Spatial recruitment and activation of the Fes kinase by ezrin promotes HGF-induced cell scattering
2007; Springer Nature; Volume: 27; Issue: 1 Linguagem: Inglês
10.1038/sj.emboj.7601943
ISSN1460-2075
AutoresAlexandra Naba, Céline Reverdy, Daniel Louvard, Monique Arpin,
Tópico(s)Multiple Myeloma Research and Treatments
ResumoArticle29 November 2007free access Spatial recruitment and activation of the Fes kinase by ezrin promotes HGF-induced cell scattering Alexandra Naba Alexandra Naba Centre National de la Recherche Scientifique (CNRS), UMR 144, Paris, France Institut Curie, Centre de Recherche, Paris, France Search for more papers by this author Céline Reverdy Céline Reverdy Hybrigenics, Paris, France Search for more papers by this author Daniel Louvard Daniel Louvard Centre National de la Recherche Scientifique (CNRS), UMR 144, Paris, France Institut Curie, Centre de Recherche, Paris, France Search for more papers by this author Monique Arpin Corresponding Author Monique Arpin Centre National de la Recherche Scientifique (CNRS), UMR 144, Paris, France Institut Curie, Centre de Recherche, Paris, France Search for more papers by this author Alexandra Naba Alexandra Naba Centre National de la Recherche Scientifique (CNRS), UMR 144, Paris, France Institut Curie, Centre de Recherche, Paris, France Search for more papers by this author Céline Reverdy Céline Reverdy Hybrigenics, Paris, France Search for more papers by this author Daniel Louvard Daniel Louvard Centre National de la Recherche Scientifique (CNRS), UMR 144, Paris, France Institut Curie, Centre de Recherche, Paris, France Search for more papers by this author Monique Arpin Corresponding Author Monique Arpin Centre National de la Recherche Scientifique (CNRS), UMR 144, Paris, France Institut Curie, Centre de Recherche, Paris, France Search for more papers by this author Author Information Alexandra Naba1,2, Céline Reverdy3, Daniel Louvard1,2 and Monique Arpin 1,2 1Centre National de la Recherche Scientifique (CNRS), UMR 144, Paris, France 2Institut Curie, Centre de Recherche, Paris, France 3Hybrigenics, Paris, France *Corresponding author. Corresponding author. Institut Curie, Centre National de la Recherche Scientifique (CNRS), UMR 144, 26, rue d'Ulm, Paris 75248, France. Tel.: +33 1 42 34 63 68; Fax: +33 1 42 34 63 77; E-mail: [email protected] The EMBO Journal (2008)27:38-50https://doi.org/10.1038/sj.emboj.7601943 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info The remodeling of epithelial monolayers induced by hepatocyte growth factor (HGF) results in the reorganization of actin cytoskeleton and cellular junctions. We previously showed that the membrane–cytoskeleton linker ezrin plays a major role in HGF-induced morphogenic effects. Here we identified a novel partner of phosphorylated ezrin, the Fes kinase, that acts downstream of ezrin in HGF-mediated cell scattering. We found that Fes interacts directly, through its SH2 domain, with ezrin phosphorylated at tyrosine 477. We show that in epithelial cells, activated Fes localizes either to focal adhesions or cell–cell contacts depending on cell confluency. The recruitment and the activation of Fes to the cell–cell contacts in confluent cells depend on its interaction with ezrin. When this interaction is impaired, Fes remains in focal adhesions and as a consequence the cells show defective spreading and scattering in response to HGF stimulation. Altogether, these results provide a novel mechanism whereby ezrin/Fes interaction at cell–cell contacts plays an essential role in HGF-induced cell scattering and implicates Fes in the cross-talk between cell–cell and cell–matrix adhesion. Introduction Ezrin is a member of the highly conserved ezrin/radixin/moesin (ERM) protein family. The ERM proteins provide a regulated linkage between cell-surface proteins and the actin cytoskeleton (Bretscher et al, 2002). In its membrane–cytoskeleton-associated form, ezrin participates in several cortical actin-based processes such as cell adhesion (Takeuchi et al, 1994; Hiscox and Jiang, 1999; Pujuguet et al, 2003), cell motility (Crepaldi et al, 1997; Ng et al, 2001) and cell (Yonemura and Tsukita, 1999; Gautreau et al, 2000) and tissue (Fiévet et al, 2007) morphogenesis. The hepatocyte growth factor (HGF) acting through its receptor Met induces a variety of biological events in epithelial cells (Stella and Comoglio, 1999). Ezrin was shown to be an essential component in HGF-induced cell scattering (Crepaldi et al, 1997; Orian-Rousseau et al, 2002, 2007) and morphogenesis (Crepaldi et al, 1997; Gautreau et al, 1999). Ezrin phosphorylation at tyrosine residues is required at different steps of epithelial cell morphogenesis in a three-dimensional matrix. Thus, tyrosine 353 signals cell survival through the activation of the PI3K pathway, whereas tyrosine 145 controls cell proliferation (Gautreau et al, 1999; Srivastava et al, 2005). The above HGF-triggered events are also mediated by cytoplasmic effectors from the Src family kinases (Rahimi et al, 1998; Tsukamoto and Nigam, 1999; Cutrupi et al, 2000). We and others have shown that ezrin is phosphorylated by Src family kinases at either tyrosine 145 or 477 (Autero et al, 2003; Heiska and Carpen, 2005; Srivastava et al, 2005). Src-dependent ezrin phosphorylation influences cell adhesion-based signaling. The spreading of epithelial cells expressing Y145F ezrin is impaired when cells are grown on fibronectin and this defect can be reverted by overexpressing activated Src (Srivastava et al, 2005). In studies performed with mammary carcinoma cells, a cooperative effect of Src and ezrin in the regulation of cell–cell contacts has been observed (Elliott et al, 2004). These findings establish a functional link between ezrin and the Src family kinases in the control of cell adhesion. To identify the signaling components involved in adhesion-mediated events downstream of ezrin, we sought to identify proteins that interact with ezrin phosphorylated by Src family kinases using a modified two-hybrid screen. We found that the kinase Fes interacts specifically with phosphorylated ezrin. The fes protooncogene encodes a 93-kDa non-receptor protein-tyrosine kinase and is a member, along with the ubiquitous kinase Fer and the testis-specific form FerT, of the fes/fps non-receptor protein kinase subfamily (Greer, 2002). Fes displays a modular structure that consists of an N-terminus Fes/CIP4 homology (FCH) domain, followed by two coiled-coil domains and an SH2 (Src homology 2) domain (Figure 1A). The C-terminus kinase domain of the protein contains the major autophosphorylation site at tyrosine 713 (Rogers et al, 1996). Figure 1.Fes interacts with ezrin phosphorylated at tyrosine 477. (A) Structural organization of the Fes kinase. Fes contains an N-terminus Fes/CIP4 homology domain (FCH), two coiled-coil domains, an SH2 domain followed by the kinase domain. Tyrosine 713 is the major autophosphorylation site in the kinase domain. The ezrin-binding domain found in the two-hybrid screen comprises amino acids 255–593. (B) GST or GST-Fes255–593 immobilized on glutathione beads was incubated with lysates from cells expressing WT ezrin and treated with pervanadate alone or with PP2 or SU6656. Immunoblot was performed with anti-VSVG antibody. For each condition, lysates (inputs) were divided in two parts to achieve incubation with the GST alone or with GST-Fes255–593 (right panel). The interaction of WT ezrin with GST-Fes255–593 is abolished when the cells were treated with Src kinase inhibitors. Lower panels: Coomassie staining of GST proteins. (C) GST or GST-Fes255–593 immobilized on glutathione beads was incubated with lysates from cells expressing WT, Y145F or Y477F ezrin. Immunoblot was performed with anti-VSVG antibody. WT and Y145F ezrin interact with GST-Fes255–593, whereas Y477F ezrin does not. Lower panels: Coomassie staining of GST proteins. (D) Tyrosine 477 in ezrin is a substrate of Src. Immunoblot with anti-ezrin or anti-pY477 ezrin antibodies was performed on lysates of LLC-PK1 cells treated with pervanadate alone or with PP2 or SU6656. (E) GST-Fes255–593 binds to a peptide comprising phosphorylated tyrosine 477 of ezrin. Purified GST (lower lane) or GST-Fes255–593 (upper lane) proteins was incubated with beads alone, the ezrin peptide (PV477YEPV), the phosphopeptide (PVp477YEPV) or a control phosphopeptide immobilized on beads. Coomassie staining of the input GST proteins and bound protein are shown. Download figure Download PowerPoint Although Fes was mainly studied in hematopoietic cells of the myeloid lineage and in endothelial cells, this protein is also present in epithelial cells (Haigh et al, 1996; Delfino et al, 2006). Here we report that in its activated form Fes has a dual localization in epithelial cells, in focal adhesions or at cell–cell contacts, depending on cell confluency. We demonstrate that ezrin plays a crucial role in the spatial recruitment and activation of Fes at the membrane, and that the correct localization of Fes is essential for the transmission of signals elicited by both HGF and extracellular matrix components. Therefore, these experiments place the ezrin/Fes interaction at the crossroad of growth factors and extracellular matrix signaling for the control of HGF-induced epithelial cell scattering. Results The SH2 domain of Fes interacts with ezrin phosphorylated at tyrosine 477 To identify partners of ezrin phosphorylated by Src family kinases, we conducted a modified two-hybrid screen with yeast strains transformed or not with Lyn kinase. As bait, we used a truncated form of ezrin lacking a 52-amino-acid sequence at the C-terminal end, since in the full-length protein most of the binding sites for ezrin partners are masked. This masking is due to a strong intramolecular interaction between the N- and C-terminal domains of the protein (Gary and Bretscher, 1995). Among the preys that were found to interact specifically with phosphorylated ezrin and not found in the screen performed in the absence of Lyn kinase, we focused our study on Fes. The region that was found in the screen to interact with ezrin corresponds to the second coiled-coil domain and the SH2 domain of Fes (Figure 1A), this comprises amino acids 255–593 (Fes255–593). Different approaches were used to confirm the interaction between ezrin and Fes and its dependence on ezrin phosphorylation by Src family kinases. First, we performed pull-down experiments with GST-Fes255–593, using extracts from LLC-PK1 cells overexpressing wild-type (WT) ezrin treated with pervanadate alone or in combination with the Src family kinases inhibitors, PP2 or SU6656. GST-Fes255–593 but not GST alone interacted specifically with WT ezrin and this interaction was disrupted when the cells were treated with either PP2 or SU6656 (Figure 1B). Two ezrin residues have been described to be phosphorylated by Src family kinases, namely tyrosines 145 and 477. To determine if the phosphorylation of one of these two residues was required for the interaction between Fes and ezrin, we performed pull-down experiments with GST-Fes255–593 on extracts from cells expressing WT ezrin, or ezrin carrying the point mutations Y145F or Y477F. GST-Fes255–593 interacted specifically with WT ezrin and Y145F ezrin but not with Y477F ezrin (Figure 1C). Altogether, these results indicated that the interaction between Fes and ezrin occurs through phosphorylated tyrosine 477 of ezrin. To confirm that in our model system the tyrosine 477 of ezrin is a substrate of Src family kinases, we have generated a phosphospecific antibody directed against the phosphotyrosine 477 of ezrin (Supplementary Figure S1). As shown in Figure 1D, this antibody recognizes a band corresponding to ezrin in lysates of cells treated with pervanadate, and failed to recognize it when the cells were treated with PP2 or SU6656, indicating that the tyrosine 477 of ezrin is a substrate of Src in LLC-PK1 cells. We next wanted to determine if the interaction between the two proteins was direct and if it required the SH2 domain of Fes. Two peptides surrounding tyrosine 477 in a phosphorylated or non-phosphorylated form were synthesized (PVY477EPV or PVpY477EPV) and incubated with purified GST-Fes255–593. A peptide comprising phosphorylated tyrosine 190 of ezrin was used as control. Whereas no interaction of either peptide was detected with GST alone, we observed an interaction of GST-Fes255–593 with the peptide phosphorylated at tyrosine 477 (Figure 1E) but not with the control phosphopeptide. These data indicate that the interaction between ezrin and Fes is direct and occurs through phosphorylated tyrosine 477 of ezrin and the SH2 domain of Fes. Activated Fes shuttles between focal adhesions and cell–cell contacts We next wanted to examine where the interaction between ezrin and Fes takes place in epithelial cells. Immunofluorescence was performed on LLC-PK1 cells with an antibody generated against a peptide corresponding to the last 15 amino acids of Fes (Supplementary Figures S2 and S3). LLC-PK1 cells derive from proximal kidney tubules and form a polarized monolayer in culture. In these cells, ezrin is mainly present in the apical microvilli but is also observed at the lateral membrane. As shown in Figure 2A, Fes displays a punctate cytoplasmic staining concentrated at the basal region of LLC-PK1 cell (Figure 2A, xz section). No obvious colocalization with ezrin was observed. Figure 2.Activated Fes localizes to adhesion sites in epithelial cells. (A) Fes displays a punctate cytoplasmic staining. Double staining was performed on confluent LLC-PK1 cells with anti-Fes (green) and anti-ezrin (red) antibodies. Images were taken with a 3D optical sectioning wide-field microscope and restored using the iterative constrained modified Gold algorithm. A 3D maximum-intensity projection (MIP) along the z-axis of the whole volume (5-μm thick) and an xz cross-section are shown. No colocalization between ezrin and Fes is observed (bar, 5 μm). (B) In non-confluent cells, activated Fes (pY713 Fes, green) is observed in focal adhesions, where it colocalizes with vinculin (red) and actin stress fibers as shown by the phalloidin staining (blue) (bar, 10 μm). (C) In confluent cells, activated Fes localizes at cell–cell contacts. pY713 Fes (green) colocalizes with E-cadherin (red) (upper panels) or ezrin (red) (lower panels). Images were taken with a 3D wide-field optical sectioning microscope and restored by deconvolution. 3D MIP along the z-axis at the level of adherens junctions (0.5 μm thick) are shown (bar, 10 μm). (D) Endogenous ezrin co-immunoprecipitates with activated Fes. Immunoprecipitation was performed on LLC-PK1 cell lysates with anti-pY713 Fes antibody or non-immune rabbit IgG. Immunoprecipitated proteins and total cell lysates (inputs) were immunoblotted with the anti-pY713 Fes or anti-ezrin antibodies. Download figure Download PowerPoint The autophosphorylation of Fes at tyrosine 713 reflects its transition from an inactive to a catalytically active kinase. We then asked where in epithelial cells the active form of Fes was localized. Using an antibody against phosphorylated tyrosine 713 of Fes (Supplementary Figures S2 and S3) we observed that the major pool of activated Fes was concentrated in two different subcellular compartments. In sparse cells, activated Fes was mainly present in focal adhesions, where it colocalized with vinculin (Figure 2B) paxillin and FAK (data not shown). However, in confluent monolayers, the localization of activated Fes was restricted to the lateral membrane. This Fes staining appeared discontinuous and overlapped with the staining of the adherens junction marker E-cadherin (Figure 2C, upper panel). A colocalization was also observed with ezrin present at the lateral surface (Figure 2C, lower panel). Given the colocalization between activated Fes and ezrin, we performed an immunoprecipitation using the anti-pY713 Fes antibody. As shown in Figure 2D, we detected an interaction between activated Fes and ezrin. Consistent with the immunofluorescence data that showed no colocalization between ezrin and Fes, we could not co-immunoprecipitate these two proteins using an anti-Fes antibody. Altogether, these results indicate that only the pool of activated Fes interacts with ezrin and this interaction occurs at the cell–cell contacts. Ezrin recruits activated Fes to the lateral membrane The localization of activated Fes at the lateral membrane and its interaction with ezrin suggested that ezrin might be responsible for the recruitment of Fes to the plasma membrane. To test this hypothesis, we used stable LLC-PK1 cell lines expressing either vesicular stomatitis virus glycoprotein (VSVG)-tagged WT (Crepaldi et al, 1997) or Y477F ezrin. Two clones (F1 and F2) that showed a fourfold increase in Y477F ezrin expression over the endogenous protein were used in all experiments and gave identical results. In these cells, the localization of Y477F ezrin was similar to that of WT ezrin (Figure 3A, right panels). Although no difference in the localization of Fes was observed in cells expressing WT or Y477F ezrin (data not shown), the distribution of activated Fes was dramatically perturbed in cells expressing Y477F ezrin. Indeed, in sparse cells it localized to the focal adhesions, which appeared enlarged (Figure 3A, left panels). In confluent cells expressing Y477F ezrin, activated Fes failed to localize at the site of cell–cell contacts and was observed in numerous focal adhesions that were not observed in cells expressing WT ezrin (Figure 3A, right panels). In line with these data, we did not observe an interaction between activated Fes and Y477F ezrin, unlike what was observed with WT ezrin (Figure 3B). These results demonstrate that the inhibition of ezrin/Fes interaction prevents the recruitment of activated Fes to the cell–cell contacts and leads to its accumulation in focal adhesions. Figure 3.Ezrin recruits Fes to cell–cell contacts. (A) pY713 Fes does not localize at the cell–cell contacts in cells expressing Y477F ezrin. Left panels: double labeling was performed with anti-pY713 Fes and anti-vinculin antibodies on non-confluent cells stably expressing WT or Y477F ezrin. Right panels: double labeling was performed with the anti-pY713 Fes and anti-VSVG antibodies on confluent cells stably expressing WT or Y477F ezrin (bar, 10 μm). (B) The ezrin/Fes interaction is abolished by the point mutation Y477F in ezrin. Immunoprecipitation with the anti-pY713 Fes antibody or non-immune rabbit IgG was performed on lysates from cells expressing VSVG-tagged WT or Y477F ezrin. Immunoblot was performed using the anti-pY713 Fes or anti-VSVG antibodies. An interaction was observed between pY713 Fes and WT ezrin but not with Y477F ezrin. (C) Activated Fes is relocated from cell–cell contacts to focal adhesions in ezrin knocked down cells. Immunostaining was performed with the anti-pY713 Fes antibody. Cells expressing scrambled (Scr) or ezrin-targeting (shEz1) shRNA are also GFP positive (bar, 10 μm). Download figure Download PowerPoint To confirm the requirement of ezrin for the recruitment of Fes to the lateral plasma membrane, we knocked down ezrin by transiently transfecting LLC-PK1 cells with a plasmid encoding short-hairpin RNA-targeting ezrin (shEz) and green fluorescent protein (GFP), which allowed us to monitor transfected cells. Expression of two different shRNA-targeting ezrin gave the same results and led to an efficient knock down (Supplementary Figure S4). In cells transfected with a non-targeting sequence (Scr), activated Fes was observed at cell–cell contacts. In contrast, the localization of activated Fes in ezrin knocked down cells was perturbed, with activated Fes only present at the sites of focal adhesions and nearly absent from cell–cell contacts (Figure 3C). Altogether, these observations demonstrate that the presence of ezrin is crucial for the recruitment of Fes at the cell–cell contacts. The ezrin/Fes interaction is required for cell spreading Since cells expressing Y477F ezrin displayed aberrant focal adhesions, we tested the ability of cells expressing WT or Y477F ezrin cells to spread on two substrates, collagen I and fibronectin. Whereas 89% of the cells expressing WT ezrin were spread 1.5 h after seeding on collagen I-coated surfaces, 81% of cells expressing Y477F ezrin remained round and failed to spread (Figure 4A). These cells showed a disorganization of the actin cytoskeleton and enlarged focal adhesion at the cell periphery that contained activated Fes and vinculin (Figure 4B). The average area of cells expressing Y477F ezrin was about 30% that of cells expressing WT ezrin (Figure 4C). Three hours after seeding, the spreading of cells expressing Y477F ezrin was still impaired as compared with cells expressing WT ezrin (data not shown); however, after 24 h, cell spreading was similar in both cell types, indicating that the lack of tyrosine 477 phosphorylation dramatically delayed the ability of cells to spread on collagen (Figure 4A) or on fibronectin (data not shown). The spreading delay resulting from the expression of Y477F ezrin was observed whether the cells were grown in media with low serum or containing 10% serum indicating that signaling from extracellular matrix requires ezrin/Fes interaction. We next determined whether this spreading defect was primarily due to an adhesion defect. We observed that cells expressing WT or Y477F ezrin adhere to the same extent on collagen I (Figure 4D) or fibronectin (data not shown), indicating that the delayed spreading of cells expressing Y477F ezrin was not due to an adhesion defect. Figure 4.Delayed spreading of LLC-PK1 cells expressing Y477F ezrin. (A) Phase-contrast images of cells expressing WT or Y477F ezrin (clone F1) 1.5 h after plating on type I-collagen (bar, 20 μm). The percentage of spread cells is shown in the bar charts and results from the counting of cells from 10 random fields in three independent experiments for each clone (F1 and F2). Data are means±s.e. Asterisk indicates P<0.05. (B) Cells were stained 1.5 h after seeding for pY713 Fes, vinculin and actin. Cells expressing Y477F ezrin remained round, displayed enlarged and only peripheral focal adhesions, and actin bundles were nearly absent (bar, 10 μm). (C) Comparison of the surface of cells expressing WT or Y477F ezrin. For each clones the area of five representative cells was measured in three independent experiments using Metamorph software. Area of cells expressing WT ezrin was taken as 100%. The data are means±standard error (s.e.). (D) Cells expressing Y477F ezrin do not display an adhesion defect. Bar charts represent the percentage of adherent cells in clones F1 and F2 as compared with cells expressing WT ezrin. Data are means±s.e. Asterisk indicates P<0.05. (E) Expression of the constitutively active form of Fes rescues the spreading defect of cells expressing Y477F ezrin. Cells were stained for pY713 Fes, vinculin and actin (bar, 10 μm). Bar charts represent the percentage of P1 and P2 cells spread 1.5 h after seeding on collagen-coated coverslips as compared with cells transfected with the vector alone (Hygro). The values in the histogram are means±s.e. Asterisk indicates P<0.05. Download figure Download PowerPoint To confirm that the inability of cells expressing Y477F ezrin to spread was due to the lack of ezrin/Fes interaction, we sought to rescue this spreading defect by overexpressing Fes. We stably transfected the F1 clone with the hygromycin vector alone or with a plasmid containing the cDNA encoding WT or a constitutively active form of Fes. This form is obtained by a point mutation in the first coiled-coil motif (L145P) that results in increased Fes tyrosine kinase activity (Cheng et al, 2001). Two different clones for each cell lines, showing a six- to an eightfold increase in Fes level over the endogenous protein, were subsequently used and gave the same results. Interestingly, the spreading defect was not rescued in cells expressing Y477F ezrin and WT Fes (data not shown). In contrast, 44% of the clones (P1 or P2) overexpressing the active form of Fes (L145P Fes) were spread (Figure 4E). This rescue is only partial as compared with cells expressing WT ezrin, but significant as compared with cells expressing Y477F ezrin and the vector alone (Hygro). As shown in Figure 4E, the overexpression of L145P Fes restored the formation of focal adhesions, comparable to that observed in LLC-PK1 cells. The same results were observed when cells were plated on fibronectin (data not shown). Therefore, these experiments indicate that Fes acts downstream of ezrin and that Fes activation is necessary to transmit signals from extracellular matrix receptors for the control of cell spreading. Ezrin and Fes cooperates to promote cell scattering Ezrin has previously been involved in HGF-induced cell scattering (Crepaldi et al, 1997; Orian-Rousseau et al, 2002, 2007). Given the colocalization of ezrin and Fes at the cell–cell contacts, we reasoned that these proteins might cooperate to promote cell scattering. To test this hypothesis, LLC-PK1 cells expressing WT or Y477F ezrin were grown either on glass or on collagen-coated surfaces. Whereas 80% of cells expressing WT ezrin scattered in response to HGF when plated either on glass (data not shown) or on collagen, 80% of cells expressing Y477F ezrin did not scatter in response to HGF either on glass or on collagen and remained as islets (Figure 5A). To rule out that this scattering defect might be due to a global effect of ezrin mutation on HGF signaling, we determined if the activation of the HGF receptor or of the MAPK pathway was perturbed in cells expressing Y477F ezrin. Indeed, it has been shown that ERM proteins are required for the activation of the Ras/MAPK pathway downstream of HGF/Met signaling (Orian-Rousseau et al, 2002, 2007). As shown in Supplementary Figure S5, the activation of HGF receptor and ERK1/2 was similar in cells expressing WT or Y477F ezrin. These results indicate that the lack of ezrin/Fes interaction impairs the HGF-induced cell scattering but does not affect the MAPK pathway. Figure 5.Ezrin/Fes interaction is required to promote HGF-induced cell scattering. (A) Cells expressing Y477F ezrin do not scatter upon HGF treatment. Cells stably expressing WT or Y477F ezrin were treated overnight with HGF and analyzed by phase-contrast microscopy (bar, 10 μm). Quantification was performed on islets of at least four cells in five random fields and in three independent experiments. Bar charts show the percentage of scattered islets. Data are means±s.e. Asterisk indicates P<0.001. (B) Ezrin is recruited to the plasma membrane upon HGF stimulation. Immunofluorescence was performed with the anti-ezrin antibody. Overlay pictures show ezrin (green) and actin (red) stainings. Arrowheads show ezrin localization at the leading edge of the cells (bar, 10 μm). (C) Fluorescence was performed with the anti-pY713 Fes antibody (green) and phalloidin (red). Upon HGF stimulation, activated Fes is recruited to cell–cell contacts and to the leading edge of cells expressing WT ezrin (arrowheads). In cells expressing Y477F ezrin, pY713 Fes remains in focal adhesions (bar, 10 μm). Download figure Download PowerPoint Ezrin-dependent localization of Fes to the lateral membrane is required for cell scattering To better understand how ezrin and Fes cooperate in HGF-induced scattering, we examined the localization of ezrin and activated Fes in cells stimulated or not with HGF. As shown in Figure 5B, HGF stimulation leads to a breakdown of the microvilli and to the recruitment of ezrin to the lateral membrane and to the leading edge of the cells. Similar localization was observed with Y477F ezrin (data not shown). Interestingly, upon HGF stimulation, activated Fes was also recruited to the leading edge and to cell–cell contacts, where it colocalized with ezrin, whereas in non-stimulated cells, Fes was only present in focal adhesions (Figure 5C, upper panels). Strikingly, in cells expressing Y477F ezrin, activated Fes remained in focal adhesions upon HGF treatment and was not recruited to the leading edge or to cell–cell contacts (Figure 5C, lower panels). This indicates that HGF stimulation is not sufficient to localize Fes at the membrane and that the interaction between ezrin and Fes is critical for the recruitment of Fes to the lateral membrane. The activation of Fes at the membrane is required for cell scattering Our results suggested that Fes acts downstream of ezrin to promote HGF-induced cell scattering. We therefore determined if the scattering defect observed in cells expressing Y477F ezrin could be reversed by expressing either WT or L145P Fes in cells expressing Y477F ezrin. Cells coexpressing Y477F ezrin and WT Fes were unable to scatter upon HGF treatment either on glass or on collagen (data not shown). However, 50% of the cells expressing Y477F ezrin and L145P Fes scattered in response to HGF only when seeded on a collagen-coated surface (Figure 6A) but not on glass (data not shown). These results demonstrate that Fes must be activated to trigger cell scattering. Moreover, since the reversion is only observed when cells are grown on collagen, it indicates that scattering also requires signals from extracellular matrix. Figure 6.The recruitment and the activation of Fes at the membrane are required for cell scattering. (A) Activated Fes rescues the scattering defect of cells expressing Y477F ezrin. Phase-contrast images of cells seeded on collagen-coated coverslips and treated or not with HGF. Bar charts represent the percentage of P1 and P2 cell islets scattered as compared with cells expressing Y477F ezrin and the vector alone (Hygro). The values in the histogram are means±s.e. and asterisk indicates P<0.001. (B) L145P Fes is recruited t
Referência(s)