Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 93 (20)

Innate Immune Response to Influenza Virus at Single-Cell Resolution in Human Epithelial Cells Revealed Paracrine Induction of Interferon Lambda 1

Affiliations

Innate Immune Response to Influenza Virus at Single-Cell Resolution in Human Epithelial Cells Revealed Paracrine Induction of Interferon Lambda 1

Irene Ramos et al. J Virol.

Abstract

Early interactions of influenza A virus (IAV) with respiratory epithelium might determine the outcome of infection. The study of global cellular innate immune responses often masks multiple aspects of the mechanisms by which populations of cells work as organized and heterogeneous systems to defeat virus infection, and how the virus counteracts these systems. In this study, we experimentally dissected the dynamics of IAV and human epithelial respiratory cell interaction during early infection at the single-cell level. We found that the number of viruses infecting a cell (multiplicity of infection [MOI]) influences the magnitude of virus antagonism of the host innate antiviral response. Infections performed at high MOIs resulted in increased viral gene expression per cell and stronger antagonist effect than infections at low MOIs. In addition, single-cell patterns of expression of interferons (IFN) and IFN-stimulated genes (ISGs) provided important insights into the contributions of the infected and bystander cells to the innate immune responses during infection. Specifically, the expression of multiple ISGs was lower in infected than in bystander cells. In contrast with other IFNs, IFN lambda 1 (IFNL1) showed a widespread pattern of expression, suggesting a different cell-to-cell propagation mechanism more reliant on paracrine signaling. Finally, we measured the dynamics of the antiviral response in primary human epithelial cells, which highlighted the importance of early innate immune responses at inhibiting virus spread.IMPORTANCE Influenza A virus (IAV) is a respiratory pathogen of high importance to public health. Annual epidemics of seasonal IAV infections in humans are a significant public health and economic burden. IAV also causes sporadic pandemics, which can have devastating effects. The main target cells for IAV replication are epithelial cells in the respiratory epithelium. The cellular innate immune responses induced in these cells upon infection are critical for defense against the virus, and therefore, it is important to understand the complex interactions between the virus and the host cells. In this study, we investigated the innate immune response to IAV in the respiratory epithelium at the single-cell level, providing a better understanding on how a population of epithelial cells functions as a complex system to orchestrate the response to virus infection and how the virus counteracts this system.

Keywords: epithelial cells; influenza; innate immunity; interferons; single cell.

Figures

FIG 1
FIG 1
MOI-dependent temporal expression dynamics of the NS1 protein (PR8-GFP virus) in human respiratory epithelial cells (A549 H2B-mCherry) using time-lapse microscopy. Green indicates NS1 expression (GFP). The cell nuclei (H2B-mcherry) are depicted in red. (A) Time course from 3 to 12 hpi. One representative field is shown for each MOI. (B) Quantification of the number of GFP-positive cells for the four MOIs used (approximately 25,000 cells/well). For MOI of 0.2, 0.6, and 2, averages of two replicate wells ± SDs are shown. For infection at an MOI of 10 and mock infection, data from one single well are shown. (C) Heat map showing the median of fluorescence intensity of the infected cells (GFP+ cells). (D) Table showing the measured percentages of infected cells at 12 hpi, as well as the theoretical percentages of cells infected either with ≥1 virion (all infected cells) or >1 virion (multiple infections).
FIG 2
FIG 2
Single-cell RNA transcriptome analysis of PR8-GFP-infected epithelial cells. A heat map shows the expression levels of the 50 markers with the highest upregulation in infected versus mock-infected cultures (mean UMI count for each sample compared to value for the time-matched mock infection).
FIG 3
FIG 3
tSNE visualization, unsupervised clustering analysis, and heat maps showing the most differentially expressed genes identified by single-cell RNA sequencing in mock-infected cultures (A) and infected cultures (C, E, F, and G) at 12 hpi. This clustering strategy resulted in 5 clusters in the mock-infected cultures (A), mostly due to differences in genes associated with cell cycle as shown by a Reactome pathway Database analysis (42) (B). The most represented pathways for each cluster (up to 4 per cluster) are depicted (P < 0.001; cluster 0 analysis did not yield any significant pathway). Some of these genes also mediated clustering in infected cultures, resulting in 6 different clusters at MOIs of 2 and 0.2 (C and F). (D) The 10 most relevant pathways (sorted by P value) identified by a Reactome pathway Database analysis of the 111 genes or entities that were differentially expressed between cluster 2 and clusters 0, 1, and 4 of the sample infected at an MOI of 2. (E and G) The three clusters generated after manual curation of the samples infected at MOIs of 2 and 0.2: infected cells producing high levels of IFN (blue), infected cells producing low levels of IFN (green), and mostly uninfected cells (red). The scale of the heat maps shows the gene expression of each gene in each cell relative to the mean expression of that gene across the sample (note that expression levels between different samples cannot be compared, as they are normalized across the cells in each sample).
FIG 4
FIG 4
Distribution of the expression of viral genes (A and B) and their correlation with cellular genes (C). (A) Total viral UMI counts per cell. (B) Cells infected with a higher MOI (i.e., 2) show more elevated levels of viral genes than those infected with a lower MOI (0.2). The ratio of the median UMI count per cell for each MOI is shown at the top of each plot. (C) Heat map showing the correlation (Pearson correlation coefficient) between viral genes and a comprehensive panel of cellular genes. Viral and cellular genes were sorted by their median correlation coefficients against the respective gene sets.
FIG 5
FIG 5
(A) Ratio of the log2 fold change mean expression in infected and bystander cells. Genes are sorted by the average value of the log2 fold change of mean expression in infected and bystander cells for the two MOIs. (B) Comparison of the mean UMI count (log2) of cellular genes in infected and bystander cells for MOIs of 2 and 0.2. Data above the horizontal line indicate increased expression in infected versus uninfected cells. Data below the horizontal line indicate decreased expression in infected versus uninfected cells. (C) Violin plots of normalized UMI count for selected genes at 12 hpi in infected and bystander cell populations at high and low MOIs. Each dot represents a cell. The red line reflects the mean value of normalized UMI count across a cell population. The vertical gray line is a box-and-whisker plot representing the median (white circles), the 25th and 75th percentiles (ends of thicker gray lines), and high and low whiskers at 1.5 times the interquartile range (ends of the thin gray lines). In order to better highlight differences in the cell populations while still including zero values, for the y axis, we applied a linear scale for values greater than or equal to 0 and less than 1 and a log10 scale for values greater than or equal to 1.
FIG 6
FIG 6
Distribution of IFN expression in infected cultures. Shown are tSNE analysis and violin plots showing the distribution of infected cells (PR8 NS), IFNB1, IFNL1, IFNL2, IFNL3, and IFNL4 at MOIs of 2 (A) and 0.2 (B). IFNL1 shows a broader distribution than the rest of the IFN genes. Levels of expression of NS or IFN are depicted in purple. The density width of the violin shapes indicates the relative frequency of cell expressing specific levels of NS or IFN in each group. A previously described statistical framework for single-cell data that applies a likelihood ratio test was used for comparisons among clusters (95).*, P value < 10−25 to 10−100; **, P value 10−100 to 10−200; ***, P value < 10−200.
FIG 7
FIG 7
Stimulation of A549 cells with TNF-α, IFN beta, IFNL1 (100 U/ml), or combinations of those cytokines or infected with PR8-GFP (MOI, 2) for 8 h. Induction of IFNL1 was not detected upon stimulation with those cytokines by qRT-PCR. Gene expression of the indicated genes (IFNs, with Mx1 and CCL5 to confirm the effectiveness of the treatments) was analyzed. Averages of biological triplicates ± SDs are shown.
FIG 8
FIG 8
Analysis of the functional antiviral response induced by IAV in differentiated NHBE cells using a H1N1/H3N2 coinfection assay. (A) Scheme of the experimental setup and time line. (B) Replication profile of H1N1 (Cal09) infection of NHBE cells. (C and D) Expression of type I and III IFN and ISGs in NHBE cells in cultures infected with the H1N1 IAV or mock infected. (E) Impact of the innate immune response induced by H1N1 IAV infection on the replication of the H3N2 virus. Averages of biological triplicates ± SEMs are shown. Two-way analysis of variance (ANOVA) and Tukey’s multiple-comparison test were used. Adjusted P values are shown as follows: *, <0.05; **, <0.01; and ***, <0.001. (F) Immunofluorescence assay of H1N1-infected NHBE cells, showing a time course analysis of the progression of viral infection.
FIG 9
FIG 9
Model of the interplay between IAV and innate immunity in epithelial cells during infection with individual virus particles (left) versus multiple virus particles per cell (right). This illustration was created with BioRender (https://biorender.com).

Similar articles

See all similar articles

Cited by 2 articles

References

    1. Ramos I, Bernal-Rubio D, Durham N, Belicha-Villanueva A, Lowen AC, Steel J, Fernandez-Sesma A. 2011. Effects of receptor binding specificity of avian influenza virus on the human innate immune response. J Virol 85:4421–4431. doi:10.1128/JVI.02356-10. - DOI - PMC - PubMed
    1. Reference deleted.
    1. Kato H, Takeuchi O, Sato S, Yoneyama M, Yamamoto M, Matsui K, Uematsu S, Jung A, Kawai T, Ishii KJ, Yamaguchi O, Otsu K, Tsujimura T, Koh CS, Reis e Sousa C, Matsuura Y, Fujita T, Akira S. 2006. Differential roles of MDA5 and RIG-I helicases in the recognition of RNA viruses. Nature 441:101–105. doi:10.1038/nature04734. - DOI - PubMed
    1. Liu Y, Olagnier D, Lin R. 2016. Host and viral modulation of RIG-I-mediated antiviral immunity. Front Immunol 7:662. doi:10.3389/fimmu.2016.00662. - DOI - PMC - PubMed
    1. Matsukura S, Kokubu F, Kurokawa M, Kawaguchi M, Ieki K, Kuga H, Odaka M, Suzuki S, Watanabe S, Homma T, Takeuchi H, Nohtomi K, Adachi M. 2007. Role of RIG-I, MDA-5, and PKR on the expression of inflammatory chemokines induced by synthetic dsRNA in airway epithelial cells. Int Arch Allergy Immunol 143(Suppl 1):80–83. doi:10.1159/000101411. - DOI - PubMed

LinkOut - more resources

Feedback