Summary The biological causes of selective pressures on coding-sequence evolution remain controversial, despite the surprising consistency of covariation between common measures of evolutionary change (substitution rates) and gene expression (mRNA levels, codon usage) across taxa. We carry out a unified analysis which reveals these conserved patterns in E. coli, yeast, worm, fly, mouse, and human, and suggests that all trends stem largely from a unified underlying selective pressure. In metazoans, these trends are strongest in tissues composed of neurons, whose structure and lifetime confer extreme sensitivity to protein misfolding. We propose, and demonstrate using a molecular-level evolutionary simulation, that selection against toxicity of misfolded proteins generated by ribosome errors suffices to create all the observed covariation. The mechanistic model of molecular evolution which emerges yields testable biochemical predictions, calls into question use of nonsynonymous-to-synonymous substitution ratios (Ka/Ks) to detect functional selection, and suggests how mistranslation may contribute to neurodegenerative disease.

Summary In eukaryotic cells, diverse stresses trigger coalescence of RNA-binding proteins into stress granules. In vitro, stress-granule-associated proteins can demix to form liquids, hydrogels, and other assemblies lacking fixed stoichiometry. Observing these phenomena has generally required conditions far removed from physiological stresses. We show that poly(A)-binding protein (Pab1 in yeast), a defining marker of stress granules, phase-separates and forms hydrogels in vitro upon exposure to physiological stress conditions. Other RNA-binding proteins depend upon low-complexity regions (LCRs) or RNA for phase separation, whereas Pab1’s LCR is not required for demixing, and RNA inhibits it. Based on unique evolutionary patterns, we create LCR mutations which systematically tune its biophysical properties and Pab1 phase separation in vitro and in vivo. Mutations which impede phase separation reduce organism fitness during prolonged stress. Poly(A)-binding protein thus acts as a physiological stress sensor, exploiting phase separation to precisely mark stress onset, a broadly generalizable mechanism.

Much recent work has explored molecular and population-genetic constraints on the rate of protein sequence evolution. The best predictor of evolutionary rate is expression level, for reasons that have remained unexplained. Here, we hypothesize that selection to reduce the burden of protein misfolding will favor protein sequences with increased robustness to translational missense errors. Pressure for translational robustness increases with expression level and constrains sequence evolution. Using several sequenced yeast genomes, global expression and protein abundance data, and sets of paralogs traceable to an ancient whole-genome duplication in yeast, we rule out several confounding effects and show that expression level explains roughly half the variation in Saccharomyces cerevisiae protein evolutionary rates. We examine causes for expression's dominant role and find that genome-wide tests favor the translational robustness explanation over existing hypotheses that invoke constraints on function or translational efficiency. Our results suggest that proteins evolve at rates largely unrelated to their functions and can explain why highly expressed proteins evolve slowly across the tree of life.evolutionary rate ͉ protein misfolding ͉ yeast ͉ translation errors ͉ gene duplication A central problem in molecular evolution is why proteins evolve at different rates. Protein evolutionary rates, quantified by the number of nonsynonymous nucleotide changes per site (dN) in the encoding genes, are routinely used to build phylogenetic trees, detect selection, find orthologous proteins among related species (1), and evaluate the functional importance of genes (2), yet we possess only hints of the biophysical cause of rate differences. Thirty years ago, Zuckerkandl (3) proposed that a protein's sequence will evolve at a rate primarily determined by the proportion of its sites involved in specific functions (or ''functional density''). Although this proposal has gained wide acceptance (2), measurement of functional density remains problematic because residues may contribute to protein function in unpredictable ways, and arduous sequence-wide saturation mutagenesis and mutant characterization studies are required to ascertain these effects.Instead, many recent studies have focused on other, more readily obtained, measures that may approximate functional density. For example, protein-protein interactions presumably constrain interfacial residues, and some reports indicate that highly interactive proteins evolve slowly (4). The intuition that a protein's overall functional importance should amplify the fitness costs of mutations at sites that make subtle functional contributions has been captured in analyses of how a gene's functional category (5, 6), its essentiality for organism survival (6-8), or the fitness effect of its deletion (or ''dispensability'') (9, 10) correlate with evolutionary rate. In all cases, the effects under consideration explain only a small fraction (Ϸ5% or less) of the observed variation in evolutionary...