Vascular Endothelial Growth Factor (VEGF) Receptor-2 Tyrosine 1175 Signaling Controls VEGF-induced von Willebrand Factor Release from Endothelial Cells via Phospholipase C-γ1- and Protein Kinase A-dependent Pathways*

There is increasing evidence that vascular endothelial growth factor (VEGF) contributes to inflammation independent of its angiogenic functions. Targeting some of the components in endothelial Weibel-Palade bodies (WPBs) effectively inhibits VEGF-induced inflammation, but little is known about how VEGF regulates WPB exocytosis. In this study, we showed that VEGF receptor-2 (VEGFR2), but not VEGFR1, is responsible for VEGF-induced release of von Willebrand factor (vWF), a major marker of WPBs. This is in good contrast to VEGF-stimulated interleukin-6 release from endothelium, which is selectively mediated through VEGFR1. We further demonstrated that VEGFR2-initiated phospholipase C-γ1 (PLCγ1)/calcium signaling is important but insufficient for full vWF release, suggesting the possible participation of another effector pathway. We found that cAMP/protein kinase A (PKA) signaling is required for full vWF release. Importantly, a single mutation of Tyr1175 in the C terminus of VEGFR2, a tyrosine residue crucial for embryonic vasculogenesis, abolished vWF release, concomitant with defective activations of both PLCγ1 and PKA. These data suggest that Tyr1175 mediates both PLCγ1-dependent and PKA-dependent signaling pathways. Taken together, our results not only reveal a novel Tyr1175-mediated signaling pathway but also highlight a potentially new therapeutic target for the management of vascular inflammation.

There is increasing evidence that vascular endothelial growth factor (VEGF) contributes to inflammation independent of its angiogenic functions. Targeting some of the components in endothelial Weibel-Palade bodies (WPBs) effectively inhibits VEGF-induced inflammation, but little is known about how VEGF regulates WPB exocytosis. In this study, we showed that VEGF receptor-2 (VEGFR2), but not VEGFR1, is responsible for VEGF-induced release of von Willebrand factor (vWF), a major marker of WPBs. This is in good contrast to VEGF-stimulated interleukin-6 release from endothelium, which is selectively mediated through VEGFR1. We further demonstrated that VEGFR2-initiated phospholipase C-␥1 (PLC␥1)/calcium signaling is important but insufficient for full vWF release, suggesting the possible participation of another effector pathway. We found that cAMP/protein kinase A (PKA) signaling is required for full vWF release. Importantly, a single mutation of Tyr 1175 in the C terminus of VEGFR2, a tyrosine residue crucial for embryonic vasculogenesis, abolished vWF release, concomitant with defective activations of both PLC␥1 and PKA. These data suggest that Tyr 1175 mediates both PLC␥1-dependent and PKA-dependent signaling pathways. Taken together, our results not only reveal a novel Tyr 1175 -mediated signaling pathway but also highlight a potentially new therapeutic target for the management of vascular inflammation.
In addition to its role in promoting angiogenesis, there is increasing evidence that VEGF contributes to inflammation independent of its angiogenic functions, although the molecular basis for this effect is incompletely understood (6 -8). VEGF is well expressed in the chronic inflammatory skin disease, psoriasis, and in synovial fluid in rheumatoid arthritis (9 -12). In addition, previous studies found an association between human severe sepsis/septic shock with elevated circulating levels of VEGF and PlGF (13,14). Using an in vitro monocyte migration assay and in vivo mouse models of arthritis, several groups, including ours, have suggested that one mechanism by which VEGF causes inflammation is by modulating the infiltration and secretion of monocytes/macrophages via the activation of VEGFR1 (11,12,15). On the other hand, emerging evidence suggests that endothelial activation is also important for VEGFinduced inflammation (6,8,9). In a mouse model of sepsis, it was demonstrated that the inhibition of VEGFR2, but not VEGFR1, attenuates sepsis mortality, possibly at least in part by suppressing vascular inflammation associated with endothelial activation (9). Consistent with this, ectopic VEGF-A expression in mice enhances leukocyte rolling and adhesion in venules mediated through the P-selectin on the surface of endothelial cells (6). These studies indicate that endothelial activation is another mechanism for VEGF-induced inflammation. P-selectin and von Willebrand factor (vWF) are the best characterized constituents of Weibel-Palade bodies (WPBs), endothelial storage granules that also contain various inflammatory mediators (16 -18). As a major component in WPBs, vWF is also involved in their biogenesis and thus is used as a marker of WPBs (18,19). WPB exocytosis, which gives rise to rapid release of vWF and other mediators such as interleukin-8 (IL-8) (17), and translocation of P-selectin from within granules to the endothelial surfaces triggering leukocyte rolling, are critical early events in endothelial activation and vascular inflammation (16). It has been reported that VEGF regulates vWF/ WPB release (20), but the precise roles of VEGF receptors and their downstream effectors in this process have not been defined. In this study, we sought to dissect the signaling pathway by which VEGF induces vWF/WPB release.

EXPERIMENTAL PROCEDURES
Additional procedures are described in the supplemental Materials and Methods.
RNA Interference-To silence phospholipase C-␥1 (PLC␥1) and protein kinase A (PKA) (catalytic ␣ and ␤ subunits), we used a lentiviral system (Sigma) for delivering short hairpin RNAs (shRNAs) into HUVECs. The target sequences (see supplemental Materials and Methods) and control scrambled sequences were selected according to an open program on the official website of Massachusetts Institute of Technology. Preparation of lentiviruses and infection of HUVECs were carried out based on a published method (21).
Measurement of vWF and IL-8 by Standard Enzyme-linked Immunosorbent Assays (ELISAs)-HUVECs or HAECs were grown until confluent on 48-well plates and serum-starved for 5 h before stimulation with agonists. The amount of vWF released into the medium was determined from American Diagnostica (Greenwich, CT) or from the Jiangsu Institute of Hematology, Soochow University, Jiangsu, China (22). The amount of IL-8 was measured by an ELISA from R&D Systems (Minneapolis, MN).

Construction of Chimeric Fusion Receptors and Mutants-
The chimeric VEGF receptors used in this study comprised the intracellular and transmembrane domains of VEGFR1 (EGLT) or VEGFR2 (EGDR) fused to the extracellular domain of the human EGF receptor (23). Construction of tyrosine-to-phenylalanine mutants of Y1175F and Y1214F was as described before (24). EGDR-Y951F was generated by PCR and cloned into the pMMP retroviral vector. See supplemental Materials and Methods for detailed primer sequences.
Measurement of Intracellular Free Ca 2ϩ and cAMP-Intracellular free Ca 2ϩ was measured by fluo-4 AM with laser scanning confocal microscopy (Leica, Heidelberg, Germany). Endothelial cells were plated onto collagen-coated coverslips and incubated for 20 min at 37°C with fluo-4 AM in medium 199 containing 2% bovine serum albumin. For pharmacological studies, cells were pretreated for 30 min with 20 M 1,2-bis-(2-aminophenoxy)-ethane-N,N,NЈ,NЈ-tetraacetic acid-acetoxymethyl ester (BAPTA-AM) before stimulation with 50 ng/ml VEGF, PlGF, or VEGF-E. Similar experiments on EGF stimulation were applied to HUVECs, which were transduced with EGLT, EGDR, and its mutants. The concentration of cAMP in HUVECs was measured according to the manufacturer's instructions (Amersham Biosciences).
Statistical Analyses-Group differences were analyzed by a standard Student's t test or by Dunnett's test (using GraphPad Prism software, otherwise mentioned in the legends of Figs. 3 and 4). All values are expressed as mean Ϯ S.D. Statistical significance was set at a p value less than 0.05 (*) or 0.01 (**).

VEGFR2, but Not VEGFR1, Mediates VEGF-induced vWF
Release-A previous study showed that VEGF stimulates WPB exocytosis from HAECs, as evaluated by vWF release as well as P-selectin translocation (20). To extend this finding, we studied the action of VEGF on vWF release from HUVECs by using a standard vWF ELISA. VEGF rapidly induced a time-and dosedependent vWF release (Fig. 1, A and B), indicating that VEGF promotes vWF secretion from human venous endothelial cells as well. Interestingly, VEGF was somewhat more powerful than thrombin on a molar basis because stimulation with 1 nM VEGF (50 ng/ml) led to 13 milliunits/ml vWF release, whereas stimulation with 38 nM thrombin (1 unit/ml) led to 7.5 milliunits/ml vWF release (Fig. 1B). A recent study has shown that histamineinduced WPBs can undergo a form of exocytosis during which vWF is retained, whereas smaller molecules, such as IL-8, are released (17). Therefore, we also examined IL-8 release from HUVECs in response to VEGF stimulation. VEGF-triggered IL-8 secretion followed a time course similar to that of vWF release (Fig. 1F), suggesting that during VEGF-induced exocytosis, the proportion of WPBs fused with plasma membrane (so-called "lingering kiss") might be too low to be estimated by our experimental system. This is consistent with the recent study in which the histamine-induced lingering kiss fusion events appeared to be 10ϳ25% (17).
A previous report showed that VEGF induced IL-6 selectively mediated through VEGFR1 but not VEGFR2 (24). To study which receptor mediates the VEGF effect on vWF release, we first examined the expression of VEGF receptors. Both VEGFR1 and VEGFR2 are expressed in HUVECs and HAECs (Fig. 1C). We then used PlGF, a VEGFR1-specific ligand, and VEGF-E, a VEGFR2-specific ligand. Treatment of HUVECs with PlGF resulted in no vWF release, although PlGF was biologically active (supplemental Fig. S1B). In contrast, treatment with VEGF-E rapidly led to vWF release comparable with that of VEGF treatment, indicating that the activation of VEGFR2 alone is sufficient to induce vWF release (Fig. 1, D and E). Similar results were obtained with HAECs (supplemental Fig. S1A). These data suggest that VEGF stimulates vWF release mainly through VEGFR2. VEGF-induced IL-8 release was also mediated via VEGFR2 (Fig. 1F).
To validate further the distinctive roles of VEGFR1 and VEGFR2 in vWF secretion, we employed a chimeric receptor approach in which the extracellular domain of the epidermal growth factor receptor (EGFR) was substituted for that of VEGFR1 (EGFR-Flt-1, designated EGLT) or VEGFR2 (EGFR-KDR, designated EGDR), and both were expressed in HUVECs by using a retrovirus system (23). This system allows independent activation of either receptor by using EGF as an agonist to stimulate the kinase activity of Flt or KDR without activating the endogenous VEGF receptors. As expected, HUVECs transduced with LacZ-expressing viruses did not respond to EGF stimulation. In response to EGF treatment, HUVECs transduced with EGDR rapidly released vWF in a pattern similar to HUVECs responding to VEGF stimulation (Fig. 1, F and G). In contrast, no response was found in HUVECs transduced with EGLT, although it was expressed at a level similar to that of EGDR and became weakly phosphorylated after EGF stimulation ( Fig. 1G and supplemental Fig. S2), which is consistent with the nature of VEGFR1 as a weak tyrosine kinase (1,2). These results confirmed that VEGFR2 plays a predominant role in VEGF-induced vWF secretion.
VEGFR2-triggered PLC␥1/Calcium Signaling Is Important but Insufficient for Full VEGF-induced vWF Release-Intracellular calcium (Ca 2ϩ ) increase triggers vWF/WPB exocytosis (16,17). Previously, several groups, including ours, found that VEGFR2 activates the PLC␥1/ Ca 2ϩ signaling pathway (23,25). Therefore, we speculated that differential activation of the PLC␥1/ Ca 2ϩ pathway by VEGFR1 and VEGFR2 may account for their different roles in the induction of vWF release.
The abilities of the two VEGF receptors to activate PLC␥1 and Ca 2ϩ mobilization were further compared by approaches using receptor-specific ligands and the chimeric receptors. As shown in Fig.  2, A-D, both approaches confirmed that the activation of VEGFR2, but not VEGFR1, led to PLC␥1 phosphorylation and elevated the level of intracellular Ca 2ϩ in HUVECs. Similar results were obtained with HAECs (supplemental Fig. S3).
A previous study using U73211 (20), a potent PLC␥1 inhibitor, showed that PLC␥1-activated Ca 2ϩ signaling is important for VEGFtriggered vWF/WPB exocytosis. Because U73122 may affect intracellular Ca 2ϩ levels independent of PLC inhibition (26 -28), we employed PLC␥1-specific shRNAs to inactivate its function to avoid nonspecific effects of U73211. PLC␥1-specific shRNAs were constructed and transduced into HUVECs by using a highly efficient lentiviral system (21). shRNA1 was identified as a suitable construct, which reproducibly reduced PLC␥1 expression with an efficiency of ϳ90% ( Fig.  2E and supplemental Fig. S4). The specificity of this construct was confirmed by the observations that it neither reduced PLC␥2 expression (Fig. 2E and supplemental Fig. S4) nor interfered with phorbol 12-myristate 13-acetate-induced vWF release (Fig. 2G), which is known to be independent of Ca 2ϩ signaling (29). Although the expression of a scrambled RNA had no effect on intracellular Ca 2ϩ mobilization, the expression of shRNA1 completely blocked the VEGF-triggered Ca 2ϩ increase at the phases of both initial peak and sustained plateau (Fig. 2F). This result indicates that VEGF-induced intracellular Ca 2ϩ mobilization is critically dependent on PLC␥1. We then assessed the effect of PLC␥1 down-regulation on vWF release.  AUGUST 28, 2009 • VOLUME 284 • NUMBER 35 VEGF-induced vWF release was decreased significantly (70% reduction) in shRNA1-expressing HUVECs when compared with that in noninfected cells or scrambled RNA-expressing cells (Fig. 2G), indicating that PLC␥1 is essential for VEGFinduced vWF release. The results indicating that inhibition of VEGFR2-triggered Ca 2ϩ signaling by down-regulating PLC␥1 is unable to block VEGF-induced vWF release completely also strongly suggest the possible participation of another effector pathway.

VEGFR2/Tyr 1175 Signaling Controls vWF Release
VEGFR2-triggered PKA Activation Is Required for Full VEGF-induced vWF Release-cAMP/PKA signaling is another distinct pathway for the regulation of vWF/WPB exocytosis from endothelial storage granules in response to cAMP-raising agonists (18,19). So far, however, PKA is known only to negatively regulate VEGF functions, such as the inhibition of VEGF-induced endothelial migration (30,31). To assess the role of the cAMP/ PKA pathway in VEGF-induced vWF release, we first examined the effect of VEGF on the intracellular cAMP level. VEGF stimulation of HUVECs elicited a small, but significant, rise in the intracellular cAMP level as measured by a standard ELISA (Fig. 3A), consistent with a previous report (31). To investigate whether intracellular cAMP increase mediates VEGF-induced vWF release, we incubated HUVECs with VEGF in the presence of cAMP analog Rp-8CPT-cAMPS (Rp). Incubation with Rp significantly reduced vWF release (Fig. 3A), indicating an involvement of cAMP increase in VEGF-induced vWF release. The increase in cAMP levels commonly leads to the activation of PKA, which subsequently results in the phosphorylation of target proteins. The activation of PKA, as estimated by measuring the phosphorylation level of the cAMP-response element-binding protein (CREB), a PKA substrate (32), was increased upon VEGF stimulation (Fig. 3B). To explore the role of PKA in VEGF-induced vWF release, two potent PKA inhibitors (H89 and myristoylated protein kinase inhibitor (14 -22)) were used at concentrations that efficiently inhibit the phosphorylation of CREB. Pretreatment of HUVECs slightly, but significantly, inhibited vWF release (20 -30% reduction) (Fig. 3C). These data strongly suggest that the cAMP/ PKA signaling pathway also plays a role in VEGF-induced vWF release, which is in striking contrast to its negative role in the regulation of endothelial migration (31,33). To explore the relationship between the PLC␥1 and PKA pathways, PLC␥1 shRNA-expressing HUVECs were pretreated with two PKA inhibitors before VEGF stimulation. PKA inactivation by either inhibitor signifi-

VEGFR2/Tyr 1175 Signaling Controls vWF Release
cantly suppressed the residual vWF release (about 25% reduction) (Fig. 3D), indicating that the PKA pathway acts in parallel with the PLC␥1/Ca 2ϩ pathway. To validate further the role of PKA in VEGF-induced vWF release, specific shRNAs targeting the catalytic ␣ and ␤ subunits of PKA were screened, and two efficient shRNAs (shRNA C␣1 and shRNA C␤1) were identified (Fig. 3E and supplemental Fig. S5). Knockdown of either the ␣ subunit or the ␤ subunit of PKA significantly attenuated VEGF-induced vWF release (Fig. 3F), suggesting that PKA is required for vWF release. To confirm that the VEGF-activated PLC␥1 and PKA pathways are distinct, the effect of PLC␥1 shRNAs on the activation of PKA and the effect of PKA shRNAs on the activation of PLC␥1 were examined. Although PKA shRNAs did not show any inhibitory effect on VEGF-induced PLC␥1 phosphorylation, PLC␥1 shRNAs slightly but insignificantly interfered with VEGF-induced activation of PKA (Fig. 3, H and I). Taken together, the results show that VEGF-induced vWF release requires both PLC␥1/Ca 2ϩdependent and PKA-dependent signaling pathways. Similarly, PLC␥1 and PKA pathways are also required for VEGF-induced IL-8 release (Fig. 3G).
Identification of Tyr 1175 in VEGFR2 as a Molecular Switch to Turn on vWF Release through PLC␥1and PKA-dependent Signaling Pathways-So far, Tyr 951 , Tyr 1175 , and Tyr 1214 in VEGFR2, the most prominent noncatalytic autophosphorylation sites (4,34), have been linked to the activation of PLC␥1 (23, 34 -36). Therefore, we investigated the roles of these residues in PLC␥1 activation and vWF release by point mutagenesis. We first showed that the various VEGFR2 mutants (Y951F, Y1175F, and Y1214F) were expressed at similar levels in HUVECs (Fig. 4A), and then we compared their capacities to stimulate PLC␥1 phosphorylation. Interestingly, a single mutation Y1175F abolished EGF-induced PLC␥1 phosphorylation (Fig. 4B). Consistently, the Y1175F mutation abolished EGF-induced Ca 2ϩ release and entry (Fig. 4C). In contrast, mutant receptors with Y951F and Y1214F retained their capacities to activate PLC␥1 (Fig. 4B), whereas mutations Y1214F and Y951F did not or only slightly affected Ca 2ϩ mobilization (Fig. 4C). We further examined the abilities of the mutants to stimulate vWF release. Interestingly, the Y1175F mutant completely lost its ability to induce vWF release, whereas Y951F and 1214F mutants were still able to activate as effectively as the wild type receptor (Fig. 4D), indicating that Tyr 1175 is essential for VEGF-induced vWF release. This result also suggests that the Y1175F mutant may fail to activate the cAMP/PKA pathway. Consistent with this, the Y1175F, but not other mutants, had a defective activation of PKA in response to EGF stimulation, which was accompanied by lower cAMP levels than wild type (Fig. 4, E and F). These data indicate that Tyr 1175 plays an essential role in vWF release through mediating both PLC␥1/Ca 2ϩ -dependent and cAMP/PKA-dependent signaling pathways.

DISCUSSION
vWF/WPB exocytosis is one of the earliest events in endothelial activation and vascular inflammation (16,19). Targeting some components of WPBs, such as P-selectin, reduced VEGF-induced inflammation (6), suggesting that VEGF stimulation of vWF/WPB exocytosis may be an important mechanism by which VEGF contributes to inflammation. Recently, potent VEGF-capturing reagents such as antibodies against VEGF or inhibitors of VEGF receptor kinase have proven to be very effective in animal inflammation models (8,9). Because VEGF receptors are critical not only for endothelial cell function but also for cellular processes such as neuronal survival (37), inhibition of specific VEGFR-induced signaling pathways may be preferable in long term treatments, especially for chronic inflammation. In this study, we clearly demonstrate that VEGF is a potent agonist that stimulates vWF/ WPB release from endothelial cells, and this stimulation is mediated by VEGFR2 through both PLC␥1dependent (in large part) and PKAdependent (to a lesser extent) dependent signaling pathways (Fig. 5). This is in contrast to the role that VEGR1 plays in inflammation by promoting the migration and secretion of monocytes/macrophages in which only VEGFR1 is expressed (11,14). In addition, this is also in good contrast to VEGFstimulated IL-6 release from endothelium, which is selectively mediated through VEGFR1 (24). We show, to our knowledge, for the first time that the cAMP/ PKA signaling pathway plays a positive role in the regulation of VEGF function (Fig. 5), in a manner opposite to its inhibitory effect on endothelial migration and tube formation (30,31). However, this is not unexpected because PKA is a critical effector for vWF/WPB exocytosis from endothelial storage granules in response to cAMP-raising agonists (18). Interestingly, unlike cAMP-raising agonists, VEGF disrupts endothelial barrier function (38), which may also contribute to inflamma-tion. When compared with the vWF release triggered by PLC␥1/Ca 2ϩ signaling, which occurs within 2-5 min after VEGF stimulation, the release induced by cAMP/PKA signaling occurs about 5 min or more after VEGF stimulation and is relatively mild. 3 The use of both signaling pathways by VEGFR2 may ensure rapid and continuous vWF/WPB release in response to VEGF stimulation.
In previous studies, in addition to Tyr 1175 , the residues Tyr 951 and Tyr 1214 were also linked to PLC␥1 activation by VEGFR2 (25, 34 -36). Our data showed that receptor with the Tyr 1175 mutant was completely incapable of activating PLC␥1 phosphorylation and intracellular Ca 2ϩ mobilization, whereas receptors with the Y951F or Y1214F mutation retained these functions (Fig. 4). The discrepancies concerning the residues responsible for PLC␥1 activation found in this study from those in the literature may be due to the different experimental systems. For example, one study on the interaction between Tyr 951 and PLC␥1 was performed in a yeast two-hybrid system (35). It has been shown that the data obtained with the yeast system are sometimes inconsistent with protein-protein interaction events in mammalian cells (39,40). Our work compared the roles of all three of these residues in VEGFR2 in the activation of PLC␥1/Ca 2ϩ signaling in primary human endothelial cells. In a recent report, an adaptor molecule TSAd was suggested to be a more important Tyr 951 -binding protein than PLC␥1 in the regulation of endothelial cell migration (31), consistent with our finding.
Tyr 1175 of VEGFR2 is a critical residue for the regulation of vascular functions. A recent study from the Shibuya group (41) showed that exchange of Tyr 1173 (corresponding to Tyr 1175 in VEGFR2/KDR) for Phe in mouse VEGFR2/Flk-1 resulted in a loss-of-function phenotype and embryonic lethality, indicating that signaling via Tyr 1175 is essential for VEGFR2 functions during mouse vasculogenesis. Our study suggests a new role that Tyr 1175 plays in the induction of vWF/WPB exocytosis from endothelial cells within adult vessels. In addition, Tyr 1175 mediates the PLC␥1/MAPK (mitogen-activated protein kinase) signaling pathway by which VEGF stimulates endothelial DNA synthesis (25) and also mediates cell migration through phosphatidylinositol 3-kinase mediated by the adaptor proteins Shb and ShcB (1-3). Therefore, Tyr 1175 is a unique residue that mediates multiple functions via several signaling pathways in both embryonic development and adult pathophysiological conditions.
The identification of cAMP/PKA signaling as a novel Tyr 1175 -mediated pathway raises new questions. For example, how does Tyr 1175 regulate the activation of the cAMP/PKA signaling pathway? Previously, several groups, including ours, found a cross-talk between G protein-coupled receptors and VEGFR2 (42). Therefore, it is possible that Tyr 1175 may transmit a signal from VEGFR2 to G protein-coupled receptors through adaptor proteins (such as Shb/ShcB) and in turn transactivate adenylate cyclase to induce cAMP/PKA signaling (Fig.  5). It is equally possible that VEGFR2 may activate PKA directly via a cAMP-independent mechanism (43).
In summary, our study demonstrates a key role of Tyr 1175 in VEGFR2 in vWF/WPB exocytosis from endothelial cells in response to VEGF stimulation and indicates its usefulness as a potential target in the future development of anti-inflammatory strategies.
Addendum-During the preparation of this manuscript, a report described the role of PKC␦ in VEGF-induced vWF secretion (44). Because PKC functions downstream of PLC␥1, it appears that PLC␥1-dependent vWF release from endothelial cells in response to VEGF is mediated through both Ca 2ϩ signaling and PKC signaling pathways.