Isoxazole to oxazole: a mild and unexpected transformation †

3-Aryltetrahydrobenzisoxazoles prepared en route to the coleo-phomone natural products and analogues, were found to undergo a remarkable base-mediated rearrangement to 2-aryltetrahydro-benzoxazoles. The scope of this unprecedented, facile transformation was probed: a range of analogues was produced, a mechanism proposed, and an application demonstrated by synthesis of a known herbicidal compound. As an

This rearrangement also took place in the absence of alkylating agent (Scheme 2); the phenolic product 6a (R = R 1-4 = H) was stable to the basic conditions, and was successfully O-allylated to give ether 5 on addition of allyl bromide. We have verified the structures of both isoxazole 4a and dimethyl product oxazole 6b (R = Me, R 1-4 = H; vide infra) through X-ray crystal structure determinations, Fig. 1a and b. ‡ We further investigated the scope of the remarkable rearrangement of benzisoxazoles 4 to benzoxazoles 6. Using isoxazole 4a, rearrangement was found to occur in aprotic solvents with reaction time of 4 h under a range of basic conditions (Table 1) including carbonates, alkoxide and amidine, but failed with tertiary amines. In the presence of water or ethanethiol (entries 12, 13) the amide products 7a,b, respectively, of ring opening of the oxazole 6a were isolated; the constitutions of the amides were confirmed by X-ray crystal structures. 7 A range of 3-(2-hydroxyphenyl)tetrahydrobenzisoxazoles 4a-i, differently substituted in the aryl and the cyclohexane ring were shown to undergo rearrangement (Table 2) using the convenient Cs 2 CO 3 conditions (THF reflux) to afford oxazoles 6a-i. § We propose the mechanism illustrated in Scheme 3 for the rearrangement. Until the oxazole structure was determined, we had supposed that a Boulton-Katritzky ring transposition 8 (similar to that reported by Suzuki et al. 9 ) was taking place,   so we retain this as the initial step in this remarkable isoxazoleto-oxazole conversion. 10 This can be followed by a Neber rearrangement 11 to give an azirine, thus overall replacing the N-O bond of the isoxazole by an N-C bond. The azirine may be envisaged to be in equilibrium with a nitrile ylide 12 stabilised at the formal negative end by the 1,3-dione system, and at the formal positive end by the electron-rich 2-hydroxyphenyl substituent. The 1,3-dipole finally collapses to the oxazole in a 6p electrocyclic ring closure.
Previous reports indicate that it is possible to form oxazoles from azirines, and also that an azirine can be generated from an isoxazole either thermally or photolytically. 13,14 However, the energies required well exceed those of our reaction conditions and thus an alternative rationale was required. The Neber rearrangement is an alternative way of generating azirines given the appropriate leaving group. 15 This mechanism implies that the base is catalytic, and this was supported by isolation of 6a (66%) from 4a using 0.1 mol equiv. of Cs 2 CO 3 (THF reflux, 1.5 h). An intermediate with m/z identical to both the isoxazole and oxazole was observed by LC-MS during the rearrangements of 4a and 4c to 6a,c, respectively, and isolated by HPLC. We were not able to unambiguously identify the structure, but NMR studies indicate the cyclohexane portion to be symmetrical, supporting either the azirine or nitrile ylide formulation. 16 An attempt to crystallise the dimethyl intermediate formed from 4c led merely to recovery of the oxazole 6c. The oxazole ring opening to form amides 7a,b is consistent with nucleophilic attack at C-5 of the oxazole.
To discount the possibility of the oxazoles being formed by retro-cycloaddition from the isoxazoles and recombination via a different connectivity, we have shown that treatment of a mixture of the two tetrahydrobenzisoxazoles 4c and 4i under the Cs 2 CO 3 -THF reflux conditions led only to the tetrahydrobenzoxazoles 6c and 6i predicted by the mechanism of Scheme 3, with no crossover products observed.
The tetrahydrobenzoxazoles prepared herein are closely related to a series of herbicides described in a patent by Ueda et al. 17 Using benzoxazole 6a we have prepared an example 8 of this group by reaction with 2-chloropyrimidine (47%) (Scheme 4).
In conclusion, we have discovered an unexpected, remarkably facile novel base-mediated rearrangement of tetrahydrobenzisoxazoles to tetrahydrobenzoxazoles, demonstrated the scope and probed the reaction mechanism of this surprising transformation. The synthetic utility of this rearrangement has been demonstrated by synthesis of a known bioactive compound.
The authors thank Loughborough University for a studentship (A. C.) and Lilly UK for financial support, and ENSIACET (Toulouse) for support for a work placement (R. M.).