Haichao Li, Haoxuan Liu
Metabolite abundances represent the ultimate functional outputs of cellular processes and provide a direct link between genotype and phenotype. While the evolutionary dynamics of mRNA and protein abundances have been extensively studied, the evolutionary forces shaping metabolite abundances remain largely unexplored. Here, we present a comprehensive framework to quantify natural selection acting on metabolite abundance using Saccharomyces cerevisiae as a model system. We quantified 1,037 metabolites across 21 mutation accumulation lines, their ancestor BY4741, and 18 wild isolates using liquid chromatography-tandem mass spectrometry (LC-MS/MS). Despite mutation accumulation lines and wild strains differing by ∼40-fold in genomic divergence, metabolite abundances were significantly more conserved among wild isolates, indicating strong natural selection acting on the metabolome. Using comparisons of variance components (Vg/Vm) and directional bias analyses, we found that the vast majority of metabolites were subject to natural selection. Approximately 28.58% of metabolites were inferred to be potentially under directional selection, with most of the remainder likely constrained by stabilizing selection. Notably, directional selection preferentially increased the abundance of high-abundance metabolites while reducing that of low-abundance metabolites, thereby increasing abundance divergence among metabolites. The direction of directional selection showed no significant bias across metabolite classes, except for amino acids and their derivatives, in which directional selection significantly favored reduced abundance of these metabolites, potentially reflecting their relatively high biosynthetic energy costs. Together, our findings reveal the patterns of natural selection acting on the metabolome and provide new insights into the evolution of the molecular processes linking gene expression to organismal phenotypes.