Shotgun proteomics data from multiple organisms reveals remarkable quantitative conservation of the eukaryotic core proteome

Genome-wide, absolute quantification of expressed proteins is not yet within reach for most eukaryotes. However, large numbers of MS-based protein identifications have been deposited in databases, together with information on the observation frequencies of each peptide spectrum ("spectral counts"). We have conducted a meta-analysis using several million peptide observations from five model eukaryotes, establishing a consistent, semi-quantitative analysis pipeline. By inferring and comparing protein abundances across orthologs, we observe: (i) the accuracy of spectral counting predictions increases with sampling depth and can rival that of direct biochemical measurements, (ii) the quantitative makeup of the consistently observed core proteome in eukaryotes is remarkably stable, with abundance correlations exceeding Rs = 0.7 at an evolutionary distance greater than 1000 million years, and (iii) some groups of proteins are more constrained than others. We argue that our observations reveal stabilizing selection: central parts of the eukaryotic proteome appear to be expressed at well-balanced, near-optimal abundance levels. This is consistent with our further observations that essential proteins show lower abundance variations than non-essential proteins, and that gene families that tend to undergo gene duplications are less well constrained than families that keep a single-copy status.