Isoforms of the Lutheran/Basal Cell Adhesion Molecule Glycoprotein Are Differentially Delivered in Polarized Epithelial Cells

Lu and Lu(v13) are two glycoprotein (gp) isoforms that belong to the immunoglobulin superfamily and carry both the Lutheran (Lu) blood group antigens and the basal cell adhesion molecule epithelial cancer antigen. Lu (85 kDa) and Lu(v13) (78 kDa) gps, which differ only in the length of their cytoplasmic domain, are adhesion molecules that bind laminin. In nonerythroid tissues, the Lu/basal cell adhesion molecule antigens are predominantly expressed in the endothelium of blood vessel walls and in the basement membrane region of normal epithelial cells, whereas they exhibit a nonpolarized expression in some epithelial cancers. Here, we analyzed the polarization of Lu and Lu(v13) gps in epithelial cells by confocal microscopy and domain-selective biotinylation assays. Differentiated human colon carcinoma Caco-2 cells exhibited a polarized expression of endogenous Lu antigens associated with a predominant expression of the Lu isoform at the basolateral domain of the plasma membrane and a very low expression of the Lu(v13) isoform at both the apical and basolateral domains. Analysis of transfected Madin-Darby canine kidney cells revealed a basolateral expression of Lu gp and a nonpolarized expression of Lu(v13) gp. Delivery of Lu(v13) to both apical and basolateral surfaces showed similar kinetics, indicating that this isoform is directly transported to each surface domain. A dileucine motif at position 608–609, specific to the Lu isoform, was characterized as a dominant basolateral sorting signal that prevents Lu gp from taking the apical delivery pathway.

The Lutheran (Lu) 1 and basal cell adhesion molecule (B-CAM) antigens were recently shown to be carried by a new glycoprotein member of the immunoglobulin superfamily with five extracellular Ig-like domains (two V set and three C2 set domains from the NH 2 terminus) (1,2). The Lu antigens belong to a highly polymorphic blood group system and were originally shown to be carried by low abundance erythrocyte membrane glycoproteins of 85-and 78-kDa species (3,4). The B-CAM antigen was first identified by monoclonal antibodies raised against human tumor cells and was shown to be overexpressed in ovarian carcinomas in vivo and up-regulated following malignant transformation in some cell types (5,6). We have recently demonstrated that the 85-and 78-kDa gps express both the Lu and B-CAM antigens and represent isoforms of the same protein encoded by 2.5-and 4.0-kb mRNAs, resulting from alternative splicing of intron 13 of the unique gene, LU (7,8). Therefore, we now refer to these molecules, which differ only in the length of their cytoplasmic tail (59 and 19 amino acids, respectively) as the Lu and Lu(v13) gps. The 40 COOH-terminal amino acids specific to the Lu gp isoform carry a prolinerich motif for binding of Src homology 3 domains and potential phosphorylation motifs that could be involved in intracellular signaling pathways (9).
Both Lu and Lu(v13) gps are adhesion molecules that bind laminin (10,11), a protein of the extracellular matrix involved in cell differentiation, adhesion, migration, and proliferation (12,13). They represent the unique receptors of laminin in normal and sickle red blood cells, as well as in progenitor erythroid cells (10,11).
The use of domain deletion mutants provided important information on the structure-function relationship within the NH 2 extracellular domain common to Lu and Lu(v13). It has been demonstrated that at least one Lu blood group antigen is located on each of the five Ig-like domains (14), whereas only the membrane proximal domain 5 is critical for laminin binding (15). However, the functional properties of the intracytoplasmic COOH-terminal domain of Lu and Lu(v13) have not yet been investigated.
Besides erythroid cells, the Lu/B-CAM antigens are expressed in a large variety of tissues. However, immunochemical and Northern blot analyses indicated that Lu and Lu(v13) expression are differentially regulated depending on the nature, the developmental stage, and the normal or pathological status of the tissues and cells investigated. Whereas both the 85-and 78-kDa gps could be detected in red blood cell membrane (3,4), the 85-kDa protein species, and accordingly the 2.5-kb mRNA, was predominantly detected in most normal tissues, with a higher expression in fetal versus adult stages (at least in liver, placenta, kidney, and lung) (1,7). Conversely, immunoprecipitation studies using an anti-B-CAM MoAb revealed both Lu and Lu(v13) gps in ovarian cancer cells (5), and Northern blot analysis revealed an up-regulation of the 4.0-kb transcript encoding Lu(v13) in the epithelial colon carcinoma HT29 cell line (7). On the other hand, immunohistochemical analysis indicated that B-CAM antigens are expressed with a basal pattern in the epithelium of normal colon, prostate, ovary, and thyroid gland and were up-regulated and nonpolarized in ovarian epithelial cancers (5).
Considered together, these results suggest that the predominant expression of the Lu isoform might account for the polarized expression of the Lu/B-CAM antigens observed in normal epithelial cells and that overexpression of the Lu(v13) isoform could be responsible for their nonpolarized expression in epithelial cancer cells.
In an attempt to test these hypotheses, we studied the sorting of Lu and Lu(v13) gp isoforms in polarized epithelial cell lines. We found that Lu gp isoform was targeted to the basolateral membrane of polarized epithelial cells, whereas the Lu(v13) isoform exhibited a nonpolarized expression pattern. In addition, we mapped the basolateral targeting signal of Lu gp to a dileucine motif in the cytoplasmic 40 amino acids specific to Lu gp.
Cell Culture and Transfection-Madin-Darby canine kidney (MDCK) and Caco-2 cells were obtained from the American Type Culture Collection (Manassas, VA) and were grown in Dulbecco's modified Eagle's medium-Glutamax I (Life Technologies, Inc.) supplemented with 10 and 20% fetal calf serum, respectively. Stable MDCK transfectants expressing wild type and mutated Lu gps were obtained after transfection of the relevant expression vectors using Lipofectin reagent (Life Technologies, Inc). Transfected MDCK cells were maintained in culture medium supplemented with 0.6 g/liter neomycin (G418). Lu b -positive cells were detected by flow cytometry using the anti-Lu b MoAb LM342 (16) and amplified by a round of selection using magnetic beads coated with anti-mouse IgG (Dynabeads-M-450, Dynal A.S, Oslo, Norway) as recommended by the manufacturer. Stable clones expressing Lu and Lu(v13) were obtained after limit dilutions. At least two clones of each were grown and then analyzed. For mutated constructs, pools with at least 90% of the cells expressing Lu antigens were analyzed.
Northern Blot Analysis-Total RNA from Caco-2 cells was extracted by Trizol (Life Technologies, Inc.) at day 10 of culture. RNA (50 g) was resolved on formaldehyde agarose gel and transferred to nylon membrane in 10ϫ SSC (1ϫ SSC ϭ 150 mM NaCl, 15 mM sodium citrate). Hybridization with a Lu cDNA probe (7) was performed using the Express Hyb TM hybridization solution (CLONTECH, Palo Alto, CA), as described by the supplier.
Immunofluorescence and Confocal Microscopy-Confluent monolayers of transfected MDCK cells were cultured on 6.5-mm-diameter Costar Transwell polycarbonate filters (pore size, 0.4 mm) (Corning Costar, Acton, MA) for 7 days prior to immunostaining. Cells were fixed in a 1:9 methanol:acetone solution, washed with ice-cold phosphatebuffered saline, and incubated with MoAb LM342 from both sides for 1 h at room temperature. Filters were washed with phosphate-buffered saline and incubated with fluorescein isothiocyanate-conjugated goat anti-mouse antibody (Jackson ImmunoResearch, West Grove, PA) (1: 200 dilution). Samples were examined by confocal microscopy using a Leica TCS equipped with DMR inverted microscope and a 63/1.4 objective.
Domain-selective Biotinylation-Wild type and transfected MDCK cells were plated at high density (5 ϫ 10 5 cells) onto 12-mm-diameter Costar Transwell polycarbonate filters and cultured for 5 days; the media were changed daily. The transepithelial resistance was measured daily using a Millipore Electrical Resistance System apparatus (Bedford, MA). Biotinylation was performed after resistance value had reached the plateau indicating that tight junctions were formed (17). Control cells were grown in parallel under the same conditions and analyzed by confocal microscopy to determine that they have formed a polarized monolayer. Cells were metabolically labeled overnight by adding 150 Ci of [ 35 S]methionine to the medium (Amersham Pharmacia Biotech). Cells were washed with cold phosphate-buffered saline-CM before incubating either the apical or the basolateral side with 0.5 mg/ml NHS-LC-biotin (Pierce) diluted in phosphate-buffered saline with calcium and magnesium (CM) at 4°C two times 30 min each; the nonbiotinylated surface was incubated with phosphate-buffered saline-CM only. Filters were washed three times with phosphate-buffered saline-CM and excised, and cells were solubilized in 1 ml of lysis buffer (1% Triton X-100, 20 mM Tris, pH 8, 150 mM NaCl, 5 mM EDTA, and 0.2% bovine serum albumin) containing protease inhibitor mixture (Roche Molecular Biochemicals) at 4°C for 1 h. 100 l of a 10% suspension of Pansorbin (Staphylococcus aureus cells) (Calbiochem, La Jolla, CA) were added for 20 min, and cell lysates were centrifuged at 15,000 ϫ g for 20 min. 10-l aliquots were counted to ensure that cells had similar total metabolical labeling. Lu gps were immunoprecipitated by incubating the supernatant with protein A-Sepharose (Amersham Pharmacia Biotech) precoated with 10 g of mouse MoAb LM342 overnight. The immunocomplexes were eluted with 10 l of 10% SDS and diluted in 600 l of lysis buffer. 30-l fractions were loaded on SDSpolyacrylamide gel electrophoresis gel, and the cpm of excised Lu gps was counted to ensure that there was no difference in the total amount of Lu proteins in the cells labeled from two different sides. The remaining 570 l were incubated overnight with immunopure-immobilized streptavidin beads. Beads were washed twice with 0.5% Triton X-100, 20 mM Tris, pH 8, 500 mM NaCl, and 0.2% bovine serum albumin, twice with 50 mM Tris, pH 8, and boiled for 5 min in 20 l of Laemmli buffer. 10-l samples were run on 8% SDS-polyacrylamide gel electrophoresis gel under nonreducing conditions, and the gel was dried and exposed for autoradiography.
Caco-2 cells were plated at high density (10 6 cells) onto 24-mmdiameter Costar Transwell polycarbonate filters and cultured for 10 days with the transepithelial resistance measured daily. Filters were used only when the resistance of the monolayer exceeded 250 ⍀⅐cm 2 . The integrity of the monolayer was assessed by measuring the appearance of [ 14 C]inulin in the basolateral medium after adding the tracer at the apical side. Less than 5% of the total radioactivity was recovered in the basolateral medium after 180 min. Biotinylation assay was performed as above.
Biotin Targeting Assays-Delivery of newly synthesized Lu and Lu(v13) gps was followed by performing a membrane targeting assay (18). Briefly, polarized MDCK monolayers expressing Lu or Lu(v13) were cultured onto 12-mm-diameter Costar Transwell polycarbonate filters and pulse labeled with [ 35 S]methionine for 20 min. Cells were incubated for different times in a methionine X-10 tissue culture medium and then chilled to 4°C in NaHCO 3 -free Dulbecco's modified Eagle's medium containing 20 mM Hepes, 0.2% bovine serum albumin. Domain-selective biotinylation and immunoprecipitation were performed as described above.

RESULTS
LU Gene Expression in Caco-2 Cells-Expression and polarization of endogenous Lu and Lu(v13) gps were studied in the human epithelial adenocarcinoma Caco-2 cell line. These cells differentiate and polarize spontaneously in vitro, between day 7 and day 10 of culture (19). Flow cytometry analyses, using LM342 and BRIC 108 anti-Lu b mouse MoAbs, revealed the presence of Lu antigens on the surface of Caco-2 cells (specific antibody binding capacity (SABC) units Ϸ 75,000 versus 2000 -4000 for red blood cells). 2 In order to identify which of the 2.5-and/or 4.0-kb Lu tran-scripts is (or are) expressed in polarized Caco-2 cells, confluent layers of Caco-2 cells were cultured on plastic dishes for 10 days. Northern blot analysis of RNA extracts revealed a strong expression of the 2.5-kb mRNA encoding Lu gp and a very weak expression of the 4.0-kb mRNA encoding the Lu(v13) isoform (Fig. 1A), strongly suggesting that the Lu gp isoform is predominantly expressed at the surface of polarized Caco-2 cells.
Polarization of Lu gp Isoform in Differentiated Caco-2 Cells-The polarization of Lu gps in differentiated Caco-2 cells was analyzed by immunofluorescence combined with confocal microscopy. Confluent Caco-2 cells that had been grown on filters for 10 days were incubated from both sides with anti-Lu b MoAb LM342, and Lu gps were visualized after incubating the cells with fluorescein isothiocyanate-conjugated goat antimouse antibody. En face views of confocal microscopy showed the presence of Lu antigens on the lateral surface of Caco-2 cells (Fig. 1B, b), and their absence from the apical and the basal surfaces, where no fluorescence was detected (Fig. 1B, a  and c). Vertical sections confirmed this observation and clearly showed a lateral membrane expression of Lu antigens (Fig. 1B, d).
To confirm the immunofluorescence staining results and analyze the steady state distribution of Lu gps, domain-selective biotinylation and immunoprecipitation assays were performed. Polarized monolayers of Caco-2 cells were metabolically labeled with [ 35 S]methionine and surface-biotinylated either apically or basolaterally. Lu gps were immunoprecipitated using purified MoAb LM342. Fig. 1C shows that Lu gp isoform of 85 kDa was immunoprecipitated from the basolateral surface but was absent from the apical surface of Caco-2 cells. Conversely, Lu(v13) isoform (78 kDa) was immunoprecipitated from both the apical and basolateral surfaces but was poorly expressed as compared with Lu isoform, which is consistent with the results of the Northern blot analysis (see above). The faint expression of Lu(v13) gp may explain the absence of signal at the apical surface in the confocal microscopy experiment.
Altogether, these results indicate that the Lu gp (85 kDa) is the predominant isoform in polarized Caco-2 cells and is responsible for the basolateral expression of Lu antigens.
Polarization of Transfected Lu and Lu(v13) gps in MDCK Cells-To confirm that Lu gp isoform was expressed in the basolateral domain of epithelial cells and gain further insights into the expression of Lu(v13), the polarization of transfected Lu gps were analyzed in epithelial MDCK cells. Lu-and Lu(v13)-transfected cells strongly expressed the two isoforms as demonstrated by flow cytometry using LM342 MoAb (not shown). Confluent monolayers of clones expressing Lu or Lu(v13) (two clones for each isoform) were filter-grown and allowed to polarize. En face views of confocal microscopy showed a basolateral labeling of Lu gp isoform and a nonpolarized expression of Lu(v13) isoform ( Fig. 2A, a-c). Vertical sections confirmed that Lu gp isoform is almost exclusively localized on the basolateral and, more precisely, on the lateral membrane. Conversely, Lu(v13) is expressed on both apical and basolateral surfaces ( Fig. 2A, d). However, it is noteworthy that Lu(v13) is overexpressed on the apical membrane when compared with its expression on the basolateral surface.
Biotinylation and immunoprecipitation experiments on polarized monolayers of the same clones revealed that the Lu gp isoform (immunoprecipitated with an apparent molecular mass of 85 kDa) is expressed mainly on the basolateral surface (94%), whereas Lu(v13) (78 kDa) is predominantly expressed on the apical surface (72%) (Fig. 2B). The overexpression of Lu(v13) at the apical surface was not due to a general sorting defect or saturation of the basolateral pathway of MDCK-transfected cells because clones used in these experiments expressed similar amounts of recombinant Lu or Lu(v13) gps. These results are in agreement with those of the confocal microscopy and indicate the presence of a basolateral sorting signal in the 40 amino acids specific to the Lu gp cytoplasmic tail, which are absent in Lu(v13) gp.
Targeting of Newly Synthesized Lu and Lu(v13) gps-It is known that nonpolar expression in polarized epithelial cells could result from two different mechanisms. First, proteins exhibiting a nonpolar expression at a steady state can be simultaneously targeted from the trans-Golgi network to the apical and the basolateral surfaces through the direct apical and basolateral pathways. Nonpolar proteins can also be directly addressed to the basolateral compartment, endocytosed, and transported to the apical surface via endosomes (for a review, see Ref. 20). As Lu(v13) exhibited a nonpolar steady state expression in polarized MDCK cells, it was interesting to examine which of these mechanisms is used. Pulse-chase membrane targeting assay was performed to monitor the delivery of newly synthesized Lu and Lu(v13) proteins using MDCK clones expressing the same amounts of each isoform (SABC Ϸ 700,000). Fig. 3A shows that newly synthesized Lu(v13)

FIG. 1. Expression of the Lu gp isoforms in Caco-2 cells. A,
Northern blot of Caco-2 cells. Confluent layers of Caco-2 cells were grown for 10 days on plastic dishes before extracting total RNA. Gel electrophoresis, transfer, and hybridization were performed as described under "Experimental Procedures" using a Lu cDNA probe of 1.5 kb. B, surface distribution of Lu antigens in Caco-2 cells by confocal microscopy. Caco-2 cells were filter-grown for 10 days and processed for indirect immunofluorescence microscopy. The cells were fixed before adding LM342 MoAb to the apical and the basolateral surfaces. Fluorescein isothiocyanate-conjugated immunoglobulin was used as a second antibody. The three top panels show xy horizontal focal sections taken through the apical (a), lateral (b), and basal (c) membranes. The bottom panel (d) shows the corresponding xz vertical section (side view). C, cell surface distribution detected by selective biotinylation. Filtergrown Caco-2 cells were metabolically labeled with [ 35 S]methionine and surface biotinylated from either the apical (Ap) or the basolateral (Bl) side. After lysis, total Lu gps were immunoprecipitated, and biotinylated proteins were isolated with streptavidin beads. The biotinylated gps were subjected to SDS-polyacrylamide gel electrophoresis, and the gel was dried before autoradiography exposure. reached apical and basolateral surfaces simultaneously within 30 min of chase and that delivery to both surfaces occurred with similar kinetics. These results indicated that Lu(v13) was directly transported to each surface domain. In contrast, the Lu gp isoform was predominantly and directly targeted to the basolateral surface domain after 30 min of chase (Fig. 3B). These data demonstrated that the carboxyl-terminal 40-amino acid fragment specific to the 85-kDa Lu gp isoform contains a basolateral sorting signal that acts as a dominant signal, preventing this isoform from taking the apical pathway. However, in contrast to its exclusive basolateral endogenous expression in Caco-2 cells (see above), a small proportion of the recombinant Lu isoform was detected at the apical domain of the transfected MDCK cells. The 10-fold overexpression of transfected Lu gp in MDCK cells (SABC Ϸ 700,000) as compared with the endogenous expression in Caco-2 cells (SABC Ϸ 75,000) might account for the delivery of small amounts of the Lu gp to the apical surface. Alternatively, because delivery to the cell surface can be influenced by both the cell type and the protein features (20), a fraction of the Lu gp might be delivered to the apical surface in MDCK but not in Caco-2 cells.
Mapping of the Basolateral Sorting Signal of Lu gp Isoform-Confocal microscopy and steady state biotinylation assays demonstrated basolateral sorting of Lu gp isoform and a nonpolar expression of Lu(v13) gp in MDCK cells. To identify the determinant(s) responsible for the sorting of Lu gp to the basolateral surface, mutations were introduced into the 40amino acid fragment specific to the cytoplasmic tail of Lu gp isoform (Fig. 4A). Pools of enriched transfected cells, with more than 90% of the cells expressing each recombinant protein, were used for immunofluorescent staining and domain-selective biotinylation assay, as described above (two assays for each mutant). In the first mutant (Lu⌬25), the cytoplasmic tail of Lu gp was deleted for the terminal 25 amino acids. Confocal microscopy (not shown) and selective biotinylation showed a nonpolar expression of Lu⌬25 gp similar to that of Lu(v13), with 68% of the protein expressed on the apical surface (Fig. 4B). The absence of the basolateral polarization of Lu⌬25 gp indicated that the basolateral sorting signal is most probably localized in the 25 amino acids that have been deleted in this mutant.
To define which of the 25 terminal amino acids of Lu gp are necessary for basolateral sorting, we constructed two other mutants in which residues were substituted into alanines (L608A/L609A and S621A) (Fig. 4A). The dileucine motif at position 608 -609 was substituted to dialanine in L608A/ L609A, because it was reported that dihydrophobic motifs, such as leucine-valine and leucine-leucine, might be involved in the basolateral sorting of several proteins (for a review, see Ref. 21). In the second construct (S621A), a substitution of the potentially phosphorylated serine to alanine at position 621 was introduced. Confocal microscopy (not shown) and surface biotinylation revealed that the S621A mutant had a polarized expression similar to Lu gp, with 95% of the protein expressed basolaterally (Fig. 4B). This indicates that potential phosphorylation of serine 621 is not required for basolateral sorting. In contrast, the L608A/L609A mutant was expressed in a nonpolarized manner on both apical and basolateral surfaces, with an overexpression at the apical surface (72%), similar to the apical expression of Lu(v13) (Fig. 4B).
In order to investigate whether complementary amino acid residues could be involved in the targeting of the Lu gp as it was described for some other proteins (22,23), we constructed two other mutants with alanine substitutions in the remaining 15 amino acids specific to the Lu gp (P590A/P593A and S596A/ S598A) (Fig. 4A). Pro-590 and Pro-593 were substituted into Ala-590 and Ala-593 because this motif is a potential binding site for Src homology 3 domains, which could be involved in intracellular signaling pathways (1). The potentially phosphorylated serines (amino acids 596 and 598) were also substituted into alanines. As shown in Fig. 4B, these two mutants exhibited a polarized expression pattern. More than 92% of the proteins were expressed basolaterally, indicating that neither the potential Src homology 3 binding motif nor the potential phosphorylation site at position 596 -598 represent complementary basolateral signals. All of these results clearly indicated the role of the dileucine motif at position 608 -609 in the basolateral expression of Lu gp isoform. DISCUSSION In this study, we have demonstrated that the two Lu gp isoforms resulting from alternative splicing of the primary LU transcript, which differ only in the size of their cytoplasmic tail, exhibit different polarization patterns in epithelial cells. We showed that Caco-2 colon carcinoma cells at day 10 of culture expressed almost exclusively the long tail Lu gp and that the localization of this isoform at the basolateral surface of the In MDCK cells, recombinant Lu gp is also expressed at the basolateral surface, whereas Lu(v13) was expressed in a nonpolarized fashion. This difference in the steady state distribution of the two Lu gp isoforms is associated with the presence of a dileucine motif within the 40 extra amino acids of Lu gp cytoplasmic tail.
The targeting features of Lu and Lu(v13) gps were also investigated. The Lu isoform was shown to be targeted directly to the basolateral surface in MDCK cells, and the Lu(v13) isoform was delivered to the apical and basolateral surfaces with the same kinetics. Therefore, Lu(v13) gp takes the direct apical pathway without transiting by the basolateral domain, as was shown for the polymeric immunoglobulin receptor (pIgR) transfected into MDCK cells (24).
Whether Lu(v13) is delivered in a nonpolarized fashion by default or because it contains both permeable apical and basolateral signals remains to be elucidated. However, Lu(v13) was found to be sorted predominantly apically in MDCK cells, whereas delivery to both surfaces by default should not give more than 50% of Lu(v13) gp on the apical surface as apical: basolateral surface area ratio in MDCK cells is close to 1 (25). It is therefore assumed that Lu(v13) might contain cryptic apical sorting information. It has been reported that the Nglycans on the extracellular domain (26,27) or amino acid sequences within the transmembrane domain (28) of integral proteins represent important determinants for sorting to the apical membrane domain of MDCK cells. Indeed, it has been shown that deletion of the cytoplasmic tail of neural cell adhesion molecule, a basolateral protein that also belongs to the Ig superfamily, resulted in a truncated soluble form that is exclusively secreted apically (29). Further experiments will be required to determine whether some or all of the five N-glycosylation sites and/or the transmembrane domain of the Lu(v13) gp could play a role in its apical sorting.
Mutations into the potential Src homology 3 binding domain, as well as mutations of three potentially phosphorylated serines in the cytoplasmic tail of Lu gp, did not alter the basolateral sorting of this isoform. These results demonstrated that these motifs, which are potentially involved in signal transduction, are not recognized as sorting sequences. In con-

FIG. 3. Surface delivery of Lu(v13) (A) and Lu (B) gps in MDCK cells.
Cells grown on filters were pulsed for 20 min with [ 35 S]methionine and then chased for 30, 45, 60, 90, and 150 min. Cells were surface biotinylated from either the apical (Ap) or the basolateral (Bl) side. immunoprecipitation and autoradiography were performed as in Fig. 1C. cpm represents radioactivity level of excised targeted proteins. trast, we showed that the basolateral sorting of Lu gp is attributable to a dominant basolateral sorting signal contributed by the dileucine motif at position 608 -609. Disruption of this dileucine lateral sorting motif results in the accumulation of Lu gp predominantly at the apical surface (72%). This dominant signal would prevent Lu gp from taking the apical delivery pathway.
Most of the basolaterally targeted proteins rely totally or partially on a tyrosine residue for their basolateral delivery (30 -34). This tyrosine residue is often part of a ␤-turn that is required for basolateral expression of these proteins and may, in some cases, be substituted to aliphatic residues, such as phenylalanine or tryptophan, without impeding the basolateral expression (35). However, some proteins were shown to need a tyrosine-independent motif for their basolateral expression. These basolateral motifs rely on a dileucine motif (36 -39), as demonstrated for the Lu gp isoform, or on motifs including leucine (40,41). Tyrosine-based, as well as dileucine-based, sorting signals can be recognized by clathrin adaptor protein complexes (for a review, see Ref. 21). Interactions between tyrosine-based signals and the medium () chain of these complexes have been demonstrated (42,43), and recent evidence indicates that dihydrophobic signals bind to the ␤-chain rather than the -chain of adaptor complexes (44). A recently described adaptor complex, AP3, also interacts with tyrosine-and dileucine-based signals (45,46). Which of these clathrin adaptor complexes interacts with the dileucine motif present in the cytoplasmic tail of Lu gp is currently under study.
In conclusion, the present demonstration that the two Lu gp isoforms exhibit different polarization expression pattern supports our previous hypothesis (7) that the nonpolarized expression of the Lu/B-CAM antigens observed in some epithelial cancers (5) should result from the overexpression of the Lu(v13) isoform. Further investigations necessary to validate this hypothesis would also help to evaluate the potential role of Lu(v13) in epithelial tumor progression and metastasis.