Renal Physiology

Skeletal muscle regulates extracellular potassium

Alicia A. McDonough, Curtis B. Thompson, Jang H. Youn


Maintaining extracellular fluid (ECF) K+ concentration ([K+]) within a narrow range is accomplished by the concerted responses of the kidney, which matches K+ excretion to K+ intake, and skeletal muscle, the main intracellular fluid (ICF) store of K+, which can rapidly buffer ECF [K+]. In both systems, homologous P-type ATPase isoforms are key effectors of this homeostasis. During dietary K+ deprivation, these P-type ATPases are regulated in opposite directions: increased abundance of the H,K-ATPase “colonic” isoform in the renal collecting duct drives active K+ conservation while decreased abundance of the plasma membrane Na,K-ATPase α2-isoform leads to the specific shift of K+ from muscle ICF to ECF. The skeletal muscle response is isoform and muscle specific: α2 and β2, not α1 and β1, levels are depressed, and fast glycolytic muscles lose >90% α2, whereas slow oxidative muscles lose ∼50%; however, both muscle types have the same fall in cellular [K+]. To understand the physiological impact, we developed the “K+ clamp” to assess insulin-stimulated cellular K+ uptake in vivo in the conscious rat by measuring the exogenous K+ infusion rate needed to maintain constant plasma [K+] during insulin infusion. Using the K+ clamp, we established that K+deprivation leads to near-complete insulin resistance of cellular K+ uptake and that this insulin resistance can occur before any decrease in plasma [K+] or muscle Na+ pump expression. These studies establish the advantage of combining molecular analyses of P-type ATPase expression with in vivo analyses of cellular K+ uptake and excretion to determine mechanisms in models of disrupted K+ homeostasis.

  • sodium, potassium-adenosine triphosphatase
  • hydrogen, potassium-adenosine triphosphatase
  • potassium clamp
  • hypokalemia
  • ion homeostasis

recent clinical studies havedemonstrated a positive association between diets rich in K+ and the control of blood pressure and prevention of stroke (4, 6, 17). This association is not surprising given the fundamental importance of K+ (the main intracellular cation) in determining cell volume as well as nerve and muscle excitability, both dependent on the steep transmembrane K+ gradient established by the ubiquitous Na+pump, Na,K-ATPase. The hydrolysis of ATP by the Na+ pump fuels the coupled uphill transport of K+ into the cell and Na+ out of the cell. Extracellular K+concentration ([K+]) is closely regulated, between 3.8 and 5 mM, by the concerted responses of two organ systems: muscle, which contains the major pool of K+ and regulates K+ distribution between the ICF and ECF compartments, and kidneys, which regulate K+ excretion by secreting or actively reabsorbing K+ (Fig.1) (12, 13, 18). The duet between these two systems is critical for both short-term and long-term K+ homeostasis.

Fig. 1.

K+ homeostasis in an average person. Extracellular fluid (ECF) K+ homeostasis is maintained by concerted regulation of kidney and muscle. ECF K+continually enters the kidneys as a function of glomerular filtration rate (GFR), and renal adjustments match K+ output to K+ input: normally 90% of the filtered load is reabsorbed. During K+ deprivation or fasting, nearly 100% of the filtered load is reabsorbed. Skeletal muscle takes up excess K+ from the ECF after a meal, driven by insulin, and also takes up excess ECF K+ during exercise, driven by catecholamines. Skeletal muscle continually loses muscle intracellular fluid (ICF) K+ to the ECF with muscle activity. During K+ deprivation, the loss of ICF K+ to the ECF is increased because the cellular uptake of K+ is decreased, buffering the fall in ECF K+ concentration ([K+]).

To frame the challenges to K+ homeostasis, one can consider commonly encountered challenges that require an acute response. Total extracellular K+ is ∼70 meq, and the average dietary K+ load is 80 meq/day (greater than the total ECF K+) (Fig. 1). Consider the consequences of eating an average cantaloupe, which contains ∼2 g or 50 meq of K+(50) that could be added to the ECF within a very short time of its ingestion. If not adjusted quickly by renal excretion or muscle uptake, ECF [K+] would increase to a dangerous 7.5 mM. However, consuming the melon increases blood glucose and stimulates insulin release, which causes muscle and fat to actively take up not only glucose but also excess K+ not excreted in the short term by the kidneys. Muscle activity releases K+chronically into the ECF and, mysteriously, the appropriate fraction is excreted by the kidneys rather than taken up again by the muscles; thus renal output is adjusted to match dietary input. As another example, one can consider K+ balance during vigorous exercise by which K+ release exceeds the capacity for muscle to actively take it up again, and muscle ECF [K+] increases within minutes by >3 mM (12, 13). In this case, nerve excitation and consequent release of catecholamine stimulate active K+ uptake into muscle via the Na,K-ATPase to restore ECF [K+] (1, 12, 13). Another example occurs when there is too much ECF [K+] during end-stage renal disease, where K+ output is effected by periodic dialysis, and dietary K+ must be sequestered in the muscle between dialysis sessions. On the occasions when life-threatening hyperkalemia develops, the therapy is acute administration of both insulin and adrenergic agonists to maximize muscle K+ uptake (1).

Complementing these examples of acute regulation, chronic adjustments must be activated when K+ intake is continuously less than output, delaying and minimizing the resulting hypokalemia (lists of causes of hypokalemia can be found in any renal physiology text). When hypokalemia results from fasting or consuming K+-deficient diets, the kidney avidly retains K+ by shifting from net K+ secretion to net K+ absorption via the apical H,K-ATPases in the distal nephron, and renal output falls to near zero (18, 51). However, K+ loss in the stools and sweat persists, so K+ must be continuously redistributed from muscle stores to the ECF, with the end result that muscle cell [K+] falls to balance the discrepancy between K+ input and output over time (20, 37, 48). A remarkable challenge to K+ homeostasis was documented in a study of soldiers in summertime basic training, who lost >40 meq K+/day in sweat alone. High muscle activity shifted plenty of K+ into the ECF so that frank hypokalemia did not develop, but after 11 days of training there was a total body K+ deficit of >400 meq. This was attributable to the fact that, besides sweat loss, K+ loss in urine persisted because daily bouts of dehydration stimulated aldosterone secretion (25).

Active transport of K+ by muscle Na,K-ATPase plays a central role in these scenarios of acute and chronic challenges to K+ homeostasis. This review will focus on the molecular mechanisms in place in muscle that contribute to K+homeostasis, in particular, muscle-specific regulation of Na+ pump isoforms, and a novel method we developed to assess cellular K+ uptake in vivo.


As introduced above, adjustments to maintain K+homeostasis are effected by a tight control of the activity and abundance of genetically related P-type ATPases: the muscle Na+ pumps (Na,K-ATPase) actively transport K+from ECF into muscle, and the renal H+/K+ pumps (H,K-ATPase) actively reabsorb K+ from the renal tubular fluid back into the ECF during K+- deficient states (45, 47). Both Na,K-ATPase and H,K-ATPase are 1:1 heteromers of ∼100 kDa α-catalytic subunits and ∼50 kDa β-glycoprotein subunits. These classes of P-type ATPases share 65% homology, and there are multiple isoforms of each type of pump (24, 45, 47).

The existence of isoforms suggests the potential for differential expression, function, and regulation. Indeed, isoform- and tissue-specific regulation of these pumps during challenges to K+ homeostasis provides the compelling rationale for the existence of these isoforms. In the kidney, both gastric and nongastric isoforms of the H,K-ATPase are found in the distal portion of the nephron (24, 45). Recent reviews have focused on the renal control of K+ excretion (18, 45, 51), and specifically on H,K-ATPases (24, 45). K+output is ultimately controlled in the collecting tubules and duct, where, during K+ restriction, there are increased abundance and activity of the nongastric, colonic isoform of the α-subunit of H,K-ATPase along with the β1- isoform classically assigned to the Na+ pump (αcβ1) in apical membranes (15, 16, 26, 27) as well as decreased abundance of the ROMK secretory K+ channel (36) (Fig. 2). These responses are responsible, at least in part, for the shift from K+ secretion to active K+ reabsorption that occurs in the collecting duct during K+-restricted states. Studies in colonic H,K-ATPase-deficient mice suggest that the decrease in K+ secretion may be a very important component, as these mice are able to reduce urinary K+ excretion during K+ deprivation to nearly the same extent as mice with normal H,K-ATPase levels (35). The remainder of this review will focus on the role of muscle in K+homeostasis.

Fig. 2.

Compartmental model of K+ homeostasis. Input and output routes between ICF, ECF, and the animal's exterior are illustrated. Red oxidative muscle fibers express plasma membrane Na+ pumps (Na,K-ATPase) composed of either α1- or α2-catalytic subunits along with β1-glycoprotein subunits, and α2β1 pumps are found in endosomal vesicles. White glycolytic muscle fibers express either α1- or α2- along with β2-glycoprotein subunits; there is as yet no evidence for endosomal pools. A number of K+ channels mediate K+ efflux from the skeletal muscle (down arrow). ECF is filtered into the renal tubule fluid by glomerular filtration. All along the nephron, basolateral Na,K-ATPase brings K+ into the renal ICF. In collecting duct principal cells, apical ROMK K+ channels secrete K+ into tubule fluid. Collecting duct intercalated cells express apical H,K-ATPase, and the colonic isoform (αcβ1) is increased during K+ deprivation to increase active K+ reabsorption.

With regard to Na+ pump isoforms, presently four α-isoforms have been identified: α1 is fairly ubiquitous, whereas α2, α3, and α4 have a limited tissue-specific expression pattern. Relevant to this review, both α1 and α2 are major isoforms in adult skeletal muscle and have ∼92% homology, whereas α3 and α4 are not appreciably expressed. There are also three isoforms of the β-glycoprotein: β1 is expressed in most tissues, β2 and β3 have a more restricted distribution, and β1 and β2 are expressed in muscle. Specific characteristics of Na+ pump isoform structure and function have been recently reviewed (7, 47).

Skeletal muscle is composed of a heterogeneous mix of muscle fiber types (46) historically classified by their metabolic (oxidative vs. glycolytic), contractile (fast twitch vs. slow twitch), and/or phenotypic (red vs. white) properties. Although most muscles contain a mixture of muscle fiber types, at the phenotypic extremes soleus contains 87% slow oxidative red fibers that are important for antigravity, weight bearing, and sustained movement, whereas white gastrocnemius contains 84% fast glycolytic white fibers important for concentrated bursts of power (2, 3). Studies in muscles at the phenotypic extremes have helped to determine whether there is a functional division of labor between muscle types, while providing information requisite to interpret function and response in mixed-fiber muscles. To understand the complexity of muscle-specific isoform expression, we assayed α1, α2, β1, and β2 abundance in a panel of five phenotypically distinct muscles by immunoblot analysis and found that, indeed, there are distinct distributions of Na+ pump α- and β-isoforms in muscle. Expression of α2 was fairly uniform across control muscles, but α1 expression was twice as high in oxidative muscles such as soleus and diaphragm than in mixed or fast glycolytic muscles such as gastrocnemius and extensor digitorum longus (49). It is difficult to assess the fraction of α1- and α2-type pumps in muscle, but estimates place the percentage of α2 protein at 40–60% (28, 48). The Klip (23) and McDonough (49) laboratories have also found that β1 is expressed without β2 in soleus and diaphragm and β2 is expressed without β1 in white gastrocnemius (23, 49). Thus the predominant isoforms in rat red oxidative muscle are α1β1 and α2β1and in white glycolytic muscle are α1β2-and α2β2-isoforms (Fig. 2); both are expressed in mixed-fiber muscles. This differential expression suggests the possibility for differential subcellular distribution or regulation. Indeed, α2β1-type Na+ pumps are also located in internal endosomal vesicle membranes in red, but not white, muscle (22) (Fig. 2), and insulin provokes a redistribution from internal membranes to the surface (29) in red muscle, which can explain, at least in part, insulin stimulation of active K+ uptake.


Skeletal muscle has an altruistic specialization to lose K+ in the face of whole body K+ deprivation to preserve ECF [K+]. This response is the converse of that observed in cultured cells. When muscle, heart, or kidney cells are placed in low-K+ medium, the drop in extracellular [K+] increases the gradient for K+ loss from the cells and limits Na,K-ATPase activity, leading to both a fall in ICF K+ and a reciprocal rise in ICF Na+. These stimuli provoke an increase in Na+ pump synthesis and a subsequent increase in the number of active Na+ pumps at the plasma membrane, which counteracts the increased K+leak from the cells (8, 30, 39, 52). In cultured kidney cells, the increase in Na+ pump abundance is driven by an increase in β1 (not α1) mRNA levels, which allows an increase in the synthesis and/or accumulation of α1β1 heteromers (30-32). We also detected an 11-fold increase in β1 protein levels (with a 3-fold increase in α1) in the outer medulla of rats deprived of K+ for 14 days and postulated that the increased pool of β1 may play a critical role in driving assembly and accumulation of H,K-ATPase induced during K+deprivation through αCβ1 heteromer formation (34) (Fig. 2). An increase in Na+pump number is also seen in erythrocytes of hypokalemic animals, which maintain normal ICF levels of Na+ and K+(reviewed in Ref. 13). The response is just the opposite in skeletal muscle in vivo, the main stores of the body's K+. Muscle loses cell [K+] during K+ deprivation and responds by decreasing Na+pump number, which facilitates further loss of cell K+ to the ECF, which buffers the fall in ECF [K+] (20,37). Other tissues such as liver and brain do not lose cell K+ during hypokalemia (20, 37). Thus the response of skeletal muscle to K+ deprivation must be studied in vivo rather than in cultured muscle cells.

Insight into the molecular mechanisms responsible for the loss of muscle K+ during K+ deprivation was provided 20 years ago by Norgaard et al. (37). In this comprehensive study, they established that rats fed a K+-deficient diet for 2 wk had a 40% decrease in ICF [K+] in gastrocnemius, a decrease in K+ influx into isolated soleus (by 86Rb uptake), and a >50% decrease in Na+pump activity along with a similar decrease in Na+ pump number measured in intact soleus muscle by the binding of3H-labeled ouabain, a specific inhibitor of Na,K-ATPase. This work preceded the cloning and identification of Na+pump isoforms, and it was subsequently established that the α1-isoform is quite ouabain resistant in rodents, so the decrease in ouabain binding likely reflects a change in the α2-isoform abundance (38). Because ouabain binding measurements were conducted in intact muscles, the measurement reflects decreased active α2-type pumps in the surface membranes. This could arise from redistribution of pumps to internal stores, inhibition of pumps in the plasma membrane, or a change in total abundance of pumps (Fig. 3). Whether there is also a change in the α1-isoform could not be assessed by ouabain binding.

Fig. 3.

Biosynthesis and regulation of Na+ pump isoforms during low-K+ diet. Top: schematic of parallel synthesis of Na,K-ATPase α- and β-subunits, assembly in the endoplasmic reticulum (ER), and processing and targeting to the plasma membrane. Potential mechanisms of regulation of mature Na+ pump activity or abundance in the plasma membrane are shown [increase (⊕) and decrease (⊝)].Bottom: results of 10 days (10d) of a low-K+diet on α2 and β2 mRNA and protein levels in gastrocnemius (49). Each lane represents a sample from a different animal, and a constant amount of protein or total RNA was applied to each lane. Samples were deglycosylated, which significantly improves immunodetection and quantitation of the heavily glycosylated β-subunits.

These findings stimulated us to examine whether the decrease in muscle Na+ pumps in the K+-deprived muscle was isoform specific. The results established an isoform and muscle specificity of the response to 10-day dietary K+ deprivation: the levels of the ubiquitous α1-isoform were unchanged, whereas α2 decreased 30% in diaphragm, 60% in soleus, and >90% in white gastrocnemius; and β1 was decreased minimally (20% fall in soleus), whereas β2 decreased 75% in white gastrocnemius (5, 48, 49). Surprisingly, α and β mRNA levels were unchanged during 2–10 days of a K+-deficient diet (48) (Fig. 3). Despite these differences, the associated fall in cell [K+] was similar in both soleus and white gastrocnemius (∼20 mM), suggesting that other factors besides αβ-pool sizes (e.g., redistribution from surface or inhibition in soleus) may be important to the response (5, 48, 49). Taken together, the results of these studies demonstrate that the expression of the muscle α2 (not α1)- and β2 (perhaps β1)-subunits are specifically depressed during K+ deprivation and hypokalemia in vivo mediated by decreased synthesis rate or increased degradation rate of the subunits. The findings are in distinct contrast to those in cultured cells deprived of K+, in which an increase in Na,K-ATPase β1 synthesis drives the assembly of additional α1β1 pumps, which facilitates K+ accumulation. These in vivo results provide evidence that the presence and regulation of the α2-isoform in skeletal muscle provided the evolutionary advantage needed by complex organisms to maintain transmembrane K+ gradients in the face of fluctuations in K+ availability (33).


The findings in K+-deprived rats support the hypothesis that a decrease in active K+ uptake by Na,K-ATPase α2 leads to a net loss of K+ from muscle and transfer to ECF, where it buffers the fall in [K+]. To establish a physiological link in vivo between the fall in muscle α2 Na+ pump expression and a decrease in active K+ uptake, we exploited the theory behind the “glucose-clamp” technique, used to measure insulin-stimulated glucose uptake in vivo, to develop a K+ clamp to measure insulin-activated K+uptake. In brief, conscious rats are infused with insulin to stimulate K+ (and glucose) uptake, and then, based on rapid and frequent assays of plasma [K+], infused with enough K+ (and glucose) to clamp plasma K+ (and glucose) at baseline (Fig. 4,A and B). The total amount of K+infused (Kinf) over a defined time period is equivalent to the sum of the insulin-stimulated portions of cellular K+uptake and K+ excretion (11). Previous studies by us (11) and others (44) showed that insulin at physiological concentrations does not significantly increase renal K+ excretion. Therefore, most of Kinfshould be attributable to insulin-stimulated cellular K+uptake. Thus the K+ infusion rate during the K+clamp appears to be a good measure of insulin-stimulated cellular K+ uptake.

Fig. 4.

The K+ clamp. A: illustration of the method as described in Choi et al. (11). B: hyperinsulinemic (5 mU · kg−1 · min−1) glucose and K+ clamps in control (●) and 2 (○)- and 10-day (+) K+-deprived rats. Adapted from Ref. 11.

We tested the hypothesis that insulin-stimulated cellular K+ uptake is a function of muscle α2abundance by comparing uptake in rats deprived of K+ for only 2 days, where there was a minor drop in plasma [K+ ] (from 4.2 to 3.8 mM) and muscle α2, with those deprived of K+ for 10 days, where plasma [K+ ] fell to 2.9 mM and mixed-fiber muscle α2 decreased by 50%. Surprisingly, the results did not support the hypothesis. After 2 days of K+ deprivation, there was an 80% reduction in insulin-stimulated cellular K+ uptake without a significant fall in Na,K-ATPase pool size or Na,K-ATPase enzymatic activity measured under maximal velocity conditions in total membranes (Fig.5), whereas insulin-stimulated glucose infusion was unchanged (Fig. 4 B). We are left to conclude that another mechanism comes into play to reduce cellular K+ uptake before the Na+ pump α2pool size falls. Possibilities include failure of insulin-induced translocation to the plasma membrane, inhibition of Na+pump activity in the plasma membrane not detected in the assay of total maximal Na,K-ATPase activity, activation of a K+ efflux route, or inhibition of a K+ uptake route unrelated to Na,K-ATPase. Ongoing studies using subcellular fractionation suggest the first possibility, that endosomal pools of Na,K-ATPase α2 are resistant to insulin-stimulated redistribution to the plasma membrane, whereas redistribution of GLUT4 is normal after 2 days of K+ deprivation (not shown) (19). After 10 days of a low-K+ diet, Kinf was further reduced to only 6% of control, likely reflecting the combined effects of resistance to insulin-stimulated redistribution and reduced pool size of Na,K-ATPase α2 (Figs. 4 B and 5). The K+ clamp technique proved to be a very valuable tool for analyzing the impact of molecular changes, testing hypotheses, and predicting new hypotheses.

Fig. 5.

Insulin-mediated plasma K+ disappearance (K Embedded Image ), Na,K-ATPase α2-subunit abundance, assessed by immunoblot of a constant amount of homogenate protein, and Na,K-ATPase activity in mixed fiber tibialis anterior muscles of control (day 0) rats and rats fed a K+- deficient diet for 2 or 10 days. Adapted from Ref.11.


The homeostatic regulation of K+ during insufficient K+ intake is often viewed as a simple negative-feedback system: as plasma K+ falls, the kidneys react by actively reabsorbing filtered K+ to minimize further K+ loss, while skeletal muscle releases cell K+, as a consequence of decreased active K+uptake, to buffer the restore falling K+ (12, 13, 40a). Although this proposed feedback suggests a causal role for plasma K+ as the error signal that drives the responses, there is evidence that regulatory responses precede changes in plasma [K+ ]. We discovered, inadvertently, that when rats were fed a high-fat diet (to induce resistance to insulin-stimulated glucose uptake), dietary K+ intake was reduced to one-third of normal (10). This regime decreased urinary K+ excretion as well as insulin-stimulated cellular K+ uptake (in addition to the decrease in insulin-stimulated glucose uptake), both of which were restored when dietary K+ was matched to that of controls. Interestingly, there was a strong correlation between urinary K+ excretion and insulin-stimulated cellular K+uptake, suggesting the possibility that the kidneys' function to excrete K+ and insulin's action to promote cellular K+ uptake are similarly (in concert) regulated in response to K+ intake. Importantly, basal plasma [K+] was not reduced at all by the threefold lower K+intake, indicating the existence of an effective homeostatic mechanism. These data suggest that the body must have a way of sensing low-K+ intake independently of plasma K+.

A number of laboratories have investigated the sensing of K+ status. Muscle adaptation to K+ deprivation is apparently not dependent on the nervous system, as the Clausen laboratory (14) established that there was the same ultimate loss of ouabain binding sites during K+deprivation in control and denervated muscles. Rabinowitz and colleagues (40, 42) theorized that K+ sensors in the gut, portal circulation, and/or liver respond to local changes in K+, secondarily to enteric changes. In sheep, they demonstrated that urinary K+ excretion increased after a meal, whether or not there was a change in plasma K+, and then progressively decreased. When sheep were fasted for 1 day between feeding days, urinary K+ fell, and, when they were refed, K+ excretion was proportionate to the K+content of the meal. “Meal-induced” kaliuresis did not change aldosterone, insulin, or glucagon levels, and the mediators of this “reflex kaliuresis” remain undetermined (40-42). The suggestion that a novel K+-sensitive receptor may exist is not unreasonable given the identification of Ca2+-sensing receptors found in a number of tissues (9) and glucosensors localized to the portal vein (21). K+-sensing receptors in the hepatic portal vein or liver could detect changes in K+ intake but may not respond to urinary K+ wasting provoked by diuretics in the face of normal diet (until plasma K+ fell).

In conclusion, the studies reviewed establish the advantage and necessity of combining molecular analyses of P-type ATPase expression with in vivo analyses of cellular K+ uptake and excretion for a determination of mechanisms and mediators in models of disrupted K+ homeostasis.


This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-57678 and DK-34316 and a Grant-in-Aid from the American Heart Association, Western States Affiliate.


  • Address for reprint requests and other correspondence: A. A. McDonough, Dept. of Physiology and Biophysics, Univ. of Southern California Keck School of Medicine, 1333 San Pablo St., Los Angeles, CA 90089-9142 (E-mail:mcdonoug{at}

  • 10.1152/ajprenal.00360.2001


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 40a.
  42. 41.
  43. 42.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
View Abstract