CF AIRWAY EPITHELIA DISPLAY EXAGERRATED HOST DEFENSE RESPONSES AND PROLONGED CILIA LOSS DURING RSV INFECTION

Abstract

BACKGROUND:

In individuals with cystic fibrosis (CF), respiratory viral infections frequently result in hospitalization and have been linked to secondary bacterial infection and colonization, highlighting viral infections as possible contributors to CF lung disease progression. We hypothesized that expression of antiviral host defense genes is dysregulated in CF airway epithelia.

METHODS:

We infected primary CF and Non-CF airway epithelia with respiratory syncytial virus (RSV) and characterized their responses at 12 hr, 24 hr, 48 hr, 72 hr, and 120 hr post infection (hpi) by RNA sequencing (RNAseq).

RESULTS:

Our analysis revealed strikingly different gene expression profiles for the CF and Non-CF epithelia over the course of the infection. While both CF and Non-CF cells exhibited an early signature for interferon signaling and antiviral defense pathways, this response was relatively exaggerated and sustained in CF epithelia. We also observed, in both genotypes, a transient downregulation of cilia-associated genes and loss of ciliary activity by 72 hpi. Interestingly, recovery of cilia activity was delayed in the CF epithelia.

CONCLUSIONS:

These findings further our understanding of innate immune dysfunction in the CF airway epithelium and suggest that virus-induced cilia injury may further compromise host defenses in CF airways.

1. Background

Cystic fibrosis (CF) is caused by mutations in the gene encoding the cystic fibrosis transmembrane conductance regulator (CFTR), an anion channel localizing to the apical surface of epithelia in multiple tissues. While CF impacts several organ systems, the majority of morbidity and mortality arises from its pulmonary manifestations. These include progressive lung disease characterized by bacterial infection and inflammation, mucus accumulation, and extensive airway remodeling. Over time, these problems lead to frequent pulmonary exacerbations and a measurable decline in lung function as measured by the forced expiratory volume in 1 second (FEV₁) and can also contribute to low body weight. CFTR modulators have significantly improved these clinical outcomes; however it is currently less clear whether CFTR correction/modulation improves long-term control of microbial pathogens and this remains a challenge in the management of CF (1).

There is evidence that, in addition to impairment of bacterial host defenses, diminished antiviral host defense is a core feature of CF airway disease (2). Clinicians have long recognized the contribution of viral infections to disease burden in the CF population, particularly in early life. Respiratory viruses such as respiratory syncytial virus (RSV), rhinovirus, influenza, metapneumovirus, parainfluenzaviruses, adenovirus, and cold-causing coronaviruses are frequently isolated from CF airways and are often associated with pulmonary exacerbations (3, 4). Clinical studies indicate that while CF and Non-CF individuals are infected with respiratory viruses at similar rates, people with CF (pwCF) often present with comparatively more severe symptoms, higher viral loads, and prolonged recovery times, as well as increased likelihood of lower respiratory tract involvement and hospitalization (5–7). Limited studies of cultured airway epithelia derived from pwCF show that CF cells fail to restrict respiratory viral replication as effectively as Non-CF, pointing to an epithelial defect in antiviral defenses (8–10).

Importantly, there is evidence that viral infections may predispose CF airways to subsequent bacterial infection, and likely contribute to disease progression. Acquisition of new bacteria in pwCF is frequently associated with recent viral/upper respiratory tract illness (11, 12), a relationship supported by epidemiological observations showing that Pseudomonas aeruginosa colonization varies seasonally coincident with the cold and flu season (13). These data suggest that viral infections - both in childhood and throughout the life of pwCF - may potentiate CF lung disease by promoting establishment of bacterial infection and inflammation in the airways.

To better understand the differing outcomes observed with viral infection in pwCF, we used RSV as a model viral pathogen to investigate the antiviral host defense responses of CF and Non-CF airway epithelia. RSV is a single-stranded, negative-sense RNA virus in the Paramyxoviridae family. It circulates seasonally in the general population, with infection rates peaking in the winter months. RSV infection causes cold-like symptoms including rhinorrhea, cough, and fever; in some individuals, infection can spread to the lower respiratory tract, leading to more serious conditions such as bronchiolitis that may require hospitalization (14). While RSV may infect a person at any age, it presents particular risks for young children and older adults. Each year in the United States, RSV infections cause an estimated 58,000–80,000 hospitalizations in children under 5 and 100,000–160,000 hospitalizations in adults older than 60 (https://www.cdc.gov/rsv/php/surveillance/index.html). In the CF population, RSV is among the viral pathogens that can be detected in CF airway secretions and is thought to contribute to pulmonary exacerbations and lung function decline over time (3–5, 15).

Here, we performed RSV infections in well-differentiated primary airway epithelial cultures derived from human CF and Non-CF lungs and grown at an air-liquid interface (termed human airway epithelia; HAE). We found that RSV replication was increased in CF HAE relative to Non-CF. We used bulk RNA sequencing to generate transcriptional profiles for the CF vs Non-CF epithelia. We observed dramatic genotype-dependent differences in gene expression at all timepoints post infection, with the most significant differences involving the linked processes of interferon signaling and regulation of airway cell ciliation. In subsequent experiments, we found that RSV-induced loss of ciliary activity was relatively prolonged in CF epithelia, a previously unrecognized phenomenon that may contribute to the severity of respiratory viral infection and predisposition to secondary bacterial infection seen in pwCF.

2. Methods

Additional detailed methods are provided in the supplementary material.

Human airway epithelial cell culture

Primary human airway epithelial cultures were established using human donor lungs obtained by the University of Iowa Cells and Tissue Core. Cultures were grown at the air-liquid interface (ALI) as described previously (16). Briefly, airway epithelial cells were dissociated from human CF and Non-CF bronchial tissue by pronase enzyme digestion. Freshly isolated epithelia were then seeded onto semi-permeable polycarbonate membrane inserts (pore size 0.4 μm; Millipore, Billerica, MA, USA) pre-coated with human placental collagen (Type IV, Sigma-Aldrich, St. Louis, MO, USA). Cells were initially seeded in culture medium consisting of DMEM/F-12 medium supplemented with 5% fetal bovine serum, penicillin (50 units/mL), streptomycin (50 μg/mL), gentamicin (50 μg/mL), fluconazole (2 μg/mL), and amphotericin B (1.25 μg/mL). Twenty-four hours after seeding, the mucosal medium was removed and cells were allowed to grow at ALI. From this point forward, cultures were maintained in DMEM/F-12 medium supplemented with 2% Ultroser G (USG; Pall Biosepra, Cergy, France) and the above listed antibiotics. Only well-differentiated cultures (>2 weeks old) were used in these studies. The presence of tight junctions was confirmed by measuring transepithelial resistance using a volt-ohm meter (World Precision Instruments, Sarasota, FL, USA; resistance >500 Ω·cm²). These studies were approved by the University of Iowa Institutional Review Board and Animal Care and Use Committee.

Viral infections and GFP imaging

Our RSV infection protocol utilized recombinant RSV strain A2 encoding a GFP reporter upstream of the NS1 open reading frame (17). Prior to infection, cultured cells were washed 3X apically using PBS containing containing Ca²⁺ and Mg²⁺. The virus was diluted in serum-free MEM and applied to the apical surface at an MOI of 1, in a total volume of 50 μL per culture. Cultures were incubated with virus at 37°C, 5% CO₂ for 2 hours. After 2 hours, unbound virus was removed and the apical surface of the cultures was again washed 3X with PBS containing Ca²⁺ and Mg²⁺. Infected cultures were maintained in a 37°C, 5% CO₂ incubator for the duration of the experiment, and cells were imaged at various timepoints post infection using an inverted stage microscope (Olympus IX70; Olympus Corporation, Center Valley, PA, USA). Image J (18) was used to process images and quantitate mean fluorescence intensity.

3. Results

3.1. RSV infection is increased in CF epithelia, relative to Non-CF

We infected CF and Non-CF HAE with recombinant RSV encoding a GFP reporter (17) at an MOI of 1, and monitored the resulting infection over a 5-day timecourse. In this model, RSV infection results in distinct fluorescent foci that become visible around 24 hr post infection (hpi) and generally reach peak fluorescence intensity by 48 hpi (Figure 1A). Notably, the abundance of these fluorescent foci, as measured by mean fluorescence intensity, was greater in the CF cells relative to Non-CF at each timepoint (Figures 1A, 1B). We also used qRT-PCR to measure the abundance of viral RNA in the infected epithelia. In Non-CF HAE, RSV nucleocapsid (N) gene copy numbers peaked at around 72 hpi then leveled off. In CF epithelia, viral RNA abundance was still climbing at 120 hpi (Figure 1C). Immunoblotting for the viral nucleoprotein indicated a trend toward relatively greater viral protein production in the CF donor cultures throughout the timecourse, resulting in markedly increased viral loads by 120 hpi (Figure 1D). Together, these results suggest that CF epithelia had greater viral burdens than the Non-CF cells, despite receiving the same inocula.

Figure 1. Human CF airway epithelia support relatively greater levels of RSV infection, compared to Non-CF epithelia.

Primary human airway epithelia (HAE), representing 4 Non-CF and 4 CF donors, were infected with RSV at MOI of 1. (A) En face views of infected cultures at the indicated timepoints post infection, showing spread of fluorescent foci over time. Each row represents an individual HAE donor. Scale bar = 1.0 mm (B) Image J was used to quantitate mean fluorescence intensity at each timepoint in the images from infected HAE cultures. Data were derived by averaging 3 infected filters/donor/timepoint. (C) Quantitative RT-PCR was used to measure levels of the RSV N gene in infected cultures over time. (D) Viral loads were also assessed by immunoblotting for the RSV nucleoprotein in lysates from infected CF and Non-CF epithelia (20 μg total protein per lane). Nucleoprotein abundance was normalized to the loading control (vinculin) for densitometric analysis. In (B), (C), and (D), data are presented as median and interquartile range, with whisker boundaries determined by Tukey’s method. Mean is represented with a “+” symbol. Multiple unpaired t-tests were performed on log transformed data to test for statistically significant differences between CF and Non-CF at each time point. *P_Adj < 0.05, **P_Adj < 0.01.

We next asked whether CF epithelia might have an increased abundance of RSV-susceptible cells. RSV preferentially infects ciliated airway epithelial cells (19). In an earlier study using this HAE model, we found that ciliated cells were modestly less abundant in CF cultures relative to Non-CF (20), suggesting that increased ciliated cell abundance in CF is an unlikely explanation for the observed difference in RSV infection. We also investigated whether there is increased abundance of RSV receptors on ciliated cells in CF. We analyzed a single cell RNAseq (scRNAseq) dataset characterizing gene expression in air-liquid interface (ALI) primary cultures of human CF and Non-CF airway epithelia (the same culture conditions used in this study; GEO accession GSE159056). Our analysis indicated that the proposed RSV receptors with the greatest expression in ciliated cells in the HAE model are insulin like growth factor 1 receptor (IGF1R) and nucleolin (NCL) (Supplementary Figure S1). While both of these molecules are known to facilitate RSV entry in airway cells, blockade of either receptor alone is not sufficient to fully prevent RSV infection, suggesting that RSV likely can make use of multiple routes of entry into airway epithelia (21, 22). We found that transcripts for IGF1R and nucleolin are similarly abundant in ciliated cells from CF and Non-CF cells (Supplementary Figure S1).

3.2. Transcriptional profiling reveals distinct responses to RSV in CF and Non-CF epithelia

We performed bulk RNAseq using RNA from the 4 CF and 4 Non-CF donors represented in Figure 1, at 0, 12, 24, 48, 72, and 120 hpi (with “0 hpi” representing the uninfected control condition for each donor). Genotype and demographic characteristics of the HAE donors used for the RNAseq study are presented in Table 1. Viral loads were investigated by quantifying transcripts for RSV genes in the RNAseq dataset. Consistent with the RSV N gene levels measured by qRT-PCR (Figure 1C), all viral RNAs were detected at relatively higher levels in the CF epithelia at each timepoint, suggesting that viral replication and/or spread was more restricted in Non-CF cells (Figure 2A and Supplementary Figure S2).

Table 1.

Characteristics of human airway epithelial donors used for transcriptional profiling

	Age	Sex	CFTR Genotype	Smoker	Chronic lung disease
NCF1	20	M	n.d.	N	-
NCF2	35	M	n.d.	N	-
NCF3	22	M	n.d.	Y	-
NCF4	39	F	n.d.	Y	asthma
CF1	24	F	CFTRΔF508/ΔF508	N	CF lung disease requiring lung transplant
CF2	32	F	CFTRΔF508/ΔF508	N	CF lung disease requiring lung transplant
CF3	26	M	CFTRΔF508/ΔF508	N	CF lung disease requiring lung transplant
CF4	38	M	CFTRΔF508/ΔF508	N	CF lung disease requiring lung transplant

(n.d. = not determined; “-“ indicates unknown or no information available)

Figure 2. CF and Non-CF HAE exhibit differing viral loads and distinct transcriptional profiles over the timecourse of RSV infection.

(A) Bulk RNAseq was performed in RSV-infected CF and Non-CF airway epithelia (n = 4 Non-CF, 4 CF), and viral transcripts were quantified at the indicated time points. Plot displays changes in viral load over time in the infected CF and Non-CF HAE. For each donor, transcript levels for all RSV genes were summed to obtain a measure of overall viral burden at each time point. Data are plotted as median and interquartile range, with whisker boundaries determined by Tukey’s method. Means are indicated by the “+” symbol. Data were tested for statistically significant differences between CF and Non-CF at each time point using multiple unpaired t-tests. *P_Adj < 0.05, ***P_Adj < 0.001. (B) Principal component analysis was performed to cluster samples based on global transcriptional patterns. In the PCA bi-plot, each data point represents a sample from an individual HAE donor at the indicated time point. (C) Plots comparing the numbers of differentially expressed genes in each genotype across the infection timecourse. Differential gene expression analysis was performed using DESeq2, with a fold change cutoff of 2 (Log₂ fold change = 1) to be considered differentially expressed. Fold changes represent a gene’s transcript abundance at a given time point post infection relative to its abundance in uninfected epithelia (0 hpi).

Distinct transcriptional profiles of host genes in RSV-infected CF and Non-CF epithelia were apparent from the earliest timepoint post infection (12 hpi). Using gene counts to perform principal component analysis, we found that the 48 samples could be loosely clustered based on global gene expression patterns (Figure 2B). The samples segregated primarily by genotype; within each genotype, samples clustered less distinctly according to timepoint. This suggests that genotype was the primary driver of the overall response to RSV infection in the HAE donors. Interestingly, the 0 hpi samples for the CF and the Non-CF donors formed separate clusters, implying the existence of some baseline genotype-dependent differences in transcriptional programs even in the absence of viral infection.

We next performed differential gene expression analysis to identify genes with increased or decreased transcript abundance at each timepoint post infection, relative to the uninfected (0 hpi) condition. Non-CF epithelia showed a robust early response to the virus, with the greatest number of gene expression changes observed at 12 hpi. The number of genes with increased or decreased mRNA then gradually tapered off (Figure 2C). In contrast, CF HAE exhibited the inverse: they had fewer transcriptional changes than the Non-CF at 12 hpi but showed increasingly greater changes over time. By 120 hpi, the transcriptional picture was strikingly different, with the Non-CF cells returning to baseline gene expression while a large number of transcriptional changes were occurring in the infected CF epithelia.

3.3. Activation of antiviral host defense pathways and release of inflammatory mediators is exaggerated in CF epithelia

We performed further analysis to identify gene pathways activated by RSV infection in the CF and Non-CF epithelia (full results for GO Biological Pathways showing enrichment at each time point are presented in Tables S1–S5). At the earliest timepoint (12 hpi) the transcriptional responses of the CF and Non-CF epithelia exhibited notable differences, with many genes uniquely regulated by RSV infection in each genotype (Figure 3A). In the Non-CF HAE, the strongest signature was an upregulation of genes mapping to pathways involved in cytoplasmic translation and protein synthesis (Figure 3B). Additionally, Non-CF epithelia showed enrichment of pathways regulating antiviral and antibacterial host defense, neutrophil chemotaxis, and mitochondrial energy production (cellular respiration, oxidative phosphorylation). The transcriptional changes observed in the Non-CF epithelia point to a coordinated defensive response involving a range of biological processes; in contrast, the responses of CF cells were dominated by induction of host defense pathways, including interferon signaling, apoptosis, and leukocyte migration (Figure 3B).

Figure 3. Both CF and Non-CF epithelia show upregulation of genes in antiviral and innate immune signaling pathways.

(A) Venn diagrams depicting the level of overlap in the significantly upregulated genes in CF and Non-CF HAE over the infection timecourse. (B) Summaries of GO Biological Pathways showing significant enrichment for upregulated genes in the CF and Non-CF epithelia, at each timepoint post infection. Pathways with P < 0.001 in the pathway analysis were grouped into general categories using Revigo (http://revigo.irb.hr/).

By 24 hpi, the pathways showing upregulation in CF and Non-CF epithelia became more similar (Figure 3B). In both genotypes, the prominent gene signature was activation of innate immune and antiviral host defense pathways, particularly interferon/inflammatory signaling and regulators of adaptive immune responses (Figure 3B). At all timepoints, in both genotypes, many of the transcripts with the greatest Log₂ fold changes were encoded by classical interferon-stimulated genes (ISGs). In infected Non-CF HAE, ISG expression increased steadily and peaked between 24–72 hpi, followed by a drop at 120 hpi. In CF HAE, expression of these genes peaked early (around 24 hpi), and remained elevated throughout the infection; additionally, the fold changes for these ISGs were generally higher in CF epithelia than in Non-CF at each timepoint (Figure 4A and Supplementary Figure S3).

Figure 4. RSV-induced antiviral responses are more pronounced in CF epithelia, relative to Non-CF.

(A) Scatterplot summarizing relative induction of selected antiviral host defense genes in Non-CF vs CF HAE at 24 hpi. For each gene, the mean Log₂ fold change in the Non-CF epithelia is plotted on the x-axis and the mean Log₂ fold change in the CF epithelia is plotted on the y-axis. (B) CF and Non-CF epithelia were infected with RSV at an MOI of 1 and multiplex cytokine array was used to measure cytokine release into the basolateral medium of infected cultures. Graph summarizes cytokine release at 24 hpi. Data were tested for statistically significant genotype-dependent differences using the Kruskal-Wallis test followed by Dunn’s multiple comparisons test. (C) CF and Non-CF HAE were infected with RSV (MOI of 1) and GC-MS was used to measure total tryptophan and kynurenine abundance over the infection timecourse. For each donor, activation of the kynurenine pathway was approximated by calculating the ratio of kynurenine/tryptophan at a given time point. In panels (B) and (C), data are plotted as median and interquartile range, with means indicted by the “+” symbol. In (C), log transformed data were tested for statistically significant differences between CF and Non-CF using multiple unpaired t-tests, with correction for multiple comparisons using the Holm-Sidak method. *P_Adj < 0.05, **P_Adj < 0.01.

These findings suggest that the innate antiviral immune response was relatively exaggerated in the CF epithelia. In support of this, we observed a trend toward relatively greater release of several antiviral and pro-inflammatory cytokines, most notably IL-6 and CXCL10/IP-10, into the basolateral medium from RSV-infected CF HAE (Figure 4B and Supplementary Figure S4). We also noted upregulation of the interferon-inducible gene indolamine 2,3-dioxygenase-1 (IDO1) and enrichment for gene pathways involved in tryptophan metabolism, particularly in the CF epithelia (Figures 3B and 4A, Supplementary Figure S3). Activation of the kynurenine pathway, which is initiated by IDO1-mediated conversion of tryptophan into its catabolite kynurenine, is a classic indicator of a virus-induced inflammatory response (23). We used metabolomic profiling to measure the abundance of tryptophan, kynurenine, and other products of tryptophan metabolism in cell extracts from RSV-infected CF and Non-CF HAE (Figure 4C and Supplementary Figure S5). Increased kynurenine/tryptophan ratios became evident around 24 hpi and continued to increase for the next 48 hours. This phenomenon was particularly pronounced in the infected CF epithelia (Figure 4C).

3.4. RSV-induced loss of ciliary activity is prolonged in CF epithelia

The genes and pathways showing downregulation in the CF and Non-CF epithelia over time were less similar (Figure 5A and Supplementary Figure S6; see full lists of pathway hits in Tables S2–S5). Interestingly, in both CF and Non-CF HAE, RSV infection altered the expression of many genes associated with ciliogenesis and cilia structure and/or function (Supplementary Figure S6 and Figure 5B). While this was observed in both genotypes, the effect was more pronounced in CF epithelia where it showed a different temporal pattern. In Non-CF epithelia, transcript abundance for cilia-associated genes decreased modestly early (12 hpi), and lessened as infection progressed. In CF cells, changes in cilia gene expression were minimal at 12 hpi, then increased steadily through the infection (Figure 5B). By 120 hpi, cilia genes were significantly downregulated in CF cells, accounting for much of the CF/Non-CF difference in numbers of differentially expressed genes at this timepoint.

Figure 5. CF and Non-CF HAE display divergent expression patterns for cilia-associated genes over the course of RSV infection.

(A) Venn diagrams depicting the level of overlap in the significantly downregulated genes in CF and Non-CF HAE over the infection timecourse. (B) Scatterplots summarizing downregulation of genes involved in cilia structure and function over the infection timecourse in Non-CF vs CF HAE. For each gene, the mean −Log₂ fold change in the Non-CF epithelia is plotted on the x-axis and the mean −Log₂ fold change in the CF epithelia is plotted on the y-axis. (C) High speed video microscopy was used to assess changes in ciliary activity over the infection timecourse in CF and Non-CF HAE (n = 4 Non-CF, 4 CF). Changes in cilia beat frequencies are presented in panel (C). The percentage of each field with actively beating cilia, expressed relative to baseline ciliary activity measured in uninfected HAE, is presented in panel (D). Data are plotted as mean ± SD and were tested for statistically significant differences between CF and Non-CF using the Kruskal-Wallis test followed by Dunn’s multiple comparisons test.

We investigated the functional consequences of this by monitoring cilia function over the course of RSV infection in CF and Non-CF HAE. As shown in Figure 5C, RSV infection caused cilia beat frequencies to drop to approximately 50% of baseline by 3–4 days post infection (dpi), with cilia beating remaining somewhat depressed out to 3 weeks post infection. Baseline cilia beat frequencies were not significantly different in CF and Non-CF epithelia, and RSV infection reduced cilia beating to similar degrees. In both genotypes, the proportion of the epithelial surface with actively beating cilia (active cilia beat area; Figure 5D) dropped dramatically between 0 and 4 dpi. The Non-CF epithelia began to show signs of recovery starting around 11 dpi, and by 3 weeks post infection were nearly back to baseline levels. In contrast, recovery of cilia beating was much slower in CF cells, which on average remained at approximately 40% of baseline 3 weeks after infection.

4. Discussion

This study demonstrates that, compared to Non-CF cells, CF patient-derived airway epithelia experience relatively greater viral loads when infected with a respiratory virus. The findings do not support the idea that CF epithelia become more heavily infected with RSV because of an increase in RSV target cells or viral receptor abundance; rather, they suggest that infection outcomes are determined by differences in antiviral responses downstream of viral entry that impact the ability of epithelia to restrict viral replication and spread. Notably, the differences in the transcriptional responses of CF and Non-CF epithelia are evident very early in the infection (as early as 12 hpi). In CF cells, this sets in motion a fundamentally different sequence of events, culminating in failure to limit viral replication or to resolve normally while amplifying the impact of virus-induced epithelial damage.

Our findings are also inconsistent with the idea that CF epithelia fail to restrict viral replication due to insufficient induction of an interferon response. In fact, both CF and Non-CF cells responded to RSV infection by markedly activating interferon and antiviral host defense, inflammatory, and adaptive immune response pathways. However, while the responses of CF and Non-CF were qualitatively similar, the magnitude of the responses differed. Overall, the RSV-induced innate immune response in CF epithelia was both more pronounced, and more sustained, than that of Non-CF cells. Although perturbation of interferon responses is a common finding in studies of viral infection in CF epithelia, the trends vary with pathogen and model system. While early studies suggested that antiviral host defenses may be blunted in CF (10, 24), subsequent reports paint a more complicated picture, with one recent study describing altered interferon signaling and a tendency toward more potent induction of some immune-related genes in response to influenza infection in CF cells (25). Studies with rhinovirus have variously observed overproduction, underproduction, or no difference in interferon/ISG and pro-inflammatory cytokine induction in infected CF epithelia relative to Non-CF (9, 26–28), and there is evidence that the directionality of this response may depend on the rhinovirus strain used (28).

Consistent with the overall more pronounced innate immune response in the CF epithelia, we found that kynurenine/tryptophan ratios were elevated to a greater degree by RSV infection in CF cells than in Non-CF. Dysregulation of tryptophan uptake and metabolism in CF epithelia has been noted in several previous studies, albeit with sometimes contradictory results as both increased (29) and decreased (30) IDO1-mediated kynurenine production have been observed at baseline in CF airway cells. In our study, differences in IDO1 expression and the abundance of kynurenine pathway metabolites were not seen in uninfected cells and only became apparent during RSV infection, suggesting that the genotype-dependent differences arise from differing responses to the infectious challenge rather than a deficiency in the baseline activity of this pathway. Kynurenine and its derivatives have a variety of immune-modulatory effects during viral infection, including regulating the activities of cells of the adaptive immune system (23). In a recent meta-analysis examining associations between tryptophan metabolism and COVID-19 outcomes, high kynurenine/tryptophan ratios were correlated with more severe disease, suggesting that inflammatory events driven by this pathway may worsen symptoms during respiratory viral illness when this pathway is activated at a high level. Considering this, our results suggest that overexuberant activation of the kynurenine pathway should be investigated as a possible contributor to the severity of symptoms during viral infection in pwCF.

How loss of CFTR function leads to dysregulated innate immune signaling and compromised antiviral host defenses in airway epithelia is unclear. We previously reported that the transcriptional profiles of airway tissues from newborn CF and Non-CF pigs were quite similar and only exhibited genotype-dependent differences with an inflammatory challenge (31, 32). Similarly, a recent single cell RNAseq study of newborn CF and Non-CF pigs found that loss of CFTR function had minimal effects on the large and small airway transcriptomes (33). In contrast, we noted significant baseline differences in the transcriptomes of the patient-derived CF and Non-CF epithelia used in this study (Figure 2B and Supplementary Figure S7). This observation is consistent with a model in which CF airway cells acquire transcriptomic abnormalities over time as lung disease progresses, due to infection and inflammation. Analysis of these baseline differences suggests that CF airway cells exhibit dysregulation of pathways broadly involved in influencing and maintaining differentiation, including ciliogenesis, extracellular matrix, cell-cell junction maintenance, cell proliferation, and cell migration pathways (Supplementary Figure S7, Table S6).

Collectively, these changes implicate epithelial-mesenchymal transition (EMT), a process in which cells transition from an epithelial to a more mesenchymal state to effect wound repair and regenerate tissue during an injury. Investigators have observed that, relative to Non-CF, CF airway epithelia grown at ALI exhibit a somewhat “de-differentiated” phenotype, characterized by delayed ciliogenesis and increased expression of mesenchymal marker proteins (34–36). These studies suggest that CF epithelia in ALI culture are not fully mesenchymal; rather, they display upregulation of some mesenchymal markers while also retaining expression of epithelial markers (36). To investigate this in the airway epithelial cells used in this study, we performed RT-qPCR to assess the expression levels of a panel of EMT-associated genes in the uninfected CF and Non-CF cultures (Supplementary Figure S8A). Similar to previous studies, our data suggest general dysregulation of genes involved in driving and maintaining EMT, with the CF cells simultaneously displaying gene expression associated with both the epithelial and the mesenchymal states. Most prominently, there was a significantly increased transcript abundance for the classic EMT marker vimentin in the CF epithelia relative to Non-CF, which was also apparent at the protein level as noted in previous studies (35, 36).

The literature suggests that this phenotype may originate in vivo, where chronic inflammation and airway remodeling trigger EMT in CF airways (35–37), and appears to be retained to some degree when patient-derived epithelia are differentiated in vitro (35, 36). Thus, the primary cultures of CF epithelia in the present study may have EMT-like characteristics that are maintained through epigenetic mechanisms. We speculate that this phenotype may influence immune responses to viral infection in CF airway epithelia. A recent report described altered antiviral host defense in human bronchial epithelia stimulated with TGF-β to model EMT (38). Interestingly, the TGF-β treated cells exhibited a significantly greater induction of innate immune genes than control cells when infected with rhinovirus, yet viral replication was enhanced in these cells relative to control. Similarly, enhanced RSV replication and TNF-α production have been observed in human airway cells pre-treated with TGF-β (39). This phenomenon – increased viral infection despite a robust antiviral response – mirrors our results: paradoxically, while the interferon response to RSV in the CF epithelia was relatively exaggerated, this response was comparatively ineffective at controlling viral replication in CF cultures. While it is unclear why antiviral defenses are less effective in this setting, the concept that adopting a less-differentiated, more proliferative phenotype promotes viral infection in the airway epithelium is supported by studies using rhinovirus (40, 41). Of note, it has been reported that RSV readily infects airway basal cells, a progenitor cell type in the airway epithelium (42–44).

The other major gene signature, in both genotypes, was downregulation of cilia-associated gene pathways in response to RSV. These transcriptional changes translated to loss of ciliary activity around 24–96 hpi, with a notably delayed recovery of cilia movement in the CF epithelia (Figure 5D). We previously reported a marked and rapid downregulation of cilia-associated genes in CF pig airway epithelia in response to a bacterial-derived inflammatory stimulus (31), suggesting that CF airway cells may be primed to exhibit cilia changes/damage when exposed to pathogens. Our findings are consistent with clinical literature indicating that in pwCF, recovery times are relatively long in the wake of respiratory viral infections. In addition, the persistence of this state in CF airways may contribute to their susceptibility to secondary bacterial infections during viral infection.

Interestingly, inducing an antiviral state in airway epithelia by treating with interferon or the dsRNA mimic poly(I:C) suppresses ciliogenesis and alters proportions of ciliated cells in the differentiated epithelium (44, 45), suggesting that regulation of cilia-associated genes during viral infection is tied to interferon signaling. Thus, the more pronounced downregulation of cilia pathway genes and increased recovery times in the infected CF epithelia may be directly related to the relatively greater intensity of the interferon response in this genotype. Additionally, there is evidence that CFTR is required for proper differentiation of ciliated airway epithelial cells, as the appearance of ciliated cells is delayed in differentiating ALI cultures from human CF donors (34, 35). A model emerges in which RSV-induced interferon signaling and ciliopathy are relatively enhanced in CF airway epithelia; further, recovery from this state and regeneration of the ciliated epithelium are delayed due to a CF phenotype that is intrinsically more inclined to a mesenchymal state.

In summary, the responses of the CF and Non-CF epithelia to RSV infection were most clearly distinguished by two features: first, the relatively exaggerated interferon responses of the CF epithelia relative to Non-CF, and second, the more severe and prolonged RSV-induced ciliopathy displayed by the CF cells. Whether these are primarily acquired characteristics resulting from chronic lung disease in the CF host, or whether these phenomena are intrinsic to CFTR-deficient cells, (or some combination of both), remains to be explored using animal models. Our findings call attention to the importance of cilia changes during viral infection and its potential to impair host defenses such as mucociliary clearance. Future studies addressing in more detail the nature of these virus-induced cilia changes in CF epithelia, and how these changes impact bacterial colonization in CF airways, will be valuable.

Supplementary Material

Supplement

Supplementary Figure S1. Expression of RSV receptors in ciliated airway epithelia. A scRNAseq dataset derived from human CF and Non-CF primary airway epithelia (GSE159056) was used to quantify expression of various receptors reported to facilitate RSV entry in airway epithelia. Violin plots summarize expression levels for the receptors in the ciliated cell population for both genotypes. Molecules reported to facilitate RSV entry include the insulin like growth factor 1 receptor (IGF1R), nucleolin (NCL), epidermal growth factor receptor (EGFR), intercellular adhesion molecule 1 (ICAM1), Toll like receptor 4 (TLR4), and C-X3-C motif chemokine receptor 1 (CX3CR1).

Supplementary Figure S2. Normalized read counts for individual RSV genes in infected CF and Non-CF airway epithelia. Read counts are expressed as transcripts per million (TPM). Data are presented as median and interquartile range, with whisker boundaries determined by Tukey’s method. Means are indicated by the “+” symbol. Data were tested for statistically significant differences between CF and Non-CF at each time point using multiple unpaired t-tests. *P_Adj < 0.05, **P_Adj < 0.01, ***P_Adj < 0.001.

Supplementary Figure S3. Exaggerated induction of interferon-stimulated genes in CF airway epithelia during RSV infection. Scatterplots summarize the relative upregulation of selected antiviral host defense genes over the infection timecourse in Non-CF vs CF HAE. For each gene, the mean Log₂ fold change in Non-CF epithelia is plotted on the x-axis and the mean Log₂ fold change in CF epithelia is plotted on the y-axis. IFI44L – interferon induced protein 44 like; CXCL10/IP10 – C-X-C motif chemokine ligand 10; RSAD2 – radical S-adenosyl methionine domain containing 2; 2’−5’-oligoadenylate synthetase like; ZBP1 – Z-DNA binding protein 1; ISG15 - interferon stimulated gene 15; BST2 – bone marrow stromal cell antigen 2; IDO1 – indoleamine 2,3-dioxygenase 1; IFIT1 – interferon induced protein with tetratricopeptide repeats 1.

Supplementary Figure S4. Pro-inflammatory and antiviral cytokine profiles in RSV-infected CF and Non-CF airway epithelia. CF and Non-CF HAE were infected with RSV (MOI = 1) and basolateral medium was collected at 24 hr intervals, followed by multiplex cytokine analysis to quantify cytokine release. Shown are cytokines in the Human Anti-Virus Response cytokine panel (Biolegend) that were detected above background levels. Data are plotted as median and interquartile range, with means indicted by the “+” symbol. Data were tested for statistically significant genotype-dependent differences using the Kruskal-Wallis test followed by Dunn’s multiple comparisons test. *P_Adj < 0.05.

Supplementary Figure S5. Products of tryptophan metabolism in RSV-infected CF and Non-CF airway epithelia. Metabolites were detected by GC-MS. Virus-induced changes in each metabolite’s abundance are expressed as the ratio of its abundance at a given time point/its abundance at time 0 (uninfected cells). Data are presented as median and interquartile range, with the mean represented by the “+” symbol. To test for significant differences between CF and Non-CF at each time point, multiple unpaired t-tests were performed on log transformed data, with correction for multiple comparisons using the Holm-Sidak method. *P_Adj < 0.05.

Supplementary Figure S6. CF epithelia show significant downregulation of cilia pathways late in the infection timecourse. Genes showing decreased transcript abundance at each time point (relative to 0 hpi) were used for pathway analysis via Enrichr as described in Methods. Tables summarize the GO Biological Pathways showing significant enrichment for downregulated genes at each time point and for each genotype. On the tables, color denotes the statistical significance of a given pathway hit, expressed as −Log₁₀ of the P-value. All gene pathways shown on these tables had P-values < 0.001.

Supplementary Figure S7. Uninfected CF and Non-CF HAE display distinct transcriptional profiles. (A) Unsupervised hierarchical clustering was performed using the gene counts for the uninfected (t = 0 hpi) CF and Non-CF samples. (B) Differential gene expression analysis was performed to identify transcripts whose baseline expression levels differ between CF and Non-CF epithelia, and gene pathway analysis was carried out via Enrichr. Bar graphs summarize pathway analysis results for 3 different pathway databases (GO Biological Pathways, MSigDB, and the ARCHS4 transcription factor coexpression database), illustrating the pathways with the greatest enrichment in our dataset. On the bar graphs, the X-axis represents the significance of the enrichment for a given gene pathway, expressed as −Log₁₀ of its P-value in the pathway analysis. The dotted vertical lines denote a P-value of 0.05.

Supplementary Figure S8. Dysregulation of factors driving epithelial-mesenchymal transition in CF airway epithelia. (A) RT-qPCR was used to measure relative transcript abundances for a panel of EMT-associated gene products in uninfected CF and Non-CF airway epithelia (n = 4 Non-CF, 4 CF). For each gene, fold differences were calculated using the 2^^−ΔΔCt method and represent transcript abundance in CF, relative to the Non-CF average. Data are presented as median and interquartile range, with the mean represented by the “+” symbol. Unpaired t-tests were performed on log transformed data to test for statistically significant differences between CF and Non-CF. TJP1 – tight junction protein 1 (also known as ZO-1/zonula occludens-1); CDH1 – cadherin 1 (also known as E-cadherin); LAMB1 – laminin subunit beta 1; SDC1 – syndecan 1; VIM – vimentin; MMP1 – matrix metallopeptidase 1; CDH2 – cadherin 2 (also known as N-cadherin); FN1 – fibronectin 1; ACTA2 – actin alpha 2, smooth muscle; TWIST1 – twist family bHLH transcription factor 1. (B) Lysates from uninfected CF and Non-CF HAE were separated by SDS-PAGE (10 μg total protein/lane) and immunoblotted for vimentin. For densitometry, vimentin abundance was normalized to that of the loading control (β-actin) and a Mann-Whitney test was performed to test for a statistically significant difference between CF and Non-CF. *P < 0.05