Novel Role for Pendrin in Orchestrating Bicarbonate Secretion in Cystic Fibrosis Transmembrane Conductance Regulator (CFTR)-expressing Airway Serous Cells*

In most HCO3−-secreting epithelial tissues, SLC26 Cl−/HCO3− transporters work in concert with the cystic fibrosis transmembrane conductance regulator (CFTR) to regulate the magnitude and composition of the secreted fluid, a process that is vital for normal tissue function. By contrast, CFTR is regarded as the only exit pathway for HCO3− in the airways. Here we show that Cl−/HCO3− anion exchange makes a major contribution to transcellular HCO3− transport in airway serous cells. Real-time measurement of intracellular pH from polarized cultures of human Calu-3 cells demonstrated cAMP/PKA-activated Cl−-dependent HCO3− transport across the luminal membrane via CFTR-dependent coupled Cl−/HCO3− anion exchange. The pharmacological and functional profile of the luminal anion exchanger was consistent with SLC26A4 (pendrin), which was shown to be expressed by quantitative RT-PCR, Western blot, and immunofluorescence. Pendrin-mediated anion exchange activity was confirmed by shRNA pendrin knockdown (KD), which markedly reduced cAMP-activated Cl−/HCO3− exchange. To establish the relative roles of CFTR and pendrin in net HCO3− secretion, transepithelial liquid secretion rate and liquid pH were measured in wild type, pendrin KD, and CFTR KD cells. cAMP/PKA increased the rate and pH of the secreted fluid. Inhibiting CFTR reduced the rate of liquid secretion but not the pH, whereas decreasing pendrin activity lowered pH with little effect on volume. These results establish that CFTR predominately controls the rate of liquid secretion, whereas pendrin regulates the composition of the secreted fluid and identifies a critical role for this anion exchanger in transcellular HCO3− secretion in airway serous cells.

pharmacological and functional profile of the luminal anion exchanger was consistent with SLC26A4 (pendrin), which was shown to be expressed by quantitative RT-PCR, Western blot, and immunofluorescence. Pendrin-mediated anion exchange activity was confirmed by shRNA pendrin knockdown (KD), which markedly reduced cAMP-activated Cl ؊ /HCO 3 ؊ exchange.
To establish the relative roles of CFTR and pendrin in net HCO 3 ؊ secretion, transepithelial liquid secretion rate and liquid pH were measured in wild type, pendrin KD, and CFTR KD cells. cAMP/PKA increased the rate and pH of the secreted fluid. Inhibiting CFTR reduced the rate of liquid secretion but not the pH, whereas decreasing pendrin activity lowered pH with little effect on volume. These results establish that CFTR predominately controls the rate of liquid secretion, whereas pendrin regulates the composition of the secreted fluid and identifies a critical role for this anion exchanger in transcellular HCO 3 ؊ secretion in airway serous cells.

HCO 3
Ϫ is a vital component of epithelial secretions. Despite the growing awareness of its importance in epithelial function, the molecular mechanism of HCO 3 Ϫ secretion remains incompletely understood. Via its buffering role, HCO 3 Ϫ controls the pH of the luminal microenvironment, a function particularly important to the physiology of many epithelial tissues, including the airways. Consistent with a role for HCO 3 Ϫ secretion in airway function, a previous study found the airway surface liquid to be acidic in cystic fibrosis (CF) 3 compared with normal cell cultures (1), and a similar finding was made in secretions from nasal submucosal glands (SMGs) from CF patients (2). Aberrant pH/HCO 3 Ϫ secretion probably contributes to CF lung pathogenesis in a number of fundamental ways. HCO 3 Ϫ is a chaotropic anion that facilitates efficient solubilization and transport of macromolecules, such as mucus (3). Also, the recent finding that HCO 3 Ϫ secretion is required for mucus secretion (4 -6) and that mucin expansion and viscosity are regulated by HCO 3 Ϫ (7, 8) strongly suggests that adequate HCO 3 Ϫ is required for proper mucus homeostasis. In addition, acidic pH has been shown to reduce ciliary beat frequency (9) and impede bacterial killing by phagocytic cells (10,11). Collectively, these consequences of inadequate HCO 3 Ϫ transport predispose the lungs to mucus blockage, bacterial infection, and disease, all hallmarks of the CF lung.
Current studies suggest that HCO 3 Ϫ exit (secretion) across the luminal plasma membrane of airway epithelial cells is mediated solely by the cystic fibrosis transmembrane conductance regulator (CFTR), the ion channel that is mutated in CF. Exactly how CFTR dysfunction leads to aberrant HCO 3 Ϫ secretion is unclear. In primary cultures of surface bronchial and tracheal epithelial cells from CF humans and pigs, a lack of CFTR is associated with reduced electrogenic HCO 3 Ϫ secretion (12,13). Likewise, in ex vivo studies of liquid/mucus secretion from intact SMGs from a range of species, cAMP-stimulated fluid * This work was supported by a Biotechnology and Biological Sciences secretion depends on both Cl Ϫ and HCO 3 Ϫ and was absent in CF glands (14 -16). Furthermore, detailed studies of anion and fluid secretion from polarized cultures of human airway Calu-3 cells, a model of human tracheobronchial SMG serous cells, also concluded that CFTR was the sole mediator of apical Cl Ϫ and HCO 3 Ϫ secretion (17)(18)(19). However, in many HCO 3 Ϫ -secreting epithelia, including the pancreas (20), salivary glands (21), and gastrointestinal (22) and reproductive tracts (23), HCO 3 Ϫ secretion is mediated by CFTR and one or more Cl Ϫ / HCO 3 Ϫ exchangers belonging to the SLC26 gene family. This family comprises 10 members, and their functional characterization indicates distinct patterns of anion specificity and transport modes. Furthermore, structural analyses indicate that CFTR and SLC26 transporters can physically interact through their regulatory and STAS domains, respectively (24), a process that is enhanced by PKA phosphorylation of the regulatory domain (24). In most cases, these molecular interactions synergize the transport activity of CFTR and the SLC26 exchanger, resulting in enhanced HCO 3 Ϫ and fluid secretion from epithelial tissues.
Therefore, SLC26 anion exchangers have a well documented role in HCO 3 Ϫ secretion in non-airway tissues. Human tracheal airway epithelial cells express abundant SLC26A3, whereas delF508 cells do not (25). Furthermore, RNA analysis has shown that human lungs also express SLC26A9 (26) which appears to act as a constitutively active CFTR-regulated Cl Ϫ channel in cultured human bronchial epithelial cells (27). However, to date, it is unknown whether SLC26 proteins are involved in HCO 3 Ϫ secretion in the airways. Hence, the purpose of this study was to investigate the potential role of SLC26 Cl Ϫ / HCO 3 Ϫ exchangers in transcellular HCO 3 Ϫ secretion in airway epithelial cells. Our results show for the first time that human airway serous cells possess a luminal cAMP/PKA-activated Cl Ϫ /HCO 3 Ϫ exchanger that exhibits functional properties consistent with those of SLC26A4 (pendrin). Short hairpin RNA (shRNA)-mediated knockdown of pendrin expression significantly reduced Cl Ϫ /HCO 3 Ϫ exchange activity and markedly lowered the HCO 3 Ϫ content of the secreted fluid. These studies therefore identify for the first time a critical role for pendrin in transcellular HCO 3 Ϫ secretion by airway serous cells.
shRNA Knockdown of SLC26A4 in Calu-3 Cells-Individual SLC26A4 and cyclophilin B (control)-deficient Calu-3 cell lines were produced using lentivirus-mediated delivery of shRNA (Sigma MISSION) knockdown with a set of four different shRNA sequences (supplemental Table S1). Lentiviral transduction particles were applied at a multiplicity of infection ratio of 1. Transduced cells were selected 48 h post-transduction using 10 g/ml puromycin. KD cell lines used for this study had V te and growth patterns similar to that of WT Calu-3 cells.
Measurement of Intracellular pH-Cells were loaded with the pH-sensitive fluorescent dye 2Ј,7Ј-bis-(2-carboxyethyl)-5(6)-carboxyfluorescein acetoxymethyl ester (10 M) for 45-60 min at 37°C in a HEPES-buffered salt solution, which consisted of 130 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 10 mM Na-HEPES, and 10 mM D-glucose set to pH 7.4. Transwells were placed in a perfusion chamber, mounted onto an inverted microscope stage (Nikon), and perfused with a HCO 3 Ϫ -buffered Krebs solution (KRB), which consisted of 115 mM NaCl, 5 mM KCl, 25 mM NaHCO 3 , 1 mM MgCl 2 , 1 mM CaCl 2 , and 10 mM D-glucose and adjusted to pH 7.4 by bubbling with a 95% O 2 , 5% CO 2 mixture at 37°C. Apical and basolateral bath volumes were 0.5 and 1 ml and were perfused at a rate of 3 and 6 ml/min, respectively. Intracellular pH (pH i ) was measured from 15-20 cells as described previously (31), using a Life Sciences Microfluorimeter System (Life Sciences Resources). Ratio values were calibrated to pH i using the high K ϩ -nigericin method (10 M), using K ϩ solutions of various pH values from 5.6 to 8.6 (31). Mean changes in pH i were estimated by calculating the average pH i over 60 s (120 data points). The initial rate of pH i change (⌬pH i /⌬t) was calculated by linear regression fitted to a minimum of 40 data points. Total buffering capacity was estimated by the ammonium pulse technique, using the Henderson-Hasselbalch equation as described previously (31). The ⌬pH i /⌬t values were converted to transmembrane efflux of HCO 3 Ϫ (ϪJ(B)) using the equation, ϪJ(B) ϭ rate of pH i change ϫ total buffering capacity. For high K ϩ KRB, KCl was increased to 115 mM, and NaCl was reduced to 5 mM. For Cl Ϫ -free KRB, NaCl was substituted with sodium gluconate, with 6 mM calcium gluconate replacing 1 mM CaCl 2 to compensate for the Ca 2ϩ buffering capacity of gluconate, and 5 mM KCl was replaced with 2.5 mM K 2 SO 4 . For Na ϩ -free KRB, 115 mM NMDG-Cl replaced NaCl, and 25 mM choline-HCO 3 replaced NaHCO 3 . Atropine (10 M) was included to block muscarinic receptors. Cl Ϫ -free HEPES-buffered solution consisted of 130 mM sodium gluconate, 2.5 mM K 2 SO 4 , 1 mM magnesium gluconate, 6 mM calcium gluconate, 10 mM HEPES (free acid), and 10 mM D-glucose. All general chemicals were purchased from Sigma-Aldrich except for forskolin (Tocris), and GlyH-101 (Calbiochem).
Transepithelial Liquid Secretion Rates and pH Measurements-After washing confluent monolayers with PBS, to remove mucus, the rate of liquid secretion (J v ) was determined by applying 0.2 and 1 ml of Krebs solution to the apical and basolateral surfaces of the cells, respectively, with the desired agonist or inhibitors. Cells were placed in a humidified CO 2 incubator at 37°C, and the volume of the apical fluid was measured using a calibrated micropipette after 24 h. For HCO 3 Ϫfree experiments, Transwells were bathed in HEPES-buffered solution and maintained in a humidified incubator at 37°C without CO 2 gassing. Correction for evaporative apical volume loss (0.08 Ϯ 0.02 l/cm 2 /h; n ϭ 12) was determined empirically by measuring the reduction in apical volume from Transwells coated with silicone gel to stop fluid leakage across the membrane. The pH of 5% CO 2 -saturated apical fluid was measured using a micro-pH electrode (Hamilton) within 60 s of removing individual Transwells from the incubator. Control experiments showed that pH drifted by 0.04 pH units/min using this approach.
Quantitative RT-PCR Analysis of SLC26 Gene Expression-Total RNA was isolated using the Qiagen RNeasy miniprep kit and quantified using the Bioanalyzer 2100 and RNA total nanochips and reagents (Agilent Technologies). cDNA synthesis was performed with 20 ng/ml total RNA/sample using the GeneAmp RT-PCR kit (Applied Biosystems). A mixed tissue standard was generated from total RNA (liver, spleen, kidney, lung, testis, pancreas, brain, and fetal brain; purchased individually from Clontech). Quantitative RT-PCR (Taqman) analysis was used to measure SLC26 gene expression, using previously optimized primers and probes (Applied Biosystems, Assayson-Demand; supplemental Table S2). Amplicons were designed to span intron-exon boundaries to preclude possible amplification of genomic DNA sequences. Reactions were performed in 96-well optical plates and analyzed on the ABI PRISM 7700 (Applied Biosystems) using the following program: 50°C for 2 min and 95°C for 10 min followed by 45 cycles of 95°C for 15 s and 60°C for 1 min. Data were quantified relative to the standard curve for each gene. Expression was normalized to GAPDH and expressed as percentage relative to the standard curve.
Immunocytochemistry-Calu-3 cells grown on Transwells were fixed and permeabilized with cold 100% methanol (Ϫ20°C) for 15 min on ice. Samples were blocked with 3% horse serum for 1 h at room temperature and then incubated with a mouse polyclonal anti-pendrin (Abnova) at 1:200 overnight at 4°C. After blocking with 3% goat serum for 1 h at room temperature, samples were incubated with a goat Alexa Fluor 488coupled anti-mouse secondary antibody (Invitrogen) at 1:1000 in 3% goat serum. The samples were also stained for zona occludens 1 (ZO-1), using a rabbit anti-ZO-1 antibody (1:500; Invitrogen) and a goat anti-rabbit secondary antibody (1:1000; Red Alexa Fluor 568, Invitrogen). Negative controls omitted primary antibody. Images were collected by confocal laserscanning microscopy, using a Leica TCS-NT system (Leica UK Ltd.) equipped with a ϫ100 oil immersion lens, using appropriate excitation and emission filter sets for dual fluorophore detection. Leica software was used to capture images, under identical conditions of imaging, illumination intensity, and photomultiplier settings.
Statistical Analysis-Results are presented as mean Ϯ S.E., where n indicates the number of experiments. Statistical analysis was performed with GraphPad Prism software (GraphPad Software Inc.), using either a paired Student's t test or one-way analysis of variance with Bonferroni's post hoc test. p values of Ͻ0.05 were considered statistically significant.

Effects of Asymmetrical Cl Ϫ Removal in Unstimulated Cells-
Under control conditions, removal of Cl Ϫ from the apical perfusate did not change pH i , indicating little resting apical anion exchange (AE) activity (Fig. 1A), which would be expected to cause an alkalinization. In marked contrast, basolateral Cl Ϫ removal produced a large, monophasic alkalinization of 0.43 Ϯ 0.01 pH units (p Ͻ 0.001; n ϭ 35). Cl Ϫ readdition reversed this response at an initial rate of 0.52 Ϯ 0.03 pH units/min (ϪJ(B) ϭ 49.1 Ϯ 7.0 mM B min Ϫ1 ). These results indicate functional AE activity in the basolateral, but not apical, membrane of Calu-3 cells under resting conditions, probably mediated by anion exchanger 2 (AE2) (32,33).
Effects of Cl Ϫ Removal Subsequent to cAMP Elevation-Exposure of cells to forskolin (fsk; 5 M) under symmetrical high Cl Ϫ conditions significantly acidified pH i by 0.13 Ϯ 0.01 units (Fig. 1A), at a rate of 0.08 Ϯ 0.01 pH units/min. Subsequent apical Cl Ϫ removal markedly alkalinized fsk-treated cells by 0.64 Ϯ 0.03 pH units (p Ͻ 0.001; n ϭ 35; Fig. 1B). The pH response was typically biphasic, consisting of a fast initial alkalinization followed by a slower secondary rise in pH i that reached a steady state after 3-4 min (Fig. 1A). Upon Cl Ϫ readdition, pH i recovered to base-line values at an initial rate of 0.77 Ϯ 0.08 pH units/min (ϪJ(B) ϭ 67.3 Ϯ 13.3 mM B min Ϫ1 : Fig. 1B). Interestingly, under fsk stimulation, removal of basolateral Cl Ϫ produced no change in pH i (Fig. 1A), which indicates that fsk treatment both inhibits basolateral AE and stimulates apical AE activity. Dideoxyforskolin, an inactive analog of fsk, did not mimic these effects, thus confirming that a change in intracellular cAMP provided the fsk-induced "switch" in the Cl Ϫ -dependent apical and basolateral pH i responses (p Ͼ 0.05 compared with non-stimulated apical and basolateral Cl Ϫ -dependent changes in pH i ; n ϭ 3; data not shown). Furthermore, both vasoactive intestinal peptide and adenosine, physiological, cAMP-mediated agonists in Calu-3 cells (34,35), changed pH i in response to apical Cl Ϫ removal, similar to the fsk-stimulated response (p Ͼ 0.05; n ϭ 4; Fig. 1C). Consistent with a cAMP/PKA-dependent activation of an apical AE, the Cl Ϫ -dependent pH i responses were reduced by the PKA inhibitor H-89 as well as by the general PK inhibitor staurosporine ( Fig. 1D; Calu-3 cells were somewhat insensitive to H-89 and required preincubation at a relatively high concentration (50 M), which may have PKA-independent effects). The changes in pH i caused by apical Cl Ϫ removal in fsk-treated cells depended entirely on HCO 3 Ϫ ; replacement of NaHCO 3 with Na-HEPES markedly reduced the Cl Ϫ -dependent alkalinization and pH i recovery (Fig. 1E). Finally, and in contrast to the results with cAMP agonists, raising cytosolic Ca 2ϩ had no effect 3 ؊ Secretion in the Airways NOVEMBER Fig. S1). Taken together, these results suggest that the observed changes in pH i are due to cAMP/PKA activation of an apical Cl Ϫ -dependent HCO 3

Role of Pendrin in HCO
Ϫ transporter, which strongly selects for HCO 3 Ϫ over OH Ϫ . Properties of cAMP-stimulated Responses to Apical Cl Ϫ Removal-The ability of different anions to support reacidification following Cl Ϫ withdrawal was next measured. Recovery of pH i was supported by a range of monovalent but not divalent anions. A clear selectivity exists among the monovalent anions, with I Ϫ and Br Ϫ exhibiting the highest rate of reacidification compared with the other anions (p Ͻ 0.01; n ϭ 5). The anion/ Cl Ϫ selectivity ratio sequence was as follows: HCOO Ϫ , 0.6; NO 3 Ϫ , 0.8; SCN Ϫ , 0.8; Cl Ϫ , 1; Br Ϫ , 1; I Ϫ , 1.3 ( Fig. 2A). Removal of Na ϩ had no significant effect on the rate of reacidification (⌬pH RA ) (p Ͼ 0.05; n ϭ 4; Fig. 2B). Similarly, hyperpolarization of the resting membrane potential (V m ) using the K ϩ channel Ϫ -buffered bathing solution on pH i , in non-stimulated and forskolin-stimulated (FSK; 5 M) conditions. B, mean ⌬pH RA following apical Cl Ϫ readdition under non-stimulated and fsk-stimulated conditions (n ϭ 35; paired observations; *, p Ͻ 0.001 compared with apical 0Cl Ϫ . †, p Ͻ 0.001 compared with basolateral 0Cl Ϫ ). C, the effect of basolateral vasoactive intestinal peptide (VIP; 150 nM) and bilateral adenosine (ADO; 10 M) on the ⌬pH RA following apical Cl Ϫ readdition. Data are expressed as a percentage of the rate obtained with fsk-treated monolayers (n ϭ 4). All agonists performed in parallel using separate monolayers. D, effect of H-89 (50 M) and staurosporine (1 M) on ⌬pH RA following apical Cl Ϫ readdition in fsk-treated Calu-3 cells (n ϭ 8). Cell cultures were pretreated with either inhibitor for 60 min. Inhibitors were present in all solutions throughout the experiment. Inhibitor-treated and untreated Calu-3 cells ran in parallel. *, p Ͻ 0.001 compared with apical 0Cl Ϫ plus forskolin. E, effect of external HCO 3 Ϫ on mean ⌬pH RA following apical Cl Ϫ readdition in fsk-stimulated Calu-3 cells. HCO 3 Ϫ -free solutions were buffered to pH 7.4 with NaHEPES (n ϭ 5; paired observations; *, p Ͻ 0.001 compared with apical 0Cl Ϫ ; †, p Ͻ 0.001 compared with apical 0Cl Ϫ plus forskolin). Error bars, S.E. 3 ؊ Secretion in the Airways opener 1-ethyl-2-benzimidazolinone (EBIO; 1 mM) likewise was without effect (Fig. 2C). However, clamping V m to 0 mV through a combination of high extracellular K ϩ plus EBIO did reduce the rate of reacidification by 23.7 Ϯ 2.3% (p Ͻ 0.05; n ϭ 4; Fig. 2C), indicating some dependence on V m . The ⌬pH RA was also insensitive to the general anion transport inhibitor 4,4Ј-diisothiocyano-1,2-diphenylethane-2,2Ј-disulfonate (H 2 -DIDS) (p Ͼ 0.05; n ϭ 4; Fig. 2D). The observed changes in pH i depicted in Figs. 1 and 2 are therefore due to the activity of a Na ϩ -independent, H 2 -DIDS-insensitive, electrogenic, monovalent anion transporter, consistent with the known properties of CFTR. Contribution of CFTR and Basolateral Cl Ϫ /HCO 3 Ϫ Exchange to pH i Responses-To determine whether CFTR contributes to the alkalinization caused by apical Cl Ϫ removal, the effect of CFTR inhibition was tested. GlyH-101 (10 M) abolished the pH i response to apical Cl Ϫ removal in fsk-stimulated Calu-3 monolayers (lack of alkalinization in the presence of GlyH-101, shown by the black trace in Fig. 3A). CFTR inh -172 produced similar results (10 M) (data not shown). Note that the initial exposure to GlyH-101 caused an alkalinization that is probably due to inhibition of HCO 3 Ϫ efflux from the cells. In the absence of a CFTR blocker, the activity of a basolateral H 2 -DIDS-sensitive "base" transporter strongly influenced both the magnitude of the pH i response to apical Cl Ϫ removal and the ⌬pH RA (supplemental Fig. S2). To ascertain whether a basolateral H 2 -DIDS-sensitive transporter also affects the Cl Ϫ -dependent pH i response in the presence of GlyH-101, Calu-3 monolayers were first exposed to basolateral H 2 -DIDS before GlyH-101. This maneuver restored apical Cl Ϫ -induced pH i changes to GlyH-101-treated cells (Fig. 3A, blue trace) and indicated the presence of an apical Cl Ϫ /HCO 3 Ϫ exchanger. In contrast to the rates of alkalinization with Cl Ϫ removal, the ⌬pH RA upon apical Cl Ϫ readdition was significantly reduced under these conditions (p Ͻ 0.001; n ϭ 4; Fig. 3C). These data suggest that apical Cl Ϫ /HCO 3 Ϫ exchange is functionally coupled to Cl Ϫ transport by CFTR or that CFTR itself directly contributes to pH changes, in addition to the exchanger.

Role of Pendrin in HCO
Knockdown of CFTR Provides Additional Evidence for a Role of both CFTR and a Coupled Anion Exchanger-The role of CFTR in regulating apical Cl Ϫ /HCO 3 Ϫ exchange was next investigated using CFTR KD Calu-3 cells (29). CFTR content of KD cells was determined to be ϳ28 Ϯ 5% that of wild-type (WT) cells (supplemental Fig. S3, A and B). Compared with WT cells, cAMP-stimulated secretory capacity in these CFTR KD cells was also reduced by ϳ25% (29). Mean pH i values did not differ significantly between WT (7.43 Ϯ 0.03; n ϭ 50) and CFTR KD (7.39 Ϯ 0.05; n ϭ 32; p Ͼ 0.05) Calu-3 cells. The switch from basolateral to apical AE activity in response to cAMP is preserved in CFTR KD cells (Fig. 4A). However, in CFTR KD cells, FIGURE 2. Ionic dependence of apical Cl ؊ -dependent HCO 3 ؊ transport in forskolin-stimulated Calu-3 cell monolayers. A, anion selectivity of the apical Cl Ϫ -dependent pH i changes measured by the ability of the replacement anion (for Cl Ϫ ) to support reacidification under fsk-stimulated conditions. Anions were added iso-osmotically to replace Cl Ϫ . (n ϭ 5; paired observations; *, p Ͻ 0.01 compared with chloride; †, p Ͻ 0.001 compared with formate, nitrate, and thiocyanate). B, the effect of bilateral Na ϩ removal on ⌬pH RA upon apical Cl Ϫ readdition in fsk-treated cells. Data are expressed as a percentage of the rate obtained with fsk-treated monolayers in the presence of bilateral Na ϩ (n ϭ 4; Na ϩ -containing and Na ϩ -free Krebs solutions applied in separate experiments, performed in parallel). C, the effect of bilateral high K ϩ (115 mM) and K ϩ channel opener EBIO (1 mM) on the percentage ⌬pH RA upon apical Cl Ϫ readdition in fsk-stimulated cells. Shown are pH i responses in high K ϩ and EBIO compared with control apical 0Cl Ϫ plus forskolin (FSK) responses. Each condition was measured in separate experiments performed in parallel (n ϭ 4; *, p Ͻ 0.05 compared with apical 0Cl Ϫ plus forskolin). D, the effect of apical H 2 -DIDS (500 M) on the ⌬pH RA upon apical Cl Ϫ readdition in fsk-stimulated cells (n ϭ 4; paired observations). Error bars, S.E. 3 ؊ Secretion in the Airways NOVEMBER  Taken together, these results suggest that both CFTR and an apical Cl Ϫ /HCO 3 Ϫ exchanger contribute to Cl Ϫ -induced pH i changes in Calu-3 cells.

Role of Pendrin in HCO
Pendrin Is Expressed in Calu-3 and Human Airway Cells-Apical SLC26A family Cl Ϫ /HCO 3 Ϫ exchangers contribute to transepithelial HCO 3 Ϫ secretion in other HCO 3 Ϫ -secreting epithelia. To identify the apical AE, quantitative RT-PCR was performed for all 10 members of the SLC26A family (36) using RNA extracted from polarized Calu-3 cells. Fig. 5A shows that a number of SLC26A transporters are expressed in Calu-3 cells, several of which are known to function as Cl Ϫ /HCO 3 Ϫ exchangers (SLC26A4, -A6, -A7, and -A9). Of these four candidates, only pendrin (SLC26A4) has properties that are consistent with our results: a monovalent anion transporter with high affinity for iodide and insensitivity to H 2 -DIDS (37)(38)(39)(40). A mouse polyclonal anti-SLC26A4 antibody detected a band migrating at ϳ100 kDa in immunoblot analysis of Calu-3 whole cell lysates (supplemental Fig. S4). The size of the detected species is consistent with that of the band detected using heterologously expressed pendrin fusion proteins as well as with previous reports (41). Confocal immunofluorescence images revealed a punctate localization of pendrin near the apical membrane of Calu-3 monolayers (Fig. 5C, top, green), when compared with ZO-1 expression (Fig. 5C, bottom, red). Immunohistochemical studies on native human tissue showed that pendrin is highly expressed in surface ciliated cells, is not present in mucoussecreting cells of submucosal glands, and is present, albeit at a lower level than in ciliated cells, in serous-like cells from SMGs (supplemental Fig. S5). This work provides additional support for pendrin expression in human airway serous cells and also confirms that pendrin is expressed in surface bronchial epithelial cells (42).
Although pendrin KD cells showed a fsk-stimulated apical Cl Ϫ -dependent change in pH i , the ⌬pH RA was reduced by 47.6 Ϯ 2.4%, compared with control KD cells (p Ͻ 0.001; n ϭ 5; Fig. 5D). Like CFTR KD cells, the profile of the apical Cl Ϫ -dependent change in pH i in pendrin KD cells was reminiscent of WT Calu-3 cells. It was only supported by monovalent anions (Fig. 5E), was H 2 -DIDS-insensitive (Fig. 5F), and was abolished by GlyH-101 addition (data not shown). Interestingly, unlike WT as well as CFTR KD cells, the pendrin KD cells showed no significant difference in the ⌬pH RA when iodide replaced chloride (p Ͼ 0.05; n ϭ 4; Fig. 5E). This observation is consistent with the KD of an AE which has a high affinity for iodide. Neither pendrin nor control KD Calu-3 cells showed significant differences in CFTR protein expression (p Ͼ 0.05; n ϭ 3; sup-plemental Fig. S3, A and B). Thus, these differences in apical Cl Ϫ -induced changes in pH i in pendrin KD cells do not reflect a change in CFTR expression. In addition, short circuit current (I SC ) measurements demonstrated a similar GlyH-101-sensitive fraction of fsk-stimulated I SC (indicative of CFTR-dependent anion transport) in pendrin and control KD Calu-3 monolayers (p Ͼ 0.05; n ϭ 9; supplemental Fig. S3C). The H 2 -DIDS insensitivity of the apical Cl Ϫ -dependent changes in pH i in WT and pendrin KD Calu-3 cells suggests that other SLC26 members highly expressed in these cells, such as SLC26A2 or SLC26A6 (both are sensitive to H 2 -DIDS (43,44)), are unlikely to be involved in these responses.
pH i Response to Cl Ϫ Removal by Fisher Rat Thyroid Cells Expressing CFTR or Pendrin-The significant changes in I Ϫ /Cl Ϫ ratio for the ⌬pH RA across the different cell types (Figs. 2A, 4C, and 5E), together with expression data and pharmacological responses to H 2 -DIDS and GlyH-101, indicate that both CFTR and pendrin contribute to the Cl Ϫ -induced pH i responses. As a further test, the pH i responses to Cl Ϫ removal and readdition were evaluated in polarized FRT monolayers stably expressing either CFTR or pendrin alone (30). Fig. 6A shows that FRT cells expressing CFTR alkalinized in response to apical Cl Ϫ removal in the presence of fsk, supporting the notion that CFTR can conduct HCO 3 Ϫ into the cell under an imposed outwardly directed Cl Ϫ gradient. The pH i response to apical Cl Ϫ was significantly augmented after fsk treatment.  3 ؊ Secretion in the Airways NOVEMBER    This increase could be completely inhibited by GlyH-101; thus, CFTR mediated the changes in pH i (Fig. 6, A and B). Anion substitution studies revealed that only monovalent anions supported reacidification, and I Ϫ and Cl Ϫ selectivity did not differ (Fig. 6C). Pendrin-expressing FRT cells responded to apical Cl Ϫ removal very differently. Significant Cl Ϫ -induced pH i responses were observed in the absence of cAMP stimulation (Fig. 6D). This Cl Ϫ /HCO 3 Ϫ AE activity was not enhanced by fsk or inhibited by GlyH-101 (Fig. 6, D and E). Importantly, the ⌬pH RA was markedly higher for I Ϫ compared with Cl Ϫ (Fig. 6F; I Ϫ /Cl Ϫ ϭ 3.2), with no ability to transport divalent anions, results that are entirely consistent for pendrin (39,40). Anion Transport Is Critical for Fluid Secretion-We next measured and compared both the rate of transepithelial liquid secretion (J v ) and the pH of secreted fluid samples from WT, CFTR KD, and pendrin KD monolayers. WT Calu-3 monolay- To determine the role played by CFTR in transepithelial HCO 3 Ϫ and liquid secretion, experiments were performed with the CFTR blocker, GlyH-101. In the presence of apical GlyH-101, non-stimulated WT Calu-3 cells absorbed rather than secreted fluid (Ϫ0.18 Ϯ 0.01 l/cm 2 /h; p Ͻ 0.001; n ϭ 3; Fig.  7B), suggesting that all basal liquid secretion involves CFTR. Forskolin-stimulated fluid secretion was substantially inhibited by GlyH-101 (60 Ϯ 7% inhibition; p Ͻ 0.001; n ϭ 3) but nonetheless achieved levels significantly greater than control (p Ͻ 0.001; n ϭ 3). Assuming complete block of CFTR in these experiments, these results suggest that although the majority of the cAMP-stimulated liquid secretion depends on CFTR, a significant CFTR-independent, cAMP-stimulated, component of liquid secretion does exist. The pH of the apical fluid from nonstimulated Calu-3 monolayers did not differ significantly from that of the bathing Krebs solution (compared with pH 7.4; p Ͼ 0.05; n ϭ 3; Fig. 7C). However, apical fluid pH significantly increased to 7.8 Ϯ 0.1 (ϳ60 mM HCO 3 Ϫ ) following fsk stimulation (p Ͻ 0.001; n ϭ 3). Surprisingly, despite reducing the rate of liquid secretion, GlyH-101 addition had no effect on the pH of the secreted fluid under either non-stimulated or forskolinstimulated conditions (p Ͼ 0.05; n ϭ 3).

Role of Pendrin in HCO
The CFTR dependence of Calu-3 liquid secretion was further explored using CFTR KD Calu-3 cells. Under non-stimulated conditions, CFTR KD Calu-3 cells failed to increase the volume of apical fluid over 24 h (p Ͼ 0.05; n ϭ 4; Fig. 7D) and secreted significantly less liquid compared with WT cells under fsk-stimulated conditions (34 Ϯ 3% compared with WT; p Ͻ 0.001; n ϭ 4). Like GlyH-101 addition in WT Calu-3 cells, CFTR KD had no significant effect on the pH of secreted apical fluid under either non-stimulated or fskstimulated conditions (p Ͼ 0.05; n ϭ 3; Fig. 7E). These results establish a critical role for CFTR in regulating the rate of transepithelial liquid secretion but not in controlling the pH ([HCO 3 Ϫ ]) of the secreted fluid. In contrast, pendrin KD did not significantly affect either the rate of transepithelial liquid secretion or the pH of secreted fluid under basal conditions, compared with control KD Calu-3 cells (p Ͼ 0.05; n ϭ 3; Fig. 7, D and E). However, although pendrin KD Calu-3 cells produced an increase in fluid volume (p Ͻ 0.05; n ϭ 3) and pH (p Ͻ 0.01; n ϭ 3) in response to fsk addition, comparatively less fluid was secreted (p Ͻ 0.05; n ϭ 3; Fig. 7D). Importantly, the secreted fluid was markedly less alkaline (pH 7.54 Ϯ 0.01; p Ͻ 0.001; n ϭ 3; Fig. 7E). Taken together, these results are consistent with pendrin mediating the majority of HCO 3 Ϫ secretion (exit) across the luminal membrane of Calu-3 monolayers via coupled Cl Ϫ /HCO 3 Ϫ exchange. Furthermore, because the osmolarity of the secreted fluid from all treatment groups did not significantly differ from the Krebs buffer (data not shown), any increase in [HCO 3 Ϫ ] indicates that the [Cl Ϫ ] of the secreted fluid must decrease proportionally to maintain constant osmolarity. From the [Cl Ϫ ] and volume of fluid secreted, it is possible to estimate total Cl Ϫ secretion capacity for the different cell types (supplemental Table S3). The results show that CFTR KD substantially reduced whereas pendrin KD increased total Cl Ϫ secretion, as would be predicted from their proposed roles in transepithelial salt and fluid secretion in Calu-3 cells (see Fig. 8). Under non-stimulated conditions, Cl Ϫ is accumulated across the basolateral membrane by the combined action of the basolateral Na ϩ -K ϩ -2Cl Ϫ cotransporter, NKCC1, together with parallel activity of the NBC and AE2 (32,33). Under these conditions, a basal level of CFTR activity drives a small amount of a Cl Ϫ -rich secretion with a pH of ϳ7.4 (25 mM HCO 3 Ϫ ) via electrogenic Cl Ϫ efflux through CFTR. Elevation of cAMP inhibits basolateral AE2 and stimulates NBC (21). At the apical membrane, cAMP/PKA further increases CFTR activity as well as activates pendrin (SLC26A4). Stimulation of CFTR activity leads to a marked rise in net transepithelial fluid secretion driven by electrogenic Cl Ϫ efflux through CFTR, whereas the co-activation of pendrin increases the HCO 3 Ϫ content of this secreted fluid to ϳ75 mM (pH 7.9), through coupled Cl Ϫ /HCO 3 Ϫ exchange, with Cl Ϫ cycling across the apical membrane via CFTR and SLC26A4.

DISCUSSION
Our findings demonstrate for the first time the presence of an apical Cl Ϫ /HCO 3 Ϫ exchange activity in cAMP-stimulated Calu-3 cells. This cAMP-activated apical AE activity is Na ϩindependent, H 2 -DIDS-insensitive, and capable of transporting a broad range of monovalent, but not divalent, anions in exchange for HCO 3 Ϫ . The selectivity profile is I ϭ Br Ͼ Cl ϭ HCOO ϭ NO 3 ϭ SCN Ͼ Ͼ OH, consistent with pendrin-mediated transport (38,45). Our results agree with data from Xenopus oocytes expressing human SLC26A4, which show that pendrin can function as a Cl Ϫ /HCO 3 Ϫ exchanger as well a Cl Ϫ /I Ϫ exchanger (39). Furthermore, mouse parotid duct cells show forskolin-stimulated I Ϫ secretion, which was absent in Slc26a4 Ϫ/Ϫ mice (40).
We measured active apical AE activity in cAMP-stimulated cells exposed to no Cl Ϫ (0Cl Ϫ ) even in the presence of the CFTR inhibitor GlyH-101 (Fig. 3). This strongly suggests that apical HCO 3 Ϫ transport can occur independently of CFTR, contrary to many previous Calu-3/SMG studies (17-19, 46, 47). However, in CFTR-transfected FRT cells, apical Cl Ϫ removal also increased pH i . This was completely GlyH-101-sensitive, indicating that HCO 3 Ϫ influx through CFTR does occur under 0Cl Ϫ conditions (Fig. 6, A and B). Therefore, the changes in pH i following removal of luminal Cl Ϫ in cAMP-stimulated Calu-3 cells could also be interpreted as being due to HCO 3 Ϫ influx through CFTR alone or in conjunction with a Cl Ϫ /HCO 3 Ϫ exchanger. We favor the latter possibility based on the monovalent anion selectivity, particularly to I Ϫ , of the pH i response in the different cell types studied. In WT Calu-3 cells, the rate of HCO 3 Ϫ -dependent reacidification displayed an anion selectivity of I Ͼ Cl ϭ HCOO, consistent with the high selectivity of pendrin for I Ϫ (validated in pendrin-transfected FRT cells; Fig. 6F). In contrast, CFTR-transfected cells transported each of these monovalent anions at equal rates (Fig. 6C). Furthermore, CFTR knockdown in Calu-3 cells (Fig. 4C) enhanced the relative I Ϫ selectivity (I/Cl ϭ 2.1), consistent with pendrin mediating a larger component of the HCO 3 Ϫ -dependent reacidification. Under physiological situations, luminal Cl Ϫ levels probably remain much higher (30 -50 mM). Thus, it is unlikely that CFTR contributes significantly to HCO 3 Ϫ flux across the apical membrane, a prediction supported by our fluid pH measurements (Fig. 7). The hypothesis that Calu-3 cells secrete HCO 3 Ϫ solely via CFTR clearly opposes our present findings. Our data do support a critical role for CFTR in "switching" Calu-3 cell HCO 3 Ϫ transport capability from pH i regulatory (via basolateral AE) to HCO 3 Ϫ secretory modes (via activation of an apical AE) in response to cAMP elevation and activation of PKA (for discussion, see the legend to Fig. 8). This dual role of CFTR is particularly evident in the CFTR KD Calu-3 cells, where the reduced CFTR expression in these cells not only lowered the rate of cAMP-stimulated apical AE activity but also led to partial relief of basolateral AE inhibition normally seen in stimulated cells. Taken together, these observations suggest that CFTR participates in both processes. Furthermore, FRT cells expressing pendrin alone (Fig. 6) show a basal level of AE activity, which could not be further enhanced by cAMP elevation (Fig. 6, D and E). These results may indicate that in Calu-3 cells, CFTR tonically inhibits pendrin activity until a cAMP stimulus is received. Exactly how this occurs requires further clarification. Previous studies on CFTR and SLC26A3 and A6 transporters postulate physical interaction between the phosphorylated regulatory domain of CFTR with the STAS domain of the SLC26 transporters, which promote enhanced channel and AE activity (24). Co-expression studies in HEK cells showed that CFTR expression led to a marked increase in Cl Ϫ /OH Ϫ and Cl Ϫ /HCO 3 Ϫ exchange mediated by SLC26A3, -A4, and -A6 (48). It is also possible that, under resting conditions, association between CFTR and pendrin blocks AE activity, a phenomenon recently described for Slc26a9 and the regulatory domain of CFTR in Xenopus oocyte co-expression studies (49).
The inhibition of the basolateral Cl Ϫ /HCO 3 Ϫ exchanger under cAMP-elevated conditions fits with a primary role of this AE in maintaining pH i . It is likely that the basolateral AE is also important for Cl Ϫ accumulation under resting conditions, as has already been suggested for Calu-3 cells (33) as well as for HCO 3 Ϫ -secreting pancreatic duct cells (50). Our results also suggest that GlyH-101 inhibition of CFTR in stimulated cells leads to the reactivation of the basolateral AE, possibly via the resulting alkalinization. This may help prevent an intracellular alkali load by shunting HCO 3 Ϫ accumulation by the basolateral Na ϩ -HCO 3 Ϫ cotransporter (NBC) (Fig. 8). The exact mechanisms involved in the cAMP/CFTR-dependent inhibition of the basolateral AE presently are not clear. AE2 is known to be very pH-sensitive (51). Thus, intracellular acidification produced by the efflux of HCO 3 Ϫ across the apical membrane upon fsk stimulation could provide the basis of one potential mechanism.
We obtained an insight into the role of the apical Cl Ϫ /HCO 3 Ϫ exchanger in Calu-3 cells by examining the composition of the secreted fluid (Fig. 7). Our pH i measurements predict fsk-stimulated, apical HCO 3 Ϫ secretion facilitated by pendrin-mediated Cl Ϫ /HCO 3 Ϫ exchange. Forskolin-stimulated pendrin KD Calu-3 cells showed reduced apical AE activity and produced a fluid of lower pH, consistent with these predictions. Although CFTR knockdown or inhibition by GlyH-101 had no effect on the pH of fsk-stimulated secreted fluid, both maneuvers reduced the volume of secreted fluid. These findings are consistent with electrogenic Cl Ϫ efflux via CFTR-driven transcellular fluid secretion and oppose the notion of HCO 3 Ϫ efflux through CFTR-driven fluid secretion, which is consistent with recent work on native serous cells isolated from pig and human SMGs (52). Cl Ϫ secretion through CFTR is predicted to support apical Cl Ϫ /HCO 3 Ϫ exchange by maintaining an inward Cl Ϫ gradient for efficient anion exchange, which has also been observed for Slc26a3 (Dra) activity in mouse duodenum (53).
Our transcellular liquid secretion measurements showed that J v was dependent on the presence of both Cl Ϫ and HCO 3 Ϫ and was increased, dose-dependently, up to ϳ5-fold by fsk, in a time-dependent fashion (supplemental Fig. S6, A and B). We also found that much of the fluid secretion could be inhibited by basolateral anion transport inhibitors, H 2 -DIDS and bumetanide (supplemental Fig. S6C), suggesting that Na ϩ , K ϩ , and Cl Ϫ uptake via the basolateral Na ϩ -K ϩ -2Clcotransporter and HCO 3 Ϫ uptake via the basolateral NBC (and/or Cl Ϫ /HCO 3 Ϫ exchange by basolateral AE2) support overall luminal fluid secretion from Calu-3 cells under cAMP-stimulated conditions

Role of Pendrin in HCO 3
؊ Secretion in the Airways (Fig. 8). These results are in good agreement with previous studies on Calu-3 cells using a similar methodology (54) or the virtual gland model (55) and also with studies of intact SMGs (56,57). In addition, we measured a maximal [HCO 3 Ϫ ] in the fsk-stimulated secreted fluid of ϳ75 mM (pH 7.9), again very similar to the values obtained by Irokawa et al. (55) (80 mM). Computer modeling studies, based on data obtained in native pancreatic duct cells, indicate that a luminal concentration of HCO 3 Ϫ of this magnitude is compatible with an electroneutral Cl Ϫ /HCO 3 Ϫ AE working in parallel with an electrogenic Cl Ϫ channel, such as CFTR (58). The results of the present study are consistent with this model. Also, the results of our fluid secretion studies agree with previous investigations using intact SMG preparations from a range of species, which concluded that the active secretion of both HCO 3 Ϫ and Cl Ϫ drives fluid secretion (16,57). In vitro measurements in excised porcine bronchi suggested that airway fluid originates primarily from the SMG (59). More recently, in vitro optical studies from individual SMGs from pig and ferrets showed that cAMP agonists induced SMG fluid/mucus secretion by a combination of Cl Ϫ and HCO 3 Ϫ secretion (56,60). This process was reduced in transgenic CF ferrets and pigs (15,61). Most studies of fluid secretion from human airway serous cells or SMGs suggest that secretion is dependent on apical Cl Ϫ and HCO 3 Ϫ secretion through CFTR (14,19,46). This is consistent with the known expression of CFTR in SMG serous cells (52,62). This also aligns with previous studies showing that the CFTR blocker, CFTR inh -172 (63), inhibited pilocarpine-and forskolin-induced airway SMG secretion in pigs and humans without CF but not in human bronchi with CF (47). Interestingly, CFTR Inh -172 had no effect on the pH of porcine SMGs gland fluid secretions under non-stimulated conditions (pH ϳ6.9) but lowered the pH of fluid secreted upon pilocarpine stimulation from ϳ7.1 to ϳ6.7, consistent with the inhibition of CFTR-dependent HCO 3 Ϫ secretion (47). However, these results could also be explained by CFTR working in parallel with an AE.
Physiologically, CFTR-regulated Cl Ϫ /HCO 3 Ϫ exchange in airway SMGs may be important in maintaining the pH of the airway surface liquid. This is necessary for antimicrobial molecules and ultimately mucociliary clearance. Its absence could explain the acidic pH reported in CF airways, which would cause further detriment of lung defense (1,2). Low pH could reduce ciliary beat frequency (9,64), compromise bacterial clearance, and reduce antimicrobial activity of immune cells (10,11). In CF, low pH could also "tighten" adhesive interactions between reported membrane-bound mucins and mucins of the mucus layer of the airway surface liquid (65). By altering the exposure of hydrophobic regions of the mucin molecules (66), the effectiveness of the mucus against harmful inhaled particles would be reduced. HCO 3 Ϫ secretion often accompanies mucus release, suggesting that HCO 3 Ϫ is required for proper mucus production (4 -6). In addition, recent work indicates that HCO 3 Ϫ ions play an essential role in determining the final rheological properties of mucus via competing with fixed anions in mucins for Ca 2ϩ . This ultimately regulates mucin swelling, cross-linking, and gel formation. Lack of HCO 3 Ϫ leads to the formation of thick, sticky, mucus, which strongly adheres to the lining epithelium (7,8). In the context of the SMG and CF, lack of HCO 3 Ϫ would therefore predispose the glands to mucus blockage and damage, a characteristic sign of CF airway disease and an important mediator of lung pathophysiology. The recent finding that pendrin expression is coupled to mucin gene expression in response to inflammatory cytokines is important in that it suggests that mucus production and mucus expansion may be tightly co-regulated (42).
Mutations in the gene encoding pendrin lead to Pendred syndrome, a condition characterized by deafness and in some cases goiter, reflecting its expression in the inner ear and thyroid (67,68). To our knowledge, Pendred syndrome patients have not been reported to develop disturbances in lung physiology. This may be because the lungs use other transporters to regulate HCO 3 Ϫ secretion, such as CFTR, in the absence of pendrin or because compensatory up-regulation of other SLC26A Cl Ϫ / HCO 3 Ϫ exchangers occurs when pendrin is not present. A similar conclusion was reached in studies on pendrin in the kidney, where it functions in acid-base status (69,70), yet no obvious renal abnormalities are observed in either Pendred patients or in Slc26a4 knock-out mice (69,70), Interestingly, the inflammatory cytokine, IL-4, up-regulates the functional expression of pendrin in primary cultures of human bronchial epithelial cells, whereby it mediates secretion of SCN Ϫ and promotes the production of OSCN Ϫ , a potent antimicrobial agent (30). Thus, in the context of airway serous SMGs, pendrin may also participate in Cl Ϫ -coupled SCN Ϫ secretion in addition to HCO 3 Ϫ secretion.
In conclusion, our results are consistent with pendrin mediating the majority of HCO 3 Ϫ secretion across the apical membrane of Calu-3 monolayers via coupled exchange of Cl Ϫ for HCO 3 Ϫ . This work establishes a critical and novel role for this anion exchanger in transporting HCO 3 Ϫ across human airway serous epithelial cells.