Truncation of the C Terminus of the Rat Brain Na+-Ca2+ Exchanger RBE-1 (NCX1.4) Impairs Surface Expression of the Protein*

The C terminus of the rat brain Na+-Ca2+ exchanger (RBE-1; NCX1.4) (amino acids 875–903) is modeled to contain the last transmembrane α helix (amino acids 875–894) and an intracellular extramembraneous tail of 9 amino acids (895–903). Truncation of the last 9 C-terminal amino acids, Glu-895 to stop, did not significantly impair functional expression in HeLa or HEK 293 cells. Truncation, however, of 10 amino acids (Leu-894 to stop; mutant C10) reduced Na+gradient-dependent Ca2+ uptake to 35–39% relative to the wild type parent exchanger, and further truncation of 13 or more amino acids resulted in expression of trace amounts of transport activity. Western analysis indicated that Na+-Ca2+ exchanger protein was produced whether transfection was carried out with functional or non-functional mutants. Immunofluorescence studies of HEK 293 cells expressing N-Flag epitope-tagged wild type and mutant Na+-Ca2+exchangers revealed that transport activity in whole cells correlated with surface expression. All cells expressing the wild type exchanger or C9 exhibited surface expression of the protein. Only 39% of the cells expressing C10 exhibited surface expression, and none was detected in cells transfected with non-functional mutants C13 and C29. Since functional and non-functional mutants were glycosylated, the C terminus is not mandatory to translocation into the endoplasmic reticulum (ER). Endoglycosidase H digestion of [35S]methionine-labeled protein derived from wild type Na+-Ca2+ exchanger and from C10 indicated that resistance to the digestion was acquired after 1 and 5 h of chase, respectively. C29 did not acquire detectable resistance to endoglycosidase H digestion even after 10 h of chase. Taken together, these results suggest that the “cellular quality control machinery” can tolerate the structural change introduced by truncation of the C terminus up to Ser-893 albeit with reduced rate of ER→Golgi transfer and reduced surface expression of the truncated protein. Further truncation of C-terminal amino acids leads to retention of the truncated protein in the ER, no transfer to the Golgi, and no surface expression.

Hydropathy analysis of the cloned NCX1 (Na ϩ -Ca 2ϩ exchanger) gene indicated that the protein can be organized into 12 transmembrane helices, the first of which was suggested to be a cleavable signal peptide (2). This was supported by microsequencing of the N terminus of the purified bovine heart Na ϩ -Ca 2ϩ exchanger (3), indicating that the first amino acid of the mature protein corresponds to amino acid 33 of the cloned gene. By carrying out a series of deletions between the initiating methionine and Asp-33 (the first amino acid of the mature protein RBE-1), 1 we have shown (4) that the signal peptide is not mandatory for functional expression of the protein in HeLa cells. Similar results have also been reported in HEK 293 cells (5) and in oocytes and Sf9 cells (6). We have also shown that the signal peptide truncated protein is glycosylated in vitro when dog pancreatic microsomes are added to a reticulocyte lysate expression system (4), suggesting that the nascent protein translocates into the lumen of endoplasmic reticulum (ER) 2 also in the absence of its signal peptide. To explore further the involvement of different domains of the Na ϩ -Ca 2ϩ exchanger in its functional expression, we have examined the role of its C terminus. To do so we have sequentially truncated 9 -29 of its 903 amino acids, starting with mutant C9 (see Table I) in which the putative extramembraneous tail of the protein was missing. Our results show that functional expression is severely impaired when 11 or more amino acids are truncated from the exchanger's C terminus. Immunoreactive protein derived from the non-functional mutants translocates to the lumen of the ER, where it is glycosylated. However, unlike the wild type exchanger protein, it does not proceed further to the Golgi and is not expressed in the surface membrane.
Mutagenesis was confirmed by sequencing 300 -600 bases upstream and downstream of the mutation. To ensure that no other mutation beyond the planned occurred, BglII-SacI fragments containing the desired mutation were subcloned into BglII-SacI-digested parent clone rbe-1. Subcloning into pcDNA3 involved preparation of HindIII-digested arms of pcDNA3 (Invitrogen) into which HindIII-digested clone rhe-1 (9) was ligated. Orientation was ensured by identification of appropriate restriction fragments. This construct was digested with Eco47III-AgeI and ligated with Eco47III-AgeI-digested rbe-1. C-termi-nal truncated mutants in pBluescript were excised with HindIII and NotI and ligated to HindIII-NotI-digested arms derived from in pcDNA3. This strategy ensured that both 5Ј-and 3Ј-untranslated regions of all the mutants were identical.
Insertion of the Flag Epitope Tag DYKDDDDK (Eastman Kodak Co.)-Insertion of the Flag epitope into the N terminus of the cloned rat brain Na ϩ -Ca 2ϩ exchanger (in pBluescript) after Gly-8 has been described (10). From this construct the cloned gene was excised with KpnI-NotI and ligated with KpnI-NotI-digested pcDNA3. This clone was digested with AgeI-NotI and ligated with the insert derived from AgeI-NotI-digested C-terminal truncated mutants. The Qiagen Midi plasmid preparation kit was used to prepare plasmid DNA.
Expression of the Cloned Na ϩ -Ca 2ϩ Exchanger-HEK 293 (human embryonic kidney) cells (ATCC CRL-1573) were transfected by the CaP i (11) method. 10 g of plasmid DNA was used to transfect about 0.8 ϫ 10 6 cells. Transfection of HeLa cells was done as described (4,8).
Determination of Na ϩ -Ca 2ϩ Exchange Activity in HEK 293 Cells-Cells were grown in 12-well or 60-mm dishes. 48 h after transfection, intracellular Na ϩ was raised in the cells by 10- 4 and their 45 Ca 2ϩ content was determined in a liquid scintillation counter. In some experiments, the Na ϩ -preloaded cells were collected by centrifugation, and suspended in a minimal volume of the Na ϩ preloading solution (without nystatin and MgCl 2 ,). Part of these cells was used to determine Na ϩ gradient-dependent Ca 2ϩ uptake, part for protein determination, and part for Western analysis. Transport was determined by suspending 1.6 l of cells (about 50 g of protein) into 50 l of either 0.16 M NaCl or 0.16 M KCl, 0.02 M Tris-HCl, pH 7.4, 25 M 45 CaCl 2 , and 1 mM ouabain. Reactions were terminated by filtration through Schleicher & Schuell BA-85 0.45-m filters, and their 45 Ca 2ϩ content was determined. Mutants were always tested in parallel with the parent wild type clone. Each transport measurement was done in triplicate.
Determination of Na ϩ -Ca 2ϩ Exchange Activity in HeLa Cells-Determination of Na ϩ gradient-dependent Ca 2ϩ uptake was carried out exactly as described previously (4,8,9).
Reconstitution of Transfected Cell Proteins-Cells expressing the wild type and C-terminal truncated mutants of the Na ϩ -Ca 2ϩ exchanger were reconstituted into exogenous brain phospholipids. Reconstitution and determination of Na ϩ gradient-dependent Ca 2ϩ uptake has been described in (4,9).
Western Blot Analysis-AbO-8, an antibody obtained from a rabbit immunized with a pentadecapeptide derived from amino acids 649 -663 of the rat heart exchanger (10), was used. Preparation, testing, and purification of the antibodies were done as described in Refs. 4 and 10. Antigen antibody complexes were detected by 125 I-Protein A (NEN Life Science Products). Quantitative analysis of the protein profile was carried out with the Fujix Bas 1000 phosphorimager using the Tina 2.07 analysis program.
Pulse-Chase Experiments-42 h after transfection, HEK293 cells were washed twice with methionine-free DMEM (Dulbecco's modified Eagle's medium) and kept with the same solution for 2 h in the 5% CO 2 incubator at 37°C. Labeling of cell proteins was carried out by replacing the medium with fresh methionine-free DMEM to which 6 Ci of [ 35 S]methionine/ml of medium was added. Unless stated otherwise, labeling was carried out for 45 min. The chase was started by aspiration of the [ 35 S]methionine-containing medium, followed by three washes with DMEM containing 10% fetal calf serum and incubation in fresh DMEM to which 10% fetal calf serum and 0.2 mg/ml methionine were added for the times specified. The cells were washed twice rapidly in 2 ml of PBS at 4°C and dissolved in a minimal volume of RIPA (0.15 M NaCl, 0.1 M Tris-HCl, pH 8.0, 0.1% SDS, 1% Triton X-100, 0.5% deoxycholate, 0.001 mM EDTA), to which 100 g/ml PMSF was added (12). The cells were kept on ice for 30 min and centrifuged for 15 min at 19,000 ϫ g at 4°C. The pellet was discarded, and the solubilized cell extract was used for immunoprecipitation.
The Effect of Tunicamycin on Na ϩ -Ca 2ϩ Exchanger Protein Expression-Tunicamycin at a final concentration of 10 ng/ml was added to the methionine-free starvation medium and to the [ 35 S]methionine labeling medium for 120 min. The cells were washed twice rapidly with 2 ml of PBS at 4°C and dissolved in a minimal volume of RIPA containing PMSF, after which they were kept on ice for 30 min and centrifuged for 15 min at 19,000 ϫ g at 4°C. The pellet was discarded, and the solubilized cell extract was used for immunoprecipitation.
Immunoprecipitation of the Na ϩ -Ca 2ϩ Exchangers-Protein A-Sepharose CL-4B beads (Sigma) loaded with purified antiserum overnight at 4°C by gentle shaking were used to immunoprecipitate the Na ϩ -Ca 2ϩ exchanger protein. For cell extract obtained from a confluent 60-mm culture dish, 30 mg of Protein A-Sepharose beads and 30 l of antiserum were used. The antibody-loaded beads were washed once in RIPA and incubated with the RIPA cell extract overnight at 4°C with gentle shaking. The beads were pelleted, washed four times with 1 ml of RIPA, and washed once with 1 ml of 0.1 M NaCl, 0.02 M Tris-HCl, pH 7.5, 0.001 M EDTA. The proteins were eluted from the beads with 10ϫ Laemmli sample buffer by boiling for 5 min and separated by SDS-PAGE.
Endoglycosidase H Digestion-Endoglycosidase H f (recombinant) was purchased from New England Biolabs. The immunoprecipitated proteins were released from the beads by boiling in 5% SDS, 2% ␤-mercaptoethanol, after which the concentrations of SDS and ␤-mercaptoethanol were reduced to 1% and 0.5%, respectively. Digestion with the enzyme was carried out as specified by the manufacturer.
Surface Biotinylation-NHS-SS-Biotin (Pierce 21331) was used to biotinylate surface membrane proteins of transfected HEK 293 cells in situ as described in Ref. 13. rbe-1-or C29-transfected cells grown in 6or 12-well plates were used to determine the surface expression by biotinylation in situ. After solubilization the entire contents of the well were loaded on streptavidin beads. The amount of NH-SS-Biotin and streptavidin needed to capture the biotinylated proteins was precalibrated by varying their amounts relative to RBE-1-transfected cells. At the end of the biotinylation reaction, excess reagent was quenched with PBS, which also contained 0.1 mM CaCl 2 , 1 mM MgCl 2 , and 100 mM glycine. The cells were lysed with a solution containing 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1% SDS, 0.1 mM PMSF, 0.01 mg/ml pepstatin A (Sigma), and 0.02 mM leupeptin (Sigma). SDS concentration was lowered by a 10-fold dilution of the lysate with the same solution but without SDS. The lysate was loaded on Immu-noPure streptavidin beads (Pierce 20349), and was gently shaken overnight at 4°C. The biotinylated proteins were released from the beads by heating 10 min at 85°C with Laemmli sample buffer, and Western analysis was carried out as described above.
Cellular Localization of the Na ϩ -Ca 2ϩ Exchanger-Indirect immunofluorescence of N-terminal Flag-tagged wild type and mutated Na ϩ -Ca 2ϩ exchangers expressed in HEK 293 cells was carried out as described for HeLa cells in Ref. 10. Nuclei were stained in 1 g/ml Hoechst (bisbenzimide, Sigma). Immunofluorescence microscopy was performed using epifluorescence optics (ZeissAxioscope) or confocal laser scanning microscopy (Zeiss LSM 410 system) attached to the Zeiss Axiovert 135M inverted microscope. The system was equipped with a 25-milliwatt air-cooled argon laser (488-nm excitation line with a 515-nm long pass barrier filter) for the excitation of green (fluorescein isothiocyanate) fluorescence. Blue fluorescence was excited with 364-nm line UV laser with LP 397 barrier filter. Double-labeled specimens were excited with both laser lines and monitored simultaneously using dual detectors and the Zeiss-supplied filter block combination with dichroic beam splitter and emission filters. The fluorescence was collected by employing a 63ϫ/1.2 C-Achromat water immersion lens (Zeiss). Autofluorescence of the specimens was set to background level. To reduce the visual noise, each confocal optical section was performed in the fast line-scan acquisition mode (512 pixels/line) by averaging of 8 images, before the final image was produced on the monitor. Images simultaneously scanned on the two channels were automatically merged to produce false single multicolor profile. The image analysis of the confocal images including pseudo color representation, brightness, and contrast level were carried out using the standard Zeiss software package and the Adobe Photoshop 4.0 program (Adobe Systems, Inc., Mountain View, CA).

RESULTS
Based on hydropathy analysis (14), the profile-fed neural network system (15), and the recently published revised topological models based on scanning cysteine accessibility method (16,17), the 31 C-terminal amino acids (Table I) of the NCX1 gene are modeled to contain the last transmembrane ␣ helix (10) and an extramembraneous tail. To examine the role of this protein segment in functional expression, sequential deletions starting with the last 9 amino acids, Glu-895 to stop (mutant C9), and ending with Leu-875 to stop (mutant C29) were carried out as described under "Experimental Procedures." Truncation of the last 9 C-terminal amino acids (C9) impaired only slightly functional expression (Fig. 1, A or B) of the cloned exchanger. Its transport activity in HeLa cells was 96.87% (S.D. ϭ 11.37) and in HEK 293 cells 82.8% (S.D. ϭ 15) when compared with that of the wild type exchanger RBE-1 (100%). Further truncation, however, of a single amino acid (mutant C10) resulted in a sharp decrease in transport activity, to about 35-39%, relative to that of the wild type exchanger. Truncation beyond the last 11 amino acids resulted in expression of trace amounts of transport activity. The reduced transport activity of the C-terminal truncated mutants did not depend on mode of transfection, cell line, or the plasmid used for expression. In Fig. 1A, transport activity was determined in the DOTAP-mediated transfection of VTF-7-infected HeLa cell expression system, using pBluescript (Stratagene) as the cloning vector. In Fig. 1B, the transport activity was measured in CaP i -mediated transfection of HEK 293 cells using pcDNA3 as the cloning vector. Similar results to those shown in Fig. 1 were obtained when proteins derived from HeLa or HEK 293 cells transfected with RBE-1 or its C-terminal truncated mutants were reconstituted into exogenous brain phospholipids. Table  II shows the relative transport activities of proteins derived from HeLa and HEK 293 cells transfected with the parent Na ϩ -Ca 2ϩ exchanger (taken in each experiment as 100%) and that of each of its C-terminal truncated mutants (presented in relative values). It can be seen that reconstitution did not rescue the Na ϩ gradient-dependent Ca 2ϩ uptake of the Cterminal truncated mutants, although following reconstitution the transport activity of C13 and C29 expressed in HEK 293 cells was somewhat higher than in HeLa cells and than in whole cells. Moreover, the relative transport activities of C9 and C10 were lower than in whole cells.
To examine whether the decrease in transport activity of the C-terminal truncated mutants resulted from impaired Na ϩ -Ca 2ϩ exchanger protein synthesis, Western analysis was carried out. Fig. 2 shows that similar amounts of immunoreactive exchanger protein were detected whether HeLa or HEK 293 cells were transfected with wild type DNA or with any of the C-terminal truncated mutant DNAs.
To study the cellular distribution of immunoreactive Na ϩ -Ca 2ϩ exchanger protein, the Flag epitope was inserted into the extracellular N terminus (10) of the wild type and C-terminal truncated clones. The transport activity of the N-Flag-tagged (FN) exchangers was similar to that of the corresponding parent clones (not shown). Fig. 3 shows the immunofluorescence detected in HEK293 cells expressing FN-RBE-1 (Fig. 3A), FN-C10 (Fig. 3B), and FN-C29 (Fig. 3C). It can be seen that only in FN-RBE-1-and in FN-C10-expressing cells is immunofluorescence detected without permeabilizing the cells (panels 1 and 2 of Fig. 3, A and B). 1 and 2, Fig. 3C). When, however, cells were permeabilized, fluorescence was detected not only in RBE-1-and C10-expressing cells (panels 3 and 4 of Fig. 3, A and B), but also in C29-expressing cells (panels 3 and 4, Fig. 3C). Studies with FN-C9-expressing cells were similar to those with FN-RBE-1expressing ones and with FN-C13-expressing cells were similar to those with C29-expressing cells (data not shown).

No such staining is detected in FN-C29-expressing cells (panels
These experiments suggest that only in cells expressing the parent exchanger RBE-1, C9, or its partially functional mutant C10 is the transporter in the surface membrane. Moreover, in RBE-1-expressing cells and in C9, similar numbers of cells were stained without and with permeabilization, whereas, in C10-expressing cells, considerably fewer cells stain in the absence of permeabilization than in its presence.  Fig. 1. We have also tested the possibility that surface expression of N-Flag-tagged C29 was not detected since its topology was "scrambled" and its N-Flag-tagged N terminus projected into the cytoplasm rather than outside the cell. Were this the case, the epitope would not be accessible to the externally added antibody and immunoreactive C29 exchanger protein in the surface membrane would not be detected. We chose, therefore, an alternative method to test the surface expression of C29, by using the impermeant reagent NHS-SS-biotin (13) to covalently modify the amino residues in the surface membrane of transfected HEK 293 cells. The Na ϩ -Ca 2ϩ exchanger RBE-1 contains 46 lysine residues and C29 contains 45 lysines. Based on experimental evidence (10) and topological analysis (16,17), 4 lysine residues are modeled to face the cell exterior. In addition, 2 lysines derived from the Flag epitope are external as well (10). All other lysines, except 2 that are intramembraneous, face the cell interior. Hence, even if the topology of C29 is "scrambled" but the mutant protein is present in the plasma membrane, some of the lysines should be accessible to the biotinylating reagent. Fig. 4 shows an immunoblot of the biotinylated cell proteins and total cell proteins derived from rbe-1and C29-transfected cells. Quantitative analysis indicates that in this experiment a 3-fold higher amount of total immunoreactive C29 was expressed than in parallel RBE-1-expressing cells. Biotinylated Na ϩ -Ca 2ϩ exchanger protein, however, is detected only in the surface of cells expressing the parent exchanger RBE-1, and none is detected in the surface of the cells expressing C29. Similar results were obtained in three additional experiments. It should be also noted that, for determination of surface expression, transfected cells derived from 1 well of a 6-well plate (about 1.2 ϫ 10 6 cells at confluence) were used, whereas, for determination of total immunoreactive protein, about 1/10th of the total protein derived from 1 well out of a 12-well plate (0.4 ϫ 10 5 cells at confluence) was used.
In order to find out the reason that the C-terminal truncated mutants beyond C11 do not exhibit surface expression and in which cellular compartment the protein accumulates, we have studied their glycosylation pattern. Fig. 5 compares the protein profile of the [ 35 S]methionine-labeled immunoprecipitated exchanger derived from RBE-1-and from C29-expressing cells, grown without and with the glycosylation inhibitor tunicamycin (18). It can be seen (Fig. 5A) that, when tunicamycin is added to the labeling medium, the molecular mass of both immunoreactive exchangers RBE-1 and C29 is lower than the corresponding immunoreactive exchangers synthesized in the absence of tunicamycin. This finding suggests that RBE-1 and C29 are glycosylated in HEK 293 cells. For comparison, we are also showing a similar experiment, except that N-Flag-tagged rbe-1 and C29 were used to transfect the cells. It can be seen (Fig. 5B) that addition of tunicamycin does not result in expression of lower molecular mass immunoreactive protein. This is consistent with the fact that introduction of the N-Flag tag after Gly-8 involved also deletion of Asn-9 (10), the single glycosylation site of the NCX1, as suggested by the in vitro experiments of Hryshko et al. (19).
Transfer of membrane proteins from the ER to the Golgi is accompanied by modification of the oligosaccharide residues resulting in acquisition of resistance to endo H digestion. To determine whether the "non-functional" C-terminal truncated mutants accumulate in the ER or whether they are transferred FIG. 1. Na ؉ gradient-dependent Ca 2؉ uptake in transfected HeLa and HEK 293 cells. A, HeLa cells were transfected in parallel with the wild type Na ϩ -Ca 2ϩ exchanger RBE-1 and with its C-terminal truncated mutants. 17 h after transfection, Na ϩ gradient-dependent 45 Ca 2ϩ influx was determined as described under "Experimental Procedures." The transport activity of RBE-1 in each experiment was defined as 100%, and that of the mutants was calculated in relative values. The data were compiled from seven separate transfection experiments; each measurement was carried out in triplicate. No transport activity was detected in mock-transfected (with pBluescript SK) cells. The numerical value of the Ca 2ϩ taken up in the presence of Na ϩ gradient in cells expressing the wild type exchanger RBE-1 is 16.6 (S.D. ϭ 3.73), and in its absence it is 2.35 (S.D. ϭ 0.86) nmol/mg of protein/10 min. B, HEK 293 cells were transfected in parallel with the wild type Na ϩ -Ca 2ϩ exchanger RBE-1 and with its C-terminal truncated mutants. 48 h after transfection, Na ϩ gradient-dependent 45 Ca 2ϩ influx was determined as described under "Experimental Procedures." The transport activity of RBE-1 in each experiment was defined as 100% and that of the mutants was calculated in relative values. The data were compiled from six separate transfection experiments; each measurement was carried out in triplicate. The numerical value of the Ca 2ϩ taken up in cells expressing RBE-1 the wild type exchanger in the presence of a driving Na ϩ gradient is 27.27 (S.D. ϭ 4.25) and in the absence of a of Na ϩ gradient is 3.16 (S.D. ϭ 1.16). The Ca 2ϩ taken up in mock-transfected (with pcDNA3) HEK 293 cells in a Na ϩ gradient-dependent manner was 2.2% (n ϭ 3) relative to the wild type exchanger RBE-1.

FIG. 2. Western analysis of NCX1 proteins derived from transfected HeLa and HEK 293 cell extracts. Transfected HeLa (A) and
HEK 293 (B) cells were suspended in PBS to which 0.2 mM PMSF was added, and their protein concentration was determined. After a brief sonication, 1/10 of 10ϫ Laemmli SDS sample buffer was added to each extract and it was boiled for 5 min. 50 g of cell protein derived from each extract were layered on a mini-Laemmli 7% SDS-polyacrylamide gel. After electrophoretic transfer onto nitrocellulose, Western analysis was carried out with AbO-8 (10), a polyclonal antibody prepared against a conserved segment of amino acids (see "Experimental Procedures") of the rat brain Na ϩ -Ca 2ϩ exchanger. 125 I-Protein A was used as secondary antibody. The relative amounts of immunoreactive exchanger protein compiled from eight different Western blots are as follows. to the Golgi, we have studied their deglycosylation pattern by endo H. We have compared their protein profile before and after exposure to endo H and compared it to that of the parent clone RBE-1. In addition, we have determined the time required to acquire resistance to endo H for RBE-1 and its Cterminal truncated mutants by metabolically labeling them with [ 35 S]methionine and chasing the label with unlabeled methionine for different times. Fig. 6 shows the protein profile of immunoprecipitated [ 35 S]methionine-labeled RBE-1 (Fig. 6A), C10 (Fig. 6B), and C29 (Fig. 6C) at different times after onset of chase with unlabeled methionine. Each sample was loaded on AbO-8 coated Protein A-Sepharose beads (see "Experimental Procedures"). The eluted proteins were divided into equal parts and incubated 90 min without and with endo H, after which they were analyzed by SDS-PAGE. It can be seen that RBE-1 acquires resistance to endo H digestion by 1 h after onset of the chase (Fig. 6A), C10 acquires resistance after 5 h (Fig. 6B), and C29 as well as C13 (data not shown) do not become resistant to endo H even after 10 h of chase (Fig. 6C).   FIG. 3. Indirect immunofluorescence of HEK 293 cells expressing N-Flag-tagged RBE-1, N-Flag-tagged C10, and N-Flag-tagged C29. HEK 293 cells were grown on coverslips and transfected with the wild type exchanger N-Flag RBE-1 (A) and its C-terminal truncated mutants N-Flag C10 (B) and N-Flag C29 (C) as described. 48 h after transfection, the cells were fixed with 4% paraformaldehyde after which cells shown in panels 1 and 2 were treated with the anti-Flag monoclonal antibody. Cells shown in panels 3 and 4 were permeabilized with 0.25% Triton X-100 before application of the anti-Flag antibody. Fluorescein isothiocyanate-conjugated goat anti-mouse IgG was used as secondary antibody (10). Nuclei were stained with bisbenzimide. Fluorescence was collected by employing a 63ϫ/1.2 C-Achromat water immersion lens (panels 1 and 3) or 3.2 ϫ 63ϫ/1.2 (panels 2 and 4) using a Zeiss laser scanning microscope as described under "Experimental Procedures." DISCUSSION Expression of transport proteins and their mutants in heterologous systems provides tools needed to study the role of structural elements that are important for functional expression. In this work we show that the C terminus of the Na ϩ -Ca 2ϩ exchanger RBE-1 (NCX1.4) is important for assembly of a functionally apt form of the protein.
Secreted and surface membrane proteins follow a multistep, well coordinated and regulated pathway (20 -23) to their final location. Defective proteins that translocate to the ER and fail to assemble or fold correctly are retained in that compartment and eventually become degraded without reaching their final cellular destination (24). Our experiments suggest that the C terminus (amino acids Leu-875 to stop) of the Na ϩ -Ca 2ϩ exchanger is not mandatory for translocation of the nascent protein to the lumen of the ER, since functional and non-functional C-terminal truncated mutants including C29, the mutant in which the entire last transmembrane segment and the ex-tramembraneous tail were truncated, are glycosylated (Figs. 5 and 6).
Glycosylation pattern, however, can play an important role in correct folding and maturation of membrane proteins (25). Of the six consensus glycosylation sequences that are identified in the primary structure of the Na ϩ -Ca 2ϩ exchanger, only Asn-9 and Asn-584 are glycosylation candidates based on distance from edges of transmembrane segments (26,27). Of the two, only Asn-9 is extracellular (10) and hence faces the lumen of the ER. Both in vitro experiments and expression in oocytes indicated that only Asn-9 is glycosylated (19) in NCX1. This holds also for expression in HEK 293 cells, for functional and non-functional mutants (Fig. 5). Deletion of Asn-9 accompanied by insertion of the N-Flag epitope (10)  were labeled with [ 35 S]methionine for 45 min, after which the label was chased with unlabeled methionine as described under "Experimental Procedures." Prior to and at different intervals after the onset of the chase, immunoreactive exchanger protein was precipitated by AbO-8-coated Protein A-Sepharose beads. After release from the beads, each sample was divided into two equal parts and incubated without or with endo H as described. SDS-PAGE was carried out to analyze the protein profile obtained. Since after 10 h of chase the label of the immunoreactive exchanger C29 was substantially reduced (Fig. 6C), the film was exposed for different periods as marked by horizontal (0 -6 h of chase) and vertical (6 -10 h of chase) frame. non-glycosylated functional N-Flag-tagged RBE-1 and a nonglycosylated non-functional N-Flag-tagged C29. Hence, glycosylation of the Na ϩ -Ca 2ϩ exchanger is not a prerequisite for acquisition of functional conformation and surface membrane expression in HEK293 cells. These experiments also suggest that the non-functional C29 translocated to the ER at least in part with similar orientation to that of the functional parent exchanger, since Asn-584 (or other consensus glycosylation candidate) did not become glycosylated.
Maturation of the protein and subsequent transfer from the ER to the Golgi is accompanied by trimming of high mannose residues and terminal glycosylation, resulting in acquisition of resistance to endo H digestion (28). As can be seen in Fig. 6, sequential truncation of C-terminal amino acids impaired one of the processes that are mandatory for the maturation of a functional Na ϩ -Ca 2ϩ exchanger, resulting in increased retention times of the impaired proteins in the ER, which led to either reduced surface expression as with C10 (Fig. 3B) or to no detectable surface expression as with C29 (Figs. 3C and 4).
Reconstitution of proteins derived from HeLa and HEK 293 cells transfected with the functionally impaired C-terminal truncated mutants into proteoliposomes did not rescue the Na ϩ gradient-dependent Ca 2ϩ uptake (Table II). This suggests that these proteins are retained in the ER in a functionally impaired form. Moreover, reconstitution experiments suggest that even C9 is partially impaired, since its relative transport activity, which in whole cells is comparable to the parent exchanger, following reconstitution is much lower. Secretion is limited in overexpression systems by the cellular processing machinery (29). The plasma membrane incorporates only a small fraction of the overexpressed transporters. The cellular processing machinery suffices to "help" that small part of transporters to acquire mature conformation. Hence, in whole cells, the transport activity of C9 is similar to that of the wild type exchanger. During reconstitution of cells expressing C9, detergent treatment probably "destabilizes" further the impaired C9 protein.
Unlike with the parent protein, removal of the detergent alone is not sufficient to restore the functional conformation of C9; therefore, we measure lower transport activity than in parallel reconstituted RBE-1-transfected cells. The consequences of the truncation of an additional amino acid and the limitations of the cellular processing machinery probably play a role in the reduced ER to Golgi transfer and reduced surface expression of mutant C10. Although we detect C10 protein only in the surface of about 40% of the expressing cells, it is possible that this is an underestimate due to limitation of surface detection in cells where the overall level of expression is not high. If such underestimate applies also to the 35-39% transport activity of C10 in whole cells, it would suggest that the "true" activity of the surface membrane C10 transporters is similar to that of the wild type exchanger and C9. Following reconstitution, the recovered transport activity of C10 reaches only 20 -30% relative to the parent exchanger, probably since it is less apt than C9 and the results of detergent treatment are more detrimental.
Gabellini et al. (30) and recently also Li and Lytton (31) reported that expression of a Na ϩ -Ca 2ϩ exchanger lacking the six C-terminal transmembrane segments exhibited considerable Na ϩ -Ca 2ϩ exchange activity in HEK 293 cells. Immunoblotting of protein derived from cells expressing the truncated isoform also revealed, in addition to the expected 70-kDa protein, higher molecular mass ones. It has been suggested (30,31) that the truncated exchanger forms dimers that could be responsible for the transport activity detected. This option is presumably not open to mutants lacking only 9 -29 amino acids.
Although the C terminus of the NCX1, NCX2, and NCX3 gene products is highly conserved and only one conservative substitution, K901R, is observed among the last 11 C-terminal amino acids (32, 33), we do not think that the conserved sequence itself is important. We have inserted the Flag epitope into the C terminus after Glu-895 instead of Ala-896 to Phe-903. The resulting C-Flag-tagged mutant exchanger, which had an extramembraneous tail of 12 different amino acids instead of 9 present in NCX1, exhibited 139% (S.D. ϭ 21) transport activity relative to that of the parent exchanger (10). We suggest that the presence of an extramembraneous tail of at least 9 -10 amino acids might be of importance. Sequential truncation of C-terminal amino acids of GLYT1 (34) resulted in progressively lower transport activities. Truncation of 64 out of a total of 76 intracellular extramembraneous C-terminal amino acids of GLYT1 resulted in 9% of transport activity relative to the parent transporter and impaired surface expression. As with the C-terminal truncated mutants of the Na ϩ -Ca 2ϩ exchanger, reconstitution did not rescue the transport activity of the impaired C-terminal truncated GLYT1 mutants (34).
One possible explanation could be that truncation of the C terminus led in both cases to exposure of hydrophobic sequences and formation in the ER of detergent-insoluble aggregates (24,35) of misfolded overexpressed proteins, as in the case of ⌬F508 mutant of the cystic fibrosis transmembrane conductance regulator or the overexpressed CFTR (24). It could also be that the presence of an extramembraneous tail is important for acquisition of functional conformation of the protein since ␣ helices are stabilized by residues flanking the helix termini (36,37). Deletion of 9 or more C-terminal extramembraneous amino acids of the Na ϩ -Ca 2ϩ exchanger could destabilize the last transmembrane helix and impair the acquisition of functional conformation of the protein by exposure of hydrophobic sequences.