A Dileucine Motif Targets E-cadherin to the Basolateral Cell Surface in Madin-Darby Canine Kidney and LLC-PK 1 Epithelial Cells*

,

E-cadherin is expressed on the lateral membranes of epithelial cells where it accumulates as a major component of the adherens junction. The cadherins in adherens junctions have central roles in establishing and maintaining cell-cell adhesion and cell polarity in epithelia and participate in morphogenesis during development (1)(2)(3)(4). The continual expression and function of E-cadherin is important in its role as a tumor suppressor and the loss of E-cadherin function contributes to tumor invasion and progression in carcinomas (5).
Mature, human E-cadherin is a 728-amino acid, single pass transmembrane protein (6). The E-cadherin ectodomain is involved in Ca 2ϩ -dependent homotypic binding to E-cadherins on adjacent cell membranes, and its cytoplasmic tail is involved in a series of protein interactions providing a link to the actin cytoskeleton. The cytoplasmic tail of E-cadherin is bound directly to ␤-catenin or plakoglobin, and thereby to ␣-catenin and actin (reviewed in Ref. 2). The binding site for ␤-catenin has been mapped by deletion mutagenesis to the distal 76 amino acids of the carboxyl terminus of E-cadherin (7), with critical residues found in the last 30 amino acid domain (8). Aside from its participation as a member of the cadherin-bound adherens junction complex, ␤-catenin is involved in the Wnt signaling pathway through its interactions with a cytoplasmic adenomatous polyposis coli complex and with the Lef/Tcf transcription factors in the nucleus (reviewed in Ref. 9).
Epithelial cells are morphologically and functionally polarized with distinct complements of cell surface proteins and lipids at the apical or basolateral poles. Maintenance of this polarity requires that newly synthesized proteins are sorted and targeted to specific membrane domains (10 -12). Sorting of proteins to the apical domain of polarized cells occurs via lipid raft interactions or through oligosaccharides (13)(14)(15), whereas specific amino acids motifs act as sorting signals to target membrane proteins to the basolateral cell surface (reviewed in Ref. 16). For example, tyrosine-based motifs target the low density lipoprotein receptor to the basolateral domain of polarized cells (17), whereas a dileucine motif is utilized by other proteins, like the Fc receptor (18). Both types of signals can also be used to direct endocytosis from the plasma membrane, although the structural requirements for each pathway may differ (19 -22). Yet other amino acid motifs can be used by selected proteins for basolateral targeting (23)(24)(25). Previous studies on adhesion molecules have defined a dileucine signal that functions in the basolateral targeting of the Lutheran glycoprotein (26) and a similar dihydrophobic signal in CD44 (27). The basolateral targeting of neural cell adhesion molecule (N-CAM) has been attributed to a cytoplasmic tail sequence without classic motif homology (23). Detailed information and further insights are needed into how other classes of adhesion proteins, including cadherins, are sorted and trafficked in polarized cells.
Recently, the polarized epithelial cells of the pig kidney LLC-PK 1 line have been distinguished as having aspects of inverted polarity. Roush et al. (28) noted that some typically basolateral proteins, such as the H,K-ATPase ␤ subunit, are delivered to the apical pole of LLC-PK 1 cells. Mis-sorting of basolateral proteins in LLC-PK 1 cells led to subsequent analysis of their AP-1 adaptor complexes. Most polarized epithelial cells express a special 1B subunit, which directs polarized sorting to the basolateral surface (29). It has now been shown that LLC-PK 1 cells express a 1A subunit but not the 1B subunit and that this results in aberrant sorting of proteins with tyrosine-based signals, a defect that can be overcome by expression of recombinant 1B protein (30). Cadherin staining in LLC-PK1 cells appears to be basolateral (31), although the trafficking of these proteins in this cell line has not been studied in detail.
The correct placement of E-cadherin on the plasma membrane is required from an early stage to help establish and maintain cell polarity (32). The mechanisms that mediate the sorting and polarized delivery of E-cadherin to the surface have not been elucidated. Previous studies have shown that ␤-catenin binds to E-cadherin early in the biosynthetic pathway, implying that the two proteins, and perhaps others, are transported to the cell surface together as a complex (33). More recently, it was suggested that ␤-catenin plays an essential role in the trafficking of E-cadherin, based on observations that mutagenized proteins, with reduced binding to ␤-catenin, were not efficiently delivered to the surface (34). It has also been previously noted that the cytoplasmic tail of E-cadherin does contain motifs with homology to known targeting signals (34,35), which could potentially function to guide its trafficking. In this study we set out to test putative signals in the cytoplasmic tail of E-cadherin for possible basolateral targeting information. Our experiments also addressed the role of ␤-catenin in this targeting. We utilized a series of chimeras to express the E-cadherin cytoplasmic tail, or mutagenized versions thereof, using Tac (the ␣ subunit of the IL-2 receptor) as an ectodomain marker. Tac is a 273-amino acid protein that has previously been used as a reporter protein for trafficking studies (36,37). These constructs were expressed in MDCK 1 cells, in which the basolateral surface expression of E-cadherin is well established, and in another epithelial cell line LLC-PK 1 cells. Using these model systems we have identified a positive targeting signal in E-cadherin that is responsible for basolateral delivery. Our results provide new insights into the trafficking of E-cadherin and its accessory proteins (catenins), and the findings are also significant to our understanding of cell polarity and sorting pathways in epithelia.

EXPERIMENTAL PROCEDURES
Cell Culture-Madin-Darby canine kidney (MDCK) and pig kidney (LLC-PK 1 ) cell lines were grown and passaged as described previously (38) in Dulbecco's modified Eagle's medium (Life Technologies, Inc., Grand Island, NY) supplemented with 10% fetal calf serum and 2 mM glutamine in 5% CO 2 and 95% air.
Antibodies-Mouse monoclonal antibodies (Transduction Laboratories, Lexington, KY) raised against a conserved region of the cytoplasmic domain of human E-cadherin and against ␤-catenin were used; Tac was recognized using a mouse monoclonal antibody (B-B10, BIO-SOURCE International, Camarillo, CA) and a rabbit polyclonal antibody directed against the green fluorescence protein (GFP) (Molecular Probes Inc., Eugene, OR) was also used for staining. Secondary antibodies included Cy3-conjugated sheep anti-mouse and goat anti-rabbit IgGs (Jackson ImmunoResearch Labs, West Grove, PA) and a horseradish peroxidase-conjugated sheep anti-mouse IgG (AMRAD, Victoria, Australia).
cDNA Construction and Expression-Chimeric fusions between human E-cadherin cDNAs and the cDNA encoding for Tac were generated in the pCDNA3 expression vector. Molecular cloning techniques were performed according to Sambrook et al. (39), using reagents from New England BioLabs (Beverly, MA). All constructs were confirmed by DNA sequencing. cDNAs encoding the transmembrane plus cytoplasmic domains (residues 554 -728), the cytoplasmic tail (residues 578 -728), or the amino-terminal half of the cytoplasmic tail (578 -653) of human E-cadherin were amplified using PCR with specific oligonucleotide primers. E-cadherin residue numbers correspond to the mature protein as defined previously (6). The oligonucleotide primers contained restriction endonuclease sites allowing generation of in-frame fusions with the Tac cDNA. Cloning the respective PCR products into pCMV-IL2R (40) using the HindIII and XbaI sites generated the pCMV-Tac/Ecad-(578 -728) and pCMV-Tac/Ecad-(578 -653) plasmids. The entire Tac/E-cadherin cDNA constructs were then subcloned into the pCDNA3 expression vector (Invitrogen, Carlsbad, CA) that also encodes for the neomycin selection marker. The resulting constructs were termed Tac/ Ecad-(578 -728) and Tac/Ecad-(578 -653). To generate the Tac/Ecad-(554 -728) plasmid, an EcoRV restriction endonuclease site was introduced using PCR at the end of the extracellular domain of the Tac cDNA within pCDNA3. This modification altered residue 239 of the Tac cDNA from an Asp to a Gln. The final plasmid was generated by cloning the E-cadherin PCR product into the EcoRV and XbaI sites.
E-cadherin-GFP encodes the full-length E-cadherin sequence with the green fluorescence protein (GFP) fused to the carboxyl terminus of the cytoplasmic domain. Initially, a SacII restriction endonuclease site was introduced at the carboxyl terminus of the full-length cDNA of E-cadherin. This was achieved by the PCR amplification of the entire E-cadherin cDNA using specific oligonucleotide primers using pCDNA3-hECad (41) as a template. The 3Ј-primer included the SacII site and a XbaI site. The resulting PCR product was digested with SgrA1 and XbaI and subcloned into pCDNA3-hECad using the same sites. The open reading frame of GFP was amplified by PCR using specific oligonucleotide primers and pEGFP-N1 (CLONTECH, Palo Alto, CA) as a template. The 5Ј-primer contained a SacII site that allowed the resulting PCR product to be cloned into the SacII site introduced into in the carboxyl terminus of E-cadherin.
Mutagenesis of the two leucine residues at positions 587 and 588 of human E-cadherin to alanines was performed using oligonucleotide cloning. The construct pCMV-Tac/Ecad-(578 -728) was digested with HindIII and XbaI to remove a 55-bp fragment that included the codons for the dileucine residues. The digested plasmid was then ligated with the annealed, phosphorylated, synthetic oligonucleotides 5Ј-AGCTTC-GTCGACGCGCGGTGGTCAAAGAGCCCGCAGCACCCCCAGAGGAT-GACAC3-Ј and 5Ј-CCGGGTGTCATCCTCTGGGGGTGCTGCGGGCTC-TTTGACCACCGCGCGTCGACGA3Ј. These complementary oligonucleotides contain a 5Ј-HindIII end and a 3Ј-XbaI end, as well as codons encoding for LRRRAVVKEPAAPPEDD (altered residues are underlined). The resulting construct was termed Tac/Ecad-(578 -728)⌬S1.
For transfection and expression of cDNAs, sub-confluent LLC-PK 1 or MDCK cells were transfected with plasmid DNA (2 g) in complex with LipofectAMINE Plus reagent (Life Technologies, Inc., Gaithersburg, MD) according to the manufacturer's guidelines. For stably expressing lines, transfected cells were passaged and maintained in media containing G418 (Geneticin, Life Technologies, Inc.); cells were kept under selection for 7-10 days and then plated at low density for ring cloning of surviving cells. Clonal cell lines were generated and then assessed by indirect immunofluorescence and immunoblotting to select lines with different levels of recombinant protein expression.
Indirect Immunofluorescence-Confluent monolayers of cells grown on glass coverslips or on Transwell polycarbonate filters (Corning Costar, Cambridge, MA) were generally fixed in 4% paraformaldehyde in PBS for 90 min and then permeabilized in PBS containing 0.1% Triton X-100 for 5 min. In one experiment LLC-PK 1 cells were fixed in ice-cold methanol for 10 min. Cells were then incubated sequentially with primary antibody (1 h) and then secondary antibodies (30 min) using PBS containing bovine serum albumin (Sigma Chemical Co., St. Louis, MO) as a blocking buffer. Cells were mounted on slides in PBS/glycerol (50/50) containing 1% n-propyl-galate. For some experiments, cells were treated with 10 M cycloheximide (Sigma), which was added to the medium for various times up to 4 h prior to fixation. Cells on coverslips were examined by epifluorescence using an Olympus Provis AX-70 microscope, and images were collected with a CCD300ET-RCX camera (DageMTI, Michigan City, IN) using National Institutes of Health IMAGE software. Cells growing on Transwell filters were examined using a Bio-Rad MRC-600 confocal laser-scanning microscope mounted on a Zeiss Axioskop, and XY and XZ sections were generated using Bio-Rad MRC-600 CoMOS software.
Immunoprecipitation and Immunoblotting-Confluent monolayers of transfected MDCK and LLC-PK 1 cells were solubilized in cold RIPA buffer (1% Triton X-100, 1% deoxycholate, 0.1% SDS, 0.15 NaCl, 5 mM EDTA, 25 mM Tris-HCl, pH 7.4) containing protease inhibitors (Roche Molecular Biochemicals, Germany) on ice. Post-nuclear supernatants were incubated with the Tac antibody for 2 h and then with washed protein G beads (Sigma) for a further 2 h. Precipitates were recovered by centrifugation then washed through several rounds of RIPA buffer 1 The abbreviations used are: MDCK, Madin-Darby canine kidney cells; GFP, green fluorescence protein; IL-2R, interleukin-2 receptor; LLC-PK 1 , pig kidney proximal tubular epithelial cell line; PAGE, polyacrylamide gel electrophoresis; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; Tac, IL-2R␣ chain; RIPA, radioimmune precipitation buffer; bp, base pair(s); CMV, cytomegalovirus. and 20 mM Tris-HCl (pH 7.4) prior to solubilization in concentrated SDS-PAGE sample buffer. Proteins in cell extracts and immunoprecipitates were separated on 8% SDS-PAGE reducing gels and then transferred to polyvinylidene difluoride Immunobilon-P membranes (Millipore, Bedford, MA) and stained with 0.1% Coomassie Brilliant Blue to ensure even protein transfer and protein loading. Membranes were immunoblotted by sequential incubations in primary antibody, horseradish peroxidase-conjugated secondary antibody followed by chemiluminescence detection with Supersignal West Pico (Pierce Chemical Co., Rockford, IL). Different luminescence exposures were collected and exposures in the linear range were used.
Surface Biotinylation-Confluent monolayers of LLC-PK 1 cells stably expressing Tac/Ecad-(578 -728) or Tac/Ecad-(578 -653) grown on filters were incubated in media containing 1.5 mg/ml Sulfo-NHS-SSbiotin (Pierce), applied to either the apical or basal side of the filter, for 60 min on ice. Filters were then washed several times in cold PBS before cells were scraped off and lysed in cold RIPA buffer. Soluble cell fractions were incubated with streptavidin beads (Sigma) in RIPA buffer, pH 7.4, for 2 h with rotation. Beads were then washed in several rounds of RIPA buffer and 20 mM Tris-HCl (pH 7.4). Biotinylated proteins bound to the streptavidin beads and unlabeled proteins in the supernatants were analyzed by SDS-PAGE, immunoblotting, and densitometry to quantitate the relative amounts of biotinylated Tac/Ecad proteins.

Basolateral
Targeting of E-cadherin-E-cadherin is delivered to the basolateral surface of polarized MDCK cells, where it gives a typical and widely documented staining pattern using specific antibodies (Fig. 1a). The same antibody did not stain E-cadherin in paraformaldehyde-fixed LLC-PK 1 cells (Fig. 1c), but it did produce cell surface staining of E-cadherin in methanol-fixed LLC-PK 1 cells (31 and Fig. 1b). Hence E-cadherin is expressed endogenously in both cell lines and is found in a polarized distribution. A tagged construct of human GFP-Ecadherin was expressed in MDCK cells, generating a clear basolateral surface staining pattern with GFP antibodies, showing that GFP-E-cadherin is targeted in a manner analo-gous to the endogenous protein (Fig. 1d). GFP-E-cadherin was also expressed in epithelial LLC-PK 1 cells, where it was also targeted in a polarized fashion to the basolateral membrane (Fig. 1e).
Targeting of Tac/E-cadherin Chimeras-For targeting studies we utilized chimeras consisting of the ectodomain of Tac fused to the cytoplasmic tail of E-cadherin ( Fig. 2A). Chimeric cDNAs expressed in epithelial cells all produced proteins of the expected molecular masses (Fig. 2B). Tac typically localizes to the apical membrane in polarized cells (Ref. 36 and Fig. 3, a  and b), therefore, any basolateral signal in the E-cadherin cytoplasmic domain is predicted to redirect Tac from apical to basolateral membranes. cDNAs for Tac alone or chimeric proteins were transfected into MDCK and LLC-PK 1 cells, and clonal, stably transfected cell lines were selected. Antibodies against Tac were used to detect the chimeric proteins and determine their localization by indirect immunofluorescence and confocal microscopy.
Chimeras containing the full cytoplasmic tail of E-cadherin were expressed and found to redirect Tac to the basolateral domain in both MDCK and LLC-PK 1 cells. Both Tac/Ecad-(578 -728) and Tac/Ecad-(554 -728), which additionally encodes the transmembrane domain of E-cadherin, were localized by epifluorescence and by confocal imaging to the basolateral membranes of MDCK and LLC-PK 1 cells (Fig. 3). Thus the presence or absence of the E-cadherin transmembrane domain had no effect on targeting. These results indicate that the Tac/E-cadherin chimeras are efficiently synthesized and transported to the cell surface and that the cytoplasmic tail of E-cadherin contains positive sorting information, capable of rerouting Tac to a basolateral trafficking pathway. MDCK cells expressing Tac/Ecad-(578 -728) showed no concomitant loss of endogenous E-cadherin staining on the basolateral surface (not shown), suggesting that the sorting and targeting machinery in these cells has not been saturated or subverted by the overexpressed protein.
Basolateral Targeting Mediated by the Membrane-proximal E-cadherin Tail-As the first step in a more detailed analysis of the cytoplasmic tail of E-cadherin, a truncation mutant was created to effectively bisect the cytoplasmic tail, leaving only the membrane-proximal portion of the tail fused to Tac. The resulting Tac/Ecad-(578 -653) construct was expressed in MDCK and LLC-PK 1 cells (Fig. 4). Immunofluorescence staining and confocal analysis showed that it was distributed in a polarized fashion. There was no staining of apical membranes when antibody was applied to either unpermeabilized cells (not shown) or permeabilized cells, however, there was staining of the basolateral membranes in LLC-PK 1 and MDCK cells expressing Tac/Ecad-(578 -653) (Fig. 4, a and c). Thus, chimeras containing either the full-length tails or only the membrane-proximal tails are trafficked similarly and have the same polarized surface distribution.
We noted that, in cells from several different clones stably expressing Tac/Ecad-(578 -653), there was intracellular staining of Tac/Ecad-(578 -653) in a perinuclear, Golgi-like pattern in addition to basolateral surface staining (Fig. 4a). This pattern was not regularly seen in cell lines expressing chimeras with full-length tails (Tac/Ecad-(554 -728) or Tac/Ecad-(578 -728)). LLC-PK 1 cells expressing Tac/Ecad-(578 -653) were treated with cycloheximide to stop protein synthesis and fixed and stained at various times after addition of the drug. There was a sequential loss of intracellular staining followed at longer times by a diminution of cell surface staining. Fig. 4 shows that after 2 h of treatment, all of the intracellular staining had disappeared, leaving only staining of Tac/Ecad-(578 -653) at the basolateral surface. From this it was concluded that intracellular staining in these cells represents a transient accumulation of newly synthesized Tac/Ecad-(578 -653) in the biosynthetic pathway and that the membrane-proximal chimera is transported to the basolateral membrane at a slower rate than Tac/Ecad-(578 -728).
The targeting of Tac/Ecad-(578 -653) was finally tested using a surface biotinylation assay to measure and quantify its appearance on the plasma membrane domains of confluent, polarized cells. Cell surface biotinylation was performed on cell lines stably expressing Tac/Ecad-(578 -728) and Tac/Ecad-(578 -653). Addition of biotin reagents to the basal sides of the monolayers labeled most of the Tac/Ecad-(578 -728) and Tac/ Ecad-(578 -653) proteins, whereas from the apical side almost no labeling occurred in either case (see Table I). Together these results show that a chimeric protein, containing only the membrane-proximal half of the E-cadherin cytoplasmic tail, has sufficient information to direct efficient sorting and targeting to the basolateral cell surface, albeit perhaps at a slower rate than constructs with the full-length cytoplasmic tail.
A Dileucine Motif Is Responsible for Basolateral Targeting of E-cadherin-Sequence analysis of the membrane-proximal Ecadherin tail encoded by the region in Tac/Ecad-(578 -653) revealed the presence of two putative targeting motifs. Sequence alignment of members of the type I cadherins, of which E-cadherin is the prototype, and type II cadherins (42), revealed that a dileucine motif at position 587 is highly conserved across species and preserved in almost all members of the family (Fig. 5A). To test this dileucine motif for targeting information, the leucines at positions 587 and 588 were changed to alanines in the chimeras encoding the full-length tail and membrane-proximal tail, using oligonucleotide cloning. The resulting mutated chimeras, termed Tac/Ecad-(578 -728)⌬S1 and Tac/Ecad-(578 -653)⌬S1, were transfected into LLC-PK 1 and MDCK cells and then localized by immunofluorescence (Fig.  5B). Tac/Ecad-(578 -728)⌬S1 and Tac/Ecad-(578 -653)⌬S1 were localized at the apical surfaces of transfected LLC-PK 1 and MDCK cells, in patterns similar to the apical Tac (Fig. 3, a  and 3b) and distinct from the basolateral non-mutated chimeras. Thus removal of the dileucine motif at 587 from the Ecadherin cytoplasmic domain resulted in a loss of basolateral targeting information. We conclude that this motif is a positive sorting signal for the basolateral membrane localization of E-cadherin and that it is necessary to direct sorting. The high level of conservation of this dileucine motif throughout the cadherins family further suggests that it has a key functional role. A second potential targeting signal of the type NPXY is present in the sequence of Tac/Ecad-(578 -653). This tyrosinebased signal at position 600 was not tested here and is not considered a likely candidate for targeting, based on experimental data from another study (34) and our observation that the motif is not conserved in cadherins across different species.
Binding of ␤-Catenin-␤-Catenin is known to bind to the cytoplasmic tail of E-cadherin early in the biosynthetic pathway and has previously been implicated in trafficking to the cell surface (34). Therefore, Tac/E-cadherin chimeras were tested for their ability to bind to ␤-catenin. The interaction of endogenous E-cadherin in MDCK cells with ␤-catenin was demonstrated by co-immunoprecipitation of the two proteins using an E-cadherin antibody (Fig. 6A). Tac/Ecad-(578 -728) was immunoprecipitated with the Tac antibody from extracts of transfected MDCK and LLC-PK 1 cells, and ␤-catenin coprecipitating in the complex was detected by immunoblotting. In extracts of both MDCK and LLC-PK 1 cells, ␤-catenin was bound to Tac/Ecad-(578 -728) (Fig. 6B). This finding suggests that the full-length cytoplasmic tail in Tac/Ecad-(578 -728) is correctly folded and processed in a manner conducive to effective trafficking and surface delivery.
The membrane-proximal Tac/Ecad-(578 -653) construct is missing the carboxyl-terminal ␤-catenin binding domain, and co-immunoprecipitation experiments confirm that, as expected, ␤-catenin was not co-precipitated with this truncated chimeric protein (Fig. 6B). Endogenous E-cadherin and chimeras with full-length E-cadherin tails were able to efficiently bind and co-immunoprecipitate ␤-catenin (Fig. 6, A and B). Deletion of the 587 dileucine targeting signal had no effect on binding of ␤-catenin, which was efficiently co-precipitated with Tac/Ecad-(578 -728)⌬S1 (Fig. 6B). The correct basolateral targeting of Tac/ Ecad-(578 -653) in the absence of bound ␤-catenin now demonstrates that ␤-catenin is not essential for basolateral sorting and targeting in either MDCK or LLC-PK 1 cells. The complexing of ␤-catenin with newly synthesized E-cadherin may be required for other roles in biosynthetic processing or trafficking, for instance, the loss of ␤-catenin binding may account for the increased  intracellular accumulation and apparently slower trafficking of the Tac/Ecad-(578 -653) mutants. DISCUSSION E-cadherin is one of the proto-typical polarized membrane proteins in epithelia. It is delivered to the basolateral membrane and is concentrated in adherens junctions where it participates in cell-cell adhesion. To address how E-cadherin is trafficked and targeted in polarized cells, we made use of Tac/ E-cadherin chimeras expressed in two epithelial cell lines. Our findings show that chimeras containing full-length or truncated cytoplasmic tails of E-cadherin were effectively sorted and trafficked to the basolateral surface domain in MDCK and LLC-PK 1 cells. Furthermore, this polarized targeting was lost after deletion of a single, dominant targeting motif in the proximal region of the tail. Three main conclusions emerged from these experiments; (i) that a single dileucine motif at 587 is able to convey basolateral targeting information in Tac/Ecadherin chimeras and that this signal is necessary for basolateral sorting; (ii) that LLC-PK 1 cells are able to correctly sort and traffic E-cadherin using a dileucine-based mechanism, in contrast to proteins sorted by tyrosine-based signals that are mis-trafficked in these cells, and (iii) that ␤-catenin is not required for basolateral targeting and surface delivery of E-cadherin.
Chimeras containing the full cytoplasmic tail of E-cadherin, Tac/Ecad-(578 -728) and Tac/Ecad-(554 -728), were efficiently targeted and trafficked to the basolateral domain of MDCK and LLC-PK 1 cells in a similar manner to that of endogenous Ecadherin or GFP-E-cadherin in either cell line. The Tac ectodomain did not interfere with the intracellular trafficking or processing in the full-length tail constructs, because there was no evidence of less efficient membrane delivery or retention within the secretory pathway. The transmembrane domain has been implicated in mediating lateral association and adhesive strength of cadherins in the plasma membrane, without affecting surface delivery (43). Amino acid sequences within transmembrane segments can, however, be involved in targeting as exemplified by the apical targeting signal in one of the transmembrane segments of the ␣-subunit of the gastric parietal H,K-ATPase (44). Our experiments directly tested the E-cadherin transmembrane domain, with the finding that this region of the protein has no role in basolateral targeting nor in sorting or membrane delivery of Tac/E-cadherin chimeras. Overall the results obtained with the Tac/E-cadherin chimeras point to the cytoplasmic tail of E-cadherin as having positive basolateral sorting information, in keeping with many other type I membrane proteins.
The correct basolateral targeting of the Tac/Ecad-(578 -653) construct first indicated that the membrane-proximal tail region alone is sufficient for membrane targeting and that distal regions of the tail are not required for targeting or surface delivery. In our hands, mutants with the truncated tail of E-cadherin, Tac/Ecad-(578 -653), were processed, targeted, and delivered to the basolateral cell surface, albeit at a slower rate than constructs with the full cytoplasmic tail. Transient accumulation of Tac/Ecad-(578 -653) in the biosynthetic pathway, seen as intracellular staining in LLC-PK 1 cells, indicated that it was trafficked less efficiently than the full-length tail. Partial accumulation of Tac/Ecad-(578 -653) was similarly noted in MDCK cells, although the truncated tail mutants did eventually reach the cell surface. Our findings are in contrast to those of a previous study in which a series of chimeras representing truncation mutants of the E-cadherin tail fused to a GP-2 ectodomain, were found to be very poorly trafficked (34). Only about 10% of the truncated GP-2 chimeric proteins in that study reached the surface of MDCK cells and were randomly sorted, whereas the majority of the proteins were blocked or degraded early in the secretory pathway (34). On the basis of the correct targeting and delivery of Tac/Ecad-(578 -653), which contains similar tail regions to some of the mutants in the previous study, we hypothesize that the poor processing and trafficking of those chimeras may have been due to the influence of the GP-2 ectodomain rather than being a function of E-cadherin trafficking or accessory proteins (see below).
The membrane-proximal tail region of E-cadherin has been implicated in a number of functions. Cadherin clustering and adhesive strength were shown to be affected by mutations in the membrane-proximal region of the C-cadherin tail (45). Binding of the p120 ctn protein has also been shown to occur in this region (45,46), and some of the amino acid sequences mapped by two-hybrid analysis, as being specifically involved in the binding of p120 ctn (47), overlap with the dileucine targeting signal identified in this study. It is therefore likely that this membrane-proximal region of the tail is involved in multiple, temporally regulated protein interactions that are important, initially for E-cadherin transport, and then for adhesive function at the cell surface. Positive sorting mediated by the Tac/Ecad-(578 -653) membrane-proximal tail allowed us to focus on putative sorting signals in this region. The dileucine motif at 587 was chosen here as a candidate targeting signal. Despite being in the non-conserved region of the tail, sequence alignments revealed that the dileucine consensus motif is highly conserved throughout cadherins of type I or II cadherin families (42). We therefore predict that this dileucine will function to target all similar transmembrane cadherins to basolateral domains of polarized epithelial cells. Cadherins in which the dileucines are not conserved include cadherins 5 (VE-cadherin) and 20. The glycosyl phosphatidylinositol-anchored cadherin 13 (T-cadherin), as expected, does not have a dileucine motif (48).
Our experimental results, therefore, confirmed that the dileucine motif in E-cadherin does function as a targeting signal. Replacing the leucines with alanines in the full-length or truncated tail constructs resulted in a complete loss of basolateral targeting. The dileucine signal in E-cadherin has an acidic amino acid cluster on its carboxyl-terminal side that is highly conserved throughout dileucine-containing cadherins and is similar to targeting motifs in other basolateral proteins, including furin (49), invariant chain (50), and low density lipoprotein receptor (22). In some cases, such as for furin, these acidic clusters have been shown to be important for the function of dileucine signals in basolateral targeting (49). Dileucine signals functioning in endocytosis typically have an acidic residue at the Ϫ4 position (D/EXXXLL) (21,51,52). In contrast, the cadherin dileucine signal typically has a basic lysine or arginine in the Ϫ4 position. It is, therefore, unlikely that this motif will also function as an endocytosis motif.
There are additional motifs sharing consensus with targeting signals encoded in the E-cadherin tail. Tyrosines at two places within the tail, one being in the membrane-proximal region and another at the carboxyl terminus, were deleted in a previous study and found to have no role in targeting of Ecadherin (34). There is a combined YXX tyrosine and dileucine motif (673-677), similar in structure to the overlapping motif, which has been shown to be responsible for basolateral targeting of the pIg F receptor in MDCK cells (53). There is also a motif belonging to the YXX group, with a tyrosine at the second X position (705-708). Both of these latter signals are adjacent to, or within, the ␤-catenin binding domain and are thus predicted to be sequestered when E-cadherin is complexed to ␤-catenin. The possibility remains open, however, that any of these additional signals might be uncovered during dynamic protein interactions and therefore could act as targeting signals during further trafficking of surface E-cadherin. One or more of these signals could, for instance, target Ecadherin to clathrin-coated vesicles for endocytosis and recycling (35) after it reaches the basolateral plasma membrane. Overall, analysis of targeting motifs points to a dileucine signal rather than tyrosine-based signals being responsible for the polarized sorting and basolateral delivery of E-cadherin.
Both MDCK and LLC-PK 1 cell lines form polarized epithelial monolayers in culture that show patent basolateral trafficking and secretion of soluble proteoglycans (54,55). Due to the reported inverse polarity of LLC-PK 1 (28,30), it was of interest to also analyze E-cadherin trafficking in these cells. Our findings verify that E-cadherin is trafficked correctly to the basolateral surface of the LLC-PK 1 cells, and we show that this targeting, as in MDCK cells, is directed by a dileucine motif. This provides new evidence to confirm that basolateral targeting via dileucine-based mechanisms functions correctly in LLC-PK 1 cells. The correct targeting of endogenous E-cadherin, GFP-E-cadherin, and Tac/Ecad constructs suggests that dileucine-based sorting is fully sufficient to direct basolateral trafficking in these cells. LLC-PK 1 cells are defective in targeting of tyrosine-based motifs due to the absence of the 1B chain (30). However, the sorting of dileucine motifs occurs through interaction with the ␤ subunits of the adaptor complex (56), allowing correct sorting of proteins such as T cell receptor sub-unit (CD3␥), Fc receptor, and E-cadherin. The correct targeting of endogenous E-cadherin and Tac/Ecad constructs in LLC-PK 1 cells acts as further evidence that this targeting does, in fact, rely on dileucine rather than tyrosine motifs. The full nature of the adaptor complexes or that required for post-Golgi transport of E-cadherin in either LLC-PK 1 or MDCK epithelial cells have yet to be characterized. Finally, the current study also provides new insights into the role of ␤-catenin in E-cadherin trafficking. ␤-Catenin is a cytoplasmic protein with affinity for the cytoplasmic tail of Ecadherin, it binds to E-cadherin early in the biosynthetic pathway, forming a stable complex that is transported to the cell surface (33). Chen and colleagues (34) concluded that ␤-catenin is required for the biosynthetic processing and trafficking of E-cadherin, based on a correlation between deletion of residues within or near the ␤-catenin binding domain and loss of surface delivery in GP-2-E-cadherin chimeras and other constructs. However, our current results suggest a different scenario. We showed that full-length tail chimeras co-precipitated ␤-catenin in similar proportion to that seen in endogenous E-cadherin-␤catenin complexes. Upon expressing the Tac/Ecad-(578 -653) chimera, which clearly did not bind ␤-catenin, we found it was correctly targeted and transported to the cell surface, suggesting that ␤-catenin is not required for sorting or delivery under these conditions. ␤-Catenin may have a role in facilitating or optimizing the transport, processing, or folding of newly synthesized E-cadherin. All of these factors might well contribute to the slower processing and transient accumulation of Tac/ Ecad-(578 -653) that we noted in the absence of ␤-catenin. Recent biophysical and biochemical analyses further suggest that ␤-catenin binding might also serve to protect the E-cad-FIG. 6. Co-immunoprecipitation of ␤-catenin. A, proteins immunoprecipitated from untransfected MDCK cells with a monoclonal antibody to E-cadherin were probed by immunoblotting with the E-cadherin antibody (top) or with a ␤-catenin antibody (bottom). Proteins at 120 and 92 kDa, respectively, were detected. B, the Tac antibody was used to immunoprecipitate chimeras from transfected LLC-PK 1 cells. The ␤-catenin antibody was then used for immunoblotting supernatants (SN lanes 1, 3, and 5) and immunoprecipitates (IP lanes 2, 4, and 6). ␤-Catenin was co-precipitated with Tac/Ecad-(578 -728) (lane 2) but not with the truncated Tac/Ecad-(578 -653) chimera (lane 4). Deletion of the dileucine targeting motif (Tac/Ecad-(578 -728)⌬S1) did not affect co-precipitation of ␤-catenin (lane 6).
herin tail from degradation (57). Our finding, that ␤-catenin is not required for basolateral targeting or surface delivery, sheds new light on ␤-catenin as having a role in facilitating, but not directing, cadherin trafficking.
The polarized targeting of E-cadherin is of seminal importance to the maintenance of epithelial polarity and function. Targeting signals and mechanisms must ensure the accurate basolateral delivery of newly synthesized E-cadherin and of internalized and recycled E-cadherin for its incorporation into adherens junctions. In this study we demonstrate one such mechanism, basolateral sorting directed via a dileucine signal. Future studies will address specific roles for additional signals and perhaps for some of the accessory proteins in cadherin/ catenin complexes.