Alteration of Endothelial Cell Monolayer Integrity Triggers Resynthesis of Vascular Endothelium Cadherin*

Although cadherins appear to be necessary for proper cell-cell contacts, the physiological role of VE-cadherin (vascular endothelium cadherin) in adult tissue has not been clearly determined. To shed some light on this question, we have disturbed the adhesive function of VE-cadherin in human endothelial cell culture using a polyclonal anti-VE-cadherin antibody. This antibody disrupts confluent endothelial cell monolayers in vitro and transiently generates numerous gaps at cell-cell junctions. The formation of these gaps correlates with a reversible increase in the monolayer permeability. We present evidence that destruction of the homotypic interactions between the extracellular domains of VE-cadherin induces a rapid resynthesis of VE-cadherin, leading to restoration of endothelial cell-cell contacts. The expression of new molecules of VE-cadherin correlates with a modest but significant increase in VE-cadherin mRNA synthesis. Altogether, these results establish a critical role for VE-cadherin in the maintenance and restoration of endothelium integrity.

The vascular endothelium is a semipermeable barrier that forms an active boundary between the bloodstream and the underlying tissues. By modulating its intercellular junctions, the endothelium allows the transmigration of various plasma constituents and regulates the movement of circulating cells between blood and inflamed tissues. Morphologically, three types of organelles constitute the endothelial cell junctions (1). At the apical side of the cells, tight junctions seal the cells to each other. Gap junctions, which are sometimes intercalated with tight junctions, allow exchange of ions and small molecules between adjacent cells. Adherent junctions, located at a more basal position, mediate the physical contacts between cells and anchor the actin cytoskeleton. In addition to proteins that are also encountered at interepithelial contacts, the junctional endothelial structures express marker proteins such as PECAM-1 and VE-cadherin (vascular endothelium cadherin; also known as cadherin-5).
PECAM-1, which appears to be able to interact both with itself through homophilic interactions and with the integrin ␣ V ␤ 3 (2, 3), has been implicated in the recruitment of leukocytes to inflammatory sites and in angiogenesis (4). VE-cadherin (5,6) belongs to the superfamily of cadherin adhesive molecules (7). The extracellular domains of these transmembrane adhesion proteins interact homotypically with cadherins of neighboring cells. For example, the prototype E-cadherin was demonstrated to be required for the tight association of cells in the epithelium. Generally, cadherins are associated via their cytoplasmic tails with ␣-, ␤-, and ␥-catenins (8,9). ␥-Catenin is in fact identical to plakoglobin (10). The sequence of assembly of the cadherin-catenin complex has recently been elucidated by demonstration of mutually exclusive association of ␤-catenin and plakoglobin with the cytoplasmic part of cadherin. There is accumulating evidence indicating that the association of ␣-catenin with cadherin is mediated by ␤-catenin or plakoglobin (11). ␣-Catenin is an F-actin bundling protein implicated in the organization of the actin filaments at the cadherin-based intercellular junctions (12). Another protein termed p120, initially identified as a substrate for various tyrosine kinases (13), was also demonstrated to associate with the E-cadherin-catenin complex (14,15). The fact that the amino acid sequence of p120 contains multiple copies of the Armadillo repeats that are also present in ␤-catenin and plakoglobin suggests that p120 is a catenin-like molecule (15).
The attachment of catenins is of crucial importance for the cell adhesive properties of cadherin (16,17). Indeed, it was demonstrated that significant changes in expression or structure of one of the components within the cadherin-catenin complex could lead to junction disassembly and consequently correlate with an increase in cell invasiveness and expression of malignancy. Some cell lines deficient in ␣-catenin (18,19) or expressing N-terminally truncated ␤-catenin (20) cannot aggregate in a cadherin-dependent manner, although the cells express E-cadherin. In contrast, the transfection of invasive carcinomas with cadherin cDNAs suppresses their invasive properties (21). This suggests that cadherin-catenin complexes have some metastasis-suppressing properties.
There is evidence to suggest that tyrosine phosphorylation is involved in the regulation of the cadherin-catenin complex. In fact, protein-tyrosine kinases have the potential to disrupt tissue architecture by phosphorylating cell junction and cytoplasmic proteins. Transformation of epithelial cells and fibroblasts with the protein-tyrosine kinase v-Src induces phosphorylation of cadherins and their associated catenins and leads to perturbation of intracellular junctions (22,23).
The capacity of the cadherin-catenin complex to be phosphorylated may correlate with the ability of catenins to directly associate with protein-tyrosine kinases. For example, whereas ␤-catenin interacts with the epidermal growth factor receptor (24), p120 associates with the cytoplasmic tyrosine kinase FER (25). Moreover, the protein-tyrosine kinase Fps/Fes, which is structurally similar to the cytoplasmic tyrosine kinase FER (25), can promote angiogenesis in transgenic mice (26). In contrast, the ability of phosphatases such as the receptor proteintyrosine phosphatase PTP to bind to the intracellular domain of cadherins is still under controversy (27,28). Nevertheless, it may be assumed that phosphatases can maintain cadherin and catenins in a dephosphorylated state, thereby stabilizing the cadherin-mediated junctions. Balance between the action of tyrosine kinases and phosphatases may therefore regulate the adhesive properties of the cell-cell junctions.
Although cadherin-catenin complexes are essential for epithelial functions, the components of these complexes are also expressed in other cell types, and their function in non-epithelial tissues is not well established. Thus, the endothelium monolayer expresses the N-, P-, and VE-cadherins and also the associated ␣-, ␤-, and ␥-catenins and p120 (6,29,30). A preliminary study indicated that VE-cadherin is functionally and structurally distinct from the other cadherins (31). Recently, it was demonstrated that VE-cadherin plays a pivotal role in vascular structure elaboration (32) by controlling endothelial cell tube formation (33). We have used a VE-cadherin-directed antibody to improve our knowledge of the physiological role of VE-cadherin in endothelial cell functioning. In this work, we demonstrate that VE-cadherin is critically involved in the maintenance of the integrity of the endothelium. Indeed, we establish that endothelial cells have the capacity to rapidly restore their intercellular junctions by a resynthesis of VEcadherin following alterations in endothelial cell monolayer integrity.

MATERIALS AND METHODS
Cells-Human endothelial cells were isolated as described previously (6).
Expression of VE-cadherin Recombinant Protein Fragment-The cDNA fragment containing the VE-cadherin sequence from nucleotides 166 to 939 was produced by polymerase chain reaction technology using the full-length VE-cadherin cDNA as a template and the following pair of oligonucleotide primers: 5Ј-GAA TTC GAT TGG ATT TGG AAC CAG ATG-3Ј and 5Ј-GT TCT AGA TCA AGC GTC CTG GTA GTC GCC CCG-3Ј. These oligonucleotides were designed to generate a 5Ј-EcoRI restriction site (underlined) and to create a 3Ј-XbaI restriction site (boldface) positioned immediately after an in-frame stop codon. The polymerase chain reaction fragment was ligated with plasmid pCRII (TA cloning kit, Invitrogen/R&D Systems, Abingdon, United Kingdom) and then inserted, after digestion with EcoRI and XbaI, into the pTG1924 expression vector (Transgène, Strasbourg, France). The resulting pTG-VE-Cad vector contained the DNA sequence coding for VE-cadherin (amino acids 1-258) fused with a nucleotide sequence encoding the N terminus of the phage CII protein at its 5Ј terminus (see Fig. 1, upper). Prior to the expression of the protein, the cDNA construct was sequenced to verify that mutations had not arisen during polymerase chain reaction.
To obtain the fragment designated as VE-Cad, Escherichia coli host strain TGE901 was transformed with the pTG-VE-Cad plasmid. Production and purification of VE-Cad was performed according to a protocol previously described (34,35). In a typical experiment, 20 mg of recombinant fragment VE-Cad were obtained from 1 liter of culture. Following purification, proper expression of the protein fragment was assayed by SDS-polyacrylamide gel electrophoresis, N-terminal sequencing, and immunoreactivity of the fusion polypeptide by Western blotting using the polyclonal anti-fusion antibody ZAFP (34).
Production of a Polyclonal Antibody Directed against the Extracellular Part of VE-cadherin-Following purification, the fragment VE-Cad was used as an antigen to raise a polyclonal anti-VE-cadherin antibody (referred to as anti-VE-Cad throughout this paper). Rabbits were injected subcutaneously with 500 g of recombinant fragment VE-Cad in Freund's complete adjuvant at 3-week intervals. Three weeks after the last immunization, rabbits were boosted three times with Freund's incomplete adjuvant. Antisera were applied to 10-ml affinity columns coupled with 30 mg of recombinant fragment VE-Cad. Bound IgG were eluted by a 0.2 M glycine solution (pH 2.2) and immediately neutralized using Tris (under solid form) and extensively dialyzed against 40 mM Tris buffer.
Confocal Microscopy-Confluent living endothelial cells were treated for 1 h with anti-VE-Cad as described for the immunofluorescence experiments. After methanol fixation and incubation with rhodaminetagged anti-rabbit IgG, a series of optical sections of the endothelial cells were obtained using a Zeiss LSM410 confocal scanning laser microscope. Fluorescein isothiocyanate fluorescence was generated using an excitatory wavelength of 488 nm of the argon laser, and the emitted light was selected using a 510-nm dichroic filter and a 510 -540-nm band-pass filter. The optical sections were imaged with a 63ϫ oil immersion objective (NA 1.4, Planapochromat, Zeiss), and the measured resolution of the optical setting was 0.21 m in the x-y plane and 0.45 m along the z axis. The microscope stage was lifted up 0.5 m between each optical section. The starting optical section (i.e. 0 m) was defined at the level of contact between cells and the coverslip. The gain and contrast of the photomultiplier detector were set in order to obtain an optimal imaging of the fluorescence.
Flow Cytometry-To determine whether the polyclonal anti-fragment antibody can recognize native VE-cadherin, its binding to VEcadherin-expressing CHO 1 cells was analyzed by flow cytometry. To preserve VE-antigen, cells were detached as described (36). Briefly, the cell layer was washed several times with Ca 2ϩ -and Mg 2ϩ -containing phosphate-buffered saline and further incubated in the same buffer for 15 min at 37°C. Trypsin (0.01% from bovine pancreas, type III, Sigma) in Hanks' balanced salt solution with 25 mM HEPES (HHBSS), 10 mM CaCl 2 , and 5 mM MgCl 2 was then added and maintained on the cells for 5-10 min, after which initial intercellular retraction was apparent. The cells were then rapidly detached by vigorous shaking of the flasks. Tryptic activity was stopped by the addition of Dulbecco's modified Eagle's medium with 10% fetal calf serum and 0.1% soybean trypsin inhibitor (Sigma). The cells were then centrifuged and resuspended in HHBSS without Ca 2ϩ and Mg 2ϩ to obtain a single cell suspension. The cells were centrifuged again and resuspended in 1% bovine serum albumin in Ca 2ϩ -and Mg 2ϩ -free HHBSS. Aliquots of 10 6 cells were pelleted and resuspended in phosphate-buffered saline containing 0.5 mM Ca 2ϩ in the presence of anti-VE-Cad (10 g/ml). After a 30-min incubation on ice, the cells were pelleted, washed once, and incubated for 30 min with the fluorescein isothiocyanate-conjugated goat antirabbit IgG secondary antibody. The cells were pelleted, resuspended in 0.5 ml of phosphate-buffered saline containing 0.5 mM Ca 2ϩ , and analyzed on a fluorescence-activated cell sorter analyzer (Becton Dickinson).

Treatment of Endothelial Cell Monolayers with Anti-VE-Cad-
To best preserve the cell monolayer, anti-VE-Cad was first diluted 50:50 in twice-concentrated medium 199 containing 40% fetal calf serum, 200 g/ml endothelial cell growth supplement (prepared from bovine brain), 200 g/ml heparin (prepared from porcine intestinal mucosa, Sigma), 100 units/ml penicillin, and 100 g/ml streptomycin. First passage cells were used exclusively. After filtration, a solution of the antibody at 200 g/ml was then applied to living confluent endothelial cell monolayers for various periods of time. The cells were fixed and permeabilized using 3% paraformaldehyde with 0.5% Triton X-100 as described earlier (6,29,37). Coverslips were first incubated with mAb TEA for 1 h at 37°C and then with a mixture of fluorescein isothiocyanate-conjugated antimouse IgG and rhodamine-conjugated anti-rabbit IgG for an additional hour. This double labeling allowed us to distinguish molecules of mAb TEA from molecules of the polyclonal antibody. In fact, the green and red labeling overlapped perfectly, demonstrating that molecules of mAb TEA and those of the polyclonal antibody have the same localization. Only rhodamine staining is presented in Fig. 3. In preliminary exper-iments, concentrations of antibody varying between 50 and 500 g/ml were also used to disturb the endothelial cell-cell contacts. We observed that the intensity of the disrupting effect increased with the concentration of the antibody. The results presented in this paper corresponded to experiments performed in the presence of 200 g/ml polyclonal antifragment antibody. When necessary, endothelial cells were preincubated with cycloheximide at 0.3 g/ml for 4 h and then cultured in the presence of both cycloheximide and anti-VE-Cad (at a final concentration of 200 g/ml) for an additional 2 or 4 h.
Transendothelial Permeability Assay-Cells (6 ϫ 10 4 at seeding) were cultured until confluent for 2 days in Transwell units (polyester filters, 0.4-m pores, Costar Corp., Poly Labo, Lyon, France). Before starting permeability experiments, the integrity of the endothelial cell monolayer was controlled by crystal blue staining of one Transwell unit. In the other Transwell units, the culture medium in the upper compartment was substituted with anti-VE-Cad (0.1 ml at 200 g/ml). At various incubation times, horseradish peroxidase-linked anti-mouse antibody (Amersham Pharmacia Biotech) was added to the upper compartment. After 1 h of incubation at 37°C, the medium in the lower compartment was assayed photometrically for the presence of peroxidase with o-phenylenediamine dihydrochloride (Sigma) as a substrate according to the supplier's instructions. Three individual Transwell units were used for each incubation time.
Cell Lysis and Western Blotting-Monolayers of endothelial cells were washed three times with serum-free medium 199 and incubated with extraction buffer (10 mM Tris-HCl (pH 7.5) and 150 mM NaCl containing 2 mM CaCl 2 , 1 mM phenylmethylsulfonyl fluoride, 40 units/ml aprotinin, 15 g/ml leupeptin, 0.36 mM 1,10-phenanthroline, 1% Nonidet P-40, and 1% Triton X-100; 500 l for a 25-cm 2 area) for 30 min on ice with occasional gentle agitation. The cell extracts were then centrifuged at 15,000 ϫ g for 5 min at 4°C. The Triton X-100-soluble fraction was electrophoresed on a 7.5% polyacrylamide gel after boiling in ␤-mercaptoethanol-containing Laemmli buffer. Polyacrylamide gels were incubated for 30 min in transfer buffer (95 mM glycine and 1 mM CaCl 2 containing Tris-HCl (pH 8.9)). Separated proteins were then electroblotted onto nitrocellulose (Amersham Pharmacia Biotech) and blocked with 10% low-fat milk in CaCl 2 and MgCl 2 containing phosphate-buffered saline (blocking buffer). The blots were incubated with mAb TEA in blocking buffer, and antigen was then detected using the ECL kit (Amersham Pharmacia Biotech) with a peroxidase-labeled goat anti-rabbit antibody.
To quantify the resynthesis of VE-cadherin, living endothelial cells were treated with the anti-fragment antibody anti-VE-Cad as described above. After various incubation times, cell monolayers were directly lysed in Laemmli buffer for 5 min on ice with gentle agitation. Following the addition of ␤-mercaptoethanol, cell extracts were heated to 100°C, electrophoresed, and transferred to membranes as described above. The presence of VE-cadherin was determined by ECL. Sequential reprobing of the membrane with specific antibodies was used to detect ␣and ␤-catenins and plakoglobin according to the supplier's instructions. The contents of VE-cadherin, ␣and ␤-catenins, and plakoglobin were also evaluated by Western blotting, with quantification being estimated from the density and the surface of the immunoblotted bands using an image analysis system (Bio-Rad molecular analysis software). Northern Blot Analysis-Total cellular RNA was isolated using the Microscale rapid RNA isolation kit (5 Prime 3 3 Prime, Inc., Boulder, CO) from endothelial cell monolayers treated for various periods of time with anti-VE-Cad as described above. The samples were separated on a 1% agarose gel in 40 mM MOPS (pH 7.0), 10 mM sodium acetate, 1 mM EDTA, and 17% formaldehyde. The gels were rinsed in water to remove the formaldehyde and then transferred to a nylon membrane (Hybond-N, Amersham Pharmacia Biotech) with 10ϫ SSC by capillary action overnight. The membrane was baked for 2 h at 80°C. Hybridizations with 32 P-labeled plasmid DNA probes were performed overnight at 65°C in 3ϫ SSC, 0.1% SDS, 5ϫ Denhardt's solution, 10% dextran sulfate, and 100 g/ml boiled salmon sperm DNA. Membranes were washed first at 65°C with 2ϫ SSC and 0.1% SDS and then under high stringency conditions (0.1ϫ SSC and 0.1% SDS for 30 min at 65°C). The blots were revealed by autoradiography on X-Omat AR film at Ϫ70°C. Northern blots were quantified using a PhosphorImager research software program (ImageQuant, Version 3.3, Molecular Dynamics, Inc.).
The VE-cadherin cDNA probe was a generous gift from Dr. P. Huber (Laboratoire de Transgénèse, Département de Biologie Moléculaire et Structurale, Commissariat à l'Energie Atomique, Grenoble, France). Plasmid containing cDNA encoding glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used to quantify mRNA levels for each condition. Probes were labeled with [␣-32 P]dCTP (Amersham Pharmacia Biotech) by a standard random-primed DNA procedure (randomprimed DNA labeling kit, Boehringer Mannheim) to a specific activity of 5-9 ϫ 10 8 cpm/g of DNA. Unincorporated precursors were removed from labeled DNA by Quick Spin Sephadex G-50 columns (Boehringer Mannheim).

Characterization of Anti-VE-Cad-
To establish the physiological role of VE-cadherin, a polyclonal anti-VE-cadherin antibody, designated as anti-VE-Cad, was raised against a recombinant fragment protein encompassing amino acids 1-258 of VE-cadherin (Fig. 1, upper). This antibody was immunopurified and was able to detect VE-cadherin fragments coated at 0.2 M by enzyme-linked immunosorbent assay (data not shown).
To demonstrate that anti-VE-Cad effectively recognized cellsurface expressed antigen, fluorescence-activated cell sorter analyses were performed on CHO cells expressing human VEcadherin. These revealed that anti-VE-Cad recognized VE-cadherin-transfected cells but failed to react with the untransfected CHO cells ( Fig. 2A). Moreover, immunofluorescence staining indicated that anti-VE-Cad selectively decorated fixed human endothelial cell monolayers at cell-cell junctions (Fig.   2B). This staining pattern was very similar to that observed with the monoclonal anti-VE-cadherin antibody TEA (6) and consisted of a thin and sharp continuous fluorescent signal highlighting the cell-cell junctions. Western blotting performed on endothelial cell lysates indicated that anti-VE-Cad recognized a single 130-kDa protein (Fig. 2C). This result is consistent with the molecular mass of VE-cadherin as already observed using mAb TEA (6). Altogether, the results demonstrate that anti-VE-Cad is able to specifically recognize native VEcadherin. In particular, it did not interact with the 140-kDa N-cadherin that is also expressed by endothelial cells (38 -41), despite a 25% amino acid sequence identity.
Reversible Disruption of Endothelial Cell Contacts by Anti-VE-Cad-The anti-fragment antibody was then used in an effort to block the adhesive property of human VE-cadherin. To evaluate its capacity to interfere with the biological activity of VE-cadherin, anti-VE-Cad was added to confluent living endothelial cell monolayers. In preliminary experiments, the cells were examined by phase-contrast microscopy after a 2-h incubation. The addition of the polyclonal antibody caused cell retraction, whereas the preimmune immunoglobulins had no effect on the confluency of the cell layer. Furthermore, following antibody treatment, cells lost their cobblestone morphology, and some of them adopted an elongated shape. No detachment of the cells from the culture surface by anti-VE-Cad treatment was detected, indicating that the interactions of the endothelial cells with the fibronectin matrix were not affected by treatment with anti-VE-Cad (data not shown).
Antibody-treated cells were also observed by indirect immunofluorescence microscopy. The corresponding images, taken 2 h after the addition of anti-VE-Cad to the living endothelial cell monolayers, revealed the formation of numerous randomly distributed intercellular gaps (Fig. 3B, arrows). This was also accompanied by an important redistribution of the VE-cadherin staining pattern. This suggests that, after incubation with anti-VE-Cad, some molecules of VE-cadherin lose their cell junction localization and are either internalized or redistributed over the whole cell surface. In contrast, preimmune immunoglobulins, at the same concentration, did not perturb the integrity of the endothelial cell monolayer (Fig. 3A), thus demonstrating that the disrupting effect observed with anti-

VE-Cad was specific.
Observation of endothelial cells at 4 h after the addition of the antibody revealed only sparse small intercellular gaps, with an almost complete loss of the zigzag staining pattern. Compared with cells treated with antibody for 2 h, cells treated for 4 h exhibited a partial restoration of the VE-cadherin intercellular staining (Fig. 3C, arrows).
Transient Permeabilization of the Endothelial Cell Monolayer by Anti-VE-Cad-The disruption of the endothelial cell monolayer following treatment with anti-VE-Cad was quantified by establishing the transmigration time course of peroxidase-conjugated anti-mouse IgG across endothelial cell monolayers that had been seeded on porous Transwell chambers (Fig. 4). Binding of anti-VE-Cad to endothelial cells induced an increase in the migration rate of peroxidase-conjugated IgG. The migration rate was maximal 1 h after the beginning of the treatment and returned to the basal level 6 h following the addition of anti-VE-Cad. In contrast, no increase in transmigration was observed with monolayers treated with preimmune immunoglobulins. The kinetic measures of endothelial cellinduced permeability correlate well with the immunofluorescence microscopy results (Fig. 3) and reflect a transient opening of the endothelial cell barrier.
Localization of VE-cadherin following Endothelial Cell Treatment with Anti-VE-Cad-Attempts to localize VE-cad-herin more accurately after treatment with anti-VE-Cad were performed by analyzing its immunofluorescent staining patterns using confocal laser scanning microscopy (Fig. 5). In contrast to untreated cells (Fig. 5A, panel 4), only weak VEcadherin staining was detected on the apical side of the cells treated with anti-VE-Cad (Fig. 5B, panel 4). This demonstrated that VE-cadherin was distributed inside the cytoplasm and not at the cell surface. After treatment of endothelial cell monolayers with anti-VE-Cad, VE-cadherin became mostly cytoplasmic and exhibited a circular staining pattern observed in intermediate cross-sections of the cells (Fig. 5B, panels 1-4). This pattern is indicative of an endoplasmic reticulum localization and suggests that treatment of endothelial cells with anti-VE-Cad triggers a neosynthesis of VE-cadherin.
Resynthesis of VE-cadherin following Anti-VE-Cad Treatment-The restoration of cell-cell contact following prolonged treatment with anti-VE-Cad may result from the recruitment of preexisting or newly synthesized molecules of VE-cadherin to adhesive junctions. To discriminate between these two possibilities, VE-cadherin was quantified by immunoblotting before and after different incubation times with anti-VE-Cad. As shown in Fig. 6A,  of VE-cadherin and ␣-, ␤-, and ␥-catenins in same cell lysates were quantified using image analyzer software (Fig. 6B). After a 2-h incubation with anti-VE-Cad, the amount of VE-cadherin decreased to 20% compared with the initial one. Extension of the incubation time up to 4 h was accompanied by an increase in VE-cadherin up to 60% of the initial pool. By contrast, the amounts of ␣and ␤-catenins and plakoglobin in the lysates were only marginally affected by anti-VE-Cad treatment (Fig.  6, A and B). These results suggest, first, that the initial decrease in VE-cadherin content might be due to a rapid degradation of antigen-antibody complexes. Second, the increase in VE-cadherin content following prolonged antibody incubation could be attributed to a resynthesis of VE-cadherin molecules. The latter conclusion is supported by our results obtained by blocking intracellular protein synthesis by cycloheximide.
Indeed, in a second series of experiments, endothelial cell monolayers were first incubated with cycloheximide prior to the addition of anti-VE-Cad (Fig. 6A, lanes 4 -6). To determine the concentration to be used in cell culture, serial dilutions of cycloheximide (10 mg/ml stock in Me 2 SO) were tested for toxicity. Incubation with cycloheximide at 0.3 g/ml for up to 14 h affected neither the morphology nor the junctions of the endothelial cells. Western blot analysis indicated that cycloheximide greatly reduced the amount of VE-cadherin in cells treated with antibody anti-VE-Cad (Fig. 6A, lanes 5 and 6). Blockage of protein synthesis prevented the increase in VEcadherin content previously observed between 2 and 4 h of incubation with anti-VE-Cad. It was noticed that the VE-cadherin content was reduced to 5% of the initial pool in the presence of cycloheximide and after a 4-h incubation with anti-VE-Cad. In contrast, cycloheximide treatment alone did not modify VE-cadherin content, confirming the slow turnover of VE-cadherin molecules (Fig. 6A, compare lanes 1 and 4). In comparison, the contents of ␣and ␤-catenins and plakoglobin diminished progressively with time in the presence of cycloheximide (Fig. 6, A (lanes 5 and 6) and B). Taken together, these Correlation of Resynthesis of VE-cadherin with Increase in VE-cadherin mRNA Level-To gain information on the origin of these neosynthesized molecules of VE-cadherin, we then examined the effect of anti-VE-Cad on the cellular level of VE-cadherin mRNA. Total RNAs were extracted from cells pretreated for various periods of time with anti-VE-Cad. Northern blotting was then performed using a VE-cadherin probe and also a GAPDH probe to normalize VE-cadherin mRNA data. As shown in Fig. 7A, transcripts of ϳ4.4 and 1.8 kilobases were detected with the VE-cadherin and GAPDH probes, respectively. The results showed that, after a 1-h incubation with anti-VE-Cad, the amount of VE-cadherin transcript remained similar to that in untreated cells (Fig. 7B). By contrast, between 1 and 2 h of incubation, a 70% VE-cadherin RNA increase was observed. Incubation with anti-VE-Cad for 3 and 6 h did not modify the level of VE-cadherin mRNA. This correlated with the time course observed for the resynthesis of VE-cadherin molecules. Indeed, following a 2-h incubation, the increase in VE-cadherin protein coincided with the increase in VE-cadherin RNA. DISCUSSION To clarify the role played by VE-cadherin at the endothelial cell junctions, we have generated anti-VE-Cad, a polyclonal antibody directed against the extracellular EC1/EC2 modules of VE-cadherin (Fig. 1). We observed that anti-VE-Cad interferes with the adhesive properties of VE-cadherin. Our results therefore indicate that, similar to E-cadherin, which supports intercellular cohesion of the epithelium, VE-cadherin is in-volved in the maintenance of the integrity of the endothelium. Our result contrasts with the finding that VE-cadherin CHO transfectants exhibit only a weak cell aggregation activity compared with the other cadherin transfectants (31,42). The discrepancy may be due to the fact that full adhesive activity of VE-cadherin requires additional components not expressed in CHO cells. Alternatively, VE-cadherin molecules may be degraded or possibly not correctly oriented at the surface of transfected CHO cells.
Binding of anti-VE-Cad to EC1 and EC2 modules leads to the destruction of intercellular contacts between endothelial cells. This indicates that the N-terminal EC1 and EC2 modules of the VE-cadherin molecule are involved in homotypic interactions. This result is in agreement with those of Shapiro et al. (43), Koch et al. (44), and Nagar et al. (45), who demonstrated that the homodimerization of E-and N-cadherins requires Nterminal EC1 modules. The absence of the homotypic recognition sequence His-Ala-Val present in classical cadherins (46) does not prevent VE-cadherin homotypic interactions.
In this work, we have presented evidence that endothelial cell junctions have the capacity to disassemble and assemble over the course of several hours. Indeed, the anti-VE-Cadinduced reversible permeability was due to an initial disorganization, followed by a reconstitution of endothelial cell-cell contacts. Such an up-regulation of cell-cell association was also observed after treatment of endothelial cells with EDTA, which is known to disrupt the Ca 2ϩ -dependent homotypic interactions between VE-cadherin molecules (29). We have correlated the disorganization of endothelial cell intercellular junctions with a decrease in the overall cellular content of VE-cadherin. The restoration of endothelial cell adhesion corresponds to an increase in the total amount of VE-cadherin. The antibodyinduced decrease in the VE-cadherin level might be due to endocytic degradation of VE-cadherin-antibody complexes. Restoration of the cell-cell contacts corresponds to the recruitment of newly synthesized molecules of cadherin to the developing adherent junctions.
We have shown that restoration of the cell-cell contacts requires the targeting of newly synthesized VE-cadherin to endothelial cell junctions. This observation may be of importance for the understanding of the regulation of VE-cadherin-mediated cell adhesion. Resynthesis and transfer of VE-cadherin to cell-cell contacts permit endothelial cells to rapidly remodel their junctions. Recently, it was hypothesized that adhesion of neutrophils might trigger intracellular signals in endothelial cells that possibly regulate VE-cadherin-catenin disorganization (52). This disorganization, which accelerates neutrophil recruitment in vivo (47), may be due to a down-regulation of VE-cadherin adhesive properties.
Taken together, our data demonstrate that the destabilization of intercellular contacts between endothelial cells initiates a transcellular signal leading to the synthesis of new molecules of VE-cadherin. The existence of such intracellular signaling was confirmed by our experiments demonstrating an increase in the VE-cadherin mRNA level after anti-VE-Cad treatment. This up-regulation of mRNA expression correlates with the time course of both the resynthesis of VE-cadherin and the reconstitution of the endothelial intercellular junctions. This is also compatible with the 6-h half-life time for VE-cadherin mRNA. 2 The first step of this signal may be initiated by the release of catenins from the cytoplasmic tail of VE-cadherin. This is supported by the finding that ␤-catenin is involved in similar signal transduction in other cell types (48). The ␤-cateninmediated signaling was demonstrated to be regulated in a 2 P. Huber, personal communication.  5 and 9). Total cellular RNA (10 g) isolated from endothelial cells was hybridized with 32 P-labeled VE-cadherin and 32 P-labeled GAPDH probes. Kb, kilobases. B, quantification of the ratio of VE-cadherin mRNA to GAPDH mRNA was done by PhosphorImager scanning. For each incubation time, one mRNA extraction was performed. The error bars were estimated from four independent successive PhosphorImager scans. Dark shaded bars, anti-VE-Cad-treated endothelial cells; light shaded bars, untreated endothelial cells. complex manner by various ␤-catenin-associated proteins (24, 49 -51). Following anti-VE-Cad treatment, detachment of catenins from VE-cadherin may permit their association with some unknown cytoplasmic targets, thus inducing the propagation of the intracellular signal. We are currently trying to identify the various molecules involved in the cadherin-mediated signal transduction pathway.