Monte Carlo Simulation Approaches to Protein Stability and Aggregation Prediction

Abstract

A protein’s sequence and set of covalent modifications determine its stability and aggregation propensity in a given environment. Given a change in sequence or covalent structure, we would like to be able to predict the change in stability and tendency to aggregate. Such knowledge would enable us to engineer more stable proteins and to better understand protein misfolding and aggregation diseases. In addition, knowledge of the protein folding pathway and aggregate structure could aid in structure-based design of therapeutics. This thesis employs Monte Carlo simulations to predict protein stability, aggregation propensity, and aggregate structure. First, we describe the use of short unfolding simulations to predict stabilized mutants of the enzyme Dihydrofolate Reductase. Next, we describe a simple model of protein domain swapping that predicts the tendency of proteins to domain swap at intermediate temperature and predicts a concentration dependence where proteins domain swap at intermediate concentration but exhibit non-specific interactions between unfolded proteins at high concentration. Finally, we predict that cataract-associated mutations within γD-crystallin destabilize the protein and that these mutations, along with an experimentally observed disulfide bond, increase the protein’s propensity to aggregate. Based on two-molecule simulations, we propose an aggregation model whereby the N-terminal hairpin of one molecule forms antiparallel beta sheet interactions with the C-terminal domain of the next molecule. We also suggest a mechanism by which the wild-type protein could accelerate mutant aggregation, an experimentally observed phenomenon. We expect our methods to be applicable to stability and aggregation prediction in other proteins.