Rac Activation upon Cell-Cell Contact Formation Is Dependent on Signaling from the Epidermal Growth Factor Receptor*

Cadherins are transmembrane receptors that mediate cell-cell adhesion. They play an essential role in embryonic development and maintenance of tissue architecture. The Rho family small GTPases regulate actin cytoskeletal dynamics in different cell types. The function of two family members, Rho and Rac, is required for the stability of cadherins at cell-cell contacts. Consistent with the published data we have found that Rac is activated upon induction of intercellular adhesion in epithelial cells. This activation is dependent on functional cadherins (Nakagawa, M., Fukata, M., Yamaga, M., Itoh, N., and Kaibuchi, K. (2001) J. Cell Sci. 114, 1829–1838; Noren, N. K., Niessen, C. M., Gumbiner, B. M., and Burridge, K. (2001) J. Biol. Chem. 276, 3305–3308). Here we show for the first time that clustering of cadherins using antibody-coated beads is sufficient to promote Rac activation. In the presence of Latrunculin B, Rac can be partially activated by antibody-clustered cadherins. These results suggest that actin polymerization is not required for initial Rac activation. Contrary to what has been described before, phosphatidylinositol 3-kinases are not involved in Rac activation following cell-cell adhesion in keratinocytes. Interestingly, inhibition of epidermal growth factor receptor signaling efficiently blocks the increased Rac-GTP levels observed after contact formation. We conclude that cadherin-dependent adhesion can activate Rac via epidermal growth factor receptor signaling.

Cadherins are transmembrane receptors that mediate calcium-dependent cell-cell adhesion. They play an essential role in embryonic development and maintenance of tissue architecture in adults (1). Cadherins interact in an anti-parallel fashion with the same type of receptors on adjacent cells (known as homophilic binding), thereby mediating formation of cell-cell contacts. Intercellular adhesion is strengthened by the clustering of cadherin receptors at junctions and the association between cadherin complexes and the actin cytoskeleton (2,3).
Cadherin-dependent adhesion is subject to regulation by cytoplasmic proteins, and recent work has focused on modulation by the Rho family of small GTPases. Members of this family of GTP-binding proteins regulate actin cytoskeletal dynamics and focal complex formation in different cell types (4). The function of two family members, Rho and Rac, is required for the stability of cadherins at cell-cell contacts (5). The exact mechanisms by which Rho and Rac operate are unclear, but we and others have shown that Rac plays a role in actin recruitment to junctions (6) and to clustered cadherin receptors (7,8).
Rho and Rac function in signal transduction cascades downstream of a variety of cell surface receptors. For example, Rho mediates stress fiber formation in response to lysophosphatidic acid (LPA) 1 stimulation, whereas Rac is required for growth factor-induced membrane ruffling (9). More recently it has been directly demonstrated that ligand binding to cell surface receptors can activate Rho and Rac (10,11). In addition, the regulation of the activity of Rho, Rac, and Cdc42 by integrin engagement has been determined and shown to be dependent on time, matrix composition, and matrix concentration (12)(13)(14)(15).
Lately, it has become apparent that cadherin-mediated adhesion can trigger intracellular signaling events including activation of phosphatidylinositol 3-kinases (PI3-kinases) and the extracellular signal-regulated kinase (ERK)/mitogen-activated protein kinase pathway (16 -18). In addition, the epidermal growth factor (EGF) receptor has been coimmunoprecipitated with the cadherin adhesion complex from epithelial cells and is activated upon cell-cell contact formation (18,19).
Furthermore, recent work has demonstrated that cell-cell contact formation can activate Rac in Madine-Darby canine kidney and Chinese hamster ovary cells transfected with C-cadherin (20,21). Spreading of E-cadherin-expressing Chinese hamster ovary cells on immobilized E-cadherin ectodomain also induces Rac activation (22). In addition, levels of GTPbound Rac are higher in VE-cadherin-expressing endothelial cells than in VE-cadherin-null cells (23). Although these studies showed that Rac activation is dependent on cadherin function, they did not establish the functional significance of Rac activation upon cell-cell contact formation, nor did they address the question of whether cadherin clustering is sufficient to enhance GTP loading on Rac. Although two studies found that PI3-kinase activity was required for full Rac activation, inhibition of PI3-kinase function did not reduce Rac activity to basal levels (20,22), suggesting that other signaling pathways may be involved in Rac activation downstream of cadherinmediated adhesion.
In light of these results, we were interested to determine whether cadherin-mediated adhesion could also activate Rac in keratinocytes. Normal keratinocytes provide the ideal system to look at Rac activation on induction of cell-cell adhesion. They can be grown to confluence in the absence of cadherin-dependent contacts (in low calcium medium), and then cell-cell adhesion can be induced by addition of calcium ions (known as the calcium switch) (24). Because cells are confluent and touching each other, they do not need to migrate to form junctions, and contact formation occurs rapidly. Using this system, we investigated the requirements for Rac activation following cell-cell contact formation in terms of (a) clustering of cadherin receptors, (b) actin polymerization, and (c) PI3-kinase and EGF receptor signaling. Our results shed light on the mechanisms via which cadherin receptors are functionally coupled to Rac activation.

EXPERIMENTAL PROCEDURES
Cell Culture and Microinjection-Normal human keratinocytes (strains Kb and Sa, passages 3-7) were cultured as described previously (25). Cells were grown to confluence in low calcium medium (25). To induce intercellular adhesion calcium chloride was added to 2 mM. Microinjection experiments and recombinant protein production were performed as described previously (7).
For antibody blocking experiments, cells were incubated with 5 g/ml HECD-1 and 2 g/ml NCC-CAD-299 for 30 min at 37°C prior to addition of calcium. As a control, cells were preincubated with 7 g/ml rabbit anti-mouse IgG. For clustering experiments, keratinocytes grown in low calcium medium were incubated for 15 min on ice with 5 g/ml HECD-1 and 2 g/ml NCC-CAD-299. Anti-mouse IgG was added to 35 g/ml, and cells were incubated for 30 min at 37°C.
For bead experiments, 15 m latex beads (Polysciences) were coated with HECD-1 described as previously (7). Latrunculin B (Calbiochem) was titrated to 0.3 M, which prevented actin cytoskeleton remodeling in keratinocytes without causing cellular retraction. Cells were treated with Latrunculin B or Me 2 SO control for 10 min and incubated with HECD-1-coated beads (2.4 ϫ 10 7 beads per dish) for 10 min before lysis. In inhibition experiments, cells were incubated with 30 M LY294002 (Sigma), 316 nM tyrphostin AG1478 (Calbiochem) (18), 35 M PD98508 (Calbiochem), or Me 2 SO vehicle for 30 min prior to stimulation with 2 mM calcium for 5 min. In EGF stimulation experiments, cells were incubated with LY294002 or Me 2 SO vehicle as described above before stimulation with 10 nM EGF for 10 min.

Activated Rac Rescues the Effect of Dominant Negative
Rac on Junctions-A number of studies have shown that a dominant negative version of Rac (N17Rac) inhibits the localization of cadherin receptors at sites of cell-cell contact (6,7). Dominant negative mutants are presumed to function by sequestering upstream activators for the small GTPases, the guanine nucleotide exchange factors (GEFs) (reviewed in Ref. 33). However, there appears to be more than 46 GEFs for Rho family GTPases, many of which are rather promiscuous in exchange activity (reviewed in Ref. 34). Therefore, it is possible that N17Rac blocks cadherin-mediated adhesion by inhibition of a Rho GTPase other than Rac.
To determine whether Rac does indeed play a role in regulation of cadherin localization at intercellular junctions, we investigated whether an activated version of Rac (L61Rac) can rescue the inhibitory effect of N17Rac on cell-cell contacts. Normal human keratinocytes cultured in the absence of intercellular contacts were injected with recombinant dominant negative Rac protein, and cell-cell adhesion was induced for 1 h. Cells were fixed and stained for E-cadherin (Fig. 1, A and B). As reported previously (7), N17Rac inhibited localization of cadherin receptors at boundaries between injected cells (Fig. 1,  A and B). Microinjection of L61Rac alone had no effect on cadherin localization at junctions after incubation for 1 h (Fig.  1, C and D). Interestingly, co-injection of L61Rac and N17Rac completely rescued the inhibitory effect of N17Rac on cadherinmediated adhesion (Fig. 1, E and F). This effect is specific for L61Rac because L61Cdc42 cannot restore junctional cadherin staining in N17Rac-injected cells. 2 Together these results provide strong evidence that N17Rac blocks localization of cadherin receptors at junctions by inhibiting Rac function. Thus, Rac plays an important role in the regulation of cadherinmediated adhesion.

Rac Is Activated upon Cell-Cell Contact Formation in Keratinocytes-Because
Rac activity is required for the stability of cadherin receptors at junctions, is Rac activated upon cell-cell contact formation in keratinocytes? To evaluate Rac activation, we performed pull-down assays utilizing the Cdc42/Rac interactive binding (CRIB) domain of the Rac effector p21-activated kinase (PAK) (32). We confirmed the specificity of the assay under our conditions by loading keratinocyte lysates with GTP␥S or GDP and assessing Rac activity (Fig. 2a).
To investigate whether cell-cell adhesion can activate the small GTPase Rac, we grew keratinocytes in low calcium medium and induced cell-cell contacts for various time periods before measuring Rac activation. We found that endogenous Rac was activated 2-4-fold within 5 min of addition of calcium, and this activation was sustained up to 120 min (Fig. 2, b and c).

Rac Activation Is Dependent on Functional Cadherin
Receptors-It is possible that the up-regulation of Rac described above was caused by stimulation of calcium-dependent intracellular signaling pathways rather than induction of cadherindependent adhesion. To distinguish between these two possibilities, we inhibited cadherin function during the calcium switch. Keratinocytes express at least two members of the cadherin family (E-and P-cadherin) (26). Cells were preincubated with anti-E-and anti-P-cadherin antibodies at concentrations known to block cadherin function in keratinocytes (24). To confirm that the antibodies blocked cadherin function, we demonstrated that, in the presence of the inhibitory antibodies, no cadherin clustering was found at intercellular boundaries (Fig. 2d). Incubation with anti-cadherin antibodies prevented Rac up-regulation upon addition of calcium ions, whereas control IgG did not (Fig. 2e). Thus, the Rac activation is dependent on functional cadherin receptors, and addition of calcium ions per se is not sufficient to trigger Rac activation.
Clustering of Cadherins Activates Rac-These results demonstrated that cell-cell adhesion mediated by cadherins could activate Rac in keratinocytes. Because junction formation is accompanied by the clustering of cadherin receptors (3), we investigated whether clustering of cadherins with antibodies was sufficient to activate Rac. Keratinocytes grown in low calcium medium were incubated with anti-E-and anti-P-cad-herin antibodies (mouse monoclonals). Clustering of the receptors was then induced by addition of anti-mouse IgG, before assaying for Rac activity. Under these conditions, we could observe clusters of cadherins on the cell surface by immunofluorescence (Fig. 3A). We found that clustering induced a modest increase in Rac activity relative to samples where the anticadherin antibodies were omitted (Fig. 3B). Quantification of the relative Rac activation in several experiments revealed that treatment of cells with anti-cadherin antibodies alone only induced a 1.2-fold increase in Rac activation relative to cells treated with anti-mouse IgG alone. Clustering of surface cadherins with two layers of antibodies induced a larger increase in Rac activation (1.7-fold; Fig. 3, B and F).
However, the degree of Rac activation after antibody clustering is lower than the activation observed following cell-cell contact formation (2-4-fold). We hypothesized that this weak Rac activation may be due to the small number of cadherin molecules per cluster formed. To overcome this problem, we incubated cells for 10 min in low calcium medium with beads coated with anti-E-cadherin antibodies. Immunofluorescence of cells treated with these beads revealed binding of anti-E-cadherin beads to the cell surface and recruitment of ␣-catenin and F-actin to the beads (Fig. 3C, arrowheads). In contrast, few beads coated with BSA alone bound to the cell surface, and those that did recruited little ␣-catenin or F-actin (Fig. 3C,   FIG. 3. Clustering of cadherin receptors induces Rac activation. A, cadherins were clustered with a mixture of mouse anti-E-and anti-P-cadherin antibodies and rabbit anti-mouse IgG. Cells were fixed and stained with fluorescein isothiocyanate-anti-rabbit secondary antibody. Bar, 32 M. B, cells in low calcium were incubated with (ϩ) or without (Ϫ) a mixture of anti-E-and anti-P-cadherin antibodies for 15 min on ice. The antibodies were clustered with anti-mouse IgG (ϩ) or left untreated (Ϫ) for 30 min at 37°C, and Rac activity was measured. Results are representative of seven experiments. C, keratinocytes were incubated with beads coated with anti-E-cadherin antibodies, anti-␣ 3 ␤ 1 -integrin antibodies, or BSA. Coated beads were added, and cells were incubated for an additional 10 min. Cells were then fixed and stained for ␣-catenin and F-actin. In some experiments, keratinocytes were treated with Latrunculin B or Me 2 SO vehicle for 10 min at 37°C, prior to incubation with beads. Arrows indicate beads that do not recruit F-actin; arrowheads indicate beads that do recruit F-actin and ␣-catenin. Asterisks show beads with actin recruitment but no ␣-catenin staining. Bar, 18.6 m. D, cells were treated as in C, but after incubation with the beads cells were lysed and Rac activation was assessed. Results are representative of three experiments. E, cells were incubated with anti-␣ 3 ␤ 1 -integrin beads or BSA beads for 10 min and lysed, and Rac activation was assessed. Results are representative of two experiments. F and G, quantification of antibody clustering and bead experiments. Activated Rac levels were calculated as a percentage of total Rac levels in the lysates. Rac activation is expressed relative to the controls; mouse IgG alone (F) or Me 2 SO (DMSO) vehicle with BSA beads (G), which were arbitrarily set as 1, are shown. arrows, top panels). We found that cells incubated with anti-E-cadherin beads showed increased Rac activation in comparison with cells treated with BSA beads (Fig. 3D). Quantification of the results of three experiments revealed that anti-Ecadherin beads enhanced Rac activation by a factor of 2.1 compared with BSA beads (Fig. 3G), suggesting that clustering of cadherins is sufficient to activate Rac.
To determine whether new actin polymerization is required for Rac activation by clustered cadherin molecules, we treated cells with Latrunculin B prior to receptor clustering. Latrunculin B inhibits actin polymerization by sequestering actin monomers (36,37). Treatment of cells with Latrunculin B for 10 min before adding anti-E-cadherin beads still allowed binding of beads to the cell surface and recruitment of ␣-catenin but blocked actin recruitment (Fig. 3C, arrows, bottom panels). Upon assaying for Rac activation, we found that Latrunculin B treatment per se caused an increase in Rac activation. Incubation with E-cadherin beads in the presence of Latrunculin B induced a further 1.4-fold increase in Rac activation (Fig. 3, D  and G). Thus, Rac activation can occur in the absence of de novo actin polymerization, but full induction of Rac activity does require formation of new actin filaments.
However, the use of beads to artificially cluster cadherin receptors may resemble the phagocytotic process. Rac activity is necessary for Fc␥-mediated phagocytosis (38). Therefore, the possibility exists that anti-E-cadherin beads may induce Rac activation by triggering early events in phagocytosis. As a control to confirm that this was not the case, we treated cells with beads coated with antibodies against another cell surface receptor, ␣ 3 ␤ 1 -integrin. As expected, these beads showed little staining for ␣-catenin but did recruit F-actin (Fig. 3C, asterisks). However, anti-␣ 3 ␤ 1 -integrin beads did not induce increased Rac activity relative to BSA control beads (Fig. 3E). Thus, the clustering of cadherin receptors can specifically activate Rac.
Rac Activation Is Dependent on EGF Receptor Signaling but Not PI3-kinases-We next explored which signaling pathways operate downstream of cadherin-dependent contact formation to induce Rac activity. We investigated three different classes of signaling molecule: PI3-kinases, the EGF receptor, and ERK1/2.
We made use of the compounds LY294002, tyrphostin AG1478, and PD98059, specific inhibitors of PI3-kinases, the EGF receptor, and mitogen-activated kinase or ERK kinase-1 (MEK1), respectively (39 -41). We confirmed that the inhibitors were functional under our experimental conditions by probing lysates for Akt phosphorylated on serine 473 (for LY294002; Fig. 4A) or activated ERK1/2 (AG1478 and PD98059; Fig. 4, B and C). As expected, phosphorylation and activation of these molecules were blocked.
Inhibition of keratinocytes with LY294002 increased Rac activity (compare with low calcium sample; Fig. 4D) but did not inhibit Rac activation induced by cell-cell adhesion. We therefore concluded that PI3-kinases do not play a major role in Rac activation downstream of cadherin adhesion in keratinocytes.
Similarly to LY294002, treatment of keratinocytes with tyrphostin AG1478 increased Rac activity (Fig. 4D). However, the drug dramatically decreased Rac activation upon induction of cell-cell adhesion. Therefore we concluded that EGF receptor function participates in cadherin-dependent Rac stimulation (Fig. 4D). Consistent with this conclusion, calcium-induced cell-cell adhesion induces EGF receptor activation in normal keratinocytes 3 as has been observed in HaCat cells (18).
Rac activation in these experiments was quantified relative to the control low calcium cultures treated with Me 2 SO ( Fig.  4G; Ϫcalcium, DMSO). However, there was some variability in the degree of Rac activation induced by drug treatment per se. For this reason Rac activity was also calculated relative to the control low calcium cultures treated with each inhibitor and was arbitrarily set at 1 (Fig. 4I; Ϫcalcium).
The inhibitors Latrunculin B, AG1478, or LY294002 can all activate Rac in keratinocytes. To investigate whether this Rac activation is a nonspecific consequence of treatment with different inhibitors, we incubated cells with the MEK1 inhibitor PD98059. Treatment with PD98059 compound did not increase the basal level of Rac activity in low calcium medium, nor did it inhibit activation of Rac induced by cell-cell contact formation (Fig. 4E). Quantification of these experiments is shown in Fig. 4, H and I. Thus, blocking signaling pathways in general does not lead to Rac activation. In addition, these results suggested that the ERK/mitogen-activated kinase cascade is not involved in induction of GTP loading on Rac downstream of cadherin-dependent adhesion.
EGF-stimulated activation of Rac is PI3-kinase-dependent in fibroblasts (42). Given our results, we were interested to determine whether a similar pathway exists in keratinocytes. Cells grown in the absence of cell-cell contacts were incubated with the PI3-kinase inhibitor LY294002 and stimulated with EGF for 10 min. Whereas in untreated cells EGF stimulation induces a strong Rac activation, this is blocked by inhibition of PI3-kinase (Fig. 4F). Thus, as in fibroblasts, EGF-induced Rac activation is PI3-kinase-dependent in keratinocytes. DISCUSSION Our results demonstrate for the first time that activated Rac can rescue the inhibitory effect of a dominant negative Rac mutant on the localization of cadherin receptors at cell-cell contact sites. This effect is specific for activated Rac and cannot be mimicked by activated Cdc42, 2 thus providing strong evidence that Rac plays a role in the regulation of intercellular junctions in keratinocytes. Similarly, our previous work has shown that active Rac cannot compensate for the loss of Rho function in cadherin-mediated adhesion and epithelial morphology (7,35). Taken together, our results indicate that Rho, Rac, and Cdc42 have distinct functions during junction formation in keratinocytes.
We have shown that induction of cell-cell adhesion in keratinocytes specifically causes an up-regulation of Rac activity. This increased Rac activation is blocked by inhibition of cadherin function, consistent with published data (20,21). Most importantly, here we have demonstrated that clustering of cadherins using antibody-coated beads per se is indeed sufficient to induce GTP loading on Rac. This response is specific to clustering of cadherins and is not achieved upon clustering of ␣ 3 ␤ 1 -integrins in keratinocytes.
We have shown previously that Rac plays a role in actin recruitment to clustered E-cadherin receptors (7). Now we demonstrate for the first time that Rac activation can occur independently of de novo actin polymerization. However, full induction of GTP loading on Rac requires formation of new actin filaments. Approximately 50% of cadherin receptors are associated with actin filaments in the absence of cell-cell adhesion (43). Thus, clustering of this pool of receptors would provide a framework for new actin polymerization. Subsequent adhesion-dependent Rac activation further enhances actin recruitment and polymerization, thereby stabilizing cadherin receptors at junctions.
In the course of these experiments, we observed that treatment of keratinocytes with two drugs that affect the actin cytoskeleton, Latrunculin B and cytochalasin D, induces Rac activation ( Fig. 3D and data not shown). Both drugs inhibit actin polymerization, but they do so by different mechanisms (36,37,44). Interestingly, cytochalasin D treatment has also been shown to activate Rho (10). In addition, drugs that affect the microtubule network have been shown to influence Rho and Rac activity (10,45). Thus, the activity of Rho family GTPases appears to be acutely sensitive to the polymerization state of microfilaments and microtubules. It is conceivable that regulatory proteins such as GEFs and GTPase activating proteins (GAPs) localize at cytoskeletal filaments and that upon disruption of the network they may be released/modified to up-regulate Rho and Rac activity. 4 Similarly to what is observed with disruption of actin filaments, inhibition of PI3-kinases (LY294002) and the EGF receptor (AG1478) increased the basal levels of Rac activation. This was not a general consequence of treating cells with in-hibitors of signaling molecules because the PD98059 inhibitor, which blocks MEK1 activation, did not enhance Rac activity. Although these results are surprising, the mechanisms via which these drugs may interfere with activation of Rho proteins are beyond the scope of this manuscript.
How can clustering of cadherin receptors promote Rac activation? To address this question we investigated which signaling pathways might be involved. In contrast to the work of Nakagawa et al. (20), we have found that Rac activation following cell-cell contact formation is not dependent on PI3kinase activity. Our results are consistent with another report in which initial Rac activation following adhesion of E-cadherin-expressing Chinese hamster ovary cells to E-cadherin ectodomain is PI3-kinase-independent (22). Another situation in which Rac activation occurs independently of PI3-kinase activity is in chemoattractant-stimulated neutrophils (46).
In contrast, we have demonstrated that EGF receptor signaling participates in the stimulation of Rac activity in kera- 4  , and MEK1 activation (C and E, PD98059) were as follows. Cells in low calcium medium were treated for 30 min with each inhibitor (ϩ) or Me 2 SO vehicle (Ϫ) and incubated in the presence (ϩ) or absence of calcium (Ϫ) for 5 min before lysis. Lysates were probed for phosphorylated Akt and total Akt (A) or diphosphorylated ERK1/2 and total ERK-2 (B and C). Rac activation was also evaluated following inhibition of PI3-kinase (D, LY294002), EGF receptor (D, AG1478), or MEK1 activation (E, PD98059). F, EGF-stimulated Rac activation is dependent on PI3-kinase. Cells in low calcium medium were incubated with LY294002 (ϩ) or Me 2 SO vehicle (Ϫ) as described above and were stimulated for 10 min with EGF (ϩ) or left unstimulated (Ϫ). Rac activation was then analyzed. G-I, quantification of Rac activation in experiments inhibiting PI3-kinase (LY294002), EGF receptor (AG1478), or MEK1 (PD98059). Activated Rac levels were calculated as a percentage of total Rac levels in the lysates and normalized in two different ways. G and H, Rac activation is expressed relative to the Me 2 SO control (DMSO) without calcium stimulation, which was arbitrarily set as 1. Note the up-regulation of Rac-GTP levels by drug treatment (LY294002 and AG1478) in the absence of cell-cell adhesion. Results represent mean Ϯ S.E. of four experiments (LY294002 and AG1478 treatment) or two experiments (PD98059 treatment). I, activated Rac levels are expressed relative to the low calcium control treated with each drug, which was arbitrarily set as 1. The same set of values used in G and H was recalculated to show the changes in the levels of Rac-GTP following addition of calcium ions. An increase in Rac activation levels is observed after cell-cell adhesion in samples treated with LY294002 and PD98059 but not with AG1478. tinocytes upon junction formation. Inhibition of EGF receptor signaling blocks Rac activation both initially (Fig. 4D) and at later time points after induction of cell-cell adhesion. 3 Consistent with this, in normal keratinocytes cell-cell adhesion activates the EGF receptor, and phosphorylated EGF receptor is recruited to cadherin clustered by beads. 3 Preliminary work suggests that EGF receptor signaling is also required for increased GTP loading on Rac induced by clustering of cadherin molecules on beads. 3 Our data are in agreement with the proposal that cadherin clustering leads to EGF receptor recruitment and activation at cell-cell contact sites (18,19).
Similar to other cell systems (e.g. fibroblasts), EGF-stimulated Rac activation is PI3-kinase-dependent in keratinocytes. This suggests that EGF receptor-induced Rac activation proceeds by two different mechanisms, depending on how the EGF receptor is activated (i.e. by clustering with cadherins or by ligand binding). In support of this theory, there is some evidence in the literature suggesting that the consequences of EGF receptor activation may depend on both the ligand and the localization of the receptor in the cell (apical or basolateral domains) (47)(48)(49).
Based on the fact that Rac function stabilizes cadherin receptors at junctions (7) and our present results, we expected that the EGF receptor inhibitor (AG1478) would interfere with establishment of cell-cell contacts. To our surprise, the inhibition of EGF receptor signaling does not block localization of cadherins at junctions upon addition of calcium ions. 3 Similar observations were made in Madine-Darby canine kidney cells; although PI3-kinase inhibition diminished Rac activation, it did not block junction formation or recruitment of GFP-Rac to junctions (20).
One explanation for these observations is that because the pull-down assay measures global Rac activation, it is possible that the results do not reflect localized changes in Rac activity at cell-cell contacts. Alternatively, EGF receptor inhibition may cause an increased rate of GTP cycling on Rac, so no net increase in GTP-bound Rac is observed (50).
Two further considerations should be taken into account here: (a) Rac activity is required for the stabilization of cell-cell contacts rather than cadherin clustering per se 5 and (b) blocking Rac function only inhibits 50% of actin recruitment to clustered cadherin receptors (7), suggesting that alternative pathways also contribute to actin polymerization at junctions. One possibility is Cdc42, which can promote actin accumulation at cell-cell contact sites and is activated upon intercellular junction formation in MCF-7 cells (16,51). Other GTPaseindependent pathways may also be involved, including calcium signaling and Ena/VASP (52,53). These pathways may compensate for the reduction in Rac activation caused by EGF receptor inhibition. Nevertheless, basal levels of activated Rac are still necessary for stable cell-cell contacts because complete inhibition of Rac activity destabilizes cadherin-mediated adhesion ( Fig. 1) (7).
If EGF receptor-dependent Rac activation does not play a role in cadherin stabilization at junctions, which other signaling events would Rac activation trigger? Our recent results suggest that when EGF receptor signaling is inhibited, there is no delay in actin recruitment to junctions or formation of other adhesive structures such as desmosomes. 3 We speculate that the EGF receptor-dependent Rac activation may play a role in other cadherin-dependent cellular processes, e.g. regulation of growth and/or differentiation of epithelia. Clearly more experiments are required to resolve this issue.
The link between EGF receptor activation and the increase in GTP loading in Rac is unclear presently. It is likely to involve activation of a Rac-GEF (promoting GTP loading on Rac) or inhibition of a Rac-GAP (thereby preventing GTP hydrolysis). Both mechanisms would result in an increase in the levels of Rac-GTP. Work is in progress to identify the GEFs or GAPs responsible for Rac activation during junction formation in keratinocytes. Evidence is accumulating for the existence of a cross-talk between cadherin-mediated adhesion and growth factor signaling. For example, non-adhesive cadherin receptors associate with insulin receptor substrate 1 (54), whereas VE-cadherin interacts with VEGFR-2 in endothelial cells (55). The communication between cadherins and growth factor receptors may be analogous to the cross-talk observed between integrins and growth factor receptors, which regulates cell survival and proliferation (56). This cross-talk may also contribute to the differences in cadherin signaling in different cells types, depending on the growth factor receptor that is able to associate with the cadherin complexes.
Here we demonstrate in epithelia the first example of crosstalk between cadherins, growth factors, and small GTPases. In keratinocytes, cadherin receptor clustering is sufficient to activate Rac, and cadherin-dependent Rac activation requires EGF receptor signaling. The functional implications of the EGF receptor-dependent Rac activation are unclear at the moment. Potential roles include the regulation of cellular morphogenesis, epithelial differentiation, and growth, events in which cadherin adhesion, Rac, and EGF receptor signaling have been implicated (1,5,57). Further work is needed to determine which of these processes are modulated by the cross-talk between cadherin receptors and signaling molecules.