Fig. 4

Promoter and enhancer signature is highly dynamic across species, but not within species. a Within a species, dynamic P/Es (red) were regions identified as an active promoter in one tissue and an enhancer in another tissue, and dynamic Es (blue) were an active enhancer in one tissue and primed in another. Numbers for each category are summed across all ten species. Within a species and across tissues, only 4% of the regulome is composed of intra-species dynamic regions. b Pairwise comparisons between maintained regulatory regions show how often regulatory signature changes between species. A substantial proportion of regulatory regions align to a region with a different regulatory signature in another species: 20% of pairwise comparisons with active promoters, 58% with active enhancers and 40% with primed enhancers are evolutionarily dynamic. Almost half of active enhancers (44%) aligned to primed enhancers. Dynamic P/Es (red) and dynamic Es (blue) almost always align to non-dynamic categories in other species (73% and 88% respectively), illustrating the evolutionary instability of this regulatory assignment. An enlargement of the intra-species dynamic regions is shown on the right for clarity. For changes of regulatory changes within the same tissue between species see Additional file 1: Figure S8B. c Evolutionary rates of changing regulatory signatures among maintained regulatory regions estimated by linear regression of pairwise comparisons. Across evolution, maintained active promoters (crosses) and active enhancers (diamonds) were more likely to change regulatory signature as evolutionary distance between species increased. d Evolutionary directionality of dynamic regulatory signatures estimated by outgroup analysis of mouse/rat/rabbit and cat/dog/horse triads. Gray inset example: in 448 cases when a genomic region is an active promoter in one ingroup species and an active enhancer in the other, the outgroup species was most likely to be an active enhancer (46%), and least likely to be an active promoter (20%). The distributions of outgroup active promoters, active enhancers and primed enhancers for each ingroup combination is statistically different from the background (All) distribution (chi-square two-tailed test, ***p < 0.001). Outgroup analysis was performed separately for each triad group, and then combined (see Additional file 1: Figure S8C and D). e Composite model based on the observed likelihood of regulatory regions changing or maintaining regulatory signatures over evolution. The thickness of the lines reflects the relative likelihood of evolutionary change, as calculated from the most parsimonious evolutionary relationships from the triad data in d and normalizing the outgoing lines from each state to one. f Validation of regulatory signature assignment using the average ChIP-seq read enrichment for evolutionarily dynamic regulatory regions and equal numbers of randomly selected control regions. Dynamic regions were the AP/AE ingroup regions identified as AE in the outgroup analysis in d. Total number of regions used are shown as insets. g Distribution of RNA-seq read counts for evolutionarily dynamic active promoters (AP) and active enhancers (AE) shown in f, and equal numbers of randomly selected control active enhancers that are not evolutionarily dynamic (p values calculated using one sided the t-test for greater expression). h Tissue distribution of evolutionarily dynamic P/Es in the species where they were an active promoter, active enhancer or primed enhancer. When showing signatures of active promoters (left; purple), they were less likely to be tissue-shared and more likely to be testis-specific than all promoters (compared to Fig. 2b). When showing enhancer signatures, they were more likely to be tissue-shared than all enhancers (Fig. 2b bottom orange and green)