Acute cigarette smoke exposure leads to higher viral infection in human bronchial epithelial cultures by altering interferon, glycolysis and GDF15-related pathways

Abstract

Background: Acute exacerbations of chronic inflammatory lung diseases, such as chronic obstructive pulmonary disease (COPD), are frequently associated with rhinovirus (RV) infections. Despite these associations, the pathogenesis of virus-induced exacerbations is incompletely understood. We aimed to investigate effects of cigarette smoke (CS), a primary risk factor for COPD, on RV infection in airway epithelium and identify novel mechanisms related to these effects.

Methods: Primary bronchial epithelial cells (PBEC) from COPD patients and controls were differentiated by culture at the air-liquid interface (ALI) and exposed to CS and RV-A16. Bulk RNA sequencing was performed using samples collected at 6 and 24 h post infection (hpi), and viral load, mediator and L-lactate levels were measured at 6, 24 and 48hpi. To further delineate the effect of CS on RV-A16 infection, we performed growth differentiation factor 15 (GDF15) knockdown, L-lactate and interferon pre-treatment in ALI-PBEC. We performed deconvolution analysis to predict changes in the cell composition of ALI-PBEC after the various exposures. Finally, we compared transcriptional responses of ALI-PBEC to those in nasal epithelium after human RV-A16 challenge.

Results: CS exposure impaired antiviral responses at 6hpi and increased viral replication at 24 and 48hpi in ALI-PBEC. At 24hpi, CS exposure enhanced expression of RV-A16-induced epithelial interferons, inflammation-related genes and CXCL8. CS exposure increased expression of oxidative stress-related genes, of GDF15, and decreased mitochondrial membrane potential. GDF15 knockdown experiments suggested involvement of this pathway in the CS-induced increase in viral replication. Expression of glycolysis-related genes and L-lactate production were increased by CS exposure, and was demonstrated to contribute to higher viral replication. No major differences were demonstrated between COPD and non-COPD-derived cultures. However, cellular deconvolution analysis predicted higher secretory cells in COPD-derived cultures at baseline.

Conclusion: Altogether, our findings demonstrate that CS exposure leads to higher viral infection in human bronchial epithelium by altering not only interferon responses, but likely also through a switch to glycolysis, and via GDF15-related pathways.

Keywords: Bronchial epithelium; Chronic obstructive pulmonary disease (COPD); Cigarette smoke; RNA-Seq; Rhinovirus infection.

Fig. 1

Effects of cigarette smoke exposure on rhinovirus infection and antiviral defenses in differentiated primary human bronchial epithelial cells. ALI-PBEC were exposed to CS or air control and then directly infected with RV-A16 (MOI 1) for 1 h and incubated for 6, 24 and 48 h. A, B Levels of RV-A16 vRNA at 6, 24 and 48hpi were measured by qPCR. Data are shown as target gene expression normalized for Ribosomal Protein L13a (RPL13A) and ATP synthase, h + transporting, mitochondrial F1 complex, beta polypeptide (ATP5B). Black symbols are data from non-COPD donors, red symbols from COPD donors. Data are mean values (blue line) ± SEM. n = 16 different donors. Analysis of differences was conducted using paired two-way ANOVA with a Tukey post-hoc test or paired two-tailed t-test. Significant differences are indicated by P < 0.05. * Indicates difference between AIR + RV and AIR, # = CS + RV vs CS, and & = CS + RV vs AIR + RV. C Heat map of interferon response gene set in ALI-PBEC at 6hpi is also shown in a separate bar graph, in which the intensity of combined gene expression is shown as average Z scores. The Z scores for individual genes are represented in green and red, while the average Z scores (underneath the heat maps) of all genes in the gene set are represented in blue and red. The Z scores shown are relative to the average gene expression of the corresponding dataset at that time point. Below the heat maps, the average Z scores in blue represent downregulated gene expression while red color shows upregulated gene expression. D ALI-PBEC were exposed to CS or air control and then directly infected with RV-A16 (MOI 1) for 1 h and cultured in presence or absence of IFN-β or IFN-λ1 for 24 h. The levels of RV-A16 vRNA at 24hpi was measured by qPCR. Data are shown as target gene expression normalized for RPL13A and ATP5B. Data are mean values ± SEM; n = 4 different donors. Analysis of differences was conducted using paired two-way ANOVA with a Tukey post-hoc test. Significant differences are indicated by *P < 0.05

Fig. 2

The modulation of antiviral responses and enhancement of rhinovirus-induced inflammatory responses by cigarette smoke. A, C, E Heat maps showing interferon response gene set, epithelial interferon gene set and inflammatory gene set in ALI-PBEC at 24hpi. For all heatmaps in this figure, the intensity of combined gene expression is shown as Z scores. Below the heat maps, blue represents downregulated gene expression while red color shows upregulated gene expression. The Z scores in all heatmaps for individual genes are represented in green and red, while the average Z scores (underneath the heatmaps) of all genes in the gene set are represented in blue and red. The Z scores shown are relative to the average gene expression of the corresponding dataset at that time point. Data are mean values ± SEM. n = 13 different donors. Analysis of differences was conducted using paired one-way ANOVA with a Tukey post-hoc test. Significant differences are indicated by *P < 0.05. B, D, F Protein levels of CXCL10, IFN-λ1 and CXCL8 were measured by ELISA. Data are mean values ± SEM; n = 16 different donors. Analysis of differences was conducted using paired one-way ANOVA or two-way ANOVA with a Tukey post-hoc test. Significant differences are indicated by *P < 0.05

Fig. 2

The modulation of antiviral responses and enhancement of rhinovirus-induced inflammatory responses by cigarette smoke. A, C, E Heat maps showing interferon response gene set, epithelial interferon gene set and inflammatory gene set in ALI-PBEC at 24hpi. For all heatmaps in this figure, the intensity of combined gene expression is shown as Z scores. Below the heat maps, blue represents downregulated gene expression while red color shows upregulated gene expression. The Z scores in all heatmaps for individual genes are represented in green and red, while the average Z scores (underneath the heatmaps) of all genes in the gene set are represented in blue and red. The Z scores shown are relative to the average gene expression of the corresponding dataset at that time point. Data are mean values ± SEM. n = 13 different donors. Analysis of differences was conducted using paired one-way ANOVA with a Tukey post-hoc test. Significant differences are indicated by *P < 0.05. B, D, F Protein levels of CXCL10, IFN-λ1 and CXCL8 were measured by ELISA. Data are mean values ± SEM; n = 16 different donors. Analysis of differences was conducted using paired one-way ANOVA or two-way ANOVA with a Tukey post-hoc test. Significant differences are indicated by *P < 0.05

Fig. 3

Enhancement of rhinovirus infection by cigarette smoke-induced oxidative stress and growth factor differentiation 15. A Heatmap of oxidative stress gene set was shown. Below the heatmaps, blue represents downregulated gene expression while red color shows upregulated gene expression. The Z scores in the heatmap for individual genes are represented in green and red, while the average Z scores (underneath the heatmaps) of all genes in the gene set are represented in blue and red. The Z scores shown are relative to the average gene expression of the corresponding dataset at that time point. B, C The mitochondrial membrane potential after CS exposure or FCCP treatment with/without mitoTEMPO was measured by JC-10 assay. Green color shows JC-10 monomers while red color is JC-10 aggregates. D The expression of GDF15 after FCCP treatment was measured by qPCR at 6 h. Data are mean values ± SEM. n = 3 different donors. Analysis of differences was conducted using paired two-tailed t test. Significant differences are indicated by *P < 0.05. E Protein levels of GDF15 after CS and RV-A16 exposure were measured by ELISA. Data are mean values ± SEM. n = 16 different donors. Analysis of differences was conducted using paired one-way ANOVA with a Tukey post-hoc test. Significant differences are indicated by *P < 0.05. F, G CS and RV-A16 exposure were done in GDF15 knockdown cells and controls and incubated for 24 h after washing apically with PBS. The levels of RV-A16 vRNA at 24hpi was measured by qPCR. Data are shown as either target gene expression normalized for RPL13A and ATP5B or delta expression between CS + RV and Air + RV groups in control and GDF15 knockdown cells. Data are mean values ± SEM; n = 4 different donors. Analysis of differences was conducted using paired two-way ANOVA with a Tukey post-hoc test or paired two-tailed t test. Significant differences are indicated by *P < 0.05

Fig. 4

The mediation of CS-induced increase in RV-A16 infection via l-lactate production. ALI-PBEC were exposed to CS or air control and then directly infected with RV-A16 (MOI 1) for 1 h and were incubated for 24 h. A, D Heat maps showing glycolysis gene set and oxidative phosphorylation gene set. Below the heat maps, blue represents downregulated gene expression while red color shows upregulated gene expression. The Z scores in all heatmaps for individual genes are represented in green and red, while the average Z scores (underneath the heat maps) of all genes in the gene set are represented in blue and red. The Z scores shown are relative to the average gene expression of the corresponding dataset at that time point. B The release of l-lactate at 24hpi was measured. Data are mean values ± SEM. n = 16 different donors. C ALI-PBEC pre-treated with l-lactate were infected with RV-A16 (MOI 1) for 1 h and collected at 24hpi. The levels of RV-A16 vRNA at 24hpi was measured by qPCR. Data are shown as either target gene expression normalized for RPL13A and ATP5B. Data are mean values ± SEM. n = 3 different donors. E The intensity of combined oxidative phosphorylation gene expression was shown as Z scores. Data are mean values ± SEM; n = 13 different donors. All the analysis of differences in this figure was conducted using paired one-way ANOVA with a Tukey post-hoc test. Significant differences are indicated by * P < 0.05

Fig. 5

Relative proportion of specific cell types after cigarette smoke exposure and RV-A16 infection in primary human bronchial epithelial cell cultures. ALI-PBEC isolated from COPD and non-COPD donors were exposed to CS or air control and then directly infected with RV-A16 (MOI 1) for 1 h and were incubated for 24 h. The relative proportion of different cell types, secretory cells (A), basal cells (B) and ciliated cells (C) in ALI-PBEC after CS and RV-A16 exposure as predicted by cellular deconvolution of the transcriptomic datasets. D Z scores of differentially expressed gene sets related to cilia-associated genes are shown at 24hpi. Data are shown as mean ± SEM; n = 13 different donors. Analysis of differences was conducted using paired two-way ANOVA with a Tukey post-hoc test or unpaired two-tailed t test. Significant differences are indicated by *P < 0.05

Fig. 6

Effects of RV-A16 infection on transcriptional responses of the nasal epithelium in vivo. Nasal brushings were obtained from non-COPD donors on day 3, 6, 9 and 13 after RV-A16 exposure and 1 day before as baseline. RNA was isolated and analyzed using RNA-Seq in a similar analysis platform as for the cultured ALI-PBEC. A–D The heatmaps of interferon response gene set, complement gene set, necroptosis and inflammatory gene sets are shown and the intensity of combined gene expression is shown as Z scores. Below the heatmaps, blue color depicts downregulated and red color depicts upregulated expression; n = 16 different non-COPD donors. The Z scores in all heatmaps for individual genes are represented in green and red, while the average Z scores (underneath the heatmaps) of all genes in the gene set are represented in blue and red. The Z scores shown are relative to the average gene expression of the corresponding dataset at that time point