Biological membranes are inherently asymmetric, consisting of various lipids and proteins that are heterogeneously distributed between membrane leaflets. The study of spatial heterogeneity in membrane bilayers is of fundamental importance in membrane biophysics. However, the accurate simulation of realistic membranes remains challenging. In all-atom (AA) modeling, the slow diffusion of lipids renders multicomponent bilayer simulations computationally demanding. In coarse-grained (CG) modeling, top-down models have been largely employed for the study of membranes; however, their implementation is not ideal due to the inaccuracies in modeling lipid–lipid and lipid–protein interactions from the point of view of statistical mechanics. In this study, we have constructed a "bottom-up" CG model of an asymmetric bilayer, in this case chosen to mimic the HIV-1 virion membrane, by following a systematic statistical mechanical route. The resulting CG model is also found to be transferable for simulating various membrane compositions, effectively capturing the cholesterol condensation effect in which higher cholesterol concentrations induce lipid tail ordering. Using this bottom-up CG model, we demonstrate that in an asymmetric bilayer, cholesterol rapidly moves from a compressed leaflet to an expanded leaflet to reduce membrane stress. The free energy landscape for interleaflet cholesterol movement was calculated in different membrane compositions. In a symmetric bilayer, the cholesterol is found to be equally stable in both leaflets. However, in an asymmetric bilayer, the stability of cholesterol depends on the overall lipid composition of the different leaflets. Overall, this study opens up a new paradigm for the systematic, bottom-up CG modeling of realistic membranes and offers insight into the nature of lipid interactions in an asymmetric bilayer.
Majumder et al. (Sat,) studied this question.