Substrate Specificity of αvβ3Integrin-mediated Cell Migration and Phosphatidylinositol 3-Kinase/AKT Pathway Activation*

The αvβ3integrin has been shown to bind several ligands, including osteopontin and vitronectin. Its role in modulating cell migration and downstream signaling pathways in response to specific extracellular matrix ligands has been investigated in this study. Highly invasive prostate cancer PC3 cells that constitutively express αvβ3adhere and migrate on osteopontin and vitronectin in an αvβ3-dependent manner. However, exogenous expression of αvβ3 in noninvasive prostate cancer LNCaP (β3-LNCaP) cells mediates adhesion and migration on vitronectin but not on osteopontin. Activation of αvβ3 by epidermal growth factor stimulation is required to mediate adhesion to osteopontin but is not sufficient to support migration on this substrate. We show that αvβ3-mediated cell migration requires activation of the phosphatidylinositol 3-kinase (PI 3-kinase)/protein kinase B (PKB/AKT) pathway since wortmannin, a PI 3-kinase inhibitor, prevents PC3 cell migration on both osteopontin and vitronectin; furthermore, αvβ3 engagement by osteopontin and vitronectin activates the PI 3-kinase/AKT pathway. Migration of β3-LNCaP cells on vitronectin also occurs through activation of the PI 3-kinase pathway; however, AKT phosphorylation is not increased upon engagement by osteopontin. Furthermore, phosphorylation of focal adhesion kinase (FAK), known to support cell migration in β3-LNCaP cells, is detected on both substrates. Thus, in PC3 cells, αvβ3mediates cell migration and PI 3-kinase/AKT pathway activation on vitronectin and osteopontin; in β3-LNCaP cells, αvβ3 mediates cell migration and PI 3-kinase/AKT pathway activation on vitronectin, whereas adhesion to osteopontin does not support αvβ3-mediated cell migration and PI 3-kinase/AKT pathway activation. We conclude therefore that αvβ3 exists in multiple functional states that can bind either selectively vitronectin or both vitronectin and osteopontin and that can differentially activate cell migration and intracellular signaling pathways in a ligand-specific manner.

Integrins are heterodimeric cell surface receptors that consist of noncovalently associated ␣ and ␤ subunits; these receptors have been shown to play a role in cell migration, prolifer-ation, and gene transcription and can affect cancer cell invasion and growth (1)(2)(3). The role of ␣ v ␤ 3 integrin in mediating cell migration and survival has been described (4 -6). Exogenous expression of ␣ v ␤ 3 has been shown to increase melanoma tumor growth and metastases (7,8), to induce conversion from radical to vertical growth phase in primary human melanoma cells, and to promote melanoma cell survival in vivo (9) and in three-dimensional collagen gels (10) indicating ␣ v ␤ 3 contribution at the level of motility and proliferation in vivo.
We have previously shown that highly invasive and metastatic human PC3 prostate cancer cells express ␣ v ␤ 3 whereas nontumorigenic and noninvasive LNCaP cells do not (5).
OPN is expressed in mature bone where prostate cancer cells preferentially metastasize. A causal role for OPN during tumor progression has been suggested by several studies, including the observation that high levels of OPN support a tumorigenic and metastatic phenotype (21,22). OPN is up-regulated in prostate cancer and other carcinomas (23,24) and increases anchorage-independent growth of prostate cancer cells (22) as well as proliferation of normal prostate cells (25). Furthermore, it is found in plasma of patients with metastatic diseases, and it increases metastatic ability of transformed cells (26,27). Its interaction with different surface receptors has been shown: specifically, ␣ 4 ␤ 1 , ␣ 8 ␤ 1 , ␣ 9 ␤ 1 , ␣ v ␤ 1 , ␣ v ␤ 5 , and CD44 on different cell types (28 -35). The ability of OPN to support haptotaxis of different cell types via ␣ v ␤ 3 has been shown (16). However, the signaling mechanisms activated via OPN-␣ v ␤ 3 interaction that support cell migration have never been described.
It has recently been shown that ␣ v ␤ 3 can be activated (36 -38) in a cell-type specific (39) manner. Its activation appears to be a sophisticated mechanism to induce adhesion to ␣ v ␤ 3 ligands, specifically to prothrombin via protein kinase C activation or ADP stimulation, to VN via either hepatocyte growth factor or AP5, an antibody to ␣ v ␤ 3 , and to OPN via either AP5 or agonists, including ADP (38, 40 -42). It has been recently shown that activated ␣ v ␤ 3 mediates cell adhesion and migration to bone sialoprotein (43). In one instance, upon activation by AP5, ␣ v ␤ 3 was shown to increase adhesion and migration of ␣ v ␤ 3 -expressing melanoma cells on OPN and VN in a comparable manner (41). However, the role of activation-dependent and activation-independent ligands of ␣ v ␤ 3 in modulating cell functions and downstream signaling events has not been described.
Several signaling molecules, specifically FAK, PI 3-kinase, and members of the MAP kinase family, play a role in modulating integrin-mediated cell migration (44). FAK is a nonreceptor tyrosine kinase localized in focal contacts that becomes tyrosine-phosphorylated and subsequently activated upon integrin-mediated cell adhesion to several matrix proteins, including VN (5,45,46). FAK phosphorylation of tyrosine 397 (Tyr 397 ) is crucial for cell migration (47). PI 3-kinase is a lipid kinase involved in proliferation, survival, and migration in response to growth factors including EGF and integrin signaling (48 -50). PI 3-kinase forms a complex with FAK via FAK-Tyr 397 in response to cell adhesion or platelet-derived growth factor stimulation (51,52), and it is known to act as a downstream effector of FAK and to control FAK-induced cell migration activated by cell adhesion to extracellular matrix proteins (53,54). AKT plays an important role in transducing survival signals in response to several growth factors and ␤ 1 integrin engagement (55,56) and very recently in supporting vascular endothelial growth factor-induced chemotaxis (57). In response to integrin engagement, AKT activation is PI 3-kinasedependent because wortmannin completely prevents AKT serine phosphorylation (50) and is also controlled by Cdc42, a member of the GTPase family (58) or by ILK (59). Integrin engagement has also been shown to stimulate activation of two members of the MAP kinase family, extracellular signal-regulated kinase-1 and -2 (ERK1/2) (2), which contribute to integrin-mediated cell migration (60,61).
In this study, we show that adhesion of invasive prostate cancer PC3 cells to OPN and VN activates the PI 3-kinase/AKT signaling pathway; however, exogenous expression of ␣ v ␤ 3 in noninvasive LNCaP cells mediates VN binding but requires EGF stimulation to mediate binding to OPN. In these cells, adhesion to OPN does not support cell migration and PI 3-kinase/AKT pathway activation, whereas ␣ v ␤ 3 mediates cell migration and PI 3-kinase/AKT pathway activation on VN. These results show that ␣ v ␤ 3 is expressed in multiple functionally different states and is able to mediate cell migration in a substrate-specific and functional state-dependent manner; finally, they show that ␣ v ␤ 3 activates intracellular signaling pathways in a selective manner in response to individual ligands.
Cell Adhesion and Migration Assays-Cell adhesion and migration assays have been described previously (5). Cell adhesion assays were performed by incubating cells with the coated substrates at 37°C for 1 or 3 h, and quantitated by measuring absorbance at 630 nm using a Titertek Multiskan Bichromatic (ICN Pharmaceuticals, Inc., Costa Mesa, CA) for crystal violet staining. In some experiments, adhesion was quantitated using cells labeled with [ 51 Cr]sodium chromate (Amersham Pharmacia Biotech). The quantitation of [ 51 Cr]sodium chromate-labeled cells was carried out by ␤-counting, and the counts/min were converted to cell numbers based on cell labeling efficiency. Cell migration assays was performed by incubating cells with the coated Transwell chamber (12-m pore size, Costar, Cambridge, MA) at 37°C for 4 h. Cell adhesion and migration assays with EGF were performed using cells that had been preincubated with 200 ng/ml EGF at 4°C for 60 min; EGF was present at the same concentration during the experiments. In the wortmannin inhibition assays, cells were harvested using trypsin (0.05%)/EDTA (0.53 mM) following neutralization in an equal volume of 0.5 mg/ml soybean trypsin inhibitor and washed twice in PBS. Wortmannin dissolved in Me 2 SO at a stock concentration of 10 mM and further diluted to the indicated concentrations in PBS was added to the cell suspension at the time of cell seeding onto the coated plates. Wortmannin was not preincubated with the cells before addition to the coated wells. Cell adhesion and migration were carried out in the presence of the indicated amounts of wortmannin. After incubation at the indicated times, unbound cells were washed away using PBS. The adherent or migrated cells were fixed using 3% paraformaldehyde at 4°C for 30 min followed by crystal violet staining at 25°C for 3 h. In the case of adhesion experiments using 51 Cr-labeled cells, bound cells were lysed in the plate without fixation or staining. Triplicate observations were performed.
AKT, Phospho-AKT, FAK, and Phospho-FAK Immunoblotting-Cells were starved in serum-free RPMI 1640 medium for 24 h, detached using 0.05% trypsin, 0.53 mM EDTA, and neutralized with 0.5 mg/ml soybean trypsin inhibitor. The cells were washed twice with serum-free RPMI 1640, resuspended in the same medium, and incubated with either 200 ng/ml EGF or serum-free medium at 4°C for 60 min. Cells were either held in suspension using 10 mg/ml BSA-coated plate or seeded on 60-mm dishes coated with FN, VN, or OPN at indicated concentrations and allowed to attach at 37°C for the indicated times. Cells were lysed in 1% Nonidet P-40 lysis buffer: 50 mM Tris, pH 7.5 (American Bioanalytical, Natick, MA), 1% Nonidet P-40 (Calbiochem), 50 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM sodium orthovanadate, 50 mM NaF, 0.2 mM EGTA, 1 mM EDTA, pH 8.0 (all from Sigma). The protein concentration of each lysate was determined using BCA protein assay reagent (Pierce). For AKT and phospho-AKT immunoblotting, 30 g of cell lysates were loaded in each lane on a 7.5% or 10% SDS-PAGE under reducing conditions. The proteins were transferred to polyvinylidene difluoride membranes (Millipore). The membrane was blocked with 5% milk in Tris-buffered saline with 0.1% Tween 20 at 25°C for 3 h. Polyclonal antibodies to AKT (0.1 g/ml) or to phospho-AKT Ser 473 (0.05 g/ml) were incubated with the membrane at 4°C for 16 h. Then, the membrane was incubated with peroxidase-coupled anti-rabbit IgG at 25°C for 1 h. The specific proteins were detected with enhanced chemiluminescence (ECL, Amersham Pharmacia Biotech). Quantitative analysis was performed using a computing densitometer (Molecular Dynamics, Sunnyvale, CA).
Analysis of FAK phosphotyrosine content was performed as described previously (5). Briefly, precleared lysates were prepared as above and then immunoprecipitated using 0.5 g of polyclonal antibody to FAK, C-20 (Santa Cruz Biotechnology, Inc). Immunoblotting analysis was performed using 1 g/ml anti-phosphotyrosine monoclonal antibody, PY20 (Transduction Laboratories, San Diego, CA). To detect immunoprecipitated proteins, membranes were stripped and reblotted using C-20 (0.1 g/ml). Experiments were repeated three times. Quantitative analysis was performed using a computing densitometer.
ERK1/2 and Phospho-ERK1/2 Immunoblotting-Immunoblotting was performed as described previously (66). Cells were harvested as described before. Cells were washed once with 0.5 mg/ml soybean trypsin inhibitor and twice with serum-free RPMI 1640 medium. Cells were resuspended in serum-free RPMI 1640 medium, incubated at 37°C without CO 2 with agitation for 15 to 30 min, and plated on 60-mm Petri dishes that had been coated with 3 g/ml either VN or FN. The concentration of human VN and FN used for coating had been previously determined to generate comparable cell attachment (data not shown). Cells were incubated at 37°C with 5% CO 2 for the indicated intervals. The attached cells were washed twice with PBS and lysed in 25 mM HEPES, pH 7.6, 0.1% Triton X-100, 300 mM NaCl, 1.5 mM MgCl 2 , 0.5 mM dithiothreitol, 1 mM sodium orthovanadate, 0.2 mM EDTA, 2 g/ml leupeptin, 2 g/ml aprotinin, 0.1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine. Thirty g of each lysate were loaded on 10% SDS-PAGE (9.93% acrylamide, 0.07% bisacrylamide) under reducing conditions, and proteins were transferred to polyvinylidene difluoride membranes. After blocking, the membrane was incubated with 0.1 g/ml polyclonal antibody to ERK1/2 (Santa Cruz Biotechnology, Inc.) or monoclonal antibody to phospho-ERK1/2 Thr 202 /Tyr 204 (1:2000) at 4°C for 16 h. The membranes were washed using 0.1% Tween 20 in Tris-buffered saline and incubated with a peroxidase-conjugated goat affinity-purified antibody to rabbit IgG at room temperature for 1 h. ERK1/2 or phosphorylated ERK1/2 was detected using enhanced chemiluminescence.
Statistical Analysis-Statistical analysis was performed using the Student's t test or one way analysis of variance, Sigma Stat (Jandel Scientific, San Rafael, CA).

EGF Mediates Cell
Adhesion to OPN via ␣ v ␤ 3 -Highly metastatic PC3 and nonmetastatic LNCaP prostate cancer cells differentially express the ␣ v ␤ 3 integrin but express comparable levels of ␤ 5 and ␤ 1 ( Fig. 1 and Ref. 5). To investigate whether ␣ v ␤ 3 mediated prostate cancer cell adhesion to OPN, we analyzed the ability of PC3 cells and LNCaP cells stably transfected with ␤ 3 cDNA (␤ 3 -LNCaP) to bind OPN. PC3 cells adhered to OPN (Fig. 2, A and B) in an ␣ v ␤ 3 -dependent manner, because LM609, a monoclonal antibody to ␣ v ␤ 3 , inhibited their adhesion to OPN ( Fig. 2A). PC3 were shown to attach equally FIG. 1. Expression of integrins in PC3, parental LNCaP, ␤ 3 -LNCaP and mock-LNCaP cells. FACS analysis was performed as described under "Experimental Procedures" using LM609 (1:500), a monoclonal antibody to ␣ v ␤ 3 (top row); P1F6 (1:500), a monoclonal antibody to ␣ v ␤ 5 (middle row); or TS2/16 (1:10), a monoclonal antibody to ␤ 1 (bottom row). Secondary anti-mouse IgG is conjugated to fluorescein isothiocyanate. 1C10 (1:500) or nonspecific hybridoma supernatant X653 (1:10) were used as negative control antibodies (filled profiles). FACS analysis FIG. 2. PC3 cells adhere to OPN in an ␣ v ␤ 3 -dependent manner. A, adhesion of 51 Cr-labeled PC3 cells to OPN (10 g/ml) was performed in the presence of LM609 (1:100), a monoclonal antibody to ␣ v ␤ 3 , or 1C10 (1:100) as negative control. No Ab represents cell adhesion to OPN in the absence of an antibody. Cell adhesion to BSA (10 mg/ml) was used as negative control. The difference between adhesion to OPN in the presence of LM609 or of 1C10 is statistically significant (p Ͻ 0.01). The experiment was repeated three times with consistent results. B, 51 Cr-labeled PC3 cell adhesion to 10 g/ml OPN in the presence (solid bars) or absence (hatched bars) of 200 ng/ml EGF is shown. Cell attachment to BSA (10 mg/ml) in the presence or absence of 200 ng/ml EGF was used as negative control. C, the number of 51  well to OPN and VN, although at the lowest concentration tested (0.1 g/ml), OPN was more active in supporting cell adhesion than VN (Fig. 2C). Surprisingly, ␤ 3 -LNCaP cells did not bind OPN (Fig. 3, A and B), although a significant amount of ␣ v ␤ 3 was expressed on the cell surface (Fig. 1), and the cells did adhere to VN in an ␣ v ␤ 3 -dependent manner (Fig. 3A) (5). We hypothesized that exogenously expressed ␣ v ␤ 3 was in a conformation that did not bind OPN and that external stimuli would be required for its activation. It has been shown that ␣ v ␤ 3 activation requires protein kinase C (38) and that EGF and its receptor activate protein kinase C (67,68). Thus, we tested the ability of EGF to activate ␣ v ␤ 3 . EGF increased ␤ 3 -LNCaP cell adhesion to OPN but had no effect on BSA (Fig.  3A). EGF stimulation did not increase ␣ v ␤ 3 expression levels in ␤ 3 -LNCaP cells (Fig. 1). In contrast, in the presence of EGF, PC3 cell adhesion to OPN was not increased (Fig. 2B). EGF- stimulated ␤ 3 -LNCaP cell adhesion to OPN was blocked by LM609 but not by P3G2, an antibody to ␣ v ␤ 5 (Fig. 3C). P3G2 previously shown to block ␣ v ␤ 5 adhesion to its ligand was active in inhibiting HeLa cell adhesion to VN (data not shown) at the concentrations used in Fig. 3C. ␤ 3 -LNCaP and mock-LNCaP transfectants express HER at comparable levels (Fig. 4); however, the effect of EGF was specific for ␤ 3 because OPN adhesion was not up-regulated either in parental LNCaP cells (Fig. 5A) or in mock-transfected LNCaP cells (Fig. 5B) that do not express ␣ v ␤ 3 . In conclusion, EGF is required to activate ␣ v ␤ 3 adhesion of noninvasive prostate cancer LNCaP cells to OPN, indicating a new level of complexity in the regulation of cell adhesion by ␣ v ␤ 3 .
Role of PI 3-Kinase in ␣ v ␤ 3 -mediated Cell Migration-To investigate whether adhesion to OPN would result in increased cell migration, PC3 and ␤ 3 -LNCaP cells were analyzed in Transwell migration assays using an EGF concentration gradient or a constant concentration of EGF in both the upper and bottom chambers. PC3 cells migrated on both OPN and VN substrates; migration on OPN occurred at a higher extent than on VN (Fig. 6, A and C). Wortmannin, a PI 3-kinase inhibitor, inhibited PC3 cell migration on OPN and VN (Fig. 6, A and C) but not adhesion on these substrates (Fig. 6, B and D). Instead, ␤ 3 -LNCaP cells did not migrate on OPN in the presence or absence of EGF ( Fig. 7A and data not shown), although these cells migrated on FN (Fig. 7A) and VN (Fig. 7B and Ref. 5). Wortmannin inhibited ␤ 3 -LNCaP cell migration on VN (Fig.  7B) but not adhesion to VN (Fig. 7C). In all experiments, concentrations of VN and OPN that gave comparable levels of adhesion were selected; OPN coating of either the bottom part or of both sides of the Transwell chamber gave comparable results (data not show). EGF was active in mediating chemotaxis of A431 cells on laminin-1-coated substrates (data not shown). These results show that a signaling step, crucial for cell migration, failed to be activated in ␤ 3 -LNCaP cell transfectants adherent to OPN but was active in PC3 cells.
Substrate Specificity of PI 3-Kinase Pathway Activation-To analyze whether a differential activation by ␣ v ␤ 3 of downstream integrin-mediated signaling events occurs in response to adhesion to a specific substrate, PC3 and ␤ 3 -LNCaP cells were allowed to attach to either OPN or VN. To examine PI 3-kinase activation, we analyzed the levels of Ser 473 phosphorylation of AKT, a downstream effector of PI 3-kinase, as a sensitive readout of PI 3-kinase activity (56). PC3 cells stimu- lated the PI 3-kinase/AKT signaling pathway on OPN more significantly than on VN through ␣ v ␤ 3 or than on FN through ␤ 1 integrins (Fig. 8A), although they attached to OPN, VN, and FN equally well (Fig. 8B). The maximum levels of AKT phosphorylation on OPN were observed between 30 and 45 min. As shown in Fig. 9A, ␤ 3 -LNCaP cell adhesion to OPN did not induce AKT Ser 473 phosphorylation, whereas adhesion to VN induced a significant increase in AKT Ser 473 phosphorylation. Densitometric analysis performed using three separate exposures in a linear range showed a 13-to 18-fold increase in AKT Ser 473 phosphorylation on VN and a 1-to 2-fold increase on OPN. AKT phosphorylation induced by EGF was detected at 20, 30, 60, and 120 min on cells attached to their extracellular matrix (not shown); however EGF did not induce AKT Ser 473 phosphorylation in ␤ 3 -LNCaP cells attached to OPN (Fig. 9A) nor did it increase AKT phosphorylation when these cells attached to VN (Fig. 9B).
A pharmacologic approach was also used to investigate the role of other downstream effectors of FAK that have been previously shown to mediate cell migration, ERK1 and -2 (60,61). An inhibitor of MEK-1 in the MAP kinase pathway, PD98059 (69), was tested to analyze its effect on ␤ 3 -LNCaP and PC3 cell migration. PD98059 had no effect on migration of either cell type on VN, although it did inhibit endothelial cell migration (data not shown and Ref. 61). Activation of ERK1/2 occurred in PC3 cells attached to both VN and OPN substrates (Fig. 10A); in contrast, ␤ 3 -LNCaP cells attached to VN and OPN did not activate ERK1/2 phosphorylation at any of the time points analyzed (Fig. 10B and not shown); ERK1/2 activation was detected in response to EGF treatment in cells attached to their matrix in tissue culture dishes (Fig. 10B). In Fig. 10A, ERK1/2 activation is longer than other cells tested in our laboratory in similar conditions; these results do not have a mechanistic explanation at this time. However, the observed sustained activation seems to be non-integrin-dependent because it is seen in cells in suspension as well after 30 min. Thus, ERK1/2 activation did not play a role or correlate with either PC3 or ␤ 3 -LNCaP cell migration.
Our previous report showed that FAK played a predominant role in mediating cell migration on VN, because migration was inhibited by expression of FAK-related non-kinase (5). It has been shown that PI 3-kinase forms a complex with FAK, acts as a downstream effector of FAK, and controls cell migration (53).
In ␤ 3 -LNCaP cells, FAK is phosphorylated in response to ␣ v ␤ 3 engagement by OPN and by VN (Fig. 11). Densitometric analysis performed using three separate exposures in a linear range showed the following increase in FAK phosphorylation: 8.5-fold increase on OPN and 9.4-fold on VN in presence of EGF and 8.1-fold on VN in the absence of EGF. EGF stimulation did induce comparable levels of HER tyrosine phosphorylation in cells in suspension or attached to VN or OPN (data not shown), thus indicating that HER is functional in all tested conditions; however, EGF did not increase FAK-tyrosine phosphorylation either on VN or in suspended cells (Fig. 11). In conclusion, on OPN substrates, ␣ v ␤ 3 -mediated signaling events fail to be activated downstream of FAK at the level of PI 3-kinase/AKT activation. DISCUSSION This study shows that ␣ v ␤ 3 is expressed in multiple functional states and that its ability to mediate cell migration and intracellular signaling pathways is substrate-specific and functional state-dependent. In PC3 cells, ␣ v ␤ 3 mediates cell adhesion, migration, and PI 3-kinase/AKT pathway activation on VN and OPN. In contrast, adhesion to OPN of noninvasive LNCaP cells upon exogenous expression of ␣ v ␤ 3 requires its activation by EGF although ␣ v ␤ 3 is in a functional state that allows adhesion to a different ligand, VN, in the absence of EGF. Furthermore, in LNCaP cells, while ␣ v ␤ 3 mediates cell migration and PI 3-kinase/AKT pathway activation on VN, adhesion to OPN fails to support cell migration and PI 3-kinase/AKT pathway activation mediated by ␣ v ␤ 3 . This is the first report that shows an integrin ligand-mediated phenotypical alteration that reverts a migratory cell into a nonmigratory cell via engagement of the same integrin. We conclude that ␣ v ␤ 3 exists in multiple functional states that can bind either VN selectively or both VN and OPN and that can differentially activate cell migration and the PI 3-kinase/AKT signaling pathway in a ligand-specific manner. The results highlight a versatile role for ␣ v ␤ 3 in the regulation of the PI 3-kinase/AKT pathway and in a substrate-dependent control of cell invasion.
We show for the first time that EGF reverts a form of ␣ v ␤ 3 that does not recognize OPN in an OPN-binding form. Although the mechanism of activation remains to be identified, EGF effect is not due to a change in integrin expression on the cell surface because EGF regulates cell adhesion to OPN with- out a significant change in integrin expression (Fig. 1). EGFdownstream players that might potentially activate ␣ v ␤ 3 are protein kinase C, known to be involved in mediating ␣ v ␤ 3 activation (38), and HER through its direct association with integrins (70); however, additional modulators, such as integrin-associated proteins (71,72), might be responsible for changes in ligand binding or post-ligand binding activities. Similar to our findings, activation-independent (fibrinogen) and -dependent (prothrombin) ligands for ␣ v ␤ 3 have been shown by Byzova and Plow (38) suggesting that a sophisticated mechanism of tight regulation and ligand selection involves The EGF receptor has been shown to synergize with ␣ v ␤ 5 to increase cell migration (73); LNCaP cells express low levels of ␤ 5 and large amounts of ␤ 1 (Fig. 1 and Ref. 5). However, these previously described OPN receptors did not play a role in the adhesion of these cells to OPN in our experimental system, since first, parental or mock-LNCaP cells that did not express ␣ v ␤ 3 did not adhere to OPN in response to EGF stimulation and second, an antibody to ␣ v ␤ 5 did not inhibit OPN adhesion of ␤ 3 -LNCaP cells. We conclude that EGF and its receptor HER synergize with ␣ v ␤ 3 in a substrate-specific manner on OPN but not on VN. This change required for ␤ 3 -LNCaP cell adhesion to OPN did not support cell migration on OPN although these cells migrated on VN. The ability of ␣ v ␤ 3 to mediate cell migration is therefore substrate-specific. Invasive PC3 cells have the ability to up-regulate cell migration through ␣ v ␤ 3 on OPN; therefore, it is conceivable that when cancer cells lose the ability to select their cell binding partners by uncoupling/deregulating the synergistic activity of ␣ v ␤ 3 integrin and HER, such as in PC3 cells, they migrate in response to engagement by multiple ␣ v ␤ 3 ligands.
Among the three known pathways that mediate cell migration and are activated by integrins: FAK, PI 3-kinase/AKT, and MAP kinase pathways, we have shown that the FAK (5) and the PI 3-kinase/AKT pathways support migration on VN in ␤ 3 -LNCaP cells and on VN and OPN in PC3 cells. The MAP kinase pathway did not play a role in either ␤ 3 -LNCaP or PC3 cell migration because PD98059 did not block cell migration (not shown). It should be pointed out that AKT has the ability to support cell migration mediated by vascular endothelial growth factor in endothelial cells (57); it is not known, however, whether this mechanism is active in other cells. We show that ␣ v ␤ 3 -OPN interaction mediates FAK tyrosine phosphorylation but this signal, although necessary, is not sufficient to mediate cell migration in noninvasive cells. PI 3-kinase, a mediator of integrin and growth factor activities including EGF (74), is known to act as a downstream effector of FAK and to control cell migration activated by cell adhe-sion to extracellular matrix proteins (53,54). Integrin-mediated adhesion to the extracellular matrix proteins stimulates the association of the p85 regulatory PI 3-kinase subunit with FAK through FAK Tyr 397 (51,52); FAK binding to PI 3-kinase has been demonstrated to activate the latter one (53). Because FAK is tyrosine-phosphorylated in response to OPN adhesion mediated by EGF, we conclude that a block at FIG. 9. Adhesion of ␤ 3 -LNCaP transfectants to OPN does not stimulate AKT Ser 473 phosphorylation. A, top panel, AKT phosphorylation of ␤ 3 -LNCaP transfectants attached to OPN (10 g/ml) or VN (3 g/ml) or held in suspension (10 mg/ml BSA) at 37°C for 3 h was measured by immunoblotting using a polyclonal antibody to phospho-AKT Ser 473 (0.05 g/ml). Protein loading control is shown on the bottom panel by AKT immunoblotting (WB) with polyclonal antibody to AKT (0.1 g/ml). B shows phospho-AKT Ser 473 (top panel) and AKT protein loading (bottom panel) of ␤ 3 -LNCaP transfectants attached to VN (3 g/ml) at 37°C for 3 h with and without EGF. The experiments were repeated at least three times with consistent results.

FIG. 10. Activation of ERK1/2 and FAK signaling pathways on OPN and VN.
A, ERK1/2 phosphorylation was tested using a monoclonal antibody to phospho-ERK1/2 (top panel), and protein loading was confirmed (bottom panel) using a polyclonal antibody to ERK1/2. The experiments were repeated at least twice with consistent results. B, cell lysates from cells attached to OPN (10 g/ml) or VN (3 g/ml) or to their matrix in tissue culture dishes (on dish) in the absence of EGF or in the presence of EGF (200 ng/ml) were analyzed using 0.1 g/ml rabbit affinity-purified antibody to ERK1/2 (bottom panel) and a monoclonal antibody to phospho-ERK1/2 (top panel). The experiments were repeated at least two times with consistent results.
FIG. 11. Tyrosine phosphorylation of FAK in ␤ 3 -LNCaP cells attached to OPN or VN. FAK tyrosine phosphorylation of ␤ 3 -LNCaP cell transfectants was measured using PY20 (1 g/ml) immunoblotting (top panels) of immunocomplexes precipitated by C-20 polyclonal antibody to FAK (0.5 g) from ␤ 3 -LNCaP transfectant detergent lysates. Lysates were prepared from cells attached to OPN (10 g/ml)or VN (3 g/ml)-coated dishes in the presence or absence of EGF (200 ng/ml). The same membrane was stripped, and FAK protein levels were analyzed using 0.1 g/ml affinity-purified antibody to FAK (bottom panels). FAK tyrosine phosphorylation is expressed as -fold increase over the levels detected in cells held in suspension. The experiments were repeated at least three times with consistent results. the level of PI 3-kinase/AKT activation downstream of FAK explains the failure of ␤ 3 -LNCaP cells to migrate, although ␣ v ␤ 3 and the PI 3-kinase/AKT pathway are fully functional in these cells upon ␣ v ␤ 3 engagement by VN. The data suggest that the generated ␤ 3 -LNCaP cells are a model system that allows the study of the ␣ v ␤ 3 effectors that mediate cell migration downstream of FAK. It remains to be analyzed whether FRNK, a negative regulator of FAK that we have shown inhibits VN-mediated migration in ␤ 3 -LNCaP cells (5), specifically inhibits FAK/PI 3-kinase interaction. It should be stressed that the PI 3-kinase/AKT pathway might also control cell adhesion, as shown by Byzova and Plow since in this study (43) wortmannin did inhibit both cell adhesion and migration after a 30-min preincubation; however, we did not observe wortmannin inhibition of cell adhesion to VN and OPN in our system due to either a cell type-specific effect or to the lack of preincubation with wortmannin in our experimental system. PTEN, a lipid phosphatase that prevents FAK and PI 3-kinase/AKT pathway activation (75,76) and down-regulates cell motility and directionality (77) is not expected to contribute to the migration of these cells, because LNCaP and PC3 cells have been shown to express a mutated and a deleted PTEN, respectively (78). In both cell types, PI 3-kinase/AKT pathway activation is controlled by integrins in the absence of an active PTEN, confirming that other molecules such as either Cdc42 (58) or ILK (59) could control integrin-mediated cell migration.
We show here that in ␤ 3 -LNCaP cells a differential activation of the PI 3-kinase/AKT pathway by ␣ v ␤ 3 occurs: OPN interaction with ␣ v ␤ 3 does not activate the PI 3-kinase/AKT pathway, whereas VN does. It should be noted that PI 3-kinase/AKT pathway is stimulated via OPN engagement of ␣ v ␤ 3 in PC3 cells (Fig. 8A) and in osteoclasts (79), thus indicating that the specific failure to activate the PI 3-kinase pathway is cell type-dependent. Because OPN-null mice generate significantly smaller metastases than wild type mice (21), it is thus conceivable that in the event LNCaP or another noninvasive cell will migrate to a metastatic site where OPN is predominantly expressed, the interaction of ␣ v ␤ 3 with OPN will not provide a migratory or, alternatively, survival signal for these cells. Recently, the role of AKT in promoting cell survival of androgen-sensitive LNCaP cells but not of androgen-insensitive PC3 cells has been shown (80). It is noted that because AKT activation promotes cell survival (48) and LNCaP cells undergo apoptosis in the presence of the PI 3-kinase inhibitor, wortmannin (80), the failure of OPN to stimulate AKT activation might result in apoptosis of these poorly tumorigenic cells. It remains to be determined whether direct ␣ v ␤ 3 integrin engagement in prostate cells prevents apoptosis by activation of the PI 3-kinase/AKT pathway. The synergistic activity of ␣ v ␤ 3 , OPN, and the downstream PI 3-kinase/AKT pathway might balance proliferative, apoptotic, and migratory stimuli, thus playing a crucial role in tumor growth and metastatic events in vivo.