HIERARCHICAL APPROACH TO PREDICTING TRANSPORT PROPERTIES OF A GRAMICIDIN ION CHANNEL WITHIN A LIPID BILAYER
A hierarchical computational approach combining molecular dynamics and continuous Poisson-Nernst-Planck theory is developed in this study to simulate transport properties of the Gramicidin A (GA) ion channel within a DMPC (dimyristoyl phosphatidylcholine) lipid bilayer. The first part of this study adopts a hybrid molecular dynamics simulation to investigate the diffusion of Na and K ions at different positions along the GA channel in both single- and double-occupied states. An analysis of the radial distribution functions suggests that the single-occupied state is more favorable for both Na and K ions than the double occupied state. In double-occupancy, K favors a state with six water molecules between the two ions, in which water-channel interactions play an important role. Self-diffusion coefficients for single Na and K ions in the GA channel were determined from molecular dynamics (CHARMM force field) to be 4.71*10-7 cm2/s and 6.22*10-7 cm2/s, respectively. The Nernst-Einstein (N-E) relation gives maximum ionic conductivities of 37 pS and 49 pS for the single Na and K occupied GA channel, respectively. These values, have the same order of magnitudes as the experimental data and, therefore, suggest that the N-E relation is useful in predicting the conductivity of an ion channel from its diffusion coefficient. The second half of the study implements a three-dimensional (3D) Poisson-Nernst-Planck (PNP) calculation to predict conductance of the GA channel in the DMPC membranes. No free parameters were used during the calculation. Partial charge distributions of the GA protein and lipid molecules are assigned using the Poisson-Boltzman module embedded in the CHARMM force field, and the diffusion coefficients obtained from the MD simulation are used. This study shows that DMPC electrostatics have significant influence on the channel conductivity. At low electrolyte concentrations, the channel can not be occupied by more than one monovalent cation. With varied diffusion coefficients along the channel, the 3D-PNP predictions replicate the experimental current-voltage relations for the channel immersed in an aqueous NaCl bath solution. Nernst potentials at two asymmetric salt conditions between two sides of membrane are also predicted to be in good agreement with theoretical values. The successful predictions for the GA system suggest that the MD and PNP simulations can be used to investigate ion transport in other biological ion channel systems.
School:University of Cincinnati
School Location:USA - Ohio
Source Type:Master's Thesis
Keywords:molecular dynamics md simulation gramicidin ion channel poisson nernst planck equation einstein relation
Date of Publication:01/01/2003