TNFα and IL-17 alkalinize airway surface liquid through CFTR and pendrin

The pH of airway surface liquid (ASL) is a key factor that determines respiratory host defense; ASL acidification impairs and alkalinization enhances key defense mechanisms. Under healthy conditions, airway epithelia secrete base (HCO3−) and acid (H+) to control ASL pH (pHASL). Neutrophil-predominant inflammation is a hallmark of several airway diseases, and TNFα and IL-17 are key drivers. However, how these cytokines perturb pHASL regulation is uncertain. In primary cultures of differentiated human airway epithelia, TNFα decreased and IL-17 did not change pHASL. However, the combination (TNFα+IL-17) markedly increased pHASL by increasing HCO3− secretion. TNFα+IL-17 increased expression and function of two apical HCO3− transporters, CFTR anion channels and pendrin Cl−/HCO3− exchangers. Both were required for maximal alkalinization. TNFα+IL-17 induced pendrin expression primarily in secretory cells where it was coexpressed with CFTR. Interestingly, significant pendrin expression was not detected in CFTR-rich ionocytes. These results indicate that TNFα+IL-17 stimulate HCO3− secretion via CFTR and pendrin to alkalinize ASL, which may represent an important defense mechanism in inflamed airways.


INTRODUCTION
The pH of airway surface liquid (ASL), the thin layer of fluid that covers the airway epithelium, is a key factor that determines airway host defense (70,85). Two of the main defense mechanisms in human airways are mucociliary clearance and antimicrobial factor-mediated bacterial killing. Previous studies have shown that abnormal acidification of the ASL impairs these processes, whereas acute alkalinization enhances them and may have therapeutic potential (2,9,19,66,77,80).
In a wide variety of respiratory disorders, inflammatory conditions can also involve airway epithelia (61). In several airway diseases [cystic fibrosis (CF), certain endotypes of asthma, chronic obstructive pulmonary disease (COPD), and non-CF bronchiectasis], the inflammatory phenotype is neutrophil predominant (7,12,18,21,24,51,75). Cytokine and chemokine-driven neutrophil influx acts as a host-protective mechanism (5,22,60). However, how pH ASL is regulated under such inflammatory conditions remains uncertain (15,35,43,58,65,73,84). For example, whether inflammation increases or decreases pH ASL is not well established and may depend on species, models, and inflammatory stimuli (30,33,44). Additionally, the acid-base transport pathways modulated by inflammation are not precisely defined and vary from study to study. Moreover, how various epithelial cell types (ciliated cells, secretory cells, etc.) contribute to HCO 3 Ϫ and H ϩ transport under inflammatory states is uncertain.
To better understand whether inflammation changes the way that airway epithelia control pHASL, we tested two key cytokines, tumor necrosis factor-␣ (TNF␣) and interleukin-17 (IL-17). These cytokines have well-established roles in promoting neutrophilic inflammation (53,56,86). TNF␣ and IL-17 expression is increased in airways of people with CF, severe asthma, and COPD (25,34,57,72,87,92). Moreover, their levels correlate with airway disease severity, rise with exacerbation, and fall with resolution of exacerbation (39,57,89). Additionally, approved agents that target TNF␣ and IL-17 pathways are clinically available, providing opportunities to test these agents for therapeutic intervention (32,55). For these reasons, understanding how TNF␣ and IL-17 regulate pH ASL is clinically relevant.
Here, we tested the hypothesis that TNF␣ and IL-17 regulate pH ASL . We used primary cultures of differentiated human airway epithelia grown at the air-liquid interface. We found that TNF␣ and IL-17 increased pH ASL and investigated the cellular and molecular pathways involved. Learning how air-way epithelia regulate pH ASL in inflamed airways may lead to a better understanding of the pathogenesis of airway disorders and thereby guide future therapeutic strategies.

MATERIALS AND METHODS
Cell culture. Primary cultures of differentiated airway epithelia were obtained without passage from multiple human donors (ages 15-73 yr; 49% male, 51% female) as previously reported (38). Briefly, donor tracheae and/or proximal bronchi were enzymatically digested and seeded together when both were available. Epithelial cells were isolated and seeded onto collagen-coated inserts (Costar no. 3470 polyester, no. 3460 polyester, no. 3413 polycarbonate). Epithelia were differentiated at the air-liquid interface for 3 wk or more before assay. All studies were approved by the University of Iowa Institutional Review Board.
To assess cytokine-induced responses, epithelia were treated on the basolateral side with 10 ng/mL TNF␣ (R&D Systems), 20 ng/mL IL-17 (R&D Systems), or both for 24 or 48 h based on dose-ranging and time-course studies. The physiologically relevant concentrations of TNF␣ and IL-17 would be those in the tissue compartment close to the epithelial basolateral membrane where receptors for these cytokines reside. That is also where studies have shown inflammatory cells expressing these cytokines in airway diseases (4,25,26,34,87). However, concentrations in that tissue compartment remain unknown. Although bronchoalveolar lavage (BAL) does not measure that space, studies of airway diseases have reported abnormally elevated concentrations of both TNF␣ and IL-17 in sputum and BAL liquid. For example, studies have reported IL-17 levels of 30 -45 pg/mL in sputum (6, 57) and 10 -15 pg/mL in BAL (87,89) and TNF␣ levels of 90 pg/mL to 26 -1,990 ng/L in sputum (39,57). Given the tiny volume of ASL (~1 L/cm 2 ) and the large volume of BAL aliquots used in bronchial segment or subsegment, ASL dilution may well be Ͼ1,000-fold. Moreover, the concentrations within the epithelial compartment are likely to be higher. Of note, these concentrations of TNF␣ (10 ng/mL) and IL-17 (20 ng/mL) are similar to those reported by others who investigated their effects (16,37,44,47,57,67). Subsequent studies revealed that these concentrations were adequate to investigate the mechanisms underlying pH ASL alkalinization.
Pharmacologic reagents. All drugs were purchased from Sigma-Aldrich except the following: GlyH-101 was a gift from the Cystic Fibrosis Foundation Therapeutics and Robert Bridges. VX-770 and VX-661 were purchased from Selleckchem and VX-445 from Med-ChemExpress.
pH ASL measurement. pHASL was assayed as previously reported (77). Briefly, we used a ratiometric pH indicator SNARF-1 conjugated to 70 kDa dextran (ThermoFisher Scientific). SNARF-1 is a single excitation (514 nm), dual emission (580 nm and 640 nm) fluorescence pH indicator with optimal range near physiologic pH. To minimize modification of ASL composition, SNARF-1, dextran was delivered as a powder to the apical side and allowed to distribute into ASL for 1 h. Fluorescence ratios were obtained on a laser scanning confocal microscope (Zeiss LSM 880) and converted to pH values using calibration curves constructed from colorless standard pH solutions. The microscope chamber housing epithelia maintained a humidified environment at 37°C. To mimic physiologic conditions, 5% CO2 was added to the chamber atmosphere whenever the basolateral side was immersed in an HCO 3 Ϫ containing buffer solution but removed when Mucosal and serosal chambers were voltage clamped, followed by recording of short-circuit current (I SC) and transepithelial conductance (G t) at baseline and in response to the sequential apical addition of (in M): 100 amiloride, 50 uridine triphosphate (UTP), 100 4,4'-diisothiocyano-2,2'-stilbenedisulfonic acid (DIDS), 10 forskolin and 100 3-isobutyl-2-methylxanthine (IBMX), and 100 GlyH-101.
Real-time PCR. Total RNA was isolated from airway epithelia using RNeasy Lipid Tissue Mini Kit (QIAGEN). Genomic DNA was removed through DNase I (QIAGEN) treatment. Quality of RNA isolation was verified on NanoDrop 2000 spectrophotometer (Ther-moFisher Scientific), and samples with 260/280 ratio Ն 1.8 were carried forward. RNA was reverse transcribed with SuperScript VILO MasterMix (Invitrogen). cDNA thus obtained was amplified using gene-specific primers (Table 1) and Fast SYBR Green Master Mix (Applied Biosystems) on QuantStudio 6 Flex Real-Time PCR System (Applied Biosystems). All reactions were performed as triplicate and gene expression was quantitated as fold change (2 Ϫ⌬⌬CT ).
siRNA knockdown. Gene knockdown in primary airway epithelia was achieved as reported previously (71). Negative control and gene-specific siRNAs were obtained from Integrated DNA Technologies (Table 2), and transfected into dissociated primary airway epithelial cells using Lipofectamine RNAiMax (Invitrogen). Epithelia were seeded onto collagen-coated inserts (Costar #3470), and differentiated at the air-liquid interface. pH ASL was measured at day 6 or 7 past seeding. The efficiency of gene knockdown was assessed with real-time PCR.
RNA-sequencing protocol and analysis. RNA-sequencing (RNAseq) was performed by the University of Iowa Genomics Division using manufacturer-recommended protocols. Briefly, 500 ng of DNase I-treated total RNA was enriched for polyA containing transcripts using beads coated with oligo(dT) primers. The enriched RNA pool was fragmented, converted to cDNA, and ligated to sequencing Pseudoalignment of raw sequencing reads and quantification of transcript-level expression were obtained using Kallisto version 0.45.0 and human transcriptome reference GRCh38.p12 (10). Gene counts were imported into R, and differential expression tests were performed using DESeq2 version 1.22.2 (52). Furthermore, gene expression modeling in DESeq2 accounted for the experimental design, acknowledging and correcting for paired control and treated samples for each donor.
To assess cytokine-induced changes in HCO 3 Ϫ transport, a gene ontology-based approach was used. Gene ontology accession for bicarbonate transport was identified with The Gene Ontology Resource (http://geneontology.org/). The accession term (GO:0015701) was used to mine BioMart (http://useast.ensembl.org/biomart/martview/6e73b1adcd14c5c772bc0ead86726df6) and obtain a data set of bicarbonate transport-related genes. This data set was further expanded and refined through a literature search. The results were visualized as a heatmap generated with Clustvis tool (https://biit.cs. ut.ee/clustvis/) (59).
Intracellular pH assay. Airway epithelia were washed thrice with the Krebs-Ringer solution and loaded with 5 M BCECF-AM (Ther-moFisher Scientific) in the presence of 2.5 mM probenecid to prevent dye extrusion. After 40 min of incubation, epithelia were washed again and immediately transferred to a custom-made chamber on the Zeiss LSM 880 confocal microscope. The basolateral side was submerged in the Krebs-Ringer solution. Using a perfusion pump, the apical side was superfused either with the Krebs-Ringer solution or a Cl Ϫ -free buffer solution containing (in mM): 118.9 Na gluconate, 25 NaHCO 3, 2.4 K2HPO4, 0.6 KH2PO4, 1.0 Mg gluconate, and 5 Ca gluconate. The microscope chamber maintained a humidified, 5% CO 2 atmosphere at 37°C. The final pH for all buffer solutions was 7.4. Imaging was performed with a ϫ40 water immersion lens, and BCECF fluorescence was continuously recorded as the apical buffer was switched from the Krebs-Ringer to Cl Ϫ -free buffer and finally back to the Krebs-Ringer.
BCECF is a dual excitation (440 nm, 490 nm), single emission (535 nm), ratiometric, pH-sensitive dye. To obtain intracellular pH (pH i) values from fluorescence emission ratios, standard curves were constructed. Briefly, after being loaded with BCECF-AM, epithelia were transferred to a high K ϩ calibration buffer containing (in mM): 120 KCl, 15 NaCl, 2.4 K 2HPO4, 0.6 KH2PO4, 1.2 MgCl2, 1.2 CaCl2, and 20 HEPES. The final pH was adjusted by adding HCl or KOH to cover a range from 6 to 8.5. To clamp pH i to the same value as the basolateral buffer, epithelia were kept in known pH buffer solution in the presence of 10 M nigericin for 10 min, and BCECF fluorescence was measured as described above. Standard curves were generated by plotting fluorescence emission ratios against known pH values.
Immunocytochemistry. Airway epithelia were washed thrice with PBS, fixed with 4% paraformaldehyde for 15 min, and permeabilized with 0.3% Triton X-100 for 20 min. To minimize nonspecific staining, epithelia were treated with SuperBlock (ThermoFisher Scientific) containing 0.5% normal goat serum for 1 h at room temperature. Primary antibodies were diluted in SuperBlock and added apically for To detect primary antibodies, the following secondary antibodies were used: goat anti-mouse, goat anti-rabbit or goat anti-rat conjugated to Alexa Flour 488 or 568 (1:1,000; ThermoFisher Scientific cat. no. A-11017, A-21069, A-11077). Actin cytoskeleton was stained with Alexa Fluor 633 phalloidin (1:300; ThermoFisher Scientific cat. no. A22284) added at the same time as secondary antibodies. Epithelia were mounted on glass slides, and Vectashield with DAPI (Vector Laboratories) was used to secure glass coverslips. Imaging was performed on the Olympus Fluoview FV 3000 confocal microscope. Z-stack images were processed with the Olympus Fluoview program.
Statistics. Testing for statistical significance was performed on GraphPad Prism 8 (GraphPad Software). Tests included unpaired or paired t test for comparing two groups, repeated measures ANOVA with posttest Tukey's or Dunnett's correction for more than two groups, and Anderson-Darling test for normal distribution. A P value of Ͻ 0.05 was considered significant. Analysis of RNA-seq data is described above.

Combination of TNF␣ and IL-17 markedly increases pH ASL .
We treated primary cultures of differentiated human airway epithelia with two key proinflammatory cytokines, TNF␣ (10 ng/mL) and IL-17 (20 ng/mL). Twenty-four hours later, we measured pH ASL in the presence of 25 mM HCO 3 Ϫ and 5% CO 2 (Fig. 1A). TNF␣ alone decreased pH ASL , whereas IL-17 alone produced no change. Because TNF␣ and IL-17 are both likely to be elevated in inflammatory airway disorders (7,45,56,57), we also tested the combination. In striking contrast to the individual cytokines, the combination of TNF␣ϩIL-17 markedly increased pH ASL . These results suggest that TNF␣ regulates pH ASL differently than IL-17 and that TNF␣ and IL-17 signaling pathways interact to produce an unexpectedly large increase in pH ASL .
In subsequent studies, we also explored the time course of effects. Thirty minutes after TNF␣ was added, pH ASL decreased and did not further change at 90 min. (Fig. 1B). This time course suggests regulation of H ϩ secretion through a posttranslational mechanism. In contrast, TNF␣ϩIL-17 increased pH ASL over a much slower time course (Fig. 1C), suggesting regulation by transcriptional mechanisms. In subsequent experiments, we applied 10 ng/mL TNF␣ and 20 ng/mL IL-17 and studied epithelia 24 h later to assess the initial response to cytokines and minimize secondary changes.
Because these studies were performed in an HCO 3 Ϫ /CO 2 environment, pH ASL could have increased due to increased HCO 3 Ϫ secretion, decreased H ϩ secretion, or both. To begin to identify the underlying transport processes, we replaced basolateral HCO 3 Ϫ with HEPES and removed CO 2 from the atmosphere. As we observed in the presence of HCO 3 Ϫ /CO 2 , TNF␣ alone decreased pH ASL , indicating that TNF␣ acidifies ASL by increasing H ϩ secretion (Fig. 1D). IL-17 alone did not change pH ASL , which suggested that it did not alter H ϩ transport. Importantly, and in contrast to the large increase observed in the presence of HCO 3 Ϫ /CO 2 , TNF␣ϩIL-17 failed to increase pH ASL . This result suggested that TNF␣ϩIL-17 increased pH ASL by increasing HCO 3 Ϫ secretion and not by reducing H ϩ secretion.
TNF␣ϩIL-17 increase CFTR-mediated HCO 3 Ϫ secretion and ASL alkalinization. HCO 3 Ϫ secretion into ASL requires an apical membrane transport mechanism. Under basal conditions, this role is performed in large part by CFTR (50,69,81). However, a calcium-activated anion channel (CaCC, TMEM16A) might also mediate HCO 3 Ϫ secretion under inflamed conditions (13,30,36). We asked whether TNF␣ϩIL-17 altered the activity of these anion channels. We mounted epithelia in modified Ussing chambers with symmetrical Krebs-Ringer solution and measured I SC and G t responses to channel activators and inhibitors. TNF␣ϩIL-17 reduced amiloride-sensitive I SC and G t , suggesting reduced epithelial Na ϩ channel (ENaC) activity (Fig. 2, A-D).
Although not statistically significant, TNF␣ϩIL-17 tended to increase the response to UTP (a CaCC activator) and DIDS (a CaCC inhibitor), suggesting that CaCC may warrant further investigation. Compared with vehicle control, epithelia treated with TNF␣ϩIL-17 had a larger change in I SC and G t in response to forskolin/IBMX, which led to CFTR phosphorylation (23,79) and activity, and GlyH-101, which inhibits CFTR channel activity (63). These results suggested that TNF␣ϩIL-17 reduced ENaCmediated Na ϩ absorption and increased CFTR-mediated anion secretion. Consistent with the electrophysical changes, TNF␣ϩ IL-17 reduced mRNA for SCNN1A (ENaC ␣-subunit) and TMEM16A but nearly doubled CFTR expression (Fig. 2E).
In addition to transcellular mechanisms, the paracellular pathway provides a route for transepithelial ion transport. We asked whether TNF␣ϩIL-17 changed paracellular conductance. Because G t is the sum of transcellular conductance (G c ) and paracellular conductance (G p ) [G t ϭ G c ϩ G p ], reducing G c to~0 provides a residual G t that approximates G p . In airway epithelia, G c is chiefly dependent on apical conductance to Na ϩ (ENaC) and Cl Ϫ (CFTR and CaCC). Therefore, residual G t after blocking ENaC, CaCC, and CFTR provides an estimate of G p . TNF␣ϩIL-17 for 24 h did not alter estimated G p (Fig. 2F). This result suggested that increased HCO 3 Ϫ secretion was likely not mediated via the paracellular route, with the caveat that the paracellular permeability to HCO 3 Ϫ was not directly measured and could have changed in the absence of a change in G p . However, the transepithelial voltage (Ϫ31 mV), the estimated apical [HCO 3 Ϫ ] (10 -12 mM at pH 7-7.1), and the basolateral [ Ϫ ] (24 mM) suggest that the electrochemical driving forces in TNF␣ϩIL-17-treated epithelia would favor paracellular HCO 3 Ϫ absorption, not HCO 3 Ϫ secretion. We asked whether CFTR contributed to the TNF␣ϩIL-17induced increase in pH ASL . After 24 h of TNF␣ϩIL-17 treatment, we added either vehicle (DMSO) or CFTR inhibitor [CFTR(inh)-172] (suspended in a volatile solvent perfluorocarbon Fluorinert FC-72) to the ASL (Fig. 3A). CFTR inhibition significantly reduced the pH ASL response but did not return it to control values without TNF␣ϩIL-17. We also applied siRNA targeted to CFTR. CFTR knockdown reduced pH ASL in both untreated and TNF␣ϩIL-17-treated epithelia (Fig. 3, B and C). These results indicate that CFTR contributed to the TNF␣ϩIL-17-induced increase in pH ASL . Because pH ASL did not return to control levels after pharmacological or siRNA inhibition of CFTR, the data raised the possibility that TNF␣ϩIL-17 might induce an additional HCO 3 Ϫ secretion mechanism. TNF␣ϩIL-17 alkalinize ASL through CFTR and non-CFTR mechanisms. Because CF epithelia lack functional CFTR, the response to TNF␣ϩIL-17 will depend entirely on non-CFTR HCO 3 Ϫ secretion. To test for a non-CFTR mechanism, we compared TNF␣ϩIL-17-induced responses in non-CF and CF epithelia. We converted pH values (log scale) to [H ϩ ] (linear scale) using the relation [H ϩ ] ϭ 10 ϪpH and calculated net alkalinization, i.e., the decrease in D[H ϩ ] ASL after TNF␣ϩIL-17 treatment. D[H ϩ ] ASL decreased in both non-CF and CF epithelia (Fig. 4A). However, the decrease was larger in non-CF epithelia. This result supported the presence of a CFTR-independent mechanism of alkalinization. It also led us to predict that restoring CFTR function to CF epithelia would further augment the TNF␣ϩIL-17-induced response. To test this possibility, we used epithelia from CF donors carrying at least one F508-CFTR allele and treated them with the triple CFTR modulator regimen (VX-445, VX-661, and VX-770) (40). TNF␣ϩIL-17 produced a greater decrease in D[H ϩ ] ASL in epithelia treated with CFTR modulators (Fig. 4B). These results indicate that TNF␣ϩIL-17-induced ASL alkalinization involved CFTR plus one or more non-CFTR HCO 3 Ϫ transporters.

TNF␣ϩIL-17 induce pendrin-mediated HCO 3 Ϫ secretion.
Both TNF␣ and IL-17 have been reported to modify transcriptional activity in epithelia (17,47). To identify HCO 3 Ϫ transport mechanisms transcriptionally upregulated by inflammation, we performed RNA-seq. TNF␣ϩIL-17 altered expression of hundreds of genes, as displayed in a volcano plot (Fig. 5A). We produced a set of 60 HCO 3 Ϫ transport-related genes using a gene ontology approach and individual gene curation including genes relevant to epithelia (Fig. 5B). TNF␣ϩIL-17 increased mRNA for a subset of the genes, including CFTR and members of the SLC26, SLC4, and carbonic anhydrase families. The HCO 3 Ϫ transport-related gene that showed the largest increase in expression was SLC26A4 (Fig. 5, A and B). To validate the RNA-seq results, we measured SLC26A4 expression with  quantitative real-time (qRT)-PCR (Fig. 5C). TNF␣ alone did not induce SLC26A4 expression, whereas IL-17 alone increased it by 74-fold. Remarkably, TNF␣ϩIL-17 increased SLC26A4 expression by 790-fold, suggesting synergy between TNF␣ and IL-17 pathways at the level of gene expression.
To test for additional functional effects of pendrin expression, we loaded epithelia with BCECF-AM and measured pH i responses using confocal microscopy (Fig. 7A). The basolateral solution was Krebs-Ringer with HCO 3 Ϫ /CO 2 . We perfused the apical side first with the Krebs-Ringer solution, then substituted Cl Ϫ with gluconate until pH i reached a plateau, and then returned to the original Krebs-Ringer solution. If apical Cl Ϫ / HCO 3 Ϫ exchange activity is present, removing apical Cl Ϫ should decrease HCO 3 Ϫ exit, increase cytosolic HCO 3 Ϫ concentration, and increase pH i . In control epithelia, substituting gluconate for Cl Ϫ slightly decreased pH i , whereas it increased pH i in TNF␣ϩIL-17-treated epithelia (Fig. 7, B and C). Changes in acid/base transport are shown in Fig. 7D as the net change in H ϩ flux (⌬[H ϩ ] i ); removing apical Cl Ϫ decreased ⌬[H ϩ ] i , i.e., alkalinized cytosol in TNF␣ϩIL-17-treated epithelia. Reintroducing apical Cl Ϫ produced the opposite changes (Fig. 7,  B-D). These results suggest that apical Cl Ϫ /HCO 3 Ϫ exchange activity was minimal under basal conditions but was significantly induced by TNF␣ϩIL-17. In conjunction with RNA-seq, qRT-PCR, and knockdown studies, these results identified pendrin as a major non-CFTR HCO 3 Ϫ secretion mechanism induced by TNF␣ϩIL-17.
TNF␣ϩIL-17 increase pendrin expression in secretory cells. We asked whether apical Cl Ϫ /HCO 3 Ϫ exchange activity was uniformly distributed across all cells that reach the apical surface by measuring the ⌬[H ϩ ] i response to apical Cl Ϫ removal in individual cells. We recorded responses from 297 control and 347 TNF␣ϩIL-17-treated cells (n ϭ 5 different donors per group) and plotted results as a frequency distribution (Fig. 7E). TNF␣ϩIL-17-treated cells achieved a greater mean reduction in D[H ϩ ] i . However, the distribution of cellular responses was not normal for either condition (P Ͻ 0.0001 by Anderson-Darling test for normal distribution). The wide range and skewness of the TNF␣ϩIL-17 distribution indicated heterogeneity of apical Cl Ϫ /HCO 3 Ϫ exchange activity and suggested that a subgroup of cells exhibited a relatively high level of Cl Ϫ /HCO 3 Ϫ exchange. Therefore, we predicted that TNF␣ϩIL-17 might induce pendrin expression in a specific cell type.
SLC26 transporters in epithelia may function in concert with CFTR to mediate HCO 3 Ϫ secretion (41,42). Previous reports suggest that interactions may be structural, functional, or both. We therefore asked whether CFTR and pendrin are expressed in the same cells of TNF␣ϩIL-17-treated epithelia. We found that like pendrin, CFTR was expressed in CC10-positive cells and that it colocalized with pendrin (Fig. 10, A-E). Recent reports have shown that CFTR is expressed at a very high level in a rare cell type called the pulmonary ionocyte (62,68). We labeled epithelia with Barttin (BSND), a pulmonary ionocyte marker (68), and confirmed intense CFTR immunostaining at the apical pole (Fig. 10, F and H). However, we did not detect high-level pendrin expression in ionocytes (Fig. 10, G and H). Thus, of cell types that reach the apical surface, the cells that express both pendrin and CFTR were predominantly secretory cells.

DISCUSSION
Our results indicate that TNF␣ and IL-17, two key regulators of neutrophilic inflammation, markedly increase pH ASL by increasing transepithelial HCO 3 Ϫ secretion. These proinflammatory cytokines increased the expression of two apical HCO 3 Ϫ transporters, CFTR anion channels and pendrin Cl Ϫ /HCO 3 Ϫ exchangers. Moreover, the maximal increase in pH ASL required the activity of both. At the tissue level, TNF␣ϩIL-17 uniformly increased HCO 3 Ϫ secretion across epithelia from different donors, whereas at the cellular level there was significant cell-to-cell variability in apical Cl Ϫ /HCO 3 Ϫ exchange activity. This heterogeneity was matched by our immunolocalization results, which showed that TNF␣ϩIL-17 increased pendrin expression predominantly in secretory cells and not ciliated cells. Immu- nostaining of TNF␣ϩIL-17-treated epithelia also revealed that secretory cells coexpressed pendrin and CFTR. These data are consistent with single-cell RNA sequence data from human airway epithelia that showed CFTR and pendrin expression in secretory cells under basal conditions (68,90). Coexpression of two apical HCO 3 Ϫ transporters suggests that secretory cells are a main cell type contributing HCO 3 Ϫ secretion under inflamed conditions. It will be important for future studies to characterize the cytoplasmic and basolateral membrane mechanisms that supply HCO 3 Ϫ for apical transporters. In this regard, our bulk RNA-seq data showed that TNF␣ϩIL-17 increased members of SLC4 and carbonic anhydrase families. A particularly interesting example is CA12. Loss of function of carbonic anhydrase XII has been linked to a CF-like pulmonary phenotype (46).
Ionocytes are another cell type that, although relatively rare, express very high levels of CFTR mRNA (62,68). Our finding of intense CFTR immunostaining in ionocytes confirmed those data. However, we did not detect significant pendrin immunostaining in ionocytes either at baseline or after TNF␣ϩIL-17 treatment. Thus, whether ionocytes secrete substantial amounts of HCO 3 Ϫ and contribute to pH ASL regulation in inflamed airways remains uncertain and may await studies using livecell markers of ionocytes.
Colocalization of pendrin and CFTR at the apical membrane of secretory cells could enable interactions between the two. Previous data indicate that pendrin can interact with CFTR, and the interaction may enhance activity of both transporters. The basis for such an interaction may be a structural association between the R domain of CFTR and the sulfate transporter and anti-factor antagonist (STAS) domain of pendrin (28,31,41,42,78). Additional studies are needed to fully understand the underlying basis of CFTR-pendrin interactions; increasing pendrin and CFTR expression with TNF␣ϩIL-17 could provide a useful model.
Previous studies found that IL-17 increased pendrin expression and HCO 3 Ϫ secretion in human airway epithelia, and IL-17 was reported to increase pH ASL in epithelia from three donors (3,44). These data are consistent with our results, with the exception that although we found that TNF␣ϩIL-17 markedly increased pH ASL , IL-17 alone did not. This difference is likely due to experimental differences between the studies. Other studies have shown increased pendrin expression after treatment with type 2 cytokines such as IL-4 or IL-13 (30,33,48,76). However, pH ASL responses across studies have not always been consistent. For example, while Gorrieri et al. increased pH ASL , Haggie et al. (33) reported the opposite effect of IL-13, i.e., pH ASL acidification after IL-13 treatment. Differences between cytokines and doses may be responsible, at least in part. Note, for example, that we found TNF␣ alone induced H ϩ secretion, but when paired with IL-17 it synergistically increased HCO 3 Ϫ secretion. The relative balance between H ϩ and HCO 3 Ϫ secretion under differing culture and experimental conditions may also explain discordant results.
Our study has limitations. First, we used primary cultures of differentiated human airway epithelia. Although this model has yielded numerous insights into airway epithelial biology and enabled the discovery and clinical translation of therapeutics, it does not replace the need for in vivo studies in humans.
Second, we studied the effect of two cytokines, TNF␣ and IL-17, relevant to neutrophil-predominant airway inflammation. Others have reported responses to type 2 cytokines (30,33,41,48). Although previous work has shown interactions between cytokine pathways, including TNF␣ and IL-17 (14,57,74), and multiple cytokines mediate inflammatory response in vivo, to our knowledge the effect of more than one cytokine has not been explored on electrolyte transport or pH ASL . However, none of these studies or our data capture the full complexity of the inflammatory environment that may obtain in vivo. Third, we mimicked inflammation using cytokines; it will be important to know the in vivo contribution of inflammatory cells (macrophages, neutrophils, etc.) and/or infectious agents to pH ASL regulation. Fourth, in addition to our study of  Fig. 10. TNF␣ϩIL-17 induces pendrin coexpression with CFTR in secretory cells but not in ionocytes. Human airway epithelia were treated with TNF␣ϩIL-17 for 48 h. A and B: pendrin was coexpressed with CFTR. A shows an en face projection, and B shows an X-Z projection. C and D: CFTR-expressing cells were CC10ϩ secretory cells. C shows an en face projection, and D shows an X-Z projection. E: colocalization of CFTR with pendrin and CC10 markers measured using the ImageJ software and reported as Pearson's correlation coefficient r (n ϭ 3-5). F and G: ionocytes were identified by labeling with BSND (Barttin) antibodies. Ionocytes showed CFTR labeling but lacked significant pendrin labeling. Scale bars: A and C, 10 m; B, D, E, and F, 5 m. H: quantification of CFTR or pendrin expression in BSNDϩ cells. BSNDϩ cells from CFTR-or pendrin-labeled epithelia were shown in a blinded fashion to 3 investigators experienced in confocal microscopy of airway epithelia. Investigators graded intensity of apical protein (CFTR or pendrin) expression on a scale: none or low ϭ 1, moderate ϭ 2, high ϭ 3. Bars indicate means Ϯ SD. ****P Ͻ 0.0001 by Mann-Whitney U test. mRNA, siRNA knockdown, pharmacological inhibition, function, and immunolocalization, quantifying amounts of pendrin and CFTR protein would be of value. Fifth, we generated epithelia from tracheae and proximal large bronchi. However, characterization of epithelial responses in distal small airways will also be of interest.
Inflammation is critical for host defense. Previous studies indicate that in the respiratory system, an acidic pH impairs, and an alkaline pH enhances two host defenses against bacteria, mucociliary transport and antimicrobial activity (1,2,9,19,66,77,80,88). Finding that TNF␣ and IL-17, important cytokine regulators of neutrophilic inflammation, increase pH ASL emphasizes the importance of an increased pH ASL in defending the respiratory tract. In this regard, it is interesting to note that humans with inherited, autosomal recessive mutations in pendrin, i.e., Pendred syndrome, have not been reported to have increased risk of adverse respiratory outcomes (54,82). It may be that CFTR alone is sufficient to support airway host defense. Our studies revealed little pendrin expression under basal conditions, with the result that pendrin knockdown did not alter pH ASL . However, TNF␣ϩIL-17 markedly increased pendrin transcripts, functional activity, and immunostaining, and pendrin knockdown reduced pH ASL . Thus, maximal pH ASL alkalinization under inflamed conditions may require pendrin as well as CFTR. Systematic large data analyses might define whether people with Pendred syndrome have an increased risk of respiratory outcomes.
The cytokine-induced increased expression of both pendrin and CFTR in serous cells and the resulting increased pH ASL may have implications for therapeutic approaches. For gene therapy/editing approaches in CF, increasing CFTR expression in serous cells or cells that could differentiate to serous cells might be beneficial for raising pH ASL . For CF and other diseases with bacterial infection or impaired mucociliary transport, increasing pendrin expression might at least partially increase pH ASL and enhance respiratory host defenses (49).

ACKNOWLEDGMENTS
We thank the University of Iowa In Vitro Models and Cell Culture Core and the Genomics Division of the Iowa Institute of Human Genetics for technical assistance.