More radical amino acid replacements in primates than in rodents: support for the evolutionary role of effective population size

Abstract

We examined the pattern of nucleotide substitution in 4933 conserved single-copy orthologous protein-coding genes of human, rhesus, mouse, and rat. Consistent with previous studies, the median ratio of the number of nonsynonymous substitutions per nonsynonymous site (d(N)) to the number of synonymous substitutions per synonymous site (d(S)) was significantly higher in the comparison between the two primates than in the comparison between the two rodents. This pattern was particularly strong in the case of genes expressed in the immune system, but also occurred in other genes, including a set of highly conserved genes involved in the regulation of transcription. Both synonymous and nonsynonymous differences occurred independently in the same codons in the primates and in the rodents to a greater extent than expected by chance, but the extent of the deviation from random expectation was much greater in the case of nonsynonymous differences. Parallel amino acid replacements occurred at the same sites in the primates and rodents far more frequently than expected by chance, but tended to involve very conservative amino acid changes. Divergent amino acid changes involved more chemically different amino acids than parallel changes, and divergent amino acid replacements between the primates were significantly more radical than those between the rodents. These results are most easily explained on the hypothesis that the evolution of these genes has been shaped largely by purifying selection, which has been less effective in primates than in rodents, presumably as a consequence of lower long-term effective population sizes in the former.

Figure 1

Examples of parallel (A) and divergent (B) patterns of amino acid replacement, illustrated by aligned amino acid sites (A) 147 and (B) 453 of 78Kd centrosomal protein (encoded by CEP78). Amino acids are indicated by the single-letter code.

Figure 2

(A) Frequency distribution of the difference between d_N /d_S in the primates and d_N /d_S in the rodents for 4933 orthologous single-copy genes from human, rhesus, mouse, and rat. (B) Median d_N /d_S between human and rhesus in orthologous genes categorized by presence (Im+) or absence (Im-) of immune system function and by presence (RT+) or absence (RT-) of function in regulation of transcription. Numbers of genes in each category are shown. The difference among categories with respect to median d_N /d_S was highly significant (P < 0.001; Kruskal-Wallis test). (C) Median d_N /d_S between mouse and rat in orthologous genes categorized by presence (Im+) or absence (Im-) of immune system function and by presence (RT+) or absence (RT-) of function in regulation of transcription. Numbers of genes in each category are as in Figure 2B. The difference among categories with respect to median d_N /d_S was highly significant (P < 0.001; Kruskal-Wallis test).

Figure 3

Percentages of genes with parallel (A) and divergent (B) amino acid replacements in orthologous genes categorized by presence (Im+) or absence (Im-) of immune system function and by presence (RT+) or absence (RT-) of function in regulation of transcription. In each case, there was a significant difference among categories: (A) χ² = 27.5; 3 d.f.; P < 0.001; (B) χ² = 41.9; 3 d.f.; P < 0.001.

Figure 4

(A) Observed vs. expected numbers of amino acid pairs involved in parallel amino acid replacements. Expected numbers were based on divergent amino acid replacements. The six amino acid pairs (single-letter code) are shown for which the observed number significantly (P < 0.05; Bonferroni-corrected) exceeded the expected number. The line is a 45° line. (B) Median MMY chemical distance between amino acids in parallel and divergent amino acid replacements. The median distance for parallel replacements was significantly different from that for divergent amino acid replacements either in rodents or in primates (Mann-Whitney test; P < 0.001 in each case). The median distance for divergent amino acid replacements in primates was significantly different from that in rodents (Sign test; P < 0.001).