PROTEIN FOLD SWITCHING IN COARSE-GRAINED MODELS

Abstract: Proteins carry out the instructions encoded in genes and are crucial for many important tasks in the living organism. Often the specific three dimensional structure, or fold, of a protein enables it to carry out a given task. It is an open question whether protein folds have arisen independently or whether mutations to existing protein sequences have caused switches to new folds. In the main part of this thesis, consisting of papers I-III, we have investigated protein fold switching in response to point mutations using computer simuations of coarse-grained mod- els. The basic idea behind coarse-grained models is that by simplifying the description of proteins, the structure of a large number of model sequences can be determined with reduced computational effort. In paper I we exhaustively enumerate protein sequence-structure space in a simple hydrophobic/polar lattice model, exceeding previous enumerations. This enables us in particular to analyze mutational pathways and their stability. In paper II we investigate how protein function changes along a mutational pathway with a sharp switch in fold, using the continuous Cβ model. We find that the switch in fold and preferred binding partner do not coincide and that the change in function is more gradual than the switch in fold. In paper III we hypothesize that fold switches between similar protein folds involve bistable sequences, while fold switches between dissimilar folds occur within a single mutation. In particular, we show that while the fold switches are driven by changes in energy, configurational entropy can play a significant role in determining when a switch occurs. In paper IV, we analyze the hybrid Monte Carlo method for biomolecular simulations, which is based on numerical integration of Newton’s equations of motion. We optimize this method for three systems and find that a nonuniform integration step can reduce the integration error.