Src Kinase Activity Is Required for Integrin αVβ3-Mediated Activation of Nuclear Factor-κB*

Integrin adhesion to extracellular matrix proteins protects adhesion-dependent cells from suspension-induced apoptosis. Previous studies indicate that activation of the transcription factor nuclear factor-κB was necessary for the integrin αvβ3 ligand osteopontin to protect endothelial cells from apoptosis caused by serum withdrawal. In this study, β3 integrins were overexpressed in smooth muscle cells. When plated on osteopontin, cells overexpressing wild-type β3 had enhanced cell adhesion, cell spreading, and nuclear factor-κB activation compared with vector control. Removal of four amino acids (759X) from the C terminus of β3 eliminated the ability of the integrin to promote these processes. Single amino acid substitutions indicated that phosphorylation at tyrosine 759 was not required for activation of the transcription factor, however this residue appeared to play a structural role, because mutation to alanine significantly inhibited nuclear factor-κB activation. The Src family of tyrosine kinases represents important transducers during integrin signaling, and the C terminus of β3 has been implicated as the binding site for Src. Immunoprecipitations demonstrated that Src associated with wild-type β3 integrins, but Src and integrins lacking the C terminus (759X) did not form a complex. Pharmacological inhibition with the Src inhibitor PP2 (4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine) or overexpression of kinase-dead c-Src blocked nuclear factor-κB activation. Mouse embryonic fibroblasts deficient for Src failed to activate nuclear factor-κB when plated on osteopontin, in contrast to control fibroblasts. Together, these experiments indicate that the C terminus of β3 and Src activity are required for integrin αvβ3-mediated nuclear factorκB activation.

Integrins are heterodimeric receptors composed of two type I transmembrane glycoproteins, ␣ and ␤. Integrins typically contain a large extracellular domain, a transmembrane domain, and a short intracellular domain with no known enzymatic function. The integrin extracellular domain binds to extracellular matrix ligands or cell surface ligands, whereas the intracellular domain provides a link to adapter proteins and the cytoskeleton. Thus, integrins act as adhesion receptors, physically linking the extracellular matrix to the cytoskeleton. Inte-grin ligation also promotes intracellular signal transduction through outside-in signaling. These signaling pathways promote complex biological functions, including cell adhesion, proliferation, migration, differentiation, and cell survival. For many cell types, integrin-mediated adhesion to the extracellular matrix is required for not only cell growth but also for cell survival (1). Disruption of adhesion causes the cells to undergo apoptosis, a process termed anoikis. Integrin antagonists also induce apoptosis both in vitro (2) and in vivo (3). It has also been suggested that non-ligated integrins can actively promote cell death (4).
The Src family of tyrosine kinases (SFKs) 1 represents important transducers during integrin-mediated signaling, and activated Src can protect cells from apoptosis. The Src family consists of eight members but only c-Src, Yes, and Fyn are ubiquitously expressed. SFKs show redundancy, therefore, inhibition of multiple SFKs is often required to demonstrate phenotypes. Mice deficient in the three ubiquitous kinases: c-Src, Yes, and Fyn, display developmental defects leading to embryonic lethality (5). Embryonic fibroblasts from mice lacking c-Src, Yes, and Fyn (SYF cells) have reduced protein phosphotyrosine levels following integrin ligation and reduced cell migration on fibronectin (5). Although c-Src is only modestly activated following integrin stimulation, c-Src has been shown to directly phosphorylate many downstream proteins, including FAK, Cas, and paxillin. In fibroblasts, SFKs regulate integrin-mediated cell attachment and spreading in response to fibronectin but not collagen (6).
The transcription factor nuclear factor-B (NF-B) is another important player in cell survival. NF-B activation protects endothelial cells, hepatocytes, B cells, osteoclasts, melanoma, and smooth muscle from apoptosis (7). The NF-B family consists of homo-or heterodimeric subunits of Rel family members. In the cytoplasm, NF-B binds to the inhibitory protein I-B (IB). Upon stimulation, IB becomes phosphorylated, ubiquinated, and degraded by the proteasome, allowing NF-B to translocate to the nucleus and promote gene expression of pro-survival genes, including but not limited to the inhibitors of apoptosis, bcl-2 family proteins, and osteoprotegrin (8).
We have previously shown that integrin ␣ v ␤ 3 ligation in rat aortic endothelial cells (RAECs) activates NF-B, protecting the cells from apoptosis (2). Blocking ␣ v ␤ 3 ligation or preventing NF-B activation induces cell death. Thus, NF-B is a downstream mediator of integrin ␣ v ␤ 3 in endothelial cell sur-vival. In the present study we provide evidence that the integrin ␣ v ␤ 3 /NF-B pathway is found in multiple cell types, perform structure/function studies to establish the region of the integrin ␣ v ␤ 3 required for NF-B activation, determine that SFKs bind to integrin ␤ 3 and the association requires the C terminus of the integrin, and identify that SFK kinase activity is required for integrin ␣ v ␤ 3 -mediated NF-B activation.
Cell Culture-RAECs were purchased from VEC Technologies Inc. and grown in MCDB 131 media (Invitrogen) supplemented with 10 mM L-glutamine (Invitrogen), 10% fetal bovine serum (HyClone, Logan, UT) and 100 units/ml each of penicillin and streptomycin (Invitrogen). Medial smooth muscle cells from Wistar Kyoto rat pups were isolated as previously described (15) and grown in Waymouth's MB 752/a media (Invitrogen) supplemented with 10% fetal bovine serum (HyClone) and 100 units/ml each of penicillin and streptomycin (Invitrogen). Cells were passaged by immersing in 0.05% trypsin.
Retroviral Infections-High titer amphotrophic retrovirus was prepared as previously described (16). Briefly, Phoenix packaging cells were transiently transfected with the retroviral plasmids. Medium was removed 10 and 24 h post-transfection. At 24-h post-transfection the cells were moved to a 32°C incubator and virus was collected between 24 and 48 h after transfection. The virus-containing media was passed through a 0.45-m filter to remove cellular debris.
Twenty-four hours prior to the infections, 10 5 Wistar Kyoto rat smooth muscle cells (WKY) or 2 ϫ 10 5 RAECs were plated per well in 6-well tissue culture plates. Cells underwent three rounds of infection. For each infection, the medium was replaced with 2 ml of retroviralcontaining medium supplemented with 4 g/ml Polybrene (Sigma-Aldrich, St. Louis, MO) and spun at 1100 ϫ g and 32°C for 1 h then returned to the 37°C incubator. After 24 h, the old medium was removed. Following the third infection, enhanced green fluorescent pro-tein-positive cells were detected by flow cytometry when applicable. For viruses encoding Src, cells were selected with 200 g/ml Hygromycin D (Calbiochem).
Flow Cytometry-For integrin expression profiling, 1 ϫ 10 6 cells were washed with phosphate-buffered saline (PBS) plus 0.2% bovine serum albumin (Sigma-Aldrich) plus 0.02% sodium azide. Cells were incubated on ice for 30 min with either non-immune mouse-IgG (Sigma-Aldrich) or the mouse monoclonal antibody SZ.21, which recognizes human ␤ 3 but not rat ␤ 3 (17). Cells were washed and incubated with phycoerythrin-conjugated anti-mouse IgG (Biomeda, Foster City, CA). Cells were washed and fixed in 1% paraformaldehyde. Samples were analyzed using the BD FACScan flow cytometry system and analyzed by CellQuest (BD Biosciences, San Jose, CA).
Recombinant OPN-Generation of histidine-tagged full-length recombinant rat OPN has been previously described (18). Briefly, Escherichia coli were transformed with the pQE30 plasmid encoding recombinant rat OPN with six histidines attached to the N terminus. The bacteria were amplified, and protein expression was induced by adding 1 mM isopropyl-␤-D-thiogalactopyranoside (Research Products International, Mt. Prospect, IL). The protein was collected using a nickelnitrilotriacetic acid-agarose resin (Qiagen, Valencia, CA), and eluted using 0.2 M imidazole (Acros Organics). Protein concentration was determined using the MicroBCA assay (Pierce, Rockford, IL). Western blots and adhesion assays verified protein purity and functionality respectively.
Adhesion Assays-Adhesion assays were performed as previously described (19). Briefly, protein solutions were diluted in PBS and adsorbed to 96-well Maxisorb plates (Nalge Nunc International, Rochester, NY) at 4°C overnight. Wells were rinsed with PBS, and nonspecific binding sites were blocked by adding 10 mg/ml bovine serum albumin and incubating at 37°C for 1 h. Cells were trypsinized, centrifuged, and washed twice with Waymouth's media without serum. Samples were plated with 60,000 cells/well and incubated at 37°C for 2 h. The plates were rinsed with PBS and fixed with 4% paraformaldehyde. The cells were stained with 0.5% toluidine blue (J. T. Baker, Phillipsburg, NJ) in 4% paraformaldehyde and rinsed in water. The dye was solubilized using 1% SDS. Adhesion was quantified by absorbance at 630 nm.
NF-B Activation Assays-Recombinant histidine-tagged rat OPN or poly-D-lysine (PDL) (Sigma-Aldrich) was adsorbed to 12-well tissue culture plates by incubating at 4°C overnight. The wells were blocked with 10 mg/ml bovine serum albumin for 1 h at 37°C. RAECs or WKY cells expressing the NF-B-dependent luciferase reporter were trypsinized and washed in serum-free media. For inhibition studies, the cells were treated with the active pharmacological Src inhibitor PP2 (4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo [3,4-d]pyrimidine, Calbiochem) or the inactive control PP3 (4-amino-7-phenylpyrazol- [3,4-d]pyrimidine, Calbiochem) and kept in suspension for 45 min. The cells were then plated on the OPN-coated surfaces and allowed to adhere for 5-7 h. Non-adherent cells were washed away and adherent cells were lysed with Reporter Lysis Buffer from Promega (Madison, WI). Protein recovery was monitored by microBCA assay (Pierce). Luciferase activity was detected using the Promega Luciferase Assay System, and activity was normalized to total protein recovered.
Western Blots-Proteins were extracted from cell monolayers in Laemmli buffer containing protease inhibitors (2 g/ml aprotinin, 2 g/ml leupeptin, 2 g/ml pepstatin, and 1 mM phenylmethylsulfonyl fluoride). After lysis and boiling, protein concentration was determined by Mi-croBCA assay. 30 g of lysate was loaded onto 10% SDS-polyacrylamide gels. Identical gels were stained with Pro-Blue staining solution (Owl Separation Systems, Portsmouth, NH) to verify uniform loading. Samples were transferred onto polyvinylidene difluoride membranes. Membranes were probed with the appropriate primary antibody, washed, and then horseradish peroxidase-conjugated secondary antibodies (Jackson Immunoresearch Laboratories, West Grove, PA). Bands were detected using the Western Lightning chemiluminescence kit (PerkinElmer Life Sciences).
Immunoprecipitations-WKY cells overexpressing c-Src and vector control or integrin ␤ 3 subunits, WT or 759X, were grown to confluence and lysed in lysis buffer containing 1% Nonidet P-40, 150 mM NaCl, 50 mM Tris, 1 mM sodium orthovanadate, 0.5 mM NaF and protease inhibitors. 800 g of lysate was immunoprecipitated with rabbit polyclonal anti-integrin ␤ 3 antibody or rabbit IgG control antibody and Protein A-Sepharose beads (Sigma-Aldrich). The immunoprecipitates or total cell lysate were subjected to Western blotting with the indicated detection antibodies.

RESULTS
Previous studies showed that integrin ␣ v ␤ 3 ligation activated the transcription factor NF-B in RAECs, and inhibition of ␣ v ␤ 3 ligation or inhibition of NF-B activation caused endothelial cell apoptosis (2). In the current study, we investigated the mechanism of ␣ v ␤ 3 -mediated NF-B activation. To determine which integrin subunit was involved in activation of NF-B, melanoma cells overexpressing ␣ v ␤ 3 or ␣ IIb ␤ 3 were plated on the ␣ v ␤ 3 ligand fibronectin or the ␣ IIb ␤ 3 ligand fibrinogen, respectively. Both cell types activated NF-B based on gel shift assays (data not shown), suggesting that the ␤ 3 subunit was the mediator for integrin-mediated NF-B activation.
Smooth Muscle Cell Adhesion through the Integrin ␣ v ␤ 3 -Activated NF-B-Medial smooth muscle cells from Wistar Kyoto rats (WKY) were identified as an excellent cell model for structure/function studies of the integrin ␤ 3 subunit, based on their high surface expression of ␣ v and low expression of ␤ 3 (data not shown). To test NF-B activation, WKY cells were infected with an NF-B-dependent luciferase retroviral construct. The cells were then plated on surfaces coated with the ␤ 3 ligand OPN or PDL, which interacts with cells through a non-integrin-mediated electrostatic interaction. Luciferase readings for cells treated on PDL correspond to basal levels of NF-B activation. Cells plated on the OPN-coated surface had enhanced luciferase activity (1.9-fold) relative to the PDL control as seen in Fig. 1, indicating that NF-B had been activated. Activation of NF-B was mediated by the ␣ v ␤ 3 integrin, because a ␤ 3 -neutralizing antibody reduced OPN-mediated luciferase activity by 78%, whereas an IgG control antibody was unable to significantly reduce luciferase activity. Collagen, a ligand for ␤ 1 integrins, failed to increase NF-B activation above basal levels (data not shown), further indicating the specificity of this pathway. Therefore, cellular adhesion to OPN through the integrin ␣ v ␤ 3 promoted NF-B activation in smooth muscle cells as well as endothelial cells.
The C Terminus of the ␤ 3 Subunit Was Necessary for Integrin-mediated NF-B Activation-Retroviral vectors containing human integrin ␤ 3 constructs, either wild-type or mutated, were created (Table I) and used to stably infect WKY cells.
Cells were incubated with an anti-human ␤ 3 antibody to detect surface expression of the integrin. All ␤ 3 constructs were efficiently expressed, with at least 80% of cells positive for surface expression of ␤ 3 . The level of surface expression was similar for each of the ␤ 3 constructs, as shown in Fig. 2. Due to difficulties acquiring antibodies that recognize rat integrins, ␤ 3 constructs were also expressed in human smooth muscle cells and surface expression of other integrins was detected by flow cytometry. Overexpression of ␤ 3 did not significantly affect ␣ v ␤ 5 , ␣ v ␤ 6 , or ␤ 1 integrin expression levels (data not shown). Assuming the same trend occurs in rat smooth muscle cells, then observed phenotypes in cells infected with the ␤ 3 constructs are due to overexpression of ␣ v ␤ 3 integrins and not changes in the expression profile of other integrins.
To test functionality of the ␤ 3 constructs, smooth muscle cells were seeded onto OPN-coated surfaces and allowed to attach for 2 h. Unattached cells were washed away, and phase contrast microscopy images of attached cells were acquired. All cell types attached to the OPN-coated surface (Fig. 3A), but substantial differences in post-receptor occupancy interactions were observed between the cell types. Overexpression of wildtype (WT) ␤ 3 dramatically increased cell spreading compared with vector control. Truncation of four amino acids from the C terminus (759X) eliminated the ability of the integrin to promote cell spreading, whereas removal of two amino acids did not impair cell spreading (761X). Substitution of tyrosine 759  with phenylalanine (Y759F) did not reduce cell spreading, but mutation to alanine (Y759A) resulted in an intermediate level of cell spreading.
To quantify these post-receptor occupancy differences, adhesion assays were performed. Compared with vector control, overexpression of WT ␤ 3 significantly enhanced cellular adhesion by 3-fold or greater at all OPN concentrations tested (Fig.  3B). 759X significantly reduced adhesion compared with WT ␤ 3 (p Ͻ 0.001), whereas 761X did not impair the ability of the integrin to promote adhesion to OPN. Substitution of tyrosine 759 with phenylalanine (Y759F) did not reduce cell adhesion, indicating that phosphorylation of this residue was not required. Alanine substitution at this residue (Y759A) allowed for adhesion, but it was impaired compared with WT ␤ 3 especially at the lowest OPN concentration tested. Thus the adhesion assay results agreed with the differences in cell spreading seen in Fig. 3A.
To test the ability of the integrins to mediate NF-B activation, smooth muscle cells expressing the NF-B-dependent luciferase reporter and the integrin ␤ 3 constructs were plated on surfaces coated with the ␤ 3 ligand OPN. After 7 h, adherent cells were lysed, and luciferase activity was measured and normalized to protein recovered. For ease of comparison, vector control samples were then normalized to a value of one. Compared with vector control cells, overexpression of full-length ␤ 3 or 761X significantly increased NF-B activation (p Ͻ 0.01), whereas 759X was unable to increase NF-B activation when plated on OPN (Fig. 3C). Integrins with the Y759F mutation retained the ability to activate NF-B, but WKY cells overexpressing Y759A integrins have significantly reduced activation of NF-B compared with WT ␤ 3 (p Ͻ 0.01). These results indicated that tyrosine 759 of integrin ␤ 3 had a critical structural role in OPN-mediated NF-B activation, because truncation at position 759 or substitution to alanine significantly inhibited activation of the transcription factor, but tyrosine phosphorylation was not required, since the Y759F mutation retained the ability to signal.
Pharmacological Inhibition of SFKs Reduced Integrin ␣ v ␤ 3 -Mediated NF-B Activation-Having identified that tyrosine 759 within the C terminus of ␤ 3 was required for NF-B activation, we performed a preliminary experiment to identify proximal downstream mediators of NF-B activation. The Src family of tyrosine kinases was recently reported to bind directly to ␤ 3 integrins through the C terminus, and its members are known proximal mediators of integrin signals (20). To test the role of SFKs in integrin-mediated NF-B activation, RAECs were pretreated with the Src family tyrosine kinase inhibitor PP2, the inactive analog PP3, or vehicle control and plated on OPN. Treatment with 10 M PP2 significantly inhibited integrin-mediated NF-B activation (p Ͻ 0.01), whereas 10 M PP3 or vehicle did not, as seen in Fig. 4. Similar results were seen when WKY smooth muscle cells were also treated with PP2 and PP3 (data not shown). Although PP2 inhibition is not complete at this dosage and higher doses of PP2 fail to yield further inhibition, 10 M PP2 does lead to greater than 75% reduction in integrin-mediated activation of the transcription factor. These results indicate that NF-B activation required SFK activity in this model.
Overexpression of Kinase-dead c-Src Blocked Integrin ␣ v ␤ 3 -Mediated NF-B Activation-To further characterize the role of SFKs, RAECs cells were infected with retroviral vectors encoding WT chicken c-Src (WT), kinase-dead c-Src (K295R), activated c-Src (Y527F), or the double mutant (K295R/Y527F) (13). As seen in Fig. 5A, lane 5, Western blots of total cell lysates indicated that the RAECs expressed endogenous SFKs (the antibody was unable to distinguish Src family members). Cells

FIG. 3. Overexpression of ␤ 3 integrin enhances cell adhesion, cell spreading, and activation of NF-B.
A, phase contrast microscopy images of rat smooth muscle cells overexpressing ␤ 3 integrins, 2 h after seeding onto OPN-coated surfaces (25 g/ml). Scale bar, 50 m. B, rat smooth muscle cells overexpressing ␤ 3 integrins (wild-type or mutated) were plated on OPN-coated surfaces. After 2 h non-adherent cells were removed, adherent cells were stained with toluidine blue, and absorbance was measured at 630 nm. C, rat smooth muscle cells overexpressing ␤ 3 integrins were plated on OPN-coated surfaces (1 g/ml). Cell lysates were harvested, luciferase activity was measured, and activity was normalized to total protein recovery. *, p Ͻ 0.01. infected with c-Src constructs have enhanced Src protein expression (Fig. 5A, lanes 1-4).
RAECs expressing the c-Src constructs were plated on PDLor OPN-coated surfaces. Expression of c-Src constructs did not affect cell adhesion to OPN (data not shown). RAECs infected with the control vector activated NF-B when plated on OPN as expected (Fig. 5B). Compared with vector control, overexpression of activated c-Src (Y527F) enhanced NF-B activation independent of substrate (p Ͻ 0.01), suggesting that c-Src was downstream of the integrin ␣ v ␤ 3 . Overexpression of kinasedead c-Src (K295R) acted as a dominant negative inhibitor, completely blocking OPN-mediated NF-B activation and reducing luciferase activity to levels seen on PDL. Kinase-dead c-Src may function either by preventing c-Src from phosphorylating a downstream substrate or by preventing autophosphorylation leading to a conformational change required for recruitment of other signaling moieties necessary for NF-B activation. The c-Src double mutant K295R/Y527F maintains an activated conformation allowing recruitment of these moieties but prevents c-Src-mediated phosphorylation of other substrates. Overexpression of the double mutant also completely eliminated NF-B activation (Fig. 5B), thus activation of the enzyme was not able to overcome the loss of its kinase activity. These results indicated that SFK kinase activity was necessary and sufficient for OPN-mediated NF-B activation in RAECs.
Genetic Ablation of SFKs Eliminated Integrin ␣ v ␤ 3 -Mediated Activation of NF-B-Embryonic fibroblasts derived from mice deficient for the Src tyrosine kinases members c-Src, Yes, and Fyn (SYF cells) have been isolated (5). We compared the ability of SYF cells and mouse embryonic fibroblasts, which express SFKs (Src8T), to induce OPN-mediated NF-B activation. Both cell types were stably infected with the retroviral vector encoding the NF-B-dependent luciferase reporter at comparable efficiencies (96% for Src8T cells versus 91% for SYF cells) and expression levels (data not shown). Plating Src8T cells on OPNcoated surfaces increased NF-B activation by Ͼ3-fold compared with PDL-coated surfaces (Fig. 6A). When plated on the ␤ 1 -ligand collagen, Src8T cells failed to activate NF-B. These results verified that mouse embryonic fibroblasts have the OPN/␣ v ␤ 3 /NF-B pathway. The SYF fibroblasts that lack SFKs failed to increase NF-B activation when plated on OPN compared with PDL, indicating that SFKs are necessary for OPNmediated NF-B activation in mouse embryonic fibroblasts.
SYF cells were infected with the mutated c-Src constructs, and expression of c-Src was verified by Western blot (data not shown). These cells were plated on PDL-or OPN-coated surfaces, and NF-B activity was measured. As seen in Fig. 6B, expression of WT c-Src restored OPN-mediated NF-B activation. Overexpression of kinase-dead c-Src (K295R) failed to rescue activation of the transcription factor, while expression of activated c-Src (Y527F) increased transcription factor activity regardless of the surface to which the cells were exposed. These results indicate that SFK kinase activity was necessary and sufficient for integrin ␣ v ␤ 3 -mediated NF-B activation in fibroblasts.
SFKs Associated with Wild-type ␤ 3 but Not 759X-A recent study in Chinese hamster ovary cells determined that SFKs interact directly with the cytoplasmic tails of integrin ␤ 3 subunits and the Y759F mutation in the ␤ 3 subunit does not inhibit the interaction (20). Thus, we hypothesized that the 759X mutation inhibited SFK association with the ␤ 3 subunit, preventing SFK-mediated signal transduction following integrin binding to OPN. To test this hypothesis, integrin ␤ 3 subunits from WKY cells expressing c-Src and ␤ 3 integrin subunits, WT or 759X, were immunoprecipitated as described under "Experimental Procedures." The lower panel of Fig. 7 demonstrates that human integrin ␤ 3 was recovered following immunoprecipitation with anti-␤ 3 antibodies from cells expressing WT ␤ 3 or 759X. Endogenous rat integrin ␤ 3 was also immunoprecipitated but was not detected by the antibody used in the Western blot (data not shown). As predicted, SFKs associated with endogenous integrin ␤ 3 as seen in lane 2 of  6 versus lane 2). These results verified that removal of four amino acids from the tail of the ␤ 3 integrin subunit prevented SFK association.
Syk Kinase Activity Was Not Required for Integrin ␣ v ␤ 3 -Mediated NF-B Activation-The non-receptor tyrosine kinase Syk also binds to the C terminus of the ␤ 3 subunit (21). To test the potential role of Syk in integrin ␣ v ␤ 3 -mediated NF-B activation, RAECs and WKY smooth muscle cells were infected with retroviral vectors encoding WT Syk or kinase-dead Syk (K402R). Overexpression of Syk protein was verified by Western blot as seen in Fig. 8A; however, neither construct was able to modulate NF-B activation in RAECs (Fig. 8B). Similar results were observed in WKY cells (data not shown). Therefore we conclude that Syk kinase was not involved in the integrin ␣ v ␤ 3 /NF-B signaling pathway. DISCUSSION RAEC binding to OPN-coated surfaces leads to integrin ␣ v ␤ 3 ligation and activation of the transcription factor NF-B (2). In this study, we demonstrated that integrin ␣ v ␤ 3 -mediated NF-B activation occurred in multiple cell types and required the C terminus of the integrin ␤ 3 subunit and SFK kinase activity. This was based on the findings that 1) smooth muscle cells activated NF-B when plated on the ␣ v ␤ 3 ligand OPN, 2) overexpression of the ␤ 3 subunit increased ␣ v ␤ 3 -mediated NF-B activation, 3) mutations to the C terminus (759X or Y759A) of the ␤ 3 subunit eliminated the ability of the integrin to activate NF-B, 4) SFKs associated with WT ␤ 3 but not 759X, and 5) pharmacological inhibition, overexpression of ki-nase-dead c-Src, or genetic ablation of SFKs inhibited activation of NF-B.
In this study, adhesion of smooth muscles cells or fibroblasts to the ␣ v ␤ 3 ligand OPN activated the transcription factor NF-B, whereas adhesion to surfaces coated with the ␤ 1 ligand collagen or PDL did not. Inhibition of integrin ␣ v ␤ 3 with a soluble neutralizing antibody for ␤ 3 prevented NF-B activation in smooth muscle cells. These results, combined with the previous finding that endothelial cells activate NF-B when plated on OPN, indicated that the OPN/integrin ␣ v ␤ 3 /NF-B pathway exists in multiple cell types. Fibronectin, a ligand for ␣ 5 ␤ 1 and ␣ v ␤ 3 integrins, has also been reported to activate NF-B in fibroblasts and smooth muscle cells (22). We found that overexpression of ␣ v ␤ 3 in smooth muscle cells led to enhanced NF-B activation on fibronectin (data not shown). Although these data did not rule out the potential role of ␣ 5 ␤ 1 in fibronectin-mediated NF-B activation, it indicated that the degree of NF-B activation correlated with ␣ v ␤ 3 expression and suggested that integrin ␣ v ␤ 3 -mediated NF-B activation was not limited to just the ligand OPN.
We believe that ␤ 3 integrin-mediated NF-B activation may be a common survival mechanism, particularly when cells adopt an invasive phenotype. Integrin ␣ v ␤ 3 and the transcription factor NF-B have already been linked to increased invasiveness and survival in melanoma cells. OPN binding to the integrin ␣ v ␤ 3 activates NF-B in melanoma cells (23,24). Blockade of integrin ␣ v ␤ 3 suppresses melanoma growth by inducing apoptosis of the tumor cells (25). Inhibition of NF-B in melanoma cells also inhibits tumor growth (26). Combining these observations, a general picture emerges where ligation of integrin ␣ v ␤ 3 , a promiscuous integrin capable of binding multiple ECM components, leads to activation of NF-B, which is necessary for cell survival during migration and invasion.
In the second part of this project, we overexpressed mutated ␤ 3 integrins to identify regions of the cytoplasmic tail that were required for NF-B activation. The cytoplasmic tail of the ␤ 3 integrin is short (47 amino acids) and has no known enzymatic function. We found that removal of four or more amino acids from the C terminus (759X) of the ␤ 3 integrin subunit eliminated its ability to promote cell adhesion, cell spreading, and NF-B activation in response to the ␤ 3 ligand OPN. Tyrosine phosphorylation of residue 759 on integrin ␤ 3 was not required, because cells overexpressing ␤ 3 integrins with the Y759F mutation were able to adhere to OPN and activate NF-B. The adhesion results were consistent with previous studies. In Chinese hamster ovary cells, overexpression of ␣ IIb ␤ 3 enhanced cell adhesion and spreading on fibrinogen, which required the cytoplasmic tail of ␤ 3 but not ␣ IIb (27). 759X significantly reduced cell spreading and adhesion plaque formation, although complete loss of function required removal of six or more amino acids (757X) (12). The Y759A mutation to the ␤ 3 integrin subunit also inhibited cell spreading and adhesion plaque formation in this model. Previous studies investigating the ␣ v ␤ 3 integrin observed similar findings. Overexpression of ␤ 3 in Chinese hamster ovary cells led to increased ␣ v ␤ 3 surface expression, enhanced cell spreading, and elevated integrin-mediated signal transduction as demonstrated by FAK phosphorylation (28). The Y759A mutation significantly impaired these processes, whereas Y759F did not. These results agreed with our finding that ␤ 3 integrins required structural integrity of the NITY motif, rather than tyrosine phosphorylation of residue 759. Finally, overexpression of isolated integrin ␤ 3 cytoplasmic tails inhibited cell spreading in a dominant negative fashion, but an alternatively spliced form of ␤ 3 that lacked the NITY domain did not (29). These results are all consistent with the role of the NITY domain in cell adhesion and outside-in signaling.
In the final part of this project, we studied the role of SFKs during integrin ␣ v ␤ 3 -mediated NF-B activation. SFKs associated with the integrin ␣ v ␤ 3 and were activated following integrin ligation in multiple cell types, including osteoclasts (30), melanoma cells (31), and platelets (32). Likewise, we found that SFKs associated with WT ␤ 3 integrin subunits in smooth muscle cells. Truncation of the C terminus of the ␤ 3 integrin (759X) eliminated this interaction. These results agreed with the findings that SFKs associated directly with ␤ 3 integrins, but not ␤ 1 or ␤ 2 integrins, and removal of the cytoplasmic tail of ␤ 3 or competition with a C-terminal ␤ 3 peptide (748 -762) eliminated the association (20). This group also observed that tyrosine phosphorylation of residue 759 in integrin ␤ 3 was not required for the interaction with SFKs. The need for SFK kinase activity in NF-B activation was then tested by pharmacological inhibition of SFKs with PP2, overexpression of kinase-dead c-Src, or genetic ablation of SFKs. All three methods inhibited integrin ␣ v ␤ 3 -mediated NF-B activation. Likewise, mutations that prevented SFK association with integrin ␤ 3 , such as 759X, also eliminated integrin ␣ v ␤ 3 -mediated NF-B activation, whereas mutations that did not affect SFK binding to ␤ 3 such as Y759F retained the ability to activate the transcription factor. Finally, overexpression of constitutively active c-Src activated NF-B independently of the substrate, suggesting that SFKs were downstream of the integrin ␣ v ␤ 3 . These results indicated that SFK kinase activity was necessary and sufficient for integrin ␣ v ␤ 3 -mediated activation of NF-B.
SFKs associated with ␣ v ␤ 3 integrins through an interaction with the C terminus of the ␤ 3 subunit. The C terminus, including the NITY motif, is unique to the ␤ 3 integrin subunit, which may explain the specificity of the interaction of c-Src with ␤ 3 tails but not other integrin tails (20). This finding may also explain the specificity of ␤ 3 integrins for NF-B activation. Overall, these results suggest a model in which integrin ␣ v ␤ 3 ligation activates SFKs, leading to SFK kinase activity, which is essential for NF-B activation. To support this model, ligation of ␤ 3 integrins with OPN has been shown to activate c-Src in melanoma cells (31) and breast cancer cells (33). The exact mechanism of SFK-mediated NF-B activation is unknown at this time, although regulators of NF-B, including IB␣, IB kinases, and NF-B-inducing kinase, have all been implicated as substrates for SFKs (34,35). Future experiments will attempt to identify the mechanism of SFKs in integrin ␣ v ␤ 3mediated NF-B activation.
In summary, integrin ␣ v ␤ 3 ligation in multiple cell types, including endothelial cells, smooth muscle cells, and fibroblasts, activated the transcription factor NF-B. Mutations to regions within the C terminus cytoplasmic tail of ␤ 3 integrins or inhibition of SFKs blocked this activation. To our knowledge, this is the first report of SFK kinase activity being required for integrin-mediated activation of NF-B, an important pro-survival transcription factor.