INTRODUCTION

Mammals closely control extracellular fluid (ECF)¹ and intracellular fluid (ICF) potassium within a very narrow range since the ratio of the ICF potassium (K⁺) to ECF K⁺ is the primary determinant of resting membrane potential. This is especially important in excitable tissues such as the heart, where disturbances in resting membrane potential can compromise cardiac contractility and become life threatening (1). ECF K⁺ is maintained by the interplay of two key organ systems: the kidneys control K⁺ excretion by actively reabsorbing or secreting K⁺ (1), while skeletal muscle, which contains the largest intracellular pool of K⁺ in the body (75% of total intracellular K⁺), adjusts ECF K⁺ by regulating K⁺ transport between the ICF and ECF compartments (2, 3).

During dietary K⁺ deprivation, K⁺ output can exceed input, leading to a fall in plasma potassium, termed hypokalemia (1). In response, skeletal muscle selectively loses ICF K⁺ into the ECF. This loss buffers the fall in ECF K⁺, minimizing the hyperpolarization of cell membranes due to the increased ratio of ICF K⁺ to ECF K⁺. There is a concomitant decrease in the number of active sodium pumps (Na,K-ATPase) in skeletal muscle plasma membranes, postulated to be the primary event leading to the transfer of ICF K⁺ into the ECF (4).

Na,K-ATPase (EC 3.6.1.37) is an intrinsic membrane-bound enzyme found in eukaryotic cells (5). For each molecule of ATP hydrolyzed, Na,K-ATPase transports two K⁺ into the cell and three Na⁺ out of the cell, generating electrical and chemical gradients of Na⁺ and K⁺ across the plasma membrane (7). Na,K-ATPase is a heterodimer, consisting of a catalytic  subunit (M_r  112,000) and a glycosylated  subunit (M_r  35,000). To date three isoforms of each subunit have been identified (5, 6): 1, 2, 3, 1, 2, and 3 (3 detected only in amphibians (5)). A putative fourth  subunit, 4, identified in rats and humans, appears isolated to the testis (8). Subunit isoforms expression is tissue-specific (6); skeletal muscles of adult rats express 1, 2, 1, and 2 mRNA and protein (9, 10). Hundal et al. (9) reported evidence of muscle type specificity for expression of : 1 predominating in slow oxidative fibers and 2 in fast glycolytic fibers.

Our previous studies provided evidence for isoform-specific regulation in response to hypokalemia: 2 protein and mRNA, but not 1 or 1, decreased in hindlimb skeletal muscle of hypokalemic rats (11). Since skeletal muscle is a heterogeneous mix of skeletal muscle tissue types that vary greatly in both metabolic and contractile characteristics (12), we aimed to investigate whether the response to hypokalemia was both isoform- and muscle type-specific.

EXPERIMENTAL PROCEDURES

To establish the relative levels of Na,K-ATPase isoform expression in a panel of distinct skeletal muscles, male Sprague-Dawley rats (300 g) were anesthetized with 0.2 ml of sodium pentobarbital/100 g of body weight. Muscles were removed, frozen in liquid nitrogen, and stored at 80 °C. To provoke hypokalemia, Sprague-Dawley rats, approximately 8 weeks of age, were placed on a K⁺-deficient diet (Harlan Teklad, TD 88239, Madison, WI) for 10 days and paired to a control group of rats fed a comparable diet with K⁺ restored (Harlan Teklad, TD 88238).

Skeletal muscle was weighed, minced into small pieces, and homogenized at an approximate 1:20 (w:v) ratio for 45-60 s using a Polytron homogenizer (Brinkmann Instruments) in a buffer containing 5% sorbitol, 25 mM histidine-imidiazole (pH 7.4), 0.5 mM Na₂EDTA, and proteolytic enzyme inhibitors: 0.5 mM phenylmethylsulfonyl fluoride, 1 µl/ml leupeptin, and 1 mM 4-aminobenzamidine dichloride. Protein concentration was determined by the method of Lowry et al. (13) after trichloroacetic acid precipitation.

A constant amount of homogenate protein (100 µg for  subunit analysis, 50 µg for  subunit analysis) was resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) as described previously (14). The gel was blotted electrophoretically onto nitrocellulose membranes or Immobilon-P membranes and then incubated overnight with one of the following antibodies: McK1 (1:200), a monoclonal specific for 1 (15), provided by K. Sweadner (Harvard Medical School); C464.6 (1:100), a monoclonal specific for 1 (16, 17), provided by M. Kashgarian (Yale Medical School); McB2, a monoclonal specific for 2 (15) provided by K. Sweadner (Harvard Medical School); anti-HERED (1:100), a polyclonal specific for 2, and anti-TED (1:200), a polyclonal specific for 3 (18), both provided by T. Pressley (Texas Technical University); anti-1 FP (1:500), a polyclonal against 1 raised in rabbit according to Shyjan and Levenson (19); anti-rat 2 (1:1,000), a polyclonal against 2 (Upstate Biotechnology, Lake Placid, NY); or SpETb2 (1:2,000), a polyclonal against human 2 (20), provided by P. Martin-Vasallo (Universidad de La Laguna, Spain). Blots probed with monoclonal antibodies were incubated for 2 h with rabbit anti-mouse IgG secondary antibodies (1:2,000). To ensure that changes in isoform abundance did not reflect loading artifacts a subset of blots was probed additionally with a mouse monoclonal specific for muscle calsequestrin, VIIID1₂ (21). All blots were processed as described previously (14), probed with ¹²⁵I-protein A, and visualized by autoradiography using Kodak XAR-5 film and DuPont Cronex Lightning Plus XL screens at 70 °C. 3 and calsequestrin blots were probed using the enhanced chemiluminescence (ECL) detection system (Amersham) using Kodak XAR-5 film. Multiple exposures of autoradiograms and assaying samples at multiple concentrations confirmed that signals were in the linear range, as established previously (22).

Immunoprecipitation of Complexes

Immunoprecipitation of nondenatured heterodimers was accomplished using techniques reported previously (23). Briefly, skeletal muscle homogenate (300 µg of protein at 3-10 µg/µl) was incubated for 1 h at 4 °C in a lysis buffer, at an approximate 1:3 (v:v) ratio, containing 50 mM Tris-buffered saline at pH 8.0, 1% Nonidet P-40, 1% bovine serum albumin, and proteolytic enzyme inhibitors: 0.2 units/ml aprotinin, and 1.0 mg/ml phenylmethylsulfonyl fluoride. The samples were centrifuged at 250 × g for 10 min, and the supernatant lysate was pretreated with mouse serum-agarose beads and nonspecific goat IgG-agarose beads at an approximate 1:5 bead to supernatant (v:v) ratio, at 4 °C for 1 h with gentle rocking. Samples were centrifuged at 2,500 × g for 30-45 s, the supernatant incubated with IEC 1/48 (24) obtained from A. Quaroni (Cornell University, Ithaca), at a approximate 1:5 antibody to supernatant (v:v) ratio, at 4 °C for 4 h with gentle rocking. This monoclonal antibody recognizes undenatured , thus it can be used to immunoprecipitate heterodimers. Goat anti-mouse IgG-agarose beads at 1:5 beads were added to supernatant and the samples incubated at 4 °C for 1 h with gentle rocking to isolate complexes. After washing, the precipitate was heated to 90 °C for 5 min to dissociate antibody-antigen complexes and  and subunits. Finally, the samples were resolved by SDS-PAGE, subjected to immunoblot analysis, probed with  and  isoform-specific antibodies, and visualized by the CDP-Star^TM chemiluminescence detection system (Tropix, Bedford, MA) for  subunits or ¹²⁵I-protein A for  subunits.

Plasma K⁺ and muscle K⁺ were measured by flame photometry. Specific muscles and cardiac ventricles were quickly dissected, flash frozen in liquid nitrogen, and stored until assayed. They were later thawed, blotted lightly to remove adherent ECF or blood, homogenized in approximately 1:50 (w:v) 0.3 M trichloroacetic acid for 1-2 min using a Tissuemizer (Tekmar, Cincinnati, OH), and centrifuged at 2,000 rpm for 20 min to remove cell debris. K⁺ and Na⁺ were measured using a Radiometer FLM 3 flame photometer (Radiometer, Copenhagen, Denmark), with lithium as internal standard (25).

Sugars were removed from  subunits with peptide N-glycosidase F (N-glycanase, Genzyme Corp., Cambridge, MA). Crude skeletal muscle homogenate (100 µg/sample) was centrifuged at 48,000 rpm for 60 min. The pelleted crude membranes were incubated with N-glycanase as described in Azuma et al. (22) to yield a final protein/enzyme ratio of 2.5 units of N-glycanase/100 µg of protein.

Autoradiograms were scanned and quantitated using the Bio-Rad GS670 Imaging Densitometer and software. All data are expressed as means ± S.E. Significance was assessed by a two-tailed Student's t test for unpaired samples, and differences were regarded as significant at p < 0.05.

Chemicals were reagent grade, spectroquality, or electrophoresis purity reagents. SDS-PAGE reagents were from Bio-Rad. Leupeptin, phenylmethylsulfonyl fluoride, 4-aminobenzamidine dichloride, SDS-PAGE molecular weight standards, and antibodies coupled to agarose beads were purchased from Sigma. Nitrocellulose transfer membrane was from Micron Separations. Immobilon-P transfer membrane was obtained from Millipore. ¹²⁵I-Protein A was from ICN. Rabbit anti-mouse secondary antibody was from Calbiochem. Anti-mouse and anti-rabbit alkaline phosphatases were from Tropix. Anti-calsequestrin antibody (IgG2b) was from Affinity Bioreagents.

RESULTS

Immunoblot analysis was used to determine the pattern of Na,K-ATPase  and  subunit expression in a panel of skeletal muscles: soleus, 87% slow oxidative fibers, with some fast glycolytic-oxidative fibers; red gastrocnemius (RG), a mixed muscle type, approximately 30% slow oxidative fibers, 62% fast glycolytic-oxidative, and 8% fast glycolytic; extensor digitorum longus (EDL), a classically fast muscle type, both fast glycolytic-oxidative (42%) and fast glycolytic (56%), with only 2% slow oxidative fibers; white gastrocnemius (WG), very fast glycolytic muscle (84%) with some fast oxidative fibers; and diaphragm, a mixed muscle type, approximately 40% slow oxidative, 27% fast glycolytic-oxidative, and 34% fast oxidative (26, 27, 28, 29).

Typical autoradiograms of rat skeletal muscle homogenates probed with isoform-specific antibodies against 1, 2, 1, and 2 are shown in Fig. 1. Whereas 1 and 2 are expressed in all skeletal muscles, 1 was very low in WG and 2 barely detected in soleus and diaphragm. The relative levels of expression of Na,K-ATPase subunits are summarized in Fig. 2. The 1, 2, and 1 subunit abundance values were normalized to abundance in soleus muscle, defined as 1. 2 was normalized to abundance in RG. 1 is 2-4-fold higher in the oxidative rich soleus and diaphragm than in glycolytic WG, EDL, and RG. 2 has much less variability than 1, is lowest in WG, and twice as high in diaphragm. 1 parallels the pattern of 1 expression: higher in both fast and slow oxidative rich fibers (soleus, diaphragm, and RG), and barely detected in WG. 2 expression, reciprocal to that of 1, is highest in muscles containing fast glycolytic fibers (WG, EDL, RG) and is below detection in oxidative dominated muscle (soleus and diaphragm). Consistent with previous observations that no 3 mRNA is expressed in mature rat skeletal muscle (6, 10), 3 protein was not detected in any individual skeletal muscle. These result suggest that 22 is the predominant heterodimer in fast glycolytic muscle (WG) and that both 11 and 21 heterodimers are expressed in muscles rich in oxidative fibers (soleus and diaphragm). Whether 21 and 22 are limited to oxidative fibers and fast glycolytic fibers, respectively, cannot be resolved in mixed muscle homogenates.

Detection of Na,K-ATPase  and  isoform abundance in skeletal muscles.

Relative levels of Na,K-ATPase  and  isoform abundance in skeletal muscles.

Characterization of Heterodimers in Soleus

The detection of 1, 2, and 1 but not 2 in soleus suggested that both  subunits form heterodimers with 1. To test this hypothesis an antibody to 1 was used to immunoprecipitate heterodimers. Immunoblot analysis of the immunoprecipitates was used to determine whether both 1 and 2 were precipitated with the anti-1 antiserum; samples of brain and kidney homogenate were assayed directly by immunoblot in parallel as positive controls. Typical autoradiograms of these immunoblots are shown in Fig. 3. As shown in panel A, the IEC 1/48 monoclonal antibody immunoprecipitated 1 from both brain and kidney, but did not immunoprecipitate 2, demonstrating that the antibody is 1-specific. As shown in panel B, the antibody coimmunoprecipitated 2 along with 1 from soleus muscle, evidence that 1 forms heterodimers with 2. 1 was also detected in these immunoprecipitated samples from soleus (not shown), evidence for 11. A corollary to the findings that both 1 and 2 form heterodimers with 1 and that 2 is not detected in this muscle is that 1 abundance may provide a measure of total sodium pump heterodimer pool size in this muscle.

Coimmunoprecipitation of 21.

lane 1

In 8-week-old rats placed on a K⁺-deficient diet for 10 days, serum K⁺ fell from 4.0 ± 0.14 mM to 2.3 ± 0.08 mM (control animals, n = 12; low K⁺ animals, n = 11). K⁺-deprived rats gained slightly less weight during the 10 days, attaining a final body weight of 267 ± 0.9 g, which was 7% lower than controls (287 ± 1.9 g). Typical autoradiograms of skeletal muscle homogenates from control and hypokalemia rats are shown in Fig. 4, and the relative changes in Na,K-ATPase  and  subunit isoforms are summarized in Fig. 5. There were no significant changes in 1 abundance. In contrast, 2 decreased significantly in all skeletal muscles and cardiac left ventricular muscle. The decline in 2 was most pronounced in fast glycolytic dominated muscles, falling to 0.06 of control in WG, followed by decreases to 0.39, 0.42, 0.44, and 0.71 of control in RG, EDL, soleus, and diaphragm, respectively. 1 decreased 25% or less in RG, EDL, and diaphragm (muscles containing oxidative fibers) to 0.74, 0.76, and 0.83 of control, respectively. However, 1 did not decrease significantly in soleus, which is puzzling given that no 2 was detected in this muscle, and 2 decreased by 56%. This result suggests a dissociation between pool sizes of  and  in this muscle. 2 decreased 60% or more in all muscles where it was detected, falling in RG, EDL, and WG to 0.27, 0.40, and 0.30 of control, respectively. These results demonstrate distinct muscle-specific responses to K⁺ deprivation: 22 decreases more than 50% in fast glycolytic muscle fibers (WG, EDL, and RG), whereas 2 alone decreases in oxidative rich soleus, with only a minor decrease in 21 in mixed oxidative diaphragm. Cardiac ventricle was assayed in parallel, and in agreement with previous findings (11) 2 but not 1 abundance decreased to 0.69 of control.

Detection of Na,K-ATPase  and  isoform abundance in skeletal muscles from control and K⁺-deprived rats.

Effect of K⁺ deprivation on Na,K-ATPase  and  isoform abundance in skeletal muscles.

Since changes in 2 abundance were so pronounced, we considered the possibility that the NH₂-terminal epitope recognized by the monoclonal McB2 had been cleaved. Blotting with an additional anti-2 polyclonal antibody, anti-HERED, which recognizes a sequence of amino acids (491-495) found on the early part of the large cytoplasmic loop, between transmembrane segments H4/H5, and unique to 2 isoforms (12, 18), confirmed that the decrease in 2 signal was not due to cleavage of the epitope (Fig. 6). The pattern of change in 2 in response to K⁺ deprivation is identical with both antibodies.

Detection of 2 isoform with two different antibodies.

Based on the distinct muscle-specific changes in Na,K-ATPase expression in hypokalemia we predicted that intracellular K⁺ would fall more in WG, which lost greater than 70% of both 2 and 2, than in diaphragm, which lost only 30% of 2 and 1. To test this hypothesis, ICF K⁺ was measured in paired muscles from the same pool of rats used for subunit abundance analysis. Fig. 7 summarizes ICF K⁺ levels expressed as µmol/g, wet weight, in skeletal muscles from control and K⁺-deprived rats. In contrast to our prediction, by day 10 all skeletal muscles lost about 20% of ICF K⁺. Additionally, ICF K⁺ measured in left ventricles from the same animals did not change. These results suggest a revised hypothesis, that muscles may have multiple mechanisms in place to regulate intracellular K⁺ loss besides a decrease in pump number, such as pump translocation to internal stores, as described previously for 21 in soleus in response to insulin stimulation (30, 31).

DISCUSSION

The existence of isoforms suggests the potential for differential isoform-specific function, tissue-specific expression, and regulation. Functional differences in enzymatic and transport properties between the 1 and 2 isoforms expressed in skeletal muscle are subtle if any (32). This study demonstrates a muscle type-specific pattern of Na,K-ATPase isoform expression and distinct muscle-specific patterns of regulation of Na,K-ATPase isoforms in response to K⁺ deprivation.

Regarding muscle-specific expression, Hundal et al. (9) reported that pooled membranes from muscles enriched in slow twitch oxidative fibers expressed 1, not 2; that pooled membranes from muscles composed of fast twitch glycolytic fibers expressed 2, not 1 isoform; and that 1 and 2 were similar in the two types of muscle membranes. This distinction prompted our examination of  and  isoform expression in five distinct skeletal muscles. 1 and 1 were detected in all muscles and were at least twice as abundant in oxidative fiber rich soleus and diaphragm than in fast twitch glycolytic WG, whereas 2 expression was equivalent in all five muscles. In agreement with Hundal (9), 2 expression was limited to fast glycolytic muscle types. Additionally, we note that 2 expression appears directly proportional to the relative percentage of fast glycolytic fibers, that is, 2 is most abundant, relative to cellular protein, in WG, which is 84% fast glycolytic; less abundant in EDL, which is 56% fast glycolytic; and below detection in slow oxidative soleus, which has no fast glycolytic fibers. In contrast to the Hundal study, we detected significant levels of 1 in EDL, an almost pure fast muscle with 42% fast glycolytic-oxidative fibers. This may be a consequence of deglycosylation of EDL samples prior to analysis, allowing for enhanced 1 subunit detection in this study. We conclude that 1, 1, and 2 are expressed ubiquitously in skeletal muscle, 1 and 1 enriched in oxidative fibers and 2 about the same relative abundance across muscle, whereas 2 is expressed in only a subset of muscles containing fast twitch glycolytic fibers.

The expression of two  and two  isoforms suggests the potential for four different heterodimer combinations unless constraints prevent an  from combining with a  subunit. We know from both purification and immunoprecipitation studies that 11 heterodimers are the active sodium pumps in kidney and epithelial cell lines (23). The immunoprecipitation of both 11 and 21 from soleus muscle in this study provides in vivo evidence that both s complex with 1. The corollary to this finding is that if nearly all the  is assembled as heterodimers, 1 subunit abundance will provide a relative measure of total sodium pump pool size in soleus, since it lacks 2. The work of Gloor et al. (33) demonstrates that glial 2 forms heterodimers with 2 but not 1 subunits. Whether 2 will be constrained to form heterodimers with only 2 in skeletal muscles remains to be established.

In our initial investigation into the effect of hypokalemia on Na,K-ATPase expression we reported decreased abundance of 2, but not 1 or 1, in pooled hindlimb skeletal muscle, heart, and brain. The changes in skeletal muscle 2 were by far the greatest: protein and mRNA levels decreased 80 and 40%, respectively, after 14 days of K⁺ deprivation (11). This study examined whether changes were restricted to a subset of muscle types found in the rat hindlimb. The results establish that 2 protein levels are also depressed more than 50% during 10 days of K⁺ deprivation in all muscles expressing this isoform (RG, EDL, and WG). The failure of our prior study to detect changes in 1 is likely a consequence of enhanced sensitivity of 1 subunit detection following deglycosylation of muscle samples prior to analysis. Overall, three distinct responses to K⁺ deprivation were observed in the panel of muscles: 1) in WG the response was the greatest, 2 and 2 decreased 94 and 70%, respectively; 2) in RG and EDL 2 and 2 decreased about 60%, and 1 decreased about 25%; and 3) in soleus and diaphragm 2 decreased 55 and 25%, respectively, with little if any accompanying change in 1 and no detection of 2. From this pattern we predicted that the WG would lose more intracellular K⁺ than soleus or diaphragm. However, intracellular K⁺ decreased equivalently in all five skeletal muscles after 10 days of K⁺ deprivation. In comparison, while we observed a 31% decrease in 2 abundance in K⁺-deprived cardiac left ventricular muscle, there was no significant decrease in cardiac intracellular K⁺, which can be attributed to the low percentage of 2 type pumps (25% or less) in the heart (6). Norgaard et al. (34) observed that after 4 weeks of K⁺-deficient diet there was only a 13% fall in left ventricle ICF K⁺ and a concomitant 43% decrease in [³H]ouabain binding (a measure of 2 abundance at the plasma membrane), the difference likely due to the longer deprivation period. Taken together, the results of this study demonstrate that the abundance of 2 and 2, and not the ubiquitous 1 isoform, is depressed in hypokalemia, and demonstrate small but significant decreases in 1 in a subset of muscles assayed, which is in contrast to our previous report of no change in 1 (11). These results, coupled with information on the relative ratios of 1 to 2 protein in various tissues, provide a molecular explanation for the significant and selective loss of K⁺ from skeletal muscle and only minor loss of K⁺ from cardiac muscle. We hypothesize that the presence and regulation of the 2 isoform in skeletal muscle provided an evolutionary advantage to complex organisms that needed to maintain transmembrane K⁺ gradients in the face of fluctuations in K⁺ availability (3).

We have hypothesized that expressing Na,K-ATPase subunits in a fiber type-specific pattern would provide an organism with the means to regulate fiber type involvement in the regulatory response to a K⁺ challenge (35). Specifically, one subset of muscles may be adapted to respond to hypokalemia by expressing isoforms that respond to insulin and rapidly move dietary K⁺ from ECF to ICF after a meal to avert cardiac complications from elevated extracellular K⁺, while another subset of muscles may be adapted to respond to hypokalemia by expressing isoforms that decrease in response to chronic K⁺ deprivation leading to the loss of K⁺ from the ICF to the ECF. Parallel specialization would allow an animal both to chronically lose K⁺ during deprivation and rapidly clear a ECF K⁺ load when a meal (and insulin surge) ends the K⁺ deprivation. Recent evidence from Lavoie et al. (31) supports this hypothesis by demonstrating that insulin increases Na,K-ATPase transport activity only in oxidative fiber type skeletal muscles (for example, soleus and most RG) by inducing translocation of 2 and 1 from intracellular to plasma membranes. Additional support comes from the study of Hsu and Guidotti (36), demonstrating that this insulin-stimulated K⁺ uptake response is still present in a hypokalemic animal with depressed sodium pump expression. Arguing against specialization during K⁺ deprivation are the findings of this study. After 10 days of K⁺ deprivation, the oxidative fiber muscle types lose as much intracellular K⁺ as the fast glycolytic muscle even though there are much smaller decreases in  and  subunit pool sizes in oxidative fiber types. One hypothesis that reconciles these findings is that the molecular mechanisms mediating the responses may be muscle fiber type-specific. For example, sodium pump transport activity may be depressed in slow oxidative fiber muscle types by both decreasing abundance and translocation of the predominant 21 heterodimers from plasma membrane to internal stores (where they will be poised to respond to insulin), while sodium pump transport activity in fast glycolytic muscle fiber types, where 22 heterodimers predominate, may be depressed solely by decreasing pool sizes of this isoform. Such a scenario would allow skeletal muscle K⁺ stores to buffer the fall in extracellular K⁺ during K⁺ deprivation while retaining the ability to clear a potassium load from the ECF into the skeletal muscle during hyperkalemia.