Increased flexibility of the SARS-CoV-2 RNA-binding site causes resistance to remdesivir

Fig 1. Identification of RDV-resistant mutations of SARS-CoV-2. (A) Schematic image of the in vitro serial passage of SARS-CoV-2 in the presence of RDV. Supernatants of virus-infected cells were passaged with gradually increased concentrations of RDV. The virus sequence was examined after 10 passages. (B) The SARS-CoV-2 genome with the locations of mutations, which demonstrated more than 80% of frequency in Miseq sequence and more than 90% of frequency in single virus sequence after 10-time passages of SARS-CoV-2 with RDV.

Fig 2. Infection kinetics of recombinant SARS-CoV-2 with NSP12 mutations. (A and B) Infectious virus production in the absence (A) and presence (B) of RDV. Supernatants of virus-infected cells were collected from 12–96 hpi and virus titers were determined by titration. Statistical significances were determined by Kruskal–Wallis test with the two stage linear step-up procedure of Benjamini, Kreiger and Yekutieli. Significant differences compared with WT virus are indicated by an asterisk (*p<0.05). (C and D) Infection rate in the absence (C) and presence (D) of RDV. Virus-infected cells were fixed and stained with antibodies for SARS-CoV-2 and cell nuclei. Infection rates were calculated by dividing the numbers of virus-positive cells by the numbers of nuclei. Statistical significances were determined by one-way ANOVA with Dunnett’s test. Significant differences compared with WT virus are indicated by asterisks (***p<0.001).

Fig 3. Quantifying antiviral effects of RDV against SARS-CoV-2 infection. (A) Schematic diagram of SARS-CoV-2 infection under RDV treatment in cell culture. The target cells are infected by infectious virus at rate β and become eclipse phase cells. The average duration of the eclipse phase is 1/k days and these eclipse phase cells start producing viruses at rate p (i.e., become virus-producing cells). The target cells and eclipse phase cells are assumed to divide at rate g, and virus-producing cells die at rate δ. Progeny infectious viruses are cleared at rate c. RDV blocks virus production by inhibiting viral replication in infected cells with inhibition rate ε. (B) Comparison of the fold-change of virus production rate, η, for eight different RDV-resistant viruses and WT virus. Orange dots are the mean value and green lines show 95% CI, which were estimated from the experiments with/without RDV. For WT, we consider η = 1, shown by the black dot. (C) Comparison of the antiviral effect of RDV, ε, for eight different RDV-resistant viruses and WT virus. Red and black dots are the mean values of the antiviral effects and blue lines show 95% CI, which were estimated from infection…

Fig 4. NSP12-RNA binding structure and comparison of thermodynamic stability. (A) Overall view of NSP12 protein and location of the RDV. (B) RMSD comparison of RNA-binding proteins. (C) Comparison of the molecular vibrations of WT and each mutant. The pink and blue spheres represent regions of large oscillations in the mutant and WT, respectively.

Fig 5. The pathogenicity and growth kinetics of RDV-resistant SARS-CoV-2 in Syrian hamsters. (A) Schematic diagram of WT and R10/E796G/C799F viruses infection and animal operations. Syrian hamsters were inoculated with saline (n = 4, uninfected control), WT (n = 6) or R10/E796G/C799F virus (n = 6). Body weight was daily examined and oral swabs were collected at indicated timepoints. (B) Change of body weight. Data are mean ± SD. Statistical significances were determined by Kruskal–Wallis test with the two stage linear step-up procedure of Benjamini, Kreiger and Yekutieli. Significant differences compared with WT virus are indicated by an asterisk (*p<0.05). (C) Viral RNA load in oral swab was measured. Data are mean ± SD. Statistical significances (*p<0.05). were determined by two-tailed Student’s t test.