The roles of carbohydrate chains of the beta-subunit on the functional expression of gastric H(+),K(+)-ATPase.

Gastric H(+),K(+)-ATPase consists of alpha and beta-subunits. The alpha-subunit is the catalytic subunit, and the beta-subunit is a glycoprotein stabilizing the alpha/beta complex in the membrane as a functional enzyme. There are seven putative N-glycosylation sites on the beta-subunit. In this study, we examined the roles of the carbohydrate chains of the beta-subunit by expressing the alpha-subunit together with the beta-subunit in which one, several, or all of the asparagine residues in the N-glycosylation sites were replaced by glutamine. Removing any one of seven carbohydrate chains from the beta-subunit retained the H(+),K(+)-ATPase activity. The effects of a series of progressive removals of carbohydrate chains on the H(+),K(+)-ATPase activity were cumulative, and removal of all carbohydrate chains resulted in the complete loss of H(+), K(+)-ATPase activity. Removal of any single carbohydrate chain did not affect the alpha/beta assembly; however, little alpha/beta assembly was observed after removal of all the carbohydrate chains from the beta-subunit. In contrast, removal of three carbohydrate chains inhibited the surface delivery of the beta-subunit and the alpha-subunit assembled with the beta-subunit, indicating that the surface delivery mechanism is more dependent on the carbohydrate chains than the expression of the H(+),K(+)-ATPase activity and alpha/beta assembly.

␤-subunits of H ϩ ,K ϩ -ATPase were prepared from rabbit gastric mucosae as described elsewhere (5). The ␣and ␤-subunit cDNAs were digested with EcoRI and XhoI. The obtained fragments were each ligated into pcDNA3 vector treated with EcoRI and XhoI.
Site-directed Mutagenesis-Introduction of site-directed mutations was carried out by sequential polymerase chain reaction steps as described elsewhere (20), in which appropriately mutated ␤-subunit cDNAs (segments between nucleotide 302 (AflII site) and 1036 (Eco47III site)) were prepared. The 5Ј-flanking sense and 3Ј-flanking antisense primers were 5Ј-GCTGAAGTCGCCAGGCGTAAC-3Ј (nucleotides 281-301) and 5Ј-CCACGGGA AGCAGCGGACGC-3Ј (nucleotides 1049 -1068), respectively. Sense and antisense synthetic oligonucleotides, each 21 bases long containing one mutated base near the center, were designed. The cDNA of H ϩ ,K ϩ -ATPase ␤-subunit in pBluescript SK(Ϫ) was used as a polymerase chain reaction template. Polymerase chain reaction was routinely carried out in the presence of 200 M each dNTP, 500 nM primers, 10 mM KCl, 10 mM (NH 4 ) 2 SO 4 , 2 mM MgSO 4 , 20 mM Tris-HCl, pH 8.8, 0.1% Triton X-100, 100 g/ml bovine serum albumin, and 2.5 units of Pfu DNA polymerase for 30 cycles. DNA sequencing was done by the dideoxy chain termination method using an Autoread and Autocycle DNA sequencing kits and an ALFexpress DNA sequencer (Amersham Pharmacia Biotech). After sequencing, the fragment amplified in the final polymerase chain reaction was digested with AflII plus HindIII or HindIII plus Eco47III and ligated back into the relevant position of the wild-type H ϩ ,K ϩ -ATPase ␤-subunit construct.
Cell Culture, Transfection, and Preparation of Membrane Fractions-Cell culture of HEK-293 was carried out as described previously (5). ␣and ␤-subunit cDNA transfection was performed by the calcium phosphate method with 10 g of cesium chloride-purified DNA/10-cm dish. Cells were harvested 2 days after the DNA transfection. Membrane fractions of HEK cells were prepared as described previously (5).

SDS-Polyacrylamide Gel Electrophoresis and Western
Blot-SDSpolyacrylamide gel electrophoresis was carried out as described elsewhere (21). Membrane preparations (30 g of protein) were incubated in a sample buffer containing 2% SDS, 2% ␤-mercaptoethanol, 10% glycerol, and 10 mM Tris-HCl, pH 6.8, at room temperature for 2 min and applied to the SDS-polyacrylamide gel. Western blot was carried out as described previously (5). Densitometric analysis of the blots was carried out using the ATTO Densitographic software (ATTO, Tokyo, Japan).
Immunoprecipitation-Immunoprecipitation was carried out as described previously (23). Membrane fractions (1 mg) of HEK cells expressing the ␣/␤ complex were incubated in 1 ml of lysis buffer containing 1% Nonidet P-40, 150 mM NaCl, 0.5 mM EDTA, and 50 mM Tris-HCl, pH 7.4, at 4°C for 30 min. After centrifugation at 16,000 ϫ g for 20 min, the supernatant was incubated with an antibody, Ab1024, at a 1:100 dilution, and 10 l of ImmunoPure immobilized protein A (Pierce) at 4°C for 12 h. After centrifugation, the pellet was four times washed with the lysis buffer followed by two washes in 0.1% Nonidet P-40, 150 mM NaCl, 0.5 mM EDTA, and 50 mM Tris-HCl, pH 7.4. The pellet was solubilized in the sample buffer for SDS-polyacrylamide gel electrophoresis and incubated at room temperature for 10 min. The proteins were separated on SDS-polyacrylamide gel and blotted. The ␤-subunit was detected by an anti-␤-subunit antibody, 2B6, in combination with a peroxidase-conjugated anti-mouse antibody, which was preabsorbed with rabbit serum.
Immunofluorescence-COS cells were cultured on glass coverslips and transfected with the ␣and ␤-subunit cDNAs as described previously (24). Within 48 h after transfection, cells were fixed in cold methanol (Ϫ20°C) for 10 min and processed for immunostaining. H ϩ ,K ϩ -ATPase ␣-subunit was detected with a rabbit polyclonal anti-body, HK9 (24), and the ␤-subunit with a monoclonal antibody, 2G11 (8). Both antibodies were used at dilutions of 1:100. Goat anti-mouse IgG conjugated to fluorescein isothiocyanate and goat anti-rabbit IgG conjugated to rhodamine were used as the secondary antibodies at dilution of 1:100. Immunostaining was visualized with a Zeiss LSM 410 laser scanning confocal microscope. Contrast and brightness settings were chosen to ensure that all pixels were in the linear range. All images were the product of 8-fold line averaging.
Glycosidase Treatment-Thirty micrograms of membrane fraction was treated with Endo H or PNGase F following the manufacturer's instructions. For Endo H digestion, 30 g of membrane fraction was treated with 0.01 units of Endo H in a solution containing 0.1% SDS, 1 M 2-mercaptoethanol, 0.5 mM phenylmethylsulfonyl fluoride, and 50 mM sodium phosphate, pH 6.0 at 37°C overnight. For PNGase F digestion, 30 g of membrane fraction was treated with 1 unit of PNGase F in a solution containing 0.1% SDS, 1% n-octylglucoside, 1 M 2-mercaptoethanol, 30 mM EDTA, and 50 mM sodium phosphate, pH 6.0, at 37°C overnight.
Assay of H ϩ ,K ϩ -ATPase Activity-ATPase activity was assayed in 1 ml of a solution containing 50 g of membrane protein, 3 mM MgCl 2 , 1 mM ATP, 5 mM NaN 3 , 15 mM KCl, 2 mM ouabain, and 40 mM Tris-HCl, pH 6.8, in the presence and absence of 50 M SCH 28080, which is an  inhibitor specific for gastric H ϩ ,K ϩ -ATPase. After incubation at 37°C for 30 min, the reaction was terminated by the addition of ice-cold stop solution containing 12% perchloric acid and 3.6% ammonium molybdate. Inorganic phosphate released was measured from the absorbance at the wavelength of 320 nm as described elsewhere (25). The K ϩ -ATPase activity was calculated as the difference between activities in the presence and absence of SCH 28080. Inorganic phosphate released in the enzyme reaction with the wild-type H ϩ ,K ϩ -ATPase was 4 -5-fold higher than the background level of inorganic phosphate released in the absence of enzyme. Protein was measured using the BCA protein assay kit from Pierce with bovine serum albumin as a standard.

RESULTS
Construction of the ␤-Subunits with Mutated N-Glycosylation Sites-Rabbit gastric H ϩ ,K ϩ -ATPase ␤-subunit contains seven putative N-glycosylation sites (Asn-Xaa-Ser and Asn-Xaa-Thr); Asn-99, Asn-103, Asn-130, Asn-146, Asn-161, Asn-193, and Asn-222 as shown in Fig. 1, which are located in the extracellular segment and are well conserved in rat and human gastric H ϩ ,K ϩ -ATPases. Therefore, it is likely that the carbohydrate chains play an important role in the function of gastric H ϩ ,K ϩ -ATPase. Klaassen et al. (19) studied the role of N-linked carbohydrate chains by expressing gastric H ϩ ,K ϩ -ATPase in insect Sf9 cells in the presence of an N-glycosylation inhibitor, tunicamycin. They reported that N-glycosylation was essential to produce a catalytically active recombinant H ϩ ,K ϩ -ATPase, whereas the expression of the H ϩ ,K ϩ -ATPase ␣and ␤-subunit proteins was observed even in the presence of tunicamycin. In a preliminary study, we found that the effect of tunicamycin on the expression of the wild-type H ϩ ,K ϩ -ATPase ␣and ␤-subunits in mammalian cells is quite different from that in insect Sf9 cells. In addition, there has been no report that tests the significance of each carbohydrate chain by mutating the putative glycosylation sites. In this study, we mutated one, several, or all of the asparagine residues in the putative N-glycosylation sites to glutamine residues, co-expressed the mutant ␤-subunit with the ␣-subunit, and studied the role of N-linked carbohydrate chains on the subunit interaction, enzyme activity, and intracellular localization of H ϩ ,K ϩ -ATPase. Of the 127 possible ␤-subunit mutants to be analyzed, we prepared mutants as shown in Table I; seven single mutants (N99Q, N103Q, N130Q, N146Q, N161Q, N193Q and N222Q); two double mutants (N99Q/N103Q and N130Q/N222Q); mutant III (N99Q/N130Q/ N222Q); mutant IV (N99Q/N103Q/N130Q/N222Q); mutant V (N99Q/N103Q/N130Q/N146Q/N222Q); mutant VI (N99Q/ N103Q/N130Q/N146Q/N193Q/N222Q); and mutant VII (N99Q/ N103Q/N130Q/N146Q/N161Q/N193Q/N222Q). Each of these mutant cDNAs was co-transfected with the ␣-subunit cDNA to HEK-293 cells.
Expression of ␣and ␤-Subunits- Fig. 2 (A and B) shows Western blot patterns of the membrane fractions of the transfectants, detected by using the anti-␤-subunit antibody, 2B6. When the cells were co-transfected with both the wild-type ␣-subunit and ␤-subunit cDNAs, a dense band with a lower molecular mass (48 kDa) (␤ c ) and a smear band with a higher molecular mass (60 -70 kDa) (␤ m ) were observed (Fig. 2, A and B, lanes 1). The ␤-subunit with a higher molecular mass (60 -70 kDa) (␤ m ) was resistant to Endo H, whereas the ␤-subunit with a lower molecular mass (40 -48 kDa) (␤ c ) was digested with Endo H (data not shown). These results indicate that the 60 -70-kDa band represents the ␤-subunit with complex-type (resistant to Endo H) carbohydrate chains and that the bands with a lower molecular mass (40 -48 kDa) represent the ␤-subunits with high mannose-type (sensitive to Endo H) carbohydrate chains (23). When the cells were co-transfected with the wild-type ␣-subunit and each of the seven single mutant ␤-subunit cDNAs, two bands, ␤ c and ␤ m , were observed ( Fig. 2A). The molecular mass of these two bands were slightly lower than those of the wild-type complex. Introducing more mutations in the ␤-subunit progressively decreased the molecular mass of the ␤ c (Fig. 2B). This result indicates that all seven consensus N-glycosylation sites are in fact modified with carbohydrate chains in the HEK cells. Mutants V, VI, and VII ␤-subunits exhibited no smear band corresponding to the ␤ m . Therefore, the wild-type ␤-subunit, all single mutants, N130Q/N222Q, and mutants III and IV were able to leave the endoplasmic reticulum (ER) compartment, whereas mutants V, VI, and VII were completely retained in the ER. Fig. 2C shows Western blot patterns detected by 2B6 anti-

FIG. 2. Western blots with anti-␤-subunit antibody of membrane fractions of HEK cells co-transfected with the wild-type
␣-subunit plus mutant ␤-subunit cDNAs. A, HEK-293 cells were co-transfected with the wild-type H ϩ ,K ϩ -ATPase ␣-subunit cDNA plus The cell membrane fractions (30 g) were applied to the gel and blotted with anti-␤-subunit antibody, 2B6. B and C, HEK-293 cells were co-transfected with the wild-type H ϩ ,K ϩ -ATPase ␣-subunit cDNA plus wild- The cell membrane fractions (30 g) were treated with (C) or without PNGase F (B) as described under "Experimental Procedures" and then applied to the gel and blotted with 2B6. body of the membrane preparations treated with PNGase F. After the treatment with PNGase F, molecular mass of the ␤-subunit in the wild-type complex decreased from 60 -70 kDa (␤ m ) and 40 -48 kDa (␤ c ) to 35 kDa (Fig. 2, B and C, lanes 1). The 35-kDa band represents the deglycosylated form (protein core) of the H ϩ ,K ϩ -ATPase ␤-subunit. No band corresponding to the deglycosylated ␤-subunit was detected in the wild-type complex in the absence of PNGase F (Fig. 2B, lane 1). Fig. 2C  (lanes 2-7) shows that PNGase F treatment shifted the ␤-subunit bands to 35 kDa in these ␣/␤ mutant complexes. However, the treatment did not shift the ␤-subunit band in the ␣/␤ mutant VII complex (Fig. 2C, lane 8). Fig. 3 shows Western blot patterns of the membrane fractions of the transfectants, detected with an anti-␣-subunit antibody, Ab1024. The ␣-subunit bands were quantified by densitometric assay using gastric microsomal H ϩ ,K ϩ -ATPase samples (from 0.05 to 1 g protein) as standards (data not shown). When the cells were transfected with the wild-type ␣-subunit cDNA in the absence of the ␤-subunit cDNA, a single weak band was detected around 95 kDa, which represents the expression of the H ϩ ,K ϩ -ATPase ␣-subunit (Fig. 3, A and B,  lanes 9). The expression of the ␣-subunit increased 7.8 times when the cells were co-transfected with the wild-type H ϩ ,K ϩ -ATPase ␤-subunit cDNA (Fig. 3B, lane 1) as previously reported (5). Similar increases in expression of the ␣-subunit were also observed when the cells were co-transfected with each one of the seven single mutant ␤-subunit cDNAs (Fig. 3A). The expression of the ␣-subunit increased 7.9, 6.1, 5.4, and 5.5 times for the N130Q/N222Q, mutants III, IV, and V, respec-tively, compared with the expression of the ␣-subunit in the absence of the ␤-subunit (Fig. 3B). However, there was only a 2.4-fold increase in expression of the ␣-subunit for the mutants VI, and little or no (1.1-fold) increase for the mutant VII.
H ϩ ,K ϩ -ATPase Activity of the Glycosylation Site Mutants-H ϩ ,K ϩ -ATPase activity was measured in the membrane fractions from the cells co-transfected with the wild-type ␣-subunit plus ␤-subunit cDNAs that were mutated at their N-glycosylation sites. Mutations introduced in only one of the glycosylation sites did not abolish the H ϩ ,K ϩ -ATPase activity of the ␣/␤ complex, indicating that there is no specific single N-linked carbohydrate chain essential for the H ϩ ,K ϩ -ATPase function (Fig. 4A). The ␣/␤ mutant IV and ␣/␤ mutant V complexes retained 64 and 39% of the H ϩ ,K ϩ -ATPase activity of the wild-type ␣/␤ complex, respectively (Fig. 4B). The ␣/␤ mutant VII complex, in which all the glycosylation sites on the ␤-subunit were mutated to glutamine, almost completely lost the H ϩ ,K ϩ -ATPase activity. These results indicate that the effects of the mutations in the N-glycosylation sites on the H ϩ ,K ϩ -ATPase activity are cumulative.
Expression of H ϩ ,K ϩ -ATPase in the Presence of Tunicamy-  cin-The expression of the H ϩ ,K ϩ -ATPase ␣and ␤-subunits in Sf9 cells was observed even in the presence of tunicamycin (19). In this study, we co-transfected the wild-type ␣and ␤-subunit cDNAs in HEK-293 cells, incubated them in the presence of tunicamycin, and prepared the membrane fractions. In contrast to the previous finding in Sf9 cells, little detectable H ϩ ,K ϩ -ATPase ␣and ␤-subunits were found in the membrane fraction of the cells incubated with 100 ng/ml or 1 g/ml tunicamycin (data not shown). Therefore, we were unable to use tunicamycin in the HEK-293 cell expression system to test the effect of this drug on the functions of H ϩ ,K ϩ -ATPase.
␣/␤ Assembly of Glycosylation Site Mutants- Fig. 5 shows Western blot patterns of the samples immunoprecipitated with anti-␣-subunit antibody, Ab1024, and detected with anti-␤subunit antibody, 2B6. When the cells were co-transfected with the H ϩ ,K ϩ -ATPase ␣-subunit plus wild-type H ϩ ,K ϩ -ATPase ␤-subunit cDNAs, proteins with molecular masses of 68 and 50 kDa, which represent the H ϩ ,K ϩ -ATPase ␤-subunits with complex-type (␤ m ) and high mannose-type (␤ c ) carbohydrate chains, respectively, were co-precipitated with the anti-␣-subunit antibody (Fig. 5A, lane 1). When the cells were co-transfected with the ␣-subunit plus each of the seven single mutant ␤-subunit cDNAs, two bands (␤ m and ␤ c ) were observed after the immunoprecipitation (Fig. 5A). The molecular mass of these two bands was slightly lower than those observed in the wild-type ␣/␤ complex. When the cells were co-transfected with the ␣-subunit plus N130Q/N222Q mutant ␤-subunit cDNAs, a similar ␤-subunit pattern was observed (Fig. 5B, lane 2). Mutants III, IV, V, and VI were also co-precipitated with the ␣-subunit (Fig. 5B, lanes 3-6). However, in the case of mutants V and VI, the amounts of precipitated ␤-subunits were significantly lower than that of the wild-type ␤-subunit. Mutant VII was hardly co-precipitated with the ␣-subunit. These results suggest that each carbohydrate chain of the ␤-subunit is not directly involved in ␣/␤ assembly of the H ϩ ,K ϩ -ATPase; however, only a small amount of ␣/␤ complex was formed after removal of all seven carbohydrate chains.
Localization of ␣and ␤-Subunits in COS Cells-The wildtype ␤-subunit, seven single mutants, N130Q/N222Q, and mutants III and IV can leave the ER compartment for the Golgi apparatus as evidenced by the observation that they have not only high mannose-type but also complex-type carbohydrate chains, as shown in Fig. 2B. We tried to examine the subcellular localization of the ␣and ␤-subunits in HEK cells directly using immunocytochemistry. It was difficult to perform subcellular localization on proteins in HEK cells because HEK cells grew in fairly dense islands, and the individual cells in these islands were small and closely packed against their neighbors. In this study, therefore, we used COS-1 cells because COS cells were larger and more spread out, which made it much easier to distinguish intracellular from cell surface distributions. We have previously shown that pump subunit proteins behave similarly with respect to surface delivery in these cell lines (26).
When the COS cells were transfected with H ϩ ,K ϩ -ATPase ␣-subunit cDNA alone, a perinuclear pattern was observed, consistent with retention of the ␣-subunit in the ER (Fig. 6B). When the H ϩ ,K ϩ -ATPase ␣-subunit was expressed with its wild-type ␤-subunit, the ␣-subunit was observed primarily at the cell surface (Fig. 6D). Therefore, the H ϩ ,K ϩ -ATPase ␣-subunit requires its ␤-subunit for efficient cell surface expression as reported previously (24). The single mutant N99Q (Fig. 6F) and the double mutants N99Q/N103Q (not shown) and N130Q/ N222Q (Fig. 6H) also supported the ␣-subunit expression on the cell surface, although the extent of the surface delivery was lower in the case of the mutant ␤-subunits compared with the wild-type ␤-subunit. However, no cell surface expression of the ␣-subunit was detected when it was co-expressed with mutant III (Fig. 6J).
When the cells were transfected with H ϩ ,K ϩ -ATPase ␤-subunit cDNA alone, the ␤-subunit was observed at the cell surface as well as in intracellular vesicles (Fig. 6A). Therefore, the H ϩ ,K ϩ -ATPase ␤-subunit reaches the cell surface with (Fig.  6C) and without (Fig. 6A) its ␣-subunit as reported previously (24). The single mutant N99Q (Fig. 6E) and the double mutants N99Q/N103Q (not shown) and N130Q/N222Q (Fig. 6G) also reach the cell surface. However, no cell surface distribution of the ␤-subunit was detected for mutant III (Fig. 6I). Thus, carbohydrate chains are also important for surface delivery of the ␣and ␤-subunits.

DISCUSSION
H ϩ ,K ϩ -ATPase ␤-subunit shows a number of structural similarities with Na ϩ ,K ϩ -ATPase ␤-subunit. Both ␤-subunits consist of a short amino-terminal cytoplasmic domain, one transmembrane domain and a large extracellular domain. This extracellular domain contains six conserved cysteine residues involved in the formation of three disulfide bonds and several carbohydrate chains. The cytoplasmic and transmembrane domains are replaceable between these two ATPases for the functional expression of H ϩ ,K ϩ -ATPase, but the majority of the extracellular domain is not replaceable (23). The number and complexity of the carbohydrate chains on the ␤-subunit are different between H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase. Na ϩ ,K ϩ -ATPase ␤1-subunit contains three N-linked carbohydrate chains. Tamkun and Fambrough (27) have shown that glycosylation of the ␤-subunit was not required for the subunit assembly and not involved in the rate of intracellular transport of the ␣/␤ complex to the plasma membrane. Furthermore, glycosylation was not required to protect the Na ϩ ,K ϩ -ATPase from degradation in the chick sensory neurons. Zamofing et al. (28) reported that Na ϩ ,K ϩ -ATPase in toad urinary bladder cells incubated with tunicamycin was active in both hydrolytic and transport activities. Takeda et al. (29) expressed Na ϩ ,K ϩ -ATPase in Xenopus oocytes by injecting the ␣and ␤-subunit cRNAs and incubating the oocytes in the presence of tunicamycin. The oligosaccharide-deficient ATPase was transported to the plasma membrane and exhibited the same Na ϩ ,K ϩ -ATPase and 86 Rb ϩ transport activities as the wild-type Na ϩ ,K ϩ -ATPase. Beggah et al. (30) abolished the three Nglycosylation sites of the ␤-subunit by site-directed mutagenesis and found that the deglycosylated ␤-subunit was able to associate with the ␣-subunit and express the functional Na ϩ ,K ϩ -ATPase ( 86 Rb uptake and [ 3 H]ouabain binding) on the cell surface of Xenopus oocytes.
H ϩ ,K ϩ -ATPase ␤-subunit contains six or seven N-linked carbohydrate chains. Klaassen et al. (19) expressed H ϩ ,K ϩ -ATPase in insect Sf9 cells in the presence of tunicamycin. Tunicamycin completely abolished the K ϩ -ATPase activity and SCH 28080-sensitive phosphorylation capacity of the enzyme expressed in Sf9 cells, whereas the expression level and assembly of ␣and ␤-subunit proteins were unaffected, indicating that the carbohydrate chains are essential for biosynthesis of functional ATPase. Therefore, the role of the carbohydrate chains on the ATPase function is quite different between Na ϩ ,K ϩ -ATPase and H ϩ ,K ϩ -ATPase.
In this study, we examined the role of carbohydrate chains of the ␤-subunit on the functions of H ϩ ,K ϩ -ATPase in mammalian cells, because there are striking differences in the glycosylation state of the ␤-subunit and the localization of ␣/␤ complex between wild-type H ϩ ,K ϩ -ATPases expressed in insect and mammalian cells. When the H ϩ ,K ϩ -ATPase was expressed in Sf9 cells, the molecular mass of carbohydrate chains was smaller than that reported in the native H ϩ ,K ϩ -ATPase (10); a large percentage of the ␤-subunit contained high mannose-type carbohydrate chains, and there was little ␤-subunit-associated complex-type carbohydrate chains (19). Nonglycosylated ␤-subunit (protein core) was also observed on the Western blot. Catalytically active H ϩ ,K ϩ -ATPase ␣/␤ complexes were localized to intracellular membranes, and the ␣-subunit was not observed on the plasma membrane of Sf9 cells (19). In contrast to these findings in Sf9 cells, we found that H ϩ ,K ϩ -ATPase ␤-subunits expressed in HEK-293 cells contained carbohydrate chains, a large fraction of which were high mannose-type carbohydrate chains. Nonglycosylated ␤-subunit was not observed on the Western blot. The ␣-subunits associated with the ␤-subunit were observed to be delivered to the plasma membrane. In the expression system using mammalian cells, no H ϩ ,K ϩ -ATPase activity was observed in the membrane fraction of HEK cells expressing the ␣/␤ mutant VII, in which all seven carbohydrate chains were removed. This result is consistent with the previous finding that no H ϩ ,K ϩ -ATPase activity was observed in Sf9 cells incubated with tunicamycin (19). However, each carbohydrate chain was not specifically essential for the H ϩ ,K ϩ -ATPase function, because any single mutation in the glycosylation sites of the ␤-subunit did not abolish the K ϩ -ATPase activity. The effect of a series of progressive removals of carbohydrate chains from the ␤-subunit on the enzyme activity was cumulative, although the effect of isocharge substitution (from Asn to Gln) on the enzyme activity could not be completely excluded. The present results suggest that the carbohydrate chains of the ␤-subunit are collectively important for the catalytic activity of H ϩ ,K ϩ -ATPase.
Each carbohydrate chain on the ␤-subunit was not directly involved in the ␣/␤ assembly. However, very small amounts of ␣/␤ complex were observed when the cells were co-transfected with the ␣-subunit plus mutant VI cDNAs. The ␣/␤ assembly was not observed after removal of all seven carbohydrate chains. Therefore, the carbohydrate chains of the ␤-subunit are collectively important for assembly between the ␣and ␤-subunits. The ␤-subunits of single mutants, double mutants, mutant III, and mutant IV contained both complex-type and high mannose-type carbohydrate chains, indicating that these subunits were able to leave the ER compartment.
The carbohydrate chains on the ␤-subunit are important for the surface delivery of the ␤-subunit and the ␣-subunit that had assembled with the ␤-subunit. In COS cells, no surface delivery occurred in the mutant III complex. This result con-FIG. 6. Immunolocalization of H ؉ ,K ؉ -ATPase ␣and ␤-subunits expressed in COS cells. COS cells were transfected with the cDNAs encoding the H ϩ ,K ϩ -ATPase ␣and ␤-subunits, either alone or in pairs. Immunolocalization was performed using a polyclonal rabbit antibody directed against the ␣-subunit, HK9, and a mouse monoclonal antibody directed against the ␤-subunit, 2G11. The distribution of the ␤-subunit is depicted in A, C, E, G, and I, and that of the ␣-subunit in B, D, F, H, and J. A and B correspond to cells expressing the wild-type ␤-subunit alone and the ␣-subunit alone, respectively. The ␤-subunit is able to reach the cell surface (A), whereas the ␣-subunit expressed without ␤-subunit is retained intracellularly (B). C and D represent cells cotransfected with the wild-type ␤-subunit and ␣-subunit, E and F represent cells co-transfected with N99Q ␤-subunit and the ␣-subunit. G and H represent cells co-transfected with N130Q/N222Q ␤-subunit and the ␣-subunit. I and J represent cells co-transfected with mutant III ␤-subunit and the ␣-subunit. Both the ␣and ␤-subunits can exhibit a cell surface distribution as long as the ␤-subunit retains at least five glycosylation sites.
trasts with the previous finding that N-glycosylation of Na ϩ ,K ϩ -ATPase ␤-subunit is not necessary for its transport to the plasma membrane (27,28). Further study of this difference between H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase would be interesting because they have different destinations in polarized cells (apical and basolateral membranes, respectively).
Rabbit gastric H ϩ ,K ϩ -ATPase ␤-subunit contains seven glycosylation sites. Therefore, theoretically, it is necessary to prepare all the combinations of mutants (127 mutants) and analyze the functions of the all mutants to precisely study the roles of each carbohydrate chain and the possible interactions between the carbohydrate chains. Here, we prepared one series of the representative plural mutants and analyzed their functions because all seven single mutants showed almost equivalent effects. It cannot be completely excluded that some specific interactions between certain combination of carbohydrate chains are more important than others for some functions of the ATPase.
In conclusion, from the present series of experiments, it has been found that different levels of glycosylation of the ␤-subunit are necessary for the ATPase activity, assembly, and the surface delivery of the ␣and ␤-subunits. Surface delivery of the ␣and ␤-subunits is more dependent on the carbohydrate chains than the expression of the H ϩ ,K ϩ -ATPase activity and ␣/␤ assembly.