Outside-in Signaling Pathway Linked to CD146 Engagement in Human Endothelial Cells*

CD146 (S-Endo 1 Ag or MUC18) is a transmembrane glycoprotein expressed on endothelial cells on the whole vascular tree. CD146 is located at the intercellular junction where it plays a role in the cohesion of the endothelial monolayer. CD146 engagement initiates an out-side-in signaling pathway involving the protein tyrosine kinases FYN and FAK as well as paxillin. Here we report that CD146 engagement by its specific monoclonal antibody in human umbilical vein endothelial cells induces a Ca 2 1 influx that is sensitive to thapsigargin and EGTA treatment, indicating that CD146 engagement initiates a store-operated calcium mobilization. In addition, bio-chemical and pharmacological analysis revealed that CD146 engagement initiates the tyrosine phosphorylation of phospholipase C- g , Pyk2, and p130 Cas . Pharmacological inhibition of Ca 2 1 flux with 1,2-bis( o -amino-phenoxy)ethane- N,N,N * ,N * -tetraacetic acetoxymethyl ester and EGTA indicated that an increase in Ca 2 1 is required for Pyk2 and p130 Cas tyrosine phosphorylation.

CD146 (S-Endo 1 Ag), also referred to as Mel-CAM or MUC18, is an integral membrane protein present on endothelial cells, regardless of their anatomical site and the size of the vessels (1-3). CD146 expression is not restricted to the endothelium but is found on other non-malignant and malignant cell types such as melanoma cells (4). In these cells, the cell surface expression of CD146 is associated with the metastatic properties of primary tumors (5,6). CD146 belongs to the Ig superfamily of cell adhesion molecules with a V-V-C2-C2-C2 structure in the extracellular portion, a single membrane-spanning domain, and a relatively short cytoplasmic tail containing 61 amino acids (7).
CD146 is involved in cell-cell adhesion through a heterophilic ligand that still remains unknown (8). In melanoma cells, the homotypic interaction of CD146 and its ligand contributes to cohesive interactions among these cells (9). Similarly, the binding of CD146 to its putative receptor on trophoblasts confers a stationary phenotype so preventing trophoblastic migration/invasion within the myometrium (10). Immunohistochemical studies have revealed that CD146 is localized at the intercellular junction in endothelial cells (3). This localization is consistent with a role of CD146 in the control of cohesive cell-cell interactions. 1 It is well known that cell adhesion molecules located at the intercellular junction control the integrity of the endothelial monolayer (11). They promote adhesion through their extracellular domain, whereas the intracytoplasmic tail is implicated in the outside-in signaling pathway that is derived from their engagement (12)(13)(14)(15). In endothelial cells, CD146 acts as a signal transduction molecule. CD146 initiates an outside-in signal cascade upon monoclonal antibody engagement. Whereas CD146 is not phosphorylated on tyrosine residues, its engagement promotes the recruitment of the Src family kinase p59 fyn (FYN) 2 as well as the tyrosine phosphorylation of a large panel of intracellular proteins including p125 FAK (FAK) and paxillin, two proteins present in focal adhesion plaques (16).
Calcium is a central second messenger that mediates a large number of cellular processes such as cell division, gene transcription, and/or cell death (17,18). Activation of multiple cell surface receptors linked to PTK activation leads to increases in intracellular calcium concentrations ([Ca 2ϩ ] i ) (19). A feature of PTK-induced increase in [Ca 2ϩ ] i involves a two-step process characterized at first by a rapid, transient release of Ca 2ϩ stored in the endoplasmic reticulum (20). This release of Ca 2ϩ from intracellular stores occurs, at least in part, via activation of phospholipase C (21,22). PLC␥ activation results in increased production of inositol 1,4,5-triphosphate and emptying of inositol 1,4,5-triphosphate receptor-gated Ca 2ϩ stores. Depletion of intracellular Ca 2ϩ stores then induces sustained extracellular calcium influx via store-operated calcium entry (SOCE) (23). Numerous endothelial cell functions are regulated by elevation in [Ca 2ϩ ] i levels such as the secretion of endothe-lial cell granules, regulation of endothelial permeability, adhesion, and transmigration of circulating cells as well as rearrangement of actin cytoskeleton (24 -28).
We show here that engagement of CD146 in HUVEC triggers a store-dependent Ca 2ϩ influx that requires the tyrosine phosphorylation (Tyr(P)) of FYN and PLC␥. In addition, CD146 engagement induces the Tyr(P) of the PTK Pyk2 (29 -31), and the adaptor protein p130 Cas (32,33), by a process involving Ca 2ϩ mobilization. Thus, these data confirm and expand the concept that besides its function as adhesive protein, CD146 is also a signaling molecule involved in the dynamics of actin cytoskeleton rearrangement.

Methods
Cell Culture-HUVECs were isolated from umbilical cord veins according to the method of Jaffe et al. (34). They were used at subconfluency after one passage. Cell monolayers were starved for 3 h in serumfree RPMI containing 0.5% BSA. For drug treatment experiments, HUVECs were pretreated with the drug for the indicated time prior to CD146 engagement.
CD146 Engagement-CD146 clustering was performed as described previously (16). Briefly, quiescent HUVECs were incubated with 10 g/ml anti-CD146 F(abЈ) 2 mAb in HBSS 30 min at 4°C; after washing, cross-linking was then performed with 20 g/ml goat anti-mouse IgG F(abЈ) 2 (GAM), and the cells were processed for calcium determinations. Control cells were incubated with isotype-matched IgG1 and crosslinked with GAM.
Inhibitor Treatments-Herbimycin, PP1, and U73122 were dissolved in Me 2 SO and incubated with HUVECs in serum-free HBSS, at the required concentration, 30 min prior to engagement of CD146. When required, HUVECs were preincubated with BAPTA-AM for 30 min at 37°C and washed twice in HBSS, and engagement of CD146 was performed as indicated above. To study the effect of EGTA on the tyrosine phosphorylation of signaling proteins, EGTA (5 mM) was added throughout the addition of GAM.
Measurements of [Ca 2ϩ ] i -Serum-starved subconfluent HUVECs grown in 96-well plates were loaded with Fluo3-AM (5 M) and 0.2 mg/ml Pluronic F-127 by incubation in loading buffer (HBSS supplemented with 10 mM HEPES, pH 7.4, 2 mM CaCl 2 , 1 mM MgCl 2 , 1% BSA) (35) for 30 min at 37°C. Cells were then washed in loading buffer and incubated for at least 15 min at room temperature, washed, and incubated again in loading buffer. CD146 engagement was then performed using anti-CD146 mAb at 4°C as described (16). Cells were then washed and incubated in Ca 2ϩ -free loading buffer to which 2 mM CaCl 2 was added when required, and cross-linking with GAM was performed as described above. Intracellular Ca 2ϩ levels were determined immediately upon addition of GAM to the Fluo3-loaded cell monolayers at 37°C using a plaque-reader spectrofluorimeter (Cytofluor series 4000 Perspective Biosystems). Levels of fluorescence from cells in individual wells were monitored at an excitation wavelength of 485 nm and an emission wavelength of 530 nm. The signal was calibrated by addition of 10 M ionomycin containing Ca 2ϩ (2 mM) to obtain maximal fluorescence (R max ). After a stable fluorescence at 530 nM (R max ), the medium was changed to one containing 5 mM EGTA in calcium-free medium with 10 M ionomycin (R min ). [Ca 2ϩ ] i concentration was calculated using the method of Grynkiewicz et al. (36) using the equation: (37). In some experiments, HUVECs were pretreated at 37°C before cross-linking for 5 min with 1 M thapsigargin for measurement of calcium release from intracellular stores. The effect of extracellular calcium was analyzed in presence of 5 mM EGTA in the loading buffer. Calcium influx was then assayed as required. All pharmacological and monoclonal antibody treatments did not alter the cell viability assessed both by treatment with 10 mM ionomycin to trigger calcium influx and by trypan blue dye exclusion demonstrating that the cells were still viable.
Immunoprecipitation and Immunoblotting-Immunoprecipitation and immunoblotting were performed as described previously (16). In brief, cell lysates (300 -600 g of total proteins/sample, i.e. 0.3 to 1 ϫ 10 7 cells/ sample) were immunoprecipitated with 4 g of mAb for 3 h at 4°C under constant agitation, followed by a 2-h incubation with protein G-Sepharose beads. After washing, the recovered proteins were subjected to SDS-PAGE and transferred to nitrocellulose Cϩ membranes. After blocking, membranes were incubated in TBS-T 5% BSA containing indicated antibodies (1 g/ml) for 1 h at room temperature. Immunoreactive bands were visualized by chemiluminescence using horseradish peroxidase-conjugated anti-mouse IgG and ECL reagent. When required, membranes were stripped and reblotted with the indicated antibodies.
Statistical Analysis-Results are expressed as mean Ϯ S.E. Calcium measurements were analyzed by the unpaired Student's t test. Differences were considered significant with p Ͻ 0.05.

CD146 Engagement-induced [Ca 2ϩ ] i Increase in HU-
VECs-To determine whether the engagement of CD146 in HUVECs triggers an increase in [Ca 2ϩ ] i , cells loaded with 5 M Fluo3-AM were pretreated with anti-CD146 mAb F(abЈ) 2 fragment (10 g/ml) at 4°C for 20 min and equilibrated at 37°C for 180 s, and [Ca 2ϩ ] i was measured after cross-linking with GAM F(abЈ) 2 antibody (20 g/ml). In the presence of 2 mM extracellular calcium, the engagement of CD146 resulted in an increase in [Ca 2ϩ ] i characterized by an initial peak followed by a phase showing a slow decrease (Fig. 1 Fig. 2A, n ϭ 4). These results indicated that the initial rise of [Ca 2ϩ ] i in response to CD146 engagement was dependent on intracellular store depletion.
Cross-linking of CD146 was then performed in the presence or absence of EGTA (5 mM). EGTA had no effect on the initial transient increase but strikingly decreased the [Ca 2ϩ ] i in the second phase from 832.5 Ϯ 106.3 to 332.9 Ϯ 45.8 nM (p Ͻ 0.002, n ϭ 4) (Fig. 2B). The role of extracellular calcium was confirmed in a subsequent experiment where CD146 cross-linking was performed in the absence of extracellular calcium. As shown in Fig. 2C, only the first transient rise in [Ca 2ϩ ] i was observed. After the peak has returned close to the basal level, subsequent addition of 2 mM Ca 2ϩ to the extracellular medium restored the second slow phase of [Ca 2ϩ ] i (Fig. 2C). Taken together, these data indicate that the second phase of slow decrease was on the dependence of extracellular calcium sources. These results suggest that CD146 engagement stimu-lates Ca 2ϩ influx into the cells that depends both on Ca 2ϩ release from internal stores and Ca 2ϩ entry from the extracellular milieu.

Involvement of Protein Tyrosine Kinases (PTK) in Calcium Flux
Induced by CD146 Engagement-For receptors coupled to PTK, calcium mobilization requires the activation of PTK and the recruitment of PLC␥ to the plasma membrane (20). Previous data have indicated that upon engagement, CD146 recruits the PTK FYN (16). To define the roles of FYN in the mobilization of Ca 2ϩ linked to CD146 engagement, cell lysates were immunoprecipitated using anti-FYN mAb, and the immunoprecipitates were analyzed by immunoblotting using anti-Tyr(P) mAb (PY20). As shown in Fig. 3 d and e). These results suggest that the Tyr(P) of PLC␥ induced by CD146 engagement is mediated by FYN activation. However, no direct association between CD146 and PLC␥ was initiated by CD146 engagement (data not shown). The involvement of FYN and PLC␥ in [Ca 2ϩ ] i mediated by CD146 engagement was confirmed by using pharmacological inhibitors. Pretreatment of HUVECs for 30 min with 2 M herbimycin (Fig. 5A) or 10 M PP1 (Fig. 5B) (Fig. 5C). Taken together, these results are consistent with a scheme according to which CD146 engagement induces the recruitment and activation of FYN, leading to the Tyr(P) of PLC␥, which in turn participates to the increase in [Ca 2ϩ ] i .
Calcium-dependent Recruitment of Pyk2 and p130 Cas by CD146 Engagement-CD146 engagement leads to the Tyr(P) of FAK, a PTK present in focal adhesion (15). PYK2, another PTK   5 mM, lane e). This phosphorylation was observed despite the absence of a molecular association be-tween CD146 and Pyk2 (data not shown). These data indicate that Pyk2 is phosphorylated on tyrosine residues following CD146 engagement by a process requiring both the activation of an Src family PTK and calcium mobilization from intra-and extracellular stores.
The adapter protein p130 Cas localizes to focal adhesion points and interacts with Pyk2 (32,33). The role of Pyk2 as a transducing molecule toward proteins of the focal adhesion plaques was then investigated in response to CD146 engagement. Similar to experiments with Pyk2, CD146 cross-linking with GAM for 15 min induced the Tyr(P) of a band with a molecular mass of Ϸ130 kDa corresponding to p130 Cas (Fig. 6B,  upper panel, lane b). No phosphorylation was observed in HU-VECs treated with a control isotype-matched (IgG1) mAb (lane a). The Tyr(P) of p130 Cas was abrogated by pretreatment of HUVECs with PP1 (10 M, lane c) and was reduced by BAPTA-AM (25 M) or EGTA (5 mM) (lanes d and e). Nevertheless, CD146 did not associate with p130 Cas (data not shown).
Pyk2 has been shown to associate with paxillin (42). Paxillin is tyrosine-phosphorylated upon CD146 engagement (16). Therefore, we investigated whether paxillin and also p130 Cas associate with Pyk2 in HUVECs upon CD146 triggering. As shown in Fig. 7, anti-Pyk2 immunoprecipitates performed on anti-CD146-stimulated HUVEC lysates contained a 130-kDa band reactive with anti-p130 Cas mAb as well as a 70-kDa band reactive with anti-paxillin mAb (lanes b). Of note, constitutive Time Course of Tyr(P) of FYN, PLC␥, Pyk2, and p130 Cas -The time course of the Tyr(P) of FYN, PLC␥, Pyk2, and p130 Cas upon CD146 engagement was then investigated. Cell lysates were immunoprecipitated with anti-Tyr(P) mAb and immunoblotted with the indicated mAbs. CD146 engagement rapidly stimulates the Tyr(P) of FYN and PLC␥ (Fig. 8) which reached their maximum 2 and 5 min, respectively, after GAM crosslinking. It should be noted that Tyr(P) of Fyn was detected as soon as 15 s upon CD146 engagement (data not shown). Maximal Tyr(P) of Pyk2 and p130 Cas were observed after 20 min. The Tyr(P) of Pyk2 was transient, whereas that of p130 Cas remained sustained after 30 min. These results indicate that CD146 engagement induces an outside-in signal pathway involving at first FYN and PLC␥ followed by the phosphorylation of Pyk2 and p130 Cas . DISCUSSION Previous data have indicated that CD146 initiates an outside-in signal transduction pathway that involves the recruitment of the Src PTK FYN and leads to the Tyr(P) of FAK and paxillin, two proteins present in the focal adhesion plaques (16). We demonstrate here that in endothelial cells, CD146 engagement promotes an increase in [Ca 2ϩ ] i both by Ca 2ϩ release from TG-sensitive Ca 2ϩ stores and entry of extracellular Ca 2ϩ . This process requires the activation of FYN and PLC␥. Moreover, Ca 2ϩ appears to serve as a second messenger by coupling Ca 2ϩ release to the Tyr(P) of the related FAK kinase, Pyk, and the adapter protein, p130 Cas , and favoring their association.
The Ca 2ϩ influx initiated by CD146 engagement in HUVECs is representative of the well known SOCE. Indeed, the initial rapid Ca 2ϩ mobilization depends upon TG-sensitive stores, whereas the long lasting decrease depends on the extracellular Ca 2ϩ . This SOCE is mediated by activation of FYN and PLC␥ (20), a key enzyme involved in calcium traffic (19). CD146 recruitment of FYN initiates the Tyr(P) of PLC␥, as evidenced by the inhibition of PLC␥ Tyr(P) by PP1, and is consistent with the time course. These data of FYN and PLC␥ phosphorylation are in agreement with previous reports indicating that FYN is involved in calcium mobilization by initiating the activation of PLC␥ (44 -46).
It should be noted that the mechanism leading to SOCE is not well elucidated. Some recent data suggest that Ca 2ϩ influx results from a secretory pathway induced by a close interaction between ER calcium stores and the plasma membrane via an involvement of the actin cytoskeleton (47,48). The secretion requires intracellular fusion events mediated by SNAP-25, a membrane protein belonging to the SNAP receptor proteins (49).
The mobilization of calcium ions in addition to the recruitment of FYN play an active role in the outside-in signaling pathway mediated by CD146. Indeed, CD146 engagement leads to Tyr(P) of Pyk2, the calcium-dependent tyrosine kinase related to FAK (reviewed in Ref. 41), by a calcium-dependent mechanism. FYN is also involved in the activation of Pyk2. It is known that several Pyk2 tyrosine residues (Tyr-402, Tyr-579, and Tyr-580 within the catalytic domain and Tyr-881 within the carboxyl terminus domain) create binding sites for the SH2 of the Src-like PTK (50,51). Pyk2 also associates with and is phosphorylated by FYN during stimulation of T cell antigen receptor (52). Nevertheless, in the case of CD146 engagement, FYN exerts an indirect effect on Pyk2 because no direct association between Fyn and Pyk2 could be found (data not shown). Moreover, Pyk2 Tyr(P) occurs more later than that of FYN, indicating that Pyk2 acts downstream of FYN.
The signaling pathway initiated by CD146 engagement includes the Tyr(P) of FAK and its substrate paxillin (16). p130 Cas belongs to the group of proteins that associate with FAK and is involved in cell migration (33). p130 Cas tyrosine phosphorylation in response to CD146 engagement strengthens the relationship between CD146 and focal adhesion points by a way that is dependent on FYN recruitment. Indeed, p130 Cas Tyr(P) depends on an Src PTK activation as demonstrated by PP1 inhibition and on calcium influx as evidenced by its partial inhibition by the calcium chelators. Src PTKs have been implicated in the Tyr(P) of p130 Cas mediated by integrins or shear stress (53,54). Nevertheless Pyk2 does not associate with p130 Cas in anti-CD146-treated HUVECs. The molecular structures implicated in the association of Pyk2 and p130 Cas upon CD146 engagement are not known, but the SH3 domain of p130 Cas and the two proline-rich sequences of Pyk2 have been involved in the formation of such a complex (55). p130 Cas is a ligand of FAK by its SH3 domain (56) and increases cell migration promoted by FAK (57). Taken together, the fact that p130 Cas plays an important role in the organization of the cytoskeletal framework (58,59) and that Pyk2 forms a constitutive complex with paxillin (42, 60) expands the concept that CD146 is in relation with the actin cytoskeleton. These results are consistent with data indicating that a part of CD146 is recovered in the Triton X-100-insoluble fraction and that CD146 colocalizes with F-actin (1). Nevertheless a direct effect of CD146 engagement on the actin reorganization has never been observed (data not shown).
In summary, the present study thus demonstrates that engagement of CD146 induces a complex signaling pathway that includes Ca 2ϩ influx as a second messenger involved in cytoskeleton dynamics. FYN plays a crucial function in the initiation of CD146 signal transduction. Upon engagement, FYN is recruited and phosphorylated by CD146. FYN allows the activation of PLC␥ which in turns can hydrolyze membrane phosphoinositides. The binding of inositol 1,4,5-triphosphate to its receptor would induce the release of Ca 2ϩ from internal stores and initiate a store-dependent entry of extracellular Ca 2ϩ . In addition, FYN regulates the Tyr(P) of targets of CD146 localized in the focal adhesion plaques by a Ca 2ϩ -dependent mechanism. FYN initiates the phosphorylation of Pyk2 which in turns associates with p130 Cas and paxillin and will allow their localization in focal adhesion plaques.
CD146 engagement leads to the Tyr(P) of FAK and paxillin and their association on the one hand (16) and Tyr(P) of Pyk2 and its association with paxillin on the other hand. It is known that FAK and Pyk2 are differentially regulated by cell adhesion and by soluble factors or develop distinct signal transduction events. Whereas FAK mainly localizes to focal contact sites, Pyk2 exhibits a punctate perinuclear distribution in FAK Ϫ/Ϫ embryo fibroblasts. Only a small fraction of Pyk2 is found in focal adhesion points (43,61,62). We can hypothesize that upon CD146 engagement, distinct signaling pathways might be developed. The cellular context might dictate which nonreceptor PTK is activated.
The localization of CD146 at the intercellular junction suggests that CD146 might mediate cell-cell and cell-extracellular matrix interactions that initiate an outside-in signaling cascade after its engagement with its ligand. Further studies are needed to determine the in vivo physiological relevance of the signaling pathways linked to CD146.