Intergenic RNA mainly derives from nascent transcripts of known genes

BackgroundEukaryotic genomes undergo pervasive transcription, leading to the production of many types of stable and unstable RNAs. Transcription is not restricted to regions with annotated gene features but includes almost any genomic context. Currently, the source and function of most RNAs originating from intergenic regions in the human genome remains unclear. ResultsWe hypothesised that many intergenic RNA can be ascribed to the presence of as-yet unannotated genes or the fuzzy transcription of known genes that extends beyond the annotated boundaries. To elucidate the contributions of these two sources, we assembled a dataset of >2.5 billion publicly available RNA-seq reads across 5 human cell lines and multiple cellular compartments to annotate transcriptional units in the human genome. About 80% of transcripts from unannotated intergenic regions can be attributed to the fuzzy transcription of existing genes; the remaining transcripts originate mainly from putative long non-coding RNA loci that are rarely spliced. We validated the transcriptional activity of these intergenic RNA using independent measurements, including transcriptional start sites, chromatin signatures, and genomic occupancies of RNA polymerase II in various phosphorylation states. We also analysed the nuclear localisation and sensitivities of intergenic transcripts to nucleases to illustrate that they tend to be rapidly degraded either on-chromatin by XRN2 or off-chromatin by the exosome. ConclusionsWe provide a curated atlas of intergenic RNAs that distinguishes between alternative processing of well annotated genes from independent transcriptional units based on the combined analysis of chromatin signatures, nuclear RNA localisation and degradation pathways.