Systematic development of computational models for the catalytic site in galactose oxidase: impact of outer-sphere residues on the geometric and electronic structures

Abstract

A systematic in silico approach has been employed to generate sound, experimentally validated active-site models for galactose oxidase (GO) using a hybrid density functional, B(38HF)P86. GO displays three distinct oxidation states: oxidized [Cu(II)-Y*]; semireduced [Cu(II)-Y]; and reduced [Cu(I)-Y]. Only the [Cu(II)-Y*] and the [Cu(I)-Y] states are assumed to be involved in the catalytic cycle, but their structures have not yet been determined. We have developed several models (1-7) for the [Cu(II)-Y*] state that were evaluated by comparison of our computational results with experimental data. An extended model system (6) that includes solvent molecules and second coordination sphere residues (R330, Y405, and W290) is essential to obtain an experimentally correct electronic structure of the active site. The optimized structure of 6 resulted in a five-coordinate Cu site with a protein radical centered on the Tyr-Cys cofactor. We further validated our converged model with the largest model (7) that included additional outer-sphere residues (Q406, H334, Y329, G513, and T580) and water molecules. Adding these residues did not affect significantly the active site's electronic and geometric structures. Using both 6 and 7, we explored the redox dependence of the active-site structure. We obtained four- and three-coordinate Cu sites for [Cu(II)-Y] and [Cu(I)-Y] states, respectively, that corroborate well with the experimental data. The relative energies of these states were validated by a comparison with experimental redox potentials. Collectively, our computational GO models well reproduce the physicochemical characteristics of the individual states, including their redox behaviors.