Synthesis of phosphonate and phostone analogues of ribose-1-phosphates

The synthesis of phosphonate analogues of ribose-1-phosphate and 5-fluoro-5-deoxyribose-1-phosphate is described. Preparations of both the α- and β-phosphonate anomers are reported for the ribose and 5-fluoro-5-deoxyribose series and a synthesis of the corresponding cyclic phostones of each α-ribose is also reported. These compounds have been prepared as tools to probe the details of fluorometabolism in S. cattleya.


Introduction
Fluoroacetate (1) and 4-fluorothreonine (2) are unusual secondary metabolites in that they contain a fluorine atom. They are elaborated by the bacterium Streptomyces cattleya as part of its defense strategy since 4-fluorothreonine (2) has antibiotic activity and fluoroacetate (1) is a toxin [1]. The biosynthetic pathway from fluoride ion to these fluorometabolites has largely been elucidated and is summarised in Scheme 1 [2].
The first committed step in the biosynthesis involves the fluorinase, an enzyme that catalyses nucleophilic attack of fluoride ion on SAM (3) to generate 5′-FDA (4). The initial fluorinated product 5′-FDA (4) is then depurinated by a purine nucleotide phosphorylase (PNP) to give 5-fluoro-5-deoxyribose-1-phosphate (5). This phosphate intermediate becomes the substrate for a ring opening isomerisation reaction converting 5-fluoro-5deoxyribose-1-phosphate (5) to 5-fluoro-5-deoxyribulose-1phosphate (6). Further processing to fluoroacetaldehyde (7) and then conversion to fluoroacetate (1) and 4-fluorothreonine (2) complete the pathway. The focus of this research involved generating a potential inhibitor of either the PNP or the subsequent isomerase enzymes as a tool, to try to accumulate biosynthesis intermediates in cell free extract studies [3]. We have already demonstrated that the S. cattleya PNP can depurinate both 5′-FDA and adenosine, and thus the presence or absence of fluorine at the C-5′-position is not a requisite for catalysis. The next enzymatic step involves a ring opening isomerisation of the ribose-1-phosphate 5. In this paper we report the synthesis of phosphonate analogues 8 and 9 of ribose-Scheme 1: Biosynthetic pathway from fluoride ion to fluoroacetate 1 and 4-fluorothreonine 2 in S. cattleya [2].
1-phosphate 5 where the linking phosphate oxygen has been replaced by a methylene group and for 9 the fluorine has been replaced by OH (see Figure 1). These phosphonates were considered as candidate inhibitors for either the PNP or the isomerase enzyme, with potential utility as tools to interrogate the biosynthesis system, and as inert candidate substrate analogues for enzyme co-crystallisation studies. We also report the synthesis of the cyclic phostone analogues 10 and 11 ( Figure 1), which were viewed also as potential substrate analogues for co-crystallisation studies, particularly with the isomerase enzyme which converts 5 to 6 during the biosynthesis.
Due to the inability to separate α−18 from β−18, route B (Scheme 3) was explored as an alternative. Following the protocol demonstrated by Meyer et al. [4], 5-O-trityl-phosphonates 21a and 21b were obtained as individual epimers after  chromatographic separation of the initial product mixture (α:β of 4:1). Detritylation of 21a, b was accomplished with ZnCl 2 [9] to generate the 5-hydroxyphosphonates 22a and 22b in a good yield. The free alcohol of both epimers was then fluorinated at the 5-position using tosylfluoride and TBAF in refluxing THF [10,11]. Sequential deprotection of 23a and 23b with TMSBr and then TFA yielded the free phosphonic acids which were neutralised with cyclohexylamine [12] to generate the non-hygroscopic salts 8a, b and 9a, b which could be recrystallised from MeOH/acetone. X-Ray structure analysis of a suitable crystal of the fluoromethyl phosphonate 8a was obtained ( Figure 2) which confirmed its structure and stereochemical relationship to intermediate ribose-1-phosphate metabolite 5. Clearly, phosphonate 8a formed a 1:1 salt with the amine. Additionally, 23a and 22a were treated in separate reactions with acetic anhydride in dry pyridine [13] to generate the cyclic phostones 10 and 11 respectively. In each case this conversion could be conveniently followed by a change in the 31 P-NMR resonance of 27 ppm (phosphonate) to 44 ppm, characteristic of a cyclic phostone. Purification by chromatography allowed the recovery of the phostones 10 and 11 in moderate yield.
Incubation of the phosphonate salts 8a and 9a with the S. cattleya PNP did not result in an inhibitory effect, and clearly the phosphonates are not good substrate analogues of the phosphates with this enzyme. We are currently exploring phosphonates 8a and 9a as inhibitors of the isomerase, the next enzyme on the pathway, and in order to probe the molecular mechanism of this enzyme, these phosphonates and the phostones 10 and 11 are being used in co-crystallisation trials with each of the overexpressed enzymes (PNP and isomerase).

Experimental
Nuclear magnetic resonance (NMR) spectra were obtained using Bruker Advance 300 and Bruker Advance II 400. All chemical shifts (δ) are reported in parts per million (ppm) and coupling constants (J) are given in Hertz (Hz). Melting points were determined in Pyrex capillaries using a Gallenkamp Griffin MPA350.BM2.5 melting point apparatus. Infra red (IR) spectra were recorded on Nicolet Avatar 360 FT-IR. High resolution mass spectrometry (HRMS) was carried out using a Micromass LCT (Manchester, UK) mass spectrometer with electrospray ionization (ESI) operating in positive and negative modes. Optical rotations were measured on Perkin-Elmer model 341 polarimeter.