SARS-CoV-2 3CLpro mutations selected in a VSV-based system confer resistance to nirmatrelvir, ensitrelvir, and GC376

Abstract

Protease inhibitors are among the most powerful antiviral drugs. Nirmatrelvir is the first protease inhibitor specifically developed against the SARS-CoV-2 protease 3CL^pro that has been licensed for clinical use. To identify mutations that confer resistance to this protease inhibitor, we engineered a chimeric vesicular stomatitis virus (VSV) that expressed a polyprotein composed of the VSV glycoprotein (G), the SARS-CoV-2 3CL^pro, and the VSV polymerase (L). Viral replication was thus dependent on the autocatalytic processing of this precursor protein by 3CL^pro and release of the functional viral proteins G and L, and replication of this chimeric VSV was effectively inhibited by nirmatrelvir. Using this system, we applied nirmatrelvir to select for resistance mutations. Resistance was confirmed by retesting nirmatrelvir against the selected mutations in additional VSV-based systems, in an independently developed cellular system, in a biochemical assay, and in a recombinant SARS-CoV-2 system. We demonstrate that some mutants are cross-resistant to ensitrelvir and GC376, whereas others are less resistant to these compounds. Furthermore, we found that most of these resistance mutations already existed in SARS-CoV-2 sequences that have been deposited in the NCBI and GISAID databases, indicating that these mutations were present in circulating SARS-CoV-2 strains.

Fig. 1.. A VSV-based non-gain-of-function system was developed to predict SARS-CoV-2 3CL^pro mutations.

(A) 3CL^pro from SARS-CoV-2 and mouse hepatitis virus (MHV) were tested in a gain-of-signal assay. Data are presented as individual points of n = 3 biologically independent replicates per condition for SARS-CoV-2 3CL^pro and n = 2 for MHV 3CL^pro, average values are represented by histogram bars. (B) Replication kinetics are shown for wild-type (wt) VSV-G-3CL^pro-L and GC376-selected F305L variant. Data are presented as SD of n = 2 biologically independent replicates per condition. (C) GC376 and nirmatrelvir dose responses are shown for wild-type (wt) VSV-G-3CL^pro-L and GC376-selected F305L variant. Data are presented as means of n = 2 (GC376) and n = 3 (Nirmatrelvir) biologically independent replicates per condition. (D) Sequence alignment of C-terminal autocleavage sites is shown for SARS-CoV-2 3CL^pro and related coronaviruses.

Fig. 2.. Sequencing of 3CL^pro escape mutants and comparison to data bases and Paxlovid EUA information.

(A) Mutants were recovered from VSV-G-3CL^pro-L wild-type (*) and the F305L variant (red **). Autocleavage site mutants are colored in turquoise, catalytic site mutants in green, near catalytic site mutants in light green, dimerization interface mutants in yellow and “allosteric” mutants in white. Viruses with more than one mutation are displayed above in a gray box and named a to f. The number of mutated sequences in the databases from NCBI and GISAID are displayed below the mutations in gray. If specific mutations were not present in the database, the residue is displayed with any mutation that occurred at this position. Multiple such different amino acid changes that were not selected in our virus are displayed with X (N203X, V204X). Mutations from the Paxlovid EUA are divided into mutations found in cell culture and mutations sequenced from treated patients. The coverage of mutation entries was obtained on June second, 2022. (B) Visualizations of mutation-affected residues are shown. Residues that were mutated one time are highlighted in yellow, two times in light orange, three times dark orange, and four times in red. The 3CL^pro protease dimer with bound nirmatrelvir (blue) was visualized in ChimeraX from the Protein Data Bank structure 7vh8 (32). Catalytic center mutations are within a range of 5 Å as visualized in dark green.

Fig. 3.. Replication kinetics and nirmatrelvir dose responses of parental VSV-G-3CL^pro-L and mutant variants.

(A to H) Replication kinetics and dose responses are shown for wild-type (A), L167F (B), L167F-2 (C), Y54C (D), N203D (E), F305L (F), G138S/F305L (G) and Q192R/F305L (H) VSV-G-3CL^pro-L. Supernatants for replication kinetics were collected at indicated time points. Supernatants for virus nirmatrelvir dose response were collected 24 hours after infection. (n = 2 biologically independent replicates per condition with individual data points shown and connecting lines of mean values). neg, without nirmatrelvir; TCID₅₀, 50% tissue culture infective dose.

Fig. 4.. Re-introduction of individual or dual 3CL^pro mutations confirms their resistance phenotype.

A graphic representation of the 3CL^pro-on and 3CL^pro-off system used to measure the inhibitory activity of the protease inhibitor against the different 3CL^pro mutants can be found in fig S3. (A) Gain-of-signal assay results are shown for single catalytic site mutations Y54C and L167F with nirmatrelvir. Data are presented as the standard deviation (SD) of n = 3 biologically independent replicates per condition. (B) Loss-of-signal assay results are shown for single catalytic site mutations Y54C and L167F with nirmatrelvir. Data are presented as the SD of n = 4 biologically independent replicates per condition. (C) Gain-of-signal assay results are shown for catalytic site mutations Y54C and L167F in combination with the Omicron 3CL^pro signature mutation P132H. Data are presented as the SD of n = 4 biologically independent replicates per condition. (D) Loss-of-signal assay results are shown for single catalytic site mutations Y54C and L167F in combination with the Omicron 3CL^pro signature mutation P132H. Data are presented as the SD of n = 4 biologically independent replicates per condition. (E) Gain-of-signal assay results are shown for double mutant L167F/F305L versus wild-type and single mutant L167F. Data are presented as the SD of n = 4 biologically independent replicates per condition. (F) Loss-of-signal assay results are shown for double mutant L167F/F305L versus wild-type and single mutant L167F. Data are presented as the SD of n = 4 biologically independent replicates per condition. (G) Gain-of-signal assay results are shown for double mutant Q192R-F305L versus wild-type and single mutant Q192R. Data are presented as the SD of n = 4 biologically independent replicates per condition. (H) Loss-of-signal assay results are shown for double mutant Q192R/F305L versus wild-type and single mutant Q192R. Data are presented as the SD of n = 4 biologically independent replicates per condition. (I) Gain-of-signal assay results are shown for mutants A194S and G138S versus wild-type. Data are presented as the SD of n = 4 biologically independent replicates per condition. (J) Loss-of-signal assay results are shown for of mutants A194S and G138S versus wild-type. Data are presented as the SD of n = 4 biologically independent replicates per condition.

Fig. 5.. Nirmatrelvir and GC376 react differently to mutants.

(A) Gain-of-signal assay results are shown for single mutants Y54C and L167F versus wild-type tested with GC376 and nirmatrelvir (Y54C: n = 4, L167F: n = 3 biologically independent replicates per condition). (B) GC376 (PDB: 7k0g) and nirmatrelvir (PDB: 7vh8) 3CL^pro crystal structures are shown with GC376 in green (and colored by heteroatom) and nirmatrelvir in light blue (and colored by heteroatom) and proximal residues in orange (within zone of 5 Å). Compound to residue distances are shown with dotted purple lines. (C) Fitting of gain-of-signal assay results are shown for single mutants Y54C and L167F versus wild-type tested with GC376 and nirmatrelvir. (D) Fitting of gain-of-signal assay results are shown for single mutant Q192R versus wild-type tested with GC376 and nirmatrelvir. Data in (C and D) are presented as the SD of n = 4 biologically independent replicates per condition.

Fig. 6.. Cross-testing mutants and validation of enzyme kinetics.

(A to C) Cross validation with cellular gain-of-signal assay based on Src-3CL^pro-Tat-Luc polyprotein. Wt Src-3CL^pro-Tat-Luc and L167F mutant were tested with nirmatrelvir (A), ensitrelvir (B), and GC376 (C). Data are presented as mean ± the standard deviation (SD) of n = 3 biologically independent replicates per condition. (D) Enzyme kinetics were measured for wt 3CL^pro and mutants with the substrate Ac-Abu-Tle-Leu-Gln↓MCA releasing the fluorogenic molecule MCA. Two biologically independent replicates per condition were used to calculate means for slopes. Slopes were used to compose a Michaelis Menten graph. Relative fluorescent units / minute (RFU/min) were used to plot the velocity of enzymes in the Michaelis Menten graph. (E) Results of a biochemical assay used for cross-validation are shown. In presence of an appropriate protease inhibitor, 3CL^pro cannot cleave the substrate and fluorescence is low. Without inhibitor, 3CL^pro cleaves the substrate peptide (KTSAVLQSGFRKME), quencher (DABCYL) and fluorogen (EDANS) are separated, and fluorescence increases. Nirmatrelvir dose responses are shown for wt versus mutant 3CL^pro variants; IC₅₀ fold changes show varying resistance of the mutant enzymes. Data are presented as mean ± the standard deviation (SD) of n = 2 biologically independent replicates per condition. (F) Recombinant wt, L167F and L167F/F305L SARS-CoV-2-mCherry (rWA1) exemplary plaques were imaged without magnification. (G) Replication kinetics are shown for recombinant wt SARS-CoV-2 (rWA1) versus L167F single and L167F F305L double mutant viruses. Data are presented as mean ± the standard deviation (SD) of n = 3 biologically independent replicates per condition. (H) Results are shown comparing the resistance of L167F single and L167F F305L double mutants to nirmatrelvir versus wt rWA1 expressing mCherry (top). The fold change in IC₅₀ values are shown on the right. Data are shown for 48 and 72 hours post infection (hpi). The dotted line indicates 50% inhibition. Data are presented as mean ± SEM from quadruplicate wells of 2 independent experiments.

Fig. 7.. Structural modelling of mutant 3CL^pro variants.

(A) Colorimetric mapping of the dAffinity value (kcal/mol) by virtual alanine scanning with MOE suite. Residues within 5 Å of the nirmatrelvir position are displayed. Colors range from blue (negative values, indicating increased protein-ligand affinity) to red (positive values, indicating decreased protein-ligand affinity). The nirmatrelvir (NV) structure is shown in light blue. (B) Colorimetric mapping of the dStability value (kcal/mol), computed as above for (A). Colors range from blue (negative values, indicating increased in the protein stability) to red (positive values, indicating decreased protein stability. (C) The catalytic center of 3CL^pro from PDB structure 7vh8 is shown with nirmatrelvir bound. Y54 (left) forms a strong hydrogen bond (HB, highlighted with a blue dashed line) with D187, whereas nirmatrelvir is at a distance of 3.5 Å (yellow dashed line). The exchange of Y54 with C (right) leads to a loss of the hydrogen bond to D187 and makes room in the nirmatrelvir binding pocket due to the smaller side-chain of cysteine versus tyrosine. (D) G138 (left) contacts H172 with a hydrogen bond. S138 (right) forms several new hydrogen bonds with the backbone hydrogen of F140, backbone oxygen of K137 and the sulfur of C128. (E) Q192 (left) forms hydrogen bonds with the oxygen and nitrogen of V186, the oxygen of R188, and stabilizes the polar contact to the CF₃ group of nirmatrelvir. R192 (right) disrupts this hydrogen bond network; subsequent rearrangement could form additional interactions with the CF₃ group.