Actin dynamics switches two distinct modes of endosomal fusion in yolk sac visceral endoderm cells

Abstract

Membranes undergo various patterns of deformation during vesicle fusion, but how this membrane deformation is regulated and contributes to fusion remains unknown. In this study, we developed a new method of observing the fusion of individual late endosomes and lysosomes by using mouse yolk sac visceral endoderm cells that have huge endocytic vesicles. We found that there were two distinct fusion modes that were differently regulated. In homotypic fusion, two late endosomes fused quickly, whereas in heterotypic fusion they fused to lysosomes slowly. Mathematical modeling showed that vesicle size is a critical determinant of these fusion types and that membrane fluctuation forces can overcome the vesicle size effects. We found that actin filaments were bound to late endosomes and forces derived from dynamic actin remodeling were necessary for quick fusion during homotypic fusion. Furthermore, cofilin played a role in endocytic fusion by regulating actin turnover. These data suggest that actin promotes vesicle fusion for efficient membrane trafficking in visceral endoderm cells.

Keywords: actin; cell biology; fusion; late endosome; lysosome; mouse; mouse embryo; yolk sac visceral endoderm.

Figure 1.. Endocytic vesicles in VE cells.

(A) Schematic diagram of the imaging setup. An E8.5 whole mouse embryo was observed with a laser confocal microscope. (B) Confocal images of VE cells stained with FM 1–43, LysoTracker Red, or anti-EEA1 antibody, or electroporated with EGFP-Rab7. The yellow dotted lines indicate the cell boundaries. The yellow arrows show some vesicles that were positive for LysoTracker Red in zone 2. (C) For descriptive purposes, the supranuclear portion of VE cells is divided into three zones. Zone 1, zone 2, and zone 3 contain early endosomes (EE), late endosomes (LE), and lysosomes, respectively. (D) Confocal images of VE cells labeled with Alexa 488-transferrin. Images taken 1, 10, 20, and 40 min after pulse labeling are shown. The yellow dotted lines indicate the cell boundaries. (E) Time-course analysis of the Alexa 488-transferrin-containing vesicles in VE cells. n=4. The scale bars indicate 4 μm (B) and 10 μm (D).

Figure 1—figure supplement 1.. Vesicular transport of dextran to lysosomes.

(A) Electron microscopic image showing the position of early endosomes, late endosomes (LE), and a lysosome (Ly) in VE cells. N indicates the nucleus. The scale bar indicates 2 μm. (B) Confocal microscopic images of rhodamine-dextran-labeled endocytic vesicles in the yolk sac VE cells of E8.5 mouse embryos. Images in zones 1–3, taken at 1, 10, 20, and 40 min after labeling, are shown. The scale bar indicates 10 μm.

Figure 2.. Homotypic fusion between late endosomes.

(A) Time-lapse imaging of endocytic vesicles in VE cells. After 5 min of pulse-labeling with Alexa 488-transferrin, homotypic fusion of late endosomes (arrows) was frequently observed in zone 2. In the upper panel, time 0 indicates the start of time-lapse imaging. In the lower panel, time 0 refers to the start of the fusion of the endosomes indicated by arrows. (B) By labeling of VE cells with FM1-43, the fusion process of the cell membranes in zone 2 was observed. Time 0 refers to the start of the fusion of the endosomes, indicated by arrows. (C) Histograms showing the size distribution of total late endosomes (black line, n=265) and the late endosomes that underwent homotypic fusion (gray line, n=37). No significant difference was observed between the two groups (Mann-Whitney U test). (D) Correlation between the size of the late endosomes that underwent homotypic fusion and the time required for completion of fusion (i.e. the time required from membrane fusion to formation of a single round vesicle). The sizes of the larger of the fused vesicles are plotted. Thirty-seven fusion events were measured. The scale bars indicate 5 μm (A, top) and 1.5 μm (A, bottom; B).

Figure 3.. Heterotypic fusion between late endosomes and lysosomes.

(A) Time-lapse imaging of VE cells. After 15 min of pulse-labeling with Alexa 488-transferrin, heterotypic fusion of late endosomes was observed: they shrank gradually and disappeared from the focal plane (arrows). In the upper panel, time 0 indicates the start of time-lapse imaging. In the lower panel, time 0 refers to the start of the fusion of the endosomes indicated by an arrow. (B) Time-lapse imaging of late endosomes in VE cells. Late endosomes and lysosomes were labeled with Alexa 488-transferrin (Tfn, green) and rhodamine-dextran (Dex, red), respectively. The bottom figures show optical sectioning through the dotted line in the upper figure. Time 0 indicates the start of time-lapse imaging. (C, D) Time course of the fluorescence intensity. In a late endosome undergoing fusion (ROI 1 in [B-C]), the fluorescence intensity of transferrin (green) slightly increased and then quickly decreased after membrane fusion. The fluorescence signal of dextran (red) in an underlying lysosome leaked into the fusing late endosome. The white box in (C) indicates the area shown in (B). The gray line in (D) indicates the diameter of the transferrin-positive late endosomes. The blue dotted lines in (D) indicate the moment the fusion pore was formed in ROI 1. ROI 2 is a negative control in a lysosome that did not undergo fusion. (E) Electron microscopic image showing pore formation between a late endosome (LE) and a lysosome (Ly). The inset shows a magnified picture of the boxed region. (F) Correlation between the size of late endosomes that underwent heterotypic fusion and the time required for completion of fusion (i.e. the time from the start of shrinkage to disappearance). Twenty-five fusion events were measured. (G) Histograms showing the size distribution of late endosomes at 5 min (gray line, n=256) and 15 min (red line, n=242) after labeling with Alexa 488-transferrin, as well as the late endosomes that underwent heterotypic fusion at 15 min (blue line, n=34). The average size of the vesicles was larger at 15 min than at 5 min (**p<0.01, Mann-Whitney U test). However, the size distribution of the late endosomes that underwent heterotypic fusion at 15 min did not differ from that of the total late endosomes at 15 min. The scale bars indicate 5 μm (A, top), 1.5 μm (A, bottom), 2 μm (B), 3 μm (C), and 1 μm (E).

Figure 4.. Mathematical modeling analysis of membrane deformation after connection of 2 vesicles.

(A) Definition of shape parameters. (B) Free-energy landscape for small vesicles. The gray arrow (D) indicates a typical orbit of membrane deformation from the initial condition $\left(w,a\right)=\left(\mathrm{0.0,0.3}\right)$. (C) Free-energy landscape for large vesicles. The solid gray arrow (E) indicates the typical orbit of membrane deformation, whereas the dashed gray arrow (D’) indicates the typical orbit of membrane deformation under fluctuated conditions. The inset is a blowup of the region ${\displaystyle w=0\sim 0.15}$, ${\displaystyle a=0.25\sim 0.35}$. (D, E) Time series of shape variation on the typical orbits indicated in (B) and (C), respectively. Note that the exact time scale cannot be estimated in this analysis. (F) Probability of the explosive fusions in the $F^{large}$ landscape (the orbit of the dashed gray arrow in [C]) for various vesicle sizes, osmotic pressure differences, and fluctuation energies, estimated by Monte-Carlo simulations. $\Gamma =2\pi \kappa \sqrt{A}/R_0$ with the bending modulus $\kappa =10\left[k_{B}T\right]$ and the minimum bending radius $R_0=100\left[nm\right]$. The green shaded region indicates the condition: ${\displaystyle \mathrm{\Delta }p\simeq 2.5\sim 25\left[Pa\right],}$ $V=4A\sqrt{A}/3\sqrt{\pi }$ with ${\displaystyle A=2\sim 25\left[\mu m^2\right]}$ and the fluctuation energy ${\displaystyle 1\left[k_{B}T\right]}$ (thermal fluctuation at room temperature). The green shaded region corresponds to late-endosome-sized vesicles (${\displaystyle 2.0\sim 25\left[\mu m^2\right]}$) under estimated osmotic pressure difference (${\displaystyle \mathrm{\Delta }p\simeq 2.5\sim 25\left[Pa\right]}$) and the thermal noise at room temperature. The blue-shaded region indicates the vesicles under larger fluctuations.

Figure 4—figure supplement 1.. Mathematical modeling of membrane deformation during membrane fusion.

(A) Definition of shape parameters for calculating the bending energy of a part of small vesicles. (B) Definition of shape parameters for calculating the volume of a part of large vesicles. (C) Calculated bending energy of a part of small vesicles using Euler-Lagrange equations (red points) and its fitting curve with a hyperbolic function (blue line). (D, E) Definition of shape parameters for calculating the bending energy of the constricted neck region of large vesicles. (D) is the entire structure and (E) is the close-up of the neck region.

Figure 5.. Actin filaments are associated with late endosomes in VE cells.

(A–C) Confocal images of E8.5 mouse embryos stained with Alexa 488-phalloidin and anti-mouse IgG. Mouse IgG was incorporated into late endosomes and used as a marker for endosomes. Punctate cytoplasmic staining of phalloidin (green) was associated with the IgG-containing endocytic vesicles (red) in zone 2 (B–B’’) and at lower frequencies with the lysosomes in zone 3 (arrows, C–C’’). (D) Optical sectioning of VE cells. The cross-sections along the x-z axis of the area indicated by the dotted line in the upper panel (x-y axis) are shown. The dotted line in the IgG panel represents the apical region of zone 2. Actin filaments, extending from the apical surface into the cytoplasm (white arrowheads), were closely associated with the endocytic vesicles (red). (E, F) Three-dimensional surface rendering of actin filaments and a late endosome. The side view (E) and basal view (F) are shown. (G) Alexa 488-phalloidin staining of mRFP-fascin-1-expressing VE cells. Actin punctates were highly colocalized with fascin-1, whereas fine actin filaments extending radially in the x-y plane from endosomes were negative for mRFP-fascin-1 (white arrowheads). The fluorescence intensity plot along the white arrow in the middle panel shows colocalization of fascin-1 and phalloidin in the large spot. (H) Deconvolution images of late endosomes and actin. Maximum intensity projection images are shown. In addition to strong phalloidin-positive spots (white arrowheads), which were observable by use of conventional confocal microscopy, actin filaments extending from the endosomal surface (orange arrowheads) were visualized in the x-y axis image. In the x-z axis image, a late endosome was surrounded by actin filaments, which extended from the apical cell surface to the basal side in the cytoplasm. The scale bars indicate 3 μm (A–D), 5 μm (G), and 1 μm (E–F), (H).

Figure 6.. Dynamic regulation of actin filaments surrounding late endosomes.

(A) Representative time-lapse images of EGFP-actin spots on a late endosome. VE cells were electroporated with EGFP-actin at E7.5, and time-lapse analysis was performed on the next day. The arrow indicates the fusion of actin spots. The arrowhead indicates the appearance of a new actin spot. Time 0 refers to an arbitrary time point in time-lapse imaging. (B) The turnover rate of actin was analyzed by use of a FRAP assay. The area (4 μm x 4 μm) covering a single late endosome at the apical region of zone 2 of VE cells expressing EGFP-actin was bleached, and the fluorescence recovery was observed by use of time-lapse microscopy. The yellow boxes indicate the laser-irradiated area, and the yellow arrowheads, actin spots recovered after photobleaching. The time laser irradiation was ended is indicated as 0. (C, D) Graphs showing the fluorescence recovery of EGFP-actin surrounding late endosomes at the apical region of zone 2 (light blue in [C], n=4) and the basal region of zone 2 (dark blue in [C], n=3) or at the cell surface of VE cells (D, n=5). Fluorescence intensity was normalized to prebleaching intensities. (E) Half-recovery time (t_1/2) of EGFP-actin fluorescence at the apical and basal regions of zone 2 and at the cell surface. Data were analyzed by use of one-way ANOVA with the Tukey multiple comparison test (*p<0.05; **p<0.01). (F) EGFP-actin localization during vesicle movement. Endocytic vesicles (red) visualized by use of rhodamine-dextran in zone 2 are shown. When a vesicle (arrow) moved in a certain direction (upward in this figure), the actin patch (arrowhead) was observed at the rear end of the movement direction. The dotted line indicates the initial position of the vesicle. Time 0 refers to an arbitrary time point in time-lapse imaging. The scale bars indicate 1 μm (A) and 3 μm (B, F).

Figure 6—figure supplement 1.. EGFP-actin electroporation and colocalization of Arp3-EGFP with actin filaments surrounding late endosomes.

(A) VE cells electroporated with EGFP-actin were incubated with 100 μg/ml rhodamine-dextran for 5 min. VE cells expressing low levels of EGFP-actin (white arrowheads) endocytosed rhodamine-dextran normally and showed normal morphology of endosomal vesicles, whereas VE cells expressing high levels of EGFP-actin (white arrows) did not endocytose normally. Thus, cells expressing low levels of EGFP-actin were selected for functional analysis. (B) Arp3-EGFP and Alexa 546-phalloidin were highly colocalized in VE cells. After electroporation of the Arp3-EGFP expression plasmid and 24 h culture, the whole embryos were stained with Alexa 546-phalloidin. In the x-y axis images, phalloidin-positive actin spots were partially positive for Arp3-EGFP. Phalloidin-positive and Arp3-EGFP-negative spots are shown by white arrowheads. The intensity plot along the border of a late endosome (indicated by a white circular arrow in the upper panel) shows partial colocalization between actin and Arp3. (C) Localization of Arp3-EGFP during vesicle movement in the x-y direction. When a vesicle (indicated by the white arrowheads) moved in a certain direction (upward in this figure), an Arp3 patch (indicated by the orange arrowheads) was observed at the rear end of the movement direction. Time 0 refers to an arbitrary time point in time-lapse imaging. The scale bars indicate 30 μm (A), 5 μm (B), and 1 μm (C).

Figure 7.. Actin dynamics regulate homotypic fusion in VE cells.

(A) Confocal microscopic images of endocytic vesicles in zone 2. E8.5 mouse embryos pretreated with 0.1 μM cytochalasin D (CD) or 0.1 μM jasplakinolide (Jasp) were labeled with Alexa 488-transferrin for 5 min. Time-lapse images were taken for 400 s from 5 min after labeling. Arrows indicate homotypic fusion. Time 0 indicates the start of time-lapse imaging. (B) Frequency of homotypic fusion calculated from the number of fusions that occurred in 400 s. Data are represented as means ± SEMs of four to nine independent experiments; individual data are shown by circular dots. Data were analyzed by use of one-way ANOVA with the Tukey multiple comparison test (**p<0.01; ***p<0.001). (C) Quantitation of types of homotypic fusion. Fusion types are classified into normal explosive fusion, slow explosive fusion (more than 60 s), and bridge fusion. Cytochalasin D treatment slowed down the fusion process in a dose-dependent manner and induced bridge fusion at 1 μM. Independent experiment numbers are shown on the top of the bars. (D) Quantitation of the time for completion of homotypic fusion. Treatment with 0.1 μM or 1 μM cytochalasin D or with 20 μM CK666, but not with 0.1 μM jasplakinolide, resulted in a significant increase in the fusion time. Data are represented as means ± SEMs of 4–30 independent experiments; individual data are shown by circular dots. Data were analyzed by use of one-way ANOVA with the Tukey multiple comparison test (**p<0.01; ***p<0.001). (E) Representative images of homotypic fusion of late endosomes in VE cells treated with 1 μM cytochalasin D. Endosomes were visualized by incubation with Alexa 488-transferrin. The arrows indicate the vesicles undergoing homotypic fusion in the bridge fusion mode. Time 0 refers to the start of fusion of the endosomes indicated by arrows. (F) Time-lapse imaging of EGFP-actin localization during homotypic fusion of late endosomes. Localization of EGFP-actin (green) and endocytic vesicles (red), visualized by use of rhodamine-dextran, was observed. When a pair of endosomes (arrows) fused, EGFP-actin was newly polymerized at the opposite positions of the docking site and its fluorescence intensity was increased (arrowheads). Time 0 refers to the start of fusion of the endosomes indicated by arrows. (G) Distribution of EGFP-actin during homotypic fusion of late endosomes. The fluorescence intensities of EGFP-actin (dark gray, n=5) and FM 1–43 (light gray, n=5) were measured during homotypic fusion. The percentages of their signals in the proximal (docking site, shown as 1), lateral (shown as 2), and distal (shown as 3) regions at the predocking, docking, rounding, and postfusion stages are shown. Data are represented as means ± SEMs (*p<0.05, unpaired t-test). (H) The time course of changes in the fluorescence intensity of EGFP-actin around a pair of late endosomes during homotypic fusion. Time 0 s indicates the moment the fusion pore of two late endosomes was formed. The graph indicates the average of five independent experiments. The scale bars indicate 10 μm (A), 2 μm (E), and 3 μm (F).

Figure 7—figure supplement 1.. Distribution of actin filaments and homotypic fusion in cytochalasin D-treated VE cells.

(A) Images of late endosomes and actin filaments at apical or basal regions in zone 2 of VE cells pretreated with 1 μM cytochalasin D for 5 min. E8.5 embryos were subsequently labeled with Alexa 594-transferrin and Alexa 488-phalloidin. Images on the right side show the maximum intensity projection (MIP) images along the x-y axis (upper) and x-z axis (bottom) of a single late endosome in the boxed area in the left side figure. After cytochalasin D treatment, filamentous structures observed on the control late endosomes (both filaments extending radially and in the apical-basal direction) decreased, and instead large clusters (white arrowheads) were observed in the cytosol, which appeared to be due to aggregation of actin filaments. (B) E8.5 mouse embryos pretreated with 0.1 μM cytochalasin D (CD) for 5 min were labeled with Alexa 488-transferrin or FM 1–43 for 5 min. The fluorescent signal was observed 5 min after labeling. Homotypic fusion of late endosomes after cytochalasin D treatment was slow when compared with that of the control nontreated cells. Time is given in s. Time 0 s indicates the moment the fusion pore of 2 late endosomes was formed. The scale bars indicate 2 μm (A, left), 0.6 μm (A, right), and 3 μm (B).

Figure 7—figure supplement 2.. Distribution of actin filaments in jasplakinolide-treated VE cells.

Images of late endosomes and actin of VE cells pretreated with 1 μM jasplakinolide for 5 min. E8.5 embryos were subsequently labeled with Alexa 594-transferrin and Alexa 488-phalloidin. Images on the right side show the maximum intensity projection (MIP) images along the x-y axis (upper) and x-z axis (bottom) of a single late endosome in the boxed area in the figure on the left. After jasplakinolide treatment, the fluorescence intensity of the phalloidin-positive mesh in zone 1 was significantly increased (see the top portion of the x-z-axis image). In addition, at the apical region of zone 2, actin filaments extending radially from the endosomal surface (orange arrowheads) were thicker when compared with those of the control cells (Figure 5H), and the fluorescence intensity of the filaments was increased. In contrast, actin filaments extending along the apical-basal axis were not changed (shown by white arrowheads). The scale bar indicates 2 μm (left side) and 0.5 μm (right side).

Figure 7—figure supplement 3.. Distribution of actin during homotypic fusion in cytochalasin D-treated VE cells.

Time-lapse imaging of EGFP-actin during homotypic fusion of late endosomes in E8.5 mouse embryos pretreated with 1 μM cytochalasin D for 5 min. Localization of EGFP-actin (green) and endocytic vesicles (red), visualized by use of Alexa 594-transferrin, was observed. EGFP-actin spots appeared to be rigid and static. When a pair of endosomes (arrows) fused, drastic changes in the distribution and fluorescence intensity of EGFP-actin signals as had been observed in the control VE cells were not seen. Time 0 refers to the start of fusion of the endosomes indicated by arrows. The scale bar indicates 3 μm.

Figure 7—figure supplement 4.. Distribution of Arp3-EGFP during fusion.

Time-lapse imaging of Arp3-EGFP localization during homotypic fusion of late endosomes. After expression of Arp3-EGFP in VE cells, localization of EGFP was observed. When a pair of endosomes (white arrowheads) fused, Arp3 was highly polymerized at the opposite position of the docking site and its fluorescence intensity was increased (orange arrowheads). Time 0 indicates the moment when the fusion pore of 2 late endosomes was formed. Time is given in s. The scale bars indicate 1 μm.

Figure 8.. Actin stabilization induces heterotypic fusion of late endosomes with lysosomes.

(A) Confocal microscopic images of endocytic vesicles in zone 2. E8.5 mouse embryos pretreated with 0.1 μM cytochalasin D (CD) or with 0.01 μM jasplakinolide (Jasp) were labeled with Alexa 488-transferrin for 5 min. Time-lapse images were taken for 400 s from 15 min after labeling. The arrows indicate heterotypic fusion with lysosomes. Time 0 indicates the start of time-lapse imaging. (B) Frequency of heterotypic fusion calculated from the number of fusions that occurred in 400 s from 5 min after labeling. Data are represented as means ± SEMs of four to seven independent experiments; individual data are shown by circular dots. Data were analyzed by use of one-way ANOVA with the Tukey multiple comparison test (**p<0.01, ***p<0.001). (C) Quantitation of heterotypic fusion frequency at 15 min after labeling of VE cells pretreated with cytochalasin D (CD), jasplakinolide (Jasp), or CK666. Data are represented as means ± SEMs of four to eight independent experiments; individual data are shown by circular dots. Data were analyzed by use of one-way ANOVA with the Tukey multiple comparison test (*p<0.05, **p<0.01). (D) Effects of myosins on heterotypic fusion. Frequency of heterotypic fusion was calculated at 15 min after labeling. Data are represented as means ± SEMs of five to eight independent experiments; individual data are shown by circular dots. Data were analyzed by use of one-way ANOVA with the Tukey multiple comparison test (*p<0.05, ***p<0.001). (E) Colocalization of EGFP-myosin IIA or EGFP-myosin Va with phalloidin in zone 2 of VE cells. The insets show the magnified images of the boxed area. Following electroporation of the EGFP-fused myosin expression plasmids and 24 hr culture, whole embryos were stained with Alexa 546-phalloidin. EGFP-myosin IIA signals overlapped with phalloidin-positive spots on late endosomes, whereas most of the EGFP-myosin Va signals were juxtaposed with phalloidin-positive spots on late endosomes (see the schemas and fluorescence intensity plots). The scale bars indicate 10 μm (A), 3 μm (E), and 1 μm (E, insert).

Figure 8—figure supplement 1.. Effects of myosin inhibitors on homotypic fusion in VE cells.

E8.5 mouse embryos pretreated with 10 μM pentachloropseudilin (PClP), 10 μM blebbistatin, 10 μM MyoVin-1, or 10 μM 2,4,6-triiodophenol (TIP) for 5 min were labeled with Alexa 488-transferrin for 5 min. Frequency of homotypic fusion was calculated at 5 min after labeling. Data are represented as means ± SEMs of four to nine independent experiments; individual data are shown by circular dots. Data were analyzed by use of one-way ANOVA with the Tukey multiple comparison test (***p<0.001).

Figure 9.. Transition from homotypic to heterotypic fusion.

(A) Time course of frequencies of homotypic (blue line) and heterotypic (red line) fusion. The graph indicates the means of 6 independent experiments. (B) Distribution of cofilin in zone 2 of VE cells. After whole embryos were electroporated with YFP-cofilin and cultured for 1 d, they were stained with Alexa 546-phalloidin. The lower graph plotting the fluorescence intensity along the white arrow in the upper image indicates colocalization of actin filaments and cofilin. (C) Optical sectioning images (right) of VE cells along the dotted line in the x-y axial panel (left). Embryos, which were electroporated with YFP-cofilin and cultured for 1 d, were stained with Alexa 546-phalloidin. The actin filaments, extending from the apical to the basal region, were highly colocalized with cofilin. Note that the white arrows and blue arrowheads indicate an actin filament that surrounds late endosomes and the cell membrane, respectively. (D) Quantification of the fluorescence intensity of phalloidin and cofilin along the apical-basal axis. The intensity of spots colocalized with phalloidin and cofilin around late endosomes observed in the x-y images was quantified along the z-axis. Fluorescence intensity was normalized to the intensities in the most apical plane (8 late endosomes from 3 VE cells). Data were analyzed by use of one-way ANOVA with the Tukey multiple comparison test and represented as means ± SEMs (**p<0.01, ***p<0.001). (E) FRAP assay to evaluate the dynamics of actin at the apical region of zone 2 in VE cell pretreated with the S3 peptide. After treatment with the S3 or RV peptide overnight, the area (4 μm x 4 μm) covering a single late endosome in VE cells expressing EGFP-actin was bleached and the fluorescence recovery was observed by time-lapse microscopy. Three independent experiments were performed. (F, G) Frequency of homotypic (F) and heterotypic fusion (G) in embryos pretreated with 15 μg/ml S3 or RV peptides. Frequencies of homotypic and heterotypic fusions were calculated according to the numbers of fusions in 400 s from 5 min (F) and 15 min (G) after labeling with Alexa 488-transferrin. Data are represented as means ± SEMs of three to four independent experiments; individual data are shown by circular dots. Data were analyzed by use of one-way ANOVA with the Tukey multiple comparison test (*p<0.05). The scale bars indicate 8 μm (B) and 2 μm (C). n.s. indicates no significant difference.

Figure 9—figure supplement 1.. Distribution of actin filaments in S3 peptide-treated VE cells.

Images of late endosomes and actin of VE cells pretreated with the S3 peptide (15 μg/ml) for 24 h and subsequently labeled with Alexa 594-transferrin and Alexa 488-phalloidin. Images on the right side show the maximum intensity projection (MIP) images along the x-y axis (upper) and x-z axis (bottom) of a single late endosome in the boxed area in the figure on the left. Actin filaments extending radially from the endosomal surface (orange arrowheads) were shorter and the fluorescence intensity was decreased by S3 peptide treatment, whereas actin filaments extending from the apical cell surface to the basal side on the surface of late endosomes were not changed (white arrowheads). The scale bars indicate 2 μm (left) and 0.6 μm (right).

Figure 10.. Summary of late endosome fusion in VE cells.

In VE cells, late endosomes (LE) show two different types of fusion with different targets. In homotypic fusion, once a membrane pore is formed between two late endosomes, the pore quickly expands and a single fused vesicle is formed. In contrast, in heterotypic fusion, the pore does not expand; instead, the late endosome gradually shrinks and disappears over time as the result of a transfer of its vesicle content into lysosomes. Two different types of actin are associated with the late endosomal membrane in VE cells: the actin filaments that extend along the apical-to-basal axis and the filaments that are radially polymerized from the membrane. Cofilin, an actin-binding protein that severs and depolymerizes actin filaments, activates actin turnover. In the apical region, late endosomes receive squeezing forces via active actin modification by cofilin and Arp2/3, which leads to homotypic fusion in an explosive mode. In contrast, in the basal region, actin is more static and late endosomes receive sliding forces via myosins.