dsRNA-induced changes in gene expression profiles of primary nasal and bronchial epithelial cells from patients with asthma, rhinitis and controls

Viral respiratory tract infections are the most common cause of asthma exacerbations [1], with rhinovirus (RV) being the most prominent virus involved. Patients with asthma are not at increased risk of a viral infection as compared to healthy controls, but in case of an upper respiratory infection they are more prone to develop a lower respiratory tract infection with more severe symptoms [2]. This suggests that host characteristics are contributing to the clinical responses to virus infections in asthma.

The majority of patients with asthma is sensitized to environmental allergens, which is associated with allergic airways inflammation [3]. There is increasing evidence of interaction between viral infection and allergic sensitization, with increased risk for a more severe exacerbation if patients with asthma are atopic [4]. Since asthma and rhinitis co-exist to variable degrees [5], the host characteristics in upper and lower airways when encountering a respiratory virus are likely to vary to a similar extent. In order to develop effective interventions for the prophylaxis and treatment of virus-induced exacerbations in asthma, it is mandatory to map the responses of both upper and lower airways to respiratory viruses in patients with asthma and/or rhinitis.

The upper and lower airway epithelial layer is constantly exposed to viruses, bacteria and allergens, and represents the first line of defence. Experiments with nasal and bronchial epithelial cells from the same individuals suggested greater susceptibility of the lower airways to RV because of differences in electrical resistance and viral replication between upper and lower airway epithelium, though no differences were observed between asthma and healthy controls [6]. However, there is increasing evidence that pre-existing asthma affects epithelial cytokine responses to RV, such as impaired interferon (IFN) production and differences in expression of genes involved in immune response and airway remodelling [7‐9]. Collectively, these studies suggest that the complex role of the airway epithelium in response to viruses is affected by asthma- and allergy-related changes in gene expression.

Double-stranded RNA (dsRNA) is produced during RV replication and is an important stimulus of the host immune response [10]. Therefore, we hypothesized a) that upper and lower airway epithelium exhibit differential responses to dsRNA infection and b) that this is modulated by the presence of asthma and allergic rhinitis. To that end, we aimed to compare dsRNA-induced gene expression profiles of cultured primary nasal and bronchial epithelial cells obtained from patients with asthma with concomitant allergic rhinitis, patients with allergic rhinitis alone, and healthy controls. The observed gene-expression profiles can be used to delineate the critical pathways that are operative in epithelial cells after virus infection in patients with and without pre-existing airways disease.

The results of the validation by independent real time PCR are presented in Additional file 1: Table S1 and Additional file 2: Figure S1 of the online supplement.

This study shows the poly(I:C)-induced gene expression profiles of both upper and lower airway epithelium of patients with asthma, allergic rhinitis and healthy controls. The transcriptional response to poly(I:C) was characterized by a strong induction of genes. Among these genes were those involved in the response to virus, apoptotic processes and antigen presentation. Although the majority of genes involved in the response to poly(I:C) were similarly induced in upper and lower airways in all groups, we also observed differential expression, in particular with regard to impaired interferon expression in asthma. Furthermore, in contrast to healthy controls and rhinitis patients, the upper and lower airways of patients with asthma did not show poly(I:C)-induced down-regulation of mitochondrial genes. These findings are indicative of mitochondrial dysfunction in airway epithelium of patients with allergic asthma, and identify genes that may play a role in the altered viral response of diseased upper as well as lower airway epithelium.

To our knowledge, this is the first study that extensively profiles gene expression by microarray of the combined upper and lower airways epithelium in response to dsRNA of healthy individuals and patients with allergic rhinitis with or without allergic asthma. In our earlier study we describe the differences in gene expression profiles at baseline levels [11]. Interestingly, differential expression between the groups is mainly observed in both upper and lower airways, resulting in primarily comparable responses to dsRNA infection by the upper and lower airway epithelium within subjects. Furthermore, only small differences are observed between healthy controls and allergic rhinitis patients. Apparently, a pre-existing inflammatory background in allergic rhinitis does not grossly affect the response to poly(I:C). This suggests that the inflammatory pathways as induced by allergens and rhinovirus are mostly diverse.

Our finding of similar induction of genes involved in inflammatory responses in the airway epithelium in all three groups extends previous results. In a previous study examining the transcriptional response of RV-infected primary bronchial epithelial cells, pro-inflammatory pathways were similarly induced in asthma and controls [7]. Our data show that this also holds for allergic rhinitis. Hence, according to these results, the majority of inflammatory genes in epithelial cells are induced by viruses in both the healthy and diseased states. However, this can not be extrapolated to COPD, in which bronchial epithelial cells demonstrated enhanced pro-inflammatory and antiviral reactions to RV as compared to healthy controls [21].

Still, we also observed several interferon related genes that were induced in healthy controls and rhinitis patients, though not in asthma patients (especially the lower airways). Interferon-β1 (IFNB1) was significantly up-regulated in all cultures of all groups except for the lower airways of asthma patients. IL28B, known as interferon-λ3, was also significantly induced in both upper and lower airways of allergic rhinitis patients and controls, though not in asthma patients. This impaired induction in asthma was not due to high baseline gene expression prior to stimulation. IFN-βs and IFN-λs play a major role in the host defense against respiratory viral infections [8, 22]. These results confirm and extend previous studies on RV-infected primary bronchial epithelial cells, demonstrating reduced IFN-β [9] and IFN-λ [8] response in asthma, suggesting a higher susceptibility to viral infections and thereby to exacerbations. This may partly explain the high correlation between viral respiratory tract infections and asthma exacerbations [1]. Nevertheless, when considering the most highly up-regulated genes, a considerable proportion of interferon-related genes appear to be induced in all groups. This fits in with very recent data in human bronchial smooth muscle cells, also showing production of IFNs by poly(I:C) [23]. Notably, the upregulation of these interferon-related genes tends to be much less in patients with asthma, although a larger fold change difference of a gene might not necessarily be linked to larger impact on a protein pathway.

The loss of mitochondrial and other metabolic gene down-regulation to dsRNA in asthma as compared to healthy controls and allergic rhinitis is in keeping with previously reported modifications in mitochondrial/metabolic function in A549 cells and animal models of allergic inflammation. These studies demonstrated that mitochondrial dysfunction exacerbates antigen-driven allergic airway inflammation by increased generation of reactive oxygen species (ROS), which induces oxidative stress in the lungs [24]. Environmental factors such as allergens, ozone, and viruses increase ROS production thereby interactively promoting allergic inflammation [25, 26]. Second, since viral replication is dependent on host resources, down-regulation of mitochondrial function in a healthy state may prevent energy production to provide this replication. A recent review describes previous studies that have observed modulation of mitochondrial functions during different viral infections [27]. That viral replication depends on mitochondrial biogenesis was previously observed during Human Cytomegalovirus infection [28]. Interestingly, this loss of down-regulated genes in our study was observed in both upper and lower airways of patients with asthma, suggesting changed host characteristics of the upper airways in patients with allergic rhinitis plus asthma as compared to those with allergic rhinitis alone. This implies at least an interaction between asthma and rhinitis in the response to respiratory viruses. Since we did not include asthma patients without allergic rhinitis we can only speculate about the role of the presence of allergic rhinitis in asthma. However, in our previous study we found a large impact of allergic rhinitis on the differences in epithelial gene expression between upper and lower airways, influencing the lower airways as well [11]. Therefore, we assume that allergic rhinitis affects both upper and lower epithelial responses to viruses in patients with asthma. This would further explain mainly similar responses to dsRNA by both nasal and bronchial epithelium within subject-groups.

Among the disease-specific genes that were induced in allergic rhinitis patients with or without asthma but not in healthy controls, there were genes (ERAP1, LAMP2) that have been shown to be involved in antigen presentation [29, 30]. Furthermore, several genes (SH3KBP1, CRNN, FLG, S100A7A) related to allergic inflammation [31‐34] were either induced in allergic rhinitis patients with or without asthma (SH3KB1) or in allergic rhinitis patients only (CRNN, FLG, S100A7A). The genes BMP6, CFB and MYLK have previously been associated with airway inflammation and hyperresponsiveness [35‐37]. Of these genes, CFB was induced in all patients, whereas BMP6 solely in allergic rhinitis patients and MYLK in allergic rhinitis patients with asthma. The AHNAK gene, induced in allergic rhinitis patients with or without asthma, was previously associated with asthma susceptibility [38]. Interestingly, there was also induction of ciliary genes (BBS1, NEBL, NME7) [39‐41] of which NEBL and NME7 in allergic rhinitis patients with or without asthma while BBS1 exclusively in patients with allergic rhinitis. Mucociliary clearance is essential for the pulmonary defense [42], and ciliary dysfunction has been previously related to asthma severity [43].

The strength of our study is that we have collected primary epithelial cells from both upper and lower airways from the same individuals in three different conditions (healthy, allergic rhinitis, allergic rhinitis and asthma). Furthermore, we were able to extensively analyse gene expression by using microarray, which was confirmed by PCR analysis. Nevertheless, the study has some limitations. Firstly, we included relatively few individuals per group. We carefully calculated this samples size by setting the false-discovery rate at approximately 5%. A larger sample size would have allowed capturing smaller differences in expression profile or greater differences in genes that display a large variation in expression per individual. As a consequence we purposely focused on paired differences before and after poly(I:C) stimulation instead of comparing unpaired expression between the different subject groups. In addition to the original power calculation mentioned, the statistics used to measure differential expression applied the Benjamini and Hochberg adjustment of p-values for multiple testing. This correction uses a smaller significance level, inevitably reducing the power of the analysis. This will have led to false-negative results on poly(I:C)-induced genes, but it purposely limited the risk of false positive discovery.

We did not use a real virus as a stimulus, but we used poly(I:C) instead. Although a real virus such as RV would have been preferable, in order to match the in vivo conditions, dsRNA is an adequate surrogate marker for RV. Toll-like receptor 3 (TLR3) is required for the sensing of RV produced dsRNA [10] and the TLR3 ligand poly(I:C) was found previously to be a very effective stimulus of airway epithelial cells [44, 45]. Since actual steroid exposure will change the expression of many genes, patients were prohibited steroid usage for 4 weeks before sampling. In fact, all allergic rhinitis patients were entirely steroid naïve and only one asthma patient used topical steroids and one asthma patient used inhaled steroids 4 weeks prior to recruitment. We cannot exclude carry-over effects of steroids even over this time span. Finally, culturing of cells will affect gene expression levels since conditions are no longer the same as in the airways. This seems to be inevitable. Alternative procedures to obtain epithelial cells such as laser capture or direct measurement after isolation might mitigate these effects of cell culturing, but will introduce new biases introduced by contamination by other cell types and/or the (enzymatic) isolation procedures themselves. However, as we compared epithelial cells from the same individuals, the standardized culturing of these cells should have affected nose and bronchial epithelial cells similarly.

The currently observed disease-related differences in viral-induced gene expression of upper and lower airways between patients with allergic rhinitis and asthma and healthy controls may have clinical implications. First, these results help to understand the mechanistic pathways of the mutual interaction between asthma and rhinitis, for which there is considerable clinical evidence [46]. Second, the current analysis identified several new genes whilst confirming other genes from previous findings. This may provide potential targets with respect to mitochondrial dysfunction and interferon involved genes for drug-discovery studies and for treatment of exacerbations in patients with combined upper and lower airway disease.

In conclusion, we demonstrated that there are differences between rhinitis patients with and without asthma in the epithelial expression of dsRNA-induced genes, which are related to interferons and mitochondrial function. This appears to be manifested in both the upper and lower airways, suggesting mainly comparable responses to dsRNA infection in upper and lower airway epithelium within subjects. These data identify host-virus interactions in asthma, rhinitis and controls, which is required for developing targeted preventative and therapeutic interventions in asthma exacerbations.