Differential transcriptomic landscapes of SARS-CoV-2-infected and bystander lung cells reveal proviral genes selectively upregulated by viral replication

ABSTRACT

Bulk transcriptomic analyses of virus-cell interactions are commonly performed on mixed populations of infected and uninfected bystander cells and may thus lead to inaccurate interpretations. Moreover, they generally focus on the expression of the coding genome and not on the total transcriptome, which is largely composed of long non-coding RNAs (lncRNAs). We performed polyA+ and whole transcriptome analysis of lung epithelial A549 cells infected with SARS-CoV-2, which were sorted based on the expression of the viral protein Spike (S). To increase the sequencing depth and ameliorate the robustness of the analysis, the samples were depleted of viral transcripts, which constituted up to 85% of total reads in the S-positive cells. Results highlighted a high number of down-regulated genes upon infection, an indication of massive host transcription shutdown. We also noticed an increased level of intronic reads in infected cells as compared to control cells, suggesting a defect in mRNA splicing. Among the upregulated coding genes in S-positive cells, we recovered candidates previously identified by analysis performed on non-sorted cells, such as CXCL8 and CCL20, but also novel candidates, including IL- 32, ITGAM and the down-regulated FEN1. Comparison of mRNA abundances of few coding, non-coding and unannotated genes between non-sorted cells and sorted ones confirmed the accuracy of the approach. We also observed that S-negative bystander cells and mock-infected control cells exhibited very similar transcriptomic profiles. This was not due to a lack of communication between infected and bystander cells transcripts related to inflammatory cytokines underwent normal splicing and maturation in infected cells, with concomitant high levels of protein secretion. The observed lack of major transcriptomic changes in the bystander population can be linked to the insufficient interferon response in infected cells. Finally, we explored the functional implications of genes selectively upregulated in the infected subpopulation and observed previously uncharacterized proviral activities of lncRNA ADIRFAS1 and endogenous Interleukin 32 (IL-32), making them attractive targets for putative therapeutic strategies. Thus, analyzing the whole transcriptome of pure populations of infected lung cells allowed the accurate identification of cellular functions that are directly affected by infection and recovery of coding and non-coding genes relevant for SARS-CoV-2 replication.