Comparison of QM-Only and QM/MM Models for the Mechanism of Tungsten-Dependent Acetylene Hydratase

We report a comparison of QM-only and QM/MM approaches for the modeling of enzymatic reactions. For this purpose, we present a QM/MM case study on the formation of vinyl alcohol in the catalytic cycle of tungsten-dependent acetylene hydratase. Three different QM regions ranging from 32 to 157 atoms are designed for the reinvestigation of the previously suggested one-water attack mechanism. The QM/MM calculations with the minimal QM region M1 (32 atoms) yield a two-step reaction profile, with an initial nucleophilic attack followed by the protonation of the formed vinyl anion intermediate, as previously proposed on the basis of QM-only calculations on cluster model M2 (116 atoms); however, the overall QM/MM barrier with M1 is much too high, mainly due to an overestimate of the QM/MM electrostatic repulsions. QM/MM calculations with QM region M2 (116 atoms) fail to reproduce the published QM-only results, giving a one-step profile with a very high barrier. This is traced back to the strong electrostatic influence of the two neighboring diphosphate groups that were neglected in the QM-only work but are present at the QM/MM level. These diphosphate groups and other electrostatically important nearby residues are included in QM region M3 (157 atoms). QM/MM calculations with M3 recover the two-step mechanism and yield a reasonable overall barrier of 16.7 kcal/mol at the B3LYP/MM level. They thus lead to a similar overall mechanistic scenario as the previous QM-only calculations, but there are also some important variations. Most notably, the initial nucleophilic attack becomes rate limiting at the QM/MM level. A modified two-water attack mechanism is also considered but is found to be less favorable than the previously proposed one-water attack mechanism. Detailed residue interaction analyses and comparisons between QM/MM results with electronic and mechanical embedding and QM-only results without and with continuum solvation show that the protein environment plays a key role in determining the mechanistic preferences in acetylene hydratase. The combined use of QM-only and QM/MM methods provides a powerful approach for the modeling of enzyme catalysis.