SLC26A9 is a Cl channel regulated by the WNK kinases

Authors


  • M. R. Dorwart and N. Shcheynikov contributed equally to this work.

Corresponding author S. Muallem: Department of Physiology, University of Texas Southwestern Medical Center at Dallas, Dallas, TX 75390-9040, USA. Email: shmuel.muallem@utsouthwestern.edu

Abstract

SLC26A9 is a member of the SLC26 family of anion transporters, which is expressed at high levels in airway and gastric surface epithelial cells. The transport properties and regulation of SLC26A9, and thus its physiological function, are not known. Here we report that SLC26A9 is a highly selective Cl channel with minimal OH/HCO3 permeability that is regulated by the WNK kinases. Expression in Xenopus oocytes and simultaneous measurement of membrane potential or current, intracellular pH (pHi) and intracellular Cl (Cli) revealed that expression of SLC26A9 resulted in a large Cl current. SLC26A9 displays a selectivity sequence of I > Br > NO3 > Cl > Glu, but it conducts Br > Cl > I > NO3 > Glu, with NO3 and I inhibiting the Cl conductance. Similarly, expression of SLC26A9 in HEK cells resulted in a large Cl current. Although detectable, OH and HCO3 fluxes in oocytes expressing SLC26A9 were very small. Moreover, HCO3 had no discernable effect on the Cl current, the reversal potential in the presence or absence of Clo and, importantly, HCO3 had no effect on Cl fluxes. These findings indicate that SLC26A9 is a Cl channel with minimal OH/HCO3 permeability. Co-expression of SLC26A9 with the WNK kinases WNK1, WNK3 or WNK4 inhibited SLC26A9 activity, and the inhibition was independent of WNK kinase activity. Immunolocalization in oocytes and cell surface biotinylation in HEK cells indicated that the WNK-mediated inhibition of SLC26A9 activity is caused by reduced SLC26A9 surface expression. Expression of SLC26A9 in the airway and the response of the WNKs to homeostatic stress raise the possibility that SLC26A9 serves to mediate the response of the airway to stress.

Cl absorption and HCO3 secretion are vital functions of epithelia. In most epithelia, like the airway (Wine, 2006), pancreas (Steward et al. 2005), salivary glands (Melvin et al. 2005), intestine (Binder et al. 2005) and the reproductive system (Chan et al. 2006) Cl absorption and HCO3 secretion are dependent on the function of the cystic fibrosis transmembrane conductance regulator (CFTR). Although CFTR functions as a Cl channel that can also conduct some HCO3 (Poulsen et al. 1994; Linsdell et al. 1997), recent work revealed that CFTR is regulated by extracellular Cl (Clo) (Shcheynikov et al. 2004; Wright et al. 2004) and is not likely to mediate most of the Cl absorption and HCO3 secretion by secretory epithelia (Wang et al. 2003).

HCO3-secreting epithelia also express coupled Cl–HCO3 exchangers at the luminal membrane. It is now clear that the luminal Cl–HCO3 exchangers are members of the SLC26 family of transporters (SLC26Ts). The family consists of 10 genes, most of which have several splice variants (Mount & Romero, 2004). Mutations in several members of the family are associated with human diseases such as diastrophic dysplasia (SLC26A2) (Superti-Furga et al. 1996), congenital Cl diarrhoea (SLC26A3) (Makela et al. 2002), Pendred syndrome (SLC26A4) (Everett et al. 1997) as well as non-syndromic forms of deafness (SLC26A5) (Liu et al. 2003). The Cl and HCO3 transport properties are specific to each member of the family. SLC26A1 and SLC26A2 function primarily as SO42− transporters (Mount & Romero, 2004), while SLC26A3 functions as a coupled 2Cl–1HCO3 exchanger (Ko et al. 2002; Shcheynikov et al. 2006). SLC26A4 transports I (Dossena et al. 2006) and also mediates Cl–HCO3 exchange (Wall, 2005), while SLC26A6 functions a coupled 1Cl–2HCO3 exchanger (Xie et al. 2002; Shcheynikov et al. 2006) and SLC26A7 functions as a pHi-regulated Cl channel (Kim et al. 2005).

Another member of the family that is expressed at high levels primarily in the airway (Lohi et al. 2002) and gastric surface epithelial cells (J. Xu et al. 2005) is SLC26A9. SLC26A9 was reported to function as a Cl–HCO3 exchanger that is inhibited by NH4+ (J. Xu et al. 2005). The mRNA level of SLC26A9 is up-regulated in the gastric mucosa of mice infected with Helicobacter pylori (Henriksnas et al. 2006). Accordingly, it was proposed that SLC26A9 in gastric surface epithelial cells mediates the HCO3 secretion needed for protection of the gastric mucosa against injury by gastric acid (J. Xu et al. 2005; Henriksnas et al. 2006). To fulfil this task SLC26A9 should be a potent HCO3 transporter and its HCO3 transport properties should be similar to the HCO3 transport properties of surface epithelial cells. This should include DIDS-sensitive Cl-dependent electroneutral and Cl-independent conductive pathways, both of which were described in gastric surface epithelia (Curci et al. 1994; Allen & Flemstrom, 2005).

A major regulator of epithelial Na+, K+ and Cl transport are the WNK (with no lysine) family of protein kinases, which were first discovered in a nested PCR screen to identify novel members of the MAP kinase family (Xu et al. 2000). The WNK name was given to these kinases because the highly conserved catalytic lysine in β strand 3 of the kinase domain is missing in the WNKs. However, it was later shown (Xu et al. 2002), and confirmed by the crystal structure of the WNK1 kinase domain (Min et al. 2004), that the catalytic lysine is contributed by β strand 2 of the kinase domain. The WNK family consists of four members that show about 45% overall identity with about 85% identity in the kinase domain (Gamba, 2005; Xie et al. 2006). To date, only WNK1, 3 and 4 have been cloned and their activity tested. The physiological role of the WNKs became evident with the discovery that mutations in WNK1 and WNK4 are linked to the familial hyperkalaemic hypertension disorder pseudohypoaldosteronism type II (PHAII) (Wilson et al. 2001). Several subsequent studies revealed that the WNKs regulate key transporters at the apical and basolateral membrane of selective renal tubular segments (reviewed in Gamba, 2005; Kahle et al. 2005; Subramanya et al. 2006; Xie et al. 2006). Thus, the WNKs regulate the Na+–Cl cotransporter NCC1 (Kahle et al. 2005; Subramanya et al. 2006), the K+–Cl cotransporters KCC1 and KCC2 (de Los Heros et al. 2006), the Na+–K+–2Cl cotransporters NKCC1 (Anselmo et al. 2006) and NKCC2 (Rinehart et al. 2005), the K+ channel ROMK (Lazrak et al. 2006; Leng et al. 2006) and the epithelial Na+ channel ENaC (Naray-Fejes-Toth et al. 2004; B. E. Xu et al. 2005). In addition, WNK4, but not WNK1, inhibits SLC26A6, but not SLC26A4 (Kahle et al. 2004). Invariably, the WNKs regulate the transporters by modulating their surface expression, rather than affecting their transport properties (see Gamba, 2005; Kahle et al. 2005; Subramanya et al. 2006; Xie et al. 2006).

Another prominent regulator of the SLC26Ts is CFTR. In a previous work we showed that CFTR regulates the activity of mouse slc26a3, human SLC26A4 and mouse slc26a6 (Ko et al. 2002). This regulation is facilitated by interaction of CFTR and the SLC26Ts with PDZ domain-containing scaffolding proteins and is mediated by interaction of the CFTR R-domain and the SLC26Ts STAS domain (Ko et al. 2004).

In the present work we show that SLC26A9 is a Cl channel with minimal OH/HCO3 permeability. The channel displays a selectivity order of I > Br > NO3 > Cl > Glu, but it conducts Br > Cl > I > NO3 > Glu, with NO3 and I inhibiting Cl conductance by SLC26A9. These properties do not support a role of SLC26A9 in HCO3 transport. We also show here that WNK1, WNK3 and WNK4 inhibit SLC26A9 expressed in Xenopus oocytes or HEK cells by reducing its surface expression. The potential significance of these findings to epithelial transport, cystic fibrosis and stress sensing is discussed.

Methods

Clones

The SLC26A9 cDNA was generously provided by Dr Juha Kere (Karolinska Institutet, Sweden), which was subcloned into the Sal I and Xho I restriction sites of the HA-pCMV vector (Clontech). Two SLC26A9 isoforms can be found in the Nucleotide database, and all the studies presented here have used variant 1 (gi:20336286). Upon sequencing the open reading frame a sequence variation in SLC26A9 (K147E) was found that has no functional significance (data not shown) and thus may represent polymorphism of SLC26A9. A double-tagged SLC26A9 construct was made by inserting a myc tag between T787 and L788 of the HA-tagged construct described above, thus preserving the PDZ ligand at the C-terminus. Full length Rattus norvegicus WNK1, Mus musculus WNK4 kinase constructs in the pCMV5 vector and Homo sapiens WNK3 in the pCMV6 vector were generous gifts from Dr Melanie Cobb (UT Southwestern Medical Center, Dallas, TX, USA), and are described in detail elsewhere (Xu et al. 2000, personal communication). The WNK1, WNK3 and WNK4 cDNAs were shuttled into the T7TS expression vector for cRNA production using standard methods. cRNA was produced using the mMESSAGE mMACHINE following the manufacturer's guidelines (Ambion). The WNK3(K159M) and WNK4(K183M) kinase-dead mutant constructs and the WNK3(1–410) kinase domain in the T7TS construct were generated using the QuickChange Site-Directed Mutagenesis Kit (Stratagene) following the manufacturer's guidelines.

Total and surface expression in HEK cells and Xenopus oocytes

Expression of SLC26A9 was assayed by Western blots and surface expression by biotinylation (HEK cells) or immunolocalization (oocytes).

Biotinylation assay

The biotinylation assay was similar to that previously described (Kim et al. 2006). In brief, HEK cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM) containing 10% fetal bovine serum (FBS). Transfections were performed with Lipofectamine Plus following the manufacturer guidelines, and the cells were used 36–48 h post transfection. Sulfo-NHE-ss-Biotin in PBS was added to cells transfected with empty vector, SLC26A9, or SLC26A9 and the WNK kinases, and the cells were incubated for 30 min at 0°C. Free biotin was quenched and washed with 1% bovine serum albumin (BSA) in PBS. The cells were used to prepare microsomes by homogenization in buffer containing (mm): 250 sucrose, 10 Hepes, 1 EDTA, 1 DTT, 0.2 PMSF. After centrifugation for 3 min at 1000 r.p.m. in an Eppendorf centrifuge, the supernatants were centrifuged for 20 min at 18 000 r.p.m. The microsomal pellet was dissolved in lysis buffer containing (mm): 20 Tris, pH 8.0, 137 NaCl, 5 NaEDTA, 5 NaEGTA, 10% glycerol, 0.5% Triton X-100, 0.2 PMSF, 50 NaF, 20 benzamidine, and kept on ice for 30 min. The lysates were cleared by centrifugation, and volume and protein were measured and equalized. Streptavidin beads in 300 μl lysis buffer were added and the mixtures were incubated overnight at 4°C. The beads were washed 5 times with lysis buffer and the bound proteins were released in sample buffer and analysed by Western blot.

Glycosylation assay

HEK cells were transfected with 1 μg of the SLC26A9 vector, and 1 μg of the appropriate WNK vector using 10 μg ml−1 of polyethylenimine (Polysciences). After 48 h the cells were lysed in RIPA buffer at 4°C. The lysates were cleared by centrifugation and 100 μl of the supernatants were incubated with 500 units of either Endoglycosidase H (EndoH) or PNGaseF (peptide-N-Glycosidase F; NEB) for 1 h at 30°C. The deglycosylation reactions were stopped by the addition of SDS-PAGE sample buffer.

Immunocytochemistry

Control and transfected oocytes were imbedded in OCT reagent, frozen in liquid N2 and 4 μm sections were cut and immobilized on poly l-lysine-coated cover slips. The sections were permeabilized by incubation with 0.5 ml of cold methanol for 10 min at −20°C. The non-specific sites were blocked with 5% goat serum, 1% BSA, and 0.1% gelatin in PBS (blocking medium) and the sections were incubated with 1 : 250 dilution of anti-STAS-domain antibodies (a gift from Phil Thomas, UT Southwestern Medical Center, Dallas, TX, USA) in blocking medium overnight at 4°C. The anti-STAS-domain antibodies were raised against recombinantly expressed and purified SLC26A3 STAS domain and were found to recognize several SLC26 transporters, including SLC26A9. After washing the bound antibodies were detected with anti-IgG tagged with fluorescein isothiocyanate (FITC). Images were captured with a Bio-Rad MRC 1024 confocal microscope.

Simultaneous measurement of current (or membrane potential), pHi and Cli in Xenopus oocytes

All procedures for maintaining the frogs and preparation of oocytes followed NIH guidelines and were approved by the Animal Care and Use Committee of UT Southwestern Medical Center. Oocytes were obtained by partial ovariectomy of female Xenopus laevis (Xenopus Express, Beverly Hills, FL, USA) that were sedated in a bath consisting of a 0.2% methanesulphonate salt of 3-aminobenzoic acid ethyl ester (Sigma) and placed on ice. The abdomen was opened and a small piece of ovary (∼1 cm2) was removed. The abdomen was then closed and sutured. The frogs were allowed to slowly recover from anaesthesia by washing with warm water while vital signs were monitored. When the frogs were fully recovered, they were transferred to the animal facility and kept in a separate tank for 24 h after surgery before returning to the home tank. Each frog was used only twice to obtain oocytes, once from each side, with an interval of 2 months before the second removal. Electrophysiological recordings were performed at room temperature with two-electrode voltage clamp or current clamp methods, as previously described (Shcheynikov et al. 2006). Briefly, SLC26A9 current was recorded with an OC-725C Oocyte Clamp Amplifier (Warner Instrument Corp.). Current and voltage were digitized at 500 Hz and then filtered at 100 Hz using a Digidata 1322A A/D converter and analysed using the Clampex 8.1 software. The current was measured by stepping the membrane potential from a holding potential of −30 mV to 0 mV, and then the membrane potentials were stepped between −100 and +60 mV at 10 mV steps for 200 ms with a 500 ms interval between steps.

Oocytes were injected with 50 nl water (controls) or 50 nl water containing between 4 and 5 ng of each of the cRNAs of interest and were used 3–5 days post injection. The oocytes were bathed in ND-96 solution or ND-96 solution in which the Cl was replaced with other anions as follows. The standard oocytes ND-96 solution contained (mm): 96 NaCl, 2 KCl, 1.8 CaCl2, 1 MgCl2, 2.5 pyruvate, 5 Hepes-Na, pH 7.5 (103.6 Cl) or a solution in which 25 mm NaCl was replaced with 25 mm sodium gluconate (78.6 Cl). The HCO3-buffered solution contained (mm): 71 NaCl, 25 NaHCO3, 2 KCl, 1.8 CaCl2, 1 MgCl2, 5 Hepes-Na, pH 7.5 (78.6 Cl, 25 HCO3). The Hepes-buffered, Cl-free media contained 96 sodium gluconate, 2 potassium gluconate, 1.8 calcium cyclamate, 1 MgSO4, 5 Hepes-Na, pH 7.5. The HCO3-buffered, Cl-free media contained 71 sodium gluconate, 25 NaHCO3, 2 potassium gluconate, 1.8 calcium cyclamate, 1 MgSO4, 5 Hepes-Na, pH 7.5. HCO3-buffered solutions were gassed with 5% CO2–95% O2. The Cl-free media were used to prepare media in which the major anion was NO3, Br or I by replacing the sodium gluconate with the respective Na+ salt so that the solution contained 96 mm Br, I or NO3. Current and voltage are digitized via an A/D converter and analysed by Clampex 8.1. For pHi measurement, the tip of the pH electrode was filled with 0.5 μl of a H+ exchanger resin and the electrodes were backfilled with ND-96 solution and calibrated in standard solutions of pH 6, 7 and 8. Calibration of the pH electrode was the same in solutions containing 110 or 10 mm Cl. The electrodes were fitted with a holder with an Ag–AgCl wire attached to a high-impedance probe of a two-channel electrometer. A second channel was used for the measurement of membrane potential and the bath was grounded via a 3 m KCl agar-bridge connected to an Ag–AgCl wire. The signal from the voltage electrode was subtracted from the voltage of the pH electrode to obtain pHi. Cli was measured with a Cl-sensitive liquid ion exchanger. The tips of the electrodes were filled with the Cl-selective ion exchanger and backfilled with 3 m KCl. The electrodes were calibrated in solutions containing 1, 3, 10, 30 and 100 mm Cl. The calibration was the same at pH 6.5 and 7.5. For simultaneous measurement of pHi and Cli, the oocytes were impaled with three electrodes. In this case, two ion-sensitive electrodes were connected to the FD-223 electrometer, and one reference microelectrode was used to record membrane potential with the OC-725C amplifier. HCO3 and Cl fluxes were determined from the first derivative of the slopes of pHi and Cli changes in response to removal of Clo.

Current measurement in HEK cells

The whole-cell Cl current was measured at room temperature in control and HEK cells transfected with SLC26A9 and the WNKs as previously described (Ko et al. 2004). The Cl current was isolated using a pipette solution containing (mm): 140 N-methyl-d-glucamine (NMDG)-Cl, 1 MgCl2, 2 EGTA, 0.5 ATP, and 10 Hepes (pH 7.3 with Tris), and a bath solution composed of (mm): 145 NMDG-Cl, 1 MgCl2, 1 CaCl2, 10 Hepes (pH 7.4 with NaOH), and 10 glucose. Cl-free solutions were prepared by replacing Cl with gluconate and using MgSO4. The current was recorded with an Axopatch 200B patch-clamp amplifier and was filtered at 2 kHz and digitized at 1 kHz. The membrane conductance was probed by stepping the membrane potential from a holding potential of 0 mV to membrane potentials between −100 and +100 mV at 20 mV steps. Pipettes had resistances between 5 and 7 MΩ when filled with an intracellular solution. The seal resistance was higher than 8 GΩ. Current recording and analysis were performed with the Clampex 8.1 software.

Results

SLC26A9 is a Cl channel

To study the properties of SLC26A9 we first expressed the protein in Xenopus oocytes, in which it is possible to simultaneously measure membrane potential or membrane current, pHi and Cli (Shcheynikov et al. 2006). Expression of SLC26A9 in the oocytes resulted in a large Cl current. Figure 1A shows the instantaneous current plots and Fig. 1B shows the I–V relationships recorded from oocytes incubated in Hepes- or HCO3-buffered media and in the presence and absence of extracellular Cl (Clo). The SLC26A9 current does not show any time- or voltage-dependent regulation in the presence or absence of HCO3 (Fig. 1A). In Hepes-buffered media the current averaged 6.2 ± 0.5 μA (mean ±s.e.m.) at +60 mV with a reversal potential of −23 ± 1 mV (n= 28). Cli in oocytes expressing SLC26A9 was measured to be 27.4 ± 2.7 mm (n= 14), which results in a calculated reversal potential of −27 ± 2.6 mV that is in good agreement with the measured reversal potential. Replacing Clo with gluconate (Glu) shifted the reversal potential to +50.6 ± 1.5 mV. Incubating the same oocytes in HCO3-buffered media had no measurable effect on the size of the current or the reversal potential in the presence and absence of Clo (Fig. 1B), providing the first evidence that SLC26A9 primarily conducts Cl and has minimal HCO3 permeability.

Figure 1.

SLC26A9 conducts Cl
A, instantaneous current recorded in oocytes bathed in Hepes- or HCO3-buffered media and in the presence or absence of Cl. B, I–V relationships of current measured in control oocytes (•) or Xenopus oocytes expressing SLC26A9 when incubated in Hepes- (○, ▵) or HCO3-buffered media (□, ∇) in which the major anion is Cl (○, □) or gluconate (0 Cl, ▵, ∇). Representative traces are shown. The mean ±s.e.m. of 28 similar experiments is given in the text.

To further characterize the SLC26A9-mediated current we measured its selectivity to different anions. Figure 2A and the summary in Fig. 2C shows that SLC26A9 conducts Br slightly better than Cl, with replacement of Clo with Bro increasing the outward current by 1.12 (±0.02)-fold (n= 9, P < 0.05). On the other hand, both I and NO3 reduced the current, indicating that SLC26A9 anion conductance followed the rank order Br > Cl > I > NO3 > Glu. Significantly, I and NO3 reduced both the outward (their own) and inward (Cl) currents. This is illustrated in Fig. 2B for I, which also shows that inhibition of the Cl current by I is fast, and completely but slowly reverses upon replacing the I with Clo.

Figure 2.

SLC26A9-mediated anion currents
A, current measured in Xenopus oocytes expressing SLC26A9 and incubated in Hepes-buffered media containing either 103.6 mm Cl (○) or 96 mm of the following anions: NO3 (◆), Br (□), I (▪), gluconate (Glu, •). Means from 9–14 oocytes are shown for the current recorded in the presence of Cl, Glu and NO3, while for clarity single traces are shown for the current recorded with Br and I. The oocyte was incubated for 10 min in Cl-containing media after exposure of the oocyte to the ion of interest for 3 min to allow recovery of Cli between solution changes. B, time course of outward and inward current inhibition by I. Current was measured by holding the membrane potential at −30 mV and every 5 s stepping to +60 and −60 mV for 200 ms. The oocytes were perfused with ND-96 containing 103.6 mm Cl and then with a solution in which Cl was replaced with I. C, current measured at +60 mV with each of the anions relative to the Cl current measured in the same oocytes using the protocol in A. For all anions the current was statistically different from that measured with Cl at P < 0.05 or better by one-way ANOVA. The mean ±s.e.m. of 9 similar experiments is given in the text.

Figure 2A shows that the anions also affected the reversal potential. This is illustrated in more detail in Fig. 3. Figure 3 also compares the selectivity of SLC26A9 and SLC26A7 for the anions. Figure 3A and B shows that while replacing Clo with Glu strongly depolarized the cells, replacing Clo with all other anions hyperpolarized the membrane potential of cells expressing SLC26A9. This suggests that the selectivity of SLC26A9 for the tested anions follows the rank order I > Br > NO3 > Cl > Glu and the reduced current observed with NO3 and I (Fig. 2) is due to reduced conductance of the channel for these anions. By contrast, SLC26A7 displays higher selectivity and conductance for NO3 over Cl. Figure 3C and D shows that replacing Clo with NO3 hyperpolarized oocytes expressing SLC26A7 and markedly increased the SLC26A7-mediated current.

Figure 3.

Anionic selectivity of SLC26A9 and SLC26A7
A, representative trace, and B, reversal potentials obtained with the indicated anions measured with oocytes expressing SLC26A9. The standard ND-96 solution contained 103.6 mm Cl and it was replaced by Cl-free solution containing 96 mm NO3, Br or I, as indicated by the bars. Note that the oocytes were incubated with solution ND-96 between solution changes. The numbers in the bars are the number of experiments performed at each condition. C, similar measurement in oocytes expressing SLC26A7, illustrating the different anionic selectivity of the two Cl channels. Although NO3 hyperpolarizes oocytes expressing SLC26A7 and SLC26A9, D shows the reduced NO3 current by SLC26A9 (•) and increased NO3 current by SLC26A7 (○) relative to Cl. Current was measured at a holding potential of +60 mV.

Sensitivity to Cl transport inhibitors

Useful information for a Cl channel is its sensitivity to common Cl channel inhibitors that can then be used to probe for the potential physiological role of the channel. Figure 4 shows that the SLC26A9-mediated Cl current is not inhibited by 100 μm diphenylamine-2-carboxylic acid (DPC), 100 μm 5-nitro-2′-(3-phenylpropylamino)-benzoate (NPPB) or 250 μm glibenclamide (n= 3 for each). SLC26A9 current is only partially inhibited by DIDS, with 0.1 mm DIDS inhibiting the current by 36 ± 5% with no further inhibition by increasing DIDS to 1 mm. This was not because of instability of the DIDS since DIDS was freshly prepared on the day of each experiment in DMSO and was diluted at least 200-fold into the perfusion media. On the other hand, both flufenamic and niflumic acids almost completely inhibited the SLC26A9-mediated Cl current with similar half-maximal inhibition at about 0.4 mm.

Figure 4.

Inhibitor sensitivity of Cl current by SLC26A9
A, representative I–V relations of Cl current in oocytes expressing SLC26A9 in the absence (•) and presence of 250 μm DPC (○), 0.1 mm DIDS (▵), 100 μm NPPB (□) or 250 μm glibenclamide (⋄). B, effect of 0.01–1 mm DIDS on SLC26A9 current. C, representative I–V relations of Cl current in oocytes expressing SLC26A9 in the absence (•) and presence of 0.5 (□) or 2.5 mm (▪) flufenamic acid (Flu) or of 0.5 (▵) or 2.5 mm (▴) niflumic acid (Nif);D, dose–response to niflumic and flufenamic acids. The results in B and D are the mean ±s.e.m. of 4 experiments.

SLC26A9 is a poor HCO3 transporter

To determine the Cl and HCO3 transport capacity of SLC26A9 we measured the effect of HCO3 on Cl current, membrane potential, Cli and pHi in the same cells. The current measurements in Fig. 1 show that HCO3 had no measurable effect on the current and the reversal potential in cells bathed in Cl-containing or Cl-free media. Hence, the HCO3/Cl permeability ratio of SLC26A9 is very low and could not be discerned by current measurements.

In a second protocol we measured the effect of HCO3 on the changes in the membrane potential, pHi and Cli in response to removal of Clo. Figure 5A and C shows that HCO3 had a minimal effect of the resting membrane potential and on the depolarization caused by removal of Clo. However, albeit very low, SLC26A9 does have a measurable HCO3 permeability, as revealed by the higher rate of HCO3 influx into oocytes expressing SLC26A9 than into control oocytes. This can be seen best in the inset in Fig. 5B with the expanded pH and time scales. SLC26A9-mediated HCO3 efflux occurred at a rate of 0.33 ± 0.04 mm min−1 (n= 5). However, the rate of Cl efflux observed on Clo removal was not affected by HCO3.

Figure 5.

Cl and HCO3 fluxes by SLC26A9
In A and B the membrane potential, pHi and Cli were measured simultaneously. In A, oocytes expressing SLC26A9 were incubated in Hepes-buffered ND-96 media containing 103.6 mm Cl and perfused with Hepes-buffered Cl-free media containing Glu (first 0 Cl, black bar), with HCO3-buffered media containing 25 mm HCO3 and 78 mm Cl (grey bar) and with HCO3-buffered Cl-free media in which Glu replaced Cl (second 0 Cl, black bar). Note that HCO3 had no measurable effect on Cl efflux. In B, control oocytes (grey traces) and oocytes expressing SLC26A9 (black traces) bathed in HCO3-buffered media were exposed to Cl-free medium for the indicated period. The inset in the pHi panel shows the changes in pHi in expanded time and pH scales. The columns in C show the mean ±s.e.m. of 4–7 experiments for the indicated conditions.

SLC26A9 is regulated by the WNK kinases

The WNK kinases (WNKs) regulate many Na+, K+ and Cl transporters (Gamba, 2005; Kahle et al. 2005; Subramanya et al. 2006; Xie et al. 2006), including slc26a6 (Kahle et al. 2004), which mediate fluid and electrolyte homeostasis. Therefore, it was of interest to determine whether the WNKs regulate SLC26A9. Oocytes were injected with 4 ng of SLC26A9 cRNA alone or together with 5 ng of either WNK1, WNK3 or WNK4. Figure 6AC demonstrates that the WNKs inhibit SLC26A9-mediated Cl current. In addition, we tested the effect of kinase-deficient WNK3 and WNK4 mutants as well as of the WNK3 kinase domain to determine if the kinase activity of the WNKs is required for the regulation of SLC26A9. The kinase-dead WNK3(K159M) and WNK4(K183) inhibited SLC26A9 to an extent similar to the wild-type WNKs. Moreover, the kinase domain of WNK3, WNK3(1–410), has no effect on SLC26A9 activity. To monitor the cell surface expression of SLC26A9, immunolocalization experiments were performed in oocytes in the presence and absence of the WNKs. Figure 6D shows that WNK1, WNK3 and WNK4 all reduce the cell surface expression of SLC26A9.

Figure 6.

Inhibition of SLC26A9 activity by WNK1, WNK3 and WNK4 expressed in Xenopus oocytes
A, I–V relationship of oocytes expressing SLC26A9 alone or SLC26A9 and WNK3, SLC26A9 and the kinase-dead WNK3(K159M) or SLC26A9 and the kinase domain of WNK3 (WNK3(1–410)). B, summary of results of the indicated number of experiments similar to those in A. In C, oocytes expressing SLC26A9 alone, SLC26A9 and WNK1, SLC26A9 and WNK4 or SLC26A9 and the kinase-dead mutant of WNK4 (WNK4(K183M)) were used to measure the SLC26A9 Cl current. In D, slices were cut from a row of 3–4 control oocytes or oocytes expressing SLC26A9 and the indicated WNKs and imbedded in O.C.T. compound. The slices were stained with anti-STAS-domain antibodies prepared against purified STAS domain.

To extend the findings of Fig. 6 to mammalian cells, we measured the SLC26A9-mediated Cl current in HEK 293 cells. Figure 7A shows that SLC26A9 expressed in HEK 293 cells generated a large Cl current. Figure 7B shows that the properties of the instantaneous current were the same as those found in Xenopus oocytes (Fig. 1). Moreover, Fig. 7A and C shows that WNK1 and WNK4 inhibited SLC26A9 current in HEK 293 cells.

Figure 7.

The WNK kinases inhibit current and surface expression of SLC26A9 in HEK cells
A, whole-cell current was measured in HEK cells with pipette and bath solutions containing 150 mm Cl (○) or bath solution containing 150 mm Glu (•). The cells were transfected with empty vector (◆), SLC26A9 alone (○, •), SLC26A9 and WNK1 (□) or SLC26A9 and WNK4 (▵). Representative I–V relationships are shown in this panel. For clarity, the I–V relationships with the WNKs are from single experiments with the means summarized in C. B shows the instantaneous SLC26A9 current in the presence of 150 mm bath Cl or 150 mm Glu for cells expressing SLC26A9 (top traces) or SLC26A9 + WNK1, and SLC26A9 + WNK4 (bottom traces). C, summary of the results from the indicated number of experiments obtained with HEK cells expressing SLC26A9 alone (open column), SLC26A9 and WNK1 (grey column), or SLC26A9 and WNK4 (black column). In D, HEK cells expressing SLC26A9 and empty vector, SLC26A9 and WNK1, or SLC26A9 and WNK4, were used to prepare extracts. The extracts were treated with EndoH or PNGaseF to analyse the glycosylation state of SLC26A9. In E, doubly tagged HA-SLC26A9-myc was expressed in HEK cells to analyse the expression and glycosylation pattern by probing the blots for the N- (anti-HA) or C-termini (anti-myc) of SLC26A9. In F, HEK cells transfected with empty vector, HA-SLC26A9, HA-SLC26A9 + WNK1, or HA-SLC26A9 + WNK4, were used to analyse the effects of the WNKs on total (input) and surface expression (biotinylated) of SLC26A9. G, summary of the percentage reduction in surface expression of SLC26A9 by the WNKs. The results are the mean ±s.e.m. of 3 (WNK1) and 4 (WNK4) experiments.

Expression of SLC26A9 N-terminally tagged with HA (HA-SLC26A9) in HEK cells invariably (n= 11) resulted in two protein products with the lower band corresponding to the SLC26A9 predicted molecular weight of 87 kDa (Fig. 7D). To determine the glycosylation state of the two products we treated HEK 293 lysates containing SLC26A9 with EndoH or PNGaseF. Surprisingly, EndoH had no effect on the mobility of either of the two bands. Notably, treatment with PNGaseF increased the mobility of both bands, suggesting that both bands represent mature forms of SLC26A9. We considered the possibility that the two bands were the result of a proteolytic cleavage event. To test this possibility, the expression pattern of HA-SLC26A9-myc, which is tagged at both termini, was analysed. Figure 7E shows the same expression and glycosylation pattern when the blot is probed with anti-HA or with anti-myc, indicating that both bands represent full length proteins.

Figure 7DF shows that coexpression of SLC26A9 with WNK1 and WNK4 had no effect on total expression or the maturation of SLC26A9. This suggests that inhibition of SLC26A9 activity by the WNKs is not due to reduced maturation of SLC26A9, but rather the WNKs affect SLC26A9 after trafficking out of the Golgi. We analysed the effect of the WNKs on surface expression of SLC26A9 by performing biotinylation assays. Figure 7F shows the results of a biotinylation assay representative of three (WNK1) and four (WNK4) other experiments. In all experiments the faster 87 kDa migrating SLC26A9 form is the predominant one biotinylated, suggesting that it is the species found in the plasma membrane and functioning as a Cl channel. WNK1 and WNK4 markedly reduced surface expression of SLC26A9.

Discussion

In the present work we found that SLC26A9 is a Cl channel regulated by the WNK kinases. Previous work used pHi measurements with the pH-sensitive fluorescent dye BCECF to conclude that SLC26A9 is a DIDS-sensitive Cl–HCO3 exchanger (J. Xu et al. 2005). The only evidence for Cl–HCO3 exchange activity was an increase in pHi in response to removal of Clo. However, the mode of transport responsible for this activity was not determined. In the present work, we measured membrane potential, current, pHi and Cli in Xenopus oocytes and current in HEK cells. Expression of SLC26A9 in the two cell types resulted in a large Cl current. Importantly, HCO3 had no effect on the reversal potential or the Cl current mediated by SLC26A9. Moreover, HCO3 had no effect on the SLC26A9-mediated Cl influx or efflux. In comparison, slc26a3 and slc26a6 mediate large HCO3 fluxes, HCO3 shifted the reversal potential according to the stoichiometry of the exchange and HCO3 markedly increased the Cl fluxes (Shcheynikov et al. 2006). These findings lead us to conclude that SLC26A9 does not function as a Cl–HCO3 exchanger or significantly contribute to epithelial HCO3 secretion. Rather, SLC26A9 functions as a Cl channel at the luminal membrane of airway secretory glands and gastric surface epithelial cells. This conclusion also means that SLC26A9 does not mediate HCO3 secretion by surface gastric epithelial cells as suggested before (J. Xu et al. 2005), but more likely controls the Cl permeability of the luminal membrane of these cells. This conclusion is further supported by the profile of the blockers' sensitivity to SLC26A9 (Fig. 4). We note that HCO3 secretion by gastric surface epithelial cells is electroneutral and is nearly abolished by 0.5 mm DIDS (Allen & Flemstrom, 2005), whereas 0.5 mm DIDS inhibited SLC26A9 activity by only 36%.

Measurement of anion current revealed a difference in the selectivity (I > Br > NO3 > Cl > Glu) and permeability (Br > Cl > I > NO3 > Glu) sequence of SLC26A9 to anions, with NO3 and I inhibiting the Cl conductance. The anomaly of a higher selectivity for NO3 but lower conductance is specific for SLC26A9. Slc26a3 (Shcheynikov et al. 2006) and SLC26A7 (Fig. 3 and Kim et al. 2005) display higher selectivity and conductance for NO3 over Cl. Figure 3 shows that replacing Clo with NO3 hyperpolarized the oocytes and markedly increased the SLC26A7-mediated current. This is in contrast to the observation with oocytes expressing SLC26A9, in which replacing Cl with NO3 resulted in lower outward current (NO3 influx), although it hyperpolarized the oocytes. We interpret these findings to indicate that NO3 can partially block SLC26A9 conductance. That is, the differential selectivity and conductance of I and NO3 by SLC26A9 may suggest that the channel pore accommodates these anions better than Cl and the anions linger within the pore longer than Cl. This interpretation is supported by the finding that both I and NO3 inhibit the Cl current and recovery of the Cl outward current from an I block is slow (Fig. 2).

SLC26A9 activity is prominently regulated by the WNK kinases (WNKs). The WNKs regulate many Na+, K+ and Cl transporters involved in fluid and electrolyte homeostasis (Gamba, 2005; Kahle et al. 2005; Subramanya et al. 2006; Xie et al. 2006), including slc26a6 (Kahle et al. 2004). The WNKs can regulate ion transporters directly by phosphorylating the transporters (for example, regulation of NKCC2 and KCC by WNK3; Rinehart et al. 2005), by phosphorylating and activating other kinases (for example, activation of ORS1 by WNK1; Anselmo et al. 2006), or by acting as a platforms or scaffolds for other proteins which may be involved in endocytosis (Yang et al. 2005). In the latter case, the platform function of the WNKs is independent of their kinase activity and is mediated by the C-terminal non-kinase domain of the WNKs (Gamba, 2005; Kahle et al. 2005; Subramanya et al. 2006; Xie et al. 2006). The present work showed that WNK1, WNK3 and WNK4 regulate SLC26A9 activity and that the regulation does not require the kinase activity of the WNKs. These findings suggest that the WNKs act as scaffolds that recruit other proteins that regulate the activity of SLC26A9, similar to the role of WNK4 in the regulation of NCC (Yang et al. 2005) and the role of WNK1 in the regulation of the ROMK K+ channel (Lazrak et al. 2006; He et al. 2007).

Regulation by the WNKs has been shown to invariably involve surface expression of the transporters (Gamba, 2005; Kahle et al. 2005; Subramanya et al. 2006; Xie et al. 2006; He et al. 2007). The vast majority of the studies describing the effects of the WNKs on the activity and surface expression of the transporters have been performed in Xenopus oocytes (Gamba, 2005; Kahle et al. 2005; Subramanya et al. 2006; Xie et al. 2006). In fact, only one study has shown an effect of a WNK in mammalian cells, a study that showed WNK4 regulated the surface expression of the NaCl cotransporter NCCT (Cai et al. 2006). It is important to demonstrate that the observed effect of the WNKs in oocytes is similar to the effect of the WNKs in mammalian cells. Moreover, protein trafficking and surface expression are markedly affected by temperature, and oocytes are maintained at 18°C. Therefore, we sought to determine if SLC26A9 functions as a Cl channel when expressed in mammalian cells. Our results indicated that SLC26A9 is a Cl channel in mammalian cells, and analysis of surface expression in HEK cells showed that WNK1 and WNK4 reduced the surface expression of the transporter.

Another surprising finding in the present work is that SLC26A9 is expressed as two mature protein products with the lower 87 kDa band corresponding to the predicted molecular weight. Only the 87 kDa band was targeted to the plasma membrane. N- and C-terminal tagging and glycosylation analysis excluded the possibility that the 87 kDa protein product arises from a proteolytic cleavage of the longer 115 kDa SLC26A9 form. Expression of SLC26A9 as two protein products with only one of them trafficking to the cell surface raises the possibility that the 115 kDa form represents a post translational modification of SLC26A9 that causes it to be retained intracellularly. Processing of the 115 kDa form may be required for its insertion into the plasma membrane where it can then function as a Cl channel. Alternatively, since generation of the two forms does not appear to involve cleavage, and both represent mature forms of the protein, it is equally possible that the two forms are stable SLC26A9 species that are targeted to different cellular compartments depending on their processing. Further work is needed to examine these possibilities.

The physiological function of SLC26A9 is not obvious from its tissue distribution or functional properties, in particular in the airway where CFTR is the dominant Cl channel at the luminal membrane that regulates the overall secretory process (Ballard & Inglis, 2004). However, the same uncertainty exists with respect to other luminal Cl channels, such as the Ca2+-activated Cl channel and the outward rectifying Cl channel. Nevertheless, a unique feature of the SLC26 transporters is that they are regulated by CFTR and they also regulate CFTR activity (Ko et al. 2002, 2004). The mutual regulation is mediated by the CFTR R domain and the SLC26 transporters STAS domain (Ko et al. 2004). Since SLC26A9 is expressed in the luminal membrane of airway secretory cells (Lohi et al. 2002), it is possible that SLC26A9 regulates CFTR activity in the luminal membrane. Moreover, SLC26A9 can function as CFTR-regulated Cl channel which could mediate a substantial portion of the Cl conductance of the luminal membrane in the stimulated cells. Indeed, the author's preliminary unpublished observations in which SLC26A9 activity was measured as Cl-dependent OH fluxes indicate that SLC26A9 is activated by CFTR (data not shown).

Regulation by the WNKs may have a general function in epithelia. The present work shows regulation of SLC26A9 by the WNKs. A recent study reported regulation of CFTR activity by WNK1 and WNK4 (Yang et al. 2007). The WNKs reduce the activity of both channels. It is possible that the WNKs are active in the resting state to prevent unnecessary fluid secretion. Another possibility is that the WNKs may mediate the action of inhibitory agonists such as substance P (Ashton et al. 1990) and AVP (Ko et al. 1999) acting on the pancreatic duct. Although only minimal evidence exists, the activity of the WNKs is regulated by receptor-mediated cell stimulation (reviewed in Subramanya et al. 2006; Xie et al. 2006).

Another possible role for the regulation of SLC26A9 by the WNKs is that it may mediate the response of the airway and surface gastric epithelial cells to stress and insult. Thus, WNK1, and perhaps other WNKs, exists in complex with the oxidative stress-responsive kinase and both kinases are required for regulation of NKCC1 (Anselmo et al. 2006). Indeed, regulation of SLC26A9 by the WNKs is independent of their kinase activity. This indicates that the WNKs function as scaffolds to recruit other proteins, perhaps other kinases, which regulate surface expression of SLC26A9. Such regulation is observed in response to cellular stress, including cell swelling and cell shrinkage. These findings suggest that the WNKs are sensors of cell stress and tonicity. Therefore, the WNKs may transmit the response of the airway and gastric gland to stress or insult by activation of SLC26A9.

Appendix

Acknowledgements

This work was supported National Institute of Dental and Craniofacial Research (NIDCR) grants DE12309 and DK38938, NIH grant GM53032, a Cystic Fibrosis Foundation grant MUALLE05G0, and by the Ruth S. Harrell Professorship in Medical Research.

Ancillary