Membrane Tension Inhibits Lipid Mixing by Increasing the Hemifusion Stalk Energy

Abstract

Fusion of biological membranes is fundamental in various physiological events. The fusion process involves several intermediate stages with energy barriers that are tightly dependent on the mechanical and physical properties of the system, one of which is membrane tension. As previously established, the late stages of fusion, including hemifusion diaphragm and pore expansions, are favored by membrane tension. However, a current understanding of how the energy barrier of earlier fusion stages is affected by membrane tension is lacking. Here, we apply a newly developed experimental approach combining micropipette-aspirated giant unilamellar vesicles and optically trapped membrane-coated beads, revealing that membrane tension inhibits lipid mixing. We show that lipid mixing is 6 times slower under a tension of 0.12 mN/m compared with tension-free membranes. Furthermore, using continuum elastic theory, we calculate the dependence of the hemifusion stalk formation energy on membrane tension and intermembrane distance and find the increase in the corresponding energy barrier to be 1.6 k_BT in our setting, which can explain the increase in lipid mixing time delay. Finally, we show that tension can be a significant factor in the stalk energy if the pre-fusion intermembrane distance is on the order of several nanometers, while for membranes that are tightly docked, tension has a negligible effect.

Keywords: continuum elasticity; membrane fusion; micropipette aspiration; optical tweezers; tension.

Figure 1

Experimental setup for lipid mixing measurements. (A) Illustration of the experiment. The optically trapped membrane-coated bead is brought into contact with the aspirated GUV (i), and confocal fluorescence microscopy scans are acquired continuously to monitor the fluorescence change caused by lipid mixing (ii). (B) (iii) Confocal fluorescence image of an aspirated GUV under high aspiration, as can be seen from the large aspirated “tongue” compared to image (i). (iv) Bright-field image of an aspirated GUV and an optically trapped bead. (C) Fluorescence intensity profile on the membrane-coated bead contact edge with the GUV. The time delay to lipid mixing is measured from the contact time to the time the fluorescence intensity increases on the bead.

Figure 2

Lipid intrinsic curvature affects the lipid mixing time delay. (A) Increasing the ratio between cholesterol and DOPC in the membrane reduced the time to lipid mixing. The solid line of each box plot is the mean lipid mixing time delay. The lipid mixing time delay of 0% cholesterol is 105 ± 28 s [n = 10], 30% is 69 ± 11 s [n = 10], and 40% is 54 ± 6 s [n = 7]. (B) The external addition of LPC increased the lipid mixing time delay from 69 ± 11 s [n = 10] at 0 μmol of LPC to 75 ± 9 s [n = 8] at 5 μmol and 108 ± 18 s [n = 7] at 10 μmol.

Figure 3

Membrane tension increases lipid mixing time delay. (A) Confocal fluorescence microscopy images of lipid mixing under different tensions. The same GUV was used in both measurements. The first frame is the contact between the bead and GUV; the white arrow points to the frame with the lipid mixing initiation. Higher tension (bottom images) increases the lipid mixing time delay. (B) Lipid mixing time delay as a function of tension (25 measurements, 12 GUVs in 8 independent experiments; for 10 GUVs, multiple measurements were performed). Black dots are the experimental results, and the solid blue line is a linear trend with a 903 ± 142 m·s/mN slope.

Figure 4

Stalk shape as a function of membrane tension. (A) Water gap between membranes as a function of membrane tension. (B) Monolayer area change, ΔA, as a function of tension. (A and B) Joining pressure is 1.57 Pa (full black line). The dashed line represents the error due to uncertainty in the joining pressure; the upper line corresponds to minimum pressure of 0.95 Pa, and the lower line to maximum pressure of 2.38 Pa. (C) Example of simulation result of stalk shape with a 11.5 nm water gap between the membranes. Parameters: monolayer bending rigidity of 17.5 k_BT, monolayer saddle-splay modulus of −8.75 k_BT, tilt rigidity of 40 mN/m, monolayer width of 1.5 nm, and spontaneous monolayer curvature of −0.18 nm^–1.

Figure 5

(A) Schematic illustration of the transition from pre-fusion to hemifusion state. The pre-fusion configuration is considered as two flat membranes whose distance is set by the balance between external forces pushing them together and repulsive undulation interaction. The second metastable configuration is a hemifusion state such as hemifusion diaphragm or elongated stalk, which are the minimal energy configurations at which the proximal monolayers are fused, but the distal monolayers are still separated. The hemifusion stalk represents the maximal energy along this pathway with the maximum amount of area pulled from the surrounding membranes and maximal elastic energy. F₀ is the stalk energy independent of tension, and F_T is the tension-dependent term. (B) Change in energy barrier to lipid mixing: theoretical prediction versus experimental results. The scattered black dots (9 vesicles, 21 measurements) are the , with being the ratio between lipid mixing time delay with tension γ to the lowest measured tension for each GUV. The error in the tension corresponds to the initial tension deviation from zero for the lowest tension measurement of the specific GUV. The continuous solid line is the theoretically predicted increase in the stalk formation energy due to tension. The dashed lines represent the validity limits of the theoretical prediction due to uncertainty in the external pressure. Bilayer bending rigidity is taken as 35 k_BT for the theoretical prediction.