Differential regulation of phosphoinositide metabolism by alphaVbeta3 and alphaVbeta5 integrins upon smooth muscle cell migration.

Smooth muscle cell migration is a key step of atherosclerosis and angiogenesis. We demonstrate that alpha(V)beta(3) and alpha(V)beta(5) integrins synergistically regulate smooth muscle cell migration onto vitronectin. Using an original haptotactic cell migration assay, we measured a strong stimulation of phosphoinositide metabolism in migrating vascular smooth muscle cells. Phosphatidic acid production and phosphoinositide 3-kinase IA activation were triggered only upon alpha(V)beta(3) engagement. Blockade of alpha(V)beta(3) engagement or phospholipase C activity resulted in a strong inhibition of smooth muscle cell spreading on vitronectin. By contrast, blockade of alpha(V)beta(5) reinforced elongation and polarization of cell shape. Moreover, Pyk2-associated tyrosine kinase and phosphoinositide 4-kinase activities measured in Pyk2 immunoprecipitates were stimulated upon cell migration. Blockade of either alpha(V)beta(3) or alpha(V)beta(5) function, as well as inhibition of phospholipase C activity, decreased both Pyk2-associated activities. We demonstrated that the Pyk2-associated phosphoinositide 4-kinase corresponded to the beta isoform. Our data point to the metabolism of phosphoinositides as a regulatory pathway for the differential roles played by alpha(V)beta(3) and alpha(V)beta(5) upon cell migration and identify the Pyk2-associated phosphoinositide 4-kinase beta as a common target for both integrins.

Integrins ␣ V ␤ 3 and ␣ V ␤ 5 have been involved in cell adhesion and migration of different cell types. Despite their ability to bind the same ligand, vitronectin, ␣ V ␤ 3 and ␣ V ␤ 5 integrins represent a good example of functional differences generated by different combinations of the integrin ␣ and ␤ subunits. Indeed, previous data showing that blocking antibodies against ␣ V ␤ 3 affect both cell spreading and migration but that those directed to ␣ V ␤ 5 inhibit only cell migration argue in favor of specific events regulated by ␣ V ␤ 3 or ␣ V ␤ 5 upon cell motility (1). Little is known on the specific signal transduction pathways triggered by ␣ V ␤ 3 or ␣ V ␤ 5 engagement. However, some differences have been shown, such as a more diffuse distribution of ␣ V ␤ 5 compared with ␣ V ␤ 3 , which is most frequently found in focal adhesion contacts (2,3). Clustering of ␣ V ␤ 5 induces a weaker colocalization of actin, ␣-actinin, and tensin than ␣ V ␤ 3 (4). Engagement of ␣ V ␤ 3 is able by itself to trigger protein kinase C or FAK 1 activation and cell migration, whereas ␣ V ␤ 5 needs the concomitant activation of growth factors or cytokines receptors for this signaling pathway (4,5).
The importance of phosphoinositides in chemotactic signaling was pointed out by the ability of receptors to chemotactic factors to directly bind and regulate the key enzymes phospholipase C (PLC) and phosphoinositide 3-kinase (PI3K) (6). Studies with mutant receptors have demonstrated that one of these two pathways might be predominant, depending on the cell status (7). The PLC activity and the PI3K activity products have common targets involved in cell migration, such as specific isoforms of protein kinase C, calcium mobilization through either internal stocks or calcium channels, and PIP 5-kinase activity allowing PIP2 formation and interaction with actinassociated proteins. Integrins have also been shown to be able to activate PLC-␥1 and PI3K IA (p85␣/p110␣) through FAK regulation (8 -10). Moreover, integrins may regulate PIP2 production either through activation of the rac-dependent PIP 5-kinase (11) or, for some integrins, through activation of the tetraspan family-associated PI4K (12). Upon cell migration, these signaling pathways contribute to an adapted response of the cell to its environment through cytoskeleton reorganization and mobilization of membrane receptors. Depending on the nature of the matrix protein and its state (intact or hydrolyzed), integrins may regulate either formation or disorganization of focal adhesions (13,14). PI3K was shown to regulate focal adhesion formation through modulation of integrin affinity (15) but also through disorganization of focal complexes (16). Because cell migration requires both focal adhesion formation at the front and focal adhesion disorganization at the rear, an important question to resolve is the nature of integrins involved at each step, and the specific enzymes activated.
Vascular smooth muscle cell (VSMC) migration is a key process in atherosclerosis, restenosis, and angiogenesis. Because VSMCs express both vitronectin receptors, ␣ V ␤ 3 and ␣ V ␤ 5 (17), we focused our studies on phosphoinositide-dependent signaling pathways upon VSMC migration, downstream of * 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.
§ Supported by fellowships from the Groupement de Réflexion et de Recherche Cardiovasculaire and the Fondation pour la Recherche Médicale.

EXPERIMENTAL PROCEDURES
Cell Culture-Pig VSMCs were prepared from thoracic aorta of 6-week-old pigs using the explant technique (18). Cells were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum. For all experiments, VSMCs were used at passages 2-5.
Cells were fed three times weekly and passaged by treatment with 0.25% trypsin and 0.02% EDTA before confluence. Cultures were maintained in a humidified incubator with 5% CO 2 /95% air at 37°C.
Solid phase assays were based on inhibition of fibrinogen or vitronectin binding to isolated human ␣ V ␤ 3 or ␣ V ␤ 5 integrins, respectively, in a solid phase enzyme-linked immunosorbent assay format. ␣ V ␤ 3 and ␣ V ␤ 5 were purified by a three-step affinity chromatography from human placenta. Microtiter plates (96 wells; MaxiSorp Immuno Plate; Nunc) were coated with 25 ng of ␣ V ␤ 3 or ␣ V ␤ 5 /well overnight at 4°C and then blocked with 3.5% BSA. In each well, 100 l of a solution of 1.5 g/ml human fibrinogen (IMCO, Stockholm, Sweden) or human vitronectin (Hoffmann-La Roche) and inhibitors (Ro 64 and Ro 66) at various concentrations in Tris-HCl buffer, pH 7.4, supplemented with 1 mM CaCl 2 , 1 mM MgCl 2 , 1 mM MnCl 2 , and 1% BSA was incubated for at least 4 h at room temperature. After three washes, anti-human fibrinogen (Dakopatts; A0080; 1:2400) or anti-human vitronectin (BIO-SOURCE; clone M2; AHE101) antibodies were added for 1 h followed by incubation with peroxidase-conjugated secondary IgG antibodies for 30 min. Absorption was measured in an enzyme-linked immunosorbent assay reader. Controls with RGDS peptides or kistrin have been performed in parallel experiments.
Adhesion assays were performed in 96-well plates coated with vitronectin (1 g/well) as follows. Stable transfectants of HEK 293 cells expressing ␣ v ␤ 3 or ␣ v ␤ 5 and parental HEK 293 cells (lacking endogenous ␣ v ␤ 3 and ␣ v ␤ 5 expression but expressing ␣ v and ␣ 5 ␤ 1 integrins) were used in this study. Cells (30,000 cells/well) were allowed to adhere for 30 min at 37°C in DMEM-0.5% BSA with 0.1 mM MnSO 4 . Ro 64 or Ro 66 was added in a concentration range between 0.001 and 100 M with a log increment of 1. After two gentle washes with PBS, HEK 293 cells were fixed with a Karnovsky solution (2% paraformaldehyde-0.1% glutaraldehyde in PBS, pH 7.4), stained with a 0.1% crystal violet solution, and rinsed four times under running water. The stain retained by cells was solubilized in 0.1 ml of 2% sodium deoxycholate, and A 595 was measured to determine cell adhesion and determine the IC 50 . Under these conditions, specific blocking antibodies directed toward ␣ v ␤ 3 (LM609) or ␣ v ␤ 5 (P1F6) completely blocked adhesion onto vitronectin of ␤ 3 -or ␤ 5 -transfected HEK 293 cells, respectively (data not shown).
Migration Assays-The haptotactic motility experiments were performed in 12-well plates using a polyester stacking system (see Fig. 1). The assay was performed in two steps: (i) seeding of VSMCs on polyester filters coated with poly-L-lysine, and (ii) migration through the polyester stacking.
(i) Seeding of Cells on Polyester-Polyester (PE) filters (21-mm diameter; 11-m pores; Lanz-Anliker, Rohrbach, Switzerland) placed into 12-well Evergreen plates (untreated; Polylabo, Strasbourg, France) were first coated for 1 min with 17 g/ml poly-L-lysine (P-Lys) (Sigma) in PBS. Under these conditions, coating efficiency for P-Lys was determined previously to be 0.2 g/PE. After PE drying at room temperature, nonspecific binding sites were blocked with PBS containing 0.5% fatty acid-free BSA (w/v) for 2 h at room temperature. Before seeding, PE filters were washed twice with PBS and once with DMEM-0.5% BSA and incubated for an additional 5 min at 37°C in the last wash.
VSMCs grown near confluence were detached with trypsin-EDTA, washed, resuspended in DMEM containing 0.5% BSA (Albumax; Life Technologies, Inc.), and finally plated into wells containing the P-Lyscoated PE, at 300,000 cells/ml in a final volume of 1 ml. Cell adhesion was allowed for 24 h at 37°C. Efficiency of seeding was checked by counting the remaining nonadherent VSMCs with a cell counter (Coulter) and was determined to be 80%.
(ii) Cell Migration-Bottom PE filters with adherent cells were washed twice in DMEM-0.5% BSA and transferred into new 12-well plates containing DMEM-0.5% BSA. Migration was initiated with an overlay of the middle and top PE filters coated with human vitronectin from Life Technologies, Inc. (1 h with 3 g/ml vitronectin in PBS followed by BSA blocking). The three PE filters were maintained in tight contact with each other by applying a monolayer of glass beads (3-mm diameter) (Fig. 1). For inhibition studies, antibodies, synthetic peptides, and small molecule antagonists were preincubated at 37°C for 1 h before overlay at appropriate concentrations established for porcine VSMCs. In some assays, 20 M LY-294002 (2-(4-morpholinyl)-8-phenyl-4H-1-benzopyran-4-one) from Biomol Research Laboratories (Plymouth Meeting, PA) or 50 or 100 nM wortmannin from Sigma, or U-73122 from Novabiochem (France Biochem, Meudon, France) was preincubated for 1 h with VSMCs to block PI3K IA/PI4K ␤ or PLC activities, respectively. When inhibitors were dissolved in Me 2 SO, the final amount of Me 2 SO did not exceed 0.6% (v/v). For in vivo phosphoinositide labeling, cells were incubated for 4 h with 0.4 mCi/ml [ 32 P]phosphate (Amersham Pharmacia Biotech) before overlay. Migration was performed in the 32 P-containing medium for 24 h at 37°C in a humidified cell culture incubator. Under our experimental conditions, VSMCs were not proliferative, as determined by cell cycle distribution analysis performed by flow cytometry after staining with a propidium iodide solution (data not shown).
At the end of the incubation period, all PE filters were rinsed twice in PBS, fixed in a Karnovsky solution for 5 min at 4°C, and dried on tissue paper. Lipids were extracted following the modified procedure of Bligh and Dyer (20) on the same PE filters used for cell migration evaluation. Control experiments were performed and showed that cell fixation before lipid extraction did not modify lipid quantification, nor was the lipid extraction detrimental to the subsequent evaluation of cell migration (data not shown). The PE filters were removed from the lipid extraction tube after the one-phase solvent extraction step and dried on tissue paper. Then, PE filters were stained in crystal violet solution for 5 min. After six washes under running water, the crystal violet staining of PE filters was dissolved by adding 500 l/well 2% deoxycholate, microwaving for 30 s at 150 W, and shaking for 2 min. Absorbance was measured at 595 nm on 200 l in 96-well plates using a HTS 7000 Bio Assay Reader (PerkinElmer Life Sciences). Cell migration was quantified by comparison of the absorbance from the middle ϩ top PE filters (cells that migrated) to the absorbance from total cells that adhered on P-Lys (the sum of absorbances from the bottom, middle, and top PE filters). Cells that migrated on the top PE filter represented ϳ30% of FIG. 1. Haptotactic cell migration system. VSMCs in DMEM-0.5% BSA were seeded on P-Lys-coated polyester filters (bottom PE filters) for 24 h. The bottom PE filters were then transferred into new wells filled with DMEM-0.5% BSA Ϯ inhibitors, and migration was started by an overlay of the middle and top PE filters coated with vitronectin. A monolayer of glass beads was added to maintain tight contact between the three PE filters. Migration was allowed to proceed for 24 h at 37°C in a humidified cell culture incubator. total cells that migrated (middle ϩ top PE filters). Each test group was performed in at least triplicate wells.
Individual lanes containing commercial standards of PtdOH, PtdIns, PtdIns-4-P, or PtdIns-4,5-P 2 were stained with iodine vapors or with zinzade spraying. The radioactive spots were visualized and quantified by a PhosphorImager 445 SI (Molecular Dynamics) after a 2-day exposure. In some experiments, whole lipid extracts were deacetylated by methylamine treatment, separated by high pressure liquid chromatography, and quantified by on line radioactivity detection (25).
In Vitro PI Kinase Assay-p85 and Pyk2 immunoprecipitates were suspended in PI kinase activity buffer (0.5 mM EDTA, 100 mM NaCl, and 50 mM Tris-HCl, pH 7.4, plus 50 M ATP and 10 mM MgCl 2 ) and incubated with PtdIns (30 g)/phosphatidylserine (60 g) vesicles and 15 Ci of [␥-32 P]ATP for 30 min at 37°C with shaking, according to Whitman et al. (26). Protein A beads associated with PI4K ␤ were incubated in a 50-l final volume containing 20 mM HEPES, pH 7.5 and 0.3% Triton X-100 plus 50 M ATP and 10 mM MgCl 2 and in the presence of 0.2 mg/ml sonicated PtdIns and 30 Ci of [␥-32 P]ATP for 20 min at 37°C with shaking (27). Reaction products were separated by TLC, visualized by autoradiography, and quantified on a Phosphor-Imager (Molecular Dynamics).
In Vitro Tyrosine Kinase Assay-Equal amounts of lysates from migrating or nonmigrating VSMCs (1.4 ϫ 10 6 cells) were subjected to immunoprecipitations with antisera against Pyk2. Immunoprecipitates were washed three times with lysis buffer and once with kinase buffer (50 mM Tris-HCl, pH 7.5, 5 mM MnCl 2 , and 5 mM MgCl 2 ). One-half of the immunoprecipitates were analyzed by SDS-polyacrylamide gel electrophoresis and immunoblotting with anti-Pyk2 antibodies, whereas the other half were incubated with 40 l of kinase buffer supplemented with 20 g of poly(Glu-Tyr) (4:1) and 40 M ATP including 15 Ci of [ 32 P]ATP for 15 min at room temperature. The reaction was stopped by the addition of SDS sample buffer, sample was boiled for 2 min, and products were resolved by 7.5% SDS-polyacrylamide gel electrophoresis.
Immune Complex Kinase Assays-24 ϫ 10 6 VSMCs in culture were washed twice with PBS without Ca 2ϩ and Mg 2ϩ and lysed in modified RIPA buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 4 mM EDTA, 1.5% Triton X-100, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 10 g/ml aprotinin). Lysates were clarified by a 10-min centrifugation at 8000 ϫ g at 4°C and then incubated with anti-Pyk2 antibodies (2.5 l/ml) or anti-PI4K ␤ antibodies (1 l/ml) for 2 h at 4°C with shaking. Pyk2 and PI4K ␤ immunoprecipitates were washed twice with modified RIPA buffer, once with 50 mM Tris-HCl, pH 7.4, and once with a protein kinase reaction buffer (50 mM Tris-HCl, 2.5 mM MgCl 2 , and 5 mM MnCl 2 , pH 7.4). The beads were resuspended in 40 l of kinase buffer containing 5 l of 100 M ATP and 10 Ci of [␥-32 P]ATP and further incubated at 30°C for 30 min with shaking. Immune complexes were dissociated by the addition of 100 l of 2ϫ elution buffer (50 mM Tris-HCl, pH 7.5, 1% SDS, and 10 mM dithiothreitol). After a 5-min centrifugation at 13,000 rpm at 4°C, the supernatant volume was adjusted to 1.5 ml by the addition of modified RIPA buffer. Anti-PI4K ␤ or anti-Pyk2 antibodies, respectively, were added and incubated overnight at 4°C on a rocking platform. After capture by protein A-Sepharose beads and three washes in modified RIPA buffer, immunocomplexes were dissolved in sample buffer, heated at 100°C for 2 min, and resolved on 7.5% SDS-polyacrylamide gel electrophoresis. Proteins were electrotransferred to nitrocellulose membranes and visualized by autoradiography on a PhosphorImager (Molecular Dynamics).
Morphological Studies-VSMCs in DMEM-0.5% BSA were seeded on glass coverslips coated with vitronectin (3 g/ml) in the presence or absence of ␣ v ␤ 3 small molecule antagonist (Ro 66) or anti-␣ v ␤ 5 blocking antibody (P1F6). In some experiments, VSMCs were pretreated with LY-294002 (20 M) or U-73122 (2 M) for 1 h to block the integrin-dependent signaling pathways. Adhesion was allowed for 4 h at 37°C, and morphological changes were directly microphotographed with a microscope.

RESULTS
␣ v ␤ 3 and ␣ v ␤ 5 Integrins Are Both Involved in VSMC Migration on Vitronectin-Expression of ␣ v ␤ 3 and ␣ v ␤ 5 integrins in cultured human or porcine VSMCs has already been demonstrated (17,28). Interaction of VSMC ␣ v ␤ 3 and ␣ v ␤ 5 integrins with their natural ligand, vitronectin, was then demonstrated in human atheromatous plaque (29). However, the respective roles of both ␣ v ␤ 3 and ␣ v ␤ 5 receptors in VSMC migration on vitronectin have not yet been elucidated. Under our experimental conditions, serum-deprived porcine VSMCs efficiently migrated on a vitronectin matrix (Fig. 2). VSMC migration was significantly detectable as early as 3 h after overlay of vitronectin-coated polyesters. To evaluate the importance of each integrin in our migration system, we used both specific blocking antibodies and RGD analog compounds from Hoffmann-La Roche, the functional characteristics of which are shown in In VSMCs, blockade of ␣ V ␤ 3 function by the specific antibody LM609 inhibited cell migration by 20% after 24 h (Fig. 3B). Ro 66 inhibited VSMC migration by 34% (Fig. 3B). Blockade of ␣ V ␤ 5 by the specific antibody P1F6 also decreased VSMC migration by ϳ35% (Fig. 3B). When LM609 ϩ P1F6, P1F6 ϩ Ro 66, or Ro 64 alone was applied on VSMCs, inhibition of cell migration was more efficient than blockade of each separate integrin (Fig. 3B). The peptide GpenGRGDSPCA (penRGD), which blocked both ␣ V ␤ 3 and ␣ V ␤ 5 integrins, induced also a strong inhibition of VSMC migration, whereas the control peptide GRGESP (RGE) was inefficient (Fig. 3B).
These data demonstrated that both integrins ␣ V ␤ 3 and ␣ V ␤ 5 are involved in VSMC migration on vitronectin. The effects of the Hoffmann-La Roche antagonists did not result from ␣ V ␤ 1 or ␣ 5 ␤ 1 integrin blockade. Indeed, it has been shown that ␣ V ␤ 1 is not involved in VSMC adhesion on vitronectin (17), and under our experimental conditions, a ␤ 1 blocking antibody (P4C10) did not affect VSMC migration (data not shown). A synergistic effect was observed on cell migration inhibition when both ␣ V ␤ 3 and ␣ V ␤ 5 were blocked. However, the LM609 antibody appeared to be less efficient than Ro 66 in blocking porcine ␣ V ␤ 3 integrin and induced a much stronger synergistic inhibition in association with P1F6 when human VSMCs were used (data not shown). It has been suggested previously that porcine ␣ V ␤ 3 integrin was poorly recognized by the LM609 antibody (28). Thus, in further experiments, we chose Ro 66 as a ␣ V ␤ 3 integrin blocker.
␣ V ␤ 3 but Not ␣ V ␤ 5 Engagement Regulates Phosphatidic Acid Production and PI3K IA Activity-Little is known about the variations of phosphoinositides in migrating cells. We took advantage of our cell migration system, which allows an easy separation of nonmigrating cells from migrating cells, to measure changes in 32 P-labeled phosphoinositides. As shown in Fig.  4B, an increase in the labeling of PtdIns-4-P, PtdIns-4,5-P 2 , 3-phosphoinositides, and PtdOH was measured after 24 h in migrating VSMCs. These data strongly suggested that the key enzymes of phosphoinositide metabolism, PI4K, PIP 5-kinase, PI3K, and PLC, were all stimulated upon VSMC migration. Because recovery of migrating cells was not sufficient to study the kinetics of phosphoinositide-dependent signal production, we measured phosphoinositide changes in P-Lys-adherent VSMCs at early times after overlay with vitronectin-coated PE (1-3 h). Interestingly, PtdIns-4-P and PtdOH production was increased (Fig. 4A). Thus, PI4K and PLC might be specific sig-nals correlated with the starting steps of cell migration (Ͻ3 h).
To evaluate the respective involvement of ␣ V ␤ 3 and ␣ V ␤ 5 integrins in the regulation of PLC and PI3K activities, we measured PtdOH production and PI3K IA (p85␣/p110) activity in the presence or absence of ␤ 3 or ␤ 5 blockers. As shown in Fig.  5A, Ro 66 inhibited PtdOH production by 26%, whereas the P1F6 antibody had no inhibitory effect. An inhibition similar to that of Ro 66 was obtained with the ␣ V -blocking peptide pen-RGD or the PLC inhibitor U-73122. Measurement of PI3K activity in p85␣ immunoprecipitates from nonmigrating or migrating VSMCs is shown in Fig. 5B. A strong stimulation of the p85-associated PI3K activity was observed upon cell migration. The ␣ V ␤ 3 blocker Ro 66 inhibited lipid kinase activity by 70%. In contrast, blockade of the ␣ V ␤ 5 integrin by P1F6 did not significantly modify PI3K activity. Incubation of cells with the PI3K inhibitor LY-294002 inhibited PI3K activity by 55%. It   50 Adhesion assay (HEK 293) IC 50 Ro 64 0.1 0.6 Ͼ100,000 84 9 5500 Ro 66 1.3 55000 Ͼ100,000 1800 232,000 597,000 Roche products Ro 64 and Ro 66 were characterized by solid phase assays and adhesion assays as described under "Experimental Procedures." For solid phase assays, inhibition of fibrinogen or vitronectin binding to isolated human ␣ v ␤ 3 or ␣ v ␤ 5 integrins, respectively, was measured by enzyme-linked immunosorbent assay in the presence or absence of Ro 64 or Ro 66 compounds. For adhesion assays, stable transfectants of HEK 293 cells (expressing ␣ v ␤ 3 or ␣ v ␤ 5 ) and parental HEK 293 cells (expressing ␣ 5 ␤ 1 ) were allowed to adhere on vitronectin or fibronectin, respectively, in DMEM-0.5% BSA and in the presence or absence of Ro 64 or Ro 66. Then, adherent HEK cells were fixed and stained with crystal violet, and absorbance was measured for cell adhesion evaluation and IC 50 determination.
should be noted that application of Ro 64 at 0.1 M inhibited VSMC migration by 40% but did not modify PtdOH production and PI3K IA activity (data not shown). Even though 0.1 M Ro 64 could not have sufficiently blocked ␣ V ␤ 3 function, its IC 50 is 10 times greater than that for ␣ V ␤ 5 (Table I). Therefore, these data reinforce the lack of ␣ V ␤ 5 involvement in PtdOH production and PI3K IA activity. Altogether, our data demonstrated that ␣ V ␤ 3 engagement, but not ␣ V ␤ 5 engagement, might act upstream of PLC and PI3K IA.
Evidence for a PI4K Activity Regulated by ␣ V ␤ 3 and ␣ V ␤ 5 Engagement-We then explored phosphoinositide signaling pathways downstream of ␣ V ␤ 5 integrin engagement upon VSMC migration. It was previously demonstrated that both ␣ V ␤ 5 and the FAK homolog Pyk2 are predominantly found outside focal adhesions (2,3,30). Moreover, Pyk2 bears putative binding sites for PI3K (31). We therefore hypothesized that Pyk2 could be a good candidate to interact with a PI kinase downstream of the ␣ V ␤ 5 integrin engagement. We assessed Pyk2 immunoprecipitates from VSMCs preincubated with ␣ V ␤ 3 or ␣ V ␤ 5 blockers. Activity assays performed on Pyk2 immunoprecipitates with poly(Glu-Tyr) or PtdIns as exogenous substrate revealed that: (i) the Pyk2-associated tyrosine kinase was increased upon VSMC migration (Fig. 6A), and (ii) a PI kinase (PI3K and/or PI4K) was associated with Pyk2 and was strongly increased upon VSMC migration (Fig. 6B). Treatment of VSMCs with ␣ V ␤ 3 or ␣ V ␤ 5 blockers decreased both Pyk2associated tyrosine kinase and PI kinase activities in migrating VSMCs (Fig. 6, A and B). Furthermore, pretreatment of intact cells with LY-294002 or the PLC inhibitor U-73122 induced an inhibition of the Pyk2-associated tyrosine and PI kinases (Fig.  6, A and B). These data demonstrate that a Pyk2-dependent signaling pathway is activated upon VSMC migration.
We further attempted to characterize this Pyk2-associated PI kinase from nonmigrating VSMCs (in culture). As shown in Fig. 7A, wortmannin (50 nM; added in the kinase assay) induced a 40% reduction of Pyk2-associated PI kinase activity, whereas the immunoprecipitated PI3K IA activity was reduced by 67%. In the same assays, LY-294002 (20 M) decreased the Pyk2-associated PI kinase activity by 28 Ϯ 6% (mean Ϯ S.E.; n ϭ 5) and the p85-associated PI3K activity by 60 Ϯ 6% (mean Ϯ S.E.; n ϭ 4) (data not shown). Furthermore, the Pyk2-associated PI kinase was able to phosphorylate PtdIns in the presence of either Ca 2ϩ or Mg 2ϩ (Fig. 7A), but the activity was decreased by 54% when Ca 2ϩ was used instead of Mg 2ϩ . By contrast, PI3K IA was not active in the presence of Ca 2ϩ (Fig.  7A). Thus, a PI kinase different from PI3K IA but still sensitive to wortmannin and LY-294002 might be isolated in Pyk2 immunoprecipitates.
As shown in Fig. 7B (left panel), using a TLC solvent allowing resolution of PtdIns-3-P and PtdIns-4-P, we detected the presence of both PI3K and PI4K products associated with Pyk2 from nonmigrating VSMCs (38% and 62% of the PIP synthesized, respectively). In migrating cells, we measured PI4K activity in Pyk2 immunoprecipitates, and no PI3K activity could be detected. This PI4K activity was inhibited by wortmannin or P1F6 treatment (Fig. 7B, right panel). Moreover, blockade of ␣ V ␤ 3 by addition of Ro 66 strongly decreased the Pyk2-associ-FIG. 6. Both ␣ v ␤ 3 and ␣ v ␤ 5 integrins regulate Pyk2-associated tyrosine kinase and PI kinase activities. VSMCs adherent on P-Lys-coated PE filters were allowed to migrate on vitronectin-coated PE filters for 24 h in DMEM-0.5% BSA. In some assays, before starting migration, VSMCs were incubated with integrin blockers (Ro 66, 650 M as an anti-␣ v ␤ 3 ; P1F6, 10 g/ml as an anti-␣ v ␤ 5 ), with a PI3K inhibitor (LY-294002, 20 M) or a PLC inhibitor (U-73122, 2 M) for 1 h at 37°C. When LY-294002, U-73122, or Ro 66 was used, control was performed in presence of the vehicle Me 2 SO. Nonmigrating and migrating control cells (NM and M) and migrating cells from each assay were lysed and incubated with an anti-Pyk2 antibody for immunoprecipitation as described under "Experimental Procedures." A, an in vitro kinase assay with poly(Glu-Tyr) as exogenous substrate was performed on Pyk2 immunoprecipitates from 1.4 ϫ 10 6 VSMCs as described under "Experimental Procedures." Results shown are from one of two representative experiments with similar results. Western blots of immunoprecipitated Pyk2 are displayed below. B, an in vitro PI kinase assay was performed on Pyk2 immunoprecipitates from 1.4 ϫ 10 6 VSMCs. Reaction products were separated by TLC and quantified on a Phos-phorImager. Results are expressed as the amount of PtdInsP produced in each assay compared with the amount of PtdInsP produced by cells that migrated in the control assay. Each bar represents the mean Ϯ S.E. of at least three experiments. Comparison to control migrating cells: ‫,ء‬ p Ͻ 0.05 (unpaired Student's t test). The cell migration percentages (mean Ϯ S.E. of at least three experiments) are shown at the bottom of the figure.   FIG. 7. Wortmannin and calcium sensitivity of the Pyk2-associated PI kinase. VSMCs in culture were lysed and incubated with anti-Pyk2 antibody or anti-p85 antibody for immunoprecipitation. An in vitro PI kinase assay was performed on Pyk2 and p85 immunoprecipitates from 9 ϫ 10 6 and 2 ϫ 10 6 VSMCs, respectively. Some assays were performed in the presence of wortmannin (50 nM) or with replacement of Mg 2ϩ by Ca 2ϩ (6 mM). Reaction products were separated by TLC and quantified on a PhosphorImager. Each bar represents the mean Ϯ S.E. of three experiments. Comparison to control conditions (two left columns): ‫,ءء‬ p Ͻ 0.01; and ‫,ءءء‬ p Ͻ 0.001 (unpaired Student's t test). Data on calcium sensitivity are representative of two independent experiments. B, PI3K and PI4K activities associated with Pyk2. Left panel, spots of PtdInsP produced by Pyk2-or p85-associated kinase activity as described above (A) were scraped, re-extracted, and separated by TLC to separate PtdIns-3-P from PtdIns-4-P as described under "Experimental Procedures" and quantified on a PhosphorImager. Right panel, VSMCs adherent on P-Lys-coated PE filters were allowed to migrate on vitronectin-coated PE filters for 24 h in DMEM-0.5% BSA in the presence or absence of the anti-␣ v ␤ 5 antibody P1F6. Lysate of migrating VSMCs (1.4 ϫ 10 6 cells) were immunoprecipitated with anti-Pyk2, and an in vitro PI kinase assay was performed as described in A in the presence or absence of wortmannin (50 nM). As a control, in the same experiment, we have performed a PI3K assay in immunoprecipitates from migrating cells (3 ϫ 10 5 VSMCs), as described in A. Reaction products were separated by TLC and quantified on a PhosphorImager. This figure is representative of two independent experiments with similar results. ated PI4K activity (data not shown). Altogether, these results showed that Pyk2-associated PI4K activity was regulated by both ␣ V ␤ 5 and ␣ V ␤ 3 integrins.
The PI4K ␤ Isoform Is Associated with Pyk2 and Activated upon ␤ 3 /␤ 5 Integrin Engagement-Wortmannin/LY-294002 sensitivity and sustained activity in the presence of calcium are two features of the PI4K type III family (32). Two distinct forms of type III PI4Ks, a 110-kDa ␤-form and a 230-kDa ␣-form, have been identified in mammalian cells (33). The yeast homolog of PI4K ␤, Pik1, and Pyk2 have both been involved in vesicular trafficking (34,35). We thus hypothesized that PI4K ␤ might be the PI4K isoform associated with Pyk2 in VSMCs. We performed kinase assays of PI4K ␤ or Pyk2 immunopre-cipitates followed by re-immunoprecipitation with antibodies to Pyk2 or PI4K ␤, respectively. As shown in Fig. 8A, phosphoproteins of 105 and 110 kDa were detected in Pyk2 and PI4K ␤ re-immunoprecipitates, respectively, whereas no immunoprecipitation of phosphoproteins was observed with nonimmune control. These data demonstrated that PI4K ␤ and Pyk2 can be isolated as complexes (Fig. 8A).
To check whether PI4K ␤ was the Pyk2-associated PI kinase regulated by ␣ V ␤ 3 and ␣ V ␤ 5 integrins, immunoprecipitates of PI4K ␤ followed by an in vitro lipid kinase assay were realized on VSMCs that migrated on vitronectin. An increase of PI4K ␤ activity was measured in migrating VSMCs compared with nonmigrating VSMCs. Accordingly, when ␣ V ␤ 3 or ␣ V ␤ 5 inhibitor was used, its activity in migrating cells was inhibited by 70% and 55%, respectively (Fig. 8B).
␣ V ␤ 3 and ␣ V ␤ 5 Integrins Are Differentially Involved in VSMC Spreading-Differences in the signaling pathways of ␣ V ␤ 3 and ␣ V ␤ 5 integrins strongly suggested that these two receptors were differentially involved upon VSMC migration. We have thus explored their role in cell spreading, a key step of cell migration. Treatment of vitronectin-adherent VSMCs with the ␣ V ␤ 3 antagonist Ro 66 induced a complete inhibition of cell spreading (Fig. 9). By contrast, the ␣ V ␤ 5 blocking antibody, P1F6, reinforced VSMC elongation and polarization (Fig. 9). We then applied LY-294002 or U-73122 on cells just before adhesion. Whereas LY-294002 did not significantly change VSMC spreading, U-73122 (a PLC inhibitor) treatment resulted in a total inhibition of spreading (data not shown). These results are in favor of the regulation of VSMC spreading by a ␣ V ␤ 3 -dependent PLC activity. However, pretreatment of VSMCs with U-73122 or LY-294002 induced an inhibition of the migration (Figs. 5 and 6), suggesting that LY-294002-sensitive PI kinases activities (PI3K IA and PI4K ␤) are required for VSMC migration.

DISCUSSION
The aim of this study was to explore phosphoinositide-dependent signaling pathways regulated by ␣ V ␤ 3 and ␣ V ␤ 5 integrins in VSMC migration onto vitronectin. We set up an original migration system allowing comparative measurement of phosphoinositides between nonmigrating and migrating cells. Moreover, we characterized new RGD analogs able to block ␣ V ␤ 3 or ␣ V ␤ 3 and ␣ V ␤ 5 functions. Here we show that ␣ V ␤ 3 and ␣ V ␤ 5 integrins are both involved in VSMC migration on vitronectin in vitro. This work is the first demonstration that endogenous vitronectin receptors transduce distinct phosphoinositide-dependent signals upon cell migration. Indeed, ␣ V ␤ 3 integrin regulates PtdOH production and PI3K IA activity in migrating VSMCs and promotes cell spreading, whereas ␣ V ␤ 5 is not involved in this signal transduction pathway and regulates other steps of cytoskeleton reorganization. Moreover, both ␣ V ␤ 3 and ␣ V ␤ 5 integrins regulate Pyk2-associated tyrosine kinase and PI4K ␤ activities.
PLC and PI3K are major signaling enzymes involved in cell migration. PLC␥1 and PI3K IA isoforms are activated down-FIG. 8. PI4K ␤ is associated with Pyk2 and regulated by ␣ v ␤ 3 and ␣ v ␤ 5 engagement. A, 24 ϫ 10 6 VSMCs in culture were lysed and incubated with anti-Pyk2 or anti-PI4K ␤ antibodies for immunoprecipitation. Pyk2 and PI4K ␤ immune complex kinase assays were performed in vitro in the presence of [␥-32 P]ATP. Immune complexes were then dissociated by SDS. Phosphorylated products were re-immunoprecipitated with the indicated antibodies and separated using 7.5% SDSpolyacrylamide gel electrophoresis. As a control, antibodies were replaced by normal rabbit serum (NRS). Phosphoproteins were detected using a PhosphorImager. The position of molecular weight standards is indicated. The positions of Pyk2 and PI4K ␤ proteins are indicated by arrowheads (p105 and p110, respectively). B, VSMCs adherent on P-Lys-coated PE filters were allowed to migrate on vitronectin-coated PE filters for 24 h in DMEM-0.5% BSA. In some assays, before starting migration, VSMCs were incubated with integrin blockers (Ro 66, 650 M as an anti-␣ v ␤ 3 ; P1F6, 10 g/ml as an anti-␣ v ␤ 5 ) for 1 h at 37°C. When Ro 66 was used, control was performed in the presence of the vehicle Me 2 SO. Nonmigrating and migrating control cells (NM and M) and migrating cells of each assay were lysed and incubated with anti-PI4K ␤ antibody for immunoprecipitation as described under "Experimental Procedures." An in vitro PI kinase assay was performed on PI4K ␤ immunoprecipitates from 1.4 ϫ 10 6 VSMCs. Reaction products were separated by TLC and quantified on a PhosphorImager. The bottom panel displays the Western blot of immunoprecipitated PI4K ␤.
FIG. 9. Effects of ␣ v ␤ 3 and ␣ v ␤ 5 integrins blockers on VSMC spreading on vitronectin. VSMCs in DMEM-0.5% BSA were allowed to adhere on vitronectin (3 g/ml)-coated 12-well plates in the presence or absence of P1F6 (10 g/ml) or Ro 66 (650 M) for 4 h at 37°C. The effects shown in Fig. 9 were observed as early as 10 min. Bars, 40 M. stream of integrin engagement through their interaction with the focal adhesion protein, FAK (8 -10). Addition of ␣ V ␤ 3 or ␣ V ␤ 5 inhibitors revealed that only ␣ V ␤ 3 integrin is implicated in the regulation of PtdOH production and PI3K IA activity. Moreover, ␣ V ␤ 3 integrin-vitronectin interaction is clearly required for VSMC spreading and seems to primarily involve PLC activation. Indeed, under our experimental conditions, PI3K inhibitors did not significantly modify VSMC spreading. The positive regulation of cell spreading by PLC activation can be explained by the regulation of PIP2/profilin association, formation of actin nucleation sites by diacylglycerols, and protein kinase C activation (6). Interestingly, our data suggested that PLC activation could be an early signal of VSMC migration.
Depending on the cellular model, ␣ V ␤ 5 was shown to be inside or outside focal adhesions or both. In vitronectin-adherent VSMCs, we found ␣ V ␤ 5 distributed mainly in diffuse spots in the cytoplasm, and ␣ V ␤ 3 in focal adhesion (data not shown). It has recently been demonstrated that integrins can transduce signals even when they are outside focal adhesion (12,36). VSMCs express both endogenous FAK and its homolog, Pyk2, and Pyk2 was found localized outside the focal adhesions (30,37). In contrast with FAK, but similar to ␣ V ␤ 5 , Pyk2 associates poorly with talin (30). Previous studies have demonstrated ␤ 1 , ␤ 2 , or ␤ 3 integrin-dependent activation of Pyk2 (38 -40). Pyk2associated tyrosine kinase activity is a good reflection of Pyk2 activation (41), and we found that it was up-regulated upon VSMC migration and by ␣ V ␤ 5 engagement. This result was in favor of activation of Pyk2-dependent signaling pathways upon VSMC migration. Indeed, a Pyk2-associated PI4K activity was regulated by ␣ V ␤ 5 engagement upon VSMC migration. PI4K was characterized by its wortmannin/LY-294002 sensitivity and its sustained activity in the presence of calcium, two features of the type III PI4K (32). Thus, it is unlikely that this PI4K is the type II PI4K described in association with the tetraspan family and ␣ 3 ␤ 1 integrin (12). In our assays, the Pyk2-associated tyrosine and PI4K activities were also regulated by ␣ V ␤ 3 engagement. Previous reports have demonstrated that Pyk2 was regulated by intracellular calcium, protein kinase C, and PI3K activity (31,42). We measured a strong inhibition of the Pyk2-associated PI4K activity after the addition of LY-294002 or U-73122. This suggests that PLC activity downstream of ␣ V ␤ 3 engagement might be involved in the regulation of the Pyk2-associated PI4K activity. The effect of LY-294002 may be due to the direct inhibition of the Pyk2associated PI4K and/or indirect inhibition of the ␣ V ␤ 3 -dependent PI3K IA activity.
We identified the Pyk2-associated PI4K as the PI4K ␤ isoform. As Pyk2, PI4K ␤ is mostly cytosolic, with a significant fraction associated with the Golgi (43). PI4K ␤ bears a prolinerich domain at the N terminus that might promote the interaction of this enzyme with SH3 domains (44). However, Pyk2 does not contain SH2 or SH3 domains, and association of PI4K with Pyk2 might be indirect. The modes of regulation of PI4Ks remain elusive, but recent data demonstrated that a type II phosphatidylinositol 4-kinase was activated through tyrosine phosphorylation by p56 lck (45). We have shown that PI4K ␤ could be tyrosine-phosphorylated in a kinase assay performed on Pyk2 immunoprecipitates (data not shown). This tyrosine phosphorylation might be the result of the action of Pyk2 or pp60 c-src . Indeed, pp60 c-src has been shown to interact with Pyk2 through its SH2 domain (46). pp60 c-src has an SH3 domain and might mediate the interaction between Pyk2 and PI4K ␤. Interestingly, it has been shown that PI4K ␤ is recruited by the small GTPase ADP-ribosylation factor, ARF, on the Golgi complex (47). On the other hand, Pyk2 has been involved in the regulation of vesicular transport through its interaction with Pap, a GTPase-activating protein, acting on the small GTPase family ARF (35). It remains to be determined whether PI4K ␤ is present in a complex with Pyk2, pp60 c-src , Pap, and ARF. PI4K ␤ phosphorylates only PtdIns to form PtdIns-4-P (43). The synthesized PtdIns-4-P might be a precursor for PtdIns-4,5-P 2 or phosphatidylinositol 3Ј,4Ј-bisphosphate to promote membrane budding or allow recruitment of other factors necessary for membrane trafficking.
We have found both PI4K and PI3K activities associated with Pyk2 from cultured VSMCs. However, only the PI4K activity was detected in Pyk2 immunoprecipitates from migrating VSMCs. The PI3K IA isoform has been shown to associate with Pyk2 in angiotensin II-stimulated VSMCs (48), in thrombin-stimulated platelets (49), and in macrophage colony-stimulating factor-stimulated macrophages (50). Under our conditions, the lack of PI3K activity in Pyk2 immunoprecipitates might be due to either a dissociation of this enzyme from the Pyk2 complex upon cell migration or to a PI3K activity below the level of detection. Thus, we cannot definitely rule out the association of Pyk2 and a PI3K upon VSMC migration.
Whereas ␣ V ␤ 3 has already been involved in vivo and in vitro in VSMC migration, recent reports based on animal models of neointima formation (51) and expression of vitronectin receptors in smooth muscle cells from human atheromatous plaque (29) have also suggested the potential role of ␣ V ␤ 5 integrin in VSMC migration. Our data clearly demonstrate that ␣ V ␤ 5 plays a major role in porcine VSMC migration on vitronectin in vitro, as well as in human VSMCs (data not shown). Moreover, both ␣ V integrins synergistically regulate this VSMC migration. Our data underline the differential involvement of both integrins in VSMC spreading, as has been shown previously for carcinoma cell spreading (52). Clearly, ␣ V ␤ 3 promoted VSMC spreading, whereas ␣ V ␤ 5 regulated VSMC elongation and polarization. The latter effect may be the result of cytoskeleton reorganization and/or a redistribution of cell adhesion receptors. Thus, we speculate that upon VSMC migration onto vitronectin, cell spreading is a step regulated by ␣ V ␤ 3 engagement through phospholipase C activation, which subsequently triggers phosphoinositide 3-kinase IA activity and Pyk2 activity. Conversely, ␣ V ␤ 5 engagement seems to be involved in dynamic changes of cytoskeleton and cell polarization, an effect likely to be due in part to its influence on the Pyk2-dependent signaling pathway. It would be important to define to what extent Pyk2/PI4K ␤ complex is involved in these processes.