Early Endosomes Act as Local Exocytosis Hubs to Repair Endothelial Membrane Damage

Abstract

The plasma membrane of a cell is subject to stresses causing ruptures that must be repaired immediately to preserve membrane integrity and ensure cell survival. Yet, the spatio-temporal membrane dynamics at the wound site and the source of the membrane required for wound repair are poorly understood. Here, it is shown that early endosomes, previously only known to function in the uptake of extracellular material and its endocytic transport, are involved in plasma membrane repair in human endothelial cells. Using live-cell imaging and correlative light and electron microscopy, it is demonstrated that membrane injury triggers a previously unknown exocytosis of early endosomes that is induced by Ca²⁺ entering through the wound. This exocytosis is restricted to the vicinity of the wound site and mediated by the endosomal soluble N-ethylmaleimide sensitive factor attachment protein receptor (SNARE) VAMP2, which is crucial for efficient membrane repair. Thus, the newly identified Ca²⁺ -evoked and localized exocytosis of early endosomes supplies the membrane material required for rapid resealing of a damaged plasma membrane, thereby providing the first line of defense against damage in mechanically challenged endothelial cells.

Keywords: VAMP2; calcium; endosomal fusion; membrane wounds; plasma membrane resealing.

Figure 1

Membrane injury leads to the disappearance of early and late endosomal vesicles close to the wound site in HUVEC. A) HUVEC transfected with GFP‐2xFYVE or LAMP1‐mGFP (green) and stained with the membrane‐impermeable dye FM4‐64 (magenta) were subjected to laser injury while being imaged by time‐lapse microscopy. Representative still images with time stamps in seconds (s) shown here. Red triangle, injury site. Boxed areas are magnified below; white arrows indicate endosomes before disappearance. Yellow arrows indicate the endosomal disappearance sites from the previous time frame. Wounded cells are outlined in white dashed lines. Scale bars, 10 µm; for zoom, 5 µm. B) Scheme illustrating the ROI‐analysis used to quantify disappearing endosomes with increasing distance from the wound site (shown as a red circle on the PM and the increment of each circle ROI from the previous circle = 20 µm in diameter). The endosomal punctae in each concentric circular ROI area were measured over time. C–I) Quantification of the disappearance of endosomal markers over time, following laser injury in HUVEC transfected with GFP‐2xFYVE (shown in purple), GFP‐Rab5 (green), LAMP1‐mGFP (pink) or pulsed with transferrin‐AF488 (orange), with t = 0 s representing the time of wounding. As in (B), endosomal disappearance is represented in circle 1 (C), circle 2 (D), circle 3 (E), circle 4 (F), circle 5 (G), circle P, the peripheral edge of the cell (H), and in the whole cell after wounding (I). Each data point shows the number of endosomes normalized to the initial number before wounding, expressed as a percentage, and mean ± SD shown here. n = 18–21 cells, pooled from 3 independent experiments. The tests used for statistical comparisons across the various endosomal markers for each circle ROI were: one‐way ANOVA with Friedman test for (C–E) and (H and I), and one‐way ANOVA with Tukey's test for (F and G). ns, not significant; **P < 0.01; ***P < 0.001.

Figure 2

LEL exocytosis does not contribute to HUVEC membrane repair. A,B) pHluorin‐LAMP1‐mApple expressing HUVEC were laser wounded in the presence of Ca²⁺ (+/− U18666A treatment) or EGTA. Still images of LAMP1‐pHluorin (green) are shown here for cells injured in the presence of Ca²⁺. White arrows highlight LEL (LAMP1) exocytosis events. LAMP1‐exocytosis events normalized to the initial count before wounding, plotted over time. C) Ionomycin‐induced LAMP1‐exocytosis events (see Figure S8A, Supporting Information) were quantified over time in the absence or presence of U18666A and normalized to the initial punctae count before ionomycin addition. D,E) Still images of laser injury of HUVEC treated with U18666A in the presence of FM4‐64 (magenta), and kinetics of membrane resealing quantified over time. F,G) Laser injury of HUVEC treated with DPA in the presence of FM4‐64 with resealing dynamics quantified in (G). Control in (D and F) refers to wounding carried out without U18666A or DPA in the presence of Ca²⁺. H,I) Resealing efficiency of cells treated with U18666A (purple) or DPA (orange) after mechanical scrape injury (H) or glass bead injury (I), represented as percentages, from 3 biological replicates. Red triangles, wound ROI. Mean ± SEM plotted for (C) and mean ± SD plotted for all other graphs. n = 22 – 25 cells (B and C), 24 – 26 cells (E), 24 – 28 cells (G), pooled from 3 independent experiments. P value is calculated using one‐way ANOVA with Kruskal–Wallis test for (B) and (C), one‐way ANOVA with Holm–Sidak's test (E), ordinary one‐way ANOVA with Tukey's test (G and H), and unpaired two‐tailed Student's t‐test (I). ns, not significant; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Figure 3

EE preferentially undergo exocytosis near the wound site. A) HUVEC transfected with GFP‐2xFYVE (green) and TfR‐pHuji (red) were laser‐wounded. Blue dashed boxes of each time point are magnified below; white arrows indicate 2xFYVE endosomes before disappearance; yellow arrows indicate TfR exocytosis events at the sites of 2xFYVE disappearance. B) The increase in TfR exocytosis events (clusters with increased pHuji signal) was quantified over time, represented in zones of increasing distance from the wound site (see Figure S11A, Supporting Information), and normalized to the baseline count in each cell per µm² of ROI area. (For absolute numbers of exocytosis events, see Figure S19A, Supporting Information). C) EE exocytosis events positive for concomitant 2xFYVE disappearance and TfR exocytotic cluster formation (see A), quantified as in (B) over time with each ROI area and initial count normalized. D) Detection of TfR exocytosis events around the wound site (orange box magnified on the right) by confocal and STED imaging, following the anti‐GFP nanobody exocytosis assay. Image representative of 4 experiments. E) HUVEC were labeled with BSA‐gold, washed, laser wounded in the presence of FM4‐64, fixed, and processed for TEM. CLEM overlay image of the relocated wounded cell (0‐250 nm section). Conventional TEM image of the consecutive 250 nm section (from 250–500 nm height) shown below with the wound site marked by a blue box and a nearby site marked by an orange box. Magnifications are shown on the top right for the wound site and the bottom right for the nearby site. Close‐up image of a BSA‐gold labeled endosome is shown on the top right of the marked TEM image with a purple box for the wound site and a green box for the near wound site. Shown next to each TEM zoom‐in is a representative tomographic slice extracted from each double‐tilt tomography series followed by the contoured slice and the 3D model of the entire tomogram (see Videos S6 and S7, Supporting Information). In the contoured slice and the model, PM (magenta) and unambiguously identified endosomes, which are vesicular (dark blue) and tubular (light blue), are traced. Yellow arrows indicate BSA‐gold labeled membranes. Representative CLEM image from n = 12 cells over 3 experiments. Red triangles, wound sites. Scale bars, 10 µm (A, D, E); 5 µm for magnification of (A); 1 µm in magnified TEM views, 500 nm in tomographic slices, and 100 nm in BSA‐gold endosome zoom‐ins of (E)]. n = 22 – 29 cells for (B) and (C), from 3 independent experiments with mean ± SD plotted here. Statistical comparisons were performed with ordinary one‐way ANOVA with Tukey's test for (B) and one‐way ANOVA with Kruskal–Wallis test for (C). **P < 0.01; ***P < 0.001.

Figure 4

Exocytosis of early endosomes occurs solely in the presence of high levels of Ca²⁺ and is dependent on SNARE‐mediated fusions. A) TfR‐pHuji expressing HUVEC (displayed in cyan) were laser wounded in the presence of EGTA and the wounding response was recorded using FM4‐64 (magenta). Representative images before and after wounding are shown. No change in the localization of TfR‐pHuji fluorescence was observed. B) HUVEC expressing GFP‐2xFYVE (green) and the Ca²⁺ indicator, R‐GECO 1.2 (red) were laser wounded. Images show Ca²⁺ entry and white arrows indicate disappearing 2xFYVE endosomes in the corresponding frames. C) Overlap of GFP‐2xFYVE disappearance (solid lines) with the GECO Ca²⁺ wave intensity increase (dashed lines) quantified for circles 1 and 2 with mean ± SD plotted. D) GFP‐2xFYVE expressing HUVEC was recorded live and 100 µm histamine was then added to induce intracellular Ca²⁺ elevation. Time‐lapse images show no change in the overall number of GFP‐2xFYVE positive endosomes and a few such examples are marked by white arrows. E) HUVEC were treated with the SNARE inhibitor, NEM (0.1 mm, 5 min), or vehicle control and subjected to laser ablation in the presence of FM4‐64 (magenta). Images pre and post wounding are shown. F,G) Colocalization of the various vesicle associated membrane protein (VAMP) proteins with respect to the EE markers, EEA1 (F), and transferrin‐AF647 labeled for 5 min (G), was calculated using Pearson's correlation coefficient, r, and represented as distribution plots with mean ± SD. Red triangles, wound sites, and white dashes indicate the laser‐ablated (A, B, E) and histamine‐stimulated (D) cells. Scale bars, 10 µm and images representative of 4 independent experiments (A, D, E). n = 22 – 29 cells for (C) and 25–30 cells for (F and G), from 3 independent experiments. Statistical comparisons were performed using two‐tailed Mann‐Whitney U test (C), and one‐way ANOVA with Kruskal–Wallis test (F and G), with ****P < 0.0001.

Figure 5

VAMP2 participates in wound‐induced EE exocytosis. A) HUVEC cotransfected with VAMP2‐SEP (green) and TfR‐pHuji (red) were laser wounded and sequential images are shown. Blue dashed boxes are magnified below; white arrows indicate VAMP2 exocytotic events at the sites of TfR exocytosis. B) Quantification of exocytosis events positive for VAMP2‐SEP fluorescence increase over time, in regions of increasing distance from the wound site (Figure S11A, Supporting Information), normalized to the circle ROI area and punctae count before wounding. (For absolute numbers of VAMP2 exocytosis events, see Figure S19B, Supporting Information). C) EE exocytosis events positive for colocalized VAMP2‐SEP and TfR‐pHuji signals of increased intensity, quantified as in (B), and normalized to the area of each ROI and baseline colocalized punctae count. D) Similar analysis as in (C) for cells subjected to low laser power ablation (non‐wounded control). E) Analysis of exocytotic events showing VAMP2‐SEP fluorescence increase that colocalized with mApple‐2xFYVE disappearance, plotted over time. F) Detection of VAMP2‐SEP exocytosis events around the wound site (magnified in blue) and far away from the wound site (orange box) by the anti‐GFP nanobody exocytosis assay. Image representative of 3 independent experiments. Red triangle, wound site. Scale bars, 10 µm; 5 µm in zooms (for A and F). Mean ± SD plotted for all graphs with n = 21 cells (C), n = 23 (D), and n = 19 (E), from 3 independent experiments, and for (B), n = 39 cells from 6 experiments. Statistical comparisons were performed using one‐way ANOVA with Kruskal–Wallis test for (B), (C), and (E), and for (D), one‐way ANOVA with Tukey's test was used. ns, not significant; *P < 0.05; **P < 0.01.

Figure 6

EE exocytosis regulated by VAMP2 is critical for HUVEC membrane resealing. A) Immunoblot showing levels of VAMP2 after siRNA treatment with siControl (siC) or siVAMP2 pool (top panel) and represented as a percentage to the loading control tubulin (bottom). Blot representative of 3 independent experiments. B,C) siC or pooled siVAMP2 transfected HUVEC in the presence of FM4‐64 (magenta) were laser‐wounded (B) and the resealing kinetics quantified (C). D) HUVEC were transfected with EmGFP miRNA control (miC) or miRNA VAMP2 and assessed for VAMP2 knockdown by western blot of cell lysates (VAMP2 levels are normalized to loading control, calnexin) and flow cytometry analysis of VAMP2 staining in EmGFP‐transfected cells (bottom panel). E,F) EmGFP‐miRNA control or miRNA VAMP2 transfected cells were laser wounded and resealing analyzed by FM4‐64 dye influx (E) and quantified (F). G,H) siC or siVAMP2 pool transfected cells were analyzed for membrane resealing after mechanical scrape injury (G) or glass bead injury (H). I) VAMP2 depletion interferes with wounding‐induced exocytosis of TfR‐positive EE. HUVEC were transfected with EmGFP miC or miR VAMP2 together with TfR‐pHuji (red) and laser wounded in the presence of FM4‐64 dye. The TfR channel is highlighted below in blue boxes, and white arrows indicate EE exocytosis events positive for TfR‐pHuji fluorescence. J) The amount of TfR exocytosis events in HUVEC transfected with EmGFP miC or miR VAMP2 was quantified over time after wounding, in regions of increasing distance closest to the wound site (Figure S11A, Supporting Information, circles 1–3), and normalized to the baseline count in each cell and circular ROI area. Lines with filled circles represent miC data and dashes with triangles indicate miR VAMP2 data. K) Model of the wounding‐induced exocytosis of EE for wound repair. Red triangles, wound sites. Mean ± SD plotted for all graphs. n = 24 – 26 cells in (C) and (F), and 31 – 44 cells for (J), pooled from 3 independent experiments. Tests used for comparisons were: unpaired two‐tailed Student's t‐test for (A), unpaired two‐tailed Welch's t‐test (D) and (H), one‐way ANOVA with Kruskal–Wallis test for (C) and (F), one‐way ANOVA with Tukey's test (G), and one‐way ANOVA with Friedman test (J). **P < 0.01; ***P < 0.001; ****P < 0.0001.