Unraveling DNA Hydrolysis: Computational Strategies for Enzyme Catalysis

Abstract

The remarkable ability of organisms to manage genetic information is powered by nucleases, enzymes that engage with nucleic acids like DNA and RNA, highlighting the intricate dance of life at a molecular level. This dissertation research focuses on the mechanisms of DNA cleavage by nucleases, particularly I-PpoI and CRISPR-Cas9, both of which have played a key role in advancing gene drive and gene editing technologies. Understanding how DNA cleavage mechanisms work is crucial for genetic regulation and genome editing, as nucleases effectively break phosphodiester bonds in DNA with high specificity. There are several unresolved questions about how these enzymes facilitate DNA cleavage at the atomic level. These include the roles of active-site residues, the influence of metal ions, and the energy barriers encountered along the catalytic pathways. Although experimentally resolved structures of I-PpoI and Cas9 have provided invaluable atomistic insights into their architecture and function, the free energy landscapes governing their catalytic mechanisms, particularly the nature of transition states and the energetic roles of individual residues and metal ions, remain incompletely understood. Motivated by these gaps, this work addresses four central questions: What are the molecular mechanisms underlying DNA cleavage by I-PpoI and Cas9? How do metal ions and point mutations modulate the catalytic free energy landscapes of these enzymes? Can the energetic contributions of individual residues be quantified to predict the impact of mutations on catalytic efficiency? Finally, from a methodological perspective, can machine learning be harnessed to accelerate free energy simulations of enzymatic reactions? To answer these questions, this work employs hybrid quantum mechanics/molecular mechanics (QM/MM) free energy simulations to investigate how metal ions and individual residues shape the free energy landscapes of DNA cleavage. By comparing these two enzymes, the general principles of phosphodiester bond hydrolysis are elucidated. In parallel, machine learning techniques are explored to accelerate QM/MM sampling, offering improved efficiency with minimal loss of chemical accuracy. Together, these mechanistic insights and methodological developments advance our understanding of nuclease function and offer new tools for predictive enzymology and rational enzyme design.