Vinexin Forms a Signaling Complex with Sos and Modulates Epidermal Growth Factor-induced c-Jun N-terminal Kinase/Stress-activated Protein Kinase Activities*

Vinexin, a novel protein that plays a key role in cell spreading and cytoskeletal organization, contains three SH3 domains and binds to vinculin through its first and second SH3 domains. We show here that the third SH3 domain binds to Sos, a guanine nucleotide exchange factor for Ras and Rac, both in vitroand in vivo. Point mutations in the third SH3 domain abolished the vinexin-Sos interaction. Stimulation of NIH/3T3 cells with serum, epidermal growth factor (EGF), or platelet-derived growth factor (PDGF) decreased the electrophoretic mobility of Sos and concomitantly inhibited formation of the vinexin-Sos complex. Phosphatase treatment of lysates restored the binding of Sos to vinexin, suggesting that signaling from serum, EGF, or PDGF regulates the vinexin-Sos complex through the Sos phosphorylation. To evaluate the function of vinexin downstream of growth factors, we examined the effects of wild-type and mutant vinexin expression on extracellular signal-regulated kinase (Erk) and c-Jun N-terminal kinase/stress-activated protein kinase (JNK/SAPK) activation in response to EGF. Exogenous expression of vinexin β in NIH/3T3 cells enhanced JNK/SAPK activation but did not affect Erk activation. Moreover mutations in the third SH3 domain abolished EGF activation of JNK/SAPK in a dominant-negative fashion, whereas they slightly stimulated Erk. Together these results suggest that vinexin can selectively modulate EGF-induced signal transduction pathways leading to JNK/SAPK kinase activation.

Growth factor receptors can associate with integrins (4,5), and signaling molecules involved in growth factor-mediated pathways have also been shown to localize at or be associated with focal adhesions. For example, receptors for basic fibroblast growth factor, platelet-derived growth factor (PDGF), 1 and epidermal growth factor (EGF) as well as downstream molecules including Sos, Raf, mitogen-activated protein kinase/extracellular signal-regulated kinase kinase (MEK), Erk, and JNK are localized at focal adhesions or fibronectin-induced adhesion complexes (6,7). Moreover, signals from growth factors and the cell adhesion/cytoskeletal complex are well coordinated. Growth factor signaling can modulate cell motility and the cytoskeleton (8,9). Conversely cell adhesion can enhance growth factor signaling (10 -13) as well as stimulate the ligandindependent tyrosine phosphorylation of growth factor receptors (14,15). Thus, there is substantial evidence for the cooperative function of cell adhesion and growth factor-mediated signaling. However, the molecular mechanisms of these types of cooperation are not fully understood.
Sos is one of the important signaling molecules recruited into focal adhesions (6). Sos has activity as a guanine nucleotide exchange factor (GEF) for Ras, and it mediates growth factor signals to the Erk kinase cascade leading to mitogenesis (16 -23). Sos is composed of multiple functional domains, including a Dbl homology domain, a pleckstrin homology domain, a Ras GEF domain, and a C-terminal proline-rich domain. The proline-rich domain is a binding site for the SH3 (Src homology 3) domains of adaptor proteins such as Grb2, Nck, and CrkII, and it mediates the intracellular translocation of Sos in response to growth factors (24 -26). The Dbl homology domain of Sos1 has been reported to have a GEF activity for another small GTPase Rac and to stimulate its downstream kinase, JNK/SAPK (27). Since Rac is thought to regulate cytoskeletal reorganization, Sos may play important roles in coordinated signaling involving both growth control and cytoskeletal organization (27).
Vinexin is a novel vinculin-binding protein localized at focal adhesion and cell-cell junctions (28). Expression of vinexin enhances actin cytoskeletal organization and cell spreading. Vinexin is transcribed into at least two alternative forms, vin-exin ␣ and ␤, both of which contain three SH3 domains. The first and second SH3 domains mediate the vinculin binding, whereas the binding partners of the third SH3 domain are unknown. Two closely related proteins, CAP/ponsin/SH3P12 and ArgBP2, have been identified. CAP/ponsin/SH3P12 also binds to vinculin as well as to other signaling molecules such as c-Cbl, insulin receptor, AF-6/afadin, and Sos, and it localizes at cell-cell and cell-matrix junctions (29 -32). ArgBP2 binds to the Abelson protein tyrosine kinases, Arg and c-Abl, and localizes along stress fibers (33). These data suggest that this protein family may function in the coordinated regulation of signaling and the cytoskeleton, although such putative functions in signaling pathways are unclear.
Vinexin ␤ has the most characteristic features of this protein family. Most of vinexin ␤ is occupied by three SH3 domains, and no apparent enzymatic domains have been identified so far (28). This feature is similar to those of other adaptor or scaffold proteins, such as Grb2, Crk, and p130 cas (34 -36). It raises the possibility that vinexin ␤ could function as a modulator of signal pathways. In the present study, we report that vinexin binds to Sos in vitro and in vivo, and this binding is mediated by the third SH3 domain of vinexin. Serum-and growth factorinduced Sos phosphorylation inhibited the formation of this vinexin-Sos complex. Furthermore, we found that vinexin modulated EGF-induced JNK/SAPK activation, in a process again dependent on the third SH3 domain of vinexin. These results indicate that vinexin ␤ associates with Sos and that it can regulate JNK/SAPK MAP kinase cascades induced by EGF.
Generation of GST Fusion Proteins-The GST fusion proteins containing vinexin domains, GST-1stSH3, GST-2ndSH3, and GST-1st2ndSH3, were described previously (28). The third SH3 domain of vinexin was amplified by polymerase chain reaction and subcloned into pGEX 4T-1 (Amersham Pharmacia Biotech) and was designated pGST-3rdSH3. Two mutants of the third SH3 domain were generated by substitutions at two of the most conserved amino acids among the SH3 domains (37). The mutations were introduced by using the Sculptor in vitro mutagenesis system (Amersham Pharmacia Biotech). The oligonucleotides 5Ј-CGATGGCTTCTTTGTGGG-3Ј and 5Ј-CCTGGAAATGT-TGTAGCCC-3Ј were used for the mutations of the tryptophan residue at position 306 of vinexin ␤ to phenylalanine (mutant WF) and the tyrosine residue at position 324 to valine (mutant YV), respectively. The mutations were confirmed by DNA sequencing. GST fusion proteins were purified as described (28).
In Vitro Binding Assays Using Affinity Precipitation-NIH/3T3 fibroblast cells were incubated for 24 h in medium containing 0.5% calf serum. The cells were then stimulated with EGF (100 ng/ml), PDGF (20 ng/ml) (Sigma), or 20% calf serum for 10 or 60 min, followed by lysis in 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 5 mM EDTA, 1.0% Triton X-100, 1 mM sodium orthovanadate, 100 mM NaF, 10 g/ml aprotinin, and 10 g/ml leupeptin. The cell lysates were affinity precipitated with 5 g of the GST fusion proteins for 3 h as described (28). Coprecipitated proteins were detected with the indicated specific antibodies and visualized using the ECL detection system (Amersham Pharmacia Biotech).
Immunoprecipitation-The expression plasmid for FLAG-tagged vinexin ␤ was described previously (28). Two mutants (mutant WF and YV) in the third SH3 domain of FLAG-tagged vinexin ␤ were generated as described above. These plasmids were transfected into NIH/3T3 cells with LipofectAMINE (Life Technologies, Inc.). Transfected cells were stimulated with serum and growth factors and lysed as described above. Equal amounts of total protein were incubated with 5 g of anti-FLAG mAb M2 for 1 h at 4°C. The immune complexes were incubated with protein G-Sepharose (Sigma) for 1 h and then precipitated and washed extensively with lysis buffer. The bound proteins were analyzed as described above.
JNK/SAPK Assay-JNK/SAPK activities were measured in an immune complex kinase assay as described (39) with minor modifications. Briefly, cells were solubilized as described above. The lysates were clarified by centrifugation. HA epitope-tagged JNK/SAPKs were immunoprecipitated with mAb 12CA5. The immunoprecipitates were resuspended in 30 l of kinase reaction mixture containing 12.5 mM MOPS (pH 7.5), 12.5 mM ␤-glycerophosphate, 7.5 mM MgCl 2 , 0.5 mM sodium orthovanadate, 3.3 mM dithiothreitol, 2 g GST-c-Jun, 20 M ATP, and 5 Ci of [␥-32 P]ATP. After incubation at 30°C for 20 min, kinase reaction products were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) and autoradiography. Quantitation was performed using an image analyzer BAS2000 (Fuji Photo Film Co., Japan).

Interaction of the Third SH3 Domain of Vinexin with Sos-
The proline-rich domain of Sos is the binding site for the SH3 domains of several adaptor proteins including those of Grb2, Nck, and CrkII. A novel cytoskeletal protein, vinexin ␤, consists of three SH3 domains with no apparent enzymatic feature so far, raising the possibility that vinexin ␤ may be an adaptor protein and bind to Sos. To examine the interaction of vinexin with Sos in vitro, various SH3 domains of vinexin were expressed and purified as GST fusion proteins (Fig. 1, A, B, and D). These GST fusion proteins, bound to glutathione beads, were incubated with cell lysates prepared from NIH/3T3 cells. The bound proteins were analyzed by immunoblotting using an anti-Sos1 mAb. As shown in Fig. 1C, the third SH3 domain of vinexin bound to Sos, whereas the first and second SH3 domains did not bind. A GST fusion protein containing both the first and second SH3 domains, a minimal vinculin binding region (28), also did not bind to Sos. These results suggested that the third SH3 domain of vinexin specifically binds to Sos.
To confirm the specificity of binding of the third SH3 domain of vinexin to Sos, we mutated the tryptophan residue at position 306 of the third SH3 domain of vinexin ␤ to phenylalanine (WF mutant) or the tyrosine residue at position 324 to valine (YV mutant) to disrupt function of the SH3 domain. These residues are highly conserved in various SH3 domains and have been shown to be required for binding to proline-rich sequences (37). Both GST fusion proteins with either of these mutated third SH3 domains completely lost Sos-binding ability (Fig. 1E), indicating that the third SH3 domain of vinexin specifically interacts with Sos.
Previous studies have demonstrated that growth factor stimulation results in the feedback serine/threonine phosphorylation of Sos, with concomitant dissociation of the complex between Sos and adaptor proteins (17, 40 -42), or that between Sos/Grb2 and receptor tyrosine kinase (43). To examine whether growth factor stimulation also modulates the affinity of Sos binding to vinexin, we first stimulated NIH/3T3 cells with 20% calf serum. As shown in Fig. 2A, serum stimulation decreased the electrophoretic mobility of Sos and concomitantly destabilized the complex between the third SH3 domain of vinexin and Sos. The effects were relatively stable, with only a slight restoration and formation of vinexin-Sos complexes 60 min after serum stimulation ( Fig. 2A). We next examined the effects of EGF and PDGF on formation of the complex of vinexin and Sos. Similar to serum stimulation, both EGF and PDGF stimulation of NIH/3T3 also decreased the electrophoretic mobility of Sos and the affinities of Sos to vinexin (Fig.  2B). These data demonstrated that serum, EGF, and PDGF all induce dissociation of the complex between vinexin and Sos.
Vinexin Interaction with Sos and Down-regulation of the Interaction by Serum Stimulation in Vivo-To determine whether the interaction between vinexin and Sos occurs in vivo, we transfected FLAG-tagged vinexin ␤ with or without point mutations in the third SH3 domain or the vector alone into NIH/3T3 cells. Lysates prepared from transfected cells were immunoprecipitated with anti-FLAG mAb, and precipitates were evaluated by immunoblotting with anti-FLAG mAb or anti-Sos1 mAb. As shown in Fig. 3, Sos was coimmunoprecipitated with FLAG-tagged vinexin ␤ but not from the lysates transfected by vector alone. Sos was also not precipitated with mutated FLAG-tagged vinexin ␤ containing mutations in the third SH3 domain. Moreover, Sos was not coimmunoprecipitated from lysates of serum-stimulated cells transfected with FLAG-tagged vinexin ␤. These data suggested that the third SH3 domain of vinexin binds to Sos in intact cells as well as in vitro and that the interaction is regulated by serum stimulation.
Regulation of the Interaction between Vinexin and Sos by the Phosphorylation of Sos-To determine whether the mobility shift of Sos and the interaction of vinexin with Sos are actually regulated by phosphorylation, serum-stimulated cell lysates were treated with alkaline phosphatase. Cell lysates incubated with or without alkaline phosphatase were precipitated with the GST fusion protein containing the third SH3 domain of vinexin. As shown in Fig. 4A, when lysates were treated with alkaline phosphatase, the serum-induced electrophoretic mobility shift was substantially reversed (Fig. 4A, lane 4, and Fig.  4B, lane 3). Similarly, the binding of Sos to vinexin was restored after the phosphatase treatment (lane 6). These data suggest that the phosphorylation of Sos following serum stimulation is required for the electrophoretic mobility shift, at least in part, and for the dissociation of the vinexin-Sos complex.
Effects of the Interaction between Vinexin and Sos on JNK/ SAPK Activities-Sos catalyzes the exchange of GDP to GTP on Ras and Rac (27). In the activated GTP-bound form, Ras and

FIG. 2. Serum, EGF, and PDGF stimulation decrease the binding of Sos to vinexin.
A, NIH/3T3 cells were serum-starved for 24 h and left untreated or were stimulated with calf serum for the indicated times. Lysates were either directly analyzed by SDS-PAGE or subjected to affinity precipitation with GST fusion proteins containing the third SH3 domain of vinexin ␤. The precipitates were washed and analyzed by SDS-PAGE. Both gels were subjected to immunoblotting using anti-Sos1 mAb. B, cells were serum-starved for 24 h and left untreated or stimulated with calf serum, 100 ng/ml EGF, or 20 ng/ml PDGF for 10 min. Affinity precipitation and immunoblotting using anti-Sos1 mAb were performed as described in A.
Rac activate several downstream targets, including Erk and JNK/SAPK, respectively (44 -46). Therefore, we examined the effect of vinexin ␤ expression on Erk2 and JNK/SAPK activation in response to EGF stimulation in NIH/3T3 cells. Cells were cotransfected with FLAG-tagged vinexin ␤, FLAG-tagged mutated vinexin ␤s, or the vector alone each with or without HA-tagged Erk2 or JNK/SAPK. Erk2 and JNK/SAPK activities were both stimulated by EGF. The highest activities were observed at 5 min for Erk2 and at 25 min for JNK/SAPK after EGF stimulation, when vector and HA-tagged kinases were cotransfected (data not shown). Therefore, the effects of overexpression of various forms of vinexin ␤ were determined at these time points. Expression of vinexin ␤ did not affect the Erk2 activation induced by EGF stimulation (data not shown). In contrast, expression of vinexin ␤ enhanced JNK/SAPK activation in response to EGF stimulation as shown in Fig. 5. This effect was specific to wild-type vinexin ␤, and no activation was observed in cells transfected by mutated vinexin ␤. In fact, transfection with mutated vinexin ␤ had marked dominantnegative effects in blocking most of the JNK/SAPK response to EGF (Fig. 5B), although expression of mutated vinexin ␤ appeared to enhance the Erk activation slightly (data not shown). Enhancement of EGF-induced JNK/SAPK activation by expression of wild-type vinexin ␤ and dominant-negative effects by expression of mutated vinexin ␤ were also observed in COS-7 cells (data not shown). These results suggest that vinexin ␤ is involved in the activation of JNK/SAPK in response to EGF and that the third SH3 domain of vinexin ␤ is important for this stimulation. DISCUSSION We have recently identified vinexin as a novel focal adhesion protein, which plays a role in cytoskeletal organization and cell spreading (28). Vinexin contains three SH3 domains. The first two SH3 domains are important for vinculin binding. In the present study, we show that the third SH3 domain of vinexin associates with Sos, a guanine nucleotide exchange factor for Ras and Rac, both in vitro and in vivo. Serum, PDGF, and EGF treatment induced dissociation of the vinexin-Sos complex by phosphorylating Sos. Moreover, expression of vinexin ␤ in NIH/ 3T3 and COS-7 cells promoted JNK/SAPK activation in response to EGF. Vinexin molecules with point mutations in the third SH3 domain displayed potent dominant-negative activity in suppression of the JNK/SAPK response to EGF. Together these results indicate that vinexin modulates not only cytoskeleton/cell spreading but also signal transduction induced by EGF and other growth factors.
We presented evidence that vinexin bound to Sos both in vitro and in vivo. We also showed that the binding site of vinexin for Sos was the third SH3 domain by using deletion mutants and point mutations in the third SH3 domain. The point mutations utilized here have been reported to disrupt the function but not the structure of SH3 domains (37), suggesting that the loss of Sos-binding ability is derived from the loss of SH3 function and not an artifact of structural alteration.
The exact binding site in Sos for vinexin remains to be determined, although the C-terminal proline-rich domain of Sos is the most plausible, because other SH3-containing adaptor molecules are known to bind in that region (24 -26). It is also intriguing that the minimal vinculin-binding region of vinexin, the first and second SH3 domains, is not necessary for Sos binding. This result suggests that a ternary complex of vinculin-vinexin-Sos may form a functional unit.
Serum-, EGF-, and PDGF-initiated signal transduction all regulated the affinity of Sos to vinexin by regulating Sos phosphorylation. Several lines of evidence have shown that serum or other growth factors can modulate Sos phosphorylation through Erk or other kinases and result in the dissociation of the adaptor-Sos complex (17, 40 -43, 47-49). In these cases, phosphorylation and subsequent dissociation of the complex are thought to provide negative feedback regulation for Ras and downstream signaling. The time course of dissociation of the vinexin-Sos complex was consistent with those of other adaptor-Sos complexes. These observations suggest that vinexin might also be involved in the regulation of signals initiated by receptor tyrosine kinases culminating in downstream MAP kinase activation.
Vinexin enhanced the activation of JNK/SAPK in response to EGF. Mutations in the third SH3 domain that abolished the ability of vinexin to bind to Sos eliminated this effect. Most importantly, both of these mutations were capable of functioning as "dominant-negative" inhibitors that strongly suppressed EGF-mediated JNK/SAPK activation in both NIH/3T3 and COS-7 cells. These findings indicate that vinexin regulates the signaling pathway from EGF stimulation to JNK/SAPK activation using its third SH3 domain. Although the precise role of vinexin in JNK/SAPK activation is not clear, it is possible that vinexin is required to recruit Sos to focal adhesions into close proximity to other signaling molecules and that this bound Sos activates Rac, leading to JNK/SAPK activation (27,45,46). Recently human vinexin ␣ (GenBank TM accession number AF037261) was reported in the GenBank TM data base to bind to PAK, an effector of Rac. Therefore, it is also possible that vinexin functions as a scaffold protein that binds to vinculin, Sos, PAK, and other molecules.
In conclusion, we have found that vinexin binds not only to the cytoskeletal protein vinculin but also to the signaling molecule Sos. Vinexin is involved in the signaling from EGF to JNK/SAPK MAP kinase, as well as in the regulation of cytoskeleton/cell spreading.