The pathophysiology of cystic fibrosis (CF) inflammatory lung disease is not well understood. CF airway epithelial cells respond to inflammatory stimuli with increased production of proinflammatory cytokines as a result of increased NF-κB activation. Peroxisome proliferator-activated receptor-γ (PPARγ) inhibits NF-κB activity and is reported to be reduced in CF. If PPARγ participates in regulatory dysfunction in the CF lung, perhaps PPARγ ligands might be useful therapeutically. Cell models of CF airway epithelium were used to evaluate PPARγ expression and binding to NF-κB at basal and under conditions of inflammatory stimulation by Pseudomonas aeruginosa or TNFα/IL-1β. An animal model of CF was used to evaluate the potential of PPARγ agonists as therapeutic agents in vivo. In vitro, PPARγ agonists reduced IL-8 and MMP-9 release from airway epithelial cells in response to PAO1 or TNFα/IL-1β stimulation. Less NF-κB bound to PPARγ in CF than normal cells, in two different assays; PPARγ agonists abrogated this reduction. PPARγ bound less to its target DNA sequence in CF cells. To test the importance of the reported PPARγ inactivation by phosphorylation, we observed that inhibitors of ERK, but not JNK, were synergistic with PPARγ agonists in reducing IL-8 secretion. In vivo, administration of PPARγ agonists reduced airway inflammation in response to acute infection with P. aeruginosa in CF, but not wild-type, mice. In summary, PPARγ inhibits the inflammatory response in CF, at least in part by interaction with NF-κB in airway epithelial cells. PPARγ agonists may be therapeutic in CF.
- nuclear factor-κB
- disease models
- in vitro
- lung diseases
- Pseudomonas aeruginosa
although cystic fibrosis (CF) is caused by defects in the chloride channel, cystic fibrosis transmembrane conductance regulator (CFTR), most patients succumb to respiratory infections with Pseudomonas aeruginosa and the exuberant inflammatory response. The pathophysiology of CF lung disease is not thoroughly understood, but inflammation contributes to the decline in pulmonary function and is a valid independent therapeutic target. Studies in infants and children (17, 24–26, 35), nearly all studies in CF mice (11, 23, 38, 39), and many studies in CF airway epithelial cell cultures and cell lines (6, 14, 21, 29, 30) indicate that the inflammatory response, either to TNFα and IL-1β or to P. aeruginosa or its products, occurs in excess in CF, compared with non-CF samples. The cytokines that are most consistently in excess in CF (e.g., IL-8 or murine equivalents KC and MIP-2, IL-6, GM-CSF), as well as other features of the CF inflammatory response such as excess ICAM expression (2) and release of MMPs (12, 34), require activation of NF-κB for upregulation, and several laboratories have shown increased activation of NF-κB in CF airway epithelial cell lines (8, 9, 14, 30, 33, 40, 41). Failure to appropriately modulate NF-κB activation could account for the excess inflammatory response in CF, and control of NF-κB activation could be of therapeutic value.
One potential regulator of NF-κB activation is the peroxisome proliferator-activated receptor-γ (PPARγ), a member of the ligand-activated nuclear receptor superfamily. For example, recently, PPARγ has been proposed as a therapeutic target in asthma, possibly by reducing NF-κB activity in the lung (22). PPARγ agonists attenuate the asthmatic response in mice, and complementary studies show that these agonists can affect airway smooth muscle cells, dendritic cells, and macrophages (3, 13, 28). Investigation of PPARγ and NF-κB may be important in CF because of reports that CF mice have reduced expression of PPARγ mRNA in organs in which CFTR expression is important, including the lung (27). However, the expression and role of PPARγ in airway epithelial cells has not been elucidated. Based on the importance of NF-κB in the CF inflammatory response, and data from other laboratories suggesting reduced PPARγ in tissues in which CFTR expression is prominent in CF mice, we hypothesized that PPARγ quantity and/or function is reduced in CF airway epithelium and that PPARγ modulates the inflammatory response of airway epithelial cells by suppressing the action of NF-κB. This hypothesis is quite appealing since PPARγ agonists are approved for human use in other contexts and therefore may be safe as therapeutic interventions.
To test our hypotheses, we investigated PPARγ expression in airway epithelial cells and found that PPARγ quantity or function is reduced in CF airway epithelial cells in culture. In addition, activation of PPARγ in airway epithelium by ligand binding can prevent excess activation of NF-κB, and treatment with PPARγ agonists in a CF infection mouse model can reduce the inflammatory response.
MATERIALS AND METHODS
Human bronchial epithelial cell pair 16HBE14o− sense and 16HBE14o− antisense cells (non-CF and CF phenotype, respectively) as well as human tracheal epithelial cell pair 9/HTEo− pCEP and 9/HTEo− pCEP-R (non-CF and CF, respectively) were grown as previously described (6, 20, 21, 29, 31).
Well-differentiated human airway epithelial cells.
Human tracheal epithelial cells were recovered from necropsy specimens with approval from the University Hospitals of Cleveland Institutional Review Board (IRB exemption EM-03-01) and grown as previously described (15, 16) at the air-liquid interface (ALI). Treatment with the CFTR inhibitor, CFTRinh-172, kindly provided by Dr. Alan Verkman, was also conducted as previously described (30). Briefly, cells were allowed to differentiate at the ALI for 3 wk in serum-containing media, switched to submerged culture (liquid-liquid interface) on day 0, and treated with either vehicle control DMSO 1:1,000 (Sigma, St. Louis, MO) or 20 μM CFTRinh-172 prepared in DMSO and diluted from a 1:1,000 stock. Drugs were added to both the basolateral and apical side, and media was replenished every 24 h. At day 3, cells were stimulated with 100 ng/ml TNFα/IL-1β for 1 h, and nuclear extracts were prepared and processed with the TranSignal TF-TF Interaction Array I (Panomics, Fremont, CA), as indicated below.
Supernatants of well-differentiated human airway epithelial cells (WD AEC) from three different donors were centrifuged for 10 min at 14,000 rpm and concentrated with an Ultra-4 Filter (Amicon, Billerica, MA) to 50–70 μl, mixed with 2× nonreducing SDS sample buffer, and subjected to SDS-PAGE gels containing 1 mg/ml gelatin. To allow MMP to renature, gels were washed twice in 2.5% Triton X-100 in sterile water, incubated in activation buffer (10 mM Tris·HCl, pH 7.5, 1.25% Triton X-100, 5 mM CaCl2, 1 μM ZnCl2) overnight at 37°C, stained with 0.25% Coomassie brilliant blue R-250 diluted in 40% methanol and 10% acetic acid for 1–2 h, and destained in 40% methanol and 10% acetic acid until clear zones of protease activity were visible against the background.
Reporter gene assays.
Cells were seeded in 24-well tissue culture dishes (Costar, Corning, NY) 24 h before transfection. Firefly luciferase plasmids driven by NF-κB or IL-8 promoters (BD Biosciences Clontech, Palo Alto, CA) and pRLTK (Promega, Madison, WI) as control were used for transfection with Lipofectamine PLUS reagent (Invitrogen, Carlsbad, CA). A final stock concentration of 10 mM troglitazone (Cayman Chemical, Ann Arbor, MI), dissolved in DMSO and diluted to various concentrations, was added immediately after the transfection reagent was withdrawn, with diluted DMSO as control. The next day, cells were stimulated with P. aeruginosa strain O1 (PAO1) 109 colony-forming units (cfu) for 1 h, lysed in 1× lysis buffer, and assayed with the Dual Luciferase Reporter Assay System (Promega) on a microplate luminometer (Berthold Detection Systems, Oak Ridge, TN).
Transcription factor arrays.
TranSignal TF-TF Interaction Array I (Panomics) was processed according to the manufacturer's instructions. Nuclear extracts from 16HBE14o− sense and antisense cells were incubated with biotin-labeled double-stranded oligonucleotides. PPARγ was immunoprecipitated with 3 μg of monoclonal antibody and magnetic protein G Dynabeads (Dynal, Brown Deer, WI). Free cis-elements and nonspecific binding proteins were washed away four times with immunoprecipitation buffer supplied by the manufacturer. PPARγ-associated biotin-labeled probes were eluted from the Dynabeads by heating to 100°C and hybridized to TranSignal Protein/DNA array membranes at 42°C overnight and washed the next day in a rotating hybridization oven. The arrays were blocked and incubated with streptavidin-HRP and developed with a chemiluminescent detection system.
16HBE14o− cell pair was stimulated with 100 ng/ml TNFα and IL-1β for 1 h, and nuclear extracts were prepared according to the transcription factor ELISA TransAM NF-κB p50 kit (Active Motif, Carlsbad, CA). Extracts were added to a microplate coated with an oligonucleotide probe corresponding to the NF-κB sequence (5′ GGGACTTTCC 3′). The plate was then incubated with an antibody specific to activated NF-κB and an HRP conjugate and developed using a colorimetric assay.
Immunoprecipitations and Western immunoblotting.
Nuclear and cytoplasmic extracts were prepared with the nuclear extraction kit from Panomics. Nuclear extracts (100 μg) and cytoplasmic extracts (400 μg) from each cell line were immunoprecipitated with polyclonal antibodies against either the p50 or p65 subunit of NF-κB (Santa Cruz Biotechnology, Santa Cruz, CA) and probed with a monoclonal antibody against PPARγ (Panomics) or immunoprecipitated with a polyclonal anti-PPARγ (Affinity Bioreagents, Golden, CO) and probed with a monoclonal antibody against NF-κB p65 (Santa Cruz). The nuclear extract was diluted to 500 μl with immunoprecipitation (IP) buffer, 1% Triton X-100, 150 mM NaCl, 10 mM Tris, pH 7.4, 1 mM EDTA, and protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO). Extracts were incubated with IP antibody and rotated 1 h or overnight at 4°C. Antibody-antigen complexes were precipitated with protein G beads (Roche, Indianapolis, IN). Beads were washed three times with cold IP buffer, eluted in SDS-PAGE sample buffer, and boiled. The supernatants were run on 10% SDS-PAGE and transferred to nitrocellulose by electroblotting. PPARγ was detected using the PPARγ Western blot detection kit from Panomics. Briefly, blots were blocked in 3% nonfat dry milk in 1× Wash Buffer II, gently rocked overnight at 4°C, and then incubated with affinity-purified monoclonal PPARγ antibody (1:300) for 2 h at room temperature. Blots were washed three times with 1× Wash Buffer II for 15 min. Anti-mouse HRP (1:1,000) was added for 1 h at room temperature, and then blots were washed 4× with 1× Wash Buffer I for 20 min and developed using a chemiluminescent detection system (Panomics). For NF-κB p65, blots were blocked in 5% nonfat dry milk in PBS plus 0.1% Tween 20, incubated with NF-κB p65 (1:1,000, Santa Cruz) for 1 h at room temperature, and probed with an anti-mouse HRP secondary antibody (1:3,000; Chemicon, Temecula, CA) for 1 h at room temperature. Chemiluminescent detection of the bands was performed using ECL (GE Healthcare, Piscataway, NJ). For Western blots in Fig. 1, A and B, equal amounts of proteins (50 μg for cytoplasmic extracts and 20 μg for nuclear extracts) were loaded and separated by 10% SDS-PAGE and transferred to nitrocellulose, and PPARγ was detected as indicated above.
9/HTEo− cell pair was stimulated or not with PAO1 109 cfu for 1 h, and nuclear extracts were prepared. Five micrograms of nuclear extracts were incubated with γ32P-labeled oligonucleotide target corresponding to the consensus binding site for PPARγ (5′ AG GTC AAA GGT CA 3′, Santa Cruz Biotechnology) for 30 min at room temperature. To test for specificity, extracts were incubated with either excess cold wild-type probe or mutant probe (5′ AG CAC AAA GCA CA 3′, Santa Cruz Biotechnology) and with an antibody to PPARγ (Santa Cruz Biotechnology) before adding the labeled probe. Complexes were resolved in a 5% nondenaturing polyacrylamide gel and visualized by autoradiography.
Animal protocols were approved by Case Western Reserve University's Animal Care and Use Committee. STOCK Cftrtm1Unc-TgN(FABPCFTR)#Jaw and B6.129P2-Cftrtm1Unc mice were bred, maintained, and infected with planktonic P. aeruginosa mucoid clinical strain M57-15, as previously described (39). Female gut-corrected Cftr knockout (CF) mice and wild-type heterozygote littermates to the Cftr knockout (wild-type) mice (6–9 wk of age) were treated with 30 mg/kg pioglitazone (Cayman Chemical) in 10% ethanol by gavage on day −1 and on day 0 (24 h apart). Control mice received diluent by gavage (10% ethanol). Mice were infected with PA M57-15 (2–7 × 107 cfu/mouse in 20 μl by insufflation under isoflurane anesthesia) 1 h after the last treatment and killed on day 1. All mice survived the procedures, and no animals developed clinical signs of disease or showed signs of distress. Three experiments were performed, and the data were combined; there was a total of 25–27 mice per group. Mice were killed by carbon dioxide, in accordance with the 2007 AVMA Panel on Euthanasia (4), and exsanguinated by direct cardiac puncture. Following death, bronchoalveolar lavage fluid (BALF) was collected, and cell numbers and cytokine concentrations were determined, as previously described (38, 39).
Lung histopathology and immunostaining.
Following bronchoalveolar lavage, the lungs were harvested and inflation-fixed with 2% paraformaldehyde for at least 48 h and embedded in paraffin. Five-micrometer sections were stained with hematoxylin and eosin and assessed for lung inflammation, as described (37). Additionally, adjacent five-micrometer sections were stained for PPARγ using a rabbit monoclonal antibody (Cell Signaling Technology, Danvers, MA). Briefly, sections were deparaffinized and hydrated with xylene and a graded alcohol series. Antigen unmasking was performed using 10 mM sodium citrate buffer (pH 6.0) and microwave heating (10 min). Slides were immersed in 3% hydrogen peroxide (10 min) to quench endogenous peroxidase, blocked for 1 h at room temperature with serum (Vector Laboratories, Burlingame, CA), and incubated overnight at 4°C with either serum blocking solution (no 1ry antibody) or anti-PPARγ rabbit monoclonal (1:800, Cell Signaling Technology). The next day, slides were incubated with biotinylated secondary antibody for 1 h at room temperature, followed by streptavidin-peroxidase for 30 min at room temperature (Vectastain Elite Rabbit IgG kit, Vector Laboratories). The presence of PPARγ was detected with exposure to 3,3′-diaminobenzene and H2O2 (Vector Laboratories). Slides were counterstained with hematoxylin.
The software packages SigmaPlot v10 (Systat Software, San Jose, CA) and SAS (SAS Institute, Cary, NC) were used to analyze the data. Data are expressed as means ± SE. If not otherwise indicated, values between two groups were analyzed using unpaired, two-tailed Student's t-test or the Mann-Whitney Rank Sum test, if normality testing failed. A probability level of P < 0.05 was considered significant. To statistically evaluate the cell count data from mice, the nonparametric van Elteren test was used to test for group differences while adjusting for different experiments. Some mouse cytokine values were below the limit of detection; they were assigned the value of the limit of detection of the assay and were considered censored data. Therefore, a stratified Wilcoxon test was used to compare groups after controlling for (stratifying by) experiments.
PPARγ protein expression and differential binding to peroxisome proliferator response element in airway epithelial cells.
Both cytoplasmic and nuclear extracts of two different cell line pairs (9/HTEo− and 16HBE14o−), as well as nuclear extracts from WD AECs, demonstrated the presence of PPARγ by Western blot (Fig. 1A), with higher expression in nuclear vs. cytoplasmic extracts, as expected. In the 9/HTEo− cell pair, there appeared to be equivalent amounts of protein in the CF and non-CF cells. For the 16HBE14o− cell pair, however, the CF member of the pair consistently expressed less PPARγ than the non-CF member in both nuclear and cytoplasmic fractions. This differential expression for the 16HBE14o− pair did not change upon PA01 stimulation (Fig. 1B). EMSA indicated that even though there was no difference in PPARγ expression between the members of the 9/HTEo− pair, the PPARγ DNA binding activity was different between them. There was a marked decrease in PPARγ DNA binding in the 9/HTEo− CF cells, which increased slightly upon PAO1 stimulation (Fig. 1D). The specificity of the binding was demonstrated by competing the binding with cold oligonucleotide, failure to compete it with a mutant oligo, and by band supershift with a PPARγ antibody (Fig. 1C). Therefore, PPARγ is expressed in human airway epithelial cells, CF and non-CF, but appears to be either less abundant, less functional in binding its target DNA sequence, or both, in CF.
PPARγ agonists reduce excess activation of NF-κB in CF cell lines.
The amount of activated NF-κB p50 in nuclear extracts of 16HBE14o− non-CF and CF matched cell pair was determined by ELISA under basal conditions and under TNFα/IL-1β stimulation (Fig. 2). Both members of the pair exhibited statistically significantly increased NF-κB p50 upon stimulation, but the CF member had significantly higher levels than the non-CF pair, both at basal and upon stimulation. In other experiments, the 16HBE14o− cell pair was transfected with firefly luciferase plasmids driven by either NF-κB or the native IL-8 promoter, and, immediately after, was exposed to an increasing concentration of troglitazone, a PPARγ agonist, for 24 h, followed by PAO1 stimulation. Promoter activity was assessed by measuring luciferase activity (Fig. 3). The CF cells had greater NF-κB or IL-8-driven luciferase expression in response to PAO1 than did the non-CF cells, and troglitazone treatment inhibited both NF-κB and IL-8 promoter-driven luciferase production in a dose-dependent fashion. Neither the 9/HTEo− pair nor the WD AECs could be studied reliably because of poor transfectability. Thus, with three independent assays in two different-matched model systems, the activation of NF-κB in CF models is in excess of that observed in non-CF models, and troglitazone can drastically reduce NF-κB activation.
PPARγ agonists inhibit cytokine and MMP-9 production by WD AEC.
To test whether these interactions with NF-κB had functional effects on WD AEC, we measured the effects on cytokine and MMP-9 release. Cells from several different donors were tested, as were different PPARγ agonists. The increased IL-8 and GM-CSF secretion from WD AECs in response to PAO1 or TNFα/IL-1β stimulation was decreased by prior treatment with 10 μM troglitazone (Fig. 4A; 4 donors, each done in triplicate); this decrease was statistically significant for the PAO1-treated cells. In a separate experiment, IL-8 secretion by TNFα/IL-1β was statistically reduced by treatment with either 10 μM troglitazone or 20 μM ciglitazone; the decrease in PAO1-treated cells did not reach statistical significance in this case (Fig. 4B; 1 donor, in triplicate).
Gelatin zymography shows that WD AECs grown at ALI secrete MMP-9 proteolytic activity, which was reduced by PPARγ agonists under either basal or stimulated conditions (Fig. 5, 3 different donors). There was no increase in MMP-9 release upon either PAO1 or TNFα/IL-1β stimulation in these samples.
NF-κB and PPARγ interaction.
Cytokine production and MMP-9 release are promoted in part by activation of NF-κB. To test whether PPARγ can interact with NF-κB, coimmunoprecipitation assays were performed. Nuclear and cytoplasmic extracts of 9/HTEo− and 16HBE14o− pairs were immunoprecipitated with antibodies to either the p50 (Fig. 6A) or the p65 (Fig. 6B) subunits of NF-κB and blotted with a monoclonal anti-PPARγ. Nuclear extracts were also IP with a polyclonal anti-PPARγ and blotted with a monoclonal anti-NF-κB p65 (Fig. 6C). Both p50 and p65 were able to pull down PPARγ, much more in the nuclear fraction in all cells. PPARγ was also able to pull down p65.
In a second, more sensitive assay, which capitalizes on the ability of biotin-labeled double-stranded oligonucleotide probes to bind specifically to their corresponding transcription factors, interaction of PPARγ with NF-κB was also identified in the 16HBEo− cell pair (Fig. 7). There was less interaction between PPARγ and NF-κB in the CF member of the pair compared with the non-CF. PAO1 stimulation caused reduction of the PPARγ-NF-κB interaction, particularly in the CF cells (Fig. 7, A and B), but this interaction could be preserved in part by troglitazone treatment of CF cells treated with TNFα/IL-1β (Fig. 7C). These data suggest that although inflammatory stimulation causes changes in NF-κB or in PPARγ that reduce their interaction, these changes can be partly abrogated by interaction of PPARγ with an agonist ligand.
The above assay was also performed in WD AECs grown at the ALI treated with CFTRinh-172, a CFTR inhibitor. This model allows one to compare the effect of CFTR inhibition on various cellular processes in WD AEC that have identical genetic endowment at all loci (30). Cells treated with CFTRinh-172 displayed less interaction between PPARγ and NF-κB, particularly following stimulation with TNFα/IL-1β (Fig. 8, A and C).
Bronchial airway epithelial cells obtained from two different CF patients homozygous for ΔF508 showed the same limited interaction between PPARγ and NF-κB as the WD AEC treated with CFTRinh-172 (Fig. 8B). These results support the concept that in CF, reduced interaction of PPARγ and NF-κB may contribute to the excess activation of genes driven by NF-κB.
Mechanism of reduced PPARγ activity in CF airway cells.
Our data indicate that PPARγ activity is reduced in CF airway epithelial cells. CF airways are characterized by an increase in RhoA and oxidant stress (19), which leads to the activation of ERK and JNK, and the subsequent phosphorylation of PPARγ by these MAP kinases (5). PPARγ is inactivated when phosphorylated (1, 7), which may explain the reduced activity seen in CF cells. To test this, we examined the effects of U0126 and PD-98059 (specific ERK inhibitors) and SP-600125 (a specific JNK inhibitor) on TNFα/IL-1β-stimulated IL-8 secretion in the presence of troglitazone in WD AEC. Figure 9 (2 different donors, each done in triplicate) shows that all inhibitors are able to decrease IL-8 secretion by TNFα/IL-1β to the same extent as troglitazone by itself, but only ERK inhibitors show synergy with troglitazone. Although both ERK inhibitors (U0126 and PD-98059) elicited a further decrease in IL-8 secretion when used in conjunction with troglitazone at both stimulus concentrations (1 μg/ml and 100 ng/ml), the additional decrease from PD-98059 was statistically significant only for the lowest stimulus concentration. The additive decrease for U0126 was statistically significant at both concentrations (see Fig. 9 legend for P values). These results indicate that the reduced PPARγ activity observed in CF cells could be due in part to PPARγ phosphorylation by ERK.
Effects of PPARγ agonists in CF mice challenged with Pseudomonas.
To test whether the effects of PPARγ agonists can be detected in the more complex in vivo model, we studied CF and wild-type mice in an acute P. aeruginosa challenge model. Mice were pretreated with pioglitazone or vehicle by gavage, challenged with M57-15 P. aeruginosa, and killed 24 h later, and lung inflammation was assessed by bronchoalveolar lavage. Changes in body weight (Fig. 10), bronchoalveolar lavage cells counts (Fig. 11), and bronchoalveolar lavage cytokine values (Fig. 12) were recorded. Pioglitazone treatment did not affect the body weight of either CF or wild-type mice (Fig. 10). Sham-treated CF mice had a significantly higher number of neutrophils (P ≤ 0.02) and lymphocytes (P ≤ 0.0018), both in relative and absolute terms (Fig. 11), and significantly elevated proinflammatory cytokines TNFα (P = 0.0206), IL-1β (P = 0.0258), IL-6 (P = 0.0001), IFNγ (P = 0.0107), MIP-2 (P = 0.0441), and KC (P < 0.0001) in BALF (Fig. 12) compared with wild-type sham-treated mice. There was no significant difference in the absolute number, yet significant difference in the relative number (P = 0.0082), of alveolar macrophages in BALF of sham-treated CF mice compared with sham-treated WT mice (Fig. 12). These results indicate an exaggerated inflammatory response to PAO1 treatment by CF mice compared with WT, as previously reported (39).
The relative and absolute numbers of neutrophils (Fig. 11), as well as levels of TNF-α, IL-1β, and MIP-2 (Fig. 12), were significantly reduced (P < 0.05) in CF mice treated with pioglitazone compared with sham-treated CF mice. There were no significant differences in lymphocyte numbers (Fig. 11), IL-6, IFN-γ, or KC (Fig. 12) between these two groups of mice. The relative and absolute numbers of neutrophils (Fig. 11), as well as levels of TNF-α, IL-1β, and MIP-2 (Fig. 12), were similar (P > 0.05) in CF mice treated with pioglitazone compared with sham-treated wild-type mice. These results indicate that pioglitazone treatment is not only capable of reducing lung inflammation in CF mice, but also that pioglitazone treatment normalizes lung inflammation.
Although the percent of neutrophils was significantly reduced (P = 0.0234), the absolute number of neutrophils was not significantly changed (P > 0.05) in wild-type mice treated with pioglitazone compared with sham-treated controls (Fig. 11). However, there was a significant increase in the percent (P = 0.0255) and absolute number (P = 0.0185) of alveolar macrophages in wild-type mice treated with pioglitazone compared with sham-treated controls (Fig. 11). TNF-α and IL-1β were significantly reduced (P < 0.04) in wild-type mice treated with pioglitazone compared with sham-treated controls (Fig. 12). Therefore, pioglitazone treatment reduced lung inflammation in WT mice.
Because in the mouse intestine both intensity of PPARγ staining and its subcellular localization is altered in CF compared with wild-type mice (27), we prepared lung sections from the above mice and stained them for PPARγ. Our results in the lung differed somewhat from those observed in the intestine. Both wild-type and CF mice treated with vehicle showed mainly diffuse staining throughout the cytoplasm and nucleus of bronchial epithelial cells, but a few cells in both models exhibited predominant nuclear staining (arrows in Fig. 13). In both mouse models, not all bronchi were stained. Pioglitazone treatment produced an overall increase in staining in the cytoplasm and nucleus of wild-type and CF mice, but the proportion of cells with predominant nuclear staining per bronchi was considerably greater (Fig. 13). As with vehicle-treated mice, not all epithelial bronchi stained for PPARγ. All no-primary antibody controls were negative.
PPARγ interacts with NF-κB in airway epithelial cells, and by this means modulates the inflammatory response. PPARγ agonists are particularly efficacious in limiting transcriptional activation by NF-κB in CF model systems, and, probably because increased NF-κB-driven transcription contributes to the excessive inflammatory response in CF (8, 9, 14, 21, 29, 30, 33, 40, 41), PPARγ agonists also inhibit the inflammatory response to infection with P. aeruginosa in CF mice in vivo. These data suggest that these drugs should be considered for anti-inflammatory therapeutic use in CF.
In these studies, we show that PPARγ is expressed in airway epithelial cells. When PPARγ ligands are administered along with or before inflammatory stimuli in two airway epithelial cell lines and one primary culture model, NF-κB-driven processes are inhibited, including the production of IL-8 and GM-CSF and the release of MMP-9 in response to Pseudomonas or cytokine stimulation. Transcription from an NF-κB luciferase construct or one in which the IL-8 promoter is used to drive luciferase is reduced by agonists of PPARγ in airway epithelial cells in a dose-dependent fashion, indicating that these agonists may exert at least a portion of their activity at the level of gene transcription. Here we show that one mechanism by which this occurs could be by direct interaction with NF-κB, or by interaction with a third protein, possibly a DNA helicase or a coactivator, to which both NF-κB and PPARγ bind. Both the p50 and the p65 subunits coimmunoprecipitated with PPARγ, and PPARγ was coimmunoprecipitated by antibodies to either p50 or p65. This interaction was confirmed by another more sensitive technique that recognizes transcription factors by DNA base pairing in their target sequences.
However, this interaction was reduced in the presence of proinflammatory stimuli (both P. aeruginosa and inflammatory cytokines), suggesting that part of the inflammatory response in airway epithelial cells is to reduce the effect of this modulator on the NF-κB system. The specific mechanisms by which PPARγ interaction with NF-κB is reduced by proinflammatory stimuli are not clear. It has been reported that phosphorylation of PPARγ by ERK results in altered conformation of PPARγ (1, 5, 7), which in turn alters its binding of DNA (which is increased), its ability to bind ligand (which is reduced) (32, 36), and its ability to be ubiquitinated and degraded (which is increased) (10). We speculate that such phosphorylation might produce a conformation less able to interact with NF-κB as well. Ligands may protect PPARγ from these changes, although these relationships may be complex and PPARγ ligands may have biological effects related to inflammation that are independent of PPARγ. Inflammatory stimuli activate ERK in airway epithelial cells and could account for the increased DNA binding of PPARγ that we observed, as well as the reduced quantity of PPARγ under inflammatory conditions. We speculate that ERK phosphorylation may also modulate the anti-inflammatory effects of PPARγ. In support of this hypothesis, we found that inhibitors of ERK, but not of JNK, were synergistic with PPARγ in inhibiting TNF-stimulated IL-8 production in WD AECs.
CF airway epithelial cells display significant abnormalities in PPARγ. The DNA binding activity of PPARγ is reduced in CF airway epithelial cells. EMSAs indicate less interaction of PPARγ with its target DNA sequence in two CF cell model systems compared with matched controls. For the 16HBE14o− cells, this could be due, at least in part, to reduced expression of PPARγ in the CF member of the pair, as demonstrated by Western blot, but in the 9/HTEo− cell pair, expression is comparable in the CF and the non-CF members of the pair. It seems most likely that the ability of PPARγ to bind to its target DNA sequence is reduced. The 9/HTEo− cell pair differs from the 16HBE14o− cell pair in that the 16HBE14o− pair displays activation of IL-8 and IL-6 production at baseline, but the 9/HTEo− cell lines are quiescent until a stimulus is applied, and the basal production of cytokines is minimal. If the continuous activation in the 16HBE14o− cells results in more rapid degradation of a posttranslationally modified PPARγ, this might account for the greater deficit in CF cells in this cell line. It is possible that the CF cell lines exist in a heightened inflammatory state and PPARγ is sensitive to this constitutive activation. Others have reported reduced quantities of PPARγ mRNA and protein in CF mouse organs in which CFTR expression is prominent, such as intestine and lung, but not in organs where CFTR is not expressed to such a high degree, such as fat or liver (27). Despite the apparent reduction in PPARγ in the 16HBE14o− antisense (CF) cell model, troglitazone was more effective in inhibiting NF-κB-driven transcription. This may occur because troglitazone can stabilize PPARγ from degradation over the time of this assay, and then the excess NF-κB activation in the CF model is more susceptible to inhibition. Alternatively, troglitazone may exhibit PPARγ-independent activities as well, such as inhibition of IKK.
In the CF mouse model, application of troglitazone is reported to result in the proper nuclear translocation of the PPARγ in the gut, which occurs without ligand in normal mice, but is not observed in the absence of ligand in CF animals. However, we observed similar PPARγ localization in the infected lungs of CF and wild-type mice both in the absence of ligand (diffuse staining and some isolated cells with preferential nuclear localization) and in the presence of ligand (increased number of cells with predominantly nuclear localization), although ligand greatly increased both the extent and intensity of detectable expression as well as the nuclear localization. It may be that in the lung, expression of PPARγ is quite sensitive to the inflammatory environment, and any changes we observe in CF may be due to heightened inflammatory tendencies. This suggestion is supported by the reduction of PPARγ in patients with asthma or alveolar proteinosis, diseases characterized by inflammation, but normal CFTR. Even if the changes in PPARγ are associated with changes in the inflammatory milieu in CF and are not related to the CF defect itself, they could contribute to the excess inflammatory response and represent a valid therapeutic target. It now appears that some, but not all, nonsteroidal anti-inflammatory drugs (NSAIDs) can ligate PPARγ. Ibuprofen, at the concentration required to observe the therapeutic effect in CF, is one of those drugs. Ligation of PPARγ might, therefore, be one mechanism of action of this proven anti-inflammatory therapeutic agent in CF (18).
To test the therapeutic potential of a more specific PPARγ agonist and one available for human use in CF animals, we utilized the acute Pseudomonas challenge model in CF and non-CF mice because it was the closest mimic of the acute Pseudomonas challenge applied to the epithelial cells in culture and because it may mimic the initial acquisition of this organism in patients. In these experiments, pioglitazone significantly limited the inflammatory response in the CF mice. However, the dose used in these studies, on a weight basis, was high compared with conventional human doses, and the drug was administered before challenge, a situation that may be difficult to duplicate in patients with CF. Nevertheless, further consideration of this class of drugs for treatment of the CF inflammatory response is warranted.
This work was supported by National Institutes of Health Grants DK-027651 and HL-073870 and a grant from the Cystic Fibrosis Foundation.
We thank Case CF Center Animal Core staff members for expert technical support, Dr. Alma Wilson and Veronica Peck for breeding and maintaining the mice, William Marcus for genotyping mice, and R. Christiaan van Heeckeren, James Poleman, and Thomas Shaw for performing the lung infection studies, tissue harvesting, sample processing, cell counts, and evaluating lung morphology. In addition, we are grateful to Yoshie Hervey for cell culture work, Inflammatory Mediator Core staff member Christopher Statt for performing cytokine assays, and Histology Core member Claudia Garner for preparing lung histology sections.
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 © 2008 the American Physiological Society