Widespread occurrence of hybrid internal-terminal exons in human transcriptomes

Abstract

Alternative RNA processing is a major mechanism for diversifying the human transcriptome. Messenger RNA isoform differences are predominantly driven by alternative first exons, cassette internal exons and alternative last exons. Despite the importance of classifying exons to understand isoform structure, there is a lack of tools to look at isoform-specific exon usage using RNA-sequencing data. We recently observed that alternative transcription start sites often arise near annotated internal exons, creating "hybrid" exons that can be used as both first or internal exons. To investigate the creation of hybrid exons, we built the HIT (Hybrid-Internal-Terminal) exon pipeline that systematically classifies exons depending on their isoform-specific usage. Using a combination of junction reads coverage and probabilistic modeling, the HIT index identified thousands of hybrid first-internal and internal-last exons that were previously misclassified. Hybrid exons are enriched in long genes with at least ten internal exons, have longer flanking introns and strong splice sites. The usage of hybrid exons varies considerably across human tissues, but they are predominantly used in brain, testis and colon cells. Notably, genes involved in RNA splicing have the highest fraction of intra-tissue hybrid exons. Further, we found more than 100,000 inter-tissue hybrid exons that changed from internal to terminal exons across tissues. By developing the first method that can classify exons according to their isoform contexts, our findings demonstrate the existence of hybrid exons, expand the repertoire of tissue-specific terminal exons and uncover unexpected complexities of the human transcriptome.