Natural evolution of SARS-CoV-2 variants in K18-ACE2 mice gives rise to more virulent virus and variant alleles associated with treatment resistance. (preprint)

Title: Natural evolution of SARS-CoV-2 variants in K18-ACE2 mice gives rise to more virulent

duplicate reads flagged. The minimum number of aligned reads for each was logged, and each 1 4 3 sample was again aligned and processed, then randomly down sampled to match this minimum 1 4 4 number of reads. All variants with a quality of less than 20 or a depth below 10 were removed 1 4 5 from the analysis. Any variant alleles that overlapped with ARTIC sequencing primers were also 1 4 6 removed. Merged VCF files, representing the combination of reads across two sequencing 1 4 7 batches for each sample, were used for the paper's main results. predictor (VEP) (https://covid-19.ensembl.org/index.html) to determine functional consequences 1 5 0 and the involved gene and amino acid change (21). Variants whose VAF changed at least once 1 5 1 across passages during the study were logged and input to the UCSC Genome Browser 1 5 2 (https://genome.ucsc.edu/), checking for annotations of antibody escape, CD8 escape, vaccine 1 5 3 targets, drug resistance, or variants of concern (22).

5 4
Antibody neutralization studies. Sera from three, healthy vaccinated individuals (ERB #2022-1 5 5 6204) were analyzed for neutralizing activity against Wuhan-like, and Beta and Delta P0 and 1 5 6 P20 viruses (n = 12). Serially diluted sera (in quadruplicate) were incubated with 100 TCID 50 of 1 5 7 virus for 1 h at 37˚C before addition to Vero E6 cells. Three days later, the 96-well plates were 1 5 8 observed using an inverted microscope for signs of infection. Neutralization titers were 1 5 9 determined and defined as the highest serum dilution preventing infection. 1 6 0 Ethics. The study was conducted in accordance with the Declaration of Helsinki, and mouse 1 6 1 protocols were approved by the Comité de protection des animaux de l'Université Laval. across P13, P17, and P20 was 34,253,784. In Beta lineage, there was a significantly lower 1 7 7 variant allele frequency (VAF) in P17 and P20 versus other passages (adjusted p-value < 0.05), 1 7 8 with a significant difference in Delta comparing P10 and P20 (Figure 2a-b). As there was no 1 7 9 significant change in allele depth across the study, these VAF differences could be due to 1 8 0 growing allelic diversity. Such significant changes, less consistent with random allelic drift, could the study (adjusted p-value > 0.05) (Figure 3), suggesting that the overall allele population did 1 8 8 not significantly change in our model without imposed selective pressures (6).

Discussion: 2 3 9
In this study, we serially infected mice with Beta or Delta lineage virus for twenty passages 2 4 0 without selective pressures. The evolved viruses developed and maintained variant alleles 2 4 1 associated with treatment resistance and later lineages like Omicron (Figure 4, Table 1), were 2 4 2 more virulent (Figure 5), grew faster and to higher levels (Delta) (Figure 6) and were less 2 4 3 sensitive to antibody neutralization (Delta) (Figure 7), demonstrating immune escape potential.

4 4
These developments occurred within roughly two months, illustrating the model's potential in 2 4 5 rapidly emulating pandemic progression that could facilitate treatment/vaccine development less 2 4 6 sensitive to viral evolution. The model can readily be adapted to mimic viral adaptation in the 2 4 7 context of immune or anti-viral selection pressures, affording rapid results in a system often 2 4 8 seen in real world conditions. 2 4 9 Our model lacked selective pressures, mimicking the evolution of SARS-CoV-2 in a naïve 2 5 0 population. Using K18-ACE2 transgenic mice eliminated the selective pressure of adapting virus 2 5 1 to the murine ACE2 receptor. The Beta P0 virus harbored the N501Y mutation previously 2 5 2 reported to enable adaptation of reference SARS-CoV-2 in non transgenic mice (12). This 2 5 3 mutation was maintained throughout passages, suggesting that this virus can infect cells 2 5 4 through the murine ACE2 receptor as well (Figure 4). Our data supports this, as non transgenic 2 5 5 C57BL6 mice were equally susceptible to infection by Beta P0 and P20 viruses. This capacity of 2 5 6 Beta to use human and murine receptors might explain, at least in part, the greater virulence of 2 5 7 the Beta isolate in ACE2 transgenic mice relative to other viral isolates, such as Wuhan and 2 5 8 Delta lacking the N501Y mutation in transgenic mouse studies (29).

5 9
The Delta P0 virus lacked the N501Y mutation and never acquired it following repeated 2 6 0 passages in ACE2 transgenic mice. This highlights that when lacking selective pressure, such 2 6 1 as neutralizing antibodies, the appearance of N501Y is not favored (Figure 4). Acquisition of 1 2 N501Y in the human population was first reported almost a year after the initial report of the 2 6 3 SARS-CoV-2 epidemic, suggesting that this mutation arose during the development of herd 2 6 4 immunity (30).

6 5
Of the 41 variant alleles that arose de-novo or changed in frequency during the study, several 2 6 6 are of clinical interest (Figure 4, Table 1). A4981G, which arose de-novo and persisted in Beta 2 6 1 4 were then sacrificed with the lung homogenate used to infect another mouse. This was repeated 3 1 8 for ten passages. Second, P10 virus for each lineage was used to infect three additional mice 3 1 9 and the process was repeated for an additional ten passages. Third, P0 and P20 virus for each 3 2 0 lineage was used to infect additional mice to compare weight loss and survival between early 3 2 1 and late passage virus, and test non-adapted mouse susceptibility to the virus. Antibody 3 2 2 neutralization of virus was also compared. 3 2 3    mouse (n = 10/group) was monitored daily throughout the experiment and reported as mean 3 3 8 percentage ± SD of weight relative to day 0. **: p<0.008 as determined using non-parametric 3 3 9 Mann-Whitney test. 3 4 0 They were then sacrificed with the lung homogenate used to infect another mouse. This was repeated for ten passages. Second, P10 virus for each lineage was used to infect three additional mice and the process was repeated for an additional ten passages. Third, P0 and P20 virus for each lineage was used to infect additional mice to compare weight loss and survival between early and late passage virus, and test non-adapted mouse susceptibility to the virus. Antibody neutralization of virus was also compared.      Table 1. Annotations of select variants that changed in frequency across the study. Figure 5. Pathogenesis of P0 and P20 viruses in K18-ACE2 mice. Mice (n = 15/group) were infected with 500 TCID50 of P0 or P20 Beta and Delta viruses. (A) On day 3 post-infection, mice (n = 5) were euthanized and lungs collected for infectious viral load determination. Results are expressed as mean ± SD SARS-CoV-2 TCID50/mg of lung tissue. (B-C) The weight of each mouse (n = 10/group) was monitored daily throughout the experiment and reported as mean percentage ± SD of weight relative to day 0. **: p<0.008 as determined using non-parametric Mann-Whitney test. Figure 6. Growth kinetics of P0 and P20 Delta viruses. P0 and two independently evolved P20 Delta viruses were used to infect Vero cells at a multiplicity of infection of 0.005. Cell-free supernatants were collected at 24 h and 48 h and used for viral titration. Results are expressed as mean ± SD TCID50/mL. *: p=0.02 as determined using an unpaired t-test. Figure 7. Neutralization of P0 and P20 viruses with sera from vaccinated subjects. (A) Neutralization assay of Wuhan-like, Beta, and Delta viruses using sera from three vaccinated subjects. Individual results and mean ± SD neutralization titer against different SARS-CoV-2 isolates are presented. ****: p<0.0001 and **: p=0.0012 determined using one way-ANOVA. ns: not statistically significant. (B) Pair-wise comparison of the sera used in A in neutralization assay against P0 and P20 Beta viruses. (C) Pair-wise comparison of the sera used in A in neutralization assay against P0 and P20 Delta viruses. *: p=0.017 determined using paired ttest.