| dc.description.abstract | Since nuclear magnetic resonance (NMR) spectroscopy data, including solution NMR from micelles and solid-state NMR from bilayers, provide valuable structural and dynamics information of membrane proteins, they are commonly used as restraints in structural determination methods for membrane proteins. However, most of these methods determine the protein structures by fitting the single-confer model into all available NMR restraints regardless of the explicit environmental effects that are determinant in the structures of membrane proteins. To develop a reliable protocol for obtaining optimal structures of membrane proteins in their native-like environments, various NMR properties were applied in the refinement approaches using explicit molecular dynamics (MD) simulations in this research. First, solution NMR NOE based-distance measurements were used as restraints in MD simulations to refine an activating immunoreceptor complex in explicit environments. Compared to the structure determined in vacuum, the resulting structures from the explicit restrained simulations yields a more favorable and realistic side-chain arrangement of a key Asp residue, which is highly consistent with mutagenesis studies on such residue. Incorporating solid-state NMR and solution NMR, MD simulations were performed in the explicit bilayers to refine the structure of membrane-bound Pf1 coat protein. Since solid-state NMR is sparse in its N-terminal periplasmic helix, the protein structure was determined by combining solid-state NMR and solution NMR. Benefiting from the sophisticated energy function and the explicit environments in MD, the orientation of Pf1's periplasmic helix can be identified in simulations restrained by solid-state NMR alone. In the simulations restrained with both solid-state NMR and solution NMR, physically irrelevant structures were frequently observed, suggesting there are conflicts between the restraints from different sample types (e.g., bilayers and micelles). As NMR data are ensemble-averaged measures, the solid-state NMR restrained explicit ensemble dynamics (ED) simulations of fd coat protein were performed in different ensemble sizes and compared to the unrestrained MD simulations. As the ensemble size increases, the violations of resulting structures from experimental NMR data decrease, while the structural variations increase to be comparable to the unrestrained MD simulations, indicating the efficacy of restrained ED in refining structures and extracting dynamics. To investigate the influence of different environments on the structures of membrane proteins, in this research, MD simulations were performed in bilayers and micelles, respectively. Since building a preassembled protein/micelle complex for MD simulation is challenging and requires considerable experience with simulation software, a web-based graphical interface Micelle Builder in CHARMM-GUI (http://www.charmm-gui.org/input/micelle) was developed to support users to build micelle systems in a automatic and simplified process. Using this interface, Pf1 coat protein was preassembled in a protein/micelle model and simulated in explicit environment. Compared to previous simulations of Pf1 coat protein in bilayers, different protein conformations were observed in these simulations due to the distinct behavior and geometry of micelles. | |
