Decreased Soluble Adenylyl Cyclase Activity in Cystic Fibrosis Is Related to Defective Apical Bicarbonate Exchange and Affects Ciliary Beat Frequency Regulation*

Human airway cilia contain soluble adenylyl cyclase (sAC) that produces cAMP upon HCO3−/CO2 stimulation to increase ciliary beat frequency (CBF). Because apical HCO3− exchange depends on cystic fibrosis transmembrane conductance regulator (CFTR), malfunctioning CFTR might impair sAC-mediated CBF regulation in cells from patients with cystic fibrosis (CF). By Western blot, sAC isoforms are equally expressed in normal and CF airway epithelial cells, but CBF decreased more in CF than normal cells upon increased apical HCO3−/CO2 exposure in part because of greater intracellular acidification from unbalanced CO2 influx (estimated by 2′,7′-bis(2-carboxyethyl)-5(6)-carboxyfluorescein (BCECF) fluorescence). Importantly, ciliated cell-specific cAMP production (estimated by FRET fluorescence ratio changes of tagged cAMP-dependent protein kinase (PKA) subunits expressed under a ciliated cell-specific promoter) in response to increased apical HCO3−/CO2 perfusion was higher in normal compared with CF cells. Inhibition of bicarbonate influx via CFTR (CFTRinh172) and inhibition of sAC (KH7) and PKA activation (H89) led to larger CBF declines in normal cells, now comparable with changes seen in CF cells. These inhibitors also reduced FRET changes in normal cells to the level of CF cells with the expected exception of H89, which does not prevent dissociation of the fluorescently tagged PKA subunits. Basolateral permeabilization and subsequent perfusion with HCO3−/CO2 rescued CBF and FRET changes in CF cells to the level of normal cells. These results suggest that CBF regulation by sAC-produced cAMP could be impaired in CF, thereby possibly contributing to mucociliary dysfunction in this disease, at least during disease exacerbations when airway acidification is common.

identified (1). Mammalian sAC is not activated by G-proteins or forskolin (1) but by HCO 3 Ϫ /CO 2 in a pH-independent manner (2,3) and by Ca 2ϩ , which synergizes with HCO 3 Ϫ (4). sAC expression has been described in many human tissues (5). We have shown that it is expressed in the airway epithelium, where it represents the only known adenylyl cyclase localized to cilia (6). sAC is important for flagellar beating in sperm (7)(8)(9), and we have shown its importance for regulating ciliary beating in human airway epithelia via cAMP production upon stimulation with HCO 3 Ϫ /CO 2 (6). There are different sources of luminal bicarbonate in the airways: it can be secreted from submucosal glands and ciliated cells (10) and can also be produced de novo from CO 2 and H 2 O by locally secreted carbonic anhydrase (11). Two proteins responsible for transporting HCO 3 Ϫ into cells, Slc26a9 (12,13) and CFTR (14), have been described in the apical membrane of airway epithelial cells. Slc26a9 is mainly a chloride channel with very low bicarbonate permeability (12); thus, the major apical HCO 3 Ϫ exchange in human airways occurs likely through CFTR (14) or is dependent on it.
In 1989, the CFTR gene was linked to cystic fibrosis (15). Initial attempts to localize the protein in the airways were based on mRNA in situ hybridization and pinpointed expression mainly to submucosal gland acini (16). Significant expression of CFTR was more recently also shown in the apical membrane of ciliated cells from healthy human beings, whereas CFTR was absent in the apical membrane of cells from CF patients homozygous for the ⌬F508 mutation (17). Expression of human CFTR (driven by the ciliated cell-specific promoter foxj1) in the trachea of a CF mouse model confirmed targeting of CFTR to the apical membrane of ciliated cells while restoring forskolinstimulated chloride secretion (18). Electrolyte conductance through CFTR is usually associated with chloride transport, but bidirectional conductance of bicarbonate through CFTR has also been demonstrated and is thought to be important in the airway and in pancreatic ducts (14, 19 -21). Regulation of HCO 3 Ϫ transport through CFTR has been related to sAC in Calu3 cells, an airway epithelial cell line (22), and in corneal endothelium (23), but there are no reports of the influence of diminished bicarbonate transport through CFTR on the activity of sAC in primary airway epithelial cells, especially as it relates to components of mucociliary transport, especially CBF. Here we investigate the role of CFTR-dependent HCO 3 Ϫ transport on the activity of sAC and its influence on CBF regulation in airway epithelial cells. We also examine the consequences of aberrant CFTR function on this process and discuss these findings in the pathophysiological context of cystic fibrosis.

EXPERIMENTAL PROCEDURES
Chemicals and Solutions-Unless stated otherwise, all of the materials were purchased from Sigma-Aldrich. The compositions of solutions used for apical and basolateral perfusions are shown in Table 1. In bicarbonate-free solutions, sodium or potassium bicarbonate was iso-osmotically replaced by sodium or potassium gluconate (see Table 1).
Air-Liquid Interface (ALI) Cell Culture-Normal human airways were obtained from organ donors whose lungs were rejected for transplant. Institutional Review Board approved consent for research with these tissues was obtained by the Life Alliance Organ Recovery Agency of the University of Miami and conformed to the declaration of Helsinki. Lungs from CF patients were obtained with IRB-approved consent at the time of transplant. All CF patients had an abnormal sweat test and the following genotypes: homozygous for ⌬F508 (n ϭ 3), G524X/⌬F508 (n ϭ 1), or 621(G3 T)/⌬F508 (n ϭ 1). For the patient with 621(G3 T)/⌬F508 mutation, Ussing chamber experiments showed chloride conductance abnormalities typical for CF. The patient with the G524X/⌬F508 mutation received a lung transplant at age 25. Based on these findings, all cells used from CF patients had, as expected from the mutations, severe defects in apical chloride conductance caused by either a lack of CFTR or malfunctioning CFTR and were used interchangeably. Airway epithelial cells were isolated and dedifferentiated through expansion. Passage 1 cells were redifferentiated at an ALI on collagen-coated 24-mm T-clear or 12-mm Snapwell filters (Costar Corning) as described previously (24 -27).
Deciliation-Fully differentiated epithelial cells, grown at the ALI, were washed with Dulbecco's modified phosphate-buffered saline. Then 500 l of deciliation buffer (20 mM Tris-HCl, pH 7.5, 50 mM NaCl, 10 mM CaCl 2 , 1 mM EDTA, 7 mM 2-mercaptoethanol, and 0.1% Triton X-100) was applied to the apical side of the culture (6,28). The cultures were gently shaken for 1 min before the apical fluid was collected and centrifuged at low speed (1,500 ϫ g) to pellet cells. The supernatant was recentrifuged at high speed (16,000 ϫ g), and the resulting pellet containing axonemes was resuspended and stored at Ϫ80°C until use.
Western Blot-ALI cultured cells were lysed (1% SDS in 10 mM Tris, pH 8.5, and 0.1 mM EDTA) in the presence of protease inhibitors and cleared from debris by centrifugation. Ten g of protein (cilia, whole, and deciliated cells) of normal and CF cell lysates and 20 g of purchased protein from testis (BD Biosciences, San Jose, CA) per lane were separated using SDS-PAGE on 12% gels (Bio-Rad) and electroblotted onto Immobilon P membranes (Millipore, Billerica, MA). The membranes were blocked with 10% nonfat dry milk and 0.05% Tween 20 in PBS and incubated with rabbit anti-sAC serum 1:10,000 (gift from W. Geng, Southwestern Medical Center, Dallas, TX). After incubating with a secondary horseradish peroxidase-la-beled antibody (KPL Inc., Gaithersburg, MD), chemiluminescence (Pierce) was used for detection and quantification on a ChemiDoc XRS system (Bio-Rad). The membranes were stripped with Restore Western blot stripping buffer (Pierce) and reprobed with rabbit anti-␤-actin (1:100) (Sigma-Aldrich) or anti-acetylated tubulin (1:1,000) (Sigma-Aldrich). Human testis protein medley was used as a positive control. The signals were quantified using ImageJ. Densities from all sAC bands were added and divided by densities of acetylated tubulin (whole cells and cilia) bands. The ratio from CF cells was divided by the ratio from normal cells giving the relative sAC expression in whole cells and cilia in CF compared with normal cells. The same procedure was performed with the 50-kDa sAC band in whole cell and cilia preparations.
Pseudotyped Lentivirus Vectors and Infection of Airway Epithelial Cells-Third generation, propagation-deficient, HIVpseudotyped lentiviruses encoding fluorescently tagged PKA subunits (RII-CFP and CAT-YFP) under transcriptional control of the ciliated cell-specific foxj promoter were used as described previously (25) for FRET. Briefly, recombinant lentiviruses were constructed using the pRRLsinPPT.CMV.MCS. Wpre vector (29). For the initial constructs, genes encoding the catalytic PKA subunit CAT and the regulatory PKA subunit RII, fused to the fluorescent proteins YFP and CFP, respectively (30), were cloned into the multiple cloning site downstream of the ciliated cell-specific foxj1 promoter for better expression in ciliated cells (31). Lentiviruses were prepared by co-transfecting HEK 293T cells with lentivirus and packaging plasmids using calcium phosphate co-precipitation. Virus-containing medium was collected, and undifferentiated airway epithelial cells, plated onto collagen-coated T-col filters in bronchial epithelial growth medium, were incubated with both viral constructs overnight at 37°C in 5% CO 2 . The following day, virus was discarded, and medium was changed to ALI medium top and bottom until cells reached confluence, when apical media were removed to establish the ALI. Expression of the fluorescently tagged proteins was monitored and captured using an inverted fluorescence microscope. Confocal images were obtained on a Zeiss LSM 510/UV confocal microscope.
Measurement of CBF and FRET in Airway Epithelial Cells-ALI cultures were placed in a customized closed chamber allowing independent perfusion of the apical and basolateral compartments, which were fully enclosed. The chamber was mounted at room temperature on the stage of an upright Nikon E600fn microscope, and water was added on top of the closed chamber for use of a 63ϫ water immersion objective with a numerical aperture of 1.0. If required, the cells were permeabilized apically or basolaterally. For basolateral permeabilization, we used Staphylococcus aureus ␣-toxin (10,000 units/ml) (25,32), and for apical permeabilization, we used nystatin (180 g/ml) as described previously (33). Solutions used for perfusion of permeabilized and nonpermeabilized cells are described in Table 1 and were used as follows. For experiments without permeabilization, solution 1 was used for apical and basolateral perfusion, whereas solution 2 was used apically for bicarbonate exposure after obtaining a base-line CBF with solution 1. For experiments with basolateral permeabilization, the cells were apically perfused with solution 1 and basolaterally with solution 3, whereas solution 4 was used basolaterally for bicarbonate exposure after obtaining a base-line CBF with solution 3. To evaluate for successful permeabilization of the basolateral membrane, the ATP free solution 5 was used at the end of the experiment. In case of permeabilization, CBF drastically slows down upon ATP withdrawal. For experiments with apical permeabilization, the cells were basolaterally perfused with solution 1 and apically with solution 3, whereas solution 4 was used apically for bicarbonate exposure after obtaining a base-line CBF with solution 3. FRET was measured as described previously (25). The cultures were exposed to CFP excitation (20-nm-wide filter, centered on 436 nm; Chroma, Bellows Falls, VT). The excitation was controlled by Metafluor TM software (Molecular Devices) via a DG4 rapid wavelength switcher (Sutter Instruments, Novato, CA). The cells were imaged with a CoolSnap Hqcooled CCD camera (Roper Scientific, Tucson, AZ) attached to the microscope using a Dual-View (Optical Insights, Tucson, AZ). The Dual-View duplicates the center 50% of the image and reflects the two images side-by-side onto the CCD chip, thereby allowing simultaneous visualization and recording of the same cell at the two different emission wavelengths of CFP and YFP (20-nm-wide filters centered on 480 for CFP and 535 for YFP; Chroma). The Metafluor TM FRET module (Molecular Devices) was used to measure the CFP/YFP ratio (FRET-RATIO) dynamically every 10 s. The difference between FRET ratios before and under perfusion with HCO 3 Ϫ /CO 2 was called ⌬FRET-RATIO (in arbitrary units). CBF was recorded according to published methods (25, 34), using infrared differential interference contrast video microscopy. The infrared signal was guided to a Sony XC-7500 CCD camera, using a Nikon dual image module, whereas the fluorescence signals below 600 nm were guided to the Dual-View as mentioned above. CBF and FRET were measured in real time and simultaneously in ciliated cells that expressed both fusion proteins.
Immunohistochemistry-Differentiated normal human airway epithelial cells grown on 0.4-m pore-sized filters at the ALI and expressing the tagged PKA subunits were fixed with 4% formaldehyde and permeabilized with 0.1% Triton X-100 and 0.05% Tween 20. The cells were blocked with 2% bovine serum albumin. The cilia were visualized with anti-acetylated tubulin antibody (1:500; Sigma) using an Alexa Fluor-555-labeled goat anti-mouse antibody as secondary label (1:2000; Molecular Probes). DAPI was added before mounting to stain nuclei. Intracellular pH Measurement with BCECF-Intracellular pH (pH i ) was measured as described previously (32). Briefly, cells plated onto collagen-coated coverslips (ALI cultures could not be reliably calibrated for these measurements) were loaded with 2.5 M BCECF-AM for 15-30 min. For fluorescence measurements, a Lambda DG4 excitation system (Sutter, Novato, CA) was used with 10-nm-wide excitation filters centered on 495 and 440 nm (Chroma Technology Corp., Brattleboro, VT). Ratio tool software from Isee Imaging (Raleigh, NC) controlled the output of the Lambda DG4. Ratiometric pH was estimated by capturing the light (535 nm) emitted from the cells through a 63ϫ water immersion objective (Nikon Instruments Inc. Melville, NY) and directed to a cooled CCD camera (CoolSnap Hq, Photometrics, Tucson, AZ). Individual ciliated cells were identified as regions of interest, and the BCECF ratio of emission intensity after excitation at 495 and 440 nm was computed within each region of interest every 10 -60 s on a pixel-by-pixel basis (after background fluorescence subtraction). The system was calibrated with 15 M of the ionophore nigericin in 130 mM KCl-containing solutions at different pH levels between 6.8 and 7.8 (32).
Ussing Chamber Experiments-Fully differentiated cells cultured on Snapwell inserts were mounted into Ussing chambers. CF and normal epithelial cells were bathed with Krebs Henseleit buffer at 37°C and studied under short circuit conditions (voltage clamped at 0 mV) with transepithelial resistance (R te ) determined by the application of intermittent (2 s) 1-mV bipolar pulses. After cultures were equilibrated, amiloride (10 M, luminal) was added to block electrogenic Na ϩ absorption. cAMP-dependent Cl Ϫ secretion was measured at the plateau phase after the addition of 10 M forskolin. After establishing a new base-line current, 10 M ATP was added for measuring calcium-dependent chloride channel activation over P 2 Y 2 activation.
Statistics-The results were compared by one-way analysis of variance and, if a significant difference was found, by the Tukey-Kramer honestly significant difference test using JMP software (SAS). p Ͻ 0.05 was accepted as significant.

Chloride and Bicarbonate Conductance of CFTR-
The lack of chloride conductance through functional CFTR was confirmed for all CF lungs used. Fig. 1 illustrates Ussing chamber experiments with cells from patients with mutations for 621(G3 T)/⌬F508 (Fig. 1A) and homozygous ⌬F508 (Fig.  1B). No significant chloride conductance was found upon apical stimulation with 10 M forskolin (tmAC agonist) in any of the tested CF cultures. In contrast, normal control cultures showed significant increases in I sc upon forskolin stimulation. The expected I sc response to ATP could be documented in both CF and normal cells (Fig. 1). Ussing chamber experiments were conducted to show absent, CFTR-dependent apical chloride current using cells from CF patients with the following CFTR mutations: 621(G3 T)/⌬F508 (A) and ⌬F508/⌬F508 (B). Note the absence of forskolininduced chloride current in both CF traces in the presence of the ENaC blocker amiloride. The cells were subsequently exposed to 10 M ATP to demonstrate CFTR-independent chloride current in CF cells.

CBF, sAC, CF, and Apical HCO 3 ؊ Exchange
Expression of Soluble Adenylyl Cyclase Protein in Airway Epithelial Cells-To examine the expression and localization of sAC in normal and CF airway epithelial cells, Western blots were performed using protein lysates from whole cells, deciliated cells, isolated cilia/axonemes, and human testis (positive control). All of the samples (except testis) were prepared from fully differentiated ALI cell cultures from donors without underlying lung disease (normal) and from the lungs of patients with cystic fibrosis. To probe the Western blots, a polyclonal rabbit antibody against a peptide (SLSEGDALLA) (5) near the N terminus of human sAC proteins was used as described previously (6). For quality and quantity controls, the same blots were stripped and reprobed with antibodies against ␤-actin and acetylated tubulin (Fig. 2). The distribution of bands for sAC found in these blots was consistent with the previously described 50-kDa and full-length 190-kDa forms (2) as well as an 80-kDa variant (5). Interestingly, the 50-kDa form seemed to localize preferentially to cilia/axonemes of airway epithelial cells. There was no difference in the expression and localization of total sAC between cultures from donors without lung disease and patients with cystic fibrosis (each n ϭ 3). The intensities of the 50-kDa band preferentially expressed in cilia, normalized for tubulin for whole cells and cilia preparations, show a nonsignificant increased expression of sAC in CF cells versus normal cells with ratios of 1.33 for whole cells and 1.35 for cilia. The sum of the intensities of the 50-, 80-, and 190-kDa sAC bands normalized to tubulin for whole cells shows a nonsignificant increased expression of sAC in CF cells versus normal cells with a ratio of 1.38 (Fig. 2).
Ciliary Beat Frequency-We have previously shown that apical and basolateral changes in HCO 3 Ϫ /CO 2 exposure influence CBF in human airway epithelial cells. sAC is stimulated by bicarbonate and subsequently produces cAMP to activate PKA (6). The response of ciliary beating to apical bicarbonate is complicated by the fact that HCO 3 Ϫ equilibrates with H 2 O and CO 2 in the presence of H ϩ and that CO 2 can freely cross the cell membrane and acidify the cytoplasm. This cytoplasmic acidification leads to a decrease of CBF (32) that is partially rescued in normal cells by the bicarbonate stimulation of sAC. Because bicarbonate is mainly conducted through CFTR at the apex of human airway epithelial cells (14, 19 -21), we hypothesized that cells from cystic fibrosis patients cannot rescue the acidification-induced CBF decrease upon increased apical HCO 3 Ϫ /CO 2 exposure because of a defective apical HCO 3 Ϫ exchange caused by absent or dysfunctional CFTR.
To investigate the effect of apical bicarbonate changes on CBF in normal and CF cells, exposure of the apical membrane was alternated from 0 to 25 mM sodium bicarbonate, whereas the basolateral perfusate did not contain bicarbonate. The bicarbonate-containing solution was equilibrated with CO 2 by bubbling the solutions with 5% CO 2 before filling into gas-impermeable syringes and tubing. Using a similar experimental setup, we showed previously that apical perfusion with bicarbonate leads to a decrease of CBF because the pH effect of CO 2 on CBF is stronger than the sAC-mediated CBF increase after stimulation by bicarbonate (6). Mean CBF base lines in normal cells (6.6 Ϯ 0.2 Hz; n ϭ 42 from six different lung donors) and in CF cells (7.1 Ϯ 0.2 Hz; n ϭ 53 from three different CF patients) were the same (p Ͼ 0.05). On the other hand, ⌬CBF from base line upon switching to apical bicarbonate-containing solutions was Ϫ23.6 Ϯ 1.3% in normal cells, significantly less than in CF cells where ⌬CBF was Ϫ30.3 Ϯ 1.7% (p Ͻ 0.05) (Fig. 3A). The absolute decrease in CBF was also significantly larger in CF cells, 2.14 Ϯ 0.2 Hz, versus normal cells, 1.57 Ϯ 0.2 (p Ͻ 0.05). These results (Fig. 3, A and B) suggest that an apical HCO 3 Ϫ exchange defect in CF could affect regulation of CBF in these cells, possibly because of a lack of sAC activation. Alternatively, the differences could be solely related to a difference in overall pH i , again caused by a defect in apical HCO 3 Ϫ exchange. To examine the latter possibility, we measured the changes in pH i , using BCECF and fluorescence microscopy (Fig. 4). In agreement with the hypothesis that bicarbonate is transported into the cells through CFTR in normal but less so in CF cells, we found greater acidification in CF cells (pH decrease by Ϫ0.49 Ϯ 0.04; n ϭ 10, two different donors) than in normal cells (pH decrease by Ϫ0.34 Ϯ 0.02; n ϭ 10, two different donors; p Ͻ 0.05).
To determine whether or not this pH difference explained the differential CBF regulation in normal versus CF cells, we tested the effects of HCO 3 Ϫ on normal and CF cells after permeabilizing the basolateral or apical membrane. Permeabilization of the membranes allows free, CFTR-independent exchange of HCO 3 Ϫ , equalizing the intracellular pH and bicarbonate concentration in normal cells and in cells from a patient with CF. Basolateral permeabilization was accomplished with ␣-toxin and confirmed by a significant CBF decrease upon basolateral perfusion with ATP-free solutions as described previously and demonstrated in Fig. 5B (6). While changing the bicarbonate concentration over the basolaterally permeabilized membrane, the apical perfusate was kept without bicarbonate in these experiments. CBF responses to changes in HCO 3 Ϫ in basolaterally permeabilized cells were statistically not significantly different between normal versus cystic fibrosis cells (Fig. 5A). In normal cells, ⌬CBF was Ϫ15.4 Ϯ 1.2% from a base line of 8.0 Ϯ 0.4 Hz (n ϭ 36, six different donors), and in CF cells, ⌬CBF was Ϫ18.9 Ϯ 2.1% from a base line of 10 Ϯ 0.45 Hz (n ϭ 29, two different donors). Although the base-line CBF of normal cells was significantly lower than that of CF cells (p Ͻ 0.05) during basolateral permeabilization, ⌬CBF was not (p Ͼ 0.05). The etiology of the different base-line frequencies remains unclear. These results were confirmed in apically permeabilized cells, where CBF was stimulated upon increases in bicarbonate, and these changes were not different between normal and CF cells. ⌬CBF upon apical HCO 3 Ϫ perfusion was Ϫ4.3 Ϯ 3.9% in normal cells (n ϭ 15, three different donors), whereas ⌬CBF was Ϫ8.3 Ϯ 2.8% (n ϭ 22, three different donors) in cells from patients with CF (p Ͼ 0.05) (Fig. 6A). The base-line CBF of the two groups was not significantly different (p Ͼ 0.05). The finding of equal CBF responses in normal and CF cells in permeabilized cells suggests similar mechanisms of CBF regulation by bicarbonate (if it can enter the cells in equal amounts) in CF and normal cells.
To further characterize the influence of sAC on CBF, we used inhibitors to block the sAC pathway at different levels; CFTR was blocked with inhibitor 172 (CFTR inh 172), the activation of sAC with the specific sAC inhibitor KH7 (35,36), and the effect of cAMP on ciliary beating by inhibiting PKA with H89 (37). To evaluate the effects of these inhibitors on CBF, we used ⌬⌬CBF as a measurement. ⌬⌬CBF is the difference between ⌬CBF with and without the use of these different inhibitors upon apical HCO 3 Ϫ /CO 2 perfusion in normal or CF cells. Inhibition of CFTR with 20 M CFTR inh 172 resulted in a ⌬⌬CBF of Ϫ2.9 Ϯ 0.9% in normal cells (n ϭ 44, four different lung donors); in contrast, it did not exaggerate the CBF response in CF cells (n ϭ 20, two different lung donors) with a ⌬⌬CBF of 2.4 Ϯ 1% (p Ͻ 0.05 compared with normal cells; Fig. 7A). Inhibition of sAC with 25 M KH7 (Fig. 7B) resulted in a ⌬⌬CBF of Ϫ4.8 Ϯ 0.8% in normal cells (n ϭ 31, three different donors), whereas KH7 resulted in no further CBF decrease in CF cells with a ⌬⌬CBF of 2.3 Ϯ 1.6% (n ϭ 10, two different CF donors; p Ͻ 0.05 compared with normal cells). Inhibiting PKA with 10 M H89 (Fig. 7C) revealed a ⌬⌬CBF of Ϫ5.6 Ϯ 1.5% in normal cells (n ϭ 36, four lungs), whereas ⌬⌬CBF in CF cells (n ϭ 24, three different lung donors) was 4.5 Ϯ 2.0% (p Ͻ 0.05 compared with normal cells). Base-line CBF was not different between all pairs of treated and untreated normal and CF cells, except that H89 decreased CBF base line in CF but not normal cells.
To ensure that the effects observed under perfusion of the apical membrane were not related to osmotic differences, all of Ϫ /CO 2 . ⌬FRET-RATIO increased significantly more in normal cells compared with CF cells, indicating larger cAMP production. D, model for apical HCO 3 Ϫ /CO 2

perfusion in nonpermeabilized cells. In normal cells (left panel), bicarbonate enters cells via CFTR (or requires functional CFTR) and stimulates sAC, which leads to a CBF increase (ϩ) via activation of PKA. CO 2 crosses the membrane and decreases CBF (Ϫ) by lowering the intracellular pH. In CF cells (right panel),
there is no functional CFTR, so the (ϩ) effect on CBF of sAC is diminished, and CBF drops by a larger degree compared with normal cells.*, p Ͻ 0.05.

FIGURE 4. CBF and pH changes in normal and CF cells upon apical HCO 3
؊ perfusion. A and B, CBF decreases significantly more upon apical HCO 3 Ϫ perfusion in normal cells (A) compared with CF cells (B). C, simultaneous recordings of pH (BCECF fluorescence) illustrate that intracellular pH (pH i ) also decreases more in CF cells compared with normal cells upon apical HCO 3 Ϫ perfusion (n ϭ 10). *, p Ͻ 0.05.

CBF, sAC, CF, and Apical HCO 3 ؊ Exchange
the solutions were osmotically balanced as shown in Table 1.
For all experiments, the solutions contained either 25 mM of sodium gluconate or 25 mM sodium bicarbonate. To make certain that the CBF changes were not nonspecific because of changes of different anions but rather specific to bicarbonate, we also performed a series of experiments exchanging the solutions between 25 mM sodium chloride and sodium gluconate (n ϭ 10, three different lungs); these experiments did not reveal a significant change in CBF upon the changes in salt solution perfusion. The solvent Me 2 SO (1‰) did not affect the CBF response. Thus, inhibiting bicarbonate transport through CFTR (CF cells, CFTR inh 172, but not DNDS), inhibiting sAC by KH7, and inhibiting PKA with H89 made normal cells behave like CF cells with respect to CBF decreases upon apical increases in HCO 3 Ϫ /CO 2 . The fact that CFTR inh 172, KH7, and H89 change the CBF response to apical HCO 3 Ϫ /CO 2 in normal but not in CF cells (Fig. 7, A-C), together with the previous data, suggests that apical HCO 3 Ϫ /CO 2 exposure not only decreases pH i but also activates sAC via HCO 3 Ϫ influx in normal cells to produce cAMP, thereby stimulating CBF via PKA activation. In CF cells, however, this mechanism of sAC activation is defective, most likely because of the inability of luminal HCO 3 Ϫ to enter the cell apically because of defective CFTR.

Real Time Measurement of cAMP Production in Single Airway Epithelial Cells with FRET-To confirm that apical HCO 3
Ϫ /CO 2 exposure increases intracellular cAMP levels upon sAC stimulation more in normal cells compared with CF cells, we used our previously developed method of measuring changes in intracellular cAMP. By measuring the cAMP-induced dissociation of fluorescently tagged subunits of PKA by FRET (25), we can estimate intracellular cAMP levels in real time. Undifferentiated normal and CF human airway epithelial cells were coinfected with two lentiviruses, each encoding a fusion protein of a PKA subunit (RII-CFP or CAT-YFP), and both constructs were driven by the ciliated cell-specific foxj promoter (Fig. 8). Cells showing FRET capability by emitting at 535 nm under excitation at 435 nm were chosen for FRET measurements. Apical perfusion with 25 mM bicarbonate increased the FRET ratio (⌬FRET-RATIO) by 0.0076 Ϯ 0.0007 arbitrary units in normal cells (n ϭ 32, three donors) but only by 0.0045 Ϯ 0.0007 arbitrary units (n ϭ 25, three different donors) in CF cells (p Ͻ 0.05) (Fig. 3C). These changes correspond to cAMP levels of ϳ100 M in normal and 50 M in CF cells (25). However, during basolateral perfusion with 25 mM bicarbonate in basolaterally permeabilized cells, the ⌬FRET-RATIO increase was similar in normal cells (n ϭ 17, two different donors) and CF cells (n ϭ 13, two different donors) (Fig. 5C). These data support the notion that sAC can be activated by apical bicarbonate exposure in normal but less so in CF cells, suggesting an important role for CFTR-mediated bicarbon- Ϫ /CO 2 . There was no significant difference in ⌬CBF after basolateral perfusion with 25 mM HCO 3 Ϫ /CO 2 between basolaterally permeabilized (S. aureus ␣-toxin) normal and CF cells (p Ͼ 0.05). B, CBF tracings of basolaterally permeabilized normal (gray) and CF cells (black) during bilateral bicarbonate exposure. At the end of the tracing, ATP was removed basolaterally, and CBF decreased, demonstrating a successful permeabilization of the basolateral membrane. C, ⌬FRET-RATIO (in arbitrary units) is the difference between FRET-RATIOs with and without apical perfusion with 25 mM HCO 3 Ϫ /CO 2 . No significant differences were found in ⌬FRET-RATIO upon basolateral perfusion 25 mM HCO 3 Ϫ /CO 2 between basolaterally permeabilized normal and CF cells (p Ͼ 0.05). D, model for basolateral HCO 3 Ϫ /CO 2 perfusion in basolateral permeabilized cells. Bicarbonate enters the normal (left panel) and CF (right panel) cells unimpeded through the permeabilized membrane, activates sAC, which produces cAMP to stimulate PKA, thereby increasing CBF (ϩ). CO 2 diffuses across the membrane similarly in both normal and CF cells and decreases intracellular pH, which decreases CBF (Ϫ). FIGURE 6. CBF response in apically permeabilized normal and CF airway epithelial cells. A, apical perfusion (gray) decreases CBF more in CF cells than in normal cells (p Ͻ 0.05). Apical permeabilization with nystatin (black) significantly reduces the CBF decrease and equalizes ⌬CBF between CF and normal cells. B, model for apical HCO 3 Ϫ /CO 2 perfusion in apically permeabilized cells. In normal cells bicarbonate enters the cells through CFTR and additionally through the permeabilized membrane. Intracellular bicarbonate stimulates sAC and subsequently CBF (ϩ) over PKA. In CF cells, bicarbonate can enter the cells through the permeabilized membrane and also stimulates sAC 3 PKA 3 CBF (ϩ). In normal and CF cells, CO 2 enters the cells directly through the membrane and decreases CBF (Ϫ) via acidification. *, p Ͻ 0.05.

CBF, sAC, CF, and Apical HCO 3 ؊ Exchange
ate entry under these conditions in normal cells, thereby helping to regulate CBF. As described above for CBF, we used CFTR inh 172, KH7, and H89 to dissect the effect of bicarbonate on cAMP production as measured by FRET. To evaluate the effects of these inhibitors on intracellular cAMP concentrations, we used the ⌬⌬FRET-RATIO as a measurement. ⌬⌬FRET-RA-TIO is the difference between ⌬FRET-RATIOs with and without the use of the different inhibitors upon apical HCO 3 Ϫ /CO 2 perfusion in normal or CF cells. Inhibition of CFTR with 20 M CFTR inh 172 resulted in a ⌬⌬FRET-RATIO of Ϫ0.0061 Ϯ 0.0012 arbitrary units in normal cells (n ϭ 9, two different lungs); in contrast, it did not exaggerate the FRET-RATIO in CF cells (n ϭ 6, two different lungs) with a ⌬⌬FRET-RATIO of Ϫ0.0011 Ϯ 0.0007 arbitrary units (p Ͻ 0.05 compared with normal cells; Fig. 7D). Inhibition of sAC with 25 M KH7 (Fig. 7E)  H89 inhibits PKA phosphorylation, but not the dissociation of the catalytic and regulatory unit of PKA, thereby still allowing FRET to occur (37). As with the CBF measurements, we repeated experiments for evaluating FRET changes with solutions containing different anions to assure that the bicarbonate effects were specific. Exchanging the solutions between 25 mM sodium chloride and sodium gluconate (n ϭ 14, two different lungs) did not reveal a significant change in FRET ratio upon the changes in salt solution perfusion. The solvent Me 2 SO (1‰) did not Ϫ /CO 2 . ⌬⌬FRET-RATIO is the difference between the ⌬FRET-RATIO before and after treating the cells with CFTR inh 172, KH7, or H89 upon apical perfusion with 25 mM HCO 3 Ϫ /CO 2 . A negative number indicates a reduced increase of the intracellular cAMP concentration. ⌬⌬FRET-RATIO upon CFTR inhibition with 20 M CFTR inh 172 is significantly reduced in normal cells, whereas no effect of CFTR inh 172 can be seen on CF cells (D). ⌬⌬FRET-RATIO upon sAC inhibition with 25 M KH7 is significantly reduced in normal cells, whereas no effect of KH7 is seen on CF cells (E). ⌬⌬FRET-RATIO upon PKA inhibition with 10 M H89 is neither affected in normal cells nor in CF cells, where ⌬⌬FRET-RATIO continues to stay lower than in normal cells (F). In general, inhibition with CFTR inh 172, KH7, and H89 in normal cells leads to a CF phenotypic behavior of CBF and most FRET changes (exception is as expected H89) upon apical exposure to bicarbonate. *, p Ͻ 0.05.

TABLE 1 Solution components
Concentrations of solution components are given in mM. Solutions 1 and 2 are for non-permeabilized cells, and solutions 3, 4, and 5 are for permeabilized cells. pH adjustment to an extracellular pH of 7.4 for nonpermeabilized cells were done with NaOH for solution 1 and with NaHCO 3 for solution 2. pH adjustment to an intracellular pH of 7.2 for permeabilized cells were done with KOH for solutions 3 and 5 and with KHCO 3 for 4. Solutions containing HCO 3 Ϫ were equilibrated with 5%CO 2 , 95%O 2 ; all others were with air. CrP, creatine phosphate disodium.

DISCUSSION
Our data demonstrate that apical bicarbonate influx requires functional CFTR and that this influx can be important for the regulation of ciliary beating in airway epithelial cells. This finding may shed new light on the discussion about the importance of CBF regulation in the development of mucociliary dysfunction in patients with cystic fibrosis, a topic usually dominated by the discussion of periciliary fluid depletion.
Cyclic AMP is one of the major regulators of CBF (38). Until recently, cAMP in airway epithelial cells was mainly thought to be produced by tmAC. Apically localized tmACs are G-proteinresponsive producers of cAMP. Despite microdomains of cAMP signaling, this cAMP is likely available to cilia; exogenous ␤-adrenergic agonists increase cAMP concentrations over stimulation of ␤ 2 -receptors and endogenous, and apically released ATP is hydrolyzed to adenosine, stimulating A2b receptors to produce cAMP (39), all via G-protein coupling to tmAC. ATP can also release intracellular calcium by stimulation of P 2 Y 2 receptors and thereby activate calcium-sensitive apical tmACs to produce cAMP. The intraciliary production of cAMP is even more important because calcium-mediated CBF increase is also dependent on the presence of at least small amounts of cAMP (38). The ciliary target for cAMP-associated CBF stimulation is PKA (38), which phosphorylates an outer dynein light chain in the axoneme (28). In contrast to tmACs located at the apical membrane (but not on the ciliary membrane), sAC is localized to human cilia (6), where it produces cAMP in close proximity to the dynein arms (Fig. 9). The activation of sAC by HCO 3 Ϫ / CO 2 is well established (2)(3)(4)40), and the sAC-mediated production of cAMP with its subsequent CBF effect has been shown using FRET-based estimation of intracellular cAMP (6). The observed changes in CBF and FRET upon HCO 3 Ϫ /CO 2 exposure were blocked by the sAC-specific inhibitors KH7 (35,36) and 2-HD-17␤ estradiol (41), whereas the tmAC specific inhibitor SQ 22536 had no effect (6). Here we dissect the proposed sAC activation pathway by using inhibitors at different steps of the pathway. CFTR inh 172 inhibits the transport of bicarbonate into normal cells as shown in Ussing chamber experiments (Fig. 1). KH7 (sAC inhibitor) blocks HCO 3 Ϫ /CO 2mediated changes in CBF and FRET, and H89, a catalytic PKA inhibitor, blocks HCO 3 Ϫ /CO 2 -mediated changes in CBF but not FRET, because the dissociation of the catalytic and regulatory unit of PKA is not inhibited, thereby allowing FRET to occur (37). Furthermore, CFTR inh 172 changes the CBF and FRET response of normal cells to apical HCO 3 Ϫ /CO 2 stimulation into a CF-like pattern. Together, these results support our hypothesis of disturbed CBF response upon apical HCO 3 Ϫ /CO 2 perfusion in CF cells, based on missing bicarbonate transport medi-FIGURE 8. Co-expression of fluorescently tagged PKA subunits in fully differentiated ciliated human airway epithelial cells obtained from normal subjects and patients with CF. The cells were infected in their undifferentiated state with lentiviral constructs driving the expression of the proteins under the ciliated cell-specific foxj1 promoter. Cilia were labeled with antiacetylated ␣-tubulin antibodies (AC. TUB) and an Alexa Fluor-555-coupled secondary antibody (red). The nuclei were stained with DAPI (pseudo-colored in white). Z-axis reconstructions and XY cuts through different levels of the cultures corresponding to cilia, apical, and basal cell compartments are shown. Both CFP-and YFP-labeled PKA subunits were mainly localized to the apical compartment of the cells, and they can be also seen inside cilia. There is significant overlap in RII-CFP and CAT-YFP expression in ciliated cells, allowing FRET measurements. There is similar expression of the fusion proteins in normal and CF cells. The bar represents 25 m. . Model of CBF regulation via the major effectors in the human airways, including cAMP, calcium, and pH. cAMP is produced in cilia by sAC. sAC is stimulated by HCO 3 Ϫ , which can enter the apical side of the cells, requiring functional CFTR. sAC can also increase apical bicarbonate conductance via PKA-mediated activation of CFTR. The effector of cAMP at the axoneme is PKA, which phosphorylates a dynein light chain, making the cilia beat faster. Calcium is mainly released from intracellular stores upon different stimulation, e.g. via G-protein-coupled receptors (GP) such as the purinergic receptor P 2 Y 2 . Calcium diffuses to the cilia, where it has a direct activating effect on the axoneme. There are interactions between calcium, sAC, and cAMP. sAC is stimulated by calcium, and calcium requires small amounts of cAMP for its action at the cilium. Intracellular pH influences CBF, probably over a direct action on the axoneme. Additionally, a decrease in pH increases the conductance of CFTR, which could in turn further stimulate sAC and increase CBF if more HCO 3 Ϫ is available.
ated by CFTR (either directly or indirectly). The importance of sAC for CBF and possibly other cellular functions is highlighted by the recent finding that chloride and bicarbonate conductance through CFTR is increased during states of low intracellular pH (42), allowing sAC activation to counteract the negative CBF effect of low pH even more. The fact that CFTR conductance is also increased by sAC itself, at least in airway epithelial cell lines (22), makes this rescue system a powerful, self-enhancing circuit. Bicarbonate comes from two major sources in the airways. Submucosal glands secrete bicarbonate as shown in pig airways (10,43). The secretion of bicarbonate is increased by IL-17A (44), a T-cell product. Production of IL-17A in the airways is increased by infections with Pseudomonas aeruginosa (45) and Klebsiella pneumonia (46), suggesting increased airway content of bicarbonate during airway disease exacerbations. The second source of airway bicarbonate comes from the reaction of H 2 O and CO 2 . The airway concentration of CO 2 can be significantly increased secondary to hypoventilation in airway diseases, because it occurs during exacerbations of asthma, chronic obstructive lung disease (COPD), CF, and pulmonary infections. The spontaneous balance between H 2 O/CO 2 and HCO 3 Ϫ is shifted to the bicarbonate side by secretion of carbonic anhydrase from serous acinar and ductal cells, as has been shown in human bronchial and tracheal submucosal glands (11) in the airways of rats (47) and guinea pigs (48). Altogether, these mechanisms indicate that there is increased availability of bicarbonate in the airway lumen during exacerbation of airway diseases. CFTR is a channel that allows bidirectional chloride and bicarbonate flux, the direction of which will depend on the driving force. Different conductance directions for chloride in sweat glands and in the airways exemplify this principle. The above mentioned increased luminal bicarbonate content during airway disease exacerbations may be a driving force for its conductance into airway epithelial cells. Submucosal glands can produce 20 meq/liter bicarbonate (10,43), which may even be increased by intraluminar production as mentioned above. Intracellular bicarbonate concentrations are described between 4 and 12 mM in different organs, depending on the predominant acid-base status (49 -51). Based on these concentrations, our hypothesis is valid, at least during disease exacerbations, when the intracellular bicarbonate concentration is lower, and the extracellular concentration is higher. Once in the cell, bicarbonate stimulates sAC, thereby producing cAMP with a subsequent increase in CBF and mucociliary clearance during non-CF airway disease exacerbations. In patients with cystic fibrosis, however, this system is disturbed at two levels based on malfunctioning CFTR: at the secretion site of bicarbonate in submucosal glands and at the absorption site of the apical membrane of ciliated cells.
Based only on one in vitro study examining solely base-line CBF, the function of cilia has been thought to be normal in CF until depletion of periciliary fluid impedes ciliary beating (52). Nourished by the development of the beta epithelial sodium channel (␤-ENaC) overexpressing mouse as a model for cystic fibrosis (53), depletion of airway surface liquid is thought to be the major reason for airway disease in cystic fibrosis (54), and this fact is not disputed by our results. Our data, however, add a ciliary mechanism by which mucociliary dysfunction is worsened at least during disease exacerbation. The fact that bicarbonate is not secreted from submucosal glands and cannot enter the ciliated cells at times when ciliary beating may need it most puts a break on an important rescue mechanism of mucociliary clearance. This is not only a problem once the airways are battling recurrent infections but also affects newborn patients with cystic fibrosis. During the initial infections of the airways, the ciliary machinery cannot counteract the slowing effect of the airway acidification. This, in addition to the altered airway surface liquid, leads to impaired mucociliary clearance and consecutive worsening of the airway inflammation.
One of the difficulties of the performed experiments is the fact that CBF is influenced by bicarbonate via sAC-mediated cAMP (6) and by changes in intracellular pH (32) that in return are also influenced by bicarbonate. In a previous study, we approached this problem by saturating the intracellular compartment with CO 2 by apical HCO 3 Ϫ /CO 2 exposure before perfusing the permeabilized basolateral membrane with HCO 3 Ϫ / CO 2 to increase the intracellular bicarbonate content (6). Here, we measure intracellular pH with BCECF, using ratiometric fluorescence microscopy, to compare the influence of apical HCO 3 Ϫ /CO 2 perfusion on the intracellular pH in CF and normal airway epithelial cells. These experiments show that the pH in CF cells in fact drops significantly more than in regular cells, confirming our hypothesis that bicarbonate entry is dysfunctional in CF cells secondary to abnormal CFTR, whereas transmembranous CO 2 diffusion is not impaired. The fact that the inhibition of PKA with H89 does significantly decrease CBF in normal cells upon exposure to HCO 3 Ϫ /CO 2 without affecting CF cells supports the hypothesis that the observed pH change is not the only factor influencing CBF under these conditions.
In summary, bicarbonate-stimulated cAMP production by sAC increases ciliary beat frequency and may serve as an important rescue mechanism of mucociliary clearance during exacerbations of airway diseases. The dependence of this pathway on functioning CFTR suggests that this compensation mechanism is defective in patients with CF, thereby possibly contributing to mucociliary dysfunction in this disease.