Regulation of Ephexin1, a Guanine Nucleotide Exchange Factor of Rho Family GTPases, by Fibroblast Growth Factor Receptor-mediated Tyrosine Phosphorylation*

Fibroblast growth factor (FGF) signal is implicated in not only cell proliferation, but cell migration and morphological changes. Several different Rho family GTPases downstream of the Ras/ERK pathway are postulated to mediate the latter functions. However, none have been recognized to be directly coupled to FGF receptors (FGFRs). We have previously reported that EphA4 and FGFRs hetero-oligomerize through their cytoplasmic domains, trans-activate each other, and transduce a signal for cell proliferation through a docking protein, FRS2α (Yokote, H., Fujita, K., Jing, X., Sawada, T., Liang, S., Yao, L., Yan, X., Zhang, Y., Schlessinger, J., and Sakaguchi, K. (2005) Proc. Natl. Acad. Sci. U. S. A. 102, 18866-18871). Here, we have found that ephexin1, a guanine nucleotide exchange factor for Rho family GTPases, constitutes another downstream component of the receptor complex. Ephexin1 directly binds to the kinase domain of FGFR mainly through its DH and PH domains. The binding appears to become weaker and limited to the DH domain when FGFRs become activated. FGFR-mediated phosphorylation of ephexin1 enhances the guanine nucleotide exchange activity toward RhoA without affecting the activity to Rac1 or Cdc42. The FGFR-mediated tyrosine phosphorylation includes, but is not limited to, the residue (Tyr-87) phosphorylated by Src family kinase, which is known to be activated following EphA4 activation. The Tyr-to-Asp mutations that mimic the tyrosine phosphorylation in some of the putative FGFR-mediated phosphorylation sites increase the nucleotide exchange activity for RhoA without changing the activity for Rac1 or Cdc42. From these results, we conclude that ephexin1 is located immediately downstream of the EphA4-FGFR complex and the function is altered by the FGFR-mediated tyrosine phosphorylation at multiple sites.

Fibroblast growth factors (FGFs) 2 have been implicated in diverse cellular processes, including apoptosis, cell survival, chemotaxis, cell adhesion, migration, differentiation, and proliferation. Currently, FGFs comprise a structurally related family of 22 molecules (FGF 1-23). The FGF ligands and the four signaling FGF receptors (FGFRs), including their alternatively spliced variants, are expressed in specific spatial and temporal patterns during the embryonic development. FGFRs contain an extracellular ligand binding domain, a single transmembrane domain, and two intracellular tyrosine kinase domains. The extracellular domain contains two or three immunoglobulinlike domains. Alternative mRNA splicing of the C-terminal portion of the third immunoglobulin-like domain creates several isoforms of FGFRs with unique ligand-binding properties. Once an FGF ligand is bound, the receptor dimerizes and phosphorylates intramolecular tyrosine residues, triggering initiation of FGFR signal transduction (1,2).
We have recently reported that EphA4 and FGFRs form a complex through their cytoplasmic interactions and trans-activate each other (3). EphA4 is a member of the Eph receptor family, the largest class of receptor tyrosine kinases with 13 members. Eph receptor family can be divided into two subfamilies (EphA and EphB) based on their structural properties and ligand binding preferences. The Ephs and their ligands, ephrins, are broadly expressed throughout the developing embryo, including the nervous system. The best characterized role of Eph signaling involves the guidance of axons during neural development. The control of axon guidance by Eph receptors is likely mediated by controlling cytoskeletal function through regulation of the Rho family GTPases: RhoA, Rac1, and Cdc42 (4 -6).
Ephexin1, a Dbl family of guanine nucleotide exchange factor (GEF) for Rho family GTPases, was identified to couple EphA receptor activation to modulation of Rho family GTPase signaling. Ephexin1 is highly expressed in the central nervous system during development and is enriched in the neuronal growth cones (7). Studies in ephexin1 knock-out mice show that ephexin1 is required for axonal outgrowth as well as for growth cone repulsion or collapse (8). In the absence of ephrin stimu-lation, ephexin1 activates RhoA, Rac1, and Cdc42, thus leading to a balanced GTPase activation that promotes axonal outgrowth. In contrast, stimulation of Ephs by ephrin induces Src family kinases (SFK)-dependent phosphorylation of ephexin1 on an evolutionarily conserved tyrosine (Tyr-87). This tyrosine phosphorylation of ephexin1 enhances the ephexin1's GDP/ GTP exchange activity specifically toward RhoA relative to Rac1 and Cdc42, thereby changing the local balance of Rho GTPases activity within cells, leading to actin cytoskeletal changes that result in growth cone collapse (8). As ephexin1 activates Rho-family GTPases that are implicated in both attraction and repulsion, the guanine nucleotide exchange activity of ephexin1 might be controlled through interactions with additional proteins. Ephexin1 may function in combination with other factors to stimulate Rac1 or Cdc42, whereas in a complex with active Eph receptors it preferentially activates RhoA (9).
In this study, we report a physical interaction between ephexin1 and FGFRs. When ephexin1 was co-expressed with FGFRs in HEK293T cells, ephexin1 was co-immunoprecipitated with FGFRs and became tyrosine-phosphorylated by the activated receptors. In vitro binding studies using a series of deletion mutants of ephexin1 revealed that ephexin1 binds to the cytoplasmic domain of activated FGFR2 mainly through its DH domain. Similar binding studies using a series of FGFR2 deletion mutants showed that the second kinase domain of FGFR2 is the preferential binding site for ephexin1. We further examined the FGFR-induced tyrosine phosphorylation sites of ephexin1 through screening ephexin1 mutants that carry Tyrto-Phe changes in several different loci. FGFR-induced tyrosine phosphorylation of exphexin1 appears to include multiple tyrosine residues. These results suggest that FGFRs are involved in the regulation of ephexin1 activity, which in turn regulates cytoskeletal reorganization at growth cones.

EXPERIMENTAL PROCEDURES
Reagents-The following are the antibodies used in the current study: mouse anti-HA monoclonal antibody 12CA5 (Roche Applied Science); mouse anti-phosphotyrosine antibody, clone 4G10 (Upstate); rabbit anti-FGFR2 polyclonal antibody (Santa Cruz Biotechnology); mouse anti-Xpress monoclonal antibody (Invitrogen); rabbit anti-phospho-Src family (Tyr416) polyclonal antibody (Cell Signaling); rabbit anti-c-Src (SRC2) polyclonal antibody (Santa Cruz Biotechnology, sc-18). A rabbit polyclonal antibody against ephexin1 was raised against the N-terminal region of ephexin1 (1-373 amino acids) in our laboratory. Unless indicated otherwise, all other materials were purchased from Sigma-Aldrich.
Cell Culture-HEK293T cells were maintained in Dulbecco's minimal essential medium (DMEM, Sigma) with 10% fetal bovine serum (Equitech-Bio) supplemented with 1% penicillin and streptomycin (Amersham Biosciences) at 37°C and 5% CO 2 . Rat L6 myoblasts were maintained in the same condition. PC12 cells were maintained in RPMI1640 (Sigma) containing 10% horse serum (Amersham Biosciences) and 5% fetal bovine serum supplemented with 1% penicillin and streptomycin. For morphological examination, culture plates were treated with 2 g/ml fibronectin at 37°C for 1 h, and washed with phosphatebuffered saline before plating cells.
For treatment with growth factors and other chemicals, HEK293T cells, 48 h after transfection, were serum-deprived for 5 h and then incubated with SU5402 or PP2 (Calbiochem) at indicated concentrations for 2 h at 37°C in 5% CO 2 prior to FGF2 stimulation. L6 cells stably expressing FGFR2 and/or ephexin1 were kept in DMEM containing 0.5% fetal bovine serum for more than 16 h for serum starvation and incubated with FGF2 or kinase inhibitors. Ephrin-A1-Fc was used after oligomerization as described previously (3).
DNA Constructs and Mutagenesis-The wild type (WT) and mutants of EphA4 and FGFRs were constructed as reported (3). Full-length cDNA of mouse ephexin1 was prepared by reverse transcription-PCR using total RNA from a mouse brain as template. They were subsequently subcloned into pCR-BluntII-Topo vector (Invitrogen). Ephexin1 mutants were prepared by applying the recombinant PCR method or site-directed mutagenesis using WT constructs as templates and a pair of appropriate mutation primers. cDNAs for Rho family GTPases were derived from the constructs previously reported (10). The sequences of all of the PCR-amplified DNAs were confirmed by sequencing after cloning into a pCR-BluntII-TOPO cloning vector according to the manufacturer's instructions (Invitrogen). For eukaryotic transient expression, all of the DNA constructs were incorporated into a pcDNA3.1 plasmid vector, into which a Myc, FLAG, or hemagglutinin (HA) epitope-encoding sequence was integrated at the 3Ј side in-frame with the coding sequences of the incorporated cDNAs. For bacterial expression, the full-length and mutant cDNAs were incorporated into a pMalC2 vector (New England Biolabs), which produces a protein N-terminally fused with maltose-binding protein (MBP), into a pGEX4T (Amersham Biosciences) expression vector that produces a protein N-terminally fused with glutathione S-transferase (GST), or a pBAD/His vector (Invitrogen) or a pTrcHisA (Invitrogen), which produces a protein fused with an Xpress epitope and six His residues at the N terminus.
Expression and Purification of Recombinant Proteins-Fusion proteins were expressed in TOP10 bacteria and purified by amylose or glutathione affinity column chromatography for MBP-fused or GST-fused proteins, respectively, according to the manufacturers' instructions. For preparation of His 6 -tagged recombinant proteins, the proteins expressed in TOP10 bacteria were purified using the BD-Talon cobalt-based affinity chromatography resin (BD Biosciences) according to the manufacturer's instruction. Eluted fractions were collected and dialyzed in a buffer containing 50 mM HEPES and 50 mM NaCl that was kept at pH 7.5. Protein concentrations were determined using the SDS-PAGE followed by Coomassie Brilliant Blue staining utilizing bovine serum albumin as a standard.
In Vitro Protein-Protein Interaction Assay-To determine the interaction between the ephexin1 deletion mutants and the FGFR cytoplasmic domain in vitro, 2 g of MBP-tagged ephexin1 deletion mutants were preloaded to amylose agarose at 4°C for 60 min. The Xpress-tagged FGFR cytoplasmic domain was then added, and the mixture was incubated at 4°C overnight. Finally beads were washed three times with a wash buffer (50 mM HEPES, pH 7.5, and 100 mM NaCl), suspended in 2ϫ SDS-PAGE sample buffer, boiled for 5 min, and fractionated using 10% SDS-PAGE. The protein was transferred to a polyvinylidene fluoride membrane and immunoblotted with an anti-Xpress antibody. In the binding assay between the DH plus PH domains of ephexin1 and the deletion mutants of the FGFR cytoplasmic domain, ephexin1 and FGFR molecules were tagged with Xpress and MBP, respectively.
Transfection, Immunoprecipitation, and Immunoblotting-Transient transfection with the various expression constructs was carried out using FuGENE 6 Transfection Reagent (Roche Applied Science). The total amount of DNA in each transfection was normalized with the pcDNA3.1 plasmid. Cells were harvested 48 h after transfection. The lysis buffer contained 50 mM HEPES buffer, 1% Triton X-100, 5 mM EDTA, 50 mM sodium chloride, 10 mM sodium pyrophosphate, 50 mM sodium fluoride, 1 mM sodium orthovanadate, and protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 M aprotinin, 1 M leupeptin, 1 M pepstatin A). In some experiments, cell lysates (15-50 g) were fractionated directly using SDS-PAGE followed by immunoblotting with an antibody. In other experiments cell lysates (150 -400 g) were used for immunoprecipitation with an appropriate antibody overnight at 4°C. The immunoprecipitates were collected by adding protein A-Sepharose, washed three times with lysis buffer, fractionated by SDS-PAGE, and immunoblotted with an antibody. Signals were detected using the enhanced chemiluminescence (ECL) reagent (Amersham Biosciences) according to the manufacturer's instruction. In some cases, membranes were stripped and re-probed with different antibodies.
In vitro guanine nucleotide exchange assays were performed as described elsewhere (10,11). GST-fused small GTPase proteins (GST-RhoA, GST-Rac1, and GST-Cdc42) were synthesized as recombinant proteins in bacteria using the same method as for FGFR and ephexin1 as describe above. The small GTPase proteins were first loaded with [ 3 H]GDP for 30 min at 30°C in a loading buffer (20 mM Tris-HCl, pH 7.5, 50 mM NaCl, 3 mM MgCl 2 , 0.1 mM dithiothreitol, 0.1 mM EDTA). MgCl 2 was then added to a final concentration of 25 mM, and the mixture was placed on ice. To start an exchange assay, 20 g of cell lysate and 3 g of [ 3 H]GDP-bound small GTPase were incubated in a 150-l final volume of exchange buffer (20 mM Tris-HCl, pH 7.5, 80 mM NaCl, 0.1 mM dithiothreitol, 0.4 mg/ml bovine serum albumin, 5 mM MgCl 2 , 1 mM GTP) at 30°C. At the indicated time points, the reactions were stopped by adding 1 ml of ice-cold termination buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 25 mM MgCl 2 ) and kept on the ice. Samples were filtered through a nitrocellulose filter (Millipore) and washed three times with the termination buffer, then dissolved in 10 ml of Filtron-X scintillation mixture (National Diagnostics). Radioactivity of bound [ 3 H]GDP was measured using a liquid scintillation counter. For data presentation, the amount of [ 3 H]GDP retained on the filter at 0 time point was taken as 1.
Retroviral Vector Construction-For making cells stably coexpressing ephexin1 with FGFR2 or EphA4, their cDNAs were subcloned into pMXs-IG vector (obtained from T. Kitamura, University of Tokyo, Tokyo) linked to EGFP through the internal ribosomal entry site sequence (12). Virus was produced as described previously (3), and viral transduction was carried out at a multiplicity of infection of 5 or more. The transduced cells were identified easily by visualization of the EGFP fluorescence.

FGF Receptor Is Involved in the Phosphorylation of Ephexin1-
Ephexin1 is known to bind to EphA4, and the function of the GEF becomes modulated by EphA4 through SFK-mediated phosphorylation (7,8,13). We have confirmed that the ephexin1 binds to EphA4 and is phosphorylated in an EphA4 dose-dependent fashion (Fig. 1A). Because we have recently reported that EphA4 and FGFR interact with and cross-phosphorylate each other (3), we initiated studies to investigate the function of FGFR in regulating the ephexin1 function. When EphA4 was overexpressed and stimulated by its ligand, ephrin-A1, ephexin1 was phosphorylated over the basal level. Addition of SU5402, an FGFR kinase inhibitor, inhibited the phosphorylation levels of both the overexpressed EphA4 and the endoge- OCTOBER 19, 2007 • VOLUME 282 • NUMBER 42 nous FGFR2 and reduced the level of ephexin1 phosphorylation (Fig. 1B). SU5402 did not affect the phosphorylation levels of SFK at Tyr-416, which is regarded as the critical phosphorylation site for SFK activation. These findings suggest that FGFR might also be involved in the regulation of ephexin1 by tyrosine phosphorylation and that the EphA4-mediated ephexin1 phosphorylation by SFK (8,13) is not the only modulator of ephexin1's guanine nucleotide exchange activity.

FGFR-mediated Ephexin1 Regulation
To evaluate the interaction of four members of FGFRs with ephexin1, we overexpressed both molecules in HEK293T cells and examined their binding and ephexin1 phosphorylation using immunoprecipitation followed by immunoblotting ( Fig.  2A). Ephexin1 bound to all the four FGFRs, whereas it was phosphorylated only by the activated FGFRs (FGFR1 and FGFR2) with FGFR2 having the highest phosphorylation effect on ephexin1. Furthermore, the overexpressed FGFR2 phosphorylated the exogenous ephexin1 in an FGFR2 dose-dependent manner. The phosphorylated portion of ephexin1 that was coimmunoprecipitated with endogenous EphA4 was almost undetectable, whereas non-phosphorylated ephexin1 was clearly co-immunoprecipitated (Fig. 2B), suggesting that FGFR2 is the driving force of ephexin1 phosphorylation in the presence of FGFR2, EphA4, and ephexin1. When FGFR2 was overexpressed in L6 cells, which barely express intrinsic FGFRs (3,14), together with ephexin1, ephexin1 was co-immunoprecipitated with FGFR2 and was phosphorylated with a peak at 60 min or later following stimulation with FGF2 (Fig. 3, A and B). Therefore, we used the time point between 30 and 60 min for evaluation of the ligand-mediated ephexin1 phosphorylation.
We further examined in HEK293T cells the effect of FGFR2 on phosphorylation of ephexin1 in the presence and absence of SU5402 following overexpression of both molecules and stimulation with FGF2. SU5402 effectively inhibited phosphorylation of FGFR2, and almost completely suppressed phosphorylation of ephexin1 at concentrations as low as 5 M. SFK phosphorylation at Tyr-416 was not altered (Fig. 4A). Because SFK is reported to be involved in the phosphorylation of ephexin1 when stimulated by ephrin (8,13), we used an SFK inhibitor PP2 and its non-functional control chemical PP3 in place of SU5402 to study the effect of SFK in the FGFR2-mediated phosphorylation of ephexin1. SFK phosphorylation was slightly inhibited by PP2 at 2.5 M or higher concentrations. PP2 barely inhibited the phosphorylation of ephexin1 at concentrations up to 2.5 M, whereas 2.5 M PP2 clearly inhibited phosphorylation of FGFR2 and PP2 at 5 M further inhibited phosphorylation of both FGFR2 and ephexin1 (Fig. 4B). PP2 at 10 M inhibited phosphorylation of overexpressed FGFR2 (without FGF2 stimulation) and completely suppressed phosphorylation of ephexin1 in HEK293T cells, whereas the phosphorylation level of SFK Tyr-416 was only slightly reduced (Fig.  4C). Taken together, FGFR2 phosphorylates ephexin1 independently of SFK activation, which is reported to be required for the effect of EphA4 on ephexin1 (8,13).
To clarify the function of PP2 on ephexin1 phosphorylation in cells other than HEK293T cells, we used L6 cells stably expressing exogenous FGFR2 and ephexin1. Under the presence of PP2 or PP3 and a phosphatase inhibitor, orthovanadate (see Fig. 9A), cells were stimulated with FGF2 (100 ng/ml) for 1 h. PP2 clearly inhibited the phosphorylation levels of both FGFR2 and ephexin1 simultaneously in a dose-dependent fashion at concentrations between 0.1 and 20 M (only the 2.5-20 M range is shown in the Fig. 5A). On the other hand, the phosphorylation level of SFK at Tyr-416 was suppressed to a certain level by low concentrations of PP2 and stayed at the constant low level (Fig. 5A). PP3 did not change the phosphorylation levels of ephexin1, SFK, or FGFR2. These findings suggest that PP2 is a potent kinase inhibitor of both FGFR2 and SFK and that the ligand-activated FGFR2 is directly involved in the phosphorylation of ephexin1.
Previous studies showed that the function of EphA4 kinase in the phosphorylation of ephexin1 is to bring ephexin1 close to SFK so that SFK becomes able to phosphorylate the RhoGEF (8,13). Both SFK and ephexin1 are known to bind to EphA4. To study whether EphA4 can directly phosphorylate ephexin1 in the absence of FGFRs, we used L6 cells expressing exogenous ephexin1 (L6/ephexin1-HA). L6 cells are known to express endogenous EphA4 (3) but do not express detectable amounts of ephexin1 or FGFRs. Exposure of L6/ephexin1-HA to 0.5 g/ml oligomerized ephrin-A1 in the presence and absence of orthovanadate or PP2 activated EphA4 as shown in Fig. 5B, whereas any such treatments did not phosphorylate ephexin1. These findings suggest that the ligand-mediated EphA4 activation, which is reported to be followed by SFK activation in other systems, does not induce ephexin1 phosphorylation in the absence of FGFRs, supporting the notion that FGF receptors play a major role in the phosphorylation of ephexin1.
Determination of Binding Domains of Ephexin1 and FGFR-We next studied the interaction between FGFR2 and ephexin1 in an in vitro binding assay. The binding site of ephexin1 to FGFR2 was determined to be the second kinase domain by studying the binding of Xpresstagged ephexin1(DH plus PH) (the domain containing DH and PH) with various MBP-tagged deletion mutants of the FGFR2 cytoplasmic domain (Fig. 6A). The region of ephexin1 responsible for binding to FGFR2 was determined by examining the binding of MBP-tagged ephexin1 deletion mutants to the Xpress-tagged cytoplasmic domain of the WT and the kinase-defective (KD) mutant (K517M) of FGFR2. Ephexin1 bound to FGFR2(KD) via the N-terminal DH domain and the PH domain and to FGFR2(WT) via a confined region of the N-terminal DH domain (Fig. 6B). We also examined FGFR3 in place of FGFR2    PP2 (B and C), an SFK inhibitor. Ephexin1 was co-transfected with FGFR2 in HEK293T cell. A and B, cells were pretreated with various concentrations of SU5402 or PP2 for 2 h, and exposed to FGF2 (100 ng/ml) for 30 min in the presence of the same inhibitor. C, cells were treated with or without PP2 (10 M) for 30 min without stimulation with FGF2. N,N-Dimethylformamide and PP3 were used as control reagents for SU5402 and PP2, respectively. Cell lysates were immunoprecipitated (IP) and immunoblotted (IB) with the indicated antibodies. SFK is an endogenous molecule in HEK293T cells. OCTOBER 19, 2007 • VOLUME 282 • NUMBER 42 using the similar binding studies. FGFR3 bound to ephexin1 through its mid-cytoplasmic region encompassing the first and second kinase domains (Fig. 7A). Ephexin1 bound to the cytoplasmic domain of FGFR3(WT) and FGFR3(KD) (N540K) through a rather broad area comprising the C-terminal DH domain, PH domain, and SH3 domain, whereas it bound to a kinase-activated (KA) mutant (K650E) of FGFR3 through a confined N-terminal DH domain area (Fig. 7B). Taken together, these findings suggest that ephexin1 binds to the cytoplasmic domain of FGFR and that the kinase-activated FGFR causes ephexin1 to make a conformation that allows it to bind less firmly to the active FGFR probably due to phosphorylation of the ephexin1 molecule.

FGFR-mediated Ephexin1 Regulation
EphA4 is reported to induce phosphorylation of one of the tyrosines (Tyr-87) mutated in M1 through activation of SFK, which causes ephexin1 to increase its guanine nucleotide exchange activity toward RhoA without changing the activity to Rac1 or Cdc42 (8). To study which tyrosines are phosphorylated by FGFR, we constructed ephexin1 mutants (M1-M9) in a pcDNA vector that carry Tyr-to-Phe mutations along the ephexin1 molecule as described in Fig. 8A. Co-expression of these mutants with FGFR2 revealed that some mutants had a low level of phosphorylation as compared with the WT: M1, M4, M6, M7, M8, and M9 had reduced phosphorylation (Fig. 8B). These findings suggest that the tyrosine residues replaced with phenylalanine in M1, M4, M6, M7, M8, and M9 are the targets of FGFR kinase and that the function of ephexin1 phosphorylated by FGFR might be changed as in the case of phosphorylation by SFK.
Ephexin1 Phosphorylated by FGFR Enhances the Guanine Nucleotide Exchange Activity on RhoA but Not on Rac1 or Cdc42-We then examined the function of ephexin1 phosphorylated by FGFR in L6 cells that stably express exogenous FGFR2 and ephexin1 (L6/FGF2 plus ephexin1) (Fig. 9). Ephexin1 phosphorylation levels were first examined using immunoblotting of the cell lysates derived from L6 cells treated differently (Fig. 9A). Because FGFR-mediated phosphorylation of ephexin1 in L6 cells was difficult to detect, we also prepared cells incubated with sodium orthovanadate, a phosphatase inhibitor, together with FGF2 for 60 min. L6 cells expressing ephexin1 alone did not show any phosphorylation of ephexin1 even in the presence of FGF2 (data not shown), and L6/FGFR2 plus ephexin1 showed barely detectable phosphorylation in the absence of FGF2. L6/FGFR2 plus ephexin1 cells showed evident ephexin1 phosphorylation after stimulation by FGF2 and enhanced phosphorylation in the presence of FGF2 and orthovanadate. The presence of orthovanadate alone moderately increased the ephexin1 phosphorylation level. SFK activity did not change with these treatments. These findings suggest that L6 cells contain a strong phosphatase system that reverses the  FGFR-dependent phosphorylation of ephexin1. A, SFK-independent ephexin1 phosphorylation following the ligand-mediated FGFR2 activation. L6 cells stably expressing FGFR2 and ephexin1 were starved in DMEM containing 0.5% serum for 20 h. Cells were pretreated with serial dilutions of PP2 or its control reagent, PP3, for 2 h, and stimulated with FGF2 (100 ng/ml) for 1 h in the presence of 0.2 mM sodium orthovanadate. PP2 or PP3 was incubated with the cells throughout the stimulation. Cell lysates were immunoprecipitated and immunoblotted with the indicated antibodies. SFK is the endogenous molecule expressed in L6 cells. Control cells were not treated with FGF2, PP2, or PP3. B, no EphA4-dependent ephexin1 phosphorylation following treatment with ephrin-A1. L6 cells stably expressing ephexin1-HA were serum-starved as described above. Cells were pre-treated with 0.5 M PP2 for 2 h and treated with 0.5 g/ml ephrin-A1 for 1 h in the presence or absence of 0.2 mM sodium orthovanadate. Cell lysates were immunoblotted with the indicated antibodies following immunoprecipitation with the antibodies shown. EphA4 is the endogenous molecule expressed in L6 cells.

FIGURE 6. Mapping of the interaction domains between FGFR2 and ephexin1 in an in vitro binding assay.
A, the ephexin1 interaction region in FGFR2. MBP-fused FGFR2 deletion mutants were synthesized as recombinant proteins, and their binding with ephexin1 (DH plus PH domains) was examined in vitro. The major ephexin1-binding region is marked by a rectangle. B, the FGFR2 binding region in ephexin1. MBP-fused ephexin1 deletion mutants were synthesized, and their binding with the kinase-defective (KD) FGFR2 mutant (FGFR2(K517M)) or FGFR2(WT) was examined in vitro. The major binding region to FGFR2(KD) is marked by clear rectangles, and the binding region to FGFR2(WT) is marked by a dark rectangle. One-fortieth of the amount of the recombinant peptides used for the binding study was fractionated using a separate gel and stained with Coomassie Brilliant Blue to show that the peptides were used in a similar amount. Stained bands of different molecular masses were put in alignment for easy comparison. Amino acid (aa) numbers of the fragments' both ends, which are counted from methionine encoded by the initiation codon, are noted following the names of the fragments. ephexin1 phosphorylation and that SFK is not involved in the ephexin1 phosphorylation by FGFR. These results are consistent with the findings described above (Fig. 5).
Using the cell lysates taken from L6/FGFR2 plus ephexin1 treated as above, we examined the GDP/GTP exchange activity on Rho family of GTPases. Cell lysates prepared from the cells incubated with FGF2 alone or with both FGF2 and orthovanadate showed increased GDP/GTP exchange activity on RhoA, and treatment with orthovanadate alone induced similar activity (Fig. 9, B-D). The phosphorylated ephexin1 did not exhibit any change in the GDP/GTP exchange activity on either Rac1 or Cdc42 as compared with the non-phosphorylated ephexin1. To examine whether the differential activation of guanine nucleotide exchange activity is caused by ephexin1 phosphorylation, we also studied the exchange activity in L6 cells expressing FGFR2 alone (L6/ FGFR2). Treatment of L6/FGFR2 with FGF2 in the presence or absence of orthovanadate did not induce any activation of Rho family GTPases as compared with the nontreated control, suggesting that the RhoA activation in L6/FGFR2 plus ephexin1 is attributed to the effect on ephexin1.
To clarify the function of ephexin1 phosphorylated at specific tyrosines, we constructed mutants simulating the phosphorylation by replacing tyrosine residues with aspartic acid as reported (15) at loci 1, 4, and 8 of Fig. 8A and designated them A1, A4, and A8. These phosphorylation-simulating mutants were expressed in L6 cells at a similar level (Fig. 10A), and the GDP/ GTP exchange activities on Rho family GTPases were examined using the same protein amount of cell lysates (Fig. 10, B-D). Cell lysates from L6 cells expressing A1 and A4 had an increased GDP/GTP exchange activity on RhoA as compared with those from the cells expressing WT or A8. The same assay using these lysates on Rac1 or Cdc42 did not show any significant changes over the control. We then expressed these ephexin1 mutants in PC12 cells, which naturally extend neurites using a retrovirus vector co-expressing EGFP (Fig.  10E). Mutants A1 and A4 reduced the neurite extension. Especially the cells expressing A1 became flattened and round without any neurites. Expression of EGFP alone, WT, or A8 did not cause any change in the cell morphology. These findings strongly support the results of the nucleotide exchange assay, wherein the FGFR-mediated phosphorylation of some tyrosines augments the guanine nucleotide exchange activity of ephexin1 on RhoA without affecting the activity on Cdc42 or Rac1.

DISCUSSION
We have shown here that ephexin1 is phosphorylated by FGFR in addition to the reported SFK-mediated phosphoryla- A, the ephexin1 interaction region in FGFR3. The binding of ephexin1 (DH plus PH domains) to MBP-fused FGFR3 deletion mutants was examined as described in Fig. 6A. B, the FGFR3 binding region in ephexin1. The binding of FGFR3(WT), FGFR3(KD) (N540K), and a kinase-active (KA) (K650E) mutant of FGFR3 to the deletion mutants of ephexin1 was examined as described in Fig. 6B. The major binding region to FGFR3(WT) and FGFR3(KD) is marked by clear rectangles, and the binding region to FGFR3(KA) is marked by a dark rectangle. One-fortieth of the amount of the recombinant peptides used for the binding study was fractionated using a separate gel and stained with Coomassie Brilliant Blue to show that the peptides were used in a similar amount. Stained bands of different molecular masses were put in alignment for easy comparison.  OCTOBER 19, 2007 • VOLUME 282 • NUMBER 42 tion that is induced by EphA4 activation (8,13). The current finding is consistent with our previous report that EphA4 and FGFR hetero-oligomerize and cross-phosphorylate each other through direct interactions between their cytoplasmic domains (3). We previously focused on the activation of FRS2␣-mediated signal transductions. In this report we have shown that a GEF for Rho family GTPases, ephexin1, is immediately downstream of the hetero-oligomer consisting of FGFR and EphA4. Non-phosphorylated ephexin1 has a balanced guanine nucleotide exchange activity toward all the members of Rho family GTPases as reported (8). The FGFR-phosphorylated ephexin1, on the other hand, has an enhanced guanine nucleotide exchange activity toward RhoA, which is similar to the reported function of the EphA4-dependent SFK-phosphorylated ephexin1. SFK is not likely to be involved in the FGFR-mediated phosphorylation. PP2, an inhibitor of SFK, surprisingly suppressed FGFR phosphorylation substantially as compared with the moderate suppression of SFK phosphorylation at Tyr-416. This might be a cell-type-specific phenomenon, but we obtained similar results in both L6 and HEK293T cells. SU5402, a specific FGFR inhibitor, showed a drastic suppression of ephexin1 phosphorylation together with a clear suppression of the ligand-induced FGFR2 phosphorylation. These findings suggest that SFK might be a predominant kinase that phosphorylates ephexin1 in cells with no FGFR expression. However, in the cells that co-express ephexin1 and FGFR, FGFR is the major player in ephexin1 phosphorylation.

FGFR-mediated Ephexin1 Regulation
FGFR has long been implicated in the extension of neurites and axons, formation of growth cones, and target recognition. Studies in neuronal cells of the developing Xenopus visual system indicate that FGFR plays a role in axon extension and target recognition of retinal ganglion cells (16,17). In addition, posterior protrusion of the early Xenopus embryo appears to be induced by EphA4 activation through FGF signaling (18). These findings suggest the presence of signal transduction pathways leading to activation of Rho family GTPases in the downstream of FGF signal. One of the GEFs located downstream of FGFR in PC12 cells is p85 ␤PIX, which has a specificity for Rac1. p85 ␤PIX is activated after phosphorylation at Ser and Thr residues by p21-activated kinase, which is located downstream of the Ras/ERK pathway (19,20). We have shown here for the first time that a Rho family GEF, ephexin1, directly binds to FGFR and is functionally modulated by FGFR-mediated tyrosine phosphorylation.
The region of FGFR binding to ephexin1 was a rather broad area encompassing the kinase domains. On the other hand, the region of ephexin1 that binds to the kinase-inactive FGFR appears to include DH, PH, and SH3 domains, whereas the kinase-active cytoplasmic domain of FGFR binds weakly to a restricted N-terminal DH domain. As such, the way of ephexin1 binding to FGFR is different from that to EphA4, in which ephexin1 constitutively binds regardless of the activation state of EphA4. Furthermore, the ephexin1 phosphorylation appears to be transient, probably due to the presence of phosphatases. These findings might have important implications that FIGURE 9. Effect of the FGFR-mediated phosphorylation of ephexin1 on the guanine nucleotide exchange activity. A, L6 cells expressing ephexin1 together with FGFR2 were starved in DMEM containing 0.5% fetal bovine serum for 20 h, then stimulated with 100 ng/ml of FGF2 for 60 min in the presence or absence of 0.2 mM sodium orthovanadate. Cell lysates were fractionated using SDS-PAGE with or without immunoprecipitation, and analyzed by immunoblotting. B-D, in vitro GDP/GTP exchange assay using cell lysates from L6 cells expressing FGF2 alone (right panel) or expressing both FGFR2 and ephexin1(left panel). Cells were treated with FGF2 and/or sodium orthovanadate. The GDP/GTP exchange assay was performed on GST-RhoA, GST-Rac1, and GST-Cdc42 as described under "Experimental Procedures." *, significantly different (p Ͻ 0.01) from the value at the 0 time point, n ϭ 4 (Student's t test). Cont, control; F, FGF2; and V, orthovanadate.
as we reported earlier (3) but in migration and morphological changes.