WNK Inhibition Increases Surface Liquid pH and Host Defense in Cystic Fibrosis Airway Epithelia

In cystic fibrosis (CF), reduced HCO3− secretion acidifies the airway surface liquid (ASL), and the acidic pH disrupts host defenses. Thus, understanding the control of ASL pH (pHASL) in CF may help identify novel targets and facilitate therapeutic development. In diverse epithelia, the WNK (with-no-lysine [K]) kinases coordinate HCO3− and Cl− transport, but their functions in airway epithelia are poorly understood. Here, we tested the hypothesis that WNK kinases regulate CF pHASL. In primary cultures of differentiated human airway epithelia, inhibiting WNK kinases acutely increased both CF and non-CF pHASL. This response was HCO3− dependent and involved downstream SPAK/OSR1 (Ste20/SPS1-related proline-alanine-rich protein kinase/oxidative stress responsive 1 kinase). Importantly, WNK inhibition enhanced key host defenses otherwise impaired in CF. Human airway epithelia expressed two WNK isoforms in secretory cells and ionocytes, and knockdown of either WNK1 or WNK2 increased CF pHASL. WNK inhibition decreased Cl− secretion and the response to bumetanide, an NKCC1 (sodium-potassium-chloride cotransporter 1) inhibitor. Surprisingly, bumetanide alone or basolateral Cl− substitution also alkalinized CF pHASL. These data suggest that WNK kinases influence the balance between transepithelial Cl− versus HCO3− secretion. Moreover, reducing basolateral Cl− entry may increase HCO3− secretion and raise pHASL, thereby improving CF host defenses.

Cystic fibrosis (CF) is an inherited, multisystem channelopathy caused by mutations in the CFTR (cystic fibrosis transmembrane conductance regulator) gene (1)(2)(3). Loss of CFTR protein function reduces anion secretion, disrupts epithelial function, and impairs airway host defense. These abnormalities result in chronic airway obstruction, inflammation, infection, tissue destruction, and bronchiectasis and limit the life span of affected individuals.
CFTR is an apical HCO 3 2 and Cl 2 channel (4-6). In airway epithelia, these transport activities control the acid-base balance and composition of the thin film of liquid, the airway surface liquid (ASL), that covers the apical membrane. The ASL interfaces with the environment and mediates at least two vital respiratory host defenses (7)(8)(9). Mucociliary clearance uses gel-forming mucins to trap inhaled particles and ciliary beating to propel them out of the airways. Secreted antimicrobial peptides disrupt bacterial cell membranes and kill inhaled pathogens. An abnormally acidic pH of the ASL (pH ASL ) resulting from reduced CFTR-mediated HCO 3 2 secretion impairs these respiratory defenses (10)(11)(12)(13)(14)(15)(16)(17)(18). Importantly, ASL alkalinization rescues these defects and may benefit individuals with CF independent of CFTR genotype (19)(20)(21).
We considered that knowledge of HCO 3 2 transport in nonairway epithelia might yield insights relevant to CF airways. In several epithelia, the WNK (with-nolysine [K]) kinases act as key regulators of anion transport (30,31). WNK kinases are serine/threonine protein kinases that modify surface expression or activity of membrane transporters. In the pancreas, which shares similarities with airway HCO 3 2 transport, WNK kinases control ductal HCO 3 2 secretion (26,32). In one study of mouse pancreatic duct, silencing of WNK kinases increased, and WNK expression decreased HCO 3 2 secretion (33). In other reports, these kinases were shown to modulate CFTR HCO 3 2 channel activity (34,35) and membrane expression of SLC26 (solute carrier 26) family transporters (36). However, whether WNK kinases coordinate HCO 3 2 secretion across human airway epithelia remains poorly understood.
In this study, we tested the hypothesis that WNK kinases regulate CF pH ASL . We studied primary cultures of differentiated human airway epithelia and applied pharmacologic and genetic interventions to elicit responses. Our results show that airway epithelia express two WNK isoforms, WNK1 and WNK2, in secretory cells and ionocytes. Importantly, reducing WNK kinase activity increases pH ASL and enhances key respiratory host defenses that are otherwise impaired in CF.

Methods
Additional details on materials and methods are in the data supplement.

Cell Culture
Airway epithelial cells were harvested from human lungs procured as postmortem specimens, as explants from patients undergoing lung transplant, or as lungs deemed unfit for transplant. Informed consent for use in research was obtained. All studies were approved by the University of Iowa Institutional Review Board. Proximal bronchi were dissected, cut into small pieces, and enzymatically digested. Epithelial cells were isolated and seeded without passage onto collagen-coated inserts (Costar, 3470; Falcon, 353180). Cell culture medium comprised a 1:1 mixture of Dulbecco's modified Eagle medium/F-12, supplemented with 2% Ultroser G (Sartorius). Epithelia were differentiated at the air-liquid interface for 3 weeks or more before assay (37). During the course of this study, new CF lung donors became scarce, partly due to more individuals taking highly active CFTR modulators. To manage this situation, epithelial cells from previous donors with CF cryopreserved at P0 were thawed and differentiated. Table E1 in the data supplement reports genotypes of CF donors included in this study. Whenever feasible, studies followed a paired design so that epithelia from the same donor were assayed under control and treatment conditions. In experiments shown in Figure 2B, differentiated airway epithelia were generated from cryostocks of transformed human airway epithelial cell lines NuLi-1 (wild-type [WT]/WT) and CuFi-4 (G551D/DF508), as previously reported (38). These cell lines were used as additional models to test CFTR dependence of pH ASL responses evoked by inhibiting WNK kinases. To assess cytokineinduced responses, epithelia were treated with a combination of 10 ng/ml TNFa (R&D Systems) and 20 ng/ml IL-17 (R&D Systems). Both cytokines were added to the Ultroser G-supplemented basolateral media for 48 hours before assessments.

Pharmacologic Reagents
WNK463 and ivacaftor were purchased from Selleckchem. Other reagents were purchased from MilliporeSigma.

Single-Cell RNA-seq and Analysis
Cells for scRNA-seq (single-cell RNA sequencing) were obtained from primary cultures of human airway epithelia. The epithelia were grown at the air-liquid interface for 3 weeks or more before assay. The cell culture methods were the same as reported above. At the time of assessment, all epithelia were visibly dry on the apical side. Electrophysiologic assessments in Ussing chambers showed a mean basal transepithelial conductance (G t ) of 3.2 mS/cm 2 , resistance of 433.3 V.cm 2 , and short-circuit current (I SC ) of 72.6 mA/cm 2 . These properties indicated well-differentiated, polarized epithelia performing electrogenic ion transport. Additional details about library preparation, sequencing methods, and bioinformatic analysis can be found in the supplement. The data are available in the National Center for Biotechnology Information's Gene Expression Omnibus (GEO) database (GEO GSE159056).

Immunocytochemistry
Airway epithelia were washed, fixed, permeabilized, and immunostained to reveal WNK1 and WNK2 expression. See supplemental methods for details. pH ASL Measurement pH ASL was measured using a fluorescent ratiometric pH indicator, SNARF-1, conjugated to 70 kD dextran (Thermo Fisher Scientific). Additional details are reported in the supplement.

Epithelial Host Defenses
Several assays were performed to assess epithelial defense mechanisms. 1) ASL viscosity was measured using the fluorescence recovery after photobleaching method (12). 2) Liquid absorption was measured using the micropipette technique (39). 3) Ciliary beat frequency was measured using phase contrast microscopy. 4) ASL antimicrobial activity was assessed using bacteria-coated grids (13). Additional details on these assays are provided in the supplement.

Electrophysiologic Studies
Airway epithelia were mounted in modified Ussing chambers (Physiologic Instruments) and bathed in symmetric Krebs buffer solution. Epithelia were voltage clamped, followed by recording of the I SC and G t . See supplemental information for details.
Bulk RNA-seq RNA isolation, library preparation, sequencing, and bioinformatics analysis were previously reported (41). RNA-seq data are available in the National Center for Biotechnology Information's GEO database (GEO GSE176121).

Statistics
Statistical significance testing was performed on GraphPad Prism 8 Software. Statistical tests included paired Student's t test for comparing two groups and one-way ANOVA with Tukey's multiple comparison test for comparing more than two groups. A P value of ,0.05 was considered significant.

Airway Epithelia Express WNK1 and WNK2
The four WNK isoforms are expressed in a tissue-specific manner (42,43). However, their expression in human airways remains relatively unexplored. Recent scRNA-seq studies have revealed considerable cellularlevel heterogeneity within airway epithelia with implications for ion transport (44)(45)(46). Notably, these studies have shown that secretory cells express nearly half the epithelial CFTR transcript, and the ionocytes, though rare, express the highest amount on a per-cell basis (44)(45)(46)(47). These cell types also express basolateral transporters involved in HCO 3 2 and Cl 2 secretion. To identify which WNK kinases might regulate anion transport across airway epithelia, we performed scRNA-seq. We studied primary cultures of differentiated airway epithelia from four different donors without CF and four different donors with CF and examined    Table E1 for cystic fibrosis transmembrane conductance regulator (CFTR) genotypes]. Dot plot showing cell type-specific expression of the four WNK isoforms, and STK39 (serine/threonine kinase 39) and OXSR1 (oxidative stress responsive kinase 1), which encode Ste20/SPS1-related proline-alanine-rich protein kinase (SPAK) and oxidative stress responsive 1 kinase (OSR1), respectively. CDH1 (e-cadherin) is included as a reference epithelial gene. For each dot, the size represents the detection rate in a particular cell type, and the color represents average gene expression for cells in which gene was detected. Data for CF ionocytes is not shown, as these cells were not detected in three out of four CF epithelia. Also see Figure E2. cell type-specific WNK gene expression ( Figures 1A and 1B). WNK1 was broadly expressed in all major cell types (i.e., secretory cells, ciliated cells, and basal cells), as well as ionocytes. WNK2 was also abundantly expressed in secretory cells and ionocytes. In contrast to WNK1, WNK2 was rarely detected in ciliated or basal cells. The remaining WNK isoforms, WNK3 and WNK4, were either not expressed, or expressed at a very low level. Importantly, WNK genes showed similar expression in CF versus non-CF epithelia ( Figure E2). We also studied the expression of the two main downstream kinases (i.e., STK39, which encodes Ste20/SPS1-related prolinealanine-rich protein kinase [SPAK], and OXSR1, which encodes oxidative stress responsive 1 kinase [OSR1]). Both genes were broadly expressed and abundantly detected in secretory cells as well as ionocytes.
To reveal WNK protein expression, we immunolabeled non-CF and CF epithelia for WNK1 and WNK2 ( Figures 1C and 1D). In agreement with scRNA-seq results, we detected WNK1 in ciliated as well as nonciliated cells and WNK2 predominantly in nonciliated cells. Further immunolocalization studies revealed WNK1 and WNK2 expression in secretory cells (labeled with anti-CC10 antibody) as well as ionocytes (labeled with anti-BSND antibody). Overall, these studies identified two WNK kinases in airway cells that secrete anions.

WNK Inhibition Increases CF pH ASL
Several HCO 3 2 and H 1 transport mechanisms integrate to determine pH ASL , and pH ASL influences host defense (13,19). To begin to understand the role of WNK kinases in regulating pH ASL , we used pharmacologic WNK inhibition. WNK463 is a selective, ATP-competitive, pan-WNK kinase inhibitor and has recently emerged as a useful tool for studying ion transport physiology (48-51). We exposed airway epithelia to either vehicle or WNK463 for 2 hours and measured pH ASL in an environment containing 25 mM HCO 3 2 and 5% CO 2 . In primary cultures of both CF and non-CF epithelia, WNK463 increased pH ASL ( Figure 2A). As an additional test, we also studied NuLi-1 (WT/WT) and CuFi-4 (G551D/DF508) epithelia and found that WNK463 elicited similar responses ( Figure 2B). Alkalinization in CF epithelia indicated that the WNK463-induced response did not require CFTR. However, exposure of WNK463-treated CuFi-4 epithelia to ivacaftor . WNK463 increases airway surface liquid pH (pH ASL ). Human airway epithelia were exposed to either vehicle or WNK463 (10 mM) for 2 hours, and pH ASL was measured using SNARF-1-dextran. (A) pH ASL responses in primary cultures of non-CF (n = 7) and CF epithelia (n = 8). (B) pH ASL responses in NuLi-1 (n = 6) and CuFi-4 epithelia (n = 6). Ivacaftor (10 mM) was applied for 2 hours in combination with WNK463. (C) pH ASL response in primary CF epithelia in the presence of HCO 3 2 /CO 2 and after replacing HCO 3 2 with HEPES and removing CO 2 from the environment (n = 4). (D) Time course of WNK463-evoked response in primary CF epithelia (n = 5). (E) Schematic showing direct versus indirect modulation of membrane transporters by WNK kinases. (F) pH ASL response in primary CF epithelia after 2-hour exposure to rafoxanide, a SPAK/OSR1 inhibitor (n = 4). In B, each data point is an epithelium derived from either NuLi-1 or CuFi-4 cells. In all other cases, each data point represents a primary differentiated airway epithelium from a different human donor. Data are shown as mean 6 SEM. Statistical significance was tested using two-way ANOVA with two-stage Benjamini, Krieger, and Yekuteli false discovery rate procedure for A, and ANOVA with post hoc Tukey's test for B, C, D, and F. *P , 0.05, ***P , 0.001, and ****P , 0.0001. Veh = vehicle.
further increased pH ASL . Ivacaftor increases open-state probability and function of CFTR-G551D channels. We concluded that inhibiting WNK supports ASL alkalinization through CFTR-independent as well as CFTRdependent mechanisms.
To further characterize the WNK463induced response, we performed additional studies in primary differentiated airway epithelia. When these epithelia were exposed to WNK463 in a nominally HCO 3 2 /CO 2free environment, the pH ASL response disappeared ( Figure 2C). This result suggested that WNK463 increased CF pH ASL by increasing HCO 3 2 secretion and not by decreasing H 1 secretion. Next, we asked whether this response was time and dose dependent. Two hours of exposure increased pH ASL , and continued exposure up to 24 hours did not further alkalinize ASL ( Figure 2D). Moreover, 10 μM WNK463 alkalinized but a lower dose (1 μM) did not alter pH ASL ( Figure E3).
WNK kinases modulate membrane transporters either directly or indirectly through their native substrates, SPAK and OSR1 (52) ( Figure 2E). In scRNA-seq data, cell types expressing WNK1 and WNK2 also expressed genes encoding SPAK and OSR1. To test the latter's involvement in controlling pH ASL , we treated CF epithelia with rafoxanide, an allosteric SPAK/OSR1 inhibitor (53). Similar to WNK463, rafoxanide applied for 2 hours also increased CF pH ASL ( Figure  2F). Taken together, these responses suggested that CF pH ASL is controlled by upstream as well as downstream kinases in the canonical WNK/SPAK/OSR1 signaling pathway.

WNK463 Enhances CF Host Defenses
Previous studies showed that alkalinizing CF ASL improves respiratory host defenses (12, [19][20][21]. Because WNK463 increased pH ASL , we tested its impact in CF epithelia. Defective mucus transport is a key feature of CF (54). In primary cultures of differentiated CF epithelia, WNK463 decreased ASL viscosity ( Figure 3A), consistent with previous studies showing that increasing pH ASL decreases viscosity. In addition, WNK463 did not alter the rate of apical liquid absorption ( Figure 3B), suggesting that a change in apical fluid volume was not involved. WNK463 also increased ciliary beat frequency, albeit modestly ( Figure 3C). Both decrease in viscosity and increase in ciliary beat frequency would improve CF mucus transport. Previous studies also indicated that CF ASL has reduced antibacterial activity (13,55). WNK463 increased ASLmediated Staphylococcus aureus killing in primary CF epithelia ( Figure 3D). Overall, these results suggested that targeting WNK kinase signaling may at least partially rescue CF host defense defects.
Either WNK1 or WNK2 Knockdown Increases CF pH ASL WNK463 is a pan-WNK kinase inhibitor (48). Because airway epithelia expressed two WNK kinases, we asked whether WNK1 or WNK2 controlled CF pH ASL . To test, we performed siRNA-mediated gene knockdown. Reducing either WNK1 or WNK2 expression increased CF pH ASL ( Figures 4A-4D). This result suggested that both isoforms, WNK1 and WNK2, participate in regulating CF pH ASL . Although analyses of additive and compensatory effects of WNK1 and WNK2 would require single-and double-knockout experiments rather than knockdown experiments, these knockdown studies suggest that if compensatory effects exist, they are incomplete.

WNK463 Reduces Electrogenic Cl 2 Secretion
CFTR is the main route for anion exit across the apical membrane of airway epithelia. Whether WNK inhibition alters CFTR activity in airway epithelia is not well established. To test, we exposed non-CF epithelia to WNK463 for 2 hours and assayed in Ussing chambers containing symmetric Krebs solution (118 mM Cl 2 and 25 mM HCO 3 2 , gassed with 5% CO 2 ). After clamping transepithelial voltage, we recorded I SC and G t and elicited responses to selective channel inhibitors or activators ( Figures 5A  and 5B). We added amiloride followed by DIDS (4,4 9 -diisothiocyano-2,2 9stilbenedisulfonic acid) to abolish ENaC (epithelial Na 1 channel)-mediated Na 1 absorption and CaCC (Ca 21 -activated Cl 2 channel)-mediated anion secretion, respectively. Next, we added forskolin to increase cellular cAMP and thereby phosphorylate and activate CFTR channels. We concluded with CFTR inh -172, an inhibitor of CFTR. We assessed the response to CFTR inh -172 and used it to estimate CFTR channel activity. WNK463 reduced DI SC -CFTR by 50% ( Figure 5C). However, DG t -CFTR remained unchanged ( Figure 5D). This result suggested that WNK463 decreased CFTR-mediated anion transport but did not alter CFTR channel activity at the apical membrane.
CF epithelia lack functional CFTR channels but express CaCC. Accordingly, we studied the effect of WNK463 on CaCCmediated anion transport in CF epithelia ( Figures 5E and 5F). After blocking ENaC with amiloride, we added uridine triphosphate, a P2Y2 purinergic receptor agonist that increases cytosolic [Ca 21 ] and thus activates CaCC. Next, we added DIDS, a nonspecific CaCC inhibitor, and recorded the change in I SC and G t . WNK463 decreased DIDS-sensitive DI SC but slightly increased DIDS-sensitive DG t (Figures 5G and 5H). Together, these findings suggested that inhibiting WNK kinases reduces anion secretion, but the effect is not on apical anion channels.
To separate the effects of WNK inhibition on Cl 2 versus HCO 3 2 transport, we repeated the studies in single anion solutions. In symmetric HCO 3 2 -free solution, WNK463 reduced DI SC -CFTR ( Figure 5I); however, in Cl 2 -free solution, DI SC -CFTR remained unchanged. Similar results were obtained for the DIDS-sensitive DI SC in CF epithelia ( Figure 5J). These data suggested that WNK inhibition reduces electrogenic Cl 2 secretion but does not alter electrogenic HCO 3 2 secretion.
Reducing Basolateral Cl 2 Entry Increases CF pH ASL Transcellular Cl 2 secretion involves the movement of Cl 2 across the apical and the basolateral membranes in series. Because studies of electrically conductive anion transport showed reduced Cl 2 secretion without major effects at the apical membrane, we asked whether a change at the basolateral membrane was involved. The loop-sensitive NKCC (Na 1 -K 1 -2 Cl 2 ) cotransporter is the main route for Cl 2 entry across the basolateral membrane, and WNK kinases are known to increase NKCC activity in renal epithelia (56,57). To further investigate the effect of WNK463 on this transport mechanism, we studied non-CF and CF epithelia in Ussing chambers. After blocking ENaC with amiloride, we added either forskolin to activate CFTR in non-CF epithelia, or uridine triphosphate to activate CaCC in CF epithelia. To estimate the contribution of NKCC1(sodium-potassiumchloride cotransporter 1), we added basolateral bumetanide and measured DI SC . WNK463 reduced bumetanide-sensitive I SC in both non-CF and CF epithelia ( Figures 6A  and 6B). This result pointed to the involvement of WNK kinases in controlling basolateral Cl 2 uptake through a bumetanide-sensitive mechanism. Previous studies in airway epithelia showed that bumetanide decreases intracellular [Cl 2 ] (58-60). This led us to hypothesize that lowering intracellular [Cl 2 ] might also increase CF pH ASL . To test, we performed two experiments: 1) We tested the effect of NKCC1 inhibition in CF epithelia. Exposure to bumetanide increased CF pH ASL ( Figure 6C, left panel). 2) We measured pH ASL in a Cl 2 -free environment. Similar to bumetanide, the removal of basolateral Cl 2 also increased CF pH ASL ( Figure 6C, right panel). Because WNK463 reduced bumetanide-sensitive I SC , and lowering NKCC1 activity or intracellular [Cl 2 ] increased pH ASL , we considered whether intracellular [Cl 2 ] was involved in the response evoked by WNK463. When introduced in the absence of Cl 2 , WNK463 failed to alkalinize CF ASL, thus indicating that the WNK463-elicited pH ASL response was Cl 2 dependent ( Figure 6D).

WNK463 Further Increases pH ASL in TNFa/IL-17-treated CF Epithelia
Airway inflammation is ubiquitous in individuals with CF after the first few weeks of life (61)(62)(63). The CF airway inflammation is characteristically neutrophil predominant, may develop in the absence of infection, and is further exacerbated by infection and colonization. Two CF-relevant inflammatory cytokines, TNFa and IL-17, drive neutrophilic inflammation (64)(65)(66)(67). In previous work, combined TNFa/IL-17 increased HCO 3 2 secretion and CF pH ASL by increasing pendrin expression (41,68). We asked if TNFa/IL-17-induced alkalinization was also accompanied by altered expression of WNK kinases. In gene expression studies, TNFa/IL-17 modestly reduced WNK1 and markedly reduced WNK2 expression (Figures 7A and 7B). In immunocytochemistry studies, TNFa/IL-17 decreased WNK2 detection, but WNK1 remained unchanged (Figures 7C and 7D). This led us to hypothesize that residual WNK kinases might continue to regulate HCO 3 2 secretion in cytokine-treated epithelia. Accordingly, exposure to WNK463 further increased pH ASL in CF epithelia treated with TNFa/IL-17 ( Figure 7E). Because WNK463 decreased Cl 2 secretion, and reducing basolateral Cl 2 entry increased CF pH ASL , we predicted a similar response to lowering basolateral Cl 2 entry in cytokine-treated CF epithelia. Exposure to bumetanide further alkalinized ASL in TNFa/IL-17-treated CF epithelia ( Figure 7F). This result suggested that TNFa/IL-17 shifted apical anion secretion in favor of HCO 3 2 over Cl 2 , and lowering basolateral Cl 2 entry further augmented this response.

Discussion
Our transcript and immunocytochemistry data for WNK1 and WNK2, and their substrates STK39 and OXSR1, indicated that these kinases are expressed in secretory cells and ionocytes, the main airway epithelial , and (J) shows DI SC response with apical DIDS in CF epithelia (n = 5-6). Each data point represents an epithelium from a different donor. Data are shown as mean 6 SEM. In some cases, error bars are hidden by symbols. Statistical significance was tested using paired Student's t test. *P , 0.05. Veh = vehicle. cells that secrete anions. Consistent with that localization, pharmacologically inhibiting WNK kinases, SPAK/OSR1 kinases, and knocking down WNK1 and WNK2 transcripts increased pH ASL . These results thus identified an important role for WNK kinases in regulating HCO 3 2 secretion across airway epithelia. Figure 8 shows a tentative model for how WNK1 and WNK2 may influence Cl 2 and HCO 3 2 secretion and the ratio between the two transport processes in CF airway epithelia. Some features in this model are unknown at present and are an opportunity for future research.
Electrophysiological studies indicated that inhibiting WNK kinases decreased the Cl 2 -mediated, but not HCO 3 2 -mediated current. Moreover, ASL alkalinization persisted in the absence of CFTR activity in CF epithelia. A clue to a potential mechanism came with the finding that inhibiting WNK kinases largely eliminated the inhibitory effect of basolateral bumetanide on I SC . Bumetanide inhibits NKCC1, the major pathway for Cl 2 entry into the cell, and thereby reduces the intracellular [Cl 2 ] (58-60). Further evidence implicating intracellular [Cl 2 ] came from studies showing that adding bumetanide alone or removing Cl 2 from the medium also alkalinized ASL in CF epithelia.
These results suggest that WNK kinases may play a key role in determining the balance between Cl 2 secretion and HCO 3 2 secretion across airway epithelia (Figure 8). Inhibiting WNK kinases decreased NKCC1 activity, which decreased Cl 2 secretion, increased HCO 3 2 secretion, and increased pH ASL . The inference that inhibiting WNK reduces NKCC1 activity is supported by the finding that bumetanide also increased pH ASL and previous reports that WNK kinases increase NKCC activity in nonairway epithelia (56,57). However, the mechanism that increases HCO 3 2 secretion is uncertain. One possibility is that WNK inhibition reduces the intracellular [Cl 2 ], thereby increasing the driving force for Cl 2 / HCO 3 2 exchange at the apical membrane and hence HCO 3 2 secretion. Finding that bumetanide replicates the effect of WNK inhibition on pH ASL is consistent with this hypothesis. However, Cl 2 -free bathing solution also induced HCO 3 2 secretion despite the fact that reduced [Cl 2 ] would initially and transiently drive Cl 2 /HCO 3 2 exchange in the opposite direction, and over the 2-hour time course of the experiment, cellular Cl 2 would be largely depleted. Thus, we favor an alternative explanation that intracellular Cl 2 is a signaling molecule that regulates membrane transport (69).
Intracellular [Cl 2 ] regulation of HCO 3 2 secretion has been reported previously. A Cl 2 -sensing motif has been identified in some HCO 3 2 transporters and other proteins (35,70). Low intracellular [Cl 2 ] was shown to increase IRBIT (IP3 receptor binding protein released with IP3)stimulated NBCe1-B (electrogenic sodium bicarbonate cotransporter 1) activity (70). Kim and colleagues showed that low intracellular [Cl 2 ] enabled structural association between WNK1 and CFTR and increased CFTR HCO 3 2 channel activity (35). Notably, this effect did not depend on WNK1 kinase activity. Yamaguchi and colleagues developed a computational model of guinea pig pancreatic duct HCO 3 2 secretion (71). In this model, maximal HCO 3 2 secretion did not depend on an increase in CFTR HCO 3 2 permeability or a change in SLC26 Cl 2 /HCO 3 2 exchange stoichiometry, but instead depended on suppression of basolateral Cl 2 uptake. The addition of NKCC1, normally missing from guinea pig pancreatic ducts, increased intracellular [Cl 2 ] and reduced secreted [HCO 3 2 ]. Our results also support intracellular [Cl 2 ]-dependent regulation of HCO 3 2 secretion. It will be important for future studies to establish underlying molecular mechanisms in airway epithelia.
We previously reported that combined TNFa/IL-17 increased production of pendrin, an apical Cl 2 / HCO 3 2 exchanger, and alkalinized CF ASL (41,68). Here, we show that TNFa/IL-17 also reduced WNK2 expression. Moreover, inhibiting the residual WNK kinase activity with WNK463 further increased pH ASL , and basolateral bumetanide mimicked the effect of WNK inhibition. These proinflammatory cytokines may thus induce HCO 3 2 secretion and increase pH ASL by at least two mechanisms, increasing pendrin During I SC recording, amiloride was added to inhibit ENaC (epithelial Na 1 channel), followed by forskolin (non-CF) or UTP (uridine triphosphate) (CF) to maximally activate CFTR or CaCC (Ca 21 -activated Cl 2 channel), respectively. At this point, basolateral bumetanide was introduced, and DI SC was recorded (n = 6 different non-CF, or 9 different CF donors). (C) Left panel: CF pH ASL response 2 hours after exposure to basolateral bumetanide (n = 6 different donors); right panel: CF pH ASL in the presence or absence of basolateral Cl 2 (n = 6 different donors). (D) CF pH ASL response in epithelia exposed to WNK463 (10 mM) for 2 hours in the presence or absence of basolateral Cl 2 (n = 6 different donors). Data are shown as mean 6 SEM. Statistical significance was tested using paired Student's t test. *P , 0.05 and **P , 0.01. Burnett = basolateral bumetanide. expression and reducing WNK2 expression. Whether WNK kinases regulate apical expression or activity of pendrin in cytokine-treated airway epithelia remains to be determined. Some reports have suggested an interaction between pendrin and apical anion channels that may increase activity of both transporters (72); additional studies are needed to fully understand whether WNK kinases modulate such interactions.
WNK kinases have an ATP-binding site that is unique among protein kinases, and WNK463 targets this site. Previous studies have found WNK463 to be highly selective (48), but it is possible that higher doses may also affect other targets. To confirm the effect of reducing WNK activity on pH ASL , we used two orthogonal approaches (i.e., pharmacologic inhibition and gene knockdown). Moreover, inhibiting downstream WNK targets (i.e., SPAK/OSR1 and NKCC1) also increased pH ASL . Overall, these results point to a key role for the WNK signaling pathway in controlling airway HCO 3 2 secretion and pH ASL .
This study has several advantages. First, we studied primary cultures of differentiated human airway epithelia from both CF and non-CF genotypes. Second, to account for biological variability, we included epithelia from multiple human donors. Third, we measured pH ASL under thin-film conditions without adding additional apical fluid. Fourth, in testing our hypothesis, we used a combination of pharmacologic, transcriptomic, gene silencing, protein immunolabeling, and electrophysiologic approaches. Fifth, although all studies and interventions were performed in primary cultures of differentiated airway epithelia, a preliminary study in established human airway epithelial cell lines, NuLi-1 and CuFi-4, yielded similar results. Finding that WNK signaling is active in these epithelia enables their use as models for studying WNK signaling.
This study also has limitations. First, we used human airway epithelia as an in vitro model, and assessing WNK kinase inhibition in vivo may be of value. However, interpretation of in vivo effects may be complicated by the fact that WNK kinases are expressed broadly in epithelial and nonepithelial cells (42,43). Newer animal models with tissue-or cell-specific WNK knockouts might help further elucidate roles of WNK kinases. Second, we did not identify the transporter directly responsible for apical HCO 3 2 exit. RNA-seq studies show that CF airway epithelia express several non-CFTR HCO 3 2 transporters, including CaCC and SLC26 family members (41,73), and WNK inhibition may affect more than one simultaneously. Third, inflammation in CF airways is a complex process, and it will be important for future studies to characterize the effects of other proinflammatory mediators (e.g., IL-8, IL-1b, etc.) on WNK signaling.
The results have implications for CF airways. First, previous studies have shown that the abnormally acidic pH ASL observed in newborns with CF increases with time and inflammation (74,75), although not to levels observed in non-CF epithelia studied under comparable conditions (41). As indicated above, our current data, together with previous results, suggest complex regulatory mechanisms are responsible. Second, loop diuretics, which inhibit NKCC, are commonly used to treat heart failure and fluid overload states (76). Yet, to our knowledge, these agents have not been shown to cause adverse airway phenotypes. This study suggests that HCO 3 2 secretion may compensate for any decrease in (C) Immunostaining for WNK1 and WNK2 in control and TNFa/IL-17-treated CF epithelia. Scale bar, 5 mm. (D) Intensity of WNK1 or WNK2 immunolabeling quantitated as integrated density using imageJ software (n = 5-6). (E and F) pH ASL responses in TNFa/IL-17-treated CF epithelia. WNK463 (10 mM) or bumetanide (100 mM) were applied for 2 hours before pH ASL measurement (n = 6-8). Each data point represents an epithelium from a different donor. Data are shown as mean 6 SEM. Statistical significance was tested using paired Student's t test (A, B, D, and F) or ANOVA with post hoc Tukey's test (E). *P , 0.05, **P , 0.01, and ****P , 0.0001.
loop-sensitive Cl 2 secretion and preserve, if not augment, host defenses (77). Third, by enhancing respiratory host defense, WNK inhibition might be a potential therapeutic target in CF and possibly in acquired CFTR dysfunction, such as that induced by cigarette smoking (78). Although WNK463 produced adverse effects in a rat model of hypertension (48), it is a nonselective WNK kinase inhibitor. Interestingly, WNK2 has a more restricted tissue expression than the ubiquitous WNK1, it is detected in airway epithelial cell types relevant for anion secretion, and its knockdown alkalinizes CF ASL. Thus, selective WNK2 inhibitors, inhibitors of downstream SPAK/OSR1 kinases, or inhibitors restricted to the airways might be pursued as potential CF therapeutics.
Author disclosures are available with the text of this article at www.atsjournals.org.  Figure 8. WNK kinases may regulate the ratio of HCO 3 2 versus Cl 2 secretion in CF airway epithelia. Model shows a CF airway epithelial cell in which the absence of CFTR function decreases HCO 3 2 and Cl 2 secretion, generates an acidic pH ASL , and impairs respiratory host defenses. Left panel: WNK1 and WNK2 regulate activity of basolateral NKCC1 (sodium-potassium-chloride cotransporter 1) via intermediate SPAK/OSR1 (Ste20/SPS1-related proline-alanine-rich protein kinase/oxidative stress responsive 1 kinase). NKCC1 imports Cl 2 into the cell. Cl 2 and HCO 3 2 exit from the cell across the apical membrane via channels and/or transporters whose identity and quantitative contributions are uncertain (indicated by question marks). Right panel: reducing WNK kinase activity, inhibiting NKCC1, and eliminating Cl 2 lower the intracellular chloride concentration and transepithelial Cl 2 secretion. At the same time, these interventions increase HCO 3 2 secretion, raise pH ASL , and improve CF host defenses. We speculate that intracellular Cl 2 acts as a signaling molecule that regulates apical and/or basolateral HCO 3 2 transporters. Some features of this model are tentative at present and are opportunities for future research. See text for details.