Protein folding and dynamics: Computer simulation, experiment and theory.

摘要

A protein must be in its unique native state to function. However, an amino acid sequence can, in principle, adopt a vast number of 3-dimensional structures. How does a protein find its native state out of so many possibilities? This “protein folding problem” is a great intellectual challenge, and its solution will have tremendous impact on artificial protein design.; In Chapter 2, we simulate folding on a computer using a lattice model with sidechains. We find that the transition state ensemble (TSE), or folding nucleus, contains non-native interactions which assist folding. This explains abnormal results from recent experiments by other researchers. Furthermore, sequence analyses of SH3-like domains suggests that non-native interactions in the TSE may have been conserved in evolution. Chapter 3 looks at circular permutations with the same lattice model. We show that a permutation at the folding nucleus shifts the nucleus to a new position, but a permutation far away from the nucleus has little effect.; We then switch to an all-atom protein model in Chapter 4. We construct the putative TSE of chymotrypsin inhibitor 2 from experimental &phgr;-values, and critically test the validity of the transition state theory in &phgr;-value analysis. Furthermore, we ask if high &phgr;-values are more important in folding than low &phgr;-values. Our study shows that, first, transition state theory seems valid in protein folding; and secondly, low &phgr;-values in β-strands may be more important than high &phgr;-values in α-helices in committing the protein to the native state.; In Chapters 5 and 7, we study the folding TSE of villin 14T with protein engineering and all-atom computer simulations. &phgr;-value analyses in Chapter 5 show that, although villin has two hydrophobic cores, only one (the “aliphatic” core) is structured in the TSE. Studies in Chapter 7 suggests that the aliphatic core is preferred because of its higher entropy.; Chapter 6 presents a mean-field theory of peptide helices in a lipid membrane, focusing on the helix tilt from the membrane normal. We apply the theory to experimental data from various proteins, and suggest that both lipid disorder and helix hydrophobicity strongly influence the helix tilt.