Computational Investigations of Coenzyme F430 Biosynthesis and Radical S-Adenosyl-L-Methionine Enzyme Catalysis

Abstract

This dissertation applies computational techniques to investigate two enzymatic systems: the structurally uncharacterized coenzyme F430 biosynthetic enzyme CfbCD and radical S-adenosyl-L-methionine (RS) enzymes. CfbCD is of particular interest due to its role in synthesizing the essential Ni-containing coenzyme F430, which is bound in the active site of methyl-coenzyme M reductase (MCR)—the enzyme catalyzing the terminal step of methanogenesis. Attempts to engineer methanogenesis into heterologous hosts more amenable to metabolic engineering have been hindered by challenges such as the functional expression of CfbCD, likely due to its unusual and partially unresolved chemistry, as well as its implied structural complexity as a member of the famously complex nitrogenase superfamily. In this work, density functional theory (DFT) and molecular dynamics (MD) simulations are employed to probe the CfbCD reaction mechanism, the structure of its uncharacterized catalytic intermediate, and its protein dynamics. These studies give insight into the catalytic mechanism, support characterization of experimentally observed intermediates, and identify putative binding pockets, catalytic residues, and strong evidence for conformational cooperativity between active sites. Regarding RS enzymes, we build on prior work by Dr. Patrick Donnan, who sought to structurally characterize the canonical intermediate Ω, a key conserved radical initiation species in diverse RS transformations across all domains of life, and intermediates I and II in the catalytic cycle of the noncanonical RS enzyme Dph2 from the diphthamide biosynthetic pathway. While previous studies proposed organometallic structures for Ω and intermediate I containing Fe–C bonds, this study applies broken-symmetry DFT (BS-DFT) to these organometallic models and simple near-attack conformers (SAM-NAC) of [4Fe–4S]-bound S-adenosyl-L-methionine, finding that in both cases SAM-NAC models are more consistent with available spectroscopic data.