The Importance of Nonsense Errors: Estimating the Rate and Implications of Drop-Off Errors during Protein Synthesis

The process of mRNA translation is both energetically costly and relatively error-prone compared to transcription and replication. Nonsense errors during mRNA translation occur when a ribosome drops off a transcript before reaching a stop codon, resulting in energetic investment in an incomplete and likely non-functional protein. Nonsense errors impose a potentially significant energy burden on the cell, making it critical to quantify their frequency and energetic cost. Here, we present a model of ribosome movement for estimating protein production, elongation, and nonsense error rates from high-throughput ribosome profiling data. Applying this model to an exemplary ribosome profiling dataset in S. cerevisiae, we find that nonsense error rates vary between codons, in conflict with the general assumption of uniform rates across sense codons. Using our parameter estimates, we find multiple lines of evidence that selection against nonsense errors is a prominent force shaping coding-sequence evolution, including that nonsense errors place an energetic burden on cells comparable to ribosome pausing. Our results indicate greater consideration should be given to the impact of nonsense errors in shaping coding-sequence evolution. Author SummaryThe process of mRNA translation is both energetically expensive and relatively error-prone. As such, natural selection is thought to shape the evolution of CDSs to reduce the cost of these errors when they occur. Nonsense errors (NSEs) occur when a ribosome stops translation prior to completing a functional protein, resulting in wasted energy on non-functional product. Despite their functional consequences, NSEs and their effects on sequence evolution are generally understudied compared to other types of translation errors. This is in part due to the challenge of quantifying these errors from omics-scale data. We present a model for quantifying codon-specific estimates of elongation and NSE rates from ribosome profiling data, which gives a snapshot of the actively translating ribosomes in a cell. Although it is well-established that sense codons vary in their elongation rates, we find evidence that codons also vary in their NSE rates. Using our parameter estimates, we find multiple lines of evidence for selection against NSEs shaping patterns of codon usage bias. Our results suggest the cost of NSEs are comparable to the cost of ribosome pausing, and thus may play a greater role in coding-sequence evolution than previously appreciated.