Post-TATA Binding Protein Recruitment Clearance of Gcn5-Dependent Histone Acetylation within Promoter Nucleosomes

Abstract

Transcriptional activation of eukaryotic genes often requires the function of histone acetyltransferases (HATs), which is expected to result in the hyperacetylation of histones within promoter nucleosomes. In this study we show that, in Saccharomyces cerevisiae, the steady-state levels of Gcn5-dependent histone acetylation within a number of transcriptionally active promoters are inversely related to the rate of transcription. High acetylation levels were measured only when transcription was attenuated either by TATA element mutations or in a strain carrying a temperature-sensitive protein component of RNA polymerase II. In addition, we show that in one case the low levels of histone acetylation depend on the function of the Rpd3 histone deacetylase. These results point to the existence of an unexpected interplay of two opposing histone-modifying activities which operate on promoter nucleosomes following the initiation of RNA synthesis. Such interplay could ensure rapid turnover of chromatin acetylation states in continuously reprogrammed transcriptional systems.

It is well established that histone modifications play pivotal roles in regulating transcription in eukaryotes. Nucleosomal histones are the targets of covalent modifications such as acetylation, phosphorylation, methylation, ADP ribosylation, and ubiquitination that either independently or in combination participate in setting up the “on” or “off” transcriptional states of particular genes or even large chromosomal regions (4, 15). Given the chemical diversity and the combinatorial potential of these modifications, it has been proposed that modified patterns of histones within nucleosomes constitute a code which dictates a given transcriptional state (18, 41). It follows that in rapidly reprogrammed transcriptional systems the dynamics of the reversion of these modifications should be part of the molecular switch which determines gene expression patterns.

In Saccharomyces cerevisiae, one of the best-studied reversible systems of histone modifications is the acetylation of lysine residues located within the amino-terminal tails of histone H3 and H4 (23). A number of nuclear histone acetyltransferases (HATs) and histone deacetylases (HDACs) are targeted to promoter regions, usually as members of multiprotein transcriptional coactivator or corepressor complexes, via interactions with DNA binding proteins, and produce modifications on these two histones that correlate with either activated or repressed transcription, respectively (34, 43). In general, these histone modifications have been linked to the process of nucleosome remodeling, which is often a prerequisite for gene activation. Among the HATs, Gcn5 is a prototype enzyme which acetylates nucleosomal histone H3. It is recruited to promoters as part of the SAGA complex (40), which is targeted by activators such as Gcn4, Swi5, and Gal4 (5, 10, 22, 25, 27). Likewise, yeast Rpd3, a prototype HDAC which deacetylates histones H3 and H4, is recruited to promoters as a complex with Sin3 by the DNA-interacting repressor Ume6 (19, 38).

Such targeted recruitments and the resulting histone modifications provide a mechanism through which a given histone acetylation state is changed in response to an activating or repressing signal. Restoration to the resting state upon removal of the signal is less well understood. Reversal of induced histone hyperacetylation following the removal of the inducing signal in a number of mammalian and yeast systems has been reported (1, 7, 8, 22, 39). Although in some instances targeted recruitment of the opposing activity offers a mechanistic explanation (19, 51), these instances are exceptions rather than the rule. In contrast, the ill-defined concept of global or untargeted histone acetylation-deacetylation has been implicated in the rapid restoration of promoter acetylation levels, which is observed upon removal of activators or repressors (12, 28, 48). Interestingly, recent studies on HDACs have shown that a large proportion of these global effects can be attributed to the association of HDACs with a much larger number of chromosomal regions than previously suspected (26, 37).

Several recent observations suggest that these rapid reversals could be the result of the coexistence of opposing enzymatic activities. HDACs are preferentially located within upstream regions of genes that direct high transcriptional activity (26), and recently it was reported that HDACs and HATs exist within the same complexes (52). In fact, balanced action of these two opposing activities could explain the lack of a clear rule for yeast that would correlate HAT involvement in transcriptional activation with increased levels of histone acetylation (11). The most notable example is the Pho4-directed transcriptional activation of the PHO8 gene, whose Gcn5 dependence does not result in any measurable acetylation of promoter histones (35). In fact, high levels of histone acetylation were measured on the PHO8 promoter only in the transcriptionally inactive swi2 strain. It was proposed that a transient function of Gcn5 on PHO8 promoter nucleosomes is sufficient to trigger the system and that this was somehow counteracted following nucleosome remodeling. Alternatively, given the fact that acetylation was not evident even at early points in the induction process, when nucleosome remodeling was not observed, it is possible that PHO8 represents an extreme situation reflecting the existence of a cotranscriptional interplay of HATs and HDACs which ensures the overall balance of chromatin acetylation levels.

In this study we present evidence supporting the existence of such a cotranscriptional mechanism that determines the steady-state levels of Gcn5-dependent histone H3 acetylation on Gcn4-, Pho4-, and Gal4-driven promoters. In fact, we show that these levels are inversely proportional to the efficiency of transcription, and we demonstrate the involvement of an HDAC activity in modulating the turnover of acetylation marks. We propose that this dynamic interplay of histone-modifying activities during transcriptional activation is part of the mechanism that ensures rapid changes in chromatin acetylation states.

MATERIALS AND METHODS

Yeast strains and media.

The gcn4Δ and gcn5Δ strains were obtained by appropriate gene disruptions of a GAL2 S288C parental strain with the genotype MATa trp1-1 leu2-3 ura3-52, as described by Georgakopoulos et al. (14). The spt3Δ and spt3Δ gcn5Δ strains were obtained by disrupting the SPT3 gene via kanamycin selection. Strain FY963 (MATa leu2Δ1 ura3-52 his4-917δ spt7Δ::LEU2) and the rpd3Δ strain (MATa trp1-1 leu2-3 ura3-52 his3Δ200 rpd3Δ::HIS3) were provided by F. Winston and K. Struhl, respectively. The rpb1ts strain genotype was MATa leu2-3 ura3-52 rpb1-1. In the rpd3Δ and rpb1ts strains, gcn5 was replaced by a kanamycin cassette. Wild-type and gcn5Δ strains bearing the improved TRP3 TATAA element or the mutated PHO5 CGTA or GAL1 CGTA element were derived by two-step replacements of their endogenous loci. DNA encoding nine Myc epitopes was inserted in frame after the last codon of the RPD3 gene at its chromosomal locus by a PCR-based strategy, with TRP1 as a selection marker. Repressed-growth conditions were either rich medium (2% glucose) or minimal medium supplemented with the required amino acids. High levels of Gcn4-dependent transcription were achieved by adding 3-aminotriazole to the medium at a concentration of 10 mM and thus eliciting histidine limitation. For Pho4-dependent induction yeast strains were grown in phosphate-free synthetic medium (44). GAL1 induction was achieved by growing cells to mid-log phase in rich media containing 2% raffinose and then shifting them to yeast extract-peptone-galactose (2% galactose) for 1 h.

Plasmid constructions.

A PHO5 reporter gene was constructed by inserting a PCR fragment containing PHO5 sequences from nucleotide −392 to +3 (which includes both upstream activating sequence 1 [UAS1] and UAS2, the TATA, the transcription start site, and the initiator Met codon) into the EcoRI-HindIII sites of a HIS3-lacZ reporter gene (the CEN3 ARS1 URA3 plasmid VS11 in reference 13). A double-stranded synthetic oligonucleotide containing the Gcn4 response element (GCRE; 5′CGATGACTCATATGCAT3′; underlining indicates the GCRE) was inserted into the BstEII site in the linker region between nucleosomes −1 and −2 to generate the bPHO5 reporter. The TATA box mutant sequences CGTA and TGTA within bPHO5 were obtained by site-directed mutagenesis using the mutagenic primers 5′CGAAATGAAACGTACGTAAGCGCTGATG3′ and 5′CGAAATGAAACGTATGTAAGCGCTGATGTTTTGCTAAG3′. The former primer was also used for the creation of the mutant TATA box of the endogenous PHO5 gene. The TATA box mutation of the GAL1 gene was obtained by site-directed mutagenesis using the mutagenic primer 5′GATCTATTAACAGACGCGTAAATGGAAAAGCTTGC3′ (underlining indicates the base alterations). The TATA box mutation of the TRP3 gene was obtained by site-directed mutagenesis using the primer 5′TTCTAATCTATAGCATATATAAAGGTACCCAAAAAGTTCGACAAGGAGC3′. The construction of the HA-tagged GCN5 gene derivative was described previously (45).

Gene expression analysis.

Total yeast RNA was isolated from yeast cells grown to an optical density at 550 nm of 0.6 to 0.8 by the hot-phenol method as described previously (17). Approximately 5 μg of RNA was loaded onto a 1.5% agarose-formaldehyde gel and hybridized in accordance with standard protocols. Measurements of β-galactosidase activities were performed as described by Syntichaki et al. (45).

In vivo nucleosome-remodeling assays.

Preparation of yeast nuclei and restriction enzyme digestions were performed as described previously (2). bPHO5 and PHO5 sensitivities to restriction endonucleases were analyzed as described by Syntichaki et al. (45) and Svaren and Horz (44), respectively. Micrococcal nuclease assays were performed using nystatin-permeabilized spheroplasts (47). Following micrococcal nuclease digestions the nucleosomal structure of the GAL1 promoter was analyzed by the indirect end labeling method. PvuII was used for secondary digestion, and the BsaI-PvuII fragment of GAL1 served as a probe. Quantifications were performed with a phosphorimager (Molecular Dynamics).

Chromatin immunoprecipitation.

Chromatin was immunoprecipitated from extracts of formaldehyde-fixed cells as described previously (24) with antisera against diacetylated H3 (K9 and K14; Upstate Biotechnology Inc.), against tetra-acetylated H4 (Chemicon), against hemagglutinin (HA) and Myc (Santa Cruz Biotechnology), or against yeast TATA binding protein (TBP; a gift from S. Buratowski). Immunoprecipitated material was eluted, and the recovered DNA was subjected to quantitative PCR analysis. For experiments with the bPHO5 promoter the set of primers used for the amplification of the DNA in the region of nucleosomes −1 and −2 were 5′CATTGGTAATCTCGAAT3′ and primer 1 described by Syntichaki et al. (45) in combination with primers corresponding to the GCRE (see Fig. 1A). For Gcn5 recruitment, primer 2 (45) was used in combination with primer 4 (5′GCACATGCCAAATTATC3′), corresponding to a sequence located 24 bp upstream of the BstEII site (bPHO5; see Fig. 2D). For the amplification of the regions located approximately 1 kb upstream or downstream of the bPHO5 promoter, primer pairs recognizing part of the β-lactamase and the β-galactosidase genes, respectively, were used. Those were 5′GAGTACTCACCAGTCACAG3′ and 5′TGCCATTGCTACAGGCATC3′ for the β-lactamase gene and 5′AACGCCGTGCGCTGTTCG3′ and 5′CCGTGGCCTGATTCATTCC3′ for the β-galactosidase gene. Primers 3 (45) and 4 were used for the amplification of a region of the endogenous PHO5 promoter encompassing nucleosome −1. Primers 5′CCCACGATACCGATATCCTAA3′ and 5′ TGGACCAGTGGAACAGGTTT3′ were used for the amplification of a region located approximately 1 kb downstream of the PHO5 promoter. The GAL1 promoter was analyzed with primers 5′CAGCGAAGCGATGATTTTTG3′ and 5′CGCTAGAATTGAACTCAGGA3′, encompassing the TATA-occluding nucleosome. For the amplification of a region located 1.5 kb downstream of the GAL1 promoter, primers 5′TTGACAAAATTTGTTCCA3′ and 5′TGCTGGTTTAGAGACGATG3′ were used. Finally, for TRP3 the primers used to amplify the promoter region were those described by Kuo et al. (25). The ACT1 coding region was used as the control in most experiments (primers 5′CCTACGAACTTCCAGATGG3′ and 5′CACTTGTGGTGAACGATAG3′). For experiments with the PHO5 and GAL1 promoters the hypoacetylated telomeric region used previously (48) served as the control (primers 5′TCATAAACATAAGCGTATCC3′ and 5′GCAACGACTTCGTCTCAG3′). PCRs were performed in the presence of [α-³²P]dATP, and products were analyzed in 1.5% agarose gel. The DNA fragments were visualized through autoradiography and quantified with a phosphorimager (Molecular Dynamics). Each chromatin immunoprecipitation (IP) was repeated at least three times, and the results showed no more than ±10% variation. For each case a representative experiment is shown.

FIG. 1.

Structure, expression, and nucleosome modifications at the bPHO5 promoter. (A) Schematic representation of the bPHO5 promoter indicating the positioned nucleosomes (−2 and −1), the Gcn4 binding site (GCRE), the TATA element, and the start site of transcription (top arrow). Also indicated are the restriction endonuclease (ClaI and DdeI) cleavage sites used to monitor nucleosomal remodeling. Small arrows, positions of primers used in chromatin immunoprecipitation (ChIP) analysis. (B) Transcriptional activation through the bPHO5 promoter measured as the amount of β-galactosidase (β-gal) activity produced by the bPHO5-lacZ derivative. These values varied ±10% between experiments. The indicated strains were grown either in minimal media (white bars) or for 12 h in histidine starvation media (black bars). WT, wild type. (C) Restriction endonuclease (ClaI and DdeI) sensitivities of the nucleosomes in wild-type and gcn5 strains grown either in minimal media (M) or for 12 h in histidine-limiting media (AT) which result in increased amounts of Gcn4. P, undigested DNA fragment; U, largest DNA fragment produced following the activity of the restriction enzyme on chromatin. (D) Representative experiment showing the detailed monitoring of histone H3 acetylation in a wild-type and gcn5 strains grown under histidine limitation conditions by ChIP using an antibody raised against the diacetylated histone H3 (H3Ac) N-terminal peptide. Input (IN) and immunoprecipitated DNA (IP) were detected following PCRs using primers which amplify a region of the DNA which marks the positions of either nucleosome −2 (left) or −1 (right) and a region of the ACT1 open reading frame, used as a control. (E) Representative experiment (as in panel D) showing Gcn5-dependent H3 acetylation levels at the −2 promoter region in spt3 and snf2 strains growing under induction conditions.

FIG. 2.

Histone H3 acetylation levels in bPHO5 are inversely related to the efficiency of TBP recruitment. (A, left) Representative experiment of chromatin immunoprecipitation (ChIP) using an antibody raised against yeast TBP. The experiment was performed with the indicated strains grown under histidine limitation and bearing bPHO5 reporters containing either the wild-type TATA element (TA) or the mutated versions TGTA (TG) and CGTA (CG). Immunoprecipitated DNA (IP) was detected following PCR amplification using primers specific for the −1 nucleosome DNA region of bPHO5 or the ACT1 open reading frame (ORF). Numbers show the quantification of this experiment expressed as the percentage of amplified −1 DNA relative to that amplified in a wild-type strain containing the wild-type bPHO5 reporter. IN, input. (Right) Representative experiment showing transcriptional activation through the bPHO5 promoter bearing the indicated mutated TATA elements and measured as the amount of β-galactosidase (β-gal) activity produced by the bPHO5-lacZ derivative. The indicated strains were grown either in minimal media (M) or for 12 h in histidine starvation media (AT). WT, wild type. (B) Representative ChIP experiments show the histone H3 acetylation levels in wild-type and gcn5 strains containing the TATA, the CGTA, or the TGTA bPHO5 derivative and growing in either minimal (M) or histidine-limiting (AT) media. Numbers represent the factors of enrichment of the amount of amplified −1 DNA relative to the input, normalized for ACT1. Identical results were obtained with primers amplifying the adjacent −2 DNA region. H3Ac, diacetylated histone H3. (C) Representative ChIP experiment showing H3 acetylation levels on chromatin located 1 kb either upstream (−1Kb) or downstream (+1Kb) from the transcription start site of bPHO5 in wild-type and gcn5 strains containing the TATA or CGTA bPHO5 derivative and growing in histidine-limiting media. Numbers represent the factors of enrichment of the amount of amplified −1 DNA relative to the input, normalized for ACT1. H3Ac, histone H3 acetylation. (D) Recruitment of the HA-tagged Gcn5 protein within the bPHO5 promoter bearing either the wild-type TATA (TA) or the CGTA (CG) element detected through ChIP. A representative experiment is shown. The bPHO5 amplified fragment encompasses three promoter nucleosomes. Numbers are as in panel B. (E) Representative experiment showing a ChIP performed with an antibody raised against tetra-acetylated histone H4 (H4Ac) N-terminal peptide. The experiment was performed for the indicated strains grown under histidine limitation and bearing either the wild-type bPHO5 reporter or the derivative with the CGTA variant. Input and immunoprecipitated DNA was detected following PCRs using primers which amplify a region of the DNA corresponding to the position of the −1 nucleosome and the ACT1 ORF region. (F) Remodeling of nucleosomes −1 and −2 as a function of the TATA element in wild-type strains transformed with bPHO5 reporters containing either the wild-type TATA element (TA) or the mutated versions TGTA (TG) and CGTA (CG) and grown in minimal (M) or histidine limitation (AT) conditions. Numbers represent the extent of remodeling expressed as percentages of digestion by ClaI (−2) or DdeI (−1). P, undigested DNA fragment; U, largest DNA fragment produced following the activity of the restriction enzyme on chromatin.

RESULTS

In a synthetic promoter, the function of Gcn5 HAT is evident only when TBP recruitment is compromised.

In the course of previous studies addressing the impact of promoter architecture on coactivator selectivity and chromatin modifications, we constructed a Gcn4-PHO5 synthetic promoter in which the Gcn4 binding site was placed in the linker region between nucleosomes −2 and −1 (Fig. 1A). In the resulting reporter (bPHO5), transcriptional activation only partially required Gcn5 (Fig. 1B). The remodeling of nucleosome −2 was completely Gcn5 dependent, whereas that of nucleosome −1 only partially required this protein (Fig. 1C). Note that identical results were obtained when a strain carrying a HAT-defective derivative, rather than a deletion of Gcn5, was utilized (data not shown). Despite these dependencies, there was no detectable histone H3 acetylation in the region encompassing each one of the promoter nucleosomes (Fig. 1D). Nevertheless, as shown in Fig. 1E, Gcn5-dependent acetylation was observed in strains in which transcription of bPHO5 was compromised, such as the swi2 and spt3 strains (Fig. 1B). Although alternative interpretations exist, these results prompted us to investigate directly the possible involvement of transcriptional rates with the apparent levels of histone acetylation. For this purpose we monitored the extent of Gcn5-dependent H3 acetylation on bPHO5 derivatives bearing mutated TATA elements. We first introduced a single base alteration in the core TATA element of bPHO5, resulting in the TGTA variant. As shown in Fig. 2A (left), TBP was recruited into this attenuated element (33, 42) but five times less efficiently than into the wild-type sequence. This was reflected also in the extent of Gcn4-dependent transcriptional activation which was dependent on Gcn5 (Fig. 2A, right). By contrast, the introduction of an additional mutation altering the core TGTA element to CGTA drastically reduced TBP recruitment and transcriptional activation (Fig. 2A). When the levels of Gcn5-dependent histone H3 acetylation on these promoters were examined, a Gcn4-dependent negative correlation with TBP recruitment was revealed (Fig. 2B). Furthermore, these increased levels were confined to the promoter region since they were not evident at a distance of ±1 kb from the transcription start site (Fig. 2C). As shown in Fig. 2D, these increased levels of promoter acetylation were not due to a difference in the amount of Gcn5 recruited. In addition, we observed that the extent of TBP recruitment did not affect the acetylation levels of histone H4 (Fig. 2E), a fact arguing against TBP-dependent differences in the extent of nucleosomal occupancy. We concluded that the steady-state level of Gcn5-dependent histone H3 acetylation on the bPHO5 promoter was inversely related to the efficiency of TBP recruitment. Finally, although the strong remodeling of nucleosome −2 was independent of the nature of the TATA element, that of nucleosome −1 was partly dependent on the extent of TBP recruitment (Fig. 2F). Thus, these results cannot totally exclude the possibility that the increased levels of H3 acetylation observed for the TATA mutated derivatives could be attributed to differences in the extent of nucleosome −1 remodeling. This possibility was clearly excluded by the results presented below.

The efficiency of TBP recruitment determines the levels of histone H3 acetylation of two native promoters.

On the basis of these observations we decided to expand them beyond the case of an episomal synthetic promoter, and thus we experimented with two endogenous promoters, PHO5 and GAL1. In accordance with published reports (3, 16, 32), both promoters required Gcn5 for transcriptional activation (Fig. 3A, top) but, as was found for bPHO5, this dependency did not translate into any measurable increases in histone H3 acetylation effected by Gcn5 (Fig. 3B; TATA-bearing promoters) (3, 11). We generated TATA-mutated versions of both promoters in their native chromosomal location and then monitored Gcn5-dependent histone H3 acetylation in relation to transcriptional activity and TBP recruitment. These TATA mutations compromised the transcriptional activation of both promoters (Fig. 3A, top), a fact which was reflected in the drastic attenuation of TBP occupancy (Fig. 3A, bottom). Interestingly, in strains growing under induction conditions, Gcn5-dependent histone H3 acetylation was apparent for the chromatin located within promoters bearing the mutated TATA elements (Fig. 3B, PRO probe) and, similar to findings for bPHO5, this acetylation was confined to the promoter region (Fig. 3B, ORF probe). Furthermore, neither the pattern of nucleosome remodeling effected under inducing conditions for both promoters (Fig. 3C) (29) nor the levels of nucleosome occupancy (data not shown) (for PHO5 see reference 6) were influenced by the TATA mutations. These experiments showed that the observed TBP recruitment-dependent differences in acetylation levels could not be attributed to differences in chromatin structure.

FIG. 3.

Gcn5-dependent acetylation on the activated PHO5 and GAL1 promoters is detected only when transcription is attenuated by TATA box mutations. (A, top) PHO5 (left) and GAL1 (right) mRNA levels are compromised by TATA mutations. RNA was isolated from wild-type (WT) and gcn5 strains containing the natural TATA element or the CGTA PHO5 or GAL1 derivative grown under inducing conditions, and DED1 was used as a loading control. (Bottom) Typical experiment showing TBP occupancy within the PHO5 (left) and GAL1 (right) promoters bearing either the natural TATA element (TA) or the CGTA (CG) version grown under inducing conditions. Numbers represent the factors of enrichment of the amount of amplified DNA relative to the input (IN), normalized for ACT1. (B) Representative experiment showing measurements of histone acetylation within the PHO5 (left) and GAL1 (right) promoters bearing either the natural TATA element or the CGTA version in a wild type or gcn5 strain grown under repressed or inducing conditions. Histone acetylation was also measured for the TATA derivatives of both genes in a region located within the open reading frame (ORF) in wild-type and gcn5 strains grown under inducing conditions. Due to the relatively high levels of Gcn5-independent histone acetylation within the PHO5 promoter, measurements were more evident when primers amplifying a hypoacetylated telomeric (TEL) region rather than the ACT1 ORF were used as the control. Numbers represent the factors of enrichment of the amount of amplified DNA relative to the input, normalized for that for the TEL region. H3Ac, diacetylated histone H3. (C) Nucleosome remodeling as a function of the TATA element. (Left) Nucleosomes −1 and −2 of the native PHO5 promoter (TATA) and its derivative (CGTA) were assayed for sensitivity to DdeI and ClaI, respectively, in cells grown in repressing (M) or inducing (LP) conditions. (Right) Nucleosomes −1 and −2 of the native GAL1 promoter (TATA) and its derivative (CGTA) were assayed by measuring micrococcal nuclease sensitivity in cells grown in repressing (Glu) or inducing conditions (Gal). The nucleosomal structure of the GAL1 promoter and the positions of the UASg and the TATA elements are shown. This is in agreement with those reported by Lohr (29).

Since the HDAC Rpd3 is involved in chromatin modifications at the PHO5 promoter (48), we decided to investigate its possible role in the above phenomena. As shown in Fig. 4, high levels of Gcn5-dependent H3 acetylation in an rpd3 strain at the PHO5 promoter (A) but not at GAL1 (B) were evident. The latter result is consistent with the fact that Rpd3 has not been reported to participate in H3 histone modifications within the GAL1 promoter (49). This increased acetylation at PHO5 did not reflect any untargeted function of Gcn5, since it was induction dependent and confined to the promoter region and since the absence of Rpd3 did not result in further increases in histone acetylation of the promoter bearing the CGTA derivative (Fig. 4A). Finally, as shown in Fig. 4C, Rpd3 was recruited to the CGTA derivative of PHO5 as efficiently as to the wild-type promoter. Thus a simple model linking increased acetylation to Rpd3 eviction was discarded. The totality of the above experiments confirmed that our initial observations with the episomal reporter can be recapitulated with promoters PHO5 and GAL1 in their native chromosomal context and furthermore pointed to the interplay of opposing histone-modifying activities on promoters. These results then predict the existence of a general mechanism which, in a TBP recruitment-dependent manner, cancels out the Gcn5-dependent histone H3 acetylation.

FIG. 4.

The Rpd3 HDAC counteracts Gcn5 on the PHO5 promoter. (A) Representative experiment showing measurements of histone acetylation within ether the PHO5 promoter (PHO5p) or the coding region (PHO5c) for the gene bearing either the wild-type (TATA) or the mutated (CGTA) TATA element in rpd3 (r3) and rpd3 gcn5 (r3g5) strains grown under the indicated conditions. IN, input; IP, immunoprecipitated DNA; H3Ac, diacetylated histone H3. (B) Representative experiment showing measurements of histone acetylation within the GAL1 promoter in rpd3 and rpd3 gcn5 strains grown under inducing conditions. (C) Representative experiment showing measurements of Rpd3 occupancy on either the wild-type (TA) or the mutated (CG) PHO5 promoter in strains grown under repressive (HP) or inducing (LP) conditions. In all experiments numbers represent the factors of enrichment of the amount of amplified DNA relative to the input, normalized for that for the telomeric region (TEL).

“Improvement” of activated transcription results in decreased levels of H3 acetylation.

To investigate the prediction that improvement of activated transcription results in decreased levels of H3 acetylation, we shifted to a completely different promoter context, namely, that of the TRP3 gene. This choice was based on the facts that this Gcn4-regulated promoter was Gcn5 dependent and that, upon induction, relatively high levels of H3 acetylation could be observed (25). We hypothesized, based on the above results, that these levels were high due to the reported weak TATA element of this promoter (30). To test this hypothesis, we replaced the TATA (GATATTA) element of the endogenous TRP3 promoter with the optimum TBP binding sequence TATATAA. This new TATA element supported increased synthesis of TRP3 mRNA (Fig. 5A) and recruited larger amounts of TBP (Fig. 5B), but, at the same time, the chromatin of this promoter exhibited reduced levels of H3 acetylation compared to the wild-type promoter (Fig. 5C). By contrast, when transcriptional activation (Fig. 5A) and TBP recruitment (Fig. 5D, bottom) at the wild-type TRP3 promoter were compromised in an spt3 strain, increased levels of H3 acetylation were detected again (Fig. 5D, top). Thus, our prediction was verified, and these results strengthen the notion that the inverse relationship between the levels of Gcn5-dependent H3 acetylation and transcriptional rates reflects general aspects of the mechanism of transcriptional activation.

FIG. 5.

Gcn5-dependent histone H3 levels as a function of the transcriptional efficiency directed by the TRP3 promoter. (A) TRP3 mRNA levels produced in the indicated strains by either the wild-type (WT) gene (TATTA) or the derivative bearing the improved TATAA element. The levels of DED1 RNA were used as a loading control, and all strains were grown under histidine limitation. (B) Representative experiment showing TBP recruitment within the TRP3 promoter bearing either the wild-type (TTA) or the mutated (TAA) TATA element. Inputs were the same as those shown in panel C for the wild-type strains. Numbers represent the factors of enrichment of the amount of amplified DNA relative to the input, normalized for ACT1. IP, immunoprecipitated DNA. (C) Representative experiment showing histone H3 acetylation levels within the TRP3 promoter bearing either the wild-type (TATTA) or mutated (TATAA) TATA elements in the indicated strains grown under histidine limitation. Numbers show the quantification of the experiment expressed as the percentages of amplified DNA relative to that amplified in a wild-type strain containing the wild-type TATTA element. IN, input; H3Ac, diacetylated histone H3. (D) Representative experiment showing measurements of histone H3 acetylation (top) and TBP recruitment (bottom) within the TRP3 promoter in wild-type, gcn5, and spt3 strains grown under histidine limitation.

Establishment of steady-state histone H3 acetylation levels requires the function of Pol II.

We next attempted to define more precisely the step which activated the pathway leading to the clearance of histone acetylation. Our results suggested that either TBP recruitment per se or events coupled with the formation of preinitiation complexes could be directly involved. Alternatively, this process might be coupled with RNA synthesis and/or elongation. To investigate some of these possibilities, we utilized a strain carrying a temperature sensitivity mutation in the large subunit of RNA polymerase II (Pol II), Rpb1. This mutation does not preclude the formation of the preinitiation complex but affects the enzymatic activity of the polymerase (21). Due to the high instability of the Gcn4 protein (31) we were able to investigate only the Pho4- and Gal4-regulated promoters. As shown in Fig. 6, increased levels of Gcn5-dependent H3 acetylation were observed at the PHO5 (Fig. 6A) and GAL1 (Fig. 6B) promoters when the temperature at which growth took place was shifted to the restrictive temperature. Most importantly, TBP recruitment was not affected by the temperature shift (Fig. 6, right). Note also that inactivation of Pol II did not affect additively the levels of histone H3 acetylation within the CGTA promoter derivative (data not shown), a fact arguing against an indirect effect other than the possibility for a highly unstable Rpd3 protein. We conclude that histone acetylation turnover was activated following TBP recruitment and preinitiation complex formation and requires the function of Pol II.

FIG. 6.

Pol II activity is required for the establishment of steady-state levels of histone acetylation on PHO5 and GAL1 promoters. Shown is a representative experiment assaying histone H3 acetylation within the PHO5 (A, left) and GAL1 (B, left) promoters in temperature-sensitive rpb1 and rpb1 gcn5 strains grown under inducing conditions, either at the permissive (23°C) temperature or for 1 h at the restrictive temperature (37°C). As a control it was confirmed that shifting wild-type (WT) and gcn5 strains to 37°C for 1 h does not result in hyperacetylation (first two lanes in each panel). The right sections (A and B) show that TBP recruitment within both promoters in the rpb1 strain is not affected by the temperature shift. Numbers represent the factors of enrichment of the amount of amplified DNA relative to the input (IN) normalized for that for the telomeric region (TEL). IP, immunoprecipitated DNA; H3Ac, diacetylated histone H3.

DISCUSSION

The results presented in this paper provide novel insights into the dynamics of the establishment of histone acetylation levels upon transcriptional activation and reveal the existence of a mechanism which operates at a step following TBP recruitment and which limits these levels. Our initial observations supporting such a mechanism were obtained with the synthetic bPHO5 reporter. Although transcription through this promoter was partly dependent on Gcn5, the remodeling of nucleosome −2 was fully dependent on the HAT activity of this protein. Yet we could not measure any Gcn5-dependent increase in histone H3 acetylation within the promoter nucleosomes. Thus, there was an apparent paradox concerning the lack of correlation between the requirement for Gcn5 and histone H3 acetylation, a consequence of its known activity. Nevertheless, using the bPHO5 TATA variants we could directly demonstrate that the levels of histone H3 acetylation specifically within promoter nucleosomes were inversely related to transcriptional efficiency. Interestingly enough, this apparent inconsistency was not unique to the synthetic bPHO5 promoter.

Indeed, motivated by these observations and using analogous methodologies, we observed in native promoters similar phenomena that reflected general features of the transcription-dependent dynamics of histone acetylation. Both the endogenous PHO5 and GAL1 promoters, whose transcriptional induction is dependent on Gcn5 (3, 16, 32), exhibited a similar behavior. Measurable levels of Gcn5-dependent histone H3 acetylation were evident only in the transcriptionally compromised promoters bearing TATA box mutations. These results were important, since they uncoupled the observed phenomenon from a given activator or a given chromosomal location. Even more important was the generalization as well as the confirmation of the predictive power of our initial observations obtained by the forward approach taken with another Gcn5-dependent promoter, TRP3. In this case, improvement of a weak TATA element resulted in increased transcriptional rates but reduced levels of Gcn5-mediated histone acetylation. In contrast, reduction of TRP3 transcriptional activation in an spt3 strain resulted in increased levels of Gcn5-dependent histone acetylation.

Recent reports showed that chromatin remodeling at the PHO5 promoter essentially translates into a dynamic unfolding of promoter nucleosomes (6, 36). Under inducing conditions nucleosome −2 is almost completely lost, whereas the unfolding of nucleosome −1 is more dynamic in that this nucleosome is present at 60% of the promoters at any given time. It follows that a trivial explanation of our results could have been a TBP recruitment-dependent difference in nucleosome loss. That this is not the case is supported by relevant measurements showing identical extents of nucleosome loss in a wild-type and a TATA-mutated PHO5 promoter (6). Using chromatin immunoprecipitation with an antibody directed against the C terminus of histone H3 we independently verified that the extents of induction-dependent nucleosome unfolding for the wild-type promoter and our TATA-mutated PHO5 promoter were the same (data not shown). Finally, we also measured (data not shown) levels of nucleosome occupancy for the induced wild-type and TATA-mutated GAL1 promoters, and again no differences were detected. Although the totality of these results excludes nucleosome loss-based explanations, the results nevertheless point to the fact that the observed acetylation dynamics for the PHO5 promoter involve nucleosome −1. In addition, they raise the possibility that the measured levels of histone acetylation under induction conditions were underestimated.

Given that nucleosome loss does not offer an explanation, what is the underlying mechanism for the observed phenomena? The PHO5 and the GAL1 promoters offered us the opportunity for additional dissections of the mechanisms involved in the establishment of histone acetylation levels. First, results obtained with a temperature-sensitive mutant strain directly affecting Pol II activity strongly indicated that events following preinitiation complex formation, not just TBP recruitment per se, were required for the function of an activity that counteracted Gcn5. Second, based on the fact that Rpd3 is localized within the PHO5 promoter chromatin (26) and is one of the determinants of its histone acetylation status (48), we identified this HDAC as the counteracting activity. Indeed, in an rpd3 strain the levels of Gcn5-dependent H3 acetylation specifically within the PHO5 promoter under inducing conditions were high. In addition, this mutation did not effect any further increase in the acetylation state of the TATA-mutated PHO5 promoter. Thus, the totality of the evidence pointed to a targeted and regulated function of Rpd3 on PHO5 rather than a global one. In summary, histone acetylation levels within the PHO5 promoter are determined by the combined action of the Gcn5 HAT and the counteracting Rpd3 HDAC, which is activated through an unknown mechanism following preinitiation complex formation. For the GAL1 promoter a different HDAC should be involved.

We can put forward two models that explain our observations. One model postulates that the Gcn5 activity is transient and in combination with other activities leads to nucleosome remodeling and transcriptional activation. Gcn5 is then somehow inactivated, and an HDAC reverses the previously generated histone modifications. Alternatively, Gcn5 remains active but is drastically counteracted by an activated HDAC. In both cases the magnitude of reversion depends on transcription rates. The second model proposes an interplay of HATs and HDACs which operates in each round of transcription initiation and reinitiation. Activation of an HDAC or inactivation of the HAT occurs at a particular point in the cycle concomitant with the initiation of RNA synthesis. Attenuated or blocked transcription initiation increases the proportion of promoters captured at a stage in the cycle prior to the activation of the HDAC (or inactivation of the HAT) and results in increased levels of histone acetylation. The cotranscriptional interplay of acetylation and deacetylation could be based on the targeted recruitment of the Gcn5 HAT and the constitutive presence of HDACs preferentially within promoter regions of actively transcribed genes (26). Note that the overall nature of each promoter is an important determinant since different levels of histone acetylation among different promoters do not obviously measure transcriptional efficiencies. Finally, it is noteworthy that, given the fact that nucleosome remodeling is independent of transcriptional rates, both models predict that at least the short term maintenance of a particular nucleosomal structure is histone acetylation independent.

The seemingly contradictory existence of activities which erase the consequences of the transcriptionally important function of Gcn5 is somewhat reminiscent of the recent findings demonstrating the Pol II holoenzyme-dependent destruction of promoter-bound activators during initiation (9, 46). Such “erasing” activities could either serve the process of transcription per se or, alternatively, could be a part of a system that safeguards the competence for rapid reprogramming of transcriptional states. For the cotranscriptional histone acetylation-deacetylation cycle, the latter alternative is consistent with the reported very rapid kinetics for the reversal of targeted acetylation (20, 50). Moreover, the biological importance of such rapid turnover rates is supported by the significant delay in the establishment of PHO5 repression in an rpd3 strain following removal of the inducing signal (48; our unpublished observations). It is interesting that the timely termination of a given transcriptional response and the overall maintenance of chromatin acetylation levels are secured by a mechanism that is built into the transcriptional-activation process itself.