Background
Chronic Obstructive Pulmonary Disease (COPD) is a chronic inflammatory respiratory disease affecting millions of people worldwide. Inhaled corticosteroids (ICS) are widely used in the management of COPD. ICS effectively reduce the number of exacerbations and improve respiratory symptoms and quality of life [
1]. However, ICS use may also increase the risk of pneumonia in COPD [
2,
3]. The TORCH study demonstrated this for the first time, comparing fluticasone propionate (FP) and placebo [
2,
3]. Findings from this study were confirmed in a meta-analysis by Singh and colleagues [
4]. Sixteen of the 18 studies included in the meta-analysis of Singh and colleagues investigated the effects of FP or FP/salmeterol, and it remained unclear whether the increased pneumonia risk would be FP specific or a class effect of ICS and also present with budesonide (BUD) treatment. More recent studies suggested that pneumonia events were lower with BUD than with FP treatment [
5,
6]. Furthemore, Suissa and colleagues reported that FP treatment is associated with a substantial increase in the risk of serious pneumonia in COPD patients, while the risk with BUD was comparatively low, even at high doses [
7]. Most recently, Suissa and colleagues reported that discontinuation of ICS use in COPD is associated with a reduction in the elevated risk of serious pneumonia, especially for FP [
8]. Thus, the increased risk to develop pneumonia in COPD may be specific to the use of FP and not the result of a class effect of ICS. The cellular mechanisms underlying these differences in safety for ICS use in COPD patients are not well understood. BUD is less lipophilic than FP and has a higher aqueous solubility, leading to a shorter retention time in the lining fluid of the airways, while after being absorbed, BUD is retained in airway tissue/epithelium for a longer time than FP [
9,
10]. It is as yet unknown how this may affect the action of BUD and FP in epithelial cells.
The bronchial epithelium forms the first continuous physical barrier to microbial infections and is part of the innate immune response, producing antimicrobial and pro-inflammatory peptides/cytokines acting on immune cells, the latter especially when the epithelial layer is damaged. In COPD, aberrant epithelial repair in response to cigarette smoking may disturb epithelial barrier function [
11] and we previously observed a reduction in epithelial barrier function upon smoke extract exposure in vitro [
12]. Compromised barrier function may render the airways more susceptible to pathogens, and accordingly, rhinovirus-induced barrier dysfunction in mice was shown to increase the risk of a secondary bacterial infection [
13]. The corticosteroid dexamethasone improves airway and corneal epithelial barrier function in vitro [
14‐
16]
.
We hypothesized that BUD is more effective than FP in protecting against airway epithelial barrier dysfunction upon damage by environmental insults. Viral infection may predispose to bacterial pneumonia, activating toll-like receptor 3 (TLR3) on airway epithelium [
13,
17]. TLR3-dependent effects have also been demonstrated for
Haemophilus Influenza, one of the most common causes of pneumonia in COPD [
18]. Therefore, we compared the effect of BUD and FP on viral mimetic poly-(I:C) and/or cigarette smoke-induced epithelial barrier function and pro-inflammatory cytokine production in both the human bronchial epithelial cell line 16HBE and cultured primary bronchial epithelial cells (PBECs) of smoking individuals with normal lung function.
Methods
Cell culture
The human bronchial epithelial cell line 16HBE was kindly provided by Dr. D.C. Gruenert (University of California, San Francisco, CA) and cultured in EMEM medium/10 % FCS (Biowhittaker, Verviers, Belgium) supplemented with 100 U/ml penicillin and 100 μg/ml streptomycin on collagen-coated flasks as described previously [
19]. PBEC cultures were obtained from bronchial brushings in six current smoking individuals with ≥10 pack-years, FEV
1/FVC > 70 % and FEV
1 > 90 % of predicted and not using inhaled corticosteroids, long-acting β
2- adrenergic agonists and long-acting anticholinergics for at least 4 weeks preceding the study. The Medical Ethics Committee of the University Hospital of Groningen approved the study. All subjects gave their written informed consent. For studies in mucociliary differentiated cells, PBECs were obtained by protease digestion from trachea-bronchial tissue of 10 non-COPD donor lungs. The study protocol was consistent with the Research Code of the University Medical Center Groningen (
http://www.rug.nl/umcg/ onderzoek/researchcode/index) and national ethical and professional guidelines (“Code of conduct; Dutch federation of biomedical scientific societies”; htttp://
www.federa.org). Cells were cultured as described previously [
20] in bronchial epithelium growth medium (BEGM, Lonza, Walkersville, MD) on collagen/fibronectin-coated flasks and stored in liquid nitrogen to be used for experiments at later time in passage 3.
Treatment of the cells
We used BUD and FP in equivalent concentrations with a dose ratio of FP:BUD = 1.6, based on the observations from clinical studies that 800 μg BUD is equivalent to 500 μg FP. Cells were pre-treated with or without BUD or FP for 2 hours and subsequently exposed to vehicle (medium), 5 or 7.5 % CSE [
12], poly-I:C (12.5 μg/ml), EGF (10 ng/ml) or GSK-3β inhibitor CT99021 (1 μM) for 1–24 hours.
Viral infection in air-liquid interface-cultured epithelial cells
To induce mucociliary differentiation at the air-liquid interface (ALI), the PBECs from non-COPD donors were grown in duplicates on semi-permeable membranes coated with 30 μg/ml collagen 10 μg/ml fibronectin and 10 μg/ml BSA in a 1:1 mixture of DMEM (Lonza) and BEGM supplemented with retinoic acid (RA, 15 ng/ml; Sigma, St. Louis, MO, USA) and exposed to air for 4 weeks as described previously [
20]. Cells were hormonally deprived overnight. For infection with live rhinovirus (RV; major receptor group RV-16, kind gift of D. Davies, University of Southampton, UK), cells in an identically seeded well were counted to calculate multiplicity of infection (MOI) and the virus concentration was adjusted to the number of cells. The apical surface was infected with 50 μl RV16 with an MOI of 1 for 24 hours at 37 °C before harvesting for RNA isolation and collection of supernatants.
Exposure to Streptococcus pneumoniae
For bacterial infection,
Streptococcus pneumoniae strain TIGR4∆cps was used.
S. pneumoniae was grown in M17 broth (Oxoid,Hamshire, UK) supplemented with 0.5 % glucose, or on blood agar plates (Mediaproducts bv, Groningen, The Netherlands) as described previously [
21]. For start inoculations in all experiments,
S. pneumoniae aliquots were made by growing
S. pneumoniae in M17 supplemented with glucose to a 600 nm optical density of ~0.25, mixed to a 10 % glycerol concentration and then frozen in 1 ml aliquots at −80 °C. Prior to infection, confluent 16HBE cell monolayers in uncoated transwell plates (Transwell, 3 μm pore-size, 6.5 mm diameter; Costar #3472, Costar Corning Inc., Cambridge, MA) were incubated for 2 hour in infection assay medium, with and without 16 nM BUD and 10 nM FP. Subsequently, ~5*10
6 CFU of
S. pneumoniae were added per well and incubated for 2–24 hours, in the presence and absence of 12.5 μg poly-(I:C). To assess
S. pneumoniae adhesion/internalization, 16HBE cells were washed with PBS/0.01 % CaCl
2 and subsequently lysed with PBS/0.1 % Triton. Colony forming units (CFUs) were determined by plating serial dilutions on blood agar plates. For analysis of transmigration, the medium was removed from the basolateral compartment after 2, 4 or 24 hours and plated for CFU determination. Before and 24 hours after bacterial infection, transepithelial resistance was measured using a volt-ohmmeter (EVOM, world precision instruments, Sarasota, FL).
Preparation of cigarette smoke extract
Cigarette smoke extract (CSE) was prepared as described previously [
22]. In short, Kentucky 3R4F research-reference cigarettes (The Tobacco Research Institute, Lexington, KY) were used without filter. Smoke from two cigarettes was bubbled through 25 ml medium (100 % CSE). The extract was prepared freshly.
Electric Cell-surface Impedance Sensing (ECIS)
Electrical resistance of submerged cultured cells was measured using ECIS (Applied Biophysics, Troy, NY) as described previously [
23,
24]. Resistance and capacitance were measured at 400 Hz and 40 kHz, respectively. In the ECIS system, all established resistance values were between 10,000-20,000 Ω in the 16HBE cultures and ~1,500 Ω in the primary cell cultures.
Western blotting
Total cell lysates were obtained and subjected to western blotting using antibodies against E-cadherin, phospho-EGF receptor (EGFR), actin, GAPDH (Santa Cruz Biotechnology, Santa Cruz, CA), zona occludens (ZO)-1 (Invitrogen, Carlsbad, CA, USA) and phospho-GSK-3β (Cell Signalling Technology, Herts, UK) as described previously [
25]. Protein levels were quantified using the gelscan program QuantityOne.
Immunofluorescent staining of ZO-1
Cells grown on LabTeks were washed with PBS/CaCl2, fixed in ice-cold acetone (90 %) for 30 min, blocked in PBS/5 % BSA for 60 min, incubated for 60 min with primary antibodies (1:200) against ZO-1 (Invitrogen) and subsequently incubated for 60 min with FITC-labeled anti-rabbit (1:200, DAKO, Glostrup, Denmark) or Rhodamine-labeled anti-mouse IgG conjugates (1:400, Jackson Immunoresearch Laboratories, West Grove, PA).
Measurement of gene expression with qPCR
RNA was isolated from 16HBE and cDNA synthesized as described previously [
26]. We analyzed the expression of E-cadherin and the housekeeping genes PPIA and β2μG. Analyses were performed by real-time PCR using Taqman according to manufacturer’s guidelines using validated probes and the TaqMan Master (Applied Biosystems, Foster City, CA).
Measurement of IL-8 levels
Protein levels were measured in cell-free supernatants using ELISA kits according to manufacturer’s guidelines (R&D systems Europe Ltd., Abingdon, UK).
Statistics
Data were analyzed using the paired Student’s t-test, the Wilcoxon-signed rank test or repeated measures ANOVA for ECIS experiments as indicated. Differences were considered statistically significant at p < 0.05.
Discussion
We hypothesized that BUD is more efficacious than FP in the protection of airway epithelial barrier function upon cigarette smoking or viral infection, potentially contributing to the ICS’ differential risk associated with pneumonia in COPD. Our results show that BUD protects more effectively than FP against CSE-induced bronchial epithelial barrier dysfunction, either alone or in combination with viral mimetic poly-(I:C), while BUD and FP equally effectively protect against poly-(I:C)-induced barrier dysfunction. Furthermore, both BUD and FP strongly suppress CSE and/or poly-(I:C)-induced pro-inflammatory cytokine production in bronchial epithelial cells lines and PBECs of smokers without significant differences between the drugs. Our findings may have important implications, since they help to explain the clinical observations that treatment with BUD is not, or seldom associated with an increased risk to develop pneumonia in patient with COPD, in contrast to FP. The present study suggests that this is unlikely a consequence of increased immunosuppression by FP. Instead, treatment of COPD patients with BUD could provide better protective effects against cigarette smoke-induced epithelial damage than FP, reinforcing the epithelial barrier. This may limit the access of pathogens upon cigarette smoking in the presence or absence of a viral infection in vivo. Indeed, we observed that BUD protected against the poly-(I:C)-induced increase in bacterial adhesion and/or internalization. This effect may be mediated by the reinforcement of the epithelial barrier, as epithelial junctions functionally segregate the basolateral from the apical site. RV-induced disruption of epithelial barrier function may thus increase the exposure of cell surface receptors for bacterial binding [
30]. Whether loss of epithelial polarity indeed acts to enhance binding of
S. pneumoniae to surface receptors will require further investigation. Our findings on the reduced expression of E-cadherin mRNA in RV16-exposed differentiated epithelium upon FP treatment suggest that the treatment with FP, but not BUD, may aggravate RV-induced barrier disruption in vivo, and thus increase the risk of a secondary bacterial infection.
Our data indicate that the differential effects of BUD and FP on epithelial barrier function are due to differences in their effect on specific pathways involved in barrier dysfunction upon CSE exposure. While BUD attenuated the EGF-induced phosphorylation of EGFR and its downstream target GSK-3β, FP was not able to do so. This inhibitory effect of BUD on EGFR signaling could be involved in the protective effect of BUD on CSE-induced barrier dysfunction, since we have previously reported that CSE-induced EGFR phosphorylation results in epithelial barrier dysfunction by delocalization of ZO-1 from tight junctions [
12]. CSE has also been described to inhibit GSK-3β activity by its phosphorylation at Ser9 in lung epithelial cells [
31]. To our knowledge, we are the first to demonstrate that pharmacological inhibition of GSK-3β results in epithelial barrier dysfunction, suggesting that the attenuation of EGFR-dependent GSK-3β phosphorylation by BUD is involved in the protective effect of BUD on epithelial barrier function. We observed that BUD was not able to restore epithelial barrier dysfunction upon pharmacological inhibition of GSK-3β, indicating that GSK-3β activation is indispensable for the responsiveness of the epithelial barrier to BUD, in line with previous findings in lymphoma cells [
32]. GSK-3β can induce degradation of transcriptional repressor Slug/Snail2 [
33], leading to upregulation of E-cadherin, ZO-1, claudins and occludin expression [
34]. Since E-cadherin mRNA expression was not affected by CSE and corticosteroids in our setting, transcriptional regulation of junctional proteins does not likely contribute to the observed effects on epithelial barrier function. Alternatively, Snail can induce disruption of tight junction complexes at the posttranslational level, by causing alternative splicing of ZO-1, resulting in higher expression of the ZO-1 isoform that is involved in junctional plasticity [
34]. Future studies will have to determine whether EGFR-dependent GSK-3β inactivation and subsequent degradation of Snail are involved in CSE-mediated disruption of epithelial junctions. In addition to its effects on barrier function, GSK-3β has been implicated in inflammatory responses to bacterial infection [
35]. Furthermore, side-stream cigarette smoke-induced inactivation of GSK-3β was shown to increase the susceptibility to adenovirus by the upregulation of its receptor in airway epithelial cells [
36], with additional implications for the susceptibility to microbial infection.
It is not fully clear why BUD is more efficient than FP in suppressing the CSE-induced EGFR/GSK-3β pathway, while these ICS are equally effective in suppressing pro-inflammatory cytokine production and providing protection against barrier dysfunction by viral mimicry. It could be speculated that different pathways are involved in these specific processes, although the pathways involved in poly-(I:C)-induced barrier dysfunction need further investigation. With respect to their physicochemical properties, BUD is less lipophilic than FP and has a higher aqueous solubility, leading to a faster dissolution rate and a shorter retention time in the lining fluid of the airways. By contrast, after absorption from the airway lumen, BUD is retained in the airway epithelium for a longer time than FP [
9,
10]. This is due to the conjugation of BUD with endogenous fatty acids, resulting in a very lipophilic ester depot from which BUD is slowly released [
9]. This prolongs anti-inflammatory effects of BUD in the airway tissue [
37] and may possibly also prolong effects of BUD on specific intracellular pathways. Fatty acid esterification of BUD has also been detected in the human lungs [
38]. Further studies are required to elucidate whether BUD and FP exert differential effects on pathways involved in pro-inflammatory and antimicrobial responses to viral infection. Additionally, recent findings in human bronchial epithelial cells show that a given glucocorticosteroid induces a unique gene expression “fingerprint” [
39], and this may explain some of the differences observed between BUD and FP in the present study.
Conclusions
Together, our data show that BUD is more efficient in the protection against cigarette smoke-induced epithelial barrier dysfunction than FP, and suggest that this is due to more efficient suppression of EGFR/GSK-3β signaling. We anticipate that this may have important implications for the reinforcement of airway epithelial barrier function upon cigarette smoking in vivo, where treatment with BUD could provide better protective effects than FP, limiting the access of pathogens.
Competing interests
None of the authors has anything to disclose.
Authors’ contributions
IH contributed to the design of the work, data analysis and interpretation, drafting and revising of the manuscript. MJ was involved in the acquisition, analysis and interpretation of the data and critical review of the manuscript. MdV helped to set up the experiments with virus. NtH and ET were involved in the inclusion of subjects and critical review of the manuscript. AvO contributed to the design of the work and critical review of the manuscript. DP contributed to the design of the work, drafting and critical revision of the manuscript. MvdB contributed to the design of the work, drafting and critical revision of the manuscript. All authors have critically read and approved the manuscript.