Electronic Theses and Dissertations

Date

2024

Document Type

Dissertation

Degree Name

Doctor of Philosophy

Department

Chemistry

Committee Chair

Nathan DeYonker

Committee Member

Abby Parrill-Baker

Committee Member

Daniel Nascimento

Committee Member

Qianyi Cheng

Committee Member

Yongmei Wang

Abstract

The increasing complexity of enzyme catalysis requires precise and reproducible computational models to understand reaction mechanisms and support advances in drug design and protein engineering. This dissertation focuses on the advancement of enzyme modeling techniques, particularly in cytochrome P450 and hydrogenase enzymes, through the development of the Residue Interaction Network Residue Selector (RINRUS), an automated toolkit that addresses a critical gap in the reproducibility of quantum mechanics (QM)-cluster models. The RINRUS toolkit automates the systematic selection of catalytically relevant residues for QM-cluster model construction by leveraging residue interaction networks (RINs). By integrating the Probe and Arpeggio tools, RINRUS enhances model robustness by capturing essential interactions. This dissertation presents two primary applications of RINRUS. The first is a detailed mechanistic investigation of [NiFe]-hydrogenase, a hydrogen evolution catalyst. Using the RINRUS toolkit, QM-cluster models were systematically constructed to analyze a specific reaction step, focusing on discrepancies between QM and "Ene" reaction energies reported in the literature. The study found that RINRUS-based models closely align with reaction energies from "Ene" and Big-QM models by rationally selecting active site residues. Compared to distance-based methods, the RINRUS approach with rules-based residue selection—especially for charged residues—achieved better energy convergence. Even at modest levels of electronic structure theory, RINRUS models resolved energy discrepancies, highlighting the importance of careful model selection in computational enzymology. The second application focuses on cytochrome P450. QM-cluster models of the P450 enzyme GcoA, complexed with various catecholates, were created using the RINRUS toolkit. Previous research in our lab using an X-ray crystal structure with guaiacol as the ligand showed strong agreement with experimental data, while this study refined QM-cluster models from molecular dynamics (MD) simulations. From the MD-derived frame, DFT on the QM-cluster model was used to compute O-demethylation capacity, examining the kinetics and thermodynamics of hydrogen transfer for seven substrates. Key residues, such as His354, Phe349, Glu249, and Pro250, contribute to stabilizing transition states, maintaining heme's planar geometry, and optimizing ligand orientation. In summary, the RINRUS toolkit advances QM-cluster modeling by ensuring reproducible results and expanding the scope of computational enzyme modeling.

Comments

Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to ProQuest.

Notes

Embargoed until 11-13-2025

Download

Available for download on Thursday, November 13, 2025

Share

COinS