Integrin-mediated activation of MEK and mitogen-activated protein kinase is independent of Ras [corrected].

The integrins are a family of cell surface receptors that mediate adhesive interactions with the extracellular matrix and also generate signals that influence cell growth and differentiation. Ligation and clustering of integrins causes activation and autophosphorylation of focal adhesion kinase (FAK), a cytoplasmic tyrosine kinase, and results in the transient activation of p42 and p44 mitogen-activated protein (MAP) kinases. Initial evidence has suggested that the integrin signaling pathway may share common elements with the canonical Ras signal transduction cascade activated by peptide mitogens such as epidermal growth factor (EGF). In this report we demonstrate that Raf-1 and MAP or extracellular signal-related kinase kinase (MEK), key cytoplasmic kinases of the Ras cascade, are activated subsequent to integrin-mediated adhesion of mouse NIH 3T3 fibroblasts. We also show that MAP kinase is downstream of MEK in the integrin signaling pathway. However, in contrast to the receptor tyrosine kinase signaling cascade, integrin-mediated signal transduction seems to be largely independent of Ras. Dominant negative inhibitors of Ras-dependent signaling failed to block integrin-mediated activation of MEK. In addition, while treatment with the peptide mitogen EGF clearly increased GTP-loading of Ras, little effect was observed in response to integrin-dependent cell adhesion. Thus, integrin-mediated activation of MEK and MAP kinase in 3T3 cells occurs primarily by a mechanism that is distinct from the Ras signal transduction cascade.

The integrins are a family of cell surface receptors that mediate adhesive interactions with the extracellular matrix and also generate signals that influence cell growth and differentiation. Ligation and clustering of integrins causes activation and autophosphorylation of focal adhesion kinase (FAK), a cytoplasmic tyrosine kinase, and results in the transient activation of p42 and p44 mitogen-activated protein (MAP) kinases. Initial evidence has suggested that the integrin signaling pathway may share common elements with the canonical Ras signal transduction cascade activated by peptide mitogens such as epidermal growth factor (EGF). In this report we demonstrate that Raf-1 and MAP or extracellular signal-related kinase kinase (MEK), key cytoplasmic kinases of the Ras cascade, are activated subsequent to integrin-mediated adhesion of mouse NIH 3T3 fibroblasts. We also show that MAP kinase is downstream of MEK in the integrin signaling pathway. However, in contrast to the receptor tyrosine kinase signaling cascade, integrin-mediated signal transduction seems to be largely independent of Ras. Dominant negative inhibitors of Ras-dependent signaling failed to block integrin-mediated activation of MEK. In addition, while treatment with the peptide mitogen EGF clearly increased GTP-loading of Ras, little effect was observed in response to integrin-dependent cell adhesion. Thus, integrin-mediated activation of MEK and MAP kinase in 3T3 cells occurs primarily by a mechanism that is distinct from the Ras signal transduction cascade.
The integrin family of cell surface adhesion receptors plays a critical role in cell to extracellular matrix interactions and in the formation and maintenance of an organized cytoskeleton (1)(2)(3)(4)(5). In addition, it has become clear that integrins are also signal-transducing receptors (6) that regulate cell growth and survival (7)(8)(9)(10)(11)(12), influence gene expression (13)(14)(15)(16), and modulate tumor behavior (9,17,18). In many cell types, clustering of integrins leads to activation and autophosphorylation of a cytoplasmic tyrosine kinase termed pp125 FAK , and the recruitment of this protein to focal adhesion complexes (19 -21). Thus, FAK 1 activation may be a key primary event for integrin signal transduction processes.
An important issue is whether the signaling events triggered by integrins involve components used by better known signal transduction cascades, such as receptor tyrosine kinase pathways which feature a key role for the Ras proto-oncogene (22)(23)(24). Recent evidence has suggested that there may be substantial overlap between the integrin signaling pathway and the Ras consensus cascade. Thus, integrin-mediated activation of FAK leads to the creation of phosphotyrosine binding sites for SH2 domain proteins, including Grb2 (25), phosphatidylinositol 3-kinase (26), and Src family kinases (27), that have been implicated in aspects of the Ras cascade. Further, integrinmediated cell adhesion has been shown to strongly activate MAP kinase, a key downstream effector of the Ras signaling pathway (25,28,29). A number of proteins involved in the Ras signal transduction path have been observed to congregate in proximity to integrins at focal adhesion sites (30). Finally, in lymphoid cells, there is evidence that anti-integrin antibodies can cause GTP loading of Ras (31). However, despite these intriguing correlations, the signaling pathway from integrins to activation of MAP kinases has yet to be clearly delineated. In this report we demonstrate that integrin-mediated adhesion of mouse fibroblasts leads to activation of the major kinases of the Ras cascade, including Raf, MEK, and MAP kinase. The activation of MAP kinase caused by integrin ligation is dependent upon activation of MEK. However, integrin-mediated activation of MEK, and subsequently of MAP kinase, seems to occur by a pathway that is largely independent of Ras.

EXPERIMENTAL PROCEDURES
Preparation of Fibronectin-coated Dishes-Substrata were prepared by allowing a 25 g/ml solution of fibronectin to bind to tissue culture dishes at room temperature overnight, followed by blocking with 2% bovine serum albumin (BSA) for 2 h at room temperature. The dishes were rinsed with phosphate-buffered saline twice prior to use.
Cell Culture, Adhesion, Preparation of Total Cell Lysates-Wild type NIH 3T3 cells were cultured in Dulbecco's minimal essential medium (DMEM) supplemented with 10% bovine calf serum, 50 units/ml penicillin, and 50 g/ml streptomycin. Confluent cells were serum starved overnight in serum-free DMEM containing penicillin and streptomycin. The cells were dissociated from culture dishes with 0.05% trypsin, 0.53 mM EDTA. Trypsin activity was extinguished with 1 mg/ml soybean trypsin inhibitor. The suspended cells were washed three times in DMEM supplemented with 2% BSA. Cell suspensions were incubated in DMEM with 2% BSA at 37°C for 45 min on a rotator to allow kinases to become quiescent (in some experiments, 30 M PD98059 was included at this step). Thereafter, cells were plated on dishes coated with fibronectin, or other ligands, and incubated at 37°C for the indicated times. Following incubations, the cells were washed twice with ice-cold phosphate-buffered saline and then lysed in a modified RIPA buffer as described previously (28). Protein was determined using the Pierce bicinchonic acid assay.
Transfections-In some cases, 3T3 cells transfected with a vector expressing the chicken ␤1 integrin subunit (32) were used. Transfections were performed with LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's instructions. Cells expressing chicken ␤1 subunit were selected using magnetic beads (DYNAL) coated with the monoclonal anti-chicken ␤1 integrin antibody W1B10 (32). The selected cells were expanded in medium containing 250 g/ml G418 (Life Technologies, Inc.). Cells were further selected twice using Micro-CELLector flasks (Applied Immune Sciences) coated with W1B10, and expanded prior to use. Transient transfections of expression constructs encoding an epitope tagged-MEK (EE-MEK) (33), co-transfection with the Raf 23-284 construct (34), with a membrane-targeted, constitutively activated Sos-1 construct (35), with empty control vectors, or with activated or dominant-negative Ras constructs in the pZIP vector (36), were also done using the LipofectAMINE procedure.
Purification of Recombinant Histidine-tagged Wild Type MEK and Kinase-defective MAP kinases-Histidine-tagged wt MEK and kinasedefective MAP kinase were prepared according to Gardner et al. (37) with modification. Briefly, bacterial cell lysates were clarified by centrifugation at 10,000 ϫ g for 15 min at 4°C, the clear supernatant was incubated with 0.5 ml of Ni 2ϩ -nitrilotriacetic acid-agarose beads (Qiagen, Chatsworth, CA) in the presence of 25 mM imidazole and 500 mM NaCl on a rotator for 1 h at 4°C. The beads then were washed 3 times with 1 ml of lysis buffer containing 25 mM imidazole and 500 mM NaCl. The histidine-tagged kinase was purified by eluting 3 times with 1 ml of 200 mM imidazole followed by a Mono Q Fast Protein Liquid Chromatography (Pharmacia) step, in which the bound proteins were eluted with a linear 0 -500 mM NaCl gradient in a buffer comprised of 25 mM Tris-HCl, pH 8.0, 0.1 mM EDTA, 0.1 mM EGTA, 10% glycerol, 0.01% Nonidet P-40, 0.1% ␤-mercaptoethanol, and 1 mM benzamidine. Protein concentration was determined using the Coomassie dye-binding assay (Bio-Rad).
Immune Complex Kinase Assays-Cell lysates were precleared with 30 l of protein G-Sepharose beads for 15 min at 4°C. Raf-1, MEK, EE-tagged MEK (33), or MAP kinase were immunoprecipitated by incubation of the lysates at 4°C for 1 h with appropriate antibodies: 0.5 g of anti-Raf-1 polyclonal antibody (C-12, Santa Cruz Biotechnology, Santa Cruz, CA) for Raf-1; 2 g of anti-MEK1 and -2 (M17030, Transduction Laboratories, Lexington, KY) for MEK1 and MEK2; 5 g of anti-polyEE monoclonal antibody (Onyx Pharmaceuticals, Richmond, CA) for EE-tagged MEK1; 1 g of anti-ERK1 polyclonal antibody (K-23, Santa Cruz Biotechnology) for p42 and p44 MAP kinases. This was followed by incubation of the immunocomplexes with 30 l of protein G-Sepharose beads for another 1 h at 4°C. The beads then were washed with cold RIPA buffer once and a cold washing buffer (0.25 M Tris-HCl, pH 7.5, 0.1 M NaCl) twice. Changes in MEK, EE-MEK, and MAP kinase activities were determined by resuspending the immunocomplexes in 40 l of kinase assay buffer comprised of 10 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM dithiothreitol, 25 M ATP, 5 Ci of [ 32 P]ATP, and 2 g of kinase-defective MAP kinase (for MEK and EE-MEK) or 10 g of myelin basic protein (MBP) (for MAP kinase) and incubating them at room temperature for 30 min. Raf-1 kinase activity was measured by incubating the resuspended immunocomplexes in 30 l of kinase assay mixture containing 13.3 mM Tris-HCl, pH 7.5, 13.3 mM MgCl 2 , 1.3 mM dithiothreitol, 32.5 M ATP, 5 Ci of [ 32 P]ATP, and 0.5 g of wt MEK at room temperature for 15 min followed by adding 10 l of 0.2 mg/ml kinase-defective MAP kinase and incubating another 15 min at room temperature. Reactions were stopped by adding 20 l of 3 ϫ SDS sample buffer and boiling for 2 min. The samples were electrophoresed on SDS-polyacrylamide gels. The phosphorylated substrate bands were quantitated either by excising the bands and scintillation counting or by phosphorimaging.
Western Blot and MAP Kinase Mobility Shift Assay-EE-MEK was immunoprecipitated by anti-polyEE antibody as described above and blotted by anti-MEK1 and -2 (0.1 g/ml). FAK was immunoprecipitated with 4 g of anti-FAK monoclonal antibody (F15020, Transduction Laboratories), and EGF receptor by 5 l of an antiserum recognizing EGF receptor cytoplasmic domain from Dr. S. Earp (University of North Carolina, Chapel Hill, NC). Tyrosine-phosphorylated FAK and EGF receptor were detected by Western blotting with an anti-phosphotyrosine antibody (1 g/ml, Transduction Laboratories). Loading of FAK was evaluated by Western blotting with an anti-FAK polyclonal antibody as described (14). In some cases, MAP kinase activity was ascertained by a mobility shift assay as described previously (28).

RESULTS
This investigation was designed to compare integrin-mediated signaling with the better known signal transduction pathway triggered through receptor tyrosine kinases and to evaluate the role of Ras in the integrin pathway. Integrin-dependent adhesion of suspended NIH 3T3 cells to a fibronectin-coated substratum led to rapid and robust tyrosine phosphorylation of pp125 FAK , but not of the 185-kDa EGF receptor (Fig. 1, A and  C). Conversely, treatment of adherent 3T3 cells with EGF caused activation and autophosphorylation of the EGF receptor, but only slightly enhanced the high basal level of FAK tyrosine phosphorylation typical of attached cells (Fig. 1, B and  D). Treatment of suspension cells with EGF strongly activated the EGF receptor, but had no effect on FAK (Fig. 1, B and D). The overall patterns of tyrosine phosphorylation in lysates of cells stimulated by integrin ligation or by peptide growth factors also showed marked differences between the two situa-FIG. 1. Comparison of tyrosine phosphorylation patterns induced by cell adhesion or EGF stimulation. NIH 3T3 cells were grown to confluence in complete medium, then were serum-starved in serum-free DMEM for 16 -24 h. Cells were dissociated with trypsin/ EDTA; trypsin activity was then extinguished with soybean trypsin inhibitor followed by washing twice with 2% BSA/DMEM. The suspended cells were incubated on a rotator for 45 min at 37°C to allow kinases to become quiescent. Cells were then either maintained in suspension (Sus) or were allowed to attach to fibronectin (Fn)-coated dishes at 37°C for 30 min. For EGF stimulation, serum-starved cells maintained under adherent (Adh) conditions were washed with fresh serum-free DMEM three times and then incubated at 37°C for 1 h; alternatively, serum-starved cells were placed in suspension (Sus) as described above. EGF was added (EGF) to cultures of adherent or suspension cells at a final concentration of 100 ng/ml and incubated for 4 min; vehicle was added (Ϫ) to control cultures. Cell lysates were prepared and protein concentrations were determined as described under "Experimental Procedures." The protein tyrosine phosphorylation of FAK and EGF receptor (EGFR) was monitored by immunoprecipitating with appropriate antibodies and Western blotting with an anti-phosphotyrosine antibody. tions (38). 2 Thus, as expected, cell adhesion and growth factor stimulation triggered distinct initial tyrosine phosphorylation events, with EGF primarily activating its cognate receptor and integrin-mediated cell adhesion primarily activating FAK.
The cytoplasmic serine/threonine kinases Raf-1 (a MAP-KKK) and MAP kinase, as well as the dual-function kinase MEK (a MAPKK), were each strongly activated by either integrin-mediated attachment of suspended NIH 3T3 cells to fibronectin or by treatment of adherent cells with EGF ( Fig. 2A). The increase in kinase activities over basal levels was usually somewhat higher for maximally effective (100 ng/ml) concentrations of EGF than for coating concentrations of fibronectin (25 g/ml) that were maximally effective in promoting adhesion. Both the 42-kDa and 44-kDa forms of MAP kinase were activated by either EGF stimulation or integrin-mediated adhesion (28). 2 The kinetics of integrin activation of Raf, MEK, and MAP kinase were essentially the same, with maximal activation within 10 min of the beginning of the adhesion process and then a gradual decline over a period of 1 to 2 h. (Fig. 2B). This is similar to previously reported kinetics for adhesion-induced MAP kinase activation in Swiss 3T3 cells (28). To confirm that the observed effect on MAP kinase was specifically due to integrins, we examined kinase activity in 3T3 cells transfected with a chicken ␤1 integrin subunit and adhering to surfaces coated with an anti-chicken ␤1 monoclonal antibody or to surfaces coated with an antibody to CD44, an abundant non-integrin cell surface protein (39); adhesion to fibronectin was used as a positive control. As seen in Fig. 2C, specific adhesion mediated by the ␤1 integrin, but not by CD44, 2 T. H. Lin and R. L. Juliano, unpublished observations. FIG. 3. Integrin-mediated MAPK activation is dependent on MEK activation. Suspended NIH 3T3 cells were treated with 30 M concentration of the selective MEK inhibitor PD98059 for 45 min, then cells were harvested immediately (Sus) or allowed to attach to a fibronectin-coated substratum for 10 min (Fn). Cell lysates were prepared and MEK and MAPK activities were determined as described in the legend of Fig. 2.   FIG. 4. Integrin-mediated cell adhesion does not significantly change Ras GTP/GDP ratios. Serum-starved, 32 P-labeled, NIH 3T3 cells were placed in suspension and then allowed to attach to a fibronectin-coated substrate as described in the legend of Fig. 1; alternatively, serum-starved adherent cells were treated with EGF. Serum-starved NIH 3T3 cells transiently expressing constitutively activated Ras (61L) were also used as another positive control. Ras GTP/GDP ratios were measured as described (41). Results are presented as the fold increase in GTP/GDP ratios compared to unstimulated control which is taken as 1.0. Data are means Ϯ S.D. of six experiments performed in duplicate. The p values from Student's t-test for each stimulation with respect to its own control are as follows: p Ͼ 0.1 for ϩFn; p Ͻ 0.005 for ϩEGF; p Ͻ 0.001 for Ras (61L).

FIG. 2. Integrin-specific adhesion of NIH 3T3 cells activates
Raf-1, MEK, and MAPK. Serum-starved cells were placed in suspension and then allowed to attach to fibronectin-coated substrata as in the legend of Fig. 1. Adherent serum-starved cells were treated with EGF (100 ng/ml). In A, Raf-1, MEK, and MAPK activities were measured by immunocomplex kinase assays from same cell lysates. Following separation of the appropriate substrates, K Ϫ MAPK (for Raf-1 assays or MEK assays) or MBP (for MAPK assays), by SDS-polyacrylamide gels, the gels were dried and autoradiographed. The substrate bands were also excised from the gels and counted by scintillation counting (not shown). B shows the parallel time courses of Raf-1, MEK, and MAPK activation induced by cell adhesion. In C, NIH 3T3 cells were transfected with a chicken ␤1 subunit. The transfected serum-starved cells were placed in suspension and then allowed to attach to surfaces coated with anti-chicken ␤1 antibody W1B10, with fibronectin, or with antibody to CD44 (Serotec USA, Washington, D. C.) as a control. Equal numbers of cells were firmly attached in each case. MAPK activation was examined using a band shift assay, as described (28). led to a robust activation of MAP kinase, as detected by a band shift assay.
We wished to ascertain possible cause and effect relationships between the proteins activated by integrin engagement and to determine whether those relationships were the same as for the consensus Ras cascade activated by receptor tyrosine kinases. To do this, we used pharmacological or molecular tools to selectively inhibit upstream events and then examined the downstream consequences. Thus, a recently described selective inhibitor of MEK (PD98059) (40) allowed us to demonstrate that integrin-mediated MAP kinase activation is blocked when MEK activation is inhibited (Fig. 3). This anticipated result confirms that this segment of the signaling cascades for receptor tyrosine kinases and for integrins are similar.
We looked for evidence of Ras activation in the integrin pathway by determining if a transient increase in active Ras GTP occurs in stimulated cells, using a well-established assay for this purpose (41). As seen in Fig. 4, transfection of cells with oncogenic (61L) Ras, or stimulation of cells with EGF, resulted in clear and substantial increases in GTP loading of Ras. However, integrin-mediated cell adhesion did not produce a significant increase in GTP loading of Ras over control levels. Thus, EGF stimulation increased Ras GTP loading by a factor of 3.1 Ϯ 1.3 over control, while adhesion to fibronectin increased Ras GTP loading by a factor of only 1.3 Ϯ 0.5 over control; the difference between the EGF and adhesion effects was significant at the p Ͻ 0.01 level. This suggested that Ras was not strongly activated in response to integrin ligation. These ex-periments cannot rule out the possibility of small increases in Ras GTP loading as a result of cell adhesion; however, minor changes may not be important in signaling to downstream effectors.
To further investigate the role of Ras in integrin signaling, we used a transient co-transfection assay that allowed measurement of MEK activation in the presence or absence of dominant inhibitors of the Ras pathway. As a positive control, we examined the effect of these dominant inhibitors on activation of MEK by oncogenic forms of Ras or Sos-1, proteins which clearly are integral components of the Ras cascade. We chose to use co-transfection with oncogenic Ras or Sos, rather than EGF stimulation, since the EGF receptor can trigger both Ras-dependent and Ras-independent activation of MAP kinase in fibroblasts (42). Ras-independent activation of Raf and of MAP kinases by tyrosine kinases has also been observed in other mammalian cell types (43,44) and in Drosophila (45).
NIH 3T3 cells were transiently transfected with an epitopetagged MEK construct (EE-MEK) whose expressed protein could be immunoprecipitated from the transfected cells and tested for its ability to phosphorylate a mutated, inactive form of MAP kinase. These cells were also co-transfected with constructs expressing a dominant interfering mutant of Raf-1 or with a control vector lacking this insert. The Raf 23-284 mutant contains the amino-terminal Ras-binding sequence of Raf, but lacks the carboxyl-terminal kinase domain (34). Therefore, it blocks Ras function by forming an inactive complex with endogenous Ras (34, 46 -48). Co-expression of Raf 23-284 did FIG. 5. Dominant-negative constructs that block Ras signaling do not affect integrin-mediated MEK activation. In A, NIH 3T3 cells were transiently co-transfected using LipofectAMINE with 0.4 g of pCMV EE-MEK and 0.4, 0.8, or 1.6 g of pCGNraf 23-284 (total amount DNA was supplemented with pCGN empty vector to 2 g for each 35-mm tissue culture dish). Transfected cells were placed in suspension and then allowed to attach to a fibronectin-coated substratum as described in the legend of Fig. 1. EE-MEK was immunoprecipitated with an anti-polyEE monoclonal antibody, and EE-MEK activity was monitored by its ability to phosphorylate K Ϫ MAPK. B shows a positive control in which NIH 3T3 cells were transiently cotransfected with 0.4 g of EE-MEK and 0.1 g of pZIPras (12V) and with 0.1 g of pCGNraf 23-284 or pCGN empty vector (total DNA was supplemented with SP70 DNA to 2 g/35-mm dish). pCGNraf 23-284 is also known as pCGNraf N4 (34). In C, NIH 3T3 cells were transiently co-transfected with 0.2 g of EE-MEK construct and 1.8 g of a construct (pZIPras (17N)) encoding the Ras (17N) dominant-negative protein, or with empty pZIP vector. The cells were allowed to attach to Fn for 10 min. In D, 3T3 cells were co-transfected with 0.2 g of EE-MEK, 0.2 g of 5Ј-Sos F construct expressing the activated form of Sos-1, or with 0.2 g of vector (a negative control), and 0.4, 0.8, or 1.6 g of pZIPras (17N) (total DNA was supplemented with pZIP vector to 2.0 g). EE-MEK activities were determined as described above. The lower inset in each set of panels shows EE-MEK expression levels measured by Western blot with a polyclonal antibody against MEK1 and -2 in the same cell lysates as for the kinase assay. not affect the robust activation of EE-MEK caused by integrinmediated adhesion of 3T3 cells (Fig. 5A). However, co-expression of Raf 23-284 very effectively inhibited activation of EE-MEK caused by simultaneous transfection with a construct expressing a constitutively activated Ras (12V) mutant (Fig.  5B). These results strongly suggest that integrin-mediated activation of MEK, and subsequent activation of MAP kinase in 3T3 cells, are largely independent of Ras activation.
To further rule out the involvement of Ras in integrin signaling, 3T3 cells were co-transfected with the EE-MEK construct and with a construct expressing a mutated form of Ras (Ras 17N) that can act in a dominant-negative manner by blocking guanine nucleotide exchange factors involved in the activation of endogenous Ras (24,35). As seen in Fig. 5C, transfection with the Ras (17N) construct had no effect on the activation of EE-MEK caused by cell adhesion to fibronectin. However, transfection with the Ras (17N) construct strongly inhibited the activation of EE-MEK caused by expression of a membrane-targeted form (5Ј-Sos F) of the Sos-1 guanine nucleotide exchange factor for Ras (Fig. 5D). Once again, this suggests that integrin-mediated MEK activation does not involve Ras.

DISCUSSION
Treatment of NIH 3T3 cells with peptide mitogens such as EGF, or ligation of integrins, both lead to activation of tyrosine kinases; however, the patterns of tyrosine phosphorylation observed are quite different. EGF activates its cognate tyrosine kinase receptor which is then strongly autophosphorylated; other proteins are tyrosine-phosphorylated as well, but there is only a very modest effect on pp125 FAK . Conversely, integrinmediated adhesion of 3T3 cells causes a pronounced activation and autophosphorylation of the pp125 FAK tyrosine kinase, without affecting EGF receptor. Although the initial patterns of tyrosine phosphorylation are quite different, both EGF treatment and integrin ligation lead to the activation of several cytoplasmic kinases known to be important in signal transduction, including Raf, MEK, and MAP kinases. While FAK activation seems to parallel the activation of Raf-1, MEK, and MAP kinase, there is no direct evidence that FAK is responsible for activation of these downstream kinases.
We have shown that integrin-mediated activation of MAP kinase depends on prior activation of MEK. Thus, as expected, these kinases stand in the same relationship to each other in the integrin signaling pathway as in the receptor tyrosine kinase pathway. The role of Raf in integrin-mediated activation of MAP kinases is less clear. Although most models suggest that Raf-1 is upstream of MEK and responsible for its activation (24), other reports suggest that 3T3 cells contain several MAPKKKs capable of activating MEK1 and -2 (49). Thus, although Raf-1 can be activated subsequent to integrin-mediated adhesion, its role in propagation of signals to MAP kinases in this pathway is uncertain at this point.
Several recent observations have suggested that integrin signals might intersect with the Ras signal transduction cascade. Thus, integrin-mediated activation and autophosphorylation of FAK leads to the binding of Grb2-Sos complexes which might then serve to activate the Ras cascade (25). Further, both peptide mitogens and integrin ligation are capable of activating MAP kinases and the nuclear Jun kinase (28,30), both known downstream elements of the Ras cascade. Despite these presumed similarities, our current data strongly suggest that integrin signaling to MAP kinases in 3T3 cells is largely independent of Ras. Both GTP loading experiments, as well as use of dominant-negative inhibitors, failed to support a major role for Ras in integrin-mediated activation of MEK. The observation that expression of the dominant-negative Raf 23-284 and Ras 17N constructs, at levels which could significantly inhibit signaling events triggered by bona fide components of the Ras pathway, failed to have any effect on integrin signaling to MEK, strongly suggests that Ras is not the major contributor to this pathway. The fact that Grb2 binds to FAK (25) suggests that integrin signaling may project through the Ras pathway to some degree. However, our current observations strongly indicate the existence of Ras-independent events that link integrins to MAP kinase activation. Thus, the integrin signal transduction cascade clearly differs from the consensus Ras signaling pathway.
There are several possibilities for Ras-independent pathways of integrin signaling. One is that another small G-protein might substitute for Ras. Cdc42 and Rac, members of the Rho subfamily of low molecular weight G-proteins, have been implicated in activation of the Jun-kinase pathway (50). However, although Cdc42 and Rac are usually thought to be downstream of Ras, they only weakly activate MAP kinases (51). Rho has been implicated in the regulation of phosphatidylinositol 4,5bisphosphate production, and possibly through this mechanism, in the regulation of actin filament assembly (52). However, it is not clear whether these events impact upon MAP kinase activation. Another possibility is that integrin ligation modulates a kinase activity that directly impinges on Raf (or another MAPKKK) without the agency of a G-protein. A similar situation seems to prevail in TNF-mediated signaling, where activation of the Jun kinase pathway occurs in a Rasindependent fashion (53), presumably via lipid-activated kinases (54). It is interesting to note that several laboratories have identified proteins that interact directly with integrin cytoplasmic domains. These include a novel serine/threonine kinase (termed ILK) that binds to ␤1 or ␤3 cytoplasmic domains (55), and a novel small protein (termed endonectin) that binds selectively to ␤3 cytoplasmic domains (56). It will be of interest to determine whether ILK, endonectin, or other proteins that bind to integrins, might play a direct role in the integrin-mediated activation of MAP kinases.