Viral sequence features of influenza A viruses and coronaviruses

Abstract

Viral genomes are shaped by different selection forces in evolution, and the resulting synonymous mutations and non-synonymous mutations can gradually accumulate and lead to changes in the phenotype. Although synonymous mutations in the nucleic acid sequence do not change the amino acid sequence, they can have significant effects on viral fitness. To date, several viral sequence features that can be caused by synonymous mutations have been identified, such as codon usage, dinucleotide usage and RNA secondary structures. The origins and outcomes of these sequence features are largely unexplored due to a lack of appropriate experimental settings that can control most of the confounding factors. The objective of this doctoral work is to identify patterns of different sequence features in influenza A viruses (IAV) and coronaviruses, and to develop and apply bioinformatic tools for guiding further experimental studies on sequence features. Results showed that dinucleotide preference, codon usage bias and direct RNA-RNA interaction were identified or predicted in viral genomes, and viral fitness was changed after introducing synonymous mutations into corresponding genomic regions.

Codon usage bias of different coronaviruses was studied by different types of correspondence analyses. Different patterns of codon usage between different virus genera were observed. While the novel SARS-CoV-2 virus was found to be similar to bat and human SARSr-CoVs at the amino acid usage level, its synonymous codon usage was most similar to bat RaTG13 virus and pangolin P1E virus. Besides codon usage bias, dinucleotide usage preference was also demonstrated to be significant in viral genomes. Dinucleotides CpG, UpA, UpG and CpA were found to be significantly under-represented or over-represented in the genomes of IAV, and evidence for dinucleotide preference directly resulting in synonymous codon usage bias was observed. These results revealed the interacting nature between codon usage and dinucleotide usage, and opened up new areas for studying mutations in viruses while controlling both codon usage and dinucleotide usage.

An R package SynMut was developed for directional reprogramming of viral genomes with specific codon usage patterns and extreme dinucleotide usage, while not affecting the original amino acid sequence or known important genomic signals. The mutant viruses designed by this tool aided in studying different effects between codon usage and dinucleotide usage to viral fitness. Another bioinformatic tool was developed to design mutant viruses for studying direct RNA-RNA interactions between viral segments in IAV. This tool minimizes the complementary base pairing in predicted RNA-RNA interacting regions by introducing synonymous mutations. Some of the mutated viruses were significantly attenuated in different cells. These results constituted important evidence on critical synonymous mutations in IAV.

Bioinformatic analysis on codon usage and dinucleotide usage in coronaviruses and influenza A viruses was conducted in this thesis. Computational methods and tools for designing mutant viruses for studying codon usage, dinucleotide usage and RNA-RNA interactions were developed. The discovery of critical synonymous mutations by experiments opens up potential future research directions on synonymous mutations in viral genomes.