Transcriptomic similarities and differences in host response between SARS-CoV-2 and other viral infections

Abstract

The pandemic 2019 novel coronavirus disease (COVID-19) shares certain clinical characteristics with other acute viral infections. We studied the whole-blood transcriptomic host response to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) using RNAseq from 24 healthy controls and 62 prospectively enrolled patients with COVID-19. We then compared these data to non-COVID-19 viral infections, curated from 23 independent studies profiling 1,855 blood samples covering six viruses (influenza, respiratory syncytial virus (RSV), human rhinovirus (HRV), severe acute respiratory syndrome coronavirus 1 (SARS-CoV-1), Ebola, dengue). We show gene expression changes in COVID-19 versus non-COVID-19 viral infections are highly correlated (r = 0.74, p < 0.001). However, we also found 416 genes specific to COVID-19. Inspection of top genes revealed dynamic immune evasion and counter host responses specific to COVID-19. Statistical deconvolution of cell proportions maps many cell type proportions concordantly shifting. Discordantly increased in COVID-19 were CD56^bright natural killer cells and M2 macrophages. The concordant and discordant responses mapped out here provide a window to explore the pathophysiology of the host response to SARS-CoV-2.

Keywords: Immunology: Bioinformatics; Molecular Biology; Transcriptomics.

Graphical abstract

Figure 1

RNA-seq data for patients with COVID-19 versus healthy control and pathway analysis of the COVID-19 signature (A–C) (A) Significance score [defined as -log10(FDR)] versus mean difference of co-normalized log2-transformed expression data between patients with COVID-19 (n = 62) and healthy controls (n = 24). The chosen cutoff of ES ≥ 1 or ≤ −1 with FDR ≤0.05% yields the 2,002 COVID-19 signatures, including 771 positively regulated genes and 1,231 negatively regulated genes. GO term enrichment analysis of positive (B) and negative (C) gene sets reveals increased neutrophil function enrichment and decreased T-cell-related pathways (gene ratios represent the number of genes in our gene set within that pathway). The gene ratio (x axis) is the ratio of the number of genes in our data enriched in a given gene set (pathway) to the total number of genes in that pathway.

Figure 2

Metaintegration of 14 non-COVID-19 viral disease data sets and pathway analysis of non-COVID-19 signature genes (A–C) (A) ROC plots of the 635 non-COVID-19 viral signatures discovered using multicohort analysis with a cutoff of ES ≥ 1 or ≤ −1 and FDR ≤0.05% resulting in 314 positively regulated genes and 321 negatively regulated genes then plotted individually for each of the 14 data sets of viral infections (n = 652) and healthy controls (n = 672) identified. The consistent and high AUC values indicate that the signature is representative of all data sets, thereby embracing the heterogeneity which will increase generalizability. GO term enrichment analysis of positive (B) and negative (C) gene sets reveals increased neutrophil function enrichment and decreased T-cell-related pathways, similar to those in Figure 1 (gene ratios represent the number of genes in our gene set within that pathway).

Figure 3

Validation of a global host immune response to viral infections (A) ROC performance of 635 non-COVID-19 signatures in 4 independent respiratory viral infection data sets including HRV, RSV, picornavirus, and influenza. (B) ROC performance in 5 additional cohorts of other viral infections to illustrate that this signature is broadly applicable to many viruses [Ebola (GSE122692), SARS CoV-1 (GSE5972), and dengue (GSE38246, EMTAB3162, GSE51808)]. (C) The signature is also tested in the 62 patients with COVID-19 and 24 HCs.

Figure 4

Comparison of COVID-19 signature with non-COVID-19 signature (A) Scatterplot of effect size for all 9,818 genes commonly present in all data sets between non-COVID-19 vs HC (x axis) and COVID-19 vs HC (y axis). Two thousand two COVID-19 signature genes from Figures 1 and 635 non-COVID-19 signature genes from Figure 2 are overlayed and colored, each of the 9 quadrants have a different color to allow for easy visualization of the overlap of Hedges' g ES from each signature. For example, teal in the top right quadrant are the genes that have an Hedges' g ES ≥ 1 for both the 2,002 COVID-19 signature genes and the 635 non-COVID-19 signature genes. Concordant host response between COVID-19 and other viral infections is reflected by 223 commonly positively (teal, top right) and 220 negatively (blue, bottom left) regulated genes in both. Discordant response is only seen in ACO1 whose expression is positively regulated in COVID-19 but negatively regulated in non-COVID and in ATL3 whose expression is negatively regulated in COVID-19 but positively regulated in non-COVID-19. (B) Using COCONUT conormalized data combined with a head-to-head comparison of COVID-19 and non-COVID-19 viral infections using Hedges' g ES ≥ 1 or ≤ −1 with FDR ≤0.05% yields 416 COVID-19-specific signatures, including 114 positively regulated genes and 302 negatively regulated genes. Significance score [defined as -log10(FDR)] vs mean difference of co-normalized log2-transformed expression data between patients with COVID-19 (n = 62) vs other viral infections (n = 652). (C) To illustrate the overlap of (A) and (B), the 416 COVID-19-specific signature genes from head-to-head comparison in (B) are shown in the same scatterplot in (A).

Figure 5

Summary of pathway analysis results Scatterplots of the significance level from pathway enrichment analysis between COVID-19 and non-COVID-19 viral infections obtained for positive genes in (A) and negative genes in (B), respectively. Significance is defined as -log10(BH-corrected p value) for each pathway. The concordance is seen in results for up-regulated genes between COVID-19 and non-COVID-19, while a degree of discordance is evident in down-regulated genes between COVID-19 and non-COVID-19. (C) Heatmap summary of pathway enrichment analysis for 15 gene sets of interest including COVID-19 vs HC (+) and (−), non-COVID-19 viral vs HC (+) and (−), COVID-19 vs non-COVID-19 viral (+) and (−), as well as gene lists from 9 groups segmented in Figure 4A as labeled in the legend key box. Values between 1 and 10 of -log10(BH-corrected p value) are plotted. ur, up-regulated; dr, down-regulated.

Figure 6

Statistical deconvolution of bulk transcriptome profiles using immunoStates of COVID-19 versus non-COVID-19 viral infections (A) Changes in cell proportions when comparing patients with COVID-19 to healthy controls. Note the trends of increased neutrophil and decreased T-cell proportions (median and interquartile range [IQR]). (B) Heatmap of changes in cell proportions of all data sets: non-COVID-19 and COVID-19. (C) Concordant and discordant changes in cellular proportions comparing COVID-19 to non-COVID-19 viral infections. Cell types that increased in COVID-19 (hence decreased in non-COVID-19) were CD56^bright NK cells, M2 macrophages, and total NK cells. Those that decreased in non-COVID-19 but increased in COVID-19 were CD56^dim NK cells, memory B cells, and eosinophils.