Since the early beginnings of chemical simulations, obtaining the activation enthalpy, entropy and free energy for molecular reactions has always been one of the major goals, in order to shed light on the molecular basis of enzymatic catalysis. In the last few decades, the ever-increasing computer power allowed the development of several methodologies such as metadynamics, Jarzynski, and umbrella sampling for the calculation of fairly accurate free energy profiles in complex systems using multi scale QM-MM schemes. However, determining accurate activation enthalpies and entropies remained a more difficult issue. In this work we present a systematic study analyzing a methodology for the calculation of the activation parameters in both water and enzymatic environments, using a DFT/Gaussian basis set QM-MM implementation for graphical processing units developed in our group. This methodology achieves the calculation of activation enthalpies by the acquisition of free energy profiles at several temperatures which are then used in a Van´t Hoff analysis.

However, QM-MM calculations at the DFT level of theory entail heavy computational costs, therefore reducing the amount of conformations sampled for a particular system. In addition, this can lead to suboptimal results, since the QM region may not have (on average) the desired global temperature. As such, we also explore the influence of said reduced sampling by means of QM-MM calculations at the semiempirical PM3 level of theory, attaining greater sampling times at the cost of lesser precision in the free energy profile.

Chorismate mutase catalyzed conversion of chorismate to prephenate has been employed as a benchmark case, since there is a significant amount of data, both experimental and theoretical, reported in literature, and allows us to compare the behavior of the same system both in water and in the enzyme.