 that phosphate residues are removed with greater difficulty than are sulphate residues in corresponding compounds. Robinson 128 initially suggested that hydrolysis of phosphoric esters is accompanied by Walden inversion and that D-galaCtOSe and D-ribose might arise in hydrolysates from natural products by decomposition of glucose 4-phosphate and xylose 3-phosphate. The alkaline hydrolysis of methyl a-D-glucoside 6-( barium phosphate) methyl D-glucofuranoside 3-(barium phosphate) and isopropylidene-D-glucose 3- and 6-( barium phosphate) was studied by Percival and Percival127 and in no case was any evidence found to support Walden inversion or anhydride formation. Levene et ~ 1 . however have claimed that treatment of fructose 3-phosphate with phenyl- hydrazine in acetic acid results in cleavage of the phosphate group with inversion since the final product is 3 6-anhydroallosazone.Further it is stated 66b that glucose 3-phosphate on treatment with phenylhydrazine also gives this anhydro-compound. On the other hand hydrolysis of glucose 3-phosphate with phosphatase and subsequent osazone formation afforded glucosazone and not allosazone. Sugar 1 -phosphates are resistant to alkali. 126 Foster Overend and Stacey J. 1951 987. 12' Percival and Percival J. 1945 874. 128Robinson Nature 1927 120 44 656. lZB Levene Raymond a d Walti J. Biol. Chrn. 1929 82 191, 84 QUARTERLY REVIEWS Only very brief mention can be made of the effect of phosphatases on glycose phosphates.Using 180 Cohn 123 demonstrated that intestinal alkaline phosphatase ruptures the oxygen-phosphorus bond a change apparently analogous to non-enzymic alkaline hydrolysis of sugar phosphates generally. Likewise prostate acid phosphatase cleaves the same bond. Phosphorylases have been used in experiments designed to determine the anomeric configuration of glycosyl phosphates but care must be exercised in interpreting the results. Changes in optical rotation have also been studied with this end in view (cf. Wolfrom et aZ.100 and Wright and Khorana it appears that assignment of anomeric configuration can be based on Hudson’s rules. A direct approach to this problem was sug- gested by the work of Dekker and Khorana 119 on the reactions of phosphate esters bearing an adjacent cis-hydroxyl function e.g.(XXIX) with dicycb- hexylcarbodi-imide. It was established that these esters give first the cyclic phosphates (XXX) which then form the phosphorylureas (XXXI). This reaction sequence may be followed readily by paper chromatography in suitable solvent systems the mobilities of the reaction products following the order (XXIX) > (XXX) > (XXXI) (examples drawn from the ribo- furanose series). Owing to the more or less planar nature of the furanose ring only the a-phosphate (XXIX) of the two anomeric ribofuranose l-phosphates is able to form a 5-membered cyclic ester and subsequently give rise to (XXXI). The anomeric configurations of synthetic /3- lo3 and enzymically produced a-ribofuranose 1 -phosphate can be assigned on the basis of these reactions. The phosphorylurea (XXXI) was much more stable than either of the samples of ribofuranose l-phosphate.It is likely that the method developed by Smith and his colleagues130 to determine the anomeric configuration of alkyl glycosides would be equally applicable for assignment of configuration in the glycose 1 -phosphate series. Properties of phosphate esters have been exploited in attempts to elucidate structural problems among natural products. To mention one example Brown et aZ. ,131 in experiments directed towards the determination of nucleotide sequence in polyribonucleotides made use of the fact that phosphates of P-aldehydo- and /3-keto-alcohols undergo elimination reactions with alkali. Reference has already been made to migration of phosphate groups and there are several observations in the literature concerning this.Tank6 and Robison 132 suggested that this might explain certain changes in optical 13O Abdel-Akher Cadotte Montgomery Smith Van Cleve and Lewis Nature 1953 131Brown Fried and Todd Chem. and Id. 1953 352; J . 1955 2206 132Tank6 and Robison Biochem. J. 1935 29 961. 171 474. FOSTER AND OVEREND CARBOHYDRATE PHOSPHATES 85 rotation of samples of fructose 6-phosphate which had been subjected to various treatments. Indirect evidence was obtained of phosphoryl migration during mild acid hydrolysis of trehalose phosphate. I n the migrations observed with glycerophosphates 133 and the " a " and " b " purine 134 and pyrimidine l35 nucleotides it is clear that the migration occurs via an inter- mediate cyclic ester. That interaction between neighbouring hydroxyl and phosphoryl groups takes place has been stressed by Kumler and Eiler,l21 who have shown that the polyol and sugar phosphates are abnormally strong acids in comparison with the monoalkyl phosphates.The difference in stability of ribo- and deoxyribo-nucleic acids towards alkali depends on the fact that only the former substance can form an internal cyclic triester. of the rates of oxidation of xylose 5- and 3-phosphate by periodic acid (and the readion of these compounds with dicyclohexyl- carbodi-imide) have led to the conclusion that xylose 3-phosphate exists in solution in the pyranose form (CI conformation) a conclusion which necessitates a re-interpretation of some of Levene and Raymond's s9 results. Marked differences have been observed in the rates of periodate oxidation of some cyclic phosphates methyl cc-D-glucoside 4 6-(phenyl phosphate) is unaffected even by prolonged treatment with periodate and methyl cc-D-glucoside 4 6-(hydrogen phosphate) is oxidised rather slowly but the rate of oxidation is greater for glucose 4 6-(hydrogen phosphate).120 The periodate oxidation of sugar phosphates has been discussed recently by Loring et ~ 1 .1 3 5 ~ A detailed description of the manifold enzymic reactions in which carbo- hydrate phosphates function as substrates is beyond the scope of this Review and only brief mention will be made of a few selected examples. Extensive investigations have established the importance of phosphate esters in carbohydrate metabolism both a t the pentose and hexose level and with the higher saccharides and polysaccharides.A recent development is the presentation of evidence that phosphoric esters of glucosamine and N-acetylglucosamine are concerned in the biosynthesis of mucopolysac- charides.136 As well as this substrate function some sugar phosphates have coenzyme activity e.g. glucose 1 6-diphosphate is a coenzyme for phosphoglucomutase and no interconversion of glucose 1 - and 6-phosphate is effected by this enzyme if the diphosphate is absent from the reaction medium. The r81e of phosphoglycosyl compounds in the biosynthesis of nucleosides and nucleotides has been reviewed by Ka1~kar.l~' Studies 133Verkade Stoppelenburg and Cohen Rec. Trm. chim. 1940 59 886. 134Brown and Todd J. 1952 52. 136 (a) Cohn J . Amer. Chem. Soc. 1950 72 2811 ; ( b ) Loring Levy Moss and las Glaser and Brown Proc. Nat. A d . Sci. U.S.A. 1955 41 253. 13' Kalckar Biochim. Biophys. Acta 1963 12 250. l'loeser J . Arner. Chem. Soc. 1956 78 3724.