Genetic analysis of variation in transcription factor binding in yeast

Abstract

Variation in transcriptional regulation is thought to be a major cause of phenotypic diversity. Although widespread differences in gene expression among individuals of a species have been observed, studies to examine the variability of transcription factor binding on a global scale have not been performed, and thus the extent and underlying genetic basis of transcription factor binding diversity is unknown. By mapping differences in transcription factor binding among individuals, here we present the genetic basis of such variation on a genome-wide scale. Whole-genome Ste12-binding profiles were determined using chromatin immunoprecipitation coupled with DNA sequencing in pheromone-treated cells of 43 segregants of a cross between two highly diverged yeast strains and their parental lines. We identified extensive Ste12-binding variation among individuals, and mapped underlying cis- and trans-acting loci responsible for such variation. We showed that most transcription factor binding variation is cis-linked, and that many variations are associated with polymorphisms residing in the binding motifs of Ste12 as well as those of several proposed Ste12 cofactors. We also identified two trans-factors, AMN1 and FLO8, that modulate Ste12 binding to promoters of more than ten genes under alpha-factor treatment. Neither of these two genes was previously known to regulate Ste12, and we suggest that they may be mediators of gene activity and phenotypic diversity. Ste12 binding strongly correlates with gene expression for more than 200 genes, indicating that binding variation is functional. Many of the variable-bound genes are involved in cell wall organization and biogenesis. Overall, these studies identified genetic regulators of molecular diversity among individuals and provide new insights into mechanisms of gene regulation.

Figure 1. Extensive Ste12 binding variations among S288c × JM789 derivatives

a, ChIP-Seq signal tracks showing Ste12 binding sites that segregate in a Mendelian (Left) or transgressive fashion (Middle and Right). The color indicates genotype background in the depicted regions: red (S96), green (HS959). Asterisks indicate peaks of interest. b, Overall genomic positions of Ste12 binding regions in S96 or HS959 under vegetative growth condition (green), pheromone treatment (blue), and most variable binding regions across segregants (red). c, Overlap of target genes (from 6644 annotated genes) between parent strains with pheromone treatment. Enriched GO categories are listed.

Figure 2. Whole-genome linkage analysis of variable Ste12 binding traits

a, Chromosomal position of significantly associated binding traits (X axis) relative to markers (Y axis) passing threshold (FDR = 0.01). See Fig. S4 for results with different thresholds. Trans-QTL regions validated by the clustering method are shown on the right. Two experimentally validated causative genes (AMN1 and FLO8) underlying the trans-QTLs are also shown. b, Histogram of markers significantly associated with multiple binding traits. c, Effect size (explained variance in quantitative traits) of cis-QTLs < 10kb to the trait (black), cis-QTLs > 10kb to the trait (red), and trans-QTLs (green).

Figure 3. Motif analysis of cis-variable binding regions

a-c, Ste12 binding signals (NormDiff scores) against the genotypes of Ste12 motif, Yhp1 motif and Yap5 motif in two target loci. Each dot plot shows strains with S96 inheritance (red dot), strains with HS959 inheritance (green dot). The mean and standard error of each group, and Ste12 consensus sequence are also shown (Red, S96; green, HS959). d-f, ChIP-Seq signal tracks of the depicted regions in dot plots a-c. The color in each track indicates genotype background in the depicted regions: red (S96), green (HS959). Additional examples are in Fig. S5.

Figure 4. Validation of two causative quantitative trait genes

a, ChIP-Seq signal tracks showing Ste12 binding of wild-type, amn1Δ, and ics2Δ parent strains, corresponding to variable binding trait B12S708456E710089. b, ChIP-Seq signal tracks showing Ste12 binding of wild-type, flo8Δ, and gle2Δ parent strains, corresponding to variable binding traits B1S202520E202741. ics2Δ and gle2Δ strains have no effect on binding. c, d, Comparison of Ste12 binding across 10 and 5 associated variable binding regions in WT and knockout strains rspectively. Error bar represents s.e.m. Additional results are in Fig. S7.

Figure 5. Ste12 binding significantly correlates with downstream gene expression

a, Examples of high correlation between binding variation and gene expression variation. Columns in the heat maps are ordered by the genotype of markers with highest association to the Ste12 binding traits. 10 additional clusters are shown in Fig. S8. b, Histogram (blue) and background histogram (yellow) of correlation coefficients between Ste12 binding and expression of nearest gene. Regions with absolute Pearson's correlation |r| > 0.335 are shaded.