Collective Coordinates Tem Perature-accelerated Molecular Dynamics Of Proteins

Accelerated sampling can be realized efficiently by Temperature-Accelerated Molecular Dynamics (TAMD) method. In this methodology, some virtual coordinates are introduced into the physical system as a virtual subsystem, and they are coupled to associated collective coordinates in the physical system. They will be simulated at high temperature, while other physical coordinates will be simulated at normal temperature. As the virtual coordinates are simulated at higher temperature, they have higher kinetic energy to cross potential barriers. For the whole system, some degrees of freedoms are at high temperature, some degrees of freedoms are at low temperature. Its distribution would deviate from the canonical Boltzmann one. The old TAMD methodology did not give solution to recover canonical distribution or ensemble averages. In this paper, under adiabatic assumption, correct canonical distribution and ensemble averages could be obtained. The amplified collective motion method or ACM employed the Anisotropy elastic Network Model (ANM) of protein to find low frequency motion modes of protein. And then put those directions at high temperature to accelerate sampling. Here we combine those two methods to build the ACM-TAMD method. At last, we use TAMD to study the LacR-IPTG unbinding.In Chapter1, we give a brief introduction to some conceptions and some ordinary accelerated or enhanced sampling methods.In Chapter2, we give a detailed introduction to the TAMD method and adiabatic reweighting method to recover canonical distribution and ensemble averages. We use the n-butane system to test the validity and accuracy of adiabatic reweighting. And then, we use the alanine dipeptide system to show the accelerated effect and the capability to recover canonical distribution and ensemble averages. Finally, we use the GB1system to test the practicability of the combined ACM-TAMD method.In Chapter3, ligand dissociating from the inducer-binding domain of the Lac repressor protein is studied by temperature-accelerated molecular dynamics (TAMD) simulations. With TAMD, ligand dissociation could be observed within relatively short simulation time. This allows many dissociation trajectories to be sampled. Under the adiabatic approximation of TAMD, all but one degree of freedom of the system are sampled from usual canonical ensembles at room temperature. Thus, meaningful statistical analyses could be carried out on the trajectories. A systematic approach is proposed to analyze possible correlations between ligand dissociation and fluctuations of various protein conformational coordinates. These analyses employ relative entropies, allowing both linear and nonlinear correlations to be considered. By applying the simulation and analysis methods to the inducer binding domain of the Lac repressor protein, we find that ligand dissociation from this protein correlated mainly with fluctuations of side-chain conformations of a few residues that surround the binding pocket. In addition, the two binding sites of the dimeric protein are dynamically coupled:occupation of one site by an inducer molecule could significantly reduce or slow down conformational dynamics around the other binding pocket.