Bisphenol S Disrupts Estradiol-Induced Nongenomic Signaling in a Rat Pituitary Cell Line: Effects on Cell Functions

1Xenoestrogens (XEs) are a diverse group of synthetic agents (e.g., pesticides, surfactants, plastics monomers) that can mimic and disrupt the actions of physiologic estrogens. Many XEs can remain in the environment for a long time, thus increasing the likelihood for human and wildlife exposure (; deJager et al. 1999;; ).

2Bisphenol A (BPA), a leachable monomer of polymerized polycarbonate plastics, has been used commercially since 1957 and is also found in food can liners and coatings on thermal cash register paper. Humans are exposed to BPA primarily from food and water (H2O) contaminated by manufactured products, particularly during the heating of plastic containers. In the National Health and Nutrition Examination Survey (NHANES), BPA levels ranged from 0.4 to 149 μg/L (1.8–660 nM) in 92.6% of urine samples from U.S. residents ≥ 6 years of age.

3Exposure to BPA in humans has been implicated in the development of chronic diseases, including diabetes, asthma, and cancer, and in causing decreased fecundity in wildlife via disrupted spermatogenesis and ovulation. The European Food Safety Authority has set a tolerable daily intake (TDI) for BPA of 0.05 mg/kg body weight/day, a value accepted by many regulatory agencies, including the. Because of increased concern over the safety of BPA,, and more recently the European Union and the U.S., have banned its use in plastic bottles for infants.

4More stringent global regulations on BPA production and use have led to the development of alternative, more heat-stable bisphenol compounds. Among these alternative compounds is 4,4´-dihydroxydiphenyl sulphone, commonly known as bisphenol S (BPS). Because of the novel nature of BPS, in vivo toxicity studies have not been reported, nor has the ability of BPS to disrupt the actions of physiologic estrogens been explored. Several studies have tested the effects of BPS via genomic mechanisms using extremely high concentrations (concentrations unlikely to be leached from BPS-containing products). At concentrations as high as 0.1–1 mM, BPS showed only slight estrogenic activity in a 4-hr recombinant two-hybrid yeast test system. In another such study, showed that 40 μM BPS had 15-fold lower genomic estrogenic activity than BPA. However, BPS was equipotent to BPA in an estrogen-response-element–driven green fluorescent protein expression system in MCF-7 breast cancer cells. Discrepancies between these studies were attributed to species (yeast vs. mammalian) differences. However, tissues frequently differ in responses, so this could also explain the discrepancies. No studies prior to ours have examined BPS for nongenomic mechanisms of action or at the low concentration ranges likely to be present in foods, environmental samples, or humans.

5BPA can potently interfere with the actions of endogenous estrogens in pituitary cells via several types of nongenomic signaling [e.g., mitogen-activated protein kinases (MAPKs), Ca2+ influx] acting via membrane estrogen receptors [mERα, mERβ, and GPER/GPR30 (G protein-coupled estrogen receptor)], and thus alter functional responses [cell proliferation, prolactin (PRL) release, and transporter function] at picomolar and subpicomolar concentrations. Physiologic estrogen actions are disrupted by BPA and other XEs for both timing and magnitude of responses—enhancing or inhibiting—depending on their concentrations. Introduction of a new active bisphenol compound (BPS) into the environment poses an unknown threat for signaling and functional disruptions.

6In the present study we examined the effects of BPS on nongenomic signaling at concentrations that allow full assessment of potency given the nonmonotonic concentration responses we expected based on our previous studies of BPA. To simulate likely exposures, we tested BPS both alone and in combination with the physiologic estrogen estradiol (E2). Using prototypic receptor inhibitors, we sought to identify the predominant mER through which BPS initiates nongenomic signaling. Effects of BPS on associated downstream (from MAPKs) functional end points were also examined, including changes in cell number (proliferation or decline) and caspase activations or inhibitions occurring via external stimuli (caspase 8) versus internal stimuli (caspase 9). Together, these mechanisms can contribute to alterations in cell number. Finally, we examined the effect of BPS on peptide hormone release (PRL). These measurements employed high-throughput plate immunoassays to facilitate quantitative comparisons between responses to different compounds and mixtures.

7Cell culture. We selected the clonal rat prolactinoma cell line GH3/B6/F10 on the basis of its naturally high expression of mERα, which enables it to respond robustly in tests for nongenomic signaling and functional end points. Cells were routinely subcultured with phenol red-free Dulbecco’s modification of Eagle’s medium (DMEM, high glucose; Mediatech, Herdon, VA) containing 12.5% horse serum (Gibco BRL, Grand Island, NY) and defined supplemented calf and fetal serum (Thermo Fisher, Waltham, MA) at 2.5% and 1.5%, respectively. Cells of passages 10–20 were used for these experiments.

8Concentration ranges selected. All concentrations for time courses and dose responses were chosen based on our previous studies that demonstrated expected potencies, efficacies, and rapidity of the responses. The chosen concentrations of BPS reflect the range of concentrations likely to be found in the environment, centering on urinary levels (0.299 ng/mL or 1.2 nM), observed in Albany, New York, residents. Lower concentrations are of interest to determine how sensitive biological systems are to presumably more widespread exposure concentrations. These concentrations of other XEs were able to activate MAPKs and caspases and disrupt PRL secretion.

9Quantitative ERK (extracellular signal regulated kinase) and JNK (c-Jun-N-terminal kinase) phosphorylation assays. To quantify phosphoactivation of ERK (pERK) and JNK (pJNK), we used a fixed cell-based immunoassay, as previously described. Cells (104/well) were plated in 96-well plates (Corning Incorporated, Corning, NY) and allowed to attach for 24 hr. The original plating medium was then replaced with DMEM containing 1% charcoal-stripped (4×) serum for 48 hr to deprive cells of serum hormones. Medium was then removed, and cells were exposed to BPS (10–15 to 10–7 M), BPA (10–15 M), or E2 (10–9 M) (all from Sigma-Aldrich, St. Louis, MO) to assess time- (0–60 min) and concentration-dependent changes. Test compounds were dissolved in ethanol (EtOH), then diluted in DMEM containing 1% charcoal-stripped serum. The vehicle control was 0.001% EtOH in DMEM. To stop mER-initiated signaling, cells were fixed with a 2% paraformaldehyde/0.2% picric acid solution (Fisher Scientific, Pittsburgh, PA) at 4°C for 48 hr. Once fixed, cells were incubated with phosphate-buffered saline (PBS) containing 0.2% fish gelatin and 0.1% Triton X-100 (Sigma-Aldrich) for 1 hr at room temperature (RT) and then incubated with primary antibodies against pERK or pJNK (1:500 in PBS/0.2% fish gelatin/0.1% Triton X-100; Cell Signaling Technology, Beverly, MA) overnight at 4°C. After washing 3 times with PBS, cells were incubated with a biotin-conjugated secondary antibody (1:500 in PBS/0.2% fish gelatin; Vector Laboratories, Burlingame, CA) for 1 hr at RT. Cells were again washed 3 times in PBS, and incubated with Vectastain ABC-AP solution (50 μL/well; Vector Laboratories) for 1 hr at RT, followed by Vectastain alkaline phosphatase substrate (pNpp solution; 50 μL/well). Plates were incubated in the dark for 30 min at 37°C, and the signal for the product para-nitrophenol (pNp) was read at A405 (absorbance of 405 nm) in a model 1420 Wallac microplate reader (PerkinElmer, Boston, MA).

10Crystal violet (CV) assays. The pNp signal was normalized to cell number, as determined by the crystal violet assay. After washing 2 times with H2O to remove the alkaline phosphatase reaction reagents, the plates were dried at RT for 1 hr. CV solution (0.1% in H2O, filtered) was added (50 μL/well), incubated for 1hr at RT, and washed 4 times with H2O. The dye was released from the cells with acetic acid (10% in H2O; 50 μL/well) at RT for 30 min, and the A590 signal was read in the Wallac microplate reader.

11Receptor inhibitor studies. We used prototypic selective receptor antagonists to determine the involvement of the three different types of mERs (ERα, ERβ, and GPR30) in ERK activation upon exposure to BPS (10–14 M). Receptor involvement in responses to BPA and NP have been determined previously. Cells (104/well) were plated in 96-well plates, allowed to attach for 24 hr, and treated with DMEM containing 1% 4× charcoal-stripped serum for 48 hr to deprive cells of serum hormones. Media were then removed and cells preincubated for 1 hr at 37°C with medium (50 µL) containing antagonists for ERα {MPP; 1,3-bis(4-hydroxyphenyl)-4-methyl-5-[4-(2-piperidinylethoxy)phenol]-1H-pyrazole dihydrochloride}, ERβ {PHTTP; 4-[2-phenyl-5,7-bis(trifluoromethyl) pyrazolo[1,5-a]pyrimidin-3-yl]phenol}, and GPER/GPR30 {G15; (3aS*,4R*,9bR*)-4-(6-bromo-1,3-benzodioxol-5-yl)-3a,4,5,9b-3H-cyclopenta[c]quinolone); all compounds were obtained from Tocris Bioscience (Bristol, UK) and target both membrane and intracellular versions of estrogen receptors. DMEM medium (50 μL) containing 10–14 M BPS was then applied to cells for 5 min, followed by fixation with a 2% paraformaldehyde/0.2% picric acid solution. The quantitative ERK phosphorylation assays were performed as described above.

12Determination of cell proliferation. As described previously,, subconfluent cells were seeded into 96-well plates coated with poly-d-lysine (5,000 cells/well) and allowed to attach overnight. Plating medium was replaced with DMEM containing 1% 4× charcoal-stripped serum for 48 hr, and then with treatment medium containing increasing concentrations of BPS or E2 (10–15 to 10–7 M) alone or BPS concentrations plus 10–9 M E2. After 3 days, cells were fixed (2% paraformaldehyde/0.1% glutaraldehyde in PBS; 50 μL/well) for 20 min at RT. Cell numbers were assessed by CV assay to compare the proliferative effects of BPS at different concentrations.

13Determination of caspase activity. Subconfluent GH3/B6/F10 cells were seeded into 96-well plates (5 × 103/well) and allowed to attach overnight. Treatments began the next day; cells were exposed for 24 hr to DMEM medium with 1% 4× charcoal-stripped serum containing one of the following: 10–14 M BPS, 10–8 M BPS, 10–14 M BPS plus 10–9 M E2; 10–8 M BPS plus 10–9 M E2. At designated times, treatment medium was suctioned off and cells were lysed with 50 µL lysis buffer (10 mM HEPES; 2 mM EDTA; 0.1% CHAPS; pH 7.4) to which 1 mM DTT (dithiothreitol; 1:2,000, freshly prepared; Sigma-Aldrich) was added. Plates were then stored at –70°C until the assay was performed. Staurosporine (500 nM; Sigma-Aldrich) dissolved in DMSO was used as a positive control for activation of caspases 8 and 9.

14For caspase assays, frozen plates were defrosted at 4°C, and assay buffer (50 mM HEPES; 100 mM NaCl; 0.1% CHAPS; 1 mM EDTA; 10% glycerol) (50 μL/well) was added. Freshly prepared 10 mM DTT and caspase 8 (Ac-IETD-AFC) or caspase 9 (Ac-LEHD-AFC) substrates (Enzo Life Sciences, Farmingdale, NY) were added to the assay buffer at final concentrations of 50 μM. Plates were then incubated in the dark at 37°C for 2 hr. The released fluorescent product, 7-amino-4-trifluoromethylcoumarin (AFC), was read using a Flexstation 3 spectrofluorometer (Molecular Devices, Sunnyvale, CA) at an excitation wavelength of 400 nm and an emission wavelength of 505 nm.

15PRL release. Our study design and conditions were based on previous studies from our laboratory. After cells (0.5–0.7 × 106) were plated into poly-d-lysine–coated 6-well plates overnight, they were hormone deprived in DMEM containing 1% 4× charcoal-stripped serum for 48 hr. Cells were then incubated for 30 min in DMEM/0.1% BSA and exposed to different concentrations of BPS alone (10–15 to 10–7 M) or in combination with E2 (10–9) for 1 min. Cells were then centrifuged at 350 × g for 5 min at 4°C; the supernatant was collected and stored at –20°C until radioimmunoassay (RIA) for PRL. Cells were then fixed with 1 mL of 2% paraformaldehyde/0.1% glutaraldehyde in PBS, and cell numbers were determined via the CV assay.

16Concentrations of PRL secreted into the media were determined using components of the rat PRL RIA kit from the National Institute of Diabetes and Digestive and Kidney Disease and the National Hormone and Pituitary Program (Baltimore, MD). We combined 100 μL of cold standard (rat PRL-RP-3) or unknown sample with 500 μL rPRL-s-9 antiserum (final dilution of 1:437,500 in RIA buffer containing 80% PBS, 20% DMEM, and 2% normal rabbit serum) and 200 μL of 125I-labeled rat PRL (15,000 counts/tube diluted in RIA buffer; PerkinElmer, Wellesley, MA). The samples were then incubated and shaken overnight at 4°C. Anti-rabbit IgG was then added (200 μL of 1:9 final dilution in RIA buffer) and the samples were incubated and shaken for 2 hr at RT. After polyethylene glycol solution (1 mL; 1.2 M polyethylene glycol, 50 mM Tris, pH 8.6) was added, samples were incubated and then shaken at RT for 15 min. The samples were centrifuged at 4,000 × g for 10 min at 4°C, the supernatants decanted, and the pellets counted in a Wizard 1470 Gamma Counter (Perkin Elmer). PRL concentrations were calculated and normalized to CV values representing cell number.

17Statistical analysis. Statistical analysis was performed using SigmaPlot, version 12 (Systat Software Inc., Chicago, IL). We applied one-way analysis of variance (ANOVA) to the dose- and time-dependent studies to assess the statistical significance of mean values produced by varying XE exposures. A Holm-Sidak comparison against vehicle control or against E2 treatment was used after the ANOVA to evaluate significance. We considered the overall α level of 0.05 to be statistically significant. In addition, we ran a Student’s t-test where the significance between values was borderline by one-way ANOVA, as noted in the figures.

18Exposure of GH3/B6/F10 cells to BPS for 5 min caused ERK activation (Figure 1A) similar to that caused by E2 both here and previously. The lowest tested BPS concentrations evoked a higher pERK response than did 10–9 M E2; the response steadily decreased with increasing BPS, indicating a nonmonotonic dose response. The combination of increasing concentrations of BPS with constant 10–9 M E2 caused a lower pERK activity than did BPS alone, and was significantly lower than the nanomolar E2 response at the highest BPS concentrations (10–100 nM). In contrast, BPS did not produce significant pJNK activation (Figure 1B); instead the highest BPS concentration (10–7 M) caused deactivation significantly below vehicle levels. However, when BPS and E2 were administered together, JNK was strongly activated—above the level seen with E2 alone—and again featured a nonmonotonic dose–response curve, with the lowest concentrations evoking the largest responses.

19We also examined the time dependence of these responses at optimal response concentrations (10–14 M BPS, 10–9 M E2; Figure 2A,B). E2 produced a typical oscillating two-peak ERK response, with the first peak within 5 min, followed by a second peak at 30 min. BPS phosphoactivated ERK within 2.5 min but did not show significant oscillation. Responses induced by BPS and E2 were significantly different from each other. The combination of 10–14 M BPS and E2 showed a slightly oscillating pattern, although differences between stimulated points were not significant. We have previously observed rephasing of responses due to XE combined with E2. Therefore, even at this very low concentration (10–14 M), BPS was able to disrupt the timing of the response to a physiologic estrogen. Even though 10–14 M BPS alone could not activate JNK at any time point tested, its combination with E2 dramatically enhanced the early and sustained pJNK response to E2 (Figure 2B).

20A prototypic chemical inhibitor for ERα (MPP, 10–8 M) was the most effective antagonist of E2 and BPS-induced responses (Figure 3). In comparison, inhibitors for ERβ (PHTTP, 10–7 M) and GPER/GPR30 (G15, 10–7 M) were much less effective in reducing the phosphoactivation of ERK by E2 and BPS. Therefore, mERα appears to be the predominant receptor that mediates this nongenomic response to BPS.

21After a 3-day exposure, 10–9 M E2 and BPS had similar effects on cell proliferation, causing a nonmonotonic stimulation, as we observed previously with E2. The combination of BPS and E2 did not stimulate cell proliferation, but instead suppressed cell numbers to levels below those of cells exposed to vehicle (Figure 4).

22Because decreases in cell number can be caused by apoptosis, we assayed caspases 8 and 9 to determine whether the extrinsic or intrinsic apoptotic pathways were activated. Caspase 8 was activated by both BPS and BPS plus E2 (10–9 M) at all time points tested (4–24 hr), regardless of the BPS concentration used (Figure 5A). In contrast, caspase 9 was significantly activated only at 24 hr, and by low concentrations of BPS (10–14 M) or its combination with E2 (Figure 5B). The positive control (staurosporine) was active at all times and on all caspases, as expected. Interestingly, nanomolar E2 alone suppressed caspase 9 activity below the level of vehicle controls at all time points, whereas inhibition below vehicle levels was observed only at the 8-hr time point for caspase 8, with some timing differences from our previous studies.

23The GH3/B6/F10 cell line secretes PRL in response to E2 and a variety of estrogenic compounds, thus making this model an excellent tool for evaluating functional responses to estrogens. After a typical exposure time of 1 min, BPS did not significantly increase PRL secretion but E2 did (Figure 6). In cells treated with BPS plus E2, the E2-induced PRL release was severely inhibited in a nonmonotonic pattern, well below that in nanomolar-E2-treated cells; at most concentrations of BPS, the PRL release was well below that of vehicle. The PRL release after treatment with BPS (10–10 M) plus E2 was not statistically different from the level of release caused by E2 alone, nor was it statistically different from vehicle because of errors for that measurement.

24Increased scrutiny and concern by government agencies and environmental advocacy groups led to the development of potential chemical replacements for BPA, such as BPS. Although BPS is less likely to leach from plastic containers with heat and sunlight, it does still escape the polymer in small quantities under normal use. Our results show that BPS is active at femtomolar to picomolar concentrations, and can alter a variety of E2-induced nongenomic responses in pituitary cells, including pERK and pJNK signaling and functions (e.g., cell number, PRL release).

25BPS had the same capability as E2 for initiating phosphoactivation of ERK across concentrations and times, with lower concentrations of BPS being more effective. BPS was also equipotent to BPA in the phosphoactivation of ERK. Such nonmonotonic dose responses are controversial and have been heavily examined recently. The fluctuation of MAPK activities with concentration and time could involve several mechanisms, including receptor desensitization due to overstimulation, activation of phosphatases, and simultaneous activation of multiple signaling pathways, thereby activating proteins at different rates. MAPK down-regulation is critical for preventing adverse effects of extended pathway stimulation. In our mixture studies (BPS plus E2), attenuation of the ERK response may protect the cell against unnecessary and perhaps dangerous estrogenic stimulation caused by the increased overall estrogenic concentration of two estrogenic compounds.

26Nongenomic and functional actions initiated in this cell line were mediated predominantly by mERα. In previous studies, chemical inhibitors effective for both mERα and mERβ (ICI 187,634) also blocked ERK responses. In contrast to the GH3/B6/F10 cells used here, GH3/B6/D9 pituitary cells expressing low mERα levels were unable to respond via E2-induced activation of MAPK signaling. In the present study, our experiments with subtype-selective antagonists also demonstrated that mERα was the predominant membrane receptor mediating these responses, as we reported previously, although, as in our past studies, ERβ and GPR30 also contributed to this ERK response to estrogens.

27Phosphoactivation of ERK and JNK has been closely associated with opposing functional end points. For example, ERK signaling (RAF→MEK1,2→ERK1,2) is often associated with cell differentiation and growth, whereas JNK signaling is usually thought to accompany the initiation of apoptosis. Simultaneous phosphoactivation of ERK and inactivation of JNK by BPS, as our data show, could simultaneously stimulate proliferation and inactivate cell death, magnifying the increase in cell number. Our BPS/E2 mixture activated both ERK and JNK, perhaps correlating with a decline we saw in cell number, if the balance of these two activities is important for the outcome. Earlier studies reported that BPS alone is capable of inducing cell proliferation in the MCF-7 cell line but noted that BPS began to show cytotoxic effects at concentrations > 10–4 M (well above the highest concentration we tested). Therefore, the proliferative/antiproliferative responses caused by BPS can happen in multiple responsive tissues.

28This is the first study to explore the ability of BPS to activate caspases. Early activation of caspase 8 (compared with caspase 9) indicates that the extrinsic pathway, which involves extracellular stimuli acting on cell-surface receptors, is the primary apoptotic pathway. The reason for later and weaker activation of caspase 9 can be explained by crossover to that pathway via a lengthy process initiated by the cleavage of Bcl2-interacting protein (BID) in the caspase 8 pathway; this results in the translocation of BID to mitochondria, where it causes later release of cytochrome c and subsequent activation of caspase 9 pathways. We previously showed increased activation of caspase 8—but not caspase 9—in phytoestrogen-treated GH3/B6/F10 cells after 24 hr of treatment.

29Cell survival versus cell death is determined by the balance of several cellular signaling responses, and the activation of capsases is only one of many factors. There are also discrepancies in the literature about the role of ERK and JNK activation in controlling cell number. Phosphoactivation of ERK can, for example, lead to the activation of the antiapoptotic protein Mcl-1, which binds to Bax protein and prevents its activation, thus inhibiting apoptosis. Activation of ERK has also been shown to inhibit caspase 9 upon phosphorylation; this is perhaps a mechanism promoting the protective effects we see with E2 both here and in past studies. Phosphoactivation of JNK can lead to activation of several pro-apoptotic proteins such as Bax, caspase 3, cyclin D1, Fas, and interleukin 1. However, JNK has also been linked to the activation of prosurvival pathways, with the final functional response dependent on the overall balance between ERK and JNK activities. More examples of these conflicting outcomes need to be studied to resolve the composite contributions of MAPKs to the control of cell number.

30BPA and other XEs are potent inducers of PRL release; in contrast, BPS caused minimal PRL release on its own. However, BPS dramatically disrupted E2-induced PRL release, as do other XEs. Disturbances in the timing or amount of PRL released can lead to a variety of physiologic complications, including electrolyte imbalance, disruptions in growth and development, metabolic dysfunctions, behavioral disturbances, reproductive failure, or lactation failure. In all, PRL regulates > 300 biological functions. The differences we have observed between these two structurally similar bisphenol compounds warrant future examination of structure–activity relationships for these responses.

31Using urine samples collected for NHANES, observed total BPA concentrations across various demographic groups in the United States, with a geometric mean of 2.6 μg/L (10 nM). In comparison, determined the occurrence of BPS in humans in seven different countries, with the highest urinary geometric mean concentrations in Japan, followed by the United States (Albany, NY), with a geometric mean of 0.299 ng/mL (1.2 nM), a concentration much higher than those used in our studies. Because earlier studies focused entirely on genomic mechanisms of BPS action in which BPS was active only in the micro- to millimolar range, those effects would be relevant only to high-dose exposures such as industrial accidents.

32Our study is the first to demonstrate that the BPA-substitute BPS can induce rapid nongenomic signaling in estrogen-responsive pituitary cells at low (femtomolar to picomolar) concentrations. Another cause for concern is that BPS also interferes with physiologic E2 signaling that leads to several functional end points. These findings highlight the need for efficient in vitro screening methods to pretest possible substitutes for XEs before they are deployed in manufacturing. As more related compounds are tested, we can establish a list of structural features likely associated with risks in this class of chemicals, and perhaps guide future designs away from these structures that can adversely affect human and animal health.