Ras activation is necessary for integrin-mediated activation of extracellular signal-regulated kinase 2 and cytosolic phospholipase A2 but not for cytoskeletal organization.

Cell adhesion to the extracellular matrix triggers a cascade of intracellular biochemical signals regulated by the integrin family of receptors. Recent evidence suggests that integrin engagement may activate a mitogen-activated protein (MAP) kinase cascade that may cooperate with more clearly defined mitogenic signaling pathways to regulate cell proliferation, adhesion, and migration. Here we report that the adhesion-dependent activation of the MAP kinase Erk2 (extracellular signal-regulated kinase 2) occurs in serum-starved NIH3T3 cells, and that this activation of Erk2 is preceded by the activation of the small GTP-binding protein Ras in fibronectin-adherent cells. Inhibition of Ras signaling by expression of a dominant-inhibitory mutant of Ras (N17Ras) in NIH3T3 cells blocked adhesion-dependent activation of Erk2, although the focal adhesion kinase (FAK) was still activated in these cells. Furthermore, activation of this Ras-MAP kinase pathway activated cytosolic phospholipase A2, leading to the release of arachidonic acid metabolites, and N17Ras also inhibited these events. However, N17Ras expression does not inhibit cell adhesion, spreading, or focal contact and stress fiber formation. These results suggest that, while integrin-dependent activation of this MAP kinase pathway is Ras-dependent, the integrin-dependent activation of FAK and several morphological events are Ras-independent. Thus, integrin-mediated signals involved in regulating cell morphology appear to diverge from those regulating MAP kinase activation at a level upstream of Ras activation.

Integrins are the major family of transmembrane receptors that mediate attachment to the extracellular matrix (ECM). 1 Engagement and clustering of integrins lead to the formation of structures, called focal adhesions, where integrins link to intracellular cytoskeletal complexes and bundles of actin filaments. These structures form a scaffolding for the association of signaling molecules that regulate signal transduction pathways leading to integrin-induced changes in cell behavior. Recently, much has been learned about the interactions between integrins and the ECM, although significantly less is known about the intracellular biochemical pathways that integrins regulate and the cellular functions that are thereby controlled (reviewed in Refs. [1][2][3]. One family of proteins whose role in integrin-mediated signaling has only recently begun to be examined is the Ras superfamily of small GTP-binding proteins. Ras, the canonical member of the superfamily, is activated in response to numerous soluble growth and serum factors. Subsequently, Ras activates a signaling pathway that includes the mitogen-activated protein (MAP) kinases Erk1 and Erk2 which eventually regulate transcription, translation, and the cytoskeleton (4). Erk1 and Erk2 are also activated when cells adhere to the ECM proteins fibronectin, laminin, collagen, and vitronectin (5)(6)(7), suggesting that integrin-matrix interactions activate a MAP kinase cascade that may be related to that activated by soluble growth factors. Adhesion to fibronectin also stimulates the tyrosine phosphorylation of the focal adhesion kinase (FAK) as well as its association with other signaling molecules (Src and Grb2/SOS) that may promote activation of the Ras signaling pathway (6). If so, then integrin-mediated activation of Ras would be expected downstream of FAK and upstream of Erk1/ Erk2 activation.
Downstream targets for Erk1/Erk2 in the Ras-MAP kinase pathway include numerous potential substrates that fall into one of several categories (4): first, protein kinases such as p90 rsk that are involved in protein translation; second, a group that includes nuclear proteins such as Elk1, a ternary complex factor involved in c-fos induction; finally, cytosolic phospholipase A 2 (cPLA 2 ), an enzyme that liberates arachidonic acid and its metabolites from glycerolphospholipids. This final group is especially interesting in light of the observation that integrin clustering in HeLa cells induces the release of arachidonic acid metabolites and cell spreading, events that are inhibited by the PLA 2 inhibitor bromophenacyl bromide (8).
What role, if any, signaling molecules of the Ras-MAP kinase pathway play in each of these events remains to be examined, although a role for Ras in integrin-mediated signaling appears likely for two reasons. First, Ras can be found at sites of integrin-clustering (9), and, second, treatment of Jurkat cells with integrin antibodies activates Ras (10). However, matrixdependent activation of Ras has not been shown, and a role for Ras in integrin-dependent activation of MAP kinases remains to be established. Therefore, the purpose of this study was to examine signaling through the Ras-MAP kinase pathway and to determine whether or not this pathway plays a role in integrin-mediated cellular responses such as cell adhesion, spreading, and focal contact and stress fiber formation. We found that Ras is rapidly and transiently activated in fibronectin-adherent NIH3T3 cells, and that this activation precedes the peak of Erk2 activation. To test if integrin-mediated Erk2 activation is Ras-dependent, we expressed a dominant-inhibitory Ras mutant (N17Ras) in NIH3T3 cells and found that it inhibits matrix-dependent activation of Erk2 and cPLA 2 and arachidonic acid release, while integrin-dependent activation of FAK is not inhibited by N17Ras. Finally, although N17Ras appears to block the integrin-mediated Ras-MAP kinase pathway, it does not inhibit the formation of focal adhesions and stress fibers. These results suggest that, while integrin-mediated adhesion can activate the Ras-MAP kinase pathway, this pathway is not necessary for integrin-dependent focal contact and stress fiber formation.

EXPERIMENTAL PROCEDURES
Reagents-Antibodies to Ha-Ras were purchased from Transduction Laboratories (for immunoblotting) and Santa Cruz Biotech (for GTPloading). A pan-Erk antibody used for immunoblotting (which recognizes the major Erk in NIH3T3 cells, Erk2) was purchased from Transduction Laboratories, and the anti-phosphotyrosine antibody (clone 4G10) was purchased from Upstate Biotechnology Inc. An agaroseconjugated anti-Erk2 antibody (Santa Cruz Biotech) was used for immunoprecipitation/kinase assays. The cPLA 2 antibody was purchased from Santa Cruz Biotech. The anti-focal adhesion kinase (FAK) antibody (CK) was a gift from J.-L. Guan (Cornell University). Reagents for immunofluorescence were obtained from Sigma (anti-vinculin antibody) and Molecular Probes (rhodamine phalloidin). Human plasma fibronectin was purchased from Collaborative Biomedicals and platelet-derived growth factor BB from Life Technologies, Inc. The dominant inhibitory Ras construct, pMMRasDN, was obtained from J. Brugge (ARIAD Pharmaceuticals). A plasmid, pLENneo, containing the neomycin resistance gene, was provided by C. M. DiPersio (MIT). Dexamethasone was purchased from Sigma. 32 PO 4 , [␥-32 P]ATP, and [ 3 H]arachidonic acid were purchased from DuPont NEN.
Cell Preparation and Adhesion-NIH3T3 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) with 10% calf serum. For adhesion experiments, the cells were washed three times in phosphatebuffered saline (PBS) solution and then maintained in DMEM and calf serum (0 -10%). After 18 h, the cells were washed with PBS, trypsinized for 1 min, washed twice with DMEM containing 0.5 mg/ml soybean trypsin inhibitor and 0.1% bovine serum albumin, and finally suspended in DMEM for 30 min at 37°C before plating the cells on polylysine (0.5 mg/ml)-or fibronectin (10 g/ml)-coated dishes (prepared as described (11)). To initiate adhesion, the cells were added to the coated dishes and centrifuged for 30 s at 50 ϫ g. NIH3T3 cells expressing the N17Ras protein under the control of the murine mammary tumor virus promoter were created by cotransfection of pM-MRasDN and pLENneo and selected in 500 g/ml G418. Neomycinresistant colonies were screened for dexamethasone-inducible expression of N17Ras by immunoblotting using an anti-Ras antibody. Cells were stimulated with 300 nM dexamethasone for 18 h to induce expression of N17Ras. In addition to the two N17Ras-expressing clonal cell lines used for this report (DN1 and DN2), two other clones expressing high levels of N17Ras show results similar to those published herein.
Cell Lysis, Immunoprecipitation, and Immunoblotting-Adherent cells were washed twice in PBS and lysed in radioimmunoprecipitation assay (RIPA) buffer (1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 158 mM NaCl, 10 mM Tris-HCl, pH 7.2, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 500 g/ml leupeptin, 1 mM sodium orthovanadate, 50 mM NaF) for 30 min on ice. The lysates were clarified by centrifugation at 16,000 ϫ g for 10 min at 4°C, and protein concentration was then determined using a Pierce Micro BCA protein assay kit. One part 4ϫ SDS sample buffer (8% SDS, 2% ␤-mercaptoethanol, 266 mM Tris, pH 7.2, 40 mM EDTA) was then added to three parts RIPA lysate, the sample was heated to 100°C for 5 min, and 10 g of protein were run on a 10% acrylamide SDS-polyacrylamide gel electrophoresis gel. FAK immunoprecipitation was performed using 50 g of RIPA lysate, the polyclonal antiserum CK, and protein A-Sepharose. Immunoblotting of total cell lysates or immunoprecipitates was performed as described (12).
Analysis of Guanine Nucleotides Associated with Ras-Serumstarved NIH3T3 cells were labeled for 6 h at 37°C with 300 Ci of 32 PO 4 per 100-mm dish, trypsinized, washed, and resuspended for 30 min at 37°C in phosphate-free DMEM containing 300 Ci of 32 PO 4 . Next, the cells were added to fibronectin-or polylysine-coated dishes, centrifuged for 30 s at 50 ϫ g, and allowed to adhere at 37°C as described above. Activation of Ras was then assessed by measuring the level of associated guanine nucleotides as described previously (13).
MAP Kinase Assay-Immunoprecipitation/kinase assays were performed using 50 g of RIPA-cell lysates prepared as described above. Erk2 was immunoprecipitated with 10 g (25 l) of agarose-conjugated antibody for 90 min at 4°C, and the immunoprecipitates were washed twice in RIPA buffer and twice in kinase wash buffer (20 mM Hepes, pH 7.2, 10 mM MgCl 2 , 1 mM dithiothreitol). The kinase assays were then performed using the MAP Kinase Assay Kit (Upstate Biotech) according to the manufacturer's directions. Duplicate immunoprecipitates of Erk2 were immunoblotted with the pan-Erk antibody to ensure that equal amounts of Erk2 protein were used in these in vitro kinase assays.
Release Assay for Arachidonic Acid Metabolites-Cells labeled with 0.4 Ci/ml [ 3 H]arachidonic acid for 18 h were washed, trypsinized, and replated on fibronectin-or polylysine-coated dishes as described above. After allowing the cells to adhere for 5 min at 37°C, the dishes were rinsed once with prewarmed DMEM, and fresh DMEM was added to the dishes. The cells were returned to the incubator for 0, 5, 10, 15, or 35 additional minutes, and 50 l of media were removed for scintillation counting. To ensure that equivalent amounts of radiolabeled lipids were present in the parental NIH3T3-and N17Ras-expressing cells, equal numbers of cells were extracted with chloroform, and the radioactivity was determined by scintillation counting.

RESULTS
Several studies have suggested a link between integrin-mediated signaling and the Ras-MAP kinase pathway. Specifically, adhesion of cells to ECM components such as fibronectin induces the activation, tyrosine phosphorylation, and nuclear translocation of MAP kinases through an integrin-dependent mechanism (5-7). However, a recent study has concluded that integrin engagement of the ECM is not sufficient to trigger the activation of Erk1 and Erk2 in serum-starved Swiss 3T3 cells (14). To examine the serum dependence of integrin-mediated MAP kinase activation, NIH3T3 cells were incubated in media containing 0, 0.5, or 10% serum for 18 h before replating in serum-free media on fibronectin-coated dishes for 0 to 40 min. MAP kinase activation was examined by mobility shift assay. As shown in Fig. 1A, we observed an upward gel mobility shift in Erk2 characteristic of its activation and phosphorylation on tyrosine and threonine residues. The shift was rapid, occurring as early as 5 min after plating the cells onto a matrix, and was maximal between 10 and 20 min. The activation of Erk2 was observed even in serum-starved (0%) cell lysates, although activation was somewhat enhanced when cells were maintained in even low levels of serum (0.5%), and higher concentrations of serum (10%) appeared to dampen integrin-mediated Erk2 activation.
The observed fibronectin-induced activation of MAP kinase suggested a potential role for activated Ras in integrin-mediated signal transduction (1). Therefore, we analyzed the activity state of Ras by quantitating the levels of GTP and GDP associated with Ras in lysates from suspended or fibronectinadherent NIH3T3 cells (Fig. 1B). A rapid and transient activation of Ras was observed in cells plated onto fibronectin, with the peak of GTP associated with Ras occurring as early as measurable (5 min). Although the percentage of GTP associated with Ras in fibronectin-adherent cells was less than that observed in PDGF-stimulated cells, it was significantly greater than that in suspended or polylysine-adherent cells (Fig. 1B). This peak of Ras activation precedes both cell spreading (which can first be observed 15 min after the cells have been plated onto fibronectin) and the peak of Erk2 activation (see Fig. 1A), suggesting that Ras could regulate integrin-mediated Erk2 activation. However, this correlation does not ensure that Erk2 activation is Ras-dependent.
To examine the regulation of fibronectin-induced MAP kinase activation by Ras, NIH3T3 cells were transfected with a dominant-inhibitory mutant of Ras (N17Ras). Expression of N17Ras in cells inhibits the activation of endogenous Ras, presumably by interfering with the exchange of GTP for GDP, and has been shown to block Ras-mediated growth factor signaling (15). Expression of N17Ras was controlled in these cells by the dexamethasone-inducible murine mammary tumor virus promoter. Parental NIH3T3 cells and two separate clones which expressed relatively low (DN1) and high (DN2) levels of N17Ras when stimulated with dexamethasone for 18 h (see Fig. 2A) were used in subsequent experiments. As shown in Fig. 2A, expression of N17Ras at relatively high levels (DN2) completely inhibited both PDGF-and matrix-dependent upward gel mobility shift characteristic of Erk2 activation. Furthermore, N17Ras inhibited the activation of Erk2 kinase activity (Fig. 2B). Expression of lower levels of N17Ras (DN1) delayed activation of Erk2 ( Fig. 2A).
The ability of the dominant-inhibitory N17Ras to inhibit fibronectin-mediated activation of Erk2 places this MAP kinase downstream of Ras in integrin as well as growth factor signaling. In an attempt to delineate further the signaling pathway from integrins to Erk2, we examined whether the activation of FAK, a potential upstream mediator of integrin-dependent Ras and MAP kinase activation (6), is Ras-dependent. Expression of N17Ras did not inhibit the tyrosine phosphorylation of FAK in fibronectin-adherent cells (Fig. 2C). In fact, when expressed at the highest levels, N17Ras enhanced the tyrosine phosphorylation of FAK.
We next examined a potential downstream target for Erk2, cPLA 2 . Phosphorylation of cPLA 2 by MAP kinase, which causes an electrophoretic mobility shift that correlates with an increase in enzymatic activity, is believed to account for the agonist-stimulated activation of cPLA 2 (16). However, several recent studies suggest that MAP kinase-independent pathways exist for cPLA 2 activation (17,18). When cell lysates from NIH3T3 cells plated onto fibronectin were examined, we observed an upward gel mobility shift characteristic of cPLA 2 activation (Fig. 3A), and this activation correlated with the release of arachidonic acid metabolites (Fig. 3B). These results are in agreement with studies in HeLa cells where integrin clustering is believed to induce integrin-dependent arachidonic acid release through activation of cPLA 2 (8). Expression of the dominant-inhibitory N17Ras completely blocked the upward gel mobility shift indicative of cPLA 2 activation and significantly inhibited the release of arachidonic acid metabolites (by 80% at the 20-min time point; see Fig. 3), indicating that Ras mediates integrin-dependent cPLA 2 activation and arachidonic acid release in NIH3T3 cells.
Having established that integrin-mediated activation of both Erk2 and cPLA 2 is Ras-dependent while FAK phosphorylation is not, we next addressed the question of how these signaling events might control morphological events such as cell spread- The slower migrating band indicates the activated (phosphorylated) form of Erk2. B, serum-starved, 32 PO 4 -labeled NIH3T3 cells in suspension (0 min), plated on fibronectin for 5 to 40 min, stimulated with 100 ng/ml PDGF BB (PDGF) in suspension for 5 min, or plated on polylysine for 5 min were lysed, and Ras was immunoprecipitated to examine by thin layer chromatography the associated guanine nucleotides. Nonradioactive GTP and GDP were used as standards. Radioactive GTP and GDP were quantitated using a Molecular Dynamics PhosphorImager. The data are expressed as the percentage of radioactive GTP [GTP/ (GTP ϩ GDP)] divided by the percentage of radioactive GTP in suspended (0 min) cells Ϯ the range. The data ranged from 19% to 43% GTP in suspended cells and from 32% to 66% GTP in cells adherent to fibronectin for 5 min.

FIG. 2. Integrin-mediated Erk2 activation is Ras-dependent, while FAK activation is Ras-independent.
NIH3T3 cells (grown in 0.5% serum for 18 h) inducibly expressing a dominant-inhibitory (N17)Ras were trypsinized, washed, and replated on fibronectin-coated dishes for 0, 5, 10, 20, 40, or 60 min or stimulated with 100 ng/ml PDGF BB (PDGF) in suspension as described under "Experimental Procedures." DN1 and DN2 are two individual clones expressing different levels of N17Ras. A, 10 g of cell lysates from parental (3T3) or N17Rasexpressing (DN1 or DN2) NIH3T3 cells were electrophoresed, and MAPK activation was analyzed by immunoblotting with a pan-Erk antibody. The slower migrating band indicates the activated (phosphorylated) form of Erk2. N17Ras expression in these cells was examined by immunoblotting the identical immunoblot with an anti-Ras antibody. B, 50 g of cell lysates were immunoprecipitated with an antibody to Erk2 and subjected to an in vitro kinase reaction using myelin basic protein as a substrate. The phosphorylated substrate was then bound to filter paper and subjected to scintillation counting. Duplicate results from three trials are shown Ϯ S.D. C, 50 g of lysates from suspension (0) or fibronectin-adherent (for 20 or 60 min) cells were immunoprecipitated with a polyclonal antiserum to FAK, subjected to electrophoresis, and immunoblotted with an anti-phosphotyrosine antibody as described under "Experimental Procedures." ing or focal contact and stress fiber formation. Since microinjection of Ras stimulates focal contact and stress fiber formation (19), it is plausible that Ras might control some aspect of integrin-mediated morphology. In addition, arachidonic acid metabolites are reported to be necessary for cell spreading (8) and actin remodeling (20), suggesting a role for Ras in these morphological changes. However, cells expressing the highest levels of N17Ras (DN2) were fully able to adhere, spread, and form focal contacts and stress fibers when plated on fibronectin (see Fig. 4). In fact, expression of N17Ras appears to enhance slightly vinculin-containing focal contacts, although heterogeneity is observed in the morphology of these cells. These results suggest that the Ras-MAP kinase-cPLA 2 -arachidonic acid pathway described in this report is not necessary for the formation of integrin-dependent morphological structures in NIH3T3 cells.

DISCUSSION
Although many cellular processes are known to be matrixdependent, the contributions of signals transduced by integrins have only begun to be explored. The ability of integrins to activate MAP kinases suggested a link to the Ras signaling pathway (5)(6)(7). In this report, we have shown directly that adhesion of fibroblastic cells to a defined matrix can indeed activate Ras. This result is in agreement with another study in which treatment of lymphoid cells with integrin antibodies activates Ras (10) and suggests a potential pathway for integrindependent activation of MAP kinases. Using a dominant-inhibitory strategy, we have shown directly that integrin-dependent, growth factor-independent activation of Erk2 in NIH3T3 cells is Ras-dependent. Furthermore, cPLA 2 appears to be a substrate for a Ras-activated kinase, presumably Erk2, and the release of arachidonic acid metabolites upon matrix adhesion is, in large part, Ras-dependent. However, in contrast with some previous suggestions, this pathway does not appear to regulate integrin-dependent morphological changes that occur when cells engage the ECM.
Why is integrin-dependent activation of the Ras-MAP kinase pathway important? First, it may play a role in regulating integrin-dependent gene expression. Adhesion of suspensionarrested cells induces the rapid expression of several genes including c-fos, a component of AP1 complexes (21). Activation of the Ras-MAP kinase cascade is known to induce MAP kinase phosphorylation of Elk1 and increase the transcriptional activation of c-fos. These events then regulate the activity of the AP-1 complex, thereby controlling cell proliferation (22). In this way, integrin-dependent activation of this pathway could regulate adhesion-dependent cell growth. Our results suggest that this integrin-dependent MAP kinase pathway converges with growth factor signaling at least at the level of Ras, perhaps through the formation of signaling complexes that integrate signals from integrins and growth factor receptors (23).
Second, Ras might be responsible for regulating cytoskeletal rearrangements involved in cell adhesion, spreading, and/or focal contact and stress fiber formation. Activated Ras can trigger a cascade which induces morphological changes such as membrane ruffling (19). However, Ras appears to do so through a MAP kinase-independent pathway (24). Furthermore, while there is considerable evidence that Ras can activate a signaling cascade (that includes the Rho family of small GTP-binding proteins) which regulates cytoskeletal dynamics (25), our observation that Ras inhibition does not prevent integrin-dependent cell adhesion, spreading, or focal contact and stress fiber formation in NIH3T3 cells suggests that a Ras-MAP kinaseindependent pathway exists for these morphological events.
Ras has also been implicated in the growth factor-induced release of arachidonic acid (26,27), and this is supported by our observation that the adhesion-dependent activation of cPLA 2 and subsequent arachidonic acid release are Ras-dependent. However, our results suggest that this pathway is not essential for integrin-dependent adhesion or spreading of NIH3T3 cells on a fibronectin matrix. This observation is in contrast with previous studies in which arachidonic acid metabolites were reported to be essential for the spreading of HeLa cells on a collagen matrix (8,28). The observed differences may be due to the different cell types, matrix proteins, and/or integrin receptors engaged in these studies; it is also possible that the reduced levels of arachidonic acid released by our cells expressing N17Ras may be sufficient to allow the cells to spread.
While our results indicate that actin rearrangements induced by engagement of integrins are Ras-independent, these events may nonetheless be mediated by Rho family members Rac and Rho which are essential for epidermal growth factor- induced actin stress fiber formation (29). Both Rac and Rho stimulate actin polymerization through regulation of phosphatidylinositol metabolism, although the precise targets remain to be identified (30,31). Hotchin and Hall (14) have recently shown that the formation of focal contacts requires an interaction of integrins with the ECM and a functionally active Rho which, in Swiss 3T3 cells, must be activated by serum factors. However, in NIH3T3 cells it appears either that serum starvation is not sufficient to turn off Rho or that Rho is activated upon engagement of integrins, or that integrin-dependent actin rearrangements are Rho-independent. Although it is presently unknown how actin rearrangements induced by integrin engagement and clustering may be mediated by Rho family members, future studies may aid in defining where integrin-dependent signaling pathways regulating cell morphology diverge from those regulating cell growth; we now know that they do so upstream of the Ras-MAP kinase pathway.