Molecular Simulation Study of Protein Conformation Change, Binding Mechanisms, and Allosteric Communication
Jason G. Pattis · OpenCommons at University of Connecticut (University of Connecticut) · 2019
Many processes in biology involve conformational changes or binding events which can be described by a pathway or ensemble of pathways. These processes are challenging to study experimentally as obtaining the temporal and spatial resolution sufficient to understand the underlying physical mechanisms can be challenging. Molecular dynamics (MD) simulations is a powerful tool that can provide atomic resolution on these processes and aid in the design and interpretation of experiments. In this thesis, I will describe MD simulations using enhanced sampling methods to investigate several biomolecular systems, including the Lassa virus nucleoprotein, phosphodiesterase enzymes and a peptide from the Flock House virus (FHV). The Lassa virus nucleoprotein (NP) has two domains and the N-terminal domain binds the single-stranded RNA genome. We initially focused on the N-terminal domain where PCA as well as metadynamics revealed a large energy barrier for NP opening the RNA binding pocket without RNA and a small barrier when bound. Anti-correlated motions in the transition state suggests NP may partially open and make initial contact with RNA, which then facilitates full opening and binding. Further studies on the full-length NP were motivated by Hydrogen/Deuterium exchange data, which suggested disruption of an NP trimer may generate opening of the RNA binding pocket. From long timescale simulations and a two-stage adaptive sampling scheme, we constructed a Markov-State Model to describes the dynamics of the full-length NP in a trimer dissociated state. The model revealed domain level conformational changes as well as changes near the RNA-binding pocket including shifting out of helix 8 and 9 which may allow room for RNA to contact the binding pocket. Phosphodiesterase 6 (PDE6) is an enzyme in the vision signaling pathway and has high sequence similiarity to PDE5 but a large difference in its catalytic rate of cGMP. Simulations revealed correlated motions between helix 12, which is far from the binding pocket, and H- and M-loops in PDE6 but not PDE5, which may explain difference in substrate access and binding. Finally we use TICA to evaluate the confidence in umbrella sampling calculations of folding of FHV gamma peptide on membranes of different lipid compositions.