DOI
https://doi.org/10.25772/6G94-FF42
Defense Date
2011
Document Type
Dissertation
Degree Name
Doctor of Philosophy
Department
Pharmaceutical Sciences
First Advisor
John Hackett
Abstract
Biophysical entities are complex systems systems with strong environmental coupling, dominated by fluctuations on a hierarchy of timescales. These properties confound simulation of ligand binding and catalysis, inflating the scale of the problem to one tractable only with a considerable outlay of resources. In an attempt to ameliorate this restriction, several techniques are developed to accelerate biomolecular simulations while collaterally lending physical insight. The first segment of this dissertation is concerned with directed simulations of ligand binding in a model system. Using the serum retinol binding protein as a prototype, the potential of mean force associated with ligand binding is calculated and dissected. Desolvation is sufficient to drive formation of an intermediate binding state; however, a combination of electrostatic and van der Waals interactions pull the intermediate into a stable configuration. Association is accompanied by a change in the conformational flexibility of the portal domains of sRBP and subsequent "stiffening" of the holo sRBP, reflecting an "order-disorder" transition in the protein. The third and fourth chapters of this dissertation entail ab initio molecular dynamics (AIMD) and quantum Monte Carlo methods (QMC) for computational enzymology. An ideal system for the application of AIMD, are the cytochromes P450 (CYP450s). Most AIMD calculations are performed using plane-wave (PW) density functional theory as an electronic structure method; conversely, computational enzymology is generally performed using calculations with Gaussian basis sets. In this scenario, no benchmark exists to comparison of PW calculations with experimental data. To clarify this situation, benchmark PW calculations are performed on CYP450 Compound I, the iron-oxo species operant in these enzymes. Finally, lattice QMC methods are developed to characterize tunneling in mean-field backgrounds. Using AIMD simulations, a potential of mean force is constructed in the limit of classical nuclei. A framework for path integral Monte Carlo is introduced in which the Euclidean functional integral is discretized on a lattice, permitting calculations of correlation functions and ultimately the action of the system. As the action is quenched, instanton solutions and their contribution to degeneracy splitting are obtained. This technique is demonstrated for malonaldehyde, a system in which proton tunneling is critical.
Rights
© The Author
Is Part Of
VCU University Archives
Is Part Of
VCU Theses and Dissertations
Date of Submission
December 2011