Identification of druggable host dependency factors shared by multiple SARS-CoV-2 variants of concern

Abstract

The high mutation rate of SARS-CoV-2 leads to the emergence of multiple variants, some of which are resistant to vaccines and drugs targeting viral elements. Targeting host dependency factors, e.g. cellular proteins required for viral replication, would help prevent the development of resistance. However, it remains unclear whether different SARS-CoV-2 variants induce conserved cellular responses and exploit the same core host factors. To this end, we compared three variants of concern and found that the host transcriptional response was conserved, differing only in kinetics and magnitude. Clustered regularly interspaced short palindromic repeats screening identified host genes required for each variant during infection. Most of the genes were shared by multiple variants. We validated our hits with small molecules and repurposed the US Food and Drug Administration-approved drugs. All the drugs were highly active against all the tested variants, including new variants that emerged during the study (Delta and Omicron). Mechanistically, we identified reactive oxygen species production as a key step in early viral replication. Antioxidants such as N-acetyl cysteine (NAC) were effective against all the variants in both human lung cells and a humanized mouse model. Our study supports the use of available antioxidant drugs, such as NAC, as a general and effective anti-COVID-19 approach.

Keywords: N-acetyl cysteine; SARS-CoV-2; antivirals; host dependency factors; variants of concern.

Figure 1

Different SARS-CoV-2 variants induce highly similar transcriptional responses. (A) Heatmap of viral transcripts in uninfected cells (mock) or cells infected with the indicated variants. The expression levels of viral transcripts are shown as row-scaled Z-scores. (B) Volcano plots showing DEGs (log₂FC>0.59 or <−0.59, adj. P-value <0.05, vs. uninfected cells) in cells infected with the indicated variants at the indicated time points. The numbers of upregulated and downregulated DEGs are indicated in red and blue, respectively. FC, fold-change; adj. P-value, adjusted P-value. (C) Heatmap of DEGs in cells infected as indicated. (D) Correlation matrix displaying Pearson's correlation coefficients among the indicated samples. (E) Gene enrichment analysis of DEGs.

Figure 2

High-stringency CRISPR-based loss-of-function screen provides biological insights into different SARS-CoV-2 variants. (A) Schematics of the screening strategy. (B) Gene enrichment analysis of the hits identified by CRISPR screen. (C) Protein network of the screen hits, generated by Stringdb. Edge width is proportional to the strength of the interaction (see Supplementary Materials and methods). (D) Left: a heatmap displaying gRNA scores of candidate genes for each variant or across all three variants. Right: mean expression levels of candidate genes at 12 h.p.i., shown as log₂-normalized expression.

Figure 3

Silencing of the hit genes identified from the CRISPR-based screen reduces the production of new viral particles. Calu-3 (A) and Caco-2 (B) cells were transfected with non-targeting siRNA (control) or the indicated siRNAs and infected with SARS-CoV-2 variants (MOI = 0.1). Infective SARS-CoV-2 particles in the supernatant were assessed by PRA. The data are presented as mean ± standard deviation (SD) of two independent replicates.

Figure 4

Compounds targeting the identified hits display antiviral activity in human lung cells. Calu-3 cells were pretreated with the indicated compounds for 24 h and infected with SARS-CoV-2 variants (MOI = 0.1). The compounds were added to fresh medium at 1 h.p.i. and remained in the medium throughout the experiment. At 48 h.p.i., the cell medium was subjected to PRA, and the viral titre was calculated and expressed as PFU/ml. The bars indicate the means of two biological replicates. Each condition was tested in triplicate per replicate. Individual technical replicates are shown as dots.

Figure 5

Decreased ROS levels impair viral replication. (A–C) ROS levels in Calu-3 cells were measured via the H₂DCFDA assay. Each condition was tested in six replicates. (A) Intracellular ROS levels were measured at different time points after IKE administration (950 nM). (B) Intracellular ROS levels were measured at different time points after SARS-CoV-2 infection in the absence or presence of IKE (950 nM). In the latter case, the cells were pretreated with IKE for 24 h prior to SARS-CoV-2 infection. (C) Intracellular ROS levels were measured after treatment with IKE (950 nM), NAC (5 mM), or GSH (300 μM) for 24 h. (D and E) Calu-3 cells were pretreated with IKE (950 nM), NAC (5 mM), or GSH (300 μM) for 24 h prior to SARS-CoV-2 infection. Viral titres of the Wuhan (D) and Omicron (E) variants were assessed in triplicate by PRA after a single cycle of replication, i.e. 30 h.p.i. The data were normalized to the untreated control and are presented as mean ± SD of two biological replicates. Individual technical replicates are shown as dots.

Figure 6

Treatment with NAC inhibits SARS-CoV-2 infection in vivo. (A) Schematic of the in vivo experiment created with Biorender.com. i.p., intraperitoneal injection. (B) qPCR analysis of viral transcripts N and RdRp (R) in lungs infected with SARS-CoV-2 and treated with NAC or IKE. n = 21 (untreated), 14 (NAC), and 6 (IKE) from two independent experiments. The data are presented as whisker plots: midline indicates the median; box indicates the 25th–75th percentile; whisker indicates minimum to maximum values. The Kruskall–Wallis corrected Dunn's test for multiple comparisons was used. (C) Representative images of IHC for the viral N protein. The quantification result is shown in Supplementary Figure S5D. Scale bar, 100 μm.