Hypotonicity Stimulates Potassium Flux through the WNK-SPAK/OSR1 Kinase Cascade and the Ncc69 Sodium-Potassium-2-Chloride Cotransporter in the Drosophila Renal Tubule*

Background: Hypotonicity stimulates transepithelial ion and water flux in the Drosophila renal tubule through unknown mechanisms. Results: The with-no-lysine (WNK) and Ste20-related proline/alanine-rich (SPAK)/oxidative stress response (OSR1) kinase cascade regulates transepithelial K+ flux through the fly sodium-potassium-2-chloride cotransporter (NKCC). Conclusion: The kinase/NKCC pathway is required for the tubule's response to hypotonicity. Significance: This provides a molecular mechanism by which flies osmoregulate. The ability to osmoregulate is fundamental to life. Adult Drosophila melanogaster maintain hemolymph osmolarity within a narrow range. Osmolarity modulates transepithelial ion and water flux in the Malpighian (renal) tubules of the fly, which are in direct contact with hemolymph in vivo, but the mechanisms causing increased transepithelial flux in response to hypotonicity are unknown. Fly renal tubules secrete a KCl-rich fluid. We have previously demonstrated a requirement for Ncc69, the fly sodium-potassium-2-chloride cotransporter (NKCC), in tubule K+ secretion. Mammalian NKCCs are regulated by a kinase cascade consisting of the with-no-lysine (WNK) and Ste20-related proline/alanine-rich (SPAK)/oxidative stress response (OSR1) kinases. Here, we show that decreasing Drosophila WNK activity causes a reduction in K+ flux. Similarly, knocking down the SPAK/OSR1 homolog fray also decreases K+ flux. We demonstrate that a hierarchical WNK-Fray signaling cascade regulates K+ flux through Ncc69, because (i) a constitutively active Fray mutant rescues the wnk knockdown phenotype, (ii) Fray directly phosphorylates Ncc69 in vitro, and (iii) the effect of wnk and fray knockdown is abolished in Ncc69 mutants. The stimulatory effect of hypotonicity on K+ flux is absent in wnk, fray, or Ncc69 mutant tubules, suggesting that the Drosophila WNK-SPAK/OSR1-NKCC cascade is an essential molecular pathway for osmoregulation, through its effect on transepithelial ion flux and fluid generation by the renal tubule.

The ability to osmoregulate is a fundamental property of living organisms across kingdoms (1). Single cell organisms as well as individual cells within multicellular organisms regulate internal osmolarity and, consequently, cell volume. Multicellular organisms will often maintain the osmolarity of extracellular fluids within a narrow range in order to avoid perturbations in cell volume and intracellular osmolarity, which may have deleterious effects on cell function.
Adult Drosophila melanogaster maintain extracellular osmolarity within a narrow range despite environmental stressors (2), but how osmoregulation is achieved is poorly understood. Insect iono-and osmoregulation is achieved through epithelial ion transport across the midgut, Malpighian (renal) tubules, and hindgut/rectum (3,4). Because the renal tubules are in direct contact with hemolymph (the blood compartment of the insect), they can directly sense changes in extracellular osmolarity. In adult Drosophila, fluid secretion rates in renal tubules vary inversely with osmolarity: decreased fluid secretion at higher osmolarity and increased secretion at lower osmolarity (5). Increased osmolarity inhibits tubule sensitivity to the diuretic effects of tyramine (5), but the mechanism by which hypotonicity stimulates fluid secretion is unknown.
Fluid secretion in the main segment of the adult fly tubule is driven by secretion of a potassium chloride-rich fluid. This occurs through the parallel transepithelial secretion of cations, primarily K ϩ , through the principal cell and Cl Ϫ secretion through the stellate cell (6 -9). We have previously shown that the fly sodium-potassium-2-chloride cotransporter (NKCC), 2 Ncc69, mediates ϳ30% of principal cell transepithelial K ϩ transport (10).
The solute-linked carrier 12 (SLC12) family of sodium-coupled cotransporters (the sodium-coupled cotransporter (NCC) and NKCC1 and NKCC2) play important roles in mammalian renal physiology. NCC and NKCC2 mediate sodium chloride reabsorption in the distal convoluted tubule and thick ascending limb, respectively (11), whereas NKCC1 plays a role in transepithelial ion transport in the intercalated cell of the collecting duct (12)(13)(14). In addition, the activity of these transporters also influences the renal handling of water, divalent cations, and acid-base homeostasis. Regulation of these transporters is therefore key in allowing the kidney to respond to variations in dietary solute and water intake, and pathogenic dysregulation leads to abnormalities in serum electrolytes, osmolarity, and blood pressure in humans and in mouse models (11).
The WNK-SPAK/OSR1 kinase cascade is a key regulator of NCC and NKCC transport activity. Mutations in WNK1 and WNK4 result in human disease, pseudohypoaldosteronism type II (also known as familial hyperkalemia and hypertension or Gordon's syndrome), characterized by high serum potassium levels and high blood pressure (15). A number of in vitro, cultured cell, and Xenopus oocyte studies support a model in which WNK kinases phosphorylate the related SPAK and OSR1 kinases on two sites, thereby activating SPAK/OSR1 (16 -18). SPAK and OSR1 then phosphorylate the N termini of NCC, NKCC1, and NKCC2 on conserved serines and threonines, increasing transporter activity (18 -26), although some studies suggest that WNK4 negatively regulates NCC (27)(28)(29)(30)(31)(32), at least under certain conditions. In vivo data are also consistent with positive regulatory roles of this kinase cascade in the regulation of NKCC2 in thick ascending limb and NCC in the distal convoluted tubule (33)(34)(35)(36)(37)(38)(39). However, although multiple studies have examined the effects of the WNK-SPAK/OSR1 kinase cascade on transporter phosphorylation in vivo, the effects on transporter activity have almost exclusively been examined in heterologous systems, such as Xenopus oocytes or HEK293 cells. However, the ion transport physiology of individual cells versus transporting epithelia may be different. Only one study, examining a dominant-negative isoform of WNK1, has directly demonstrated a role for WNK regulation of NKCC in renal tubule transepithelial ion flux (40).
Our goal was to examine roles for the WNK-SPAK/OSR1-NKCC pathway in the Drosophila renal tubule, a physiologically accessible transporting epithelium that also offers ease of genetic manipulation. Here, we demonstrate that Drosophila WNK and the SPAK/OSR1 homolog Fray regulate transepithelial K ϩ flux in the main segment of the fly renal tubule through their actions on Ncc69. Activation of the WNK-SPAK/OSR1-NKCC pathway in hypotonic conditions results in increased K ϩ flux and fluid secretion by the Malpighian tubules. Because the tubules generate an isosmotic fluid that is subsequently modified (similar to the glomerular filtrate), increased fluid generation by the tubules is the first step in clearance of a water load and the homeostatic maintenance of extracellular osmolarity.

Chemicals and Reagents
All chemicals and reagents were from Sigma or Fisher unless otherwise specified.

Fly Stocks and Genetics
The following D. melanogaster strains were used: wBerlin (wild-type), obtained from Dr. Adrian Rothenfluh (University of Texas Southwestern Medical Center, Dallas, TX); w;Ncc69 r2 , obtained from Dr. William Leiserson (Yale University, New Haven, CT) (41); w;c42-GAL4, expressing GAL4 in the principal cells of the main and lower segments as well as bar-shaped cells in the initial and transitional segments (42), obtained from Dr. Julian Dow (University of Glasgow, Glasgow, UK) and outcrossed for five generations to wBerlin. w;UAS-WNK RNAi , an RNAi line targeting Drosophila wnk, was obtained from the Vienna Drosophila Resource Center (Vienna, Austria; line 106928) (43) and was outcrossed for five generations to wBerlin; upon outcrossing, two eye colors became apparent, one with a "gradient" eye in heterozygotes and one with a heterozygous orange eye. The gradient eye stock was used for all experiments. All transgenic fly lines generated (described below) were outcrossed for five generations to wBerlin.
Flies were reared at 28°C unless otherwise noted on cornmeal/yeast/molasses food prepared in a central kitchen at University of Texas Southwestern Medical Center. Female flies were collected within 48 h of eclosion and kept on standard food for 3-5 days before tubule dissection.

Cloning and Plasmid Construction
Primer sequences are listed in Table 1. wnk Cloning-RNA was isolated from whole wBerlin adult flies using TRIzol reagent (Invitrogen). cDNA was prepared using SuperScript II reverse transcriptase (Invitrogen) and an oligo(dT) 20 primer. Initial attempts to obtain a full-length wnk clone based on the annotated sequence were unsuccessful. Therefore, PCR was performed to clone three segments of the AGTTGTAGACGGAGGAGCCT open reading frame separately, which were then ligated together. All amplifications were performed with high-fidelity Pfx polymerase (Invitrogen). Initial attempts to clone the 5Ј-end, using primers W5837D and W9433U, revealed a misannotated splice site. A more extensive analysis of the cDNA was therefore undertaken. PCR was performed with primers W5024D and W6483U, revealing an alternatively spliced exon within the 5Ј-UTR as well as redefining the splicing events in the 5Ј-UTR and suggesting a "new" start codon. Based on these findings, the 5Ј-end was amplified with primers W6777D and W9433U. The second segment was amplified with W9291D and W15754-2U. The third segment was amplified with primers W13711D and W15754U. The three segments were individually cloned into pENTR (Invitrogen) using the Topo reaction (Invitrogen). pENTR containing segments two and three was digested with PciI and BglI (New England Biolabs, Ipswich, MA) and ligated together using T4 DNA ligase (New England Biolabs, Ipswich, MA). The resultant product, as well as pENTR containing segment one, was then digested with ApaLI (New England Biolabs, Ipswich, MA) and ligated together to yield the final full-length wnk cDNA in pENTR. The cDNA was sequenced in its entirety, and the sequence has been deposited in GenBank TM . Generation of WNK D420A -In order to mutate Asp-420 to alanine, base pairs 1258:1260 of the wnk open reading frame were mutated from GAC to GCC. The top strand mutation was introduced by performing PCR with primers W7655D and W8332U. The bottom strand mutation was introduced by performing PCR with primers W8307D and W9433U. Top and bottom strands were then "zippered" together by performing PCR with primers W7655D and W9433U. This fragment was then cloned into pENTR using the TOPO reaction (Invitrogen). All PCR was performed with Pfx polymerase (Invitrogen), and the resulting mutant cDNA sequence was confirmed by sequencing. pENTR plasmids containing full-length wild-type wnk and the mutant wnk fragment were cut with SpeI and BstBI (New England Biolabs, Ipswich, MA). The mutant wnk fragment was ligated into the full-length wnk backbone using T4 DNA ligase (New England Biolabs, Ipswich, MA).
Ncc69 Cloning-Clone GH27027 was obtained from the Drosophila Genomics Resource Center (Bloomington, IN). The open reading frame was amplified by PCR using Pfx polymerase (Invitrogen) and primers N16285D and N26955U, cloned into pENTR using the TOPO reaction (Invitrogen), and sequenced in its entirety. To generate the GST-tagged N terminus of Ncc69 (amino acids 1-204) for use in the in vitro kinase assay, primers Ncc69N204F and Ncc69N204R were used with Pfx polymerase (Invitrogen), and the amplified fragment was cloned into pENTR using the TOPO reaction (Invitrogen). The Ncc69 fragment was then recombined into pDEST15 using the clonase reaction (Invitrogen), and the sequence was confirmed.
fray Cloning-A full-length cDNA encoding fray (RE53265) was obtained from the Drosophila Genomics Resource Center (Bloomington, IN). The full-length open reading frame was obtained by PCR with Pfx polymerase (Invitrogen) and primers F1D and F1656U and cloned into pENTR using TOPO (Invitrogen). In order to introduce the T206E mutation, base pairs 616:618 were mutated from ACC to GAG. The top strand mutation was introduced by using primers F1D and F628U, and the bottom strand mutation was introduced by using primers F606D and F1656U. Top and bottom strands were then "zippered" together by performing PCR with primers F1D and F1656U, and the resulting PCR product was cloned into pENTR using the TOPO reaction (Invitrogen). All PCR was performed with Pfx polymerase (Invitrogen), and the resulting cDNA sequence was confirmed by sequencing. A similar strategy was used to introduce the D185A kinase-dead mutation into wild type or fray T206E , using primers F548D and F569U. To generate GST-tagged Fray for use in the in vitro kinase assay, wild-type and mutant fray were recombined into pDEST15 using the clonase reaction (Invitrogen).

Expression of GST-tagged Fusion Proteins
pDEST15 constructs were transformed into BL21(DE3) Escherichia coli. Protein expression and purification were undertaken as described in the GST Gene Fusion System Handbook of GE Healthcare. Briefly, cells were grown at 37°C in 2ϫ YT medium to an A 600 of 1-2 OD. Isopropyl-D-galactosidase was added to a final concentration of 0.1 mM to induce protein expression, and the cells were cultured for another 16 h at 20°C. Cells were harvested by centrifugation and lysed by sonication in PBS with SIGMAFAST TM protease inhibitor. GST-tagged proteins were purified from the lysates using GSTrap TM FF (GE Healthcare) and eluted from the column in 50 mM Tris-HCl, pH 8.0, with 10 mM glutathione. After gel electrophoresis, Coomassie Blue-stained proteins were compared with a BSA standard to determine approximate protein concentrations.

Generation of UAS-fray RNAi
Primers targeting the 3Ј-UTR of fray were designed using the DSIR program, siRNA 21 nucleotides. The top hit was chosen, and BLASTn was performed against the Drosophila genome, which showed 16-bp off-target hits to heterochromatin/U only and additional 14-bp off-target hits. The FRS2 and FRR2 primers were designed based on the TRiP protocol (45). These were then annealed and ligated into pWALIUM20 provided by the TRiP project (Harvard University, Boston, MA) after EcoRI/ NheI digestion (New England Biolabs, Ipswich, MA). DNA was injected into y 1 w 67c23 ; P{CaryP}attP2 (Bloomington stock no. 8622) by Rainbow Transgenic Flies, Inc. The presence of the insert was confirmed by genomic DNA PCR and sequencing.

Ramsay Assay and Ion-specific Electrodes
The Ramsay assay was set up as described previously (10,48). One tubule of the dissected pair remained in the droplet, whereas the other tubule of the pair was wrapped around a Minutien pin (Fine Science Tools, Foster City, CA) as an anchor. The secreted fluid droplet was examined at ϳ2 h, and its diameter was measured using an ocular micrometer in a dissecting stereomicroscope (Nikon, Melville, NY) at ϫ50 magnification. The volume of the droplet was calculated assuming spherical geometry (4/3 r 3 ). The secretion rate was calculated for each tubule by dividing volume/time.
Ion-specific and reference electrodes were prepared according to the method of Maddrell et al. (49) as described previously (10). Unfilamented borosilicate glass capillaries with an outside diameter of 1.2 mm (Harvard Apparatus, Holliston, MA) were washed for 5 min with nitric acid, rinsed 3-5 times with deionized water, and dried on a hot plate set to 200°C for a minimum of 20 min. Pipettes were pulled to a tip size of 1-2 m using a vertical pipette puller (Narishige, East Meadow, NY). They were then dried for at least 10 min on the hot plate and lightly silanized by application of a drop of dichlorodimethylsilane inside a 15-cm Pyrex dish, which was inverted over the pipettes on the hot plate for a minimum of 20 min. Silanized pipettes were stored over silica gel (Fisher) until use. For measuring K ϩ flux, backfill solution of 0.5 M KCl was added to the pipette, and a small amount of potassium ionophore I mixture B was aspirated into the tip of the pipette by application of negative pressure. The reference electrode was prepared from filamented borosilicate glass capillaries with an outside diameter of 1.2 mm (Harvard Apparatus), pulled in a manner similar to the ionspecific electrode. The tip and shank were filled with 1 M sodium acetate, and the electrode was backfilled with 3 M KCl. Selectivity of the potassium ion-specific electrode for K ϩ compared with Na ϩ is greater than 10 3.9 (50).
K ϩ activity was measured as described previously (10). Calibration drops consisting of 15, 75, 150, and 200 mM KCl were measured by immersing the reference and ion-specific electrodes into the fluid drop and recording the potential using a Digidata 1200 amplifier (Axon Instruments, Union City, CA) and a FD223a dual-channel electrometer (World Precision Instruments, Sarasota, FL). The ion-specific electrode was calibrated before and after each set of experimental measurements. Slope/decile change in K ϩ concentration was calculated using the Nernst equation for the difference between 15 and 150 mM, 75 and 150 mM, and 150 and 200 mM, and the average slope was then calculated. K ϩ activity in the experimental fluid was measured, and concentration was calculated according to Equation 1, where [K ϩ ] e is the potassium concentration of the experimental drop, [K ϩ ] c is the potassium concentration of the calibration drop, V is the change in potential (mV) between the experimental drop and the calibration drop, and s is the slope (mV) for a 10-fold change in K ϩ concentration, determined by measurements from the calibration drops. [K ϩ ] c was determined by the mean of the two 200 mM calibration drops (pre-and postexperiment). The K ϩ flux of each tubule was calculated by multiplying K ϩ concentration by the secretion rate for each tubule.

Statistics
Results comprising two groups were compared using an unpaired two-tailed t test. For results with three or more groups, one-way analysis of variance (ANOVA) was used. Bonferroni's test or Newman-Keuls test was used for post hoc testing of one-way ANOVA results. Two-tailed Fisher's exact test was used for categorical variables. Significance level was set at p Ͻ 0.05. All statistical analyses were performed using GraphPad Prism, version 5.0 or 6.0 (GraphPad Software, La Jolla, CA).

RESULTS
We have previously shown that the Drosophila NKCC, Ncc69, is required in the cation-conducting principal cell of the fly renal tubule for normal transepithelial K ϩ flux (10). Here we examined whether WNK-SPAK/OSR1 signaling regulates this NKCC. Drosophila wnk is an essential gene. We therefore used the GAL4-UAS system (51) to knock down wnk in the tubule principal cell using the GAL4 driver c42-GAL4 (42), allowing analysis of fluid secretion and K ϩ flux in the tubule of the adult fly using the Ramsay assay (48). Because the initial segment of the tubule is not fluid-secreting, and the lower segment of the tubule lies outside the bathing droplet in our experiments, this assay measures main segment fluid secretion and ion flux (48). The effect of wnk knockdown on fluid secretion rate, K ϩ concentration, and K ϩ flux is shown in Table 2, and the effect on K ϩ flux is illustrated in Fig. 1. wnk knockdown decreased K ϩ flux by about 30% (Fig. 1A), similar to the decrease in K ϩ flux seen in tubules homozygous for a null mutation in Ncc69 (10).
Mutating Asp-420 in WNK to Ala abolishes kinase activity (46). Overexpression of WNK D420A phenocopies wnk loss of function in flies (46), indicating that this kinase-dead variant has dominant-negative properties. We therefore expressed WNK D420A in the principal cell of the tubule and found that this decreased K ϩ flux to a similar extent as seen with wnk knockdown (Fig. 1B). Together, these data indicate that WNK positively regulates K ϩ flux in the Drosophila renal tubule.
To confirm that wnk transcripts were indeed decreased in the knockdown tubules, we performed quantitative RT-PCR. Compared with control tubules, wnk transcript levels in knockdown tubules were decreased by 63% (Fig. 1C). This probably underestimates the degree of wnk knockdown in the principal cells because c42-GAL4 is expressed in only a subset of the tubule cells examined.
Mammalian WNKs activate the Sterile 20 kinases SPAK and OSR1 by phosphorylating them (16 -18). Similarly, Sato and Shibuya (46) (41), we hypothesized that Fray also phosphorylates Ncc69. To demonstrate this, we performed in vitro kinase assays. We observed autophosphorylation of wild-type Fray, as has previously been observed for bacterially expressed SPAK and OSR1 (25,53), but wild-type Fray did not phosphorylate the N terminus of Ncc69 (Fig. 2). Similar to previous studies with bacterially purified SPAK and OSR1 (25,53), after bacterial expression and purification of full-length Fray, we observed both full-length and truncated Fray isoforms, both of which were autophosphorylated. Interestingly, we observed less autophosphorylation of wild-type Fray in the presence of Ncc69, suggesting that in the absence of activation by WNK, Fray may be non-productively binding Ncc69. Mutation of the T-loop threonine of SPAK and OSR1, a target of WNK phosphorylation that is required for SPAK/OSR1 activation, to a phospho-mimicking glutamic acid results in constitutive SPAK/OSR1 activity in vitro (26) and in vivo in C. elegans (54). We therefore introduced a similar mutation into Fray, T206E. Fray T206E phosphorylated the N terminus of Ncc69 in the absence of WNK (Fig. 2), indicating that this mutation renders Fray constitutively active and that Ncc69 is a target of Fray phosphorylation. No kinase activity, either autophosphorylation or Ncc69 phosphorylation, was seen when a mutation predicted to render the kinase inactive, D185A, was introduced into wild-type or constitutively active Fray (Fig. 2).
Like wnk, fray is an essential gene in the fly (55). We therefore knocked down fray in the principal cell of the tubule using the GAL4-UAS system. Again, we observed a decrease in K ϩ flux in the fray knockdown tubules ( Table 2 and Fig. 3A), similar to the phenotype seen with wnk knockdown. Quantitative RT-PCR showed a 51% decrease in whole-tubule fray transcript levels (Fig. 3B).
Expression of the constitutively active Fray T206E isoform in the principal cell of the tubule did not increase K ϩ flux (Fig. 3C), suggesting that increasing Fray activity is not sufficient to increase ion flux. A possible explanation for this is that mouse protein-25/calcium-binding protein 39 (Mo25/Cab39) is limit- All values shown are mean Ϯ S.E. "Slope" is the mean slope per 10-fold difference in ͓K ϩ ͔ of the electrodes used for each experiment. Note that K ϩ flux was calculated separately for each tubule analyzed, and therefore the value of mean secretion rate ϫ mean ͓K ϩ ͔ may differ slightly from measured K ϩ flux, due to rounding. a In this experiment, ͓K ϩ ͔ was measured in a subset of tubules; n for secretion is followed by n for ͓K ϩ ͔ and K ϩ flux.  The GAL4-UAS system, using the principal cell GAL4 driver c42-GAL4, was used to knock down wnk in the principal cells of the tubule or to express a dominant-negative WNK isoform. Fluid secretion rates were measured in isolated tubules using the Ramsay assay, and K ϩ concentration was determined using ion-specific electrodes, allowing calculation of K ϩ flux. A, K ϩ flux (pmol/min/tubule) was measured in control tubules (w;c42-GAL4/ϩ and w;UAS-WNK RNAi /ϩ) versus wnk principal cell knockdown tubules (w;UAS-WNK RNAi /ϩ; c42-GAL4/ϩ). n ϭ 38 -42 tubules/genotype. There was no significant difference in K ϩ flux between the two control genotypes. K ϩ flux was decreased in the wnk knockdown tubules compared with both controls. B, K ϩ flux (pmol/min/tubule) was measured in control tubules (w;c42-GAL4/ϩ and w;UAS-WNK D420A /ϩ) versus tubules expressing dominant-negative, kinase-dead WNK in the principal cells (w;UAS-WNK D420A /ϩ;c42-GAL4/ϩ). n ϭ 30 -31 tubules/genotype. There was no significant difference in K ϩ flux between the two control genotypes. K ϩ flux was decreased in the tubules expressing kinase-dead WNK compared with both controls. C, wnk transcript levels were measured by quantitative RT-PCR in control tubules (w;UAS-WNK RNAi /ϩ) versus wnk knockdown tubules (w;UAS-WNK RNAi /ϩ;c42-GAL4/ϩ), n ϭ 3 sets of tubules, 40 tubules/genotype in each set. All data are presented as the mean Ϯ S.E. (error bars) *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001.

Genotype
ing in wild-type tubules. Mo25/Cab39 has been shown to dramatically increase the in vitro kinase activity of SPAK/OSR1, including the SPAK and OSR1 constitutively active mutants analogous to Fray T206E (56), perhaps by facilitating domain swapping of SPAK/OSR1 dimers (57), and Fray and Mo25 also act together in Drosophila neuroblast asymmetric cell division (58). Nevertheless, expression of Fray T206E was sufficient to rescue the decreased K ϩ flux seen in wnk knockdown tubules back to control levels (Fig. 3D). This indicates that Fray acts downstream of WNK in the positive regulation of K ϩ flux and that WNK probably acts by phosphorylation of Fray on the T-loop threonine Thr-206.
Because Fray is acting downstream of WNK and Fray phosphorylates Ncc69, we predicted that the WNK-Fray kinase cascade is regulating K ϩ flux in the tubule through the regulation of the Ncc69 NKCC. To test this, we examined the effect of wnk knockdown in tubules homozygous for a null mutation in Ncc69, w;Ncc69 r2 /Ncc69 r2 (10,41). Unlike in the wild-type background, wnk knockdown in the Ncc69 r2 /Ncc69 r2 mutant background had no effect on K ϩ flux ( Table 2 and Fig. 4A), indicating that Ncc69 is downstream of WNK signaling. To verify that wnk transcripts were decreased in the Ncc69 mutant background, we performed quantitative RT-PCR. wnk transcript levels were decreased 48% in knockdown tubules compared with control (Fig. 4B), and the decrease in wnk transcript levels was not significantly different in the wild-type and Ncc69 r2 /Ncc69 r2 mutant backgrounds (p ϭ 0.5413). Similarly, knocking down fray in the principal cell of the tubule in the Ncc69 r2 /Ncc69 r2 mutant background also had no effect on K ϩ flux (Fig. 4C), indicating that Ncc69 is also downstream of Fray signaling.
failed to secrete fluid at all, again consistent with an inhibitory effect of hypertonicity on secretion; 29 of 56 (52%) w;c42-GAL4/ϩ tubules bathed in hypertonic medium failed to secrete, and 25 of 49 (51%) yw/w;UAS-fray RNAi /c42-GAL4 tubules failed to (p ϭ 1.000, Fisher's exact test). Fluid secretion rates increase when tonicity of bathing medium is decreased (5). We replicated this result in w;c42-GAL4/ϩ control tubules and showed that K ϩ flux is also increased ( Table 2 and Fig. 6A). However, the increase in K ϩ flux was not observed in wnk or fray knockdown tubules (Fig.  6A). Similarly, hypotonic bathing medium stimulated K ϩ flux in a second control strain, wBerlin, but not in w;Ncc69 r2 / Ncc69 r2 mutant tubules lacking the NKCC transporter (Fig.  6B). Thus, whereas Ncc69 mediates ϳ20 -30% of K ϩ flux under baseline conditions, under hypotonic conditions, Ncc69 mediates ϳ50% of K ϩ flux (Table 2 and Fig. 6B) (10). Similarly, the WNK-and Fray-sensitive components of K ϩ flux are increased under hypotonic conditions. These results indicate that hypotonic bathing medium stimulates K ϩ flux by activating the WNK/Fray/Ncc69 pathway. Lytle and Forbush (62) demonstrated in isolated shark rectal gland, a transporting epithelium, that both hypotonic and hypertonic conditions activate NKCC, and multiple studies have shown that in isolated cells, WNK-SPAK/OSR1 signaling is activated in hypertonic and hypotonic media (17, 23, 24, 53, 59 -61). Why would extracellular hypertonicity and hypotonicity, which cause opposite effects on cell volume (shrinkage and swelling, respectively), both activate NKCC activity? Activation of the WNK-SPAK/OSR1-NKCC pathway under hypertonic conditions increases intracellular Cl Ϫ ; the accompanying inward flow of osmotically obliged water restores cell volume after hypertonic shrinkage. What, then, is the purpose of hypotonic activation of the WNK-SPAK/OSR1-NKCC pathway? A clue to this comes from the same study of Lytle and Forbush (62), in which transepithelial tubule secretion only occurred under hypotonic and not under hypertonic conditions despite NKCC activation in both situations. Under hypertonic conditions, NKCC activation in the absence of net transepithelial transport would result in restoration of tubule epithelial cell volume. However, under isotonic or hypotonic conditions, NKCC activation on the basolateral membrane of this secreting epithelium would result in cell swelling if it were not coupled to apical ion efflux; thus, under hypotonic conditions, both basolateral NKCC and apical transport mechanisms are stimulated, resulting in net fluid secretion. Thus, it is likely that activation of NKCC by hypertonicity primarily defends cell volume (i.e. regulatory volume increase after cell shrinkage), whereas activation by hypotonicity is functionally designed to stimulate transepithelial solute and fluid fluxes. Here, we have directly proven for the first time that in a transporting epithelium, hypotonicity does result in increased transepithelial ion flux through the activation of the WNK-SPAK/OSR1-NKCC pathway.

DISCUSSION
The Malpighian tubules of insects are renal epithelia that play an essential role in iono-and osmoregulation (3,4). Because the fly tubules are bathed in hemolymph, they can directly sense changes in its osmolarity. Indeed, in isolated tubules, changes in bathing solution osmolarity directly regulate fluid secretion and ion flux (5,63). Here, we demonstrate an inverse relationship between tubule bathing osmolarity and K ϩ flux in the Drosophila renal tubule. We previously demonstrated a role for the fly NKCC, Ncc69, in transepithelial K ϩ flux (10). We therefore examined whether Ncc69 is required for the increased K ϩ flux seen in hypotonic conditions. We demonstrate that both Ncc69 and its kinase regulators, WNK and Fray, are required for the response to hypotonicity. Our results are consistent with a model in which WNK activates the SPAK/ OSR1 homolog Fray, which then phosphorylates the N terminus of Ncc69 to result in transporter activation, K ϩ uptake into the principal cell, and eventual K ϩ secretion across the apical membrane and into the urine (Fig. 6C). Hypotonicity activates the WNK-Fray-Ncc69 pathway because no stimulation of K ϩ flux is seen in tubules with decreased wnk, fray, or Ncc69. Increased KCl secretion drives an increase in osmotically obliged water flux, presumably through aquaporins expressed in the tubule (64), resulting in increased fluid secretion.
In contrast, as in the shark rectal gland, hypertonicity does not result in increased fluid secretion or transepithelial ion flux. In fact, K ϩ flux and fluid secretion decrease, and this occurs independently of the WNK-SPAK/OSR1-NKCC pathway. Blumenthal (5,65) has previously demonstrated that tyramine is a positive regulator of transepithelial fluid secretion, by increasing chloride flux through the stellate cells. Increased osmolarity decreases tubule sensitivity to tyramine, resulting in decreased fluid secretion (5,65). Our results suggest that this process occurs independently of WNK-SPAK/OSR1 signaling.
How is activation of the WNK-SPAK/OSR1 pathway occurring in tubules exposed to hypotonicity? The typical cellular response to hypotonicity is cell swelling, followed by a decrease in cellular volume (regulatory volume decrease), typically achieved by extrusion of KCl (66). Swelling followed by regulatory volume decrease in renal epithelial cells of the proximal tubule and cortical collecting duct was first demonstrated in the 1970s, and it was proposed to occur through loss of KCl (67). More recently, intracellular chloride concentrations were directly measured during cell swelling and regulatory volume decrease in the A6 Xenopus renal epithelial cell line. An immediate decrease in intracellular chloride occurred at the time of cell swelling due to dilution, followed by an ongoing decrease in intracellular chloride as cell volume decreased toward normal, suggesting that extrusion of intracellular chloride was serving as a mechanism of regulatory volume decrease (68). Similarly, measurement of intracellular chloride in the renal tubules of another insect, the grasshopper-like New Zealand alpine weta, demonstrated a decrease in intracellular chloride when the tubules were bathed in hypotonic saline as well as an increase in fluid secretion rate and K ϩ flux (63), as seen in our study of Drosophila tubules.
WNKs have been proposed to function as chloride sensors (69,70), and very recently, the structural basis for chloride inhibition of WNK autophosphorylation and activation has been described (71). The sequence of the chloride-binding site of mammalian WNK1 is perfectly conserved in Drosophila WNK. Thus, a decrease in intracellular chloride occurring during cell swelling and regulatory volume decrease could stimulate WNK through relief of chloride inhibition, followed by Fray and Ncc69 activation, stimulating an increase in transepithelial ion and water flux. Indeed, intracellular Cl Ϫ concentration in the   . Hypotonicity stimulates K ؉ flux by activating the WNK/Fray/NKCC pathway. A,K ϩ flux (pmol/min/tubule) was measured in control (w;c42-GAL4/ϩ), principal cell wnk knockdown (w;UAS-WNK RNAi /ϩ;c42-GAL4/ϩ), and principal cell fray knockdown (yw/w;UAS-fray RNAi /c42-GAL4) tubules in standard bathing medium (measured osmolality ϭ 335 mosM/kg) or hypotonic bathing medium (standard bathing medium diluted with water to a measured osmolality of 257 mosM/kg). n ϭ 32-38 tubules/genotype/condition. K ϩ flux was stimulated in control tubules under hypotonic conditions but was not stimulated in wnk or fray knockdown tubules. B, K ϩ flux (pmol/min/tubule) was measured in control (wBerlin) and Ncc69 homozygous null mutant (w;Ncc69 r2 /Ncc69 r2 ) tubules in standard bathing medium (osmolality ϭ 335 mosM/kg) or hypotonic bathing medium (osmolality ϭ 257 mosM/kg). Flies in this experiment were reared at room temperature, ϳ23°C. n ϭ 20 -22 tubules/genotype/condition. K ϩ flux was stimulated in control tubules under hypotonic conditions, but was not stimulated in Ncc69 null tubules. All data are presented as the mean Ϯ S.E. (error bars). *, p Ͻ 0.05; ****, p Ͻ 0.0001. C, model for NKCC regulation. Our previous data are consistent with a role for Ncc69 in K ϩ uptake from the hemolymph into the principal cell; additional K ϩ uptake pathways are indicated with question marks. K ϩ is then secreted across the apical membrane in exchange for H ϩ , which is extruded into the lumen by the apical vacuolar H ϩ -ATPase. The data presented here are consistent with a positive regulatory role for the WNK-SPAK/OSR1 (Fray) kinase cascade in NKCC (Ncc69) regulation. Hypotonicity stimulates K ϩ flux by activating the WNK-SPAK/OSR1-NKCC pathway.
fly tubule is 30 mM in isotonic conditions (72), slightly higher than the apparent median inhibitory concentration of 20 mM for Cl Ϫ inhibition of mammalian WNK1 autophosphorylation (71). This suggests that tubule WNK-Fray signaling is poised to respond to changes in intracellular Cl Ϫ . A similar mechanism probably explains the activation of sodium-coupled chloride cotransporters expressed in Xenopus oocytes. For example, Ponce-Coria et al. (73) observed activation of NKCC2 by WNK3-SPAK signaling by either bathing Xenopus oocytes in low Cl Ϫ hypotonic medium or by decreasing intracellular Cl Ϫ via coexpression of the potassium-chloride cotransporter, KCC2. Similarly, WNK3 powerfully stimulates both NCC and NKCC1 in Xenopus oocytes even under hypotonic conditions, in which cotransporter activity is otherwise suppressed (70,74,75).
Hypotonic cell swelling is thought to activate a volumeregulated anion channel (also called I Clswell ). Recently, LRRC8A has been identified as a key component of this channel in vertebrate cells (76,77), whereas two groups have shown that in Drosophila S2 cells, Bestrophin-1 encodes I Clswell (78 -80). Bestrophin-1 is robustly expressed in adult fly tubule (FlyAtlas database) (81), suggesting a possible mechanism for lowering of intracellular Cl Ϫ in response to hypotonicity, with subsequent activation of the WNK-Fray signaling cascade.
In the adult fly, fluid secretion by the main segment is followed by reabsorption of ϳ30% of the fluid in the lower segment (82). In addition, although it has not been studied in adult Drosophila, in other adult insects, including Dipterans, the hindgut and rectum are major sites of water and ion reabsorption regulating the final osmolarity of the excreta (3,83,84). Similarly, in Drosophila larvae, the anal papillae play a role in osmoregulation (85). Thus, following a hypotonic meal that results in decreased hemolymph osmolarity, water clearance could be achieved by first increasing the rate of fluid secretion by the main segment, as observed in our study; preferential reabsorption of ions over water in more "downstream" segments, such as the lower segment of the tubule, hindgut, and rectum, would then allow excretion of a water load without excess ion loss.
In summary, we have shown that the WNK-SPAK/OSR1 kinase cascade regulates the fly NKCC to modulate transepithelial ion flux in the Drosophila renal tubule. We provide the first direct demonstration that this pathway is required for the increased transepithelial ion flux that occurs upon hypotonic stimulation. The resulting increase in urine generation by the Malpighian tubule provides a means for clearance of a water load in order to maintain homeostasis of hemolymph osmolarity.