Cells intricately tune gene expression to maintain homeostasis. Translation initiation is an important regulatory step in gene expression, whereby the concerted actions of proteins prime an mRNA to recruit a ribosome for protein synthesis. The 5′ untranslated region (UTR) of mRNAs has a central role in translation initiation: for example, the 5′ cap promotes initiation, whereas upstream open reading frames suppress initiation, and terminal oligopyrimidine sequences tune initiation in response to different cell states. However, such canonical features cannot explain the variation in 5′ UTR initiation activity among different mRNAs, suggesting undiscovered regulatory elements across the transcriptome.

With DART, we observed hundred- to thousand-fold differences in initiation activity across yeast and human 5′ UTRs and identified the regulatory features responsible. For instance, yeast 5′ UTRs containing eukaryotic initiation factor 4G (eIF4G) binding motifs drive ribosome recruitment. Human UTRs containing C-rich motifs across the RNA are repressive, whereas cap-adjacent AU-rich elements enhance translation initiation. Counterintuitively, we found that the coding sequence also has an important role in ribosome recruitment, expanding the potential regulatory landscape of initiation. Moreover, because tested RNAs are synthesized in vitro, non-physiological RNA features can be tested with DART. N1-methylpseudouridine (m1Ψ), a chemical modification used in mRNA vaccines, was incorporated into RNAs and tested using DART, where it generally enhanced initiation.