Abstract
The causes of rate variation among sites within proteins are as yet poorly understood. Here, we compare the spatial autocorrelation of non-synonymous substitutions among species within diverse phylogenetic groups: Saccharomyces, Drosophila, Arabidopsis, and primates. Across these taxa, we find that amino acid substitutions exhibit excess clustering that extends over a 20-30 codon length (10-20 Angstrom distance) scale. We show that these substitutions cluster more strongly and exhibit compensatory dynamics within species lineages but exhibit patterns of convergent evolution between lineages. We evaluate a simple model of thermodynamic constraints on protein folding and conclude that it is unable to recapitulate the observed spatial clustering of substitutions. While pairs of substitutions with the strongest epistasis tend to spatially cluster in these simulations, the magnitude and length scale are smaller than that observed in real data. Additionally, we show that the pattern of convergent substitution is also not expected under this model, suggesting it is likely caused by factors other than these simple thermodynamic constraints. Our results support a prevalent role for epistasis and convergent evolution in shaping protein evolution across the tree of life.
Competing Interest Statement
The authors have declared no competing interest.
