Atomistic simulation of mixed-lipid bilayers: mixed methods for mixed membranes

Abstract

Biomembranes are characterised by diversity in the composition of the lipids they contain and the local environments these lipids occupy. The principles that govern the lateral distribution of lipids in a heterogeneous system have important implications for membrane structure and function. Modelling equilibrium lateral distributions with atomistic detail using molecular dynamics (MD) simulation is challenging due to the very long trajectories needed for lipid lateral diffusion. A mixed MD/Monte Carlo (MC) approach relying on configuration-bias MC mutation moves allows fluctuations in local composition to occur without diffusion and has been applied to several simple mixtures. Mixtures of saturated-tail lipids differing in tail length by four carbons per tail showed lateral distribution in the fluid phase indistinguishable from random mixing, but with a distinct tendency for short-tailed lipids to cluster together within the gel phase. Local environment influenced local composition strongly at edge defects, weakly in the neighbourhood of a transmembrane helical peptide under conditions of hydrophobic mismatch and to no measurable extent in a curved bilayer. The use of an isomolar semi-grand canonical ensemble also enables insight into the phase coexistence between the gel and fluid phases in a binary mixture and between the liquid-ordered and liquid-disordered systems in ternary, cholesterol-containing system.