Relative exchangeabilities (REs) between amino acid pairs are major components in protein sequence substitution models, which is core to molecular evolution analyses such as inference of phylogeny and positive selection. Contradicting conventional assumption of its consistency across a phylogenetic tree, recent studies have discovered that REs show considerable variation among different clades of species on the Tree of Life. However, the underlying mechanism of this RE heterogeneity is largely unknown. Investigating 79 closely related pairs of genomes widely sampled across all prokaryotes, we found that their REs can be clustered into three groups. Remarkably, these groups exhibit notable differences in features like genome size and gene number, despite their statistically indistinguishable divergence levels and overall selection strengths (dN/dS). Functional annotation suggests that different groups enrich prokaryotes with different lifestyles, e.g., generalists versus specialists. Comparing between these groups, we found that different lifestyles may favor radical amino acid substitutions of different physicochemical properties, thus leading to between-clade RE difference. We further investigated a group of 20 eukaryotic clades to find unique RE and conservativeness patterns. Hence, RE patterns among species clades correspond to their distinct lifestyles, providing a mechanistic explanation of this proteome evolution heterogeneity.
Species clades with different lifestyles exhibit different relative amino acid exchangeability in protein evolution