ZEB1 induces ER-α promoter hypermethylation and confers antiestrogen resistance in breast cancer

Antiestrogen resistance is a major obstacle to endocrine therapy for breast cancers. Although reduced estrogen receptor-α (ER-α) expression is a known contributing factor to antiestrogen resistance, the mechanisms of ER-α downregulation in antiestrogen resistance are not fully understood. Here, we report that ectopic zinc-finger E-box binding homeobox 1 (ZEB1) is associated with ER-α deficiency in breast cancer cells and thus confers antiestrogen resistance. Mechanistically, ZEB1 represses ER-α transcription by forming a ZEB1/DNA methyltransferase (DNMT)3B/histone deacetylase (HDAC)1 complex on the ER-α promoter, leading to DNA hypermethylation and the silencing of ER-α. Thus, ectopic ZEB1 downregulates ER-α expression and subsequently attenuates cell growth inhibition by antiestrogens, such as tamoxifen and fulvestrant. Notably, the depletion of ZEB1 by RNA interference causes ER-α promoter demethylation, restores ER-α expression, and increases the responsiveness of breast cancer cells to antiestrogen treatment. By studying specimens from a large cohort of subjects with breast cancer, we found a strong inverse correlation between ZEB1 and ER-α protein expression. Moreover, breast tumors that highly express ZEB1 exhibit ER-α promoter hypermethylation. Using a nude mouse xenograft model, we further confirmed that the downregulation of ZEB1 expression restores the responsiveness of breast cancer cells to antiestrogen therapy in vivo. Therefore, our findings suggest that ZEB1 is a crucial determinant of resistance to antiestrogen therapies in breast cancer.

Estrogen receptor-α (ER-α) is a key transcriptional regulator that plays critical roles in normal breast development and breast tumorigenesis. [1][2][3][4] Approximately 70% of breast cancers are ER-α positive and are treated with targeted antiestrogen therapy by using effective ER blocking agents, [5][6][7] such as tamoxifen and fulvestrant. Thus, ER-α expression is an important prognostic marker that is predictive for tumor response to antiestrogen treatment. However, intrinsic or acquired resistance to antiestrogen therapy presents a major challenge. [8][9][10][11] Antiestrogen resistance is believed to be caused primarily by alterations in the expression and function of ERα. 12,13 Therefore, an improved understanding of the molecular mechanisms that control ER-α activity may reveal new molecular targets that could be exploited to more effectively treat and eradicate breast cancers.
While deletions, insertions, rearrangements, or polymorphisms of the ER-α gene are uncommon and are not generally associated with loss of ER-α, 14,15 there is increasing evidence that ER-α deficiency is a result of CpG island hypermethylation within the ER-α promoter. [16][17][18][19] An abnormal methylation pattern can account for the transcriptional inactivation of the ER-α gene and consequent antiestrogen resistance in breast cancer cell lines and tumors. Other epigenetic events, such as histone deacetylation, are also involved in the complex mechanisms that regulate the transcription of the ER-α promoter. 20,21 Notably, DNA methyltransferase (DNMT) and histone deacetylase (HDAC) inhibitors, which are candidates for new cancer therapeutics, can synergistically reactivate ER-α expression in ER-α-negative breast cancer cells and restore responsiveness to antiestrogen therapy. 22 Zinc-finger E-box binding homeobox 1 (ZEB1) is a crucial member of the zinc-finger-homeodomain transcription factor family involved in the regulation of cell fate determination 23,24 and in the differentiation of several lineages, such as in myogenesis 25 and osteogenesis. 26 Beyond its physiological roles, ZEB1 is overexpressed in breast cancers, 27,28 regulates cell adherence and polarity, 29,30 modulates chemoresistance and radioresistance, 24,31 and promotes the generation of breast cancer stem cells. 32,33 Notably, a growing body of evidence has implied a potential role of ZEB1 in epigenetic regulation during tumorigenesis. [34][35][36] For example, ZEB1 interacts with HDAC1 and HDAC2 at the E-cadherin promoter in either an miR-200-dependent 35 or an miR-200-independent manner, 34,36 resulting in the transcriptional silencing of E-cadherin and cancer progression. However, the cellular and molecular mechanisms by which ZEB1 regulates dynamic epigenetic modification remain incompletely understood.
In this study, we demonstrate a mechanism for ZEB1/ER-αmediated antiestrogen resistance in breast cancer by which ER-α promoter methylation and histone deacetylation are established. ZEB1 represses ER-α expression by forming a ZEB1/DNMT3B/HDAC1 complex on its promoter. Notably, the downregulation of ZEB1 restores ER-α activity and thus increases the sensitivity of breast cancer cells to antiestrogen treatment in vitro and in vivo. We therefore suggest that ZEB1 or pathways downstream of ZEB1 may be viable therapeutic targets. The inhibition of ZEB1 to restore ER-α expression, in combination with methylation inhibitors and/or HDAC inhibitors, will represent a new strategy for overcoming antiestrogen resistance in breast cancer.

Results
Ectopic expression of ZEB1 results in the promoter hypermethylation and silencing of ER-α in breast cancer cells. To determine whether ZEB1-regulated ER-α expression in breast cancer is correlated with DNA methylation, we performed a search using the CpG island prediction database MethPrime and identified an upstream CpG-rich region at position − 4138/ − 3872 of the ER-α promoter ( Figure 1a). Two canonical E 2 -box elements (CACCTG) to which ZEB1 can potentially bind were found within this region. 29 Subsequently, we sought to assess ZEB1-regulated ER-α expression at the messenger RNA and protein levels in ZEB1/ MCF-7 (Figures 2a and b) and shZEB1/231 cells (Figures 2c and d) by qPCR and immunoblotting, and we demonstrated the downregulation of ER-α expression by ZEB1. Mechanistically, gene regulation via promoter methylation is accompanied, in some cases, by an increase in the activities of DNMT and HDAC. [20][21][22] Thus, ZEB1/MCF-7 cells were treated with the demethylating agent 5-aza-2′-deoxycytidine (AZA), the HDAC inhibitor valproic acid (VPA), or a combination of the two. The results indicated that treatment with either AZA or VPA abolished the ZEB1-mediated downregulation of ER-α at the messenger RNA and protein levels (Figures 2e and f). The combination of AZA and VPA rescued ER-α to a greater degree than either agent alone.
To confirm this finding, breast cancer cell lines were evaluated for ZEB1 and ER-α expression. The results revealed an inverse correlation between ZEB1 and ER-α at both messenger RNA and protein levels within the cell lines tested (Figures 2g and h). Collectively, the above observations indicate a potential role for ZEB1 in the regulation of ER-α promoter hypermethylation, thus altering ER-α expression in breast cancer cells.
ZEB1 represses ER-α transcription by recruiting DNMT3B and HDAC1 to the ER-α promoter. Next, we performed promoter-reporter assays to elucidate the molecular mechanism by which ZEB1 regulates ER-α transcription. As shown in Figure 3a, the wild-type -4184/-1951 promoter, ER-wtE 2 , has two canonical E 2 -box elements (CACCTG) at positions − 4099/ − 4094 and − 3935/ − 3930. The results of the luciferase assay indicated that ZEB1 overexpression decreased the promoter activity of the ER-wtE 2 reporter by~46% relative to the control without ZEB1 transfection in MCF-7 cells (Figure 3a). The E 2 -box elements were then manipulated by site-directed mutagenesis, individually or in combination. The luciferase assays showed that the mutation of either E 2 -box did not affect the ZEB1-mediated repression of ER-α promoter activity. However, simultaneous mutations within both E 2 -box elements completely eliminated the transcriptional repression of the ER-α promoter by ZEB1. In addition, the chromatin immunoprecipitation (ChIP) assays revealed that ZEB1 could bind to the ER-α promoter during basal conditions in an E 2 -boxdependent manner (Supplementary Figure S2). ZEB1 overexpression increased its binding to the endogenous ER-α promoter ( Figure 3b). Moreover, significant binding was observed in the area of E 2 -box I. These observations indicated that ZEB1 binds directly or as part of a complex to the endogenous ER-α promoter, resulting in the transcriptional repression of ER-α.
To further elucidate the mechanism by which ZEB1 represses ER-α promoter activity, we investigated the recruitment of DNMT3B and HDAC1. 38 The results showed that DNMT3B and HDAC1 each co-immunoprecipitated with ZEB1 in ZEB1/MCF-7 cells (Figure 3c). The ChIP experiments further demonstrated that DNMT3B and HDAC1 were recruited to both E 2 -box elements of the ER-α promoter, whereas maximum ChIP binding was seen in the area of E 2 -box I (Supplementary Figure S2; Figures 3d and e). Next, DNMT3B and HDAC1 expression levels were decreased by corresponding specific shRNAs (Figure 3f), and the ER-wtE 2 reporter was transfected. Importantly, luciferase assays showed that the knockdown of DNMT3B or HDAC1 significantly attenuated ZEB1 repression of the ER-α promoter ( Figure 3g). These observations point to an important role for the E 2 -box elements, especially E 2 -box I, in the regulation of ER-α by ZEB1 via interaction with DNMT3B and HDAC1.
Furthermore, we used ER-wtE 2 and ZEB1 zinc-finger deletion mutants to determine the functional domains of ZEB1 involved in binding to the ER-α promoter. As observed in Figure 3h, the deletion of either zinc-finger domain did not affect the ZEB1-mediated repression of ER-α promoter activity, whereas simultaneous deletions within both zincfinger domains completely abolished the transcriptional repression of the ER-α promoter by ZEB1. This finding indicated that ZEB1 binding to the ER-α promoter is abrogated when the functional zinc-finger domains are excised.
ZEB1 is correlated with ER-α expression and promoter hypermethylation in breast cancer patients. To better understand the correlation between ZEB1 and ER-α, we divided 248 cases of human breast carcinoma into two groups based on ZEB1 expression scores ( Figure 4a). ER-α expression in each group was represented by its expression score or the percentage of positive cases. Our results demonstrated that ER-α expression was lower in tumors with high ZEB1 expression compared to tumors with lower ZEB1 expression ( Figure 4b). Similarly, the positive percentage analysis for ER-α demonstrated a negative correlation with ZEB1 expression (Figure 4c). Importantly, ZEB1 expression was positively correlated with DNA hypermethylation of the ER-α promoter in 19 randomly selected samples (Figure 4d). We also examined the correlation between ZEB1 and PR, which is a downstream target of activated ER-α, 2,3 and obtained similar results (Figures 4e and f).
To further assess the clinical relevance of ZEB1 and ER-α downregulation, we investigated the relationship between ZEB1 and ER-α in different grades of tumors. The results showed that ZEB1 expression was relatively lower ( Figure 4g) and that ER-α expression was higher ( Figure 4h) in low-grade tumors compared to high-grade tumors, which is consistent with a previous report showing that low ER-α expression is often associated with high-grade breast cancer tumors and clinical resistance to hormone therapy. 39 Ectopic expression of ZEB1 confers antiestrogen resistance in breast cancer cells. To further determine whether ZEB1-induced loss of ER-α caused antiestrogen resistance in breast cancer cells, ZEB1/MCF-7 or shZEB1-/231 cells were treated with tamoxifen, and cell viability was measured. ZEB1 overexpression rendered MCF-7 cells less sensitive to tamoxifen treatment ( Figure 5a). An  Figure S3), revealing that ectopic ZEB1 confers antiestrogen resistance in breast cancer cells.
In addition, we investigated whether ER-α downregulation is important for ZEB1-induced resistance to antiestrogen treat-ment. An ER-α expression plasmid was transfected into ZEB1/ MCF-7 cells (Figure 5e) prior to treatment with tamoxifen. The EdU cell proliferation assay showed that the rescue of ER-α expression in ZEB1/MCF-7 restored sensitivity to tamoxifen ( Figure 5f) and fulvestrant (Supplementary Figure S4), which demonstrated that the downregulation of ER-α is involved in ZEB1-mediated antiestrogen resistance.
Downregulation of ZEB1 increases the antiestrogen sensitivity of breast cancer in vivo. Next, we assessed whether ZEB1 downregulation in breast cancer cells would influence tumor response to antiestrogen treatment in vivo. shZEB1/231 or shCtrl/231 cells were injected into the mammary fat pads of female BALB/c nude mice to establish tumors as a xenograft model, and mice were subsequently treated with tamoxifen ( Figure 6a). Immunohistochemical   (Figure 6b). Notably, upon treatment with tamoxifen, tumor volumes and weights were significantly decreased in mice injected with shZEB1/231 cells compared with those of mice injected with shCtrl/231 cells (Figures 6c and d). Ki-67 expression was also reduced in tumors from shZEB1/231 mice compared to that of shCtrl/231 ZEB1 induces ER-α promoter hypermethylation J Zhang et al mice ( Figure 6e). Conversely, ZEB1 overexpression in MCF-7 cells resulted in decreased cell sensitivity to tamoxifen in tumor xenografts (Supplementary Figure S5). These data collectively suggest a major role for ZEB1 in affecting the responsiveness of breast cancer cells to antiestrogen treatment in vivo.

Discussion
Resistance to antiestrogen therapy is one of the major barriers to the successful treatment of breast cancer, and ER-α expression is currently the main biomarker of response to antiestrogen treatment. 5,6,9,40,41 It has been well established that a combination of genetic, epigenetic, and transcriptional controls regulates ER-α expression. However, the ontogeny of tumor progression that leads to the formation of the ER-αnegative and/or antiestrogen-resistant state is not clearly understood. Our work reveals a key role for ZEB1 in antiestrogen resistance in breast cancer. First, we found that ZEB1 interacts with DNMT3B and HDAC1 at the ER-α promoter, leading to DNA hypermethylation and the downregulation of ER-α in breast cancer cells. Second, ZEB1 expression is higher than and is inversely correlated with the amount of ER-α protein in breast cancer patients. Third, the downregulation of ZEB1 considerably increases the responsiveness of breast cancer cells to antiestrogen therapy in vitro and in vivo, and this effect was ER-α dependent. Therefore, our study indicates that ZEB1 may act as a determinant of antiestrogen resistance in breast cancer. Although adjuvant antiestrogen therapy is recommended for all women with ER-α-positive breast cancer, 450% of ER-αpositive tumors that initially respond to antiestrogen treatment will eventually develop resistance. [8][9][10]41 Possible causes for the intrinsic and acquired resistance include the pharmacological properties of antiestrogen treatments, alterations in the expression and function of ER-α, interactions of tumors with the local microenvironment, and genetic alterations of tumor cells. [42][43][44] In the present study, we found that the transcriptional regulation of ER-α can be modulated by ZEB1, a protein that is upregulated in high-grade breast cancer phenotypes. 27,28 Thus, the abundance of ER-α protein is relatively low in tumors with high ZEB1 expression. These cases, despite their ER-α-positive status, are less likely to benefit from antiestrogen treatment. The importance of this finding is that the loss of ER-α in breast cancer patients is indicative of a poor prognosis, 10,13 and this reduced expression or absence of ER-α caused by ZEB1 may have implications for ER-α-negative and/or antiestrogen-resistant breast cancers. Our findings help to explain why some ER-αpositive tumors respond poorly to antiestrogen treatment. Therefore, restoring ER-α expression by inhibiting ZEB1 provides a potential therapeutic strategy for restoring antiestrogen sensitivity in breast cancer.
A number of causes have been identified to account for ER-α inactivation, such as homozygous deletion, loss of heterozygosity, or ER gene mutation. [14][15][16][17][18][19] Here, we further demonstrated that ectopic ZEB1 represses ER-α expression in breast cancer cells and thus recapitulates a loss of response to antiestrogen treatment in vitro and in vivo. Mechanistically, ZEB1 interacts with DNMT3B and HDAC1 at the ER-α promoter, causing promoter hypermethylation, histone deacetylation, and transcriptionally silenced chromatin. [20][21][22] We were able to reverse this state using the methylation inhibitor AZA and the HDAC inhibitor VPA. The binding of ZEB1 to the ER-α promoter in breast cancer cells provided further evidence that the downregulation of ER-α by ZEB1 is a possible functional mechanism for eliminating ER-α in ZEB1-positive breast cancer cells. Therefore, these results indicate that ZEB1 can regulate ER-α expression by inducing promoter methylation and chromatin remodeling to achieve transcriptional repression. [35][36][37]45 Functional hallmarks of ER-α-negative cells include their ability to proliferate without estrogen stimulation and their antiestrogen resistance. 10,18 This state is consistent with our results in the present study showing that ZEB1/MCF-7 cells were resistant to tamoxifen and fulvestrant without significant G 1 /S arrest compared to Ctrl/MCF-7 cells. Moreover, ZEB1 tumors were able to grow without estrogen (data not shown) and were unaffected by tamoxifen treatment in the xenografted nude mice. These experiments strongly indicated that ZEB1 expression caused estrogen independence in breast cancer cells that possibly led to antiestrogen resistance. Notably, ZEB1 expression is increased in high-grade human breast tumors, and a high percentage of these tumors are ER-α negative. 27 We observed similar results in breast cancer patients; a negative correlation exists between ZEB1 and ERα expression. Therefore, ZEB1 overexpression may provide a mechanistic link between the development of aggressive breast cancer and the loss of ER-α expression and may provide a method to elucidate the ontogeny of ER-α-negative and/or antiestrogen-resistant breast cancer.
In this study, we demonstrated an alternative mechanism for ZEB1/ER-α-mediated antiestrogen resistance in breast cancer that supplements the promoter methylation and deacetylation of ER-α. Considering that estrogen induces ZEB1 expression while regulating breast cancer development and progression, 45 we therefore suggest a potential role for ZEB1 at an intersection between intrinsic and acquired resistance to antiestrogen therapy. The inhibition of ZEB1 to restore ER-α expression, in combination with methylation inhibitors and/or HDAC inhibitors, will represent a new strategy for overcoming antiestrogen resistance in breast cancer.

Materials and Methods
Cell culture and transfection. Human breast cancer cell lines were maintained in DMEM (MDA-MB-231, MCF-7, and SUM-159) and RPMI 1640 (ZR-75-1) supplemented with 10% fetal bovine serum, 100 IU penicillin, and 100 mg/ml streptomycin. For antiestrogen treatment, cells were cultured in phenol red-free minimal essential media with 5% fetal bovine serum and then treated with different concentrations of tamoxifen or fulvestrant. Cells were transfected using Lipofectamine 2000 (Invitrogen, CA, USA) according to the manufacturer's protocol.   Immunoblotting assay. The preparation of total cell extracts and immunoblotting with appropriate antibodies were performed as previously described. 37 The appropriate antibodies were used as indicated in Supplementary Data. Labeled proteins were visualized with an ECL chemiluminescence kit (Millipore).   Immunohistochemical analysis. Immunohistochemical analysis of paraffin-embedded sections was performed using the Envision Kit (Dako, Denmark) according to the manufacturer's protocol. Sections were boiled in retrieval solutions to expose antigens. The specific antibodies (see Supplementary Data) were applied to the sections. Slides were counterstained with hematoxylin, dehydrated, and mounted. Immunostaining was independently evaluated by 2 pathologists.
Cell proliferation assay. Cells were seeded onto a 96-well plate at a density of 4 × 10 3 cells/well and treated with different concentrations of tamoxifen or fulvestrant for 24-96 h. Cell viability was assessed using the CCK-8 assay according to the manufacturer's protocols (Dojindo, Tokyo, Japan). Six parallel replicates were measured for each sample.
5-ethynyl-2′-deoxyuridine cell proliferation assay. Cells growing in 24-well plates were cultured in the presence of tamoxifen or fulvestrant at the indicated time points and then assayed with the Cell-Light EdU Apollo488 In Vitro Imaging Kit according to the manufacturer's instructions (RiboBio, Guangzhou, China). Images were taken and analyzed using a Confocal FV1000 microscope (Olympus, Tokyo, Japan). EdU-positive cells were calculated as (EdU add-in cells/ Hoechst stained cells) × 100%. At least 200 cells were counted per well.
Tumor xenograft experiments. All experimental procedures involving animals were performed according to the institutional ethical guidelines for animal experiments and approved by the Ethics Committee for Animal Use at the Medical College of Nankai University. In brief, cells were collected and suspended in 200 μl phosphate-buffered saline at a concentration of 5 × 10 6 cells/ml and then injected into the mammary fat pads of female BALB/c nude mice. Tumor development was allowed to reach a volume of~100 mm 3 . The mice were then randomized into 2 groups (5 mice per group), and tamoxifen pellets (5 mg/pellet, Innovative Research of America) or placebo pellets were subcutaneously embedded for another 3 weeks. Tumor volumes (V) were calculated by measuring the length (L) and width (W) of the tumor with calipers and applying the following formula: V = (L × W 2 ) × 0.5. Tumor tissues were also processed and sectioned for histological evaluation.
Statistical analysis. Statistical analyses were performed using SPSS 13.0 software (SPSS, Chicago, IL, USA). Data are presented as the means ± S.D. and represent three independent experiments. Spearman's rank correlation test was used to analyze the correlation of gene expression in tissue samples. One-way analysis of variance was used to compare means between treatment groups. Where appropriate, Student's t-test for unpaired observations was applied. A P-valueo0.05 was considered significant.

Conflict of Interest
The authors declare no conflict of interest.