Molnupiravir maintains antiviral activity against SARS-CoV-2 variants and exhibits a high barrier to the development of resistance

ABSTRACT Molnupiravir, an oral prodrug of N-hydroxycytidine (NHC), previously demonstrated broad in vitro antiviral activity against multiple RNA viruses and has shown a high barrier to the development of resistance. Here, we present the antiviral activity of NHC against recent SARS-CoV-2 variants and the results of resistance selection studies to better understand the potential for viral resistance to NHC. NHC activity against SARS-CoV-2 variants omicron (BA.1, BA.1.1, BA.2, BA.4, BA.4.6, BA.5, BQ.1.1, XBB.1, and XBB.1.5), alpha (B.1.1.7), beta (B.1.351), gamma (P.1), delta (B.1.617.2), lambda (C.37), and mu (B.1.621) was evaluated in Vero E6 cells using cytopathic effect assays. Resistance selection studies were performed by passaging SARS-CoV-2 (WA1) in the presence of NHC or a 3C-like protease inhibitor (MRK-A) in Vero E6 cells. Supernatants from cultures exhibiting a cytopathic effect score of ≥2 were re-passaged, and IC50 values were estimated. Whole-genome deep sequencing was performed on viral RNA isolated at each passage. NHC demonstrated similar potency against all SARS-CoV-2 variants evaluated. No evidence of SARS-CoV-2 phenotypic or genotypic resistance to NHC was observed following 30 passages. A random pattern of nucleotide changes was observed in NHC cultures, consistent with the drug’s mechanism of action. In contrast, resistance was readily selected in all three MRK-A control cultures with the selection of a T21I substitution in the 3C-like protease. In conclusion, molnupiravir maintains antiviral activity across all major SARS-CoV-2 variants. Furthermore, no evidence of viral resistance to NHC was observed, supporting previous reports that NHC has a high barrier to developing resistance.

NHC has also demonstrated a high barrier to the development of resistance in other RNA viruses.In previous studies with mouse hepatitis virus (coronavirus model) (20), Middle Eastern respiratory syndrome coronavirus (20), Venezuelan equine encephalitis virus (21), respiratory syncytial virus, and influenza virus (6,22), NHC demonstrated only modest shifts (i.e., twofold) in susceptibility after prolonged exposure to the viruses in vitro (6,(20)(21)(22).It is currently unknown whether this high barrier to antiviral resistance against NHC also extends to SARS-CoV-2.

In vitro antiviral activity of NHC against SARS-CoV-2 variants
In cell culture, NHC displayed potent antiviral activity against all evaluated SARS-CoV-2 omicron lineages compared with earlier variants alpha, beta, gamma, delta, lambda, and mu.The IC 50 values ranged from 0.28 to 5.50 µM (Table 1) and were comparable (within twofold) to that against the original WA1 isolate [IC 50 = 1.23 µM (range: 0.57-2.26)].

In vitro susceptibility of SARS-CoV-2 to NHC following 30 passages
The NHC IC 50 values against serial-passaged SARS-CoV-2 WA1 remained generally unchanged (less than 2.5-fold change) at each of the 30 passages, for all triplicate cultures (Fig. 1).In contrast, the susceptibility of SARS-CoV-2 WA1 to the control protease inhibitor, MRK-A, declined incrementally after several passages with an approximately 30-fold increase in IC 50 values by passage 16 and beyond (Fig. 1).
To confirm the observed results from the initial IC 50 estimates performed during the passaging experiment, susceptibility assessments were repeated for virus cultures from passages 5, 9, 15, 20, 25, and 30 using the Vero E6 cell (a subclone of Vero cells expressing high levels of ACE2; ATCC, Manassas, VA, USA) inhibitory assay with CPE scoring and with quantitative reverse transcription polymerase chain reaction (RT-PCR) method (data not shown).The results of this re-analysis confirmed that there were no increases in the NHC IC 50 values against any of the passaged viruses evaluated over the 30 passages (data not shown).

Genotypic characterization of passaged virus: analysis of viral sequence changes detected across the SARS-CoV-2 genome
Next-generation sequencing was performed on viral RNA from each replicate culture at each passage, and sequence changes relative to the original WA1 isolate were determined.As expected, viruses cultured in the presence of NHC accumulated nucleotide substitution variants throughout the viral genome (Fig. S1), which in turn caused random amino acid changes following 30 passages in culture (Table 2).Most nucleotide changes observed in NHC cultures were transitions, not transversions, consistent with the mechanism of action of NHC (Table 3).Although some amino acid substitutions were observed in the viral polymerase protein (nsp12) (Table 2), all substitutions were unique, and none were observed in more than one culture.Changes in the viral spike protein were noted in multiple replicates, including W64R, R682Q/W, K814R, and K986E.Currently, none of these substitutions are known to be associated with changes in antibody binding or viral infectivity.In contrast, viruses passaged in the presence of the MRK-A protease inhibitor were all rapidly selected for a T21I mutation in the viral nsp5 (3CL) protease after 10 passages and were fixed in the population by passage 30 (Table 2).

Phenotypic characterization of passaged viruses: viral growth kinetics and susceptibility to neutralizing antibodies
Viral isolates from passages 5, 9, 15, 20, 25, and 30 of untreated control and NHC cultures were assessed for viral growth kinetics on Vero E6 cells.The relative slope of the viral growth curve during exponential growth is plotted for NHC-treated and control (no-drug) cultures (Fig. S2).Following five passages, both control virus and NHC-passaged cultures showed an increase in viral replication compared with the original unpassaged WA1 isolate in Vero E6 cells.This increased replication is likely due to tissue culture adaptation of the virus because it occurred in both the control and the NHC-treated cultures.However, additional passaging did not result in further enhancement of viral replication capacity in either NHC or control cultures (Fig. S2).
To determine if amino acid substitutions observed in the spike protein following passaging with NHC affected neutralizing antibodies, viral isolates from NHC-treated and control SARS-CoV-2 WA1 passage 30 cultures were tested for susceptibility to two neutralizing antibodies, S309 (sotrovimab) and REGN10933 (casirivimab), targeting the SARS-CoV-2 spike protein.Results showed that NHC-passaged viruses remained as susceptible to both neutralizing antibodies as the passaged control virus (Fig. 2).(5,(23)(24)(25)(26).The suscepti bility of the passaged SARS-CoV-2 to the control inhibitor, MRK-A, declined incrementally over 30 passages, confirming the utility of this method for resistance selection.In contrast, the susceptibility of SARS-CoV-2 to inhibition by NHC did not significantly change over 30 passages, as demonstrated by the NHC IC 50 values of less than twofold change from baseline.This finding is consistent with data from studies of several other RNA viruses demonstrating a high barrier to resistance against NHC (6,21,22).

DISCUSSION
This high barrier to resistance suggests that molnupiravir should have long-term clinical utility against current and future variants (thus far, primarily differentiated by changes in the spike protein).This assumption is further supported by our in vitro data confirming NHC activity against omicron and other variants, indicating the potential of molnupiravir for the treatment of existing and future SARS-CoV-2 variants, including multiple omicron lineages.
Sequence analysis of the passaged virus confirmed the mechanism of action of molnupiravir.Each culture passaged with NHC showed an accumulation of random amino acid changes across the genome over time, which were not associated with changes in viral susceptibility to NHC.For NHC-passaged cultures, no amino acid substitutions in the viral replicase gene products were observed in more than one culture; however, several amino acid substitutions were observed in two or more cultures for the spike protein, that is, W64R, R682Q/W, K814R, and K986E.Currently, none of these substitutions are known to be associated with changes in antibody binding or viral infectivity (27,28).Moreover, following 30 passages with NHC, viruses from all three NHC-treated cultures remained susceptible to several monoclonal antibodies.In this study, passaging experiments were conducted with only SARS-CoV-2 wild-type virus but not with more recent variants.Therefore, the effect of existing mutations in currently circulating isolates on the resistance development could not be determined.However,  all SARS-CoV-2 variants we examined to date demonstrated an equivalent sensitivity to molnupiravir in cell culture.Furthermore, existing mutations in RNA-dependent RNA polymerase (nsp12) in the variant viruses do not overlap with those developed over passage.
To assess the impact of accumulated amino acid substitutions on viral growth, replication capacity studies were performed.The results of these studies demonstrated that viral replication kinetics increased after five passages in both the control and NHC cultures.This result is likely due to tissue culture adaptation of the virus during initial rounds of passaging and not due to the impact of NHC as no further increase in replication kinetics was observed during subsequent passaging despite the accumula tion of additional amino acid substitutions in the NHC cultures.To determine if amino acid substitutions observed in the spike protein following passaging with NHC affected neutralizing antibodies, viral isolates from NHC and SARS-CoV-2 WA1 passage 30 cultures were tested for susceptibility to two neutralizing antibodies, sotrovimab and casirivimab, targeting the SARS-CoV-2 spike protein.Sotrovimab neutralizes SARS-CoV-2 by binding to a strongly conserved epitope outside the receptor-binding motif (29).Casirivimab neutralizes SARS-CoV-2 by binding to the "up" and "up and down" conformation of the viral receptor-binding domain (30).In this study, NHC-passaged viruses remained as susceptible as the control virus (SARS-CoV-2 WA1) to both neutralizing antibodies, which  further suggests that the amino acid substitutions observed did not impact the activity of neutralizing antibodies.A T21I amino acid substitution in nsp5 was noted in all MRK-A triplicate cultures.This substitution is involved in substrate binding to the 3CL protease, which is the main protease involved in the replication cycle in SARS-CoV-2 (31).Of note, this same substitution was recently shown to contribute to resistance to nirmatrelvir during in vitro resistance studies and likely contributed to the loss of susceptibility to the MRK-A compound in these cultures (32).However, additional phenotyping studies are required to confirm the impact of this mutation on the susceptibility of SARS-CoV-2 to MRK-A.
In conclusion, the SARS-CoV-2 data presented here are consistent with previous studies with other RNA viruses and further support the conclusion that molnupira vir/NHC has a high barrier to the development of SARS-CoV-2 resistance, which is expected to translate to durable clinical efficacy.In addition, NHC was effective against all evaluated SARS-CoV-2 variants in vitro, further demonstrating that changes in the viral spike proteins associated with novel variants do not impact antiviral activity of NHC.

Evaluation of NHC antiviral activity against SARS-CoV-2 variants
The antiviral activity of NHC against the ancestral WA1 isolate and multiple SARS- ) was evaluated by assessing the ability of NHC to reduce viral CPEs in Vero E6 cells.All viruses were propagated in Vero E6 cells for one passage.Virus infectious titer for each stock was determined by an end point dilution CPE assay on Vero E6 cells as previously described (33).Antiviral assays were initially performed at Columbia University Medical Center using previously described methods (32,34).More recently, assays were conducted at the Research Laboratories of Merck & Co., Inc., Rahway, NJ, USA, where protection from virus-induced CPE was measured using the CellTiter-Glo 2.0 cell viability assay kit (Promega, Madison, WI, USA) to assess antiviral activity of NHC against SARS-CoV-2 variants WA1 and omicron lineages BA.4 and BA.5 in Vero E6/TMPRSS2 cells (BPS Bioscience, San Diego, CA, USA).Prior to assay start, Vero E6/TMPRSS2 cells were removed from maintenance flasks, counted, and diluted in assay media to 4.0 × 10 4 cells/mL.Prepared cells were placed into conical tubes for bulk infection with each SARS-CoV-2 isolate at a targeted multiplicity of infection (MOI) of 0.1.serial dilutions (12 points) of compounds (NHC or remdesivir) were dispensed into 384-well assay plates with the highest final drug concentration of 10 µM.Dimethyl sulfoxide (DMSO)-only (no drug control) wells were also prepared.After bulk infection, 50 µL (2,000 cells) of the infected cells was added to each well of the assay plates.Uninfected cells (50 µL at 4.0 × 10 4 cells/mL, i.e., 2,000 cells per well) were added to designated wells on the assay plates; these served as the minimum CPE control, whereas DMSO-treated infected cells served as the maximum CPE control.The assay plates were then incubated in a humidified chamber at 37°C, 5% carbon dioxide, for 3-4 days.
Cell viability in compound-treated wells was assessed using the CellTiter-Glo 2.0 kit (Promega, Madison, WI, USA) as per the manufacturer's protocol.The assay signal (luminescence) was read using an EnSight multimode plate reader (PerkinElmer, Inc., MA, USA).IC 50 values were calculated with a four-parameter, variable-slope dose-response curve fit of the assay data for each compound using GraphPad Prism 8.0 software (GraphPad Software Inc., San Diego, CA, USA).

In vitro selection and characterization of SARS-CoV-2 resistance to NHC versus a protease inhibitor
Prior to conducting, this study was reviewed and approved by the Columbia Univer sity Institutional Biosafety Committee (IBC) under appendices APA-AQNR6252 and APA-BDYW7534.All passaging experiments were conducted by trained personnel in Biosafety Level-3 (BSL-3) laboratories at Columbia University.
An in vitro culture system was used to select potential SARS-CoV-2 strains with reduced susceptibility to NHC or to an experimental 3CL protease inhibitor (MRK-A).In this system, parallel virus (SARS-CoV-2, US WA1/2020) cultures were serially passaged in cell culture in the presence of varying concentrations (0-100 µM) of NHC or MRK-A, and the development of drug resistance was monitored for each passage of culture.The passaged virus cultures were characterized phenotypically (i.e., drug susceptibility, viral growth kinetics, and susceptibility to neutralizing antibodies) and genotypically (i.e., sequencing of viral RNA to identify changes in viral proteins potentially associated with resistance).

Genotypic characterization of passaged virus
Viral RNA from culture supernatants was collected for each replicate culture, at each passage, for both NHC and MRK-A, and purified using the PureLink Pro 96 Viral RNA/DNA Purification Kit (ThermoFisher Scientific, Waltham, MA, USA).Whole genome SARS-CoV-2 sequencing libraries were prepared via the QIAseq DIRECT SARS-CoV-2 Kit (Qiagen, Hilden, Germany).
Sequencing was performed at the DNA Sequencing & Genotyping Center, University of Delaware, Ammon-Pinizzotto Biopharmaceutical Innovation Center (Newark, DE, USA), on the Illumina platform (San Diego, CA, USA), yielding 2 × 151 nucleotide paired-end reads.Genomic variants were identified using the following procedure: PCR primers were removed from the reads with the open-source program removePrimer (accessed via github.com/jsh58/AmpliconTools)with default parameters and "-ef 2 -fp −1,1." Only read pairs in which both primers were identified, and subsequently removed, were further analyzed.Read pairs generated from likely PCR chimeras (having primers removed on the same strand, in the wrong orientation, or more than 1 kb apart) were also filtered out; samples with excessive (>5%) chimeric fragments were removed from further analysis.For the remaining samples, the paired reads were merged together using NGmerge v0.3 (35) with default parameters and "-d." Reads that were successfully merged were aligned as unpaired full-length fragments by bowtie2 v2.3.5.1 (36) to the SARS-CoV-2 reference genome (GenBank ID NC_045512.2) (37).Reads that failed merging by NGmerge were aligned as regular paired-end reads by bowtie2 with default parameters and "-X 1000." The alignment files were combined and converted to binary form by SAMtools v1.9 (38).Pileup files were produced from the alignment files, counting only bases with quality scores of at least 30.Variants from the reference genome were identified at sites with a minimum read coverage of 50 and a minimum allele fraction of 0.10.Spurious variants due to large deletions were investigated and manually corrected.Variants for multiple samples were combined into a single table and annotated for the predicted effects on SARS-CoV-2 proteins.

Phenotypic characterization of passaged virus: viral growth kinetics and susceptibility to neutralizing antibodies
The passaged viruses were propagated in Vero E6 cells in 75 cm 2 flasks in the absence of drugs.The supernatant of each culture with a CPE score of 3+ or higher (>40% CPE in well) was harvested through centrifugation at 2,500 rpm for 10 min (Thermo Sorvall Legend XTR Refrigerated Centrifuge; Marshall Scientific, Hampton, NH, USA), aliquoted, and stored at −80°C.The propagated viruses (cultures with a CPE score 3+ or higher) were used to characterize viral growth kinetics or fitness in Vero E6 cells (39) at specific time points (passages 0, 5, 10, 15, 20, 25, and 30), and to assess susceptibility to NHC, MRK-A, and other agents, such as neutralizing antibodies, at passage 30.The infectious titer was determined by the end point dilution culture method as described by Cresta et al. (40).
For the assessment of the replication capacity of the passaged viruses, Vero E6 cell cultures, which were plated at 1.5 × 10 4 cells/mL per well a day before infection, were inoculated, in triplicate, with viruses from NHC-passaged culture at 200 TCID 50 units per well.Free virions were removed by a complete change of medium at 6 hours post infection.The supernatant was collected at 11, 23.5, 32.5, and 48 hours follow ing inoculation, and viral RNA was isolated using the PureLink Pro 96 Viral RNA/DNA Purification Kit (ThermoFisher Scientific).The viral RNA copy number was determined by reverse transcription quantitative PCR (RT-qPCR) using the TaqPath 1-Step RT-qPCR Master Mix, CG (ThermoFisher Scientific), with the CDC N1 primer and probe from the 2019-nCoV RUO Kit (Integrated DNA Technologies, Coralville, IA, USA) and using Applied Biosystems 7500 Fast Dx Real-Time PCR instrument (ThermoFisher Scientific).The ascending slope during the initial exponential growth phase on Vero E6 cells was estimated based on the increase in the viral RNA copy number in the culture superna tant.
To evaluate the susceptibility of the NHC-passaged viruses to neutralizing antibodies S309 (sotrovimab) and REGN10933 (casirivimab), Vero E6 cells were cultured in 96-well plates at 1.5 × 10 4 cells per well overnight.The following day, virus samples were mixed with neutralizing antibody in a total volume of 100 µL (in triplicate): 1,500 TCID 50 units of each of the three lineages (A, B, and C) of NHC-p30 (virus passaged for 30 times in the presence of NHC) along with the original virus, WA1-p0 (zero passages), and two no-drug control-passaged viruses, WA1-p30.1 and WA1-p30.2.The neutralizing antibody consisted of a fivefold dilution series starting at 10 µg/mL (i.e., 10, 2, 0.4, 0.08, 0.016, 0.0032, 0.00064, and 0 µg/mL).The virus-antibody mixture was incubated at 37°C for 1 hour and inoculated onto individual Vero E6 cell culture wells.Three days following infection, the CPE of each culture well was scored (as described previously), and the percentage of the CPE score was used to estimate the neutralizing antibody IC 50 values using GraphPad Prism v9.4.1.

TABLE 1
Antiviral activity of SARS-CoV-2 viral isolates a

TABLE 2
SARS-CoV-2 amino acid changes detected at >90% frequency at passage 30 in cultures exposed to NHC or MRK-A a,d,e

TABLE 3
Numbers of transitions and transversions in NHC-and MRK-A-passaged cultures a,b,c cNucleotide changes detected at ≥10% frequency at passage 30.