| It's about 40 years since the emergence of molecular dynamics(MD) simulations, it has become a very popular scientific tool in physics, chemistry, material and the like subjects. However, this method has two unresolved issues: firstly, the time of ergodic phase space; secondly, the accuracy of the force field. The most important part of MD is a potential energy function, whose independent variable is the atomic coordinate. When we study the simulations of protein folding as well as the interaction of the protein-ligand, we try to look for their thermodynamics equilibrium of the systems, in other words, we are looking for the location where is the minimum value of the potential energy. Because of the limitations of laboratory equipment and the simulation time, we can't calculate the potential energy surface by quantum method, completely. There are many popular standard force fields, such as AMBER and CHARMM. In these force fields, the potential energy is replaced by the energy of the change of bonds, angles, dihedral angles and nonbond interaction. They consider the same residue in the different proteins has the same charge, and the charge doesn't change whether the environment alters. To some extent, this method is reasonable, and the calculation is very simple. In fact, these standard force field can't show the accurate electronic environment, which affects the accuracy of simulations. Several years ago, polarized force field has solved this issue well. This force field is more complex than the standard force fields, more and more people has applied this method in the MD simulations. We use polarized protein-specific charge(PPC) in our study. With the development of technology, the performance of CPU is better than before. Also, GPU and other equipment have been applied to the field of biological system simulations, the length of time in protein folding we can simulate becomes longer and longer. If we want to reduce the time of ergodic phase space, we should improve the method except for the equipment. Recently, accelerated molecular dynamics(aMD) was proposed by McCammon group. They add a bias potential to the real potential energy, so the energy barriers become lower, the simulation process is accelerated.There are four parts in this work. In the first part, we give a brief summary of protein folding and the interaction between protein and ligand. In the second part, we introduce some basic theories of MD simulations, such as the molecular force fields, solvent models, polarization effect and the method of aMD and molecular fractionation with conjugate caps approach(MFCC). My main research results during the graduate school are shown in the third part. This part include two sections, one is about the effect of electronic polarization to human ?-thrombin, the other is about all-atom direct folding simulation for proteins using the accelerated molecular dynamics in implicit solvent model. In the last part, I summarize my results, and imagine the future of our research field. The third part is the main part and it is summarized as follows:The first section introduces MFCC incorporated the Poisson-Boltzmann model. We use this method to calculate the polarized protein-specific charges. Then we study the effect of polarization by comparing PPC and AMBER charge. The results show that the PPC can correctly describe the polarized state of the thrombin and L86. Especially, the RMSD of backbone atoms and the hydrogen bonds are more stable using PPC than AMBER charge. The present results indicate that protein polarization plays critical roles in maintaining the compact structure of thrombin.In the second section, we report the results of protein folding(2I9M, C34, N36, 2KES, 2KHK) by the method of accelerated molecular dynamics(aMD) at room temperature with the implicit solvent model. In general, the potential energy barriers are always very high during the molecular dynamics(MD) simulations of protein folding, so the systems are trapped in one or another local minimum for long simulation time. And the rates of the systems moving from one potential energy basin to another are quite low. In order to increase the rates, the aMD which is based on the MD is proposed by McCammon group. The primary difference between them is that the aMD adds a robust bias potential energy to the real one, so the heights of local barriers are reduced. Starting from the linear structures, these proteins successfully fold to the native structure in a 100ns-aMD simulation. In contrast, they are failed under the traditional MD simulation in the same simulation time. Here, we investigate evolutionary process of folding under aMD, we find different proteins have various pathways of folding. Then we find the lowest root mean square deviation(RMSD) of helix structures from the native structures are much lower by aMD contrasting to those by MD, and their values are 0.36?, 0.63?, 0.52?, 1.1?, 0.78? vs 1.8?, 4.2?, 2.8?, 2.4?, 2.3? respectively. For further study, the cluster analysis shows that the representative structures which are the top three occupied clusters of each protein are native structures by aMD. The time of proteins folding from linear configurations to their corresponding native structures are shorter by aMD than those by MD for these proteins. The percentages of native contacts of the five proteins from native structures fluctuate between 90% and 100% by aMD. Similarly, the energy analysis of the aMD is in agreement with the experiment results. All analyses show the aMD is a wonderful method, and we may apply it to the field of drug design.In the third section, we introduce the general process in drug design and we mainly analysis the influence of water in the CDK2 and X64 complex's binding pocket. We analysis the RMSD of protein's backbone atoms from the native structure, and we find that the RMSD of CDK2 using PPC is more stable. We also analysis the occupancy of H-bonds in the water bridge, also, PPC is better than AMBER. |
PDF Full Text Request