Advances in Coarse-grained Models for Protein Folding and Protein-protein Interactions

Abstract

Almost all biological functions rely on the dynamics of proteins. Protein folding and protein-protein interactions are the two most fundamental problems of protein dynamics. Molecular dynamics (MD) simulation is a powerful tool to elucidate the mechanism of protein folding and protein-protein interactions by providing atomic-level resolution. However, the timescales of protein folding and protein association-dissociation are often not accessible by all-atom MD simulations due to the high computational cost. Coarse-grained models effectively address this issue by reducing the degrees of freedom of the system to only a few that are essential for the properties to be studied. In this dissertation, I describe the developments and applications of coarse-grained modeling of protein folding and protein-protein interactions through several case studies. I first present a computational study of the folding mechanism of a triosephosphate isomerase (TIM) barrel protein using a coarse-grained model. This is the first time this model was used to study a large protein with more than 200 amino acid residues. From the simulations, we proposed a 3-channel folding mechanism with one major and two minor folding pathways. The simulations show overall good agreements with the experiments in capturing the regions that are first to fold and capturing a rate-limiting intermediate state found in the major folding channel. The simulations advance our understanding of the folding mechanism of this TIM barrel proteins by directly providing structural details of the protein folding intermediates as suggested by experiments. In the realm of protein-protein interactions, I developed a new sampling method based on Hamiltonian replica exchange (HREX) that allows efficient calibration of the coarse-grained model and fast calculation of the dissociation constant. This HREX method was used to study the protein-protein interaction in the context of the allosteric regulations in the KIX and TAZ1 domain of the CBP/P300 transcription coactivator. The simulations captured both the positive/cooperative allosteric effect in the KIX domain and the negative allosteric effect in the TAZ1 protein switch with two vastly different allosteric mechanisms. The simulations suggest the positive allosteric regulation in KIX, in which a prebound ligand favors the binding of the second ligand, is due to a favorable entropic change. Whereas the negative allosteric regulation in TAZ1, in which two ligands compete for binding the same target, is mainly driven by long-range electrostatic forces. The simulations also suggest the importance of electrostatics for the coarse-grained model to be generally successful in modeling the allosteric effect involving intrinsically disordered proteins. These studies advance our understanding of the allosteric effect by generating testable hypotheses that help quantify the problem.