Integrin α_vβ₃ Controls Activity and Oncogenic Potential of Primed c-Src Free

Interactions of tumor cells with their microenvironment are important for cancer development and progression (1). Tumor cells connect with the extracellular matrix through various members of the integrin family of adhesion receptors and upon malignant transformation cells often undergo specific changes in the expression levels of integrins. High levels of integrin α_vβ₃ correlate with growth and/or progression of melanoma (2, 3), neuroblastoma (4), and multiple different carcinomas (5–9). Moreover, individuals homozygous for the β₃^L33P polymorphism that enhances the affinity of β₃ integrins have an increased risk to develop breast cancer, ovarian cancer, and melanoma (10). Despite the fact that α_vβ₃ in the tumor vasculature has been identified as a valuable drug target, endothelial α_vβ₃ is dispensable for tumorigenesis (11, 12). It remains unclear if and how increased levels of α_vβ₃ on tumor cells contribute to cancer development.

Following ligand binding, integrins cluster and organize into multiprotein complexes termed cell-matrix adhesions that connect to the actin cytoskeleton through a variety of cytoskeletal linker proteins. Cell-matrix adhesions also contain various signaling intermediates, including non–receptor tyrosine kinases (non-RTK) such as focal adhesion kinase (FAK) and c-Src (13). Integrin-mediated adhesion stimulates FAK and c-Src activities, and, in turn, c-Src modulates the stability of cell-matrix adhesions through phosphorylation of several components, including integrin cytoplasmic tails (14–16). In addition, the FAK/c-Src complex is involved in the transmission of information from the extracellular matrix into the cell to regulate cellular signaling cascades in control of apoptosis and proliferation (17).

In unstimulated cells, c-Src is folded into a closed, autoinhibitory conformation. Its activation requires dephosphorylation of the COOH-terminal Tyr⁵³⁰ residue (amino acid numbering used in this study is for human c-Src) to disrupt intramolecular binding of this residue to the Src homology 2 (SH2) domain. A disruption of the interaction between the SH3 domain and prolines in the linker region further contributes to the formation of an unfolded or “primed” conformation. Finally, for full enzymatic activity, primed c-Src must be phosphorylated in its kinase domain at residue Tyr⁴¹⁹ by transphosphorylation (18, 19). The oncogenic product of Rous sarcoma virus (v-Src) is constitutively activated through amino acid substitutions in the SH3 domain and the kinase domain as well as a deletion of the regulatory COOH-terminal tyrosine (18, 20). Although Rous sarcoma virus is avian specific, c-Src plays a critical role in cancer development (21, 22). Indeed, levels of c-Src activity are frequently increased in human melanoma and carcinomas of the breast, colon, and other epithelia (23–25). It is incompletely understood how c-Src activity is enhanced in tumors. Increased levels of c-Src and binding of overexpressed RTKs to the c-Src SH2 domain may enhance c-Src priming. In addition, mutations in the SRC gene stabilizing a primed conformation of c-Src through truncation of the regulatory COOH terminus have been detected in colon and endometrial cancer, although such mutations seem to be rare (26–29).

Because (a) c-Src selectively mediates signaling by β₃ integrins (30), (b) null mutations in the Src or the Itgb3 gene give rise to partially overlapping abnormalities (31, 32), and (c) increased expression or activity of α_vβ₃ or c-Src has been associated with growth, progression, or poor prognosis of the same types of cancer (2–9, 23–25), we hypothesized that a functional interaction of α_vβ₃ with c-Src may contribute to cancer development. In this report, we show that the activity and oncogenic potential of primed c-Src is in fact subject to a remarkably tight regulation by integrin α_vβ₃. Our findings identify the β₃ cytoplasmic domain as a critical regulator of c-Src–mediated oncogenic signaling.

Cell lines and plasmids. The HBL100 cell line was obtained from the American Type Culture Collection (Rockville, MD). The β₁ integrin–deficient cell lines GD25 and GE11 were provided by Dr. Reinhard Fässler (Max Planck Institute, Martinsried, Germany) and have been described previously (33, 34). All cell lines were cultured in DMEM supplemented with 10% FCS, penicillin, and streptomycin. A Myc tag was added at the 3′ end of the cDNA encoding c-Src^Y530F (the plasmid encoding mouse-chicken c-Src in which the COOH-terminal regulatory Tyr was replaced by Phe was purchased from Upstate Biotechnology, Lake Placid, NY); a hemagglutin (HA) tag was added at the 3′ end of the cDNA encoding mouse c-Src (Upstate Biotechnology); and the tagged constructs were cloned into the LZRS retroviral vector. Retroviral expression plasmids encoding integrin β₁ or β₃ subunits, β₁^exβ₃ⁱⁿ and β₃^exβ₁ⁱⁿ chimeras, and those encoding the extracellular and transmembrance region of the non-signaling interleukin 2 receptor α (IL2Rα) subunit alone or fused to the integrin β₁ cytoplasmic domain were described before (33, 34). To generate the LZRS-IL2Rβ₃ plasmid, the β₁ cytoplasmic domain in IL2Rβ₁ was replaced with the β₃ cytoplasmic domain. The retroviral expression plasmid encoding ts72v-Src (35) was provided by Dr. Irwin H. Gelman (Roswell Park Cancer Institute, Buffalo, NY). The LZRS retroviral construct expressing chimeric vSrc/Src^YF was generated by substituting the first 131 amino acids of Src^YF with the same region from ts72v-Src. The cDNA encoding human epidermal growth factor receptor (EGFR) was provided by Dr. Frank Furnari (Ludwig Institute for Cancer Research, La Jolla, CA) and cloned into the pMSCV retroviral expression plasmid by Sophia Bruggeman (The Netherlands Cancer Institute, Amsterdam, The Netherlands). The retroviral H-Ras^G12V expression plasmid (Ras^GV) was provided by Dr. John Collard (The Netherlands Cancer Institute). The β₃^Y747A, β₃^Y759A, and β₃^Δ759 mutants were provided by Dr. Jari Ylänne (University of Oulu, Finland) and subcloned into the LZRS retroviral vector (36). All cDNAs were transfected into amphotrophic or ecotrophic packaging cells to generate virus-containing culture supernatants that were used for retroviral transduction of HBL100, GD25, and GE11 cells. Subsequently, Src^YF, c-Src, ts72v-Src, vSrc/Src^YF, or Ras^GV expressing clones were transduced with retroviral constructs encoding wild-type, mutant, and chimeric integrin subunits or EGFR. Positive cells were bulk sorted at least twice by fluorescence-activated cell sorting for human integrin, IL2Rα, or EGFR surface expression.

Antibodies and other materials. Anti-human β₁ monoclonal antibodies were TS2/16, clone 18 (BD Transduction Laboratories, Lexington, KY), and K20 (Biomeda, San Francisco, CA). Anti-human β₃ monoclonal antibodies were C17 (provided by Dr. Ellen van der Schoot, Sanquin, Amsterdam, The Netherlands), 23C6 (provided by Dr. Michael Horton, University College London, United Kingdom), and SSA6 (provided by Dr. Sanford Shattil, University of California San Diego, CA). Other monoclonal antibodies were anti-c-Src (B-12; Santa Cruz Biotechnology, Santa Cruz, CA), anti-α-tubulin (B-5-1-2; Sigma, St. Louis, MO), anti-β-actin (AC-15; Sigma), anti-EGFR (Ab-1 clone 528; Calbiochem, La Jolla, CA), anti–phosphorylated signal transducer and activator of transcription 3(Y⁷⁰⁵) [anti-p-Stat3(Y⁷⁰⁵); 3E2; Cell Signalling Technology], and anti-bromodeoxyuridine (anti-BrdUrd; Bu20a; DAKO, Carpinteria, CA). The following rabbit polyclonal antibodies were used: anti–phosphorylated Src(Y⁴¹⁹) [anti-p-Src(Y⁴¹⁹); Biosource, Camarillo, CA], anti-c-Src (SRC 2; Santa Cruz Biotechnology), anti-myc (A-14; Santa Cruz Biotechnology), anti-HA (GeneTex, Inc., San Antonio, TX), anti-FAK (C-20; Santa Cruz Biotechnology), anti–phosphorylated FAK(Y⁹²⁵) [anti-p-FAK(Y⁹²⁵); Biosource], anti-human β₁ (provided by Dr. Ulrike Mayer, University of Manchester, United Kingdom), anti-Stat3 (K-15; Santa Cruz Biotechnology), anti-human IL2Rα (N-19; Santa Cruz Biotechnology), and polyclonal goat anti-human β₃ (N-20; Santa Cruz Biotechnology). Texas Red–conjugated phalloidin was purchased from Molecular Probes (Eugene, OR). Human plasma fibronectin was prepared as described previously (34).

Flow cytometry, immunofluorescence, and Western blot analysis. These experiments were done as described previously (34).

Immunoprecipitations. For immunoprecipitations, cells were lysed for 15 min at 4°C in lysis buffer [1% NP40, 50 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl, 1 mmol/L sodium vanadate, 0.5 mmol/L sodium fluoride, and protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO)]. Lysates were clarified by centrifugation at 14,000 rpm for 15 min at 4°C and precleared with protein A-Sepharose (Amersham Pharmacia Biotech AB, Uppsala, Sweden) for 2 h at 4°C. Proteins were immunoprecipitated o/n at 4°C with antibodies to c-Src (B-12), β₁ (K20), or β₃ (SSA6), coupled to protein A-Sepharose.

For in vitro Src kinase assays, cells were lysed for 15 min at 4°C in lysis buffer [0.5% NP40, 50 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl, 5 mmol/L MgCl₂, 1 mmol/L sodium vanadate, and protease inhibitor cocktail (Sigma-Aldrich)]. Lysates were clarified by centrifugation at 14,000 rpm for 15 min at 4°C. Src^YF protein was isolated from the clarified lysates by immunoprecipitation with 5 μg anti-myc antibody for 2 h at 4°C, and immune complexes were collected with protein A-Sepharose. Kinase activity of isolated Src^YF protein was determined by use of a Src kinase assay kit (Upstate Cell Signaling Solutions, Charlottesville, VA).

Soft agar and tumorigenicity assays. For soft agar assays, six-well plates were first coated with 0.6% low melting point (LMP) agarose (Roche, Indianapolis, IN). Subsequently, 100,000 cells were suspended in culture medium containing 0.35% LMP agarose and seeded on top of the 0.6% LMP agarose layer. For tumorigenicity assays, cells were harvested, washed, and resuspended in 0.2 mL sterile PBS per injection. Female 6-week-old athymic BALB/c mice were then s.c. injected into the left and right flanks. After cell inoculation, tumor volumes were measured using calipers at the indicated times. All animal experiments were approved by the animal welfare committee of the Netherlands Cancer Institute.

Terminal deoxynucleotidyl transferase–mediated nick-end labeling staining and BrdUrd incorporation assays. Cells (75,000) were plated in culture medium on fibronectin coated coverslips, and after 4 h, the cells were either kept in culture medium or switched to serum-free medium for 24 or 48 h. The cells were labeled with 15 μmol/L BrdUrd (Sigma) for 4 h before fixation in 2% paraformaldehyde. For BrdUrd staining, cells were permeabilized in 0.5% Triton X-100; DNA was denatured with 2 mol/L HCl and neutralized with 0.1 mol/L sodium borate; and coverslips were labeled with anti-BrdUrd antibody followed by FITC-conjugated secondary antibody. For terminal deoxynucleotidyl transferase–mediated nick-end labeling (TUNEL) staining, cells were permeabilized in 0.1% Triton X-100 with 0.1% sodium citrate in PBS and stained with an in situ cell death detection kit (Roche). For both procedures, nuclei were visualized with TOPRO-3 (Molecular Probes), and preparations were mounted in MOWIOL 4-88 solution supplemented with DABCO (Calbiochem).

Integrin α_vβ₃ supports primed c-Src–mediated tumor growth. A c-Src mutant in which a primed conformation is induced by the substitution of Tyr⁵³⁰ by Phe (Src^YF) was introduced alone or in combination with the β₃ integrin subunit in HBL100, GD25, and GE11 cells. Although Src^YF-transformed HBL100 cells were tumorigenic, tumors grew much faster when the surface expression levels of α_vβ₃ were increased (Supplementary Fig. S1). The cooperation between α_vβ₃ and Src^YF was even more striking in GD25 and GE11: whereas cells expressing β₁ integrins (GDSrc^YFβ1 and GESrc^YFβ1) were virtually unable to grow in soft agar or form tumors in mice, cells lacking β₁ integrins but expressing high levels of α_vβ₃ (GDSrc^YFβ3 and GESrc^YFβ3) were highly tumorigenic (Fig. 1; Supplementary Fig. S2A; see ref. 34). Ectopic expression of β₁ integrins in GESrc^YFβ3 cells did not affect tumor growth, indicating that α_vβ₃ supports Src^YF-mediated tumor formation, irrespective of the expression of β₁ integrins (Fig. 2A). Moreover, when a 1:1 mixture of GESrc^YFβ1 and GESrc^YFβ3 cells was injected s.c., GESrc^YFβ3 cells (recognized by an antibody directed against human β₃) were readily detectable in the resulting tumors, whereas GESrc^YFβ1 cells (recognized by an antibody directed against human β₁) were virtually absent, indicating that α_vβ₃ supports Src-mediated tumor formation in a cell-autonomous fashion (Fig. 2B).

In contrast to Src^YF, Ras^GV-mediated tumor growth was not affected by the expression of α_vβ₃, indicating that this integrin is specifically required for Src-mediated tumorigenesis (Fig. 2C). In several human cancers, overexpressed and/or activated EGFR can stimulate priming of c-Src through (in)direct interactions with its SH2 domain (19, 23). We analyzed if α_vβ₃ can also support c-Src–mediated tumor growth under such conditions. Moderate overexpression of c-Src was not sufficient by itself to induce tumor formation even in the presence of α_vβ₃ (Fig. 2D; Supplementary Fig. S2B). However, although tumors grew slow compared with those induced by Src^YF, α_vβ₃ significantly increased tumor growth when c-Src and EGFR were coexpressed (Fig. 2D; Supplementary Fig. S2C).

Together, these findings indicate that α_vβ₃ specifically cooperates with primed c-Src in a cell-autonomous fashion to stimulate the formation of tumors by fibroblasts and epithelial cells.

α_vβ₃ supports Src^YF-mediated survival, proliferation, and tumor formation upstream of FAK and Stat3 by enhancing Src^YF activation. The activation of the transcription factor Stat3 by phosphorylation at Tyr⁷⁰⁵ is strongly enhanced in cells transformed by v-Src (37, 38). Src^YF only moderately increased Stat3 activity in the presence of β₁ integrins, whereas phosphorylation was clearly enhanced in the presence of α_vβ₃ (Fig. 3A). This suggested that α_vβ₃ may enhance Src^YF activity, which, like wild-type c-Src, requires transphosphorylation of Tyr⁴¹⁹ in its catalytic domain to acquire full enzymatic activity. Indeed, whereas expression of Src^YF led to increased levels of p-Src(Y⁴¹⁹) in the presence of either β₁ or β₃ integrins, phosphorylation was much stronger in GESrc^YFβ3 than in GESrc^YFβ1 cells (Fig. 3B). Stimulation of the levels of p-Src(Y⁴¹⁹) and of two known Src substrates [p-FAK(Y⁹²⁵) and p-Stat3(pY⁷⁰⁵)] in the presence of α_vβ₃ was also observed when cells were cultured under conditions that may mimic the tumor environment (serum-free or nonadherent; Fig. 3C; data not shown). On the other hand, equal levels of Src activity were detected in in vitro Src kinase assays on Src^YF immunoprecipitates from GESrc^YFβ1 and GESrc^YFβ3 lysates (Fig. 3D). These data raise the possibility that in vivo activation of Src^YF is regulated by α_vβ₃, possibly by enhanced clustering and subsequent transphosphorylation in the kinase domain. Concentration of Src^YF on the beads in the in vitro Src kinase assays might result in activation of Src^YF in a similar way.

To investigate the role of the β₃ cytoplasmic tail in Src^YF activation and oncogenic potential, we expressed a chimeric integrin subunit consisting of the β₃ extracellular and transmembrane regions fused to the β₁ cytoplasmic region (β₃^exβ₁ⁱⁿ) in GESrc^YF cells (Supplementary Fig. S2A, left). Unlike wild-type β₃, β₃^exβ₁ⁱⁿ did not enhance the levels of p-Src(Y⁴¹⁹), p-Stat3(Y⁷⁰⁵), and p-FAK(Y⁹²⁵) (Fig. 3C). On the other hand, an inverse chimeric β₁^exβ₃ⁱⁿ integrin subunit, unlike wild-type β₁, strongly increased the degree of phosphorylation of Src, Stat3, and FAK (Fig. 3C; Supplementary Fig. S2A, right). Moreover, surface expression of β₁^exβ₃ⁱⁿ supported Src^YF-mediated tumor formation, whereas expression of β₃^exβ₁ⁱⁿ did not (Fig. 3E).

We next investigated if increased levels of α_vβ₃ promote survival and proliferation of Src^YF-transformed cells. After 24 and 48 h of serum deprivation, there were significantly fewer TUNEL-positive GESrc^YFβ3 than GESrc^YFβ1 cells (Fig. 4A; Supplementary Fig. S3A; P < 0.05, Student's t test). In agreement with the role of the β₃ cytoplasmic tail in tumor formation induced by Src^YF, the sensitivity to serum starvation was suppressed in the presence of the β₁^exβ₃ⁱⁿ but not the β₃^exβ₁ⁱⁿ chimeric integrin subunit (P < 0.05). Furthermore, following serum deprivation the expression of β₃ or β₁^exβ₃ⁱⁿ correlated with high proliferation rates, whereas a large proportion of the cells expressing β₁ or β₃^exβ₁ⁱⁿ underwent cell cycle arrest (P < 0.05; Fig. 4B; Supplementary Fig. S3B).

From these findings, we conclude that (a) the primed Src^YF mutant is in fact subject to tight regulation in vivo; (b) as part of a functional integrin, the β₃ cytoplasmic tail is required and sufficient to support the activity of Src^YF under conditions that mimic the tumor environment; and (c) the ability of the β₃ cytoplasmic tail to support Src^YF activation is correlated with its ability to support Src^YF-mediated survival, proliferation, and tumor formation.

Functional, spatial, and molecular association of the β₃ cytoplasmic tail with Src^YF. Having shown that the β₃ cytoplasmic tail controls the activity and oncogenic potential of Src^YF, we asked two questions: Is Src^YF-mediated phosphorylation of the β₃ tail involved? And can the β₃ tail on its own support oncogenic signaling by primed Src? Both β₃^Y747A and β₃^Y759A mutants stimulated tumor growth of GESrc^YF cells to the same extent as wild-type β₃, indicating that recruitment of signaling and adaptor proteins to phosphorylated tyrosine residues in the β₃ cytoplasmic tail is not required (Fig. 5A; Supplementary Fig. S4A). On the other hand, an IL2R fusion construct containing the β₃ cytoplasmic tail did not enhance tumor growth of GESrc^YF cells compared with IL2R- or IL2R-β1, indicating that the cooperation between the β₃ cytoplasmic tail and Src^YF requires the context of a functional integrin (Fig. 5B; Supplementary Fig. S4B).

We next investigated if Src^YF formed a complex with the β₃ subunit but did not detect an interaction using coimmunoprecipitation, whereas an interaction of Src^YF with endogenous FAK was readily detectable (Fig. 5C; Supplementary Fig. S4C; data not shown). To investigate if a possible weak interaction may occur, we analyzed the subcellular localization of Src^YF and β₃ integrins. Irrespective of the type of integrins expressed, Src^YF induced the formation of podosomes, adhesive structures that are characteristic for Src-transformed cells. Notably, this indicates that the activity of Src^YF in cells lacking high amounts of α_vβ₃ is sufficient to cause morphologic but not oncogenic transformation. β₁ and β₃ integrins were partially colocalized with Src^YF in podosomes (Fig. 5D,, left and middle), making it possible that the β₃ cytoplasmic tail locally interacts with Src^YF and enhances Src^YF-mediated oncogenic signaling. In line with this idea, IL2Rβ3 did not colocalize with Src^YF in podosomes, which may explain its inability to support Src^YF-mediated oncogenic transformation (Fig. 5D , right).

The last four amino acids of the integrin β₃ tail have been reported to mediate binding of α_IIbβ₃ to the SH3 domain of c-Src (16). A similar interaction may explain α_vβ₃-mediated control of the oncogenic potential of primed c-Src. Experiments using v-Src, which contains multiple mutations in its SH3 domain (18), showed efficient phosphorylation on Tyr⁴¹⁹ and colony outgrowth in soft agar irrespective of α_vβ₃ expression levels (Fig. 6A and B). However, v-Src also contains activating mutations in its kinase domain (18) that may make the interaction with the β₃ cytoplasmic domain redundant. Therefore, we generated a v-Src/Src^YF chimera in which only the NH₂-terminal region including the SH3 domain was derived from v-Src. This construct failed to induce oncogenic transformation even in the presence of high levels of α_vβ₃ (Fig. 6C; Supplementary Fig. S4D). Moreover, expression of a β₃^Δ759 mutant that lacks the four most COOH-terminal amino acids required for binding the c-Src SH3 domain (16) failed to support Src^YF-mediated tumor formation (Fig. 6D).

Together, these data support the idea that oncogenic activity of primed Src variants containing a wild-type kinase domain depends on SH3-mediated interactions with the β₃ cytoplasmic domain.

In human melanomas and carcinomas of the breast, colon, pancreas, and other organs, the activity of c-Src is frequently increased compared with that in surrounding tissue (23–25). It is not fully understood how high levels of c-Src activity contribute to human cancer. The fact that activating mutations in the SRC gene seem to be rare argues against a role in tumor initiation. Increased c-Src activity may contribute to invasion and metastasis by promoting tumor cell scattering, migration, proteolytic activity, and anoikis resistance (39, 40). For colon cancer, increased c-Src activity may also contribute to tumor growth (41), perhaps by stimulating vascular endothelial growth factor–mediated angiogenesis (42). Our findings clearly show that elevated c-Src activity promotes tumorigenicity of immortalized cells where the p53 and Rb tumor suppressor pathways are suppressed, as is almost invariably the case in human cancer. We find that elevated c-Src activity promotes tumor growth in a cell-autonomous fashion by stimulating survival and proliferation. This may be especially important during early stages of cancer development. The contribution of elevated levels of c-Src activity to tumor growth may decrease as tumors progress and acquire additional mutations (e.g., those activating Ras).

The molecular mechanism responsible for the increased activity of c-Src in human cancers is incompletely understood. Overexpression of RTKs has been proposed to induce a primed conformation of c-Src by disrupting the intramolecular binding of the SH2 domain to phosphorylated Tyr⁵³⁰ (19, 23). In addition, mutations in the COOH-terminal region of the SRC gene that lead to a primed conformation of c-Src have been detected in a small subset of carcinomas of the colon and the endometrium (26, 27). Whatever the mechanism of priming, our findings show that the oncogenic potential of primed c-Src can be strongly enhanced by integrin α_vβ₃. The notion that α_vβ₃ and c-Src may cooperate in human cancer is supported by a number of reports showing that an increase in the expression of α_vβ₃ is associated with growth and/or progression of various cancers in which c-Src activity is frequently enhanced (2–9, 23–25). The loss of α₅β₁ or other β₁ integrins has also been associated with oncogenic transformation and tumor growth (43, 44). We observed that β₁-deficient Src^YF transformed cells were slightly more tumorigenic than their β₁-expressing derivatives (data not shown). However, expression of β₁ in Src^YFβ3 cells did not reduce their tumorigenic capacity, indicating that α_vβ₃ supports oncogenic signaling by primed Src, irrespective of the expression of β₁ integrins. The expression of α_vβ₃ will be important for Src-mediated aspects of cancer development, whereas α_vβ₃ may be dispensable for those aspects that are driven by oncogenic Ras (our findings) or other oncogenes such as c-Neu (11).

The interaction between the β₃ cytoplasmic tail and the c-Src SH3 domain has been shown by others (16), but we were unable to detect the interaction of the β₃ tail and primed c-Src by coimmunoprecipitation (possibly due to the much lower levels of expression). Nevertheless, several lines of evidence support a model in which this interaction controls the oncogenic potential of primed c-Src: (a) only those integrins and integrin chimeras that contain the β₃ cytoplasmic tail promote oncogenic signaling by primed c-Src; (b) in contrast to full-length β₃, an IL2Rβ₃ fusion construct fails to colocalize with primed Src in podosomes and fails to support tumor growth; (c) the YRGT motif in the β₃ cytoplasmic domain that was reported to interact with the SH3 domain of c-Src (16) is required for the functional interaction of α_vβ₃ with primed c-Src; and (d) α_vβ₃ cannot stimulate primed c-Src variants containing the SH3 domain of v-Src despite the fact that both SH2- and SH3-mediated autoinhibition is prevented. The β₃ subunit has a tendency to form homo-oligomers and clustering of α_vβ₃ in the plane of the membrane may cocluster primed c-Src, leading to enhanced activation through cross-phosphorylation in the kinase domain (45, 46). Indeed, such intermolecular autophosphorylation is considered the major mechanism underlying c-Src activation (47). In addition to clustering, α_vβ₃ may support conformational alterations in the Src protein or recruit additional proteins that contribute to oncogenic signaling. In this respect, our results using Tyr to Ala mutants argue against a role for the recruitment of signaling or adaptor proteins to the conserved NpxY/NxxY motifs in the β₃ cytoplasmic tail.

In conclusion, a functional interaction with the β₃ cytoplasmic tail augments the activity and oncogenic potential of primed c-Src. Phosphorylation of FAK and Stat3 are enhanced in the presence of α_vβ₃, but it remains to be investigated if these or other downstream pathways underlie the synergistic effect of primed c-Src and α_vβ₃ on survival, proliferation, and tumor growth. As overexpression of α_vβ₃ and elevated levels of c-Src activity occur in the same types of tumors, the interaction of these two proteins may be an important event in cancer development and/or progression. Interfering with their interaction might therefore be a valuable therapeutic approach in melanomas and carcinomas of the breast, colon, and several other tissues. Moreover, a combinatorial analysis of the levels of integrin α_vβ₃ and c-Src may be useful to predict cancer development and/or progression.

Grant support: Dutch Cancer Society grant NKI 2003-2858 (S. Huveneers and E.H.J. Danen) and NKI 2001-2488 (I. van den Bout).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Sophia Bruggeman, John Collard, Reinhard Fässler, Frank Furnari, Irwin Gelman, Michael Horton, Ulrike Mayer, Sanford Shattil, Ellen van der Schoot, and Jari Ylänne for their generous gifts of reagents; Ingrid Kuikman for the generation of the LZRS-β₃^Δ759 vector; and Anton Berns, John Collard, Ed Roos, Marc Vooijs, Bob van de Water, and members of the Sonnenberg group for critical reading of the article.