Repurposing screen identifies novel candidates for broad-spectrum coronavirus antivirals and druggable host targets

Abstract

Libraries composed of licensed drugs represent a vast repertoire of molecules modulating physiological processes in humans, providing unique opportunities for the discovery of host-targeting antivirals. We screened the Repurposing, Focused Rescue, and Accelerated Medchem (ReFRAME) repurposing library with approximately 12,000 molecules for broad-spectrum coronavirus antivirals and discovered 134 compounds inhibiting an alphacoronavirus and mapping to 58 molecular target categories. Dominant targets included the 5-hydroxytryptamine receptor, the dopamine receptor, and cyclin-dependent kinases. Gene knock-out of the drugs' host targets including cathepsin B and L (CTSB/L; VBY-825), the aryl hydrocarbon receptor (AHR; Phortress), the farnesyl-diphosphate farnesyltransferase 1 (FDFT1; P-3622), and the kelch-like ECH-associated protein 1 (KEAP1; Omaveloxolone), significantly modulated HCoV-229E infection, providing evidence that these compounds inhibited the virus through acting on their respective host targets. Counter-screening of all 134 primary compound candidates with SARS-CoV-2 and validation in primary cells identified Phortress, an AHR activating ligand, P-3622-targeting FDFT1, and Omaveloxolone, which activates the NFE2-like bZIP transcription factor 2 (NFE2L2) by liberating it from its endogenous inhibitor KEAP1, as antiviral candidates for both an Alpha- and a Betacoronavirus. This study provides an overview of HCoV-229E repurposing candidates and reveals novel potentially druggable viral host dependency factors hijacked by diverse coronaviruses.

Keywords: CRISPR/Cas9; HCoV-229E; SARS-CoV-2; antivirals; coronavirus; host-targeting antiviral therapy; repurposing.

Fig 1

Genetic diversity of human pathogenic coronaviruses, ReFRAME library composition, and screening workflow. (A) A section of the coronavirus phylogeny based on an IQ-TREE maximum-likelihood tree showing representative viruses of 20 species representing all 18 species of the genus Betacoronavirus and two of the genus Alphacoronavirus (adapted from (1)). Branch support was estimated using the Shimodaira-Hasegawa (SH)-like approximate likelihood ratio test with 1,000 replicates (SH = sequence homology). (B) Workflow of our approach to simultaneously identify broad-spectrum coronavirus antivirals and host targets thereof. Created with BioRender.com. (C) Composition of the ReFRAME drug repurposing library.

Fig 2

Results of ReFRAME library screening against the Alphacoronavirus HCoV-229E. (A) Results of the dual luciferase-based HCoV-229E/R-Luc screening of the ReFRAME library in Huh-7.5/F-Luc cells. Renilla luciferase (R-Luc) activity is proportional to virus infection, whereas firefly luciferase (F-Luc) activity corresponds to cell numbers and viability. R-Luc and F-Luc values detected in DMSO-treated, HCoV-229E-infected cells were set to 100% and corrected for background values in the case of R-Luc. The data were generated in one experiment (n = 1) with single luciferase measurements. (B) Stage of development of confirmed HCoV-229E hit compounds. (C) Fifty-eight distinct host targets are annotated to these hits. The number before the slash indicates the number of antiviral hits with this target annotation, the number after the slash is equivalent to the total number of molecules within this target category in the ReFRAME library. The number preceding the compound name indicates the IC₅₀ rank (see also Fig. S2 and Table S1). (D–G) Dose-dependent antiviral activity against HCoV-229E of the compound categories (D) compounds with diverse mechanisms, (E) CDK inhibitors, (F) regulators of inflammation, and (G) cathepsin inhibitors. One experiment with means of two independent luciferase measurements normalized to DMSO control is presented (n = 1).

Fig 3

CRISPR/Cas9 screening for potential HCoV-229E host factors. (A) Targets for CRISPR-Cas9-mediated knock-outs were selected based on the effect of their corresponding compound on cell viability (≥100% at a concentration of 5 µM) in uninfected Huh-7.5 cells, antiviral effect on HCoV-229E (IC₅₀), and RNA expression in Huh-7.5 cells (RPKM ≥1) (n = 1). (B) Effect of CRISPR/Cas9-mediated knock-outs for 32 putative host targets of a selected number of our hit compounds on HCoV-229E infection (R-Luc, light gray) and cell viability (F-Luc, dark gray). Cells were infected with hCoV-229E for 48 h. The data were normalized to the non-targeting control (nt). The bars represent the mean of the data acquired with three sgRNA per host factor. Each dot represents the mean of all data acquired using one sgRNA with between one and six biological replicates for each sgRNA (1 ≤ n ≤ 6) as indicated after each host factor name. Each biological replicate was composed of three technical replicates. (C) Ratio of HCoV-229E infection (R-Luc) to cell viability (F-Luc) normalized to the non-targeting control with data from (B). The mean (bar) as well as the individual values (dots) from four to six independent experiments (4 ≤ n ≤ 6) are shown for each sgRNA (#1-#3). One-way ANOVA, Dunnett’s multiple comparisons test, only significant differences compared to the non-targeting control are indicated by asterisks, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig 4

Dose-dependent antiviral activity against HCoV-229E infection of squalene synthases-targeting drugs. (A-E) Twelve-step dose-titration of compounds targeting squalene synthases on Huh-7.5/F-Luc/Cas9 cells. Serial 1:3 dilutions of compounds starting at 0.1 µM. The means of three independent experiments with two technical replicates, each, normalized to the DMSO controls are shown (n = 3). Dose-titration of (A) P-3622, (B) Zaragozic acid, (C) Remdesivir, (D) YM53601. (E) IC₅₀ (nM) and CC₅₀ (µM) values of the titrated drugs using non-linear curve fit calculations shown in A–E (GraphPad Prism 9).

Fig 5

Counter-screening of confirmed HCoV-229E hits with SARS-CoV-2. Calu-3 cells were infected with SARS-CoV-2 (strain SARS-CoV-2/München-1.2/2020/984, P4) in the presence or absence of the indicated compounds (final concentration 5 µM, one technical replicate). (A) Infection efficiency was analyzed 48 hours post-inoculation by quantification of viral genome equivalents in the culture fluid of cells. qRT-PCR measurements were performed in duplicate. Culture fluid of cells inoculated with heat-inactivated SARS-CoV-2 served as background control (light blue circles). Compounds that reduced virus load more than two standard deviations below the mean value of the DMSO control were recognized as candidate antivirals. (B) Immunofluorescence analysis of infected cells. DNA was stained with DAPI (blue), and SARS-CoV-2 nucleoprotein with a monoclonal antibody (green) (scale bar 125 µm). (C) Imaging-based quantification of total cell numbers and infected cell numbers. Images were automatically quantified as outlined in the methods section. The relative number of infected cells (N protein expressing cells) is correlated against the number of total cells. Means of four images per well for one experiment (n = 1) are presented. DMSO and heat-inactivated virus controls are depicted with light green and light blue circles, respectively. Data from Remdesivir-treated cells are depicted with red circles. Validated broad-spectrum coronavirus antivirals meeting our inclusion criteria are shown with dark blue circles. The inclusion criterion was a reduction of virus-infected cell numbers by more than 50% and a residual total cell number of more than 75% of DMSO solvent control. The data were generated in one experiment due to the limited availability of compounds.

Fig 6

Dose titration of candidates with broad-spectrum coronavirus activity in Calu-3 cells. (A-G) Calu-3 cells were seeded 48 hours prior to infection with SARS-CoV-2 and were pre-treated with different doses of the candidate antivirals selected from previous screenings 2 hours before infection. Forty-eight hours post-infection viral genome copies in the supernatant were quantified via qRT-PCR. Cytotoxicity was quantified via LDH assay using the supernatant of drug-treated, non-infected cells 48 hours after the start of drug treatment. Both infectivity (black dots) and cytotoxicity (clear squares) were normalized to the DMSO control. Mean values of three independent experiments (n = 3) with two technical replicates each are shown with a non-linear curve fit including constraints to bottom = 0% and top = 100% (infectivity) and bottom = 100% (cytotoxicity). (H) IC₅₀, CC₅₀, IC₉₀, and the specificity index (SI) for each compound calculated from the curve fits shown in (A-G).

Fig 7

Validation of candidates with broad-spectrum coronavirus activity in well-differentiated primary human lung cells. Primary cells were differentiated and grown as pseudostratified epithelium under air-liquid interface culture conditions. Cells were pre-treated for 2 hours from the basolateral compartment with twofold serial dilutions of indicated compounds starting at 10 µM prior to inoculation with SARS-CoV-2 in the presence of compounds (apical and basal treatment). Virus inoculum was removed 1 hour post-infection and cells were washed once with HBSS and further cultured under air-liquid interface conditions. The drugs in the basolateral compartment were present throughout the experiment. Seventy-two hours post-infection, the new progeny viruses were collected by HBSS washes of the apical compartment prior to lysis of the cells. The antiviral activity of the respective compounds was determined by qRT-PCR analyses of purified RNA from culture fluids (A) and cell extracts (B). (C) Cell viability was assessed by quantification of GAPDH mRNA. Mean values (bar) and individual values (points) of three to six donors are given (n = 3 for AXT-914, Phortress, Flavopiridol, Bardoxolone, Omaveloxolone, P-3622; n = 6 for Remdesivir). The data were normalized to the respective DMSO solvent control. Dotted lines indicate DMSO solvent control set to 100% and dashed lines give assay background from cells inoculated with non-replicating, heat-inactivated virus. The qRT-PCR analysis from each sample was performed in one technical replicate for AXT-914, Phortress, Flavopiridol, Bardoxolone, and Omaveloxolone and two technical replicates for P-3622 and Remdesivir.