A hybrid between Na+,K+-ATPase and H+,K+-ATPase is sensitive to palytoxin, ouabain, and SCH 28080.

Na(+),K(+)-ATPase is inhibited by cardiac glycosides such as ouabain, and palytoxin, which do not inhibit gastric H(+),K(+)-ATPase. Gastric H(+),K(+)-ATPase is inhibited by SCH28080, which has no effect on Na(+),K(+)-ATPase. The goal of the current study was to identify amino acid sequences of the gastric proton-potassium pump that are involved in recognition of the pump-specific inhibitor SCH 28080. A chimeric polypeptide consisting of the rat sodium pump alpha3 subunit with the peptide Gln(905)-Val(930) of the gastric proton pump alpha subunit substituted in place of the original Asn(886)-Ala(911) sequence was expressed together with the gastric beta subunit in the yeast Saccharomyces cerevisiae. Yeast cells that express this subunit combination are sensitive to palytoxin, which interacts specifically with the sodium pump, and lose intracellular K(+) ions. The palytoxin-induced K(+) efflux is inhibited by the sodium pump-specific inhibitor ouabain and also by the gastric proton pump-specific inhibitor SCH 28080. The IC(50) for SCH 28080 inhibition of palytoxin-induced K(+) efflux is 14.3 +/- 2.4 microm, which is similar to the K(i) for SCH 28080 inhibition of ATP hydrolysis by the gastric H(+),K(+)-ATPase. In contrast, palytoxin-induced K(+) efflux from cells expressing either the native alpha3 and beta1 subunits of the sodium pump or the alpha3 subunit of the sodium pump together with the beta subunit of the gastric proton pump is inhibited by ouabain but not by SCH 28080. The acquisition of SCH 28080 sensitivity by the chimera indicates that the Gln(905)-Val(930) peptide of the gastric proton pump is likely to be involved in the interactions of the gastric proton-potassium pump with SCH 28080.

The sodium pump (Na ϩ ,K ϩ -ATPase) 1 and the gastric proton pump (H ϩ ,K ϩ -ATPase) are ion-transporting ATPases, both hydrolyzing ATP to actively pump ions against their electrochemical gradients (1,2). Unlike other ion-transporting ATPases, they both require a highly glycosylated smaller peptide, the ␤ subunit, for their function, in addition to a larger ATP-recognizing ␣ subunit. The ␣ subunits of proton and sodium pumps have ϳ60% identical primary sequences and form 10 membrane-spanning domains, denoted M1-M10, with their amino and carboxyl termini localized in the cytosol (3)(4)(5)(6)(7)(8)(9). The ␤ subunits are 30% identical and form a single membrane span with the amino terminus on the extracellular side of the membrane. Despite this limited identity of the primary structure, the tertiary structure of both types of ␤ subunits must be very similar, since the proton pump ␤ subunit was shown in numerous investigations to form functional complexes with the ␣ subunit of the sodium pump when these proteins were expressed together (10 -14).
Both enzymes are pharmacological receptors for clinically relevant drugs. Thus, the pumping activity of the gastric proton pump is specifically inhibited by imidazopyridines like 8-benzyloxy-3-cyanomethyl-2-methyl-imidazo[1,2-a]pyridine (SCH 28080) or omeprazole (15)(16)(17). Derivatives of the latter are widely used clinically for controlling hyperacidity or peptic ulcer (18,19). The sodium pump is specifically inhibited by a series of naturally occurring steroids, such as ouabain and digitalis (20,21). Based on their clinical use, these substances are also referred to as cardiac glycosides or cardioactive steroids; their application helps to increase muscular contractility of the failing heart (22)(23)(24).
In recent years, the work of many investigators has been focused on the identification of amino acids or peptides of the proton and sodium pumps involved in interaction with omeprazole and SCH 28080, or ouabain and related cardiac glycosides, respectively. Analysis of a series of sodium pump mutants helped to identify several amino acids involved in the recognition of ouabain (25)(26)(27)(28), but little is known about amino acids or peptides of the gastric proton pump involved in the recognition of omeprazole or SCH 28080. Amino acids in both the aminoand carboxyl-terminal halves of the ␣ subunit of gastric H ϩ ,K ϩ -ATPase have been suggested to participate in the binding sites for these inhibitors. Blostein et al. (29) expressed a chimeric polypeptide consisting of the amino-terminal 519 amino acids of gastric H ϩ ,K ϩ -ATPase and the COOH-terminal 507 amino acids of Na ϩ ,K ϩ -ATPase in LLC-PK1 cells and reported inhibition of potassium influx by SCH 28080. Munson et al. (30) labeled a peptide from the M1-M2 region of gastric H ϩ ,K ϩ -ATPase with a photoactive analog of SCH 28080. Labeling experiments have also identified Cys 892 within the extracellular loop connecting the M7 and M8 membrane spans of the gastric proton-potassium pump ␣ subunit as a component of the binding site for omeprazole (17). Since omeprazole binding to the proton pump can be inhibited by SCH 28080 (31), this result raises the possibility that amino acids within the COOHterminal half of the gastric pump ␣ subunit might also partic-ipate in interactions of the gastric proton pump with SCH 28080.
To test this hypothesis, we made use of the highly specific interactions between the sodium pump and palytoxin to investigate whether the M7/M8 extracellular peptide of the gastric proton pump ␣ subunit is involved in SCH 28080 recognition. Palytoxin from marine corals of the genus Palythoa, like the cardioactive steroids, is a specific inhibitor of the sodium pump (32,33). Unlike the cardioactive steroids, however, which inhibit ion flow through the pump, palytoxin converts the enzyme into an open channel that allows ions to flow down their concentration gradient (34). The palytoxin-induced channels have been studied with electrophysiological tools (35,36) and have been found to have a single-channel conductance of about 10 picosiemens in crayfish giant axons. Single-channel recordings have also been obtained from Na ϩ ,K ϩ -ATPase incorporated into planar lipid bilayers (36). Expression of Na ϩ ,K ϩ -ATPase in the yeast Saccharomyces cerevisiae causes the yeast cells to lose intracellular K ϩ when exposed to palytoxin, whereas yeast cells without Na ϩ ,K ϩ -ATPase are not affected by palytoxin (11,37,38). Taking advantage of the high specificity of the palytoxin/sodium pump interaction, we used the expression of Na ϩ ,K ϩ -ATPase in yeast to address the following questions. Does palytoxin also induce a K ϩ efflux from yeast cells that express a hybrid sodium pump consisting of a sodium pump ␣ subunit and a gastric proton pump ␤ subunit? Does a similar phenomenon occur when yeast cells express a chimeric sodium pump ␣ subunit containing the extracellular sequence Gln 905 -Val 930 from the M7/M8 loop of the gastric proton pump ␣ subunit? Finally, is any palytoxin-induced K ϩ efflux observed sensitive to ouabain or SCH 28080?

EXPERIMENTAL PROCEDURES
Vectors-The yeast expression vectors used in this study and the various combinations used for transformation of yeast cells are summarized in Table I. The methods applied for their construction have been described previously. Therefore, a brief summary of the most important features is given here. The yeast expression plasmid YEp1PT was used as a vector for the cDNA coding for the rat sodium pump ␣3 subunit. This vector is denoted YEpr␣3 (39). The same expression plasmid was also used for the insertion of the cDNA coding for a chimera of the sodium pump ␣3 subunit that was made to contain Gln905-Val 930 of the rat gastric proton pump ␣ subunit in place of the original Asn 886 -Ala 911 (13). This vector is named YEpNGH26. Vectors pG1T-r␤1 and pG1T-HK␤ were used for the expression of the rat sodium pump ␤1 subunit and the rat gastric proton pump ␤ subunit, respectively (39).
Yeast Cells Used for Expression-The S. cerevisiae strain 30-4 (MAT-␣, trp1, ura3, Vn2, GALϩ) was used for transformation with either of the yeast expression vectors YEpNGH26 (coding for the chimeric sodium pump ␣ subunit) or YEpr␣3 (coding for the wild-type sodium pump ␣3 subunit of the rat) together with either of the vectors pG1T-r␤1 or pG1T-HK␤ coding for the ␤ subunits of rat Na ϩ ,K ϩ -ATPase and gastric H ϩ ,K ϩ -ATPase, respectively. Thus, coexpression of the various vectors results in the four different combinations of subunits given in Table I. The conditions for the selection of transformants are described elsewhere (40). Microsomes from transformed or nontransformed yeast cells were prepared as described previously (40). Protein content of the microsomal preparations was determined by the method of Lowry (41).
Detection of NK␣3, NGH26, and HK␤ Subunits in a Western Blot-The conditions for electrophoresis and immunoblot detection of ␣3 or NGH26 subunits have been described previously in detail (13). Briefly, 100 g of microsomal protein isolated from yeast expressing either the NK␣3/NK␤ or NGH26/HK␤ heterodimers was run on a 7.5% SDSpolyacrylamide gel and was then electrotransferred onto nitrocellulose membranes by semidry blotting at 0.8 mA/cm 2 . The ␣3 and NGH26 subunits were visualized by using monoclonal antibody 5 against sodium pump ␣ subunits (1:400 dilution) and the commercially available enhanced chemiluminescence kit (ECL) following the protocol of the provider.
HK␤ subunits in membrane preparations from cells expressing either NK␣3/HK␤ or NGH26/HK␤ were detected in a similar way. Here, however, 35 g of microsomal protein were sufficient for electrophoresis and semidry blotting. The monoclonal antibody 2/2E6 served as a primary antibody against the gastric pump ␤ subunit (1:2000 dilution). All other conditions were the same as above.
Estimation of the Gastric Proton Pump ␤ Subunits in Yeast Membrane Preparations by an Antibody Capture Assay-An antibody capture assay (42) was used with some minor modifications (43) to estimate the expression level of the HK␤ in membrane preparations from yeast expressing either NK␣3/HK␤ or NGH26/HK␤. Microsomal proteins from nontransformed cells served as a control.
A total of 35 g of yeast microsomal protein in 50 l of phosphatebuffered saline containing 0.1% Tween 20 (PBS-T) was added to the wells of a microtiter plate. After incubation for 2 h at room temperature, the microtiter wells were washed twice with PBS-T. Afterward, 150 l of blocking buffer composed of 3% bovine serum albumin in PBS-T with 0.02% sodium azide were added overnight to saturate the remaining protein binding sites of the well. Then, the wells were washed once again as described before. To each of the wells 50 l of a monoclonal antibody (2/2E6) against the gastric proton pump ␤ subunit were added at a 1:2000 dilution in PBS-T and incubation continued for 2 h at room temperature. Unbound antibody was then removed by four washes with PBS-T, and afterward 50 l of an alkaline phosphatase-conjugated anti-mouse IgG were added to the wells. After 2 h of incubation at room temperature, the wells were washed four times with PBS-T, and then twice with 10 mM diethanolamine, pH 9.5, containing 5 mM MgCl 2 . All further steps concerning the time course of the formation of p-nitrophenolate from p-nitrophenyl phosphate were carried out as described (42).
Ouabain Binding to Yeast Microsomal Preparations-A total of 250 g of yeast microsomal protein was incubated for 1 h at 30°C with 5 mM PO 4 (Tris form), 5 mM MgCl 2 , 10 mM Tris/HCl, pH 7.5, and various concentrations of [ 3 H]ouabain. The incubation volume was 250 l. Thereafter, the protein was pelleted by centrifugation for 5 min at 12,000 ϫ g. After washing twice with 1 ml of ice-cold water, the pellet was dissolved in 250 l of 1 M NaOH by incubation for 15 min at 80°C. The NaOH was neutralized by 250 l of 1 M HCl, and the radioactivity was determined by scintillation counting after the addition of scintillation fluid.

ATP-promoted Binding of [ 3 H]
Ouabain-A total of 125 g of microsomal protein was incubated for 5 min at 30°C in 10 mM Tris/HCl, pH 7.5, 50 nM [ 3 H]ouabain, 5 mM MgCl 2 , 50 mM NaCl, and various concentrations of ATP (Tris form). The total volume of each sample was 250 l. Thereafter, the protein was pelleted by centrifugation at 12,000 ϫ g for 5 min, washed twice with H 2 O at 4°C, dissolved in 250 l of 1 M NaOH, and processed as described at the end of the preceding paragraph.
Effect of SCH 28080 on the ATP-promoted Binding of [ 3 H]Ouabain-To investigate whether SCH 28080 inhibits binding of [ 3 H]ouabain to yeast membranes, the above experiment was repeated at 0, 1, and 10 M ATP in the presence of 25 M SCH 28080. All other conditions were the same as described above.
Palytoxin-induced K ϩ Efflux from Yeast Cells Expressing Wild-type and Chimeric Na ϩ ,K ϩ -ATPase-Single yeast colonies were incubated at 30°C in a cell incubator with vigorous shaking overnight in 5 ml of SD growth medium (6.7 g/liter Bacto Yeast nitrogen base without amino acids) supplemented with d-galactose (20 g/liter SD medium) as the only carbon source. The cell suspension was then transferred to 200 ml of SD medium and incubation continued for an additional 24 h under the same conditions. Thereafter, cells were collected by centrifugation for 5 min at 3000 ϫ g, washed twice with 30 ml of NaGHBC (200 mM NaCl, 100 mM glucose, 10 mM HEPES, 0.5 mM boric acid, 1 mM CaCl 2 , adjusted to pH 7.5 with Tris), and suspended in NaGHBC to a final concentration of 10 6 cells/ml (A 600 ϭ 6).
For the measurement of the palytoxin-induced K ϩ loss from yeast cells, 450 l of the cell suspension were incubated with various concentrations of palytoxin in the absence or presence of ouabain or SCH 28080 for 2 h at 30°C. Since SCH 28080 was dissolved in dimethyl sulfoxide:ethanol (1:1, v/v), the content of the solvent was kept constant in all samples; the final volume of each sample was 500 l. To determine the total cellular K ϩ content (100% value), 450 l of the yeast cell suspension were incubated for 1 h at 70°C with 50 l of 1% lithium dodecyl sulfate (w/v). Afterward, cells were centrifuged at 12,000 ϫ g for 2 min in a bench-top centrifuge. The supernatants were collected and measured for their K ϩ content in a flame photometer.
Initial rates of palytoxin-induced K ϩ efflux were measured in a flow ionometer. Yeast cells were suspended at a concentration of 15 ϫ 10 6 cells/ml in NaGHBC buffer containing 0.2, 0.5, or 5 M palytoxin for cells with NK␣3/NK␤1; 0.5, 2, or 5 M palytoxin for cells expressing NK␣3/HK␤; and 1, 2.5, or 5 M palytoxin for cells expressing the NGH26/HK␤ hybrid. The total volume of the solution was 1 ml. Using a peristaltic pump, the reaction mixture was transported to a chamber containing a K ϩ -sensitive electrode. The time elapsed between palytoxin addition and signal recording was 7 min, and recording was continued for an additional 5 min. The electric signals derived from the K ϩ -sensitive electrode were digitized and analyzed by a computer program provided by the supplier of the flow ionometer.
Data Analysis-Data are plotted in the figures as mean values Ϯ standard deviations. The lines through the data are the best fits of appropriate equations obtained using the programs InPlot 4.0 or Prism (GraphPad Software, San Diego, CA).
Materials-Growth media were purchased from Difco (Detroit, MI). Palytoxin was obtained from Dr. L. Bèress (Christian-Albrechts-Universitä t, Kiel, Germany). The yeast strain 30-4 was obtained from Dr. R. Hitzeman (Genentech, CA). Restriction endonucleases were from MBI Fermentas (Vilnius, Lithuania) or Amersham Buchler (Braunschweig, Germany). Amersham Bucher was also the provider of [ 3 H]ouabain (28 Ci/mmol) and of the enhanced chemiluminescence Western blot analysis system. The nitrocellulose membranes were from Schleicher & Schuell (Dassel, Germany). The microtiter plates Max-iSorb were from Nunc (Wiesbaden, Germany). Alkaline phosphataseconjugated anti-mouse IgG was purchased from SeroTec (Oxford, United Kingdom). The hybridoma cells producing monoclonal antibody 5 were purchased from Developmental Studies Hybridoma Bank Project, University of Iowa (Iowa City, IA). The antibody 2/2E6 against the gastric proton pump ␤ subunit was a gift of Dr. C. Okomoto (University of Southern California, Los Angeles, CA). The flow ionometer is a product of ZABS GmbH (Marburg an der Lahn, Germany).

Binding of [ 3 H]Ouabain-In the presence of phosphate (P)
and Mg 2ϩ , Na ϩ ,K ϩ -ATPase (E 2 ) becomes phosphorylated (E 2 -P) and binds ouabain with high affinity [E 2 *-P⅐ouabain]. Use of radioactive ouabain enables one to determine the affinity of the enzyme for ouabain under these conditions. Thus, as expected, the rat wild-type sodium pump ␣3 and ␤1 subunits (NK␣3/NK␤1) in microsomal yeast membranes bind [ 3 H]ouabain with high affinity (Fig. 1). The K d determined from a Scatchard plot is 7.7 Ϯ 0.3 nM (Table II). A similar affinity for ouabain was obtained with membranes from cells expressing the NK␣3/HK␤ subunits (K d of 6.4 Ϯ 1.4 nM). The B max for ouabain binding to the NK␣3/HK␤ hybrid (0.85 Ϯ 0.07 pmol/ mg) is slightly reduced compared with that of the wild-type sodium pump (1.1 Ϯ 0.02 pmol/mg) (Table II). Ouabain binding to microsomal membranes from cells expressing the chimeric sodium pump ␣ subunit (NGH26) together with the sodium pump ␤1 subunit (NK␤1) is almost undetectable (data not shown).
Replacement of the sodium pump ␤1 by the proton pump ␤ subunit results in the formation of NGH26/HK␤ complexes capable of binding ouabain (Fig. 1). The equilibrium dissociation constant K d for ouabain binding to the NGH26/HK␤ heterodimer (21.1 Ϯ 2.5 nM; Table II) is slightly increased compared with complexes assembled with NK␣3, indicating a reduction in affinity. In addition, the maximum amount of ouabain bound by NGH26/HK␤ is also reduced (0.18 Ϯ 0.01 pmol/mg), accounting for only a fraction (17%) of the total binding obtained with the wild-type sodium pump ( Fig. 1; Table II).
Expression of NK␣3, NGH26, and HK␤ Subunits in Yeast-To investigate the relative level of expression of the ␣3 and NGH26 subunits, microsomal proteins isolated from transformed yeast were probed in a Western blot with a monoclonal antibody raised against the sodium pump ␣ subunit. Fig. 2 shows that both subunits are expressed in the yeast in comparable quantities. As shown in the same figure, the antibody does not recognize any protein of ϳ100 kDa corresponding to the ␣3 or NGH26 subunit in microsomes isolated from nontransformed yeast cells.
The expression levels of the gastric proton pump ␤ subunit (HK␤) were measured in yeast membrane preparations from cells expressing either NK␣3/HK␤ or NGH26/HK␤ using a monoclonal antibody raised against the gastric proton pump ␤ subunit. The Western blot presented in Fig. 3A shows that the antibody recognizes two protein bands of about 40 and 43 kDa in membranes from cells expressing the NK␣3/HK␤ or NGH26/ HK␤ heterodimers. Since corresponding bands are not detected in membranes from nontransformed cells, these bands at 40 and 43 kDa are gastric pump ␤ subunits, possibly glycosylated to various extents. It is apparent from the figure that the abundance of the HK␤ subunit in membrane preparations from cells expressing the NK␣3/HK␤ complex is higher than in membranes from cells expressing the NGH26/HK␤ hybrid.
Quantification of the HK␤ subunit expression levels in yeast microsomes from cells expressing either the NK␣3/HK␤ or the NGH26/HK␤ hybrid cannot easily be addressed by the Western blotting method. This was done instead by applying a variation of an antigen capture assay (42). In this assay the antigen (here HK␤) is affixed to the wells of a microtiter plate. This affixed antigen is then incubated with an antibody specific for the antigen (here antibody 2/2E6 against HK␤). Subsequently, an alkaline phosphatase-conjugated secondary antibody (here an anti-mouse IgG) is added to the well. With sufficient washes between additions, the antigen, the primary antibody, the secondary antibody, and alkaline phosphatase should be present in equal amounts. Thus, the amount of p-nitrophenyl phosphate hydrolyzed by the alkaline phosphatase is directly pro- portional to the number of HK␤ subunits present in each well, and, therefore, the amount of the p-nitrophenolate anion produced by the alkaline phosphatase can be used as measure of the relative abundance of HK␤ subunits. Fig. 3B shows the results of this experiment. After either 40 or 60 min of incubation with substrate, the amount of the p-nitrophenolate anion formed in samples containing the HK␤ from cells expressing the NGH26/HK␤ heterodimers accounts for only 13-17% of the value obtained from the microsomes from cells expressing the NK␣3/HK␤ subunit combination.
Binding of [ 3 H]Ouabain in the Presence of ATP, Na ϩ , and Mg 2ϩ -The sodium pump can be phosphorylated either by inorganic phosphate in the presence of Mg 2ϩ (the conditions of the experiment in Fig. 1) or by ATP in the presence of Na ϩ and Mg 2ϩ (44). In both cases, ouabain binds to the E 2 -P conformational state of the enzyme and forms a stable and easily measurable [E 2 *-P⅐ouabain] complex. By measuring the formation of the [E 2 *-P⅐ouabain] complex as a function of the ATP concentration, one can determine whether ATP is hydrolyzed to yield the phosphoenzyme, and, if so, it is possible from these experiments to obtain an EC 50 value indicating the relative affinity of the enzyme for ATP (43,45). Fig. 4  Palytoxin-induced Potassium Efflux from Yeast Cells-Yeast cells expressing the NK␣3/NK␤1, NK␣3/HK␤, or NGH26/HK␤ subunits were incubated for 2 h with various concentrations of palytoxin. Thereafter, cells were removed by centrifugation and K ϩ in the supernatant was determined by flame photometry. As shown in Fig. 5, the interaction of palytoxin with the NK␣3/NK␤1, NK␣3/HK␤, and NGH26/HK␤ complexes results in a loss of up to 80% of cytosolic K ϩ levels in yeast. The EC 50 value for the palytoxin-induced K ϩ efflux from cells expressing the NK␣3/NK␤1 subunits is 136 nM. The corresponding value obtained with cells expressing the NK␣3/HK␤ subunits is approximately 2-fold higher (313 nM), and the EC 50 obtained with cells expressing the NGH26/HK␤ subunits is about 6-fold higher (822 nM) than for NK␣3/NK␤1. Palytoxin has no effect on nontransformed yeast cells and only a small loss of K ϩ is observed from yeast cells that express the chimeric NGH26 subunit together with the rat sodium pump ␤1 subunit (data not shown).
To estimate the initial rates of the palytoxin-induced efflux, cells expressing either NK␣3/NK␤1, NK␣3/HK␤, or NGH26/ HK␤ complexes were incubated at various concentrations of palytoxin (see "Experimental Procedures") and K ϩ was measured in the extracellular medium using a flow ionometer. During the period of recording, the initial rate of K ϩ efflux (V in ) at each palytoxin concentration was linear and increased with increasing palytoxin concentrations. The rate of efflux can be described by the formula V in ϭ k K ϩ ⅐N⅐P o where k K ϩ is the single channel conductance, N is the number of channels and is equal to the B max for ouabain binding, and P o is the probability that the channels are open. heterodimers were separated by SDS-polyacrylamide gel electrophoresis and then electrotransferred onto nitrocellulose membranes. Expressed proteins were visualized by using the monoclonal antibody 2/2E6 against HK␤ and the enhanced chemiluminescence Western blot analysis system. B, a total of 35 g of microsomal protein was attached to the bottom of the wells of microtiter plates and was incubated first with the 2/2E6 antibody against HK␤ and then with an alkaline phosphatase-conjugated secondary antibody. After either 40 or 60 min of incubation with p-nitrophenyl phosphate, the p-nitrophenolate anion was determined by measuring the absorbance at 405 nm using a molar absorbance coefficient ⑀ ϭ 18,500 liters/mol/cm. Microsomal protein from nontransformed cells served as a control, and values obtained with the control microsomes were subtracted from the rest.  Table II as mean values for all palytoxin concentrations.
To investigate the effect of the gastric proton pump-specific inhibitor SCH 28080, yeast cells expressing NK␣3/NK␤1, NK␣3/HK␤, or NGH26/HK␤ complexes were incubated at 400 nM palytoxin with various concentrations of SCH 28080. After an incubation period of 2 h, K ϩ in the supernatant was determined as described under "Experimental Procedures." As shown in Fig. 7A, only cells expressing the NGH26/HK␤ complexes are sensitive to SCH 28080. The palytoxin-induced K ϩ efflux from these cells is almost completely inhibited by 50 M SCH 28080 with an IC 50 of 14.3 Ϯ 2.4 M (Fig. 7B). DISCUSSION Because of their importance as receptors for clinically relevant drugs, much effort has been put into identifying sequences of the proton and sodium pumps that participate in the recognition of omeprazole, SCH 28080, and ouabain and related cardiac steroids. Analysis of the properties of sodium pump mutants allowed for the identification of several sites involved in recognition of ouabain (27,28). Comparatively little is known, however, about amino acids or peptides involved in the binding of omeprazole or SCH 28080 to the gastric proton pump. The omeprazole binding site contains Cys 892 within the extracellular loop connecting the M7 and M8 membrane spans of the proton pump ␣ subunit (17). Since omeprazole binding to the proton pump can be inhibited by SCH 28080 (31), it is possible that the binding sites for both substances are identical or in close proximity to each other. Thus, the M7/M8 extracellular peptide of the gastric pump ␣ subunit might be part of the binding site for SCH 28080. This peptide corresponds to a sequence of the sodium pump ␣ subunit that has been shown to contain amino acids involved in the binding of ouabain (28). The corresponding region of the rat gastric proton pump ␣ subunit contains the peptide Gln 905 -Val 930 , which is involved  in assembly with the gastric proton pump ␤ subunit (13), much as the corresponding Asn 886 -Ala 911 peptide of the sodium pump ␣ subunit is involved in assembly with the sodium pump ␤ subunit (46).
To investigate a possible involvement of the M7/M8 extracellular peptide of the gastric proton pump ␣ subunit in SCH 28080 recognition, wild-type sodium pump ␣ subunits or ␣ subunit chimeras containing Gln 905 -Val 930 of the gastric proton pump ␣ subunit were coexpressed with either sodium or proton pump ␤ subunits in yeast (Table I) and were investigated with respect to their ability to recognize either the sodium pump-specific inhibitors ouabain and palytoxin or the gastric proton pump-specific inhibitor SCH 28080.
Ouabain Binding-Microsomal membranes prepared from yeast cells expressing the various subunit combinations bind [ 3 H]ouabain with high affinity ( Fig. 1; Table II). The NGH26/ HK␤ heterodimer was characterized by a reduced ouabain binding affinity (K d ϭ 21.1 Ϯ 2.5 nM; Table II), and a lower maximum ouabain binding capacity (0.18 Ϯ 0.01 pmol/mg) than the NK␣3/NK␤ complex. The ouabain binding capacity of the NGH26/HK␤ heterodimer is only 17% of the total binding obtained with the wild-type sodium pump ( Fig. 1; Table II), and is similar to the expression level that has previously been reported for the NGH26/HK␤ complex (13). To investigate the reasons for the lower ouabain binding capacity, the expression levels of the NK␣3, NGH26, and HK␤ were measured in microsomal preparations by Western blots and antibody capture assays. Fig. 2 shows that the level of expression of the NGH26 subunit is approximately the same as the level of expression of the ␣3 subunit. Fig. 3, however, indicates that HK␤ is present in microsomal membranes from yeast cells expressing the NGH26/HK␤ complex at lower levels than HK␤ is present in microsomes from yeast expressing NK␣3/HK␤. Results of the antibody capture assay indicate that only about 13-17% of the amount of p-nitrophenolate obtained from NK␣3/HK␤ was formed by membranes containing the NGH26/HK␤ complex (Fig. 3B). Thus, it appears that the reduced expression level of HK␤ in membranes from cells expressing the NGH26/HK␤ heterodimer limits the number of NGH26/HK␤ complexes, and the low level of ouabain binding shown in Fig. 1 is a measure of the number of these complexes in the membranes. The limiting effect of HK␤ expression on the number of assembled pumps can be explained by the requirement of the expression system for two different plasmids to direct the synthesis of the ␣ and ␤ subunits independently in the yeast cells. Transformation of yeast with the two plasmids results in cells with different copy numbers for the two plasmids.
Extraction of the yeast membranes with SDS is necessary to measure ouabain-sensitive ATP hydrolysis because the detergent inactivates endogenous yeast ATPases and enriches yeast membranes in the heterologously expressed Na ϩ ,K ϩ -ATPase (40). The NGH26/HK␤ complex is easily denatured when yeast membranes containing this complex are incubated with SDS, however, and it was not possible to measure ouabain-inhibited ATPase activity in yeast membranes containing this chimera. Thus, to show that the NGH26/HK␤ complex binds ATP and is autophosphorylated by ATP, we measured binding of ouabain in the presence of Na ϩ , Mg 2ϩ , and ATP. Under these conditions (see "Experimental Procedures"), ATP hydrolysis and autophosphorylation of the enzyme by ATP promotes binding of ouabain (44).
ATP promoted binding of [ 3 H]ouabain by all ␣␤ complexes examined. The maximum amount of ouabain bound by membrane preparations containing either NK␣3/NK␤1 or NK␣3/ HK␤ was nearly the same, and was similar to the maximum amount of ouabain bound when the pumps are phosphorylated by inorganic phosphate (Fig. 1). The maximum amount of ouabain bound by the NGH26/HK␤ heterodimer was only about 17% of the amount bound by NK␣3/NK␤1 or NK␣3/HK␤, however, consistent with the results obtained from the experiments shown in Figs. 1 and 3. The EC 50 of the NGH26/HK␤ chimera for ATP (1.13 Ϯ 0.51 M) is also very similar to the EC 50 of the wild-type Na ϩ ,K ϩ -ATPase for the nucleotide in the same assay (Table II). These data confirm that the NGH26/HK␤ chimera is capable of forming a phosphoenzyme both from inorganic phosphate and from ATP, although they do not address the question whether the NGH26/HK␤ chimera is capable of completing the catalytic cycle.
Ouabain binding to yeast membranes containing the different pumps was measured at 0, 1, and 10 M ATP in the presence of 25 M SCH 28080. SCH28080 is a specific inhibitor of gastric H ϩ ,K ϩ -ATPase and inhibits the binding of omeprazole to this enzyme. At this concentration, SCH 28080 had no effect on the binding of ouabain to membranes containing either NK␣3/NK␤1 or NK␣3/HK␤ hybrids (data not shown). This result is in agreement with data published previously about the effect of SCH 28080 on sodium pump activity (11,47). 25 M SCH 28080 caused a small (25%) reduction in ouabain binding to NGH26/HK␤, but because of the uncertainties associated with the measurements this result was not statistically significant.
Palytoxin-induced Potassium Efflux from Yeast Cells-Palytoxin has no effect on nontransformed yeast cells, and only a small loss of K ϩ is observed from yeast cells that express the chimeric NGH26 subunit together with the rat sodium pump ␤1 subunit (data not shown). This small effect on cells expressing NGH26/NK␤1 may be due to interaction of palytoxin with the very low number of functional pumps in the yeast membrane (13). Alternatively, the assembly of NGH26 and NK␤1 may affect the binding of palytoxin or the conduction of ions through the channel after palytoxin binds to the heterodimer. Because of the small magnitude of the K ϩ loss, however, investigation of the effects of palytoxin on this subunit combination was not continued any further.
When yeast cells expressing either the NK␣3/NK␤1, NK␣3/ HK␤, or NGH26/HK␤ subunits are exposed to various concentrations of palytoxin, they lose up to 80% of their intracellular K ϩ within 2 h (Fig. 5). At all concentrations of palytoxin, the K ϩ concentration inside the cell at the end of the incubation period represents a steady state between K ϩ efflux through the palytoxin-induced channel and K ϩ influx through the yeast TRK1 and TRK2 uptake pathways. At the end of the incubation was not possible to test for reduced palytoxin binding affinity since radiolabeled palytoxin is not available, analysis of the initial rates of K ϩ efflux from the cells at various palytoxin concentrations shows that a difference in the number of channels is sufficient to explain the data. From the data presented in Table II it can be seen that dividing kЈN by the number of channels (B max ) yields nearly identical values of the apparent single-channel conductance kЈ for all of the ␣␤ heterodimers. These values are 0.23 Ϯ 0.04 for NK␣3/NK␤1, 0.17 Ϯ 0.09 for NK␣3/HK␤, and 0.27 Ϯ 0.02 for NGH26/HK␤. The similarity of these values indicates that the product of the single channel conductance (k K ϩ ) and the constant of proportionality (k) between palytoxin concentrations and the open channel probability is the same. Although values for each of these constants cannot be determined independently from these measurements, it is possible that the single-channel conductance and the open probability are similar for all of the heterodimers.
Inhibition of Palytoxin-induced Potassium Efflux by Ouabain and SCH28080 -The palytoxin-induced K ϩ efflux from yeast cells expressing any of the subunit combinations NK␣3/ NK␤1, NK␣3/HK␤, or NGH26/HK␤ is completely inhibited by ouabain (Fig. 6). At 400 nM palytoxin, ouabain inhibits the K ϩ efflux from cells expressing the different heterodimers with IC 50 values of 23-28 M (Table II). The small differences between the IC 50 values for ouabain inhibition of palytoxin-induced K ϩ efflux in the different heterodimers is similar to the pattern of K d values for ouabain binding to the heterodimers (13).
As expected from the known insensitivity of Na ϩ ,K ϩ -ATPase to SCH 28080, there was Ͻ10% inhibition by SCH 28080 of K ϩ efflux induced by 400 nM palytoxin from yeast cells expressing either the NK␣3/NK␤1 or the NK␣3/HK␤ subunits (Fig. 7A). In contrast, the palytoxin-induced K ϩ efflux from yeast cells expressing the NGH26/HK␤ subunits is almost completely inhibited by 50 M SCH28080. The IC 50 value for the SCH 28080 inhibition of the palytoxin-induced K ϩ efflux is 14.3 Ϯ 2.4 M (Fig. 7B). This value should be compared with the IC 50 values determined for the inhibition of proton transport into porcine gastric vesicles (0.5 M) or for SCH 28080 inhibition of aminopyrine uptake in rabbit gastric glands (0.2 M) (16). Thus, sensitivity to SCH 28080 can be conferred upon Na ϩ ,K ϩ -ATPase by substitution of 26 amino acids from gastric H ϩ ,K ϩ -ATPase (Gln 905 -Val 930 ) for the homologous residues in the extracellular loop connecting transmembrane segments 7 and 8 of Na ϩ ,K ϩ -ATPase ␣3 subunit.
It is not known whether the extracellular loop connecting transmembrane segments 7 and 8 is directly involved in binding SCH28080 or whether the binding site is induced elsewhere in the protein as a result of chimera formation. If the extracellular loop is part of the binding site, it is likely that the binding site also consists of additional parts of the polypeptide. Using chimeric ATPases, Blostein et al. (29) concluded that the amino-terminal half of the gastric proton pump must be involved in SCH 28080 binding, and this is in good agreement with the results of Munson et al. (30), who labeled a 9.9-kDa tryptic peptide starting at Gln 104 of the gastric proton pump ␣ subunit with a photoactivated derivative of SCH 28080. An aminoterminal localization of the SCH 28080 binding site was further supported by the observation that the rat gastric proton pump peptide Val 115 -Ile 126 from the M1 membrane-spanning domain confers SCH 28080 sensitivity to the sodium pump (11). In that report, however, SCH 28080 did not inhibit palytoxin-induced K ϩ efflux from yeast cells expressing this chimera. The results of that investigation have recently been questioned by Asano et al. (48), however, who found that replacement of the Val 115 -Ile 126 sequence of gastric H ϩ ,K ϩ -ATPase with the corresponding region of Na ϩ ,K ϩ -ATPase ␣ subunit did not abolish SCH 28080 sensitivity. A possible explanation for this apparent discrepancy is suggested by the possibility that SCH28080 binding is induced indirectly by chimera formation and that the different chimeras have different properties because of their different sequences. Thus, mutations within Val 115 -Ile 126 do not necessarily have to preclude interactions of SCH 28080 with the protein, since the other points of attachment are possibly enough to stabilize binding. Asano et al. implicated Thr 825 and Pro 829 as possible determinants of the affinity of the enzyme for SCH 28080 (48), together with Glu 822 in the M6 membrane-spanning domain of gastric H ϩ ,K ϩ -ATPase (49).
The finding that the extracellular peptide Gln 905 -Val 930 of the proton ATPase confers SCH 28080 sensitivity to Na ϩ ,K ϩ -ATPase is consistent with previous findings demonstrating that SCH 28080 competes with omeprazole and that omeprazole reacts with Cys 892 from the M7/M8 connecting loop (31,50). A recent investigation suggested that the binding sites for omeprazole and SCH 28080 are not identical but do overlap (51). Taking our results into consideration, one can imagine that the interaction of SCH 28080 with the Gln 905 -Val 930 peptide reduces the accessibility of Cys 892 for omeprazole. Furthermore, conformational changes induced by either inhibitor binding or chimera formation may easily account for participation of other benzimidazole-reactive cysteines in SCH28080 binding  (51). The close proximity between the Gln 905 -Val 930 peptide and Cys 892 , the fact that both SCH 28080 and omeprazole inhibit H ϩ secretion catalyzed by the proton pump, and the demonstration that SCH 28080 inhibits the palytoxin-induced K ϩ efflux from cells expressing the NGH26/HK␤1 heterodimer implicates the M7/M8 loop of ␣ subunit in ion conduction by the proton pump. The homologous region of the sodium pump has also been suggested to be involved in ion binding, occlusion, or translocation (34,45,52). Further investigation of this possibility and the identification of additional peptides and amino acids within the Gln 905 -Val 930 sequence that are important for interactions between the protein and the transported ions, will require additional experiments with new chimeras and mutants. The approach developed in this study may be useful, therefore, in elucidating the mechanisms that lead to SCH 28080 inhibition of H ϩ secretion by the gastric proton pump.