4F2 (CD98) Heavy Chain Is Associated Covalently with an Amino Acid Transporter and Controls Intracellular Trafficking and Membrane Topology of 4F2 Heterodimer*

4F2, also termed CD98, is an integral membrane protein consisting of a heavy chain linked to a light chain by disulfide bond. We have generated a monoclonal antibody to the mouse 4F2 light chain and cloned the cDNA. It encodes a mouse counterpart of rat L-type amino acid transporter-1, and induces system L amino acid transport in Xenopus oocytes in the presence of 4F2 heavy chain. Transfection studies in mammalian cells have indicated that the 4F2 heavy chain is expressed on the plasma membrane on its own, whereas the 4F2 light chain can be transported to the surface only in the presence of 4F2 heavy chain. 4F2 heavy chain is expressed diffusely on the surface of fibroblastic L cells, whereas it is localized selectively to the cell-cell adhesion sites in L cells expressing cadherins. These results indicate that the 4F2 heavy chain is associated covalently with an amino acid transporter and controls the cell surface expression as well as the membrane topology of the 4F2 heterodimer. Although 4F2 heavy and light chains are expressed coordinately in most tissues, the light chain is barely detected by the antibody in kidney and intestine, despite the presence of heavy chain in a complex form. The results predict the presence of multiple 4F2 light chains.

4F2, also termed CD98, is an integral membrane protein consisting of a heavy chain linked to a light chain by disulfide bond. We have generated a monoclonal antibody to the mouse 4F2 light chain and cloned the cDNA. It encodes a mouse counterpart of rat L-type amino acid transporter-1, and induces system L amino acid transport in Xenopus oocytes in the presence of 4F2 heavy chain. Transfection studies in mammalian cells have indicated that the 4F2 heavy chain is expressed on the plasma membrane on its own, whereas the 4F2 light chain can be transported to the surface only in the presence of 4F2 heavy chain. 4F2 heavy chain is expressed diffusely on the surface of fibroblastic L cells, whereas it is localized selectively to the cell-cell adhesion sites in L cells expressing cadherins. These results indicate that the 4F2 heavy chain is associated covalently with an amino acid transporter and controls the cell surface expression as well as the membrane topology of the 4F2 heterodimer. Although 4F2 heavy and light chains are expressed coordinately in most tissues, the light chain is barely detected by the antibody in kidney and intestine, despite the presence of heavy chain in a complex form. The results predict the presence of multiple 4F2 light chains.
4F2 antigen, also called CD98, has been originally identified as an activation antigen of lymphocytes (1). It is known to be rather ubiquitously expressed in many types of cells and notably in almost all tumor cell lines (2). Early biochemical studies have revealed that 4F2 antigen is a heterodimer consisting of a type 2 glycosylated integral membrane protein of around 80 kDa (heavy chain (H-chain)) 1 and a protein with apparent molecular mass of 37 kDa (light chain (L-chain)) linked by disulfide-bond (2,3). Although 4F2 H-chain cDNA has been previously cloned (4,5), 4F2 L-chain remained unidentified. 4F2 H-chain shares some 30% homology with a broad specificity amino acid transporter (BAT), which activates system b 0,ϩlike amino acid transport (6,7), and is shown to induce system y ϩ L amino acid transport when expressed in Xenopus oocytes (8,9). Since 4F2 H-chain has been indicated subsequently to induce multiple amino acid transport systems (10), there has been speculation that 4F2 H-chain is a specific activator of the transport systems rather than a carrier (11). Besides its relation to amino acid transport, a variety of functional implications have been made on 4F2. For instance, anti-4F2 H-chain antibody has been reported to inhibit growth of some tumor cells (12) and hematopoietic progenitor cells (13). Also, it was indicated to be involved in the virus-induced syncytium formation as well as cell fusion of normal monocytes in the absence of virus infection (14,15). More recently, 4F2 H-chain has been shown to reverse the dominant negative effect of overexpressed cytoplasmic domain of ␤ 1 integrin on the ligand affinity of integrin (16). Although these results imply the involvement of 4F2 antigen in diverse cellular activities, exact mechanisms underlying them remain unknown.
In the present study, we first have generated a monoclonal antibody to mouse 4F2 L-chain and isolated the 4F2 L-chain cDNA by expression cloning using it. 4F2 L-chain consists of 512 residues and is predicted to be a very hydrophobic protein with 11 or possibly 12 membrane-spanning regions. Gen-Bank TM search has revealed that the cDNA is a mouse counterpart of the most recently reported rat gene termed L-type amino acid transporter-1, LAT1. LAT1 cRNA could induce system L amino acid transport in Xenopus oocytes in the presence of rat 4F2 H-chain and has been suggested to be a 4F2 L-chain (17). We have confirmed that the mouse 4F2 L-chain induces high affinity amino acid transport with features of system L in the presence of mouse 4F2 H-chain and have proved that it is associated covalently with 4F2 H-chain by a disulfide bond via cysteine at position 103 of the latter. The 4F2 H-chain has been indicated to be expressed on the cell surface as a monomer on its own, whereas 4F2 L-chain is transported to the plasma membrane only in the presence of 4F2 H-chain. 4F2 H-chain is expressed on the epithelial cell surface of most embryonic tissues in vivo, and the analysis on cultured cells has indicated further that 4F2 H-chain is expressed selectively at cell-cell adhesion sites generated by cadherins. The present results thus reveal a critical role of 4F2 H-chain in the control of intracellular trafficking as well as the cell surface topology of the 4F2 heterodimer and provide a new clue to delineate the mechanisms for its involvement in diverse cellular functions. We also present the results predicting the presence of addi-tional 4F2 L-chains that is distinct from that of LAT1 in some normal epithelial tissues such as kidney and intestine.

EXPERIMENTAL PROCEDURES
Antibodies and Cell Lines-Anti-4F2 H-chain monoclonal antibody (mAb), 14.37, has been reported previously (3). To raise anti-4F2 mAbs for multiple purposes, Armenian hamsters were immunized with pooled SDS-PAGE gel slices corresponding to the 80-kDa 4F2 H-chain or 37-kDa L-chain from the P3U1 cell lysates that had been immunoprecipitated with 14.37 mAb. Hybridoma supernatants were screened by two independent assays: immunoprecipitation of 125 I-labeled P3U1 cell lysates, and immunoblotting of the cell lysates immunoprecipitated with 14.37 mAb. By these procedures, two additional anti-4F2 H-chain mAbs were obtained: 10.10 mAb, which is capable of efficiently immunoprecipitating the 4F2 heterodimer from P3U1 cells as well as the 4F2 H-chain expressed by cDNA transfection, and 10.4 mAb, which is effective for the detection of 80-kDa 4F2 H-chain by immunoblotting. The 10.10 mAb could be used for immunoprecipitation, immunostaining, and immunohistochemistry, whereas the 10.4 mAb could be used for immunoblotting. Another mAb, 10.7, was also capable of immunoprecipitating the 4F2 heterodimer from P3U1 cells. It specifically reacted to the 37-kDa band by the immunoblotting and was indicated to recognize the 4F2 L-chain (see also text). The 10.7 mAb has been shown to stain the cells only after the permeabilization, suggesting that its epitope is in the cytoplasmic region. Anti-E-cadherin (ECCD-2) was purchased from Takara Co.Ltd, Kyoto, Japan, and anti-N-cadherin (NCD-2) was provided by Dr. Takeichi, Kyoto University, Kyoto, Japan. Anti-Myc antibody (9E10) was purified in our laboratory. L cells and those stably transfected with E-cadherin (EL) and N-cadherin cDNA (NL) were also provided by Dr. Takeichi. All cell lines were maintained in Dulbecco's modified minimal essential medium supplemented with 10% fetal calf serum.
cDNA Transfection and Expression Cloning-cDNA library of BAL 17.2 mouse B lymphoma cells was constructed in the expression vector pPISC, 2 and COS cells were transfected with the cDNA library (10 g/5 ϫ 10 6 cells) by electroporation. Cells were harvested by trypsinization 3 days after the transfection and fixed and permeabilized by Fix and Parm (Caltag, S. San Francisco, CA), as instructed by manufacturer. Cells were stained with 10.7 mAb followed by FITC-conjugated anti-hamster IgG (Caltag, S. San Francisco, CA). Positively stained cells were collected by cell sorting with fluorescence-activated cell sorter Vantage (Becton Dickinson, Mountain View, CA). Episomal plasmids were directly recovered from such collected cells by the method described by Davis et al. (18). Briefly, the sorted cells were treated with buffer containing 100 mM EDTA, 10 mM Tris-Cl, pH 8.0, 0.1% SDS, and 100 g/ml of proteinase K at 55°C overnight followed by phenol/chloroform extraction and ethanol precipitation. Samples were suspended in 2 l of water and transformed into bacteria. Plasmids were purified from liquid culture of bacteria and again transfected into COS cells. After three cycles of the procedure, a single plasmid clone, p10.7, was isolated and sequenced. The cDNA transfection into COS and HeLa cells was done using electroporation and CaPO 4 method, respectively.
Plasmid Construction-4F2 L-chain cDNA tagged with Myc epitope at the C terminus was constructed by subcloning synthetic oligonucleotides encoding the epitope tag and a part of cDNA into the 3Ј-end of the cDNA. Single residue mutants of 4F2 H-chain (cysteine at position 103 being substituted for serine, C103S, and cysteine at 325 for serine, C325S) were constructed by the two-step PCR and confirmed by DNA sequencing. For cRNA synthesis, cDNAs of both 4F2 H-chain (3) and 4F2 L-chain were subcloned into pSP73 vector (Promega, Madison, WI). After linearizing the plasmids with XhoI, cRNAs were synthesized by using mMASSAGE, mMACHINE SP6 kit (Ambion, Austin, TX), as instructed by the manufacturer.
Immunofluorescence Staining and Immunohistochemistry-Cells were cultured on the coverslips, rinsed with TBS ϩ (Tris-buffered saline, pH7.4, supplemented with 10 mM CaCl 2 ), fixed with 3% formaldehyde in TBS ϩ , and blocked with 2% bovine serum albumin in TBS ϩ . For double staining, the cells were incubated with anti-4F2 H-chain mAb (10.10) and anti-E-cadherin or anti-Myc for 1 h at room temperature and then with biotin-conjugated goat anti-hamster IgG (Caltag) and FITC-conjugated rabbit anti-rat IgG or anti-mouse IgG (Caltag) for 1 h at room temperature followed by Texas red-avidin (Biomeda, Foster City, CA). The samples were dried, mounted in ProLong antifade kit (Molecular Probes, Inc., Eugene, OR), and analyzed with a confocal laser microscopy (Olympus, Osaka, Japan). Immunohistochemistry was performed as described before (19). Briefly, whole embryos (E14) were fixed with 4% paraformaldehyde at 4°C for 30 min. Frozen sections at 10 -16-mm thickness were preblocked, incubated with 10.10 mAb, and then with biotin-conjugated goat anti-hamster IgG followed by ABC kit.
Measurement of Amino Acid Uptake in Xenopus Oocytes-Amino acid uptake was measured as described (6) with slight modifications. Briefly, five to seven Xenopus oocytes per condition were washed twice in amino acid-free uptake solution (100 mM choline chloride, 2 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , and 10 mM Hepes, pH 7.5). Oocytes injected with cRNAs or water as a control were incubated with 200 l of uptake solution containing 50 M radiolabeled amino acids at 370 KBq/ml for 30 min at 25°C. Amino acid competition experiments were performed by adding 5 mM inhibitors to the uptake solutions. After incubation, oocytes were washed five times with 1 ml of ice-cold wash solution (80 M choline chloride, 20 mM L-arginine, 20 mM L-leucine, 2 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , and 10 mM Hepes, pH 7.5). Each oocyte was then transferred to a vial, dissolved with 200 l of 10% SDS, followed by addition of 3 ml of scintillation solution for scintillation counting.

4F2 Heavy Chain Is Associated Covalently with a System L Amino Acid Transporter by Disulfide
Bond-An anti-mouse 4F2 mAb, 10.7, was produced, that was capable of immunoprecipitating a band at 120 kDa in nonreducing condition and two bands at 80 and 37 kDa positions in reducing condition from the surface-labeled P3U1 cell lysates (data not shown). When the immunoprecipitate of P3U1 lysate with either anti-4F2 H-chain (14.37) or 10.7 mAb was blotted with 10.7, a 37-kDa band was detected (Fig. 1a, left), and conversely, immunoprecipitation with 10.7 resulted in the coprecipitation of 80 kDa 4F2 H-chain (Fig. 1a, right). The results indicate that 10.7 mAb recognizes the light chain of 4F2 heterodimer. We then isolated a 4F2 L-chain cDNA, p10.7, by expression cloning using the mAb as described under "Experimental Procedures." When COS cells were transfected with either 4F2 H-chain (p14.37, Ref. 3) or p10.7 cDNA, an 80-or 37-kDa band, respectively, was immunoprecipitated only with the corresponding mAb as expected. On the other hand, when COS cells were cotransfected with both cDNAs, anti-4F2 H-chain mAb could immunoprecipitate the 37-kDa band reactive to 10.7 mAb in addition to the 80-kDa 4F2 H-chain band (Fig. 1b). Because 4F2 H-chain has only two cysteines at positions 103 and 325 in the extracellular region (5), a single residue mutation for each cysteine to serine was introduced, C103S and C325S, and cotransfected with p10.7 cDNA. As also shown in Fig. 1b, the C103S mutation of 4F2 H-chain abrogated the coprecipitation of the 37-kDa band by anti-4F2 H-chain mAb, whereas the C325S mutation did not. These results have proven that the cDNA indeed encodes the 4F2 L-chain.
The p10.7 cDNA consists of 3456 bp and contains an open reading frame (nucleotides 27-1565) encoding 512 residues (GenBank TM /DDBJ accession number AB17189). The deduced amino acid sequence is highly homologous (98% identity) to the recently reported rat LAT1 (17). The hydrophobicity profile shown in Fig. 2a predicts at least 11 and possibly 12 helical transmembrane domains. As shown in Fig. 2b, injection of 4F2 L-chain cRNA alone into Xenopus oocytes induced negligible Na ϩ -independent uptake of Leu or Arg, whereas 4F2 H-chain cRNA induced Arg uptake as reported previously (8,9). When both cRNAs were coinjected, potent Na ϩ -independent Leu uptake was induced, whereas Arg-uptake tended to be suppressed as compared with that induced by 4F2 H-chain cRNA alone. The Leu uptake in the double cRNA transfectants was almost completely inhibited by Ile, Val, His, and Phe as well as by 2-(Ϫ)-endoamino-bicycloheptane-2-carboxylic acid (BCH), a specific inhibitor of system L transport. A kinetic study revealed that the Na ϩ -independent Leu uptake was saturable and high affinity, with the K m calculated to be around 25 M (Fig. 2b).  ϩ hc), and assayed for the uptake of indicated amino acids 3 days after the injection. Control oocytes (Ϫ) received water. The uptake of amino acids was measured by incubating oocytes with 50 M indicated radiolabeled amino acids for 30 min at 25°C in the uptake solution containing 100 mM choline chloride in place of NaCl. Means Ϯ S.D. of five to seven oocytes are indicated. Center panel, the Na ϩ -independent L-leucine uptake was determined as above in the presence of 5 mM indicated inhibitors using the oocytes that had been injected with both 4F2 L-chain and 4F2 H-chain cRNAs (2.5 ng of each) 3 days before. Means Ϯ S.D. of five to seven oocytes are indicated. Right panel, oocytes that had been injected with both 4F2 L-and 4F2 H-chain cRNAs (2.5 ng of each) 3 days before were incubated with varying concentrations of L-leucine for 30 min, and the uptake was measured as above. The mean base-line uptake of control oocytes was subtracted from that of cRNA-injected oocytes at each concentration. Inset, Eadie-Hofstee plot of the data.

4F2 H-chain Guides the 4F2 L-chain to Plasma Membrane
That Is Independent of Disulfide Linkage-We then examined the intracellular trafficking of each protein. COS cells transfected with either 4F2 L-or H-chain cDNA alone, or with both, were surface-labeled with biotin, lysed, and immunoprecipitated with anti-4F2 H-or L-chain mAb followed by the detection of biotinylated proteins with ABC system. As controls, aliquots of the same cell lysates (one-fourth) were immunoprecipitated similarly and blotted with the corresponding mAbs. The level of biotinylated 4F2 L-chain was found to be marginal as compared with the total 4F2 L-chain in the L-chain single transfectants, whereas the vast majority of 4F2 L-chain was estimated to be expressed on the cell surface in the H-chain/ L-chain double transfectants (Fig. 3a, left panel). In contrast, comparable levels of biotinylated 4F2 H-chain were detected in both H-chain single and H-chain/L-chain double transfectants (Fig. 3a, right panel). Biotinylated 4F2 H-chain in the former was detected as a monomer without covalently associated molecule as 4F2 H-chain (C103S) (Fig. 3b), eliminating the possi-bility that mouse 4F2 H-chain was associated with the endogenous 4F2 L-chain in COS cells and expressed on the cell surface. 4F2 H-chain (C103S), that failed to form disulfidelinkage with 4F2 L-chain, however, was capable of inducing cell surface expression of 4F2 L-chain as efficiently as wild type 4F2 H-chain (Fig. 3a). These results were confirmed by immunofluorescence staining. When 4F2 H-or L-chain cDNA was singly transfected into HeLa cells, 4F2 H-chain was expressed on the cell surface, whereas 4F2 L-chain was remained mostly in the cytosol particularly in the Golgi area (Fig. 4, a versus b). With the cotransfection of 4F2 H-and L-chain cDNAs, on the other hand, 4F2 L-chain was expressed on the cell surface with the same pattern as 4F2 H-chain (Fig. 4, c versus d). An essentially similar effect was obtained by the cotransfection with 4F2 H-chain (C103S) cDNA as well (Fig. 4, e versus f).
These results indicate that 4F2 H-chain functions as a "guidance molecule" for 4F2 L-chain to the plasma membrane, for which the covalent linkage by a disulfide bond is not essential.
4F2 H-and 4F2 L-chains Are Coordinately Induced in Normal Lymphocytes following Activation-In normal mouse lymphocytes, the 4F2 L-chain transcript is induced rapidly following the mitogenic stimulation in vitro with concanavalin A in a coordinated manner as with that of 4F2 H-chain (Fig. 5a). Also, both transcripts are expressed in all leukemic cell lines examined and with similar relative intensities (data not shown). In Fig. 5b, expression profiles of 4F2 H-and 4F2 L-chain transcripts in normal adult organs are shown. Although both mRNAs are expressed rather ubiquitously, the level of 4F2 L-chain mRNA appears to be disproportionally low as compared with that of 4F2 H-chain in kidney, small intestine, and liver (see below).

4F2 H-chain Is Sorted Specifically to the Cell-Cell Adhesion
Sites Generated by Cadherins-The expression pattern of 4F2 on the cells was then investigated. In OTF9 embryonic carcinoma cells, endogenous 4F2 H-chain was found to be located selectively at the cell-cell adhesion sites (Fig. 6b). Because the distribution was found to be nearly identical to that of Ecadherin (Fig. 6a), we intended to examine directly the effect of cadherins on 4F2 H-chain distribution at the cell surface using fibroblastic L cells and those stably transfected with E-cadherin (EL) or N-cadherin (NL). In L cells, 4F2 H-chain was stained diffusely on the surface and E-cadherin was undetectable (Fig. 6, c and d). In EL and NL cells, which exhibited significant cell-cell adhesion, 4F2 H-chain was concentrated at the cell-cell adhesion sites and colocalized with the cadherins (Fig. 6, e, f and g, h). Immunoprecipitation analysis revealed that 4F2 H-chain was associated with 4F2 L-chain in all these cells (data not shown). These results thus suggest that 4F2 H-chain is sorted specifically to the cell-cell adherent membrane sites, most likely together with 4F2 L-chain, once the stable cell adhesion is generated by cadherins.

Expression of 4F2 H-chain in Various Normal Tissues, Localization at Cell-Cell Adhesion Sites in Polarized Epithelial
Cells and Implication for Multiple 4F2 L-chain(s)-Finally, we examined the cellular localization of 4F2 H-chain in various normal tissues. Immunohistochemistry of mouse embryos has indicated that 4F2 H-chain is expressed on the surface of epithelial cells of most tissues, including epidermis (Fig. 7a), the choroid plexus in the brain (b), retina (c), as well as intestinal (d), renal (e), and thymic epithelium (f). In polarized epithelial cells such as in intestine and kidney, 4F2 H-chain expression is restricted apparently at the lateral adhesion sites (Fig. 7, d and  e), consistent with the colocalized expression with E-cadherin in cell lines. We wished to know whether 4F2 H-chain on the cell surface of these tissues is associated with the 4F2 L-chain. Thus far, the only available anti-4F2 L-chain antibody, 10.7 mAb, worked poorly in immunohistochemistry. Therefore, immunoblotting analysis was performed. The 4F2 H-chain is expressed in all embryonic and adult tissues examined (Fig. 8A). The 4F2 L-chain, however, is detected barely in the kidney, intestine, and adult liver, whereas it is expressed strongly in others including brain, testis, and spleen as well as fetal liver (Fig. 8a). Nonetheless, the 80-kDa 4F2 H-chain in kidney and intestine is detected still as a 120-kDa complex in nonreducing condition (Fig. 8b), strongly implying that the 4F2 H-chain forms complex with 4F2 L-chain(s) that is distinct from the LAT1 in these polarized epithelial tissues. DISCUSSION In the present study, we have generated a monoclonal antibody to the mouse 4F2 L-chain and cloned its cDNA. The deduced amino acid sequence has revealed that 4F2 L-chain consists of 512 residues with at least 11 or possibly 12 helical transmembrane domains, calculated molecular mass being 56 kDa. Results 3 indicate that both N-and C-terminal ends of the coding region are retained in the 4F2 L-chain, and thus much faster migration of 4F2 L-chain in SDS-PAGE at 37-kDa position appears to be because of the intrinsic structural features. 4F2 L-chain is highly homologous (98% identity) to the very recently reported rat LAT1 (17) and thus considered to be its mouse counterpart. Rat LAT1 has been shown to induce system L amino acid transport in Xenopus oocytes in the presence of 4F2 H-chain. Based on the functional dependence on 4F2 Hchain, LAT1 is suggested to be a 4F2 H-chain. Our present results have confirmed that the mouse 4F2 L-chain can mediate amino acid transport with typical features of high affinity system L when expressed in Xenopus oocytes together with the 4F2 H-chain. We have indicated further that the 4F2 L-chain is associated covalently with 4F2 H-chain in the cells by a disulfide bond via cysteine at position 103, which is conserved in the 3  4F2 H-chain of mouse, rat, and human (4,5). Thus, it has been proved that 4F2 H-chain is associated covalently with a system L amino acid transporter.
We then addressed the molecular basis for the functional dependence of 4F2 L-chain on 4F2 H-chain. Present results have indicated that 4F2 H-chain alone is expressed efficiently on the cell surface as monomer. In contrast, 4F2 L-chain is expressed minimally at the plasma membrane in mammalian cells, remaining mostly in the Golgi area, and requires 4F2 H-chain to be sorted to the cell surface. The results thus indicate that one of the functions of 4F2 H-chain is to guide 4F2 L-chain to the plasma membrane. Rather unexpectedly, the guidance effect is independent of disulfide-linkage, implying the involvement of noncovalent steric association. A similar mechanism has been proposed for the heterodimeric P-type cation-exchange ATPases, in which a ␤-subunit is responsible for the correct intracellular trafficking of ␣/␤ heterodimeric holoenzymes from ER to the cell surface (20).
Amino acid permeases in lower eukaryocytes are expressed usually as monomeric proteins (21). Membrane expression of permeases in yeast, however, is shown to be controlled by a unique ER-resident protein, SHR3, without which the transport of permeases from ER to plasma membrane is impaired selectively (22). In this aspect, it is noted that a mutation of BAT (M467T), most commonly detected in the patients of type I cysteinuria, results in the defective expression of BAT protein on the cell surface (23). The BAT is reported also to be associated with an as yet undefined 50-kDa protein in the kidney cells (24). It thus seems possible that 4F2 H-chain/BAT family, often called "transport-related" proteins, represents specific "guidance molecules" for selected amino acid transporters to the plasma membrane in mammalian cells. At present, it remains to be seen whether 4F2 H-chain has additional functions as an integral part of the transport carrier.
In normal mouse embryos, 4F2 H-chain has been shown to be expressed prominently in the epithelial cells of most tissue, in addition to the vascular and lymphohematopoietic cells. In the polarized epithelial cells such as in kidney and intestine, 4F2 H-chain expression appears to be restricted at the lateral sites, which primarily depended on cadherins (25). Indeed, immunofluorescence analysis of OTF9 embryonic carcinoma cells in culture has indicated clearly that the 4F2 H-chain is expressed selectively at the cell-cell adhesion sites and colocalizes with E-cadherin. Furthermore, in L cells stably transfected with Eor N-cadherin cDNA, 4F2 H-chain is expressed at the cell-cell adhesion sites, whereas it is expressed diffusely on the surface of L cells without cadherin, indicating that the membrane topology of 4F2 H-chain is regulated by cadherins. have indicated that E-cadherin is coimmunoprecipitated with 4F2 heterodimer from OTF9 cells lysed with mild detergents, suggesting that 4F2 complex is included in the membrane domain generated by E-cadherin. Similar cell-cell adhesion-dependent restriction of the cell surface topology has been reported for Na ϩ /K ϩ -ATPase and Cl Ϫ /HCO3 Ϫ channel (26,27).
Our present results have also implicated the presence of multiple 4F2 L-chain(s) that are associated covalently with 4F2 H-chain. In cells of most tissues including lymphoid cells and most cancer cells, both 4F2 H-and L-chain are detected at comparable levels. In the intestine and kidney, however, 4F2 L-chain is detected barely by the 10.7 mAb. Nonetheless, the 4F2 H-chain is present mostly as a 120-kDa complex rather than an 80-kDa monomer form also in these tissues, suggesting the presence of distinct 4F2 L-chain(s). Molecular heterogeneity in the system L transport activity has been previously demonstrated (28). Because 4F2 H-chain in these polarized epithelial cells is localized selectively on the lateral, but not apical, surface, the yet undefined 4F2 L-chain(s) may be expected to exhibit unique functions, directional solute transports for instance.
The 4F2 antigen has been suggested to be involved in a wide variety of cellular functions, including cellular growth (12,13), virus-induced cell aggregation, and fusion (14,15), and affinity regulation of ␤1-integrins (16). It has been reported also that anti-4F2 H-chain antibodies affect the Ca 2ϩ -influx in sarcolemmal vesicles (29). Although system L transport of the 4F2 heterodimer should play certainly important roles in proliferative cells, including tumor cells, to meet critical nutritional requirements, the relation of it to many other suspected functions remains obscure. Further studies on the guidance mechanisms of 4F2 H-chain for 4F2 L-chain(s) and possibly other proteins, with or without covalent linkage, to the cell surface as well as the analysis on the mechanisms for regulation of membrane topology of 4F2 heterodimer by cell adhesion molecules might provide new clues to delineate the multiple functions of 4F2 antigen in various cell types.