Functional characterization of enhancer evolution in the primate lineage

Abstract

Background: Enhancers play an important role in morphological evolution and speciation by controlling the spatiotemporal expression of genes. Previous efforts to understand the evolution of enhancers in primates have typically studied many enhancers at low resolution, or single enhancers at high resolution. Although comparative genomic studies reveal large-scale turnover of enhancers, a specific understanding of the molecular steps by which mammalian or primate enhancers evolve remains elusive.

Results: We identified candidate hominoid-specific liver enhancers from H3K27ac ChIP-seq data. After locating orthologs in 11 primates spanning around 40 million years, we synthesized all orthologs as well as computational reconstructions of 9 ancestral sequences for 348 active tiles of 233 putative enhancers. We concurrently tested all sequences for regulatory activity with STARR-seq in HepG2 cells. We observe groups of enhancer tiles with coherent trajectories, most of which can be potentially explained by a single gain or loss-of-activity event per tile. We quantify the correlation between the number of mutations along a branch and the magnitude of change in functional activity. Finally, we identify 84 mutations that correlate with functional changes; these are enriched for cytosine deamination events within CpGs.

Conclusions: We characterized the evolutionary-functional trajectories of hundreds of liver enhancers throughout the primate phylogeny. We observe subsets of regulatory sequences that appear to have gained or lost activity. We use these data to quantify the relationship between sequence and functional divergence, and to identify CpG deamination as a potentially important force in driving changes in enhancer activity during primate evolution.

Fig. 1

Schematic of Experimental Design. a We identified potential hominoid-specific enhancers by intersecting hominoid-specific ChIP-seq predicted enhancers from primary human liver with ChromHMM-predicted strong enhancers in HepG2 cells (screenshot from http://genome.ucsc.edu) [54]. We then tiled across each candidate enhancer using 194 nt sequences and identified 697 tiles that were active in the STARR-seq reporter assay in HepG2 cells. b We located orthologous sequences in 11 primates and computationally reconstructed 9 ancestral sequences for 348 of the active tiles, using New World monkeys as an outgroup. c We then cloned all 20 present-day or ancestral orthologs per tile and performed STARR-seq again in HepG2 cells. After collecting DNA and RNA from cells, we calculated enrichment scores as the log₂ ratio of RNA to DNA for each ortholog. Each shade of red represents a different ortholog tested

Fig. 2

Performance of Computational Predictions. a We trained the gapped-kmer support vector machine classifier (gkm-SVM) on an independent reporter assay experiment conducted in HepG2 cells. We then predicted the functional activity of all of our human sequence tiles and found a modest correlation with our functional data. b The distributions of differences in predicted gkm-SVM score between the human vs. marmoset, vervet, or rhesus ortholog for all active human tiles. c Predicted scores for all orthologs of the 348 human-active enhancer tiles, normalized to the human ortholog. Clades are denoted by colored lines (green: hominoid, orange: Old World monkeys, purple: New World monkeys). Cyan bar below dendrogram denotes a group of 108 enhancer tiles that follows expectations for hominoid-specific enhancers as predicted by ChIP-seq comparative genomics

Fig. 3

Functional Scores for Orthologs and Ancestral Sequences. a The average pairwise Spearman correlation of functional scores between any two orthologs across all enhancer tiles tested. b Correlation between the STARR-seq enrichment scores, normalized to the enrichment score of its human ortholog (log₂[non-human score/human score]), and gkm-SVM predicted scores, similarly normalized to the predicted score of its human ortholog (log₂[non-human prediction/human prediction]). c Functional scores normalized to human for all orthologs of the 220 enhancer tiles. Black bars represent missing data. Clades are denoted by vertical colored lines (green: hominoid, orange: Old World monkeys, purple: New World monkeys). Groups are denoted by horizontal colored lines below the dendrogram; gray: relatively higher activity in in Old World monkeys, orange: relatively lower in Old World monkeys, green: relatively higher activity in either humans or hominoids

Fig. 4

Common Patterns of Enhancer Modulation over the Primate Phylogeny. a Functional scores for all enhancer tiles normalized to the MRCA of hominoids and Old World monkeys (N2). Black bar graph in the center contains the N2 score for each tile. Color bars above the heatmap indicate subsets of enhancer tiles exhibiting coherent patterns with respect to gain/loss of activity across the primate phylogeny, including: increased in NWM (yellow), increased in hominoid (gray), decreased in NWM (green), decreased in hominids (orange), and decreased in OWM (purple). b The average score normalized to N2 for each species across the group of 27 enhancer tiles with increased activity restricted to the outgroup of New World monkeys. Gray “+” / “-” indicates that there could be either a gain-of-activity event at the “+” or a loss-of-activity event at the “-”. Error bars are one standard error. c Same as (b) for a group of 22 enhancer tiles with increased activity within the hominoid clade. Red “+” indicates a gain-of-activity event. d Same as (b) for a group of 29 enhancer tiles with decreased activity restricted to the outgroup of New World monkeys. e Same as (b) for a group of 22 enhancer tiles with decreased activity in hominids. Blue “-” indicates a loss-of-activity event. f Same as (b) for a group of 36 enhancer tiles with decreased activity in OWM

Fig. 5

Molecular Characterization of Enhancer Modulation. a For every branch along the tree, we calculated the nucleotide and functional divergence. The number of nucleotide changes is on the x-axis and the absolute value of the difference in the logged functional activity between the daughter and ancestral node is on the y-axis. b The fraction of indels, A → C and T → G mutations, A → G and T → C mutations, A → T and T → A mutations, C → A and G → T mutations, C → G and G → C mutations, and C → T and G → A mutations in our set of 84 prioritized mutations (those associated with a significant functional difference) in black and 2537 background mutations (those associated with a non-significant functional difference) in gray. Asterisk represents a p-value < 0.05 (Fisher’s exact test). First seven tests use a Bonferroni correction. CpG deamination was calculated separately from the mutational spectra, and therefore not corrected for multiple testing