Computational Approaches to Understanding the Structure, Dynamics, Functions, and Mechanisms of Various Bacterial Proteins

The 3D structure of a protein can be fundamentally useful for understanding protein function. In the absence of an experimentally determined structure, the most common way to obtain protein structures is to use homology modeling, or the mapping of the target sequence onto a closely related homolog with an available structure. However, despite recent efforts in structural biology, the 3D structures of many proteins remain unknown. Recent advances in genomic and metagenomic sequencing coupled with coevolution analysis and protein structure prediction have allowed for highly accurate models of proteins that were previously considered intractable to model due to the lack of suitable templates. Structural models obtained from homology modeling, coevolution-based modeling, or crystallography can then be used with other computational tools such as small molecule docking or molecular dynamics (MD) simulations to help understand protein function, dynamics, and mechanism.Here coevolution-based modeling was used to build a structural model of the HgcAB complex involved in mercury methylation (Chapter I). Based on the model it was proposed that conserved cysteines in HgcB are involved in shuttling mercury, methylmercury, or both. MD simulations and docking to a homology model of E. coli inosine monophosphate dehydrogenase (IMPDH) provided insights into how a single amino acid mutation could relieve inhibition by altering protein structure and dynamics (Chapter II). Coevolution-based structure prediction was also combined with docking, and experimental activity data to generate machine learning models that predict enzyme substrate scope for a series of bacterial nitrilases (Chapter III). Machine learning was also used to identify physicochemical properties that describe outer membrane permeability and efflux in E. coli and P. aeruginosa and new efflux pump inhibitors for the E. coli AcrAB-TolC efflux pump were identified using existing physicochemical guidelines in combination with small molecule docking to a homology model of AcrA (Chapter IV). Lastly, quantum mechanical/molecular mechanical simulations were used to study the mechanism of a key proton transfer step in Toho-1 beta-lactamase using experimentally determined structures of both the apo and cefotaxime-bound forms. These simulations revealed that substrate binding promotes catalysis by enhancing the favorability of this initial proton transfer step (Chapter V).