Rhinovirus C15 infection induces airway epithelial cell remodeling and robust inflammatory responses: potential implications for airway obstruction in children

Abstract

Despite recognition of rhinovirus-C (RV-C) as a cause of severe respiratory exacerbations, little is known about the pathogenesis of RV-C infections. We infected mucociliary-differentiated primary tracheobronchial epithelial cells with RV-C15 or RV-A16. Initial RNASeq data showed that, compared to RV-A16, RV-C15 decreased expression of genes related to ciliary function while increasing expression of genes associated with mucus secretion and inflammation. Using different airway epithelial cell isolates, we confirmed greater reduction in DNAI2 and FOXJ1 (regulates production of motile cilia) and increased FOXA3 (regulates mucin-related gene expression) after RV-C15 infection compared to RV-A16. Similarly, nasal swab samples from children with RV-C but not RV-A infections showed significantly decreased DNAI2 and FOXJ1 mRNA compared to controls . While both RV-C15 and RV-A16 infection of airway epithelial cells increased mRNA expression and secretion of MUC5AC, RV-C15 induced greater airway surface liquid thickness, as measured by FITC-dextran staining. DAPT, a Notch inhibitor, reversed the effects of RV-C15 on DNAI2, FOXJ1 and FOXA3 expression. RV-C15 induced loss of α-acetyl tubulin, extrusion of airway epithelial cells, dissociation of ZO-1 from tight junctions, reduced ciliary beat frequency and decreased epithelial cell transepithelial electrical resistance. Finally, protein abundance of pro-inflammatory cytokines in cell supernatants and nasal samples also tended to be higher after RV-C infection. We conclude that RV-C causes significant disruptions in airway epithelial cell ciliary function which may lead to airway obstruction. Such disruptions may play a role in the severity of RV-C respiratory tract infections.

Introduction

Infections with human rhinovirus (RV) are a common cause of asthma exacerbations ^1–5. Accumulating evidence indicates that infections with a newly-discovered species of rhinovirus, RV-C, are associated with severe respiratory tract infections and asthma exacerbations often requiring hospitalization ^6–19. In contrast to RV-A and RV-B viruses, which utilize intercellular adhesion molecule (ICAM)-1 and low density lipoprotein family receptors (LDL-R), RV-C utilizes cadherin related family member 3 (CDHR3) as a receptor. Individuals with CDHR3 C529Y variants (AG and AA genotype) are more susceptible to RV-C infection, as this variant is localized on the airway epithelial cell surface, where it is accessible to viral infection, in contrast to the more common GG genotype which is predominantly localized to the cytoplasm ^20–22.

In addition to its contribution to asthma exacerbation, recent data suggest a possible role for early-life RV-C infections in asthma development. A multicenter prospective study of U.S. infants hospitalized for bronchiolitis found that only infants with RV-C had a higher risk of physician-diagnosed asthma at 4 years of age ²³.The presence of allergic sensitization was additive but not necessary for asthma development. These investigators also found that nasal aspirates from patients with severe bronchiolitis had greater type 2 cytokine levels than samples from patients with less severe disease ²⁴.

Despite recognition of RV-C as a cause of severe exacerbation, little is known about the pathogenesis of RV-C infections. RV-C has been refractory to study, in part because its receptor, CDHR3, is only expressed in differentiated airway epithelial cells. Accordingly, RV-C has been grown in primary mucociliary-differentiated human airway epithelial cells at air-liquid interface ^{25, 26} and HeLa cells transduced with the CDHR3 AA allele (HeLa-E8 cells) ²⁷.

Airway epithelial dysfunction plays a crucial role in the pathogenesis of asthma. Ciliated cells, responsible for the clearance of mucus and inhaled particles from the airways, are often reduced in number and function in asthmatic patients ²⁸, impairing mucociliary clearance and contributing to airway obstruction. In contrast, goblet cells, which produce and secrete mucus, frequently undergo hyperplasia in asthmatic airways, leading to excessive mucus production, mucus plugging and further airway obstruction ²⁹. However, little is known about the effects of RV infection, in particular RV-C, on airway epithelial function. In a study of RV-infected nasal polyp epithelial cells, RV-C15 but not RV-B48 downregulated expression of cilia-related genes and ciliary beat frequency ²⁶. Infection of nasal ³⁰ and human epithelial cells ³¹ with RV-A16 reduced expression of genes related to ciliary function. RV-A has been shown to induce mucin 5AC in cell culture models ³¹, mice ^{32, 33} and human volunteers ³⁴. In an animal model of RV-C infection, we found that, compared to RV-A1B-infected mice, RV-C induces greater type 2 cytokine expression and eosinophilic inflammation ³³.

We undertook this study to directly compare the effects of RV-A and RV-C on the airway epithelium, focusing on ciliary function, mucus and inflammation. We conclude that RV-C may cause significant disruptions in airway epithelial cell function. Such disruptions may play a role in the severity of respiratory tract infections.

Results

Infection of mucociliary differentiated human tracheobronchial epithelial cells.

Human airway epithelial cells were isolated from tracheobronchial tissue of lungs deemed unacceptable for lung transplantation. Airway epithelial cells obtained from up to nine donors were used (Table 1). We employed the air-liquid-interface cultures of primary human tracheobronchial cells as an in vitro model for assessing the replication dynamics of RV genotypes A16 and C15 (Figure 1A). As a preliminary experiment, using cells isolated from a single donor, we found that replication of both viruses reached a stationary phase 24 hours post-infection, maintaining its peak infectious load from 24 to 48 hours (Figure 1B). As shown previously ³⁵, viral copy number was about ½ log less for RV-C15 compared to RV-A16. Replication of RV-C15 and RV-A16 was further confirmed by the presence of dsRNA by immunofluorescent microscopy with anti-dsRNA antibody (Figure 1C).

Table 1.

Description of airway epithelial cell isolates.

Donor #	Sex	CDHR3 C529Y variant	Age (year)
H197	M	G/G	68
H284	M	G/A	43
H278	M	G/G	36
H310	M	G/A	52
H101	F	G/G	26
H102	M	G/G	59
H103	M	G/G	43
H104	F	G/G	41
H105	F	A/A	14

Figure 1. Next generation sequencing of human tracheobronchial epithelial cells infected with RV.

A. Experimental design. B. Cells were inoculated with RV-C15 or RV-A16 (MOI=1). RNA was purified and analyzed at 4, 24, and 48 h post-infection. Viral copy number was assessed by qPCR. Data points represent the average value of two replicates. C. Immunofluorescent staining for dsRNA (red) and DAPI (blue) in cells infected with RV-A16 or RV-C15 48 h after infection. D. Results of RNASeq. In this preliminary experiment, cells from a single donor were infected with either sham, RV-A16, or RV-C15 and collected 24 hours later (four technical replicates for each condition). Top 10 biological process gene ontology groups (sorted by average ranks of false discovery rates and fold enrichment) for RV-C15 versus sham. Gene ontology analysis was conducted using the online tool ShinyGO 0.82. E. Genes in cilia- and mucin-associated gene ontology groups were compared for RV-A16 and RV-C15. Data represent log2 fold changes relative to the sham group. F. Computational deconvolution of RNA-seq data reveals changes in basal, ciliated, and goblet cells in sham, RV-A16, or RV-C15 infected cells. Data shown are mean ± SEM and were analyzed by one-way ANOVA.

Comparison of RV-C15 and RV-A16 infection by bulk RNA-seq analysis.

To characterize the distinct host responses to RV-C15 and RV-A16 infections, we utilized bulk RNA-seq analysis to evaluate differential gene expression in cultured human tracheobronchial cells from this single donor. Principal component analysis revealed a significant shift in gene expression profiles 24 hours after infection, with clear separation between each group (Supplemental Figure 1A). A total of 9,244 genes displayed differential expression (p<0.05) in response to RV-C15 infection, while 2,555 genes showed differential expression in response to RV-A16 infection. Of these, RV-C15 and RV-A16 upregulated 5228 and 1873 genes respectively, with a shared upregulation of 900 genes between both serotypes. In contrast, 4016 and 682 genes were downregulated in response to RV-C15 and RV-A16, respectively, with a common downregulation of 320 genes across both infections (Supplemental Figure 1B). Gene ontology analysis identified distinct enriched biological processes among differentially expressed genes (DEGs) in human tracheobronchial epithelial cells infected with RV-A16 and RV-C15 (Figure 1D, Supplemental Figure 1C and Supplemental Table). For RV-C15, the top 10 most enriched gene ontology groupings related to cilium organization and assembly. Also enriched were groupings for proliferation, migration and differentiation, and signaling pathways related to stress-activated protein kinase and EGF receptor. Both genotypes encriched biological processes related to virus and viral genome replication as well as signaling pathways associated with cytokines-, IL-1 and type 1 IFNs.

Given the strong enrichment of cilia- and epithelial-related genes in RV-C15-infected cells, we performed a comparative analysis focusing on ciliary genes and mucus production. Relative to RV-A16, RV-C15 infection led to decreased expression of genes involved in cilium assembly, movement, and organization, while enhancing the expression of genes associated with mucus production, secretion and glycosylation (Figure 1E). Individual genes representing these gene ontology groupings are displayed in Supplemental Figure 1D. Our data suggest that RV-C15 triggers a transition away from the ciliated cell state towards the secretory cell state. Accordingly, computational deconvolution of the gene expression data suggested that, relative to sham-treated controls, RV-C15 decreased the proportion of ciliated cells and increased the proportion of secretory cells (Figure 1F). RV-A16 also increased the proportion of secretory cells.

Measurement of epithelial cell markers in different cell donors.

Based on the association between ciliary dysfunction, mucus secretion and asthma exacerbation, we assessed the mRNA and protein expression of the transcription factors FOXJ1, which governs differentiation of ciliated epithelial cells ^36–38 and FOXA3, which regulates goblet cell differentiation.³⁹ PCR primers employed in this study are shown in Table 2. qPCR analysis from nine different donors demonstrated that RV-C15 infection induces a reduction in expression of DNAI2 and FOXJ1 while both RV-C15 and RV-A16 increased expression of FOXA3 and MUC5AC (Figures 2A). Additionally, immunofluorescent staining (four donors) and Western blot analysis (three donors) revealed that RV-C15 but not RV-A16 infection induces a reduction in FOXJ1 expression while increasing FOXA3 expression (Figures 2 B-E and Supplemental Figure 2). Notably, some cells exhibited overlapping immunofluorescent staining of FOXA3 and FOXJ1, suggesting a potential transdifferentiation of ciliated cells into goblet cells (Figure 2E). Next, we stained airway epithelial cell cultures with PAS and anti-MUC5AC. PAS and MUC5AC staining showed that RV-C15 caused extrusion of ciliated cells from the airway epithelial layer and secretion of PAS- and MUC5AC-positive material (Figure 2F). To determine whether RV-C15 infection induces greater mucin production, we measured secreted MUC5AC by ELISA and airway surface liquid height by FITC-dextran staining (four donors each). Both viruses increased MUC5AC secretion, but RV-C15 led to a greater increase in the depth of airway surface liquid (Figure 2G and 2H).

Table 2.

PCR primers used in this study.

Primers	Sequence
Human GAPDH (Forward)	5’-TTCGACAGTCAGCCGCAT-3’
Human GAPDH (Reverse)	5’-CGACCAAATCCGTTGACTCC-3’
Human ACTB (Forward)	5’-CACCATTGGCAATGAGCGGTTC-3’
Human ACTB (Reverse)	5’-AGGTCTTTGCGGATGTCCACGT-3’
Human DNAI2 (Forward)	5’-ACCCGAAGAACTTCCTGACG-3’
Human DNAI2 (Reverse)	5’-ACACCCATTGTCCTTCAAGCTG-3’
Human FOXJ1 (Forward)	5’-ACTCGTATGCCACGCTCATCTG-3’
Human FOXJ1 (Forward)	5’-GAGACAGGTTGTGGCGGATTGA-3’
Human FOXA3 (Forward)	5’-CTCGCTGTCTTTCAACGACTGC-3’
Human FOXA3 (Reverse)	5’-CGCAGGTAGCAGCCATTCTCAA-3’
Human MUC5AC (Forward)	5’-CAGACTTGGGCACTGGTGCTG-3’
Human MUC5AC (Forward)	5’-TTCTCCGGCCTCATCTTCTCC-3’
Human HES1 (Forward)	5’-GGAAATGACAGTGAAGCACCTCC-3’
Human HES1 (Reverse)	5’-GAAGCGGGTCACCTCGTTCATG-3’
Human HES2 (Forward)	5’-AACCAGAGCCTGAGCCAGCTTA-3’
Human HES2 (Reverse)	5’-TGCAGGAAGCGCACGGTCATTT-3’
Human CDHR3 (Forward)	5’- TAAAGACCTCCCCCAGAGCA-3’
Human CDHR3 (Reverse)	5’- TGTTGGTGGCCTGGATTCTG-3’

Figure 2. Effects of RV infections on ciliary and mucus gene expression.

A. mRNA expression of DNAI2, FOXJ1, FOXA3, and MUC5AC in cells infected with sham, RV-C15, or RV-A16 at 24 h after infection (n=9 donors across multiple experiments). Cells from donors with CDHR3 variants are depicted by red triangles. B. A representative image of immunofluorescent staining for FOXA3 (red), FOXJ1 (green) and DAPI (blue). C. Quantification of FOXJ1⁺ or FOXA3⁺ cells as a percentage of total DAPI⁺ nuclei (n=4 donors from two independent experiments). D. A representative image of immunoblot showing the effects of RV-C15 and RV-A16 infection on levels of FOXJ1 and FOXA3 at 48 h after infection. E. Densitometry analysis of western blot for the expressions of FOXJ1 and FOXA3 across three groups. GAPDH expression was used as control. Densitometry analysis was performed using Image J program and normalized by the control level. Group mean data from three isolates are shown (two independent experiments). F. PAS staining (left) and immunofluorescent staining for MUC5AC (right, red) in sections of air-liquid interface cultures 48 h after infection with sham, RV-C15 or RV-A16. G, H. MUC5AC release and airway surface liquid depth were quantified 48 h after sham, RV-C15 or RV-A16 infection (n=4 donors from two independent experiments). Data shown are mean ± SEM and were analyzed by one-way ANOVA.

Measurement of epithelial cell markers in nasal swabs from children with RV infections.

We also had the opportunity to obtain nasal swabs and aspirates from children with RV-A or RV-C infections. At C.S. Mott Children’s Hospital, many patients suspected of having a respiratory virus undergo testing of nasopharyngeal swabs using the FilmArray Respiratory Panel (Biofire Diagnostics, Salt Lake City, UT). Children hospitalized in with enterovirus infections were eligible for study (Table 3). Swabs were taken from the surface of the nasal inferior turbinate using a floxed swab, and therefore likely represent epithelial cell gene expression. Similar to the cell culture results, DNAI2 and FOXJ1 expression were decreased in subjects with RV-C but not RV-A infection, indicating a downregulation of ciliated cell-associated genes (Figure 3). There were no significant changes in FOXA3 or MUC5AC.

Table 3.

Description of patient samples.

Nasal swabs
ID	Age (mo)	Viral serotype	Diagnosis
2410	192	-	control
2505	44	-	control
2609	99	-	control
2612	114	-	control
2806	182	-	control
2807	200	-	control
1136	10	C51	acute hypoxic respiratory failure
1516	108	C11	status asthmaticus
2104	11	C46	acute respiratory failure
2105	10	C15	reactive airway disease with acute exacerbation
2107	11	C15	bronchiolitis
2309	17	C17	asthma exacerbation
2311	52	C40	status asthmaticus
2314	45	C11	status asthmaticus
2315	44	C55	acute respiratory failure with hypoxia
2318	33	C45	hypoxia
2322	52	C41	status asthmaticus
2515	105	C7	status asthmaticus
1124	6	A20	acute respiratory failure with hypoxia
1126	9	A101	bronchiolitis
1510	83	A21	acute on chronic respiratory failure
1517	143	A101	status asthmaticus
2110	11	A20	bronchiolitis
2316	33	A71	status asthmaticus
2507	18	A67	status asthmaticus
2702	74	A49	status asthmaticus
Nasal aspirates
ID	Age (mo)	Viral serotype	Diagnosis
2403	33	-	control
2405	51	-	control
2606	84	-	control
2610	88	-	control
1108	1	Cpat19	bronchiolitis
1114	10	C24	bronchiolitis
1301	17	C1	bronchiolitis
1303	22	C6	bronchiolitis
1306	13	C23	bronchiolitis
1307	19	C6	bronchiolitis
1310	16	C10	bronchiolitis
1703	188	C11	acute hypoxic respiratory failure
2102	5	C4	acute on chronic respiratory failure
1109	3	A80	acute respiratory failure
1112	5	A61	acute on chronic respiratory failure
1116	11	A53	acute respiratory failure with hypoxia
1315	21	A34	acute on chronic respiratory failure
1504	127	A80	acute on chronic respiratory failure
2501	99	A1B	acute on chronic respiratory failure
1124	6	A20	bronchiolitis
1322	13	A78	acute on chronic hypoxic respiratory failure
1509	131	A54	status asthmaticus
1510	83	A21	acute on chronic respiratory failure
2320	21	A49	acute hypoxic respiratory failure

Figure 3. Effects of RV infection on ciliary and mucus gene expression in hospitalized children.

A. Human nasal swab mRNA samples from controls and children infected with RV-C or RV-A (control, n=6; RV-C, n=12; RV-A, n=8). Data shown are mean ± SD and were analyzed by one-way ANOVA.

The Notch signaling pathway contributes to RV-C15-induced cellular remodeling.

Previous studies suggest that Notch signaling controls the balance of ciliated and secretory cell fates in developing airways ^{40, 41}. To investigate the potential involvement of Notch signaling pathways in mediating the cellular alterations induced by RV-C15, we examined RNA-seq data and found the majority of Notch-associated genes were upregulated during RV-C15 infection (Figure 4A). We used qPCR analysis and found increased mRNA expression of the Notch target genes HES1 and HES2 in response to RV-C15 in three donors (Figure 4B). To explore the requirement of Notch signaling to RVC15-induced cellular changes, we treated infected cells with the γ-secretase inhibitor DAPT (N-[N-(3, 5-difluorophenacetyl)-l-alanyl]-s-phenylglycinet-butyl ester), an inhibitor of Notch signaling. DAPT treatment restored DNAI2 and FOXJ1 (Figure 4C). In addition, DAPT reduced the number of FOXA3-positive cells while increasing the number of FOXJ1-positive cells (Figure 4D). A similar effect of DAPT was also observed in the sham-infected cells (Supplemental Figure 3). These findings suggest that Notch signaling contributes to RV-C15-induced epithelium remodeling.

Figure 4. RV-C15 epithelial remodeling is reversed by a Notch signaling inhibitor.

A. Genes selected from the Notch signaling pathway (GO: 0007219) were compared based on log2 fold change vs sham for RV-A16 and RV-C15 infection. B. mRNA expression of HES1 and HES2 in cells infected with sham or RV-C15 (n= three donors across two independent experiments). C. mRNA expression of DNAI2, FOXJ1 and FOXA3 in cells treated with sham, RV-C15, or RV-C15 plus DAPT (N-[N-(3,5-Difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester) (n=three donors across two independent experiments). D. Immunofluorescence staining for FOXJ1 (green) and FOXA3 (red) in RV-C15-infected cells treated with vehicle or DAPT at 48 h after infection. Data shown are mean ± SEM and were analyzed by t test (panel B) or one-way ANOVA (panel C).

RV-C15 infection causes dissociation of ZO-1 from the plasma membrane and reduced ciliary beat frequency.

We examined the effects of RV-C15 and RV-A16 on cellular localization of zona occludins (ZO)-1, a tight function protein, in one cell donor. RV-C15 disrupted ZO-1 staining (Figure 5A), in contrast to RV-A16 infection which did not. Staining cell cultures for α-acetyl tubulin, a cilia marker, we again noted shedding or extrusion of ciliated cells after RV-C15 infection (Figure 5B). We measured the effects of RV-C15 and RV-A16 on ciliary beat frequency, transepithelial resistance and LDH release using cells from three-to-four different donors. RV-C15 compromised human bronchial epithelial cell ciliary beating frequency greater than RV-A16 (Figure 5C and 5D). Finally, compared to RV-A16, RV-C15 infection resulted in significantly significantly decreased transepithelial resistance (Figure 5E) and higher levels of LDH release (Figure 5F). Together, these results suggest that RV-C15 negatively impacts airway epithelial integrity and function.

Figure 5. Effects of RV-C15 and RV-A16 infection on epithelial function.

Confocal images of immunofluorescent staining for (A) zona occludins (ZO)-1 (red) or (B) acetyl-tubulin (red) and DAPI (blue). C. A representative kymograph from high-speed video microscopy for each group. D. Comparison of ciliary beat frequency (n=4 donors across two independent experiments). E. Comparison of transepithelial electrical resistance (TEER) across three groups. F. Lactate dehydrogenase (LDH) release in the apical and basal media was measured 48 h after sham, RV-C15 or RV-A16 infection (n=four donors from two independent experiments). Data shown are mean ± SEM and were analyzed by one-way ANOVA.

RV-C15-infected bronchial epithelial cells exhibit higher levels of inflammatory cytokine expression and release.

Preliminary bulk RNA-seq data from one donor showed higher expression of cytokines and interferons in human bronchial epithelial cells infected with RV-C15 compared to cells infected with RV-A16 and uninfected controls (Figure 6A). To further quantify cytokine release into the medium, multiplex ELISA assays were performed on cells from four donors (Figure 6B). The results demonstrated that RV-C15-infected cells exhibited significantly higher levels of IL-6, IL-10, CCL2, CXCL1, CXCL2, CXCL5, CXCL10, CCL2 and CCL3. Tracheobronchial epithelial cells infected with RV-A16 showed higher release of IL-10, IFN-β and CCL20. These findings demonstrate that RV-C15 infection induces a robust pro-inflammatory response.

Figure 6. Effects of RV-C15 and RV-A16 infection on expression of pro-inflammatory cytokines.

A. RNASeq data from cultured tracheobronchial epithelial cells infected with sham, RV-C15 or RV-A16. Heatmap comparing cytokines and interferon expression among three groups. Each column represents one gene and each row represents one sample. Samples are grouped by sham (top), RV-C15 (middle) and RV-A16 (bottom). Red-labeled genes correspond to protein levels assayed in panel B. B. Effects of RV infection on cytokine release into the apical and basal media 48 h after infection. IL-6, IL-10, IL-1β, TGF- β, CXCL1, CXCL2, CXCL5, CXCL6, CXCL8, CXCL10, CCL2, CCL3, CCL11 and CCL20 are shown (n=four donors across two independent experiments). Data shown are mean ± SEM and were analyzed by one-way ANOVA.

Measurement of soluble cytokines in nasal aspirates from children with RV infections.

We also had the opportunity to collect nasal aspirates from children with RV-A or RV-C infections (Table 3). To collect nasal aspirates, 1 mL of sterile saline was instilled into each nare prior to suction aspiration using a flexible red rubber suction catheter. . Cytokines were measured by Meso Scale Discovery immunoassay (Rockville, MD). RV-C, but not RV-A, significantly increased IL-1β, IFN-λ, IL-2 and IL-10 levels (Figure 7). Both viruses increased IFN-β. Finally, we found that two cytokines not produced by airway epithelial cells, IFN-γ and IL-13, were also significantly increased after RV-C infection. Collectively, these data support the notion that RV-C infection triggers a robust initial chemokine response in epithelial cells, which subsequently amplifies immune responses.

Figure 7. Soluble cytokines and interferons in human nasal aspirates from controls and children hospitalized with RV-C or RV-A infections.

Levels of IFN-α, IFN-β, IFN-λ, IFN-γ, IL-1β, IL-2, IL-4, IL-5, IL-8, IL-10, IL-12, IL-13, and TNF-α were assayed by ELISA (control, n=4, RV-C, n=11; RV-A, n=9). Data shown are mean ± SD and were analyzed by one-way ANOVA.

Discussion

Despite accumulating evidence that RV-C infections are associated with severe respiratory tract infections, asthma exacerbations ^6–19 and asthma development ²³, little is known about the pathogenesis of RV-C infections. We infected mucociliary-differentiated primary tracheobronchial epithelial cells with RV-C15 or RV-A16 and examined responses by RNASeq, qPCR, histology, immunofluorescence imaging, and measurement of ciliary beat frequency. We found using next generation sequencing of one human bronchial epithelial cell isolate that RV-C15 highly enriched gene ontology pathways related to ciliary assembly and function. Further examination revealed downregulation of cilia-related genes. qPCR showed corresponding reductions in DNAI2, an mRNA encoding a representative member of the dynein intermediate chain family, and FOXJ1, a master regulator of the motile ciliogenic program ^36–38 . These results were replicated in epithelial cell isolates from different donors, as well as nasal swabs from children hospitalized with RV infections. Together these data suggest that, compared to RV-A16, RV-C15 has a more significant suppressive effect on ciliary gene expression.

Previous in vitro studies have revealed that cytotoxic respiratory viruses, including RSV, hold a common transcriptomic signature which includes the global downregulation of cilium-related genes ^42–44. However, less is known about the relative effects of RV-C and RV-A on ciliary gene expression. In a study of RV-infected nasal polyp epithelial cells, RV-C15 but not RV-B48 downregulated expression of cilia-related genes ²⁶.

In addition to downregulating the transcription of cilia-relate genes, it is also conceivable that RV-C15 induces the loss of airway epithelial cells after infection. Computational deconvolution of RNASeq data showed a small but significant reduction in ciliated epithelial cells. We also found that RV-C15 caused evidence of epithelial damage in cultured cells. RV-C15 caused dissociation of ZO-1 from the plasma membrane and shedding or extrusion of ciliated cells. Infection with RSV has been shown to cause similar extrusion of cultured cells ⁴⁵. RV-C15-infected human bronchial epithelial cells exhibited more compromised ciliary beating frequency than RV-A16. RV-C15 infection resulted in higher levels of LDH release. Together, these results suggest that, compared to RV-A16, RV-C15 more negatively impacts airway epithelial integrity and function, i.e., it is more cytotoxic.

Computational deconvolution of RNASeq data showed increases in secretory cells after both RV-A16 and RV-C15 infection, with RV-C15 having the most significant impact. RV-A16 and RV-C15 infection of cultured human bronchial epithelial cells also increased mRNA expression of FOXA3, a transcription factor which regulates goblet cell differentiation ³⁹ .

RV-A16 and -C15 each increased FOXA3 protein abundance and the number of FOXA3-expressing cells. In addition, we found using immunofluorescence microscopy that, upon RV-C15 infection, some epithelial cells expressed both FOXA3 and FOXJ1, suggesting potential transdifferentiation of ciliated cells into goblet cells. Although it is generally thought that a ciliated cell cannot transdifferentiate into a mucus secretory cell, this result is analogous to ssRNASeq data showing MUC5AC transcripts in human ciliated airway epithelial cells treated with IL-13 ^{46, 47}. Indeed, the pattern of gene expression observed after RV-C15 infection -- downregulation of ciliogenesis genes and upregulation of mucus-related genes, is similar to that observed after IL-13 treatment. Previous studies have shown that RV-A14 ⁴⁸ and -A16 ³⁴ increase MUC5AC expression in cultured airway epithelial cells and human volunteers ³⁴ and RV-A1B causes mucous metaplasia adult FVB/N ⁴⁹ and immature BALB/c ⁵⁰ and C57BL/6 mice ⁵¹.

RNASeq suggested that RV-C infection induces a greater inflammatory response than RV-A, and this was to some extent confirmed by protein levels in airway epithelial cell conditioned media and nasal aspirates from children hospitalized with RV infections. Children in our study with RV-C infection tended to have higher nasal cytokine levels. With regard to type 2 cytokines, children with RV-C tended to have greater IL-4 and IL-5 levels, and had significantly increased IL-13 levels compared to children infected with RV-A. These data are consistent with results from our mouse model showing that, compared to RV-A1B-infected mice, RV-C induces greater type 2 cytokine expression and eosinophilic inflammation ³³. They are also consistent with a previous clinical study showing that infants with RV-C infection, Moraxella colonization and a high T2 cytokine profile that had a significantly higher risk of developing asthma after recurrent wheeze ⁵².

The mechanisms underlying the differential effects of RV-C and RV-A on airway epithelial function remain unknown. In normal individuals, the major group RV-A receptor ICAM-1 is expressed at low levels in airway epithelium and increased with inflammation ^{53, 54}. On the other hand, the RV-C receptor CDHR3 is highly expressed in ciliated airway epithelial cells and associated with the cilia of these cells ⁵⁵. It is therefore conceivable that RV-C may have a more pronounced effect on ciliary function than RV-A due to localization of its receptor. We did not notice a clear effect of CDHR3 variants on the response to RV infection.

Given the importance of ciliary function for airway clearance and prevention of airway obstruction, our data showing that RV-C induces downregulation of cilium-related genes and the loss of epithelial integrity and function may explain increased morbidity in patients with RV-C infections. Notch signaling controls the balance of ciliated and secretory cell fates in developing airways ⁴⁰. In our study. RV-C15 infection increased mRNA expression of the Notch target genes HES1 and HES2, and the γ-secretase inhibitor DAPT reversed the effects of RV-C15 on DNAI2 and FOXJ1 mRNA expression, as well as the number of FOXA3- and FOXJ1-expressing cells. Targeting of the Notch pathway may therefore represent a future strategy for the treatment of viral-induced asthma exacerbations. On the other hand, in a hamster model of RSV, while shedding of infected epithelial cells caused acute distal airway obstruction, it also accelerated the reduction of virus titers.

We conclude that RV-C causes significant disruptions in airway epithelial cell ciliary function which may lead to airway obstruction. Such disruptions may play a role in the severity of RV-C respiratory tract infections.

Methods

Generation of RV-A16 and RV-C15.

RV-A16 strain 11757 was purchased from ATCC and propagated in H1 HeLa cells. The identity of RV-A16 was confirmed by semi-nested PCR amplification of the P1-P2 region from gel purified PCR products ⁶. Full length cDNA encoding RV-C15 was reverse transcribed and resulting full-length vRNA transfected into HeLa cells using lipofectamine (Invitrogen, Waltham, MA). After three freeze-and-thaw cycles, virus was concentrated and partially purified from infected HeLa-E8 cell lysates by means of ultrafiltration with a 100-kDa cutoff filter assay, as described previously ⁵⁶. Purified RV copies will be quantified by PCR using specific primers and cDNA plasmids. To quantify viral copy number, qPCR for viral RNA was conducted using pan-RV primers (forward primer CTAGCCTGCGTGG; reverse primer AAACACGGACACCCAAAGTAGT) and probes (6FAM-TCCTCCGGCCCCTGA-MGB- NFQ) ⁵⁷.

Cell culture.

Human airway epithelial cells were isolated from tracheobronchial tissue of lungs deemed unacceptable for lung transplantation. Airway epithelial cells obtained from up to nine donors were used (Table 1). Primary airway epithelial cells were cultured in Transwells (Corning, Glendale, AZ) at air-liquid interface as described previously ^{58, 59}. Briefly, airway epithelial cells were cultured under submerged conditions in PneumaCult^™-Ex Plus medium (Stemcell Technologies, Vancouver, CA) until they reached 90% confluency. Cells were then shifted to PneumaCult^™-ALI medium and maintained at air-liquid interface for 28 days. For infection, virus was applied apically at MOI of 1 in 50 μl PBS, or an equivalent volume of PBS alone for mock treatment. Cells were incubated at 33°C for 2 hours. Following incubation, the apical solutions were removed, and cells were washed twice with PBS. Cultures were then maintained at 33°C and harvested at designated time points for downstream analysis. For DAPT treatment, cells infected with RV-C15 were cultured in medium containing 10 μM DAPT (N-[N-(3,5-Difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester, Sigma-Aldrich) or an equivalent concentration of DMSO (vehicle control), and maintained in this condition until sample collection.

Next generation RNA sequencing (RNAseq).

RNA library generation and next-generation sequencing were performed by the University of Michigan Advanced Genomic Core. Total RNA was extracted from cell cultures from a single donor carrying the CDHR3 C529Y genotype (G/G), using TRIzol Reagent (Invitrogen) according to the manufacturer’s protocol. Cells were infected with either sham, RV-A16, or RV-C15 and collected 24 hours later (four technical replicates for each condition). Total RNA was prepared using Trizol extraction (Invitrogen). RNA was double purified using RNEasy spin columns (Qiagen, Germantown, MD) and enriched for mRNA using a Poly(A) mRNA Magnetic Isolation Module (New England Biolabs, Ipswitch, MA). cDNA libraries for next-generation sequencing were prepared with a xGen Broad-Range RNA Library Prep (IDT, Coralville, IA) with xGen Normalase UDI Primers (IDT). Samples were subjected to 151 bp paired-end sequencing using a NovaSeq S4 flow cell (Illumina) according to the manufacturer’s protocol. Reads were mapped to the reference genome GRCm38 (ENSEMBL 102) using STAR v2/7.8a ⁶⁰ and assigned count estimates to genes with RSEM v1/3.3 ⁶¹. Computational deconvolution of RNA-seq data was performed using the Human Lung Cell Atlas reference matrix ⁶². Gene ontology was analyzed by ShinyGO 0.82 ⁶³ and sorted by average ranks of false discovery rates and fold enrichment.

Real-time quantitative PCR.

RNA was extracted using a RNeasy kit (Qiagen). cDNA was synthesized from 500 ng (cells) or 10 ng (human nasal swab samples) of RNA using a High-Capacity cDNA Synthesis kit (Applied Biosystems, Waltham, MA) and subjected to quantitative real-time PCR (qPCR) using specific primers (Table 2) for mRNA. The expression levels were normalized to GAPDH or ACTB. PCR was performed on an Eppendorf Mastercycler machine, using the comparative threshold cycle (Ct) method. Specific primers employed in this study are listed in Table 2.

CDHR3 C529Y genotyping.

CDHR3 was amplified from cDNA described above using the High-Fidelity 2X PCR Master Mix (New England Biolabs, Ipswich, MA) with primers listed in Table 2. The resulting 170 bp PCR product was purified using the PCR & DNA Cleanup Kit (New England Biolabs) and submitted for Sanger sequencing to Genewiz (South Plainfield, NJ).

Western blot analysis.

Cells were lysed in RIPA buffer (Thermo Scientific) containing Complete Protease Inhibitor (Roche, San Francisco, CA), centrifuged and the supernatant collected. Equal amounts of protein were subjected to SDS-PAGE, and proteins were transferred to nitrocellulose membranes by wet transfer for 1 h. Membranes were blocked with 5% fat-free milk and probed with anti-FOXJ1 (R&D Systems, Minneapolis, MN) or anti-FOXA3 (Novus Biologicals, Centennial, CO) and anti-GAPDH (Sigma-Aldrich, St. Louis, MO)(Table 4). Bound antibody was detected with an appropriate secondary antibody conjugated with horseradish peroxidase. The signal was detected by enhanced chemiluminescence.

Table 4.

Antibodies used in this study.

Antigen	Host	Dilution	Source	Use
dsRNA	Mouse	1:250	Nordic Mubio (10010200)	IF
ZO-1	Rabbit	1:500	Abcam (EPR19945-296)	IF
α-acetylated tubulin	Mouse	1:5000	Sigma-Aldrich (T7451)	IF
MUC5AC	Mouse	1:2000	Abcam (ab3649)	IF
FOXJ1	Rabbit	1:1000	Novus Biologicals (NBP1-87928)	IF
FOXA3	Rabbit	1:1000	Novus Biologicals (NBP2-87461)	IF & WB
GAPDH	Mouse	1:1000	Sigma-Aldrich (G8795)	WB
FOXJ1	Goat	1:500	Abcam (AF3619)	WB

Histology and immunofluorescence microscopy.

Cells were fixed with 100% methanol at −20°C. Paraffin-embedded blocks were sectioned at 500-μm intervals at a thickness of 5 μ, and each section was deparaffinized, hydrated, and stained. To visualize mucus, samples were stained with periodic acid Schiff (Sigma-Aldrich) and observed by light microscopy. Staining with anti-FOXJ1 (Novus Biologicals), anti-FOXA3 (Novus Biologicals), anti-Muc5ac (Abcam, Waltham, MA), anti-α-acetyl-tubulin (Sigma-Aldrich), anti-ZO1 (Abcam) and anti-dsRNA (Nordic Mubio, Susteren, The Netherlands) were visualized by immunofluorescence microscopy or immunofluorescence confocal microscopy.

Ciliary beat assay.

The ciliary beat assay was performed using a Nikon Ti2 Widefield microscope. Briefly, the membrane containing the cells was carefully peeled from the insert and placed on a glass slide. Ciliary beat was recorded as 10-second video clips. The recorded videos were analyzed using the “Kymograph” function in the Fiji toolbar of NIH ImageJ image processing software ⁶⁴. Beating frequency was calculated by measuring the wavelength of each wave observed in the videos.

Lactase dehydrogenase (LDH) and enzyme-linked immunosorbent assays (ELISA).

The basal media were incubated on the apical side of the inserts at 37°C for 5 minutes. Cytotoxicity was assessed using the LDH Cytotoxicity Detection Kit (Takara, Fort Worth, TX). Levels of MUC5AC was assayed by ELISA kit (Invitrogen, Waltham, MA). Cytokine protein concentrations in airway epithelial cell conditioned media were measured by multiplex ELISA (Luminex, Austin, TX). These studies were performed by the University of Michigan Rogel Cancer Center Immunology Core.

Transepithelial electrical resistance (TEER) assay.

TEER measurement was performed using an EVOM meter (World Precision Instruments, Sarasota, FL) following the manufacturer’s instructions. Prior to use, the electrode probe was disinfected by immersion in 70% ethanol for 10 minutes, then rinsed in sterile PBS. The culture medium was removed. Subsequently, 200 μL of PBS was added to the apical chamber and 500 μL to the basolateral chamber. Readings were taken until a stable resistance value was obtained.

Airway liquid surface (ASL) depth measurement.

ASL depth measurements were performed as previously described ⁶⁵, with slight modifications. Briefly, the ASL was strained with FITC-dextran (Sigma-Aldrich, St. Louis, MO) by depositing dextran solution on the apical site of the culture at 37°C for 2 h. For measurements, the culture inserts were removed and scanned with the confocal microscopy. ASL depth was determined from z-image stacks.

Measurement of nasal epithelial mRNA expression and nasal aspirate cytokines from human subjects.

The study was approved by the Institutional Review Board (IRB) at the University of Michigan (HUM00133414). At C.S. Mott Children’s Hospital, many patients suspected of having a respiratory virus undergo testing of nasopharyngeal swabs using the FilmArray Respiratory Panel (Biofire Diagnostics, Salt Lake City, UT). Children hospitalized in with enterovirus infections were eligible for study (Table 3). Uninfected children undergoing elective dental or otolaryngologic procedures were enrolled as control subjects. Nasal aspirate collection was performed using an established protocol ⁶⁶. To collect nasal aspirates, 1 mL of sterile saline was instilled into each nare prior to suction aspiration using a flexible red rubber suction catheter. Nasal swab samples were taken from the surface of the nasal inferior turbinate using a floxed swab (Copan), placed in RNAlater (ThermoFisher Scientific, Rochester, NY) and brought to the laboratory for processing and analysis.

Nasal aspirate sample quality was assessed by evaluating the number of cytokines with measured levels below the lower limit of detection, the number of live cells in the sample, and the total protein concentration of the sample. Samples were excluded if they met two of the three criteria: greater than or equal to 7 of the 15 measured cytokines below the lower limit of detection, a live cell count < 400,000, or undetectable total protein. Control samples were also excluded if Novaplex multiplex PCR respiratory panels (Seegene, Irvine, CA) showed an asymptomatic viral infection.

RV typing was performed on the cell supernatant by semi-nested PCR amplification of the P1-P2 region from gel purified PCR products ⁶. The identity of each sequence was determined by comparison to known 5′ sequences using BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Soluble cytokine concentrations were analyzed by electrochemiluminescence (Meso Scale Discovery).

Data analysis.

Data were represented as mean ± SEM for cell culture experiments and mean ± SD for human samples. Statistical significance was assessed by t test or analysis of variance (ANOVA), as appropriate. A P value of ≤0.05 was considered statistically significant.

Data availability.

Complete RNASeq data from cultured cells will be uploaded to NCBI Gene Expression Omnibus.