Ribosomal frameshifting is a carefully tuned modifier of proteins

Abstract

Ribosomal frameshifting, occurring during protein translation, is one of the most consequential recoding events, as it leads to large changes to the protein sequence. However, beyond a few well-characterized cases, its prevalence and consequences across organisms remain poorly understood. To address this, we modeled and experimentally tested frameshifting and systematically characterized its evolutionary impact. We developed SLIPPERRS, a mechanistic model that identifies "slippery" frameshift sites achieving [~]70% accuracy. Building on SLIPPERS predictions and on our mass spectrometry and fluorescence-based assay, we characterized 165 short mRNA sequences. We identified dozens of novel sites with frameshift probabilities that often exceed those of known programmed frameshift cases, and found evidence for both translocation and decoding mechanisms. In Gammaproteobacteria, an average of [~]100 genes per species harbor frameshifting sites, which alter proteins by truncating, extending, and fusing canonical proteins. We detect selection for shorter and less disordered frameshifted proteins to mitigate their deleterious impact. We found negative correlation between GC-content and the occurrence of frameshifting sites. In GC-rich bacteria, where frameshift products are expected to be particularly long and disordered, there is selection for cryptic stop codons. While frameshifting is prevalent and expands protein diversity, it is carefully tuned to avoid its harmful consequences.