The WNK-SPAK-NCC cascade revisited: WNK1 stimulates the activity of the NaCl cotransporter via SPAK, an effect antagonized by WNK4

Abstract

The With No lysine kinases WNK1 and WNK4 are key regulators of blood pressure. Their mutations lead to Familial Hyperkalemic Hypertension (FHHt), associated with an activation of the Na-Cl cotransporter, NCC. Although it is clear that WNK4 mutants activate NCC via SPAK (Ste20 Proline-Alanine rich Kinase), the mechanisms responsible for WNK1-related FHHt and alterations in NCC activity are not as clear. We tested whether WNK1 modulates NCC through WNK4, as predicted by some models, by crossing our recently developed WNK1-FHHt mice (WNK1^+/FHHt) with WNK4^−/− mice. Surprisingly, the activated NCC, hypertension, and hyperkalemia of WNK1^+/FHHt mice remain in the absence of WNK4. We demonstrate that WNK1 powerfully stimulates NCC in a WNK4-independent and SPAK-dependent manner. Moreover, WNK4 decreases the WNK1 and WNK3-mediated activation of NCC. Finally, the formation of oligomers of WNK kinases through their C-terminal coiled-coil domain is essential for their activity towards NCC. In conclusion, WNK kinases form a network in which WNK4 associates with WNK1 and WNK3 to regulate NCC.

INTRODUCTION

Arterial hypertension is now the largest contributor to premature morbidity and mortality worldwide¹. Although increased renal salt reabsorption is recognized as a crucial pathogenic factor, the underlying pathophysiological mechanisms remain unknown in most patients. In order to identify pathways, the disturbance of which could cause hypertension, one of the strategies used is the genetic analysis of rare monogenic hypertensive or salt-losing disorders, such as Familial Hyperkalemic Hypertension (FHHt; OMIM 145260; Gordon’s syndrome or pseudohypoaldosteronism type II). FHHt is a rare disease of hypertension, hyperkalemia and metabolic hyperchloremic acidosis². Genetic analysis of FHHt patients showed that the disease results from mutations in genes encoding the serine/threonine kinases WNK1 or WNK4 (With No lysine (K) 1 and 4)³ and components of an ubiquitin-ligase complex, KLHL3 (Kelch-like 3) and Cullin3⁴. These proteins seem to belong to the same blood pressure regulating pathway, as WNK1 and WNK4 are recruited by KLHL3 as substrates for ubiquitination by the Cul3-based E3 RING-type ubiquitin-ligase complex⁵.

The high sensitivity of patients to thiazides suggested that FHHt was caused by an activation of NCC, the Na⁺-Cl⁻ transporter of the Distal Convoluted Tubule (DCT) inhibited by this class of diuretics. Subsequent to the discovery of WNK kinases, the efforts of several groups focused on characterizing the mechanisms underlying the regulation of NCC by WNK1 and WNK4. The results were however often contradictory and did not allow the establishment of a clear and unifying model to explain how WNKs regulate the expression and/or activity of the co-transporter.

We first reported that WNK4 inhibits NCC activity in Xenopus laevis oocytes^{6, 7}, an effect reversed by angiotensin II⁸. This inhibitory effect was also observed in mouse DCT cells⁹, endogenously expressing WNK4 and NCC, suggesting that it is not an overexpression artifact. The mechanisms by which WNK1 regulates NCC are less well defined. Biochemical studies showed that the catalytically active WNK1 isoform (L-WNK1) activates the Ste20 Proline-Alanine rich Kinase SPAK^{10, 11}, responsible for phosphorylating and activating NCC directly¹². However, these data were not supported by studies performed in Xenopus oocytes. In this system, L-WNK1 failed to activate NCC but relieved NCC from WNK4-mediated inhibition⁷, suggesting that L-WNK1 lies upstream of WNK4 to activate NCC. The apparent discrepancies between results obtained using different model systems have generated substantial confusion concerning WNK effects.

In order to identify the pathway(s) involved in vivo in WNK1-dependent activation of NCC, we generated a mouse model of FHHt caused by WNK1 mutations (WNK1^+/FHHt mice)¹³. This model exhibited a full FHHt phenotype, resulting from an increased expression of L-WNK1 in the DCT and, to a lesser extent, in the CNT. As expected, the abundance of total and phosphorylated NCC at the apical membrane of DCT cells was enhanced. We also showed that phosphorylated SPAK abundance is increased near the apical membrane of DCT cells of mutant mice. These data support the existence of a WNK1/SPAK/NCC cascade, but do not indicate whether SPAK/NCC is stimulated directly by WNK1, or whether the effects result from relief of WNK4 inhibition.

The present study therefore tested the role of WNK1-WNK4 interaction in vitro and in vivo. Our results provide evidence that WNK1 is a powerful activator of NCC, independent of WNK4. In addition, we show that WNK4 antagonizes the activating effect of WNK1 on the cotransporter. Together, these results help resolve existing controversies and elucidate the mechanisms by which L-WNK1 and WNK4 signal to NCC.

METHODS

A detailed Methods section is available in the online Data Supplement at http://hyper.ahajournals.org.

RESULTS

WNK1-FHHt mice retain the full FHHt phenotype in the absence of WNK4

Mutations identified at the WNK1 locus in FHHt patients are large deletions of the first intron³. We recently developed a mouse model for WNK1-FHHt by deleting the first intron of mouse WNK1 (WNK1^+/FHHt mice)¹³. This model recapitulates the FHHt phenotype, with elevated arterial pressure, hyperkalemia and hyperchloremic metabolic acidosis. We used these mice to test the interaction between WNK1 and WNK4 in vivo, by crossing the WNK1^+/FHHt animals with WNK4^−/− mice. The latter displayed decreased abundance of total and phosphorylated NCC, associated with hypochloremic metabolic alkalosis and increased renin and hematocrit^{14, 15}. If WNK1 can activate SPAK and NCC, independent of WNK4, as suggested by biochemical studies, the phenotype of WNK1^+/FHHt might persist when WNK4 is inactivated. As shown in Table 1 and Figure 1, WNK1^+/FHHt:WNK4^−/− animals retained the FHHt phenotype, including elevated systolic blood pressure (WNK1^+/i1lox: 98 ± 2.6, WNK1^+/FHHt: 109 ± 3 and WNK1^+/FHHt:WNK4^−/−: 112 ± 8, P<0.05). The increased extracellular volume was also evidenced by a decreased renin expression in both WNK1^+/FHHt and WNK1^+/FHHt:WNK4^−/−. Moreover, the increased NCC expression and phosphorylation seen in WNK1^+/FHHt mice were still present in the double mutants (Figure 1B). These data show that WNK1 mutations cause FHHt by increasing the level of expression of L-WNK1 in the distal nephron, which then activates NCC independently of WNK4. As prior studies by others and ourselves^{7, 16} found that L-WNK1 did not regulate NCC directly in Xenopus oocytes, we decided to revisit the effects of L-WNK1 in this system and in cells.

Table 1. WNK1^+/FHHt:WNK4^−/− mice still exhibit FHHt metabolic abnormalities.

Blood was sampled from the retro-orbital sinus and blood gas and electrolytes were measured using a Radiometer® ABL80 Flex analyzer.

Plasma	WNK4−/− (n=4-5)	Controls (n=14-22)	WNK1+/FHHt (n=15-29)	WNK1+/FHHt:WNK4−/− (n=13)
Na+ (mmol/L)	144.5 ± 0.5	147.4 ± 0.4	147.7 ± 0.3	149.0 ± 0.4*
K+ (mmol/L)	3.9 ± 0.1	4.1 ± 0.1	4.9 ± 0.1***	5.0 ± 0.1***
Cl− (mmol/L)	107.0 ± 0.7***	111.1 ± 0.4	114.6 ± 0.3***	115.3 ± 0.5***
pH	7.39 ± 0.01**	7.32 ± 0.01	7.28 ± 0.01**	7.28 ± 0.02*
HCO3− (mmol/L)	27.2 ± 1.5**	24.6 ± 0.4	22.7 ± 0.4*	23.6 ± 0.6
Hematocrit	47.2 ± 0.5**	42.8 ± 0.6	39.7 ± 0.5***	39.3 ± 0.6**

^*P<0.05,

^**P<0.01 and

^***P<0.001 compared to controls (One-way ANOVA followed by Dunnet’s multiple comparisons test).

Figure 1. L-WNK1 induces hyperkalemic hypertension and NCC phosphorylation independently of WNK4 in mice.

(A) Renin transcripts level was measured by quantitative real-time PCR in the kidney cortex of WNK4^−/− (n=4), WNK1^+/i1lox (n=5), WNK1^+/FHHt (n=6) and WNK1^+/FHHt:WNK4^−/− (n=3) males. Results are expressed in arbitrary units relative to the expression of ubc. The control group (WNK1^+/i1lox mice) has been arbitrarily set to 1. Data are mean ± s.e.m. *: P<0.05 compared to the control group. (B) Left: Immunoblots performed on membrane-enriched fraction of renal cortex dissected from WNK4^−/− (n=4), WNK1^+/i1lox (n=6), WNK1^+/FHHt (n=6) and WNK1^+/FHHt:WNK4^−/− (n=5) males, incubated with antibodies against anti-NCC or its T53 phosphorylated residue (pNCC-T53). Right: Densitometric analysis of the immunoblots. Data, expressed as a percentage of the average signal intensity measured in the control group, set arbitrarily to 100%, are mean ± s.e.m. *: P<0.05, **: P<0.01 and ***: P<0.001 compared to the control group (Student’s unpaired t-test).

The kidney-enriched L-WNK1 isoform activates NCC in a SPAK-dependent manner

Several isoforms of WNK1 are generated by alternative splicing of eight exons in humans and mice, and we showed that the isoform lacking only exon 11 (L-WNK1-Δ11) is the major variant in the nephron (about 70% of all WNK1 isoforms)¹⁷. This isoform was, however, not the one used in the studies aimed at characterizing the regulation of NCC by L-WNK1⁷, which used the isoform lacking both exons 11 and 12¹⁸. Therefore, the lack of L-WNK1 effect on NCC in expression systems could be due to the use of this isoform. We tested the effect of the L-WNK1 variants upon NCC activity in Xenopus oocytes and HEK-293 cells. L-WNK1-Δ11 induced a 3.25 fold increase in NCC activity (Figure 2A), associated with increased membrane abundance of total and phosphorylated NCC (pNCC; Figure S1). With the exception of L-WNK1-Δ9, variants containing all exons or a combination of the alternatively spliced exons 9, 11 and 12 also activated NCC, but significantly less than L-WNK1-Δ11 (Figure 2A). This effect was kinase dependent as it was absent when WNK1 D368A, known to lack kinase activity¹⁸, was employed (Figure 2B). Previous studies did not observe the direct activation of NCC by the rat L-WNK1-Δ11-12 variant first because it is less potent than L-WNK1-Δ11 (Figure 2A) and, second, because it carries a previously unsuspected mutation in the C-terminal domain, identified by sequencing (G2120S; Figure S2A). This mutation reduced the effect of L-WNK1 on the cotransporter. Replacing the serine residue by a glycine in the rat L-WNK1-Δ11-12 cDNA restored NCC activation (Figure S2B), while mutating the glycine into a serine in the human L-WNK1-Δ11 cDNA prevented NCC activation (Figure S2C).

Figure 2. L-WNK1 is a powerful activator of NCC activity in Xenopus laevis oocytes.

(A) NCC activity was measured in Xenopus oocytes overexpressing NCC alone or in combination with different L-WNK1 splicing variants. All L-WNK1 variants, except the one deleted from exon 9 (L-WNK1-Δ9), stimulated NCC activity. (B) Effect of L-WNK1-Δ11 point mutations affecting its kinase activity (Δ11-DA), SPAK binding (Δ11-F316A), dimer formation (Δ11-HQ) or both (Δ11-F316A-HQ) on NCC activity. For (A) and (B), NCC activity in oocytes injected with NCC cDNA alone (black bar) was arbitrarily set to 100%. Data are mean ± s.e.m. ***: P<0.001 and ****: P<0.0001 vs. NCC alone (Student’s unpaired t-test). (C) Immunoblots from NCC-HEK293 cells transiently transfected with NCC and human L-WNK1-Δ11-myc constructs, and probed with antibodies indicated. Data, indicated below the blots, are mean ± s.e.m. **: P<0.01 and ***: P<0.001 vs. the control (Student’s unpaired t-test). (D) The inhibition of L-WNK1 expression by RNA-interference decreased NCC phosphorylation. Immunoblots from NCC-HEK-293 cells transfected with control and WNK1-specific siRNA, probed with the antibodies indicated.

In order to test the requirement for SPAK in L-WNK1-mediated activation of NCC, we disrupted the L-WNK1/SPAK interaction by eliminating a SPAK-binding motif in L-WNK1-Δ11. Disruption of a similar motif in WNK3 blocks its ability to activate NCC¹⁹. This mutant (L-WNK1-Δ11-F316A) not only failed to stimulate NCC, but also inhibited the co-transporter (Figure 2B). Because oocytes express L-WNK1²⁰ and L-WNK1 oligomerizes²¹, we hypothesized that L-WNK1-Δ11-F316A exerted this effect through interactions with endogenous L-WNK1. Thastrup and colleagues showed that a small HQ motif (HIQEVVSLQT), located in the C-terminal coiled-coil domain, is essential for the formation of homo and heteromers of WNK kinases²². The replacement of the HIQEVVSLQT sequence by AIQEVVSLAT impairs the interaction of WNK kinases²². Introduction of these mutations into L-WNK1-Δ11 (Δ11HQ/AA) prevented the activation of NCC. Moreover, mutation of the HQ motif of L-WNK1-Δ11-F316A abrogated its inhibition (Δ11-F136A-HQ/AA; Figure 2B). This suggests that basal NCC activity in vitro is dependent on endogenous L-WNK1 and that L-WNK1-Δ11-F316A acts as dominant negative form. To confirm that basal NCC activity in vitro requires endogenous L-WNK1 activity, we used NCC transfected mammalian kidney cells, which express, glycosylate, and phosphorylate NCC in a manner that resembles mammalian kidney (Figure S3). Overexpression of human L-WNK1-Δ11 increased the abundance of total and phosphorylated NCC (Figure 2C). Conversely, L-WNK1 siRNA-mediated knockdown (Figure 2D) reduced pNCC abundance. These results demonstrate that kidney-enriched isoform of L-WNK1 is a powerful activator of NCC through a WNK4-independent mechanism, since WNK4 is not expressed in HEK293 cells (Figure S3).

L-WNK1 effect on NCC is antagonized by WNK4

WNK4 inhibits NCC in Xenopus oocytes^{6, 7}, mammalian cells²³ and mice²⁴. Additionally, WNK4 prevents the positive effect exerted by WNK3 on NCC²⁵ (Figure 3A). We observed the same effect when WNK4 and L-WNK1-Δ11 were co-expressed, in Xenopus oocytes and HEK293 cells (Figure 3A-B). The decreased NCC activity observed when WNK4 and L-WNK1Δ11 are co-expressed could result from the inhibition of L-WNK1-Δ11 by WNK4 or from the parallel action of an activator (L-WNK1-Δ11) and an inhibitor (WNK4). The former could be caused by a reduced L-WNK1-Δ11 abundance mediated by WNK4. Figures 3C and 3D show that WNK4 significantly reduced the abundance of L-WNK1-Δ11 in both oocytes and HEK293 cells. The effects to reduce L-WNK1 were not simply the result of an overexpression artifact. Indeed, overexpression of WNK3 did not modify L-WNK1-Δ11 (Figure S4A) and overexpression of WNK4 did not affect SPAK or pSPAK abundance in HEK293 cells (Figure S4B).

Figure 3. WNK4 inhibits the activation of NCC by L-WNK1 and WNK3.

(A) NCC activity was measured in Xenopus oocytes injected alone or in combination with L-WNK1-Δ11 (WNK1), WNK3 and/or WNK4 as stated. WNK4 not only inhibited NCC activity but also prevented the L-WNK1-Δ11 or WNK3-dependent stimulation of NCC (N=15 independent experiments per combination). (B) Immunoblots from NCC-HEK293 cells transiently overexpressing L-WNK1-Δ11 and WNK4. WNK4 overexpression prevented the activation of NCC by L-WNK1-Δ11. (C-D) Immunoblots from NCC HEK293 cells transiently transfected with myc-tagged L-WNK1-Δ11 and WNK4 constructs (C) or from Xenopus oocytes injected with myc-tagged L-WNK1-Δ11 alone or in combination with WNK4 (D). WNK4 overexpression significantly decreased abundance of L-WNK1-Δ11 in both systems. *: P<0.05 and **: P<0.01 for the comparison between L-WNK1-Δ11 alone and L-WNK1-Δ11 + WNK4 (Student’s unpaired t-test). (E) NCC activity alone or in combination of wild type or mutant-HQ WNKs as stated. The inhibition of the formation of WNKs dimers by mutating the C-terminal HQ motif relieved L-WNK1-Δ11 or WNK3 from WNK4-mediated inhibition. For oocytes experiments, the activity measured in oocytes injected with NCC alone was arbitrarily set to 100% and data are mean ± s.e.m. **: P<0.01 for the comparison between wild type WNK4 and WNK4-HQ (Student’s unpaired t-test).

We next assessed the requirement for the formation of WNK4 oligomers by mutating the C-terminal HQ motif²². The resulting WNK4-HQ/AA mutant lost the capacity to inhibit NCC (Figure 3E) and precluded the ability of WNK4 to inhibit L-WNK1- or WNK3-induced activation of NCC in oocytes (Figure 3E). These functional observations indicate that interactions of WNKs kinases via the HQ motif are necessary for their effects on NCC.

DISCUSSION

Since the discovery of WNK kinases and their relation to FHHt, an inherited disease featuring arterial hypertension, a great deal of effort has been directed at understanding how WNK kinase mutations produce the disease. As all FHHt symptoms are corrected by a thiazide treatment and as the expression of the Na-Cl cotransporter NCC is increased in FHHt mouse models, several groups aimed at understanding how the WNK kinases modulate NCC activity. However, the results thus obtained failed to generate an unifying model. In the present study, we show data supporting a revised model of WNK kinase action in signalling to NCC, in which WNK1 and WNK3 mediate activation of NCC through SPAK, and WNK4 antagonizes WNK1 and WNK3 (Figure 4).

Figure 4. Proposed new model for the regulation of NCC by WNKs.

Note that WNK3 is not shown and that L-WNK1 is indicated instead of L-WNK1-Δ11 to simplify the Figure. At baseline, L-WNK1-Δ11 forms homomers that activate NCC in a SPAK-dependent manner. WNK4 could antagonize the effect of L-WNK11-Δ11 on NCC by either forming inactive WNK4-WNK1 heteromers or inhibitory WNK4 homomers.

This model can explain several contradictory observations. First, the demonstration that L-WNK1 is a strong stimulator of NCC activity clarifies the pathophysiology of FHHt (Figure S5), since increased L-WNK1 expression in the DCT is sufficient to promote the development of the FHHt phenotype¹³. Second, it fits with the in vitro observations that L-WNK1 is a more powerful activator of SPAK than WNK4¹¹. Third, the new model provides a plausible explanation for the discrepancy between the mild phenotype consequent to WNK3 deletion in vivo, and the powerful ability of WNK3 to activate NCC in vitro^{25, 26}. Since both WNK1 and WNK3 can activate NCC, the increased WNK1 expression observed in WNK3^−/− mice likely compensates for the loss of WNK3^{27, 28}.

Previous studies failed to demonstrate the WNK1 effect on NCC for two reasons. First, those studies did not use the L-WNK1 isoform lacking exon 11, the most abundant isoform along the nephron¹⁷ and the most powerful NCC activator. Instead, the studies used an isoform lacking both exons 11 and 12, which represents only 20% of WNK1 isoforms in the kidney¹⁷. Second, the previously used rat L-WNK1-Δ11-12 cDNA carries a mutation in the C-terminal region of the protein that reduces its effect on NCC. This cDNA corresponds to the first isolated L-WNK1 cDNA, which was subcloned from a rat brain library¹⁸, where the variant missing both exons 11 and 12 is the predominant variant¹⁷.

Co-expression of L-WNK1-Δ11 and WNK4 results in a decreased NCC activity compared to L-WNK1-Δ11 alone. This could result from the inhibition of L-WNK1-Δ11 by WNK4 or from the parallel action of an activator (L-WNK1-Δ11) and an inhibitor (WNK4). WNK kinases can form homo or heteromers via a conserved carboxyl-terminal domain^{21, 29}. Co-immunoprecipitation experiments indicated that L-WNK1 interacts with both WNK4 and WNK3 in HEK293 cells through a C-terminal conserved HQ motif²². We show here that the interaction between L-WNK1-Δ11, WNK3 and WNK4 is essential for the regulation of NCC, as the mutation of their HQ motif abrogates the effect of all three kinases. Our study does not however allow us to determine which homo or heteromers are required for NCC regulation. The disruption of the interaction by mutating the HQ motif results in a decreased capacity of WNK3 to autophosphorylate and phosphorylate WNK1 on the T-loop residue (Ser382)²², required for WNK1 activation. Inhibiting the formation of activating WNK3 homomers and/or L-WNK1/WNK3 heteromers could therefore prevent the activation of NCC by L-WNK1 or WNK3 in oocytes. Similarly, preventing the formation of WNK4 homomers reduced its autophosphorylation. Interestingly, Thastrup and colleagues showed that, while WNK4 can interact with WNK1, it is unable to phosphorylate its T-loop residue²². The expression of WNK4 in oocytes could thus result in the formation of inhibitory WNK4 homomers and/or L-WNK1/WNK4 or WNK3/WNK4 heteromers, in which L-WNK1 or WNK3 is not activated (Figure 4). The expression of WNK4-HQ mutant in Xenopus oocytes would then restore NCC activity by preventing the formation of these inhibitory homomers and/or heteromers.

The present study reconciles in vivo and in vitro observations about the mode of L-WNK1 action towards NCC, but controversy still remains concerning intrinsic WNK4 activity. Most in vitro studies, including the present one, concluded that WNK4 is an inhibitor of NCC, either when overexpressed in Xenopus oocytes and mammalian cells^{6, 7, 23} or when endogenous WNK4 is knocked down in mDCT15 cells⁹. These results are supported by some but not all in vivo studies. Lalioti et al. first reported that WNK4 overexpression in transgenic mice results in Gitelman-like syndrome²⁴, caused by inactivating mutations of NCC, again suggesting that WNK4 inhibits the cotransporter. This conclusion is however not supported by two subsequent studies. First, WNK4 inactivation in mice also leads to the development of a mild Gitelman syndrome^{14, 15}. Second, Wakabayashi and colleagues observed that WNK4 overexpression caused FHHt in mice³⁰, even though they used the same transgene than the one used in Lalioti’s study. These two studies suggest that WNK4 is an activator of NCC, in vivo.

How could we reconcile the WNK4 inhibitory effect observed in vitro and in vivo with the activating effect observed in vivo? NCC inhibition by WNK4 in Xenopus can be abrogated by angiotensin II⁸. Similarly, the phosphorylation of SPAK by WNK4 is stimulated by an increased Ca²⁺ concentration³¹. One could therefore speculate that WNK4 inhibitory and stimulatory actions coexist and can be modulated by intracellular signals and/or that the environment of in situ DCT cells at baseline favours the activating mode of WNK4 while the culture conditions of oocytes or mammalian cells mostly favours its inhibitory mode.

Perspectives.

We present evidence that the kidney-enriched isoform of L-WNK1 stimulates NCC expression and activity in a SPAK-dependent manner and not through modulation of an inhibitory effect of WNK4, as previously reported. In addition, we show that WNK4 antagonizes WNK1- or WNK3-mediated activation of the co-transporter. A lot of questions, however, remain open before we can fully understand how the WNK kinases interact with each other to regulate the activity of the Na-Cl cotransporter. We believe that ones of the most important issues are to determine the WNKs relative expression level in the DCT and the characterization of the oligomers truly formed by WNK1, WNK3 and WNK4 in vivo.

Novelty and Significance.

Since the discovery of WNK kinases and their relation to FHHt, an inherited disease featuring arterial hypertension, numerous studies have aimed at understanding how WNK kinase mutations produce the disease. Most of the FHHt phenotype is due to activation of NCC. Yet none of the models for WNK actions on NCC fully account for experimental results. In the present study, we show data supporting a revised model of WNK kinase action in signalling to NCC, in which WNK1 and WNK3 activate NCC through SPAK, and WNK4 antagonizes WNK1 and WNK3.