Constitutive expression of AhR and BRCA-1 promoter CpG hypermethylation as biomarkers of ERα-negative breast tumorigenesis

Background Only 5–10 % of breast cancer cases is linked to germline mutations in the BRCA-1 gene and occurs early in life. Conversely, sporadic breast tumors, which represent 90-95 % of breast malignancies, have lower BRCA-1 expression, but not mutated BRCA-1 gene, and tend to occur later in life in combination with other genetic alterations and/or environmental exposures. The latter may include environmental and dietary factors that activate the aromatic hydrocarbon receptor (AhR). Therefore, understanding if changes in expression and/or activation of the AhR are associated with somatic inactivation of the BRCA-1 gene may provide clues for breast cancer therapy. Methods We evaluated Brca-1 CpG promoter methylation and expression in mammary tumors induced in Sprague–Dawley rats with the AhR agonist and mammary carcinogen 7,12-dimethyl-benzo(a)anthracene (DMBA). Also, we tested in human estrogen receptor (ER)α-negative sporadic UACC-3199 and ERα-positive MCF-7 breast cancer cells carrying respectively, hyper- and hypomethylated BRCA-1 gene, if the treatment with the AhR antagonist α-naphthoflavone (αNF) modulated BRCA-1 and ERα expression. Finally, we examined the association between expression of AhR and BRCA-1 promoter CpG methylation in human triple-negative (TNBC), luminal-A (LUM-A), LUM-B, and epidermal growth factor receptor-2 (HER-2)-positive breast tumor samples. Results Mammary tumors induced with DMBA had reduced BRCA-1 and ERα expression; higher Brca-1 promoter CpG methylation; increased expression of Ahr and its downstream target Cyp1b1; and higher proliferation markers Ccnd1 (cyclin D1) and Cdk4. In human UACC-3199 cells, low BRCA-1 was paralleled by constitutive high AhR expression; the treatment with αNF rescued BRCA-1 and ERα, while enhancing preferential expression of CYP1A1 compared to CYP1B1. Conversely, in MCF-7 cells, αNF antagonized estradiol-dependent activation of BRCA-1 without effects on expression of ERα. TNBC exhibited increased basal AhR and BRCA-1 promoter CpG methylation compared to LUM-A, LUM-B, and HER-2-positive breast tumors. Conclusions Constitutive AhR expression coupled to BRCA-1 promoter CpG hypermethylation may be predictive markers of ERα-negative breast tumor development. Regimens based on selected AhR modulators (SAhRMs) may be useful for therapy against ERα-negative tumors, and possibly, TNBC with increased AhR and hypermethylated BRCA-1 gene.

Agonists of the aromatic hydrocarbon receptor (AhR) are ubiquitous in the environment and include dietary compounds, metabolites of fatty acids, industrial xenobiotics, and skin photoproducts generated through exposure to ultraviolet radiation [33]. Importantly, the expression of the AhR and downstream gene targets such as CYP1B1 are increased in human and rodent mammary tumors [34,35]. Consequently, the use of selective modulators of the AhR (SAhRMs) has been proposed in breast cancer therapy [36].
Previously, we reported that AhR agonists repressed estradiol (E2)-dependent BRCA-1 transcription in human breast cancer cells [37][38][39][40][41]. This repressive effect was linked to increased recruitment to the BRCA-1 promoter of the activated AhR and other factors associated with the epigenetic machinery [42] including DNA methyl-transferase-1, (DNMT-1), DNMT-3a and -3b; methyl-binding domain protein-2 (MBD2); and placement of histone-3 trimethylation marks on lysine-9 (H3K9me3) [43]. In AhR-activated human breast cancer cells, the pattern of BRCA-1 promoter CpG methylation [44] coincided with the one detected in human sporadic breast tumors [45,46]. Recently, using a rodent model we found that gestational activation of the AhR increased CpG methylation of the Brca-1 gene while reducing BRCA-1 expression in mammary tissue of female offspring. The latter changes were overridden by gestational pretreatment with an AhR antagonist [47]. These cumulative data draw attention to the fact alterations of AhR expression and activity may play a role in the etiology of breast tumorigenesis. Nevertheless, the connection between higher AhR expression and/or activation and BRCA-1 promoter hypermethylation in breast tumors has not been investigated.
This study reports that rat mammary tumors induced with the AhR-agonist 7,12-dimethyl-benzo(a)anthracene (DMBA) [48] had augmented CpG methylation of the Brca-1 gene; higher expression of Ahr, Cyp1b, and proliferation markers (Cdk4, Ccnd1); and diminished expression of BRCA-1 and ERα. In cell culture experiments, the treatment with α-naphthoflavone (αNF), a prototype SAhRM, exerted cell line-specific effects: in ERαnegative human UACC-3199 sporadic breast cancer cell line, it rescued BRCA-1 and ERα expression, while inducing CYP1A1; in ERα-positive MCF-7 breast cancer cells, αNF antagonized E2-dependent stimulation of BRCA-1 without affecting ERα expression. Finally, we document that human TNBC had higher AhR expression and BRCA-1 promoter CpG methylation compared to human luminal-A (LUM-A), LUM-B, and HER-2-positive breast tumors. We conclude that constitutive high expression of AhR associated with BRCA-1 gene hypermethylation may be prognostic markers of ERα-negative breast tumor development. Therapies based on SAhRMs may hold promise for rescue of BRCA-1 and ERα expression in ERα-negative breast cancers.

Animal experiments
Weaned female Sprague-Dawley rats and AIN-76A diet were purchased from Harlan Laboratories (Houston, Texas). At day 50 of age, 8 animals/group (n = 8) were assigned to either a sesame oil vehicle control group, or a treatment group receiving 10 mg/animal of DMBA (Sigma-Aldrich, St. Louis, MO) by oral gavage [48]. Animals were palpated weekly, and mammary tumors were collected when they reached a diameter of 1 cm. Animals were sacrificed according to a protocol approved by the IACUC Committee of the University of Arizona. Mammary gland tissues and tumors were collected and stored frozen until further analysis.

Cell culture experiments
Human MCF-7 and UACC-3199 breast cancer cells were obtained from the American Type Culture Collection (Manassas, VA). MCF-7 and UACC-3199 cells were maintained, respectively, in Dulbecco's Modified Eagles Media (DMEM) or RPMI 1640 media (Mediatech, Manassas, VA) supplemented with 10 % fetal calf serum (FCS) (Hyclone Laboratories, Logan, UT). αNF and E2 for cell culture experiments were obtained from Sigma-Aldrich (St. Louis, MO). For experiments with αNF and E2, cells were plated in 6-well plates at a density of 5 × 10 5 /cells/well. Then, after 24 h cells were cultured for an additional 72 h in phenol-red free DMEM (MCF-7) or RPMI (UACC-3199) supplemented with 10 % charcoal-stripped FCS plus 2 μM αNF in the presence or absence of 10 nM E2 [42]. For Western blotting, at the end of the incubation period, cells were washed with ice-cold phosphate buffer saline (PBS) and scraped with cold lysis buffer containing protease inhibitor. For mRNA studies, at the end of the incubation period, cells were washed with ice-cold PBS. Extraction of RNA was carried out using Triazol Reagent (ThermoFisher Scientific, Grand Island, NY). Cell extracts and RNA samples were stored frozen at -20°C until further use.

Breast tumor collection
Human normal and breast tumor tissue sections were obtained de-identified from the University of Arizona Cancer Center Tissue Acquisition and Cellular/Molecular Analysis Shared Resource with the approval from the Institutional Review Board of the University of Arizona, Approval Form#F309. No patient-level correlations between gene activation information and individual patientdata were performed, according to U.S. Department of Human Health Services and Federal Drug Administration regulations, and in compliance with the World Medical Association Declaration of Helsinki (http://www.wma.net/ en/30publications/10policies/b3/index.html). The presence of tumor in each sample was confirmed by a staff pathologist and classified according to the following criteria: (i) TNBC: basal-like, and cytokeratin-, ERα-, PR-, HER-2-, and epidermal growth factor-negative; (ii) HER-2-positive: HER-2-positive and ERα-negative; (iii) LUM-A: ERα-positive and/or PR-positive, and HER-2-negative; and (iv) LUM-B: ERα-positive and/or PR-positive, and HER-2positive. As controls, we also obtained sections of nontumor tissue from the region surrounding TNBC and LUM-B tumors.

mRNA analyses
Sections of normal mammary gland and mammary tumor tissues from 8 animals/group (n = 8) were homogenized (1 mL/40 mg of tissue) of QIAzol Reagent (Invitrogen, Carlsbad, CA). Total RNA was purified using RNeasy Lipid Tissue Mini Kit as per manufacturer's instructions (Qiagen, Valencia, CA) [47]. Concentrations and quality of RNA were verified using the Nanodrop1000 Spectrophotometer (Thermo Scientific, Wilmington, DE). Equal amounts of total RNA (500 ng) were transcribed into cDNA using ISCRIPT supermix kit (Bio-Rad Laboratories, Hercules, CA). Next, cDNA aliquots were analyzed by qPCR using the SYBR Green PCR Reagents kit (Life Technologies, Grand Island, NY). Briefly, reactions were run at a final volume of 25 μL consisting of the following master mix: 12.5 μL of SYBR Green mix, 1 μL each of forward and reverse primers, 9.5 μL nuclease-free water, and 1 μL cDNA. Amplification

Statistical methods
Densitometry after Western blotting and CpG methylation analyses were performed using Kodak ID Image Analysis Software (Eastman Kodak Company, Rochester, NY). Statistical analyses were performed using Prism 5.0 (GraphPad Software Inc., La Jolla, CA) [47]. Data were analyzed by 1-way ANOVA. Post-hoc multiple comparisons among all means were conducted using Tukey's Test after main effects and interactions were found to be significant at P ≤ 0.05. Data were presented as means ± SEM and statistical differences highlighted with different letters or asterisks.

BRCA-1 expression in mammary tumors
Previous studies documented a high degree (~75 %) of correlation between loss of BRCA-1 and reduced ERα expression in human breast tumors [27,28]. Results of BRCA-1 and ERα protein expression in control mammary tissue, and in adjacent normal mammary tissues and tumors obtained from animals treated with DMBA are presented in Fig. 1. Compared to control mammary gland, BRCA-1 expression (Fig. 1a) was reduced by an average 50 % in peritumoral mammary tissue (Fig. 1b). BRCA-1 protein levels were reduced by an additional~40 % in DMBA-induced mammary tumors. Similarly, we found that compared to control mammary tissue, ERα levels were reduced by an average 40 % and 70 % respectively, in DMBA-treated but apparently normal mammary gland, and mammary tumors.

Brca-1 promoter CpG methylation in mammary tumors
To examine if the reduction in BRCA-1 expression in DMBA-treated animals was related to changes in Brca-1 promoter CpG methylation status, we extracted genomic DNA from control mammary gland, and adjacent mammary tissues and tumors from DMBA-treated animals.
In control experiments, we ascertained that rat bisulfonated genomic DNA obtained from control mammary tissue was amplified in the linear range with U-and Mspecific Brca-1 oligonucleotides, and Brca-1 amplicons were of the expected size (142 bp) (Fig. 2a). Turning to changes in Brca-1 promoter CpG methylation (Fig. 2b), we found that compared to control, the adjacent mammary gland isolated from DMBA-treated animals had an average 1.9-fold increase in Brca-1 promoter CpG methylation (Fig. 2c), which was increased on average an additional~1.0-fold in DMBA-induced tumors (Fig. 2c). These data suggested that the Brca-1 gene was a target for repression via CpG methylation in mammary tissue of animals treated with the AhR agonist and mammary carcinogen, DMBA.

AhR expression and activation in mammary tumors
Focusing on measurements of Ahr expression and activation, we first examined changes in Ahr in mammary tissue of control animals, and peritumoral and tumor tissues obtained from DMBA treated animals (Fig. 3). Compared to control, levels of Ahr were increased~2.7fold in peritumoral tissues; Ahr expression was increased an additional~4.5-fold in DMBA-induced mammary tumors. In human breast cancer cell lines, higher CYP1B1 expression over CYP1A1 has been related to higher AhR expression and ERα-negative status [50]. Therefore, we measured changes in expression of Cyp1a1 and Cyp1b1 as controls for AhR pathway activation. Basal Cyp1a1 was reduced by 30 and 70 % respectively in adjacent mammary gland and mammary tumors (Fig. 4a). Conversely, Cyp1b1 levels were markedly increased, on average~5.0 and 14.0-fold of control, respectively in peritumoral tissue and mammary tumors (Fig. 4b).
These data indicated that constitutive overexpression of the Ahr in rat mammary tumors was coupled with differential regulation on the Cyp1a1 (repression) and Cyp1b1 (activation) target genes.

Proliferation markers in mammary tumors
Previous investigations documented that increased expression and activation of the AhR may be associated with mitogenic responses [51,52], and increased Cdk4  [47] and human [53] mammary cells. Based on this information, we compared Ccnd1 (cyclin D1) and Cdk4 expression in control mammary gland, and adjacent mammary tissues and tumors obtained from animals treated with DMBA (Fig. 5). We noticed that compared to control, in DMBA-treated animals levels of Cdk4 (Fig. 5a) and Ccnd1 (Fig. 5b) were increased respectively an average~3.0-and 1.0-fold in adjacent mammary tissues; and an additional~6.0-and 12.0-fold increase was seen, respectively, for Cdk4 and Ccnd1, in mammary tumors. Taken together, animal results suggested that constitutive high Ahr expression and pathway activation on the Cyp1b1 gene were linked to induction of mammary tumorigenesis associated with reduced expression of BRCA-1 and ERα.

Targeting of AhR with αNF in human breast cancer cells
Increased expression and activation of the AhR may contribute to epigenetic remodeling during early breast carcinogenesis [54], whereas loss of BRCA-1 associates with ERα-negativity in hereditary and sporadic breast tumors [27]. Therefore, we compared the expression of BRCA-1 and AhR in human ERα-positive MCF-7, and ERα-negative UACC-3199 sporadic, breast cancer cells. We selected these cell lines because MCF-7 cells express wild-type BRCA-1 and are ERα-positive. Conversely, UACC-3199 cells have wild-type but hypermethylated, BRCA-1 [21,55], and express low levels of ERα [56]. Results of Western blots informed that expression of BRCA-1 was~5.0-fold higher in MCF-7 compared to UACC-3199 cells (Fig. 6a). Conversely, the expression of the AhR was notably higher (~15.0-fold) in UACC-3199 compared to MCF-7 cells.
In previous studies with MCF-7 cells, we used αNF to reverse the repressive effects of AhR agonists on BRCA-1 expression [38]. We then extended these studies to UACC-3199 breast cancer cells. Results depicted in Fig. 6b revealed that the treatment with αNF increased (~1.4-fold of control) BRCA-1 mRNA; this change was associated with a~2.0-fold upregulation of BRCA-1 and ERα expression (Fig. 6c). Turning to other biological changes that occurred in UACC-3199 cells along with reactivation of BRCA-1 by αNF, we detected a large increase (~32-fold of control) in CYP1A1 expression with only modest effects (~1.5-fold increase compared to control) on CYP1B1 (Fig. 6d). Then, we compared the effects of αNF on  Fig. 7 confirmed that αNF increased~2.0-and 3.0-fold of control respectively, BRCA-1 and ERα in UACC-3199 cells, which were however refractory to the treatment with E2 alone or in combination with αNF. On the other hand, as previously reported by our group [41,42], the treatment of MCF-7 cells with αNF antagonized the E2-dependent induction of BRCA-1. Overall, these cell culture studies implied that the effects of αNF, selected as a prototype AhR antagonist, were influenced by cell-context and ERα status, i.e. αNF rescued BRCA-1 and ERα expression in sporadic and ERα-negative UACC-3199 breast cancer cells carrying hypermethylated BRCA-1. Conversely, αNF antagonized E2-dependent stimulation of BRCA-1 expression in ERαpositive MCF-7 breast cancer cells.

BRCA-1 promoter methylation and AhR expression in human breast tumors
Next, we wished to explore if differential expression of AhR and BRCA-1 promoter CpG methylation associated  with pathological classification of human breast tumor subtypes based on receptor status. Therefore, we compared the level of BRCA-1 promoter CpG methylation in genomic DNA obtained from control breast tissue and various breast tumor subtypes including TNBC, LUM-A, HER-2-positive, and LUM-B. On average, we observed that BRCA-1 promoter methylation (M/U ratio) was increased~6.6-fold in TBNC compared to nontumor breast tissue (Fig. 8). Conversely, compared to non-tumor tissue, there were no differences in the amount of BRCA-1 promoter methylation in LUM-A, LUM-B, and HER-2-positive breast tumors. Interestingly, the increased BRCA-1 promoter methylation in TNBC correlated with increased expression (~3.0-fold of control) of AhR. Overall, these results denoted that coordinated increase in AhR expression and BRCA-1 gene hypermethylation may be molecular markers of TNBC.

Discussion
Earlier studies documented that the AhR is overexpressed and constitutively activate in rodent and human mammary tumors [35]. These findings attributed to environmental and endogenous factors that activate the AhR a role in breast tumorigenesis. Our prior cell culture [37][38][39][40][41][42][43][44] and rodent [47] model investigations of breast cancer provided evidence that the BRCA-1 gene was a molecular target for the AhR and various chromatin remodeling factors. Specifically, the recruitment of the activated AhR, DNMTs, and MBD-2 to the BRCA-1 gene culminated with placement of repressive histone (H3K9me3) and DNA (CpG methylation) marks, and downregulation of BRCA-1 expression.
The first objective of this study was to investigate the association between AhR expression and/or activation and Brca-1 promoter methylation status in mammary tumors. For this purpose, we adopted the DMBA-rat mammary tumor model based on the knowledge DMBA is a strong AhR agonist [33] and mammary carcinogen [48,57]. The upregulation of Ahr and Cyp1b1 were paralleled by increased Brca-1 CpG methylation, and reduced expression of BRCA-1 and ERα in mammary tumors induced with DMBA. Also, the reduction in BRCA-1 expression observed in peritumoral tissue suggested that Brca-1 CpG methylation may be an epigenetic event that occurs prior to overt mammary tumor formation linked to Ahr overexpression and/or activation. This interpretation may have prognostic value since adjacent non-tumor mammary tissue from DMBAtreated animals had also increased expression of the proliferation markers Cdk4 and Ccnd1 (cyclin D1). Overall, results of animal experiments linked higher AhR expression and activity on the Cyp1b1 gene to increased risk of mammary tumorigenesis [34,48,54,57,58] via epigenetic silencing of Brca-1.
The reduction in ERα expression observed in adjacent mammary gland and mammary tumors of DMBAtreated animals was consistent with previous reports of reduced ERα in familial BRCA-1 tumors [25,26], and sporadic breast cancers with hypermethylated BRCA-1 [28]. The ERα and the BRCA-1 participate in a positive feed-back loop whereby the ERα upregulates BRCA-1 [38], which in turn stimulates ERα expression [27]. Therefore, AhR-dependent repression of BRCA-1 via increased CpG methylation may disrupt this positive feedback loop between BRCA-1 and ERα and favor the development of ERα-and BRCA-1-negative breast tumors.
Turning to markers of AhR activation, we measured increased Cyp1b1 in adjacent mammary gland and mammary tumors of DMBA-treated animals. This accumulation was consistent with previous studies reporting stimulation of Cyp1b1 in rat models of mammary tumorigenesis [34,48]. The CYP1B1 enzyme catalyzes the production from E2 of mutagenic 4-hydroxy-E2 (4OH-E2) [59,60]. It is feasible that the constitutive activation of the AhR/CYP1B1 axis may have the synergistic effect of increasing DNA damage via increased production of mutagenic 4OH-E2 while impairing DNA repair functions controlled by BRCA-1. Conversely, we found that Cyp1a1 was reduced in adjacent mammary gland and mammary tumors of DMBA-treated animals. Consistent with these findings, earlier studies documented preferential repression of Cyp1a1 in DMBA-induced mammary tumors [48], as well as in human invasive ductal carcinomas [61,62] and breast cancer cells lacking the ERα [50,63]. Furthermore, reduced CYP1A1 enzymatic activity has been linked to constitutive activation of the AhR [64] and resistance of breast cancer cells to apoptosis induced by DMBA [65].
To further elucidate the cross-talk between expression and/or activation of AhR, and BRCA-1 regulation, we turned to cell culture experiments using UACC-3199 sporadic breast cancer cells, which possess hypermethylated BRCA-1 promoter [21,55] and express low ERα [56]. Compared to MCF-7 cells, UACC-3199 cells had higher basal AhR, but lower BRCA-1. Therefore, we tested whether or not treatment of UACC-3199 cells with the AhR antagonist αNF rescued BRCA-1 expression. The rationale for this approach was based on our previous studies showing that BRCA-1 silencing by AhR agonists was reversed by cotreatment with α-NF [38]. The mechanisms of action of αNF as an AhR antagonist and anticarcinogen have been related respectively, to reduction of transcriptionally active nuclear AhR complexes [66,67], and inhibition of 4OH-E2 production by CYP1B1 [68]. The rescue of BRCA-1 and ERα by αNF in UACC-3199 breast cancer cells were biological changes associated with preferential induction of CYP1A1. Conversely, αNF did not affect ERα levels, but antagonized E2-dependent activation of BRCA expression, in ERα-positive MCF-7 cells. The latter findings were in accord with our previous reports documenting repression by αNF and 3-methoxy-4-naphthoflavone, another antagonist of the AhR, of E2-dependent transcriptional activation of the BRCA-1 gene [42]. These differential effects of αNF on BRCA-1 and ERα expression could be attributed to interactions between agonist/antagonist activities on the AhR and ERα status [69]. This AhR-ERα cross-talk could be exploited for the development of strategies aimed at the reactivation of BRCA-1 and ERα in ERα-negative and AhR-overexpressing tumors.
We further extended our studies of BRCA-1/AhR cross-talk to human breast tumors, and found that compared to LUM-A, LUM-B, and HER-2-positive tumors, TNBC had higher AhR and BRCA-1 CpG methylation. These observations provided additional support to the hypothesis that constitutive AhR expression may be associated with hypermethylation of the BRCA-1 promoter and the development of TNBC. It remains unknown whether the reduced ERα expression in DMBA-induced tumors, UACC-3199 cells, and TNBC tumors may be due to hypermethylation, or disruption of expression of transcription factors that regulate transcription, of the ERα (ERS1) gene. Answering these queries may assist with the development of strategies for coordinate epigenetic reactivation of BRCA-1 and ESR1 in ERα-negative breast tissues.

Conclusions
Many studies have effectively utilized the AhR-agonist and mammary carcinogen DMBA to examine the molecular pathways that contribute to breast cancer and efficacy of therapies [57]. To our knowledge, this is the first study linking constitutive overexpression of the AhR to BRCA-1 promoter hypermethylation in DMBAinduced mammary tumors and human TNBC. The potential prognostic significance of the current findings is underscored by the fact the AhR is constitutively active in ERα-negative human breast tumor cells [34,35,50,61]. Ongoing studies in our laboratory are using in vitro and vivo models to explore the effects of AhR knockout on epigenetic regulation of BRCA-1 and ESR1 (ERα), and the preventative effects of AhR antagonists. Progress in these areas may help clarifying a causative role for the AhR in breast tumorigenesis and assist with the development of risk models for BRCA-1 mutation carriers [70,71] and sporadic TNBC, for which therapy options remains an intensive area of investigation [72,73].

Competing interests
The authors declare that they have no competing interests.
Authors' contributions OIS, AJP, and DFR contributed to the conception and design of animal experiments, collection of animal tissues and analyses, and cell culture experiments. CL contributed to the collection of human breast tumors and tumor data interpretation. DFR and OIS wrote the manuscript. All authors read and approved the final version of this manuscript.