Bisphenol BPAF and BPC are agonists for estrogen receptor ERα but antagonists for N-terminal domain-lacking ERα Xiaohui Liu, Miki Shimohigashi, Yasuyuki Shimohigashi ABSTRACT In the DNA complex of coactivator-bound estrogen receptor alpha (ERα), the N-terminal domain (NTD) and C-terminal ligand-binding domain (LBD) interact specifically. ERα-NTD and LBD constitute activation function 1 (AF-1) and activation function 2 (AF-2), respectively. We recently revealed that, despite the complete inactivity of ERα-NTD(AF-1), desNTD(AF-1)-ERα exhibited approximately 65% of the activity of the natural estrogen 17β-estradiol (E2) for wild-type full-length ERα. It remains unclear how a deficiency of NTD(AF-1) influences the activity of desNTD(AF-1)-ERα especially with respect to other estrogens and xenoestrogens. The major objective of this study is to evaluate the ligand specificity of desNTD(AF-1)-ERα for a series of xenoestrogens, including bisphenol A (BPA), BPAF, and BPC, together with E2, and the antagonists 4-hydroxytamoxifen (4-OHT) and ICI 182,780 (ICI). The receptors were transiently expressed in HeLa cells, and receptor activation activity was evaluated by luciferase reporter gene assay. Antagonist activity was examined for ERα and desNTD(AF-1)-ERα using E2 as a reference agonist. E2 exhibited full agonist activity for both ERα and desNTD(AF-1)-ERα, whereas 4-OHT and ICI were completely inactive and exhibited antagonist activity for E2 in both ERα and desNTD(AF-1)-ERα. All bisphenols were active for full-length ERα. Surprisingly, however, BPAF and BPC were almost completely inactive for desNTD(AF-1)-ERα, whereas BPA was fully active. BPAF and BPC exhibited distinct antagonist activity for E2 in desNTD(AF-1)-ERα, with pA2 values of 7.62 and 7.86, respectively. The present results revealed that the presence of the N-terminal NTD(AF-1) domain substantiates the agonist activity of halogen-containing BPAF and BPC in wild-type full-length ERα. ERα-NTD(AF-1) plays an essential role in determining the agonist/antagonist activity of BPAF and BPC for estrogen receptor ERα. Introduction When the quaternary structure of the DNA complex of coactivator-bound estrogen receptor alpha (ERα) was elucidated using cryo-electron microscopy, the adjacency of the N-terminal domain (NTD) and C-terminal ligand-binding domain (LBD) was observed in each ERα protein molecule ( Fig 1A ) [ 1 ]. ERα-NTD and ERα-LBD constitute activation function 1 (AF-1) and activation function 2 (AF-2), respectively, which are essential for transcription activation of wild-type full-length ERα [ 2 , 3 ] ( Fig 1A and 1B ). This spatial proximity of AF-1 and AF-2 was observed for the first time and judged to be essential for ERα to recruit specific steroid receptor coactivator (SRC) proteins [ 1 , 4 ]. Because NTD and LBD include AF-1 and AF-2, respectively, we refer to them as NTD(AF-1) and LBD(AF-2) henceforth. We term the spatial proximity or interaction between NTD(AF-1) and LBD(AF-2) the N/C-intramolecular interaction. 10.1371/journal.pone.0350499.g001 Fig 1 Structure of estrogen receptor ERα, desNTD(AF-1)-ERα, and their ligands. ( A ) Schematic of the structure of the ERα-coactivator (SRC-3, p300)-DNA complex. ( B ) Schematic bar representation of domain structures of full-length ERα, ERα-NTD(AF-1), and desNTD(AF-1)-ERα. ( C ) Structures of a series of estrogen receptor ligands: a . E2 (17β-estradiol); b. 4-OHT (4-hydroxytamoxifen); c. ICI (ICI 182,780); d . BPA (bisphenol A); e. BPAF (bisphenol AF); and f . BPC (bisphenol C). ERα, a member of the nuclear receptor (NR) superfamily, regulates a number of physiological processes, including reproduction, development, embryogenesis, and metabolism, and is activated by the sex hormone 17β-estradiol (E2) [ 5 – 8 ]. ERα consists of 595 amino acid residues and, like other NRs, has a domain structure comprising an N-terminal NTD(AF-1) domain, a DNA-binding domain (DBD), and a C-terminal LBD(AF-2) domain [ 9 , 10 ]. In wild-type ERα, these domains span residues 1–180 for NTD(AF-1), 181–263 for DBD, and 303–595 for LBD(AF-2), which includes the F-domain ( Fig 1B ). A flexible hinge region (residues 264–302) connects the DBD and LBD(AF-2). These domains are designated alphabetically as A/B, C, D, E, and F from the N-terminus to the C-terminus ( Fig 1B ). E2 binds specifically to a site within the ligand-binding pocket (LBP) of the dimeric LBD(AF-2) [ 11 ]. desNTD(AF-1)-ERα (residues 181–595) exhibits surprisingly high residual transcriptional activity. Indeed, this NTD(AF-1)-lacking ERα retains approximately 65% of wild-type ERα activity [ 12 ], suggesting that wild-type full-length ERα and desNTD(AF-1)-ERα recruit different sets of SRC coactivator proteins. The C-terminal LBD(AF-2) domain appears to exist in different states in wild-type ERα and desNTD(AF-1)-ERα. This difference in the state of LBD(AF-2) likely affects the binding of chemical substances to the LBP of ERα-LBD(AF-2). Therefore, it is important to analyze the activity of a series of estrogens and xenoestrogens [ 13 – 15 ] for both full-length ERα and desNTD(AF-1)-ERα. In this study, the presence or absence of N/C-intramolecular interaction is crucial for evaluating the ligand‒receptor interaction. We previously reported that potent endocrine-disrupting chemicals such as bisphenol AF (BPAF) and bisphenol C (BPC) ( Fig 1C ), as well as other bisphenols, function as agonists for ERα but as antagonists for ERβ [ 16 – 19 ]. ERβ possesses an NTD(AF-1) domain that is 36 residues shorter than that of ERα-NTD(AF-1). BPAF and BPC were found to bind to ERα and ERβ considerably strongly, and thus they were with a robust concern for their endocrine-disturbing action in the human body [ 20 ]. Although the amino acid sequences of the NTD(AF-1) domains are dissimilar, the DBD and LBD(AF-2) domains of ERα and ERβ are highly similar. We hypothesized that the ERα-agonist/ERβ-antagonist activity of BPAF and BPC might be due to a difference in the N-terminal NTD(AF-1) domains. This hypothesis prompted us to examine the activity of a series of estrogens and xenoestrogens against desNTD(AF-1)-ERα. In the present study, we evaluated the intrinsic receptor characteristics of ERα and desNTD(AF-1)-ERα using a traditional agonist and antagonists for wild-type ERα: E2, 4-hydroxytamoxifen (4-OHT), and ICI 182,780 (ICI) ( Fig 1C ). We also assayed bisphenol A (BPA), a chemical compound primarily used in the manufacture of various plastics. BPA is a xenoestrogen and endocrine disruptor [ 21 ] that mimics the effects of estrogen in the body. We found that BPA binds strongly to human estrogen-related receptor gamma (ERRγ), but only weakly to ERα [ 22 , 23 ]. We further demonstrated that ERRγ, on of 48 human nuclear receptors, is most abundantly expressed in the human placenta [ 24 ]. In these a few decades, serious concern is the leakage of BPA and its substitutes from chemical products. In particular, a strong apprehension has been raised about crucial endocrine-disrupting effects to human health, especially as the adverse impacts on the development of the reproductive systems and fetal brain. BPAF and BPC bind to ERRγ fairly strongly, and now ERα, ERβ, and ERRγ appear to be the major targets of bisphenols among human endocrine hormone receptors [ 17 , 19 , 25 – 27 ]. In the present study, BPA, BPAF and BPC were tested against ERα and desNTD(AF-1)-ERα. We describe the receptor binding affinity, transcriptional activation activity, and antagonist activity of these estrogens and xenoestrogens, and discuss the molecular mechanisms of receptor responses in full-length ERα and desNTD(AF-1)-ERα. Materials and methods Chemicals 17β-Estradiol (E2; CAS No. 50-28-2; 98.9% purity) was purchased from Research Biochemicals International (Natick, MA, USA). BPA (2,2-bis(4-hydroxyphenyl)propane; CAS No. 80-05-7; >99% purity), BPAF (2,2-bis(4-hydroxyphenyl)hexafluoropropane; CAS No. 1478-61-1; >98% purity), and ICI 182,780 (ICI; 7α,17β-[9-[(4,4,5,5,5-pentafluoropentyl)-sulfinyl]nonyl]estra-1,3,5(10)-triene-3,17-diol; CAS No. 129453-61-8; >98% purity) were purchased from Tokyo Chemical Industry (TCI) Co., Ltd, Tokyo, Japan. BPC (bis(4-hydroxyphenyl)-2,2-dichlorethylene; CAS No. 14868-03-2; 98% purity) and 4-OHT (4-[( Z )-1-{4-[2-(dimethylamino)ethoxy]phenyl}- 2-phenylbut-1-enyl]phenol; CAS No. 68047-06-3; >99.0% purity) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Receptor-binding assays for estrogen receptor ERα Preparation of GST-fused ERα-LBD protein: For receptor-binding assays, the glutathione S -transferase (GST)-fused ERα-LBD (GST-ERα-LBD) was prepared as previously described [ 20 ]. Briefly, a cDNA fragment encoding ERα-LBD (residues 303‒595) was amplified by PCR and cloned into the pGEX-6p-1 vector (GE Healthcare BioSciences, Piscataway, NJ, USA) for GST fusion, and the resulting GST-ERα-LBD construct was expressed in E. coli BL21α. The receptor protein was purified using an affinity column (10 × 100 mm) with Glutathione Sepharose 4B (GE Healthcare BioSciences), followed by gel filtration on a Sephadex G-10 column (15 × 100 mm; GE Healthcare BioSciences). Purity was confirmed by SDS-PAGE using 12.5% polyacrylamide gel and Coomassie brilliant blue staining. Protein concentrations were determined by the Bradford method [ 28 ]. The quality of the purified receptor protein was ensured by a saturation binding assay using [ 3 H]E2 (120.5 Ci/mmol; PerkinElmer Life Sciences, Boston, MA, USA) [ 16 , 20 , 29 ]. Radioligand binding assay for competitive binding: The competitive binding assay to measure the binding affinity of a ligand to the ERα-LBD receptor was performed as previously reported [ 17 , 19 , 20 ]. Briefly, compounds were dissolved in N , N -dimethylsulfoxide (DMSO) and serially diluted in binding buffer (50 mM Tris-HCl [pH 7.4], 1 mM EDTA, 1 mM EGTA, 1 mM sodium vanadate(V), 10% glycerol) containing 2 mg/mL γ-globulin using a half-logarithmic 3.16-fold dilution method, maintaining the DMSO concentration below 0.3%. γ-Globulin, but not bovine serum albumin, was used to block nonspecific adsorption to the plasticware. Compounds were evaluated for their ability to inhibit the binding of [ 3 H]E2 (final concentration: 1 nM) to GST-ERα-LBD (60 ng). The assay solutions were incubated at 25°C for 1 h. Bound/free separation was then performed using 100 μl of 0.4% dextran-coated charcoal (Sigma-Aldrich) in PBS (pH 7.4) via direct vacuum filtration using a 96-well filtration plate (MultiScreen HTS HV, 0.45 μm pore size; Millipore, Billerica, MA, USA) for 2‒10 min at 4°C. Radioactivity was determined using a liquid scintillation counter (TopCount NXT; PerkinElmer Life Sciences Japan, Tokyo, Japan). To estimate binding affinity, IC 50 (half-maximal inhibitory concentration) values were calculated from dose–response curves generated using GraphPad Prism 10 (GraphPad Software, Inc., Boston, MA, USA). Each assay was performed in duplicate and repeated at least three times. In vitro biological assays for ERα and desNTD(AF-1)-ERα Plasmid construction for ERα and desNTD(AF-1)-ERα : The expression plasmid of full-length ERα/pcDNA3.1 was generated by PCR amplification of an ERα cDNA clone (OriGene Technologies, Rockville, MD, USA), followed by subcloning into the pcDNA3.1(+) vector (Invitrogen, Carlsbad, CA, USA) as reported previously [ 16 , 17 ]. Using the cDNA clone of full-length ERα as a template, the expression plasmid desNTD(AF-1)-ERα/pcDNA3.1 was similarly constructed by PCR followed by subcloning into pcDNA3.1(+). Cell culture and luciferase reporter gene assay: HeLa cells (RCB0007; RIKEN BRC, Tsukuba, Japan) were seeded at a density of 5 × 10 5 cells per well in 6-well microplates and cultured for 24 h at 37°C in a 5% CO 2 atmosphere as reported previously [ 18 , 19 , 25 – 27 ]. Cells were maintained in modified Eagle’s Minimum Essential Medium (Nissui, Tokyo, Japan) supplemented with 10% (v/v) charcoal-treated fetal bovine serum. For transient transfection, three plasmids were co-transfected in a total volume of 2.0 mL medium: (i) the firefly luciferase reporter plasmid pGL3 (Promega, Madison, WI, USA) containing three tandem estrogen response elements (3 × ERE/pGL3; ERE = AGGTCAnnnTGACCT) (1500 ng/well), (ii) an expression plasmid full-length ERα/pcDNA3.1 or desNTD(AF-1)-ERα/pcDNA3.1 (500 ng/well), and (iii) the internal control plasmid pGL4.74 (Promega) containing the Renilla luciferase gene (100 ng/well) [ 30 ]. Transfection was performed using Lipofectamine TM 3000 (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s protocol. Following transfection, cells were cultured for an additional 24 h under the same conditions. For the luciferase reporter gene assay, cells were harvested using 0.05% trypsin/0.02% EDTA solution and plated onto 96-well microplates at a density of 5 × 10 4 cells/well. Following compound administration at various concentrations, the cells were incubated for 24 h. Firefly/ Renilla luciferase activities were measured and normalized using the Dual-Luciferase Reporter Assay System (Promega), and luminescence was measured on a Fluoroskan FL microplate reader (Thermo Fisher Scientific). Each assay was performed in duplicate and repeated independently at least three times. EC 50 values were estimated from dose–response curves generated using GraphPad Prism 10 (GraphPad Software). Luciferase reporter gene assay for transcription inhibitory activity Reporter gene assays for measuring antagonist activity : For compounds with no transcriptional activity, antagonist activity was initially assessed qualitatively using in vitro luciferase reporter gene assay [ 17 , 19 ]. A serial dilution of each compound (10 −13 to 10 −5 M final concentration) was tested for full-length ERα or desNTD(AF-1)-ERα activity in HeLa cells in the presence of 10 nM E2. This E2 concentration elicits near-maximal activation (approximately 95%) of both full-length ERα and desNTD(AF-1)-ERα. Each assay was performed as described above for transcription activation activity. Apparent IC 50 values were estimated from dose–response curves generated by GraphPad Prism 10 (GraphPad Software) when activity decreased in a dose–dependent manner. Schild plot analysis for antagonist activity : For compounds identified as antagonists, inhibitory activity was quantified using Schild plot analysis [ 31 , 32 ]. Seven antagonist concentrations (0.01–10 μM) were prepared via half-logarithmic (3.16-fold) serial dilutions, and a serial dilution of E2 (10 −13 to 10 −5 M final concentration) was assayed in the presence of each antagonist concentration. The assay was performed twice in duplicate, and data were analyzed from each dose–response curve generated using GraphPad Prism 10 (GraphPad Software). Using the obtained EC 50 values, the pA 2 value, an indicator of antagonist affinity for a receptor, was calculated for each antagonist using Schild plot analysis [ 32 , 33 ]. Analysis of X-ray crystal structure of ERα-LBD in complex with compounds Molecular structure information on experimentally determined three-dimensional (3D) structures was downloaded from the RCSB Protein Data Bank (PDB) ( https://www.rcsb.org/ ). All molecular structure studies were carried out using BIOVIA Discovery Studio Visualizer molecular modeling software (Dassault Systèmes, Vélizy-Villacoublay, Paris, France). Statistical analysis All assays were performed at least three times (n ≥ 3) in each independent experiment, with each sample concentration in duplicate. Data analyzed by the software GraphPad Prism 10 are presented as the mean ± standard deviation (SD) for the indicated number of independent experiments. Statistical significance was determined using a two-tailed Student’s t -test with p < 0.001 considered significant. Results Receptor binding affinity of a series of estrogens and xenoestrogens Because wild-type full-length ERα and N-terminal NTD(AF-1)-lacking desNTD(AF-1)-ERα share the same ligand-binding domain LBD(AF-2), all ligands would be expected to bind to the ligand-binding pocket (LBP) with the same binding affinity for both full-length ERα and desNTD(AF-1)-ERα. The receptor protein used was GST-ERα-LBD, which possesses a fully open binding pocket. The receptor-binding affinities are shown in Table 1 . The natural estrogen agonist E2 exhibited the highest affinity (IC 50 = 0.74 nM) to ERα-LBD, and the synthetic antagonists 4-OHT [ 34 ] and ICI [ 35 ] showed similar binding affinities (1‒3 nM). Among the bisphenols tested, BPC showed the highest affinity (3.02 nM), followed by BPAF and BPA. 10.1371/journal.pone.0350499.t001 Table 1 Receptor-binding affinity of estrogens and xenoestrogens for ERα. Compounds Binding affinity a IC 50 (nM) b Relative affinity E2 (17β-estradiol) 0.74 ± 0.13 100 4-OHT (4-hydroxytamoxifen) 1.28 ± 0.22 58 ICI (ICI 182,780) 2.96 ± 0.32 25 BPA (bisphenol A) 1060 ± 78.8 0.07 BPAF (bisphenol AF) 49.6 ± 5.20 1.5 BPC (bisphenol C) 3.02 ± 0.22 24 a The receptor used for the competitive binding assay was GST-ERα-LBD. b The data of half-maximal inhibitory concentration (IC 50 ) are presented as the mean ± standard deviation (SD) assessed from at least three independent experiments (n ≥ 3). In vitro biological activity for ERα and desNTD(AF-1)-ERα A series of compounds, including E2, 4-OHT, and ICI, were evaluated for their activity on wild-type full-length ERα and N-terminal NTD(AF-1)-lacking desNTD(AF-1)-ERα using a luciferase reporter gene assay. As expected, E2 exhibited strong full transcriptional activation activity (EC 50 = 0.63 nM) for full-length ERα ( Fig 2A ) ( Table 2 ). All of the tested bisphenols exhibited full activation activity, as shown in Fig 2A . The order of activity was BPA (EC 50 = 1,140 nM)