R-equol, a synthetic metabolite of the dietary estrogen daidzein, modulates the nongenomic estrogenic effects of 17β-estradiol in pituitary tumor cells

ABSTRACT

We examined nongenomic signaling and functional effects (cell proliferation; prolactin release) of R-equol (R-eq), a synthetically produced metabolite of the phytoestrogen daidzein that preferentially binds to estrogen receptor-α, across a broad concentration range (10^–16 to 10^–7M). Membrane estrogen receptor-α, via which nongenomic signaling occurs, is enriched in the GH3/B6/F10 pituitary tumor cells used in these studies. We previously reported that other phytoestrogens, including daidzein, are potent inducers of nongenomic signaling acting via membrane receptors for estrogens, resulting in changes to these same functional responses. In the present studies, R-eq activated typical rapid signaling pathways normally evoked by estrogens, but with some differences in response timings and functional outcomes. Levels of R-eq (at 15 nM, a blood level typical for dietary phytoestrogens) were compared to a female physiological level of estradiol (E₂; 1 nM). Both estrogens activated multiple mitogen-activated protein kinases (ERK, JNK and p38) by phosphorylation within 2.5–15 min, with subsequent oscillations, as observed previously for other estrogens. Like E_2, R-eq also augmented intracellular calcium levels and caused prolactin release; in contrast to E₂, it did not produce a dose-dependent increase in cell proliferation, as estrogens that activate ERK often do. R-eq and E₂, both alone and in combination, activated G_αi by GTP-charging. However, R-eq suppressed 1 nM E₂-activated ERK, JNK and p38, as well as cell proliferation (most pronounced at typical phytoestrogen blood levels of 10^–10 to 10^–7M). We conclude that R-eq, like E₂, is a rapid activator of nongenomic signals, but when combined with E_2, can interfere with E₂-induced nongenomic estrogenic effects .

Keywords:

• calcium
• ERα
• ERK activation
• G_αi
• JNK activation
• kinases
• membrane estrogen receptors
• non-monotonic dose effects
• p38 activation
• prolactin
• phytoestrogen

Introduction

Soy-based foods are a major source of human exposure to phytoestrogens^1,2 that structurally resemble endogenous estrogens like estradiol (E₂), as well as a variety of synthetic estrogens. Daidzein is a plentiful and common phytoestrogen available from soy-based foods^3,4 that can be converted into 2 chiral forms of equol; gut micro-organisms produce S(-) equol (S-eq).^5-7 The core ring structures of the phytoestrogens daidzein and R(+)-equol (R-eq; the chiral form used in this study) closely resemble those of E₂ (see Fig. 1) and other estrogens. The binding affinity of S-eq to estrogen receptor-β (ERβ) is 13 times higher than that of R-equol (Ki 50nM), whereas R-eq showed a preference for binding to estrogen receptor-α (ERα) 4 times higher than that of S-eq; both have higher affinity for ERs than daidzein.⁸ Therefore, it is not surprising that phytoestrogens can act as selective estrogen receptor (ER) modulators (SERMs), acting differentially as agonists or antagonists for ERs depending upon the cell type involved.^9-12 Modifications in both the compounds’ skeletal structure and the end hydroxyls (red-white moieties in Fig. 1) can affect the affinity for ERα, as well as subsequent interactions of the liganded receptor. For example, daidzein sulfation at the 4’ position suppresses estrogen agonist activity, while at the 7’ position it increases estrogen agonist activity.¹³ Differences in patterns of estrogenic signaling, and disturbances of E₂-induced signaling when challenged with a phytoestrogen, can cause alterations of important estrogen-driven physiological actions.^14-16 Estrogenic activity via the nuclear forms of ERs α and β reflect their different binding affinities,⁹ but differential binding and activities at membrane forms of ERs (mERs) that drive nongenomic signaling actions are largely unknown.

Figure 1. Structure of E₂, daizein and R-eq. Upper Row: Two-dimensional (2D) structure comparisons of 17β estradiol, daidzein, and R-eq. Note hydroxyl groups at opposite ends of the molecules that are thought to be essential for binding into the ligand-binding pocket on ERα.⁸ Lower Row: Three-dimensional (3D) structures showing differences in the spatial arrangements of these hydroxyl groups (red-white) in the same compounds. The asymmetric center of R-eq is shown at the rightmost position (red dotted circle)^9,10.

Phytoestrogens such as coumestrol, daidzein, genistein, and trans-resveratrol have been suggested as possible safe post-menopausal hormonal replacement alternatives,¹⁷ though they may be riskier for breast cancer patients due to potential promotion of the growth of already established mammary tumors.^2,18,19 Asians eating diets high in soy are less susceptible to the disease consequences of estrogen overexposure (affecting initial induction of cancers of reproductive tissues)^2,20-22 or underexposure (causing osteoporosis, hot flashes, heart disease, and depression).^2,22-24 Interestingly, the incidences of these diseases in the Asian population become similar to Westerners when they adopt a Western diet.^{2,20-23,25-30} So there are likely some physiology- and disease-modifying activities of such dietary estrogens.

Estrogens, including phytoestrogens, can activate both the slower (hours to days) classical genomic pathway of transcriptional actions, and the non-classical rapid (seconds to minutes) nongenomic signaling mechanisms³¹ using a wide variety of signaling processes and protein mediators.^32,33 Our laboratory has previously investigated the lesser-known non-genomic signaling activities of phytoestrogens such as daidzein, coumestrol, genistein, and trans-resveratrol,^15,16,34 as well as a number of other estrogen mimetics.^31,33,35,36 We found phytoestrogens to be active at or below concentrations generally provided by dietary exposures. However, the daidzein equol metabolites have not previously been studied for actions through this non-classical pathway, and have interesting potential for dissimilar actions because of differential metabolic availability.

Daidzein serves as the microbial substrate for equols (mainly S-eq) when commonly introduced from various food sources such as soy beans, tofu, food additives, and milk substitutes. The range of equol concentrations in the plasma of humans varies widely depending upon dietary habits, geographic location, plant active constituents, gut microbiomes, and the use of commercially available supplement formulations.^3-8,37 However, human genetic variations prevalent in some subpopulations or individuals can result in dissimilar pharmacokinetics that affect attainable blood levels of S-eq. Only ∼35% of humans are high S-eq producers from daidzein metabolism; therefore, body fluid concentrations can vary significantly even with equivalent consumption.^3,37 Because of this, and because the chiral forms are differentially active and differentially produced by gut microflora, equols might be more easily controlled by direct dietary supplementation (instead of providing the precursor daidzein). It is also important to study the effects of equols over broad concentration ranges to fully consider their possible therapeutic uses or dangers, because equols can be present in humans at such widely varying doses due to both ingestion and metabolism. In addition, compounds that mimic hormones, like the hormones themselves, often display non-monotonic concentration dependence,^{15,33,35,38-41} making examination of detailed concentration curves necessary to fully understand their health consequences.

Because R-eq preferentially binds to ERα, and S-eq to ERβ,⁸ we evaluated R-eq for non-genomic effects in these studies using GH3/B6/F10 pituitary tumor cells that express relatively high levels of mERα.^42,43 We examined a variety of cellular signaling responses known to be elicited by estrogens, and 2 functional responses induced by estrogens in the pituitary.

Results

Time dependence of 1 nM E₂ vs. 15 nM R-eq MAPK activations

E₂ at a female physiological concentration (1 nM) caused rapid activation of ERKs (extracellular signal regulated kinases), initiating a response by 2.5 min that peaked at 10 min, then declined slightly, and rose again at 60 min (Fig. 2A), similar to what we observed previously.^44-46 R-eq at a concentration that approximates physiological S-eq concentrations resulting from dietary exposure to daidzein (15 nM) also significantly increased pERK by 2.5 min, peaking at 5 min, and sustaining this activation through 60 min with minor fluctuations. E₂ significantly activated JNKs (jun N-terminal kinases) (Fig. 2B) at all time points (2.5 min – 60 min), with peaks at 2.5 min and 30 min. Treatment with R-eq caused a slightly delayed JNK activation, peaking at 10 min and sustained through 60 min. Finally, E₂ significantly activated p38 kinase (Fig. 2C) by 2.5 min, peaking at 5 min, followed by a slight decline. Similarly, though not identically, R-eq activated p38 kinase by 5 min, peaking at 15 min, and this activity was sustained up to 60 min.

Figure 2. Time-dependent effects of E₂ and R-eq on MAPK activations. Cells were exposed to physiologic levels of E₂ (1nM), and dietary levels of R-eq (15nM) for 2.5 - 60 min. (A) Effects of E₂ and R-eq on ERK phosphorylation (pERK). For E₂ n = 56 spread over 6 experiments; for R-eq n=56 over 4 experiments. (B) Effects of E₂ and R-eq on JNK phospho-activation (pJNK). For E₂ or R-eq, n = 38 over 3 experiments. (C) Effects of E₂ and R-eq on p38 phospho-activation (p-p38). For E₂ n = 38 over 4 experiments; for R-eq n = 46 over 4 experiments. * = significant at p< 0.05 vs. control (vehicle) group.

Concentration dependence of E₂ vs. R-eq treatment on MAPK activations at 5 min

Cells were exposed to log incremental concentrations (10⁻¹⁶ to 10⁻⁷ M) of E₂ or R-eq for 5 min (Fig. 3). E₂ activated ERK (Fig. 3A) with a pattern similar to that we observed previously, showing a non-monotonic response with 2 different dose optima (10⁻¹⁵ and 10⁻⁸ M) and a midpoint non-effective dose (10⁻¹³ M). E₂ did not produce significant activation at the highest dose (10⁻⁷ M). R-eq also activated ERK with 2 dose optima (10⁻¹⁵ and 10⁻¹⁰ M), with ineffective doses in between and at the highest concentration. E₂ and R-eq significantly increased pJNK across the same concentration ranges, with a simpler concentration-dependent pattern at all doses of 10⁻¹⁴M and above (Fig. 3B). E₂ treatment caused the activation of p38 at all concentrations tested (10⁻¹⁵-10⁻⁷ M) except 10⁻¹⁴ M. R-eq also increased p-p38 (phospho-p38), but only at concentrations above 10⁻¹⁰ M (Fig. 3C). So both a typical female physiological concentration (1nM) of E₂ and a common dietary concentration for naturally occurring phytoestrogens (15nM) of R-eq activated all 3 kinases (ERK, JNK, and p38), with ERK activation being non-monotonic for both estrogens. We recognize that preparing concentrations at such a low levels can be more error-prone. In the past some experiments have shown significant results at this lowest concentration, and some did not. We have inspected our individual experiments to see if non-monotonic patterns are usually revealed, and we see them often, though not in every case. We have also seen non-monotonic dose-dependence patterns in many of our past published studies.^{15,33,35,44-62}

Figure 3. Dose-dependent effects of E₂ and R-eq on MAPK activations. Cells were exposed for 5 min to various concentrations of E₂ and R-eq. A), B) and C) line graphs demonstrating the effects of E₂ and R-eq on pERK, pJNK and p-p38, respectively. * = p< 0.05 vs. control (vehicle) group.

Dose-dependent effects of adding R-eq to E₂-induced phosphorylation of MAPKs

Next we wanted to determine if adding R-eq on top of already existing physiological E₂ (1 nM) would alter the response to E₂. Treating cells for 5 min with either 1nM E₂ (see horizontal bars, Fig. 4) or R-eq (10⁻¹⁶ to 10⁻⁷ M) alone caused phosphorylation of all MAPKs (ERK, JNK and p38). R-eq elicited a non-monotonic concentration pattern. However the addition of varying concentrations of R-eq to 1 nM E₂ suppressed the E₂-induced increase in ERK (Fig. 4A), JNK (Fig. 4B), and p38 phosphorylation (Fig. 4C), at higher concentrations. The activated state of p38 was the most sensitive to these effects, showing a suppression at as little as 10⁻¹⁵ M R-eq.

Figure 4. Dose-dependent effects of added R-eq on E₂-induced MAPK activities. Cells were exposed for 5 min to a mixture of E₂ (1nM, physiological concentration) and R-eq (range 10⁻¹⁶ to 10⁻⁷M). pERK (A), pJNK (B) and p-p38 (C) were measured compared to the activity level of the vehicle-treated controls (*= p< 0.05 vs. control) and the 1nM E₂-induced (see horizontal stippled bars) pERK, pJNK and p-p38 levels (^# = p < 0.05 vs E₂).

Effects of E₂ and R-eq on G_αi activation

We then tested R-eq for another recently demonstrated rapid estrogenic signaling effect, activation of G_αi protein by GTP charging⁵⁶ (Fig. 5). With the addition of GTPγS to block dissociation (by irreversible binding of this analog), we measured the amount of GTP-bound G protein that accumulated over 5 min. Both E₂ (1nM) and R-eq (15nm) significantly increased GTP-charged G_αi compared to vehicle controls, though R-eq did so less. The mixture of the 2 estrogens activated G_αi to the same extent as had E₂ alone.

Figure 5. E2 and R-eq, alone or in combination – effects on GTP-bound G_αi. Cells were exposed to E₂ (1nM) or R-eq (15nM) or both for 5 min, using the irreversible binding of the GTPγS analog to measure cumulative binding. GTPγS-bound G_αi levels were detected using an Ab specific for the complex. Vehicle n = 36; E₂ n = 20, and R-Eq n = 36, over 2 experiments. * = significance at p < 0.05 vs. vehicle; ^# = significance vs. E₂; $ = significance vs. R-eq.

Effects of E₂ and R-eq on intracellular free calcium levels and PRL release

We next examined another pathway of rapid estrogenic signaling responses, changes in intracellular calcium levels (Fig. 6), to see if they would be affected by R-eq in a dietary phytoestrogen concentration range in comparison to a physiologic level of E₂ (1nM). Exposure to either compound for 1 min significantly increased the intracellular calcium levels compared to controls, though E₂ was more effective (Fig 6A). We then tested a functional consequence of elevated calcium levels in these cells, PRL release (Fig. 6B). A 1 min exposure to 1nM E₂ significantly increased PRL secretion into the medium by ∼15%, as we have seen previously.^15,53 However, R-eq applied at a 15nM typical phytoestrogen dietary concentration increased secreted PRL by ∼70%.

Figure 6. Effects of E₂, R-Eq, and their combination on intracellular calcium concentrations and PRL release. Cells were exposed to E₂ (1nM) or R-eq (15nM) or both for 1 min. (A) Effects of E₂ and R-eq on intracellular calcium levels compared to vehicle control. Vehicle n = 142; E₂ n = 72; R-eq n = 25, over 4 experiments. (B) Effects of E₂ and R-eq on PRL release. The concentration of PRL released into the medium was measured by radioimmunoassay. Vehicle n = 36; E₂ n = 30 and R-eq n = 30, over 3 experiments. *=significance vs. vehicle, ^#=significance vs. E₂ at p<0.05.

Dose dependence of cell proliferation on E₂, R-eq and the mixture

Cells treated for 3 d with E₂ (10⁻¹⁶ to 10⁻⁷ M) showed increases in cell number, as we observed previously.^34,46,54,63 However, R-eq treatment did not change cell numbers over the entire concentration range tested. Addition of this entire range of R-eq concentrations to a 1nM physiological E₂ level (shaded horizontal bar) suppressed the E₂-induced increase in cell number (Fig 7B).

Figure 7. Dose-dependent effects of E₂, R-eq, and their combination on cell proliferation. Cells were exposed to various doses of E₂, R-eq and a mixture of E₂ and R-eq for 3 d. (A) E₂ vs. R-eq at a range of doses for each (10⁻¹⁶ to 10⁻⁶ M). For E₂ n = 42 over 7 experiments; for R-eq n = 30 over 5 experiments. All E₂ treated cells were significantly different than vehicle controls or R-eq at p< 0.05 (*). (B) A mixture of various concentration of R-eq (10⁻¹⁶ to 10⁻⁷ M) with a physiological E₂ concentration (1nM). For E₂ + R-eq n = 24 in 3 experiments; for R-eq n = 30 over 5 experiments. * = p < 0.05 vs. control (vehicle) group. ^# = p < 0.05 vs E₂.

Discussion

R-eq increased the activities of all MAPKs, elevated intracellular calcium, activated G_αi, and elicited PRL release in time frames consistent with rapid nongenomic signaling pathway actions; it thus joins the ranks of phytoestrogens known to act via these nonclassical (extranuclear) pathways. Yet R-eq had no effect on cell proliferation, and inhibited the proliferative effects of E₂, making it promising for therapeutic interventions without exacerbating the growth of tumor tissues for which estrogens can be a risk factor. This constellation of different effects on estrogen-inducible signaling pathways illustrates how complex is estrogen signaling in general, how estrogens can act differentially, and how the control of cell proliferation involves multiple estrogenic signaling inputs.

Many physiological estrogens and xenoestrogens can act potently via liganded mERα in multiple tumor cell types (breast,^48,64-66 pituitary,^{15,47,58,67-69} prostate,^19,60,70 pheochromocytoma⁷¹). We previously reported the ability of the R-eq precursor daidzein to activate nongenomic actions in mER-α enriched pituitary tumor cells.^15,34 Here we document that the metabolite R-equol can also initiate rapid (< 15 min) non-genomic signaling in pituitary cells at concentrations as low as femtomolar. All signaling pathways tested (3 types of MAPK activations, calcium influx, G protein activation) were actuated, as we saw previously with other phytoestrogens,^15,34,48 environmental estrogens^{33,49,58,63,71} and their metabolites,⁵⁷ and physiological estrogens.^{31,46,50,53,54,72} Though R-eq evoked signals in all of these pathways, it did not exactly duplicate the timing and concentration-dependence patterns elicited by E₂. The R-eq concentration dependence of ERK activations was non-monotonic, though not as strikingly as for some other xenoestrogens previously examined.^{15,33,44,49,53,63,71,72} These data show once again that different estrogens are unique in the signaling profiles they induce.

One of the most rapid nongenomic actions we have studied to date (occurring within 15 seconds) is the charging of G_αi regulatory subunit with its activating nucleotide GTP, while mERα physically associates with G_αi proteins in the cell membrane of these pituitary cells.⁵⁶ Though R-equol increased GTP binding to G_αi, combining a common dietary phytoestrogen level of R-eq with E₂ did not blunt the E₂-induced response, as it did for other responses measured (MAPKs activations). The lack of an effect at one dose could reflect previous observations where some concentrations of a xenoestrogen used to challenge the physiological estrogens E₂, estrone, and estriol enhanced their actions, while others (often higher) blunted the response, and some had no effect.^{34,45,53,54,58,71} However, if R-eq does act on this particular signaling pathway with a non-monontonic dose-response pattern, only detailed concentration range studies may reveal any enhancing or inhibiting effects at other possible exposure levels.

We measured 2 functional responses, and found them to be associated with the expected signaling responses, though not perfectly matching with the expected dose and signaling contributions. Secretion of peptide hormones like PRL is thought to be triggered by increases in cellular calcium levels, which both R-eq and E₂ elicited. However, R-eq caused a larger PRL release, while evoking a smaller change in intracellular calcium compared to E₂. We and others have seen similar deviations from simple correlations of peptide secretion magnitude with calcium levels. There are certainly other controls over the final secreted peptide levels, such as voltage-dependent calcium channel activity, vesicle docking and membrane fusion, protein kinases A and C activities, and nucleotide levels (ATP, cAMP and GTP)^72,73 which we have not tested here. PRL is also known as a growth factor,⁷⁴ but its release by R-eq apparently did not provoke cell proliferation in this system.

The MAPKs have been strongly implicated in influencing cell number changes,⁷⁵ and R-eq caused the activation of all 3 kinase classes assessed here similarly, though not identically to E₂.Though it is generally thought that ERK activation should predict a proliferative response, this is not always the case, as we^{34,46,51,58,59} and others^76,77 have shown previously, and here. Some of the additional downstream consequences of ERK and p38 activation, such as the induction of reactive oxygen species,⁷⁸ can lead to cell death and so reduce cell numbers. In addition, the alterations in the balance of ERK, JNK (usually eliciting apoptosis), and p38 responses at certain timing interval oscillations may be more predictive. Clearly each signaling or functional endpoint must be assessed independently and in detail to build a composite mingled signaling profile of xenoestrogen actions on a given cell type. How they may interfere with different endogenous estrogen actions on the same endpoints will predict their effects in a hormone-replete animal or human.

The sensitivity of our model system to physiological as well as other estrogens is high. We sometimes see no effects at the 10⁻¹⁵ to 10⁻¹⁶ M levels,^34,51 and at other times we do.^{45,46,54,58,64} That is why we have examined concentration ranges at increasingly lower levels recently. We think it is very rare that femtomolar sensitivity is tested by others, for several reasons: 1) There has not been general recognition of such potent responses because most researchers do not have experimental systems that allow such a wide and detailed range of concentrations to be tested at the same time; the relatively high through-put nature of our assay allows that. 2) Along with that, there has been an under-appreciation of often non-monotonic concentration dependence caused by steroids and their mimics. When a non-responsive dose is found, few ever test responses below that dose. 3) Our well-controlled in vitro settings that allow sensitivities like these to be observed involve exhaustive charcoal stripping of our sera to remove endogenous steroids (see Methods). This allows us to observe previously unrecognized non-monotonic patterns involving these much lower concentrations where few have measured responses before. Non-monotonic response patterns are very prevalent in these rapid signaling responses in our hands and in a cell model that was selected for the robust expression of mERα.⁴²

Cell proliferation is a complex endpoint that weighs a variety of upstream inputs before it executes. It is not surprising that this includes both nongenomic and genomic signaling pathways of estrogens, and cannot be predicted by any one signaling stream. Estrogen-mediated cell proliferation occurs over a longer time frame (stably measured by 3 d after exposure^34,46,75), and the cell-cycle checkpoints must meet many requirements before allowing the cell to divide.⁷⁹ Our study showed that R-eq was itself unable to cause cell proliferation, but could suppress E₂-induced pituitary cell proliferation. By contrast, daidzein, the precursor of R-eq, was able to increase cell proliferation at 10⁻⁷ to 10⁻⁶ M, an achievable blood level.⁸⁰ Some individuals do not metabolize daidzein to S-eq (for which we do not yet have these types of data). R-eq, as a cell proliferation-inert compound that inhibits the growth-enhancing properties of other estrogens, but is not produced naturally in humans, may potentially be a therapy against estrogen-fueled tumor growth. This would require providing the synthesized metabolite directly as a dietary supplement, while limiting dietary exposures from daidzein. We note that R-eq was found to reduce palpable rodent mammary tumors in one study.⁸¹ If S-eq were to be found in the future to have any of these cancer-cell proliferation capabilities, then it might be a natural preventative for those who consume daidzein and produce S-eq naturally.

Though phytoestrogens in general have been touted as therapeutic alternatives for replacing missing physiological estrogens after surgical or natural menopause, not much is known about the mechanistic effects of many individual compounds or their combinations with levels of endogenous estrogens such as developing/young adult women and men would experience. As we have shown here, and in our past studies,^45,53,54,71 either enhancements or disruptions are possible, depending upon the kind of response examined. It will be essential to understand the mechanisms driving biological effects in each type of cell or tissue, for each type of phytoestrogen and its metabolites, so that unique patterns of therapeutic or harmful actions can be understood. Because levels of physiological estrogens vary with age and sex, the presence of R-eq and other phytoestrogens could cause alterations (either enhancements or interference) differentially in different sex and life-stage categories of people. That is why the consumption of a phytoestrogen or precursor such as daidzein by infants and children as milk substitutes may present an entirely different situation when compared to adults or the elderly, or those worried about estrogen-driven tumor cell proliferation, or pregnant women. Our study speaks to the issue of R-eq interactions with only one physiological estrogen at a concentration representing a mid-cycle adult woman (E₂, 1nM). Understanding the features of phytoestrogens that drive these disruptive vs. enhancing mechanisms will allow us to choose or redesign exogenous estrogen sources that exclude estrogenic or anti-estrogenic activities when needed. Perhaps we may eventually be more successful at predicting outcomes based on details of estrogens’ chemical structures, but we will need to train such an analysis system with many more examples of the multiplicity of signaling pathways affecting specific endpoints before this can be truly useful in preventive and therapeutic medicine.

Materials and methods

We purchased phenol red-free Dulbecco's Modified Eagle Medium (DMEM) from Mediatech, horse serum from Gibco BRL, and defined supplemented calf sera and fetal bovine sera from Hyclone. Paraformaldehyde and glutaraldehyde were purchased from Fisher Scientific. Antibodies (Abs) used in the ERK, JNK, and p38 phosphorylation studies (individually described below) were purchased from Cell Signaling. All other materials were purchased from Sigma (now Merck). R-eq (Santa Cruz Biotechnology) and 17-β E₂ were dissolved in absolute ethanol and this stock was then diluted to the desired concentration with DMEM media containing 1% charcoal-stripped serum. The final concentration of ethanol vehicle was 0.001%.

GH3/B6/F10 (mERα-enriched) rat pituitary tumor cell sublines⁴² were routinely cultured in DMEM containing 12.5% horse serum, 2.5% defined supplemented calf serum, and 1.5% fetal calf serum. They were then switched to hormone-free media in Corning multi-well plates prior to our experiments, as described below. Cells were used between passages 10 and 20.

Hormone images

ZINC is a free database of commercially available compounds for virtual screening. It gives a unique external ID for a particular compound (ZINC ID's). The 3D visualization files were obtained from PubChem (an open chemistry database). 3D visualization tools (RASMOL, JMOL etc) were used to draw 3D structures of R-eq, E₂, and daidzein. The ZINC ID for R-eq is ZINC00388660 (PubChem CID: 6950272).

MAPK assays

We originally developed this assay to assess activated ERK 1 and 2 levels in fixed GH3/B6/F10 cells,⁴⁷ and subsequently adapted it to equivalent assays for p38s and JNKs.^34,82 Briefly, cells were plated at a density of 10,000 cells/poly-D-lysine-coated well of 96-well plates. The following day growth media were replaced with DMEM containing 1% charcoal-stripped (4×) serum (to deprive cells of steroids) for 48 hr. For time-dependent experiments, cells were then treated with physiological/dietary levels of estrogens (either 1 nM E₂ or 15 nM R-Eq or vehicle) for 2.5 to 60 min. For dose-dependent experiments cells were treated with a wide range (10⁻¹⁶ to 10⁻⁷ M) of E₂ or R-eq doses (or vehicle) for 5 min. The cells were then fixed with 2% paraformaldehyde/0.2% picric acid in phosphate-buffered saline (PBS) at 4°C for 48 hr. After fixation, the cells were permeabilized with PBS containing 0.2% fish gelatin and 0.1% Triton X-100 for 1 hr at RT (room temperature), and washed 3X with PBS. Primary Abs against phosphorylated ERKs (p-Thr202/Tyr204) or phosphorylated JNK (p-Thr183/Tyr185), or phosphorylated p38 (p-Thr180/Tyr182) were added at 1:500 dilutions. After overnight incubation at 4°C, the cells were washed 3X with PBS; a biotin-conjugated secondary Ab (Vector Labs, 1:300) in PBS/0.2% fish gelatin was then added for 1 hr at RT. The cells were again washed 3X with PBS, incubated with Vectastain ABC-AP solution (Vector Labs) for 1 hr at RT, and again washed 3X with PBS, followed by addition of Vectastain alkaline phosphatase substrate plus levamasole (an endogenous phosphatase inhibitor). Plates were incubated in the dark for 30 min at 37°C, and the signal for the product of the phosphatase action (para-nitrophenol) was read at A₄₀₅ in a model 1420 Wallac microplate reader (Perkin Elmer). The number of cells in each well was estimated by the crystal violet (CV) assay^64,82,83 and used to normalize the antigen levels of these phosphorylated enzymes to cell numbers in individual wells. Briefly, the cells were stained with 0.1% CV for 30 min and destained with distilled water. The dye was released with 10% acetic acid at RT for 30 min, and the A₅₉₀ signal of the extract was read in the microplate reader. The experiments were repeated at least 3 times using different passages of cells on different days.

Cytosolic free Ca²⁺ imaging

GH3/B6/F10 cells were plated at a density of 15,000/well in black-sided, clear bottom 96-well tissue culture plates. The next day, the growth media were changed to 1% 4X charcoal-stripped fetal calf serum for 48 hrs. Ca_i²⁺ release was determined using a Fluo-4 No-wash Calcium Assay Kit (cat# F36205, Molecular Probes, Inc.) following the company's protocol. Briefly, culture media were removed and 50 μL of the dye loading solution (containing 2.5 mM probenecid) was added to each well. Plates were incubated for 60 min at 37^oC and an additional 60 min at RT in the dark. Fluorescence (excitation at 485 nm and emission at 525 nm) was measured with a Flexstation 3 instrument (Molecular Devices). A baseline was established for each well during the initial segment of each run for 17 sec, followed by addition of 50 μL of 2X concentrated treatment agents (estrogens or vehicle). Fluorescence was recorded every 1.7 second for 60 sec. Maximum peak height was determined for each well with Flexstation software (SoftMax Pro 5.2).

G_αi activation assay

GH3/B6/F10 cells were plated in poly-D-lysine-coated 96-well plates (10,000/well) and the next day incubated with media containing 1% charcoal-stripped FBS, continuing for 48 hrs. Cells were then incubated with DMEM medium alone for 2 hrs. The cells were then permeabilized with saponin (0.03mg/ml) and GTPyS (100μM) was added for 10 min at RT. Following a 5 min estrogen treatment the cells were fixed with 4% paraformaldehyde, centrifuged at 500 xg for 2 min and then incubated for an additional 8 min at RT. Cells were then blocked for 1 hr at RT with PBST (PBS with 0.1% Tween 20) + 1% gelatin. Next the cells were incubated with a mouse monoclonal Ab to the active (GTP-charged) form of G_αi, (1:1000; New East Biochemicals) overnight at 4°C. Cells were then washed 3X with PBS, and a biotin-conjugated secondary Ab (Vector Labs, 1:300) in PBS/1% fish gelatin was added for 1 hr at RT. The cells were again washed 3X with PBS, incubated with Vectastain ABC-AP solution (Vector Labs) for 1 hr at RT, and again washed 3X with PBS. Plates were incubated in the dark for 30 min at 37°C, and the signal for the product of the phosphatase action (para-nitrophenol) was read at A₄₀₅ in a model 1420 Wallace microplate reader (Perkin Elmer). The data were normalized to cell number in each well as estimated by the CV assay.^64,82,83

Cell proliferation assays

Subconfluent cells growing in serum-containing medium were seeded into 96-well plates (5000 cells/well) coated with poly-D-lysine, allowed to attach overnight, and the next day treated with E₂ or R-eq in DMEM medium containing 1% 4× charcoal-stripped FBS for 3 d. Cell numbers were assessed by the CV assay (described above) to compare the proliferative effects of E₂ or R-eq at different concentrations. We used ∼6 samples for each treatment group, and experiments were repeated at least 3 times using different passages of cells. We chose a day 3 assessment time point based on our previous studies.^34,46

Prolactin (PRL) release

Cells (0.5–0.7 × 10⁶) were plated in poly-D-lysine-coated 6-well plates. After serum deprivation in ITS defined medium [5 μg/mL insulin, 5 μg/mL transferrin, 5 ng/mL sodium selenite, and 0.1% bovine serum albumin (BSA)] for 48 h, this medium was replaced with DMEM/0.1% BSA for a 1-hr incubation at 37°C. The cells were treated with E₂ or R-eq for 1 min and centrifuged at 4°C (350×g for 5 min); the supernatant was then collected and stored at -20°C. The PRL concentrations 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. Briefly, RIA buffer [80% PBS, 20% DMEM, 2% normal rabbit serum], 100 μL cold standard (rat PRL-RP-3) or unknown sample, rPRL-s-9 antiserum (final dilution of 1:437,500 in RIA buffer), and [¹²⁵I]-rat-PRL (PerkinElmer; using 15,000 counts per tube diluted in RIA buffer) were combined and incubated with shaking overnight at 4°C. Anti-rabbit IgG (R-0881; Sigma) was added to a final dilution of 1:9, and the samples were incubated with shaking at RT for 2 hr. One mL of polyethylene glycol solution [1.2 M polyethylene glycol (P-6667; Sigma), 50 mM Tris, pH 8.6] was then added; the samples were incubated with shaking at RT for 15 min, then centrifuged at 4,000 × g for 10 min at 4°C. The supernatant was decanted, and the pellet counted in a Wizard 1470 Gamma Counter (PerkinElmer). The PRL concentration was then calculated and normalized to the CV values that estimated the number of cells in each well,^82,83 as described above. These measurements were done in at least 3 different experiments on different days.

Statistical analysis

Data from all experiments were analyzed by one-way analysis of variance (ANOVA) followed by multiple comparisons versus control group (Holm-Sidak method). The Sigma Stat 3 program (Systat Software, Inc.) was used for all statistical analysis, and significance was accepted at p < 0.05.

Abbreviations

Abs	=	antibodies
ANOVA	=	one-way analysis of variance
ATP	=	adenosine tri-phosphate
BSA	=	bovine serum albumin
cAMP	=	cyclic adenosine mono-phosphate
DMEM	=	Dulbecco's Modified Eagle Medium
E2	=	estradiol
ER	=	estrogen receptor
ERK	=	extracellular-regulated kinase
Gαi	=	G protein α subclass i
GTP	=	guanine nucleotide tri-phosphate
GTPγS	=	guanine nucleotide tri-phosphate γS
JNK	=	jun N-terminal kinase
mER	=	membrane estrogen receptor
PBS	=	phosphate-buffered saline
PRL	=	prolactin
R-eq	=	R-equol
S-eq	=	S-equol

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Dr. David Konkel (UTMB Institute for Translational Sciences) for critically editing the manuscript.

Funding

These studies were supported by NIH grant R01ES15292.