Abstract
Sphingolipids are involved in the pathogenesis of inflammatory diseases. The central molecule is ceramide, which can be converted into ceramide-1-phosphate (C1P). Although C1P can exert anti- and pro-inflammatory effects, its influence on cigarette smoke (CS)-induced lung inflammation is unknown. We aimed to clarify the role of C1P in the pathogenesis of CS-triggered pulmonary inflammation and emphysema in humans and mice.
The effects of C1P were addressed on CS-induced lung inflammation in C57BL/6 mice, CS extract-triggered activation of human airway epithelial cells (AECs) and neutrophils from patients with chronic obstructive pulmonary disease. Differential cell counts in bronchoalveolar lavage fluid were determined by flow cytometry and pro-inflammatory cytokines were measured by ELISA. Expression and DNA binding of nuclear factor (NF)-κB and neutral sphingomyelinase (nSMase) were quantified by PCR, electrophoretic mobility shift and fluorometric assays.
C1P reduced CS-induced acute and chronic lung inflammation and development of emphysema in mice, which was associated with a reduction in nSMase and NF-κB activity in the lungs. nSMase activity in human serum correlated negatively with forced expiratory volume in 1 s % predicted. In human AECs and neutrophils, C1P inhibited CS-induced activation of NF-κB and nSMase, and reduced pro-inflammatory cytokine release.
Our results suggest that C1P is a potential target for anti-inflammatory treatment in CS-induced lung inflammation.
Abstract
Ceramide-1-phosphate, a potential target for anti-inflammatory treatment in cigarette smoke-induced lung inflammation http://ow.ly/E7tUS
Introduction
Chronic obstructive pulmonary disease (COPD) is the third most common cause of death in the western community. Airflow limitation and chronic inflammatory response in the airways and the lungs are triggered by the inhalation of noxious particles or gases, commonly cigarette smoke (CS) [1]. This inflammation is characterised by the accumulation of macrophages and neutrophils in the lungs. Oxidative stress in the airways is an essential component of disease pathogenesis. It activates the transcription factor nuclear factor (NF)-κB in inflammatory and structural cells, inducing the production of pro-inflammatory cytokines and chemokines, such as tumour necrosis factor (TNF)-α, interleukin (IL)-1β and IL-8 [2]. The chronic inflammatory cascade leads to alveolar wall destruction and the development of emphysema [3, 4].
Recent reports suggest that sphingolipids play a role in the pathogenesis of airway inflammation [5, 6]. The central molecule ceramide triggers apoptosis of pulmonary epithelial cells, and increased pulmonary levels of ceramide lead to the development of emphysema in mice [5, 7]. Furthermore, elegant in vitro and in vivo experiments emphasise that ceramide and its derivatives modulate chronic pulmonary inflammation in bronchial asthma [8, 9] and cystic fibrosis [6]. Ceramide is de novo generated by serine palmitoyl transferase, ceramide synthase or via membrane sphingomyelin hydrolysis by sphingomyelinase (SMase). It can be directly phosphorylated by ceramide kinase to ceramide-1-phosphate (C1P) [10]. Exogenous C1P prevents ceramide generation and inhibits the apoptotic effect of ceramide in alveolar macrophages [11, 12] and may block the neutral (n)SMase2-dependent apoptosis of human airway epithelial cells (AECs) [13]. Furthermore, phospho-ceramide analogue-1, an analogue of C1P, inhibits TNF-α release in RAW264.7 mouse macrophages [14]. However, a detailed study unravelling the role of C1P in CS-induced lung inflammation in mice and humans is still missing.
The aim of this study was to investigate the effects of natural and synthetic C1P products on murine models of CS-induced emphysema, inflammatory response of human AECs and blood neutrophils from COPD patients. The underlying working mechanisms of C1P were explored by measuring DNA binding and mRNA expression of NF-κB and by assessing nSMase activity.
Materials and methods
A detailed description of the materials and methods can be found in the supplementary material.
Reagents
C8-C1P was purchased from Cayman Chemical (Ann Arbor, MI, USA) and aliquoted in ethanol. C16-C1P was acquired from Matreya LLC (Pleasant Gap, PA, USA) and solubilised in sterile nanopure water on ice using a probe sonicator as previously described [15]. Fresh lipopolysaccharide (LPS)-free cigarette smoke extract (CSE) was generated by bubbling smoke from one commercial cigarette (Marlboro; Philip Morris GmbH, Munich, Germany) at a constant rate through 10 mL RPMI with 0.1% fetal calf serum and penicillin/streptomycin. One burned cigarette is equivalent to 100% of CSE medium.
Mice
C57BL/6 mice (6–8 weeks old) were bred at animal facilities of the University Hospital Freiburg (Freiburg, Germany). All experiments were approved by the animal ethics committee of the German government.
Mouse models of acute and chronic CS-induced lung inflammation
CS-induced lung inflammation was induced as previously described [16, 17]. C1P was administered as either intratracheal or oral instillations.
Histological analysis of lungs
Formalin-fixed, paraffin-embedded lung sections were stained with haematoxylin and eosin. Mean linear intercept (Lm) and internal surface area of the lung (ISA) were calculated as previously described [17].
Cell culture experiments
Human bronchial BEAS-2B and alveolar A549 epithelial cells were cultured as previously described [18]. 5×105 cells·well−1 were seeded in a six-well plate and pre-treated with C1P prior to CSE stimulation. 6 h later, supernatants were collected for cytokine measurement. To measure NF-κB activation, cells were stimulated with 5% CSE for 10 min.
Isolation and in vitro stimulation of human neutrophils
Neutrophils were isolated using Pancoll gradient (PAN-Biotech GmbH, Aidenbach, Germany), as described previously [5], from 29 stable COPD patients (Global Initiative for Chronic Obstructive Lung Disease (GOLD) stages I, II, III and IV: n=6, n=8, n=6 and n=9, respectively), eight healthy ex-smokers and 15 healthy nonsmokers at the Dept of Pulmonology, University Hospital Freiburg. This study was approved by the local ethics committee and all participants gave written informed consent (subject characteristics are shown in the supplementary material). The purity of neutrophils was >98% as assessed on cytospin slides. Cells were seeded into 24-well plates (1×106 cells·well−1) in RPMI1640 culture medium and treated with C1P or vehicle for 1 h prior to 5% CSE for 18 h. Supernatants were collected for cytokine measurements.
Sphingomyelinase fluorometric assay
nSMase activity was measured in murine lung tissue and in human sera with the SMase fluorometric assay kit (Cayman Chemical).
Other molecular and cellular analyses
Detailed information can be found in the supplementary material.
Statistical analysis
Groups were compared using one-way ANOVA followed by the Bonferroni post hoc test; correlation was analysed with Pearson correlations (GraphPad Prism, San Diego, CA, USA). A p-value <0.05 was considered significant.
Results
Exogenous C1P inhibits CS-induced lung inflammation
Similar to previous findings, mice exposed to the smoke of five cigarettes per day for 3 consecutive days displayed higher numbers of leukocytes (neutrophils and macrophages) (fig. 1a), higher total number of cells (mean±SD 113 400±16 486) and increased levels of pro-inflammatory cytokines in bronchoalveolar lavage fluid (BALF) (fig. 1b and c) compared to air-exposed controls with a total mean±SD of cells in BALF of 33 000±10 816. Intrapulmonary application of C8-C1P (2.0 µg·kg−1) or C16-C1P (1.6 µg·kg−1) 30 min prior to CS exposure attenuated inflammation as shown by a decrease in BALF leucocytosis and total BALF cell number (smoke/C16-C1P: mean±SD 44 400±16 211; smoke/C8-C1P: mean±SD 50 400±14 449), and a fall in concentrations of BALF cytokines, notably IL-1β, IL-6, keratinocyte chemoattractant (KC) and macrophage inflammatory protein (MIP)-2. Moreover, leukocyte infiltration in the tissue was also reduced (fig. 1d–g).
Of note, the effect of C1P analogues was maintained, even when applied after the induction of CS-induced lung inflammation (starting at the fourth session of CS exposure) (supplementary material). We observed lower levels of CS-induced total BALF leukocytes (data not shown), neutrophils and macrophages (fig. 2a and c) when mice were administered with C1P before and after CSE exposure over a period of 3 days, suggesting that exogenous C1P can prevent and also treat early stages of established CS-induced lung inflammation (fig. 2a–d).
C1P partially prevents development of emphysema in a chronic model of CS-induced lung inflammation
Next, to study the involvement of C1P in the pathogenesis of emphysema, a mouse model of chronic CS-induced lung inflammation (characterised by the development of pulmonary emphysema) was used. Since repeated anaesthesia (3 days per week for 3 months) for intratracheal treatment would have been fatal for many animals, oral treatment with C8-C1P (40 µg·kg−1) or C16-C1P (50 µg·kg−1) was used. Prior to this, the relative potency of intratracheal versus oral treatment in the acute model of CS-induced lung inflammation was established. Accordingly, the oral dose of the drug with a comparable reduction in airway inflammation was more than 20-fold higher by intratracheal administration (data not shown). As shown in figure 3, animals exposed to CS for 7 months demonstrated foci of emphysema disseminated throughout the lung parenchyma (fig. 3b) compared to air-exposed mice (fig. 3a). Oral treatment for the last 3 months with C16-C1P and C8-C1P (fig. 3c and d) significantly reduced the signs of lung emphysema. The precise morphometric assessment of lung emphysema was performed as previously described [17], and the results are given in table 1.
After 7 months of smoke exposure, the average interalveolar distance (Lm) was significantly increased and the ISA was considerably decreased compared to the respective air-exposed group. The reduction of CS-induced lung emphysema and inflammation following C1P treatment was accompanied by a reduced number of total BALF leukocytes, as well as BALF macrophages and neutrophils (fig. 3e), and reduced cytokine levels of IL-1β, IL-6, MIP-2 and KC (fig. 3f and g). Furthermore, the mRNA expression level of matrix metalloproteinase (MMP)-12 was significantly higher after 7 months in CS-exposed mice compared to air-exposed mice. Administration of C1P attenuated MMP-12 mRNA expression in CS-induced lung inflammation (fig. 3h). Additional immunohistochemistry staining of CD3 and CD20 identified lymphocytes and the appearance of lymphoid aggregates and follicles in lung tissue of mice. No lymphoid follicles were found in air-exposed mice (fig. 3i). The presence of lymphoid aggregates in CS-exposed mice was reduced by treatment with C1P (fig. 3i). Our study showed an increase in the B-cell attracting chemokines CXCL13 and CXCR5, which are found in areas of inducible bronchus-associated lymphoid tissue (iBALT), in the lungs of CS-exposed mice after 7 months. In general, we were not able to detect the typical formation of iBALT in C1P treated mice (fig. 3i). Hence, C1P reduces the development of iBALT in a mouse model of chronic CS-induced lung inflammation due to attenuated mRNA expression of CXCL13 (fig. 3h).
C1P reduces CS-induced NF-κB expression and activity in lung tissue and human AEC lines
CS exposure is associated with the activation of the NF-κB signalling pathway in the lungs of humans and mice [19, 20]. Thus, we next questioned whether C1P can decrease CS-induced NF-κB DNA binding in murine lung tissue. As shown in figure 4a and b, both C8-C1P and C16-C1P partially inhibited the pronounced NF-κB DNA binding in the lung tissue induced by acute CS exposure. Furthermore, quantified PCR analysis revealed that the protective role of C1P can also be linked to the reduced mRNA expression of NF-κB1 (fig. 4c) and NF-κB target genes, including MMP-9, MYD88 and TNF receptor-associated factor (TRAF) 2 expression (fig. S1).
To exclude an indirect effect of C1P, we analysed the influence of C1P on NF-κB DNA binding in A549 and BEAS-2B cells. For this purpose, cells were pre-incubated with the indicated concentration of C8-C1P, C16-C1P or vehicle prior to stimulation with CSE (5%) for 10 min. Upon CSE stimulation, strong NF-κB DNA binding was observed in BEAS-2B (fig. 4d) and A549 (fig. 4e) cells compared to vehicle-treated cells. Interestingly, both C1P analogues significantly reduced CSE-induced NF-κB DNA binding in a dose-dependent manner (fig. 4f and g).
Effect of C1P on CS-induced cytokine production and reactive oxygen species release in human AEC lines
Based on the observation that C1P reduces CS-induced NF-κB DNA binding in human AECs in vitro, we then investigated whether C1P consequently modulates production of pro-inflammatory cytokines (IL-6/IL-8) in AECs. Therefore, A549 and BEAS-2B cells were either treated with the indicated concentration of C8-C1P, C16-C1P or vehicle for 1 h prior to CSE stimulation. Supernatants were collected after 6 h and cytokines were quantified. As shown in figure 5, C8-C1P and C16-C1P significantly inhibited the CSE-induced production of IL-6 and IL-8. Additional reactive oxygen species (ROS) production by human AECs was determined. C1P was able to diminish CSE-induced ROS release in A549 and BEAS-2B (fig. S2).
C1P attenuates CSE-induced NF-κB expression and inhibits IL-8 production in human neutrophils
Besides AECs, neutrophils are also a major source of airway IL-8 in COPD. Therefore, we investigated the effect of C1P on CSE-induced IL-8 production in human neutrophils isolated from nonsmoking controls, ex-smoking controls and COPD patients. In line with the findings from AECs, C16-C1P and C8-C1P dose-dependently inhibited CSE-induced release of IL-8 by neutrophils of control subjects and COPD patients with mild-to-moderate (GOLD stages I and II) and severe (GOLD stages III and IV) disease (fig. 6a–d). The efficacy of C1P treatment in blocking IL-8 production was similar between the control subjects and COPD patients with mild-to-moderate (GOLD stages I and II) and severe (GOLD stages III and IV) disease. Transcriptional analysis revealed that C1P also reduced CSE-induced NF-κB1 level in neutrophils of COPD patients (fig. 6e).
Increased nSMase activity/expression in murine lung tissue with emphysema and CS-exposed human granulocytes and sera
It has previously been reported that CSE increases nSMase activity in AECs, resulting in an increased level of ceramide and epithelial cell apoptosis [5, 13]. Thus, we analysed nSMase activity in the lungs of mice with CS-induced lung inflammation/emphysema, as well as in the serum of 22 COPD patients, eight ex-smoker controls and 11 nonsmoker controls. As shown in figure 7, increased nSMase activity was detected in lung tissue of CS-exposed mice compared to air-exposed littermates (fig. 7a) and also in the sera of COPD patients compared to nonsmokers and ex-smoker controls (fig. 7b). Interestingly, there was a significant negative correlation between human serum nSMase levels and post-bronchodilator forced expiratory volume in 1 s (FEV1) % predicted in COPD patients (r= −0.57; p<0.01) (fig. S3a) and also in the entire study population (r= −0.75; p<0.001) (fig. S3b).
Finally, we addressed the modulatory effect of C1P on CS-triggered nSMase activity/expression in the lung tissue of mice and human neutrophils. C8-C1P and C16-C1P significantly attenuated CS-induced nSMase activity in murine lung (fig. 7a) and nSMase2 mRNA expression in human neutrophils of COPD patients (fig. 7c).
Discussion
Sphingolipids, such as sphingosine-1-phosphate (S1P) and C1P, have been implicated in the regulation of immune response during infections, as well as in inflammatory disorders such as atherosclerosis, rheumatoid arthritis, asthma, acute respiratory distress syndrome or cystic fibrosis [9, 21–24]. It is remarkable that while the role of S1P and S1P receptor-agonists/antagonists in the pathogenesis of inflammatory disorders has been extensively studied in vivo using translational mouse models of human diseases [9, 10, 25], the knowledge concerning the function of exogenous C1P in modulating immune responses is, at present, mostly limited to in vitro observations [26, 27]. Both pro- and anti-inflammatory behaviour of exogenous C1P has been described. For instance, C1P stimulates eicosanoid (e.g. prostaglandin E2) biosynthesis, and promotes phagocytosis of neutrophils, migration of macrophages and degranulation of mast cells [15, 22, 24, 28]. Conversely, C1P inhibits LPS-triggered production of pro-inflammatory cytokines such as TNF-α, IL-6 or IL-5/IL-13 in macrophages or mast cells, as well as reducing acid SMase activity [10, 22, 23, 29].
Here, we report for the first time that both the synthetic C1P analogue C8-C1P and the natural analogue C16-C1P exert anti-inflammatory effects in a mouse model of CS-induced lung inflammation and emphysema. Intrapulmonary and systemic oral treatment attenuates CS-induced recruitment of leukocytes to the lung and production of pro-inflammatory cytokines, such as KC, MIP-2, IL-6 and IL-1β, as well as development of lung emphysema. In addition, CS-induced mRNA levels of MMP-9 and MMP-12 are diminished by administration of C1P. Therefore, C1P could prevent tissue injury and remodelling due to the reduced matrix proteolysis. Furthermore, we recognised a link between iBALT formation and emphysema development that supports the study of Bracke et al. [30], who described an increase of CXCL13 in patients with COPD and how crucial CXCL13 is in the development of CS-induced lymphoid follicles in mice. Treatment with C1P attenuates the presence of CS-induced accumulation of lymphocytes and CXCL13, and its receptor CXCR5.
Mechanistically, we provide evidence that C1P reduces CS-induced expression and DNA binding of NF-κB in structural lung cells in vivo and in vitro. In connection with this, nSMase2 has recently been reported to play a role in CS-triggered apoptosis of AECs [31]. Therefore, it is of significance that C1P inhibits CS-stimulated nSMase activity and expression in murine lungs and human neutrophils, respectively. We observed increased nSMase activity in the serum of patients with COPD compared to corresponding controls, which was negatively correlated to FEV1 % pred. Overall, our findings suggest that synthetic and natural C1P analogues might be considered as new therapeutic agents for CS-induced lung inflammation and emphysema.
NF-κB is one of the key transcription factors involved in cellular responses to CS in the lungs [19, 20, 32]. Its activation leads to the induction of multiple genes associated with innate immunity, cellular inflammation and repair processes. In the present study, CS-induced NF-κB DNA binding and NF-κB expression in the lungs were reduced when mice were treated with C8-C1P or C16-C1P, suggesting that C1P might exert its anti-inflammatory capacity by suppressing the NF-κB pathway. However, the decrease in NF-κB DNA binding might also be due to the attenuation in CS-induced lung inflammation observed upon C1P treatment. Thus, to better define the capacity of C1P to modulate CS-induced NF-κB transformation, in vitro experiments with CSE-stimulated human A549 and BEAS-2B AECs were performed. Strikingly, both C8-C1P and C16-C1P were dose-dependently able to attenuate CSE-triggered translocation of NF-κB into the nucleus in epithelial cell lines, implying that C1P, via interfering with the NF-κB pathway, regulates CS-induced lung inflammation. In this context, C1P reduces CS-triggered induction of NF-κB target genes expression such as Mmp9, Myd88 and Traf2 (supplementary material). Unfortunately, TNF receptor 1 and 2 were not significantly increased in our mouse models (data not shown), thus it was not possible to determine the effect of C1P on mRNA expression.
Interestingly, similar effects of C8-C1P and/or C16-C1P on LPS-induced NF-κB activation in human embryonic kidney cells and primary human peripheral blood mononuclear cells have been demonstrated [33]. However, other studies reported that high concentrations of C16-C1P (15 or 20 µM) stimulated NF-κB activation in the alveolar rat macrophage cell line NR8383 [12] and murine macrophage cell line J774A.1 [15]. Nevertheless, Goldsmith et al. [14] did not observe any effect of phospho-ceramide analogue-1, a synthetic C1P analogue, on LPS-induced NF-κB activation in the murine macrophage RAW264.7 cell line.
An explanation for this discrepancy might be the different C1P concentration (1–20 µM) [12, 15, 33] and the C1P analogue, as well as the different cell types used in these studies. Of note, the solvent in which C1P is dispersed has been reported to be critical for its pro- or anti-inflammatory action. For instance, C1P was often dissolved in either methanol/dodecane [11] or ethanol/dodecane [34], and dodecane is known to influence the effect of C1P [29]. To avoid this, we used C16-C1P sonicated in sterile water or C8-C1P dispersed in ethanol/PBS or medium (1:2000–1:200 000 dilution).
Since NF-κB pathway activation leads to the induction of pro-inflammatory cytokines in various cell types [4, 18, 23, 35], we investigated whether the CSE-induced production of IL-8 and IL-6 by A549 and BEAS-2B cells is also regulated by C1P. Indeed, C8-C1P and C16-C1P dose-dependently inhibited CSE-triggered production of IL-8 and IL-6 in AECs.
Furthermore, we have shown an increase of ROS level produced by A549 and BEAS-2B when they were stimulated with CSE. The reduced ROS release through C1P stimulation could prevent cellular damage or accumulation of oxidative stress.
Nevertheless, oxidative stress in the airways following CS exposure also causes activation of an endogenous membrane nSMase. Chen et al. [35] found that peptidoglycan- or LPS-stimulated RAW264.7 macrophages release ceramide through the activation of nSMase, which further initiates the NF-κB pathway [35]. Thus, the stimulation of nSMase seems to be required for CS-induced apoptosis of AECs [13, 31]. Of note, Filosto et al. [31] found higher levels of nSMase2 in lung tissue of emphysema patients in comparison to control subjects. Thus, nSMase has recently gained attention as a new therapeutic strategy for CS-induced lung inflammation and lung injury. Strikingly, we found elevated nSMase activity in the lungs of mice chronically exposed to CS and, more importantly, significantly higher levels of nSMase in the serum of patients with COPD compared to never- and former smokers. Notably, enzyme activity was negatively correlated with airflow limitation (FEV1 % pred), further underlining the potential role of nSMase in the pathogenesis of COPD. Moreover, treatment of mice with both C1P analogues decreased nSMase activity in the lungs, demonstrating an additional anti-inflammatory action of C1P. We suggest that the inhibition of nSMase by C1P decreases ceramide level and consequently influences NF-κB signalling and the development of emphysema. However, to fully clarify whether ceramide or one of its metabolites regulates these effects, further studies investigating the total ceramide level in the context of CS-induced lung inflammation and the treatment with C1P are needed. Interestingly, an inhibitory effect of C1P has previously been reported for another isoform of the enzyme, the acid SMase in murine bone marrow macrophages [11], although its activity is not modified in CS-induced lung inflammation [31]. Furthermore, it has been shown that C1P can directly interact with serine palmitoyl transferase to block its activity [12]. Hence, we suggest C1P might also inhibit the nSMase isoform due to interaction with this ceramide-generating enzyme. Further analysis could reveal the level of C1P, ceramide, S1P, ceramide kinase, SMase and glycosylceramide synthase in the lungs and plasma of healthy controls compared to COPD patients, leading us to fully understand the complex metabolism of sphingolipids. Additionally, it would be interesting to investigate how exogenous C1P modifies the level of these lipids so that a concrete target can be found for future drug development.
It is worth mentioning that the anti-inflammatory effect of C1P on CS-induced lung inflammation is not limited to the modulation of structural cells, as both C8-C1P and C16-C1P also reduce CSE-triggered upregulation of nSMase and NF-κB expression, and consequently IL-8-production, in human neutrophils. Neutrophils are known to play an essential role in the pathogenesis of COPD and, moreover, the modulation of the CXCR2/IL-8 axis has previously been implicated as a therapeutic strategy for COPD [36]. Thus, our observation that C1P inhibits CS-induced IL-8 production in AECs and neutrophils further strengthens our assumption that C1P might be a new therapeutic target for COPD. The potential of C1P to dampen pro-inflammatory cytokine production has also been reported for LPS-induced TNF-α, MIP-2, IL-6 and IL-8 in human embryonic kidney cells, human blood mononuclear cells, endothelial cells and mastocytes [27, 33, 37].
COPD is a complex obstructive pulmonary disease characterised by irreversible and progressive airflow limitation. Although the effect of C1P on the progression of COPD was characterised by pulmonary inflammation, air space enlargement and parenchymal destruction, there is still need for further analysis of the respiratory function of mice to consider the clinical relevance of how C1P treatment affects respiratory mechanics.
In summary, we have demonstrated for the first time that C1P analogues inhibit CS-induced lung inflammation and emphysema in mouse models of COPD. Mechanistically, we showed that C1P interferes with CS-induced activation of the NF-κB pathway, consequently inhibiting pro-inflammatory cytokine production (such as IL-6 and IL-8) by human AECs and blood neutrophils, and also attenuating CS-induced induction of nSMase activity in the lungs and neutrophils of COPD patients. Thus, C1P analogues might hold promise as novel treatment options in COPD.
Acknowledgements
We would like to thank Barbara Schuler (Dept of Pharmaceutical Biology and Biotechnology, University Hospital Freiburg, Freiburg, Germany) for assistance in performing the electrophoretic mobility shift assays and Madelon Hossfeld (Dept of Pneumology, COPD and Asthma Research Group, University Hospital Freiburg, Freiburg, Germany) for support with the immunohistochemistry.
Footnotes
This article has supplementary material available from erj.ersjournals.com
Support statement: This study was funded by the Deutsche Forschungsgemeinschaft DFG (grant ID 7/8-1). Z. Lazar was a recipient of the European Respiratory Society Long-term Research Fellowship (grant ID 50-2012). Funding information for this article has been deposited with FundRef.
Conflict of interest: None declared.
- Received April 30, 2014.
- Accepted November 6, 2014.
- Copyright ©ERS 2015