Detecting intragenic trans-splicing events with hybrid transcriptome sequencing in cancer cells

Trans-splicing can generate non-co-linear (NCL) transcripts that consist of exons in an order topologically inconsistent with the corresponding DNA template. Detecting trans-spliced RNAs (ts-RNAs) may be interfered by false positives from experimental artifacts, circular RNAs (circRNAs), and genetic rearrangements. Particularly, intragenic ts-RNAs, which are derived from separate precursor mRNA molecules of the same genes, are often mistaken for circRNAs through analyses of high-throughput transcriptome sequencing (RNA-seq) data. In addition, the biogenesis and function of ts-RNAs remain elusive. Here we developed a bioinformatics pipeline, NCLscan-hybrid, with the integration of long and short RNA-seq reads to minimize false positives and identify intragenic ts-RNAs. We utilized two features of long reads, out-of-circle and rolling circle, to distinguish intragenic ts-RNAs from circRNAs. We also designed multiple experimental validation steps to examine each type of false positives and successfully confirmed an intragenic ts-RNA (ts-ARFGEF1) in breast cancer cells. On the basis of ectopic expression and CRISPR-based endogenous genome modification experiments, we confirmed that ts-ARFGEF1 formation was significantly dependent on the reverse complementary sequences in the flanking introns of the NCL junction. Subsequent in vitro and in vivo experiments demonstrated that ts-ARFGEF1 silencing can significantly inhibit tumor cell growth. We further showed the regulatory role of ts-ARFGEF1 in p53-mediated apoptosis through affecting the PERK/eIF2a/ATF4/CHOP signaling pathway in breast cancer cells. This study thus described both bioinformatics procedures and experimental validation steps for rigorous characterization of transcriptionally non-co-linear RNAs, expanding the discovery of this important but understudied class of RNAs.