Effects of Benzophenone-3 and Propylparaben on Estrogen Receptor–Dependent R-Loops and DNA Damage in Breast Epithelial Cells and Mice

Abstract

Background:

Endocrine-disrupting chemicals have been shown to have broad effects on development, but their mutagenic actions that can lead to cancer have been less clearly demonstrated. Physiological levels of estrogen have been shown to stimulate DNA damage in breast epithelial cells through mechanisms mediated by estrogen-receptor alpha ($\text{ER}\mathrm{\alpha }$). Benzophenone-3 (BP-3) and propylparaben (PP) are xenoestrogens found in the urine of $>96\%$ of U.S. population.

Objectives:

We investigated the effect of BP-3 and PP on estrogen receptor–dependent transactivation and DNA damage at concentrations relevant to exposures in humans.

Methods:

In human breast epithelial cells, DNA damage following treatment with $17\mathrm{\beta }\text{-estradiol}$ (${\mathrm{E}}_2$), BP-3, and PP was determined by immunostaining with antibodies against $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ and 53BP1. Estrogenic responses were determined using luciferase reporter assays and gene expression. Formation of R-loops was determined with DNA: RNA hybrid–specific S9.6 antibody. Short-term exposure to the chemicals was also studied in ovariectomized mice. Immunostaining of mouse mammary epithelium was performed to quantify R-loops and DNA damage in vivo.

Results:

Concentrations of $1\mathrm{\mu }\mathrm{M}$ and $5\mathrm{\mu }\mathrm{M}$ BP-3 or PP increased DNA damage similar to that of ${\mathrm{E}}_2$ treatment in a $\text{ER}\mathrm{\alpha }\text{-dependent}$ manner. However, BP-3 and PP had limited transactivation of target genes at $1\mathrm{\mu }\mathrm{M}$ and $5\mathrm{\mu }\mathrm{M}$ concentrations. BP-3 and PP exposure caused R-loop formation in a normal human breast epithelial cell line when $\text{ER}\mathrm{\alpha }$ was introduced. R-loops and DNA damage were also detected in mammary epithelial cells of mice treated with BP-3 and PP.

Conclusions:

Acute exposure to xenoestrogens (PP and BP-3) in mice induce DNA damage mediated by formation of $\text{ER}\mathrm{\alpha }\text{-dependent}$ R-loops at concentrations 10-fold lower than those required for transactivation. Exposure to these xenoestrogens may cause deleterious estrogenic responses, such as DNA damage, in susceptible individuals. https://doi.org/10.1289/EHP5221

Introduction

Endocrine-disrupting chemicals (EDCs) alter the endocrine system by binding directly to the receptors and modulating downstream signaling pathways. Xenoestrogens are structurally diverse EDCs that affect estrogen receptor (ER) signaling pathways. BP-3 (oxybenzone, or 2-Hydroxy-4-methoxybenzophenone, CAS No. 131-57-7) is a UV-filter used in personal care products, such as sunscreens, cosmetics, and lotions, with concentrations up to 0.148% (Liao and Kannan 2014) and a maximum allowed concentration of 6% by Food and Drug Administration (FDA) and European commission (EC 2017). BP-3 was detected in the urine samples of 96.8% of U.S. population in the 2003–2004 National Health and Nutrition Examination Survey (NHANES) conducted by the Centers for Disease Control and Prevention (CDC) (Calafat et al. 2008). Similarly, PP (propyl parahydroxybenzoate, CAS No. 94-13-3) is widely used as an antimicrobial agent in food and personal care products. Although the FDA limits PP to 0.1% in food, currently there is no specific limit for preservatives in personal care products. PP is banned as a food preservative, and maximum permissible levels in personal care products is 0.4% in the European Union (EU) (Snodin 2017; EC 2014). PP was detected in the urine samples of $>96\%$ of U.S. population surveyed during 2003–2005 by the CDC (Ye et al. 2006).

Estrogenic responses are determined by the action of two distinct estrogen receptor (ER) subtypes, estrogen receptor $\mathrm{\alpha }$ ($\text{ER}\mathrm{\alpha }$) and estrogen receptor $\mathrm{\beta }$ ($\text{ER}\mathrm{\beta }$). Ligand-activated ER recruits coactivators to estrogen response elements (ERE) in promoters of target genes leading to transcription initiation (Shang et al. 2000; Yi et al. 2017). In $\text{ER}\mathrm{\alpha }$ expressing breast cancer cells, proliferation is among the types of cellular responses (Henderson et al. 1988; Musgrove and Sutherland 1994). Hence, estrogenic responses to putative xenoestrogens is most often determined by transactivation of ERE-reporters, endogenous gene expression and cell proliferation in ER-expressing MCF-7 and T47D cell lines, where $\text{ER}\mathrm{\alpha }$ is the dominant subtype (Buteau-Lozano et al. 2002; Vladusic et al. 2000). These studies showed BP-3 was a weak agonist of ER at $1\mathrm{\mu }\mathrm{M}$ (Kerdivel et al. 2013; Schlotz et al. 2017; Schlumpf et al. 2001). BP-3 was found in the urine samples of 25 volunteers who used sunscreen containing 4% BP-3 twice a day for 5 d, suggesting it was readily absorbed through skin (Gonzalez et al. 2006). Metabolites of BP-3, such as 2,4-diOH-BP and 2,3,4-triOH BP, were shown to form by oxidation in rat and human liver microsomes (Okereke et al. 1994; Watanabe et al. 2015). 2,4-diOH-BP was detected in the urine samples of women scheduled to undergo a diagnostic and/or therapeutic laparoscopy or laparotomy as part of the ENDO study (Kunisue et al. 2012) and was shown to have higher ER transactivation potential in comparison with BP-3 (Watanabe et al. 2015). BP-3 and BP-3 metabolite 4,4′-dihydroxybenzophenone were also detected in 27 of the 79 breast milk samples from mothers who had normal pregnancy and delivery, and who participated in the Breast Milk Bank of the Blood and Tissue Bank of Catalonia (Spain) (Molins-Delgado et al. 2018). Exposure of BP-3 during pregnancy and lactation in mice resulted in altered mammary gland ductal architecture that persisted for weeks after exposures ended (LaPlante et al. 2018). Long-term exposure of MCF-7 breast cancer cells to $100\mathrm{\mu }\mathrm{M}$ BP-3 for $>20\text{\hspace{0.17em} weeks}$ increased the motility of these cells (Alamer and Darbre 2018). This increase was also observed in estrogen nonresponsive cell line MDA-MB-231, suggesting alternate pathways of BP-3 actions at this dose. Similarly, PP was shown to be an effective ER-agonist with 1.3-fold induction of gene expression using reporter assays (ERE-CAT reporter) at $10\mathrm{\mu }\mathrm{M}$, increased expression of estrogen-responsive gene Trefoil Factor 1 (TFF1, also known as pS2) and increased proliferation of MCF-7 cells at $1\mathrm{\mu }\mathrm{M}$ (Byford et al. 2002). Proliferation induced by PP was inhibited by ER antagonist (fulvestrant) indicating dependence on $\text{ER}\mathrm{\alpha }$. PP also increased cell motility (increased scratch closure) in both short-term (7-d) and long-term (20-wk) treatments in the MCF-7 cell line (Khanna et al. 2014).

In addition to stimulating cell proliferation and motility, estrogen also induces genotoxicity and DNA damage and is considered a major risk factor in breast cancer etiology (Roy and Liehr 1999; Yager and Davidson 2006). Estrogen has been shown to induce DNA damage by a) metabolic activation of estrogen and b) hormonal carcinogenesis (Santen et al. 2009). ${\mathrm{E}}_2$ is metabolized to form catechol estrogens (${16\mathrm{\alpha }\text{-OHE}}_2$ or ${2\text{-OHE}}_2$ and ${4\text{-OHE}}_2$), which can be oxidized to form reactive semiquinone (SQ) intermediates and quinone derivatives. Two such compounds, ${\mathrm{E}}_2$-3-4-Q and ${\mathrm{E}}_2$-2-3-Q form stable DNA adducts or depurinating adducts, such as ${4\text{-OHE}}_2$-1-N7Gua and ${4\text{-OHE}}_2$-1-N3Ade, which were associated with increased breast cancer risk, but micromolar levels of ${\mathrm{E}}_2$-3-4-Q and ${\mathrm{E}}_2$-2-3-Q were required to show DNA adduct formation in vitro (Cavalieri and Rogan 2016). The SQ and quinone derivatives can also generate ROS through redox cycling, which can be genotoxic (Fussell et al. 2011; Wang et al. 2010). Similarly, ER–independent DNA damage was shown in $\text{ER}\mathrm{\alpha }\text{-negative}$ cell lines using the COMET assay (Rajapakse et al. 2005), cII mutagenesis assay (Zhao et al. 2006), or LOH (Huang et al. 2007; Russo et al. 2003). The concentrations of ${\mathrm{E}}_2$ or ${4\text{-OHE}}_2$ used in these studies were $\ge 70\text{\hspace{0.17em} nM}$, with the exception of Russo et al. 2003, who reported increased clonal efficiency of MCF10F cells at $0.007\text{\hspace{0.17em} nM}$. However, the median ${\mathrm{E}}_2$ level during pregnancy is $74\text{\hspace{0.17em} nM}$ and $<2\text{\hspace{0.17em} nM}$ in normal cycling women (Table 1), and the level of circulating estradiol metabolites are 100-fold lower (Xu et al. 2007; Ziegler et al. 2010). Clinical data show that in postmenopausal women with ER-positive early breast cancer, endocrine therapy with an aromatase inhibitor was associated with significantly lower recurrence than tamoxifen (TAM) therapy (EBCTCG 2015), which could be because of either lower levels of estrogen metabolites or reduced ER activation. Epidemiological data show that for a given level of total estrogen, increased levels of ${4\text{-OHE}}_2$, ${2\text{-OHE}}_2$ and ${16\text{-OHE}}_2$ are associated with reduced risk in breast cancer (Dallal et al. 2014; Moore et al. 2016; Sampson et al. 2017) or no independent association with risk (Sampson et al. 2017); however, an earlier study reported ${4\text{-OHE}}_2$ levels to be associated with higher breast cancer risk (Fuhrman et al. 2012). Hence, the impact of metabolic activation of estrogen at physiologically relevant concentrations on DNA damage remains to be demonstrated.

Table 1.

Estimation of estrogen and xenoestrogens concentrations of estradiol (${\mathrm{E}}_2$), benzophenone-3 (BP-3), or propylparaben (PP) in urine/blood samples of women and female mice.

Ligand	Median ($\mathrm{\mu }\mathrm{M}$)	90th or 95th percentile ($\mathrm{\mu }\mathrm{M}$)	Relative transactivation activity at 90th or 95th percentile (% RTA vs. ${\mathrm{E}}_2$)	References
BP-3 (urine)
Non-Pregnant	0.137	6.70b	$18.91\pm 6.62\%$	Woodruff et al. 2011
Pregnant	0.47	29.5b	$27.16\pm 6.2\%$	Philippat et al. 2013
PP (urine)
Non-Pregnant	0.161	1.98b	$64.27\pm 20.5\%$	Calafat et al. 2010
Pregnant	0.253	3.26b	$104.07\pm 20.98\%$	Philippat et al. 2013
${\mathrm{E}}_2$ (blood)
Human
Ovulatory	0.0003–0.0018			Clarke et al. 1997
Luteal	0.0002–0.0008			O'Leary et al. 1991
Pregnant	0.074	0.118a		Schock et al. 2016
Mouse	$<0.0003$			Majewski et al. 2018

^a90th percentile of exposure in the given population.

^b95th percentile of exposure in the given population.

Hormonal carcinogenesis is postulated to act through ER to initiate lesions as well as stimulate progression of tumors. ${\mathrm{E}}_2$ treatment stimulated renal tumors in male Syrian hamsters (Liehr et al. 1988). TAM reduced tumors but did not alter levels of DNA-adducts, suggesting the primary effect of ${\mathrm{E}}_2$ being mediated by ER. Similarly, blockage of ER activation through selective estrogen receptor modulators (SERMs), such as TAM and raloxifene, reduced the incidence of breast cancer by 50%–75% in women (Cummings et al. 1999; Cuzick and International Breast Cancer Intervention Study 2001; Martino et al. 2004). Bilateral oophorectomy and hysterectomy in women under 40 years of age reduced breast cancer later in life by 75% (Feinleib 1968). Administration of aromatase inhibitor (exemestane) for 35 months to a cohort of postmenopausal women with Gail score of 1.66 and prior atypical ductal/lobular hyperplasia or ductal carcinoma in situ treated with mastectomy but noncarriers for BRCA1/2 and no prior invasive ductal carcinoma resulted in 65% relative reduction of breast cancer (Goss et al. 2011). Mobley and Brueggemeier showed that 8-oxo-dG production with buthionine sulfoximine (BSO), ${\mathrm{E}}_2$ ($10\text{\hspace{0.17em} nM}$) and H2O2 treatment could be reduced with TAM treatment in ER-positive MCF7 cells but not in ER-negative MDA-MB-231 cells, suggesting DNA damage was at least partially ER-mediated (Mobley and Brueggemeier 2004). Stork et al. showed lack of DNA damage marker $\mathrm{\gamma }\text{-}\mathrm{H}2\mathrm{A}\mathrm{X}$ (phosphorylated H2AX) in MCF10A cells following treatment of $10\text{\hspace{0.17em} nM}$ and $100\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$ for $24\mathrm{h}$ (Stork et al. 2016). In T47D cells, ${\mathrm{E}}_2$-mediated $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ was diminished with treatment of ER inhibitors like TAM or fulvestrant (Periyasamy et al. 2015). ER signaling stimulates proliferation, which was causally linked to tumorigenesis, by increasing the probability of replication errors, which are propagated in daughter cells (Henderson and Feigelson 2000; Preston-Martin et al. 1990). Therefore, ${\mathrm{E}}_2$ can be considered as a carcinogen through its actions on progression of cancer that was initiated by other factors.

The studies involving DNA damage by ${\mathrm{E}}_2$ have used different cell lines, tissues, and end points. Therefore, there is no consistent way to discriminate the contribution of ER-dependent and ER-independent mechanisms across published studies. It is possible that both mechanisms contribute to ${\mathrm{E}}_2$-mediated carcinogenesis.

Recent studies have shown that ER stimulation leads to transcription-coupled DNA damage, suggesting a distinct mechanism. Interaction of $\text{ER}\mathrm{\alpha }$ with chromatin forms transcriptional coactivator/corepressor complexes to initiate transcription (Chao et al. 2002; Fullwood et al. 2009; Shang et al. 2000). The open chromatin in these $\text{ER}\mathrm{\alpha }$ complexes were susceptible to DNA damage by formation of RNA:DNA triplex structures, called R-loops (Stork et al. 2016). Therefore, estrogen can stimulate carcinogenesis by initiating direct DNA damage mediated by $\text{ER}\mathrm{\alpha }$ and proliferation that expands the population of breast cells.

Bioassays of transcriptional activities have been valuable in rapidly assessing the risk posed by xenoestrogens. However, it is unclear whether the transcriptional activities of xenoestrogens reflect their potential mutagenic activity mediated by $\text{ER}\mathrm{\alpha }$. DNA damage by selective $\text{ER}\mathrm{\alpha }$ agonists such as diethylstilbestrol (DES) and 4,4′,4′′-(4-Propyl-[1H]-pyrazole-1,3,5-triyl) trisphenol (PPT) (Periyasamy et al. 2015) suggest that transcriptional DNA damage needs to be assessed to determine potential breast cancer risk posed by xenoestrogens. In this study, we evaluated effects of two xenoestrogens, BP-3 and PP, which differ in structure and transcriptional potency and compared these with ${\mathrm{E}}_2$.

Methods

Cell Culture

T47D (ATCC #HTB-133), T47DKBluc (ATCC #CRL-2865) and MCF-7 (ATCC #HTB 22) cells were passaged in growth media containing phenol red–free (PRF) DMEM-F12 (Sigma #D6434) or MEM 1X (Gibco #51200-038) with 10% heat-inactivated FBS (Omega Scientific #FB-02) and $10\mathrm{\mu }\mathrm{g}/\text{ml}$ insulin (Sigma #9278), $2\text{\hspace{0.17em} mM}$ L-glutamine (Hyclone #SH30034.01), gentamycin $15\mathrm{\mu }\mathrm{g}/\text{ml}$ (Gibco #15750-060), and 1X antibiotics/antimycotics (AB/AM, Gibco #15240-062) and incubated at 37°C with 5% ${\mathrm{CO}}_2$. For experiments, cells were grown in clearing media with charcoal-stripped serum (CSS) (MEM 1x with 10% charcoal dextran–treated FBS (Omega Scientific #FB-04), $10\mathrm{\mu }\mathrm{g}/\text{ml}$ insulin, and $2\text{\hspace{0.17em} mM}$ L-glutamine) for $24–72\mathrm{h}$ before being plated for experiments.

The 76N-Tert cell line, a human mammary epithelial cell line immortalized with expression of human telomerase reverse transcriptase (TERT), was a gift from Dr. Vimla Band (Zhao et al. 2010). These cells were grown in F-media [$250\mathrm{mL}$ DMEM (-pyruvate) (Gibco #11965-092), $250\mathrm{mL}$ Ham’s F12 (Gibco #11765-054), 5% FBS, $250\text{\hspace{0.17em} ng}/\mathrm{mL}$ hydrocortisone (Sigma #H4001), $10\text{\hspace{0.17em} ng}/\mathrm{mL}$ human epidermal growth factor (Tonbo Biosciences #21-8356-U100), $8.6\text{\hspace{0.17em} ng}/\mathrm{mL}$ cholera toxin vibrio (Millipore Sigma #227035), $1\mathrm{\mu }\mathrm{g}/\mathrm{mL}$ human insulin solution, and 1X antibiotic/antimycotic] and passaged every 2–3 d.

Generation of 76N-Tert-ESR1 Cells

An inducible $\text{ER}\mathrm{\alpha }$ (ESR1) construct was generated using the pINDUCER14 vector (Meerbrey et al. 2011). Specifically, FLAG tag sequence was amplified from pFLAG-CMV-2 (Andersson et al. 1989) using forward primer 5′-ATACCGGTACCATGGACTACAAAGACGATGACGAC-3′ and reverse primer 5′-TCGACCGGTACGCGTGCGATCGCTGAATTCGCGGCAAG-3′. The amplified FLAG sequence was then cleaned using the Monarch PCR & DNA Cleanup Kit (NEB #T1030) and ligated into pINDUCER14 by digesting both plasmids with AgeI, performing dephosphorylation with shrimp alkaline phosphatase (NEB #M0371S) and gel electrophoresis and extracting from agarose gel (DNAland Scientific #GP1001). Sequencing of pINDUCER14-FLAG confirmed that the FLAG sequence was inserted.

ESR1 was amplified from a plasmid expressing ESR1 made in our lab (pIRES-hrGFPII-ESR1, unpublished data). pIRES-hrGFPII-ESR1 contained the ESR1 cDNA sequence (Open Biosystems #MHS6278-211691051) in the multiple cloning site of the pIRES-hrGFPII vector (Stratagene #240157). ESR1 was amplified from pIRES-hrGFPII-ESR1 using forward primer 5′- GCAGAAATGACCATGACCCTCCACACCAAAGC-3′ and reverse primer 5′- TAAACGCGTTCAGACCGTGGCAGGGAAACCCT-3′. Ligation of ESR1 into pINDUCER14-FLAG was done by digesting both plasmids with EcoRI and MluI and then performing dephosphorylation, cleanup, and extraction as described above. Two linker sequences (Linker A: 5′-AATTGCGCGATCGCGG-3′ and Linker B: 5′-AATTCCGCGATCGCGC-3′) between FLAG and ESR1 were added to keep the ESR1 sequence in frame. Sequencing of this final pINDUCER14-FLAG-ESR1 confirmed that all inserts were in the correct orientation relative to the vector, both FLAG and ESR1 were in frame, and the ESR1 sequence was identical to the Homo sapiens ESR1 gene (Sequencing Primers: F: 5′-CGGTGGGAGGCCTATATAAG-3′, M: 5′-GCTACCATTATGGAGTCTGG-3′, and R: 5′-ACTTATATACGGTTCTCCCC-3′). This final construct was referred to as pIND-ESR1 and expressed a constitutive GFP reporter as well as $\text{ER}\mathrm{\alpha }$ with N-terminal FLAG tag.

In addition, 293T cells were cultured in DMEM:F12 (Sigma #D8900) supplemented with 10% FBS, $15\mathrm{\mu }\mathrm{g}/\mathrm{mL}$ gentamycin (Gibco #15750-060), and 1x antibiotic/antimycotic. Cells were lifted with 0.05% trypsin and plated in $60\text{-}\mathrm{mm}$ tissue culture dishes at $2.5\times {10}^6$ cells per dish for next-day use. 293T cells were then transfected with $3.5\mathrm{\mu }\mathrm{g}$ pIND-ESR1, $3\mathrm{\mu }\mathrm{g}$ psPAX2 (Addgene #12260) (gag, pol, and rev packaging vector), and $2\mathrm{\mu }\mathrm{g}$ pMD2.G (Addgene #12259) (vsv-g packaging vector) in antibiotic-free media using Lipofectamine 2000 (Thermo Fisher Scientific). Media was refreshed after $24\mathrm{h}$, and viral media was collected at 48 and $54\mathrm{h}$ post initial transfection. Viral media from transfected 293T was filtered using a $0.45\mathrm{\mu }\mathrm{M}$ filter (Corning #431220) and added to 76N-Tert cells twice, $6\mathrm{h}$ apart, in a 1:1 ratio with F-media. After $24\mathrm{h}$, viral media was removed and replaced with F-media. Following cell expansion, the cells were pooled and resuspended in 1% FBS in PBS. Selection of the stably transduced cells was performed by FACS for GFP-positive cells using FACSAria II (Becton-Dickinson). 76N-Tert uninfected cells were used as a control to set the background fluorescence. Approximately 5% cells were GFP-positive, suggesting pIND-ESR1 expression. The GFP-positive cells were collected to 90% purity. These cells were expanded and referred to as 76N-Tert-ESR1.

Luciferase Reporter Assay

T47DKBluc cells were grown in clearing media for $72\mathrm{h}$ and plated in a 24-well plate at $10\times {10}^5$ cells/well density. After $24\mathrm{h}$, cells were treated with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$ ($17\mathrm{\beta }\text{-estradiol}$, Sigma #E2758), $10\text{\hspace{0.17em} nM}$ fulvestrant (F, ICI 182, 780, Tocris #1047), 0.5 to $50\mathrm{\mu }\mathrm{M}$ BP-3 (Sigma #H36206) or 0.5 to $50\mathrm{\mu }\mathrm{M}$ PP (Sigma #P53357). Stock solutions were prepared in DMSO (Sigma #D8418), then diluted to working concentrations in media. Luciferase assays were performed using the Promega Dual-Luciferase Reporter Assay (Promega #E1910). Cells were lysed in 1X Passive Lysis Buffer after treatment for $24\mathrm{h}$ and then stored at $–20°\mathrm{C}$. Luciferase activity was determined in lysates by using the Polar Star OPTIMA plate reader (BMG Labtech) and expressed in relative light units (RLU). Treatments were compared with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$ included on the plate, and relative transactivation activity (RTA) is defined as percent transactivation in comparison with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$.

RT-qPCR

RNA from T47D cells, MCF-7 cells, or flash-frozen fourth mammary gland was isolated with TRIzol (ThermoFisher Scientific #15596018) and Direct-zol RNA MiniPrep Plus (Zymo Research #R2072). cDNA was prepared from $1\mathrm{\mu }\mathrm{L}$ of RNA in $20\mathrm{\mu }\mathrm{L}$ reaction mix with Protoscript II First Strand cDNA Synthesis Kit (New England Biolabs #E6560S), following the standard protocol provided by the manufacturer. qPCR for TFF1, progesterone receptor (PGR/Pgr), and Amphiregulin (AREG/Areg) was performed using primers in Table S1 (Integrated DNA Technology) and iTaq Universal SYBR Green Supermix (Biorad, #1725121) on CFX96 Real-Time System thermocycler (Bio-Rad). Each run (96-well qPCR plate) included an inter-run calibrator to normalize across experiments. No housekeeping gene was included in the experiment to avoid possible variation due to treatments. Results represent average of three experiments. Data was analyzed with $\mathrm{\Delta }\mathrm{\Delta }\text{Ct}$ method, and relative fold change in expression of target gene was compared among control and treatments.

Cell Proliferation Assay

T47D cells grown in clearing media for $72\mathrm{h}$ was plated as $100\mathrm{\mu }\mathrm{l}$ of cells suspension having 5,000 or 10,000 cells per well on five 96-well plates (one for each day). The 96-well plate had 12 cell-free wells for a blank and 7 wells per treatment on each plate. After $24\mathrm{h}$, media was changed in appropriate wells on each plate to reach the desired final concentration of ${\mathrm{E}}_2$ ($0.5\text{\hspace{0.17em} nM}$), BP-3 (5, $50\mathrm{\mu }\mathrm{M}$) or PP (1, $10\mathrm{\mu }\mathrm{M}$) in the given wells. All plates were maintained in a ${37}^{\mathrm{o}}\mathrm{C}$, 5% ${\mathrm{CO}}_2$ incubator until media were exchanged, on one plate per day, for 10% Alamar Blue in plating media. Plates were read at the same time each day at $4\mathrm{h}$ and $8\mathrm{h}$ after media exchange on a BioTek Synergy 2 plate reader (Bio-Tek) at $570\text{\hspace{0.17em} nm}$ and $600\text{\hspace{0.17em} nm}$. Percent Alamar Blue reduction was calculated as per the Alamar Blue protocol:

$$\text{Percent reduced}=\frac{(\mathrm{117,216}\times \text{test \hspace{0.17em} well \hspace{0.17em}}{\mathrm{A}}_{570})-(\mathrm{80,586}\times \text{test \hspace{0.17em} well \hspace{0.17em}}{\mathrm{A}}_{600})}{[\mathrm{155,677}\times \text{mean \hspace{0.17em}}(\text{negative \hspace{0.17em} control \hspace{0.17em}}{\text{well \hspace{0.17em}}\mathrm{A}}_{600}\left)\right]-[\mathrm{14,652}\times \text{mean \hspace{0.17em}}({\text{negative \hspace{0.17em} control \hspace{0.17em} well \hspace{0.17em}}\mathrm{A}}_{570}\left)\right]}\times 100$$

Immunostaining

T47D, MCF7, 76N-Tert, or 76N-Tert-ESR1 cells were grown in clearing media for $48\mathrm{h}$ and plated on $20\mathrm{mm}$ glass uncoated coverslips in 12-well plates with a density of $2\times {10}^5$ cells/well. After $24\mathrm{h}$ of growth, cells were treated with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$, 1 or $5\mathrm{\mu }\mathrm{M}$ BP-3, and 1 or $5\mathrm{\mu }\mathrm{M}$ PP with or without $1\mathrm{\mu }\mathrm{M}$ fulvestrant for $24\mathrm{h}$. For $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}/53\text{BP}1/\text{ER}\alpha$, cells were fixed in ice-cold methanol (100%) for 10 min and quenched with $0.1\mathrm{M}$ glycine for 15 min. Cells were washed with 1X PBS, blocked in 2% BSA/PBS with 0.1% Triton-X 100 for $1\mathrm{h}$ at room temperature (RT), incubated overnight with anti-$\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ antibody (Cell Signaling #9718S), $\text{anti-ER}\mathrm{\alpha }$ antibody (Santa Cruz Biotechnology #sc-8002) or anti-53BP1 antibody (Abcam #ab36823) at 4°C, followed by $1\mathrm{h}$ with anti-rabbit AlexaFluoro 488–conjugated secondary antibody (Cell Signaling #8889S) or anti-mouse AlexaFluoro 488–conjugated (Cell Signaling #4408S) at RT. For S9.6, cells were fixed in ice-cold 100% methanol for 10 min at $–20°\mathrm{C}$, permeabilized in 100% acetone for 1 min at RT, blocked for 30 min in saline sodium citrate pH 7 (SSC, 4X), 3% BSA, and 0.1% Triton-X and incubated with S9.6 antibody (Kerafast #ENH001) for $2\mathrm{h}$ at RT, followed by $1\mathrm{h}$ with anti-mouse AlexaFluoro 596–conjugated secondary antibody (Life Technologies #A11062) or antimouse AlexaFluoro 488–conjugated (Cell Signaling #4408S). For each treatment, two replicates of slides were stained with one set of replicates treated with RNase H (NEB #M0297L) for $4\text{\hspace{0.17em} hr}$ at 37°C prior to incubation with primary antibody. Stained cells were mounted with Vectashield mounting medium containing DAPI (Vector Laboratories #H-1,200). Slides were imaged at 60X (immersion oil) with Nikon A1 spectral confocal microscope. Analysis of $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ and S9.6 intensity per nucleus or foci per nucleus was calculated using Nikon analysis software, where DAPI was used as a mask for the nucleus.

Western Blot

Cells from MCF7 grown in growth media, 76N-Tert (parental), 76N-Tert-ESR1, and 76N-Tert-ESR1 grown in F-media treated with doxycycline for $24\mathrm{h}$ and 76N-Tert-ESR1 treated with doxycycline and $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$ for $24\mathrm{h}$ were lysed with ice-cold RIPA lysis buffer [$50\text{\hspace{0.17em} mM}$ Tris–HCl, pH 8.0; $150\text{\hspace{0.17em} mM}$ NaCl; $1\text{\hspace{0.17em} mM}$ EDTA; 1% Triton X-100; 1% Sodium deoxycholate; 0.1% SDS; 1% protease inhibitors (Sigma-Aldrich #P8340), 1% phosphatase inhibitor #2 (Sigma-Aldrich #P5726), and 1% phosphatase inhibitor #3 (Sigma-Aldrich #P0044)]. Homogenate was centrifuged at $\mathrm{13,000}\text{\hspace{0.17em} rpm}$ for 15 min at 4°C to remove cellular debris. Protein quantification was performed using BCA protein assay (Thermo Scientific #23225). Equal amounts of protein ($28\mathrm{\mu }\mathrm{g}$) were separated by SDS-PAGE on 10% acrylamide under denaturating conditions and then blotted onto PVDF membrane (Millipore #IPVH00010). Nonspecific binding was blocked with 5% nonfat dry milk in TBST (Tris-buffered saline and Tween® 20 containing $10\text{\hspace{0.17em} mM}$ Tris-HCl, pH 7.5; $150\text{\hspace{0.17em} mM}$ NaCl; 0.05% Tween® 20) for $1\mathrm{h}$. The blot was incubated with 1:100 $\text{anti-ER}\mathrm{\alpha }$ antibody (Abcam #ab16660) overnight at 4°C. After incubation, the blot was washed with TBST and then incubated with HRP-conjugated secondary antibody (1:5000, GE Healthcare #NA934V) for $1\mathrm{h}$. Bands were detected using enhanced chemiluminescence solution and visualized using G-box imaging system (Syngene). The blot was washed with TBST and incubated with $\text{anti -}\mathrm{\beta }$ actin antibody (1:5000, Sigma #A1978) overnight at 4°C. After washing with TBST and HRP secondary antibody incubation for $1\mathrm{h}$ (1:5,000, GE Healthcare #NA931C), bands were detected with enhanced chemiluminescence and G-box system. Expected molecular weights were 67 kDa ($\text{ER}\mathrm{\alpha }$) and 42 kDa ($\mathrm{\beta }$ actin).

Animal Treatment

Forty mature BALB/c female mice (8 wk old) were purchased from Jackson Laboratory and housed in temperature-controlled facilities with a set temperature of $17.8–26.11°\mathrm{C}$ and humidity of 30%–70%, $12\text{-}\mathrm{h}$ alternating day/night light cycle and fed LabChow 5058 ad libitum. All procedures were in accordance with the national guidelines for the care and use of animals and approved by the University of Massachusetts Amherst’s Institutional Animal Care and Use Committee.

The mice were ovariectomized before treatment. Briefly, each mouse was anesthetized with a mix of isoflurane and oxygen. The flanks were shaved, sterilized with povidone-iodine (Betadine) and cleaned with alcohol. An incision was made to the skin on the right flank. The underlying muscle layer was nicked to reveal a small hole through which the ovary was pulled out by grasping the periovarian fat. A Serrefine clamp was used to hold the ovary. After ensuring that the blood vessels were constricted to prevent breeding, the ovary was cut from the uterine horn. The periovarian fat was restored into the peritoneum. The peritoneum was closed with one or two stiches and the skin was closed with $9\text{-}\mathrm{mm}$ wound clips. The procedure was repeated on the contralateral side. The mouse was monitored for a week post procedure, and wound clips were removed after 10 d. After 1 wk of recovery, the mice were randomized to four groups and began an acute oral treatment via pipette with vehicle control (tocopherol-stripped corn oil) ($n=7$) or one of three different compounds ${\mathrm{E}}_2$ ($n=8$), BP-3 ($n=12$), and PP ($n=12$) for 4 d. Each mouse was administered $1\mathrm{\mu }\mathrm{L}$ of oil per gram of body weight to deliver $250\mathrm{\mu }\mathrm{g}/\mathrm{kg}/\mathrm{d}$ ${\mathrm{E}}_2$, $\mathrm{3,000}\mathrm{\mu }\mathrm{g}/\mathrm{kg}/\mathrm{d}$ BP-3, or $\mathrm{10,000}\mathrm{\mu }\mathrm{g}/\mathrm{kg}/\mathrm{d}$ PP or vehicle control. For BP-3 and PP, these doses represent the toxicologically no-adverse-effect-level (NOAEL) doses for each compound based on development and reproductive toxicity assays (Scientific Committee on Consumer Products 2005, 2008; Soni et al. 2001).

Six hours prior to sacrifice, all of the mice were treated with a 5-gray (Gy) dose of gamma radiation. Then 2 h before sacrifice, all mice were injected intraperitoneally with $70\mathrm{\mu }\mathrm{g}/\mathrm{g}$ body weight of BrdU (Sigma Aldrich; Cat. #B5002) that was previously prepared at $10\mathrm{mg}/\text{ml}$ in PBS and filter sterilized. The mice were sacrificed using carbon dioxide followed by cervical dislocation. Whole blood was collected by cardiac puncture and tissues were harvested. One of the fourth mammary gland was fixed in 10% NBF and transferred to 70% alcohol prior to paraffin embedding. The other fourth mammary gland was cleared of lymph node and stored in $–70°\mathrm{C}$. The whole blood was allowed to coagulate at RT for 20 min and then spun down at $2000\times g$ for 10 min at $4°\mathrm{C}$ to retrieve the serum.

Immunostaining of Mouse Mammary Gland

Freshly cut $4\text{-}\mathrm{\mu }\mathrm{M}$ paraffin-embedded sections were deparaffinized/rehydrated with 100% xylenes 3 times for 5 min each, 2 times with 100% ethanol for 5 min each, 95% ethanol for 3 min, and 70% ethanol for 3 min. Samples were rinsed with PBS. Antigen unmasking was performed by boiling the samples in $1\text{\hspace{0.17em} mM}$ EDTA for $1\text{\hspace{0.17em} hr}$. Samples were cooled down to RT and then treated with SSC 0.2X with gentle shaking at RT for 20 min. Samples were blocked in 3% BSA in PBS with 0.5% Tween® 20 for $1\mathrm{h}$ at RT. Primary antibody incubation was done with monoclonal S9.6 antibody (Kerafast #ENH001) or anti-H2AX antibody (Cell Signaling #9718S) for overnight at 4°C. After primary incubation, samples were washed 3 times with PBS containing 0.5% Tween® 20 and then incubated with anti-mouse AlexaFluoro 488–conjugated (Cell Signaling #4408S) or anti-rabbit AlexaFluoro 488–conjugated secondary antibody (Cell Signaling #8889S) for $1\mathrm{h}$ at RT. Samples were washed 2 times with PBS containing 0.5% Tween® 20 and 2 times with PBS and then mounted with Vectashield mounting medium containing DAPI. Slides were imaged at 60× with Nikon A1 spectral confocal microscope. Analysis of S9.6 intensity per nucleus or foci per nucleus were calculated using Nikon analysis software, where DAPI was used as a mask for the nucleus. IHC for Ki67 was performed on a DakoCytomation autostainer using 1:1,000 D2H10 primary antibody (cell signaling #9027T) and the Envision HRP detection system (Dako). Positive cells were counted using ImageJ software. A total of 1,200 cells were counted per slide to determine percent Ki67 positive.

ELISA

The serum from whole blood that was harvested from all the mice were quantified using a ${\mathrm{E}}_2$–specific enzyme-linked immunosorbent assay (ELISA) (Calbiotech #ES180S-100).

Statistical Analyses

Unless specified, data were analyzed by one-way analysis of variance (ANOVA) followed by Tukey's honestly significant difference (HSD) multiple-range test using GraphPad Prism 8 statistical analysis software or R program (version 3.6.0; R Development Core Team). The difference between control and fulvestrant/RNase H treated groups were evaluated with two-way ANOVA followed by Bonferroni correction. Results are presented as mean ± standard error of the mean (SEM). Data were considered statistically significant at $p<0.05$. Growth curves were fitted to linear regression model, and slopes were compared between control and treatment conditions. Slopes and 95% confidence interval are reported in Table 2.

Table 2.

Slopes of growth curve showing effect of estradiol (${\mathrm{E}}_2$), benzophenone-3 (BP-3), or propylparaben (PP) on T47D cells.

Growth Curve	Slope	95% CI
Control DMSO	0.0107	−0.006811, 0.02821
$0.5\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$	0.08495	0.06604, 0.1039
$1\mathrm{\mu }\mathrm{M}$ PP	0.01856	0.003943, 0.03318
$10\mathrm{\mu }\mathrm{M}$ PP	0.06387	0.05225, 0.07550
$5\mathrm{\mu }\mathrm{M}$ BP-3	0.0202	0.008131, 0.03226
$50\mathrm{\mu }\mathrm{M}$ BP-3	0.01581	0.0009721, 0.03064

Results

DNA Damage and TFF1 Gene Expression in Cells Treated with $E_2$, BP-3, or PP

We monitored $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ foci as a measure of DNA damage in T47D cells treated with the compounds for $24\mathrm{h}$. A dose-dependent increase in $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity was observed following ${\mathrm{E}}_2$ treatment (Figure 1A). Treatment with either BP-3 or PP also led to an increase in $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity. Treatment with BP-3 at 1 or $5\mathrm{\mu }\mathrm{M}$ increased $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity in comparison with the control ($p<0.0001$) although we did not observe a dose-dependent increase ($1\mathrm{\mu }\mathrm{M}$ BP-3 vs $5\mathrm{\mu }\mathrm{M}$ BP-3, Figure 1B). PP treatment also resulted in significantly increased $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity at 1 and $5\mathrm{\mu }\mathrm{M}$ in comparison with the control ($p<0.0001$). The $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity due to PP treatment was dose-dependent, similar to that of ${\mathrm{E}}_2$ treatment ($1\mathrm{\mu }\mathrm{M}$ PP vs. $5\mathrm{\mu }\mathrm{M}$ PP, $p<0.0001$) (Figure 1C). We also observed a dose-dependent increase in nuclear $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity in MCF-7 with treatment of ${\mathrm{E}}_2$ ($10–100\text{\hspace{0.17em} nM}$), BP-3 ($1–30\mathrm{\mu }\mathrm{M}$) and PP ($1–30\mathrm{\mu }\mathrm{M}$) (Figure S1). We confirmed the DNA damage with immunostaining of 53BP1 (P53-binding protein 1), a DNA damage response factor, which localizes to the sites of DNA damage and forms ionization radiation–induced foci. Similar to $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity, we observed dose-dependent increases in 53BP1 nuclear intensity following treatment with ${\mathrm{E}}_2$ ($10–100\text{\hspace{0.17em} nM}$) and PP ($1–5\mathrm{\mu }\mathrm{M}$) incomparison with control in both T47D and MCF-7. BP-3 treatment ($1–5\mathrm{\mu }\mathrm{M}$) showed increased nuclear 53BP1 intensity over control in both T47D and MCF-7, but only MCF-7 showed dose-dependent increase (Figure 1D and E).

Figure 1.

Evaluation of DNA damage in cells treated with $17\mathrm{\beta }\text{-Estradiol}$ (${\mathrm{E}}_2$), Benzophenone-3 (BP-3), or Propylparaben (PP) for $24\mathrm{h}$. Immunofluorescence (upper panel) and quantification (lower panel) of nuclear $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity in T47D cells treated with (A) 10 or $100\text{nM}$ ${\mathrm{E}}_2$, (B) 1 or $5\mathrm{\mu }\mathrm{M}$ BP-3, and (C) 1 or $5\mathrm{\mu }\mathrm{M}$ PP. (D) Immunofluorescence of 53BP1 staining with 10 or $100\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$, 1 or $5\mathrm{\mu }\mathrm{M}$ BP-3, and 1 or $5\mathrm{\mu }\mathrm{M}$ PP in T47D (upper panel) and MCF-7 (lower panel). (E) Quantification of nuclear 53BP1 of treatments in (D) in T47D (left panel) and MCF-7 (right panel). ***$p<0.0001$, *$p<0.01$ compared with control with treatments using one-way analysis of variance (ANOVA) followed by Tukey's honestly significant difference (HSD) multiple-range test. $n=3$ biological replicates. Scale $\text{bar}=50\mathrm{\mu }\mathrm{M}$ (A–C), $10\mathrm{\mu }\mathrm{M}$ (D). All graphs show mean ± SEM.

The effect of these compounds on $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ was contrasted with the mRNA expression of estrogen-responsive gene TFF1. Treatment with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$ stimulated a 13.1-fold increase in expression of the estrogen-responsive gene TFF1, whereas responses to $5\mathrm{\mu }\mathrm{M}$ BP-3 or PP did not differ significantly from the control (Figure 2A). The transcriptional responses to ${\mathrm{E}}_2$ were blocked by treatment with fulvestrant (ICI 182780, $1\mathrm{\mu }\mathrm{M}$), demonstrating the dependence on ER. Blocking ER with fulvestrant also significantly reduced the effect of ${\mathrm{E}}_2$ on $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity (Figure 2B, $p<0.0001$) and inhibited $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity in response to $5\mathrm{\mu }\mathrm{M}$ BP-3 ($p<0.0001$), suggesting that the induction of DNA damage was, in part, dependent on ER. However, the $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ foci induced by ${\mathrm{E}}_2$ and BP-3 was incompletely blocked by fulvestrant in comparison with its inhibition of TFF1 expression.

Figure 2.

TFF1 expression and $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity in T47D cells treated with $17\mathrm{\beta }\text{-estradiol}$ (${\mathrm{E}}_2$), benzophenone-3 (BP-3), or propylparaben (PP) for $24\mathrm{h}$ with or without the ER antagonist fulvestrant. (A) Inhibition of TFF1 expression following treatment of $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$, $5\mathrm{\mu }\mathrm{M}$ BP-3, and $5\mathrm{\mu }\mathrm{M}$ PP when cotreated with fulvestrant (ICI 182 780, $1\mathrm{\mu }\mathrm{M}$) compared with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$, $5\mathrm{\mu }\mathrm{M}$ BP-3, and $5\mathrm{\mu }\mathrm{M}$ PP treatments without fulvestrant. (B) Quantification of nuclear $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ following cotreatment of fulvestrant ($1\mathrm{\mu }\mathrm{M}$) with ${\mathrm{E}}_2$ ($10\text{\hspace{0.17em} nM}$), BP-3 ($5\mathrm{\mu }\mathrm{M}$), or PP ($5\mathrm{\mu }\mathrm{M}$) compared with ${\mathrm{E}}_2$ ($10\text{\hspace{0.17em} nM}$), BP-3 ($5\mathrm{\mu }\mathrm{M}$), or PP ($5\mathrm{\mu }\mathrm{M}$) without fulvestrant treatment, respectively. ***$p<0.0001$ compared control to xenoestrogens treatment and ^###$p<0.001$ compared with negative fulvestrant and with positive fulvestrant using multiple comparison for 2-way ANOVA. $n=3$ biological replicates. All graphs show mean ± SEM.

Estrogenic Response in Cells Treated with BP-3 and PP

Reporter assays provide a sensitive means to evaluate estrogenic activity on a minimal promoter, whereas endogenous genes containing estrogen-responsive elements provide physiologically relevant targets. To saturate ER responses in these assays, $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$ is sufficient; hence, it was used as positive control that is relevant to physiologic ${\mathrm{E}}_2$ levels ($2–70\text{\hspace{0.17em} nM}$) in women (Table 1). T47D-KBluc cells harbor an integrated ERE-luciferase reporter in which BP-3 showed a lowest-observed-effect at $5\mathrm{\mu }\mathrm{M}$ with transactivation increasing to a maximum 37% relative transactivation activity (RTA) in comparison with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$ (Figure 3A). In contrast, PP showed 4.7% RTA at $0.5\mathrm{\mu }\mathrm{M}$ and increased to 288% at $50\mathrm{\mu }\mathrm{M}$. To estimate the transactivation activity of the compounds at levels that are relevant to human exposure, we used the published urinary levels of BP-3 and PP (Table 1). At concentrations measured in the 95th percentile of pregnant women, BP-3 had $27.16\pm 6.2\%$, and PP had $104.07\pm 20.98\%$ RTA (Figure S2, white and black arrows, respectively). Expression of endogenous ER target genes AREG and PGR were also quantified in T47D and MCF-7 cell lines (Figure 3B and C). Treatment of BP-3 and PP at $1\mathrm{\mu }\mathrm{M}$ resulted in no significant changes in mRNA expression of AREG and PGR, a concentration that led to significant increases in DNA damage in both T47D and MCF7 cells (Figure 1). Proliferation induced by these compounds was also compared with control treatment to provide an additional measure of their bioactivity (Figure 3D, Table 2). PP stimulated significant proliferation of T47D cells at $10\mathrm{\mu }\mathrm{M}$, but not at $1\mathrm{\mu }\mathrm{M}$ PP. However, BP-3 had marginal effect at 5 or $50\mathrm{\mu }\mathrm{M}$. Low concentrations of BP-3 and PP only marginally increased cell numbers in comparison with control.

Figure 3.

Evaluation of estrogen receptor transactivation and proliferation in cells treated with $17\mathrm{\beta }\text{-estradiol}$ (${\mathrm{E}}_2$), benzophenone-3 (BP-3), or propylparaben (PP) for $24\mathrm{h}$. (A) Transactivation response (in relative light unit, RLU) of T47D-KBluc cells in response to $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$ (green), $0.5–100\mathrm{\mu }\mathrm{M}$ BP-3 (brown), and $0.5–100\mathrm{\mu }\mathrm{M}$ PP (violet) treatment. Expression of endogenous genes AREG (B) and PGR (C) with ${\mathrm{E}}_2$ ($10\text{\hspace{0.17em} nM}$), BP-3 (1 or $5\mathrm{\mu }\mathrm{M}$), or PP (1 or $10\mathrm{\mu }\mathrm{M}$) treatment as relative fold change over control in T47D (left panel) and MCF-7 (right panel). *$p<0.05$ and ***$p<0.0001$ compared control with treatments using one-way analysis of variance (ANOVA) followed by Tukey's honestly significant difference (HSD) multiple-range test. $n=3$ biological replicates. (D) Proliferation of 47D cell as percent of Alamar Blue reduction in response to ${\mathrm{E}}_2$ ($0.5\text{\hspace{0.17em} nM}$), PP (1 or $10\mathrm{\mu }\mathrm{M}$), BP-3 ($5\mathrm{\mu }\mathrm{M}$), or control. The confidence intervals of the slope are reported in Table 2. All graphs show mean ± SEM.

R-Loop Formation in T47D Cells Treated with $E_2$, BP-3, or PP

R-loop formation was investigated as a possible mechanism of DNA damage using the S9.6 antibody to specifically detect DNA:RNA hybrids. Although we observed a basal level of R-loop foci in the vehicle-treated control in T47D cells, nuclear S9.6 foci were significantly increased with $5\mathrm{\mu }\mathrm{M}$ of BP-3 or PP treatment and comparable with responses with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$. Addition of RNase H to the cells treated with $5\mathrm{\mu }\mathrm{M}$ BP-3 or PP or $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$ abolished the S9.6 intensities, confirming the specificity of S9.6 nuclear staining (Figure 4A and B). Similarly, increase of R-loops formation was obtained with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$, $5\mathrm{\mu }\mathrm{M}$ BP-3, or $5\mathrm{\mu }\mathrm{M}$ PP treatment of MCF-7 cells, which was abrogated following RNase H addition post fixation (Figure 4C).

Figure 4.

R-loop formation in T47D and MCF7 cells treated with $17\mathrm{\beta }\text{-estradiol}$ (${\mathrm{E}}_2$), benzophenone-3 (BP-3), or propylparaben (PP) or vehicle with or without RNase H. (A) Immunostaining of R-loop with S9.6 antibody and DAPI in T47D cells treated with ${\mathrm{E}}_2$ ($10\text{\hspace{0.17em} nM}$), BP-3 ($5\mathrm{\mu }\mathrm{M}$) or PP ($5\mathrm{\mu }\mathrm{M}$) without and with RNase H treatment following fixation. Scale $\text{bar}=20\mathrm{\mu }\mathrm{M}$. (B) Quantification of the nuclear S9.6 intensity in T47D. (C) Quantification of nuclear S9.6 intensity in MCF-7. ***$p<0.0001$ compared control with xenoestrogens treatment and ^###$p<0.001$ compared negative RNase H and with positive RNase H using multiple comparison for 2-way ANOVA. $n=3$ biological replicates. All graphs show mean ± SEM.

R-Loop Formation in Normal Breast Epithelial Cell Line Treated with $E_2$, BP-3, or PP

Next, we asked whether R-loops form in normal breast epithelial cells in response to exposures of BP-3 and PP (Figure 5). The 76N-Tert cells do not express endogenous ESR1, providing a null background to test $\text{ER}\mathrm{\alpha }\text{-stimulated}$ R-loops. The cells were stably infected with an inducible human ESR1 (pINDUCER-ESR1) (Figure 5A) $\text{ER}\mathrm{\alpha }$ expression in 76N-Tert-ESR1 was confirmed with Western blot (Figure 5B). MCF-7 cell lysate was used as a positive control. Immunofluorescence showed 90% of the 76N-Tert-ESR1 cell population were GFP-positive ($\text{ER}\mathrm{\alpha }$ expressing) (Figure S3).

Figure 5.

Characterization of 76N-Tert-ESR1 and R-loop formation in 76N-Tert-ESR1 following treatment with $17\mathrm{\beta }\text{-estradiol}$ (${\mathrm{E}}_2$), benzophenone-3 (BP-3) or propylparaben (PP) with and without RNase H. (A) Map of pIN-ESR1 construct ESR1 insertion next to doxycycline(dox) inducible ${\text{TRE}}_2$ promoter. (B) Western blot $\text{ER}\mathrm{\alpha }$ (upper panel) with MCF-7 as positive control (lane 1), 76N-Tert parental (lane 2), 76N-Tert-ESR1 without dox (lane 3), 76N-Tert-ESR1 with dox (lane 4), and 76N-Tert-ESR1 with dox and ${\mathrm{E}}_2$ ($10\text{\hspace{0.17em} nM}$) treatment and $\mathrm{\beta }\text{-actin}$ as loading control (lower panel). (C) Immunostaining with S9.6 antibody and DAPI with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$, $5\mathrm{\mu }\mathrm{M}$ BP-3, or $5\mathrm{\mu }\mathrm{M}$ PP treatment to parental 76N-Tert cells (upper panel), to 76N-Tert-ESR1 with dox induction (middle panel) without or with RNase H treatment (lower panel). Scale $\text{bar}=20\mathrm{\mu }\mathrm{M}$. (D) Quantification of nuclear S9.6 intensity in (C). ***$p<0.0001$ compared control with xenoestrogens treatment and ^###$p<0.001$ compared among 76N-Tert Parental, 76N-Tert with dox and ${\mathrm{E}}_2$ ($10\text{\hspace{0.17em} nM}$) negative RNase H and 76N-Tert with dox and ${\mathrm{E}}_2$ ($10\text{\hspace{0.17em} nM}$) positive RNase H using multiple comparison for 2-way ANOVA. $n=3$ biological replicates. All graphs show mean ± SEM.

In the parental 76N-Tert cell line, which does not express $\text{ER}\mathrm{\alpha }$, treatment with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$, $5\mathrm{\mu }\mathrm{M}$ BP-3, or $5\mathrm{\mu }\mathrm{M}$ PP showed low nuclear S9.6 staining. After induction of $\text{ER}\mathrm{\alpha }$ with doxycycline, $5\mathrm{\mu }\mathrm{M}$ BP-3 or PP increased the number of nuclear S9.6 foci significantly over vehicle-treated control and was comparable with $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$ treatment. RNase H treatment reduced nuclear S9.6 foci in $10\text{\hspace{0.17em} nM}$ ${\mathrm{E}}_2$ treated as well as $5\mathrm{\mu }\mathrm{M}$ BP-3 or PP treated 76N-Tert-ESR1 cell line induced with doxycycline ($p<0.0001$, Figure 5C and D).

Evaluation of R-Loop Formation and DNA Damage in Mice Treated with $E_2$, BP-3, or PP

To evaluate the relevance of exposure to xenoestrogens in vivo, we treated ovariectomized BALB/c mice orally with ${\mathrm{E}}_2$ ($250\mathrm{\mu }\mathrm{g}/\mathrm{kg}/\mathrm{d}$), BP-3 ($\mathrm{3,000}\mathrm{\mu }\mathrm{g}/\mathrm{kg}/\mathrm{d}$) or PP ($\mathrm{10,000}\mathrm{\mu }\mathrm{g}/\mathrm{kg}/\mathrm{d}$) for 4 d (Figure 6A). These doses were used in experiments evaluating effects of chronic exposures on mammary gland development (LaPlante et al. 2018). We observed 3.8-fold higher nuclear S9.6 staining in the mammary epithelium of ${\mathrm{E}}_2$-treated animals over control-treated animals. Exposure to BP-3 also induced 2.5-fold higher nuclear S9.6 staining in the mammary epithelial cells, whereas PP induced 3.8-fold higher in comparison with control-treated animals (Figure 6B and C). Nuclear $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ intensity in the mammary gland of ${\mathrm{E}}_2$- and BP-3-treated animals was significantly higher than animals treated with vehicle control (Figure 6D). Although oral treatment of ${\mathrm{E}}_2$ stimulated proliferation as shown by higher Ki67 straining and transcriptional activation of ER-target genes (Areg and Pgr) in the mammary gland, neither BP-3 nor PP elicited significant responses (Figure 6 E–H). Similarly, elevated serum levels of ${\mathrm{E}}_2$ and uterine weight was observed only in ${\mathrm{E}}_2$-treated mice (Figure S4).

Figure 6.

Acute exposure of xenoestrogens in mice. (A) Schematic of experimental design and exposure period. (B) Immunostaining of mouse mammary epithelium with S9.6 antibody harvested from mice treated with ${\mathrm{E}}_2$, BP-3, or PP. Each image shows a ductal structure with luminal and myo-epithelial cell nucleus (blue) and R-loop (green) inside the nucleus. Scale $\text{bar}=10\mathrm{\mu }\mathrm{M}$. Quantification of the immunostaining data for S9.6 (C) and $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ (D). Expression of Areg (E) and Pgr (F) from mouse mammary gland. $n=3$ biological replicates. Ki67 straining of luminal epithelial cells (G) and percent of Ki67 strained cells per luminal cells counted (H). Scale $\text{bar}=50\mathrm{\mu }\mathrm{M}$ [Number of biological replicates (n): control (5), ${\mathrm{E}}_2$ (8), BP-3 (12), PP (10)] ***$p<0.0001$, **$p<0.01$ compared control with treatments using one-way analysis of variance (ANOVA) followed by Tukey's honestly significant difference (HSD) multiple-range test. All graphs show mean ± SEM.

Discussion

Exposure of xenoestrogens was implicated in breast cancer risk (Pastor-Barriuso et al. 2016) as well as resistance to breast cancer treatment (Goodson et al. 2011; Warth et al. 2018) due to their endocrine actions. The median urinary level of BP-3 was $0.137\mathrm{\mu }\mathrm{M}$, and PP was $0.161\mathrm{\mu }\mathrm{M}$ in nonpregnant women in participating in the NHANES by the CDC (Calafat et al. 2010; Woodruff et al. 2011). The serum levels of BP-3 was reported to be approximately $0.87\mathrm{\mu }\mathrm{M}$ ($200\mathrm{\mu }\mathrm{g}/\mathrm{L}$) following exposure in women (Janjua et al. 2004; Matta et al. 2019; Tarazona et al. 2013). In addition, the urinary concentrations of xenoestrogens observed in pregnant women were higher than in the general population, with median urinary concentrations of BP-3 and PP being $0.47\mathrm{\mu }\mathrm{M}$ and $0.253\mathrm{\mu }\mathrm{M}$, respectively, and the 95th percentile concentrations in pregnant women being $29.5\mathrm{\mu }\mathrm{M}$ BP-3 and $3.26\mathrm{\mu }\mathrm{M}$ PP (Table 1). This finding raises the possibility that women may have higher exposure during pregnancy due to use of creams and lotions or that absorption and metabolism may be altered in pregnancy. These compounds were also found in normal tissues of women undergoing mastectomy for primary breast cancer (Barr et al. 2012; Barr et al. 2018) and in milk collected during the period of sunscreen use from three different cohorts of mothers of singleton children (Schlumpf et al. 2010). However, based on measures of transcriptional activity in MCF7 human breast cancer cell lines (Byford et al. 2002; Kerdivel et al. 2013), typical exposures to BP-3 and PP would appear to pose a minimal risk for breast cancer through ER-mediated transcriptional activation of target genes.

Estrogens and their metabolites have been shown to induce direct DNA damage. However, DNA damage by catechol estrogens from ER-negative cell lines requires concentrations that are 100-fold greater than the average circulating concentrations in women (Cavalieri and Rogan 2016; Savage et al. 2014; Xu et al. 2007). BP-3 and PP were shown to have the potential to cause DNA damage independent of ER transactivation, based on experiments on ER-negative cell lines. For example, treatment of BP-3 ($10\mathrm{\mu }\mathrm{M}$) induced $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ foci in normal human keratin cell lines (Kim et al. 2018) and PP ($50\mathrm{\mu }\mathrm{M}$) showed 8-Hydroxy-2-deoxyguanosine (8-OHdG) release in Vero cells (derived from monkey kidney) (Pérez Martín et al. 2010). However, these levels exceeded typical concentrations measured in human populations (Table 1).

In the breast, epithelial cells with functional $\text{ER}\mathrm{\alpha }$, we observed DNA damage at physiologic concentrations of ${\mathrm{E}}_2$. BP-3 and PP also caused DNA damage at low concentrations ($1–5\mathrm{\mu }\mathrm{M}$) (Figure 1). Both the nuclear $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ and 53BP1 foci were diminished by fulvestrant, suggesting $\text{ER}\mathrm{\alpha }$ dependency of DNA damage. At these low concentrations ($1\mathrm{\mu }\mathrm{M}$ of PP and $1–5\mathrm{\mu }\mathrm{M}$ of BP-3), we did not observe ER-mediated transcriptional response in target genes. Instead, we observed R-loop formation. We also observed increases in R-loops and $\mathrm{\gamma }\text{-}\mathrm{H}2\text{AX}$ in the mammary epithelial cells of ovariectomized BALB/c mice orally treated with BP-3 or PP at doses designed to model environmental exposures in humans (Figure 6D). The doses of BP-3 and PP used in mice were not sufficient to affect transcription of Areg or Pgr (Figure 6E–F) or proliferation of mammary epithelium (Figure 6G–H) in comparison with control treatments, nor were they sufficient to alter uterine weights in comparison with the control treatment in ovariectomized mice (Figure S4). This finding for BP-3 was supported by a previous study (LaPlante et al. 2018). These results with BP-3 and PP are consistent with the idea that in mammary epithelial cells of human and mice, the formation of R-loops and DNA damage is ER-dependent but is separable from gene transcription and proliferative responses.

ER-mediated DNA double-strand breaks was shown to form by collision of R-loop formed during active transcription (cotranscriptional R-loop) and replication fork in MCF7 cells (Stork et al. 2016). Alternatively, R-loop formation can occur with RNA Polymerase II pausing, which results in no increase of gene expression but leads to DNA damage (Hatchi et al. 2015; Shivji et al. 2018; Zhang et al. 2017). Indeed, our results showed that, BP-3 and PP induced formation of R-loops and DNA damage (Figure 7) but did not lead to detectable increases in full-length transcripts of TFF1, AREG, or PGR.

Figure 7.

A schematic model for ER-dependent DNA damage. ${\mathrm{E}}_2$ or xenoestrogens binding to the ER recruit ER to the estrogen response element (ERE) in the promoter and forms R-loop. Persistence of R-loop in the promoter introduces DNA damage.

Experiments performed using a normal breast epithelial cell line 76N-Tert expressing inducible $\text{ER}\mathrm{\alpha }$ treated with ${\mathrm{E}}_2$, BP-3, and PP provided a) additional evidence that the R-loop formation and DNA damage were $\mathrm{E}\mathrm{R}\mathrm{\alpha }\text{-dependent}$ and b) that normal breast epithelial cells were susceptible to DNA damage by xenoestrogens. This finding raises the possibility that a subset of women bearing variants of R-loop processing factors may be particularly susceptible to the genotoxic effects of xenoestrogens such as BP-3 and PP. More than 300 R-loop binding proteins have been identified (Wang et al. 2018). A number of such factors were recently shown to be involved in the resolution of R-loops to limit DNA damage, including TopI (Tuduri et al. 2009), BRCA1 (Hatchi et al. 2015), BRCA2 (Shivji et al. 2018), SETX (Cohen et al. 2018; Hatchi et al. 2015), Aquaris (Sollier et al. 2014), THO/THREX complex (Bhatia et al. 2014; Gómez‐González et al. 2011), BuGZ, and Bub (Wan et al. 2015). For example, recruitment of BRCA1/SETX was important for R-loop mediated transcriptional termination. As a consequence, the mutational rate of termination regions where BRCA/SETX colocalize was higher in BRCA1-deficient tumors in comparison with BRCA1-WT tumors (Hatchi et al. 2015). Premalignant breast lesions such as atypical hyperplasia expressed higher levels of $\text{ER}\mathrm{\alpha }$ (Gregory et al. 2019) and thus may be especially sensitive to the genotoxic effects of these xenoestrogens. Therefore, limiting exposure to personal care products and foods containing these chemicals may be valuable for this subset of women.

However, the present studies do not show a direct risk of exposure to these compounds on subsequent breast cancer. Although chronic exposure to low levels of DNA damage have the potential to induce mutations that either initiate or promote carcinogenesis, the experiments were not designed to demonstrate a causal effect of BP-3 or PP on mammary tumors or breast carcinogenesis. The DNA damage observed was associated with the formation of $\mathrm{E}\mathrm{R}\mathrm{\alpha }\text{-dependent}$ R-loops, but it is unclear whether $\text{ER}\mathrm{\beta }$ also contributes to the formation of R-loops or may mitigate this. Although many tissues express ERs, they vary in the levels of $\text{ER}\mathrm{\alpha }$ and $\text{ER}\mathrm{\beta }$ as well as expression of DNA repair factors and proficiency of resolving R-loops. Therefore, this mechanism of DNA damage may be limited to the breast epithelium of a subset of individuals. Also unclear is how combinations of environmental xenoestrogens may interact to augment or dissipate the genotoxicity through competing actions on ER. Nonetheless, the data presented here reveal a need to consider the unique potential for genotoxicity of environmental xenoestrogens in tissues expressing ERs.

These studies demonstrated that xenoestrogens possessed the potential for genotoxic activity that was mediated by $\text{ER}\mathrm{\alpha }$ through the formation of R-loops and DNA double-strand breaks. These genotoxic effects were observed at concentrations well below those necessary for detectable transcriptional activation. Therefore, R-loop forming capacity provides a valuable end point to consider when evaluating the safety and activity of environmental chemicals. The inducible expression of $\text{ER}\mathrm{\alpha }$ in normal breast cells provides a tool with which to quantify the variation in sensitivity to these compounds among individuals and to determine whether a subset of individuals is preferentially susceptible to the genotoxic activities.