Genome-Wide Mapping of Binding Sites Reveals Multiple Biological Functions of the Transcription Factor Cst6p in Saccharomyces cerevisiae

ABSTRACT In the model eukaryote Saccharomyces cerevisiae, the transcription factor Cst6p has been reported to play important roles in several biological processes. However, the genome-wide targets of Cst6p and its physiological functions remain unknown. Here, we mapped the genome-wide binding sites of Cst6p at high resolution. Cst6p binds to the promoter regions of 59 genes with various biological functions when cells are grown on ethanol but hardly binds to the promoter at any gene when cells are grown on glucose. The retarded growth of the CST6 deletion mutant on ethanol is attributed to the markedly decreased expression of NCE103, encoding a carbonic anhydrase, which is a direct target of Cst6p. The target genes of Cst6p have a large overlap with those of stress-responsive transcription factors, such as Sko1p and Skn7p. In addition, a CST6 deletion mutant growing on ethanol shows hypersensitivity to oxidative stress and ethanol stress, assigning Cst6p as a new member of the stress-responsive transcriptional regulatory network. These results show that mapping of genome-wide binding sites can provide new insights into the function of transcription factors and highlight the highly connected and condition-dependent nature of the transcriptional regulatory network in S. cerevisiae.

mids (10), while its overexpression results in chromosome instability (11). Currently, the only well-studied direct target of Cst6p is NCE103 (12), which encodes a carbonic anhydrase converting CO 2 to HCO 3 Ϫ , the latter serving as a substrate of several carboxylation reactions. Under low-CO 2 conditions, Cst6p binds to the promoter of NCE103 and activates its expression. In some other fungal species, the orthologs of Cst6p have pleiotropic functions. Thus, in Candida albicans, Rca1p also directly activates carbonic anhydrase gene expression (12), with other phenotypes being altered hyphal formation, membrane ergosterol content, antifungal responses, and chemical resistance (12,13). In Candida glabrata, Cst6p negatively regulates the expression of the main adhesin gene EPA6 and, thereby, biofilm formation (14).
Despite extensive phenotypic studies on CST6 mutants, the genome-wide in vivo targets of Cst6p still remain unknown. The ChIP-chip study of 203 TFs in S. cerevisiae (4) included Cst6p, but the target list identified seems questionable, because the known target NCE103 was not found and no consensus binding motif could be enriched from the target sequences. We therefore mapped the binding sites of Cst6p during growth on both glucose and ethanol, using ChIP with lambda exonuclease digestion followed by sequencing (ChIP-exo), which allows for identifying the location of DNA-binding proteins at high resolution (15). Following this, we measured the regulatory effect of Cst6p on its target genes, and the biological functions of these regulatory effects were investigated. Our results provide deeper understanding of the function of Cst6p and how it integrates with the transcriptional regulatory network in S. cerevisiae.

Identification of Cst6p binding sites by ChIP-exo.
A cst6⌬ strain was reported to have a more severe growth defect on respiratory carbon sources (including ethanol and galactose) than on glucose (5,16), indicating a stronger role of this TF during growth on ethanol than during growth on glucose. We therefore aimed to identify the binding targets of Cst6p during growth on ethanol. To enable immunoprecipitation of Cst6p, we tagged it with CBP-ProtA in situ. The tag does not seem to affect the function of Cst6p during growth on ethanol, as growth was unaffected (see Fig. S1 in the supplemental material). Using ChIP-exo (Fig. 1A), we identified 40 binding sites distributed on 14 chromosomes when the cells were grown in synthetic medium with ethanol as the carbon source in batch cultivation (see Table S1 in the supplemental material). A binding site motif of 5= GTGACGT 3= was enriched from the bound region sequences (discovered in 32 of the 40 regions) (Fig. 1B). Compared with the motif determined by an in vitro protein binding microarray study (5= TGACGT 3=) (17), an additional conserved guanine nucleotide was found, indicating different preferences of DNA binding between Cst6p and Sko1p (5= A TGACGT 3=) (18). For growth on glucose, only 6 binding events with low signal-to-noise ratios were detected, of which 4 were also found during growth on ethanol (see Table S1), indicating that Cst6p hardly binds to its targets during growth on glucose. In the following analysis, we therefore only focus on the targets identified during growth on ethanol.
Of the 40 binding sites identified for growth on ethanol, 38 were located in putative promoter regions of genes, suggesting a general regulatory function of these bindings. Due to the presence of divergent promoters, a total of 59 protein-coding genes were identified as putative targets for Cst6p regulation (see Table S2 in the supplemental material). When this gene list was compared with the 106 Cst6p targets (P value Յ0.01) identified by ChIP-chip for growth in yeast extract-peptone-dextrose (YPD) medium (4), no overlap was observed. We suppose that the Cst6p tagging or the ChIP process did not work properly in the ChIP-chip study, though there is also the possibility that the targets of Cst6p on YPD are totally different from those on synthetic medium which we used. We also compared our ChIP-exo gene list with a list of orthologue genes in C. albicans found to be bound by Rca1p, the Cst6p orthologue in this yeast, during growth in YPD medium by using ChIP-chip with tiling microarray (12). Only NCE103 was shared by the two datasets. Whether this large difference is due to the condition-dependent characteristic of Cst6p binding or to the rapid evolution of regulatory networks in fungi (19) is worth exploration.
The exonuclease digestion in ChIP-exo allowed highresolution mapping of Cst6p locations. Overall, the binding sites have an average distance of 591 bp from the start codons (Fig. 1C). The previously suggested binding site on the NCE103 promoter (12) was clearly found among the sequencing read peaks on both the forward and reverse strands (Fig. 1D). Multiple peaks were observed on the promoter region of ROX1, which encodes a transcriptional repressor of hypoxic genes, probably due to the presence of two binding sites of Cst6p located close to each other ( Fig. 1E) or, perhaps, to the occupancy of other proteins at this area (20). Using the results of ChIP-exo, it is possible to validate or exclude computationally predicted binding sites on gene promoters. For example, the Cst6p binding motif 5= GTGACGT 3= located 982 bp upstream from the start codon of ACC1 was identified as a real binding site, while no binding was found at another putative binding site (5= TTGACGT 3=, 807 bp upstream from ACC1).
Gene ontology analysis of the 59 target genes ( Fig. 2; see also Table S3 in the supplemental material) revealed two notable features: (i) 10 of the targets (CBF2, GAT2, HAP4, PHD1, RMI1, ROX1, SIR4, SOK2, YAP1, and YAP6) encode DNA-binding or transcriptional regulatory proteins, and (ii) 16 of the targets encode mitochondrial proteins, including six located in the mitochondrial envelope. In terms of cellular processes, the target genes are categorized into cellular respiration, gluconeogenesis, stress response, and pseudohyphal growth. Interestingly, in addition to NCE103, Cst6p binds to the promoters of ACC1 and PYC1, whose protein products are major HCO 3 Ϫ -consuming enzymes in S. cerevisiae.
Cst6p functionally regulates the expression of its targets. To investigate the regulatory effects of the Cst6p binding events identified, the expression levels of 17 target genes in the cst6⌬ strain were compared with their expression levels in the wild type. The CST6 deletion mutant in the CEN.PK genetic background is viable on ethanol (see below), allowing the measurement of gene expression under this growth condition where we found Cst6p to be most active. As expected, the expression of CST6 itself was not detected in the cst6⌬ strain (data not shown). The expression of NCE103 had a 92.6% decrease in the cst6⌬ strain relative to its expression in the wild type, while a smaller decrease of 50.2% was observed during growth on glucose (Fig. 3). For the other target genes investigated, seven (AHP1, PHD1, YAP6, ACC1, ROX1, HAP4, and PYC1) had significantly decreased expression (approximate 15% to 45%) relative to their expression in the wild type for growth on ethanol. RPS3 had a 30.9% increase in expression, while the other eight genes tested did not show significant changes in expression between the cst6⌬ strain and the wild type. No relation could be found between the extent of the expression changes of the target genes and the binding strength (signal-to-noise ratio of ChIP-exo peaks) of Cst6p on their promoters.
The function of Cst6p during growth on ethanol is dependent on NCE103 expression. In a medium with 1% (vol/vol) ethanol as the sole carbon source, the cst6⌬ strain had a longer lag phase, whereas the specific growth rate was only slightly lower later in the cultivation (Fig. 4A). This growth defect is less severe than that reported earlier. Thus, Garcia-Gimeno and Struhl observed no growth on ethanol (5), probably due to differences in strain backgrounds and in ethanol concentrations (3% ethanol was used in their study). In a glucose medium, the growth of the cst6⌬ strain was similar to that of the wild type and respiratory growth after the diauxic shift was not affected (Fig. 4B).
Since the expression of NCE103 was decreased to a greater extent than the expression of the other target genes with growth on ethanol (Fig. 3), we evaluated whether the low expression of this gene was contributing to the slower growth of the cst6⌬ strain. Constitutive expression of NCE103 using the TEF1 promoter in the cst6⌬ strain partially restored the growth on ethanol at the early stage (Fig. 4A), and this implies that the carbonic anhydrase activity or the subsequent HCO 3 Ϫ concentration in the cst6⌬ strain is the limiting factor for the initial growth of the cst6⌬ strain on ethanol. The physiological role of Nce103p has been well documented: the NCE103 deletion mutant is inviable in atmospheric air but can be complemented by supplementing the medium with is a dubious gene upstream from ROX1. Data shown in panels B to E are based on the binding events that occurred when cells were grown on ethanol. In panels D and E, the coverage graph showing the 5= ends of reads was created from the aligned BAM file using "Depth Graph (Start)" in IGB (46). The putative binding sites of Cst6p are marked by triangles. The core binding sequences of the ATF/CREB family TFs (consensus sequence, 5= TGACGT 3=) are underlined. some nutrients whose biosynthesis needs HCO 3 Ϫ , such as fatty acids, L-aspartic acid (providing oxaloacetate), uracil, and L-arginine ( Fig. 4C) (21). Since the supplementation of L-aspartic acid is not essential for the complementation of the nce103⌬ strain grown on ethanol (21), we chose fatty acids, uracil, and L-arginine to test whether they could also improve the growth of the cst6⌬ strain on ethanol. While the growth of the wild type was slightly affected by the supplementation of nutrients, the cst6⌬ strain did show enhanced growth (Fig. 4D).
Cst6p is a member of the stress-responsive transcriptional regulatory network. Since Cst6p and Aca1p bind to a sequence similar to the sequence bound by the stress-responsive Sko1p in vitro, they were at first supposed to be involved in response to stress. However, phenotypic and gene expression analyses showed that Cst6p and Aca1p are not involved in the responses to various stresses (5). Another study reached a similar conclusion based on studies of osmotic stress but showed that the cst6⌬ strain is more sensitive to oxidative stress than the wild type (22). Both these studies used YPD medium, and an effect of the brand of yeast extract and peptone on phenotype was noticed in the latter study. As our ChIP-exo results show that Cst6p binds to some stressrelated genes during growth on ethanol, we wondered if Cst6p is involved in stress response under this condition. Growth test on ethanol showed that the cst6⌬ strain is more sensitive than the wild type to H 2 O 2 (Fig. 5A) and high concentrations of ethanol (Fig. 5B). On glucose, the cst6⌬ strain showed sensitivity to H 2 O 2 similar to that of the wild type (Fig. 5C). Hypersensitivity to ethanol stress was also observed for the cst6⌬ strain grown on glucose but was less apparent than when it was grown on ethanol (Fig. 5D).
This indicates that Cst6p is indeed a member of the stressresponse regulatory network in S. cerevisiae. We therefore explored the relationship between Cst6p and other members of this extensively studied network (23,24). As mentioned above, YAP1 and YAP6, encoding stress-responsive TFs, are direct targets of Cst6p. Furthermore, overlap analysis of the Cst6p targets with the known bound targets of all the TFs in S. cerevisiae suggests that Cst6p coregulates gene expression with 106 other TFs (see Table S4 in the supplemental material). Notably, among the top 10 TFs with the highest number of shared targets with Cst6p ( Fig. 6A), five (Sko1p, Msn2p, Skn7p, Cin5p, and Yap6p) are known as stress-responsive TFs and two (Ste12 and Sok2p) are involved in the regulation of pseudohyphal growth. The targets of Cst6p with extensive coregulation include the peroxiredoxinencoding AHP1, heat shock protein-encoding HSP12, molecular chaperone-encoding MDJ1, and stress-responsive TF genes like YAP6 (Fig. 6B).
Considering that Cst6p and Sko1p bind to the same sequence in vitro, we compared the binding sites for Cst6p that we identified in the work presented here with those identified by chromatin immunoprecipitation sequencing (ChIP-seq) for Sko1p (8). For most of the shared targets, the ChIP-exo peak of Cst6p was found inside the broader ChIP-seq peak region of Sko1p. Combined with the result of binding motif search, the two TFs appear to bind to the same site ( Fig. 6C shows an example). On the other hand, some well-studied direct targets of Sko1p, such as GRE2 (22), were not identified as direct targets of Cst6p.

DISCUSSION
TFs in S. cerevisiae have been studied extensively for their functions and the underlying molecular mechanisms. However, knowledge of the specific functions of some TFs is still limited, and therefore, we performed a genome-wide binding-site mapping of Cst6p. Using this information, we revealed its role in regulating  Table S3 in the supplemental material. diverse cellular processes that can explain different phenotypes reported earlier and further studied here. For example, the binding of Cst6p to chromosome segregation-related genes CBF2, GIP3, and RMI1 could be linked to the chromosome instability of a CST6 overexpression mutant. Furthermore, we find that the regulatory function of Cst6p is highly condition dependent, as almost no binding event was identified in cells grown on glucose but several were identified in cells grown on ethanol.
From our analysis, the carbonic anhydrase gene NCE103 was for the first time identified as a direct target of Cst6p in S. cerevisiae, although a putative Cst6p binding site on its promoter has been shown to be responsible for the regulation by CO 2 (12). From quantitative analysis of gene expression, we showed that Cst6p contributes to most of the activation of NCE103 expression for growth on ethanol (Fig. 3). This is consistent with the result in C. albicans, where Rca1p (ortholog of Cst6p) was identified as the only TF activating NCE103 expression by screening a TF knockout mutant library (12). Furthermore, the Cst6p regulation of NCE103 expression is the major determinant of the phenotype of a CST6 deletion strain grown on ethanol (Fig. 4A). While NCE103 is essential for HCO 3 Ϫ supply and growth under atmospheric air, the phenotype of the cst6⌬ strain is only obvious at the early stage of the growth on ethanol. When ethanol is used as the sole carbon source, the slow CO 2 production by respiration (Fig. 4C) and the markedly decreased Nce103p level may not be able to provide enough HCO 3 Ϫ for key biosynthetic reactions required for cell growth. The growth on glucose is almost not affected, probably due to the fact that NCE103 expression is not significantly affected during growth on this carbon source (Fig. 3) or because the rapid CO 2 production from fermentation ensures sufficient provision of HCO 3 Ϫ . In C. albicans, growth delay in the early stage was also observed for the rca1⌬ mutant grown on glucose, especially when a synthetic medium was used (13). Taken together, the results suggest that Cst6p/Rca1p is important for the growth of yeast under conditions with low carbonate levels and that this is due to its activation of the conserved target NCE103. From a biotechnological point of view, the expression level of the carbonic anhydrase gene NCE103 is a potential target for metabolic engineering to increase the production of related metabolites (Fig. 4C), and Cst6p could be used to design regulatory circuits that are responsive to CO 2 levels.
The binding to stress-related genes and the results of stress resistance tests establish Cst6p as a stress-responsive TF. As seen by the results in Fig. 6B, the binding of several TFs to target genes associated with stress response makes the stress-responsive regulatory network in S. cerevisiae complex. Genetic analyses have given greater functional understanding of this network: the contributions of different TFs to the transcription of a specific gene are quite different, and the regulatory effect is critically dependent on the growth condition and type of stress (8,22). From our results, Cst6p is positively involved in the expression of its stressrelated targets during growth on ethanol, where oxidative stress might arise from respiration in the mitochondria (25). Except for NCE103, the target genes identified have moderate decreases in expression in the cst6⌬ strain (Fig. 3), indicating the presence of other transcriptional activators or repressors controlling the expression of these genes. The number of binding sites, binding strengths, and regulatory effects of Cst6p may increase if the cells are exposed to a higher level stress (e.g., the addition of H 2 O 2 ).
Previous studies showed that Cst6p, Aca1p, and their paralog Sko1p have the same binding sequence in vitro, and the activating effects of Cst6p and Aca1p on basal expression of stressresponsive genes are only observable in the sko1⌬ background (5,22). Thus, a competition model between the three TFs was proposed. Here, we provide in vivo evidence for this model by showing that Cst6p and Sko1p can bind at the same site on target promoters (Fig. 6C). Considering that other TFs, such as Yap1p and Skn7p, also bind directly to many of these genes (Fig. 6B), integrated ChIP-exo and transcriptome analyses of all the TFs involved under various stress conditions (26) would provide new insight into this complex combinatorial regulation. On the other hand, collected microarray data showed that the expression of CST6 itself is hardly affected by environmental changes (27), which suggests that Cst6p is connected to the transcriptional regulatory network in S. cerevisiae mainly through its targets.
Cst6p/Rca1p was identified as a negative regulator of genes involved in hyphal growth and cell adhesion in the human pathogens C. albicans (13) and C. glabrata (14). In C. albicans, the targets of Rca1p identified from transcriptome analysis have a considerable overlap with the targets of Efg1p, which is a major activator of hyphal growth. Here, the functionally conserved homolog of EFG1 in S. cerevisiae, PHD1 (28), was identified as a direct target of Cst6p. This provides a hint that Cst6p/Rca1p may mediate the suppression of hyphal growth by low levels of CO 2 (13,29) through Phd1p/Efg1p, which could be tested in Candida species. While Cst6p appears to activate the expression of PHD1 (Fig. 3), both Cst6p and Phd1p may act as repressors in other yeasts or under specific conditions (6).
Here, we demonstrated that ChIP-exo can be used as an efficient method for mapping the precise locations where TFs bind to the genome. The present work offers three lessons about the understanding of transcriptional regulatory network: (i) it is important to use functionally active conditions to map the binding sites of TFs, (ii) genome-wide mapping of targets can reveal novel functions of TFs and suggest additional conditions for target mapping, and (iii) the transcriptional regulatory network may be far more complex than previously imagined due to the extensive combinatorial regulation and existence of TF cascades. Integrative analysis of data from different types of study is therefore critically needed in order to reconstruct complete transcriptional regulatory networks.

Strains.
All the S. cerevisiae strains used in this study (Table 1) are derived from the uridine auxotrophic strain CEN.PK 113-5D (30), provided by P. Kötter (Frankfurt, Germany). The cst6⌬ strain was constructed by transforming CEN.PK 113-5D with a Kluyveromyces lactis URA3 (KlURA3) marker gene (from vector pWJ1042 [31]) flanked by 45-bp upstream and downstream sequences of the CST6 coding region and screening on synthetic complete medium lacking uracil (SCϪUra; Formedium). Similarly, the prototrophic wild-type strain was constructed by integrating KlURA3 into the ura3-52 locus of CEN.PK 113-5D by homologous recombination. The tagged strain CST6-TAP was constructed by integrating a tagging cassette containing the tandem affinity purification (TAP) tag CBP-ProtA coding sequence and KlURA3 (as described in reference 32) into the CST6 locus, allowing the tag to be fused in-frame to the C-terminal end of Cst6p, connected by a six glycine linker (G6). For constitutive expression of NCE103 in the CST6 deletion background, CST6 in CEN.PK 113-5D was first replaced by a 159-bp fragment downstream from CST6, followed by KlURA3 to obtain strain cst6⌬U, and then the KlURA3 marker gene in cst6⌬U was looped out by the homologous recombination of the two direct-repeat 159-bp fragments on 5-fluoroorotic acid medium to obtain strain cst6⌬L. Plasmid pRS416-NCE103, constructed by inserting the NCE103 gene flanked by the TEF1 promoter and its own terminator between the XbaI and KpnI sites of a centromere plasmid, pRS416 (33), was transformed into cst6⌬L to get the cst6⌬(NCE103) strain. All the primers for cassette constructions and PCR identifications are listed in Table S5 in the supplemental material.
Media and cultivations. Single colonies from fresh agar plates were inoculated into 3 ml YPD medium or minimal medium (as described below) with 2% (wt/vol) glucose in tubes and grown for 12 to 24 h. Cells were harvested by centrifugation and resuspended in sterile water to obtain precultures. For growth in liquid medium (total volume of 20 ml in 100-ml shake flasks), strains were cultivated on a rotary shaker at 200 rpm and 30°C in minimal medium containing (liter Ϫ1 ) 7.5 g (NH 4 ) 2 SO 4 , 14.4 g KH 2 PO 4 , 0.5 g MgSO 4 ·7H 2 O, vitamins and trace metals as used in reference 34, and carbon sources as indicated above. For nutrient supplementation, a 100ϫ stock containing 50 mM palmitic acid and 50 mM stearic acid dissolved in 1:1 (vol/vol) ethanol-Tween 20 was added to get a final concentration of 0.5 mM of each fatty acid, and uracil and L-arginine were added to a final concentration of 20 mg liter Ϫ1 of each compound. For growth assays on agar plates, cell suspensions were adjusted to an optical density at 600 nm (OD 600 ) of 1.0 and diluted four times in 10-fold series. Three microliters of each dilution was spotted onto SCϪUra plates (containing yeast nitrogen base and complete supplement mixture without uracil; Formedium) supplemented with carbon sources as indicated above, and photographs were taken after 2 to 3 days of growth at 30°C.
ChIP-exo. Cells were cultivated in shake flasks in minimal medium with 2% (wt/vol) glucose or 1% (vol/vol) ethanol as the sole carbon source to an OD 600 of 1.5 to 1.8 (mid-log phase). Formaldehyde with a final concentration of 1% (wt/vol) and distilled water were added to the cultures to cross-link protein-DNA complexes with an OD 600 of 1.0 and a total volume of 100 ml. Cross-linking was performed for 12 min at room temperature with shaking and quenched by adding 2.5 M glycine to a final concentration of 125 mM. After 5 min, cells were washed twice with 20 ml cold TBS (10 mM Tris-HCl, 150 mM NaCl, pH 7.5) and frozen with liquid nitrogen. ChIP-exo was performed according to the methods in references 35 and 36, with some modifications. Briefly, cells were disrupted with glass beads on a FastPrep 24 (MP Biomedicals) and the crude cell lysate was sonicated to shear chromatin, using a Branson digital Sonifier 250 (Branson Ultrasonics). After centrifugation, the supernatant, containing chromatin fragments, was applied to IgG Sepharose 6 fast flow   Table S6 in the supplemental material).
To elute and reverse cross-link the bound complexes, TE buffer containing 1% SDS was added to the beads and samples were incubated overnight at 65°C. After protease K (Thermo Scientific) digestion and DNA extraction with phenol-chloroform-isoamyl alcohol (Amresco, USA), the single-strand DNAs were subjected to primer extension using phi29 DNA polymerase (New England Biolabs). The products were given d(A) tails and ligated with the second adapter using the same reagents as in on-bead reactions and then amplified by PCR for 20 to 22 cycles using Phusion high-fidelity DNA polymerase (New England Biolabs). The GeneRead size selection kit (Qiagen) was used to purify DNA before and after the second incorporation of d(A) tails, second adapter ligation, and PCR. The final DNA samples were measured by using a Qubit double-stranded DNA (dsDNA) high-sensitivity (HS) assay kit (Thermo Scientific) and 2200 TapeStation (Agilent Technologies), pooled in equimolar amounts, and sequenced on the NextSeq 500 system (2 ϫ 75 bp, mid-output mode; Illumina). All adapters and primers used in ChIP-exo are listed in Table S6. ChIP-exo sequencing reads were mapped to reference genome assembly R64-2-1 of S. cerevisiae S288C with Bowtie2 (37) using the default settings to generate Sequence Alignment/Map (SAM) files. SAM files were treated with SAMtools (38) option -q 20 to remove low-quality reads and then converted to sorted BAM files. BAM files were trimmed 70 bp from the 3= end using trimBam (http://genome.sph.umich.edu/wiki/BamUtil: _trimBam) to increase the resolution. To identify peaks and compare biological duplicates, the program GEM (39) was used. The noise level was calculated from averaged noise throughout each replicate. Binding events were manually curated in the Integrative Genomics Viewer (40) to remove redundant events representing the same binding and false-positive events with poor peak shape or poor duplication (41). Identification of target genes was done with BEDTools (42) using the closest function with the parameters -io -iu -s or -io -id -S to find any downstream or upstream targets, respectively. Gene targets with a distance of more than 1,200 bp from the binding event were sorted out.
Bioinformatic analyses. The binding site motif was enriched by analyzing the 100-bp sequences upstream and downstream from binding sites using MEME-ChIP 4.10.2 (43). The putative binding sites of Sko1p on promoters were predicted on the YeTFaSCo website (18). The target genes were annotated to gene ontology terms using GO Slim Mapper from SGD (27). The overlaps between Cst6p targets and targets of other TFs were obtained using the "Rank Genes by TF" tool in YEASTRACT (44).
Quantitative real-time PCR. Cells were cultivated to an OD 600 of 1.5 to 1.8 (mid-log phase) in minimal medium with 2% (wt/vol) glucose or 1% (vol/vol) ethanol as the sole carbon source, harvested by centrifugation after mixing with crushed ice, frozen in liquid nitrogen, and then stored at Ϫ80°C. A FastPrep-24 homogenizer (MP Biomedicals) was used to disrupt cells, and total RNA was isolated using the RNeasy minikit (Qiagen). cDNA was synthesized using the QuantiTect reverse transcription kit (Qiagen). For quantitative PCR (qPCR), the DyNAmo flash SYBR green qPCR kit (Thermo Scientific) was used with a reaction mixture volume of 20 l. Previously published primers for CST6 and NCE103 (12) were used, while primers for other genes were designed using IDT's PrimerQuest tool. All the primers in qPCR are listed in Table S5 in the supplemental material. The PCR was performed on the Mx3005P qPCR system (Agilent Technologies). The thermal program consisted of an initial denaturation of 15 min at 95°C, 40 cycles of 10 s at 95°C and 30 s at 60°C, and a final segment of 1 min at 95°C and 30 s at 55°C followed by a ramp up to 30 s at 95°C (for the dissociation curve). ACT1 was used as the reference gene for gene expression level comparison using the cycle threshold (⌬⌬C T ) method (45). Statistical significance tests were done with a one-tailed equal variance (homoscedastic) t test in Microsoft Excel 2013.
Sequence data accession number. The ChIP-exo data have been deposited in the Gene Expression Omnibus database under the accession number GSE76154.