35 GHz ENDOR characterization of the "very rapid" signal of xanthine oxidase reacted with 2-hydroxy-6-methylpurine (13C8): evidence against direct Mo-C8 interaction.

Xanthine oxidase is a molybdenum-containing enzyme that catalyzes the hydroxylation of xanthine and a wide variety of other aromatic heterocycles. In the course of the reaction with xanthine and substrates such as 2-hydroxy-6-methylpurine (HMP), the enzyme gives rise to a Mo(V) EPR signal, denoted "very rapid", that arises from an authentic catalytic intermediate. The two alternative catalytic mechanisms proposed for this enzyme differ critically in whether the distance between Mo and C8 of the purine nucleus in this intermediate is short enough to admit a direct bonding interaction. To examine this distance, we have performed 13C ENDOR measurements of the "very rapid" EPR signal generated by xanthine oxidase during reaction with 13C8-HMP. The resulting (13)C8 hyperfine tensor, A = [10.2(1), 7.0(1), 6.5(1)] MHz, is discussed in the framework of a detailed consideration of factors involved in extracting metrical parameters from an anisotropic hyperfine interaction composed of contributions from multiple sources, in particular, the effect of the local contributions from spin density on (13)C8. The analysis presented here gives a Mo...C distance whose value is expected to be ca. 2.7-2.9 A in the "very rapid" intermediates formed with both xanthine and HMP, consistent with plausible bond lengths for a Mo-O-C8 fragment where C8 is a trigonal-planar aromatic carbon. The difference from earlier conclusions is explained. The data thus do not support the existence of a direct Mo-C bond in the signal-giving species. This conclusion supports a mechanism that does not involve such an interaction and which begins with base-assisted nucleophilic attack of the Mo(VI)-OH group on the C-8 of substrate, with concomitant hydride transfer to the Mo=S group to give Mo(IV)-SH; the EPR-active "very rapid" species then forms by one-electron oxidation and deprotonation to yield the EPR-detectable Mo(V)OS(OR) species. We further discuss the complexities and limitations of the semiempirical method used to arrive at these conclusions.