Computational Studies of the Temperature Dependence of Enzyme Catalysis

Abstract: Enzymes are known to adapt to low temperatures by lowering the enthalpy of activation of the key reaction steps. A lesser known property of some psychrophilic (cold-active) enzymes is an anomalous temperature optimum. For most enzymes the temperature optimum is determined by their denaturation temperature, but anomalous optima can occur well below the denaturation temperatures and there must thus be an alternative mechanism for their inactivation. A proposed mechanism examined in this thesis posits that there is a small conformational change or local unfolding which renders the enzyme inactive above its temperature optimum.In paper I this hypothesized mechanism of inactivation was used to guide the design of variants of a bacterial α-amylase from Pseudoalteromonas haloplanktis. This enzyme exhibits an anomalous optimum and it has been suggested that the breaking of a specific interaction is responsible for its inactivity at high temperatures. Using computational methods several designs were evaluated. When experimentally tested the best performing design raised the temperature optimum by about 6 °C. This finding demonstrates the validity of the assumed mechanism and the utility of computational modeling in enzyme design.In Paper II and III two closely related bacterial lipases were investigated, namely Lipase A from Bacillus subtilis and a psychrophilic homologue found in Bacillus pumulus. By combining computational methods and experiments, it was found that a single mutation site is responsible for the difference in activity and the mechanism of its effect on the catalysis was determined. The lipase found in Bacillus pumulus also exhibits an anomalous temperature optimum. Our preliminary findings regarding the mechanism of this optima are also presented.In the last publication, paper IV, computation of heat capacities of biomolecular systems from molecular dynamics simulations is investigated. A non-zero heat capacity of activation has been suggested to be responsible for curved Arrhenius plots in some enzymes. To further investigate these ideas, clear and robust methods for the evaluation of heat capacity must be employed. Our results show that a method relying on mean potential energies calculated at various temperatures converges quicker then alternative methods and is less sensitive to the choice of thermostat and other simulation parameters.