Susceptibility of β1 Na⁺-K⁺ Pump Subunit to Glutathionylation and Oxidative Inhibition Depends on Conformational State of Pump

Background: Glutathionylation of a cysteine in the membrane Na⁺-K⁺ pump β subunit occurs despite its lipid bulk phase location in the currently known structure of the pump molecule.

Results: Glutathionylation was dependent on the conformational changes that occur in the catalytic cycle of the Na⁺-K⁺ pump.

Conclusion: Na⁺-K⁺ pump cycle phase determines glutathionylation.

Significance: Cysteine glutathionylation can depend on protein conformational state.

Abstract

Glutathionylation of cysteine 46 of the β1 subunit of the Na⁺-K⁺ pump causes pump inhibition. However, the crystal structure, known in a state analogous to an E2·2K⁺·P_i configuration, indicates that the side chain of cysteine 46 is exposed to the lipid bulk phase of the membrane and not expected to be accessible to the cytosolic glutathione. We have examined whether glutathionylation depends on the conformational changes in the Na⁺-K⁺ pump cycle as described by the Albers-Post scheme. We measured β1 subunit glutathionylation and function of Na⁺-K⁺-ATPase in membrane fragments and in ventricular myocytes. Signals for glutathionylation in Na⁺-K⁺-ATPase-enriched membrane fragments suspended in solutions that preferentially induce E1ATP and E1Na₃ conformations were much larger than signals in solutions that induce the E2 conformation. Ouabain further reduced glutathionylation in E2 and eliminated an increase seen with exposure to the oxidant peroxynitrite (ONOO⁻). Inhibition of Na⁺-K⁺-ATPase activity after exposure to ONOO⁻ was greater when the enzyme had been in the E1Na₃ than the E2 conformation. We exposed myocytes to different extracellular K⁺ concentrations to vary the membrane potential and hence voltage-dependent conformational poise. K⁺ concentrations expected to shift the poise toward E2 species reduced glutathionylation, and ouabain eliminated a ONOO⁻-induced increase. Angiotensin II-induced NADPH oxidase-dependent Na⁺-K⁺ pump inhibition was eliminated by conditions expected to shift the poise toward the E2 species. We conclude that susceptibility of the β1 subunit to glutathionylation depends on the conformational poise of the Na⁺-K⁺ pump.

Introduction

Glutathionylation is a reversible oxidative modification in which a disulfide bond forms between the cytosolic tri-peptide GSH and a cysteine residue on a protein. The negatively charged GSH adduct can change the structure and function of the protein in a manner reminiscent of the effect of phosphorylation. As is firmly established for phosphorylation, glutathionylation plays an important role in cell signaling (1). We have reported that oxidative signaling induces glutathionylation of the β1 subunit of the membrane Na⁺-K⁺ pump, whereas we did not detect glutathionylation of the α1 subunit (2–4). The β1 subunit contains seven cysteine residues. Six of these, located in the extracellular domain, are linked by three disulfide bonds (5). This leaves only one cysteine, Cys-46, with a free sulfhydryl group as a candidate for glutathionylation. Mutagenesis of α1/β1 Na⁺-K⁺ pump heterodimers expressed in Xenopus oocytes confirmed that Cys-46 was reactive and established that glutathionylation of it was causally related to Na⁺-K⁺ pump inhibition. The β2 and β3 subunit isoforms only have the six cysteine residues linked by disulfide bonds, and pump heterodimers in which these subunits were expressed showed no oxidation-induced inhibition of function (2).

The crystal structure of Na⁺-K⁺-ATPase (EC 3.6.1.3), determined in a state analogous to an E2·2K⁺·P_i configuration, indicates that the side chain of Cys-46 in the β1 subunit is located in the lipid bulk phase of the membrane (6, 7) and hence not expected to be accessible to the hydrophilic GSH in the cytosol. However, P-type ATPases, including the Na⁺-K⁺ pump, undergo large changes in molecular structure during their catalytic cycle, as first indicated by proteolytic cleavage studies and subsequently supported by studies on their crystal structures (8). The sulfhydryl group of Cys-46 can form a disulfide bridge to a cysteine in the α subunit when a trypsin digest of Na⁺-K⁺-ATPase that includes the transmembrane domain of the β1 subunit is exposed to Cu²⁺-phenanthroline. This is not a common feature of membrane-buried cysteines (9) and may be explained by access of the Cys-46 side chain to a hydrophilic milieu under some circumstances. Use of Tris in the cross-linking experiments suggests the enzyme was in an E1-like conformation.

We have examined whether changes in Na⁺-K⁺ pump conformation affect susceptibility of the β1 subunit to glutathionylation and hence susceptibility of pump activity to oxidative stress. We measured glutathionylation, trypsin digestion, and activity of Na⁺-K⁺-ATPase-enriched membrane fragments suspended in solutions commonly used to preferentially stabilize different conformational states. We also measured β1 subunit glutathionylation in lysate of isolated cardiac myocytes that had been incubated under different conditions designed to shift the conformational state of in situ pumps by altering concentrations of Na⁺-K⁺ pump ligands. The whole cell voltage clamp technique was used to control transmembrane concentrations of Na⁺-K⁺ pump ligands and measure electrogenic Na⁺-K⁺ pump current (I_p)⁴ in intact cardiac myocytes under different conditions expected to affect the conformational poise of the Na⁺-K⁺ pump.

EXPERIMENTAL PROCEDURES

Isolation of Na⁺-K⁺-ATPase-enriched Membrane Fragments

Na⁺-K⁺-ATPase from the outer medulla of pig kidney was prepared according to the method of Jorgensen (10) using SDS extraction followed by differential centrifugation. The specific activity was ∼1800 μmol·mg⁻¹·h⁻¹ at 37 °C. Sarcolemmal microsomes from pig heart were prepared essentially as described by Klodos et al. (11) for preparation of kidney plasma membranes. The microsomes were further purified on a discontinuous sucrose density gradient (from 8.5 to 60%) followed by incubation overnight with SDS. The specific activity was 50–200 μmol·mg⁻¹·h⁻¹ at 37 °C.

Glutathionylation of Na⁺-K⁺-ATPase β1 Subunit

Two independent techniques were used to detect glutathionylation. For the Biotin-GSH technique, we prepared biotinylated GSH ester as described (12) and incubated Na⁺-K⁺-ATPase preparations with 0.5 mm biotin-GSH for 15 min at 37 °C (13). Streptavidin-Sepharose beads were then added. After incubation for 1 h at 4 °C, we washed the beads five times with buffer containing 0.1% SDS. The glutathionylated proteins were released, and immunoblotting was used to detect the β1 subunit as described (2). When indicated, the enzyme was exposed to 0.5 mm ONOO⁻ before biotin-GSH ester was added 5 s later (13). The concentration of ONOO⁻ can be regarded as nominal only because of the short half-life of the compound (seconds). For the GSH antibody technique, we immunoprecipitated the β1 subunit with a monoclonal antibody and probed the immunoprecipitate with a GSH antibody using Western blotting (4). Because the Na⁺-K⁺-ATPase preparations contain GSH at a level that gives clear signals for detection of glutathionylation with this technique (2), additional GSH was not added. We used solutions of different compositions as indicated in the figure legends to stabilize Na⁺-K⁺-ATPase in E1ATP, E1Na₃, and E2 conformations according to the nomenclature of the Albers-Post scheme (14). The solutions stabilize the Na⁺-K⁺ pump molecule in the sense that there is a higher probability of it assuming E1ATP, E1Na₃, or E2 than other conformations, but fluctuations into other states are not eliminated. In experiments examining the effects of incubating pig Na⁺-K⁺-ATPase with recombinant human glutaredoxin 1 (hGrx1), we used Western blot techniques (4) to detect co-immunoprecipitation of the β1 subunit with hGrx1 and the α1 subunit.

Measurement of Na⁺-K⁺-ATPase Activity

The specific enzyme activity was measured using the method of Baginski et al. (15), and protein was determined by the method of Peterson (16). The optimum activity (V_max) was measured 1–4 min later at 23 °C in a test medium that contained 130 mm NaCl, 20 mm KCl, 4 mm MgCl₂, 3 mm Tris-ATP, 20 mm histidine (pH 7.4).

Trypsin Cleavage of Na⁺-K⁺-ATPase

Controlled proteolysis of pig kidney Na⁺-K⁺-ATPase was performed in solutions that stabilize E1Na₃ or E2 conformations. The preparations were incubated in 30 mm histidine and 0.5 mm GSH (E2) or 30 mm histidine, 100 mm NaCl, 4 mm MgCl₂, and 0.5 mm GSH (E1Na₃) for 5 min at 10 °C followed by a 10-min incubation with either 0.5 mm ONOO⁻ or water (controls). Proteolysis was initiated by the addition of 5 μg of trypsin to controls and 0.125 μg of trypsin to ONOO⁻-treated enzyme, and it was allowed to continue for 5, 10, 20, or 30 min at 24 °C. It was then terminated with SDS sample buffer containing 1% trichloroacetic acid to irreversibly inhibit trypsin. Forty μg of protein was loaded onto 8% SDS-PAGE, and the gel was stained with Coomassie Blue. Identification of cleavage products was performed after blotting onto a PVDF membrane by Edman degradation analysis (Alphalyse, Odense, Denmark).

Myocytes

Ventricular myocytes were isolated from male White New Zealand rabbits. Details of anesthesia, excision of the heart, and cell isolation techniques have been described previously (17). Myocytes were used on the day of isolation only and stored at room temperature in Krebs-Henseleit buffer solution.

Glutathionylation of Myocyte β1 Na⁺-K⁺ Pump Subunit

Myocytes were incubated in modified Tyrode's solution. We varied the extracellular K⁺ concentration from 0 to 14 mm with the aim of shifting the conformational poise of Na⁺-K⁺ pumps. The solution contained 140 mm NaCl, 0–14 mm KCl, 1 mm MgCl₂, 10 mm glucose, 0.44 mm NaH₂PO₄, and 10 mm HEPES. Mg²⁺- and glucose-free solutions used when indicated were buffered with 10 mm Na₂HPO₄ and 1.4 mm NaH₂PO_4. With another approach, we aimed to modify the conformational poise by exposing myocytes to the Na⁺ ionophore monensin to enhance Na⁺ influx and increase its intracellular concentration (18). We also exposed myocytes to ouabain that binds to the Na⁺-K⁺ pump in E2 and E2P conformations (19).

Glutathionylation of β1 Na⁺-K⁺ pump subunits was induced from base line by exposing myocytes to the chemical oxidant ONOO⁻ or to angiotensin II (Ang II). The latter induces receptor-coupled oxidative stress via activation of NADPH oxidase (20). Biotin-GSH and anti-GSH antibody techniques for detection of β1 subunit glutathionylation (4) and co-immunoprecipitation of α1 and β1 subunits in myocyte lysate were performed as described (2, 4).

Measurement of Electrogenic Na⁺-K⁺ Pump Current (I_p)

Myocytes were initially superfused with a modified Tyrode's solution that was similar to those used for studies on glutathionylation except that it contained 2.16 mm CaCl₂ and 5.6 mm KCl. We switched to a solution that was nominally Ca²⁺-free and contained 2 mm Ba²⁺ and 0.2 mm Cd²⁺ when the whole cell configuration was established. This solution contained 100 nm Ang II when indicated. When stable holding currents were assured after 6–8 min, we switched to a solution that was identical except that it contained 100 μm ouabain to identify I_p at a holding potential of −40 mV. Patch pipettes used for measurement of I_p were wide-tipped (4–5 μm) and filled with solutions containing: 5 mm HEPES, 2 mm MgATP, 5 mm ethylene glycol-bis(β-aminoethylether)-N,N,N′,N′-tetraacetic acid, 70 mm potassium glutamate, 10 μm l-arginine, 10 or 80 mm sodium glutamate, and 80 or 10 mm tetramethylammonium chloride. Recombinant FXYD1 protein was included in pipette solutions as described previously (4) when indicated. To eliminate a theoretical risk of FXYD1 induced Na⁺ influx along its electrochemical gradient and secondary stimulation of I_p (4), I_p was measured in Na⁺-free superfusates when the pipette Na⁺ concentration was 10 mm. We used Na⁺-containing superfusates when pipette solutions contained 80 mm Na⁺ because the electrochemical gradient eliminates any possible Na⁺ influx at this concentration and because the experiments are technically easier to perform in Na⁺-containing superfusates, reducing use of a our limited supply of recombinant FXYD1 protein. For measurement of effects of extracellular K⁺ concentrations on the membrane potential, pipette solutions contained 10 mm NaCl and 140 mm KCl. Details of protocols and equipment used to define and measure I_p have been described (3, 21).

Statistical Analysis

The results are expressed as the means ± S.E. One-way analysis of variance was used for analysis of co-immunoprecipitation data and Student's t test for paired data between experiments and controls. Student's t test for unpaired data were used for comparison of mean levels of I_p. Inhibition of ATP-dependent Na⁺-K⁺-ATPase activity was compared with Wilcoxon matched signed ranks test. p < 0.05 is regarded as significant in all comparisons.

RESULTS

E1/E2 Conformational States and β1 Subunit Glutathionylation in Pig Kidney Na⁺-K⁺-ATPase

The β1 subunit of pig kidney Na⁺-K⁺-ATPase is glutathionylated under base-line conditions in the absence of experimentally induced oxidative stress (2). Large signals for glutathionylation were detected with the biotin-GSH technique when Na⁺-K⁺-ATPase-enriched membrane fragments were suspended in a solution that stabilizes the E1ATP conformation, whereas the signal was much smaller in a solution that stabilizes the E2 conformation (Fig. 1A). Results were similar for glutathionylation detected with the GSH antibody technique (Fig. 1B). There was no difference in glutathionylation between the E2 and K-bound E2 conformations (Fig. 1B). The signal for glutathionylation was stronger in the E1Na₃ than in the E2 conformation (Fig. 1C). To examine whether glutathionylation is also dependent on conformation under conditions of oxidative stress, we incubated Na⁺-K⁺-ATPase with ONOO⁻ in a nominal concentration of 0.5 mm for 15 min. Glutathionylation remained dependent on E1ATP/E2 conformations (Fig. 1D).

FIGURE 1.

Conformation dependence of β1 subunit glutathionylation of pig kidney Na⁺-K⁺-ATPase. A, glutathionylation after incubation of Na⁺-K⁺-ATPase-enriched membrane fragments in solution containing 0.1 mm EDTA, 25 mm histidine, and 2 mm Na₂ATP at a pH of 7.1 (E1ATP) or in solution containing 0.1 mm EDTA and 25 mm histidine at a pH of 7.1 (E2). Glutathionylation was measured with the biotin-GSH technique. B, glutathionylation in E1ATP and E2 conformations with and without 10 mm KCl in 30 mm histidine solution. Glutathionylation was measured with the GSH antibody technique. C, glutathionylation after incubation of Na⁺-K⁺-ATPase in 100 mm NaCl, 4 mm MgCl₂, and 20 mm histidine at a pH of 7.35 (E1Na₃). The solution used to promote the E2 conformation contained 20 mm histidine at a pH of 7.35 in these experiments. Glutathionylation was measured with the GSH antibody technique. D, glutathionylation induced by ONOO⁻ (ONOO) in E1ATP and E2 conformations. The solution used to promote the E2 conformation contained 0.1 mm EDTA and 25 mm histidine at a pH of 7.1 in these experiments. Glutathionylation was measured with the biotin-GSH technique. E, effect of pre-exposure to ouabain before exposure to ONOO⁻ on glutathionylation in the E2 conformation (0.1 mm EDTA, 25 mm histidine, pH of 7.1). F, glutathionylation with pre-exposure to ONOO⁻ before exposure to ouabain. The histograms show mean densitometry ± S.E. of five experiments. *, p < 0.05. IB, immunoblot; IP, immunoprecipitation.

Na⁺-K⁺-ATPase-enriched membrane fragments were suspended in solution that stabilizes the E2 conformation and then exposed to 100 μm ouabain for 5 min. They were subsequently exposed to ONOO⁻ for an additional 5 min before biotin-GSH was rapidly added. Glutathionylation was measured 15 min later. Ouabain caused a decrease in β1 subunit glutathionylation when membrane fragments were exposed to ONOO⁻ (Fig. 1E). There was a similar decrease when the timing of exposure to ONOO⁻ and ouabain was reversed (Fig. 1F).

Oxidative Stress and Proteolysis of Na⁺-K⁺-ATPase in E1Na₃ and E2 Conformational States

We examined whether glutathionylation causes changes in the trypsin digestion pattern of pig kidney Na⁺-K⁺-ATPase-enriched membrane fragments. Na⁺-K⁺-ATPase was stabilized in E2 or in E1Na₃ conformations. Exposure to ONOO⁻ dramatically increased the susceptibility of the enzymes to trypsin cleavage (Fig. 2A). There were several major conformation-dependent cleavage products in control enzymes in E1 as well as in E2 conformations. The trypsin digestion pattern of enzyme stabilized in the E2 conformation was characterized by a doublet of bands migrating at 57 and 48 kDa and by a triplet of bands migrating at molecular masses at ∼38 kDa. These bands correspond to cleavage products following cleavage at the E2-specific trypsin cleavage site 1 after Arg-438 in pig kidney enzyme (Arg-445 in shark) according to Jørgensen and Collins (22). In the E1 conformation-specific cleavage, products migrated at 71, 60, and 28 kDa. In contrast, a much more complete digestion was observed after ONOO⁻ treatment where hardly any cleavage products were left when the enzymes in E1 or E2 conformations, even at the 40 times lower trypsin concentration. Reducing the duration of exposure to trypsin from 20 to 5 min resulted in a similar comprehensive cleavage for the ONOO⁻-treated enzyme emphasizing the large increase in trypsin sensitivity after ONOO⁻ treatment. One band, running at ∼68 kDa was preferentially seen in ONOO⁻ treated enzymes in the E2 conformation. Edman degradation analysis of this band indicates the sequence IATLASGLEGG that can be identified on a small helix (3′) connecting the A domain with M3 (7). This locates the cleavage site between Arg-262 and Ile-263 in the pig kidney sequence and corresponding to Arg-269 and Ile-270 in the shark sequence (Fig. 2B) (7). This is the trypsin cleavage site specific for the E1 form as previously demonstrated (22), indicating that ONOO⁻ stabilizes an E1 conformation. A similar and even more pronounced 68-kDa band was present after ONOO⁻ treatment of the shark enzyme stabilized in E2 (not shown). Treatment with ONOO⁻ also caused some cleavage of the β subunit migrating as a smear at ∼55 kDa. Without exposure to ONOO⁻, the E1Na₃ conformation appears to be the more trypsin-sensitive form as judged from the remaining volume of the undigested α band (Fig. 2A, first and third lanes). The relative intense bands of undigested enzyme in the presence of ONOO⁻ are due to the presence of a 40 times lower trypsin concentration in these preparations. The trypsin digestion pattern of enzyme stabilized in the E2 and K-bound E2 conformations were similar. The intensity of the E1-specific 68-kDa band increases slightly with time both without and with K⁺ (Fig. 2C).

FIGURE 2.

A, SDS-PAGE of trypsin-treated pig Na⁺-K⁺-ATPase-enriched membrane fragments. The E2 conformation was stabilized by incubation in 30 mm histidine at a pH of 7.4 alone. The E1Na₃ conformation (E1) was stabilized by the further addition of 4 mm MgCl₂ and 100 mm NaCl. The Na⁺-K⁺-ATPase was exposed to 0.5 mm ONOO⁻ and 0.5 mm GSH before exposure to trypsin for 20 min as indicated and compared with controls. Controls received 5 μg of trypsin/100 μg of enzyme, whereas ONOO⁻/GSH-treated enzyme received only 1 μg of trypsin/40 μg of enzyme (0.125 μg). The positions of the α and β subunits are indicated on the left side, and the molecular mass of major cleavage products is indicated (kDa). B, the shark Na-K-ATPase cytoplasmic headpiece (Protein Data Bank 2ZXE) with E1- and E2-specific trypsin cleavage sites indicated (red ×). The cleavage site in the E1 form is on the 3′ helix after Arg-269 (Arg-262 in pig) in the A domain and in the E2-form after Arg-445 (Arg-438 in pig) on a short helix in the N domain. The salt bridge between Glu-223 in the A-domain and Arg-551 in the N-domain, which is probably broken by ATP binding in connection with E2 to E1 transition, is indicated by the stippled orange line (7). C, SDS-PAGE of trypsin-treated pig Na⁺-K⁺-ATPase-enriched membrane fragments. The E2 and K-bound E2 conformations were stabilized by incubation in 30 mm histidine with and without 10 mm KCl at a pH of 7.4. Trypsin treatment was initiated with 0.125 μg of trypsin/100 of μg enzyme and incubated for 0 (H₂O) to 30 min before gel electrophoresis.

Conformational States and Oxidant-induced Inhibition of Na⁺-K⁺-ATPase

It is impossible from the trypsin digestion patterns (Fig. 2A) to decide whether there is any conformation-dependent change in trypsin sensitivity to ONOO⁻ because of the very fast cleavage. However, if trypsin exerts most of its effect on the Na⁺-K⁺ pump molecular complex nominally stabilized in the E2 conformation after ONOO⁻ has shifted the poise toward an E1-like conformation (2), the similar cleavage patterns after exposure to ONOO⁻ in E1Na₃ and E2 conformations (Fig. 2A) are to be expected because proteolysis is irreversible. We examined whether a difference in sensitivity to ONOO⁻ of Na⁺-K⁺-ATPase stabilized in E2 or in E1Na₃ conformations is evident in a functional assay. We used preparations from pig kidney and pig heart. They were exposed to 0.5 mm ONOO⁻ for 5 min at 10 °C before 0.5 mm GSH was added, and incubation continued for further 5 min. The hydrolytic activity was measured at increasing concentrations of Tris-ATP (Fig. 3). V_max and K_0.5 for ATP activation were derived for the fitting of a simple hyperbola to the data. The values for V_max and K_0.5 for ATP activation are shown in Table 1.

FIGURE 3.

Conformation dependence of oxidation-induced decrease in Na⁺-K⁺-ATPase activity. Na⁺-K⁺-ATPase-enriched membrane fragments were incubated in solutions containing 20 mm histidine at a pH of 7.35 to stabilize an E2 conformation or in 100 mm NaCl, 4 mm MgCl₂, and 20 mm histidine at a pH of 7.35 to stabilize an E1Na₃ conformation during exposure to ONOO⁻/GSH before measurement of activity. Preparations isolated from pig kidney (A) or pig hearts (B) were used. Means of triplicate determinations were fitted with a simple hyperbola. Standard errors are contained within the symbols. Comparison of data pooled for all ATP concentrations indicated that ONOO⁻ induced a significantly greater inhibition when either enzyme had been exposed to ONOO⁻ in the E1Na₃ than in the E2 conformation and that the kidney preparation was more susceptible to inhibition than the heart preparation.

TABLE 1.

Oxidation-induced inhibition of pig kidney and heart Na⁺-K⁺-ATPase

	Control		ONOO−/GSHa
	Vmax	K0.5	Vmax	K0.5
	μmol·mg−1·h−1	μm	μmol·mg−1·h−1	μm
Kidney
E2	352 ± 10	629 ± 45	228 ± 3	493 ± 19
E1Na3	331 ± 6	791 ± 34	125 ± 2	359 ± 14
Heart
E2	19.8 ± 0.3	471 ± 17	16.0 ± 0.3	384 ± 18
E1Na3	17.4 ± 0.2	472 ± 17	9.4 ± 0.1	259 ± 9

^a ONOO⁻/GSH values of V_max and K_0.5 are all significantly different from controls (p < 0.05).

The inhibition under V_max conditions after exposure of Na⁺-K⁺-ATPase to ONOO⁻/GSH in the E1Na₃ conformation was significantly greater than inhibition after exposure in the E₂ conformation in both preparations. When comparing activity at all ATP concentrations, the exposure induced a greater inhibition of kidney than heart enzyme. Generally, ONOO⁻ increased the apparent ATP affinity. However, in kidney enzyme the reduction in K_0.5 for ATP activation was less for both main conformations than the decrease in V_max, as also found previously (2). In contrast, for heart enzyme, the decrease in K_0.5 was exactly paralleled by the decrease in V_max, both in E1Na₃ and in E2. This could indicate that a different mechanism for the effect of ONOO⁻ on the apparent ATP affinity exists in the two enzyme preparations. β1 is the principal subunit in the kidney (23), whereas β1 as well as β2 subunits are expressed in the heart (24). The lack of susceptibility of the latter to glutathionylation (2) may confer a degree of resistance of the heart preparation to oxidative modification. The effect of FXYD1, predominantly expressed in the heart, but not FXYD2 expressed in the kidney, to mediate reversal of β1 subunit glutathionylation (4) may also contribute to the differences. Treatment of Na⁺-K⁺-ATPase with ONOO⁻ did not change the high affinity ATP activation estimated by measuring the ATPase activity in the absence of K⁺ (Na-ATPase activity; data not shown), suggesting that the effect of ONOO⁻ on low affinity ATP activation is due to a shift in the conformational poise toward the E1 conformation rather than a change in the ATP binding affinity itself.

Glutaredoxin1, β1 Subunit Glutathionylation, and α1/β1 Subunit Co-immunoprecipitation in Na⁺-K⁺-ATPase

We determined α1/β1 subunit co-immunoprecipitation of pig kidney Na⁺-K⁺-ATPase-enriched membrane fragments suspended for 15 min in solutions that stabilize E1ATP or E2 conformational states. The solutions contained 1 μm recombinant hGrx1 to catalyze deglutathionylation of the β1 subunit, or they were hGrx1-free. The exogenous hGrx1 co-immunoprecipitated with the β1 subunit (Fig. 4A) as does endogenously expressed Grx1 in rabbit cardiac myocyte lysate (4). Co-immunoprecipitation of hGrx1 and the β1 subunit was detected for Na⁺-K⁺-ATPase suspended in solutions stabilizing E1ATP or E2 conformational states, but the signal for co-immunoprecipitation was greater for the E1ATP than the E2 conformation (Fig. 4B). As also shown in Fig. 1, glutathionylation of the β1 subunit was greater in the E1ATP than the E2 conformation. Glutathionylation in both conformations was reduced by hGrx1 (Fig. 4C), supporting the functional significance of the co-immunoprecipitation of exogenous hGrx1 and β1 subunit. The pattern of α1/β1 subunit co-immunoprecipitation reflected the hGrx1-dependent β1 subunit glutathionylation in an inverse manner (Fig. 4, compare C and D).

FIGURE 4.

Effect of glutaredoxin1 on Na⁺-K⁺-ATPase β1 subunit glutathionylation. A, immunoblots of α1 subunit, GSH, and hGrx1 on β1 subunit immunoprecipitate of pig kidney Na⁺-K⁺-ATPase. Na⁺-K⁺-ATPase-enriched membrane fragments from kidney were incubated with 1 μm hGrx1 for 15 min in 0.1 mm EDTA, 25 mm histidine, and 2 mm ATP at a pH of 7.1 (E1ATP) or 0.1 mm EDTA and 25 mm histidine at a pH of 7.1 (E2) B, histogram showing co-immunoprecipitation of hGrx1 with β₁ subunit. C, effect of hGrx1 on β1 subunit glutathionylation. D, effect of hGrx1 on α1 and β1 subunit co-immunoprecipitation. The mean densitometries ± S.E. of five experiments are shown. *, p < 0.05. IB, immunoblot; IP, immunoprecipitation.

Extracellular K⁺ and β1 Subunit Glutathionylation in Cardiac Myocytes

We measured the membrane potential, a determinant of conformational poise, in rabbit myocytes suspended in solutions containing 0, 4.3, 7, or 14 mm K⁺ (Fig. 5A). The measured potentials were in good agreement with their known K⁺ dependence. At external K⁺ concentrations of more than ∼2 mm, they are mainly determined by the K⁺ permeability through K⁺ channels (25), and as the external K⁺ concentration is increased from 4.3 to 14 mm with little change in internal K⁺ concentration of ∼140 mm, the membrane depolarizes because of a reduction in the electrochemical driving force for K⁺ diffusion. However, K⁺ channels are activated by extracellular K⁺ (26), and the membrane depolarizes in K⁺-free solutions when the K⁺ channels are closed. Glutathionylation of the β₁ subunit measured with the biotin-GSH technique was dependent on the extracellular K⁺ concentration (Fig. 5B). We also exposed myocytes to 0–14 mm extracellular K⁺ in the presence of 2 mm Ba²⁺. Ba²⁺ blocks the inwardly rectifying K⁺ channels and induces membrane depolarization and spontaneous action potentials in isolated ventricular myocytes (27). It is therefore expected to eliminate voltage-dependent effects of extracellular K⁺ on the poise. Glutathionylation of the β1 subunit was independent of extracellular K⁺ in Ba²⁺-containing solutions (Fig. 5C).

FIGURE 5.

Dependence of β1 subunit glutathionylation in cardiac myocytes on the extracellular K⁺ concentration. A, membrane potentials of myocytes at 0–14 mm extracellular K⁺. B, Na⁺-K⁺ pump β1 subunit glutathionylation in lysate of myocytes exposed to 0–14 mm K⁺ (Biotin-GSH technique). There was a statistically significant decrease in glutathionylation with an increase in K⁺ concentration beyond 4.3 mm. C, K⁺ dependence of glutathionylation with exposure of myocytes to Ba²⁺ (Biotin-GSH technique). D, K⁺ dependence of glutathionylation in Mg²⁺-free solutions using the biotin-GSH technique. E, K⁺ dependence of glutathionylation in Mg²⁺-free solution using the GSH antibody technique. F, K⁺ dependence of glutathionylation in Mg²⁺-free solution with exposure of myocytes to Ba²⁺. The histograms show the mean membrane potentials of myocytes or mean densitometry of myocyte lysate ± S.E. of five experiments. *, p < 0.05. The K⁺-dependent increases in glutathionylation in D and E were statistically significant. IB, immunoblot; IP, immunoprecipitation.

The increase in glutathionylation with hyperpolarization (Fig. 5, A and B) is consistent with stabilization of an E1-like conformation because the inside negative potential would inhibit release of Na⁺ extracellularly from E2P, driving the reaction backwards, probably to E1P. Because ATP and Mg²⁺ are necessary for phosphorylation (28), we examined whether eliminating Mg⁺ from extracellular solutions affected the K⁺-dependent effect on β1 subunit glutathionylation. Myocytes were incubated for 30 min in Mg²⁺-free solutions that had external K⁺ concentrations from 0 to 14 mm. The solutions were also glucose-free to reduce cellular ATP levels. Although ATP and Mg²⁺ cannot be totally eliminated from the cytoplasm, a reduction in their concentrations might reduce the rate of phosphorylation and enzyme cycling, resulting in an accumulation of the enzyme in the unphosphorylated E1Na₃ state. Increasing the K⁺ concentration increased β1 subunit glutathionylation, as detected by the biotin-GSH technique (Fig. 5D). Similar results were obtained when we measured glutathionylation with the GSH antibody technique (Fig. 5E). Ba²⁺ also eliminated the dependence of β1 subunit glutathionylation on the extracellular K⁺ concentration in Mg²⁺-free solutions (Fig. 5F).

Oxidation-induced β1 Subunit Glutathionylation in Myocytes

We examined whether ouabain decreases β1 subunit glutathionylation induced by ONOO⁻ in cardiac myocytes as it does in Na⁺-K⁺-ATPase-enriched membrane fragments. Myocytes were loaded with biotin-GSH in Tyrode's solution that included 4.3 mm K⁺. The solution was ouabain-free or it included 100 μm ouabain. The myocytes were exposed to ONOO⁻ in the nominal concentration of 0.25 mm after 15 min, or they remained suspended in ONOO⁻-free solutions. They were lysed after an additional 15 min, and glutathionylation of the β1 subunit was determined. Ouabain decreased glutathionylation of the β1 subunit of myocytes exposed to solutions free of ONOO⁻ throughout the 30-min period and abolished an ONOO⁻-induced increase (Fig. 6A).

FIGURE 6.

β1 subunit glutathionylation in myocytes exposed to oxidative stress. A, glutathionylation measured in lysate of myocytes preincubated with ouabain and then ONOO⁻ (ONOO) before lysis. Glutathionylation was measured with the biotin-GSH technique. B, glutathionylation in lysate of myocytes preincubated with 5 μm monensin for 15 min before exposure to Ang II for an additional 15 min. The histograms show mean densitometry ± S.E. of five experiments. *, p < 0.05. IB, immunoblot.

We incubated myocytes for 30 min in Tyrodes's solution containing monensin. They were exposed to Ang II for 15 min to induce receptor-coupled oxidative stress (20) or to control solutions. Ang II increased β1 subunit glutathionylation detected in lysates from myocytes that had been incubated in control solutions but not in lysates from myocytes that had been preincubated in solutions containing 5 μm monensin for 15 min (Fig. 6B).

Extracellular K⁺, β1 Subunit Glutathionylation, and α1/β1 Subunit Co-immunoprecipitation

Because glutathionylation of the β1 subunit induced by oxidant stress decreases its co-immunoprecipitation with the α1 subunit (2), we examined whether the effect of extracellular K⁺ on β1 subunit glutathionylation (Fig. 5) affects α1/β1 subunit co-immunoprecipitation. We measured β1 subunit glutathionylation with the GSH antibody technique in these experiments with solutions that contained Mg²⁺ or that were Mg²⁺-free. The effects of extracellular K⁺ on glutathionylation reproduced results shown in Fig. 5, and changes in glutathionylation were associated with inverse changes in α1/β1 subunit co-immunoprecipitation in Mg²⁺-containing (Fig. 7, compare A and B) and Mg²⁺-free solutions (Fig. 7, compare C and D) as was the case for Na⁺-K⁺-ATPase exposed to hGrx1 (Fig. 4).

FIGURE 7.

Dependence of β1 subunit glutathionylation and α1/β1 subunit co-immunoprecipitation on extracellular K⁺ concentration. A, β1 subunit glutathionylation measured with GSH antibody technique. B, co-immunoprecipitation of α1/β1 subunits in lysate from the batches of myocytes used for data in A. C, dependence of β1 subunit glutathionylation on extracellular K⁺ concentration in Mg²⁺-free solutions. D, co-immunoprecipitation of α1/β1 subunits in lysate from the batches of myocytes used for data in C. The histograms show mean densitometry ± S.E. of five experiments. There was a statistically significant inverse correlation between β1 subunit glutathionylation and α1/β1 subunit co-immunoprecipitation shown in A versus B and C versus D. IB, immunoblot; IP, immunoprecipitation.

Dependence of Ang II-induced Inhibition of I_p on Transmembrane Na⁺ and K⁺ Gradients

Ang II inhibited I_p of ventricular myocytes voltage-clamped with patch pipettes that included 10 mm Na⁺ and 70 mm K⁺ in the filling solutions that perfuse the intracellular compartment, consistent with Ang II-induced glutathionylation of the β1 Na⁺-K⁺ pump subunit (20). Because Fig. 6B suggests that Ang II does not cause glutathionylation when the intracellular Na⁺ concentration is high, we examined whether Ang II reduces I_p when the Na⁺ concentration in the patch pipette solution is 80 mm. The solution also included 70 mm K⁺. The experimental protocol and an example of a membrane current trace used in the measurements of I_p are shown in Fig. 8A. The previously published results using 10 mm Na⁺ and 70 mm K⁺ in pipette solutions are shown in Fig. 8B. In contrast to the previous results, Ang II had no effect on I_p when the Na⁺ concentration in the pipette solutions was increased to 80 mm (Fig. 8C). Treatment of rabbits with an angiotensin-converting enzyme inhibitor or an Ang II receptor antagonist to disrupt in vivo Ang II signaling causes an increase in I_p of voltage-clamped myocytes measured with pipette solutions that include 10 mm Na⁺ when the solutions also include K⁺ but not when they are K⁺-free (29). We examined the effect of in vitro exposure to Ang II when pipette solutions were K⁺-free. The solutions included 10 mm Na⁺. Ang II had no effect on I_p measured using the K⁺-free pipette solutions (Fig. 8C).

FIGURE 8.

Dependence of Ang II-induced inhibition of Na⁺-K⁺ pump current (I_p) on intracellular Na⁺ and K⁺. A, experimental protocol and example of holding currents. The patch pipette solution perfusing the intracellular compartment included 80 mm Na⁺ and 70 mm K⁺, and the myocyte was exposed to Ang II in the example shown. The vertical arrow indicates when the whole cell configuration was established. Ca²⁺ subsequently remained included in the superfusate for ∼3 min before switching to a Ba²⁺- and Cd²⁺-containing superfusate that was Ca²⁺-free. I_p was identified by the inward shift in current with exposure to ouabain. This decrease was normalized for the membrane capacitance (in pF) of each myocyte. B, Ang II-induced decrease in two independent sets of experiments performed in Na⁺ containing and Na⁺-free superfusates (previously published data). Patch pipette solutions contained 10 mm Na⁺ and 70 mm K⁺ in both sets of experiments. *, p < 0.05. C, effect of Ang II when pipette solutions were K⁺-free or contained a high Na⁺ concentration to shift the conformational poise toward E₂ state. D, effect on I_p of FXYD1 included in pipette solutions that also contained Na⁺ in a concentration near physiological intracellular levels. The FXYD1-induced increase in I_p is consistent with its effect to mediate deglutathionylation of the β1 subunit. E, effect of FXYD1 when the pipette Na⁺ concentration was high. The absence of any effect of FXYD1 to increase I_p is consistent with an already low level of β1 subunit glutathionylation with high intracellular Na⁺ levels implied by results in Fig. 6B.

The association of FXYD proteins with the Na⁺-K⁺ pump is important for the functional stability of α1/β1 subunit interaction (30). FXYD1 promotes reversal of β1 subunit glutathionylation and, when included in patch pipette solutions, abolishes oxidation-induced inhibition of I_p in cardiac myocytes (4). When included in patch pipette solutions containing 10 mm Na⁺, FXYD1 caused an increase in I_p (Fig. 8D), although it had no effect when patch pipette solutions contained 80 mm Na⁺ (Fig. 8E). These results are consistent with the effect of a monensin-induced increase in the intracellular Na⁺ concentration, resulting in a decrease β1 subunit glutathionylation from base line (Fig. 6B).

DISCUSSION

Glutathionylation of the β1 subunit of Na⁺-K⁺-ATPase stabilized in E1 conformations was easily detectable under base-line conditions and was further increased by exposure to ONOO⁻. Signals for glutathionylation were much weaker after incubation in solutions stabilizing the E2 conformation. The functional significance of such conformation-dependent susceptibility to glutathionylation was indicated by greater ONOO⁻-induced inhibition of activity of Na⁺-K⁺-ATPase stabilized in E1Na₃- than E2 conformations. Ouabain reduced the level of β1 subunit glutathionylation during oxidative stress. It binds to the α subunit in E2 and E2P conformations by interacting with transmembrane helices M1–M4. A change in the tryptic digest with ouabain binding suggests a structural change in the Na⁺-K⁺ pump complex (31). This might reduce access of GSH to Cys-46 in the β1 subunit. However, there is no evidence for such a ouabain-induced change in the β subunit in the crystal structure (32). An alternative explanation for the decrease in glutathionylation is that ouabain further stabilizes the E2 conformation compared with the 25 mm histidine, 0.1 mm EDTA solution otherwise used. By reducing fluctuations from E2 into conformations that allow access of GSH to Cys-46, ouabain might have caused the decrease in glutathionylation that we observed.

The crystal structure of Na⁺-K⁺-ATPase shows that the transmembrane domain of the β1 subunit is somewhat detached from α subunit domains and nearly parallel to αM7. Cys-46 of the β1 subunit is on the opposite side of its helix with its side chain facing the lipid bulk phase of the membrane (7). If structural changes in Na⁺-K⁺-ATPase with the E1Na⁺₃ → E2P transition are similar to those in the closely related Ca²⁺-ATPase, large scale changes in the position of Cys-46 may occur. Ca²⁺-ATPase does not have a β subunit that allows a firm prediction for Na⁺-K⁺-ATPase. However, it can be anticipated that changes in the coordination of the Na⁺ ion associated with the transition will be transmitted to the phosphorylation domain of α, αM3, and the C terminus. Because of an extensive hydrogen-bonding network to the β subunit, its position is also expected to change (7), consistent with results of early proteolytic studies on Na⁺-K⁺-ATPase (33).

Peroxynitrite shifts the poise toward E1-like conformations as we reported previously (2) and as also suggested by the effect of ONOO⁻ on low affinity ATP activation in this study. This may be due to the GSH adduct to Cys-46 disrupting interaction of the β1 subunit with αM7, reminiscent of the effect of the β1-(Y39W,Y43W) mutation to disrupt interaction with αM7 and shift the poise toward E1P (34). This may effectively stabilize the enzyme in E1-like conformations regardless of the starting conditions because the side chain of Cys-46 seems unlikely to return to a location in the lipid membrane bulk phase in an E2-like conformation if it has acquired a 305-Da charged adduct with exposure to ONOO⁻. A ONOO⁻-induced shift toward an E1-like conformation is in agreement with the trypsin cleavage patterns that we observed. Na⁺-K⁺-ATPase was more trypsin-sensitive when stabilized in the E1Na₃ than in the E2 conformation but even more trypsin sensitive in either of the two conformations after exposure to ONOO⁻. The digestion of the β subunit after exposure to ONOO⁻ suggested that the subunit is accessible to trypsin in an E1-like conformation once glutathionylation has occurred, and because proteolysis is irreversible, the digestion measured experimentally becomes complete regardless of starting conditions. The large increase in trypsin sensitivity induced by ONOO⁻ (Fig. 2A) indicates a decrease in the overall structural stability of the enzyme. This may in part explain the observed decrease in αβ co-immunoprecipitation after ONOO⁻ treatment at least in the presence of detergent (2).

Recombinant hGrx1 co-immunoprecipitated with the β1 subunit of Na⁺-K⁺-ATPase stabilized in the E1ATP conformation and induced a large decrease in glutathionylation, indicating that it had access to Cys-46 of the subunit. It also decreased glutathionylation when Na⁺-K⁺-ATPase was nominally stabilized in the E2 conformation. However, if the location of Cys-46 in the E2 conformation is similar to its known location in a state analogous to E2·2K⁺·P_i (6, 7), the ∼11-kDa hydrophilic Grx1 is not expected to have access to it. Fluctuations from E2 into other conformations would be much more likely to account for the access. The exact conformational state(s) in which glutathionylation of the β1 Na⁺-K⁺-ATPase subunit occurs cannot be determined at present. However, taken together, our data indicate that conformational state is a critical determinant and can account for glutathionylation that would be counterintuitive in light of the currently known crystal structure.

In addition to access of the hydrophilic GSH to Cys-46, it is a requirement for glutathionylation that the sulfhydryl group of the cysteine has a pK_a that allows deprotonation at physiological pH, typically a consequence of proximity to basic residues in the amino acid sequence of the protein (1). However, there are no neighboring basic amino acids for Cys-46 in the primary sequence of the β1 subunit. In the three-dimensional structure, however, a cluster of four arginines and one lysine near the C terminus of M10 is located ∼15 Å from the Cys-46 side chain (7). This cluster has been suggested to act as a voltage sensor that moves in response to changes in the electrical field with electrogenic binding of Na⁺. Movement of cluster and Cys-46 toward each other may lower the pK_a of the sulfhydryl group.

Apart from such spatial movements to proximity to basic residues in the three-dimensional structure of a macromolecular complex (35), cooperativity with other cysteines, hydrogen bonding to serine or histidine residues, the presence of an adjacent proline, and coordination to a divalent cation (30) are feasible mechanisms. Determining the crystal structure of the Na⁺-K⁺ pump complex in E1 conformations will be critical for understanding the structural features that facilitate the glutathionylation.

Specific Na⁺-K⁺ pump conformational states in intact cells cannot be easily imposed experimentally, and we used a variety of complementary approaches to examine whether the conformation dependence of β1 subunit glutathionylation status in Na⁺-K⁺-ATPase-enriched membrane fragments is also likely to occur for in situ Na⁺-K⁺ pumps in cardiac myocytes. Levels of glutathionylation measured with the biotin-GSH or GSH antibody techniques were higher after incubation of myocytes in solution containing 4.3 mm K⁺ than in K⁺-free solutions. This may be due to a shift toward E1 conformations that depends on extracellular K⁺, consistent with the shift that occurs when Xenopus oocytes in K⁺-free solutions are re-exposed to K⁺ (36). With an increase beyond 4.3 mm K⁺, depolarization reduces the voltage-dependent rebinding of Na⁺ at extracellular sites, and binding of K⁺ causes a shift of the poise toward E2 species. Consistent with this, there was a decrease in glutathionylation at high extracellular K⁺ concentrations.

If phosphorylation of E1Na₃ becomes rate-limiting in Mg²⁺-free solutions, a voltage-dependent accumulation of E1 conformations with an increase in the extracellular K⁺ concentration should be prevented. This would be consistent with the difference in glutathionylation between Mg²⁺-containing and Mg²⁺-free solutions shown in Fig. 5. The effect of Ba²⁺ to abolish the dependence of glutathionylation on the K⁺ concentration may be due to elimination of the K⁺ dependence of membrane voltage. However, an effect of Ba²⁺ to inhibit Mg²⁺-dependent phosphorylation of E1Na⁺₃ (37) may also contribute to an increase in E1 species and hence susceptibility of the β1 subunit to glutathionylation. Glutathionylation was decreased by exposure of myocytes to ouabain, consistent with the shift from E1 toward E2 conformations in Xenopus oocytes exposed to ouabain (36). Monensin also decreased glutathionylation, and it prevented an Ang II-induced increase. This is consistent with a high intracellular Na⁺ concentration favoring outward Na⁺ pumping and hence a shift from E1 toward the E2 conformations and reduced susceptibility of Cys-46 to glutathionylation.

Ang II decreased I_p when patch pipettes contained 10 mm Na⁺ and 70 mm K⁺ but not when they were K⁺-free. Modeling suggests that cytosolic K⁺ can inhibit partial reactions of the Na⁺-K⁺ pump in E1 conformations and limit the overall forward reaction rate (38), a conclusion that is consistent with effects of K⁺ in patch pipettes on I_p in voltage-clamped myocytes (39). Because under our conditions the forward E1 → E2P reaction is rate-limiting and cannot occur with K⁺ bound to transport sites in the E1 state, cytosolic K⁺ is expected to inhibit this reaction, causing an accumulation of the pump in the E1 state. This is consistent with the dependence of the Ang II-induced decrease in I_p on the K⁺ concentration in patch pipette solutions. However, with 80 mm rather than 10 mm Na⁺ in patch pipettes, K⁺ competes less effectively for binding, and a shift toward the E2 states may account for the absence of an effect of Ang II-induced oxidative stress on I_p we found with 80 mm Na⁺ and 70 mm K⁺ in patch pipettes.

The implication of glutathionylation stabilizing E1-like states is that a small initial shift toward these states may cause progressive accumulation of Na⁺-K⁺ pump molecules with glutathionylated β1 subunits. However, Grx1 that mediates deglutathionylation (1) co-immunoprecipitates with β1 subunits in myocytes lysate, and recombinant Grx1 included in patch pipette solutions abolishes oxidant induced Na⁺-K⁺ pump inhibition in voltage-clamped myocytes (2). The increased interaction of Grx1 with the β1 subunit in E1 compared with E2 states we now report for Na⁺-K⁺-ATPase-enriched membrane fragments may reflect a mechanism that prevents progressive amplification of E1-dependent β1 subunit glutathionylation and an associated excessive pump inhibition in the intact cell.

E1/E2 conformation-dependent β1 subunit glutathionylation adds complexity to interpretation of intrinsic properties of in situ Na⁺-K⁺ pumps in cells. Pump activation at different concentrations of intracellular and extracellular Na⁺ and K⁺, with or without control of membrane voltage, has been widely used to implicate in situ pump characteristics. These interventions may alter E1/E2 conformation-dependent β1 subunit glutathionylation and hence Na⁺-K⁺ pump activity, independent of intrinsic ligand affinities or voltage dependence that has been the focus of such studies. Misinterpretation that may occur can be exemplified by the effect we reported on I_p in myocytes isolated from rabbits with experimentally induced diabetes (40), a condition associated with increased oxidative stress. A diabetes-induced decrease in I_p was dependent on the K⁺ concentration in patch pipettes, and we implicated an increase in the rate of the backward E1 + K⁺ → E2(K⁺) reaction. We did not identify a mechanism for this, nor is one intuitively apparent. We now offer the relatively simple explanation that effects of the oxidative stress characteristic of diabetes were only detectable when intracellular K⁺ blocked binding of Na⁺ to E1ATP and effectively slowed down a rate-limiting E1 → E1P transition. This would increase the concentration of E1 pump species susceptible to glutathionylation. Difficulties with identifying functional Na⁺-K⁺ pump characteristics in cells independent of poise-dependent glutathionylation may account for characteristics that are often poorly compatible between the many published studies. Taking poise-dependent glutathionylation and its functional equivalent into account may produce more consistent results between future studies.