Steroid hormone receptor-mediated histone deacetylation and transcription at the mouse mammary tumor virus promoter.

Acetylation of lysines in histones H3 and H4 N-terminal tails is associated with transcriptional activation and deacetylation with repression. Our studies with the mouse mammary tumor virus (MMTV) promoter in chromatin show significant levels of acetylation at promoter proximal and distal regions prior to transactivation. Upon activation with glucocorticoids or progestins, promoter proximal histones become deacetylated within the region of inducible nuclease hypersensitivity. The deacetylation lags behind the initiation of transcription, indicating a role in post-activation regulation. Our results indicate a novel mechanism by which target promoters are regulated by steroid receptors and chromatin modification machinery.

The association of histone acetylation with transactivation, and deacetylation with repression, was first suggested in 1964 by Allfrey et al. (1). Such associations are now well documented in numerous studies (2)(3)(4), including several with steroid hormone-regulated promoters (5)(6)(7). Acetylation of specific lysine residues in histone N-terminal tails decreases their net positive charge, which has been proposed to cause an electrostatic repulsion between the histones and the negatively charged phosphate backbone of DNA (8) that results in a more open chromatin conformation. There are, however, data that conflict with this simple model. Mutskov et al. (9) have shown that hyperacetylated and hypoacetylated histone tails can associate almost equally well with DNA in chromatin at physiological salt concentrations. Mizuguchi et al. (10) show that hyperacetylation of histones alone does not stimulate transcription from the adenovirus E4 promoter. An alternative model proposes that the pattern and nature of histone tail modification provides a "code" that can be recognized by specific factors that then associate with the tail, thereby determining the functional consequence of the histone modification (11). The two models are not mutually exclusive.
Circumstantial evidence suggests that transactivation of some promoters is associated with deacetylation. Treatment with histone deacetylase (HDAC) 1 inhibitors, which results in hyperacetylation of histones, can decrease rather than increase transactivation at some promoters or result in no change at others (12). The HDAC inhibitor sodium butyrate inhibits transactivation of the ovalbumin promoter by estrogen receptors (ERs) (13), and of the tyrosine aminotransferase promoter (14) and the MMTV-LTR by glucocorticoid receptors (GRs) (15). Furthermore, Deckert and Struhl (16) have recently shown that in Saccharomyces cerevisiae both acetylated and deacetylated histones H3 and H4 can be associated with transcriptionally active promoters.
Our results demonstrate that at the MMTV-LTR there is a significant level of basal acetylation at the promoter when it is inactive that decreases during hormone activation. This differs from what has generally been found at mammalian gene promoters characterized thus far (7,17), but it has been described at some yeast promoters (Refs. 16,18, and others). The experiments we describe significantly advance our understanding of the role of deacetylation in transcriptional regulation at the MMTV-LTR. The results suggest a mechanism by which steroid hormone-targeted promoter activity is regulated by the acetylation state of histones in addition to other proteins found at the promoter in response to stimuli that induce transactivation.

EXPERIMENTAL PROCEDURES
Cell Culture-The cell lines used in these experiments, 1470.2 and 3017.1, are derived from the same mouse adenocarcinoma parent line, C127i, have multiple copies of stably integrated MMTV-LTR (19,20), and constitutively express mouse GR. Cells were grown in Dulbecco's modified Eagle's medium containing either 10% FBS or charcoalstripped 10% FBS for 16 -20 h prior to treatment with the synthetic glucocorticoid dexamethasone (Dex) (100 nM) or the progestin R5020 (30 nM).
Chromatin Immunoprecipitation (ChIP) Assays-Following treatment with hormone, cells were cross-linked with 0.5% formaldehyde at 37°C for 10 min, nuclei were isolated, and the DNA was digested to predominantly monosomes with micrococcal nuclease (0.1units/g nucleic acid) at 37°C for 10 min. Chromatin immunoprecipitation with antibodies to acetylated histones H3 and H4 was done essentially as described by the Upstate Biotechnology ChIP protocol (Upstate Biotechnology Inc., Lake Placid, NY). After overnight incubation at 4°C, protein A-Sepharose was added for 3 h to pull out immune complexes. Washes were also as described but with the addition of 0.5% deoxycholate. Following cross-link reversal by incubation at 65°C for 4 -6 h, DNA was purified by repeated phenol/chloroform extraction and quantified by fluorimetry (Amersham Pharmacia Biotech). All steps starting with the isolation of nuclei to the cross-link reversal step were done in the presence of 5 mM sodium butyrate. DNA was analyzed by PCR (20 cycles) with primers that amplify MMTV nucleosome B (100-base pair product), 5Ј-TTAAGTAAGTTTTTGGTTACAAACT-3Ј and 5Ј-TCAGAG-CTCAGATCAGAACCTTTGATACC-3Ј; or nucleosome F (120-base pair product), 5Ј-GAGGAAGTTGGCTGTGGTCCTTG-3Ј and 5Ј-TTCGTGC-TCGCAGGGCT-3Ј. PCR products were visualized on 7% nondenaturing polyacrylamide gels stained with SyberGreen (Molecular Probes, Eugene, OR). The linear range of the PCR reaction for each primer set was characterized by titration with a known amount of MMTV-LTR plasmid DNA and was used to determine the appropriate amounts of total immunoprecipitated DNA to use in the PCR reactions. This allows assessment of the acetylation status at a particular nucleosome in the MMTV-LTR in response to hormone treatment by quantification with a Storm FluorImager and ImageQuant analysis (Molecular Dynamics, Sunnyvale, CA).
Nuclear Run-ons-Nuclear run-ons were done as described by Pennie et al. (21). Briefly, nuclei were isolated from treated cells, washed, and resuspended in reaction buffer (10 mM Tris, pH 8.0, 5 mM MgCl 2 , 200 mM KCl, 200 units/ml RNasin (Promega), 1 mM CTP, 1 mM GTP, 2 mM ATP, 2 M UTP, 5 mM dithiothreitol). The reaction was started with the addition of 20 l of [␣-32 P]UTP (3000 Ci/mmol, Amersham Pharmacia Biotech) and allowed to proceed at 26°C for 1 h. Labeled RNA was purified with phenol and RNAzol (Teltest, Inc.). Membranes were prepared by slot blot with fragmented ␤-actin, MMTV-LTR, or pUC18 DNA and hybridized overnight at 60°C. After washing, filters were analyzed using a PhosphorImager and ImageQuant software to quantify the amount of transcript hybridized.

RESULTS
To better understand the mechanisms that underlie chromatin remodeling and transactivation by steroid hormone receptors, we investigated the acetylation status of histones H3 and H4 in the MMTV promoter in response to steroid hormone receptor activation. The MMTV-LTR has six well defined positioned nucleosome families, designated nucleosomes (Nuc) A-F. NucA overlaps the TATA region and the transcription start site, and NucF is the farthest 5Ј to the transcription start site (Ϫ1 kilobase) (Fig. 1A) (22,23). Four GR binding sites or hormone response elements (HREs) are located in the NucB region (Fig. 1A). A nuclease hypersensitive site develops across NucB in response to glucocorticoid treatment, which indicates chromatin reorganization or remodeling at this nucleosome (24 -26). Just 5Ј to NucB there are two HREs in the NucC region, and nuclease hypersensitivity extends into this nucleosome (26,27). Nuclease hypersensitive sites in promoters are generally associated with histone acetylation and with transcriptionally active promoters (28 -30).
We determined the acetylation level of histones H3 and H4 at NucB and NucF, located inside and outside the hypersensitive site, respectively (25). A difference in the acetylation status of histones at these two nucleosomes can indicate whether any observed changes are associated with ATP-dependent chromatin remodeling. The mouse adenocarcinoma cell line 1470.2 was treated with 100 nM Dex for 1 h. Nuclei were isolated and subjected to micrococcal nuclease digestion. Acetylation status was determined using ChIPs assays with antibodies to acetylated lysines in H3 (Lys-9 and -14) and H4 (Lys-5, -8, -12, and -16). There is a significant level of basal acetylation of histone N-terminal tails H3 and H4 at both NucB and NucF prior to treatment with Dex (Fig. 1B, ϪDex). Dex treatment results in a decrease in H3 and H4 acetylation at NucB to about 50% of that seen in the untreated cells (Fig. 1, C and D). In striking contrast, at both H3 and H4 acetylation of NucF is not decreased but is somewhat increased by hormone treatment, indicating that deacetylation is associated with NucB and the nuclease hypersensitive site. It is unlikely that the decrease in acetylation is due to loss or sliding of NucB, as it was shown that GR-induced chromatin remodeling at the MMTV promoter does not involve either of these processes (26).
To understand the kinetic relationship between transcription and deacetylation, a time course of both was carried out (Fig. 2). Treatment of cells with Dex for as little as 15 min results in some deacetylation at NucB at histones H3 and H4; however, transcription has already reached its maximum. Deacetylation reaches its maximum at the 50% level after 30 min of treatment and persists after transcription decreases. By 15 min of treatment, histone acetylation at NucF is somewhat higher than that observed in untreated cells, as shown in Fig.  1, and does not change significantly at any time point (data not shown). These results indicate that deacetylation at NucB is associated with hormone-induce transcription but lags behind transcriptional activation, as deacetylation is still decreasing when transcription is maximal.
The progesterone receptor (PR) is closely related to the GR, binds to the same HREs, and can activate transcription from the MMTV-LTR. Like GR, stably expressed PR is able to induce hypersensitivity at NucB (19). To determine whether the deacetylation at histones H3 and H4 is specific to the GR, cell line 3017.1 (19), which expresses both GR and PR, was treated for 1 h with either Dex or the synthetic progestin R5020, and ChIP assays were done (Fig. 3). Deacetylation is observed at both histones H3 and H4 when either GR or PR is activated in this cell line. We see very little change at NucF after hormone addition, as we observed in the 1470.2 cells. These results suggest that at the MMTV-LTR, related steroid hormone receptors regulate chromatin remodeling and transcription by a common mechanism that involves histone deacetylation. DISCUSSION Histone acetylation within the proximal MMTV promoter region is significant prior to hormone activation, decreases as maximally activated transcription proceeds, and persists after the decline in transcription begins. The relatively high level of basal histone acetylation may facilitate the rapid onset of steroid hormone receptor-activated transcription and nuclease hypersensitivity at the MMTV promoter. Miziguchi et al. (10) report that histone hyperacetylation at an in vitro assembled chromatin template is not sufficient to cause increased transcription but synergistically facilitates transcription induced by the binding of an activator and the activity of the ATP-dependent chromatin remodeling factor, NURF (nucleosome remodeling factor). Additionally, maximal histone deacetylation at the MMTV promoter is achieved just prior to the decrease in transcription, which suggests that deacetylation plays a role in the down-regulation of activated transcription, consistent with the generally observed correlation between deacetylation and transcriptional repression. HDAC inhibitors can, however, decrease rather than increase transcription at the MMTV promoter (15) but not in a time-frame consistent with repression. 2 It is thus possible that the HDAC inhibitors target a nonhistone protein in which acetylation is inhibitory to transcriptional initiation. Acetylation of the non-histone coactivator protein ACTR (activator of thyroid and retinoic acid receptors) coincides with inhibition of ER-mediated transcription (5).
The dynamic pattern of histone acetylation we observed at the MMTV promoter is different from two other systems examined in kinetic detail. Experiments with the estrogen-responsive cathepsin D and the pS2 promoters (5,7) show low levels of histone acetylation prior to treatment with estradiol, which then rise and reach a peak just prior to the maximum of transcription at 60 min. Elevated levels of acetylation then persist to some degree as transcription declines and increases again. Reinke et al. (18) describe a transient hyperacetylation at the PHO8 promoter in yeast that occurs prior to chromatin remodeling and transcriptional activation. However, unlike our observation at the MMTV-LTR, levels of acetylation did not drop below those observed prior to activation. The variety of acetylation patterns and the dynamics of timing observed in these promoter systems suggests that different genes utilize histone and non-histone protein acetylation in distinct ways to regulate transcriptional activity. The mechanisms by which acetylation functions in transcriptional regulation are likely more complex than simply electrostatic repulsion between acetylated histones and DNA, or a histone code that directs non-histone proteins to the promoter, and are not yet fully understood.
FIG. 3. PR can mediate deacetylation of histones H3 and H4 at the MMTV promoter. 3017.1 cells were treated for 1 h with either 100 nM Dex or 30 nM R5020. It was shown previously that R5020 does not activate the GR in these cells at the concentration used in these experiments (19). A, after hormone treatment, ChIP assays with antibodies to acetylated histones H3 and H4 and PCR analysis of various fractions were done as described for Fig. 1. Control represents cells treated with vehicle only and shows the basal level of histone acetylation. "None" is the no antibody control. B and C, statistical analysis of the results from four independent experiments. Values from each experiment were normalized and expressed as described in Fig. 1 legend. Error bars represent S.E.