Molecular dynamics studies of peptide, nanoparticle, and lipid interactions using multiscale simulations.

Molecular dynamics (MD) simulations of peptide-monolayer, nanoparticle-bilayer, and peptide-peptide interactions were performed to predict experimentally measured properties and to understand the atomic-scale interactions. Firstly, simulations of lung-surfactant peptide SP-B25 in a palmitic acid monolayer were performed with various initial configurations. Starting with initial peptide orientation perpendicular to the monolayer, the predicted final tilt angles average 54° ∼ 62° with respect to the monolayer normal, similar to those measured experimentally. Hydrogen bonding analyses show that Arg-12 and Arg-17 help anchor the peptide to the monolayer, and Y-7 and Q-19 help control the tilt of the peptides in the monolayer, suggesting that the factors controlling orientation of peptides can be uncovered through MD simulations. Secondly, to simulate the interactions of polyamidoamine dendrimers with a lipid bilayer in water, a coarse-grained model based on that of Marrink was developed, and validated by showing that the radius of gyration predicted for the coarse-grained dendrimers agrees with the values from atomistic simulations and from experiments. Simulations of the interactions of these dendrimers with lipid bilayers show that pore formation is induced by unacetylated (cationic) generation 5 (G5) dendrimers in the absence of salt, but not by G3 dendrimers nor by fully acetylated (uncharged) dendrimers nor by unacetylated G5 in the presence of high salt concentrations. All of these results are in agreement with experiments, suggesting that coarse-grained MD simulations can predict membrane-disrupting properties of nanoparticles at the microsecond timescale. Thirdly, coiled-coil peptides were simulated, and their alpha-helicities, and inter- and intra-helical electrostatics were compared with experimental data. Although the effects of alpha-helicity and electrostatic interactions are similar to what is seen experimentally, the values of helicity are not in quantitative agreement with experimental values, suggesting that simulations cannot yet accurately predict helical propensity either because of the timescale limitation of simulations or inadequacy of the forcefields. These results suggest that MD simulations are useful to verify experimental data as well as to understand atomic-scale interactions. However, there are still limitations of simulation accuracy and timescale, and thus more accurate forcefields and advanced methods for improving equilibration should be developed in the future.PhDBiological SciencesBiophysicsMolecular physicsPhysical chemistryPure SciencesUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/126475/2/3253323.pd