WNKs are potassium-sensitive kinases

Abstract

With no lysine (K) (WNK) kinases regulate epithelial ion transport in the kidney to maintain homeostasis of electrolyte concentrations and blood pressure. Chloride ion directly binds WNK kinases to inhibit autophosphorylation and activation. Changes in extracellular potassium are thought to regulate WNKs through changes in intracellular chloride. Prior studies demonstrate that in some distal nephron epithelial cells, intracellular potassium changes with chronic low- or high-potassium diet. We, therefore, investigated whether potassium regulates WNK activity independent of chloride. We found decreased activity of Drosophila WNK and mammalian WNK3 and WNK4 in fly Malpighian (renal) tubules bathed in high extracellular potassium, even when intracellular chloride was kept constant at either ∼13 mM or 26 mM. High extracellular potassium also inhibited chloride-insensitive mutants of WNK3 and WNK4. High extracellular rubidium was also inhibitory and increased tubule rubidium. The Na⁺/K⁺-ATPase inhibitor, ouabain, which is expected to lower intracellular potassium, increased tubule Drosophila WNK activity. In vitro, potassium increased the melting temperature of Drosophila WNK, WNK1, and WNK3 kinase domains, indicating ion binding to the kinase. Potassium inhibited in vitro autophosphorylation of Drosophila WNK and WNK3, and also inhibited WNK3 and WNK4 phosphorylation of their substrate, Ste20-related proline/alanine-rich kinase (SPAK). The greatest sensitivity of WNK4 to potassium occurred in the range of 80–180 mM, encompassing physiological intracellular potassium concentrations. Together, these data indicate chloride-independent potassium inhibition of Drosophila and mammalian WNK kinases through direct effects of potassium ion on the kinase.

INTRODUCTION

With no lysine (K) (WNK) kinases are important regulators of epithelial ion transport (1). WNKs are serine-threonine kinases that phosphorylate and activate the downstream Ste20-related proline/alanine-rich kinase (SPAK) and oxidative stress response (OSRI) kinases, which in turn phosphorylate and activate the sodium chloride cotransporter (NCC) and the related sodium-potassium-2-chloride cotransporters (NKCC1 and NKCC2) (2). Sodium chloride is reabsorbed through NCC and NKCC2 in the distal convoluted tubule and thick ascending limb of the mammalian kidney. Consequently, genetic perturbations of WNKs in mice and humans result in abnormalities in electrolyte homeostasis and blood pressure (1). We have shown that in Drosophila, WNK regulates epithelial ion transport in the Malpighian (renal) tubule by stimulating the SPAK/OSR1 homolog, Fray, and the fly NKCC (3).

NKCC transporters are activated by phosphorylation (4). Low intracellular chloride also activates NKCCs by stimulating NKCC phosphorylation (5–9). These findings predicted a chloride-inhibited protein kinase. In chloride-secreting epithelia, stimulation of apical chloride exit lowers intracellular chloride, activating the chloride-sensitive kinase. The subsequent phosphorylation and activation of the basolateral NKCC increases basolateral chloride entry, coupling apical, and basolateral chloride transport (10). The discovery that chloride directly inhibits WNK activation (11) provided a molecular mechanism linking intracellular chloride to the WNK-SPAK/OSR1-NKCC pathway and transepithelial chloride transport.

The Drosophila Malpighian tubule consists of cation-secreting principal cells and chloride-secreting stellate cells (12). We demonstrated that the basolateral NKCC, regulated by WNK and Fray, is required for principal cell potassium secretion (3, 13). Bathing Malpighian tubules in hypotonic bathing medium (standard bathing medium diluted with water) increases potassium secretion in a WNK-, Fray-, and NKCC-dependent manner (3). Under these conditions, intracellular chloride decreased from ∼27 mM, in standard bathing medium, to ∼15 mM, in hypotonic medium, and Drosophila WNK activity increased twofold over 30–60 min (14). The increase in WNK activity is consistent with chloride inhibition of WNK kinases. However, expression of a chloride-insensitive WNK mutant did not increase potassium secretion unless coexpressed with the scaffold protein, Mouse protein 25 (Mo25) (14). Furthermore, the principal cell secretes potassium. By analogy to intracellular chloride regulation of WNK in chloride-secreting epithelia, we wondered whether intracellular potassium also regulates WNK, which would allow coupling of apical potassium exit with basolateral potassium entry. Indeed, differences in sodium chloride and potassium chloride on WNK1 thermal stability suggested the possibility of potassium binding to WNKs (11). Based on the breakthrough in understanding that chloride directly regulates WNKs by binding to the kinase domain and inhibiting autophosphorylation (11), we hypothesized that potassium directly regulates WNK kinase activity.

Here, we demonstrate that high potassium concentrations inhibit the activity of Drosophila and mammalian WNKs expressed in the fly Malpighian tubule. Intracellular chloride was directly monitored using a transgenic chloride sensor and intracellular potassium was measured using substituted rubidium ion and inductively coupled plasma mass spectrometry (ICP-MS). Using experiments designed to allow potassium and chloride concentrations to be independently varied, we observed that potassium and chloride have additive inhibitory effects on WNK activity. In vitro studies demonstrate that these effects are due to direct regulation of kinase activity by potassium ion.

MATERIALS AND METHODS

Fly Stocks

Fly lines used are shown in Table 1. Flies expressing ClopHensor in the tubule were reared at 26°C. Other crosses were performed at 28°C, except in the case of experiments using tub-GAL80^ts20, in which case crosses were performed at 18°C. Adult females were collected within 1–2 days and tubules were dissected 3–5 days later. For experiments using tub-GAL80^ts20, flies were shifted to 28°C for 7 days before experimentation to induce adult-specific gene knockdown and expression.

Table 1.

Drosophila strains used

Genotype	Description	Source	Reference
w; c42-GAL4	Principal cell GAL4 driver	Dow/Davies lab	(15)
w; tub-GAL80ts20	Temperature-sensitive GAL4 repressor	BDSC 7019	(16)
w; UAS-DmWNKRNAi	RNAi against Drosophila WNK	BDSC 42521	(14)
w; UAS-ClopHensor c304	pH/chloride sensor	Krämer lab	(14)
w; UAS-SPAKD219A	WNK substrate	Rodan lab	(14)
w; UAS-WNK3WT	Human WNK3	Rodan lab	(17)
w; UAS-WNK3L295F	Chloride-insensitive human WNK3	This study
w; UAS-WNK4WT	Mouse WNK4	Rodan lab	(17)
w; UAS-WNK4L319F	Chloride-insensitive mouse WNK4	This study
w; UAS-Mo25	Mouse Mo25	Rodan lab	(14)

BDSC, Bloomington Drosophila Stock Center, Indiana University, Bloomington, IN; DmWNK, Drosophila WNK; Mo25, mouse protein 25; WNK, with no lysine (K) kinase.

Generation of transgenic fly lines: UAS-WNK3^L295F and UAS-WNK4^L319F.

The generation of UAS-WNK3^WT and UAS-WNK4^WT transgenic lines has been described (17). Open reading frames of pUAS plasmids were mutated using the QuikChange II Site-Directed Mutagenesis kit (Agilent) as per the manufacturer’s protocol. To generate the WNK3^L295F mutation, plasmid pUAS-HsWNK3^WT was mutated using primers 5′-ACTGGATCTGTGAAGATTGGTGATTTCGGATTAGCCACCTTAATG-3′ and 5′-CATTAAGGTGGCTAATCCGAAATCACCAATCTTCACAGATCCAGT-3′. To generate the WNK4^L319F mutation, plasmid pUAS-MmWNK4^WT was mutated using primers 5′-GTCAAAATCGGAGACTTCGGACTGGCCACGC-3′ and 5′-GCGTGGCCAGTCCGAAGTCTCCGATTTTGAC-3′. Mutations were confirmed by Sanger sequencing. Midiprep DNA was sent to Rainbow Transgenic Flies for microinjection into stock line No. 24483 (M{vas-int.Dm}ZH-2A, M{3xP3-RFP.attP}ZH-51 D). Transgenic lines were generated from single male transformants, and PCR confirmation of UAS-transgene was performed using sequence-specific primers. Transgenic lines were outcrossed for five generations to the Rodan laboratory wBerlin genetic background.

Intracellular Chloride and pH Measurements Using ClopHensor

pH calibration.

Tubules from female flies expressing ClopHensor in the principal cell (w; c42-GAL4/UAS-ClopHensor c304) were dissected from 3- to 5-day-old flies in Drosophila saline, consisting of (in mM): NaCl 117.5, KCl 20, CaCl₂ 2, MgCl₂ 8.5, NaHCO₃ 10.2, NaH₂PO₄ 4.3, HEPES 15, and glucose 20, pH 7.0. Tubules were attached to the bottom of 35-mm glass bottom dishes with 14-mm microwell/No. 1.5 cover glass (Cellvis) coated with poly-lysine, and the solution exchanged to pH-varied solution containing: 100 mM Na⁺ gluconate, 50 mM K⁺ gluconate, 8.5 mM MgCl₂, 2 mM CaCl₂, 20 mM glucose, 15 mM HEPES (varied pH), 10 µM tributyltinchloride (Sigma), 5 µM nigericin (Invitrogen), 5 µM carbonyl cyanide 3-chlorophenylhydrazone (Sigma), and 5 µM valinomycin (Sigma). After equilibration for at least 1 h, cells were imaged using a Zeiss LSM510 confocal microscope, with excitation at 488 nm (green emission) and 458 nm (cyan emission). Individual cells were then outlined and pixel intensity measured using ImageJ without image manipulation. The ratios of green/cyan versus pH were entered into GraphPad Prism, and a sigmoidal curve interpolated using the function “sigmoidal, 4PL, X is log(concentration).” This provided the values for the following equation, used to calculate intracellular pH (pHi) in the tubule epithelial cells (18):

$$p{H}_i=p{K}_a-\frac{1}{p}\mathrm{ }\times \log \left(\frac{B2-B1\mathrm{ }}{R_{pH}-B1}-1\right)$$

where R_pH is the experimentally derived green/cyan ratio, pK_a = 7.352, p = power (Hill slope, 1.656), and B1 (1.021) and B2 (3.172) are the minimum and maximum asymptotic values of R_pH. A separate calibration was performed for the pH measurements in the higher chloride (∼26 mM) baths, as these were performed at a different time. In that case, pK_a = 7.201, p = power (Hill slope, 3.835), and B1 (1.110) and B2 (1.826). pH calibration curves are shown in Supplemental Fig. S1 (all Supplemental material is available at https://doi.org/10.6084/m9.figshare.c.4961201).

Chloride calibration.

Tubules expressing ClopHensor in the principal cells (w; c42-GAL4/UAS-ClopHensor c304) were dissected from 3- to 5-day-old flies in Drosophila saline. Tubules were attached to the bottom of 35-mm glass bottom dishes with 14-mm microwell/No. 1.5 cover glass (Cellvis) coated with poly-lysine, and the solution exchanged to the chloride calibration solution, consisting of: 100 mM NaCl/gluconate, 50 mM K-Cl/gluconate, 2 mM Ca-Cl/gluconate, 8.5 mM Mg-Cl/gluconate, 20 mM glucose, 15 mM HEPES pH 7.2, 10 µM tributyltinchloride (Sigma), 5 µM nigericin (Invitrogen), 5 µM carbonyl cyanide 3-chlorophenylhydrazone (Sigma), and 5 µM valinomycin (Sigma). Cl/gluconate anions were adjusted to achieve varying chloride concentrations. After 1 h equilibration, tubules were imaged using a Zeiss LSM510 confocal microscope, with excitation at 543 nm (red emission) and 458 nm (cyan emission). Individual renal tubule epithelial cells were outlined and pixel intensity measured in ImageJ without image manipulation. The ratios of cyan/red versus chloride were entered into GraphPad Prism, and a sigmoidal curve interpolated using a logistic dose-response sigmoidal fit. This provided the values for the following equation, used to calculate intracellular chloride ([Cl⁻]_i) (18):

$$[\mathrm{C}{\mathrm{l}}^{-}{]}_i=K_d\mathrm{ }\times {\left(\frac{A1-A2}{R_{\mathrm{C}\mathrm{l}}-A2}-1\right)}^{\frac{1}{\mathrm{p}}}$$

where R_Cl is the experimentally derived cyan/red ratio, K_d = 11.26, p = power (Hill slope 1.616), and A1 (3.312) and A2 (1.673) are the maximum and minimum asymptotic values of R_Cl. This calibration curve was used to calculate intracellular chloride concentrations in the low intracellular chloride (∼13 mM) baths. A separate calibration was performed for the measurements of intracellular chloride in the higher chloride (∼26 mM) baths, as these were performed at a different time. In that case, K_d = 17.33, p = 1.622, A1 = 3.504, and A2 = 1.897. Chloride calibration curves are shown in Supplemental Fig. S2.

Measurement of R_pH and R_Cl and determination of varying potassium bathing media with intracellular chloride concentrations of ∼13 mM and 26 mM.

Tubules from adult female flies expressing ClopHensor in the principal cell (w; c42-GAL4/UAS-ClopHensor c304) were dissected in Drosophila saline at 3–5 days of age. Tubules bathed in varying potassium/chloride bathing solutions for 1 h were imaged as described in pH calibration and Chloride calibration. Individual epithelial cells were outlined in ImageJ, excluding the nucleus, and pixel intensity captured for each emission channel. The ratios of green/cyan and cyan/red were used to calculate pH and chloride as described in pH calibration and Chloride calibration; measurements of both ratios were made in the same cells. Bathing medium ion concentrations were then adjusted to achieve the same intracellular chloride concentrations in the varying potassium baths. For example, if intracellular chloride was too low in a particular bath, extracellular chloride was increased and the intracellular chloride of the complete set of low-, normal-, and high-potassium baths was measured again. An example of baths resulting in significantly different intracellular chloride concentrations is shown in Supplemental Fig. S3. Osmolality was adjusted using N-methyl-d-glucamine (NMDG)⁺ and gluconate⁻ salts and was measured in triplicate on an Advanced Instruments Osmometer, model 303 (Norwood, MA), according to the manufacturer’s instructions. Ultimately two sets of baths were designed with lower (∼13 mM) and higher (∼26 mM) intracellular chloride. The salines used for low-intracellular chloride baths are shown in Table 2. These were mixed 1:1 with Schneider’s medium, consisting of (in mM): glycine, 3.33; l-arginine, 2.3; l-aspartic acid, 3.01; l-cysteine, 0.496; L-cystine, 0.417; l-glutamic acid, 5.44; l-glutamine, 12.33; l-histidine, 2.58; l-isoleucine, 1.15; l-leucine, 1.15; l-lysine hydrochloride, 9.02; l-methionine, 5.37; l-phenylalanine, 0.909; l-proline, 14.78; l-serine, 2.38; l-threonine, 2.94; l-tryptophan, 0.49; l-tyrosine, 2.76; l-valine, 2.56; β-alanine, 5.62; CaCl₂, 5.41; MgSO₄, 15.06; KCl, 21.33; KH₂PO₄, 3.31; NaHCO₃, 4.76; NaCl, 36.21; Na₂HPO₄, 4.94; α-ketoglutaric acid, 1.37; d-glucose, 11.11; fumaric acid, 0.862; malic acid, 0.746; succinic acid, 0.847; trehalose, 5.85; and yeastolate, 2,000 mg/L. The 1:1 mixture was then diluted by adding 277 µL H₂O to 300 µL saline/Schneider’s mix. The salines used for higher intracellular chloride baths are shown in Table 3. These were mixed 4:1 with Schneider’s medium, then diluted by adding 80 µL H₂O to 300 µL saline/Schneider’s mix.

Table 2.

Salines for making varying K⁺ baths, low [Cl⁻]_i (∼13 mM)

	Low K+, mM	Normal K+, mM	High K+, mM
NaCl	115.5	104	60
Na gluconate	40	51.5	95.5
K gluconate		40	140
CaCl2	5.5	5.5
Ca gluconate			5.5
MgCl2	20	20
Mg gluconate			20
NaHCO3	17	17	17
NaH2PO4	7.5	7.5	7.5
HEPES	22	22	22
Glucose	35	35	35
NMDG gluconate	148	125

NMDG, N-methyl-d-glucamine.

Table 3.

Salines for making varying K⁺ baths, high [Cl⁻]_i (∼26 mM)

	Low K+, mM	Normal K+,, mM	High K+, mM
NaCl	117.5	117.5	91
Na gluconate			17
KCl	1	25.5
K gluconate			65.1
CaCl2	2	2	2
MgCl2	8.5	8.5	8.5
NaHCO3	10.2	10.2	10.2
NaH2PO4	4.3	4.3	4.3
HEPES	15	15	15
Glucose	20	20	20
NMDG Cl	27
NMDG gluconate	25	27

NMDG, N-methyl-d-glucamine.

Because of the mathematical transformation in the equations used to calculate R_Cl, cases in which R_Cl was approaching the asymptotic value of the calibration curve resulted in extreme (and physiologically implausible) values of chloride (e.g., 431 mM). Therefore, outliers were identified and removed in GraphPad Prism, v. 8, using the robust regression and outlier removal (ROUT) method to remove definitive outliers with high stringency (Q = 0.1%). This process resulted in the removal of the following number of values: for higher intracellular chloride, 4 of the low potassium, 3 of the normal potassium, and 1 of the high potassium values, and for lower intracellular chloride, 1 of the low potassium, 2 of the normal potassium, and 3 of the high potassium values.

Measurement of Tubule WNK Activity

Fifteen pairs of anterior renal tubules, expressing kinase-dead rat SPAK^D219A as a substrate for endogenous Drosophila WNK (14), or transgenically expressed wild-type or chloride-insensitive mammalian WNK3 or WNK4, were dissected from adult females in Drosophila saline. Tubule pairs were transferred to 300 µL normal potassium bath (with low- or high-intracellular chloride) for 1 h (Drosophila WNK and WNK3) or 2 h (WNK4) in a 9-well Pyrex dish. To prevent evaporation, Parafilm was used to cover the wells. After 1 h equilibration, the tubules were transferred to low- or high-potassium baths (with the same intracellular chloride) for 15–60 min. Tubules expressing Drosophila WNK and WNK3 were incubated at room temperature, ∼22°C. Tubules expressing WNK4 were incubated at 37°C, as this resulted in greater WNK4 activity and better ability to detect phosphorylated SPAK. Tubules were then transferred to 100 µL 2× Laemmli sample buffer.

30–50 µL lysate was used to detect phosphorylated SPAK (pSPAK) and total SPAK (tSPAK) by Western blotting, as previously published (14). Primary antibodies were used at 1:1,000 dilution: pSPAK (Ser373)/phospho-OSR1 (Ser325) (rabbit, Millipore No. 07–2273) and tSPAK [mouse, Cell Signaling Technology No. 2281, Abcam anti-STK39 (2E10) No. 117982, or GeneTex anti-STK39 (2E10) No. GTX83543]. We previously validated the specificity of the anti-pSPAK and anti-tSPAK antibodies (14). Secondary antibodies were used at 1:2,500 dilution: fluorophore-conjugated anti-rabbit AzureSpectra 800 (VWR, No. AC2134) or anti-mouse AzureSpectra 700 (VWR, No. AC2129). Protein bands were visualized using a c600 Azure Biosystems instrument. Because the primary antibodies were from different species, the pSPAK and tSPAK bands could be visualized on the same blot. Band intensities were quantified in ImageJ by manually outlining the bands and subtracting background pixel intensities from a nearby region. To account for day-to-day variability in the Western blotting procedure, the pSPAK/tSPAK ratio in each Western blot was normalized to the normal potassium control. Linearity of antibodies was validated by performing Western blots on varying quantities of tubule lysates (Supplemental Fig. S4).

In another series of experiments, the effect of rubidium on Drosophila WNK activity was evaluated. These were performed by substituting rubidium for potassium in the normal, low and high potassium/rubidium baths with intracellular chloride of ∼13 mM.

In experiments to examine the effect of ouabain on Drosophila WNK activity, tubules expressing kinase-dead rat SPAK^D219A were dissected from adult females in Drosophila saline, and then bathed for 1 h in standard bathing medium [1:1 mix of Drosophila saline and Schneider’s medium, (14)], with or without 100 µM ouabain or vehicle (DMSO) control. Western blotting for pSPAK and tSPAK was then performed.

Cellular Potassium and Rubidium Measurements Using Inductively Coupled Plasma Mass Spectrometry

Fourteen pairs of tubules expressing kinase-dead rat SPAK (w; UAS-SPAK^D219A/+; c42-GAL4/+) were placed into normal rubidium bath for 1 h, then transferred to low-, normal-, or high-rubidium bath for 30 min. A fourth group of tubules was incubated in modified Drosophila saline, containing (in mM): NaCl 117.5, RbCl 20, MgCl₂ 8.5, CaCl₂ 2, glucose 20, NaHCO₃ 10.2, NaH₂PO₄ 4.3, HEPES 8.6, and glutamine 5, pH 7.0, as this solution (containing potassium instead of rubidium) was previously used to measure intracellular potassium concentrations using a different method (19) and served as an additional control. Tubules were washed by dipping in 50 µL deionized water and then immediately frozen at −20°C before analysis by ICP-MS. For each ICP-MS replicate, 14 pairs of tubules were digested using a 5:1 mixture of nitric acid (OPTIMA grade, 70%, Thermo Fisher Scientific) and ultrapure hydrogen peroxide (ULTREX II, 30%, Thermo Fisher Scientific). This mixture was allowed to digest overnight, heated until dry, and resuspended in 2% nitric acid for analysis. An Agilent 7900 ICP-MS was operated in helium (He) collision cell gas mode for all measurements. Elements were measured at the following isotopes: ⁸⁵Rb, ⁶³Cu, ⁶⁶Zn, ⁵⁶Fe, ³⁹K, ⁷²Ge, and ⁴⁵Sc. Calibration standards and samples were prepared in an acid matrix of 2% OPTIMA Grade Nitric Acid. Solutions of Agilent Multi-element Calibration Standard 2A were prepared to obtain an eight-point calibration curve. A rubidium stock solution was used as an independent measure. Agilent Germanium (or Scandium) Standard(s) were added online to calibration standards, blanks, and samples, and were used to correct for potential sample matrix and/or nebulization effects (isotopes ⁷²Ge or ⁴⁵Sc). Control digestions were used to measure background.

Protein Reagents for Differential Scanning Fluorimetry and Autophosphorylation Assays

His-tagged Drosophila WNK (DmWNK) kinase domain, residues 261–534, and rat WNK1 kinase domain, residues 194–483 were cloned, expressed, and purified as previously described (14, 20). As purified from Escherichia coli, WNK1 is phosphorylated, and was dephosphorylated to obtain uWNK1 (unphosphorylated WNK1), also as previously described (20). Human WNK3, residues 118–409 was codon optimized for bacterial expression (Genscript, NJ), and subcloned into a pET29b vector. Expression and purification were as reported for the kinase domain of WNK1 (20). Like WNK1, the WNK3 kinase domain is phosphorylated as expressed in E. coli and was dephosphorylated with λ phosphatase and PP1cγ using 10/1 and 5/1 ratios, respectively, to produce uWNK3 (unphosphorylated WNK3). Phosphatases were removed using Ni-NTA and gel filtration chromatography. WNK1 and WNK3 dephosphorylation was confirmed by mass spectrometry. Residual phosphatase activity was checked using p-nitrophenyl phosphatase (G-Biosciences, St. Louis, MO) following published protocols (21). uWNK3 was buffer exchanged into 50 mM HEPES, pH 7.4, 150 mM NaCl. DmWNK purified from E. coli is unphosphorylated at the activation loop, as assessed by mass spectrometry (data not shown). This may reflect decreased activity of DmWNK as compared with uWNK1 or uWNK3 (see results). The control p38 MAP kinase was expressed in E. coli and purified as published previously (22). p38 is unphosphorylated as expressed. A plasmid for apoptosis signal-regulated kinase 1 (ASK1) (659–951) was obtained from Stefan Knapp, Structural Genomics Consortium at Oxford (23). ASK1 is phosphorylated as expressed in E. coli. Expression, purification, and dephosphorylation was conducted as published (23).

Differential Scanning Fluorimetry

Briefly, 5 µM DmWNK, uWNK1, and uWNK3 kinase domains, and control kinases p38 and ASK1, were treated similarly. The protein concentration was 5 µM. Kinases were incubated with 10 mM Tris (pH 8.0), 5× SYPRO Orange (Invitrogen), water, and the specified salt concentration to make a 20-µL reaction volume for a well on a 96-well clear bottom plate. The temperature was increased from an initial 4°C to 80°C using 0.5°C increments in a Bio-Rad CFX96 real-time PCR machine. The fluorescence intensity of SYPRO Orange was probed using the fluorescein amidite channel of the PCR machine.

In Vitro Autophosphorylation Assays

Phosphorylation assays.

Phosphorylation assays were carried out at 30°C under standard conditions of 20 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM ATP, and varying potassium gluconate concentrations as indicated in Fig. 5. Reactions were initiated by the addition of 20 µM DmWNK or uWNK3 kinase domain, with aliquots removed at each time point. WNK autophosphorylation was terminated by addition of guanidine-HCl to 1 M before proteolysis for HPLC-MS analysis. The design allows for uniform sample handling, improving the comparison of the mass spectrometry signals for each time point.

Figure 5.

Potassium inhibits Drosophila with no lysine (K) kinase (WNK) and human WNK3 autophosphorylation in vitro. A: the time course of Drosophila WNK kinase domain (amino acids 261-534) in vitro autophosphorylation, in the presence of varying concentrations of potassium gluconate, is shown. Autophosphorylation at Ser 434 was assessed by LC-MS at the indicated times. B: human WNK3 kinase domain (amino acids 118-409) autophosphorylation at Ser 308 was followed as in A. In A and B, n = 3 independent experiments. Means ± standard deviation shown; in most cases standard deviation was smaller than the symbols. There were significant effects of potassium on autophosphorylation. In two-way ANOVA with mixed-effects analysis, P < 0.0001 for time, potassium concentration, and interaction. In Tukey’s multiple-comparison testing, all potassium concentrations were significantly different (P < 0.05) from all other concentrations at each time point, except for 80 mM vs. 120 mM potassium at 1 h in A. Raw data sets are available in the Supplemental Material.

Peptide standards for mass spectrometry.

Isotopically labeled WNK peptide standards corresponding to phosphorylated and unphosphorylated chymotrypsin-derived activation loop peptides (AKSVIGTPEFMAPEMY and AKS*VIGTPEFMAPEMY, with S* denoting phospho-serine) were synthesized (21st Century Biochemicals, Marlboro, MA) (24). This peptide is the same for DmWNK and WNK3. These standards were used to optimize HPLC peptide separation, determine elution times, and for relative quantitation. Representative ion traces are shown in the Supplemental Material, “WNK Ion Extraction Traces.”

Protein digestion and HPLC peptide separation.

Chymotrypsin protease reaction mix [containing 100 mM Tris-HCl, pH 8.3, 10 mM CaCl₂, and sequencing grade chymotrypsin (Roche Applied Science)] was used to digest DmWNK and uWNK3. All proteolysis reactions were conducted at a 100:1 protein:protease molar ratio. Activation loop peptides were separated by an Agilent 1100 series LC system (Agilent Technologies, Palo Alto, CA) with an RP-C18 microbore HPLC column (Phenomenex Aeris WIDEPORE 150 2.1 mm, 3-µm particle size, 200 Å pore diameter). Peptides were eluted using a water/acetonitrile gradient with 0.2% formic acid. DmWNK and WNK3 activation loop peptides were eluted at 21%–23% acetonitrile. Peptides were separated from salts on the HPLC column by diverting early eluents.

LC-MS/MS analysis.

HPLC-MS/MS analysis was performed on an LCQ DECA XP ion trap mass spectrometer (ThermoFinnigan) with the HPLC in-line to an orthogonal electrospray ionization source. Integration under ion traces corresponding to mass ranges for the activation loop peptides were used to acquire raw MS detector responses. Time courses for the appearance of peptides were conducted in triplicate. The raw MS responses were integrated in Xcalibur 2.0.7 (ThermoFinnigan) and scaled to peptide standards, to obtain a ratio of phosphorylated to unphosphorylated peptide and percent phosphorylation. MS/MS spectra were acquired in a data-dependent manner for each MS scan (ThermoFinnigan). MASCOT (v. 2.2, Matrix Science Ltd., London, UK) (25) and MassMatrix (26) software was used for peptide confirmation from tandem mass spectrometry spectra, using a custom search database specific for the recombinant WNK proteins studied.

Kinase-Glo in Vitro Kinase Assay

Kinase-Glo assays were performed in 25 μL volumes containing 50 mM HEPES, pH 7.4, 20 mM MgCl₂, 110 mM NaCl, 240 mM potassium gluconate or sodium gluconate, 7 μM ASK1, 60 μM myelin basic protein, and 50 μM ATP. Bovine myelin basic protein was purified from brain acetone powder (Sigma) as previously described (27). Zero minute time point reactions were used to define maximum luminescence. Reactions performed without ATP were used to define minimum luminescence. Assays were stopped at indicated time points by the addition of 20 μL Kinase-Glo Reagent (Promega), then transferred to a 96-well plate and centrifuged. Luminescence was read on a CLARIOstar plate reader (BMG Labtech, Ortenberg, Germany). The data were processed using MARS data analysis software (BMG Labtech).

WNK4 and WNK3 In Vitro Kinase Assays

Cloning of HA-WNK4 and HA-WNK3 for expression in Drosophila S2 cells.

To generate the pAc5-HA-WNK plasmids, mouse Wnk4 and human WNK3 were PCR amplified from existing clones in the laboratory [originally obtained from Chou-Long Huang, UT Southwestern, see Stenesen et al. (17)]. An HA tag was added to the 5′ end by PCR. The primers used for WNK4 were: 5′-ATGTACCCATACGATGTTCCAGATTACGCTATGCTAGCACCTCGAAATAC-3′ and 5′-CATCCTGCCAATATCCCCGGCGAATG-3′. The primers used for WNK3 were: 5′-ATGTACCCATACGATGTTCCAGATTACGCTATGGCCACTGATTCAGGGGATCC-3′ and 5′-TTTAGGACCAGGAGGGATTGTGGCAGG-3′. Then, a second round of PCR was performed to allow Gibson assembly. For WNK4, the primers used were: 5′-CCTACTAGTCCAGTGTGGTGGAATTCGCCACCATGTACCCATACGATGTTCCAGATT-3′ and 5′-CCGCATGTTAGAAGACTTCCTCTGCCCTCAAGCATCCTGCCAATATCCCCGG-3′. The primers used for WNK3 were: 5′-CCTACTAGTCCAGTGTGGTGGAATTCGCCACCATGTACCCATACGATGTTCCAGATT-3′ and 5′-CCGCATGTTAGAAGACTTCCTCTGCCCTCAAGTTTAGGACCAGGAGGGATTG-3′. The pAc5 STABLE2 Neo plasmid backbone, obtained from Addgene (plasmid No. 32426) (28), was amplified with 5′-CTTGAGGGCAGAGGAAGTCTTCTAACATGC-3′ and 5′-GTGGTGGAATTCGCCACC-3′. PCR was performed with Phusion DNA polymerase and HF buffer (New England Biolabs). Gibson assembly was performed with New England Biolab HiFi DNA assembly followed by transformation into E. coli α5 (Invitrogen). Clones were verified by restriction digest and Sanger sequencing.

Preparation of immunoprecipitated WNK3 and WNK4 for in vitro kinase assays.

Drosophila S2-GAL4 cells were maintained at 26°C in Schneider medium (Gibco) with 10% heat-inactivated FBS (Gibco). Cells (20 × 10⁶) were transfected with pAc5-HA-WNK4 or pAc5-HA-WNK3 (20 µg) using the TransIT insect transfection reagent (Mirus Bio) and harvested 48 h after transfection by centrifugation. The pellet was resuspended in 500 µL Thermo Co-IP lysis/binding/wash buffer (Invitrogen) with 2× protease inhibitor cocktail (Invitrogen) and incubated on ice for 5 min, followed by centrifugation. About 50 µL anti-HA magnetic beads (Invitrogen) were equilibrated with Tris buffered saline-0.1% Tween-20 (TBS-T). The TBS-T was removed and 500 µL cell lysate supernatant was added to the equilibrated magnetic beads. The suspension was gently rotated at 4°C for 30 min. The beads were then immobilized and the unbound sample removed and saved. Beads were washed three times with wash buffer (Invitrogen, No. 1861279), followed by a final wash with 15 mM Tris/HCl pH 8.0. To verify effective immunoprecipitation by Western blotting, 50 µL Laemmli buffer (Bio-Rad) was added to immobilized beads, the mixture was heated at 95°C for 5 min, and the beads were removed by magnetic immobilization. Briefly, 20 µL of eluate was separated by 8% SDS-PAGE and Western blotting was performed using anti-HA primary antibodies (mouse, 1:1,000; Invitrogen, No. 26183) and goat anti-mouse secondary antibodies (1:5,000, Li-Cor, No. 926–68070). Protein bands were visualized using a c600 Azure Biosystems instrument.

Preparation of GST-SPAK^D219A.

The rat SPAK^D219A open reading frame was recombined into the pDEST15 vector (Thermo Fisher Scientific No. 11802014) from pENTR RnSPAK^D219A (14) using the LR Clonase II reaction (Invitrogen). Plasmids were verified by restriction digest.

For protein expression, pDEST15 RnSPAK^D219A plasmid was transformed into competent E. coli (New England Biolabs No. C2527H). About 2 L of LB medium with 100 µg/mL ampicillin was inoculated with 80 mL from a clonal E. coli BL21 strain carrying pDEST15 RnSPAK^D219A grown in LB-amp for 16 h at 37°C. When the 2-L culture reached an A600 of 0.6, protein expression was induced with 500 µM IPTG (GoldBio) and the cultures incubated at 37°C for an additional 4 h before harvesting by centrifugation. The cell pellets were resuspended in 20 mL lysis buffer (1× PBS, 1% Triton X-100, 1 mM DTT, 1 mM benzadimine, 1× protease inhibitor cocktail (Invitrogen No. 1861279), 0.1% lysozyme (Thermo No. 89833) at 4°C for 30 min and then sonicated for 6 × 10 s (80% power, Branson 450 sonifier). The lysate was centrifuged at 15,000 rpm for 30 min at 4°C and the supernatant was collected. Ammonium sulfate (25% and 50% saturation) was successively added to clarified lysate, followed by pelleting at each step. The final pellet was redissolved in 20 mL lysis buffer lacking lysozyme and added to a glutathione agarose gravity-flow column (Thermo, No. 16101) that had been pre-equilibrated with PBS. The slurry was allowed to gently shake at 4°C for 1 hour. The flow-through was drained, and the column was washed with 200 mL of PBS. Proteins were eluted in 20 mL elution buffer (15 mM Tris/HCl pH 8.0, 15 mM NaOH, 15 mM reduced glutathione, 1 mM DTT, 0.5 mM benzamidine) and the eluate was concentrated to 4 mg/mL using a centrifugal filter unit (EMD Millipore). Protein purity was assessed using SDS-PAGE (8%) followed by Coomassie staining.

WNK3/WNK4 in vitro kinase assays.

Each 20 µL kinase reaction contained 2.5 µL anti-HA magnetic beads (1/20th of the preparation) supplying HA-WNK4 or HA-WNK3, 5 µg GST-SPAK^D219A, 10 mM HEPES pH 8.0, 7.5 mM magnesium gluconate, 1 mM DTT, 150 µM ATP, and varying concentrations of potassium gluconate. Reactions proceeded for 30 min at 37°C. To quench the reaction, an equivalent volume of 2× Laemmli buffer was added and boiled for 5 min at 95°C. Beads were then removed by magnetic immobilization and 20 µL supernatant was separated by 8% SDS-PAGE gel followed by Western blotting with primary antibodies as described in Measurement of Tubule WNK Activity. Secondary antibodies used were goat anti-rabbit (1:5,000, Li-Cor, No. 926–32211) and anti-mouse (1:5,000, Li-Cor, No. 926–68070). Pixel intensity of the upper (dominant) band was measured using ImageJ, with subtraction of background. The 0 mM potassium gluconate pSPAK/tSPAK ratio was assigned a value of 1, and the pSPAK/tSPAK ratios of other potassium gluconate concentrations were normalized to the ratio of the 0 mM bands from the same gel.

Statistics and Analysis

When data sets were sufficiently large, normality was determined using the D’Agostino & Pearson normality test. Normally distributed data sets were compared by unpaired, two-sided t test (2 comparisons) or ANOVA (>2 comparisons), and nonnormal data sets were compared by Kruskal–Wallis test. Normality was assumed for small data sets. For experiments in which data were normalized to control, two-tailed one sample t tests to a theoretical mean of 1 were performed. When multiple comparisons were made, the P value was adjusted accordingly; details of the correction test used are given in the figure legends. GraphPad Prism, v. 8 was used for statistical testing. Nonlinear regression using the log(inhibitor) versus response–variable slope (four parameters) function in GraphPad Prism, v.8, was used to analyze the relationship between in vitro WNK4 activity and potassium gluconate concentration.

RESULTS

High Extracellular Potassium Inhibits Drosophila WNK Activity in the Malpighian Tubule at Fixed Intracellular Chloride Concentrations

To determine whether there are effects of extracellular potassium on WNK activity that are independent of intracellular chloride, we designed baths in which extracellular potassium and chloride were adjusted to keep intracellular chloride constant with varying extracellular potassium. We designed two sets of baths with intracellular chloride concentrations similar to those we previously studied. In the first set of baths, we designed a “normal potassium” bath with an extracellular potassium of ∼17 mM, similar to the previously used hypotonic bath that stimulates WNK activity (14), as well as “low” (6.4 mM) and “high” (43 mM) baths of the same osmolality (Table 4). We then measured intracellular pH and chloride in Malpighian tubule epithelial cells using the transgenic sensor ClopHensor (14, 18, 29). There were no significant differences in intracellular pH or chloride in tubule epithelial cells bathed in the three varying potassium baths. Intracellular chloride was ∼13 mM, similar to the intracellular chloride concentration we measured previously in hypotonic conditions (14) (Fig. 1, A and B).

Table 4.

Ion composition and osmolality of baths with [Cl⁻]_i ∼ 13 mM

	Low K+, mM	Normal K+, mM	High K+, mM
Na+	60.0	60.0	60.0
Cl−	61.1	58.1	33.4
K+	6.4	16.8	42.8
Ca2+	2.8	2.8	2.8
Mg2+	9.1	9.1	9.1
HCO3−	5.7	5.7	5.7
H2PO4−	2.8	2.8	2.8
HPO42−	1.3	1.3	1.3
NMDG+	38.5	32.5
gluconate−	48.9	56.3	67.9
Osm (mosm/kg)	272	272	265

Additional nonionic components are not shown. See materials and methods for details. Osmolality was measured in triplicate; the mean is shown. NMDG, N-methyl-d-glucamine.

Figure 1.

High-potassium bath inhibits Drosophila with no lysine (K) kinase (WNK) activity in the Malpighian tubule at low, fixed intracellular chloride concentrations (∼13 mM). The intracellular pH (A) and chloride concentration (B) of Malpighian tubule principal cells was measured using the transgenic sensor, ClopHensor, expressed under the control of the c42-GAL4 driver (w; c42-GAL4/UAS-ClopHensor c304). Tubules were bathed for 1 h in baths containing low, normal, or high potassium (Table 4). Individual data points for each cell analyzed are shown, as well as means (bars) ± SE(error bars, standard error of the mean). A: n = 27 cells from nine tubules for each bathing medium. There was no significant effect of bathing medium on intracellular pH, P = 0.8628 by one-way ANOVA (analysis of variance). NS, nonsignificant. B: n = 25, 21, and 24 cells from nine tubules in low-, normal-, and high-potassium baths, respectively. There was no significant effect of bathing medium on intracellular chloride, P = 0.4574 by Kruskal–Wallis test. Malpighian tubules expressing kinase-dead rat Ste20-related proline/alanine-rich kinase (SPAK)^D219A in the principal cell (w; UAS-SPAK^D219A/+; c42-GAL4/+) were bathed in normal-potassium bath for 1 h, followed by low-potassium bath (C) or high-potassium bath (D) for 15, 30, or 60 min. Endogenous Drosophila WNK activity was estimated from the ratio of phosphorylated SPAK (pSPAK)/total SPAK (tSPAK), as determined by Western blotting and ImageJ analysis. In this and subsequent experiments, the phosphorylated SPAK (pSPAK)/total SPAK (tSPAK) ratio was set to 1 in the control (in this case, 0 min), and the other time points were normalized to the control. The molecular weight marker is 75 kDa. n = 6 independent experiments, as represented by the individual data points. Means ± SE shown. One sample t tests were performed to a theoretical mean of 1 with Bonferroni correction for multiple comparisons, adjusted α = 0.0167. C: minor effects of low-potassium bath on WNK activity. P values: 15 min, 0.0178 (NS), 30 min, 0.4083 (NS), 60 min, 0.0089 (*). D: high-potassium bath inhibited WNK activity at 30 and 60 min. P values: 15 min, 0.4379 (NS), 30 min, <0.0001 (***), 60 min, <0.0001 (***). Full gels for all independent experiments for this and subsequent figures are available in the Supplemental Material.

To measure Drosophila WNK activity in tubules bathed in varying extracellular potassium baths, we transgenically expressed kinase-dead rat SPAK^D219A in the Malpighian tubule principal cell, where it serves as a substrate for the endogenous Drosophila WNK. The ratio of phosphorylated SPAK (pSPAK) to total SPAK (tSPAK) on Western blots of tubule lysates reflects WNK activity (14). Bathing tubules in low-potassium bath for 15, 30, or 60 min after previous incubation in normal potassium bath for 1 h had minimal effects on Drosophila WNK activity (Fig. 1C). However, tubule WNK activity was inhibited in tubules bathed in high potassium bath for 30 or 60 min (Fig. 1D). The kinetics of inhibition at 30 and 60 min are similar to the WNK activation we observed at 30 and 60 min when intracellular chloride decreased in tubules bathed in hypotonic medium (14).

Since chloride inhibits WNK activity, we designed a second set of varying potassium baths (Table 5) to determine whether the high potassium inhibitory effect is still observed at higher intracellular chloride. Intracellular pH and chloride were similar in the three baths (Fig. 2, A and B). Intracellular chloride concentrations were ∼26 mM, similar to the intracellular chloride concentrations we previously measured in standard bathing medium (14). As with the low-intracellular chloride baths, low-potassium bath had no effect on tubule Drosophila WNK activity, and high-potassium bath had an inhibitory effect at 30 and 60 min (Fig. 2, C and D). These results indicate that high-potassium baths inhibit WNK activity at both low- and high-intracellular chloride concentrations, suggesting additive inhibitory effects of potassium and chloride on WNK activity.

Table 5.

Ion composition and osmolality of baths with [Cl⁻]_i ∼ 26 mM

	Low K+, mM	Normal K+, mM	High K+, mM
Na+	91.4	91.4	85.4
Cl−	116	114.4	81.5
K+	4.5	20.0	45.0
Ca2+	2.1	2.1	2.1
Mg2+	7.7	7.7	7.7
HCO3−	7.2	7.2	7.2
H2PO4−	3.2	3.2	3.2
HPO42−	0.8	0.8	0.8
NMDG+	32.8	23.7
Gluconate−	15.8	23.7	51.9
Osm (mosm/kg)	296	309	299

Additional nonionic components are not shown. See materials and methods for details. Osmolality was measured in triplicate; the mean is shown. NMDG, N-methyl-d-glucamine.

Figure 2.

High-potassium bath inhibits Drosophila with no lysine (K) kinase (WNK) activity in the Malpighian tubule at higher intracellular chloride concentrations (∼26 mM). The intracellular pH (A) and chloride concentration (B) of Malpighian tubule principal cells was measured using ClopHensor in w; c42-GAL4/UAS-ClopHensor c304 flies. Tubules were bathed in low-, normal-, or high-potassium baths with higher extracellular chloride concentrations (Table 5) to achieve higher intracellular chloride. Individual data points for each cell analyzed, means and SE shown. A: n = 21 cells from seven tubules for each bathing medium. There was no significant effect of bathing medium on intracellular pH, P = 0.2849 by one-way ANOVA. B: n = 15, 18, and 20 cells from seven tubules in low-, normal-, and high-potassium baths, respectively. There was no significant effect of bathing medium on intracellular chloride, P = 0.3552 by Kruskal–Wallis test. Malpighian tubules expressing kinase-dead rat Ste20-related proline/alanine-rich kinase (SPAK)^D219A in the principal cell (w; UAS-SPAK^D219A/+; c42-GAL4/+) were bathed in normal potassium bath for 1 h, followed by low potassium bath (C) or high potassium bath (D) for 15, 30, or 60 min. The phosphorylated SPAK (pSPAK)/total SPAK (tSPAK) ratio was determined by Western blotting followed by ImageJ analysis. n = 6 independent experiments, as represented by the individual data points. Means ± SE shown. One sample t tests were performed to a theoretical mean of 1 with Bonferroni correction for multiple comparisons, adjusted α = 0.0167. C: there was no significant effect of low potassium bath on Drosophila WNK activity. P values: 15 min, 0.2644 (NS), 30 min, 0.4370 (NS), 60 min, 0.4759 (NS). D: high-potassium bath inhibited WNK activity at 30 and 60 min. P values: 15 min, 0.2343 (NS), 30 min, <0.0001 (***), 60 min, <0.0001 (***).

To assess whether intracellular potassium changes when tubules are bathed in varying extracellular potassium, we attempted to measure tubule potassium using inductively coupled plasma mass spectrometry (ICP-MS). Background was high, however, with resultant low signal-to-noise (data not shown). Since rubidium is transported by potassium-transporting channels and transporters, and given the fast transport rates of Drosophila Malpighian tubules (30), we reasoned that tubule potassium content would be rapidly replaced by rubidium when the tubules were bathed in rubidium-substituted baths. To confirm this, we measured tubule potassium and rubidium content in triplicate in tubules bathed in salines containing either potassium or rubidium. Tubules bathed in potassium-containing baths had 39.08 ± 9.70 (means ± SD) ppb (parts per billion) potassium and 0.01 ± 0.01 ppb rubidium, whereas tubules bathed in rubidium-containing baths had 1.26 ± 1.16 ppb potassium and 79.45 ± 31.90 ppb rubidium, indicating that rubidium effectively replaces potassium in Malpighian tubule epithelial cells.

Next, we assessed whether extracellular rubidium influences WNK activity similar to potassium. We substituted rubidium for potassium in the low-intracellular chloride baths (Table 4), and measured WNK activity after bathing tubules for 1 h in normal rubidium bath, followed by 30 min in low-, normal-, or high-rubidium bath. There was no significant effect of the low-rubidium bath on Drosophila WNK activity, whereas high-rubidium bath inhibited WNK, as observed with low- and high-potassium baths (Fig. 3A). We then measured tubule rubidium using ICP-MS in tubules bathed in low-, normal-, and high-rubidium baths, as well as in a rubidium-substituted Drosophila saline that was previously used to measure intracellular potassium using double-barreled ion-specific electrodes (19). (Of note, measurement of intracellular potassium using that technique in Drosophila Malpighian tubules was technically challenging, with 10 successful impalements of 150 microelectrodes fabricated, and therefore we did not attempt to reproduce those measurements for the current studies.) Tubule rubidium was similar in the normal rubidium and Drosophila saline conditions, lower (but not significantly) in the low-rubidium bath, and significantly higher in the high-rubidium bath (Fig. 3B).

Figure 3.

High-rubidium bath increases tubule rubidium and inhibits Drosophila with no lysine (K) kinase (WNK) activity, and ouabain increases Drosophila WNK activity. A: Malpighian tubules expressing Ste20-related proline/alanine-rich kinase (SPAK)^D219A (w; UAS-SPAK^D219A/+; c42-GAL4/+) in the principal cell were bathed in “normal” rubidium bath for 1 h, followed by “low” rubidium or “high” rubidium bath for 30 min. These baths mimicked the low-, normal-, and high-potassium baths used in Fig. 1 except that potassium was replaced by rubidium. Like potassium, high rubidium bath inhibits Drosophila WNK activity. n = 3 independent experiments. Means ± SE shown. One sample t tests were performed to a theoretical mean of 1 with Bonferroni correction for multiple comparisons, adjusted α = 0.025. P values: low rubidium, 0.2921 (NS), high rubidium, <0.0001 (***). B: tubule rubidium was measured by inductively coupled plasma mass spectrometry in tubules bathed in low-, normal-, or high-rubidium baths (as in A), or in Drosophila saline in which potassium was replaced by rubidium. n = 10 independent replicates, as indicated by the individual data points. Means ± SE shown. P = 0.0038, one-way ANOVA. P values in Dunnett’s multiple comparisons test (α = 0.05), comparing each bathing medium to the normal rubidium bath: Drosophila saline, 0.9978 (NS); low rubidium, 0.5478 (NS); high rubidium, 0.0285 (*). C: Malpighian tubules expressing SPAK^D219A (w; UAS-SPAK^D219A/+; c42-GAL4/+) in the principal cell were bathed in standard bathing medium with or without 100 µM ouabain. n = 5 independent experiments. Means ± SE shown. One sample t tests were performed to a theoretical mean of 1. **P = 0.0012.

Because the low-rubidium baths did not achieve a statistically significant reduction in tubule rubidium, we used an alternative approach to lower intracellular potassium. In most cells, the inward pumping of potassium by the Na⁺/K+-ATPase maintains high-intracellular potassium concentrations. Na⁺/K+-ATPase is expressed on the basolateral membrane of the Malpighian tubule principal cell (19, 31–36). We previously demonstrated that treating tubules with 100 µM ouabain, which inhibits the Na⁺/K+-ATPase, decreases the concentration of secreted fluid potassium, and increases the concentration of secreted fluid sodium, consistent with the expected effect of ouabain to decrease intracellular potassium and increase intracellular sodium in the principal cell (13). We therefore incubated tubules for 1 h in 100 µM ouabain and compared Drosophila WNK activity in ouabain-treated versus control tubules. We observed an increase in Drosophila WNK activity in ouabain-treated tubules (Fig. 3C). This suggests that lowering intracellular potassium stimulates WNK activity and that ouabain may result in lower intracellular potassium concentrations than those achieved in the low-potassium baths.

Direct Effects of Potassium on WNK Thermal Stability and WNK Autophosphorylation

Although the ICP-MS results indicate that tubule potassium likely changes in the varying potassium baths, changes in extracellular potassium (or ouabain treatment) could also affect cellular physiology in other ways, for example by changing membrane potential. We, therefore, turned to in vitro studies to better define effects of precisely controlled potassium concentrations on WNK activity, independent of other factors. Since chloride ion directly interacts with WNK to regulate its activity (11), we hypothesized that potassium also interacts directly with WNK kinases. To test this, we first performed differential scanning fluorimetry of the purified kinase domains of Drosophila WNK, and mammalian WNK1 and WNK3, in the presence of varying concentrations of potassium chloride or potassium gluconate. Potassium increased the melting temperature of Drosophila WNK (Fig. 4, A and B), WNK1 (Fig. 4, C and D), and WNK3 (Fig. 4E), indicating that potassium stabilizes Drosophila and mammalian WNK kinases through direct binding. Two control kinases, p38 and apoptosis signal-regulated kinase 1 (ASK1), were also tested, and showed no shift in melting temperature with increasing potassium concentrations (Fig. 4, F and G).

Figure 4.

Potassium increases the thermal stability of Drosophila with no lysine (K) kinase (WNK) and mammalian WNK1 and WNK3 kinase domains. The kinase domains of Drosophila WNK (amino acids 261-534) (A and B), WNK1 (amino acids 194-483) (C and D) and WNK3 (amino acids 118-409) (E), and p38 kinase (F) and the apoptosis signal-regulated kinase 1 (ASK1) kinase domain (amino acids 659-951) (G), were incubated with varying concentrations of potassium chloride (A, C, F, and G) or potassium gluconate (B, D, and E) in vitro. Relative fluorescence units (RFU) were measured using differential scanning fluorimetry with SYPRO orange fluorescence. Potassium concentrations are shown. The dRFU/dt is plotted vs. temperature; the minima are taken as the melting temperature. WNK melting temperature increases with increasing potassium concentrations, indicating that potassium stabilizes the kinase.

Chloride ion binding inhibits the autophosphorylation of Drosophila and mammalian WNKs (11, 14). We, therefore, assessed the effect of potassium on Drosophila WNK and human WNK3 kinase domain autophosphorylation in vitro, as measured by mass spectrometry. The effect of potassium gluconate on autophosphorylation of Drosophila WNK Ser434 (Fig. 5A) and WNK3 Ser308 (Fig. 5B) was measured over time. WNK3 was more active than Drosophila WNK. However, this may not reflect the activity of the full-length proteins, particularly since we did not attempt to optimize the boundaries of the Drosophila kinase domain. Both kinases showed potassium-sensitive inhibition over a range of potassium concentrations from 0 mM to 240 mM.

The decrease in Drosophila WNK and human WNK3 autophosphorylation with increasing potassium concentrations indicates that potassium, like chloride, directly inhibits WNK autophosphorylation. Since autophosphorylation is required for WNK activation (37), these results indicate that potassium is a negative regulator of WNK activity. In contrast, 240 mM potassium gluconate had no effect on the activity of the control kinase ASK1, as assayed by phosphorylation of myelin basic protein (Fig. 6).

Figure 6.

Potassium has no effect on apoptosis signal-regulated kinase 1 (ASK1) activity. Progress curves of ASK1 phosphorylation of myelin basic protein in the presence of 240 mM potassium gluconate (gray) or sodium gluconate (black); the curves are overlapping. Activity was measured using Kinase Glo as a function of ATP consumption over time. Time points were measured in triplicate with means ± standard deviation shown (in some cases smaller than the symbols). In two-way repeated-measures ANOVA, there was a significant effect of time (P < 0.0001) but no effect of ion (P = 0.8203).

Effects of Varying Potassium Bath on Mammalian WNK3 and WNK4 Expressed in the Malpighian Tubule

Since our in vitro studies indicated that potassium binds to and inhibits the autophosphorylation of both Drosophila and mammalian WNKs, we next assessed the effects of varying potassium bath on full-length mammalian WNKs in tubule epithelial cells. Knocking down the single Drosophila WNK in the Malpighian tubule principal cell (14) and expressing mammalian WNKs in its stead allows analysis of individual mammalian WNK paralogs. As with Drosophila WNK, there was no effect of low-potassium bath on WNK3 activity, and high-potassium bath inhibited WNK3 (Fig. 7A). To further characterize whether the potassium effect is independent of chloride, we next expressed the chloride-insensitive WNK3^L295F mutant (11) in Drosophila WNK knockdown tubules. As expected, WNK activity was increased in WNK3^L295F-expressing tubules compared with tubules expressing wild-type WNK3 (Fig. 7B). Low-potassium bath did not affect chloride-insensitive WNK3 activity and high-potassium inhibited activity (Fig. 7C). These results further indicate effects of potassium independent of WNK chloride regulation and suggest that potassium and chloride do not bind to the same site.

Figure 7.

High-potassium bath inhibits wild-type and chloride-insensitive WNK3 expressed in the Malpighian tubule. A: Malpighian tubules in which Drosophila WNK was knocked down and human WNK3 was expressed with SPAK^D219A in principal cells (w; UAS-DmWNK^RNAi UAS-WNK3/UAS-SPAK^D219A; c42-GAL4/+) were bathed for 1 h in normal potassium (low-intracellular chloride) bath, followed by 30 min in low-, normal-, or high-potassium bath. There was no effect of low-potassium bath, and high-potassium bath inhibited WNK3 activity. n = 6 independent experiments. Means ± SE shown. One sample t tests were performed to a theoretical mean of 1 with Bonferroni correction for multiple comparisons, adjusted α = 0.025. P values: low potassium, 0.3045 (NS); high potassium, <0.0001 (***). B: chloride-insensitive WNK3 is more active than wild-type WNK3. WNK3 activity was measured in Malpighian tubules in which Drosophila WNK was knocked down and wild-type human WNK3 was expressed with SPAK^D219A in principal cells (w; UAS-DmWNK^RNAi UAS-WNK3/UAS-SPAK^D219A; c42-GAL4/+), or in which Drosophila WNK was knocked down and chloride-insensitive WNK3^L295F was expressed with SPAK^D219A (w; UAS-DmWNK^RNAi UAS-WNK3^L295F/UAS-SPAK^D219A; c42-GAL4/+). n = 4 independent experiments. Means ± SE shown. One sample t test was performed to a theoretical mean of 1, *P = 0.0159. C: Malpighian tubules in which Drosophila WNK was knocked down and human chloride-insensitive WNK3^L295F was expressed with SPAK^D219A in principal cells (w; UAS-DmWNK^RNAi UAS-WNK3^L295F/UAS-SPAK^D219A; c42-GAL4/+) were bathed for 1 h in normal potassium (low-intracellular chloride) bath, followed by 30 min in low-, normal-, or high-potassium bath. There was no effect of low-potassium bath, and high-potassium bath inhibited chloride-insensitive WNK3 activity. n = 6 independent experiments. Means ± SE shown. One sample t tests were performed to a theoretical mean of 1 with Bonferroni correction for multiple comparisons, adjusted α = 0.025. P values: low potassium, 0.0467 (NS); high potassium, < 0.0001 (***). WNK, with no lysine (K) kinase.

WNK4 is the dominant WNK regulating NCC in the distal convoluted tubule (1, 38). We, therefore, examined the effect of varying potassium baths on tubule-expressed WNK4. High potassium bath inhibited both wild-type WNK4 and chloride-insensitive WNK4^L319F expressed in Drosophila WNK knockdown tubules (Fig. 8), similar to the effects of high-potassium bath on Drosophila WNK and WNK3 (Figs. 1, 2, and 7). Mouse Mo25 was coexpressed with WNK4 in these experiments as preliminary experiments showed increased WNK4 activity in the presence of Mo25 coexpression (data not shown). Significant increases in WNK4 activity were observed in low-potassium bath.

Figure 8.

High potassium inhibition and low potassium activation of wild-type and chloride-insensitive WNK4 in tubules. A: Malpighian tubules in which Drosophila WNK was knocked down and mouse Wnk4 was coexpressed with mouse protein 25 (Mo25) and Ste20-related proline/alanine-rich kinase (SPAK)^D219A in adult principal cells (w; UAS-DmWNK^RNAi UAS-WNK4/tub-GAL80^ts20 UAS-SPAK^D219A; UAS-Mo25/c42-GAL4) were bathed for 2 h in normal-potassium (high intracellular chloride) bath, followed by 30 min in low-, normal-, or high-potassium bath. Low-potassium bath stimulated, and high-potassium bath inhibited WNK4 activity. n = 5 independent experiments. Means ± SE shown. One sample t tests were performed to a theoretical mean of 1 with Bonferroni correction for multiple comparisons, adjusted α = 0.025. P values: low potassium, 0.0184 (*); high potassium, 0.0086 (*). B: Malpighian tubules in which Drosophila WNK was knocked down and chloride-insensitive mouse WNK4^L319F was coexpressed with mouse Mo25 and SPAK^D219A in adult principal cells (w; UAS-DmWNK^RNAi UAS-WNK4^L319F/tub-GAL80^ts20 UAS-SPAK^D219A; UAS-Mo25/c42-GAL4) were bathed for 2 h in normal-potassium (high intracellular chloride) bath, followed by 30 min in low-, normal-, or high-potassium bath. Low-potassium bath stimulated, and high-potassium bath inhibited chloride-insensitive WNK4 activity. n = 6 independent experiments. Means ± SEM shown. One sample t tests were performed to a theoretical mean of 1 with Bonferroni correction for multiple comparisons, adjusted α = 0.025. P values: low potassium, 0.0137 (*); high potassium, 0.0031 (**). WNK, with no lysine (K) kinase.

Potassium Inhibits WNK4 and WNK3 In Vitro

In renal epithelial cells, small changes in intracellular potassium have been measured in rats fed low- or high-potassium diets (39–41). We, therefore, next addressed the degree to which WNK activity changes in response to changes in physiologically relevant potassium concentrations. We focused on WNK4, since it is the most extensively characterized WNK in mammalian renal physiology, and compared it with WNK3, since we observed differences in potassium effects on tubule-expressed WNK3 and WNK4 (Figs. 7 and 8).

To better assess the sensitivity of WNK3 and WNK4 to a broad range of potassium concentrations, full-length HA-tagged WNK3 or WNK4 were expressed in Drosophila S2 cultured cells, immunoprecipitated, and assayed for kinase activity using purified glutathione S-transferase (GST)-SPAK^D219A as a substrate. HA-tagged WNK4 was observed at a molecular weight between 150 and 250 kDa (Fig. 9A), similar to the previously reported molecular weight of WNK4 in mDCT cells (42). As expected, WNK3, which is larger than WNK4, was observed at a higher molecular weight (Fig. 9A), as also observed by others (37). GST-SPAK^D219A appeared as a doublet (Fig. 9A), as also observed by other investigators purifying SPAK and OSR1 after bacterial expression (24, 43).

Figure 9.

High potassium inhibits full-length WNK4 and WNK3 activity in vitro, and low potassium stimulates WNK4. A: full-length hemagglutinin (HA)-tagged WNK4 or WNK3 were expressed in S2 cultured cells and immunoprecipitated. Anti-HA Western blotting demonstrating immunoprecipitated HA-WNK4 (left) and HA-WNK3 (middle). Migration of molecular weight standards (kDa) is shown to the left. The vertical line between the molecular weight standards and the HA-WNK3 lane indicates removal of intervening lanes. Glutathione S-transferase (GST)-SPAK^D219A was expressed in Escherichia coli and purified. About 10 µg total protein was resolved by SDS-PAGE and the gel stained with Coomassie (right). MW, molecular weight standards (kDa). B and C: immunoprecipitated WNK4 (B) and WNK3 (C) were assayed for activity using purified GST-tagged SPAK^D219A as a substrate in the presence of varying potassium gluconate concentrations. Phosphorylated SPAK (pSPAK) and total SPAK (tSPAK) were detected by immunoblotting. The molecular weight marker is at 100 kDa. Representative gels are shown; not all concentrations could be run on the same gels due to lane limitations. Therefore, 0 mM potassium reactions were run on every gel and the pSPAK/tSPAK ratio set to 1 (indicated as 100% in the graph), with other ratios normalized to the 0 mM potassium. In the merged images, yellow/green colors indicate a higher pSPAK/tSPAK ratio (green channel dominates), and orange colors indicate a decreasing pSPAK/tSPAK ratio (red channel increasingly dominates). Means ± SD shown. For WNK4, three independent experiments were performed for 20, 40, 80, 100, 150, 180, 240, and 300 mM potassium gluconate, and seven independent experiments for potassium gluconate concentrations of 110, 120, 130, and 140 mM. A sigmoidal log(potassium) vs. response curve with variable slope was fitted in GraphPad Prism, r² = 0.8177, log IC50 = 2.088 (122 mM). For WNK3, n = 3 independent experiments. SPAK, Ste20-related proline/alanine-rich kinase; WNK, with no lysine (K) kinase.

Increasing concentrations of potassium gluconate inhibited WNK4 activity across a broad range of concentrations (Fig. 9B). This inhibition is unlikely due to changes in osmolarity. In a separate study examining the activity of phosphorylated WNK1 and WNK3 toward OSR1, we found no effect of the osmolyte sucrose at concentrations up to 600 mM on WNK1 phosphorylation of OSR1. Furthermore, increasing concentrations of the osmolyte PEG400 stimulated (rather than inhibited) WNK1 and WNK3 phosphorylation of OSR1 (44). Therefore, WNK4 inhibition by potassium gluconate is likely due to increasing potassium concentrations rather than increasing osmolarity.

In the Drosophila Malpighian tubule principal cell, the intracellular potassium concentration is 121 mM, as measured using ion-specific electrodes (19). In the distal convoluted tubule, connecting tubule, and cortical collecting duct principal cells of the rat nephron, intracellular potassium was ∼135 mmol/kg wet wt, as measured using electron microprobe analysis (39). “Free” potassium concentrations in distal nephron epithelial cells may be lower; for example, the intracellular potassium concentration was 95 mM in principal cells in the rabbit cortical collecting duct, as measured using ion-specific electrodes (45). Intracellular potassium concentrations have been estimated at 90–100 mM in the thick ascending limb (46). In our in vitro studies, the steepest part of the dose-response curve for potassium inhibition of WNK4 was in the range of 80–180 mM (log concentration, 1.9–2.26), with an IC₅₀ of 122 mM (Fig. 9B). Thus, WNK4 is most sensitive to changes in potassium concentration within the physiological range of intracellular potassium in renal epithelial cells.

Like WNK4, WNK3 was also inhibited by high-potassium concentrations. Interestingly, maximal WNK3 activity was observed at 100 mM potassium, that is, physiological intracellular potassium concentrations (Fig. 9C). These findings mirror WNK3 and WNK4 activity in the tubule, in which WNK4 was uniquely stimulated in low-potassium baths (Figs. 7 and 8). Thus, WNK4 appears especially tuned to respond to lowering of intracellular potassium.

DISCUSSION

Here, we demonstrate that high-extracellular potassium concentrations inhibit Drosophila WNK and mammalian WNK3 and WNK4 expressed in the Malpighian tubule. This inhibition is independent of changes in intracellular chloride. Low-extracellular potassium activates WNK4, also independent of changes in intracellular chloride. High-potassium bath was associated with an increase in tubule potassium, as inferred from substituted rubidium. In addition, ouabain, which inhibits the Na⁺/K+-ATPase and is therefore expected to lower intracellular potassium, also resulted in stimulation of Drosophila WNK activity. In vitro experiments show that potassium binds to Drosophila and mammalian WNK kinase domains to directly inhibit their autophosphorylation and activity toward the WNK substrate, SPAK. Concentration-response curves with WNK4 indicate that the greatest effect of potassium on WNK4 activity occurs in the range of 80–180 mM. This encompasses physiological intracellular potassium concentrations (19, 39, 45–47). Thus, WNKs are potassium-sensitive kinases.

A potential physiological role for potassium regulation of WNKs is suggested by two mammalian secretory epithelia, the sweat and salivary glands. As in the Malpighian tubule, these epithelia rely on a basolateral NKCC for the uptake of sodium, potassium, and chloride from the blood into the epithelial cells (4). WNK1 and WNK4 are expressed in both glands (48–50). Intracellular potassium in these epithelia falls dramatically in response to hormonal (cholinergic) stimulation: from 92 mM to 32 mM in sweat gland and from 116 to 81 mM in salivary gland (51–53). Our data (Fig. 9B) predict that these decreases in intracellular potassium would increase WNK4 activity by ∼15%–30%. In sweat gland, a concomitant fall in intracellular chloride (52) would be predicted to further activate WNK, based on our findings that chloride and potassium have independent effects on WNK activity (Figs. 1, 2, 7, and 8). In contrast, in submandibular and parotid salivary glands, intracellular chloride concentrations stayed the same or even increased as intracellular potassium fell (53, 54). This is remarkable, because it indicates that intracellular chloride cannot always be the signal for WNK regulation in chloride-secreting epithelia. However, relief of potassium inhibition of WNK would allow NKCC1 activation and increased gland secretion even when intracellular chloride is constant or increasing.

The low-potassium diet prevalent in modern society increases the risk for hypertension, cardiovascular disease, stroke, and chronic kidney disease (55–59). One consequence of a low-potassium diet is the phosphorylation-dependent activation of the sodium chloride cotransporter (NCC) in the distal convoluted tubule of the kidney. NCC activation decreases sodium delivery to the downstream aldosterone-sensitive distal nephron, decreasing sodium-dependent potassium secretion (60). Thus, NCC phosphorylation regulates renal potassium excretion. NCC phosphorylation in low-potassium conditions is WNK4 and SPAK/OSR1-dependent (61–65). Consistent with a role for WNKs in regulating potassium homeostasis, human patients with WNK1 or WNK4 gain-of-function mutations are hyperkalemic (66), and Wnk4 knockout in mice results in hypokalemia (38).

Terker et al. (63) proposed that decreased extracellular potassium results in basolateral membrane hyperpolarization of distal convoluted tubule epithelial cells, lowering of intracellular chloride, and the activation of WNK4. In support of this model, acute changes in extracellular potassium result in transient changes in intracellular chloride in isolated distal convoluted tubules (67). Furthermore, when chloride insensitive mutations of WNK4 were knocked into mice, increased NCC phosphorylation and activity were observed under a normal-potassium diet, with no further increases with low-potassium diet (68).

Could intracellular potassium play a role in regulating WNK activity in renal epithelia, together with intracellular chloride? Existing evidence suggests that intracellular potassium remains constant in renal epithelial cells when extracellular potassium is changed acutely, except at very high (50 mM) extracellular potassium concentrations (69). However, several studies demonstrate changes in intracellular potassium in response to chronic changes in dietary potassium. Intracellular potassium decreased in proximal and distal tubules of rats chronically fed low-potassium diet (40, 41, 70). Conversely, in rats fed a high-potassium diet, intracellular potassium increased in the distal nephron, including in distal convoluted tubule and connecting tubule (39, 41).

In most of these studies, feeding rats low- or high-potassium diets resulted in relatively small changes in renal epithelial cell intracellular potassium, between 7 mM and 14 mM (39–41). Are these small changes sufficient to result in changes in WNK activity and ion transport? In addressing this question, we first considered changes in WNK activity that occur in response to changes in chloride concentration. As with potassium, few measurements of intracellular chloride have been made in renal epithelial cells. As measured using electron microprobe analysis, high-potassium diet resulted in a statistically significant, but small, increase in intracellular chloride in the distal convoluted tubule, from 9.6 mM to 11 mM (39). In modeling studies, a change in plasma potassium from 2 mM to 4 mM was estimated to increase intracellular chloride in the distal convoluted tubule from ∼13 mM to 17 mM, and to ∼19 mM at a plasma potassium of 6 mM (63). An increase in chloride concentration from 13 mM to 17 mM inhibits in vitro WNK4 activity (SPAK phosphorylation) by ∼3% (71). Based on our own data (Fig. 9B), an 8 mM increase in potassium concentration, as measured by Beck et al. (39) in the distal convoluted tubule of rats fed a high-potassium diet, inhibits WNK4 activity by ∼4%. These data imply that physiologically relevant changes in intracellular potassium and chloride concentrations have similar effects on WNK activity. These, apparently small, effects may be amplified by the downstream signaling cascades. Furthermore, since our data show additive effects of chloride and potassium, greater changes in WNK activity may be achieved with parallel changes in potassium and chloride than would occur with changes in one ion alone. Ultimately, additional measurements of intracellular potassium and chloride in the mammalian nephron under different conditions would be helpful, but are technically challenging. Identification of potassium-insensitive WNK mutants could also help resolve these questions.

In prior studies, we found that chloride binds to a highly conserved binding site in an inactive conformer of WNK1 (11). We hypothesize that potassium also binds to inactive WNKs. The data presented in Figs. 7 and 8, in which high-potassium inhibits both wild-type WNK3 and WNK4 and WNK3 and WNK4 in which the chloride binding site is mutated, favor independent binding sites for chloride and potassium. Furthermore, high potassium inhibits Drosophila WNK, WNK3, and WNK4 at both low and high concentrations of intracellular chloride in the Malpighian tubule (Figs. 1, 2, 7, and 8). Structural studies of potassium-binding sites will be pursued in the future. We anticipate that structural studies will also inform understanding of the differences in low potassium activation of WNK4, as compared with WNK3 and Drosophila WNK (Figs. 1, 2, 7, 8, and 9).

The hypothesis that WNK exists in an equilibrium between inactive and active conformers, with the inactive conformer stabilized by potassium and chloride, can also explain the phenotype of the chloride-insensitive WNK4 mice. These mice have active WNK4 despite hyperkalemia (68). If the mutation in the chloride-binding site mimics 0 mM chloride, this could be sufficient to favor the active WNK conformer even in the face of mildly elevated extracellular potassium (plasma potassium of 4.75 mM in the chloride-insensitive WNK4 knockin mice, vs. 4.37 mM in wild-type). However, in physiological conditions, intracellular chloride does not decrease to 0 mM, and therefore, these results do not exclude a role for potassium regulation of WNKs (39, 63, 67). In addition, although NCC is not dephosphorylated in the chloride-insensitive WNK4 mice in response to an acute potassium load, it is dephosphorylated in mutant mice chronically fed a high-potassium diet (68). This suggests that this could be a physiological situation in which potassium regulation of WNKs plays a greater role, consistent with the effects of chronic changes in dietary potassium on intracellular potassium (39–41, 70).

In summary, the earlier discovery that WNKs are direct sensors of chloride (11) made it possible for us to hypothesize that WNKs are directly regulated by potassium. Here, we have demonstrated WNK regulation by potassium in renal tubule epithelial cells even when intracellular chloride is kept constant. The in vitro assays indicate that this is a direct effect of potassium on WNK kinase domains. Thus, WNK is a multimodal sensor of the intracellular ionic environment, which may allow integration of, and response to, ionic perturbations influencing cellular function.

GRANTS

Stocks obtained from the Bloomington Drosophila Stock Center (NIH P40OD018537) were used in this study. ICP-MS was performed at the Iron and Heme Core Facility at the University of Utah, supported, in part, by NIH National Institute of Diabetes and Digestive and Kidney Diseases Grant U54DK110858. This work was supported by the National Institutes of Health (National Institute of Diabetes and Digestive and Kidney Diseases Grant DK110358 to A.R.R. and E.J.G.) and the Welch Foundation Grant I1128 (to E.J.G.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

E.J.G. and A.R.R. conceived and designed research; J.M.P., L.N., R.A., J.M.H., H.H., Q.S., F.Z., J.S-P., D.E.M., J.N.S., and L.K.J. performed experiments; J.M.P., L.N., R.A., J.M.H., F.Z., J.S-P., and A.R.R. analyzed data; R.A., J.M.H., E.J.G., and A.R.R. interpreted results of experiments; D.E.M. and A.R.R. prepared figures; A.R.R. drafted manuscript; J.M.P., L.N., J.M.H. E.J.G., and A.R.R. edited and revised manuscript; J.M.P., L.N., R.A., J.M.H., H.H., Q.S., F.Z., J.S-P., D.E.M., J.N.S., L.K.J., E.J.G., and A.R.R. approved final version of manuscript.