Regulation of membrane scission in yeast endocytosis

Abstract

During clathrin-mediated endocytosis, a flat plasma membrane is shaped into an invagination that undergoes scission to form a vesicle. In mammalian cells, the force that drives the transition from invagination to vesicle is primarily provided by the GTPase dynamin that acts in concert with crescent-shaped BAR domain proteins. In yeast cells, the mechanism of endocytic scission is unclear. The yeast BAR domain protein complex Rvs161/167 (Rvs) nevertheless plays an important role in this process: deletion of Rvs dramatically reduces scission efficiency. A mechanistic understanding of the influence of Rvs on scission, however, remains incomplete. We used quantitative live-cell imaging and genetic manipulation to understand the recruitment and function of Rvs and other late-stage proteins at yeast endocytic sites. We found that arrival of Rvs at endocytic sites is timed by interaction of its BAR domain with specific membrane curvature. A second domain of Rvs167-the SH3 domain-affects localization efficiency of Rvs. We show that Myo3, one of the two type-I myosins in Saccharomyces cerevisiae, has a role in recruiting Rvs167 via the SH3 domain. Removal of the SH3 domain also affects assembly and disassembly of actin and impedes membrane invagination. Our results indicate that both BAR and SH3 domains are important for the role of Rvs as a regulator of scission. We tested other proteins implicated in vesicle formation in S. cerevisiae and found that neither synaptojanins nor dynamin contribute directly to membrane scission. We propose that recruitment of Rvs BAR domains delays scission and allows invaginations to grow by stabilizing them. We also propose that vesicle formation is dependent on the force exerted by the actin network.

FIGURE 1:

rvs167Δ but not vps1Δ changes coat movement. (A) Left: One thousand millisecond exposures from time-lapse movies of WT, vps1Δ, and rvs167Δ cells with endogenously tagged Sla1-eGFP. Right: Kymographs of Sla1-eGFP or Rvs167-eGFP in WT, vps1Δ, and rvs167Δ cells. (B) Scission efficiency in WT, vps1Δ, and rvs167Δ cells. Error bars are SD, p values from two-sided t test, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. (C) Average centroid positions of Sla1-eGFP in WT and vps1Δ cells. (D) Number of Rvs167 molecules in WT and vps1Δ cells. (E) Average centroid positions of Sla1-eGFP in WT and successful and retracted Sla1-eGFP positions in rvs167Δ cells. (F) Average centroid positions of Rvs167-eGFP in WT and vps1Δ cells. All centroids were coaligned with Abp1-mCherry so that time = 0 s corresponds to Abp1 intensity maximum and y = 0 nm corresponds to nonmotile positions of Sla1 in the respective strains. Shading on plots shows 95% confidence intervals. Dashed red lines indicate Abp1 intensity maxima in respective strains.

FIGURE 2:

Synaptojanin-like proteins do not significantly influence endocytosis. (A) One thousand millisecond exposures from time-lapse movies of cells with endogenously tagged Inp51-, Inp52-, and Inp53-eGFP. (B) Inp52 centroid trajectory aligned in space and time to other endocytic proteins. (C) Scission efficiency in WT, rvs167Δ, inp51Δ, and inp52Δ cells. Error bars are SD, with p values from two-sided t test, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. (D) Centroid positions of Sla1-eGFP in WT, inp51Δ, and inp52Δ cells. (E) Centroid positions of Rvs167-eGFP in WT, inp51Δ, and inp52Δ cells. All centroids were coaligned with Abp1-mCherry so that time = 0 s corresponds to Abp1 intensity maximum and y = 0 nm corresponds to position of nonmotile Sla1 in the corresponding strains. Shading on plots represents 95% confidence interval.

FIGURE 3:

Rvs167 BAR domains need membrane curvature to localize to endocytic sites. (A) Schematic of Rvs protein complex with and without the SH3 domain. (B) One thousand millisecond exposures from time-lapse movies, showing localization of full-length Rvs167 and Rvs167-sh3Δ in WT, sla2Δ, LatA treated, and LatA-treated sla2Δ cells. Bars indicate percentage of cells with Rvs167 cortical patches and percentage of cells that contain both Rvs167 and Abp1.

FIGURE 4:

Myo3 likely recruits Rvs167 in the absence of membrane curvature. Two hundred millisecond exposures from time-lapse movies showing Rvs167-eGFP localization in untreated and LatA-treated WT, vrp1Δ, syp1Δ, bbc1Δ, myo5Δ, and myo3Δ cells. Percentages are averaged number from two experiments. They indicate number of LatA-treated cells in which Rvs167-eGFP is localized at the plasma membrane.

FIGURE 5:

Endocytic dynamics is changed in rvs167-sh3Δ cells. (A, B) Centroid positions of Sla1 and Rvs167 in WT and rvs167-sh3Δ cells. (C, D) Numbers of molecules in WT and rvs167-sh3Δ cells. Centroids were aligned so that time = 0 s corresponds to Abp1 intensity maximum in the respective strains and y = 0 nm corresponds to nonmotile Sla1 position. Shading represents 95% confidence interval.

FIGURE 6:

Increased BAR domain expression increases Rvs recruitment, Sla1 movement, and Abp1 recruitment. (A, B) Molecule numbers and centroid positions of Rvs167. Centroid positions were coaligned with Abp1-mCherry so that time = 0 s corresponds to Abp1 intensity maximum. (B) y = 0 nm is based on Sla1 positions for the respective strains. (C) Sla1 centroid positions, aligned so that the centroids begin inward movement at the same time. y = 0 nm corresponds to their nonmotile centroid position. Dashed lines correspond to the Sla1 centroid positions when intensity of Abp1 in the corresponding strain is at maximum. (D) Abp1 molecule number, aligned so that time = 0 s corresponds to Abp1 intensity maximum. Shading represents 95% confidence interval.

FIGURE 7:

Model for yeast endocytic scission. Membrane at an endocytic site is bent by forces derived from actin polymerization. BAR domains arrive at a tubular invagination and scaffold the membrane, delaying scission. Actin forces eventually overcome the influence of BAR scaffolding, and the membrane breaks, resulting in vesicle formation.