A Network Approach For Studying Protein Folding

The protein folding is an important problem in the life science. A protein molecule can fulfill their biological function, only when they fold into their specific three-dimensional structure. Protein misfolding can lead to disease, e.g., misfolding and aggregate prion protein leads to the mad cow disease. Therefore, understanding the protein folding mechanism would greatly help us to study the protein self-assembling and protein design. Therefore it is very significant to study the protein folding pathways and transition states efficiently and accurately.The main works in this thesis is as follows:(1)Protein folding network analysis is an effective approach to investigate the high-dimensional free energy surface of peptide and protein folding and the method can avoid the limitations of simple projection of free energy surface based on a few order parameters. We present improvements in the effectiveness and accuracy to the folding network analysis method based on Markov Cluster (MCL) algorithm. We applied this approach to investigate the folding free energy landscape of the beta-hairpin peptide trpzip2 and found that the folding network method is able to determine the basins and folding paths of trpzip2 more clearly and accurately than the two-dimensional free energy projection method.(2) Identifying folding transition state structures is crucial to study protein folding pathways and folding dynamics. Protein folding transition state structures cannot be obtained directly from experiments but can be investigated by using molecular dynamics simulations. Although, there are various methods that utilize data from molecular dynamics simulations to identify the folding transition state structures their results often different notably. In this paper we present a novel approach to find the folding transition state structures based on the free energy along the generalized path length of the folding network. By using this approach we analyzed the folding transition state structures of the beta-hairpin peptide trpzip2 and the results show that our method can find different kinds of the folding transition state structures identified by other methods. This shows that our method can provide more complete information for the folding transition states of proteins.(3) The FEPL method based on the folding network has also been applied to study the downhill folding process of the beta-hairpin trpzip2. The results show that there are three types of folding mechanisms of the trpzip2 in the same folding conditions:the multistate folding, two-state folding and downhill folding. The downhill folding process of trpzip2 has not been discovered in the previous studies. The result further demonstrates that the network approach can study the protein folding completely.