Extracellular K+ rapidly controls NaCl cotransporter phosphorylation in the native distal convoluted tubule by Cl−‐dependent and independent mechanisms

High dietary potassium (K+) intake dephosphorylates and inactivates the NaCl cotransporter (NCC) in the renal distal convoluted tubule (DCT). Using several ex vivo models, we show that physiological changes in extracellular K+, similar to those occurring after a K+ rich diet, are sufficient to promote a very rapid dephosphorylation of NCC in native DCT cells. Although the increase of NCC phosphorylation upon decreased extracellular K+ appears to depend on cellular Cl− fluxes, the rapid NCC dephosphorylation in response to increased extracellular K+ is not Cl−‐dependent. The Cl−‐dependent pathway involves the SPAK/OSR1 kinases, whereas the Cl− independent pathway may include additional signalling cascades.


Introduction
A high dietary potassium (K + ) intake has antihypertensive effects and improves cardiovascular outcomes O'Donnell et al. 2014). These beneficial effects of a K + rich diet are probably related to a negative Na + balance as a result of enhanced renal Na + excretion (Sorensen et al. 2013;Mente et al. 2014;Buendia et al. 2015;Penton et al. 2015). Indeed, oral K + loading promotes a rapid natriuresis that coincides with a dephosphorylation, and thus inactivation, of the thiazide-sensitive NaCl cotransporter (NCC) in the renal distal convoluted tubule (DCT) (Sorensen et al. 2013). Conversely, dietary K + restriction increases NCC phosphorylation (Vallon et al. 2009) and cell surface abundance of NCC (Frindt & Palmer, 2010), and also promotes a salt-sensitive rise in blood pressure (Vitzthum et al. 2014) that depends on the presence of NCC (Terker et al. 2015). Although an increase in NCC abundance was not always observed with dietary K + restriction (Nguyen et al. 2012), the critical role of NCC for the maintenance of Na + homeostasis and the control of blood pressure is indictated by thiazide diuretics being a mainstay of anti-hypertensive therapy. Furthermore, genetic diseases featuring loss-of-function mutations in NCC cause hypotension, while mutations in NCC-regulating kinases [with no Lysine (K) (WNK) WNK1 and WNK4] and ubiquitin-protein ligase complexes (KLHL3, cullin) are related to hypertension (Simon et al. 1996;Wilson et al. 2001).
The WNK kinases mediate their effects via phosphorylation of the STE20/SPS-1-44 related prolinealanine-rich protein kinase (SPAK) and the related oxidative stress-related kinase (OSR1), which directly activate NCC by N-terminal phosphorylation (Hoorn et al. 2011a). Based on studies in heterologous expression systems, it was proposed that the WNK and SPAK/OSR1 kinase pathway is critically involved in the effect of dietary K + intake on NCC activity. In this model, altered dietary K + intake impacts on the extracellular K + concentration ([K + ] ex ), which leads to changes in basolateral plasma membrane voltage and thus intracellular [Cl − ] ([Cl − ] i ). Changes in [Cl − ] i have been shown to modulate the activities of both WNK1 (Piala et al. 2014) and WNK4 (Bazúa-Valenti et al. 2015), which ultimately regulate SPAK/OSR1 phosphorylation and activity, and thus NCC (Terker et al. 2015). However, given the lack of an appropriate cell model that resembles the physiological complexity of native DCT cells, the relevance of this pathway for the DCT in the kidney in vivo is unclear. Indeed, the results from SPAK knockout (Wade et al. 2015), SPAK knock-in (Castañeda-Bueno et al. 2014) and SPAK/OSR1 knockout mouse models (Terker et al. 2015) have indicated that additional mechanisms must contribute as well. Moreover, the rapidity of NCC dephosphorylation in response to an increased dietary K + load suggests that not only an inactivation of the phosphorylating kinase pathway, but also an activation of yet undefined dephosphorylating protein phosphatases (PPs), probably contributes towards the control of NCC phosphorylation by plasma [K + ]. Several PPs have been implicated in the modulation of NCC phosphorylation, including PP1 (Picard et al. 2014), PP3 (calcineurin) ) and PP4 (Glover et al. 2010). Nevertheless, to date, no specific physiological function has been assigned to any of these phosphatases in the DCT.
In the present study, we investigated the response of native DCT cells to changes in [K + ] ex by combining in vivo and ex vivo methods. Our data support the hypothesis that the regulation of NCC phosphorylation triggered by changes in plasma K + occurs very rapidly and is mediated by Cl − -dependent and independent mechanisms that involve SPAK/OSR1 and probably additional signalling pathways.
Technologies (Beverly, MA, USA). Total amount of NCC protein (tNCC), pT53NCC, pT58NCC and pS71NCC antibodies have been described previously (Wagner et al. 2008;Sorensen et al. 2013;Picard et al. 2014). The pT96-pT101 NKCC2 antibody was generated by immunization of rabbits with a phospho-peptide (NH2-CLQ(pT)FGHN(pT)MD-CONH2) corresponding to the amino acids of mouse NKCC2 and subsequent affinity purification of the antibody (Pineda, Berlin, Germany). The specificity of the Ab towards pT96-pT101 NKCC2 was confirmed by immunohistochemistry. The antibody specifically stains the apical plasma membrane of the thick ascending limb. Pre-incubation of the antibody with the above mentioned phospho-peptide, but not with the corresponding dephospho-peptide abolished the immunostaining (data not shown). β-actin antibody was purchased from Sigma-Aldrich. Anti-Serine/threonine-protein kinase 39 (referred to as tSPAK in the present study) and anti-phospho-SPAK (Ser 373)/phospho-OSR1 (Ser 325) (referred to as pSPAK-pOSR1 in the present study) antibodies were purchased from Millipore (Millipore, Temecula, CA, USA).

Animals
Experiments were conducted in C57Bl/6 wild-type mice aged between 3 and 8 months. Mice were maintained under a 12:12 h light/dark cycle and had access to standard chow (3430; Provimi-Kliba, Penthalaz, Switzerland) and water ad libitum. Animals were age-, weight-and sex-matched for each experimental series.

KCl I.V. injections
Twelve hours before the experiments, age-and weight-matched female animals were fasted, although free access to water was maintained. Animals were fixed in a supine position for surgery under anaesthesia with isoflurane (Attane, Piramal, India). After median laparotomy, the vena cava was punctured using a 26-G needle. Either 100 μl of 80 mmol l -1 KCl (in distilled H 2 O with NaCl added to 300 mosmol kg -1 ) or 0.9% NaCl as control were randomly injected. Either 5 or 15 min later, blood was withdrawn via the right ventricle. Blood was directly centrifuged, and the plasma collected and stored at -20°C for further analysis. Animals were then perfused with 20 ml of ice-cold PBS via the abdominal aorta, and then the kidneys were removed, decapsulated and snap frozen for western blot analysis.

Isolated perfused mouse kidney
Isolated mouse kidney perfusion was performed at 37°C in a small animal perfusion system (Hugo Sachs Elektronik, March, Germany) as described previously (Schweda et al. 2003). Age-and weight-matched male animals were used. In all experiments, kidneys were first perfused for 30 min (pressure-fixed 90 mmHg) with buffer (for precise buffer composition, see Supporting information, Table S1) containing 4.5 mmol l -1 K + . Afterwards, buffer [K + ] was randomly adapted by adding KCl to a K + concentration ranging between 3.2 and 10.38 mmol l -1 , and kidneys were perfused for additional 30 min. After 1 h, perfusion was stopped and kidneys were snap frozen in liquid nitrogen for western blot analysis. NCC phosphorylation in perfused kidneys was normalized to NCC phosphorylation in contralateral kidneys from same animals, which were removed and snap frozen at the beginning of the procedure. For immunohistochemistry studies, after perfusion with either 4.5 or 10 mmol l -1 [K + ] for 1 h, 3% paraformaldehyde was added to the perfusion buffer. Then, 50 mL of buffer + fixative were infused at 200 mmHg and washed out with 50 mL of buffer without fixative at 120 mmHg. The kidneys were post-fixed in 1.5% paraformaldehyde at 4°C overnight and processed as described previously for immunofluorescence microscopy (Sorensen et al. 2013).

Kidney slice preparation
Sex-, age-and weight-matched mice were used for preparation of kidney slices. Both kidneys were quickly removed under deep isoflurane anaesthesia (Attane) and 280 μm thick slices were cut with a vibrating microslicer (Vibratome, Microm; Thermo Scientific, Waltham, MA, USA). Slicing was performed in ice-cold Ringer-type solution (in mmol l -1 ): 98.5 NaCl, 35 NaHCO 3 , 3 KCl, 1 NaH 2 PO 4 , 2.5 CaCl 2 , 1.8 MgCl 2 and 25 glucose. The slices were incubated for equilibration in a similar Ringer-type solution (25 mmol l -1 NaHCO 3 ) for 30 min at 30.5°C. Afterwards, the slices were transferred to the incubation chambers containing the effector or control solutions. All solutions were continuously bubbled with 95% O 2 and 5% CO 2 . In experiments in which [K + ] ex was changed, osmolality and [Cl − ] ex were kept constant by addition of equal amounts of NaCl (or Na-Gluconate in low [Cl − ] ex experiments). Some slices exposed to varying [K + ] ex were incubated also in the presence isoproterenol (100 nmol l -1 ) or calyculin A (20 nmol l -1 ) dissolved in DMSO, or tacrolimus (10 μmol l -1 ) dissolved in ethanol. In parallel, slices were incubated with either DMSO or ethanol in the appropriate amount and used as control vehicle. After 30 min of incubation with the effectors, slices were snap frozen in liquid nitrogen and processed for immunoblotting.

Reversibility of NCC dephosphorylation
After equilibration under control conditions (3 mmol l -1 [K + ] ex ) for 30 min, kidney slices were transferred to J Physiol 594.21 incubation chambers containing either 1, 3 or 10 mmol l -1 [K + ] ex (low, control, high [K + ] ex respectively) and incubated for another 30 min, as described previously. Next, slices previously incubated with 10 mmol l -1 [K + ] ex were transferred to the 1 mmol l -1 [K + ] ex chamber and incubated for a further 30 min. After a total incubation time of 90 min, slices from each incubation condition were collected and processed for immunoblotting.

Aldosterone measurements
For aldosterone measurements, 10 μl of plasma was diluted in 40 μl of 0.9% NaCl and measured in triplicates using an Aldosterone ELISA (enzyme-linked immunosorbent assay) Kit (CAN-ALD-450; Diagnostics Biochem Canada Inc., Dorchester, ON, Canada) and a Tecan plate reader (Tecan infinite F200 Pro; Tecan Group AG, Männedorf, Switzerland) in accordance with the manufacturer's instructions.

Plasma electrolyte analysis
Plasma electrolyte levels were measured in 100 μl of undiluted plasma using an EFOX 5053 flame photometer (Eppendorf, Hanburg, Germany).

Immunoblotting
Kidney samples were homogenized in a Precellys 24 homogenizer (Bertin Technologies, Rockville, MD, USA) or by ultrasound in lysis buffer containing protease inhibitor cocktail (Complete; Roche Diagnostics, Basel, Switzerland), phosphatase inhibitor (Phosstop; Roche Diagnostics), and (in mmol l -1 ) 200 mannitol, 80 Hepes and 41 KOH (pH corrected to 7.4). Homogenized samples without any further manipulation, or the supernatant from the samples centrifuged for 10 min 4000 g at 4°C (whole animal studies or isolated perfused kidney), were then used for western blotting. Protein concentration was assessed via Bradford assay (CooAssay Protein Dosage Reagent; Uptima, Montluçon, France). Then, 25 μg (or 50 μg for the detection of SPAK and pSPAK-pOSR1) of protein was solubilized in loading buffer (31.5 mmol l -1 Tris-HCl, 1% SDS, 0.005% Bromphenol blue, 12.5% glycerol and 5% β-mercaptoethanol, pH 6.5), run on 8% SDS-PAGE and transferred to nitrocellulose membranes. Membranes were blocked for 20 min in blocking buffer (Odyssey blocking buffer; Li-Cor, Lincoln, NE, USA) and incubated with primary antibodies at 4°C overnight. Membranes were further incubated for 2 h at room temperature with the appropriate secondary antibodies and imaged using a Li-Cor infrared scanner (Li-Cor). Optical density of bands was quantified using ImageJ (http://imagej.nih.gov/ij) and normalized to the corresponding band densities in a Coomassie-stained gel (ProtoBlue Safe; National Diagnostics, Hessle, UK) that was loaded and run in parallel. A protein dilution series was performed for each antibody and confirmed that signal detection and analysis were performed in the linear range of the system. Detection of β-actin was used to visualize equal loading.

Statistical analysis
An unpaired Student's t test was used to compare between two groups. For multiple comparison, one-way ANOVA with Bonferroni's multiple comparison post hoc test was performed using OriginPro, 2015 (OriginLab Corp., Northampton, MA, USA).

Small variations of plasma [K + ] rapidly modulate NCC phosphorylation
In previous experiments in mice, we found that oral K + loading rapidly increased plasma [K + ] and dephosphorylated NCC (Sorensen et al. 2013). Similarly, further analysis of rat and mouse models suggested an inverse correlation between plasma [K + ] and NCC phosphorylation (Rengarajan et al. 2014;Terker et al. 2016). To test whether altered plasma [K + ] could directly account for the observed rapid changes of NCC phosphorylation in the kidney, we performed two types of experiments. First, we I.V. injected 100 μl of either vehicle or 80 mmol l -1 KCl in mice. The KCl injection increased   Table S2). G, immunofluorescence staining of pT53NCC (upper images) and the closely-related cotransporter pNKCC2 phosphorylated in T96 and T101 (lower images) in kidneys perfused with 4.5 or 10 mmol l -1 [K + ] ex . Scale bar = 250 µm.
plasma [K + ] significantly (Fig. 1A), whereas the plasma concentrations of Na + and aldosterone remained largely unchanged within the time frame analysed ( Fig. 1B and C). NCC phosphorylation at all tested sites (T53, T58 and S71) was decreased 5 min after KCl injection; the effect was even more pronounced at 15 min (Fig. 1D). Next, we perfused isolated mouse kidneys ex vivo with buffers containing different concentrations of K + , ranging from 3.2 to 10.38 mmol l -1 . To counterbalance the inter-individual variation, NCC phosphorylation in the perfused kidney was normalized to NCC phosphorylation in the intact contralateral kidney from the same animal (Fig. 1E).
We observed a clear inverse dose-response relationship between [K + ] and NCC phosphorylation ( Fig. 1F; see also Supporting information, Table S2) with the most pronounced changes in NCC phosphorylation occurring between 3 and 5 mmol l -1 of [K + ]. Consistent with previous in vivo data (Sorensen et al. 2013;Rengarajan et al. 2014;Terker et al. 2015), the effect of K + on the isolated perfused kidney ex vivo was specific for NCC because the phosphorylation of the closely-related transporter NKCC2 in the thick ascending limb of the loop of Henle remained stable, regardless of [K + ] (Fig. 1G).

The kidney slice model reproduces the in vivo findings
The isolated perfused mouse kidney (IPK) model allowed us to assess the effect of K + on the DCT in an intact organ at the same time as avoiding the systemic effects of confounding factors such as hormones and renal innervation. Nevertheless, with this system, we could not exclude the influence of changes in tubular flow and intrarenal haemodynamics on DCT function. Given the lack of an appropriate DCT cell model suitable to assess the direct effect of [K + ], we used simple kidney slice preparations to study ex vivo acute changes in NCC phosphorylation in native DCT cells. First, we studied the behaviour of tNCC and phosphorylated NCC in kidney slices incubated with control Ringer-type buffer (3 mmol l -1 [K + ] ex ) during the course of 1 h. Under these conditions, the abundance of tNCC remained stable between 15 and 60 min of incubation, whereas pT53NCC decreased continuously with time ( Fig. 2A and B). As we observed a stable NCC phosphorylation between 30 and 60 min of incubation, all further experiments were performed during this time frame, unless stated otherwise.
First, we analysed the effect of elevated [K + ] ex . Although [K + ] did not affect tNCC, the drop in pT53NCC over time was steeper in slices incubated with 10 mmol l -1 [K + ] ex compared to those incubated in 3 mmol l -1 [K + ] ex buffer ( Fig. 2C and D).

NCC phosphorylation is modified by manipulations that affect DCT membrane voltage
The membrane voltage of DCT cells is considered to be controlled by the activity of the basolateral K + channel Kir 4.1 (KCNJ10 gene) (Reichold et al. 2010;Zhang et al. 2014), which functions as heterotetramer with Kir 5.1 (KCNJ16). Under physiological conditions, KCNJ10/16 is assumed to hyperpolarize DCT basolateral plasma membrane voltage (Paulais et al. 2011). Thus, increasing [K + ] ex should depolarize DCT cells, whereas hyperpolarization will occur under low [K + ] ex . We investigated whether manipulations promoting changes in the membrane voltage of DCT cells were sufficient to modulate NCC phosphorylation. The phosphorylation of NCC was assessed by immunoblotting after incubating kidney slices for 30 min with different [K + ] ex . Although tNCC remained stable under all [K + ] ex investigated ( Fig. 3A and B), pT53NCC showed an inverse dose-response curve in response to changes in [K + ] ex ( Fig. 3A and C). In agreement with the results obtained in vivo and with the IPK model, physiologically relevant variations in [K + ] ex were already sufficient to induce statistically significant changes in NCC phosphorylation ([K + ] ex 3 mmol l -1 = 100 ± 4.7% of pT53NCC; [K + ] ex 5 mmol l -1 = 56.0 ± 7.0 % of pT53NCC; P < 0.001). Furthermore, NCC dephosphorylation could be partially reversed by transferring the slices from high [K + ] ex (10 mmol l -1 ) to low [K + ] ex (1 mmol l -1 ) (Fig. 3D). Moreover, the addition of the potassium channel blocker BaCl 2 (5 mmol l -1 ) was sufficient to promote the dephosphorylation of NCC ( Fig. 3E and F).
DCTs express K + channels on both the apical and the basolateral membrane (Bandulik et al. 2011) and these channels could potentially sense changes in K + intake. To assess whether changes of [K + ] ex on the basolateral (blood) side are sufficient to promote changes in NCC phosphorylation, DCTs were isolated by hand and exposed at the basolateral cell side to either 1 or 10 mmol l -1 [K + ] ex . The lumen was microperfused with 2 mmol l -1 K + and 50 mmol l -1 Na + , which recapitulates the physiological luminal K + and Na + concentrations, as observed previously by in vivo micropuncture measurements (Weinstein, 2012). Subsequently, changes in the phosphorylation of NCC were assessed by immunoblotting. As shown in Figure 3G, lowering basolateral [K + ] ex to 1 mmol l -1 triggered the phosphorylation of NCC within 10 min; at 10 mmol l -1 of basolateral [K + ] ex , the phosphorylation of NCC was almost completely abolished.

Low [Cl − ] ex clamps SPAK/OSR1 phosphorylation but does not prevent NCC dephosphorylation in response to high [K + ] ex
Changes in intracellular [Cl − ] i have been proposed to underlie the modifications of NCC phosphorylation in response to altered [K + ] ex Terker et al. 2015). According to this hypothesis, [K + ] ex affects membrane voltage, which changes the driving force for Cl − to exit the cell through chloride channels in the basolateral membrane of DCT cells. Thus, a hyperpolarized cell membrane will lead to an increased Cl − exit, therefore lowering its intracellular concentration; in contrast, depolarized cells will retain Cl − . These secondary changes in [Cl − ] i were proposed to regulate the phosphorylation of SPAK and NCC as a result of the inhibitory effect of [Cl − ] i on WNK1 and WNK4 (Alessi et al. 2014). Indeed, increasing [K + ] ex promotes both NCC and SPAK/OSR1 dephosphorylation (Fig. 4A, C and D).
To test the hypothesis that cellular Cl − fluxes and altered SPAK activity contribute to the regulation of NCC by [K + ] ex , we compared kidneys slices incubated with either a standard buffer (110 mmol l −1 [Cl − ] ex ) or a buffer in which almost all Cl − was replaced by using sodium   (Fig. 4A, C and D). Lowering [Cl − ] ex to 5 mmol l -1 significantly increased NCC phosphorylation (Fig. 4C). Under this low [Cl − ] ex condition, we did not see any increase in either SPAK/OSR1 or NCC phosphorylation in response to lowering [K + ] ex from 3 (control) to 1 (low) mmol l -1 (Fig. 4B, C and D). Nevertheless, NCC phosphorylation under low [Cl − ] ex conditions could still be enhanced by isoproterenol, which was previously described as an activator of NCC phosphorylation (Terker et al. 2014) (Fig. 4G). This indicates that the lack of response to low [K + ] ex is not explained by an already maximal stimulation of NCC. By contrast to the diminished response to low [K + ] ex , NCC was still dephosphorylated when [K + ] ex was increased from 3 (control) to 10 (high) mmol l -1 ( Fig. 4B and C). The extent of the dephosphorylation was comparable to the one at baseline [Cl − ] ex (80.9 ± 11.8% ] ex (n = 6-7 slices per treatment, two mice). C, changes in NCC phosphorylation at position T53 upon incubation with different [K + ] ex (n = 8-17 slices per treatment, three to seven mice). * * P < 0.01, * * * P < 0.001; unpaired Student's t test compared to control [K + ] ex (3 mmol l -1 ). D, representative experiment showing the reversibility of NCC dephosphorylation triggered by high [K + ] ex . Control, high and low stand for 3, 10 and 1 mmol l -1 [K + ] ex respectively. This experiment was repeated twice (two mice) with similar results. E and F, effect of BaCl 2 (5 mmol l -1 ) on NCC phosphorylation (n = 6 slices per treatment, two mice). G, changes in NCC phosphorylation in hand isolated DCTs microperfused with 2 mmol l -1 [K + ] ex from the luminal side and either 1 or 10 mmol l -1 [K + ] ex from the basolateral side for 10 min. Each band represents a pool of 10 DCTs. * * * P < 0.001, ns, non-significant using an unpaired Student's t test. In (B), (C) and (F), points represent individual experiments (slices) and bars represent the mean of the given treatment.

. Effect of [K + ] ex on NCC and SPAK/OSR1 phosphorylation in native DCT cells is affected by changes in Cl − conductances
A, representative immunoblot showing the changes in tNCC and pT53NCC, as well as SPAK and pSPAK-pOSR1, in kidney slices treated with low (1 mmol l -1 ), control (3 mmol l -1 ) or high (10 mmol l -1 ) [K + ] ex under control (110 mmol l -1 ) [Cl − ] ex conditions. B, same experiment as in A but under 5 mmol l -1 [Cl − ] ex . C, summary of the effect of [K + ] ex on NCC phosphorylation at position T53 under control or low [Cl − ] ex conditions (n = 9-18 slices per treatment, three to six mice). D, summary of the effect of [K + ] ex on SPAK/OSR1 phosphorylation at position T53 under control or low [Cl − ] ex conditions (n = 6 slices per treatment, two mice). C and D, * P < 0.05, * * P < 0.01 and * * * P < 0.001, compared to the control condition ([K + ] ex 3 mmol l -1 ) of each [Cl − ] ex ; &P < 0.05 compared to the control condition of 110 mmol l -1 [Cl − ] ex using ANOVA with Bonferroni's multiple comparison post hoc test. E, representative experiment showing the changes in NCC expression and phosphorylation at T53 in the presence or absence of the [Cl − ] channel blocker DIDS upon treatment with control (3 mmol l -1 ) or high (10 mmol l -1 ) [K + ] ex . F, densitometric analysis of (E) (n = 3 slices per treatment, one mouse). * P < 0.05 compared to the control for each condition using an unpaired Student's t test. G, stimulation of NCC phosphorylation by isoproterenol (100 mmol l -1 ) under low (5 mmol l -1 ) [Cl − ] ex . J Physiol 594.21 vs. 72.5 ± 19.0%; under 110 vs. 5 mmol l -1 [Cl − ] ex , respectively). Interestingly, the dephosphorylation of NCC occurred independently of any significant change in SPAK/OSR1 phosphorylation, which remained clamped at a high level ( Fig. 4B and D). Consistent with the Cl − replacement studies, inhibition of plasma membrane Cl − fluxes with the Cl − channel blocker DIDS (0.5 mmol l -1 ) did not prevent NCC dephosphorylation in response to increasing [K + ] ex from 3 (control) to 10 (high) mmol l -1 (Fig. 4E and F).

High [K + ] ex induced dephosphorylation of NCC is not blocked by inhibition of PP1, PP2A and PP3
Given the evidence described above for SPAK independent effects of [K + ] ex on NCC, as well as the rapid kinetics of NCC dephosphorylation, we investigated whether some of the known NCC regulating PPs might be involved in mediating the effect of high [K + ] ex on NCC. Neither incubation of tissue slices with the PP1 and PP2A inhibitor calyculin A (Sorensen et al. 2013), nor the PP3 (calcineurin) inhibitor tacrolimus (Hoorn et al. 2011b), prevented the dephosphorylation of NCC in response to increased [K + ] ex on standard [Cl − ] ex (Fig. 5).

Discussion
Dietary K + intake has a strong impact on the phosphorylation, and hence activity, of NCC in the DCT. Previous studies have correlated these effects with altered plasma K + levels and suggested a model in which the extracellular K + concentration, via altered membrane voltage and intracellular Cl − concentrations, regulates the WNK-and SPAK-dependent phosphorylation of NCC. In the present study, we employed three different ex vivo models (i.e. the isolated perfused kidney, the isolated perfused tubule and kidney slices) to experimentally test this hypothesis on the native DCT. Our data provide clear evidence for rapid and direct effects of [K + ] ex on NCC phosphorylation and suggest that, apart from the Cl − -dependent WNK/SPAK kinase pathway, a Cl −and SPAK independent pathway also contributes to the regulation of NCC by [K + ] ex . Rengarajan et al. (2014) showed that 3 h of I.V. infusion of KCl reduces phosphorylation of NCC and marginally that of SPAK in parallel to increased plasma [K + ]. Similarly, for various mouse models, Terker et al. (2016) described a clear inverse relationship between plasma [K + ] and NCC phosphorylation, suggesting that plasma [K + ] may have direct effects on the native DCT. However, these data were correlative and did not allow for clear cause-effect conclusions. The possible regulatory effect of humoral kaliuretic factors and/or nerval reflexes (Rabinowitz, 1996) could not be excluded. By using three independent ex vivo approaches, we provide consistent and conclusive evidence that [K + ] ex directly modulates NCC phosphorylation in native DCT cells. This regulation occurs very rapidly (<30 min) and is particularly prominent in the range of physiological plasma [K + ], suggesting that transient and small variations in plasma (K + ), as they may occur in response to a meal, may already be sufficient to modulate NCC activity and hence renal Na + reabsorption. However, the present ex vivo data contrast with our previous data on isolated DCT preparations in which we failed to observe any significant NCC dephosphorylation after increasing extracellular K + from 5 to 10 mmol l -1 (Sorensen et al. 2013). As we demonstrate in the present study, 5 mmol l -1 is already on the upper [K + ] limit rendering almost all NCC dephosphorylated. Nevertheless, DCTs are still able to respond to decreased [K + ] ex , as demonstrated by the isolated perfused DCT technique.
Based on experiments using whole animal and heterologous expression system models, Terker et al. (2015) recently hypothesized that DCT cells depolarize upon an increase in plasma [K + ]. The subsequent increase in [Cl − ] i leads to decreased phosphorylation of NCC via the inhibition of the WNK-SPAK/OSR1 kinase pathway. In agreement with this hypothesis, in native DCTs, we found that both an increase of [K + ] ex and the addition of the K + channel blocker BaCl 2 to the incubation buffer triggered a pronounced dephosphorylation of NCC. By contrast, removal of extracellular Cl − , which presumably lowers intracellular Cl − , significantly increased NCC phosphorylation, and blocked the stimulatory effect of low [K + ] ex on NCC. However, although these data are suggestive for an involvement of plasma membrane voltage and intracellular [Cl − ], we could not directly analyse these parameters in our ex vivo preparations for technical reasons, including an insufficient up-take of Cl − -sensitive dyes into DCT cells. Other mechanisms, such as changes in cell volume, remain to be considered. Indeed, hypo-osmotic cell swelling stimulates SPAK-dependent NCC activation (Richardson et al. 2008) and recent data suggest that WNK4 is phosphorylated in response to osmotic stress (Maruyama et al. 2016).
Interestingly, when we reduced [Cl − ] ex to very low levels, thereby clamping SPAK-OSR1 phosphorylation to very high levels, only the response of NCC to low [K + ] ex , and not high [K + ] ex , was blocked. Similarly, blocking Cl − channels with DIDS did not prevent the dephosphorylation of NCC under high [K + ] ex , suggesting that additional mechanisms to the proposed pathway via Cl − -dependent changes in WNK-SPAK/OSR1 activity are also involved in the regulation of NCC phosphorylation by K + . It was already evident from the data provided by Terker et al. (2015) that the SPAK/OSR1 single and double knockout mice were still able to modulate NCC phosphorylation to some extent in response to low dietary K + intake. Moreover, Castañeda-Bueno et al. (2014) reported changes in NCC and its phosphorylation in response to changes in dietary K + in WNK4 knockout mice. Furthermore, Wade et al. (2015) reported an increase in NCC phosphorylation in response to low K + diet in SPAK knockout mice.
The rapidity of NCC dephosphorylation in response to K + infusion, already evident within 5 min, suggests that, in addition to a shut-down of NCC phosphorylation by kinases, an activation of NCC dephosphorylation via phosphatases may also contribute to the observed effects. Previous studies had implicated various phosphatases in the control of NCC function (Glover et al. 2010;Hoorn et al. 2011b;Picard et al. 2014). PP1 and PP3 are highly abundant in the DCT and inhibition of PP1 and PP3 by calyculin A and tacrolimus, respectively, increases NCC phosphorylation (Hoorn et al. 2011b;Picard et al. 2014). PP3 inhibition also probably explains the NCC-dependent salt-sensitive hypertension in patients under immunosuppressive therapy with calcineurin inhibitors (Hoorn et al. 2011b;Borschewski et al. 2016). Similarly, a kidney-specific deletion of FK506 binding protein, crucial for the effects of tacrolimus on PP3, attenuates tacrolimus-induced arterial hypertension . However, in our experiments, inhibition of PP1 and PP3 did not block the [K + ] ex -induced dephosphorylation of NCC, suggesting that other signalling cascades are involved.
In conclusion, in the present study, we show that [K + ] ex directly controls NCC phosphorylation in native DCT cells. The effect of low [K + ] ex on NCC probably involves altered transmembrane Cl − fluxes and an activation of SPAK/OSR1, whereas the rapid effects of high [K + ] ex do not critically depend upon transmembrane Cl − permeability and may include both inhibition of SPAK/OSR1 and activation of as yet unidentified signalling pathways. Based on the rapidity of NCC dephosphorylation in response to [K + ] ex , we hypothesize that PPs are involved. Our data, might stimulate further research to identify these additional mechanism. of the data and Ms Julia Graf for her contribution to the initial immunofluorescence characterization of the pNKCC2 antibody. We also thank Dr Eilidh Craigie for carefully proofreading the manuscript submitted for publication. A set-up for the isolated perfused kidney was kindly provided by Professor Wolf-Georg Forssmann.

Supporting information
The following supporting information has available in the online version of this article.