Accumulation of cis- and trans-regulatory variations is associated with phenotypic divergence of a complex trait between yeast species

Abstract

Gene regulatory variations accumulate during evolution and alter gene expression. While the importance of expression variation in phenotypic evolution is well established, the molecular basis remains largely unknown. Here, we examine two closely related yeast species, Saccharomyces cerevisiae and Saccharomyces paradoxus, which show phenotypical differences in morphology and cell cycle progression when grown in the same environment. By profiling the cell cycle transcriptome and binding of key transcription factors (TFs) in the two species and their hybrid, we show that changes in expression levels and dynamics of oscillating genes are dominated by upstream trans-variations. We find that multiple cell cycle regulators show both cis- and trans-regulatory variations, which alters their expression in favor of the different cell cycle phenotypes. Moreover, we show that variations in the cell cycle TFs, Fkh1, and Fkh2 affect both the expression of target genes, and the binding specificity of an interacting TF, Ace2. Our study reveals how multiple variations accumulate and propagate through the gene regulatory network, alter TFs binding, contributing to phenotypic changes in cell cycle progression.

Keywords: gene regulation; genome evolution; transcription factors binding.

Figure 1

Trans-dominant expression program follows the growth pattern. (A) Microscopy images of S. cerevisiae, S. paradoxus, and their hybrid grown on YPD. Saccharomyces cerevisiae display bipolar budding and G1 daughter cell delay. Saccharomyces paradoxus displays unipolar and synchronized budding and remains attached after cytokinesis. The hybrid grows like S. cerevisiae, while exhibiting a short daughter-delay. Black arrows mark new buds. (B) Experimental layout: Cells were synchronized to early S-phase using HU, washed, and sampled for transcriptional profiling and DNA staining for 3 h in 5-min intervals. (C) Synchronized progression was monitored by measuring the DNA content using flow-cytometry (full profiles are shown in Supplementary Figure S1A). (D) Co-expression of cycling genes is conserved between species. Shown are gene–gene correlation matrices (Pearson) of periodic genes ordered by their expression peak along the cell cycle (Santos et al. 2015), S. cerevisiae is shown in the lower triangle and S. paradoxus in the top triangle. (E) Histograms showing the distributions of co-expression correlation coefficients between orthologous genes and between random pairs. (F) Periodicity of gene expression is synchronized with cell cycle transitions. Shown are the average fold changes of periodic genes (relative to median) classified to three groups based on expression time [defined in Santos et al. (2015)]. Black panel indicates percentage of cells in G2-M by DNA content profiles. Correlation values in bottom indicate the Pearson correlation coefficient between expression of the G2-M module to percent cells in G2-M by DNA content. (G) Differences in cell cycle phase dynamics: Shown is the duration of the cell cycle phases along the experiment based on (F).

Figure 2

Cell cycle regulators accumulated regulatory variations that bias expression in support of the respective phenotype (A) Expression of cyclins. Left: gene expression differences between the species and between hybrid alleles. Presented fold change values are: log₂(reads in S. cerevisiae/reads in S. paradoxus). Genes ordered by their expression time along the cell cycle. Asterisk marks significant difference in expression (corrected P-value < 0.05 and log₂FC > 0.5). Right: Gene expression levels for example genes: CLN3 (early G1), CLB6 (early S), and CLB1 (G2) along the cell cycle, in synchrony with changes in DNA content (bottom panel). (B) Expression of cell cycle TFs, same as in (A). (C) log₂ FC in TFs target genes: Fkh1/2 (Zhu et al. 2000), Ace2 (Voth et al. 2007), and Swi5 (Voth et al. 2007). Black dots indicate the group’s mean. Targets defined by MacIsaac et al. (2006) are shown in Supplementary Figure S2A. Ace2 targets are shown in detail in Supplementary Figure S2B (D) Divergence in cell cycle regulators. A scheme of cell cycle regulation, color indicates inter-species expression differences. (E) DNA contents profiles of WT, ace2Δ/Δ and swi5Δ/Δ, in both species and hybrid, taken during exponential growth in YPD. Grey area marks cells in G1 phase.

Figure 3

Cis- and trans-regulatory variations affect Ace2 binding (A) Meta-gene profiles of Ace2 and Swi5 in the two species. All genes were aligned by their TSS and the signal was averaged. Small panel shows the sequence logos of the DNA-motif bound by each factor based on top 10 scoring motifs. (B) ChEC-seq binding profiles of Ace2 between species. Top: plotted is Ace2 sum of signal on each promoter after Z-score standardization. Pearson correlation values of top 100 promoters are shown, genes involved in cell separation are marked in red. Bottom: Relative Ace2 binding on cell-separation genes. (C) Binding profiles of Ace2 within the hybrid, same as in (A). (D) Ace2 differentially binds PRY3 promoter. Top: Ace2 binding on PRY3 promoter, black circles and squares represent Ace2 binding motif (CCAGC) on + and − strand, respectively. Background line represents nucleosome occupancy (Tirosh et al. 2010). Dashed line represent the TSS (Pelechano et al. 2013), x-ticks are bp relative to TSS. Bottom: Sequence changes leading to loss or gain of Ace2 motif. (E) Expression of PRY3 along the cell cycle, note the cis-effect between hybrid alleles. (F and G) Binding profiles of Swi5 between species and within hybrid, same as in (B and C).

Figure 4

Divergence in binding of Fkh1 and Fkh2 on Ace2 targets (A) Pearson correlation matrix between the top 100 promoters of Ace2, Swi5, Fkh1, and Fkh2. Hyc and Hyp refer to Hybrid S. cerevisiae genome and S. paradoxus genome, respectively. Note that Fkh1 shows low-positive correlation with S. cerevisiae Ace2. (B–E) Binding profiles of Fkh1 and Fkh2 between species and within the hybrid, same as Figure 3B, marked in blue are Ace2 targets (Voth et al. 2007). (F) Binding of Ace2 and Fkh1 on the cell-separation gene DSE2. Circles and squares represent Ace2 binding motif (CCAGC, black), and Fkh1 binding motif (GTAAACA, yellow) on + and − strand, respectively. Note trans-effects both in Ace2 and in Fkh1. (G) Binding on the top promoters of Ace2 and Swi5. Top: Shown are relative sum of signal on promoters for each factor in the species and hybrid. Note overlap of Fkh1 and Fkh2 with Ace2 promoters. Bottom: Differences in expression levels. Cell-separation genes are marked in red.

Figure 5

Fkh1 and Fkh2 directly mediate Ace2 binding to cell-separation genes (A) Ace2 binding is mediated by Fkh1 and Fkh2. Shown are normalized sum of signal on promoters for Swi5 and Ace2 in WT S. cerevisiae compared to fkh1Δ, fkh2Δ, and fkh1Δfkh2Δ. Color represents correlation of top 100 promoters to WT. (B) Cis- and trans-effects on Ace2 binding. Scatter plot of relative Ace2 binding differences between species (x-axis) and between hybrid alleles (y-axis). Color represents the differences in Ace2 binding between S. cerevisiae WT to fkh1Δfkh2Δ. Note that cell-separation genes show a strong trans-effect and are strongly affected by fkh1Δfkh2Δ. (C) S. paradoxus’ FKH2 is sufficient to restore Ace2 binding in S. cerevisiae. Scatter plots of normalized sum of signal on promoters of Ace2 in S. cerevisiae WT vs. S. cerevisiae fkh1Δ expressing S. paradoxus FKH2. Red dots represent cell-separation genes. (D) Scheme for proposed mechanism driving cell cycle variation: in S. cerevisiae, higher expression of Fkh1 and Fkh2 leads to higher expression of target genes and faster G2-M transition. Fkh1 (and Fkh2, to lower extent) binds Ace2 target genes and recruits Ace2, leading to higher expression of target genes, G1 daughter-cell delay and cell separation. In S. paradoxus, lower expression of Fkh1 and Fkh2 leads to reduced expression of target genes (including Fkh1 itself and Ace2) and longer G2. Downstream, Fkh1 exhibits reduced binding to Ace2 targets (trans-effect) leading to reduced binding of Ace2, lower expression of target genes, shorter G1-phase and reduced separation.