Estrogen Receptor β-Selective Transcriptional Activity and Recruitment of Coregulators by Phytoestrogens*

Estrogens used in hormone replacement therapy regimens may increase the risk of developing breast cancer. Paradoxically, high consumption of plant-derived phytoestrogens, particularly soybean isoflavones, is associated with a low incidence of breast cancer. To explore the molecular basis for these potential different clinical outcomes, we investigated whether soybean isoflavones elicit distinct transcriptional actions from estrogens. Our results demonstrate that the estrogen 17β-estradiol effectively triggers the transcriptional activation and repression pathways with both estrogen receptors (ERs) ERα and ERβ. In contrast, soybean isoflavones (genistein, daidzein, and biochanin A) are ERβ-selective agonists of transcriptional repression and activation at physiological levels. The molecular mechanism for ERβ selectivity by isoflavones involves their capacity to create an activation function-2 surface of ERβ that has a greater affinity for coregulators than ERα. Phytoestrogens may act as natural selective estrogen receptor modulators that elicit distinct clinical effects from estrogens used for hormone replacement by selectively recruiting coregulatory proteins to ERβ that trigger transcriptional pathways.

Estrogens are used in hormone replacement therapy (HRT) 1 to prevent hot flashes, urogenital atrophy, and osteoporosis in postmenopausal women (1,2). HRT also may prevent heart disease (3), Alzheimer's disease (4), and colon cancer (5). Unfortunately, HRT has not lived up to its potential to improve the health of women, because estrogens have been associated with an increased incidence of breast (6,7) and endometrial cancer (8). This relationship has hampered compliance with HRT severely and has sparked an intense pursuit for selective estrogen receptor modulators (SERMs) that have a safer profile (9,10). Recently, raloxifene has been approved for the prevention and treatment of osteoporosis (11). Raloxifene is classified as a SERM because it exhibits agonist activity in some tissues such as the bone (12,13) and acts as an antagonist in other tissues including the breast (14). Although these effects are extremely desirable, raloxifene also increases hot flashes (15), is weaker than estrogens at increasing bone mineral density (16), and does not improve cognitive function (17) or prevent hip fracture (13). Thus, the quest for superior SERMs for HRT continues to be intense.
There also is a growing interest in using dietary natural plant estrogens (phytoestrogens), particularly those found in soy products, as a potential alternative to the estrogens in HRT (18). Interest in phytoestrogens has been fueled by observational studies showing a lower incidence of menopausal symptoms, osteoporosis, cardiovascular disease, and breast and endometrial cancers in Asian women who have a diet rich in soy products (19 -24). Consistent with epidemiological studies are the findings that soy phytoestrogens prevent mammary tumors (25,26) and bone loss (27,28) in rodents and atherosclerosis of coronary arteries in monkeys (29). Soy protein relieves hot flashes in postmenopausal women (30) and attenuates bone loss in the lumbar spine of perimenopausal women (31). Furthermore, a high intake of dietary phytoestrogens is associated with a lower incidence of breast cancer in women (19). Many postmenopausal women are taking phytoestrogens in an effort to alleviate menopausal symptoms without increasing their risk of developing breast cancer. Moreover, many women with a history of breast cancer take phytoestrogens to control menopausal symptoms (32,33) because estrogens are contraindicated.
The isoflavones, genistein, daidzein, and biochanin A, which are abundant in soybeans (34) and available widely as herbal tablets, are especially popular among postmenopausal women. Despite their popularity and putative health benefits it is clear that we need to know much more about the molecular mechanisms, safety, and efficacy of isoflavones before they can be recommended to postmenopausal women as an alternative to estrogens for HRT. However, it is clearly important to elucidate the molecular mechanisms whereby isoflavones may elicit distinct clinical actions from estrogens used in HRT. Isoflavones have a structure similar to that of 17␤-estradiol (E 2 ) and are capable of binding to the two known estrogen receptors, ER␣ and ER␤ (35)(36)(37). Compared with ER␣, ER␤ exhibits a 7-30fold greater binding affinity for genistein, whereas E 2 binds to ER␣ and ER␤ with equal affinity (38,39). The relatively selective binding of genistein to ER␤ indicates that isoflavones may produce distinct clinical effects from estrogens by selectively triggering ER␤-mediated transcriptional pathways or differentially triggering transcriptional activation or repression pathways by ER␤.
To test this hypothesis, we compared the effects of isoflavones and E 2 on transcriptional repression and activation in the presence of ER␣ or ER␤. Our data demonstrate that isoflavones selectively trigger the transcriptional pathways of ER␤, particularly transcriptional repression. In addition to selectively binding to ER␤, our results suggest that the ER␤ selectivity of isoflavones involves their capacity to induce an activation function-2 (AF-2) surface of ER␤ that has greater affinity for coregulators such as glucocorticoid interacting receptor protein 1 (GRIP1) (40) compared with ER␣. Phytoestrogens may act as natural SERMs by selectively recruiting coregulators that trigger ER␤-mediated transcriptional pathways.

MATERIALS AND METHODS
Plasmids-Human ER␣ and ER␤ were provided by P. Chambon and J.-A. Gustafsson, respectively (41). Gal-GRIP1 and GST-GRIP1 were provided by M. Stallcup (42). Three copies of the Ϫ125 to Ϫ82 human TNF-␣ promoter fragment (43) or one copy of the ERE from the frog vitellogenin A2 gene (5Ј-TCAGGTCACAGTGACCTGA-3Ј; vitA2-ERE) were ligated into the polylinker upstream of Ϫ32 to ϩ45 herpes simplex thymidine kinase (tk) promoter linked to luciferase (TNF-RE tkLuc and ERE tkLuc, respectively). A synthetic oligonucleotide containing the 17-nucleotide Gal-responsive element (5Ј-CGGAGTACTGTCCTCCG-3Ј) was inserted in between the G and C of the AP-1-like site (5Ј-TGAGCTCA-3Ј) at the Ϫ105 to Ϫ95 region of the TNF-RE and cloned upstream of the Ϫ32 to ϩ 45 tk promoter (Gal-TNF-RE tkLuc).
Cell Culture, Transfection, and Luciferase Assays-U937, U2OS, MDA-MB-435, and MCF-7 cells were obtained from the cell culture facility at the University of California, San Francisco. U937 cells were maintained as described previously (44), whereas U2OS, MDA-MB-435, and MCF-7 cells were maintained and subcultured in phenol red-free Dulbecco's modified Eagle's medium/F-12 media containing 5% fetal bovine serum, 2 mM glutamine, 50 units/ml penicillin, and 50 g/ml streptomycin. For experiments, cells were collected, transferred to a cuvette, and then electroporated with a Bio-Rad gene pulser as described previously (41) using 3 g of reporter plasmid and 1 g of ER␣ or ER␤ expression vectors. After electroporation, the cells were resuspended in media and plated at 1 ml/dish in 12-well multiplates. The cells were treated with E 2 , genistein, daidzein, or biochanin A (Sigma-Aldrich) 3 h prior to exposure to 5 ng/ml TNF-␣ (R & D Systems) for 24 h at 37°C. Cells were solubilized with 200 l of 1ϫ lysis buffer, and luciferase activity was determined using a commercially available kit (Promega). The concentration of hormone required to produce a halfmaximal induction (EC 50 ) or inhibition (IC 50 ) of luciferase activity was calculated with the Prism curve-fitting program (Graph Pad Software, version 2.0b). For proliferation studies, parental MCF-7 cells were subcloned at 1 cell/well in the presence of 0.1 nM E 2 , and the fastest growing clone was selected for experiments. These cells expressed exclusively ER␣ as determined by reverse transcription polymerase chain reaction (RT-PCR). The cells were plated in duplicate at a density of 25,000 cells/35-mm plate in tissue culture medium containing 3% stripped fetal bovine serum. One day after plating they were treated with increasing concentrations of E 2 or genistein. The medium was changed every other day, and E 2 or genistein was added to the medium. After 8 days the cells were counted with a Coulter counter. All experiments presented in the figures were performed at least three times, and the data were similar between experiments.
MDA-MB-453 Stable Cell Line-The ER-negative human breast cancer cell line, MDA-MB-453 (45), was transfected by electroporation with pcDNA 6/V5-His (Invitrogen) vector containing human ER␣. The cells were maintained in 10 g/ml blastocidin (Invitrogen) until resistant colonies formed. Individual clones were obtained after the cells were plated into 96-well dishes at 1 cell/well in the presence of blastocidin. The expression of ER␣ and blastocidin-S deaminase, which confers resistance, was confirmed by RT-PCR in the clonal stable cell line.
Glutathione S-Transferase Pull-down Assays-GST pull-down assays were performed as described previously (46). Briefly, human ER␣ and ER␤ were transcribed and translated in vitro using the TNT T7 Quick Coupled Transcription/Translation system (Promega) and [ 35 S]methionine. For each binding reaction, a 2-l aliquot of translation product was incubated with Escherichia coli-expressed GST-GRIP1 immobilized to glutathione-Sepharose beads (Amersham Pharmacia Biotech) in the presence of vehicle control (0.1% ethanol), E 2 , or genistein. The samples were rocked gently at 4°C for 2 h. After extensive washing of the beads, the labeled proteins were eluted with SDSpolyacrylamide gel electrophoresis loading buffer and separated on a 12% SDS-polyacrylamide gel. The radiolabeled bound ERs were detected by autoradiography and analyzed using the Storm phosphorimaging system and ImageQuant software (Molecular Dynamics).

Estrogens Selectively Repress the TNF-␣ Promoter through
ER␤-To investigate the effects of isoflavones on transcriptional repression, we used the Ϫ125 to Ϫ82 region (43) of the TNF-␣ promoter (TNF-␣-responsive element, (TNF-RE)) because this region mediates TNF-␣ activation and E 2 repression (41). E 2 produced a profound dose-dependent repression of TNF-␣ activation of the TNF-RE upstream of a minimal tk promoter (TNF-RE tkLuc) with either transfected ER␣ (Fig.  1A) or ER␤ (Fig. 1B) in U937 cells. Daidzein and biochanin A had no effect on TNF-␣ activation of the TNF-RE with ER␣, whereas genistein produced a minor repression at 1 M (Fig.  1A). In contrast, all three isoflavones produced a large repression (30 -60%) of TNF-␣ activation of TNF-RE in the presence of ER␤ (Fig. 1B). Genistein is the most potent isoflavone and is about 65-fold weaker than E 2 at repression (IC 50 ϭ 8.5 versus 0.13 nM). The isoflavones are more effective also at triggering transcriptional activation of a classical estrogen response element (ERE) in U937 cells with ER␤ ( Fig. 2B) compared with ER␣ ( Fig. 2A). However, isoflavones are about 10 -300-fold more potent at triggering transcriptional repression compared with transcriptional activation with ER␤ (genistein, IC 50  Genistein Decreases TNF-␣ mRNA in Bone Cells-The effect of genistein on endogenous TNF-␣ gene expression was investigated in a human osteosarcoma cell line (U20S) because these cells express ER␣ and ER␤, as demonstrated by RT-PCR (data not shown), and TNF-␣ is involved in the pathogenesis of osteoporosis (47). U20S cells were treated with E 2 or genistein for 24 h and then exposed to TNF-␣ for 1 h. TNF-␣ produced a profound induction of TNF-␣ mRNA as determined by RT-PCR that was repressed markedly by E 2 or genistein (Fig. 3A). The observation that genistein inhibits endogenous TNF-␣ mRNA in untransfected cells demonstrates that repression of TNF-␣ transcription by genistein is physiological and not caused by nonspecific squelching of transcriptional factors by transfected ERs. To determine which ER isoform is responsible for repressing the endogenous TNF-␣ gene, we transfected U2OS cells with ER␣ or ER␤. Although the endogenous ERs are capable of repressing the native TNF-␣ gene, they are not present in high enough levels to repress the large number of transfected plasmids containing the TNF-RE. Fig. 3B shows that genistein is very effective at repressing the TNF-RE in cells transfected with ER␤ but not ER␣. These results indicate that genistein represses the endogenous TNF-␣ gene through ER␤ even though U2OS cells also express ER␣.
Isoflavones Are Weak ER␣ Agonists in Breast Cancer Cells-Our results indicate that isoflavones selectively promote ER␤mediated transcription. To explore the activity of genistein on ER␣ in breast cancer cell lines, we compared the effects of E 2 and genistein on ER␣ activation of ERE tkLuc in an ERnegative breast cancer cell line (MDA-MB-453) stably transfected with ER␣ and on the proliferation of MCF-7 cells, which express endogenous ER␣ but not ER␤ as determined by RT-PCR (data not shown). Similar to transiently transfected U937 and U20S cells, genistein is much weaker than E 2 at activating an ERE in the ER␣ MDA-MB-453 stable cells (Fig. 4A) and stimulating the proliferation of MCF-7 cells (Fig. 4B). Thus, genistein is a weak ER␣ agonist in cells transiently (U937 and U20S) or stably (MDA-MB-453) transfected with ER␣ and in cells that express endogenous ER␣ (MCF-7).
Isoflavones Selectively Recruit GRIP1 to ER␤-A potential explanation for ER␤-selective activity is that isoflavones induce a functional AF-2 surface in ER␤ but not ER␣ because we showed previously that the AF-2 surface is required for repression (41). Consistent with this hypothesis is the observation that an ER␤ with a mutation in helix 3 (K314A) of the AF-2 surface failed to promote repression in response to genistein (Fig. 5). Because binding of coregulatory proteins (48,49) to the AF-2 surface is required for repression by E 2 (41), we compared the effects of E 2 and genistein on functional interactions that occur at the TNF-RE between the coregulator, GRIP1 (40), and ER␣ or ER␤. For these studies, a Gal response element was inserted in the center of an AP-1-like site in the TNF-RE (Gal-TNF-RE), which is essential for TNF-␣ activation and E 2 repression (41). Gal-GRIP1 was used for these studies instead of Gal-ER because ERs do not bind directly to the TNF-RE (41) FIG. 1. Isoflavones selectively repress transcription of the TNF-RE through ER␤. Three copies of the Ϫ125 to Ϫ82 region of the TNF-␣ promoter were cloned upstream of the minimal tk promoter (TNF-RE tkLuc). U937 cells were transfected with 3 g of TNF-RE tkLuc and 1 g of expression vector for human ER␣ (A) or ER␤ (B). After transfection, the cells were treated for 24 h with TNF-␣ (5 ng/ml) in the presence of increasing concentrations of E 2 , genistein, daidzein, or biochanin A, and luciferase activity was measured. TNF-␣ activated the TNF-RE tkLuc by 5-10-fold in the absence of drugs. Each data point represents the mean of triplicate samples Ϯ S.E. RLU, relative luciferase units.

FIG. 2. Isoflavones selectively activate transcription of an ERE through ER␤.
A single copy of the vitellogenin A2 ERE upstream of the minimal tk promoter (ERE tkLuc) was transfected into U937 cells with either 1 g of expression vector for human ER␣ (A) or ER␤ (B). After transfection, the cells were treated for 24 h with increasing concentrations of E 2 , genistein, daidzein, or biochanin A, and luciferase activity was determined. Each data point represents the mean of triplicate samples Ϯ S.E. RLU, relative luciferase units. but may be tethered to the TNF-RE through coregulators such as GRIP1. Gal-GRIP1 activated Gal-TNF-RE tkLuc ϳ20-fold (data not shown). E 2 is extremely potent at inhibiting Gal-GRIP1 activation of Gal-TNF-RE tkLuc in the presence of either ER␣ (Fig. 6A) or ER␤ (Fig. 6B) (IC 50 ϭ 28.5 pM for ER␣, and IC 50 ϭ 1.5 pM for ER␤). In contrast, genistein is much more potent at repressing Gal-GRIP1 activation with ER␤ (IC 50 ϭ 49 pM) compared with ER␣ (IC 50 ϭ 1.8 M). Furthermore, at saturating levels (10 M), genistein produced a 33% repression with ER␣ compared with a maximal 72% repression with ER␤ at only 10 nM.
These results suggest that genistein creates an AF-2 surface in ER␤ that permits the binding of GRIP1 more efficiently compared with ER␣. To investigate this hypothesis directly, glutathione S-transferase-GRIP1 pull-down assays were performed with either 35 S-labeled ER␣ or ER␤ in the presence of E 2 or genistein. A similar dose-dependent increase in binding of ER␣ or ER␤ to GRIP1 was observed with E 2 (Fig. 7A). In contrast, genistein is more effective at enhancing the interaction between GRIP1 and ER␤ (Fig. 7B). At 10 M, binding of ER␤ to GRIP1 is 2-fold greater than with ER␣. These findings demonstrate that genistein creates an AF-2 surface in ER␤ that has a higher affinity for GRIP1 than that in ER␣. DISCUSSION Estrogens in HRT improve menopausal symptoms but are associated with an increased risk of breast (6, 7) and endometrial cancer (8). To overcome the uterotropic effects of estrogens, women with a uterus are treated also with progesterone in HRT regimens. Unfortunately, the addition of progesterone may increase the risk of breast cancer further (50,51) and attenuate potential benefits of estrogens on the cardiovascular system (52). The current challenge is to discover estrogens that retain their ability to prevent menopausal symptoms without promoting breast cancer or requiring progesterone for endometrium protection. The development of more ideal estrogens for HRT requires a greater understanding of how different estrogenic compounds differentially regulate gene activation and repression by ER␣ and ER␤.  3. A, genistein inhibits endogenous TNF-␣ gene expression in human osteosarcoma cells. U20S cells were treated with 10 nM E 2 , 1 M genistein (Gen) or ethanol (Cont) for 24 h and then exposed to TNF-␣ (5 ng/ml) for 1 h. Total RNA was isolated, and the expression of TNF-␣ mRNA was determined by RT-PCR. Ethidium bromide staining shows a 444-base pair PCR product of the TNF-␣ gene and a 217-base pair product for glyceraldehyde-3-phosphate dehydrogenase (GAPDH), which was used as internal control for the quality of RNA prepared. B, genistein selectively represses transcription of the TNF-RE in U2OS cells through ER␤. U2OS cells were transfected with TNF-RE tkLuc (3 g) and 1 g of expression vector for either ER␣ (f) or ER␤ (OE). The cells were treated for 24 h with increasing concentrations of genistein and then assayed for luciferase activity. Each data point represents the mean of triplicate samples. The S.E. was Ͻ 10%.

FIG. 4.
A, genistein is weak at activating an ERE in ER␣-stable MDA-MB-453 cells. ERE tkLuc (3 g) was transfected into MDA-MB-453 stably transfected with ER␣. Cells were treated for 24 h with increasing concentrations of E 2 or genistein, and luciferase activity was measured. B, genistein is weak at stimulating the growth of MCF-7 breast cancer cells. MCF-7 cells were plated at a density of 25,000 cells/35-mm plate in tissue culture medium containing 3% stripped fetal bovine serum. One day after plating they were treated with increasing concentrations of E 2 or genistein. After 8 days the cells were counted with a Coulter counter. Each data point represents the mean of duplicate samples. The S.E. was Ͻ 10%. RLU, relative luciferase units.

FIG. 5.
Genistein requires a functional AF-2 surface for repression by ER␤. U937 cells were transfected with 3 g of TNF-RE tkLuc and 1 g of human wild-type ER␤ or a helix 3 activation function-2 mutant (K314A in human ER␤530). Cells were treated for 24 h with TNF-␣ (5 ng/ml) in the absence or presence of 100 nM genistein, and luciferase activity was measured. Each data point represents the mean of triplicate samples Ϯ S.E. RLU, relative luciferase units.
We have shown that isoflavones elicit distinct transcriptional actions from estrogens. E 2 effectively triggers both ER␣and ER␤-mediated transcriptional activation or repression pathways. In contrast, our results demonstrate that isoflavones are weak ER␣ agonists and potent ER␤ agonists because they are effective only at triggering transcriptional activation or repression with ER␤. The key question is how do isoflavones elicit distinct transcriptional actions from estrogens despite the fact they both bind to the same binding pocket of ER␣ and ER␤ (53)(54)(55)? One possibility is that isoflavones bind to ER␤ more effectively than to ER␣. In fact, ER␤ has a 30-fold greater affinity for genistein compared with ER␣ (39). However, this difference in binding affinity is unlikely to account entirely for the distinct transcriptional actions of isoflavones because we observed that isoflavones were over a 1,000-fold more potent at triggering transcriptional activity with ER␤ compared with ER␣. Furthermore, at saturating levels (10 M), genistein was less effective at repressing GRIP1 activation of Gal-TNF-RE tkLuc with ER␣ and recruiting GRIP1 to ER␣, compared with ER␤. These studies indicate that the divergent transcriptional actions of estrogens and isoflavones probably also result from differences in their ability to recruit coregulators and trigger transcriptional functions of ER␣ or ER␤. These data are consistent with the discoveries that coregulator proteins (48,49) are required for both transcriptional activation and repression by ERs (41,56,57). E 2 nonselectively recruits coregulators to ER␣ and ER␤, whereas isoflavones selectively recruit coregulators to ER␤. By recruiting coregulators such as GRIP1 to both ERs, E 2 effectively triggers transcriptional activation and repression pathways for both ER␣ and ER␤. Undoubtedly, E 2 elicits its full spectrum of beneficial and adverse effects by triggering all transcriptional pathways of ERs. In contrast, at physiological FIG. 6. Genistein selectively represses Gal-GRIP1 activation of Gal-TNF-RE through ER␤. A synthetic oligonucleotide that contained the 17-nucleotide Gal-responsive element was inserted in between the G and C of the AP-1-like site (5Ј-TGAGCTCA-3Ј) at the Ϫ105 to Ϫ95 region of the TNF-RE. Because the AP-1-like site is destroyed, this construct is inactive in the absence of Gal-GRIP1. Cells were transfected with Gal-GRIP1, Gal-TNF-RE tkLuc, and ER␣ (A) or ER␤ (B) and then treated with increasing concentrations of E 2 or genistein for 24 h, and luciferase activity was determined. Gal-GRIP1 activated Gal-TNF-RE tkLuc by about 20-fold in the absence of E 2 or genistein. Each data point represents the mean of triplicate samples Ϯ S.E. RLU, relative luciferase units.

FIG. 7.
Genistein is more effective at recruiting GRIP1 to ER␤. [ 35 S]Methionine-labeled ER␣ or ER␤ was synthesized in an in vitro transcription/translation system and then incubated with E. coli-expressed glutathione S-transferase-GRIP1 in the presence of increasing concentrations of E 2 (A) or genistein (B). [ 35 S]methionine-labeled ER␣ or ER␤ bound to glutathione S-transferase-GRIP1 was separated by SDS-polyacrylamide gel electrophoresis. The dried gels were exposed to x-ray film (top) and a phosphorscreen and then scanned with a Phos-phorImager (bottom). The first lane on the autoradiograms represents the input of [ 35 S]methionine-labeled ER␣ or ER␤ in the binding reaction. This is a single experiment that is representative of six experiments that produced similar results. levels (0.55-0.86 M) (58) genistein is very weak at recruiting GRIP1 to ER␣, but it is potent at recruiting GRIP1 to ER␤. By selectively recruiting coregulators to ER␤, isoflavones would only trigger ER␤-mediated transcriptional pathways. These results suggest that isoflavones should be effective at eliciting the clinical effects that are mediated by ER␤ but not ER␣. Moreover, isoflavones are 10 -300-fold more potent at triggering transcriptional repression compared with activation. These results indicate that it may be possible to develop transcriptional activation or repression-selective estrogens for HRT. It is unclear why genistein recruits GRIP1 more effectively to ER␤ than to ER␣. However, the binding of GRIP1 may stabilize the genistein-ER␤ complex more effectively than the genistein-ER␣ complex because the binding of coregulators has been shown to slow the rate of dissociation of an agonist from the ER-coregulator complex (59).
The lack of regulation of ER␣-mediated genes and the potent repression of ER␤-mediated genes by isoflavones may account for the low incidence of menopausal symptoms, osteoporosis, cardiovascular disease, and breast and endometrial cancer in Asian countries (19,(21)(22)(23)(24). For example, our studies suggest that ER␤-mediated repression of the TNF-␣ gene may be an important mechanism whereby isoflavones may prevent osteoporosis because excessive production of TNF-␣ after menopause is thought to lead to osteoporosis (47). Genistein also protects against vascular injury in ovariectomized female rats through ER␤ (60). We have shown also that E 2 produces a robust stimulation of proliferation of breast cancer (MCF-7) cells. ER␣ undoubtedly mediates this effect because these cells only express ER␣. Furthermore, it is likely that ER␣ mediates the proliferative effects on endometrial cells because these cells do not express ER␤ (61). Based on these findings, we hypothesize that ER␤-selective estrogens such as isoflavones may prevent some menopausal symptoms and conditions but will be less likely to elicit stimulatory effects on breast and endometrial cells compared with estrogens present in current HRT regimens that also trigger ER␣-transcriptional pathways. Consistent with this hypothesis are the observations that isoflavone-rich soy protein relieves menopausal symptoms (30, 62) but does not exert estrogenic effects on the endometrium in postmenopausal women (62,63). Furthermore, isoflavone-rich soy protein does not induce proliferation in endometrial and mammary tissue in postmenopausal female macaques (64).
Understanding how natural estrogens and synthetic SERMs elicit selective clinical effects is a key to the development of safer estrogens for HRT. We have shown that isoflavones elicit distinct transcriptional actions from estrogens by selectively recruiting coregulators to ER␤. These data are consistent with the observation that helix 12 of the AF-2 surface exists in a different position when genistein is bound to ER␤ (55) compared with E 2 -bound ER␣ (53) or ER␤ (55). Our results suggest that isoflavones may act as natural SERMs, which may be safer than estrogens in current HRT regimens because they selectively trigger the transcriptional pathways of ER␤. Estrogens in HRT also trigger ER␣ transcriptional pathways, which may promote the proliferation of breast and endometrial cells.