A Novel Mechanism of G Protein-dependent Phosphorylation of Vasodilator-stimulated Phosphoprotein*

Vasodilator-stimulated phosphoprotein (VASP) is a major substrate of protein kinase A (PKA). Here we described the novel mechanism of VASP phosphorylation via cAMP-independent PKA activation. We showed that in human umbilical vein endothelial cells (HUVECs) α-thrombin induced phosphorylation of VASP. Specific inhibition of Gα13 protein by the RGS domain of a guanine nucleotide exchange factor, p115RhoGEF, inhibited thrombin-dependent phosphorylation of VASP. More importantly, Gα13-induced VASP phosphorylation was dependent on activation of RhoA and mitogen-activated protein kinase kinase kinase, MEKK1, leading to the stimulation of the NF-κB signaling pathway. α-Thrombin-dependent VASP phosphorylation was inhibited by small interfering RNA-mediated knockdown of RhoA, whereas Gα13-dependent VASP phosphorylation was inhibited by a specific RhoA inhibitor botulinum toxin C3 and by a dominant negative mutant of MEKK1. We determined that Gα13-dependent VASP phosphorylation was also inhibited by specific PKA inhibitors, PKI and H-89. In addition, the expression of phosphorylation-deficient IκB and pretreatment with the proteasome inhibitor MG-132 abolished Gα13- and α-thrombin-induced VASP phosphorylation. In summary, we have described a novel pathway of Gα13-induced VASP phosphorylation that involves activation of RhoA and MEKK1, phosphorylation and degradation of IκB, release of PKA catalytic subunit from the complex with IκB and NF-κB, and subsequent phosphorylation of VASP.

Vasodilator-stimulated phosphoprotein (VASP) 4 is purified and characterized as a 46-kDa membrane-associated protein that can be phosphorylated by cAMP-and cGMP-dependent protein kinases (PKA and PKG, respectively) (1). VASP is the founding member of the Ena/ VASP family of proteins, which consists of VASP, Drosophila Enabled (Ena), a mammalian Ena homolog Mena, and Ena-VASP-like protein EVL (2)(3)(4). Similarly to other members of the Ena/VASP family of pro-teins, VASP contains the central proline-rich region flanked with Ena/ VASP homology domains 1 and 2 (2).
VASP is highly expressed in vascular endothelial cells, platelets, smooth muscle cells, and fibroblasts where it can be found in focal adhesions, along stress fibers, and in the areas of highly dynamic membrane activity, such as extending lamellipodia and filopodia (2,5,6). VASP plays an important role in physiological and pathophysiological responses of platelets and endothelial cells. Recent studies (7) showed that VASP is an essential factor that negatively regulates platelet adhesion; in VASP knockout mice, platelet adhesion to the vessel wall was enhanced. Moreover, the loss of VASP increases platelet adhesion to the atherosclerotic endothelium and subendothelial matrix (7). More importantly, phosphorylated VASP may participate in the endothelial barrier function and tight junction regulation. Phosphorylated VASP localizes to endothelial cell-cell junctions and may be co-immunoprecipitated with zonula occludens 1 after PKA activation (8).
There is a growing body of evidence that VASP has complex effects on cytoskeletal organization and cell motility by regulating actin network geometry and thereby controlling the protrusive behavior of the cell (9). It was shown that VASP could induce polymerization of G-actin into F-actin bundles in in vitro assays. VASP was also shown to stabilize F-actin (10), to nucleate F-actin assembly (11,12), to antagonize capping protein activity and promote actin filament elongation (13), and to facilitate Arp2/3-dependent actin polymerization (14,15). More importantly, the lamellipodial protrusion rate was shown to correlate positively with the intensity of green fluorescent protein-VASP at the leading edge (16). All these molecular events are involved in the regulation of cell motility. VASP phosphorylation is a modulator for VASP-dependent regulation of the actin cytoskeleton. VASP can be phosphorylated at three sites (human VASP), serine 157, serine 239, and threonine 278 (corresponding to serine 153, serine 235, and threonine 274 in murine protein). Phosphorylation of serine 157 leads to a marked retardation in electrophoretic mobility and shift in apparent molecular mass from 46 to 50 kDa in SDS-PAGE (17).
PKA is an important effector enzyme that is commonly activated by cAMP in response to cAMP-elevating agents. The established mechanism of PKA activation in response to various agonists involves the stimulatory G protein, G s , which activates adenylyl cyclase resulting in a production of cAMP. Binding of cAMP to PKA regulatory subunits (PKAr) leads to the release and activation of catalytic PKA subunits (PKAc) (18). More importantly, our laboratory reported two novel cAMP-independent mechanisms of PKA activation. One study showed that vasoactive peptides endothelin-1 and angiotensin II activate PKA by inducing phosphorylation and degradation of the inhibitor of B (IB), subsequently releasing PKAc from inhibition by IB (19). A similar mechanism was reported for PKA activation upon lipopolysaccharide (20) and thrombin stimulation (21). In addition, we reported that G␣ 13 , upon interaction with protein kinase A-anchoring protein 110 (AKAP110), induces release of the PKAc from the AKAP110-PKAr complex, resulting in the PKA activation (22). G␣ 13 protein regulates multiple signaling pathways by transducing signals from the cell surface to activate diverse cellular responses. It regulates the activity of the Na ϩ /H ϩ exchanger NHE1 (23), apoptosis (24), and the extracellular signal-regulated kinase and c-Jun NH 2 -terminal kinase pathways (25) and participates in embryonic development (26). It also induces mitogenesis and neoplastic transformation (27,28). It was shown that important effectors in G␣ 13 -mediated downstream signaling are RhoA, Rac, and Cdc42, members of a Rho family of small GTPases (25,29). Activation of RhoA by G␣ 13 results in actin stress fiber formation and cell retraction (29). Rho is activated by G␣ 13 upon direct interaction of G␣ 13 with the Rho-specific guanine nucleotide exchange factor, p115 RhoGEF (30,31). Recent studies from our laboratory showed two additional proteins directly interacting with G␣ 13 . One is radixin, the member of ERM (ezrin/radixin/moesin) family of proteins, whose function is regulated by direct interaction with G␣ 13 (32). The other is AKAP110, as described above (22). Multiple receptors are reported to couple to G␣ 13 , including the thrombin PAR-1 receptor (33).
Here we show that both thrombin and G␣ 13 can activate PKA and induce VASP phosphorylation in HUVECs and HEK-293 cells. Moreover, we described a novel mechanism of G␣ 13 -induced cAMP-independent PKA activation and VASP phosphorylation that involves activation of RhoA and MEKK1. Activated RhoA and MEKK1, in turn, lead to phosphorylation and subsequent degradation of IB, thus releasing PKAc from the complex with IB and NF-B. PKAc then phosphorylates VASP. As VASP proteins are multifunctional organizers of actin cytoskeleton (34), this study opens the possibility that thrombin and G␣ 13 are involved in subtle regulation of actin cytoskeleton.  [2-(p-bromocinnamylamino)ethyl]-5-isoquinolinesulfonamidedihydrochloride (H-89), and eNOS inhibitor N-nitro-L-arginine methyl ester (L-NAME) were purchased from Sigma. Anti-G␣ 12 , -G␣ 13 , -HA, -RhoA, -ROCK-2, and -IB␣ antibodies were from Santa Cruz Biotechnology. RPMI 1640, glutamine, antibiotic-antimycotic, fetal bovine serum (FBS), Hanks' balanced salt solution (HBSS), and Lipofectamine-2000 reagent were from Invitrogen. Endothelial growth medium (EGM-2) BulletKit and EBM-2 basal medium were purchased from Cambrex. Superfect was purchased from Qiagen.

Reagents-Murine
Cell Culture and DNA Transfection-The human embryonic kidney 293 (HEK-293) cells (ATCC) were maintained in RPMI 1640 medium, supplemented with 2 mM glutamine, 100 units/ml penicillin G, 100 units/ml streptomycin sulfate, 25 g/ml amphotericin B, and 10% FBS. The human umbilical vein endothelial cells (HUVECs) were cultured for up to 8 passages in EGM-2 BulletKit. Transient transfections of HEK-293 cells and HUVECs were performed using Lipofectamine-2000 and SuperFect, respectively, according to the protocols from the man-ufacturers. The HEK-293 cells were serum-starved in 0.2% FBS for 16 h. HUVECs were maintained in 1% serum 1-2 h before the experiment.
Small Interfering RNA (siRNA)-RhoA-specific siRNA duplexes were purchased from Dharmacon. Nonsilencing control siRNA was from Qiagen. The siRNA transfection was performed using siRNA transfection reagent and siRNA transfection medium (Santa Cruz Biotechnology) according to the manufacturer's protocol. The final siRNA concentration was 60 nM.
Immunoblotting-The HEK-293 cells were lysed 24 -30 h after transfection. The quiescent HUVECs were lysed after stimulation with ␣-thrombin. Lysis buffer contained 25 mM HEPES (pH 7.5), 150 mM NaCl, 10% glycerol, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 5 mM EDTA, 10 mM NaF, 100 M Na 3 VO 4 , 1 mM dithiothreitol, and 5 l/ml mammalian protease inhibitor mixture (Sigma). The insoluble material was removed from the lysates by centrifugation at 15,000 ϫ g for 10 min, and cleared lysates were subjected to SDS-PAGE, transferred to polyvinylidene difluoride membrane, and analyzed by immunoblotting with appropriate antibodies. Phosphorylationdependent electrophoretic mobility shift of FLAG-tagged VASP was detected by immunoblotting with the anti-FLAG antibody, whereas the shift of endogenous VASP was detected by immunoblotting with the anti-VASP antibody. The percentage of phosphorylated VASP was calculated by using densitometry analysis of VASP bands detected with anti-VASP antibody from scanned Western blot images using the ImageJ program.
Luciferase Reporter Gene Assays-SRE-, B-, and CREB-dependent gene expression was determined by the SRE.L reporting system, B-luciferase reporting system, and CREB "PathDetect" trans-reporting system (Stratagene), respectively. Briefly, HEK-293 cells grown on 24-well plates at 90% confluency were transiently transfected with the following plasmids (per well): 50 ng of SRE.L, luciferase cDNA under the control of the serum-response factor-responsive element from the c-fos promoter (reporter plasmid), for assessment of SRE activation, 50 ng of B-driven luciferase reporter plasmid, for assessment of NF-B activation, or 75 ng of pFR-Luciferase (reporter plasmid) and 4 ng of pFA2-CREB (fusion trans-activator plasmid), for assessment of CREB activation, together with 50 ng of pCMV-␤-gal (transfection efficiency control plasmid) and 100 ng of empty vector or plasmid as indicated. Cells were starved for 16 h, washed twice with cold phosphate-buffered saline, and lysed with protein extraction reagent, and the cleared lysates were assayed for SRE-, B-, and CREB-dependent expression of firefly luciferase and ␤-galactosidase activity using corresponding assay kits (Promega). Luciferase activities were measured with a Sirius luminometer (Berthold Detection Systems). Luciferase activity of each sample was normalized to ␤-galactosidase activity to correct for the differences in the transfection efficiency and expressed as the folds increase over the control. The data represent mean Ϯ S.D. of triplicate determinations. The experiments were repeated two to three times with similar results.
Kinase Assay-Kinase activity of Myc-tagged-LIMK1 was determined in the COS-7 cells grown on the 60-mm plates and transfected with Myc-tagged LIMK1. Cells were harvested 48 h after transfection and lysed in the lysis buffer containing 50 mM Tris⅐HCl (pH 7.5), 100 mM NaCl, 5 mM EDTA, 40 mM Na 4 P 2 O 7 , 1% Triton X-100, 1 mM dithiothreitol, 200 M Na 3 VO 4 , and 5 l/ml mammalian protease inhibitor mixture (Sigma). The lysates were normalized for protein concentration, and Myc-tagged LIMK1 was immunoprecipitated from 100 l of cell lysates by incubation with anti-Myc antibody at 4°C for 2 h and protein A/G-agarose for 30 min. Immunoprecipitates were washed three times in lysis buffer and two times in kinase buffer containing 20 mM HEPES (pH 7.5), 10 mM MgCl 2 , 150 mM KCl, 10 mM MnCl 2 , 1 mM dithiothreitol, 50 M Na 3 VO 4 , and 5 l/ml mammalian protease inhibitor mix-ture. Thereafter, the immunoprecipitate was incubated with ATP buffer containing 20 M ATP, and 5 Ci of [␥-32 P]ATP, together with 4 g of purified recombinant glutathione S-transferase-cofilin for 20 min at 30°C. The reaction was terminated with SDS-PAGE sample buffer, followed by SDS-PAGE, autoradiography, and PhosphorImager analysis.
Cyclic AMP Assay-cAMP production was determined using the cAMP Biotrack enzyme immunoassay system according to the manufacturer's protocol (Amersham Biosciences). Briefly, cell lysates were incubated with antiserum, containing anti-cAMP antibody, followed by the incubation with cAMP-peroxidase conjugate, washing, and incubation with peroxidase substrate. The reaction was stopped by the addition of 1 M sulfuric acid, and the absorbance was determined at 450 nm. The results were determined from the standard curve and represent the mean Ϯ S.D. from triplicate determinations.
Immunostaining and Confocal Microscopy-HUVECs were transfected for 24 h where indicated and maintained in the EGM-2 medium supplemented with 10% FBS. Two hours before the stimulation with ␣-thrombin, the cells were maintained in EBM-2 basal medium supplemented with 1% FBS. Thereafter, the cells were washed with HBSS, fixed with 2% paraformaldehyde, and permeabilized with 0.1% Triton X-100. Nonspecific binding was blocked with blocking solution containing 0.2% fish skin gelatin (Sigma) and 1% bovine serum albumin in HBSS followed by incubation with appropriate primary and secondary antibody dissolved in the blocking solution. The coverslips were then mounted using ProLong antifade kit (Molecular Probes). Images were taken by laser-scanning confocal microscopy on a Zeiss LSM 510 micro-scope equipped with 63ϫ water-immersion objective and laser excitations at 488 and 543 nm. For quantitative analysis of the immunofluorescence data, HUVECs expressing FLAG-tagged VASP wild type and mutants were scored based on the localization of VASP in the cell. The percentage of the cells with VASP accumulated at the periphery was calculated from at least 200 FLAG-positive cells counted for each transfection.

␣-Thrombin Stimulation and G␣ 13 Q226L Expression Induce VASP Phosphorylation and Phosphorylation-dependent Translocation-We
tested the effect of thrombin stimulation on the phosphorylation of VASP by using HUVECs. We stimulated confluent HUVECs with 25 nM ␣-thrombin for different periods of time and subjected cleared cell extracts to SDS-PAGE and immunoblotting with anti-VASP antibody. ␣-Thrombin induced prolonged phosphorylation of VASP (Fig. 1A) that lasted for at least 4 h after stimulation.
Targeting Ena/VASP proteins to the membrane and focal adhesions is required for cell adhesion, motility, and formation of filopodia (35), indicating that VASP localization in the cell is critical for its function. Therefore, we next analyzed the distribution of VASP in the HUVECs stimulated with ␣-thrombin. We stimulated confluent and quiescent HUVECs with 25 nM ␣-thrombin for 5 min followed by fixation, permeabilization, and blocking of the nonspecific binding sites. Thereafter, VASP localization in the cells was detected by staining with polyclonal anti-VASP antibody, whereas F-actin was stained with phalloidin. Our HUVECs were serum-starved for 1 h followed by stimulation with 25 nM ␣-thrombin for the indicated time points. Thereafter, the cells were lysed, and cell lysates were subjected to SDS-PAGE and immunoblotting with anti-VASP antibody. B, ␣-thrombin induces translocation of endogenous VASP to the periphery of the cell. Confluent HUVECs, grown on the gelatin-coated coverslips, were stimulated with 25 nM ␣-thrombin for 5 min, fixed with 2% formaldehyde for 15 min, permeabilized with 0.1% Triton X-100 for 5 min, and blocked with blocking solution for 30 min. Subsequently, coverslips were incubated with rabbit anti-VASP antibody followed by anti-rabbit Alexa Fluor 488 (green fluorescence) antibody to detect the localization of endogenous VASP and phalloidin Alexa Fluor 594 (red fluorescence) to detect actin fibers. C, translocation of VASP upon thrombin stimulation is phosphorylationdependent. HUVECs grown on the gelatin-coated glass coverslips on 12-well plates were transfected with 400 ng of FLAG-tagged wild type VASP (WT-VASP), VASP-S153A, VASP-S235A, or phosphorylation-deficient VASP (PD-VASP) using SuperFect reagent according to the manufacturer's instructions. Twenty four hours after transfection, cells were serumstarved for an additional 1 h and stimulated with 25 nM ␣-thrombin for 5 min. Thereafter, the cells were treated as described above and stained with monoclonal anti-FLAG antibody followed by anti-mouse Alexa Fluor 488 antibodies and phalloidin Alexa Fluor 594. For each transfection, the percentage of the cells with FLAG-tagged VASP translocated to the periphery of the cell was calculated from 200 FLAG-positive cells. Data represent the mean Ϯ S.D. of three independent experiments. D, WT-VASP but not PD-VASP accumulates on the periphery of endothelial cells after ␣-thrombin stimulation. Images were taken by laser-scanning confocal microscopy on a Zeiss LSM 510 microscope equipped with 63ϫ water-immersion objective. studies showed that in HUVECs endogenous VASP is located at the tips of the actin fibers and along the actin fibers. After ␣-thrombin stimulation, the cells retracted, rounded up, and formed stress fibers. Most interestingly, in these cells VASP accumulated at the periphery of the cell (Fig. 1B). To test if VASP phosphorylation was required for the ␣-thrombin-induced translocation, we generated three FLAG-tagged VASP mutants where serine 153 and serine 235 were mutated into alanine, individually and jointly. We transfected HUVECs with FLAGtagged wild type VASP (WT-VASP), VASP-S153A, VASP-S235A, or VASP-S153A/S235A (PD-VASP) and stimulated either with HBSS or 25 nM ␣-thrombin for 5 min and fixed them. The transfected cells were detected by anti-FLAG antibody and scored based on the localization of FLAG-tagged VASP in the cell. We calculated the percentage of cells with VASP accumulated at the periphery from at least 200 FLAG-positive cells counted for each transfection. The percent of the cells with VASP accumulated at the periphery of the cell was ϳ52% in the case of WT-VASP, whereas it was 20, 23, and 14% for VASP-S153A, VASP-S235A, and PD-VASP (Fig. 1C), respectively. These results suggested that ␣-thrombin-induced phosphorylation of VASP on both serine 153 and serine 235 was required for VASP translocation. The representative images of localization of WT-VASP and PD-VASP in control and ␣-thrombin-stimulated cells are shown in Fig. 1D. Both FLAG-tagged WT-and PD-VASP had similar localization to that of endogenous VASP. However, after ␣-thrombin stimulation only WT-VASP trans-located to the periphery of the cells, and localization of PD-VASP was not affected by thrombin.
The thrombin receptor is coupled to members of the G i , G q , and G 12,13 families of proteins (reviewed in Ref. 36). To dissect G protein(s) that mediate ␣-thrombin-induced VASP phosphorylation, we used regulators of G protein signaling (RGS proteins). RGS proteins function as GTPase-activating proteins (GAPs) thus serving as negative regulators of G protein signaling (37). RGS domain of Rho-specific guanine nucleotide exchange factor p115RhoGEF was shown to have GAP activity specifically for G␣ 12 and G␣ 13 (30). RGS3 protein was shown to exhibits GAP activity for G␣ i and G␣ q (38), whereas RGS10 exhibits GAP activity for G␣ i (39). We tested if transfection of HUVECs with RGS domain of p115RhoGEF, RGS3, or RGS10 could affect ␣-thrombin-induced VASP phosphorylation. Our results showed that expression of the RGS domain of p115RhoGEF but not RGS3 and RGS10 abolished VASP phosphorylation induced by ␣-thrombin ( Fig. 2A), suggesting that it is likely that G␣ 12 and/or G␣ 13 are involved in this pathway. We next transiently transfected constitutively active mutants of G␣ 12 and G␣ 13 that lack GTPase activity, G␣ 12 Q229L and G␣ 13 Q226L, into HEK-293 cells along with FLAG-tagged VASP. Our result showed that G␣ 13 Q226L expression induced VASP phosphorylation of both serine 153, as it was detected by electrophoretic mobility shift of VASP, and serine 235, as it was detected by phosphospecific antibody (Fig. 2B). Most interestingly, expression of G␣ 12 Q229L did not induce VASP HUVECs grown on 12-well plates were transfected with 400 ng of empty vector, RGS domain of p115 RhoGEF, RGS3, or RGS10 using SuperFect reagent according to the manufacturer's protocol. Twenty four hours after transfection, cells were serum-starved for 1 h before stimulation with 25 nM thrombin for 5 min. Thereafter the cells were lysed, and cleared cell extracts were subjected to SDS-PAGE and immunoblotting with anti-VASP antibody. B, G␣ 13 but not G␣ 12 induces phosphorylation of VASP at both serine 153 and serine 235. HEK-293 cells, grown on 12-well plates, were co-transfected with 100 ng of FLAG-tagged VASP cDNA, and the indicated amounts of either empty vector, G␣ 13 Q226L, or G␣ 12 Q229L cDNA, serum-starved overnight and lysed. Cleared cell extracts were subjected to immunoblotting with anti-FLAG, anti-phospho-VASP(S239), anti-G␣ 12 , and anti-G␣ 13 antibodies. C, G␣ 12 Q229L and G␣ 13 Q226L induce SRE activation. Experiment was performed as described under "Materials and Methods." D, expression of G␣ 13 Q226L induces phosphorylationdependent translocation of VASP to the periphery of the cell. HUVECs were transfected with 400 ng of FLAG-tagged wild type VASP (WT-VASP), VASP-S153A, VASP-S235A, or phosphorylation-deficient VASP (PD-VASP) and 400 ng of empty vector or cDNA encoding for G␣ 13 Q226L. Twenty four hours after transfection, cells were serum-starved for an additional 1 h. Thereafter the cells were treated as described above and stained with monoclonal anti-FLAG antibody followed by anti-mouse Alexa Fluor 488 antibody. For each transfection, the percentage of the cells with FLAG-tagged VASP translocated to the periphery of the cell was calculated from 200 FLAG-positive cells. Data represent mean Ϯ S.D. of three independent experiments. E, representative images of WT-VASP and PD-VASP localization in the cells transfected with empty vector or G␣ 13 Q226L. Images were taken by laser-scanning confocal microscopy on a Zeiss LSM 510 microscope equipped with 63ϫ water-immersion objective. phosphorylation (Fig. 2B), indicating that this phenomenon is specific for G␣ 13 . It is known that G␣ 12 regulates gene expression by activation of distinct transcriptional control elements such as the SRE (40). Therefore, we confirmed functional activity of expressed G␣ 12 Q229L by its ability to activate SRE-dependent reporter gene transcription (Fig. 2C).
As we observed that thrombin stimulation induced phosphorylationdependent VASP accumulation at the periphery of the cell, we determined how G␣ 13 Q226L affected localization of WT-VASP, VASP-S153A, VASP-S235A, and PD-VASP in HUVECs. The cells transfected with wild type or mutants of VASP together with empty vector or G␣ 13 Q226L were analyzed as described above. The percentage of G␣ 13 Q226L-expressing cells with VASP accumulated at the periphery of the cell was ϳ70% for WT-VASP, whereas it was 30, 33, and 26% for VASP-S153A, VASP-S235A, and PD-VASP (Fig. 2D), respectively. Similarly to thrombin stimulation, these results suggested that G␣ 13 Q226Linduced phosphorylation of VASP on either serine 153 or serine 235 was required for VASP translocation. The representative images of localization of WT-VASP and PD-VASP in empty vector and G␣ 13 Q226Ltransfected cells are shown in Fig. 2E. In the control cells localization of FLAG-tagged WT-VASP and PD-VASP was similar to localization of endogenous VASP shown in Fig. 1B. In the cells transfected with G␣ 13 Q226L, WT-VASP but not PD-VASP was localized at the periphery of the cell. These results suggested that thrombin functions, at least in part, through G␣ 13 to induce VASP phosphorylation and phosphorylation-dependent translocation.
To address the direct involvement of RhoA in thrombin-mediated VASP phosphorylation, the effect of thrombin was examined in confluent endothelial cells with RhoA depleted by using siRNA. To establish the efficiency of siRNA in decreasing the level of RhoA, we first performed the titration (30 -120 nM) of siRNA in HUVECs. The protein level of RhoA was significantly decreased at 60 nM of siRNA (Fig. 3A). Depletion of RhoA did not affect the levels of ␤-actin or nonphosphorylated VASP (Fig. 3A).
We stimulated cells transfected with either nonsilencing control siRNA or RhoA-specific siRNA with ␣-thrombin for 5 min. More importantly, down-regulation of RhoA abolished the VASP phosphorylation induced by ␣-thrombin, suggesting that this phenomenon is mediated by RhoA (Fig. 3A).
Studies from our laboratory and other laboratories have shown that members of Rho family of small GTPases are involved in the downstream signaling initiated by G␣ 13 (25,29). To investigate if RhoA is involved in the G␣ 13 -VASP pathway, we tested if botulinum C3 exoenzyme, which ADP-ribosylates and specifically inactivates RhoA, can inhibit G␣ 13 Q226L-induced phosphorylation of VASP. Expression of C3 exoenzyme diminished VASP phosphorylation induced by G␣ 13 Q226L (Fig. 3B), which suggested that RhoA is downstream from G␣ 13 in the pathway connecting G␣ 13 and VASP. We next tested if activated forms of RhoA, Rac, and Cdc42 (RhoA V14, Rac V12, and Cdc42 V12) could induce VASP phosphorylation. Our result showed that RhoA V14 but not Rac V12 and Cdc42 V12 induced VASP phosphorylation, supporting the notion that RhoA may mediate G␣ 13 -induced VASP phosphorylation (Fig. 3C). To control for the functional activity of the expressed constructs, we determined that RhoA, Rac, and Cdc42 could induce stimulation of SRE-dependent gene transcription to a similar degree (42), suggesting that RhoA-dependent VASP phosphorylation was specific (Fig. 3D).
RhoA stimulates serine/threonine kinase Rho kinase (ROCK) (43), which in turn phosphorylates and activates another serine/threonine kinase, LIM kinase (LIMK) (44). Therefore, we tested if ROCK-2 or LIMK1 could induce VASP phosphorylation. Our experiments showed that neither ROCK-2 nor LIMK1 was capable of inducing VASP phosphorylation (Fig. 4, A and C). Furthermore, the specific inhibitor of ROCK, Y-27632, which competes with ATP for binding the catalytic site, did not affect G␣ 13 -induced VASP phosphorylation (Fig. 4A). We confirmed functional activity of ROCK-2 by its ability to induce SRE activation and functional activity of Y-27632 by its ability to abolish ROCK-2-induced SRE activation (Fig. 4B). Functional activity of LIMK1 is confirmed by its ability to phosphorylate cofilin in the in vitro kinase assay ( Fig. 4D). Together, these data suggested that neither ROCK-2 nor LIMK1 were involved in G␣ 13 -induced VASP phosphorylation.
␣-Thrombinand G␣ 13 -induced VASP Phosphorylation Is Mediated by PKA Activation-As VASP is a major substrate of PKA (45), we examined if VASP phosphorylation induced by ␣-thrombin, G␣ 13 , and RhoA is mediated by PKA. Initially, we used the PKA inhibitor H-89, which acts by binding to the ATP-binding site in the PKA catalytic site. Treatment with the H-89 inhibitor abolished ␣-thrombin-induced VASP phosphorylation in HUVEC in a dose-dependent manner, bringing VASP phosphorylation to the basal level at a concentration of 30 M (Fig. 5A). More importantly, H-89 also abolished G␣ 13 - (Fig. 5B) and RhoA-induced VASP phosphorylation in HEK-293 cells (Fig. 5C).
To corroborate the results obtained with H-89, we used the specific cell-permeable PKA inhibitor 14-22 amide (PKI), which acts as PKA pseudosubstrate, binds, and inactivates the catalytic subunit of PKA (46). Cells expressing G␣ 13 Q226L and FLAG-tagged VASP were incubated with different concentrations of PKI (Fig. 5B, lower panel). PKI inhibited VASP phosphorylation induced by expression of G␣ 13 Q226L in a dose-dependent manner. Together, these results suggested that ␣-thrombin-, G␣ 13 -, and RhoA-induced VASP phosphorylation was mediated by PKA.
To determine whether PKA was activated in the cells expressing the active mutants of G␣ 13 and RhoA, we performed cyclic AMP-response element-binding protein (CREB) assay using PathDetect system (Stratagene) as a readout of the PKA activation. Briefly, the system contains the trans-activator protein, pFA2-CREB, in which the activation domain of CREB transcriptional activator is fused with the yeast GAL4 DNA binding domain and the reporter plasmid, pFR-Luc, in which a synthetic promoter with GAL4-binding sites controls the expression of the firefly luciferase gene. Activated PKA phosphorylates CREB at serine residue 133 (47), thus activating the transcription of the luciferase gene from the reporter plasmid. Data showed that both G␣ 13 Q226L and RhoA V14 induced activation of CREB-dependent expression of firefly luciferase, which was abolished by H-89 (Fig. 5D), indicating that both G␣ 13 and RhoA can induce PKA activation. In contrast, expression of G␣ 12 Q229L did not activate CREB-dependent gene transcription (Fig. 5D), thus explaining the inability of G␣ 12 Q229L to induce VASP phosphorylation.
Because PKA is activated in response to cAMP-elevating agents as a result of cAMP binding to PKAr, leading to the release and activation of PKAc (18), we next tested if ␣-thrombin stimulation induces cAMP production in HUVEC. Stimulation of HUVEC with ␣-thrombin did not induce cAMP production, whereas forskolin, which served as a positive control, induced an ϳ15-fold elevation in intracellular cAMP (Fig. 5E, upper panel). Most interestingly, under our experimental conditions ␣-thrombin stimulation of HUVECs induced phosphorylation of ϳ28% of the total VASP protein; forskolin induced phosphorylation of ϳ55% of total VASP protein, whereas combined ␣-thrombin and forskolin stimulation induced phosphorylation of ϳ85% of the total VASP protein (Fig. 5E, lower panel), suggesting that they may use different mechanisms to induce VASP phosphorylation. Transiently transfected G␣ 13 Q226L and RhoA V14 in HEK-293 cells did not increase cAMP levels, whereas G␣ s R201C increased intracellular cAMP levels ϳ5-fold (Fig. 5F). These data indicated that ␣-thrombin-, G␣ 13 -, and RhoA-induced VASP phosphorylation is cAMP-independent.
It is well known that thrombin induces increase in intracellular Ca 2ϩ (for review see Ref. 36). Elevation of intracellular levels of Ca 2ϩ activates eNOS (NOS-III) (48) to produce NO and to activate the production of cGMP leading to activation of PKG (49). To test if eNOS activation is involved in ␣-thrombin-induced VASP phosphorylation, we used the eNOS inhibitor L-NAME, which acts as an L-arginine analog (50). We incubated quiescent HUVECs with L-NAME for 30 min prior to stimulation with ␣-thrombin. L-NAME did not abolish ␣-thrombin-induced VASP phosphorylation, suggesting that eNOS is not likely to be involved in this pathway (Fig. 5G). This finding was in accordance with the finding that PKG expression could not be detected in HUVECs (51).
To address the question if G␣ 13 -induced VASP phosphorylation was based on phosphorylation and degradation of IB, we employed phosphorylation-deficient dominant negative mutant of IB␣, IB␣-S32A/ S36A (IB␣m). Overexpression of IB␣m abolished G␣ 13 -induced VASP phosphorylation, indicating that IB phosphorylation is the underlying mechanism of VASP phosphorylation induced by expression of G␣ 13 Q226L (Fig. 6B). Furthermore, expression of IB␣m abol-ished RhoA-induced VASP phosphorylation (Fig. 6C), suggesting that G␣ 13 functions through RhoA to induce phosphorylation and degradation of IB␣, leading to VASP phosphorylation.
Degradation of IB leads to release of NF-B, which translocates to the nucleus and activates B-dependent gene transcription (53). To confirm that G␣ 13 and RhoA activation induced IB degradation, we performed B assay. Data showed that both G␣ 13 Q226L and RhoA V14 induced very pronounced NF-B activation in HEK-293 cells (Fig. 6D), which represents additional evidence that these proteins are involved in pathways leading to IB degradation. More importantly, G␣ 13 Q226Linduced NF-B activation was ϳ10-fold higher than that of G␣ 12 Q229L. This difference may serve as an additional explanation of the failure to induce VASP phosphorylation by G␣ 12 Q229L. Thereafter, the cells were lysed, and cleared extracts were subjected to immunoblotting with anti-VASP antibodies. B and C, HEK-293 cells were transfected with 100 ng of FLAG-tagged VASP and 100 ng of G␣ 13 Q226L (B) or RhoA V14-HA (C) and treated with of H-89 or PKA inhibitor 14-22 amide (PKI) in the concentrations as indicated. Cleared cell extracts were subjected to immunoblotting with anti-FLAG and anti-G␣ 13 (B) or anti-HA (C) antibodies. D, G␣ 13 Q226L and RhoA V14-HA but not G␣ 12 Q229L induce CREB activation that is inhibited by the PKA inhibitor H-89. The data of the CREB assay are expressed as luciferase activity over ␤-galactosidase activity and represent mean Ϯ S.D. of triplicate determinations. The experiment was performed as described under "Materials and Methods" and repeated three times with similar results. E, upper panel, ␣-thrombin does not affect cAMP level in HUVECs. HUVECs were stimulated with HBSS, 25 nM ␣-thrombin, or 1 M forskolin in HBSS for 15 min. Thereafter, the cells were lysed in the manufacturer-supplied lysis reagent and proceeded according to manufacturer's protocol. The experiment was performed in duplicate, and the results are represented as the mean fold increase Ϯ S.D. in cAMP production over the control. E lower panel, ␣-thrombin and forskolin have additive effects on VASP phosphorylation. Confluent HUVECs were stimulated for 15 min with 1 M forskolin prior to the addition of 25 nM ␣-thrombin for an additional 15 min. Cleared cell lysates were subjected to SDS-PAGE and immunoblotting with anti-VASP antibody. F, G␣ 13 Q226L and RhoAV14 do not affect the cAMP level in HEK-293 cells. HEK-293 cells were transiently transfected with 300 ng of empty vector, G␣ 13 Q226L, RhoA V14, or G␣ s R201C for 24 h. Thereafter, the cells were lysed in the manufacturer-supplied lysis reagent and proceeded according to the manufacturer's protocol. The experiment was performed in duplicate, and the results are represented as the mean fold increase Ϯ S.D. in cAMP production over the control. G, ␣-thrombin-induced VASP phosphorylation is not inhibited by L-NAME. Confluent HUVECs were preincubated with 0.5 or 1 mM L-NAME for 30 min as indicated followed by stimulation with 25 nM ␣-thrombin for 5 min. Thereafter, the cells were lysed, and cleared extracts were subjected to immunoblotting with anti-VASP antibodies. 13 but Not Rho Activation Is Mediated by MEKK1-We have reported previously (25) that G␣ 13 can activate MEKK1. As MEKK1 is shown to activate IB kinase (IK) (54) and phosphorylation and degradation of IB lead to release of PKAc (20), we next tested if activation of MEKK1 can induce VASP phosphorylation. Overexpression of the activated form of MEKK1 led to VASP phosphorylation (Fig. 7A), which was inhibited by co-expression of dominant negative IB␣ (Fig. 7B), indicating that MEKK1-induced VASP phosphorylation was mediated by phosphorylation and degradation of IB␣. More importantly, expression of the dominant negative mutant of MEKK1 (MEKK1-DN) abolished G␣ 13 -induced VASP phosphorylation in a dose-dependent manner (Fig. 7C). These data suggest that MEKK1 is involved in the G␣ 13 -VASP pathway. Most interestingly, dominant negative MEKK1 did not inhibit RhoA-induced VASP phosphorylation (Fig. 7D), and dominant negative RhoA did not affect MEKK1-induced VASP phosphorylation (Fig. 7E). Together these data indicated that RhoA may employ mechanisms other than through MEKK1 to induce VASP phosphorylation.

IB Degradation Induced by G␣
Based on these results we proposed the signaling pathway underlying G␣ 13 -induced VASP phosphorylation as shown in Fig. 8. G␣ 13 activates MEKK1 and RhoA via two independent pathways, which merge to activate IK leading to phosphorylation and degradation of IB and subsequent release and activation of PKAc leading to VASP phosphorylation.

DISCUSSION
In this study we present evidence that thrombin and the G␣ subunit of the heterotrimeric G 13 protein induce activation of PKA and phosphorylation of VASP, suggesting a novel pathway of cAMP-independent VASP phosphorylation.
Phosphorylation of VASP by ␣-Thrombin Stimulation and Active G␣ 13 -Our immunofluorescence studies showed that both ␣-thrombin stimulation and G␣ 13 Q226L expression in endothelial cells induced phosphorylation-dependent translocation of VASP to the periphery of the cells (Figs. 1 and 2). As VASP localization to the membrane and focal adhesions is required for cell adhesion, motility, and formation of filop-odia (35), it is likely that thrombin stimulation and G␣ 13 activation may modify VASP function by its phosphorylation and translocation. Additionally, these findings are interesting because G␣ 13 protein is known to induce cell retraction and actin stress fiber formation (29), whereas VASP phosphorylation may serve as a negative regulator of actin polymerization (55,56). We speculate that VASP phosphorylation induced by G␣ 13 activation can be an autoregulatory mechanism by which G␣ 13 limits its own responses. It is also possible that G␣ 13 has dual action on the actin cytoskeleton, and the balance between cell retracting and relaxing actions determines the overall result. Most interestingly, G␣ 13 is shown to activate radixin (32) that acts as barbed-end actin filament capping protein (57), whereas VASP is shown to act as an anti-capping protein that antagonizes the function of capping proteins (13), suggesting that G␣ 13 protein may be involved in a complex regulation of actin cytoskeleton architecture. Additionally, it is known that G␣ 13 protein induces Rho activation that in endothelial cells results in the actin stress fiber formation and cell retraction, thus enhancing the endothelial permeability (58). Because both PKA activation and VASP phosphorylation are implicated in maintaining the integrity of endothelial barrier (8,59,60), based on our data we hypothesize that signaling complexes regulated by G␣ 13 create the molecular basis for ligand-dependent loss as well as the restoration of endothelial barrier function. Currently we are testing these intriguing possibilities.
RhoA GTPase Is Involved in ␣-Thrombin and G␣ 13 -induced VASP Phosphorylation-As thrombin signaling is in many cases mediated by RhoA (41), we tested if RhoA mediates ␣-thrombin-induced VASP phosphorylation by employing RhoA-specific siRNA (Fig. 3). Additionally, as G␣ 13 can stimulate the members of the Rho family of small GTPases, Rho, Rac, and Cdc42, we tested if any of these GTPases could mediate G␣ 13 -induced VASP phosphorylation (Fig. 3). We concluded that RhoA mediates ␣-thrombinand G␣ 13 -induced VASP phosphorylation based on the following evidence: (i) RhoA-specific siRNA inhibited VASP phosphorylation induced by ␣-thrombin; (ii) C3 exoenzyme, the Rho-specific inhibitor, abolished G␣ 13 -induced VASP phosphorylation; and (iii) the activated form of RhoA, but not that of Rac and Cdc42, induced VASP phosphorylation. Most interestingly, two downstream effectors of RhoA, serine/threonine kinases ROCK-2 and LIMK1, are not involved in these RhoA effects as they failed to induce VASP phosphorylation under our experimental conditions (Fig. 4).
␣-Thrombin-, G␣ 13 -, and RhoA-induced VASP Phosphorylation Is Mediated by PKA Activation-G␣ 13 -induced VASP phosphorylation was inhibited by two different PKA inhibitors, the cell-permeable PKA inhibitor 14-22 amide (PKI) and H-89, suggesting that G␣ 13 -VASP pathway is mediated by PKA activation. Moreover, PKA appears to be the target kinase for the pathways leading from the activation of the thrombin receptor or RhoA to VASP phosphorylation as both ␣-thrombin-and RhoA-induced VASP phosphorylation were abolished by H-89 treatment (Fig. 5). These results are supported by the evidence that activated forms of both G␣ 13 and RhoA induce PKA activation (determined by CREB trans-reporter assay) that can be inhibited by H-89. More importantly, ␣-thrombin stimulation or expression of activated G␣ 13 and RhoA did not induce cAMP production over the basal level (Fig. 5), suggesting that ␣-thrombin-, G␣ 13 -, and RhoA-induced VASP phosphorylation is cAMP-independent. At this point we are not excluding the possibility that other tissues have different AKAPs that could release PKAc under certain conditions. Also, the involvement of other protein kinases such as PKG and PKC was not excluded.
␣-Thrombin-, G␣ 13 -, and RhoA-induced VASP Phosphorylation Is Mediated by IB Phosphorylation and Degradation-Certain pools of PKAc, instead of binding the PKAr, bind to IB to form the complex NF-B-IB-PKAc. Association between PKA and NF-B-IB blocks the catalytic activity of PKA. More importantly, this pool of PKAc is not sensitive to changes in intracellular cAMP levels (20). Upon treatment with lipopolysaccharide (20) or ET-1 (19), IB␣ is phosphorylated, ubiquitinated, and degraded by the proteasome pathway (52), leading to the release of NF-B and PKAc from the inhibition by IB. Here we identify G␣ 13 as a possible target G protein that may mediate ligandinduced cAMP-independent PKA activation. Using phosphorylationdeficient IB␣ mutant in order to prevent IB␣ degradation, we demonstrated that both G␣ 13 and RhoA employ the NF-B signaling pathway to activate PKA and phosphorylate VASP. These findings are supported by the evidence that expression of both G␣ 13 Q226L and RhoA V14 in HEK-293 cells induced NF-B-dependent gene transcription (Fig. 6).
We reported that G␣ 13 could activate PKA by direct interaction with AKAP110 (22). However, as AKAP110 distribution is restricted to cerebellum and spermatozoids, the mechanism of G␣ 13 -iduced PKA activation reported here could be a general mechanism employed by many different cell types. ␣-Thrombin-induced VASP phosphorylation was inhibited by the proteasome inhibitor MG-132, suggesting that thrombin-induced VASP phosphorylation is mediated by proteasomedependent protein degradation, presumably IB degradation (Fig. 6).   13 -induced VASP phosphorylation. G␣ 13 activates MEKK1 and RhoA via two independent pathways, which induce phosphorylation and degradation of IB␣, presumably through activation of IB kinase (IK), leading to release and activation of PKA catalytic subunit leading to VASP phosphorylation.
More importantly, IB␣ is retranscribed after its degradation, serving as an autoregulatory feedback loop (62), which could mean that IB␣ degradation-dependent PKA activation is expected to be transient. This makes prolonged VASP phosphorylation in response to ␣-thrombin stimulation even more complex, involving either inhibition of phosphatases that dephosphorylate VASP or additional pathways of PKA or activation of other kinases.
IB Degradation Induced by G␣ 13 but Not RhoA Activation Is Mediated by MEKK1-MEKK1, which belongs to the family of mitogenactivated protein kinase kinase kinases, is shown to activate c-Jun NH 2terminal kinase and p38 signaling cascades and to induce apoptosis (63,64). It is also shown that MEKK1 induces NF-B activation through phosphorylation and degradation of IB␣ (54,65,66), identifying MEKK1 as the common signal transduction component for NF-B activation and mitogen-activated protein kinase cascades activation. MEKK1 was shown before to be involved in G␣ 13 -mediated regulation of cellular responses (24). Our data represent additional evidence that MEKK1 is involved in G␣ 13 signaling, in this case in G␣ 13 -induced VASP phosphorylation (Fig. 7). However, MEKK1 is not involved in RhoA-induced VASP phosphorylation (Fig. 7), which was in accordance with the report that MEKK1 is not involved in RhoA-induced NF-B activation (67). Additionally, we determined that RhoA is not downstream from MEKK1 in MEKK1-induced VASP phosphorylation by failure of dominant negative Rho to inhibit MEKK1-induced VASP phosphorylation. These findings indicated that there are two different pathways mediating G␣ 13 -induced VASP phosphorylation, RhoA-and MEKK1-mediated pathways, merging presumably at the IB kinase that phosphorylates IB␣ targeting it to proteasome-dependent degradation. Most interestingly, blocking only one pathway, either RhoA with C3 toxin (Fig. 3) or MEKK1 with DN-MEKK1 (Fig. 7), almost completely blocks the phosphorylation of VASP induced by expression of activated form of G␣ 13 . This apparent contradiction can be reconciled if we postulate that these two pathways operate in concert and that both have to be intact for full response. "Coincidence detectors are ubiquitous in biology . . . These proteins integrated convergent signaling pathways by responding in a characteristic fashion to distinct but coincident signals" (73). One of the best understood examples includes stimulation of adenylyl cyclase II by G␤␥ subunits, which is strictly conditional requiring simultaneously two input signals G␤␥ and G␣ s (61).
In summary, we have shown that stimulation with ␣-thrombin and activation of G␣ 13 induced VASP phosphorylation. Moreover, we propose the mechanism underlying G␣ 13 -induced VASP phosphorylation (Fig. 8). It involves activation of RhoA and MEKK1 via two different pathways that ultimately lead to IB␣ degradation and release and activation of PKAc resulting in VASP phosphorylation.