|Home | About | Journals | Submit | Contact Us | Français|
After the Women’s Health Initiative found that the risks of hormone therapy outweighed the benefits, a need for alternative drugs to treat menopausal symptoms has emerged. We explored the possibility that botanical agents used in Traditional Chinese Medicine for menopausal symptoms contain ERβ-selective estrogens. We previously reported that an extract containing 22 herbs, MF101 has ERβ-selective properties. In this study we isolated liquiritigenin, the most active estrogenic compound from the root of Glycyrrhizae uralensis Fisch, which is one of the plants found in MF101. Liquiritigenin activated multiple ER regulatory elements and native target genes with ERβ but not ERα. The ERβ-selectivity of liquiritigenin was due to the selective recruitment of the coactivator steroid receptor coactivator-2 to target genes. In a mouse xenograph model, liquiritigenin did not stimulate uterine size or tumorigenesis of MCF-7 breast cancer cells. Our results demonstrate that some plants contain highly selective estrogens for ERβ.
Since the average age of the population in the United States is rising, there are an increasing number of menopausal women. Menopause is often associated with hot flashes, night sweats, mood changes, urogenital atrophy and loss of bone density that have traditionally been treated with hormone therapy (HT) to restore estrogen levels. The Women’s Health Initiative (WHI) trial, however, found that estrogen plus progestin increased a woman’s risk of heart disease, breast cancer, and dementia [1–5]. In addition, a second arm of the WHI found that using estrogen alone increased the risk of stroke and dementia [6,7]. While some of the unfavorable results on heart disease from the WHI are likely due to the late time when HT was initiated in relation to the onset of menopause  the adverse impact of HT on breast cancer and blood clots indicated that new strategies are needed to treat menopausal symptoms.
Alternative drugs to traditional HT could potentially include selective estrogen receptor modulators (SERMs), such as tamoxifen and raloxifene. However, while current SERM therapy has some favorable effects, such as improved bone mineral density [9,10] and the prevention of breast cancer, SERMs exacerbate hot flashes . Other pharmacological options for hot flashes include antidepressant therapy using serotonin and norepineprhine reuptake inhibitors, as well as other neuro-modulators, such as gabapentin . However, the overall benefit of these treatments is unclear considering their moderate efficacy , potential significant side effects [14,15] and lack of benefits on other menopausal symptoms, such as vaginal atrophy and osteoporosis.
Many patients rely on botanical dietary supplements (BDS) used in Traditional Chinese Medicine (TCM) to relieve their menopausal symptoms. It has been reported that about 25% of women use botanical extracts to treat menopausal symptoms . Recently, there has been an increased interest in using isoflavones, which are one of several classes of phytoestrogens, as an alternate therapy for menopausal symptoms. Daidzein and genistein, the major isoflavones found in soy products, have been studied at length with inconclusive results from clinical trials. Some studies with daidzein have shown modest improvement in menopausal symptoms as compared to placebo [17,18], while a large meta-analysis did not show that daidzein-rich isoflavones improved symptoms over placebo . Similarly, studies with genistein, have shown inconclusive results [19,20]. The selective estrogen receptor (SERM) DT56a is an enzymatic isolate of soybeans that has been shown to improve vasomotor symptoms and increased bone mineral density in post-menopausal women, with no effect on sex hormone levels or endometrial thickness [21–23]. While these data suggest that phyto-SERMs from soy have promise to safely treat menopausal symptoms and osteoporosis their effectiveness still needs to be evaluated in larger placebo-controlled randomized trials.
In addition to soy, it has been found that Chinese herbs contain compounds that have estrogenic activity [24,25]. We have been performing basic and clinical research on one botanical extract, MF101, which is composed of 22 individual plants . A Phase 2 clinical trial with 217 postmenopausal women found that 5 gm and 10 gm MF101 were safe for short term use and more effective than placebo after 12 weeks of treatment (data not shown). We also demonstrated that MF101 acted as an ERβ-selective agonist by regulating gene transcription via ERβ pathways . In addition, we showed that MF101 does not stimulate MCF-7 breast cancer cell proliferation or uterine growth in a mouse xenograft model . These findings suggested that plants used in TCM might be a source for the discovery of estrogen receptor subtype selective drugs to safely treat menopausal symptoms.
The estrogens used in current HT regimens activate both known estrogen receptor subtypes, ERα and ERβ. Although the precise roles of both ERs are not known, the specific activation of each subtype results in different biological outcomes. ERα and ERβ knockout mice have different phenotypes . In addition, estradiol (E2,) activation of ERα versus ERβ results in different gene regulation patterns . Estrogen acts as an agonist on ERα and ERβ in all tissue types, which likely explains the beneficial aspects of HT, but this non-selective action also likely causes the adverse side effects unveiled by the WHI. On the other hand, drugs that selectively activate ERα or ERβ might mimic some of the beneficial effects while avoiding the untoward effects. Since ERα has been shown to cause the proliferation of breast cancer cells and ERβ has been demonstrated to be a tumor suppressor [28,29], it is conceivable that ERβ-selective agonists may serve as safer long-term alternative treatment to traditional HT.
Due to the large unmet need encountered as a result of the WHI, we have developed a parallel strategy to discover drugs for the treatment of menopausal symptoms. Our first approach was to determine the mechanism of action of standardized crude plant extracts and to test them in controlled clinical trials. To this end, we showed that MF101 had ERβ-selective properties  and a Phase 2 clinical trial with MF101 for the treatment of hot flashes was done to further evaluate its safety and efficacy (http://clinicaltrials.gov/show/NCT00119665). Our next step was to isolate and characterize individual chemical entities from MF101 as potential drugs for menopausal symptoms. This paper summarizes the results from a study representing our second approach as we isolated an ERβ-selective compound, liquiritigenin, from the root of G. uralensis, and determined its biologic activity on estrogen receptors in cells and animal models.
Dry, powdered G. uralensis roots were extracted with 9:1 water—methanol (18 h, constant mixing) at a 10:1 solvent to mass ratio. The filtrate was recovered after suction filtration (Whatman #1 filter), concentrated by rotary evaporation to remove the methanol, and partitioned with an equal volume of ethyl acetate (repeated once). The combined ethyl acetate layers were dried with anhydrous sodium sulfate, concentrated to dryness by rotary evaporation in vacuo, and resuspended in a small volume of ethyl acetate. The sample was loaded onto a fritted glass column packed with silica gel (200–400 mesh, 60Å) and eluted with a hexane/ethyl acetate gradient, starting with 100% hexane. Liquiritigenin eluted from the silica column with 60–80% ethyl acetate in hexane. The liquiritigenin fractions recovered off the silica column were further purified by preparative reverse phase HPLC (Delta 600 system, Waters Corporation, Milford, MA) on a C8 column (SymmetryPrep, 19 × 150 mm, Waters Corporation) with UV detection (λ = 254 nm). A gradient elution from 35–40 % acetonitrile in water over 15 min at a flow rate of 12 mL/min was utilized to isolate liquiritigenin. Mass spectrometry analysis was performed on a HP 1100 LC/MS (Agilent Technologies, Santa Clara, CA), and yielded the expected molecular ion of m/z 255 [M-H]−. The 1H and 13C spectra were recorded on a Bruker 500 MHz nuclear magnetic resonance spectrometer (Bruker, Fallanden, Switzerland). NMR spectra were acquired in methanol-d4 and were consistent with published data for liquiritigenin isolated from Glycyrrhiza species .
The human osteosarcoma U2OS cell line, human breast MCF-7 cancer cell line, and human cervical HeLa cancer cell line were obtained from the cell culture facility at the University of California, San Francisco. The MCF-7 cells express ERα, whereas the U2OS and WAR5 do not express endogenous ERs. The prostate cancer cell line, WAR5 cells was prepared as previously described . All cell lines were maintained and subcultured as previously described . Transfections were carried out with a Bio-Rad gene pulser. Cells were electroporated with 3 μg of ERE, CECR6, NKD, or NKG2 tk-Luciferase reporter vectors along with 1 μg of ERα or ERβ expression vectors. After electroporation, the cells were plated and treated with estradiol (E2), liquiritigenin (Liq), 2,3-bis(4-Hydroxyphenyl)-propionitrile (DPN, Tocris Bioscience, Ellisville, Missouri) or 4,4′,4″-(4-Propyl-[1H]-pyrazole-1,3,5-triyl)trisphenol (PPT, Tocris Bioscience, Ellisville, Missouri) for 18 h. Cells were then solubilized and luciferase activity was determined with an assay kit (Promega, Madison, WI). Experiments were performed at least three times and the mean ± S.E. was calculated and statistical analysis was performed using the Prism curve-fitting program (Graph Pad Software, version 3.03).
U20S cells expressing a tetracycline-inducible ERα or ERβ cDNA were prepared as previously described . Cells were treated with doxycycline (1 μg/ml) for 18 hours and then with E2 or liquiritigenin for 3 hours. Total RNA was isolated using the Aurum Total RNA kit (Bio-Rad, Hercules, CA) and reverse transcription (RT) reactions were performed using an iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA). Real-time quantitative PCR was performed using SYBR Green Supermix with an iCycler thermal cycler (Bio-Rad). Experiments were performed at least three times and the mean ± S.E. was calculated and statistical analysis was performed using the Prism curve-fitting program (Graph Pad Software, version 3.03). We used the following PCR primers:
The relative binding affinity of liquiritigenin to pure full-length ERα and ERβ was determined using ERα and ERβ competitor assay kits, according to the manufacturer’s instructions (Invitrogen Life Technologies, Carlsbad, CA). Fluorescence polarization of the fluorophore-tagged estrogen bound to ERα and ERβ in the presence of increasing amounts of competitor ligand or extract was determined (10 readings per well; 0.02 millisecond integration time; G factor = 1.1087) using the GENios Pro microplate reader (Tecan Systems Inc., San Jose, CA) with fluorescein excitation (485nM) and emission (530 nM) filters. Each liquiritigenin dose was performed in triplicate and the relative error was determined by calculating the standard error of three values from the mean.
Following treatment with liquiritigenin or E2, stably transfected U2OS-ERα and U2OS-ERβ cells were fixed with 1% formaldehyde and ChIP was done as previously described . Immunoprecipitations were performed overnight at 4 °C with anti-SRC-2 (ab9261, Abcam, Cambridge, MA) antibodies. DNA fragments were purified (QIAquick PCR Purification Kit, Qiagen, Valencia, CA) and PCR-amplified. The primers used for ChIP are: CECR6 forward 5-TGATAAATGCTAGTGAGGTGCC-3, reverse 5-AGAACCGCCTGCTCCTAACAAT-3. NKD forward 5-GGGTCAGGACGAGTGTTTTCTT-3, reverse 5-ACCCCGGACCAAATTTCAGTTA-3. NKG2E Forward 5′-AGCCACCCAAAGTCTCCTAT-3′; Reverse 5′-TTCAGTGGAGAGGTCAGGTT-3′. PCR reactions for non-immune assays served as negative controls (data not shown).
MCF-7 (250,000) cells were aggregated in suspension and then resuspended in 200 μL neutralized collagen, as previously described . The cells were then grafted under the kidney capsule of nude mice as described and illustrated in detail at: http://mammary.nih.gov/tools/mousework/Cunha001/index.html. Five mice per group were treated with a continuous infusion using osmotic pumps (Alzet, Cupertino, CA) containing vehicle, E2 (0.4 mg) or liquiritigenin (2 mg) that infused 2.5 μl/h for 1 month. After one month of treatment, the tumors and uteri were removed and analyzed. These animal studies were carried out with approval from the University of California, San Francisco Committee on Animal Research.
Our previous results showed that MF101 contains ERβ-selective activity . The 22 herbs constituting MF101 were individually screened for estrogenic activity in transfection assays. Of the 22 herbs, G. uralensis contained high estrogenic activity. Activity-guided isolation of the compounds from the G. uralensis was initiated using ERE tk-Luc and an expression vector for ERβ. These studies resulted in the identification of the flavanone, liquiritigenin (Fig. 1). To assess the relative activity of liquiritigenin via ERα or ERβ, we used transfection assays with increasing concentrations of liquiritigenin. Liquiritigenin produced a dose-response activation of ERE tk-Luc in the U2OS cells transfected with ERβ, but not ERα (Fig. 2A). The activation first occurred at 1 nM and the maximal activation was observed at 500 nM. The EC50 of liquiritigenin (36.5 nM) for the activation of ERE tk-Luc was about 80-fold less than the EC50 of E2 (0.45 nM) (Fig. 2B). The ER subtype selectivity of liquiritigenin was compared to the ERβ-selective agonist, DPN, and the ERα agonist, PPT. Whereas no activation of ERE tk-Luc by ERα was observed with liquiritigenin, DPN produced a 20-fold activation and PPT produced a 74-fold activation, which was similar to the activation by E2 (Fig. 2C, left panel). In contrast, E2, liquiritigenin and DPN produced a similar activation of ERE tk-Luc with ERβ, whereas PPT had no effect (Fig. 2C, right panel). The activation of ERE tk-Luc by liquiritigenin in the presence of ERβ was blocked by the ER antagonists, ICI 182780, raloxifene and tamoxifen, demonstrating that the activation is mediated by ERβ (Fig. 2D). The ERβ-selectivity of liquiritigenin was also observed in HeLa cells and the prostate cancer WAR5 cell line (Fig. 2E, 2F, respectively). Liquiritigenin did not activate other nuclear receptors including the androgen receptor, progesterone receptor B, glucocorticoid receptor or thyroid hormone receptor in transfection assays (Fig. 3A, 3B, 3C and 3D, respectively). These results demonstrate that liquiritigenin is more selective than DPN with ERβ in U2OS cells and is an ERβ-selective agonist in multiple cell lines.
In addition to the traditional ERE, it is important to determine if liquiritigenin selectivity activates ERβ in elements derived from native ER target genes. We used ER regulatory elements from the cat eye syndrome chromosome region candidate 6 (CECR6), killer cell lectin-like receptor (NKG2E) and the naked cuticule homolog (NKD) genes that are activated by E2 [35,36]. The CECR6 element has an ERE and the NKG2E regulatory region contains a composite element containing a c-jun, heat-shock factor 2, and CCAAT/enhancer-binding protein beta and a unique variant ERE [35,36]. The NKD element does not contain an ERE or other known alternative regulatory elements . These elements were transfected into U2OS cells with expression vectors for ERα or ERβ. Liquiritigenin produced a dose-dependent activation of the CECR6 (Fig. 4A), NKG2E (Fig. 4B) and NKD (Fig. 4C) with ERβ but not with ERα. To determine if liquiritigenin exhibits ERβ-selectivity on native genes, we examined its effect in the previously characterized U2OS cell lines that are stably transfected with ERα or ERβ. Liquiritigenin produced a time-dependent increase in CECR6 (Fig. 4D), NKG2E (Fig. 4E) and NKD (Fig. 4F) mRNA in the U2OS-ERβ cells, but not the U2OS-ERα cells. These studies demonstrate that liquiritigenin is an ERβ-selective agonist on multiple ER regulatory elements and native target genes.
Estrogen receptor ligands have been shown to have different affinities for ERα or ERβ. For example, E2 binds with equal affinity to both ERα and ERβ, whereas some phytoestrogens, such as the isoflavone genistein bind with a higher affinity to ERβ compared to ERα[37,38]. One of the possible mechanisms for the ERβ-selectivity of liquiritigenin is that it binds with higher affinity to ERβ than to ERα. However, competition binding curves show that ERβ only has only a 20-fold greater affinity for liquiritigenin compared to ERα (Fig 5A), which is not likely sufficient to explain the differences in transcriptional regulation. A more plausible explanation for the ERβ-selectivity is that liquiritigenin recruits coactivators only to ERβ. To test this hypothesis, U2OS-ERα and U2OS-ERβ cells were incubated with liquiritigenin for increasing times and then ChIP was done with an antibody to the coactivator SRC-2. We chose to focus on SRC-2 because we previously showed that E2 recruited only SRC-2 to multiple ER regulatory elements, such as NKG2E, CECR6 and NKD genes . Liquiritigenin caused the recruitment of SRC-2 to the CECR6 (Fig 5B), NKG2E (Fig 5C), and NKD (Fig 5D) genes in the U2OS-ERβ cells, but not the U2OS-ERα cells. These results demonstrate that liquiritigenin acts as an ERβ-selective agonist because it only recruits coactivators to ERβ.
The major concern with estrogens for menopausal symptoms is the proliferation of breast and endometrial cells causing an increased risk for breast and uterine cancer. To determine if liquiritigenin has a proliferative effect on breast cancer cells, we grafted MCF-7 breast cancer cells under the kidney capsule of nude mice. Using a subcutaneous osmotic pump designed to deliver a steady dose of drug, we treated the mice for 30 days with vehicle, E2, or liquiritigenin. Large tumors developed in the mice treated with E2 (Fig. 6B), while there was essentially no tumor growth in the mice treated with vehicle (Fig. 6A) or liquiritigenin (Fig. 6C). There were also no differences in the weights of the tumors in mice treated with liquiritigenin compared to the control mice (Fig. 6D). In addition, after 30 days of treatment, liquiritigenin did not increase uterine horn mass, whereas E2 did (Fig. 6E).
We have been exploring plants as a source for ER subtype selective estrogens based on their historical use for treating menopausal symptoms in TCM and on the findings that soybeans contain isoflavones that selectively bind to ERβ  and preferentially activate it in transfection studies . We first tested a crude plant extract, MF101, for ER subtype selectivity. Despite the fact that MF101 is comprised of 22 different botanical agents and a multitude of compounds, our studies found MF101 exhibited ERβ-selectivity and did not exhibit proliferative effects on human breast cancer cells or the mouse uterus . The purpose of this study was to isolate active compounds from the individual plant components of MF101 as potential drugs to treat menopausal symptoms. Individual compounds have the potential to be safer than the crude herbal formulation since some of the non-therapeutic compounds might elicit adverse affects. In clinical trials, however, there were no major adverse effects detected after MF101 treatment (data not shown). Active compounds can also be synthesized and quantified, allowing for the administration of known amounts and higher doses of the active drug. In this study, we isolated the ERβ-active compound, liquiritigenin, from G. uralensis, one of the herbs present in MF101. We further detailed the binding and transcriptional activity of this purified compound through the ER.
Liquiritigenin induced only ERβ-specific pathways in transfection assays. Liquiritigenin activated ERE-tk-luciferase, as well as three native ER regulatory elements (NKG2E, CECR6, and NKD) in cells transfected with ERβ but not with ERα. The ERβ-selectivity was also observed with the native NKG2E, CECR6, and NKD genes since these genes were activate activated in the U2OS-ERβ cells, but not in the U2OS-ERα cells. The mechanism for the ERβ-selectivity is unlikely related to differences in the binding to ERα and ERβ, because ERβ only has a 20-fold higher affinity for liquiritigenin compared to ERα. Our ChIP studies showed that liquiritigenin recruits SRC-2 to the NKG2E, CECR6, and NKD genes only in U2OS-ERβ cells. These findings suggest that the selectivity of liquiritigenin is due to the differential recruitment of coactivators to ERβ.
The major problem with HT is not a lack of efficacy, but rather its proliferative effects on breast cancer cells. Therefore, it is essential to rule out a proliferative action for any alternative drug for HT. In a mouse xenograft model, liquiritigenin did not stimulate breast cancer tumor formation after 30 days of treatment, as compared to therapeutic doses of E2, which caused the formation of large tumors. In addition, unlike E2, liquiritigenin did not increase the size of the uterus. The lack of stimulation of breast and uterine cells by liquiritigenin is consistent with the findings that the synthetic ERβ-selective drug, ERB-041 also does not elicit any proliferative effects on mammary and uterine tissue in rats . The data with liquiritigenin, MF101 and ERB-041, as well as the findings that ERβ acts as a tumor suppressor in breast cancer cells [28,29], indicate that ERβ-selective agonists might have a safer profile than the estrogens currently used in HT that activate both ERα and ERβ.
While plants are known to contain many estrogenic compounds , their selectivity for the ER subtypes remains largely unstudied. The isoflavone genistein binds better to ERβ than ERα , and exhibits ERβ-selectivity in transfection studies . Liquiritigenin was more selective than genistein in our studies. Genistein at 1 μM produced a large activation of ERE-tkLuc  and activated numerous genes in U2OS-ERα cells (data not shown). In contrast, liquiritigenin did not activate multiple ER regulatory elements or endogenous genes at the same concentration (data not shown). A related compound, isoliquiritigenin has been shown to activate ERα in MCF-7 cells . We found that isoliquiritigenin is a non-selective agonist that activates both ERα and ERβ transcriptional pathways (data not shown). The structural differences between liquiritigenin, genistein, and isoliquiritigenin that result in higher ERβ selectivity for liquiritigenin are currently being investigated.
We previous showed that the crude botanical mixture MF101 acted as a selective ERβ-agonist by inducing a functional conformational change in the ERβ receptor that causes the recruitment of coactivators . Here we identified liquiritigenin as a major ERβ-selective compound from one of the plants in MF101. These studies suggest that liquiritigenin might be a viable drug candidate to selectively activate ERβ in humans. However, the major question that remains to be answered is whether ERβ-selective estrogens, like liquirtigenin, will be effective at treating menopausal symptoms. In a rat hot flash model ERB-041 did not show any efficacy, suggesting that ERβ-selective compounds might not be useful for treating hot flashes . However, it is possible that other ERβ-selective ligands, such as liquirtigenin will exert different biological effects than ERB-041. Whereas our studies with the U2OS osteosarcoma cells indicate that liquirtigenin regulates genes in bone cells it is unclear if they will increase bone mineral density. Future clinical trials with liquiritigenin are required to determine if it is effective at treating hot flashes, other menopausal symptoms or osteoporosis. While the precise effects that are mediated by ERβ are unclear and the indications of ERβ-selective ligands need to be determined, our study indicates that plants are a source of highly selective ERβ compounds.
We thank Pierre Chambon and Jan-Åke Gustafsson for providing plasmids and Roanna Medina for technical assistance. This work was supported by a grant from the National Center for Complementary and Alternative Medicine (AT002173).
J. E. M. and N. L. contributed equally to this work
Disclosure Statement: J.E.M., N.L., T.Z. and W.A.R. have nothing to declare. R.E.S, S.B., S.C., M.T., and I.C., are employed by and have equity interests in Bionovo, Inc. L.F.B. is on the Scientific Advisory Board of Bionovo, Inc. D.C.L. is on the Scientific Advisory Board and has received financial support for research from Bionovo, Inc.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.