Functional Dynamics and Free Energy Evaluation of Biomolecular Systems
Jung-Hsin Lin1,2,3,4*
1Research Center for Applied Sciences, Academia Sinica, Taipei, Taiwan
2Institute of Biomedical Sciences, Academia Sinica, Taipei, Taiwan
3School of Pharmacy, College of Medicine, National Taiwan University, Taipei, Taiwan
4College of Engineering Sciences, Chang Gung University, Taoyuan, Taiwan
* Presenter:Jung-Hsin Lin,
The elucidation of the functional dynamics of biomolecular systems and the evaluation of the free energies involved in the reactions thereof are the principal aims of molecular dynamics simulations. As an outstanding example, I will describe our studies on the functional dynamics of Type IIA topoisomerases (Top2s), which are key DNA manipulating enzymes in both eukaryotes and bacteria, powered by ATP binding and hydrolysis to facilitate the crossing of two DNA duplexes and topological conversion of DNA supercoiling configuration. To analyze the large scale coordinated dynamics of Top2-DNA ligation process, we conducted molecular dynamics simulations at the microsecond time scale, and we observed the slow DNA configurational changes consistent for Top2 catalysis. There were also global functional collective motions between the enzymes and the DNA in the concomitant resealing process. With the steered molecular dynamics simulations, we were able to characterize the gating mechanism of the DNA passage manipulated by Top2. We found the accompanying conformational transition is very similar to the rocker-switch-type movement of membrane transporters in the major facilitator superfamily. Recently, we developed a new computational scheme for evaluating the standard free energy of binding for biomolecular systems, based on calculation of the potential of mean force (PMF) from umbrella simulations along the reaction coordinates of the curvilinear dissociation pathway. We provided a theoretical framework based on statistical mechanics of binding equilibrium for such simulations. A simple variational principle is applied to determine the lower bound PMF, which is subsequently used to derive the standard free energy of binding. The theoretical predicted binding free energies for the benchmark systems agree excellently with the experimental value. Our approach has also been used to discriminate the decoys from protein-protein docking studies. We have also applied this method to protein-small molecule and protein-peptide systems for cancer drug design. Very recently, we have constructed several deep learning neural networks for evaluation of binding free energy for protein-ligand systems, and we have also used such networks to learn the energetics of biomolecules from quantum mechanical calculations for carrying out molecular dynamics simulations.

Keywords: Collective Dynamics, Generalized Correlation, Standard Free Energy of Binding, Deep Learning Neural Network, Molecular Quantum Mechanics