Synergistic Activation of Functional Estrogen Receptor (ER)-α by DNA Methyltransferase and Histone Deacetylase Inhibition in Human ER-α-negative Breast Cancer Cells

Introduction

ER 5 is a ligand-activated nuclear receptor that regulates transcription of estrogen-responsive genes in diverse target cells (1) . ER and its ligand, 17β-estradiol, not only play a critical role in normal breast development but also have long been linked to mammary carcinogenesis, breast tumor progression, and outcome of breast cancer patients (2) . Given the fact that 17β-estradiol stimulates the growth of ER-positive breast tumors via functional ER, endocrine therapy such as antiestrogen or ovarian ablation has been established as an important part of breast cancer management (3) . However, up to one-third of breast carcinomas lack ER at the time of diagnosis, and a fraction of cancers that are initially ER positive lose ER during tumor progression (4) . Genetic alterations, such as homozygous deletion, loss of heterozygosity, or ER gene mutation, have not been reported to play a major role in loss of ER expression. There is increasing evidence that epigenetic alterations play a role in inactivation of ER gene expression (5 , 6) . As demonstrated by Southern and MSP analyses, the ER CpG island is unmethylated in normal breast tissue and most ER-positive tumor cell lines, whereas it is methylated in ∼50% of unselected primary breast cancers and most ER-negative breast cancer cell lines (7) . The methylation of these CpG cluster sites is associated with either reduced or absent ER expression.

The DNMT inhibitor 5-aza-dC is widely used to study the reexpression of genes silenced by promoter methylation (8) . 5-aza-dC exerts its demethylating function through sequestration of DNMT1 to 5-aza-dC-substituted DNA by the irreversible binding of the cysteine in the catalytic domain of the DNMT1 enzyme to the 6 position of the cytidine ring (9 , 10) . In our previous study, treatment of ER-negative human breast cancer cells with the methyltransferase inhibitors 5-aza-cytidine or 5-aza-dC led to partial demethylation of the ER CpG island, reexpression of ER mRNA, and synthesis of functional ER protein. Restoration of ER function was documented by eliciting ERE-driven promoter activity from an exogenous plasmid as well as the expression of the ER-responsive gene, PR (11) .

Inactive chromatin built on methylated CpG clusters is emerging as an important molecular mechanism to silence a variety of methylated tissue-specific and imprinted genes (12 , 13) . This process involves methyl CpG-binding proteins to recruit HDAC family members. These HDACs remove acetyl groups from lysine residues of core histones, particularly H₃ and H₄, to increase ionic interactions between positively charged lysines of histones and negatively charged DNA, thereby generating a more compact nucleosome structure that limits gene activity. In our previous study, TSA, a potent and reversible HDAC inhibitor, could induce ER mRNA by 5-fold. This transcriptional activation is associated with increased sensitivity to DNase I at the ER locus without alteration of the methylated CpG sites, suggesting that an accessible chromatin structure is important in ER expression (6) .

Recently, in vitro studies have shown that DNMT1 interacts physically with either HDAC1 or HDAC2 (14 , 15) , in addition to its ability to methylate hemimethylated CpG sites in DNA. These findings suggested that the ability of DNMT1 to repress transcription after replication through its regulatory NH₂ terminus is partially dependent on HDAC activity. A study using 5-aza-dC followed by TSA treatment of human colon cancer and leukemia cells robustly reactivated multiple methylated genes such as MLH1, TIMP3, CDK2B, and CDK2A. This finding supports the essential roles of both DNMT and HDAC in silencing expression of endogenous methylated genes (16) .

Here we show that cotreatment with DNMT and HDAC inhibitors can synergistically induce ER gene expression in ER-negative breast cancer cells. Induced expression of ER mRNA and protein is associated with expected ER function on estrogen-responsive targets. First, ERE-driven reporter activity from an exogenous plasmid is induced by the treatments and blocked by the ER antagonist ICI 182,780. Secondly, the treatments stimulate expression of the ER-responsive gene, PR, which is again suppressed by ICI 182,780. Our data show that synergistic induction of ER occurs in conjunction with reduced soluble DNMT1 expression and DNMT activity, partial demethylation of the ER CpG island, and increased acetylation of histones H₃ and H₄. These data suggest that pharmacological intervention against both DNMT and HDAC can synergistically reactivate the methylated ER gene expression and restore ER function.

Materials and Methods

Cell Lines, Reagents, Antibodies, and Treatment Protocols.

The human breast cancer cell lines (MDA-MB-231, MDA-MB-435, MCF-7/wt, and T-47D) and their culture conditions were as reported previously (17) . Rabbit antihuman DNMT1 antisera were described previously (18) . Specific anti-acetylated H₃ and H₄ rabbit polyclonal antibodies and PCNA monoclonal antibody were purchased from Upstate Biotechnology (Lake Placid, NY) and Oncogene Research Products (Cambridge, MA), respectively. The anti-ER-α antibody, 1D5, was obtained from Coulter Immunotech (Marseille, France). The pure antiestrogen ICI 182,780 was a generous gift from AstraZeneca Pharmaceuticals (Macclesfield, United Kingdom). For treatment, cells were seeded at a density of 5 × 10⁵ cells/100-mm tissue culture dish. After 24 h of incubation, the culture media were changed to media containing 5-aza-dC for up to 96 h or TSA for 12 h. For the combination study, 5-aza-dC was present in culture for 96 h, and TSA was added for the last 12 h.

RNA Isolation, RT-PCR Analysis of ER and PR, and Quantitative Competitive PCR Analysis of ER.

Total cellular RNA was extracted by Trizol reagent (Life Technologies, Inc., Rockville, MD). RT-PCR was carried out according to our previously described method (6) . RNAs under comparison were simultaneously reversibly transcribed to achieve equal efficiency for reverse transcription. ER mRNA expression was initially screened by RT-PCR. If ER mRNA was present, a previously reported quantitative competitive PCR assay was performed to determine the level of ER transcript in treated MDA-MB-231 and MDA-MB-435 cells as compared with the expression level in untreated cells or the control ER-positive MCF-7 cells (6) . PR RT-PCR primers were described previously (11) , and the PCR reaction was performed with an Advantage GC2 PCR kit (Clontech, Palo Alto, CA) according to the protocol recommended by the supplier.

Genomic DNA Isolation and MSP Analysis of ER CpG Island.

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 uracils as described previously (19) . Methylation status of the bisulfite-modified DNA at the ER locus was characterized by MSP using our previously reported method (7) .

Western Blot Analysis of ER, PCNA, or DNMT1 Expression.

Proteins from detergent-lysed cells were quantified by using the BCA protein assay kit (Pierce, Rockford, IL). The Western blot procedure was reported previously (20) . Briefly, after protein separation by electrophoresis and transfer to nitrocellulose, blots were probed for 1 h with 2 μg/ml either mouse monoclonal antibody 1D5 (ER specific) or Ab-1 (PCNA specific) or affinity-purified rabbit polyclonal IgG (DNMT1 specific) at a 1:1000 dilution in PBS containing 5% dry milk. After incubation with peroxidase-labeled secondary rabbit antimouse or goat antirabbit antibodies, the immunoreactive proteins were detected by the enhanced chemiluminescence method as recommended by the manufacturer (Amersham Pharmacia Biotech, Piscataway, NJ). Protein band intensities were quantified by densitometry (EagleSight Software of Eagle Eye II Imaging System; Stratagene, La Jolla, CA).

DNMT Activity Assay.

Lysates from cells growing in the presence or absence of 5-aza-dC, TSA, or 5-aza-dC and TSA were prepared and assayed for enzyme activity by the incorporation of [³H]methyl-S-adenosylmethionine into poly(dI-dC:dI-dC) as described previously (21) .

Analysis of Acetylated Histones.

Cellular histone extraction was performed using 2 × 10⁶ cells according to the procedure of Yoshida et al. (22) , with the following modifications. The acid (H₂SO₄)-soluble supernatant was precipitated with 10 volumes of cold acetone. After overnight precipitation, histones were collected by centrifugation. The pellet containing histones was dissolved in 50 μl of H₂O, and proteins were quantified by using the BCA protein assay kit.

The extracted histones were separated by electrophoresis on a 15% SDS-PAGE gel and transferred to nitrocellulose membrane. After sequential incubations with rabbit anti-acetylated H₃ and H₄ polyclonal sera and goat antirabbit horseradish peroxidase-conjugated secondary antibody, the immunoreactive proteins were detected by the enhanced chemiluminescence method.

Transfections and Luciferase Assays.

The ERE₂-tk-luciferase/SV-neo plasmid has been described previously (11) . MDA-MB-231 cells were transfected with the plasmid using LipofectAMINE reagent and selected with G418 (1 mg/ml; Life Technologies, Inc.) for 2 weeks. Isolated colonies of G418-resistant cells were pooled and used for analysis. The pooled, transfected cells were maintained in culture media with 1 mg/ml G418. Luciferase activities were determined by Bright-Glo Luciferase Assay System (Promega, Madison, WI), and the activities were normalized for protein concentrations.

Results and Discussion

Induced ER Expression by DNMT and HDAC Inhibitors.

The ER- and PR-negative cell line MDA-MB-231, which has densely methylated ER and PR CpG islands, was used as a cell model to test whether both DNMT and HDAC inhibition could synergistically reactivate ER expression. To test the hypothesis that pharmacological modulation of DNMT and HDAC by 5-aza-dC and TSA could synergistically activate ER gene expression, we first performed dose-response and time-course studies to characterize the effects of each individual drug. Maximal ER gene re-expression was achieved with 100 ng/ml TSA (0.33 μm) for 48 h or with 2.5 μm 5-aza-dC for 96 h (data not shown) when cells were treated with TSA or 5-aza-dC alone. Therefore, to test the hypothesis that combination treatment might have additive or synergistic effects, a treatment strategy of an optimal dose of 5-aza-dC (2.5 μm) for 96 h with the addition of TSA at 100 ng/ml for the last 12 h was used. We postulated that the reactivating effects of the demethylation agent may be potentiated by a short treatment with the HDAC inhibitor. Thus, for the combination studies, MDA-MB-231 cells were treated with 5-aza-dC for 96 h, and TSA was added for the last 12 h. As shown in Fig. 1A ⇓ , ER mRNA was induced 31-fold by treatment with 5-aza-dC for 96 h as compared with vehicle-treated cells, whereas treatment with TSA alone for 12 h had little effect. The combined treatment led to a ∼400-fold induction in ER transcript. Thus, the brief addition of TSA to optimal 5-aza-dC treatment increased ER mRNA by about 13-fold, suggesting a synergistic effect.

To ascertain whether both DNMT and HDAC activities could play a role in repression of ER expression more generally, the dual treatment strategy was evaluated in another ER-negative human breast cancer cell line, MDA-MB-435. Optimal ER gene re-expression was observed after treatment of these cells with either 50 ng/ml (0.165 μm) TSA for 48 h or 0.6 μm 5-aza-dC for 96 h when these drugs were used as single agents (data not shown). The same combination treatment used for MDA-MB-231 cells was used for MDA-MB-435 cells. As shown in Fig. 1B ⇓ , treatment with 5-aza-dC alone induced ER mRNA by 36-fold, whereas treatment with TSA alone for 12 h had little effect. The combined treatment led to a 318-fold induction of ER transcript compared with that of vehicle-treated MDA-MB-435 cells (Fig. 1B) ⇓ . Thus, two different ER-negative cell lines showed evidence of synergistic reactivation of ER by the combined approach of DNMT and HDAC inhibitors.

We next studied how the levels of ER reactivation compared with endogenous ER expression in the human breast cancer cell line MCF-7. As shown in Fig. 1C ⇓ , 5-aza-dC induced ER transcripts in MDA-MB-231 and MDA-MB-435 cells to 4% and 2%, respectively, of the expression level found in MCF-7 cells. The addition of TSA to the 5-aza-dC-treated cultures induced ER mRNA to 50% and 20% of the MCF-7 expression levels, respectively. Thus, the combination treatment consistently and synergistically induced ER transcript in ER-negative breast cancer cell lines, supporting our hypothesis that the activities of both DNMT and HDAC are key regulators in ER gene silencing. The synergistic effect with an optimal exposure to DNMT inhibitor and a suboptimal exposure to HDAC inhibitor suggested a layered action on ER activation.

However, the combination of DNMT and HDAC inhibitors apparently did not reactivate ER mRNA expression to the levels seen in MCF-7 cells. This finding could reflect experimental variation or differences in genotypes between cell lines. Alternatively, it is possible that other mechanisms might also play a role in the silencing process. For example, other transcriptionally repressive complexes such as methyl-CpG binding protein/HDAC could also be involved in the silencing process. Also, some corepressor proteins in various repressive complexes are partially HDAC independent. For instance, a component of the repression mediated by the MeCP2 transcriptional repression domain is partially HDAC independent, and mSin3A could retain some ability to repress transcription, even in the absence of associated HDACs (12) . Also, DMAP-1, a corepressor in the DNMT1/HDAC gene transcriptional repression complex, is HDAC independent (14) . Therefore, simultaneous inhibition of several components in the methylation-associated repressive complexes might be necessary to achieve maximal reactivation of repressed genes.

This effect appeared to be relatively specific for the methylated ER gene because the treatments (alone or in combination) had no apparent effect on the level of expression of the unmethylated housekeeping gene, β-actin, using a competitive quantitative PCR assay (data not shown). However, the combined treatment of the ER-negative breast cancer cells did reactivate expression of other genes silenced by methylation such as RARβ2 or cyclin D2 (data not shown) in addition to ER.

Next, we investigated whether the activated ER transcript was translated into ER protein by Western analysis. The ER protein signal in 5-aza-dC-treated MDA-MB-231 cells was compared with that in the combination-treated cells after normalizing with β-actin protein. As shown in Fig. 1D ⇓ , 5-aza-dC treatment of MDA-MB-231 cells reactivated the M_r 67,000 ER protein, whereas TSA had little effect. The amount of ER protein in combination-treated cells was about twice that in cells treated with 5-aza-dC alone. These data provide further evidence that ER gene expression was enhanced by the combined treatment with 5-aza-dC and TSA as compared with that seen with either drug alone.

Analyses of Activated ER Function in MDA-MB-231 Cells.

The ability of the re-expressed ER protein to mediate an estrogen response was tested next. For this study, a plasmid that contains a firefly luciferase gene under the transcriptional regulation of two EREs driving a thymidine kinase promoter was stably transfected into MDA-MB-231 cells. These cells were treated with 5-aza-dC or TSA alone and with the combination of both (see “Materials and Methods”) in estrogen-containing culture media. As shown in Fig. 2A ⇓ , treatment with 5-aza-dC alone led to a 1.4-fold increase in promoter activity. TSA alone induced the reporter activity by 5-fold, an unexpected finding because 12 h of TSA induced little ER mRNA or protein. Possibly, this is due to the chromatin remodeling effect of TSA instead of ER involvement. Nevertheless, cotreatment with 5-aza-dC and TSA elicited a more than additive induction of reporter activity (about 8-fold; Fig. 2A ⇓ ). Thus, the combination treatment enhanced ERE-driven reporter transcription in MDA-MB-231 cells when compared with single agents.

To confirm that drug-activated reporter activity was a specific estrogen effect mediated through the action of the activated ER, the ER antagonist ICI 182,780 (5 μm) was used to block estrogen effects in the culture (23) . As shown in Fig. 2A ⇓ , ICI 182,780 suppressed 5-aza-dC-stimulated ERE-mediated reporter activity but had no effect on TSA-mediated transcription. These results indicate that 5-aza-dC-activated reporter transcription (5-aza-dC alone or the combination) is a specific hormone effect and suggest that the effect of treatment by TSA alone on this reporter activity is not completely related to estrogen response. Indeed, the exact nature of the effect of TSA on this reporter needs to be further explored.

We further investigated the ability of the drug-induced ER to activate expression of the endogenous ER-responsive gene, PR. As shown in Fig. 2B ⇓ , treatment of MDA-MB-231 cells with 5-aza-dC or the combination induced PR mRNA expression, whereas ICI 182,780 or TSA alone had little or no effect. Cotreatment of MDA-MB-231 cells with 5-aza-dC/ICI 182,780 suppressed PR expression. Similarly ICI 182,780 treatment of cells growing in 5-aza-dC and TSA led to a marked decrease in PR expression. Together, these data suggest that drug-induced ER is functionally active.

Changes in DNMT1 Expression, ER CpG Island Methylation, and Increased H₃ and H₄ Acetylation Introduced by DNMT and HDAC Inhibitors.

To study the underlying mechanisms associated with the synergistic activation of the methylated ER gene, we examined the expression and activity of the methylation maintenance enzyme, DNMT1. The results in Fig. 3A ⇓ showed that addition of 2.5 μm 5-aza-dC to the culture of MDA-MB-231 cells for 96 h markedly depleted soluble DNMT1 protein expression and total DNMT activity from the whole cell protein lysates (Fig. 3B) ⇓ . Interestingly, TSA also down-regulated soluble DNMT1 protein expression with some reduction in DNMT activity. Down-regulation of DNMT1 expression by TSA was consistently obtained in several independent experiments. The combination treatment further reduced the amount of DNMT1 protein and total DNMT activity. Although the effect of 5-aza-dC on soluble DNMT1 protein and total DNMT activity was expected because it sequesters DNMTs after its incorporation into genomic DNA and inhibits its enzyme activity (10) , the effect of TSA was not. One possible explanation is that the effect of TSA on DNMT1 expression and DNMT activity reflects its ultimate antiproliferative effects. A second explanation is that a possible functional or physical interaction exists between DNMT1 and HDAC. Because the cell proliferation marker PCNA interacts physically with DNMT1 (24) , we examined its expression by immunoblot. As shown in Fig. 3A ⇓ , PCNA expression was not obviously modulated by any treatment. Thus, our data suggest a possible functional connection between DNMT1 and HDAC or physical dependency for protein stabilities because no obvious down-regulation of DNMT transcript was observed by the various treatments (data not shown). However, the discrepancy between the level of DNMT1 depletion and reduction in total DNMT activity induced by TSA could also reflect a role of other DNMT family members that are not affected by TSA.

Next, the MSP assay was used to examine the methylation status of the ER CpG island in MDA-MB-231 cells after the treatments (Fig. 3C) ⇓ . The ER CpG island remained methylated in MDA-MB-231 cells treated with vehicle or TSA (100 ng/ml for 12 h). However, treatment with 5-aza-dC alone or in combination with TSA led to partial demethylation of the CpG island, in conjunction with ER gene re-expression. These data are consistent with our previous findings and reaffirm that DNA methylation is a participant in the regulation of ER gene expression (11) .

We then studied whether HDAC inhibition directly acetylates histones and whether the increased histone acetylation is associated with ER gene activation. As shown in Fig. 3D ⇓ , the HDAC inhibitor TSA up-regulated H₃ and H₄ acetylation as expected, although the magnitude of ER gene activation was not clearly proportional to the degree of histone H₃ and H₄ acetylation.

In summary, a demethylated ER promoter sequence with increased histone H₃ and H₄ acetylation by sequential inhibition of both DNMT and HDAC is correlated with the synergistic activation of the ER gene. These data indicate that pharmacological intervention against DNMT and HDAC can lead to alteration of ER gene methylation and chromatin conformation change characterized by hyperacetylated histones. Future studies using coimmunoprecipitation or chromatin-immunoprecipitation assay to elucidate the individual components that form the transcriptionally repressive complex at the ER gene locus will be pursued, and the effects of pharmacological manipulation will be further explored.

Because ER is a critical growth-regulatory gene in breast cancer, it is important to better understand its transcriptional regulation. Our data suggest that a combination drug regimen of both DNMT and HDAC inhibitors can synergistically activate functional ER protein. Additional studies will be necessary to evaluate the clinical implications of this finding with regard to sensitivity to selective ER modulators such as tamoxifen or ICI 182,780.