Innovative drug design technologies from computational mapping of energy, entropy, and free energy in proteins and protein-ligand systems

Project Summary/Abstract This project aims to advance our understanding of structural mechanisms underlying protein-small molecule binding – and ultimately of protein function – by developing novel computational methods for thermodynamic mapping of proteins, ligands, and the surrounding aqueous solvent. Despite significant progress in structural and computational research, the determinants of protein-ligand binding thermodynamics are not fully understood, so it is difficult to reliably design ligands that will bind a targeted protein with high affinity. State-of- the-art free energy perturbation (FEP) and related methods of computing protein-ligand binding free energies can predict binding affinities with useful accuracy, but do not provide the insights needed to guide successive rounds of ligands with improved affinity. Moreover, experimental studies have revealed poorly understood couplings between binding sites and distant protein regions, highlighting opportunities to develop a deeper understanding of how energy and entropy redistribute throughout proteins when they bind other molecules, including small molecule drugs. We aim to address these challenges by generalizing concepts of thermodynamic densities from liquids to proteins and ligands, creating a new window into fundamental biomolecular processes. We will develop theory, algorithms, and open-source software for generating structure-based maps of free energy, entropy, and energy in proteins and surrounding water by analyzing molecular dynamics (MD) simulation trajectories. This work builds upon recent advances unifying the theory of the entropy density of a liquid with the mutual information expansion – hitherto only applicable to solutes – and defining the potential energy density for arbitrary potentials. The software will be shared on GitHub for community evaluation, use, and development. We will then apply these novel tools to map thermodynamic determinants of ligand affinity and guide computational ligand optimization. By assigning thermodynamic contributions to binding to specific system components, such as ligand functional groups and protein sub-pockets, the new method will provide guidance toward the design of higher-affinity ligands. We will also explore use of the new method to help identify ligandable cryptic sites in proteins, testing two key hypotheses: 1) cryptic binding site formation is favored in high local free energy regions of the apoprotein, and 2) ligands bind with highest affinity to pockets filled with thermodynamically unfavorable solvent but with favorable intrinsic protein thermodynamics. These concepts also are expected to be useful in broader aspects of molecular biophysics, such as understanding protein stability, energy storage and release in molecular motors, allostery, and the free energy barriers controlling biomolecular kinetics.