Non-random clustering of stress-related genes during evolution of the S. cerevisiae genome

BACKGROUND: Coordinately regulated genes often physically cluster in eukaryotic genomes, for reasons that remain unclear. RESULTS: Here we provide evidence that many S. cerevisiae genes induced by starvation and other stresses reside in non-random clusters, where transcription of these genes is repressed in the absence of stress. Most genes essential for growth or for rapid, post-transcriptional responses to stress in cycling cells map between these gene clusters. Genes that are transcriptionally induced by stresses include a large fraction of rapidly evolving paralogues of duplicated genes that arose during an ancient whole genome duplication event. Many of these rapidly evolving paralogues have acquired new or more specialized functions that are less essential for growth. The slowly evolving paralogues of these genes are less likely to be transcriptionally repressed in the absence of stress, and are frequently essential for growth or for rapid stress responses that may require constitutive expression of these genes in cycling cells. CONCLUSION: Our findings suggest that a fundamental organizing principle during evolution of the S. cerevisiae genome has been clustering of starvation and other stress-induced genes in chromosome regions that are transcriptionally repressed in the absence of stress, from which most genes essential for growth or rapid stress responses have been excluded. Chromatin-mediated repression of many stress-induced genes may have evolved since the whole genome duplication in parallel with functions for proteins encoded by these genes that are incompatible with growth. These functions likely provide fitness effects that escape detection in assays of reproductive capacity routinely employed to assess evolutionary fitness, or to identify genes that confer stress-resistance in cycling cells.