Tissue-specific versus isoform-specific differences in cation activation kinetics of the Na,K-ATPase.

The experiments described in this report reconcile some of the apparent differences in isoform-specific kinetics of the Na,K-ATPase reported in earlier studies. Thus, tissue-specific differences in Na+ and K+ activation kinetics of Na,K-ATPase activity of the same species (rat) were observed when the same isoform was assayed in different tissues or cells. In the case of alpha1, alpha1-transfected HeLa cell, rat kidney, and axolemma membranes were compared. For alpha3, the ouabain-insensitive alpha3*-transfected HeLa cell (cf. Jewell, E. A., and Lingrel, J. B. (1991) J. Biol. Chem. 266, 16925-16930), pineal gland, and axolemma (mainly alpha3) membranes were compared. The order of apparent affinities for Na+ of alpha1 pumps was axolemma approximately rat alpha1-transfected HeLa > kidney, and for K+, kidney approximately alpha1-transfected HeLa > axolemma. For alpha3, the order of apparent affinities for Na+ was pineal gland approximately axolemma > alpha3*-transfected HeLa, and for K+, alpha3*-transfected HeLa > axolemma approximately pineal gland. In addition, the differences in apparent affinities for Na+ of either kidney alpha1 or HeLa alpha3* as compared to the same isoform in other tissues were even greater when the K+ concentration was increased. A kinetic analysis of the apparent affinities for Na+ as a function of K+ concentration indicates that isoform-specific as well as tissue-specific differences are related to the apparent affinities for both Na+ and K+, the latter acting as a competitive inhibitor at cytoplasmic Na+ activation sites. Although the nature of the tissue-specific modulation of K+/Na+ antagonism remains unknown, an analysis of the nature of the beta isoform associated with alpha1 or alpha3 using isoform-specific immunoprecipitation indicates that the presence of distinct beta subunits does not account for differences of alpha1 of kidney, axolemma, and HeLa, and of alpha3 of axolemma and HeLa; in both instances beta1 is the predominant beta isoform present or associated with either alpha1 or alpha3. However, a kinetic difference in K+/Na+ antagonism due to distinct betas may apply to alpha3 of axolemma (alpha3beta1) and pineal gland ( alpha3beta2).

The sodium potassium adenosine triphosphatase (Na,K-ATPase) or sodium pump is responsible for maintaining the electrochemical gradient of Na ϩ and K ϩ across the plasma membrane of animal cells. It normally couples the hydrolysis of one molecule of ATP to the transport of three Na ϩ ions out and two K ϩ ions into the cell (for reviews, see Refs. [1][2][3][4]. This cation pump is a heterodimer comprised of a catalytic ␣ subunit (Ϸ105 kDa) and a highly glycosylated ␤ subunit (45-55 kDa), and may (5,6) or may not (7,8) form larger oligomers. The ␣ subunit contains the binding sites for Na ϩ , K ϩ , ATP, and the highly specific cardiac glycoside inhibitors such as ouabain, as well as the site of phosphorylation (1). The function of the ␤ subunit is not completely understood; it appears to be essential for the normal delivery and correct insertion of ␣ into the plasma membrane (9) and to have some influence on the catalytic activity of ␣ (10 -14). A third peptide subunit known as the ␥ subunit (6.5 kDa) appears to exist in association with ␣ and ␤, at least in certain tissues, although its role is yet to be determined (15).
In mammals, three isoforms of the ␣ subunit (␣1, ␣2, and ␣3) and two of the ␤ subunit (␤1 and ␤2) are known to exist (4). Isoforms of the ␣ subunit are expressed in a tissue-specific manner: ␣1 is present ubiquitously; ␣2 is detected mainly in skeletal muscle, heart, and certain neuronal cells (neurons and astrocytes); and ␣3 is mainly in neurons (4,16).
Earlier studies of cation activation of the Na,K-ATPase by Sweadner (17) using rat kidney and axolemma and later studies by Shyjan et al. (18) using kidney, brain, and pineal gland indicated a higher affinity for Na ϩ in preparations now known to be predominantly ␣3. Thus, the order of apparent affinities for Na ϩ in the former study was axolemma (predominantly ␣3) Ͼ kidney (␣1 only) and in the latter, pineal gland (predominantly ␣3) Ն brain (a mixture of ␣1, ␣2, and ␣3) Ͼ kidney. In contrast, Jewell and Lingrel (19), using membranes isolated from HeLa cells transfected with the individual ␣ isoforms, reported that the order of apparent affinities for Na ϩ is ␣1 Ϸ ␣2* Ͼ ␣3*, and for K ϩ , ␣3* Ͼ ␣2* Ϸ ␣1, where ␣2* and ␣3* denote ouabain-resistant mutants of ␣2 and ␣3, respectively. Moreover, studies of pump-mediated K ϩ (Rb ϩ ) influx into these individual isoform-transfected cells confirmed the general conclusions drawn from the aforementioned work, except that considerably larger kinetic differences among the isoforms were observed (20). Interestingly, in experiments carried out with kidney and axolemma microsomal membranes delivered by membrane fusion into red cells, the order of apparent affinities for cytoplasmic Na ϩ and K ϩ resembled those of ␣1and ␣3*-transfected HeLa cells, respectively (20).
The aim of the experiments described in this study was to reconcile the discordant results obtained in the foregoing studies, as well as numerous earlier reports, regarding the order of apparent affinities for Na ϩ and K ϩ of different tissues and/or isoforms (for review, see Ref. 4). In particular, the question of isoform-specific versus tissue-specific properties of the rat Na,K-ATPase has been addressed by studying the same isoform, either ␣1 or ␣3, in the membranes of various cells. Thus, the properties of the ␣1 isoform were examined in kidney, axolemma and rat ␣1-transfected HeLa cells, and those of the ␣3 isoform, in axolemma, pineal gland, and ␣3*-transfected HeLa cells. The results provide evidence for isoform-independent, tissue-specific modulation of the kinetic behavior of the Na,K-ATPase, the most striking being the differences in the effects of intracellular K ϩ as a competitive inhibitor of Na ϩ at cytoplasmic Na ϩ activation sites.
Cell Culture and Membrane Preparations-Rat kidney microsomes were prepared as described by Jørgensen (23) and stored in a sucrosehistidine-EDTA buffer (SHE buffer: 0.25 M sucrose, 0.03 M histidine, 1.0 mM Tris-EDTA, pH 7.5) at Ϫ70°C. Rat axolemma membranes were prepared as described by Sweadner (24) and stored at Ϫ70°C in a solution comprising 0.315 M Sucrose, 10 mM Tris, and 1 mM EDTA, at pH 7.4. Rat pineal gland membranes were prepared as described by Ceñ a et al. (25), with the following modifications. After sonication (Braun-Sonic 1510 sonicator) four times at low setting for 3 s in SHE buffer, the protein was collected by centrifugation at 100,000 ϫ g for 30 min at 4°C using a TLA100 rotor in a Beckmann TL-100 centrifuge, resuspended in SHE buffer (Ϸ500 l/10 mg of original tissue), and stored at Ϫ70°C. Membranes were isolated from rat ␣1and ␣3*transfected HeLa cells as described elsewhere (19,26) and stored at Ϫ70°C. Protein concentrations of the tissue preparations were determined using the Lowry assay as modified by Markwell et al. (27). Specific activities are indicated in the legends to Figs. 1 and 2.
Enzyme Assays-Membranes were permeabilized as described by Forbush (28). Briefly, they were diluted to 0.06 -0.5 mg/ml and treated for 10 min at 22°C with 1% BSA, 0.65 mg/ml SDS and 25 mM imidazole, after which they were diluted 6-fold with 0.3% bovine serum albumin, 25 mM imidazole. ATP hydrolysis was measured as described previously (29), in a final volume of 100 l containing 30 mM Tris-HCl (pH 7.4), 1 mM EDTA, 3 mM MgCl 2 , and, unless indicated otherwise, concentrations of NaCl varying from 0.5 to 100 mM with KCl kept constant at 10 mM, or KCl concentrations varying from 0.2 to 50 mM with NaCl kept constant at 100 mM, with choline chloride added so that ([NaCl] ϩ [KCl] ϩ [ChCl]) was constant at 150 mM. Prior to the assay, membranes were preincubated in the reaction medium without or with 10 M or 5 mM ouabain for 10 min at 37°C. The reaction was initiated by adding [␥-32 P]ATP (final concentration of 1 mM) and NaCl, KCl, and choline chloride to the concentrations listed above.
Immunoprecipitation and Immunoblotting-Axolemma membranes (0.4 mg/ml) were solubilized for 20 min at room temperature in solubilizing buffer comprising 1% Triton X-100 or 1% CHAPS, 1 0.32% bovine serum albumin, 5 mM EDTA, dissolved in phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 10.1 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 ). The insoluble material was removed by centrifugation for 1 min at 500 ϫ g, and the supernatant was then incubated for 1 h at 4°C with monoclonal mouse antibodies specific for ␣1 or ␣3 (6 g/150 l solubilized axolemma). Protein G covalently linked to agarose beads (Pharmacia Biotech Inc.), pretreated for 3 h at 4°C with goat anti-mouse antibody (100 l of antibody added to 125 l of dry beads), were added to the antibody-treated solubilized axolemma (25 l of the original dry beads added to 150 l of solubilized axolemma) and incubated overnight at 4°C. After several washes of the beads (suspension in 200 l of solubilizing buffer, centrifugation for 1 min at 500 ϫ g), the protein was eluted with 60 l of sample buffer (0.06 M Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, 5% ␤-mercaptoethanol, 0.00125% bromphenol blue), incubated at 37°C for 5 min, and then separated on a 12% SDSpolyacrylamide gel using a Bio-Rad mini gel apparatus as described by Laemmli (30). Proteins were transferred to a polyvinylidine difluoride membrane (Millipore) which was then blocked for 1 h at 37°C in blocking buffer (PBS containing 5% milk powder and 0.1% Tween 20) and probed overnight at 4°C with polyclonal rabbit anti-␣1, -␣3, -␤1, or -␤2 antisera diluted in blocking buffer. After several washes with 0.1% Tween 20 in PBS, the membranes were probed (1 h, 37°C) with horseradish peroxidase-labeled donkey anti-rabbit antibody (diluted 1:5000 in blocking buffer) and exposed for 1 min to the Enhanced ChemiLuminescence (ECL) reagents obtained from Amersham Corp. Densitometry was carried out on several exposures of the Kodak imaging film using a SciScan 5000 scanner and software (U. S. Biochemical Corp.).
Analysis of Kinetic Data-Results are expressed as percentages of V max and were analyzed using the Kaleidagraph computer program with either (i) the equation for the noncooperative, three-site model described by Garay and Garrahan (31): where K Na Ј is the apparent affinity for Na ϩ , and [Na] is the Na ϩ concentration or (ii) the cooperative, n-site model described by the following form of the Hill equation (cf. Ref. 32): where "cat" represents the cation (K ϩ or Na ϩ ). Equation 2 was fitted to the experimental points and the values of V max , K, and n obtained.

RESULTS
Na ϩ and K ϩ Activation Profiles of Distinct ␣ Isoforms-To gain insight into the basis for the discrepancies in apparent cation affinities, a series of experiments were carried out in which the cation activation profiles of pumps of the same ␣ isoform but from different tissues were compared. This comparison was confined to ␣1 and ␣3 of the rat. It was technically not feasible to include ␣2 in the analysis since there are virtually no suitable tissues with predominantly this isoform. (Although ␣2 may predominate in adult skeletal muscle, a high background Mg-ATPase activity precludes meaningful kinetic analysis of Na,K-ATPase.) For ␣1, the tissues compared were kidney, ␣1-transfected HeLa cells, and axolemma. The activity of ␣1 in axolemma was determined by taking advantage of the low sensitivity of the rodent ␣1 isoform to cardiac glycosides. Thus, axolemma ␣1 was assayed in the presence of 10 M ouabain which effectively inhibits ␣2 and ␣3 (33). The ␣3-rich tissues compared were pineal gland, ␣3*-transfected HeLa cells, and axolemma. The difference in activity observed in the absence and presence of 10 M ouabain was ascribed mainly to ␣3 since the proportion of ␣2 in axolemma is relatively low (20). In the case of the pineal gland, we have confirmed the report by Shyjan et al. (18) showing that the predominant ␣ isoform detected in immunoblots of the adult rat pineal gland is ␣3 (results not shown). In addition, ␣1 is also detected, but the immunoblots do not provide information regarding the relative activities of the two isoforms. Therefore, assays to quantify ATPase activity sensitive to low (10 M) versus high (5 mM) ouabain concentrations were carried out, and the results indicated that the activity of ␣1 is less than 5% that of ␣3 in pineal gland (experiment not shown).
The results of kinetic experiments carried out with ␣1-containing membranes isolated from rat kidney, axolemma, and rat ␣1-transfected HeLa cells are shown in Fig. 1, and the results for membranes rich in ␣3, in Fig. 2. The data are expressed as percentages of V max and the curves are best fits to Equation 2. The insets in Figs. 1A and 2A represent the same data fitted to Equation 1.
As shown in Fig. 1A, the apparent Na ϩ affinity of ␣1 from kidney, with the K ϩ concentration held constant at 10 mM, appears somewhat lower than that of ␣1 from either axolemma or HeLa; K 0.5(Na) Ј values were 6.6 Ϯ 0.6, 4.7 Ϯ 0.9, and 5.0 Ϯ 0.3 mM for the three tissues, respectively. At a higher K ϩ concentration (20 mM; cf. Ref. 18), the difference in the K 0.5(Na) Ј value for ␣1 of kidney became greater as indicated below (see Fig. 4). In the case of K ϩ activation (Fig. 1B), the order of apparent affinities (assayed at 100 mM Na ϩ ) are as follows: axolemma Ͻ kidney Ϸ HeLa, with K 0.5(K) Ј values of 2.4 Ϯ 0.5, 0.9 Ϯ 0.1, and 1.1 Ϯ 0.3 mM, respectively. For the ␣3 isoform of the enzyme, Fig. 2A shows that the apparent affinity for Na ϩ of ␣3* from HeLa cells is markedly lower than that of ␣3 from either pineal gland or axolemma; K 0.5(Na) Ј values were 11.1 Ϯ 1.5, 4.9 Ϯ 0.8, and 5.7 Ϯ 0.9 mM, respectively. The K ϩ -activation profiles of ␣3 pumps (Fig. 2B) show that the apparent affinity for K ϩ of HeLa ␣3* pumps is higher than that of other tissues (K 0.5(K) Ј values were 0.7 Ϯ 0.1, 1.6 Ϯ 0.1, and 1.4 Ϯ 0.4 mM, respectively). The kinetic constants and Hill coefficients (n) are shown in Table I.
In other experiments (not shown), the possibility that the results from axolemma membranes were flawed by an incomplete distinction of ␣1 from ␣3 was tested. Thus, since axolemma ␣1 activity is measured as the difference in ATPase activity in the presence of low and high ouabain concentrations, and ␣3, as the difference in activity in the absence and presence of low ouabain, a spuriously lower-than-true apparent affinity of axolemma ␣1 and higher-than-true apparent affinity of ␣3 for K ϩ may have resulted from the well documented K ϩ -mediated decrease in ouabain binding (34). In other words, incomplete inhibition of ␣3 at high K ϩ concentrations would effect an apparent affinity decrease in the former and affinity increase in the latter curves relating activity to K ϩ concentration. In order to rule out this possibility, even though this K ϩ /ouabain antagonism is minimal in rat brain preparations (34), an experiment was carried out in which the K ϩ concentration was varied in the presence of different concentrations of ouabain (5, 10, and 20 M). It was found that K 0.5(K) Ј for the ␣1 enzyme remained constant at all three ouabain concentrations. Moreover, it should be noted that the enzyme was preincubated with  Table I. A, activation by Na ϩ at 10 mM KCl (V max values are 3.72 Ϯ 0.07, 0.260 Ϯ 0.004, and 0.107 Ϯ 0.003 mol/(mg⅐min) for kidney, axolemma, and HeLa cells, respectively); B, activation by K ϩ at 100 mM NaCl (V max values are 4.32 Ϯ 0.14, 0.330 Ϯ 0.030, and 0.108 Ϯ 0.003 mol/(mg⅐min) for kidney, axolemma, and HeLa cells, respectively). q, kidney; E, axolemma; Ç, transfected HeLa cells. In the insets the data were fitted to Equation 1.
FIG. 2. Activation by Na ؉ and K ؉ of rat ␣3 Na,K-ATPase from pineal glands, axolemma, and transfected HeLa cells. Assays were carried out and analyzed as described in Fig. 1, except that ouabain-sensitive activities attributed mainly to ␣3 are the differences between hydrolysis measured in the absence and presence of 5 mM ouabain (pineal gland) or the absence and presence of 10 M ouabain (axolemma) or in the presence of 10 M and 5 mM ouabain (␣3*transfected HeLa cells). A, activation by Na ϩ at 10 mM KCl (V max values are 0.353 Ϯ 0.011, 3.34 Ϯ 0.07, and 0.078 Ϯ 0.002 mol/(mg⅐min) for pineal gland, axolemma, and HeLa cells, respectively); B, activation by K ϩ at 100 mM NaCl (V max values are 0.465 Ϯ 0.008, 5.19 Ϯ 0.07, and 0.075 Ϯ 0.003 mol/(mg⅐min) for pineal gland, axolemma, and HeLa cells, respectively). q, pineal gland; E, axolemma; Ç, transfected HeLa cells. In the insets the data were fitted to Equation 1. ouabain in the absence of K ϩ (see "Experimental Procedures"), and that the enzyme activity measured thereafter remained constant as a function of time.
K ϩ Interactions at Cytoplasmic Na ϩ Binding Sites-One of the inherent problems in kinetic studies of Na,K-ATPase in membrane fragments is the lack of control of the composition of cations at the cytoplasmic versus extracellular milieu. Specifically, it has been shown that K ϩ binding and inhibition at the cytoplasmic Na ϩ activation sites alters the enzyme's apparent affinity for Na ϩ (31). To determine whether Na ϩ /K ϩ interactions are, indeed, distinct for the sodium pumps of different tissues, a series of activity measurements were carried out at varying K ϩ concentration and Na ϩ maintained constant at a low 5 mM rather than 100 mM concentration. The results shown in Fig. 3, A and B, indicate that the extent of K ϩ -inhibition at the presumably cytoplasmic Na ϩ binding site is at least partly affected by the nature of the tissue. As shown in Fig. 3A, the kidney ␣1 enzyme is significantly more sensitive to K ϩ inhibition than ␣1 from either HeLa cells or axolemma; at 20 mM KCl, a concentration at which the [K ϩ ]/[Na ϩ ] ratio of 4 is still lower than the normal physiological value (Ͼ10), the activity of kidney ␣1 is reduced by 40%, whereas that of the other tissues is minimally affected. The results for ␣3 (Fig. 3B) show that the transfected HeLa enzyme is much more sensitive to inhibition by K ϩ than either the axolemma or pineal gland enzymes.
It has been observed that the antagonistic effect of vanadate, a potent inhibitor of the sodium pump, is facilitated by the presence of K ϩ ions (35,36). To ensure that the K ϩ -mediated inhibition observed in this study is not the result of vanadate present in the kidney preparation, two control experiments were performed: (i) in one, assays were carried out in the presence of 2.5 mM norepinephrine, which reverses the effect of vanadate (36), and (ii) in the other, the assay time was reduced 10-fold and the amount of kidney microsome sample increased 10-fold; if vanadate were present in the microsome suspension, such an increase in endogenous vanadate concentration should result in greater inhibition of activity. Neither of these conditions altered the K ϩ -inhibition profiles, which argues against an apparent K ϩ inhibition secondary to the presence of vanadate in the kidney preparation.
Based on the Albers-Post model of the Na,K-ATPase reaction mechanism and, more specifically, on the model which assumes random binding of Na ϩ and K ϩ to (the same) three equivalent sites on the cytoplasmic side of the enzyme, Garay and Garrahan (31), in their studies on Na ϩ efflux in red cells, and Sachs (37), in studies of ouabain-sensitive ATPase activity in broken red cell ghosts, showed that activity adhered closely to the following relationship: where [Na] in and [K] in are the cytoplasmic concentrations of Na ϩ and K ϩ , respectively. In accordance with this model, the plot of K Na Ј , the apparent affinity for sodium, as a function of K ϩ concentration yields the linear relationship K Na Ј ϭ K Na (1 ϩ [K] in /K K ) (Equation 4; see Ref. 37). From this plot, K Na , the apparent affinity for Na ϩ when the K ϩ concentration is zero, as well as K K , the apparent affinity for K ϩ at the cytoplasmic Na ϩ binding site, are readily obtained (cf. Ref. 37). In order to apply this analysis to our system, a series of experiments were carried out in which K Na Ј was determined at various K ϩ concentrations for each of the tissues studied in Figs. 1 and 2. It should be noted that the plots of K Na Ј shown in Fig. 4, A and B, are best fits to the model described by Equation 1 (see insets of Figs. 1A and 2A), which adheres to the noncooperative assumptions of Garay and Garrahan, while the data of Figs. 1 and 2 were best fits to Equation 2. The values of K Na and K K , representing the cytoplasmic binding constants for Na ϩ and K ϩ , respectively, and calculated from the Fig. 4 plots, are shown in Table II.
In comparing these values, it is evident that the main difference between the ␣1 pumps of kidney and those of HeLa and axolemma is the higher apparent affinity for K ϩ as a competitive inhibitor of Na ϩ . In the case of the ␣3 isoforms, however, the lower apparent affinity for Na ϩ characteristic of transfected HeLa cells ( Fig. 2A) is a function of both a lower affinity for Na ϩ as an activator as well as a higher affinity for K ϩ as a competitive inhibitor, as compared to either the pineal gland or axolemma ␣3 enzymes. Thus, the ratio K Na /K K reflects the ability of K ϩ to compete with Na ϩ for the cytosolic cation binding site and the larger the ratio, the more susceptible the enzyme is to competitive inhibition by K ϩ . A large K Na /K K ratio explains the greater K ϩ inhibition, at low Na ϩ concentration, observed in the case of kidney pumps and HeLa ␣3* pumps, as depicted in Fig. 3, A and B. As well, since K ϩ /Na ϩ antagonism should decrease as the Na ϩ concentration is increased, the curves of enzyme activity versus Na ϩ concentration shift to the right, resulting in the lower apparent affinities for Na ϩ , as Tissue-specific Cation Activation of Na,K-ATPase noted in Figs. 1A and 2A.
Are the Tissue-specific Kinetic Differences the Result of ␣ Associations with Different ␤ Isoforms?-The question as to whether differences in cation activation are due, at least to some extent, to differences in the ␤ isoform which associates with ␣ was approached by carrying out a series of experiments involving immunoprecipitation of ␣1 and ␣3 from axolemma, followed by immunoblotting with ␣ and ␤ isoform-specific antibodies to determine the nature of the associated ␤ subunits. The results of these experiments are shown in Fig. 5 and summarized below. This question is relevant only to axolemma membranes since only ␤1 is present in kidney and only ␤2, in the adult pineal gland. In addition, HeLa cells contain ␤1 message (38,39) and ␤1 protein has been detected by Western blotting 2 ; neither ␤2 message nor protein were detected by polymerase chain reaction, Northern analysis, or Western blotting. 3 When a mouse monoclonal antibody specific for ␣1 was used to immunoprecipitate the enzyme of Triton X-100-solubilized axolemma membranes, the only subunit isoforms detected on Western blots using rabbit polyclonal antisera specific for ␣1, ␣3, ␤1, and ␤2 as primary antibodies, were ␣1 and ␤1 (Fig. 5,  A-D, lanes 3). Further, when a mouse monoclonal antibody specific for ␣3 was used, ␣3 and ␤1 were detected along with a barely visible band corresponding to ␤2 (Figs. 5, A-D, lanes 4). Control experiments carried out omitting the precipitating antibody showed minimal amounts of nonspecific binding of the axolemma Na,K-ATPase subunits to the protein G-linked agarose beads (Fig. 5, A-D, lanes 5). The fact that the appearance of two bands, one of slightly higher mobility than ␣1 and ␣3 (Fig. 5, A and B, lane 3), the other at Ϸ50 kDa (Fig. 5C, lanes  3-5), reflect nonspecific reactions was evidenced in the following controls (not shown). (i) The first band was present even when the primary detecting antibody was omitted, thus indicating that it is the result of nonspecific binding of the secondary blotting antibody to the primary immunoprecipitating antibodies. (ii) The second nonspecific band appeared even when 2 W. J. Ball, unpublished results. 3 R. Levenson, personal communication.  4. Dependance of K Na on K ؉ concentration for rat pumps from kidney, axolemma, pineal gland, and transfected HeLa cells. Assays were carried out as described in Fig. 1, but at varying concentrations of KCl (5, 10, 20, 35, and 50 mM). K Na Ј were first determined by fitting the data obtained for each Na ϩ -activation curve to Equation 1 and were then plotted as a function of KCl concentration. Each point represents an average Ϯ S.D. of at least three separate experiments, and the values of K Na and K K obtained are shown in Table  II. A, ␣1 pumps: q, kidney; E, axolemma; å, transfected HeLa cells. B, ␣3 pumps: q, pineal gland; E, axolemma; å, transfected HeLa cells. the primary antibody (mouse anti-␣1 or -␣3) was omitted (Fig.  5C) or when the procedure was carried out in the absence of solubilized axolemma (not shown), indicating that it probably represents a nonspecific reaction involving the primary blotting antibody to ␤1 and goat anti-mouse antibodies.
␣-␤ Stoichiometries in Axolemma-The stoichiometry of the ␣-␤ associations was then evaluated in order to assess whether these associations may have been disrupted as a result of the membrane solubilization procedure. These analyses were done utilizing different exposures of the Western blots from several replicate experiments which were quantified as described under "Experimental Procedures." In this work, two assumptions were made; first, that the ␣1:␤1 subunit stoichiometry of the kidney enzyme, as based on studies of the purified enzyme, is 1:1 (for example, see Ref. 40) and second, that heterodimers comprising ␤1 are not preferentially immunoprecipitated compared to those comprising ␤2.
The estimate of ␣1:␤1 stoichiometry in axolemma was based on a comparison of the densities of the ␣ and ␤ bands of axolemma immunoprecipitated with anti-␣1 monoclonal antibody with those of unprecipitated kidney microsomes, following exposures to anti-␣1 and anti-␤1 antisera as shown in Fig. 5, A and C (lanes 2 and 3). The ratio of ␣1 to ␤1 in precipitated axolemma was found to be 1.08 Ϯ 0.17 (S.E. for five independent experiments).
Because there is no tissue in which ␣3 and ␤1 have been shown to be expressed in a 1:1 ratio, the question of possible ␣3-␤1 versus ␣3-␤2 associations was evaluated as follows. The ␣3:␤1 ratio as detected in anti-␣3-immunoprecipitated axolemma samples was compared to that found in unprecipitated axolemma membranes, after correcting the ratio for the proportion of ␤1 presumed to associate with ␣1. The ratio in the immunoprecipitate of axolemma sample was found to be reasonably close to the "corrected" ratio observed in the unprecipitated membranes. Thus, if the corrected ␣3:␤1 ratio of unprecipitated axolemma is normalized at 1.00, the ratio of the precipitate is 1.08 Ϯ 0.09 (S.E. for four independent experiments).
It should also be mentioned that in other experiments (not shown) aimed to determine whether the detergent Triton X-100 interfered with subunit interactions, immunoprecipitations were also carried out with a 4-fold lower concentration of Triton X-100 (0.25%) and with 1% CHAPS. Under both conditions, the ␣:␤ ratios obtained were not significantly different from those observed with 1% Triton X-100 (data not shown).

DISCUSSION
In this study we show that the divergent results regarding the relative affinities of the Na,K-ATPase of the different rat isoforms as reported in different laboratories are not simply accounted for by differences in the experimental conditions used. Thus, we have reproduced the relative cation affinities for ␣1 versus ␣3 as reported by Jewell and Lingrel (19) and Munzer et al. (20) on the one hand, and those of Sweadner (17) and Shyjan et al. (18), on the other. To gain insight into the basis for this dichotomy, we have assessed the apparent cation affinities of pumps of the same catalytic isoform, either ␣1 or ␣3, but from different tissues and, therefore, membrane environments. Marked differences in the apparent affinities for both Na ϩ and K ϩ were observed in pumps of the same ␣ isoforms isolated from different cellular sources. These data and previous work in lamb (22) and dog (41) tissues are consistent with the conclusion that factors other than the type of FIG. 5. Coimmunoprecipitation of the ␤ subunit with ␣1 and ␣3 from rat axolemma. Membranes were prepared, solubilized in 1% Triton X-100 and immunoprecipitated with either an ␣1or ␣3-specific monoclonal antibody as described under "Experimental Procedures." Following SDS-polyacrylamide gel electrophoresis, the proteins were analyzed by Western blotting using polyclonal antisera specific for: A, ␣1; B, ␣3; C, ␤1; and D, ␤2. Lanes are: 1, axolemma; 2, kidney; 3, Immunoprecipitate from axolemma using ␣1-specific monoclonal antibody 6H; 4, Immunoprecipitate from axolemma using ␣3-specific monoclonal antibody M7-PB-E9; and 5, control: immunoprecipitation performed in the absence of the primary antibody. Molecular masses are given in kilodaltons.

TABLE II
Affinities for Na ϩ and K ϩ binding to Na ϩ activation sites of the Na,K-ATPase of various tissues K Na and K K values were calculated from the plots shown in Fig. 4 using the relationship KЈ Na ϭ K Na (1 ϩ [K] in /K K ), where KЈ Na is the observed apparent affinity for Na ϩ in the presence of K ϩ . K Na is the apparent affinity for Na ϩ as an activator in the absence of potassium, and was obtained from the y intercept. K K is the apparent affinity for K ϩ as a competitive inhibitor of Na ϩ at the cytoplasmic binding site, and is the negative of the x intercept. Tissue-specific Cation Activation of Na,K-ATPase catalytic isoform influence interactions of the pump with Na ϩ and K ϩ . The most obvious tissue-specific protein component which interacts with the pump is the ␤ subunit. In fact, effects of different ␤ subunits on both K ϩ (10 -12) and Na ϩ (13, 14) affinities have been described. Accordingly, one question is whether the kidney enzyme's lower apparent affinity for Na ϩ and higher apparent affinity for K ϩ as compared to other ␣1 pumps are the result of interactions of ␣1 with different ␤ subunits. To address this question, particularly in axolemma, in which both ␤1 and ␤2 have been identified, the nature of the ␤ subunit which coimmunoprecipitates with the distinct ␣ subunits was assessed. The results of these experiments indicate that ␤1 associates with ␣1 in axolemma and that the stoichiometry of the association is close to 1.0. The determination of ␣/␤1 stoichiometries imply that little, if any, ␤2 associates with either ␣1 or ␣3. Whether ␤2 associates preferentially with ␣2 in axolemma remains to be determined. Although HeLa cells contain human ␤1, while kidney cells contain rat ␤1, these subunits are 95% identical (39). A difference in both type and amount of glycosylation has been observed between kidney and brain ␤1 (42) and is presumably the basis for the differences in mobilities in immunoblots of kidney and axolemma as shown in Fig. 5C. The possibility remains that these differences are at least partly responsible for the distinct kinetics, even though there is evidence that the oligosaccharides are not essential for primary function (reviewed in Ref. 43).
It is unlikely that the presence of distinct ␤s account for differences of ␣1 of kidney, axolemma, and HeLa, and of ␣3 of axolemma and HeLa; in both instances, ␤1 is the predominant ␤ isoform present or associated with ␣1 or ␣3. However, tissuespecific differences in cation affinities, despite similar ␣␤ pairing, does not imply that the ␤ subunit has no effect on function. In fact, a kinetic difference due to the distinct ␤ subunits is observed in the case of ␣3. Thus, as shown in Table II, the ratio K Na /K K is 1.7-fold lower in pineal gland compared to axolemma, reflecting the 1.9-fold difference in K K . That this difference is a result of ␤2 association with ␣3 in the pineal gland and of ␤1 with ␣3 in axolemma is supported by a recent report showing a 1.6-fold higher apparent affinity for Na ϩ of ␣3␤2 compared to ␣3␤1 in Sf-9 cells transfected with these isoform pairs (14). In that study, the Na ϩ activation kinetics from which the kinetic constants were obtained were carried out in the presence of 30 mM K ϩ so that the difference in apparent affinity for Na ϩ may also reflect a difference in K K . Other kinetic differences were not detected. Taken together, these results are consistent with a role for the distinct ␤s in modulating K ϩ interactions at cytoplasmic Na ϩ sites.
An important observation regarding the coimmunoprecipitation studies presented here is the isoform specificity of the reactions as evidenced in the coimmunoprecipitation of ␣ with ␤, but not of ␣1 with ␣3. This lack of coimmunoprecipitation between the different ␣ isoforms is in contradiction with reports that pumps coimmunoprecipitate as ␣ heterodimers in rat brain and in bacculovirus infected Sf-9 cells (5). Although it is possible that the detergent (1% CHAPS) used in that study (5) did not fully solubilize the membranes, or that the Triton X-100 used in this study disrupted ␣-␣ interactions, these are unlikely explanations, since we confirmed our results using 1% CHAPS.
It can be argued that certain methodological procedures may be responsible for some of the differences observed for the ␣3 enzyme of axolemma as compared to that of other tissues. As described above, the activity ascribed to ␣3 in axolemma is that which is sensitive to 10 M ouabain. Unfortunately, the ouabain-affinities of ␣2 and ␣3, both present in axolemma, are quite similar (4), and it was technically difficult to distinguish the two on that basis. However, the amount of ␣2 in axolemma is relatively low (Ϸ25%) (20), so that its effect on the kinetic behavior cannot account for the magnitude of the differences in the observed kinetic constants as discussed below. As well, the similar fold difference in apparent affinity for external K ϩ ascribed to ␣1 versus ␣3 in two separate systems (transfected HeLa cells compared to axolemma-and kidney-fused red blood cells; see Ref. 20) argues against a substantial contribution of ␣2 to the behavior of the ouabain-sensitive pumps of axolemma. For the same reasons, it is unlikely that the mutation of ␣3 to render it ouabain-resistant in HeLa cells alters its behavior.
There has been some evidence that detergents, such as SDS used here to increase the permeability of membrane vesicles to substrates, can have an effect on cation activation kinetics of the Na,K-pump (44). Although such an effect was not observed in the case of the Na ϩ -activation profile of axolemma enzyme, or in the case of the Na ϩ and K ϩ activation profiles of transfected HeLa cells (data not shown), it is entirely possible that other pumps might react differently to SDS treatment. However, the SDS concentration and the SDS:protein concentration ratio were identical in all of our experiments. Therefore, if SDS affects the different enzymes to varying extents, it should be the result of differential interactions with the surrounding environment, which would be consistent with the notion that the catalytic behavior of the pump does not depend solely on the isoform of the ␣ subunit.
The most intriguing results of this study concern tissuedistinct K ϩ /Na ϩ antagonism. Differences in K K , the apparent affinity for K ϩ at (cytoplasmic) Na ϩ activation sites, underlie tissue-specific differences in the sodium-activation profiles noted in both the present and earlier studies (17)(18)(19)(20). There is evidence also of differences in apparent Na ϩ affinity, independent of K ϩ concentration, between pumps of the same catalytic isoform, although this difference was slight in the case of the kidney enzyme compared to other ␣1 pumps (see Table II). In general, the ability of K ϩ to act as a competitive inhibitor of Na ϩ binding is reflected in the ratio of K Na to K K , the cytosolic FIG. 6. Comparison of Na ؉ -activation curves derived from kinetic constants in Table II (20). Data points for ␣1 (q) and ␣3 (å) were taken from Fig. 6 of Munzer et al. (20) and are also expressed as percent of V max . The dashed curves were derived from the K 0.5 Ј values for intracellular Na ϩ shown in Table II  Tissue-specific Cation Activation of Na,K-ATPase binding constants for Na ϩ and K ϩ , respectively. It is apparent from Table II that, of the membrane systems examined, ␣1 in the kidney and ␣3 transfected into the HeLa cell have the largest K Na /K K ratios when compared to other pumps with the same ␣ isoforms. It is these two enzyme preparations which exhibit the greatest sensitivity to inhibition by K ϩ as depicted in Fig. 3. Specifically, it seems that in the case of kidney ␣1, this inhibition is due to its relatively lower K K compared to the other ␣1 pumps, whereas with HeLa ␣3 pumps, it is due mainly to a higher K Na (Table II). Whether these characteristics are intrinsic to the enzyme, for example due to tissue-specific co-or postranslational modification(s), or rather, the result of modulation of the enzyme by another associated protein remains unresolved.
The kinetic analyses of the sigmoid activation kinetics described in this and previous studies (17)(18)(19)(20) are based on conventional cooperative or noncooperative models. Using the noncooperative model, the apparent affinities of ␣1 and ␣3 for intracellular Na ϩ and K ϩ of the same tissue (HeLa) derived in the present study can account for the low affinities of ␣3 compared to ␣1 observed in studies of Munzer et al. (20) using intact cells, with the following provisos. Those authors pointed out that their data points for Na ϩ activation of ␣3 pumps could be obtained only in the region of the curve well below saturation due to the technical difficulty of raising intracellular Na ϩ above Ϸ45 mM. This precluded a reliable estimate of the kinetic constants for ␣3 when using the noncooperative model. However, the data fitted well to a cooperative model (Equation 2 in Ref. 20) giving K 0.5 Ј values for intracellular Na ϩ of 17.6 mM for ␣1 and and 63.5 for ␣3. In the present study, the K Na and K K values for rat ␣1 and ␣3 in HeLa membranes (Table II) were used to obtain the observed apparent affinity, K Na Ј , at 135 mM intracellular K ϩ , a concentration approximating that of the intact cells used by Munzer et al. (20). Using these values of K Na Ј (7.1 mM for ␣1 and 30.3 mM for ␣3) we derived curves of pump activation as a function of varying intracellular Na ϩ using the cooperative model (Fig. 6). The curves (solid lines) and the values of K 0.5 Ј thus obtained (19.2 mM for ␣1 and 75.2 mM for ␣3) are similar to those derived from the data of previous studies (20) with intact cells (Fig. 6, dashed lines). Therefore, these results indicate that the physiologically significant extremely low apparent affinity of ␣3 reflects its much greater sensitivity to inhibition by intracellular K ϩ .
Modulation of pump behavior by the membrane environment is most likely the explanation for the discrepancies among previous reports concerned with isoform-specific behavior. A slightly higher Na ϩ affinity in axolemma compared to kidney was reported first by Sweadner (17), and later by others including Shyjan et al. (18) who compared brain and kidney. This observation was replicated in the present study of kidney and axolemma Na,K-ATPase, with values very close to those reported by Sweadner, i.e. K Na values of 0.72 mM and 1.02 mM, respectively (Table II)  Tissue-specific as well as isoform-specific behavior is evident not only in apparent affinities for cytoplasmic Na ϩ and K ϩ but also for K ϩ at extracellular sites (Table I). Thus, it was only in the same (red cell or HeLa cell) environment that a higher affinity for extracellular K ϩ of ␣3 or axolemma compared to kidney (␣1) was observed (19,20). It may be relevant that exogenous kidney pumps fused into red cells and endogenous red cell pumps behave identically with respect to the apparent affinity for extracellular K ϩ , K K(ext) (45). (The greater difference between ␣1 and ␣3 observed in studies with intact cells may reflect the limitation of kinetic studies with unsided preparations).
The foregoing considerations argue in favor of the conclusion that the primary structure of the ␣ isoform is not the sole determinant of the magnitude of cation affinity/selectivity. Our observations are consistent with the existence of some pump modulator, for example one which interacts and effects a greater sensitivity of ␣1 to K ϩ inhibition in the kidney compared, for example, to ␣1 of axolemma; in the microsome-fused red cell system, association of the ␣ subunit with the putative regulator may be interrupted following its association with other components of the new (red cell) environment. This kind of regulation is reminiscent of the effects of an intrinsic red cell membrane (blood group) antigen, Lp, found in genetically low potassium sheep red blood cells. This protein or glycoprotein interacts with the pump and effects K ϩ -inhibition (for review, see Ref. 46). Interestingly, when kidney pumps are delivered from microsomes into low potassium sheep red cells, the Lp antigen effects susceptibility to K ϩ inhibition (47), which supports the idea that fusion into red blood cells confers a new membrane environment for the pump.
The study described in this report provides evidence in support of the conclusion that factors in addition to the primary structure of the ␣ isoforms dictate the kinetic behavior of the Na,K-ATPase. Likely candidates include other membranebound components or modulation by co-or post-translational modifications of either subunit.