Cell Surface Expression of Calnexin, a Molecular Chaperone in the Endoplasmic Reticulum*

The folding and assembly of nascent proteins in the endoplasmic reticulum are assisted by the molecular chaperone calnexin, which is itself retained within the endoplasmic reticulum. It was up to now assumed that calnexin was selectively expressed on the surface of immature thymocytes because of a particular characteristic of the protein sorting machinery in these cells. We now report that a small fraction of calnexin is normally expressed on the surface of various cells such as mastocytoma cells, murine splenocytes, fibroblast cells, and human HeLa cells. Surface biotinylation followed by chase culture of living cells revealed that calnexin is continuously delivered to the cell surface and then internalized for lysosomal degradation. These results suggest that there is continuous exocytosis and endocytosis of calnexin, and the amount of calnexin on the plasma membrane results from the balance of the rates of these two events. To study the structural requirement of calnexin for surface expression, we created deletion mutants of calnexin and found that the luminal domain, particularly the glycoprotein binding domain, is necessary. These findings suggest that the surface expression of calnexin depends on the association with glycoproteins and that calnexin may play a certain role as a chaperone on the plasma membrane as well.

A quality control mechanism in the endoplasmic reticulum (ER) 1 ensures that properly folded polypeptides and completely assembled oligomeric complexes are transported to the Golgi and beyond. Calnexin, an ER-resident molecular chaperone, binds transiently to nascent proteins, assisting the folding and assembly while retaining proteins that persist in a mal-folded or incompletely assembled state.
The large luminal domain of calnexin contains two sets of repeated structures (1) that recognize mono-glucosylated glycans (2)(3)(4). In the ER quality control machinery, calnexin acts as a constituent of the glucosylation/deglucosylation cycle. During this cycle, monoglucosylated glycoproteins bind transiently to calnexin.
ER chaperones are thought to be localized in the ER by their cognate receptor, which constantly retrieves escaped chaperones from a dynamic intermediate compartment between the ER and the Golgi complex. The C-terminal Lys-Asp-Glu-Leu (KDEL) tetrapeptide for luminal chaperones (5) and the Cterminal di-lysine (Lys-Lys-X-X) motif for membrane-bound chaperones (6) are generally known as the ER retention motif. Calnexin has previously been believed to localize in the ER, because this membrane-bound chaperone has C-terminal charged residues (RKPRRE) that are similar to the ER retention consensus motif and were proven to function as a retention signal (6,7).
However, immature CD4-CD8 thymocytes were recently found to express calnexin on their surface in association with the CD3 complex (8,9). In this regard, calnexin associated with Ig␣/Ig␤ heterodimer is also expressed on the surface of pro-B cells (10). This escape of calnexin to the cell surface was thought to be implicated in the defective ER retention of immature cells that allows many, but not all, resident ER proteins to reach the cell surface (8). Thus, the surface expression of calnexin/receptor complexes appears to be a common feature of both T and B precursor cells, even though its physiological meaning in early lymphocyte development remains to be determined.
We here report that a small fraction of calnexin is normally expressed on the cell surface regardless of cell type, lineage, or maturation stage of the cell. Calnexin on the cell surface is dynamically turned over by endocytosis. The glycoprotein binding domain is suggested to be prerequisite. Our results suggest that a small fraction of calnexin can escape from the ER and be transported to and expressed on the cell surface, probably by interaction with glycoproteins.

EXPERIMENTAL PROCEDURES
Animals-C57BL/6 mice were purchased from Japan CLEA Inc. (Tokyo, Japan) Cells and Antibodies-The immature thymocyte cell line KKF was derived from Gross's leukemia virus-infected BALB/k thymocytes and expresses the pre-T cell receptor complex as described previously (11,12). P815 is a mastocytoma cell line expressing FcR␥ on the cell surface (13). These cell lines were maintained in RPMI 1640 (Life Technologies, Inc.) supplemented with 5% fetal bovine serum (FBS), 100 units/ml penicillin, 100 g/ml streptomycin, and 50 M 2-mercaptoethanol. DAP-3, a mouse fibroblast L cell clone (14), and HeLa cells (ATCC CCL185) were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% FBS, 100 g/ml L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 50 M 2-mercaptoethanol. An anti-mouse CD3⑀ mAb (145-2C11) was kindly provided by Dr. J. Bluestone (University of Chicago, Chicago, IL). SPA-860, a polyclonal antibody to the C terminus of canine calnexin, and anti-␣-adaptin mAb * This work was supported by grants-in-aid for Scientific Research from the Ministry of Education and a grant from Human Frontier Science Program. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Cell Surface Biotinylation, Immunoprecipitation, SDS-PAGE Analysis, and Western Blotting-Cell surface biotinylation, immunoprecipitation, SDS-PAGE, and Western blotting were performed as described previously (12). Cells were solubilized in a lysis buffer (1% digitonin or 1% Nonidet P-40, 50 mM Tris-HCl (pH 7.6), 300 mM NaCl, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 10 mM idoacetamide). To determine the quantity of surface calnexin, biotin-labeled proteins were sequentially precipitated twice with streptavidin-agarose beads (Pierce). The supernatant of the beads was collected, and precipitated proteins were eluted with the SDS sample buffer from streptavidin pellets. Volumes of the samples were adjusted so that each sample was derived from equivalent amounts of starting material. In the blocking experiment of glycosylation, KKF cells were incubated with or without tunicamycin (0.5 g/ml) (Wako Pure Chemical Industries, Osaka, Japan) for 10 h at 37°C.
Endocytosis and Exocytosis Experiments-Endocytosis experiments were performed with chase culture after surface biotinylation. Biotinylated splenocytes were cultured in RPMI 1640 medium containing 10% FBS for 4 h at 0°C or at 37°C. In some experiments, the cells were incubated in the presence or absence of the inhibitor of lysosomal degradation, methylamine (50 mM) or bafilomycin A 1 (0.5 M) (Sigma). Before solubilization with the 1% Nonidet P-40 lysis buffer, dead cells were removed by pelleting through a cushion of lympholyte M (Cedarlane Laboratories, Hornby, Ontario, Canada). Viabilities were consistently greater than 97%.
For exocytosis assay (8), C57BL/6 splenocytes were pretreated with 2 g/ml brefeldin A (BFA) (Wako Pure Chemical Industries) for 1 h at 37°C and then washed twice in Hanks' balanced salts solution (HBSS). The cells were resuspended either in HBSS only or in HBSS containing 2 mg/ml Pronase (Sigma) and incubated at 37°C for 15 min. Pronase treatment was quenched with an equal volume of ice-cold HBSS containing 5% FBS. After three more washes in HBSS containing 5% FBS, any remaining Pronase was inactivated with 0.5 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 10 g/ml leupeptin for 10 min on ice. Pronase-treated cells were resuspended at 10 6 /ml in RPMI 1640 medium containing 10% FBS and cultured for 4 h on ice or at 37°C in the presence or absence of 2 g/ml BFA. After culturing, the cells were washed three times with HBSS and surface-labeled with biotin as above. Before solubilization with 1% Nonidet P-40 lysis buffer, dead cells were removed by lympholyte M treatment as above.
DNA Construction-The expression vector pMX-IRES-GFP was provided by Dr. T. Kitamura (University of Tokyo, Tokyo, Japan), and pcDNA3 neo was obtained from Invitrogen Corp. (San Diego, CA). Polymerase chain reaction (PCR) was employed to construct Tac chimeras and FLAG-tagged deletion mutants of calnexin. Template cDNA was prepared from mouse 2B4 T cell hybridomas with a Superscript II kit (Life Technologies, Inc.). The vector containing cDNA for Tac (interleukin-2 receptor ␣ chain) has been previously described (15). Recombinant Tac-E19 was provided by Dr. J. S. Bonifacino (16). TTC was made by ligating the PCR product of the cytoplasmic tail of calnexin to the 10 amino acids downstream from the transmembrane region of Tac. TCC was also constructed by fusing the PCR product of the calnexin transmembrane domain and cytoplasmic tail to the extracellular domain of Tac. The cDNA of Tac and Tac chimeras were cloned into the pMX-IRES-GFP vector. Constructs containing FLAG epitope tag two bases after the signal sequence cleavage site were made by PCR. F-WT is the full-length calnexin construct. F-⌬N1 and F-⌬N2 were the mutants of calnexin lacking the glycoprotein binding domain (17). F-⌬N1 lacks the region encoding amino acids 3-391 of calnexin, whereas F-⌬N2 lacks amino acids 254 -456 of calnexin. F-⌬C deletes the Cterminal RKPRRE-ER retention motif of calnexin. The resulting cDNA of each FLAG-calnexin construct was cloned into pcDNA3 neo vector. The sequences of all constructs were confirmed with an automatic DNA sequencer (ABI PRISM TM 310 Genetic Analyzer; Perkin-Elmer).
Transfection and Flow Cytometry-Transfections of HeLa cells were performed using the calcium phosphate method (18). 48 h after transfection, cells were removed from dishes by washing once with phosphate-buffered saline and incubating with 20 mM EDTA in phosphatebuffered saline. Cell surface staining was performed using phycoerythrin-conjugated anti-human CD25 (Tac) mAb (Pharmingen, San Diego, CA) or anti-FLAG mAb followed by fluorescein isothiocyanate-conjugated anti-mouse IgG Ab (Organon Teknika, Westchester, PA). For intracellular staining, cells were fixed in 4% paraformaldehyde, phosphate-buffered saline for 10 min at reverse transcription and then permeabilized with 0.1% Triton X-100, phosphate-buffered saline for 10 min at 4°C before incubation with the antibodies. Flow cytometry was performed using a FACScan TM flow cytometer (Becton Dickinson, San Jose, CA), and 10 4 cells were analyzed by the CELLQuest TM program (Becton Dickinson).
Reverse Transcriptase-PCR Analysis-Total RNA was extracted from each transfectant by the AGPC (acid guanidinium-phenol-chloroform) method. The cDNA was prepared with random hexamer primers and reverse transcribed with Superscript II kit (Life Technologies, Inc.). The reverse transcription products were amplified with the specific primers for ␤-actin and Bip under the condition that the PCR reactions were performed in a linear range.

Surface Expression of Calnexin on Various Cells-Calnexin
has not been reported on the surface of cells except for T and B precursor cells (9,10). To test whether this notion is true, we performed surface biotinylation of splenic T cells and KKF, an immature thymocyte cell line. Cell lysates were immunoprecipitated with anti-CD3 mAb or anti-calnexin Ab. As previously shown (12), calnexin was found to be assembled with the pre-T cell receptor⅐CD3 complex on KKF but not on splenic T cells (Fig. 1A, lanes 2 and 5). Unexpectedly, calnexin was expressed on the surface of not only KKF cells but also splenic T cells ( To extrapolate from this observation to other types of cells, we performed similar experiments with surface biotinylation using P815 mastocytomas, murine splenocytes, and DAP-3 fibroblast cells and found calnexin on all of these cells as well (Fig.  1C). Since surface biotinylation might also label proteins present inside dead cells whose plasma membrane was not intact, it is possible that contamination of a small number of dead cells could result in a similar observation. In addition to the use of lympholyte M to separate only live cells after biotinylation, we used an Ab recognizing ␣-adaptin as a control to exclude this possibility. ␣-Adaptin is a subunit of a clathrin-associated protein complex, AP-2, that associates with the cytoplasmic surface of the plasma membrane so that it could not be labeled in live cells by our biotinylation protocol. As expected, the experiment by immunoprecipitation and blotting with anti-␣-adaptin Ab revealed that no biotinylated ␣-adaptin was detected, confirming that calnexin detected by the biotinylation experiments represents the protein present on the surface of the plasma membrane (Fig. 1B).
A Small Fraction of Calnexin Is Expressed on the Cell Surface-We next determined the fraction of calnexin expressed on the plasma membrane. p815, murine splenocytes, and DAP-3 were surface-biotinylated, and the labeled proteins were sequentially precipitated twice with streptavidin-agarose beads to isolate and remove the biotinylated proteins. Precipitated proteins eluted by the SDS sample buffer from streptavidin pellets as well as the residual lysates were subjected to SDS-PAGE and immunoblot analysis with anti-calnexin Ab. Volumes of the samples were adjusted so that each sample was derived from an equal amount of starting material. Surface calnexin was detected in the first streptavidin precipitation (Fig. 2, lanes 1, 4, and 7), whereas no calnexin was detected in the second precipitates (Fig. 2, lanes 2, 5, and 8), indicating that virtually all of the biotinylated calnexin was in the first streptavidin precipitates. The percentages of calnexin on the plasma membrane calculated from this type of experiment were 1.3% in p815 (Fig. 2, lanes 1 and 3), 8.9% in splenocytes (Fig. 2, lanes 4 and 6), and 5.4% in DAP-3 (Fig. 2, lanes 7 and  9), respectively, of the total calnexin existing in the cell. Therefore, it is conceivable that a small percentage of calnexin is expressed on the cell surface regardless of cell type or maturation stage.
Delivery of Intracellular Calnexin to the Cell Surface-The fact that calnexin is expressed on the plasma membrane suggests that the molecule is delivered to the cell surface through the secretory pathway. To test this possibility, cells were first treated with Pronase to digest surface proteins present on cells and then placed into culture at 37°C to allow reestablishment of the surface expression of the proteins (8). After Pronase treatment, calnexin completely disappeared from the cell surface (Fig. 3, lane 1 and 2) and was re-expressed after culturing for 4 h at 37°C (Fig. 3, lane 3). Reprobing the same membrane with the anti-calnexin Ab detected similar amounts of the total cellular calnexin in each lane, suggesting that equivalent amounts of cells were subjected to the experimental procedures. Comparison of the amounts of biotinylated calnexin in lanes 1 and 3 of Fig. 3 revealed that 30ϳ50% of calnexin of the original level on the cell surface reached the plasma membrane during the 4-h incubation. To confirm that the reappeared calnexin was delivered through the ER-Golgi pathway after new synthesis, the transport to the plasma membrane was blocked by BFA during the reappearance stage. As shown in Fig. 3, lane 4, BFA significantly inhibited about 80% of the reappeared calnexin on the cell surface, whereas the total cellular level of calnexin remained at the same level. These data indicate that reappeared calnexin on the cell surface was mainly transported through Golgi after new synthesis. Disappearance of Surface-biotinylated Calnexin-We next examined whether the surface calnexin was stably expressed or continuously endocytosed. Surface-biotinylated cells were cultured for 4 h at 37 or 0°C. After the culture at 37°C, most of the calnexin (70ϳ95%) disappeared from the cell surface (Fig.  4A, lane 3). This was in striking contrast with the T cell receptor ␣/␤ chains, which remained mostly on the cell surface during the same period of time (Fig. 4A, lane 6). The disappearance of calnexin was not observed by incubation at 0°C (Fig. 4, lane 2), demonstrating that calnexin disappeared from the cell surface by dynamic metabolism. The disappearance of the biotinylated surface calnexin by incubation at 37°C could result from endocytosis and degradation within the cells or secretion into the medium. We could not detect any biotinylated calnexin by immunoprecipitation of the culture supernatant after incubation (data not shown). Then, to confirm that surface calnexin is endocytosed and degraded after internalization, the surfacebiotinylated cells were treated by methylamine or bafilomycin, both of which are known to inhibit lysosomal degradation by induce neutralization of the lysosomal acid compartments. Indeed, treatment with either reagent inhibited degradation of biotinylated calnexin (Fig. 4B). 60 and 85% of the disappeared calnexin was recovered onto the cell surface by the treatment of methylamine (lane 3) and bafilomycin (lane 4), respectively.
The results indicate that a significant proportion of the surface calnexin was continuously endocytosed and degraded in the lysosomal compartment. These results suggest that continuous exocytosis and endocytosis of calnexin take place and that the amount of calnexin on the plasma membrane results from the balance of the rates of these two events.
ER Retention Motif of Calnexin Effectively Retains the Molecule in the ER-The surface expression of calnexin could result from the relative shortage of the ER retention receptors for the di-lysine motif. Alternatively, the di-lysine motif of calnexin might not be strong enough to sequester calnexin in the ER efficiently. To provide supporting evidence for at least one of these possibilities, we transfected Tac (interleukin-2 receptor ␣ chain) and Tac-tagged chimeric constructs by using retrovirus vector pMX containing IRES-GFP (Fig. 5A) into HeLa cells. HeLa cells were also shown to express calnexin on the cell surface (Fig. 5B). Tac-E19 has the luminal and transmembrane domain of Tac fused to adenovirus gene product E19, which contains the C-terminal KKMP-ER retention motif (16). TTC has the calnexin cytoplasmic tail with the transmembrane portion and luminal region of Tac, whereas TCC has the transmembrane and cytoplasmic domain of calnexin with the luminal region of Tac. Transient transfection of these chimeric molecules into HeLa cells resulted in similar levels of GFPpositive population and chimeric protein expression, as detected by intracellular staining with anti-Tac Ab (Fig. 5C, right  panels). To estimate the expression levels of Tac chimera proteins, particularly TTC and TCC, transfected HeLa cells were analyzed by immunoprecipitation and blotting with anti-C terminus calnexin Ab. Tac chimera proteins for TTC and TCC were calculated to be expressed 8 and 11 times higher than the endogenous calnexin, respectively. (Fig. 5D). Surface staining with anti-Tac Ab revealed that neither of them, besides Tac, was expressed on the cell surface (Fig. 5C, left panels). Surface biotinylation experiments also confirmed that the surface expression of endogenous calnexin did not change by transfection of Tac-calnexin or Tac-E19 chimeras (data not shown). These results suggest that the ER retention motifs of calnexin and E19 efficiently retain molecules, containing them in the ER,  4 -6), and DAP-3 (lanes 7-9) were lysed with 1% Nonidet P-40 and then sequentially precipitated twice with streptavidin-agarose beads. Streptavidin pellets were eluted in SDS sample buffer, and the eluates were diluted with lysis buffer to adjust the volumes to their residual supernatants. Equal aliquots derived from the samples were subjected to SDS-PAGE and immunoblot analysis with anti-calnexin Ab.  30.4% for Tac, 28.0% for Tac-E19, 37.2% for TTC, and 26.6% for TCC) were stained with phycoerythrin-anti-Tac mAb, and their histograms were shown. Intracellular staining of each transfectant with phycoerythrin-anti-Tac mAb is shown in the histograms in the right panel. It was confirmed that GFP levels were correlated with the expression of Tac proteins. D, aliquots of transfected HeLa cells used in C were lysed in 1% Nonidet P-40 lysis buffer and immunoprecipitated (IP) and blotted with anti-C-terminal calnexin Ab (CNX-C). Densitometric analysis showed that the expression of transfected Tac-chimeras was 3-fold higher than the endogenous calnexin for both TTC and TCC. Since the percentage of Tac chimera-positive cells in the total population was 37.2% for TTC and 26.6% for TCC, the Tac chimeras were expressed at 8 and 11 times higher level than the endogenous calnexin for TTC and TCC, respectively. and that there were a sufficient number of ER retention receptors.
The Glycoprotein Binding Domain Is Involved in Expression of Calnexin on the Cell Surface-The above observation suggested that the luminal domain of calnexin is important for its surface expression. When KKF cells were surface-biotinylated after treatment with tunicamycin, calnexin expression on the cell surface was reduced to approximately 40% that of cells with no treatment (Fig. 6A). This result also suggests that the surface expression of calnexin seem to be regulated by protein glycosylation. Therefore, we postulated that the glycoprotein binding domain of calnexin might regulate the cell surface expression. To test this possibility, we constructed the FLAGtagged full-length calnexin (F-WT), the FLAG-tagged calnexin with the N-terminal deletion including the glycoprotein binding sites (F-⌬N1 and F-⌬N2), and the FLAG-tagged calnexin with the C-terminal deletion including the di-lysine like motif (F-⌬C) (Fig. 6B). These constructs were transiently transfected into HeLa cells, and the surface and intracellular expressions of calnexin were analyzed. As compared with F-WT and F-⌬C, F-⌬N1 and F-⌬N2 were expressed on the cell surface at a strikingly lower level (Fig. 6, C and D) despite the fact that the total cellular levels of chimeric proteins were similar (Fig. 6C, right panels). In this experiment, it remains the possibility that F-⌬⌬N1 and F-⌬⌬N2 were unfolded states. To clarify this point, we investigated whether overexpression of FLAG-tagged constructs could induce the unfolded protein response (19,20), which results in the up-regulation of genes encoding ER-resident chaperones such as calnexin, calreticulin, and Bip (GRP78). Since the endogenous calnexin could not be distinguished from the transfected calnexin in the transfectants, we examined unfolded protein response by analyzing the expression of Bip. The expression levels of Bip mRNA in all transfectants were increased by the treatment with tunicamycin, which is often used as a positive inducer of ER stress (Fig. 6E). In contrast, the level of Bip expression was not changed in transfectants expressing FLAG-tagged calnexins (Fig. 6E, lanes 1, 3,  5, and 7). These results suggest that the expression of FLAGtagged constructs did not alter the quality control systems in the ER. Thus, the glycoprotein binding domain is suggested to be important for the expression of calnexin on the cell surface. DISCUSSION Recent reports have shown that several soluble ER resident proteins are found on the cell surface. These are calreticulin (CRT), protein disulfide isomerase (PDI), 78-kDa glucose-regulated protein (GRP78), and 94-kDa glucose regulated protein (GRP94) (21)(22)(23)(24). These ER luminal proteins commonly have an ER retention motif, KDEL, at the C terminus, suggesting that this type of ER retention is not efficient enough to keep the KDEL-containing proteins strictly in the ER. In this report, we demonstrate that calnexin, a membrane-spanning protein having another type of ER retention motif, di-lysine, can also escape from the ER to be expressed on the plasma membrane of various cells. Calnexin has previously been reported to be expressed on the cell surface of immature lymphocytes (8 -10, 12). In these cells, calnexin was found to predominantly associate with pre-T or pro-B cell receptor complexes. It has been postulated, therefore, that the surface expression of ER resident proteins is a characteristic of immature lymphocytes and that the surface expression of calnexin might have an important role in the development and/or physiology of these cells. In this regard, re-evaluation might be needed for the meaning of the surface expression of calnexin on immature lymphocytes in light of our finding of its expression on other cell types as well.
Surface biotinylation followed by chase culture of living cells revealed that there is a continuous endocytosis of calnexin. The amount of calnexin expressed on the cell surface is only several percent of the total cellular calnexin, most of which reside within the ER. Since biotinylated calnexin was hardly detectable after the chase culture (Fig. 4), it is likely that the internalized calnexin is not subjected to recycling to the surface or returning to the ER but, rather, is delivered to lysosomes for degradation. This was confirmed by the observation that neutralization of lysosomal acidic compartments with neutralizing reagents prevented internalized calnexin from degradation and retained the protein level of calnexin. COPI, cytosolic coat complex, is known to directly interact with the ER retention motif (25,26). COPI coats are known to play a significant role in the retrograde transport of ER resident proteins, and it functions in many different traffic pathways including early to late endosomes (27,28). Therefore, an intriguing possibility would be that the endocytosed calnexin is delivered to the lysosomal compartment by association with COPI; further experiments will be required to elucidate the mechanism of endocytosed calnexin and for lysosomal targeting.
It is known that a small fraction of residual proteins of intracellular organelles other than the ER is also expressed on the plasma membrane. These include a lysosomal residual protein, lamp-1 (29 -31), and a resident of the trans-Golgi network (TGN), TGN38 (32)(33)(34). These proteins contain the tyrosine-based sorting signal in their cytoplasmic tails, which interacts with clathrin-adaptor complexes. The intracellular localization of proteins mediated by these tyrosine signals is known to be saturable (18). In contrast, the present study showed that this is not the case for the di-lysine motif. The overexpression of exogenous di-lysine motif-containing proteins resulted in the increased surface expression of neither exogenous proteins themselves nor endogenous proteins with a di-lysine motif (Fig. 5). 2 Since di-lysine signals interact with COPI, components of another class of coated vesicles, the results suggest that COPI is not easily saturated and probably exists much more abundantly than the AP complex in cells. Another attributable difference between these two sorting systems is their sites of function in the cell. In the case of clathrincoated vesicles, their site of sorting is between trans-Golgi network and the plasma membrane, which resides near the end of the secretory pathway. By contrast, COPI is mostly concentrated on the cis-Golgi membrane, and this location is at the starting site of the secretory pathway, apart from the plasma membrane.
A major question then is how a fraction of calnexin escapes from the ER to reach the plasma membrane. Our data using several deletion mutant calnexins suggest that the glycoprotein binding site in the luminal domain of calnexin appears important for its exit from the ER. The possibility that the anti-FLAG mAb is shown as the histograms in the right panel. Each numeral indicates the percentage of gated cells. D, relative evaluation of FLAG expression on the transfected cells. The mean fluorescence intensity (MFI) of non-stained cells was subtracted from that of each FLAG-stained transfectant and is shown as bar charts. E, unfolded protein responses in FLAG-tagged calnexin-transfected cells. HeLa cells were mocktransfected and transfected with F-WT, F-⌬N1, and F-⌬N2, respectively. 2 days after transfection, cells were treated with (ϩ) or without (Ϫ) 0.5 g/ml tunicamycin for 40 h as a positive inducer of ER stress. The expression of endogenous ␤-actin and Bip were determined by reverse transcription-PCR analysis. failure of such deletion mutant calnexins to be expressed on the cell surface might be related to the unfolded protein structure of these mutants is unlikely because the expression of the mutants did not induce any unfolded protein response. Therefore, the simplest explanation would be that calnexin can escape from the ER through a specific binding to glycoproteins as a chaperone, and this association may cover the retention signal. The association of the CD3 complex in immature thymocytes as well as the Ig␣/Ig␤ heterodimer in pro-B cells may hinder the di-lysine motif of calnexin from the binding with COPI. In this study, however, we could not detect any specific molecule(s) associated with calnexin at a semi-stoichiometric amount on the surface of various cells under the condition that detected the association of calnexin with the CD3 complex on the immature thymocyte cell line. It is possible that calnexin associates with a small amount, but yet a variety, of glycoproteins in various type of cells. It provides a striking contrast to the situation in immature lymphocytes in which the surface calnexin largely associates with pre-T or pro-B receptor complexes 2 and may emphasize the biological importance of the specific interaction of calnexin-immune receptor complexes.
It was recently reported that the expression of calnexin is up-regulated when cells contact a substrate that induces cell adhesion (35). Moreover, calreticulin, a luminal homologue of calnexin, is known to be essential for integrin-mediated calcium signaling and cell adhesion (36 -38). There is considerable overlap of the substrate glycoproteins between calnexin and calreticulin, and they are able to associate with the same protein simultaneously or sequentially. A small fraction of calnexin, which could escape from ER retention to reach the cell surface, may act as an essential modulator both for integrinadhesive functions and for integrin-initiated signaling.
In conclusion, the present study demonstrates that a variety of cells previously thought to keep calnexin exclusively within the ER express calnexin at a low level on their surface and that the luminal glycoprotein binding region seems to be responsible for the expression. Further experiments will be required to elucidate its physiological implications.