Targeted disruption of the Wnk 4 gene decreases phosphorylation of NaCl cotransporter , increases Na excretion and lowers blood pressure

We recently generated Wnk4 knockin mice and found that a major pathogenesis of pseudohypoaldosteronism type II was the activation of the OSR1/SPAK kinase-NaCl cotransporter (NCC) phosphorylation cascade by the mutant WNK4. However, the physiological roles of wild-type WNK4 on the regulation of Na excretion and blood pressure, and whether wild-type WNK4 functions positively or negatively in this cascade, remained to be determined. In the present study, we generated WNK4 hypomorphic mice by deleting exon 7 of the Wnk4 gene. These mice did not show hypokalemia and metabolic alkalosis, but they did exhibit low blood pressure and increased Na and K excretion under low-salt diet. Phosphorylation of OSR1/SPAK and NCC was significantly reduced in the mutant mice as compared with their wild-type littermates. Protein levels of ROMK and Maxi K were not changed, but epithelial Na channel appeared to be activated as a compensatory mechanism for the reduced NCC function. Thus, wild-type WNK4 is a positive regulator for the WNK-OSR1/SPAK-NCC cascade, and WNK4 is a potential target of anti-hypertensive drugs.


INTRODUCTION
Pseudohypoaldosteronism type II (PHAII) is an autosomaldominant disorder characterized by hyperkalemia, metabolic acidosis and hypertension (1 -3).In 2001, Wilson et al. found that WNK1 and WNK4 gene mutations cause PHAII (4).However, since the physiological function of WNK kinases had not been clarified, the molecular pathogenesis of PHAII was not clear at that time.Since PHAII was known to be caused by dysregulation of NaCl reabsorption and potassium excretion in the kidney (5), many studies have focused on how WNK kinases regulate renal channels and transporters (6 -10).Most of these initial studies used the Xenopus oocyte expression system (6 -10).The Na-Cl cotransporter (NCC) has been postulated as a major target of WNK kinases since PHAII is very sensitive to thiazide diuretics (5).Yang et al. (6) and Wilson et al. (7) found that wild-type WNK4 inhibited NCC function in the oocyte expression system, but a disease-causing mutant WNK4 lacked this inhibitory effect.On the basis of these data, they speculated that the pathogenesis of PHAII might come from a loss-offunction effect of the mutant.On the other hand, in previous studies with MDCK cells, we found that the inhibitory effect of WNK4 was not specific to NCC but was observed in other transporters not related to PHAII (11).Furthermore, the mutant WNK4 also retained its inhibitory effect in our experiments.Accordingly, as Richardson et al. (12,13) recently suggested, we decided to use genetically engineered mice instead of a heterologous overexpression system to clarify the physiological roles of WNK kinases.Recently, mutant WNK4 transgenic mice carrying two additional mutant WNK4 alleles were generated, and they exhibited the PHAII phenotypes (14), clearly indicating that the mutant WNK4 acts as a gain-of-function mutant rather than a loss-of-function mutant.We recently generated Wnk4 D561A/þ knockin mice, an ideal mouse model of PHAII, and found that the major pathogenesis of PHAII in vivo is the activation of the phosphorylation cascade of OSR1/SPAK kinase and NCC by the mutant WNK4 (15).Accordingly, the next question was whether the disease-causing mutant WNK4 was a dominant-active or -negative mutant.In other words, was wild-type WNK4 a positive or negative regulator for this cascade?This simple question has not yet been answered, but it is an important issue for further research on WNK kinases and also a crucial problem when considering the development of anti-hypertensive drugs focusing on WNK kinases.To answer this question, generation and analysis of Wnk4 null mice is necessary.However, homozygous mice in which the kinase domain of the Wnk4 gene was targeted showed embryonic lethality (unpublished observation).Therefore, in the present study, we targeted the Wnk4 gene at different exons and successfully generated WNK4 hypomorphic mice.These mice have lower blood pressure and a reduced NaCl reabsorption capacity as compared with their wild-type littermates, and they also have decreased OSR1/SPAK and NCC phosphorylation.

Characterization of the targeted allele of the Wnk4 gene
Since we suspected that WNK4 null mice might exhibit embryonic lethality, we planned to generate WNK4 hypomorphic mice.Accordingly, we designed a targeting vector to delete a single exon (exon 7) of the Wnk4 gene outside the kinase domain (Fig. 1).Homologous recombination was confirmed by PCR and Southern blotting in three ES cell lines, from which the heterozygous mutant mice were successfully generated (Fig. 2A).We observed no phenotypical difference among the three mouse lines.We selected one of the lines for further analyses, and examined the mutant WNK4 mRNA and protein produced in the kidneys of this line.To examine the mRNA, we prepared several primer sets flanking exon 7 and performed RT -PCR.As we expected, the deletion of exon 7 yielded the splicing from exon 6 to 8 (Fig. 2B).In addition, the smaller amount of transcript derived from the splicing from exon 6 to 9 was also detected.No other types of transcripts were detected by RT -PCR with other primer sets.We then used immunoblots to determine if these transcripts were translated into proteins in the kidney.For this purpose, we generated two new antibodies to WNK4 that recognized the amino and carboxy terminal portions of WNK4.The specificity of the antibodies was determined by antigen absorption tests (Supplementary Material, Figure S1).The transcript derived from the splicing from exon 6 to 8 had a frame-shift and should yield a C-terminally truncated WNK4 protein of 63 kD.The transcript derived from the splicing from exon 6 to 9 had no frame-shift, which should result in a 14 kD smaller protein lacking the middle portion of WNK4 (WNK4DEx7-8) than wild-type WNK4.Both antibodies detected a band of the same molecular weight in the mutant mouse kidney.The difference of apparent molecular weight of the bands between the mutant mice and the wild-type littermates were consistent with that estimated from the calculated molecular weight (14 kD) (Fig. 2C and D).On the basis of these observations, we concluded that the mutant mice expressed the WNK4DEx7-8 protein.The C-terminally truncated WNK4 protein was not detected in the mutant mice.The WNK4DEx7-8 protein was also detected by immunofluorescence.As shown in Figure 2E, the signal for WNK4DEx7-8 was co-localized with the signals for parvalbumin (marker of distal convoluted tubule 1) or calbindin-D28K (marker of distal convoluted tubule 2, connecting tubule and cortical collecting duct), similarly to the signal for wild-type WNK4.Co-localization with parvalbumin plus calbindin-D28K was also confirmed to be similar (data not shown).These results indicate that the intrar-   enal localization of WNK4DEx7-8 is the same as that of the wild-type WNK4.

Functional characterization of WNK4DEx7-8 protein
To determine if the mutant mice constituted a WNK4 hypomorphic model, we characterized the kinase activity of the WNK4DEx7-8 protein in vitro (HEK293 cell) and in vivo (kidney).Both autophosphorylation of WNK4 and phosphorylation of SPAK (the only physiological substrate of WNK4 identified thus far) were greatly reduced by the deletion (Fig. 3), confirming that the mutant mice were indeed WNK4 hypomorphic mice.

Blood and urine analyses
We observed no obvious differences between the WNK4 hypomorphic mice and the wild-type littermates in survival, gross physical appearance and organ morphology.There were no significant differences in the pH, K and HCO3-levels, plasma renin activity and aldosterone levels (Table 1).Hypocalciuria and hypomagnesemia previously observed in the Ncc null mice (16) were also not observed in the WNK4 hypomorphic mice (urine Ca: 0.071 + 0.007 mg/g Cre in the wildtype mice and 0.073 + 0.008 mg/g Cre in the hypomorphic mice, n ¼ 8; serum Mg: 1.80 + 0.06 mg/dl in the wild-type mice and 1.72 + 0.03 mg/dl in the hypomorphic mice, n ¼ 8).Urinary excretion of Na and K under normal condition was also not significantly affected in the WNK4 hypomorphic mice (Fig. 4A and B).To more thoroughly characterize the phenotypes, we fed the hypomorphic and wild-type mice low-salt (0.01% NaCl) and high-salt (4% NaCl) diets, and monitored urinary Na and K excretion, particularly focusing on the transition periods in changing diets.As shown in Figure 4A, Na excretion in the hypomorphic mice was significantly higher on days 5, 6 and 7 under the low-salt diets, and during the transitional period from low-to high-salt diet.On days 5, 6 and 7 under low-salt diets, wild-type mice reduced Na excretion to levels estimated to be equivalent to the reduced Na intake.However, the hypomorphic mice could not fully adapt to the reduced NaCl intake.At this phase, both groups of mice were presumably exhibiting their maximum capacity of Na conservation, and the difference in and the WNK4DEx7-8 protein were immunoprecipitated from the kidney with anti-WNK4 antibody generated against the N-terminus and used for a kinase assay (as described above).Little autophosphorylation of WNK4DEx7-8 was observed in the mutant mice, and phosphorylation of GST-SPAK was significantly reduced.this capacity was also evident when the mice were abruptly switched to high-salt diet.Although we could not detect any differences in the serum K levels between wild-type and hypomorphic mice, even under low-K diet (serum K: 2.94 + 0.14 mEq/l in wild-type mice and 3.20 + 0.20 mEq/l in hypomorphic mice, n ¼ 8), increased K excretion was also detected in hypomorphic mice in the metabolic cage study (Fig. 4B).

Blood pressure
We used radiotelemetric devices to measure blood pressure under physiological conditions.As shown in Figure 5, although no significant difference in blood pressure was detected in the resting period (day time) between the two groups, the systolic blood pressure of the hypomorphic mice was significantly lower than that of their wild-type littermates during the period of high-activity (118.

Reduced phosphorylation of OSR1 and NCC in the WNK4 hypomorphic mice
We investigated the phosphorylation status of the downstream targets of WNK4.As expected from the in vivo kinase assay (Fig. 3), phosphorylation level of OSR1 at the specific site phosphorylated by WNK kinase were significantly decreased in the WNK4 hypomorphic mice as compared with their wildtype littermates (Fig. 5A).Phosphorylations of NCC at T53, T58 and S71 were also dramatically reduced in the hypomorphic WNK4 mice, although the total amount of NCC was unchanged (Fig. 5B).Since in vitro studies reported that kidney K and Na channels were regulated by WNK4, we also investigated the protein levels of these channels by immunoblots.Protein levels of ROMK and maxi-K were not changed in the hypomorphic mice (Fig. 6A).As for the epithelial Na channel (ENaC), alpha-, beta-and gamma-ENaC plus cleaved alpha-and gamma-ENaC were significantly increased in the hypomorphic mice (Fig. 6B).

DISCUSSION
In this study, we successfully generated and analyzed WNK4 hypomorphic mice.The expression of WNK4DEx7-8 protein in the mutant mice prevented the embryonic lethality observed in total WNK4 knockout mice.We performed in vitro and in vivo kinase assays that demonstrated the decreased kinase activity of WNK4DEx7-8 protein.Furthermore, the phenotypes were only observed in the homozygous mice, not in the heterozygous mice (data not shown), indicating that the phenotypes are caused by the reduced activity of WNK4, not by some unexpected dominant-negative functions of the mutant WNK4 on other WNK kinases or OSR1/SPAK.Therefore, we regard the mutant mice as WNK4 hypomorphic mice.The generation of this mutant enabled us to investigate in vivo the consequence of reduced WNK4 activity in the kidney function.The hypomorphic mice did not show the typical phenotypes (hypokalemia, alkalosis, hypomagnesemia and increased renin activity and aldosterone) of human Gitelman syndrome under normal diet.However, even the Ncc-null mice did not completely mimic the physiological findings observed in Gitelman's patients (16).As the NCC protein was conserved and NCC phosphorylation was not completely absent in hypomorphic mice, the phenotypes were expected to be milder than those in Ncc-null mice.Nonetheless, we detected clear reductions in peak ability to conserve Na under Na-deprived and subsequent Na-loaded conditions (Fig. 4A).The difference in K excretion between hypomorphic and wild-type mice was more evident than that in Na excretion (Fig. 4B) on day 2, thus suggesting that the decreased function of NCC by in WNK4 hypomorphic mice increases Na delivery to the downstream nephron segment where Na/K exchange occurs.In addition, the lower blood pressure and decreased OSR1 and NCC phosphorylation in WNK4 hypomorphic mice (Figs 5  and 6) clearly demonstrate that WNK4 positively regulates NCC function in vivo through the OSR1/SPAK-NCC phosphorylation cascade.
The study of Wnk4 D561A knockin mice suggested that WNK-OSR1/SPAK-NCC constitutes a kinase cascade in in vivo kidney.However, whether WNK4(D561A) was acting as a dominant-active or -negative mutant in this cascade could not be determined, since the role of wild-type WNK4 in this cascade was not clarified.Our in vitro phosphorylation study (17) showed that WNK4 as well as WNK1 phosphorylates their substrates, OSR1 and SPAK kinases, suggesting that wild-type WNK4 sends a positive signal at least to OSR1/SPAK.Accordingly, it may be reasonable to speculate that wild-type WNK4 is a positive regulator of NCC.In the present study, we demonstrate that a reduction of WNK4 kinase activity in vivo leads to reduced NCC phosphorylation, lower blood pressure and reduced capacity of NaCl reabsorption.We previously showed that phosphorylation was important for the localization of NCC on the apical plasma membranes in the distal convoluted tubules, and Pacheco-Alvarez et al. (18) reported that phosphorylation itself was necessary for the functional activation of NCC.Whether the regulation of NCC by phosphorylation is a functional activation of the transporter itself or is mediated by intracellular sorting mechanisms or both, both data clearly indicates that the phosphorylation of NCC up-regulates its function in vivo.In this respect, we can conclude that wild-type WNK4 is a positive regulator for NCC at least through the WNK-OSR1/SPAK-NCC cascade.In contrast, other studies have suggested that wild-type WNK4 is a negative regulator for NCC.However, such data on NCC regulation by WNK kinases relied on overexpression systems including Xenopus oocytes (6,7) and transgenic mice (14), and the scheme of NCC regulation by WNK kinases was depicted in the absence of OSR1/SPAK kinases and phosphorylation of NCC [see reviews by McCormick (19) and Richardson (12)].Wilson et al. (7) reported that the inhibitory effect of wild-type WNK4 on NCC was dependent on WNK4 kinase activity.On the other hand, Yang et al. (6) showed that it was not kinase-dependent, instead it was mediated by the carboxy terminal portion of WNK4.If it is a kinase-dependent process, NCC must be up-regulated in the WNK4 hypomorphic mice since the mice had reduced WNK4 kinase activity.However, the phenotypes that we observed do not match this hypothesis, suggesting that the inhibitory function of WNK4 observed in oocytes occurs only minimally in vivo.However, if the inhibitory effect observed in oocytes is not dependent on WNK4 kinase activity, we must wait until conditional WNK4 null mice are generated to answer this question, since the carboxy terminal portion of WNK4 remains in the WNK4 hypomorphic mice.This situation is similar to determining if ENaC is a direct target of WNK4, since Ring et al. (9) reported that wild-type WNK4 inhibited ENaC activity in oocytes in a manner that was not kinase activity-dependent.However, we believe that the up-regulation of ENaC observed in the hypomorphic mice is not a primary effect of the WNK4 hypomorph; rather, it is a compensatory mechanism for the reduction of NCC function as long as the phenotypes in the hypomorphic mice are not hypertension but low blood pressure and NaCl-losing.Accordingly, we think that the activation of ENaC provides evidence that NCC function is indeed decreased in vivo.
Although the issue of whether WNK4 is a positive or negative regulator of NCC was settled, as described above, there are some unresolved issues regarding the phenotypes of hypomorphic mice.Increased K excretion was still observed after Na excretion reached a steady state under the low-salt diet (days 5 -7).Even if we consider ENaC activation in the hypomorphic mice, the difference in Na excretion may not have been sufficient to account for the differences in K excretion by Na/K exchange alone.This suggests that K excretion mechanism(s) other than Na/K exchange are increased directly or indirectly in WNK4 hypomorphic mice.With regard to the lower blood pressure in the hypomorphic mice, we were unable to observe further decreases with a low-salt diet.Thus, our observation period may not have been sufficiently long (up to 10 days under low-salt diet) to detect the difference.
WNK kinases have been thought to be target molecules for anti-hypertensive drugs since the discovery of mutations in the WNK kinases in PHAII.However, the kinase domains of WNK kinases (WNK1-4) are highly similar, making it difficult to develop isoform-specific inhibitors.If the effect of each WNK on NCC is variable (20,21) (i.e.WNK4 is inhibitory and WNK3 is stimulatory), inhibitors to WNK kinases may not be promising as anti-hypertensive drugs.In this respect, the present study, along with the study of the Wnk1 þ/2 mice (22), tells us that the inhibition of WNK kinases may be beneficial for reducing blood pressure.
In conclusion, the generation and analyses of the WNK4 hypomorphic mice clarified that WNK4 kinase activity is important for maintaining NCC function by phosphorylation.

Targeted disruption of the Wnk4 gene
To perform targeted disruption of the Wnk4 gene, we prepared the targeting vector by using PCR-amplified segments of the Wnk4 gene after verifying the sequences (Fig. 1).The targeting vector was then transfected into J1 ES cells by electroporation, as described previously (23).After selection with 150 mg/ml G418 and 2 mM ganciclovir, targeted embryonic stem cell clones were selected by PCR and Southern blotting.Chimeric male mice were bred with C57BL/6 females to produce the heterozygous floxed mice, and the neo cassette was then deleted by crossing the mice with Cre recombinase-expressing transgenic mice (24).This experiment was approved by the Animal Care and Use Committee of Tokyo Medical and Dental University.

Blood pressure measurements
We measured blood pressure by a radiotelemetric method (25) in which a blood pressure transducer (Data Sciences International, St Paul, MN, USA) was inserted into the left carotid artery.Ten days after transplantation, each mouse was housed individually in a standard cage on a receiver under a 12-h light-dark cycle.Systolic and diastolic blood pressure, heart rate and activity were recorded every minute by radiotelemetry.Mice showed alternating periods of highactivity (10 p.m. -4 a.m.) and low-activity (10 a.m -4 p.m.), as reported previously (26).For each mouse, we measured blood pressure values for more than three consecutive days and calculated the mean + SEM of all values during both high-and low-activity periods.

Blood and urine measurements
Blood for electrolyte analyses was obtained from the retroorbital sinus under light ether anesthesia.Electrolyte levels were determined with an i-STAT analyzer (Fuso, Osaka, Japan).The serum aldosterone level and plasma renin activity were measured by the SRL Clinical Laboratory Service (Tokyo, Japan).Low-salt (0.01% NaCl), high-salt (4% NaCl) and low-potassium (0.03%) diets were obtained from Oriental Yeast Co., Ltd (Tokyo, Japan).Urine samples were analyzed by DRI-CHEM (Fujifilm, Tokyo, Japan).

Immunoblotting and immunofluorescence
Semiquantitative immunoblotting was performed as described previously (15) by using whole kidney homogenate without the nuclear fraction (600g) or the crude membrane fraction (17 000g).The intensity of bands was analyzed by using Image-Gauge software (Fujifilm).For immunofluorescence staining, kidneys were fixed by perfusion through the left ventricle with periodate lysine (0.2 M) and paraformaldehyde (2%) in PBS.Tissue samples were soaked for several hours in 20% sucrose in PBS, embedded in Tissue-Tek OCT Compound (Sakura Finetechnical Co., Ltd, Tokyo, Japan), and snap frozen in liquid nitrogen.We generated two anti-WNK4 antibodies, recognizing the N-terminus (MLAPRNTETGVPMS) and the C-terminal portion (QRSDNLPGPGIMRRN) of mouse WNK4.The specificity of these anti-WNK4 antibodies was verified by antigen absorption tests (Supplementary Material, Figure S1).The antibody recognizing the N-terminus of WNK4 was better than the C-terminal antibody at detecting WNK4 in immunofluorescence assays (Supplementary Material, Figure S2).After we used our entire initial batch of anti-OSR1/SPAK and anti-phosphorylated OSR1/SPAK antibodies, we produced a second batch with the same antigen peptides (17).The initial batches recognized multiple bands around 70 kD in addition to bands around 50 kD (15).We tentatively labeled the bands around 70 and 50 kD as SPAK and OSR1, respectively, on the basis of the previous study (27).In this study, we re-evaluated the specificity of several OSR1/SPAK antibodies, including our own, using T7-tagged OSR1 and SPAK as controls.As shown in Supplementary Material, Figure S3, the new batch of anti-total OSR1/SPAK antibody recognized both OSR1 and SPAK, but the sensitivity was slightly better for OSR1 than for SPAK.The major band detected by this antibody in mouse kidney was identified as OSR1 by comparing the results obtained by OSR1-specific (M09; Abnova, Taipei, Taiwan) and SPAK-specific (Cell Signaling Technology, Danvers, MA, USA) antibodies.With regard to anti-phosphosrylated-OSR1/SPAK antibody, the new batch scarcely detected p-SPAK, as shown in Supplementary Material, Figure S4.Accordingly, the band we detected in the kidney samples were regarded as p-OSR.As in the case with

Figure 1 .
Figure 1.Targeting strategy for generating Wnk4 hypomorphic mice.This diagram shows the targeting construct, the wild Wnk4 locus, and the targeted locus before and after Cre recombination.Exon 7 was flanked with two loxP sites.

Figure 2 .
Figure 2. Characterization of the targeted Wnk4 allele.(A) Verification of homologous recombination by Southern blotting.SpeI-digested genomic DNA derived from embryonic stems cell clones (left panel) and mouse tails (right panel).The 5.2-kb band is from the wild-type allele, and the 4.3-kb band (left panel) is from the floxed allele.The 3.8-kb band (right panel) is from the mutated allele after Cre recombination.(B) RT-PCR products with a primer set designed for exons 6 and 9. Kidney mRNA was isolated from the mutant mice and their wild-type littermates.Sequencing of RT -PCR products revealed that the mutant mouse had two patterns of splicing: the major upper band consisted of the mRNA spliced from exon 6 to 8, and the minor lower band was the transcript consisting of exons 6 and 9.The upper transcript has a shift of the open reading frame in the region corresponding to exon 8, but the open reading frame was preserved after the deletion of the exons 7 and 8 in the lower transcript.(C) Immunoblots of WNK4.Two anti-WNK4 antibodies generated against the N-terminus and C-terminal portion of WNK4 were used for immunoblots.Both antibodies detected the same band of WNK4 in the wild-type mice (155 kD).However, both antibodies recognized a band of lower molecular weight (140 kD) in the mutant mice.This size difference matched the expected size of WNK4 protein in which exons 7 and 8 were deleted (WNK4DEx7-8).(D) The amino acid sequence deleted in the WNK4DEx7-8.The letters in green indicate a putative coil domain, and the amino acids in red are mutated in PHAII patients.E. Immunofluorescence of WNK4 in the kidney.The anti-WNK4 antibody generated against the N-terminus was used for immunofluorescence.Intrarenal localization of the wild-type WNK4 and WNK4DEx7-8 (red) was determined by co-staining with parvalbumin and calbindin-D28K (green), and they did not appear to be different.Bar ¼ 50 mm.

Figure 3 .
Figure 3. Kinase assay of WNK4DEx7-8in vitro and in vivo.In an in vitro assay (left panels), FLAG-tagged WNK4 and FLAG-tagged WNK4 devoid of exons 7 and 8 produced in HEK293 cells by transient transfection were immunoprecipitated with M2-FLAG agarose, incubated with GST-SPAK (17) in a kinase buffer containing g 32 P-ATP, separated by SDS-PAGE and visualized by autoradiography.In the in vivo assay (right panels), WNK4 and the WNK4DEx7-8 protein were immunoprecipitated from the kidney with anti-WNK4 antibody generated against the N-terminus and used for a kinase assay (as described above).Little autophosphorylation of WNK4DEx7-8 was observed in the mutant mice, and phosphorylation of GST-SPAK was significantly reduced.

Figure 5 .
Figure 5. Telemetric measurement of blood pressure.Systolic blood pressure of the hypomorphic mice during their active period was significantly lower than that in their wild-type littermates (P , 0.05).

Figure 6 .
Figure 6.Phosphorylation status of OSR1 and NCC.(A) Although the amounts of total OSR1 and SPAK were not changed in the hypomorphic mice, the phosphorylated OSR1 was significantly decreased ( Ã P , 0.05) in the hypomorphic mice.(B) The amounts of phosphorylated NCC at three sites (T53, T58, S71) were significantly reduced in the hypomorphic mice without an apparent change in the total NCC content.

Figure 7 .
Figure 7. Immunoblots of ROMK1, Maxi-K and ENaC channels.(A) No differences in the amount of potassium channels were seen between the hypomorphic mice and their wild-type littermates.(B) The protein expression of full-length ENaCs (a, b, and g at 85 kD) and cleaved-ENaCs (a at 30 kD and gat 70 kD) was increased in the hypomorphic mice.Equal loading of protein in each lane was verified by b-actin immunoblot shown in A.