Hotspots of biased nucleotide substitutions in human genes

Abstract

Genes that have experienced accelerated evolutionary rates on the human lineage during recent evolution are candidates for involvement in human-specific adaptations. To determine the forces that cause increased evolutionary rates in certain genes, we analyzed alignments of 10,238 human genes to their orthologues in chimpanzee and macaque. Using a likelihood ratio test, we identified protein-coding sequences with an accelerated rate of base substitutions along the human lineage. Exons evolving at a fast rate in humans have a significant tendency to contain clusters of AT-to-GC (weak-to-strong) biased substitutions. This pattern is also observed in noncoding sequence flanking rapidly evolving exons. Accelerated exons occur in regions with elevated male recombination rates and exhibit an excess of nonsynonymous substitutions relative to the genomic average. We next analyzed genes with significantly elevated ratios of nonsynonymous to synonymous rates of base substitution (dN/dS) along the human lineage, and those with an excess of amino acid replacement substitutions relative to human polymorphism. These genes also show evidence of clusters of weak-to-strong biased substitutions. These findings indicate that a recombination-associated process, such as biased gene conversion (BGC), is driving fixation of GC alleles in the human genome. This process can lead to accelerated evolution in coding sequences and excess amino acid replacement substitutions, thereby generating significant results for tests of positive selection.

Figure 1. Cumulative Bias in Different Classes of Nucleotide Substitutions in the Exons with the Highest Degree of Acceleration on the Human Lineage

(A) The proportion of W→S substitutions compared to W→S and S→W substitutions on the human lineage. (B) The proportion of nonsynonymous substitutions compared to all substitutions in each gene on the human lineage. (C) The proportion of W→S substitutions compared to W→S and S→W substitutions on the human lineage in the flanking noncoding regions. Dashed lines represent averages for the entire dataset.

Figure 2. Genes Containing the Most Accelerated Exons

Exon boundaries are marked with black lines. S→W substitutions on the human lineage are marked with blue lines, W→S substitutions on the human lineage are marked with red lines, and all other substitutions on the human lineage are marked with grey lines.

Figure 3. Patterns of Base Substitution in Flanking Regions

Average W→S bias of base substitution patterns in noncoding regions surrounding the top 20 accelerated exons (blue bars) compared with W→S bias in noncoding regions surrounding all exons in the dataset (red bars). 95% confidence intervals were estimated by bootstrapping with 1,000 replicates.

Figure 4. Cumulative Average Human Recombination Rate in the Regions Surrounding the Most Accelerated Exons

(A) Cumulative average male (red), female (green), and sex-averaged (black) recombination rates. (B) Cumulative average distance to the nearest recombination hotspot. The dashed line represents the average for the entire dataset.

Figure 5. Venn Diagram Showing Overlap between Three Different Subsets of Fast-Evolving Genes

Genes with evidence for accelerated d_N/d_S on the human lineage based on a LRT p<0.05 using a chi-squared test are shown in one circle. Genes containing exons with evidence for accelerated evolutionary rate in humans based on simulations with a FDR p < 0.05 are in the second circle. Genes with evidence for a significant McDonald-Kreitman test [39] with p < 0.05 are in the third circle.

Figure 6. Predicted Effect of a Bias Towards Fixation of W→S Mutations, Considered as a Selective Coefficient (f), on Coding Sequences under Purifying Selection

(A) Effect of f on the pattern of nucleotide substitutions (B) Effect of f on the d_N/d_S ratio The line colors represent ancestral GC content (0.3–0.4, black; 0.4–0.5, red; 0.5–0.6, green; 0.6–0.7, blue).