Limited SARS-CoV-2 diversity within hosts and following passage in cell culture

Gage Kahl Moreno; Katarina M Braun; Peter J Halfmann; Trent M Prall; Kasen K Riemersma; Amelia K Haj; Joseph Lalli; Kelsey R Florek; Thomas C Friedrich; Yoshihiro Kawaoka; David H O'Connor

To characterize patterns of sub-consensus diversity, we looked at SNVs at or above 1% 129 frequency in only the Illumina reads. We previously established that this conservative cutoff 130 ensures that only bona fide mutations are considered [15, 16] . All minor variant analyses and 131 figures were completed using the Illumina SNV data as these data are higher average quality 132 and ideal for analysis involving low-frequency variants (Fig 3) . Seventy-five percent of all minor 133 variants we identify fall in ORF1a and ORF1b, which together take up 72.8% of the length of the 134 28kb coding genome. ORF1a and ORF1b encode the replicase machinery [7] . We account for 135 differences in gene size by normalizing variants to kilobase gene length (variants / kb-gene-136 length -"v/kbgl") [10] . The highest density of variants was reported in smaller genes like 137 envelope, ORF7a, and ORF10 (S2 Table) . We also show that through each passage, variant 138 density in ORF1a and ORF1b increases. There were no SNVs ≥ 1% in the spike gene in the 139 primary NP swab, but low-frequency SNVs (all <5%) were identified in spike following passage 140 in cell culture (Fig 3) . Outside of ORF1a and ORF1b, the other genes in the primary NP swab 141 . CC-BY 4.0 International license author/funder. It is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.20.051011 doi: bioRxiv preprint are clonal above the 1% threshold, with the exception of one low-frequency SNV In addition to assessing the fate of individual minor variants, we were also interested in 188 evaluating population dynamics using diversity metrics. Specifically, we calculated genewise 189 diversity using π , the average number of pairwise differences per nucleotide site among a set of 190 . CC-BY 4.0 International license author/funder. It is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.20.051011 doi: bioRxiv preprint sequences, for each gene in each sample. Overall, genewise nucleotide diversity is very low 191 compared to other RNA viruses, consistent with low mutation rates in coronaviruses due to RNA 192 proofreading machinery [18, 19] . Genewise diversity was very low in the primary NP swab and 193 was only measurable in ORF1a (9 SNVs), ORF1b (5 SNVs) and N (1 SNV). Genewise diversity 194 is more varied in the passaged samples (Fig 5) . Interestingly, π is highest in ORF7a in these 195 samples -although this signal seems to be primarily driven by the small size of this gene. To Vero 76 ORF10, p1 Vero E6 envelope (E), and p1 Vero STAT-1 KO ORF3a. 205