Rational Design Of RNA Structures That Predictably Tune Eukaryotic Gene Expression

Abstract

Predictable tuning of gene expression is essential for engineering genetic circuits and for optimising enzyme levels in metabolic engineering projects. In bacteria, gene expression can be tuned at the stage of transcription, by exchanging the promoter, or at stage of translation by altering the ribosome binding site sequence. In eukaryotes, however, only promoter exchange is regularly used, as the tools to modulate translation are lacking. Working in S. cerevisiae yeast, we here describe how hairpin RNA structures inserted into the 5 untranslated region (5UTR) of mRNAs can be used to tune expression levels by altering the efficiency of translation initiation. We demonstrate a direct link between the calculated free energy of folding in the 5UTR and protein abundance, and show that this enables rational design of hairpin libraries that give predicted expression outputs. Our approach is modular, working with different promoters and protein coding sequences, and it outperforms promoter mutation as a way to predictably generate a library where a protein is induced to express at a range of different levels. With this tool, computational RNA sequence design can be used to predictably fine-tune protein production, providing a new way to modulate gene expression in eukaryotes.