Cis-element acquisition converts pervasive transcripts into translated mRNAs

The mammalian genome is pervasively transcribed into thousands of unstable non-coding RNAs, including promoter-upstream transcripts (PROMPTs) and enhancer RNAs (eRNAs). These pervasive transcripts are degraded by the nuclear exosome within minutes of synthesis and are never translated. Coding and non-coding transcripts differ systematically in their content of splicing and polyadenylation signals, yet whether these cis-elements are sufficient to redirect a pervasive transcript toward an mRNA-like fate has not been tested at endogenous loci. Here, using CRISPR-Cas9 knock-in at three endogenous non-coding loci pervasively transcribed in murine cancer cells, we show that appending a splicing signal together with a polyadenylation signal is sufficient to render the resident transcript stable, polyadenylated, exported, and translated into full-length protein. Through quantitative analysis of isogenic single-cell clones, we find that the polyadenylation signal acts principally on transcript stability, extending half-life from ∼20 minutes to ∼12 hours, whereas splicing acts principally on nuclear export, with splicing efficiency quantitatively predicting both export and protein output. Remarkably, the untranslated regions of naturally intronless genes substitute for splicing, conferring coding capacity predominantly by enhancing export rather than stability. A genome-wide CRISPR screen on the intronless-gene reporter nominates nuclear export, 3′-end processing, and translation factors including NXF1, PABPN1, and EIF3G as required for conversion. Our results demonstrate that the acquisition of defined cis-elements is sufficient to convert pervasive transcripts into mRNAs, with nuclear export as the convergent rate-limiting step, providing an experimental reconstitution of a route by which non-coding loci may be recruited into the coding genome.