Structural Determinants of the Energetic Barrier of Phospholipid Flip-Flop
Spontaneous phospholipid flip-flop sets the background against which flippases, floppases, and scramblases create and dissipate membrane asymmetry. How the associated free-energy barrier depends on lipid chemistry and on the surrounding membrane remains unclear, because experiments are few and contradictory and previous simulations covered only a small part of the lipidome. Here, we calculate Martini 3 flip-flop free-energy barriers for over 700 membrane--lipid combinations, the largest such dataset to our knowledge. A systematic single component bilayer panel of five headgroups, three linkages, and 35 tail families is complemented by mixed bilayers, from binary mixtures to organelle-inspired compositions, that place the same lipid in different membranes and thus separate the effects of lipid chemistry and the membrane environment. In single component bilayers, the barrier rises with tighter tail packing and depends strongly on headgroup and tail unsaturation but only weakly on linkage. Moving a lipid into a mixed bilayer shifts its barrier as much as most headgroup substitutions do, and the shift follows the change in local tail order. Within a shared membrane, headgroup differences are mostly smaller than between single component bilayers, so part of a headgroup effect arises from the membrane that the lipid forms. A Bayesian model combining chemical descriptors with local membrane structure predicts barriers for withheld lipid families and membranes with cross-validated errors of 2.8 and 3.8 kJ/mol. The calculated trends agree qualitatively with native-lipid vesicle experiments, although Martini 3 underestimates the barrier differences between lipids, gives lower barriers than atomistic models, and overstates the effect of cholesterol.