Sequence determinants of intron-mediated enhancement learned from thousands of random introns

Spliceosomal introns are a ubiquitous feature of eukaryotic genes, whose presence often boosts the expression of their host gene, a phenomenon known as intron-mediated enhancement (IME). IME has been noted across diverse genes and organisms, but remains mysterious in many respects. For example, how does intron sequence affect the magnitude of IME? In this study, we performed a massively parallel reporter assay (MPRA) to assess the effect of varying intron sequence on gene expression in a high-throughput manner, using tens of thousands of synthetic introns with natural splice sites and randomized internal sequence. We observe that most random introns splice efficiently and enhance gene expression as well as or better than fully natural introns. Nearly all introns stimulate gene expression ~eight-fold above an intronless control, at both mRNA and protein levels, suggesting that the primary mechanism acts to increase mRNA levels. IME strength is positively associated with splicing efficiency and with the intronic content of poly-uridine stretches, which we confirm using reporter experiments. In sum, this work elucidates sequence determinants of IME from tens of thousands of random introns. Overall design: We constructed a plasmid library with two reporter genes, GFP and dTomato, in which the GFP carries a single intron in its 5'UTR, comprising natural splice sites flanking 160nt of random nucleotide sequence. We also included an intron-specific barcode in the downstream exon of GFP as well as in the dTomato in order to normalize GFP expression to dTomato expression from each barcode. We integrated this library into the genome of a specialized HEK293T cell line with a recombinase landing pad (Khandelia et al 2011) to create a stable transgenic pool of cells. From these cells we extracted total RNA and generated an amplicon library of GFP and dTomato 5'UTRs using gene-specific RT-PCR, which we sequence to measure steady-state mRNA abundance and splicing status, in ten biological replicates. We also measure effects on protein expression via GFP and dTomato fluorescence intensity by iteratively selecting cells with the most extreme fluorescence ratios and performing the same RNA-seq protocol on these subpopulations to identify which introns they contain. This was done for the top 10% most green and most red cells, in three successive stages, in two replicate trajectories. Finally, we synthesized and tested a second smaller library with mutated versions of introns from the original library to validate the effect of polyU content on intron-mediated enhancement.