1. Introduction
Breast cancers (BC) are the most common malignancies and causes of cancer mortality in women worldwide [
1]. Triple negative breast cancers (TNBC) account for ~15–20% of BC cases and refer to a heterogenous group of BC that lacks expression of estrogen receptor (ER)α, progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) [
2]. Systemic chemotherapy is the main treatment option for TNBC, since these tumors lack molecular targets for therapy. Unfortunately, only ~20–30% of TNBC patients have a pathological complete response to neoadjuvant therapy [
3]. Therefore, key research priorities related to the management of TNBC include (1) deriving methods for effective prevention, and (2) identifying compounds for use in the adjuvant setting to enhance therapeutic response.
The
BRCA1 gene encodes a 220-kDa nuclear phosphoprotein (BRCA1) that functions as a tumor suppressor through involvement in DNA damage repair, cell cycle control, transcriptional regulation, apoptosis, and mRNA splicing [
4]. Women who inherit
BRCA1 mutations have a ~72% lifetime risk of developing BC [
5], the majority of which are TNBC [
6]. Similar to
BRCA1 mutation carriers, the hypermethylation of
BRCA1 is associated with a BRCA1-deficient phenotype (i.e., BRCAness) [
7] and increased odds of developing sporadic breast tumors that are TNBC [
8]. The hypermethylation of
BRCA1 is reported in ~20–65% of sporadic TNBC [
9,
10,
11] and contributes to the biallelic inactivation of functional alleles in tumors from
BRCA1 mutation carriers [
12,
13,
14].
Dietary factors are considered to play a key role in both the prevention and progression of BC [
15]. The consumption of genistein [GEN (i.e., 5,7-dihydroxy-3-(4-hydroxyphenyl)chromen-4-one)], a non-toxic naturally occurring isoflavone found in soy, has been suggested to lower rates of BC in Eastern Asian countries [
16,
17]. Epidemiological studies indicated that BC risk may be decreased by ~40% with higher consumption of soy and soy isoflavones [
18,
19,
20,
21,
22,
23,
24]. The results of mechanistic studies in vitro show that the anti-tumorigenic activity of GEN in BC cells is largely attributable to the preferential induction of ERβ, which suppresses ERα signaling [
25]. In ERα-positive MCF-7 cells, the overexpression of ERβ enhances the antiproliferative effects of GEN [
26]. Other anti-tumorigenic effects of GEN in BC cells include inhibition of protein tyrosine kinase [e.g., epidermal growth factor receptor (EGFR), platelet-derived growth factor receptor (PDGFR)] and nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) signaling [
25]. In addition, GEN induces cell cycle arrest in the G2/M phase of the cell cycle [
27] and inhibits DNA methyltransferase (DNMT) activity [
28,
29,
30,
31] in TNBC cells.
Dietary studies in rodents demonstrate that the dose and timing of exposure to GEN influence tumor response. For example, soy isoflavones administered during the perinatal (i.e., gestation and lactation) or postweaning (4 weeks of age onward) periods increase respectively tumor burden and the onset of mammary adenocarcinoma development in mouse mammary tumor virus (MMTV)-neu transgenic mice [
32,
33]. On the other hand, GEN administered before puberty appears to delay mammary tumor development in rats [
34] and decrease 7,12-dimethylbenz[a]anthracene (DMBA)-induced mammary tumor incidence and aggressiveness in mice [
35]. Interestingly, the anti-mammary tumor effects of GEN are not observed in
Brca1+/− mice, possibly suggesting a requirement for coincident
Brca1 expression.
The
BRCA1 promoter harbors multiple cognate binding sequences for the aryl hydrocarbon receptor (AHR), which are referred to as xenobiotic response elements (XRE) [
36]. The AHR is a highly conserved ligand-activated transcription factor of the basic helix–loop–helix–PER–ARNT–SIM (bHLH/PAS) family [
37]. It regulates a gene battery involved in the metabolism and conjugation of steroids, drugs, and other xenobiotics [
38]. Under normal conditions,
BRCA1 is induced by estradiol (E2) through a non-canonical mechanism of ERα, whereby an activator protein (AP)/ERα transcription complex containing the unliganded AHR assembles at the proximal
BRCA1 promoter [
39]. Upon activation with exogenous ligands [e.g., 2,3,7,8-tetrachlorodibenzene(p)dioxin (TCDD)], the AHR colocalizes at the
BRCA1 promoter with DNMT3a, DNMT3b, and DNMT1 [
36,
40,
41]. This leads to hypermethylation at a cytosine-guanine dinucleotide (CpG) island proximal to exon 1a of
BRCA1 and the suppression of E2-dependent
BRCA1 transactivation.
We recently reported that GEN prevents
BRCA1 hypermethylation in ERα-positive MCF7 BC cells treated with the AHR agonist TCDD, and reverses constitutive
BRCA1 CpG methylation in ERα-negative HER2-enriched cells with constitutive high levels of AHR [
42]. Due to the association between the TNBC phenotype and increased
AHR expression and
BRCA1 CpG methylation, we hypothesized that GEN counteracts the AHR-dependent repression of
BRCA1. We show that a GEN-enriched diet administered to mice from conception through lactation, weaning, and adult life decreases basal
Brca1 methylation and AHR activity in the adult mammary gland. We also document that in HCC38 TNBC cultured cells, the AHR is overexpressed and constitutively active. Conversely, the treatment of HCC38 cells with GEN and selected AHR antagonists increases BRCA1 protein levels via CpG demethylation and decreased recruitment of the AHR at the
BRCA1 promoter. The latter effects are observed in parallel with increased ERα expression, leading to the sensitization of HCC38 TNBC cells to the growth inhibitory effects of 4-hydroxytamoxifen (4-OHT).
4. Discussion
Triple negative breast cancers are clinically aggressive [
2], prone to visceral and central nervous system metastasis [
53,
54], and currently lack targeted therapeutics [
55]. The CpG hypermethylation of
BRCA1 is a common epigenetic aberration in sporadic TNBC [
9,
10,
11] and contributes to the silencing of wild-type
BRCA1 alleles in tumors from germline
BRCA1 mutation carriers [
12,
13,
14]. Our group has documented that
AHR is overexpressed and
BRCA1 is hypermethylated in primary tumors from TNBC patients compared with other BC subtypes and non-malignant tissue [
46]. Investigations by other groups have also reported
AHR overexpression is associated with the TNBC phenotype [
56]. High levels of AHR protein are also found in rodent mammary tumors and pre-malignant tissue [
46,
47,
57,
58], human BC cell lines [
42,
47], and primary tumors [
59,
60]. In normal human mammary epithelial cells, overexpression of the AHR induces malignant transformation [
61], whereas AHR knockdown in TNBC cells attenuates tumorigenicity in vitro and in orthotopic mouse models [
62]. Our group has characterized the role of AHR activation in the epigenetic silencing of
BRCA1 [
36,
40,
41]. Specifically, we found that GEN prevented
BRCA1 CpG hypermethylation in ERα-positive BC cells treated with an AHR agonist as well as reversed constitutive
BRCA1 CpG methylation in ERα-negative HER2-enriched cells with high levels of AHR [
42].
In this study, we first investigated the effect of GEN on the AHR-dependent epigenetic regulation of
BRCA1 in vivo. Dietary GEN is thought to contribute to the lower rates of BC seen in Eastern Asian populations [
16,
17]. In both rodents and humans, intestinal absorption of GEN occurs rapidly and efficiently due to its small molecular weight (~270 kDa) and lipophilic properties [
17]. Studies in rodents have shown the absorption efficiency of total genistein ranges from ~46% to 100%, depending on the animal model, source of GEN, and sex [
63,
64,
65,
66,
67]. The major pathways of GEN metabolism are glucuronidation and sulfation, and the predominant plasma metabolites are genistein-7-glucuronide-4′-sulfate (G-7G-4′S) and genistein-4′,7-diglucuronide (G-4′,7-diG) [
68,
69,
70]. In mice, after a single oral administration of GEN (20 mg/kg), ~80% was converted to glucuronides or sulfates, whereas ~20% was aglycone GEN [
69]. In the present study, we administered GEN-enriched diets (4 and 10 ppm) to breeding pairs, pregnant and lactating mothers, weanlings, and adult offspring. A previous study in female mice administered a GEN-enriched diet (6 ppm) over a similar time course (gestation through lifetime) and found that plasma GEN levels reached ~51.1 nM [
33]. In female rats, a 5 ppm GEN diet administered from gestation through the lifetime produced serum levels of ~0.1 and 0.02 μM in adult offspring (PND 140) and weanlings (PND 21), respectively [
71]. Here, we show a dose-dependent effect of dietary GEN administered over the entire lifetime on basal
Brca1 CpG methylation in the mammary gland of adult mice. Compared with mice on a control diet (0 ppm GEN), mice fed the low-dose (GEN4) and high-dose (GEN10) diets had ~15% and ~50% less
Brca1 promoter methylation, respectively. This epigenetic effect was linked to the decreased expression of
Cyp1b1 (~50% decreased in GEN10 mice), which confirmed antagonistic effects of GEN toward the AHR. Women from populations with habitual high soy diets are presumed to be exposed to GEN in utero, which has been suggested to reduce BC risk later in life potentially by priming the mammary gland to differentiation [
72]. Studies with rodent models have also documented that the protective effect of GEN against mammary tumorigenesis may be dependent on pre-pubertal exposure, particularly starting at conception [
72]. Conversely, soy isoflavones (130 ppm) administered in periods that did not comprise the entire lifetime (i.e., not gestational through lifetime) actually increased spontaneous tumor multiplicity and mass in mammary tumor models (i.e., MMTV-neu transgenic mice) [
32,
33]. Pre-pubertal GEN exposure (500 ppm) was shown to increase
Brca1 expression and decrease DMBA-induced mammary tumor incidence and aggressiveness in mice [
35]. However, these effects were not observed in
Brca1+/− mice, possibly suggesting a dependency on
Brca1 expression for the protective effects of GEN against mammary tumorigenesis [
35]. Our data suggest that lifetime GEN may decrease BC risk in mice by decreasing basal
Brca1 promoter methylation.
Based on these in vivo results, a second objective of this study was to characterize the effects of GEN in a cell culture model of TNBC with constitutively active AHR. Previously, our group reported that activation of the AHR induced its colocalization at the
BRCA1 promoter with DNMT1, DNMT3a, and DNMT3b, leading to hypermethylation of a CpG island proximal to the
BRCA1 exon 1a [
40,
41]. Non-quantitative MSP and bisulfite sequencing was used by other investigators to show that CpG methylation at 30/30 CpG sites of the
BRCA1 promoter associated with decreased levels of BRCA1 protein and mRNA in HCC38 cells [
45]. Using real-time MSP, we documented here that the ratio of
mBRCA1/
umBRCA1 was ~10-fold higher in HCC38 compared to MCF7 cells. The
BRCA1 hypermethylation was observed in parallel with the marked upregulation of AHR expression (protein and mRNA). When compared to MCF7 cells, expression of the AHR target gene
CYP1A1 was decreased in HCC38 cells, whereas
CYP1B1 expression was elevated. This trend was in line with previous reports documenting that constitutively active AHR in rodent and human mammary tumors associated with elevated
CYP1B1, but not
CYP1A1 or mRNA [
47]. Earlier studies revealed that AHR overexpression and constitutive activity in TNBC cells was likely due to a positive amplification loop, whereby the AHR-dependent induction of tryptophan 2,3-dioxygenase (TDO2) caused the accumulation of endogenous AHR ligands in the form of tryptophan metabolites (e.g., kynurenine, kynurenic acid) [
73]. Alternatively, constitutive AHR expression and activity could be due to the loss of AHR repressor (AHRR) expression, which imparts a negative feedback regulation on AHR signaling and activity [
74].
We observed that the treatment of HCC38 cells with GEN and various high-affinity AHR antagonists (NF, CH-223191, GAL) increased cellular levels of the BRCA1 protein. This effect was consistent with the stimulatory effects of GEN and NF on BRCA1 expression in the UACC3199 cell line, which was due in part to CpG demethylation of the
BRCA1 promoter [
42]. In agreement with these data, we show here that GEN and GAL decrease CpG methylation at the
BRCA1 promoter in HCC38 cells. The dose of GEN (10 μM) used in this study was similar to the one required to CpG demethylate the
BRCA1 promoter in UACC3199 cells [
42]. In humans, micromolar levels of GEN are achievable in the blood either through prolonged dietary exposure or supplementation. Serum levels of GEN in adults consuming a soy-rich Asian diet (~50 mg isoflavones/d) have been shown to approach concentrations of ~0.1–1.2 μM [
75,
76]. Moreover, in a phase I clinical trial administering GEN (600 mg/d) to post-menopausal women over an 84-d intervention, mean serum levels were ~11.1 μM, and in some subjects, levels reached >30 μM [
77].
Several reports [
28,
29,
30,
31] support the capacity for GEN to demethylate tumor suppressor genes and reactivate their expression at concentrations similar to the one used here. In MCF7 (ERα-positive), MDA-MB-231 (TNBC), and MCF-10a (non-tumorigenic) cells, methylated DNA immunoprecipitation coupled to PCR amplification was used to determine that GEN (18.5 μM) treatment over 48 h demethylated both
BRCA1 and
BRCA2 [
28]. In another study, methylation-sensitive restriction analysis (MRSA) showed that 10 μM GEN over a 96-h period decreased methylation and increased the expression of
RARβ in both MC7 and MDA-MB-231 cells [
29]. These dose and time points mimic those used in the current investigation to demethylate
BRCA1 in HCC38 cells. A relatively lower dose (3.125 μM) of GEN, administered over a six-day period was also shown to decrease methylation (determined by MSP) and increase the expression of
GSTP1 in MDA-MB-468 (TNBC) but not MCF7 cells [
30]. Computational studies have demonstrated that the antagonistic effect of GEN against DNMT activity may be due to competitive binding with hemimethylated DNA at the catalytic site of DNMT1 [
31].
A study by Xu and colleagues reported on the inability of the DNMT inhibitor 5-azacytidine to demethylate and restore the expression of
BRCA1 in HCC38 cells [
45]. Thus, it is possible that the effect of GEN on
BRCA1 methylation and expression in HCC38 cells may not be due to the inhibition of DNMT1, but rather antagonism toward the AHR. Several lines of evidence support this speculation. First, the influence of GEN on
BRCA1 methylation and expression in HCC38 cells is analogous to that of known AHR antagonists, suggesting that the activation of
BRCA1 could be related to inhibitory effects on AHR binding/activity at the
BRCA1 promoter. Second, the treatment with GEN decreases the constitutive binding of AHR at
BRCA1 exon 1a, which is a response that is similar to the one elicited by the high-affinity AHR antagonists GAL and CH-223191. Third, in addition to antagonizing constitutively active AHR, both GEN and GAL prevent TCDD-induced binding of AHR at
BRCA1 exon 1a in MCF7 cells. This is consistent with our previous studies showing that GEN prevents TCDD-induced CpG methylation and the downregulation of
BRCA1 in MCF7 [
42].
Previous investigations suggest that GEN is a weak AHR antagonist [half maximal inhibitory concentration (IC
50) >50 μM]. However, in vitro gel mobility shift assays using cytosolic AHR from rat livers were used as opposed to whole cell model systems [
48]. In murine hepatoma Hepa-1c1c7 cells, GEN dose-dependently inhibited (0.1–20 μg/mL) TCDD-mediated activation of an XRE-driven reporter system, and in human HepG2 cells, GEN (50 μM) repressed the basal and TCDD-dependent expression of
CYP1A1 [
78]. In Caco2 colon cancer cells, GEN (50 μM) decreased nuclear AHR levels and prevented TCDD-induced AHR nuclear localization [
50]. In MCF7 BC cells, GEN dose-dependently (1–20 μM) decreased the basal expression of
CYP1A1 and
CYP1B1 [
79], but did not antagonize the TCDD-dependent (5 nM) activation of a reporter system at 1 and 10-μM doses [
80]. Studies in T47D ER+ human BC cells indicate that GEN may be a partial agonist/antagonist in BC [
51]. Alone, GEN (40 μM, max dose used in this study) was shown to act as a weak agonist, eliciting a reporter gene response <20% of that elicited by 10 nM TCDD. However, in T47D cells, 20 and 40-μM doses of GEN decreased the TCDD-mediated activation of the AHR-driven reporter system. This activity is similar to that of GAL in MCF7 cells. Alone, GAL increased
CYP1A1 expression, but attenuated TCDD-dependent and DMBA-dependent induction in co-treatment experiments [
81]. Our data suggest that in BC cells with activated AHR (either constitutive or induced), GEN exerts antagonism toward AHR-dependent CpG methylation of
BRCA1.
A significant clinical burden for TNBC patients is a lack of targeted therapeutics and reliance on systemic chemotherapy as the mainline neoadjuvant treatment option [
55]. Approximately, only 20–30% of TBNC patients have a pathological complete response to neoadjuvant therapy [
3]. The BRCA1 protein is a necessary factor for the transactivation of
ESR1. Previous studies demonstrated that the transfection of ERα-positive BC cells with siRNA against
BRCA1 silenced the expression of
ESR1, whereas the ectopic expression of a wild-type BRCA1 construct into the
BRCA1-mutated/ERα-negative HCC1937 cell line rescued the ERα protein [
52]. The latter outcome was shown to modulate the response of HCC1937 cells to the growth inhibitory effects of the antiestrogen fulvestrant. In the present study, we show that the upregulation of BRCA1 by GEN occurs in parallel with increased ERα expression. Moreover, cells pretreated with GEN were sensitized to the antiproliferative effects of 4-OHT, which is the active metabolite of the antiestrogen tamoxifen. Similar effects were observed with the control compound GAL. The capacity for GEN to sensitize TNBC cells to tamoxifen through ERα upregulation has been previously demonstrated in the MDA-MB-231 cell line, which is a model of TNBC [
82]. Moreover, in rats, the dietary administration of GEN starting from PND15 was shown to improve the response of DMBA-induced mammary tumors to tamoxifen therapy [
34]. Although previous studies in MDA-MB-231 cells linked this sensitization effect to decreased
ESR1 methylation [
82], in the present study, we showed no difference in
ESR1 methylation between HCC38 and MCF7 cells. These results lend support to the possibility that
BRCA1 hypermethylation may drive an ERα-negative phenotype through loss of the BRCA1-dependent transactivation of
ESR1, suggesting that compounds (i.e., GEN) that reduce
BRCA1 CpG methylation may hold promise as both preventive and adjuvant therapeutics for TNBC.