Goblet Cell Hyperplasia Requires High Bicarbonate Transport To Support Mucin Release

Goblet cell hyperplasia, a feature of asthma and other respiratory diseases, is driven by the Th-2 cytokines IL-4 and IL-13. In human bronchial epithelial cells, we find that IL-4 induces the expression of many genes coding for ion channels and transporters, including TMEM16A, SLC26A4, SLC12A2, and ATP12A. At the functional level, we find that IL-4 enhances calcium- and cAMP-activated chloride/bicarbonate secretion, resulting in high bicarbonate concentration and alkaline pH in the fluid covering the apical surface of epithelia. Importantly, mucin release, elicited by purinergic stimulation, requires the presence of bicarbonate in the basolateral solution and is defective in cells derived from cystic fibrosis patients. In conclusion, our results suggest that Th-2 cytokines induce a profound change in expression and function in multiple ion channels and transporters that results in enhanced bicarbonate transport ability. This change is required as an important mechanism to favor release and clearance of mucus.

with goblet cell hyperplasia. Using global gene expression profiling, short circuit current recordings, intracellular pH measurements, and protein immunodetection, we investigated the effects of IL-4 on ion transport at the functional and molecular levels. The results reveal a profound change in expression and function in multiple ion channels and transporters that results in enhanced bicarbonate transport ability. Importantly, CFTR appears to play a key role in this process since its loss of function impairs the mechanism of mucin release.

Modulation of ion transport by IL-4.
For our study, we used bronchial epithelial cells from two individuals, BE37 and BE63, which required lung transplantation due to pulmonary hypertension and idiopathic pulmonary fibrosis, respectively. We chose these cells as the closest to those of healthy controls. Indeed, the two diseases affect the distal part of the lungs and do not damage the epithelium of the main bronchi. We measured transepithelial ion transport properties in cells treated with IL-4 (10 ng/ml) for 24 and 72 hours. Figure 1 shows data obtained from short-circuit current experiments on well differentiated bronchial epithelia (cells plated on porous membrane and kept under air-liquid condition for three weeks). After blocking Na + absorption with amiloride (not shown), cells were stimulated with CPT-cAMP to induce phosphorylation and hence activation of CFTR (Fig. 1A). The resulting current was strongly sensitive to CFTR inh -172, a potent and selective CFTR inhibitor. In the presence of this inhibitor, apical application of UTP generated a very fast current increase that reached a maximum in a few seconds and then declined to pre-stimulation levels in 10-20 minutes (Fig. 1B). The effect of UTP is mediated by intracellular Ca 2+ mobilization that leads to transient activation of TMEM16A Cl − channels 15 ).
Treatment with IL-4 promoted a marked increase in cAMP-and Ca 2+ -activated Cl − secretion, as indicated by the amplitude of the currents blocked by CFTR inh -172 (Fig. 1A) or activated by UTP (Fig. 1B), respectively. CFTR-dependent current was increased 2.6-fold at 24 hours and 3.5-fold at 72 hours, with the two values being significantly different (Fig. 1A). The increase in UTP-dependent current was instead essentially the same (nearly 10-fold) at the two times of IL-4 treatment (Fig. 1B). It is interesting to note that in cells receiving IL-4 for 72 hours, CFTR currents initially showed oscillations that progressively disappeared (Fig. 1A).
We looked at the level of expression of CFTR and TMEM16A proteins with immunofluorescence and western blot techniques (Fig. 1C). With confocal microscopy, we found that IL-4 treatment elicited a strong upregulation of TMEM16A. Immunofluorescence detection of CFTR also revealed an increase in signal although less dramatic than TMEM16A. Both proteins appeared to be localized at the apical side of cells but, importantly, never within the same cell (Fig. 1C, left). In agreement with immunofluorescence and functional data, western blot experiments revealed that TMEM16A protein was strongly upregulated by IL-4 ( Fig. 1C, right, and Supplementary Fig. 1). In contrast, the extent of CFTR protein expression by western blot appeared to be unaltered by the cytokine. We only noted a slight change in mobility that could reflect a modification of the pattern of glycosylation (Fig. 1C, right). We quantified mRNA levels by real time RT-PCR. CFTR mRNA was not altered by IL-4, whereas TMEM16A mRNA was upregulated 7-fold.
To further support the conclusion that the cAMP-dependent current is due to CFTR, we used cell from CF patients. In particular, we used cells from patients homozygous for the F508del mutation, which causes a severe defect in CFTR protein trafficking and a more than 90% decrease in CFTR function. As expected, the cAMP-dependent current was markedly reduced in CF cells, including those treated with IL-4 ( Fig. 1D). However, we noted that IL-4 treatment for 72 hours induced a nearly three-fold increase in CFTR function, an effect proportionally similar to that of IL-4 in non-CF cells. By immunofluorescence, we found that IL-4 increases CFTR protein expression also in CF cells (Fig. 1D). Close inspection of microscopic images reveals that a large fraction of CFTR signal is intracellular given the perinuclear pattern of staining (Fig. 1D, enlarged image).

Upregulation of ion channels and transporters by IL-4.
To determine the extent and time-course of gene expression changes elicited by IL-4, we extracted the RNA from cells treated at different time points, from 6 to 72 hours. Given the complexity of the study, involving four different time points and three separate cell preparations, we chose to use the cells from a single individual, BE 37. The RNA samples were analyzed by microarray hybridization (Fig. 2; GEO Access Number: GSE78914). We were particularly interested in the expression of genes involved in transepithelial ion transport. As shown previously by us, IL-4 markedly increases the expression of the TMEM16A chloride channel 15 and of SLC26A4 (a.k.a. pendrin), an electroneutral anion exchanger 16 . The new data reveal that TMEM16A and pendrin upregulation is already detectable at 6 hours and that expression continues to increase although with a different time-course (Fig. 2). Pendrin has the strongest expression at 72 hours with a 37-fold increase (FDR < 10 −6 ) compared to untreated cells. Instead, TMEM16A expression shows a peak at 24 hours (18-fold, FDR < 10 −4 ) followed by a slight decrease at 72 hours. Importantly, microarray analysis revealed the upregulation of other ion transport systems. Some genes showed a relatively delayed response to IL-4 with a particular hyperexpression at 72 hours. This is the case of the genes coding for ATP12A, SLC31A1, KCNMB4, and SLC7A1 (Fig. 2). ATP12A is the non-gastric form of H + /K + -ATPase, responsible for H + secretion at the apical membrane of epithelial cells 17 . SLC31A1 is a high-affinity copper transporter. KCNMB4 serves as a β subunit of the large conductance Ca 2+ -activated K + channel. SLC7A1 is a high affinity cationic amino acid transporter. Other genes showed a more rapid induction by IL-4. This list includes genes coding for: KCNJ16, a pH-sensitive K + channel 18 ; SLC39A8, a bicarbonate-dependent zinc and iron transporter 19 ; SLC24A3, a K + -dependent Na + /Ca 2+ exchanger 20 ; KCNK3, the TASK-1 two-pore K + channel inhibited by extracellular acid 21 ; SLC12A2, the NKCC1 co-transporter 22 ; SLC6A14, the ATB 0,+ amino acid transporter 23 ; SLCO1B3, an organic anion transporter. Importantly, we also noted that IL-4 strongly induced the expression of the cytosolic carbonic anhydrase 2 (CA2), with a more than 20-fold (FDR < 10 −5 ) upregulation already reached at 6 hours (Fig. 2). In contrast, another carbonic anhydrase, the membrane bound CA12, was downregulated. A more general list of top  By immunofluorescence, we investigated the expression of selected proteins whose upregulation was suggested by microarray data. Whenever permitted by the compatibility of primary antibodies, we also stained the cells for acetylated tubulin or MUC5AC, markers of ciliated and goblet cells, respectively. Similarly to TMEM16A, SLC26A4 protein was markedly upregulated by IL-4, with localization in the apical membrane (Fig. 3). The two proteins showed co-localization in some cells but separate expression in many others. Immunodetection of CA2 revealed strong expression induced by IL-4, with a preferential localization in MUC5AC-positive cells (Fig. 3). Interestingly, SLC12A2 and ATP12A showed a more uniform distribution: they were strongly upregulated by IL-4 in most cells although with different subcellular localization (Fig. 3). As expected, SLC12A2 was in the basolateral membrane. ATP12A appeared in the apical region, with a more marked expression in non-ciliated cells. Expression of TMEM16A, SLC26A4, CA2, SLC12A2, and ATP12A is also shown in Supplementary Fig. 2 at a different scale of view. Upregulation of CA2 and SLC26A4 was also confirmed by western blot analysis ( Supplementary Fig. 3).
Bicarbonate transport in IL-4 treated epithelia. As indicated by microarray analysis, IL-4 stimulates the expression of several proteins involved in anion transport, possibly resulting in enhanced bicarbonate secretion. Therefore, we were particularly interested in the contribution of bicarbonate to cAMP-and Ca 2+ -activated currents. Accordingly, we carried out experiments in a Cl − -free solution, a condition that has been used to estimate net bicarbonate transport 24 . In untreated cells, the absence of extracellular Cl − decreased the cAMP-and Ca 2+ -dependent currents to 11% and 5%, respectively (Fig. 4A,C). Treatment of cells for 72 hours with IL-4 changed the percentage of current remaining after Cl − removal to 20% and 41%, respectively, representing a nearly two-fold and eight-fold proportional increase compared to untreated cells (Fig. 4B,C). Given the transient behavior of the Ca 2+ -dependent current, we also estimated its magnitude not as the size of peak but as area under the curve, AUC (Fig. 4D). Using this parameter, the UTP-activated Cl − -independent component appeared even larger in interleukin-treated compared to control cells (13% vs. 1% of the values measured in the presence of Cl − ).
To further investigate the changes elicited by IL-4, we used bumetanide as the blocker of the NKCC1 co-transporter and S0859 as a general inhibitor of basolateral bicarbonate transporters 25 . First, we studied the cAMP-activated current. As expected, given the role NKCC1 in supporting Cl − secretion, bumetanide markedly decreased the cAMP-activated current in untreated cells (Fig. 5A). Subsequent addition of CFTR inh -172 allowed estimation of the residual CFTR-dependent component remaining after bumetanide. By considering the relative effects of bumetanide and CFTR inh -172, we calculated that NKCC1 inhibition removed ~60% of total secretion mediated by CFTR (Fig. 5A). In another set of experiments, we added S0859 after bumetanide. The representative trace in Fig. 5A shows that S0859 effect consisted of a small transient peak followed by a modest decrease of the current. The total inhibition obtained by S0859 plus bumetanide was only slightly higher than that elicited by bumetanide alone (Fig. 5A). In cells treated with IL-4 for 72 hours, the relative effects of bumetanide and S0859 were modified. The current inhibited by bumetanide was larger in absolute terms but the percent inhibition of total current was significantly smaller (< 40%) if compared to that measured in untreated cells (Fig. 5A,B). This result suggested a higher contribution of other types of basolateral anion transporters. In agreement with this interpretation, S0859 appeared to induce a more marked inhibition in cells treated with IL-4 ( Fig. 5B).
We also studied the effect of basolateral transporter inhibitors on the Ca 2+ -activated current. Apical UTP was given in the presence and absence of bumetanide alone or in combination with S0859. In untreated cells, the UTP-activated current, particularly if measured as AUC, was partially inhibited by bumetanide and almost totally blocked by bumetanide plus S0859 (Fig. 5C). A similar behavior (additive effect of bumetanide and S0859) was also seen in cells treated with IL-4. However, in treated cells the residual response to UTP in the presence of the two inhibitors was sensibly higher (Fig. 5D).
We noted that, even in the presence of bumetanide plus S0859, there were still residual cAMP-and Ca 2+ -dependent currents, particularly in cells treated with IL-4. To further inhibit transepithelial anion transport, we tested acetazolamide as an inhibitor of carbonic anhydrase. When given after bumetanide and S0859, acetazolamide further decreased the CFTR-dependent current ( Supplementary Fig. 4). Total inhibition obtained with the three compounds together was ~85%. The UTP-activated current was also sensitive to acetazolamide. The peak induced by UTP in the presence of the triple combination of inhibitors was smaller than that measured with bumetanide plus S0859 (27.8 ± 2.2 vs. 47.9 ± 3.1 μ A/cm 2 ; Supplementary Fig. 4). As a control experiment, we first added CFTR inh -172 followed by CPT-cAMP. Importantly, after totally blocking CFTR with its selective inhibitor, no further decrease was seen with bumetanide, S0859, and acetazolamide ( Supplementary Fig. 4). The only visible response remaining in the presence of CFTR inh -172 was the small transient current elicited by S0859. This response may be due to activation of Ca 2+ -activated Cl − channels since it was inhibited by CaCC inh -A01 ( Supplementary Fig. 4).
As indicated by our study, pendrin/SLC26A4 is one of the most upregulated genes by IL-4. However, pendrin activity cannot be detected in short-circuit current recordings since its mechanism of ion transport is electroneutral 16 . Therefore, we looked for evidence of bicarbonate transport by measuring intracellular pH with the BCECF fluorescent probe as done previously in Calu-3 cells 26 . To this aim, polarized cell monolayers loaded with BCECF were mounted in a specially-designed cuvette which allowed independent perfusion of the apical and basolateral sides. To assess the presence of a pendrin-mediated Cl − /HCO 3 − exchange, we perfused the apical side with a Cl − -free solution. To remove the contribution of CFTR, which is also permeable to bicarbonate, we used CF cells and no cAMP stimulation. As shown in Fig. 1D, there is negligible contribution of CFTR under these conditions. In untreated cells under resting conditions, apical Cl − removal caused no change in intracellular pH (Fig. 6A).
In contrast, a significant alkalinization was observed in cells treated with IL-4 thus indicating the enhanced expression of an apical bicarbonate exchanger (Fig. 6B). Indeed, in the absence of extracellular Cl − (replaced by gluconate) the exit of bicarbonate through an exchange mechanism is impeded. In the same experiments, we subsequently stimulated the cells with apical UTP. Under this condition, Cl − replacement elicited an intracellular alkalinization that was larger in cells treated with IL-4 (Fig. 6A,B). This effect may result from entry of bicarbonate through Ca 2+ -activated Cl − channels. To explain this effect, we have to consider that the electrogenic exit of Cl − through the channels is increased after extracellular Cl − removal. This causes a depolarization of membrane potential (inner side more positive) that causes enhanced bicarbonate entry. Importantly, in the presence of the inhibitor CaCC inh -A01, which blocks Ca 2+ -activated Cl − channels, UTP-dependent alkalinization was not different from that elicited by Cl − removal alone (Fig. 6C) suggesting that TMEM16A works synergistically with pendrin in secreting HCO 3 − . To more directly test the ability of bronchial epithelial cells to secrete bicarbonate, we carried out experiments in which the apical side of epithelia was covered with 150 μ l of Krebs solution containing 128 mM Cl − and 24 mM HCO 3 − . After 48 hours of incubation at 37 °C in a humidified 5% CO 2 atmosphere, the apical fluid was collected to measure the ion composition (Fig. 7). We noted a remarkable difference between cells treated with and without IL-4. In untreated cells, Cl − increased to 143 mM and HCO 3 − decreased to 16 mM with respect to the initial concentration of each ion in the original solution. With IL-4, Cl − concentration decreased to 107 mM whereas HCO 3 − was accumulated to 42 mM (Fig. 7). In agreement with the high HCO 3 − concentration, the apical fluid of cells treated with IL-4 was more alkaline as indicated by two different types of pH measurements (Fig. 7). Another significant change regarded K + : from the initial value of 4.6 mM, K + concentration was moderately modified in control cells (3.2 mM) but markedly decreased (0.3 mM) in IL-4 treated cells (Fig. 7). In contrast, Na + and Ca 2+ concentration were not significantly altered. We were concerned that addition of liquid on the apical side for 48 hours could perturb the behavior of epithelia in terms of response to IL-4. Supplementary Fig. 5 shows that submerged cells still respond to IL-4 with a marked goblet cell hyperplasia. Control short-circuit current recordings, done after removal of apical fluid, also confirmed upregulation of CFTR and TMEM16A currents by IL-4.
We asked whether the upregulation of anion transport in cells treated with IL-4 plays a role in mucus release. For this purpose, we devised an assay in which ATP plus fluorescent nanospheres (in 50 μ l saline solution) were added to the apical side of epithelia. Epithelia were kept tilted during the addition of fluid to allow unidirectional flow of solution by gravity. After removal of excess fluid, mucus stained by nanospheres was visualized by fluorescence microscopy. In control experiments, done in the absence of ATP, very little fluorescence was visible in cells either treated with and without IL-4 (Fig. 8A,B). This finding indicates that nanospheres do not bind to epithelial surface if mucus release is not stimulated. When ATP was included in the solution, epithelia not treated with IL-4 showed appearance of fluorescent filaments suggesting release of mucus (Fig. 8C). Filaments were strongly increased in number and intensity in cells treated with IL-4 for 72 hours (Fig. 8D). This pattern suggested formation of a complex network of mucus strands upon stimulation with ATP. Importantly, a marked reduction in mucus strands was detected when epithelia treated with IL-4 were previously exposed for three hours to a bicarbonate-free basolateral solution (Fig. 8E) or when experiments were carried out on CF epithelia treated with IL-4 (Fig. 8F). Summary of data is shown in Fig. 8G. Upregulation of SLC6A14. Although not directly connected to bicarbonate transport, we were intrigued by the upregulation of SLC6A14 by IL-4 (Fig. 2). SLC6A14 is responsible for Na + -dependent uptake of basic and neutral amino acids at the apical membrane 23,27 , and has recently been identified as a modifier of lung disease in CF 28 . We carried out short-circuit current experiments in which the apical side of epithelia was exposed to lysine or arginine (Supplementary Fig. 6). Amino acids elicited very small responses in untreated cells. In contrast, epithelia treated with IL-4 showed currents that rapidly appeared after amino acid addition.

Discussion
Prolonged treatment of bronchial epithelial cells with IL-4 or IL-13 promotes marked morphological changes that recapitulate the goblet cell hyperplasia occurring in individuals affected by bronchial asthma 14,29 . Our study reveals that treatment with IL-4 results in a profound modification of ion transport mechanisms as indicated by analysis of gene expression with microarrays and by functional assays. At the transcriptome level, IL-4 causes the upregulation of several genes coding for ion channels and transporters. In addition to TMEM16A Cl − channel and to pendrin/SLC26A4 anion exchanger, IL-4 increases the expression of SLC12A2 which codes for NKCC1. This basolateral co-transporter, whose upregulation was previously observed in vivo in asthmatic subjects 30 , generates intracellular accumulation of Cl − by coupling its uptake with that of Na + and K + . Therefore, NKCC1 is an important determinant of the driving force for Cl − secretion. Importantly, as revealed by immunofluorescence, upregulation of NKCC1 by IL-4 occurs in ciliated as well as in goblet cells. Interestingly, other genes upregulated by IL-4 include the ATP12A proton pump, the pH-sensitive KCNJ16 and KCNK3 K + channels, and the CA2 carbonic anhydrase, all proteins that may be involved in pH and bicarbonate homeostasis.
At the functional level, IL-4 strongly upregulates Ca 2+ -and cAMP-dependent Cl − secretion. While the former process appears to be tightly correlated with overexpression of TMEM16A protein, the latter one appears surprising since CFTR protein expression was found to be unaltered in lysates of IL-4 treated cells. Furthermore, CFTR is specifically expressed in ciliated cells that actually undergo a significant reduction in epithelia treated with IL-4 for 72 hours 14 . To explain the upregulation of cAMP-dependent secretion, it can be hypothesized that the decrease in the number of CFTR-expressing cells is counteracted by upregulation of basolateral Cl − transporters (e.g. NKCC1) that results in a higher driving force for Cl − . The higher driving force for Cl − could also explain the increase in cAMP-dependent Cl − secretion measured in CF cells. However, we cannot exclude that IL-4 also increases CFTR protein expression at the cell surface, a phenomenon that could depend on enhanced trafficking of the protein to the apical surface and/or reduced internalization.
Several lines of evidence suggest that a major effect of changes induced by IL-4 is the enhanced ability of epithelia to secrete bicarbonate. In particular, removal of extracellular Cl − revealed residual transepithelial currents, representing bicarbonate secretion, that were upregulated by IL-4. Also, S0859 and acetazolamide, which affect intracellular bicarbonate concentration in different ways, caused additive inhibitory effects on cAMP-and Ca 2+ -dependent transepithelial currents. Furthermore, measurements of intracellular pH revealed an enhanced permeability to bicarbonate in IL-4 treated cells that may be due, at least in part, to pendrin expression in the apical membrane. Finally, measurements of apical fluid composition showed highly increased bicarbonate concentration, largely exceeding the one on the basolateral side, and, accordingly, a significant alkalinization. This finding remarks the increased capacity of epithelia treated with IL-4 to secrete and generate an asymmetrical distribution of bicarbonate.
Importantly, we found that bicarbonate transport in IL-4 treated cells has an important role in the mechanism of mucus release. When stimulated with a purinergic agonist, cells rapidly released mucus as indicated by staining with fluorescent nanospheres. This mechanism was significantly inhibited by keeping the cells in a bicarbonate-free basolateral solution. Furthermore, mucus release was similarly inhibited when the assay was carried out on CF cells. This result indicates that CFTR activity is particularly important under goblet cell hyperplasia conditions. Interestingly, a defect in mucus release was also recently shown in intestinal goblet cells of CF mice 31 . Such findings seem counterintuitive since CF disease is actually characterized by extracellular accumulation of mucus. However, it is possible that a CFTR-dependent primary defect impairs normal release of mucus that is later expelled in a more condensed state. In this respect, it is intriguing that CFTR has such an important role in rapid mucus release despite being selectively expressed in ciliated and not in goblet cells. It can be hypothesized that CFTR affects mucus by modulating the ion composition of the extracellular milieu. Alternatively, it is possible that a low but physiologically significant expression of CFTR in mucin granules of goblet cells 32 is important for mucus release.
Summarizing, our results reveal profound changes in ion transport mechanisms occurring in bronchial epithelia under conditions that mimic goblet cell hyperplasia. Such changes appear to favor bicarbonate secretion and mucus release. Figure 9 shows a tentative model to depict the process of Cl − and bicarbonate transport. Part of the information shown in the figure is supported by findings of this and previous studies. Other information, particularly the site of expression of some of the proteins induced by IL-4 (i.e. apical vs. basolateral membrane, ciliated vs. non-ciliated/goblet cells), remains to be resolved in future studies. Upregulation of NKCC1, essentially in all cells, indicates a general increase in the basolateral uptake of Cl − that promotes its secretion through apical channels (CFTR and TMEM16A). Cl − may be then recycled back through pendrin in exchange for bicarbonate. In this respect, it has previously been shown in Calu-3 cells that pendrin forms together with CFTR a functional unit that promotes bicarbonate secretion 26 . Regarding the source of intracellular bicarbonate, serving as a substrate for pendrin, our results suggest the contribution of S0859-sensitive basolateral transporters and CA2-mediated conversion from CO 2 . In the context of the model represented in Fig. 9, resulting in net bicarbonate secretion, the upregulation of ATP12A is particularly intriguing. ATP12A, a non-gastric form of H + / K + -ATPase, is expressed in the apical membrane of airway epithelia where it is involved in acidification of apical fluid coupled to K + reabsorption 33,34 . The enhanced expression of ATP12A that we found in cells treated with IL-4 appears to be in contrast with the high bicarbonate levels and the alkaline pH measured in the apical fluid. However, we also found a low apical K + concentration that could limit the activity of ATP12A thus allowing accumulation of HCO 3 − . Therefore, we can hypothesize that K + exit at the apical membrane is the rate-limiting step controlling H + secretion. The channel responsible for K + secretion remains to be identified although possible candidates are among the genes upregulated by IL-4. How K + and H + secretion regulates apical pH, bicarbonate concentration in the airway surface fluid, and mucus dynamics will require future studies.
In conclusion, our study reveals the complexity of events triggered by IL-4. Since our experiments were done on cells from a small number of individuals, future studies will need to confirm these observations in different cell/animal models and ex vivo samples from patients with different lung diseases. Elucidation of molecular mechanisms underlying goblet cell hyperplasia may help to understand pathological alterations occurring in asthma, cystic fibrosis, and other chronic respiratory diseases characterized by mucus hypersecretion, thus leading to possible development of novel therapeutic strategies.
Scientific RepoRts | 6:36016 | DOI: 10.1038/srep36016 Methods Cell culture. The procedures for isolation and culture of human bronchial epithelial cells were described in detail in a previous study 14 . Briefly, mainstem human bronchi, derived from CF and non-CF individuals undergoing lung transplant were dissected, washed, and incubated overnight at 4 °C in protease XIV solution. Epithelial cells were then detached mechanically, dissociated by trypsinization, and cultured in flasks in a serum-free medium (LHC9/RPMI 1640). After 4-5 passages, cells were seeded at high density (500,000/cm 2 ) on Snapwell 3801 porous inserts. After 24 hours from seeding, the medium was switched to DMEM/F12 (1:1) plus 2% New Zealand fetal bovine serum (Life Technologies), hormones, and supplements 14 . The medium was replaced daily on both sides of permeable supports up to 8-10 days (liquid-liquid culture, LLC). Subsequently the apical medium was totally removed and the cells received nutrients only from the basolateral side (air-liquid culture, ALC). This condition favored a further differentiation of the epithelium. Cells were maintained under ALC for 3 weeks. Cells were obtained from two non-CF subjects (BE37: patient with pulmonary hypertension; BE63: patient with idiopathic pulmonary fibrosis) and three CF subjects (BE43, BE49, BE91: patients with F508del/F508del genotype). For CF cells, the LCH9/RPMI 1640 medium also contained in the first 4 days additional antibiotics to eradicate bacterial contamination. For this purpose the mixture of antibiotics (usually colistin, piperacillin, and tazobactam) and dosage were designed on the basis of the antibiogram of bacteria isolated from the most recent expectorate of the patient.
All procedures related to the use of human epithelial cells were carried out in accordance with the approved guidelines. In particular, the protocols to isolate, culture, store, and study bronchial epithelial cells from patients undergoing lung transplant was approved by the Ethical Committee of Gaslini Institute under the supervision of the Italian Ministry of Health. Written informed consent was obtained from all patients using a form that was also approved by the same Ethical Committee.

Microarray analysis.
For the analysis of gene expression, microarray hybridization experiments were performed on three separate preparations of bronchial epithelial cells (BE37) differentiated on Snapwell inserts and treated with and without IL-4 for 6, 12, 24, and 72 hours. Total RNA for each condition was used for hybridization to the Affymetrix GeneChip Human Genome 133A2 array using standard protocols as previously described 35 .
Differentially expressed genes were detected by a Bayesian t-test method, Cyber-t 36 particularly suited when the number of replicates is limited. The Benjamini-Hochberg procedure was used to calculate the False Discovery Rate (FDR). The thresholds used were FDR < 0.05 unless otherwise stated. Microarray data are publicly available (GEO Access Number: GSE78914).
Short-circuit current recordings. Snapwell supports carrying differentiated bronchial epithelia were mounted in a vertical chamber resembling an Ussing system with internal fluid circulation. Both apical and basolateral hemichambers were filled with 5 ml of a Krebs bicarbonate solution containing (in mM): 126 NaCl, 0.38 Figure 9. Mechanisms of anion secretion in bronchial epithelial cells exposed to IL-4. For simplicity, the cartoon shows all channels and transporters within the same cell although some components (e.g. CFTR and TMEM16A) are localized in separate cell types. The NKCC1 transporter (SLC12A2) promotes the intracellular accumulation of Cl − that is then secreted through TMEM16A and CFTR Cl − channels. Bicarbonate is accumulated inside the cell by means of basolateral transporters and by conversion from CO 2 . Pendrin then mediates the exchange of extracellular Cl − with intracellular HCO 3 − . The apical membrane also contains the ATP12A K + /H + pump and possibly a K + channel. Secretion of K + could be the mechanism controlling the acidification of apical fluid by ATP12A. KH 2 PO 4 , 2.13 K 2 HPO 4 , 1 MgSO 4 , 1 CaCl 2 , 24 NaHCO 3 , and 10 glucose. Both sides were continuously bubbled with a gas mixture containing 5% CO 2 -95% air and the temperature of the solution was kept at 37 °C. The transepithelial voltage was short-circuited with a voltage-clamp (DVC-1000, World Precision Instruments) connected to the apical and basolateral chambers via Ag/AgCl electrodes and agar bridges (1 M KCl in 1% agar). The offset between voltage electrodes and the fluid resistance were canceled before experiments. The short-circuit current was recorded with a PowerLab 4/25 (ADInstruments) analogical to digital converter connected to a Macintosh computer.
All antibodies were dissolved in 5% skimmed-milk in TBS-T. Protein bands were visualized using the Super Signal West Femto Substrate (Thermo Fisher Scientific Inc). Direct recording of the chemiluminescence was performed using the Molecular Imager ChemiDoc XRS System (Biorad).
Following incubation with primary antibody, cells were rinsed three times in PBS and incubated with 200 μ l of a solution of secondary Alexa Fluor conjugated antibodies (Invitrogen) diluted 1:200 in PBS-BSA 1% for 1 hour in the dark. After further 3 washes in PBS, the porous membrane carrying the cells was cut from the plastic support of the Snapwell, placed on microscope slides and mounted with Fluoroshield with 4′ ,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich) to stain cell nuclei.
Confocal microscopy was performed using a laser scanning confocal microscope TCS SP8 (Leica Microsystems, Heidelberg, Germany). Image analysis was performed using Leica and ImageJ software. Intracellular pH (pH i ) measurements. Measurements were performed using a pH-sensitive fluorescent probe, BCECF/AM (2′ ,7′ -bis(2-carboxyethyl)-5(6)-carboxyfluorescein acetoxymethylester; Life Technologies) and the fluorescence intensity was recorded at excitation wavelengths of 500 and 440 nm and at emission wavelength of 535 nm by a computer controlled spectrofluorometer (Cary Eclipse Varian). Briefly, bronchial epithelial cells grown on permeable supports in polyethylene terephthalate (BD Falcon) at ALC conditions, and loaded with BCECF, were mounted in a cuvette which allowed independent perfusion of the apical and basolateral sides. The solution contained (in mM): 115 NaCl, 5 KCl, 25 NaHCO 3 , 1 MgCl 2 , 1 CaCl 2 , 10 D-glucose. Solution pH was adjusted to 7.4 by bubbling with 95% O 2 and 5% CO 2 . For Cl − -free conditions, NaCl was substituted with sodium gluconate, CaCl 2 with 6 mM calcium gluconate, and KCl with 2.5 mM K 2 SO 4 . Experiments were carried out at 37 °C.
Calibration of fluorescence ratio to pH i was performed after each experiment using nigericin (10 μ M) and potassium ions (150 mM) at various external pH values varying between 6.0 and 8.0 as previously reported 37 . The value of ∆ pH i was calculated as the difference between the maximum value reached upon the perfusion with Cl-free solution and the value obtained by averaging the last 20 data points recorded before perfusion with the Cl − -free solution.