Computational Modeling of the Mechanisms and Selectivity of Organophosphate Hydrolases

Computational modeling is becoming an increasingly integral part of (bio)chemistry, providing a powerful complementary view into the dynamics, binding, and reactivity of biochemical systems. In particular, molecular simulations based on multiscale models are now regularly employed in studies of enzymatic reactions, offering invaluable mechanistic insight through the lens of molecular energy landscapes. In this thesis, I used the empirical valence bond (EVB) and related methods to study the mechanisms and selectivity of organophosphate hydrolases. Organophosphate hydrolases are a diverse class of enzymes capable of degrading some of the most toxic compounds known to mankind, including pesticides and chemical warfare agents. They are particularly interesting from a mechanistic and evolutionary point of view, having evolved the ability to catalyze the hydrolysis of compounds which were introduced to nature less than a century ago. Moreover, they show promise as effective organophosphate decontamination agents and a thorough understanding of their function is fundamental to the future design of efficient and selective biocatalysts. As organophosphate hydrolases are metal-dependent enzymes, a reliable metal model was a prerequisite to our simulations. First, I present the development of force-field independent parameters for several alkaline-earth and transition-metal ions described using the nonbonded cationic dummy model. The model was subsequently employed in EVB simulations to probe the origin of metal-ion activity and selectivity patterns observed in methyl parathion hydrolase (MPH) and to provide mechanistic insight into its paraoxonase and promiscuous arylesterase activities. I further set out to resolve open mechanistic questions surrounding diisopropyl fluorophosphatase (DFPase) by performing extensive simulations of two mechanistic pathways proposed in literature, including calculating the effects of mutations, temperature, and protonation states on the rate of hydrolysis. Using this knowledge, I address the origin of cross-selectivity between DFPase and a structurally similar enzyme serum paraoxonase 1 (PON1). Finally, I present the latest developments in the software used to perform the simulations