Population genomics and transcriptional consequences of regulatory motif variation in globally diverse Saccharomyces cerevisiae strains

Abstract

Noncoding genetic variation is known to significantly influence gene expression levels in a growing number of specific cases; however, the patterns of genome-wide noncoding variation present within populations, the evolutionary forces acting on noncoding variants, and the relative effects of regulatory polymorphisms on transcript abundance are not well characterized. Here, we address these questions by analyzing patterns of regulatory variation in motifs for 177 DNA binding proteins in 37 strains of Saccharomyces cerevisiae. Between S. cerevisiae strains, we found considerable polymorphism in regulatory motifs across strains (mean π = 0.005) as well as diversity in regulatory motifs (mean 0.91 motifs differences per regulatory region). Population genetics analyses reveal that motifs are under purifying selection, and there is considerable heterogeneity in the magnitude of selection across different motifs. Finally, we obtained RNA-Seq data in 22 strains and identified 49 polymorphic DNA sequence motifs in 30 distinct genes that are significantly associated with transcriptional differences between strains. In 22 of these genes, there was a single polymorphic motif associated with expression in the upstream region. Our results provide comprehensive insights into the evolutionary trajectory of regulatory variation in yeast and the characteristics of a compendium of regulatory alleles.

Keywords: adaptive evolution; evolution; regulatory variation; yeast.

Fig. 1.

Examples of highly divergent regulatory regions across Saccharomyces cerevisiae strains. (A) Histogram of the mean pairwise percentage of variables motifs across 37 S. cerevisiae strains for each of the 5,468 intergenic regions. (B) Predicted motif calls for 37 S. cerevisiae strains are plotted for the intergenic region upstream of the gene AAH1. Each row is a strain, and colored boxes represent motif calls. Different colors represent distinct motifs. A phylogeny for the strains is shown to the left, as constructed from the motif calls for that region. (C) Predicted motif calls for a section of the intergenic region upstream of the gene FLO1. The region shown represents 1,000 bp upstream of the gene, out of 7,218 total upstream bases. A divergent clade is highlighted in gray, and within this clade wild strains are marked with a red dot in the phylogeny, whereas laboratory strains are marked with a blue dot.

Fig. 2.

Evolutionary forces acting at intergenic regions. (A) –log(NI) scores for three classes of sites (experimentally determined sites, noncoding sites falling within predicted motifs, and noncoding sites falling outside predicted motifs) are plotted. –log(NI) scores were obtained by summing information across all sites of a particular class and using synonymous sites within genes as putatively neutral sites. Confidence intervals were obtained by bootstrapping (see Materials and Methods). 95% CI for nonsynonymous sites are shown as in gray. (B) –log(NI) scores for each of 133 motifs, sorted from lowest –log(NI) to highest –log(NI). –log(NI) scores were obtained by summing information across all sites genome-wide falling within a particular motif, and comparing with all synonymous sites. Motifs with low numbers of polymorphic and divergent sites were excluded due to low power to detect differences with such low counts (<15 total sites). Confidence intervals were obtained by bootstrapping (see Materials and Methods).

Fig. 3.

Examples of motifs effecting gene expression. (A) NAR1 expression in strains containing the two labeled sequences at the motif REB1 in the upstream intergenic region. Substitutions to the consensus motif sequence for REB1 are marked in blue. A sequence logo for REB1 representing the PSSM is shown in the upper left corner. (B) YER186C expression in strains containing the two labeled sequences at the motif AFT2 in the upstream intergenic regions. Substitutions to the consensus motif sequence are marked in blue. A sequence logo for AFT2 representing the PSSM is shown to the upper left of the plot.

Fig. 4.

Effects of variants at specific motifs on gene expression. For each motif, a box plot of the difference in expression Z scores between strains containing the motif and strains not containing the motif at all genes with a variable motif are plotted. Motifs are sorted by mean difference in expression. Motifs significant in our test for genome-wide differences in expression for showing lower expression when the motif is present are colored in blue, and motifs significant for showing greater expression when the motif is present are colored in magenta.