Primary airway epithelial cells grown in air-liquid interface differentiate into cultures that resemble native epithelium morphologically, express ion transport similar to those in vivo, and secrete cytokines in response to stimuli. Comparisons of cultures derived from normal and cystic fibrosis (CF) individuals are difficult to interpret due to genetic differences besides CFTR. The recently discovered CFTR inhibitor, CFTRinh-172, was used to create a CF model with its own control to test if loss of CFTR-Cl− conductance alone was sufficient to initiate the CF inflammatory response. Continuous inhibition of CFTR-Cl− conductance for 3–5 days resulted in significant increase in IL-8 secretion at basal (P = 0.006) and in response to 109 Pseudomonas (P = 0.0001), a fourfold decrease in Smad3 expression (P = 0.02), a threefold increase in RhoA expression, and increased NF-κB nuclear translocation upon TNF-α/IL-1β stimulation (P < 0.000001). CFTR inhibition by CFTRinh-172 over this period does not increase epithelial sodium channel activity, so lack of Cl− conductance alone can mimic the inflammatory CF phenotype. CFTRinh-172 does not affect IL-8, IL-6, or granulocyte/macrophage colony-stimulating factor secretion in two CF phenotype immortalized cell lines: 9/HTEo− pCEP-R and 16HBE14o− AS, or IL-8 secretion in primary CF cells, and inhibitor withdrawal abolishes the increased response, so CFTRinh-172 effects on cytokines are not direct. Five-day treatment with CFTRinh-172 does not affect cells deleteriously as evidenced by lactate dehydrogenase, trypan blue, ciliary activity, electron micrograph histology, and inhibition reversibility. Our results support the hypothesis that lack of CFTR activity is responsible for the onset of the inflammatory cascade in the CF lung.
- cystic fibrosis transmembrane conductance regulator
- epithelial sodium channel
even though the primary defect in cystic fibrosis (CF) is the impaired chloride conductance, the CF patient in the end dies as a consequence of the relentless, progressive lung infection and inflammation, mainly due to Pseudomonas aeruginosa. The relationship between mutant CFTR (defective chloride transport) and the exuberant inflammation/infection in the airways of CF patients has been controversial. Data have been presented (and refuted) indicating that defective CFTR (impaired Cl− secretion) causes abnormalities in the airway surface liquid (ASL), and this in turn reduces the bactericidal activity of ASL, therefore promoting infection (24, 30). Alternatively, the “isotonic, low ASL volume” hypothesis proposes that defective CFTR causes a decrease in ASL volume by direct or indirect interaction of CFTR with the epithelial sodium channel (ENaC), which in turn impairs mucociliary clearance and promotes infection (16, 26). It is not yet clear whether both defects, impaired Cl− secretion and increased Na+ absorption, are required and responsible for the onset of the inflammatory response observed in CF patients or whether one or the other is sufficient. Increased sodium absorption could accompany the absence of CFTR but have no direct involvement in the inflammatory response.
Recently, Mall et al. (15), using a mouse model that overexpresses the Scnn1b subunit of ENaC, have shown that overexpression of this subunit alone (without impairment of chloride conductance) is sufficient to mimic much of the lung pathophysiology observed in CF, thus supporting the idea that increased activity of ENaC leading to reduction in ASL volume accounts for most of the CF lung pathophysiology. On the other hand, Thiagarajah et al. (28), using pig submucosal airway glands and human bronchi, have shown that loss of CFTR activity alone is sufficient to reduce airway submucosal gland fluid secretion and increase fluid viscosity and protein concentration, with no apparent change in either [Na+] or [Cl−]. They speculated that this decreased fluid secretion and increased fluid viscosity is the cause of the initial mucus plugging and decreased mucociliary clearance seen in CF airways. Also, Salinas et al. (22) found submucosal gland fluid hyperviscosity in biopsies from CF subjects with minimal or no disease, suggesting defective submucosal gland function as an intrinsic defect in CF.
Determination of which of these theories is correct is important since the requirements for the restoration of normal ionic and metabolic milieu and function will vary accordingly. The system used to evaluate these hypotheses is critical. The best model for such studies would be one that closely resembles the native epithelium, yet is isolated from the influences of infection and inflammatory cells, at least at the outset. Airway epithelial cells grown on filters at the air-liquid interface differentiate into cultures that resemble native epithelium morphologically, have cilia, goblet cells, and basal cells, secrete mucus and transport it via the cilia, express ion transport systems that closely resemble those observed in vivo, and secrete matrix material, proteases, and cytokines in response to appropriate stimuli. However, cultures developed from different individuals differ at many genetic loci besides the CF gene, which increases variability and makes it difficult to ascribe responses unambiguously to defective CFTR function rather than some other (less obvious or controlled) genetic difference. Identification of CFTRinh-172 (14, 17), a selective inhibitor of CFTR, makes it possible to inhibit CFTR continuously and thereby create a CF model with its own control (cultures treated with vehicle alone). We have used this system to examine the hypothesis that absence of CFTR function by itself is sufficient to mimic the inflammatory profile observed in the airways of CF patients and therefore responsible for the onset of the inflammatory response in CF.
Human tracheal epithelial cells (HTE) recovered from necropsy specimens, as previously described (5, 10), and with the approval from the University Hospitals of Cleveland Institutional Review Board (IRB exemption EM-03-01), were grown in an air-liquid interface (ALI) on collagen-coated, semipermeable membranes (either 7 × 106 cells/4.5 cm2 filter or 1 × 106 cells/1 cm2 filter, Transwell-clear polyester membrane; Costar, Corning, NY) as previously described (8) and allowed to differentiate in serum-containing media for 3 or 4 wk.
At 3 or 4 wk, on day 0, cells were switched to submerged culture (liquid-liquid interface, LLI) and treated with either DMSO 1:1,000 (vehicle control, normal cells; Sigma, St. Louis, MO) or 20 μM CFTRinh-172 (synthesized as described, Ref. 18) prepared in DMSO and diluted from a 1:1,000 stock. Drugs were added to both the basolateral (1- or 2-ml volume, according to filter size) and the apical side (0.35 or 1.5 ml), and media was replenished every 24 h. At day 3 or 5, cells were committed to either electrical measurements or stimulated with diverse challenges as indicated below. Cells committed to P. aeruginosa strain O1 (PAO1) stimulation were switched to serum-free media 24 h before PAO1 stimulation and kept in serum-free media until the end of the experiment, and media was replenished every 12 h during that time. Serum-free media contained 1:1 DMEM-Ham's F-12, pH 7.2, 2.5 mM l-glutamine, 100 units·100 μg−1·ml−1 penicillin/streptomycin, 50 μg/ml gentamicin, 1.25 μg/ml amphotericin B (Gibco/Invitrogen, Carlsbad, CA), 2 μg/ml fluconazole (Diflucan, Pfizer), 5 μg/ml transferrin, 5 μM hydrocortisone, 5 μg/ml insulin, 20 μg/ml endothelial cell growth supplement, and 1 mg/ml BSA (Sigma). Serum-containing media had the same antibiotics present as the serum-free media. Antibiotics were not present during PAO1 stimulation.
9/HTEo− pCEP and pCEP-R Cell Lines
The development and maintenance of this matched pair of human tracheal epithelial cells derived from simian virus 40-transformed human tracheal epithelial cells (9/HTEo−, kindly provided by Dieter Gruenert, Univ. of California, San Francisco) have been described previously (20). Cells transfected with the R domain of CFTR, pCEP-R, retain expression of endogenous CFTR mRNA (12) but have completely inhibited cAMP-stimulated chloride transport while retaining Ca2+-stimulated chloride transport. These cells display increased levels of aGM1, increased P. aeruginosa binding, and increased cytokine production in response to PAO1 (4, 12, 13, 19). Cells were conditioned to grow in media without serum for three passages before use in these experiments [DMEM:Nutrient Mixture F-12 (DMEM/F-12), 1:1, pH 7.2, 2.5 μM l-glutamine, and 50 μg/ml hygromycin were from Gibco/Invitrogen and 5 μg/ml transferrin, 5 μM hydrocortisone, 25 ng/ml epidermal growth factor, 25 mM HEPES (DMEM/F-12 already contains 12 mM HEPES and 2.438 g/l NaHCO3), and 1 mg/ml BSA were from Sigma]. Cells were plated on collagen-coated 24-well plates at 0.3 × 106 cells/well and allowed to reach confluency, and then cells were treated with either DMSO (1:1,000) or 10 μM CFTRinh-172 every 12 h for 3 days and subsequently stimulated with PAO1 as indicated below.
16HBE14o− S and AS Cell Lines
The development and maintenance of these cell lines have been described previously. Sense and antisense constructs containing the first 131 nucleotides of human CFTR were transfected into 16HBE14o− cells (provided by D. Gruenert), which formed tight junctions. The antisense (16HBE14o− AS) cell line has no cAMP-stimulated chloride efflux in response to forskolin and is deemed CF phenotype. The sense transfected cell line (16HBE14o− S) retains the cAMP-stimulated chloride response and is the normal phenotype matched line (13, 21). The antisense cells display increased cytokine production in response to PAO1. Cells were plated on collagen-coated 24-well plates (0.6 × 106 cells/well for AS and 0.3 × 106 cells/well for S) and allowed to reach confluency, and then cells were treated with either DMSO (1:1,000) or 10 μM CFTRinh-172 every 12 h for 3 days and subsequently stimulated with PAO1 as indicated below.
Four-week-old cells plated on 1-cm2 collagen-coated filters were kept for 3 days in either ALI or LLI (2 filters/condition) and subsequently mounted in a temperature-controlled Ussing chamber that allowed for the separate perfusion of the apical and basolateral compartments with a Krebs-Ringer bicarbonate solution as previously described (29). At 1-min intervals, the transepithelial voltage was clamped to a near-zero value to measure transepithelial resistance. After baseline recording, amiloride (100 μM, Sigma) was added to apical side, followed by addition of forskolin/IBMX (F/I; 10/100 μM, respectively, Sigma) to basal side, and then CFTRinh-172 was added to either apical or basal side at either 1 or 5 μM.
Delta Conductance Calculation
To determine minimum concentration of CFTRinh-172 and duration of effect for continuous CFTR inhibition while keeping CFTRinh-172 present all the time, so as not to confound the issue with the reversibility of CFTRinh-172, transepithelial voltage (VT) and transepithelial resistance (RT) were measured using the EVOM Epithelial Voltohmmeter (World Precision Instruments, Sarasota, FL) inside an Isotemp 500 series incubator from Fisher (Pittsburgh, PA) so we could control temperature and CO2 (37°C, 5% CO2) during measurement. VT and RT as measured were used to calculate transepithelial conductance (GT) and delta transepithelial conductance (ΔGT). In high-resistance epithelia like these, paracellular conductance is low, and changes in apical conductance are readily measured. Four-week-old cells grown in ALI were switched to LLI for 3 or 5 days in the presence of DMSO 1:1,000 (control vehicle, normal cells) or either 5 μM or 20 μM CFTRinh-172, all prepared in DMSO and all diluted from a 1:1,000 stock. Fresh media was added for both DMSO- and 5 μM CFTRinh-172-treated cells every 6 or 24 h. For the 20 μM CFTRinh-172-treated cells and their controls, media was replenished every 6, 12, or 24 h. Measurements were always done in parallel from same-donor DMSO-treated cells and CFTRinh-172-treated cells. The resistance and voltage were measured every minute for 50 min. All drugs were added from the apical side. The first 15 min were a baseline, after which 100 μM amiloride was added, and measurements were recorded for 10 min. A cocktail of 10 μM forskolin plus 100 μM IBMX was then added and recorded for 15 min. Finally, an acute addition of 5 or 10 μM CFTRinh-172 was added and recorded for 10 min. Values are expressed as the changes in conductance seen on amiloride (ΔGAmil), forskolin/IBMX (ΔGF/I), and CFTRinh-172 acute addition (ΔGCFTRinh-172).
The laboratory strain of PAO1 (kindly supplied by Alice Prince) was streaked into a MacConkey agar plate and allowed to grow overnight. A single colony was transferred to 4 ml of the same media in which cells were grown, incubated overnight at 37°C, and transferred to 100 ml of the same media until an optical density of 600 nm of 1.0 was reached [∼1 × 109 colony-forming units (CFU)/ml]. PAO1 was washed twice with HBSS before being resuspended to the final dilution. After 3 or 5 days of growing cells in the presence of either DMSO or 20 μM CFTRinh-172, cells were washed twice with HBSS, exposed to either HBSS at one dose (1 × 109 CFU/ml) or a dose curve (1 × 109 to 1 × 105 CFU/ml) of PAO1, or TNF-α plus IL-1β (either 100 ng/ml or 1 μg/ml for each) for 1 h at 37°C, and washed three times with HBSS and growth media containing 100 μg/ml tobramycin (Sigma) or 100 μg/ml gentamicin (Gibco/Invitrogen) plus DMSO or CFTRinh-172. Media was collected for cytokine assay every 12 h.
Media were collected separately from the apical and basolateral compartments from each filter of well-differentiated cells or from the 24-well plates for the immortalized cell lines and frozen at −80°C until they were assayed for IL-8, IL-6, and granulocyte/macrophage colony-stimulating factor (GM-CSF) using ELISA kits (R&D Systems, Minneapolis, MN). Protein was not measured in the majority of cases since cell lysates were destined for other experiments, so values are expressed as picograms per milliliter of medium.
Three-week-old cells grown in ALI were switched to LLI in serum-containing media for 2 days in the presence of either DMSO (1:1,000) or 20 μM CFTRinh-172 (15 filters/condition from 4 different donors), with media replaced every 24 h. Cells were dissociated from filters by exposing them to 0.02% trypsin (E-pet; Biofluids, Rockville, MD) for 30 min at room temperature (RT), which were then neutralized with soybean trypsin inhibitor (Biofluids). Dissociated cells from five filters/condition were pooled and replated in two Lab-Tek II glass chamber slides (Rochester, NY), four chambers each, and allowed to attach for 24 h in the continued presence of either DMSO or CFTRinh-172. This maneuver was necessary to visualize the nuclei clearly without interference from other layers of cells in the pseudostratified epithelium.
Cells were then stimulated with TNF-α/IL-1β (100 ng/ml each) for 15 min at 37°C or treated with HBSS. Immediately, cells were washed three times with PBS (without calcium or magnesium, pH 7.4), fixed with cold methanol for 15 min at RT, washed three times in PBS, permeabilized with 0.05% saponin in the presence of 0.1% ovalbumin and 0.02% azide for 15 min at RT, washed three times with PBS, and blocked with 1% BSA for 30 min at RT. Cells were then incubated for 1 h at RT with either no primary antibody (control) or 1:50 NF-κB p65 (mouse; BD Transduction Laboratories, San Jose, CA). After being washed, cells were incubated with the secondary antibody, Alexa Fluor 488 goat anti-mouse (1:50; Molecular Probes, Eugene, OR), for 1 h at RT. Cells were washed six times with PBS, 15 min per wash, stained with Hoechst 33258 (50 ng/ml) for 2 min to label the nuclei, and washed three times with PBS.
Confocal Scanning Laser Microscopy and Analysis of Fluorescence Images
All confocal images were acquired using a Zeiss LSM 510 META inverted laser-scanning confocal microscope in the Ireland Comprehensive Cancer Center Confocal Microscopy Core Facility at Case Western Reserve University and University Hospitals with a ×63 numerical aperture of 1.4 oil-immersion plan-apochromat objective. For NF-κB p65, images of Alexa Fluor 488 were collected using a 488-nm excitation light from an argon laser, a 488-nm dichroic mirror, and a 500- to 550-nm band pass barrier filter. All Hoechst-stained nuclear images were collected using a Coherent Mira-F-V5-XW-220 (Verdi 5W) Ti:sapphire laser tuned at 750 nm, a 700-nm dichroic mirror, and a 390- to 465-nm band pass barrier filter. Images were collected as 12-bit data with a gray scale ranging from 0 to 4,690.
From each treatment, five to seven Z-sections, collected at random under same-power conditions, were transferred into the MetaMorph Imaging System software (Universal Imaging/Molecular Devices, Downingtown, PA) for quantitative analysis. Hoechst-stained nuclei were used to determine the area to be digitized in cells labeled with p65 or control (no primary antibodies). This allowed us to monitor the appearance (movement) of p65 into the nuclei under the diverse conditions examined. Fluorescence in each of the nuclei was digitized, and an average gray value for each nucleus was determined. Initial comparisons were made between cells from same donor; eventually, all the data were pooled and shown as means ± SE.
Three-week-old cells grown in ALI were switched to LLI in serum-containing media for 4 days in the presence of either DMSO (1:1,000) or 20 μM CFTRinh-172, and media were replaced every 6 h. Cells were then lysed in 100 μl of ice-cold lysis buffer [50 mM Tris, pH 7.5, 1% Triton X-100, 100 mM NaCl, 50 mM NaF, 200 μM Na3VO4, 10 μg/ml pepstatin and leupeptin (Sigma)] for 30 min at 4°C. Plates were scraped to suspend cells, and cells were transferred to 1-ml microcentrifuge tubes and centrifuged for 10 min at 14,000 rpm at 4°C. The supernatant was extracted and frozen at −80°C for at least 24 h. Protein concentration of samples was measured using the Bio-Rad Dc protein assay system (Bio-Rad, Hercules, CA). Proteins were separated using SDS-PAGE through either a 7.5% (for Smad3) or 12% (for RhoA) acrylamide gel. Gels were transferred to Immobilon-P membrane (Millipore, Bedford, MA). Blots were blocked in PBS solution mixed with 5% nonfat dehydrated milk and 0.1% Tween 20 (Sigma) overnight at 4°C. Antibodies against Smad3 (rabbit), RhoA (mouse), and Erk1 (rabbit) were obtained from Santa Cruz Biotechnologies (Santa Cruz, CA). Erk1 was used as a control for protein loading. Blots were incubated with 1:1,000 dilution Smad3 for 3 h or RhoA (1:1,000) for 1 h at RT and washed three times with PBS with 0.1% Tween 20. Secondary antibody conjugated to horseradish peroxidase (1:4,000 dilution) was added to the blots for 1 h at RT and washed at least three times with PBS with 0.1% Tween 20 (Sigma). Signal was visualized by incubating with SuperSignal chemiluminescent substrate (Pierce, Rockford, IL) for 8 min at RT. The membranes were exposed to Fuji scientific imaging film (Fuji, Tokyo, Japan).
HTE cells, 4-wk-old, grown in ALI, were switched to LLI for 5 days in the presence of nothing added, DMSO, or 20 μM CFTRinh-172 (media replaced every 24 h), at the end of which cells were fixed for electron micrographs (EM) (1 filter/condition). As a control, a filter that was kept in ALI with nothing added to the media was also fixed. To preserve the mucous layer, a nonaqueous method was used to fix the cells as previously described (23). Briefly, cells were washed three times with perfluorocarbon, 10 min each (Fluorinert; 3M Industrial Chemical Products, St. Paul, MN), incubated for 1 h at RT with 1% osmium tetroxide (Sigma) prepared in perfluorocarbon, washed again three times with perfluorcarbon, 10 min each, and kept in perfluorocarbon until they were embedded in resin. Three grids/condition were prepared using standard procedures. Grids were examined in their entirety with a scanning electron microscope at magnifications specified, and four to five micrographs were taken for each one.
Cell viability after CFTRinh-172 and DMSO treatment was determined by lactate dehydrogenase (LDH) release in the media and by trypan blue exclusion. Filters from two different culture preparations were treated with either DMSO or 20 μM CFTRinh-172 for 5 days in LLI (serum-containing media replaced every 24 h). LDH activity was measured in the supernatant and cell lysates of 8 DMSO-treated filters and 10 CFTRinh-172-treated filters following the instructions in the TOX-7 kit from Sigma. For the trypan blue exclusion assay, five DMSO-treated filters and four CFTRinh-172-treated filters were trypsinized for 15 min at 37°C, resuspended in PBS, incubated with 0.4% trypan blue (prepared in PBS) for 10 min, and then counted. For both assays, four filters kept in LLI in the absence of either DMSO or CFTRinh-172 were examined as a further control.
Data are expressed as means ± SE. Values between treatments were compared using Student's t-test. A probability level of P < 0.05 was considered significant.
CFTR Inhibition by CFTRinh-172
Short-circuit measurements were performed to determine the electrical properties of monolayers grown in LLI and ALI culture conditions and the lack of difference between the two preparations. Under short-circuit conditions (no serum present), 1 μM CFTRinh-172 added to the apical side of 4-wk-old HTE cells grown in serum-containing media greatly inhibits CFTR-dependent Cl− conductance. A final concentration of 5 μM is necessary to completely abolish CFTR activity acutely from the apical side (Fig. 1A′, LLI, and 1C′, ALI). CFTR-dependent Cl− conductance was not inhibited by the addition of either 1 or 5 μM CFTRinh-172 to the basal side; in these preparations, addition of 3.5 μM CFTRinh-172 to apical side inhibited CFTR (Fig. 1B′, LLI, and 1D′, ALI). Since CFTR inhibition by CFTRinh-172 is reversible, to establish a “CF phenotype,” CFTRinh-172 will need to be present constantly. To preserve ALI culture conditions, initially CFTRinh-172 was added to the basal side (1- or 2-ml total volume) and to the apical side (50-μl total volume), but this volume was quickly absorbed, and the apical concentration of CFTRinh-172 was difficult to control. For this reason, cultures were converted from ALI to LLI. Figure 1, B, B′, D, and D′, indicate that after differentiation has been established by growing cells in ALI for 4 wk, switching cells to LLI for the duration of the experiment (in this case 3 days) does not alter the responses of the cells. In separate experiments (data not shown), we have shown that the apparent initial difference in sodium absorption between the two systems (Fig. 1, A and B vs. C and D) is due to the change in temperature, not to any actual difference in sodium absorption between the two systems.
Dose and treatment interval experiments were performed to determine the minimum concentration of CFTRinh-172 and frequency of drug addition necessary for the continuous inhibition of CFTR in the presence and absence of serum. To access apical membrane sodium conductance and cAMP-stimulated Cl− conductance of the monolayer without changing media, measurements of transepithelial voltage and resistance were made under static conditions (see methods). After 4 wk on ALI, cells were grown for 3 days in LLI in the presence of DMSO 1:1,000 (control vehicle, normal cells) or either 5 μM or 20 μM CFTRinh-172, and media were replenished every 6, 12, or 24 h. Three filters each were treated with either DMSO or 5 μM CFTRinh-172 for a short period of time, 1 h. Figure 2A indicates that in serum-containing media, 5 μM CFTRinh-172 nearly completely inhibits CFTR at 1 h, but after 6 h, the fractional inhibition of CFTR is only ∼50%. After 24 h of treatment with 5 μM CFTRinh-172, inhibition of CFTR is only ∼20%, probably due to binding to plastic walls. In contrast, 20 μM CFTRinh-172 inhibits CFTR when administered as infrequently as every 24 h (Fig. 2A). A trace of the conductance changes for one of those filters treated with 20 μM CFTRinh-172 every 24 h as well as the trace for a filter treated with DMSO are shown in Fig. 2B. It should be noted that the delta conductance for F/I (ΔGF/I) response varies depending on the donor, but whatever the starting level of CFTR activity is, CFTRinh-172 is able to inhibit it, and the comparisons (dose and time of exposure) were always done with cells from same donor. The residual conductance observed in cells treated with 5 μM CFTRinh-172 every 6 or 24 h is indeed CFTR dependent since the acute addition of 10 μM CFTRinh-172 at the end of the experiment is able to abolish it (data not shown).
For certain experiments, cells need to be grown in the absence of serum. Since preliminary experiments indicate that CFTRinh-172 binds to serum proteins, most likely albumin, we calculated that 0.1% of added BSA would mimic the serum-buffering capacity. HTE cells, 3-wk-old, were placed in LLI in serum-free media with or without 0.1% BSA (1 filter/condition) for ∼5 h. After amiloride and F/I addition, increasing concentrations of CFTRinh-172 were added. In the presence of 0.1% BSA, neither 1 nor 5 μM CFTRinh-172 completely inhibited CFTR, but both 10 and 20 μM fully inhibited CFTR (Fig. 2C). To determine the concentration of CFTRinh-172 in serum-free media plus 0.1% BSA that would maintain continuous inhibition, 3-wk-old HTE cells were switched to LLI and treated for 5 days with either DMSO (1:1,000) or 10 μM CFTRinh-172 (replenished every 24 h) or 20 μM CFTRinh-172 (replenished every 12 or 24 h). Figure 2D shows %inhibition of CFTR-dependent Cl− conductance by CFTRinh-172. To maintain continuous inhibition of CFTR in this media, 20 μM CFTRinh-172 must be administered every 12 h (Fig. 2D).
There was no permanent change in the electrical properties of the well-differentiated cells following these treatments. CFTR activity was restored to 59 ± 4.6% of baseline 1 h after removal of CFTRinh-172 (n = 3) and to 70 ± 1.4% of baseline by 4 h (n = 3). Treated cells were washed and placed back in ALI without drugs, and after 7 days, basal resistance values (1.22 for CFTRinh-172-treated cells and 1.27 for DMSO-treated cells, kohmscm2) were very similar to those observed before the experiment (1.14 for CFTRinh-172-treated cells and 1.12 for DMSO-treated cells, kohmscm2). The same set of cells can therefore be tested repeatedly.
There was also no evidence of cell damage from any of these treatments. LDH release by DMSO-treated cells was 6.1 ± 0.7% and 7.4 ± 0.4% for CFTRinh-172-treated cells, similar to LDH release in media from cells in the absence of either DMSO or CFTRinh-172 (5.8 ± 0.3%). Trypan blue staining of DMSO-treated filters indicated 90.4 ± 2.2% viability, and CFTRinh-172-treated filters indicated 85.6 ± 2.8%, not statistically different from untreated cells. EM pictures of CFTRinh-172-treated cells (Fig. 3, representative EM of each filter) show no morphological difference between any of the treatments except that the mucous layer has washed away in submerged cultures. In addition, visual inspection indicated no change in ciliary activity by CFTRinh-172 treatment.
Effects of CFTRinh-172 Treatment on Amiloride-Sensitive Conductance
We also examined the effects of CFTRinh-172 on amiloride-sensitive conductance. Figure 4A shows that CFTRinh-172 concentrations that failed to inhibit CFTR also did not affect ΔGAmil; on the other hand, CFTRinh-172 concentrations that resulted in constant CFTR inhibition resulted in a significant decrease in amiloride-sensitive conductance. This decrease remained statistically significant when treatment was extended for 5 days (data not shown). We also examined the basal resistance values of these cultures (Fig. 4B). Concentrations of CFTRinh-172 that failed to inhibit CFTR-dependent Cl− conductance also did not affect basal RT with respect to vehicle-treated cells. However, CFTRinh-172 concentrations that inhibit CFTR-dependent Cl− conductance caused an increase in basal RT, indicating either inhibition of basal CFTR-dependent Cl− secretion or decrease in sodium absorption or both.
Inflammatory Profile in Response to CFTRinh-172 Treatment
Basal cytokine secretion.
Figure 5A shows that treatment of HTE cells with 20 μM CFTRinh-172 results in an increase in IL-8 basal secretion in both serum-free and serum-containing media (P = 0.006 for basolateral secretion in serum-containing media; P = 0.035 for basolateral secretion in serum-free media). GM-CSF was also increased in the apical side of cells in serum-containing media (Fig. 5B, P = 0.04). Basal secretion of GM-CSF was below the level of detection of the assay in most of the basolateral samples for serum-containing media and in the majority of the apical and basal side for serum-free media. IL-6 secretion was below levels of detection for almost all filters in both conditions.
Stimulated cytokine secretion.
To test the effects of CFTRinh-172 treatment on cytokine production in response to P. aeruginosa by differentiated HTE cells, ALI cells were switched to LLI for 6 days, during which they were treated every 24 h with either DMSO or 20 μM CFTRinh-172. At day 4, cells were switched to serum-free media, and media were replenished every 12 h thereafter. On day 5, cells were stimulated with either PAO1 (1 × 109 to 1 × 107 CFU/ml), TNF-α plus IL-1β (100 ng/ml), or HBSS as control for 1 h. Media were collected at day 6 from the basolateral side, 12 h after stimulation; CFTRinh-172 was also present during this period. Eight filters per condition from two different culture preparations were studied. Figure 6 shows that CFTRinh-172 treatment for 5 days produced significantly increased responses of IL-8 to 109 CFU/ml PAO1 stimulation and under basal conditions. Although treatment with either 108 to 107 CFU/ml PAO1 or 100 ng/ml TNF-α plus IL-1β caused an increase in IL-8 secretion with respect to basal levels, there was no significant difference between DMSO- and CFTRinh-172-treated cells. In these experiments, GM-CSF secretion was below the levels of detection of the assay for most conditions for both DMSO- and CFTRinh-172-treated cells.
Specificity of CFTRinh-172 effect on cytokine production.
CFTRinh-172 itself does not affect cytokine production independently of its action on CFTR. This issue was addressed by 1) determining the effects of CFTRinh-172 on cytokine production in two CF phenotype immortalized cell lines, 9/HTEo− pCEP-R and 16HBE14o− AS and on primary bronchial epithelial cell cultures grown in LLI derived from bronchus of a ΔF508 homozygous CF patient; and 2) examining the reversal of inflammation after the withdrawal of CFTRinh-172 from primary airway epithelial cells. CFTR activity is already absent in 9/HTEo− pCEP-R, 16HBE14o− AS, and cells derived from a CF patient (12, 13, 19, 20), therefore, if CFTRinh-172 actions on cytokine secretion are solely due to CFTR inhibition, we do not expect to see a further increase in cytokine production, even upon PAO1 stimulation, when these cells are treated with CFTRinh-172. The pCEP-R and AS cells were treated for 3 days with DMSO or 10 μM CFTRinh-172 in serum-free media replaced every 12 h, stimulated with either HBSS (control), PAO1 109 or 107, or TNF-α plus IL-1β (100 ng/ml each) for 1 h, and washed, and media with appropriate drugs were added back and collected 12 h later for cytokine assays. Experiments were done in triplicate. Figure 7 shows that stimulation of IL-8, IL-6, and GM-CSF by PAO1 or TNF-α plus IL-1β, in these two CF cell lines, was not significantly affected by CFTRinh-172. CF primary epithelial cultures derived from harvested lung of a CF patient were treated with DMSO or 20 μM CFTRinh-172 for 3 days in serum-containing media replenished every 24 h, stimulated with either HBSS (control), PAO1 109 to 108 to 107, or TNF-α plus IL-1β (0.1 or 1 μg/ml) for 1 h and washed, and media were collected 12 h later from the basolateral side for cytokine assays; n = 3/condition. Figure 8 shows that primary CF cells treated with CFTRinh-172 did not secrete more IL-8 than CF cells treated with DMSO after either PAO1 or TNF-α plus IL-1β stimulation. Furthermore, basal IL-8 secretion was not affected by CFTRinh-172 treatment.
We also tested the effects of CFTRinh-172 on the normal counterparts of these CF phenotype immortalized cell lines: 16HBE14o− sense and 9/HTEo− pCEP cells. Since these cells do express CFTR, we expected that treatment with the inhibitor would make them CF-like and therefore exhibit an increase in cytokine secretion. Cells were treated for 3 days with DMSO or 10 μM CFTRinh-172 in serum-free media replaced every 12 h, stimulated with either HBSS (control), PAO1 108 or 107, or TNF-α plus IL-1β (100 ng/ml each) for 1 h, and washed, and media with appropriate drugs were added back and collected 12 h later for cytokine assays. Experiments were done in triplicate. Figure 9 indicates that our prediction holds true. Both cell lines showed an increase in IL-8, IL-6, and GM-CSF secretion at basal and after stimulation with either PAO1 or TNF-α/IL-1β when they were treated with CFTRinh-172. In many cases, the increase was statistically significant. It should be noted that the effect was more pronounced in the 16HBE14o− than in the 9/HTEo− cells.
To test the reversibility of the increased cytokine production, HTE cells were switched to LLI for 5–7 days, and 16 filters were treated every 24 h with either DMSO or 20 μM CFTRinh-172 (8 filters each) for 3 days, after which the inhibitor was withdrawn for 48 h (reversibility filters). At the same time, on day 3, 32 additional filters were switched to LLI and began treatment with DMSO or inhibitor (16 filters each) for 3 days (treated filters); on day 5 all filters were switched to serum-free media, and media were replenished every 12 h thereafter (with or without CFTRinh-172). On day 6, all filters were stimulated with either PAO1 (1 × 109 CFU/ml) or HBSS as control for 1 h (8 filters/condition from 2 different culture preparations). Basolateral media were collected at day 7, 12 h after stimulation. Figure 10 shows greater IL-8 secretion in CFTRinh-172-treated cells upon PAO1 stimulation than in DMSO-treated cells, although this difference did not reach statistical significance. Following withdrawal of drug, no difference in stimulated secretion was seen. Cytokine secretion increased significantly with time in cultures exposed to DMSO (data not shown).
NF-κB p65 translocation to the nucleus in response to CFTRinh-172 treatment.
Since IL-8 and IL-6 are NF-κB-regulated genes, and several laboratories have presented data indicating that NF-κB is upregulated in CF (3, 6, 7), we decided to examine the effects of CFTRinh-172 treatment on the translocation of NF-κB p65 from the cytoplasm to the nucleus of primary airway cells grown in ALI. Briefly, cells were treated with either DMSO or 20 μM CFTRinh-172 for a total of 3 days and stimulated for 15 min with TNF-α/IL-1β (100 ng/ml each). As indicated in methods, to have a single monolayer of cells and be able to quantify the fluorescence of the nuclei without the interference of other cell layers, cells needed to be dislodged from filters and placed in chamber glass slides where they were allowed to recuperate for 1 day (in the continuous presence of CFTRinh-172 or not) and then stimulated for 15 min and immediately fixed and stained. Figure 11A shows a representative image of each condition, and Fig. 11B shows the summarized data. TNF-α/IL-1β treatment was able to increase the amount of p65 translocated to the nuclei in both DMSO-treated cells (P < 0.000001) and CFTRinh-172-treated cells (P < 0.000001). At basal state, CFTRinh-172 treatment resulted in a minor decrease in the level of p65 in the nuclei, although it was statistically quite significant (P < 0.01). On the other hand, CFTRinh-172 treatment caused a statistically significant increase (P < 0.000001) in the translocation of p65 into the nucleus upon stimulation with TNF-α/IL-1β compared with the increase seen upon stimulation in the DMSO-treated cells.
These findings were corroborated using TransAM NF-κB p65 transcription factor ELISA assay kits (Active Motif, Carlsbad, CA). We prepared nuclear extracts from 3-wk-old cells grown in ALI after treating them for 3 days with either DMSO or 20 μM CFTRinh-172 cells (1 donor) and stimulating them for 20 min with 100 ng/ml each TNF-α/IL-1β. A positive control from the kit was included, and binding of NF-κB p65 was competed by using the wild-type consensus oligonucleotide also provided in the kit. Assay was done in duplicate and with cells from one donor. Mean optical density for NF-κB binding in basal state without CFTRinh-172 was 0.19; basal state + CFTRinh-172 0.29; TNF-α/IL-1β, no inhibitor 0.69; and TNF-α/IL-1β + CFTRinh-172 0.85. All binding in all cases was competed by the wild-type oligonucleotide of p65.
RhoA and Smad3 protein levels in response to CFTRinh-172 treatment.
To further test whether CFTRinh-172 treatment of well-differentiated cells results in the creation of a CF phenotype, we measured RhoA and Smad3 protein levels. Two different CF models, our 9/HTEo− pCEP-R (CF phenotype) cells and excised nasal epithelium from CFTR−/− mice, exhibit an increase in total RhoA protein expression (11). In the same two CF models, Kelley et al. (9) have shown decreased Smad3 protein.
HTE cells were grown in serum-containing media in LLI for 4 days and treated with either DMSO (1:1,000, 2 filters) or 20 μM CFTRinh-172 every 6 h. Total RhoA protein was measured at the end of the experiment. A representative blot (Fig. 12A) shows that cells treated with CFTRinh-172 (CF phenotype) for 4 days had markedly increased total levels of RhoA compared with nontreated cells (ERK1 was also measured to control for protein loading). Figure 12B shows the densitometry analysis of RhoA expression relative to ERK1 protein content (each replicate shown as a separate point). In each case, total RhoA was higher in CFTRinh-172-treated cells. Smad3 expression was measured in a separate set of filters than those used for RhoA, also treated for 4 days with 20 μM CFTRinh-172 in LLI. Figure 12C shows a blot representative of eight filters, four treated with DMSO and four with CFTRinh-172; Smad3 expression is markedly decreased in CF cells (CFTRinh-172 treated), as further seen in Fig. 12D (P = 0.02). Again, ERK1 was used to control for protein loading. Thus this experimental system replicates the CF inflammatory phenotype after 3–5 days of treatment with CFTRinh-172, at least with respect to RhoA, Smad3, and cytokine responses to PAO1.
Most studies show that there are excessive inflammatory responses in CF vs. non-CF airway epithelial cell lines, and the better matched the cell lines are and the less the perturbation involved in matching them (i.e., long-term corrected cells vs. acutely corrected with adenoviral vectors), the clearer the CF/non-CF differences. However, not all experiments show such differences. In particular, human airway epithelial cells grown in primary culture at the ALI most closely resemble epithelia in the native state. However, only some, not all, studies using this model show that CF and non-CF cells differ in their expression of inflammatory proteins (1, 2).
Several issues may be important in explaining these differences. First, it is possible that cells retrieved from severely ill (dead or dying) CF patients have been altered by the milieu from which they were retrieved (a badly inflamed, infected airway) and have acquired characteristics that are the result not of the CF gene but of the altered environment. Second, polymorphisms in genes relevant to inflammation are well known, and some of them may be relevant to the course of CF. Such polymorphisms occur in genes such as IL-8, IL-10, TNF-α, and many others and can alter the expression of the proteins by as much as eightfold (TNF) or two- to threefold (IL-10). Thus, when one compares CF and non-CF individuals, many other genetic differences besides CF may affect the measurement at hand. Thus some means of altering CFTR function in well-differentiated cultures from a single individual would be desirable, to assure a common genetic background when inflammatory responses are compared. However, careful consideration must be given to how such a matched CF/non-CF pair of cells from the same individual might be generated. Using viral vectors to deliver wild-type CFTR to CF airway cells imposes the genetic program of the virus itself as well as the changes entrained by expression of CFTR, and such studies must be very carefully controlled to be interpreted unambiguously.
Recently, Ma et al. (14) identified a low-molecular-weight inhibitor of CFTR (CFTRinh-172) that works rapidly and is reversible, voltage independent, and apparently quite specific. Patch-clamp studies showed that CFTRinh-172 reduces open channel probability without affecting single channel conductance: it increases the time the channel spends in the closed state without affecting mean open time (27). At concentrations where CFTR Cl− conductance was blocked, the inhibitor did not inhibit other Cl− channels (Ca2+-activated chloride secretion and volume-activated Cl− currents), MDR-1, ATP-K+-sensitive channels, aquaporin 1 channel, urea transporter, Na+/H+ exchanger, or Cl−/HCO3− exchanger. At a concentration of 5 μM, it did not affect cellular cAMP production or phosphatase activity. CFTRinh-172 was not toxic to FRT cells at concentrations up to 100 μM for 24 h, and it was not toxic to mice when administered twice daily (25). We have taken advantage of this inhibitor and the availability of well-differentiated cultures of human airway epithelial cells at the ALI to test whether inhibition of CFTR produces inflammatory changes observed in matched cell lines.
The well-differentiated cells have several advantages over in vivo systems: flexibility, control of experimental conditions, and greater opportunities for interventions. Another strength is that this strategy will only test the effect of inhibition of transport by CFTR. The protein is still present in the apical membrane (as shown by the immediate restoration of activity when the inhibitor is removed) so its binding to other proteins should be intact, and, unlike most CF patients' cells, there will be no excess of misprocessed CFTR. This allows us to isolate the effect of CFTR activity on the inflammatory process, an important advantage. However, one disadvantage associated with the system is the possible variability among donors. This issue was addressed by always comparing DMSO-treated cells and CFTRinh-172-treated cells from the same donor and by using as many donors as possible. In addition, we and others (2) found some basal secretion of IL-8 by these cultures, a finding not thought to replicate the in vitro situation in normal subjects whose airway epithelial cells are thought to be quiescent. Another disadvantage may be the switch to LLI to maintain the desired concentration of inhibitor. Nevertheless, the advantages of the well-differentiated cells outweigh the disadvantages.
These data show that CFTRinh-172 is capable of inhibiting CFTR activity continuously with appropriate doses at appropriate intervals. Treatment of cells for 5 days with CFTRinh-172 does not affect the cells deleteriously as evidenced by LDH assays, trypan blue exclusion, continued ciliary activity, intact histology by EM, and reversibility of inhibition. CFTR inhibition by CFTRinh-172 for as little as 4–5 days produces an inflammation profile that resembles that observed in CF patients: increase in IL-8 and GM-CSF secretion upon stimulation by PAO1; increase in IL-8 secretion at basal state; increased nuclear transport of NF-κB in response to stimulus; increased total RhoA protein levels as well as decreased total Smad3 protein levels in the basal state. Not every stimulus or condition resulted in significant increase in cytokine production. This is probably due to substantial variation from donor to donor. The effect of CFTRinh-172 on cytokine secretion seems to be mediated through its CFTR inhibition since it did not affect cytokine secretion in our two CF phenotype immortalized cell lines (9/HTEo− pCEP-R and 16HBE14o− AS), nor did it affect IL-8 secretion in primary CF cells derived from a ΔF508 homozygous CF patient. After withdrawal of the inhibitor, cells reverted to the same IL-8 secretion upon PAO1 stimulation as controls. Notably, this inflammatory profile exists in the absence of increased sodium absorption, so reduced Cl− secretion, in and of itself, is sufficient to produce the inflammation profile typically associated with CF patients in isolated airway epithelial cells. The mechanism by which active CFTR downregulates ENaC activity is unknown, so the mechanism in CF by which ENaC upregulation occurs is also not known. However, upregulation fails to occur with CFTRinh-172 in this system over 4–5 days. Possibly, the treatment period is too short for upregulation to occur.
Nevertheless, in our model, a system that closely resembles the in vivo system both morphologically and physiologically, CFTR activity is decreased and ENaC is normal or slightly decreased, yet we observe the inflammatory profile associated with CF. Therefore, decreased CFTR and not increased ENaC activity accounts for the onset of the inflammatory CF phenotype in this cell model, and increased ENaC activity is not an obligatory intermediate in the process.
This work was supported by Cystic Fibrosis Foundation Grants (DAVIS00VO, DAVISOOZO, and RDP) and NIH grants (P30-DK-27651, HL-73856, and P30-CA-43703-12).
We thank Yoshie Hervey for establishing the initial air-liquid interface protocol at Case Western Reserve University facility.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society