Genome-Wide Identification of Regulatory Sequences Undergoing Accelerated Evolution in the Human Genome

Abstract

Accelerated evolution of regulatory sequence can alter the expression pattern of target genes, and cause phenotypic changes. In this study, we used DNase I hypersensitive sites (DHSs) to annotate putative regulatory sequences in the human genome, and conducted a genome-wide analysis of the effects of accelerated evolution on regulatory sequences. Working under the assumption that local ancient repeat elements of DHSs are under neutral evolution, we discovered that ∼0.44% of DHSs are under accelerated evolution (ace-DHSs). We found that ace-DHSs tend to be more active than background DHSs, and are strongly associated with epigenetic marks of active transcription. The target genes of ace-DHSs are significantly enriched in neuron-related functions, and their expression levels are positively selected in the human brain. Thus, these lines of evidences strongly suggest that accelerated evolution on regulatory sequences plays important role in the evolution of human-specific phenotypes.

Keywords: DHS; accelerated evolution; regulatory sequence..

Fig. 1

(a) Phylogenetic tree of an ace-DHS and its local ancient repeat elements. (b) Bar plot for the odds ratio of the overlap between ace-DHSs and conserved non-coding sequences found to be under accelerated evolution by four other studies. A P value < 0.001 is indicated by ‘***’.

Fig. 2

(a) Scatterplots of the percentage of background DHSs (blue) and ace-DHSs (red) that are active in more than a given number of cell lines. Inner box shows the boxplots of the number of cell lines in which a DHS is active. (b) The top box shows the heat map after bi-clustering of ace-DHSs (row) and cell lines (column). The bar color to the left of the heat map corresponds to different clusters of ace-DHSs. The bottom box shows the scatterplots of the mean percentage of ace-DHSs in a given cluster that are active in a specific cell line. Each scatterplot represents a cluster of ace-DHSs and is colored the same as that in the top box. The text below the x axis represents the BTO ID of each cell line. (c) Bar plot of the −log10(P value) for the enrichment of epigenetic marks in ace-DHSs in comparison to that in background DHSs in the CD4 cell line. The three sub-figures on the right show the distributions of H4K20ME1, nucleosome, and H3K9me3 ChIP-Seq reads within ±1 kb of the center of ace-DHSs and background DHSs, respectively.

Fig. 3

The GO enrichment results for the target genes of ace-DHSs. Local, Distal and PSG represent the target genes for local ace-DHSs, the target genes for distal ace-DHSs, and positively selected genes, respectively. The enriched GO terms are organized into different clusters according to their relatedness, and the most significant GO terms of the top 10 clusters are shown in bar plots, in which the bar length corresponds to −log10(P value).

Fig. 4

Gene expression analysis of the target genes of ace-DHSs. (a) The expression levels of the target genes of ace-DHSs in comparison to the target genes of background DHSs in six organs of human (top) and chimpanzee (middle) and the change of expression level of the target genes of ace-DHSs in six organs from chimpanzee to human (bottom). The genes in chimpanzee refer to the orthologous genes of human ace-DHSs target genes. The red color indicates that the target genes of ace-DHSs are expressed at significantly higher levels than those of background ace-DHSs or are up-regulated in human. The blue color indicates the opposite. (b) and (c) show two examples of ace-DHSs within which the transcription factor-binding site contains a mutation in the human ace-DHSs, causing increased binding affinity for the corresponding transcription factor. The content in the box is the JASPAR ID for the corresponding transcription factor.

Fig. 5

(a) Boxplot for the mean DAF (derived allele frequency) of SNPs within ace-DHSs and within background_DHSs. Three red stars indicate that the difference of the mean DAF between ace-DHSs and background DHSs is significant, with P values < 0.001. (b) Percentage of ace-DHSs and background DHSs that contain positively selected SNPs.