Src Kinase Regulates the Activation of a Novel FGD-1-related Cdc42 Guanine Nucleotide Exchange Factor in the Signaling Pathway from the Endothelin A Receptor to JNK*

Small GTPases act as binary switches by cycling between an inactive (GDP-bound) and an active (GTP-bound) state. Upon stimulation with extracellular signals, guanine-nucleotide exchange factors (GEFs) stimulate the exchange of GDP to GTP to shift toward the active forms of small GTPases, recognizing the downstream targets. Here we show that KIAA0793, containing substantial sequence homology with the catalytic Dbl homology domain of the faciogenital dysplasia gene product (FGD1), is a specific GEF for Cdc42. We, therefore, tentatively named it FRG (FGD1-related Cdc42-GEF). Src kinase directly phosphorylates and activates FRG, as Vav family GEFs. Additionally, FRG is involved in the signaling pathway from the endothelin A receptor to c-Jun N-terminal kinase, resulting in the inhibition of cell motility. These results suggest that FRG is a member of Cdc42-GEF and plays an important role in the signaling pathway downstream of G protein-coupled receptors. Rho family small GTPases comprise a large branch within the Ras family of low molecular weight guanine nucleotide-binding proteins (1–4). One of the well known roles of Rho family GTPases is the regulation of the actin cytoskeleton the DH and PH domains (491–827), and the second PH domain (821– 1055) were amplified from cDNA of FRG as a template and ligated into the mammalian FLAG-tag expression vector pCMV-FLAG. The cDNAs of RhoA, Rac1, and Cdc42 were amplified from total RNA of human kidney 293T cells using the method of reverse transcription polymerase chain reaction and were inserted the Escherichia coli GST-tag expres- sion vector pET42a. pCMV-FLAG-RhoA, pCMV-FLAG-Rac1, pCMV-FLAG-Cdc42, pCMV-FLAG-RhoG14V, pCMV-FLAG-RacG12V, pCMV- FLAG-Cdc42G12V, pCMV-FLAG-RhoT19N, pCMV-FLAG-RacT17N, pCMV-FLAG-Cdc42T17N, and pCMV-FLAG-MKK4K95R were con- structed as described previously (30, 39). pUSE-CA-Src (a constitutively activated mutant of c-Src) was purchased from Upstate Biotechnology. The E. coli expression plasmids encoding the RhoA binding domain (RBD) of mDia1 and the Rac1 and Cdc42 binding domain (CRIB) of (cid:1) Pak were constructed as described previously (32, 40–41). pME-ETA receptor-enhanced green fluorescence protein was generously provided by Dr. T. Sakurai (University of Tsukuba, Ibaraki, Japan) (42). pGEX2T-c-Jun (amino acids 1–221) and SR (cid:1) -HA-JNK1 were kindly provided by Dr. M. Karin (University of California, San CA). All DNA sequences were confirmed using DNA sequencer L-4200L (LI-COR). Recombinant Proteins— Recombinant GST-tagged mDia1RBD, (cid:1) Pa- kCRIB, Rho family GTPases, and c-Jun proteins were purified using E. (DE3) pLysS, as described m m M 0.1 (cid:2) g/ml leupeptin, 0.1 m M EGTA, (cid:2) M Na 3 VO 4 , and 2 m M (cid:3) -glycerophosphate) for Src kinase, or an exchange buffer (20 m M HEPES-NaOH 7.5), 5 m M MgCl 2 , 150 m M NaCl, 1 m M dithiothre-itol, 0.1 m M phenylmethanesulfonyl fluoride, 0.1 (cid:2) g/ml leupeptin, and 3 m M EDTA). To compare the amounts of the immunoprecipitates or expressed proteins in the cell lysates in each transfection, the precipi- tates or aliquots of the cell lysates were boiled in a Laemmli sample buffer and then separated on 8–15% SDS-polyacrylamide gels. The electrophoretically separated proteins were transferred to polyvinyli-dene difluoride membranes. The membranes were blocked with phos- phate-buffered saline containing 0.1% Tween 20 and 10 mg/ml bovine serum albumin and immunoblotted with antibodies. The bound anti- bodies were detected using the ECL system (Amersham Biosciences) with anti-mouse or anti-rabbit IgG antibodies conjugated with horse- radish peroxidase according to the manufacturer’s protocol. Images of protein bands were captured using Adobe Photoshop 5.0 plug-in soft- ware and an EPSON GT-7000U scanner. The band intensity of kinases was semiquantified using NIH Image 1.61. Representatives of at least three separate experiments are in

Rho family small GTPases comprise a large branch within the Ras family of low molecular weight guanine nucleotidebinding proteins (1)(2)(3)(4). One of the well known roles of Rho family GTPases is the regulation of the actin cytoskeleton (5,6). Rho family GTPases RhoA, Rac, and Cdc42 induce the formation of stress fibers, lamellipodia, and filopodia, respectively (5). Furthermore, Rho family GTPases have been implicated in the control of diverse physiological responses, such as cell proliferation and cell motility as well as in physiopathological processes such as transformation and metastasis (7,8).
Rho family GTPases adopt either an active GTP-bound or an inactive GDP-bound conformational state (2)(3)(4). Their activity is controlled positively by guanine nucleotide exchange factors (GEFs), 1 which catalyze the replacement of GDP with GTP, and negatively by GTPase-activating proteins, which accelerate the endogenous GTPase activity (4,9). Dbl family proteins are GEFs for Rho family GTPases and share a Dbl homology (DH) domain adjacent to a pleckstrin homology (PH) domain structure (4,9). The DH domain is responsible for catalytic activity (10), and the PH domain appears to be necessary for proper localization and full activation (11)(12)(13). Numerous members of Rho family GEFs have been identified in mammals (9). Some GEFs are specific for each Rho family GTPase, whereas others show a broad activity (9). For example, p115RhoGEF/Lsc and Tiam1 are specific GEFs for RhoA (14 -17) and Rac1 (18), respectively, and Vav acts as the GEF for RhoA, Rac1, and Cdc42 (19 -25). The molecular mechanism by which GEFs are activated is largely unknown, although it has been shown that Vav family GEFs are directly tyrosine-phosphorylated and activated by Src kinase (20 -22).
The G protein-coupled receptors (GPCRs) are the largest family of human cell-surface receptors (26 -29). Some GPCRs activate c-Jun N-terminal kinase (JNK), a subfamily of mitogen-activated protein kinases (28). This pathway depends on Src kinase and Rho family GTPases, Src kinase often acting upstream of Rho family GTPases (28, 30 -32). We previously showed that endothelin, a ligand of GPCRs, activates JNK via the signaling pathway dependent upon Src kinase and Rho family GTPases Rac1 and Cdc42 (33). In this signaling pathway, Src kinase acts as an upstream regulator of Rac1 and Cdc42. However, the detailed mechanism linking Src kinase to Rho family GTPases remains unclear (34).
To clarify the mechanism of GPCR-mediated activation of Rho family GTPases, we performed a BLAST search using Dbl big sister (Dbs), a GEF involved in GPCR-mediated activation of Rho family GTPases (35), as a query. As a result, we found a novel Rho family GEF gene, KIAA0793. The KIAA0793 protein shows sequence similarity to a human faciogenital dysplasia gene product (FGD1), which is a specific activator of Cdc42 (36) and appears to be a family of related Cdc42-GEFs including mouse Fgd2 and Fgd3 (37) and mouse Frabin (38).
In the present study, we show that KIAA0793 is a specific GEF toward Cdc42, thus termed FRG (FGD1-related Cdc42-GEF). FRG is directly phosphorylated by Src kinase, which increases the Cdc42-GEF activity of FRG. We also demonstrate that FRG is involved in the signaling pathway linking the endothelin A (ETA) receptor to JNK, leading to the inhibition of cell motility. Taken together, our results indicate that FRG is a novel member of Cdc42-GEFs and may act as a direct regulator linking the Src kinase and Rho family GTPases in the chemorepellant GPCR-JNK-signaling pathway.

EXPERIMENTAL PROCEDURES
Antibodies-A mouse monoclonal antibody M2 against a FLAG-peptide was obtained from Sigma-Aldrich. A mouse monoclonal antibody 12CA5 against the hemagglutinin (HA) epitope was purchased from Roche Diagnostics. A mouse monoclonal antibody 1E4 against Aequorea victoria green fluorescence protein was obtained from Medical and Biological Laboratories, Co., Ltd. (Nagoya, Japan). A mouse monoclonal anti-phosphorylated tyrosine antibody 4G10 was purchased from Upstate Biotechnology (Lake Placid, NY). A rabbit polyclonal antibody against SRC2 and a mouse monoclonal antibody 26C4 against RhoA were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Mouse monoclonal antibodies against Rac1 and Cdc42 were purchased from BD Biosciences (San Jose, CA). Anti-mouse and anti-rabbit IgG antibodies conjugated with horseradish peroxidase were obtained from Amersham Biosciences.
Immunoprecipitation and Immunoblotting-After the addition of endothelin-1, cells were lysed in 600 l of lysis buffer A (20 mM HEPES-NaOH (pH 7.5), 3 mM MgCl 2 , 100 mM NaCl, 1 mM dithiothreitol, 1 mM phenylmethanesulfonyl fluoride, 1 g/ml leupeptin, 1 mM EGTA, 1 mM Na 3 VO 4 , 10 mM NaF, 20 mM ␤-glycerophosphate, and 0.5% Nonidet P-40) for a 60-mm dish. The lysates were centrifuged at 15,000 rpm for 10 min at 4°C. Aliquots (400 g) of the supernatants were mixed with protein G-Sepharose CL-4B preabsorbed with 0.5 g of anti-FLAG or 0.5 g of anti-HA antibody. The immune complexes were precipitated by centrifugation and washed twice with lysis buffer A and twice with kinase buffer A (20 mM HEPES-NaOH (pH 7.5), 10 mM MgCl 2 , 1 mM dithiothreitol, 0.1 mM phenylmethanesulfonyl fluoride, 0.1 g/ml leupeptin, 0.1 mM EGTA, 10 M Na 3 VO 4 , and 2 mM ␤-glycerophosphate) for JNK, kinase buffer B (20 mM HEPES-NaOH (pH 7.5), 10 mM MgCl 2 , 3 mM MnCl 2 , 1 mM dithiothreitol, 0.1 mM phenylmethanesulfonyl fluoride, 0.1 g/ml leupeptin, 0.1 mM EGTA, 10 M Na 3 VO 4 , and 2 mM ␤-glycerophosphate) for Src kinase, or an exchange buffer (20 mM HEPES-NaOH (pH 7.5), 5 mM MgCl 2 , 150 mM NaCl, 1 mM dithiothreitol, 0.1 mM phenylmethanesulfonyl fluoride, 0.1 g/ml leupeptin, and 3 mM EDTA). To compare the amounts of the immunoprecipitates or expressed proteins in the cell lysates in each transfection, the precipitates or aliquots of the cell lysates were boiled in a Laemmli sample buffer and then separated on 8 -15% SDS-polyacrylamide gels. The electrophoretically separated proteins were transferred to polyvinylidene difluoride membranes. The membranes were blocked with phosphate-buffered saline containing 0.1% Tween 20 and 10 mg/ml bovine serum albumin and immunoblotted with antibodies. The bound antibodies were detected using the ECL system (Amersham Biosciences) with anti-mouse or anti-rabbit IgG antibodies conjugated with horseradish peroxidase according to the manufacturer's protocol. Images of protein bands were captured using Adobe Photoshop 5.0 plug-in software and an EPSON GT-7000U scanner. The band intensity of kinases was semiquantified using NIH Image 1.61. Representatives of at least three separate experiments are shown in the figures.
Kinase Assays-After the addition of 100 nM endothelin-1 for 20 min, transfected cells were lysed in 600 l of lysis buffer A for a 60-mm dish, and the lysates were centrifuged as described above. Aliquots (400 g) of the supernatants were mixed with protein G-Sepharose CL-4B preabsorbed with anti-HA antibody. The activities of the proteins were measured as the 32 P radioactivity incorporated into recombinant c-Jun using a BAS 2500 imaging analyzer (Fujifilm, Tokyo, Japan) as described previously (30 -33, 38). Activities were normalized to the amounts of immunoprecipitated JNK, as estimated from these band intensities.
Pull-down Assays of Rho Family GTPases-Cells were stimulated with or without 100 nM endothelin-1 for 15 min and then lysed in lysis buffer B (20 mM HEPES-NaOH (pH 7.5), 20 mM MgCl 2 , 150 mM NaCl, 1 mM dithiothreitol, 1 mM phenylmethanesulfonyl fluoride, 1 g/ml leupeptin, 1 mM EGTA, 1 mM Na 3 VO 4 , 10 mM NaF, and 0.5% Nonidet P-40). To detect active GTP-bound Rho family GTPases in the cell lysates, we performed pull-down assays using recombinant GST-tagged mDia1RBD (for RhoA) and ␣PakCRIB (for Rac1 and Cdc42) as described previously (32, 40 -41). Aliquots (400 g) of the supernatants mixed with glutathione-Sepharose 4B with 10 g of GST-mDia1RBD or 20 g of GST-␣PakCRIB were precipitated by centrifugation, and complexes were boiled in a Laemmli sample buffer and then separated on 15% SDS-polyacrylamide gels. The separated proteins were immunoblotted using each antibody against Rho family small GTPases.
Exchange Assays-After the addition of endothelin-1 for 15 min, cells were lysed in 900 l of lysis buffer A for a 10-cm dish, and the lysates were centrifuged as described above. Aliquots (800 g) of the supernatants were used for the exchange assay. For releasing assays, the [ 3 H]GDP-bound Rho family GTPases were obtained by incubation with an exchange buffer containing 125 ng/l Rho GTPases, 250 ng/l bovine serum albumin, 5 mM EDTA, and 0.3 M [ 3 H]GDP (0.3 Ci/l) at 30°C for 90 min. The reaction was stopped by adding 5 mM MgCl 2 , and mixtures were immediately cooled on ice. The immunoprecipitated DH domains of FRG and the FRG protein were incubated in 30 l of an exchange buffer containing 16 ng/l Rho GTPases⅐[ 3 H]GDP, 33 ng/l bovine serum albumin, and 3 M cold GDP at 30°C for 0, 10, 20, and 30 min. The reactions were stopped by adding 1 ml of an ice-cold wash buffer (20 mM HEPES-NaOH (pH 7.5) and 10 mM MgCl 2 ) and filtered through nitrocellulose membranes. The membranes were immediately washed twice with an ice-cold wash buffer and air-dried. The radioactivity remaining on each membrane was measured by an LS6500 liquid scintillation counter (Beckman Coulter Inc., Palo Alto, CA). For the binding assay, the immunoprecipitated DH domain of FRG and the FRG protein were incubated in 30 l of an exchange buffer containing 16 ng/l Rho GTPases, 33 ng/l bovine serum albumin, and 3 M [ 32 P]GTP (0.1 Ci/l) at 30°C for 0, 10, 20, and 30 min. The reactions were stopped by adding an ice-cold wash buffer and filtered through nitrocellulose membranes. The membranes were immediately washed twice with a wash buffer and air-dried. The radioactivity remaining on each membrane was measured as before. Activities were normalized to the amounts of FRG polypeptides in the immunoprecipitates.
In Vitro Tyrosine Phosphorylation-Cells were lysed in 900 l of lysis buffer A for a 10-cm dish, and the lysates were centrifuged, as described above. Aliquots (800 g) of the supernatants were used for an in vitro phosphorylation reaction using recombinant c-Src (6 units) in 30 l of kinase buffer B containing 20 M ATP at 30°C for 15 min and then chilled on ice. The tyrosine-phosphorylated FRG protein was washed twice with an exchange buffer and used for an exchange reaction for Cdc42.
Cell Motility Assay-Cell motility was measured using a 24-well Boyden chamber (BD Biosciences) according to the manufacturer's protocol. Briefly, upper wells with polyethylene terephthalate filters (8-m pore size) were coated with 10 g/ml extra cellular matrix E-C-L (Upstate). Serum-starved cells (2 ϫ 10 5 cells in 500 l of Dulbecco's modified Eagle's medium per well) were loaded into upper wells, which were immediately plated on the chamber containing 165 nM endothelin-1 (750 l of Dulbecco's modified Eagle's medium per well). After incubation at 37°C for 5 h, upper filters were stained with a Diff-Quick staining kit (Biochemical Sciences Inc., Sterling Height, MI) according to the manufacturer's protocol. Cells that had not migrated were wiped away from the inner surface of the upper wells. Using an optical microscope, the number of migrated cells was counted in at least three independent experiments.
Statistical Analysis-Statistical analysis was performed using SAS StatView 5.0. Values shown represent the mean Ϯ S.E. from at least three separate experiments. A Student's t test was carried out for intergroup comparisons (*, p Ͻ 0.01).

KIAA0793 Shows a Significant Homology to Cdc42-specific
GEFs-We performed a BLAST search using the DH domain of Dbs because Dbs was implicated in the signaling pathway linking GPCRs to Rho family GTPases (35). The search revealed the appearance of several uncharacterized proteins exhibiting a putative DH domain. One of them, KIAA0793, was of particular interest and further characterized. Fig. 1 shows the structural features of KIAA0793 in comparison with those of other Rho family GEFs. This molecule contains a tandem of DH and two PH domains, the DH domain being closely related to those of CDEP (chondrocyte-derived ezrin-like domain-containing protein) (43, 44) (54% identity, 71% similarity), Frabin (38) (30% identity, 47% similarity), FGD1 (36) (28% identity, 51% similarity), Fgd2 (30% identity, 51% similarity), and Fgd3 (37) (26% identity, 46% similarity). KIAA0793 contains an N-terminal region exhibiting extensive homology to the FERM domain, also called band 4.1 homology domain of ezrin, radixin, and moesin (ERM) proteins, which function as the cross-linker between plasma membranes and actin filaments (45,46). KIAA0793 exhibits an additional structural feature, a prolinerich region (amino acids 398 -498) (47).
KIAA0793 Activates Cdc42 but Not RhoA and Rac1-To examine which Rho family GTPases are activated by the DH and PH domains of KIAA0793, the ability of KIAA0793 to induce the dissociation of [ 3 H]GDP from recombinant RhoA, Rac1, and Cdc42 proteins was measured. It has been demonstrated that the isolated fragment containing DH and PH domains of Rho family GEFs acts as a limited catalytic domain toward Rho family GTPases (12,48). As shown in Fig. 2A, the isolated DH and PH fragment of KIAA0793 promoted nucleotide exchange on Cdc42 but not on RhoA or Rac1. In agreement with the release of [ 3 H]GDP from Cdc42, the fragment of KIAA0793 promoted the binding of [ 32 P]GTP to Cdc42 but not to RhoA and to Rac1 in a time-dependent manner (Fig. 2B). These results suggested that KIAA0793 may be the GEF toward Cdc42.
To confirm further whether KIAA0793 acts as the Cdc42specific GEF in vivo, we measured the GTP-bound form of endogenous and exogenous RhoA, Rac1, and Cdc42 in 293T cells expressing the isolated DH and PH fragment. The RhoA binding domain of mDia1 was used to precipitate GTP-bound RhoA from the cell lysates (40). The Rac1 and Cdc42 interactive binding domain of ␣PakCRIB was used to pull down activated GTP-bound Rac1 or Cdc42 from the cell lysates (41). Fig. 2C shows that transfection of the DH and PH fragment of KIAA0793 to cells activates Cdc42 but not RhoA and Rac1, suggesting that KIAA0793 acts as the Cdc42-specific GEF. As discussed above, the DH domain is responsible for the nucleotide exchange activity for Rho family GTPases, and FGD1 and Frabin have been shown to behave as the Cdc42-specific GEFs (36,38). Therefore, this newly identified molecule KIAA0793 is tentatively named FRG (FGD1-related Cdc42-GEF).
JNK Activation by Endothelin Is Mediated by Src Kinase and Cdc42-It has been shown that GPCR-mediated activation of JNK involves Src kinase and Rho family GTPases (8, 30 -33). To confirm whether the ETA receptor activates JNK dependent on Src kinase and Rho family GTPases in 293T cells, we added endothelin to cells that were transfected transiently with the plasmid encoding cDNA of the ETA receptor and HA-tagged JNK. Using an anti-HA antibody, the epitope-tagged JNK was immunoprecipitated from the cell lysates, and the in vitro kinase activity was assessed as the radioactivity incorporated into recombinant c-Jun. The peak of JNK activation was observed 20 min after the addition of endothelin (Fig. 3A).
Next we examined the involvement of Src kinase and Rho family GTPases in endothelin-induced JNK activation. As shown in Fig. 3B, pretreatment of cells with PP1 or PP2, specific inhibitors of Src kinase, inhibited the activation of JNK induced by endothelin. Pretreatment with C. difficile toxin B, which glycosylates Rho family GTPases RhoA, Rac1, and Cdc42 and inhibits their intracellular functions, also blocked JNK activation elicited by endothelin (Fig. 3C). Additionally, the endothelin-induced activation of JNK was suppressed only by the dominant-negative mutant of Cdc42 but not by the dominant-negative mutants of RhoA and Rac1. These results suggest that endothelin may activate JNK through Src kinase and Cdc42.
To investigate whether Rho family GTPases are activated by endothelin, we measured the active form of Rho family GTPases in cells. As shown in Fig. 3D, stimulation of endothelin activates Cdc42 but not RhoA and Rac1. Pretreatment of cells with PP1 inhibited the endothelin-induced activation of Cdc42. Again, these results indicate that endothelin induces Cdc42 activation, which may be regulated by Src kinase.
FRG Is Involved in the Signaling Pathway from the ETA Receptor to JNK-To examine the involvement of FRG in the signaling pathway induced by endothelin, we co-transfected the plasmids encoding the ETA receptor with FRG (⌬DHPH) or FRG (⌬DH). When transfected into cells, FRG (⌬DHPH), which is the deletion mutant lacking the DH and PH domains, and FRG (⌬DH), which is the deletion mutant lacking the DH domain, were detectable with an anti-HA antibody (Fig. 4A). It is known that the mutant lacking DH and PH domains and the mutant carrying the mutation within the DH domain of Rho family GEF show a dominant inhibitory effect (49,50). As shown in Fig. 4B, the expression of the deletion mutants of FRG suppressed endothelin-induced activation of Cdc42, suggesting that FRG is involved in the activation of Cdc42 after the stimulation of endothelin. Likewise, an in vitro kinase assay revealed that JNK activation induced by endothelin was inhibited by co-transfection of FRG (⌬DHPH) or FRG (⌬DH) (Fig. 4C). Taken together with the data of Fig. 4B, these results indicate that FRG may be a candidate of Rho family GEF for the signaling pathway from the ETA receptor to JNK.
FERM and Proline-rich Regions of FRG May Be Responsible for the Endothelin-induced JNK Activation-To identify the functional domains of FRG responsible for the signaling pathway from the ETA receptor to JNK, we utilized the variants including the FRG-FERM domain (FRG-FERM), the FRG-proline-rich domain (FRG-Proline), the FRG-DH and -PH domains (FRG-DHPH), and the FRG-second PH domain (FRG-2ndPH). Each fragment was detected with an anti-FLAG antibody in the transfected cells (Fig. 4A). As shown in Fig. 4D, the endothelin-induced activation of JNK was blocked by FERM and the proline-rich regions of FRG. On the other hand, the DH and PH domains of FRG induced an increase of JNK activity without endothelin stimulation (Fig. 4E), supporting the idea that the catalytic DH and PH domains directly activate Cdc42 (Fig. 2). Other variants, FRG-FERM, FRG-Proline, and FRG-2ndPH, did not induce JNK activation (Fig. 4E). The inhibitory effect of FERM and the proline-rich regions of FRG indicates that both regions might be involved in JNK activation induced by endothelin.

Involvement of FRG in the Inhibition of Cell Motility Dependent on the JNK Pathway Induced by the ETA Receptor-We
previously reported that the endothelin inhibits cell motility through the JNK pathway (33). To examine whether FRG inhibits the cell motility, we transfected the plasmid encoding the catalytic DH and PH domains of FRG into cells. As shown in Fig. 4F, the DH and PH domains inhibited cell motility. In addition, MKK4K95R, a kinase-deficient variant that inhibits the transmission of signals from GPCRs to JNK (30 -32), rescued the FRG-DH and -PH domain-induced inhibition of the cell motility. These results suggest that FRG suppresses cell motility through the JNK pathway.
As indicated before, the endothelin-induced activation of the JNK pathway involved FRG. Thus, we investigated the involvement of FRG in the endothelin-induced inhibition of the cell motility. As shown in Fig. 4G, co-transfection of FRG (⌬DHPH) or FRG (⌬DH) with the ETA receptor into cells blocked the inhibition of the cell motility induced by endothelin. Taken together with the results of Fig. 4F, these results indicate FRG is involved in the endothelin-induced inhibition of the cell motility through the JNK pathway.
Furthermore, to explore whether other portions of FRG are involved in the endothelin-induced inhibition of the cell motility, we transfected the variants of FRG, FRG-FERM, FRG-Proline, and FRG-2ndPH into cells. As shown in Fig. 4H, FRG-FERM rescued the inhibition of cell motility induced by endothelin, suggesting the involvement of the FERM region of FRG in the inhibition of the cell motility signaling pathway linking the ETA receptor to JNK.
The Cdc42-GEF Activity of FRG Was Enhanced by the ETA Receptor-To investigate whether the GEF activity of FRG is activated upon the stimulation by endothelin, we measured the GEF activity of the immunoprecipitated FRG by dissociation of [ 3 H]GDP from recombinant RhoA, Rac1, and Cdc42. Upon the stimulation by endothelin, the Cdc42-GEF activity of FRG was specifically enhanced, whereas no change was observed in RhoA and Rac1 (Fig. 5A). Additionally, as shown in Fig. 5B, a Cdc42-GEF activity of FRG to induce the binding of [ 32 P]GTP was promoted by endothelin in a time-dependent manner. These results suggest that stimulation by endothelin enhances the Cdc42-GEF activity of FRG.
Src Kinase Directly Regulates the Activation of the Cdc42-GEF Activity in Cells and in Vitro-As shown in Fig. 3D, we showed that Src kinase is required for the activation of Cdc42 in the pathway downstream of the ETA receptor. Thus, we investigated whether Src kinase activates the Cdc42-GEF activity of FRG in cells. We transfected the plasmid encoding CA-Src, a constitutively activated form of c-Src, with FRG into cells. Transfection of CA-Src with FRG activated the Cdc42-GEF activity of FRG (Fig. 6A). Additionally, the activation of the Cdc42-GEF activity by CA-Src was prevented by pretreatment with PP1.
Next we examined the possibility that Src kinase induces tyrosine phosphorylation of FRG in cells. As shown in Fig. 6A, CA-Src was able to induce tyrosine phosphorylation of FRG in cells. PP1 inhibited the CA-Src-induced tyrosine phosphorylation of FRG in cells. These results suggest that Src kinase contributes to the activation and tyrosine phosphorylation of FRG in cells.
The well known mechanism by which Rho family GEFs are activated is that of Vav family GEFs, which are directly tyrosine-phosphorylated and activated by Src kinase (20). Thus, we next examined whether the immunoprecipitated FRG is directly tyrosine-phosphorylated and activated by recombinant c-Src in vitro. As shown in Fig. 6B, in vitro tyrosine-phosphorylated FRG exhibited and enhanced the Cdc42-GEF activity of FRG. This result suggests that Src kinase directly phosphorylates FRG and activates the Cdc42-GEF activity of FRG in vitro.
Then we investigated whether the endothelin-induced activation of the Cdc42-GEF activity of FRG requires Src kinase. As shown in Fig. 6C, the endothelin-induced activation of the Cdc42-GEF activity of FRG was blocked by pretreatment with PP1 and was inhibited by PP2 (data not shown). Thus, it is likely that Src kinase regulates the activation of the Cdc42-GEF activity of FRG in cells.
Additionally, we examined whether endothelin causes tyrosine phosphorylation of FRG in cells. Fig. 6C indicated that endothelin stimulated the tyrosine phosphorylation of FRG. The endothelin-induced tyrosine phosphorylation was inhib- ited by pretreatment with PP1. This result suggests that Src kinase is necessary for the endothelin-induced tyrosine phosphorylation of FRG in cells.
The N-terminal Region and the DH and PH Domains Are Responsible for the Tyrosine Phosphorylation-To further elucidate which region of FRG is the site for tyrosine phosphorylation, we transfected the plasmid encoding CA-Src with the various truncated mutants of FRG into cells. As shown in Fig.  6D, CA-Src was able to induce the tyrosine phosphorylation of FRG-FERM, FRG-Proline, and FRG-DHPH, whereas FRG-2ndPH was not tyrosine-phosphorylated by Src kinase. Additionally, the tyrosine phosphorylation of FRG-FERM, FRG-Proline, and FRG-DHPH by CA-Src was prevented by pretreatment with PP1. These results suggest that Src kinase may tyrosine phosphorylate the N-terminal region and the DH and PH domains of FRG in cells.
Next we investigated whether the FRG-FERM, FRG-Proline, and FRG-DHPH are directly phosphorylated by recombinant c-Src in vitro. The immunoprecipitated fragments of FRG-FERM, FRG-Proline, and FRG-DHPH were tyrosine-phosphorylated by Src kinase in vitro (Fig. 6E). Taken together with those of Fig. 6D, these results suggest that the FERM, proline-rich, and DH and PH domains of FRG provide the tyrosine phosphorylation sites and/or structural features required for the regulation of Cdc42-GEF activity. These possibilities are under current investigation. DISCUSSION Although there are many Rho family GEFs (9), it remains unclear how these Rho family GEFs are activated by extracellular signals. In the present study, we performed a BLAST search using the DH domain of Dbs, a GEF involved in the GPCR-Rho family GTPase signaling pathway, as a query. We selected one putative GEF, KIAA0793, which has homology with the FGD1-related family. It has been shown that FGD-1 and Frabin, the members of FGD1-related family, are the GEFs for Cdc42 (36,38). Using in vivo and in vitro assays, we demonstrated that KIAA0793 is also a Cdc42-specific GEF. We, therefore, refer to this GEF as FRG (FGD1-related Cdc42-GEF). FRG is phosphorylated and activated by Src kinase, which increases the Cdc42-GEF activity of FRG. Moreover, FRG is involved in the signaling pathway linking the ETA receptor to JNK, leading to the inhibition of cell motility. On the basis of these findings, we summarized the proposed sig- naling pathway in Fig. 7. It is noteworthy that CDEP, which shows a significant homology with FRG, acts as a RhoA-specific GEF (44). It is of interest to examine whether FRG may activate RhoA under a certain condition, such as upon stimulation of receptor other than GPCRs.
GPCRs or G␣␤␥ has been reported to stimulate Rho family GEFs. For example, lysophosphatidic acid or sphingosine-1phosphate activates G13 through their GPCRs (16). The GTPbound G␣13 directly induces the activation of p115RhoGEF, which is a RhoA-specific GEF and belongs to a subfamily of the regulator of G protein signaling domain-containing GEF, such as PDZ-RhoGEF (51,52) and LARG (53)(54)(55)(56). Recently, it has been reported that PDZ-RhoGEF forms a complex with G␣q⅐GTP, which appears to regulate its GEF activity (57). Furthermore, G␤␥ directly activates a novel type of the Rac-GEF P-Rex1 synergistically with phosphatidylinositol 3,4,5triphosphate (58). However, it remains to be investigated whether these Rho family GEFs act downstream of GPCRs/ G␣␤␥ in Src kinase-dependent or -independent manner.
Vav family GEFs, Vav, Vav2, and Vav3, are representative of Rho family GEFs that are directly tyrosine-phosphorylated and activated by tyrosine kinases of the Src family (20 -23). However, it remains unknown whether Vav family GEFs are regulated generally downstream of GPCRs/G␣␤␥. The activation of Vav by the N-formyl peptide receptor, a GPCR expressed in neutrophils and monocytes, has been reported. Ma et al. (23) show that G␤␥ and phosphoinositide 3-kinase ␥ activate the Rac1-GEF activity of Vav in a manner apparently independent of the tyrosine kinase. As far as we know, this is the first report of the Cdc42-specific GEF, which is directly phosphorylated and activated by Src kinase in the signaling pathway downstream of GPCR.
The DH and PH domains of FRG were tyrosine-phosphorylated by Src kinase in vivo and in vitro (Fig. 6, D-E). Structure analysis revealed that the DH domain of Vav is autoinhibited by its N-terminal region and that the Tyr-174 in this region is phosphorylated by Src family kinase (21). Phosphorylation or deletion of this region results in the stimulation of GEF activity (22)(23). Thus, tyrosine phosphorylation relieves autoinhibition by opening the Rho family GTPase interaction surface of the DH domain (21)(22)(23). It is conceivable that tyrosine phosphorylation of the DH and PH domains of FRG may also play an essential role in the activation of GEF activity of FRG. However, the FERM and proline-rich domains of FRG were also tyrosine-phosphorylated by Src kinase (Fig. 6, D-E). Therefore, these regions are likely to provide the tyrosine phosphorylation sites required for the stimulation of the exchange activity of FRG. Further study should help reveal details of the regulatory mechanism of FRG by Src kinase.
Kubo et al. (59) recently characterized KIAA0793 as a signaling molecule of neurite remodeling. They showed that KIAA0793 is a Rac1-specific GEF, whereas its activation mechanism has not been described. It seems contradictory that we showed KIAA0793 as the Cdc42-specific GEF. This may be because of the difference of the signaling pathway and/or conditions of the experiments. It will be interesting to examine whether FRG is activated by Src kinase in the signaling pathway of neurite remodeling.
In this study, we identified a new signaling molecule, FRG, which directly links between Src kinase and Rho family GTPase Cdc42 in the GPCR/JNK signaling pathway. Recently, it was indicated that tyrosine kinases are involved in the G protein-induced activation of Rho family GEFs (60,61). Further study is necessary to clarify how Src kinase regulates Rho family GEFs in the pathway linking GPCRs to Rho family GTPases. Additionally, such studies might promote our understanding of the general mechanism by which Src kinase regulates Rho family GEFs and thereby help to elucidate the cellular function of the pathway.