Localization of nectin-free afadin at the leading edge and its involvement in directional cell movement induced by platelet-derived growth factor

Afadin is a protein that binds nectin and actin filaments (F-actin) and thus connects nectins to the actin cytoskeleton (Mandai et al., 1997; Takahashi et al., 1999). Afadin comprises multiple domains: two Ras-binding (RA) domains, a dilute (DIL) domain, a PDZ (post synaptic density protein, Drosophila disc large tumor suppressor, and zonula occludens-1 protein) domain, three proline-rich domains, and an F-actin-binding (FAB) domain in this order from the N-terminus (Takai et al., 2003). Nectins are Ca²⁺-independent immunoglobulin-like cell-cell adhesion molecules and consist of four members: nectin-1, nectin-2, nectin-3 and nectin-4 (Ogita and Takai, 2008; Takai and Nakanishi, 2003). Our previous studies have revealed the roles and modes of actions of nectins and afadin in the formation of adherens junctions (AJs) (Ogita and Takai, 2008; Sakisaka et al., 2007; Takai et al., 2008a). The formation of AJs begins with trans-interaction of nectins between neighboring cells. This trans-interaction induces the activation of the small G proteins Rap1, Cdc42 and Rac through the activation of c-Src. Activated Cdc42 and Rac bind to the GTPase activator protein IQGAP1 (also known as p195) and these molecules cooperatively reorganize the actin cytoskeleton to recruit the E-cadherin–catenin complex to nectin-based cell-cell contact sites in epithelial cells. At this stage, E-cadherin molecules have only weak adhesion activity and do not trans-interact with each other. However, afadin binds activated Rap1 and then associates with p120^ctn, which prevents the endocytosis of non-trans-interacting E-cadherin and enhances the adhesion activity of E-cadherin to facilitate its trans-interaction, eventually establishing AJs. In addition, afadin functions as a connector between the cadherin-catenin and nectin-afadin systems at AJs by interacting with ponsin, LIM domain only 7 (LMO7), and afadin DIL domain-interacting protein (ADIP), or by directly binding to α-catenin (Asada et al., 2003; Mandai et al., 1999; Ooshio et al., 2004; Pokutta et al., 2002; Tachibana et al., 2000).

In addition to its role in the formation of AJs, afadin is involved in cell polarization (together with Par complex proteins) and the formation of tight junctions (TJs) by assembly of TJ components, such as claudin, at the apical site of AJs (Komura et al., 2008; Ooshio et al., 2007; Yamada et al., 2006). The nectin-afadin complex also plays a key role in cell survival by supporting the platelet-derived growth factor (PDGF)-induced activation of the phosphatidylinositol 3-kinase (PI3K)-Akt pathway (Kanzaki et al., 2008). Moreover, afadin is indispensable for morphogenesis in embryonic development because the lack of afadin in mice causes embryonic lethality owing to disorganized cell-cell junctions, improper differentiation of cells and impaired cell migration during gastrulation (Ikeda et al., 1999; Zhadanov et al., 1999). Therefore, afadin has multiple important functions in the formation of cell-cell junctions, as shown by in vitro and in vivo experiments.

Afadin has been shown to be present in the cytosolic fraction as well as at cell-cell adhesion sites (Mandai et al., 1997). We recently demonstrated that, in sparsely cultured cells that do not contact with other cells, afadin prevents PDGF-induced hyperactivation of PDGF receptor and its downstream signaling molecules Ras and ERK by increasing the phosphatase activity of SHP-2 (Nakata et al., 2007). The Ras-ERK signaling pathway is involved in the regulation of cytoskeletal dynamics and directional cell movement (Chernyavsky et al., 2005; Ho et al., 2001; Huang et al., 2004; Matsubayashi et al., 2004). However, the role of afadin in the cytosolic fraction is not fully understood. Thus, in this study, we investigated the role of afadin in moving cells and found that afadin was localized at the leading edge of moving cells and was involved in directional cell movement. To promote directional cell movement, nectin-like molecule-5 (Necl-5; also known as poliovirus receptor), a cell-adhesion molecule structurally similar to nectins, forms a complex with αvβ3 integrin and PDGF receptor, and enhances the signaling pathways needed for the formation of leading-edge structures (Amano et al., 2008; Minami et al., 2007; Nagamatsu et al., 2008; Takahashi et al., 2008). On the basis of these results, we also examined the association of afadin with these molecules to elucidate the molecular mechanism by which afadin is involved in PDGF-induced directional cell movement.

We first examined the effect of afadin on cell movement in response to PDGF stimulation using wild-type and afadin-knockdown NIH3T3 cells. The expression level of afadin in both cell types is shown in Fig. 1A. Wild-type and afadin-knockdown NIH3T3 cells were sparsely plated on μ-Slide VI Flow dishes pre-coated with vitronectin, an extracellular matrix protein that binds to αvβ3 integrin (Schvartz et al., 1999), and were directionally stimulated with PDGF. Wild-type NIH3T3 cells became polarized with the well-spreading leading edge toward the higher concentration of PDGF, whereas afadin-knockdown NIH3T3 cells showed elongated shapes and had a small leading edge that was randomly directed and was independent of the direction of the higher concentration of PDGF (Fig. 1B).

These results led us to assume that afadin is involved in directional cell movement. To examine this assumption, we performed a wound-healing assay by scratching the confluent monolayer of wild-type and afadin-knockdown NIH3T3 cells in the presence or absence of PDGF. In the absence of PDGF, the wound was still open at 8 hours after scratching in both wild-type and afadin-knockdown NIH3T3 cells, and the degree of the wound closure was similar between both types of cells (supplementary material Fig. S1A). In the presence of PDGF, however, the wound made by the scratch gradually closed in wild-type NIH3T3 cells (Fig. 1C). In this process, cells efficiently formed protrusions, and the immunofluorescence signal for afadin was observed at the leading edge of these protrusions as well as at the cell-cell adhesion sites. By contrast, the wound did not close in the afadin-knockdown cell monolayer. Formation of the leading edge in afadin-knockdown cells was less obvious than that in wild-type cells and the organization of actin stress fibers appeared more prominently in afadin-knockdown than wild-type cells due to the increased activation of RhoA in afadin-knockdown cells (Miyata et al., 2009). The velocity of cell proliferation and the degree of apoptosis in wild-type and afadin-knockdown cells during the experimental period (8 hours) were not significantly different, as shown in our previous report (Nakata et al., 2007) and in our unpublished results. We further analyzed the alignment of the Golgi complex in cells at the wound edge because the reorientation of the Golgi complex in moving cells was reported to correlate with directional cell movement (Kupfer et al., 1982; Nobes and Hall, 1999). The ratio of the Golgi complex facing the wound was lower in afadin-knockdown than wild-type cells (Fig. 1D). In addition, the Boyden chamber assay showed that afadin-knockdown NIH3T3 cells were less responsive to PDGF than wild-type cells when PDGF was added into only the bottom chamber (Fig. 1E). However, the number of wild-type NIH3T3 cells crossing the membrane was similar to that of afadin-knockdown cells when PDGF was added into both upper and lower sides of chambers, suggesting that afadin might not contribute to random chemokinesis. Furthermore, the area of cell movement of afadin-knockdown NIH3T3 cells was similar to that of wild-type cells, as assessed by the phagokinetic track motility assay that visualizes the tracks of moving cells (Fig. 1F), assuming that cell movement is not impaired by knockdown of afadin. Taken together, these results indicate that afadin regulates the directionality of cell movement in moving cells in response to PDGF.

This hypothesis was further confirmed by time-lapse microscopy. Wild-type NIH3T3 cells formed a large leading edge in the direction of the higher concentration of PDGF and this leading-edge formation was persistent, keeping the direction of cell movement constant (supplementary material Movie 1). By contrast, formation of the leading edge in afadin-knockdown NIH3T3 cells was less stable and was unrelated to the direction of PDGF stimulation, resulting in impaired directional cell movement (supplementary material Movie 2). Tracking of cell movement also revealed that the PDGF-induced directional cell movement was markedly perturbed by knockdown of afadin in NIH3T3 cells (Fig. 2).

To prove that the impairment of leading-edge formation and directional cell movement in afadin-knockdown NIH3T3 cells is dependent on the reduction of afadin itself, we used afadin-knockdown cells stably re-expressing green fluorescent protein (GFP)-tagged full-length afadin (GFP-afadin), which was resistant to RNAi against afadin. The expression level of GFP-afadin was similar to that of endogenous afadin in wild-type NIH3T3 cells (Fig. 3A). Formation of the leading edge induced by stimulation with PDGF was rescued by re-expression of GFP-afadin in afadin-knockdown NIH3T3 cells (Fig. 3B). Similarly to endogenous afadin, GFP-afadin was recruited to the leading edge. The accumulation of PDGF receptor at the leading edge was also recovered by re-expression of GFP-afadin in afadin-knockdown NIH3T3 cells (Fig. 3C). Consistent with these results, directional cell movement assessed by the wound-healing assay, the reorientation of the Golgi complex and the Boyden chamber assay were restored by re-expression of GFP-afadin in afadin-knockdown NIH3T3 cells (Fig. 3D-F).

We have reported that Necl-5, αvβ3 integrin, and PDGF receptor form a complex at the leading edge in moving cells, and that this complex is important for cell movement and proliferation (Amano et al., 2008; Ikeda et al., 2004; Kakunaga et al., 2004; Minami et al., 2007; Nagamatsu et al., 2008; Takahashi et al., 2008). Based on this observation, we examined the association of these molecules with afadin in NIH3T3 cells stimulated with PDGF. The immunofluorescence signals for αvβ3 integrin, PDGF receptor, and Necl-5 were concentrated at the leading edge formed in the direction of the higher concentration of PDGF, as previously described, and colocalized with that for afadin (Fig. 4A). In GFP-expressing NIH3T3 cells, the signal for N-cadherin was detected on the entire plasma membrane, but was not highly concentrated at the leading edge, as previously reported (Fujito et al., 2005). The ectopically expressed GFP signal was not observed on the cell periphery, providing the notion that the accumulation and colocalization of αvβ3 integrin, PDGF receptor and Necl-5 with afadin at the leading edge are not merely dependent on the thickening of the cell front. This colocalization was further confirmed by the bead-cell contact assay. When microbeads coated with vitronectin were incubated with NIH3T3 cells, the signal for αvβ3 integrin was preferentially observed at the bead-cell contact sites and the signal for afadin also accumulated there, although these signals were not observed at the contact sites between cells and the control Concanavalin A (ConA)-coated beads (Fig. 4B). Furthermore, the signals for PDGF receptor and Necl-5 were also accumulated at the contact sites between cells and vitronectin-coated beads; however, the signal for nectin-1 was not accumulated there (Fig. 4C). This suggests that afadin, which localizes at the bead-cell contact sites, associates with αvβ3 integrin, PDGF receptor and Necl-5, but not with nectin. These results are also consistent with our previous report that the localization and function of nectins and Necls are different, even though their molecular structures are similar (Takai et al., 2008b). The wound-healing assay further confirmed that afadin at the leading edge was free from nectin. Just after scratching the wild-type NIH3T3 cell monolayer, the leading edge was not formed and afadin and nectin-3 colocalized at the cell-cell adhesion sites, whereas at 2 hours after scratching, afadin, but not nectin-3, was accumulated at the leading edge (supplementary material Fig. S1B). At this time, the immunofluorescence signals for afadin and nectin-3 became weaker at the cell-cell adhesion sites behind the wound because of the loose cell-cell adhesion, but their colocalization was still observed at several cell-cell adhesion sites. These results indicate the different intracellular behavior between afadin and nectin.

We then examined whether afadin is necessary for the accumulation of αvβ3 integrin, PDGF receptor and Necl-5 at the leading edge. When afadin was knocked down in NIH3T3 cells, the signal for PDGF receptor was hardly detected at the leading edge in response to PDGF (Fig. 4D). Similar results were obtained for the signals for αvβ3 integrin and Necl-5 (data not shown). These results indicate that afadin plays a role in recruiting and tethering αvβ3 integrin, PDGF receptor and Necl-5 to the leading edge of moving cells to promote the PDGF-induced directional cell movement.

Our recent study showed that Rap1, which is activated in response to PDGF, is recruited to the leading edge and is crucial for PDGF-induced formation of leading-edge structures (Takahashi et al., 2008). We, and another research group, have also reported that activated Rap1 binds afadin (Hoshino et al., 2005; Linnemann et al., 1999). To confirm this, we conducted immunoprecipitation and in vitro binding assays. The association of afadin with Rap1 was dependent on the activation of Rap1 because a constitutively active mutant of Rap1 (V12Rap1), but not Rap1 inactivated by Rap1GAP, could be co-immunoprecipitated with afadin (Fig. 5A). Rap1GAP is a GTPase-activating protein (GAP) for Rap1. In the in vitro binding assay, we used pure recombinant proteins of afadin fused to maltose-binding protein (MBP-afadin) and V12Rap1 tagged with six repeats of His (His-V12Rap1). When His-V12Rap1 was incubated with MBP-afadin immobilized on amylose resin beads, the direct binding of His-V12Rap1 to MBP-afadin was dose-dependent (Fig. 5B). As negative controls, His-V12Rap1 did not bind to MBP, nor did His-wild-type Rap1 (Rap1WT), on which GTPγS was not loaded, bind to MBP-afadin. In addition, the Ras association (RA) domain of afadin as well as full-length afadin were co-immunoprecipitated with V12Rap1, but afadinΔRA, which lacks the RA domain, was incapable of being co-immunoprecipitated with V12Rap1 (Fig. 5C). This indicates that the binding of afadin to the active form of Rap1 is mediated by the RA domain of afadin.

On the basis of these results, we examined whether the binding of Rap1 to afadin is required for its localization at the leading edge, the formation of the leading edge, and directional cell movement in response to PDGF. When GFP-afadinΔRA was re-expressed in afadin-knockdown NIH3T3 cells, this re-expression failed to rescue the formation of the leading edge that was observed in wild-type NIH3T3 cells expressing GFP and in afadin-knockdown NIH3T3 cells re-expressing GFP-afadin (Fig. 3B). GFP-afadinΔRA did not accumulate at the cell edge, indicating a role of Rap1 in the recruitment of afadin to the leading edge. Similarly, PDGF receptor did not accumulate at the cell edge (Fig. 3C). As expected, re-expression of GFP-afadinΔRA in afadin-knockdown NIH3T3 cells did not restore directional cell movement (Fig. 3D-F). These results suggest that afadin promotes the formation of the leading edge and PDGF-induced directional cell movement by binding to Rap1.

The activation of Rap1 has been reported to be mediated by an adaptor protein Crk, which directly interacts with phosphorylated PDGF receptor, and by the Rap1 guanine exchange factor C3G, which is activated downstream of Crk (Ichiba et al., 1997; Matsumoto et al., 2000). Therefore, we hypothesized that accumulation of the Rap1-afadin complex is involved in the association of Rap1 with PDGF receptor through Crk and C3G, and performed the immunoprecipitation assay to test this. V12Rap1 was co-immunoprecipitated with PDGF receptor in the presence of Crk and C3G (Fig. 5D). However, this co-immunoprecipitation was perturbed by the dominant-negative mutant of Crk. This perturbation was partial but statistically significant. Furthermore, such co-immunoprecipitation was observed at the endogenous protein level, and was inhibited by Rap1GAP (Fig. 5E). The results from these immunoprecipitation assays and the immunofluorescence microscopy shown in Fig. 3 indicate that afadin associates with PDGF receptor through active Rap1 for its accumulation at the leading edge.

As described above, the RA domain of afadin is essential for the binding of afadin to Rap1. Therefore, we examined whether the RA domain itself inhibits the interaction between afadin and Rap1. When GFP-afadin and GFP-afadin-RA were simultaneously transfected with FLAG-V12Rap1 in HEK293 cells, GFP-afadin-RA was preferentially co-immunoprecipitated with FLAG-V12Rap1 and inhibited the interaction between GFP-afadin and FLAG-V12Rap1 (Fig. 6A), indicating an inhibitory effect of the RA domain of afadin on the afadin-Rap1 interaction. We next examined whether blockade of the afadin-Rap1 interaction by overexpression of the RA domain affects the formation of the leading edge and directional cell movement. After stimulation with PDGF, NIH3T3 cells expressing GFP formed the leading edge, whereas NIH3T3 cells expressing GFP-afadin-RA hardly formed the leading edge, similar to the situation observed in afadin-knockdown NIH3T3 cells (Fig. 6B). Consistent with this, the PDGF-induced directional cell movement determined by the Boyden chamber assay was impaired in NIH3T3 cells expressing GFP-afadin-RA compared with cells expressing GFP (Fig. 6C). These results suggest that the interaction between afadin and activated Rap1 is involved in formation of the leading edge, which is important for directional cell movement.

In the last set of experiments, we investigated how the Rap1-afadin complex regulates directional cell movement. We recently demonstrated that, in response to PDGF, SHP-2 binds to the tyrosine-phosphorylated form of afadin, and that formation of the SHP-2–afadin complex is necessary for the interaction between SHP-2 and the tyrosine-phosphorylated form of PDGF receptor and for the proper regulation of the Ras-MEK-ERK pathway (Nakata et al., 2007). Appropriate activation of this signaling pathway controls the formation of leading-edge structures through reorganization of the actin cytoskeleton, resulting in enhanced directional cell movement (Chernyavsky et al., 2005; Ho et al., 2001; Huang et al., 2004). However, it remains elusive how the SHP-2–afadin complex, which crucially regulates the PDGF-induced Ras-ERK signaling, is recruited to PDGF receptor and how this complex is associated with Rap1.

To explore these issues, we examined the intracellular localization of afadin, SHP-2 and PDGF receptor in NIH3T3 cells in response to PDGF. As expected, SHP-2 was recruited to the leading edge and colocalized with PDGF receptor and afadin in wild-type NIH3T3 cells, whereas this colocalization was absent in afadin-knockdown NIH3T3 cells (Fig. 7A). When GFP-afadin was re-expressed in afadin-knockdown NIH3T3 cells, the cells formed the leading edge, and the colocalization of SHP-2 with PDGF receptor and afadin at the leading edge was rescued (Fig. 7B). By contrast, when GFP-afadinΔRA was re-expressed instead of GFP-afadin, the cells failed to form the leading edge and SHP-2 did not colocalize with PDGF receptor. These results suggest that the interaction of afadin with Rap1 is necessary for the recruitment of SHP-2 to PDGF receptor and that the associations between SHP-2, afadin and Rap1 are crucial for PDGF-induced formation of the leading edge.

We further examined whether the associations between SHP-2, afadin and Rap1 are involved in the proper regulation of ERK activation. As previously reported, ERK is hyperactivated in afadin-knockdown cells compared with wild-type NIH3T3 cells (Nakata et al., 2007). This hyperactivation was canceled by re-expression of GFP-afadin in afadin-knockdown NIH3T3 cells, but not by re-expression of GFP-afadinΔRA (Fig. 7C). In addition, the association between PDGF receptor and Necl-5, which plays a crucial role in the formation of the leading edge, was reduced by expression of GFP-afadinΔRA compared with that of GFP-afadin in HEK293 cells (Fig. 7D). Taken together, these results indicate that afadin, which is recruited to the leading edge by binding to Rap1, supports the association of PDGF receptor with Necl-5 at the leading edge and that afadin, together with Rap1, SHP-2 and ERK, positively regulates the formation of the leading edge, thus facilitating PDGF-induced directional cell movement.

Moving cells form protrusions, such as lamellipodia and filopodia, ruffles, focal complexes and focal adhesions, by reorganization of the actin cytoskeleton at the leading edge. This reorganization of the actin cytoskeleton is performed by several F-actin-binding proteins, including N-WASP, WAVEs, IQGAP1 and cortactin (Briggs and Sacks, 2003; Daly, 2004; Takai et al., 2001; Takenawa and Miki, 2001). In addition, many of these F-actin-binding proteins function downstream of Cdc42 and Rac, the activation of which is induced by growth factors in an integrin-dependent manner (Arthur et al., 2004; DeMali et al., 2003; Ridley, 2001). We recently demonstrated that PDGF-activated Rap1 subsequently induces the activation of Rac and its downstream molecules and is crucial for the formation of the leading edge and for cell movement (Takahashi et al., 2008). A pool of afadin that localizes in the cytosol, but not at the cell-cell adhesion, is also shown to be involved in PDGF-induced formation of the leading edge by regulating the activation of ERK through SHP-2 (Nakata et al., 2007). However, the role of afadin in cell movement was unclear: does afadin affect the mobility of cells itself or does afadin only affect the directionality of cell movement? It was also unclear how afadin, Rap1 and SHP-2 are involved in cell movement. Here, we show that afadin localizes at the leading edge of moving NIH3T3 cells where Necl-5, αvβ3 integrin and PDGF receptor (all of which are important for cell movement) also localize, and that afadin preferentially controls the directionality of cell movement but does not affect the mobility of cells.

We also revealed the molecular mechanisms involved in PDGF-induced and afadin-dependent formation of the leading edge. It has been shown that Necl-5 physically interacts with αvβ3 integrin and induces the clustering of αvβ3 integrin and PDGF receptor for the formation of leading edge structures including lamellipodia, peripheral ruffles, focal complexes and focal adhesions, eventually enhancing cell movement (Ikeda et al., 2004; Minami et al., 2007). These actions of Necl-5 are dependent on PDGF-induced activation of Rac and the binding of αvβ3 integrin to its specific extracellular matrix proteins, such as vitronectin and fibronectin (Minami et al., 2007). The extracellular region of Necl-5 directly interacts with the extracellular region of αvβ3 integrin and is necessary for directional cell movement, but not for random cell movement, whereas the cytoplasmic region of Necl-5 is necessary for both directional and random cell movement (Ikeda et al., 2004). In addition, previous studies have clarified that PDGF receptor physically interacts with αvβ3 integrin (Schneller et al., 1997; Woodard et al., 1998), suggesting the formation of a ternary complex with PDGF receptor, αvβ3 integrin and Necl-5. This ternary complex is formed in response to PDGF at the leading edge, where Rap1 and Rac are activated (Takahashi et al., 2008). Considering our recent and present results, on one hand, activated Rap1 causes reorganization of the actin cytoskeleton together with Rac and other F-actin-binding proteins (Miyata et al., 2009). On the other hand, activated Rap1 recruits the afadin–SHP-2 complex to PDGF receptor and this complex regulates PDGF-induced activation of ERK and formation of the leading edge (Fig. 8). These signaling networks in turn enhance the clustering of the Necl-5–αvβ3-integrin–PDGF receptor complex in a feedback amplification manner and stabilize the formation of the well-spreading leading edge, resulting in the facilitation of directional cell movement. Although there is no direct evidence that ERK promotes the formation of the ternary complex with Necl-5, αvβ3 integrin and PDGF receptor at the leading edge, ERK participates in reorganization of the actin cytoskeleton by regulating the activity of myosin light chain (MLC) kinase and MLC phosphorylation (Klemke et al., 1997), and contributes to integrin activation and focal adhesion turnover through the phosphorylation of focal adhesion kinase and paxillin (Chou et al., 2003; Hunger-Glaser et al., 2003; Liu et al., 2002). Thus, ERK might indirectly support and stabilize the formation of the ternary complex, but its precise role and mechanism of action remain to be clarified.

Our previous study demonstrated that PDGF-induced and PDGF-receptor-mediated activation of Rap1 through the Crk-C3G complex and the subsequent activation of Rac are essential for the formation of the leading edge and for directional cell movement (Takahashi et al., 2008). In addition to these observations, this present study has revealed that afadin, SHP-2 and ERK are also necessary for formation of the leading edge and for directional cell movement. Considering this and previous studies, it is clear that several types of molecules participate in and regulate directional cell movement. Although we have extensively investigated the mechanism of directional cell movement, the underlying mechanism seems to be quite complicated and is not yet fully understood. Therefore, further studies are required to better understand the underlying mechanisms.

In contrast to the results of our present study, one study has shown that knockdown of afadin enhances cell movement and accelerates wound healing in epithelial MCF10A cells (Lorger and Moelling, 2006). In that report, when afadin is knocked down, the intercellular adhesion mediated by E-cadherin is impaired because the association of E-cadherin with p120^ctn and F-actin becomes weaker. This impaired intercellular adhesion loosens the cell-cell connection and increases the directionality of cell movement in MCF10A cells. Similar to the results of Lorger and Moelling, the signal for N-cadherin at the cell-cell contact sites is reduced in afadin-knockdown NIH3T3 cells (data not shown). However, the directionality of cell movement is attenuated in these NIH3T3 cells. Although the cell types used in the experiments differ, the reason for this discrepancy remains to be elucidated, and further studies are therefore necessary.

There is a report that p120^ctn regulates RhoA activation through p190RhoGAP (Wildenberg et al., 2006). Both afadin and p120^ctn are peripheral membrane proteins and directly interact with nectins and cadherins, respectively, and they can also interact with each other (Hoshino et al., 2005). The knockdown of p120^ctn in NIH3T3 cells induces the enhancement of RhoA activation and formation of actin stress fibers and suppresses the velocity of the wound closure: these phenotypes are similar to those observed in afadin-knockdown NIH3T3 cells, as shown in this work and in our recent study (Miyata et al., 2009). Wildenberg and colleagues propose that p120^ctn regulates RhoA activation through p190RhoGAP, which interacts with p120^ctn (Wildenberg et al., 2006), whereas we have recently demonstrated that afadin regulates the RhoA activation through Rap1 and ARAP1 (Miyata et al., 2009). ARAP1 is a Rap1-dependent RhoGAP. Afadin prevents the inactivation of Rap1, which is activated by PDGF at the leading edge, and promotes the binding of activated Rap1 to ARAP1. This results in enhancement of the ARAP1 function in modulating the activation state of RhoA. Although the detailed molecular mechanisms are different, it might be of interest that afadin and p120^ctn, which are both classified as peripheral membrane proteins, have similar effects on cellular phenotypes.

Afadin located at the leading edge is unlikely to bind to nectin, because nectin was not concentrated at the leading edge or at the contact sites between the vitronectin-coated beads and the cells where afadin was concentrated (Fig. 4C). However, afadin is concentrated and binds nectin at AJs after the initial formation of cell-cell contact and the bound afadin supports the nectin-induced formation of AJs (Asakura et al., 1999; Takahashi et al., 1999). On the basis of these results, we conclude that afadin is localized in at least two pools in two different states: free afadin at the leading edge and nectin-bound afadin at the cell-cell contact site. Thus, afadin plays key roles, not only in the formation of cell-cell junctions between contacting cells, but also in directional cell movement in moving cells.

When normal cells become confluent, they cease movement and proliferation, and start to form cell-cell junctions (Abercrombie, 1970; Zegers et al., 2003). This phenomenon is referred to as `contact inhibition' of cell movement and proliferation. Contact inhibition is also important for mesenchymal-epithelial transition (MET), which is observed during organogenesis in embryonic development. The mechanism underlying contact inhibition is complicated, but Necl-5 and nectin are at least involved in this mechanism (Fujito et al., 2005). In moving cells, Necl-5 predominantly localizes at the leading edge and enhances cell movement (Ikeda et al., 2004). However, Necl-5 is downregulated from the cell surface by its trans-interaction with nectin-3 at the cell-cell contact sites, resulting in the reduction in cell movement and proliferation by inhibiting the signaling mediated by PDGF receptor and αvβ3 integrin (Fujito et al., 2005). We have shown here that afadin at the leading edge of moving cells regulates cell movement in addition to its important role in nectin-induced formation of cell-cell junctions. During the formation of cell-cell junctions, afadin might also help Necl-5 to recruit nectin-3 as well as help nectin-3 to recruit nectin-1 for their trans-interaction at the cell-cell adhesion sites. Therefore, afadin might play a crucial role in MET through these functions.

NIH3T3 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% calf serum. Afadin-knockdown NIH3T3 cells were generated as previously described (Kanzaki et al., 2008). Briefly, pEB-H1-afadin vector containing a short hairpin RNA (shRNA) sequence against afadin was transfected into NIH3T3 cells, followed by selection with 500 μg/ml G418 (Nacalai Tesque). Afadin-knockdown NIH3T3 cells stably expressing GFP were generated by additionally transfecting pCAGpuro-EGFP-N3 into afadin-knockdown NIH3T3 cells, followed by selection with both 500 μg/ml G418 and 10 μg/ml puromycin (Sigma-Aldrich). NIH3T3 cells stably expressing GFP were generated by transfecting pCAGpuro-EGFP-N3 into wild-type NIH3T3 cells, followed by selection with 10 μg/ml puromycin. HEK293 cells were cultured in DMEM supplemented with 10% fetal calf serum. For DNA transfection, Lipofectamine 2000 (Invitrogen) and an Amaxa Nucleofector kit (Amaxa) were used.

For rescue experiments, expression vectors for RNAi-resistant GFP-tagged rat full-length afadin (amino acids 1-1829; pMSCVpuro-GFP-afadin^*) and GFP-afadinΔRA (amino acids 351-1829; pMSCVpuro-GFP-afadinΔRA) were created. Afadin-knockdown NIH3T3 cells stably expressing GFP-afadin and GFP-afadinΔRA were generated by retrovirus-mediated introduction and selection with both 500 μg/ml G418 and 10 μg/ml puromycin, as previously described (Kakunaga et al., 2004; Kanzaki et al., 2008). AfadinΔRA cDNA did not contain the target sequence for afadin shRNA.

V12Rap1B, in which a glycine residue at amino acid 12 of bovine Rap1B was replaced by a valine residue (V), was a constitutively active mutant of Rap1 and prepared using the QuickChange site-directed mutagenesis kit (Stratagene). The cDNA of Rap1GAP was a gift from Patrick Casey (Duke University, Durham, NC). Expression vectors for FLAG-Rap1WT, FLAG-V12Rap1, GFP-V12Rap1, Myc-Rap1GAP, FLAG-CrkI and Myc-C3G were prepared as previously described (Fukuyama et al., 2005). An expression vector for a FLAG-tagged dominant-negative mutant of CrkI (pIRM21-FLAG-CrkI-W169L) was kindly provided by Michiyuki Matsuda (Kyoto University, Kyoto, Japan). Expression vectors for GFP-afadin and GFP-afadinΔRA were prepared as described (Nakata et al., 2007). Expression vectors for FLAG–Necl-5 and HA-PDGF receptor were prepared as described (Amano et al., 2008). To construct an expression vector for GFP-afadin-RA, the rat afadin cDNA fragment that corresponds to amino acids 1-350 was subcloned into pEGFP-C1 (Clontech).

A rabbit anti-afadin polyclonal antibody (pAb) and a mouse anti-afadin monoclonal Ab (mAb) were prepared as described (Sakisaka et al., 1999). A rat mAb against the extracellular region of Necl-5 (mAb-i, 1A8-8) was prepared as previously described (Ikeda et al., 2003). Hybridoma cells expressing a mouse anti-Myc mAb (9E10) were obtained from American Type Culture Collection and prepared as described (Kodama et al., 2000). Hamster anti-integrin αv and β3 mAbs (H9.2B8 and 2C9.G2, respectively) were purchased from BD Biosciences. The following mouse mAbs were purchased from commercial sources: anti-GM130 (Pharmingen), anti-N-cadherin (Pharmingen), anti-SHP-2 (Pharmingen), anti-actin (Chemicon), anti-phospho-ERK1/2 (Cell Signaling Technology), anti-FLAG (Sigma-Aldrich), anti-HA (Babco) and anti-GFP (Clontech). Rabbit anti-PDGF receptor β (Y92) and anti-GAPDH (14C10) mAbs were purchased from Abcam and Cell Signaling Technology, respectively. The following rabbit pAbs were purchased from commercial sources: anti-PDGF receptor β (Santa Cruz Biotechnology and Upstate), anti-Rap1GAP (Santa Cruz Biotechnology), anti-ERK1/2 (Cell Signaling Technology), and anti-GFP (MBL). Fatty-acid-free BSA, trypsin inhibitor and 4,6-diamidino-2-phenylindole (DAPI) were purchased from Sigma-Aldrich. Horseradish-peroxidase-conjugated and fluorophore-conjugated secondary Abs were purchased from GE Healthcare and Chemicon, respectively. Pre-immune IgG was purchased from Jackson ImmunoResearch Laboratories. Vitronectin was purified from human plasma (Kohjinbio) as previously described (Yatohgo et al., 1988).

To generate a concentration gradient of PDGF, a μ-Slide VI Flow (uncoated; Ibidi) was used as previously described (Minami et al., 2007). Briefly, cells were seeded at a density of 5×10³ cells/cm² on the vitronectin-coated μ-Slide VI Flow, cultured for 18 hours, and serum-starved with DMEM containing 0.5% BSA for 1 hour. The concentration gradient of 30 ng/ml PDGF was made according to the manufacturer's protocol. After 30 minutes, cells were fixed with acetone-methanol (1:1), incubated with 1% BSA in PBS, and then incubated with 20% BlockAce in PBS, followed by immunofluorescence microscopy. Tracking of cell movement was analyzed by the Chemotaxis and Migration Tool (version 1) (Ibidi), which was plugged into the ImageJ software (NIH, Bethesda, MD). The directionality of cell movement was defined as the endpoint y-value divided by the accumulated distance. Data are expressed as means ± s.e.m. of 13 randomly selected wild-type or afadin-knockdown NIH3T3 cells.

Polarization of the Golgi complex in the direction of movement was assessed as previously described (Nobes and Hall, 1999). Briefly, correctly polarized cells were defined as those that could reorient their Golgi complex to the 120° sector of the cells facing the wound front. Random polarization would mean that 33% of the cells had a Golgi complex in any sector. Wider wounds (∼200 μm wide) were made in monolayers of cells, and the cells fixed for 2 hours after wounding. In each experiment, at least 100 cells within the row of cells adjacent to the wound were examined. Data are expressed as means ± s.e.m. of three independent experiments. Paired Student's t-test was performed for statistical analysis.

The Boyden chamber assay was performed as previously described (Fujito et al., 2005). Falcon cell-culture inserts (8.0-μm pores; Becton Dickinson) were coated with 3 μg/ml vitronectin at 37°C for 1 hour. The inserts were then blocked with 1% BSA at 37°C for 30 minutes. The cells, which had been serum-starved in DMEM supplemented with 0.5% fatty-acid-free BSA for 1 hour, were detached with 0.05% trypsin and 0.53 mM EDTA and then treated with a trypsin inhibitor. Cells were re-suspended in DMEM supplemented with 0.5% fatty-acid-free BSA and seeded at a density of 2.5×10⁴ cells per insert. The cells were incubated at 37°C for 12 hours in the presence or absence of 30 ng/ml PDGF. After incubation, the inserts were washed with PBS and the cells fixed using 3.7% formaldehyde and subsequently stained with DAPI (Sigma-Aldrich). After removing the cells on the upper part of the filter using cotton sticks, the number of stained cells that crossed the filter was counted in five randomly chosen fields per filter using a microscope. Data are expressed as means ± s.e.m. of three independent experiments.

After being washed with ice-cold PBS, cells were harvested using pre-warmed Laemmli buffer containing 1 mM Na₃VO₄, 10 mM NaF and a phosphatase inhibitor cocktail (Sigma-Aldrich), boiled for 5 minutes, and sonicated three times for 10 seconds with 20-second cooling periods. The protein concentrations of the samples were determined using an RC DC protein assay kit (Bio-Rad) with BSA as a reference protein. The samples were separated by SDS-PAGE, and this was followed by western blotting with the indicated Abs. The band intensity measured by densitometry was analyzed using ImageJ software. For the immunoprecipitation assay, cells expressing various combinations of the indicated molecules were lysed with Buffer A (20 mM Tris-HCl at pH 7.5, 150 mM NaCl, 1 mM CaCl₂, 1 mM MgCl₂, 1% NP-40, 1 mM Na₃VO₄, 1 mM APMSF, 3 μg/ml leupeptin and 5 μg/ml aprotinin). The cell lysates were centrifuged at 20,000 g for 15 minutes, and the supernatant was then incubated with the anti-FLAG mAb at 4°C for 2 hours, followed by incubation with protein-G-Sepharose beads at 4°C for 2 hours. After the beads had been extensively washed with Buffer A, bound proteins were eluted from the beads by boiling in SDS sample buffer for 5 minutes and were subjected to SDS-PAGE, followed by western blotting with the indicated Abs.

Recombinant maltose-binding protein (MBP) and MBP-fused full-length afadin (MBP-afadin) were prepared as described (Hoshino et al., 2005; Nakata et al., 2007; Sakisaka et al., 2008). cDNA of wild-type bovine Rap1B or V12Rap1B was inserted into the pQE-30 vector to express His-tagged Rap1WT (His-Rap1WT) or V12Rap1 (His-V12Rap1) in E. coli. Purified His-V12Rap1 was preloaded with GTPγS as previously described (Yamada et al., 2005). To examine the affinities of Rap1WT (GTPγS-unloaded) and V12Rap1 (GTPγS-loaded) for afadin, Rap1WT or His-V12Rap1-GTPγS were incubated with MBP (20 pmol) or MBP-afadin (20 pmol) immobilized on 20 μl of amylose resin beads in 400 μl of Buffer B (25 mM Tris-HCl at pH 7.5, 150 mM NaCl, 5 μM GTPγS, 48 mM MgCl₂, 8 mM EDTA, 1 mM DTT and 0.08% CHAPS) at 4°C for 2 hours. After extensively washing the beads with Buffer B, the bound proteins were eluted by boiling the beads in the SDS sample buffer. The samples were then subjected to SDS-PAGE and stained with Coomassie brilliant blue.

Assays for phagokinetic track motility, wound healing and bead-cell contact were performed as previously described (Ikeda et al., 2004). In the wound-healing assay, the width of the wound space at 0, 2 and 8 hours after scratching the confluent cell monolayer was measured at at least five different points in each experiment to quantify the extent of wound closure (Goldfinger et al., 1999). Data are expressed as means ± s.e.m. of three independent experiments.