Comment on "The pentacovalent phosphorus intermediate of a phosphoryl transfer reaction".

The claim by Lahiri et al. (1) to have identified the pentacovalent phosphorus intermediate of a phosphoryl transfer reaction addresses an important, long-disputed problem: whether the enzyme-catalyzed transfer of phosphate is a dissociative or associative process. The apparent observation of a pentaoxy-phosphorane suggests that the transfer occurs through an associative, addition– elimination process and contrasts with the solution mechanism, which is dissociative in character and has little bond order to the nucleophile and leaving group in the transition state (2). However, the apical P–O bond lengths for this trigonal bipyramidal intermediate (2.0 and 2.1 Å) are significantly longer than those in small-molecule phosphorane structures [e.g., 1.65 Å and 1.67 Å for pentacyclohexyloxy-phosphorane (3)]. These distances are not unreasonable for a transition state structure with partial apical bonds (4), but by definition this is not a stable species. Even at 93 K, the half-life for decomposition of an activated complex is 10 11 s, so this cannot be a kinetically trapped transition state. To be thermodynamically viable, such a TS complex would need to be comparable in stability with the starting and product complexes, which implies that the kinetic barrier to reaction has disappeared (and so the transfer reaction should become ultrafast at low temperature) and that the thermodynamic properties of the transition state are very unusual (because the change in temperature from 291 to 93 K would have to lead to a decrease in G of some 12 kcal mol 1 for the reaction within the ES complex). From a structural point of view, we were also concerned that some of the attributes of various Fourier maps described by Lahiri et al. (1) did not seem consistent with a PO3 n– moiety. Is there an alternative, better interpretation of these data? The crystallization conditions used by Lahiri et al. (1) contained 10 mM Mg and 100 mM NH4F. Our published results (5) have shown that MgF3 – is a good mimic of the transition state of phosphoryl transfer reactions and can be formed under conditions such as those employed by Lahiri et al. (1). Our crystal structure of a small GTPase complexed with its GTPase-activating protein (GAP) shows that MgF3 – sits in the active site and adopts a trigonal bipyramidal arrangement with similar apical bond lengths (2.0 and 2.3 Å) to the nucleophile and leaving group as those assigned to the “phosphorane” above. It therefore seems plausible that the PGM structure contains an MgF3 – rather than a PO3 n– species. We further note that fluoride catalyses the dephosphorylation of a phosphoenzyme without affecting activity (6), furnishing a kinetic pathway for loss of phosphate from the PGM complex. We have previously confirmed that our G protein complex really contained the MgF3 –