Rapid estrogenic signaling activities of the modified (chlorinated, sulfonated, and glucuronidated) endocrine disruptor bisphenol A

Abstract

An automated in-vitro, medium-throughput screening method was designed to compare the rapid actions of 10⁻⁹M estradiol to those of modified bisphenol A (BPA) compounds: chlorinated (created due to waste water treatment) and phase II BPA metabolites. GH₃/B₆/F₁₀ prolactinoma cells were treated with increasing concentrations (10⁻¹⁵‒10⁻⁷M) of BPA or modified BPA compounds for 5 min, and phospho-activated kinases assayed via a fixed-cell plate immunoassay. Mono- and di-chlorinated BPAs phospho-activated the extracellular signal-regulated kinase (ERK), while tri-chlorinated BPA dephosphorylated ERK to a level below vehicle-treated controls. When c-Jun-N-terminal kinase (JNK) was examined, mono-chlorinated compounds caused no responses (similar to unmodified BPA), while di- and tri-chlorinated BPAs caused extensive dephosphorylation. When deconjugation of di-sulfated and glucuronidated metabolites were inhibited, these stably conjugated compounds were unable to activate either ERK or JNK, but could inactivate them. This suggests that the modified versions of BPA (extensively chlorinated or conjugated metabolites) may alter the ability of membrane estrogen receptor-α (which mediates these rapid signaling responses) to partner with other signaling proteins. While other studies have examined the estrogenicity of modified (chlorinated and conjugated) BPA compounds in gene expression assays, this is the first to examine their actions via rapid, membrane-initiated signaling pathways.

Keywords: :

• non-monotonic
• polychlorinated
• non-genomic
• estradiol
• membrane estrogen receptor
• phosphorylated

Introduction

Bisphenol A (BPA) is a well-known and environmentally pervasive endocrine-disrupting chemical (EDC) used commercially for the production of polycarbonate plastic water and food containers, as a monomer in the formation of anti-corrosive epoxy inner linings of food cans, and as an ingredient in coatings for thermal paper cash register receipts and faxes.^1-4 BPA had long been considered a weak estrogenic compound compared with estradiol (E₂) due to its low affinity (1,000-fold difference) for intracellular estrogen receptors (iERs).⁵ However, we determined BPA to be equipotent to E₂ in initiating rapid signaling via membrane estrogen receptors (mERs)^6-8 and, thus, capable of altering associated functional endpoints at low, environmentally relevant concentrations. However, it is unknown if common environmental derivatives of BPA (chlorinated versions) and endogenous phase II metabolites of BPA (glucuronidated and sulfated versions) (see Fig. 1) can act via these membrane-initiated signaling pathways.

Figure 1. Structure of BPA, its variably chlorinated versions, and BPA phase II metabolites. (A) Bisphenol A; (B) Mono-, (C) Di-, and (D) Tri-chlorobisphenol A; (E) Bisphenol A β-d-glucuronide; (F) Bisphenol A disulfate.

Typical human exposures to BPA occur from consumption of food and drink that have come into contact with packaging containing BPA, especially during the heating of polycarbonate plastic containers.⁹ Levels of urinary BPA range from 0.4–149 μg/L (1.8–660 nM) in 92.6% of U.S. residents ≥ 6 y of age.¹⁰ Several in vivo rodent studies have linked BPA to developmental alterations in reproductive organs,^11-13 as well as changes in hormonal signaling.^14,15 The presence of BPA has also been associated with the development of chronic diseases such as diabetes,¹⁶ asthma,¹⁷ ovarian dysfunction, and the onset of obesity.¹⁸ In July 2012, the FDA banned BPA from use in the production of plastic baby bottles and drinking cups.¹⁹

A major health concern is the presence of BPA in drinking water. Release of BPA into aquatic environments, particularly to waste water treatment plants, occurs primarily from plastic-to-water migrations such as when water runs through poly-vinyl chloride (PVC) pipes in water supply systems, or from discarded plastic in the environment.²⁰ Once in the drinking water, BPA is rapidly biodegraded by microorganisms (e.g., Pseudomonas), with a half-life of 2.5–4 d.^21-24 However, because of BPA’s presence in many forms of consumer products and refuse, it is rapidly and repeatedly replaced.

The addition of chlorine to water supply systems as a means of preventing water borne diseases, along with BPA’s constant replenishment from the environment, can result in its reaction with free chlorine, giving rise to a variety of chlorinated forms (Fig. 1, left panels).^25-27 Depending on the amount of chlorine present in an aqueous medium, it can rapidly transform BPA within 4 h, producing variably chlorinated forms with half-lives of 10‒20 h (in a controlled environment).²⁸ While an abundance of evidence does exist about other poly-chlorinated biphenols having endocrine-disrupting capabilities, little is known about the estrogenic potential of poly-chlorinated BPA compounds, and nothing about their influence via rapid estrogenic signaling pathways. Our present study therefore uses the well-differentiated GH₃/B₆/F₁₀ rat prolactinoma cell line as our in-vitro model to examine the ability of progressively chlorinated versions of BPA to initiate rapid kinase signaling.

In addition, we also assessed the ability of phase II glucuronidated and sulfated metabolites of BPA (Fig. 1, right panels) to induce membrane-initiated signaling. While most estrogenic metabolism is mediated via cytochrome P450 enzymes (e.g., E₂ to estriol), estrogens also undergo phase II conjugations, which increase their water solubility and facilitate their excretion.^29,30 These conjugations prevent endogenous estrogens³⁰ and exogenous estrogens like BPA^31-33 from binding to intracellular ERs (iERs), thus rendering the compounds biologically inactive in the direct genomic signaling pathway. Previous studies from our group,^34-36 and others,^37,38 have documented that iERα is the same protein as mERα. Localization of mER’s in lipid membranes (as opposed to iERs’ aqueous environment in the cytosol or nucleus) may yield conformational differences in the receptor that alter the specificity of ligand binding sites. This could allow metabolites of BPA modified by conjugation, while unable to enter cells, still able to bind to mERα and cause membrane-initiated signaling.

As in many of our previous studies, we used a fixed cell-based 96-well plate immunoassay to evaluate protein phosphorylation and dephosphorylation of the mitogen-activated protein kinases (MAPKs).^6-8,39-42 While this method has yielded excellent results, it can be a very time consuming process with multiple reagent incubations and the attendant wash steps. To increase efficiency and reproducibility, we designed an automated medium-throughput screening system using a BIOMEK FX^P workstation for liquid handling.

Results

Concentration-dependent changes in phospho-activation of MAPKs by BPA and its progressively chlorinated forms

BPA caused ERK activation (Fig. 2A) in GH₃/B₆/F₁₀ cells within 5 min, with a non-monotonic response pattern, as expected.^6-8 Exposure to mono- and di-chlorinated compounds caused ERK activation at most concentrations, and at some concentrations the responses were equivalent to that of 10⁻⁹M E₂ (Fig. 2B and C). The dose-responses did not gradually rise and plateau at the highest concentrations, as is seen with classical responses, but instead showed non-monotonic patterns. The tri-chlorinated BPA form did not cause phospho-activation of ERK, but instead led to a dramatic dephosphorylation, to below vehicle control levels (Fig. 2D).

Figure 2. Dose-response analysis of ERK phosphorylation (pERK) in GH₃/B₆/F₁₀ cells after exposure to BPA and its progressively chlorinated versions. The cells were exposed to a range of concentrations (in log increments) of (A) BPA, (B) mono-, (C) di-, and (D) tri-chlorinated BPA. pERK was measured by plate immunoassay at a 5 min exposure time. The widths of the vehicle and E₂ (10⁻⁹M) bars represent the means ± SE (n = 24 over three experiments) * = p < 0.05 when compared with vehicle (V). # = p < 0.05 when compared with 10⁻⁹M E₂. E₂ (10⁻⁹M) is significantly different from vehicle.

Though unmodified (parent) and mono-chlorinated BPAs had no significant effects on JNK activation (Fig. 3A and B, respectively), di- and tri-chlorinated BPAs caused significant dephosphorylation of JNK (Fig. 3C and D). Thus, chlorination to different degrees had different effects, with the more extensively modified compounds actually disrupting basal levels of MAPK phosphorylation.

Figure 3. Dose-response analysis of JNK phospho-activation (pJNK) in response to varying concentrations of BPA and its progressively chlorinated versions. GH₃/B₆/F₁₀ cells were exposed to increasing concentrations (in log increments) of (A) BPA, (B) mono-, (C) di-, and (D) tri-chlorinated BPA. pJNK was measured by plate immunoassay at a 5 min exposure time. The widths of the vehicle and E₂ (10⁻⁹M) bars represent the means ± SE (n = 24 over three experiments) * = p < 0.05 when compared with vehicle (V). # = p < 0.05 when compared with 10⁻⁹M E₂. E₂ (10⁻⁹M) is significantly different from vehicle.

Concentration-dependent changes in phospho-activation of MAPKs by BPA phase II metabolites

Cells were exposed to BPA phase II metabolites ± β-glucuronidase or sulfatase inhibitors. The inhibitors alone had no significant effect on kinase activation levels. In the presence of the inhibitor STX-64, sulfated-BPA was only able to activate ERK at 10⁻⁷M (Fig. 4A). However, in the absence of STX-64, the sulfated compound phospho-activated ERK with a non-monotonic dose-response, suggesting some conversion of the compound to the unconjugated form in these pituitary tumor cells. In the presence of the inhibitor D-glucaric acid-1, 4-lactone, stably glucuronidated BPA did not activate ERK, but instead inactivated ERK to significantly below the baseline (vehicle-treated) pERK level, indicating a different and unexpected activity. In the absence of inhibitor, and likely partial deconjugation, we observed significant ERK activation at the lowest concentration (Fig. 4B), for which we have previously found BPA to be the most active.^6,8,43,44 Variability in the uninhibited activation profiles of these two forms of conjugated compounds may suggest partial deconjugations, as well as differences in the relative rates of bond cleavage by sulfatases vs. glucuronidases.

Figure 4. Dose-response analysis of ERK phospho-activation (pERK) by BPA phase II metabolites. GH₃/B₆/F₁₀ cells were exposed to increasing concentrations (in log increments) of (A) Bisphenol A- disulfate and (B) Bisphenol A β-d-glucuronide. The cells were pre-incubated ± D-glucaric acid-1,4-lactone (20 mM) or STX-64 (10 nM) to inhibit β-glucuronidase and sulfatase, respectively. pERK was measured by plate immunoassay at a 5 min exposure time. Controls for the inhibitor administered alone are shown by a symbol at the single concentration at which it was used. The widths of the vehicle and E₂ (10⁻⁹M) bars represent the means ± SE (n = 24 over three experiments) * = p < 0.05 when compared with vehicle (V). # = p < 0.05 when compared with 10⁻⁹M E₂. E₂ (10⁻⁹M) is significantly different from vehicle.

Both conjugated compounds in either the presence or absence of their respective deconjugation inhibitors were unable to activate JNK signaling above vehicle-treated levels, but instead deactivated JNK at some concentrations in a non-monotonic pattern (Fig. 5A and B). The disulfated compound was consistently far more effective in this deactivation. Thus, phase II enzymatic conjugation of BPA with both sulfate and glucuronic acid allows it to suppress MAPK activation levels, disrupting normal actions via mERs and introducing new ones.

Figure 5. Dose-response analysis of JNK phospho-activation (pJNK) by BPA phase II metabolites. GH₃/B₆/F₁₀ cells were exposed to increasing concentrations (in log increments) of (A) Bisphenol A-disulfate and (B) Bisphenol A β-d-glucuronide. The cells were pre-incubated ± D-glucaric acid-1,4-lactone or STX-64 to inhibit β-glucuronidase and sulfatase, respectively, prior to analysis. Controls for the inhibitor administered alone are shown by a symbol at the single concentration at which it was used . pJNK was measured by plate immunoassay at a 5 min exposure time. The widths of the vehicle and E₂ (10⁻⁹M) bars represent the means ± SE (n = 24 over three experiments) * = p < 0.05 when compared with vehicle (V). # = p < 0.05 when compared with 10⁻⁹M E₂. E₂ (10⁻⁹M) is significantly different from vehicle.

Comparison of automated plate assay results with prior manual assays

In designing an automated program for these assays using the BIOMEK software, many adjustments were necessary to match the automated results as closely as possible to our previous manual results.^{6-8,39, 41,44,45} Liquid-handling parameters were optimized, such as the speed of aspiration and dispensing, and determining the adequate depth of tip immersion into the liquid reservoirs. Such adjustments decreased the amount of unwanted liquid carryover that could ultimately build up and increase the final liquid volume in each well, thereby causing added variability. Time adjustments were necessary, taking into consideration the elapsed time from the moment the probe grippers moved or replaced tip boxes, to the point at which the multi-channel probe picked up pipette tips and then finally dispensed the liquids into the wells of the plates. These adjustments made it possible to meet our target times for exposure to estrogens and to precisely terminate the response by fixation. In general, the results of these optimized automated assays agreed very well with previously published MAPK activation/deactivation measurements for both E₂ and BPA.^{6-8,39,41,44,45}

Discussion

Drinking water systems that rely on the addition of chlorine as an inexpensive disinfectant are also abundant sources of EDCs, including BPA, which is rapidly chlorinated to other forms.²⁸ Our study demonstrates that environmentally relevant concentrations (femtomolar-picomolar) of chlorinated derivatives of BPA can initiate or alter membrane-initiated actions via mERs in prolactinoma cells. Glucuronidated and sulfated phase II conjugates of BPA often depressed MAPK activities, instead of activating them. Our results thus provide further insights into these metabolites’ biological activities, which differed from those of their parent compounds.^30-33 Such actions on MAPKs have the potential to cause inappropriate cellular signaling in humans and other animals, and also to interfere with the actions of natural estrogens (such as E₂ and other physiologic estrogenic compounds). Altering these pathways may lead to health changes at downstream functional endpoints such as cell proliferation, and the release of other hormones like prolactin.^6,8,43,44 Our automated protocol increased assay efficiency and reproducibility, very important advantages when examining the large number of environmental compounds, and their chemically modified forms that are of concern, for their disruptive actions on kinase activities.

Dose-response studies of BPA and its chlorinated conjugates generated curves with atypical shapes (Figs. 2 and 3), now more often recognized as a usual consequence of examining EDCs or endogenous hormones.^6,8,43,44 Such non-monotonic dose relationships⁴⁶ are commonly seen when the range of concentrations examined is extended to femtomolar to picomolar levels in sensitive assays. However, the molecular bases for non-monotonicity are largely unknown, and are still a source of much controversy. They could involve protective downregulation of MAPK activities to eliminate unnecessary pathway stimulation; the presence of multiple receptor subtypes that bind the same ligand, each generating a different response pattern (stimulatory or inhibitory); concentration-dependent receptor downregulation or desensitization; and/or stimulation of multiple pathways from the same receptor where signaling can subsequently be redundant, divergent, and convergent.^46-50

Phospho-activation of ERK and JNK is often associated with opposing functional responses. For instance, ERK signaling promotes cell growth and differentiation by activating pro-survival enzymes⁵¹ and inhibiting apoptotic enzymes.^52-54 Conversely, JNK signaling is often associated with inflammation or the initiation of cell death by activating pro-apoptotic proteins, including Bax, Fas, or caspases.^55-59 Opposing responses of ERK and JNK can be appreciated in our data by comparing MAPK responses to di-chlorinated BPA where ERK is activated above the vehicle-treated level, and JNK is inactivated. Simultaneous activation of ERK and deactivation of JNK by EDCs have previously been correlated with severe stimulatory effects on downstream functional endpoints such as cell proliferation.⁶⁰ In contrast, tri-chlorinated BPA suppressed the activities of both MAPKs. Suppression of multiple kinases has been seen with another polychlorinated compound, 2, 3, 7, 8-tetrachlorodibenzo-p-dioxin (TCDD),^61,62 where 0.03 and 0.3 nM treatments rapidly (15 min) inhibited phospho-activation of ERK and JNK in CH12.LX B lymphocytes, correlating with decreased cell viability and “impaired plasmacytic differentiation.” Dual suppression of ERK and JNK by tri-chlorinated BPA suggests that it might impair the actions of a common target upstream of ERK and JNK, and probably led to the cancellation of many downstream actions for which these kinases are responsible, dramatically disrupting endocrine function. Therefore, final functional outcomes may depend on the overall balance between ERK- and JNK-related activities.^63,64

Early structure-activity studies noted that the 17β-hydroxyl group of E₂ was necessary for binding within the ligand-binding pocket of iER.^65,66 Therefore, any chemical mimicking this feature of E₂, such as in BPA, has the possibility of binding to an ER and inducing an estrogenic response.⁶⁷ The addition of multiple chlorine atoms at or near these sites might then disrupt this interaction and block or alter downstream responses. Such a trend was shown using a green fluorescent protein expression reporter system to assess nuclear ER actions;^68,69 when the number of chlorine substituents added to BPA or nonylphenol increased, their estrogenic activity decreased. Increasing brominated substitutions on BPA also cause attenuation in estrogenic cell proliferation.⁷⁰

However, our observations are not consistent with merely blocking of ligand binding to the receptor, because these modified BPAs do have additional effects in some cases (e.g., inactivation of MAPKs). Phase II conjugation of both endogenous and exogenous chemicals changes them from hydrophobic to hydrophilic molecules, which while facilitating their elimination from the body, also likely changes their ability to bind certain ER conformations or to alter the receptor shape once bound. The addition of chlorine, sulfate, or glucuronide moieties surrounding the phenolic-hydroxyl groups of BPA could direct different conformations of the ER with resulting changes in the ability to partner with other proteins. In such cases, interacting phosphatases affecting ERK and JNK could be activated instead.^71-74

We and others have explored the protein identities of mERα and iERα.^34,37 A close protein similarity was established when nine iERα-specific antibodies (Abs) recognized (epitopes on) mERα,^34,75 the iERα-specific Ab H151 elicited or blocked membrane-initiated responses to estrogenic ligands,^35,75 and disrupting the ERα mRNA (via antisense or siRNA methods), resulted in decreased expression of both the membrane and intracellular versions of ERα.^76,77 Phase II conjugates failed to activate rapid kinase responses in our study, indicating another similarity between mER and iER.^29,30 However, their ability to inactivate kinases suggests action via an alternative signaling partner. Even though iERs and mERs are modified forms of the same protein, they may differ in conformation, due to their post-translational modifications⁷⁸ and their different chemical environments (soluble vs. membrane), and so may accommodate a wider variety of estrogen-like molecules⁷⁹ and different signaling partner proteins.

These changes seen in the phospho-activation of the kinases are smaller compared with some responses, for instance, to epidermal growth factor. However biological responses modified in such small but significant increments, and are no less biologically important than larger changes for initiating functionally consequential activations or de-activations. Such small but significant changes in signaling are known to build by amplification as they progress down a cascade.⁸⁰ Furthermore, multiple small changes can add up to large ones, with MAPKs acting as response summation nodes.

The studies presented here demonstrate that some chlorinated conjugates of BPA can induce rapid membrane-initiated signaling in estrogen-responsive prolactinoma cells at low (femtomolar-picomolar) concentrations, but that more extensively altered forms (tri-chlorinated, glucuronidated, sulfated) inactivate the same kinases. Further structure-activity studies could provide additional insights and identify the endocrine-disruptive potential of any new chemicals or their modifications.⁸¹ With our implementation of the BIOMEK FX^P Laboratory Automation Workstation protocol, we have developed a much more efficient and reliable automated screening tool for testing and identifying potential EDCs and their variants, and facilitating large-scale structure-activity studies.

Materials and Methods

Cell culture conditions and experimental compounds

The well-differentiated GH₃/B₆/F₁₀ rat prolactinoma cell line was used to represent responses of lactotroph anterior pituitary cells, and was selected on the basis of its naturally high expression of mERα.^39,82,83 Cells were routinely sub-cultured with phenol red-free Dulbecco’s modification of Eagle’s medium (DMEM, high glucose; 90013PB) containing 12.5% horse serum (SH30074.02), defined supplemented calf (SH30072.03), and fetal serum (SH30088.03) at 2.5% and 1.5%, respectively, all from Fisher Scientific. Cells of passages 10–20 were used for these experiments.

BPA chlorinated compounds and glucuronidated or sulfated metabolites were kindly provided by the National Institute for Environmental Health Sciences (NIEHS). Chemical purity of these compounds was determined by mass spectrometry and nuclear magnetic resonance and ranged from 98–99.7% as indicated by the data sheets provided. Before use on cells, they were dissolved in ethanol, and then diluted in DMEM containing 1% charcoal-stripped serum.

Automated ERK and JNK phosphorylation assays

A fixed cell-based immunoassay was used to quantify phospho-activation of extracellular-regulated kinase (ERK/pERK) and c-jun N-terminal kinases (JNK/pJNK), which we previously developed and optimized, as detailed in references 39 and 40. These previous studies also showed that ERα is the primary receptor involved in initiating these responses and the very rapid time course of a 5 min activation peak.^{7,34,39,84-86} This assay automation is simply a mechanical adaption of our previous method, and no changes in the assay parameters were introduced intentionally. 10⁻⁹M E₂ was used as our positive control in all experiments. Automation of this assay employed the BIOMEK FX^P Laboratory Automation Workstation operated by Biomek Software (Biomek Coulter). Our automated assay was programmed to consist of six phases: (1) ligand exposure, (2) Ab incubation, (3) washing, (4) color signal production during the signal amplification incubation, (5) plate reading of signal, and (6) washing to prepare for the subsequent cell staining and extraction steps to estimate cell number. These phases are detailed in the Results section. An example of the “exposure” phase can be viewed in the following link http://www.youtube.com/watch?v=6Jb7xBjWTtA.

The software code used to automate these protocol phases are protected by copyright (2012).

Progressively chlorinated BPA

Cells (10⁴/well) were plated in 96-well plates (Corning Inc.) and allowed to attach for 24 h. The cells were then incubated with DMEM containing 1% charcoal-stripped (4X) serum for 48 h to deprive them of serum hormones. The medium was then removed and cells were exposed to increasing concentrations (10⁻¹⁵‒10⁻⁷M) of individual mono-, di-, and tri- chlorinated BPAs for 5 min. The short time point of 5 min was chosen to observe membrane-initiated phospho-activations without interference from later genomic responses. The concentration range was chosen to bracket environmentally relevant concentrations and test very low concentrations that we have shown to activate these responses. Test compounds were dissolved in ethanol and then diluted in DMEM containing 1% charcoal-stripped serum. The vehicle control (V) was 0.001% ethanol in DMEM. To stop mER-initiated signaling, cells were fixed with a 2% paraformaldehyde/0.2% picric acid solution (Ricca Chemical Company; 5860-32) and incubated at 4 °C for 48 h. The cells were then incubated in phosphate-buffered saline (PBS) containing 0.2% fish gelatin (G7765) and 0.1% Triton X-100 (9002931) from Sigma-Aldrich, for 1 h at room temperature (RT), followed by overnight incubation at 4 °C with primary Abs against pERK (4370S) or pJNK (4668S) from Cell Signaling Technology (1:500 in PBS/0.2% fish gelatin/0.1% TritonX-100). The cells were then washed with PBS (3X) before incubation with biotin-conjugated secondary Ab (Vector Labs; BA 2001) (1:500 in PBS/0.2% fish gelatin) for 1 h at RT, then washed again in PBS (3X). The preparation was then incubated with Vectastain ABC-AP solution (50 μL/well) from a kit (Vector Labs; AK5000) for 1 h at RT, followed by addition of 50 μL/well alkaline phosphatase substrate (pNpp solution, Thermo Scientific; 34045). The plates were incubated in the dark for 30 min at 37 °C and the A₄₀₅ signal due to the para-nitrophenol (pNp) product was measured. The pNp signal was then normalized to the cell number, estimated by the crystal violet (CV) assay⁴⁰ and measured at A₅₉₀. Absorbance signals for both assays were read in a model 1420 Wallac microplate reader (Perkin Elmer).

Phase II BPA metabolites

The anterior pituitary is known to contain β-glucuronidase⁸⁷ and sulfatases,^29,88 which can cause enzymatic hydrolysis and deconjugate glucuronic acid and sulfate from the parent compound, respectively. To maintain the integrity of our conjugated compounds during the analysis, we pre-incubated our cells with 20 mM D-glucaric acid-1,4-lactone (Carbosynth; MG67426), a potent inhibitor of β-glucuronidase⁸⁹ or 10 nM 667 Coumate/STX-64 (Sigma-Aldrich; S1950), a sulfatase inhibitor,²⁹ each for 1 h at 37 °C (50 µl/well). Conjugates were then added to wells (50µl/well) at final concentrations of 10⁻¹⁵‒10⁻⁷M, incubated for 5 min, and the signaling stopped with 2% paraformaldehyde/0.2% picric acid solution as above. Automated immunoassay methods proceeded as above for the chlorinated compounds.

Statistical analysis

Statistical analysis was performed using Sigmaplot version 12.3 (Systat Software Inc.). One-way analysis of variance (ANOVA) was applied to the dose-dependent and timed studies to assess the statistical significance of mean values produced by varying exposures. A Holm-Sidak comparison against vehicle control or against 10⁻⁹M E₂ treatment was used to evaluate significance. The overall α level selected for the statistical analysis was 0.05.

Abbreviations:
Ab	=	antibody
BPA	=	bisphenol A
JNK	=	c-Jun-N-terminal kinase
CV	=	crystal violet
DMEM	=	Dulbecco’s modified eagle medium
EDC	=	endocrine-disrupting compounds
ERKs	=	extracellular signal regulated kinases
E2	=	estradiol
ER	=	estrogen receptor
iERα	=	intracellular estrogen receptor-α
mERα	=	membrane estrogen receptor-α, MAPKs, mitogen-activated protein kinases
PBS	=	phosphate-buffered saline
pERK	=	phosphorylated ERK
pJNK	=	phosphorylated JNK
RT	=	room temperature

Acknowledgments

These studies were supported by National Institutes of Health grant ES015292, NIH Ruth L. Kirschstein National Research Service Award F31ES021164-01, and the Colgate-Palmolive Award for Student Research Training in Alternative Methods. The parent BPA and all of the conjugates were provided by the NIEHS. This work utilized the University of Texas Medical Branch/Gulf Coast Consortium Core in High Throughput Screening for Chemical Biology, which is supported in part by the John S. Dunn Foundation through the Gulf Coast Consortium for Chemical Genomics. We also thank Dr David Konkel for critically editing the manuscript.

Disclosure of Potential Conflicts of Interest

The authors declare that they have no conflict of interest.