Coronavirus Susceptibility to the Antiviral Remdesivir (GS-5734) Is Mediated by the Viral Polymerase and the Proofreading Exoribonuclease

Abstract

Emerging coronaviruses (CoVs) cause severe disease in humans, but no approved therapeutics are available. The CoV nsp14 exoribonuclease (ExoN) has complicated development of antiviral nucleosides due to its proofreading activity. We recently reported that the nucleoside analogue GS-5734 (remdesivir) potently inhibits human and zoonotic CoVs in vitro and in a severe acute respiratory syndrome coronavirus (SARS-CoV) mouse model. However, studies with GS-5734 have not reported resistance associated with GS-5734, nor do we understand the action of GS-5734 in wild-type (WT) proofreading CoVs. Here, we show that GS-5734 inhibits murine hepatitis virus (MHV) with similar 50% effective concentration values (EC₅₀) as SARS-CoV and Middle East respiratory syndrome coronavirus (MERS-CoV). Passage of WT MHV in the presence of the GS-5734 parent nucleoside selected two mutations in the nsp12 polymerase at residues conserved across all CoVs that conferred up to 5.6-fold resistance to GS-5734, as determined by EC₅₀ The resistant viruses were unable to compete with WT in direct coinfection passage in the absence of GS-5734. Introduction of the MHV resistance mutations into SARS-CoV resulted in the same in vitro resistance phenotype and attenuated SARS-CoV pathogenesis in a mouse model. Finally, we demonstrate that an MHV mutant lacking ExoN proofreading was significantly more sensitive to GS-5734. Combined, the results indicate that GS-5734 interferes with the nsp12 polymerase even in the setting of intact ExoN proofreading activity and that resistance can be overcome with increased, nontoxic concentrations of GS-5734, further supporting the development of GS-5734 as a broad-spectrum therapeutic to protect against contemporary and emerging CoVs.IMPORTANCE Coronaviruses (CoVs) cause severe human infections, but there are no approved antivirals to treat these infections. Development of nucleoside-based therapeutics for CoV infections has been hampered by the presence of a proofreading exoribonuclease. Here, we expand the known efficacy of the nucleotide prodrug remdesivir (GS-5734) to include a group β-2a CoV. Further, GS-5734 potently inhibits CoVs with intact proofreading. Following selection with the GS-5734 parent nucleoside, 2 amino acid substitutions in the nsp12 polymerase at residues that are identical across CoVs provide low-level resistance to GS-5734. The resistance mutations decrease viral fitness of MHV in vitro and attenuate pathogenesis in a SARS-CoV animal model of infection. Together, these studies define the target of GS-5734 activity and demonstrate that resistance is difficult to select, only partial, and impairs fitness and virulence of MHV and SARS-CoV, supporting further development of GS-5734 as a potential effective pan-CoV antiviral.

Keywords: RNA polymerases; SARS-CoV; antiviral agents; antiviral resistance; coronavirus; nucleoside analogs; pandemic.

FIG 1

GS-441524 and GS-5734 inhibit MHV with minimal cytotoxicity. (A) GS-441524 is a 1′-cyano 4-aza-7,9-dideazaadenosine C-adenosine nucleoside analogue. (B) Change in viral titer of MHV compared to vehicle control after treatment with GS-441524. The data represent the results from 2 independent experiments, each with 3 replicates. Error bars represent standard error of the mean (SEM). (C) Viral titer data from panel B presented as the percentage of uninhibited control. The EC₅₀ of GS-441524 was calculated to be 1.1 μM. (D) Cell viability normalized to the vehicle control after treatment with GS-441524. The data represent the results from 3 independent experiments, each with 3 replicates. Error bars represent SEM. (E) GS-5734 is a monophosphoramidate prodrug of GS-441524. (F) Change in viral titer of MHV compared to vehicle control after treatment with GS-5734. The data represent the results from 4 independent experiments, each with 3 replicates. Error bars represent SEM. (G) Viral titer data from panel F presented as the percentage of uninhibited control. The EC₅₀ of GS-5734 was calculated to be 0.03 μM. (H) Cell viability normalized to vehicle control after treatment with GS-5734. The data represent the results from 3 independent experiments, each with 3 replicates. Error bars represent SEM.

FIG 2

Antiviral activity of GS-441524 and GS-5734 and modeled therapeutic efficacy of GS-5734 against SARS-CoV and MERS-CoV in HAE cultures. (A) Mean EC₅₀ values of SARS-CoV and MERS-CoV-infected HAE cultures from three different patient isolates treated with GS-441524 or GS-5734. (B) Viral titers of SARS-CoV-infected HAE cultures when treated with various doses of GS-5734 24 h postinfection. (C) Viral titers of MERS-CoV-infected HAE cultures when treated with various doses of GS-5734 24 h postinfection.

FIG 3

GS-5734 acts at early times postinfection to decrease viral RNA levels. (A) MHV viral titer after single-cycle infection and treatment with 2 μM GS-5734 at the indicated times postinfection. The data represent the results from 2 independent experiments, each with 3 replicates. Error bars represent SEM. Statistical significance compared to addition of GS-5734 at 0 h postinfection (p.i.) was determined by one-way analysis of variance (ANOVA) with Dunnett’s post hoc test for multiple comparisons and is denoted by asterisks: *, P < 0.05; **, P < 0.01; ***, P < 0.001. (B) Change in viral titer (black bars) and viral RNA levels (hatched bars) normalized to vehicle control 10 h postinfection after treatment with GS-5734. The data represent the results from 2 independent experiments, each with 3 replicates. Error bars represent SEM. Statistical significance compared to DMSO-treated samples was determined by one-way ANOVA with Dunnett’s post hoc test for multiple comparisons and is denoted by asterisks: **, P < 0.01; ***, P < 0.001.

FIG 4

Viruses lacking ExoN-mediated proofreading are more sensitive to GS-5734 inhibition. (A) Change in viral titer of WT and ExoN(−) viruses normalized to vehicle control after treatment with GS-5734. The data represent the results from 2 independent experiments, each with 3 replicates. Error bars represent SEM. Statistical significance compared to WT at each concentration was determined by t test using the Holm-Sidak method to correct for multiple comparisons and is denoted by asterisks: ***, P < 0.001. (B) Viral titer reduction from panel A represented as percentage of vehicle control, resulting in a WT EC₅₀ value of 0.087 μM and an ExoN(−) EC₅₀ of 0.019 μM.

FIG 5

Two mutations in the predicted fingers domain of the nsp12 RdRp, F476L and V553L, arose after 23 passages in the presence of GS-441524, and these residues are completely conserved across CoVs. (A) Schematic of the MHV genome displaying proteins with mutations identified after passage with GS-441524. The nsp12 RdRp is shown in yellow, nsp13-helicase in purple, ns2 in green, and spike in blue. (B) Linear schematic of nsp12 showing the locations of F476L and V553L within the predicted fingers of the RdRp core domain. (C) The previously described (32) Phyre2 model of the MHV RdRp core domain was used to map the predicted locations of the F476L and V553L residues, shown here in orange. The SDD active site residues are shown in yellow, the palm in red, the fingers in blue, and the thumb in green. (D) Amino acid conservation of F476 and V553 residues across CoVs demonstrating that both of these residues are completely conserved.

FIG 6

The F476L and V553L mutations mediate resistance to GS-5734 and are associated with a fitness defect. (A) Change in viral titer of WT, F476L, V553L, F476L + V553L, and p23 viruses normalized to the vehicle control after treatment with GS-5734. The data represent 2 independent experiments, each with 3 replicates. Error bars represent SEM. Statistical significance compared to WT was determined by Kolmogorov-Smirnov test and is denoted by asterisks: *, P < 0.05. (B) The change in genomic RNA levels of WT, F476L, V553L, and F476L + V553L MHV normalized to vehicle control after treatment with GS-5734. The data represent the results from 2 independent experiments, each with 3 replicates. Error bars represent SEM. Statistical significance compared to WT at each concentration was determined by one-way ANOVA with Dunnett’s post hoc test for multiple comparisons and is denoted by asterisks: *, P < 0.05; **, P < 0.01. (C) Multi-cycle replication kinetics of WT, F476L, V553L, or F476L + V553L MHV. The data represent the results from 2 independent experiments, each with 3 replicates. Error bars represent SEM. (D) Coinfection competition assay of WT and F476L V553L MHV at the indicated ratios. The percentage of the population of each mutation was assessed after four successive passages. The data are representative of 2 independent experiments each with 2 replicates. Error bars represent standard deviation (SD).

FIG 7

MHV resistance mutations confer resistance and are attenuated in SARS-CoV. (A) Change in luciferase activity normalized to vehicle control of WT or F480L + V557L SARS-CoV containing the NanoLUC reporter. The data are representative of the results from 2 independent experiments, each with 3 replicates. Error bars represent SEM. (B) Viral titer data from panel A presented as the percentage of vehicle control. This EC₅₀ value was calculated as 0.01 μM for WT and 0.06 μM for F480L + V557L virus, which represents a 6-fold increase in resistance. (C) Percent starting weight of BALB/c mice inoculated with WT or F480L + V557L SARS-CoV containing the NanoLUC reporter at 10³ or 10⁴ PFU. The data are representative of the results from 2 independent experiments, each with 10 to 12 animals per group. Error bars represent SEM. Statistical significance was determined by 2-way ANOVA and is denoted by asterisks: *, P < 0.05. (D) Lung titers from animals in panel C 2, 4, and 7 days postinfection. The data are representative of the results from 2 independent experiments, each with 3 animals per group. Error bars represent SEM. Statistical significance was determined by Wilcoxon test and is denoted by asterisks: *, P < 0.05.