U.S. Department of Agriculture: Animal and Plant Health Inspection Service


Date of this Version



Infection, Genetics and Evolution 90 (2021) 104505



U.S. gov't work


Isolation and cultivation of wild-type viruses in model organism cells or tissues is standard practice in virology. Oftentimes, the virus host species is distantly related to the species from which the culture system was developed. Thus, virus culture in these tissues and cells basically constitutes a host jump, which can lead to genomic changes through genetic drift and/or adaptation to the culture system. We directly sequenced 70 avian influenza virus (Orthomyxoviridae) genomes from oropharyngeal/cloacal swabs collected from wild bird species and paired virus isolates propagated from the same samples following isolation in specific-pathogen-free embryonated chicken eggs. The data were analyzed using population genetic approaches including evaluation of single nucleotide polymorphism (SNP) frequencies and divergence with pooled-sequencing analyses, consensus sequence placement in neighbor-joining trees, and haplotype reconstruction and networks. We found that propagation of virus in eggs leads to skewed SNP mutation spectra with some SNPs going to fixation. Both synonymous and nonsynonmous SNP frequencies shifted. We found multiple consensus sequences that differed between the swabs and the isolates, with some sequences from the same sample falling into divergent genetic clusters. Twenty of 23 coinfections detected had different dominant subtypes following virus isolation, thus sequences from both the swab and isolate were needed to obtain full subtype data. Haplotype networks revealed haplotype frequency shifts and the appearance or loss of low-frequency haplotypes following isolation. The results from this study revealed that isolation of wild bird avian influenza viruses in chicken eggs leads to skewed populations that are different than the input populations. Consensus sequence changes from virus isolation can lead to flawed phylogenetic inferences, and subtype detection is biased. These results suggest that for genomic studies of wild bird influenza viruses the biological field should move away from chicken egg isolation towards directly sequencing the virus from host samples.