Identification of a Novel Cadherin (Vascular Endothelial Cadherin-2) Located at Intercellular Junctions in Endothelial Cells*

Endothelial cells express two major cadherins, VE- and N-cadherins, but only the former consistently participates in adherens junction organization. In heart microvascular endothelial cells, we identified a new member of the cadherin superfamily using polymerase chain reaction. The entire putative coding sequence was determined. Similarly to protocadherins, while the extracellular domain presented homology with other members of the cadherin superfamily, the intracellular region was unrelated either to cadherins or to any other known protein. We propose for this new protein the name of vascular endothelial cadherin-2. By Northern blot analysis, the mRNA was present only in cultured endothelial cell lines but not in other cell types such as NIH 3T3, Chinese hamster ovary, or L cells. In addition, mRNA was particularly abundant in highly vascularized organs such as lung or kidney. In endothelial cells and transfectants, this cadherin was unable to bind catenins and presented a weak association with the cytoskeleton. This new molecule shares some functional properties with VE-cadherin and other members of the cadherin family. In Chinese hamster ovary transfectants it promoted homotypic Ca2+ dependent aggregation and adhesion and clustered at intercellular junctions. However, in contrast to VE-cadherin, it did not modify paracellular permeability, cell migration, and density-dependent cell growth. These observations suggest that different cadherins may promote homophilic cell-to-cell adhesion but that the functional consequences of this interaction depend on their binding to specific intracellular signaling/cytoskeletal proteins.

Endothelial permeability to plasma proteins and circulating cells is controlled in part by intercellular junctions. Besides their role in promoting homotypic cell adhesion, emerging evidence suggests that intercellular junctions can transfer cellcell signals and be responsible for complex cellular responses such as contact inhibition of cell growth and cell polarity.
The molecular organization of intercellular junctions in the endothelium has been only partially elucidated in the last few years. At least three types of complex structures have been described: tight junctions, adherens junctions, and complexus adhaerentes. All of these structures are formed by specific transmembrane proteins, which through their extracellular region promote homotypic cell-to-cell adhesion and through the cytoplasmic tail bind to a complex network of cytoskeletal and signaling proteins (1)(2)(3). Outside of these junctional structures, other adhesive proteins such as PECAM 1 (4) or S-Endo-1/ Muc-18 (5) have been found to be clustered at intercellular contacts. The intracellular molecules that associate with adherens junctions are different from those that associate with tight junctions and from those that link other junctional adhesion proteins such as PECAM, suggesting that a certain specificity in signaling should exist (3).
In adherens junctions, the transmembrane proteins responsible for cell-to-cell adhesion belong to the cadherin superfamily of adhesive proteins. In endothelial cells, one of the major cadherins is VE-cadherin or cadherin-5 (VE-cad), which is consistently present at adherens junctions and is cell-specific (2,3,6). Similarly to the other members of the family, the short intracellular tail of VE-cad is linked to three cytoplasmic proteins called catenins: ␤-catenin, plakoglobin, and p120. ␤-Catenin and plakoglobin bind ␣-catenin, which in turn promotes the anchorage of the complex to the actin cytoskeleton. As the other known cadherins, the extracellular domain of VE-cad promotes homotypic, calcium-dependent adhesion.
Endothelial cells also express N-cadherin, which in human endothelium does not colocalize with VE-cad at cell contacts but remains diffuse on the cell membrane (7,8).
Besides these well characterized cadherins, some indirect evidence suggests that other cadherin-like structures may be present in the endothelium (9). Therefore, we started an investigation to test this possibility. Using a polymerase chain reaction method previously introduced by Suzuki et al. (6) and Sano et al. (10), we identified a new protein that, on the extracellular domain, presents homology with cadherins. This protein is concentrated at intercellular junctions and expresses adhesive properties, but, in contrast to VE-cad, does not bind to catenins and does not modify paracellular permeability, cell migration, or growth. This indicates that several proteins may participate in the molecular organization of interendothelial junctions, but each molecule may play a specific functional role and possibly transfer different intracellular signals. For its localization in endothelial cells and for its homology with the cadherin and protocadherin family, we propose for this new protein the name of vascular endothelial cadherin-2 (VE-cad-2).

MATERIALS AND METHODS
All reagents were purchased from Sigma unless indicated otherwise.
PCR-PCR was performed as described previously by Suzuki et al. (6) and Sano et al. (10). Template cDNA was synthesized using mouse heart microvascular endothelial cell (H5V) (11) total RNA according to the protocol of the GeneAmp RNA PCR kit (Perkin-Elmer). Two different sets of degenerated oligonucleotide primers were used in this study. One set corresponds to two well conserved sequences in the cytoplasmic domain of cadherin; the upstream and downstream primers were 5Ј-G-AATTCAC ( The PCR reaction consisted of 33 cycles at 95°C for 1 min, 45°C for 2 min, and 72°C for 3 min using Taq DNA polymerase obtained from Boehringer Mannheim. The PCR products were subcloned into the pCR II plasmid by use of the TA cloning kit (Invitrogen Co., San Diego, CA) and sequenced according to the dideoxynucleotide chain termination method (12).
Screening of a cDNA Library and Sequence Analysis-About 8 ϫ 10 5 plaques of a gt10 cDNA library from postnatal day 4 -8 mouse brain microvasculature (13) were screened for clone 1 and clone 14 by a plaque hybridization method, as described previously (14), using as probes the cDNAs obtained from PCR experiments. cDNAs were radiolabeled with [ 32 P]␣-dCTP (Amersham International, Buckinghamshire, UK), using a random primer DNA labeling kit obtained from Boehringer Mannheim. Plaques showing a strong positive hybridization signal were screened four times to obtain a single phage clone.
Phage inserts were subcloned in the pBluescript vector and sequenced by Genomic Express S.A. (Grenoble, France). The nucleotide and the deduced protein sequences were screened against the data bank as described previously (15).
Plasmid Construction and Transfection-The full-length open reading frame for VE-cad-2 was cut with EcoRI, and the insert was subcloned into the pECE eukaryotic expression vector (16) to yield the pECE-VE-cad-2 construct. CHO cells were plated at 5-6 ϫ 10 5 cells/ 10-mm Petri dish in Dulbecco's modified Eagle's medium with 10% fetal calf serum. After about 24 h, the cells were transfected by calcium phosphate precipitation with 20 g of pECE-VE-cad-2 and 2 g of pCMVneo plasmid. Cells were washed 24 h later with PBS and cultured for another 24 -36 h in Dulbecco's modified Eagle's medium with 10% fetal calf serum. They were then cultured in presence of 600 g/ml G418 (Geneticin; Life Technologies, Inc.). After about 10 days in selective medium, the surviving colonies were ring-cloned. G418-resistant clones were screened for expression by Northern blot analysis and indirect immunofluorescence microscopy. Control cells, CHO cells transfected with pCMVneo were selected, cloned, and cultured in the same way.
RNA Expression and Northern Blot Analysis-Total RNA was extracted and purified by use of the Rapid Total RNA isolation kit (5 Prime 3 3 Prime, Inc., Boulder, CO) and 20 g were run in a standard formaldehyde/agarose gel, blotted onto Hybond N membrane (Amersham International), fixed at 80°C for 2 h, and hybridized at 65°C in a buffer containing 10% dextran sulfate, 3ϫ SSC, 5ϫ Denhardt's solution, 10% SDS, 100 g/ml denatured salmon sperm DNA. The membranes were then washed twice with 2ϫ SSC and 0.1% SDS at room temperature for 10 min, twice with 0.5ϫ SSC and 0.1% SDS at 65°C for 15 min, and once with 0.1ϫ SSC and 0.1% SDS at 65°C for 10 min and then subjected to autoradiography.
Sterile plasticware was from Falcon (Becton Dickinson, Lincoln Park, NJ); both culture medium and serum were from Life Technologies, Inc.
Antibodies-A rabbit antiserum was raised against a recombinant fragment spanning the extracellular domain of VE-cad-2 (74 -335 amino acids). The fragment was generated by PCR. The primers were designed to create at the 5Ј-end a BamHI site and at the 3Ј-end a HindIII restriction site. The cDNA fragment was then subcloned into the BamHI-HindIII site of the expression vector pQE30 (Qiaexpressionist kit; Qiagen, Chatsworth, CA) in the correct reading frame and sequenced to verify that no mutation had arisen during PCR. The resulting pQE30-VE-cad-2 vector was then introduced into M15 (pREP4) cells by a single step transformation method. The fusion protein was induced by the addition of isopropyl-␤-D-thiogalactopyranoside and was purified from the extract by nickel-nitrilotriacetic acid resin affinity chromatography, as described by the manufacturer (Qiaexpressionist kit; Qiagen). Polyclonal antibody against VE-cad-2 was produced in rabbit by injecting 0.5 mg of the fusion protein in Freund's complete adjuvant at three subcutaneous sites. Subsequent injections were in Freund's incomplete adjuvant with 0.5 mg of the fusion protein. The fragment antiserum was affinity-purified by affinity chromatography on the corresponding fragment affinity column CN-Br-Sepharose 4B (Pharmacia LKB Biotechnology, Uppsala, Sweden). The antiserum was further characterized for its positive reaction with endothelial cells (H5V) and VE-cad-2 transfectants by enzymelinked immunosorbent assay, immunoprecipitation, Western blot, and immunofluorescence staining of fixed cell monolayers.
Fluorescein-and rhodamine-conjugated secondary antibodies (reactive with either mouse or rabbit IgG) were purchased from Dakopatts (Glostrup, Denmark).
Goat anti-mouse IgG peroxidase-conjugated antibody and protein A peroxidase-conjugated antibody used for immunoblotting detection were from Pierce.
Immunofluorescence Microscopy-Cells were grown on glass coverslips, rinsed in PBS, and fixed in methanol. The cells were then rinsed and incubated for 45 min at 37°C with the relevant primary antibodies, washed three times with PBS, and incubated for 30 min with the fluorophore-conjugated secondary antibodies. Coverslips were then mounted in Mowiol 4 -88 (Calbiochem) and examined with a Zeiss Axiophot microscope. Photographs were taken using T-Max P3200 films.
EGTA Treatment-EGTA was used for chelating calcium ions in the culture medium as described previously (21). A buffer stock solution of 100 mM EGTA was used to obtain a final concentration of 5 mM. Cells grown to confluence on glass coverslips were incubated with 5 mM EGTA at 37°C for 30 min, fixed, and processed for indirect immunofluorescence as described above.
Cytochalasin Treatment-Cells were cultured to confluence on glass coverslips and treated with 1 g/ml cytochalasin D in culture medium. 30 min later, cells were fixed with 3% paraformaldehyde, permeabilized with 0.5% Triton X-100, and processed for indirect immunofluorescence as described above.
Cell Surface Biotinylation-Biotinylation of cell surface proteins was performed as described elsewhere (22) using sulfonitrohydroxysuccinimido-biotin (Pierce). Samples were analyzed by electrophoresis on a 7.5% SDS-polyacrylamide gel and transferred to nitrocellulose membrane. The membranes were blocked with 10% low fat milk and then incubated in fresh blocking solution with horseradish peroxidase-conjugated streptavidin (Biospa Division, Milano, Italy) for 1 h at room temperature. After three washes with PBS containing 0.1% Tween 20, peroxidase-conjugated streptavidin was visualized using the ECL kit as described under "Blot and Immunoprecipitation." Western Blot and Immunoprecipitation-Whole cell extracts were obtained from confluent cells as described previously (23). Detergent solubilization was carried out essentially as reported previously in detail (24). Different cell extracts were adjusted to 1ϫ Laemmli sample buffer and fractionated under reducing conditions on 7.5% SDS-polyacrylamide gels (25).
Western blot analyses of the various cell extracts were carried out essentially as described (24). After blocking with 10% low fat milk, the proteins of interest were detected by specific monoclonal or polyclonal antibodies at the optimal dilution in blocking buffer. This was sequentially followed by incubation with goat anti-mouse IgG peroxidaseconjugated antibody (1 mg/ml) for monoclonal antibodies or protein A peroxidase-conjugated antibody (1 mg/ml) (Pierce) for polyclonal antibody and further development of peroxidase activity using an ECL kit (Amersham Biotech Pharmacia International) and autoradiography.
Immunoprecipitation of the cadherin-catenin complex was per-formed using the nonionic detergent-soluble fraction of cells, as previously reported (24) with some modifications. Briefly, cell extracts were precleared by incubation with uncoupled protein G-or protein A-Sepharose CL-4B (Amersham Biotech Pharmacia) for 2 h. After centrifugation, the precleared supernatants were incubated with protein G-or protein A-Sepharose coupled to mAb TEA 1.31 or polyclonal antibody against VE-cad-2 during 1 h. Immunocomplexes were collected by centrifugation; washed five times in a buffer containing 0.5% Triton X-100, 0.1% bovine serum albumin, 50 mM Tris-HCl, pH 7.4, 0.1 M NaCl, and 2 mM CaCl 2 ; and finally resuspended in 30 l of 1ϫ Laemmli sample buffer and boiled for 5 min. Samples were analyzed by electrophoresis, transferred to nitrocellulose membranes, and immunoblotted sequentially with polyclonal antibody to VE-cad-2 or mAb TEA 1.31 to VE-cad or mAbs to ␣and ␤-catenin as described above.
Cell Adhesion-CHO, CHO-VE-cad-2, or CHO-VE-cad cells were cultured in 96-well plates and grown for 5 days to confluency. All three types of cells were labeled with [ 125 I]iododeoxyuridine (1 mCi/ml) overnight prior to the cell adhesion experiment. 12 h later, the cells were detached as described above and resuspended at 3 ϫ 10 5 cells/ml in Dulbecco's modified Eagle's medium with 10% fetal calf serum. 100 l of labeled cell suspension were added to different adherent cell monolayers (CHO, CHO-VE-cad-2, and CHO-VE-cad) and incubated for 1 h at 37°C. After three washes with PBS containing 10% fetal calf serum, the cells were solubilized with 0.5 M NaOH, 0.1% SDS and counted in a ␥-counter.
Permeability Assay-Permeability across cell monolayers was measured in Transwell units (with polycarbonate filter, 0.4-m pore; Corning Costar Corp., Cambridge, MA) as described previously (15). Briefly, CHO transfectants were cultured to confluency for 5 days. Then culture medium was replaced with serum-free medium, and horseradish peroxidase conjugated to goat immunoglobulins (8 mg/ml initial concentration in the upper chamber; minimal calculated molecular mass, 200 kDa; specific activity, 28 units/mg) was added to the upper chamber. At 2 h, 100-l aliquots were collected from the lower compartment and assayed photometrically for the presence of enzymatic activity. In some experiments, EGTA (5 mM, final concentration) was added both to the lower and upper compartments for 2 h at the same time as immunoglobulins.
Cell Migration-Cell migration was estimated as described previously (15). Briefly, the cell monolayer was wounded with a plastic tip. Four diameters, regularly distanced by about 45°, were removed. The remaining cells were washed twice with culture medium to remove cell debris and incubated at 37°C in culture medium. At the indicated time intervals, cells were fixed with Fast Green FCF (0.02% in methanol) and stained with crystal violet (0.5% in a 20:80 mixture of methanol/ water). The distance migrated by the cells was measured using a micrograduate scale (Nikon) adapted in the ocular of a Nikon inverted microscope under phase contrasts (magnification ϫ 100).
Cell Growth Assay-Cell growth was evaluated as described previously (27). Cells were plated at 1 ϫ 10 4 /ml (1 ml/well) in 24-well plates (2 cm 2 /well). Culture medium was not changed for the duration of the experiment (96 h). Cell number was evaluated after trypsinization of the cells and counting (four replicates) in a Bü rker chamber.

Cloning of a New Member of the Cadherin Superfamily-A
PCR method was applied to identify new members of the cadherin superfamily. As primers we first used two degenerated oligonucleotides corresponding to two highly conserved sequences in the cytoplasmic domain of cadherins as previously described by Suzuki et al. (6). PCR was carried out using cDNA obtained from mouse heart microvascular endothelial cells (H5V). The resulting 160-bp products were then subcloned in the pCRII plasmid and sequenced. Of 35 clones sequenced, three clones encoded the amino acid sequence of N-cadherin, and two clones encoded the amino acid sequence of VE-cad. The other cDNA clones encoded amino acid sequences that did not present homology with the cytoplasmic domain of cadherins.
In the second part of the research, we used as primers degenerated oligonucleotides corresponding to two conserved sequences of the extracellular domain of cadherins, as described previously by Sano et al. (10).
PCR from the cDNA of the cell line H5V yielded four major bands of 450, 370, 300, and 130 bp in size. The 450-and 130-bp bands correspond to the distance between the two primer sites in classic cadherins and the two primer sites in protocadherins, respectively (10). The 370-and 300-bp bands would not be predicted from any of the known members of the cadherin superfamily and were therefore discarded.
The 130-bp product was subcloned into the pCR II vector, and 30 clones were isolated and sequenced. Two cDNAs (clones 1 and 14) presented a novel sequence and were considered good candidates to be new putative members of the cadherin superfamily.
A cDNA library of postnatal day 4 -8 mouse brain capillaries (13) was screened by using clones 1 and 14 as probes. Four clones of 5.8, 4.0, 2.4, and 1.8 kb were obtained by screening with clone 1, while only a clone of 700 bp was obtained using clone 14 as probe. The sequence analysis of the 700-bp clone revealed a partial sequence that did not correspond to any previously identified sequence (data not shown).
The sequences of the two cDNA clones of 5.8 and of 4.0 kb obtained by screening with clone 1 overlap and appear to contain the full-length open reading frame of a novel member of the cadherin superfamily. The nucleotide and deduced amino acid sequences of the 4.0-kb clone are shown in Fig. 1. The 3868-bp sequence contains 298 bp of putative 5Ј-untranslated region, an open reading frame of 3540 nucleotides encoding 1180 amino acids, and a short 3Ј-untranslated region of 29 bp. At position 299, the cDNA sequence contains a translation initiation site that matches the Kozak criteria (28). The polyadenylation signal was not identified.
The sequence presents a signal peptide, an extracellular region that can be divided into six domains (EC1-EC6), a transmembrane domain, and a large cytoplasmic region of 443 amino acids. The repeats EC1-EC5 are 104 -109 residues long and present homology to the cadherin repeats found in the extracellular region of classic cadherins. The EC6 domain is 144 amino acids long and does not contain characteristic features of the domains of cadherin superfamily members. EC1-EC5 contain repeated sequences that are highly conserved in the cadherin family, such as DXD, LD(R/Y)E, and DXNDNXP. In the domains EC2, EC3, EC4, and EC5, between the sequences LD(R/Y)E and DXNDNXP, there is the motif AXDXG, conserved among members of the protocadherin family. The cytoplasmic domain is distinct from any cytoplasmic region of members of cadherins superfamily and from any other sequences present in the data bank.
A homology search in the data bank revealed that the clone was related to the protocadherin family (about 40% amino acid identity with human protocadherin 68, protocadherin 2, protocadherin 3, and protocadherin 4). Amino acid identity is always restricted at the extracellular region of the molecule. Comparison of the amino acid sequence of the first five extracellular domains with other cloned cadherin superfamily members shows that the clone has about 25% of amino acid identity with several components of the superfamily such as human and Drosophila fat, human cadherin 6, and mouse cadherin 5/VEcad. Based on these results, we conclude that this new protein is similar to cadherins and, more specifically, to protocadherins. The cDNA sequence, however, did not correspond to any of the previously identified cadherin (6, 29 -31) or protocadherin (10, 32) cDNA sequences. Since it is localized in endothelial cells and it is homologous with the cadherin and the protocadherin family, we propose that the name of this new protein might be VE-cad-2.

VE-cad-2 Is Expressed in Different Mouse Tissues and Endothelial Cell
Lines-We examined the expression of VE-cad-2 in different mouse tissues and cell lines by Northern blot analysis and immunofluorescence staining. As hybridization probe, we used a cDNA stretch (from nucleotide 2100 to 3868) corresponding to the cytoplasmic tail of VE-cad-2 to avoid aspecific hybridization with other cadherin mRNAs. A single band of about 7-kb mRNA was detected in highly vascularized organs such as lung, heart, liver, and kidney. VE-cad-2 mRNA was barely detectable in brain and thymus ( Fig. 2A). VE-cad-2 mRNA was found in four endothelial cell lines of different origin but not in other cell types such as L929 and 3T3 fibroblasts or in cultured epithelial cells (PDV) (Fig. 2B). We then performed immunofluorescence staining of cultured endothelial cells using an affinity-purified rabbit polyclonal antibody directed to a recombinant fragment of VE-cad-2, corresponding to the extracellular region between amino acid residues 74 -335. In preliminary experiments, we found that this antibody did not recognize mouse VE-, N-, or E-cadherin transfectant cells by flow cytometry, thus excluding the possibility of crossreactivity with other cadherins potentially expressed in endothelial cells (1). The antiserum recognized a band of 160 kDa in H5V cells and in VE-cad-2 transfectants (Fig. 3B).
The antiserum concentrated at the intercellular contacts. VE-cad-2 staining was similar to that of other junctional proteins such as VE-cad (see Fig. 5, A and B, respectively).
Characterization of VE-cad-2 Transfectants-In order to investigate the functional properties of VE-cad-2, its cDNA was subcloned into the pECE expression vector and transfected in CHO cells. Successful transfection and selection were determined by Northern blot and Western blot analysis of VE-cad-2 transfectants (Fig. 3, A and B).
As shown in Fig. 3A, a major band of 4.5 kb was evident in three VE-cad-2 transfectant clones but was absent in CHO control cells (CHO cells transfected with the neomycin resistance gene and the empty pECE plasmid). The size of mRNA observed corresponds to the coding cDNA sequence introduced in the pECE expression vector.
Expression of VE-cad-2 protein was analyzed by cell surface biotinylation. As reported in Fig. 3B, a band of 160 kDa was detected in transfectants and in endothelial cells, while no detectable VE-cad-2 expression was found in control transfectants. This size is larger than the molecular weight predictable from the deduced amino acid sequence. This discrepancy is commonly found in members of the cadherin superfamily and is probably due to the state of glycosylation and specific structural properties of cadherins.
To investigate the possibility that, like other members of the cadherin family, VE-cad-2 could associate with catenins through the cytoplasmic tail, cell extracts of VE-cad-2 transfectants were immunoprecipitated with the VE-cad-2 antibody and then blotted with anti-VE-cad-2 and anti-␣-and ␤-catenin antibodies (Fig. 4A). No specific bands corresponding to catenins could be detected in VE-cad-2 transfectant cells. As a control, in VE-cad transfectants, VE-cad immunoprecipitates contained ␣and ␤catenins. Similar results were obtained when H5V endothelial cells were used (data not shown).
Analysis of solubilization of a membrane protein in detergent is considered an indirect indication of their association with the cytoskeleton. Western immunoblot of the Triton X-100 soluble and insoluble fractions was carried out with VE-cad-2 antibody (Fig. 4B). The majority of the protein was found in the Triton X-100-soluble fraction. In contrast, for comparison, in VE-cad transfectants VE-cad was equally distributed in the two fractions. These data suggest that VE-cad-2 is not tightly associated with the cytoskeleton.
By immunofluorescence staining, the VE-cad-2 antiserum decorated cell-to-cell contacts (Fig. 5C). In a way similar to endothelial cells, VE-cad-2 distributed along the intercellular junctions with irregular intensity. As previously reported, in VE-cad transfectants, catenins codistributed with VE-cad at intercellular junctions (see also Fig. 5, D and E). Consistent with immunoprecipitation data, in VE-cad-2 cells ␣and ␤-catenins were not detected at intercellular contacts (see ␣-catenin as an example; Fig. 5D).
Similarly to VE-cad (26), VE-cad-2 staining at intercellular contacts was Ca 2ϩ -dependent, since treatment with EGTA caused dispersion of the molecule from intercellular contacts (Fig. 5, G and H). The integrity of actin cytoskeleton was not required for maintaining VE-cad-2 clustered at junctions, since cell treatment with cytochalasin D only slightly modified VEcad-2 staining (Fig. 5, I and L).
Functional Properties of VE-cad-2: Comparison with VEcad-VE-cad-2 has five cadherin-specific motifs in its extracel-  CHO-VE-cad-2 and CHO-VE-cad cells were extracted and immunoprecipitated with antibodies to VE-cad-2 and to VE-cad, respectively. The immunoprecipitates were separated by SDS electrophoresis, transferred to nitrocellulose by Western blotting, and reacted with antibodies to VE-cad-2, VE-cad, and ␣and ␤-catenins. A band of 160 kDa corresponding to VE-cad-2 and a band of 130 kDa corresponding to VE-cad were detected probing with antibodies to VE-cad-2 or VE-cad, respectively. Two bands of 100 and 96 kDa corresponding to ␣and ␤-catenins, respectively, were detected in VE-cad immunoprecipitates, while no signal for these proteins was detected in the immunoprecipitate of VE-cad-2. B, partition of VE-cad-2 and of VE-cad between the detergent-soluble and detergent-insoluble fractions. Confluent CHO-VEcad-2 and CHO-VE-cad cells were separated into Triton X-100-soluble and -insoluble fractions, as described under "Materials and Methods." After protein separation by SDS-PAGE electrophoresis, the fractions were probed with antibodies to VE-cad-2 or VE-cad, respectively. Under these conditions, the majority of VE-cad-2 was found in the Triton X-100-soluble fraction. For comparison in VE-cad transfectants, VE-cad was equally distributed in the two fractions. We used 5 ϫ 10 6 cells for each immunoprecipitate and soluble/insoluble Triton X-100 fractions. The positions of molecular mass markers are shown on the right. lular region with putative internal Ca 2ϩ binding sequences but, as described above, lacks the capacity to bind catenins. Since binding to catenins may influence cadherin functional behavior, we studied the adhesive properties of VE-cad-2 transfectants in comparison with VE-cad.
Transfection of CHO cells with VE-cad-2 confers calcium-dependent aggregating activity (Fig. 6, A and B) in a way comparable with, or slightly more effective than, transfection with VE-cad. Aggregation was blocked by the addition of EGTA, while cell treatment with cytochalasin D (in order to disrupt actin cytoskeleton) did not affect both VE-cad-2 and VE-cadinduced aggregation. Similar data were obtained when cells were transfected with the truncated VE-cad mutant lacking the cytoplasmic domain responsible for catenin binding (26).
To determine whether the aggregation between VE-cad-2 transfectants was homophilic, we performed aggregation assays mixing VE-cad-2 transfectants with either VE-cad or control CHO cells. Aggregation assays were performed by labeling VE-cad-2 with a fluorescent vital dye to distinguish between the two cell types (26). Aggregates were exclusively formed by cells of the same cell type (either VE-cad-2 or VE-cad transfectants), and mixed aggregates were not observed. CHO control cells formed aggregates neither with themselves nor with VEcad-2 transfectants (data not shown).
To further confirm the homophilic properties of VE-cad-2, we seeded a suspension of VE-cad-2 transfectants on monolayers of VE-cad-2, VE-cad, or control CHO cells. As shown in Fig. 6C, FIG. 5. Immunofluorescence localization of VE-cad-2 and VEcad in endothelial cells and immunofluorescence staining of VE-cad-2 and ␣-catenin in CHO transfected cells. The H5V endothelioma cell line was grown to confluence on glass coverslips, fixed, and immunostained with polyclonal antibody to VE-cad-2 (A) or mAb TEA 1.31 to VE-cad (B). All of the antisera concentrated at the intercellular contacts of endothelial cells. Some diffuse staining on the cell surface was apparent with the antisera. CHO-VE-cad-2 and CHO-VE-cad cells were grown to confluence on glass coverslips, fixed, and immunostained with polyclonal antibodies to VE-cad-2 (C) or to ␣-catenin (D and F) or with mAb TEA 1.31 to VE-cad (E). VE-cad-2 was localized at intercellular contacts in CHO transfected cells (C); the staining at junctions was irregular and different from the fine and continuous distribution of VE-cad antibodies (E). ␣-Catenin codistributed at cell-to-cell junctions only in CHO-VE-cad cells (F), whereas it retained a diffuse distribution in CHO-VE-cad-2 cells (D). Cells grown to confluence were treated with EGTA (5 mM) for 30 min (G and H) or with cytochalasin D (1 g/ml) for 30 min (I and L), before fixation and immunofluorescence staining. VE-cad-2 transfectants were double-labeled with the polyclonal antibody to VE-cad-2 followed by a secondary rhodamine-labeled anti-rabbit antibody (G and I) and fluorescein-phalloidin for actin staining (H and L). After EGTA treatment, VE-cad-2 showed a disperse localization on the cell surface (G) without a significant cell retraction as seen by actin staining (H). After cytochalasin D treatment, actin cytoskeleton was disrupted (L), but VE-cad-2 maintained in large part its localization at intercellular contacts (I). Bar, 30 m. VE-cad-2 transfectants adhered more efficiently to VE-cad-2 than to control and VE-cad transfectants. As previously reported (15), VE-cad transfectants adhered in an homophilic way to VE-cad transfectant monolayers but not to VE-cad-2 cells or to controls. Control CHO cells adhered poorly to any cell type tested.
We than tested the capacity of VE-cad-2 transfectants to detach from the neighboring cells and migrate onto a "wounded" area (15). As reported in Fig. 7A, VE-cad-2 transfection did not reduce the migratory capacity of the cells, but, as expected (15), this property was markedly inhibited by VE-cad transfection.
Furthermore, as shown in Fig. 7B, VE-cad-2 transfectant cells seeded on Transwell filters were unable to limit the passage of horseradish peroxidase conjugated to goat immunoglobulins, while VE-cad transfectants reduced this parameter by more than 60%.
Finally, it has been reported that VE-cad can induce contact inhibition of cell growth in CHO cells (27). Analysis of VE-cad-2 transfectant growth curves showed that the expression of this protein did not significantly modify this parameter (Fig. 7C). DISCUSSION In this paper, we describe the cloning and expression of a new member of the cadherin superfamily. Because of its expression in endothelial cells and its homology with the previously characterized cadherins, we propose for this protein the name VE (vascular endothelial) cadherin-2 (VE-cad-2).
The extracellular domain of VE-cad-2 contains five extracellular repeats and five copies of the four and five amino acid residues LDRE and DXNDN, which are typically repeated three or four times in cadherin (33,34). The spacing of these sequences in this new cadherin is similar to that observed in other cadherins (34).
However, the cytoplasmic domain is not homologous to cadherins or to any other sequence in the available data bases. Within the cadherin superfamily, members of the protocadherin group have similar characteristics. All protocadherins have an extracellular region homologous to cadherins but a variable intracellular domain (35). In VE-cad-2, this region is not homologous to any known protocadherin.
We have been unable to detect VE-cad-2 association with catenins, which are typically bound to the cytoplasmic domain of classic cadherins (36 -38). These proteins are responsible for cadherin binding to the actin cytoskeleton and possibly for intracellular signaling (39). Consistent with this observation, VE-cad-2 is in large part associated with the detergent-soluble fraction of cell extracts. However, we cannot exclude the possibility that catenins might interact weakly with VE-cad-2, even if such association might be undetectable in our experimental system. VE-cad-2 displays functional properties similar to those described in VE-cad and other cadherins. It clusters at intercellular junctions in endothelial cells and transfectants and promotes homotypic aggregation and adhesion. These properties are Ca 2ϩ -dependent and, as expected (26), do not require an intact actin cytoskeleton, since cytochalasin D is ineffective.
In contrast, VE-cad-2 is unable to promote other activities previously ascribed to VE-cadherin, such as reduction of paracellular permeability (15), inhibition of cell migration from a confluent monolayer (15), and contact inhibition of cell growth (15).
Interestingly, the functional behavior of VE-cad-2 is similar to that of a truncated VE-cad mutant lacking the cytoplasmic region responsible for binding to catenins (26). Like VE-cad-2, the truncated VE-cadherin promotes homotypic clustering and aggregation but cannot significantly affect cell migration, permeability, and growth (26).
These data are consistent with the idea that the cadherinlike extracellular region has the structural requirements for promoting homophilic cell recognition but that the functional consequences of this interaction are related to the association with cytoplasmic partners. Cadherin association with catenins and actin is required for strengthening adhesion (40,41) and for transferring signals that limit cell growth and motility (26). Different cytoplasmic structures may be responsible for different cellular responses.
We still do not have direct information about the intracellular partners of VE-cad-2. Immunoprecipitation analysis of metabolically labeled cells showed that the VE-cad-2 antiserum immunoprecipitates two major bands of about 100 and 200 FIG. 7. Effect of VE-cad-2 transfection on cell monolayer permeability, wound repair, and cell growth. A, time course of the migration distance covered by control (Ⅺ), VE-cad-2 (छ), and VE-cad transfectants (E). Layers of CHO control, CHO-VE-cad-2, or CHO-VEcad cells were wounded, and migration of cells out of the wounded edge was observed at different times (15). Migration values for VE-cad-2 transfectants were comparable with those for control cells. Migration values of VE-cad transfectants were, at all times, statistically different (p Ͻ 0.01 by Duncan test) from the corresponding values of control cells. B, confluent cell monolayers grown on Transwell filters were tested for permeability to horseradish peroxidase conjugated to goat immunoglobulins (measured as OD). Monolayers of VE-cad-2 transfectants did not significantly reduce permeation of peroxidase in comparison with control cells. Transfection of VE-cad reduced CHO permeability, and this effect was lost after treatment of cells with EGTA (5 mM). Open bar, filter alone; filled bar, CHO control; striped bar, CHO VE-cad-2; dashed bar, CHO VE-cad. C, cell growth of CHO control (Ⅺ), CHO-VE-cad-2 (f), and CHO-VE-cad cells (E) was estimated as reported under "Materials and Methods." Transfection of VE-cad-2 did not significantly change the cell proliferation capacity. In contrast, transfection of VEcad significantly (p Ͻ 0.01 Duncan test versus control cells) inhibits cell growth.
kDa. 2 We are currently trying to identify these structures.
The other members of the protocadherin group have variable cytoplasmic tails that do not bind catenins and do not associate with the actin cytoskeleton. Another protocadherin (protocadherin 2, Pcdh-2) was found to coprecipitate two bands of different molecular mass (180 and 50 kDa) than those found in association with VE-cad-2 (32). This suggests that different protocadherins may link different cytoplasmic proteins.
The expanding list of members of the cadherin superfamily shows that several of them display an homologous extracellular domain but divergent intracellular tails. For instance, LI-cadherin has a short tail (42,43), and T-cadherin lacks the cytoplasmic region and is linked to the membrane by a glycosyl phosphatidylinositol anchor (44). Desmogleins and desmocollins have an intracellular tail different from that of classic cadherins; they do not bind catenins, and they associate with intermediate filaments rather than actin (45).
The presence of this new cadherin in endothelial cells indicates that the organization of intercellular junctions is complex and that several proteins may cooperate to promote homotypic interaction and signaling (3).
It is possible that some of these proteins play a relevant role in the control of vascular permeability, while others are more important for intercellular signaling or transport functions. Future studies will help us to understand their reciprocal interaction and specific biological relevance.