Key words
Erythrina poeppigiana
- Fabaceae - erythroidines - alkaloids - phytoestrogens - reporter gene assay
Introduction
Medicinal plants used in traditional medicine can be considered as an almost permanent
source of medication of significant importance. A high percentage of individuals,
particularly in third world countries, depend on medicinal plant-based preparations
for their primary health care. In particular, about 80 % of the population in developing
countries still resorts to medicinal plants for various reasons. Most important, they
are affordable and, according to their long-term use in humans, they are usually considered
to be safe, although their consumption may pose risks undetected
so far [1], [2]. Unfortunately, very often the knowledge base for traditional use is empirical.
More importantly, it is usually not known whether the plants or plant-derived products
contain bioactive compounds, which may account for the traditional use.
Among these traditional medicinal plants, Erythrina poeppigiana, from the Erythrina species (Fabaceae), is of particular interest since it is widely used for the treatment
of various disorders and complications including hormone-related conditions [3], [4]. It is distributed and utilized in traditional remedies in South America, Africa,
and Asia. Irrespective of the geographical site of use, this plant has been reported
as being used in the folk empirical system of medicine as an abortive agent and for
the
treatment of amenorrhea, microbial infections, jaundice, agitation, and insomnia [5]. E. poeppigiana is particularly rich in secondary metabolites with estrogen-like activities commonly
referred to as phytoestrogens [6]. In a recent study, the dichloromethane extract of E. poeppigiana was found to be exceptionally rich in isoflavones, among them eight specifically
prenylated genistein derivatives of which seven exhibited significant estrogenic activities
in vitro
[6], [7].
Chemically speaking, the large majority of the newly discovered phytoestrogens belongs
either to flavonoids (flavones, flavanones, chalchones), isoflavonoids (isoflavones,
coumestanes, arylbenzofurans), or lignans, with isoflavones being the best characterized
class [8]. Reports on plant secondary metabolites for which the estrogenic activity originates
from an alkaloid structure are rather scarce. A study published in 1994 by Ng et al.
reported for the first time estrogenic activity of an alkaloid. Yuehchukene, a bis-indole alkaloid isolated
from Murraya paniculata (Rutaceae), exhibited mixed estrogenic and antiestrogenic properties in vivo and in vitro
[9]. In 2001, Nazrullaev et al. reported the estrogenic activity of several quinoline
alkaloids [10]. Recently, Allred et al. described the estrogenic activity of trigonelline, an alkaloid
from coffee beans structurally related to vitamin B3 [11]. A possible reason for alkaloids being neglected towards the exploration of their
estrogenicity could be that they are generally considered toxic. However, the striking
structural variation of a vast number of natural alkaloids is the basis of a wide
spectrum of quite diverse activities. Moreover, several pharmacological studies have
shown that several alkaloids belonging to different chemical subgroups exhibit only
a low toxicity [12], [13].
Along these lines and in a continuing effort to characterize the chemical and biological
properties of medicinal plants traditionally used for the treatment of gynecological
problems, we investigated the methanolic extract of E. poeppigiana. The alkaloid-rich fraction was targeted and isolated, and the corresponding constituents
were purified and structurally elucidated using chromatographic and spectroscopic
methods, respectively. Moreover, various estrogen receptor (ER)-dependent test systems,
including receptor binding and reporter gene assays, in two independent
cellular systems, namely MVLN breast cancer cells (a bioluminescent variant of MCF-7
breast cancer cells) [14] and U2OS human osteosarcoma cells [15], were used. These cell lines were utilized for determination of potential estrogen-like
activities of the isolated alkaloids. We finally evaluated the impact of the most
potent constituents of the methanolic extract of E. poeppigiana on the expression of the endogenous estrogen regulated genes SGK3 (serum/glucocorticoid
regulated kinase 3) and trefoil factor-1 (TFF1)
in MCF-7 cells. In addition, to further substantiate observed estrogenic activities
of α- and β-erythroidine in silico, docking calculations were performed. The interactions of isolated erythroidines to
the ligand binding domain (LBD) of the estrogen receptor alpha (ERα) have been studied considering ER plasticity as expressed by different conformational
preferences derived by crystallography.
Results
The relative binding affinities (RBA) to ERα and ERβ for 1–4 ([Fig. 1]) were determined using purified recombinant proteins and a fluorescence polarization
approach ([Table 1] and Fig. 1S, Supporting Information). Compounds 1 and 2 displayed RBA values for ERα of 0.015 ± 0.010 and 0.005 ± 0.010, respectively. For 3 and 4 at the concentrations tested, no dose-dependent competition curve could be obtained.
Therefore, no RBA could be determined, implying a
very low or no affinity for ERα. The same holds for binding to ERβ, except for 2, for which a relatively low RBA of 0.006 ± 0.010 to ERβ could be determined. IC50 values of estradiol, 1, and 2 for ERα were 8.31 × 10−9 M, 5.73 × 10−5 M, and 1.11 × 10−4 M, respectively, whereas the IC50 of 2 to ERβ was 9.61 × 10−5 M as compared to 5.47 × 10−9 M for estradiol.
Fig. 1 Chemical structures of isolated compounds 1 (α-erythroidine), 2 (β-erythrodine), 3 (8-oxo-α-erythrodine), and 4 (8-oxo-β-erythrodine) on MVLN cells (breast cancer cell line derivative).
Table 1 Ligand binding data.
|
RBA (%)
|
IC50 (µM)
|
|
Erα
|
ERβ
|
Erα
|
ERβ
|
|
* n. m. = not measurable
|
|
Estradiol
|
100
|
100
|
0.0083
|
0.0055
|
|
α-Erythroidin
|
0.015 ± 0.010
|
n. m.
|
57.3
|
n. m.
|
|
β-Erythroidin
|
0.005 ± 0.010
|
0.006 ± 0.010
|
111
|
96.1
|
|
8-Oxo-α-erythroidin
|
n. m.*
|
n. m.
|
n. m.
|
n. m.
|
|
8-Oxo-β-erythroidin
|
n. m.
|
n. m.
|
n. m.
|
n. m.
|
To verify whether this binding affinity to the ER translates into receptor transactivation,
we performed reporter gene assays. As toxicity of alkaloids is a concern, we additionally
performed the MTT cytotoxicity assay in U2OS osteosarcoma cells. This test showed
that both erythroidines are void of cytotoxicity in doses up to 10 µM (Fig. 13S, Supporting Information).
Estrogenic activity of the alkaloids was first assayed in a derivative of ER-positive
MCF-7 breast cancer cells (MVLN). In accordance with the aim of our study, we assessed
the estrogenic activity of the four isolated alkaloids as well as a positive (E2 10 nM)
and a negative (DMSO 0.1 %) control in MVLN cells, and verified observations for the
most potent molecules in U2OS human osteosarcoma cells. Compound 1 induced significant and dose-dependent luciferase reporter gene activity in MVLN
cells, starting from 10 nM, while compound 2 induced significant estrogenic
activity starting from 1 µM ([Fig. 2]). On the other hand, compounds 3 and 4, which are the corresponding 8-oxo derivatives of 1 and 2, failed to induce any significant estrogenic activity in MVLN cells ([Fig. 2]). From these results, we made two decisions for additional experiments. First, only
1 and 2 exhibited sufficient activity to be studied in depth. Second, we decided to continue
to investigate interactions of erythroidines with ERα only to address the issue of
structure-function relationships.
Fig. 2 Estrogenic activity of erythroidines 1 and 2 activated ERα-dependent reporter gene expression in a statistically significant manner, reaching
a maximum response at 10 µM, while 3 and 4 had no effects. Significance was calculated against DMSO (set to 100 %): ** p < 0.01,
*** p < 0.001. Data represent the mean ± SD of at least three independent triplicate
experiments. MVLN cells: breast cancer cell line derivative.
As a consequence of this decision, compounds 1 and 2 have been investigated for their estrogenic potential in a cell line of a different
organ, namely, bone-derived U2OS human osteosarcoma cells expressing ERα (U2OS-ERα). Compound 1 induced significant and dose-dependent reporter gene activity in U2OS cells starting
from 10 nM ([Fig. 3]), while 2 also induced a significant and dose-dependent activation of luciferase reporter gene
activity in U2OS-ERα cells starting from 10 nM ([Fig. 3]). As a general observation, the maximum induction in response to compound 2 is by far less pronounced than that of compound 1.
Fig. 3 ERα-dependent luciferase reporter gene activation by α-erythroidine (1) and β-erythroidine (2) in U2OS cells. Compound 1 appears to be more potent than 2 in U2OS-ERα cell systems. Significance was calculated against DMSO (set to 100 %): ** p < 0.01,
*** p < 0.001. Data represent the mean ± SD of at least three independent triplicate
experiments. U2OS-ERα: human osteosarcoma cell line expressing estrogen receptor α.
In order to verify whether the activation of the reporter gene activity by 1 and 2 is mediated through the ER, U2OS-ERα cells were coincubated with the higher effective dose of 1 and 2, and the pure antiestrogen Fulvestrant. Fulvestrant completely inhibited the induction
of the reporter gene transcription by these alkaloids, indicating that their effect
is ER-dependent ([Fig. 4]).
Fig. 4 Inhibition of the ER-dependent reporter gene activity of E2, 1, and 2 by Fulvestrant. Inhibition of estrogenic responses by Fulvestrant is indicative for
the mediation of the effect by the ER. Significance comparing results with 0.1 % DMSO
(= 100 %): *** p < 0.001; significance comparing combined effects against individual
treatment: ### p < 0.001. Data represent the mean ± SD of at least three independent, triplicate
experiments.
Aiming to further characterize the estrogenicity of these alkaloids, we investigated
whether the erythroidines are also capable of inducing endogenous target genes in
MCF-7 breast cancer cells. The impact of 1 and 2 on the expression of the estrogen-regulated genes TFF1 ([Fig. 5 A] and [B]) and serum and glucocorticoid-inducible kinase 3 (SGK3; [Fig. 5 C] and [D]) was evaluated in MCF-7 breast cancer cells. The expression of both genes was upregulated
in response to 10−8 M of estradiol and to 10−5 M concentrations of both test substances ([Fig. 5]). This effect was completely inhibited by coincubation with the antiestrogen fulvestrant
(data not shown).
Fig. 5 Regulation of the TFF1 gene (pS2; A, B) and serum and glucocorticoid-inducible kinase 3 (C, D) gene expression by α- and β-erythroidine (A, C) and (B, D) in MCF-7 cells. Significance was calculated against DMSO (set to 1): * p < 0.05,
** p < 0.01, *** p < 0.001. Data represent the mean ± SD of at least three independent
experiments. MCF-7 cells: breast cancer cell line.
According to the chemical structure of erythroidines, their affinity for the ER was
not expected. Most molecules showing strong estrogenic activity usually possess two
OH groups in a distance of 10–12 Å. However compounds, both natural and semisynthetic,
possessing only one OH group has showed a significant affinity to the ER [16]. Likewise, compounds 1 and 2, although they lack OH groups, exhibited low but surprising binding affinities. This
affinity was proven to be translated into a significant activity regarding ER-dependent
reporter gene expression in cells of different organs and regarding regulation of
endogenous gene expression. So we considered it necessary to perform molecular docking
studies in silico to further understand the interaction of erythroidines with the ER LBD on a molecular
level. Our hypothesis was that the nitrogen atom existing in 1 and 2 could trigger binding to the ER. This nitrogen can be protonated and form a hydrogen
bond with known important hydrophilic amino acids such as His 524, Asp353, and Arg394
inside the binding pocket of the receptor [16]. One the other hand, the nitrogen in oxo-erythoidine derivatives cannot be protonated
and, thus, the binding ability is prohibited in compounds 3 and 4.
Specifically, in order to investigate the binding of compounds 1 and 2 to the ER, all possible conformations and stereochemistry patterns considering the
nitrogen inversion were investigated [17], [18]. The theoretical pKa value of each nitrogen atom was calculated as 9.26 and 9.12,
respectively, using ChemAxon software (www.chemaxon.com) suggesting that at pH 7,
the nitrogen atom is protonated and each compound can exist in an R/S nitrogen configuration. Additionally, calculated logD values on pH 7
were found to be − 1.43 and − 1.86, respectively, showing very low lipophilicity.
After 1000 runs of a conformational search using the Monte Carlo/Low Mode (MC/LMOD)
algorithm, the global minimum of each compound was depicted (Fig. 14S, Supporting Information). These structures reveal that in the S nitrogen configuration, the NH group is better exposed to the solvent or to the interaction
with the receptor pocket, while in the R configuration, NH is sheltered by the core of the molecule.
Furthermore, docking calculations were performed considering the lowest energy structure
obtained for the interaction of each compound with the ERα LBD. Starting from the crystal structure of LBD-ERα in a complex with diethylstilbestrol (DES) (PDB entry 3ERD) or ortho-trifluoromethylphenylvinyl
estradiol (EZT) (PDB entry 2P15), the crystallographic ligand was replaced manually
by compounds 1 and 2. Both ligands fit within the ligand binding cavity of the 3ERD LBD structure of ERα, exhibiting only weak Van Der Waals (VdW) interactions. In the
case of the 2P15 ERα LBD structure, the binding pocket is larger, so both ligands were suitably fitted.
In both cases, the S enantiomer (regarding the N+ configuration) forms a hydrogen bond with His 524 since
the NH is better exposed to the receptor as described above ([Fig. 6] and [7]).
Fig. 6 Minimum energy structure of ERα-LBD (2P15) in complex with α-erythroidine (S configuration) (A) and α-erythroidine (R configuration) (B). (Color figure available online only.)
Fig. 7 Minimum energy structure of ERα-LBD (2P15) in a complex with β-erythroidine (S configuration) (A) and β-erythroidine (R configuration) (B). (Color figure available online only.)
Discussion
In third world countries, people depend on traditional medicine for their primary
health care, which comprises medicinal plants as a fundamental element. Efficacy is
mostly evidenced by the traditional use without knowing the molecular basis of the
activity. Likewise, (toxic) side effects presumably often remain unnoticed. With our
studies on estrogenic activities of compounds from E. poeppigiana, we aimed to contribute to the evaluation of the efficacy and safety of this medicinal
plant.
Generally, plants belonging to Erythrina species are very rich in various subgroups of flavonoids. The estrogenic activity
of E. poeppigiana could initially be linked to derivatives of the soy isoflavone genistein, which were
isolated from its dichlomethane extract [6], [7]. Nevertheless, information about the possible estrogenicity of more polar compounds
of the plant, including the estrogenicity of erythroidines, was missing. Erythroidine
alkaloids represent major constituents of the methanol extract in
Erythrina species. Thus, the aim of the present study was to phytochemically investigate the
polar constituents of the stem bark of E. poeppigiana and to evaluate their estrogenic properties using in vitro and in silico models.
Along these lines, we first characterized the relative binding affinities of the isolated
erythroidines 1, 2, 3 and 4 to the ERs in comparison to 17β-estradiol. Overall, we showed that two (1, 2) of the four compounds bound to ERα and, additionally. one of them (2) to ERβ. With RBA values ≤ 0.1 %, α- and β-erythroidines have to be characterized as weak binders of the ERs; however, with
this relative binding affinity, they are still in the range of the binding affinities
of the red clover isoflavones biochanin A and
formononetin or the endocrine disruptor bisphenol A [19]. In contrast, both oxo-derivatives, 8-oxo-α-erythroidine (3) and 8-oxo-β-erythroidine (4), showed no affinity for the ERs.
Considering that the binding affinity to a receptor does not always correlate with
receptor transactivation [20], the estrogenic activity of these four alkaloids was additionally assessed on cell
culture models of human origin. Prior to hormonal stimulation experiments, we performed
cytotoxicity assays. The MTT test did not reveal cytotoxic properties of erythroidines
(Fig. 13S, Supporting Information).
The initial reporter gene assays were performed in MVLN cells, ERα-expressing MCF-7 breast cancer cell line-derived reporter cells [14]. The results exactly matched the ligand binding data, meaning measurable activity
of 1 and 2 and neglectable activity of 3 and 4. This observation could also be correlated with the docking results. The oxo-erythroidine
derivatives 3 and 4 have lost their ability for nitrogen protonation, so they lose their only group to
form a hydrogen bond with His 524 as shown for
compounds 1 and 2. Since His 524 is not so important for ER recognition compared to Arg394 and Asp353
[16], binding only to His 524 could explain the low binding affinity for compounds 1 and 2.
In order to further investigate the estrogenic activity of the two active alkaloids
1 and 2, they were assayed in the U2OS-ERα variant of the bone-derived human U2OS osteosarcoma cell line [15]. The rationale for this procedure stems from the observation that compounds which
are structurally different from estradiol, but still exhibit estrogen-like properties,
may have organ-specific functional qualities, like, for example, the synthetic estrogen
receptor modulator Tamoxifen. As shown by the results in [Fig. 3], α-erythroidine (1) and β-erythroidine (2) significantly induced reporter gene activity in U2OS-ERα cells, with a more pronounced magnitude of stimulation for 1. The stimulation pattern of the reporter genes in [Fig. 2] and [3] did not result in a steadily increasing dose-response pattern. So far, we are missing
a mechanistic clue, but nonlinear dose-response patterns have been extensively discussed
for endocrine disrupting chemicals (for review see [21]). The pure antiestrogen Fulvestrant® completely inhibited the estrogenic activity
of these alkaloids ([Fig. 4]), implying that the observed effect was primarily mediated through the ER. In essence,
α- and β-erythroidines are capable of inducing ER-mediated reporter gene activity in cells
originating from mammary glands and bones, thereby not exhibiting organ selective
properties.
Reporter gene constructs usually contain minimal or reduced promotor element arrangements
with a low number of DNA base pairs comprising the respective estrogen response element
and, therefore, do not entirely mimic complex promotors of responsive genes, which
often comprise several thousand base pairs. For weak estrogenic compounds, it is therefore
important to test whether they are capable of triggering the ER-dependent regulation
of expression of endogenous genes. We therefore investigated the impact of 1 and 2 on the regulation of expression of specific
estrogen-regulated genes in MCF-7 cells, namely TFF1 and SGK3, by semiquantitative
real-time PCR. TFF1, formerly known as pS2, was the gene from which an estrogen response
element as a molecular switch for the ER was initially described [22], [23]. It also represents a gene which, in the presence of estradiol, is significantly
upregulated both in human MCF-7 breast cancer cells and in breast cancer biopsies
of estrogen-dependent breast cancers, and, therefore, qualifies as an E2 responsive
marker gene [22], [23], [24]. In our study, E2, as well as both erythroidines, induced a significant upregulation
of TFF1 ([Fig. 5 A] and [B]), thus confirming their estrogenic activity, which appears to be ER-dependent because
it was completely inhibited by Fulvestrant®. It has been shown that E2 dose-dependently
induces the expression of the SGK3 gene [25], which promotes ER-positive mammary adenocarcinoma cell
survival. In the current study, E2, 1, and 2 induced an upregulation of SGK3 ([Fig. 5 C] and [D]), thus confirming the estrogenic properties of these alkaloids even on a complex
promotor in a natural genomic organization. Regarding the regulation of expression
of these primary estrogen response genes, α-erythroidine seemed to be slightly more potent than β-erythroidine.
In order to decipher the molecular mechanism by which these alkaloids bind the ER,
molecular docking simulations were performed starting from two distinct crystal structures
of ERα. In both structures, ERα adopts the so-called “agonist conformation” regarding the orientation of the C-terminal
H12 helix. The first structure is the one derived in the complex with DES, a classical
estradiol agonist, usually adopted for this type of docking calculation. In these
conditions, ERα did not show any remarkable interaction with both erythroidine molecules. However,
the crystal structure of ERα in the complex with 17β-estradiol (ΕΖΤ) had proven that the LBD cavity can display a remarkable structure
plasticity, notably on loop Asp411-Val418, in order to accommodate the bulky analog
of EZT. Considering this specific conformational snapshot of the ERα structure, α/β-erythroidines (1 and 2) could adequately fit and, more importantly, form a hydrogen bond between His 524
and the nitrogen atom (in the S configuration). However, one has to consider that ERα-LBD is lipophilic, while
α/β-erythroidines are protonated at a physiological pH exhibiting negative logD values.
This definitely would influence the entropic term (ΔS) of the systemʼs free energy
of binding, lowering the overall binding affinity. Overall molecular docking calculations
support binding of α/β-erythroidines for ERα. As shown at [Fig. 6] and [7], the S nitrogen configuration of both 1 and 2 forms a hydrogen bond with His 524 inside the receptor pocket. The resulting
interactions provide evidence for the in vitro binding affinity and activity of these alkaloids toward the ERα.
Overall, based on ER-dependent test systems including competitive binding analyses,
reporter gene assays, and gene expression as well as in silico studies, we provide various pieces of evidence for the estrogenicity of α- and β-erythroidine, a prominent class of Erythrina species alkaloids contained in the methanol extract of the stem bark of E. poeppigiana. Thus, we suggest that α- and β-erythroidine alkaloids contribute to the estrogenic activity of the medicinal plant
E. poeppigiana. Whether it contributes to the
estrogenic properties recently shown in an experimental rat model [26] remains open. From our in vitro study, it is not possible to assess the overall contribution of the erythroidines
to the total estrogenic activity of the plant or how it is relative to the less polar
isoflavones [6], [7], [27] or arylbenzofurans [28]. Overall, our results from four different in vitro test systems justify studies in animal models to
better define efficacy and safety profiles of the test compounds.
In conclusion, our study presented here is important in the field of molecular endocrinology
as well as in the area of alkaloid biochemistry, disclosing the existence of potentially
estrogenic agents with this particular chemistry.
Materials and Methods
General experimental procedures
Analytical TLC was performed on Merck precoated silica gel 60 F254 plates. Spots were visualized by fluorescence extinction using UV light and vanillin-sulfuric
acid reagent. Column chromatography was carried out using Si gel 0.04–0.06 mm (Merk).
Amberlite XAD-4 resin (Rhom and Hass) was used for the preparation of the enriched
fraction of the MeOH extract. A Thermo Finnigan HPLC system connected to a spectral
system UV2000 PDA detector was employed for the profiling of the extracts, and ChromQuest
2.1 software was used for the operation of the system and data
management. High-speed countercurrent chromatography was performed using fast centrifugal
partition chromatography equipment (FCPC®, Kromaton) equipped with a 1000-mL rotor
and a preparative pump (LabAlliance®). 1 and 2D NMR spectra (COSY, COSYLR, HSQC-DEPT,
HMBC) were recorded in deuterated chloroform (CDCl3 – Merck) on a Bruker Avance III spectrometer (Bruker Biospin GmbH) operating at 600.11 MHz
for 1H and at 150.11 MHz for 13C, with a 5-mm inverse detection probe. The residual 1H (7.26 ppm) and 13C (77.0 ppm)
signals of CDCl3 were used as an internal standard. 1 and 2D NMR experiments were performed with standard
pulse programs, at room temperature
Plant material
The stem bark of E. poeppigiana was collected in November 2006 in Sancta Cruz (Bolivia), and identified by Ing. Mario
Saldias Paz. A specimen was deposited in the Museo de Historia Natural, Facultad de
Ciencias Agricolas de Santa Cruz-Bolivia under the voucher number USZ: 71775.
Extraction, isolation, and identification
Air-dried and pulverized bark of the plant (1.66 kg) was extracted at room temperature,
successively with CH2Cl2 (3 × 2 L), MeOH (3 × 2 L), and H2O (3 × 2 L), for 48 h per extraction. The MeOH extract was concentrated to give a
residue (90.5 g) from which 50 g was subjected to amberlite resin XAD-4HP (Rhom and
Hass) dissolved in hot water and remained for two days with controlled and smoothed
shaking. After filtration, the resin was washed with MeOH (2 L) and the eluent was
concentrated to give 14.5 g of enriched extract. From this enriched
extract, 10 g were subjected to FCPC-based separation using a biphasic system consisting
of EtOAc/isopropanol/H2O (ethyl acetate/isopropanol/water) in the proportion of 3/2/5. The rotor speed was
set at 1000 cycles/min and the flow rate at 15 ml/min. From this analysis, 119 fractions
of 50 ml each were collected. Fractions 63–67 were put together to provide 4.1 g,
which were further subjected to column liquid chromatography (4.5 cm Ø) on Si gel
(0.015–0.04 mm) leading to the isolation of 1 (40.0 mg) and 2 (60.0 mg). Fraction 19 of the last
separation was subjected to preparative TLC using a dichloromethane/methanol, 95/5,
solvent system, leading to the purification of compound 4 (5.5 mg), while fractions 17–18 (131.2 mg) were concentrated together and subjected
to column liquid chromatography (2.1 cm Ø) on Si gel (0.015–0.04 mm), leading to 3 (7.0 mg). The structural elucidation of the purified alkaloids was carried out using
1 and 2D NMR and comparison with literature data [17], [29], [30]. The isolated
compounds ([Fig. 1]) were identified as α-erythroidine (1), β-erythroidine (2), and their oxo-derivatives, 8-oxo-α-erythroidine (3) and 8-oxo-β-erythroidine (4).
Receptor binding assay
The fluorescent derivative of E2, recombinant human ERα, and recombinant human ERβ were purchased from Invitrogen®, while EZT (purity > 98 %) was obtained from Sigma-Aldrich.
The receptor binding affinities, RBA of 1–4, were assessed using a fluorescence polarization approach as previously described
[31], [32]. In brief, the concentration for ΕΖΤ and 1–4 that inhibited the binding of fluorescent estrogen ES2 (1 nM) (Invitrogen®) to isolated
recombinant human ERα or
ERβ (Invitrogen) by 50 % (IC50) was determined and used to derive the receptor binding affinity values: [RBA = (IC50 17β-estradiol/IC50 compound) × 100].
Cells and plasmids
U2OS cells were obtained from ATCC/Promochem. U2OS cells stably transfected with ERα (U2OS-ERα cells) and the (ERE)2-tk-Luc reporter plasmid were kindly provided by Dr. Luisella Toschi (Schering AG).
MVLN cells, a derivative of an ER-positive MCF-7 cell line stably transfected with
the vitellogenin-A2-promoter/luciferase reporter construct, were from Dr. Michel Pons
(INSERM U439). ERα-positive MCF-7 cells were from the German national center for biomedical material
and resources (DSMZ).
3-(4,5-Dimethylthiazol-2-Yl)-2,5-diphenyltetrazolium bromide assay
MTT was from AppliChem. The MTT test, as such, was performed as described in the literature
[33], with 7500 U2OS cells per well of a 96-well plate.
Reporter gene assays
U2OS cells were routinely cultured in phenol red-free DMEM-F12 medium containing 10 %
fetal calf serum (FCS) and 0.5 mg/ml gentamycin (Calbiochem/VWR). Experiments were
performed in phenol red-free DMEM-F12 medium containing 5 % dextran-coated charcoal
(DCC) stripped FCS, and 0.5 mg/ml gentamycin. For transfection, U2OS-ERα cells were plated in a 24-well plate (30 000 cells/well) and transfected with 100 ng
of the (ERE)2-tk-Luc reporter plasmid using the liposomal protocol (DOTAP; Roth) and DOTAP : DNA
in a ratio of 3 : 1 [15].
MVLN cells as well as MCF-7 cells were cultured in DMEM-F12 medium as previously described
[14], while experiments were performed using DMEM-F12 supplemented with 1 % DCC in 24-well
plates (80 000 cells/well).
Treatment with the test substances and luciferase assay
All tested substances were serially diluted in DMSO (final DMSO concentration in test
well plate of 0.1 % v/v). All assays were performed dose-dependently, thereby specifically
adapting the test substance concentration to the respective assay in order to obtain
a reliable dose-response curve. Ten nM E2 was used as a positive control, while 0.1 %
DMSO was used as a negative control. To investigate whether the estrogenic activity
of the isolated neutral alkaloids is mediated by ER activation, cells were incubated
with the effective doses of the substances in the absence or
presence of the pure antiestrogen Fulvestrant, also referred to as Faslodex and ICI
182 780 (purity > 99 %; Tocris) at the test concentration of 500 nM [34]. In all of the reporter gene experiments, cells were exposed to the test substances
24 h prior to the measurement of luciferase activity. All experiments were done in
triplicate and were independently repeated three times.
RNA isolation, cDNA synthesis, and mRNA quantification using realtime PCR
After a 24-h treatment period, the total cytoplasmic ribonucleic acid (RNA) was extracted
from adherent MCF-7 cells using Trizol® reagent (PeqLab) according to the manufacturerʼs
protocol. RNA samples were qualitatively examined on a 1 % agarose-formaldehyde gel.
DNA contamination was enzymatically eliminated by digestion (deoxyribonuclease I,
Ambion). Absence of genomic DNA was checked by PCR. M–MLV reverse transcriptase (Promega)
and oligo (dT) 12–18 primers were used for the first-strand cDNA synthesis. Quantitative
real-time PCR using Platinum® Taq DNA polymerase (Life
Technologies) and a thermal cycler with an iQ real-time detection system (BioRad)
was performed for mRNA quantification. SybrGreen I (Sigma-Aldrich) was used as a detection
probe. The reactions were run three times in triplicate. After vortexing, 50 µl aliquots
of the total mix were pipetted to each well of the 96-well PCR plate (BioRad). PCR
reactions consisted of a first denaturing cycle at 95 °C for 3 min, followed by 50
cycles of 10 s at 95 °C, 15 s at 60 °C, and 30 s at 72 °C. Fluorescence was quantified
at the end of the 60 °C annealing step and product identity was
confirmed by a melting curve analysis (60–95 °C). Primer sequences are summarized
in Table 1S, Supporting Information. For all genes measured, we obtained three different biological
replicates (mRNA preparations) from three independent cell culture experiments resulting
in at least three independently synthesized cDNAs from independent cell culture experiments.
Each cDNA was subjected to qPCR analysis in triplicate. The relative mRNA amounts
of target genes TFF1/pS2 and SGK3 were calculated after normalization to an endogenous
reference gene (ribosomal protein 18,
RPS18). Results are expressed as the relative amount of mRNA of the gene of interest
compared to the mRNA levels of the housekeeping gene hRPS18, using the 2-ΔΔCT formula [35].
Molecular simulation
All calculations were run using Macromodel 9.0 (Schrödinger, Inc.). Compoundsʼ 1 and 2 virtual structures (both R/S enantiomers regarding the NH group) were generated using Maestro 9.3.5 (Schrödinger,
Inc.). The full search in the conformational space for each molecule was achieved
using the OPLS2005 force field with an MC/LMOD search algorithm. One thousand starting
conformers were produced and minimized using the TNCG algorithm (rmsG < 0.01 kJ/mol
Å). No solvent model was used.
The ligand binding domains of ERα in the agonist conformation (PDB entries 3ERD and 2P15) were chosen as starting structures
for docking calculations. On each complex, the crystallographic ligand was replaced
by compound 1 or 2 (R/S enantiomers). Docking calculations were performed using a 1000 step search of the
mixed MC/LMOD search algorithm as implemented in Macromodel with a ratio of 0.5 and
an OPLS2005* force field. A distance-dependent dielectric “constant” of 4 r was used.
All residues within 6.0 Å from the ligand were allowed to
move freely, while the remaining residues were treated as “frozen atoms”. After each
successful run, the complex was minimized using the TNCG algorithm (rmsG < 0.01 kJ/mol
Å). Unique conformations were stored only if they were within the lowest 50 kJ/mol.
Data presentation and statistical analysis
All data from the receptor binding affinity are expressed as mean ± standard deviation
of at least three independent experiments. Data from luciferase reporter gene assays
were obtained from three independent cell culture experiments, within which treatments
were performed in triplicate. Data from real-time PCR experiments were obtained from
three different cell culture experiments, RNA extractions, and cDNA syntheses. Statistical
analysis included one-way analysis of variance (ANOVA) followed by Bonferroniʼs post
hoc test in order to determine significant differences.
Results are defined as significant at * p ≤ 0.05, ** p ≤ 0.01, and *** p ≤ 0.001.
Supporting information
Competition curves of ligand binding assays and 1 and 2D NMR data of isolated compounds,
an HPLC-DAD chromatogram of the methanol extract, data of the MTT test, and global
minimum structures of α- and β-erythroidine using molecular docking simulations are available as Supporting Information.
Acknowledgements
A part of this research project was supported by EU-FP7REGPOT-2011 project INsPiRE
(284460).
