Microscopic pKa analysis of Glu286 in cytochrome c oxidase (Rhodobacter sphaeroides): toward a calibrated molecular model.

As stringent tests for the molecular model and computational protocol, microscopic pK(a) calculations are performed for the key residue, Glu286, in cytochrome c oxidase (CcO) using a combined quantum mechanical/molecular mechanical (QM/MM) potential and a thermodynamic integration protocol. The impact of the number of water molecules in the hydrophobic cavity and protonation state of several key residues (e.g., His334, Cu(B)-bound water, and PRD(a3)) on the computed microscopic pK(a) values of Glu286 has been systematically examined. To help evaluate the systematic errors in the QM/MM-based protocol, microscopic pK(a) calculations have also been carried out for sites in a soluble protein (Asp70 in T4 lysozyme) and a better-characterized membrane protein (Asp85 in bacteriorhodopsin). Overall, the results show a significant degree of internal consistency and reproducibility that support the effectiveness of the computational framework. Although the number of water molecules in the hydrophobic cavity does not greatly influence the computed pK(a) of Glu286, the protonation states of several residues, some of which are rather far away, have more significant impacts. Adopting the standard protonation state for all titratable residues leaves a large net charge on the system and a significantly elevated pK(a) for Glu286, highlighting that any attempt to address the energetics of proton transfers in CcO at a microscopic level should carefully select the protonation state of residues, even those not in the immediate neighborhood of the active site. The calculations indirectly argue against the deprotonation of His334 for the proton pumping process, although further studies that explicitly compute its pK(a) are required for a more conclusive statement. Finally, the deprotonated Glu286 is found to be in a stable water-mediated connection with PRD(a3) for at least several nanoseconds when this presumed pumping site is protonated. This does not support the proposed role of Glu286 as a robust gating valve that prevents proton leakage, although a conclusive statement awaits a more elaborate characterization of the Glu286-PRD(a3) connectivity with free energy simulations and a protonated PRD(a3). The large sets of microscopic simulations performed here have provided useful guidance to the establishment of a meaningful molecular model and effective computational protocol for explicitly analyzing the proton transfer kinetics in CcO, which is required for answering key questions regarding the pumping function of this fascinating and complex system.