Simulation of redox processes in complex environments
Embargo Date
2025-04-01
OA Version
Citation
Abstract
The predictive simulation of redox processes in solvated systems is a challenge even for modern computational chemistry methods, due to the need for reliable description of interactions that occur on multiple time- and length- scales. One has to properly take into account both the reactant as well as environment dynamical structural fluctuations, and accurately describe short-range specific and long-range electrostatic interactions. Moreover, it has been shown that environment polarization is crucial for the numerical accuracy of computed potentials.
This work addresses these fundamental challenges as well as the feasibility and automation of a computational infrastructure for the predictive simulation of redox processes for complex systems. We use a linear response approximation (LRA) based protocol that combines molecular dynamics (MD) for configurational sampling with a polarizable quantum mechanics/molecular mechanics (QM/MM) embedding model for energies' evaluation. We focus on two classes of model systems: (i) small organic molecules in homogeneous solvents, and (ii) co-factors in proteins. Small redox-active molecules have been used as model systems to fine-tune the multi-step protocol and to explore the role of environment polarization, in particular, the dependence of the shifts in the computed redox potentials on the extent of hole delocalization. The newly automated infrastructure was then utilized to study redox-active protein systems of biochemical interest: a Green Fluorescent Protein (GFP) and azurin. Our simulations resulted in the first estimate of the GFP redox potential that properly accounts for protein structural fluctuations, and additionally revealed the significant effect of the concentration of counterions used in MD simulations on the computed redox potential. Our preliminary studies of the azurin redox potential point to a significant contribution from the electrostatic interactions of the redox-active center and the protein environment to the overall reduction free energy and reduction potential. The results of this work align with previous studies pointing to the key role of the environment polarization for quantitative estimates of redox potentials both in homogeneous solvents and in protein environments, and a correlation has been drawn between the extent of polarization effects and the delocalization of the hole upon ionization. Additionally, a flexible, customizable computational infrastructure has been developed and fully documented for application of the protocol used in this work for in-house use by the Bravaya Group.