Lipid-dependence of target membrane stability during influenza viral fusion

Abstract

Although influenza kills about a half million people each year, even after excluding pandemics, there is only one set of antiviral drugs: neuraminidase inhibitors. By using a new approach utilizing giant unilamellar vesicles and infectious X-31 influenza virus, and testing for the newly identified pore intermediate of membrane fusion, we observed ∼30-87% poration, depending upon lipid composition. Testing the hypothesis that spontaneous curvature (SC) of the lipid monolayer controls membrane poration, our Poisson model and Boltzmann energetic considerations suggest a transition from a leaky to a non-leaky fusion pathway depending on the SC of the target membrane. When the target membrane SC is below approximately -0.20 nm^-1 fusion between influenza virus and target membrane is predominantly non-leaky while above that fusion is predominantly leaky, suggesting that influenza hemagglutinin (HA)-catalyzed topological conversion of target membranes during fusion is associated with a loss of membrane integrity.

Keywords: Fusion; Membrane; Poration; Spontaneous curvature; Virus.

Fig. 1.

Poration assay. (A) Scheme of the experiment. GUVs with the desired lipid compositions labeled with DiD (blue) were incubated with R-18-labeled X-31 influenza virus (red) at 37°C for 30 min in pH 7.4 buffer containing Alexa-488 (green). This was followed by a change in pH 7.4 to pH 5. After 10 min of incubation at pH 5, numbers of filled (Alexa-488 positive) and unfilled (Alexa-488 negative) vesicles were counted. See Materials and Methods for details. (B) Poration of GUVs. Influx of Alexa-488 (green) to POPC GUVs in presence of the pore-forming peptide melittin. The line profile in green indicates the intensity distribution of Alexa-488 in the absence and presence of melittin. The blue profile indicates DiD intensity. (C) Representative examples of leakage (dye influx) and lipidic-dye transfer from viral to GUV membranes. Images shown are confocal section GUV equatorial planes. Intensity profiles of DiD (blue), Alexa-488 (green) and R-18 (red) along the radial axis of GUVs (dotted line). The line profile in the green channel indicates leakage and in the red channel indicates lipidic dye transfer. (D) Average intensity of R-18 along the radial distance at pH 7.4 and pH 5. Error bars represent s.e.m. [sample size (n) is indicated in the figure]. Note the increase in intensity due to lipid transfer (hemi-fusion) between viral and GUV membranes and consequent dequenching due to dye transfer from viral to the target membrane. See Materials and Methods section for details. Scale bars: 10 µm. a.u., arbitrary units.

Fig. 2.

Dependence of poration on lipid composition and SC. (A) Dependence of leakage on the concentration of non-bilayer-forming lipids in bilayer. Fractions of filled GUVs are plotted against the POPC mol fraction of target membranes. The POPC mole fraction decreases as the mole fraction of non-bilayer-forming lipids characterized by negative SC increases (see Table 1). (B) Variation of leakage with SC. Spontaneous curvature values were obtained through the linear combination of constituent lipids (as in A) weighted by their mole fraction. See Table 1 for details of the compositions. Typically, 100 vesicles are counted in each case and error bars represent s.e.m. of at least three independent measurements. Note that individual lipid series tend to converge on the SC axis.

Fig. 3.

Poisson analysis of target membrane leakage. 〈n〉 analysis of lipid compositions showing the postulated exponential relationship (linear on a log scale) between Poisson-distributed ‘leakage machines’ and SC for (A) POPC:CHOL:POPE:TG, (B) POPC:CHOL:TG, (C) POPC:POPE:TG and (D) POPC:DOG:TG membranes. See main text and Table 1 for lipid compositions and Table S1 for analysis and fitting parameters.

Fig. 4.

Energetics analysis of the target membrane leakage. Poisson leakage model expressed as energy changes using the probabilities for leakage and no-leakage and the calculated energy changes derived from the measured leakage fraction. ΔE (phenomenological energy difference between leaky and non-leaky) was obtained by Boltzmann analysis of the probabilities of occurrence of leaky versus non-leaky vesicles. The energy crossover point occurs at an SC of approximately −0.2 nm⁻¹. Lipid composition of cumulative SC >−0.21 nm⁻¹ corresponds to the leaky ‘rupture-insertion’ regime, while <−0.21 nm⁻¹ corresponds to the non-leaky ‘hemifusion-stalk’ regime. A target membrane with a lipid composition corresponding to the rupture-insertion regime is likely to exhibit content leakage or influx.

Fig. 5.

A diagrammatical representation of the possible interaction between the viral and the target membrane. In the presence of lipids with a negative SC for the monolayer in the target membrane, stalk formation is favored. Since the perimeter of a stalk is negative, it is stabilized in the presence of lipids with negative SC (cone-shaped lipids, shown in the diagram) of the monolayer and can extend to form a hemifusion diaphragm. The hemifusion diaphragm eventually forms the fusion pore, merging the two compartments initially separated by the viral and the target membranes. By contrast, when the SC of the contacting monolayer of the target membrane is less negative or positive, stalk formation and elongation is not favored energetically, and the target membrane ruptures to form a leakage pore. A leakage pore (in the target membrane) does not connect the two compartments but causes efflux of contents from the target membrane or influx of aqueous markers (as detected in the current assay). A leakage pore may have similar topology to a fusion pore (positive curvature at the edge), which is stabilized in the presence of lipids with a positive (or less negative) SC of the monolayer (inverse cone-shaped lipids, shown in the diagram). Note that, in our current assay, fusion pore formation was not monitored, and only the probability of leakage pore formation had been determined.