Abstract
The introduction of a single, non-optimal codon cluster into a transcript was found to cause ribosomes to queue upstream of it and decrease the transcript’s half-life through the action of the dead box protein Dhh1p, which interacts with stalled ribosomes and promotes transcript degradation. Naturally occurring transcripts often contain multiple slow codon clusters, but the influence of these combinations of clusters on a transcript’s half-life is unknown. Using a kinetic model that describes translation and mRNA degradation, we find that the introduction of a second slow codon cluster into a transcript near the 5′-end can have the opposite effect than that of the first cluster by increasing the mRNA’s half-life. We experimentally validate this finding by showing that S. cerevisiae transcripts that only have a slow codon cluster towards the 3′-end of the coding sequence have shorter half-lives than transcripts with non-optimal clusters at both 3′ and 5′ -ends. The origin of this increase in half-life arises from the 5′-end cluster suppressing the formation of ribosome queues upstream of the second cluster, thereby decreasing the opportunities for translation-dependent degradation machinery, such as the Dhh1p protein, to interact with stalled ribosomes. We also find in the model that in the presence of two slow codon clusters the cluster closest to 5′-end is the primary determinant of mRNA half-life. These results identify two of the rules governing the influence of slow codon clusters on mRNA half-life, demonstrate that multiple slow codon clusters can have synergistic effects, and indicate that codon usage bias can play a more nuanced role in controlling cellular protein levels than previously thought.
