Influence of Frozen Residues on the Exploration of the PES of Enzyme Reaction Mechanisms

In
this work, we studied one of the very widely used approximations
in the prediction of an enzyme reaction mechanism with computational
methods, that is, fixing residues outside a given radius surrounding
the active site. This avoids the unfolding of truncated models during
MD calculations, avoids the expansion of the active site in cluster
model calculations (albeit here only specific atoms are frozen), and
prevents drifting between local minima when adiabatic mapping with
large QM/MM models is used. To test this, we have used the first step
of the reaction catalyzed by HIV-1 protease, as the detrimental effects
of this approximation are expected to be large here. We calculated
the PES with shells of frozen residues of different radii. Models
with free regions under a 6.00 Å radius showed signs of being
overconstrained. The QM/MM energy barrier for the remaining models
was only slightly sensitive to this approximation (average of 0.8
kcal·mol–1, maximum of 1.6 kcal·mol-1).
The influence over the energy of reaction was almost negligible. This
widely used approximation seems safe and robust. The resulting error
is on average below 1.6 kcal·mol–1, which is
small when compared with others deriving from, for example, the choice
of the density functional or semiempirical MO/SCC-DFTB method, the
basis set used, or even the lack of sampling or incomplete sampling.