Cdc42 and Rac1 Regulate the Interaction of IQGAP1 with β-Catenin*

IQGAP1, a target of Cdc42 and Rac1 small GTPases, directly interacts with β-catenin and negatively regulates E-cadherin-mediated cell-cell adhesion by dissociating α-catenin from the cadherin-catenin complex in vivo (Kuroda, S., Fukata, M., Nakagawa, M., Fujii, K., Nakamura, T., Ookubo, T., Izawa, I., Nagase, T., Nomura, N., Tani, H., Shoji, I., Matsuura, Y., Yonehara, S., and Kaibuchi, K. (1998) Science 281, 832–835). Here we investigated how Cdc42 and Rac1 regulate the IQGAP1 function. IQGAP1 interacted with the amino-terminal region (amino acids 1–183) of β-catenin, which contains the α-catenin-binding domain. IQGAP1 dissociated α-catenin from the β-catenin-α-catenin complex in a dose-dependent manner in vitro. Guanosine 5′-(3-O-thio)triphosphate (GTPγS)·glutathioneS-transferase (GST)-Cdc42 and GTPγS·GST-Rac1 inhibited the binding of IQGAP1 to β-catenin in a dose-dependent manner in vitro, whereas neither GDP·GST-Cdc42, GDP·GST-Rac1, nor GTPγS·GST-RhoA did. The coexpression of dominant active Cdc42 with IQGAP1 suppressed the dissociation of α-catenin from the cadherin-catenin complex induced by the overexpression of IQGAP1 in L cells expressing E-cadherin (EL cells). Consistent with this, the overexpression of either dominant negative Cdc42 or Rac1 resulted in the reduction of E-cadherin-mediated cell adhesive activity in EL cells. These results indicate that Cdc42 and Rac1 negatively regulate the IQGAP1 function by inhibiting the interaction of IQGAP1 with β-catenin, leading to stabilization of the cadherin-catenin complex.

is a well known calcium-dependent cell-cell adhesion molecule. The cytoplasmic domain of cadherin binds to ␤-catenin, and this complex is linked to the actin cytoskeleton by ␣-catenin. It is well known that this linkage is essential for the cadherinmediated cell-cell adhesion (for a review see Ref. 4). Therefore, it is likely that dynamic rearrangement of the cadherin-catenin complex is crucial for the above phenomena. However, little is known about the regulatory mechanism underlying the rearrangement of the cadherin-catenin complex.
We have recently found that IQGAP1 regulates the cadherinmediated cell-cell adhesion (10). IQGAP1 interacts with ␤-catenin and E-cadherin both in vitro and in vivo. The overexpression of IQGAP1 induces the dissociation of ␣-catenin from the cadherin-catenin complex and results in reduction of the Ecadherin-mediated cell adhesive activity in EL cells but not in L cells expressing E-cadherin mutant in which the cytoplasmic domain is deleted and replaced by the carboxyl-terminal half of ␣-catenin (nE␣CL cells) (26). The inhibitory effect of IQGAP1 on the E-cadherin-mediated cell-cell adhesion is counteracted by the coexpression of dominant active Cdc42 (Cdc42 Val12 ). Thus, IQGAP1 together with Cdc42 and Rac1 appear to regulate the cell-cell adhesion through the rearrangement of the cadherin-catenin complex. However, how Cdc42 and Rac1 regulate the IQGAP1 function remains to be clarified.
In the present study, we investigated how Cdc42 and Rac1 regulate the IQGAP1 function. We found that IQGAP1 bound to the amino terminus of ␤-catenin and thereby dissociated ␣-catenin from the ␤-catenin-␣-catenin complex. GTP␥S⅐GST-Cdc42 and GTP␥S⅐GST-Rac1, which interacted with IQGAP1, inhibited the binding of IQGAP1 to ␤-catenin in vitro. We also found that dominant active Cdc42 suppressed the dissociation of ␣-catenin from the cadherin-catenin complex induced by IQGAP1 in vivo. These results indicate that Cdc42 and Rac1 negatively regulate the IQGAP1 function by inhibiting the interaction of IQGAP1 with ␤-catenin, resulting in the stabilization of the cadherin-catenin complex.

EXPERIMENTAL PROCEDURES
Materials and Chemicals-EL cells and nE␣CL cells were kindly provided by Drs. A. Nagafuchi and S. Tsukita (Kyoto University, Kyoto, Japan) and were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum containing 0.1 mg/ml G418 (26). The mouse neuropilin-1-expressing L cells, designated as LAP86 cells, were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (27). The cDNAs encoding mouse ␣-catenin and mouse ␤-catenin were kindly provided by Drs. A. Nagafuchi and S. Tsukita. Anti-E-cadherin and anti-␤-catenin antibodies were purchased from Transduction Laboratories (Lexington, KY). Anti-␣-catenin antibody was purchased from Sigma. Anti-IQGAP1 and anti-maltose-binding protein (MBP) polyclonal antibodies were generated against GST-IQGAP1-(aa 1-216) and MBP, respectively (10). Dithiobis(succinimidyl propionate) (DSP) was obtained from Pierce. Cytochalasin D was purchased from Sigma. All materials used in the nucleic acid study were purchased from Takara Shuzo Co., Ltd. (Kyoto, Japan). Other materials and chemicals were obtained from commercial sources.
Preparation of Recombinant Proteins-The expression and purification of various GST and MBP fusion proteins were done as described (10). GST-IQGAP1 was purified from overexpressing Spodoptera frugiperda insect cells as described (28).
Interaction of MBP-␤-Catenin Mutants with GST-IQGAP1-The interactions of various MBP-␤-catenin mutants with GST-IQGAP1 or GST-Rho GDI were examined as described (10). Briefly, indicated MBP-␤-catenin mutants were mixed with affinity beads coated with GST, GST-IQGAP1, or GST-Rho GDI in Buffer A (20 mM Tris/HCl at pH 7.4, 1 mM EDTA, 1 mM dithiothreitol, 500 mM NaCl, 0.1% (w/v) Triton X-100, 0.1% (w/v) CHAPS, 10 M (p-amidinophenyl)-methanesulfonyl fluoride, 10 g/ml leupeptin). The beads were then washed with Buffer A, and the bound proteins were coeluted with GST fusion proteins by the addition of Buffer A containing 10 mM reduced glutathione. The eluates were subjected to SDS-PAGE, followed by immunoblotting with anti-MBP antibody.
Inhibition of ␤-Catenin Binding of ␣-Catenin by IQGAP1-Various concentrations of GST-␣-catenin were mixed with the indicated amounts of GST-IQGAP1. The mixture was incubated with MBP-␤catenin-N-2-(aa 1-183) (5 pmol)-immobilized beads for 1 h at 4°C. After the beads had been washed, the immunocomplex was subjected to SDS-PAGE, followed by immunoblotting with anti-␣-catenin and anti-IQGAP1 antibodies.
Immunological Collection of Transiently Transfected Cells-EL cells, nE␣CL cells, or LAP86 cells (3 ϫ 10 6 /10-cm dish) were seeded. The indicated plasmids together with pME18-tAic2A (29,30), an interleukin (IL)-3␤1 receptor lacking a cytoplasmic domain, were transfected into EL cells, nE␣CL cells, or LAP86 cells using LipofectAMINE (Life Technologies, Inc.). After a 20-h incubation, the cells were scraped from 5 to 10 dishes with silicon-rubber policemen and transferred to a plastic dish precoated with anti-Aic2 antibody (HC) (29,30). The immunologically collected cells (5 ϫ 10 5 cells) were then seeded onto a 48-well culture dish and incubated for 24 h. The recombinant small GTPases and IQGAP1 were expressed in 80 -90% of the collected cells and at a level 2-3-fold higher than that of endogenous proteins as described previously (10). Cell Dissociation Assay-By using immunologically collected cells (5 ϫ 10 5 cells) as described above, the cell dissociation assay with EL cells or nE␣CL cells was performed after treatment of the cells with trypsin in the presence of either Ca 2ϩ (TC treatment) or EGTA (TE treatment) as described (10,26). The total particle numbers after TC or TE treatment were designated as N TC and N TE , respectively. The cell dissociation assay with LAP86 cells was performed after treatment of the cells with EDTA either in the presence (TEd treatment) or absence (Ed treatment) of trypsin as descried (27,31). The total particle numbers after Ed or TEd treatment were designated as N Ed and N TEd , respectively. The extent of dissociation of cells was represented by the cell dissociation index N TC /N TE or N Ed /N TEd . The lower the values of these indexes, the higher the activity of cell adhesion. In some experiments, confluent cultures were pretreated with 1 M cytochalasin D in culture medium for 2 h and then the cell dissociation assay was performed in the presence of 1 M cytochalasin D as described (32).
Immunoprecipitation of E-cadherin from EL Cells-The immunologically collected cells (5 ϫ 10 6 cells) were transferred to a 6-well culture dish, and after incubation for 24 h, immunoprecipitation was performed as described previously (10). Briefly, the cells were harvested and suspended with phosphate-buffered saline in the presence of 0.75 mM DSP. Next the cells were incubated for 20 min at room temperature, and then the DSP activity was quenched by the addition of 50 mM glycine in phosphate-buffered saline. Finally, the cells were lysed with lysis buffer (50 mM Tris/HCl at pH 7.4, 50 mM NaCl, 10 M (p-amidinophenyl)-methanesulfonyl fluoride, 10 g/ml leupeptin, 0.5% (w/v) Triton X-100, 1 mM CaCl 2 ), and the lysates were mixed with indicated antibody and incubated for 1 h at 4°C. The immunocomplex was subjected to SDS-PAGE, followed by immunoblotting with the indicated antibodies.
We examined whether Cdc42 or Rac1 regulates the interaction of IQGAP1 with ␤-catenin in vitro. A mixture, containing GST-IQGAP1 with various small GTPases, was added to affinity beads coated with either MBP or MBP-␤-catenin-N-2-(aa 1-183). GTP␥S⅐GST-Cdc42 inhibited the binding of IQGAP1 to ␤-catenin-N-2 in a dose-dependent manner (Fig. 3A). A similar result was obtained when GTP␥S⅐GST-Rac1 was used instead of GTP␥S⅐GST-Cdc42 (Fig. 3B). In the presence of either GDP⅐GST-Cdc42, GDP⅐GST-Rac1, or GTP␥S⅐GST-RhoA, the interaction of IQGAP1 with ␤-catenin-N-2 was not affected (Fig. 3B). These results indicate that activated Cdc42 or Rac1 inhibits the interaction of IQGAP1 with ␤-catenin in vitro. We have previously shown that IQGAP1 interacts with E-cadherin as well as ␤-catenin both in vitro and in vivo. Therefore, we tried to examine whether Cdc42 or Rac1 affects the interaction of IQGAP1 with E-cadherin under the same conditions as ␤-catenin. However, neither Cdc42 nor Rac1 affected the interaction of IQGAP1 with E-cadherin (data not shown). The role of Cdc42 and Rac1 in the interaction of IQGAP1 with E-cadherin remains to be clarified. We have previously shown that Cdc42 Val12 counteracts the inhibitory effect of IQGAP1 on the E-cadherin-mediated cell-cell adhesion (10). Taken together, these observations suggest that the Cdc42 or Rac1 activation leads to stable cadherin-catenin complex formation in vivo through inhibition of the binding of IQGAP1 to ␤-catenin.
To examine the effect of Cdc42 and IQGAP1 on the cadherincatenin complex formation in vivo, EL cells were used because these cells are mouse L fibroblasts expressing E-cadherin and adhere to each other in an E-cadherin-dependent fashion (26). We have already found that IQGAP1 induces the dissociation of ␣-catenin from the cadherin-catenin complex in EL cells (10). We here examined whether activated Cdc42 suppresses the dissociation of ␣-catenin from the cadherin-catenin complex induced by IQGAP1 in EL cells. To enrich the EL cells transiently expressing IQGAP1 and Cdc42, the cells were transfected with plasmids encoding either Myc, Myc-IQGAP1, or Myc-IQGAP1 plus HA-Cdc42 Val12 together with a plasmid encoding tAic2A, an IL-3␤1 receptor lacking a cytoplasmic domain. The transfected cells were immunologically collected by anti-Aic2 antibody (10,29,30). E-cadherin was then immunoprecipitated from those cells. When E-cadherin was immunoprecipitated from the cells expressing Myc alone, both ␤-catenin and ␣-catenin were coimmunoprecipitated with E-cadherin. When E-cadherin was immunoprecipitated from the cells expressing Myc-IQGAP1, ␤-catenin, but not ␣-catenin, was coimmunoprecipitated with E-cadherin as described previously (10). When E-cadherin was immunoprecipitated from the cells expressing both Myc-IQGAP1 and HA-Cdc42 Val12 , both ␤-catenin and ␣-catenin were coimmunoprecipitated with E-cadherin (Fig. 4). These results indicate that the Cdc42 activation leads to stable cadherin-catenin complex formation by suppressing IQGAP1 function in vivo as well as in vitro.
We and others (7)(8)(9)(10)(11) have reported that Cdc42, Rac1, and RhoA are involved in the regulation of the cadherin-mediated cell-cell adhesion. However, direct evidence has yet to be presented that Cdc42, Rac1, and RhoA regulate the cadherinmediated cell-cell adhesion. We, therefore, examined whether these small GTPases affect the cadherin-mediated cell-cell adhesion using the cell dissociation assay (26), one of the adhesion assays for the evaluation of the cadherin activity. To obtain an enriched population of EL cells expressing recombi- The domain organization of ␤-catenin is presented (41). B, indicated MBP-␤-catenin mutants were mixed with affinity beads coated with either GST, GST-IQ-GAP1, or GST-Rho GDI. After the beads had been washed, the bound proteins were coeluted with GST fusion proteins by the addition of glutathione. The eluates were subjected to SDS-PAGE, followed by immunoblotting with anti-MBP antibody. The results are representative of three independent experiments. nant small GTPases, EL cells were transfected with plasmid encoding the various small GTPases together with a plasmid encoding tAic2A, and the transfected cells were immunologically collected by an antibody to Aic2 as described above. In the cell dissociation assay, EL cells expressing dominant negative Cdc42 (Cdc42 Asn17 ) were easily dissociated, and many single cells were observed (Fig. 5A). A similar result was obtained when the cells expressing dominant negative Rac1 (Rac1 Asn17 ) were used instead of the cells expressing Cdc42 Asn17 , whereas the control cells or the cells expressing Cdc42 Val12 , dominant active Rac1 (Rac1 Val12 ), or dominant active RhoA (RhoA Val14 ) displayed big aggregates (Fig. 5A). These results indicate that cells expressing Cdc42 Asn17 or Rac1 Asn17 show weaker E-cadherin-mediated adhesive activities than the control cells and the cells expressing Cdc42 Val12 , Rac1 Val12 , or RhoA Val14 . When the cell dissociation assay was performed using the cells expressing dominant negative RhoA (RhoA Asn19 ), the cells displayed smaller aggregates than the control cells, although single cells were hardly observed. The N TC /N TE index, which reflects the cadherin activity, of Cdc42 Asn17 -or Rac1 Asn17 -expressing EL cells was higher than that of cells transfected with the control plasmid, indicating that Cdc42 Asn17 and Rac1 Asn17 reduced the E-cadherin-mediated adhesive activity (Fig. 5B). RhoA Asn19 also reduced the E-cadherin-mediated adhesive activity to a lesser extent than Cdc42 Asn17 and Rac1 Asn17 (Fig. 5B).
These results are, however, not sufficient to conclude that Cdc42, Rac1, or RhoA affects the E-cadherin-mediated adhesive activity through the cadherin-catenin complex. Since Cdc42, Rac1, or RhoA regulates the reorganization of the actin cytoskeleton (5,6), it is also possible that these small GTPases affect actin cytoskeleton and then indirectly affect the cadherin function. To exclude the possibility that these small GTPases indirectly affect the cadherin activity, nE␣CL cells (26), which are L cells expressing E-cadherin-␣-catenin chimeric protein, were used. In nE␣CL cells, the cadherin-catenin complex is not remodeled (26), and it has been shown that the disorganization of actin cytoskeleton by treatment of EL cells and nE␣CL cells with cytochalasin D, which prevents actin from polymerizing, inhibits the E-cadherin-and nE␣C-mediated cell-cell adhesion (32,34) (Fig. 6A). In the cell dissociation assay, nE␣CL cells expressing Cdc42 Asn17 or Rac1 Asn17 formed aggregates (Fig. 5,  A and B). The effects of Cdc42 Asn17 and Rac1 Asn17 on nE␣CL cells were apparently weaker than those on EL cells. In contrast, the effect of RhoA Asn19 on nE␣CL cells was similar to that on EL cells (Fig. 5, A and B). These results suggest that Cdc42 or Rac1 affects the E-cadherin-mediated adhesive activ-ity mainly by acting on the cadherin-catenin complex, presumably through the regulation of the interaction of IQGAP1 with ␤-catenin and that RhoA affects the E-cadherin-mediated adhesive activity presumably through actin cytoskeleton rather than the cadherin-catenin complex.
Moreover, to strengthen our conclusion that Cdc42 or Rac1 regulates the E-cadherin-mediated adhesive activity mainly by acting on the cadherin-catenin complex, the effect of Cdc42 Asn17 or Rac1 Asn17 on the neuropilin-1-mediated cell-cell adhesion was examined using LAP86 cells (27), which are L cells expressing mouse neuropilin-1. It has been shown that neuropilin-1 is a Ca 2ϩ -independent cell adhesion molecule, and it mediates adhesion by heterotypic interaction of neuropilin-1 with protease-sensitive molecules on the cell surface (27,31). The neuropilin-1-mediated cell-cell adhesion does not require catenins. At first, the requirement of actin cytoskeleton for the neuropilin-1-mediated cell-cell adhesion in LAP86 cells was examined using cytochalasin D. Under the conditions in which the E-cadherin-and nE␣C-mediated cell-cell adhesions were inhibited by cytochalasin D, neuropilin-1-mediated cell-cell adhesion was not affected at all (Fig. 6A). Thus, neither actinbased cytoskeleton nor catenins are required for the neuropilin-1-mediated cell adhesive activity. By using this cell line, the effect of Cdc42 Asn17 , Rac1 Asn17 , or RhoA Asn19 on the neuropilin-1-mediated cell-cell adhesion was examined by the cell dissociation assay. Cdc42 Asn17 , Rac1 Asn17 , or RhoA Asn19 had no effect on the neuropilin-1-mediated cell adhesive activity (Fig. 6B). Cdc42 and Rac1 appear to regulate the cadherin-mediated cell-cell adhesion mainly by acting on the cadherin-catenin complex through IQGAP1 and partially by acting on the actin cytoskeleton. DISCUSSION We have recently found that IQGAP1, a target of Cdc42 and Rac1, regulates the cadherin-mediated cell-cell adhesion (10). However, it is not known how Cdc42 and Rac1 regulate the IQGAP1 function. In this study, we demonstrated that Cdc42 and Rac1 regulate the interaction of IQGAP1 with ␤-catenin. Based on the present study together with previous observations (10), we propose the mechanism by which Cdc42 and Rac1 FIG. 4. Effect of Cdc42 Val12 on the dissociation of ␣-catenin from the cadherin-catenin complex induced by IQGAP1. pEF-BOS-Myc or pEF-BOS-Myc-IQGAP1 together with pEF-BOS-HA or pEF-BOS-HA-Cdc42 Val12 were cotransfected with pME18-tAic2A, IL-3␤1 receptor lacking a cytoplasmic domain, into EL cells. EL cells expressing Myc-IQGAP1 and/or HA-Cdc42 Val12 were immunologically collected using anti-IL-3␤1 receptor antibody. By using the immunoisolated cells, E-cadherin was immunoprecipitated using anti-E-cadherin antibody or control IgG. The immunocomplex was subjected to SDS-PAGE, followed by immunoblotting with indicated antibodies. The results are representative of three independent experiments.

FIG. 5.
Effect of Cdc42 Asn17 , Rac1 Asn17 , and RhoA Asn19 on the E-cadherin-mediated cell-cell adhesion. A, indicated expression vectors of the small GTPases were transfected into either EL cells or nE␣CL cells. The transfected cells were immunologically collected as described above. By using the collected cells, the cell dissociation assay was performed after treatment of the cells with trypsin in the presence of Ca 2ϩ (TC treatment). Bar, 50 m. B, the cell dissociation assay was performed after treatment of the cells with trypsin in the presence of either Ca 2ϩ (TC treatment) or EGTA (TE treatment). The total particle numbers after TC or TE treatment were designated as N TC and N TE , respectively. together with IQGAP1 regulate the cadherin-mediated cell-cell adhesion (Fig. 7). When Cdc42 and Rac1 are in the GDP-bound inactive forms, Cdc42 and Rac1 cannot interact with IQGAP1, and IQGAP1 interacts with ␤-catenin, thereby dissociating ␣-catenin from the cadherin-catenin complex. This state confers the weak adhesive activity. In contrast, when Cdc42 and Rac1 are in the GTP-bound active forms at sites of cell-cell contact, Cdc42 and Rac1 interact with IQGAP1 and thereby prohibit IQGAP1 from interacting with ␤-catenin. This results in the stabilization of the cadherin-catenin complex. Thus, Cdc42 and Rac1 positively regulate cadherin-mediated cell-cell adhesion by the suppression of the activity of IQGAP1 to perturb the cadherin-catenin complex. This state confers the strong adhesive activity. Thus, Cdc42 and Rac1 and IQGAP1 can serve as positive and negative molecular switches of the cadherin activity, respectively.
We and others (7)(8)(9)(10)(11) have shown that Cdc42, Rac1, and RhoA are involved in the regulation of the E-cadherin-mediated cell-cell adhesion. Braga et al. (7) have shown that microinjection of Rac1 Asn17 or Botulinus C3 ADP-ribosyltransferase (C3), which inactivates Rho, into human keratinocytes inhibits the localization of E-cadherin at the sites of cell-cell adhesion, unlike microinjection of either RhoA Val14 or Rac1 Val12 . Very recently, they have also shown that the effect of RhoA and Rac1 on the cadherin-mediated cell-cell adhesion depends on the cellular context and the maturation status of the junctions (11). Takaishi et al. (8) have shown that, in the stable transformants of Madin-Darby canine kidney (MDCK) cells expressing Rac1 Val12 , the immunofluorescent levels of E-cadherin, ␤-catenin, and actin filament at the sites of cell-cell adhesion increase, whereas in the cells expressing Rac1 Asn17 those of E-cadherin, ␤-catenin, and actin filament decrease. Microinjection of C3 into wild-type MDCK cells inhibits cadherinmediated cell-cell adhesion, but microinjection of C3 into MDCK cells expressing Rac1 Val12 does not (8). Moreover, we have found that microinjection of Rho GDI, a negative regulator of the Rho family members, into MDCK cells induces the perturbation of cell-cell adhesion, and co-microinjection of Cdc42 Val12 or Rac1 Val12 , but not RhoA Val14 , reverses the inhibitory action of Rho GDI (9). These previous observations strongly suggest that the activity of Cdc42, Rac1, and RhoA is essential for maintaining the cadherin-mediated cell-cell adhesion and that Cdc42 and Rac1 regulate it presumably through a mechanism distinct from that involving RhoA. However, direct evidence has yet to be presented that Cdc42, Rac1, and RhoA regulate the cadherin-mediated cell-cell adhesion. Here, we showed that using a cell dissociation assay, the expression of Cdc42 Asn17 or Rac1 Asn17 reduced the E-cadherin-mediated FIG. 6. Effect of Cdc42 Asn17 , Rac1 Asn17 , and RhoA Asn19 on the neuropilin-1-mediated cell-cell adhesion. A, EL cells, nE␣CL cells, or LAP86 cells were preincubated either in the absence (open column) or presence (closed column) of 1 M cytochalasin D for 2 h. The cell dissociation assay with EL cells or nE␣CL cells was performed as described in Fig. 5. The photographs shown are representative of three independent experiments after TC treatment. The cell dissociation assay with LAP86 cells was performed after treatment of the cells with EDTA either in the presence (TEd treatment) or absence (Ed treatment) of trypsin. The photographs shown are representative of three independent experiments after Ed treatment. The total particle numbers after Ed or TEd treatment were designated as N Ed and N TEd , respectively. The extent of dissociation of cells was represented by the cell dissociation index N TC /N TE or N Ed /N TEd . These ratios indicate the adhesive activity. The lower the value of N TC /N TE or N Ed /N TEd , the higher the activity of cell adhesion. The values shown are means Ϯ S.E. of triplicates. B, indicated expression vectors of the small GTPases were transfected into LAP86 cells. The transfected cells were immunologically collected as described above. By using the collected cells, the cell dissociation assay was performed as described in A. The photographs shown are representative of three independent experiments after Ed treatment. and Rac1 are in the GDP-bound inactive forms, Cdc42 and Rac1 cannot interact with IQGAP1, and IQGAP1 interacts with ␤-catenin, thereby dissociating ␣-catenin from the cadherin-catenin complex. This state confers the weak adhesive activity. When Cdc42 and Rac1 are in the GTP-bound active forms at sites of cell-cell contact, Cdc42 and Rac1 interact with IQGAP1. Then, IQGAP1 is unable to interact with ␤-catenin, resulting in the stabilization of the cadherin-catenin complex. Thus, Cdc42 and Rac1 positively regulate cell-cell adhesion by the suppression of the activity of IQGAP1 to perturb the cadherin-catenin complex. Therefore, Cdc42, Rac1, and IQGAP1 can cyclically regulate cell-cell adhesion by remodeling the cadherin-catenin complex. cell adhesive activity in EL cells and that the expression of RhoA Asn19 also reduced the activity to a lesser extent. The inhibitory effects of Cdc42 Asn17 and Rac1 Asn17 on the cell adhesive activity in nE␣CL cells, which require actin cytoskeleton for the adhesion like EL cells but not catenins unlike EL cells, were apparently weaker than those in EL cells, whereas the inhibitory effect of RhoA Asn19 in nE␣CL cells was similar to that in EL cells. Cdc42 Asn17 , Rac1 Asn17 , and RhoA Asn19 had no effects on the neuropilin-1-mediated cell-cell adhesion which requires neither catenins nor actin cytoskeleton. It has been shown that Cdc42, Rac1, and RhoA regulate the actin cytoskeleton (5,6). Taken together, these observations suggest that Cdc42 and Rac1 regulate the E-cadherin-mediated adhesive activity mainly by acting on the cadherin-catenin complex and partially by acting on the actin cytoskeleton, whereas RhoA regulates the E-cadherin-mediated adhesive activity presumably through actin cytoskeleton or other components.
Target molecules for Cdc42 and Rac1 have been identified to be p21-activated kinase (12)(13)(14), WASP (15,16), and IQGAPs (17)(18)(19)(20), and those for Cdc42 to be N-WASP (21) and MRCK-␣, -␤ (22). Recently, MRCK-␣ has been shown to be localized at cell-cell contact sites when MRCK-␣ together with Cdc42 Val12 is coexpressed in HeLa cells and to regulate the actin cytoskeleton through phosphorylation of myosin light chain (22). This result suggests that MRCK-␣ is also involved in the regulation of cell-cell adhesion possibly through the regulation of actin filament rearrangement. In addition to MRCK, the WASP family has been shown to regulate reorganization of actin (15,16,21,35). Therefore, it is possible and reasonable that the cadherin activity is dually regulated by Cdc42 and Rac1 through the direct regulation of the cadherin-catenin complex by IQGAP1 and the indirect regulation of actin filaments by MRCK and WASP family members.
Despite numerous attempts to clarify the regulation of the cadherin function, it is not fully understood. Increased tyrosine phosphorylation of ␤-catenin appears to correlate with dysfunction of cadherin-mediated cell-cell adhesion, induced by the expression of v-Src (36,37) or treatment with growth factors, such as epidermal growth factor or hepatocyte growth factor (38). It has been shown that treatment of cells with pervanadate, a potent inhibitor of tyrosine phosphatases, induces perturbation of the cadherin-mediated cell-cell adhesion. In accordance with dysfunction of cadherin-mediated cell-cell adhesion, tyrosine phosphorylation of ␤-catenin and dissociation of ␣-catenin from the cadherin-catenin complex are observed (39). Therefore, the tyrosine phosphorylation of ␤-catenin has been thought to be important for the regulation of the cadherin function. Taken together with our model, it is tempting to speculate that tyrosine-phosphorylated ␤-catenin recruits IQGAP1 or GTPase-activating proteins for Cdc42 and/or Rac1 to the cell-cell contact sites, resulting in the reduction of cadherin activity through the inhibition of Cdc42 and/or Rac1 activity. However, it has been reported that the tyrosine phosphorylation of ␤-catenin is not required for the reduction of cadherin activity induced by v-Src, since the expression of v-Src reduces the cadherin function of nE␣CL cells as well as EL cells (40). Thus, the physiological role of the tyrosine phosphorylation of ␤-catenin in cadherin function remains to be clarified.
In conclusion, we showed how Cdc42 and Rac1 regulate the cadherin-mediated cell-cell adhesion through IQGAP1. However, it remains to be clarified in which physiological situations this system operates. The cadherin-mediated cell-cell adhesion is dynamically rearranged in various situations, such as compaction of early embryogenesis, cell scattering, wound-induced cell migration, and tumorigenesis (1-3). It is plausible that Cdc42 and Rac1 together with IQGAP1 regulate these processes. We are currently investigating whether IQGAP1, downstream of Cdc42 and Rac1, participates in these physiological phenomena.