Comments
Abstract
The lung epithelium constitutes a selective barrier that separates the airways from the aqueous interstitial compartment. Regulated barrier function controls water and ion transport across the epithelium and is essential for maintaining lung function. Tight junctions (TJs) seal the epithelial barrier and determine the paracellular transport. The properties of TJs depend especially on their claudin composition. Steroids are potent drugs used to treat a variety of airway diseases. Therefore, we addressed whether steroid hormones directly act on TJ properties in lung epithelia. Primary human tracheal epithelial cells and NCI-H441 cells, both cultivated under air–liquid interface conditions, were used as epithelial cell models. Our results demonstrate that glucocorticoids, but not mineralocorticoids, decreased paracellular permeability and shifted the ion permselectivity of TJs toward Cl−. Glucocorticoids up-regulated claudin 8 (cldn8) expression via glucocorticoid receptors. Silencing experiments revealed that cldn8 is necessary to recruit occludin at the TJs. Immunohistochemistry on human lung tissue showed that cldn8 is specifically expressed in resorptive epithelia of the conducting and respiratory airways but not in the alveolar epithelium. We conclude that glucocorticoids enhance lung epithelia barrier function and increase paracellular Cl− selectivity via modulation of cldn8-dependent recruitment of occludin at the TJs. This mode of glucocorticoid action on lung epithelia might be beneficial to patients who suffer from impaired lung barrier function in various diseased conditions.
Glucocorticoid treatment is recommended in most chronic and inflammatory lung diseases. Our findings provide new insights into how glucocorticoids improve lung epithelial barrier function and identify a new role of claudin 8 in tight junction organization. They also imply a new mode of glucocorticoid action in the lung, which is beneficial for patients with chronic obstructive pulmonary disease, asthma, polytrauma, and acute respiratory distress syndrome.
Epithelia separate compartments of the organism. They establish and maintain different ion and/or osmotic gradients between them. Hence, they form a barrier to prevent an uncontrolled molecule flux. The paracellular permeability reflects epithelial barrier function and is limited by tight junctions (TJs). TJs have a complex heteromeric protein composition, including cytoskeletal proteins, scaffold proteins, and barrier-forming proteins, such as occludin and claudins (1). Although occludin has been shown to decrease paracellular epithelial permeability and maintain lipid polarity of epithelial cells (2), studies in occludin−/− knockout mice revealed that a lack of occludin does not affect TJ properties in all epithelia (3). Thus, other proteins besides occludin are thought to define the paracellular permeability of TJs. Claudins have been identified as such barrier-determining TJ proteins (4). To date, 23 protein homologs of the claudin superfamily have been identified (5, 6). Claudins form TJs via two types of interaction: either with claudins of adjacent cells (trans interaction) or with claudins among the membrane of one cell (cis interaction). Both cis and trans interactions have been shown to be involved in TJ formation (7). Claudin interaction is not restricted to homomeric protein assembly. In fact, quite complex heteromeric interactions within the claudin superfamily are described either via cis (8) or trans interaction (9, 10). Both, the number of homologous proteins and the complex interaction pattern contribute to the functional variability of TJs, especially with respect to their permselectivity. The specific combination of different claudins enables the fine tuning of TJ properties in accordance with the specific requirements of a certain epithelium. This has been impressively demonstrated for the different segments of the nephron (11–13). Impaired claudin function resulting in an inappropriate adjustment of TJs to epithelial requirements is the cause of multiple diseases associated with defects of paracellular transport (14). This demonstrates the physiological relevance of claudins.
Because fine tuning of TJs is essential for an appropriate epithelial function, it appears conclusive that claudins underlie hormonal control to regulate the properties of TJs in response to changes in physiological conditions. In keeping with this concept, it was shown that ovarian hormones adjust uterine TJs (15) and hormonal control of amnion TJs during pregnancy is involved in amniotic fluid volume regulation (16, 17). Intestinal paracellular permeability is adjusted via epidermal growth factor (EGF) and aldosterone (Aldo), which is hypothesized to adjust Na+ uptake in response to a low-Na+ diet (18). Furthermore, claudin (cldn) 4 modulation supports Aldo-dependent adjustment of Na+ resorption in the distal nephron segments (19). Restoration of epithelial barrier function is improved by steroids, such as estrogen (20, 21), glucocorticoids, and mineralocorticoids (18, 19, 22–26). Hence, it has been proposed that steroids support maintenance of TJs during diseases, such as inflammatory bowel disease (26).
In lung epithelia, claudins determine paracellular permeability (27). Their dysregulation is associated with several lung diseases. Impaired expression of several different claudins has been shown to be involved in interstitial pneumonia, sarcoidosis, and metaplastic epithelium (28). Impaired epithelial paracellular permeability has also been observed in lung epithelia obtained from septic animals (29). Cldn4 was recently linked with intact alveolar fluid clearance (30), and its elevation is involved in compensating lung edema in early stages of acute lung injury (31). Chronic alcohol abuse increases susceptibility to acute lung injury and increases TJ permeability in the alveolar epithelium via impairing cldn1, -5, and -7 expression (32). Compromised TJs in lung epithelia have also been observed in biopsies from patients with asthma (33).
Glucocorticoid treatment is a standard treatment in various forms of obstructive pulmonary diseases, such as asthma or chronic obstructive pulmonary disease (COPD) (34–36). Steroids are potent drugs, and have been shown to affect epithelial barrier function in many tissues. Surprisingly, relatively little is known about their impact on the barrier function of the lung epithelium. Based on the existing knowledge of claudin function in epithelia and the effects of steroids, we herein tested whether steroids regulate claudins directly to modulate TJ function in lung epithelia. Primary cultivated human tracheal epithelial cells (hTEpCs) and NCI-H441 cells, both cultivated at air–liquid interface (ALI), served as epithelial models of the lung. Our results show that glucocorticoids significantly affect transepithelial electrical resistance (TEER) and paracellular permeability, and induce the recruitment of occludin at the TJs via regulating cldn8. Cldn8 is expressed in epithelia along the conducting and respiratory airways, but not within the alveolar epithelium of the human lung. Possible implications of these effects for lung function will be discussed.
A detailed description of materials and methods is given in the online supplement.
NCI-H441 cells (ATCC, Bielefeld, Germany) and primary human tracheal epithelial cells (Promocell, Heidelberg, Germany), both cultivated at ALI, were used as epithelial cell models. Steroids and compounds were added to the basal medium. If not stated otherwise, cultivation time is always given as days after seeding cells on filter. Epithelia were exposed to compounds for the entire cultivation time at ALI, unless stated otherwise.
Paracellular permeability was quantified by impedance spectrometry as the TEER. Permselectivity of TJs was investigated via dilution potential measurements in open-circuit configuration of Ussing chamber experiments. To estimate the paracellular permeability only, the apparent permeability coefficient (Papp) to Na+-fluorescein was measured.
Quantification of expression levels were performed in semiquantitative RT-PCR experiments using the RT2 profiler PCR array “tight junction” and the RT2 SYBR Green PCR master mix. Subsequent RT-PCR experiments were performed using the quantitect primer assays: hydroxymethylbilane synthase (HMBS) (internal housekeeping gene control) = QT0014462; cldn8 = QT00212268; epithelial sodium channel α-subunit (α-ENaC) = QT00022883; epithelial sodium channel β-subunit (β-ENaC) = QT00051597; epithelial sodium channel γ-subunit (γ-ENaC) = QT00063217; T-cell lymphoma invasion and metastasis 1 (TIAM1) = QT01009568; intercellular adhesion molecule 1 (ICAM1) = QT00074900; guanine nucleotide binding protein alpha inhibiting activity polypeptide 1 (GNAI1) = QT00023555 (all from Qiagen, Hilden, Germany).
Cldn8 and -4 protein expression was analyzed in Western blotting experiments using the following primary antibodies: rabbit anti-cldn8 (LS-B5505) antibody (Life Span Bioscience, Seattle, WA) and rabbit anti-cldn4 (ab53156) antibody (Abcam, Cambridge, UK), both diluted 1:500, and mouse anti–β-actin antibody (Sigma-Aldrich GmbH, Steinheim, Germany) diluted 1:2,500.
Intracellular localization of cldn8, zonula occludens 1 (ZO1), and occludin were performed by immunocytochemistry using the following primary antibodies: rabbit anti-cldn8 (LS-B5505) antibody (Life Span Bioscience), goat anti ZO1 (ab99462) antibody (Abcam), and mouse anti-occludin antibody (MAB7074) (R&D Systems, Minneapolis, MN), both diluted 1:300.
Trachea tissue was obtained from a combined heart–lung transplant of a patient suffering from severe pulmonary–arterial hypertension. Healthy lung tissue was obtained from volume-reduction surgery of donor lungs. Cldn8 was detected in formalin-fixed, paraffin-embedded human lung slices using cldn8 antibody (LifeSpan BioSciences) and stained using the ZytoChem-Plus HRP kit (ZYTOMED Systems GmbH, Berlin, Germany). Hematoxylin and eosin staining was prepared for orientation. Human tissue was obtained with approval of the local ethics committee (Medizinische Hochschule Hannover approval 2699-2015) and in accordance with the donors.
NCI-H441–based cell lines stably silenced for cldn8 expression were generated by lentiviral particles either expressing cldn8-specific small hairpin RNA (shRNA) (V3LHS_406694; GE Healthcare Dharmacon, Little Chalfont, UK), or noncoding shRNA (RHS4384; GE Healthcare Dharmacon) as negative control. Cell populations were enriched for stably transfected cells by cultivating NCI-H441 cells in the presence of 4 μg/ml puromycin. Cultivation in the presence of puromycin did not affect formation of ALI epithelia.
NCI-H441 cells (Figure 1A) were cultivated on filters in a submerged configuration until they formed a confluent monolayer, upon which an ALI was imposed. Basal medium was replaced by ALI medium supplemented with 300 nM dexamethasone (Dex) as a glucocorticoid, 1,000 nM hydrocortison (HyCort) as a steroid with intermediate gluco- and mineralocorticoid function, or 300 nM Aldo as a mineralocorticoid. Mifepristone (Mif; 300 and 3,000 nM) and spironolacton (Spiro; 300 nM) were used as inhibitors of glucocorticoid and mineralocorticoid receptors, respectively. To assess sealing properties of TJs, TEER was measured on Day 11 after seeding cells on filters. Dex and HyCort, both induced an increase of TEER, whereas Aldo had no effect on TEER. The glucocorticoid receptor antagonist Mif diminished the steroid induced increase in TEER partly in case of Dex and almost completely in case of HyCort. The mineralocorticoid receptor antagonist, Spiro, had no effect either on control or on steroid-treated epithelia. To test if the increase in TEER depends on an altered transcellular ion conductibility, we measured the Papp (Figure 1B) for Na+-fluoescein. In accordance with the TEER measurements, Dex and HyCort decreased Papp. This effect on Papp was partially diminished by Mif. Evidently, glucocorticoids modulate predominantly the paracellular pathway.
Figure 1. Effects of steroids on paracellular permeability and gene expression. (A) NCI-H441 cells were cultivated at air–liquid interface (ALI), and paracellular permeability was quantified as transepithelial electrical resistance (TEER). Aldosterone (Aldo), cells cultivated in the presence of 300 nM Aldo; Control, cells cultivated in the absence of steroids; Dex, cells cultivated in the presence of 300 nM dexamthasone; HyCort, cells cultivated in the presence of 1,000 nM hydrocortison; Mif, mifepristone; Spiro, spironolacton (numbers give concentrations in nM; n = 7–9, one-way ANOVA with Holm Sidaks correction for multiple comparison). (B) Apparent permeability coefficients (Papp) for sodium fluorescein determined for ALI-cultivated NCI-H441 epithelia cultivated at control conditions (control) in the presence of 300 nM Dex and 100 nM HyCort without or together with 3,000 nM Mif (n = 11–12, one-way ANOVA with Holm Sidaks correction for multiple comparison). (C) Differential gene expression of tight junction (TJ) and TJ-associated proteins in NCI-H441 epithelia. Scatter plot: “Control sample” represents cells cultivated in the absence of steroids; “Dex treated sample,” cells cultivated in the presence of 300 nM Dex. Data points give the mean of relative expression levels (n = 3). Black line indicates identical expression levels in both cell populations. Red lines represent threefold difference in expression levels. Transcripts with differences in expression levels exceeding the threefold threshold are highlighted. gnai1, guanine nucleotide binding protein α inhibiting activity polypeptide 1; icam1, intercellular adhesion molecule 1; tiam1, T-cell lymphoma invasion and metastasis 1. (D) Bar diagram: the transcripts assigned in C were analyzed in subsequent experiments. Cells were cultivated at ALI in the presence of 300 nM Dex, 1,000 nM HyCort, or in the absence of steroids (control). Expression levels are given as expression levels relative to control cells using the ΔΔCt method. Only changes in expression levels of claudin (cldn) 8 exceed the threefold threshold for both steroids. Red lines, threefold expression change; black line, no change in expressional level (all HyCort/Dex versus control). (E) Effect of glucocorticoid receptor modulators on cldn8 expression in NCI-H441 epithelia. Cells cultivated in the absence of steroids (control), and in the presence of 300 nM Dex and 1,000 nM HyCort. Mif was added at concentrations of 300 nM (Mif 300) and 3,000 nM (Mif 3,000) to antagonize glucocorticoid receptor activation (ANOVA with Holm Sidaks correction for multiple comparison; n = 5–6). Claudin 8 expression (cldn8 expr) is given as relative expression to hydroxymethylbilane synthase (rel. HMBS). (F) Glucocorticoids increase the paracellular permselectivity for Cl−. NCI-H441 cells were cultivated at ALI conditions in the absence of steroids (control) and in the presence of 300 nM Dex and 1,000 nM HyCort. Paracellular permeabilities for Na+ and Cl− were measured on the basis of their dilution potentials. The relative permeability was calculated as the ratio of Na+ permeability over Cl− permeability (n = 8, one-way ANOVA with Holm Sidak’s correction for multiple comparison). (G) TEER of primary cultivated human tracheal epithelial cells (hTEpCs) cultivated at ALI in the absence (control) and presence of 300 nM Dex (n = 8, Mann-Whitney test). (H) Differential gene expression of TJs and TJ-associated proteins in hTEpC epithelia. Scatter plot: “Control sample” represents cells cultivated in the absence of steroids; “Dex treated sample,” cells cultivated in the presence of 300 nM Dex. Data points give the mean of relative expression levels (n = 3). The black line indicates identical expression levels in both cell populations. Red lines represent threefold difference in expression levels. Transcripts encoding claudins with differences in expression levels exceeding the threefold threshold are highlighted. (I) Cldn8 expression relative to HMBS. Control, hTEpCs cultivated in the absence of steroids; Dex, hTEpCs cultivated in the presence of 300 nM Dex; HyCort, hTEpCs cultivated in the presence of 1,000 nM HyCort; and Aldo, cells cultivated in the presence of 300 nM Aldo. ANOVA with Bonferroni correction for multiple comparison. All data are given as mean ± SEM. Significance levels are given as: *P < 0.05; **P < 0.01; ***P < 0.001, and ****P < 0.0002.
[More] [Minimize]To further investigate the molecular mechanisms responsible for the glucocorticoid-induced decrease in paracellular permeability, we next analyzed whether TJ protein expression, a major determinant of paracellular permeability, was regulated by glucocorticoids. We compared expression levels of TJ protein genes in NCI-H441 epithelia, cultivated in the presence and absence of 300 nM Dex, in semiquantitative RT-PCR experiments. This initial expression screen (Figure 1C; see also Table E1 in the online supplement) revealed altogether four transcripts with expression levels exceeding the threshold of threefold up- or down-regulation: namely, cldn8, gnai1, icam1, and tiam1. In these experiments, the 4-fold standard deviation corresponded with a threefold change in expression levels and, thus, we considered those changes that exceeded a threefold threshold as relevant. To confirm the initial screen, we performed subsequent semiquantitative RT-PCR experiments with PCR primers, different from those used in the initial screen. In these experiments, we compared the expression levels between control cells (untreated epithelia) and epithelia cultivated either in the presence of 300 nM Dex or 1,000 nM HyCort (Figure 1D). We observed that only cldn8 was more than threefold up-regulated by both HyCort and Dex. In the cases of gnai1, icam1, and tiam1, a more than threefold change in expression levels was observed only for Dex, but not for HyCort-treated epithelia. Thus, we focused on cldn8 and tested whether glucocorticoid receptors regulate cldn8 expression, while investigating the effects of Dex, HyCort, and Mif in semiquantitative RT-PCR experiments (Figure 1E). Both steroids increased cldn8 expression in H441 epithelia significantly. In line with the TEER measurements, Mif diminished steroid-induced increase in cldn8 expression. These results suggest that glucocorticoids act via their glucocorticoid receptors on cldn8 expression. Cldn8 is a barrier-forming claudin, which decreases the TJ permeability for anions and cations, but more strongly affects cations (37) by interacting with cldn4 (26, 38, 39). Although cldn4 expression was not affected by Dex treatment, on either the transcriptional or protein levels (see Figure E1), we addressed whether glucocorticoids affect TJ permselectivity using dilution potential measurements (Figure 1F). Both Dex and HyCort decreased the permeability coefficient for Na+ (PNa+) from 1.17 ± 0.16 × 10−5 cm/s (control without steroids) to 0.18 ± 0.02 × 10−5 cm/s (Dex-treated cells) and 0.19 ± 0.02 × 10−5 cm/s (HyCort treated cells), respectively. The permeability coefficient for Cl− (PCl−) was also decreased from 1.23 ± 0.15 × 10−5 cm/s (control without steroids) to 0.22 ± 0.03 × 10−5 cm/s (Dex treated cells) and to 0.23 ± 0.02 × 10−5 cm/s (HyCort treated cells), respectively. The fractional permeability was calculated as the ratio of PNa+/PCl− and was found to be 0.999 ± 0.045 (control without steroids), 0.857 ± 0.031 (Dex-treated cells), and 0.821 ± 0.016 (HyCort-treated cells). This indicates that glucocorticoids reduce the paracellular permeability for all ions, but the TJ permeability of glucocorticoid-treated epithelia was more selective for Cl− than for Na+. This increase in Cl− selective permeability was significant for Dex- and for HyCort-treated epithelia.
In line with the observations in NCI-H441 epithelia, Dex also significantly increased TEER (Figure 1G) in epithelia from hTEpCs. An expression screen for genes encoding TJs and TJ-associated proteins revealed a much stronger and more complex effect on gene expression in primary human tracheal cells. Among proteins involved directly in the formation of TJ barrier, cldn3 and cldn8 were up-regulated by Dex, whereas cldn17 expression was reduced (Figure 1H). Because cldn8 was the only claudin, which behaved the same in NCI-H441 epithelia as well as in hTEpC epithelia, we focused on cldn8 in subsequent experiments. To confirm cldn8 up-regulation in primary cultivated hTEpCs, we investigated its glucocorticoid-dependent up-regulation in hTEpCs obtained from an additional donor (Figure 1I). In these epithelia, Dex induced a cldn8 up-regulation. Although this up-regulation was significant, it was reduced by two orders of magnitude compared with the epithelia from the donor used for the expression screening shown in Figure 1H. We conclude that glucocorticoids up-regulate cldn8 expression also in primary hTEpCs cultivated at ALI. The observed variability in glucocorticoid-induced changes in expression levels might be due to interdonor variability.
We next aimed to further characterize the relationship between cldn8 expression and barrier function. NCI-H441 epithelia were cultivated in the presence of increasing Dex concentrations, ranging from 0 to 1,000 nM, and HyCort, ranging from 0 to 10 μM (Figure 2). Whereas 1 nM Dex already increased TEER, the concentration dependence of HyCort on TEER was shifted far to the right, as HyCort induced TEER increases at concentrations of 100 nM and above. There is a remarkable additional difference in concentration dependence of TEER to these steroids. Dex increased TEER (Figure 2A) up to concentrations of 30 nM, at which point it showed its maximum effect. TEER decreased again at Dex concentrations of 100 nM and above. HyCort (Figure 2B) resulted in an almost unimodal concentration-dependent TEER increase. The concentration dependence of cldn8 expression to Dex and HyCort follows the same pattern as the TEER (Figures 2C and 2D). Thus, it is most likely that both effects are mediated via the same glucocorticoid receptor–mediated pathway, and that cldn8 expression causes the increase in TEER. It again supports that glucocorticoids affect TJs via modulation of cldn8 expression. In lung epithelia, ENaC subunit expression is modulated via glucocorticoid receptors (40–42). Thus, we investigated the concentration response of ENaC subunit expression (Figures 2E and 2F). The concentration response of ENaC expression to Dex observed here replicated previously published observations exactly (42). Importantly, the concentration response of ENaC subunit expression to Dex and HyCort follows the concentration response of cldn8 expression to both glucocorticoids. To test whether the expression of ENaC subunits is regulated via glucocorticoid receptors, we investigated the effect of Mif on Dex-dependent ENaC regulation (Figure 2G). Mif prevented the Dex-induced up-regulation of ENaC subunits almost completely. This observation, together with the identical concentration dependencies, strongly suggests that cldn8 and ENaC subunit expression are regulated via the same receptor.
Figure 2. Concentration response relationship to glucocorticoids. NCI-H441 cells were cultivated at ALI in the presence of Dex and HyCort, both at the given concentrations. (A and B) Concentration response relationship of TEER to Dex and HyCort. (C and D) Concentration response relationship of cldn8 expression to Dex and HyCort. (E and F) Concentration response relationship of α-, β-, and γ-epithelial sodium channel (ENaC) subunits to Dex and HyCort. All expression levels are given as relative expression levels normalized to Hmbs. (G) Effect of Mif on Dex-induced up-regulation of α-, β-, and γ-ENaC subunits in NCI-H441 epithelia. Bars represent relative expression levels as mean ± SEM.
[More] [Minimize]When NCI-H441 cells were cultivated on filters, the TEER did not rise steadily with cultivation time (Figure 3A). Instead, we observed that, independently from Dex treatment, the TEER increased from Day 6 onward, which was always 2 days after establishing ALI. The TEER reached quasi—steady state levels from Day 9 onward. During this time, the TEER was higher in Dex-treated epithelia than in control epithelia. The time course of TEER reflected several distinguishable states of NCI-H441 epithelia’s functional properties. Stable ALI on Days 6–7 and completed barrier formation from Day 9 onward. To investigate the time course of cldn8 up-regulation, we seeded NCI-H441 cells on filters. On Day 4, medium was changed against ALI medium without Dex, and cells were either kept at submerged (submerged; Figure 3B) or ALI configuration (ALI; Figures 3B and 3C). ALI cells were exposed to Dex on Days 7 and 10 (D7 ALI and D10 ALI, respectively; Figures 3B and 3C) and submerged cells on Day 10 only for the depicted exposure time. Expression levels of γ-ENaC (Figure 3B) and cldn8 (Figure 3C) were measured at given time points after Dex exposure. Expression of the γ-ENaC subunit, which served as an indicator for the functional expression of glucocorticoid receptors, was initiated 1 hour after Dex exposure in all tested NCI-H441 epithelia. Thus, glucocorticoid receptors are functional in all tested epithelia, regardless of ALI. Importantly, cldn8 became up-regulated only in ALI conditions at Day 10. In these experiments, the time course of increase in cldn8 expression levels after Dex exposure was similar to that of γ-ENaC. All other epithelia (submerged and Day 7 ALI) did not respond to Dex exposure with an up-regulation of cldn8 expression. Even though glucocorticoid receptors are functional in fully functionally maturated ALI epithelia, as well as in immature ALI and submerged epithelia, cldn8 up-regulation depends on a fully functional maturated ALI phenotype.
Figure 3. Time course of glucocorticoid-induced gene expression. (A) Time course of TEER in NCI-H441 epithelia cultivated in the absence of Dex (control) and in the presence of 300 nM Dex. Cultivation time represents cultivation time after seeding of cells on filters. ALI was established on Day 4. (B and C) NCI-H441 cells were cultivated on filters at submerged conditions (submerged) until Day 7 after seeding to ensure confluent cell layers at air–liquid conditions (D7 ALI) and for 10 days after seeding at ALI conditions (D10 ALI) to ensure tightness of the epithelium. Afterwards, 300 nM Dex was added to cell cultures for the given time intervals. Expression of γ-ENaC subunits (B) and cldn8 expression (C). All expression levels are given as relative expression levels normalized to expression levels of cells cultivated under submerged conditions exposed to Dex for 0 hours using the ΔΔCt method. All data are given as mean ± SEM.
[More] [Minimize]To test whether Dex affects organization and/or composition of TJ in lung epithelia, we investigated its effect on intracellular localization of TJ proteins with respect to cldn8 localization (Figure 4). The marker for TJ formation, ZO1, formed a belt-like structure, which surrounded the cells at the apical side of their lateral membranes. The ZO1 localization was independent of Dex treatment, and it always colocalized with cldn8 (Figures 4A and 4B). In untreated control NCI-H441 epithelia, occludin did not colocalize with cldn8, but accumulated in intracellular stores, where it showed a more punctuated staining pattern. In these cells, its colocalization with cldn8 was negligible (Figure 4C). Dex treatment resulted in occludin colocalization with cldn8 at the lateral membrane (Figure 4D). These observations applied also to hTEpC epithelia. In control hTEpC epithelia (Figure 4E), occludin colocalization with cldn8 at the lateral membrane was reduced compared with Dex-treated hTEpC epithelia (Figure 4F). However, in control hTEpCs, occludin did not accumulate in intracellular pools. This demonstrates that glucocorticoids modulate the incorporation of occludin into TJ.
Figure 4. Immunofluorescent staining of cldn8. Epithelia were cultivated at ALI. (A and C) NCI-H441 control epithelia cultivated in the absence of steroids and (B and D) in the presence of 300 nM Dex, all visualized using structured illumination. (E) Primary human tracheal epithelia cultivated in the absence of steroids and (F) in the presence of 300 nM Dex, both visualized using confocal microscopy. ZO1, zonula occludens protein 1, margenta channel in A and B; Ocln, occludin1, magenta channel in C–F; cldn8 is the green channel. Gray scale images of each channel are shown as small insets at the left of each image. Color image represents red/green/blue image of merged channels. Scale bars, 10 μm. con, control.
[More] [Minimize]To elucidate the role of cldn8 in TJ formation, we established a stable cldn8 silencing cell line on the basis of NCI-H441 cells. NCI-H441 cells stably expressing a nontargeting shRNA were used as control epithelia. Cldn8 silencing was confirmed in RT-PCR and Western blot experiments (Figures 5A and 5B). Cells in which cldn8 was silenced (cldn8−/−) formed epithelia with stable ALI. However, in contrast to control epithelia, cldn8−/− epithelia did not respond to Dex treatment with an increase in TEER (Figure 5C). In control epithelia, ZO1 colocalized with cldn8 at the plasma membrane independent of Dex treatment (Figures 5D and 5F), whereas occludin colocalization with cldn8 at the TJ depended on Dex treatment (Figures 5E and 5G). In cldn8−/− epithelia, ZO1 also localized to the lateral membranes independent of Dex treatment (Figures 5H and 5J); however, in contrast to control epithelia, Dex treatment did not result in recruitment of occludin to the TJs. However, in these epithelia, occludin localized at intracellular stores, even when cells were treated with Dex (Figures 5I and 5J). This proves cldn8 to play a pivotal role in recruiting occludin at the TJs.
Figure 5. Silencing of cldn8 using small hairpin RNA (shRNA) constructs. (A) Relative cldn8 expression in ALI-cultivated NCI-H441 cells stably expressing noncoding control shRNA (neg con) or silencing shRNA (cldn8 kd) in the absence of Dex (wo Dex) or in the presence of 300 nM Dex (Dex). (B) Cldn8 detection in Western blot experiments of ALI-cultivated NCI-H411 cells. Upper panel: NCI-H441 cells stably expressing noncoding shRNA in the absence and presence of 300 nM Dex. Lower panel: NCI-H441 cells expressing cldn8-silencing shRNA in the absence and presence of 300 nM Dex. Actin (act) served as an internal control to confirm protein loading. (C) Effect of Dex on TEER in NCI-H441 ALI epithelia (neg con, noncoding control; cldn8 kd, cldn8-silencing shRNA) cultivated at ALI in the absence and presence of 300 nM Dex (wo Dex and +Dex, respectively). (D-K) Confocal detection of cldn8 (green), ZO1 (magenta), and ocln (magenta) in NCI-H441 epithelia cultivated at ALI in the absence and presence of 300 nM Dex. Scale bars, 10 μm. Data are given as mean ± SEM.
[More] [Minimize]To confirm the relevance of our cell culture experiments, we investigated cldn8 protein distribution in human lung (Figure 6). As expected, cldn8 was detected in epithelial cells of conducting airways, such as the trachea, proximal and distal bronchioli, and bronchili respiratorii, but was completely absent in alveolar epithelial cells. The staining pattern for cldn8 contoured the epithelial cells and confirms its localization at the TJs in native tissues. Furthermore, it proves that the herein investigated cell models resemble the in vivo situation of the human lung.
Figure 6. Immunohistochemical staining of cldn8 in human lung sections. Cldn8 staining can be shown in typical distribution in the apicolateral membrane respiratory epithelial cells in the trachea (A; arrowheads point to typical dot-like staining). The same pattern can be observed in large bronchi (B), peripheral bronchioli (C), and respiratory bronchioli (D). Net-like staining patterns appear in areas of tangentially sectioned tissue. No staining is detectable in type I pneumocytes of the alveolar septum (asterisks in D). Scale bars, 50 μm.
[More] [Minimize]TJs regulate the paracellular transport pathway. They form a belt around the apical pole of the lateral membranes (cis interaction) and link adjacent lateral membranes between epithelial cells via trans interactions (7, 10). Claudins determine the functional properties of TJs. Their variability (5, 6), along with their tissue-specific expression and complex interaction pattern (7, 9, 10, 43), enables optimization of TJ properties in accordance with the requirements of certain epithelia. Steroids modulate claudin expression in several tissues, and hence affect TJ properties (18–26). Herein, we address the effect of steroids on lung epithelia. In both investigated epithelia, NCI-H441 and hTEpCs, we observed that glucocorticoids, but not mineralocorticoids, increase TEER. We demonstrate that ENaC channel expression was up-regulated as well, and, hence, we can assume that transcellular ion permeability will increase rather than decrease. Therefore, the effect of glucocorticoids on the transcellular ion transport would result in an opposing effect on TEER, namely a TEER decrease, rather than the one we observed. The measured decrease in Papp for Na+-fluorescein confirmed that the observed TEER increase in Dex-treated epithelia is indeed due to a drop in paracellular permeability. This effect was mediated via glucocorticoid receptor–dependent pathway, as we demonstrated that Mif, a specific glucocorticoid receptor antagonist, diminished this effect. Screening expression levels revealed that glucocorticoids up-regulate cldn8. Previous studies showed that cldn8 interaction with cldn4 is required to decrease paracellular permeability and to shift TJs’ permselectivity toward Cl− (26, 38, 39). Although we did not observe any expressional changes in cldn4 upon glucocorticoid treatment, our results are in accordance with these reports, as we observed that Dex induces an up-regulation of the cldn4 interacting partner, cldn8, as well as an increase in Cl− permselectivity of TJs. However, by identifying cldn8’s role in Dex-induced recruitment of occludin to the TJs, we demonstrate that cldn8 induces a complex change in TJ organization. Hence, we conclude that the change in TJ permselectivity results from multiple cldn8-dependent modulations of TJ organization. Cldn8 is the only glucocorticoid-regulated claudin within the investigated NCI-H441 epithelia. The effect of glucocorticoids on TJ in hTEpCs is, by far, more complex when compared with NCI-H441 epithelia. The expression screening on these epithelia revealed a much stronger up- and down-regulation of several TJs and TJ-associated proteins, but only in the case of cldn8 was the Dex-induced effect on gene regulation similar in both investigated epithelia.
Our results from immunocytochemistry experiments further demonstrate that cldn8 plays a pivotal role in the recruitment of occludin to the TJ. ZO1, which links claudins with the cytoskeleton, localizes at the apical pole of the lateral membranes, independent of Dex treatment. This proves TJ formation, even in the absence of Dex. On the other hand, occludin is recruited only in Dex-treated epithelia. The fact that cldn8 silencing abolished occludin recruitment demonstrates that Dex decreases TJ permeability via cldn8-mediated recruitment of occludin. This is in agreement with early observations that occludin decreases TJ permeability (2).
Cultivation of epithelia in the absence of Dex reduced cldn8 abundance rather than completely ablating it. The apicolateral localization of cldn8 also remained unaffected by Dex. This may point toward the possibility that, aside from the demonstrated quantitative cldn8 regulation, other mechanisms, which also depend on cldn8, are involved in regulating TJ organization. Phosphorylation of claudins has been demonstrated to modulate TJ assembly and claudin dynamics (44–46). Possibly, cldn8 phosphorylation or post-transcriptional protein modification might also be involved in cldn8-mediated occluding recruitment.
hTEpCs behaved similarly upon Dex treatment. In the absence of Dex, occludin already colocalized with cldn8 at the TJ, but to a much lesser extent than in Dex-treated hTEpCs. The initial expression screening on hTEpC epithelia revealed a larger number of Dex-regulated genes compared with NCI-H441 cells. In particular, cldn3 and -8 were up-regulated, whereas cldn17 was down-regulated. Possibly, cldn17 replaces cldn8 to recruit occludin in hTEpC epithelia, cultivated in the absence of Dex. The immunohistology on human lung specimen revealed cldn8 expression in TJs of epithelial cells along the conducting as well as respiratory airways, but not in the alveolar epithelium.
The effect of Dex on cldn8 expression depends on the ALI. As we demonstrated, glucocorticoids modulate cldn8 expression via the Mif-sensitive glucocorticoid receptor. While testing the Dex effect on expression of γENaC subunits, we confirmed that the glucocorticoid receptor is functional in epithelial cells either cultivated as submerged cell layers or at the ALI (47, 48). We observed almost identical concentration dependencies of Dex on cldn8 and ENaC subunit expression. Both effects were also sensitive to Mif. This indicates that regulation of these genes depend on the same glucocorticoid receptor. Such a scenario requires yet unknown factors, which are involved in cldn8 regulation and mediate its ALI dependence. Glucocorticoids are critical during lung maturation, in particular for epithelial cell maturation (49). Currently, glucocorticoid treatment is recommended in preterm infants to avoid acute respiratory distress syndrome and to improve lung maturation (50). The dependence of glucocorticoid-mediated TJ modulation on ALI adds an additional unresolved issue to the understanding of steroid modulation of lung development and maturation.
We recently demonstrated that NCI-H441 epithelia maintain ALI configuration even when cultivated in the absence of glucocorticoids. However, the apical liquid layer in these cells was increased several fold when compared with cells cultivated at Dex concentrations of 3 nM and above (51). The apical surface liquid volume in lung epithelia is regulated via modulating water resorption. The results presented herein suggest that glucocorticoids regulate apical surface liquid (ASL) volume, not only by increasing the osmotic driving force for water resorption, but also by optimizing paracellular transport, eventually by decreasing the paracellular back-leak of ions from blood to lumen.
In the intestine, cldn8 regulation was shown to depend on Aldo (38), but not on glucocorticoids (26). We show that cldn8 regulation in lung epithelia is independent of Aldo, but depends on glucocorticoids. As a mineralocorticoid, Aldo acts rather as a hormone that regulates electrolyte homeostasis. Conclusively, Aldo affects TJs in intestinal epithelia, which are important for electrolyte uptake (26, 38), or in epithelia of the nephron, which are involved in electrolyte excretion (19, 39). Epithelial function in other organs, which are not involved in electrolyte homeostasis, may likely remain unaffected by Aldo. Immunohistology on lung tissues revealed cldn8 expression in epithelial cells along the conducting and respiratory airways, but not in the alveolar epithelium. These cldn8-positive epithelia are thought to be involved in stabilizing and reducing the ASL, rather than in ASL production, as is presumed for the alveolar epithelium (52). The mineralocorticoid-independent, but glucocorticoid-dependent, TJ modulation via cldn8, especially along airway epithelia, would adjust resorptive epithelial transport and barrier function independently from the electrolyte state of the organism.
Several lung diseases are associated with defects of TJs (29–32). In particular, asthma patients suffer from compromised TJs (33). Cldn8 has already been identified to be down-regulated in epithelia upon cigarette smoke exposure (CSE). Those epithelia responded to CSE with a drop in TEER (53). Our data provide a possible mechanistic scenario for this observation, as we identified cldn8 as a key player in TJ formation. Its down-regulation during pathophysiological conditions, such as CSE or COPD and asthma, possibly explains the decrease in epithelial barrier function. Guidelines for asthma and COPD treatment indicate topical, but also systemic, glucocorticoid medication (34–36), predominantly to suppress inflammatory responses. Our study proposes another possible mode of action, that glucocorticoid medication may be beneficial to patients by directly enhancing the barrier function of the respiratory epithelium. This may not only apply to inflammatory states, but also to conditions of multifactorial organ failure, such as blunt chest trauma, hemodynamic edema, or acute respiratory distress syndrome.
| 1. | Van Itallie CM, Anderson JM. Architecture of tight junctions and principles of molecular composition. Semin Cell Dev Biol 2014;36:157–165. |
| 2. | Balda MS, Whitney JA, Flores C, González S, Cereijido M, Matter K. Functional dissociation of paracellular permeability and transepithelial electrical resistance and disruption of the apical–basolateral intramembrane diffusion barrier by expression of a mutant tight junction membrane protein. J Cell Biol 1996;134:1031–1049. |
| 3. | Saitou M, Furuse M, Sasaki H, Schulzke JD, Fromm M, Takano H, Noda T, Tsukita S. Complex phenotype of mice lacking occludin, a component of tight junction strands. Mol Biol Cell 2000;11:4131–4142. |
| 4. | Furuse M, Sasaki H, Fujimoto K, Tsukita S. A single gene product, claudin-1 or -2, reconstitutes tight junction strands and recruits occludin in fibroblasts. J Cell Biol 1998;143:391–401. |
| 5. | Overgaard CE, Daugherty BL, Mitchell LA, Koval M. Claudins: control of barrier function and regulation in response to oxidant stress. Antioxid Redox Signal 2011;15:1179–1193. |
| 6. | Hewitt KJ, Agarwal R, Morin PJ. The claudin gene family: expression in normal and neoplastic tissues. BMC Cancer 2006;6:186. |
| 7. | Piontek J, Winkler L, Wolburg H, Müller SL, Zuleger N, Piehl C, Wiesner B, Krause G, Blasig IE. Formation of tight junction: determinants of homophilic interaction between classic claudins. FASEB J 2008;22:146–158. |
| 8. | Koval M. Differential pathways of claudin oligomerization and integration into tight junctions. Tissue Barriers 2013;1:e24518. |
| 9. | Daugherty BL, Ward C, Smith T, Ritzenthaler JD, Koval M. Regulation of heterotypic claudin compatibility. J Biol Chem 2007;282:30005–30013. |
| 10. | Piontek J, Fritzsche S, Cording J, Richter S, Hartwig J, Walter M, Yu D, Turner JR, Gehring C, Rahn H-P, et al. Elucidating the principles of the molecular organization of heteropolymeric tight junction strands. Cell Mol Life Sci 2011;68:3903–3918. |
| 11. | Hou J, Goodenough DA. Claudin-16 and claudin-19 function in the thick ascending limb. Curr Opin Nephrol Hypertens 2010;19:483–488. |
| 12. | Hou J. Regulation of paracellular transport in the distal nephron. Curr Opin Nephrol Hypertens 2012;21:547–551. |
| 13. | Balkovetz DF. Claudins at the gate: determinants of renal epithelial tight junction paracellular permeability. Am J Physiol Renal Physiol 2006;290:F572–F579. |
| 14. | Krug SM, Schulzke JD, Fromm M. Tight junction, selective permeability, and related diseases. Semin Cell Dev Biol 2014;36:166–176. |
| 15. | Nicholson MDO, Lindsay LA, Murphy CR. Ovarian hormones control the changing expression of claudins and occludin in rat uterine epithelial cells during early pregnancy. Acta Histochem 2010;112:42–52. |
| 16. | Kobayashi K, Kadohira I, Tanaka M, Yoshimura Y, Ikeda K, Yasui M. Expression and distribution of tight junction proteins in human amnion during late pregnancy. Placenta 2010;31:158–162. |
| 17. | Kobayashi K, Miwa H, Yasui M. Progesterone maintains amniotic tight junctions during midpregnancy in mice. Mol Cell Endocrinol 2011;337:36–42. |
| 18. | Miró L, Pérez-Bosque A, Maijó M, Amat C, Naftalin RJ, Moretó M. Aldosterone induces myofibroblast EGF secretion to regulate epithelial colonic permeability. Am J Physiol Cell Physiol 2013;304:C918–C926. |
| 19. | Le Moellic C, Boulkroun S, González-Nunez D, Dublineau I, Cluzeaud F, Fay M, Blot-Chabaud M, Farman N. Aldosterone and tight junctions: modulation of claudin-4 phosphorylation in renal collecting duct cells. Am J Physiol Cell Physiol 2005;289:C1513–C1521. |
| 20. | Shin JA, Yang SJ, Jeong SI, Park HJ, Choi Y-H, Park E-M. Activation of estrogen receptor β reduces blood–brain barrier breakdown following ischemic injury. Neuroscience 2013;235:165–173. |
| 21. | Burek M, Arias-Loza PA, Roewer N, Förster CY. Claudin-5 as a novel estrogen target in vascular endothelium. Arterioscler Thromb Vasc Biol 2010;30:298–304. |
| 22. | Kelly SP, Chasiotis H. Glucocorticoid and mineralocorticoid receptors regulate paracellular permeability in a primary cultured gill epithelium. J Exp Biol 2011;214:2308–2318. |
| 23. | Kröll S, El-Gindi J, Thanabalasundaram G, Panpumthong P, Schrot S, Hartmann C, Galla H-J. Control of the blood–brain barrier by glucocorticoids and the cells of the neurovascular unit. Ann N Y Acad Sci 2009;1165:228–239. |
| 24. | Tenenbaum T, Matalon D, Adam R, Seibt A, Wewer C, Schwerk C, Galla H-J, Schroten H. Dexamethasone prevents alteration of tight junction–associated proteins and barrier function in porcine choroid plexus epithelial cells after infection with Streptococcus suis in vitro. Brain Res 2008;1229:1–17. |
| 25. | Felinski EA, Cox AE, Phillips BE, Antonetti DA. Glucocorticoids induce transactivation of tight junction genes occludin and claudin-5 in retinal endothelial cells via a novel cis-element. Exp Eye Res 2008;86:867–878. |
| 26. | Fischer A, Gluth M, Weege F, Pape U-F, Wiedenmann B, Baumgart DC, Theuring F. Glucocorticoids regulate barrier function and claudin expression in intestinal epithelial cells via MKP-1. Am J Physiol Gastrointest Liver Physiol 2014;306:G218–G228. |
| 27. | Coyne CB, Gambling TM, Boucher RC, Carson JL, Johnson LG. Role of claudin interactions in airway tight junctional permeability. Am J Physiol Lung Cell Mol Physiol 2003;285:L1166–L1178. |
| 28. | Kaarteenaho-Wiik R, Soini Y. Claudin-1, -2, -3, -4, -5, and -7 in usual interstitial pneumonia and sarcoidosis. J Histochem Cytochem 2009;57:187–195. |
| 29. | Cohen TS, Gray Lawrence G, Margulies SS. Cultured alveolar epithelial cells from septic rats mimic in vivo septic lung. PLoS One 2010;5:e11322. |
| 30. | Rokkam D, Lafemina MJ, Lee JW, Matthay MA, Frank JA. Claudin-4 levels are associated with intact alveolar fluid clearance in human lungs. Am J Pathol 2011;179:1081–1087. |
| 31. | Wray C, Mao Y, Pan J, Chandrasena A, Piasta F, Frank JA. Claudin-4 augments alveolar epithelial barrier function and is induced in acute lung injury. Am J Physiol Lung Cell Mol Physiol 2009;297:L219–L227. |
| 32. | Fernandez AL, Koval M, Fan X, Guidot DM. Chronic alcohol ingestion alters claudin expression in the alveolar epithelium of rats. Alcohol 2007;41:371–379. |
| 33. | Xiao C, Puddicombe SM, Field S, Haywood J, Broughton-Head V, Puxeddu I, Haitchi HM, Vernon-Wilson E, Sammut D, Bedke N, et al. Defective epithelial barrier function in asthma. J Allergy Clin Immunol 2011;128:549–556. |
| 34. | Rottier BL, Duiverman EJ. Anti-inflammatory drug therapy in asthma. Paediatr Respir Rev 2009;10:214–219. |
| 35. | Price D, Bousquet J. Real-world perceptions of inhaled corticosteroid/long-acting β2-agonist combinations in the treatment of asthma. Respir Med 2012;106:S4–S8. |
| 36. | Sethi S, Mahler DA, Marcus P, Owen CA, Yawn B, Rennard S. Inflammation in COPD: implications for management. Am J Med 2012;125:1162–1170. |
| 37. | Yu ASL, Enck AH, Lencer WI, Schneeberger EE. Claudin-8 expression in Madin-Darby canine kidney cells augments the paracellular barrier to cation permeation. J Biol Chem 2003;278:17350–17359. |
| 38. | Amasheh S, Milatz S, Krug SM, Bergs M, Amasheh M, Schulzke J-D, Fromm M. Na+ absorption defends from paracellular back-leakage by claudin-8 upregulation. Biochem Biophys Res Commun 2009;378:45–50. |
| 39. | Hou J, Renigunta A, Yang J, Waldegger S. Claudin-4 forms paracellular chloride channel in the kidney and requires claudin-8 for tight junction localization. Proc Natl Acad Sci USA 2010;107:18010–18015. |
| 40. | Zeissig S, Fromm A, Mankertz J, Zeitz M, Fromm M, Schulzke JD. Restoration of ENaC expression by glucocorticoid receptor transfection in human HT-29/B6 colon cells. Biochem Biophys Res Commun 2006;344:1065–1070. |
| 41. | Schulz-Baldes A, Berger S, Grahammer F, Warth R, Goldschmidt I, Peters J, Schütz G, Greger R, Bleich M. Induction of the epithelial Na+ channel via glucocorticoids in mineralocorticoid receptor knockout mice. Pflugers Arch 2001;443:297–305. |
| 42. | Venkatesh VC, Katzberg HD. Glucocorticoid regulation of epithelial sodium channel genes in human fetal lung. Am J Physiol 1997;273:L227–L233. |
| 43. | Anderson JM, Van Itallie CM. Physiology and function of the tight junction. Cold Spring Harb Perspect Biol 2009;1:a002584. |
| 44. | D’Souza T, Agarwal R, Morin PJ. Phosphorylation of claudin-3 at threonine 192 by cAMP-dependent protein kinase regulates tight junction barrier function in ovarian cancer cells. J Biol Chem 2005;280:26233–26240. |
| 45. | Van Itallie CM, Tietgens AJ, LoGrande K, Aponte A, Gucek M, Anderson JM. Phosphorylation of claudin-2 on serine 208 promotes membrane retention and reduces trafficking to lysosomes. J Cell Sci 2012;125:4902–4912. |
| 46. | Twiss F, Oldenkamp M, Hiemstra A, Zhou H, Matheron L, Mohammed S, de Rooij J. HGF signaling regulates Claudin-3 dynamics through its C-terminal tyrosine residues. Tissue Barriers 2013;1:e27425. |
| 47. | Nakamura K, Stokes JB, McCray PB Jr. Endogenous and exogenous glucocorticoid regulation of ENaC mRNA expression in developing kidney and lung. Am J Physiol Cell Physiol 2002;283:C762–C772. |
| 48. | Watt GB, Ismail NA, Caballero AG, Land SC, Wilson SM. Epithelial Na+ channel activity in human airway epithelial cells: the role of serum and glucocorticoid-inducible kinase 1. Br J Pharmacol 2012;166:1272–1289. |
| 49. | Bird AD, McDougall AR, Seow B, Hooper SB, Cole TJ. Glucocorticoid regulation of lung development: lessons learned from conditional GR knockout mice. Mol Endocrinol 2015;29:158–171. |
| 50. | Justus G, Sloboda DM, Henrich W, Plagemann A, Dudenhausen JW, Braun T. Avoiding the prenatal programming effects of glucocorticoids: are there alternative treatments for the induction of antenatal lung maturation? J Perinat Med 2015;43:503–523. |
| 51. | Neubauer D, Korbmacher J, Frick M, Kiss J, Timmler M, Dietl P, Wittekindt OH, Mizaikoff B. Deuterium oxide dilution: a novel method to study apical water layers and transepithelial water transport. Anal Chem 2013;85:4247–4250. |
| 52. | Althaus M, Clauss WG, Fronius M. Amiloride-sensitive sodium channels and pulmonary edema. Pulm Med 2011;2011:830320. |
| 53. | Shaykhiev R, Otaki F, Bonsu P, Dang DT, Teater M, Strulovici-Barel Y, Salit J, Harvey B-G, Crystal RG. Cigarette smoking reprograms apical junctional complex molecular architecture in the human airway epithelium in vivo. Cell Mol Life Sci 2011;68:877–892. |
*These authors contributed equally to this work.
This work was supported by Ministry of Science, Research, and the Arts of Baden-Württemberg grant Az: 32-7533-6-10/15/5, and by Deutsche Forschungsgemeinschaft grants DI1402/3-1 and SFB1149/1.A05.
Author Contributions: Conception and design—F.K., H.S., M.F., P.D., and O.H.W.; performed experiments—F.K., H.S., P.B., V.E.W., K.E.T., and O.H.W.; data analysis and interpretation—F.K., H.S., P.B., V.E.W., and O.H.W.; manuscript—F.K., H.S., K.E.T., M.F., P.D., and O.H.W.
This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1165/rcmb.2015-0071OC on October 16, 2015
Author disclosures are available with the text of this article at www.atsjournals.org.
