Building Quantum Calculation-Based Protein Polarization Effect Into Protein-Inhibitor Binding Dynamics for Lead Molecule Prioritization

Abstract

Reliable quantitative description of protein structure and dynamics of protein-inhibitor binding is quintessential for the evaluation of novel inhibitor designs and electrostatic interactions play a critical role in achieving this. The local electrostatic environment inside a folded protein is not homogeneous but largely determined by amino-acid residue location, specific conformation and dielectric environment. However, conventional molecular mechanical force fields (like AMBER) employ fixed atomic charges, which makes them incapable of accurately modeling biomolecular interactions. Therefore, current research focus is to enable incorporation of polarization effects into force field based molecular simulations.

Towards this end, we assembled a robust simulation pipeline for attempting quantum mechanical calculation of true polarization state of protein near its native structure, using linear-scaling molecular fragmentation coupled with continuum-solvent model. Briefly, electron densities of protein’s residue-based fragments were obtained by quantum mechanical calculations. Partial charges thus generated were used to calculate solvent reaction field at protein surface using Poison-Boltzmann method. Induced surface charges mimicking the solvation effect, were then added as additional background charges in subsequent quantum calculations of fragments. Protein and solvent charges polarized each other until convergence was reached which were then successfully employed in AMBER ff03-based MD simulations for correctly representing protein-ligand electrostatics in MMGBSA calculations.

This pipeline, well suited for parallel computation,enabled full quantum mechanical calculation of large proteins and also ensured an improved contribution of electrostatic interactions, in protein-inhibitor MD simulations, towards total binding free energy. Developed protocol was validated by ranking binding affinities of known pre-clinical inhibitors for a Rheumatoid arthritis druggable target, which notably provided a good correlation with their reported IC50 values. It was also employed for optimization of novel lead molecule series for the same target, which now await experimental validation.