Entire genome transcription across evolutionary time exposes non-coding DNA tode novogene emergence

Even in the best studied Mammalian genomes, less than 5% of the total genome length is annotated as exonic. However, deep sequencing analysis in humans has shown that around 40% of the genome may be covered by poly-adenylated non-coding transcripts occurring at low levels 1 . Their functional significance is unclear 2,3 , and there has been a dispute whether they should be considered as noise of the transcriptional machinery 4,5 . We propose that if such transcripts show some evolutionary stability they will serve as substrates for de novo gene evolution, i.e. gene emergence out of non-coding DNA 6–8 . Here, we characterize the phylogenetic turnover of low-level poly-adenylated transcripts in a comprehensive sampling of populations, sub-species and species of the genus Mus , spanning a phylogenetic distance of about 10 Myr. We find evidence for more evolutionary stable gains of transcription than losses among closely related taxa, balanced by a loss of older transcripts across the whole phylogeny. We show that adding taxa increases the genomic transcript coverage and that no major transcript-free islands exist over time. This suggests that the entire genome can be transcribed into polyadenylated RNA when viewed at an evolutionary time scale. Thus, any part of the “non-coding” genome can become subject to evolutionary functionalization via de novo gene evolution.