Synthesis of-D-ribofuranosylbenzazoles using nitrile oxide addition chemistry

A nitrile oxide based route to 2--D-ribofuranosylbenzazoles has been developed. Tri-Obenzoyl--D-ribofuranosylformonitrile oxide (14) was generated from the corresponding carbaldoxime 16 by treatment with NCS/pyridine, followed by base-induced dehydrochlorination of the resulting hydroximoyl chloride. Reaction of the nitrile oxide with 1,2-diaminobenzene afforded 2-(tri-O-benzoyl--D-ribofuranosyl)benzimidazole (21), from which 2-(-Dribofuranosyl)benzimidazole (22) was prepared by treatment with Et3N/MeOH. 2-Aminophenol reacted similarly to yield 2-(tri-O-benzoyl--D-ribofuranosyl)benzoxazole (18). In the absence of a co-reactant dimerisation of the nitrile oxide afforded 3,4-di(tri-O-benzoyl--Dribofuranosyl)-1,2,5-oxadiazole-2-oxide (17). The carbaldoxime starting material 16 was prepared from tri-O-benzoyl--D-ribofuranosyl cyanide by reaction with semicarbazide to form the semicarbazone, followed by transimination with hydroxylamine.


Introduction
Nucleoside analogues (C-nucleosides) in which the ribose unit is attached to a carbon atom of a heterocycle have attracted much attention, both from a synthetic perspective and on account of their potential biological activities. 1 Two general approaches have been employed for their synthesis: introduction of an intact heterocycle at the anomeric centre of the furanose, and creation of the heterocycle at that position, for example via cycloaddition reactions.3][4][5] A key step in the preparation by Albrecht, Repke and Moffat of 3-(-D-ribofuranosyl)isoxazole-5carboxamide (1) was the cycloaddition of nitrile oxide 2 to ethyl propiolate. 2 The Kozikowski group used nitrile oxide − isoxazoline chemistry 6 to synthesise various C-nucleoside analogues, e.g. the ribofuranosyl pyrazole 3 and the indole compound 4; 3,4 and the same method was used by Baraldi et al. for the pyridazinone 5. 5 Whereas the cycloaddition reactions of nitrile oxides have been extensively used over many years for heterocycle synthesis, their 1,3-nucleophilic addition reactions have received less attention. 7We have, however, recently established that nitrile oxide addition chemistry can be used to prepare 2-pyranosylbenzazoles (benzimidazoles 6, benzoxazoles 7, benzothiazoles 8), 8 and we anticipated that a similar approach might be used for their furanosyl analogues, as outlined in Scheme 1.
Such furanosylbenzazoles show a range of useful biological activities.They are of interest as antiviral compounds, 9,10 and have been identified as potential therapeutic agents for the treatment of diseases such as cystic fibrosis. 11In particular it has been reported that they show excellent agonist potency and selectivity for the P2Y2 receptor, and that they also have enhanced metabolic stability compared with N-linked analogues and the natural ligand ATP. 11Methods for their synthesis have included addition of 2-lithiated benzimidazoles/benzothiazoles to sugar lactones and deoxygenation of the resulting hemiacetals, [12][13][14] and reaction of 2-substituted anilines with aldonic acids, 11,15 and with acetimidates/thioacetimidates derived from furanosyl cyanides. 16Forcing conditions were sometimes required, mixtures of isomers formed and overall yields were variable.Alternative approaches are therefore desirable, particularly for the benzoxazoles and benzimidazoles.We selected nitrile oxide addition chemistry as it had the potential to provide furanosyl benzimidazoles, benzoxazoles and benzothiazoles all from the same source.

Results and Discussion
In order to test the effectiveness of the method we selected the known -D-ribose-derived nitrile oxide 14 2 as a representative example that should provide access to a range of benzazole Cnucleoside derivatives.As most nitrile oxides are short-lived compounds that readily dimerise to 1,2,5-oxadiazole N-oxides (furoxans), it is common practice in synthetic applications for the nitrile oxides to be generated in situ in the presence of the co-reactant.The two main methods, both of which have been used for carbohydrate synthesis, 17 involve either base-induced dehydrohalogenation of hydroximoyl halides (usually the chloride) or dehydration of nitromethyl compounds. 6,18In the present case the shortest route is conversion of D-ribose into the nitromethyl compound 11 via initial reaction condensation with nitromethane (the 'Fischer-Sowden reaction'), followed by Mukaiyama-type dehydration of 11 with an isocyanate, e.g.PhNCO 19 or tolylene diisocyanate. 20However, in our hands and those of others, 21,22 the literature route to the nitromethylribose afforded a mixture of the and -isomers which were time-consuming to separate and gave only a moderate yield of the target -product.We therefore turned to the longer but ultimately more reliable route via the cyanide 9, the oxime 16 and the hydroximoyl chloride 15, as originally developed by the Moffat group. 2 Of the various literature methods for the preparation of the starting material we chose that reported by Morelli et al., 23 involving reaction of tri-O-benzoyl--D-ribofuranosyl acetate with trimethylsilyl cyanide in the presence of catalytic boron trifluoride etherate, and this afforded the cyanide 9 in 86% yield.As direct hydrolysis of 9 to the aldehyde 13, and thence the carbaldoxime 16, was reported to be problematic, the cyanide 9 was first converted into the imidazolidine derivative 10 (59%) by reaction with 1,2-dianilinoethane (Wanzlick's reagent) and sodium hypophosphite in acetic acid and pyridine.Treatment of the product with tosylic acid afforded the crude aldehyde 13, which was not isolated but reacted immediately with hydroxylamine-pyridine to afford the required aldoxime 16 (45%).Although the oxime had thus been obtained by the Moffat group route, the procedure was not considered satisfactory for larger scale synthesis.Toth and Somsak have reported 24 an alternative method which they used to convert pyranosyl cyanides into pyranosyl aldoximes via the corresponding semicarbazones, and we considered that this approach should also be applicable for their furanosyl counterparts.This proved to be the case.Reaction of the ribofuranosyl cyanide 9 with semicarbazide afforded the semicarbazone 12 (85%), which was then converted to the aldoxime 16 (81%) as a mixture of Eand Z-isomers by transimination with hydroxylamine.This latter route thus provided the oxime in 69% overall yield from the cyanide, compared with 27% via the imidazolidine.Attempts to convert the aldoxime 16 into the hydroximoyl chloride 15 with chlorine in diethyl ether, using the technique that we had previously used for pyranosyl hydroximoyl chlorides, 25 were only partially successful and the product was not isolated in pure form.A milder chlorination procedure, originally reported by Larsen and Torssell, 26 employing Nchlorosuccinimide in pyridine was therefore investigated.The hydroximoyl chloride was not isolated; instead the nitrile oxide was generated from the crude product by addition of triethylamine.The reaction was first carried out in the absence of a co-reactant in order to provide access to 3,4-di(tri-O-benzoyl--D-ribofuranosyl)furoxan (17) (72%), which is of interest both in its own right and is also a likely by-product of reactions involving the nitrile oxide (Scheme 3).The product was identified from its spectroscopic properties.In the 13 C NMR spectrum in CDCl3 there are, in addition to the expected resonances for the two tri-Obenzoylribofuranose units, characteristic peaks at 112.6 (C-3) and 155.0 ppm (C-4) for the carbon atoms of the 1,2,5-oxadiazole ring.These values are typical for unstrained furoxans. 27In the proton NMR spectrum the two non-equivalent furanosyl substituents give rise to overlapping and only partially resolved signals.There are, however, distinct doublets at 5.46 (H-1) and 5.37 ppm (H-1´) for the protons bonded to the carbons adjacent to C-3 and C-4, respectively, of the furoxan ring; these assignments were made by comparison with data reported for other dialkylfuroxans. 28

Scheme 3
Having isolated the furoxan 17 in good yield, and thus established that nitrile oxide 14 could be generated efficiently from the oxime 16 in a one-pot procedure, the synthesis of ribofuranosyl benzazoles was investigated.As an efficient route from glycosyl cyanides to glycosyl benzothiazoles is already well established, 29 we concentrated our attention on the benzimidazoles and benzoxazoles.To obtain the benzoxazole 18 nitrile oxide 14 was reacted with 2aminophenol.A mixture of the oxime 16 (0.30mmol), N-chlorosuccinimide and pyridine in chloroform was heated at 40 °C for 45 minutes and, after cooling and removal of the solvent, 2aminophenol (0.75 mmol) in ethanol was added and the solution heated at reflux for five hours.From the reaction mixture was isolated 2-(tri-O-benzoyl--D-ribofuranosyl)benzoxazole (18) (90%), which was identified from its spectroscopic properties.The carbon NMR spectrum in CDCl3 showed characteristic signals for the benzoxazole moiety [162.5 (C-2), 152.0 (C-7a), 141.7 (C-3a), 126.9 (C-5), 125.7 (C-6), 121.7 (C-4) and 112.0 ppm (C-7)], which were very similar to those previously reported for 2-pyranosylbenzoxazoles 7, 8 and are typical of 2-alkylsubstituted benzoxazoles. 30The presence of the perbenzoylribofuranose unit was evident in both the 13 C and 1 H NMR spectra.The reaction pathway is believed to proceed by initial dehydrochlorination of the hydroximoyl chloride 15 by one equivalent of the 2-aminophenol to generate the nitrile oxide 14, followed by nucleophilic addition of the anilino group of a second equivalent of 2-aminophenol to form the amidoxime 20; nucleophilic displacement of hydroxylamine by the phenolic group then leads to the benzoxazole, as illustrated in Scheme 3.
The approach used to prepare the ribofuranosylbenzimidazole was similar to that described above for the benzoxazole, but using 1,2-diaminobenzene (DAB) instead of 2-aminophenol as the nucleophilic co-reactant.Reaction of the nitrile oxide 14 with DAB (1:2.5 molar ratio) in ethanol at reflux for five hours afforded, after work-up, the per-benzoylated -D-ribofuranosylbenzimidazole 21 in 92% yield.Work-up of the reaction mixture was aided by washing with 4% aq.CuSO4, which removed excess DAB as its copper complex, 31 thus reducing time-consuming chromatographic separations.The 1 H and 13 C NMR spectra in CDCl3 showed the presence of both the ribofuranose and benzimidazole units.The proton spectrum had the expected benzimidazole peaks at 7.90−7.94(H-4,7) and 7.23−7.27ppm (H-5,6).In the carbon spectrum the signals for benzamidazole unit were similar to those reported for the pyranosyl analogues 6. 8 There were two distinct peaks at 151.3 ppm and 123.2 ppm for C-2 and C-5/6 respectively, with the other carbons giving broad signals.This effect has been reported previously for imidazoles and benzimidazoles, 8,32 and is attributed to rapid proton exchange between N-1 and N-3.Debenzoylation using triethylamine/methanol yielded the parent 2--D-ribofuranosylbenzimidazole ( 22) (91%).The overall yield from the oxime 16 to the benzimidazole 22 was 82%.In conclusion, these results also show that nitrile oxide addition chemistry can provide an effective alternative to current synthetic routes to furanosylbenzazoles.For example, 2-(-Dribofuranosyl)benzimidazole can be prepared in four steps from the corresponding commercially-available cyanide in 58% overall yield.The same approach should also be applicable for the synthesis of 2-furanosylbenzothiazoles.An improved procedure for the conversion of furanosyl cyanides into furanosyl carbaldoximes has also been established.