Urokinase Stimulates Human Vascular Smooth Muscle Cell Migration via a Phosphatidylinositol 3-Kinase-Tyk2 Interaction*

Janus kinases Jak1 and Tyk2 play an important role in urokinase-type plasminogen activator (uPA)-dependent signaling. We have recently demonstrated that both kinases are associated with the uPA receptor (uPAR) and mediate uPA-induced activation of signal transducers and activators of transcription (Stat1, Stat2, and Stat4) in human vascular smooth muscle cells (VSMC). Janus kinases are not only required for Stat activation but may also interfere with other intracellular signaling pathways. Here we report that in VSMC, Tyk2 interacts with a downstream signaling cascade involving phosphatidylinositol 3-kinase (PI3-K). We demonstrate that uPA induces PI3-K activation, which is abolished in VSMC expressing the dominant negative form of Tyk2. The regulatory subunit p85 of PI3-K co-immunoprecipitates with Tyk2 but not with Jak1, Jak2, or Jak3, and uPA stimulation increases the PI3-K activity in Tyk2 immunoprecipitates. Tyk2 directly binds to either of the two Src homology 2(SH2)p85 domains in a uPA-dependent fashion. We provide evidence that the Tyk2-mediated PI3-K activation in response to uPA is required for VSMC migration. Thus, two unrelated structurally distinct specific inhibitors of PI3-K, wortmannin and LY294002, prevent VSMC migration induced by uPA. No migratory effect of uPA was observed in VSMC expressing the dominant negative form of Tyk2. Our results underscore the versatile function of Tyk2 in uPA-related intracellular signaling and indicate that PI3-K plays a selective role in the regulation of VSMC migration.

A major pathway for signal generation by cytokines, growth factors, and polypeptide hormones involves activation of the Janus tyrosine kinase family (Jak kinases) 1 and tyrosine phosphorylation of signal transducers and activators of transcription (Stat) proteins (1,2). Jak activation results in tyrosine phosphorylation of several Stat proteins that form homo-and heterodimers and translocate to the nucleus to regulate gene transcription by binding to specific promoter sequences of stimulated genes (3). Which members of the Jak and Stat families are activated varies greatly among different agonists and in different cell systems (4,5). We have recently demonstrated that the urokinase receptor (uPAR) utilizes the Jak/Stat pathway for intracellular transmission in human vascular smooth muscle cells (VSMC) and endothelial cells (6,7). uPA/uPAR is a multifunctional system involved in wound healing, tissue remodeling, immune response, and cancer by affecting cell migration, adhesion, and proliferation (8,9). Some of these functions require uPAR-dependent signal transduction, which remains imperfectly defined. uPAR is associated with two Janus kinases, Jak1 and Tyk2, which become activated in response to uPA and subsequently induce formation and activation of Stat1, Stat2, and Stat4 complexes (6,7,10). Both kinases co-localized with the uPAR to the leading edge of migrating VSMC (6), implying a possible contribution of the Jaks to cell migration, most likely via the interactions with other kinases or signaling molecules. Several reports have shown that Jaks and Stats interfere with multiple signaling cascades, such as Ras/mitogen-activated protein kinase pathway, phosphatidylinositol 3-kinase (PI3-K), Pyk2, and Src kinases. These cascades couple Jak/Stat to pathways with different downstream signaling functions (5,11).
We reasoned that PI3-K is a candidate for mediating uPAinduced VSMC migration. PI3-K phosphorylates the D3 position of phosphatidylinositol, and the phosphorylated lipid products of this enzymatic reaction may act on multiple downstream effectors (12,13). PI3-K is composed of two subunits, a regulatory p85 subunit and a 110-kDa catalytic subunit. The regulatory p85 subunit contains two Src homology 2 (SH2) domains, which bind to specific phosphotyrosine-containing motifs and have been implicated in mediating proteinprotein interactions (14). The ability of p85 SH2 domains to associate with other proteins links PI3-K to distinct signaling cascades required for control of cell growth and proliferation, adhesion and motility, differentiation, and survival (15). Moreover, recent reports underscore the importance of PI3-K in cell migration (16 -18). In this study, we investigated the extent to which PI3-K and Janus kinases are involved in the uPA-induced cell migration. Our findings demonstrate that in human VSMC, Tyk2, besides activation of Stat proteins in response to uPA, is the main uPA-dependent pathway of PI3-K activation. Tyk2 interacts with PI3-K via the SH2 domains of p85 catalytic subunit that, in turn, leads to p85 tyrosine phosphorylation and to PI3-K activation. The association of both kinases is critical to provide VSMC cytoskeletal reorganization in response to uPA required for cell migration.
Antibodies-Mono-and polyclonal anti-phosphotyrosine antibodies were from Affinity Research Products Ltd. (Exeter, UK) and Pierce; mono-and polyclonal anti-p85 PI3-K antibodies were from Transduction Laboratories (Lexington, KY) and Upstate Biotechnology, Inc. Mono-and polyclonal anti-Jak-kinase antibodies were purchased from Transduction Laboratories and Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Peroxidase-conjugated affinipure goat anti-mouse and goat anti-rabbit IgG and Cy3-conjugated anti-rabbit and anti-mouse IgG were purchased from Dianova (Hamburg, Germany). Alexa 488-conjugated phalloidin was from Molecular Probes, Inc. (Eugene, OR).
Cell Culture and Expression Constructs-Human VSMC from coronary artery were obtained from Clonetics (San Diego, CA). The cells were grown in SmGM2 medium (Clonetics) supplemented with 5% fetal bovine serum and were used between passages 3 and 6. For uPA stimulation experiments, the cells were cultured for 24 h in serum-free medium and were then treated with uPA as described (6). Human transformed kidney epithelial cells HEK 293 were obtained from Microbix Biosystems (Toronto, Canada).
Total RNA from human VSMC was extracted according to the acid phenol extraction protocol (19). Full-length wild type Tyk2 cDNA was performed by PCR on two pairs oligonucleotides (sense, 5Ј-GATGC-CCGGGTCTGTGCTGAATG; antisense, 5Ј-CACGGCCAAGAAAGAAA-AATAAGT) according to the high fidelity PCR protocol (Roche Diagnostics). The product of the reverse transcription reaction with tRNA was used as a template for PCR, and the PCR product was cloned into pCR2.1TOPO plasmid. The structure of Tyk2 was confirmed by sequencing. The recombinant adenovirus DNA for Tyk2 expression was generated by homologous recombination in Escherichia coli BJ5183 essentially as described previously (20). Briefly, an HindIII-XbaI fragment of the plasmid pCRIITyk2 encoding wild type human Tyk2 was ligated into XbaI-HindIII-digested pBKTGCMV. The resulting construct was cleaved with restriction endonucleases PacI and EheI to linearize it and to transform together with the linear adenoviral vector pAD1 in E. coli strain BJ5183. The recombinant adenovirus construct was cleaved with PacI and transfected into the packaging cell line HEK 293 by calcium phosphate precipitation (21). The concentration of the recombinant adenovirus was assessed based on the absorbance at 260 nm and on limiting dilution plaque assay (22). An expression construct that encodes Tyk2 that is mutated in its active center at residue 3079 (Lys into Glu) was generated by site-directed mutagenesis using the QuickChange protocol from Stratagene (La Jolla, CA), and the corresponding recombinant adenovirus was prepared as indicated above.
VSMC Infection, Tyk2 Overexpression, RNA Preparation, and Northern Blotting-VSMC were grown to 50% confluency and infected for 1 h with recombinant AdTyk2 adenovirus stock at a multiplicity of infection of 500 plaque-forming units/cell. The efficiency of infection was assessed by immunological staining with anti-Tyk2 antibody. Cells were serum-starved overnight 1 day after infection and used for experiments on the second day after infection. Total RNA was prepared from VMSC at the indicated times after Ad5Tyk2 infection after 6 h of serum starvation using standard protocol with Trizol reagent (Life Technologies, Inc.). The RNA was electrophoresed through an agarose-formaldehyde gel, transferred to a gene-screen plus (DuPont, Bad Homburg, Germany), and hybridized with nick-translated BamHI fragment from Tyk2 cDNA as described (23). An actin probe was used to confirm equal RNA loading on the gel.
Cell Stimulation and Immunoprecipitation-Subconfluent and serum-starved VSMC were treated with 1 nM uPA for 5-180 min at 37°C, lysed, and precleared as described previously (6). In some experiments, cells were pretreated with 100 nM wortmannin for 20 min at 37°C. For immunoprecipitation, precleared cell lysates containing 800 -1000 g of protein were incubated overnight at 4°C with 4 -5 g of antibody coupled to protein A-agarose. Precipitates were washed in PBS-Tween buffer and were used for PAGE and Western blotting. The blots were developed with the indicated antibodies, and the immune complexes were visualized by an enhanced chemiluminescence detection system. Membrane stripping was performed using 200 M ␤-mercaptoethanol, 62.5 mM Tris-HCl, pH 6.8, and 2% SDS for 45 min at 50°C.
In Vitro Phosphorylation Assay and Reimmunoprecipitation-For the in vitro phosphorylation assay, 400 -800 g of protein of precleared cell lysates were incubated with 4 -5 g of the indicated antibody coupled to protein A-agarose for 2 h. Precipitates were washed four times in lysis buffer and twice in kinase buffer (20 mM HEPES, pH 7.4, 10 mM MgCl 2 , 5 mM MnCl 2 , 1 mM dithiothreitol, 300 M sodium orthovanadate, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride). The kinase assay was performed in 40 l of kinase buffer containing 3-5 Ci of [␥-32 P]ATP (3000 Ci/mmol) for 10 min at 30°C. Precipitates were washed once with lysis buffer and twice in PBS-Tween buffer. The reaction product was eluted in 150 l of lysis buffer containing 0.5% SDS and 5 mM sodium orthovanadate and then subjected to a second round of immunoprecipitation with 1 g of the indicated antibody coupled to protein A-agarose overnight at 4°C. Phosphorylated proteins were analyzed by SDS-PAGE and autoradiography. In some experiments, gels containing phosphorylated proteins were soaked in 1 N KOH at 55°C for 2 h to hydrolyze phosphate on serine and threonine.
PI3-K Assay-Precleared cell lysates containing 600 -800 g of protein were incubated for 2 h at 4°C with 4 g of anti-PI3-K antibody against the p85 regulatory subunit and then precipitated on protein A-agarose. Precipitates were washed once with lysis buffer containing 300 M sodium orthovanadate and 1 mM phenylmethylsulfonyl fluoride and then four times with 10 mM HEPES, 100 mM NaCl, pH 7.4, and then subjected to the in vitro kinase reaction in a final volume of 50 l of reaction mixture containing 1 mg/ml L-␣-phosphatidylinositol, 3 mM MgCl 2 , 15 mM ATP, and 3-5 Ci of [␥-32 P]ATP (3000 Ci/mmol) for 10 min at 30°C. Phase separation of lipids was performed in two steps by n-hexane/isopropyl alcohol (26:14) and 2 N KCl/HCl (8:0.25). Phosphorylated lipids were separated by thin layer chromatography on aluminum sheets silica gel 60W (Merck, Germany) in chloroform/methanol/ H 2 O/25% ammonium (45:35:7.5:2.8; v/v/v/v) and visualized by autoradiography.
Fusion Protein Precipitation Assay-GST fusion proteins containing the single C-or N-terminal p85 SH2 domain (GST-p85-C-SH2 and GST-p85-N-SH2) bound to glutathione-agarose beads were used for affinity precipitation. Precleared cell lysates containing 400 -800 g of protein were incubated with 3-4 g of SH2 domain conjugates for 2 h at 4°C. Precipitates were washed three times with PBS-Tween buffer. Precipitated proteins were eluted with 2-fold Laemmli sample buffer containing 20 mM dithiothreitol and 10 mM glutathione and were used for PAGE and Western blotting.
Chemotaxis Assay-Chemotaxis assay was performed using modified Boyden chambers with polyvinylpyrrolidone-free polycarbonate filter membranes, 8-m pore size, as described (24). 25,000 -30,000 cells in serum-free SmGM2 medium were added to the upper well of the Boyden chamber. uPA was diluted in serum-free SmGM2 and added to the lower well of the Boyden chamber, and migration was allowed for 4 h. When chemotaxis was performed in the presence of the PI 3-K inhibitors wortmannin or LY294002, these substances were added to the upper well. All experiments were performed in triplicates. Cell migration was quantified by densitometry, and cell migration in the absence of chemoattractant was taken as 100%.
Wounding Assay and Time Lapse Videomicroscopy-Wounding experiments using confluent VSMC monolayer were performed as described previously (6). Before wounding, cells were cultured in serumfree SmGM2 medium for 6 h. In migration experiments using PI3-K inhibitors, cells were pretreated with wortmannin or LY294002 in serum-free medium for 30 min before wounding, and then inhibitors were added again to the serum-free medium after wounding together with or without uPA. Cell migration was monitored by time lapse imaging using an endoscope (telecam PAL 20210036, Storz, Germany) attached to an Axioplan microscope (Zeiss) and acquisition and analysis software (Avid Videoshop, Avid Desktop Software Inc., Microsoft Excel). The wounds were viewed inside an environmental chamber under constant temperature (37°C) and humidified in 5% CO 2 and air (CTI Controller 3700; Zeiss). Microscopic recordings were started immediately after wounding, and then further images were taken every 30 min for 9 h following wounding. Results are the mean number of migrated cells Ϯ S.D. at indicated time points or -fold stimulation of at least five separate experiments.
Immunofluorescence Microscopy-Cells were seeded and cultured on glass coverslips, and wounding was performed as indicated above. Cells were allowed to migrate for 8 h at 37°C and then treated for different time periods with appropriate inhibitors or stimulators. After incubation, cells were fixed with 4% paraformaldehyde in PBS for 20 min at room temperature, washed three times with PBS, and permeabilized with 0.1% Triton X-100 in PBS for 3 min at room temperature. Then cells were stained with Alexa 488-conjugated phalloidin for 20 min at room temperature. Coverslips were mounted with Vectashield mounting medium. Immunofluorescent staining for Tyk2 was performed as described previously (6). Color images were captured using a SensiCam 12-bit CCD camera (PCO Computer Optics, Kelheim, Germany) and acquisition and analysis software (AxioVision and KS 300; Carl Zeiss, Vision, Munich, Germany) running on a Power PC 586/200MMX-64MB (Inteq, Berlin, Germany). Images for Alexa 488 and Cy3 staining were captured digitally and were imported as TIF files into Illustrator for analysis and printing.

RESULTS
uPA Induces PI3-K Activation and Tyrosine Phosphorylation of p85-To address the potential role of PI3-K in uPA-mediated signaling, we first assayed PI3-K activity directly by measuring the levels of the product, phosphatidylinositol phosphate, in serum-starved VSMC stimulated with uPA at different time points. As shown in Fig. 1A, uPA induced a significant increase (up to 3-fold) in PI3-K activity in a time-dependent manner with a sustained PI3-K activation observed after 3 h. The PI3-K-specific inhibitor wortmannin, which binds covalently to the catalytic p110 subunit of PI3-K and inhibits PI3-K irreversibly at nanomolar concentrations (12) completely blocked the uPA-induced stimulation of enzyme activity (Fig. 1B). To assess a potential contribution of uPA's proteolytic activity to the mechanisms of PI3-K activation, the amino-terminal fragment of uPA, ATF, was used for cell stimulation. ATF provided the same effect (Fig. 1C), confirming the involvement of proteolytically inactive uPA in signaling events. We also examined the requirement for uPAR upon PI3-K activation. For this purpose, we used soluble recombinant uPAR, which is a known competitor for the uPA-and ATF-uPAR binding. The data presented in Fig. 1C clearly demonstrate that the pretreatment of VSMC with soluble recombinant uPAR significantly decreased the uPA/ATF-induced PI3-K stimulation (shown for ATF).
We next examined PI3-K p85 subunit phosphorylation after uPA stimulation using an in vitro kinase assay combined with two rounds of immunoprecipitation with anti-p85 antibody (Fig. 1D, upper panel). The results demonstrate that uPA markedly increased phosphorylation of the p85 protein. The band was resistant to alkaline hydrolysis, indicating tyrosine phosphorylation (Fig. 1D, middle panel). The equal loading of precipitated material on the gel was confirmed by immunoblotting (Fig. 1D, lower panel). A, VSMC were treated with 1 nM uPA for the indicated times at 37°C, and the PI3-K assay was performed with subsequent thin layer chromatography and autoradiography visualizing phosphatidylinositol 1,4,5trisphosphate (PtdIns(3)P) products. B, cells were untreated or treated with uPA for 30 min and/or with wortmannin, as indicated. C, VSMC were stimulated with 1 nM ATF for 45 min in the presence or absence of 20 nM soluble recombinant uPAR. Recombinant soluble uPAR was expressed and purified as described previously (23). Quantification of the results by densitometry is shown below each panel. D, phosphorylation of p85 was assayed in control VSMC and after cell activation with 1 nM uPA by in vitro kinase assay combined with two rounds of immunoprecipitation (IP and re-IP) with anti-p85 polyclonal antibody (Ab), as described under "Experimental Procedures." Phosphorylated proteins were subjected to SDS-PAGE and autoradiography before (upper panel) and after (middle panel) KOH treatment. The equal loading of precipitated proteins on the gel was confirmed by immunoblotting (lower panel). WB, Western blot.
Tyk2 and PI3-K Are Associated-To examine the potential link between PI3-K and uPA-mediated Jak/Stat signaling, we first performed immunoprecipitation with anti-Jak antibodies and looked for the p85 subunit of PI3-K in the immunoprecipitates. Proteins that were co-precipitated with Jaks were subjected to in vitro kinase assay as above, and the phosphorylated proteins were reimmunoprecipitated with anti-p85 antibody. As a control for p85 recovery, both rounds of immunoprecipitation were performed with anti-p85 antibody. As shown in Fig.  2A, upper panel, one band corresponding to p85 was detected in Tyk2, but not in Jak1, Jak2, or Jak3 immunoprecipitates. The identity of the immunoprecipitated p85 band was confirmed by immunoblotting with anti-p85 antibody ( Fig. 2A, lower panel). To verify that antibodies to Jak1, Jak2, and Jak3 were behaving appropriately under our experimental conditions, the additional positive controls were performed, confirming that all three kinases might be precipitated from the VSMC lysates (Fig. 2B). These data demonstrate that p85 can directly and constitutively associate with Tyk2 in VSMC. Moreover, Tyk2 immunoprecipitates possessed PI3-K activity, which was enhanced by uPA stimulation (Fig. 2C).
To further characterize this interaction, we overexpressed the wild type Tyk2 in VSMC using adenoviral cell infection. The overexpression enabled us to obtain cell lysates enriched in Tyk2, which is otherwise expressed at a low level in native VSMC. Immunofluorescent staining of control uninfected cells and Ad5Tyk2-VSMC with anti-Tyk2 antibody confirmed the efficiency of cell infection with the recombinant adenovirus under these conditions (Fig. 3A). Northern and Western blot analysis were further used to confirm the expression of Tyk2 mRNA and protein in VSMC infected with adenovirus and cultured for different time periods. As shown in Fig. 3B, VSMC infection with Ad5Tyk2 resulted in a dramatic increase in Tyk2 mRNA and protein levels relative to uninfected cells. The Tyk2 protein expression reached a maximum 2 days after infection and decreased within the next 3 days. In further experiments, we therefore used VSMC cultured for 2 days after infection with Ad5Tyk2.
To investigate the interaction mechanisms between Tyk2 and p85, in vitro binding experiments were performed. We used two GST fusion proteins containing the single C-and N-termi-nal Src homology 2 (SH2) domains of the human p85 subunit of PI3-K (GST-p85-C-SH2 and GST-p85-N-SH2) immobilized on the glutathione-agarose beads. The beads were incubated with the lysates of unstimulated and uPA-stimulated Ad5Tyk2-VSMC, and the precipitated proteins were analyzed by immunoblotting with anti-Tyk2 antibody. As shown in Fig. 3C, one major band corresponding to Tyk2 was detected in both GST-p85-C-SH2 and GST-p85-N-SH2 precipitates from uPA-stimulated cells. Tyk2 bands identified in unstimulated cells were significantly weaker. These results strongly suggest that Tyk2 directly associates with PI3-K through the binding to both Cand N-terminal SH2 domains of p85 and that this association is uPA-inducible.
PI3-K and Tyk2 Are Required for uPA-induced VSMC Migration-Recent reports imply that PI3-K is involved in cell migration control (16 -18). Although uPA is known to regulate cell migration (25), the underlying molecular mechanisms are unclear. Therefore, we next examined the possible impact of PI3-K on VSMC migration in response to uPA. For this purpose, two main approaches were used, namely directional VSMC migration in Boyden chambers and cell movement in a wounded VSMC monolayer monitored by time lapse imaging (see "Experimental Procedures" for details). Fig. 4A displays the data on directional migration of VSMC along the uPA gradient tested in a microchemotaxis Boyden chamber. After 4 h, cell migration in response to uPA was significantly enhanced about 3-fold, compared with migration in the presence of medium alone. To examine the requirement of PI3-K for uPA-promoted cell migration, VSMC were subjected to migration assays in the presence of two unrelated structurally distinct specific PI3-K inhibitors, wortmannin and LY294002. Cell treatment with both inhibitors decreased uPA-related cell migration with no statistically significant effect on the basal migration (Fig. 4A).
To further characterize VSMC migration, an in vitro injury assay was performed. This cell migration model was used to assess the potential effects of constant uPA doses and to analyze the role of PI3-K in this process. The cell migration rate was calculated from the cell number migrating into the wounded area within 9 h after wounding. In these experiments, VSMC migration significantly increased in the presence of uPA FIG. 2. Tyk2 and PI3-K are associated in VSMC. A, proteins co-precipitated (IP) with antibodies (Ab) to Jak1, Jak2, Jak3, or Tyk2 were subjected to in vitro kinase assay and reimmunoprecipitated (re-IP) with anti-p85 antibody (upper panel). The left lane shows the results of the in vitro kinase assay combined with two rounds of immunoprecipitation with anti-p85 antibody as control for p85 recovery. Phosphorylated proteins were subjected to SDS-PAGE and autoradiography. The identity of the immunoprecipitated p85 band was confirmed by immunoblotting with anti-p85 antibody (lower panel). B, Janus kinases were immunoprecipitated from VSMC lysates under the same conditions as used in A and were identified by immunoblotting as indicated. C, PI3-K activity was measured in Tyk2 immunoprecipitates of unstimulated (control) or uPA-stimulated (uPA) VSMC (upper panel). Products of PI3-K were separated by thin layer chromatography and visualized by autoradiography. The lower panel shows the results of densitometric quantification. WB, Western blot. compared with migration with medium alone as early as 2 h after wounding (Fig. 4B). Furthermore, VSMC treatment with either wortmannin or LY294002 completely abolished the effects of uPA. Thus, uPA regulation of cell migration is PI3-K-dependent.
The next series of experiments were performed to determine whether Tyk2 is required for PI3-K-mediated VSMC migration in response to uPA. VSMC were adenovirally infected to express the dominant negative form of Tyk2 devoid of kinase activity, as described under "Experimental Procedures." The efficiency of cell infection was confirmed by immunostaining, Northern and Western blotting (Fig. 5, A and B). As shown in Fig. 5C, uPA-induced PI3-K activation was blocked in Ad5Tyk2KE-expressing cells, whereas in wild type Ad5Tyk2expressing VSMC, PI3-K activity increased after uPA stimulation. These results demonstrate that Tyk2 is required for the uPA-related PI3-K activation and that Tyk2 functions presumably upstream of PI3-K in VSMC.
To elucidate whether Ad5Tyk2KE mutant also inhibits PI3-K-mediated VSMC migration in response to uPA, cell migration assays were performed using the wounding model, as described above. As shown in Fig. 5D, VSMC expressing wild type Tyk2 demonstrated an approximately 2-fold increase in cellular migration in the presence of uPA. By contrast, uPA did not affect cell migration in Ad5Tyk2KE-expressing VSMC.
Association of Tyk2 and PI3-K Is Essential for uPA-related Cytoskeletal Reorganization in Migrating VSMC-Cell motility is generally associated with polarization of initially unpolarized cells. An active leading edge facing the cell-free area is formed. In addition, focal contacts, stress fibers, and actin filaments are rearranged (26). To examine the cytoskeletal reorganization resulting from uPA-induced activation of Tyk2 and PI3-K in migrating VSMC in the wounding model, we again performed time lapse videomicroscopy and immunofluorescent staining with Alexa 488-conjugated phalloidin. As reported by others (24) and as shown in Fig. 6, VSMC migrating in the presence of uPA developed a polarized morphology, with extended lamellipodia extrusions at the leading edge facing the open space (Fig. 6B). Formation of actin-rich lamellipodial extensions at the leading edge and abundant stress fibers in response to uPA was observed also in VSMC expressing wild type Tyk2 (Fig. 6F). The dynamic cytoskeletal activity of both cell types migrating without uPA was significantly weaker (Fig. 6, A and E). By contrast, lamellipodial activity was completely different in VSMC expressing a Tyk2 mutant form. Ad5Tyk2KE-VSMC did not display a strong morphological polarity with clear leading edge and revealed in the presence of uPA mainly small lateral protrusions localized at the sides and not at the leading edge of migrating cells (Fig. 6J).
To examine the contribution of PI3-K to the observed cytoskeletal rearrangements, cells were treated with PI3-K inhibitors. VSMC treatment with wortmannin caused a complete loss of uPA-induced polarity (Fig. 6D). In Ad5Tyk2-VSMC treated with wortmannin, cell polarity was also disturbed; the protrusive leading edge, clearly developed in the presence of uPA and absence of wortmannin, was damaged (Fig. 6H). These cells came to be similar to Ad5Tyk2KE-VSMC, whose uPA-induced lamellipodial activity can be seen all around the cells' periphery (Fig. 6, J and L). Moreover, wortmannin treatment of VSMC expressing wild type and the mutant form of Tyk2 resulted in the reorganization of stress fibers. Interestingly, these effects were especially pronounced in the presence of uPA. As can be seen in Fig. 6, H and L, uPA stimulation of wortmannin-pretreated VSMC resulted in disassembly of stress fibers, which lost organization into typical long bundles of parallel or radially directed stress fibers.
We previously demonstrated that in migrating VSMC, Tyk2 is polarized to the leading edge (6). The strong polarity in Tyk2 intracellular redistribution was also observed in migrating VSMC expressing wild type Tyk2 (Fig. 7A). By contrast, in VSMC expressing the Tyk2 mutant form, Tyk2 was randomly distributed within the cell that might explain the loss of cytoskeletal polarity of these cells (Fig. 7, C and D).

DISCUSSION
We provide evidence implying novel roles for the Janus kinase Tyk2 and PI3-K in human coronary VSMC. We demonstrate that in addition to known regulation of cell transcription via Stat proteins, Tyk2 associates with PI3-K through the Src homology 2 domains of p85 subunit, thereby providing a crucial link between two signaling pathways. Probably the most important finding is that the revealed Tyk2 and PI3-K association within the cell and their activation are urokinase-responsive and are required for the uPA-related VSMC migration. This observation is the first demonstration that PI3-K is essential for at least some of the uPA/uPAR functions in VSMC and that Tyk2 is the main uPA-dependent pathway of PI3-K activation.
VSMC proliferation and migration into the intima after vascular injury, as well as their formation of neointima, contribute to vessel narrowing and are pivotal to the atherosclerotic process (27). uPA and uPAR are active participants in these processes by regulating wound healing, tissue remodeling, and immune responses (8). In addition to the effects mediated by proteolysis, uPA and uPAR also display biological functions that are not directly attributable to the formation of plasmin but rather to the induction of cell migration and proliferation control (28). The link between uPA/uPAR and cell motility was established over a decade ago (29 -31). Current data confirm the uPA-dependent cell migration in a wide variety of cell types (32)(33)(34)(35)(36)(37)(38). Nevertheless, the proposed molecular mechanisms have been conflicting. The uPA/uPAR-related migratory responses seem to be highly cell specific, implying some structural specificity and diversity of underlying signaling events. Thus, in human breast cancer cells MCF-7 and HT 1080 fibrosacroma cells, uPA-initiated cellular motility required activation of signaling cascade including Ras, mitogen-activated protein kinase kinase, extracellular signal-regulated kinase, and myosin light chain kinase as downstream effectors (35,36). On the contrary, migration of human epithelial cells seemed to involve the uPA-activated protein kinase C (33), whereas in cells of monocytic lineage these effects were attributed to the activation of protein tyrosine kinases of the Src family (34,39). In addition, uPA/uPAR-related cell migration is integrin-dependent (9,40,41).
The molecular machinery of uPA/uPAR-related VSMC migration remains sparsely explored, although the recent generation of transgenic mice deficient in uPA and uPAR demonstrated that VSMC migration is dependent on the fibrinolytic system and is decisive for the severity of vascular damage (42,43). Separate recent findings confirm migration of rat VSMC initiated by uPA/uPAR binding and suggest involvement of Src tyrosine kinase and G protein in this process (24). We suggested a role for the Janus kinases in VSMC migration and showed that Jak1 and Tyk2 were co-localized with the uPAR to the leading edge of migrating VSMC (6). However, to generate migratory responses, Jaks were expected to signal independently of Stat activation most likely through a link to an additional unknown pathway.
PI3-K is central to cell migration processes regulated by cytokines and growth factors in diverse cell types including VSMC (44). Moreover, PI3-K possesses a high capacity to cooperate with other signaling pathways to mediate a required functional response (15). Interestingly, several recent reports provide evidence for the interference of PI3-K and Jak/Stat signaling cascade. Thus, the PI3-K p85 regulatory subunit was shown to bind directly to Stat3 protein, which served as an adapter to couple the PI3-K signaling pathway to the interferon receptor in Daudi cells (45). A similar interaction was demonstrated for the p85 subunit and Stat5 protein in a bone marrowderived Ba/F3 cell line, where both pathways cooperated to mediate interleukin-3-dependent suppression of apoptosis (46). Several other reports provided further evidence for coordinated activation of PI3-K and Stat proteins leading to the functional cooperation of both signaling cascades (47,48). Moreover, recent studies demonstrated the ability of Jaks to associate with and to regulate the p85 subunit of PI3-K, as was shown for Jak3 in human T cells (49), Jak1 in cardiac myocytes (50), and Jak2 in human neutrophils (51).
Consistent with these reports, we find that Janus kinase Tyk2, but not Jak1, Jak2, or Jak3, is specifically associated with PI3-K in human VSMC. Moreover, this association is uPA-dependent. The fact that Jak1 was not co-immunoprecipitated with PI3-K is of interest because in our previous studies we did not observe any difference between Tyk2 and Jak1 in terms of uPA stimulation, uPAR association, and polarization to the leading edge of migrating cells (6,7). The inability to reveal Jak1-PI3-K complexes suggests that this kinase is most likely involved exclusively in the transcriptional regulation via Stat proteins or might perform an additional function in the uPA/uPAR-related signaling. Although we found no association between Jak1 and PI3-K, their cooperation cannot be completely excluded. Nevertheless, two kinases performing the same task would be redundant.
To address the molecular basis of Tyk2-PI3-K interaction, we performed an adenovirus-mediated Tyk2 overexpression in VSMC. We performed Tyk2 pull-down assays using GST-p85 fusion proteins composed of either the C-or N-terminal SH2 domains of p85. Tyk2 bound to both N-and C-terminal SH2 domains of the p85 regulatory PI3-K subunit. This result is consistent with the findings of others demonstrating that these domains mediate specific protein-protein interactions (14). Tyk2 has a candidate motif, YXXM, which is a potential site for p85 binding. However, determination of whether or not SH2 p85 domains are responsible for the interaction with Tyk2 in the context of above motif requires further verification. Our data are the first to demonstrate that following uPA stimulation, Tyk2 and PI3-K form a specific protein complex in VSMC. Although Tyk2 constitutively associates with p85, uPA induces a strong activation of this interaction. Moreover, uPA induces PI3-K activation within the complex, which we believe is mediated by tyrosine phosphorylation of the PI3-K p85 subunit. This hypothesis, which we are currently pursuing, is supported by the fact that growth factor-promoted p85 phosphorylation increases PI3-K activity (52). However, the increase in PI3-K activity we measured in Tyk2 precipitates in response to uPA was significantly less than the uPA-induced PI3-K activation in whole cell lysates. An intriguing question is raised by these observations, which imply additional still unknown uPA/uPARdependent PI3-K functions.
Our results highlight the importance of PI3-K in the regulation of cell migration in response to uPA. We demonstrate the involvement of PI3-K in uPA-induced signaling leading to cell migration in two in vitro models, the Boyden chamber and the wounding assay. Both techniques allowed us to analyze different parameters of cell migration. Specific PI3-K antagonists, LY294002 and wortmannin, completely abrogated the uPAresponsive increase in cell motility. The observed ability of uPA to initiate sustained PI3-K activation might facilitate the long term migration process.
One of the issues raised by these observations is whether PI3-K lies upstream or downstream of Tyk2 in the reavealed pathway. Our demonstration that in VSMC expressing the Tyk2 dominant negative form PI3-K activation in response to uPA is completely blocked implicates that PI3-K activation acts broadly downstream of Tyk2 in the signal transduction cascade. Moreover, these data support the concept that in VSMC Tyk2 is central to the PI3-K regulation. This positioning is further supported by our finding that a dominant negative Tyk2 blocks uPA-related cell migration. Upon stimulation, Tyk2, as certain other players in signaling machineries, relocates to the leading edge of the cell membrane that might enhance reaction-limited signal transduction. The results of our study on VSMC morphology reported here imply that polarization of Tyk2 might contribute to the lamellipodial activity required for a spatial asymmetry of migrating cells. However, it appears that the loss of PI3-K activity has similar consequences on lamellipodial activity and migration process as blockage of Tyk2 by its mutant. Overall, these data favor the view that a functional relationship between two kinases is decisive for VSMC cytoskeletal reorganization and migration. Our data are consistent with other reports demonstrating the central role of PI3-K in initiating actin cytoskeletal rearrangements, cell polarization, and cell migration in several cell systems (53,54). During these processes, PI3-K is generally translocated from the cytosol to the cytoskeleton-associated fraction presumably via a link to additional proteins (55). It is likely that Tyk2-PI3-K association via the SH2 domains of the p85 subunit is responsible for bringing PI3-K to the cytoskeletonassociated subcellular fraction, since loss of Tyk2 polarization in Ad5Tyk2KE-VSMC correlates with the loss of PI3-K activation and of leading edge formation. However, it is also possible that other adaptor molecules may facilitate translocation of PI3-K. In rat VSMC, Src kinase was presumed to cooperate functionally with uPAR and integrins at the leading edge upon the migration process, since uPA-caused a c-Src redistribution from the cytoplasm to plasma membrane (24). These data, coupled with the recent demonstration that the integrin-dependent cell migration requires both Src family kinases and PI3-K (55), as well as with our findings, suggest the existence of a complex functional unit formed at the leading cell membrane in response to uPA. The localized regulation of signaling molecules, such as Src, Tyk2, and PI3-K, by uPA/uPAR and integrins may provide an efficient mechanism for targeting downstream functional effects of these kinases as well as for the balance between these pathways.