Combinatorial Statistical Mechanics of Protein Folding

DESCRIPTION (provided by applicant): Here, we address 2 questions: (1) By what kinetic processes do proteins fold up? (2) How can we use that information to develop more efficient computational sampling methods? We focus on theory and computer modeling. Specifically, master equations have recently become the method of choice for understanding folding energy landscapes. (1) We will use them to elucidate the concepts of Transition State and Rate-Limiting Step in folding, to understanding why some proteins fold slowly with complex kinetics and others fold quickly with simple kinetics, and why the folding rates of helical proteins varies more than is predicted by Plaxco-Baker-type correlations. (2) We are combining replica-exchange molecular dynamics (REMD) with conformational zippers (CZ) to search the conformational spaces of all-atom models (such as AMBER7 with the 1996 version of the Cornell et al. forcefield and GB/SA implicit solvation) to find native states. We have exciting preliminary results, showing that this search method is folding three small proteins (GB1, protein A, SH3) to within 2 Angstroms each of their respective native states, currently about 3-4 orders of magnitude faster than Folding@Home simulations. Such work is relevant to the public health because it contributes to understanding the basic properties of proteins in health and disease, and because computational protein modeling is now an important part of computer-based drug discovery, leading to increased rates and reduced costs for developing new drugs.