Functional Association between K+-Cl- Cotransporter-4 and H+,K+-ATPase in the Apical Canalicular Membrane of Gastric Parietal Cells*

We studied whether K+-Cl- cotransporters (KCCs) are involved in gastric HCl secretion. We found that KCC4 is expressed in the gastric parietal cells more abundantly at the luminal region of the gland than at the basal region. KCC4 was found in the stimulation-associated vesicles (SAV) derived from the apical canalicular membrane but not in the intracellular tubulovesicles, whereas H+,K+-ATPase was expressed in both of them. In contrast, KCC1, KCC2, and KCC3 were not found in either SAV or tubulovesicles. KCC4 coimmunoprecipitated with H+,K+-ATPase in the lysate of SAV. Interestingly the MgATP-dependent uptake of 36Cl- into the SAV was suppressed by either the H+,K+-ATPase inhibitor (SCH28080) or the KCC inhibitor ((R)-(+)-[(2-n-butyl-6,7-dichloro-2-cyclopentyl-2,3-dihydro-1-oxo-1H-inden-5-yl)oxy]acetic acid). The KCC inhibitor suppressed the H+ uptake into SAV and the H+,K+-ATPase activity of SAV, but the inhibitor had no effects on these activities in the freeze-dried leaky SAV. These results indicate that the K+-Cl- cotransport by KCC4 is tightly coupled with H+/K+ antiport by H+,K+-ATPase, resulting in HCl accumulation in SAV. In the tetracycline-regulated expression system of KCC4 in the HEK293 cells stably expressing gastric H+,K+-ATPase, KCC4 was coimmunoprecipitated with H+,K+-ATPase. The rate of recovery of intracellular pH in the KCC4-expressing cells after acid loading through an ammonium pulse was significantly faster than that in the KCC4-non-expressing cells. Our results suggest that KCC4 and H+,K+-ATPase are the main machineries for basal HCl secretion in the apical canalicular membrane of the resting parietal cell. They also may contribute in part to massive acid secretion in the stimulated state.

It has been assumed that Cl Ϫ moves passively down its electrochemical gradient through apical channels. Although the intracellular Cl Ϫ concentration of the parietal cell has not been reported, it is speculated to be much lower than that of the luminal secreted HCl (160 mM). Thus, Cl Ϫ transport through Cl Ϫ channels would require a large electrical potential difference across the apical membrane (e.g. 60 mV, inside negative against the luminal side). But to date the reports of this electrical potential differences have been low, such as 20 -25 mV (9), hinting that the Cl Ϫ secretory mechanism may be more complex than previously assumed.
Electroneutral K ϩ -Cl Ϫ cotransporters (KCCs) belong to a cation-chloride cotransporter gene family (SLC12). KCCs contribute to transepithelial transport and to the regulation of cell volume (10 -12). At least four KCC isoforms (KCC1-KCC4) have been identified to date. KCC3 has three splicing variants: KCC3a-KCC3c. KCC1 is widely expressed, whereas KCC2 is restricted to neurons. KCC3a and KCC4 are mainly expressed in epithelial-type cells (13,14). Recently we found that KCC3a is expressed in the basolateral membrane of gastric parietal cells located at the luminal region of gastric glands and coimmunoprecipitated with Na ϩ ,K ϩ -ATPase (15). Exogenous expression of KCC3a in LLC-PK1 cells up-regulates Na ϩ ,K ϩ -ATPase activity in lipid rafts (15).
If KCCs are present in the luminal membrane of gastric parietal cells, they may be involved in the electroneutral cotransport of K ϩ -Cl Ϫ , which would be driven by the electrochemical gradient for K ϩ across the luminal membrane established by the H ϩ ,K ϩ -ATPase (a very low luminal K ϩ concentration) and Na ϩ ,K ϩ -ATPase (a very high intracellular K ϩ concentration). So far, there have been no reports describing the presence of any KCCs in the luminal membrane. In the present study, we found that KCC4 is predominantly expressed and associated with H ϩ ,K ϩ -ATPase in the apical canalicular membrane of gastric parietal cells and that KCC4 is an important molecule for maintaining H ϩ ,K ϩ -ATPase activity in the canalicular membrane.
Isolation of Gastric Tissues and Cells-Mice, rats, and rabbits were humanely killed in accordance with the guidelines presented by the Animal Care and Use Committee of the University of Toyama, and their gastric mucosae were isolated from the stomachs. The cell suspension rich in parietal cells was prepared from isolated rabbit gastric mucosa as described previously (17). Hog gastric mucosa was prepared from the stomach obtained from Toyama meat center (Toyama, Japan). Human gastric mucosa was obtained from surgical resection of Japanese patients at Toyama University Hospital in accordance with the recommendations of the Declaration of Helsinki and with ethics committee approval. All of the patients gave informed consent.
Preparation of Hog Gastric Vesicles (Stimulation-associated Vesicles (SAV) and Tubulovesicles (TV))-Two kinds of gastric vesicles (heavy and light vesicles) were prepared simultaneously from hog gastric mucosa as described previously (18). Briefly the fundic region of the mucosa was scraped and homogenized in 250 mM sucrose, 1 mM EGTA, and 5 mM Tris-HCl (pH 7.4). The suspension was centrifuged at 1,000 ϫ g for 10 min, and the supernatant was further centrifuged at 13,500 ϫ g for 30 min. The pellet, resuspended in the buffer solution, was applied to the top of a 7% Ficoll shelf on a 12% Ficoll step gradient and centrifuged in an RPV-50T rotor (Hitachi Koki Co., Tokyo, Japan) at 132,000 ϫ g for 1 h. Heavy vesicles were collected from the interface between the 7 and 12% Ficoll layers, and they were washed with 250 mM sucrose to remove the Ficoll. Then the sample was centrifuged at 120,000 ϫ g in an SW41 Ti rotor (Beckman) for 20 h laying on a discontinuous sucrose gradient (10,20,30, and 50% sucrose), and 10 fractions of 1 ml each were collected from the top of the gradient. Fraction 5 was used as the SAV that contained the apical canalicular membranes of parietal cells (rich in H ϩ ,K ϩ -ATPase and least contaminated with Na ϩ ,K ϩ -ATPase) (supplemental Fig. 1). On the other hand, the supernatant after 13,500 ϫ g centrifugation was centrifuged at 100,000 ϫ g for 30 min. The samples were then applied to a 250 mM sucrose and 7% Ficoll step gradient and centrifuged at 132,000 ϫ g for 1 h. Light (microsomal) vesicles were collected from the interface between the 250 mM sucrose and 7% Ficoll layers, and the sample was used as the TV that contained the intracellular microsomal membranes of parietal cells. All procedures were carried out at 4°C. When indicated, freeze-dried SAV and TV were prepared by lyophilization, which increased leakiness of the vesicular membranes.
Northern Blotting-Poly(A) ϩ RNA of the cells was prepared by using PolyATtract mRNA isolation system II (Promega, Madison, WI), and 2.5 g of it was separated on a 1% agarose, formaldehyde gel and transferred onto a nylon membrane (Zeta-probe GT, Bio-Rad). Northern blotting was performed as described previously (15) using the 32 P-labeled rabbit KCC4 probe that is 713 bp long and corresponds to nucleotides 682-1394 of the KCC4 cDNA.
Western Blotting-Preparation of membrane fractions and Western blotting were carried out as described previously (19). The signals were visualized with the ECL Plus system (GE Healthcare). To quantify the chemiluminescence signals on the membranes, a FujiFilm LAS-1000 system and MultiGauge software were used. Anti-H ϩ ,K ϩ -ATPase antibodies were used at 1:5,000 (1H9 and 2B6) or 1:4,000 dilution (Ab1024). Anti-KCC1, anti-KCC2, anti-KCC3, anti-KCC4, and anti-␤-actin antibodies were used at 1:1,000 dilution. Anti-Na ϩ ,K ϩ -ATPase ␣1 subunit antibody was used at 1:10,000 dilution. Anti-Xpress antibody was used at 1:5,000 dilution. Anti-ezrin and anti-syntaxin-1 antibodies were used at 1:500 dilution. Anti-Rab11 antibody was used at 1:200 dilution. For the negative control, 1 volume of each primary antibody was preincubated with 5 volumes of the corresponding blocking peptide. Horseradish peroxidase-conjugated anti-mouse, anti-rabbit, or anti-goat IgG was used as a secondary antibody (1:2,500 dilution).
Immunohistochemistry-The gastric mucosa isolated from rat stomach was embedded in the optimum cutting temperature compound (Sakura Finetechnical Co., Tokyo, Japan) and was cut at 8 m. The sections were fixed in ice-cold methanol for 7 min at room temperature and were pretreated with 1.5% bovine serum albumin for 1 h at room temperature to block nonspecific binding of antibody. Then the sections were incubated with anti-mouse KCC4, anti-H ϩ ,K ϩ -ATPase (1H9), anti-Na ϩ ,K ϩ -ATPase ␣1 subunit, or anti-AQP4 antibody (1:100 dilution) overnight at 4°C. Alexa Fluor 488-conjugated and Alexa Fluor 546-conjugated anti-IgG antibodies (1:100 dilution) were used as secondary antibodies. Immunofluorescence images were visualized using a Zeiss LSM 510 laser scanning confocal microscope.
Immunocytochemistry-HEK293 cells were fixed with icecold methanol for 7 min at room temperature and permeabilized with phosphate-buffered saline containing 0.3% Triton X-100 and 0.1% bovine serum albumin for 15 min at room temperature. Nonspecific binding was blocked by 3% bovine serum albumin. The permeabilized cells were incubated with the antimouse KCC4 or anti-H ϩ ,K ϩ -ATPase (1H9) antibody (1:100 dilution) overnight at 4°C and then with the Alexa Fluor 488conjugated and Alexa Fluor 546-conjugated anti-IgG antibodies (1:100 dilution) for 1 h at room temperature. Immunofluorescence images were visualized using a Zeiss LSM 510 laser scanning confocal microscope.
Immunoprecipitation-Membrane fractions of SAV (100 g of protein) and the HEK293 cells stably expressing both KCC4 and H ϩ ,K ϩ -ATPase (2 mg of protein) were solubilized in lysis buffer (phosphate-buffered saline containing 0.5% Triton X-100, 0.1% bovine serum albumin, and 1 mM EDTA) for 30 min on ice and centrifuged at 90,000 ϫ g for 30 min at 4°C. The lysate was precleared with protein A-agarose beads, and the supernatant was incubated in the presence and absence of anti-KCC4 antibody (1:50 dilution) or anti-His tag antibody (1:100 dilution) for 12 h at 4°C with end-over-end rotation. Antibodyantigen complexes were incubated with protein A-agarose beads and incubated for 4 h at 4°C with end-over-end rotation. Then the beads were washed three times with the lysis buffer and suspended in SDS sample buffer. The samples were used for Western blotting to check the expression of KCC4, H ϩ ,K ϩ -ATPase ␣-subunit, or ␤-actin.
Measurement of 36 Cl Ϫ Uptake into Hog Gastric Vesicles (SAV and TV)-SAV or TV (100 g of protein) were preincubated with a solution of 150 mM KCl, 250 mM sucrose, and 40 mM PIPES-Tris (pH 6.8) for 5 min or 20 h at 4°C. Then the vesicle (SAV or TV) was incubated with a solution containing 150 mM KCl, 4 mM MgSO 4 , 250 mM sucrose, 2 mM ATP, 5 Ci/ml Na 36 Cl, and 40 mM PIPES-Tris (pH 6.8) for 5 min at 25°C. The reaction mixtures were rapidly filtered through a 0.45-m HAWP filter (Millipore Co., Bedford, MA). To calibrate nonspecific binding of Na 36 Cl to the vesicles and the filter, the experiment was performed in the absence of ATP. The filter was washed with solution containing 150 mM KCl, 250 mM sucrose, and 40 mM PIPES-Tris (pH 6.8); transferred to a counting vial; and solubilized with 5 ml of ACS II scintillant. Then the radioactivity of 36 Cl Ϫ was countered.
Measurement of H ϩ Uptake into SAV and TV-H ϩ uptake into hog gastric vesicles was assessed by measuring the quenching of acridine orange fluorescence (20). The reaction was started by addition of ATP (350 M) to the mixture containing SAV (20 g/ml) or TV (5 g/ml), 150 mM KCl, 2 mM MgCl 2 , 5 M acridine orange, 10 g/ml valinomycin, and 20 mM PIPES-NaOH (pH 7.4). In the experiments to test anion selectivity, 150 mM KCl was replaced with 150 mM KX (X ϭ Br, I, H 2 PO 4 , or gluconate). When indicated, SCH28080 (20 M), a specific inhibitor of H ϩ ,K ϩ -ATPase, or DIOA (10 M), an inhibitor of KCCs, was added. Fluorescence of acridine orange was measured in a Shimadzu RF-5000 spectrofluorometer at 25°C (excitation, 495 nm; emission; 530 nm). The H ϩ uptake was expressed as change of fluorescence intensity from 0 to 3 min after addition of ATP.
Measurement of H ϩ ,K ϩ -ATPase Activity-H ϩ ,K ϩ -ATPase activities of SAV and TV were measured in a pyruvate kinaselactate dehydrogenase-linked system where hydrolysis of ATP is coupled with oxidation of NADH (21). The reaction mixture contained TV (10 g/ml) or SAV (40 g/ml), 150 mM KCl, 3 mM MgSO 4 , 200 M NADH, 1 mM ATP, 0.8 mM phosphoenolpyruvate, 11 IU/ml lactate dehydrogenase, 4 IU/ml pyruvate kinase, 10 g/ml valinomycin, and 5 mM PIPES-NaOH (pH 7.4) in the presence or absence of 10 M SCH28080. In the experiments to test anion selectivity, 150 mM KCl was replaced with 150 mM KX (X ϭ Br, I, H 2 PO 4 , or gluconate). When indicated, DIOA (10 M) was added. The decrease in the amount of NADH was measured by a Beckman spectrophotometer in a dual wavelength mode at 340 and 500 nm at 25°C.
H ϩ ,K ϩ -ATPase activities of freeze-dried vesicles prepared from TV and SAV (10 g of protein) and the membrane fraction of HEK293 cells (30 g of protein) were measured in a 1-ml solution containing 15 mM KCl, 3 mM MgSO 4 , 1 mM ATP, 5 mM NaN 3 , 2 mM ouabain, and 40 mM Tris-HCl (pH 6.8) in the presence or absence of 50 M SCH28080. After incubation for 10 min (freeze-dried vesicles) or 30 min (HEK293 cells) at 37°C, the reaction was terminated by addition of the ice-cold stop solution containing 12% perchloric acid and 3.6% ammonium molybdate, and the inorganic phosphate released was measured (22). Plasmid Construction-Full-length cDNA encoding rat KCC4 was inserted into the pcDNA4/His vector (Invitrogen) by using EcoRI and XbaI restriction sites (KCC4-pcDNA4/His vector). The KCC4 cDNA (with Xpress epitope) cut from the KCC4-pcDNA4/His vector was inserted into the pcDNA5/TO vector by using AflII and XbaI restriction sites (KCC4-pcDNA5/TO vector).
Tetracycline-regulated Expression System of KCC4 in HEK293 Cells Stably Expressing H ϩ ,K ϩ -ATPase-HEK293 cells stably expressing ␣and ␤-subunits of the gastric H ϩ ,K ϩ -ATPase were established as described previously (23). The cells were cotransfected with the KCC4-pcDNA5/TO and pcDNA6/TR vectors (Invitrogen) using Lipofectamine 2000 and cultured for 24 h. The transfected cells were selected in the presence of 400 units/ml hygromycin B and 6 g/ml blasticidin S.
Measurement of Intracellular pH-Intracellular pH (pH i ) of the HEK293 cells was measured by monitoring the fluorescence of BCECF as described previously (24,25). The cells (1 ϫ 10 5 cells), seeded on coverslips (13 ϫ 7 mm; Matsunami Glass, Osaka, Japan) coated with poly-L-lysine and rat tail collagen I (Invitrogen), were cultured for 24 h. Then the cells were treated with or without tetracycline (2 g/ml) for an additional 24 h. Expression of KCC4 was confirmed by Western blotting and immunocytochemistry.
The cells were incubated with BCECF-AM (10 M) in buffer containing 145 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1. Statistics-Results are shown as means Ϯ S.E. Differences between groups were analyzed by one-way analysis of variance, and correction for multiple comparisons was made by using Tukey's multiple comparison test. Comparison between the two groups was made by using Student's t test. Statistically significant differences were assumed at p Ͻ 0.05.

Expression of KCC4 in Gastric Parietal
Cells-First we examined whether KCC4 mRNA was expressed in gastric parietal cells. Northern blotting showed significant expression of KCC4 mRNA (ϳ4.9 kb) in rabbit gastric parietal cells (Fig. 1A). Western blotting showed that anti-mouse KCC4 antibody reacted with the 165-kDa protein in mouse and rat gastric mucosa and that anti-human KCC4 antibody reacted with the 165-kDa protein in hog and human gastric mucosa (Fig. 1B, upper panels). Both antibodies reacted with the 165-kDa bands in the membrane fraction of HEK293 cells transfected with rat KCC4 (Fig.  1B, cloned KCC4). The specificities of these antibodies for the 165-kDa bands were confirmed by using the corresponding blocking peptide (Fig. 1B, lower panels).
In the immunohistochemistry of isolated rat gastric mucosa, KCC4 was found to be colocalized with H ϩ ,K ϩ -ATPase, which is expressed in the intracellular tubulovesicles and the apical canalicular membrane of parietal cells (Fig. 2, A-F). The specificity of anti-KCC4 antibody for positive staining was confirmed by using the blocking peptide (Fig. 2, G-I). On the other hand, the distribution pattern of KCC4 was apparently different from that of the Na ϩ ,K ϩ -ATPase ␣1 subunit (Fig. 2, J-L). It has been reported that younger parietal cells in the luminal region of the glands much more actively secrete acid than the older parietal cells in the basal region (26 -28). AQP4 has been reported to be localized in the basolateral membrane of the parietal cells at the basal region of the gastric glands (29). Interestingly the present double immunostaining of KCC4 and AQP4 in the gastric mucosa showed that KCC4 is expressed in the parietal cells more abundantly at the luminal region of the glands than at the basal region. (Fig. 2, M-O).
Expression of KCC4 in the SAV That Are Derived from the Apical Canalicular Membranes of Gastric Parietal Cells-To determine whether KCC4 is expressed in the intracellular tubulovesicles and the apical canalicular membrane of parietal cells, two types of gastric vesicles (SAV and TV) were prepared from hog gastric mucosa. H ϩ ,K ϩ -ATPase ␣and ␤-subunits were highly expressed in both TV and SAV (Fig. 3A). The expression level of the Na ϩ ,K ϩ -ATPase ␣1 subunit in TV and SAV was much lower than that in the gastric mucosa (Fig. 3A).
It has been reported that Rab11 is present in the H ϩ ,K ϩ -ATPase-rich vesicular membrane and related to the vesicular trafficking machinery in gastric parietal cells (30). ␤-Actin has been reported to be associated with ezrin in the apical membrane of gastric parietal cells (31,32). Here the expression level of Rab11 in TV was much higher than that in SAV. In contrast, the expression levels of ␤-actin and ezrin in SAV were much  . Expression of KCC4 in the SAV. A, Western blotting was performed with hog tubulovesicles (5 g of protein) and stimulation-associated vesicles (5 g of protein) with anti-H ϩ ,K ϩ -ATPase ␣-subunit (HK␣), anti-H ϩ ,K ϩ -ATPase ␤-subunit (HK␤), and anti-Na ϩ ,K ϩ -ATPase ␣1 subunit (NaK) antibodies. The specific bands for H ϩ ,K ϩ -ATPase ␣-subunit, H ϩ ,K ϩ -ATPase ␤-subunit, and Na ϩ ,K ϩ -ATPase ␣1 subunit were observed at 95, 80, and 100 kDa, respectively. As a positive control for detecting Na ϩ ,K ϩ -ATPase ␣1 subunit, hog gastric mucosa (5 g of protein) was used (mucosa). B, expression of Rab11 (27 kDa), ␤-actin (45 kDa), ezrin (85 kDa), syntaxin-1 (35 kDa), KCNQ1 (70 kDa), KCNE2 (60 kDa), and CFTR (150 kDa) in TV and SAV (30 g of protein). C, Western blotting was performed with TV, SAV, and gastric mucosa of hogs (30 g of protein) using anti-human KCC4 antibody. A band of 165 kDa was observed in SAV but not in TV (upper panel). The 165-kDa bands disappeared in the presence of the corresponding blocking peptide (ϩBP; lower panel). Upper and lower panels were derived from a membrane blotted with two same sets of the three samples. The membrane was cut into two pieces and used for two separate immunoblotting shown in the upper and lower panels. D, Western blotting was performed with TV, SAV, and gastric mucosa of hogs (30 g of protein) using antibodies for KCC1, KCC2, and KCC3. The 180-kDa band of KCC3 detected in the mucosa is due to KCC3a in the basolateral membrane of the gastric parietal cell (15). A membrane fraction of pig kidney LLC-PK1 cells was used as a positive control for KCC1, and that of hog brain was used as a positive control for KCC2 (control). The specific bands for KCC1, KCC2, and KCC3 were observed at 130, 138, and 180 kDa, respectively. E, immunoprecipitation (IP) was performed with the detergent extracts of SAV (100 g of protein) using anti-KCC4 antibody and protein A-agarose in the lysate of SAV (IP:KCC4, ϩ). In control experiments, preimmune serum instead of the antibody was used (IP:KCC4, Ϫ). The detergent extracts (input; 1 ⁄33 (for KCC4) or 1 ⁄100 (for H ϩ ,K ϩ -ATPase ␣-subunit (HK␣) and ␤-actin) of total protein) and immunoprecipitation samples (IP:KCC4; 1 ⁄100 (for KCC4) and 1 ⁄50 (for H ϩ ,K ϩ -ATPase ␣-subunit (HK␣) and ␤-actin) of the samples) were detected by Western blotting (WB) using antibodies for KCC4 (top panel), H ϩ ,K ϩ -ATPase ␣-subunit (middle panel), and ␤-actin (bottom panel). Top and middle panels were derived from a membrane blotted with two same sets of the three samples. The membrane was cut into two pieces and used for two separate immunoblotting shown in the top and middle panels. The immunoprecipitation shown is representative of three independent experiments. higher than those in TV (Fig. 3B). Syntaxin-1 was expressed in both TV and SAV as reported previously (33) (Fig. 3B). KCNQ1/KCNE2 K ϩ channels and CFTR Cl Ϫ channel were predominantly expressed in TV (Fig. 3B). These results confirmed that SAV and TV are derived from the apical canalicular membranes and the intracellular microsomal membranes of the gastric parietal cell, respectively.
Interestingly KCC4 (165 kDa) was predominantly expressed in SAV but not in TV (Fig. 3C). The specificity of the antibody for the 165-kDa band was confirmed by using the corresponding blocking peptide (Fig. 3C, lower panel). No significant expression of other KCCs such as KCC1, KCC2, and KCC3 was observed in either TV or SAV (Fig. 3D).
To study whether KCC4 is associated with H ϩ ,K ϩ -ATPase in SAV, immunoprecipitation was performed using an anti-KCC4 antibody. The subsequent Western blotting of the immune pellets with an anti-H ϩ ,K ϩ -ATPase ␣-subunit antibody gave a clear band corresponding to the H ϩ ,K ϩ -ATPase ␣-subunit (95 kDa), whereas blotting with an anti-␤-actin antibody as a negative control gave no ␤-actin band (45 kDa) (Fig. 3E). These results suggest that KCC4 is associated with H ϩ ,K ϩ -ATPase in the SAV.
Note that the apparent difference in affinity of the anti-human KCC4 antibody for KCC4 protein between Fig. 3, C and E, originated from the pretreatment of SAV with (Fig. 3E) and without (Fig. 3C) the lysis buffer. This buffer effect was confirmed by a separate experiment shown in supplemental Fig. 2.
Inhibition of Cl Ϫ Transport by the H ϩ ,K ϩ -ATPase Inhibitor in SAV-Here we measured Cl Ϫ uptake into TV and SAV using 36 Cl Ϫ in a solution containing 150 mM KCl, 4 mM MgSO 4 , 250 mM sucrose, 2 mM ATP, 5 Ci/ml Na 36 Cl, and 40 mM PIPES-Tris (pH 6.8). DIOA is known as a potent inhibitor of KCCs (34). Although high concentrations of DIOA suppressed H ϩ ,K ϩ -ATPase activity (IC 50 ϭ 75-97 M), no inhibitory effect of this drug on H ϩ ,K ϩ -ATPase was observed at a lower concentration (Ͻ20 -30 M) (35). In Fig. 4, DIOA (10 M) significantly inhibited Cl Ϫ uptake in SAV (Fig. 4B) but not in TV (Fig. 4A), reflecting the presence of KCC4 in SAV and its absence from TV. SCH28080 (10 M), a specific inhibitor of H ϩ ,K ϩ -ATPase, significantly inhibited Cl Ϫ uptake in both SAV and TV (Fig. 4). In the SAV, inhibitory effects of DIOA and SCH28080 were not significantly different from that of DIOA plus SCH28080 (Fig.  4B). These results suggest that the DIOA-sensitive Cl Ϫ transport in SAV may be mediated by the H ϩ ,K ϩ -ATPase activity.
Inhibition of H ϩ Transport by the KCC Inhibitor in SAV-Next the inhibitory effects of DIOA on H ϩ uptake into SAV and TV by H ϩ ,K ϩ -ATPase were studied. SCH28080 (20 M) markedly inhibited the H ϩ uptake in TV and SAV as expected (Fig. 5,  A and B). On the other hand, no H ϩ uptake was observed in the freeze-dried TV and SAV (Fig. 5, A and B). Interestingly DIOA (10 M) significantly inhibited the H ϩ uptake into SAV, whereas it had no significant effect in TV (Fig. 5, C and D), indicating that the inhibition of KCC4 resulted in inhibition of H ϩ uptake by H ϩ ,K ϩ -ATPase.
Inhibition of H ϩ ,K ϩ -ATPase Activity by the KCC Inhibitor in SAV-We studied whether DIOA affects the SCH28080-sensitive K ϩ -ATPase activity (H ϩ ,K ϩ -ATPase activity). DIOA (10 M) significantly inhibited the H ϩ ,K ϩ -ATPase activity in SAV (Fig. 6B), whereas it had no significant effects in TV (Fig. 6A). In the freeze-dried leaky TV and SAV, no significant effects of DIOA (10 M) on the H ϩ ,K ϩ -ATPase activity were observed (supplemental Fig. 3, A and B). These results suggest that DIOA indirectly inhibits H ϩ ,K ϩ -ATPase activity via inhibition of the ion transport by KCC4.
Anion Selectivity for the DIOA-sensitive H ϩ Transport and H ϩ ,K ϩ -ATPase Activity in SAV-To study whether the DIOA (10 M)-sensitive H ϩ transport and H ϩ ,K ϩ -ATPase activity in SAV depends on the species of anion, H ϩ transport and enzyme  Fig. 4). We could not examine the effect of I Ϫ on H ϩ uptake because KI quenches the fluorescence of acridine orange (36). The order of anion selectivity for the H ϩ ,K ϩ -ATPase activity was Cl Ϫ Ͼ Br Ϫ Ͼ I Ϫ Ͼ PO 4 3Ϫ ϭ gluconate (Fig. 7B).
These orders were similar to that reported for the anion selectivity for K ϩ transporting activity of KCC4 (Cl Ϫ Ͼ Br Ϫ Ͼ PO 4 3Ϫ ϭ I Ϫ Ͼ gluconate) (37).
Anion Selectivity for the DIOAsensitive K ϩ Uptake into the SAV-To study whether the K ϩ uptake into SAV by K ϩ -Cl Ϫ cotransport depends on anion species, we measured the DIOA (10 M)-sensitive 86 Rb ϩ uptake in solutions of 150 mM KX (where X ϭ Cl, Br, H 2 PO 4 , or gluconate). The order of anion selectivity for 86 Rb ϩ uptake into SAV was Cl Ϫ Ͼ Br Ϫ Ͼ PO 4 3Ϫ ϭ gluconate (Fig. 7C). This selectivity is quantitatively the same as those of the DIOA-sensitive H ϩ transport (Fig. 7A) and H ϩ ,K ϩ -ATPase activity (Fig. 7B), indicating that proton transport by H ϩ ,K ϩ -ATPase depends on ion transport by KCC4. The SCH28080 (10 M)-sensitive 86 Rb ϩ uptake in solutions of 150 mM KX (where X ϭ Cl, Br, H 2 PO 4 , or gluconate) was also assessed. The order of anion selectivity was Cl Ϫ Ͼ Br Ϫ Ͼ PO 4 3Ϫ ϭ gluconate (supplemental Fig. 5).
Stable Coexpression of KCC4 and H ϩ ,K ϩ -ATPase in the HEK293 Cells-The tetracyclineregulated expression system of KCC4 was constructed in HEK293 cells that stably expressed ␣and ␤-subunits of gastric H ϩ ,K ϩ -ATPase. No significant expression of endogenous KCC4 was observed in control HEK293 cells (data not shown). In this heterologous expression system, exogenous expression of KCC4 protein was assessed by using an anti-Xpress antibody. Expression of KCC4 (165 kDa) was observed in the cells treated with tetracycline (Tet-On cells), whereas no significant expression of KCC4 was observed in the cells treated without tetracycline (Tet-Off cells) (Fig. 8A). Expression levels of H ϩ ,K ϩ -ATPase ␣-subunit in the Tet-On cells were not significantly different from those in the Tet-Off cells (Fig. 8B). Both KCC4 and H ϩ ,K ϩ -ATPase were found to be present in the plasma membrane of the Tet-On cells (Fig. 8C). To check whether KCC4 is associated with H ϩ ,K ϩ -ATPase in the Tet-On cells as is the case in the SAV (Fig. 3E), immunoprecipitation was performed by using an anti-His tag antibody (for KCC4). The subsequent Western blotting of the immune pellets with an anti-H ϩ ,K ϩ -ATPase ␣-subunit antibody gave a band for H ϩ ,K ϩ -ATPase ␣-subunit (Fig. 8D), indicating association between KCC4 and the H ϩ ,K ϩ -ATPase ␣-subunit.
KCC4-induced Stimulation of H ϩ Transport by H ϩ ,K ϩ -ATPase in the Heterologous Expression System-In the membrane fractions of both Tet-Off and Tet-On cells (i.e. cell-free condition), DIOA (10 M) did not inhibit the H ϩ ,K ϩ -ATPase activity (supplemental Fig. 3, C and D) as found for the freezedried TV and SAV (supplemental Fig. 3, A and B). This would reflect the fact that the membrane samples obtained from the HEK293 cells were not tightly sealed as has been described elsewhere (38). In Fig. 9, we studied the capacity of the acid extrusion (H ϩ transport activity) in the Tet-On and Tet-Off cells. The capacity was assessed by measuring the rate of recovery of pH i in the cells after acid loading through an ammonium pulse. The pH i recovery was monitored in the absence of Na ϩ to exclude the contribution of the Na ϩ /H ϩ exchanger. The pH i recovery process in the Tet-On cells (Fig. 9, B and G) was significantly faster than that in the Tet-Off cells (Fig. 9, A and G). Interestingly DIOA (10 M) significantly decreased the recovery rate of pH i in the Tet-On cells (Fig. 9, D and G) but not in the Tet-Off cells (Fig. 9, C and G). SCH28080 (10 M) significantly decreased the recovery rate of pH i in both the Tet-On and Tet-Off cells (Fig. 9, E-G). These results suggest that KCC4 may stimulate the H ϩ transport activity of H ϩ ,K ϩ -ATPase in the Tet-On cells.

DISCUSSION
In the present study, we found the following. 1) KCC4 is expressed in the apical canalicular membrane of gastric parietal cells more abundantly at the luminal region of the glands than at the basal region. In contrast, KCC1, KCC2, and KCC3 are not significantly expressed in the apical canalicular membrane. KCC4 is absent from the intracellular tubulovesicles. 2) KCC4 is associated with H ϩ ,K ϩ -ATPase in the apical canalicular membrane. 3) In vesicles of the apical canalicular membrane (SAV), the H ϩ ,K ϩ -ATPase inhibitor suppresses the Cl Ϫ transport activity, which is sensitive to a KCC inhibitor (DIOA; 10  M). 4) The KCC inhibitor suppresses the H ϩ ,K ϩ -ATPase activity in the SAV but not in the freeze-dried leaky SAV and the TV. 5) In the SAV, the anion selectivity of the DIOA-sensitive H ϩ ,K ϩ -ATPase activity and the DIOA-sensitive H ϩ and K ϩ transports are similar to that for the K ϩ transporting activ-FIGURE 8. Tetracycline-regulated expression system of KCC4 in the HEK293 cells stably expressing gastric H ؉ ,K ؉ -ATPase. The tetracyclineregulated expression system of KCC4 was introduced to the HEK293 cells that stably express ␣and ␤-subunits of gastric H ϩ ,K ϩ -ATPase. The cells were treated with (on) or without (off) 2 g/ml tetracycline. A, the expression of KCC4 in the membrane fraction of the cells (30 g of protein) was confirmed by Western blotting using anti-Xpress antibody. B, the expression level of H ϩ ,K ϩ -ATPase ␣-subunit in the Tet-On cells (on) was compared with that in the Tet-Off cells (off). In the upper panel, a representative picture of Western blotting is shown. In the lower panel, the score for the Tet-Off cells is normalized as 1. n ϭ 5. NS, p Ͼ 0.05. C, a-c show the same cells under a microscope (as do d-f and g-i). Double immunostaining was performed with the Tet-On cells (a-f) and the Tet-Off cell (g-i) using anti-KCC4 plus anti-H ϩ ,K ϩ -ATPase ␣-subunit antibodies. Localizations of KCC4 (a, d, and g), H ϩ ,K ϩ -ATPase (HK) (b, e, and h), and KCC4 plus H ϩ ,K ϩ -ATPase (merged images; c, f, and i) are shown. In d-f, anti-KCC4 antibody was pretreated with the blocking peptide (ϩBP). Positive KCC4 staining disappeared. Scale bars, 10 m. D, immunoprecipitation (IP) was performed with the detergent extracts of the KCC4-expressing Tet-On cells using anti-His tag antibody for KCC4 and protein A-agarose. The detergent extract (input) and the immunoprecipitation samples obtained with (IP:His(KCC4), ϩ) and without (IP:His(KCC4), Ϫ) the antibody were detected by Western blotting (WB) using anti-Xpress antibody for detecting KCC4 (upper panel) and anti-H ϩ ,K ϩ -ATPase ␣-subunit antibody (HK␣; lower panel). The immunoprecipitation shown is representatives of three independent experiments. ity of KCC4 (37). 6) The expression of KCC4 stimulates H ϩ transport across the HEK293 cell membranes that coexpress H ϩ ,K ϩ -ATPase.
Recently we found that KCC3a is expressed in the basolateral membrane of gastric parietal cells, that KCC3a is associated with Na ϩ ,K ϩ -ATPase ␣1 subunit (␣1NaK), and that KCC3a up-regulates ␣1NaK activity in the membrane fraction of the KCC3a-expressing LLC-PK1 cells and rabbit gastric mucosa. KCC3a may directly activate the ␣1NaK activity in cell-free conditions (15). On the other hand, in this study, the functional association of KCC4 with H ϩ ,K ϩ -ATPase could be seen only in intact tightly sealed SAV and HEK293 cells and not in the freeze-dried leaky SAV and the leaky membrane fraction of HEK293 cells. Our present results suggest that KCC4 indirectly increases H ϩ ,K ϩ -ATPase activity by effectively supplying K ϩ to the luminal surface of this ATPase.
Gastric acid secretion by the parietal cells is accompanied with dramatic morphological changes. In resting parietal cells, tubulovesicles are present in intracellular compartments underlying the apical canalicular membrane and form a reticulated meshwork. Upon stimulation, the tubulovesicles translocate and connect with the canalicular membrane, resulting in massive acid secretion (39 -41). So far, several Cl Ϫ and K ϩ channels have been identified in gastric parietal cells (7,42). CFTR (5), CLIC-6 (parchorin) (6, 7), and SLC26A9 (8) are candidates that could be involved in the luminal Cl Ϫ efflux for gastric acid (HCl) secretion. KCNQ1/KCNE2 (1, 2) and Kir4.1 (3,4) are candidates that could be involved in the luminal K ϩ efflux for the K ϩ recycling. In the present study, CFTR was confirmed to be localized predominantly in the tubulovesicles (Fig. 3B). KCNQ1 was found to be expressed mainly in the tubulovesicles (Fig. 3B) as reported previously (4,43). It has been reported that CLIC-6 is distributed throughout the cytosol (6) and that Kir4.1 is localized in the tubulovesicles (4). Therefore, the distribution pattern of these K ϩ and Cl Ϫ channels is apparently different from that of KCC4 that is predominantly expressed in the apical canalicular membrane.
On the basis of our present findings, we suggest that these two membranes may not mix but remain separate and distinct when the tubulovesicular membrane is connected with the apical canalicular membrane upon stimulation. Fig. 10 summarizes the putative mechanisms of HCl secretion in these two membranes. In the resting state, KCC4 together with H ϩ ,K ϩ -ATPase that is present in the apical canalicular membrane is involved in the basal acid secretion. Upon stimulation, CFTR, CLIC-6, SLC26A9, KCNQ1/KCNE2, Kir4.1, and KCC4 are involved in the KCl transport for massive gastric acid secretion.
Interestingly it has been reported that KCC1, KCC2, and KCC3 are inhibited at pH Ͻ7.5 when these are expressed in Xenopus oocytes, whereas KCC4 is activated at acidic pH (44). This finding suggests that KCC4 may be specialized to operate in an acidic environment. In fact, we found that KCC4 is expressed in the apical canalicular membranes, which face the gastric acid.
In the kidney, KCC4 has been reported to be crucial for Cl Ϫ extrusion across the basolateral membrane of acid-secreting ␣-intercalated cells and that the loss of KCC4 leads to deafness associated with renal tubular acidosis (45). In contrast to this basolateral localization in renal ␣-intercalated cells, KCC4 is localized in the apical membrane of gastric parietal cells. It will be necessary to clarify the mechanism of the recruitment of KCC4 in the apical canalicular membrane of the parietal cells.
Gastric parietal cells migrate from the luminal to the basal region of the glands, and the luminal parietal cells more actively secrete acid than do the basal parietal cells (26 -28). KCC4 is expressed in the parietal cells more abundantly at the luminal region of the glands than at the basal region. It is noted that KCC3a is located in the luminal region (15). KCC3a in the basolateral membrane of the parietal cell increases the Na ϩ ,K ϩ -ATPase activity. The electrochemical potential gradient for K ϩ across the apical membrane drives the ion transport by KCC4 and is established by mainly Na ϩ ,K ϩ -ATPase and H ϩ ,K ϩ -ATPase. Thus, both KCC3a and KCC4 are involved in acid secretion.
In conclusion, KCC4 is functionally expressed in the apical canalicular membrane of gastric parietal cells, and its K ϩ -Cl Ϫ cotransport is coupled with H ϩ ,K ϩ -ATPase activity in the canalicular membrane. KCC4 may function as a K ϩ -supplying molecule for H ϩ ,K ϩ -ATPase and also as a Cl Ϫ -transporting molecule for HCl secretion. In the resting state of the parietal cells, KCC4 and H ϩ ,K ϩ -ATPase in the apical canalicular membrane of the parietal cell are the main machineries for the basal gastric acid secretion. In the stimulated state, they also contribute to acid secretion together with other K ϩ and Cl Ϫ channels and H ϩ ,K ϩ -ATPase that are present in tubulovesicles.