ABSTRACT:

Organophosphate esters (OPEs) can exhibit various toxicities including endocrine disruption activity. Unfortunately, the low-dose endocrine-disrupting effects mediated by estrogen
receptors (ERs) are commonly underestimated for OPEs and their metabolites. Here, structure-oriented research was performed to investigate the estrogenic/antiestrogenic effect of 13 OPEs (including three metabolites) and the potential mechanism. All of the OPEs exerted antiestrogenic activities in both E-screen and MVLN assays. OPEs with bulky substituents, such as phenyl rings (triphenyl phosphate (TPP), tricresyl phosphate (TCP), diphenylphosphoryl chloride, and diphenylphosphite) or relatively long alkyl chains (dibutylbutylphosphonate (DBBP)), exerted relatively strong ER antagonism potency at micromolar concentrations. The established quantitative structure−activity relationship indicated that the antiestrogenic activities of the OPEs mainly depended on the volume, leading eigenvalue, and hydrophobicity of the molecule. Molecular docking revealed that the three OPEs with the bulkiest substituents on the phosphate ester group (TPP, TCP, and DBBP) have a similar interaction mode to the classical ER antagonist 4-hydroxytamoxifen. The correlation between the antiestrogenic activity and the corresponding ER binding affinity was statistically significant, strongly suggesting that the OPEs possess the classical antagonism mechanism of interfering with the positioning of helix 12 in the ER.

. INTRODUCTION

Organophosphate esters (OPEs) are a type of emerging environmental pollutant that are mainly used as flame pounds by 2019,2,3 and it is expected that the use of OPEs will increase because they are being used to replace brominated (TBP), tris(2-chloroisopropyl)phosphate, and tris(2-chloroethyl) phosphate (TCEP) exert ER antagonism potency in MVLN cells.20 Using the same testing assay, Liu et al.21 found that tris-(1,3-dichloro-2-propyl) phosphate (TDCPP),triphen-yl phosphate (TPP), and tricresyl phosphate (TCP) acted as ER antagonists,but TCEP did not. By contrast, based on an in vivo model, TDCPP, TPP, and TCP have been found to flame retardants in many applications.4−6 Consequently, OPEs study found that OPEs and their metabolites (e.g., TPP, TCP, and critically, the presence and biomagnification of OPEs have even been found in the ecosystem of Antarctica,10 which has raised scientific and public concern. Among their various acute toxicities, the best-described action of OPEs arises from inhibition of the enzyme increasing evidence that OPEs have endocrine disruption the levels of OPEs and their metabolites have close associations with endocrine-related health outcomes for humans (e.g., sperm concentration, low birth weight, and disrupting effects mediated by estrogen receptors (ERs) are commonly underestimated for OPEs and their metabolites, and moreover, the potential ER activity of certain OPEs is still controversial. For example, OPEs such as tributylphosphate and diphenyl phosphate (DPHP)) acted as ER antagonists in a yeast two-hybrid assay and showed estrogenic activity inan E- screen assay, but they only induced a minor response in MVLN cells.23 However, it should be noted that chemicals usually confer different ER responses with different tissues and animals because tissue/animal-specific cofactors are always involved in allosteric regulation, which may lead to different strong antiestrogeniceffects in human cells, but it functions as an agonist in yeast cells,24 whereas 4-n-nonylphenol exhibits estrogenic activity in human cells, but it acts as an antagonist in a yeast assay,25 which is due at least in part to the coregulatory proteins that help confer these bioactivities. Unfortunately, there is a lack of consistent results among investigations on the potential ER activity of OPEs and their metabolites, which could lead to underestimation of the ER activities for some OPE compounds. Thus, to exclude the effects of the cofactors and, moreover, to largely mimic the human condition, assays that are specific to the same human cell line/tissue are needed. Generally, OPEs have classical structural characteristics in their molecules, a phosphate ester linkage with three substituents (e.g., aromatic rings, halogenated alkyl chains, and alkyl chains). The abovementioned OPEs show the same basic structural characteristics but possess variations in one or more of the structural features. All of the evidence suggests that the cellular estrogen signaling pathway might be perturbed by OPE analogue exposure via structural-based ER activation, and this conclusion is important to assess the environmental and human health risk of an alternative structure. However, the mode of action of the estrogenic-like and/or antiestrogenic activity of these chemicals through ER-mediated pathways is not well established. In this context, studies that rapidly screen and systematically analyze the ERα-dependent estrogenicantiestrogenic activity of OPE analogues are urgently needed.
According to the 21st Embedded nanobioparticles century toxic testing report, in vitro assays combined within silico models are highly recommended for toxicity exploration of environmental pollutants.27 In
particular, molecular docking, quantitative structure−activity relationship (QSAR) prediction models, and well-designed invitro experiments have been successfully applied to determine and predict the molecular structural properties that mainly contribute to the interaction between compounds and nuclear estrogenic-like and/or antiestrogenic effects of 13 OPEs (including three metabolites) with different substituent groups, including phenyl rings, halogenated alkyl chains, and alkyl chains, we performed a comprehensive study by combining three in vitro methods that are specific to human breast tissue:E-screen assay, MVLN assay, and detection of estrogen- responsive genes in MCF-7 cells. Furthermore, a QSAR model was developed to identify and describe the structural features that contribute to the ability of the 13 analytes to interact with hERα. Finally, molecular docking was performed to obtain a better understanding of the interaction mode between the OPEs and hERα and to explore their structural requirements for ER activity.

MATERIALS AND METHODS

Chemicals.

TCP (90% purity, CAS No. 1330-78-5), diphenylphosphoryl chloride (DPPC, 99% purity, CAS No. 2524-64-3), dibutyl phosphate (DBP, 97% purity, CAS No. 107-66-4), TPP (99% purity, CAS No. 115-86-6), diphenyl- phosphite (DPP, 85% purity, CAS No. 4712-55-4), dibenzyl phosphate (DBzP, 99% purity, CAS No. 1623-08-1), DPHP (99% purity, CAS No. 838-85-7), dibutylbutylphosphonate (DBBP, 99% purity, CAS No. 78-46-6), methyl dichlorophos- phate (MDCP, 85% purity, CAS No. 677-24-7), diisopropyl- chlorophosphate (DICP, 97% purity, CAS No. 2574-25-6), triisopropyl phosphate (TiPP, 95% purity, CAS No. 513-02-0), diethyl cyanomethylphosphonate (DECP, 98% purity, CAS No. 2537-48-6), diethyl allylphosphonate (DEAP, 98% purity, CAS No. 1067-87-4), 17β-estradiol (E2, 99% purity), and 4- hydroxytamoxifen (OHT, 98% purity) were purchased from Sigma Chemical Company (St. Louis, MO). All of the tested chemicals, E2, and OHT were dissolved in dimethyl sulfoxide. All of the stock solutions were stored at −20 。C.

E-Screen Assay.

Generally, in response to ERα agonists, the mitotic effect leads to proliferation of MCF-7/BUS cells. In this study, we performed the E-screen assay utilizing MCF-7BUS cells, which were a generous gift from A. M. Soto and C. Sonnenschein (Tufts University School of Medicine, Boston, MA), following a method from the protocol reported by Wang et al.31 with minor modification. The MCF-7/BUS cells were cultured in 100 mm culture dishes in Dulbecco’s modified Eagle’s medium (DMEM) (Cellgro, Manassas, VA) consisting of fetal bovine serum (10%), streptomycin−penicillin (100 UmL), L-glutamine (2 mM), and insulin−transferrin−selenium supplement (1%) (all from Gibco, Grand Island, NY). The MCF-7/BUS cells were maintained in a humidified atmos- phere of 5% CO2 at 37 。C. The MCF-7/BUS cells were then trypsinized and placed in 12-well plates (12 000 cells/well). To minimize the basal hormonal activity during the assays, the cells were starved in a steroid-free (SF) medium for 48 h before each experiment. The SF medium contained phenol red-free DMEM (Hyclone) supplemented with dextran- charcoal-treated fetal bovine serum (5%;Hyclone), strepto- mycin−penicillin (100 U/mL), and L-glutamine (2 mM). The cells were treated with serial dilutions of the test chemicals (10 nM to 5 mM) in the SF medium. E2 (from 0.1 pM to 1 nM) was used as a positive control. Cell proliferation was measured by DNA content measurement, where the WST-1 assay was based on the measurement of an enzymatic reaction that is generally considered to be proportional to the number of cells. Here, a WST-1 proliferation kit (Roche Diagnostics, Mannheim, Germany) was used to determine the number of MCF-7 cells, which also indicates the cell proliferation effect, after 6 days of exposure according to the kit instructions. Consequently, the cell proliferation effect was obtained from the solvent control (0.1% dimethyl sulfoxide (DMSO))- corrected absorbance and expressed as the percentage of the maximal absorbance of the positive control. Three replicates of each experiment were performed.

MVLN Cell Assay.

The MVLN assay is based on human breast cancer cell line MCF-7/BUS stably transfected with the luciferase reporter gene and estrogen-responsive element derived from the Xenopus vitellogenin A2 gene. ER agonists/antagonists can induce/inhibit production of luciferase in MVLN cells, so the MVLN assay has been mainly used the MVLN cells,which were a generous gift from Prof. John P.Giesy (University of Saskatchewan, Saskatoon, Canada), were cultured in 100 mm culture dishes in DMEM (Cellgro). To minimize the basal hormonal activity, the cells were starved in an SF medium for 24 h before each experiment. Subsequently, the cells were seeded in the interior 40 wells of a 96-well ViewPlate (4 × 104 cells/well) (Packard Instrument Company, Boston, MA) and starved in the SF medium for another 24 h. A concentration range of E2 (from 0.1 pM to 10 nM) was used as a positive control, whereas the exposure concentration of the test chemicals ranged from 10 nM to 5 mM. The MVLN cells were incubated with the tested OPEs for 48 h. Subsequently, the cells were rinsed with phosphate-buffered saline (pH 7.4) and lysed with passive lysis buffer (25 μLwell), and then a Luciferase Reporter Assay kit (Promega, Madison) was used to measure the luciferase activity according to the manufacturer’s protocol. The luminescence was measured by a microplate reader for a 10 s luminescence signal (Varioskan Flash, Thermo Fisher Scientific, Waltham). The total protein content was measured by the Bradford assay (Tiangen, Beijing, China), so as to normalize the luminescent units. Accordingly, the results here are given as the relative luminescent unit per microgram of protein. The maximal luminescence induction of the positive control, which was corrected for the solvent control (0.1% DMSO), was set to 100%, and accordingly, the responses of the other tested compounds were converted to a percentage of the maximum level. To verify that the exposure concentration range of each target compound would not affect the cell viability (<20%),33 a WST-1 kit (Roche Diagnostics) was utilized to evaluate the cell viability of the 13 OPEs in parallel, and the cell morphology was also observed under a microscope. Each experiment was performed 3 times.

RNA Isolation and Reverse Transcription Polymerase Chain Reaction (RT-PCR).

MCF-7/BUS cells were added to 6-well plates at a density of 1 × 106 cells/well. Before the experiment, the cells were starved for 48 h in an SF medium and then incubated with the test OPEs for another 48 h. The complete methods for RNA isolation and RT-PCR are provided in the Supporting Information.

Molecular Docking.

Molecular docking calculations were performed to determine whether the tested molecules bind to hERα-LBD and to investigate their binding modes using the Molegro Virtual Docker (MVD) software package (provided by Prof. Shulin Zhuang, Zhejiang University, China). Before the molecular docking calculations, the tested molecules were geometry optimized with Gaussian 09W using the B3LYP hybrid functional and 6-31G(d) basis set. For fast and accurate identification of the potential binding modes, the heuristic search algorithm MolDock SE (simplex evolution) combined with MolDock Score [GRID] was used to evaluate the binding affinity of the ligands in the active pocket of hERα-LBD. After docking, energy optimization of the hydrogen bonds was permitted and the binding pose with the best rerank docking score was retained for post-analysis. During all of the docking process, the other parameters were set to the default values. The natural ligand OHT was docked in the crystal structure (PDBid:3ERT) with arerank score of −109.5, and a very low all-atom RMSD (0.05 Å), which was superior to the previous results of other docking software for the corresponding crystal complex.34 The natural ligand OHT docked in the crystal structure with arerank score of −109.5 and a very low all-atom root mean square deviation (0.05 Å). In the MVD calculation model, the rerank score is negatively correlated with the binding energy.

Structural Molecular Descriptors and Development of Predictive Relationships.

A DELL Precision 370 workstation was used to perform all of the molecular simulations, and the molecular descriptors were calculated using Dragon 6.0, except for the molar volume (MV), which was calculated by Gaussian 09W. In the model, PEC50 (−logEC50) of proliferation and the luciferase activity in both the single and coexposure assays were defined as the activity data. The details of the process of QSAR development and statistical significance verification can be found in our previous studies (see the Supporting Information).28,35

Statistical Analysis.

The data were analyzed for statistical significance by one-way analysis of variance and Tukey’s multiple ranges. All of the data are expressed as the mean ± standard deviation (SD). The difference was considered to be significantif the p-value was ≤0.05. The four-parameter logistic curve regression analysis method was used to analyze the dose−response relationship using EC50 according to the following model response value = minimum + (maximum − minimum) ÷ [1 + (x/EC50)Hill l_slope ] (1)
In this equation, x is the log concentration of the test compound and EC50 is the median effective concentration. The EC50 values were obtained from this nonlinear regression model. All of the statistical analyses were performed using Sigma Plot (version 12.5, Systat Software Inc., San Jose, CA) and SPSS 19.0 (SPSS Inc., Chicago, IL).

Results

Proliferation of MCF-7.

Two exposure assays, single exposure of the individual chemicals and combined exposure of the individual chemicals with E2, were performed to investigate the estrogenic-like and/or antiestrogenic activity of the 13 OPEs in MCF-7 cells by the E-screen assay. None of the OPEs showed proliferation effects in the single exposure assay (<±5%) (Figure S1). E2 was used as a positive control, and it induced the maximum cell proliferation at 1 nM with EC50 = 13.9 pM (Figure S1). OHT inhibited E2-induced proliferation in a dose-related Biogenic synthesis manner with EC50 = 11.8 nM and amaximum inhibition value of −19.31% (Table 1). The data of the proliferative effect caused by coexposure to a concentration range of the individual chemicals with E2 are provided in Table 1. DPPC, DPCP, DPP, TPP, TCP, and DBBP exerted strong antiestrogenic activities with EC50 values at micromolar concentrations of 34, 6.0, 14, 7.0, and 32 μM, respectively, which indicated that the antiproliferation effects of the OPEs were dependent on the substituents on the phosphate ester groups. For example, except for DBBP, the OPEs with phenyl ring substituents showed more potential for antiproliferation than the OPE with an alkyl substituent (Figure 1). All of these compounds were thus considered to act like antiestrogens in the E-screen assay.

Transactivation in MVLN Cells.

We used the MVLN assay to further investigate the ER activities of the OPEs. E2 was used as a positive control, and it induced the maximum ERE-luciferase value at 1 nM (EC50 = 79.7 pM) (Table 1). The steroidal antiestrogen OHT was used as a negative control, and it inhibited the luciferase activity in a dose-related manner with EC50 = 84.9 pM, and the maximum inhibition value was −16.2% (Table 1 and Figure 2). The response of the luciferase activity by the individual OPEs was in a dose-related manner (Figure 2). Accordingly, TCP, DPPC, DPP, and TPP with phenyl ring substituents showed relatively strong antiestrogenic potency with EC50 values of 7.68, 88.6, 97.6, and 28.8 μM, respectively. DBBP with three long alkyl chains on the phosphate ester group also showed relatively strong antiestrogenic activity with EC50 = 400 μM. For the metabolites, DBP and DBzP could inhibit the luciferase activities in dose-related manners with EC50 values of 2.0 and 9.7 mM, respectively. However,DPHP, which is the metabolite of TPP, only showed a minor response in this assay (<±5%). Therefore, all of the tested OPEs, except forDPHP, can inhibit the ERE-luciferase activity of MVLN in a dose-related manner, indicating that these compounds are ER antagonists.

Further experiments involving cotreatment with the E2 (1 nM) were performed in MVLN cells. OHT inhibited the E2- induced luciferase activity in a dose-related manner with EC50 = 16.4 nM, and the maximum inhibition value was −21.7% (Table 1 and Figure S2). All of the tested OPEs were able to antagonize the estrogenic effect of E2 in a dose-dependent manner (Figure S2). TPP (with three phenyl ring substituents) exerted the most potent antiestrogenic activity in this cotreatment assay, with EC50 = 29.1 μM and a maximum effect of 22.2%. DPP (with two phenyl ring substituents) was the next most potent with EC50 = 46.8 μM (Table 1). Additionally, TCP (three phenyl ring substituents), DPPC (two phenyl ring substituents), and DBBP (three alkyl substituents) also exerted relatively strong ER antagonism potency with EC50 values of 105, 218, and 304 μM, respectively. These results indicated that all of the tested OPEs can inhibit the E2-induced ER activity and their corresponding antiestrogenic activities are structure depend- ent, which is consistent with results obtained from the E-screen assay.

Expression Alterations of Estrogen-Related Genes.

Trefoil factor 1 (TFF1/PS2) and early growth response protein 3 (EGR3) are estrogen-regulated genes. In particular, the TFF1 which belongs to the early growth response family,38 plays an
important role in estrogen-dependent induction of the immune evasion system. Commonly, both endogenous and exogenous estrogens can induce the expression of TFF1 and EGR3 genes analysis experiment was performed to determine if OPE treatment disrupted ER-mediated signaling. We treated MCF-7 cells with either 1 nM E2, the neat OPE, or 1 nM E2 and the corresponding OPE for 48 h, and the expression of the estrogen-regulated genes (TFF1 and EGR3) was evaluated. The results showed that the expression of TFF1 and EGR3 was significantly stimulated by E2 treatment (Figure 3). The mRNA levels of TFF1 and EGR3 were not significantly different after treatment with the 13 OPEs alone. By contrast, all of the tested OPEs significantly attenuated the TFF1 and EGR3 genes in a different range when they were administered with E2 (p<0.05). This provides further evidence that the OPEs can inhibit the estrogenic potency of E2, which would disrupt the ER-mediated signaling pathway, and they actas ER antagonists.

Development of Predictive Relationships.

The activity data and descriptors of the 13 OPEs are given in Table S1. Accordingly, pEC50(1) of the proliferation effect in the coexposure assay, pEC50(2) of the luciferase activity in the single exposure MVLN cell assay, and pEC50(3) of the luciferase activity in the coexposure MVLN cell assay were defined as the dependent variables. Using the molecular structural parameters as predictor variables gave the optimal QSAR models (Table 2). All of the candidates share common characteristics, that is, phosphate ester groups. The molecular descriptors, edge adjacency indices, find more and getaway indices were maintained. Monte Carlo simulation gave a pseudo-coefficient of correlation at the 95% confidence interval (R*2) of 0.450 for the proliferation activity, which was much less than the original R2 (0.672) (Figure S3). Accordingly, Monte Carlo simulation gave pseudo-coefficients of correlation at the 95% confidence interval (R*2) of 0.447 for the luciferase activity in the single exposure assay and 0.449 for the luciferase activity in the coexposure assay, which were much less than the original R2 values (0.777 and 0.708, respectively) (Figures S4 and S5). Therefore, it was confirmed that in our three models, the difference between the model prediction capacity and the random prediction was statistically significant and chance correlation was avoided.
AMR (Ghose−Crippen molar refractivity, molecular proper- ties) and SPmax_AEA (leading eigenvalue from the augmented edge adjacency matrix weighted by the bond order, edge adjacency indices) were entered into the final model for the proliferation effects in the coexposure assay model. The AMR is used to describe the molecular “shape” with three basic descriptors,40 and the eigenvalues represent a fraction of the information residing in the eigen system of the adjacency matrix.41 Similarly, the MV also encodes molecular shape information, and Hy is also a function of the number of hydrophilic groups and the carbon atoms in the molecule.42 Hy (hydrophilic index) and MV (molar volume) were entered into the final model for the luciferase activity in the single exposure assay, whereas AMR and SPmax_AEA were entered into the final model for the luciferase activity in the coexposure assay model. The credible model of the predictive relationship between the luciferase activity and the molecular descriptors (MV and Hy) suggested that the molecular volume and hydrophobicity properties can be used as reliable descriptors to predict the antiestrogenic activity of OPE exposure alone. Interestingly, the cell proliferation and luciferase activity coexposure model had the same final encode parameters (AMR and SPmax_AEA), which further indicated that the molecular shape and the leading eigenvalue were the governing molecular descriptors for the antiestrogenic potential of the OPEs for cotreatment with E2. Based on this, it is believed that the chemical properties, including AMR, SPmax_AEA, MV, and Hy, of the target chemicals might be the critical structural factors that affect the antiestrogenic activities of OPEs.
Docking Calculation by MVD. The docking results showed that all of the tested OPEs can bind with ER-LBD in a different range (Figures 4 and S6). The corresponding rerank scores of the 13 OPEs are given in Table S1. The OPEs with phenyl ring substituents gave relatively low scores, suggesting more potential in binding with ER. DBBP with alkyl substituents on the phosphate ester also gave a relatively low rerank score. Notably, three of the OPEs (TCP, TPP, and DBBP), which are structurally similar to the known estrogen antagonist (OHT), had a similar binding mode to OHT in the antagonist conformation of hERα-LBD. However, the H- bonding interactions between Glu353 or Arg394 and the ligands were not found for the tested OPEs, which lead to decreases of their binding affinities with ER compared to OHT (Figure 4). The OPEs with three substituents (TCP, TiPP, DBBP, DEAP, and TPP) were chosen to analyze the relationship between the chemical structure and the anti- estrogenic potency. The correlation coefficients (R2) between the antiestrogenic activities pEC50(1), pEC50(2), and pEC50(3) (Table S1) and ER binding scores of these analytes were 0.917 (p = 0.010), 0.843 (p = 0.028), and 0.984 (p = 0.001), respectively. These significant positive associations suggested that they may possess the classical antagonism mechanism by interfering with the positioning of helix 12 (H12) in the ER (Figure 4).

. DISCUSSION

The endocrine-disrupting effects induced by environmental pollutants are of great concern. Antagonizing the effect of endogenous hormones is one of the pathways by which xenobiotics exert their endocrine effects.43 A compound is considered to be a true estrogen antagonist when (a) in the absence of E2, it cannot induce a change of the cell number, (b) it inhibits the cell proliferation induced byE2, and (c) this inhibition is reversed by gradually increasing the concentration of E2 while the dose of the tested compound is unchanged.43 In this study, we used the E-screen assay to test the ER activities of the 13 OPEs. The results showed that all of the OPEs did not affect the cell number with a single exposure. However, in the coexposure assay, all of the tested OPEs inhibited proliferation of MCF-7 cells induced by E2, and we also observed a dose dependency of this inhibition by increasing the exposure dose of the individual OPEs while keeping the dose of E2 constant at 1 nM. It should be noted that we modulated the chemical concentration here instead of E2. E2 induced the maximal cell proliferation at 1 nM in the E-screen assay system of this study, which suggested that E2 could completely bind with ER-LBP in MCF-7/BUS cells at a concentration of 1 nM. Thus, chemicals that inhibited the ER activities in a dose-related manner in the presence of 1 nM E2 are considered to be ER antagonists. Additionally, all of the tested OPE compounds could inhibit ER transactivation in MVLN cells,when exposed alone or in cotreatment with 1 nM E2. Furthermore, the results from RT-PCR analysis indicated that the OPEs inhibited E2-induced TFF1 and EGR3 gene transcription to different extents, suggesting that the OPEs can attenuate the endogenous ERα function in MCF-7 cells. This result further confirmed that the OPEs exerted antiestrogen activity through ER-mediated pathways. Of note, the EC50 values for the inhibition effects of certain OPEs on AChE are also at the micromolar concentration level, for example, the EC50 values for tetraethyl pyrophosphate and phoxim ethyl therefore reasonable for us to raise a concern about the ER antagonism activities of the OPEs because they may exert antiestrogenic effects at a concentration comparable with that in their anticholinesterases.
Considerable evidence has indicated that a phenyl ring substituent is an important structural feature because it anchors to the molecular recognition site of ER receptors.46 Thus, of particular interest in this study was the important structural elements in the interactions of the 13 structurally similar OPEs with the ER. The critical element of the ER’s ligand-dependent transcriptional activation function (AF-2) conformation switch is a short helical region (H12), and different classes of ligands influence the orientation of H12, which will directly determine the conformation of the ER (agonist or antagonist). Generally, ER antagonist ligands inhibit the AF-2 activity by interfering with the positioning of H12 resulting in blocking of coactivator recruitment.47 Accordingly, the results from the E-screen and MVLN cell assays indicated that the structural features of the phosphate ester group are crucial for the antiestrogenic potency of OPEs. In particular, compounds with three side chains on the phosphate ester group showed better antiestrogenic activity than those with two side chains. OPEs with three phenyl ring substituents on the different sides of the phosphate ester group showed better antiestrogenic activities than those with three alkyl substituents. The in vitro findings were corroborated by in silico modeling. The molecular descriptors obtained from the QSAR model suggested that compounds with some degree of molecular bulk,a leading eigenvalue, and hydrophobicityare favorable for the enhanced antiestrogenic activity. Moreover, molecular docking provided evidence to support the fact that OPEs with three side chain substituents on the phosphate ester group have a similar interaction mode to the classical ER antagonist (OHT), that is, they interfere with the positioning of H12 in the ER.
In conclusion, we have demonstrated that all of the tested OPEs are estrogen antagonists based on both comprehensive in vitro and in silico evaluation experiments, they confer their activities at micromolar concentrations, and their potential mechanism is structure based. The structural features of the phosphate ester group substituents and the side chain number are crucial for the antiestrogenic potency of the OPE. The antiestrogenic activities of the OPEs are mainly dependent on the molecular bulk, leading eigenvalue, and hydrophobicity. Furthermore, OPEs with three large side chains on the phosphate ester group may have a classical ER antagonism mechanism by interfering with the positioning of H12 in the ER. Special attention should be paid to the structural properties that determine the antiestrogenicity of OPEs, including their metabolites, to ensure a sound ambient environment. Overall, we first provide data from a new perspective to predict the low-dose endocrine-disrupting toxicity endpoints as well as to evaluate the exposure risk of OPEs and/or OPE-like chemicals, including their metabolites, which is extremely meaningful.