A COMPUTER-SIMULATION OF FREE-VOLUME DISTRIBUTIONS AND RELATED STRUCTURAL-PROPERTIES IN A MODEL LIPID BILAYER

A novel combined approach of molecular dynamics (MD) and Monte Carlo simulations is developed to calculate various free-volume distributions as a function of position in a lipid bilayer membrane at 323 K. The model bilayer consists of 2 x 100 chain molecules with each chain molecule having 15 carbon segments and one head group and subject to forces restricting bond stretching, bending, and torsional motions. At a surface density of 30 angstrom2/chain molecule, the probability density of finding effective free volume available to spherical permeants displays a distribution with two exponential components. Both pre-exponential factors, p1 and p2, remain roughly constant in the highly ordered chain region with average values of 0.012 and 0.00039 angstrom-3, respectively, and increase to 0.049 and 0.0067 angstrom-3 at the mid-plane. The first characteristic cavity size V1 is only weakly dependent on position in the bilayer interior with an average value of 3.4 angstrom3, while the second characteristic cavity size V2 varies more dramatically from a plateau value of 12.9 angstrom3 in the highly ordered chain region to 9.0 angstrom3 in the center of the bilayer. The mean cavity shape is described in terms of a probability distribution for the angle at which the test permeant is in contact with one of and does not overlap with anyone of the chain segments in the bilayer. The results show that (a) free volume is elongated in the highly ordered chain region with its long axis normal to the bilayer interface approaching spherical symmetry in the center of the bilayer and (b) small free volume is more elongated than large free volume. The order and conformational structures relevant to the free-volume distributions are also examined. It is found that both overall and internal motions have comparable contributions to local disorder and couple strongly with each other, and the occurrence of kink defects has higher probability than predicted from an independent-transition model.