Stop-codon read-through arises largely from molecular errors and is generally nonadaptive

Stop-codon read-through refers to the phenomenon that a ribosome goes past the stop codon and continues translating into the otherwise untranslated region (UTR) of a transcript. Recent ribosome-profiling experiments in eukaryotes uncovered widespread stop-codon read-through that also varies among tissues, prompting the adaptive hypothesis that stop-codon read-through is an important, regulated mechanism for generating proteome diversity. Here we propose and test a competing hypothesis that stop-codon read-through arises mostly from molecular errors and is largely nonadaptive. The error hypothesis makes distinct predictions about the probability of read-through, frequency of sequence motifs for read-through, and conservation of the read-through region, each of which is supported by genome-scale data from yeasts and fruit flies. Thus, except for the few cases with demonstrated functions, stop-codon read-through is generally nonadaptive. This finding, along with other molecular errors recently quantified, reveals a much less precise or orderly cellular life than is commonly thought. Author summary The stop codon gives the translating ribosome the signal for the termination of peptide synthesis, but occasionally the ribosome goes past the stop codon and continues translating into the otherwise untranslated region of a transcript. Stop-codon read-through generates an elongated peptide, which could be beneficial under certain circumstances. Although stop-codon read-through was thought to be rare, recent ribosome-profiling experiments in eukaryotes discovered hundreds of genes that undergo stop-codon read-through at a detectable rate. It is unclear whether most of these observed read-through events have biological functions or reflect cellular errors. The error hypothesis makes a set of distinct predictions about the probability of read-through, frequency of sequence motifs for read-through, and conservation of the read-through region. Our analysis of genome-scale data from yeasts and fruit flies verifies each of these predictions, suggesting that most stop-codon read-through events are nonadaptive cellular errors. These and related findings of various molecular errors in transcription and posttranscriptional modification paint a much less precise or orderly cellular life than is commonly portrayed.