Translation accuracy in E. coli

Translation is a fundamental process of life, yet methods to systematically investigate its fidelity have been limited. Most previous estimates of translation-error rates have relied on reporter assays that evaluate only a single codon and fail to capture the full spectrum of translation errors. Here, we present a proteome-wide analysis of mass spectrometry data that directly estimates nearly all pairwise amino-acid substitution rates, revealing mistranslation rates and spectra per amino acid and per codon. Applying this method to ribosomal variants of Escherichia coli reported to differ in translation fidelity, we found no significant differences among their overall error rates, estimated here at 2 per 1000 amino acids. Instead, each variant exhibited unique mistranslation profiles; the putative error-prone variant preferentially misread near-cognate codons at the third position, with a bias that likely led to prior overestimates of its error rate. We also tested the translational-accuracy hypothesis of codon usage, which predicts that codons enriched in highly expressed genes are selected for translational accuracy. Contrary to that prediction, codons favored in highly expressed genes are not translated more accurately. These results underscore the necessity of proteome-wide measures of translation accuracy and highlight the limitations of single-codon approaches for characterizing translation fidelity.