Characterization of the human transcriptome diversity and its implication in identifying cancer-specific regions of transcripts

Abstract

The emergence of exons and introns in eukaryotes has increased the potential of gene products and their functional diversity to a greater extent than in prokaryotes. In humans, the transcriptome has the highest percentage of multi-exon genes that contain at least one alternatively spliced exon. At the exon level, however, it is the constitutive exons that have been observed more often in gene transcripts. This trend has been shifting with the increasing numbers of alternative exons per gene. Here, we estimate the differentially used exons (DEXs) of human RNA transcripts at a saturation level among 1,540 pairs of 56 nominally normal tissues with DEXs comprising at least 72% of the human exome. Our result has shifted the paradigm of exon usage from mainly constitutive to differentially used exons in the human transcriptome, suggesting a more diverse and dynamic transcript repertoire of the human transcriptome than previously reported. We used this comprehensive catalogue of DEXs of the normal transcriptome as background for further analysis against the cancer transcriptomes. We demonstrated that 99% of the aberrantly processed exons of adenocarcinoma also exist in the normal transcriptome blueprint of DEXs showing similar processing pattern as normal in other non-cancerous tissues. A small group of authentic ‘cancer-specific’ exons that were found only in cancer patients harboring specific gene mutations, were not found in any of the normal tissues. Thus, our results also reveal the ectopic expression of nearly all ‘cancer-specific’ exons of adenocarcinoma compared to the corresponding normal tissue of origin. This will help selection for authentic ‘cancer-specific’ target exons for diagnostic or therapeutic purposes.