Study On Protein Folding With United-Residue Model

Study protein folding and protein folding mechanisms is a very difficult but also attractive task. In the past decades, protein folding research has achieved lots of successes, but there are still many problems need to be solved. My research is focused on studying protein folding mechanisms with United-Residue (UNRES) model and improving UNRES force field. Our research can help people understanding both small peptides and real proteins folding mechanisms. We also did some researches about UNRES force field. More detiales about my research are listed as following:1) We employed UNRES to study a short peptide: Trptophan Zipper 2(Trpzip2) folding mechanisms. We find that this small peptide folds from extended structure to native structure following two folding mechanisms: zipper and simultaneously zipper & hydrophobic collapse. Each folding mechanism appears with different probabilty. The "Zipper" mechanism appears with a probabilty around 72%, this may explain why only "zipper" mechanism observed in the experimental research, since the experimental research always oberved the most probable events. We also find that the folding process of trpzip2 is not only a two states process but also a three states process. Our results may help to clarify the inconsistencies in the current picture of the folding mechanisms ofβhairpins.2) In the following, we extended the application of UNRES to real protein. We employed UNRES approach to study folding processes of a Six-helix protein domain (the C-terminal domain of Ku86 protein, PDB ID: 1Q2Z). We simulated 48 110-ns independent molecular dynamics trajectories with UNRES starting from extended conformations of this protein. This protein successfully folds into its native state and the results show that its folding process is relatively simple: firstly the secondary structures form very fast with the hydrophobic collapse, then helix pairs form and finally these pairs assemble into the tertiary structure. We also find the first-half (the first three helices) and last-half (the last three helices) parts of this protein can fold into their native conformations independently and this suggests that this protein may be evolved from smaller polypeptides or proteins. 3) We cut six helix-bundle protein (PDB ID:1Q2Z) into slices by different ways. Then a series of molecular dynamics simulations with these slices were carried out by UNRES. The results shows that the slices containing three helices are more stable than the slices containing two helices. As shown in our sequence alignment results, the first half of 1Q2Z has sequence similarity to the second half. So we have reason to suppose that 1Q2Z may consist of two parts/fragments. Besides, all these slices show the same folding behaves as the whole 1Q2Z protein.4) We also did something on UNRES force field. UNRES force field is more and more elaborate, lots of energy term parameters have been changed, so this new UNRES force field needs to be reparameterized. Based on Prof. Liwo's unpublished work, we knew the parameter space of UNRES. We introduced a method which can be used to decide initial parameter sets for UNRES optimization. Recently, temperature-dependence and new energy term parameters have been applied to UNRES in order to develop a more powerful and reliable force field and thus the UNRES force field parameters (weight of energy terms) must be reparameterized. Therefore, a good initial parameter set for UNRES force field optimization will be needed. With this goal, we randomly generated parameter sets in the parameter space and applied a hierarchical selection procedure to obtain appropriate initial parameter sets for UNRES force field optimization. A training set which contains 13 proteins was used to test the folding ability of these parameter sets. The results show that most proteins in our training set reach native-like structures with the parameter sets selected by our procedure. Our results suggest that it is possible to randomly search the force field parameter space to find good initial parameter sets and our procedure could be used for gross selection of good initial parameter sets for UNRES force field optimization. We also suggest that it is possible to find a better force field directly though further extensively searching of UNRES parameter space.This work is supported by the National Natural Science Foundation of China under Grant No.30525037 and No.30470412.