A synthesis of 1 ,2-cis-configurated, non-isosteric phosphonate analogues of aldose-I-phosphates is described. Treatmciit of 1-0-acyl-glycoses 1, 7, 13, and 19 with trialkyl phosphitc in the presence of trirnethylsilyl trifluoromethanesulfonate gave the 1,2-ci.s-configurated glycosylphosphonates 2,4,8, 10, 14, 16, 20, and 22 as the major anomers and the 1,2-~rons-contigurated glycosylphosphonates 3, 5 , 9 , I I , 15, 17,21, and 23 as the minor anomers. The 1,2-cis-configurated phosphonates 4, 10, 16, and 22 were deprotected to give the (a-u-glucopyranosy1)phosphonate 6 , the (/?-mnannopyranosy1)phosphonate 12, the (a-D-ribofu~dnosyl)phosphonate 18, and the (/I-o-arabinofuranosy1)phosphonate 24, respectively, in high yields. The preferred formation of 1,2-cis-configurated phosphonates is explained by postulating an equilibrium between the anomeric phosphonium-salt intermediates (such as 25 and 26) and a stabilization of the cis-contigurdted salts through formation of a pentacoordinated species (such as 28).Introduction. -There is no general method for the synthesis of non-isosteric but isopolar glycosylphosphonate analogues of aldose-1 -phosphates, i.e. compounds carrying a phosphono group at the anomeric center such as 6 (Scheme 1 ) . Paulsen [l] has described some derivatives of these compounds, which could not be transformed into the desired analogues, and the scope of our multistep synthesis of the u-D-mannofuranosy1)phosphonate seems to be limited [2], so that an evaluation of the biological properties of these phosphate analogues has not been possible so far.We now report a short and efficient route to glycosylphosphonates. Our scheme is based on the known reactions of carbocations with trialkyl phosphites leading to dialkyl phosphonates [3] [4]. In a similar way, 1-0-acyl-glycoses should react via the corresponding oxonium ions to give glycosylphosphonates. This was checked by treating benzylated 1-0-acetyl-glycoses with P(OMe), and P(OEt), in the presence of a suitable catalyst. The choice of the glycoses was dictated both by the importance of their phosphates (glucose and ribose) and by the desire to probe the influence of the configuration at C(2) (mannose and arabinose) on the stereochemical outcome of the phosphonoylation. The benzyl group is known as a 'non-participating' and easily removable protective group in the synthesis of glycosides [S]. Preliminary experiments established the advantage of trimethylsilyl trifluoromethanesulfonate [6-81 as catalyst in combination with a C( 1) acetoxy group in obtaining high yields of the desired phosphonates.Results and Discussion. -Treatment of a mixture of the glucopyranoses 1 (a/p = 4:l)[9] (Scheme 1 and Table I ) with 1.5 equiv. of P(OMe), and 1.2 equiv. of trimethylsilyl trifluoromethanesulfonate in CH,CI, at room temperature gave the 1,2-cis-configurated dimethyl (a-D-glucopyranosy1)phosphonate 2 (92%) and its anomer 3 ( 5 %). Under similar conditions, 1 reacted with P(OEt), to yield the diethyl phosphonates 4 (93%) and

SummaryThe I-C-nitroglycosyl chloride 1 reacted with the anions from 2-nitropropane, nitromethane, and diethyl malonate, to give the chain-extended products 2 (81 %), 5 (72%), and 6 (83%0), respectively. Treatment of the 1-C-nitroglycosyl bromide 7 by the lithium salt obtained from 8 gave the dodecodiulose derivative 9 (76%). The /?+-configuration of 2 and 9 was inferred from their NMR and CD spectra. Treatment of 2 and 9 with sodium sulfide gave the enol ethers 3 (96 %) and 10 (92 %), respectively. The (2)-configuration of 10 was deduced from the configuration of its hydrogenation product 11.

Summary 2,3-O-Cyclohexylidene-~-erythruronic acid (6) available in 83% yield from D-ribonolactone (7), was treated with phenylmagnesium bromide to give the D-rib0 and L-lyxo derivatives 10 and 11 in high yields (Scheme I and 2). The diastereoselectivity depended on the temperature and mode of operation ( Table 1). The absolute configuration of 10 and 11 was determined by correlation with (R)-and ( S ) -phenylethanediol (17 and 16) respectively, excluding intramolecular hydride shifts during formation of 10 and 11. Reaction of 6 with methoxymethoxyphenyllithium gave the lactones 18 and 19. The ~-l y x o isomer 19 was transformed in high yields into the D-rib0 lactone 18. Compound 10 was transformed into the adenosine analoge 24 by reduction with Diisobutylaluminium hydride, hydrolysis, acetylation and nucleoside synthesis according to Vorbruggen (Scheme 3). Its structure was deduced from its UV., NMR. and CD. data and from those of the isopropylidene derivative 25. Similarly, 18 was transformed into the adenosine analog 29 and into the isopropylidene derivative 30.

Treatment of the chloro-nitro-ribofuranose 7 with KPO(OMe)2 gave the 0-amino phosphate 8 (5%) and the nitrile 9 (62%). Compound 9 was also obtained by the reaction of 8 with KPO(OMe),, and its structure was established by X-ray analysis. Treatment of the chloro-nitro-mannofuranose 10, the bromo-nitro-ribofuranose 14, or the bromo-nitro-mannofuranose 16, respectively, with the K or Na salt of HPO(OMe)2 lead also to 0-amino phosphates and nitriles. The (1-C-nitrog1ycosyl)phosphonate 22 was obtained (21 %) together with the nitrile 21 (5 1 %) from the chloro-nitro-mannofuranose 10 and KPO(OEt)2. The reaction of the I-C-nitroglycosyl sulfone 25 (NO2-group endo) with KPO(OEt), gave the (I-C-nitroglycosy1)phosphonate 22 (61 %) and the nitrile 21 (1 1 YO), whilst the anomeric sulfone 26 (NO,-group exo) gave 22 (15%) and 21 (58%). In the presence of [18]crown-6, a mixture of the anomers 25 and 26 gave the (I-C-nitroglycosy1)phosphonate 22 in 67% yield together with 21 (13%). These findings are rationalized as the result of a competition between a nucleophilic attack of the dialkyl-phosphite anions on the NO2-group leading ultimately to the nitrile 21 and a single-electron transfer reaction leading to the (1-C-nitroglycosy1)phosphonate 22.

SummaryPartially protected 4-or 5-hydroxy-sugar oxinies were transformed into 5-or 6-membered 1-C-nitroglycosyl chlorides, respectively, by reaction with NaOCl under phase-transfer conditions. With the exception of the oxidation of the gluco-derivative 1 giving the anomers 6 and 7, the reactions were completely diastereoselective. We report a convenient preparation of protected 1-C-nitroglycosyl chlorides directly from the corresponding sugar oximes which should facilitate investigations of the reactivity of halonitro ethers. The oxidation of 4-or 5-hydroxy-sugar oximes to lactone oximes very probably occurs via the tautomeric cyclic hydroxylamines. It should proceed easily also with NaOCl as oxidant. Since Corey & Estreicher [4] have described a two-step one-pot transformation of oximes into chloronitro compounds by oxidation first with HOCl in a biphasic system at pH 5.5 and then with NaOCl in a biphasic system at pH 10 in the presence of a phase-transfer catalyst, a direct oxidation of appropriate hydroxy oximes to chloronitro ethers appeared feasible. In the event, treatment of the D-gluco-and D-manno-configurated 5-hydroxy oximes 1 and 2, respectively, with NaOCl in a biphasic system at pH 11-12 in the presence of a phase-transfer catalyst gave the corresponding chloronitro ethers 6-8 in a good yield (Scheme). The 4-hydroxy oximes 3 and 4 were transformed by a similar treatment first at a pH of ca. 7 and then at pH 11-12 into the chloronitro ethers 9 and 10, respectively, while the dihydroxy oxime 5 gave the novel nitroacetal 11 by double neighbouring-group participation. With the exception of the oxidation of the D-gluco oxime 1 giving a 1 : 1 mixture of 6 and 7, the reactions were completely diastereoselective. The anomeric configura-') 7th Communication: [l]