Fast evolution of core promoters in primate genomes

Abstract

Despite much interest in regulatory evolution, how promoters have evolved remains poorly studied, mainly owing to paucity of data on promoter regions. Using a new set of high-quality experimentally determined core promoters of the human genome, we conducted a comparative analysis of 2,492 human and rhesus macaque promoters and their neighboring nearly neutral regions. We found that the core promoters have an average rate of nucleotide substitution substantially higher than that at 4-fold degenerate sites and only slightly lower than that for the assumed neutral controls of neighboring noncoding regions, suggesting that core promoters are subject to very weak selective constraints. Interestingly, we identified 24 core promoters (at false discovery rate = 50%) that have evolved at an accelerated rate compared with the neutral controls, suggesting that they may have undergone positive selection. The inferred positively selected genes show strong bias in molecular function. We also used population genetic approaches to examine the evolution of core promoters in human populations and found evidence of positive selection at some loci. Taken together, our results suggest that positive selection has played a substantial role in the evolution of transcriptional regulation in primates.

FIG. 1.—

Different evolutionary patterns between core promoters and coding regions. (A) Proportion of core promoters that have evolved significantly faster (black bars) or more slowly (striped bars) than the 4-fold degenerate sites in the same genes. The proportion expected from multiple testing is shown (gray bars). (B) Proportion of genes that have evolved significantly faster (black bars) or more slowly (striped bars) at the second codon positions than at the 4-fold degenerate sites in the same genes. Only 2 genes have K_2nd/K₄ > 1 at P value < 0.05 and no gene at P value < 0.01 or 0.005. The total number of core promoters studied is 2,492 (in 2,333 genes). The expected numbers from multiple testing are calculated using the P value × the total number of promoters (genes) in the analysis.

FIG. 2.—

The observed distribution of core promoters that have evolved significantly faster than neutral controls. Striped bars represent the observed numbers at different P value intervals, and the gray square represents the uniform distribution expected from chance alone. The light gray square highlights the positively selected promoters identified in this study (P < 0.005, false discovery rate = 50%).