Rapid incorporation of Favipiravir by the fast and permissive viral RNA polymerase complex results in SARS-CoV-2 lethal mutagenesis

Abstract

The ongoing Corona Virus Disease 2019 (COVID-19) pandemic, caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), has emphasized the urgent need for antiviral therapeutics. The viral RNA-dependent-RNA-polymerase (RdRp) is a promising target with polymerase inhibitors successfully used for the treatment of several viral diseases. We demonstrate here that Favipiravir predominantly exerts an antiviral effect through lethal mutagenesis. The SARS-CoV RdRp complex is at least 10-fold more active than any other viral RdRp known. It possesses both unusually high nucleotide incorporation rates and high-error rates allowing facile insertion of Favipiravir into viral RNA, provoking C-to-U and G-to-A transitions in the already low cytosine content SARS-CoV-2 genome. The coronavirus RdRp complex represents an Achilles heel for SARS-CoV, supporting nucleoside analogues as promising candidates for the treatment of COVID-19.

Fig. 1. Antiviral effects of T-705 on SARS-CoV-2.

a In vitro effects of T-705 on SARS-CoV-2. Distribution of the mutations along the SARS-CoV-2 genome and number of mutations observed in presence or absence of T-705. A 3-fold increase in the presence of the drug is observed (P < 0.001, Pearson’s χ² test with Yates’ continuity correction). b Quantification of the antiviral effect of T-705 by genome copy number, virus-mediated CPE and number of infectious particles. Mean ± standard deviation (SD) shown (n = 3). Source data are provided as a Source Data file.

Fig. 2. Polymerisation modes of SARS-CoV nsp12.

a, b Steady-state data showing extension of hairpin (HP) and annealed primer-template (PT) RNA substrates by the nsp12:7L8:8 complex. c Fraction of primer (P, blue), intermediate (black) and full-length (FL, red) species over time, as analysed from gel shown in panel b. Additional extension products beyond the full-length observed on the PT substrate are attributed to partially-denatured forms of the full-length product and were included in the analysis of the FL fraction. Source data are provided as a Source Data file.

Fig. 3. Incorporation of nucleoside analogues T-705 and T-1105.

a Structures of GTP, T-705 and T-1105 in their ribonucleotide triphosphate forms. b Elongation of the PT10/20 substrate in the presence of various analogue concentrations and the absence of ATP or GTP. Red dots show analogue incorporation sites at the omitted nucleotide positions, incorporation of correct nucleotides are indicated by blue letters and GTP:U mismatches by red letters. c Left panel shows products obtained with 100 µM of each analogue after 5 min. Right panel shows time-course in the presence of 10 µM of each analogue and absence of GTP. Site assignments for the hairpin substrate are not shown due to drastic differences in RNA migration attributed to changes in residual RNA structure following analogue incorporation. Full time-course series at multiple analogue concentrations are shown in Supplementary Fig. 5b. Source data are provided as a Source Data file.

Fig. 4. Comparison of T-1105-RTP:U and natural GTP:U mismatch incorporation levels.

a Elongation reactions with the PT substrate done in the absence of ATP with 50 µM each GTP, UTP and CTP show rapid addition of the first uracil followed by slow misincorporation of a GTP:U mismatch. In the presence of 1 µM T-1105-RTP (with 50 µM each GTP, UTP and CTP) the analogue incorporation is on a similar timescale as the native GTP:U mismatch. b Analysis of incorporation levels show a burst phase followed by linear product buildup over a 60-s timeframe that is consistent with the measured lifetime of the nsp12-7L8-8 elongation complex. Both the burst amplitude and linear rate indicate that 1 µM T-1105-RTP is incorporated ~5-fold faster than the natural GTP:U mismatch at 50 µM GTP. Quantitation reflects slopes and intercepts with standard errors obtained from a linear curve fit to the initial rate data (0–60 s). Expanded concentration series shown in Supplementary Fig. 5b, bottom panel. Source data are provided as a Source Data file.

Fig. 5. Pre steady-state elongation by nsp12-7L8-8 complex.

a EDTA quench-flow data showing rapid elongation to +7 products on the PT10/20 substrate in the absence of CTP. b, c Fitting of the quench-flow data to models with discrete rates for each elongation step (b, colour key in a) or with a single average incorporation rate for each step c. d Nucleotide concentration dependence yielding maximal elongation rates of 150 ± 30 s⁻¹ on the HP substrate and 91 ± 4 s⁻¹ on the PT substrate based on error weighted fits to the observed average elongation rates. Experiments were run at three NTP concentrations with 20 timepoints per experiment, including repeats of 100 ms timepoints (n = 3). Supplementary Table 1 lists rates ± standard error (S.E) for both an overall average rate and for each individual incorporation step. Source data are provided as a Source Data file.

Fig. 6. Comparison of viral polymerase active site structures and sequences.

a Analogous post-catalysis, pre-translocation structures of CoV nsp12 and poliovirus 3D^pol after incorporation of remdesivir (RDV) and deoxycytosine (dCYT), respectively. The CoV polymerases have replaced a motif F glutamate residue (3D^pol Glu161) with an alanine (nsp12 Ala547), removing a highly conserved interaction that positions the motif F arginine for interactions with the NTP and pyrophosphate. b Alignment of representative viral RdRp sequences showing the large genome nidoviruses have a SDD instead of GDD sequence in the palm domain motif C (tan) and alanine, serine or glutamine in place of the aforementioned glutamate (blue) found in other positive strand RNA viruses. The NTP-interacting arginine (yellow) is conserved, but the overall length of the motif F loop is shorter in the nidoviruses (numbers reflect omitted residues).