Transcriptional Activation of Estrogen Receptor α in Human Breast Cancer Cells by Histone Deacetylase Inhibition¹

Interaction between 17β-estradiol and ERα³ plays an important role in breast carcinogenesis and breast cancer treatment. That estrogen stimulates the growth of certain breast cancers is well established, and hormonal therapy via estrogen depletion or antiestrogen administration is widely used to block the action of estrogen in women with breast cancer. However, patients whose breast cancers lack ER seldom respond to endocrine therapy; therefore, a potential mechanism for hormone resistance is de novo or acquired loss of ER gene expression at the transcriptional level during disease progression (1, 2).

One possible mechanism for loss of ER in ER-negative breast cancers is cytosine methylation of the ER CpG island in the 5′regulatory region of the gene (3). Indeed the ER gene CpG island is extensively methylated in ER-negative breast cancer cells, and ∼50% of unselected primary breast tumors but remains unmethylated in normal breast tissue and many ER-positive tumors and ER-positive cancer cell lines (4, 5). The functional importance of this finding is demonstrated by the fact that treatment of ER-negative human breast cancer cells with the demethylating agent, 5-aza-dC, led to reactivation of expression of ER mRNA and functional ER protein (6). Recent studies indicate that silencing of a gene by methylation involves the generation of an inactive chromatin structure characterized by deacetylated histones. An abundant chromosomal methyl CpG-binding protein, MeCP2, was the first protein identified to link methylated DNA and a HDAC-containing transcriptionally repressive complex for gene silencing. Subsequently, several MBD proteins have been identified that, similar to MeCP2, couple methylated DNA to HDAC (7, 8). More recently, the well-known maintenance methyltransferase,Dnmt1, was found to interact physically with HDAC through its N terminus, thereby forming a transcriptionally inactive chromatin structure that represses transcription (9, 10). All of these findings demonstrate the important role of HDAC in transcriptional regulation. The HDACs deacetylate lysine groups of histones H3 and H4, allowing ionic interactions between positively charged lysines and negatively charged DNA, which result in a more compact nucleosome structure that limits transcription. The availability of specific HDAC inhibitors such as TSA (11)permits the study of the role of HDAC in silencing a variety of tissue-specific methylated genes (7, 12).

Here, we have tested whether loss of ER expression in some breast cancers is associated with transcriptional repression through HDAC activity on the methylated ER gene. Our data demonstrate that specific HDAC inhibition via TSA treatment can reactivate ER transcription in the presence of the methylated DNA. The activated gene transcription is associated with increased sensitivity of the ER promoter to DNase I treatment. These data suggest that inactive chromatin mediated by HDAC is critical to ER gene silencing.

Human breast cancer cells (Hs578t, MCF-7/WT,MCF-7/Adr^R, T-47D, and MDA-MB-231) were grown in DMEM supplemented with 10% fetal bovine serum and 2 mml-glutamine. TSA was purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan), dissolved in absolute ethanol at a stock concentration of 1 mg/ml (3.30 mm), and stored at−20°C. DNase I was purchased from Pharmacia Biotech (Pharmacia Biotech, NJ). Cells (MDA-MB-231, Hs578t, T-47D, or MCF-7/Adr^R) were seeded at a density of 6 × 10⁵ cells/100-mm tissue culture dish (8 × 10³cells/cm²). After 24 h of incubation, the culture medium was changed to different concentrations of TSA- or vehicle (ethanol)-containing medium. Either total cellular RNA or genomic DNA was isolated after 0, 24, 48, or 72 h of TSA exposure.

Total cellular RNA was isolated from cell lines with TRIzol reagent according to the recommendations of the supplier (Life Technologies,Inc., Rockville, MD). RNA (3 μg) was reversibly transcribed by Moloney murine leukemia virus reverse transcriptase (Life Technologies)using OligoDT₁₅ primer (Promega Corp., Madison,WI) in a final volume of 50 μl. Four % of synthesized cDNA (2 μl,derived from 150 ng of initial RNA) was used for PCR amplification of ER and the constitutively expressed housekeeping gene β-actin (13). Specific sense and antisense PCR primers used for the amplifications across the seventh intron of ER and the first intron of β-actingenes, yielding 470 and 400 bp of PCR products respectively, were described previously (6). PCR products were resolved by 2% agarose gel electrophoresis and visualized by ethidium bromide staining.

A quantitative assay was performed to determine the level of ER mRNA in TSA- and vehicle-treated MDA-MB-231 cells as compared with the expression levels in the known ER-positive MCF-7 and T-47D cells using the method of Wang and Rowley (14). This assay involves coamplification of a wild-type target cDNA (wER) of unknown amount and a competitive template (cER) in known amounts. A truncated competitive template was generated with a sense primer containing a 22-bp deletion(primer 2) and the same antisense primer (primer 3) as the wild-type(Fig. 2 A). After amplification, the competitive template was separated, gel purified, and quantified by Spectrophotometer DU 640(Beckman, CA). Thus, target wild-type and known amounts of competitive templates can be coamplified with the pair of wild-type primers (primers 1 and 3) and differentiated by size. Because the most accurate results are obtained when wild-type and competitive templates are amplified at nearly equivalent concentrations, resulting in the signal ratio of wER:cER equivalent to 1, we first performed an initial titration in log and then in 2-fold dilutions to determine the approximate concentration of the wild-type ER cDNA in our experimental samples. RNAs under comparison were simultaneously reversibly transcribed to achieve equal efficiency for reverse transcription. The PCR reactions were carried out with 0.5 μm of wild-type sense and antisense primers for 35 cycles. The wild-type and competitive PCR products were fractionated on 2.5% agarose gel,stained with ethidium bromide, and scanned by Densitometer (EagleSight Software of Eagle Eye II Imaging System; Stratagene, La Jolla, CA). The ratio between wild-type and control templates was determined and used to calculate the amount of target wild-type cDNA because the input of competitive template is known.

DNA was isolated by standard phenol-chloroform extraction. Isolated DNA was subjected to modification by sodium bisulfite to convert unmethylated cytosines but not methylated cytosines to uracil as described previously (15). Methylation status of the bisulfite-modified DNA at the ER locus was characterized by methylation-specific PCR using a method described previously (5).

This assay was performed according to the method of Keshet et al. (16) using cells from TSA- or vehicle-treated MDA-MB-231 and MCF 7 cells (1 × 10⁷ cells/each) with the following modifications. The isolated DNA was digested with EcoRI, the recognition sites of which flank the ER promoter region to yield a 3.1-kb fragment that was separated by 1% agarose gel electrophoresis. DNAs blotted on nylon membrane were probed with a PCR-amplified, 561-bp DNA fragment corresponding to the ER CpG island (Fig. 4 A). The sense and antisense oligonucleotides used to amplify the fragment are 5′-AGACCAGTACTTAAAGTTGGAGGCC-3′ and 5′-GGGAAACCCCCCAGG-3′. The amplified DNA was cloned into pCR2.1-TOPO vector (Invitrogen,CA) according to the manufacturer’s protocol. Colonies containing amplified DNA sequence, determined by Mini-Prep (Promega Wizard Mini-Prep kits), were grown and purified. The purified plasmids were sequenced via automated sequencing (Johns Hopkins Sequencing Core Facility). The specific ER CpG probe was prepared from sequence-confirmed plasmid and labeled with bio-16-dUTP (Boehringer Mannheim, IN) by PCR using the above-mentioned primer set. The 3.1-kb DNA band containing the ER promoter region was visualized by chemiluminescence using a streptavidin-conjugated,alkaline phosphatase-catalyzed substrate, CDP-star. The band signals that reflect the resistance to DNase I were quantified by densitometry(Stratagene), and the cumulative DNase I sensitivity was calculated as follows. The band density at each DNase I concentration was divided by the density of the control band and then multiplied by 100%. This value was subtracted from 100% to yield the percentage of DNase I sensitivity. The sum of the percentage of DNase I sensitivity at each dose was considered to be the cumulative DNase I sensitivity.

A growing body of data demonstrates the importance of histone acetylation and deacetylation and corresponding structural alteration of chromatin in gene transcriptional regulation (7, 8, 10). The ER-negative cell line, MDA-MB-231, the ERCpG island of which is densely methylated, was used as a cell model to test whether HDAC activity contributes to repression of ER expression in ER-negative breast cancer cells. TSA, a specific and potent HDAC inhibitor, was used as a pharmacological tool. Exposure of MDA-MB-231 cells to increasing concentrations of TSA led to induction of ER mRNA synthesis in a dose-dependent manner (Fig. 1,A). A detectable level of ER mRNA, as demonstrated by a RT-PCR product with predicted 470-bp size,was noted after treatment with 50 ng/ml (160 nm)TSA for 48 h, and ER transcript was clearly present after 100 ng/ml (330 nm) TSA for the same duration. A time course analysis showed that a weak signal could be seen after 24 h of 100 ng/ml TSA treatment, whereas ER transcript was readily observed after 48 or 72 h (Fig. 2 B). Multiple experiments were done to achieve optimal conditions for TSA induction of ER mRNA in MDA-MB-231 cells. These showed that maximal ER reactivation was achieved with 100 ng/ml TSA for 48 h using an initial seeding density of 8 × 10³cells/cm². Higher inoculating cell densities reduced ER transcript signal (data not shown).

To ascertain whether HDAC activity could play a role in repression of ER expression more generally, the dose response and time course studies described above were extended to other ER-negative human breast cancer cells lines with methylated ER CpG islands. As shown in Fig. 1 C, TSA treatment led to re-expression of ER mRNA in all three ER-negative cell lines tested. Optimal ER gene re-expression was observed after treatment of MDA-MB-231 cells with 100 ng/ml (330 nm), Hs578t cells with 400 ng/ml (1.32μ m), and MCF-7/Adr^R with 25 ng/ml (82.5 nm) TSA for 48 h. Therefore,TSA treatment consistently induced ER re-expression in the panel of ER-negative cell lines, supporting a role for HDAC in ERgene silencing.

A quantitative competitive PCR assay was used to assess the magnitude of TSA-induced ER mRNA transcript in MDA-MB-231 cells. Fig. 2,A shows the design of the primer sets used, and validation of the quantitative competitive PCR is shown in Fig. 2,B. As shown in Fig. 2,C, a 5-fold increase in ER transcript was obtained after TSA exposure in MDA-MB-231 cells (100 ng/ml for 48 h). This effect was specific for the methylated ER promoter because TSA treatment (50 or 100 ng/ml for 48 h) of ER-positive, unmethylated MCF-7 cells had no effect on the level of ER mRNA expression using the same quantitative assay (data not shown). However, TSA treatment of MDA-MB-231 cells did not restore ER mRNA expression to the levels seen in cell lines with endogenous ER expression, as shown in Fig. 2 C. Quantitative assay suggested that the level of ER transcript seen with TSA treatment of MDA-MB-231 cells represented about 1 and 10% of that seen in the ER-positive MCF-7 and T-47D cell lines, respectively. Several possibilities might account for this:

(a) It is possible that only partial reactivation is seen because only a fraction of cells responded to the treatment. Indeed, a similar pattern of partial reactivation was seen in MDA-MB-231 cells exposed to a demethylating agent, 5-aza-dC, in our previous study(6).

(b) It has been shown that a component of the repression mediated by MeCP2 transcriptional repression domain is partially HDAC independent; mSin3A could retain some ability to repress transcription,even in the absence of associated HDACs (7).

(c) Simultaneous inhibition of several components in the methylation-associated repressive complexes might be necessary to achieve maximal reactivation of the repressed genes (12).

The ability of TSA to reactivate ER expression raised the question of whether the ER CpG island remained methylated. Our previous studies with 5-aza-dC demonstrated that ER re-expression was associated with demethylation of the ER CpG island. However, a parallel study of the progesterone receptor in progesterone receptor-negative human breast cancer cells showed that the ligand-bound ER could overcome methylation-related repression of the progesterone receptor,even in the continuing presence of a methylated progesterone receptor CpG island (17). We therefore examined the methylation status of the ER CpG island in TSA-treated MDA-MB-231 cells using a sensitive MSP assay that allowed examination of methylation status across the ER CpG island (Fig. 3,A). As shown in Fig. 3,B, the ER CpG island was completely methylated across the entire CpG island in MDA-MB-231 cells treated with vehicle or TSA (100 ng/ml for 48 h). A single primer set, ER 5, was used to confirm this finding in the other TSA-treated, ER-negative human breast cancer cells, Hs578t and MCF-7/Adr^R (Fig. 3 C). As expected, the unmethylated ER-positive, MCF-7 cells demonstrated an unmethylated pattern using all four primer sets that span the ER CpG island. In summary, TSA treatment of ER-negative breast cancer cells can lead to re-expression of ER mRNA without an apparent alteration in the methylation status of the ER CpG island.

Because acetylated histones are generally associated with transcriptionally active chromatin whereas deacetylated histones are often found in conjunction with an inactive chromatin state(18), we next studied whether HDAC inhibition could alter chromatin structure at the ER gene locus. Because nuclease susceptibility is one of the characteristics of active chromatin(19), we used a DNase I sensitivity assay to examine chromatin conformation of the ER gene in ER-negative MDA-MB-231 cells in the presence or absence of the HDAC inhibitor, TSA. Cells were treated with 100 ng/ml of TSA for 48 h, a treatment course shown previously to result in optimal re-expression of ER mRNA. Equal amounts of purified nuclei from control and TSA-treated MDA-MB-231 cells were exposed to increasing concentrations of DNase I as described in “Materials and Methods.” Nuclei isolated from ER-positive MCF-7 cells served as a DNase I accessible control. As expected, the ER locus in MCF-7 cells was a highly DNase I sensitive region, whereas the ER locus in control MDA-MB-231 cells was relatively resistant to DNase I digestion (Fig. 4,B). TSA treatment of MDA-MB-231 cells resulted in an ∼2-fold increase in DNase I sensitivity (Fig. 4 C), suggesting that inhibition of HDAC activity leads to a more open chromatin conformation, even in the presence of CpG island methylation.

Recently, the interaction between DNA methylation and histone deacetylation linked by methyl-binding proteins (MBDs), or the direct interaction of Dnmt1 with HDAC as well as other corepressors, has been an area of active study. More recently, a nucleosome-stimulated ATPase Mi2, a part of chromatin remodeling machinery, was also shown to bind the methylated DNA through MBD3 and deacetylase in Xenopus laevis and mammalian cells, further illustrating the role of HDAC on gene transcription regulation (8). In addition to evaluating the role of HDAC in in vitro studies, it is of importance to study its role in silencing endogenous methylated genes. In some cases, HDAC inhibition alone seems to be sufficient to reactivate a methylated gene. For example, both sodium butyrate and TSA can restore transcription from methylated and silenced plant rRNA genes(20). Also, reactivation of transcription of the methylated FMR1 gene was achieved by treatment with 4-phenylbutyrate, sodium butyrate or TSA (21). However, in other cases, both demethylation and HDAC inhibition appear to be necessary. For example, certain hypermethylated genes like MLH1,TIMP3, CDKN2B, and CDKN2A can be transcriptionally activated in colon cancer cells by TSA only after Dnmt1 inhibition by 5-aza-dC, suggesting an important role of Dnmt1 in transcription repression although recruiting HDAC is essential (12).

ER is a critical growth-regulatory gene in breast cancer, and its expression status is tightly linked to the prognosis and treatment outcome of breast cancer patients. Thus, it is important to understand its regulation. Our work suggests that histone deacetylation and DNA methylation may both play a role in ER transcription, and further studies will focus on the effects of TSA on ER protein expression. This is critical because it is possible that activation of the silenced ER by HDAC inhibition could open a new avenue for management of a subset of advanced breast cancer with hormonal resistance. Studies using primary breast cancers have shown that the antiestrogen, tamoxifen, confers a benefit to women whose breast cancer expresses ER by immunohistochemistry in as few as 1–10%of tumor cells. Thus, even partial re-expression of ER could be of clinical benefit (22).