WNK bodies cluster WNK4 and SPAK/OSR1 to promote NCC activation in hypokalemia

Abstract

K⁺ deficiency stimulates renal salt reuptake via the Na⁺-Cl⁻ cotransporter (NCC) of the distal convoluted tubule (DCT), thereby reducing K⁺ losses in downstream nephron segments while increasing NaCl retention and blood pressure. NCC activation is mediated by a kinase cascade involving with no lysine (WNK) kinases upstream of Ste20-related proline-alanine-rich kinase (SPAK) and oxidative stress-responsive kinase-1 (OSR1). In K⁺ deficiency, WNKs and SPAK/OSR1 concentrate in spherical cytoplasmic domains in the DCT termed “WNK bodies,” the significance of which is undetermined. By feeding diets of varying salt and K⁺ content to mice and using genetically engineered mouse lines, we aimed to clarify whether WNK bodies contribute to WNK-SPAK/OSR1-NCC signaling. Phosphorylated SPAK/OSR1 was present both at the apical membrane and in WNK bodies within 12 h of dietary K⁺ deprivation, and it was promptly suppressed by K⁺ loading. In WNK4-deficient mice, however, larger WNK bodies formed, containing unphosphorylated WNK1, SPAK, and OSR1. This suggests that WNK4 is the primary active WNK isoform in WNK bodies and catalyzes SPAK/OSR1 phosphorylation therein. We further examined mice carrying a kidney-specific deletion of the basolateral K⁺ channel-forming protein Kir4.1, which is required for the DCT to sense plasma K⁺ concentration. These mice displayed remnant mosaic expression of Kir4.1 in the DCT, and upon K⁺ deprivation, WNK bodies developed only in Kir4.1-expressing cells. We postulate a model of DCT function in which NCC activity is modulated by plasma K⁺ concentration via WNK4-SPAK/OSR1 interactions within WNK bodies.

INTRODUCTION

With-no-lysine (WNK) kinases are essential for the regulation of epithelial ion transport in various tissues. WNKs act as the penultimate link of a signal cascade controlling the activity of cation Cl⁻ cotransporters (CCCs); they phosphorylate either Ste20-related proline-alanine-rich kinase (SPAK) or its homolog, oxidative stress-responsive kinase-1 (OSR1). SPAK/OSR1 then phosphorylate CCCs, including the renal Na⁺-K⁺-2Cl⁻ cotransporter (NKCC2) and Na⁺-Cl⁻ cotransporter (NCC). NCC is exclusively located in the distal convoluted tubule (DCT) of the kidney and is activated by phosphorylation. A DCT-specific property of the WNK-SPAK/OSR1 cascade is that its components can concentrate within spherical cytoplasmic domains lacking a delineating membrane, which recently have been termed “WNK bodies” (6). In the DCT, the WNK-SPAK/OSR1 pathway contributes to systemic K⁺ homeostasis. During K⁺ deficiency, the state most strongly linked to WNK body formation, active SPAK/OSR1 increases electroneutral NaCl reabsorption via NCC. This limits K⁺ wasting in downstream nephron segments, where electrogenic Na⁺ reabsorption via the epithelial Na⁺ channel is balanced by K⁺ secretion into the urine to maintain electroneutrality (34). Conversely, dietary K⁺ excess has also been proposed as a stimulant of WNK bodies (6).

The WNK family is phylogenetically ancient, with homologs in unicellular eukaryotes (13), plants (14, 19, 35), and simple invertebrates (12, 15, 23, 26). Specific alterations of wnk genes arose in coelacanths and are conserved across terrestrial vertebrates (6); a sequence termed exon 4a became integrated into the wnk1 locus, encoding a truncated, kinase-inactive isoform, kidney-specific (KS-)WNK1 (10, 21). KS-WNK1 is expressed along the distal nephron but is especially enriched in the DCT, where it is the most abundant WNK1 isoform by a large margin (32). Boyd-Shiwarski et al. (6) showed that KS-WNK1 is an essential component of WNK bodies, thereby providing a link between WNK body formation and the unique proteome of the DCT.

Genetically altered mouse models have demonstrated a role of KS-WNK1 in the DCT in the maintenance of systemic K⁺ balance. Deletion of KS-WNK1 leads to activation of NCC and a mild elevation in blood pressure (11, 17). Recently, it has been shown that deletion of KS-WNK1 abrogates the suppression of NCC activity by K⁺ loading (36). In a preliminary report (5), KS-WNK1 deletion blunted NCC activation upon K⁺ deprivation. The NCC response to dietary K⁺ intake may therefore be linked to formation of the KS-WNK1-dependent WNK bodies.

In addition to WNKs/SPAK/OSR1, 60S ribosomal subunit protein-22 (RPL22) has been identified as a WNK body component (6). In a previous study, discrete, non-membrane-bound subdomains of the murine DCT were documented after 7 dys of aldosterone administration by Cheema et al. (8), who identified RPL22 and the 20S proteasomal subunit therein but did not examine their kinase content. Ultrastructural similarities between these aldosterone-induced structures and WNK bodies are apparent. Thus, it is highly likely that they are equivalent, as noted by Boyd-Shiwarski et al. (5) themselves. However, the structures were interpreted differently by the two groups. Cheema et al. (8) categorized the structures as pathological protein aggregates resulting from aldosterone-induced metabolic stress, whereas Boyd-Shiwarski et al. (5) postulated a function in WNK-SPAK/OSR1 signaling. Since WNK body function remains unclear, we were interested in identifying essential factors for WNK body generation and in clarifying whether WNK bodies are involved in activating WNK-SPAK/OSR1 signaling in the DCT. To this end, we examined WNK body formation in mice fed diets of varying K⁺ and NaCl content and in genetically engineered mice carrying mutations of WNK-SPAK/OSR1 pathway proteins.

MATERIALS AND METHODS

Animal experiments.

Nine- to twelve-week-old male and female Pax8-rtTA/LC1/Kcnj10^fl/fl mice in the C57Bl/6J strain background were used for experiments involving 72-h time courses of K⁺-deficient diets and 48-h high-K⁺ diets. Where indicated, kidney-specific Kir4.1^−/− mice were generated as previously described (9). wnk4^−/− mice were rederived from cryopreserved sperm (7) at Charles River onto the C57Bl/6NCrl background. Mice were fed a control diet (1% K⁺), K⁺-deficient (15–30 ppm K⁺) diet, or high-K⁺ (5% K⁺) diet. Control or high-K⁺ diets were prepared by adding KCl to the K⁺-deficient diet (TD.88239, Envigo Teklab Diets, Collinsville, IL) in a gel format. Mice were fed the K⁺-deficient diet for 12, 24, 48, and 72 h. The 12-h time point was carried out in the dark part of the cycle. For the rescue protocol, mice were fed the K⁺-deficient diet for 48 h followed by 48 h of the high-K⁺ diet. Mice were housed in a temperature- and humidity-controlled facility on a 12:12-h light-dark cycle with access to food and water ad libitum. Mice were kept in individual metabolic cages with ad libitum access to deionized drinking water and the assigned diet. Urine was collected every 24-h period except for the 12-h time. After the indicated time point, mice were anesthetized with isoflurane for blood collection and euthanized by cervical dislocation for immediate blood and tissue collection. Plasma K⁺ concentration was evaluated using an I-STAT blood analyzer system (Abbott, Chicago, IL). All aforementioned animal experiments were conducted according to the National Institutes of Health Guidelines for the Care and Use of Experimental Animals and were approved by the Institutional Animal Care and Use Committee of the Oregon Health and Science University (Protocol IP00286). For experiments involving mice fed 10-day high-salt/K⁺-deficient, high-salt, and control diets, a K⁺-deficient diet was obtained from Teklad (TD.88239, see above). High-salt/K⁺-deficient diets were prepared by adding 6% NaCl, high-salt diets by adding 6% NaCl and 1% KCl, and control diets by adding 1% KCl in a gel format. Male C57BL/6 mice were used for experiments involving the 10-day diets. These experiments were approved by the Berlin Regional Office for Health and Social Affairs (LAGESO, project no. G0220/12). Transgenic WNK4-PHAII mice were generated by the group of Richard P. Lifton (16).

Perfusion fixation and tissue processing.

Mice were anesthesized with isoflurane and pentobarbital sodium and perfused via the abdominal aorta, first with 3% hydroxyethyl starch in sodium cacodylate (cacodylate) buffer for 20 s and then 3% paraformaldehyde with 3% hydroxyethyl starch in cacodylate buffer for 5 min. Alternatively, mice were perfused first with PBS and then with 3% paraformaldehyde in PBS. Kidney tissue was dissected into slices of 0.5−3 mm thickness for various embedding techniques. For paraffin embedding, tissue was postfixed in 3% paraformaldehyde and 3% hydroxyethyl starch in cacodylate buffer overnight at 4°C and then transferred to cacodylate buffer supplemented with 300 mosmol sucrose and 0.02% sodium azide until embedding. For cryoembedding, tissue was left in 800 mosmol sucrose in cacodylate buffer at 4°C overnight, infiltrated with TissueTek OCT Compound (ThermoFisher Scientific, Walthman, MA), and then snap frozen in 2-methylbutane cooled with liquid nitrogen. For transmission electron microscopy (TEM), tissue was postfixed overnight at room temperature in 1.5% glutaraldehyde, 1.5% paraformaldehyde, and 0.05% picric acid in cacodylate buffer and then in 1% osmium tetroxide and 0.8% potassium hexocyanoferrate in cacodylate buffer for 1.5 h at room temperature. Tissue was dehydrated via an ascending ethanol series and embedded in epoxy resin.

Antibodies.

Rabbit NCC and guinea pig NKCC2 polyclonal antibodies were generated within our research groups (4, 20, 25). To detect WNK1, a commercial goat polyclonal antibody (Santa Cruz Biotechnology, Dallas, TX) directed against a COOH-terminal sequence present in both full-length WNK1 and KS-WNK1 was used. The specificity of the rabbit WNK4 polyclonal antibody has been previously validated in WNK4 knockout mice (28). Rabbit NCC phosphorylated at Thr⁵³ (pT-NCC) polyclonal antibody has been previously validated (18). SPAK/OSR1 phosphorylated at Thr^243/185 (pT-SPAK/OSR1) (38) and WNK phosphorylated at Ser³⁸² of WNK1 cross-reactive with the equivalent WNK4 phosphoacceptor (pS-WNK) (30) sheep polyclonal antibodies were produced by Dario Alessi (Dundee, UK). To label SPAK/OSR1 phosphorylated at Ser^383/325 (pS-SPAK/OSR1), a sheep polyclonal antibody generated by Dario Alessi and a rabbit polyclonal antibody (Millipore, Darmstadt, Germany) were used. Three antibodies to SPAK were used, of which two were produced by commercial suppliers (rabbit polyclonal, Cell Signaling Technologies, Danvers, MA, and mouse monoclonal, Millipore) and one by Eric Delpire (Nashville, TN) (rabbit polyclonal) (22). For an overview of all primary antibodies used in this study and information on antibody validation, see Supplemental Table S1 (all Supplemental Data are provided online at https://doi.org/10.6084/m9.figshare.8104793.v1).

Western blot analysis.

Proteins were extracted in lysis buffer supplemented with protease and phosphatase inhibitors. Proteins (40 µg) were loaded onto 4–15% Mini-PROTEAN TGX Stain-Free precast gels (Bio-Rad, Hercules, CA) for separation and transferred onto polyvinylidene difluoride membranes using the Trans-Blot Turbo transfer system (Bio-Rad); 5% skim milk in PBS was applied for 30 min as a blocking buffer. Membranes were then incubated with antibodies to pT-NCC, NCC, and β-actin in blocking buffer at 4°C overnight (see Supplemental Table S1 for details). Membranes were probed with horseradish peroxidase-conjugated secondary antibodies (1:5,000, Life Technology, Waltham, MA) in blocking buffer for 1 h at room temperature. Bands were detected by enhanced chemiluminescence using ECL Plus (Perkin-Elmer, Waltham, MA). A PXi 4 GeneSys (Syngene, Frederick, MD) system was used for imaging. Densitometric analysis was performed in FIJI (v.2.0, National Institutes of Health, Bethesda, MD).

Immunofluorescence.

Perfusion-fixed, paraffin-embedded sections or OCT-embedded cryosections were used for immunofluorescent labeling. Paraffin-embedded tissue was cut to 4 µm thickness, dewaxed in xylene, and rehydrated via graded ethanol series. For epitope retrieval, sections were steamed in citrate buffer (10 mM sodium citrate, pH 6.0) in a pressure cooker for 6 or 10 min. Sections were incubated with various blocking buffers for 30 min (see Supplemental Table S1). Primary antibodies were dissolved in blocking buffer and incubated overnight at 4°C. Fluorescently labeled secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA, and Invitrogen, Carlsbad, CA) were dissolved in the same and applied for 1 h at room temperature. For cryosections, the same protocol was followed except that sections were cut to 5−6 µm thickness, epitope retrieval was omitted, and sections were permeabilized with 0.5% Triton X-100 in Tris-buffered saline or PBS before being blocked. In labelings prepared for stimulated emission depletion (STED) microscopy, fluorescently labeled secondary antibodies (Abberior, Göttingen, Germany) were incubated for 2.5 h. Samples were mounted in 1:9 PBS-glycerol for confocal microscopy or in Abberior Mount Liquid Antifade for STED microscopy.

Confocal and STED microscopy.

Immunofluorescent labelings were examined with excitation laser lines of 405-, 488-, 543-, and 633-nm wavelengths under an LSM 5 Exciter confocal microscope (Carl Zeiss Jena, Jena, Germany) equipped with Plan-Neofluar ×40/numerical aperture (NA) 1.3 and Plan-Neofluar ×63/NA 1.4 objectives. Alternatively, a Zeiss LSM 880 with Airyscan function was used, equipped with 405-, 488-, 595-, and 633-nm excitation laser lines and a Plan-Apochromat ×63/NA 1.4 objective. For superresolution fluorescence microscopy, we used the Abberior STEDYCON system mounted to a Zeiss AxioImager.Z2 confocal microscope with a Plan-Apochromat ×100/NA 1.46 objective; 561- and 640-nm excitation lasers were used along with a 775-nm depletion laser. Images and z-stacks were processed in FIJI (v.2.0). Maximum intensity projections of z-stacks were generated, brightness and contrast were adjusted, and background was subtracted by rolling ball function (27). For quantitative evaluation in FIJI (v.2.0), confocal micrographs were used without prior modification. For each animal, 15 images containing DCTs were taken at ×400 magnification. Regions of interest containing DCTs were generated by drawing around the basolateral and apical membranes of each DCT and subtracting the nuclei, which were identified by automatically subtracting DAPI-labeled regions. The fraction of DCT cytoplasmic area occupied by pS-SPAK/OSR1-positive punctate signals was then calculated via the Analyze Particles algorithm. DAPI and pS-SPAK/OSR1 signals were identified by Huang’s and Otsu’s automatic thresholding methods, respectively. Imaging and evaluation were conducted in a blinded manner.

Preparation of epoxy resin-embedded samples for TEM.

Semithin sections of epoxy resin-embedded tissue were cut at 0.5 µm thickness and stained with Richardson staining solution for light microscopy. Ultrathin sections were then cut at 70 nm thickness on an Ultracut S microtome (Leica, Wetzlar, Germany), transferred to Formvar-coated copper grids, and stained first with 4% uranyl acetate in 50% ethanol for 15 min and then Reynold’s lead citrate for 4 min.

Immunogold labeling.

Tissue blocks were snap frozen in OCT, sectioned at 80 nm thickness using a cryoultramicrotome (Leica Ultracut E fitted with a Leica EM FCS low-temperature sectioning system) and transferred to Formvar-coated copper grids. Sections were permeabilized with 0.5% Triton X-100 in PBS for 30 min; 5% milk in PBS was applied as a blocking buffer for 60 min and used as antibody dilutant. Primary antibodies to SPAK were incubated for 1 h at room temperature and then overnight at 4°C. Gold particle-conjugated secondary antibodies (5 nm) were applied for 1 h at room temperature. For postfixation, 2.5% glutaraldehyde solution in PBS was used for 15 min. After signal enhancement with a colloidal silver kit (BioCell, Cardiff, UK), sections were stained with 4% uranyl acetate for 4 min at 4°C and then with 0.4% uranyl acetate in 2% methylcellulose at 4°C.

Transmission electron microscopy.

A Zeiss EM 906 transmission electron microscope was used for image acquisition. Brightness and contrast of electron micrographs were adjusted in FIJI (v.2.0).

Statistical analysis.

GraphPad Prism (v.7) was used for statistical analysis. Comparisons between multiple groups were performed using one-way ANOVA followed by a Tukey posttest. P values of <0.05 were considered statistically significant.

RESULTS

Low-K⁺ and high-K⁺ diets rapidly alter NCC phosphorylation.

First, we aimed to induce a state of hypokalemia and NCC activation in mice as a model to examine WNK body formation. To this end, we fed mice previously maintained on a normal-salt/normal-K⁺ (NS/NK) diet with a normal-salt/low-K⁺ (NS/LK) diet for 12, 24, 48, and 72 h. Plasma K⁺ concentration was significantly lower (P < 0.05) compared with baseline at all four time points (Fig. 1A). The degree of activating NCC phosphorylation at Thr⁵³ (pNCC) and total NCC (tNCC) abundance was evaluated by Western blot analysis and densitometry. The pNCC-to-tNCC ratio (pNCC/tNCC) was significantly increased throughout the investigated time points (+350, +360, +300, and +320% for 12, 24, 48, and 72 h, respectively, P < 0.05; Fig. 1B).

Fig. 1.

Inverse effects of short-term low-K⁺ (LK) or high-K⁺ (HK) diets on plasma K⁺ concentration and Na⁺-Cl⁻ cotransporter (NCC) phosphorylation in mice. NS, normal salt; NK, normal K⁺. A: mice that received the NS/NK diet were switched to the NS/LK diet for 72 h (n = 6 at 0 h and n = 7 at each subsequent time point). Plasma K⁺ concentration was significantly lower within 12 h (−0.831 mM, 95% confidence interval: −0.4503 to −1.212). No further significant changes to plasma K⁺ concentration occurred over a further 60 h under the NS/LK diet. B: the ratio of NCC phosphorylated at Thr⁵³ (pNCC) to total NCC (pNCC/tNCC) significantly increased within 12 h of switching mice from NS/NK to NS/LK diets. It then remained stable for the next 60 h under the NS/LK diet (+350, +360, +300, and +320% at 12, 24, 48, and 72 h of the NS/LK diet, respectively). The original Western blot using antibodies to pNCC and tNCC is shown at the right. C: mice (n = 7) that received the NS/NK diet were switched to the NS/LK diet for 48 h and then to the NS/HK diet for the next 48 h. Plasma K⁺ concentration was lower in animals on the NS/LK diet than on the NS/NK diet (−0.917 mM, 95% confidence interval: −0.573 to −1.26) and higher in those on the NS/HK diet than on the NS/NK diet (+0.898 mM, 95% confidence interval: +1.241 to +0.554). D: the switch from the NS/LK diet to the NS/HK diet reduced pNCC/tNCC. The original Western blot is shown at the right. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 by one-way ANOVA followed by a Tukey posttest.

To investigate the reversibility of the effects of the NS/LK diet, mice were fed the NS/LK diet for 48 h and then switched to a normal-salt/high-K⁺ (NS/HK) diet for a further 48 h. Control groups included mice fed the NS/LK diet for 48 h and mice maintained on the NS/NK diet. The lowered plasma K⁺ concentration induced by 48 h on the NS/LK diet was restored by 48 h on the NS/HK diet (from 3.3 ± 0.1 to 5.1 ± 0.1 mmol/L, P < 0.05 for NS/HK vs. NS/LK diets and P < 0.05 for NS/HK vs. NS/NK diets; Fig. 1C). pNCC/tNCC was reduced after the NS/HK diet (−69%, P < 0.05 for NS/HK vs. NS/LK diets; Fig. 1D). pNCC/tNCC was not significantly difference for NS/HK versus NS/NK diets.

Low-K⁺ diets provoke the formation of puncta containing WNK4, SPAK, and pSPAK/OSR1, which are dispersed by high-K⁺ diets.

The WNK bodies described by Boyd-Shiwarski et al. (6) are strikingly similar to our previous observation of punctate accumulations of WNK/SPAK kinases in mice that received a NS/LK diet (29). To characterize the content of the puncta and kinetics of their formation in greater detail, we evaluated the intracellular distribution of WNK4, SPAK, and pS-SPAK/OSR1 by immunofluorescence in mice that received the NS/LK diet for 12, 24, 48, or 72 h or the NS/NK diet. K⁺ depletion induced punctate and increased apical signals for all three antibodies in the DCT compared with the NS/NK diet (Fig. 2A). Since WNK body formation has also been reported after dietary K⁺ loading (6), we performed analogous immunolabelings in mice fed the NS/HK diet for 48 h after spending the previous 48 h on the NS/LK diet. In contrast to the previously published findings, WNK4, SPAK, and pS-SPAK/OSR1 puncta were absent from the DCT after the NS/HK diet (Fig. 2B).

Fig. 2.

Low-K⁺ (LK) diet induces putative with no lysine (WNK) bodies containing Ste20-related proline-alanine-rich kinase (SPAK), phosphorylated SPAK/oxidative stress-responsive kinase-1 (OSR1), and WNK4 in the cytoplasm of the distal convoluted tubule (DCT). NS, normal salt; NK, normal K⁺; LK, low K⁺; HK, high K⁺; pS-SPAK/OSR1, SPAK/OSR1 phosphorylated at Ser^383/325. A: immunofluorescent labeling of several Na⁺-Cl⁻ cotransporter (NCC)-activating kinases, SPAK, pS-SPAK/OSR1, and WNK4, showed an increasing presence of punctate foci and intensified apical signals in the DCT of mice switched from the NS/NK diet to the NS/LK diet for 72 h (n = 3, 2 male mice and 1 female mouse; male mouse is shown). B: the NS/LK diet was stopped after 48 h, and animals (n = 3) were then fed the NS/HK diet for 48 h; this abolished the punctate signals of all three antibodies. DCT profiles are indicated by dashed red lines.

High-salt/low-K⁺ diets induce the formation of WNK bodies in which catalytically active WNK/SPAK isoforms accumulate.

By itself, salt loading reduces NCC activity (24, 37); however, in mice fed a high-salt/low-K⁺ (HS/LK) diet, NCC activity is increased (29). These findings show that the NCC-stimulating effect of K⁺ deprivation takes precedence over the NCC-suppressing effect of salt loading. We examined whether the redistribution of WNK/SPAK kinases to puncta upon K⁺ deprivation is similarly unaffected by concomitant salt loading. We fed HS/LK, high-salt/normal-K⁺ (HS/NK), and NS/NK diets to mice for 10 days. Using an antibody to pS-SPAK/OSR1, punctate signals were observed in mice fed the HS/LK diet but were virtually absent in mice fed the HS/NK and NS/NK diets. Puncta were localized in the DCT but absent from the thick ascending limb (TAL), as demonstrated by coimmunolabeling of NKCC2 and NCC as marker proteins of the TAL and DCT (Fig. 3A). In mice fed the HS/LK diet and the two control groups, the fraction of DCT cytoplasmic area occupied by pS-SPAK/OSR1-positive puncta was calculated. The area occupied by puncta was significantly greater in mice fed the HS/LK diet compared with the HS/NK or NS/NK diets (Fig. 3B).

Fig. 3.

Diets with high-salt (HS) and low-K⁺ (LK) content induce ultrastructurally distinct with no lysine (WNK) bodies in the distal convoluted tubule (DCT), containing phosphorylated WNK/Ste20-related proline-alanine-rich kinase (SPAK)/oxidative stress-responsive kinase-1 (OSR1) isoforms. NS, normal salt; NK, normal K⁺; HK, high K⁺; pS-SPAK/OSR1, SPAK/OSR1 phosphorylated at Ser^383/325. A: immunofluorescent labeling of pS-SPAK/OSR1 produced a punctate cytoplasmic signal in the DCT of mice fed the HS/LK diet for 10 days (n = 5, all male mice). This signal was reduced in mice fed the HS/NK or NS/NK diets for 10 days (each n = 5). The apical membrane of the DCT (dct) was identified by Na⁺-Cl⁻ cotransporter (NCC) immunolabeling, and the apical membrane of the thick ascending limb (TAL, tal) was identified by Na⁺- K⁺-Cl⁻ cotransporter 2 (NKCC2) immunolabeling. The punctate signal (arrows) in the HS/LK diet-fed group appeared in the first DCT cells beyond the DCT-TAL border (white line). pS-SPAK/OSR1 localized to DCT and TAL apical compartments in all three experimental groups. By differential interference contrast (DIC), DCT and TAL morphology was normal in all three groups. B: quantification of the area fraction of the DCT cytoplasm occupied by pS-SPAK/OSR1-positive puncta (n = 5 animals/dietary group). ***P < 0.001 by ANOVA and a Tukey post hoc test. C: immunogold labeling (silver enhanced) of SPAK in mice fed the HS/LK diet. Gold particle accumulation is shown in a perinuclear hypodense region (frame), i.e., the ultrastructural correlate of WNK bodies.

The equivalence of these immunoreactive puncta and the WNK bodies described in the previously published literature appeared likely based on their common link to K⁺ intake and frequent perinuclear location. To confirm this, we examined the DCT of HS/LK diet-fed mice by TEM. In immunogold labeling of ultrathin cryosections, antibody to SPAK labeled hypodense, perinuclear cytoplasmic regions, i.e., the ultrastructural equivalent of WNK bodies (Fig. 3C).

To investigate which NCC-regulating kinases are found in WNK bodies and to determine their activation state, we conducted immunofluorescent labelings using antibodies to several members of the WNK-SPAK/OSR1 pathway and their phosphorylated species. The presence of a DCT-specific, punctate signal in HS/LK diet-fed mice indicated that WNK1, WNK4, SPAK, OSR1, catalytically active (pS-)WNK, and catalytically active (pT-)SPAK/OSR1 were WNK body components (Fig. 4, A–F). Antibodies to catalytically active WNKs and SPAK/OSR1 delivered stronger apical signals in the DCT compared with the non-phospho-specific antibodies. Colocalization of the punctate signals shown in Figs. 3 and 4 was confirmed by double immunolabelings (see Supplemental Fig. S1, available online at https://doi.org/10.6084/m9.figshare.9897836.v1).

Fig. 4.

Under the high-salt (HS)/low-K⁺ (LK) diet, punctate immunofluorescence in the distal convoluted tubule (DCT) cytoplasm was produced by with no lysine (WNK)1 (A), WNK4 (B), Ste20-related proline-alanine-rich kinase (SPAK; C), and oxidative stress-responsive kinase-1 (OSR1; D) antibodies. E and F: antibodies to phosphorylated WNK1/WNK4 (Ser^383/332) (E) and to SPAK/OSR1 phosphorylated at Thr^243/185 (pT-SPAK/OSR1; F) produced punctate cytoplasm along with a strong apical signal. DCT profiles are indicated by dashed red lines.

WNK bodies locate to the vicinity of the endoplasmic reticulum, microtubules, and Golgi apparatus.

Defining the spatial relationships between WNK bodies, cellular organelles and the cytoskeleton could provide hints toward WNK body function. However, they remain poorly characterized, particularly as ultrastructural detail is limited in published TEM micrographs of WNK bodies due to preparation of samples for immunogold labeling (5, 6). In samples embedded in epoxy resin to achieve greater morphological preservation, we observed that WNK bodies had formed at the border of the endoplasmic reticulum. WNK bodies displayed no delineating membrane (Fig. 5A). We further examined colabelings of WNK body constituents and microtubule- or Golgi-specific proteins by STED superresolution and conventional confocal microscopy. Labeling of β-tubulin and WNK4 showed WNK bodies arranged alongside microtubules (Fig. 5B). WNK bodies labeled by an antibody to SPAK were located close to Golgi dictyosomes, as visualized by Gm130 immunolabeling (Fig. 5C). Colocalization between 20S proteasomes and putative WNK bodies has also been previously reported (8). In contrast, we observed disparate punctate signals produced by antibodies to 20S proteasomes and pS-SPAK/OSR1 (Fig. 4D). WNK bodies further did not contain the aggresomal proteins histone deacetylase-6 (HDAC6; Supplemental Movie S1, available online at https://doi.org/10.6084/m9.figshare.8104790.v2) and p62/sequestosome 1 (Supplemental Movie S2, available online at https://doi.org/10.6084/m9.figshare.8104796.v1) or the intermediate filament vimentin (Supplemental Movie S3, available online at https://doi.org/10.6084/m9.figshare.8104787.v1), which is enriched around aggresomes.

Fig. 5.

Subcellular localization of with no lysine (WNK) bodies within distal convoluted tubule (DCT) cells. All images are of mice fed a high-salt (HS)/low-K⁺ (LK) diet for 10 days (n = 5, all male mice). OSR1, oxidative stress-responsive kinase-1; SPK, Ste20-related proline-alanine-rich kinase; pS-SPAK/OSR1, SPAK/OSR1 phosphorylated at Ser^383/325. A: electron micrograph of epoxy resin-embedded tissue. The representative, hypodense WNK body (*) next to the nuclear membrane (nm) is not membrane bound and borders on endoplasmic reticulum (er) profiles. B: WNK bodies, here immunostained for WNK4, were located alongside microtubules visualized by β-tubulin immunofluorescence. C: WNK bodies, here immunostained for SPAK, were typically located near Golgi dictyosomes visualized by Gm130 immunofluorescence. D: WNK bodies, here immunostained for pS-SPAK/OSR1, were not colocalized with smaller cytoplasmic puncta containing proteasomal 20S subunit-immunoreactive signal. In B and C, conventional confocal overviews are on the left and stimulated emission depletion superresolution fluorescence micrographs of insets are on the right. D: conventional confocal micrograph.

WNK4-deficient mice develop large WNK bodies lacking catalytically active WNK/SPAK/OSR1.

The presence of phosphorylated WNK/SPAK/OSR1 isoforms within WNK bodies raises the possibility that WNKs may activate SPAK/OSR1 in the WNK body environment. In oocytes, KS-WNK1/WNK4 dimers are strong activators of SPAK, and subsequently NCC, by facilitating WNK4 autophosphorylation (3). Based thereon, and as KS-WNK1 and WNK4 are WNK body components, we speculated that recruitment of WNK4 into WNK bodies might be required for WNK4 and SPAK phosphorylation. To examine this hypothesis, we used WNK4-deficient mice, which display a phenotype of reduced NCC phosphorylation, renal K⁺ wasting, and hypokalemia. In the DCT of WNK4-deficient mice fed standard chow, we detected WNK bodies using antibodies to WNK1 (Fig. 6A), SPAK (Fig. 6B), and OSR1 (Fig. 6C). At the laser intensity showing prominent WNK bodies in WNK4-deficient mice, wild-type (WT) controls showed little detectable WNK1 signal. WT mice displayed a weaker punctate SPAK signal and no punctate OSR1 signal. Apical SPAK and OSR1 were suppressed in WNK4-deficient animals compared with WT control animals. Antibodies to pS-WNK, pS-SPAK/OSR1, and pT-SPAK/OSR1 did not label the WNK bodies of WNK4-deficient animals, and apical signals in the DCT were weak or absent (Fig. 6, D–F). All three phospho-antibodies showed a stronger apical signals in the DCT of control animals. These findings suggest that WNK4 may mediate SPAK/OSR1 phosphorylation within WNK bodies and that this reaction may be required for SPAK/OSR1 to exit WNK bodies toward the apical membrane to phosphorylate NCC. Since the formation of WNK bodies in WNK4-deficient mice may be driven by hypokalemia, we next fed the NS/HK diet to WNK4-deficient mice to normalize their plasma K⁺ levels. After 7 days, two of the four WNK4-deficient mice involved in the experiment exhibited plasma K⁺ concentrations comparable with those of WT mice on the NS/NK diet, whereas the other two mice remained hypokalemic (Supplemental Fig. S2, available online at http://doi.org/10.6084/m9.figshare.9897842). In contrast to the rapid disappearance of WNK bodies in WT mice fed the NS/HK diet (Fig. 2B), WNK4-deficient mice still exhibited abundant WNK bodies after the NS/HK diet independently of their plasma K⁺ concentration (see Supplemental Fig. S2). These were predominantly located in the early DCT (DCT1). These results suggest that WNK4 plays an important role in adaptations to changes in dietary K⁺ load, especially in DCT1.

Fig. 6.

Presence of large with no lysine (WNK) bodies in WNK4-deficient [knockout (KO)] mice. OSR1, oxidative stress-responsive kinase-1; SPK, Ste20-related proline-alanine-rich kinase; pS- SPAK/OSR1, SPAK/OSR1 phosphorylated at Ser^383/325; pT- SPAK/OSR1, SPAK/OSR1 phosphorylated at Thr^243/185. A–C: WNK4-KO mice (n = 3, 2 male mice and 1 female mouse; female mouse shown) displayed large WNK bodies containing WNK1 (A), SPAK (B), and OSR1 (C) immunoreactive signals in the distal convoluted tubule (DCT), whereas apical SPAK and OSR1 signals were reduced compared with wild-type (wt) control mice. Note the small size of SPAK- and OSR1-immunoreactive puncta in wt mice. D: apical pS-WNK signals were weaker, and WNK bodies were undetectable, in WNK4-KO versus wt mice. E: pS-SPAK immunoreactivity was nearly absent in WNK4-KO mice but located apically in wt mice. F: pT-SPAK/OSR1 immunoreactivity was nearly absent in WNK4-KO mice, whereas apical signal was observed in wt mice.

Basal WNK body abundance is reduced in hyperkalemic WNK4 transgenic mice with a pseudohypoaldosteronism type II phenotype.

To test whether WNK body formation correlates with WNK4 activity, we examined WNK4-transgenic mice carrying a gain-of-function mutation of wnk4 mirroring the human pathology of pseudohypoaldosteronism type II (WNK4-PHAII). WNK4-PHAII mice display a phenotype of increased NCC phosphorylation, renal K⁺ retention, and hyperkalemia (16). Compared with WT control animals, WNK4-PHAII animals showed reduced punctate SPAK signals in the DCT (Supplemental Fig. S3, available online at https://doi.org/10.6084/m9.figshare.9897851.v1). Thus, hyperkalemia caused by PHAII-inducing wnk4 mutations or high-K⁺ diets were associated with reduced punctate SPAK accumulation, which supports the notion that WNK body formation is inversely correlated with plasma K⁺. These results further suggest that formation of WNK bodies does not necessarily coincide with increased WNK4 or NCC activity.

Kir4.1 is required for the formation of WNK bodies after NS/LK diets.

Previously, we (9) demonstrated that the basolateral K⁺ channel Kir4.1 functions as a K⁺ sensor required for the appropriate adjustment of NCC phosphorylation to plasma K⁺ concentration. We hypothesized that WNK bodies could be essential for this mechanism and therefore that WNK body formation might depend on Kir4.1. We evaluated the distribution of known WNK body constituents in kidney-specific Kir4.1-deficient mice fed the NS/LK diet for 72 h. Labeling of Kir4.1 revealed that Kir4.1 deletion was incomplete, resulting in a mosaic pattern of Kir4.1-deficient and Kir4.1-expressing DCT cells. A comparison of this pattern with serial sections labeled with antibodies to WNK4 (Fig. 7A), SPAK (Fig. 7B), or pS-SPAK/OSR1 (Fig. 7C) showed that WNK bodies were induced only in Kir4.1-expressing cells, indicating that WNK body formation requires sensing of plasma K⁺ concentration. DCT cells were identified by concomitant labeling of parvalbumin.

Fig. 7.

Basolateral K⁺ channel Kir4.1 is required for with no lysine (WNK) body formation upon dietary K⁺ restriction. OSR1, oxidative stress-responsive kinase-1; SPK, Ste20-related proline-alanine-rich kinase; pS- SPAK/OSR1, SPAK/OSR1 phosphorylated at Ser^383/325. A: conditional kidney-specific Kir4.1 knockout (KO) mice (n = 3, 2 male mice and one female mouse; male mouse shown) displayed a mosaic pattern of Kir4.1-positive and -negative cells in the distal convoluted tubule (DCT). After 72 h on the normal-salt/low-K⁺ diet, Kir4.1-expressing DCT cells (arrows) contained WNK4-immunoreactive WNK bodies. This signal was absent in Kir4.1-negative DCT cells (arrowheads), as determined by labeling of serial sections. DCT cells were identified by staining of parvalbumin (PVA). B and C: SPAK (B) and pS-SPAK/OSR1 antibodies (C) also displayed WNK body signals in Kir4.1-positive cells (arrows), whereas WNK bodies were absent in Kir4.1-negative cells (arrowheads).

DISCUSSION

WNK kinases form key signaling components of a renal “K⁺ switch”. They act, in part, by phosphorylating the downstream kinases SPAK and OSR1, which, in turn, phosphorylate and activate NCC. This reduces distal Na⁺ delivery and therefore K⁺ secretion. When plasma K⁺ concentration is low, WNKs, SPAK, and OSR1 form punctate structures, recently named WNK bodies by Boyd-Shiwarski et al. (6). That group demonstrated that KS-WNK1 is essential for the formation of WNK bodies and suggested that they serve a physiologically important function.

When speculating about the influence of WNK bodies on NCC, a salient question regards the phosphorylation state of the kinases they contain. The published literature is divided on this issue. Several publications have shown that phosphorylated WNK1 is not found in WNK bodies (1, 2, 6). Our previous data (29), on the other hand, identified phosphorylated WNKs/SPAK/OSR1 as WNK body components, which we were able to reproduce in this study. Although the antibody to phosphorylated WNKs we used is not isoform specific, we showed that it does not label WNK bodies developed by WNK4-deficient mice, indicating that the most abundant activated WNK isoform in WNK bodies is WNK4. Full-length WNK1 would be expected to be phosphorylated in WNK4-deficient mice due to their hypokalemic phenotype (7). The absence of phosphorylated WNKs in the WNK bodies of WNK4-deficient mice further suggests that the predominant WNK1 isoform in WNK bodies is KS-WNK1, which is not autophosphorylated.

The activity of WNK4 is determined by the intracellular Cl⁻ concentration ([Cl⁻]_i), with high [Cl⁻]_i exerting an inhibitory effect. However, recent data obtained in Xenopus laevis oocytes suggest that the sensitivity of WNK4 to [Cl⁻]_i may be reduced by its heterodimerization with KS-WNK1, thereby allowing WNK4 autophosphorylation and subsequent SPAK/OSR1 phosphorylation despite high [Cl⁻]_i (3). Those authors hypothesized that this might be of relevance for WNK body function. The spatial concentration of KS-WNK1, WNK4, and SPAK/OSR1 within WNK bodies may facilitate the assembly of active KS-WNK1/WNK4 dimers in proximity to their substrates SPAK/OSR1. After being phosphorylated, SPAK/OSR1 could be exported to the apical membrane to activate NCC. Impairment of SPAK/OSR1 export out of WNK bodies represents a possible explanation of the enlargement of WNK bodies observed in WNK4-deficient mice, in which nonphosphorylated SPAK/OSR1 accumulate, suggesting that phosphorylated SPAK/OSR1 are specifically targeted for export. The reduced apical presence of SPAK/OSR1 in the DCT displayed by WNK4-deficient mice is consistent with this theory. The arrangement of WNK bodies along microtubules observed in this study and in patients with hypokalemia (31) may be pertinent to apical trafficking of SPAK/OSR1.

WNK4 deficiency is associated with hypokalemia (7). To explore whether this represents the driving force of WNK body formation in WNK4-deficient mice, we attempted to normalize their plasma K⁺ concentration by dietary K⁺ loading. In contrast to WT mice, which consistently display increases of plasma K⁺ upon K⁺ loading, two of four WNK4-deficient mice remained hypokalemic, likely due to persistent potassium wasting in the connecting tubule/collecting duct as a result of reduced DCT function. Whereas K⁺-deprived WT mice responded to normalization of plasma K⁺ concentration with a resolution of WNK bodies, WNK4-deficient mice showed persisting WNK bodies independently of plasma K⁺ concentration, which suggests that WNK4 is critical for the regulation of WNK body dynamics.

To explore the possibility that WNK bodies might represent a step toward degradation of excessively produced NCC-activating kinases, we tested the localization of the aggresomal proteins HDAC6, vimentin, and aggresomal/autophagosomal protein p62 in mice fed the HS/LK diet. All three products were not detected in WNK bodies, which is in contrast with our previous documentation of HDAC6 in WNK bodies of patients with hypokalemia (31). Therefore, species-related differences in WNK body composition are possible. Alternatively, the presence of HDAC6 in WNK bodies of patients with hypokalemia may reflect pathophysiological changes occurring in hypokalemic nephropathy (31). Our study supports previous publications (6, 8) stating that WNK bodies are neither aggresomes nor autophagosomes and rather suggests that WNK bodies are organelle-like structures permitting adaptations of NCC function to changes in K⁺ homeostasis.

WNK bodies become prominent when NCCs are activated by dietary K⁺ restriction (29), yet genetically engineered models have shown that NCC activity and WNK body formation can be dissociated. For example, NCC activity is high when WNK bodies are absent in mice lacking KS-WNK1 or in models of PHAII due to wnk1 or wnk4 mutations (11, 33, 36). To date, the relevance of WNK bodies to WNK-SPAK/OSR1 signaling has remained elusive. Wu et al. (36) provided functional data showing that the inhibition of NCC upon dietary K⁺ loading is blunted in mice lacking KS-WNK1. This observation is consistent with preliminary data provided by Boyd-Shiwarski et al. (5), who found that the ability of KS-WNK1-deficient mice to reduce NCC phosphorylation in hyperkalemia and to increase NCC phosphorylation in hypokalemia are both compromised. This bidirectional role of KS-WNK1 suggests that the formation and resolution of WNK bodies mediates adaptations of NCC function to the needs of K⁺ homeostasis. In line with this idea, we failed to detect any significant punctate accumulation of WNKs and SPAK/OSR1 in the DCT of normokalemic WT mice, although antibodies to SPAK/OSR1 produced small puncta. Presumably, these puncta also may contain WNKs but their detection may have been unsuccessful due to the low signal-to-noise ratio of the respective antibodies and the relatively modest baseline expression of the kinases.

The present data indicate that WNK bodies do not form when DCT cells cannot sense plasma K⁺ concentration due to genetic deletion of the basolateral K⁺ channel-forming protein Kir4.1 (8). Conversely, when NCC is overactivated in hyperkalemic WNK1- or WNK4-mutated mice, in which K⁺ sensing is presumably preserved, WNK body formation is suppressed. Thus, WNK body formation appears to specifically mediate the response of the DCT to changes in plasma K⁺ concentration, whereas other stimuli may not require WNK bodies for the modulation of NCC activity. Since the effects of Kir4.1 are thought to be mediated by Cl⁻ efflux (9), WNK bodies may represent a structural correlate of [Cl⁻]_i sensing in the DCT.

To conclude, we provide evidence that WNK bodies play a key functional role during changes in plasma K⁺ concentration via WNK4-induced SPAK/OSR1 activation. We found that 1) WNK bodies form and disperse rapidly upon dietary K⁺ perturbation; 2) WNK body formation requires the ability of DCT cells to sense plasma K⁺ concentration; and 3) WNK4-deficient mice develop enlarged WNK bodies but no apical SPAK/OSR1, suggesting that translocation of SPAK/OSR1 from WNK bodies to the apical membrane requires their activation by WNK4. A schematic interpretation of these results is shown in Figs. 8 and 9. Coupled with earlier data, the present work suggests that WNK bodies may be key structures permitting plasma K⁺ concentration to be transduced to NaCl transport activation.

Fig. 8.

Distribution of with no lysine (WNK)/Ste20-related proline-alanine-rich kinase (SPAK)/oxidative stress-responsive kinase-1 (OSR1) kinases in the distal convoluted tubule (DCT) under normal-K⁺ or K⁺-deficient diets or upon genetic WNK4 deletion. Labeling of WNK1 delivered no apical or punctate signals in mice fed normal-K⁺ diets, whereas K⁺-deprived mice developed WNK1-positive WNK bodies (*). WNK4-deficient [knockout (KO)] mice displayed larger WNK1-positive bodies. WNK4 was not detectable in puncta or in the apical compartment in mice fed normal-K⁺ diets. In K⁺ deficiency, labeling of WNK4 showed WNK bodies and a discrete apical signal. SPAK and OSR1 were detected in small WNK bodies and in the apical compartment in normal-K⁺ diet-fed mice. K⁺-deficient mice displayed larger WNK bodies and intensified apical signals for both SPAK and OSR1. In WNK4-KO mice, SPAK and OSR1 located to especially large WNK bodies, whereas apical signals were suppressed. Antibodies to phosphorylated (p)SPAK/OSR1 and pWNK produced apical signals in mice fed normal-K⁺ diets. In K⁺-deficient mice, WNK bodies were detected by both pSPAK/OSR1 and pWNK antibodies and apical signals were intensified. WNK4-KO mice displayed no signal in either pSPAK/OSR1 or pWNK labeling (*). Numerous small WNK bodies were visualized under normal-K⁺ diets with antibodies to SPAK and OSR1. Since kidney-specific WNK1 is an essential component of WNK bodies, absence of WNK1 labeling in normal-K⁺ diet-fed mice likely reflects a poorer signal-to-background ratio of the WNK1 antibody rather than a complete absence of WNK bodies.

Fig. 9.

Hypothesized signaling in distal convoluted tubule (DCT) cells in normokalemia versus hypokalemia. A: in normokalemia, K⁺ efflux via the basolateral Kir4.1 channel and subsequent Cl⁻ efflux via the basolateral Cl⁻ channel Kb (ClCKB channel) are low, with no lysine (WNK)4 therefore inhibited by high intracellular Cl⁻ concentration ([Cl⁻]_i). Consequently, Ste20-related proline-alanine-rich kinase (SPAK) and Na⁺-Cl⁻ cotransporter (NCC) activation are impeded. B: in hypokalemia, K⁺ and Cl⁻ efflux are increased, and kidney-specific (KS-)WNK1 forms WNK bodies (yellow circle). WNK4 and SPAK are subsequently recruited to WNK bodies, in which WNK4 and KS-WNK1 form dimers, enabling WNK4 autophosphorylation by relieving inhibition of WNK4 by Cl⁻. WNK4 then activates SPAK by phosphorylation (p), thereby enabling trafficking of SPAK to the apical membrane to activate NCC. In parallel, basolateral Cl⁻ efflux disinhibits free cytoplasmic WNK4 not bound to KS-WNK1, providing an alternative pathway to NCC activation. Oxidative stress-responsive kinase-1 (OSR1) may take the place of SPAK. [K⁺]_e, extracellular K⁺ concentration.

GRANTS

This work was financially supported by the Deutsche Forschungsgemeinschaft (to S. Bachmann; Grants MU 2924/2-1,2 and BA 700/22-1,2 and INST 335/596-1 FUGG), National Institute of Diabetes and Digestive and Kidney Diseases Grants R01 DK-098141 (to J. A. McCormick), and R01 DK-051496 and R01 DK-054983, Department of Veterans Affairs Grant I01 BX002228, and Transatlantic Network of ExcellenceGrant 17CVD05 from Fondation Leducq (to D. H. Ellison).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

M.N.T., L.N.M., J.S., X.-T.S., C.-L.Y., J.A.M., J.H., D.H.E., S.B., and K.M. conceived and designed research; M.N.T., C.A.C., T.M.B., L.N.M., J.S., R.J.C., X.-T.S., J.H., D.H.E., S.B., and K.M. performed experiments; M.N.T. and C.A.C. analyzed data; M.N.T., C.A.C., L.N.M., J.S., X.-T.S., C.-L.Y., J.A.M., J.H., D.H.E., S.B., and K.M. interpreted results of experiments; M.N.T., L.N.M., J.A.M., J.H., and D.H.E. prepared figures; M.N.T., L.N.M., J.S., X.-T.S., C.-L.Y., J.A.M., J.H., D.H.E., S.B., and K.M. drafted manuscript; M.N.T., C.A.C., T.M.B., C.D., L.N.M., J.S., R.J.C., X.-T.S., C.-L.Y., J.A.M., J.H., D.H.E., S.B., and K.M. approved final version of manuscript; C.A.C., T.M.B., C.D., J.S., X.-T.S., C.-L.Y., J.A.M., and J.H. edited and revised manuscript.