Is the bias introduced in a FEP calculation by parameterizing a QM reaction acceptable? Comparison with Car-Parrinello MD/AMBER results for the second proton transfer in triosephosphate isomerase (TIM)

The intramolecular mechanism for the second proton transfer in TIM has been subjected to quantum mechanics (QM)/free energy perturbation (FEP) simulations in the presence of the protein environment with His95 in its ε form. The results confirm and eventually strengthen earlier FEP simulations in the presence of δ-His95 [G. Alagona, C. Ghio, P.A. Kollman, J. Comp. Chem. 24 (2003) 46] producing a shallow energy profile for the intramolecular path. Attempts to model the His-assisted pathway ab initio, for determining the parameters to be used in subsequent intermolecular FEP simulations, failed because HF/3-21+G calculations in vacuo suggested Glu165 instead as the only proton acceptor/donor in the active site. In order to assess whether the intramolecular mechanism modeled in the FEP simulations produced an unwanted bias, Car-Parrinello molecular dynamics simulations (including in the QM model the substrate intermediate, GluH165, His95 and Lys12), coupled to a classical description of the protein environment, have been carried out. The His95-assisted mechanism turns out to be the preferential pathway regardless which model of the environment is considered. Barrier heights, obtained using the blue moon ensemble method, indicate that the intramolecular proton transfer is about twice as much less favorable than the His95-assisted one.