Rab5-regulated endocytosis plays a crucial role in apical extrusion of transformed cells

Significance

At the initial stage of carcinogenesis, transformation occurs in a single cell within the epithelium. However, it is not clearly understood what happens at the interface between the newly emerging transformed cells and the surrounding normal epithelial cells. Here, using mammalian cell culture and zebrafish embryo systems, we demonstrate that Rab5, an important regulator of endocytosis, is accumulated and that endocytosis is enhanced in RasV12-transformed cells surrounded by normal cells. The elevation of endocytosis disrupts E-cadherin–based cell–cell adhesions with the surrounding normal cells and modulates signaling pathways, eventually leading to apical elimination of the transformed cells. This report demonstrates that endocytosis plays a crucial role in cell competition between normal and transformed epithelial cells in mammals.

Abstract

Newly emerging transformed cells are often eliminated from epithelial tissues. Recent studies have revealed that this cancer-preventive process involves the interaction with the surrounding normal epithelial cells; however, the molecular mechanisms underlying this phenomenon remain largely unknown. In this study, using mammalian cell culture and zebrafish embryo systems, we have elucidated the functional involvement of endocytosis in the elimination of RasV12-transformed cells. First, we show that Rab5, a crucial regulator of endocytosis, is accumulated in RasV12-transformed cells that are surrounded by normal epithelial cells, which is accompanied by up-regulation of clathrin-dependent endocytosis. Addition of chlorpromazine or coexpression of a dominant-negative mutant of Rab5 suppresses apical extrusion of RasV12 cells from the epithelium. We also show in zebrafish embryos that Rab5 plays an important role in the elimination of transformed cells from the enveloping layer epithelium. In addition, Rab5-mediated endocytosis of E-cadherin is enhanced at the boundary between normal and RasV12 cells. Rab5 functions upstream of epithelial protein lost in neoplasm (EPLIN), which plays a positive role in apical extrusion of RasV12 cells by regulating protein kinase A. Furthermore, we have revealed that epithelial defense against cancer (EDAC) from normal epithelial cells substantially impacts on Rab5 accumulation in the neighboring transformed cells. This report demonstrates that Rab5-mediated endocytosis is a crucial regulator for the competitive interaction between normal and transformed epithelial cells in mammals.

At the initial stage of carcinogenesis, transformation occurs in a single cell within the epithelium. However, it is not clearly understood what happens at the interface between the newly emerging transformed cells and surrounding normal epithelial cells. In previous studies, we and other groups have demonstrated that when cells with an oncogenic mutation, such as RasV12 or v-Src, are surrounded by normal epithelial cells, the transformed cells are apically extruded from the epithelial monolayer (1–5). During this process, normal epithelial cells can recognize and actively eliminate the neighboring transformed cells through dynamic regulation of the cytoskeletal protein filamin, a phenomenon called EDAC (epithelial defense against cancer) (6), implying a notion that the normal epithelium has antitumor activity that does not involve immune systems. In addition, EPLIN (epithelial protein lost in neoplasm) is accumulated in transformed cells when they are surrounded by normal epithelial cells, and the EPLIN accumulation plays a crucial role in apical extrusion of the transformed cells (7). However, the molecular mechanisms of elimination of transformed cells still remain largely unknown, including the link between EDAC and EPLIN accumulation.

Endocytosis is a process by which cells take up macromolecules from the plasma membrane and extracellular space. Through endocytosis, internalized molecules are first transported into early endosomes and often further targeted into late endosomes and lysosomes, leading to degradation of the cargo molecules. This endocytic event is involved in various cellular processes, such as cell motility, cell proliferation, and oncogenesis (8–11). Previous studies in Drosophila have demonstrated that endocytosis also plays a crucial role in cell competition where cells with different properties compete with each other for survival (12–14). However, it is not known whether and how endocytosis is involved in the interaction between normal and transformed epithelial cells in vertebrates.

In this study, using mammalian cultured cells and zebrafish embryos, we have demonstrated that Rab5-mediated endocytosis is enhanced in Ras-transformed cells that are surrounded by normal epithelial cells, which positively regulates the elimination of the transformed cells from epithelia by linking EDAC and EPLIN.

Results

Rab5 Accumulates in Ras- or Src-Transformed Cells That Are Surrounded by Normal Cells.

To explore the involvement of endocytosis in the interaction between normal and transformed epithelial cells, we first examined the localization of Rab5 that plays a crucial role in the internalization and transport of endocytic vesicles to early endosomes and in the endosomal fusion (15–17). To this end, we used Madin-Darby canine kidney (MDCK) cells stably expressing GFP-tagged oncogenic Ras (RasV12) in a tetracycline-inducible manner (1). We found that Rab5 was substantially accumulated in RasV12-transformed cells when they were surrounded by normal epithelial cells (Fig. 1 A–C and Fig. S1A). The accumulation of Rab5 was not observed when normal cells or RasV12-transformed cells were cultured alone, or expression of RasV12 was not induced in the absence of tetracycline (Fig. 1 A–C). The analyses of confocal images at higher magnification showed that the number and total immunofluorescence intensity of intracellular Rab5 puncta were significantly increased in RasV12 cells surrounded by normal cells, compared with those in RasV12 cells cultured alone (Fig. S1 B and C). Comparable noncell-autonomous accumulation was also observed for exogenously expressed Rab5 (Fig. S1 D and E). These data indicate that the presence of surrounding normal epithelial cells induces accumulation of Rab5⁺ vesicles in RasV12-transformed cells. In addition, the noncell-autonomous Rab5 accumulation also occurred in Src-transformed cells that were surrounded by normal epithelial cells (Fig. 1 D–F). Furthermore, we found that immunofluorescence intensity of the early endosomal marker EEA1 was enhanced in RasV12 cells surrounded by normal cells (Fig. 1G), suggesting that endosomal formation is elevated in a noncell-autonomous fashion. In contrast, the Golgi marker GM130 or the recycling endosome marker Rab11 was not accumulated in RasV12-transformed cells that were surrounded by normal cells (Fig. S2), indicating the specific accumulation of Rab5.

Fig. 1.

Rab5 is accumulated in Ras- or Src-transformed cells when they are surrounded by normal epithelial cells. (A, B, D, and E) Immunofluorescence of Rab5 shown by confocal images of xy (A and D) or xz (B and E) sections. The confocal images are xy images if not indicated. MDCK-pTR GFP-RasV12 cells (A and B) or MDCK-pTR cSrcY527F-GFP cells (D and E) were mixed with normal MDCK cells or cultured alone on collagen gels. Cells were fixed after 16-h incubation with tetracycline and stained with anti-Rab5 antibody (red) and Hoechst (blue). For Tet (-), MDCK-pTR GFP-RasV12 cells or MDCK-pTR cSrcY527F-GFP cells were prestained with CMFDA (green; white asterisks), and the mixture of cells was incubated without tetracycline. (Scale bars, 10 μm.) (C and F) Quantification of the fluorescence intensity of Rab5 for A and D, respectively. Data are mean ± SD from four (C) or three (F) independent experiments. *P < 0.05; n = 156, 147, 156, and 155 cells (C) or n = 155, 103, 154, and 153 cells (F). Values are expressed as a ratio relative to MDCK cells. (G) Immunofluorescence of EEA1 shown by confocal images of xz sections. Cells were incubated in the same way as described above and stained with anti-EEA1 antibody (red) and Hoechst (blue). (Scale bar, 10 μm.)

Fig. S1.

Exogenously expressed Rab5 also accumulates in RasV12-transformed cells that are surrounded by normal epithelial cells. (A) Accumulation of endogenous Rab5 in an extruding single RasV12 cell that is surrounded by normal cells. Immunofluorescence of endogenous Rab5 shown by confocal images of xy or xz sections. MDCK-pTR GFP-RasV12 cells were mixed with normal MDCK cells or cultured alone on collagen gels. Cells were fixed after 16-h incubation with tetracycline and stained with anti-Rab5 antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) This result indicates that accumulation of Rab5 also occurs in a single RasV12 cell surrounded by normal cells. Exogenously expressed RasV12 proteins are often accumulated at the endoplasmic reticulum (ER) or Golgi compartments because of the limited capacity of the prenylation machinery. Thus, GFP-RasV12–decorated ER or Golgi is observed in some planes of confocal images. (B) Fluorescence images of intracellular Rab5 puncta. Cells were fixed after 16-h incubation with tetracycline and stained with anti-Rab5 antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) (C) Quantification of intracellular Rab5. The number of Rab5 granules (Left) or the total fluorescence intensity (Right) in each RasV12 cell was depicted as a dot. The red bars indicate the mean of the results. *P < 0.05, **P < 0.001; n = 30 cells for each condition. (D) Immunofluorescence of exogenously expressed Rab5 shown by confocal images of xy or xz sections. MDCK-pTR GFP-RasV12 FLAG-Rab5WT cells were mixed with normal MDCK cells or cultured alone on collagen gels. Cells were fixed after 16-h incubation with tetracycline and stained with anti-FLAG antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) (E) Quantification of the fluorescence intensity of FLAG-Rab5 for Fig. S1D. Data are mean ± SD from three independent experiments. *P < 0.05; n = 98 and 99 cells. Values are expressed as a ratio relative to Ras alone.

Fig. S2.

The Golgi marker GM130 or the recycling endosome marker Rab11 does not accumulate in RasV12-transformed cells that are surrounded by normal cells. (A) Immunofluorescence of GM130 shown by confocal images. MDCK-pTR GFP-RasV12 cells were mixed with normal MDCK cells or cultured alone on collagen gels. Cells were fixed after 16-h incubation with tetracycline and stained with anti-GM130 antibody (red) and Hoechst (blue). For Tet (-), MDCK-pTR GFP-RasV12 cells were prestained with CMFDA (green) (asterisks), and the mixture of cells was incubated without tetracycline. (Scale bar, 10 μm.) (B) Immunofluorescence of exogenously expressed Rab11 shown by confocal images of xy (Left) or xz (Right) sections. MDCK-pTR GFP-RasV12 HA-Rab11WT cells were mixed with normal MDCK cells or cultured alone on collagen gels. Cells were fixed after 16-h incubation with tetracycline and stained with anti-HA antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) It should be noted that in MDCK cells Rab11 accumulates around the centrosome immediately beneath the apical plasma membrane (45). (C) Establishment of MDCK-pTR GFP-RasV12 cells stably expressing HA-Rab11WT or HA-Rab11DN. Cell lysates were examined by Western blotting with the indicated antibodies.

Rab5 Plays a Positive Role in the Apical Extrusion of RasV12-Transformed Cells.

We next examined the endocytic uptake by analyzing the incorporation of transferrin. Cells were incubated with fluorescence-conjugated transferrin, and intracellularly incorporated transferrin was quantified (18). We found that both the number and total intensity of transferrin granules were significantly elevated in RasV12-transformed cells surrounded by normal cells, compared with those in RasV12 cells cultured alone (Fig. 2 A and B), suggesting that clathrin-dependent endocytosis is enhanced under the former condition. We further studied whether the noncell-autonomous up-regulation of endocytosis influences the behavior of transformed cells. Incubation with chlorpromazine, an inhibitor of clathrin-mediated endocytosis, suppressed apical extrusion of RasV12 cells that were surrounded by normal cells (Fig. 2 C and D and Fig. S3A). In addition, expression of a dominant-negative mutant of Rab5 (Rab5S34N, hereafter referred to as Rab5DN) in RasV12 cells substantially diminished their apical extrusion (Fig. 2 E and F and Fig. S3 A–C), whereas expression of a dominant-negative mutant of Rab11 did not (Figs. S2C and S3 D and E).

Fig. 2.

Rab5-regulated endocytosis plays a positive role in apical extrusion of RasV12-transformed cells. (A and B) Internalization of Alexa Fluor-647–conjugated transferrin into MDCK-pTR GFP-RasV12 cells that are cultured alone or mixed with normal MDCK cells. After 16 h of tetracycline addition, cells were incubated with Alexa Fluor-647–conjugated transferrin, followed by acid-wash to remove surface-attached transferrin. (A) Fluorescence images of internalized Alexa Fluor-647–conjugated transferrin (Tf; red). (Scale bar, 10 μm.) (B) Quantification of internalized transferrin. The number of transferrin granules (Left) or the total fluorescence intensity (Right) in each RasV12 cell was depicted as a dot. The red bars indicate the mean of the results. **P < 0.001, *P < 0.05; n = 30 cells for each condition. (C and D) Effect of chlorpromazine on apical extrusion of MDCK-pTR GFP-RasV12 cells surrounded by normal MDCK cells. (C) After incubation with tetracycline in the presence or absence of chlorpromazine for 24 h, cells were stained with Alexa Fluor-568 phalloidin (red) and Hoechst (blue). (Scale bar, 10 μm.) (D) Quantification of apical extrusion of RasV12 cells. Data are mean ± SD from three independent experiments. **P < 0.001; n = 280 and 292 cells. (E and F) Effect of expression of Rab5DN on apical extrusion of MDCK-pTR GFP-RasV12 cells. MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12-HA-Rab5S34N (Rab5DN) cells (clone 1) were mixed with normal MDCK cells, and incubated with tetracycline for 24 h. (E) Cells were stained with Alexa Fluor-568 phalloidin (red) and Hoechst (blue). (Scale bar, 10 μm.) (F) Quantification of apical extrusion of RasV12 cells. Data are mean ± SD from four independent experiments. *P < 0.05; n = 230 cells for each condition. As to MDCK-pTR GFP-RasV12-HA-Rab5S34N (Rab5DN) cell lines, clone 1 and clone 2 showed comparable phenotypes (e.g., Fig. S2C); thus, hereafter the data with clone 1 are presented.

Fig. S3.

Expression of Rab5DN, but not Rab11DN, suppresses apical extrusion of RasV12-transformed cells. (A) Effect of chlorpromazine or Rab5DN coexpression on colocalization between EEA1 and internalized transferrin. MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12-Rab5DN cells (clone 1) were mixed with normal MDCK cells, followed by tetracycline incubation with or without chlorpromazine for 16 h. Cells were further incubated with Alexa Fluor-568–conjugated transferrin acid-washed to remove surface-attached transferrin, and stained with anti-EEA1 antibody. Fluorescence intensity of transferrin overlapping EEA1 was quantified. Data are mean ± SD from three independent experiments. *P < 0.01; n = 92, 92, and 92 cells. Values are expressed as a ratio relative to Ras (-). (B) Establishment of MDCK-pTR GFP-RasV12 cells stably expressing HA-Rab5S34N. Cell lysates were examined by Western blotting with the indicated antibodies. (C) Effect of expression of Rab5DN on apical extrusion of MDCK-pTR GFP-RasV12 cells. MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12-HA-Rab5S34N (Rab5DN) cells (clone 2) were mixed with normal MDCK cells, and incubated with tetracycline for 24 h. Data are mean ± SD from three independent experiments. *P < 0.05; n = 301 and 307 cells. The two clones have given comparable phenotypes, and hereafter we mainly showed data using clone 1. (D and E) Effect of expression of Rab11DN on apical extrusion of MDCK-pTR GFP-RasV12 cells. MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12-HA-Rab11S25N (Rab11DN) cells were mixed with normal MDCK cells, and incubated with tetracycline for 24 h. (D) Cells were stained with Alexa Fluor-647 phalloidin (red) and Hoechst (blue). (Scale bar, 10 μm.) (E) Quantification of apical extrusion of RasV12 cells. Data are mean ± SD from three independent experiments. n = 306 and 309 cells. n.s., not significant.

In the previous studies, we have demonstrated that Src-transformed cells are apically extruded from a monolayer of the enveloping layer (EVL), the outermost epithelium of zebrafish embryos (2, 6). By using this experimental system, we demonstrated that Rab5 was accumulated in Src-transformed cells that emerged in a mosaic-manner within the normal epithelium (Fig. 3A), and that coexpression of Rab5DN significantly suppressed apical extrusion of the transformed cells (Fig. 3 B and C). Collectively, these data suggest that Rab5-mediated endocytosis plays a positive role in the elimination of transformed cells from epithelia.

Fig. 3.

Rab5 plays a crucial role in apical extrusion of v-Src–expressing cells in the EVL of zebrafish embryos. (A) Confocal images of a Rab5-YFP–expressing zebrafish embryo (at 10–11 h postfertilization, 3 h after tamoxifen treatment) with RFP–v-Src in a mosaic manner. (Scale bar, 20 μm.) (B and C) Effect of expression of Rab5DN on apical extrusion of v-Src–expressing cells. Embryos were injected with the GFP–v-Src–expressing vector with or without the Rab5DN cassette. Representative images are shown in B. Arrows indicate apically extruded v-Src–expressing cells. (Scale bar, 10 μm.) (C) Quantification of apical extrusion of v-Src–expressing cells. Data are mean ± SD from four independent experiments. *P < 0.05; n = 490 and 364 cells.

Vesicle Transport Through Late Endosomes Is Also Involved in Apical Extrusion of Transformed Cells.

Some of the cargo molecules transported into early endosomes are destined for lysosomal degradation via vesicle transport through multivesicular bodies/late endosomes (11, 19). Tsg101 localizes at the multivesicular body/late endosome and is involved in maturation of late endosomes as a component of the endosomal sorting complex required for transport (ESCRT)-I complex (20). We found that Tsg101 was accumulated in RasV12-transformed cells surrounded by normal epithelial cells, but not in RasV12 cells cultured alone (Fig. 4A). Comparable noncell-autonomous accumulation was also observed for another ESCRT-I component, Vps28 (Fig. S4A). Expression of Rab5DN significantly suppressed Tsg101 accumulation (Fig. 4 B and C), suggesting that accumulation of Tsg101 is dependent on Rab5 activity. Rab7 is another late endosome marker, which is required for cargo transport from the late endosome to the lysosome (11). Rab7 was also specifically accumulated in RasV12 cells surrounded by normal cells (Fig. S4B), and the accumulation was diminished by expression of Rab5DN (Fig. S4C). Bafilomycin inhibits the activity of a V-type ATPase in the endosome membrane, thereby increasing pH within endosomal compartments. Treatment with bafilomycin blocks the transport of endocytosed cargos from early to late endosomes (21). We showed that addition of bafilomycin significantly suppressed apical extrusion of RasV12-transformed cells that were surrounded by normal cells (Fig. 4 D and E). Next, we established Tsg101-knockdown RasV12 cells (Fig. 4F). Knockdown of Tsg101 strongly attenuated Vps28 accumulation (Fig. S4D) (22), and substantially suppressed apical extrusion of RasV12-transformed cells (Fig. 4G). Taken together, these data suggest that vesicle transport through late endosomes also plays an important role in apical extrusion of transformed cells.

Fig. 4.

Vesicle transport through late endosomes is involved in apical extrusion of transformed cells. (A) Accumulation of Tsg101 in RasV12-transformed cells that are surrounded by normal epithelial cells. MDCK-pTR GFP-RasV12 cells were mixed with normal MDCK cells or cultured alone on collagen gels. Cells were fixed after 16-h incubation with tetracycline and stained with anti-Tsg101 antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) (B and C) Effect of coexpression of Rab5DN on the accumulation of Tsg101. MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12-Rab5DN cells were mixed with normal MDCK cells, and incubated with tetracycline for 16 h. (B) Cells were stained with anti-Tsg101 antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) (C) Quantification of the fluorescence intensity of Tsg101 for B. Data are mean ± SD from three independent experiments. *P < 0.02; n = 104 and 110 cells. Values are expressed as a ratio relative to RasV12 cells. (D and E) Effect of bafilomycin on apical extrusion of RasV12-transformed cells surrounded by normal epithelial cells. MDCK-pTR GFP-RasV12 cells were mixed with normal MDCK cells, and incubated with tetracycline in the presence or absence of bafilomycin for 24 h. (D) Cells were stained with Alexa Fluor-568 phalloidin (red) and Hoechst (blue). (Scale bar, 10 μm.) (E) Quantification of apical extrusion of RasV12 cells. Data are mean ± SD from three independent experiments. *P < 0.002; n = 376 and 306 cells. (F) Establishment of MDCK-pTR GFP-RasV12 cells stably expressing Tsg101-shRNA. Cell lysates were examined by Western blotting with the indicated antibodies. (G) Effect of Tsg101-knockdown on apical extrusion of RasV12 cells. MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12 Tsg101-shRNA 1 or -shRNA 2 cells were mixed with normal MDCK cells, and incubated with tetracycline for 24 h. Data are mean ± SD from three independent experiments. *P < 0.05; n = 316, 319, and 316 cells.

Fig. S4.

Vesicle transport through late endosomes is involved in apical extrusion of transformed cells. (A) Coaccumulation of Vps28 and Tsg101 in RasV12-transformed cells that are surrounded by normal epithelial cells. MDCK-pTR GFP-RasV12 cells were mixed with normal MDCK cells or cultured alone on collagen gels. Cells were fixed after 6-h incubation with tetracycline and stained with anti-Vps28 (white) and anti-Tsg101 (red) antibodies and Hoechst (blue). (Scale bar, 10 μm.) (B) Accumulation of Rab7 in RasV12 cells surrounded by normal cells. Cells were incubated as described above, and were fixed after 16-h incubation with tetracycline, followed by staining with anti-Rab7 antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) (C) Effect of coexpression of Rab5DN on Rab7 accumulation in RasV12 cells surrounded by normal cells. MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12-Rab5DN cells were mixed with normal MDCK cells, and incubated with tetracycline for 16 h. Cells were stained with anti-Rab7 antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) (D) Effect of Tsg101-knockdown on Vps28 accumulation in RasV12 cells surrounded by normal cells. MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12 Tsg101-shRNA 1 cells were mixed with normal MDCK cells, and incubated with tetracycline for 6 h. Cells were stained with anti-Vps28 (white) and anti-Tsg101 (red) antibodies and Hoechst (blue). (Scale bar, 10 μm.)

Rab5-Regulated Endocytosis Acts Upstream of EPLIN in the Interaction Between Normal and Transformed Epithelial Cells.

In the previous studies, we have reported that E-cadherin–based cell–cell adhesions are dynamically modulated at the boundary between normal and transformed cells (1, 2). Moreover, E-cadherin has been reported to be endocytosed via Rab5 (23–26). Thus, we next examined the link between Rab5-mediated endocytosis and E-cadherin. When normal or RasV12-transformed cells were cultured alone, E-cadherin mainly localized at cell–cell contact sites (Fig. 5A) (1). However, when RasV12 cells were surrounded by normal cells, cytoplasmic puncta of E-cadherin substantially increased (Fig. 5 A and B), which partially overlapped EEA1 immunofluorescence (Fig. S5A). Cytoplasmic puncta were not observed for the tight junction marker occludin or the cell-adhesion marker integrin αVβ3 (Fig. S5 B and C), suggesting the specific effect on E-cadherin, at least to a certain extent. By superresolution microscopic analyses, we further demonstrated that the number and total immunofluorescence intensity of intracellular E-cadherin puncta were significantly increased in RasV12 cells surrounded by normal cells, compared with those in RasV12 cells cultured alone (Fig. 5 C and D). On the other hand, intracellular localization of E-cadherin was substantially suppressed in Rab5DN-expressing RasV12 cells (Fig. 5 A and B). In addition, E-cadherin immunofluorescence became often blurred at cell–cell contact sites between normal and RasV12 cells, but remained intact between normal and Rab5DN-expressing RasV12 cells (Fig. S6A). These data suggest that E-cadherin undergoes internalization via Rab5-mediated endocytosis, which is—at least partly—involved in the dynamic regulation of cell–cell adhesions between normal and transformed cells. Intracellular E-cadherin puncta were also observed in about 20% of the surrounding normal cells (Fig. S6B), which may be passively induced by the loss of E-cadherin from the plasma membrane on the adjacent transformed cell.

Fig. 5.

Rab5-mediated endocytosis of E-cadherin is enhanced in RasV12-transformed cells surrounded by normal cells. (A and B) Effect of coexpression of Rab5DN on cytoplasmic E-cadherin puncta in RasV12-transformed cells that are surrounded by normal epithelial cells. MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12-Rab5DN cells were mixed with normal MDCK cells or cultured alone on collagen gels. (A) Cells were fixed after 16-h incubation with tetracycline and stained with anti–E-cadherin antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) (B) Quantification of the intracellular fluorescence intensity of E-cadherin. Data are mean ± SD from four independent experiments. *P < 0.05, **P < 0.02; n = 126, 128, and 127 cells. (C and D) Superresolution microscopic analyses of intracellular E-cadherin puncta. (C) Fluorescence images of intracellular E-cadherin puncta (red). Cells were fixed after 16-h incubation with tetracycline and stained with anti–E-cadherin antibody (red). The area in the white dashed box is shown at 2.5-fold magnification (Right). (Scale bar, 10 μm.) (D) Quantification of intracellular E-cadherin. The number of E-cadherin granules (Left) or the total fluorescence intensity (Right) in each RasV12 cell was depicted as a dot. The red bars indicate the mean of the results. **P < 0.01; n = 30 cells for each condition.

Fig. S5.

Endocytosis is enhanced for E-cadherin, but not for the tight junction marker occludin or the cell-adhesion marker integrin αVβ3. (A) Partial colocalization of intracellular E-cadherin puncta and the early endosomal marker EEA1. MDCK-pTR GFP-RasV12 cells were mixed with normal MDCK cells on collagen gels. Cells were fixed after 16-h incubation with tetracycline and stained with anti–E-cadherin (red) and anti-EEA1 (green) antibodies and Hoechst (blue). GFP-RasV12 is shown in white. Arrows and arrowheads indicate cytoplasmic E-cadherin puncta with or without colocalization with EEA1, respectively. The white dashed line in the xz panel denotes the cross-sections represented in the above xy panels. The confocal images are xy images if not indicated. The area in the white dashed box is shown at 3.9-fold magnification, Right. (Scale bar, 10 μm) (B and C) Immunofluorescence of occludin (B) and integrin αVβ3 (C) shown by confocal images of xy or xz sections. MDCK-pTR GFP-RasV12 cells were mixed with normal MDCK cells on collagen gels. Cells were fixed after 16-h incubation with tetracycline and stained with antioccludin or -integrin αVβ3 antibody (red) and Hoechst (blue). (Scale bars, 10 μm.)

Fig. S6.

Detailed analyses of E-cadherin localization at the interface between normal and RasV12-transformed cells. (A) Disrupted E-cadherin–based cell–cell adhesions between normal and RasV12-transformed cells. MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12-Rab5DN cells were mixed with normal MDCK cells on collagen gels. Cells were fixed after 16-h incubation with tetracycline and stained with anti-EPLIN (red) and anti–E-cadherin (green) antibodies. GFP-RasV12 is shown in white. Arrows indicate disrupted E-cadherin–based cell–cell contact sites. (Scale bar, 10 μm.) (B) Intracellular puncta of E-cadherin are occasionally observed in the neighboring normal cells. MDCK-pTR GFP-RasV12 cells were mixed with normal MDCK cells or cultured alone on collagen gels. Cells were fixed after 16-h incubation with tetracycline and stained with anti-E-cadherin antibody (red) and Hoechst (blue). The area in the white dashed box is shown at 2.6-fold magnification, Right. (Scale bar, 10 μm.) The white line in the right panels indicates the contour of a RasV12 cell.

Next, we examined EPLIN, which forms a complex with E-cadherin at cell–cell adhesions (27). EPLIN has also been reported to play an active role in apical extrusion of the transformed cells through the activation of protein kinase A (PKA) (7, 28). When normal or RasV12-transformed cells were cultured alone, EPLIN was mainly localized at cell–cell adhesions (Fig. 6A) (7, 27). In contrast, in RasV12-transformed cells that were surrounded by normal cells, EPLIN was translocated from cell–cell contacts to the cytoplasmic region (Fig. 6 A–C) (7). Intracellularly accumulated EPLIN was only partially colocalized with the cytoplasmic E-cadherin puncta (Fig. 6B). The quantitative immunofluorescence analysis showed that 45% of EPLIN⁺ vesicles were not costained with E-cadherin (n = 1,800 EPLIN⁺ intracellular puncta). In addition, 81% of Rab7/E-cadherin double-positive vesicles were EPLIN⁻ [n = 100 Rab7 (+) E-cadherin (+) puncta], suggesting that during endocytosis, EPLIN often dissociates from the E-cadherin complex. We then found that coexpression of Rab5DN significantly suppressed intracellular accumulation of EPLIN in RasV12 cells surrounded by normal cells (Fig. 6 A and C). Addition of chlorpromazine also diminished the accumulation of EPLIN (Fig. S7A). Furthermore, PKA activity was elevated in RasV12 cells surrounded by normal cells (Fig. 6 D and E) (7), but was significantly suppressed by coexpression of Rab5DN (Fig. 6 D and E). PKA inhibitor KT5720 significantly suppressed apical extrusion of RasV12 cells (Fig. S7B), indicating a crucial role of PKA in this process. In contrast, knockdown of EPLIN did not affect Rab5 accumulation (Fig. S7 C and D). Collectively, these data suggest that Rab5-regulated endocytosis functions upstream of the EPLIN/PKA pathway.

Fig. 6.

Rab5-regulated endocytosis acts upstream of the EPLIN/PKA pathway in the interaction between normal and transformed epithelial cells. (A–C) Effect of coexpression of Rab5DN on EPLIN accumulation in RasV12-transformed cells that are surrounded by normal epithelial cells. MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12-Rab5DN cells were mixed with normal MDCK cells or cultured alone on collagen gels. (A) Cells were fixed after 16-h incubation with tetracycline and stained with anti-EPLIN (red) and anti–E-cadherin (green) antibodies and Hoechst (blue). GFP-RasV12 is shown in white. The white dashed line in the xz panel denotes the cross-sections represented in the above xy panels. The confocal images are xy images if not indicated. (Scale bar, 10 μm.) The area in the white dashed box is shown at 2.3-fold magnification in B. (B) Arrows and arrowheads indicate cytoplasmic E-cadherin puncta with or without colocalization with EPLIN, respectively. (C) Quantification of the intracellular fluorescence intensity of EPLIN. Data are mean ± SD from four independent experiments. *P < 0.05; n = 137, 133, 138, 134, and 135 cells. (D and E) Effect of coexpression of Rab5DN on PKA-catalyzed phosphorylation in RasV12 cells surrounded by normal cells. (D) Cells were incubated as described above except tetracycline treatment for 20 h, and stained with anti–phospho-PKA substrate (pPKAsub) antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) (E) Quantification of the fluorescence intensity of pPKAsub. Data are mean ± SD from three independent experiments. *P < 0.05, **P < 0.02; n = 103, 101, 101, 101, and 101 cells.

Fig. S7.

Effect of EPLIN-knockdown on Rab5 accumulation in RasV12-transformed cells surrounded by normal epithelial cells. (A) Effect of chlorpromazine on EPLIN accumulation. MDCK-pTR GFP-RasV12 cells were mixed with normal MDCK cells on collagen gels. After incubation with tetracycline in the presence or absence of chlorpromazine for 16 h, cells were stained with anti-EPLIN antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) (B) Quantification of the effect of the PKA inhibitor KT5720 on apical extrusion of RasV12 cells. Data are mean ± SD from three independent experiments. *P < 0.01; n = 280 and 292 cells. (C and D) MDCK-pTR GFP-RasV12 cells or MDCK-pTR GFP-RasV12 EPLIN-shRNA cells were mixed with normal MDCK cells on collagen gels. Cells were fixed after 16-h incubation with tetracycline and stained with anti-Rab5 antibody (red) and Hoechst (blue), shown by xy (C) or xz (D) sections. The confocal images are xy images if not indicated. (Scale bars, 10 μm.)

EDAC from the Surrounding Normal Cells Induces Rab5 Accumulation in RasV12-Transformed Cells.

We have previously reported that normal epithelial cells can recognize and actively eliminate the neighboring transformed cells through dynamic regulation of the cytoskeletal protein filamin, a process called EDAC (6). Accumulation of filamin in normal cells at the boundary with the neighboring transformed cells induces up-regulation of EPLIN in the transformed cells, eventually leading to their apical extrusion (7). In addition to filamin, E-cadherin in the surrounding normal cells also plays a vital role in apical extrusion of transformed cells (1). We found that when RasV12-transformed cells were surrounded by filamin-knockdown cells, accumulation of Rab5 was significantly suppressed (Fig. 7 A–C). Similarly, knockdown of E-cadherin in the surrounding normal cells also substantially diminished Rab5 accumulation in RasV12 cells (Fig. 7 D and E). Taken together, these data suggest that EDAC from normal epithelial cells plays a positive role in the regulation of endocytosis in the neighboring transformed cells.

Fig. 7.

EDAC from the surrounding normal cells induces Rab5 accumulation in RasV12-transformed cells. (A–C) Effect of filamin-knockdown in normal cells on Rab5 accumulation in the neighboring RasV12-transformed cells. MDCK-pTR GFP-RasV12 cells were mixed with normal or filamin-knockdown MDCK cells and incubated with tetracycline for 16 h. (A and B) Cells were stained with anti-Rab5 antibody (red) and Hoechst (blue). (Scale bars, 10 μm.) (C) Quantification of the fluorescence intensity of Rab5 in RasV12 cells. Data are mean ± SD from three independent experiments. *P < 0.05; n = 101 and 93 cells. Values are expressed as a ratio relative to MDCK:Ras = 50:1. (D and E) Effect of E-cadherin knockdown in normal cells on Rab5 accumulation in the neighboring RasV12 cells. MDCK-pTR GFP-RasV12 cells were mixed with normal or E-cadherin knockdown MDCK cells and incubated with tetracycline for 16 h. (D) Cells were stained with anti-Rab5 antibody (red) and Hoechst (blue). (Scale bar, 10 μm.) (E) Quantification of the fluorescence intensity of Rab5 in RasV12 cells. Data are mean ± SD from three independent experiments. *P < 0.05; n = 104 and 98 cells. Values are expressed as a ratio relative to MDCK:Ras = 50:1. (F) A schematic model of the molecular regulation at the interface between transformed and the neighboring normal cells. When transformed cells are surrounded by normal cells, Rab5-mediated endocytosis is up-regulated in the transformed cells, leading to E-cadherin endocytosis, EPLIN accumulation, and PKA activation, thereby inducing apical elimination of transformed cells. In addition, EDAC from the neighboring normal cells, via E-cadherin–based cell–cell adhesions and filamin accumulation, positively regulates Rab5-mediated endocytosis.

Discussion

In this study, we have presented several lines of evidence indicating that Rab5-mediated endocytosis and the following transport to late endosomes are enhanced and play an active role in the apical elimination of the transformed cells. First, Rab5 is accumulated in RasV12 cells surrounded by normal cells where clathrin-dependent endocytosis is elevated. Second, addition of chlorpromazine or coexpression of Rab5DN profoundly suppresses apical extrusion of transformed cells. Third, in the EVL of zebrafish embryos, Rab5 is accumulated in v-Src–expressing epithelial cells, and coexpression of Rab5DN significantly attenuates their apical delamination. Fourth, Tsg101 and Rab7 are accumulated in RasV12 cells surrounded by normal cells. Fifth, addition of bafilomycin or knockdown of Tsg101 suppresses the elimination of transformed cells. Collectively, these data demonstrate that endocytosis is a crucial regulator for the interaction between normal and transformed epithelial cells, which substantially influences the behavior and fate of transformed cells in the cell community.

It has been previously reported in Drosophila that endocytosis is involved in the interaction between normal and transformed epithelial cells and affects the outcome of cell competition between them (12–14). For example, Igaki et al. reported that in the eye disk epithelium, Rab5-mediated endocytosis is enhanced in scribble mutant cells that are surrounded by wild-type cells and that coexpression of Rab5DN strongly suppresses apoptosis of scribble mutant cells (13). Moreno and Basler showed that overexpression of Rab5 in wild-type epithelial cells diminishes their apoptosis at the interface with the neighboring myc-overexpressing cells in the wing discs (12). In addition, Ballesteros-Arias et al. reported that knockdown of Rab5 can induce cell competition in the wing discs in a cell-context–dependent manner (14). Herein, we demonstrate that endocytosis plays an important role in cell competition between normal and transformed epithelial cells in vertebrates as well. The data from Drosophila show that the Jun N-terminal kinase (JNK) pathway acts downstream of endocytosis (13, 14); however, we have not observed the noncell-autonomous activation of this pathway, suggesting the involvement of distinct cellular processes or signaling pathways in the elimination of RasV12-transformed cells. Here, we have revealed three molecular mechanisms whereby endocytosis regulates cell competition: (i) involvement of EDAC: filamin accumulation in the surrounding normal cells induces elevated endocytosis of transformed cells in a non-cell-autonomous fashion; (ii) dynamic modulation of cell–cell adhesions: endocytosis of E-cadherin is enhanced in transformed cells surrounded by normal cells; and (iii) specific up-regulation of E-cadherin endocytosis at the interface: endocytosis of other cell-adhesion proteins, such as occludin and integrin, is not enhanced. Hence, these findings provide insights into still enigmatic phenomena of cell competition and shed light on the unexplored events at the initial stage of carcinogenesis.

We show that E-cadherin is one of the cargo proteins via Rab5-mediated endocytosis, as reported by other groups using the different experimental conditions (23–26). In addition to E-cadherin, localization of EPLIN is also regulated by endocytosis. In a steady status, EPLIN forms a protein complex with E-cadherin via β-catenin and α-catenin at cell–cell adhesions (27). However, in RasV12 cells surrounded by normal cells, EPLIN translocates into the cytoplasm where it just partially colocalizes with internalized E-cadherin (Fig. 6B). Together with other data, these results show that it is plausible that after internalization, EPLIN dissociates from the E-cadherin complex, which is consistent with the previous reports that a certain fraction of catenins dissociates from E-cadherin during endocytosis (29–31). Free EPLIN would then form a complex with other binding partners, thereby activating signaling pathways, such as PKA, and promoting the apical extrusion event (Fig. 7F). However, it is likely that there are a number of proteins, in addition to E-cadherin, of which endocytosis is modulated by Rab5 in transformed cells; there could be additional adhesion molecules and signaling pathways that are regulated by endocytosis, and the sum of those overall orchestrated effects would modulate the behavior of RasV12-transformed cells.

In addition to PKA, EPLIN functions upstream of myosin-II; the activity of myosin-II is elevated in RasV12-transformed cells in a noncell-autonomous fashion, which is regulated by EPLIN (7). The activation of myosin-II generates pulling-forces exerted at the interface between normal and transformed cells, thereby promoting apical extrusion of the transformed cells (6). Along the same line, recent studies have revealed that physical forces or mechanical tensions can play a crucial role in cell competition between normal and transformed epithelial cells (32, 33). Thus, the EPLIN–myosin-II pathway could be another endocytosis-mediated regulatory mechanism for apical extrusion.

Another question is: What are the molecular mechanisms that induce the noncell-autonomous up-regulation of endocytosis? Our results suggest that EDAC from the surrounding normal cells regulates the elevation of Rab5-mediated endocytosis, implying that transformed cells sense the modulated conditions in the neighboring normal cells and accordingly respond to them by activating the endocytic pathways. To explore the molecular mechanisms whereby EDAC induces Rab5-mediated endocytosis, we have examined the effect of various inhibitors on Rab5 accumulation in RasV12-transformed cells that are surrounded by normal cells (Table S1). Among the tested inhibitors, the MEK inhibitor U0126 diminishes accumulation of Rab5. MEK is a component of the MAPK signaling pathway that is the cell-autonomous downstream target of Ras. Thus, this result suggests that among the downstream pathways of Ras, the MAPK pathway is involved in the accumulation of Rab5. Still, it is not clear at present whether the MAPK pathway is also directly involved in the further downstream, noncell-autonomous processes. The other inhibitors have no effect on Rab5 accumulation (Table S1); thus, so far we have been unable to identify a key signaling pathway that plays a direct role in the up-regulation of Rab5-mediated endocytosis, which needs to be identified in future studies.

Table S1.

Effect of various inhibitors on apical extrusion and Rab5 accumulation in RasV12-transformed cells that are surrounded by normal cells

		RasV12
Inhibitor	Target	Extrusion	Rab5 accumulation
KT5720	PKA	↓*	No effect
Blebbistatin	Myosin-II	↓*#	No effect
Cytochalasin D	Actin polymerization	↓*#	No effect
Rapamycin	mTOR	ND	No effect
U0126	MAPK	↓*#	↓*
Y27632	Rho kinase	↓*#	No effect
3-AT	Catalase	↓*	No effect

mTOR, mammalian target of rapamycin; ND, no data; #: our published observations.

^*Statistically significant.

Several studies have reported that Rab5 is involved in cancer development and progression. Elevated expression of Rab5 is observed in various types of cancer (34–36), and Rab5 plays a positive role in tumor invasion/metastasis (37–39). Our results demonstrate that Rab5-mediated endocytosis is also involved in the apical extrusion of transformed cells from the normal epithelium at the initial stage of carcinogenesis. Thus, what is the fate of the apically extruded transformed cells? For transformed cells to metastasize into distant organs, they have to leave the epithelium basally and invade the underlying matrix. In turn, in the apical lumen, cells are generally subjected to harsh physical conditions (e.g., stool, urine, digestive fluids). Hence, apical extrusion is the opposite direction from metastasis, and thus can be regarded as a cancer-preventive process. Indeed, our recent result using a cell competition mouse model has demonstrated that apically extruded cells do not form a tumorous mass and eventually disappear from the intestine tissues (40), although the pathological consequence of apical extrusion of transformed cells still remains controversial at present (4). To fully address this issue, the fate of apically extruded cells needs to be extensively studied in various epithelial tissues in vivo. By further clarifying the functional significance and molecular mechanisms of the Rab5-mediated endocytosis, it is expected that we could develop a novel type of cancer preventive medicine.

Materials and Methods

A complete description of the methods is provided in SI Materials and Methods. This description includes antibodies and materials, cell culture, immunofluorescence and Western blotting, transferrin-uptake assay, superresolution microscopy, microinjection, and confocal imaging of zebrafish embryos, and data analyses.

SI Materials and Methods

Antibodies and Materials.

Rabbit anti-Rab5 (ab18211), rabbit anti-Rab7 (ab77993), and mouse anti-Vps28 (ab139345) antibodies were purchased from Abcam. Mouse anti-EEA1 (610456) and mouse anti-GM130 (610822) antibodies were from BD Biosciences. Mouse anti-EPLIN (sc-136399) and goat anti-Tsg101 (sc-6037) antibodies were from Santa Cruz Biotechnology. Mouse anti-GAPDH (MAB374), mouse antiintegrin αVβ3 (MAB1976), and mouse anti-HA (05-904) antibodies were from Merck Millipore. Rabbit antiphospho-(Ser/Thr) PKA substrate (#9621S) antibody was from Cell Signaling Technology. Mouse anti-FLAG (M2; F3165) antibody was from Sigma-Aldrich. Mouse antioccludin (33-1500) antibody was from Thermo Fisher Scientific. Rat anti–E-cadherin (DECMA1; ab11512 and ECCD2; 13–1900) antibodies were from Abcam and Thermo Fisher Scientific, respectively. Anti–E-cadherin antibody (ECCD2) was used for superresolution microscopic analyses, whereas anti–E-cadherin antibody (DECMA1) was used in the other experiments. Alexa Fluor-568– or -647–conjugated phalloidin (Life Technologies) was used at 1.0 U/mL Alexa Fluor-568– and -647–conjugated secondary antibodies were from Life Technologies. Hoechst 33342 (Life Technologies) was used at a dilution of 1:5,000. For immunofluorescence, primary antibodies were used at 1:100, except for antiphospho-(Ser/Thr) PKA substrate antibody that was used at 1:25, anti-FLAG antibody at 1:1,000, and antioccludin, antiintegrin αVβ3, and anti–E-cadherin (ECCD2) antibodies at 1:50, and all secondary antibodies were used at 1:200. For Western blotting, peroxidase-conjugated anti-rabbit and anti-mouse secondary antibodies were from GE Healthcare and Jackson ImmunoReseach, respectively. Transferrin Alexa Fluor-647 conjugate (T23366) and Alexa Fluor-568 conjugate (T23365) were from Life Technologies. To fluorescently stain living MDCK-pTR GFP-RasV12 cells or MDCK-pTR cSrcY527F-GFP cells, CMFDA (green dye) (Life Technologies) was used according to the manufacturer’s instructions. Where indicated, the following inhibitors were used: bafilomycin A1 (5 nM; Calbiochem), KT5720 (4 μM; Calbiochem), (S)-(-)-blebbistatin (30 μM; Calbiochem), Cytochalasin D (4 μM; Sigma-Aldrich), Rapamycin (10 μM; Sigma-Aldrich), U0126 (40 μM; Promega), Y27632 (20 μM; Calbiochem), and 3-amino-1, 2, 4-triazole (80 mM; Tokyo Chemical Industry Co). For chlorpromazine (Sigma-Aldrich), 10 μg/mL was used, except for Fig. S5A, where 5 μg/mL was used.

Cell Culture.

MDCK cells were cultured in DMEM containing 10% (vol/vol) FBS and penicillin/streptomycin at 37 °C in ambient air supplemented with 5% (vol/vol) CO₂. MDCK-pTR GFP-RasV12 cells and MDCK-pTR cSrcY527F-GFP cells were cultured in the medium containing 10% (vol/vol) FBS, 5 μg/mL of blasticidin (Invivogen), and 400 μg/mL of zeocin (Invivogen) (1, 7). To establish MDCK-pTR GFP-RasV12 cells stably expressing HA-Rab5S34N, MDCK-pTR GFP-RasV12 cells were transfected with pCIneo-HA-canine Rab5S34N (a kind gift from Akira Kikuchi, Osaka University, Osaka, Japan) using Lipofectamine 2000 (Life Technologies), followed by selection in the medium containing 800 μg/mL of G418 (Gibco), 5 μg/mL of blasticidin, and 400 μg/mL of zeocin. We established two independent MDCK-pTR GFP-RasV12-HA-Rab5S34N cell lines (clone 1 and clone 2). The two clones have given comparable phenotypes, and we mainly showed data using clone 1. To construct PB-EF1-FLAG-Rab5aWT-IRES-Neo and PB-EF1-HA-Rab11WT/S25N-IRES-Neo, the cDNAs of canine Rab5aWT and rat Rab11WT/S25N were amplified from pCIneo-HA-Rab5WT and pCIneo-HA-rat Rab11WT/S25N (both constructs are gift from A. Kikuchi, Osaka University, Osaka, Japan) by PCR and inserted into the EcoRI/BamHI and BamHI/NotI site of PB-EF1-IRES-Neo, respectively. To establish MDCK-pTR GFP-RasV12 cells stably expressing FLAG-Rab5aWT or HA-Rab11WT/S25N, MDCK-pTR GFP-RasV12 cells were transfected with PB-EF1-FLAG-Rab5aWT-IRES-Neo or PB-EF1-HA-Rab11WT/S25N-IRES-Neo, respectively, using nucleofection (nucleofector 2b Kit L, Lonza), followed by selection in the medium containing 800 μg/mL of G418 (Gibco), 5 μg/mL of blasticidin, and 400 μg/mL of zeocin. MDCK-pTR GFP-RasV12 cells stably expressing EPLIN-shRNA were cultured as previously described (7). MDCK-pTR GFP-RasV12 cells stably expressing Tsg101-shRNA were established as follows. Tsg101-shRNA sequences (Tsg101-shRNA 1, 5′-GATCCCCGCAACAGGGCCACCAAATATTCAAGAGATATTTGGTGGCCCTGTTGCTTTTTC-3′ and 5′-TCGAGAAAAAGCAACAGGGCCACCAAATATCTCTTGAATATTTGGTGGCCCTGTTGCGGG-3′ or Tsg101-shRNA 2, 5′-GATCCCCGAGTAATAGATCTGGATGTTTCAAGAGAACATCCAGATCTATTACTCTTTTTC-3′ and 5′-TCGAGAAAAAGAGTAATAGATCTGGATGTTCTCTTGAAACATCCAGATCTATTACTCGGG-3′) were cloned into the BglII/XhoI site of pSUPER.neo+gfp (Oligoengine). MDCK-pTR GFP-RasV12 cells were transfected with pSUPER.neo+gfp Tsg101-shRNA using Lipofectamine 2000, followed by selection in the medium containing 800 μg/mL of G418, 5 μg/mL of blasticidin, and 400 μg/mL of zeocin. MDCK-pTR filamin A-shRNA cells and MDCK-pTR E-cadherin-shRNA cells were established and cultured as previously described (1, 6). For MDCK-pTR GFP-RasV12 cells or MDCK-pTR cSrcY527F-GFP cells, 2 μg/mL tetracycline (Sigma-Aldrich) was added to induce the expression of GFP-RasV12 or cSrcY527F-GFP. For Fig. 7 A–E, to induce sufficient knockdown of filamin A or E-cadherin protein, MDCK-pTR filamin A-shRNA cells or MDCK-pTR E-cadherin shRNA cells were incubated with 2 μg/mL tetracycline for 48 h or 96 h, respectively, before coincubation with MDCK-pTR GFP-RasV12 cells. All inhibitors were added together with tetracycline. For immunofluorescence, cells were plated onto collagen gel-coated coverslips. Type-I collagen (Cellmatrix Type I-A) was obtained from Nitta Gelatin and was neutralized on ice to a final concentration of 2 mg/mL, according to the manufacturer’s instructions.

Immunofluorescence and Western Blotting.

For immunofluorescence, MDCK-pTR GFP-RasV12 cells or MDCK-pTR cSrcY527F-GFP cells were mixed with MDCK cells at a ratio of 1:50 and cultured on the collagen matrix, as previously described (1). The mixture of cells was incubated for 8–12 h, followed by tetracycline treatment for 15–20 h (when various noncell-autonomous changes occur before apical extrusion) for immunofluorescence analyses, or for 24 h for analyses of apical extrusions. Cells were fixed with 4% (wt/vol) paraformaldehyde (PFA) in PBS and permeabilized as previously described (1), except for occludin and integrin αVβ3, for which cells were fixed in methanol at −20 °C for 3 h, followed by blocking in 10% (wt/vol) BSA/PBS for 1 h. Immunofluorescence images were analyzed at 0.5-μm xz-intervals by a scanning confocal microscope (FV1000 or FV1200 system; Olympus) equipped with a 60× oil-immersion objective (1.35 NA). For the images of occludin and Rab11, z-stacked images are presented. Acquisition software was Olympus FV10-ASW. For Fig. 2 A and B, fluorescence images were acquired using an inverted microscope (IX71; Olympus) on 60× oil-immersion objective (1.35 NA) and charge-coupled device camera (EXi Aqua; QImaging) with the Metamorph software (Molecular Devices). Images were quantified by the MetaMorph software. The confocal images in the figures are xy images if not indicated.

Western blotting was performed as previously described (41). Primary antibodies were used at 1:1,000 except anti-GAPDH antibody at 1:2,000. Stained gels and Western blotting data were analyzed using ImageQuant LAS4010 (GE Healthcare).

Transferrin-Uptake Assay.

For transferrin-uptake assay in Fig. 2 A and B, cells were preincubated in serum-free DMEM for 3 h, then incubated in serum-free DMEM containing 50 μg/mL of transferrin-Alexa Fluor-647 for 1 h at 4 °C for binding and for 3 min at 37 °C for internalization, then washed three times with ice-cold PBS (pH 3.0), followed by fixation with 4% (wt/vol) PFA in PBS. For transferrin-uptake assay in Fig. S2A, cells were treated as described above except that they were incubated with chlorpromazine for the last 30 min of preincubation where indicated, followed by incubation with transferrin-Alexa Fluor-568 for 1 h at 4 °C for binding and for 20 min at 37 °C for internalization.

Superresolution Microscopy.

Cells were cultured on collagen-gel–coated coverslips and fixed as described above, except using 0.17-mm-thick coverslips. For immunostaining of E-cadherin, cells were incubated with anti–E-cadherin antibody (ECCD2) for 2 h, followed by incubation with Alexa Fluor-568-conjugated secondary antibody at a dilution of 1:100 for 30 min. The stained samples were mounted onto ProLong Diamond Antifade Mountant (P36965, Thermo Fisher Scientific) on a glass slide. For the calibration of multicolor signals, TetraSpeck beads (T7280, Thermo Fisher Scientific) were added at a ratio of 1:50 in the mounting medium to generate the channel alignment file.

The SIM images were obtained with the Zeiss ELYRA PS.1 system on 100× objective (NA 1.46; Zeiss) and EM-CCD camera. SIM images were calculated using default settings with theoretically predicted point spread function parameters. Images were quantified by the MetaMorph software.

Microinjection and Confocal Imaging of Zebrafish Embryos.

The transgenic lines we used were as follows: krt18:KalTA4-ERT2 (6) and krt18:Rab5-Venus (YFP) generated in this study using the Tol2 system (42). We used mosaic expression in the EVL using the bidirectional promoter, dUAS, which allows us to express two different effectors or fluorescently labeled markers from a single unit of 5×UAS in a tamoxifen-inducible manner (43). For Fig. 3A, dUAS:Cherry–v-Src was coinjected with 20 pg of Tol2 RNA at the one- to two-cell stage in embryos obtained from crossing between krt18:KalTA4-ERT2 and krt18:Rab5-Venus fish, as previously described (6). For Fig. 3 B and C, embryos were treated as described above, except that either dUAS:GFPvv-Src or dUAS:GFP–v-Src;Rab5S36N was coinjected into embryos obtained from crossing between krt18:KalTA4-ERT2 and wild-type AB strain fish. The injected embryos were bathed in 0.5 μM Z-4-hydroxytamoxifen (Sigma-Aldrich H7904: a stock of 5 mM in ethanol) starting at 60% epiboly and were fixed in 4% (wt/vol) PFA at 1–2 somites stage (3 h after tamoxifen treatment) for whole-mount antibody staining. Whole-mount immunohistochemistry was performed as previously described with a minor modification (44). Stained embryos were mounted in 1% low-melting agarose (Sigma-Aldrich) in PBS. Confocal images were taken using a 20× or 40× water-immersion lens on Leica DM2500 microscope with the TCS SPE confocal system. The xy and xz projection images were produced using the Volocity software.

Data Analyses.

For data analyses, two-tailed Student’s t tests were used to determine P values except in Fig. 6E, where one-tailed Student’s t test was used. For quantification of apical extrusion or fluorescence intensity, more than 50 or 25 cells were analyzed for each experimental condition, respectively.