INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

S-Endo-1 antigen (CD146) is an integral membrane protein present on human endothelial cells. It is identical to MUC18/MCAM, an antigen found on melanoma cells. CD146 is ubiquitously present on the endothelium along the whole vascular tree and is highly expressed on human umbilical vein endothelial cells (HUVECs)¹ (1, 2). CD146 expression is not restricted to the endothelium; it is also detected in nonmalignant and malignant cells from other tissues (3-7).

CD146 (113-119 kDa) is a highly glycosylated monomer that belongs to the immunoglobulin (Ig) superfamily of cell adhesion molecules (1, 8, 9). CD146 contains five extracellular Ig-like homology domains (V-V-C2-C2-C2), one transmembrane segment, and a short cytoplasmic tail (10). CD146 molecular cloning has revealed significant homology to other Ig superfamily adhesion molecules such as B-CAM, ALCAM, BEN/DM-GRASP/SC1, KG-CAM, the chicken HEMCAM, and gicerin (11-16).

CD146 function is still not elucidated, although recent data suggest its involvement in cell-cell adhesion. Indeed, the expression of CD146 on melanoma cells correlates with an invasive phenotype, indicating that CD146 may be used as a cell surface marker of tumor progression and metastasis formation (17-19). Indeed, CD146 has been shown to be involved in tumor-endothelial cell interactions that might lead to extravasation of tumor cells (20, 21). Finally, it has been reported that CD146 mediates a Ca²⁺-independent homotypic melanoma cell adhesion by promoting heterophilic interaction through a still unknown ligand (19, 21).

CD146 is located at the interendothelial junctions. It is found in cytoskeletal protein-rich fractions and colocalizes with -actinin (22).² It is well documented that proteins present at the interendothelial junctions promote adhesion through their extracellular domain and mediate intracellular signaling to the complex network of cytoskeletal proteins through their intracytoplasmic part (23). CD146 possesses potential recognition sites for protein kinases in its cytoplasmic tail (9) which might be involved in signal transduction. Nevertheless, CD146 involvement in outside-in signaling in HUVECs has not been investigated.

In this present report, we investigated the signaling pathways initiated by CD146 engagement. Because the ligand of CD146 is unknown, CD146 clustering was performed using anti-CD146 monoclonal antibody (mAb) on HUVECs. Results indicate that cross-linking of CD146 induces tyrosine phosphorylation of a large panel of proteins, among which p125^FAK and paxillin associate with each other. However, CD146 is not directly bound to these proteins but becomes associated with p59^fyn. Thus, our results indicate that CD146 engagement may initiate a reorganization of the cytoskeleton through a p59^fyn- and p125^FAK-dependent pathway.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Antibodies-- The following mAbs and polyclonal antibodies were used: S-Endo-1 (F(ab')₂ fragment) and 7A4 mAbs, both specific for CD146 and kindly given by Biocytex (Marseilles, France); anti-Tyr(P) 4G10 (Upstate Biotechnology, Lake Placid, NY) and PY20 (Transduction Laboratories, Lexington, KY) mAbs; anti-p125^FAK, anti-paxillin, anti-p53/56^lyn, p59^fyn mAbs (Transduction Laboratories); rabbit anti-p72^syk (kindly given by Dr. R. Geahlen, IN); isotype-matched IgG1 or IgG2a (Sigma, St. Louis, MO); horseradish peroxidase-labeled goat anti-rabbit, goat anti-mouse (GAMIg) and its F(ab')₂ fragment (Jackson laboratories, Palo Alto, CA).

Cell Culture-- HUVECs were isolated from cord blood (24). They were used at confluence after one passage. They were rendered quiescent by incubation in RPMI 1640 medium containing 1% fetal calf serum 2 h before cell activation.

Cell Activation and Inhibitor Treatment-- Quiescent HUVECs were incubated in Hanks' balanced salt solution for 30 min at 4 °C with either isotype-matched IgG1 (control cells) or 10 µg/ml S-Endo-1 F(ab')₂ mAb for CD146 clustering; when indicated, cells were further stimulated with 20 µg/ml F(ab')₂ GAMIg for 30 min at 37 °C. After three washes at 4 °C, the cells were lysed for immunoblotting.

Immunoprecipitation-- After activation, the cells were lysed at 4 °C in 500 µl of lysis buffer (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, 2 mM sodium orthovanadate, 50 mM sodium fluoride, 1 mM phenylmethylsulfonyl fluoride, 25 µg/ml aprotinin, 2 µg/ml leupeptin, 2 µg/ml pepstatin). After adjustment to a protein concentration of 300-500 µg, i.e. 3-6 × 10⁶ cells (bicinchoninic acid assay, Pierce, Rockford, IL), cell lysates were precleared by incubation with irrelevant IgG1 or IgG2a mAbs and protein G-Sepharose (Pharmacia Biotech, Uppsala, Sweden) for 1 h at 4 °C. Precleared samples were immunoprecipitated with 2 µg for 3 h at 4 °C followed by a 2-h incubation with protein G-Sepharose. Immunoprecipitates were analyzed by SDS-PAGE and immunoblotting. Immunoprecipitations using anti-CD146 and anti-Tyr(P) were performed, respectively, with S-Endo-1 and 4G10 mAbs.

Immunoblotting-- Immunoprecipitates or total cell lysates were subjected to SDS-PAGE and transferred to nitrocellulose C⁺ filters. After blocking in 10 mM Tris-HCl buffer, pH 7.5, 0.15 M NaCl, 0.1% Tween 20, 3% bovine serum albumin (TBS-T BSA), they were incubated in TBS-T BSA containing protein-specific antibodies (1 µg/ml) for 1 h at room temperature. Immunoreactive bands were visualized by chemiluminescence using horseradish peroxidase-conjugated anti-mouse or rabbit IgG and ECL reagent. Western blotting using anti-CD146 was performed with 7A4 mAb. When required, membranes were stripped in 62.5 mM Tris-HCl, pH 6.8, 2 mM EDTA, 2% SDS, 100 mM -mercaptoethanol for 30 min at 60 °C and reblotted with the indicated antibodies.

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

CD146 Engagement Induces a Pattern of Tyrosine-phosphorylated Proteins in HUVECs-- Engagement of CD146 was obtained by incubating HUVECs with S-Endo-1 F(ab')₂ mAb and subsequently by cross-linking the complexes with GAMIg. Whole cell lysates were analyzed on a 5-15% gradient SDS-PAGE followed by immunoblotting with anti-Tyr(P) mAb. In HUVECs, engagement of CD146 induced the tyrosine phosphorylation of a large panel of proteins (Fig. 1, lane 2), which was increased further upon cross-linking with GAMIg at 37 °C for 30 min (Fig. 1, lane 3). The complex pattern of tyrosine-phosphorylated proteins consistently includes proteins at apparent molecular masses of 46, 55-60, 70-80, 100, 125, and 150 kDa. A 2-h pretreatment of HUVEC with 50 µM genistein (Fig. 1, lane 4) or 2 µM herbimycin (Fig. 1, lane 5) greatly reduced the tyrosine phosphorylation induced by CD146 engagement. Kinetic analysis of the induction of tyrosine phosphorylation upon CD146 engagement revealed a rapid onset after 5 min, a maximum at 30 min, followed by a decrease thereafter (data not shown). In contrast, in HUVECs treated with a control isotype-matched (IgG1) antibody cross-linked with GAMIg, a pattern of constitutive tyrosine phosphorylation was barely detectable (Fig. 1, lane 1).

View larger version (68K): [in this window] [in a new window]   Fig. 1.   Induction of tyrosine phosphorylation by CD146 engagement in HUVECs. HUVECs were treated by S-Endo-1 (F(ab')2) mAb fragment alone or cross-linked with GAMIg as described under "Experimental Procedures." Genistein (50 µM) or herbimycin (2 µM) was added 2 h before cell activation. After cell lysing, 50 µg of proteins was analyzed on a linear 5-15% gradient SDS-PAGE. Immunoblotting was performed with PY20 anti-Tyr(P). The position of prestained standards and their molecular masses in kDa are indicated. C indicates IgG1 isotype-matched control cells. Results are representative of five independent experiments.

CD146 Clustering Did Not Induce CD146 Phosphorylation-- The presence of a tyrosine residue at position 641 (Tyr-Ile-Asp-Leu) in CD146 cytoplasmic tail suggests a potential site of phosphorylation of the molecule (9). The capacity of CD146 engagement to phosphorylate Tyr⁶⁴¹ was studied. As shown in Fig. 2 (upper panel), anti-Tyr(P) immunoblotting of anti-CD146 immunoprecipitation revealed that CD146 is not tyrosine-phosphorylated, either in isotype-matched control (lane 1) or in CD146-stimulated cells (lane 2). Reciprocal experiments, using anti-CD146 immunoblotting of anti-Tyr(P) immunoprecipitation, led to similar results (Fig. 2, upper panel, lane 4). As a control, immunoprecipitation and immunoblotting with anti-CD146 mAbs revealed the presence of CD146 in these samples (Fig. 2, lower panel).

View larger version (30K): [in this window] [in a new window]   Fig. 2.   CD146 is not a tyrosine-phosphorylated protein. HUVECs were either activated by CD146 engagement or incubated with 0.1 mM NaV, as described under "Experimental Procedures." Lysates (300 µg) were immunoprecipitated (IP) using S-Endo-1 or 4G10 mAbs, separated on 7% SDS-PAGE, and immunoblotted (WB) using 7A4 or PY20 mAbs as indicated (upper panel). Blots were then stripped and reblotted with 7A4 mAb (lower panel). C indicates IgG1 isotype-matched control cells.

CD146 Engagement Induces Tyrosine Phosphorylation of p125^FAK and Paxillin-- The colocalization of CD146 with cytoskeletal proteins² and the tyrosine phosphorylation of a band in the molecular weight range of 125,000 suggested that p125^FAK could be tyrosine-phosphorylated upon CD146 engagement (26). To examine whether CD146 engagement stimulates tyrosine phosphorylation of p125^FAK in HUVECs, cell lysates were immunoprecipitated with anti-p125^FAK mAb and immunoblotted with anti-Tyr(P) mAb. Fig. 3A shows that a constitutive tyrosine phosphorylation of p125^FAK occurs in isotype-matched control cells (lane 1). CD146 engagement caused a marked increase in the tyrosine phosphorylation of p125^FAK (Fig. 3A, lane 2). Disruption of the actin network by a 2-h pretreatment of the HUVEC monolayer with 5 µM cytochalasin D inhibited the p125^FAK tyrosine phosphorylation mediated by CD146 clustering (Fig. 3A, lane 3).

View larger version (37K): [in this window] [in a new window]   Fig. 3.   CD146 engagement stimulates p125FAK tyrosine phosphorylation and paxillin association to p125FAK in HUVECs. CD146 engagement was performed as described under "Experimental Procedures." Lysates (300 µg) were immunoprecipitated, separated on 7% SDS-PAGE, and immunoblotted as indicated. C indicates IgG1 isotype-matched control cells. Panel A, tyrosine phosphorylation of p125FAK induced by CD146 engagement in absence or after pretreatment with 5 µM cytochalasin D. Panel B, lack of association of p125FAK with CD146. Lysates were immunoprecipitated using S-Endo-1 and immunoblotted with anti-FAK mAbs (upper panel). Blots were then stripped and reblotted with 7A4 mAb (lower panel). Panel C, tyrosine phosphorylation of paxillin and its association with p125FAK.

CD 146 Clustering Results in the Association of p59^fyn with CD146-- Because CD146 engagement induces the tyrosine phosphorylation of p125^FAK and paxillin, we investigated whether CD146 engagement results in the recruitment of PTK(s) in the vicinity of CD146. In this regard, p125^FAK possesses several binding sites for Src or Src-related kinases (26). In addition, CD146 engagement induces tyrosine phosphorylation of proteins with molecular masses (55-80 kDa) close to those of Src-related kinases such as p53/56^lyn and p59^fyn, as well as other PTKs such as p72^syk, all known to bind p125^FAK. We next determine whether these PTKs are involved in CD146 signal pathway. No association between p53/56^lyn or p72^syk with CD146 was detected (Fig. 4, upper panel), although reprobing with anti-CD146 mAb confirmed the presence of CD146 in anti-CD146 immunoprecipitates (Fig. 4, lower panel). These results indicated that CD146 cross-linking induces the recruitment of p59^fyn kinase to CD146.

View larger version (54K): [in this window] [in a new window]   Fig. 4.   Association of CD146 with p59fyn. CD146 engagement was performed as described under "Experimental Procedures." Lysates (500 µg) were immunoprecipitated using S-Endo-1 mAb. Immunoprecipitates were first immunoblotted with the indicated anti-PTK antibodies (upper panel) and reblotted with 7A4 mAb (lower panel).

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

The results reported here demonstrate that in cultured HUVECs, engagement of CD146 initiates a PTK-dependent signaling cascade. Activation of this pathway results in the tyrosine phosphorylation of a complex pattern of proteins, including p125^FAK and paxillin as well as the association of p59^fyn with CD146. To our knowledge, this is the first report demonstrating an outside-in signaling pathway downstream of CD146.

Dimerization of cell surface receptors represents a key event in signal transduction (28). Binding of S-Endo-1 mAb to the extracellular part of CD146 induces a dimerization of CD146 sufficient to promote the tyrosine phosphorylation of intracellular proteins. Cross-linking of CD146 dimers by a secondary antibody leads to an oligomerization of CD146 which subsequently increases the intensity of the phosphorylation events. This process is time- and dose-dependent and requires genistein- and herbimycin-sensitive kinases.

Phosphorylation of tyrosine residues plays an important role in signal transduction by creating docking sites for SH2 domains of signaling molecules (29). Although the CD146 molecule contains a tyrosine residue (Tyr⁶⁴¹) in its cytoplasmic tail (10), we cannot detect any tyrosine phosphorylation of CD146 upon anti-CD146 mAb or upon NaV treatment. These results indicate that CD146 does not serve as a docking site for an SH2 adaptor/effector signaling molecule.

We also show here that upon aggregation, CD146 associates with p59^fyn, a nonreceptor PTK belonging to the Src family kinases (30). p59^Fyn possesses adjacent Src homology SH2 and SH3 domains involved in the binding to target proteins (31). The lack of tyrosine-phosphorylable CD146 excludes the binding by the SH2 domain of p59^fyn (32). In an other way, it is known that the SH3 domains bind to a proline-rich sequence. Although the CD146 cytoplasmic tail contains proline residues, no consensus sequence for SH3 binding site is found (33). Therefore, the molecular basis of the p59^fyn interaction with CD146 is still unknown.

Consistent with the recruitment of p59^Fyn to the membrane upon CD146 engagement, we observe the tyrosine phosphorylation of two major proteins involved in the formation of focal adhesion plaques, p125^FAK and paxillin. These results indicate that CD146 cross-linking induces downstream events that activate p125^FAK and mediate its association with its substrates such as paxillin. Both p125^FAK tyrosine phosphorylation and its association with paxillin depend on the integrity of the cytoskeleton as they are inhibited by pretreatment of cytochalasin D (34).

It is well known that p125^FAK possesses high affinity binding sites for members of Src family kinases, including p59^fyn, which in turn can phosphorylate other tyrosines on p125^FAK (26). p59^fyn participates in the phosphorylation of Tyr at positions 397, 576, 577, and 925, present in the catalytic and COOH-terminal domains of p125^FAK (27, 35, 36). These phosphorylations create multiple binding sites for substrate proteins (26, 38) including paxillin, tensin, and p130^Cas (39-41). Paxillin is a cytoskeletal protein that localizes to sites of adhesion and binds to two sequences present in the carboxyl terminus of p125^FAK (42, 43). Its association with p125^FAK is thought to modulate the localization of p125^FAK to focal adhesion plaques (44-46). Taken together, the results suggest that CD146 engagement mediates the formation of a complex among CD146, p59^fyn, p125^FAK, and paxillin, which promotes focal adhesion assembly.

It is tempting to speculate that CD146 engagement does not only involve an outside-in PTK pathway in HUVEC but also initiates an inside-out PTK-dependent signaling pathway. In mice, p125^FAK gene inactivation impairs cell motility (47), whereas its overexpression stimulates cell migration (48). In addition, p125^FAK and paxillin phosphorylations are increased by cell density (37). Preliminary results² indicate that CD146 expression on the plasma membrane is also increased by cell density. CD146 might therefore participate in the establishment of the endothelial monolayer by initiating downstream events that promote cell proliferation and monolayer formation.

Taken together, the data presented here indicate a role of CD146 in signaling. They suggest that interaction of CD146 with its still unknown ligand might lead to a similar signaling pathway.