Computer modelling of enzyme reactions

Abstract

We give an overview on modern methods and atomistic models for the computation of enzyme mechanisms. It has to be stressed that models and methods must be at the same level of sophistication. If essential components (H-bonding side chains, metal ion, cofactor, etc.) of the active site are absent from the model, even the largest basis set may provide false results. On the other hand, the largest model with dozens of explicit atoms may lead to artefacts if the accuracy of the computational method (minimal basis set ab initio, semi-empirical or molecular force field) is unsatisfactory. Enzyme reactions are especially adequate for the application of an embedding scheme. While the active site should be treated by a high-level quantum mechanical method, distant protein atoms may be considered at a lower level of accuracy, e.g. as fixed point charges. The biophase is best treated by the DelPhi method, which is based on the solution of the linearised Poisson–Boltzmann equation and considers important counter-ion shielding effects on charged surface side chains. An integrated version of the embedding scheme is the combination of quantum mechanical and molecular mechanical methods (QM/MM) allowing treating all protein atoms in a calculation, however, at different levels of sophistication. Here we may have problems for atoms at the boundary, for which we present our approach offering a natural partition of the active site and protein core by applying strictly localised molecular orbitals. Two enzyme mechanisms, those of HIV integrase and xylose isomerase will be discussed in more detail in order to illustrate the principles outlined above.