WNK1 Affects Surface Expression of the ROMK Independent of WNK4

Georgina Cope, Meena Murthy, Amir P. Golbang, Abbas Hamad, Che-Hsiung Liu, Alan W. Cuthbert, and Kevin M. O’Shaughnessy Department of Medicine, University of Cambridge, Cambridge, United Kingdom

The WNK (with no lysine ) are a novel class of serine/threonine kinases that lack a characteristic lysine residue for ATP docking. Both WNK1 and WNK4 are expressed in the mammalian , and in either can cause the rare familial syndrome of hypertension and (Gordon syndrome, or type 2). The molecular basis for the action of WNK4 is through alteration in the membrane expression of the NaCl co-transporter (NCCT) and the renal outer-medullary K channel KCNJ1 (ROMK). The actions of WNK1 are less well defined, and evidence to date suggests that it can affect NCCT expression but only in the presence of WNK4. The results of co-expressing WNK1 with ROMK in Xenopus oocytes are reported for the first time. These studies show that WNK1 is able to suppress total current directly through ROMK by causing a marked reduction in its surface expression. The effect is mimicked by a kinase-dead mutant of WNK1 (368D>A), suggesting that it is not dependent on its catalytic activity. Study of the time course of ROMK expression further suggests that WNK1 accelerates trafficking of ROMK from the membrane, and this effect seems to be dynamin dependent. Using fragments of full-length WNK1, it also is shown that the effect depends on residues in the middle section of the (502 to 1100 WNK1) that contains the acidic motif. Together, these findings emphasize that the molecular mechanisms that underpin WNK1 regulation of ROMK expression are distinct from those that affect NCCT expression. J Am Soc Nephrol 17: 1867–1874, 2006. doi: 10.1681/ASN.2005111224

he WNK (with no lysine kinase) kinases WNK1 and interaction extends to the effects of WNK on other transporters WNK4 are widely expressed in mammalian transport- or ion channels. Lifton’s laboratory has shown, for example, T ing epithelia (1,2), and expression studies in Xenopus that WNK4 also inhibits expression of the Na-K-Cl cotrans- oocytes suggest that they are able to modify the expression of porter SLC12A2 (NKCCl) and Cl/base exchanger SLC26A6 several co-transporters and ion channels (3,4). The details of the (CFEX) transporters and the renal outer-medullary K channel interaction are best understood for WNK4, which reduces sur- KCNJ1 (ROMK) (1,11). The effect on ROMK is intriguing be- face expression of the thiazide-sensitive NaCl co-transporter cause of the part that it plays in K secretion in the distal (NCCT; symbol SLC12A3) in Xenopus oocytes (5–8). This and a reduction in its expression in the collecting duct effect of WNK4 depends on its serine-threonine (S/T) kinase could explain the hyperkalemia that is seen in patients with activity as well as a highly conserved downstream acidic motif Gordon syndrome. We recently confirmed this effect of WNK4 (EPEEPEADQH). Mutations that cause charge-changing amino (8), but, to date, there have been no reports on WNK1 effects on acid substitutions within this motif abolish the inhibitory effect ROMK expression. To address this, we studied their coexpres- of wild-type WNK4 and cause the phenotype of hypertension sion using Xenopus oocytes. We show that in this system, and hyperkalemia that characterizes Gordon syndrome WNK1 directly affects ROMK trafficking independent of (pseudohypoaldosteronism type 2 [PHA2]; OMIM #145260) (9). WNK4 and that this effect is not kinase dependent. Using WNK1 mutations also can cause this phenotype, but published truncations or fragments of WNK1, we further show that the C data suggest that WNK1 protein is effective only in regulating terminal of WNK1 is not necessary for this effect and that the NCCT trafficking when coexpressed with WNK4 (5,10). This acidic motif and coiled-coil regions alone may be sufficient. suggests that the WNK may form a multimeric complex with This has important implications for the physiology of WNK1 NCCT and that protein–protein interactions between WNK1 because its predominant isoform in the kidney is an N-terminal and WNK4 are key to the functionality of WNK1. truncation that lacks a functional kinase S/T domain. It is not known whether this paradigm of WNK1–WNK4 Materials and Methods Received January 6, 2006. Accepted April 20, 2006. Cloning and cRNA Synthesis Full-length cDNA for ROMK2 was identified in an IMAGE clone Published online ahead of print. Publication date available at www.jasn.org. (clone no. 4611308) and then subcloned in-frame into pEGFP-C1 to G.C. and M.M. contributed equally to this work. produce a GFP fusion protein with the N-terminal of ROMK2. A full-length clone for rat WNK1 was a gift of Dr. Melanie Cobb (12). The Address correspondence to: Dr. Kevin M. O’Shaughnessy, Clinical Pharmacology Ͼ Unit, Level 6, ACCI, Box 110, Addenbrooke’s Hospital, Cambridge, CB2 2QQ, UK. WNK1 sequence was mutated to produce 637Q E (to reproduce the Phone: ϩ44-1223-762578; Fax: ϩ44-1223-762576; E-mail: [email protected] WNK4 565QϾE disease ) and 368DϾA (kinase-dead) mutants

Copyright © 2006 by the American Society of Nephrology ISSN: 1046-6673/1707-1867 1868 Journal of the American Society of Nephrology J Am Soc Nephrol 17: 1867–1874, 2006 using site-directed mutagenesis. The WNK1 protein fragments were microscope. All images were captured using an equatorial section produced by PCR amplification off the full-length WNK1 template to through each oocyte. Images were collected using a ϫ10 objective lens produce a product with a 5Ј EcoRI site and a 3Ј XhoI site for direct with brightness and contrast settings kept constant for all oocytes in cloning into pET29b (Novagen, EMDBiosciences, Damstadt, Germany). each injection series. The fluorescence signal in the membrane was

This created an in-frame N-terminal (His)6 fusion and a C-terminal quantified using Leica confocal software (version 2.61 of LCS Lite, Leica S-Tag. Details of the primers that were used to generate them are Microsystems, Heidelberg, Germany) with sampling made at 16 equi- shown in Table 1. Full-length sequence for WNK4 was PCR-amplified spaced points on the circumference and averaged to give mean total from mouse kidney cDNA and cloned into pcDNA3. Wild-type and fluorescence intensity in arbitrary fluorescence units. K44A dynamin clones were gifts of Dr. Peter Friedman (13). All clones were verified by sequencing before being used to run off cRNA. Copy S-Tag–Agarose Pulldown Method RNA was transcribed in vitro from linearized plasmids using either the Recombinant fusion of truncated WNK1 proteins with N- T7 or the SP6 mMESSAGE mMACHINE kit (Ambion, Austin, TX) and terminal (His)6-Tag and C-terminal S-Tag were generated using Esche- quantified by ultraviolet absorption spectroscopy. richia coli protein expression system, cells were lysed using lysozyme (10 mg/ml), and proteins were extracted and purified using IMAC Expression in Xenopus oocytes (Akta Prime, GE Healthcare, Uppsala, Sweden). Purified proteins were Xenopus laevis eggs were harvested and de-folliculated as detailed dialyzed on a Slide-a-lyser dialysis cassette (Pierce, Rockford, IL) with previously (8). Briefly, the cRNA (10 ng of ROMK) was injected in a a molecular weight cutoff of 10 kD and concentrated using Centricon total volume of 100 nl per oocyte, and for co-injections involving WNK1 filters (Millipore Corp., Billerica, MA). Protein concentration was esti- constructs, an additional 10 ng of cRNA was added to the injectate mated using the BCA assay (Pierce). again in a total volume of 100 nl. Water-injected oocytes were used as For pulldown assays, 20 ␮g of fusion protein was adsorbed onto controls throughout. Oocytes then were incubated in ND96 that con- S-Tag agarose (Novagen) and washed once in wash buffer (20 mM tained 2 mM sodium pyruvate and 0.1 mg/ml gentamicin at 18°C for Tris-HCl [pH 7.5], 1.5 M NaCl, and 1% Triton X-100). In preliminary 2 d unless stated otherwise. experiments, binding of fusion protein to the beads was confirmed by both Coomassie staining and Western blotting with an anti-polyhis Two-Electrode Voltage Clamp Recording antibody (Novagen). Bead-bound protein was exposed to 200 ␮lofa Two-electrode voltage clamp used microelectrodes that were filled lysate from oocytes that were injected with ROMK cRNA and pro- with 3 M KCl (1 to 2 M⍀) for voltage sensing and current passing. cessed using a lysis buffer that contained 50 mM Tris-HCl (pH 7.5), 150 External voltage and current electrodes consisted of fine, chloride- mM NaCl, 2 mM EDTA, 1% Triton X-100, and 10 ␮l of Protease coated silver wires. Oocytes were held in a small chamber and perfused Inhibitor Cocktail Set III (Calbiochem, La Jolla, CA). Beads were incu- continuously with ND96 (2 ml/min) at room temperature (18 to 20°C). bated at 4°C for4honarotating stage. They then were pelleted and After impaling, oocytes were held at a holding potential of Ϫ60 mV. washed twice with affinity wash buffer before SDS-PAGE separation Current-voltage (I-V) plots were obtained from voltage step protocols and Western blotting. that ranged from Ϫ140 mV to 40 mV in 20-mV increments. The oocytes were held at each voltage step for 500 ms with 100-ms intervals be- Western Blotting tween the voltage steps. For the I-V plots, the steady-state current at Oocytes were placed in lysis buffer (50 mM Tris-HCl [pH 7.5], 150 each voltage step was used. Initially, oocytes were clamped at Ϫ60 mV mM NaCl, 2 mM EDTA, 1% Triton X-100, and 10 ␮l of Protease in ND96, and when stable, the perfusing solution was switched to high Inhibitor Cocktail Set III from Calbiochem) and left for 10 min on ice ϩ ϩ ϩ K ND96 (K 45 mM) and I-V plots were obtained. The high-K ND96 before vigorous pipetting and vortexing. Oocytes then were sonicated, ϩ was changed to high-K ND96 that contained 2 mM BaCl2, and a and the yolk and the cellular debris were pelleted by centrifugation for second set of I-V plots were obtained. The ROMK current was defined 10 min at 10,000 rpm at 4°C. The supernatant was removed, 5 vol of as the difference between these two curves. The clamp potential of the acetone was added, and the Eppendorf tube was left on ice for 10 min, two-electrode voltage clamp amplifier (OC-725C amplifier; Warner before being centrifuged at 10,000 rpm for 5 min to remove any final Instruments, Hamden, CT) was controlled using the program Pulse and yolk and debris. Oocyte lysate was stored at Ϫ70°C until required. Pulsefit (HEKA Electronic, Lambrecht, Germany) in conjunction with Thawed lysates (from 10 oocytes) were solubilized in sample buffer an ITC-16 interface (Computer Interface Instrutech Corp., Long Island, NY). (100 mM Tris-HCl [pH 7.6], 5% glycerol, 2% SDS, 5% ␤-mercaptoetha- Data were filtered at 1 kHz. nol, and 0.02% Bromophenol blue) by heating at 90°C for 3 min before loading onto an 8% PAGE gel. For the pulldowns, an equal volume of Quantitative Confocal Microscopy SDS-PAGE buffer (10% SDS, glycerol, 1 M Tris [pH 6.8], 1 M dithio- Membrane surface expression of EGFP-ROMK was determined by threitol, and Bromophenol blue) was added to pelleted beads to elute laser-scanning confocal microscopy with a Leica DMRXA confocal proteins before polyacrylamide electrophoresis. Proteins were blotted

Table 1. Primer sequences used to generate the WNK1 (with no lysine kinase) fragments

Primer Name 5Ј-3Ј sequence Product size

1–501WNK1F CGTAACGAATTCGCTACGTATTGAAGATATTAA 1.5 Kb 1–501WNK1R TTGCCTCTCGAGTCCTTCTAATTTCTGAGAACCTG 1.5 Kb 502–1100WNK1F CTAAAGGAATTCATGTCTGACGGCACCGCAGAG 1.8 Kb 502–1100WNK1R TGGCTACTCGAGAATTTGATAGCTATCTTTTTCT 1.8 Kb 1101-CTWNK1F GGCTATTCTCGAGCTAGGTGGTCCGTAGGTTGGA 4.2 Kb 1101-CTWNK1R TTTGCGCGAATTCGACACCACCACTTTCCAGTGC 4.2 Kb J Am Soc Nephrol 17: 1867–1874, 2006 Independent Effect of WNK1 on ROMK Expression 1869 onto polyvinylidene difluoride membranes using semidry blotting ap- negative approach using a C-terminal fragment of WNK4 (620 paratus and blocked with 5% milk powder in PBS overnight at 4°C with to 1222) that lacks both a kinase domain and an acidic motif. ϫ agitation and then washed 5 15 min in 0.05% PBS Tween 20. The Injection of cRNA for this fragment had no effect on ROMK membranes were probed with a rabbit anti-mouse ROMK antibody current injected either alone or with WNK1 (Figure 1B). (Sigma, St. Louis, MO) or anti-poly(his)6 antibody (Novagen) at 1:1500 dilution for3hat4°C. Wash procedures were repeated and then reprobed for 45 min with a goat anti-rabbit horseradish peroxidase Studies of WNK1 on ROMK Surface Expression conjugated secondary antibody 1:2000 (Amersham Biosciences). After ROMK expression in the surface membrane of Xenopus oo- final washing, horseradish peroxidase was visualized by ECL detection cytes was assessed using an EGFP-ROMK fusion protein. Oo- (Amersham Biosciences, Uppsala, Sweden). cytes that were injected with cRNA for the EGFP-ROMK fusion protein showed an easily quantifiable green fluorescence signal Statistical Analyses in the oocyte membrane when assessed 40 h after injection For all oocyte experiments, at least 10 to 15 oocytes were injected for (Figure 2). This signal was markedly reduced by co-injection of each cRNA used. Differences between groups were compared by one- cRNA for full-length WNK1, and co-injection of both WNK1 way ANOVA with post hoc comparison by t testing. Figures show and WNK4 cRNA produced no further inhibition of EGFP- representative experiments that were replicated using at least three different batches of oocytes from different donor animals. Data points ROMK expression in the surface membrane (Figure 2). represent mean Ϯ SEM, and the SPSS statistical package (version 11; To explore further the mechanism for the effect of WNK1 on SPSS, Inc., Chicago, IL) was used throughout with significance defined ROMK expression, we used WNK1 mutants. This showed that as P Ͻ 0.05. the inhibitory effect of WNK1 on ROMK expression does not require intact S/T kinase activity because the effect could be Results reproduced with a kinase-dead mutant of WNK1 (368DϾA; Effect of Full-Length WNK1 on ROMK Current Figure 2). To look at the effect of charge conservation within the To study the effect of WNK1 on total current through ROMK, acidic motif, we injected a mutant WNK1 cRNA (637QϾE) that we injected cRNA for ROMK into Xenopus oocytes either alone carries one of the acidic motif mutations that in WNK4 (as or in combination with WNK1 or WNK1 plus WNK4. Injection 565QϾE) causes Gordon syndrome (PHA2) (9). This mutant of cRNA for a full-length WNK1 protein causes profound in- WNK1 637QϾE proved to be as effective as wild-type WNK1, hibition of the Ba-sensitive currents through ROMK in voltage- suggesting that charge conservation within this highly con- clamped Xenopus oocytes (Figure 1A). Co-injection of cRNA for served motif is not important for WNK1 effects on ROMK full-length WNK4 with WNK1 produced no further inhibition expression. of ROMK current. We also explored the effect of WNK1 on the time course of Endogenous WNK4 may be necessary for the effect of WNK1 the fluorescence signal in the surface membrane after injection on ROMK current. To investigate this, we used a dominant of EGFP-ROMK cRNA. Oocytes that expressed EGFP-ROMK

Figure 1. ROMK current-voltage (I-V) plots for voltage-clamped Xenopus oocytes. (A) ROMK was expressed alone (ⅷ)or coexpressed with WNK1 (Œ) or WNK1 ϩ WNK4 (). (B) ROMK (ⅷ) coexpressed with WNK1 (Œ) and a nonfunctional WNK4 ϩ fragment (WNK4 620 to 1222; f, ) as indicated in the legend. The plots are the recorded Ba2 -sensitive currents in the oocytes (I; ␮A) at each holding membrane potential (V; mV). 1870 Journal of the American Society of Nephrology J Am Soc Nephrol 17: 1867–1874, 2006

pathway involving -coated vesicles. To assess the role of dynamin in WNK1 suppression of ROMK, we first studied the effect of injecting dynamin 1 cRNA alone. This produced the same suppression as WNK1, and coexpressing WNK1 and dynamin produced no further inhibition of membrane expres- sion (Figure 4). A GTPase-defective form of dynamin (14), K44A, was used to explore the specificity of the dynamin effect. Expressed in Xenopus oocytes, this mutant exerts a profound dominant negative effect on dynamin-dependent endocytosis (11,15). Injection of the dominant negative K44A mutant caused substantial reversal of the effect of WNK1 on EGFP-ROMK expression (Figure 4). Figure 2. Quantitative confocal fluorescence of Xenopus oocytes that expressed enhanced green fluorescent protein (EGFP)- Use of WNK1 Fragments to Define Critical Domains for ROMK. Oocytes expressed the EGFP-ROMK fusion protein ROMK Interaction alone or with WNK1 proteins as indicated. A confocal image To try to isolate further the domains that are important in through a typical oocyte that expressed EGFP-ROMK alone is WNK1 inhibition of ROMK membrane expression, we isolated shown in the inset. The WNK1 368DϾA is a kinase-dead mu- the main domains using three fragments of the full-length tant (WNK1 KD), and 637QϾE has a mutation in the WNK1 sequence WNK1 (Figure 5). These then were coexpressed in acidic motif. Water-injected controls are shown in the right- oocytes as small poly-(His)6 fusion proteins with EGFP-ROMK hand column. All WNK1 proteins were significantly different to monitor their effect on membrane expression. The results are from EGFP-ROMK alone; P Ͻ 0.0001. shown in Figure 5. The fragment that contains the kinase do- main (1 to 501 WNK1) has a small inhibitory effect on the alone showed little membrane fluorescence before 15 h (Figure EGFP-ROMK fluorescence signal, but the fragment that con- 3). It then rises in two phases to an eventual plateau at approx- tains the acidic motif (502 to 1100 WNK1) produced a much imately 40 h. This time course does not reflect cRNA translation larger effect and the C-terminal fragment (1101-CT WNK1) had because the level of ROMK protein in total oocyte lysates was no significant effect on EGFP-ROMK expression. constant up to 40 h after injection (Figure 3, inset). The co- We also looked at the effect of the 1 to 501 and 502 to 1100 injection of WNK1 or WNK4 cRNA did not affect the early fragments on ROMK currents directly (Figure 6A). The N- appearance of the signal (15 to 20 h), but this early phase was terminal fragment (1 to 501 WNK1) had no measurable effect, not maintained subsequently and there was a steady decline in but 502 to 1100 WNK1 substantially inhibited the total ROMK expression. The kinetics of this late-phase decline seemed to be current. We also looked at the effect of the kinase-dead and 637 the same in oocytes that were injected either with WNK1 alone or with WNK1 and WNK4 together.

Effect of Dynamin of WNK1 Interaction with ROMK The removal of membrane proteins can occur through sev- eral pathways, including a dynamin-dependent endocytosis

Figure 3. Time course of EGFP-ROMK expression in Xenopus oocytes. Oocytes were injected with cRNA for EGFP-ROMK Figure 4. Role of dynamin in EGFP-ROMK interaction with alone (F) or plus WNK1 () or WNK1 ϩ WNK4 (Œ), and the WNK1. Oocytes were injected with cRNA for EGFP-ROMK quantitative confocal fluorescence (plotted logarithmically) is alone or plus WNK1, dynamin 1 (DNM), the combination of measured at various time points. Time points that are signifi- WNK1 and DNM, or WNK1 and the dominant negative form cantly different from EGFP-ROMK alone are indicated: *P Ͻ (K44A DNM) as indicated. The inset shows representative con- 0.05; **P Ͻ 0.01; ***P Ͻ 0.0001. The inset figure shows total focal images for each injection. Significant differences are oocyte protein Western blotted for ROMK at various time shown: *P Ͻ 0.0001 versus EGFP-ROMK alone; **P Ͻ 0.005 points between 15 and 40 h. WNK1/DNM versus WNK1/K44A DNM. J Am Soc Nephrol 17: 1867–1874, 2006 Independent Effect of WNK1 on ROMK Expression 1871

protein fragments using lysates from oocytes that were injected with ROMK cRNA. The N-terminal fragment (1 to 501 WNK1) was able to produce measurable pulldown of ROMK, but this was minimal compared with the 502 to 1100 WNK1 fragment (Figure 7). No pulldown was detectable with the C-terminal fragment in keeping with its lack of effect on EGFP-ROMK expression (Figures 5 and 6).

Discussion The hypertension in Gordon syndrome now can be explained by loss-of-function mutations in WNK4, which cause increased expression of NCCT and salt retention in the distal nephron. Significantly, these same mutations show gain of function in their interaction with the K channel, ROMK, and it is the suppression of ROMK expression in the collecting duct that probably causes hyperkalemia in Gordon patients. WNK1 mutations also can cause Gordon syndrome, but ex- pression studies with NCCT suggested that WNK1 could only affect surface expression of the transporter in the presence of WNK4 (5). This effect seems to rely on phosphorylation, be- cause Ellison’s laboratory recently reported that kinase-dead WNK1 is not effective as an antagonist of WNK4-mediated suppression of NCCT (10). This contrasts with our findings here with ROMK expression, in which WNK1 alone is able to suppress almost completely its surface expression. This effect does not require coexpression of WNK4 or seem to require intact S/T kinase activity; the kinase-dead 368DϾA mutant has the same effect as wild-type WNK1, and its complete lack of kinase activity is well documented (12). The results with kinase- dead WNK1 also are consistent with our findings that deletion of the first 500 residues of WNK1 (in 502 to 1100 WNK1) does not affect the WNK1-ROMK interaction. Previous functional associations of WNK1 have been linked to its S/T kinase do- main, such that catalytically active WNK1 phosphorylates Syn- aptotagmin 2 (16), SPAK/OSR1 (17), and SGK1 (18,19). This is important because of the large number of potential WNK1 proteins in vivo reflecting an alternative promoter and splice Figure 5. Effect of WNK1 fragments on membrane expression of variation (20). Several of these, including the predominant kid- EGFP-ROMK. (A) Major domains contained within each frag- ney-specific isoform, are N-terminal truncations that lack an ment (not to scale): AM, acidic motif; CC, coiled-coil; NLS, S/T kinase domain (21). Hence, if these isoforms have a role in nuclear localization signal; PEST motif. (B) Representative regulating ion transport, then they must be mediated by re- Western blots for ROMK are shown for oocytes that expressed gions outside the kinase domain. Our findings with the inter- Ϫ ROMK alone ( ) or with full-length (FL) WNK1 or one the action of WNK1-ROMK in the oocyte provide further support fragments as indicated. (C) Representative Western blots of for a functional role of N-terminal truncations of WNK1 in vivo. oocytes that expressed the various WNK1 fragments as indi- That WNK1 has actions independent of protein phosphory- cated and blotted using an anti-poly(His) antibody. (D) Fluo- 6 lation has been documented before in other kinases, for exam- rescence signals from oocytes that expressed the EGFP-ROMK fusion protein alone or with WNK1 protein fragments as indi- ple, in the Ste-20–related kinase, SPAK, which also shares cated. Significant difference from EGFP-ROMK alone are indi- homology to the WNK kinase family and interacts with NKCCl cated: *P Ͻ 0.02; **P Ͻ 0.0001. (22). It also is important to note that the interaction of WNK4 with ROMK is itself kinase independent (11 and our own unpublished observations) and that both WNK4 and WNK1 QϾE WNK1 mutants and found that they behaved like full- share several other domains that may be important for protein– length WNK1 and completely suppressed ROMK current in protein interactions. Foremost of these is the acidic motif, which keeping with their effect on EGFP-ROMK expression (Figure is a run of 10 predominantly negatively charged amino acids 6B). within the midsection of WNK1. This motif is unique to WNK To look for evidence of direct interaction of the fragments family members and the disease mutations in WNK4 cluster with ROMK, we attempted pulldowns with the WNK1 fusion within it. Our 502 to 1100 WNK1 fragment that contains the 1872 Journal of the American Society of Nephrology J Am Soc Nephrol 17: 1867–1874, 2006

Figure 6. Effect of WNK1 fragments and mutant WNK1 on ROMK current. ROMK was expressed alone (F) or with a WNK1 ϩ fragment (A) or a mutant full-length WNK1 (B) as indicated in the legends. The plots are the recorded Ba2 -sensitive currents in the voltage-clamped oocytes (I; ␮A) at each holding membrane potential (V; mV).

This early phase presumably reflects trafficking of the receptor to the membrane, and this process seems to be intact. Only at later time points does ROMK disappear from the membrane in the presence of WNK1, which is likely to represent increased removal of ROMK protein probably through the clathrin-me- diated pathway suggested previously. Dynamin-dependent en- docytosis usually is taken to imply endocytosis through clath- rin-coated vesicles. The evidence for this is extensive, but dynamin also regulates membrane turnover through caveolae Figure 7. Interaction of WNK1 fragments with ROMK. The in mammalian cells (24,25). Xenopus expresses caveolin WNK1 protein fragments that were immobilized on S-Tag– (26), and there is evidence that caveolae may operate in Xenopus agarose beads were incubated with oocyte lysate and pelleted, and bound protein was eluted and Western blotted using an oocytes(27,28). Therefore, it is possible that caveolae have a role anti-mouse ROMK antibody. The fragments used in each lane in the accelerated endocytosis of ROMK that we observed in the are shown. presence of WNK1. Despite the genetic heterogeneity in pedigrees with PHA2, hyperkalemia is a consistent finding, and although the molec- ular mechanisms are distinct, the pathophysiology probably is acidic motif was almost as effective as full-length WNK1, but the same: A reduction in ROMK expression and hence secretion other domains within the middle section of the protein may of K into the collecting duct. In individuals with PHA2A, this contribute, because introducing a charge-changing disease mu- arises from gain-of-function mutations in WNK4 that affect the tation (QϾE) into the acidic motif of WNK1 does not affect its inhibitory action on ROMK expression. Contrast this with acidic motif of the protein. These seem to accelerate trafficking WNK4, where the same mutation profoundly disturbs WNK4 of ROMK from the surface membrane (8,11). In individuals function (8). The coiled-coil domains are possible candidates, with PHA2C, the mutations are within the PRKWNK1 gene although there is not as yet a crystal structure for full-length and are large intronic deletions rather than missense mutations WNK protein to know how these domains are displayed and in the acidic motif. The consequences of this within cells that hence might play a role in protein–protein interactions (23). line the (DCT) still is unknown, but on The previous report on WNK4 interaction with ROMK sug- the basis of RNA measurements in leukocytes from affected gested that the inhibition of ROMK expression by WNK4 was individuals, these deletions may cause overexpression of full- dependent on clathrin-mediated endocytosis (11). ROMK has length WNK1 (9). Our results suggest that this also could an NPXY recognition motif for AP2 that binds amphiphysin accelerate trafficking from the membrane of ROMK reducing and clathrin, and deletion or mutation of the motif blocked the ROMK expression. Other factors operate in Gordon syndrome effect of WNK4. This mechanism is relevant to our time-course to modulate the deficit in K secretion imposed by reduced study, which shows that early ROMK expression is unaffected. ROMK expression. For example, mutant WNK4 in PHA2A also J Am Soc Nephrol 17: 1867–1874, 2006 Independent Effect of WNK1 on ROMK Expression 1873 increases NCCT expression in the DCT and hence reduces nase structure, downstream targets, and potential roles in delivery of NaCl for electrogenic Na uptake in the collecting hypertension. Res 15: 6–10, 2005 duct. Overexpression of full-length WNK1 in PHA2C also may 4. Cope G, Golbang A, O’Shaughnessy KM: WNK kinases reduce ENaC expression directly in the DCT (29). and the control of . Pharmacol Ther 106: If ROMK expression is suppressed in the collecting duct of 221–231, 2005 5. Yang CL, Angell J, Mitchell R, Ellison DH: WNK kinases patients with Gordon syndrome, then it is reasonable to ask regulate thiazide-sensitive Na-Cl cotransport. J Clin Invest why individuals with nonfunctional forms of ROMK (causing 111: 1039–1045, 2003 type 2 Bartter) develop hypokalemia (30). In fact, these individ- 6. Wilson FH, Kahle KT, Sabath E, Lalioti MD, Rapson AK, uals do have hyperkalemia, but it is transient and confined to Hoover RS, Hebert SC, Gamba G, Lifton RP: Molecular the neonatal period (31). They subsequently lose large amounts pathogenesis of inherited hypertension with hyperkale- of Na and K because in the thick ascending limb, activity of the mia: The Na-Cl cotransporter is inhibited by wild-type but ϩ Na,K,2Cl transporter (NKCC2) is limited by K returning to the not mutant WNK4. Proc Natl Acad Sci U S A 100: 680–684, lumen through ROMK. There may be several reasons for why 2003 this is not seen in PHA2. First, the effect of the WNK may be 7. Kahle KT, Wilson FH, Lifton RP: Regulation of diverse ion confined to the DCT and collecting duct. Alternatively, residual transport pathways by WNK4 kinase: A novel molecular ROMK function may be sufficient to return K to the lumen switch. Trends Endocrinol Metab 16: 98–103, 2005 because suppression is not complete and hence K rate limiting. 8. Golbang AP, Murthy M, Hamad A, Liu CH, Cope G, Van’t Hoff W, Cuthbert A, O’Shaughnessy KM: A new kindred There also could be isoform selectivity, because isoforms 2 and with pseudohypoaldosteronism type II and a novel muta- 3 of ROMK usually are mutated in type 2 Bartter (30). Given the tion (564DϾH) in the acidic motif of the WNK4 gene. rarity of individuals with PHA2, let alone their tissues, the Hypertension 46: 295–300, 2005 development of animal models of PHA2 are needed to resolve 9. Wilson FH, Disse-Nicodeme S, Choate KA, Ishikawa K, the precise role of ROMK in the final phenotype. Nelson-Williams C, Desitter I, Gunel M, Milford DV, Lip- kin GW, Achard JM, Feely MP, Dussol B, Berland Y, Un- Conclusion win RJ, Mayan H, Simon DB, Farfel Z, Jeunemaitre X, We have shown for the first time that WNK1 can modulate Lifton RP: Human hypertension caused by mutations in WNK kinases. Science 293: 1107–1112, 2001 ROMK directly in the Xenopus oocyte expression system. [Since 10. Yang CL, Zhu X, Wang Z, Subramanya AR, Ellison DH: submission of our manuscript, Lazrak et al. reported similar Mechanisms of WNK1 and WNK4 interaction in the reg- findings when ROMK and full-length WNK1 were coexpressed ulation of thiazide-sensitive NaCl cotransport. J Clin Invest in HEK cells. However, unlike our study, they report that the 115: 1379–1387, 2005 effect of WNK1 depends on intact kinase activity (32).] The 11. Kahle KT, Wilson FH, Leng Q, Lalioti MD, O’Connell AD, details of this interaction contrast sharply with the WNK1– Dong K, Rapson AK, MacGregor GG, Giebisch G, Hebert WNK4 interaction that regulates surface expression of the SC, Lifton RP: WNK4 regulates the balance between renal NCCT transporter. These findings serve to emphasize the com- NaCl reabsorption and Kϩ secretion. Nat Genet 35: 372– plexity of the molecular mechanisms that underlie Gordon 376, 2003 syndrome. 12. Xu B, English JM, Wilsbacher JL, Stippec S, Goldsmith EJ, Cobb MH: WNK1, a novel mammalian serine/threonine protein kinase lacking the catalytic lysine in subdomain II. Acknowledgments J Biol Chem 275: 16795–16801, 2000 This work was supported by grants from the British Heart Founda- 13. Syme CA, Friedman PA, Bisello A: Parathyroid hormone tion (UK) and the Isaac Newton Trust (Cambridge, UK). A.G. and G.C. receptor trafficking contributes to the activation of extra- have PhD studentships from the British Heart Foundation and Medical cellular signal-regulated kinases but is not required for Research Council (UK), respectively. regulation of cAMP signaling. J Biol Chem 280: 11281– Parts of this work have appeared in an abstract to the summer 11288, 2005 meeting of the British Hypertension Society, Queen’s College Cam- 14. van der Bliek AM, Redelmeier TE, Damke H, Tisdale EJ, bridge, September 19 to 21, 2005 (33). Meyerowitz EM, Schmid SL: Mutations in human dynamin block an intermediate stage in coated vesicle formation. J Cell Biol 122: 553–563, 1993 References 15. Veyhl M, Wagner CA, Gorboulev V, Schmitt BM, Lang F, 1. Kahle KT, Grimenez I, Hassan H, Wilson FH, Wong RD, Koepsell H: Downregulation of the Na(ϩ)-D-glucose co- Forbush B, Aronson PS, Lifton RP: WNK4 regulates apical transporter SGLT1 by protein RS1 (RSC1A1) is dependent and basolateral Cl(Ϫ) flux in extrarenal epithelia. Proc Natl on dynamin and protein kinase C. J Membr Biol 196: 71–81, Acad Sci U S A 101: 2064–2069, 2004 2003 2. Choate KA, Kahle KT, Wilson FH, Nelson-Williams C, 16. Lee BH, Min X, Heise CJ, Xu BE, Chen S, Shu H, Luby- Lifton RP: WNK1, a kinase mutated in inherited hyperten- Phelps K, Goldsmith EJ, Cobb MH: WNK1 phosphorylates sion with hyperkalemia, localizes to diverse ClϪ-trans- synaptotagmin 2 and modulates its membrane binding. porting epithelia. Proc Natl Acad Sci U S A 100: 663–668, Mol Cell 15: 741–751, 2004 2003 17. Moriguchi T, Urushiyama S, Hisamoto N, Iemura SI, 3. Xu BE, Lee BH, Min X, Lenertz L, Heise CJ, Stippec S, Uchida S, Natsume T, Matsumoto K, Shibuya H: WNK1 Goldsmith EJ, Cobb MH: WNK1: Analysis of protein ki- regulates phosphorylation of cation-chloride-coupled co- 1874 Journal of the American Society of Nephrology J Am Soc Nephrol 17: 1867–1874, 2006

transporters via the STE20-related kinases, SPAK and 26. Razani B, Park DS, Miyanaga Y, Ghatpande A, Cohen J, OSR1. J Biol Chem M510042200, 2005 Wang XB, Scherer PE, Evans T, Lisanti MP: Molecular 18. Xu B, Stippec S, Chu PY, Lazrak A, Li XJ, Lee BH, English cloning and developmental expression of the caveolin gene JM, Ortega B, Huang CL, Cobb MH: WNK1 activates SGK1 family in the amphibian Xenopus laevis. Biochemistry 41: to regulate the epithelial sodium channel. Proc Natl Acad 7914–7924, 2002 Sci U S A 102: 10315–10320, 2005 27. Kwak JO, Kim HW, Oh KJ, Kim DS, Han KO, Cha SH: 19. Xu B, Stippec S, Lazrak A, Huang CL, Cobb MH: WNK1 Co-localization and interaction of organic anion trans- activates SGK1 by a phosphatidylinositol 3-kinase-depen- porter 1 with caveolin-2 in rat kidney. Exp Mol Med 37: dent and non-catalytic mechanism. J Biol Chem 280: 34218– 204–212, 2005 34223, 2005 28. Sadler SE: Low-density caveolae-like membrane from Xe- 20. Delaloy C, Lu J, Houot AM, Disse-Nicodeme S, Gasc JM, nopus laevis oocytes is enriched in Ras. J Cell Biochem 83: Corvol P, Jeunemaitre X: Multiple promoters in the WNK1 21–32, 2001 gene: One controls expression of a kidney-specific kinase- 29. Naray-Fejes-Toth A, Snyder PM, Fejes-Toth G: The kid- defective isoform. Mol Cell Biol 23: 9208–9221, 2003 ney-specific WNK1 isoform is induced by 21. O’Reilly M, Marshall E, Speirs HJ, Brown RW: WNK1, and stimulates epithelial sodium channel-mediated Naϩ a gene within a novel blood pressure control pathway, transport. Proc Natl Acad Sci U S A 101: 17434–17439, tissue-specifically generates radically different isoforms 2004 with and without a kinase domain. J Am Soc Nephrol 14: 30. Jeck N, Derst C, Wischmeyer E, Ott H, Weber S, Rudin C, 2447–2456, 2003 Seyberth HW, Daut J, Karschin A, Konrad M: Functional 22. Piechotta K, Garbarini N, England R, Delpire E: Character- heterogeneity of ROMK mutations linked to hyperprostag- ization of the interaction of the stress kinase SPAK with the Kidney Int Naϩ-Kϩ-2ClϪ cotransporter in the nervous system: Evi- landin E syndrome. 59: 1803–1811, 2001 dence for a scaffolding role of the kinase. J Biol Chem 278: 31. Finer G, Shalev H, Birk OS, Galron D, Jeck N, Sinai- 52848–52856, 2003 Treiman L, Landau D: Transient neonatal hyperkalemia in 23. Min X, Lee BH, Cobb MH, Goldsmith EJ: Crystal structure the antenatal (ROMK defective) Bartter syndrome. J Pediatr of the kinase domain of WNK1, a kinase that causes a 142: 318–323, 2003 hereditary form of hypertension. Structure 12: 1303–1311, 32. Lazrak A, Liu Z, Huang CL: Antagonistic regulation of 2004 ROMK by long and kidney-specific WNK1 isoforms. Proc 24. Henley JR, Krueger EWA, Oswald BJ, McNiven MA: Dy- Natl Acad Sci U S A 103: 1615–1620, 2006 namin-mediated internalization of caveolae. J Cell Biol 141: 33. Cope G, Golbang A, Murthy M, Hamad A, Liu C-H, Hoff 85–99, 1998 WV, Cuthbert A, O’Shaughnessy KM: The origin of hy- 25. Oh P, McIntosh DP, Schnitzer JE: Dynamin at the neck of perkalaemia in Gordon’s syndrome: WNK1 interacts di- caveolae mediates their budding to form transport vesicles rectly with the potassium channel ROMK to reduce its by GTP-driven fission from the plasma membrane of en- surface expression in Xenopus oocytes [Abstract 5.3]. J Hum dothelium. J Cell Biol 141: 101–114, 1998 Hypertens 19: 832, 2005