The Roles of Compensatory Evolution and Constraint in Aminoacyl tRNA Synthetase Evolution

Abstract

Mitochondrial protein translation requires interactions between transfer RNAs encoded by the mitochondrial genome (mt-tRNAs) and mitochondrial aminoacyl tRNA synthetase proteins (mt-aaRS) encoded by the nuclear genome. It has been argued that animal mt-tRNAs have higher deleterious substitution rates relative to their nuclear-encoded counterparts, the cytoplasmic tRNAs (cyt-tRNAs). This dynamic predicts elevated rates of compensatory evolution of mt-aaRS that interact with mt-tRNAs, relative to aaRS that interact with cyt-tRNAs (cyt-aaRS). We find that mt-aaRS do evolve at significantly higher rates (exemplified by higher dN and dN/dS) relative to cyt-aaRS, across mammals, birds, and Drosophila. While this pattern supports a model of compensatory evolution, the level at which a gene is expressed is a more general predictor of protein evolutionary rate. We find that gene expression level explains 10-56% of the variance in aaRS dN/dS, and that cyt-aaRS are more highly expressed in addition to having lower dN/dS values relative to mt-aaRS, consistent with more highly expressed genes being more evolutionarily constrained. Furthermore, we find no evidence of positive selection acting on either class of aaRS protein, as would be expected under a model of compensatory evolution. Nevertheless, the signature of faster mt-aaRS evolution persists in mammalian, but not bird or Drosophila, lineages after controlling for gene expression, suggesting some additional effect of compensatory evolution for mammalian mt-aaRS. We conclude that gene expression is the strongest factor governing differential amino acid substitution rates in proteins interacting with mitochondrial versus cytoplasmic factors, with important differences in mt-aaRS molecular evolution among taxonomic groups.

Keywords: compensatory evolution; gene expression; mitochondrial-nuclear coevolution; protein evolution.

Fig. 1.

Protein translation requires physical interactions between aaRS proteins and tRNAs. The nuclear genome encodes aaRS targeted to the cytosol (cyt-aaRS) that interact with the nuclear-encoded tRNAs (cyt-tRNAs) and aaRS targeted to the mitochondria (mt-aaRS) that interact with mitochondrial-encoded tRNAs (mt-tRNAs).

Fig. 2.

Phylogenetic relationships among species sampled in our analyses. Gene trees, based on gene trees inferred from RAxML and used for the estimation of d_N/d_S (ω) in PAML, rarely conflicted with the known species tree for birds and flies. However, all mammalian gene trees were discordant. The mammalian tree shown represents the known species tree for the full set of mammals used in our analysis (Bininda-Emonds et al. 2007). For some aaRS genes, only subsets of these species were included, based on availability, quality, and length of sequence data.

Fig. 3.

Estimates of d_N/d_S (ω) for nuclear-encoded mt-aaRS (gray boxes) and cyt-aaRS (white boxes). mt-aaRS evolve more rapidly than cyt-aaRS in all taxa analyzed. Estimates of ω were generated using codeml (model = 0, NSsites = 0) in PAML. P values indicate significant differences based on Mann–Whitney U tests.

Fig. 4.

Transcript levels (FPKM) for mt-aaRS (gray boxes) and cyt-aaRS (white boxes). cyt-aaRS are expressed at higher levels than mt-aaRS for all tissues analyzed. Statistical significance: **P < 0.001 and ***P < 0.0001 based on Mann–Whitney U tests.

Fig. 5.

The relationship between transcript level (FPKM) and d_N/d_S (ω) for aaRS genes. A significant negative relationship exists between FPKM and d_N/d_S in chicken and D. melanogaster, but not in mouse. GLM regression lines are shown separately for mt-aaRS (closed circles, red dash-dot line) and cyt-aaRS (open triangles, blue dashed line) and for all aaRS genes (solid black line). R² and P values are from regressions using data from all aaRS genes.

Fig. 6.

The residuals from GLM regressions of d_N/d_S (ω) on gene expression (FPKM) using data from all aaRS genes for cyt-aaRS (white boxes) and mt-aaRS (gray boxes). In mouse, but not chicken or flies, the pattern of greater d_N/d_S in mt-aaRS proteins persists after the effects of gene expression are removed. Statistical significance: **P < 0.001 based on Mann–Whitney U tests.