Studies directed towards conformationally restricted nucleosides

5-Nitroimidazoles can be transformed into a variety of heterocyclic systems via reduction to the corresponding 5-aminoimidazoles: 1-(β-D-ribofuranosyl)-5-nitroimidazole is a precursor of nucleosides and their analogues. Using this approach, potential routes to conformationally restricted 1-(β-D-ribofuranosyl)-5-nitroimidazole derivatives, which employ radical cyclisation as a key ring forming step, are investigated. Novel radical precursors which, because of the extreme acid sensitivity of 1-(β-D-ribofuranosyl)-5-nitroimidazoles, have to be prepared and manipulated under basic conditions are described and radical cyclisation investigated.


Introduction
In previous studies we have shown that simple 5-aminoimidazoles 1 can be conveniently prepared from 5-nitroimidazoles 2 and that these amines are useful building blocks for the preparation of novel heterocyclic systems, including nucleoside analogues. [1][2][3] Recently we have described a convenient synthetic approach to the biosynthetic purine precursor AIR (aminoimidazole ribonucleotide) via the 5-nitroimidazole 3. 4 The availability of the derivative 3 led us to investigate routes to the conformationally restricted 5-nitroimidazole 4 which is of interest as a potential precursor to nucleoside analogues that are constrained to the anticonformation, 5 which may be the preferred conformation for some enzyme substrates. In particular we have investigated intramolecular cyclisations of the nucleophilic radical 5 on to the 2-position of the electron-deficient imidazole ring. In order to generate the desired radical 5 it was necessary to protect the secondary alcohol functions of the ribofuranose derivative 3 prior to manipulation of the primary alcohol. Since we have previously established 4 that the 5-nitroimidazole 3 is extremely sensitive to the presence of small traces of acid, which catalyses the formation of 4(5)-nitroimidazole 2 (R = H), this functionalisation needed to be carried out under basic conditions. In this paper we describe the preparation of some suitably functionalised novel 5-nitroimidazole derivatives and attempts to achieve the desired cyclisation.

Results and Discussion
Our initial approach to protecting the secondary alcohol functions at positions 2' and 3' was to form a cyclic derivative but all attempts were unsatisfactory. Formation of the acetal 6 using 2,2dimethoxypropane and pyridinium p-toluenesulphonate as catalyst, which in contrast to ptoluene sulphonic acid is reported to be compatible with acid sensitive groups, 6 gave no identifiable products. Attempts to form the cyclic carbonate 7 using trichloromethylchloroformate in dry pyridine were equally unsuccessful. Analyses of the reaction mixture by 1 H NMR and IR spectroscopy suggested that the desired product 7 had formed but only in very low yield. In a final attempt to form a cyclic derivative, compound 3 was treated with triethyl orthoformate in dry THF. After chromatography the desired product 8 (pale yellow syrup; R f 0.33) was obtained as a mixture of diastereoisomers (40% yield) ( δH C1'-H 6.41 and 6.47) but attempts to improve the yield and avoid the formation of alternative products were unsuccessful and this approach was abandoned. Having achieved selective blocking of the 5'-position we next investigated protection of the 2' and 3' hydroxy groups under basic conditions using protecting groups that would be stable under conditions required for removal of the 5'-silylether group. Compound 9 was acetylated using excess acetic anhydride and DMAP and under these conditions we obtained a quantitative yield of the derivative 10 which was isolated as a syrup and used without further purification. The structure and purity of product 10 was fully supported by 1 H NMR spectroscopy including the observation of two singlets at δH 2.11 and 1.99 which were assigned to the 2' and 3' acetoxy substituents.
Treatment of a solution of compound 10 with tetra-n-butylammonium fluoride (TBAF) in THF solution resulted in complete disappearance of starting material (tlc) and formation of a mixture. The major component was isolated using short column chromatography and identified as the primary alcohol 12 (82% yield) which was fully characterised. In principle the preparation of this derivative 12 provides the opportunity to introduce other functional groups at the 5' position of the ribofuranose fragment. However, attempts to scale-up this preparation resulted in much lower yields of the desired alcohol 12 and analysis of the mixture showed that the monoand tri-acetyl derivatives 14 and 15 were present in significant amounts. TBAF appeared to be promoting ester cleavage and also acyl group migration. 8 In an earlier study of protected ribofuranosyl derivatives, Reese et al determined that benzoate esters are less prone to basecatalysed migration and cleavage than acetate groups. 9 We therefore decided to employ more robust benzoyl protecting groups. Reaction of compound 9 with a large excess of benzoic anhydride in pyridine gave the diester 11 in 87% yield. Subsequent removal of the silylether protecting group using TBAF gave a single product that was isolated and identified as the desired alcohol 13 (76%), identical to the sample prepared by the previous route. Since iodo functions are excellent radical triggers we next converted the alcohol 13 into the corresponding iodide. Compound 13 was dissolved in freshly dried pyridine and treated with methylsulphonyl chloride. Work-up and chromatography gave a syrup that was essentially a single product and was identified as the 5'-O-methylsulphonyl derivative 17 (70% yield). The 1 H NMR spectrum showed a singlet at δ 3.15 corresponding to the mesyl ester. In a similar manner the 2',3'-diacetoxy derivative 12 was converted to the mesyl ester 16 (86% yield).

17; R = COPh
PhO S Treatment of the mesyl ester 17 with sodium iodide in hot, dry acetone gave after chromatography a single crystalline product. This was identified as the iodide 19, m.p. 178-180 o C (70% yield) which was fully characterised. In a similar manner the iodide 18 was also obtained as a gummy syrup. However, we were not satisfied with the overall yield of the 2',3'dibenzoyloxy-5'-iodo derivative 19 and an alternative procedure was investigated. In particular we dissolved the alcohol 13, triphenylphosphine and imidazole in acetonitrile-diethyl ether and a solution of iodine in the same solvent was added with stirring. Work-up gave the iodide 19 (60%) which was identical to an authentic sample.
In addition to compound 19, a second derivative containing a different radical trigger was prepared. Thus treatment of the alcohol 13 with phenoxythiocarbonyl chloride and DMAP in dry acetonitrile gave the thiocarbonate derivative 20 which was obtained as colourless needles, m.p. 126-129 o C (71% yield) and fully characterised.
Barton and co-workers 10 have described the preparation of the conformationally restricted nucleoside precursors of adenosine 22 and uridine 24 using carbon radicals generated from the acyclic precursors 21 and 23 (X = TeAn). Earlier workers 11 had made the adenosine analogue 22 by reductive cyclisation of the iodide 21 (X = I) and Matsuda and co-workers have made the cycloderivatives of adenosine and guanosine by photoirradiation of 5'-phenylthio derivatives. 12 It therefore seemed reasonable to attempt the formation of the cyclo-derivative 25 using the radical precursors 19 and 20. Reaction of the iodide 19 with tributyltin hydride (2 equiv.) in the presence of AIBN (0.3 equiv.) in degassed benzene at reflux temperature gave, after chromatography, one major component. This had a well defined 1 H NMR spectrum and appeared to have a molecular weight (FAB) of ca 579 mass units. This product remains unidentified but appears to still contain iodine and it is not the desired product 25. When the thiocarbonate 20 was treated with tributyltin hydride under similar conditions a complex mixture was obtained and no products were identified or isolated. We conclude that the carbon radical generated from the precursors 19 and 20 does not readily cyclise to the desired product 25. We now believe that the 5-nitro substituent interferes with the desired process and is not compatible with the radical methodology.

Experimental Section
General Procedures. All solvents were purified and distilled before use. Acetonitrile was doubly distilled from P 2 O 5 and then stored in a dark bottle over 4Å molecular sieves. THF was ISSN 1424-6376 Page 223 © ARKAT USA, Inc dried by refluxing over sodium until benzophenone held a purple colour. Thin layer chromatography (tlc) was carried out using foil backed alumina or silica plates. Flash column chromatography was carried out using Janssen silica gel (particle size 0.035-0.07mm). Chromatotron chromatography was performed on plates prepared using silica gel 60 PF 254 Melting points were determined on a Kofler melting point microscope apparatus and are uncorrected. Nuclear Magnetic Resonance (NMR) spectra were determined on a Jeol GSX 270 MHz FT spectrometer. Unless otherwise stated, NMR spectra were run in CDCl 3 or d 6 -DMSO. Microanalyses were obtained on a Perkin Elmer 240 elemental analyser. Mass spectra were obtained on a AEI MS12 mass spectrometer at 70 eV. High resolution and Fast Atom Bombardment (FAB) mass spectra were recorded by the EPSRC Mass Spectrometry Service (Swansea). Infrared (IR) absorption spectra were recorded using either a Perkin-Elmer 881 or Perkin-Elmer Paragon FT-IR spectrophotometer: unless otherwise stated all solid samples were incorporated into KBr discs and liquid samples were dissolved in CHCl 3 . Spectroscopic details were recorded for all compounds made; in those cases where a compound was obtained in more than one reaction and where starting materials were recovered, identification was confirmed by m.p. and IR.