Synthesis of novel nucleoside–carbohydrate hybrids

Abstract

A substitution approach for the synthesis of 8-(mannosyloxy)adenosines, novel nucleoside–carbohydrate hybrids found in a new family of chitin synthase inhibitor—guanofosfocins, has been investigated. When an N⁶-benzoyl-8-bromoadenosine derivative was exposed to the sodium alkoxides derived from di-O-isopropylidene-D-mannoses or cyclohexanols, substitution occurred at room temperature to yield 8-(mannosyloxy)-adenosines and their 8-cyclohexyloxy derivatives. The 8-mannosyloxy products were isolated solely as α anomers and adopted a syn conformation with respect to the anomeric C–N bond. Whilst the hybrid compounds were obtainable in good yields, the new glycosidic linkage was acid sensitive to furnish 8-oxoderivatives under deprotecting conditions for the dimethoxytritylgroup. The synthesis of structurally unique nucleoside analogs is of considerable importance in the development of antiviral, antibiotic, antitumor and antifungal agents Naturally occurring compounds often provide inspiration for the creation of such analogs and a recently isolated family of chitin synthase inhibitor, guanofosfocins, stimulated our interest. The guanofosfocins, isolated from the fermentation broths of Streptomycessp. and Trichodermasp., have a unique cyclic structure containing a glycosidic bond between the 8-position of guanosine and a D-mannose moiety (Fig. 1) Despite their promising therapeutic effects against fungal diseases, further investigation of these fascinating molecules has been hindered by their low stability Since the ethereal bond between a glycosyl moiety and the 8-position of a purine nucleoside found in guanofosfocin is previously unknown, we planned to develop methodology for the efficient construction of this linkage in order to synthesise novel nucleoside–carbohydrate hybrids for screening and to evaluate their stability.

One possible strategy for the synthesis of an 8-(glycosyloxy)purine nucleoside would be to substitute a protected carbohydrate-type nucleophile into a purine substrate bearing a leaving group at the 8-position. To date, simple 8-alkoxypurine nucleosides (8-methoxy and 8-benzyloxy substituted derivatives) have been prepared by such a method–6 However, the reaction conditions (e.g. non-protected nucleosides are treated with excess alcohol and sodium metal in DMSO at 65°C or higher) are unsuitable for sensitive carbohydrate-type nucleophiles. Therefore, prior to the introduction of a carbohydrate moiety, we endeavored to establish a general substitution procedure using an easily accessible purine nucleoside substrate, 8-bromoadenosine, and a simple secondary alcohol, cyclohexanol.

As the substitution reaction conditions will be basic, the protecting groups employed on 8-bromoadenosine must be base stable. We first prepared a 2′,3′-O-isopropylidene-5′-O,N⁶-ditrityl substrate 1a, which was treated with sodium cyclohexyloxide generated insitu from cyclohexanol and sodium hydride in DMF.‡ Whilst no reaction occurred at room temperature, on heating the mixture at 70°C the reaction proceeded and, after 24 h, the starting nucleoside was almost consumed to provide 8-(cyclohexyloxy)adenosine derivative 2a in 65% yield along with a 22% yield of 8-oxoadenosine 4 (run 1 in Table 1). A similar tendency was observed in the reaction of N⁶ free substrate 1b (run 2). By contrast, the reaction of N⁶-benzoylated substrate 1c was completed at room temperature in 4 h to afford 8-cyclohexyloxy product 2c in 75% yield (run 3). It has been reported that an N²-acyl group on guanosine derivatives enhances the rate of intramolecular cyclization at the 8-position The benzoyl group at N⁶ of adenosine increases the rate of substitution by a similar electronic effect.

The formation of 8-oxo byproduct 4 is due to hydrolysis of an addition intermediate 3 (Fig. 2) as attempts to form 4 directly by treating the substrates with sodium hydride alone in wet DMF were unsuccessful. Prolonged reaction, however, didn't lead to improvement in the yield of 2.

Since selective deprotection at the 5′-position is required for further studies, other protecting groups at this position were briefly examined. 5′-O-tert-Butyldiphenylsilyl-protected substrate 1d underwent desilylation under the reaction conditions to yield 5′-hydroxyl product 2d (run 4). 5′-O-Dimethoxytrityl (DMTr) derivative 1e, in which the DMTr group can be removed more readily than the Tr group reacted with cyclohexanol to furnish the desired product 2e in good yield (run 5).

Several other base–solvent combinations were also investigated (DMSO as an alternative solvent and Bu^tOK/18-crown-6, Et₃N as an alternative base), but sodium hydride in DMF gave the most satisfactory results.

With a suitable substrate in hand, we turned our attention to the substitution reaction of carbohydrate-type nucleophiles (Table 2). When 2,3,4,6-tetra-O-benzylmannopyranose (5) was allowed to react with 1e, 8-(benzyloxy)adenosine 9 was isolated (run 1). From this result, it was deduced that base catalyzed ring opening of 5 was occurring and subsequent β-elimination of benzyloxide resulting in the formation of the 8-benzyloxy product. We anticipated that protection of the 2- and 3-hydroxy groups as a cyclic acetal would restrain this process, allowing the desired reaction to proceed. Indeed, when 2,3;4,6-di-O-isopropylidene-D-mannopyranose (6), prepared by a known route was used as a nucleophile, 8-(mannopyranosyloxy)adenosine derivative 11 was successfully obtained in good yield (run 2). Similarly, the corresponding 8-(mannofuranosyloxy)adenosine derivative 12 and the 8-(mannopyranosylthio)adenosine derivative 13 have been prepared using 2,3;5,6-di-O-isopropylidene-D-mannofuranose (7) and 2,3;4,6-di-O-isopropylidene-1-thio-D-mannopyranose (8), respectively (runs 3 and 4).

The mannose–adenosine hybrids obtained here were formed as single isomers. Generally, the configuration of glycosidic linkages in pyranoses can be determined from ¹³C NMR spectra by measuring the ¹J_CH coupling at C-1 In hybrid 11, the coupling constant is 178 Hz, showing the new linkage to be exclusively α^10,11 which is consistent with the mannosyl bond in guanofosfocins.

In addition to the new carbohydrate–nucleoside hybrids, a substrate derived form myo-inositol was employed in the reaction to create a novel 8-inositol substituted adenosine compound 14 (run 5).

Next, we briefly examined deprotection at the 5′-position. Dimethoxyltrityl group can be removed by trichloroacetic acid in nitromethane–methanol solvent system with little or no concomitant depurination However, when these conditions were applied to the nucleoside–carbohydrate hybrids 11–13, the glycosidic bond was cleaved to give the 8-oxo or 8-thioxo product 17 (runs 1–3 in Table 3). The 8-cyclohexyloxy derivatives 2e and 14 were acid stable, so that the dimethoxytrityl group could be removed without difficulty (runs 4 and 5). Efforts to investigate the selective deprotection conditions, not affecting the glycosidic bond, are currently underway.

One intriguing feature of the guanofosfocins (Fig. 1) is that they clearly adopt an anti conformation about the anomeric C–N bond though purine nucleosides with bulky substituent at C-8 are known to have a syn conformation–15 It is well documented that the chemical shifts of H(2′) protons in syn and anti adenosines differ through changes in the torsion angle around the glycoside bond–15 It is therefore possible to predict the glycosyl conformation of the hybrids by comparing their shifts with compounds of known orientation. 8-Bromoadenosines are known to adopt syn conformations with the H(2′) signal characteristically located 0.5–0.6 ppm downfield from the H(3′) signal In this respect, hybrid compounds 11 and 12 show patterns similar to 1e, indicating a syn orientation. The 8-(cyclohexyloxy)adenosine 2e also shows a typical syn pattern. However, on removal of the bulky dimethoxytrityl group, the H(2′) signal of compound 15 shifted upfield to give a ¹H NMR spectra consistent with an anti relationship. This finding is significantly important because an anti relationship is beneficial for future attempts to prepare guanofosfocin type compounds.

In summary, by successfully forming the purine–carbohydrate bond found in guanofosfocin we have prepared a new type of nucleoside–carbohydrate hybrid. The approach will be compatible with the use of a variety of nucleophiles. The glycosidic linkage is extremely acid labile and this may contribute to the instability of the natural products. Further study on the application of this approach to the preparation of a wide variety of nucleoside hybrid molecules, including the use of guanosine substrates, is in progress. Acknowledgement