Preferred synonymous codons are translated more accurately: Proteomic evidence, among-species variation, and mechanistic basis

Abstract

A commonly stated cause of the widespread phenomenon of unequal uses of synonymous codons is their differential translational accuracies. However, this long-standing translational accuracy hypothesis (TAH) of codon usage bias has had no direct evidence beyond anecdotes. Analyzing proteomic data from Escherichia coli, we observe higher translational accuracies of more frequently used synonymous codons, offering direct, global evidence for the TAH. The experimentally measured codon-specific translational accuracies validate a sequence-based proxy; this proxy provides support for the TAH from the vast majority of over 1000 taxa surveyed in all domains of life. We find that the relative translational accuracies of synonymous codons vary substantially among taxa and are strongly correlated with the amounts of cognate tRNAs relative to those of near-cognate tRNAs. These and other observations suggest a model in which selections for translational efficiency and accuracy drive codon usage bias and its coevolution with the tRNA pool.