Regulation of specific target genes and biological responses by estrogen receptor subtype agonists.

Another potential way to make estrogens safer for drug therapy is to add a second drug to alter the biological properties of estrogens after they interact with ERα. We screened plant extracts and found that a chalcone derivative dramatically changed the gene expression profile by E 2 in U2OS cells expressing ERα. We termed the chalcone an ERα coagonist ( Figure 1E ), because it was inactive by itself, but it caused E 2 to regulate genes that it did not activate in its absence and it potentiates the regulation of E 2 on some genes. The coagonist blocked E 2 -mediated proliferation of MCF-7 cells, suggesting that the coagonist changes the proliferative response of E 2 by causing ERα to regulate a different set of genes. While the mechanism of the coagonist is unclear, our studies suggest the possibility that it binds to ERα as heteroligand with one subunit binding to E 2 and the other subunit binding to the chalcone ( Figure 1E , right panel). The combination of two different ligands bound to ERα simultaneously likely produces different conformation then when ERα is bound to only E 2 ( Figure 1E , left panel) or the chalcone. While the effects of the coagonist on E 2 -mediated bone loss, weight loss, and mammary gland and endometrial cell proliferation in animals need to be investigated, it may be possible that coagonist compounds can alter the clinical responses to estrogens and make them safer.

Estrogens have important actions in non-reproductive tissues, including the brain, urogenital tract and bone. Because of their actions in these tissues, estrogens have been used for over 50 years to prevent and treat a variety of conditions affecting postmenopausal women, including hot flashes, urogenital atrophy and osteoporosis. Estrogens would be the clear drug of choice for treating menopausal symptoms if they did not cause some serious adverse effects. The most troublesome side-effect of estrogens is the increased risk of breast and endometrial cancer [ 1 , 2 ]. Estrogens also increase blood clotting which can lead to venous thromboembolisms, and possibly strokes and heart disease, particularly in older women [ 1 ]. Estrogens in hormone therapy (HT) were formulated long before there was a significant understanding of the mechanism of action of estrogens. The identification of ERα and ERβ ( Figure 1 ) and the crystal structures of their ligand binding domain (LBD), the discovery of variety of coregulatory proteins involved in the genomic pathway and the demonstration of the nongenomic actions of estrogens [ 3 , 4 ] provide an extraordinary opportunity to design a new generation of estrogens that are safer and more selective. Estrogen receptor subtype agonists (ERSAs) [ 5 – 10 ] have been identified ( Figure 2 ) which might represent new classes of drugs to treat menopausal conditions. Here we will review ERα and ERβ regulation of genes and the actions of several ERSAs and their potential clinical applications.

It is well established that estrogens exert tissue-specific effects, but the mechanism is unclear. Tiling arrays identified 1,090 ERα binding sites on chromosomes 1 and 6 in MCF-7 cells whereas 1,137 binding sites were found in U2OS cells [ 46 ]. Only 172 ERα binding sites were common to both cell types. The cell specific recruitment of ERα is mediated by the binding of the pioneer factor, FoxA1 that recognizes monomethylated and dimethylated histone H3. Once FoxA1 recognizes these methylated histones near an ER binding site it interacts with ER to open up chromatin structure and facilitate the recruitment of transcription factors leading to increased transcription [ 46 ]. Because FoxA1 is expressed in MCF-7 cells, but not U2OS cells, the genes regulated by ERα are different [ 46 ]. These findings suggest that it might be possible to design tissue selective ERα modulators that mimic the agonist activity of E 2 in some tissues, but not in other tissues. We identified two plant extracts (PEs), Radix Glycyrrhiza and Radix Pueraria that behave as tissue selective ERα agonists ( Figure 1D ). These PEs activate ERα in transfection assays using an ERE upstream of the luciferase reporter and bind to purified ERα (In preparation). To test the effects of the PEs on weight loss, ovariectomized mice were fed a high fat diet (HFD). After the mice gained weight, they were treated orally for 6 weeks with the PEs separately while being maintained on the HFD. The vehicle treated control mice continued to gain weight, whereas the E 2 -treated mice, which served as positive controls, lost 20.5% of their weight. The body weight and abdominal fat of both PE treated mice was significantly reduced to levels similar to mice treated with E 2 . In contrast, no significant proliferative effects were found in the mammary gland and uterus. While further characterization and studies are needed with the PEs these studies suggest that it might be possible to develop tissue selective ERα agonists that retain the beneficial effects mediated by ERα without promoting breast and endometrial cancer.

Gene expression data, tiling arrays and ChIP-seq data shows that ERα and ERβ regulate different genes by binding to distinct regulatory elements and interacting with different coactivators and transcription factors. Animal studies demonstrated that ERα- and ERβ-selective agonists produce different biological effects. Three classes of ERβ-selective agonists have been identified; ERβ binder, ERβ activator and ERβ binder/activator. ERβ-selective agonists might be clinically useful for preventing breast cancer and treating hot flashes and inflammatory conditions associated with menopause. Because the proliferative effects of estrogens are mediated through ERα, the impetus to design ERα-selective agonists for clinical use has not been strong as ERβ-selective agonists. However, ERα is clearly important for preventing osteoporosis, weight gain and insulin resistance. Tissue selective ERα agonists or ERα coagonists may provide a safer approach if proven to activate ERα in tissues that are beneficial, such as the bone and adipose tissue, but not the mammary gland and uterus. While many additional studies are needed evaluate the safety and efficacy of ERα- and ERβ-selective agonists they offer a new therapeutic approach for preventing and treating specific menopausal conditions.

ERs are composed of three major modular domains; an A/B domain, a DNA binding domain (DBD), and a LBD. Several features differ between ERα and ERβ that might be important for designing ERSAs. First, the size of the ERα and ERβ binding pocket for ligands are different providing a structural basis for designing ligands that selectively bind to each ER. Second, the two activation function (AF-1 and AF-2) domains that are responsible for regulating gene transcription are located in the least homologous regions ( Figure 1 ). The A/B domain contains AF-1 has only 17% homology, whereas the LBD which contains the AF-2 is 55% homologous. Differences in AF-1 and AF-2 could allow drugs to be designed that recruit different cofactors to ERα and ERβ, and thereby causing a different pattern of genes regulated. ERα and ERβ have distinct cellular actions, which provide a rationale for developing ERSAs. This has been demonstrated with microarrays that showed ERα and ERβ regulate different genes [ 11 – 14 ]. Only 40% of genes regulated by estradiol (E 2 ) in U2OS cells that express ERα are also regulated by ERβ [ 12 ]. Furthermore, ERα and ERβ regulate different classes of genes suggesting that the two ERs have distinct physiological roles. Another feature that distinguishes ERβ from ERα is that ERβ regulates three classes of genes, whereas ERα regulates a single class of genes [ 15 ]. U2OS cell lines stably transfected with a doxycycline-inducible ERα or ERβ [ 15 ] were used to measure the effects of unliganded ER in cells treated only with doxycycline or liganded-ER when cells were treated with both doxycycline and E 2 . Unliganded ERα produced a small upregulation of only 1 gene and downregulation of 3 genes, whereas the addition of E 2 to doxycycline treated U2OS-ERα cells resulted in the activation of 518 genes and repression of 157 genes. These data indicate that ERα requires the ligand to regulate gene transcription in U2OS cells. In contrast, three classes of genes were regulated in U2OS-ERβ cells. 453 genes were regulated by unliganded ERβ (Class I genes). 258 genes were not regulated by unliganded ERβ, but regulated by E 2 -bound ERβ (Class II genes). 83 genes were regulated by unliganded ERβ and potentiated by the addition of E 2 (Class III genes). The unliganded effect of ERβ is mediated by AF-2, because it is lost when the ERβ AF-2 is replaced by the ERα AF-2 [ 16 ]. These results demonstrate that intrinsic differences in AF-2 of ERα and ERβ can lead to a different set of genes regulated.

The major concern for developing ERα agonists is that they will cause cell proliferation and increase the risk of cancer. In fact, PPT stimulates the proliferation of HC11 mouse mammary epithelial cells [ 45 ] and increases uterine weight in rats [ 42 ]. These findings indicate that ERα-selective binders ( Figure 1C ), like PPT might not be useful drugs for hormone therapy. Another strategy would be to design tissue selective ERα agonists that activate ERα in some tissues, such as the bone and adipose tissue, but not in the mammary gland and uterus ( Figure 1D ). An alternative strategy is to combine estrogens with other compounds that block the proliferative effects of estrogens in the mammary and uterus ( Figure 1E ). Progestins are effective at blocking the proliferative effects of estrogens in the uterus, but unfortunately they exacerbate the proliferative effects in the mammary gland.

One important action of estrogens that is relatively unappreciated is their anti-inflammatory effects. A number of diseases during menopause have an inflammatory component to their pathogenesis. These conditions include osteoporosis, cardiovascular disease, Alzheimer’s disease, obesity and atrophic vaginitis. Estrogens in HT are very effective at preventing osteoporosis and atrophic vaginitis, but controversy exists regarding their effects on cardiovascular disease, obesity and Alzheimer’s disease. The anti-inflammatory action of ERB-041 have been examined in multiple inflammatory rodent models, including endometriosis, rheumatoid arthritis, inflammatory bowel and sepsis [ 6 , 37 , 38 ]. These studies demonstrated that ERB-041 was very potent at blocking inflammation in these models and suggested that ERβ-selective agonists might be important drugs to treat a variety of disorders associated with inflammation. MF101 and synthetic ERβ agonists, including ERB-041 are potent repressors of pro-inflammatory genes [ 8 , 39 ], indicating that estrogens can produce anti-inflammatory actions through ERβ. The effects of ERβ on inflammatory conditions associated with menopause, such as osteoporosis, obesity, cardiovascular disease and atrophic vaginitis is unclear. ERB-041 did not prevent ovariectomy-induced bone loss or weight gain in rats [ 6 ], suggesting that ERα mediates these effects. DPN decreases the size of infarcts in mouse hearts subjected to ischemia and reperfusion similar to E 2 [ 40 ]. This cardioprotective effect of DPN was abolished in ERβ knockout mice [ 40 ]. These findings indicate that ERβ agonists might be useful for preventing cardiovascular disease. Another possible clinical indication for ERβ agonists, where an anti-inflammatory effect could be therapeutic is atrophic vaginitis. Our pre-clinical studies with mice indicate that ERβ agonists may play a role in the treatment of postmenopausal vaginal atrophy and dryness.

Multiple ERβ-selective agonists have been synthesized [ 10 ]. 2,3-bis(4-hydroxyphenyl)-propionitrile (DPN) has 70-fold higher relative binding affinity and 170-fold higher relative potency in transfection assays with ERβ compared to ERα [ 7 ]. Wyeth synthesized a number of ERβ-selective compounds [ 29 ]. ERB-041 has been the most studied. It has over a 200-fold greater selectivity for binding to ERβ compared to ERα [ 6 ]. In addition to synthetic compounds, a plant extract, MF101 contains ERβ-selective agonists [ 8 ], several of which have been identified, including liquiritigenin and nyasol [ 9 , 13 ]. Based on binding and functional studies, we proposed that these compounds can be grouped into three classes [ 13 ] ( Figure 1 ). One class is represented by ERB-041 which is selective because it binds to ERβ at a much higher affinity than ERα ( Figure 1A ). We termed it an ERβ binder. MF101, LIQ and nysasol bind to both ERα and ERβ similarly, but they only activate ERβ [ 13 ]. When these compounds bind to ERα they produce an inactive conformation that prevents ERα from forming a functional complex and recruiting coactivators [ 8 , 9 ] ( Figure 1B ). These are termed ERβ activators. DPN is selective because it binds ERβ with higher affinity, but also more potently activates ERβ than ERα. We termed it an ERβ binder/activator. While most genes regulated by DPN, ERB-041, MF101, LIQ and nyasol are the same, these three classes of ERβ agonists regulate some different genes [ 13 ]. Importantly, many genes regulated by these ERβ agonists in U2OS-ERβ cells are distinct from those regulated by E 2 . This observation is consistent with the finding that ERβ binding sites are different when it is bound to ERB-041 compared to E 2 in MCF-7 cells [ 17 ]. From these results, it can be expected that different classes of ERβ agonists will produce different biological and clinical effects from one another and non-selective estrogens used in HT.