This project focuses on the study of membranes, proteins and carbohydrates by molecular dynamics computer simulation. Progress is reported under each Aim listed above Aim 1. Understand Model Membranes. Methods for evaluating the lateral stress profile of lipid bilayers and inverted micelles have been developed and introduced into CHARMM. The radius of curvature (the inverse of the spontaneous curvature) of POPC and POPE calculated from simulation are in near quantitative agreement with experiment. This not only provides strong validation of the recently developed C36 lipid force field (3), but also underscores the complexity of the profile. Radii of curvature for polyunsaturated lipids obtained from simulation are near to those of POPC, a somewhat surprising result given their high flexibility.
Aim 2. Develop Simulation Methodology. A thorough evaluation of the concentration dependence of TIP3P water/carbohydrate solutions was published (1). The results provide bounds on the limits for viscosity scaling when comparing, among other quantities, simulated and experimental NMR relaxation times. Efforts to extend the C36 lipid force field (3) focused on charged lipids, including phospatidylserine, cardiolipin, and phosphoinositols. Preliminary simulations indicated that the interaction of Na+ and lipid head group atoms is too strong with the current default CHARMM parameters, leading to an underestimate of the experimentally obtained zeta potentials and overestimates of the chain deuterium order parameters. A modification to the interaction parameters based on the concentration dependent osmotic pressure of model solutes in ionic solutions reduced the binding. This in turn increased the surface area of the bilayers, and improved agreement with experimental zeta potentials and order parameters. This work greatly expands the reach of membrane simulations. A web-based version of Grand Canonical Monte Carlo/Brownian Dynamics (CCMC/BD) for the study of ion channels was designed. Simulation results for the voltage dependent anion channel (VDAC), alpha-Hemolysin, and the protective antigen pore of the anthrax toxin (PA), were presented to illustrate system setup, input preparation, and typical output (conductance, ion density profile, ion selectivity, and ion asymmetry). The accuracy of a hydrodynamics based model of diffusion through a tube was demonstrated. While simulation and experimental results are in good agreement for VDAC and alpha-Hemolysin, simulated ion conduction values overestimate experimental values for PA by a factor of 1.5 to 7 (depending on His protonation state and the transmembrane potential). This implies that the currently available computational model of this protein requires further structural refinement. The paper has been accepted for publication, but is not yet in press.
Aim 3. Simulate Complex Membranes The study of coarse grained (CG) simulations of different ratios of PEGylated and normal lipids described in last years report was submitted and published (5). The combined experimental and theoretical study of tethered pleckstrin homology (PH) domains was also published (2). Subsequent CG simulations of LacY in a DPPC bilayer yielded the same factor of two difference in the diffusion constants of monomers and dimers connected with 60 Angstrom tether as published for tethered lipids (to be submitted). These results further confirm the inapplicability of Saffman-Delbruck model with the usual parameters. Molecular Dynamics (MD) simulations (3) of ion transport in the Voltage Dependent Anion Channel (VDAC) yielded excellent agreement with experimental conductance and ion selectivity. The simulations showed probable paths of ions as they crossed the channel, leading to a proposed mechanism for anion selectivity: the rate for K+ is smaller than that for Cl- because of the attractive interactions between K+ and residues on the channel wall. Grand Canonical Monte Carlo/Brownian Dynamics (CCMC/BD) simulations (4) of VDAC using diffusion constants extracted from the MD simulations were used to access 20 NMR models, and assorted mutants. Based on the results, it is likely that some of the NMR models are flawed (their predicted conductances and ion selectivities do not agree well with experiment). It was also shown that the difference in ion selectivity between the wild-type and the mutants is the result of altered potential of mean force profiles that are dominated by the electrostatic interactions. Based on the success of the CCMC/BD simulation of VDAC, the method was applied to examine pore blockage of alpha-Hemolysin by polyethylene glycol (PEG). Different rigid solution structures of PEG29 were placed in the pore and ion conductance calculated. Blockage by rigid PEG is significantly less than observed experimentally, indicating that either flexibility of PEG is critical for blockage, or that the solution and channel conformations are substantially different. All atom MD simulations of alpha-Hemolysin/PEG systems are in progress, as are extensions of BD simulation method to include flexibility. A combined experimental and simulation study showed definitively that corralling by a fence is the underlying mechanism for pooling of PIP2 on the surface of nascent phagosomes (7). The nature of the fence remains unknown. Simulations and experiments ruled out actin. Further simulations indicate that septin can retard PIP2 diffusion, but only when inserted at least 15 Angstroms in the membrane (to be submitted).
Showing the most recent 10 out of 52 publications
