Potassium conservation is impaired in mice with reduced renal expression of Kir4.1

Potassium conservation is impaired in mice with reduced renal expression of Kir4.1. Am J Physiol F1271–F1282, 2018. better understand the role of the inward-rectifying K channel Kir4.1 (KCNJ10) in the distal nephron, we initially studied a global Kir4.1 knockout mouse (gKO), which demonstrated the hypokalemia and hypomagnesemia seen in SeSAME/EAST syndrome and was associ- ated with reduced Na/Cl cotransporter (NCC) expression. Lethality by ~3 wk, however, limits the usefulness of this model, so we developed a kidney-speciﬁc Kir4.1 “knockdown” mouse (ksKD) using a cadherin 16 promoter and Cre-loxP methodology. These mice appeared normal and survived to adulthood. Kir4.1 protein expression was decreased ~50% vs. wild-type (WT) mice by immunoblotting, and immunoﬂuorescence showed moderately reduced Kir4.1 staining in distal convoluted tubule that was minimal or absent in connecting tubule and cortical collecting duct. Under control conditions, the ksKD mice showed metabolic alkalosis and relative hypercalcemia but were normokalemic and mildly hypermagnesemic despite decreased NCC expression. In addition, the mice had a severe urinary concentrating defect associated with hypernatremia, enlarged kidneys with tubulocystic dilations, and reduced aquaporin-3 expression. On a K/Mg-free diet for 1 wk, however, ksKD mice showed marked hypokalemia (serum K: 1.5 (cid:2) 0.1 vs. 3.0 (cid:2) 0.1 mEq/l for WT), which was associated with renal K wasting (transtubular K gradient: 11.4 (cid:2) 0.8 vs. 1.6 (cid:2) 0.4 in WT). Phosphorylated-NCC expression increased in WT but not ksKD mice on the K/Mg-free diet, suggesting that loss of NCC adaptation underlies the hypokalemia. In conclusion, even modest reduction in Kir4.1 expression results in impaired K conservation, which appears to be mediated by reduced expression of activated NCC.


INTRODUCTION
Maintaining potassium homeostasis in the face of decreased K intake or increased gastrointestinal (GI) or renal K losses is critical to survival. Hypokalemia is associated with ventricular arrhythmias, muscle weakness, chronic kidney disease from interstitial fibrosis, and growth retardation in children (23,53). A number of adaptations occur both systemically as well as in GI and renal tissue to defend the serum [K] in the face of reduced intake or increased losses. In the setting of hypokalemia, potassium moves from intracellular stores (principally muscle), where 98% of the total body K resides, to replete extracellular compartments (56). Reduced plasma [K] results in hyperpolarization of adrenal glomerulosa cells, leading to decreased aldosterone secretion (33,67), resulting in reduced K secretion in colon and distal nephron (37,50,66). GI adaptation also includes increased distal colonic K reabsorption via the colonic H/K ATPase, which appears to be mediated by increased expression of the ␤-subunit (13,41,58).
Renal adaptation to K deficiency also involves both reduced secretion and increased K reabsorption. K secretion through ROMK, the main renal K secretory channel in the distal nephron, is decreased with dietary K restriction (17), which is associated with increased ROMK endocytosis (7). Multiple mechanisms may contribute to this including inhibition by superoxide-induced protein tyrosine kinase and mitogen-activated protein kinase signaling (3,77), WNK1/KS-WNK1 kinase inhibition (30,34,79), inhibition by angiotensin II (78), and low luminal K itself (15). Although aldosterone does not directly affect ROMK, its reduced plasma level with K restriction is associated with decreased epithelial sodium channel (ENaC) expression, which reduces the driving force for K secretion (14,16). A low-K diet also results in increased K reabsorption in distal nephron intercalated cells via increased H/K ATPase activity (22,44). Furthermore, there is evidence for increased K reabsorption ("recycling") in the medulla with dietary K restriction (81). As a result of these adaptations, urinary and stool K losses as low as 6 and 4 mEq/day, respectively, have been reported in human volunteers on a K-restricted diet (1,68).
Recently, there has been increased recognition of the role of the distal convoluted tubule (DCT) in K conservation. Dietary K restriction results in increased NCC expression (14,72,74), reducing distal delivery of Na and Na-dependent K losses in the connecting tubule (CNT) and cortical collecting duct (CCD). There has also been increasing evidence supporting the role of the inward-rectifying K channel Kir4.1 (KCNJ10) in the regulation of NCC (20,73,83).
Kir4.1 is present in brain, inner ear, retina, and kidney where it has numerous important functional roles (28,38,43). In the kidney, Kir4.1 is distributed along the basolateral surface of the distal nephron where it is likely involved, as a heteromer with Kir5.1, in setting the basolateral membrane potential and recycling K brought across this membrane by the Na-K-ATPase (25,29,35,79). Loss-of-function mutations of this channel underlie the SeSAME/EAST syndrome, a neurorenal condition whose renal phenotype includes hypokalemic, hypomagnesemic metabolic alkalosis with hypocalciuria (6,61). Many of the disease mutations resulted in disrupted pH gating or surface expression (51,57,80). Global Kir4.1 knockout mice show growth retardation and early lethality (43). Urinary Na wasting and hypocalciuria were seen in these pups (6), but no serum measurements have been reported.
The hypokalemia and hypomagnesemia seen with SeSAME/ EAST syndrome are reminiscent of Gitelman's syndrome (18,65), suggesting that NCC function was impaired in this condition. To test this hypothesis and gain further insight into the pathogenesis of SeSAME/EAST syndrome, we investigated the renal phenotype of the global Kir4.1 knockout mouse (gKO). The gKO mice showed hypokalemia and hypomagnesemia, associated with urinary K and Mg wasting, as well as a trend toward increased serum bicarbonate level, consistent with SeSAME/EAST syndrome (19,20). In addition, we found decreased NCC immunofluorescence in kidney sections from gKO mice (20). This raised the intriguing question of how loss of Kir4.1 on the basolateral surface was coupled to reduced apical NCC expression. Zhang et. al. (83) subsequently showed that loss of Kir4.1 function in the gKO mice led to a marked reduction in DCT whole cell K current, membrane depolarization, decreased SPAK expression, and ultimately reduced NCC expression.
Unfortunately, early lethality [100% mortality by~3 wk (43)] limits the studies that can be done with the gKO mice. A gKO mouse also shows 100% mortality by~4 wk (10), suggesting that neuropathology largely underlies this demise. We therefore used Cre-loxP technology (31,45) to develop a mouse where renal Kir4.1 expression was significantly reduced, termed the kidney-specific Kir4.1 knockdown (ksKD) mouse, to better understand the role of Kir4.1 in the nephron under basal conditions as well as in various homeostatic and disease states. Results from both the gKO and ksKD mice are presented below.

Mice
The global Kir4.1 knockout mouse (gKO), Kir4.1 Ϫ/Ϫ , in which the KCNJ10 gene was disrupted by insertion of a neomycin-resistance construct within its coding region (28), was obtained from the Mutant Mouse Resource & Research Centers repository (https://www.mmrrc.org/). Mice were bred and maintained in strict accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and all experimental procedures were approved by the University Committee on Animal Resources (UCAR, University of Rochester).

Blood and Urine Collection
For the gKO mice, 3-to 7-day-old pups were anesthetized [ketamine (100 mg/kg) ϩ xylazine (10 mg/kg) ip] and decapitated by guillotine, and blood was collected via hematocrit tube, which was then spun (1,000 g for 5 min) to isolate serum. For the ksKD mice, 5to 9-mo-old animals were anesthetized (as above), and blood was collected by cardiac puncture. Urine was obtained by bladder aspiration. Serum and urine Na and K were determined by flame photometry, and the other electrolytes were measured via ion-sensitive electrodes at the Univeristy of Rochester Medical Center Clinical Research Laboratory except for serum bicarbonate in the gKO mice, which was measured at the University of Massachusetts Mouse Metabolic Phenotyping Center (Worcester, MA). Osmolality was determined using a vapor pressure osmometer. The transtubular K gradient (TTKG), an index of K secretion in the CCD, was calculated as [UK/(Uosm/Sosm)]/SK, where UK, SK, Uosm, and Sosm are the urine and serum K concentrations and osmolalities, respectively (11).

Immunofluorescence
For gKO mice, kidneys were excised and immersed in Nakane fixative (40) for 4 h, washed in phosphate-buffered saline (PBS), and immersed overnight in 30% sucrose at 4°C, and 4 -5 m sections were cut. Sections were subsequently washed in PBS and blocked for 30 min with dilution buffer (PBS with 0.1% BSA and 10% donkey serum; Jackson Immuno Research Laboratories, West Grove, PA). For ksKD mice, excised kidneys were embedded in OCT (Sakura Finetek, Torrance, CA), frozen over a liquid nitrogen bath for 10 s, and immediately stored at Ϫ20°C. Sections were subsequently cut, fixed in acetone at Ϫ20°C for 10 min, and then washed in PBS ϩ 0.1% BSA before blocking as above. Primary antibodies, incubated in dilution buffer for 1 h at room temperature, included rabbit anti-Kir4.

Statistics
Comparisons between gKO or ksKD and WT mice were done with age-matched animals, typically littermates. The two-tailed Student's t-test was used to determine whether differences between two groups were significant, and one-way ANOVA followed by the post hoc Tukey test was used for comparing more than two groups. Data are presented as means Ϯ SE. Differences were considered significant for P Ͻ 0.05.

gKO Mice
The gKO mice show hypokalemia and hypomagnesemia consistent with SeSAME/EAST syndrome. The gKO mice are much smaller and weigh less (Table 1) than their WT littermates. They fail to grow normally and die typically by 2-3 wk of age (43). The gKO kidneys appear normal grossly and by light microscopy (Fig. 1A). However, abnormalities in ultrastructure have been reported in SeSAME/EAST patients (51). gKO pups (at 3-7 days old) show relative hypokalemia (serum K: 5.1 Ϯ 0.3 vs. 6.8 Ϯ 0.5 mEq/l in WT) and hypomagnesemia (serum Mg: 2.3 Ϯ 0.1 vs. 2.9 Ϯ 0.3 mEq/l in WT; Table 1). This was associated with a transtubular potassium gradient, an index of K secretion in the CCD (see MATERIALS AND METHODS) that was inappropriately elevated (TTKG: 13.4 Ϯ 1.0 vs. 8.9 Ϯ 0.7 in WT), indicative of renal K wasting. Similarly, urinary [Mg], corrected for urinary concentration differences, was not decreased [urine Mg/(U osm /S osm ): 9.8 Ϯ 0.7 vs. 10.7 Ϯ 0.6 mEq/l in WT] in the setting of hypomagnesemia, consistent with renal Mg wasting. No hemolysis was seen, so the shift in serum K values to higher levels in all mice is likely due to intracellular release of K with decapitation (see MATERIALS AND METHODS). The gKO mice also showed an increased serum bicarbonate level (30 Ϯ 2 vs. 26 Ϯ 1 mEq/l in WT) that was nearly statistically significant (P ϭ 0.06), consistent with the metabolic alkalosis seen in SeSAME/EAST syndrome. The very low urine Na levels for all mice likely reflect normal/appropriate Na avidity in these neonates (2). In addition, the gKO mice have a mild urinary concentrating defect (U osm : 333 Ϯ 4 vs. 403 Ϯ 15 mosmol/kgH 2 O in WT) associated with moderate hemoconcentration (hematocrit: 42 Ϯ 3 vs. 34 Ϯ 2% in WT) and a trend toward hypernatremia. No significant differences in blood or urine electrolytes were observed between WT and heterozygous (Kir4.1 ϩ/Ϫ ) mice despite significant reduction in Kir4.1 protein expression (Fig. 1A), consistent with the recessive inheritance of SeSAME/EAST.

Kir5.1 surface expression is decreased in the gKO mice.
Kir4.1 likely forms the predominant basolateral K channel in the distal nephron as a heteromer with Kir5.1 (29,35). Kir5.1 does not form functional homomeric channels in heterologous systems (47,80) although these have been reported in brain (70). Instead, it typically requires coexpression with Kir4.1 or other Kir channels for surface expression (71). Thus loss of Kir4.1 expression would be expected to reduce surface expression of Kir5.1. Indeed, while bright basolateral staining in the distal nephron was seen in kidney sections of WT mice probed with Kir5.1 antibody (Fig. 1A), with a distribution similar to that seen with Kir4.1, only a diffuse signal, slightly brighter than background, was seen in the gKO mice, consistent with a recent report (83).
NCC expression is decreased in the gKO mice. Loss of functional Kir4.1 in SeSAME/EAST syndrome results in a hypokalemic, hypomagnesemic metabolic alkalosis, akin to Gitelman's syndrome or to the effect of thiazide diuretics. This suggests that the hypokalemia may be mediated, at least in part, by decreased NCC function in the DCT, leading to increased Na-dependent K secretion in the CNT and CCD. Indeed, immunofluorescence shows decreased apical NCC expression in kidney sections probed with NCC antibody in gKO vs. WT mice (Fig. 1B) as we previously reported in abstract form (20). Substantially decreased NCC expression is also seen on immunoblots of kidney lysates from gKO vs. WT mice and significantly reduced expression is seen in heterozygotes (Fig.  1B), consistent with recent reports (83).

ksKD Mice
The gKO mice typically die by~3 wk of age, before weaning (43) very challenging. A gKO mouse showed similar growth retardation and early lethality (10), suggesting that death in these mice was likely due to neurological complications. To better understand the role of Kir4.1 in health and disease, we therefore sought to generate a kidney-specific Kir4.1 knockout mouse using Cre-loxP methodology. Transgenic mice carrying the cadherin 16 promoter, which drives Cre recombinase expression in the distal nephron, that were bred to also be heterozygous for Kir4.1, were crossed with mice homozygous for floxed KCNJ10, the gene that codes for Kir4.1 (see MATE-RIALS AND METHODS). The resulting ksKD mice (4.1 f/-,Cre ϩ ) were outwardly indistinguishable from their WT littermates, grew normally to adulthood and had similar weights ( Table 2). ksKD kidneys, however, were grossly enlarged ( Table 2).
Kir4.1 protein expression in the distal nephron is decreased, but not abolished, in the ksKD mice. Immunoblots of kidney lysates probed with Kir4.1 antibody reveal a~50% reduction in Kir4.1 protein expression in the ksKD vs. WT mice ( Fig. 2A). WT kidney sections probed with Kir4.1 antibody showed basolateral staining that was bright in the DCT, mild to marginal in CNT and CCD (Fig. 2B), and absent in the medulla (data not shown), consistent with previous reports (29,51). The ksKD mice, relative to WT, had moderately reduced Kir4.1 intensity in the DCT and minimal if any CNT and CCD staining.
On control diet, the ksKD mice had normal serum K, increased serum Mg, metabolic alkalosis, relative hypercalcemia, and a concentrating defect. Loss of Kir4.1 function in SeSAME/EAST patients is associated with hypokalemia, hypomagnesemia, metabolic alkalosis, and hypocalciuria (6,61). Hypokalemia and hypomagnesemia were also demonstrated in the gKO mice (as above). In contrast, the ksKD mice show normokalemia and a mildly increased serum Mg level (2.9 Ϯ 0.2 vs. 2.2 Ϯ 0.1 mEq/l in WT; Table 2). Increased serum bicarbonate concentration, consistent with metabolic alkalosis, and increased serum Ca level, within the normal range, are also seen. The ksKD mice may well also have hypocalciuria but this could not be demonstrated because the urine [Ca] for these mice was below the laboratory threshold of detection (i.e., Ͻ1 mg/dl) due to their dilute urine. Only an upper limit to the urine Ca/creatinine ratio could therefore be calculated (Ͻ0.12 vs. 0.13 Ϯ 0.01 for WT); the actual ratio for ksKD mice may be much lower. Urinary electrolyte levels, normalized by urinary creatinine, were otherwise similar to WT. Finally, a marked concentrating defect (U osm : 450 Ϯ 25 vs. 1,553 Ϯ 279 mosmol/kgH 2 O in WT) is observed, associated with mild hypernatremia.
NCC expression is decreased in the ksKD mice. Immunoblots of kidney lysates showed significantly decreased NCC expression in the ksKD vs. WT mice (Fig. 3), as was seen with the gKO mice (Fig. 1B). Note, however, that unlike the gKO mice, which were hypokalemic, the ksKD mice showed no difference in serum K concentration ( Table 2), suggesting that a moderate loss of NCC expression is not sufficient to cause hypokalemia under control conditions. ksKD kidneys are enlarged and have dilated/cystic tubules. ksKD kidneys are 23% longer and weigh 61% more than their WT littermates ( Table 2). Numerous tubulocystic dilations are seen at the cortico-medullary junction of these mice (Fig. 4A). The tubule dilation likely results from presumed chronic polyuria associated with the significant concentrating defect in these mice. No colocalization of the dilated tubules with NCC or aquaporin-2 (AQ-2) antibodies was seen by immunofluorescence on ksKD kidney sections (data not shown), suggesting that either the tubules did not arise from DCT, CNT or CCD or that the dilated/cystic tubules lost expression of these transporters.
The concentrating defect in the ksKD mice is associated with reduced AQ-3 expression. The marked urinary concentration defect, associated with hypernatremia and tubule dilation, was seen in the ksKD mice under control conditions, where serum [K] was normal ( Table 2). A more subtle concentrating defect was also seen in gKO pups (Table 1). Disrupted urinary concentration might arise from reduced surface expression of AQ-2 or AQ-3, the major apical and basolateral water channels, respectively, in principal cells of the CNT and CCD (36). Alternatively, this could result from disruption of the medullary concentration gradient established by the NKCC2 in the TAL. Indeed, AQ-3 expression in ksKD mice was decreased 40% vs. WT in immunoblots of kidney lysates (Fig. 4B). No associated changes in cellular localization were seen in ksKD kidney sections probed with AQ-3 antibody. No change in AQ-2 expression by immunoblot was seen in the ksKD mice but kidney sections probed with AQ-2 antibody showed immunofluorescence staining that was similar to slightly more apically distributed in ksKD vs. WT mice (Fig. 4C), perhaps representing a compensation. Finally, no significant difference in total NKCC2 expression was seen in immunoblots of kidney lysates from ksKD vs. WT mice although the lower molecular weight (presumed monomeric) band was decreased (Fig. 4D).  On a K-and Mg-free diet, the ksKD mice showed marked hypokalemia associated with renal K wasting but not hypomagnesemia. No renal K or Mg wasting was seen in the ksKD mice on control diet (as above). To look more closely for abnormalities in K and Mg conservation, the mice were placed on a K and Mg free diet for 1 wk. We chose a combined K/Mg-free diet to simultaneously screen for both hypokalemia and hypomagnesemia in these initial studies. The K/Mg-free diet also mimics many common, clinically relevant states such as starvation, anorexia, alcohol abuse, and chronic diuretic therapy. On the K/Mg-free diet, serum K was markedly decreased in the ksKD mice (1.5 Ϯ 0.1 vs. 3.0 Ϯ 0.1 mEq/l for WT; Table 3), and these mice appeared listless with reduced muscle tone. Thus, while a modest, decrease in serum K in WT mice on K/Mg-free diet was seen [from 3.8 Ϯ 0.2 mEq/l on control diet (Table 2) to 3.0 Ϯ 0.1 mEq/l], a decrease more than threefold greater was seen with the ksKD mice on the K/Mg-free diet (from 4.2 Ϯ 0.2 mEq/l on control diet to 1.5 Ϯ 0.1 mEq/l). In WT mice, the TTKG decreased substantially [from 10.2 Ϯ 1.1 on control diet ( Table 2) to 1.6 Ϯ 0.4 on K/Mg-free diet (Table 3)], reflecting appropriate, maximal renal K conservation. In contrast, no renal K adaptation was seen with the ksKD mice (TTKG: 9.9 Ϯ 0.6 on control diet and 11.4 Ϯ 0.8 on K/Mg-free diet). The ksKD mice continued to show hypermagnesemia despite the K/Mg-free diet (Mg: 2.7 Ϯ 0.2 vs. 1.9 Ϯ 0.1 mEq/l for WT; Table 3). Furthermore, while a modest decrease in serum Mg was seen in WT mice on the K/Mg-free diet [from 2.2 Ϯ 0.1 mEq/l on control diet ( Table 2) to 1.9 Ϯ 0.1 mEq/l on K/Mg-free diet], the decrease was no greater for the ksKD mice on the K/Mg-free diet [from 2.9 Ϯ 0.2 mEq/l on control diet (Table 2) to 2.7 Ϯ 0.2 mEq/l on K/Mg-free diet], suggesting that renal Mg conservation is fully intact in the ksKD mice. Finally, the ksKD mice on the K/Mg-free diet again showed a concentration defect associated with hypernatremia, as was seen on control diet, but now also had significant weight loss relative to baseline weight before starting the diet (Ϫ19 Ϯ 1 vs. Ϫ2 Ϯ 1% for WT; Table 3). This may reflect increased urinary losses due to the superimposed effects of hypokalemia seen with the K/Mg-free diet (52). pNCC expression increases in WT but not ksKD mice on K/Mg-free diet. NCC expression has been shown to play a homeostatic role in renal K adaptation. On low-K diet, NCC expression increases (16,72,74), reducing distal Na delivery and limiting Na-dependent K secretion in the CNT and CCD. Consistent with this, increased pNCC expression was seen with WT mice on the K/Mg-free diet but not with ksKD mice (Fig.  5A). This suggests that failure to increase expression of pNCC, the activated form of the transporter (46,54), underlies the K wasting seen in the ksKD mice on K/Mg-free diet. We did not, however, see an increase in total NCC expression in either WT or ksKD mice on K/Mg-free diet (Fig. 5B). A possible explanation for this is that total NCC expression has been shown to decrease on Mg-deficient diet (75). Thus the lack of Mg in the combined K/Mg-free diet may have prevented observing the expected increase in total NCC expression in the WT mice with K restriction.

gKO Model
The gKO mice showed hypokalemia and hypomagnesemia, associated with increased urinary K and Mg losses, as well as a trend toward increased serum bicarbonate level that just missed statistical significance. Scant blood volumes in these 3to 7-day-old pups limited further analysis but significantly elevated serum bicarbonate levels were demonstrated in the ksKD mice (Table 2), where only partial ablation of Kir4.1 occurred, and hypocalciuria has been reported previously in gKO mice (6). Thus the gKO mouse appears to recapitulate the cardinal features of SeSAME/EAST syndrome. We also found decreased NCC expression in the gKO mouse [as previously reported by our group and others (20,83)], which provides a mechanism for the observed hypokalemia since reduced NCC function would be expected to result in increased Na-dependent K secretion distally in principal cells of the CNT and CCD.
Greater understanding of the molecular mechanism coupling the loss of basolateral Kir4.1 function with decreased apical NCC function has recently emerged. It has been proposed that loss of Kir4.1 in DCT would inhibit basolateral Na/K-ATPase function due to reduced K recycling as well as depolarize the basolateral membrane, leading to reduced basolateral Cl reabsorption through ClC-Kb channels, resulting in increased intracellular [Cl] (61). This, in turn, would be anticipated to increase Cl-sensitive inhibition of WNK/SPAK signaling (4,72) and ultimately to reduced NCC surface expression. Indeed, Zhang et. al. (83) showed that loss of Kir4.1 function in the gKO mouse led to marked reduction in DCT whole cell K current, membrane depolarization, decreased SPAK expression and ultimately reduced NCC expression, lending support to this model.

ksKD Model
Lethality by 2-3 wk of age limits the usefulness of the gKO model, which motivated us to pursue generating a ksKD mouse. The resultant mice had Kir4.1 protein expression that was decreased~50% vs. WT ( Fig. 2A). On the control diet (Table 2), the ksKD mice showed increased serum bicarbonate, relative hypercalcemia, and a urinary concentrating defect associated with hypernatremia. Serum K was normal and serum Mg was mildly increased despite a moderate reduction in NCC expression (Fig. 3). On K/Mg-free diet however, marked hypokalemia was seen (serum: K 1.5 Ϯ 0.1 vs. 3.0 Ϯ 0.1 mEq/l for WT) associated with renal K wasting (Table 3). Serum Mg remained mildly increased despite the dietary restriction. Furthermore, the increase in pNCC expression seen in WT mice on K/Mg-free diet was not seen with ksKD mice (Fig. 5A), suggesting that the reduced Kir4.1 expression in these mice leads to loss of this K conserving mechanism, resulting in the observed hypokalemia.
The reason hypokalemia and hypomagnesemia were not observed on control diet is likely due, at least in part, to the fact that the ksKD mice have significant residual Kir4.1 expression (~50%, Fig. 2A) in contrast to the gKO mice, where there is complete ablation of Kir4.1, and the SeSAME/EAST syndrome, where mutant Kir4.1 channels have minimal residual function (51,57,80). Nevertheless, given the moderately reduced NCC expression here (~40%, Fig. 3), one might have expected to see hypokalemia (and hypomagnesemia). However, even complete ablation of NCC, as in the NCC knockout mouse, yields normokalemia (with hypomagnesemia) on control diet; hypokalemia is seen only on low K diet (42,62). This suggests that normal dietary K intake is sufficient to offset renal K losses associated with reduced NCC function. In addition, compensation in other nephron segments and in the GI tract may be acting to mitigate the K (and Mg) losses.
Hypernatremia is seen in the ksKD mice on control diet, presumably due to the marked concentrating defect (Table 2), and appears to be even more pronounced on K/Mg-free diet (Table 3) where the superimposed hypokalemia (not seen with control diet) may be exacerbating the free water losses (52). A trend toward hypernatremia is also seen with the gKO pups, which have a more subtle concentrating defect ( Table 1), suggesting that the defect progresses over time. The concentrating defect in the ksKD mice on control diet is associated with decreased AQ-3 expression (Fig. 4B). AQ-3 is the main basolateral water channel isoform in the CNT and CCD, and its genetic ablation has been shown to result in nephrogenic diabetes insipidus (36). That AQ-2 is not also decreased (and in fact shows a suggestion of increased, possibly compensatory, surface expression) argues against this being simply a nonspecific loss of distal nephron function. If reduced AQ-3 expression is indeed responsible for the observed concentrating defect, how might this result from the decreased Kir4.1 expression in these mice? AQ-3 and Kir4.1 are both localized to the basolateral surface of principal cells so physical interaction is possible. Indeed, it has been proposed, based on immunoprecipitation experiments, that Kir4.1 and the AQ-4 isoform are tethered together via a dystrophin-glycoprotein multiprotein complex in retina (8) although the functional significance of this has been questioned (55).
We cannot completely rule out a reduction in medullary concentration also contributing to the concentrating defect. While total NKCC2 expression did not change in the ksKD mice [although the lower molecular weight (presumed monomeric) band did have decreased signal], it is certainly possible that changes in TAL function occur in these mice independent of NKCC2 expression, i.e., via changes in other TAL transporters. However, it is likely that loss of TAL function would also impair paracellular Mg (and Ca) uptake in this segment where the majority of filtered Mg is reabsorbed (48), leading to hypomagnesemia. That, in fact, hypermagnesemia (and hypercalcemia) is seen in the ksKD mice, even on K/Mg-free diet, makes significant disruption of TAL function less likely. In addition, Kir4.1 has not been detected in the medulla and loss of Kir4.1 in cortical TAL does not lead to apparent functional changes in gKO mice (82), although see Ref. 9.
The functional significance of the concentrating defect in humans is unclear. Gitelman's syndrome is associated with polyuria (65), but a concentrating defect does not occur in these patients or in the NCC knockout mouse (42). An overt concentrating defect was not apparent in SeSAMAE/EAST children (U osm : 501-834 mosmol/kgH 2 O, n ϭ 5) under basal conditions (6). Note, however, that while the concentrating defect was dramatic in the adult ksKD mice, it was subtle in the gKO pups. Thus a concentrating defect in the SeSAME/EAST children may be apparent under water deprivation conditions or in any patients that survive past childhood.
The marked hypokalemia seen in the ksKD mice on K/Mgfree diet likely arises from loss of Kir4.1 function in the DCT. In the DCT, NCC expression is modulated in response to alterations in dietary K. On a low-K diet, for example, NCC expression in WT rodents has been shown to increase (14,72,74), presumably reducing distal Na-dependent K secretion in the CNT and CCD. New insights into the role of Kir4.1 in mediating this physiologic K homeostasis have recently come to light. Based on a well-conceived series of experiments, Terker et al. (73) proposed that Kir4.1 acts as a "K sensor" in the DCT, where hypokalemia, for example, will lead to basolateral DCT membrane hyperpolarization, increased basolateral Cl reabsorption, reduced intracellular [Cl], and increased WNK4-associated NCC phosphorylation, resulting in increased DCT Na reabsorption and ultimately to appropriately reduced Na-dependent K secretion in the CNT and CCD. Thus, while pNCC expression increases in WT mice on K/Mg-free diet, the failure of this to occur with the ksKD mice (Fig. 5A) likely underlies the observed hypokalemia.
In addition, it is also possible that loss of Kir4.1 function in the CNT and CCD contributes to the impaired K conservation. Although several different K channels have been detected on the basolateral surface of principal cells (21,24,76), a predominant conductance consistent with Kir4.1/Kir5.1 has been reported in mouse (29). Furthermore, an 8-mV depolarization of the resting potential of gKO principal cells as well as increased ␤and ␥-ENaC expression was recently reported, both of which may contribute to increased K loss in the CNT and CCD (69). We have not yet tested for K wasting in ksKD principal cells.
In comparing the gKO and ksKD models, note that Kir4.1 expression in the gKO heterozygote pups is~30% of WT (Fig.  1A) while expression in the adult (ksKD) heterozygotes is similar to WT ( Fig. 2A). This suggests that Kir4.1 expression may be developmentally regulated. In addition, it is striking that the gKO heterozygote pups do not show electrolyte abnormalities (consistent with the recessive inheritance of SeSAME/EAST patients) despite a~40% reduction in NCC expression that is very similar to that of the ksKD mice, which do show abnormalities. A likely explanation for this apparent greater role of Kir4.1 in the adult than in the pups is that Na-dependent K secretion in the CNT/CCD is not yet significantly active in the first few weeks of life so that modulation of NCC expression may have minimal impact on distal K secretion. Indeed, it has been shown that Na reabsorption in 2-wkold rabbit CCDs is 60% of adult rates and that K secretion is absent until the 4 wk of life (60). Furthermore, ultrastructural studies have shown immaturity in terms of reduced number of organelles, mitochondria and basolateral infoldings in the CCD in early postnatal life (12). Although the distal nephron thus appears relatively insensitive to modulation of NCC by Kir4.1 in the first few weeks of life, complete absence of Kir4.1, and the associated marked reduction in NCC, as in the gKO mice, do have significant phenotypic implications.
In contrast to the hypomagnesemia seen in the gKO mice and SeSAME/EAST patients, we found hypermagnesemia in the ksKD mice on both control diet and, remarkably, even on K/Mg-free diet. We considered that the significant residual Kir4.1 expression in the ksKD mice may have prevented us from observing the hypomagnesemic phenotype. However, a recent report (Ref. 9, see below) of an inducible ksKD mouse in which essentially all the Kir4.1 expression was ablated in adult mice showed normomagnesemia. This raises the possibility that loss of Kir4.1 function has a greater impact on Mg transport in the developing mouse or human than in the adult. That the ksKD mice were actually hypermagnesemic, rather simply normomagnesemic, may be due to mild volume depletion (as evidenced by hypernatremia; Table 2) associated with the concentrating defect in these mice, which may have led to increased Mg reabsorption in the loop of Henle (49). The apparent metabolic alkalosis in the ksKD mice may also have contributed to increased urinary Mg reabsorption (49). The relative hypercalcemia seen in the ksKD mice (Table 2) may also reflect increased Ca reabsorption in the setting of mild volume depletion (5).
While this article was in revision, a report of an inducible, kidney-specific Kir4.1 knockout mouse was published (9). Nearly complete ablation of Kir4.1 protein expression in adult mice was achieved resulting in hypokalemia on control diet. That hypokalemia in our model was only seen on K/Mg-free diet likely reflects the significant degree of residual Kir4.1 protein expression in the ksKD mice. Cuevas et al. (9) found increased pNCC and total NCC expression in WT mice on K restricted diet that was not seen in the knockout mice, consistent with our findings. (As discussed above, the lack of increased total NCC expression with K deficiency in our model likely reflects our use of the combined K/Mg-free diet.) In addition, hypomagnesemia was not seen (consistent with our findings) and mild dehydration was likely present, as evi-denced by increased urine volume and hematocrit. The more severe dehydration we saw, associated with frank hypernatremia and tubulocystic dilations, may reflect the increased chronicity in our mice (5-9 mo vs. up to 1 mo of increased free water losses). Clearance studies showed an impaired response to hydrochlorothiazide and furosemide and increased ENaCdependent Na reabsorption in the inducible knockout mice. Finally, decreased total NKCC2 expression was seen, whereas we saw only a reduction of a lower (~150 kDa) NKCC2 band. Again, this difference may result from the more mild reduction in Kir4.1 expression in our model. The study by Cuevas et al. lends strong support for the role of Kir4.1 as a DCT K sensor, regulating distal Na delivery. The work here demonstrates qualitatively similar effects with more mild, physiological changes in Kir4.1 expression.

Conclusion
Even modest reduction in Kir4.1 expression results in impaired K conservation, which appears to be mediated by reduced expression of activated NCC.

ACKNOWLEDGMENTS
We acknowledge Dr. Lise Bankir (INSERM, Paris, France) for very helpful physiological discussions and input regarding the concentration defect in these mice.

GRANTS
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants K08-DK-069346 and K08-DK-069346 -06S1 (to D. A. Gray) and R01-DK-099284 (to T. Wang).

DISCLOSURES
No conflicts of interest, financial or otherwise, are declared by the authors.