INTRODUCTION

Polarized epithelia separate biological compartments and regulate the vectorial transport of ions and solutes. Epithelial cells are polarized into an apical and basolateral pole (1), each characterized by a distinct protein and lipid composition. The induction of epithelial cell differentiation and polarization and the molecular signals responsible for selective protein sorting to distinct membrane domains can be studied in vitro using MDCK¹ cells (2).

The organization of MDCK cells into an apical and a basolateral pole depends on extracellular matrix and cell-cell interactions (3, 4, 5). Formation of cell-cell contacts, induced through adhesive interactions mediated by E-cadherin (6), results in the differentiation of MDCK cells into a polarized epithelium and the gradual restriction of specific proteins to the apical or basolateral membrane domain (5, 6, 7, 8). E-cadherin is a classical transmembrane, calcium-dependent cadherin cell adhesion molecule, which is characterized by five extracellular structural repeats and a highly conserved cytoplasmic region. The cadherin cytoplasmic region interacts with the catenins, a group of cytoplasmic proteins that connects cadherins with the actin-based cytoskeleton (9, 10). In epithelial cells in vivo, E-cadherin is a major constituent of adherens junctions, where it mediates calcium-dependent adhesion and links cortical actin filaments between adjacent cells (7, 11).

T-cadherin is a member of the cadherin family that shares the ectodomain organization with classical cadherins and is anchored to the membrane through a GPI moiety (12, 13). In contrast to classical cadherins that generally do not mediate adhesive interactions without their conserved cytoplasmic domain (14, 15), T-cadherin induces calcium-dependent, homophilic aggregation between transfected cells (13). In the animal, T-cadherin is a component of specific cell populations both within and outside the nervous system. In the nervous system, T-cadherin demarcates specific neuron populations and axon pathways (12, 16, 17), suggesting a role in axon patterning. Outside the nervous system, T-cadherin is expressed on skeletal muscle surfaces and is specifically excluded from myoneural junctions (17), which are demarcated by N-cadherin (19). The mechanisms of how T-cadherin functions in mediating cell-cell interactions and axon guidance are not understood.

To gain insights into the principal function of T-cadherin, we have examined whether T-cadherin is localized to cell-cell junctions like classical cadherins. Two approaches were used: immunohistochemical staining of the chick intestinal epithelium in vivo and heterologous expression of T-cadherin in polarized MDCK cells in vitro. Our results revealed T-cadherin on apical cell surfaces of epithelial cells, in contrast to the basolateral localization of classical cadherins. Examination of the mechanisms responsible for differential cadherin localization identified direct targeting as well as selective removal from specific membrane domains as key principles.

EXPERIMENTAL PROCEDURES

Madin-Darby canine kidney cells, clone II/8, were obtained from Dr. James Nelson (Stanford University, Stanford, CA) (20) and maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), penicillin, and streptomycin.

Fertilized White Leghorn chicken eggs (McIntyre Poultry Farm, Lakeside, CA) were incubated in a force-draft incubator until the desired developmental stages. Embryos were dissected in phosphate-buffered saline (PBS, pH 7.2), fixed for 6 h in 4% formaldehyde in PBS, and rinsed in several changes of PBS. The tissue was cryoprotected overnight in 30% sucrose in PBS and embedded and frozen in OCT embedding medium (Miles Scientific, Elkhart, IN). Sections of 15 µm were cut, collected on chromalum-coated slides, and labeled with anti-T-cadherin antiserum as described below.

Chick T-cadherin (13) or N-cadherin pcDNA1 expression plasmids were cotransfected with the selectable marker plasmid pSV2neo into MDCK cells as described below. Chicken N-cadherin cDNA (a gift from Dr. C. Kintner, Salk Institute, La Jolla, CA) was excised from the SP72 plasmid vector (Promega, Madison, WI) with EcoRV and XbaI, and the coding region was inserted into the EcoRV- and XbaI-digested pcDNA1 expression vector (Ncad-pcDNA1; Invitrogen, La Jolla, CA). Three mutants, N-T_GPI, Ncyt, and T-Ncyt (see Fig. 5), were constructed by reverse transcription-polymerase chain reaction amplification and ligation of the respective T- and N-cadherin DNA fragments into pcDNA1. N-T_GPI, comprising N-cadherin ectodomain amino acids 1-660 (amino acid numbering according to Hatta et al. (21)) and the GPI anchor of T-cadherin (amino acids 615-690, numbering according to Ranscht and Dours-Zimmermann (12)), was generated by reverse transcription-polymerase chain reaction using N-cadherin primers 5-(NotI)ATGTGCCGGATAGCGGGAA and 3-(SalI)AATGGTCCAATTCCTCTTAAT (restriction sites underlined) and T-cadherin primers 5-(SalI)CTTAACAATACTCATGCCCAG and 3-(XbaI)CTACAGACAAAATAAACTGAA. The N-cadherin cytoplasmic domain deletion mutant Ncyt (amino acids 1-755) was generated using N-cadherin 5-(NotI)ATGTGCCGGATAGCGGGAA and 3-(XbaI)TCAACGGCGCTTCATCCATACTAC primers. Last, the T-Ncyt chimera comprising the T-cadherin ectodomain (amino acids 22 to 616) and the N-cadherin cytoplasmic tail deletion (amino acids 667-755) was generated using T-cadherin primers 5-(NotI)ATGCAGCACAAAACTCAACTT and 3-(SalI)GTTAAGCTTGTTGATTCTCCA and N-cadherin primers 5-(SalI)CATGCCCAGCTCTCTTTAAGG and 3-(XbaI)CAACGGCGCTTCATCCATACTAC. All polymerase chain reaction fragments were cloned into the pCRII vector (Invitrogen) and digested with restriction enzymes specific for the sites at their respective 5- and 3-ends. Ligation into the pcDNA1 expression plasmid (Invitrogen) was carried out in a single step reaction with one or two DNA inserts.

MDCK cells were transfected by calcium phosphate precipitation (22) with either of the above plasmids in the presence of the selectable marker plasmid pSV2neo using CellPhect (Pharmacia Biotech Inc.) according to the manufacturer's directions. Colonies were selected in the presence of G418 and enriched by fluorescence-activated cell sorting. For T- and N-cadherin co-expression, T-cadherin-expressing MDCK cells were transfected with Ncad-pcDNA1 and pPGK-hygromycin (kindly provided by R. Oshima, The Burnham Institute) and selected in G418 (0.5 mg/ml) and hygromycin (0.25 mg/ml).

MDCK cells were grown on Costar Transwell^TM filters (Costar Corp., Cambridge, MA), coated with type I collagen (25 µg/ml; Collaborative Research Inc., Bedford, MA), at densities close to saturation (2 × 10⁵ cells/cm²). After 2-3 days, cells were washed with PBS containing 1 mM MgCl₂ and 0.1 mM CaCl₂ (PBS/CM) and fixed with 2% paraformaldehyde in PBS/CM for 30 min at room temperature. The cells were incubated with 50 mM NH₄Cl in PBS/CM for 10 min at room temperature and permeabilized for 10 min with 0.075% (w/v) saponin in PBS/CM containing 0.2% bovine serum albumin. T-cadherin was detected by indirect immunofluorescence with rabbit anti-T-cadherin antiserum (1:100; Ranscht and Dours-Zimmermann (12)), followed by fluorescein-conjugated fluorescein isothiocyanate goat anti-rabbit IgG (1:500; Cappel, West Chester, PA) in PBS/CM containing 0.075% (w/v) saponin and 0.2% bovine serum albumin. E-cadherin was visualized with the mouse monoclonal antibody rr1 (1:500; Gumbiner and Simons (22); a generous gift from Dr. Barry Gumbiner, Sloan Kettering Cancer Center, New York, NY). For N-cadherin staining, cells were fixed with ethanol (20 °C) for 2 min and then labeled with mouse monoclonal anti-N-cadherin antibody GC4 (1:50; Volk and Geiger (11); Sigma). Both rr1 and GC4 were visualized with fluorescein-conjugated fluorescein isothiocyanate goat anti-mouse IgG (1:200 in PBS/CM containing 1% heat-inactivated goat serum; Antibodies, Inc., Davis, CA).

For counterstaining, nuclei were labeled with propidium iodide (1 µg/ml; Sigma) in PBS/CM for 15 min after pretreatment of the cells with RNase A (50 µg/ml; Boehringer Mannheim) for 30 min at room temperature. Vertical optical sections were obtained by laser scanning confocal microscopy (XZ sections in 0.25-µm intervals; Zeiss LSM 410 inverted laser scan microscope).

Monolayer cultures of MDCK cells, grown to confluency on polycarbonate Transwell^TM filter chambers, were fixed for 1 h with 4% paraformaldehyde and 0.2% glutaraldehyde in 0.13 M phosphate buffer and blocked for 1 h at room temperature with 5% goat serum in Tris-buffered saline (TBS; 0.1 M Tris and 77 mM NaCl, pH 7.6) containing 0.1% Photo-flo (Eastman Kodak Co.). Cells were stained overnight at 4 °C with anti-T-cadherin antiserum (1:500) in TBS containing 1% goat serum and 0.1% Photo-flo. The cultures were then labeled with goat anti-rabbit IgG (1:50; Cappel) for 1 h at room temperature, followed by rabbit peroxidase-antiperoxidase for 1 h at room temperature (1:100; Sternberger Monoclonals, Baltimore, MD). All of the immunoreagents were diluted in TBS containing 1% goat serum, and the cells were rinsed after each incubation step in three changes of TBS. Cells were reacted in a chromogen mixture of 0.05% diaminobenzidine and 0.005% hydrogen peroxide. The immunolabeled cells were then postfixed in 2% OsO₄ and 15% potassium ferricyanide in 0.1 M phosphate buffer for 1 h on ice, dehydrated with a series of ethanol, flat embedded in a TAAB/Epon (1:1) resin embedding mixture, and polymerized for 2 days at 65 °C. Ultrathin sections were cut and examined with a Hitachi 600E transmission electron microscope.

Cells from a confluent 35-mm plate were lysed for 20 min at 4 °C in 100 µl of lysis buffer (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 50 µM leupeptin, 5 µM pepstatin, and 4 µg/ml aprotinin; all protease inhibitors from Boehringer Mannheim). Insoluble material was pelleted, and 14 µl of the supernatant was analyzed by Western blotting of proteins separated by SDS-PAGE and transferred onto Immobilon polyvinylidene difluoride membranes (Millipore Intertech, Bedford, MA; Ref. 23). The blots were blocked for 1 h at room temperature with 5% fish gelatin (Sigma) in TBST (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 0.05% Tween 20) and incubated with either rabbit anti-T-cadherin (1:600), mouse anti-E-cadherin (rr1, 1:500), or rat anti-N-cadherin (NCD2, 1:200; a gift from M. Takeichi, Kyoto University, Kyoto, Japan; Ref. 21) in TBST for 1 h at room temperature. Primary antibodies were detected by chemiluminescence (ECL system, Amersham Life Science, Inc.) after incubation of the blots with horseradish peroxidase-conjugated sheep anti-rabbit, sheep anti-mouse, or sheep anti-rat IgG, respectively.

T-cadherin was removed with phosphatidylinositol-specific phospholipase C (a gift from M. G. Low, Columbia University, New York, NY) from MDCK and Cos7 cells as described previously (13). Proteins were separated by SDS-PAGE and analyzed by immunoblotting with anti-T-cadherin or anti-N-cadherin antibodies as described above.

For metabolic labeling, MDCK cell cultures grown on Transwell^TM filters were rinsed twice with PBS. The cells were incubated with DMEM/FBS without methionine for 15 min at 37 °C. [³⁵S]Methionine/cysteine (250 µCi; DuPont NEN) was added to the basolateral side of cells on the Transwell^TM filter in a total volume of 100 µl DMEM/FBS without methionine; DMEM/FBS without methionine (1 ml) was added to the apical compartment to keep the cells submerged. Metabolically labeled proteins were biotinylated for targeting assays as described below. For assays measuring cadherin removal from specific membrane domains, cells were incubated in DMEM/FBS without methionine for 15 min, labeled with [³⁵S]methionine/cysteine as above, and chased for the indicated times in cold DMEM/FBS.

Biotinylation was performed essentially as described by Rodriguez-Boulan et al. (24). Briefly, MDCK cells were plated at high density on 24-mm Costar Transwell^TM plates (as described above). After 6-8 days in culture, the permeability of the monolayer was measured by adding 0.2 µCi of [³H]inulin to the apical compartment of the filter chamber. After 2 h at 37 °C, [³H]inulin was measured in the basolateral compartment. If more than 1% of the added inulin permeated, the monolayers were discarded. Monolayers with less than 1% inulin permeability were washed with PBS/CM and biotinylated by adding 0.5 mg sulfo-NHS-biotin (stock solution, 200 mg/ml in dimethylsulfoxide; Pierce) in 1 ml of PBS/CM either to the apical or basolateral compartment of the filter chamber. Compartments not receiving sulfo-NHS-biotin were filled with an equal volume of PBS/CM. After 30 min of agitation at 4 °C, filters were washed three times with Tris-saline/phenylmethylsulfonyl fluoride (15 mM Tris, pH 7.5, 120 mM NaCl, and 1 mM phenylmethylsulfonyl fluoride) and extracted with 100 µl of lysis buffer (1% Nonidet P-40, 60 mM octylglucoside, 10 mM Tris, pH 8.0, 150 mM NaCl, 2 mM CaCl_2, 1 mM phenylmethylsulfonyl fluoride, 2 µg/ml pepstatin, 50 µM leupeptin, and 4 µg/ml aprotinin). The cells were scraped from the filter with a rubber policeman and sedimented in a microfuge for 5 min.

Immunoprecipitation was essentially performed as described by Wollner et al. (8). Samples were analyzed by SDS-PAGE, and gels were processed for fluorography using Amplify (Amersham). Protein bands were quantitated by scanning densitometry (Pharmacia).

RESULTS

Both E-cadherin and LI-cadherin are concentrated on the basolateral cell surface of intestinal epithelial cells, E-cadherin in adherens junctions (25, 26) and LI-cadherin in lateral contact zones excluding junctional regions (27). To test the localization of T-cadherin, sections of intestine from chick embryos just before hatching were immunohistochemically labeled for T-cadherin expression. T-cadherin was localized on the apical surface of chick intestinal epithelium and displayed the most intense staining signal in the villus, in areas where enterocytes approach the apical extrusion zone of the villus (Fig. 1). Little staining was detected on epithelial cells of the crypt and on cells close to the crypt-villus junction. Thus, in contrast to the basolateral localization of E-cadherin and LI-cadherin, T-cadherin is localized on the apical surface of epithelial cells in vivo.

To address whether heterologously expressed, GPI-anchored T-cadherin and classical cadherins are differentially targeted in polarized epithelial cells in vitro, MDCK cells were stably transfected with either chicken T-cadherin (13) or N-cadherin cDNA expression plasmids and pSV2 neo. N-cadherin was chosen because MDCK cells express endogenous E-cadherin (28, 29). Cell lines expressing either cadherin were selected from neomycin-resistant colonies, enriched by fluorescence-activated cell sorting, and grown to confluency on collagen-coated Transwell^TM polycarbonate filters. Cadherin distribution was examined by confocal microscopy following indirect immunostaining with rabbit anti-chick T-cadherin antiserum (12) and monoclonal anti-N-cadherin antibody GC4. In optical sections in the vertical plane (XZ section), T-cadherin was only apparent on apical MDCK cell surfaces (Fig. 2A). In contrast, endogenously expressed E-cadherin, detected with monoclonal anti-canine E-cadherin antibody rr1 (22), was expressed at the lateral surface both in wild type MDCK cells as described previously (28, 29; not shown) and in T-cadherin-transfected cells (Fig. 2B). Exclusive T-cadherin localization on the apical cell surface was confirmed by immuno-electron microscopy after immuno-peroxidase staining (Fig. 2C).

To verify that the anticipated GPI-anchored T-cadherin protein is expressed in MDCK cells, transfected cells were treated with phosphatidylinositol-specific phospholipase C (PI-PLC), an enzyme that specifically cleaves GPI-linked anchors (30). PI-PLC released a large fraction of T-cadherin from the cell surface into the culture medium (Fig. 2D). Some proportion of T-cadherin was resistant to PI-PLC treatment, possibly due to incomplete digestion. In controls without enzyme, T-cadherin remained associated with cell membranes.

T- and N-cadherin are expressed in an overlapping pattern on cultured sympathetic neurons and immature muscle cells (17, 31), raising the possibility that one cadherin affects the localization of the other at specific membrane domains. To test this possibility, MDCK cells were double transfected to stably express T- and N-cadherin protein. Both 95-kDa mature T-cadherin and 120-kDa N-cadherin were detected by Western blotting (not shown). As in the singly transfected cells, T-cadherin was localized to the apical surface (Fig. 3A), whereas N-cadherin was detected basolaterally (Fig. 3B). Apical T-cadherin and basolateral N-cadherin expression was already prominent on subconfluent groups of nonpolarized MDCK cells, which are much flatter in morphology (not shown). As the distribution of T- and N-cadherin in the double transfected cells was indistinguishable from that in the single transfected cells, these experiments provide evidence that the localization of these cadherins is independent of each other.

The localization of T-cadherin was biochemically quantitated by immunoprecipation of surface biotinylated proteins from either the apical or basolateral membrane domain. MDCK cell monolayers with <1% permeability for the [³H]inulin tracer were used for these experiments. Confluent monolayers of T-cadherin-transfected cells were labeled with sulfo-NHS-biotin (a tight junction-impermeable probe) from either the apical or the basolateral side of the filter compartment and extracted with lysis buffer containing 1% Nonidet P-40. As GPI-linked proteins are insoluble in nonionic detergents such as Nonidet P-40 (32, 33, 34, 35), T-cadherin was immunoprecipitated from both Nonidet P-40-soluble and Nonidet P-40-insoluble fractions. Precipitated proteins were separated by SDS-polyacrylamide gel electrophoresis, and biotinylated T-cadherin was detected with peroxidase-coupled streptavidin on immunoblots (Fig. 4). T-cadherin was predominantly (93%) localized on the apical cell surface (Fig. 4), and approximately 50% of the protein was soluble in Nonidet P-40 extracts (Fig. 4, lane 1), whereas the remainder was resistant to nonionic detergent extraction (Fig. 4, lane 3). A small percentage of T-cadherin (7%) was localized basolaterally, probably due to misrouting. Thus, both immunohistochemical and biochemical analyses localize the majority of T-cadherin at the apical MDCK cell surface.

Wild Type N-Cadherin and Cytoplasmic Tail Deletion Mutant Ncyt Are Localized at the Basolateral Domain in MDCK Cells

classical cadherins contain in their cytoplasmic domain signals for basolateral sorting (50). To investigate whether deletion of these signals is sufficient to shuttle cadherins apically like cytoplasmic tail deletion mutants of the low density lipoprotein receptor (37), we have generated and tested for localization and targeting an N-cadherin cytoplasmic tail deletion mutant, Ncyt (Fig. 5). The Ncyt protein lacks all but three amino acids (KRR) of the N-cadherin cytoplasmic domain. The localization of the Ncyt mutant was analyzed by confocal microscopy of confluent, stably transfected MDCK cell monolayers stained by indirect immunofluorescence. Optical sections in the XZ axis revealed the majority of the Ncyt protein on basolateral MDCK cell surfaces (Fig. 6A). To biochemically quantitate Ncyt protein distribution, monolayers of MDCK cells were biotinylated from either the apical or basolateral surface, extracted with lysis buffer containing 1% Nonidet P-40, and immunoprecipitated with monoclonal anti-N-cadherin antibodies. The majority (87%) of the biotinylated Ncyt protein was detected in extracts biotinylated from the basolateral surface (Fig. 6B, Ncyt). This distribution was closely similar to that of wild type N-cadherin (Fig. 6B, N). Thus, in contrast to the prominent apical localization of deletion mutants of other basolaterally targeted proteins, Ncyt was stably expressed on the basolateral domain.

Steady state distribution of Ncyt at the basolateral surface of MDCK cells.

To determine whether newly synthesized T- and N-cadherin are directly sorted to the apical or basolateral membrane, the kinetics of cadherin arrival at each membrane domain were measured between 4 and 96 h after induction of cell-cell contact. For synchronization, MDCK cells were kept on Transwell^TM polycarbonate filters in DMEM/FBS under low Ca²⁺ conditions (37). Formation of intercellular contacts was induced by raising the Ca²⁺ concentration to 1.8 mM. At various times after Ca²⁺ induction, cells were labeled for 1 h with [³⁵S]methionine/cysteine and biotinylated from either the apical or basolateral filter compartment (see above), and cadherins were immunoprecipitated with specific antibodies. The immune complexes were dissociated, and biotinylated cadherins were identified in either the apical or the basolateral fraction by reprecipitation with streptavidin-agarose. After 4 h, 84% of total T-cadherin was detected on the apical membrane domain (Fig. 7A). With further polarization, the percentage of T-cadherin delivered to the apical membrane increased, reaching a maximum of 96% after 96 h (Fig. 7A). Thus, T-cadherin was directly targeted to the apical membrane domain already 4 h after induction of cell-cell contact.

Delivery of newly synthesized cadherins and cadherin mutant Ncyt to specific membrane domains.

Targeting assays of wild type N-cadherin in transfected MDCK cells demonstrated that 86% (96 h after induction of cell-cell contact) of newly synthesized N-cadherin was delivered directly to the basolateral surface of fully polarized cells (Fig. 7B, N). This is comparable with the targeting of endogenous E-cadherin in MDCK cells, 82% of which is directly transported to the basolateral membrane domain 96 h after induction of cell-cell contact (Fig. 7B, E). A small percentage of N- and E-cadherin is missorted to the apical surface. As apically transported wild type N- and E-cadherin do not accumulate at the apical surface, they must be rapidly removed, either by proteolytic degradation or transcytosis.

N-Cadherin Cytoplasmic Tail Deletion Mutant Ncyt Is Transported to the Apical and Basolateral Domain in MDCK Cells and Is Removed at the Apical Surface

To address whether the Ncyt mutant protein is directly targeted to the basolateral domain, the arrival of metabolically labeled Ncyt protein at each membrane surface was measured in targeting assays as described above. Analysis of proteins at both membrane domains revealed that mutant protein was transported in similar amounts to both the apical and basolateral membrane domains (Fig. 7B, Ncyt).

The random transport of the Ncyt mutant raised the possibility that this protein is preferentially removed from the apical membrane surface. To test this hypothesis, the residence time of newly synthesized protein at each membrane domain was determined. Monolayer cultures were metabolically labeled for 15 min or 1 h with [³⁵S]methionine/cysteine, and the label was chased with medium containing an excess of unlabeled methionine. At various time points during the chase period, duplicate cultures were biotinylated for 15 min at 4 °C from either the apical or basolateral filter chamber as described above. Cells were extracted with Nonidet P-40 lysis buffer and immunoprecipitated with monoclonal anti-N-cadherin antibody followed by avidin-agarose. After a chase period of 4 h, 96% of Ncyt protein was found on the basolateral membrane (not shown). After chase of 120 h, 87% of the Ncyt protein was detected on the basolateral surface domain (Fig. 7C). Thus, the Ncyt mutant protein is preferentially removed at the apical membrane surface and accumulates basolaterally.

The regions within N- and T-cadherin responsible for their respective basolateral and apical distribution in MDCK cells were examined by studying the distribution of additional mutants (see Fig. 5).

Expression of T-Ncyt, composed of the T-cadherin ectodomain and the N-cadherin transmembrane domain and truncated cytoplasmic domain, was achieved only in a small number of cells and was unstable. Examination by indirect immunofluorescence revealed that the T-Ncyt mutant was localized on the basolateral MDCK cell surface (Fig. 8), identical to that of the Ncyt protein. Expression of the reverse chimera, N-T_GPI, containing the N-cadherin extracellular portion followed by the 76 T-cadherin carboxyl-terminal amino acids, was restricted to the apical membrane domain, as shown by confocal microscopic analysis of transfected cells after indirect immunofluorescence (Fig. 9A). As expression of the N-T_GPI mutant was unstable in MDCK cells, membrane attachment of the mutant protein was examined after transient transfection of the corresponding DNA construct into Cos7 cells. Proteins of 95 and 110 kDa were detected with monoclonal anti-N-cadherin antibody by Western blotting of transfected cell lysates (Fig. 9B). The sizes of these proteins correspond to those of T-cadherin, which comprises the mature 95-kDa protein and the unprocessed 110-kDa precursor (12, 13). Treatment of N-T_GPI-expressing Cos7 cells with PI-PLC released both the 95- and 110-kDa proteins from the cell surface into the culture medium, whereas in controls without PI-PLC, chimeric proteins were associated with cell membranes (Fig. 9B). Thus, the 76 carboxyl-terminal amino acids of T-cadherin include the signal for GPI anchor attachment.

Steady state distribution of T-Ncyt at the basolateral surface of MDCK cells.

Taken together, these results demonstrate that the carboxyl-terminal 76 T-cadherin amino acids contain the signal for GPI anchor attachment and are responsible for apical targeting of T-cadherin in MDCK cells.

DISCUSSION

In this article, we have examined whether GPI-linked T-cadherin is a component of cell-cell junctions like classical cadherins. In contrast to the basolateral distribution of classical cadherins, T-cadherin was transported and stably expressed at the apical domain of polarized epithelial cells. Apical targeting signals reside within the carboxyl-terminal 76 amino acids of T-cadherin and contain signals for GPI anchor attachment. The signal for GPI-linkage generally consists of an uncharged amino acid followed by a stretch of hydrophobic amino acid residues (38, 39). The T-cadherin carboxyl-terminal 76 amino acid region when fused to the ectodomain of N-cadherin produces a GPI-linked protein that is distributed apically. Therefore, the T-cadherin carboxyl-terminal amino acids are sufficient for GPI anchor attachment and apical targeting.

Classical cadherins contain in their cytoplasmic domain a putative basolateral targeting signal (50) that resembles that of the low density lipoprotein receptor and other basolaterally targeted proteins (36). The targeting motif comprises a tyrosine and a downstream cluster of three negatively charged amino acid residues, structurally organized into a type I -turn, and is related to clathrin-coated pit localization signals (40). Removal or inactivation of this signal results in the transport of basolaterally targeted proteins to the apical cell surface (41, 42, 43). Classical cadherins such as E-, P-, and N-cadherin contain in their cytoplasmic domain two such motifs, Y-(X)₆-EED and Y-(X)₃-EDD. However, N-cadherin lacks one of the three negatively charged amino acids in the distal signal, and the putative N-cadherin-targeting signals have not been probed for activity. The results reported here demonstrate that heterologously expressed N-cadherin is targeted basolaterally. Thus, the signals contained within the N-cadherin cytoplasmic domain are sufficient for basolateral targeting.

We demonstrate here that the N-cadherin tail minus mutant Ncyt is localized at the basolateral membrane in MDCK cells. This mutant is randomly transported to both the apical and basolateral membrane surfaces but rapidly and specifically removed from the apical pole, either by proteolytic degradation or transcytosis. The random transport of the Ncyt mutant is unexpected, as deletion of the basolateral targeting signals of proteins such as the low density lipoprotein receptor, the Fc receptor, lysosomal glycoprotein 120, and polyimmunoglobulin receptor results in the predominant targeting and stable expression of the mutant proteins at the apical membrane domain (41, 42, 43, 44). The fast removal of the Ncyt mutant may be explained by the lack of homophilic binding at the apical surface, which renders the protein accessible to proteolytic degradation. This is in contrast to the stable expression of both GPI-linked wild type T-cadherin and the N-T_GPI mutant at the apical surface. As the N-T_GPI chimera differs from the Ncyt mutant only in its carboxyl-terminal 76 amino acids, our results strongly argue that this region provides signals not only for apical targeting but also for stable expression at the apical domain. One possible explanation is that T-cadherin and the N-T_GPI chimera are clustered as GPI-linked proteins in caveolae and are therefore less accessible to proteolytic degradation. T-cadherin was recently isolated from the heart as a major protein of caveolae preparations (45), where it is hypothesized to serve as a calcium store and contribute to specific cardiac functions.

The apical localization of T-cadherin in MDCK cells correlates with its apical expression on intestinal epithelial cells in vivo. The highest T-cadherin concentration was detected on the apical extrusion zone where enterocytes exfoliate. Expression on the apical domain is complementary to that of E-cadherin (25, 26) and LI-cadherin (27), which occupy the basolateral domain and are distributed in adherens junctions and lateral contact zones, excluding junctional regions, respectively.

What is the possible function of T-cadherin at the apical surface of epithelial cells? One hypothesis is that T-cadherin is clustered in caveolae as in the heart and serves as a calcium store. Alternatively, and perhaps in addition, T-cadherin may play a role in receiving and transducing signals from the luminal space. A number of GPI-linked proteins of lymphocytes participate in T-cell activation putatively through a signaling cascade involving Src-related kinases (46, 47, 48, 49). In epithelial cells in vivo, apically expressed T-cadherin may play a role in signaling events required for specific epithelial functions. Together with the observation that T-cadherin mediates weak adhesive interactions² and repulses growth cones from specific neuron populations (18), its localization at the apical cell surface of polarized epithelial cells is consistent with a function as a recognition rather than a cell adhesion molecule. The nature of T-cadherin function and its mechanism of operation at the apical epithelial cell surface are important issues to be addressed concerning this unique member of the cadherin family.