Molecular Cloning and Characterization of a Novel Human Classic Cadherin Homologous with Mouse Muscle Cadherin*

We used a novel cDNA cloning method based on the cadherin- b -catenin protein interaction and identified a new human classic-type cadherin, which we named cad-herin-15, from adult brain and skeletal muscle cDNA libraries. Sequence analysis revealed that this cadherin was closely related to mouse muscle cadherin and seemed to be its human counterpart. However, its deduced amino acid sequence differed from that of mouse muscle cadherin in that it had an extra 31-amino acid sequence at its C terminus that has been found neither in mouse muscle cadherin nor in any other known classic cadherin. Analysis of cadherin-15 protein expressed in L fibroblasts showed that it was cleaved proteolyti-cally, expressed on the cell surfaces as a mature form of about 124-kDa, and functioned as a cell-cell adhesion molecule in a homophilic and specific manner, but Ca 2 1 did not protect it against degradation by trypsin. Our findings also suggest that cadherin-15 mediates cell-cell adhesion with a binding strength comparable to that of E-cadherin.

and is indispensable for association with catenins, the ensuing linkage to the cytoskeleton and full functioning as a Ca 2ϩ -dependent cell-cell adhesion molecule (13)(14)(15). It is now understood that these cadherins play essential roles in various morphogenetic events in multicellular organisms (1). The first classic cadherins to be identified were E-, N-, and P-cadherins, as a result of the establishment of their respective blocking antibodies (16 -23), and then V-cadherin was identified using a blocking monoclonal antibody (24). Thereafter, several cadherins were identified (25-28) by determining their cross-reactivities with antibodies raised against conserved peptide sequences or cross-hybridization with cDNA fragments of known cadherins. Over the past few years, the existence of more classic cadherin molecules has been demonstrated by PCR-based cDNA cloning methods (3, 29 -34).
Full cDNA cloning of nine independent human classic cadherin molecules has been reported, i.e. E-, N-, and P-cadherins and cadherin-4, -5, -6, -8, -11 (OB-cadherin), and -12 (3,29,30,31,(35)(36)(37). We are interested in how many classic cadherin molecules actually exist in humans and how these molecules function and cooperate in the development and maintenance of the integrity of the human body, as well as in various pathogenic states such as cancers. In an attempt to find novel human classic cadherin molecules that have not been identified by the aforementioned methods, we devised a new strategy based on the cadherin-catenin interaction. Our technique is a protein interaction cloning method using ␤-catenin, which binds directly to the cytoplasmic domains of classic cadherins (13,38). By using this method, we found two novel human classic cadherins, which we named cadherin-14 and -15. The former is a novel type II classic cadherin expressed widely in the central nervous system, and recently, we reported its cDNA cloning (39), and the latter closely resembles mouse M (muscle)-cadherin and appears to be its human counterpart. However, this molecule has an additional peptide sequence at its C terminus not possessed by M-cadherin. In this report, we describe molecular cloning and functional and biochemical analyses of this human cadherin-15 molecule.

EXPERIMENTAL PROCEDURES
cDNA Cloning-To use ␤-catenin as a probe for the first cDNA cloning procedure based on the protein-protein interaction, we constructed and purified ␤-catenin-glutathione S-transferase fusion protein, as described previously (39). This protein was radiolabeled in vitro with [␥-32 P]ATP using bovine heart kinase and was used to screen a human adult brain gt11 expression cDNA library (CLONTECH). Positive clones were plaque-purified and sequenced.
To isolate a cadherin-15 cDNA covering the entire open reading frame, we screened a human adult skeletal muscle gt10 cDNA library using a [␥-32 P]dCTP-labeled PCR probe corresponding to the 227-bp nucleotide sequence at the 5Ј-end of the cDNA clone yielded by the first cloning procedure. Positive clones were purified by several rounds of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) D83542.
rescreening and subjected to the following sequence analysis.
DNA Sequence Analysis-The cDNA inserts were excised from the purified phage DNAs by EcoRI digestion and subcloned into the EcoRI site of the pBluescript II SK(Ϫ) or (ϩ) phagemid vector, and overlapping subclones were prepared by the stepwise deletion method (40). The cDNA sequences on both strands were determined by an ABI PRISM 377 DNA sequencer (Perkin-Elmer) using a Dye Primer Cycle Sequencing Kit (Perkin-Elmer). To identify the 5Ј-end of cadherin-15 mRNA, 5Ј RACE (41) was performed using human adult skeletal muscle poly(A) ϩ RNA and the 5Ј RACE System (Life Technologies, Inc.), according to the manufacturer's instructions. The nucleotide and amino acid sequences were analyzed using the GeneWorks software package (IntelliGenetics) and the BLAST and FASTA programs.
RNA Blot Analysis-Poly(A) ϩ RNAs of cultured cells were purified using the QuickPrep Micro mRNA Purification Kit (Amersham Pharmacia Biotech), and poly(A) ϩ RNAs of normal human tissue were prepared in the same manner from specimens obtained during surgery or autopsy or purchased from CLONTECH. RNA blottings were performed as described previously (35). Premade filters (human brain multiple tissue Northern blots II and III and human muscle multiple tissue Northern blot) purchased from CLONTECH were also used. To avoid cross-hybridization with cadherins of other subclasses, a 143-bp nucleotide sequence (positions 1669 -1811 in Fig. 1) located within the EC5, where homologies with other cadherin subclasses are below 30% (Table  I), was chosen, amplified by a PCR-labeling procedure (42), and used as a probe.
Expression Vector Construction and Transfection-To express human cadherin-15 in mouse L fibroblasts, an expression vector, pBAT15H, was constructed by replacing the mouse E-cadherin cDNA of pBATEM2 (43) with a cDNA fragment of cadherin-15 containing the entire open reading frame. Transfection of pBAT15H into L cells was performed using LipofectAMINE reagent (Life Technologies, Inc.) together with pSTneoB carring the neomycin resistance gene (44), according to the manufacturer's instructions. The transfected cells were selected in DMEM supplemented with 10% calf serum in the presence of 400 g/ml G418 in a humidified atmosphere comprising 5% CO 2 , 95% air at 37°C for about 2 weeks. Then, the G418-resistant colonies were isolated, screened for cadherin-15 expression by RNA blotting, and maintained under the above condition. Mouse E-cadherin transfectants were also obtained using pBATEM2 and pSTneoB and used as controls. Mouse L fibroblasts (LTKϪ) were supplied by the Riken GenBank.
Immunoprecipitation and N-terminal Peptide Sequence Determination-Cells were lysed in 1% Nonidet P-40, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml leupeptin, 100 kallikrein inhibitor units/ml aprotinin, 10 mM HEPES (pH 7.4), 150 mM NaCl, 2 mM CaCl 2 with gentle pipetting on ice. The cell lysate was clarified by centrifugation at 6,200 ϫ g at 4°C for 10 min twice and preadsorbed with an anti-mouse IgG antibody coupled to Sepharose 4B (Organon Teknika Corporation) at 4°C for 1 h followed by removal of the beads by centrifugation and passing through a filter. The resulting lysate was incubated with an anti-␤-catenin monoclonal antibody (Transduction Laboratories) at 4°C for 1 h and then with the anti-mouse IgG antibody coupled to Sepharose 4B at 4°C for 1 h. The beads were washed five times with the lysis buffer and then three times with distilled water. The bound materials were eluted from the beads with 1 M acetic acid, lyophilized, redissolved in Laemmli's sample buffer (45), heat-denatured, separated by 7.5% SDS-PAGE, electroblotted onto PVDF membranes (Millipore), and visualized by staining with Coomassie Brilliant Blue R-250 (Sigma) or AuroDye forte (Amersham Pharmacia Biotech). The cadherin-15 bands were cut out from some membranes, and the N-terminal amino acid sequence was determined using the HP G1005 Protein Sequencing System (Hewlett-Packard).
Cell Aggregation-Short term cell aggregation experiments were performed as described previously (14) with some minor modifications. Briefly, dispersed cell suspensions were obtained by treating cells with HCMF (HCMF, 10 mM HEPES-buffered Ca 2ϩ ,Mg 2ϩ -free Hanks' solution) containing 0.01% trypsin and 5 mM CaCl 2 at 37°C for 15 min at 80 rpm. Fifty thousand cells suspended in 0.5 ml of HCMF with or without 5 mM CaCl 2 containing 1% bovine serum albumin were placed in each well of a 24-well plastic plate (Ultra Low Cluster, Costar) and allowed to aggregate at 37°C for 60 min at 80 rpm. The extent of cell aggregation was represented by the index (n 0 Ϫ n 60 )/n 0 , where n 60 and n 0 were the total numbers of particles after incubation for 60 min and at the start of incubation, respectively.
Long term cell aggregation experiments were performed as follows. Single cell suspensions were obtained by treating cells with phosphatebuffered saline containing 0.05% trypsin and 0.02% EDTA at 37°C for 15 min, washed twice with DMEM supplemented with 10% calf serum, and resuspended in DMEM supplemented with 10% calf serum and 70 units/ml DNase I (Takara) at a cell density of 2 ϫ 10 5 cells/ml. Onehundred thousand cells (0.5 ml) were placed in each well of a 24-well plastic plate (Ultra Low Cluster, Costar) and allowed to aggregate at 37°C for 24 h at 100 rpm in a humidified atmosphere comprising 5% CO 2 , 95% air. For mixed cell aggregation experiments using two cell lines, one line was labeled with 40 g/ml DiI (Molecular Probes) in DMEM supplemented with 10% calf serum for 1 h, and the other was unlabeled; the cells were suspended, as described above, and equal numbers of cells of the two lines were mixed and allowed to aggregate for 12 h, as described above.
Other Biochemical Procedures-Determination of the trypsin sensitivity and detergent solubility of cadherin-15 and immunoblot analysis were performed as described previously (14). The ECCD-2 monoclonal antibody (46) and a commercially available polyclonal anti-mouse Mcadherin antibody (M-cadherin (1), Santa Cruz Biotechnology) were used to detect mouse E-cadherin and cadherin-15, respectively, by immunoblotting. Exposition of cadherin-15 molecules on the surfaces of cadherin-15-transfected L cells was examined by labeling membrane proteins of the transfectants with a membrane-impermeable reagent, EZ-Link Sulfo-NHS-Biotin (Pierce), as described by Nelson and coworkers (47), immunoprecipitating the biotinylated cadherin-15 proteins with catenins using an anti-␤-catenin monoclonal antibody (Transduction Laboratories), separation by 7.5% SDS-PAGE, and transfer to PVDF membranes, as described above. Then, the immunoprecipitates were stained with AuroDye forte (Amersham Pharmacia Biotech) to detect all the components or with diaminobenzidine using the avidin-biotin-peroxidase complex (Elite ABC, Vector Laboratories) to detect components exposed on the external cell surfaces.

Molecular Cloning of Human Cadherin-15-To identify
novel classic cadherin molecules that associate with ␤-catenin, a human adult brain gt11 expression cDNA library was screened with a radiolabeled ␤-catenin fusion protein. Approximately 10 6 recombinants were screened, and 23 positive clones were isolated. DNA sequence analysis of these clones disclosed 11 clones of N-cadherin, 1 of cadherin-11, 8 of adenomatous polyposis coli tumor suppressor protein, which also associates with ␤-catenin (48,49), and 3 of novel proteins, 2 of which showed high sequence similarities to each other and to known cadherin molecules. One of the two cadherin-related molecules was named cadherin-14 and was reported recently by our group (39), and the other, designated cadherin-15, was analyzed in detail in this study. This clone contained a 1656-bp cDNA insert, which showed the highest homology with mouse Mcadherin (30). However, comparison of its nucleotide sequence with that of mouse M-cadherin revealed that it lacked the part encoding the translation initiation codon, signal peptide, precursor region, and EC1-3.
As preliminary RNA blot analysis showed that this clone was expressed strongly in skeletal muscle (data not shown), a human adult skeletal muscle library was screened using a PCR probe located at the 5Ј-end of the cDNA clone yielded by the first cloning procedure. About 1.7 ϫ 10 5 phages were screened, and five positive clones were isolated. A clone containing the longest cDNA insert was selected, and the cDNA was subjected to sequence analysis. This clone comprised 2833 bp with a poly(A) tail and covered the entire open reading frame. To identify the 5Ј-end of the full mRNA, 5Ј RACE was performed, and an additional 24-nucleotide sequence was obtained. The combined nucleotide and deduced amino acid sequences are shown in Fig. 1. The open reading frame begins with an ATG codon at positions 78 -80, terminates at a TGA codon at positions 2520 -2522, and consists of 2442 nucleotides encoding 814 amino acids. The nucleotide sequence of the former cDNA clone isolated from a brain library differed from this sequence (Fig. 1) at three points as follows: A at position 1827, A at position 2048, and T at position 2818 were replaced by C, G, and G, respectively. These replacements altered Lys at amino acid position 584 to Gln in the amino acid sequence but did not affect the open reading frame.
Sequence Analysis of Cadherin-15-The deduced amino acid sequence consists of two hydrophobic sequences corresponding to the signal peptide and transmembrane domain, a long extracellular domain containing five cadherin extracellular subdomain repeats, and a relatively short cytoplasmic domain, which are structural characteristics of the classic cadherins. The cleavage site of the signal peptide was deduced according Nielsen et al. (50). We expected this protein to undergo further proteolytic cleavage at amino acid positions 45-46 (51) and to be expressed on cell surfaces as a mature and functional protein, 769 amino acids long with 4 consensus sites for N-linked glycosylation (52). The proteolytic cleavage and exposition on the cell surfaces were confirmed, as described below.
A search for homologies with known classic cadherins revealed that human cadherin-15 resembled mouse M-cadherin most closely, and the homologies of cadherin-15 with mouse M-cadherin and the other known human classic cadherins are summarized in Table I. The putative mature protein of human cadherin-15 shows 83% homology with that of mouse M-cadherin but much lower homologies with those of the human classic cadherins reported so far, suggesting that cadherin-15 is a human counterpart of mouse M-cadherin. However, alignment of human cadherin-15 and mouse M-cadherin revealed marked differences between the two molecules. Cadherin-15 has an additional 31-amino acid stretch at its C terminus, which has not been found in any other classic cadherin. The fact that both the cDNA clones isolated from brain and skeletal muscle libraries encode this sequence indicates that this is not due to a cloning artifact or individual variations.
Cadherin-15 Expression in Human Tissues- Fig. 2 shows cadherin-15 expression in various normal human tissues. Skeletal muscle showed intense expression of a transcript of about 2.9 kb and faint expression of two of about 5.8 and 6.7 kb. Expression of cadherin-15 transcripts was faint in the brain and very faint in the placenta, prostate, spinal cord, and thyroid. Next, we performed RNA blotting of several muscle and brain tissues (Fig. 3). As shown in Fig. 3A, of the muscle tissues examined only skeletal muscle expressed cadherin-15; no cadherin-15 transcripts were detected in smooth or cardiac muscle. In various brain sections, cadherin-15 transcripts were detected only in the cerebellum (Fig. 3B), suggesting that the faint expression in the whole brain (lanes 2 and 13 in Figs. 2 and 3B, respectively) derived from the cerebellum.
Cadherin-15 Transfection and Expression in Mouse L Fibroblasts-To analyze the molecular nature and functional characteristics of cadherin-15, cadherin-15 cDNA placed downstream from the chicken ␤-actin promoter was introduced into L cells, which are mouse fibroblasts deficient in cadherin activity (53). Over 30 G418-resistant colonies were isolated and screened for cadherin-15 expression by RNA blotting (data not shown). In this study, a transfectant clone designated L15-1, which showed the highest level of cadherin-15 expression, was used for further analysis.
Cadherin-15 Protein Expressed in L Cells-To detect the cadherin-15 protein, the cadherin-catenin complex was immunoprecipitated from L15-1 cells with the anti-␤-catenin monoclonal antibody. Cadherin-15 protein was detected as a single band of approximately 124 kDa (Fig. 4). This band stained with Coomassie Blue was cut out and subjected to N-terminal amino acid sequencing. The sequence was Ala-Trp-Val-Ile-Pro-Pro-Ile-Ser-Val-Ser-Glu-Asn (Fig. 1), which agreed with that expected in the light of the sequences of other classic cadherins (51). Cadherin-15 molecule expression in L15-1 was also detected by immunoblotting with a polyclonal anti-mouse Mcadherin antibody, which cross-reacted weakly with human cadherin-15 (Fig. 5A). Transfection of classic cadherin cDNAs into L cells is known to be accompanied by up-regulation of catenin proteins, probably because association of catenins with cadherins retards their turnover (54). Consistent with this observation, L15-1 cells expressed far more ␤-catenin protein than the parent L cells (Fig. 5B). As a control for further functional analysis, a mouse E-cadherin-transfectant, LE-1 cells, which express similar amounts of ␤-catenin protein, was selected, and the results are also shown in Fig. 5B. Assuming that cadherin-15 and E-cadherin associate with catenins and are processed and degraded in the same manner, L15-1 cells can be considered to express virtually the same number of cadherin molecules per cell as LE-1 cells. These two transfectants expressed similar amounts of ␣-catenin protein, and they both expressed more than the parent L cells (data not shown).
Cell-Cell Binding Activity and Biochemical Properties of Cadherin-15-First, short term cell aggregation experiments were performed to examine whether cadherin-15 functions as a cell-cell adhesion molecule. Unexpectedly, L15-1 cells showed only weak Ca 2ϩ -dependent aggregation, whereas LE-1 cells aggregated strongly in the presence of Ca 2ϩ under the same conditions (Table II). There are three possible explanations for this low Ca 2ϩ -dependent aggregation rate of L15-1 cells as follows: first, cadherin-15 molecules were not exposed on the cell surfaces; second, unlike E-cadherin, cadherin-15 was not protected by Ca 2ϩ against trypsin treatment and was degraded during cell suspension preparation; and third, the cell-cell binding activity of cadherin-15 was much weaker than that of E-cadherin.
As shown in Fig. 6, cadherin-15 molecules expressed in L15-1 cells were labeled extracellularly with sulfo-NHS-biotin, as  were E-cadherin molecules. Cadherin-15 also showed a detergent solubility comparable to that of E-cadherin in LE-1 cells (Fig. 7); a considerable amount of cadherin-15 could not be extracted with Nonidet P-40. Next, we compared the trypsin sensitivities of cadherin-15 and E-cadherin. As shown in Fig. 8, E-cadherin expressed in LE-1 cells showed the characteristic resistance to trypsin treatment in the presence of Ca 2ϩ that has been documented to be a key property of the classic cadherins (54). Cadherin-15 expressed in L15-1 cells, however, was not fully protected by Ca 2ϩ against trypsin; most of the cadherin-15 appeared to be degraded by trypsin in the presence of Ca 2ϩ (Fig. 8). Interestingly, even in the absence of Ca 2ϩ , a few cadherin-15 molecules, a similar number to those after trypsin treatment in the presence of Ca 2ϩ (Fig. 8), remained intact, suggesting that the cadherin-15 molecules that were resistant to trypsin irrespective of the presence or absence of Ca 2ϩ might not have been exposed on the cell surfaces. Taking these findings together, we conclude that the majority of the cadherin-15 molecules were exposed on the external cell surfaces and linked to the cytoskeletal system via catenins, in a similar manner to E-cadherin, and that the low Ca 2ϩ -dependent aggregation rate of L15-1 cells in the short term cell aggregation experiments was attributable to the susceptibility of cadherin-15 to trypsin in the presence of Ca 2ϩ . Therefore, we conducted long term cell aggregation experi-  ments in an attempt to establish whether cadherin-15 really does act as a cell-cell adhesion molecule. The influence of trypsin treatment was considered negligible in this assay, because cadherin-15 protein expression in L15-1 cells recovered to its initial level within 3 h of the trypsin and EDTA treatment described under "Experimental Procedures" (data not shown). After incubation for 24 h, L15-1 cells formed definite aggregates almost identical in size and cell-cell adhesiveness to the LE-1 aggregates, whereas virtually no parent L cell aggregation under the same conditions was observed (Fig. 9), demonstrating that cadherin-15 really does function as a cell-cell adhesion molecule. Assuming that L15-1 and LE-1 cells express equivalent numbers of cadherin molecules per cell, as discussed above, we conclude that the cell-cell binding strengths of cadherin-15 and E-cadherin are virtually the same.
To determine the cell-cell binding specificity of cadherin-15, equal numbers of DiI-labeled L15-1 cells and unlabeled L, LE-1, or L15-1 cells were mixed and allowed to aggregate for 12 h. As can be seen in Fig. 10, L15-1 cells did not interact with the parent L cells, indicating that cadherin-15 mediates cellcell adhesion in a homophilic manner, as do the other classic cadherins (55). When L15-1 cells were mixed with LE-1 cells, each transfectant aggregated separately, and chimeric aggregates were never found (Fig. 10). In contrast, when DiI-labeled and unlabeled L15-1 cells were mixed, aggregates containing random labeled and unlabeled cells were formed (Fig. 10). L15-1 cells did not interact with P-cadherin, cadherin-6, or cadherin-14 transfectants in another series of mixed cell aggregation experiments (data not shown). These results indicate that the cell-cell binding specificity of cadherin-15 is unique and distinct from those of the other known classic cadherins. DISCUSSION Many molecules classified as cadherin superfamily members have been identified in the past few years. Full cDNA cloning of nine human classic cadherin molecules has been accomplished so far (3,29,30,31,(35)(36)(37), but exactly how many members belong to this family is unknown. We employed a novel cDNA cloning method based on the protein interaction between classic cadherins and ␤-catenin and identified two novel human classic cadherin molecules, cadherin-14 (39) and cadherin-15, from a human adult brain cDNA library. Therefore, this method is considered useful for searching for new members of the classic cadherin family as well as for unknown molecules that associate with ␤-catenin, which is a multifunctional protein involved in both the cadherin cell adhesion and receptormediated intercellular signal transduction systems (56). We have tested this method on only one library, but it is possible that applying this method to other expression cDNA libraries derived from different sources will lead to the discovery of more novel molecules that interact with ␤-catenin.
Our cDNA sequence analysis of cadherin-15 revealed that, of the known cadherins, cadherin-15 showed a very close resemblance to mouse M-cadherin, which was first identified in muscle cells and is considered to be involved in the fusion of myoblasts to myotubes (30). Mouse M-cadherin has been reported to be expressed in skeletal muscle and cerebellum but not in cardiac or smooth muscle (30,(57)(58)(59), an expression pattern in complete agreement with that of cadherin-15 that we observed. These two findings suggest strongly that cadherin-15 is a human homologue of mouse M-cadherin. However, the two molecules differ markedly, cadherin-15 has a unique 31-amino acid sequence at its C terminus that has been found neither in mouse M-cadherin nor in the other known classic cadherins. Data base searches revealed no peptide sequences similar to this sequence, and what properties its addition confers on the function and molecular nature of cadherin-15 are unknown. The linkage with catenins and the cell-cell binding function of cadherin-15, at least, did not appear to be affected by this

cells and transfectants
Cell suspensions obtained after treatment with trypsin in the presence of Ca 2ϩ were placed in a 24-well plastic plate without (Ca(Ϫ)) or with (Ca(ϩ)) 5 mM CaCl 2 and allowed to aggregate at 37°C for 60 min at 80 rpm. The extent of cell aggregation was represented by the aggregation index (n 0 -n 60 )/n 0 , where n 60 and n 0 are the total numbers of particles after incubation for 60 min and at the start of incubation, respectively. sequence, and no cytoplasmic components that interacted with cadherin-15 other than catenins were detected in repeated immunoprecipitation experiments. Molecular biological approaches, including site-directed mutagenesis, should clarify the significance of this sequence.
The classic cadherins have been proposed to be divided into two subgroups, types I and II, on the basis of their overall sequence similarities and conservation of several motifs and aromatic amino acid residues in their extracellular domains (3,29). The human classic cadherins, E-, N-, and P-cadherin and cadherin-4, have been classified as type I and cadherin-6, -8, -11, -12, and -14 as type II. Although cadherin-5 was initially reported to be a type II cadherin (29), it was later found not to resemble the type II cadherins very closely (33,39). Thus, it would appear that cadherin-5 cannot be classified as either type I or II. Cadherin-15 is more similar to type I than type II cadherins (Table I). However, it should be noted that there is an important difference between cadherin-15 and the other four type I cadherins. All the type I cadherins have the HAV tripeptide motif in EC1 which, together with its flanking amino acids, is intimately involved in the adhesive function and binding specificities of these cadherins (60 -62) but has been replaced by the FAL tripeptide at amino acid position 123-125 in cadherin-15 (Fig. 1). Incidentally, the type II cadherins cad-herin-6, -8, -11, -12, and -14 have QAI or QAD instead of the HAV motif, and cadherin-5 has VIV at the corresponding position. Cadherin-15 also exhibited a definitive biochemical difference from type I cadherins as follows: Ca 2ϩ did not protect it against trypsin, suggesting the structures of the extracellular domains of cadherin-15 and type I cadherins differ. Similar trypsin sensitivity has been reported only for cadherin-5 (63). Moreover, the LDRE motif found in EC4 of the other human classic cadherins has been replaced by LSPA in cadherin-15. Therefore, we propose that cadherin-15 and cadherin-5 cannot be classified as either type I or II cadherins.
As there is no appropriate method for quantifying cadherins at present, the binding strength of each cadherin subclass cannot be evaluated precisely. In a cDNA transfection system using L fibroblasts, which lack cadherin activity and express very little catenin at the protein level but large amounts at the RNA level, ectopic cadherin expression induces accumulation of catenin proteins, probably because they are stabilized by association with cadherins. In this study, we paid particular attention to the expression level of ␤-catenin protein, which associates directly with the cytoplasmic domain of classic cadherins (13,38), in transfectants. Assuming that any classic cadherin subclass influences the preservation of ␤-catenin protein in L cells similarly, the binding strength of each cadherin FIG. 9. Long term aggregation of L15-1 cells. L, LE-1, and L15-1 cells were trypsinized completely in the presence of EDTA to produce single cell suspensions and suspended in DMEM supplemented with 10% calf serum and 70 units/ml DNase I. One-hundred thousand cells (0.5 ml) were placed in each well of a 24-well plastic plate and allowed to aggregate at 37°C for 24 h at 100 rpm in a CO 2 incubator, and phase contrast micrographs of unfixed aggregates were taken. Scale bars, 100 m. Equal numbers of DiI-labeled L15-1 cells and unlabeled L, LE-1, or L15-1 cells were mixed in DMEM supplemented with 10% calf serum and 70 units/ml DNase I and allowed to aggregate, as described in the legend to Fig. 9, for 12 h. The resulting aggregates were fixed with formaldehyde, mounted, and photographed. The upper panels show phase contrast micrographs of individual mixed cell aggregates, and the corresponding fluorescence micrographs in the same fields are shown in the lower panels. Note that although DiI-labeled and unlabeled L15-1 cells formed randomly mixed aggregates, L15-1 cells did not form aggregates with unlabeled L or LE-1 cells (shown by arrowheads). Scale bars, 100 m. subclass can be compared using transfectants that express equal amounts of ␤-catenin protein. Therefore, we compared the aggregation of a cadherin-15 transfectant with that of an E-cadherin transfectant expressing almost the same amounts of ␤-catenin protein. The sizes and cell-cell adhesiveness of the aggregates were indistinguishable. On the assumption stated above, it is conceivable that the cell-cell binding strengths of cadherin-15 and E-cadherin are virtually equivalent. Finally, our mixed cell aggregation assays showed that cadherin-15 mediates cell-cell adhesion in a homophilic manner and exhibits cell-cell binding specificity, i.e. it did not interact with the molecules expressed on L cells, mouse E-cadherin, human Pcadherin, cadherin-6, or cadherin-14.
In conclusion, we have isolated a full-length human cadherin-15 cDNA using a novel cDNA cloning method and characterized the cadherin-15 molecule using an L fibroblast cDNA transfection system. We hope that the technique we have developed and results of this study will be useful for further investigations into cell-cell interactions.