Abstract
Synonymous codon usage (codon bias) greatly influences not only translation but also mRNA stability. In vertebrates, highly expressed genes preferentially use codons with an optimal tRNA adaptation index (tAI) that mostly end in C or G. Surprisingly, the codon usage of viruses infecting humans often deviates from optimality, showing an enrichment in A/U-ending codons, which are generally associated with slow decoding and reduced mRNA stability. This observation is particularly evident in RNA viruses causing respiratory illnesses in humans. This review analyzes the mutational and selective forces that shape nucleotide composition and codon usage drift in human RNA viruses, as well as their impact on translation, viral fitness, and evolution. It also describes how some viruses overcome suboptimal codon usage to outcompete host mRNA for translation. Finally, the roles of viral tropism and host adaptation in codon usage bias of prototypical viruses are discussed.