Development Of A Well-balanced Protein Force Field And Application In Protein Folding Study

It has been about37years for the first molecular dynamics simulation, which opens up the "structure to function" area in biophysics. Molecular dynamics simulations unravel the properties of kinetics and thermodynamics of biomolecules along time on atomic level, which provide a powerful tool for revealing important biological phenomena and exploring the mechanism of interaction of biomolecules. In addition, molecular dynamics simulation also become a significant complementary tool for experiments. Many problems, such as ionization constant of amino acids and the barrier of enzyme reaction, have been calculated and solved by experimental scientists, utilizing molecular dynamics simulations. Currently, there are two bottlenecks molecular dynamics simulations encountering:1) the simulation time, which is the sampling in the entire phase;2) the accuracy of force fields. By the development of computational power, the first problem is less limited in simulations. The simulation time for BPTI protein, which has58amino acids, has reached1ms, which realizes the transition from microscope to mesoscope. Thus, whether force fields are accurate enough determines whether the molecular dynamics simulations are valid. A little error in the energy surface of force fields may cause the enormous conformational change of systems. The development of molecular force fields plays an important role in calculating the dynamical properties and investigating the folding/unfolding, aggregation, and conformational change of biomolecules.Although molecular force fields have been widely used, there are still some inherent deficiencies, which is more and more clearly realized. In previous research in our group, the calculated binding energy between protein and ligand deviates severely from experimental value by utilizing current AMBER force fields. The simulation of WW-domain (all (3structure) performed by Schulten et. al., utilizing CHARMM22force field, turns the system to all helix structure when it is at its equilibrium. It has been proved that the deficiencies lie in electrostatics and torsional terms.Biomolecules are generally in solution, so the environment plays an important effect in the stability and other dynamical properties. The polarization effect is very remarkable in polar dielectrics, especially in water. However, the polarization effect is not included in current force fields, such as AMBER, CHARMM, GROMOS etc., in which the atomic charges of amino acid residues are fixed, no matter how the environment change. This causes unauthentic results of molecular dynamics simulations. The deficiency in torsional term has been known in recent years. During the parameterizing of force fields, torsional term are the last one to be parameterized, which makes all kinds of errors (systemic errors and other term errors) are included to torsional term. It is informally called the garbage terms. In current force fields, torsional term (φ(C-N-Ca-C),Ψ(N-Ca-C-N)) is described as one dimension Fourier expansion, and the parameterization is mainly focused on several important areas (usually low energy areas) of the energy surface, which causes the inaccurate barrier between two different conformations, and makes the force field biases the specific secondary structure.Based on the deficiencies of electrostatic and torsional terms, we first propose a simple and effective polarizable charge model, which is fitted on a pair of alanine dipeptides along the hydrogen bond distance. By employing the charge scheme, we successfully investigate the folding mechanisms of a short helix in implicit water and a long helix in explicit water. However, it is not enough to fold β structure to its native state if only polarization effect is taken into consideration due to the inappropriate secondary population distribution of current force fields. Employing the high level quantum chemical calculation, we discard the torsional format of current force fields, and propose a new force field (2D force field) based on2-dimension Fourier expansion. The new developed force fields have been proved to be better than current force fields by simulations of different benchmark systems, including dipeptides, tripeptide, tetrapeptide, and proteins. The development of the new force fields provides a powerful tool for molecular dynamics simulations.