Available structures of unphosphorylated WNK kinase domains possess networks of bound water molecules. In contrast, phosphorylated active WNKs are monomerIc (Akella et al., 2021). ProBis-H2O (Jukič et al., 2020) was initially used to find water networks across WNK1 structures; then the networks were manually curated in PyMOL (see Methods). CWN1 was observed previously (Akella et al., 2021), and is in the active site between the Catalytic Loop and Activation Loop (Figure 1A). Of the 14 waters that make up CWN1 in PDB file 6CN9, nine are conserved in PDB 5DRB (Figure 1B). A second water network was uncovered, CWN2, found by superimposing the N-terminal domains of 6CN9 and 5DRB. CWN2 is in the subunit interface in 6CN9 (a lattice contact in 5DRB, Figure 1C). Of the seven CWN2 waters in 6CN9, six are conserved in 5DRB.

PEG400 is a strong osmolyte we have shown activates uWNK (Akella et al., 2021). PEG400 has a destabilizing effect both on WNK1/SA (Figure 2A) and WNK3/SA (not shown) as observed by differential scanning fluorimetry. To test the effects of PEG400 on WNK structure, we formed crystals and cryoprotected with either PEG400 or glycerol (as previously reported, see Methods).

Crystallographic data collection parameters and refinement statistics of the two structures are given in Supplementary file 1a. The structure observed in glycerol-soaked crystals was very similar to the previously determined uWNK1/SA structure 6CN9 (r.m.s.d. 0.19 Å), including a few glycerol molecules. No PEG400 was observed in uWNK1/SA/PEG400. Remarkably, PEG400 induced a space group change from P1 to P2₁. Together with the space group change, uWNK1/SA in PEG400 monomerizes. To understand the conformational change, the WNK1/SA/PEG400 structure was indexed in P1 and origin selected to place the 6CN9 dimer in the asymmetric unit (see Supplementary file 1b). The change in structure induced by PEG400 is shown in Figure 2B, giving an overall r.m.s.d. of 3.2 Å, between the 6CN9 dimer and the two subunits in WNK1/SA/PEG400. Superpositions of monomers show that the PEG400-induced monomer is more similar to the A-chain than the B-chain of 6CN9 (r.m.s.d of 0.79 versus 1.3 Å, Supplementary file 1b). Figure 2C gives the conformational change as a function of sequence. Prominent changes occur in helix C and β5, with the largest change occurring in the Activation Loop (including the preceding 3/10 chloride-binding helix). Changes in the same loci are induced by the osmolyte sucrose in phosphorylated WNK1 (Akella et al., 2020).

Figure 3 is a closeup of the active site comparing the PEG400-induced structure with 6CN9. PEG400 affects CWN1, the chloride-binding 3/10 helix, and helix C (Figure 3A). PEG400 induces significant losses in the CWN1, with only 5 waters remaining out of the 14 in 6CN9 (Figure 3B, missing waters in gray). An active configuration of WNK1, pWNK1 (PDB file 5W7T), has even fewer bound waters (Figure 3B). The loss of water is accompanied by a corresponding reduction in the cavity size (Figure 3C). Furthermore, PEG400 induces a large rotation in the 3/10 helix and neighboring helix C (Figure 3D). The change in the 3/10 helix eliminates the chloride-binding site and gives rise to big conformational changes at K375 and E388 in the Activation Loop (Figure 3D). Mutations at these sites are discussed below. The water network at the dimer interface, CWN2, is completely lost in PEG400 with the loss of the interface.

CWN1 in PDB 6CN9 is surrounded by a conserved cluster of charged residues. These residues are in the active site and emanate from the Activation Loop and Catalytic Loop (referred to here as the AL-CL Cluster) (Figure 4A). Figure 4B shows the high conservation of these residues across human WNK isoforms and WNKs from other species including D. melanogaster. Activation Loop residues E388, K381, and K375 of WNK1 interact with Catalytic Loop residues WNK1/D349, K351, and D353, respectively, forming a cage that houses CWN1 (Figure 4A). WNK1/D349 and WNK1/K351 are pan-kinase catalytic residues. Additional residues in the active site cavity, WNK1/K310, K375, K381, and Y420, contribute to the AL-CL Cluster. WNK1/E388 (WNK3/E314) is noteworthy because it is on the Activation Loop that undergoes large conformational changes upon phosphorylation (Akella et al., 2020).

We used mutational analysis to explore the role the AL-CL Cluster (and by proxy, CWN1), in WNK activity and regulation. Mutational analysis was carried out primarily in WNK3, the most active WNK in our hands. Activation Loop residues, WNK3/E314 (WNK1/E388), and K307 (WNK1/K381), and a residue in the Catalytic Loop, WNK3/D279 (WNK1/D353) were mutated. Additional AL-CL Cluster residues K236, M301 (lysine in WNK1), and Y346 were also mutated. AL-CL Cluster mutants were expressed in Escherichia coli, purified, and the state of phosphorylation assessed. Mutant expression levels varied, but all were sufficient for assays (Supplementary file 1c). Activation Loop phosphorylation at WNK3/S308 (WNK1/S382) and WNK3/S304 (WNK1/S378) is necessary for WNK activity (Xu et al., 2002). Chymotrypsin-derived peptides were monitored by liquid chromatography–mass spectrometry (LC–MS) to determine Activation Loop phosphorylation state (Supplementary file 1d). Each of the mutants was about 95% phosphorylated as expressed, suggestive of proper folding (Supplementary file 1c,d). Mutant proteins were dephosphorylated to form uWNKs. A range of autophosphorylation phenotypes (both more and less active than wild-type) was observed (Figures 5 and 6).

The higher activity mutants, WNK3/E314A and WNK3/E314Q are consistent with our model that CWN1 in uWNKs and the AL-CL Cluster that supports CWN1 are inhibitory. To determine whether mutation at this position is activating in other WNKs, we generated WNK1/E388A. As with WNK3, WNK1/E388A is more active than wtWNK1 (Figure 5E, F, Supplementary file 1c). (Also as expected, the basal autophosphorylation activity of wt uWNK1 is less than uWNK3, as discussed in the Introduction.) To determine whether chloride inhibition is retained in the higher activity mutants, the effects of chloride on uWNK3 (S308) and uWNK1 (S382) autophosphorylation were tracked. Under the conditions used, chloride is apparently less inhibitory of WNK3/E314A autophosphorylation (Figure 5D). Using WNK1/E388A, and in a slower reaction regime, the reduction in chloride inhibition is more obvious (Figure 5E, F). These data were fit to a simple autocatalytic kinetic model in Dynafit (Kuzmič, 2009) (see Methods). The modeling recapitulated the higher apparent activity and weaker chloride binding of WNK1 (Figure 5E, F, Supplementary file 1f). The high activity of the mutants WNK3/E314Q, WNK3/E314A, and WNK1/E388A suggests a loss of the CWN1-stabilized inhibitory conformation.

Figure 6 describes the activity and chloride sensitivity of wild-type uWNK3 (Figure 6A) and mutants that have either autophosphorylation activity similar to wild-type (Figure 6B, C); or lower activity than wild-type (Figure 6D–F). WNK3/K236 is on the periphery of the AL-CL Cluster and the mutant WNK3/K236A exhibits little change from wild-type. K307A also is very similar to wild-type, even though it is adjacent to the phosphorylation site at Ser308. The mutants WNK3/D279N, M301A, and Y346F show related and complex phenotypes. These mutants are less active than WT and much more sensitive to chloride (Figure 6D–F). In addition to WNK3/S308, WNK3/S304 is also autophosphorylated (extent described in Akella et al., 2021). The effects of mutation on S304 phosphorylation mirrored the effects on S308 phosphorylation: wtWNK3 and WNK3/E314A were the most active, K236A and K307A were similar to wild-type, and M301A, Y346F, and D279N were the least active (Figure 5—figure supplement 1). S304 phosphorylation also appears to be more sensitive to chloride than S308 phosphorylation in the less active mutants (Figure 5—figure supplement 1).

The osmolyte PEG400 was used to put a demand on solvent in vitro as a mimic of osmotic pressure effects in cells (Kuznetsova et al., 2014). We previously observed that PEG400 stimulates ATP consumption by WNK3 in the presence of substrate GST-OSR1 (Akella et al., 2021). Unlike the LC–MS method used to study chloride inhibition by measuring Activation Loop autophoshorylation, ADP-Glo measures total ATP consumption and is compatible with PEG400.

We used ADP-Glo to measure GST-OSR1 substrate and autophosphorylation in all six WNK3 mutants (Figure 7). The ADP-Glo assay results are generally similar to the LC–MS autophosphorylation assays in that mutation of WNK3/E314 and WNK3/K236 show high activity whereas other AL-CL Cluster mutants are less active. The mutant WNK3/K307A showed lower activity in the ADP-Glo assays than in the MS-based autophosphorylation assays. Osmotic effects on WNKs measured in vitro with osmolytes are generally smaller than the chloride effects (Akella et al., 2021; Piala et al., 2014). None of the active single-point mutants were sufficient to eliminate the activating effect of osmolyte.

SLS was performed on wild-type and mutant uWNK3 between 0.8 and 1.8 mg/ml at room temperature as described previously (Akella et al., 2021; Supplementary file 1g). At 0.8 mg/ml, wild-type uWNK3 exhibits an apparent mass of ~60 kD, suggestive of a 50/50 mixture of 40 kD monomer and 80 kD dimer. At 1.8 mg/ml uWNK3 is dimeric. In contrast, wild-type uWNK1 is monomeric at both concentrations. The WNK3 mutants generally exhibited 50/50 mixtures at 0.8 mg/ml, similar to wild-type uWNK3. Two of the mutants, WNK3/E314Q and WNK3/D279N were monomeric. The appearance of monomers on mutating the AL-CL Cluster support the idea that the waters in CWN1 promote the dimer.

To address whether conformational changes induced by the mutations correlate with the effects of PEG400, WNK3/SA/E314A and the corresponding WNK1/SA/E388A were crystalized. Crystals of WNK1/SA/E388A did not diffract. WNK3/SA/E314A crystallized, affording limited diffraction to 3.4 Å (see Materials and methods and Supplementary file 1h). WNK3/SA/E314A crystals were in space group P1, and adopts a unique dimer in which the Activation Loops are swapped. The significance of this domain swapping is unclear. Excluding the Activation Loops, comparisons of uWNK3/SA/E314A, uWNK1/SA, and uWNK1/SA/PEG400 show that uWNK3/SA/E314A is more similar to uWNK1/SA/PEG400 than uWNK1/SA (with r.m.s.d.’s of 0.7 versus 0.85, respectively). Furthermore, the orientations of the 3/10 helix and helix C most resemble WNK1/SA/PEG400 (Figure 8A). The greater similarity of uWNK3/SA/E314A to uWNK1/SA/PEG400 than to uWNK1/SA is apparent in the plot of conformational differences as a function of sequence (blue trace smaller than cyan trace, Figure 8B).

Purification of WNK1 kinase domain for crystallization was performed following protocols in Min et al., 2004. The kinase domain of WNK3 (WNK3(118-409)) expresses well in BL21(DE3) E. coli cells. WNK3 and mutants were expressed and purified in pET29b using modifications of the original protocol (Min et al., 2004). Colonies were grown in terrific broth (TB) containing the antibiotics kanamycin and chloramphenicol on an INFORS shaker overnight at 37°C. 800 μl of cultured cells were transferred to 50 ml of TB at 37°C and grown to OD₆₀₀ = 1.8; then 16 ml were transferred to 1 l of TB and grown to OD₆₀₀ = 1.3. Protein expression was induced with 0.5 mM isopropyl thiogalactopyranoside. Cultures were grown for 18 hr, at 18°C and 180 rpm. Cells were pelleted by centrifugation and flash frozen in liquid N₂. Cells were lysed in an Avastin cell disruptor, and the lysate ultracentrifuged. The WNK3 was extracted from the supernatant on a NiSO₄-charged Sepharose column (GE) which was loaded in 50 mM Tris–HCl pH 8, 500 mM NaCl, 5 mM imidazole, and eluted in 250 mM imidazole. After dialysis in 50 mM Tris–HCl pH 8, 50 mM NaCl, 1 mM etheylenediamine tetra acetic acid (EDTA), and1 mM dithiothreitol, the WNK3s were further purified a Mono-Q 10/100 GL column (GE) eluted with 1 M NaCl. Samples were concentrated using an Amicon (Millipore), and buffer exchanged to 50 mM hydroxyethyl piperazine ethanesulfonic acid (HEPES) pH 7.4, 50 mM NaCl, 1 mM EDTA, and 1 mM Tris carboxyethyl phosphene. The HEPES buffer was made using HEPES free acid (Fisher) and adjusted to pH 7.4 using Tris base (Fisher). Gel filtration was performed on a Superdex-75 16/60 column. WNK1, WNK1/E388A used for assays were expressed and purified using similar protocol as WNK3.

WNK1/3 is phosphorylated (pWNK1/3) as purified from E. coli. The extent of phosphorylation of the different mutants is in Supplementary file 1c. Phosphorylated WNK1/3 (pWNK1/3) mutants were dephosphorylated using PP1cγ and λ-phosphatase in a 10:1 ratio, in 0.5 mM MnCl₂. Phosphatases were removed using Ni-NTA and gel filtration chromatography as described (Akella et al., 2021). The uWNK1/3 was stored in 50 mM HEPES pH 7.4 and 150 mM NaCl.

Crystals of WNK1/SA were obtained as reported previously (Akella et al., 2021; Min et al., 2004), grown in 24% PEG 2000, 300 mM NaCl, and 100 mM HEPES, pH 7.5, and cryoprotected in 15% glycerol, 26% PEG2000, and 300 mM NaCl, pH 7.5. Crystals were in space group P1, with cell constants a = 38.30 Å, b = 57.8 Å, c = 65.7 Å, α = 91.3°, β = 89.99°, γ = 90.89°. Recently, we obtained crystals under different condition of reduced chloride containing 200 mM potassium formate and 20% vol/vol PEG3350. These crystals when cryoprotected with 15% glycerol diffracted to 2.0 Å. The unit cell dimensions of a = 38.2 Å, b = 57.7 Å, c = 65.6 Å, α = 89.0°, β = 89.6°, γ = 89.2° were similar (0.19 Å r.m.s.d.) to that reported earlier (PDB 6CN9). These crystals were then exposed to 25% PEG400 in the same buffer for 25 min to determine the effect of PEG400 on the crystals. The space group changed to P2₁ with altered lattice constants, a = 38.3 Å, b = 56.8 Å, c = 65.3 Å, β = 95.4° without significant loss of resolution.

The unphosphorylatable mutant uWNK3/S308A/E314A was crystallized using tag-free protein. Crystallization was carried out in 8% Tacsimate, pH 8.0 10% PEG3350 (Hampton Research). These crystals were cryoprotected in 25% glycerol. The space group was P2₁, and the cell dimensions were a = 50.2 Å, b = 113.6 Å, c = 67.5 Å, β = 101.4°.

Assays were conducted in 20 mM HEPES, pH 7.4, 20 mM MgCl₂, 5 mM ATP and usually in 150 mM NaCl 30°C. Reactions were initiated by the addition of uWNK3 or uWNK1 to 4 μM. Substrate phosphorylation assays contained 40 μM gOSR1. Reactions were stopped by addition of guanidine-HCl to a final concentration of 1 M. The content of the assay buffer is given in Supplementary file 1e.

Chloride sensitivity of in vitro phosphorylation assays was caried out in a water bath at 30°C, in 20 mM HEPES pH 7.4, 20 mM Mg-gluconate and 5 mM ATP. Varying concentrations NaCl were added to the reaction mix. Chloride inhibition was measured at 50, 150, and 250 mM total Cl⁻. Reactions were started by adding 4 μM of uWNK3 or uWNK3 mutant. Aliquots of 50 μl were removed at time points: 0, 4, 10, 15, and 20 min and stopped by addition of guanidine-HCl to 1 M. Assay contents in Supplementary file 1e.

Chloride sensitivity of in vitro autophosphorylation assays was carried out at 30°C, in 20 mM HEPES pH 7.4, 20 mM Mg-gluconate, and 5 mM ATP. Varying concentrations NaCl were added to the reaction mix. Chloride inhibition was measured at 50, 150, and 250 mM total Cl⁻. Reactions were started by adding 4 μM of uWNK1 or uWNK1/E388A. Aliquots of 50 μl were removed at time points: 0, 10, 20, 30, 40, and 60 min and stopped by addition of guanidine-HCl to a final concentration of 1 M.

ADP-Glo reagent (Promega Inc) was used as a readout for autophosphorylation as described previously (Akella et al., 2021). This reagent is compatible with the high ATP concentrations required for uWNK autophosphorylation assays, and can be used, unlike mass spectrometry, in the presence of PEG400. 50 µl reactions contained 40 mM HEPES (pH 7.4), 10 mM MgCl₂, 4 µM pWNK, and 40 µM gOSR1. Final chloride concentration was maintained at 150 mM. The reaction was started by the addition of 5.2 mM ATP. Reactions were stopped after 15 min with 50 µl of ADP-Glo reagent in the presence of 100 nM WNK463 pan WNK inhibitor (Yamada et al., 2016). Manufacturer’s protocols were followed for the remaining steps of ATP depletion (40 min), conversion of ADP to ATP (1 hr). 100 μl aliquots were transferred to a 96-well plate and centrifuged for 2 min at 800 rpm. Luminescence was read on a CLARIOstar plate reader and data analyzed using MARS software (both reader and software, BMG Labtech, Ortenberg, GER). Data were further processed using GraphPad-Prism software.

SLS was conducted on a Wyatt DynaPro Nanostar DLS, at 25°C. Samples of uWNK3 were prepared at four concentrations (0.8, 1.2, 1.8, and 2.4 mg/ml) in 50 mM HEPES pH 7.4 and 150 mM NaCl. Samples were centrifuged at 16,100 × g for 10 min to remove aggregates and particles, before 5 μl were loaded into a quartz cuvette. Light scattering at 90° was monitored until the detector voltage was stable, and then the scattering was monitored for ten times for 5 s, with three replicates. Data were analyzed using Dynamics version 7.5.0.17 (Wyatt Technologies).

Differential scanning fluorimetry was used to measure the protein melt temperature using the lipophilic dye (Sypro Orange) (Pantoliano et al., 2001). Stock solutions (1) 200 µM WNK1/SA (194-483) in 50 mM HEPES, 50 mM NaCl, pH 7.4, (2) buffer (50 mM HEPES, 150 mM NaCl, pH 7.4), (3) 100% PEG400, and (4) 0.2 µl of SyproOrange (labeled 2.5X, Life Technologies Inc) were mixed. Final 100 µl solutions were 5 µM WNK1 and varying concentrations of PEG400. 25 µL were added to 3 wells in a Bio-Rad Multiplate 96-well clear PCR plate. Plates were read on a Bio-Rad CFX96 Real-Time PCR system. The temperature was increased from 4 to 80°C in 0.5°C steps and fluorescence measurements taken in the 6-FAM Fluorescein channel. The fluorescence response to the heat curve was fit to a binding isotherm with the inflection point taken as the melt temperature T_m.