Interaction of TFAP2C with the Estrogen Receptor-α Promoter Is Controlled by Chromatin Structure

Breast cancer is currently the most common cancer and the second most common cause of cancer deaths among women in the United States (1). Predicting the biological response to hormonal treatment and chemotherapy depends upon the overexpression of certain markers, including estrogen receptor-α (ERα) and ErbB2/HER2/neu (2, 3). Approximately 75% of breast cancers overexpress ERα, and overexpression of the receptor predicts an improved prognosis and response to hormonal therapy compared with cancers with low or absent ERα expression (4, 5). Given the effectiveness and low complications associated with hormonal therapy, the possibility of inducing hormone response in ERα-negative cancer remains an attractive treatment strategy.

Recent work on the regulation of ERα expression in breast cancer suggests that both epigenetic chromatin structure and trans-active factors play a role in the regulation of receptor expression. Differences in expression of the receptor comparing ERα-positive and ERα-negative cell lines can be explained by differences in transcription of the ERα gene (6). Transcription also seems to be the mechanism of control in approximately one third of ERα-negative primary breast cancers, and tumors with low or absent ERα mRNA have a worse tumor grade (7). Methylation of the CpG island in the ERα promoter has been found in certain ERα-negative breast carcinoma cell lines and tumors (8, 9). Treatment of the ERα-negative breast cancer cell line MDA-MB-231 with 5-aza-2′-deoxycytidine (AZA) was reported to reactivate ERα expression, and a synergistic induction of ERα expression was possible with the combination of AZA and trichostatin A (TSA; refs. 10, 11). However, other groups examining the MDA-MB-231 cell line failed to confirm that ERα expression was induced with AZA or TSA treatment (12, 13). Another line of investigation has focused on the activity of certain transcription factors involved in regulating ERα expression in breast cancer. The forkhead transcription factors FoxM1 and FOXO3a have been implicated in the regulation of the ERα promoter (14, 15). TFAP2C has been shown to activate expression of the ERα promoter, and binding of AP-2 factors induces a prominent hypersensitive site in the chromatin near the transcriptional start site (16, 17). More recently, it has been reported that elimination of TFAP2C in a variety of ERα-positive cell lines significantly reduces the expression of ERα and eliminates functional hormone response (18).

Finding evidence for the regulation by both epigenetic structure and trans-active factors raises an interesting question as to how these two mechanisms might interact in the control of ERα expression. At the present time, there has not been a consistent and reproducible means for inducing ERα expression in ERα-negative breast carcinomas. We hypothesize that epigenetic chromatin structure can influence the functional activity of TFAP2C at the ERα promoter. To address this possibility, we have examined how the combination of epigenetic structure of the ERα promoter and forced overexpression of TFAP2C participate in control of ERα promoter activity.

Cell lines. Cell lines were obtained from American Type Culture Collection and maintained as previously described (19). The MCF7-5C cell line was a generous gift from Dr. Craig Jordan (20). AZA and TSA (Sigma Aldrich) were used at the final concentrations of 2.5 μmol/L and 100 ng/mL, respectively (11). For experiments with drug treatment, cells were treated with AZA for 4 to 5 d and with TSA for 12 to 24 h before analysis. The adenoviral vector expressing TFAP2C was previously described (17).

Analysis of CpG island methylation. CpG DNA methylation was determined by cloning and sequencing multiple clones recovered from bisulfite-treated DNA using the MethylSEQr reagent kit (Applied Biosystems), which was used according to the manufacturer's instructions.

Chromatin immunoprecipitation analysis. Chromatin immunoprecipitation (ChIP) experiments were carried out as previously described with minor modifications (18, 21, 22). Twenty million cells were cross-linked for 10 min at 37°C using 1% formaldehyde. Cells were washed twice with PBS, pelleted by centrifugation, and resuspended in 500 μL sonication buffer [50 mmol/L Tris-Cl (pH 8.1), 10 mmol/L EDTA, and 1% SDS + protease inhibitors] and sonicated using conditions determined empirically for each cell line to achieve an optimal fragment length between 1,000 to 300 bp. After sonication, samples were centrifuged at 20,000 × g for 10 min at 4°C. The supernatant containing cross-linked DNA/histones was diluted 1:10 using IP dilution buffer [0.01% SDS, 1.1% Triton-X 100, 1.2 mmol/L EDTA, 16.7 mmol/L Tris-Cl (pH 8.1), 167 mmol/L NaCl] plus protease inhibitors. Samples were precleared using 75 μL of protein A agarose (Upstate Biotech) for 1 h at 4°C. For histone H3 analysis, about one tenth of the sample was removed as input control, whereas the remainder was split into two reactions for immunoprecipitation with acetylated histone 3 lysine 9 (H3K9; Upstate Biotech) or control nonspecific IgG (Sigma). For TFAP2C analysis, anti-TFAP2C antibody SC-12762X (Santa Cruz Biotechnology) was used; for RNA polymerase II (Pol II), anti–Pol II antibody CTD4H6 (Upstate/Millipore) was used. Antibodies were allowed to recognize their antigens overnight at 4°C with agitation. Protein/antibody/DNA complexes were collected using 70 μL protein A agarose (Upstate Biotech) followed by washing and elution according to the manufacturer's suggested protocol. Protein/DNA cross-links were reversed using 200 mmol/L NaCl at 65°C for 4 h. DNA was recovered from input ChIP reactions using the Qiagen DNeasy kit (Qiagen) according to manufacturer's suggested protocol. Purified DNA was quantified by using a NanoDrop ND-1000 (NanoDrop). Samples were analyzed by SYBR green quantitative real-time PCR with primers designed to glyceraldehyde-3-phosphate dehydrogenase (GAPDH; GAPDHf 5-CGTCCTTGACTCCCTAGTGTCCT-3, GAPDHr 5-CCTACTTTCTCCCCGCTTTTTT-3) and ERα (ERαPf 5-TAACCTCGGGCTGTGCTCTT-3, ERαPr 5-TTCCCTTGGATCTGATGCAGTAG-3). For PCR reactions analyzed by gel electrophoresis, PCR amplification of the ERα gene promoter was done with the primer pair: ER CHIP 2 5′(5′-gctgtgctctttttccaggt-3′), ER CHIP 2 3′ (5′-ttgctgctgtccaggtacac-3′). For quantitative ChIP, amplification of target amplicons was monitored as a function of increased SYBR green fluorescence. An analysis threshold was set, and the cycle threshold (Ct) was computed for each sample. Fold enrichment of target sequence was calculated using the following formula (fold enrichment = 2^{[(Ct Ip'ed) − (Ct Input)]}.

Chromatin accessibility. The accessibility of cis-regulatory elements within GAPDH and ERα were carried out as previously described with minor changes (23). Two million cells were washed using PBS, collected by centrifugation, and resuspended, and nuclei were collected using RBS buffer [10 mmol/L Tris (pH 7.5), 10 mmol/L NaCl, 3 mmol/L MgCl₂, and 0.05% NP40]. Nuclei preparations were collected by centrifugation and washed with the digestion buffer [50 mmol/L Tris (pH 7.5), 100 mmol/L NaCl, 10 mmol/L MgCl₂, and 1 mmol/L DTT] to remove any residual RBS buffer. Isolated nuclei were resuspended into 100 μL of digestion buffer. Half of the nuclei were left untreated to serve as uncut control nuclei, whereas the remaining nuclei were treated with 20 units of RQ1 DNase (Promega) and digested at 37°C for 5 min. After digestion was completed, the reactions were stopped by adding 15 mAU proteinase K and digesting reactions at 65°C for 15 min. DNA was purified using Qiagen DNeasy blood and tissue kit (Qiagen) and quantified by spectrophotometry. The accessibility indices of cis-regulatory regions were determined using quantitative real-time PCR, in conjunction with the same primer sets used for quantitative ChIP experiments. Triplicate reactions were setup (5 pmol of each primer, 10 ng of DNA, and 2× SYBR Green PCR master mix) and run using a ABI 7000 sequence detection system (Applied Biosystems) using the following conditions: 10-min denaturation at 95°C, followed by 40 cycles of 94°C for 30 s and 60°C for 30 s. Amplification of both target amplicons was monitored as a function of increased SYBR green fluorescence. The accessibility index for each amplicon was determined by the following formula: AI = 2^{[(Ct Dnase treated) − (Ct Uncut)]}).

Reverse transcription-PCR. Real-time quantitative reverse transcription-PCR (RT-PCR) was done by isolating RNA from the using the RNeasy Mini protocol (Qiagen). cDNA was synthesized using SuperScript III first-strand synthesis for RT-PCR (Invitrogen), Assay-on-Demand Gene Expression FAM/MGB Probe, nonprimer limited probes for human ERα (Hs01046818_m1) or human β-actin endogenous control (4333762F), and TaqMan Universal PCR Master Mix. No AmpErase UNG (Applied Biosystems) were used to perform quantitative PCR using ABI Prism Model 7700 Sequence Detector. In each qPCR assay, samples were run in triplicate, along with three no-template controls and a standard curve. Data were analyzed using ABI Prism 7700 Sequence Detection System software (Applied Biosystems).

Statistical analysis. Statistical analysis of ChIP and chromatin accessibility was determined using a one-way ANOVA analysis with Tukey's multiple comparison posttest. All statistical analysis was done using GraphPad Prism Version 4.0 (GraphPad Software, Inc.).

The pattern of expression for TFAP2A, TFAP2C, and ERα. Previous studies established that TFAP2C controls hormone response in breast carcinoma cells through the regulation of multiple pathways of estrogen signaling. One mechanism of estrogen response involves the ability of TFAP2C to activate ERα expression by binding to AP-2 response elements in the ERα promoter. We hypothesized that overexpression of TFAP2C might be able to activate ERα expression in ERα-negative breast carcinoma cell lines. A number of breast carcinoma cell lines were examined for their pattern of expression of TFAP2A, TFAP2C, and ERα. As seen in Fig. 1, the ERα-negative breast carcinoma cell lines MCF7-5C and MDA-MB-231 do not express significant amounts of TFAP2C protein compared with the ERα-positive cell lines MCF7 and T47-D. Of particular interest was the pattern of AP-2 expression in the MCF7-5C cell line, which was derived from the MCF7 parental line by long-term culture in the absence of estrogen (20). The pattern of AP-2 expression in MCF7-5C was reevaluated compared with the MCF10A cell line, which is a commonly used model for normal mammary epithelial cells. The data confirm that hormone-independent MCF7-5C cells have lost TFAP2C expression in concert with their loss of ERα expression compared with the parental MCF7 cell line. Interestingly MCF7-5C was also found to overexpress TFAP2A compared with MCF7, whereas MDA-MB-231 was noted to lack expression of either TFAP2A or TFAP2C. The findings in the ERα-negative cell lines suggested the possibility that overexpression of TFAP2C might induce expression of ERα in MCF7-5C and MDA-MB-231. However, forced overexpression of TFAP2C delivered by adenoviral vector failed to induce expression of ERα in these ERα-negative cell lines (see below and additional data not shown).

Epigenetic marks of the ERα promoter in ERα-negative cell lines. The inability for TFAP2C to induce ERα expression in ERα-negative cells suggested the possibility that epigenetic changes in the chromatin structure of the ERα gene might play a role in blocking the transcriptional effects of TFAP2C. First, we examined the DNA methylation pattern of the CpG island in the proximal region of the ERα gene promoter. The promoter region analyzed encompassed 28 potential CpG methylation sites and included CpGs incorporated into the AP-2 binding sites, which are required for TFAP2C-mediated activation of the ERα promoter. As seen in Fig. 2, the CpG island in the ERα-positive cell lines MCF7 and T47-D was completely unmethylated. By comparison the region in MCF7-5C cells was found to have complete DNA methylation except for one site in each of two clones. Various degrees of CpG methylation was found in MCF10A and MDA-MB-231 with all clones from these two cell lines showing some CpG site methylation, save for one clone derived from MDA-MB-231. Hence, it seemed plausible that DNA methylation might be one factor inhibiting TFAP2C from inducing ERα expression in ERα-negative cell lines.

Another chromatin mark associated with differences in transcriptional activity is histone acetylation. We examined H3K9 acetylation in the proximal ERα gene promoter using quantitative ChIP analysis. As shown in Fig. 3, there were significant differences in Ac-H3K9 marks comparing ERα-positive and ERα-negative cell lines. Chromatin from MCF7 and T47-D showed Ac-H3K9 marks, whereas the ERα-negative cell lines MCF10A, MDA-MB-231, and MCF7-5C had no evidence for acetylated histones in the ERα promoter. As a positive control, all cell lines showed Ac-H3K9 marks in the GAPDH promoter region, and there were no significant differences comparing the ERα-positive and ERα-negative cell lines.

The finding that the chromatin at the ERα CpG island in ERα-negative cells was more highly DNA methylated and associated with deacetylated histones suggests that the ERα promoter might not be accessible to factors such as TFAP2C. To address this possibility, we compared the ERα promoter region in the different breast cancer cell lines for DNA accessibility. As seen in Fig. 4A, the relative DNA accessibility of the ERα promoter region was significantly lower in the ERα-negative cell lines compared with the ERα-positive lines. However, DNA accessibility for the GAPDH gene was equivalent for all the cell lines examined. AZA can be used to inhibit CpG methylation, and TSA is a histone deacetylase inhibitor that can induce histone acetylation. We examined whether treatment with AZA and TSA could increase the DNA accessibility at the ERα promoter in the ERα-negative cell lines. As seen in Fig. 4B, treatment with AZA and TSA successfully increased the DNA accessibility at the ERα promoter in all the ERα-negative cell lines. Interestingly, there was a trend for AZA/TSA treatment to decrease the accessibility index at the GAPDH locus, which did not reach statistical significance.

AZA/TSA alters ERα inducibility. Based on the finding that treatment with AZA and TSA increased DNA accessibility, we examined whether treatment with these drugs altered the ability for TFAP2C to activate the ERα promoter. In the absence of AZA/TSA, TFAP2C had a modest effect on ERα expression in MDA-MB-231 cells (see Fig. 5A). Drug treatment with AZA alone induced a moderate increase in ERα expression of ∼10-fold and the combined treatment with AZA/TSA increased expression of ERα by ∼15-fold. These results are consistent with other published works showing that combined treatment with AZA/TSA had a synergistic effect on ERα induction (11). Similar results were obtained with TFAP2C overexpression in the presence of each drug individually. However, in the presence of both AZA/TSA, TFAP2C induced an 80-fold increase in ERα expression in MDA-MB-231 cells. The level of ERα expression achieved by the combination of AZA/TSA and TFAP2C overexpression was ∼1% to 5% the level of ERα mRNA in MCF7 cells. Because the level of ERα expression was still relatively low compared with MCF7, the expression of the ERα-responsive gene RERG was also examined (Fig. 5B). The expression pattern for RERG mirrored the increase in ERα expression. TFAP2C alone had no significant effect on RERG expression compared with untreated MDA-MB-231 cells. Overexpression of TFAP2C in the presence of either TSA or AZA alone had a modest effect, whereas overexpression of TFAP2C in the presence of both AZA/TSA resulted in a significant induction of RERG expression. These data indicate that induction of ERα in this system has functional effects on the expression of hormone responsive genes. The effect of TSA and AZA treatment on TFAP2C induction was similarly examined in MCF7-5C cells (Fig. 5C). In the absence of drug, TFAP2C had no significant effect on ERα expression and AZA/TSA alone had a modest effect. However, in the presence of AZA/TSA, overexpression of TFAP2C induced a 16-fold increase in ERα expression.

The ability for AZA/TSA to increase chromatin accessibility and to allow functional activity of TFAP2C at the ERα promoter suggested that drug treatment induced alterations of chromatin structure, allowing TFAP2C to bind to the AP-2 regulatory region of the ERα promoter. We formally investigated this possibility by examining TFAP2C binding to the ERα promoter by ChIP analysis. Consistent with earlier results, binding of endogenous TFAP2C to the ERα promoter can be detected in ERα-positive cells, but not ERα-negative cells (see Fig. 6A). Because MCF7-5C do not express endogenous TFAP2C (see Fig. 1), ChIP analysis with anti-TFAP2C failed to enrich for the ERα promoter and treatment with AZA/TSA did not alter this finding. Similarly, in the absence of AZA/TSA, ChIP analysis failed to detect an interaction between TFAP2C and the ERα promoter even with overexpression of TFAP2C. However, ChIP analysis in cells with overexpression of TFAP2C in the presence of AZA/TSA treatment showed binding of TFAP2C to the AP-2 regulatory region of the ERα promoter. Quantitative ChIP analysis was used to estimate the relative fold enrichment of pull-down of the ERα promoter with anti-TFAP2C antibody (Fig. 6B). Compared with MCF7-5C with overexpression of TFAP2C, AZA/TSA induced a 6-fold increase in binding of TFAP2C detected by ChIP analysis. For comparison, ChIP analysis in MCF7 using endogenous TFAP2C showed a 35-fold increased enrichment. The induction of ERα expression with binding of TFAP2C was associated with a 25-fold increase in loading of RNA Pol II, as determined by quantitative ChIP analysis. We conclude that the chromatin in the region of the ERα locus in ERα-negative cells is inaccessible to TFAP2C binding but that treatment with AZA/TSA increased chromatin accessibility allowing TFAP2C to bind to the AP-2 regulatory region resulting in TFAP2C-mediated loading of Pol II and induction of the ERα promoter. These findings establish a functional relationship between epigenetic chromatin marks and the activity of trans-active factors at the ERα gene promoter.

Recent work has shown that epigenetic chromatin alterations play a critical role in the regulation of gene expression particularly in processes related to development, differentiation, and oncogenesis (24–28). Methylation of CpG islands near the transcriptional start sites of genes is associated with gene silencing (24). Various histone modifications, including acetylation and methylation, are associated with differential transcriptional states and have lead to the concept of a “histone code,” which establishes heritable alterations of chromatin structure with functional effects on gene expression (25, 29). Particular attention has been given to methylation and acetylation of histone H3. Acetylation and methylation of lysine 4 (H3K4ac and H3K4me) and acetylation of lysine 9 (H3K9ac) tends to be associated with activation, whereas methylation of lysines 9 and 27 (H3K9me and H3K27me) tend to be repressive (25, 26, 29). In addition, there can be associations between CpG island methylation and specific histone marks. For example, functional differences have been described between dimethylated and trimethylated lysine 4 of H3 (H3K4me2+ versus H3K4me3+), and patterns of H3K4 methylation have been found in ES cells to be restricted to genes with CpG islands (27).

Mechanisms through which chromatin modifications alter patterns of gene expression are currently an intensive area of investigation, and one important question is how gene specificity is established. The ability of sequence specific DNA binding factors to interact with enzymes capable of altering chromatin structure is clearly one mechanism that allows for gene-specific epigenetic chromatin modifications (30). A model system that we have extensively studied is the role of TFAP2C in the regulation of genes involved in hormone response including the ERα gene. Herein, we have showed that the chromatin of the ERα gene promoter is inaccessible to TFAP2C in ERα-negative breast cell lines due at least in part to alterations in epigenetic structure, which includes both CpG island methylation and histone H3 deacetylation. Furthermore, treatment of ERα-negative cells with AZA and TSA increased chromatin accessibility and allowed TFAP2C to bind to the regulatory region of the ERα promoter resulting in induction of ERα mRNA expression. Thus, a strategy that induces chromatin accessibility and overexpression of TFAP2C may be a more robust approach for altering the phenotype of ERα-negative breast carcinomas.

One potential limitation of the finding is the apparent low level of ERα induction compared with MCF7 cells. However, it is not known if the level of induction (1-5% of MCF7 ERα mRNA levels) represents relatively low level of expression in all cells or if this represents high ERα expression in a small fraction of the cells. The experiments were conducted over a several-day period, and it is possible that the epigenetic alterations required to allow ERα activation may not have had time to occur in the majority of cells. On the other hand, the levels of expression of ERα were sufficient to allow for induction of hormone-responsive genes showing functional effects of ERα expression. If the effects are limited to a subset of cells, it may be possible to delineate mechanisms that render some cells resistant to induction and thereby expand the effects to all cells in a tumor. Alternatively, the relatively low level of ERα expression may indicate that additional trans-active factors are necessary for full activation.

AP2 family members typically bind to G + C-rich cis-elements, and their binding in vitro is inhibited by cytosine methylation within their cognate elements (31–33). Our current findings extend those results by showing that TFAP2C DNA binding in vivo is inhibited in association with DNA methylation and histone deacetylation of its binding site in the ERα promoter and that TFAP2C DNA binding in vivo can be activated by pretreating breast carcinoma cells with the epigenetic modifying drugs AZA and TSA. These findings indicate that TFAP2C, and likely other AP-2 family members, requires an unmethylated, open, and accessible chromatin structure for optimal DNA binding and transactivation in vivo. Our results also indicate that the functional consequence of TFAP2C interaction with a promoter, such as ERα is to increase Pol II loading and subsequent activation of the basal transcriptional complex.

One particularly interesting finding resulted from an examination of the hormone-independent MCF7-5C cell line, which was derived from the parental hormone-responsive MCF7 cell line after a long-term culture under conditions of estrogen deprivation (20). The MCF7-5C cells have lost ERα expression, and we have identified extensive epigenetic chromatin modifications in the AP-2 regulatory region of the ERα promoter, including a nearly complete methylation in the CpG island and significant increases in the presence of deacetylated histone H3. One intriguing finding is that MCF7-5C cells have also lost expression of TFAP2C and show overexpression of TFAP2A compared with MCF7 parental cells. Our previous work identified a role for TFPA2C in regulating expression of a set of genes involved in control of hormone response (18). If epigenetic alterations of certain TFAP2C target genes (such as ERα) were to occur first, then TFAP2C would be functionally inhibited and one might expect there to be no further selective pressure to alter TFAP2C expression. However, the loss of TFAP2C expression implies that inhibition of TFAP2C may have preceded the epigenetic chromatin changes and may be functionally linked to the development of the epigenetic marks in the ERα promoter. We are examining additional TFAP2C target genes to determine if the findings presented for the ERα gene can be found in other AP-2 regulatory regions. The data presented suggest that TFAP2C may be involved in maintaining an epigenetic structure associated with active transcription of AP-2 regulated genes.

The alteration in the pattern of TFAP2C expression has implications for breast cancer progression. TFAP2C is known to control a set of genes associated with the hormone responsive breast cancer phenotype. Finding that the expression of TFAP2C is lost with progression from a hormone responsive to a hormone unresponsive phenotype in a breast cancer cell line model raises the possibility that regulation of TFAP2C expression may be a mechanism related to breast cancer progression and development of hormone independence. Recognizing the limitation of cell line models, an examination of primary cancer specimens will need to be done to determine if loss of TFAP2C is a common mechanism involved in progression to hormone independence. A recent study examined TFAP2C expression in a panel of breast cancer cell lines using immunohistochemistry (34). TFAP2C was found to be an independent predictor of poor survival and a decreased duration of response to antihormonal therapy. Interestingly, in ERα-positive patients, the association between TFAP2C expression and time to progression was lost. These findings support a role for TFAP2C in the regulation of hormone response in primary breast cancers. However, in light of our findings, attempting to make a direct correlation between TFAP2C expression levels and function of TFAP2C does not take into account that epigenetic modification can alter the pattern of functional activity of AP-2 factors. Additional studies will need to be done to understand the role of TFAP2C in ERα-negative breast carcinomas.

No potential conflicts of interest were disclosed.