Unraveling the influences of sequence and position on yeast uORF activity using massively parallel reporter systems and machine learning

Abstract

Upstream open reading frames (uORFs) are potent cis-acting regulators of mRNA translation and nonsense-mediated decay (NMD). While both AUG- and non-AUG initiated uORFs are ubiquitous in ribosome profiling studies, few uORFs have been experimentally tested. Consequently, the relative influences of sequence, structural, and positional features on uORF activity have not been determined. We quantified thousands of yeast uORFs using massively parallel reporter assays in wildtype and ∆upf1 yeast. While nearly all AUG uORFs were robust repressors, most non-AUG uORFs had relatively weak impacts on expression. Machine learning regression modeling revealed that uORF functions are strongly impacted by both their sequences and locations within transcript leaders. Indeed, alternative transcription start sites highly influenced uORF activity. These results define the scope of natural uORF activity, identify features associated with translational repression and NMD, and suggest that the locations of uORFs in transcript leaders are nearly as important as uORF sequences.

Yeast eIF2A has a minimal role in translation initiation and uORF-mediated translational controlin vivo

“…It was reported that among the 407 mRNAs containing a single functional AUG-uORF wherein mutating the uORF significantly altered reporter expression by > 1.5-fold, all but 13 of the uORFs functioned to inhibit reporter expression. Among the 144 mRNAs containing a single functional NCC-uORF, only 13 of the uORFs influenced reporter expression either positively or negatively by > 1.5-fold, reflecting the weaker effects of NCC-versus AUG-initiated uORFs on downstream translation (May et al 2023). Our analysis of the 394 mRNAs containing a single inhibitory AUG-uORF—the only group large enough for statistical analysis of TE changes—revealed little or no change in median TE in response to SM treatment or to the eIF2AΔ mutation in the presence or absence of SM (Figure 6A(iii), cols.…”

“…uORFs with mean RPF counts below two or mean CDS RPF counts below 32 in the combined samples (two replicates each) were excluded from the analysis. The compilations of annotated translated uORFs (Martin-Marcos et al 2017), evolutionarily conserved translated uORFs (Spealman et al 2018), or functional uORFs whose elimination by start codon mutation conferred a consistent and statistically significant alteration in translation of the downstream GFP CDSs in reporter constructs (May et al 2023), were all published previously and are provided in Figure 6-source data 5-7.…”

“…Figure 6-source data 7.Spreadsheet tabulates the functional uORFs , chromosome coordinates, start codon of uORF, distances of the uORF AUG from the 5' end of the mRNA and the main CDS start codon, and the gene name(May et al 2023). …”

Current limitations in predicting mRNA translation with deep learning models

“…Previous studies reached different conclusions concerning the impact of IF uAUGs on translation [15,20,27,28]. To clarify this, we determined the relationship between the location of OOF and IF uAUGs in the 5'UTR and the translation output of the mRNAs, in both yeast and human sequences, synthetic or endogenous.…”

NaP-TRAP, a novel massively parallel reporter assay to quantify translation control

“…These assays have improved our understanding of translation control and mRNA stabil ity. Previous works have characterized Kozak strength, 23 uORFs, [24][25][26] IRESs, 27,28 codon optimality, 29,30 RNA struc ture, 31 microRNA binding sites, 32 and the effects of vari ation in human UTRs. [33][34][35][36] These assays have also iden tified novel sequence motifs driving translation and de cay.…”