Glyconothio‐O‐lactones. Part I. Preparation and reactions with nucleophiles

Abstract

AbstractFuranoid and pyranoid glyconothio‐O‐lactones were prepared by photolysis of S‐phenacyl thioglycosides or by thermolysis of S‐glycosyl thiosulfinates, which gave better results than the thionation of glyconolactones with Lawesson's reagent. Thermolysis of the thiosulfinates obtained from the dimannofuranosyl disulfide 7 or the manofuranosyl methly disulfide 8 (Scheme 2) gave low yields of the thio‐O‐lactone 2. However, photolysis of the S‐phenacyl thioglycoside 6 obtained by in situ alkylation of the thiolato anion derived from 5 led in 78–89% to 2. Similarly, the dithiocarbonate 10 was transformed, via 11a, into the ribo‐thio‐O‐lactone 12 (79%). Thermolysis of the peracetylated thiosulfinates 14 (Scheme 3) led to the intermediate thio‐O‐lactone 15, which underwent facile β‐elimination of AcOH (→ 16, 75%) during chromatography. The perbenzylated S‐glucopyranosyl dithiocarbonate 18 (Scheme 4) was transformed either into the S‐phenacyl thioglucoside 19 or into a mixture of the anomeric methyl disulfides 21a/b. Whereas the photolysis of 19 led in moderate yield to 2‐deoxy‐thio‐O‐lactone 20, oxidation of 21b and thermolysis of resulting thiosulfinates gave the thio‐O‐lactone 4 (79%), which was transformed into 20 (36%) upon photolysis. The pyranoid manno‐thio‐O‐lactone 26 was prepared in the same way and in good yields from 22 via the dithiocarbonate 24b and the disulfide 25. The ring conformations of the δ‐thio‐O‐lactones, flattened 4C1 for 15 and 4 and B2,5 for 26, are similar to the ones of the O‐analogous oxo‐glyconolactones. The reaction of 2 (Scheme 5) with MeLi and then with MeI gave the thioglycoside 27 (29%) and the dimeric thio‐O‐lactone 29 (47%). The analogous treatment of 2 with lithium dimethylcuprate (LiCuMe2) and MeI led to a 4:1 mixture (47%) of 31 and 27. The structure of 2 was proven by an X‐ray analysis, and the configuration at C(6) and C(5) of 29 was deduced from NOE experiments. Substitution of MeI by CD3I led to the CD3S analogues of 27, 29, and 31, i.e. 28, 30, and 32, respectively, evidencing carbophilic addition and ‘exo’‐attack on 2 by MeLi and the enethiolato anion derived from 2. The preferred ‘endo’‐attack of LiCuMe2 is rationalized by postulating a single‐electron transfer and a diastereoselective pyramidalization of the intermediate radical anion.