A molecular docking study of phytochemical estrogen mimics from dietary herbal supplements

Purpose The purpose of this study is to use a molecular docking approach to identify potential estrogen mimics or anti-estrogens in phytochemicals found in popular dietary herbal supplements. Methods In this study, 568 phytochemicals found in 17 of the most popular herbal supplements sold in the United States were built and docked with two isoforms of the estrogen receptor, ERα and ERβ (a total of 27 different protein crystal structures). Results The docking results revealed six strongly docking compounds in Echinacea, three from milk thistle (Silybum marianum), three from Gingko biloba, one from Sambucus nigra, none from maca (Lepidium meyenii), five from chaste tree (Vitex agnus-castus), two from fenugreek (Trigonella foenum-graecum), and two from Rhodiola rosea. Notably, of the most popular herbal supplements for women, there were numerous compounds that docked strongly with the estrogen receptor: Licorice (Glycyrrhiza glabra) had a total of 26 compounds strongly docking to the estrogen receptor, 15 with wild yam (Dioscorea villosa), 11 from black cohosh (Actaea racemosa), eight from muira puama (Ptychopetalum olacoides or P. uncinatum), eight from red clover (Trifolium pratense), three from damiana (Turnera aphrodisiaca or T. diffusa), and three from dong quai (Angelica sinensis). Of possible concern were the compounds from men’s herbal supplements that exhibited strong docking to the estrogen receptor: Gingko biloba had three compounds, gotu kola (Centella asiatica) had two, muira puama (Ptychopetalum olacoides or P. uncinatum) had eight, and Tribulus terrestris had six compounds. Conclusions This molecular docking study has revealed that almost all popular herbal supplements contain phytochemical components that may bind to the human estrogen receptor and exhibit selective estrogen receptor modulation. As such, these herbal supplements may cause unwanted side effects related to estrogenic activity.


Background
The use of alternative medicines in the United States, particularly herbal supplements, has dramatically increased since the beginning of the 21st century ( Figure 1). Filling American minds with promises of enhanced beauty, sharper senses, and optimum organ functions, herbal supplements claim to increase, or improve almost all issues a person could have with their body. Without a doubt it is appealing to have problems solved by simply swallowing a pill or drinking a tea, not much effort required, however it has been widely ignored the consecutive consequences these supplements can provide (Cupp 1999).
Two major factors play a part in the ongoing, unnoticed herbal supplement crisis: Regulations for herbal supplements and uneducated consumers. Beginning with the first, the United States does not classify herbal supplements as drugs, and therefore supplements are not required to undergo the extensive testing that pharmaceutical drugs do before put on the market. Courtesy of the "Dietary Supplement Health and Education Act of 1994", herbal supplements are not evaluated by the Food and Drug Administration (Calixto 2000) making it easy for supplement companies to rapidly introduce new supplements to consumers, with or without the knowledge of possible harmful side effects. Unspecified drugs, contaminations, toxins, and/or heavy metals (Au et al. 2000) can be included in an herbal supplement, and since companies are not required to subject their products to quality analysis, this spectrum of harmful compounds could be digested by a consumer and induce adverse effects. As for the second, biologically uneducated consumers do not understand or simply do not consider the concept that plants are not always beneficial. They believe anything that is natural must be good for their health and safe to consume (Stonemetz 2008), which is far from the truth. Plants contain hundreds of phytochemicals, some of which are indeed toxic to the human body. One class of phytochemicals of major concern, which is the focus of this study, phytoestrogens, can interfere and react with the human estrogen receptors, which regulate neural, skeletal, cardiovascular, and reproductive tissues. This interference, however, is not always adverse. For example, some phytoestrogens can promote carcinogenic growth, while others can inhibit the growth.
The purpose of this study was to identify potential estrogen mimics or anti-estrogens in phytochemicals found in popular dietary herbal supplements. The data gathered can only suggest the possibility of a phytochemical to be an anti-estrogen or a mimic, not confirm its estrogenic properties. It is our hope that the discoveries made during this study can help to identify the estrogenic activity of the phytochemicals examined. This information can then lead to the health benefits or hazards associated with the phytochemicals, which in turn could greatly affect the increasingly popular herbal supplement movement.

Molecular docking
Protein-ligand docking studies were carried out based on the crystal structures of human estrogen receptor α [ERα: PDB 1X7E (Manas et al. 2004a), PDB 1X7R (Manas et al. 2004b), and PDB 3ERD (Shiau et al. 1998)] and human estrogen receptor β [ERβ: PDB 1U3Q, 1U3R, 1U3S (Malamas et al. 2004), 1U9E, 1X7B, 1X76, 1X78 (Manas et al. 2004a), and 1X7J (Manas et al. 2004b)]. Prior to docking all solvent molecules and the co-crystallized ligands were removed from the structures. Molecular docking calculations for all compounds with each of the proteins were undertaken using Molegro Virtual Docker v. 6.0 (2013). Potential binding sites in the protein structures were identified using the grid-based cavity prediction algorithm of the Molegro Virtual Docker (2013) program. The location of the volume used by the docking search algorithm was positioned at the center of the cavity and a sphere (15 Å radius) large enough to encompass the entire cavity of the binding site of each protein structure was selected in order to allow each ligand to search. If a cocrystallized inhibitor or substrate was present in the structure, then that site was chosen as the binding site. If no co-crystallized ligand was present, then suitably sized (>50 Å 3 ) cavities were used as potential binding sites. The docking searches were constrained to those cavities. Standard protonation states of the proteins based on neutral pH were used in the docking studies. Each protein was used as a rigid model structure; no relaxation of the protein was performed. Assignments of charges on each protein were based on standard templates as part of the Molegro Virtual Docker (2013) program (Thomsen and Christensen 2006); no other charges were necessary to be set. Flexible ligand models were used in the docking and subsequent optimization scheme. As a test of docking accuracy and for docking energy comparison, co-crystallized ligands were re-docked into the protein structures (see Table 1). Additionally, as positive controls, the known estrogenic compounds 17β-estradiol and α-zearalenone were docked with each protein structure in order to compare docking energies with the herbal phytochemicals. Different orientations of the ligands were searched and ranked based on their energy scores. The RMSD threshold for multiple cluster poses was set at <1.00 Å. The docking algorithm was set at maximum iterations of 1500 with a  Tables 2, 3 , 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 and 15.

Alkaloids
The alkaloid ligands examined in this study are shown in Figure 2. The molecular docking results for the alkaloids are summarized in Table 2. Of the alkaloids examined in this study, cis-and trans-clovamide, with docking energies of −119.8 and −113.6 kJ/mol, respectively, and N-transferuloyltyramine (E dock = −103.1 kJ/mol) were found to dock well with ERα. Their docking energies were more exothermic than those of estradiol, −92.0 kJ/mol and the corresponding co-crystallized ligand genistein, −93.4 kJ/mol, and the clovamides were more exothermic than zearalenone (E dock = −104.1 kJ/mol). The co-crystallized ligand, genistein, and the clovamide and feruloylyramine ligands have similar positions in the binding site ( Figure 44). Phe 404, Leu 525, Leu 346, Leu 387, and Leu 391 form a hydrophobic pocket around the docked alkaloids. Phe 404 exhibited edge-to-face π-π interactions between the phenyl substituent of Phe with the caffeic or ferulic substituents of the alkaloids and with the hydroxyphenyl substituent of genistein. Notable hydrogen bonds in the lowest-energy docked pose of cis-clovamide were the 3-OH and 4-OH of the cis-caffeic moiety with the carboxylate residue of Glu 353 and the 3-OH group with the guanidine residue of Arg 394 ( Figure 45).  Figure 3 Chalcone ligands examined in this work.
Similarly, cis-clovamide, trans-clovamide, and N-trans-feruloyltyramine were the alkaloids that docked well with ERβ. Their docking energies (−124.9, −122.0, and −113.8 kJ/mol, respectively) were more exothermic than those of estradiol, −100.0 kJ/mol, zearalenone, −104.9 kJ/mol, and the corresponding co-crystallized ligand 2-(3-fluoro-4-hydroxyphenyl)-7-vinyl-1,3-benzoxazol-5-ol, −107.9 kJ/mol. The alkaloid and the co-crystallized ligand occupied similar positions in the binding site, a hydrophobic pocket formed by Leu 298, Phe 356, Leu 339, and His 475. Phe 356 exhibited edge-to-face π-π interactions with the caffeic or ferulic substituents of the docked alkaloid ligands as well as with the hydroxyphenyl substituent of the co-crystallized ligand. There were two notable hydrogen bonds formed between the 4-OH group on the ferulyl substituent of N-trans-feruloyltyramine and the guanidine group of Arg 346, and the carboxylate of Glu 305 ( Figure 46). These same two residues formed hydrogen bonds with the 4-hydroxyphenyl group of the co-crystallized ligand. The caffeoyl group of cis-clovamide formed hydrogen bonds with Glu 305 and Leu 298. trans-Clovamide, however, formed hydrogen bonds with Leu 339, Arg 346, and Glu 305.

Coumarins
The MolDock docking energies of the coumarins are summarized in Table 4 Figure 5 Additional coumarin ligands examined in this work. and Leu 391. Furthermore, there are edge-to-face π-π interactions between Phe 404 and the chromene benzene rings of the ligands, as well as hydrogen bonds between the 7-OH group of the chromene and the guanidine group of Arg 394 and the carboxylate of Glu 356, analogous to the co-crystallized ligand genistein. Additionally, one of the carboxylates of pratenol B forms a hydrogen bond with imidazole substituent NH group of His 475.
Phe 356 exhibited edge-to-face π-π interactions with the hydroxychromene substituent of glabrene. There was a hydrogen bond between the imidazole substituent NH group of His 475 and the 5′-OH group of glabrene, and three hydrogen bonding interactions were seen between the 7-OH group of glabrene and the carbonyl group of Leu 339, the guanidine moiety of Arg 346, and the carboxylate of Glu 305.

Diterpenoids
The structures for the diterpenoid ligands examined in this work are shown in Figures 6,7,8,and 9, and the docking energies are listed in  Figure 6 Diterpenoid ligands examined in this work.

Lignans
The structures and the docking energies of the lignans are shown in Figure 21 and  -OH groups hydrogen bonded to argenine in the binding pocket (Arg 394 in ERα; Arg 346 in ERβ) and the other phenolic -OH group hydrogen bonded to the histidine (His 524 in ERα; His 475 in ERβ).

Phenanthrenoids
The docking energies and structures of phenanthrenoids are shown in Table 9 and Figure 22, respectively. None of the phenanthrenoids examined in this work showed docking energies lower than estradiol or zearalenone for ERα or ERβ.

Miscellaneous phenolics
The docking energies for miscellaneous herbal phenolic compounds are listed in Analogous to other phenolic compounds (see above), the cinnamate moiety of the lowest-energy pose of cimicifugic acid F in ERα shows edge-to-face π-π interactions with Phe 404, and hydrogen bonding between the -OCH 3 group and the guanidine group of Arg 394. Additionally, the 3-carboxylate group of the ligand is hydrogen-bonded to His 524, and the phenolic group fits into a hydrophobic pocket formed by Met 388, Met 421, and Ile 424. The lowest-energy docked pose of the aglycone of agnucastoside C with ERβ, as with other phenolic compounds (see above), has the p-coumarate phenolic -OH group hydrogen bonded to the carbonyl group of Leu 339 and the guanidine group of Arg 346, and edge-toface π-π interactions of the phenolic ligand with Phe 356. The cyclic hemiacetal group is hydrogen bonded to His 475.

Stilbenoids
Structures and docking energies for the stilbenoid ligands are shown in Figures 35 and 36, and Table 13, respectively.

Triterpenoids
Docking energies are presented in Table 14   triterpenoid ligands that showed good docking with either ERα or ERβ.

Miscellaneous phytochemicals
Several miscellaneous phytochemicals found in herbal supplements were included in this study (

Angelica sinensis
Dong quai (Angelica sinensis root) has been used in Chinese traditional medicine for thousands of years for various female health conditions (e.g., dysmenorrhea, pelvic pain, symptoms of menopause) (Chye 2006;Al-Bareeq et al. 2010;Fang et al. 2012). In spite of its history, dong quai provided no clinical relief of menopausal symptoms (Hirata et al. 1997). In fact, dong quai has been shown to stimulate the growth of MCF-7 (ER+ human mammary carcinoma) cells (Lau et al. 2005), but does not bind either ERα or ERβ (Liu et al. 2001). The plant contains several miscellaneous phytochemicals, only two of which have notable docking energies, 10-angeloylbutylphthalide (−107.1 kJ/mol with ERβ) and angeliferulate (−110.7 and −121.5 kJ/mol with ERα and ERβ, respectively).

Centella asiatica
Centella asiatica (gotu kola) has been used in Ayurvedic traditional medicine for cognitive enhancement (Rao et al. 2005), to alleviate symptoms of anxiety and promote relaxation (Wijeweera et al. 2006), as well as for headache, body ache, asthma, ulcers, and wound healing (Kumar and Gupta 2002). Animal studies have revealed cognitive enhancement (Kumar and Gupta 2002;Rao et al. 2005), neuroprotective (Subathra et al. 2005), and anxiolytic (Wijeweera et al. 2006) effects. To our knowledge, there have been no reports on the estrogenic activity of C. asiatica.
C. racemosa extracts have revealed several triterpenoids , phenylpropanoids (Chen et al. 2002) and caffeic acid derivatives (Li et al. 2003). Very few of the C. racemosa triterpenoids showed negative docking energies and are, therefore, unlikely estrogen receptor binding agents. Several C. racemosa phenolic compounds did show remarkable docking affinities for both ERα and ERβ: cimicifugic acid A, cimicifugic acid B, cimicifugic acid G, cimiciphenol, cimiciphenone, cimiracemate A, cimiracemate B, cimiracemate C, cimiracemate D, and fukinolic acid. It is likely that any estrogenic activity of C. racemosa extract (Seidlová- Wuttke et al. 2003a) is due to the presence of phenolic components rather than triterpenoids.

Echinacea spp.
Echinacea (E. angustifolia, E. pallida, and E. purpurea) is one of the most popular herbal supplements sold in the United States and has been used as a treatment for the common cold, coughs, bronchitis, upper respiratory infections, and inflammatory conditions (Percival 2000). Recent studies have demonstrated Echinacea to exhibit immunesystem-stimulating activity (Block and Mead 2003). Phytochemicals that have been isolated from Echinacea spp. include chicoric acid and monomethyl and dimethyl ethers, trichocarpinine, cinnamoylechinadiol, cinnamoylechinaxanthol, cinnamoylepoxyechinadiol, cinnamoyldihydroxynardol, caftaric acid, caffeoyl-p-coumaroyltartaric acid, burkinabin A, burkinabin B, kaempferol, luteolin, and quercetin. Although Echinacea has not shown estrogenic activity (Zava et al. 1998), six phytochemicals were identified in this docking study that showed strong docking to the estrogen receptor: the flavonoid quercetin; the phenolic compounds caffeoyl-p-coumaroyltartaric acid, caftaric acid, and chicoric acid; and the sesquiterpenoids cinnamoylechinadiol and cinnamoylepoxyechinadiol.

Gingko biloba
G. biloba is commonly used as a supplement to improve cognitive abilities (Kennedy et al. 2000), and for women specifically, it has been used to treat some of the side effects accompanying menopause (Oh and Chung 2004).
Licorice (Glycyrrhiza glabra) root has been used for thousands of years by different cultures and for a variety of reasons (Fenwick et al. 1990). Although licorice root has been suggested as a treatment for symptoms of menopause (Ojeda 2003), G. glabra root extracts have been shown to be inactive in terms of ERα or ERβ binding (Liu et al. 2001). Nevertheless, however, fractionation of G. glabra extracts has revealed several ER-modulating components (Khalaf et al. 2010;Simons et al. 2011). Glabrene binds to human ER and shows estrogenic activity (Tamir et al. 2001;Simons et al. 2011). Our in-silico docking study shows glabrene to be a strongly docking ligand to both ERα and ERβ (−104.8 and −114.9 kJ/mol, respectively). Glabridin, on the other hand, displayed ERαselective antagonism (Simons et al. 2011), in contrast to the docking results that showed glabridin to have ERβ docking selectivity (−15.8 and −92.9 kJ/mol, respectively). The chalcones isoliquiritigenin (Tamir et al. 2001;Maggiolini et al. 2002) and licochalcone A (Rafi et al. 2000), the flavonoid quercetin (Kuiper et al. 1998), and the isoflavonoid genistein (Ososki and Kennelly 2003) also bind to human ERα and show estrogenic activity. These ligands all show negative docking energies with ERα (range from −93.2 to −99.9 kJ/mol) and ERβ (−98.9 to −107.8 kJ/mol).

Lepidium meyenii
Maca (Lepidium meyenii) is native to the central Andes of Peru (3500-4500 m asl) (Wang et al. 2007). The root has been used by native Amerindians to improve fertility, as an aphrodisiac for both men and women. Maca was found to increase sperm counts and gonadal mass in a rat model (Chung et al. 2005), to improve copulatory performance of male mice and rats (Zheng et al. 2000;Cicero et al. 2001), and to increase litter size (Ruiz-Luna et al. 2005) and pregnancy rates in female mice (Kuo et al. 2003). In adult human males, maca treatment led to increased semen volume and sperm count (Gonzales et al. 2001) and increased sexual desire (Gonzales et al. 2002;Stone et al. 2009). In addition, maca reduced sexual dysfunction in postmenopausal women (Brooks et al. 2008) and inhibited estrogen-deficient osteoporosis in ovariectomized rats (Zhang et al. 2006), but maca extracts have not shown estrogenic activity (Brooks et al. 2008). None of the L. meyenii phytochemicals investigated in this in-silico study showed remarkable docking energies, consistent with the non-estrogenic activity previously reported. Ptychopetalum olacoides, P. uncinatum Muira puama (bark and root extracts of P. olacoides or P. uncinatum) has been used in Amazonian Brazil during highly stressful periods, to treat CNS-related ailments, neuromuscular problems, "nervous weakness", sexual debility, frigidity, impotence, and rheumatism (Schultes and Raffauf 1990;Siqueira et al. 1998;Duke et al. 2009). Consistent with these traditional uses, P. olacoides ethanol root extract has shown memory retrieval improvement in young and aging mice (da Silva et al. 2004), in-vitro acetylcholine esterase inhibitory activity (Siqueira et al. 2003), and prevention of stress-induced hypothalamic-pituitaryadrenal hyperactivity (Piato et al. 2008). In addition, Muira puama formulations have demonstrated efficacy in treating male erectile dysfunction and low libido (Waynberg 1994) and low sex drive in women (Waynberg and Brewer 2000). A number of clerodane diterpenoids have been isolated from P. olacoides bark (Tang et al. 2008;Tang et al. 2009;Tang et al. 2011). Several of these have given excellent docking energies with the estrogen receptor, but two in particular, ptycho-6α,7α-diol and ptycholide IV had remarkable docking to ERβ (−122.9 and −114.7 kJ/mol, respectively). To our knowledge, the estrogenic effects of Muira puama have not been investigated.

Rhodiola rosea
R. rosea is reputed to strengthen the nervous system, fight depression, enhance memory, and improve energy levels (Brown et al. 2002), which has been attributed to adaptogenic properties of the herb ( . The flavonoids gossypetin, herbacetin, and rhodiolin, and the lignan (+)-lariciresinol, have been identified in R. rosea. Lariciresinol showed strong docking to both ERα and ERβ. There are conflicting reports on the potential estrogenic effects of R. rosea, however (Eagon et al. 2004;Kim et al. 2005).

Silybum marianum
Milk thistle, Silybum marianum, extracts (silymarin) have been used for centuries to treat liver diseases (Flora et al. 1998

Tribulus terrestris
Tribulus terrestris has been used to contribute to physical and sexual strength (De Combarieu et al. 2003;Neychev and Mitev 2005). The plant is rich in steroidal glycosides (Wu et al. 1996;Yan et al. 1996;De Combarieu et al. 2003;Dinchev et al. 2008) and T. terrestris extracts have shown androgenic effects in animal models (Gauthaman et al. 2002;Gauthaman and Ganesan 2008), but had no influence on androgen production (Neychev and Mitev 2005) or gains in strength or muscle mass (Rogerson et al. 2007) in young men. To our knowledge, there have been no reports on the estrogenic effects of T. terrestris.

Trifolium pratense
The alkaloids cis-and trans-clovamide, several coumarins, flavonoids, and isoflavonoids have been identified in red clover (Trifolium pratense). Although there have been no ethnobotanical reports to support it, the presence of isoflavones has led to suggest that red clover may serve as a phytochemical alternative to post-menopausal hormone replacement therapy (Coon et al. 2007). Red clover extract has been shown to exhibit weakly estrogenic effects in a rat model (Burdette et al. 2002) and does show in-vitro ERα and ERβ binding ability (Dornstauder et al. 2001;Beck et al. 2003;Overk et al. 2005 (Beck et al. 2003;Pfitscher et al. 2008) and these compounds did show a binding preference for ERβ. Molecular docking of these ligands also showed preference for ERβ. They were not, however, the best docking ligands of T. pratense phytochemicals. The   alkaloids cis-and trans-clovamide, and the coumarin pratenol B showed good docking to both ERα and ERβ. Of the T. pratense isoflavonoids, calycosin and pseudobaptigenin had more exothermic docking energies to ERβ than did biochanin A, daidzein, formononetin, or genistein.

Conclusions
This molecular docking study has revealed that almost all popular herbal supplements contain phytochemical components that may bind to the human estrogen receptor and exhibit selective estrogen receptor modulation. As such, these herbal supplements may cause unwanted side effects related to estrogenic activity. For example, estrogenic agents may be effective and potent growth stimulators of estrogen-receptor positive tumors and pose a hazard to patients with breast cancer who have ER-positive tumors and who are being treated with antiestrogens. The strongest docking (most exothermic docking energies) phytochemical ligands were phenolic compounds and the weakest docking ligands were triterpenoids. A common binding motif for phenolic ligands in ERα is the hydrophobic pocket of Leu 387, Phe 404, Met 388, and Leu 391, along with edge-to-face π-π interactions with Phe 404, and hydrogen bonds between the phenolic -OH group and the  There are several limitations to these docking results: Some of the herbal phytochemicals examined may not be bioavailable due to limited solubility, membrane permeability; This docking study has only examined docking of the natural ligands (or their aglycones) and does not take into account in-vivo hydrolysis or other metabolic derivatization; The docking studies do not account for synergism in the estrogen receptor binding; The molecular docking method itself suffers from inherent limitations (e.g., the protein is modeled as a rigid structure without flexibility, solvation of the binding site and the ligand is excluded, and free-energy estimation of protein-ligand complexes is largely ignored) (Yuriev et al. 2011;Yuriev and Ramsland 2013). Docking energies do not provide information about whether strongly binding ligands may function as agonists or antagonists of the estrogen receptor.