Novel Ylidic Phosphoryl Compounds from Halogenated Furan-2,5-Diones with Trivalent Phosphorus Esters: Application of this Approach to New Trisphosphonates Containing a Geminal Bisphosphonate Unit

Abstract The reactions of trivalent phosphorus esters, including trialkyl phosphites, dialkyl phosphonites, and alkyl phosphinites, with 3-halo- and 3,4-dihalo-furan-2,5-diones has been shown to lead to the formation of novel phosphorus ylides possessing additional phosphoryl-containing groups. For the reaction of 3,4-dihalo-furan-2,5-diones with trialkyl phosphites, the products are trialkoxyphosphonium ylides containing an adjacent geminal bisphosphonate unit. These can be used to provide a convenient route to novel 2,3,3-tris(dialkoxyphosphoryl)-substituted propionate esters which can be hydrolyzed to give the corresponding novel trisphosphonic monocarboxylic acid. GRAPHICAL ABSTRACT


INTRODUCTION
The synthesis of systems containing a geminal bisphosphonate unit has attracted considerable interest because of the ability of some of these systems to bind to the surface of bone or to affect bone metabolism. We have had a long-standing interest in those geminal bisphosphonates that bind to bone in the context of developing radiopharmaceutical agents for both skeletal imaging 1 and therapy. 2 On the other hand, bisphosphonates that affect bone metabolism can be used to treat bone related disorders such as Paget's disease 3 and osteoporosis. 4 We have therefore maintained an interest in reactions that could be exploited to generate systems containing a geminal bisphosphonate unit. In our studies of the reactions of by the phosphorus reagent to give the phosphoryl systems 16a-e. While pathways analogous to both Routes A and B (Scheme 2) could then explain the subsequent formation of 18a-e from 16a-e, DFT calculations (see Supplemental Materials) suggest that the presence of the initially introduced phosphorus group would encourage subsequent attack at the remaining chlorinated carbon. This gives the phosphonium-phosphoryl systems 17a-e which are converted to the observed ylidic-bisphosphonates 18a-e in a manner analogous to that proposed for the conversion of 5 to 7 (Route A, Scheme 2). However, we were unable to confirm this mechanism since, as previously noted, we were unable to observe the presence of either of the intermediate compounds 16 and 17 by NMR during the reaction. This may, however, simply indicate that their subsequent reaction with phosphorus reagent is faster than their rate of formation.
The ylidic bisphosphoryl compounds 18a-e, like the ylidic monophosphoryl compounds 7 and 12, were sufficiently stable to be purified by column chromatography and this also enabled the two diastereomeric forms of the ylide 18d [(±) δP 32.9 (d, JPP = 22 Hz), 33.8 (d, JPP = 22 Hz) and 64.1 (s); and meso δP 34.2 (x2)(s) and 64.1 (s)] to be successfully separated. It is interesting to note that while the non-equivalence of the phosphonate resonances in 31 P NMR spectrum of the (±)-diastereoisomer of 18d gives rise to the observation of a 2 JPP coupling between them, neither of these resonances show a 3 JPP coupling to the ylidic phosphorus atom. Once again the IR bands associated with the anhydride units in the ylidic bisphosphoryl compounds 18 showed a shift to lower wavenumber relative to the corresponding trisphosphoryl systems 19. Thus, for example, whereas the ylide 18c exhibited bands at 1714 and 1790 cm -1 the corresponding trisphosphonate 19c, formed by passing anhydrous HCl into a solution of the ylide in DCM, showed bands at 1781 and 1861 cm -1 .
Interestingly, heating the ylides 18a-c with their corresponding alcohol under reflux led not only to decomposition of the ylide group but also ring opening of the anhydride and subsequent decarboxylation to give the highly phosphorylated propionates 20a-c. The same products could also be prepared by heating the trisphosphonates 19a-c with the corresponding alcohol.

General details 13
NMR spectra were recorded on JEOL EX-270, Bruker AMX400, AV400 and AV600 spectrometers. 31 P NMR spectra are referenced to 85% phosphoric acid, 1 H NMR spectra are

CONCLUSIONS
We have shown that the reactions of trivalent phosphorus esters, such as trialkyl phosphites, dialkyl phosphonites and alkyl phosphinites, with 3-halofuran-2,5-diones follow a quite different reaction pathway to that observed when phosphines are used. 7,8 With the trivalent phosphorus esters the reactions proceeded cleanly with the formation of novel ylides 7 that possess an adjacent phosphoryl-containing group. In the parent 3-bromofuran-2,5dione 1 case, DFT calculations indicate that there is a lower energy barrier for the initial attack by the phosphorus ester at the C-4 position than at the C-3 position on the ring. 13 If so, this would suggest the reaction may proceed via the carbene mechanism shown in Scheme 2, Route B rather than by an initial substitution of the halogen by the phosphorus reagent (Scheme 2, Route A). On the other hand, when a substituent, such as a benzyl group, is placed on the furan-2,5-dione adjacent to the halogen, as in 10a, we observe products, such as the benzylidene-substituted isomers 13 and 14, which are better explained in terms of an initial attack by the phosphorus ester at the halogenated carbon. This would be consistent with the benzyl substituent inhibiting access to its adjacent carbon by the phosphorus ester. We have also shown that by starting with 3,4-dihalo-furan-2,5-diones the scope of the reaction can be extended to produce novel ylides 18 that possess an additional geminal bisphosphoryl system. Thus, for example, the use of triethyl phosphite and 3,4-dichlorofuran-2,5-dione 3 gave the novel ylidic bisphosphonate 18b containing a geminal bisphosphonate unit. Although reactions mechanism analogous to both Routes A and B (Scheme 2) can be drawn to explain the formation of the ylides 18, DFT calculations 13 indicate that the presence of the initially introduced phosphorus group in 16 serves to encourage the subsequent attack by the phosphorus ester at the remaining halogenated carbon thus giving the mechanism shown in Scheme 4 rather than one involving a carbene intermediate. The ylidic bisphosphonates 18a-c have been shown to be valuable precursors for the formation of the corresponding novel trisphosphonates 20a-c and the trisphosphonic monocarboxylic acid 20; R = H. Preliminary studies also indicate that the ylidic bisphosphonates 18a-c will be useful precursors for a range of other highly phosphorylated compounds.