Synthesis of 5-O-benzyl-2-C-β -fluoromethyl-1,2,3-tri-O-acetyl-D-ribofuranose

The novel, protected fluoromethylribose ( 10 ) was prepared in 7 steps (26% overall yield) from commercially available D -ribonolactone. First, the three hydroxyl groups were protected as the 2,3-isopropylidene-5-benzyl derivative. Reduction of the resulting fully protected ribonolactone to the lactol was achieved by using Cp 2 TiF 2 -catalysed hydrosilylation, followed by hydrolysis. Reaction with formaldehyde installed the 2-C - β -hydroxymethyl group. Treatment with DAST gave the 1-fluoro-2-C - β -fluoromethyl derivative, which, on hydrolysis and acetylation, afforded 5-O -benzyl-2- C - β -fluoromethyl-1,2,3-tri-O -acetyl- D -ribofuranose.


Introduction
The introduction of fluorine atoms into molecules may change biological activity, vulnerability towards metabolism, lipophilicity and pKa with minimal change in steric bulk, and is therefore of particular interest to the pharmaceutical industry. 1For example, fluorinated carbohydrates have been investigated for their biological properties, and as tools to study enzyme-carbohydrate interactions. 2tarting with the pioneering work of Walton, 3 who prepared 2´-C-methyladenosine (an inhibitor of adenosine demethylase and of KB cells in culture) and 2´-C-methylcytidine (1), (an inhibitor of hepatitis C virus RNA polymerase), much interest has been shown in the synthesis of modified nucleosides as anti-tumour and anti-viral agents. 4It is not surprising therefore that the fluorinated modifications of nucleosides have also been investigated. 5For example, Piccirilli and co-workers have recently reported the synthesis of 2´-C-β-trifluoromethylribonucleosides (2). 6In the particular case of 2´-C-methyl nucleosides bearing a single fluorine substituent, however, the only successful synthetic strategy so far has been to introduce the fluoromethyl group after the nucleoside base (Scheme 1), but these approaches suffer from moderate yields and lack the flexibility to incorporate the nucleoside base at a late stage in the synthesis.In the first example, ring-opening of epoxide (3) under harsh conditions gave 2´-C-α-fluoromethyl derivative (4), which was converted to 2´-C-β-fluoromethyl-2-deoxycytidine in six steps. 7The fluorine was introduced similarly in the second example, and three further steps were required to afford 2´-C-β-fluoromethyluridine (5). 8wo patents 4c,9 describe remarkably mild fluoride displacement of tosylate (6) and triflate (7) to give compounds (8) and (9), respectively.

Results and Discussion
Our synthesis commenced with the protection of the three hydroxyl groups of commercially available (D)-ribonolactone (11) (Scheme 3).Thus, treatment of 11 with acetone in the presence of anhydrous cupric sulphate gave acetonide (12) in almost quantitative yield. 10 Benzyl ether (13) has been previously described. 11In our own hands, benzylation proceeded more cleanly and rapidly with sodium hydride and benzyl bromide in dimethylformamide rather than tetrahydrofuran solvent.
We then sought a convenient method to reduce 13 to lactol (14).Although this can be achieved with diisobutylaluminium hydride at low temperature, precedent suggested a significant excess of reagent would be required, which is disadvantageous when working on a large scale. 12e were therefore drawn to Buchwald's hydrosilylation, which is achieved by using inexpensive polyhydromethylsiloxane and a number of titanium catalysts under ambient conditions. 13,14We chose dicyclopentadienyltitanium difluoride as the catalyst since it is easy to prepare from the commercially available dichloride and sodium fluoride in water at 50°C, followed by thorough drying. 14,15he catalyst was dissolved in anhydrous tetrahydrofuran and the polyhydromethylsiloxane added.The mixture was heated to 50°C, whereupon there was brief, rapid gas evolution (presumably hydrogen) and the mixture turned blue, indicative of the formation of a titanium (III) species. 16The lactone (13) was then added at room temperature and the mixture stirred until the reduction was complete.We employed 4 mol% catalyst as the reaction did not go to completion with 2 mol%.Work-up consisted of dilution with seven volumes of tetrahydrofuran and slow addition of dilute aqueous sodium hydroxide, which hydrolyses the silylated lactol and destroys excess hydride equivalents, liberating hydrogen.Work-up as a dilute solution is essential to avoid precipitation of a silicone polymer that occludes product and leads to lower yields.We were thus able to prepare >100g of 14, as a 4:1 mixture of anomers, in 80% yield.

Scheme 3
With lactol (14) in hand, we then performed an aldol reaction with aqueous formaldehyde to give 15 as a single stereoisomer at C-2 in 55% yield.This transformation has been described for the related 5-O-trityl derivative. 17The reaction was quite slow, presumably because the formaldehyde and open-chain ribose aldehyde are in low concentration.The product (15) was isolated as a pair of anomers (2:1 ratio).The stereochemistry obtained at C-2 can be readily explained by Figure 1, in which the formaldehyde preferentially approaches the enol syn to the adjacent hydrogen, that being the smaller substituent.The anomeric hydroxyl of 16 was protected as the pivalate ester from which a single anomer (17) was obtained by recrystallisation.Heating 17 with tetrabutylammonium fluoride in tetrahydrofuran failed to yield any 18, which was perhaps not entirely unexpected given the steric hindrance and electronic deactivation by the oxygen substituents.Other conditions were tried (e.g.caesium fluoride as fluoride source, DMF or DMSO solvent, at different temperatures) but all gave complex mixtures.There was evidence for both loss and migration of the pivalate ester.Although compound (17) was synthetically a dead-end, it was nevertheless useful for establishing the relative stereochemistry at C-1 and C-2 relative to C-5.Measurement of a series of nuclear Overhauser effects and long-range proton-carbon correlations were employed to confirm that the aldol reaction (14 -15) had indeed proceeded with the intended selectivity.Spectra were measured using a Varian Inova 500 MHz spectrometer in deuterochloroform.Assignment of the 1 H and 13 C nmr spectra was made using a combination of proton-proton correlation experiments (COSY/TOCSY) and one bond proton-carbon correlations (HSQC).Stereochemical information was gathered from long-range proton-carbon (HMBC), 1D n.O.e. and 2D n.O.e.(ROESY/NOESY) experiments.The proton and carbon shift assignments are in the experimental section.The key correlations used to assign the stereochemistry are shown in Figure 2.

Figure 2
We therefore turned our attention to DAST (diethylaminosulfur trifluoride), which is a powerful reagent for converting alcohols into fluorides, 18 including those imbedded in a monosaccharide, without cleavage of ketal protecting groups. 19

Scheme 5
Attempts to convert 19 directly into 10 were unsuccessful.A mixture of acetic anhydride and acetic acid at room temperature only returned starting material, and addition of a strong acid such as boron trifluoride or sulphuric acid to this mixture produced 20 without cleaving the ketal.However, hydrolysis under relatively forcing conditions (50% aq.TFA, 20°C or HOAc/H 2 O/Dowex ion-exchange resin H + form, 50°C) yielded triol (21) quantitatively.Acetylation then gave 10 in 72% overall yield from 15. 12 The anomers of 10 were readily separated by column chromatography, thereby facilitating their characterisation (see Experimental Section).

Conclusions
We have developed a relatively short, high yielding synthesis of a novel protected ribose, which should be a convenient intermediate for preparing a range of nucleoside analogues.

Experimental Section
General Procedures.Melting points were determined using open glass capillary tubes and a Gallenkamp melting point apparatus and are uncorrected.Spectroscopic data were recorded on a Perkin-Elmer 983 (IR), Finnigan Mat.Navigator (LRMS, either positive (ES + ) or negative (ES -) electrospray mode), and Varian Unity Inova ( 1 H NMR 300, 400 or 500 MHz) instruments and are consistent with the assigned structures.Combustion analyses were performed by Exeter Analytical (UK) Limited, Uxbridge, Middlesex, U.K. Optical rotations were performed by ARKAT Warwick Analytical Service Ltd., Coventry, U.K. using a Perkin-Elmer 341 polarimeter.Accurate mass determinations for molecular ions were obtained using a commercially available Apex II Fourier Transform Mass Spectrometer (Bruker Daltonics, Inc. Billerica, MA, USA) equipped with a 4.7 Tesla, passively shielded, superconducting magnet and an electrospray ionisation source (ESI), used in positive ion mode (Analytica of Branford, Branford, CT, USA) and calibrated using sodium trifluoroacetate.Ether refers to diethyl ether.All reactions were conducted under a positive pressure of dry nitrogen unless stated otherwise.Anhydrous solvents were purchased from Sigma-Aldrich and used directly.Flash chromatography refers to column chromatography on silica gel (Kieselgel 60, 230-400 mesh, from E. Merck, Darmstadt).Kieselgel 60 F 254 plates from E. Merck were used for TLC, and compounds were visualised using u.v.light or 0.5% aqueous potassium permanganate solution.

5-O-Benzyl-2-C-β-fluoromethyl-D-ribofuranose (21).
A solution of 19 (5.00 g, 15.9 mmol) in 50% aqueous trifluoroacetic acid (16 mL) was stirred at 20°C for 3 d.The mixture was neutralised by the addition of solid sodium bicarbonate and the solvent was removed under reduced pressure.The residue was adsorbed on silica gel and applied to the top of a silica gel pad.Elution with ether gave 21 as a gum (4.55 g, 100%), which was carried on to the next step without further purification.δ H (CDCl 3 , 400 MHz) approx.2.5:1 mixture of anomers.