Activator-independent transcription of Snf1-dependent genes in mutants lacking histone tails

Summary

Transcriptional regulation of Snf1-dependent genes occurs in part by histone-acetylation-dependent binding of the transcription factor Adr1. Analysis of previously published microarray data indicated unscheduled transcription of a large number of Snf1- and Adr1-dependent genes when either the histone H3 or H4 tail was deleted. Quantitative real time PCR confirmed that the tails were important to preserve stringent transcriptional repression of Snf1-dependent genes when glucose was present. The absence of the tails allowed Adr1 and RNA Polymerase II to bind promoters in normally inhibitory conditions. The promoters escaped glucose repression to a limited extent and the weak constitutive ADH2 transcription induced by deletion of the histone tails was transcription factor- and Snf1-independent. These effects were apparently due to a permissive chromatin structure that allowed transcription in the absence of repression mediated by the histone tails. Deleting REG1, and thus activating Snf1 in the H3 tail mutant enhanced transcription in repressing conditions, indicating that Snf1 and the H3 tail influence transcription independently. Deleting REG1 in the histone H4 tail mutant appeared to be lethal, even in the absence of Snf1, suggesting that Reg1 and the H4 tail have redundant functions that are important for cell viability.

Introduction

Chromatin structure poses a barrier to transcription by RNA Polymerase II (RNA Pol II). Eukaryotic organisms have evolved mechanisms that use this barrier to transcription to regulate gene expression (Wu & Grunstein, 2000, Jenuwein & Allis, 2001). The structure of chromatin is modulated by post-translational modifications of histones, the highly conserved proteins that constitute the central structure of the nucleosome core particle. Many proteins interact with histones and induce these modifications that in turn affect the dynamic properties of nucleosome core particles (Wu & Grunstein, 2000, Saha et al., 2006). These proteins include histone acetyl transferases (HATs), histone deacetylases (HDACs), and ATP-dependent chromatin remodeling complexes (CRCs) and target primarily the unstructured N-terminal tails of the histones (Millar & Grunstein, 2006). HATs and HDACs regulate the charge of the tails to create a dynamic equilibrium between the presence and absence of acetyl groups on lysine residues. The modulation of charge influences the interactions between histones and DNA that in turn can change the location of the nucleosomes on the DNA (Zheng & Hayes, 2003). The tails interact in vitro with the DNA to enhance the stability of the nucleosome at a particular translational position (Yang et al., 2007). This translational movement can affect the accessibility of specific binding sites to transcription factors (Brower-Toland et al., 2005). For example, specific patterns of acetylation in the histone tails are recognized by CRCs that alter the translational position of the nucleosome core particle (Saha et al., 2006), making a previously occluded binding site accessible to a DNA-binding protein or vice versa. Both DNA sequence and chromatin binding proteins can influence the position of nucleosomes on the DNA (discussed in (Caserta et al., 2009, Travers et al., 2009)).

We analyzed the consequences of deleting or mutating the histone tails for gene expression and promoter architecture of glucose-repressed genes in Saccharomyces cerevisiae (S. cerevisiae). The genes analyzed were those whose derepression in the absence of glucose is mediated by the transcription factors Adr1 and Cat8, which act downstream of the Snf1 kinase, the yeast homolog of the mammalian AMP-activated protein kinase (AMPK). Adr1 activates more than 100 genes whose encoded enzymes are required for non-fermentative metabolic pathways (Young et al., 2003). In the presence of glucose the activity of Adr1 is inhibited by phosphorylation of the DNA binding and regulatory domains (Kacherovsky et al., 2008; Ratnakumar et al., 2009; Ratnakumar & Young, 2010). In the absence of glucose, Snf1 promotes the dephosphorylation and activation of Adr1. The yeast 14-3-3 or Bmh proteins bind to the regulatory domain when S230 is phosphorylated and this inhibits a nearby activation domain (Parua et al., 2010). By promoting Ser230 dephosphorylation Snf1 counteracts the inhibitory role of Bmh. At the ADH2 promoter (ADH2prm), Adr1 and Cat8 cooperate to bind DNA, recruit coactivators, remodel chromatin and activate transcription (Tachibana et al., 2005, Biddick et al., 2008). At the promoters of genes encoding glycolytic and gluconeogenic enzymes, such as the FBP1 promoter (FBP1prm), only Cat8 is needed for close to maximal derepression. A HAT-dependent increase in histone H3- and H4-tail acetylation facilitates both chromatin remodeling and transcriptional activation shortly after glucose is exhausted in a set of co-regulated promoters including ADH2prm (Agricola et al., 2006).

Surprisingly, Adr1 and Cat8 can bind in repressing conditions in a mutant lacking two HATs, Hda1 and Rpd3 (Δhdac) (Tachibana et al., 2007, Verdone et al., 2002). Activator binding in the Δhdac mutant in repressing conditions is associated with HAT-dependent hyperacetylation of promoter nucleosomes similar to that observed in derepressing conditions (Verdone et al., 2002), suggesting that histone hypoacetylation inhibits activator binding in repressing conditions. Although activator binding and promoter remodeling take place, and a stable pre-initiation complex (PIC) is recruited in the Δhdac mutant in repressing conditions, there is very little transcription (Tachibana et al., 2007, Verdone et al., 2002). The inactive “poised” PIC contains SAGA (Spt-Ada-Gcn5 Acetyltransferase complex), Mediator, Swi/Snf (Swi/Snf chromatin remodeling complex), and RNA Pol II partially phosphorylated on Ser5 of the carboxy-terminal domain (Tachibana et al., 2007). The inactive “poised” complex can be activated in the Δhdac mutant in repressing conditions by activating Snf1 and by activating Adr1 (Tachibana et al., 2007). Thus PIC recruitment is necessary but not sufficient for efficient transcription.

Although increased acetylation of the histone tails is associated with increased activator binding in the Δhdac mutant, it is possible that enhanced acetylation of other proteins contributes to or is responsible for activator binding and PIC formation. To test this possibility we analyzed activator and RNA Pol II binding, gene expression and chromatin remodeling in mutants lacking the histone H3 and H4 N-terminal tails or containing Gln substitutions of the Lys residues in the tails. Our analysis shows that the absence of the H3- or H4-tails leads to specific up-regulation of Snf1-dependent genes. Adr1 binding sites became accessible at the ADH2 promoter when the histone tails were altered and well-positioned nucleosomes were lost or became unstable, allowing the binding of Adr1. However, the low level of constitutive (glucose-resistant) ADH2 expression induced by these alterations was independent of the presence of Adr1 and Cat8 and independent of Snf1, indicating that RNA Pol II is able to actively engage Adr1-dependent promoters in the absence of an activator if the chromatin structure is permissive.

Results

Deletion of either H3- or H4-tail causes constitutive expression of glucose-repressed genes

Since there was previous evidence for a role of the histone H3- and H4-tails in glucose regulation (Agricola et al., 2006, Tachibana et al., 2007, Verdone et al., 2002), we looked for the effect on genes whose expression is dependent on either the transcription factors Adr1 and Cat8 or the upstream kinase Snf1 in the dataset of the first transcriptome analysis published for mutants deleted for these histone tails (Sabet et al., 2003). This genome-wide expression study showed up-regulation of a significant number of genes when the H3-tail was deleted, suggesting that the H3-tail has a global role in repression in glucose-rich media (Table 1). In contrast, deletion of the H4-tail either increased or decreased expression of a very similar proportion of genes.

Table 1.

Genes whose expression was affected more than two-fold by deletion of the histone tails in microarray analysis

Subset (number of genes)	up-regulated genes		down-regulated genes
	Δh3t	Δh4t	Δh3t	Δh4t
Whole genome	590 (10%)	273 (5%)	47 (0.8%)	387 (6%)
Adr1-dependent (105)a	50 (48%)	26 (25%)	1 (1%)	6 (6%)
Cat8-dependent (247)a	61 (25%)	38 (15%)	10 (4%)	15 (6%)
Snf1-dependent (430)a	153 (36%)	111 (26%)	16 (4%)	22 (5%)
Adr1 direct targets (28)b	18 (64%)	5 (18%)	0	0
Cat8 direct targets (23)b	14 (61%)	1 (4%)	2 (9%)	2 (9%)
Adr1 & Cat8 direct targets (14)b	9 (64%)	2 (14%)	0	1 (7%)
Unfolded protein response (169)c	28 (17%)	29 (17%)	5 (3%)	3 (2%)

^aThe list of genes for the different subsets was obtained from (Young et al., 2003).

^bThe list of genes for the different subsets was obtained from (Tachibana et al., 2005).

^cThe list of genes for the different subsets was obtained from (Travers et al., 2000).

We analyzed the data corresponding to the specific subset of genes whose derepression depends on Adr1, Cat8, or Snf1 (Young et al., 2003) and found a significant difference between the observed number of genes affected by either tail deletion and the expected distribution if the different subsets followed the same trend observed genome-wide (Table 1 and Table S2). Specifically, in each of these subsets there was a significantly higher proportion of genes whose expression was up-regulated by the deletion of either the H3- or H4-tail than expected. We also looked for the effect of the histone tail deletions in a subset of genes whose regulation is totally independent of the carbon source in order to estimate the specificity of the contribution of the tail deletions to the Snf1-dependent genes. The genes chosen for this subset are regulated by the unfolded protein response and are up-regulated when the endoplasmic reticulum is subjected to acute stress (Travers et al., 2000). This subset was also up-regulated by both tail deletions in yeast growing in the absence of stress (Table 1). However, the proportion of genes in this subset that was affected was lower than the proportion of glucose-regulated genes. When we only considered those genes in which direct binding and regulation of transcription by Adr1, Cat8, or both factors have been confirmed (Tachibana et al., 2005), the array data of Sabet (Sabet et al., 2003) showed a very high proportion of genes up-regulated by deletion of the H3-tail. It also showed a significant but lower proportion of genes up-regulated by deletion of the H4-tail only when Adr1 participates in their regulation (Table 1).

The data reported above suggest that both tails are specifically involved in the glucose-repression pathway controlled by the Snf1 kinase and the downstream transcriptional activators Adr1 and Cat8. We quantified the mRNA levels of several Snf1-dependent genes in mutants lacking either histone H3- or H4-tail (Δh3t and Δh4t, respectively) by quantitative real-time reverse transcriptase PCR (qPCR). We confirmed that both tail deletions cause the low-level, constitutive activation of ADH2, ACS1, ADY2, FBP1, MLS1, and perhaps MDH2, which are representatives of Snf1-dependent genes. Fig. 1A shows their expression in repressing growth conditions relative to a strain containing wildtype (wt) histone genes. The increase in ADH2 mRNA levels in the tail mutants is about 6-fold. Derepression results in an ~400-fold increase in ADH2 mRNA levels in a strain with wt histone tails (Fig. 1B). Thus, the ADH2 mRNA levels in the Δh3t mutant represent only about 1.5% of the derepressed level. The extent of relief from repression of the other genes is also minor. MDH2 repression was not affected by the tail deletions in the array data and was the least affected as determined by qPCR. Derepressed expression of these genes was reduced in the tail mutants. Fig. 1B shows their derepressed expression relative to wt in repressing conditions. ADH2 was the least affected (two-fold reduced) and the strongly Cat8-dependent genes FBP1 and MLS1 were more strongly affected (about four-fold reduced). In summary deletion of the H3 and H4 tails alters regulation of Snf1-dependent genes in two ways. The genes are released to a limited extent from the stringent glucose repression that normally prevails and their maximal derepression is reduced.

Figure 1. Histone tail deletions alter transcriptional regulation of Adr1- and Cat8 -dependent genes.

mRNA levels were determined by qPCR, normalized to ACT1 mRNA levels and expressed as fold-increase over wt, repressed. A. Repressed growth conditions. The dotted line represents the level of wt expression (1.0). B. Derepressed growth conditions. Error bars represent sd of three biological replicates. Wt, Δh3t and Δh4t indicate wt H3-H4 and deletions of amino acids 4-30 in H3 and amino acids 4-28 in H4, carried on plasmids in strains JJY200, JJY430, and JJY428, respectively. C. Repressed mRNA levels measured in strains deleted for the H3-H4 genes and carrying plasmids for expression of histone tails in which all of the lysines have been replaced by glutamines in H3 and H4 (plasmids pRM200 HHT2KtoQ-HHF2 and pRM200 HHT2-HHF2KtoQ, respectively). The dotted line represents the level of wt expression (1.0).

Deletion of the histone H3 and H4 tails allows Adr1, but not Cat8, to bind in repressing conditions

A repressive chromatin structure due to hypoacetylated histones appears to prevent Adr1 from binding in the presence of glucose (Verdone et al., 2002, Tachibana et al., 2007). When the HDACs Rpd3 and Hda1 are absent, Adr1 and Cat8 can bind in the presence of glucose but transcription is still strongly glucose-repressed, suggesting that hyperacetylation of the histone tails allows Adr1 and Cat8 binding but post-binding step(s) in glucose repression are still operative. Hypoacetylated histone tails interact with neighboring DNA and increase the stability of a specific nucleosome position (Yang et al., 2007). If hyperacetylation of the histone tails, and not acetylation of some other substrate allows Adr1 to bind in repressing conditions, deletion of the tails might also create a permissive chromatin structure.

Adr1 binding in the tail mutants was studied by chromatin immunoprecipitation (ChIP) after epitope tagging Adr1 at its C-terminus. As expected, Adr1 binding was not detectable in the wildtype (wt) parent strain grown in repressing conditions and could be readily detected after derepression (Fig. 2A). However, in a Δh3t mutant Adr1 was bound to several promoters (ADH2, ADY2, FBP1; Fig. 2A) at similar levels in repressing and derepressing conditions, a dramatic difference compared to its binding in cells with wt histones where binding is restricted to derepressing conditions. Moreover, Adr1 binding in the Δh3t mutant in repressing conditions was at a level similar to its binding in cells with wt histones in derepressing conditions. A similar result was observed with Δh4t (data not shown). These results indicate that the restriction on Adr1 binding in repressing conditions is absent when the histone H3 or H4 tails are missing, as was observed in a mutant in which the Rpd3 and Hda1 deaectylases are deleted (Verdone et al., 2002, Tachibana et al., 2007),

Figure 2. Binding and activity of transcription components in histone tail mutants.

A. ChIP for Adr1-HA at the ADH2, ADY2 and FBP1 promoters in wt or Δh3t strains (JJY200A-HA and JJY430A-HA, respectively; Table 3) grown in repressing (R) or derepressing conditions for 6h (DR). B. ChIP for Cat8-TAP-HA at ADH2prm and FBP1prm in wt or Δh3t strains (LLTY79 and LLTY80, respectively; Table 3) grown in repressed or derepressed conditions for 6h. C. ADH2 and FBP1 mRNA levels in wt and tail mutants (Δh3t and Δh4t) in combination with either Δsnf1 or Δadr1Δcat8 (see Table 3 for strains). Cells were grown in repressing conditions and mRNA levels were determined as in Fig. 1. D. ChIP for Rpb1 at ADH2, ADY2 and FBP1 promoters. Cells were grown in repressing conditions. Error bars represent the sd of three biological replicates in each panel.

These results suggest that the H3- and H4-tails restrict the accessibility of the Adr1-binding site during glucose repression. Since Adr1 binding also occurs in repressive growth conditions in the Δhdac mutant (Verdone et al., 2002, Tachibana et al., 2007), hypoacetylation of the H3- and H4-tails may contribute to glucose repression. In agreement with this interpretation, substitution of all lysines by glutamines in either tail, which mimics the acetylated state of lysine, also led to weak constitutive activation of ADH2 at a level similar to that observed in the tail-less mutants (Fig. 1C). Expression of ADY2, ACS1, FBP1, and MLS1 was less affected by the K-to-Q substitutions than by deletion of the tails. This apparent promoter specificity may indicate that transcription factor binding to the ADH2 promoter is more sensitive to histone acetylation than the other promoters tested.

Several genes dependent on Cat8 for their expression were weakly activated in repressing conditions in the tail mutants (Table 1, Table S2, and Fig. 1A). Therefore, we asked whether deletion of the H3-tail also allowed Cat8 binding in repressing conditions. Cat8 was epitope-tagged at its C-terminus and binding was measured by ChIP. In derepressing conditions Cat8 was detected at the promoters of ADH2 and FBP1 as expected (Tachibana et al., 2005). However, in repressing conditions we found that, unlike for Adr1, the ΔH3t deletion did not allow Cat8 binding to any of the promoters tested at a level above background (Fig. 2B).

Low-level, constitutive ADH2 activation caused by the tail deletion is independent of the transcription factors Adr1 and Cat8 and the Snf1 kinase

The high level of Adr1 binding in repressing conditions in Δh3t was surprising in light of the low level of gene expression. A similar phenomenon, a high level of promoter occupancy but very weak transcription, was observed in a Δhdac mutant: (Verdone et al., 2002, Tachibana et al., 2007). An important difference between the two situations is the high level of Cat8 binding in the Δhdac mutant (Tachibana et al., 2007) but lack of detectable Cat8 binding in the histone tail mutants (Fig. 2B). Since ADH2 expression is dependent on both Adr1 and Cat8, the lack of Cat8 binding to the ADH2prm in the tail mutants in repressing conditions might explain the relatively weak transcription. To determine whether transcription in the tail mutants in repressing conditions is transcription factor-dependent, we deleted both Adr1 and Cat8 in a strain lacking the H3-tail. Surprisingly, ADH2 transcription in the H3-tail mutant in repressing conditions was unaffected by the deletions and FBP1 transcription was only reduced about 30% (Fig. 2C).

Binding and activity of both Adr1 and Cat8 in derepressing conditions are dependent on Snf1 (Schuller, 2003, Young et al., 2003, Young et al., 2002) and their binding is Snf1-dependent in repressing conditions in the Δhdac mutant (Tachibana et al., 2007). Although Snf1 is primarily inactive in the presence of glucose (Hardie et al., 1998, Hong & Carlson, 2007, Rubenstein et al., 2008, Orlova et al., 2008), the Snf1-dependent binding of Adr1 and Cat8 in the Δhdac mutant suggests that loss of HDAC activity may constitute a stress that activates Snf1 (Hong & Carlson, 2007). Surprisingly, expression of ADH2 in the H3-and H4-tail mutants in repressing conditions was independent of Snf1 (Fig. 2C). In contrast FBP1 expression was reduced about two-fold in the tail mutants in the absence of Snf1 but it was still significantly higher than in a strain with wt histones. These results are consistent with those showing that ADH2 expression is independent of the transcription factors Adr1 and Cat8 (Fig. 2C), both of which are Snf1-dependent for activity. When the tail mutants were derepressed by depleting glucose from the media, the increase in ADH2 activity was Snf1-dependent (data not shown), indicating that the major pathway of derepression, which is Snf1-dependent, is still functional in the absence of the H3 tail.

In the Δhdac mutant Adr1 binding is accompanied by recruitment of an inactive PIC containing RNA Pol II (Tachibana et al., 2007). To determine if similar recruitment of RNA Pol II occurs in the Δh3t mutant we did ChIP for the largest subunit of RNA Pol II, Rpb1, to measure promoter occupancy at ADH2, ADY2, and FBP1 (Fig. 2D). The level of RNA Pol II in the H3-tail mutant in repressing conditions was significantly higher than the background binding in the wt strain in repressing conditions at the promoters of ADH2 and ADY2. Upon derepression RNA Pol II levels increased in the tail mutants and reached a level similar to that in the wt parent at the ADH2prm but not at the promoter of ADY2, in agreement with the lower level of derepression of ADY2 (Fig. 1B). The occupancy by RNA Pol II at the ADH2prm in repressing conditions was higher than would be predicted based on the low level of ADH2 mRNA (Fig. 2A). This observation suggests that RNA Pol II is bound but that some step in transcription is still repressed, as it is in the Δhdac mutant (Tachibana et al., 2007). In contrast, the tail deletion did not lead to a detectable increase in RNA Pol II binding at the FBP1prm in repressing conditions despite the observation of significant transcription. The reason for this difference might be partial stabilization of RNA Pol II binding or PIC formation because of the presence of Adr1 in the promoters of ADH2 and ADY2 but the absence of the major activator of FBP1, Cat8, at the FBP1prm in the tail mutant (Fig. 2B).

In summary, our results suggest that although Adr1 is bound to several promoters in repressing conditions in the tail mutants, it is not necessary to recruit RNA Pol II. Rather, transcription occurred independently of Adr1, Cat8, and Snf1. However, RNA Pol II at the ADH2prm in repressing conditions appeared to be mostly inactive as indicated by the low level of transcription (Fig. 1A) compared to RNA Pol II occupancy in the H3-tail mutant (Fig. 2D), suggesting that some coactivator is absent or modification of the complex cannot occur in the presence of glucose, similar to the situation in an Δhdac mutant (Tachibana et al., 2007). However, the Snf1-independent Adr1 binding in the tail mutants is in contrast with the situation in the Δhdac mutant in which Adr1 binding is Snf1-dependent (Tachibana et al., 2007). This difference suggests that Snf1 has a role in Adr1 binding that is not essential when the tails are deleted. Another important difference is the extent of derepresion in the two situations. Derepression occurs normally in the Δhdac strain but is severely restricted in the histone tail mutants. Thus, the tails have an important positive as well as negative role in Adr1-dependent gene regulation.

Snf1 and the histone tails affect different aspects of ADH2 repression

The low level of ADH2 transcription in the histone tail mutants despite the presence of Adr1 and RNA pol II at the promoter suggests that a major step of glucose repression is still functional when either histone tail is missing. A major pathway of repression of Snf1-dependent genes is mediated by the essential protein phosphatase Glc7. The Glc7 phosphatase activity relevant to glucose repression can be eliminated by deleting Reg1, a non-essential targeting subunit. In the absence of Glc7 phosphatase activity, Snf1 is constitutively phosphorylated and active in the presence of glucose (Sanz et al., 2000) although it may not be fully activated (Orlova et al., 2008). We deleted REG1 in strains lacking the H3- or H4-tail to assess genetically the interaction of the tail mutants with Snf1. As expected (Dombek et al., 1993), deleting REG1 in a strain with intact H3 resulted in significant ADH II activity in repressing conditions as detected by in situ staining for ADH activity after electrophoresis of cell extracts (Fig. 3). The constitutive ADH II activity increased dramatically when the H3-tail was deleted in the Δreg1mutant (Fig. 3), suggesting that the H3-tail has a role in repressing transcription that cannot be overcome by activation of Snf1. Synergistic relief from glucose repression when Snf1 was activated by deleting REG1 in the absence of the H3 tail was confirmed by β-galactosidase assays using a lacZ reporter gene under the control of the ADH2prm (Fig. 3, bottom). Similar results were obtained when ADH2 expression was measured by qPCR in strains with a REG1 deletion in the presence and absence of the H3-tail (Table 2). The enhanced ADH2 expression caused by deleting REG1 in the Δh3t mutant required Snf1 because the expression was reduced to that of the Δh3t single mutant when SNF1 was deleted (Table 2). These results are similar to those in the Δhdac mutant where the Δreg1Δhdac combination synergistically activated ADH2 transcription (Tachibana et al., 2007).

Figure 3. Synergistic effect of Δh3t and Δreg1 deletions on the derepression of ADH2.

In-gel ADH assay of extracts from yeast cells growing in repressing synthetic medium lacking uracil. ADHI, glucose-induced ADH isozyme derived from ADH1; ADHII, glucose-repressed ADH isozyme derived from ADH2. Indicated at the bottom of each lane □□□□□□β-galactosidase activities expressed in Miller units (average and sd of three biological replicates) of a lacZ reporter gene driven by the ADH2 promoter on plasmid pBGM18 carried in each strain. Strains and plasmids are listed in Table 3.

Table 2.

Effect of combining REG1 and histone tail deletions on expression of Snf1-dependent genes

Relevant genotype1			Expression2
REG1	SNF1	Histones	ADH2	FBP1
+	+	wt	1.0	1.0
−	+	wt	15	37
−	+	Δ h3t	23	29
−	+	H3-S10A	16	24
−	−	wt	1.3	1.1
−	−	Δ h3t	3.7	2.4

¹Strains are listed in Table 3. Wt is JJY200; Δreg1 is JJY200r ; Δreg1 Δh3t is JJY430r; Δreg1 H3-S10A is JJYS10Ar; Δreg1 Δsnf1 is JJY200rs; Δreg1 Δsnf1 Δh3t is JJY430rs.

²Yeast cultures were grown in repressing-rich medium (YPD). ADH2 and FBP1 mRNA levels were determined by qPCR, normalized to ACT1, and are expressed relative to expression of the same gene in wt cells grown in YPD. Values are the average of three biological replicates and the standard deviation was under 25% of the average.

We also observed that FBP1 expression was enhanced in repressing conditions when REG1 was deleted (Table 2), suggesting that activation of Snf1 activates Cat8 which in turn activates FBP1. However, combining activation of Snf1 (by deleting REG1) with deletion of the H3-tail did not increase expression of FBP1 suggesting that activation of Snf1 can fully overcome the restriction to transcriptional activation imposed by the presence of the H3-tail. The contrast between the effect of deleting REG1 in the presence and absence of the H3 tail at these two promoters suggests that the histone tails may act differently at different promoters.

Phosphorylation of H3-Ser10 by Snf1 and subsequent K14 acetylation by Gcn5 has been proposed to relieve repression of INO1 expression (Lo et al., 2001). To test whether this mechanism could be involved in the activation of ADH2 expression we measured ADHII activity and the activity of an ADH2/lacZ reporter gene in a strain in which the H3-S10A allele was the only copy of histone H3. The activity of ADH2 was unaffected as determined by both assays (data not shown). We also combined a deletion of REG1 with a histone H3-S10A allele to test the possibility that the S10A allele in the presence of active Snf1 might allow Adr1 binding and transcriptional activation as in the Δhdac mutant (Tachibana et al., 2007). However, ADH2 was expressed to the same extent in the Δreg1 mutants carrying either a wt-H3 or the H3-S10A mutant histone (Table 2) suggesting that Snf1 promotes transcriptional activation of ADH2 independently of H3-Ser10 phosphorylation.

To test the possibility that the histone H4 tail contains a target of Snf1 we attempted to delete REG1 in a strain with a Δh4t allele. However, we were unable to delete REG1 in this strain unless it also carried a plasmid with an intact H4 (HHF1) gene (Fig. 4 and data not shown). This result suggests that activating Snf1 in a strain lacking the H4-tail causes a severe growth defect. In support of this conclusion, we observed that the H4-tail mutant was defective for growth on glycerol, a carbon source that activates Snf1. The H3-tail mutant was not glycerol-negative (data not shown) and the slow-growth phenotype of a Δreg1Δh3t double mutant was cured by deleting Snf1 (Fig. 4). In contrast, the apparent growth defect of a Δreg1Δh4t double mutant was not cured by deleting SNF1 (Fig. 4). These results indicate that the growth defect caused by deleting REG1 in the H3-tail mutant acts through Snf1, but in the case of the H4-tail mutant Reg1 may have a Snf1-independent function.

Figure 4. Interaction between activation of Snf1 on glucose and deletion of the H3- and H4-N terminal tails.

Yeast cells with the indicated genotypes (see Table 3 for strains) were grown for ten generations in complete repressing medium containing uracil and then plated at various dilutions on either synthetic medium with glucose (SCD) or the same medium with FOA in order to test their ability to lose the URA3-plasmid carrying wt-histones (plasmid pMS329). The 10X or 1X dilutions are shown. The higher dilutions showed that approximately the same number of cells were plated from each strain.

Activator-independent remodeling of ADH2prm

A significant perturbation of the nucleosome architecture of the ADH2prm in the Δh3t mutant might explain binding of Adr1 and RNA Pol II in repressing conditions, as well as the weak constitutive activation of transcription. The ADH2prm has three well-positioned nucleosomes as indicated by Micrococcal nuclease (MNase) mapping (Di Mauro et al., 2002, Verdone et al., 1996, Tachibana et al., 2007) (see Fig. 6A and Fig. S1 for a depiction of the ADH2 promoter nucleosomes). The promoter nucleosomes are remodeled in an Adr1- and Cat8-dependent, but transcription-independent manner upon derepression (Verdone et al., 1997, Biddick et al., 2008b) including a small 3′-translation of the N(−1) TATA-box containing nucleosome (Di Mauro et al., 2002, Verdone et al., 1996, Tachibana et al., 2007). A supercoiling assay (Simpson et al., 1985) using the ADH2prm on a small episomal plasmid indicated that on average about one nucleosome is lost upon derepression in a wt strain and a similar loss occurred in a Δhdac mutant in repressing conditions (Tachibana et al., 2007).

Figure 6. Chromatin architecture of the ADH2prm and the effect of histone-tail deletions and activation of Snf1 in repressing conditions.

A. The location of ADH2prm nucleosomes in repressed cells. The white box indicates the location of the TATA box; white circles, Adr1 binding sites; black dot, Cat8 binding site. The dashes show the predicted positions of nucleosomes based exclusively on DNA sequence (Segal et al., 2006). B. Nucleosome scanning assays of MX1-4C (wt histones) in repressing and derepressing conditions compared to the Δh3t and Δh4t strains in repressing conditions. A NuSA of a strain containing deletions of ADR1 and CAT8 is shown for comparison. C. Nucleosome scanning assays of Δreg1 in repressing conditions compared to its wt parent W303 in repressing and derepressing conditions. Relative protection was calculated by normalization to a CEN3 amplicon and then to the wt repressed control sample (see Experimental Procedures for details). Two-to-five biological replicates were performed for each strain. The average standard deviation was less than 25%. See Fig. S1 for a detailed description of the sequence of the promoter and the nucleosome locations.

To address the influence of histone tails on the nucleosome density at the ADH2prm and to compare this situation with the Δhdac mutation, we performed a supercoiling assay using DNA prepared from the Δh3t mutant and its parent strain, each carrying an ADH2prm-containing minichromosome (Fig. 5A and B). There was a decrease in nucleosome density of the ADH2prm-containing minicircle upon derepression of the parent strain, albeit a smaller change than was seen in the strain studied previously (Fig. 5B inset). Reminiscent of the previous results with the Δhdac mutant, the supercoiled plasmid isolated from the Δh3t strain grown in repressing conditions had about the same nucleosome density as the plasmid from the wt strain after derepression, and even lower nucleosome density in derepressed conditions. Since Adr1 can bind in both the Δh3t (Fig. 2) and Δhdac mutants in repressing conditions (Verdone et al., 2002), we asked whether the change in nucleosome density in the Δhdac strain was Adr1-dependent. In the strain with wt histones, the change in nucleosome density that accompanied derepression was Adr1-dependent (Fig. 5C and D). In contrast, the repressed Δadr1Δhdac strain had a lower nucleosome density in repressing conditions and further loss occurred upon derepression (Fig. 5E and F), suggesting that if the histone tails are hyperacetylated, Adr1 is not necessary to recruit activities that remodel the nucleosomes at the promoter. Thus, in both the Δh3t and Δhdac mutant nucleosome eviction or alteration apparently occurs without requiring an activator.

Figure 5. Loss of the histone H3 tail is sufficient for chromatin remodeling on glucose at the ADH2prm.

Supercoiling analysis of a plasmid with the ADH2 promoter (pLLTY1) was performed as described in Experimental Procedures. A. Samples from W303-CH1a (LLYT4), MX1-4C (wt; JJY200) strain or Δh3t (JJY430) strain were repressed (R) or derepressed for 6h (DR). B. Quantitation of supercoiling assays in a gel similar to that shown in A. The topoisomer distributions were quantified using phosphorimaging. The data are presented as the relative intensity for each of eight bands. Error bars represent the s.d. of two independent experiments in this panel and in panels D and F. A shift to the right represents a decrease in overall nucleosome density. The inset shows the results from the W303 strain, repressed (R, dashed line) and derepressed for 4h (DR, solid line). C, D. Samples from wt (LLTY4) or Δadr1 (LLTY6) grown in either repressing (R) or depressing conditions for 4h (DR). E, F. Samples from wt (W303; LLTY4) or Δadr1Δhdac (LLTY111) grown in either repressing (R) or depressing conditions for 6h (DR).

We also analyzed the ADH2prm using a nucleosome-scanning assay (NuSA) because both position and relative occupancy can be determined by this technique (Sekinger et al., 2005). We first used NuSA combined with ChIP (ChIP-NuSA) to analyze the histone composition of individual promoter nucleosomes (Sekinger et al., 2005). ChIP-NuSA for histones H3 (Fig. S2), H4, H2B, and the variant histone Htz1 confirmed that the N(−1), N(−2), and N(+1) nucleosomes contain these histones as expected (data not shown). When the ADH2prm in the Δh3t and Δh4t strains on glucose was analyzed, the results indicated that it was as remodeled in the tail mutants as in the derepressed parent strain (Fig. 6B) as judged by the decrease in relative nucleosome occupancy (decrease in resistance to MNase digestion) and altered positions of the N(−1) and N(−2) nucleosomes. As a control we performed a NuSA using a derepressed culture of a strain lacking both ADR1 and CAT8. This NuSA pattern matched that of the wt strain grown in repressing conditions, confirming that the remodeling is transcription-factor-dependent when the histone tails are present. These results confirm those obtained with the supercoiling assay (Fig. 5) showing that when the H3 or H4 tail was deleted, complete remodeling was observed on glucose. In a Δreg1 strain the loss of protection of DNA bound by the nucleosomes in repressing conditions was intermediate between that observed in repressing and derepressing conditions in a wt strain (Fig. 6C). In particular the TATA-containing nucleosome N(−1) did not experience the same increase in accessibility in repressing conditions as was observed when the cells were derepressed. This might explain the inability of a REG1 deletion to induce full derepression of ADH2 on glucose (Table 2). These results suggest that additional derepression-dependent remodeling may be needed for full activation of the ADH2prm.

A small peak of MNase-resistant DNA is usually observed in the nucleosome-free (nfr) region between the N(−1) and N(−2) nucleosomes (Fig. 6B, and (Tachibana et al., 2007, Biddick et al., 2008)). The ChIP-NuSA results (Fig. S2 and data not shown) indicate that this DNA is protected by histones. Its MNase-resistance is intermediate between that of a nucleosome and free DNA and it appears to contain a shorter stretch of DNA than is contained within a nucleosome (data not shown). It could represent an altered nucleosome either alone or complexed with another protein as was recently reported for a nucleosome-RSC complex in the GAL1,10 promoter (Floer et al., 2010). Its presence is unaffected by deletion of the H3 and H4 tails.

Discussion

These results demonstrate the importance of the histone tails in maintaining a repressive state of chromatin at Adr1-dependent promoters. The similar phenotypes of the histone tail mutants and the Δhdac mutant with regard to Adr1 binding and recruitment of a mostly inactive RNA pol II suggest that hyperacetylation of the histone tails and not increased acetylation of another protein most likely explains activator binding and PIC recruitment in the Δhdac mutant.

Although the transcriptional activator Adr1 and RNA Pol II were able to gain access to glucose-repressed promoters when the histone tails were altered (Fig. 2), they were incapable of inducing the high level of transcription that is observed when glucose is absent from the media, showing that the mechanism of glucose repression operates at several levels including a step after RNA Pol II has been recruited (Tachibana et al., 2007). When the H3 or H4 tail was deleted, the binding site occluded by the presence of nucleosome N(−1) apparently became accessible because Adr1 occupied the ADH2prm in the presence of glucose to the same extent that it binds in a derepressed wt strain (Fig. 2A and data not shown). However, although the chromatin state of the ADH2prm in the repressed Δh3t and Δh4t strains resembled that of a derepressed wt strain (Fig. 6B) and the Adr1 binding site appeared to be fully occupied (Fig. 2A), ADH2 transcript levels reached only 1.5% of the level attained upon derepression (Fig. 1). Surprisingly, we found that this weak transcriptional activity was independent of Adr1, Cat8, and Snf1 (Fig. 2C). Dispensability of transcriptional activators, and Adr1 in particular, has been observed when histone reassembly is altered by mutation of Spt6, a histone H3-H4 chaperone (Adkins & Tyler, 2006). Both defective reassembly, and loss of the H3 and H4 tails appears to lead to an altered chromatin state that allows activator-independent RNA Pol II binding and transcription. In the former case, a high level of transcription may be maintained after inactivating Spt6 because Adr1 was previously activated by derepression.

The permissive chromatin structure at the ADH2prm created by deletion of the H3 tail led to a transcriptional state that resembles the state of genes that are occupied by RNA Pol II but remain inactive (Zeitlinger et al., 2007, Ivaldi et al., 2007, Tachibana et al., 2007). In these complexes RNA Pol II appears to be paused downstream of the start site. In the tail deletion mutants grown in repressing conditions, we observed Adr1-independent transcription at a lower level than would be predicted by the level of RNA Pol II bound to the promoter (Fig. 2D). The inability of Adr1 to more effectively activate transcription was not due solely to the absence of the histone tail because a high level of ADH2 mRNA was observed in the Δh3t mutant when glucose was removed from the media (Fig. 1B) and an ADH2/lacZ reporter showed a high level of Snf1-dependent derepression in the Δh3t mutant strain (data not shown). In a Δhdac mutant, both Adr1 and Cat8 are bound to the ADH2prm, have recruited a PIC consisting of coactivators and RNA Pol II, but are unable to activate transcription (Tachibana et al., 2007). This poised and inactive PIC can be activated by activating Snf1, or by activating Adr1. Indeed, we found similarly that activation of Snf1 by deleting REG1 could increase the weak constitutive expression of ADH2 in the tail mutants (Table 2 and data not shown).

Glucose-resistant ADH2 transcription is Snf1-independent in the histone H3 and H4 tail mutants (Fig. 2C), but Snf1-dependent in the Δhdac mutant (Tachibana et al., 2007), suggesting that there may be an important Snf1 target within the tails. Ser10 of Histone H3 seemed a likely possibility because Snf1 appears to act by phosphorylating H3-Ser10 to induce INO1 expression (Lo et al., 2001). However, the S10A mutation had no effect on ADH2 regulation, apparently eliminating its phosphorylation by Snf1 as a prerequisite of Adr1 binding and transcription activation. HIS3 expression is also dependent on Snf1 but, analogous to ADH2 expression, the dependence on Snf1 does not involve phosphorylation of Ser10 on histone H3 (Liu et al., 2005). Recent studies have not reproduced the dependence of INO1 induction on Ser10 phosphorylation (Shirra et al., 2005) so its importance, if any, is uncertain. An alternative site of Snf1 action in the histone tails is suggested by research demonstrating that Ser36 in the tail of mammalian H2B is phosphorylated by AMPK in response to stress and that this modification is related to gene activation (Bungard et al., 2010). In yeast H2B Threonine is present at an analogous position but its phosphorylation status has not been reported and the importance of the H2B tail for expression of Snf1-dependent genes is unknown.

Some aspect of the histone H4 tail might play a role in growth on a non-fermentable carbon source because its deletion, but not deletion of the H3 tail, prevents growth on glycerol (data not shown). This role may involve the Glc7 phosphatase because deleting REG1, which activates Glc7, was lethal in the absence of the H4 tail, but not in the absence of the H3 tail. Surprisingly, deleting Snf1 did not rescue the inviability of the Δreg1Δh4t double mutant. Thus, Reg1 appears to have a role independent of Snf1 that might be related to chromatin. This role may involve histone gene expression because a histone gene transcription factor, Spt10, is synthetically lethal in the absence of Reg1 (Eriksson et al., 2005, Hess & Winston, 2005).

The chromatin architecture of the ADH2prm is significantly altered in the histone tail deletion mutants and this may account for the binding of Adr1 and RNA Pol II. Similar changes in chromatin architecture occur in the Δhdac mutant and these changes were independent of Adr1. In both cases promoter nucleosomes have apparently been evicted. Since recruitment of the CRCs SAGA and Swi/Snf to the ADH2prm is normally dependent on Adr1 (Biddick et al., 2008b), these results suggest that significant perturbations in promoter architecture can occur without these chromatin modifying activities. Alternatively, in the absence of the histone tails the CRCs may be able to bind and alter the chromatin without activator-dependent recruitment. The increase in accessibility of DNA occluded by nucleosomes induced by the histone tail deletion may accomplish the role of the missing step in chromatin remodeling that normally occurs downstream of Snf1 activation in derepressing conditions. This would explain the synergistic effect on ADH2 expression described in Table 2 and Fig. 3.

In summary we have shown that deletion of the H3 or H4 tail allows Adr1 and RNA Pol II to bind the promoters in repressing conditions when chromatin structure normally precludes such binding. Activator- and Snf1-independent transcription occurred at a low level in the absence of the H3 and H4 tails. However, the RNA Pol II bound at the ADH2prm appeared to be mostly inactive, suggesting that, as observed in a Δhdac mutant, there is something missing from the PIC that is necessary for efficient transcription. Alternatively or in addition, incomplete chromatin remodeling might prevent complete activation. We conclude that when glucose is present, hypoacetylated histone H3 and H4 tails are essential to avoid Snf1-independent transcriptional noise and the unscheduled binding of activators. However, the inactive state of the Snf1 kinase prevents the formation of highly active transcriptional complexes on glucose even when chromatin is not a barrier for transcription. At the same time, Snf1-independent factors, or Snf1 activated to a higher level in the absence of glucose may be required to complete chromatin remodeling of the ADH2prm.

Experimental Procedures

Yeast strains, plasmids, and growth conditions

The S. cerevisiae strains used in all experiments with histone mutants were derived from MX1-4C, in which both chromosomal copies of the histone H3 and H4 genes have been deleted (Morgan et al., 1991) and the only source of histones is a centromeric plasmid containing one copy of the histone H3 and H4 genes encoding either wt or mutant versions of the histones (Table 3). Other strains are derived from W303. The plasmids used in this study are listed in Table 3. Deletion and tagging of chromosomal genes used published methods and plasmids (Knop et al., 1999, Guldener et al., 1996). For glucose-repression, yeast strains were grown in YP media (1% Yeast extract, 2% Peptone) containing 5% glucose and kept under OD₆₀₀=1.0. Glucose-repression was maintained in synthetic media containing 5% glucose to maintain selection for plasmids (Sherman, 1991). Derepression medium was YP containing 0.05% glucose. Growth conditions on plates for plasmid selection or with 5-fluoro-orotic acid (FOA) were described elsewhere (Young et al., 2003, Voronkova et al., 2006). All cultures were grown at 30 °C.

Table 3.

Strains and plasmids

Strain	Description
W303-CH1a	MATa ade2-1 can1-100 ura3-1 leu2-3,112 trp1-1 his3-11,115
MX1-4C	MATa ura3-52 leu2-3,112 trp1-289 his31 Δ(hhf1-hht1)Δ(hhf2-hht2) pMS329 (Sabet et al., 2003)
CKY11-2 (Δreg1)	W303-CH1a Δreg1::NAT1
JJY200 (wt)	MATa ura3-52 leu2-3,112 trp1-289 his31 Δ(hhf1-hht1)Δ(hhf2-hht2pRM200) (Sabet et al., 2003)
JJY430 (Δh3t)	MATa ura3-52 leu2-3,112 trp1-289 his31 Δ(hhf1-hht1)Δ(hhf2-hht2pRM430 (Sabet et al., 2003)
JJY428 (Δh4t)	MATa ura3-52 leu2-3,112 trp1-289 his31 Δ(hhf1-hht1)Δ(hhf2-hht2) pGF29 (hhf2Δ4-28) (Sabet et al., 2003)
JJY200Hs	JJY200 Δsnf1::ura3::HIS3
JJY430Hs (Δh3t)	JJY430 Δsnf1::ura3::HIS3
JJY428Hs (Δh4t)	JJY428 Δsnf1::ura3::HIS3
JJY200ac	JJY200 Δadr1::NAT1, Δcat8::KanMX
JJY430ac	JJY430 Δadr1::NAT1, Δcat8::KanMX
JJY200r	JJY200 Δreg1::KanMX
JJY430r	JJY430 Δreg1::KanMX
JJYS10Ar	MX1-4C Δreg1::KanMX pS10A
JJY200rs	JJY200 Δreg1::kanMxΔsnf1::NatMX
JJY430rs	JJY430 Δreg1::kanMx, Δsnf1:: NatMX
JJY428rs	MX1-4C Δreg1::kanMx, Δsnf1:: NatMX pGF29 (hhf2Δ4-28)
JJY200A-HA	JJY200 ADR1-3HA-KanMX
JJY430A-HA	JJY430 ADR1-3HA-KanMX
JJY428A-HA	JJY428 ADR1-3HA-KanMX
JJY200 pMS329	MX1-4C pRM200
JJY430 pMS329	MX1-4C pRM430
JJY428 pMS329	MX1-4C pGF29 (hhf2Δ4-28)
LLTY4	W303-CH1a (pLLTY1,TRP1)
LLTY6	W303-CH1a adr1::LEU2,( pLLTY1, TRP1)
LLTY79	JJY200 CAT8-3HA::KanMX
LLTY80	JJY430 CAT8-3HA::KanMX
LLTY111	W303-CH1a Δrpd3::LEU2 Δhda::NatMX adr1::KanMX (pLLTY1, TRP1) (aka Δhdac) (Tachibana et al., 2007)
LLTY122	MX1-4C (pLLTY1,TRP)
LLTY124	MATa ura3-52 leu2-3,112 trp1-289 his3-11,115Δ(hhf1-hht1)Δ(hhf2-hht2)(pLLTY21-HIS3) (pLLTY1,TRP1)

Plasmid	Description
pMS329	CEN4, ARS1, URA3, HHT1-HHF1 (Sabet et al., 2003)
pRM200	CEN4, ARS1, TRP1, HHT2-HHF2 (Sabet et al., 2003)
pRM430	CEN4, ARS1, TRP1, hht2Δ4-30-HHF2 (Sabet et al., 2003)
pGF29	CEN4, ARS1, TRP1, (HHT2-hhf2 Δ 4-28)
pS10A	CEN4, ARS1, TRP1, hht2 (H3 S10A)-HHF2 (Lo et al., 2001)
pBGM18	pRS316-based, URA3 CEN4 ADH2 (from codon 109)-lacZ (Young et al., 2000)
pLLTY21	CEN4, ARS1, URA3, hht2 Δ 4-30-HHF2
pNS460	CEN4, ARS1, TRP1, hht1K4,9,14,18,23,27Q-HHF1 (Sabet et al., 2003)
pRM491	CEN4, ARS1, TRP1-HHT1-hhf1K5,8,12,16Q (Sabet et al., 2003)

Chromatin Immunoprecipitation (ChIP) and mRNA quantification

ChIP was performed as described (Tachibana et al., 2005). One mg of whole-cell extract was immunoprecipitated with 2 μg of F7-antihemagglutinin monoclonal antibody (Santa Cruz Biotechnology). Quantification of specific sequences in both input and immunoprecipitated (IP) DNA preparations was done by qPCR using Power SYBR Green master mix (Applied Biosystems) in a PTC-200 Thermocycler coupled to a Chromo 4 continuous fluorescence detector (MJ-Research). Opticon 3 software (MJ-Research) was used for the data analysis. Occupancy of a protein at the promoter is expressed as fold-increase of the IP/input ratio of the amount of the specific amplicon for the promoter sequence over the IP/input ratio corresponding to the amplicon for the telomeric sequence TEL-VI-R. For the analysis of gene expression, total RNA samples were purified as described (Collart & Oliveiro, 1993) and DNAse I-treated with the Ambion DNAse free Kit. We used SSII-reverse transcriptase (Invitrogen) for the reverse-transcription of 1μg of RNA as described (Collart & Oliveiro, 1993). Quantification of mRNA levels was done by qPCR as described above and in (Tachibana et al., 2007). The primers used for both ChIP and expression analyses are in Table S1.

In-gel ADH assays and β-galactosidase assays

In-gel ADH assays were performed by in situ staining after electrophoresis of whole cell extracts on non-denaturing polyacrylamide gels (Dombek et al., 1993). β-galactosidase assays were performed on permeabilized cells (Guarente, 1983). The ADH2/lacZ reporter plasmid was pBGM18 (Table 3; (Young et al., 2000)). Activities are expressed in Miller units.

Nucleosome Scanning Assays (NuSA)

NuSA was performed as described in Tachibana (Tachibana et al., 2007) using 200 ml cultures of cells grown in either repressing or derepressing medium for 4 hr or 6 hr for W303 cells or cells derived from MX1-4C, respectively. Fold-enrichment values were normalized to either total DNA, or amplicons from CEN3 or ACT1 as described in Biddick (Biddick et al., 2008). The amplicon used for ACT1 normalization was demonstrated to be within a region in the ACT1 gene that was protected by nucleosomes during the MNase digestion (data not shown). No detectable difference was noted using the different procedures for normalization. The total DNA concentration was determined using Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen). When doing NuSA-ChIP assays, the MNase digestions were stopped by shifting the reactions to 4 °C and adding 18 μl of 0.5M EDTA and 7 μl of 200 mM EGTA. The samples were frozen at −70 °C until the next step. SDS was not added at this point since it would interfere with the antibody binding in subsequent steps. One 600 μl-aliquot from the MNase digestion step was used for each NuSA- ChIP. As outlined in the ChIP procedure referenced above protease inhibitors and PMSF were added to each 600μl sample. Appropriate amounts of stock solutions were added to make the final volume of each aliquot 800 μl in 1X of Buffer L (ChIP Buffer; 50mM Hepes, pH 7.5, 140mM NaCl, 1% triton X-100, 0.1M sodium deoxycholate and a mixture of protease inhibitors). The samples were sonicated (4X 20 sec pulses with low, constant power using a microprobe), centrifuged for 10min at 10000 xg, 4 °C in a table-top microfuge, and the supernatant was transferred to a new tube with PMSF. The antibody was added and incubated overnight at 4 °C with rocking. For histone immunoprecipitation, 5 μg of antibody per sample were used (H3: Upstate mAb 05-499; H2B: Upstate mAb 07-371; H4: Abcam ab 7311-200). For HA–immunoprecipitation, 10μl of HA antibody (#SC7392, Santa Cruz Biotechnology) were used. From this step on the ChIP procedure referenced above was used. The final ChIP samples were diluted approximately 20-fold in 1mM Tris pH 8.0. These diluted NuSA-ChIP samples were analyzed using the same method described for the NuSA samples.

Supercoiling assays

Supercoiling assays were performed as described in Tachibana (Tachibana et al., 2007) using the pLLTY1 plasmid, carrying the −640 to +135 region of the ADH2 gene. Repressing and derepressing media were the same as described above. Derepression was carried out for 4 hr (W303-derived strains) or for 6hr for cells derived from the MX1-4C strain.