All-atom Simulations Of Protein Folding And Stability

The folding and stability mechanisms of proteins are unsolved key problems in life science and have been a hotspot of researches. The folding mechanism is referred to the process that protein folds from one-dimensional sequence to the 3-dimensional structure. Understanding this mechanism would greatly accelerate our study on the protein self-assembling, and moreover, be helpful to protein design and to the treatment of various diseases relevant to protein misfolding (such as mad cow disease and Alzheimer disease). And the stability of native and designed proteins is the base of protein function. So it is also very significant to study the stability mechanism of proteins.This thesis includes works of three aspects in protein folding and stability:First, we proposed a new method of increasing the sampling efficiency of protein simulation. Currently, the sampling efficiency in traditional methods, such as Monte Carlo and molecular dynamics, is the big obstacle to simulate protein folding. Up to now, many novel methods have been put out. The most famous of them are Essential Dynamics Sampling (EDS), Amplified Collective Motion (ACM), and Replica Exchange Method (REM). Based on EDS and ACM, we proposed a new sampling method: Directed Essential Dynamics (DED). The main idea of DED is to use the principal component analysis (PCA) to determine six slowest collective motions of peptide every 20fs during the folding process and then add an additional weak force along the combined direction of this motions to adjust the folding direction. This method can make the peptides avoid trapping in the local minima for long time and enhance the sampling efficiency in conformational space during the simulation. As an application, one S-peptide with 15 amino acids is used to demonstrate the DED method. The results show that DED can lead the S-peptide fold quickly into the native state, while the traditional molecular dynamics needs more times to do this.Second, we found that theβ-hairpin trpzip2 can fold into its native state through multiple pathways. Due to the large degrees of freedom, it is difficult to simulate the entire folding processes of large proteins. Therefore, smallβ-hairpins become the focus since they have similar long-range interactions and hydrophobic cores in the protein structures and their folding mechanism would be very helpful to understand those of entire proteins. Forβ-hairpin folding mechanisms, different models have been proposed, including the famous zip-out and hydrophobic cluster models. We studied the trpzip2 and obtained 10 successful entire folding trajectories. The results show the trpzip2 could fold into the native state though multiple pathways, depending on the ways of the formation of the hydrophobic core. This means that the previously proposed folding mechanisms are not in paradox and they could be described in a unified way. Furthermore, we find that the native state of the hairpin does not have the lowest entropy and some non-native states exhibit even lower entropies.Third, we proposed a new method to define inter-residue contact in proteins. Generally, protein stability is thought to be related to its inter-residue contact network. Inter-residue contact is usually defined by the distance between the residues. Our analysis suggests it is more accurate to define the contact with all-atom interaction energy. Our following applications support this. We study fifteen groups of mesophilic and thermophilic proteins and find that the contact energy and contact number are better criterions, comparing with other methods, to distinguish thermophilic proteins from their mesophilic counterparts. Additionally it indicates that the charged-polar and charged-nonpolar contacts are also the key factors for the protein stability, besides the charged-charged contacts. We also apply our method in identifying the key residues in the Gβprotein domain from transducin. We find that most key residues in this protein can be located by the lowest contact energies, but not by distance-defined contacts. This gives a new approach to predict and analyze the key residues in proteins.