Roles of curvature and hydrophobic interstice energy in fusion: studies of lipid perturbant effects.

We have examined the effects of small amounts (1-4 mol %) of lipids of different molecular shapes, long chain lipids, and hydrocarbon on the kinetics of PEG-mediated fusion of 1,2-dioleoyl-3-sn-phosphatidylcholine/1,2-dioleoyl-3-sn-phosphatidylethanolamine/sphingomyelin/cholesterol (DOPC/DOPE/SM/CH, 35:30:15:30) sonicated vesicles. The effects of these lipid perturbants were different for different steps in the fusion process and varied with the ratio of the cross-sectional areas of headgroup to acyl chain moieties. For lipids with a ratio <1 (negative intrinsic curvature), a decrease in this ratio led to a dramatic increase in the initial rate of vesicle contents mixing but left the initial rate of lipid mixing roughly unchanged. For lipids with ratios >1 (positive intrinsic curvature), the initial rates of both lipid and contents mixing decreased mildly with increasing ratio. In the context of the "stalk model" for fusion, lipid mixing reflects mainly formation of the initial fusion intermediate (stalk), while contents mixing reflects conversion of this intermediate either to a second intermediate or to a fusion pore. Results with positively curved lipids (ganglioside, GM1; lysophosphatidylcholine, LPCs) and negatively curved lipids (dioleoylglycerol, DOG, and 1,2-diphytanoyl-sn-glyvero-3-phosphatidylcholine, DPhPC) can be taken as supportive of the usual interpretation of the stalk model in terms of bending energy, but enhancement of fusion in the presence of long-chain phospholipids, hexadecane, as well as a mixture of GM1 plus hexadecane could not be explained by their curvature alone. We propose that the ability of a lipid perturbant to compensate for lipid packing mismatch, that is, to lower "void" energy, must be taken into account, along with intrinsic curvature, to explain the ability of lipid perturbants to promote pore formation.